Prevalence of Distal Symmetrical Polyneuropathy by Diabetes Prevention Program Treatment Group, Diabetes Status, Duration of Diabetes, and Cumulative Glycemic Exposure

Diabetic neuropathies include a heterogeneous group of neuropathic conditions that can occur in patients with diabetes after other causes have been excluded, and distal symmetric polyneuropathy (DSPN) is the most common form of diabetic neuropathy (1). The prevalence of DSPN among adults with diabetes in the U.S. is estimated to be 28%. Approximately 50% of adults with diabetes will develop DSPN over their lifetimes (2). DSPN can cause symptoms that adversely impact quality of life and increase the risk for both foot ulcers and nontraumatic lower-extremity amputations (3,4).

Given the substantial morbidity associated with diabetic DSPN, better strategies are needed to prevent this complication. Currently, the few medications approved for the treatment of diabetic DSPN target only the troubling symptoms and not the causes of DSPN. Results of clinical trials in populations with both type 1 and type 2 diabetes have demonstrated that improved glycemic control can delay or prevent the development of diabetic DSPN (5,–,7). Since DSPN can also develop in people with prediabetes and metabolic syndrome, we questioned whether DSPN can be prevented early in the course of dysglycemia (8,9).

The Diabetes Prevention Program (DPP) enrolled adults ≥25 years of age with overweight or obesity, elevated fasting glucose, and impaired glucose tolerance who were at high risk for developing type 2 diabetes. Participants were randomized to an intensive lifestyle (ILS), metformin, or placebo intervention to reduce progression to diabetes. Both active interventions were effective in preventing or delaying the onset of diabetes, as previously reported (10,11). Those randomized to ILS were taught behavioral self-management strategies to lose ≥7% of initial body weight and engage in 150 min of moderate-intensity physical activity per week. Because during DPP, ILS lowered weight, central adiposity, and blood pressure, and improved glycemia and lipids (compared with metformin and placebo), all factors previously demonstrated to be independent risk factors for DSPN, we hypothesized that this intervention might delay or prevent DSPN (9,12,13). The potential impact of metformin on DSPN was less clear. Although effective in preventing or delaying the onset of diabetes, it is known to result in lower vitamin B₁₂ levels, which could contribute to DSPN (10,14). Investigators of observational studies of metformin and meta-analyses of those studies have reported either no effect or adverse effects of metformin on DSPN, but few clinical trials have adequately addressed this question (15,16). Therefore, the DPP and Diabetes Prevention Program Outcomes Study (DPPOS) presented an opportunity to evaluate the effects of both ILS and metformin in comparisons with a placebo intervention on DSPN. Accordingly, ∼21 years after DPP randomization, a comprehensive evaluation of DSPN was performed that included assessment of DSPN symptoms, assessment of pinprick sensation to evaluate small nerve fiber function, vibration sensation to assess large fiber function, and 10-g monofilament testing to assess protective sensation. DSPN was considered to be present if DSPN symptoms were present or if small fiber, large fiber, or protective sensation was abnormal. Our goal was to evaluate the impact of the DPP interventions and other risk factors on the development of clinically determined DSPN.

In the DPP, 3,234 participants at high risk for developing type 2 diabetes were randomized to ILS, masked metformin 850 mg twice daily, or a matching placebo twice daily. Details of the enrollment criteria, design, and methods have previously been published (17,– 19). The institutional review boards at each of the 27 study sites reviewed and approved the study, and all participants provided written informed consent. The DPP interventions were stopped by the study sponsor on the advice of the Data Safety Monitoring Board in 2001 due to the efficacy of the active interventions compared with placebo in reducing the risk of incident diabetes. All participants were informed of the results, metformin and placebo were unmasked, and all were offered a modified lifestyle program over 6 months; subsequently, 2,779 participants consented to continue participation in the DPPOS. All DPPOS participants were offered quarterly lifestyle information sessions. More ILS reinforcement sessions were offered semiannually to those originally randomized to ILS. Open-label study metformin 850 mg twice daily was provided to those originally randomized to receive metformin, and the placebo was discontinued.

