bmjopen.bmj.com · For peer review only Technical Appendix 3 Covariates BaselineHbA1c Baseline...

For peer review only

A novel model-based meta-analysis to indirectly estimate the comparative

efficacy of two medications: an example using DPP-4 inhibitors, sitagliptin and linagliptin, in treatment of type 2

diabetes mellitus

Journal: BMJ Open

Manuscript ID: bmjopen-2012-001844

Article Type: Research

Date Submitted by the Author: 20-Jul-2012

Complete List of Authors: Gross, Jorge; Hospital de Clinicas de Porto Alegre, Endocrine Division, Department of Internal Medicine Rogers, James; Metrum Research Group, Polhamus, Daniel; Metrum Research Group, Gillespie, William; Metrum Research Group,

Friedrich, Christian; Boehringer Ingelheim, Gong, Yan; Boehringer Ingelheim, Monz, Brigitta; Boehringer Ingelheim, Patel, Sanjay; Boehringer Ingelheim, Staab, Alexander; Boehringer Ingelheim, Retlich, Silke; Boehringer Ingelheim,

<b>Primary Subject Heading</b>:

Pharmacology and therapeutics

Secondary Subject Heading: Diabetes and endocrinology

Keywords: General diabetes < DIABETES & ENDOCRINOLOGY, Diabetes & endocrinology < INTERNAL MEDICINE, STATISTICS & RESEARCH METHODS

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml

BMJ Open on A

pril 6, 2020 by guest. Protected by copyright.

http://bmjopen.bm

j.com/

BM

J Open: first published as 10.1136/bm

jopen-2012-001844 on 5 March 2013. D

ownloaded from

http://bmjopen.bmj.com/


1


efficacy of two medications: an example using DPP-4 inhibitors, sitagliptin and

linagliptin, in treatment of type 2 diabetes mellitus

Jorge L Gross,1 James Rogers,2 Dan Polhamus,2 William Gillespie,2 Christian Friedrich,3 Yan

Gong,4 Brigitta Monz,4 Sanjay Patel,5 Alexander Staab,3 Silke Retlich3

1Hospital de Clínicas de Porto Alegre, Universidade Federal do Rio Grande do Sul, Porto

Alegre, RS, Brazil

2Metrum Research Group, Tariffville, Connecticut, USA

3Boehringer Ingelheim, Biberach, Germany

4Boehringer Ingelheim, Ingelheim, Germany

5Boehringer Ingelheim, Bracknell, Berkshire, UK

Correspondence to Silke Retlich, [email protected]

Running title: Novel MBMA for indirect comparison of diabetes treatments

Keywords: dipeptidyl peptidase-4 inhibitors, HbA1c, model-based meta-analysis, type 2

diabetes mellitus

Page 1 of 48


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960

on April 6, 2020 by guest. P

rotected by copyright.http://bm

jopen.bmj.com

/B

MJ O

pen: first published as 10.1136/bmjopen-2012-001844 on 5 M

arch 2013. Dow

nloaded from



2

Previous Presentations: Abstracts based on this study have been presented as posters at the

72nd Scientific Sessions of the American Diabetes Association, 8–12 June, 2012, Philadelphia,

USA, and at the Population Approach Group Europe (PAGE) conference, Venice, Italy, 5–8

June 2012.

Page 2 of 48


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



3

ABSTRACT

Objectives: To develop a longitudinal statistical model to indirectly estimate the comparative

efficacies of two drugs, using model-based meta-analysis. Comparison of two oral dipeptidyl

peptidase-4 inhibitors, sitagliptin and linagliptin, for the treatment of type 2 diabetes mellitus

(T2DM) was used as an example.

Design: A systematic review with model-based meta-analysis (MBMA).

Data sources: MEDLINE, Embase, publications on www.ClinicalTrials.gov, Cochrane review of

DPP-4 inhibitors for T2DM, sitagliptin trials on FDA website to December 2011, and individual

patient data from the manufacturer of linagliptin.

Eligibility criteria for selecting studies: Double-blind, randomised, controlled, clinical trials, at

least 12 weeks in duration, that analysed the efficacy of sitagliptin or linagliptin as changes in

glycated haemoglobin (HbA1c) levels, in adults with T2DM and HbA1c >7.0%, irrespective of

background medication.

Model development and application: A Bayesian model was fitted (Markov Chain Monte

Carlo method). The final model described HbA1c levels as function of time, dose, baseline

HbA1c, washout status/duration and ethnicity. Other covariates showed no major impact on

model parameters and, therefore, were not included in the final model. For the indirect

comparison, a population of 1000 patients was simulated from the model with a racial

composition reflecting the average racial distribution of the linagliptin trials, and a baseline

HbA1c of 8.0%.

Page 3 of 48


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



4

Results: Longitudinal data from 11 234 patients (10 linagliptin and 15 sitagliptin trials) were

included. Simulations showed that both linagliptin 5 mg and sitagliptin 100 mg reduced HbA1c

by 0.8% (placebo-adjusted) at week 24. Credible intervals for participants without washout were

–0.88 to –0.74 (both groups), and for those with washout were –0.91 to –0.75 (linagliptin) and –

0.90 to 0.75 (sitagliptin), when administered for 24 weeks.

Conclusions: This model seems a valid approach for indirect drug comparisons. The results

show sitagliptin and linagliptin have virtually indistinguishable efficacies in HbA1c reduction in

patients with T2DM.

Page 4 of 48


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



5

ARTICLE SUMMARY

Article focus

• In the absence of evidence from head-to-head trials, indirect and mixed treatment

comparisons can be used for drug comparisons.

• The aim of this study was to develop an approach, using Bayesian methodology (Markov

Chain Monte Carlo method), to indirectly estimate the comparative efficacy of two

compounds, incorporating longitudinal and/or dose–response data.

Key messages

• A longitudinal statistical model was developed to compare the efficacy of two oral DPP-4

inhibitors, sitagliptin and linagliptin, with respect to changes in HbA1c levels in patients

with type 2 diabetes mellitus (T2DM).

• The model demonstrated that both linagliptin and sitagliptin reduced HbA1c levels by

0.8% (placebo-adjusted) when administered to patients with T2DM for 24 weeks,

irrespective of background medications.

Strengths and limitations of this study

• This study represents a novel use of longitudinal model-based meta-analysis in the field

of diabetes treatment, being the only instance to date that adequately accounts for

longitudinal correlations in each treatment arm, which is a prerequisite to the correct

characterisation of uncertainty in estimation of drug effects.

• When relevant head-to-head comparisons are not available, the model described in this

study could have an important role in treatment decision-making.

Page 5 of 48


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



6

• Although the analysis included a large sample of 11 234 patients with T2DM, its

applicability to the general population of patients with T2DM might be limited by the

relatively selected patient populations in the included trials. Additionally, while our

analysis adjusts for key differences in study designs, there remains the possibility of bias

attributable to covariate effects that could not be estimated with the available data.

Page 6 of 48


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



7

INTRODUCTION

Ideally, head-to-head, randomised, controlled trials should be conducted to estimate the

comparative efficacy of different treatments. However, it is not always feasible to conduct direct

comparisons among all available treatment options. Indirect comparisons and network meta-

analysis (mixed treatment comparisons) have been used to estimate relative efficacy when

there are no direct comparative data, to provide the best available evidence to facilitate

decision-making by physicians and other stake-holders, such as payers. However, these

approaches have certain limitations, including the risk of bias arising from inherent differences in

the designs of the included studies, and the difficulties of finding appropriate summary statistics

to compare the findings of individual trials.1 2

An approach, recently described as model-based meta-analysis (MBMA), has been

developed to estimate the comparative efficacy of two medications. MBMA can be used to

provide a mechanism for integrating information from heterogeneously designed trials and, thus,

to evaluate outcomes with different drugs that have not been compared directly.3 Model-based

meta-analysis is distinguished from the methodology of conventional meta-analysis by the

manner in which it incorporates longitudinal and/or dose–response data. By modelling the

response as a parametric function of time, MBMA allows the integration of information from

trials of different durations and with different sampling time points. This enables the use of less

restrictive inclusion/exclusion criteria for study selection, and more efficient use of data from the

studies that are selected, therefore resulting in a particularly comprehensive summary of all

relevant data.3

In response to the growing worldwide epidemic of diabetes mellitus, new anti-

hyperglycaemic agents are continuously being developed. The dipeptidyl peptidase (DPP)-4

inhibitors are a relatively new class of oral anti-hyperglycaemic drugs developed for the

treatment of type 2 diabetes mellitus (T2DM) that are increasingly being used in clinical practice

Page 7 of 48


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



8

because of their clinically meaningful efficacy, promising tolerability, safety and convenience –

in particular, a virtually absent risk of hypoglycaemia or weight gain.4 Although several DPP-4

inhibitors are already available in many countries, to date, only one published trial has been

conducted to directly compare individual drugs within this class.5 Therefore, further research is

needed to understand the comparative effects of the drugs within this class.

The model developed in this study incorporates Bayesian methodology and aims to

provide a valid approach to estimate the comparative efficacy of different compounds. Bayesian

approaches are acknowledged by the Cochrane Collaboration to have a role in meta-analysis,

particularly in the setting of indirect comparison.1

Objective

To use an MBMA approach to develop a longitudinal statistical model for the comparison of the

efficacy of two oral DPP-4 inhibitors, shown by changes in glycated haemoglobin (HbA1c)

levels, in patients with T2DM who had started treatment with one of two DPP-4 inhibitors,

regardless of background medication. The two drugs evaluated were linagliptin, which has

recently been approved for clinical use in several jurisdictions, and sitagliptin, the most

commonly used DPP-4 inhibitor.

METHODS

Data sources

Sitagliptin studies were identified from a systematic search in MEDLINE, Embase, studies listed

on www.ClinicalTrials.gov that included a reference to publication, the latest-date Cochrane

review of DPP-4 inhibitors for T2DM6 and details of sitagliptin trials on the Food and Drug

Administration (FDA) website, to December 2011.7 Details of the search strategy used are

provided in the Appendix (supplementary table 1).

Page 8 of 48


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



9

Results of the relevant studies for linagliptin were obtained from the manufacturer’s database,

several of which have been subsequently published as full papers8-11 or abstracts.12-14

Study selection

Included studies were double-blind, controlled, randomised trials of at least 12 weeks’ duration

that analysed the efficacy of sitagliptin or linagliptin in the reduction of HbA1c levels in adults

with T2DM and HbA1c >7.0%, irrespective of background medication. Excluded studies were:

open-label studies (and data from open-label extensions to double-blind studies) and extension

studies that used patient response in the initial study to determine eligibility in the extension

phase of the study (eg, if the extension phase included only those who did not require rescue

medication during the initial study). Other excluded study types were special population studies

(eg, studies in patients with declining renal function) and phase IV studies or study arms in

which patients were randomly assigned to initial combination therapies.

Two independent reviewers extracted aggregated data from all selected studies,

according to treatment arm (sitagliptin, linagliptin or placebo), by. We extracted data on: the first

author’s name, year of publication of the trial, comparator, dose(s) of sitagliptin or linagliptin

evaluated, trial duration, number of participants, and their gender, ethnicity, duration of T2DM,

mean age, baseline HbA1c (%), HbA1c at evaluated time points, baseline body mass index

(BMI, kg/m2), fraction of patients on previous anti-hyperglycaemic therapy, presence and

duration of washout and concomitant medication. A common data template was defined. The

main outcome of interest was HbA1c, the primary end point of all included studies. Intention-to-

treat (ITT) populations were included and group means, as reported, were used or were

calculated, using the last observation carried forward (LOCF) approach. The analyses were

conducted using the maximum licensed dose of sitagliptin (100 mg) and the licensed dose of

linagliptin (5 mg). However, when data at other dose levels were available, they were included

Page 9 of 48


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



10

in the analysis, and appropriate adjustments were made via the dose–response terms in the

model.

Data selection process

For the linagliptin studies, the dataset was built from the original Boehringer Ingelheim database

using SAS scripting. The quality of the dataset was assured by an independent script review.

For the sitagliptin studies, the dataset was built manually by collecting information given in the

different source publications. If the results were available as numbers in the publications, these

numbers were included in the dataset. Where the results were only available as graphics, the

corresponding data were collected using GetData Graph Digitilizer, version 2.24 software

(http://www.getdata-graph-digitizer.com). The quality of the manually built sitagliptin dataset was

assured by an independent second reviewer. The initial dataset consisted of HbA1c data,

presented as either the change from baseline and/or the actual HbA1c measurements,

depending on the information provided in the publication. R scripting (R version 2.10.1, The R

Foundation for Statistical Computing, Vienna, Austria) was then used to obtain an analysis-

ready dataset with consistent encoding of information (eg, baseline values were added to

changes from baseline in order to obtain actual HbA1c measurements for all records).15

Statistical analysis

Model development

Initial exploratory data analyses were used to derive a suitable parametric (algebraic)

description of the average HbA1c trends as a function of time, dose, washout status/duration

and ethnic origin. Qualitative prior information was also used to guide the initial selection of

parametric forms. The following assumptions were made: (1) Given the known properties of

measured HbA1c, it was assumed that in the absence of additional interventions, HbA1c levels

for patients washing out prior antidiabetes medication (during the study washout/run-in phase)

Page 10 of 48


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



11

would rise for some time until achieving a plateau, and (2) the incremental (placebo-adjusted)

effect of DPP-4 inhibitors on HbA1c was expected to approach a plateau during the time frame

of interest (24 weeks).

Inter-arm random effects were incorporated in order to reflect correlations amongst

observations at different time points, for both linagliptin and sitagliptin as well as for placebo.

Study-level random effects were also incorporated in recognition of potential inter-study

heterogeneity. Sample size-based adjustments for model precision terms were incorporated, in

a manner that accounts for longitudinal corrections, following an approach described

elsewhere.3

The model was fitted using Bayesian Markov Chain Monte Carlo (MCMC) methodology.

The computations were carried out using OpenBUGS version 3.2.1 (2010) software (Free

Software Foundation, Boston, MA, USA). The model was adjusted for baseline HbA1c and

washout status/duration. Other covariates considered were: standard covariates including

demographics, such as ethnicity, age, BMI and gender, anti-hyperglycaemic background

medication, duration of T2DM and the fraction of patients who underwent washout of previous

anti-hyperglycaemic therapy. The WinBUGS code is available from the authors, on request.

Model selection and evaluation

Following a ‘full model estimation approach’,16 17 initial preference was given to a full model,

meaning one that includes all terms of potential interest. In order to achieve stable parameter

estimation, selective simplifications were applied, guided by exploratory data analysis, to the full

model until we obtained satisfactory convergence diagnostics. Covariates were excluded from

the model for the purpose of achieving stable parameter estimation; however, each excluded

covariate was evaluated graphically to ensure that it was not associated with model residuals

(differences between the observed values and those predicted by the model). A graphic

Page 11 of 48


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



12

representation of the final model, for patients with or without a pre-randomisation washout

period, is shown in figures 1A and 1B.

The final model was evaluated, using posterior predictive check methodology, as reported

elsewhere,18 and demonstrated that it adequately described the data. This model inherently

adjusted for baseline HbA1c and washout status/duration. The other covariates (see above),

with the exception of ethnicity, showed no major impact on the model parameters and were

therefore not included in the final model.

Application of the model: comparison of linagliptin and sitagliptin

In order to assess the efficacy of the two DPP-4 inhibitors in comparable patients under similar

conditions, a population of 1000 patients was simulated from the model under reference

conditions. Data for each patient were simulated as if arising from an individual trial, so that the

resulting inference represents an average over the expected range of inter-trial variation. The

population size of 1000 was chosen to reduce sampling error to negligible levels. The reference

racial composition for this simulated population was 61.5% white, 1.5% black and 37.0% Asian,

reflecting the average enrolled distribution in linagliptin trials. The simulated average baseline

HbA1c (%) in this population was 8.01. Results are expressed as mean differences, with 95%

credible intervals (the Bayesian equivalent of confidence intervals).

RESULTS

A total of 31 sitagliptin studies were assessed for eligibility for inclusion in the analysis, and 16

were excluded on the basis of the study design that did not meet our inclusion criteria (figure 2;

supplementary table 2). A further 10 linagliptin studies were included.

Page 12 of 48


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



13

The included studies were between 12 and 26 weeks in duration, with one exception

(the study by Seck et al. 201019 lasted 104 weeks) (Table 1).

Data from a total of 11 234 participants were included in the analysis, arising from 25

randomised trials (10 linagliptin and 15 sitagliptin) (figure 2). The mean age at baseline of all

study participants was 56.5 years, with reported means for treatment arms of the included

studies ranging from 50.9 to 62.0 years; the proportion of females across all study participants

was 45.5%, with reported proportions for study groups ranging from 22.8% to 64.0%; the mean

BMI was 29.7 kg/m2, with reported means for treatment arms ranging from 24.1 to 32.7. Mean

baseline HbA1c was 8.0%, with reported means for treatment arms ranging from 7.49% to

8.87%. The most commonly used background medication was metformin monotherapy.

Metformin was also used in combination with glimepiride or pioglitazone, and one study30

included patients receiving initial monotherapy with pioglitazone.

Figure 3 depicts the application of the statistical model to each individual study,

demonstrating that the observed data from the studies fall mostly within the 90% prediction

interval (between 5% and 95% prediction bounds), with no overall systematic over- or under-

prediction. The simulations performed using the model show that both linagliptin 5 mg and

sitagliptin 100 mg reduce HbA1c levels by 0.8% (placebo-adjusted), at week 24, when

administered to patients with T2DM for 24 weeks (figures 4A and 4B). Credible intervals for

participants without washout were –0.88 to –0.74 (both treatment groups), and for those who

underwent washout of previous anti-hyperglycaemic therapy, –0.91 to –0.75 (linagliptin) and –

0.90 to –0.75 (sitagliptin).

As a post hoc assessment, a t test was used to compare the HbA1c difference from

placebo residuals (unexplained variations after fitting of the model) for linagliptin and sitagliptin.

A p-value of 0.14 was generated, suggesting no evidence of a systematic bias in favour of

linagliptin by conventional thresholds (p<0.05).

Page 13 of 48


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



14

DISCUSSION

The model developed in this study incorporates Bayesian methodology and provides a novel

approach to indirectly estimating the comparative efficacy of two compounds. The findings

presented suggest that the model developed in this study provides a valid alternative approach

to indirect drug comparisons. The findings of this MBMA show that linagliptin is equally effective

as sitagliptin in the reduction of HbA1c levels, both showing a mean, placebo-adjusted reduction

of approximately 0.8% after 24-week treatment of patients with T2DM. In this study, evidence

was gathered from the results of randomised, double-blind trials of sitagliptin and linagliptin.

Sensitivity analyses performed in this study, using various prior distributions, support the

robustness of the model. As far as we know, the present MBMA represents one of the first such

applications of these techniques in the field of diabetes treatment.

There might be some limitations in applying the findings of the present analysis to the

general population of patients with T2DM because of the relatively selected patient populations

in the included trials, which included mostly white, middle-aged patients with mean baseline

HbA1c <9%. The participants in the analysed trials would have been further restricted during

pre-trial run-in periods, which would exclude those with poor treatment adherence. Furthermore,

the analysis was performed retrospectively, using data from different trials. As with all meta-

analyses based on published data, there is a potential for publication bias. In the context of the

present analysis, this potential bias pertains only to our estimates of the effects of sitagliptin, as

our linagliptin data sources were not subject to publication selection. However, this is unlikely to

have a substantial impact on the findings for sitagliptin, as current practice in clinical research

mandates that all clinical trials are published regardless of their results, and as several sources

were searched, including trial registries and documents used in the regulatory process.

Page 14 of 48


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



15

The model includes the assumption that HbA1c levels are maintained after the full effect

of treatment has been reached. This is based on observations in previous 24-week trials, where

HbA1c levels have been shown to be maintained for this period,9-11 36 and the known

pharmacological properties of DPP-4 inhibitors.4 37 38 The final model was adjusted for baseline

HbA1c, ethnic origin and washout duration. Other covariates (concurrent medications, fraction

of patients on previous oral antidiabetic drugs, BMI, age, gender, duration of T2DM) were not

included in the final model because they did not show significant impact on the model

parameters. Reasons for this might be either that only mean covariate values were available, or

that some covariates are confounded (eg, BMI was shown to vary as a function of ethnic origin,

making it difficult to isolate the independent effects of these covariates). It is important to

recognise that these covariates might be of clinical importance, and their exclusion from the

model could simply reflect an inability to reliably estimate the independent effect of these factors

with the data available.

