Subclinical Neuropathy Among Diabetes Control and Complications Trial Participants Without Diagnosable Neuropathy at Trial Completion: Possible predictors of incident neuropathy?

The DCCT design and eligibility criteria have been described elsewhere (1). Briefly, 1,441 subjects with 1- to 15-year histories of type 1 diabetes who did not have neuropathy requiring medical intervention or treatment and who had only minimal or no microvascular complications were eligible to participate. Subjects were randomly assigned to intensive therapy (administering insulin three or more times daily by injection or by an external insulin pump) or conventional therapy (one to two injections of insulin daily) and followed for 4–9 years (mean 6.5) (1,9). At DCCT baseline, the two treatment groups were similar, with only minor exceptions noted below, in terms of their clinical and electrodiagnostic results. Among the seven categorical clinical or nerve conduction summary measures, only the peroneal nerve conduction was different between treatment groups at baseline (Table 1). That measure showed a lower prevalence of an abnormal result in the intensive compared with the conventional treatment group (35.4 vs. 42.3%, P = 0.008). None of the 10 continuous nerve conduction measures showed significant group differences (Table 2). After 5 years of treatment, numerous significant nerve conduction differences were observed between groups, all favoring better function (faster sensory and motor conduction velocities and shorter F-wave latencies) in the intensive treatment group (8).

Distal symmetrical polyneuropathy was defined in the DCCT using clinical and electrodiagnostic criteria. Board-certified neurologists who were masked to treatment group assignment performed the clinical neurological evaluations and identified causes of neuropathy other than diabetes (1). Nerve conduction studies evaluated the dominant median (motor and sensory), peroneal (motor), and sural nerves using standard techniques and specified anatomical landmarks or stimulation-to-recording electrode distances for each study. Absolute threshold levels for the individual attributes were defined as the median of the upper or lower limits provided by participating laboratories (8).

As shown in Fig. 1, the DCCT study end point of confirmed clinical neuropathy required the presence of both definite clinical neuropathy by the neurologist's examination (defined by at least two of the following: positive responses among symptoms, sensory signs, or absent or hypoactive reflexes consistent with a distal symmetrical polyneuropathy) and an abnormal nerve conduction study consistent with a distal symmetrical polyneuropathy (value above or below the absolute threshold of normal for amplitude, conduction velocity, distal latency, or F-wave latency in at least two anatomically distinct nerves) (1,8). (The original DCCT definition of confirmed clinical neuropathy permitted confirmation of neuropathy based on unequivocally abnormal autonomic test results. However, only 12 subjects fulfilled that criterion, and confirmation of distal symmetrical polyneuropathy in these analyses is based only on appropriate nerve conduction study abnormalities). Subjects with only one abnormal finding among symptoms, sensory signs, or absent or hypoactive reflexes, with or without abnormal nerve conduction studies, were classified as having possible clinical neuropathy. The term subclinical neuropathy was used to describe subjects who had an abnormal nerve conduction study suggestive of a distal symmetrical polyneuropathy but who did not fulfill criteria for definite clinical neuropathy.

For these analyses, we stratified the DCCT cohort at study completion by systematically excluding subjects with progressively less severe degrees of neuropathy. We first excluded those who fulfilled the primary DCCT end point of confirmed clinical neuropathy and sequentially excluded subjects with definite clinical neuropathy, possible clinical neuropathy, and subclinical neuropathy. In other words, we began by excluding subjects with the strongest evidence of neuropathy and eventually excluded all subjects with any clinical or electrodiagnostic evidence of neuropathy. Treatment groups comprised of the remaining subjects after exclusion of those fulfilling any of the definitions of neuropathy were compared in terms of the results of their neurologic and electrophysiologic evaluations. Treatment group differences were compared with the contingency χ² test for categorical (or qualitative) measures of neuropathy and Wilcoxon's rank-sum test for ordinal and numeric (or quantitative) nerve conduction measures. When the sample size was small, Fisher's exact test was used. A global evaluation of treatment groups across all nerve conduction measures was made for subgroup 1 using the O'Brien rank-sum test.

