Rethinking Neuropathy in Type 1 Diabetes: Had We Lost Sight of What Matters Most?

Although animal and human physiological studies have informed potential mechanisms of diabetic neuropathy (1), the real‐world variability in clinical presentation and complexities in measurement have severely hindered a clear understanding of its epidemiology, its impact on the population with type 1 diabetes, and the development of therapies. The most common form—and therefore the one usually referred to simply as “diabetic neuropathy”—is the diffuse, symmetrical, slowly progressive, length‐dependent damage to the peripheral and autonomic nervous system classified technically by the term “diabetic distal symmetrical polyneuropathy” (1). It remains asymptomatic for years, may first present with abnormal sensation symmetrically at the tips of the toes, and may over time spread to the stocking‐and‐glove distribution. It involves injury to different anatomical nerve types that show variable clinical manifestations between individuals. Nerves can be classified as small, thinly myelinated or unmyelinated fibers that make up the autonomic as well as the pain and temperature sensory fibers, while large, myelinated fibers are responsible for other sensory and skeletal muscle functions. Some people have a large fiber–predominant pattern and experience numbness, tingling, or imbalance, while others have small fiber–predominance with burning and stabbing pain, impairment in sensing heat and cold, or a propensity toward clinical autonomic abnormalities like lightheadedness or gastroparesis (1). Some people experience a combination, and regardless of the pattern, people can be asymptomatic for extended periods even though their physical examination or specialized testing reveals marked impairments in nerve structure and function. With progression to foot muscle weakness or clinical autonomic manifestations like dry feet from limited sweat production, the subsequent risk of ulcer, infection, and amputation intensify. Complications like these are feared by people with diabetes more than death itself (2), and of great concern is the recent evidence of a resurgence in amputations (3–5).

To study the outcomes that matter most, a clinical researcher must recognize this variability in the presentation of neuropathy but also acknowledge that an emphasis on diabetic neuropathy alone is inadequate. Loss of protective sensation is only one of three core component causes of foot complications (Fig. 1). Tissue ischemia from peripheral vascular disease, and the minor trauma to skin that incites ulceration and infection induced by inadequacies in footwear and general foot care, are equally important. In practice, evaluation can be accomplished by way of very simple annual examinations for loss of protective sensation (testing pressure sensation with a monofilament, testing vibration sensation, or frankly an even simpler “Touch the Toes” approach adopted by Diabetes UK) (6–8), for arterial patency (pedal pulses and skin changes), and for the presence of foot abnormalities such as calluses or deformities that indicate repetitive minor trauma. Depending on the number and severity of abnormalities, their presence should initiate a number of interventions beginning with self–foot care education and professionally fitted therapeutic footwear and continuing to referral for wound management and surgical consultation (7,9).

Though we could criticize a study that does not implement specialized testing or we could debate the most objective test for identifying diabetic neuropathy, the issue is that there is no single objective test that is all‐encompassing for the three core component causes. Though perhaps not sensitive or specific enough to serve as a biomarker response to a therapeutic intervention in a clinical trial, one strategy is to implement a composite score, such as a self‐reported symptom questionnaire (10).

In this issue of Diabetes Care, two articles report on the prevalence and factors associated with diabetic neuropathy using this simple measurement approach. The first, administered as a substudy nested within the 25,000‐person T1D Exchange Clinic Network in the U.S., cross‐sectionally administered the questionnaire component of the Michigan Neuropathy Screening Instrument (MNSIQ) among 6,000 people representative of the overall cohort (11). The average person from this study population was 40 years of age, had 18 years of diabetes duration and a current A1C of 8.1%, and had low prevalence of diabetes complications. Symptomatic neuropathy was present in 11%, associated with recognized cardiometabolic risk factors such as older age and longer diabetes duration, smoking, dyslipidemia, higher body mass, and presence of other complications. The second study, administered in the Scottish Diabetes Research Network Type 1 Bioresource, is an extraordinary and innovative example of population‐level sampling with linked health services research databases, including a direct patient evaluation component coupled with clinical electronic records (12). Compared with the former, this study population was slightly older with higher glycemic exposure. Neuropathy, measured by the same questionnaire, was present in 13% and was associated with similar cardiometabolic risk factors. What emerges collectively from these two ambitious projects is the following: First, prevalence of symptomatic diabetic neuropathy in the population with type 1 diabetes is substantial, even when using a technically insensitive tool for identifying neuropathy, and is remarkably consistent regardless of cohort study selection, approach, and national setting. Second, there is confirmation of the probable causal role of cardiometabolic risk factors (13). Finally, it is new and inescapable knowledge that a person with type 1 diabetes who has some measure of social disadvantage, such as a low level of education, a lower income level, or residence in a deprived geographical community, has a dramatically higher likelihood of diabetic neuropathy.

