nejmoa032782

n engl j med

352;4

www.nejm.org january

27, 2005

The

new england journal

of

medicine

341

original article

Vascular Risk Factors and Diabetic Neuropathy

Solomon Tesfaye, M.D., Nish Chaturvedi, M.D., Simon E.M. Eaton, D.M., John D. Ward, M.D., Christos Manes, M.D., Constantin Ionescu-Tirgoviste, M.D.,

Daniel R. Witte, Ph.D., and John H. Fuller, M.A., for the EURODIAB Prospective Complications Study Group*

From the Diabetes Research Unit, RoyalHallamshire Hospital, Sheffield, UnitedKingdom (S.T., S.E.M.E., J.D.W.); the De-partment of Epidemiology and PublicHealth, Faculty of Medicine, Imperial Col-lege of Science, Technology and Medicine,London (N.C.); the Department of Inter-nal Medicine and Diabetic Center, Papa-georgiou General Teaching Hospital,Thessaloniki, Greece (C.M.); the Instituteof Diabetes, Nutrition, and Metabolic Dis-eases, Bucharest, Romania (C.I.-T.); andEURODIAB, Department of Epidemiologyand Public Health, Royal Free and Univer-sity College Medical School, London(D.R.W., J.H.F.). Address reprint requeststo Dr. Tesfaye at the Diabetes ResearchUnit, Royal Hallamshire Hospital, SheffieldS10 2JF, United Kingdom, or at [email protected].

*The investigators in the European Diabe-tes (EURODIAB) Prospective Complica-tions Study Group are listed in the Ap-pendix.

N Engl J Med 2005;352:341-50.

Copyright 2005 Massachusetts Medical Society.

background

Other than glycemic control, there are no treatments for diabetic neuropathy. Thus,identifying potentially modifiable risk factors for neuropathy is crucial. We studiedrisk factors for the development of distal symmetric neuropathy in 1172 patientswith type 1 diabetes mellitus from 31 centers participating in the European Diabetes(EURODIAB) Prospective Complications Study.

methods

Neuropathy was assessed at baseline (1989 to 1991) and at follow-up (1997 to 1999),with a mean (SD) follow-up of 7.30.6 years. A standardized protocol included clini-cal evaluation, quantitative sensory testing, and autonomic-function tests. Serum lipidsand lipoproteins, glycosylated hemoglobin, and the urinary albumin excretion rate weremeasured in a central laboratory.

results

At follow-up, neuropathy had developed in 276 of 1172 patients without neuropathy atbaseline (23.5 percent). The cumulative incidence of neuropathy was related to the gly-cosylated hemoglobin value and the duration of diabetes. After adjustment for thesefactors, we found that higher levels of total and low-density lipoprotein cholesterol andtriglycerides, a higher body-mass index, higher von Willebrand factor levels and uri-nary albumin excretion rate, hypertension, and smoking were all significantly associat-ed with the cumulative incidence of neuropathy. After adjustment for other risk factorsand diabetic complications, we found that duration of diabetes, current glycosylated he-moglobin value, change in glycosylated hemoglobin value during the follow-up period,body-mass index, and smoking remained independently associated with the incidenceof neuropathy. Cardiovascular disease at baseline was associated with double the riskof neuropathy, independent of cardiovascular risk factors.

conclusions

This prospective study indicates that, apart from glycemic control, the incidence ofneuropathy is associated with potentially modifiable cardiovascular risk factors, includ-ing a raised triglyceride level, body-mass index, smoking, and hypertension.

abstract

The New England Journal of Medicine Downloaded from nejm.org on January 27, 2015. For personal use only. No other uses without permission.

Copyright 2005 Massachusetts Medical Society. All rights reserved.

n engl j med

352;4


27

,

2005

The

new england journal

of

medicine

342

iabetic neuropathy is a common

cause of morbidity and death among pa-tients with diabetes, generating a huge

economic burden.

1

Apart from tight glycemic con-trol, no other evidence-based treatments are knownto ameliorate or prevent neuropathy.

The duration and level of hyperglycemia are im-portant determinants of microvascular complica-tions of diabetes, including neuropathy.

2

The rela-tive effect of cardiovascular risk factors specificallyassociated with diabetes (e.g., hypertension, dys-lipidemia, and increased weight) or not associatedwith diabetes (smoking) on the development ofneuropathy are incompletely elucidated. The Dia-betes Control and Complications Trial (DCCT) re-ported a 60 percent reduction in neuropathy in theintensively treated groups after five years,

2

but thecumulative incidence of neuropathy (15 to 21 per-cent) and abnormal nerve conduction (40 to 52 per-cent) remained substantial. Such findings suggestthat neuropathy can develop, despite intensivecontrol of the glucose level.

2

Thus, risk factors be-sides hyperglycemia are probably involved in theevolution of neuropathy. Identifying them, partic-ularly if they are modifiable, might lead to newrisk-reduction strategies. The European Diabetes(EURODIAB) Prospective Complications Studyaimed to define and assess the relative importanceof other, potentially modifiable factors that increasethe risk of distal symmetric neuropathy in patientswith type 1 diabetes mellitus.

patients and baseline investigations

The baseline examination included 3250 patients(1668 men and 1582 women; mean [SD] age,32.710.2 years; mean duration of diabetes, 14.79.3 years).

