Peripheral Neuropathy and NerveDysfunction in Individuals at HighRisk for Type 2 Diabetes: ThePROMISE CohortDiabetes Care 2015;38:793–800 | DOI: 10.2337/dc14-2585

OBJECTIVE

Emerging evidence suggests that peripheral neuropathy begins in the early stages ofdiabetes pathogenesis. Our objective was to describe the prevalence of peripheralneuropathy and nerve dysfunction according to glucose tolerance and metabolicsyndrome status and examine how these conditions are associatedwith neurologicalchanges in individuals at risk for type 2 diabetes.

RESEARCH DESIGN AND METHODS

We studied 467 individuals in the longitudinal PROMISE (Prospective Metabolismand Islet Cell Evaluation) cohort. Peripheral neuropathy was defined by MichiganNeuropathy Screening Instrument (MNSI) scores (>2), and the severity of nervedysfunction was measured objectively by vibration perception thresholds (VPTs)using a neurothesiometer. Metabolic syndrome was defined using the InternationalDiabetes Federation/American Heart Association harmonized criteria.

RESULTS

The prevalence of peripheral neuropathy was 29%, 49%, and 50% for normalglycemia, prediabetes, and new-onset diabetes, respectively (P < 0.001 for trend).The mean VPT was 6.5 V for normal glycemia, 7.9 V for prediabetes, and 7.6 V fornew-onset diabetes (P = 0.024 for trend). Prediabetes was associated with higherMNSI scores (P = 0.01) and VPTs (P = 0.004) versus normal glycemia, independentof known risk factors. Additionally, progression of glucose intolerance over 3 yearspredicted a higher risk of peripheral neuropathy (P = 0.007) and nerve dysfunction(P = 0.002). Metabolic syndrome was not independently associated with MNSIscores or VPTs.

CONCLUSIONS

In individuals with multiple risk factors for diabetes, prediabetes was associatedwith similar risks of peripheral neuropathy and severity of nerve dysfunction asnew-onset diabetes. Prediabetes, but not metabolic syndrome, was indepen-dently associated with both the presence of peripheral neuropathy and the se-verity of nerve dysfunction.

Peripheral neuropathy is a serious complication of diabetes. It plays a major contrib-utory role in the initiationof foot ulcerationand the subsequent development of lower-extremity amputation, resulting in severe disability, reduced quality of life, and asignificant economic burden to the health care system (1). Peripheral neuropathy is

1Department of Nutritional Sciences, Universityof Toronto, Toronto, ON, Canada2Division of Endocrinology, Department of Medi-cine, University of Toronto, Toronto, ON, Canada3Centre for Studies in Family Medicine, Univer-sity of Western Ontario, London, ON, Canada4Lunenfeld-Tanenbaum Research Institute,Mount Sinai Hospital, Toronto, ON, Canada5Division of Endocrinology and Metabolism andthe Population Health Research Institute, De-partment of Medicine, McMaster University,Hamilton, ON, Canada6Dalla Lana School of Public Health, University ofToronto, Toronto, ON, Canada

Correspondingauthor: Anthony J. Hanley, anthony.hanley@utoronto.ca.

Received 31 October 2014 and accepted 5January 2015.

This article contains Supplementary Data onlineat http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc14-2585/-/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.

C. Christine Lee,1 Bruce A. Perkins,2

Sheena Kayaniyil,1 Stewart B. Harris,3

Ravi Retnakaran,2,4 Hertzel C. Gerstein,5

Bernard Zinman,2,4 and

Anthony J. Hanley1,2,6

Diabetes Care Volume 38, May 2015 793

EPIDEM

IOLO

GY/H

EALTH

SERVICES

RESEA

RCH

classically recognized as a complication oflong-standing diabetes, with risk factorsincluding increased duration of diabetesandpoor glycemic control (2,3). However,accumulating evidence suggests that theprevalence of peripheral neuropathy ismarkedly elevated at the time of diabetesdiagnosis, with estimates varying from11.5 to 48% in population-based observa-tional studies (4,5). In contrast, a recentstudy using British and German adminis-trative databases showed that peripheralneuropathy based on diagnostic codeswas present in only 2.4–5.7% of patientswith newly diagnosed diabetes in primarycare practices (6). These discrepancies inthe prevalence of peripheral neuropathyin new-onset diabetes could be related todifferences in the methods used to ascer-tain this condition.The presence of peripheral neuropa-

