Which Test for Diagnosing Early Human Diabetic Neuropathy?

Epidemiological studies show that diabetic neuropathy can affect 50–90% of patients with diabetes. Diabetes increases the risk of foot ulceration and amputation more than 23-fold (1), and neuropathy is the major contributory factor to this increased risk. Neuropathy presents as painful neuropathy in ∼20% of patients (2), and it independently predicts all-cause and diabetes-related mortality (3).

Several longitudinal studies have shown that glycemic control and vascular risk factors are important in the development of diabetic neuropathy in patients with both type 1 (4) and type 2 (5) diabetes. However, the time of onset and exact deficits associated with early diabetic neuropathy have not been established using a comprehensive neuropathy assessment. In an early study of 132 newly diagnosed patients with type 2 diabetes, symptoms and clinical signs of neuropathy had a low prevalence (1.5 and 2.3%, respectively), while polyneuropathy assessed by electrophysiology was identified in 15.2% of patients and autonomic neuropathy based on an abnormal expiration-to-inspiration ratio was increased threefold compared with healthy control subjects (6). In a 5-year follow-up of this cohort, there was only a minimal deterioration in electrophysiology (7).

The latest recommendations continue to advocate a multimodal approach to assessing diabetic neuropathy. This should include symptoms and signs, quantitative sensory testing, and electrophysiology (8). There is no doubt that symptoms and neurological deficits matter most to patients. However, the excessive variability and poor reproducibility (9) of these measures limit their utility in observational studies as well as in clinical trials. Although neurophysiology has been considered to be the most objective measure of diabetic neuropathy, it assesses only large myelinated fibers, which constitute about 10% of peripheral nerves. Further, the reproducibility of this approach has been questioned (10).

Small fibers constitute the majority of peripheral nerve fibers, are the earliest to be damaged, and have direct pathophysiological relevance to generation of nociceptive signals, leading to both neuropathic pain (11) and impaired pressure-induced vasodilation (12) predisposing to foot ulceration. Despite these factors, small nerve fibers receive considerably less attention than large ones. Perhaps not surprisingly, a reduction in intraepidermal nerve fiber density (IENFD) has been demonstrated in subjects with impaired glucose tolerance, which improved after lifestyle intervention (13). In contrast, Dyck et al. (14) found no evidence of neuropathy in subjects with impaired glycemia. However, their diagnosis of neuropathy was based on composite abnormality in nerve conduction. Although small fibers can be assessed objectively by quantifying IENFD in skin biopsies, this is an invasive procedure requiring expert laboratory assessment (15).

We have pioneered corneal confocal microscopy (CCM), a rapid and noninvasive ophthalmic technique. The technique has been shown to be reproducible and highly sensitive and specific in the diagnosis of early nerve damage, even in patients without symptoms or signs and in those without neurophysiologic abnormalities (16) (Fig. 1). The utility of this technique was confirmed recently by demonstrating that small-fiber dysfunction and corneal nerve loss predated neurophysiologic abnormalities in patients with type 1 diabetes (17).

In this issue, the new report by Ziegler et al. (18) is timely because it uses neurophysiology and quantitative sensory (vibration and thermal perception thresholds) and autonomic function testing, along with a detailed assessment of small-fiber pathology by quantifying IENFD and corneal nerve fiber morphology using CCM. Despite the relatively short duration of diagnosed diabetes among study participants (2.1 ± 1.8 years) and their excellent glycemic control (HbA_1c 6.8 ± 1.1% or 50.8 ± 12.0 mmol/mol), these patients showed evidence of abnormalities in a number of measures, including lower-limb electrophysiology, vibration and cold detection thresholds, and expiration-to-inspiration and Valsalva ratios. A reduction below the 2.5th percentile for corneal nerve fiber density was shown in 21% of the patients. Although reduced IENFD was present in 14% of the patients, they did not all exhibit reduced corneal nerve fiber density.

Although there were some weak, but statistically significant, correlations between CCM and IENFD and electrophysiology measures, there was no correlation between CCM and quantitative sensory testing or autonomic function testing. This is in contrast to another recent study showing highly significant correlations between corneal nerve fiber length and cold detection threshold, laser Doppler imaging flare technique, and heart rate variability (19). Clearly, only a longitudinal follow-up study will be able to identify which of these parameters—particularly corneal nerve fiber density or IENFD—best predict development and progression of diabetic neuropathy. Furthermore, the prevalence of microalbuminuria (15.1%) and background retinopathy (2.3%) were lower than the prevalence of neuropathy, an observation that is consistent with another recent study showing corneal nerve loss even in patients with type 2 diabetes who do not have diabetic retinopathy (20). These findings challenge the dogma that the earliest microvascular complications of diabetes are retinopathy and microalbuminuria. Ziegler et al. (18) used a novel corneal nerve imaging and quantification technique that enabled analysis of a larger corneal area, potentially enhancing the utility of CCM. However, the diagnostic utility of this technique compared with the faster and established technique involving central corneal image acquisition with automated image analysis (16) remains to be established.

The study by Ziegler et al. (18) and other ongoing longitudinal studies, such as the Longitudinal Assessment of Neuropathy in type 1 Diabetes using novel ophthalmic Markers (LANDMark) trial (21) in type 1 diabetes, which have used a battery of established tests, including symptoms and deficits, neurophysiology, quantitative sensory testing, autonomic function tests, and the newer more precise small-fiber tests such as IENFD and CCM, will enable us to objectively establish the utility of each of these tests. This information will provide insight into the most effective early diagnostic tests as well as the best surrogate end points for clinical trials of diabetic neuropathy. To continue to advocate measures that do not fulfill the very basic definition of a surrogate end point for the diagnosis of diabetic neuropathy (8) is questionable. This is especially important if the diagnostic tests are aimed at identifying the earliest phases of nerve damage in longitudinal studies of risk factors for development, progression, or regression of diabetic neuropathy. These considerations warrant careful attention if we are to resurrect translational research and U.S. Food and Drug Administration approval of new therapies for human diabetic neuropathy.