3

These patients were randomly selectedin a stratified manner from 31 diabetes clinicsacross Europe. Selection criteria and methods havebeen described previously.

3

Written informed con-sent was obtained from all patients, and the insti-tutional review board at each center approved thestudy. Trained physicians performed all measure-ments with standardized procedures and equip-ment according to a predefined protocol. Baselineexaminations were conducted from 1989 to 1991,and follow-up examinations from 1997 to 1999.

Baseline blood samples, obtained after an over-night fast, if possible, were sent to central laborato-ries. Measurements included total cholesterol, high-

density lipoprotein (HDL) cholesterol, and triglyc-erides.

4-6

Levels of low-density lipoprotein (LDL)cholesterol were calculated.

7

The reference rangefor glycosylated hemoglobin was 2.9 to 4.8 per-cent.

8

In order to compare the results with those ofthe DCCT, measured glycosylated hemoglobin val-ues were converted as previously reported (DCCTglycosylated hemoglobin=[1.0289the glycosylat-ed hemoglobin measurements for this study]+1.5263).

9

DCCT-converted glycosylated hemoglo-bin values were used in all further analyses.

Glycosylated hemoglobin values, measured cen-trally according to the follow-up protocol, wereavailable for 1119 of 1172 patients (95.5 percent).Additional glycosylated hemoglobin values, deter-mined at one to eight (median, five) routine inter-im visits (measured and standardized according tocenter) between baseline and the final follow-upvisit, were available for 794 participants (67.7 per-cent). All available glycosylated hemoglobin mea-surements were included in estimates of the changein glycosylated hemoglobin during follow-up. Base-line and final follow-up glycosylated hemoglobinvalues were standardized for the entire group. Lin-ear regression for each patient provided the changein glycosylated hemoglobin per patient per year offollow-up and was included in the analyses to as-sess a possible association with neuropathy beyondthe baseline glycosylated hemoglobin value.

assessment and definition of diabetic complications

The urinary albumin excretion rate was measuredfrom a single 24-hour urine collection.

10

A urinaryalbumin excretion rate of 20 to 200

g per minutewas defined as microalbuminuria; a rate greaterthan 200

g per minute was defined as macroalbu-minuria.

The presence and severity of diabetic retinopathywere assessed from retinal photographs (two fieldsper eye) obtained with a wide-angle camera andscored centrally by one expert in reading changes ofretinopathy.

11

Retinopathy was classified as non-proliferative (background) or proliferative.

Cardiovascular disease was defined as either ahistory of physician-diagnosed cardiovascular dis-ease (e.g., previous myocardial infarction, angina,coronary-artery bypass grafting, or stroke) or is-chemic changes detected on a 12-lead electrocar-diogram (classified by two observers according tothe Minnesota code).

12

To assess the role of insulin resistance, an esti-

d

methods



n engl j med

352;4


27, 2005

vascular risk factors and diabetic neuropathy

343

mated glucose disposal rate was calculated on thebasis of the waist-to-hip ratio, the presence of hy-pertension, and the glycosylated hemoglobin val-ue.

13

Fibrinogen and von Willebrand factor werealso measured.

14

assessment of distal symmetric polyneuropathy

Physicians experienced in taking a neurologic his-tory assessed patients for distal symmetric polyneu-ropathy. Patients with features suggesting otherforms of diabetic neuropathy or polyneuropathy dueto causes other than diabetes were excluded.

symptoms and signs of neuropathy

The presence or absence of the following symp-toms within the previous six months was ascer-tained: asleep numbness or dead feeling in thefeet, a prickling sensation in the feet, deep achingor burning pains in the legs, unusual difficulty inclimbing stairs, difficulty with bladder control, andnocturnal diarrhea. Symptoms were evaluated care-fully to rule out other causes. In addition, reflexes ofthe ankle and knee tendons were assessed, with re-inforcement if necessary.

perception of vibration

The threshold of perception of vibration was mea-sured by centrally calibrated biothesiometers (Bio-medical).

15

Three readings on the right big toe andright medial malleolus were obtained and averagedfor the analysis.

autonomic function

Autonomic function was assessed according to twocardiovascular reflex responses after the patientshad rested in a supine position for at least five min-utes a change in heart rate and a change in sys-tolic blood pressure on standing, as determinedwith a Hawksley random-zero sphygmomanome-ter. The RR ratio was calculated by one observer.The RR ratio is the ratio of the longest electrocar-diographic RR interval between the 28th and 32ndbeats after standing to the shortest interval be-tween the 13th and 17th beats.

definition of neuropathy

Neuropathy was diagnosed in patients with two ormore of the following four measures: the presenceof one or more symptoms, the absence of two ormore reflexes of the ankle or knee tendons, a vibra-tion-perception threshold that was abnormal for

the patients age,

15

and abnormal autonomic func-tion (loss of heart rate variability with an RR ratio ofless than 1.04, postural hypotension with a fall insystolic blood pressure of 20 mm Hg or more, orboth).

