The increased conversion of glucose to sorbitol that occurs through the polyol pathway in the hyperglycemic state has been suggested to play an important role in the pathogenesis of diabetic neuropathy [2]. Reported benefits with aldose-reductase inhibitors, which act by interrupting the accumulation of sorbitol, have been disappointing [3-7]. These discouraging results may have occurred because patients with severe degrees of polyneuropathy, in whom a decrease of nerve function is likely to be too advanced, were investigated.

To determine whether treatment for mild neuropathy results in benefit, we conducted a randomized, placebo-controlled, double-blind trial of the aldose-reductase inhibitor tolrestat (Alredase, Wyeth-Ayerst) in patients with asymptomatic diabetic autonomic neuropathy. Tolrestat significantly improved autonomic nerve function and peripheral neuropathy, when present. No important side effects were observed.

Methods

Top Methods Results Discussion Author & Article Info References

Patients

All diabetic patients participating in this study were referred from the outpatient department of the Institute of General Medicine at the University of Naples, Italy and gave written, informed consent. The study, which began in early 1990, was carried out in accordance with the Helsinki Declaration II and was approved by the ethics committee of our institution.

Men and postmenopausal women were included in the study if they had noninsulin-dependent diabetes mellitus with a known duration of disease of at least 1 year, stable diabetes (no change of therapy or stable glycated hemoglobin values during the previous six months), and asymptomatic diabetic autonomic neuropathy defined by the presence of at least one pathologic cardiovascular reflex test result. Patients with other possible causes of neuropathy, including liver and renal disease, thyroid hypofunction, low cobalamine levels, malignant disease, or excessive alcohol intake, were excluded. We also excluded patients affected by diseases known to interfere with cardiovascular reflexes (ischemic heart disease, heart failure, valvular heart disease, and major cardiac arrhythmias), as well as those receiving cardiac glycosides, anticholinergics, sympathomimetics, ß-blockers, or other drugs affecting heart-rate variability. Eligibility criteria were investigated during a 2-week period.

Study Design

A randomized, double-blind, parallel-group design was used. All patients were given placebo during a 4-week run-in period (single-blind). At the end of this period, baseline neurologic functions were measured, and patients were randomly assigned either to continue to receive placebo or to receive treatment with tolrestat (200 mg/d given in the morning). Patient compliance was assessed by counting of returned tablets; 85% of pills were taken in both patient groups.

During the study, patients continued their standard antidiabetic therapy, diet, oral tablets, or insulin. Change of dosage was avoided during the study to prevent varying diabetic control from influencing nerve function [8]. No treatment for peripheral neuropathy, if present, was permitted except for occasional non-narcotic analgesic drugs given as needed.

Procedures

All tests assessing autonomic and peripheral neuropathy were done by persons blinded to patient treatment regimen.

Autonomic Neuropathy

Autonomic nervous function was measured using four cardiovascular reflex tests, each done 3 hours after patients ate lunch: heart rate response to six forced respirations in 1 minute (deep breathing, maximum/minimum heart rate expressed by the expiration/inspiration ratio between the mean of the three longest R-R intervals in expiration and the mean of the three shortest R-R intervals in inspiration), heart rate response to standing (modified 30:15 ratio; ratio of the 30th to the 15th R-R interval after changing from supine to upright position), heart rate response to a standardized Valsalva maneuver (ratio of the longest R-R interval after the maneuver to the shortest interval during the maneuver), and blood pressure change after standing (change in systolic blood pressure beginning 30 seconds after assuming the upright posture) [9, 10]. Heart rate was recorded continuously with a Lifepak 6 (PhysioSystem, Milan, Italy) recorder. Each value was calculated as the mean of at least two consecutive tests that were done in the same sequence with a few minutes rest between each test. Results from each of these tests are known to worsen with age; reference values for a nondiabetic sample were used [11].

Peripheral Neuropathy

The involvement, if any, of peripheral nerves was assessed by measuring vibration perception thresholds at four sites using a biothesiometer (Biomedical Instrument Company, Newborn, Ohio). The sites studied were the right and left medial malleoli and the right and left halluces. The presence of peripheral neuropathy was defined by vibration perception threshold values of more than two standard deviations above the age-related normal range at two or more sites [12].

Clinical Assessment

A questionnaire was completed on symptoms of peripheral neuropathy if present. Patients were asked to grade the severity of neuropathic symptoms of either pain (burning, deep ache, tenderness) or paresthesiae (ns and needles, asleep feelings) using a visual analog scale graded from 0 to 10, in which 0 corresponded to the absence of symptoms, and 10 corresponded to very severe symptoms.