thy in newly diagnosed diabetes and thetime gap between the onset of diabetesand clinical diagnosis (7) support the no-tion that neuropathic changes could beinitiated in the early stages of diabetespathogenesis. In the Cooperative HealthResearch in the Region of Augsburg(KORA) F4 study, the prevalence of pe-ripheral neuropathy defined as bilater-ally impaired foot vibration perceptionand/or foot pressure sensation was sim-ilar in patients with known diabetes(22%) and in those with impaired fastingglucose and impaired glucose tolerance(24%) (8). The U.S. National Health andNutrition Examination Study (1999–2004) reported that the prevalence of pe-ripheral neuropathy, as measured by atleast one insensate area on 10-g mono-filament testing, was 11.9% in individualswith impaired fasting glucose (9). Usingthe Michigan Neuropathy Screening In-strument (MNSI), peripheral neuropathywas found in 13.0% of subjects with im-paired glucose tolerance and 11.3% ofthose with impaired fasting glucose inthe MONICA (Monitoring Trends and De-terminants on Cardiovascular Diseases)/KORA Augsburg Surveys (10). However,these studies were limited by the lack oforal glucose tolerance testing to ascertainglycemic status in theNationalHealth andNutrition Examination Survey (9) andthe relatively small sample size in theMONICA/KORA Augsburg Surveys (10).Age, sex, ethnicity, height, duration of

diabetes, and glycemic control are estab-lished risk factors for diabetic neuropa-thy (2,3). However, the risk factors for

peripheral neuropathy in prediabetesare not well documented. Emerging evi-dence suggests that components of themetabolic syndrome, in particular ab-dominal adiposity and dyslipidemia, areassociatedwith peripheral neuropathy inpatients with prediabetes (10,11). In thecurrent study, we aimed to describe theprevalence of peripheral neuropathy andnerve dysfunction and to investigate theassociation of glucose tolerance statusand metabolic syndrome with theseneurological changes using data fromthe Prospective Metabolism and IsletCell Evaluation (PROMISE) study, a well-characterized cohort of individuals athigh risk for type 2 diabetes.

RESEARCH DESIGN AND METHODS

Study PopulationThe study population comprised partici-pants in PROMISE, a prospective observa-tional study that explores the longitudinalrelationships of insulin resistance and pan-creatic b-cell dysfunction with known andemerging risk factors in individuals at highrisk for type 2 diabetes. Between May2004 and December 2006, PROMISE re-cruited 712 participants without diabetesfrom Toronto and London, Ontario, Can-ada, basedon the presenceof one ormorerisk factors for type 2 diabetes, includingobesity, hypertension, family history of di-abetes, and history of gestational diabetesor birth of a macrosomic infant (12).Among these participants, 496 returnedfor the 3-year follow-up examination,when assessments of neuropathy andnerve dysfunction were conducted. Therewere no significant differences in BMI ormeasures of glucose homeostasis in par-ticipants who attended the follow-up ex-amination compared with those who didnot. After excluding participants in whomdiabetes developed after baseline andbefore the 3-year follow-up examination(n = 7), those with missing data on glucosetolerance status at the 3-year follow-upexamination (n = 2), and those with miss-ing data on measurements of peripheralneuropathy and nerve function (n = 20),the sample size of the current analysis was467 participants without known diabetesat the time of the 3-year examination whohad peripheral neuropathy and nervefunction measurements. The study proce-dures detailed in this article were per-formed at the 3-year examination, andthe data were analyzed cross-sectionally.The institutional review boards at each

study center approved the study protocol,and all participants provided written in-formed consent.

Measurement of PeripheralNeuropathyPeripheral neuropathy was measuredusing the MNSI. The MNSI consisted oftwo parts: a self-administered question-naire on clinical signs and symptoms ofneuropathy and a clinical examinationof both feet involving 1) an inspectionto detect deformities, dry skin, calluses,infection, fissures, or ulcerations; 2) agrading of ankle reflexes; 3) a semiquan-titative assessment of vibration percep-tion at the dorsum of the great toe; and4) 10-g monofilament testing. In thisanalysis, we used only the clinical exam-ination component of the MNSI, with amaximum possible score of 10. An MNSIscore of.2 defined the presence of pe-ripheral neuropathy (13). The MNSI hasbeen validated against the Mayo ClinicNeuropathy Disability Score, nerve con-duction measures, quantitative vibra-tion threshold tests, and the MichiganDiabetic Neuropathy Score. In these val-idation studies, the maximum possiblescore was 8 because the score at thattime did not include the 10-g monofila-ment test (14).

Measurement of Nerve DysfunctionWe used a neurothesiometer to quantifythe vibration perception threshold(VPT), a measure of the severity of nervedysfunction. The neurothesiometer is abattery-operated diagnostic instrumentthat assesses sensitivity thresholds at var-ious sites on the body surface. It has beenvalidated against nerve conduction stud-ies (15). On the basis of the method oflimits, participants were asked to indicatewhen they first perceived vibration sen-sation after stimulus was applied to thedistal pulp of their left first toe. The in-tensity of the stimulus was gradually in-creased from null to a voltage at whichvibration was first detected (16). Threeseparate tests were conducted with par-ticipants’ eyes closed. The average of thethree VPTs was used for analysis. A nullstimulus test was added randomly to en-sure participant adherence and under-standing of the test requirements. Eachcenter followed standardized proceduresfor the measurement of peripheral neu-ropathy and nerve dysfunction using iden-tical sets of equipment. To ensure that theoutcomes were measured consistently,

794 Neuropathy in Individuals at High Risk for Type 2 Diabetes Diabetes Care Volume 38, May 2015

research assistants were trained centrallywith annual review sessions.