16

Pure peripheral neuropathy was definedas distal neuropathy without autonomic symptomsor abnormal autonomic-function tests.

The cutoff points and the definition of neuropa-thy were established before any examination of theoutcome data. The four criteria were evaluated atthe same times (the day of the baseline visit and theday of the follow-up visit) for each patient.

Figure 1. Patients Analyzed for Progression to Neuropathy in the EURODIAB Study.

Initial and follow-up assessments were available for 1272 of 3250 patients. Of these patients, 100 did not have the neuropathy component assessed at follow-up, yielding a final study population of 1172 patients.

3250 Patients

28 Died508 Were lost to follow-up

11 Had an unknownstatus of life or death

1819 Qualified for follow-up

1272 Reached follow-up

100 Were not assessedfor neuropathy

1172 Were assessedfor neuropathy

927 Had neuropathyat baseline

59 Were not assessed forneuropathy at baseline

437 Were at a center thatdid not participate in the follow-up

8 Did not meet entrycriteria



n engl j med

352;4


27

,

2005

The

new england journal

of

medicine

344

Of 3250 patients examined, 927 had neuropathyat baseline,

17

whereas 1172 patients without neu-ropathy at baseline had neuropathy assessed at fol-low-up (Fig. 1). Baseline investigations and assess-ment of neuropathy were repeated after a mean of7.3 years of follow-up.

statistical analysis

Statistical analyses were performed with Stata soft-ware (version 7.0). The characteristics of patientswho did and those who did not undergo assessmentfor neuropathy were compared with the use of Stu-dents t-test or the MannWhitney U test where ap-propriate. The possibility of a threshold effect in therelation between glycosylated hemoglobin valuesand the risk of neuropathy was evaluated by com-paring the linear and exponential trend lines accord-ing to quintiles of glycosylated hemoglobin, as pre-viously reported.

9,18

The relation between riskfactors and the incidence of neuropathy was ana-lyzed with the use of logistic-regression models.Odds ratios for dichotomous variables express therisk of neuropathy for patients with the risk factorsas compared with those without the risk factors.

Odds ratios for continuous risk factors were stan-dardized, thus expressing the risk associated witha 1-SD increase in the continuous risk factors. Alllogistic-regression models were adjusted for theduration of diabetes and glycosylated hemoglobinvalues. The relation between the change in glycosy-lated hemoglobin values and the incidence of neu-ropathy was assessed with the use of logistic regres-sion, with adjustments for potential confounders.The odds ratio for the incidence of neuropathy as-sociated with each increase of one quintile in theglycosylated hemoglobin value, as compared withthe lowest quintile (the reference category), was alsoassessed.

All relevant risk factors were included in a mul-tivariate logistic-regression model to calculate mu-tually adjusted, standardized odds ratios (Model 1).In addition, complications of diabetes were addedto the model (Model 2).

We used logistic-regression models to assess thedevelopment of neuropathy during follow-up, rath-er than an analysis of failure time (time to diagno-sis of neuropathy) for current-status data, becausethere was little variation in follow-up time (mean,

* Plusminus values are means SD. Alternatively, in cases of skewed distributions, data are medians (5th percentile, 95th percentile). P values are derived from Students t-test for data normally distributed or, in cases of skewed distributions, from a MannWhitney U test. To convert values for cholesterol to milligrams per deciliter, divide by 0.02586. To convert values for triglycerides to milligrams per deciliter, divide by 0.01129.

The blood-pressure data exclude patients who were undergoing antihypertensive therapy.

The body-mass index is the weight in kilograms divided by the square of the height in meters.

Table 1. Comparison of Baseline Data in 1819 Patients According to Whether There Was an Assessment for Neuropathy at Follow-up.*

Variable Assessed for Neuropathy at Follow-up P Value

No Yes

No. of patients 647 1172

Age (yr) 30.79.1 30.78.8 0.91

Duration of diabetes (yr) 12.78.5 12.48.1 0.45

Glycosylated hemoglobin (% of hemoglobin) 8.31.9 8.01.8 0.01

Insulin dose per kg of body weight (IU) 0.70.2 0.70.2 0.55

Systolic blood pressure (mm Hg) 118.0 (96, 148) 117.0 (96, 142) 0.03

Diastolic blood pressure (mm Hg) 74.0 (57, 92) 74.0 (58, 91) 0.95

Weight (kg) 66.710.6 66.810.5 0.85

Body-mass index 23.32.7 23.42.7 0.58

Waist-to-hip ratio 0.80.1 0.80.1 0.64

Albumin excretion rate (g/min) 9.5 (3.2, 148.2) 9.9 (3.3, 147.1) 0.32

Total cholesterol (mmol/liter) 5.281.04 5.161.05 0.02

Triglycerides (mmol/liter) 0.90 (0.48, 2.22) 0.86 (0.48, 2.20) 0.03

von Willebrand factor (U/ml) 1.08(0.51, 1.95) 1.11 (0.55, 2.21) 0.07



n engl j med

352;4


27, 2005


345

7.30.6 years); the response is therefore the cumu-lative incidence of neuropathy during the follow-up period. All P values are two-sided.

