Abstract
Aims/Introduction
We investigated the classification of diabetic peripheral neuropathy (DPN) patients by subjective symptoms, and identification of the relationship between the patterns and intensities of symptoms and the clustered groups of DPN patients.
Materials and Methods
This multicenter study analyzed epidemiological data and sensory symptoms of 649 patients with DPN. Cluster analysis was carried out to identify subgroups of patients with characteristic symptom profiles. Factor analysis was carried out to investigate the symptom patterns of the clustered groups of DPN patients.
Results
Three clusters of patients with DPN were identified: severe symptoms with decreased quality of life (cluster 1, n = 119, 18.3%), predominantly insensate symptoms with relatively good quality of life (cluster 2, n = 318, 49.0%), and moderate pain intensity and decreased quality of life (cluster 3, n = 204, 31.4%). The frequency of symptoms on each item of the Michigan Neuropathy Screening Instrument questionnaire showed a similar distribution according to pain intensities along with the three clusters.
Conclusions
Our study supports the hypothesis that diversity in sensory symptoms exists in patients with DPN. Heterogeneity in DPN patients should be taken into account for a more stratified or individualized treatment approach. Based on a multicenter study, we identified three clusters of patients with DPN. Our research supports the hypothesis that diversity in sensory symptoms exists in patients with DPN. Heterogeneity in DPN patients should be taken into account for a more stratified or individualized treatment approach.
Keywords: Diabetes mellitus, type 2; Diabetic neuropathies; Phenotype
Based on a multi‐center study, we identified three clusters of patients with diabetic peripheral neuropathy. Our research supports the hypothesis that diversity in sensory symptoms exists in patients with diabetic peripheral neuropathy. Heterogeneity in diabetic peripheral neuropathy patients should be taken into account for a more stratified or individualized treatment approach.
INTRODUCTION
Diabetic peripheral neuropathy (DPN) is one of the most common complications of diabetes, and is a significant cause of morbidity, mortality and diminished quality of life (QOL). A previous study reported that the prevalence of neuropathy in patients with type 2 diabetes was 33.5% 1 , and the prevalence of painful DPN is 43.1% in patients with DPN in Korea 2 . DPN is a heterogeneous disease with widely varying pathology, suggesting differences in perception and recognition of sensory symptoms among patients. Although numerous pharmacological agents have been proven to be effective for patients with DPN, they are often ineffective and poorly tolerated. This is mainly due to difficulties in objectifying DNP symptoms in real practice as a result of its diverse symptoms or pain patterns and the subjective description of symptoms by patients. Therefore, a comprehensive approach for DPN management is necessary by classifying or subtyping patients according to subjective symptoms and objective test results. However, few studies have classified DPN patients according to subjective symptoms and identified the characteristics of the patient groups with similar symptoms.
The present study aimed to establish clustered groups of DPN patients with a composite of subjective symptoms, and clinical impacts on pain severity, sleep disturbance and QOL. Furthermore, our objective was to identify the relationship between the intensity levels of symptom patterns and clustered groups of DPN patients, and to evaluate the clinical significance of the currently used composite symptom score questionnaire.
METHODS
Study population and design
Patients with type 2 diabetes aged >19 years from 10 hospitals in Korea were included in this nationwide, multicenter, cross‐sectional, observational study from January 2017 to May 2017. Inclusion criteria comprised of the following: (i) patients aged >19 years, classified as type 2 diabetes mellitus according to the 2017 American Diabetes Association guidelines or those who are being treated with diabetes medications, including insulin; (ii) patients who have access to medical records and data during the entire study period; and (iii) patients whose symptoms and QOL have been assessed or documented by several questionnaires. Exclusion criteria comprised of: (i) type 1 diabetes; (ii) causes of neuropathy other than diabetes (such as vitamin B12 deficiency, chronic alcohol abuse, drug‐induced neuropathy); (iii) severe renal disease or liver disease; (iv) advanced malignancy, active infection; (v) cervical or lumbar spondylosis; and (vi) pregnancy, lactation or childbearing age without use of safe contraception. Participants filled in questionnaires about demographic data (such as age, sex, diabetes duration in years), alcohol, smoking, glycemic control and comorbidities. Clinical history of major cardiovascular events (ischemic heart disease, stroke and peripheral vascular disease) and diabetes complications (retinopathy, nephropathy) were collected from medical records. The list of medications for diabetes mellitus and DPN, and laboratory data obtained within 3 months before enrollment were collected from available sources.
Measures
Data regarding neuropathy symptoms, QOL and perception of overall health status of the patients were collected by trained healthcare workers using specific questionnaires in each hospital. All questionnaires were administered through face‐to‐face interview with healthcare professionals.
Symptoms of DPN were assessed using the Michigan Neuropathy Screening Instrument questionnaire (MNSI), a structured examination involving inspection of the feet and evaluation of fine touch (using a Semmes–Weinstein 5.07 10‐g monofilament), vibration perception (using a 128‐Hz tuning fork) and ankle reflexes. In addition, the diagnosis of neuropathic pain was made using another validated screening tool, the painDETECT questionnaire. It evaluates pain intensity, pain pattern and pain quality. By summing up the scores given in each domain, a final score between −1 and 38 can be achieved, with higher scores indicating more likely neuropathic pain 3 . Trained technicians examined the feet of participants for deformities, dry skin, callus and ulcerations. Also, muscle strength reflexes were tested using a hammer at the Achilles tendon on both ankles. Vibration sense tests were carried out on the interphalangeal joints of both great toes, and the monofilament test at the dorsum of both great toes 4 .