To date, four standard meta-analyses of the DPP-4 inhibitor class have been published,

none of which has provided any results on the comparative efficacies of linagliptin and

sitagliptin.6 39-41 These analyses confirm the efficacy of DPP-4 inhibitors, in terms of HbA1c

reduction, and their tolerability, in particular resulting from the absence of weight gain and low

risk of hypoglycaemia associated with monotherapy. The findings also indicate that therapy with

DPP-4 inhibitors reduces HbA1c reductions to a similar extent to comparator drugs.39 Several of

the limitations associated with traditional meta-analysis arise from the fact that only study end

point data are used in these analyses. For example, difficulties in selecting an appropriate

summary statistic are often encountered because the treatment effect of interest varies as a

function of the duration of treatment. Similarly, it might be difficult to appropriately adjust for the

effect of covariates on treatment response when response is assessed at different time points in

different studies. To address the limitations of traditional meta-analysis, a general methodology

has recently been proposed for the statistically valid use of MBMA.3 This analytic approach has

Page 15 of 48


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



16

been used in one other recent study, by Gibbs et al. 2011,42 which evaluated the relationship

between DPP-4 inhibition and HbA1c reduction using data obtained from clinical trials of four

drugs in this class.42 The advantage of this approach, also used in the present study, is that it

enabled the synthesis of longitudinal data from multiple studies with different durations and

different sampling schedules, resulting in analyses that are both more comprehensive (including

a greater number of studies), and more efficient (incorporating more of the relevant data within

each study) than previous methods. The unique MBMA approach in the current study also

allows adjustment for covariates (eg, differences in the use of washout or racial composition in

individual trials) to allow comparison of treatment response in comparable patients under similar

conditions. One limitation of the study by Gibbs et al. 201142 was that the MBMA used did not

account for correlations across time points within treatment arms, which could lead to an

overestimation of the inter-trial variability in drug effect. In contrast, our approach takes account

of longitudinal correlations, in accordance with previously published methods,3 which is a

prerequisite to the correct characterisation of uncertainty in the estimation of drug effects.

As the clinical use of DPP-4 inhibitors increases, patients, prescribers and payers will

require information on the relative benefits of the individual drugs within this class. This study

demonstrated that the efficacy of the two DPP-4 inhibitors, sitagliptin and linagliptin, is virtually

indistinguishable, in terms of changes in mean HbA1c levels, in patients with T2DM treated with

a range of background anti-hyperglycaemic therapies. Both linagliptin and sitagliptin act by

inhibiting the DPP-4 enzyme that rapidly inactivates the intestinal hormone, glucagon-like

peptide (GLP)-1. GLP-1 stimulates insulin secretion in a glucose-dependent manner. Sitagliptin

is largely excreted via the kidneys, with the majority of an oral dose (87%) being excreted in the

urine.43 Unlike sitagliptin and other DPP-4 inhibitors, linagliptin has a largely non-renal route of

excretion (only ~5% excreted renally), with the majority being eliminated via the bile and gut;44 45

it therefore does not require dose adjustment in patients with declining renal function.46 In view

of the similar efficacy of these two drugs, treatment choices might, therefore, be made on the

Page 16 of 48


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



17

basis of other differences between the drugs and consideration of patient clinical characteristics,

such as the patient’s renal function.

Broadening the use of MBMA has the potential to improve the comparison of individual drug

therapies, compared with older statistical methods, and could provide a new way of generating

results for populations that have not yet been studied.

Page 17 of 48


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



18

Contributors

All authors were fully responsible for all content and editorial decisions, and were involved at all

stages of manuscript development, including reviewing and revising the manuscript for scientific

content and have approved the final version. In addition: JG contributed to the data analysis and

interpretation of the findings. JR, DP and WG shared primary responsibilities for developing the

statistical analysis plan, executed all statistical analyses (including model development, model

selection and model summary), and interpreted the findings. SP monitored data collection, and

contributed to data selection, the statistical analysis plan and interpretation of the results. CF

contributed to the analysis concept, the statistical analysis plan and interpretation of the

findings. BM contributed to data collection and the statistical analysis plan, and interpreted the

findings. YG contributed to the interpretation of the findings. AS contributed to the analysis

strategy, the statistical analysis plan and interpretation of results. SR contributed to analysis

strategy, the statistical analysis plan, data collection and the interpretation of results.

Funding

Medical writing assistance, supported financially by Boehringer Ingelheim, was provided by

Jennifer Edwards, MB BS, of Envision Scientific Solutions during the preparation of this article.

Page 18 of 48


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



19

Competing interests

All authors have completed the Unified Competing Interest form at

www.icmje.org/coi_disclosure.pdf (available on request from the corresponding author) and

declare: JG has received fees for Board membership from Boehringer Ingelheim, Novo Nordisk

and Eli Lilly, and has received research grants from Boehringer Ingelheim, Eli Lilly,

GlaxoSmithKline and Janssen. JR, DP and WG have received fees for participation in review

activities, and for manuscript writing and reviewing from Boehringer Ingelheim. CF, YG, BM, SP,

AS and SR are employees of Boehringer Ingelheim, the manufacturer of linagliptin.

Data sharing statement

No additional data available.

Page 19 of 48


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



20

References

1 Higgins J, Deeks J, Altman D. Special topics in statistics. In: Higgins JPT, Green

S, editors. Cochrane Handbook for Systematic Reviews of Interventions Version

5.1.0: The Cochrane Collaboration, 2011.

2 Glenny AM, Altman DG, Song F, et al. Indirect comparisons of competing

interventions. Health Technol Assess 2005;9:1–134.

3 Ahn JE, French JL. Longitudinal aggregate data model-based meta-analysis with

NONMEM: approaches to handling within treatment arm correlation. J

Pharmacokinet Pharmacodyn 2010;37:179–201.

4 Scheen AJ. Pharmacokinetics of dipeptidylpeptidase-4 inhibitors. Diabetes Obes

Metab 2010;12:648–58.

5 Scheen AJ, Charpentier G, Ostgren CJ, et al. Efficacy and safety of saxagliptin in

combination with metformin compared with sitagliptin in combination with

metformin in adult patients with type 2 diabetes mellitus. Diabetes Metab Res

Rev 2010;26:540–9.

6 Richter B, Bandeira-Echtler E, Bergerhoff K, et al. Dipeptidyl peptidase-4 (DPP-

4) inhibitors for type 2 diabetes mellitus. Cochrane Database Syst Rev

2008:CD006739.

Page 20 of 48


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



21

7 Merck & Co. Inc. US Food and Drug Administration Drug Approval Package.

Januvia (sitagliptin phosphate) tablets.

http://www.accessdata.fda.gov/drugsatfda_docs/nda/2006/021995s000TOC.cfm

8 Del Prato S, Barnett AH, Huisman H, et al. Effect of linagliptin monotherapy on

glycaemic control and markers of beta-cell function in patients with inadequately

controlled type 2 diabetes: a randomized controlled trial. Diabetes Obes Metab

2011;13:258–67.

9 Forst T, Uhlig-Laske B, Ring A, et al. Linagliptin (BI 1356), a potent and selective

DPP-4 inhibitor, is safe and efficacious in combination with metformin in patients

with inadequately controlled Type 2 diabetes. Diabet Med 2010;27:1409–19.

10 Owens DR, Swallow R, Dugi KA, et al. Efficacy and safety of linagliptin in

persons with Type 2 diabetes inadequately controlled by a combination of

metformin and sulphonylurea: a 24-week randomized study. Diabet Med

2011;28:1352–61.

11 Taskinen MR, Rosenstock J, Tamminen I, et al. Safety and efficacy of linagliptin

as add-on therapy to metformin in patients with type 2 diabetes: a randomized,

double-blind, placebo-controlled study. Diabetes Obes Metab 2011;13:65–74.

12 Gallwitz B, Uhlig-Laske B, Bhattacharaya S, et al. Linagliptin has similar efficacy

to glimepiride but improved cardiovascular safety over 2 years in patients with

T2DM inadequately controlled on metformin. Abstract 39-LB. American Diabetes

Association 71st Scientific Sessions. San Diego, CA, USA, 2011.

Page 21 of 48


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



22

13 Patel S, Barnett A, Harper R, et al. 1 yr Linagliptin monotherapy is well tolerated

& sustains improvement in glycaemic control in patients for whom metformin is

inappropriate. Abstract D-0920. 21st World Congress Diabetes. Dubai, United

Arab Emirates, 2011.

14 Rafeiro E, Ross S, Meinicke T, et al. Efficacy and safety of 5 mg daily dosing

regimens with linagliptin in patients with type 2 diabetes inadequately controlled

on metformin. 47th Annual Meeting of the European Association for the Study of

Diabetes. Lisbon, Portugal, 2011.

15 Hornick K. The R FAQ. http://CRAN.R-project.org/doc/FAQ/R-FAQ.html.

16 Gastonguay MR. A full model estimation approach for covariate effects:

Inference based on clinical importance and estimation precision. AAPS Journal

2004;6:Abstract W4354.

17 Gastonguay MR. Full covariate models as an alternative to methods relying on

statistical significance for inferences about covariate effects: a review of

methodology and 42 case studies. Abstract 2229. Annual Meeting of the

Population Approach Group in Europe. Athens, Greece, 2011.

18 Gelman A, Meng X-L, Stern H. Posterior predictive assessment of model fitness

via realized discrepancies. Statistica Sinica 1995;6:733–807.

Page 22 of 48


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



23

19 Seck T, Nauck M, Sheng D, et al. Safety and efficacy of treatment with sitagliptin

or glipizide in patients with type 2 diabetes inadequately controlled on metformin:

a 2-year study. Int J Clin Pract 2010;64:562–76.

20 Aschner P, Kipnes MS, Lunceford JK, et al. Effect of the dipeptidyl peptidase-4

inhibitor sitagliptin as monotherapy on glycemic control in patients with type 2

diabetes. Diabetes Care 2006;29:2632–7.

21 Bergenstal RM, Wysham C, Macconell L, et al. Efficacy and safety of exenatide

once weekly versus sitagliptin or pioglitazone as an adjunct to metformin for

treatment of type 2 diabetes (DURATION-2): a randomised trial. Lancet

2010;376:431–9.

22 Charbonnel B, Karasik A, Liu J, et al. Efficacy and safety of the dipeptidyl

peptidase-4 inhibitor sitagliptin added to ongoing metformin therapy in patients

with type 2 diabetes inadequately controlled with metformin alone. Diabetes Care

2006;29:2638–43.

23 Goldstein BJ, Feinglos MN, Lunceford JK, et al. Effect of initial combination

therapy with sitagliptin, a dipeptidyl peptidase-4 inhibitor, and metformin on

glycemic control in patients with type 2 diabetes. Diabetes Care 2007;30:1979–

87.

24 Hanefeld M, Herman GA, Wu M, et al. Once-daily sitagliptin, a dipeptidyl

peptidase-4 inhibitor, for the treatment of patients with type 2 diabetes. Curr Med

Res Opin 2007;23:1329–39.

Page 23 of 48


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



24

25 Hermansen K, Kipnes M, Luo E, et al. Efficacy and safety of the dipeptidyl

peptidase-4 inhibitor, sitagliptin, in patients with type 2 diabetes mellitus

inadequately controlled on glimepiride alone or on glimepiride and metformin.

Diabetes Obes Metab 2007;9:733–45.

26 Iwamoto Y, Taniguchi T, Nonaka K, et al. Dose-ranging efficacy of sitagliptin, a

dipeptidyl peptidase-4 inhibitor, in Japanese patients with type 2 diabetes

mellitus. Endocr J 2010;57:383–94.

27 Mohan V, Yang W, Son HY, et al. Efficacy and safety of sitagliptin in the

treatment of patients with type 2 diabetes in China, India, and Korea. Diabetes

Res Clin Pract 2009;83:106–16.

28 Nonaka K, Kakikawa T, Sato A, et al. Efficacy and safety of sitagliptin

monotherapy in Japanese patients with type 2 diabetes. Diabetes Res Clin Pract

2008;79:291–8.

29 Raz I, Hanefeld M, Xu L, et al. Efficacy and safety of the dipeptidyl peptidase-4

inhibitor sitagliptin as monotherapy in patients with type 2 diabetes mellitus.

Diabetologia 2006;49:2564–71.

30 Rosenstock J, Brazg R, Andryuk PJ, et al. Efficacy and safety of the dipeptidyl

peptidase-4 inhibitor sitagliptin added to ongoing pioglitazone therapy in patients

with type 2 diabetes: a 24-week, multicenter, randomized, double-blind, placebo-

controlled, parallel-group study. Clin Ther 2006;28:1556–68.

Page 24 of 48


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



25

31 Scott R, Wu M, Sanchez M, et al. Efficacy and tolerability of the dipeptidyl

peptidase-4 inhibitor sitagliptin as monotherapy over 12 weeks in patients with

type 2 diabetes. Int J Clin Pract 2007;61:171–80.

32 Scott R, Loeys T, Davies MJ, et al. Efficacy and safety of sitagliptin when added

to ongoing metformin therapy in patients with type 2 diabetes. Diabetes Obes

Metab 2008;10:959–69.

33 Boehringer Ingelheim. Boehringer Ingelheim Study 1218.05. A randomized,

double-blind, placebo-controlled, five parallel group study investigating the

efficacy and safety of BI 1356 BS (0.5 mg, 2.5 mg and 5.0 mg administered orally

once daily) over 12 weeks in drug naive and treated patients with type 2 diabetes

with insufficient glycemic control (study includes an open-label metformin

treatment arm).

http://clinicaltrials.gov/ct2/show/NCT00328172?term=1218.5&rank=1.

34 Araki E, Kawamori R, Inagaki N, et al. Long-term safety of linagliptin

monotherapy in Japanese patients with type 2 diabetes. IDF World Diabetes

Congress. Dubai, United Arab Emirates, 2011.

35 Lewin A, Arvay L, Liu D, et al. Safety and efficacy of linagliptin as add-on therapy

to a sulphonylurea in inadequately controlled type 2 diabetes. Diabetologia

2010;53(Suppl 1):S326.

36 Gomis R, Espadero RM, Jones R, et al. Efficacy and safety of initial combination

therapy with linagliptin and pioglitazone in patients with inadequately controlled

Page 25 of 48


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



26

type 2 diabetes: a randomized, double-blind, placebo-controlled study. Diabetes

Obes Metab 2011;13:653–61.

37 Bohannon N. Overview of the gliptin class (dipeptidyl peptidase-4 inhibitors) in

clinical practice. Postgrad Med 2009;121:40–5.

38 Eckhardt M, Hauel N, Himmelsbach F, et al. 3,5-Dihydro-imidazo[4,5-d]pyridazin-

4-ones: a class of potent DPP-4 inhibitors. Bioorg Med Chem Lett 2008;18:3158–

62.

39 Esposito K, Cozzolino D, Bellastella G, et al. Dipeptidyl peptidase-4 inhibitors

and HbA1c target of <7% in type 2 diabetes: meta-analysis of randomized

controlled trials. Diabetes Obes Metab 2011;13:594–603.

40 Monami M, Cremasco F, Lamanna C, et al. Predictors of response to dipeptidyl

peptidase-4 inhibitors: evidence from randomized clinical trials. Diabetes Metab

Res Rev 2011;27:362–72.

41 Monami M, Iacomelli I, Marchionni N, et al. Dipeptydil peptidase-4 inhibitors in

type 2 diabetes: a meta-analysis of randomized clinical trials. Nutr Metab

Cardiovasc Dis 2010;20:224–35.

42 Gibbs JP, Fredrickson J, Barbee T, et al. Quantitative model of the relationship

between dipeptidyl peptidase-4 (DPP-4) inhibition and response: meta-analysis

of alogliptin, saxagliptin, sitagliptin, and vildagliptin efficacy results. J Clin

Pharmacol 2011:doi 10.1177/0091270011420153.

Page 26 of 48


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



27

43 Vincent SH, Reed JR, Bergman AJ, et al. Metabolism and excretion of the

dipeptidyl peptidase 4 inhibitor [14C]sitagliptin in humans. Drug Metab Dispos

2007;35:533–8.

44 Blech S, Ludwig-Schwellinger E, Grafe-Mody EU, et al. The metabolism and

disposition of the oral dipeptidyl peptidase-4 inhibitor, linagliptin, in humans. Drug

Metab Dispos 2010;38:667–78.

45 Huttner S, Graefe-Mody EU, Withopf B, et al. Safety, tolerability,

pharmacokinetics, and pharmacodynamics of single oral doses of BI 1356, an

inhibitor of dipeptidyl peptidase 4, in healthy male volunteers. J Clin Pharmacol

2008;48:1171–8.

46 Graefe-Mody U, Friedrich C, Port A, et al. Effect of renal impairment on the

pharmacokinetics of the dipeptidyl peptidase-4 inhibitor linagliptin. Diabetes Obes

Metab 2011;13:939-46.

[[Legends]]

Table 1 Summary of design and demographics of studies in the analysis dataset

Figure 1 A A graphic representation of the components of the final model, for study arms

that included patients who were treatment-naïve or had completely washed out their prior anti-

hyperglycaemic medication.

Page 27 of 48


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



28

Figure 1 B A graphic representation of the components of the final model, for study arms

that included patients washing out their prior anti-hyperglycaemic medication.

Figure 2 Study selection: PRISMA flow diagram for linagliptin and sitagliptin searches.

*Only sitagliptin records were used for this analysis. FDA, Food and Drug Administration.

Figure 3 Difference from placebo values (percentage points) of the 21 studies with relevant

treatment arms (ie, studies with linagliptin 5 mg, or sitagliptin 100 mg and placebo arms) over

time: comparison of observed and predicted HbA1c difference from placebo. Filled dots

represent observed data, the shaded regions show the unconditional 90% prediction intervals,

and the central line represents the median prediction.

Figure 4

A Estimated drug effects on HbA1c for reference population, with no pre-treatment washout,

over 24 weeks (difference from placebo).

B Estimated drug effects on HbA1c for reference population, with 4-week washout plus 2-week

placebo run-in period, over 24 weeks (difference from placebo).

Reference population of 1000 participants, baseline HbA1c: 8.0%, racial composition: 61.5%

white, 1.5% black, 37.0% Asian.

Page 28 of 48


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



29


Study Drug Dose

(mg/day)

Treatment

duration

(weeks)

Patients

(n)

Baseline

age

(years)

Female

(%)

Baseline

HbA1c

(%)

Baseline

BMI

(kg/m2)

Washout

duration

(weeks)

Concomitant

medications

Aschner

200620

Placebo NA 24 244 54.3 48.6 8.03 30.8 14 NA

Sitagliptin 100 24 229 53.4 42.9 8.01 30.3 14 NA

200 24 238 54.9 53.2 8.08 30.3 14 NA

Bergenstal

201021

Sitagliptin 100 26 166 52.0 48.0 8.50 32.0 0 Metformin

Charbonnel

200622

Placebo NA 24 224 54.7 40.5 8.03 31.5 18 Metformin


Goldstein

200723

Placebo NA 24 165 53.3 47.2 8.68 32.5 14 NA

Sitagliptin 100 24 175 53.6 48.0 8.87 31.2 14 NA

Hanefeld

200724

Placebo NA 12 107 55.9 36.9 7.59 31.4 8 NA

Sitagliptin 25 12 107 55.1 48.6 7.71 31.9 8 NA

Page 29 of 48


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960

on April 6, 2020 by guest. Protected by copyright. http://bmjopen.bmj.com/ BMJ Open: first published as 10.1136/bmjopen-2012-001844 on 5 March 2013. Downloaded from



30

50 12 107 55.3 54.5 7.60 31.6 8 NA

50 12 108 55.2 55.9 7.79 32.7 8 NA

100 12 106 56 44.5 7.78 31.6 8 NA

Hermansen

200725

Placebo NA 24 106 55.2 45.3 8.43 30.7 16 Glimepiride

Sitagliptin 100 24 106 54.4 47.2 8.42 31.0 16 Glimepiride


+ metformin


+ metformin

Iwamoto

201026

Placebo NA 12 73 60.2 31.5 7.74 24.1 8 NA

Sitagliptin 25 12 80 59.9 36.3 7.49 25.0 8 NA

50 12 72 60.2 34.7 7.57 24.5 8 NA

100 12 70 58.3 48.6 7.56 24.2 8 NA

200 12 68 60.6 41.2 7.65 24.4 8 NA

Mohan

200927

Placebo NA 18 169 50.9 40.0 8.70 24.9 8 NA

Sitagliptin 100 18 339 50.9 43.0 8.70 25.1 8 NA

Page 30 of 48


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960




31

Nonaka

200828

Placebo NA 12 75 55.0 34.0 7.69 25.1 8 NA

Sitagliptin 10 12 75 55.6 40.0 7.54 25.2 8 NA

Raz 200629

Placebo NA 18 103 55.5 37.3 8.05 32.5 14 NA

Sitagliptin 100 18 193 54.5 46.3 8.04 31.8 14 NA

200 18 199 55.4 49.5 8.14 32.0 14 NA

Rosenstock

200630

Placebo NA 24 174 56.9 46.9 8.00 31.0 18 Pioglitazone

Sitagliptin 100 24 163 55.6 42.1 8.05 32.0 18 Pioglitazone

Scheen

20105

Saxagliptin 5 18 334 58.8 52.9 7.68 31.1


Seck

201019


Scott

200731


Sitagliptin 10 12 122 55.1 50.4 7.89 30.8 8 NA

25 12 122 56.2 52 7.85 30.5 8 NA

Page 31 of 48


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960




32

50 12 120 55.6 42.3 7.89 31.4 8 NA

100 12 121 55.1 47.6 7.96 30.4 8 NA

Scott

200832

Placebo NA 18 88 55.3 41.0 7.68 30.0 8 NA

Sitagliptin 100 18 91 55.2 45.0 7.75 30.3 0 NA

Boehringer Placebo NA 12 63 59.0 49.2 8.27 30.9 0 NA

Ingelheim Linagliptin 0.5 12 57 58.0 22.8 8.24 31.0 6 NA

Study

1218.533

2.5 12 55 60.0 52.7 8.38 31.5 6 NA

5 12 54 56.0 42.6 8.38 31.2 6 NA

Forst

20109

Placebo NA 12 70 60.0 38.6 8.37 32.2 6 NA

Linagliptin 1 12 64 59.0 43.8 8.24 32.2 6 Metformin

5 12 62 60.0 46.8 8.46 31.6 6 Metformin

10 12 66 62.0 47.0 8.35 31.7 6 Metformin

Del Prato

20118


Linagliptin 5 24 333 56.0 51.4 8.00 29.0 6 NA

Taskinen Placebo NA 24 175 57.0 42.3 8.02 30.1 6 NA

Page 32 of 48


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960




33

201111


Owens

201110



+ SU

Gallwitz

201112


+ SU

Araki

201134


Linagliptin 5 12 159 60.0 30.2 8.07 24.6 4 NA

10 12 160 61.0 30.0 7.98 25.0 4 NA

Lewin

201035

Placebo NA 18 82 56.0 39.0 8.60 28.1 4 NA

Linagliptin 5 18 158 57.0 52.5 8.61 28.3 6 SU

Patel

201113

Placebo NA 18 73 56.0 57.5 8.06 30.0 6 SU

Linagliptin 5 18 147 57.0 64.0 8.11 29.0 6 NA

Rafeiro Placebo NA 12 43 59.0 51.2 7.92 28.6 6 NA

Page 33 of 48


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960




34

201114


SU, sulphonylurea.