baseline characteristics of the patients without neuropathy at baseline

Table 1 compares baseline data among the 1819 pa-tients without neuropathy at baseline. Those whowere lost to follow-up or died or in whom neurop-athy was not assessed (647 patients) had higher

baseline glycosylated hemoglobin, systolic bloodpressure, total cholesterol, and triglyceride levels.Of the remaining 1172 patients without neuropa-thy at baseline, neuropathy had developed in 276 atfollow-up a cumulative incidence of neuropathyof 23.5 percent. There was little variation in the in-cidence of defining variables: symptoms of neurop-athy developed in 24.4 percent of patients, absentreflexes in 20.5 percent, abnormal vibration-percep-tion threshold in 20.4 percent, and abnormal auto-nomic-function results in 17.3 percent. There wereno significant differences between centers in the in-

results

* Plusminus values are means SD. Alternatively, in cases of skewed distributions, data are medians (5th percentile, 95th percentile). P values are derived from Students t-test or, in cases of skewed distributions, from a MannWhitney U test. To convert values for cholesterol to milligrams per deciliter, divide by 0.02586. To convert values for triglycerides to mil-ligrams per deciliter, divide by 0.01129.

The blood-pressure data exclude patients who were undergoing antihypertensive therapy.

Table 2. Baseline Characteristics of 1819 Patients According to the Incidence of Neuropathy.*

VariableProgression

to NeuropathyNo Progression to Neuropathy P Value

No. of patients 276 896

Age (yr) 33.610.0 29.88.1

n engl j med

352;4


27

,

2005

The

new england journal

of

medicine

346

cidence of neuropathy. At baseline, 570 of the 927subjects defined as having neuropathy (61.5 per-cent) had an abnormal measurement of the vibra-tion-perception threshold.

baseline characteristics according to the incidence of neuropathy

Table 2 shows baseline characteristics of the studypopulation grouped according to whether neurop-athy did or did not develop during the study. Pa-tients in whom neuropathy developed were on av-erage 3.8 years older, had had diabetes for 3.3 yearslonger, and had poorer blood glucose control (a gly-cosylated hemoglobin value 0.5 percent higher) atbaseline than those in whom neuropathy did notdevelop.

In unadjusted comparisons, patients in whomneuropathy developed also had higher baseline lev-els of total and LDL cholesterol and fasting triglyc-erides, a higher body-mass index, higher von Will-ebrand factor levels and albumin excretion rate,

and a lower estimated glucose disposal rate. Morepatients in whom neuropathy developed had base-line hypertension, microalbuminuria or macroal-buminuria, any retinopathy, a history of cardiovas-cular disease, and a history of smoking.

There was no evidence of a significant thresholdeffect of glycosylated hemoglobin at baseline andprogression to neuropathy. An abnormal vibration-perception threshold was present in 166 (60.1 per-cent) of the 276 patients in whom neuropathy de-veloped.

risk factors for neuropathyafter adjustment for glycosylated hemoglobin and duration of diabetes

Adjustment for possible confounding by the twomajor known determinants of neuropathy dura-tion of diabetes and glycosylated hemoglobin value(Table 3) attenuated or abolished some associa-tions with progression to neuropathy. The risk fac-tors that remained significantly associated with thedevelopment of neuropathy were baseline cardio-vascular risk factors (body-mass index and levels oftotal cholesterol, LDL cholesterol, and triglycerides).The urinary albumin excretion rate at baseline andthe von Willebrand factor level were also elevatedin patients in whom neuropathy developed (Table3), as was the insulin resistance (but the estimatedglucose disposal rate was lower).

After adjustment for glycosylated hemoglobinvalue and duration of diabetes, we found that thepresence of hypertension, albuminuria (either mi-croalbuminuria or macroalbuminuria), any retinop-athy, history of cardiovascular disease, and historyof smoking at baseline were significantly associat-ed with newly diagnosed neuropathy (odds ratio formicroalbuminuria or macroalbuminuria, 1.48; andfor history of cardiovascular disease, 2.74). Bothformer and current smoking at baseline were sig-nificantly related to neuropathy when included sep-arately, with adjustment for glycosylated hemoglo-bin values and duration of diabetes. Since bothsmoking groups had similar risk, they are present-ed jointly.

change in glycosylated hemoglobin levels during follow-up

An increase in glycosylated hemoglobin was asso-ciated with an odds ratio of 1.45 (95 percent confi-dence interval, 1.23 to 1.72; P

n engl j med

352;4


27, 2005


347

ratio for each quintile of change in the glycosylatedhemoglobin value, as compared with the lowestquintile of change (the reference category), with ad-justment for the baseline glycosylated hemoglobinvalue and duration of diabetes. The odds ratio forthe development of neuropathy was 2.48 (95 per-cent confidence interval, 1.50 to 4.11) for patientswith the highest rate of deterioration of glycosylat-ed hemoglobin (the highest quintile).

multivariate models

In a multivariate logistic-regression model (Table 4,Model 1) that mutually adjusted for the presence ofall other risk factors, we found that duration of dia-betes, glycosylated hemoglobin value, change inglycosylated hemoglobin value, triglyceride level,body-mass index, history of smoking, and presenceof hypertension at baseline remained significantlyassociated with the incidence of neuropathy. Therelation between cholesterol level and the incidenceof neuropathy in Model 1 did not reach statisticalsignificance, and albumin excretion rate was unre-lated to neuropathy. The strongest relation was withhypertension, which carried an odds ratio of 1.57.Addition of the estimated glucose disposal rate tothis model did not materially change the odds ratiosfor the other variables, and the glucose disposal rateitself was not associated with newly diagnosed neu-ropathy (odds ratio, 0.99; 95 percent confidence in-terval, 0.90 to 1.08; P=0.75).