Follow-up

The study was conducted on an outpatient basis with visits at 1-month intervals. Clinical assessments described previously were repeated at 6, 9, and 12 months. Results of hemoglobin, full blood count, urea and electrolytes, liver function, creatinine, urinalysis, fasting and postprandial glucose, and glycated hemoglobin (HbA_1c) tests were assessed each month.

The diurnal plasma glucose profile represents the mean of the plasma glucose values measured every 2 hours during the day and every 4 hours during the night. Plasma glucose was measured using a glucose-oxidase method on a Beckman glucose analyzer. Glycated hemoglobin was measured by column chromatography (Auto A1c Analyzer, Kogaku Kioto, Menarini Diagnostici, Italy); normal values for our laboratory are 4.5% to 5.7%.

Statistical Analysis

All data are presented as mean ±SD; 95% confidence intervals (CIs) are provided where appropriate. Statistically significant differences in the vibration perception threshold at each site have been reported. Although the treatment by site interaction was not statistically significant for vibration perception thresholds (repeated measures analysis of variance), we decided to not use the mean value for the four sites; the values of the two different sites of the dominant leg are reported and used for analysis. Statistical significance was assessed by paired, two-tailed t-tests within groups (that is, for differences between baseline and follow-up measures in the tolrestat and the placebo groups) and the two-tailed test for a putative treatment effect (that is, comparison of the differences between baseline and follow-up values in the tolrestat and placebo groups).

Results

Top Methods Results Discussion Author & Article Info References

Of the fifty patients who entered the study, five dropped out: two for intercurrent disease apparently unrelated to the study drug (one in the tolrestat group and one in the placebo group), two for unstable diabetes (one in each group), and one in the tolrestat group for adverse reactions (nausea and diarrhea) that resolved rapidly after withdrawal of the study drug. Skin rashes, dizziness, and weight gain were not detected or reported by patients during treatment. Four of these five patients dropped out before randomization and the remaining patient, after randomization. Analysis of the results (both as randomized and as treated) did not differ whether the analysis was done according to intention to treat or according to actual treatment.

Forty-five patients (25 in the tolrestat group and 20 in the placebo group) completed the trial. Their clinical and metabolic characteristics are presented in Table 1. The two groups were similar in the number of patients, duration of diabetes, and metabolic control.

The baseline values of HbA_1c were 8.2 ± 1.1 (mean ±SD) in the tolrestat group and 8.4 ± 1.1 in the placebo group (P = 0.2). Glycemic control remained stable during the study period (repeated measures analysis of variance) in both groups: at 12 months, changes in daily glycemic profile were 0.4 (CI, –0.1 to 0.7) in the tolrestat group and 0.1 (CI, –0.3 to 0.5) in the placebo group (P = 0.1).

Both groups had comparable values of cardiovascular reflex tests in the basal state. Patients treated with tolrestat showed an improvement of all tests, which appeared to begin after 6 months of treatment and persisted until the end of the study (Table 2). In this group, deep breathing increased by 0.026 expiration/inspiration ratio (95% CI, 0.015 to 0.036, P = 0.001) from baseline to 12 months. In the placebo group, the same test decreased by a value of 0.012 expiration/inspiration ratio (95% CI, –0.024 to –0.004, P = 0.03).

Figure 1 shows the changes observed at the end of the study in the tolrestat and placebo groups, respectively. If one looks at the results of the deep breathing test, 20 patients improved, 3 deteriorated, and 2 remained unchanged compared with 4, 14, and 2 patients in the placebo group. Similar results were obtained for the other tests.

View larger version (33K): [in this window] [in a new window]   Figure 1. Individual values of the four cardiovascular reflex tests.Top. In the basal state (Basal) and after 12 months of treatment with the aldose-reductase inhibitor tolrestat. Patients are noninsulin-dependent diabetics; means (SD) of the basal and post-treatment values are represented by a larger filled circle. Bottom. Individual values of the four cardiovascular reflex tests in 20 noninsulin-dependent diabetic patients assigned to the placebo treatment. The units of measurement for deep breathing are for expiration/inspiration ratio; for lying-to-standing, 30/15 ratio; and for Valsalva, longest/shortest R-R ratio.

The effects of treatment on vibration perception threshold are shown in Table 2. The vibration perception threshold improved in the tolrestat group at the end of treatment, with a mean decrease of 2.1 volts (CI, 1.01 to 3.86 volts; P = 0.008) for vibration perception threshold at the malleolus and a mean decrease of 1.8 volts (95% CI, 1.09 to 3.69; P = 0.009) at the great toe. The improvement of vibration perception threshold was independent of the starting values: No relation was found between the baseline volt level and the changes observed during treatment. The positive influence of tolrestat on vibration perception threshold at both sites was already evident from the ninth month compared with baseline (P = 0.04) and remained significant until the end of the study. On the contrary, in the placebo group there was a progressive increase of vibration perception threshold at both sites studied (P < 0.05).

