Sclerostin and Insulin Resistancein Prediabetes: Evidence of a CrossTalk Between Bone and GlucoseMetabolismDiabetes Care 2015;38:1509–1517 | DOI: 10.2337/dc14-2989

OBJECTIVE

A genemutation of theWnt/b-catenin signaling cascade is present in rare patientswith the insulin resistance syndrome. Sclerostin is a circulating peptide inhibitingWnt/b-catenin signaling. Our aims were to evaluate serum sclerostin in subjectswith prediabetes and to analyze its relationship with insulin resistance and b-cellfunction.

RESEARCH DESIGN AND METHODS

We performed a cross-sectional study including 43 healthy normal glucose-tolerant(NGT) individuals and 79 individuals with impaired glucose regulation (IGR), whichincluded subjects with impaired fasting glucose (IFG), impaired glucose tolerance(IGT), and combined IFG-IGT, undergoing oral glucose tolerance test (OGTT) anddual-energy X-ray absorptiometry. A subgroup of 18 with NGT and 30 with IGR alsounderwent a euglycemic–hyperinsulinemic clamp with tracer.

RESULTS

Sclerostin levels were higher in IGR compared with NGT (50.86 2.4 vs. 38.76 2.3pmol/L; P = 0.01), positively correlatedwith HOMA-insulin resistance (IR) (r = 0.62;P < 0.001), and negatively correlated with insulin-mediated total body glucosedisposal (r =20.40; P < 0.001). Fasting endogenous glucose production (EGP) andhepatic and adipose tissue insulin resistance indexes were positively correlatedwith sclerostin levels (r = 0.48, r = 0.62, and r = 0.61, respectively; P < 0.001).Fasting and OGTT insulin clearance were inversely correlated with sclerostin se-rum levels (r =20.52 and r =20.44, respectively; both P < 0.001). Sclerostin levelswere not correlated with b-cell function parameters. In multiple linear regressionanalysis, the addition of sclerostin levels to the traditional risk factors for insulinresistance improved the r2 associated with HOMA-IR (r2 change: 0.055; F change:28.893; P = 0.001) and insulin-mediated total body glucose disposal (r2 change:0.059; F change: 4.938; P = 0.033).

CONCLUSIONS

Sclerostin levels are increased in individuals with prediabetes and correlated withinsulin resistance in skeletal muscle, liver, and adipose tissue. The correlationbetween sclerostin and insulin clearance at fasting state and during OGTT is novel;thus, studies are needed to explore the potential causal relationship.

1Department of Medicine, Diabetes Division,University of Texas Health Science Center atSan Antonio, San Antonio, TX2Institute of Neuroscience, National ResearchCouncil, Padua, Italy3Department ofMedicine, University of TexasHealthScience Center at San Antonio, San Antonio, TX4Departamento de Clinica Medica, Faculdade deCiencias Medicas, Obesity and Comorbidities Re-search Center, Universidade Estadual de Cam-pinas, Campinas, Brazil

Corresponding author: Franco Folli, folli@uthscsa.edu.

Received 16 December 2014 and accepted 28April 2015.

This article contains Supplementary Data onlineat http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc14-2989/-/DC1.

© 2015 by the American Diabetes Association.Readers may use this article as long as the workis properly cited, the use is educational and notfor profit, and the work is not altered.

Giuseppe Daniele,1 Deidre Winnier,1

Andrea Mari,2 Jan Bruder,3

Marcel Fourcaudot,1 Zuo Pengou,1

Devjit Tripathy,1 Christopher Jenkinson,1

and Franco Folli1,4

Diabetes Care Volume 38, August 2015 1509

PATH

OPHYSIO

LOGY/COMPLIC

ATIO

NS

Type 2 diabetes mellitus (T2DM) is aworldwide epidemic, importantly re-ducing life expectancy and quality oflife because of chronic complicationssuch as cardiovascular diseases, renalfailure, and blindness (1). Impaired glu-cose regulation (IGR), a condition thatincludes impaired fasting glucose (IFG)and/or impaired glucose tolerance(IGT), precedes the onset of T2DM (2).IGR, also defined as prediabetes, is

characterized by insulin resistance and al-tered insulin secretion, two key patho-physiological features preceding thedevelopment of T2DM (3). The compo-nents of the insulin resistance syndrome,such as obesity, IGR, hypertension, dysli-pidemia, endothelial dysfunction, and hy-perinsulinemia, are associated with ahigher risk of developing cardiovasculardiseases (4–6). Furthermore, 10–15% ofindividuals with prediabetes alreadyhave evidence of diabetic microvascularcomplications, suggesting that the mech-anisms responsible for diabetes complica-tions are already activewhen T2DM is notyet clinically manifested (7).Bone has recently emerged as a new

important organ in the regulation of glu-cose metabolism (8). Sclerostin, a 190-amino acid glycoprotein secretedmainlyby osteocytes, is a circulating hormonethat antagonizes the Wnt (Wingless-type mouse mammary tumor virus inte-gration side) pathway, thus suppressingosteoblast activity and downregulatingbone turnover (9). Interestingly, a genemutation in the Wnt signaling pathwayhas been associated with hyperlipid-emia, hypertension, and early coronaryartery disease in patients with meta-bolic syndrome (10,11).The role of sclerostin in insulin resis-

tance in humans has not been explored.The aim of this studywas to evaluate therelationship between sclerostin andwhole-body glucose metabolism in indi-viduals with prediabetes compared withnormal glucose-tolerant (NGT) subjects.