The primary outcome for this analysis was defined as the presence of symptoms and/or signs of DSPN at DPPOS year 17. Symptoms of DSPN were assessed with the Michigan Neuropathy Screening Instrument (MNSI), a self-administered 15-question questionnaire. A score of ≥4 abnormal responses indicates the presence of clinical neuropathy (20). Signs of DSPN were assessed with pinprick, vibratory, and 10-g monofilament testing. These assessments have been validated with nerve conduction studies and neurothesiometer (21,22). Pinprick sensation was assessed with the Owen Mumford Neuropen Neurotip applied four times in an arrhythmic manner to the dorsal aspect of the great toes bilaterally. For each undetected application, 1 point was given. A score ≥5 out of a maximum score of 8 was considered abnormal. Vibratory sensation was tested with a Rydel-Seiffer 128-Hz graduated tuning fork applied twice over the dorsal aspect of the distal interphalangeal joint of the great toes bilaterally. Each test was scored as follows: 2, no vibration was felt; 1, vibration sensation was lost when the tuning fork indicator was ≤4; and 0, vibration sensation was lost with the tuning fork indicator >4. A cumulative score of ≥5 out of a maximum score of 8 was considered abnormal. Protective sensation was assessed with a 10-g Owen Mumford Neuropen 10-gm monofilament, applied 10 times to the dorsal aspect of each great toe bilaterally. Protective sensation was considered absent if fewer than eight applications were detected at each great toe. Signs of DSPN were considered to be present if any test of pinprick, vibration, or protective sensation was abnormal.

Self-reported demographic variables, medical history, and lifestyle factors were ascertained with questionnaires administered at annual and semiannual study visits from DPP baseline through DPPOS year 17. Data collected from these questionnaires between baseline and DPPOS year 17 were collated for ascertainment of any reported history of gastric surgery, thyroid disease, B₁₂ supplement use, and cumulative smoking history. Cancer diagnoses (excluding nonmelanoma skin cancer) were adjudicated according to the Surveillance, Epidemiology, and End Results (SEER) Program guidelines. Information about concomitant prescription medication use was collected at annual study visits, including use of metformin prescribed by out-of-study providers, and cumulative metformin use from DPP baseline through DPPOS year 17 was quantified with use of these data and data on study drug compliance for participants randomized to metformin. Height and weight were measured with standardized procedures at year 17 for calculation of BMI. Waist circumference was measured at DPPOS year 16. Information on alcohol consumption (drinks per week) was also collected at year 17. Blood was drawn at DPPOS year 16 for vitamin B₁₂ and fasting triglyceride levels and at DPPOS year 17 for assessment of estimated glomerular filtration rate (eGFR), which was calculated with the Chronic Kidney Disease Epidemiology Collaboration equation (23). Fasting plasma glucose levels were obtained semiannually through 2014, and fasting and 2-h post–75-g oral glucose load plasma glucose levels were obtained annually throughout the study. Diabetes was diagnosed according to fasting glucose ≥126 mg/dL or 2-h glucose ≥200 mg/dL, and a repeat test was required for confirmation. Hemoglobin A_1c (HbA_1c) was measured at baseline, 6 months, and annually throughout the DPP and DPPOS. Duration of diabetes was calculated, and time-weighted HbA_1c was defined as the mean HbA_1c throughout the DPP and DPPOS.