Symptoms of diabetic peripheral neuropathy (either pain or paresthesia or both) were minor (a score of less than 5 points for all diabetics in the visual analog scale) and were present in only 4 patients in the tolrestat group and in 3 in the placebo group. The small number of diabetics complaining of symptoms precluded any statistical analysis; however, three patients improved and one remained unchanged in the tolrestat group, whereas all three patients worsened in the placebo group at the last visit (12th month). No patient in both groups had new onset of symptoms during the study period.

Tolrestat was well tolerated by all patients. Neither statistically significant changes from baseline nor differences in mean changes between groups were noted for any of the safety measures considered. Both groups had minor and statistically insignificant reductions in erythrocyte indices and total proteins, whereas renal function indices and liver indices (alanine aminotransferase, alkaline phosphatase, and glutamyl transferase values) showed no significant change.

Discussion

Top Methods Results Discussion Author & Article Info References

We found that tolrestat exerts a positive effect on nerve function in patients with mild neuropathy resulting from diabetes mellitus. Results of both cardiovascular reflex tests for autonomic neuropathy and vibration perception threshold improved after 12 months of active treatment with the aldose-reductase inhibitor. This finding contrasts with the progression of neuropathy in our diabetic patients receiving placebo, as shown by the decreased expiration/inspiration ratio during deep breathing, which, among the other tests, is a sensitive, reproducible, and specific test for one of the earliest defects, cardiac vagal denervation [13]. The vibration perception threshold also deteriorated by approximately 1.5 volts at the end of the observation period.

Polyneuropathy affecting the somatic and autonomic nervous systems is a common complication of diabetes mellitus [14]. Therapy for both forms is said to be symptomatic, supportive, or palliative. So far, the only two etiologic approaches used in the hope to halt or possibly reverse diabetic neuropathy are strict glycemic control and aldose-reductase inhibitors. Although improved metabolic control may result in some improvement of nerve conduction velocity and the sense of vibration [15, 16], few patients are able to establish normal levels of circulating glucose throughout the day or to maintain normal levels of HbA_1c [17, 18]. Evaluating nerve function after successful pancreatic transplantation in a large series of type-I diabetics, Kennedy and colleagues [19] found that the progression of diabetic neuropathy might be halted through the restoration of the euglycemic state. Moreover, a small degree of improvement led the authors to suggest that intervention is necessary at an earlier stage of disease, before possible structural changes of the nerve have occurred.

The need for an intervention early in the course of neuropathy also is suggested by results from most studies that have evaluated the effect of different aldose-reductase inhibitors (sorbinil, tolrestat, and ponalrestat) on symptomatic sensorimotor diabetic neuropathy [5, 7, 20]. In human diabetic neuropathy, aldose-reductase inhibitors have been of doubtful efficacy perhaps because many of the studies were hampered by the short period of pharmacologic treatment, the heterogeneity of the groups under investigation, and the severe degree of polyneuropathy of the diabetics evaluated. Moreover, concern has arisen about the incidence and severity of systemic side effects and skin eruptions seen with sorbinil [21]. Tolrestat is an aldose-reductase inhibitor with a favorable safety profile [22] and appears to be free of major side effects.

The concept of irreversible diabetic autonomic neuropathy has been challenged in part by the results of metabolic and pharmacologic intervention trials. Although a large percentage (80%) of the diabetics we studied showed some improvement of autonomic neuropathy at the end of the study, most of the values did not normalize. Although this finding may indicate that a longer treatment period is required for a more consistent improvement, no final agreement has been reached on the role of the sorbitol pathway in diabetic neuropathy. Other views on the pathogenesis of diabetes are emerging, such as glycation of proteins [23] and endoneural hypoxia [24]. Thus, depending on the contribution of the different mechanisms in the single patient and on the balance between the reversible and irreversible components of neuropathy, the benefits of aldose-reductase inhibition are variable.

Although the mortality among diabetics with autonomic neuropathy is not as high as previously thought [13], it still remains greater than that in diabetic patients without such a complication. Recently, Navarro and colleagues [25] reported the data from a large series of diabetic patients (n = 152) with motor-nerve conduction defects, autonomic nerve dysfunction, or both. These patients showed decreased survival rates compared with 57 patients without measurable defects. Interestingly, the patients with abnormalities in autonomic nerve function who successfully received transplants had an improved survival rate compared with nontransplanted patients.

Although the treatment effects observed in this study were small and are perhaps of little clinical significance, our findings that the progression of mild diabetic autonomic and peripheral neuropathy may be halted or even reversed by aldose-reductase inhibition suggest that this class of compounds may improve a metabolic component of nerve function. More and longer studies, done in an earlier phase of the disease, will help to determine the role of aldose-reductase inhibition in the prevention of diabetic neuropathy.