Diagnosis of DPN
DPN was diagnosed based on the presence of either of the following criteria: (i) an MNSI score ≥3; or (ii) symptoms of pain, burning, tingling and/or loss of sensation, and abnormal or decreased/absent ankle reflexes; and (iii) at least one of the tests for neuropathy is positive. Such criteria are based on the recommendations of the Toronto Diabetic Neuropathy Expert Group 5 . Most of the patients in this study were classified as probable DPN.
Patient‐reported outcomes measures
Quality of life was assessed using the following questionnaires: the modified Korean version of the Brief Pain Inventory‐Short Form (BPI‐SF) to evaluate the severity and interference of pain with daily functioning; the six‐item Medical Outcomes Study (MOS) Sleep Scale to measure overall sleep quality using a sleep problem index; the Korean version of the EuroQoL Health (EQ‐5D) with a standardized five‐item measure of health profiles; and the visual analog scale (VAS) to assess pain intensity.
BPI‐SF
The modified BPI‐SF Korean version used in the present study contains a four‐item pain severity scale (worst, least, average and current pain) and a seven‐item pain interference scale (general activity, mood, walking ability, normal work, relationships with others, sleep and enjoyment of life). Each BPI item is equidistantly bounded on a 0–10 numeric rating scale, having 0 as ‘no pain’ and 10 as ‘pain as bad as you can imagine’ for severity, or ‘does not interfere’ to ‘completely interferes’ for interference 2 .
Six‐item MOS‐Sleep scale
A six‐item version of the MOS‐Sleep Scale was used in the present study. The six‐item summary scale includes trouble falling asleep (MOS‐7), awaken during sleep (MOS‐8), awaken short of breath or with headache (MOS‐5), enough sleep to feel rested (MOS‐4), amount of sleep needed (MOS‐12) and trouble staying awake during day (MOS‐9). The MOS‐Sleep Scale measured sleep quality and problems over the previous 4 weeks. The patients responded on a 6‐point scale, ranging from 1 for ‘all of the time’ to 6 for ‘none of the time’. To represent a measure of overall sleep quality, a sleep problem index was calculated as the sum of points of all six sleep items 2 .
EQ‐5D
The EQ‐5D was used to measure the health‐related QOL of the patients. The descriptive system consists of five dimensions: mobility, self‐care, usual activities, pain/discomfort and anxiety/depression. Each dimension has three response levels indicating ‘no problem’ (or 1), ‘some problems’ (or 2) and ‘severe problems’ (or 3). The sum of the scores of the five domains were used as a single preference‐based index score (EQ‐5D index) 2 .
VAS
The VAS is based on self‐reported measures of symptoms that are a 10‐cm line with anchor statements on the left (no pain) and right (extreme pain). The patients were asked to mark their current pain level on the line. Measurements from the “no pain” anchor point of the scale to the patients' marks are recorded in centimeters and are interpreted as their pain. The score 10 represents the worst imaginable health state and 0 represents the best imaginable health state.
Statistical analysis
K‐means cluster analysis was carried out using the scores from the following scales: MNSI (total score, 13); BPI‐SF (total score, 40); six‐item MOS sleep scale (total score, 36), EQ‐5D (total score, 15) and VAS score (total score, 10). Two analytical approaches were applied to improve the reliability of the best cluster solution for the number of clusters in the dataset. First, we used Ward's method and dendogram for hierarchical cluster analysis to estimate the number of likely clusters within the studied population. Next, we identified the cluster number to minimize the total within‐cluster sum of squares by iterating for all of the clusters of the K‐means algorithm 6 . Results from both algorithms indicated that the optimal number of clusters was three. We then examined each variable for differences among the clusters.
Exploratory factor analysis was carried out to identify neuropathic symptom patterns using 12 of 15 items of the MNSI self‐administered questionnaire (three questions excluded: questions number 4, 8 and 9 were excluded due to low communality in factor analysis).
The varimax rotation method was used to simplify the interpretations of summary factors. The Kaiser–Mayer–Olkin measure and Bartlett's test of sphericity were carried out to find sampling adequacy. Factors with eigenvalues >1 were considered. The significance level for interpretation of factor loadings was 0.40. Cronbach's alpha was calculated to measure the reliability of these factors. Finally, the summary factors were classified according to the loaded symptoms on a specific factor. Data analyses were carried out using SPSS Statistics for Windows, version 26.0 (SPSS Inc., Chicago, IL, USA); significance was set at P < 0.05. Continuous variables were presented as the mean ± standard deviation, and categorical data as frequencies and percentages were compared between clusters by anova tests, or χ2 or Fisher's exact tests.