RESEARCH DESIGN AND METHODS

The cross-sectional study included 43healthy NGT individuals and 79 individ-uals with IGR, which included individualswith IFG, IGT, and combined IFG-IGT.According to the American Diabetes As-sociation, the diagnosis of prediabetes ismade on the basis of one of the follow-ing clinical biochemistry criteria: 1) 2-hplasma glucose (PG) during an oral

glucose tolerance test (OGTT) of140–199 mg/dL (IGT); 2) fasting plasmaglucose (FPG) of 100–125 mg/dL (IFG);or 3) HbA1c of 5.8–6.4% (12).

Exclusion criteria were 1) diagnosis ofT2DM; 2) blood pressure .140/90mmHg; 3) serum creatinine .1.6mg/dL; 4) hematocrit ,35%; 5) evidenceof major organ system disease, includingcancer, as determined by medical history,physical examination, androutinescreeningblood chemistries; and 6) use of medica-tions known to affect glucose metabolism.All study subjects had a stable bodyweightfor at least 3 months and had not partici-pated in strenuous exercise before enroll-ment. All subjects underwent an OGTT(75 g), dual-energy X-ray absorptiometry(DXA) whole-body scan, and determina-tionof plasmasclerostin levels. A subgroupof 18 with NGT and 30 with IGR also un-derwent a euglycemic–hyperinsulinemicclamp combined with tritiated glucoseto directly measure hepatic glucoseproduction and suppression after insulininfusion.

The study protocol was approved bythe institutional review board of theUniversity of Texas Health Science Cen-ter and the South Texas VeteransHealthcare System, Audie Murphy Hos-pital at San Antonio, TX. The study pro-tocol was conducted in accordance withthe guidelines of the Declaration ofHelsinki. Written and oral informed con-sent was obtained from all participantsenrolled in this study.

Study ProceduresAll metabolic studies were done in themorning at The Bartter Research Unit ofthe South Texas Veterans Healthcare Sys-tem after an overnight fast of 10–12 h.

OGTT: A catheter was placed in an ante-cubital vein and blood samples werecollected at 230, 215, 0, 30, 60, 90,and 120 min for determination of PG,C-peptide, insulin, glucagon, and freefatty acid (FFA) concentrations.

Hyperinsulinemic–euglycemic clamp:At 0600 h (2180 min), after a 10-hovernight fast, a prime (25 mCi 3FPG/100) continuous (0.25mCi/min) in-fusion of [3-3H]-glucose was startedvia a catheter placed into an antecu-bital vein and continued throughoutthe study. A second catheter wasplaced retrogradely into a vein onthe dorsum of the hand, which was

then placed in a heated box (608C).Baseline arterialized venous bloodsamples for determination of plasma[3-3H]-glucose radioactivity, PG, insulin,C-peptide, glucagon, FFA, and sclerostinconcentrations were drawn at 230,220, 210, 25, and 0 min. At time 0,insulin was infused at 60 mU/kg z min.Arterialized blood samples were col-lected every 5 min for PG determina-tion, and a 20% glucose infusion wasadjusted to maintain the PG concen-tration at 100 mg/dL. Throughout theinsulin clamp, blood samples for de-termination of PG concentrationwere drawn every 10–15 min for de-termination of plasma insulin concen-tration and [3-3H]-glucose specificactivity.

DXA: DXA whole-body scan was per-formed to determine fat and leanbody mass and whole-body minushead bone mineral density (Hologic,Waltham, MA).

Calculationsb-Cell function was assessed from theOGTT using a model describing the rela-tionship between insulin secretion (ISR;expressed in pmol zmin21 zm22) and glu-cose concentration as the sum of twocomponents (13,14). The first compo-nent represents the dependence of ISRon glucose concentration through adose-response function relating thetwo variables. From the dose-response,b-cell glucose sensitivity (the slope) andISR at a fixed glucose concentration of5.2 mmol/L for NGT and 5.8 mmol/L forIGR were calculated. The dose-responseis modulated by a potentiation factor,accounting for various mechanisms (pro-longed hyperglycemia, nonglucose sub-strates, gastrointestinal hormones,neural modulation). The potentiationfactor averages 1 during the test and ex-presses relative potentiation or inhibi-tion of ISR; its excursion is quantifiedby the ratio between the 2-h and thebaseline value (potentiation ratio). Thesecond ISR component represents thedependence of ISR on the rate of changeof glucose concentration and is deter-mined by a single parameter (rate sensi-tivity), which is related to early insulinrelease. The model parameters wereestimated from glucose and C-peptideconcentrations (using C-peptide decon-volution [15]), as previously described(14). The integral of ISR during the whole

1510 Sclerostin and Insulin Resistance Diabetes Care Volume 38, August 2015

test (total ISR) was also calculated.Because a frequently sampled OGTTwas performed, a dynamic index of in-sulin sensitivity (SI) derived from mini-mal model analysis was calculated (16).Basal insulin clearance was calculated

as the ratio of insulin secretion and in-sulin concentration in the basal period;insulin clearance during the OGTT wasdetermined as the ratio of mean insulinsecretionand insulin concentrationduringthe OGTT. Under steady-state postab-sorptive conditions, the rate of endoge-nous glucose appearance was calculatedas the [3-3H]-glucose infusion rate(DPM/min) divided by the steady-stateplasma [3-3H]-glucose specific activity(DPM/mg). During the euglycemic insu-lin clamp, the rate of glucose appear-ance was calculated with the Steeleequation, using a distribution volumeof 250 mL/kg. Endogenous (primarilyreflects hepatic) glucose production(EGP) was calculated by subtractingthe exogenous glucose infusion ratefrom the rate of glucose appearance.The rate of insulin-mediated total-bodyglucose disposal (TGD/SSPI) was deter-mined by adding the rate of residualEGP to the exogenous glucose infusionrate and dividing it for the steady stateof plasma insulin during the last 30min ofthe clamp (17). Hepatic insulin resistanceindex (HIRI) was the product of fastingEGP and plasma insulin concentration.The adipose tissue IRI (ATIRI) was theproduct of fasting plasma FFA and fastingplasma insulin levels.

