Sweat gland nerve fiber density and association with sudomotor function, symptoms, and risk factors in adolescents with type 1 diabetes

The prevalence of autonomic neuropathy among adolescents with type 1 diabetes (T1D) ranges widely, with estimates ranging from 4 to 75%, based primarily on cardiovascular autonomic function tests [1, 2]. However, diabetic autonomic dysfunction can impact multiple organs, including the cardiovascular, urogenital, gastrointestinal, pupillomotor, thermoregulatory, and sudomotor systems. Among these, sudomotor dysfunction is a common and early aspect of diabetic autonomic neuropathy [3] and typically causes decreased sweating in the distal parts of the body. Abnormal autonomic sudomotor innervation can lead to hypohidrosis, anhidrosis, and hyperhidrosis [4], with the last being a compensatory phenomenon [3]. Research by Gibbons et al. revealed a significant reduction in sudomotor innervation in adults with diabetes (n = 36, mean diabetes duration between 4.6 and 13 years depending on sex and diabetes type) compared with healthy controls. The study group also suggested that sweat gland nerve fiber quantification can be done rapidly and reproducibly [5] and that sweat gland nerve fiber density (SGNFD) can differentiate between adults with mild diabetic neuropathy and healthy controls [6]. Additionally, Luo et al. found lower sweat gland innervation indices in adults with type 2 diabetes (n = 35) [7], and Liu et al. observed a decrease in SGNFD over a 1-year period without any progression in intraepidermal nerve fiber density (IENFD) [8].

Quantifying SGNFD using 3D reconstruction of sudomotor innervation has shown great potential for diagnosis with high reproducibility [9]. The method provides an evaluation of structural nerve damage, while the quantitative sudomotor axon reflex test (QSART) assesses the functional sweat response after nerve stimulation. QSART has been validated as an effective method in autonomic testing [10]; however, it has not yet been applied to adolescents with T1D.

The aim of this study was to determine SGNFD in adolescents with T1D and to investigate the association between SGNFD, sweat response obtained by QSART, and potential risk factors for abnormal sudomotor function and sweat gland nerve fiber loss.

The study was a cross-sectional study.

Sixty adolescents and 23 controls were enrolled in the study. None of the participants had a history of neuropathic symptoms. Twelve out of the 60 adolescents with T1D (20%) used pens to inject insulin, and the remaining 80% had an insulin pump. All used continuous glucose monitors (CGM) except for five participants. The median of self-monitoring of daily blood glucose frequency was 7 (range 1–25).

The selection process for the study population and the data available for final analysis are shown in Fig. 2, and the clinical characteristics of the participants are displayed in Table 1.

We report, to our knowledge for the first time, that adolescents with T1D had significantly reduced maximum and mean values of SGNFL and SGNFD compared to healthy control subjects. Additionally, to the best of our knowledge, this is the largest study of individuals with diabetes in which their sudomotor innervation has been quantified. A detailed morphological assessment of sweat glands is crucial for understanding morphological changes in autonomic structures such as sweat glands in diabetes. Thus, we evaluated not only the nerve fiber innervation in the sweat glands but also the length of the nerve fibers and the volume of the sweat glands.

The biological question of whether sweat glands shrink when they become dysfunctional is of interest. Loavenbruck et al. found a decrease in total SGv per biopsy in 10 cases with anhidrosis diagnosed with thermoregulatory sweat test compared to controls. They also found that the total nerve fiber length per biopsy volume decreased, but because of a greater change in SGv, SGNFD increased [18]. This contradicts our findings with similar SGv and a reduced SGNFL as the main explanation for the reduction of SGNFD, highlighting the importance of detailed morphological assessments of sweat glands, as the effect of different diseases on their structure and function may vary.

Our findings of lower mean values of SGNFL in the T1D group compared to controls support the notion of a reduction in the number of nerves, similar to findings when quantifying IENFD in people with diabetes [19]. The clinical impact of assessing only one affected SG parameter was of lesser significance; however, a combination of reduced SGv, SGNFL, and SGNFD was observed to have clinical consequences including either no sweat response or reduced sweat response.