Clinical Measurements and ProcedureGlycemic status (i.e., normal glucose tol-erance, prediabetes, diabetes) was cat-egorized using 1999 World HealthOrganization criteria (17). Prediabeteswas defined as impaired glucose toler-ance and/or impaired fasting glucose.Plasma glucose, HDL cholesterol, andtriglyceride levels were measured infasting blood samples by standard labo-ratory methods. We defined metabolicsyndrome and its components using theInternational Diabetes Federation/American Heart Association harmonizedcriteria (18).Anthropometricmeasurements, includ-

ing height, weight, and waist circumfer-ence, were taken in duplicate accordingto standardized procedures, and the aver-ages of thesemeasurements were used inthe analyses. BMI was calculated asweight in kilograms divided by height insquare meters. Resting systolic and dia-stolic blood pressures were recordedtwice using an automated sphygmoma-nometer after a 5-min rest. Informationon demographics (age, sex, ethnicity), life-style factors (smoking, alcohol consump-tion), and family history of diabetes werecollected in standardized questionnairesby self-report (12). Ethnicity was catego-rized into four groups for the analysis:Caucasian, South Asian, Hispanic, andother. An identical protocol for theseclinical measurements and procedureswas used at both the baseline and the3-year follow-up examinations.

Statistical AnalysisWe summarized the participant charac-teristics stratified by the presence or ab-sence of peripheral neuropathy definedby an MNSI score of.2 out of a total of10, using medians with interquartileranges for continuous variables and per-centages for categorical variables. Weused ANOVA, Kruskal-Wallis, and x2

tests to determine whether continuousand categorical variables differed by thepresence of peripheral neuropathy. Weplotted the prevalence of peripheralneuropathy and the means and SEs ofVPTs by glucose tolerance status andmetabolic syndrome.In light of the well-recognized variabil-

ity in 2-h glucose concentrations afteroral glucose tolerance test (19), we inves-tigated the impact of glucose tolerance

status on the prevalence of neuropathyusing two separate approaches: 1)We cat-egorized glucose tolerance status usingdata from the 3-year follow-up examina-tion (i.e., normal glucose tolerance, predi-abetes, and diabetes were categorizedusing glucose concentrations at the3-year examination only), and 2) we cate-gorized participants into four groupsaccording to progression of glucose intol-erance between the baseline and 3-yearexaminations (i.e., sustained normalglucose tolerance at both baseline and3-year examination, progression to predia-betes, sustained prediabetes, progressionto diabetes). Because of the small numberof participants who regressed from predi-abetes at baseline to normal glucose tol-erance at 3-year examination (n = 1), thisdata point was included in the sustainednormal glucose tolerance group.

We investigated the impact of themet-abolic syndrome on the prevalence ofneuropathy using three approaches: 1)we analyzed the metabolic syndromeas a dichotomous variable (i.e., presencevs. absence); 2) we analyzed the numberof metabolic syndrome disorders (0–5);and 3) we calculated the mean of the zscores of metabolic disorders, which wasanalyzed as a continuous variable (20).Because the distribution of VPTs wasskewed, we loge-transformed this vari-able to achieve normality before sub-sequent analyses. In the subset ofparticipants without diabetes at the 3-year follow-up examination, we usedmultiple regression analyses to investi-gate the association of glucose tolerancestatus and metabolic syndrome withMNSI scores and VPTs. The regression co-efficients (b) for linear regression werepresented along with their 95% CIs. Weincluded covariates in the analyses ifthey were of a priori clinical relevance orif they were associated with both theexposure and the outcome. Covariatesincluded age, sex, ethnicity, and height.

In sensitivity analyses, we used theoriginal MNSI definition (.2 out of totalscore of 8) to define peripheral neurop-athy in prevalence estimates and regres-sion models to assess the consistency ofthe findings across the different defini-tions of MNSI. We also conducted astepwise backward regression analysiswith covariates being removed fromthe multivariable-adjusted model ifP . 0.05. In the stepwise models, wesubstituted fasting blood glucose and

2-h postload blood glucose for predia-betes to determine their specific contri-butions to neuropathic outcomes. Inaddition, we defined peripheral neurop-athy by VPT .12 V (21) and .15 V (22)and estimated its prevalence. Statisticalanalyses were performed using Stata12.0 software (StataCorp, College Sta-tion, TX).