Biochemical and Hormone AnalysesCirculating sclerostin was measured induplicate using a commercially availableELISA (Biomedica, Vienna, Austria) ac-cording to the manufacturer’s instruc-tions. The intra-assay and interassayvariability was ,5%. Sclerostin mea-surements were reported in picomolesper liter, and the lower limit of detectionwas ,10 pmol/L. Plasma glucose levelswere measured using the glucose oxi-dase method (GM9; Analox Instru-ments, London, U.K.). Plasma insulinand C-peptide levels were measured byradioimmunoassay method (SiemensMedical Solutions Diagnostics, Tarrytown,NY). Plasma glucagon levels were mea-sured by a radioimmunoassay method(Linco Research). Plasma [3-3H]-glucoseradioactivitywasmeasured in Somogyi se-rum precipitates.

Statistical AnalysisValues were calculated as mean 6 SEMor as median (interquartile range) forvariables with a skewed distribution.The difference between means in NGTand IGR was compared with the two-sided unpaired t test and repeated-measures ANOVA, with time and groupas factors. To further explore the as-sociation between sclerostin and glu-cose metabolism alterations, we usedANOVA and post hoc analysis to alsoperform a subgroup analysis of IFG,IGT, and combined IFG-IGT subjects.Variables that were not normally distrib-uted were log-transformed before anal-ysis. A P value of ,0.05 (two-tailedanalysis) was considered to be statisti-cally significant. Pearson correlation(normal distribution) and Spearman cor-relation analyses (nonnormal distribution)were used to assess the correlations be-tween serum sclerostin and other contin-uous parameters, and we used partialcorrelations to correct the possible influ-ence of age, sex, BMI, and body fat onsclerostin values. Multivariate analyseswere performed to evaluate the contribu-tion of sclerostin in the association withinsulin resistance, insulin sensitivity, andinsulin clearance. Data were analyzed us-ing SPSS 20 software (IBM Corp, Armonk,NY).

RESULTS

Baseline Characteristics of the StudyPopulationThe clinical, biochemical, metabolic, anddensitometric data of NGT and IGR sub-jects are summarized in Table 1. The twostudy groups were comparable in age,sex, BMI, fat mass, lean body mass,and bone density. IGR subjects hadhigher FPG and 2-h PG, fasting insulin,and HbA1c (all P , 0.05). IGR subjectswere insulin resistant in skeletal muscle,liver, and adipose tissue and had lowerinsulin clearance at baseline and duringOGTT compared with NGT subjects (allP , 0.05). Also, as expected, IGR sub-jects showed a higher insulin secretionrate despite higher PG during the OGTTand a reduction of the glucose dose-response compared with NGT subjects(Supplementary Fig. 1A–C). A subgroupanalysis showed that subjects withisolated IFG or IGT or with combinedIFG-IGT had similar age, BMI, fat con-tent, and bone density although theymanifested typical glucose metabolism

alterations associated with isolated IFGor IGT or combined IFG-IGT (Supplemen-tary Table 1). Subjects with isolated IGTand combined IFG-IGT had a higher in-sulin secretion rate (P = 0.02), whereasb-cell glucose sensitivity was consistentlyreduced throughout the three subgroups(Supplementary Table 1). Sclerostin se-rum levels were higher in IGR than inNGT (50.8 6 2.4 vs. 38.7 6 2.3 pmol/L;P = 0.01). However, when subjects weredivided according to glucose tolerancestatus, isolated IFG and combined IFG-IGTmanifested significantly higher sclerostinserum levels compared with NGT (55.862.9 and 50.86 5.1 vs. 38.76 2.3 pmol/L;P = 0.02, respectively), but no differencewas found in sclerostin serum levels be-tween isolated IGT and NGT. When sub-jects were further divided according tosex or ethnicity, serum sclerostin levelswere similar in men and women in boththe NGT and IGR groups (P = 0.54 and P =0.45, respectively; data not shown), andno differences were found in subgroupanalysis.

Relationship of Serum SclerostinLevels With Insulin Sensitivity inMuscle, Liver, and Adipose TissueSerum sclerostin levels were positivelycorrelated with HOMA-insulin resis-tance (IR) (r = 0.62; P , 0.001) (Fig.1A) and negatively with insulin-mediatedtotal body glucose disposal as measuredby the hyperinsulinemic–euglycemicclamp (Ln TGD/SSPI: r = 20.40; P ,0.001) (Fig. 1B).Multiple linear regressionmodeling was performed using HOMA-IR,SI, and TGD/SSPI as dependent variables(Table 2). In model 1, independent var-iables were BMI, body fat, and HbA1c. Inmodel 2, sclerostin was added as anindependent variable to model 1 to as-sess the association of sclerostin withinsulin resistance. In model 1, BMI andHbA1c were independently associatedwith HOMA-IR (r2 = 0.192; P , 0.001).In model 2, sclerostin was indepen-dently associated with HOMA-IR and in-creased the r2 compared with model 1(r2 change: 0.055; F change: 28.893; P =0.001). Moreover, the SI index was sep-arately used as a dependent variable(model 1) to evaluate the associationof sclerostin with insulin sensitivity. Inmodel 1, BMI and 2-h PG were indepen-dently associated with SI (r2 = 0.293). Inmodel 2, sclerostin was independentlyassociated with SI and increased the r2

care.diabetesjournals.org Daniele and Associates 1511

(r2 change: 0.032; F change: 4.766; P =0.031).Similarly, a multiple linear regression

analysis was performed using TGD/SSPI,a direct measure of insulin sensitivityin humans, as a dependent variable. Inmodel 1, independent variables wereBMI, body fat, HbA1c, FPG, 2-h PG, andfasting plasma insulin. In model 2,

sclerostin was added as an independentvariable to model 1 to assess the contri-bution of sclerostin to the associationwithinsulin sensitivity. BMI and fasting plasmainsulin were independent predictors of in-sulin sensitivity (r2 = 0.518). Consistently,in model 2, sclerostin was strongly inde-pendently associated with insulin sensitiv-ity and increased the r2 (r2 change: 0.059;