We have illustrated four possible conditions that may explain the relationship between SG parameters (SGNFL, SGv, SGNFD) and sweat response in Fig. 4. Our possible explanation is that dysfunctional nerves may lead to a shrinkage of sweat glands, resulting in a reduced sweat response. A smaller number of nerves (lower SGNFL) may also impact sweat gland function but to a lesser extent. As illustrated in Fig. 4, four possible conditions exist: (1) a condition with a normal number of nerves with normal nerve function leading to normal SGv, SGNFD, and sweat response, (2) a condition with fewer nerves, but with normal nerve function, leading to a limited lowering of SGv and sweat response given a reduced SGNFD, (3) a condition with a normal number of nerves, but they are dysfunctional, leading to a shrinkage of sweat glands and a lower sweat response, but higher SGNFD, and (4) a condition with fewer nerves, which are also dysfunctional, leading to a pronounced shrinkage and low or no sweat response with normal, reduced, or increased SGNFD depending on the degree of reduction of SGNFL relative to SGv.

In our study, the majority of affected cases displayed sweat gland parameters consistent with condition 2 in Fig. 4: reduced SGNFD with hypothetically well-functioning nerves. This also explains why we found an association between lower SGNFL and lower sweat response.

A few of the patients in this study with abnormal SGNFD showed all three SG parameters (SGNFL, SGv, SGNFD) affected, similar to condition 4 in Fig. 4. Those cases exhibited the most severe reduction or absence of sweat response.

In theory, all patients could exhibit a combination of all conditions illustrated in Fig. 4. When conducting QSART on a patient, all nerves (A, B, C, D) are stimulated with acetylcholine, and an abnormal sweat response may be a mixture of conditions 2, 3, and 4, whereas abnormal SGNFD will only occur in conditions 2 and 4. The lack of association between QSART and SGNFD may be due to the fact that abnormal QSART also encompasses condition 3, leading to an even higher SGNFD. Additionally, acetylcholine in QSART does not stimulate adrenergic nerves, which are taken over if the parasympathetic system is damaged [20].

Two key points we want to emphasize if our explanatory model proves correct are that (1) reduced, normal, and increased SGNFD can all indicate an abnormal nerve status, depending on SGv, and (2) incorporating the quantification of sudomotor innervation into the QSART diagnostic method could help us distinguish between different conditions and evaluate both somatic and autonomic C-fibers. If the intra-individual variation in SGv, SGNFL, and SGNFD is low, patients may present a reduced mixture of conditions.

Our study found no association between SGNFD and age, although SGv increased with age. Considering that distal IENFD is known to decrease with advancing age [21], we suspect that our data may be best explained by the limited age range of our study population.

Our study found that most adolescents with T1D have measurable structural nerve changes and to a lesser extent reduced sweat response. Even fewer reported subjective changes in sweating. The idea that adolescents have preserved the ability to sweat in spite of structural changes in innervation seems plausible given their young age, fewer comorbidities, and fewer other confounding factors such as medications.

We found several lifestyle factors negatively associated with sweat gland innervation, including alcohol consumption, low-density lipoprotein, cholesterol, BMI, waist-to-height ratio, and smoking. Gender, BMI, and HbA1c have been previously described as factors that lower SG innervation in people with diabetes [8]. In addition, alterations in lipid metabolism and immuno-inflammatory factors are also known to be associated with diabetic neuropathy [22, 23]. It is known that abnormal lipoprotein levels can affect blood vessels supplying nerves and disrupt the lipid balance in the myelin sheath. Hyperlipidemia, particularly hypertriglyceridemia, is associated with the development and progression of neuropathy [24]. Also, excess glycated protein can contribute to nerve damage, forming advanced glycation end products (AGEs) that impair nerve function [25].

Our study showed that higher blood pressure correlates with higher sweat response, possibly due to increased activity of the sympathetic nervous system affecting both the blood vessels and SGs when the nerve function is intact. This could explain why blood pressure does not correlate with our structural nerve findings. Notably, none of our included adolescents in both groups were obese, which is of relevance as high blood pressure, obesity, and increased sweat response often coexist. Despite these facts, we still found that the waist-to-height ratio was positively correlated with SGNFL in the control group, but not in the T1D group. A higher waist-to-height ratio is a measure of central obesity, and whether the need for more sweating results in a higher SGNFL in healthy individuals with no risk of nerve damage remains to be investigated.