Page 34 of 48


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960




PRISMA 2009 ChecklistPRISMA 2009 ChecklistPRISMA 2009 ChecklistPRISMA 2009 Checklist

Section/topic # Checklist item Reported on page #

TITLE

Title 1 Identify the report as a systematic review, meta-analysis, or both. 1

ABSTRACT

Structured summary 2 Provide a structured summary including, as applicable: background; objectives; data sources; study eligibility criteria, participants, and interventions; study appraisal and synthesis methods; results; limitations; conclusions and implications of key findings; systematic review registration number.

3–6

INTRODUCTION

Rationale 3 Describe the rationale for the review in the context of what is already known. 7, 8

Objectives 4 Provide an explicit statement of questions being addressed with reference to participants, interventions, comparisons, outcomes, and study design (PICOS).

8

METHODS

Protocol and registration 5 Indicate if a review protocol exists, if and where it can be accessed (e.g., Web address), and, if available, provide registration information including registration number.

N/A

Eligibility criteria 6 Specify study characteristics (e.g., PICOS, length of follow-up) and report characteristics (e.g., years

considered, language, publication status) used as criteria for eligibility, giving rationale. 9, 10

Information sources 7 Describe all information sources (e.g., databases with dates of coverage, contact with study authors to identify additional studies) in the search and date last searched.

9, Supplementary Table 1

Search 8 Present full electronic search strategy for at least one database, including any limits used, such that it could be repeated.

9, Supplementary Table 1

Study selection 9 State the process for selecting studies (i.e., screening, eligibility, included in systematic review, and, if

applicable, included in the meta-analysis). 9, 10

Data collection process 10 Describe method of data extraction from reports (e.g., piloted forms, independently, in duplicate) and any processes for obtaining and confirming data from investigators.

9, 10, Fig 2

Data items 11 List and define all variables for which data were sought (e.g., PICOS, funding sources) and any assumptions and simplifications made.

12, 13

Risk of bias in individual studies

12 Describe methods used for assessing risk of bias of individual studies (including specification of whether this was done at the study or outcome level), and how this information is to be used in any data synthesis.

13

Summary measures 13 State the principal summary measures (e.g., risk ratio, difference in means). 11, 12

Page 35 of 48


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960





Synthesis of results 14 Describe the methods of handling data and combining results of studies, if done, including measures of consistency (e.g., I

2) for each meta-analysis.

10–12

Page 1 of 2

Section/topic # Checklist item Reported on page #

Risk of bias across studies 15 Specify any assessment of risk of bias that may affect the cumulative evidence (e.g., publication bias, selective reporting within studies).

13

Additional analyses 16 Describe methods of additional analyses (e.g., sensitivity or subgroup analyses, meta-regression), if done,

indicating which were pre-specified. N/A

RESULTS

Study selection 17 Give numbers of studies screened, assessed for eligibility, and included in the review, with reasons for exclusions at each stage, ideally with a flow diagram.

Fig 2 Supplementary Table 2

Study characteristics 18 For each study, present characteristics for which data were extracted (e.g., study size, PICOS, follow-up period) and provide the citations.

Table 1

Risk of bias within studies 19 Present data on risk of bias of each study and, if available, any outcome level assessment (see item 12). 13

Results of individual studies 20 For all outcomes considered (benefits or harms), present, for each study: (a) simple summary data for each intervention group (b) effect estimates and confidence intervals, ideally with a forest plot.

N/A

Synthesis of results 21 Present results of each meta-analysis done, including confidence intervals and measures of consistency. N/A

Risk of bias across studies 22 Present results of any assessment of risk of bias across studies (see Item 15). 13

Additional analysis 23 Give results of additional analyses, if done (e.g., sensitivity or subgroup analyses, meta-regression [see Item 16]).

N/A

DISCUSSION

Summary of evidence 24 Summarize the main findings including the strength of evidence for each main outcome; consider their relevance to key groups (e.g., healthcare providers, users, and policy makers).

14–17

Limitations 25 Discuss limitations at study and outcome level (e.g., risk of bias), and at review-level (e.g., incomplete retrieval of identified research, reporting bias).

14–15

Conclusions 26 Provide a general interpretation of the results in the context of other evidence, and implications for future research.

16–17

FUNDING

Funding 27 Describe sources of funding for the systematic review and other support (e.g., supply of data); role of funders for the systematic review.

18

Page 36 of 48


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960





From: Moher D, Liberati A, Tetzlaff J, Altman DG; The PRISMA Group. Preferred reporting items for systematic reviews and meta-analyses: The PRISMA statement. PLoS Med 2009;6(7):e1000097.

For more information, visit: www.prisma-statement.org.

Page 2 of 2

Page 37 of 48


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960




Time (days)

HbA

1c (

%)

7.5

8.0

8.5

−50 0 50 100 150

washout

HbA1cprior

HbA1c∞

HbA1cbase

HbA1cplacebo

HbA1cdrug = HbA1cplacebo(1 − Edrug)

Page 38 of 48


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



Time (days)

HbA

1c (

%)

7.5

8.0

8.5

−50 0 50 100 150

washout

HbA1cprior

HbA1c∞

HbA1cbase

HbA1cplacebo

HbA1cdrug = HbA1cplacebo(1 − Edrug)

Page 39 of 48


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



Sitagliptin records identified through search of FDA Drug Approval Package, Cochrane

Review, clinical trials.gov registry, and manual searching

(n = 48) Linagliptin records identified in sponsor’s library (n = 10)

Sc

ree

nin

g

Inc

lud

ed

E

lig

ibil

ity

Ide

nti

fic

ati

on

Sitagliptin and linagliptin* records identified in Embase searching, Cochrane library,

sponsor’s library, clinical trials.gov registry, Australian and

New Zealand Clinical Trial registry, and manual searching

(n = 1008)

Records after duplicates removed (n = 41)

Records screened (n = 45)

Records excluded (n = 992)

Full-text articles assessed for eligibility

(n = 41)

Full-text articles excluded, with reasons

(n = 16)

Studies included in qualitative synthesis

(n = 25)

Studies included in quantitative synthesis

(meta-analysis) (n = 25 )



Page 40 of 48


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from


For peer review onlyWeek

HbA

1c d

iffe

ren

ce f

rom

pla

cebo

(per

cen

tage

po

ints

)

−1.0

−0.5

0.0

●

●

●

●

Araki 201128

0 5 10152025

●●●

●

●

Aschner 20068

●●

●

●●

BI Study 1218.522

0 5 10152025

●●●

●

●

Charbonnel 200610

●●●

●

●

Del Prato 201124

●

●

●

●

Forst 201023

●

●

●

●

●

Goldstein 200711

●●

●

●

●

Hanefeld 200712

●●

●

●

●

Hermansen 200713

−1.0

−0.5

0.0

●

●

●

●

●

Iwamoto 201014

−1.0

−0.5

0.0●●

●

●

Lewin 201029

●

●

●

●

Mohan 200915

●

●

●

●

●

Nonaka 200816

●●●

●

●

Owens 201126

●●

●

●

Patel 201130

●

●

●

Raz 200617

●●

●

●

Rafeiro 201131

●●●

●

●

Rosenstock 200618

●

●

●

●

●

Scott 200720

−1.0

−0.5

0.0

●●

●

●

Scott 200821

−1.0

−0.5

0.0

0 5 10152025

●●●

●

●

Taskinen 201125

Linagliptin 5 mg Sitagliptin 100 mg● ●

Page 41 of 48


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



No washout

Time after start of treatment (weeks)


HbA

1c d

iffer

ence

from

pla

cebo

(%)

HbA

1c d

iffer

ence

from

pla

cebo

(%)

−1.0

−0.8

−0.6

−0.4

−0.2

0 5 10 15 20 25

Linagliptin 5 mg

Sitagliptin 100 mg

Washout: 6 weeks

−1.0

−0.8

−0.6

−0.4

−0.2

0.0

A

B

0 5 10 15 20 25

Shaded areas show 90% prediction intervals

Point estimates for linagliptin and sitagliptin (the two overlap)

Page 42 of 48


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



Suggested Additional Information

Table 1 Details of search strategies

Table 2 Summary of excluded references and reasons for their exclusion

Page 43 of 48


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



Supplementary Table 1

Details of search strategies

Database

(search date)

Search string Citations

identified

Embase

(8 November 2010)

1. linagliptin OR ‘ondero‘/exp OR ondero OR ‘bi-1356‘/exp

OR ‘bi- 1356‘ OR ‘bi1356‘/exp OR bi1356 OR ‘bi

1356‘/exp OR ‘bi 1356‘

72

2. ‘sitagliptin‘/exp OR sitagliptin OR ‘januvia‘/exp OR

januvia OR ‘sitagliptin‘/exp OR sitagliptine OR ‘mk

0431‘/exp OR ‘mk 0431‘ OR ‘km0431‘/exp OR mk0431

OR ‘mk431‘/exp OR mk431 OR ‘mk 431‘/exp OR ‘mk

431‘

1545

3. Search 1 OR 2 1582

4. ‘diabetes‘/exp OR diabetes OR ‘diabetic‘/exp OR diabetic 526 269

5. Search 3 AND 4 1450

6. ‘comparative study‘/exp OR ‘comparative study‘ OR

‘clinical trial‘/exp OR ‘clinical trial‘ OR ‘randomised

controlled trial‘/exp OR ‘randomisation‘/exp OR ‘single

blind procedure‘/exp OR ‘single blind procedure‘ OR

‘double blind procedure‘/exp OR ‘double blind procedure‘

OR ‘triple blind procedure‘ OR ‘crossover procedure‘/exp

OR ‘crossover procedure‘ OR ‘placebo‘/exp OR ‘placebo‘

OR ‘random‘ OR rct OR ‘single blind‘ OR ‘single blinded‘

OR ‘double blind‘ OR ‘double blinded‘ OR ‘treble blind‘

OR ‘treble blinded‘ OR ‘triple blind‘ OR ‘triple blinded‘ OR

‘prospective study‘/exp OR ‘prosepctive study‘

2 215 299

Page 44 of 48


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from




8. Search 7 AND [humans]/lim 807

Cochrane library

(8 November 2010)

(linagliptin OR ondero OR sitagliptin OR januvia) AND

diabetes

[Search All Text]

48

IDEA

(10 November 2010)

23

Linagliptin OR ondero OR sitagliptin OR januvia

[Study type: interventional studie; Conditions: diabetes;

Recruitment: closed studies]

130

Manual searching 0

Total 1008

Page 45 of 48


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from




Summary of excluded references and reasons for their exclusion

Reference Reason for exclusion

Aaboe 20101 <20 patients per arm and/or crossover design

Bragz 20072 <20 patients per arm and/or crossover design

Chan 20083 Renal insufficiency population

Herman 20064 <20 patients per arm and/or crossover design

Nauck 20075 Seck et al. 2010

6 is extension study to Nauck et al.

20075, as the Seck et al. 2010

6 article reports the

results of the full analysis dataset (in addition to

those of the per-protocol dataset) whereas in

Nauck et al. 20075 only those of the per-protocol

group are given, only data that referred to the full

analysis dataset reported in Seck et al. 20106

were used

Nonaka 20097 Only 4-week treatment duration

Prately 20108 Open-label design

Raz 20089 Phase IV study in poorly controlled subjects

Retnakararn 201010

<20 patients per arm and/or crossover design

Rigby 201011

Open-label design

Williams-Herman 200912


These extension studies included only those

patients from the previous study that had not

required rescue medication (introducing a likely

selection bias)

Merck Study Code P01514 Lack of suitable data

Page 46 of 48


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



Merck Study Code P01414

Merck Study Code RC43120114


Lack of suitable data


Renal insufficiency population

References

1 Aaboe K, Knop F, Vilsbøll T, et al. Twelve weeks treatment with the DPP-4 inhibitor, sitagliptin,

prevents degradation of peptide YY and improves glucose and non-glucose induced insulin

secretion in patients with type 2 diabetes mellitus. Diab Obes Metab 2010;12:323–33.

2 Brazg R, Xu L, Dalla Man C, et al. Effect of adding sitagliptin, a dipeptidyl peptidase-4 inhibitor, to

metformin on 24-h glycaemic control and b-cell function in patients with type 2 diabetes. Diabetes

Obes Metab 2007;9:186–193.

3 Chan JC, Scott R, Arjona Ferreira JC, et al. Safety and efficacy of sitagliptin in patients with type

2 diabetes and chronic renal insufficiency. Diabetes Obes Metab 2008;10:545–55.

4 Herman GA, Bergman A, Yi B, et al. Tolerability and pharmacokinetics of metformin and the

dipeptidyl peptidase-4 inhibitor sitagliptin when co-administered in patients with type 2 diabetes.

Curr Med Res Opin 2006;22:1939–47.

5 Nauck MA, Meininger G, Sheng D, et al. Efficacy and safety of the dipeptidyl peptidase-4

inhibitor, sitagliptin, compared with the sulfonylurea, glipizide, in patients with type 2 diabetes

inadequately controlled on metformin alone: a randomized, double-blind, non-inferiority trial.


6 Seck T, Nauck M, Sheng D, et al. Safety and efficacy of treatment with sitagliptin or glipizide in

patients with type 2 diabetes inadequately controlled on metformin: a 2-year study. Int J Clin

Pract 2010;64:562–76.

7 Nonaka K, Tsubouchi H, Okuyama K, et al. Effects of once-daily sitagliptin on 24-h glucose

control following 4 weeks of treatment in Japanese patients with type 2 diabetes mellitus. Horm

Metab Res 2009;41:232–7.

Page 47 of 48


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



8 Pratley RE, Nauck M, Bailey T, et al. Liraglutide versus sitagliptin for patients with type 2 diabetes

who did not have adequate glycaemic control with metformin: a 26-week, randomised, parallel-

group, open-label trial. Lancet 2010;375:1447–56.

9 Raz I, Chen Y, Wu M, et al. Efficacy and safety of sitagliptin added to ongoing metformin therapy

in patients with type 2 diabetes. Curr Med Res Opin 2008;24:537–50.

10 Retnakaran R, Qi Y, Opsteen C, et al. Initial short-term intensive insulin therapy as a strategy for

evaluating the preservation of beta-cell function with oral antidiabetic medications: a pilot study

with sitagliptin. Diabetes Obes Metab 2010;12:909–15.

11 Rigby SP, Handelsman Y, Lai YL, et al. Effects of colesevelam, rosiglitazone, or sitagliptin on

glycemic control and lipid profile in patients with type 2 diabetes mellitus inadequately controlled

by metformin monotherapy. Endocr Pract 2010;16:53–63.

12 Williams-Herman D, Johnson J, Teng R, et al. Efficacy and safety of initial combination therapy

with sitagliptin and metformin in patients with type 2 diabetes: a 54-week study. Curr Med Res

Opin 2009;25:569–83.

13 Williams-Herman D, Johnson J, Teng R, et al. Efficacy and safety of sitagliptin and metformin as

initial combination therapy and as monotherapy over 2 years in patients with type 2 diabetes.


14 Merck & Co. Inc. US Food and Drug Administration Drug Approval Package. Januvia (sitagliptin

phosphate) tablets.


Page 48 of 48


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from





diabetes mellitus

Journal: BMJ Open

Manuscript ID: bmjopen-2012-001844.R1


Date Submitted by the Author: 06-Dec-2012





Secondary Subject Heading: Diabetes and endocrinology



BMJ Open on A


http://bmjopen.bm

j.com/

BM


jopen-2012-001844 on 5 March 2013. D

ownloaded from



1







Alegre, RS, Brazil








diabetes mellitus

Page 1 of 95


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



2




June 2012.

Page 2 of 95


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



3

ABSTRACT



peptidase-4 inhibitors, sitagliptin and linagliptin, for treatment of type 2 diabetes mellitus (T2DM)

was used as an example.














composition reflecting the average racial distribution of the linagliptin trials, and baseline HbA1c

of 8.0%.

Page 3 of 95


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



4

Results: Longitudinal data from 11 234 patients (10 linagliptin, 15 sitagliptin trials) were


by 0.81% (placebo-adjusted) at week 24. Credible intervals for participants without washout

were –0.88 to –0.75 (linagliptin) and –0.89 to –0.73 (sitagliptin)., and for those with washout, –

0.91 to –0.76 (linagliptin) and –0.91 to –0.75 (sitagliptin), when administered for 24 weeks.



patients with T2DM.

Page 4 of 95


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



5

ARTICLE SUMMARY

Article focus






Key messages














Page 5 of 95


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



6






Page 6 of 95


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



7

INTRODUCTION









to compare the findings of individual trials.1 2 In particular, endpoint-based approaches cannot

be sensibly applied when the studies involved in the review vary substantially with respect to

treatment duration.











relevant data.3



Page 7 of 95


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



8












Objective







METHODS

Data sources




Page 8 of 95


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



9





Study selection











according to treatment arm (sitagliptin, linagliptin or placebo). We extracted data on: the first







treat (ITT) populations were included whenever possible and group means, as reported, were

used or were calculated, using the last observation carried forward (LOCF) approach. The

Page 9 of 95


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



10

analyses were conducted using the maximum licensed dose of sitagliptin (100 mg) and the

licensed dose of linagliptin (5 mg). However, when data at other dose levels were available,

they were included in the analysis, and appropriate adjustments were made via the dose–

response terms in the model.
















Model development

The statistical models that were considered represent a particular class of nonlinear mixed-

effects models in which model precision terms are scaled according to sample sizes. Sample

size adjustments are carried out in a manner that approximately estimates and adjusts for

longitudinal correlations, following an approach described elsewhere.3

Page 10 of 95


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



11









of interest (24 weeks). Bayesian prior distributions for parameters describing the magnitude and

onset of drug effects were specified separately and independently for linagliptin and sitagliptin.

Magnitudes of drug effect were parameterised as fractional reductions from baseline and were

assigned uniform prior distributions between zero and one, implying that both drugs have some

beneficial effects (a defensible assumption for marketed drugs) and that neither can reduce

HbA1c levels below zero (patently true), and assigning equal likelihood to all possibilities

between these two extremes.



Software Foundation, Boston, MA, USA). Final inferences were based on 1,000 approximately

independent draws from the posterior (after discarding burn-in samples and thinning to de-

correlate samples16). The model was adjusted for baseline HbA1c and washout status/duration.

Other covariates considered were: standard covariates including demographics, such as

ethnicity, age, BMI and gender, anti-hyperglycaemic background medication, duration of T2DM

and the fraction of patients who underwent washout of previous anti-hyperglycaemic therapy.

The OpenBUGS code is available from the authors, on request.


Page 11 of 95


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



12









period, is shown in Figures 1A and 1B.

The final model was evaluated using posterior predictive check methodology16 in order to

assess whether the observed data were consistent with the range of expectation implied by the

model. This model inherently adjusted for baseline HbA1c and washout status/duration. The

other covariates (see above), with the exception of ethnicity, showed no major impact on the

model parameters and were therefore not included in the final model. Further details of the

mathematical and statistical specifications of the final model are presented in the online

Technical Appendix.

Model summary and inference

Since mean predicted values are not directly available as model parameters, these were

estimated by taking averages of values that were simulated from the fitted model. In the same

way that variances can be appropriately scaled according to sample size during model fitting,

variances were scaled during simulation to simulate trial arms of different sizes. This included

scaling simulation variances to correspond to n=1, which we conceptualised as the simulation of

an individual patient.