RESULTS
Pain location and treatment pattern of the study participants
In the questionnaire regarding pain areas described by patients, painful sensation in the front of feet was reported by 69.9% (n = 448) of the patients with DPN, followed by 40.9% (n = 262) in the back of feet, and approximately 20% in the anterior legs between the ankle and knee (Appendix A). Most of the DPN patients complained of bilateral pain mainly in the lower extremities. In the treatment pattern of patients with DPN, anticonvulsants, such as gabapentin and pregabalin, were the most commonly used medications for pain control, whereas alpha‐lipoic acid was most commonly used as a potentially disease‐modifying therapeutic drug, followed by gamma‐linoleic acid (Appendix B).
Clinical characteristics of the three identified clusters
The vast majority of patients (98.7%) were included in the clusters resulting from the present cluster analysis. The three identified clusters (clusters 1, 2 and 3) included 119 (18.3%), 318 (49.0%) and 204 (31.4%) patients, respectively. Baseline characteristics per cluster are summarized in Table 1. Cluster 1 had a higher proportion of women, and patients had a higher body mass index and fasting blood glucose (FBG) and triglyceride compared with the patients in cluster 2 and 3. Patients in cluster 1 had longer duration of diabetes and diabetic neuropathy than those in cluster 2 or cluster 3. There were no significant differences among groups in any of the baseline parameters, such as age, glycated hemoglobin, low‐density lipoprotein cholesterol, creatinine, systolic blood pressure and treatment pattern of diabetes.
Table 1.
Baseline characteristics of the three clusters
| Variable |
Total n = 649 |
Cluster 1 n = 119 |
Cluster 2 n = 318 |
Cluster 3 n = 204 |
P‐value |
|---|---|---|---|---|---|
| Age (years) | 65.8 ± 12.0 | 66.3 ± 12.2 | 65.6 ± 11.7 | 66.1 ± 12.1 | 0.087 |
| Female, n (%) | 355 (54.7) | 82 (68.9)*** | 165 (51.9) | 104 (51.0) | 0.003 |
| Duration of diabetes (years) | 14.6 ± 9.5 | 16.6 ± 9.3** | 15.1 ± 9.2 | 13.1 ± 9.7 | 0.003 |
| Duration of DPN | 6.1 ± 5.2 | 7.0 ± 4.3* | 5.5 ± 4.3 | 6.6 ± 6.7 | 0.015 |
| BMI | 25.0 ± 3.8 | 26.2 ± 3.9*** | 25.0 ± 3.6 | 24.5 ± 4.0 | 0.001 |
| Diabetes treatment | |||||
| Diet and exercise | 5 (0.8) | 1 (0.8) | 2 (0.6) | 2 (1.0) | 0.110 |
| OHA | 339 (52.2) | 56 (47.1) | 165 (51.9) | 112 (54.9) | |
| Insulin | 40 (6.2) | 13 (10.9) | 14 (4.4) | 11 (5.4) | |
| Insulin + OHA | 250 (38.5) | 44 (37.0) | 133 (41.8) | 73 (35.8) | |
| GLP‐1 RA + OHA + insulin | 15 (2.3) | 5 (4.2) | 4 (1.3) | 6 (2.9) | |
| FBS | 153.0 ± 60.8 | 174.1 ± 83.7*** | 144.2 ± 47.3 | 152.8 ± 58.6 | <0.001 |
| HbA1c | 8.00 ± 1.72 | 8.0 ± 1.7 | 7.9 ± 1.5 | 8.1 ± 2.0 | 0.376 |
| SBP | 125.2 ± 15.00 | 125.4 ± 14.7 | 124.1 ± 14.4 | 126.8 ± 15.7 | 0.139 |
| DBP | 71.6 ± 11.3 | 72.8 ± 10.0* | 70.4 ± 11.0 | 72.7 ± 12.2 | 0.031 |
| Total cholesterol | 155.6 ± 89.3 | 148.8 ± 40.4 | 154.8 ± 38.5 | 149.1 ± 35.7 | 0.216 |
| Triglyceride | 149.2 ± 84.0 | 164.2 ± 102.0** | 151.5 ± 84.4 | 134.9 ± 67.8 | 0.009 |
| HDL cholesterol | 48.0 ± 14.0 | 46.6 ± 13.7 | 48.1 ± 14.6 | 48.9 ± 13.4 | 0.962 |
| LDL cholesterol | 79.5 ± 32.4 | 79.8 ± 33.7 | 79.2 ± 31.2 | 79.0 ± 33.3 | 0.383 |
| Creatinine | 1.2 ± 1.3 | 1.4 ± 1.4 | 1.1 ± 1.2 | 1.1 ± 1.3 | 0.164 |
| History of diabetic foot ulcer | 17 (2.6) | 3 (2.5) | 7 (2.2) | 7 (3.4) | 0.700 |
| History of amputation of lower limb | 12 (1.8) | 2 (1.7) | 4 (1.3) | 6 (2.9) | 0.363 |
Data are expressed as means ± standard deviation for continuous variables and frequency (%) for categorical variables.
P < 0.05 versus cluster 2.
P < 0.05 versus cluster 3.
P < 0.05 versus cluster 2 and 3.
BMI, body mass index; DBP, diastolic blood pressure; DPN, diabetic peripheral neuropathy; FBS, fasting blood glucose; GLP‐1 RA, glucagon‐like peptide‐1 receptor agonist; HbA1c, glycated hemoglobin; HDL, high‐density lipoprotein; LDL, low‐density lipoprotein; OHA, oral hypoglycemic agent(s); SBP, systolic blood pressure. [Correction added on September 15 2022, after first online publication: Values in ”Total” and “Cluster 1” columns in Table 1 have been corrected.]