When complications of diabetes were added tothe model (Table 4, Model 2), odds ratios for tri-glycerides and hypertension were not statisticallysignificant. Duration of diabetes, glycosylated he-moglobin value, change in glycosylated hemoglo-bin value, body-mass index, and history of smokingremained significantly related to the incidence ofneuropathy. In addition, the presence of cardiovas-cular disease at baseline was independently relatedto the incidence of neuropathy (odds ratio, 2.12).The relation between any retinopathy and neuropa-thy was insignificant (P=0.06) when change in theglycosylated hemoglobin value was included in themodel but reached significance (P=0.04) whenchange in the glycosylated hemoglobin value wasexcluded.

Models 1 and 2 were repeated without autonom-ic symptoms and the results of autonomic-functiontests. The only major difference in the results wasthat the relationship between hypertension andpure peripheral neuropathy was no longer statis-tically significant.

During a period of seven years, the incidence of neu-ropathy in the EURODIAB cohort of patients withtype 1 diabetes was 23.5 percent, confirming thepreviously reported strong contributions of glyce-mic control and duration of diabetes to the risk ofneuropathy.

17,19,20

We observed that the rate of de-terioration of glycemic control during follow-upcontributed markedly to the risk of neuropathy, in-dependently of the glycosylated hemoglobin valueat baseline. Furthermore, cardiovascular risk fac-tors, such as hypertension, smoking, obesity, andelevated triglyceride levels, and the presence of car-diovascular disease at baseline appear to be relatedto newly diagnosed neuropathy.

This study used a definition of neuropathy thatwas similar to that in other protocols,

2,21,22

and28.5 percent of patients had neuropathy at base-line,

17

a frequency similar to that observed in otherstudies.

22

Of the patients with neuropathy at base-line, approximately 60 percent had an abnormalvibration-perception threshold. Moreover, 500 ran-domly selected subjects within the cohort under-went nerve-conduction studies of the arms and legsat follow-up; there was a significant relationshipbetween the slowing of nerve conduction and the

discussion

Figure 2. Odds Ratios for the Development of Neuropathy per Quintile of Change in Glycosylated Hemoglobin.

The odds ratios have been adjusted for glycosylated hemoglobin at baseline and for duration of diabetes. Quintile 1 served as the reference category. Ver-tical bars represent 95 percent confidence intervals. The P value for trend across quintiles is 0.009.

Odd

s R

atio

for

the

Dev

elop

men

t of

Neu

ropa

thy

3.5

2.5

3.0

2.0

1.5

1.0

0.5

0.0

Quintile of Change and Median Change in Glycosylated Hemoglobin

1 2 3 4 5(1.96

[reference])(0.58) (0.18) (0.84) (1.90)

4.5

4.0



n engl j med

352;4


27

,

2005

The

new england journal

of

medicine

348

number of abnormal criteria used to define neu-ropathy.

Assessing neuropathy involves measures that areprone to varying degrees of subjectivity or observerbias. However, these measurement errors are likelyonly to underestimate the statistical significance ofthe association between risk factors and disease.

Elucidating the risk factors for neuropathy is im-portant, given the association between the risk fac-tors and increased diabetes-related morbidity andmortality.

23

Mortality in patients with neuropathy ishigh, and the cause of death is often coronary heart

disease.

23,24

Previous cross-sectional studies havereported associations between neuropathy andheight,

25

cigarette smoking,

26

low levels of HDLcholesterol,

19

and hypertension

19,27

in addition toestablished risk factors, level of glycemia, and du-ration of diabetes. Hypertension was observed to bea strong risk factor for neuropathy in 463 youngpatients with type 1 diabetes.

21

A follow-up of sub-jects with either type 1 or type 2 diabetes suggestedthat both type 1 and type 2 diabetes and markers ofmicrovascular disease were associated with the se-verity of neuropathy.

28

The baseline cross-sectional data in the presentstudy

17

are evidence of a strong association betweenneuropathy and other microvascular complications,suggesting a common pathogenic mechanism.

29-31

We excluded all patients with neuropathy at base-line in order to identify new cases, resulting in a lessrobust but still significant association.

Several of the risk factors associated with neu-ropathy in this study are markers of insulin resis-tance. Both an elevated triglyceride level and obesityare strong predictors of microalbuminuria and ret-inopathy in type 1 diabetes.

9,18

Insulin resistance,present in type 1 diabetes,

32

may be involved in thedevelopment of both microvascular and macrovas-cular complications.