The comparisons of the neuropathy subgroups by treatment are shown in Table 1 (dichotomous clinical and nerve conduction measures) and Table 2 (continuous nerve conduction results). Analyses of subgroup 1, which excluded subjects who had a confirmed clinical neuropathy, showed numerous significant treatment group differences. All of the significant differences reflected better function among the intensive treatment group compared with the conventional treatment group in terms of the prevalence of sensory symptoms, decreased or absent reflexes, or the dichotomous nerve conduction abnormalities beyond the upper or lower limits of normal for the individual nerves. Similarly, the continuous nerve conduction study results were significantly better in the intensive treatment group than the conventional treatment group for 8 of the 10 measures, including all lower-extremity measures. The median motor and peroneal conduction velocities averaged >2 and 3 m/s faster, respectively, among intensive treatment subjects compared with conventional treatment subjects. Peroneal and sural amplitudes also showed significantly better performance (higher values), favoring the intensive treatment group.

Treatment group differences persisted, as additional subjects with progressively lesser degrees of neuropathy were eliminated from the comparisons, although the number of significant treatment group differences and the magnitude of the average group differences diminished. After eliminating subjects with definite clinical neuropathy (subgroup 2) and subjects with definite clinical neuropathy or possible clinical neuropathy (subgroup 3), numerous group differences remained, all of which favored the intensive treatment group. The final column (subgroup 4) in Tables 1 and 2 represents only those subjects who did not meet any of the DCCT definitions of clinical or subclinical neuropathy. That is, subgroup 4 includes subjects without any symptoms, signs, or electrodiagnostic abnormalities possibly attributable to a distal symmetrical polyneuropathy. As expected, none of the dichotomous indictors of neuropathy showed significant treatment group differences, but the continuous electrodiagnostic measures showed significant group differences in median and peroneal motor conduction velocities, reflecting faster conduction velocities in the intensive treatment group compared with the conventional treatment group (55.8 vs. 55.4 m/s, P < 0.001 and 46.0 vs. 44.4 m/s, P < 0.001, respectively).

Among subjects who did not meet any of the DCCT definitions of neuropathy at DCCT completion (subgroup 4 in Tables 1 and 2), those in the conventional treatment group had greater levels of subclinical neuropathy than subjects in the intensive treatment group. Subclinical neuropathy in this context, as opposed to the DCCT definition, represents a form of asymptomatic neuropathy that has not yet produced clinical signs or abnormal electrophysiology but that is inferred by inferior performance on sensitive nerve conduction studies that, nonetheless, remained within the range of normal. At DCCT completion, 8 of the 10 nerve conduction measures were significantly better in the intensive treatment group compared with the conventional treatment group after excluding subjects with confirmed clinical neuropathy, the primary DCCT end point (O'Brien rank-sum test across all nerve conduction measures, P < 0.0001). Because significant treatment group differences remained in the nerve conduction results regardless of the definitions of neuropathy used to exclude subjects from analysis, the two groups consisting of subjects without diagnosable neuropathy at the DCCT completion cannot be considered comparable for future studies of neuropathy.

Dichotomous classifications of neuropathy are well accepted but have limitations, and subjects receiving the same classification may differ substantially in important ways. Figure 2 depicts the temporal change in a hypothetical continuous attribute of neuropathy, such as the number of functioning axons per nerve, for two subjects who are initially identical but who deteriorate at different rates. In this figure, levels of axonal loss that remain above a critical threshold are insufficient to fulfill criteria for a diagnosis of neuropathy. Axonal loss from any cause, including age-related axonal attrition that remains above the threshold, nonetheless reduces the safety margin for developing neuropathy. In terms of the timing of DCCT/EDIC evaluations shown in Fig. 2, the two subjects differed substantially at DCCT completion in terms of the number of axons per nerve but neither fulfilled the dichotomous definition for “neuropathy.” The subject deteriorating most rapidly fulfilled diagnostic criteria for neuropathy before the other subject who was deteriorating more slowly and would have done so even if their rates of deterioration had become similar after DCCT completion. Although the representation is hypothetical, it is potentially relevant to explaining the appearance of new-onset neuropathy among DCCT/EDIC subjects.