According to foot examination findings, there were more patients with foot deformities in cluster 2 than cluster 1 and 3. There was no significant difference regarding patients with foot ulcers among the three groups. Cluster 1 had more patients with abnormal ankle reflexes and first toe vibration sensations than the other clusters. Also, cluster 1 had significantly more patients with abnormal findings in the 10‐g monofilament test. The overall foot examination score was significantly higher for patients in cluster 1 (2.8 ± 2.4) than cluster 2 and 3 (1.5 ± 1.6, 1.9 ± 2.0, respectively, P < 0.01; Table 2).
Table 2.
Characteristics of patient foot examination
| Variable |
Cluster 1 n = 119 |
Cluster 2 n = 318 |
Cluster 3 n = 204 |
P‐value |
|---|---|---|---|---|
| Foot deformity (right) | ||||
| Absent | 90 (75.6) | 200 (63.3) | 153 (75.4) | 0.004 |
| Present | 29 (24.4) | 116 (36.7) | 50 (24.6) | |
| Foot deformity (left) | ||||
| Absent | 92 (77.3) | 205 (64.9) | 160 (78.8) | 0.001 |
| Present | 27 (22.7) | 111 (35.1) | 43 (21.1) | |
| Foot ulcer (right) | ||||
| Absent | 113 (95.0) | 311 (98.4) | 195 (96.1) | 0.103 |
| Present | 6 (5.0) | 5 (1.6) | 8 (3.9) | |
| Foot ulcer (left) | ||||
| Absent | 114 (95.8) | 313 (99.1) | 196 (97.5) | 0.088 |
| Present | 5 (4.2) | 3 (0.9) | 5 (2.5) | |
| Ankle reflex (right) | ||||
| Present | 25 (25.3) | 196 (66.9) | 99 (52.9) | 0.000 |
| Reduced | 61 (61.6) | 90 (30.7) | 79 (42.2) | |
| Absent | 13 (13.1) | 7 (2.4) | 9 (4.8) | |
| Ankle reflex (left) | ||||
| Present | 26 (26.3) | 197 (67.2) | 106 (56.7) | 0.000 |
| Reduced | 62 (62.6) | 87 (29.7) | 72 (38.5) | |
| Absent | 11 (11.1) | 9 (3.1) | 9 (4.8) | |
| Vibration in 1st toe (right) | ||||
| Normal | 25 (25.3) | 226 (77.1) | 102 (54.5) | 0.000 |
| Impaired | 44 (44.4) | 53 (18.1) | 55 (29.4) | |
| Absent | 30 (30.3) | 14 (4.8) | 30 (16.0) | |
| Vibration in 1st toe (left) | ||||
| Normal | 24 (24.2) | 226 (77.1) | 108 (57.8) | 0.000 |
| Impaired | 45 (45.5) | 54 (18.4) | 48 (25.7) | |
| Absent | 30 (30.3) | 13 (4.4) | 31 (16.6) | |
| 10 g‐Monofilament (right) | ||||
| Normal | 44 (44.4) | 242 (82.6) | 137 (73.7) | 0.000 |
| Impaired | 34 (34.3) | 43 (14.7) | 36 (19.3) | |
| Absent | 21 (21.2) | 8 (2.7) | 14 (7.5) | |
| 10 g‐Monofilament (left) | ||||
| Normal | 44 (44.4) | 242 (82.9) | 139 (74.7) | 0.000 |
| Impaired | 35 (35.4) | 42 (14.4) | 32 (17.2) | |
| Absent | 20 (20.2) | 8 (2.7) | 15 (8.1) | |
| Total score | 2.8 ± 2.4*** | 1.5 ± 1.6 | 1.9 ± 2.0 | 0.000 |
Data are expressed as frequency (%) for categorical variables.
P < 0.05 versus cluster 2 and 3.
Comparison of QOL according to symptom clusters
The BPI‐SF measures, total score of the items for pain severity (worst, weakest, average, current pain) were numerically higher in cluster 1 than the other clusters. The score of each item for pain interference (general activity, mood, walking, normal work, relationship, sleep and enjoyment of life) was significantly higher in cluster 1 than the other clusters. The sleep problem index was significantly lower in cluster 1 than cluster 2 and 3. Therefore, patients in cluster 1 had more sleep disturbances than those of the patients in cluster 2 and 3. EQ‐5D index and VAS were significantly higher in cluster 1 than the other clusters. Patients in cluster 2 had mild pain and relatively normal sleep patterns and QOL compared with patients in clusters 1 and 3 (Table 3).
Table 3.