DSPN was considered to be present if symptoms were present or if small fiber, large fiber, or protective sensation was abnormal. Descriptive statistics for relevant variables were computed with mean (SD) or median (interquartile range) for numerical variables and frequency (%) for categorical variables. Bivariate tests for association, including ANOVA, Kruskal-Wallis, and Pearson χ² test for independence, between each variable and treatment group were conducted. The prevalence of DSPN at DPPOS year 17 was calculated for each randomized group. The odds of DSPN associated with the DPP randomized treatment was assessed with logistic regression models including adjustment for age, sex, and race/ethnicity. This association was examined further in multivariable logistic regression models with adjustment additionally for weight, height, smoking history, eGFR, triglycerides, and low serum B₁₂ or B₁₂ supplement use. Effect modification by age, sex, and race/ethnicity on the association between DSPN at DPPOS year 17 (∼21 years since enrollment in the DPP) and DPP treatment group was planned a priori to be evaluated in logistic regression models with results presented for significant interactions indicated by P values <0.05. The associations between DSPN and diabetes status, diabetes duration, and cumulative HbA_1c were also explored with use of multivariable regression models that included adjustment for age, sex, race/ethnicity, DPP randomization group, height, weight, smoking history, eGFR, triglycerides, and low serum B₁₂ or B₁₂ supplement use. Because metformin was often prescribed by participants’ primary care providers after participants developed diabetes, a secondary analysis was performed for evaluation of the association between cumulative metformin exposure and DSPN with use of a logistic regression model that included adjustment for age, sex, race/ethnicity, height, weight, smoking history, eGFR, triglycerides, and low serum B₁₂ or B₁₂ supplement use. Relevant odds ratios (ORs) along with their corresponding 95% CIs and P values are reported. For the association between DSPN and cumulative metformin use, diabetes duration, and cumulative HbA_1c, we report both unstandardized (in text) and standardized (i.e., scaling continuous predictors by their sample SDs prior to entering them in the models [Table 3]) model-based ORs.

By the DPPOS year 17 study visit (20–23 years after enrollment in the DPP; mean 21 years), 348 participants had died and 53 participants had withdrawn from DPPOS. The 1,792 participants who attended the DPPOS year 17 visit and had DSPN assessments were included in this study (Supplementary Fig. 1). Compared with participants included in this analysis, those not included were at baseline on average older and more likely to be male, to be White, and to report a history of smoking. Mean weight and height was higher and mean eGFR was lower for those not included (Supplementary Table 1).

At the year 17 visit, mean (SD) age was 70.4 (9.1) years, 70.6% of participants were women, and 50.9% were White (Table 1). Participants in the metformin group reported marginally more alcohol use and had a higher prevalence of either low B₁₂ levels or use of B₁₂ supplements than the ILS or placebo group participants. Self-reported histories of thyroid disease (10.0%) and gastric surgery (4.7%) as well as adjudicated cancers (14.2%) were not significantly different across randomized DPP treatment groups. The overall prevalence of diabetes was 68.5%. There was a significantly higher prevalence of diabetes, longer duration of diabetes, and higher cumulative HbA_1c exposure in the placebo group compared with the metformin and ILS groups, consistent with the DPP and DPPOS results (10,11).

The prevalence of DSPN in the entire cohort was 21.7%. Neither the crude prevalence of DSPN nor the components of DSPN differed across the randomized DPP treatment groups (Table 2). The crude prevalence was marginally higher among participants with diabetes (22.7%) compared with those without diabetes (19.6%). Symptoms of DSPN were more prevalent than abnormal pinprick, vibration, and protective sensation.

There was no overall association between DPP treatment group and DSPN based on the logistic regression model with adjustment for age, sex, and race (ILS OR 0.95 [95% CI 0.71, 1.26], and metformin OR 0.92 [95% CI 0.69, 1.22]) (Supplementary Table 2). However, there was significant interaction by age for the association between DPP treatment group and DSPN (P = 0.046). Figure 1 shows how the estimated ORs for DSPN varied by age at the DPPOS year 17 visit in comparisons of ILS and placebo groups (left panel) and metformin and placebo groups (right panel). The OR for DSPN decreased by 17.4% (95% CI 3.0, 29.3; P = 0.019) with each 5-year increase in age for ILS participants in comparisons with placebo participants. The OR for DSPN for ILS compared with placebo was <1 for older ages and >1 for younger ages as shown in Fig. 1, suggesting a protective effect of ILS on DSPN for older participants and a possible detrimental effect of ILS relative to placebo on DSPN for younger participants, although the sample was not large enough for assessment of the statistical significance of treatment effects in specific subgroups. Age did not influence the null association between metformin versus placebo and DSPN (P = 0.592). In addition, there was no significant association between cumulative metformin use and DSPN in multivariable logistic regression models (OR per SD 1.09 [95% CI 0.97, 1.23]) (Table 3).