Despite the method being based on staining and automatic tracing, the technique of quantifying sudomotor innervation has limitations. Firstly, it requires a skin biopsy, which is an invasive procedure. In our study, the procedure was well-tolerated by adolescents when performed under local anesthesia. Additionally, deviations in the method steps can affect the final results, including the handling of the skin biopsy (fixative and storing), the staining protocol, the antibodies used, the choice of SG for analysis, the type of confocal microscope and its settings (magnification, image resolution, bit depth, gain settings), the program for removing background, and methods for tracing nerves. This highlights the importance of standardizing the entire process. We tried different automatic settings to minimize limitations before analyzing the included SG, and outlined optimized methodology in the methods section. Additionally, structures other than nerves that could lead to mis-tracing were excluded for the SG contour analysis. These structures include the contour of lipid droplets, fractionated or folded pieces of tissue, and arrector pili muscles, which have relatively dense autonomic innervation [26, 27]. It is important to note that the contour peripherally may overestimate the SGv, and using a ×10 objective may risk missing small nerve branches. The number of SGs, size of SGs, and density of innervation may vary depending on location, emphasizing the need for well-matched normative data. Additionally, it is important that the skin biopsy is taken deep enough to include the dermis.

Further research into normal physiological variations in SGv and sudomotor innervation is needed. Our study had more females than males in the control group, which can be considered a limitation. Furthermore, our small study population restricts us to adjust for risk factors; for example, sex differences are of importance. Healthy males tend to exhibit higher sweating rates than females [28], although the findings are not always uniform in all age groups [29]. The sex differences are primarily attributed to hormonal variations, body composition, and differences in metabolic heat production [30]. Park et al. showed that there is no difference in total activated sweat glands in males and females in their early 20s [31]; however, additionally, research is needed to establish clear conclusions. Our data showed that SGNFD_maximum was higher in healthy females than in healthy males, mainly caused by a nonsignificant lowering of SGv. Women tend to have more SGs but men's SGs are more active. Some studies have supported lower sweat output per SG in women than men [32]. We had no data on where the young women were in their menstrual cycle, which could be of relevance due to the impact of hormones on sweating [33]. Estrogen has a tendency to enhance heat dissipation responses, which can lead to lower body temperatures, whereas progesterone tends to promote higher body temperature, thereby influencing the autonomic mechanism of thermoregulation [34]. By periodically assessing resting sweat response, researchers can observe changes in sweat composition or rate that may occur naturally or due to various factors, such as aging, change in hormone profile, and disease progression. However, when we are specifically interested in a functional investigation of the adrenergic nerves, QSART might be superior, as the nerves are stimulated.

Another limitation of our study is that we had no information about diet, because nutrition is known to impact sweating [35]. However, on the test day, they received the same standard meal as part of another test [13], which might have minimized this factor. Finally, the test only evaluates the SG innervation at the exact spot where the skin biopsy was taken. This spot was not exactly the same location as where the distal QSART capsule was placed.

Despite these limitations and critical evaluation of the methods, our study demonstrated that quantifying sweat gland innervation in a 3D manner is possible. The analysis of IENFD and SGNFD can be performed using the same skin biopsy, and many laboratories already have established protocols and equipment. Altogether, the quantification of SG parameters may complement the results obtained from QSART. However, it is important to note that the test requires proper handling of the skin biopsy and trained personnel for analysis, and it may not be available in many places. The number of selected SGs per individual can be debated, but, for clinical use, approximately 5–6 SGs per individual appears to be sufficient, as shown in other studies of IENFD [36]. Currently, there is no clinical significance in routinely performing skin biopsies and quantifying sudomotor innervation in all adolescents with diabetes. The study has provided new insights into the field. However, individuals with autonomic symptoms are considered to benefit from general autonomic testing including QSART [37].

The strength of this study lies in its large population of 60 adolescents with T1D, whose clinical parameters were well documented. Clinical tests were standardized by the same trained healthcare professionals using the same equipment on all participants (both T1D and control subjects). The entire SGNFD analysis process, from start to finish, was performed by a single person to minimize the risk of inconsistent processing due to multiple observers.

In conclusion, SGNFD and QSART were found to be useful in evaluating sudomotor innervation and function in adolescents. Results showed that adolescents with T1D had reduced SGNFL and SGNFD compared to controls. Both SGNFL and SGNFD were found to be associated with SGv, which was identified as an important factor in the association with sweat response. Notably, reduced, normal, and increased SGNFD can all be a sign of neuropathy, depending on the SGv, if our explanatory model is proven true. The addition of more diagnostic tests may be desirable in mapping the distribution of neuropathy, differentiating between structural and functional nerve damage, and further advancing our understanding of the underlying causes of sudomotor failure. However, further research is needed to understand its clinical relevance, before it can be recommended to be incorporated in clinical practice.