Background: Previous studies have illustrated clinical associations between diabetic peripheral neuropathy (DPN) and diabetic kidney disease (DKD). Quantitative sensory testing (QST) can accurately detect thermal perception abnormalities and aid in the early diagnosis of asymptomatic small-fiber DPN in patients with type 2 diabetes. The aim of this study was to determine the predictive value of thermal perception abnormalities by QST to detect DKD. Methods: We prospectively enrolled 432 patients with type 2 diabetes (50.2% male, mean age 57.2 years, and average duration of diabetes 9.9 years) at our hospital between 2016 and 2017. Demographic and clinical data of the patients were recorded and analyzed. Diagnosis and staging of DKD were determined by urinary albumin excretion rate and estimated glomerular filtration rate. The presence of thermal perception abnormalities was determined by QST. Multiple logistic regression and receiver operating characteristic (ROC) analyses were performed to investigate the relationships between thermal perception abnormalities and DKD in these patients. Results: In multiple regression analysis, abnormal cold perception in the lower limbs was associated with an increased risk of advanced DKD. Area under the ROC curve analysis revealed that four-limb cold perception abnormalities had the best discriminatory power (0.741 ± 0.053) to predict advanced DKD. Conclusions: Our results demonstrate the value of using thermal perception abnormalities to identify patients with type 2 diabetes also at risk of DKD.
Type 2 diabetes mellitus (T2DM) is one of the most important health problems worldwide and is associated with a wide variety of macrovascular and microvascular complications and increased morbidity and mortality [1-4]. Diabetic peripheral neuropathy (DPN) and diabetic nephropathy are 2 major chronic microvascular complications in T2DM patients and can lead to serious consequences including foot ulceration, lower extremity amputation, and renal failure. Thus, screening and early diagnosis of these T2DM complications are critical to provide an opportunity for effective interventions and treatment to prevent progression to more advanced stages [5, 6].
DPN is characterized by pain, tingling, and numbness, with prevalence rates ranging from 30 to 50% . It is often associated with many comorbidities, including cognitive impairment, depression, autonomic neuropathy, peripheral artery occlusive disease (PAOD), nephropathy, retinopathy, cardiovascular disease, and medial arterial calcification [8, 9]. The diagnosis of DPN is most often made based on suggestive clinical symptoms and neurologic tests. However, many patients with DPN have no obvious neuropathic symptoms despite showing evidence of neurologic deficits on nerve conduction studies (NCSs) or electromyography. NCSs are considered to be the gold standard to diagnose DPN, but they cannot be performed in all diabetic patients due to their difficulty to conduct, time-consuming, and cost [10, 11]. In addition, NCSs only assess the function of large myelinated fast-conducting Aa and Ab fibers, and the test is usually normal if diabetic neuropathy predominately involves small-fiber neuropathy . Due to the insidious nature of symptoms, DPN is often not diagnosed in the early stages. The severity of thermal perception abnormalities can reflect the extent of small-fiber neuropathy . Quantitative sensory testing (QST) is a cheaper, portable, painless, and easy-to-use tool that can accurately detect thermal perception abnormalities. It is suggested that thermal perception abnormalities screening by QST can aid in the early diagnosis of asymptomatic small-fiber DPN .
Diabetic kidney disease (DKD) has been associated with a rapid progression in renal impairment and is now the most significant etiology of chronic kidney disease (CKD) and end-stage renal disease globally. The progression of DKD can be attributed to not only diabetic nephropathy, but also some other diabetes-related comorbid conditions, such as neurologic bladder, frequent urinary tract infection, macrovascular angiopathy, and adverse effects of medication. Patients with advanced DKD are at high risk of kidney failure and death. A recent study had showed that estimated glomerular filtration rate (eGFR) <45 mL/min/1.73 m2 (CKD stage 3b) at the time of nephrologist referral was an important risk factor for adverse outcomes in diabetes patients. Fortifying timely recognition and further optimization of nephroprotection measures for patients at risk of advanced DKD is an important issue in patients with diabetes [15-17].
Previous studies have indicated that DKD is significantly associated with DPN, and that patients with DPN have a higher incidence of albuminuria and lower eGFR than patients without DPN [18, 19]. Sheen et al.  had reported that thermal perception abnormalities of the lower limbs by QST are significantly associated with lower eGFR and higher albuminuria in type 2 diabetes. Given that the chronology of the occurrence of DPN and DKD in patients with T2DM is still unclear, we hypothesized that the severity of thermal perception abnormality of the 4 limbs by QST can help to detect patients at risk of kidney failure. Therefore, in this study, we aimed to determine the predictive value of thermal perception abnormality to detect the presence of advanced DKD in patients with T2DM.