Page 12 of 95


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



13



conditions and the average HbA1c level was computed at each time-point for this simulated

population. Data for each patient were simulated as if arising from an individual trial, so that the


simulation of this population average was then repeated for each of the 1000 different

parameter configurations represented in the posterior sample (the entire posterior simulation

therefore involved a total of 106 simulated patients), resulting in inferences that reflect posterior

parameter uncertainty as well as inter-trial and inter-patient variation. The reference racial

composition for this simulated population was 61.5% white, 1.5% black and 37.0% Asian,

reflecting the average enrolled distribution in linagliptin trials. The median simulated baseline



RESULTS


were excluded on the basis of the study design that did not meet our inclusion criteria (Figure 2;

supplementary Table 2). A further 10 linagliptin studies were included.




randomised trials (10 linagliptin and 15 sitagliptin) (Figure 2). The mean age at baseline of all


Page 13 of 95


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



14








Figures 3A and 3B depict the application of the statistical model to each individual study,



prediction. Both change from baseline and placebo corrected change from baseline HbA1c

percentage points are presented to demonstrate longitudinal model performance for each

therapy. Similarly, Figures 4A through 4D show the 90% credible intervals at the endpoint for

the linagliptin and sitagliptin change from baseline and placebo corrected change from baseline,

demonstrating accurate prediction of the effect, on average.

The simulations performed using the model show that both linagliptin 5 mg and


administered to patients with T2DM for 24 weeks (Figures 5A and 5B). Credible intervals for

participants without washout were –0.88 to –0.75 (linagliptin) and –0.89 to –0.73 (sitagliptin).

For those who underwent washout of previous anti-hyperglycaemic therapy, the credible

intervals were –0.91 to –0.76 (linagliptin) and –0.91 to –0.75 (sitagliptin). Figure 6 shows

simulated differences in the true effect at 24 weeks between linagliptin 5 mg and sitagliptin 100

mg with no washout, demonstrating that the model predicted difference lies almost entirely

within 0.2 percentage points, less than previously suggested margins for non-inferiority of 0.3–

0.4 percentage points.36 37

Page 14 of 95


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



15





DISCUSSION



















Page 15 of 95


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



16



























Page 16 of 95


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



17



























Page 17 of 95


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



18










Page 18 of 95


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



19

Contributors













Funding



Page 19 of 95


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



20

Competing interests










Page 20 of 95


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



21

References

1. Higgins JPT, Deeks JJ, Altman DG. Special topics in statistics. In: Cochrane

Handbook for Systematic Reviews of Interventions Version 510. Higgins JPT,

Green S, editors. The Cochrane Collaboration, 2011.

2. Glenny AM, Altman DG, Song F, Sakarovitch C, Deeks JJ, D'Amico R, et al.

Indirect comparisons of competing interventions. Health Technol Assess

2005;9(26):1-134, iii-iv.

3. Ahn JE, French JL. Longitudinal aggregate data model-based meta-analysis with


Pharmacokinet Pharmacodyn 2010;37(2):179-201.

4. Scheen AJ. Pharmacokinetics of dipeptidylpeptidase-4 inhibitors. Diabetes Obes

Metab 2010;12(8):648-58.

5. Scheen AJ, Charpentier G, Ostgren CJ, Hellqvist A, Gause-Nilsson I. Efficacy

and safety of saxagliptin in combination with metformin compared with sitagliptin

in combination with metformin in adult patients with type 2 diabetes mellitus.

Diabetes Metab Res Rev 2010;26(7):540-9.

6. Richter B, Bandeira-Echtler E, Bergerhoff K, Lerch CL. Dipeptidyl peptidase-4

(DPP-4) inhibitors for type 2 diabetes mellitus. Cochrane Database Syst Rev

2008(2):CD006739.

Page 21 of 95


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



22

7. Merck & Co. Inc. US Food and Drug Administration Drug Approval Package.

Januvia (sitagliptin phosphate) tablets. 2006.


Accessed 11 October, 2011.

8. Del Prato S, Barnett AH, Huisman H, Neubacher D, Woerle HJ, Dugi KA. Effect

of linagliptin monotherapy on glycaemic control and markers of beta-cell function

in patients with inadequately controlled type 2 diabetes: a randomized controlled

trial. Diabetes Obes Metab 2011;13(3):258-67.

9. Forst T, Uhlig-Laske B, Ring A, Graefe-Mody U, Friedrich C, Herbach K, et al.

Linagliptin (BI 1356), a potent and selective DPP-4 inhibitor, is safe and

efficacious in combination with metformin in patients with inadequately controlled

Type 2 diabetes. Diabet Med 2010;27(12):1409-19.

10. Owens DR, Swallow R, Dugi KA, Woerle HJ. Efficacy and safety of linagliptin in


metformin and sulphonylurea: a 24-week randomized study(1). Diabet Med

2011;28(11):1352-61.

11. Taskinen MR, Rosenstock J, Tamminen I, Kubiak R, Patel S, Dugi KA, et al.

Safety and efficacy of linagliptin as add-on therapy to metformin in patients with


Obes Metab 2011;13(1):65-74.

Page 22 of 95


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



23

12. Gallwitz B, Uhlig-Laske B, Bhattacharaya S, Patel S, Woerle H-J. Linagliptin has

similar efficacy to glimepiride but improved cardiovascular safety over 2 years in

patients with T2DM inadequately controlled on metformin. Abstract 39-LB. Paper

presented at: American Diabetes Association 71st Scientific Sessions; 24-28

June, 2011; San Diego, CA, USA.

13. Patel S, Barnett A, Harper R, Toorawa R, Larbig M, von Eynatten M, et al. 1 yr

Linagliptin monotherapy is well tolerated & sustains improvement in glycaemic

control in patients for whom metformin is inappropriate. Abstract D-0920. Paper

presented at: 21st World Congress Diabetes; 4-8 December, 2011; Dubai, United

Arab Emirates.

14. Rafeiro E, Ross S, Meinicke T, Toorawa R, Woerle H-J. Efficacy and safety of 5

mg daily dosing regimens with linagliptin in patients with type 2 diabetes

inadequately controlled on metformin. Paper presented at: 47th Annual Meeting

of the European Association for the Study of Diabetes; 12-16 September, 2011;

Lisbon, Portugal.

15. Hornick K. The R FAQ. 2011. http://CRAN.R-project.org/doc/FAQ/R-FAQ.html.

Accessed 23 January, 2012.

16. Gelman A, li Meng, X. and Stern, H. . Posterior predictive assessment of model

fitness via realized discrepancies. Statistica Sinica 1995;6:733-807.

Page 23 of 95


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



24

17. Gastonguay MR. A full model estimation approach for covariate effects:


2004;6(S1):Abstract W4354.

18. Gastonguay MR. Full covariate models as an alternative to methods relying on


methodology and 42 case studies. Abstract 2229. Paper presented at: Annual

Meeting of the Population Approach Group in Europe; 7-10 June, 2011; Athens,

Greece.

19. Seck T, Nauck M, Sheng D, Sunga S, Davies MJ, Stein PP, et al. Safety and

efficacy of treatment with sitagliptin or glipizide in patients with type 2 diabetes

inadequately controlled on metformin: a 2-year study. Int J Clin Pract

2010;64(5):562-76.

20. Aschner P, Kipnes MS, Lunceford JK, Sanchez M, Mickel C, Williams-Herman

DE. Effect of the dipeptidyl peptidase-4 inhibitor sitagliptin as monotherapy on

glycemic control in patients with type 2 diabetes. Diabetes Care

2006;29(12):2632-7.

21. Bergenstal RM, Wysham C, Macconell L, Malloy J, Walsh B, Yan P, et al.

Efficacy and safety of exenatide once weekly versus sitagliptin or pioglitazone as

an adjunct to metformin for treatment of type 2 diabetes (DURATION-2): a

randomised trial. Lancet 2010;376(9739):431-9.

Page 24 of 95


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



25

22. Charbonnel B, Karasik A, Liu J, Wu M, Meininger G. Efficacy and safety of the

dipeptidyl peptidase-4 inhibitor sitagliptin added to ongoing metformin therapy in

patients with type 2 diabetes inadequately controlled with metformin alone.

Diabetes Care 2006;29(12):2638-43.

23. Goldstein BJ, Feinglos MN, Lunceford JK, Johnson J, Williams-Herman DE.

Effect of initial combination therapy with sitagliptin, a dipeptidyl peptidase-4

inhibitor, and metformin on glycemic control in patients with type 2 diabetes.

Diabetes Care 2007;30(8):1979-87.

24. Hanefeld M, Herman GA, Wu M, Mickel C, Sanchez M, Stein PP. Once-daily

sitagliptin, a dipeptidyl peptidase-4 inhibitor, for the treatment of patients with

type 2 diabetes. Curr Med Res Opin 2007;23(6):1329-39.

25. Hermansen K, Kipnes M, Luo E, Fanurik D, Khatami H, Stein P. Efficacy and

safety of the dipeptidyl peptidase-4 inhibitor, sitagliptin, in patients with type 2

diabetes mellitus inadequately controlled on glimepiride alone or on glimepiride

and metformin. Diabetes Obes Metab 2007;9(5):733-45.

26. Iwamoto Y, Taniguchi T, Nonaka K, Okamoto T, Okuyama K, Arjona Ferreira JC,

et al. Dose-ranging efficacy of sitagliptin, a dipeptidyl peptidase-4 inhibitor, in

Japanese patients with type 2 diabetes mellitus. Endocr J 2010;57(5):383-94.

27. Mohan V, Yang W, Son HY, Xu L, Noble L, Langdon RB, et al. Efficacy and

safety of sitagliptin in the treatment of patients with type 2 diabetes in China,

India, and Korea. Diabetes Res Clin Pract 2009;83(1):106-16.

Page 25 of 95


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



26

28. Nonaka K, Kakikawa T, Sato A, Okuyama K, Fujimoto G, Kato N, et al. Efficacy

and safety of sitagliptin monotherapy in Japanese patients with type 2 diabetes.

Diabetes Res Clin Pract 2008;79(2):291-8.

29. Raz I, Hanefeld M, Xu L, Caria C, Williams-Herman D, Khatami H. Efficacy and

safety of the dipeptidyl peptidase-4 inhibitor sitagliptin as monotherapy in patients

with type 2 diabetes mellitus. Diabetologia 2006;49(11):2564-71.

30. Rosenstock J, Brazg R, Andryuk PJ, Lu K, Stein P. Efficacy and safety of the

dipeptidyl peptidase-4 inhibitor sitagliptin added to ongoing pioglitazone therapy

in patients with type 2 diabetes: a 24-week, multicenter, randomized, double-

blind, placebo-controlled, parallel-group study. Clin Ther 2006;28(10):1556-68.

31. Scott R, Wu M, Sanchez M, Stein P. Efficacy and tolerability of the dipeptidyl


type 2 diabetes. Int J Clin Pract 2007;61(1):171-80.

32. Scott R, Loeys T, Davies MJ, Engel SS. Efficacy and safety of sitagliptin when

added to ongoing metformin therapy in patients with type 2 diabetes. Diabetes

Obes Metab 2008;10(10):959-69.

33. Boehringer Ingelheim. Boehringer Ingelheim Study 1218.05. A Randomized,

Double-blind, Placebo-controlled, Five Parallel Group Study Investigating the

Efficacy and Safety of BI 1356 BS (0.5 mg, 2.5 mg and 5.0 mg Administered

Orally Once Daily) Over 12 Weeks in Drug Naive and Treated Patients With Type

2 Diabetes With Insufficient Glycemic Control (Study Includes an Open-label

Page 26 of 95


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



27

Metformin Treatment Arm). 2011.

http://clinicaltrials.gov/ct2/show/NCT00328172?term=1218.5&rank=1. Accessed 11

January, 2012.

34. Araki E, Kawamori R, Inagaki N, Watada H, Hayashi N, Horie Y, et al. Long-term

safety of linagliptin monotherapy in Japanese patients with type 2 diabetes.

Paper presented at: IDF World Diabetes Congress; 4-8 December, 2011; Dubai,

United Arab Emirates.

35. Lewin AJ AL, Liu D, Patel S, Woerle H.-J. Safety and efficacy of linagliptin as

add-on therapy to a sulphonylurea in inadequately controlled type 2 diabetes. .

Diabetologia 2010;53 (Suppl 1):S326.

36. United States Deprtment of Health and Human Services, Food and Drug

Administration, Center for Drug Evaluation and Research. Guidance for Industry.

Diabetes Mellitus: Developing Drugs and Therapeutic Biologics for Treatment

and Prevention. 2008. www.fda.gov/downloads/Drugs/.../Guidances/ucm071624.pdf.

Accessed 29 November 2012.

37. European Medicines Agency. Guideline on clinical investigation of medicinal

products in the treatment or prevention of diabetes mellitus. 2012.

www.ema.europa.eu/docs/en_GB/...guideline/.../WC500129256.pdf. Accessed 29

November 2012.

38. Gomis R, Espadero RM, Jones R, Woerle HJ, Dugi KA. Efficacy and safety of

initial combination therapy with linagliptin and pioglitazone in patients with

Page 27 of 95


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



28

inadequately controlled type 2 diabetes: a randomized, double-blind, placebo-

controlled study. Diabetes Obes Metab 2011;13(7):653-61.

39. Bohannon N. Overview of the gliptin class (dipeptidyl peptidase-4 inhibitors) in

clinical practice. Postgrad Med 2009;121(1):40-5.

40. Eckhardt M, Hauel N, Himmelsbach F, Langkopf E, Nar H, Mark M, et al. 3,5-

Dihydro-imidazo[4,5-d]pyridazin-4-ones: a class of potent DPP-4 inhibitors.

Bioorg Med Chem Lett 2008;18(11):3158-62.

41. Esposito K, Cozzolino D, Bellastella G, Maiorino MI, Chiodini P, Ceriello A, et al.

Dipeptidyl peptidase-4 inhibitors and HbA1c target of <7% in type 2 diabetes:

meta-analysis of randomized controlled trials. Diabetes Obes Metab

2011;13(7):594-603.

42. Monami M, Cremasco F, Lamanna C, Marchionni N, Mannucci E. Predictors of

response to dipeptidyl peptidase-4 inhibitors: evidence from randomized clinical

trials. Diabetes Metab Res Rev 2011;27(4):362-72.

43. Monami M, Iacomelli I, Marchionni N, Mannucci E. Dipeptydil peptidase-4

inhibitors in type 2 diabetes: a meta-analysis of randomized clinical trials. Nutr

Metab Cardiovasc Dis 2010;20(4):224-35.

44. Gibbs JP, Fredrickson J, Barbee T, Correa I, Smith B, Lin SL, et al. Quantitative

model of the relationship between dDipeptidyl peptidase-4 (DPP-4) inhibition and

Page 28 of 95


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



29

response: Meta-analysis of alogliptin, saxagliptin, sitagliptin, and vildagliptin

efficacy results. J Clin Pharmacol 2011:doi 10.1177/0091270011420153.

45. Vincent SH, Reed JR, Bergman AJ, Elmore CS, Zhu B, Xu S, et al. Metabolism

and excretion of the dipeptidyl peptidase 4 inhibitor [14C]sitagliptin in humans.

Drug Metab Dispos 2007;35(4):533-8.

46. Blech S, Ludwig-Schwellinger E, Grafe-Mody EU, Withopf B, Wagner K. The

metabolism and disposition of the oral dipeptidyl peptidase-4 inhibitor, linagliptin,

in humans. Drug Metab Dispos 2010;38(4):667-78.

47. Huttner S, Graefe-Mody EU, Withopf B, Ring A, Dugi KA. Safety, tolerability,



2008;48(10):1171-8.

48. Graefe-Mody U, Friedrich C, Port A, Ring A, Retlich S, Heise T, et al. Effect of

renal impairment on the pharmacokinetics of the dipeptidyl peptidase-4 inhibitor

linagliptin(*). Diabetes Obes Metab 2011;13(10):939-46.

[[Legends]]


Page 29 of 95


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



30

Figure 1 A Graphic representation of the components of the final model, for study arms that

included patients washing out their prior anti-hyperglycaemic medication in the run-in period.

Figure 1 B Graphic representation of the components of the final model, for study arms that

included patients who were treatment-naïve or had completely washed out their prior anti-

hyperglycaemic medication before enrolment.



Figure 3 Drug effects (as HbA1c percentage points) of the 21 studies with relevant treatment

arms (ie, studies with linagliptin 5 mg, or sitagliptin 100 mg and placebo arms) over time: (A)

comparison of observed and predicted HbA1c change from baseline and (B) difference from

placebo.

For visual clarity, Hermansen 2007 is represented only for the arms that excluded metformin

background; both sets of arms are shown in Figure 4.

Filled dots represent observed data, the shaded regions show the unconditional 90% prediction

intervals, and the central line represents the median prediction.

Figure 4 Drug effects (as HbA1c percentage points) of the relevant studies at their respective

endpoints. Filled dots represent observed data, horizontal lines show the 90% unconditional

prediction intervals, and the horizontal lines represent the median predicted value.

Figure 4A Linagliptin change from baseline.

Figure 4B Sitagliptin change from baseline.

Page 30 of 95


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



31

Figure 4C Linagliptin difference from placebo.

Figure 4D Sitagliptin difference from placebo.

Figure 5







Figure 6 Posterior distribution for the difference in effect estimates between linaglitpin (5mg)

and sitagliptin (100mg) at 24 weeks. Reference population of 1000 participants (therefore

involving 106 simulated patients), baseline HbA1c: 8.0%, racial composition: 61.5% white, 1.5%

black, 37.0% Asian.

Page 31 of 95


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



32


Study Drug Dose

(mg/day)

Treatment

duration

(weeks)

Patients

(n)

Baseline

age

(years)

Female

(%)

Baseline

HbA1c

(%)

Baseline

BMI

(kg/m2)

Washout

duration

(weeks)

Concomitant

medications

Aschner

200620

Placebo NA 24 244 54.3 48.6 8.03 30.8 14 NA

Sitagliptin 100 24 229 53.4 42.9 8.01 30.3 14 NA

200 24 238 54.9 53.2 8.08 30.3 14 NA

Bergenstal

201021


Charbonnel

200622



Goldstein

200723

Placebo NA 24 165 53.3 47.2 8.68 32.5 14 NA

Sitagliptin 100 24 175 53.6 48.0 8.87 31.2 14 NA

Hanefeld

200724

Placebo NA 12 107 55.9 36.9 7.59 31.4 8 NA

Page 32 of 95


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960




33

Sitagliptin 25 12 107 55.1 48.6 7.71 31.9 8 NA

50 12 107 55.3 54.5 7.60 31.6 8 NA

50 12 108 55.2 55.9 7.79 32.7 8 NA

100 12 106 56 44.5 7.78 31.6 8 NA

Hermansen

200725




+ metformin


+ metformin

Iwamoto

201026

Placebo NA 12 73 60.2 31.5 7.74 24.1 8 NA

Sitagliptin 25 12 80 59.9 36.3 7.49 25.0 8 NA

50 12 72 60.2 34.7 7.57 24.5 8 NA

100 12 70 58.3 48.6 7.56 24.2 8 NA

200 12 68 60.6 41.2 7.65 24.4 8 NA

Mohan

200927

Placebo NA 18 169 50.9 40.0 8.70 24.9 8 NA

Sitagliptin 100 18 339 50.9 43.0 8.70 25.1 8 NA

Page 33 of 95


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960




34

Nonaka

200828

Placebo NA 12 75 55.0 34.0 7.69 25.1 8 NA

Sitagliptin 10 12 75 55.6 40.0 7.54 25.2 8 NA

Raz 200629 Placebo NA 18 103 55.5 37.3 8.05 32.5 14 NA

Sitagliptin 100 18 193 54.5 46.3 8.04 31.8 14 NA

200 18 199 55.4 49.5 8.14 32.0 14 NA

Rosenstock

200630



Scheen

20105

Saxagliptin 5 18 334 58.8 52.9 7.68 31.1


Seck

201019


Scott

200731


Sitagliptin 10 12 122 55.1 50.4 7.89 30.8 8 NA

25 12 122 56.2 52 7.85 30.5 8 NA

50 12 120 55.6 42.3 7.89 31.4 8 NA

100 12 121 55.1 47.6 7.96 30.4 8 NA

Page 34 of 95


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960




35

Scott

200832

Placebo NA 18 88 55.3 41.0 7.68 30.0 8 NA

Sitagliptin 100 18 91 55.2 45.0 7.75 30.3 0 NA



Study

1218.533

2.5 12 55 60.0 52.7 8.38 31.5 6 NA

5 12 54 56.0 42.6 8.38 31.2 6 NA

Forst

20109

Placebo NA 12 70 60.0 38.6 8.37 32.2 6 NA


5 12 62 60.0 46.8 8.46 31.6 6 Metformin

10 12 66 62.0 47.0 8.35 31.7 6 Metformin

Del Prato

20118


Linagliptin 5 24 333 56.0 51.4 8.00 29.0 6 NA

Taskinen

201111

Placebo NA 24 175 57.0 42.3 8.02 30.1 6 NA


Owens Placebo NA 24 262 58.0 53.2 8.14 28.2 6 Metformin

Page 35 of 95


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960




36

201110


+ SU

Gallwitz

201112


+ SU

Araki

201134


Linagliptin 5 12 159 60.0 30.2 8.07 24.6 4 NA

10 12 160 61.0 30.0 7.98 25.0 4 NA

Lewin

201035

Placebo NA 18 82 56.0 39.0 8.60 28.1 4 NA

Linagliptin 5 18 158 57.0 52.5 8.61 28.3 6 SU

Patel

201113

Placebo NA 18 73 56.0 57.5 8.06 30.0 6 SU

Linagliptin 5 18 147 57.0 64.0 8.11 29.0 6 NA

Rafeiro

201114

Placebo NA 12 43 59.0 51.2 7.92 28.6 6 NA


SU, sulphonylurea.