Clinical characteristics and clinical impacts of diabetic peripheral neuropathy on pain, sleep and quality of life in three clustered groups
| Variable |
Cluster 1 n = 119 |
Cluster 2 n = 318 |
Cluster 3 n = 204 |
P‐value |
|---|---|---|---|---|
| PainDETECT score | 17.9 ± 7.0*,** | 5.2 ± 3.7** | 9.5 ± 5.6 | 0.000 |
| MNSI score | 9.2 ± 1.8*,** | 4.3 ± 1.7** | 6.1 ± 2.1 | 0.000 |
| BPI | ||||
| Worst | 7.1 ± 1.9*,** | 2.3 ± 1.9** | 5.8 ± 2.1 | 0.000 |
| Weakest | 5.0 ± 2.8*,** | 1.1 ± 1.0** | 3.2 ± 2.3 | 0.000 |
| Average | 6.6 ± 2.0*,** | 2.0 ± 1.5** | 5.0 ± 1.7 | 0.000 |
| Current | 6.3 ± 2.3*,** | 1.1 ± 1.1** | 4.4 ± 2.1 | 0.000 |
| BPI total | 25.0 ± 7.5*,** | 6.5 ± 4.0** | 18.4 ± 6.5 | 0.000 |
| Treatment response (in the past 24 h) | 30.2 ± 25.3 | 26.1 ± 30.5** | 34.0 ± 28.0 | 0.010 |
| Pain interference items † | ||||
| General activity | 5.5 ± 2.8*,** | 0.8 ± 1.3** | 2.8 ± 2.3 | 0.000 |
| Mood | 6.1 ± 2.3*,** | 1.2 ± 1.7** | 2.9 ± 2.7 | 0.000 |
| Walking | 5.6 ± 2.6*,** | 0.9 ± 1.4** | 2.5 ± 2.4 | 0.000 |
| Normal work | 5.3 ± 2.7*,** | 0.7 ± 1.2** | 2.5 ± 2.4 | 0.000 |
| Relationship | 4.3 ± 3.0*,** | 0.5 ± 0.9** | 1.2 ± 1.8 | 0.000 |
| Sleep | 6.2 ± 2.9*,** | 0.9 ± 1.5** | 2.7 ± 2.7 | 0.000 |
| Enjoyment of life | 5.0 ± 3.1*,** | 0.8 ± 1.2** | 2.3 ± 2.5 | 0.000 |
| MOS sleep scale ‡ | ||||
| Enough sleep to feel rested | 2.7 ± 1.58*** | 3.6 ± 1.7 | 3.7 ± 1.6 | 0.000 |
| Awaken short of breath or with headache | 5.3 ± 1.1*** | 5.7 ± 0.7 | 5.7 ± 0.9 | 0.000 |
| Feel drowsy during day | 3.0 ± 1.2*** | 4.2 ± 1.3 | 3.9 ± 1.3 | 0.000 |
| Trouble falling asleep | 3.0 ± 1.7*** | 4.9 ± 1.5 | 4.4 ± 1.6 | 0.000 |
| Awaken during sleep | 3.1 ± 1.5*** | 4.9 ± 1.4 | 4.7 ± 1.6 | 0.000 |
| Trouble staying awake during day | 3.7 ± 1.5*** | 4.9 ± 1.3 | 4.8 ± 1.3 | 0.000 |
| Snoring during sleep | 3.7 ± 1.8*** | 4.4 ± 1.7 | 4.2 ± 1.8 | 0.000 |
| Sleep problem index | 19.2 ± 5.4*** | 26.8 ± 5.2 | 25.8 ± 5.37 | 0.000 |
| EQ‐5D § | ||||
| Mobility | ||||
| Some | 89 (74.8) | 44 (13.8) | 73 (35.8) | 0.000 |
| Severe | 6 (5.0) | 0 (0) | 1 (0.5) | |
| Self‐care | ||||
| Some | 43 (36.1) | 6 (1.9) | 11 (5.4) | 0.000 |
| Severe | 1 (0.8) | 0 (0) | 1 (0.5) | |
| Usual activity | ||||
| Some | 77 (64.7) | 11 (3.5) | 53 (26.0) | 0.000 |
| Severe | 7 (5.9) | 0 (0) | 0 (0) | |
| Pain/discomfort | ||||
| Some | 81 (68.1) | 170 (53.5) | 169 (82.8) | 0.000 |
| Severe | 38 (31.9) | 2 (0.6) | 16 (7.8) | |
| Anxiety/depression | ||||
| Some | 72 (60.5) | 32 (10.1) | 76 (37.3) | 0.000 |
| Severe | 21 (17.6) | 0 (0) | 0 (0) | |
| EQ‐5D index § | 9.3 ± 1.8*,** | 5.8 ± 0.9** | 7.1 ± 1.3 | 0.000 |
| VAS ¶ | 6.2 ± 2.1*,** | 1.2 ± 1.1** | 4.4 ± 1.7 | 0.000 |
Data are expressed as means ± standard deviation for continuous variables and frequency (%) for categorical variables.
P < 0.05 versus cluster 2.
P < 0.05 versus cluster 3.
P < 0.05 versus cluster 2 and 3.
Items were derived from the Brief Pain Inventory‐Short Form (BPI). A 0–10 numeric rating scale was anchored at 0 for “no pain” and 10 for “pain as bad as you can imagine.”
Item response on a 6‐point scale ranging from 1 for “all of the time” to 6 for “none of the time;” dimensions of sleep quantity, “get the amount of sleep you needed;” and sleep adequacy, “get enough sleep to feel rested upon waking in the morning” were calculated backwards.