Materials and Methods
Study Design and Population
This study was performed between October 2016 and December 2017 at Chang Gung Memorial Hospital, Keelung, Taiwan. Patients with T2DM between the ages of 18 and 70 years who attended an outpatient clinic in the hospital were recruited for the study. The exclusion criteria were patients with a urinary tract infection over the past 3 months, poorly controlled hypertension (HTN), thyroid function disorder, pregnancy, malignancy, liver cirrhosis, polycystic kidney disease, end-stage renal disease, systemic lupus erythematosus, and confirmed neuropathy owing to nondiabetic causes (alcoholism, nutrition deficiency, or uremia). All of the patients completed a survey including demographic information, common biochemical parameters, and assessments of macro- and microangiopathies. All of the study subjects were treated according to the American Diabetes Association diabetes mellitus treatment guideline .
The presence of DKD was identified by assessing albuminuria and eGFR. Albuminuria was identified by calculating urine albumin/creatinine ratio (UACR, mg/g) in first-voided spot urine samples, and eGFR (mL/min/1.73 m2) was calculated using the Modification of Diet in Renal Disease study equation [21, 22]. The patients were defined as having advanced DKD if they had an eGFR <45 mL/min/1.73 m2[16, 23]. Coronary artery disease and cerebral vascular accidents were recorded by chart review. PAOD was diagnosed according to the ankle-brachial index (ABI) and was defined as the lowest values of right and left ABIs as measured by Doppler ultrasound. ABI is a measure of the blood pressure in the arteries supplying legs relative to central, aortic pressure (approximated by measuring the blood pressure in the arm). The right and left ABIs were determined as the right and left ankle systolic pressures divided by the highest brachial systolic pressure, respectively. PAOD was defined as an ABI ≤0.9 or >1.3 . Retinopathy was diagnosed based on the findings of an ophthalmic examination, including the best-corrected visual acuity and fundus examination by indirect ophthalmoscopy conducted by a retinal specialist. Macular edema was judged to be present if retinal thickening or hard exudates were detected at or within a 1-disc diameter of the center of the macula. Severity of retinopathy was classified according to the most severe changes in the worst-affected eye . Neuropathy was judged to be present based on the following: (1) clinical symptoms of burning pain, paresthesia, hyperesthesia, tingling, and painful cramps involving the lower limbs; and (2) diminished fine touch sensation and impaired vibratory sensation based on the findings of nerve tests, including pinprick, vibration sensation, 10-g monofilament tests, and absence of Achilles reflex. In brief, pinprick sensation was measured using sterile Neurotip (Owen Mumford, Oxford, UK) 4 times on the same site. Vibration testing by the on-off method was conducted using a 128-Hz tuning fork applied to the bony prominence just proximal to the nail bed on the dorsal side of the big toe. The patient reported perception of both the start of the vibration sensation and the cessation on dampening, conducted twice on each toe. Vibration testing by the timed method is performed by the patient reporting the time at which vibration is reduced to an undetectable level. The tuning fork was then applied to the back of the phalanx of the examiner’s thumb. Then, the time (in seconds) at which the vibration sensation is weakened from the sides to beyond the examiner’s perception was added to provide a score. The Vibration Perception Threshold test is quantitatively measured by a limit method using Medoc equipment (Medoc Advanced Medical Systems, Durham, NC, USA). Monofilament tests were performed bilaterally using a 10-g monofilament. First, a reference stimulation is applied to the forehead or sternum. With the patients’ eyes closed, the monofilament is applied to the area just proximal to the nail bed on the noncallused dorsal side of the big toe using a smooth motion: touching the skin, bending the monofilament for a full second, and then lifting it from the skin. This maneuver was repeated 4 times per foot in a random arrhythmic manner. Each test was performed by an examiner blinded to results of all other examinations [20, 26]. Doppler ultrasound, fundoscopic examinations, and nerve tests were routinely performed when the subjects were enrolled in this study .