There was a higher likelihood of DSPN associated with the diagnosis of diabetes at DPPOS year 17 (OR vs. no diabetes 1.40 [95% CI 1.07, 1.84]; P < 0.001), longer duration of diabetes (OR 1.04 per year [95% CI 1.02, 1.05]; P < 0.001), and higher cumulative mean HbA_1c exposure (OR 1.85 per 1% increase [95% CI 1.54, 2.21]; P < 0.001) after adjustment for covariates. Cumulative HbA_1c exposure conferred the greatest risk based on the standardized ORs (Table 3). Supplementary Tables 2 and 3 provide ORs with 95% CIs from logistic models with adjustment for different sets of covariates.

The DPPOS, which is the long-term follow-up of the DPP, provides an opportunity to better understand risk factors for the development of DSPN and to evaluate the efficacy of the DPP interventions in mitigating the risk of developing DSPN. In this population of adults at risk for and with diabetes, the overall prevalence of DSPN was 21.7% (24). Individuals with diabetes and those with longer duration of diabetes and greater cumulative glycemic exposure had higher odds of DSPN. However, despite the fact that DPP interventions were effective in reducing incident diabetes, we saw only limited evidence of treatment group differences in the prevalence of DSPN.

Among the 68.5% of DPPOS participants who developed diabetes by the DPPOS year 17, the prevalence of DSPN was 22.7%. This falls between a range of DSPN prevalence in prior reports from 8.3% in younger adults with newly diagnosed diabetes, with use of nerve conduction studies, to 44% in older adults with longer durations of diabetes, with use of both symptoms and clinical exam findings (25–29). Like these prior reports, our findings showed a higher prevalence of DSPN associated with a longer duration of diabetes, which may explain why the prevalence of DSPN for participants who had developed diabetes in this study (median diabetes duration 9 years) was higher than that reported for younger adults with newly diagnosed diabetes but lower than that reported among older adults with a longer duration of diabetes. The method of DSPN assessment can also contribute to differences in DSPN prevalence. The observed overall prevalence of DSPN of 21.7% in the entire DPPOS study population at DPPOS year 17 is higher than the prevalence of neuropathy we reported previously, consistent with longer follow-up and more comprehensive assessment of DSPN with the MNSI and pinprick and vibratory and protective sensation tests at DPPOS year 17 rather than the prior definition of neuropathy, based only on protective sensation as assessed with a monofilament exam (24). While in prior reports investigators noted greater likelihood of DSPN with poorer glycemic control, this study adds to the literature with longitudinal assessments of HbA_1c. Greater cumulative glycemic exposure conferred an even higher risk for DSPN than a diagnosis of diabetes or duration of diabetes. Lower cumulative glycemic exposure may have contributed to the lower DSPN prevalence of 19.6% among participants without diabetes compared with the DSPN prevalence (22.7%) among participants with diabetes at DPPOS year 17. Since this is only marginally lower than the prevalence of DSPN among those who had developed diabetes, greater awareness and screening for DSPN may be indicated for individuals with prediabetes.

Given the increased risk of DSPN with greater cumulative glycemic exposure, interventions to lower glucose exposure would be expected to mitigate this risk. The Kumamoto study showed that mean HbA_1c lowered to 7.1% with multiple insulin injection therapy compared with HbA_1c of 9.4% with conventional insulin injection therapy in patients with type 2 diabetes resulted in significant increases in both motor and sensory median nerve conduction velocities in contrast to decreases observed in the conventional insulin injection therapy group (30). Subsequently, the results of the Action to Control Cardiovascular Risk in Diabetes (ACCORD) trial demonstrated that the intensive glucose-lowering intervention targeting HbA_1c ≤6% compared with standard glucose lowering with a target HbA_1c of 7.0–7.9% reduced the risk of neuropathy as assessed with the MNSI, light touch sensation, and ankle reflexes (7). Neither the UK Prospective Diabetes Study (UKPDS) nor the Veterans Affairs Diabetes Trial (VADT) showed significant reductions in the risk of DSPN with intensive glucose control interventions; however, these studies did not include comprehensive or sensitive assessments of DSPN (31,32). In the Bypass Angioplasty Revascularization Investigation in Type 2 Diabetes (BARI 2D) study, investigators assessed DSPN with the MNSI and also found no difference in the prevalence of DSPN between insulin sensitizer treatment and insulin-providing treatments after 4 years; however, in a subgroup analysis of participants without DSPN at baseline, they reported a greater reduction in incident DSPN with insulin-sensitizing compared with insulin-providing treatments even after adjustment for differences in glycemia (33). While this finding raises the possibility of a benefit of insulin sensitizers in prevention of DSPN, in our study we did not find a difference in DSPN between metformin and placebo groups using a more comprehensive assessment of neuropathy that included components of the MNSI.