Both of the reported studies are reasonably representative of the contemporary type 1 diabetes population. Rather than simple recognition of the average 11–13% prevalence estimate, of greater clinical relevance for risk communication is the lifetime, or long-term, neuropathy risk. In both studies, older subgroups had prevalence estimates approximating 20% (disregarding the oldest extreme category most susceptible to survivorship bias). Approximate cumulative incidence from previous cohort studies estimated long-term risk as low as 30% (14) and as high as 70% (15) depending on secular periods from type 1 diabetes onset in the 1960s to the 1980s and also depending on the sensitivity of the methods used to define the neuropathy. Two interpretations are possible: first, prevalence has decreased over time such that cohorts like these, whose participants on average had diabetes diagnosed in the 1990s, show the lowest prevalence estimates; second, a questionnaire may not compare with the more sensitive methods used in earlier, smaller‐scale studies that included physical examination (15) and objective testing (14). As examples, in the Diabetes Control and Complications Trial (DCCT)/Epidemiology of Diabetes Interventions and Complications (EDIC) study cohort or in work detailing neuropathy measures in longstanding type 1 diabetes, clinical manifestations of neuropathy were observed in only half of those with abnormal objective testing by nerve conduction studies (14,16,17). By extrapolation from current results, the contemporary long-term risk of 20% may represent 40% neuropathy prevalence if objective testing were implemented. Furthermore, the threshold score recommended by the researchers who developed and validated the MNSIQ has only 30% sensitivity, while lower threshold scores have greater sensitivity (to 50%) without substantially limiting specificity (10). Applying these thresholds provides prevalence estimates of 35–39% (11,12). To overcome this diagnostic uncertainty, the investigators could have implemented, at least in a subset of participants, a measurement such as gold standard nerve conduction studies (18,19), physical examination scales (10,20–22), established quantitative sensory tests (23), morphological evaluation of intraepidermal nerve fibers (23), or newer methods such as the proxy examination of corneal nerve fibers (23–25). These are less practical for such large‐scale studies, and although they may be more sensitive, objective, and specific for identifying peripheral nerve injury (and therefore more appropriate for most clinical trials evaluating neuropathy‐specific interventions), it is advantageous that the MNSIQ questionnaire attempts to measure presence of different patterns of diabetic neuropathy and, additionally, the aspects of other core component causes. Items such as whether legs hurt when walking (perhaps indicating vascular disease) or whether there is a history of amputation certainly represent the outcomes that matter most to people with diabetes, even if they are not specific to neuropathy and fail to identify asymptomatic disease.

The probable causal role of cardiometabolic risk factors for neuropathy (13) is further confirmed by the two studies reported in this issue of Diabetes Care (11,12). New and provocative, though, is the association with socioeconomic disadvantage. It has long been identified as a risk factor for amputation, perhaps by way of association with the third core component cause, inadequacies in general foot care, which may relate to access to appropriate footwear, foot care, and personal foot hygiene. However, the authors of the current two studies (11,12) make the consistent claim that socioeconomic disadvantage is fundamentally associated with the first core component, diabetic neuropathy itself. The association was independent of current cardiometabolic risk factors, but historical exposures may represent unmeasured confounding variables in this cross‐sectional analysis. Furthermore, the MNSIQ is a composite outcome measure that includes history of amputation.

These innovative designs for large‐scale research for both type 1 diabetes and neuropathy, a condition and a complication that are generally understudied, represent transformative groundwork that should now be converted into longitudinal cohort structures to evaluate the causal framework for socioeconomic disadvantage (26) and other exposures for neuropathy incidence, progression, regression, and their secular trends. The research community has worked tirelessly, without success, to find a disease‐modifying therapy for diabetic neuropathy other than targeting glycemic and cardiometabolic control (25,27 28). As providers, researchers, advocates, policy makers, and members of the health care industry that aspire to have tangible impact on the lives of people with type 1 diabetes, we must challenge ourselves to recognize socioeconomic disadvantage and ask ourselves what we can do to genuinely and creatively support the vulnerable. This likely matters the most.