33,34

A potential link betweeninsulin resistance and microvascular complicationsincluding neuropathy is their association with en-dothelial dysfunction.

35

Other complications of di-abetes linked to endothelial dysfunction may alsopredict the risk of microalbuminuria,

18

retinopa-thy,

9

and neuropathy. The patients in whom neu-ropathy developed had a lower estimated glucosedisposal rate at baseline (higher insulin resistance).However, multivariate analysis indicates that obe-sity, glycemic control, and the level of triglycerideswere more predictive individually as compared withthe combined summary measure.

When autonomic symptoms and results of au-tonomic-function tests were excluded from the def-inition of neuropathy, the effect of hypertension onthe incidence of neuropathy lost statistical signifi-cance. This is consistent with our previous reportthat systolic blood pressure was a risk factor for thedevelopment of cardiac autonomic neuropathy.

36

Apart from the use of medication to induce glu-cose values approaching normoglycemia, no othertreatments currently exist to reduce the progressionto neuropathy.

29

Aggressive treatment of hyperten-sion is now standard clinical practice in the man-agement of nephropathy and retinopathy, and the

* CI denotes confidence interval. To convert values for cholesterol to milligrams per deciliter, divide by 0.02586. To convert values for triglycerides to milli-grams per deciliter, divide by 0.01129.

Model 1 represents 1101 observations with urinary albumin excretion rate but without retinopathy and cardiovascular disease.

Model 2 represents 932 observations with urinary albumin excretion rate and

complications of diabetes (retinopathy and cardiovascular disease).

Table 4. Odds Ratios for Associations between Key Risk Factors and the Incidence of Diabetic Neuropathy with the Use of Two Logistic-Regression Models.*

VariableOdds Ratio

(95% CI) P Value

Model 1

Duration of diabetes (yr) 1.40 (1.211.63)

n engl j med

352;4


27, 2005


349

results of the present study make a case for clinicaltrials to confirm the efficacy of antihypertensiveagents and possibly other strategies for cardiovas-cular risk reduction in slowing the progression ofneuropathy.

The EURODIAB Prospective Complications Study was supportedby grants from the Wellcome Trust, the European Union, and Diabe-tes UK.

Dr. Tesfaye reports having received consulting and lecture feesfrom Eli Lilly, consulting fees from AstraZeneca and Pfizer, and lec-ture fees from Takeda and GlaxoSmithKline.

appendix

In addition to the authors, the following investigators participated in the EURODIAB Prospective Complications Study:

Greece

B. Kara-manos, A. Kofinis, K. Petrou (Hippokration Hospital, Athens); N. Papazoglou, C. Manes (General Hospital Papageorgiou of Thessaloniki,Thessaloniki);

Italy

F. Giorgino, G. Picca, A. Angarano, G. De Pergola, L. Laviola, R. Giorgino (Endocrinologia e Malattie Metaboliche,DETO, Universit Degli Studi di Bari, Bari); M. Songini, A. Casu, M. Pedron, S. Pintus, M. Fossarello (Ospedale San Michele, Cagliari); G.Pozza, R. Mangili, V. Asnaghi (Ospedale San Raffaele, Milan); F. Santeusanio, G. Rosi, V. DAlessandro, C. Cagini, P. Bottini, G.P. Reboldi(Istituto di Patologia Medica, Policlinico, Perugia); R. Navalesi, G. Penno, S. Bandinelli, R. Miccoli, M. Nannipieri (Dipartimento di Endo-crinologia e Metabolismo, Pisa); G. Ghirlanda, C. Saponara, P. Cotroneo, A. Manto, A. Minnella (Universita Cattolica del Sacro Cuore,Rome); M. Borra, P. Cavallo Perin, S. Giunti, G. Grassi, G.F. Pagano, M. Porta, R. Sivieri, F. Vitelli, D. Ferrari, M. Veglio (Dipartimento di Me-dicina Interna, Universit di Torino and ASO CTO/CRF/Maria Adelaide, Turin); M. Muggeo, M. Iagulli (V Cattedra di Malattie del Metabo-lismo, Verona);

Ireland

J.B. Ferriss, G. Grealy, D.O. Keefe (Cork Regional Hospital, Cork);

Germany

M. Toeller, C. Arden (Diabetes Re-search Institute, Heinrich Heine University, Dusseldorf ); E. Standl, B. Schaffler, H. Brand, A. Harms (City Hospital Schwabing, Munich);

Belgium

R. Rottiers, C. Tuyttens, H. Priem (University Hospital of Ghent, Ghent);

Finland

P. Ebeling, M. Kylliinen, V.A. Koivisto (Uni-versity Hospital of Helsinki, Helsinki);

Poland

B. Idzior-Walus, J. Sieradzki, K. Cyganek, B. Solnica (Jagiellonian University, Krakow);

theNetherlands

H.H.P.J. Lemkes, J.C. Lemkes-Stuffken (Leiden University Medical Center, Leiden);

Portugal

J. Nunes-Correa, M.C. Rogado,L. Gardete-Correia, M.C. Cardoso, A. Silva, J. Boavida, M. Machado Sa Marques (Portuguese Diabetic Association, Lisbon);