In contrast to these prior trials in adults with established type 2 diabetes, the DPP enrolled a population at high risk for diabetes, and it is likely that the magnitude of differences in severity and glycemic burden of incident diabetes across randomized groups was insufficient to allow detection of an intervention effect on DSPN. Although the ILS and metformin groups had lower cumulative glycemic exposure than the placebo group, the degree of glycemic separation in the DPPOS was much less than in the intensive glucose-lowering trials in patients with established type 2 diabetes. It is also possible that the greater prevalence of low B₁₂ levels and higher use of B₁₂ supplements or the marginally higher alcohol consumption in the metformin group could have abrogated any glucose-lowering benefit of metformin for DSPN. However, the likelihood of having DSPN within the metformin group compared with placebo remained unchanged even in multivariable models accounting for differences in diabetes status, duration, cumulative glycemic exposure, B₁₂ levels, and vitamin B₁₂ supplement use.

Small trials of aerobic and resistive exercise in adults with type 2 diabetes have shown a benefit for diabetic neuropathy, possibly through improvement in glycemia or vascular function (34–37). However, the evidence for lowering the risk of DSPN with lifestyle interventions for individuals with prediabetes is more limited. In the China Da Qing Diabetes Prevention Outcomes Study (CDQDPOS) there was not any difference in the prevalence of neuropathy, based on monofilament testing or history of lower-extremity amputation, gangrene, or ulceration, between the combined lifestyle interventions groups compared with the control group after 20 years (38). However, a smaller trial of 32 adults with prediabetes showed improvements in more sensitive neuropathy measures including cutaneous reinnervation and foot sweat volume with a 1-year lifestyle intervention (39). Overall, we found no difference in the prevalence of DSPN for ILS, which included a physical activity component, compared with placebo. However, there was a significant age interaction in this association, with a lower OR for DSPN with ILS compared with placebo among older adults and a higher OR among younger adults. This finding may be explained, in part, by the greater efficacy of ILS in diabetes prevention for older adults who also had the greatest weight loss and physical activity and the lower efficacy of ILS in younger participants who had the least weight loss and increase in physical activity (10,40). However, the difference in ORs by age for DSPN with ILS compared with placebo should be viewed as exploratory because subsets of participants defined according to age, with smaller sample sizes, were limited for evaluating effects of ILS on DSPN.

The strengths of this study include the implementation of comprehensive measures of DSPN including assessment of symptoms, small fiber and large fiber function, and protective sensation. Our longitudinal assessments of glycemia allowed us to explore the role of cumulative glycemic exposure in the risk of DSPN for individuals who did or did not develop diabetes. Furthermore, the randomized DPP interventions provided an opportunity to determine the effect of ILS and metformin on the likelihood of developing DSPN.

There are also limitations to this study. DSPN was not assessed at baseline. It is possible that differences in the baseline prevalence of DSPN across DPP treatment groups may have limited the ability to detect a difference in DSPN by DPP treatment group at DPPOS year 17. The participants included in this analysis were healthier at baseline than those who were not available for the analysis; therefore, the overall prevalence of DSPN reported in this manuscript may be lower than the true prevalence due to survival bias. Not all secondary causes of peripheral neuropathy were directly assessed and excluded; therefore, DSPN results reported in this manuscript cannot be interpreted as reflecting “diabetic DSPN.” However, reports of thyroid disease, gastric surgeries, and cancer that was potentially treated with neurotoxic chemotherapeutic agents were not highly prevalent. The frequencies of these conditions were not different across the randomized treatment groups and should not confound the association between DPP treatment groups and DSPN. There was crossover use of metformin in the ILS and placebo groups as a first-line agent for treatment of diabetes once incident diabetes was detected; however, this was tracked as a concomitant medication and in the secondary analysis that we performed here we did not find a significant association between cumulative metformin exposure and the likelihood of DSPN. While this null finding may seem reassuring, indicating that there are no adverse neurological effects of metformin due to vitamin B₁₂ deficiency, this is not conclusive because vitamin B₁₂ levels were measured in those randomized to metformin or placebo in DPPOS year 9, several years prior to this DSPN assessment, and those with low B₁₂ levels were advised to take B₁₂ supplements. Another limitation includes the provision of a modified lifestyle modification program and quarterly lifestyle sessions to all participants after the DPP ended. This may have reduced our ability to detect an effect of the ILS on DSPN (40).

