In our study cohort of 95 adult individuals with T1D, patients with sudomotor dysfunction had features of lower TIR and nocturnal TIR, higher TAR and nocturnal TAR in comparison with patients without sudomotor dysfunction. Group patients by tertiles of TIR and nocturnal TIR, we found the prevalence of sudomtor dysfunction decreased with increasing tertiles. Correlation analysis revealed that the relationship between nocturnal TIR and FESC was stronger than that between TIR and FESC with correlation coefficient was respectively 0.362 and 0.356. Multiple regression analysis revealed that both TIR and nocturnal TIR were inversely associated with the risk of sudomotor dysfunction detected by SUDOSCAN regardless of sex, age, diabetes duration, BMI, Alb, smoking BUN, DN and HbA1c, notably, more statistically significant results were obtained between nocturnal TIR and sudomotor dysfunction.
DPN is characterized by broad spectrum of clinical presentations and can stay asymptomatic for a long time, that hinder the prompt diagnosis of this disease [17]. And meanwhile, it is sad to say that the measures we use for detection of DPN in routine clinical practice either are crude, or detect the disease very late when DPN has been well established. For example, scored clinical assessments such as the Michigan Neuropathy Screening Instrument (MNSI), the Neuropathy Impairment Score (NIS), and the Neuropathy Disability Score (NDS), remain subjective, heavily reliant on the examiners’ interpretations [18]. Besides,tests purport to diagnose DPN—including the 10 g monofilament, the Ipswich Touch Test, and vibration perception threshold testing—are still dependent on patients’ subjective response [19]. Although nerve conduction studies (NCS) are the current sensitive and specific tool for DPN, they are not only labour intensive, time consuming, and costly, but are also just an assessment of large nerve fibers function [20].
Small nerve fibers, which account for 70–90% of the peripheral nerve fibers, have been proved to be the first damaged part with skin biopsies by assessing intraepidermal nerve fiber density (IENFD), including the fibers that innervate the sweat glands [21]. Thus, sudomotor function represents an attractive tool to evaluate early neuropathy in people with DM. SUDOSCAN is an FDA approved point of care device which provides a quantitative measurement of sudomotor function within 3 min [22]. Solomon et al. assessed the diagnostic performance of SUDOSCAN for DPN in T1DM. Their study defined DPN in terms of established American Academy of Neurology consensus criteria using NCS and clinical examinations and found that the sensitivity and specificity of FESC were respectively 87.5 % and 76.2 % with the area under the ROC curve (AUC) was 0.85 [23]. Jin et al. also explored the diagnostic efficiency of SUDOSCAN in Chinese with a sensitivity and specificity of 85.6% and 76.2 % (AUC = 0.859) [24]. In addition, study also showed the robust repeatability and reproducibility of this machine in person with or without diabetes [25].
A series of large studies evidenced that blood sugar management, as represent by HbA1c, is an effective method for reducing the occurrence or progression of DPN in T1DM. During the mean 6.5-year follow-up, the Diabetes Control and Complications Trial (DCCT) concluded that the prevalence of DPN increased substantially in the conventional participants (from 5 % to 17 %) and only slightly among the intensive group participants (from 7 % to 9 %) with HbA1c values between the two groups were 9.1% and 7.4%, respectively. Adjusting for the presence of confirmed DSP at baseline, the risk reduction for incident DSP with intensive glucose control during DCCT was 64 % [26]. The European Diabetes (EURODIAB) Prospective Complications Study confirmed that HbA1c level during follow-up contributed especially to the risk of neuropathy, independently of the baseline HbA1c [27]. However, we should not ignore the reality that in DCCT the cumulative incidence of neuropathy (15–21%) and abnormal nerve conduction (40% to 52 %) remained substantial [28]. Besides, through 14 years of the Epidemiology of Diabetes Interventions and Complications (EDIC) study, 25% of subjects in the former intensive treatment group and 35% of subjects in the former conventional therapy group developed confirmed DPN [29]. Such thought-provoking results hint that neuropathy can develop even the ‘ideal glycemic control’ tested by HbA1c.