Luxembourg

G. Michel, R. Wirion, S. Cardillo (Centre Hospitalier, Luxembourg City);

France

B. Soussan, O. Verier-Mine, P. Fallas, M.C. Fallas (CentreHospitalier de Valenciennes, Valenciennes); G. Cathelineau, A. Bouallouche, B. Villatte Cathelineau (Hospital Saint-Louis, Paris);

UnitedKingdom

J.H. Fuller, J. Holloway, L. Asbury, D.J. Betteridge (University College London, London); J.D. Ward, S. Tesfaye, S. Eaton, C. Mody(Royal Hallamshire Hospital, Sheffield); S. Walford, J. Sinclair, S. Hughes, V. McLelland, J. Ward (New Cross Hospital, Wolverhampton);

Austria

K. Irsigler, H. Abrahamian (Hospital Vienna-Lainz, Vienna);

Croatia

G. Roglic, Z. Metelko, Z.R. Pepeonik (Vuk Vrhovac Insti-tute for Diabetes, Zagreb);

Romania

C. Ionescu-Tirgoviste, A. Coszma, C. Guja (Clinic of Diabetes, Nutrition & Metabolic Diseases, Bu-charest);

Coordinating center

J.H. Fuller, N. Chaturvedi, J. Holloway, D. Webb, L. Asbury, M. Shipley, S.J. Livingstone (University CollegeLondon, London);

Central laboratories

G.-C. Viberti, R. Swaminathan, P. Lumb, A. Collins, S. Sankaralingham (Guys and St. ThomasHospital, London).

references

1.

Vinik AI, Park TS, Stansberry KB, Pit-tenger GL. Diabetic neuropathies. Diabeto-logia 2000;43:957-73.

2.

The Diabetes Control and Complica-tions Trial Research Group. The effect of in-tensive diabetes therapy on the developmentand progression of neuropathy. Ann InternMed 1995;122:561-8.

3.

Microvascular and acute complicationsin IDDM patients: the EURODIAB IDDMComplications Study. Diabetologia 1994;37:278-85.

4.

Siedel J, Hagele EO, Ziegenhorn J,Wahlefeld AW. Reagent for the enzymaticdetermination of serum total cholesterolwith improved lipolytic efficiency. ClinChem 1983;29:1075-80.

5.

Bucolo G, David H. Quantitative deter-mination of serum triglycerides by the use ofenzymes. Clin Chem 1973;19:476-82.

6.

Warnick GR, Albers JJ. A comprehen-sive evaluation of the heparin-manganeseprecipitation procedure for estimating highdensity lipoprotein cholesterol. J Lipid Res1978;19:65-76.

7.

Friedewald WT, Levy RI, FredricksonDS. Estimation of the concentration of low-density lipoprotein cholesterol in plasma,without the use of the preparative ultracen-trifuge. Clin Chem 1972;18:499-502.

8.

John WG, Gray MR, Bates DL, BeachamJL. Enzyme immunoassay: a new technique

for estimating hemoglobin A

1

c

. Clin Chem1993;39:663-6.

9.

Chaturvedi N, Sjoelie AK, Porta M, et al.Markers of insulin resistance are strong riskfactors for retinopathy incidence in type 1diabetes. Diabetes Care 2001;24:284-9.

10.

Kearney EM, Mount JN, Watts GF, SlavinBM, Kind PRN. Simple immunoturbidimet-ric method for determining urinary albuminat low concentrations using Cobas-Bio cen-trifugal analyser. J Clin Pathol 1987;40:465-8.

11.

Aldington SJ, Kohner EM, Meuer S,Klein R, Sjolie AK. Methodology for retinalphotography and assessment of diabetic ret-inopathy: the EURODIAB IDDM Complica-tions Study. Diabetologia 1995;38:437-44.

12.

Rose GA, Blackburn H, Gilum RF, Prin-eas RJ, eds. Cardiovascular survey methods.2nd ed. Geneva: World Health Organiza-tion, 1982.

13. Williams KV, Erbey JR, Becker D, Arsla-nian S, Orchard TJ. Can clinical factors esti-mate insulin resistance in type 1 diabetes?Diabetes 2000;49:626-32.14. Ford I, Malik RA, Newrick PG, PrestonFE, Ward JD, Greaves M. Relationships be-tween haemostatic factors and capillarymorphology in human diabetic neuropathy.Thromb Haemost 1992;68:628-33.15. Wiles PG, Pearce SM, Rice PJS, MitchellJMO. Vibration perception threshold: influ-