Few would argue with the concept of an asymptomatic or subclinical neuropathy. Patients without symptoms or signs of neuropathy are frequently found to have electrodiagnostic evidence of neuropathy during evaluation of an unrelated condition. Similarly, electrodiagnostic deterioration within the normal range may reflect a subclinical neuropathy. This concept has application in the context of chemotherapy-induced toxic neuropathy, one of the few situations in which asymptomatic, neurologically intact patients are evaluated sequentially as they develop neuropathy. In such trials, declining sensory amplitudes are used to monitor the onset and progression of neuropathy during chemotherapy. Among these patients, the existence of interval subclinical neuropathy that precedes the onset of symptoms and signs is only inferred. It is reasonable, however, to assume a continuum from normal nerve function to subclinical neuropathy to clinically evident neuropathy that is detectable on neurological examination, despite the predictably small window of severity that can be described as subclinical. For example, patients receiving cisplatin chemotherapy demonstrate a progressive decline in sensory response amplitudes within the normal range early in the course of cisplatin neuropathy, often before symptoms or objective evidence of sensory neuropathy develop (10,11). In studies involving thalidomide, an average decline of 40% from baseline in the sensory amplitude has been recommended as a physiological level associated with the onset of clinically detectable sensory neuropathy (12). This criterion does not depend on the cause of the neuropathy. Using this criterion, patients with baseline sensory amplitudes near or above the normal mean could experience a substantial amplitude decline with neither clinical nor electrodiagnostic evidence of neuropathy. It is these subjects with varying degrees of subclinical neuropathy who we found overrepresented in the conventional treatment group compared with the intensive treatment group and who contribute to the significant group differences we identified.

The DCCT used generic normal values to identify electrodiagnostic evidence of neuropathy. Use of normal values specific for age, sex, and anthropometric features would have likely identified additional subjects with confirmed clinical neuropathy or with subclinical neuropathy because what is normal for an elderly, heavy, tall male will not be normal for a young, slender, short female (13). Had additional subjects been classified with nerve conduction evidence of neuropathy, the pool of subjects without neuropathy may not have shown the same treatment group discrepancies we report.

The question remains whether the small group differences in nerve conduction results are important. In diabetic neuropathy, nerve conduction abnormalities reflect the severity of neuropathic symptoms and signs among patients with diabetic neuropathy (14). Nerve conduction changes associated with diabetic neuropathy include declining response amplitude and conduction velocity, the latter finding differentiating diabetic neuropathy from axonal loss neuropathies. Nevertheless, small changes in conduction velocity could reflect a deleterious effect of relative hyperglycemia on the nerve membrane not associated with axonal degeneration. Clinically detectable and meaningful changes in the neurologic disability score correspond to a change in peroneal amplitude and conduction velocity of ∼0.7 mV and 2.0 m/s, respectively (14). Meaningful changes in sensory amplitude (median and sural nerves) are ∼3.9 μV. The treatment group differences we found after excluding subjects with confirmed clinical neuropathy for the peroneal nerve (0.6 mV and 3.1 m/s) were near or exceeded the values considered clinically significant for clinical trials. Although the median motor amplitude values did not differ between the treatment groups, the conduction velocity difference of 2.1 m/s did exceed the level considered clinically significant. The average change we observed in the median sensory and sural amplitude (1.3 μV) was less than that considered clinically detectable among diabetic subjects (3.9 μV) (14).

Our results suggest that different levels of subclinical neuropathy in the treatment groups at DCCT completion could contribute to the subsequent onset of neuropathy during EDIC (7). Namely, the timing of incident neuropathy could reflect, at least in part, previous subclinical diabetic-induced neuronal injury, as opposed to predetermined ongoing axonal damage or augmented neuronal apoptosis. In an ongoing EDIC study, the same measures performed in DCCT are being repeated on average 13 years after the completion of DCCT. In the planned analyses, we will adjust for the electrodiagnostic differences between treatment groups at DCCT completion and determine whether prior intensive therapy resulted in sustained improvements in neuropathy independent of treatment group differences at DCCT completion.

This work was supported by contracts with the Division of Diabetes, Endocrinology and Metabolic Diseases of the National Institutes of Diabetes and Digestive and Kidney Diseases and the General Clinical Research Centers Programs, National Center for Research Resources.