RESULTS

Of the 467 participants with data on pe-ripheral neuropathy and nerve dysfunc-tion at the 3-year examination, 344 hadnormal glycemic control (74%), 101 hadprediabetes (21%), and 22 had new-onset diabetes (5%) ascertained by 75-goral glucose tolerance test at that exam-ination. Among these 467 participants,73% were women and 63% reportedhaving a family history of diabetes. Theaverage age of the study population was53 years (interquartile range 46–60years), and the ethnic distribution forCaucasians, South Asians, Hispanics,and other races was 71%, 12%, 10%,and 7%, respectively. Approximately34% of the participants had evidence ofperipheral neuropathy as defined by anMNSI score .2. The prevalence of pe-ripheral neuropathy according to in-creasing MNSI score was 65.7% for ascore #2, 19.2% for a score .2–3,11.8% for a score.3–4, 2.6% for a score.4–5, and 0.7% for a score .5–6. Themedian VPT was 5.7 V in all participants(interquartile range 4.0–8.3 V).

Characteristics of the study popula-tion stratified by the presence ofMNSI-defined peripheral neuropathyare shown in Table 1. Participants withperipheral neuropathy were more likelyto be older. They also had a higher levelof fasting blood glucose and 2-h post-load blood glucose as well as higherVPTs. No substantial ethnic or sex dif-ferences existed between those withand without MNSI-defined peripheralneuropathy.

The prevalence of peripheral neurop-athy was 29% for participants with nor-mal glucose tolerance, 49% for thosewith prediabetes, and 50% for thosewith newly diagnosed type 2 diabetesat the 3-year examination (P , 0.001)(Fig. 1A). There were significant differ-ences in the prevalence of neuropathyacross categories of glucose tolerancestatus progression between baselineand the 3-year examination. Specifically,

care.diabetesjournals.org Lee and Associates 795

prevalence was ;29% for those withsustained normal glucose tolerance,48% among those who progressedfrom normal glucose tolerance to pre-diabetes, 42% in those with sustainedprediabetes, and 83% among thosewho progressed from prediabetes to di-abetes (P , 0.001) (Fig. 1B). Resultswere similar when we defined neuropa-thy using the original MNSI definition(.2 out of total score of 8).The average VPT was 6.5, 7.9, and

7.6 V for participants with normal glu-cose tolerance, prediabetes, and newlydiagnosed type 2 diabetes, respectively(P = 0.02) (Fig. 2A). The average VPT was6.5 V for participants who had sustainednormal glucose tolerance betweenbaseline and the 3-year examination,7.9 V among those who progressedfrom normal glucose tolerance to pre-diabetes, 6.6 V in those who had sus-tained prediabetes, and 11.4 V amongthose who progressed from prediabetesto diabetes (P = 0.006) (Fig. 2B).On the basis of International Diabetes

Federation/American Heart Associationharmonized criteria, approximatelyone-third of the study population hadmetabolic syndrome, as defined by thepresence of three or more disorders.Although the prevalence of peripheralneuropathy was not significantly

different between those with and with-out metabolic syndrome (33% vs. 37%,respectively; P = 0.4), the average VPTwas higher, specifically, 7.6 V for partic-ipants with and 6.5 V for those withoutmetabolic syndrome (P = 0.01) (Fig. 2C).The average VPT was greater with thepresence of a higher number of compo-nents of the metabolic syndrome, from5.5 V with one component to 8.8 V withfour or more components (P = 0.0005)(Fig. 2D).

Compared with those with normalglycemia, participants with prediabeteshad significantly higher MNSI scores andVPTs (Table 2). In addition, progressionof glucose intolerance compared withsustained normal glucose control wassignificantly associated with a higherrisk of peripheral neuropathy and nervedysfunction (Table 2). There was no as-sociation between metabolic syndromeand MNSI scores (Supplementary Table1). Initial associations of metabolic syn-drome with VPTs were attenuated tononsignificance after adjusting for age,sex, ethnicity, and height (Supplemen-tary Table 1).

In sensitivity analyses, using the origi-nal MNSI definition (.2 out of total scoreof 8) to define peripheral neuropathy didnot change the prevalence estimates(Fig. 1) and magnitude of association in

regression models materially (data notshown). In backward stepwise multipleregression models, age, height, and pre-diabeteswere all significant, independentpositive correlates of MNSI scores andVPTs (all P , 0.05).

When prediabetes was replacedby fasting blood glucose and 2-h post-load blood glucose in the stepwiseregression model of neuropathic out-comes, only 2-h postload blood glucosereached a significance level of 0.05 andwas retained in the final model (datanot shown). We evaluated an alterna-tive definition of peripheral neuropa-thy using neurothesiometer cut pointsof VPT .12 V and .15 V as reportedpreviously (21,22). In the entire studysample, neuropathy prevalence was9.7% for VPT .12 V and 5.1% for VPT.15 V, although there were no signifi-cant differences in prevalence underthese definitions according to glycemicor metabolic syndrome status (data notshown).