Determination of Thermal Perception Abnormalities by QST
QST was performed using a Medoc device (TSA2001/VSA3001) following previously published procedures. All measurements were performed on the thenar eminence of the dominant hand and the lateral distal aspect of the foot dorsum of the same side. The adaptation temperature of the probe was 32°C, and its contact area was 30 × 46 mm2. Using the method of limits, a threshold was determined as the average of 4 successive stimuli for cold and warm sensations. Probe temperature change rates of 1°C were used for cold and warm thresholds. Five seconds elapsed before switching to the next sensory stimuli of the nonpainful modalities. All of the samples were measured twice, and the mean value was used for further statistical analysis [9, 27]. Upper limb warmth perception abnormality (ULWPA) was defined as abnormal warmth perception in the right or left upper limbs (0–2 points). Upper limb cold perception abnormality (ULCPA) was defined as abnormal cold perception in the right or left upper limbs (0–2 points). Lower limb warmth perception abnormality (LLWPA) was defined as abnormal warmth perception in the right or left lower limbs (0–2 points). Lower limb cold perception abnormality (LLCPA) was defined as abnormal cold perception in the right or left lower limbs. Upper limb warmth and cold perception abnormality (ULWCPA) was defined as a combination of both warmth and cold perception abnormalities in the upper limbs (0–4 points). Lower limb warmth and cold perception abnormality (LLWCPA) was defined as a combination of both warmth and cold perception abnormalities in the lower limbs (0–4 points). Four-limb warmth perception abnormality (FLWPA) was defined as abnormal warmth perception in the right or left/upper or lower limbs (0–4 points). Four-limb cold perception abnormality (FLCPA) was defined as abnormal cold perception in the right or left/upper or lower limbs (0–4 points). Four-limb warmth or cold perception abnormality (FLWCPA) was defined as a combination of both warmth and cold perception abnormalities in all 4 limbs (0–8 points).
Continuous data were expressed as means and standard deviations, and categorical data were expressed as percentages. In the primary analysis, patients with and without advanced DKD were compared. Continuous variables were tested for normal distribution using the Kolmogorov-Smirnov test. Student’s t test was used to compare the mean values of continuous variables and normally distributed data; in other cases, the Mann-Whitney U test was used. Categorical data were tested using the χ2 test. We assessed the risk factors for advanced DKD using univariate analysis, and the variables that were statistically significant (p < 0.05) in the univariate analysis were included in the multivariate analysis. A multiple logistic regression model and backward elimination of data were used to analyze these variables.
Correlations between paired groups of variables were assessed using Spearman’s rank correlation analysis. Discrimination was calculated using the area under receiver operating characteristic (AUROC) curves. The AUROC values were compared using a nonparametric approach. AUROC analysis was also used to calculate the cutoff values, sensitivity, specificity, and overall correctness. Finally, cutoff points were calculated by calculating using the Youden index (sensitivity + specificity − 1). All statistical analyses were 2-tailed, and a p value of <0.05 was considered to be statistically significant. All data were analyzed using the Statistical Package for Social Sciences software, version 20.0 for Windows (SPSS, Inc., Chicago, IL, USA).
Baseline Study Characteristics
A total of 432 patients were included in this study (mean age 57.2 years; 50.2% male), of whom 53 (12.3%) had advanced DKD. The baseline characteristics of the advanced and nonadvanced DKD groups are shown in Table 1. The advanced DKD group was older, had a longer duration of diabetes, and higher percentages of HTN, PAOD, retinopathy, and neuropathy. There were no significant differences in gender, BMI, the prevalence of coronary artery disease, cerebral vascular accidents, smoking, and family history of diabetes between the 2 groups. The average UACR and triglyceride levels were significantly higher in the advanced DKD group, while the average levels of eGFR, low-density lipoprotein (LDL) cholesterol, and alanine aminotransferase were significantly higher in the nonadvanced DKD group. There were no significant differences in the levels of total cholesterol, high-density lipoprotein cholesterol, HbA1C, high-sensitivity C-reactive protein, and homocysteine between the 2 groups.
Risk Factors for Advanced DKD
Univariate analysis showed that age, diabetes duration, HTN, PAOD, retinopathy, neuropathy, UACR, LDL cholesterol, triglycerides, alanine aminotransferase, and upper and lower extremity warmth and cold perception abnormalities were good prognostic indicators for advanced DKD. To adjust for imbalances in baseline characteristics and multiple associations between temperature perception abnormalities in the extremities and other variables, we used 2 multivariate models to evaluate the impact of temperature perception abnormalities in the extremities of patients with advanced DKD. The results showed that age, diabetes duration, HTN, triglycerides, and LLCPA had independent prognostic significance for advanced DKD (Table 2). Regression coefficients of these variables were used to calculate the odds of advanced DKD in each patient as follows:
Logarithm of the odds of advanced DKD = −12.870 + 0.757 × LLCPA + 0.096 × age + 0.006 × diabetes duration + 2.634 × HTN + 0.005 × triglycerides.