F change: 4.938; P = 0.033). Fasting EGPand the HIRI were positively correlatedwith sclerostin serum levels (r = 0.48 andr = 0.62, respectively; P, 0.001 after cor-rection for age, sex, BMI, body fat, andFPG) (Fig. 1C and D). Moreover, the ATIRIwas positively correlated with serum scle-rostin levels (r = 0.61; P, 0.001) (Supple-mentary Fig. 2C).

Table 1—Characteristics of study participants in the whole population and the clamp subgroup

Whole population

Characteristic NGT (N = 43) IGR (N = 79) P value

Age (years) 44.0 6 1.9 46.0 6 1.4 0.42

Sex 0.54Men (n) 11 21Women (n) 32 58

Weight (kg) 82.5 6 3.6 84.5 6 3.2 0.61

BMI (kg/m2) 31.1 6 1.1 31.9 6 1.2 0.55

Fat content (%) 35.7 6 1.5 38.0 6 1.5 0.22

Lean mass (%) 64.3 6 1.5 62.0 6 1.6 0.21

Bone density (g/m2) 0.976 6 0.02 0.951 6 0.01 0.23

FPG (mg/dL) 94 6 1 105 6 1 ,0.001

2-h PG (mg/dL) 109 6 1 150 6 4 ,0.001

HbA1c (%) (mmol/mol) 5.60 6 0.04 (38 6 0.4) 5.70 6 0.04 (39 6 0.4) 0.03

Fasting plasma insulin (mU/L) 5.5 6 0.6 7.8 6 0.6 0.01

FFA (mmol/L) 0.49 6 0.04 0.52 6 0.04 0.59

Matsuda Index (mmol z min21 z kg21) 8.8 6 1.3 4.5 6 1.3 ,0.001

OGIS (mL z min21 z m22) 402 6 8 346 6 5 ,0.001

HOMA-IR (mmol/L z mU/L) 1.3 6 0.1 2.0 6 0.2 ,0.001

Fasting insulin clearance (L z min21 z m22) 4.8 6 0.7 3.5 6 0.8 0.02

OGTT insulin clearance (L z min21 z m22) 2.5 6 0.5 1.7 6 0.5 0.04

ATIRI (mmol/L z mU/L) 2.6 6 0.3 4.4 6 0.3 ,0.001

Sclerostin (pmol/L) 38.7 6 2.3 50.8 6 2.4 0.01

Clamp subgroup analysis

NGT (n= 18) IGR (n= 30) P value

Age (years) 43.2 6 2.8 47.1 6 2.2 0.29

Sex 0.56Men (n) 6 9Women (n) 12 21

Weight (kg) 80.5 6 6.2 89.5 6 2.9 0.15

BMI (kg/m2) 29.6 6 1.4 32.4 6 1.0 0.11

Fat content (%) 36.9 6 2.2 36.8 6 1.6 0.95

Lean mass (%) 63.0 6 2.1 63.2 6 1.6 0.94

FPG (mg/dL) 94 6 1 106 6 1 ,0.001

2-h PG (mg/dL) 112 6 4 145 6 5 ,0.001

HbA1c (%) (mmol/mol) 5.5 6 0.1 (37 6 1.1) 5.7 6 0.1 (39 6 1.1) 0.19

Fasting plasma insulin (mU/L) 4.7 6 0.6 8.6 6 1.0 0.01

FFA (mmol/L) 0.49 6 0.04 0.52 6 0.04 0.59

HOMA-IR (mmol/L z mU/L) 1.1 6 0.1 2.3 6 0.3 ,0.001

TGD/SSPI (mg z kg21 z min21/mU z L21) 8.4 6 0.9 5.9 6 0.6 0.02

Fasting EGP (mg z kg21 z min21) 2.0 6 0.1 2.1 6 0.2 0.72

Clamp EGP (mg z kg21 z min21) 0.4 6 0.2 1.1 6 0.5 0.03

HIRI [(mg z kg21 z min21) z (mU/L)] 9.1 6 1.1 14.9 6 1.1 ,0.001

Sclerostin (pmol/L) 39.2 6 3.7 58.1 6 6.1 0.03

Data are expressed as mean 6 SEM unless indicated otherwise.

1512 Sclerostin and Insulin Resistance Diabetes Care Volume 38, August 2015

Insulin clearances at baseline (Ln fast-ing insulin clearance) andduringOGTT (LnOGTT insulin clearance) were inversely

correlated with sclerostin serum levels(r = 20.52 and r = 20.44, respectively;both P , 0.001 after correction for age,

sex, BMI, and TGD/SSPI) (Fig. 1E and F).To further examine the correlation be-tween sclerostin serum levels and insulin

Figure 1—Partial correlation between serum sclerostin levels controlled for age, sex, BMI, body fat percentage, and HOMA-IR (A), LnTGD/SSPI (B),fasting EGP (C), hepatic insulin resistance (D), Ln fasting insulin clearance (E), and Ln OGTT insulin clearance (F).