Page 36 of 95


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960




1







Alegre, RS, Brazil








diabetes mellitus

Page 37 of 95


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



2




June 2012.

Page 38 of 95


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



3

ABSTRACT



peptidase-4 inhibitors, sitagliptin and linagliptin, for the treatment of type 2 diabetes mellitus

(T2DM) was used as an example.














composition reflecting the average racial distribution of the linagliptin trials, and a baseline

HbA1c of 8.0%.

Page 39 of 95


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



4

Results: Longitudinal data from 11 234 patients (10 linagliptin, and 15 sitagliptin trials) were


by 0.81% (placebo-adjusted) at week 24. Credible intervals for participants without washout

were –0.88 to –0.754 (both groups)(linagliptin) and –0.89 to –0.73 (sitagliptin)., and for those

with washout, were –0.91 to –0.765 (linagliptin) and –0.910 to –0.75 (sitagliptin), when

administered for 24 weeks.



patients with T2DM.

Page 40 of 95


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



5

ARTICLE SUMMARY

Article focus






Key messages














Page 41 of 95


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



6






Page 42 of 95


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



7

INTRODUCTION











treatment duration.











relevant data.3



Page 43 of 95


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



8












Objective







METHODS

Data sources




Page 44 of 95


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



9





Study selection











according to treatment arm (sitagliptin, linagliptin or placebo) by. We extracted data on: the first









Page 45 of 95


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



10




















Model development





Page 46 of 95


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



11
















Inter-arm random effects were incorporated in order to reflect correlations amongst

observations at different time points, for both linagliptin and sitagliptin as well as for placebo.

Study-level random effects were also incorporated in recognition of potential inter-study

heterogeneity. Sample size-based adjustments for model precision terms were incorporated, in

a manner that accounts for longitudinal corrections, following an approach described

elsewhere.3






Formatted: Indent: First line: 0.5"

Page 47 of 95


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



12




The OpenWinBUGS code is available from the authors, on request.










period, is shown in figures Figures 1A and 1B.

The final model was evaluated, using posterior predictive check methodology, as reported

elsewhere,16 and demonstrated that it adequately described the datain order to assess whether

the observed data were consistent with the range of expectation implied by the model. This

model inherently adjusted for baseline HbA1c and washout status/duration. The other

covariates (see above), with the exception of ethnicity, showed no major impact on the model

parameters and were therefore not included in the final model. Further details of the


Technical Appendix.


Page 48 of 95


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



13







Application of the model: comparison of linagliptin and sitagliptin






population size of 1000 was chosen to reduce sampling error to negligible levelsThe simulation

of this population average was then repeated for each of the 1000 different parameter

configurations represented in the posterior sample (the entire posterior simulation therefore

involved a total of 106 simulated patients), resulting in inferences that reflect posterior parameter

uncertainty as well as inter-trial and inter-patient variation. The reference racial composition for

this simulated population was 61.5% white, 1.5% black and 37.0% Asian, reflecting the average

enrolled distribution in linagliptin trials. The median simulated average baseline HbA1c (%) in

this population was 8.01. Results are expressed as mean differences, with 95% credible

intervals (the Bayesian equivalent of confidence intervals).

RESULTS

Formatted: Font: Not Italic

Page 49 of 95


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



14


were excluded on the basis of the study design that did not meet our inclusion criteria (Ffigure 2;

supplementary Ttable 2). A further 10 linagliptin studies were included.




randomised trials (10 linagliptin and 15 sitagliptin) (Ffigure 2). The mean age at baseline of all

















Page 50 of 95


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



15



administered to patients with T2DM for 24 weeks (Ffigures 54A and 54B). Credible intervals for

participants without washout were –0.88 to –0.754 (linagliptinboth treatment groups) and –0.89

to –0.73 (sitagliptin)., and fFor those who underwent washout of previous anti-hyperglycaemic

therapy, the credible intervals were –0.91 to –0.765 (linagliptin) and –0.910 to –0.75 (sitagliptin).

Figure 6 shows simulated differences in the true effect at 24 weeks between linagliptin 5 mg and

sitagliptin 100 mg with no washout, demonstrating that the model predicted difference lies

almost entirely within 0.2 percentage points, less than previouslythe suggested margins for non-

inferiority margin of 0.3–-0.4 percentage points.36 37





DISCUSSION









Page 51 of 95


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



16


























Page 52 of 95


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



17



























Page 53 of 95


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



18





















Page 54 of 95


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



19

Contributors













Funding



Page 55 of 95


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



20

Competing interests










Page 56 of 95


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



21

References






2005;9(26):1-134, iii-iv.





Metab 2010;12(8):648-58.







2008(2):CD006739.

Page 57 of 95


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



22
















2011;28(11):1352-61.




Obes Metab 2011;13(1):65-74.

Formatted: Default Paragraph Font, Font:

(Default) Times New Roman, 12 pt, English

(U.S.)

Page 58 of 95


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



23










Arab Emirates.





Lisbon, Portugal.





Formatted: Default Paragraph Font, Font:

(Default) Times New Roman, 12 pt, English(U.S.)

Page 59 of 95


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



24



2004;6(S1):Abstract W4354.





Greece.




2010;64(5):562-76.




2006;29(12):2632-7.





Page 60 of 95


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



25




Diabetes Care 2006;29(12):2638-43.




Diabetes Care 2007;30(8):1979-87.














Page 61 of 95


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



26
















Obes Metab 2008;10(10):959-69.






Page 62 of 95


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



27


http://clinicaltrials.gov/ct2/show/NCT00328172?term=1218.5&rank=1. Accessed 11

January, 2012.











and Prevention. 2008. www.fda.gov/downloads/Drugs/.../Guidances/ucm071624.pdf.

Accessed 29 November 2012.



www.ema.europa.eu/docs/en_GB/...guideline/.../WC500129256.pdf. Accessed 29

November 2012.



Formatted: Default Paragraph Font, Font:(Default) Times New Roman, 12 pt, English

(U.S.)


(U.S.)


(U.S.)

Page 63 of 95


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



28











2011;13(7):594-603.









Page 64 of 95


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



29












2008;48(10):1171-8.




[[Legends]]


Page 65 of 95


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



30

Figure 1 A GA graphic representation of the components of the final model, for study arms

that included patients who were treatment-naïve or had completely washinged out their prior

anti-hyperglycaemic medication in the run-in period.

Figure 1 B GA graphic representation of the components of the final model, for study arms

that included patients who were treatment-naïve or had completely washeding out their prior

anti-hyperglycaemic medication before enrolment.



Figure 3 Difference from placebo valuesDrug effects (as HbA1c percentage points) of the 21

studies with relevant treatment arms (ie, studies with linagliptin 5 mg, or sitagliptin 100 mg and

placebo arms) over time: (A) comparison of observed and predicted HbA1c change from

baseline and (B) difference from placebo.









Page 66 of 95


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



31




Figure 54










black, 37.0% Asian.

Formatted: Superscript

Page 67 of 95


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



32


Study Drug Dose

(mg/day)

Treatment

duration

(weeks)

Patients

(n)

Baseline

age

(years)

Female

(%)

Baseline

HbA1c

(%)

Baseline

BMI

(kg/m2)

Washout

duration

(weeks)

Concomitant

medications

Aschner

200620

Placebo NA 24 244 54.3 48.6 8.03 30.8 14 NA

Sitagliptin 100 24 229 53.4 42.9 8.01 30.3 14 NA

200 24 238 54.9 53.2 8.08 30.3 14 NA

Bergenstal

201021


Charbonnel

200622



Goldstein

200723

Placebo NA 24 165 53.3 47.2 8.68 32.5 14 NA

Sitagliptin 100 24 175 53.6 48.0 8.87 31.2 14 NA

Hanefeld

200724

Placebo NA 12 107 55.9 36.9 7.59 31.4 8 NA

Formatted Table

Page 68 of 95


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960




33

Sitagliptin 25 12 107 55.1 48.6 7.71 31.9 8 NA

50 12 107 55.3 54.5 7.60 31.6 8 NA

50 12 108 55.2 55.9 7.79 32.7 8 NA

100 12 106 56 44.5 7.78 31.6 8 NA

Hermansen

200725




+ metformin


+ metformin

Iwamoto

201026

Placebo NA 12 73 60.2 31.5 7.74 24.1 8 NA

Sitagliptin 25 12 80 59.9 36.3 7.49 25.0 8 NA

50 12 72 60.2 34.7 7.57 24.5 8 NA

100 12 70 58.3 48.6 7.56 24.2 8 NA

200 12 68 60.6 41.2 7.65 24.4 8 NA

Mohan

200927

Placebo NA 18 169 50.9 40.0 8.70 24.9 8 NA

Sitagliptin 100 18 339 50.9 43.0 8.70 25.1 8 NA

Formatted Table

Formatted Table

Formatted Table

Page 69 of 95


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960




34

Nonaka

200828

Placebo NA 12 75 55.0 34.0 7.69 25.1 8 NA

Sitagliptin 10 12 75 55.6 40.0 7.54 25.2 8 NA

Raz 200629 Placebo NA 18 103 55.5 37.3 8.05 32.5 14 NA

Sitagliptin 100 18 193 54.5 46.3 8.04 31.8 14 NA

200 18 199 55.4 49.5 8.14 32.0 14 NA

Rosenstock

200630



Scheen

20105

Saxagliptin 5 18 334 58.8 52.9 7.68 31.1


Seck

201019


Scott

200731


Sitagliptin 10 12 122 55.1 50.4 7.89 30.8 8 NA

25 12 122 56.2 52 7.85 30.5 8 NA

50 12 120 55.6 42.3 7.89 31.4 8 NA

Formatted Table

Formatted Table

Page 70 of 95


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960




35

100 12 121 55.1 47.6 7.96 30.4 8 NA

Scott

200832

Placebo NA 18 88 55.3 41.0 7.68 30.0 8 NA

Sitagliptin 100 18 91 55.2 45.0 7.75 30.3 0 NA



Study

1218.533

2.5 12 55 60.0 52.7 8.38 31.5 6 NA

5 12 54 56.0 42.6 8.38 31.2 6 NA

Forst

20109

Placebo NA 12 70 60.0 38.6 8.37 32.2 6 NA


5 12 62 60.0 46.8 8.46 31.6 6 Metformin

10 12 66 62.0 47.0 8.35 31.7 6 Metformin

Del Prato

20118


Linagliptin 5 24 333 56.0 51.4 8.00 29.0 6 NA

Taskinen

201111

Placebo NA 24 175 57.0 42.3 8.02 30.1 6 NA


Formatted Table

Formatted Table

Page 71 of 95


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960




36

Owens

201110



+ SU

Gallwitz

201112


+ SU

Araki

201134


Linagliptin 5 12 159 60.0 30.2 8.07 24.6 4 NA

10 12 160 61.0 30.0 7.98 25.0 4 NA

Lewin

201035

Placebo NA 18 82 56.0 39.0 8.60 28.1 4 NA

Linagliptin 5 18 158 57.0 52.5 8.61 28.3 6 SU

Patel

201113

Placebo NA 18 73 56.0 57.5 8.06 30.0 6 SU

Linagliptin 5 18 147 57.0 64.0 8.11 29.0 6 NA

Rafeiro

201114

Placebo NA 12 43 59.0 51.2 7.92 28.6 6 NA


SU, sulphonylurea.

Formatted Table

Page 72 of 95


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960




37

Page 73 of 95


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



For peer review onlyTime (days)

HbA

1c (%

)

7.5

8.0

8.5

−50 0 50 100 150

run-in

HbA1cprior

HbA1c∞

HbA1c0

HbA1cplacebo

HbA1cdrug = HbA1cplacebo(1−Edrug)

Page 74 of 95


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from


For peer review onlyTime (days)

HbA

1c (%

)

7.5

8.0

8.5

−50 0 50 100 150

run-in

HbA1cprior

HbA1cbase

HbA1c∞HbA1cplacebo

HbA1cdrug = HbA1cplacebo(1−Edrug)

Page 75 of 95


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from






Scre

enin

g In

clud

ed

Elig

ibili

ty

Iden

tific

atio

n Sitagliptin and linagliptin* records identified in Embase searching, Cochrane library,



(n = 1008)





(n = 41)


(n = 16)


(n = 25)





Page 76 of 95


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from


For peer review onlyHbA

1c c

hang

e fro

m b

asel

ine

(per

cent

age

poin

ts)

Linagliptin 5 mg Sitagliptin 100 mg Placebo● ● ●●

Week

−1.0−0.5

0.00.5

●

●

● ●

●●

●● ●

Araki 201134

0 5 10 15 20 25

Aschner 200620

●

●

●●

● ●●

Bergenstal 201021

0 5 10 15 20 25

BI Study 1218.533

●

●

● ● ●

●●● ● ● ●

Charbonnel 200622

Del Prato 20118

●

●● ●

●●● ●●

●

Forst 20109

●

●● ● ●

Gallwitz 201112

●

●

● ● ●

●●

● ● ● ●

Goldstein 200723

−1.0−0.50.00.5

●

●● ● ●

●●● ● ● ●

Hanefeld 200724

−1.0−0.5

0.00.5

●●

● ●

● ●●

●

●

Hermansen 200725

●

●

●

●●

●●●● ● ●

●

Iwamoto 201026

●

●

● ●

●●●

● ●

Lewin 201035

●

●

● ●

●●

● ●●

Mohan 200927

●●

●

●●

●●●

●● ●

●●

Nonaka 200828

●

●

● ●●

●●

● ● ● ●

Owens 201110

●

●● ●

●●

●●

●

Patel 201113

●

●

●

●●

● ●

Rafeiro 201114

●

●● ●

●●● ● ●

Raz 200629

−1.0−0.50.00.5

●

●

● ● ●

●●● ● ● ●

Rosenstock 200630

−1.0−0.5

0.00.5

0 5 10 15 20 25

●

●

● ● ●

Scheen 20105

●● ● ●

●

●●

●●

● ●

Scott 200731

0 5 10 15 20 25

●

● ● ●

●●

●

● ●

Scott 200832

●

●

● ● ●

●● ● ●● ●

Taskinen 201111

●

●● ● ●

●●● ● ● ●

● ●

●● ●

●● ●●

●

● ●● ● ●

●

●● ● ●

●●● ● ● ●

Page 77 of 95


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from


For peer review onlyWeek

HbA

1c d

iffer

ence

from

pla

cebo

(per

cent

age

poin

ts)

−1.0

−0.5

0.0

0 5 10 15 20 25 0 5 10 15 20 25

−1.0

−0.5

0.0

−1.0

−0.5

0.0

−1.0

−0.5

0.0

−1.0

−0.5

0.0

0 5 10 15 20 25

Linagliptin 5 mg Sitagliptin 100 mg●● ●

●●

●

●

●●●

●

●

●●

●

●●

●●●

●

●

●●●

●

●

●●

●

●

●

Goldstein 200723

●●●

●

●

Hanefeld 200724

●●●

●

●

Hermansen 200725

●

●

●

●

●

Iwamoto 201026

●

●

●

●

Mohan 200927

●

●

●

●

●

Nonaka 200828

●●

●

●

Patel 201113

●●

●

Raz 200629

●●●

●

Rafeiro 201114

●●●

●

●

Rosenstock 200630

●●

●●

●

Scott 200731

●●●

●

●

Del Prato 20118

Charbonnel 200622

BI Study 1218.533

Aschner 200620

Araki 201134

Taskinen 201111

●●

●

●

Forst 20109

●●●

●

Lewin 201035

●●●

●

●

Owens 201110

●●

●

●

Scott 200832

Page 78 of 95


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



Linagliptin studies

HbA1c change from baseline (%)

Stu

dy

Araki 201134

BI Study 1218.533

Del Prato 20118

Forst 20109

Gallwitz 201112

Lewin 201035

Owens 201110

Patel 201113

Rafeiro 201114

Taskinen 201111

Araki 201134

BI Study 1218.533

Del Prato 20118

Forst 20109

Lewin 201035

Owens 201110

Patel 201113

Rafeiro 201114

Taskinen 201111

−2 −1 0 1

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●Placebo

Linagliptin, 5 mg

Sitagliptin studies

HbA1c change from baseline (%)S

tudy

Aschner 200620

Charbonnel 200622

Goldstein 200723

Hanefeld 200724

Hermansen 200725

Iwamoto 201026

Mohan 200927

Nonaka 200828

Raz 200629

Rosenstock 200630

Scott 200731

Scott 200832

Aschner 200620

Bergenstal 201021

Charbonnel 200622

Goldstein 200723

Hanefeld 200724

Hermansen 200725

Iwamoto 201026

Mohan 200927

Nonaka 200828

Raz 200629

Rosenstock 200630

Scheen, 20105

Scott 200731

Scott 200832

Seck 201019

Placebo

Sitagliptin, 100 mg

−2 −1 0 1

●●

●●

●●●

●●

●●

●●

●●

●●

●●

●●●●

●●

●●

●●

Linagliptin studies

HbA1c difference from placebo (%)

Stu

dy

Araki 201134

BI Study 1218.533

Del Prato 20118

Forst 20109

Lewin 201035

Owens 201110

Patel 201113

Rafeiro 201114

Taskinen 201111

−2 −1 0 1

●

●

●

●

●

●

●

●

●

Sitagliptin studies

HbA1c difference from placebo (%)

Stu

dy

Aschner 200620

Charbonnel 200622

Goldstein 200723

Hanefeld 200724

Hermansen 200725

Iwamoto 201026

Mohan 200927

Nonaka 200828

Raz 200629

Rosenstock 200630

Scott 200731

Scott 200832

−2 −1 0 1

●

●

●

●

●

●

●

●

●

●

●

●

●

A. B.

C. D.

Page 79 of 95


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



A

B

No washout



HbA

1c d

iffer

ence

from

pla

cebo

(%)

−1.0

−0.8

−0.6

−0.4

−0.2

0.0

0 5 10 15 20 25

Washout: 6 weeks

HbA

1c d

iffer

ence

from

pla

cebo

(%)

−1.0

−0.8

−0.6

−0.4

−0.2

0.0

0 5 10 15 20 25

Linagliptin 5 mg

Sitagliptin 100 mg

Shaded areas show 90% prediction intervals

Point estimates for linagliptin (red) and sitagliptin (blue) (the two lines overlap)

Page 80 of 95


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



Difference in mean drug effects at 24 weeks (% HbA1c)

Per

cent

of T

otal

0

5

10

15

20

−0.3 −0.2 −0.1 0.0 0.1 0.2 0.3

Favours linagliptinFavours sitagliptin

Page 81 of 95


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



Technical Appendix to:A novel model-based meta-analysis to indirectly

estimate the comparative efficacy of twomedications: an example using DPP-4 inhibitors,sitagliptin and linagliptin, in treatment of type 2

diabetes mellitus

December 3, 2012

Jorge L Gross1, James Rogers2, Dan Polhamus2, William Gillespie2, Christian Friedrich3, Yan Gong4, BrigittaMonz4, Sanjay Patel5, Alexander Staab3, Silke Retlich3

(1) Hospital de Clínicas de Porto Alegre, Universidade Federal do Rio Grande do Sul, Porto Alegre, RS, Brazil(2) Metrum Research Group, Tariffville, Connecticut, USA(3) Boehringer Ingelheim, Biberach, Germany(4) Boehringer Ingelheim, Ingelheim, Germany(5) Boehringer Ingelheim, Bracknell, Berkshire, UK

Page 82 of 95


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



Technical Appendix

This appendix is intended primarily to provide a mathematical / statistical specification of the final modelemployed in our analysis. We additionally mention several variants of the model that were considered dur-ing model development.

1 Base model: structural form

The structural form of the model (i.e., the parametric form relating the central tendency of predictions totime, dose, and covariates) remained largely unchanged across all model-fitting iterations described in thepresent report. This structural form may conceptualized in terms of a latent (random-effect) baseline HbA1c(HbA1cbase) that is modified as a function of dose (D) and time (t ) by multiplicative terms for the effect ofwashout, placebo intervention, and drug effect (where applicable).

HbA1c(t , D) = HbA1cbase

×Iwashout

�

1+∆W∞�

1− e−kW (t+twash)��

1+∆W∞�

1− e−kW twash� (washout effect)

×�

1+∆P∞�

f P (t )��

(placebo effect)

×�

1−Emax,drugD

E D50,drug+D

�

1− e−kdrugt�

�

(drug effect)

We now consider each of the model terms in slightly more detail:

• Washout term�

1+∆W∞�


1+∆W∞�

1− e−kW twash�

Although response observations from the (pre-treatment) washout period were not modeled, the du-ration of pre-treatment washout twash is used to determine how much of the post-baseline trend canbe attributed to washout. One may better understand the washout term in terms of the following twoextreme scenarios:

– The washout term for a group with an extremely long pre-treatment washout (twash ≈∞) wouldbe essentially 1, i.e., no adjustment.

– The washout term for a group discontinuing prior medication only at the time of the first doseof the randomized treatment (twash ≈ 0) would be

�

1+∆W∞�

1− e−kW (t )��

, a term that rises ex-ponentially from 1 (no adjustment at baseline) toward a horizontal asymptote of (1+∆W∞) as tbecomes very large.