Items were from three levels indicating “no problem” (or 1), “some problems” (or 2) and “severe problems” (or 3), and EuroQoL Health (EQ‐5D) index was the sum of scores of five dimensions.
Values from 0 to 10, where 0 represents the worst imaginable health state and 10 represents the best imaginable health state.
MOS, Medical Outcomes Study; MNSI, Michigan Neuropathy Screening Instrument; VAS, visual analog scale.
Factor analysis
The Kaiser–Meyer–Olkin measure of sampling adequacy was 0.909, and Bartlett's test of sphericity was significant ( χ2 = 1,773.38, P < 0.01) thus supporting the suitability of data for factor analysis. Five eigenvalues were >1, and this determined the five factors computed for all of the patients. Table 4 shows the five factors and their item loadings, with absolute values >0.4 identified. The results of factor analysis explained 78.37% of the total variance in the patients with DPN. Factor 1, capturing 18.35% of rotated variance, included number 14 (“Is the skin on your feet so dry that it cracks open?”), number 11 (“Are your symptoms worse at night?”) and number 5 (“Do you ever have any prickling feelings in your legs or feet?”). Factor 2, capturing 16.84% of rotated variance, included number 10 (“Do you feel weak all over most of the time?”) and number 13 (“Are you able to sense your feet when you walk?”). Factor 3, capturing 16.37% of rotated variance, included number 3 (“Are your feet too sensitive to touch?”), number 6 (“Does it hurt when the bed covers touch your skin?”) and number 2 (“Do you ever have any burning pain in your legs and/or feet?”). Factor 4, capturing 16.16% of rotated variance, included number 15 (“Have you ever had an amputation?”) and number 12 (“Do your legs hurt when you walk?”). Factor 5, capturing 10.66% of rotated variance, included number 1 (“Are your legs and/or feet numb?”) and number 7 (“When you get into the bath or shower, are you able to tell the hot water from the cold water?”). The Cronbach's alpha of the factors 1, 2, 3, 4 and 5 were 0.785, 0.999, 0.653, 0.965 and 0.577, respectively.
Table 4.
Factor loadings for Michigan Neuropathy Screening Instrument questionnaire items
| Variable | Factor 1 | Factor 2 | Factor 3 | Factor 4 | Factor 5 | Cronbach's alpha |
|---|---|---|---|---|---|---|
| Number 14. Is the skin on your feet so dry that it cracks open? | 0.954 † | 0.081 | 0.074 | 0.042 | 0.142 | 0.785 |
| Number 11. Are your symptoms worse at night? | 0.954 † | 0.081 | 0.074 | 0.042 | 0.142 | |
| Number 5. Do you ever have any prickling feelings in your legs or feet? | 0.456 † | 0.007 | 0.411 | 0.170 | −0.107 | |
| Number 10. Do you feel weak all over most of the time? | 0.078 | 0.984 † | 0.034 | 0.081 | 0.106 | 0.999 |
| Number 13. Are you able to sense your feet when you walk? | 0.078 | 0.984 † | 0.034 | 0.081 | 0.106 | |
| Number 3. Are your feet too sensitive to touch? | 0.015 | 0.044 | 0.801 † | 0.207 | 0.144 | 0.653 |
| Number 6. Does it hurt when the bed covers touch your skin? | −0.003 | −0.068 | 0.784 † | 0.145 | 0.231 | |
| Number 2. Do you ever have any burning pain in your legs and/or feet? | 0.308 | 0.134 | 0.639 † | 0.050 | −0.173 | |
| Number 15. Have you ever had an amputation? | 0.068 | 0.063 | 0.189 | 0.953 † | 0.088 | 0.965 |
| Number 12. Do your legs hurt when you walk? | 0.087 | 0.106 | 0.212 | 0.941 † | 0.081 | |
| Number 7. When you get into the tub or shower, are you able to tell the hot water from the cold water? | −0.027 | 0.173 | 0.188 | −0.019 | 0.763 † | 0.577 |
| Number 1. Are your legs and/or feet numb? | 0.231 | 0.029 | −0.039 | 0.177 | 0.710 † | |
| Factor's name | Painful | Insensate | Hypersensitivity | Damaged | Hypoesthesia | |
| Eigenvalue | 2.202 | 2.020 | 1.965 | 1.939 | 1.279 | |
| % Of variance explained | 18.352 | 16.836 | 16.371 | 16.155 | 10.656 | |
| % Cumulative variance | 18.352 | 35.189 | 51.560 | 67.715 | 78.370 |
Kaiser‐Meyer‐Olkin Measure of Sampling Adequacy = 0.909. Bartlett's Test of Sphericity. Chi‐Square χ2 = 1773.382 (df = 21, ***P < 0.01). Extraction Method: The principal axis factor, Rotation Method: Varimax rotation.
Factor loadings ≥0.40.