care.diabetesjournals.org Daniele and Associates 1513

clearance independently from whole-body insulin sensitivity, we performeda multiple linear regression analysisin a model including age, sex, BMI,and TGD/SSPI as independent varia-bles (Supplementary Table 2). Whenfasting insulin clearance was usedas a dependent variable, TGD/SSPI re-mained independently associated withfasting insulin clearance (model 1) (r2 =0.245).In model 2, sclerostin serum levels

were independently associated withfasting insulin clearance and increasedr2 (r2 change: 0.183; F change: 13.477;P = 0.001). Similarly, when OGTT insulinclearance was used as a dependent var-iable, TGD/SSPI remained indepen-dently associated with OGTT insulinclearance (model 1), and in model 2,sclerostin serum levels were indepen-dently associated with OGTT insulinclearance and increased the r2 (r2

change: 0.120; F change: 7.445; P =0.009). Sclerostin levels were also neg-atively correlated with IRIs derivedfrom OGTT as OGIS and the Ln Matsuda

Index, and positively with ATIRI (r =20.38, r = 20.42, and r = 0.61, respec-tively; all P , 0.001) (SupplementaryFig. 2A–C).

Serum sclerostin levels were posi-tively correlated with the fasting insulinsecretory rate (fasting ISR) (r = 0.19; P =0.04), although no correlation was ob-served with the total ISR, b-cell glucosesensitivity, potentiation factor, and ratesensitivity (data not shown). Interest-ingly, when subjects were further dividedaccording to their glucose tolerance sta-tus, those with isolated IFG were mostlyaccountable for the correlations betweenserum sclerostin levels and insulin sensi-tivity in skeletal muscle (r = 0.63; P, 0.01for HOMA-IR; r = 20.55; P = 0.03 for LnMatsuda Index; r = 20.64; P , 0.01 forOGIS; r =20.57; P = 0.05 for Ln TGD/SSPI),liver (r = 0.61; P = 0.03 for fasting EGP),and adipose tissue (r = 0.71; P , 0.01for ATIRI) (Supplementary Table 3). Sim-ilarly, in subjects with isolated IFG, se-rum sclerostin levels were stronglycorrelated with fasting insulin clearance(r = 20.44; P = 0.02), and a significant

correlation between serum sclerostinlevels and OGTT insulin clearancewas found in subjects with combinedIFG-IGT (r =20.37; P = 0.03) (Supplemen-tary Table 3).

Relationship of Serum SclerostinLevels With Glucose LevelsSerum sclerostin levels were positivelycorrelated with FPG (r = 0.22; P = 0.01)(Fig. 2A), fasting and 2-h plasma insulin(r = 0.63 and r = 0.45, respectively; P,0.001) (Fig. 2B and C), and HbA1c (r =0.18; P = 0.04) (Fig. 2D). No correla-tions were found between serum scle-rostin levels and plasma glucagonlevels at fasting and during OGTT (P =0.56 and P = 0.67, respectively; datanot shown). When subjects were fur-ther divided according to the glucosetolerance status, subjects with isolatedIFG and combined IFG-IGT were ac-countable for the correlation betweenserum sclerostin levels and fastingplasma insulin (r = 0.6; P , 0.01 andr = 0.52; P , 0.01, respectively) and2-h plasma insulin (r = 0.72; P , 0.01

Table 2—Multivariate analysis with HOMA-IR (whole population), SI (whole population), and TGD/SSPI (clamp subgroup) asdependent variables

Multivariate analysis r2 (P value) r2 change F (P value) F change (P value) b P value

HOMA-IR (mmol/L z mU/L)Independent variablesModel 1 0.192 (,0.001) 7.430 (,0.001)

Body fat (%) 0.041 0.643BMI (kg/m2) 0.374 ,0.0012-h PG (mg/dL) 0.065 0.491HbA1c (%) 0.208 0.026

Model 2 0.247 (,0.001) 0.055 13.291 (,0.001) 28.893 (0.001)Sclerostin (pmol/L) 0.423 ,0.001

SI (mL z min21 z kg21/pmol z L21)Independent variablesModel 1 0.293 (,0.001) 9.810 (,0.001)

Body fat (%) 20.148 0.084BMI (kg/m2) 20.290 0.0012-h PG (mg/dL) 20.321 0.001HbA1c (%) 20.103 0.249

Independent variableModel 2 0.325 (,0.001) 0.032 10.681 (,0.001) 4.766 (0.031)

Sclerostin (pmol/L) 20.182 0.031

TGD/SSPI (mg z kg21 z min21/mU z L21)Independent variablesModel 1 0.518 (,0.001) 6.447 (,0.001)

Body fat (%) 20.193 0.126BMI (kg/m2) 20.367 0.010FPG (mg/dL) 20.480 0.7462-h PG (mg/dL) 20.129 0.352FPI (mU/L) 20.318 0.022HbA1c (%) 20.039 0.765

Model 2 0.577 (,0.001) 0.059 6.830 (,0.001) 4.938 (0.033)Sclerostin (pmol/L) 0.320 0.030

1514 Sclerostin and Insulin Resistance Diabetes Care Volume 38, August 2015

and r = 0.29; P = 0.08, respectively)(Supplementary Table 3).

CONCLUSIONS

This study shows that serum sclerostin lev-els are increased in subjects with prediabe-tes and strongly correlated with plasmainsulin and insulin resistance derived fromOGTT and euglycemic–hyperinsulinemicclamp. Our results suggest, although donot prove a direct causality, that in earlystages of glucose intolerance and insulin re-sistance, sclerostin might play a role in in-sulin action and clearance in humans.Sclerostinmighthavea role inWnt signalingin hepatic glucose metabolism by actingthrough the same pathways. Interestingly,we show that fasting EGP and the HIRI arestrongly correlated with serum sclerostinlevels. A recent study demonstrated thatb-catenin, the downstreammediator of ca-nonical Wnt signaling, modulates hepaticinsulin signaling and glucose production(18). Also, a cross-sectional study including

patients with liver cirrhosis found serumsclerostin levels were increased comparedwith control subjects and correlated withmarkers of liver dysfunction (19), sug-gesting a possible role of sclerostin in livermetabolism.