As a conceptual and notational device, we also define HbA1cprior, the inferred HbA1c level at the be-ginning of the washout period:

HbA1cprior ≡HbA1cbase

1+∆W∞�


Page 1 of 7

Page 83 of 95


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



Technical Appendix

• Placebo term�

1+∆P∞�

f P (t )��

. Both exponential and a bi-exponential variants were considered forf P :

f P (t ) = 1− e−kP t ("exponential" variant)

f P (t ) = e−kP0t − e−kP t ("bi-exponential" variant)

The placebo term is intended to characterize all aspects of the longitudinal profile that are commonto all placebo-treated patients. In principal at least, this may include more than a mere “placeboeffect”. For example, it might also partially reflect longitudinal changes that would occur even in theabsence of placebo intervention. In contrast to the washout term, the placebo term does not varysystematically as a function of washout duration (in particular, in contrast to the washout term, theplacebo “effect” does not approach a limit of zero as washout duration approaches infinity), so thatthe two effects are distinctly identifiable.

The exponential and bi-exponential variants of the placebo function have different implications withrespect to the limiting placebo effect P∞ at large time values. The bi-exponential variant, sometimesreferred to as a “Bateman” function is a typical choice in the pharmacology domain for modelingplacebo effects, and is applicable when placebo effects are expected to return to zero at some dura-tion [4]. By contrast, our “exponential” variant implies a non-zero limiting effect due to placebo. Inpractice, the predictive implications for the fitted model may be similar for both functions over a finiteduration of interest, since the estimated parameters in the Bateman function may (if supported by thedata) imply a positive first derivative over most or all of that duration (in our case, over a 24 weekduration). To the extent that data do not support any return to zero for the placebo effects over theduration of interest, both convergence diagnostics and model selection criteria will tend to favor thesimpler “exponential” placebo function. In our application the bi-exponential / Bateman functionalform appeared to be reasonably well estimated and so was therefore employed in the final model.

As a conceptual and notational device, we also define HbA1cplacebo(t ), the expected time course for anindividual randomized to placebo:

HbA1cplacebo(t ) = HbA1cbase

×Iwashout

�

1+∆W∞�


1+∆W∞�


×�

1+∆P∞�

f P (t )��

(placebo effect)

Further, we define HbA1c∞ to be the limiting value of HbA1cplacebo(t ) as t becomes very large.

• Drug effect term

(1−Edrug)≡�

1−Emax,drugD

E D50,drug+D

�

1− e−kdrugt�

�

This term varies as a function of both dose (D) and time (t ). The term equals 1 (no adjustment) whent = 0 and/or D = 0, and approaches a horizontal asymptote of (1− Emax,drug) as both D and t becomevery large. The parameter kdrug, which describes the onset of drug effects, was provisionally assumed

Page 2 of 7

Page 84 of 95


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



Technical Appendix

to take the same value for both linagliptin and sitagliptin, consistent with the expected pharmacol-ogy of the two DPP4 inhibitors (empirical support for this assumption is provided by the posteriorpredictive checks included in the primary manuscript).

The dose-response component of our model may be described as a simplified Emax model, a moregeneral expression of which is:

E =EmaxDα

E Dα50+Dα

.

This functional form, also known as the Hill Equation, is widely used in the pharmacology domain todescribe both concentration-response and dose-response relationships [5]. Originally proposed onthe basis of drug receptor theory to describe concentration-response analyses, its use in describingdose-response relationships may also be theoretically justified when linear pharmacokinetics applyand dose is therefore an appropriate proxy for concentration. Theoretical support is more tenuouswhen nonlinear pharmacokinetics apply (as is the case for linagliptin), but the model remains a rea-sonable initial default. The parameterα, sometimes referred to as the Hill coefficient or the sigmoidic-ity parameter, has been set to a value of 1 in our implementation. This simplification was initially in-troduced on a tentative basis, in consideration of the limited degree of dose-response information inthe data, and based on prior experiences of the modeling team suggesting that, empirically, the Hillcoefficient for dose-response relationship is generally close to one. This simplification does imply adose-response relationship that has a non-zero gradient at D = 0, however this implication is consis-tent with observed dose-response relationships (anecdotally) for most drugs. Final acceptance of thisaspect of the model was based on the observed predictive performance of our model at all studieddose levels.

We may now use the terms introduced above to define the conditional expectation for the mean HbA1c onthe i th occasion in the j th group and k th study:

ÚHbA1ci j k = HbA1cprior,j k

×�

1+∆W∞,j k

�

1− e−kW (t i j k+twash,j k )��

×�

1+∆P∞,j k

�

f P (t i j k )��

×

1−Emax,drugj k

D j k

E D50,drugj k+D j k

�

1− e−kdrugt i j k�

!

Note that the circumflex, or “hat”, has not been used here to indicate an estimate (as would be commonin the statistical literature), but refers rather to an expected value (as is common in the pharmacometricliterature).

2 Base model: stochastic structure

Random effects associated with washout and placebo effects were implemented using a re-parameterizationof the model. The parameters ∆W∞ and ∆P∞ represent conceptual steady state asymptotes for very large

Page 3 of 7

Page 85 of 95


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



Technical Appendix

time values (t →∞). Since it is not clear that direct estimation of such parameters is supported by the data,the model was implemented using truncated parameterizations that reformulate the effect parameter as aneffect at a time t ∗ that is richly represented in the data set. For this purpose we define ∆W ∗ and ∆P∗ as thefractional changes at reference time t ∗ = 24 weeks due to washout and placebo, respectively. That is,

∆W ∗ =�

1+∆W∞�

1− e−kW (t ∗+twash)��

1+∆W∞�

1− e−kW twash� −1

∆P∗ = ∆P∞ f P (t ∗).

The random-effect structure for the i th visit for the j th arm (or group) in the k th study may then be describedas:

Inter-study variation:

log�

1+∆P∗study,k

�

∼ N�

log�

1+Ô∆P∗�

,ψ2∆P

�

Study-level random effects are additionally considered for mean baseline HbA1c and for the washout mag-nitude∆W ∗, but convergence diagnostics suggested that this level of variation in these parameters was notidentifiable based on the available data.

All study-level random effects are assumed to be independent.

Inter-arm variation:

log�

1+∆P∗j k

�

∼ N�

log�

1+∆P∗study,k

�

,ω2∆P/n 1j k

�

log�

HbA1cbase,j k

�

∼ N�

log�

HbA1cbase-study,k

�

,ω2HbA1cbase

/n 1j k

�

All arm-level random effects are assumed to be conditionally independent, given study-level random effects.Our scaling of random effect variances according to sample size follows a recently published rationale [1],and is similar to an approach used in model-based meta-analysis of Alzheimer’s Disease progression [3].Arm-level random effects were additionally considered for the washout magnitude ∆W ∗, but convergencediagnostics suggested that this level of variation was also not identifiable based on the available data.

Modeling baseline HbA1c values using Lognormal distributions constrains these to positive values. Model-ing the ∆ parameters using shifted Lognormal distributions allows the placebo and washout factors in themodel to take positive and negative (up to negative one) values. Taken together with constraints on the drugeffects (ensured via the prior for the drug Emax values), and the multiplicative construction of the model, thepreceding constraints ensure that all HbA1c predicted values are positive.

Residual variation:

HbA1ci j k ∼N

�

ÚHbA1ci j k ,σ2

n i j k

�

.

Page 4 of 7

Page 86 of 95


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



Technical Appendix

3 Covariates

Baseline HbA1cBaseline HbA1c is implicitly included as a covariate in the model in the sense that the washout, placebo, anddrug effect terms all operate multiplicatively on baseline HbA1c. In general, larger changes from baselineare implied by larger baseline values.

WashoutThe inclusion of the washout term in the base model implies that washout status and washout duration actas covariates in the sense that they modify an individual’s predicted change from baseline.

Other covariates

Whereas the covariate effects of baseline HbA1c and washout duration are implied by the longitudinal struc-ture of the base model, the remaining candidate covariates were considered as potential modifiers of drugeffect in a manner dependent on the covariate variable’s distributional properties.

• For univariate covariates taking discrete values at the arm-level (as was the case with backgroundmedication status), the covariate effect was introduced via an exponentiated linear predictor so that,with x j k denoting the dichotomous variable level on arm j in study k , the following substitution ismade:

Emax,drugj k← Emax,drugj k

× exp¦

βdrugj kx j k

©

(This approach is equivalent, modulo re-parameterization, to simply using a different Emax,drug foreach level of the covariate.)

• For univariate covariates taking continuous values at the arm-level including age, body mass index(BMI), gender (recall that at the arm-level, gender is represented as a proportion), and duration ofdiabetes, the candidate covariate was centered at its average value and an exponentiated linear pre-dictor was again used. In this case, letting x j k be the continuous covariate, the following substitutionis made:


× exp¦

βdrugj k(x j k −x ··)

©

.

• Race, like gender, is represented at the arm-level using proportions. However, unlike gender, the racialcomposition of a each study arm must be represented with multiple proportions (specifically, propor-tion white, proportion black, proportion asian, and, in certain cases, proportion “other”), precludingany interpretable use of the continuous covariate parameterization that was used for gender. Instead,a separate E r

max,drugj kwas defined for each race r (separately for each drug). Race-specific terms in

the model then imply race-specific conditional expectationsÚHbA1cr

i j k , and the aggregate conditionalexpectation is computed using a weighted average:

ÚHbA1ci j k =∑

r∈(white, black, asian, other)

F ri j kÚHbA1c

r

i j k

where F r is the fraction of the arm identifying with race r .

In order to improve parameter estimation for race-based covariate effects, racially subsetted data wereused where possible (this was possible in all cases for linagliptin records).

Page 5 of 7

Page 87 of 95


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



Technical Appendix

As noted in the primary manuscript, the effects of covariates other than race were not sufficiently well esti-mated to justify inclusion in the model. Race was therefore the only explicit in the final model (in additionto the implicit covariates, baseline HbA1c and washout duration).

4 Priors

The probababilistic specification of the prior is provided in Table 1. With regard to the primary researchquestions of interest, the most consequential components of the prior are the distributional statements re-lating to the magnitudes of drug effects, Emax, linagliptin and E ∗max, sitagliptin. Since those parameters representasymptotic drug effects (at theoretically infinite times), we re-parameterize in terms of effects at 24 weeks,E ∗linagliptin and E ∗sitagliptin, using the same reasoning as was discussed for placebo and washout effects. Sincethese drug effect parameters represent fractional reductions from a hypothetical untreated state, it is nat-ural that they should be bounded between zero and one, implying that both drugs have some beneficialeffect (a defensible assumption for marketed drugs), neither or which may reduce HbA1c levels below zero(patently true). The use of Uniform (flat) densities between these two extremes implies that all intermedi-ate values are considered (a priori) equally likely. Perhaps most importantly, the distributions for E ∗linagliptinand E ∗sitagliptin are specified as being independent. In combination with the use of Uniform densities, theprior independence of these two parameters allows for potential findings of either substantial similarity orsubstantial difference between the two drugs, as determined by the data.

Priors for ancillary (non-drug-effect) parameters were chosen largely for analytical convenience, but wereverified to be diffuse in comparison to thier corresponding posterior distributions, suggesting that these el-ements of the prior were sufficiently non-informative. Additionally, alternative distributions were evaluatedfor a number of elements of the prior (e.g. for variance components), and did not result in any substantialdifferences in conclusions.

Page 6 of 7

Page 88 of 95


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



Technical Appendix

parameter priorkW (w−1) Unif (0.001, 7)kP0 (w−1) Unif (0.001, 7)kP (w−1) Unif (0.001, 7)kdrug (w−1) Unif (0.001, 7)ÚHbA1cbase Log Normal(0, 100)Ö∆W ∗ Unif (−1, 10)log�

1+Ô∆P∗�

Unif (−10, 5)ψ∆P Unif (0, 10)ω∆P Unif (0, 10)ωHbA1c0 Unif (0, 5)σ Unif (0, 100)b linagliptin Unif (0, 100)bsitagliptin Unif (0, 100)E ∗linagliptin Unif (0, 1)E ∗sitagliptin Unif (0, 1)Emax,linagliptin ( 1

b linagliptin+1) ∗E ∗linagliptin

Emax,sitagliptin ( 1bsitagliptin

+1) ∗E ∗sitagliptin

E D50,linagliptin10

b linagliptin

E D50,sitagliptin200

bsitagliptin

Table 1: Prior distribution for parameters in base model

References

[1] Ahn, J.E. and French, J.L. Longitudinal aggregate data model-based meta-analysis with NONMEM: ap-proaches to handling within treatment arm correlation. J Pharmacokinet Pharmacodyn 37 (2010):179–201.

[2] Oehlert, G. A Note on the Delta Method. The American Statistician 46 (1992):27–29.

[3] Gillespie, W.R., Rogers, J.A., Ito, K. and Gastonguay, M.R. Population Dose-Response Model for ADAS-cog Scores in Patients with Alzheimers Disease by Meta-Analysis of a Mixture of Summary and IndividualData. In American Conference on Pharmacometrics (Mashantucket, CT, 2009).

[4] Mould, D.R. Developing Models of Disease Progression (John Wiley & Sons, Inc., 2006), pages 547–581.URL http://dx.doi.org/10.1002/9780470087978.ch21

[5] Goutelle, S., Maurin, M., Rougier, F., Barbaut, X., Bourguignon, L., Ducher, M. and Maire, P. The Hill equa-tion: a review of its capabilities in pharmacological modelling. Fundam Clin Pharmacol 22 (2008):633–48.

Page 7 of 7

Page 89 of 95


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from

http://dx.doi.org/10.1002/9780470087978.ch21






Page 90 of 95


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from





Database

(search date)


identified

Embase

(8 November 2010)



1356‘/exp OR ‘bi 1356‘

72





431‘

1545

3. Search 1 OR 2 1582














2 215 299

Page 91 of 95


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from





Cochrane library

(8 November 2010)


diabetes

[Search All Text]

48

IDEA

(10 November 2010)

23




130

Manual searching 0

Total 1008

Page 92 of 95


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



















were used




Retnakararn 201010


Rigby 201011

Open-label design






selection bias)


Page 93 of 95


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from









References






Obes Metab 2007;9:186–193.












Pract 2010;64:562–76.



Metab Res 2009;41:232–7.

Page 94 of 95


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from
















Opin 2009;25:569–83.





phosphate) tablets.


Page 95 of 95


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from





diabetes mellitus

Journal: BMJ Open

Manuscript ID: bmjopen-2012-001844.R2


Date Submitted by the Author: 31-Jan-2013





Secondary Subject Heading: Diabetes and endocrinology, Research methods



BMJ Open on A


http://bmjopen.bm

j.com/

BM


jopen-2012-001844 on 5 March 2013. D

ownloaded from



1







Alegre, RS, Brazil








diabetes mellitus

Page 1 of 99


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



2




June 2012.

Page 2 of 99


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



3

ABSTRACT


efficacies of two drugs, using model-based meta-analysis (MBMA). Comparison of two oral

dipeptidyl peptidase-4 inhibitors, sitagliptin and linagliptin, for type 2 diabetes mellitus (T2DM)

treatment was used as an example.

Design: Systematic review with MBMA.

Data sources: MEDLINE, Embase, www.ClinicalTrials.gov, Cochrane review of DPP-4

inhibitors for T2DM, sitagliptin trials on FDA website to December 2011, and linagliptin data

from the manufacturer.

Eligibility criteria for selecting studies: Double-blind, randomised, controlled, clinical trials, ≥

12 weeks’ duration, that analysed sitagliptin or linagliptin efficacies as changes in HbA1c levels,

in adults with T2DM and HbA1c >7.0%, irrespective of background medication.




model parameters andwere not included. For the indirect comparison, a population of 1000

patients was simulated from the model with a racial composition reflecting the average racial

distribution of the linagliptin trials, and baseline HbA1c of 8.0%.

Results: The model was developed using longitudinal data from 11 234 patients (10 linagliptin,

15 sitagliptin trials), and assessed by internal evaluation techniques, demonstrating that the

Page 3 of 99


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



4

model adequately described the observations. Simulations showed both linagliptin 5 mg and

sitagliptin 100 mg reduced HbA1c by 0.81% (placebo-adjusted) at week 24. Credible intervals

for participants without washout were –0.88 to –0.75 (linagliptin) and –0.89 to –0.73 (sitagliptin),

and for those with washout, –0.91 to –0.76 (linagliptin) and –0.91 to –0.75 (sitagliptin).

Conclusions: This study demonstrates the use of longitudinal MBMA in the field of diabetes

treatment. Based on an example evaluating HbA1c reduction with linagliptin versus sitagliptin,

the model used seems a valid approach for indirect drug comparisons.

Page 4 of 99


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



5

ARTICLE SUMMARY

Article focus




Chain Monte Carlo method) to indirectly estimate the comparative efficacy of two

compounds, incorporating longitudinal dose–response data.

Key messages

• A longitudinal statistical model was developed for the indirect comparison of two

pharmaceutical compounds (oral DPP-4 inhibitors linagliptin and sitagliptin), with respect

to changes in HbA1c levels in patients with type 2 diabetes mellitus (T2DM)

• The model was evaluated by comparing model predictions to observed values











Page 5 of 99


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



6






Page 6 of 99


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



7

INTRODUCTION



comparisons among all available treatment options. Network meta-analysis (mixed treatment

comparisons) has been used to estimate relative efficacy when there are no direct comparative

data, to provide the best available evidence to facilitate decision-making by physicians and

other stake-holders, such as payers. However, these approaches have certain limitations,

including the risk of bias arising from inherent differences in the designs of the included studies,

and the difficulties of finding appropriate summary statistics to compare the findings of individual

trials.1 2 In particular, endpoint-based approaches cannot be sensibly applied when the studies

involved in the review vary substantially with respect to treatment duration.


developed and used to estimate the comparative efficacy of two medications. MBMA can be

used to provide a mechanism for integrating information from heterogeneously designed trials

and, thus, to evaluate outcomes with different drugs that have not been compared directly.3

Model-based meta-analysis is distinguished from the methodology of conventional meta-

analysis by the manner in which it incorporates longitudinal and/or dose–response data. By

modelling the response as a parametric function of time, MBMA allows the integration of

information from trials of different durations and with different sampling time points. This enables

the use of less restrictive inclusion/exclusion criteria for study selection, and more efficient use

of data from the studies that are selected, therefore resulting in a particularly comprehensive

summary of all relevant data.3




Page 7 of 99


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



8











This approach to drug comparison employs a mathematical model to describe the time-course

of HbA1c levels, and is being increasingly used to characterise longitudinal data. The general

meta-analytic methodology of Ahn and French3 has previously been used to successfully

describe longitudinal meta-data from clinical trials in Alzheimer’s disease6 7, rheumatoid

arthritis8, lipid disorders,9 glaucoma,10 and chronic obstructive pulmonary disease.11 Similar

approaches have been used to perform dose-response meta-analyses in a range of therapeutic

areas, including migraine,12 post-operative anticoagulant therapy,13 and rheumatoid arthritis.14

This analytic approach has also been used in the field of diabetes in a recent study by Gibbs et

al. 2011,15 which evaluated the relationship between DPP-4 inhibition and HbA1c reduction

using data obtained from clinical trials of four drugs in this class.

Objective



Page 8 of 99


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



9





METHODS

Data sources








Study selection








Page 9 of 99


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



10


























Page 10 of 99


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



11







Model development


















Page 11 of 99


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



12

























Page 12 of 99


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



13




Technical Appendix.





















Page 13 of 99


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



14

RESULTS






















Page 14 of 99


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



15

















DISCUSSION

The model developed in this study incorporates Bayesian methodology and provides a tangible






Page 15 of 99


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



16



robustness of the model. Although the use of MBMA is relatively new in the field of diabetes

therapy, this method is nonetheless being increasingly recognized as an important tool in the

evaluation of pharmaceutical therapies48 49 .






















Page 16 of 99


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



17

















has recently been proposed for the statistically valid use of MBMA.3 The advantage of this

approach, also used in the present study, is that it enabled the synthesis of longitudinal data

from multiple studies with different durations and different sampling schedules, resulting in

analyses that are both more comprehensive (including a greater number of studies), and more

efficient (incorporating more of the relevant data within each study) than previous methods. The

unique MBMA approach in the current study also allows adjustment for covariates (eg,

differences in the use of washout or racial composition in individual trials) to allow comparison of

treatment response in comparable patients under similar conditions. One limitation of the study

by Gibbs et al. 201115 was that the MBMA used did not account for correlations across time

points within treatment arms, which could lead to an overestimation of the inter-trial variability in

Page 17 of 99


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



18

drug effect. In contrast, our approach takes account of longitudinal correlations, in accordance

with previously published methods,3 which is a prerequisite to the correct characterisation of

uncertainty in the estimation of drug effects.


require information on the relative benefits of the individual drugs within this class. Based on the

model developed in this study, it is apparent that the efficacy of the two DPP-4 inhibitors,

sitagliptin and linagliptin, is virtually indistinguishable, in terms of changes in mean HbA1c

levels, in patients with T2DM treated with a range of background anti-hyperglycaemic therapies.

