Randomized Controlled Trial of Physical Exercise in Diabetic Veterans With Length-Dependent Distal Symmetric Polyneuropathy

Abstract

Rationale: Physical exercise is an essential adjunct to the management of patients with type 2 diabetes mellitus. Therapeutic interventions that improve blood flow to peripheral nerves, such as exercise, may slow the progression of neuropathy in the diabetic patient. Aims: This randomized clinical trial was conducted to determine whether a structured program of aerobic, isokinetic strength, or the combination of aerobic-isokinetic strength exercise intervention alters peripheral nerve function in glycemic-controlled diabetic patients with advanced length-dependent distal symmetric polyneuropathy. Methods: Forty-five patients with type 2 diabetes mellitus exhibiting tight glycemic control (HbA1c intergroup range 7.2-8.0%) were randomized by block design across four experimental groups: sedentary controls (n = 12), aerobic exercise (n = 11), isokinetic strength (n = 11), or the combination of aerobic-isokinetic strength training (n = 11). Patients randomized to training groups exercised 3× per week for 12 weeks, whereas patients randomized to the sedentary control group received standard of care. To minimize attention and educational bias, all patients attended a 12-session health promotion educational series. At baseline, immediately following intervention, and again at 12-week post-intervention, detailed nerve conduction studies were conducted as a primary outcome measure. At these same intervals, all patients completed as secondary measures quantitative sensory testing, symptom-limited treadmill stress tests, and a Short-Form 36-Veterans Questionnaire (SF-36V). Results: Of the 45 patients randomized into this study, 37 (82%) had absent sural nerve responses, 19 (42%) had absent median sensory nerve responses, and 17 (38%) had absent ulnar sensory nerve responses. By comparison, responses from tibial nerves were absent in only three (7%) subjects while responses from peroneal nerves were absent in five (11%) subjects. Eleven (92%) of 12 patients that had volunteered to be biopsied exhibited abnormal levels of epidermal nerve fiber densities. Exercise, regardless of type, did not alter sensory or motor nerve electrodiagnostic findings among those patients exhibiting measurable responses (ANOVA). There was, however, a modest (p = 0.01) beneficial effect of exercise on sensory nerve function (Fisher's Exact Test). Importantly, the beneficial effect of exercise on sensory nerve function was enhanced (p = 0.03) during the post-intervention interval. In addition, three of six patients that had undergone exercise intervention exhibited a marked 1.9 ± 0.3-fold improvement in epidermal nerve fiber density. By comparison, none of three sedentary patients whom agreed to be biopsied a second time showed improvement in epidermal nerve fiber density. Compared to baseline values within groups, and compared with sedentary values across groups, neither aerobic, isokinetic strength, or the combination of aerobic-isokinetic strength exercise intervention altered peak oxygen uptake. Patients that underwent aerobic or the combined aerobic-isokinetic strength exercise intervention, however, demonstrated an increase in treadmill test duration that was sustained over the 12-week post-intervention period. Conclusion: A 12-week course of physical exercise, regardless of type, does not alter sensory or motor nerve electrodiagnostic findings. In a subset of patients, a short-term structured program of aerobic exercise may selectively improve sensory nerve fiber function. Large-scale exercise lifestyle intervention trials are warranted to further evaluate the impact of aerobic exercise on sensory nerve fiber function in diabetic neuropathic patients. Clinical Trial Registration: www.ClinicalTrials.gov, identifier NCT00955201.

Keywords: diabetes; exercise; human; metabolic; nerve conduction studies; peripheral nerve.

Figures

FIGURE 1

Flow diagram of study design.

FIGURE 2

Flow diagram of patient enrollment, randomization, and participation.

FIGURE 3

Exercise selectively elicits a sustained increase in SF-36V patient self-reported physical component score. Data shown are the means ± SD of self-reported responses from patients randomized to sedentary control (open bars) or combined exercise (solid bars) experimental groups. (A) Physical or (B) mental component scores are expressed as a percentage of weighted score (Supplementary Table S2) determined at entry into the study (baseline), immediately following intervention, and again at 12-week post-intervention of N responding patients, as indicated. Non-parametric data were analyzed by ANOVA on ranks (Kruskal–Wallis test). N.S., not significant. Note that one patient randomized to exercise did not complete the 12-week post SF-36V questionnaire.