Frequency of symptom patterns in each cluster
Participants’ scores on the MNSI and painDETECT were significantly higher in cluster 1 than cluster 2 and 3 (Table 3). Figure 1 shows the frequencies of symptoms represented by results of factor analysis in the three cluster groups. Patients in cluster 1 reported high intensity on average for factor 3 and 4 dimensions (hypersensitivity, damaged), whereas patients in cluster 2 had high intensity for factor 2 and 5 dimensions (insensate, hypoesthesia). Patients in cluster 3 showed moderate intensity in all dimensions on average. Cluster 3 showed significantly more patients with reduced pain in response to symptomatic treatment in the past 24 h than other clusters (P = 0.01; Table 3). Based on cluster and factor analysis, the characteristics of the three clusters were defined as follows: cluster 1: high pain intensity, predominant positive symptoms of neuropathic pain, decreased QOL; cluster 2: predominant negative symptoms of neuropathic pain, relatively good QOL; and cluster 3: moderate pain intensity and moderate decreased QOL.
Figure 1.
Frequency (%) distribution of symptoms on the Michigan Neuropathy Screening Instrument questionnaire (MNSI) items among three clusters according to subgroups of symptom patterns. a P < 0.01. There was a significant difference across the clusters in all items.
DISCUSSION
The present study showed that patients with DPN were clustered into three groups based on subjective neuropathic symptoms, and the clinical impacts of DPN on pain, sleep and QOL. Cluster 1 had a higher proportion of women, higher FBG, longer duration of diabetes and diabetic neuropathy than the other clusters, and were associated with the highest pain intensity and lowest QOL. Cluster 2 had a higher proportion of patients with negative neuropathic symptoms and showed relatively good QOL compared with other cluster patients. Patients in cluster 3 showed moderate pain intensity and QOL, and included a higher proportion of patients with reduced pain in response to symptomatic treatment than other clusters. These findings are consistent with previous studies in a large cohort of patients with DPN, in that patients presenting heterogeneous symptoms could be classified into several subgroups according to their neuropathic symptoms 7 , 8 . Regardless of cluster classification, patients mostly complained of pain in the lower extremities.
Approximately 30–50% of patients with diabetic neuropathy develop neuropathic pain, which can present as several different patterns 9 . The reason why some patients with diabetic neuropathy develop neuropathic pain whereas others do not is not fully understood. Neuropathic pain in DPN seems to be associated with female sex 10 and increasing age 11 . Metabolic problems including obesity 12 , elevated glycated hemoglobin 13 , type and duration of diabetes, and high alcohol intake, might increase the risk of developing neuropathic pain. Sensory phenotype and severity of neuropathic symptoms are also associated with DPN 14 . The present findings show that the patient cluster that complained of the most severe pain had a higher proportion of women, higher FBG and a longer duration of diabetes. This result is mostly consistent with the factors related to the development of neuropathic pain in DPN suggested in previous studies 15 , 16 , 17 , 18 . In particular, FBG was significantly higher in cluster 1 than other clusters. In general, DPN symptoms get worse at night. It is associated with an upregulated stress response during sleep by activation of sympathetic vascular control and poor subjective sleep quality. This result could be explained by high pain intensity playing a detrimental role in contributing to elevated FBG in cluster 1 than other clusters.
Regarding foot examinations, the proportion of patients who showed abnormal findings in ankle reflex, vibration sensation and response to 10‐g monofilament in cluster 1 was higher than cluster 2 and 3. This observation is consistent with the finding of a large‐scale study carried out in the UK that showed the characteristics of painful DPN in patients assessed with the neuropathy symptom score and neuropathy disability score 15 . However, patients in cluster 2 had a significantly higher proportion of patients with foot deformities. A substantial percentage of cluster 2 patients reported numbness or diminished sensation, and such negative symptoms might make it difficult to detect foot deformity and act as a risk factor for foot ulcers 19 . Diabetic neuropathy is well known as a risk factor for diabetic foot ulcers 20 . In the present study, the proportion of patients who had diabetic foot ulcers did not show a significant difference in each cluster. However, there is a possibility that the detection of changes in foot appearance might be delayed in patients with negative symptoms, thus more detailed foot observation and foot care education are needed in such patients 21 .
The patients in cluster 1 showed higher scores than patients in cluster 2 and 3 in all items of the painDETECT score, followed by cluster 3 and 2, respectively. In addition, the BPI‐SF measurements showed a similar pattern. In the MOS Sleep Scale, cluster 1 had significantly worse sleep quality than clusters 2 and 3, whereas clusters 2 and 3 did not show a significant difference in sleep quality. According to these findings, study participants with predominantly positive symptoms and high pain intensity had higher pain interference scores, and more frequently reported sleep disturbance than participants with moderate or mild pain intensity. Patient responses to items of the MOS Sleep Scale in the present study supported the potential impact of nocturnal exacerbations due to pain on the various dimensions of sleep. EQ‐5D index showed a significant difference between each cluster, and it was demonstrated that the proportion of patients in cluster 1 who had problems with motor skills, self‐management and daily activities was higher than the patients in clusters 2 and 3. In particular, the proportion of patients complaining of anxiety or depression was significantly higher in cluster 1 compared with other clusters. Based on these findings, a detailed evaluation of psychological performance might be required for patients who complain of positive symptoms and a high level of pain intensity. In addition, this result suggests that DPN phenotype classification can be useful for the attention and appropriate treatment of specific symptoms of each subtype.