In conclusion, 21.7% of participants enrolled in the DPPOS had DSPN at year 17, and the crude prevalence of DSPN was only nominally higher for those who had developed diabetes than for those who had not developed diabetes. Cumulative glycemic exposure confers greater risk for DSPN than diabetes status and diabetes duration. Many participants who had levels of glycemia remaining below those used to diagnose diabetes still developed DSPN. This may warrant initiation of DSPN screening for those with prediabetes. ILS was more effective for older adults at risk for diabetes than in younger participants in decreasing the incidence of diabetes (10,11). This might account for the potentially greater benefit of ILS for DSPN in older than in younger participants. In further studies evaluating the effects of lifestyle interventions on DSPN, the possibility of effect modification by age should be considered.

Acknowledgments. The Diabetes Prevention Program Research Group gratefully acknowledges the commitment and dedication of the participants of the DPP and DPPOS. A complete list of centers, investigators, and staff can be found in Supplementary Material.

S.E.K. is an editor of Diabetes Care but was not involved in any of the decisions regarding review of the manuscript or its acceptance.

Funding. Research reported in this publication was supported by the National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK) of the National Institutes of Health (NIH) under award nos. U01 DK048489, U01 DK048339, U01 DK048377, U01 DK048349, U01 DK048381, U01 DK048468, U01 DK048434, U01 DK048485, U01 DK048375, U01 DK048514, U01 DK048437, U01 DK048413, U01 DK048411, U01 DK048406, U01 DK048380, U01 DK048397, U01 DK048412, U01 DK048404, U01 DK048387, U01 DK048407, U01 DK048443, and U01 DK048400, by providing funding during the DPP and DPPOS to the clinical centers and the Coordinating Center for the design and conduct of the study and for the collection, management, analysis, and interpretation of the data. Funding was also provided by the Eunice Kennedy Shriver National Institute of Child Health and Human Development, the National Institute on Aging, the National Eye Institute, the National Heart, Lung, and Blood Institute, the National Cancer Institute, the Office of Research on Women’s Health, the National Institute on Minority Health and Health Disparities, the Centers for Disease Control and Prevention, and the American Diabetes Association. The Southwestern American Indian Centers were supported directly by the NIDDK, including its Intramural Research Program, and the Indian Health Service. The General Clinical Research Center program, National Center for Research Resources, and the Department of Veterans Affairs supported data collection at many of the clinical centers. Merck KGaA provided medication for the DPPOS. The DPP and DPPOS have also received donated materials, equipment, or medicines for concomitant conditions from Bristol-Myers Squibb, Parke-Davis, LifeScan, Health o meter, Hoechst Marion Roussel, Merck-Medco Managed Care, Merck and Co., Nike Sports Marketing, Slim Fast Foods Co., and Quaker Oats Company. McKesson BioServices, Matthews Media Group, and the Henry M. Jackson Foundation provided support services under subcontract with the Coordinating Center. The sponsor of this study was represented on the Steering Committee and played a part in study design, how the study was done, and publication.

The opinions expressed are those of the study group and do not necessarily reflect the views of the funding agencies. The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH.

Duality of Interest. C.G.L. was an employee in the Division of Diabetes, Endocrinology and Metabolic Diseases at the NIDDK at the NIH at the time this research was conducted and is now an employee of Pfizer. No other potential conflicts of interest relevant to this article were reported.

Author Contributions. C.G.L., A.C., S.L.E., J.P.C., D.D., R.B.G., S.E.K., W.C.K., M.T.M., N.H.W., and W.H.H. had access to all data. A.C. and S.L.E. are the guarantors of this work and, as such, had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Handling Editors. The journal editor responsible for overseeing the review of the manuscript was Matthew C. Riddle.