Both linagliptin and sitagliptin act by inhibiting the DPP-4 enzyme that rapidly inactivates the

intestinal hormone, glucagon-like peptide (GLP)-1. GLP-1 stimulates insulin secretion in a

glucose-dependent manner. Sitagliptin is largely excreted via the kidneys, with the majority of

an oral dose (87%) being excreted in the urine.56 Unlike sitagliptin and other DPP-4 inhibitors,

linagliptin has a largely non-renal route of excretion (only ~5% excreted renally), with the

majority being eliminated via the bile and gut;57 58 it therefore does not require dose adjustment

in patients with declining renal function.59 In view of the similar efficacy of these two drugs,

treatment choices might, therefore, be made on the basis of other differences between the

drugs and consideration of patient clinical characteristics, such as the patient’s renal function.




Page 18 of 99


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



19

Contributors













Funding



Page 19 of 99


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



20

Competing interests










Page 20 of 99


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



21

References






2005;9(26):1-134, iii-iv.





Metab 2010;12(8):648-58.





6. Ito K, Ahadieh S, Corrigan B, French J, Fullerton T, Tensfeldt T. Disease

progression meta-analysis model in Alzheimer's disease. Alzheimers Dement

2010;6(1):39-53.

Page 21 of 99


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



22

7. Rogers JA, Polhamus D, Gillespie WR, Ito K, Romero K, Qiu R, et al. Combining

patient-level and summary-level data for Alzheimer's disease modeling and

simulation: a beta regression meta-analysis. J Pharmacokinet Pharmacodyn

2012;39(5):479-98.

8. Demin I, Hamren B, Luttringer O, Pillai G, Jung T. Longitudinal model-based

meta-analysis in rheumatoid arthritis: an application toward model-based drug

development. Clin Pharmacol Ther 2012;92(3):352-9.

9. Mandema JW, Hermann D, Wang W, Sheiner T, Milad M, Bakker-Arkema R, et

al. Model-based development of gemcabene, a new lipid-altering agent. Aaps J

2005;7(3):E513-22.

10. Luu KT, Raber SR, Nickens DJ, Vicini P. A model-based meta-analysis of the

effect of latanoprost chronotherapy on the circadian intraocular pressure of

patients with glaucoma or ocular hypertension. Clin Pharmacol Ther

2010;87(4):421-5.

11. Renard D, Looby M, Kramer B, Lawrence D, Morris D, Stanski DR.

Characterization of the bronchodilatory dose response to indacaterol in patients

with chronic obstructive pulmonary disease using model-based approaches.

Respir Res 2011;12:54.

12. Mandema JW, Cox E, Alderman J. Therapeutic benefit of eletriptan compared to

sumatriptan for the acute relief of migraine pain--results of a model-based meta-

analysis that accounts for encapsulation. Cephalalgia 2005;25(9):715-25.

Page 22 of 99


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



23

13. Mandema JW, Boyd RA, DiCarlo LA. Therapeutic index of anticoagulants for

prevention of venous thromboembolism following orthopedic surgery: a dose-

response meta-analysis. Clin Pharmacol Ther 2011;90(6):820-7.

14. Mandema JW, Salinger DH, Baumgartner SW, Gibbs MA. A dose-response

meta-analysis for quantifying relative efficacy of biologics in rheumatoid arthritis.

Clin Pharmacol Ther 2011;90(6):828-35.







2008(2):CD006739.









Page 23 of 99


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



24








2011;28(11):1352-61.




Obes Metab 2011;13(1):65-74.









Page 24 of 99


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



25


Arab Emirates.





Lisbon, Portugal.







2004;6(S1):Abstract W4354.





Greece.

Page 25 of 99


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



26




2010;64(5):562-76.




2006;29(12):2632-7.








Diabetes Care 2006;29(12):2638-43.




Diabetes Care 2007;30(8):1979-87.

Page 26 of 99


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



27




















Page 27 of 99


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



28










Obes Metab 2008;10(10):959-69.







http://clinicaltrials.gov/ct2/show/NCT00328172?term=1218.5&rank=1. Accessed

11 January, 2012.



Page 28 of 99


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



29









and Prevention. 2008.

www.fda.gov/downloads/Drugs/.../Guidances/ucm071624.pdf. Accessed 29

November 2012.



www.ema.europa.eu/docs/en_GB/...guideline/.../WC500129256.pdf. Accessed

29 November 2012.

48. Mould DR. Model-based meta-analysis: an important tool for making quantitative

decisions during drug development. Clin Pharmacol Ther 2010;92(3):283-6.

49. Mandema JW, Gibbs M, Boyd RA, Wada DR, Pfister M. Model-based meta-

analysis for comparative efficacy and safety: application in drug development and

beyond. Clin Pharmacol Ther 2011;90(6):766-9.

Page 29 of 99


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



30













2011;13(7):594-603.







Page 30 of 99


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



31










2008;48(10):1171-8.




[[Legends]]




Page 31 of 99


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



32









placebo.












Page 32 of 99


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



33

Figure 5










black, 37.0% Asian.

Page 33 of 99


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



34


SU, sulphonylurea.

Study Drug Dose

(mg/day)

Treatment

duration

(weeks)

Patients

(n)

Baseline

age

(years)

Female

(%)

Baseline

HbA1c

(%)

Baseline

BMI

(kg/m2)

Washout

duration

(weeks)

Concomitant

medications

Aschner

200630

Placebo NA 24 244 54.3 48.6 8.03 30.8 14 NA

Sitagliptin 100 24 229 53.4 42.9 8.01 30.3 14 NA

200 24 238 54.9 53.2 8.08 30.3 14 NA

Bergenstal

201031


Charbonnel

200632



Goldstein

200733

Placebo NA 24 165 53.3 47.2 8.68 32.5 14 NA

Sitagliptin 100 24 175 53.6 48.0 8.87 31.2 14 NA

Hanefeld Placebo NA 12 107 55.9 36.9 7.59 31.4 8 NA

Page 34 of 99


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960




35

200734

Sitagliptin 25 12 107 55.1 48.6 7.71 31.9 8 NA

50 12 107 55.3 54.5 7.60 31.6 8 NA

50 12 108 55.2 55.9 7.79 32.7 8 NA

100 12 106 56 44.5 7.78 31.6 8 NA

Hermansen

200735




+ metformin


+ metformin

Iwamoto

201036

Placebo NA 12 73 60.2 31.5 7.74 24.1 8 NA

Sitagliptin 25 12 80 59.9 36.3 7.49 25.0 8 NA

50 12 72 60.2 34.7 7.57 24.5 8 NA

100 12 70 58.3 48.6 7.56 24.2 8 NA

200 12 68 60.6 41.2 7.65 24.4 8 NA

Mohan

200937

Placebo NA 18 169 50.9 40.0 8.70 24.9 8 NA

Page 35 of 99


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960




36

Sitagliptin 100 18 339 50.9 43.0 8.70 25.1 8 NA

Nonaka

200838

Placebo NA 12 75 55.0 34.0 7.69 25.1 8 NA

Sitagliptin 10 12 75 55.6 40.0 7.54 25.2 8 NA

Raz 200639

Placebo NA 18 103 55.5 37.3 8.05 32.5 14 NA

Sitagliptin 100 18 193 54.5 46.3 8.04 31.8 14 NA

200 18 199 55.4 49.5 8.14 32.0 14 NA

Rosenstock

200640



Scheen

20105

Saxagliptin 5 18 334 58.8 52.9 7.68 31.1


Seck

201029


Scott

200741


Sitagliptin 10 12 122 55.1 50.4 7.89 30.8 8 NA

25 12 122 56.2 52 7.85 30.5 8 NA

50 12 120 55.6 42.3 7.89 31.4 8 NA

Page 36 of 99


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960




37

100 12 121 55.1 47.6 7.96 30.4 8 NA

Scott

200842

Placebo NA 18 88 55.3 41.0 7.68 30.0 8 NA

Sitagliptin 100 18 91 55.2 45.0 7.75 30.3 0 NA



Study

1218.543

2.5 12 55 60.0 52.7 8.38 31.5 6 NA

5 12 54 56.0 42.6 8.38 31.2 6 NA

Forst

201019

Placebo NA 12 70 60.0 38.6 8.37 32.2 6 NA


5 12 62 60.0 46.8 8.46 31.6 6 Metformin

10 12 66 62.0 47.0 8.35 31.7 6 Metformin

Del Prato

201118


Linagliptin 5 24 333 56.0 51.4 8.00 29.0 6 NA

Taskinen

201121

Placebo NA 24 175 57.0 42.3 8.02 30.1 6 NA


Page 37 of 99


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960




38

Owens

201120



+ SU

Gallwitz

201122


+ SU

Araki

201144


Linagliptin 5 12 159 60.0 30.2 8.07 24.6 4 NA

10 12 160 61.0 30.0 7.98 25.0 4 NA

Lewin

201045

Placebo NA 18 82 56.0 39.0 8.60 28.1 4 NA

Linagliptin 5 18 158 57.0 52.5 8.61 28.3 6 SU

Patel

201123

Placebo NA 18 73 56.0 57.5 8.06 30.0 6 SU

Linagliptin 5 18 147 57.0 64.0 8.11 29.0 6 NA

Rafeiro

201124

Placebo NA 12 43 59.0 51.2 7.92 28.6 6 NA


Page 38 of 99


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960




1







Alegre, RS, Brazil








diabetes mellitus

Page 39 of 99


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



2




June 2012.

Page 40 of 99


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



3

ABSTRACT


efficacies of two drugs, using model-based meta-analysis (MBMA). Comparison of two oral

dipeptidyl peptidase-4 inhibitors, sitagliptin and linagliptin, for treatment of type 2 diabetes

mellitus (T2DM) treatment was used as an example.

Design: A sSystematic review with model-based meta-analysis (MBMA).



patient linagliptin data from the manufacturer of linagliptin.

Eligibility criteria for selecting studies: Double-blind, randomised, controlled, clinical trials,

≥at least 12 weeks’ in duration, that analysed the efficacy of sitagliptin or linagliptin efficacies as

changes in glycated haemoglobin (HbA1c) levels, in adults with T2DM and HbA1c >7.0%,

irrespective of background medication.






composition reflecting the average racial distribution of the linagliptin trials, and baseline HbA1c

of 8.0%.

Page 41 of 99


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



4

Results: The model was developed using lLongitudinal data from 11 234 patients (10 linagliptin,

15 sitagliptin trials) were included, and assessed by internal evaluation techniques,

demonstrating that the model adequately described the observations. Simulations showed that

both linagliptin 5 mg and sitagliptin 100 mg reduced HbA1c by 0.81% (placebo-adjusted) at

week 24. Credible intervals for participants without washout were –0.88 to –0.75 (linagliptin) and

–0.89 to –0.73 (sitagliptin)., and for those with washout, –0.91 to –0.76 (linagliptin) and –0.91 to

–0.75 (sitagliptin), when administered for 24 weeks.

Conclusions: This study demonstrates the use of longitudinal MBMA in the field of diabetes

treatment. Based on an example evaluating HbA1c reduction with linagliptin versus sitagliptin,

the model used seems a valid approach for indirect drug comparisons. The results show

sitagliptin and linagliptin have virtually indistinguishable efficacies in HbA1c reduction in patients

with T2DM.

Page 42 of 99


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



5

ARTICLE SUMMARY

Article focus





compounds, incorporating both longitudinal and/or dose–response data.

Key messages

• A longitudinal statistical model was developed for the indirect comparison of two

pharmaceutical compounds (, oral DPP-4 inhibitors linagliptin and sitagliptin), with

respect to changes in HbA1c levels in patients wit h type 2 diabetes mellitus (T2DM).

• The model was evaluated by comparing model predictions to observed valuesto

compare the efficacy of two oral DPP-4 inhibitors, sitagliptin and linagliptin, with respect

to changes in HbA1c levels in patients with type 2 diabetes mellitus (T2DM).









Page 43 of 99


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



6








Page 44 of 99


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



7

INTRODUCTION



comparisons among all available treatment options. Indirect comparisons and nNetwork meta-

analysis (mixed treatment comparisons) hasve been used to estimate relative efficacy when







treatment duration.


developed and used to estimate the comparative efficacy of two medications. MBMA can be

used to provide a mechanism for integrating information from heterogeneously designed trials

and, thus, to evaluate outcomes with different drugs that have not been compared directly.3

Model-based meta-analysis is distinguished from the methodology of conventional meta-

analysis by the manner in which it incorporates longitudinal and/or dose–response data. By

modelling the response as a parametric function of time, MBMA allows the integration of

information from trials of different durations and with different sampling time points. This enables

the use of less restrictive inclusion/exclusion criteria for study selection, and more efficient use

of data from the studies that are selected, therefore resulting in a particularly comprehensive

summary of all relevant data.3



Page 45 of 99


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



8












This approach to drug comparison employs a mathematical model to describe the time-course

of HbA1c levels, and is being increasingly used to characterise longitudinal data. The general

meta-analytic methodology of Ahn and French3 has previously been used to successfully

describe longitudinal meta-data from clinical trials in Alzheimer’s disease6 7, rheumatoid

arthritis8, lipid disorders,9 glaucoma,10 and chronic obstructive pulmonary disease.11 Similar

approaches have been used to perform dose-response meta-analyses in a range of therapeutic

areas, including migraine,12 post-operative anticoagulant therapy,13 and rheumatoid arthritis.14

This analytic approach has also been used in the field of diabetes in a recent study by Gibbs et

al. 2011,15 which evaluated the relationship between DPP-4 inhibition and HbA1c reduction

using data obtained from clinical trials of four drugs in this class.

Comment [A1]: New citations added (changes to

references not shown as tracked changes)

Comment [A2]: Moved from Discussion

Page 46 of 99


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



9

Objective







METHODS

Data sources








Study selection






Page 47 of 99


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



10


























Page 48 of 99


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



11









Model development

















Page 49 of 99


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



12

























Page 50 of 99


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



13





Technical Appendix.



















Page 51 of 99


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



14



RESULTS





















Page 52 of 99


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



15


















DISCUSSION


tangible approach to indirectly estimating the comparative efficacy of two compounds. The

findings presented suggest that the model developed in this study provides a valid alternative

approach to indirect drug comparisons. The findings of this MBMA show that linagliptin is

equally effective as sitagliptin in the reduction of HbA1c levels, both showing a mean, placebo-

Page 53 of 99


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



16

adjusted reduction of approximately 0.81% after 24-week treatment of patients with T2DM. In

this study, evidence was gathered from the results of randomised, double-blind trials of

sitagliptin and linagliptin. Sensitivity analyses performed in this study, using various prior

distributions, support the robustness of the model. As far as we know, the present MBMA

represents one of the first such applications of these techniques in the field of diabetes

treatment. Although the use of MBMA is relatively new in the field of diabetes therapy, this

method is nonetheless being increasingly recognized as an important tool in the evaluation of

pharmaceutical therapies48 49 .



















Comment [A3]: Sentence revised and 2 new

citations added

Page 54 of 99


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



17



























Comment [A4]: Moved to Introduction

Page 55 of 99


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



18










require information on the relative benefits of the individual drugs within this class. Based on the

model developed in this study, it is apparentThis study demonstrated that the efficacy of the two

DPP-4 inhibitors, sitagliptin and linagliptin, is virtually indistinguishable, in terms of changes in

mean HbA1c levels, in patients with T2DM treated with a range of background anti-

hyperglycaemic therapies. Both linagliptin and sitagliptin act by inhibiting the DPP-4 enzyme

that rapidly inactivates the intestinal hormone, glucagon-like peptide (GLP)-1. GLP-1 stimulates

insulin secretion in a glucose-dependent manner. Sitagliptin is largely excreted via the kidneys,

with the majority of an oral dose (87%) being excreted in the urine.56 Unlike sitagliptin and other

DPP-4 inhibitors, linagliptin has a largely non-renal route of excretion (only ~5% excreted

renally), with the majority being eliminated via the bile and gut;57 58 it therefore does not require

dose adjustment in patients with declining renal function.59 In view of the similar efficacy of these

two drugs, treatment choices might, therefore, be made on the basis of other differences

between the drugs and consideration of patient clinical characteristics, such as the patient’s

renal function.

Page 56 of 99


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



19




Page 57 of 99


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



20

Contributors













Funding



Page 58 of 99


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



21

Competing interests










Page 59 of 99


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



22

References






2005;9(26):1-134, iii-iv.





Metab 2010;12(8):648-58.





6. Ito K, Ahadieh S, Corrigan B, French J, Fullerton T, Tensfeldt T. Disease

progression meta-analysis model in Alzheimer's disease. Alzheimers Dement

2010;6(1):39-53.

Page 60 of 99


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



23

7. Rogers JA, Polhamus D, Gillespie WR, Ito K, Romero K, Qiu R, et al. Combining

patient-level and summary-level data for Alzheimer's disease modeling and

simulation: a beta regression meta-analysis. J Pharmacokinet Pharmacodyn

2012;39(5):479-98.

8. Demin I, Hamren B, Luttringer O, Pillai G, Jung T. Longitudinal model-based

meta-analysis in rheumatoid arthritis: an application toward model-based drug

development. Clin Pharmacol Ther 2012;92(3):352-9.

9. Mandema JW, Hermann D, Wang W, Sheiner T, Milad M, Bakker-Arkema R, et

al. Model-based development of gemcabene, a new lipid-altering agent. Aaps J

2005;7(3):E513-22.

10. Luu KT, Raber SR, Nickens DJ, Vicini P. A model-based meta-analysis of the

effect of latanoprost chronotherapy on the circadian intraocular pressure of

patients with glaucoma or ocular hypertension. Clin Pharmacol Ther

2010;87(4):421-5.

11. Renard D, Looby M, Kramer B, Lawrence D, Morris D, Stanski DR.

Characterization of the bronchodilatory dose response to indacaterol in patients

with chronic obstructive pulmonary disease using model-based approaches.

Respir Res 2011;12:54.

12. Mandema JW, Cox E, Alderman J. Therapeutic benefit of eletriptan compared to

sumatriptan for the acute relief of migraine pain--results of a model-based meta-

analysis that accounts for encapsulation. Cephalalgia 2005;25(9):715-25.

Page 61 of 99


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



24

13. Mandema JW, Boyd RA, DiCarlo LA. Therapeutic index of anticoagulants for

prevention of venous thromboembolism following orthopedic surgery: a dose-

response meta-analysis. Clin Pharmacol Ther 2011;90(6):820-7.

14. Mandema JW, Salinger DH, Baumgartner SW, Gibbs MA. A dose-response

meta-analysis for quantifying relative efficacy of biologics in rheumatoid arthritis.

Clin Pharmacol Ther 2011;90(6):828-35.







2008(2):CD006739.









Page 62 of 99


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



25








2011;28(11):1352-61.




Obes Metab 2011;13(1):65-74.









Page 63 of 99


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



26


Arab Emirates.





Lisbon, Portugal.







2004;6(S1):Abstract W4354.





Greece.

Page 64 of 99


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



27




2010;64(5):562-76.




2006;29(12):2632-7.








Diabetes Care 2006;29(12):2638-43.




Diabetes Care 2007;30(8):1979-87.

Page 65 of 99


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



28




















Page 66 of 99


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



29










Obes Metab 2008;10(10):959-69.







http://clinicaltrials.gov/ct2/show/NCT00328172?term=1218.5&rank=1. Accessed

11 January, 2012.



Page 67 of 99


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



30









and Prevention. 2008.

www.fda.gov/downloads/Drugs/.../Guidances/ucm071624.pdf. Accessed 29

November 2012.



www.ema.europa.eu/docs/en_GB/...guideline/.../WC500129256.pdf. Accessed

29 November 2012.

48. Mould DR. Model-based meta-analysis: an important tool for making quantitative

decisions during drug development. Clin Pharmacol Ther 2010;92(3):283-6.

49. Mandema JW, Gibbs M, Boyd RA, Wada DR, Pfister M. Model-based meta-

analysis for comparative efficacy and safety: application in drug development and

beyond. Clin Pharmacol Ther 2011;90(6):766-9.

Page 68 of 99


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



31













2011;13(7):594-603.







Page 69 of 99


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



32










2008;48(10):1171-8.




[[Legends]]




Page 70 of 99


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



33









placebo.












Page 71 of 99


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



34

Figure 5










black, 37.0% Asian.

Page 72 of 99


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



35


SU, sulphonylurea.