CONCLUSIONS

In a cohort at high risk for type 2 diabe-tes, we observed that participants withprediabetes had a similar neuropathyprevalence as those with new-onset di-abetes. In addition, the severity of nervedysfunction, as measured objectivelyby a neurothesiometer, was directlyproportional to the burden of metabolicsyndrome disorders. In individuals with-out diagnosed diabetes, we observedthat age, height, and prediabetes, par-ticularly impaired glucose tolerance,were independently associated withboth peripheral neuropathy and nervedysfunction. The findings extend thecurrent literature by documenting thatthe prevalence of peripheral neuropa-thy in individuals with prediabetes is ashigh as among those with new-onset di-abetes and by providing additional evi-dence from a large, well-characterizedcohort that prediabetes is an indepen-dent risk factor for neuropathy andnerve dysfunction.

The prevalence of peripheral neurop-athy observed in individuals with predi-abetes was higher than previouslyreported (9,10). This could be relatedto differences among study populationsbecause PROMISE participants were re-cruited based on their risk for type 2 di-abetes, and thus, potential risk factorsfor peripheral neuropathy (including

Table 1—Baseline characteristics of participants stratified by the presence ofperipheral neuropathy in the PROMISE cohort

Neuropathy 2 Neuropathy +* P value

n (%) 307 (65.7) 160 (34.3) d

Age (years) 51 (45–58) 57 (48–64) 0.0001

Female sex 75.6 67.5 0.06

Ethnicity 0.08Caucasian 70.4 71.9South Asian 5.54 8.75Hispanic 14.7 7.50Other 9.45 11.9

Current smoker 4.6 5.6 0.3

Heavy drinker 11.8 13.1 0.7

Family history of diabetes 60.1 67.5 0.1

BMI (kg/m2) 30.4 (26.6–34.4) 30.7 (27.8–35.1) 0.3

Height (cm) 164 (159–171) 166 (160–174) 0.07

Waist circumference (cm) 98.2 (88.3–108) 101 (91.0–110) 0.04

Systolic blood pressure (mmHg) 125 (115–135) 125 (117–136) 0.3

Plasma triglycerides (mmol/L) 1.23 (0.95–1.65) 1.27 (1.00–1.81) 0.3

HDL cholesterol (mmol/L) 1.3 (1.1–1.6) 1.3 (1.0–1.5) 0.4

Fasting blood glucose (mmol/L) 5.1 (4.8–5.6) 5.3 (4.9–5.7) 0.03

2-h postload blood glucose (mmol/L) 5.9 (4.8–7.3) 6.7 (5.4–8.5) 0.0001

VPT (V) 5.2 (3.8–7.2) 6.5 (5.1–9.9) 0.0001

Data are median (interquartile range) or % unless otherwise indicated. *The presence ofneuropathy was defined as .2 out of a total MNSI score of 10.

796 Neuropathy in Individuals at High Risk for Type 2 Diabetes Diabetes Care Volume 38, May 2015

glycemia, a family history of diabetes,and metabolic syndrome disorders)were highly prevalent. In addition, vari-ous diagnostic criteria could account forthe difference in the prevalence of neu-ropathy in the various subgroups of glu-cose abnormalities. For example, theKORA F4 study used bilaterally impairedfoot vibration perception and foot pres-sure sensation to detect more severeperipheral neuropathy (8). A high prev-alence of peripheral neuropathy was

also observed in those with normal glu-cose tolerance, an observation that maybe explained by the substantial burdenof diabetes risk factors in this group.

Previous studies consistently demon-strated an association of age and heightwith diabetic neuropathy (2,23,24). Inthe current study, we observed thatthese nonmodifiable risk factors werealso associated with peripheral neurop-athy in prediabetes. Duration of diabe-tes has been shown to be a risk factor of

peripheral neuropathy, with prevalenceranging from 20.8% in patients with di-abetes for ,5 years to 36.8% in thosewith diabetes for .10 years (23). In ad-dition, poor glycemic control increasesthe risk and progression of diabetic neu-ropathy (2,3). The PROMISE cohort com-prised participants at risk for type 2diabetes. The alarmingly high preva-lence of peripheral neuropathy in indi-viduals with prediabetes suggests thatthe duration and degree of dysglycemiabefore the clinical diagnosis of diabetesmay be a crucial factor in the early initi-ation of neurological changes.

Despite well-established evidence re-garding the risk factors for diabeticneuropathy, a literature gap exists fornormal glucose tolerant and predia-betic populations. Nonglycemic pa-rameters of the metabolic syndrome,including abdominal adiposity and dys-lipidemia, have been shown to be asso-ciated with peripheral neuropathy inpatients with prediabetes (10,11). Incontrast, we did not observe indepen-dent associations of metabolic parame-ters with peripheral neuropathy ornerve dysfunction, which may be re-lated to the high prevalence of theserisk factors in the current study popula-tion. However, the findings confirm pre-vious observations that prediabetes, inparticular impaired glucose tolerance, isindependently associated with both pe-ripheral neuropathy and nerve dysfunc-tion (25,26). Hyperglycemia may lead toneurological pathology in part throughincreasing oxidative stress and activa-tion of the protein kinase C and polyolpathways, both of which exert neuro-toxic effects (27,28). Besides hypergly-cemia, dyslipidemia, microvasculardysfunction, and metabolic syndromehave all been shown to initiate neurop-athy in prediabetes through a commonmechanism of oxidative stress (29).