Interestingly, we observed a strongcorrelation between serum sclerostinlevels and insulin clearance at fastingand during the OGTT, which was de-creased in IGR individuals. This is arather novel observation obtained byusing an established method for mea-suring insulin clearance in humans(20,21).

Insulin clearance by the liver is an im-portant component of the insulin deg-radation pathway, which has beenrecently suggested as a potential targetfor T2DM treatment (22). The liver re-moves ;80% of endogenous insulin,and the remainder is cleared by the kid-neys and muscles (23). However, clear-ance rates for insulin decrease in

glucose intolerance (24), obesity, ab-dominal obesity in particular (25), hy-pertension (26), hepatic cirrhosis, andnonalcoholic fatty liver disease (27),implicating a possible link between in-sulin resistance and reduced insulinclearance (28). Reduced insulin clear-ance has important implications in thepathophysiology of T2DM and IGR. Forexample, animal models have shownthat decreased insulin clearance servesas a compensatory mechanism to pre-serve b-cell function and maintain pe-ripheral insulin levels in the states ofinsulin resistance (29). It is temptingto hypothesize that sclerostin mighthave a role in modulating insulin clear-ance by acting on the Wnt signalingpathway in the liver.

IFG and IGT subjects have distinctpathophysiological abnormalities. Onone hand, IFG subjects manifest hepaticinsulin resistance associated with nor-mal insulin-mediated glucose disposal

Figure 2—Partial correlation between serum sclerostin levels controlled for age, sex, BMI, body fat percentage, and FPG (A), fasting plasma insulin(B), 2-h plasma insulin (C), and HbA1c (D) in NGT and IGR.

care.diabetesjournals.org Daniele and Associates 1515

(reflecting mainly skeletal muscle glu-cose uptake). On the other hand, IGTsubjects conversely manifest almostnormal hepatic insulin sensitivity withreduced insulin-mediated glucose dis-posal (3). This is the potential involve-ment in the modulation of the insulinclearance, which occurs prevalently al-though not exclusively in the liver, fur-ther reinforced by the findings thatsclerostin levels are significantly higherin the subgroups of patients with IFGand combined IFG-IGT in which insulinresistance in the liver is present.Serum sclerostin levels positively cor-

related with FPG and 2-h PG and HbA1c.In a small recent study evaluating agroup of healthy offspring of patientswith T2DM, serum sclerostin levelswere consistently correlated with FPG,although they were not correlated with2-h PG (30). Another study demon-strated that compared with healthy con-trols, circulating levels of sclerostinwere increased in patients with T2DM,positively associated with duration ofT2DM and glycated hemoglobin, andinversely related to bone turnovermarkers (31,32). Similarly, serum sclero-stin levels were elevated in subjectswith type 1 diabetes compared withnormal control subjects and correlatedwith HbA1c (33). Hyperglycemia has adetrimental effect on bone health, act-ing directly on bone cells and indirectlythrough the formation of advanced gly-cation end products that have beenshown to reduce bone strength and in-crease the risk of falls and fractures.Moreover, hyperglycemia is directly in-volved in T2DMmicrovascular complica-tions and accelerated atherosclerosis.Circulating sclerostin is further in-creased in patients with T2DM with ath-erosclerotic lesions, such as abnormalintima-media thickness, carotid pla-ques, and aortic calcification, comparedwith patients with T2DM without ath-erosclerotic lesions (34). Therefore, ourresults suggest that a possible detrimen-tal role of sclerostin on bone metabo-lism and atherosclerosis might beginearly in the pathogenesis of glucose me-tabolism alterations, at the stage of IGR.A recent study showed that serum scle-rostin levels were correlated withHOMA-IR but not with HOMA–b-cellfunction in healthy offspring of T2DMpatients (30). In patients with T2DM,the increase of serum sclerostin is

associated with a significant decreasein the b-catenin serum concentration,suggesting, indeed, that sclerostin sig-naling is modulated in diabetes (31).On one hand, this gives support to thehypothesis that bone, which producesalmost all the available sclerostin, mayplay a role also in glucose metabolismand insulin resistance (8,32,35). On theother hand, it can be speculated thatanti-sclerostin antibody treatment,which is a promising option for severeosteoporosis and surgical implant fixa-tion, might have a role in T2DM treat-ment (36).

In subcutaneous and visceral whiteadipose tissue from lean and obese sub-jects, the Wnt pathway was altered inobese subjects, and its inhibition wasassociated with insulin resistance (37).Therefore, sclerostin-mediated inhibi-tion of Wnt signaling might directlycause adipogenesis and adipose tissueinsulin resistance, which are present inT2DM and in prediabetes (38). Giventhat obesity is a major risk factor for de-veloping T2DM, changes in componentsof theWnt signaling pathway could con-ceivably contribute to the increased riskof diabetes development by affectingadipogenesis.

Although a strong correlation be-tween serum sclerostin levels, bonedensity, and body fat was previouslydemonstrated in healthy subjects andin patients with T2DM (31,39,40), ourresults could not confirm these findings.Our study is limited by the sole mea-surement of whole-body minus headbone density and not the individual sitessuch as the spine, hip, and forearm. Ourstudy has other limitations. First, thecross-sectional design does not allowthe study of potential causes of changesin sclerostin levels during the develop-ment of IGR, thus providing a strongerlink between insulin resistance and in-creased sclerostin levels. Second, thesample size is relatively small and mightaffect the statistical power, althoughthe study groups were well matchedfor age, sex, BMI, and body composition,including bone density.