FIGURE 4

Aerobic exercise selectively elicits improved treadmill endurance times. Data shown are the median, interquartile range, and minimum–maximum range of patient treadmill endurance times at (A) entry into the study (baseline), (B) immediately following intervention, and again at (C) 12-week post-intervention of N responding patients, as indicated. Parametric data were analyzed by two-way ANOVA with Tukey’s multiple comparison post hoc analysis. At baseline, all groups were statistically indistinguishable. Patients that had received aerobic training exhibited a strong statistical trend (p = 0.07) toward prolonged treadmill endurance times that was sustained (p = 0.02) when compared to those patients that had received strength training alone.

FIGURE 5

Group analyses of the effect of exercise on ulnar sensory nerve evoked electrodiagnostic parameters. Data shown are the median, interquartile range, and minimum–maximum range of latency, amplitude, and calculated conduction velocity elicited from antidromic stimulation of ulnar sensory nerves at entry into the study (baseline), immediately following intervention, and again at 12-week post-intervention of responding patients, as indicated. Parametric data were analyzed by two-way ANOVA with Tukey’s multiple comparison post hoc analysis. The total number of patients with measurable responses at baseline (sedentary controls, N = 8; exercise, N = 20), immediately following intervention (sedentary controls, N = 6; exercise, N = 16), and at 12-week post-intervention (sedentary controls, N = 7; exercise, N = 21) were group analyzed. At all three time points evaluated, there was no significant differences observed within and across experimental groups.

FIGURE 6

Group analyses of the effect of exercise on median sensory nerve evoked electrodiagnostic parameters. Data shown are the median, interquartile range, and minimum–maximum range of latency, amplitude, and calculated conduction velocity elicited from antidromic stimulation of median sensory nerves at entry into the study (baseline), immediately following intervention, and again at 12-week post-intervention of responding patients, as indicated. Parametric data were analyzed by two-way ANOVA with Tukey’s multiple comparison post hoc analysis. The total number of patients with measurable responses at baseline (sedentary controls, N = 5; exercise, N = 21), immediately following intervention (sedentary controls, N = 4; exercise, N = 18), and at 12-week post-intervention (sedentary controls, N = 6; exercise, N = 17) were group analyzed. At all three time points evaluated, there was no significant differences observed within and across experimental groups.

FIGURE 7

Group analyses of the effect of exercise on tibial nerve evoked electrodiagnostic parameters. Data shown are the median, interquartile range, and minimum–maximum range of latency, amplitude, and calculated conduction velocity elicited from antidromic stimulation of tibial nerves at entry into the study (baseline), immediately following intervention, and again at 12-week post-intervention of responding patients, as indicated. Parametric data were analyzed by two-way ANOVA with Tukey’s multiple comparison post hoc analysis. The total number of patients with measurable responses at baseline (sedentary controls, N = 11; exercise, N = 31), immediately following intervention (sedentary controls, N = 9; exercise, N = 27), and at 12-week post-intervention (sedentary controls, N = 8; exercise, N = 26) were group analyzed. At all three time points evaluated, there was no significant differences observed within and across experimental groups.

FIGURE 8

Group analyses of the effect of exercise on peroneal nerve evoked electrodiagnostic parameters. Data shown are the median, interquartile range, and minimum–maximum range of latency, amplitude, and calculated conduction velocity elicited from antidromic stimulation of peroneal nerves at entry into the study (baseline), immediately following intervention, and again at 12-week post-intervention of responding patients, as indicated. Parametric data were analyzed by two-way ANOVA with Tukey’s multiple comparison post hoc analysis. The total number of patients with measurable responses at baseline (sedentary controls, N = 10; exercise, N = 30), immediately following intervention (sedentary controls, N = 8; exercise, N = 26), and at 12-week post-intervention (sedentary controls, N = 7; exercise, N = 26) were group analyzed. At all three time points evaluated, there was no significant differences observed within and across experimental groups.