In clinical practice, the neurothesiom-eter has been used to monitor the pro-gression of peripheral neuropathyrather than as a tool for diagnosing pe-ripheral neuropathy. To this point, ele-vated VPTs have been shown to beassociated with an increased risk of sub-sequent foot ulceration (21,22). Theneurothesiometer, being minimally in-vasive and returning results in ,3 min,may be more sensitive for the diagnosisof peripheral neuropathy, particularly inpatients with mild neuropathy, such as

Figure 1—Prevalence of peripheral neuropathy by glucose tolerance status comparing differentdefinitions of MNSI at 3-year follow-up (A) and by progression of glucose intolerance betweenbaseline and 3-year follow-up (B). The P value refers to the overall comparison across subgroups.DM, diabetes mellitus; NFG, normal fasting glucose; NGT, normal glucose tolerance.

care.diabetesjournals.org Lee and Associates 797

those with prediabetes (21). However,no standardized VPT cutoff value to de-fine the presence of peripheral neurop-athy has been published to date.The findings of this study have impor-

tant research and clinical implications.From a research perspective, althoughthe documentation of impaired glucosetolerance in patients with idiopathicneuropathy (11) has raised the concernthat clinically significant nerve dysfunc-tionmay be present before diabetes on-set, the current study adds that thepresence of risk factors for diabetesonset alone conveys a high risk for

neuropathy. This risk is further exacer-bated by the presence of impaired glu-cose tolerance or new-onset diabetes.These findings emphasize the need todetermine etiological factors for neurop-athy onset independent of glucose me-tabolism and to determine the thresholdof glucose exposure below that of a di-agnosis of impaired glucose intolerancethat may promote nerve injury. From aclinical perspective, previous evidencehas suggested that older patients withdiabetes or prediabetes are often un-aware of their peripheral neuropathyand that they receive infrequent foot

examinations and inadequate preventiveeducation on foot care (30). These obser-vations coupledwith the high prevalenceof peripheral neuropathy and nerve dys-function stress the urgent need for earlydetection and intervention not only inpopulations with established diabetesbut also in those with risk factors suchas prediabetes or metabolic syndrome.

The strengths of the PROMISE studyinclude a well-characterized multiethniccohort as well as detailed measurementof peripheral neuropathy and nerve dys-function. However, several potentiallimitations should also be considered.

Figure 2—Severity of nerve dysfunction by glucose tolerance status at 3-year follow-up (A), by progression of glucose intolerance between baselineand 3-year follow-up (B), by the presence of the metabolic syndrome (C), and by the number of metabolic disorders (D). The P value refers to theoverall comparison across subgroups. DM, diabetes mellitus; MetS, metabolic syndrome; NFG, normal fasting glucose; NGT, normal glucosetolerance.

Table 2—Association of glucose tolerance status with peripheral neuropathy and nerve dysfunction in the PROMISE cohort

MNSI score** VPT**

b (95% CI) P value b (95% CI) P value

Prediabetes vs. NFG/NGT (cross-sectional at 3-year follow-up)Unadjusted 0.38 (0.11, 0.65) 0.005 0.20 (0.076, 0.32) 0.001Multivariable adjusted* 0.34 (0.075, 0.61) 0.01 0.14 (0.044, 0.24) 0.004

Progression of glucose intolerance vs. sustained NFG/NGTUnadjusted 0.40 (0.14, 0.67) 0.003 0.20 (0.085, 0.32) 0.001Multivariable adjusted* 0.37 (0.10, 0.63) 0.007 0.15 (0.052, 0.25) 0.002

NFG, normal fasting glucose; NGT, normal glucose tolerance. *Adjusted for age, sex, ethnicity, and height. **MNSI scores (with a total score of 10)and VPTs were modeled as continuous variables.

798 Neuropathy in Individuals at High Risk for Type 2 Diabetes Diabetes Care Volume 38, May 2015

Using a single oral glucose tolerance testto diagnose prediabetes has beenshown to have low reproducibility (31).Although the neurothesiometer hasbeen validated to assess nerve dysfunc-tion, limited evidence supports usingonly VPT to assess the severity of periph-eral neuropathy. In addition, we did nothave information on other knowncauses of peripheral neuropathy or col-lect information on whether partici-pants were symptomatic. Finally, theresults may apply only to individualswith similar demographic andmetaboliccharacteristics.In conclusion, this study documents

that the prevalence of peripheral neu-ropathy in patients with prediabetesand additional risk factors for diabetesonset is as high as in those with newlydiagnosed (or early onset) diabetes andconfirms that risk factors for diabeticneuropathy, such as age, height, andglycemia, are associated with peripheralneuropathy, even in the absence of aclinical diagnosis of diabetes. The resultssupport efforts toward early detectionand intervention as a means to reducethe incidence of peripheral neuropathyand its downstream consequences inhigh-risk populations.