HbA1c estimates blood glucose concentrations over 2–3 months, but fails to predict the day-to-day glucose excursions [10]. Beck et al. re-analyzed the DCCT’s data to search for the role TIR played in diabetic microvascular outcomes. They found TIR was higher in the intensively treated group than in the conventionally treated group (52 vs. 31%), the per 10% decrease in TIR led to a 64% raise in retinopathy and a 40% increase in microalbuminuria, though resources are a 7-point glucose profile from capillary blood [30]. Our study showed that abnormal sudomotor function group had perceptibly lower TIR and nocturnal TIR, higher TAR, nocturnal TAR, M value and ADDR, while HbA1c level between groups was comparable (9.94 ± 2.79% vs. 9.70 ± 2.30%; P = 0.672). Above calculated results signified that HbA1c may be inferior to CGM metrics in building relationships with sudomotor dysfunction. Importantly, comparisons of groups expressing similar HbA1c values but different short-term glucose fluctuation values and ‘time in ranges’ may be helpful to establish the role of these dysglycemia components in DPN. When it comes to people with T2D, former studies documented that TIR derived from data of CGM has a significantly positive correlation with peripheral nerve function [31], a significantly negative correlation with painful diabetic polyneuropathy [32]. Moreover, our previous work demonstrated a negative correlation between TIR and sudomoror dysfunction in T2D [16]. Consistent with those findings, we now observed the prevalence of abnormal sudomoror function in the study population declined with the ascending of TIR tertiles. The linear and binary logistic regression analysis both indicated the independently negative association between TIR and sudomotor dysfunction even after adjusting for several risk factors.
Overnight glycemic control turns out to be a portion linked with the risk of diabetes outcomes. A prospective observational study was conducted on 162 pregnant women with gestational diabetes mellitus to investigate the association between temporal change of glucose and large for gestational age (LGA) infants [13]. By analysis of 7-day CGM data, the authors summarized that individuals who delivered LGA infants underwent significantly higher glucose concentrations during the 6-hour period of the night (00:00–06:00 am) compared with mothers who did not have LGA infants (6.0 ± 1.0 mmol/L vs. 5.5 ± 0.8 mmol/L; P = 0.005) [13]. Another research found mean nocturnal glucose levels, rather than diurnal glucose values or glucose variability, were independently related to the seriousness of vascular remodeling [14]. For the first time, based on the results of data analysis, we draw a conclusion that overnight TIR assured of its place in sudomotor dysfunction. Possible explanations for patients with suboptimal overnight TIR may lies in: (1) the overnight period lacks the glycemic excursions due to meals and exercise that occurs during the day; (2) under normal conditions of nighttime sleep, the elevated levels of growth hormone and cortisol cause the increment of the blood glucose [33]; (3) our selected samples were patients who just came to the hospital for undesirable glycemic control in the past. One can saw this with the mean level of HbA1c was more than 9% in study objects.
Explaining exact mechanisms of the above scenario are difficult, however, we can try to understand it with rational speculation. Clinical evidences suggest that overnight glycaemic control is susceptible to the daytime blood sugar levels of the person with diabetes and vise verse [34]. Poor nocturnal glycemic control can cause a reduced sensitivity of insulin at liver and other tissues, leading to fasting hyperglycemia [35]. Fasting blood glucose is the cornerstone of all-round glycemic control throughout the day. So we guess that lower TIR in the overnight period exacerbates the whole-day TIR thereby exert an influence on diabetic outcomes. Anyway our result extends the clinical benefits of TIR by demonstrating that nocturnal TIR is also involved in diabetes outcomes.
Given that GV is involved in the natural process of DPN in subjects with T1D [36] and we used to verify M value contributed to sudomotor dysfunction in T2D [14], we also analysed the effect that M value, ADDR, MAGE, SD and CV, as an index of short-term GV, had on DPN, represented by abnormal sudomotor function. Again M value was proved to be a strong and stable predictive element for sudomotor dysfunction in T1D. By contrast, the other indicators are less effective predictors. ADDR was statistically associated with sudomotor dysfunction in the crude logistic model, but after adjustment of several confounding factors like sex, age, diabetes duration, BMI, Alb, smoking, BUN, DN, and HbA1c, the relationship was inexistent with borderline significance. MAGE, SD and CV showed no relationship with sudomotor function firmly (data not shown).
Our study had a certain degree of limitation. In essence, this retrospective study cannot draw a cause-and-effect conclusion but can only describe the correlation between TIR and sudomotor function. Apart from that, all participants received 3 d of CGM, which perhaps made our results unrepresentative. Last but not least, despite relatively small number of samples in our cohort, patients with T1D are often regarded more homogeneous than patients with T2D. These people usually lack the other components of metabolic disorder so common for patients with T2D [37]. And we strictly conformed to inclusion and exclusion criteria to avoid the baseline characteristics of subjects vary enormously, which facilitates the chance of disclosing an association between CGM glucometrics and early diabetic complications in T1D.