Strengths of our study are the evalu-ation for the first time of circulating se-rum sclerostin in subjects with NGT andIGR and the extensive and detailed evalu-ation of biochemical and clinical parame-ters of glucose metabolism, derived fromOGTT and the gold standard, euglycemic

clamp with glucose tracer to evaluate he-patic glucose production.

In summary, sclerostin circulates inhigher amounts in individuals with pre-diabetes, particularly in IFG and IFG-IGT,and is strongly correlated with whole-body insulin sensitivity and EGP as wellas insulin clearance, supporting the hy-pothesis that sclerostin action is impor-tant not only in the regulation of boneformation and resorption but also in theregulation of glucose metabolism in hu-mans. Further studies are needed toclarify the mechanisms by which sclero-stin could affect insulin action in vivoand to evaluate the role of sclerostin inthe natural history of glucose intoler-ance from prediabetes to T2DM.

Acknowledgments. The authors thank JohnKincade and James King, University of TexasHealth Science Center at San Antonio, TX, forprecious help with the euglycemic clamp. Theauthors thank Prof. Ralph A. DeFronzo, Univer-sity of Texas Health Science Center at SanAntonio, TX, for his continuous and generoussupport. F.F. acted as a Visiting Professor at theObesity and Comorbidities Research Center,Universidade Estadual de Campinas, Campinas,Brazil. F.F. dedicates this article to his belovednephew, Matteo Folli (17 April 1995–2 April2015), who left us in the prime time of his life.Funding. F.F. received grants from the SãoPaulo Research Foundation.Duality of Interest. No potential conflicts ofinterest relevant to this article were reported.Author Contributions. G.D. researched andanalyzed data, conceived the study, and wrotethe manuscript. D.W. researched and analyzeddata. A.M. researched data, contributed to thediscussion, and reviewed the final version of themanuscript. J.B. contributed to the discussionand reviewed the final version of the manu-script. M.F., Z.P., D.T., and C.J. researched data.F.F. conceived the study, researched, analyzed,and interpreted the data, wrote themanuscript,contributed to the discussion, and reviewed thefinal version of the manuscript. All authors gavefinal approval of the version of the manuscriptto be published. F.F. is the guarantor of thiswork and, as such, had full access to all the datain the study and takes responsibility for theintegrity of the data and the accuracy of thedata analysis.

References1. Dall TM, Yang W, Halder P, et al. The eco-nomic burden of elevated blood glucose levelsin 2012: diagnosed and undiagnosed diabetes,gestational diabetes mellitus, and prediabetes.Diabetes Care 2014;37:3172–31792. Garber AJ, Handelsman Y, Einhorn D, et al.Diagnosis and management of prediabetes inthe continuum of hyperglycemia: when do therisks of diabetes begin? A consensus statementfrom the American College of Endocrinology

1516 Sclerostin and Insulin Resistance Diabetes Care Volume 38, August 2015

and the American Association of Clinical En-docrinologists. Endocr Pract 2008;14:933–9463. Abdul-Ghani MA, Jenkinson CP, RichardsonDK, Tripathy D, DeFronzo RA. Insulin secretionand action in subjects with impaired fastingglucose and impaired glucose tolerance: re-sults from the Veterans Administration GeneticEpidemiology Study. Diabetes 2006;55:1430–14354. Velloso LA, Folli F, Sun XJ, White MF, SaadMJ, Kahn CR. Cross-talk between the insulin andangiotensin signaling systems. Proc Natl AcadSci U S A 1996;93:12490–124955. Ferrannini E, Buzzigoli G, Bonadonna R, et al.Insulin resistance in essential hypertension. NEngl J Med 1987;317:350–3576. Fadini GP, Ceolotto G, Pagnin E, deKreutzenberg S, Avogaro A. At the crossroadsof longevity andmetabolism: themetabolic syn-drome and lifespan determinant pathways. Ag-ing Cell 2011;10:10–177. Plantinga LC, Crews DC, Coresh J, et al.; CDCCKD Surveillance Team. Prevalence of chronickidney disease in US adults with undiagnoseddiabetes or prediabetes. Clin J Am Soc Nephrol2010;5:673–6828. Wei J, Ferron M, Clarke CJ, et al. Bone-specificinsulin resistance disrupts whole-body glucosehomeostasis via decreasedosteocalcin activation.J Clin Invest 2014;124:1–139. Winkler DG, Sutherland MK, Geoghegan JC,et al. Osteocyte control of bone formation viasclerostin, a novel BMP antagonist. EMBO J2003;22:6267–627610. Singh R, De Aguiar RB, Naik S, et al. LRP6enhances glucose metabolism by promotingTCF7L2-dependent insulin receptor expressionand IGF receptor stabilization in humans. CellMetab 2013;17:197–20911. Mani A, Radhakrishnan J, Wang H, et al.LRP6 mutation in a family with early coronarydisease and metabolic risk factors. Science2007;315:1278–128212. American Diabetes Association. Standardsof medical care in diabetesd2014. DiabetesCare 2014;37(Suppl. 1):S14–S8013. Mari A, Tura A, Gastaldelli A, Ferrannini E.Assessing insulin secretion by modeling inmultiple-meal tests: role of potentiation. Diabe-tes 2002;51(Suppl. 1):S221–S22614. Mari A, Schmitz O, Gastaldelli A,Oestergaard T, Nyholm B, Ferrannini E. Mealand oral glucose tests for assessment of beta-cellfunction: modeling analysis in normal subjects.Am J Physiol Endocrinol Metab 2002;283:E1159–E116615. Van Cauter E, Mestrez F, Sturis J, PolonskyKS. Estimation of insulin secretion rates from