Acknowledgments. The authors thank thestudy subjects for their ongoing participationand the research nurses for their expert tech-nical assistance.Funding. R.R. is supported by a Heart and StrokeFoundation of Ontario Mid-Career InvestigatorAward. A.J.H. holds a Tier II Canada ResearchChair in Diabetes Epidemiology. PROMISE issupported by grants from the Canadian DiabetesAssociation, the Connaught Fund from the Uni-versity of Toronto, and the Canadian Institutes ofHealth Research.Duality of Interest. B.A.P. participated on anadvisory board for Neurometrix, Inc. S.B.H. hasreceived consultation fees from or participated onadvisory boards for Merck, Novo Nordisk, Sanofi,Boehringer Ingelheim, Lilly, AstraZeneca, Janssen,Takeda, and Abbott and received funding fromSanofi, Novo Nordisk, and AstraZeneca. H.C.G.has received consulting fees from Sanofi, NovoNordisk, Lilly, Bristol-Myers Squibb, Roche,AstraZeneca, Abbot, GlaxoSmithKline, andBayer; lecture fees from Sanofi and Bayer;and support for research or continuing educa-tion through his institution from Sanofi, Lilly,Merck, Novo Nordisk, Boehringer Ingelheim,Bristol-Myers Squibb, and AstraZeneca. No otherpotential conflicts of interest relevant to thisarticle were reported.Author Contributions. C.C.L. contributed tothe discussion and writing, review, and editingof the manuscript. B.A.P., S.K., and H.C.G.

contributed to the discussion and review andediting of the manuscript. S.B.H., R.R., and B.Z.researched data and contributed to the discus-sion and review and editing of the manuscript.A.J.H. researched data and contributed to thediscussion and writing, review, and editing ofthe manuscript. A.J.H. 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.Prior Presentation. Parts of this study werepresented in abstract form at the 74th ScientificSessions of the American Diabetes Association,San Francisco, CA, 13–17 June 2014.

References1. Boulton AJ, Vileikyte L, Ragnarson-TennvallG, Apelqvist J. The global burden of diabeticfoot disease. Lancet 2005;366:1719–17242. Adler AI, Boyko EJ, Ahroni JH, Stensel V,Forsberg RC, Smith DG. Risk factors for diabeticperipheral sensory neuropathy. Results of theSeattle Prospective Diabetic Foot Study. Diabe-tes Care 1997;20:1162–11673. Sorensen L, Molyneaux L, Yue DK. Insen-sate versus painful diabetic neuropathy: theeffects of height, gender, ethnicity andglycaemic control. Diabetes Res Clin Pract2002;57:45–514. UK Prospective Diabetes Study (UKPDS)Group. Intensive blood-glucose control with sul-phonylureas or insulin compared with conven-tional treatment and risk of complications inpatients with type 2 diabetes (UKPDS 33). Lan-cet 1998;352:837–8535. Spijkerman AM, Dekker JM, Nijpels G, et al.Microvascular complications at time of diagno-sis of type 2 diabetes are similar among diabeticpatients detected by targeted screening and pa-tients newly diagnosed in general practice: theHoorn screening study. Diabetes Care 2003;26:2604–26086. Kostev K, Jockwig A, Hallwachs A, RathmannW. Prevalence and risk factors of neuropathy innewly diagnosed type 2 diabetes in primary carepractices: a retrospective database analysis inGermany and U.K. Prim Care Diabetes 2014;8:250–2557. Harris MI, Klein R, Welborn TA, KnuimanMW. Onset of NIDDM occurs at least 4-7 yrbefore clinical diagnosis. Diabetes Care 1992;15:815–8198. Bongaerts BW, Rathmann W, Kowall B,et al. Postchallenge hyperglycemia is posi-tively associated with diabetic polyneuropathy:the KORA F4 study. Diabetes Care 2012;35:1891–18939. Gregg EW, Gu Q, Williams D, et al. Preva-lence of lower extremity diseases associatedwith normal glucose levels, impaired fastingglucose, and diabetes among U.S. adults aged40 or older. Diabetes Res Clin Pract 2007;77:485–48810. Ziegler D, Rathmann W, Dickhaus T,Meisinger C, Mielck A; KORA Study Group. Prev-alence of polyneuropathy in pre-diabetes anddiabetes is associated with abdominal obesityand macroangiopathy: the MONICA/KORAAugsburg Surveys S2 and S3. Diabetes Care2008;31:464–469