Correlations among Sensory Abnormalities in the Extremities and Diabetes-Associated Complications
Spearman’s rank correlation analysis revealed that all extremity thermal perception abnormalities were significantly negatively correlated with eGFR and that ULWPA, LLWPA, and LLCPA were significantly positively correlated with retinopathy. However, only LLCPA was significantly positively correlated with UACR (Table 3). The correlations between the severity of thermal perception abnormality and the features associated with advanced DKD, including eGFR, UACR, BMI, LDL, HbA1C, and HTN, are shown in Figure 2. It was observed that the degree of FLWCPA was positively correlated with that of UACR, HbA1C, and HTN, while negatively correlated with that of eGFR, BMI, and LDL.
Discrimination of the Results of QST Tests
Comparisons between the discriminatory abilities of the extremity sensory abnormalities determined by QST tests and other diabetes-associated complications are shown in Table 4. The discriminatory powers of FLCPA and FLWCPA were similar to that of UACR and significantly higher than those of PAOD and retinopathy. Among all the extremity abnormalities, AUROC analysis showed that FLCPA had the best discriminatory power.
The prevalence of temperature perception abnormalities among different CKD stages is shown in Figure 1. A progressive and significant increase in the prevalence of all extremity sensory abnormalities was correlated with increasing CKD stage of the patients. Among all extremity temperature perception abnormalities, ULWPA had the highest prevalence in our patients. In further comparisons, FLWPA had a higher prevalence than FLCPA, while ULWCPA had a higher prevalence than LLWCPA.
In this study, the mean diabetes duration of our patients was 9.9 years, and the prevalence of neuropathy was 23.8%, which is consistent with the findings of previous reports . Among the 432 patients, 53 (12.3%) had advanced DKD. Our analysis revealed that age, diabetes duration, HTN, triglycerides, and abnormal cold perception in the lower limbs were independent predictors of advanced DKD in the patients with T2DM (Table 2). Our results also showed that abnormal cold perception in the lower limbs was a good predictor of advanced DKD (AUROC 0.718). Moreover, abnormal cold perception in the 4 limbs had an even better discriminatory power, which was similar to that of UACR (AUROC 0.741 vs. 0.830, p = 0.094) and significantly better than that of PAOD, retinopathy, neuropathy, and other temperature perception abnormalities by QST (Table 4).
Previous investigations had reported that the duration of diabetes, age, triglycerides, and HTN were linked to the development and progression of diabetic small-fiber neuropathy [29-31]. The correlations between these 4 factors and the occurrence of renal complications have also been well described in type 1 diabetes [32, 33]. In the current study, we further demonstrated that these 4 variables were also independent predictors of advanced DKD in patients with T2DM (Table 2). Moreover, our findings confirmed the good discriminative power and independent predictive value of abnormal cold perception in the lower limbs for prediction of advanced DKD in these patients (Table 4). The correlations among all temperature perception abnormalities by QST were good, and the correlation between each temperature perception abnormality and eGFR was better than neuropathy (Table 3).
In this study, we also investigated the relationships between temperature perception abnormalities by QST and different stages of CKD in an attempt to elucidate whether temperature perception abnormalities by QST can be used as an early warning sign of DKD development. Our results showed that the prevalence of temperature perception abnormalities significantly increased with the progression of CKD (Fig. 1). DPN is characterized by progressive distal and symmetrical peripheral neuropathy of sensory nerve fibers, followed by autonomic and motor involvement . Small-fiber neuropathy with abnormal thermal perception is the most frequent sensory impairment . Abnormal cold perception can reflect dysfunction of myelinated Ad small fibers, and abnormal warmth perception can reflect dysfunction of unmyelinated C small fibers . In this study, the prevalence of abnormal warmth perception was higher than that of abnormal cold perception in all 4 limbs. Our data are consistent with those of Weintrob et al.  who reported that unmyelinated C small fibers are more vulnerable than myelinated Ad small fibers in diabetes [12, 36]. Furthermore, previous studies have reported that diabetic small-fiber neuropathy is length dependent [37, 38]. Longer nerves are more easily destroyed, so the distal nerves in the lower extremities are more vulnerable than those in the hands. Interestingly, our data showed that the prevalence of abnormal cold and warmth perception in the lower extremities was lower than that in the upper extremities by QST (Fig. 1). However, the difference in the prevalence of abnormal thermal perception between the upper and lower extremities was not significant in all different CKD stages. It is well known that multiple interrelationships exist among thermal perception abnormalities, and many potential confounders have been studied. Only abnormal cold perception in the lower extremities was found to be an independent predictor for advanced DKD after we performed multiple logistic regression analysis.