ence of age, height, sex, and smoking, andcalculation of accurate centile values. DiabetMed 1991;8:157-61.16. The Consensus Committee of the Amer-ican Autonomic Society and the AmericanAcademy of Neurology. Consensus state-ment on the definition of orthostatic hypo-tension, pure autonomic failure, and multi-ple system atrophy. Neurology 1996;46:1470.17. Tesfaye S, Stevens LK, Stephenson JM,et al. Prevalence of diabetic peripheral neu-ropathy and its relation to glycaemic controland potential risk factors: the EURODIABIDDM Complications Study. Diabetologia1996;39:1377-84.18. Chaturvedi N, Bandinelli S, Mangili R,Penno G, Rottiers RE, Fuller JH. Microalbu-minuria in type 1 diabetes: rates, risk factorsand glycemic threshold. Kidney Int 2001;60:219-27.19. Maser RE, Steenkiste AR, Dorman JS, etal. Epidemiological correlates of diabeticneuropathy: report from Pittsburgh Epide-miology of Diabetes Complications Study.Diabetes 1989;38:1456-61.20. Franklin GM, Kahn LB, Baxter J, Mar-shall JA, Hamman RF. Sensory neuropathyin non-insulin-dependant diabetes mellitus:the San Luis Valley Diabetes Study. Am J Epi-demiol 1990;131:633-43.21. Forrest KY, Maser RE, Pambianco G,



n engl j med 352;4 www.nejm.org january 27, 2005350


Becker DJ, Orchard TJ. Hypertension as arisk factor for diabetic neuropathy: a pro-spective study. Diabetes 1997;46:665-70.22. Young MJ, Boulton AJM, MacLeod AF,Williams DR, Sonksen PH. A multicentrestudy of the prevalence of diabetic peripher-al neuropathy in the United Kingdom hospi-tal clinic population. Diabetologia 1993;36:150-4.23. Forsblom CM, Sane T, Groop PH, et al.Risk factors for mortality in Type II (non-insulin-dependent) diabetes: evidence of arole for neuropathy and a protective effect ofHLA-DR4. Diabetologia 1998;41:1253-62.24. Coppini DV, Bowtell PA, Weng C, YoungPJ, Sonksen PH. Showing neuropathy is re-lated to increased mortality in diabetic pa-tients a survival analysis using an acceler-ated failure time model. J Clin Epidemiol2000;53:519-23.25. Gadia MT, Natori N, Ramos LB, AyyarDR, Skyler JS, Sosenko JM. Influence ofheight on quantative sensory, nerve-con-duction, and clinical indices of diabetic pe-ripheral neuropathy. Diabetes Care 1987;10:613-6.26. Mitchell BD, Hawthorne VM, Vinik AI.

Cigarette smoking and neuropathy in dia-betic patients. Diabetes Care 1990;13:434-7.27. Harris M, Eastman R, Cowie C. Symp-toms of sensory neuropathy in adults withNIDDM in the U.S. population. DiabetesCare 1993;16:1446-52.28. Dyck PJ, Davies JL, Wilson DM, ServiceFJ, Melton LJ III, OBrien PC. Risk factors forseverity of diabetic polyneuropathy: inten-sive longitudinal assessment of the Roches-ter Diabetic Neuropathy Study cohort. Dia-betes Care 1999;22:1479-86.29. Cameron NE, Eaton SE, Cotter MA, Tes-faye S. Vascular factors and metabolic inter-actions in the pathogenesis of diabetic neu-ropathy. Diabetologia 2001;44:1973-88.30. Tesfaye S, Harris N, Jakubowski J, et al.Impaired blood flow and arterio-venousshunting in human diabetic neuropathy:a novel technique of nerve photography andfluorescein angiography. Diabetologia 1993;36:1266-74.31. Eaton SE, Harris ND, Ibrahim S, et al.Increased sural nerve epineurial blood flowin human subjects with painful diabeticneuropathy. Diabetologia 2003;46:934-9.

32. DeFronzo RA, Hendler R, Simonson D.Insulin resistance is a prominent feature ofinsulin-dependent diabetes. Diabetes 1982;31:795-801.33. Soedamah-Muthu SS, Chaturvedi N,Toeller M, et al. Risk factors for coronaryheart disease in type 1 diabetic patients inEurope: the EURODIAB Prospective Com-plications Study. Diabetes Care 2004;27:530-7.34. Erbey JR, Robbins D, Forrest KY, Or-chard TJ. Low-density lipoprotein particlesize and coronary artery disease in child-hood-onset type 1 diabetes population. Me-tabolism 1999;48:531-4. [Erratum, Metab-olism 1999;48:1337.] 35. Zenere BM, Arcaro G, Saggiani F, RossiL, Muggeo M, Lechi A. Noninvasive detec-tion of functional alterations of the arterialwall in IDDM patients with and without mi-croalbuminuria. Diabetes Care 1995;18:975-82.36. Witte DR, Tesfaye S, Chaturvedi N, et al.Risk factors for cardiac autonomic neuropa-thy in Type 1 diabetes mellitus. Diabetologia(in press).Copyright 2005 Massachusetts Medical Society.

electronic access to the journals cumulative index

At the Journals site on the World Wide Web (www.nejm.org), you can search an index of all articles published since January 1975 (abstracts 19751992, full text 1993present). You can search by author, key word, title, type of article, and date. The results will include the citations for the articles plus links to the full text of articles published since 1993. For nonsubscribers, time-limited access to single articles and 24-hour site access can also be ordered for a fee through the Internet (www.nejm.org).



nejmoa032782

Documents

Transcript of nejmoa032782