11. Smith AG. Impaired glucose tolerance andmetabolic syndrome in idiopathic neuropathy.J Peripher Nerv Syst 2012;17(Suppl. 2):15–2112. Hanley AJ, Retnakaran R, Qi Y, et al. Associ-ation of hematological parameters with insulinresistance and b-cell dysfunction in nondiabeticsubjects. J Clin Endocrinol Metab 2009;94:3824–383213. Moghtaderi A, Bakhshipour A, Rashidi H.Validation ofMichigan neuropathy screening in-strument for diabetic peripheral neuropathy.Clin Neurol Neurosurg 2006;108:477–48114. Feldman EL, Stevens MJ, Thomas PK,Brown MB, Canal N, Greene DA. A practicaltwo-step quantitative clinical and electrophys-iological assessment for the diagnosis andstaging of diabetic neuropathy. Diabetes Care1994;17:1281–128915. Bril V, Kojic J, Ngo M, Clark K. Comparisonof a neurothesiometer and vibration in measur-ing vibration perception thresholds and rela-tionship to nerve conduction studies. DiabetesCare 1997;20:1360–136216. Claus D, Mustafa C, Vogel W, Herz M,Neundorfer B. Assessment of diabetic neuropa-thy: definition of norm and discrimination ofabnormal nerve function. Muscle Nerve 1993;16:757–76817. World Health Organization. Definition, Di-agnosis and Classification of Diabetes Mellitusand Its Complications. Part 1: Diagnosis andClassification of Diabetes Mellitus. Geneva,World Health Organization, 199918. Alberti KG, Eckel RH, Grundy SM, et al.; In-ternational Diabetes Federation Task Force onEpidemiology and Prevention; National Heart,Lung, and Blood Institute; American Heart Asso-ciation; World Heart Federation; InternationalAtherosclerosis Society; International Associa-tion for the Study of Obesity. Harmonizing themetabolic syndrome: a joint interim statementof the International Diabetes Federation TaskForce on Epidemiology and Prevention; Na-tional Heart, Lung, and Blood Institute; Ameri-can Heart Association; World Heart Federation;International Atherosclerosis Society; and Inter-national Association for the Study of Obesity.Circulation 2009;120:1640–164519. SacksDB. A1C versus glucose testing: a com-parison. Diabetes Care 2011;34:518–52320. Kelly AS, Steinberger J, Jacobs DR, Hong CP,Moran A, Sinaiko AR. Predicting cardiovascularrisk in young adulthood from the metabolic syn-drome, its component risk factors, and a clusterscore in childhood. Int J Pediatr Obes 2011;6:e283–e28921. Bril V, Perkins BA. Comparison of vibrationperception thresholds obtained with the Neuro-thesiometer and the CASE IV and relationship tonerve conduction studies. Diabet Med 2002;19:661–66622. YoungMJ,Breddy JL, VevesA, BoultonAJ. Theprediction of diabetic neuropathic foot ulcerationusing vibration perception thresholds. A prospec-tive study. Diabetes Care 1994;17:557–56023. Young MJ, Boulton AJ, MacLeod AF,Williams DR, Sonksen PH. A multicentre studyof the prevalence of diabetic peripheral neurop-athy in the United Kingdom hospital clinic pop-ulation. Diabetologia 1993;36:150–15424. Cheng YJ, Gregg EW, Kahn HS, Williams DE,De Rekeneire N, Narayan KM. Peripheral

care.diabetesjournals.org Lee and Associates 799

insensate neuropathyda tall problem forUS adults? Am J Epidemiol 2006;164:873–88025. Dyck PJ, Dyck PJ, Klein CJ, Weigand SD.Does impaired glucose metabolism causepolyneuropathy? Review of previous stud-ies and design of a prospective controlledpopulation-based study. Muscle Nerve 2007;36:536–54126. Sumner CJ, Sheth S, Griffin JW, CornblathDR, Polydefkis M. The spectrum of neuropathy

in diabetes and impaired glucose tolerance.Neurology 2003;60:108–11127. Yagihashi S, Mizukami H, Sugimoto K.Mechanism of diabetic neuropathy: where arewe now and where to go? J Diabetes Investig2011;2:18–3228. Dobretsov M, Romanovsky D, Stimers JR.Early diabetic neuropathy: triggers and mecha-nisms. World J Gastroenterol 2007;13:175–19129. Papanas N, Vinik AI, Ziegler D. Neurop-athy in prediabetes: does the clock start

ticking early? Nat Rev Endocrinol 2011;7:682–69030. Bongaerts BW, Rathmann W, Heier M,et al. Older subjects with diabetes and pre-diabetes are frequently unaware of havingdistal sensorimotor polyneuropathy: theKORA F4 study. Diabetes Care 2013;36:1141–114631. American Diabetes Association. Standardsof medical care in diabetesd2014. DiabetesCare 2014;37(Suppl. 1):S14–S80

800 Neuropathy in Individuals at High Risk for Type 2 Diabetes Diabetes Care Volume 38, May 2015