Study Drug Dose

(mg/day)

Treatment

duration

(weeks)

Patients

(n)

Baseline

age

(years)

Female

(%)

Baseline

HbA1c

(%)

Baseline

BMI

(kg/m2)

Washout

duration

(weeks)

Concomitant

medications

Aschner

200630

Placebo NA 24 244 54.3 48.6 8.03 30.8 14 NA

Sitagliptin 100 24 229 53.4 42.9 8.01 30.3 14 NA

200 24 238 54.9 53.2 8.08 30.3 14 NA

Bergenstal

201031


Charbonnel

200632



Goldstein

200733

Placebo NA 24 165 53.3 47.2 8.68 32.5 14 NA

Sitagliptin 100 24 175 53.6 48.0 8.87 31.2 14 NA

Hanefeld Placebo NA 12 107 55.9 36.9 7.59 31.4 8 NA

Page 73 of 99


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960




36

200734

Sitagliptin 25 12 107 55.1 48.6 7.71 31.9 8 NA

50 12 107 55.3 54.5 7.60 31.6 8 NA

50 12 108 55.2 55.9 7.79 32.7 8 NA

100 12 106 56 44.5 7.78 31.6 8 NA

Hermansen

200735




+ metformin


+ metformin

Iwamoto

201036

Placebo NA 12 73 60.2 31.5 7.74 24.1 8 NA

Sitagliptin 25 12 80 59.9 36.3 7.49 25.0 8 NA

50 12 72 60.2 34.7 7.57 24.5 8 NA

100 12 70 58.3 48.6 7.56 24.2 8 NA

200 12 68 60.6 41.2 7.65 24.4 8 NA

Mohan

200937

Placebo NA 18 169 50.9 40.0 8.70 24.9 8 NA

Page 74 of 99


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960




37

Sitagliptin 100 18 339 50.9 43.0 8.70 25.1 8 NA

Nonaka

200838

Placebo NA 12 75 55.0 34.0 7.69 25.1 8 NA

Sitagliptin 10 12 75 55.6 40.0 7.54 25.2 8 NA

Raz 200639

Placebo NA 18 103 55.5 37.3 8.05 32.5 14 NA

Sitagliptin 100 18 193 54.5 46.3 8.04 31.8 14 NA

200 18 199 55.4 49.5 8.14 32.0 14 NA

Rosenstock

200640



Scheen

20105

Saxagliptin 5 18 334 58.8 52.9 7.68 31.1


Seck

201029


Scott

200741


Sitagliptin 10 12 122 55.1 50.4 7.89 30.8 8 NA

25 12 122 56.2 52 7.85 30.5 8 NA

50 12 120 55.6 42.3 7.89 31.4 8 NA

Page 75 of 99


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960




38

100 12 121 55.1 47.6 7.96 30.4 8 NA

Scott

200842

Placebo NA 18 88 55.3 41.0 7.68 30.0 8 NA

Sitagliptin 100 18 91 55.2 45.0 7.75 30.3 0 NA



Study

1218.543

2.5 12 55 60.0 52.7 8.38 31.5 6 NA

5 12 54 56.0 42.6 8.38 31.2 6 NA

Forst

201019

Placebo NA 12 70 60.0 38.6 8.37 32.2 6 NA


5 12 62 60.0 46.8 8.46 31.6 6 Metformin

10 12 66 62.0 47.0 8.35 31.7 6 Metformin

Del Prato

201118


Linagliptin 5 24 333 56.0 51.4 8.00 29.0 6 NA

Taskinen

201121

Placebo NA 24 175 57.0 42.3 8.02 30.1 6 NA


Page 76 of 99


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960




39

Owens

201120



+ SU

Gallwitz

201122


+ SU

Araki

201144


Linagliptin 5 12 159 60.0 30.2 8.07 24.6 4 NA

10 12 160 61.0 30.0 7.98 25.0 4 NA

Lewin

201045

Placebo NA 18 82 56.0 39.0 8.60 28.1 4 NA

Linagliptin 5 18 158 57.0 52.5 8.61 28.3 6 SU

Patel

201123

Placebo NA 18 73 56.0 57.5 8.06 30.0 6 SU

Linagliptin 5 18 147 57.0 64.0 8.11 29.0 6 NA

Rafeiro

201124

Placebo NA 12 43 59.0 51.2 7.92 28.6 6 NA


Page 77 of 99


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960




181x90mm (300 x 300 DPI)

Page 78 of 99


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



182x90mm (300 x 300 DPI)

Page 79 of 99


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from






Scre

enin

g In

clud

ed

Elig

ibili

ty

Iden

tific

atio

n Sitagliptin and linagliptin* records identified in Embase searching, Cochrane library,



(n = 1008)





(n = 41)


(n = 16)


(n = 25)





Page 80 of 99


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



102x90mm (300 x 300 DPI)

Page 81 of 99


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



102x90mm (300 x 300 DPI)

Page 82 of 99


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



90x100mm (300 x 300 DPI)

Page 83 of 99


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



90x184mm (300 x 300 DPI)

Page 84 of 99


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



96x90mm (300 x 300 DPI)

Page 85 of 99


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



Technical Appendix to:A novel model-based meta-analysis to indirectly

estimate the comparative efficacy of twomedications: an example using DPP-4 inhibitors,sitagliptin and linagliptin, in treatment of type 2

diabetes mellitus

December 3, 2012

Jorge L Gross1, James Rogers2, Dan Polhamus2, William Gillespie2, Christian Friedrich3, Yan Gong4, BrigittaMonz4, Sanjay Patel5, Alexander Staab3, Silke Retlich3

(1) Hospital de Clínicas de Porto Alegre, Universidade Federal do Rio Grande do Sul, Porto Alegre, RS, Brazil(2) Metrum Research Group, Tariffville, Connecticut, USA(3) Boehringer Ingelheim, Biberach, Germany(4) Boehringer Ingelheim, Ingelheim, Germany(5) Boehringer Ingelheim, Bracknell, Berkshire, UK

Page 86 of 99


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



Technical Appendix

This appendix is intended primarily to provide a mathematical / statistical specification of the final modelemployed in our analysis. We additionally mention several variants of the model that were considered dur-ing model development.

1 Base model: structural form

The structural form of the model (i.e., the parametric form relating the central tendency of predictions totime, dose, and covariates) remained largely unchanged across all model-fitting iterations described in thepresent report. This structural form may conceptualized in terms of a latent (random-effect) baseline HbA1c(HbA1cbase) that is modified as a function of dose (D) and time (t ) by multiplicative terms for the effect ofwashout, placebo intervention, and drug effect (where applicable).

HbA1c(t , D) = HbA1cbase

×Iwashout

�

1+∆W∞�


1+∆W∞�


×�

1+∆P∞�

f P (t )��

(placebo effect)

×�

1−Emax,drugD

E D50,drug+D

�

1− e−kdrugt�

�

(drug effect)

We now consider each of the model terms in slightly more detail:

• Washout term�

1+∆W∞�


1+∆W∞�


Although response observations from the (pre-treatment) washout period were not modeled, the du-ration of pre-treatment washout twash is used to determine how much of the post-baseline trend canbe attributed to washout. One may better understand the washout term in terms of the following twoextreme scenarios:

– The washout term for a group with an extremely long pre-treatment washout (twash ≈∞) wouldbe essentially 1, i.e., no adjustment.

– The washout term for a group discontinuing prior medication only at the time of the first doseof the randomized treatment (twash ≈ 0) would be

�

1+∆W∞�

1− e−kW (t )��

, a term that rises ex-ponentially from 1 (no adjustment at baseline) toward a horizontal asymptote of (1+∆W∞) as tbecomes very large.

As a conceptual and notational device, we also define HbA1cprior, the inferred HbA1c level at the be-ginning of the washout period:

HbA1cprior ≡HbA1cbase

1+∆W∞�


Page 1 of 7

Page 87 of 99


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



Technical Appendix

• Placebo term�

1+∆P∞�

f P (t )��

. Both exponential and a bi-exponential variants were considered forf P :

f P (t ) = 1− e−kP t ("exponential" variant)

f P (t ) = e−kP0t − e−kP t ("bi-exponential" variant)

The placebo term is intended to characterize all aspects of the longitudinal profile that are commonto all placebo-treated patients. In principal at least, this may include more than a mere “placeboeffect”. For example, it might also partially reflect longitudinal changes that would occur even in theabsence of placebo intervention. In contrast to the washout term, the placebo term does not varysystematically as a function of washout duration (in particular, in contrast to the washout term, theplacebo “effect” does not approach a limit of zero as washout duration approaches infinity), so thatthe two effects are distinctly identifiable.

The exponential and bi-exponential variants of the placebo function have different implications withrespect to the limiting placebo effect P∞ at large time values. The bi-exponential variant, sometimesreferred to as a “Bateman” function is a typical choice in the pharmacology domain for modelingplacebo effects, and is applicable when placebo effects are expected to return to zero at some dura-tion [4]. By contrast, our “exponential” variant implies a non-zero limiting effect due to placebo. Inpractice, the predictive implications for the fitted model may be similar for both functions over a finiteduration of interest, since the estimated parameters in the Bateman function may (if supported by thedata) imply a positive first derivative over most or all of that duration (in our case, over a 24 weekduration). To the extent that data do not support any return to zero for the placebo effects over theduration of interest, both convergence diagnostics and model selection criteria will tend to favor thesimpler “exponential” placebo function. In our application the bi-exponential / Bateman functionalform appeared to be reasonably well estimated and so was therefore employed in the final model.

As a conceptual and notational device, we also define HbA1cplacebo(t ), the expected time course for anindividual randomized to placebo:

HbA1cplacebo(t ) = HbA1cbase

×Iwashout

�

1+∆W∞�


1+∆W∞�


×�

1+∆P∞�

f P (t )��

(placebo effect)

Further, we define HbA1c∞ to be the limiting value of HbA1cplacebo(t ) as t becomes very large.

• Drug effect term

(1−Edrug)≡�

1−Emax,drugD

E D50,drug+D

�

1− e−kdrugt�

�

This term varies as a function of both dose (D) and time (t ). The term equals 1 (no adjustment) whent = 0 and/or D = 0, and approaches a horizontal asymptote of (1− Emax,drug) as both D and t becomevery large. The parameter kdrug, which describes the onset of drug effects, was provisionally assumed

Page 2 of 7

Page 88 of 99


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



Technical Appendix

to take the same value for both linagliptin and sitagliptin, consistent with the expected pharmacol-ogy of the two DPP4 inhibitors (empirical support for this assumption is provided by the posteriorpredictive checks included in the primary manuscript).

The dose-response component of our model may be described as a simplified Emax model, a moregeneral expression of which is:

E =EmaxDα

E Dα50+Dα

.

This functional form, also known as the Hill Equation, is widely used in the pharmacology domain todescribe both concentration-response and dose-response relationships [5]. Originally proposed onthe basis of drug receptor theory to describe concentration-response analyses, its use in describingdose-response relationships may also be theoretically justified when linear pharmacokinetics applyand dose is therefore an appropriate proxy for concentration. Theoretical support is more tenuouswhen nonlinear pharmacokinetics apply (as is the case for linagliptin), but the model remains a rea-sonable initial default. The parameterα, sometimes referred to as the Hill coefficient or the sigmoidic-ity parameter, has been set to a value of 1 in our implementation. This simplification was initially in-troduced on a tentative basis, in consideration of the limited degree of dose-response information inthe data, and based on prior experiences of the modeling team suggesting that, empirically, the Hillcoefficient for dose-response relationship is generally close to one. This simplification does imply adose-response relationship that has a non-zero gradient at D = 0, however this implication is consis-tent with observed dose-response relationships (anecdotally) for most drugs. Final acceptance of thisaspect of the model was based on the observed predictive performance of our model at all studieddose levels.

We may now use the terms introduced above to define the conditional expectation for the mean HbA1c onthe i th occasion in the j th group and k th study:

ÚHbA1ci j k = HbA1cprior,j k

×�

1+∆W∞,j k

�

1− e−kW (t i j k+twash,j k )��

×�

1+∆P∞,j k

�

f P (t i j k )��

×

1−Emax,drugj k

D j k

E D50,drugj k+D j k

�

1− e−kdrugt i j k�

!

Note that the circumflex, or “hat”, has not been used here to indicate an estimate (as would be commonin the statistical literature), but refers rather to an expected value (as is common in the pharmacometricliterature).

2 Base model: stochastic structure

Random effects associated with washout and placebo effects were implemented using a re-parameterizationof the model. The parameters ∆W∞ and ∆P∞ represent conceptual steady state asymptotes for very large

Page 3 of 7

Page 89 of 99


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



Technical Appendix

time values (t →∞). Since it is not clear that direct estimation of such parameters is supported by the data,the model was implemented using truncated parameterizations that reformulate the effect parameter as aneffect at a time t ∗ that is richly represented in the data set. For this purpose we define ∆W ∗ and ∆P∗ as thefractional changes at reference time t ∗ = 24 weeks due to washout and placebo, respectively. That is,

∆W ∗ =�

1+∆W∞�

1− e−kW (t ∗+twash)��

1+∆W∞�

1− e−kW twash� −1

∆P∗ = ∆P∞ f P (t ∗).

The random-effect structure for the i th visit for the j th arm (or group) in the k th study may then be describedas:

Inter-study variation:

log�

1+∆P∗study,k

�

∼ N�

log�

1+Ô∆P∗�

,ψ2∆P

�

Study-level random effects are additionally considered for mean baseline HbA1c and for the washout mag-nitude∆W ∗, but convergence diagnostics suggested that this level of variation in these parameters was notidentifiable based on the available data.

All study-level random effects are assumed to be independent.

Inter-arm variation:

log�

1+∆P∗j k

�

∼ N�

log�

1+∆P∗study,k

�

,ω2∆P/n 1j k

�

log�

HbA1cbase,j k

�

∼ N�

log�

HbA1cbase-study,k

�

,ω2HbA1cbase

/n 1j k

�

All arm-level random effects are assumed to be conditionally independent, given study-level random effects.Our scaling of random effect variances according to sample size follows a recently published rationale [1],and is similar to an approach used in model-based meta-analysis of Alzheimer’s Disease progression [3].Arm-level random effects were additionally considered for the washout magnitude ∆W ∗, but convergencediagnostics suggested that this level of variation was also not identifiable based on the available data.

Modeling baseline HbA1c values using Lognormal distributions constrains these to positive values. Model-ing the ∆ parameters using shifted Lognormal distributions allows the placebo and washout factors in themodel to take positive and negative (up to negative one) values. Taken together with constraints on the drugeffects (ensured via the prior for the drug Emax values), and the multiplicative construction of the model, thepreceding constraints ensure that all HbA1c predicted values are positive.

Residual variation:

HbA1ci j k ∼N

�

ÚHbA1ci j k ,σ2

n i j k

�

.

Page 4 of 7

Page 90 of 99


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



Technical Appendix

3 Covariates

Baseline HbA1cBaseline HbA1c is implicitly included as a covariate in the model in the sense that the washout, placebo, anddrug effect terms all operate multiplicatively on baseline HbA1c. In general, larger changes from baselineare implied by larger baseline values.

WashoutThe inclusion of the washout term in the base model implies that washout status and washout duration actas covariates in the sense that they modify an individual’s predicted change from baseline.

Other covariates

Whereas the covariate effects of baseline HbA1c and washout duration are implied by the longitudinal struc-ture of the base model, the remaining candidate covariates were considered as potential modifiers of drugeffect in a manner dependent on the covariate variable’s distributional properties.

• For univariate covariates taking discrete values at the arm-level (as was the case with backgroundmedication status), the covariate effect was introduced via an exponentiated linear predictor so that,with x j k denoting the dichotomous variable level on arm j in study k , the following substitution ismade:


× exp¦

βdrugj kx j k

©

(This approach is equivalent, modulo re-parameterization, to simply using a different Emax,drug foreach level of the covariate.)

• For univariate covariates taking continuous values at the arm-level including age, body mass index(BMI), gender (recall that at the arm-level, gender is represented as a proportion), and duration ofdiabetes, the candidate covariate was centered at its average value and an exponentiated linear pre-dictor was again used. In this case, letting x j k be the continuous covariate, the following substitutionis made:


× exp¦

βdrugj k(x j k −x ··)

©

.

• Race, like gender, is represented at the arm-level using proportions. However, unlike gender, the racialcomposition of a each study arm must be represented with multiple proportions (specifically, propor-tion white, proportion black, proportion asian, and, in certain cases, proportion “other”), precludingany interpretable use of the continuous covariate parameterization that was used for gender. Instead,a separate E r

max,drugj kwas defined for each race r (separately for each drug). Race-specific terms in

the model then imply race-specific conditional expectationsÚHbA1cr

i j k , and the aggregate conditionalexpectation is computed using a weighted average:

ÚHbA1ci j k =∑

r∈(white, black, asian, other)

F ri j kÚHbA1c

r

i j k

where F r is the fraction of the arm identifying with race r .

In order to improve parameter estimation for race-based covariate effects, racially subsetted data wereused where possible (this was possible in all cases for linagliptin records).

Page 5 of 7

Page 91 of 99


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



Technical Appendix

As noted in the primary manuscript, the effects of covariates other than race were not sufficiently well esti-mated to justify inclusion in the model. Race was therefore the only explicit in the final model (in additionto the implicit covariates, baseline HbA1c and washout duration).

4 Priors

The probababilistic specification of the prior is provided in Table 1. With regard to the primary researchquestions of interest, the most consequential components of the prior are the distributional statements re-lating to the magnitudes of drug effects, Emax, linagliptin and E ∗max, sitagliptin. Since those parameters representasymptotic drug effects (at theoretically infinite times), we re-parameterize in terms of effects at 24 weeks,E ∗linagliptin and E ∗sitagliptin, using the same reasoning as was discussed for placebo and washout effects. Sincethese drug effect parameters represent fractional reductions from a hypothetical untreated state, it is nat-ural that they should be bounded between zero and one, implying that both drugs have some beneficialeffect (a defensible assumption for marketed drugs), neither or which may reduce HbA1c levels below zero(patently true). The use of Uniform (flat) densities between these two extremes implies that all intermedi-ate values are considered (a priori) equally likely. Perhaps most importantly, the distributions for E ∗linagliptinand E ∗sitagliptin are specified as being independent. In combination with the use of Uniform densities, theprior independence of these two parameters allows for potential findings of either substantial similarity orsubstantial difference between the two drugs, as determined by the data.

Priors for ancillary (non-drug-effect) parameters were chosen largely for analytical convenience, but wereverified to be diffuse in comparison to thier corresponding posterior distributions, suggesting that these el-ements of the prior were sufficiently non-informative. Additionally, alternative distributions were evaluatedfor a number of elements of the prior (e.g. for variance components), and did not result in any substantialdifferences in conclusions.

Page 6 of 7

Page 92 of 99


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



Technical Appendix

parameter priorkW (w−1) Unif (0.001, 7)kP0 (w−1) Unif (0.001, 7)kP (w−1) Unif (0.001, 7)kdrug (w−1) Unif (0.001, 7)ÚHbA1cbase Log Normal(0, 100)Ö∆W ∗ Unif (−1, 10)log�

1+Ô∆P∗�

Unif (−10, 5)ψ∆P Unif (0, 10)ω∆P Unif (0, 10)ωHbA1c0 Unif (0, 5)σ Unif (0, 100)b linagliptin Unif (0, 100)bsitagliptin Unif (0, 100)E ∗linagliptin Unif (0, 1)E ∗sitagliptin Unif (0, 1)Emax,linagliptin ( 1

b linagliptin+1) ∗E ∗linagliptin

Emax,sitagliptin ( 1bsitagliptin

+1) ∗E ∗sitagliptin

E D50,linagliptin10

b linagliptin

E D50,sitagliptin200

bsitagliptin

Table 1: Prior distribution for parameters in base model

References

[1] Ahn, J.E. and French, J.L. Longitudinal aggregate data model-based meta-analysis with NONMEM: ap-proaches to handling within treatment arm correlation. J Pharmacokinet Pharmacodyn 37 (2010):179–201.

[2] Oehlert, G. A Note on the Delta Method. The American Statistician 46 (1992):27–29.

[3] Gillespie, W.R., Rogers, J.A., Ito, K. and Gastonguay, M.R. Population Dose-Response Model for ADAS-cog Scores in Patients with Alzheimers Disease by Meta-Analysis of a Mixture of Summary and IndividualData. In American Conference on Pharmacometrics (Mashantucket, CT, 2009).

[4] Mould, D.R. Developing Models of Disease Progression (John Wiley & Sons, Inc., 2006), pages 547–581.URL http://dx.doi.org/10.1002/9780470087978.ch21

[5] Goutelle, S., Maurin, M., Rougier, F., Barbaut, X., Bourguignon, L., Ducher, M. and Maire, P. The Hill equa-tion: a review of its capabilities in pharmacological modelling. Fundam Clin Pharmacol 22 (2008):633–48.

Page 7 of 7

Page 93 of 99


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from

http://dx.doi.org/10.1002/9780470087978.ch21






Page 94 of 99


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from





Database

(search date)


identified

Embase

(8 November 2010)



1356‘/exp OR ‘bi 1356‘

72





431‘

1545

3. Search 1 OR 2 1582














2 215 299

Page 95 of 99


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from





Cochrane library

(8 November 2010)


diabetes

[Search All Text]

48

IDEA

(10 November 2010)

23




130

Manual searching 0

Total 1008

Page 96 of 99


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from



















were used




Retnakararn 201010


Rigby 201011

Open-label design






selection bias)


Page 97 of 99


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from









References






Obes Metab 2007;9:186–193.












Pract 2010;64:562–76.



Metab Res 2009;41:232–7.

Page 98 of 99


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from
















Opin 2009;25:569–83.





phosphate) tablets.


Page 99 of 99


BMJ Open

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960



jopen.bmj.com

/B

MJ O


arch 2013. Dow

nloaded from


bmjopen.bmj.com · For peer review only Technical Appendix 3 Covariates BaselineHbA1c Baseline...

Documents

Transcript of bmjopen.bmj.com · For peer review only Technical Appendix 3 Covariates BaselineHbA1c Baseline...