QST is a simple, reproducible, and noninvasive evaluation tool that has high diagnostic validity for sensory abnormality, especially diabetic small-fiber neuropathy [13, 14]. To the best of our knowledge, this is the first study to demonstrate that QST can help generate objective information for patients and physicians and aid in the early detection of advanced DKD. Patients with advanced DKD are known to have significant risks of various morbidities and mortality . The early detection and timely referral to a nephrology specialist are critical and may aid in the prevention and/or management of these complications. These are important strategies to minimize suffering and the costs associated with this disease. Our study indicates that QST may be a valuable screening tool to identify individuals at risk of advanced DKD, especially in remote areas with a shortage of health resources.
In the past decades, emerging evidence has indicated central nervous system involvement in DPN, and the association between cognitive impairment and DPN in diabetes had been reported [40-42]. It is suggested that dysregulation of glycemic control may lead to microvascular dysfunction in the blood-brain barrier, autoimmune damage to neurologic diseases, and changes in the synthesis, availability, or reuptake of neurotransmitters. In the pathogenesis of DPN, there is an interaction between the peripheral pathway and the central pathway. A study investigating 36 type 2 diabetes patients revealed that the alteration of the central pathway of DPN documented by MRI may precede clinically evident cognitive impairment . The presence and extent of albuminuria is associated with the cognitive changes in CKD . In the current study, we confirmed that the prevalence of temperature perception abnormalities by QST increased with the progression of CKD, and the severity of temperature perception abnormality was also positively correlated to the degree of UACR and glycemic control, which might reflect endothelial damage (Fig. 2). Considering diabetes and CKD are correlated with increased risk of reversible cognitive dysfunction [41, 42], future personalized and specialized screening protocols for cognitive impairment in diabetes patients with thermal perception abnormality might help to allow timely interventions. Further well-powered research is needed to study this issue.
In spite of the encouraging results observed in this study, several potential limitations should be recognized. First, this study was conducted at a single medical center, and so the generalizability of our findings to other hospitals with different patient populations may be limited. Second, the results of the QST were based on subjective responses, so they were valid only when the patient was cooperative. Third, for lacking information about medication and diet control, we did not adjust these 2 important factors in the logistic regression models. Fourth, sequential measurements of thermal perception using QST (e.g., semiannually and annually) may reflect the dynamic aspects of clinical diseases and thus may have provided better information about the risk of advanced DKD risk in our patients. Finally, the predictive accuracy of logistic regression models has its own limitations.
In conclusion, the present study provides clinical evidence demonstrating an association between abnormal thermal perception by QST and an increased risk of advanced DKD in patients with type 2 diabetes. Our analysis also showed that abnormal cold perception in the lower limbs was an independent predictor of advanced DKD in these patients. Our results suggest that screening for DKD should start soon after the development of abnormal cold perception in the lower limbs by QST. Further prospective clinical studies with a larger sample size and long-term follow-up are warranted to validate the potential role of QST in reducing the burden on healthcare systems, especially in remote areas with a shortage of health resources.
The authors thank the staff of the Community Medicine Research Center of Keelung CGMH. They also express their sincere gratitude to all participants of the Taiwan Consortium for Acute Kidney Injury and Renal Diseases (CAKs).
Statement of Ethics
This study was conducted in full compliance with the ethical principles of the Declaration of Helsinki, the Good Clinical Practice guidelines, and the applicable local regulatory requirements. Subjects were invited to participate in this study when they visited our outpatient clinic. A trained endocrinologist examined all of the patients during screening and informed them about the consent procedure. Written informed consent was obtained from all subjects prior to their participation. This study was approved by the Institutional Review Board of Chang Gung Memorial Hospital (Approval No. 103-5291B).
Conflict of Interest Statement
The authors have no conflicts of interest to disclose.
This study was supported by the Chang Gung Memorial Hospital Research Program (CMRPG2F0171, CMRPG2F0172, CMRPG2F0173, CRRPG2H0161, CRRPG2H0162 and CMRPG2J0261) and the Ministry of Science and Technology of Taiwan (MOST 108-2321-B-182-003 and MOST 109-2321-B-182-001).
W.C.F., H.C.P., and K.M.C. contributed to the conception, design, and interpretation of data. W.C.F., K.M.C., and H.C.P. contributed to data collecting and manuscript drafting. C.Y.S., W.I.W., C.C.L., and Y.C.C. provided patient information, participated in the design and coordination, and helped draft the manuscript. C.C.L., C.Y.S., W.I.W., Y.C.C., and staff of the Community Medicine Research Center provided intellectual content for the work and were involved in editing and revising the manuscript. All authors discussed, contributed to, and approved the final manuscript version.
Wei-Ching Fang and Kuei-Mei Chou contributed equally to the manuscript.