C-peptide levels. Comparison of individual andstandard kinetic parameters for C-peptide clear-ance. Diabetes 1992;41:368–37716. Breda E, Cavaghan MK, Toffolo G, PolonskyKS, Cobelli C. Oral glucose tolerance test mini-mal model indexes of beta-cell function and in-sulin sensitivity. Diabetes 2001;50:150–15817. Steele R. Influences of glucose loading andof injected insulin on hepatic glucose output.Ann N Y Acad Sci 1959;82:420–43018. Liu H, Fergusson MM, Wu JJ, et al. Wntsignaling regulates hepatic metabolism. Sci Sig-nal 2011;4:ra619. Rhee Y, Kim WJ, Han KJ, Lim SK, Kim SH.Effect of liver dysfunction on circulating sclero-stin. J Bone Miner Metab 2014;32:545–54920. Polonsky KS, Given BD, Hirsch L, et al. Quan-titative study of insulin secretion and clearancein normal and obese subjects. J Clin Invest 1988;81:435–44121. Tillil H, Shapiro ET, Miller MA, et al. Dose-dependent effects of oral and intravenous glu-cose on insulin secretion and clearance in normalhumans. Am J Physiol 1988;254:E349–E35722. Maianti JP, McFedries A, Foda ZH, et al.Anti-diabetic activity of insulin-degrading en-zyme inhibitors mediated by multiple hor-mones. Nature 2014;511:94–9823. Duckworth WC, Bennett RG, Hamel FG. In-sulin degradation: progress and potential. En-docr Rev 1998;19:608–62424. Bonora E, Zavaroni I, Coscelli C, Butturini U.Decreased hepatic insulin extraction in subjectswith mild glucose intolerance. Metabolism1983;32:438–44625. Peiris AN,Mueller RA, Smith GA, StruveMF,Kissebah AH. Splanchnic insulin metabolism inobesity. Influence of body fat distribution. J ClinInvest 1986;78:1648–165726. Salvatore T, Cozzolino D, Giunta R, GiuglianoD, Torella R, D’Onofrio F. Decreased insulin clear-ance as a feature of essential hypertension. J ClinEndocrinol Metab 1992;74:144–14927. Goto T, Onuma T, Takebe K, Kral JG. The in-fluence of fatty liver on insulin clearance and in-sulin resistance in non-diabetic Japanese subjects.Int J Obes Relat Metab Disord 1995;19:841–84528. Jones CN, Abbasi F, Carantoni M, PolonskyKS, Reaven GM. Roles of insulin resistance andobesity in regulation of plasma insulin concen-trations. Am J Physiol Endocrinol Metab 2000;278:E501–E50829. Mittelman SD, Van Citters GW, Kim SP, et al.Longitudinal compensation for fat-induced insu-lin resistance includes reduced insulin clearanceand enhanced beta-cell response. Diabetes 2000;49:2116–212530. Ebenibo S, Edeoga C, Crisler J, Dagogo-JackS. Plasma FGF-21 and sclerostin levels predict

glycemia and insulin sensitivity in healthy sub-jects with parental type 2 diabetes. Presented atthe Determinants of Insulin Resistance & Asso-ciated Metabolic Disturbance session of Endo2013, 15-18 June 2013, San Francisco, Calif. Ab-stract SUN-78031. Gaudio A, Privitera F, Battaglia K, et al. Scle-rostin levels associated with inhibition of theWnt/b-catenin signaling and reduced boneturnover in type 2 diabetes mellitus. J Clin En-docrinol Metab 2012;97:3744–375032. Garcıa-Martın A, Rozas-Moreno P, Reyes-Garcıa R, et al. Circulating levels of sclerostinare increased in patients with type 2 diabetesmellitus. J Clin Endocrinol Metab 2012;97:234–24133. Neumann T, Hofbauer LC, Rauner M, et al.Clinical and endocrine correlates of circulatingsclerostin levels in patients with type 1 diabe-tes mellitus. Clin Endocrinol (Oxf) 2014;80:649–65534. Morales-Santana S, Garc ıa-Fontana B,Garcıa-Martın A, et al. Atherosclerotic diseasein type 2 diabetes is associated with an increasein sclerostin levels. Diabetes Care 2013;36:1667–167435. van Lierop AH, Hamdy NA, van der MeerRW, et al. Distinct effects of pioglitazone andmetformin on circulating sclerostin and bio-chemical markers of bone turnover in menwith type 2 diabetes mellitus. Eur J Endocrinol2012;166:711–71636. Virdi AS, Irish J, Sena K, et al. Sclerostinantibody treatment improves implant fixationin a model of severe osteoporosis. J Bone JointSurg Am 2015;97:133–14037. Ehrlund A, Mejhert N, Lorente-Cebrian S,et al. Characterization of the Wnt inhibitors se-creted frizzled-related proteins (SFRPs) in hu-man adipose tissue. J Clin Endocrinol Metab2013;98:E503–E50838. Tripathy D, Daniele G, Fiorentino TV, et al.Pioglitazone improves glucose metabolismand modulates skeletal muscle TIMP-3-TACEdyad in type 2 diabetes mellitus: a rando-mised, double-blind, placebo-controlled,mechanistic study. Diabetologia 2013;56:2153–216339. Urano T, Shiraki M, Ouchi Y, Inoue S. As-sociation of circulating sclerostin levels withfat mass and metabolic diseasedrelatedmarkers in Japanese postmenopausal women.J Clin Endocrinol Metab 2012;97:E1473–E147740. Amrein K, Dobnig H, Wagner D, et al.Sclerostin in institutionalized elderly women: as-sociations with quantitative bone ultrasound,bone turnover, fractures, and mortality. J AmGeriatr Soc 2014;62:1023–1029

care.diabetesjournals.org Daniele and Associates 1517