In the present study, pain intensity seems to be the most important factor in classifying DPN patients into three clusters. The frequency of distribution of each symptom on the MNSI had comparable tendencies regarding pain intensities. The patients in cluster 1 had prominent symptom intensities, and a higher number of patients with symptoms related to ‘painful’ and ‘hypersensitivity’. Such data are consistent with those of previous studies showing the heterogeneity of neuropathic pain in patients whose pain was assessed with the painDETECT 22 or the MNSI 23 , which showed that patients with varied symptoms might be classified into several phenotypes based on neuropathy‐related symptom profiles.
Although the present study was based on a large multicenter study, it had several limitations. First, as the diagnosis for DPN was not assessed with more reliable and quantitative methods, such as nerve conduction study, relatively more patients with possible or probable DPN might have been included than patients with confirmed DPN. Second, the effect of the type or dose of concurrent neuropathic medication on various neuropathic symptoms could not be evaluated. The cross‐sectional design of this study made it difficult to explore the causal relationships between various DPN symptoms and the effects of medication or associated sequelae. Despite such limitations, the results of this study suggest that sensory phenotypes of DPN might lead to more individualized and effective treatment of patients with DPN. Future studies are required to replicate the findings of the present study. Also, whether a specific treatment approach is systematically associated with the altered clinical characteristics of DPN over time is worth investigating.
In conclusion, the present cluster and factor analyses support the hypothesis that diversity in sensory symptoms exists in patients with DPN. We identified three cluster groups based on sensory profiles and effects of DPN on QOL. The identification of patient subtypes with distinct symptom characteristics at baseline is important and shows that heterogeneity in patients with DPN should be taken into account for a more stratified or individualized treatment approach.
DISCLOSURE
The authors declare no conflict of interest.
Approval of the research protocol: This study was approved by the local ethics committees of each participating hospital (CUH 2017‐06‐004).
Informed consent: All participants enrolled provided written informed consent to the procedures in accordance with the Declaration of Helsinki.
Registry and the registration no. of the study/trial: N/A.
Animal studies: N/A.
ACKNOWLEDGMENTS
We thank all the participants for taking part in the current study. This study was supported by a grant (KYJ, 2017S‐3) from the Korean Diabetes Association.
Appendix A.
The location of neuropathic pain using on schematic diagram representing the front and back views of the whole body region
| Pain location | Patient (n) | Pain location | Patient (n) | Pain location | Patient (n) | Pain location | Patient (n) |
|---|---|---|---|---|---|---|---|
| 1 | 448 | 17 | 3 | 39 | 84 | 61 | 8 |
| 2 | 448 | 18 | 1 | 40 | 85 | 62 | 4 |
| 3 | 82 | 19 | 79 | 41 | 16 | 63 | 2 |
| 4 | 92 | 20 | 72 | 42 | 16 | 64 | 2 |
| 5 | 91 | 21 | 13 | 43 | 20 | 65 | 0 |
| 6 | 88 | 22 | 8 | 44 | 20 | 66 | 14 |
| 7 | 39 | 23 | 9 | 45 | 10 | 67 | 23 |
| 8 | 34 | 24 | 3 | 46 | 7 | ||
| 9 | 23 | 25 | 3 | 47 | 5 | ||
| 10 | 21 | 26 | 2 | 48 | 5 | ||
| 11 | 0 | 27 | 2 | 49 | 3 | ||
| 12 | 0 | 28 | 1 | 50 | 1 | ||
| 13 | 2 | 29 | 15 | 51 | 39 | ||
| 14 | 3 | 30 | 10 | 52 | 36 | ||
| 15 | 0 | 31 | 2 | 53 | 4 | ||
| 16 | 0 | 32 | 2 | 54 | 4 | ||
| 17 | 3 | 33 | 0 | 55 | 3 | ||
| 18 | 1 | 34 | 2 | 56 | 2 | ||
| 19 | 79 | 35 | 262 | 57 | 1 | ||
| 20 | 72 | 36 | 261 | 58 | 1 | ||
| 21 | 13 | 37 | 53 | 59 | 0 | ||
| 22 | 8 | 38 | 55 | 60 | 0 |
Appendix B.
The proportion of medication use for diabetic peripheral neuropathy in the entire patients
| Medication | Total (n = 498) |
|---|---|
| Antidepressants | 58 (11.6) |
| Anticonvulsants | 330 (66.3) |
| Alpha‐lipoic acid | 264 (53.0) |
| Gamma‐linoleic acid | 149 (30.0) |
| Opioids | 15 (3.0) |
| None | 54 (10.8) |
Data are expressed as frequency (%).
Contributor Information
Jong Chul Won, Email: drwonjc@gmail.com.
Tae Sun Park, Email: pts@jbnu.ac.kr.
REFERENCES
- 1. Won JC, Kwon HS, Kim CH, et al. Prevalence and clinical characteristics of diabetic peripheral neuropathy in hospital patients with type 2 diabetes in Korea. Diabet Med 2012; 29: e290–e296. [DOI] [PubMed] [Google Scholar]
- 2. Kim SS, Won JC, Kwon HS, et al. Prevalence and clinical implications of painful diabetic peripheral neuropathy in type 2 diabetes: results from a nationwide hospital‐based study of diabetic neuropathy in Korea. Diabetes Res Clin Pract 2014; 103: 522–529. [DOI] [PubMed] [Google Scholar]
- 3. Rio J, Bittencourt JV, Corrêa LA, et al. Cross‐cultural adaptation of the painDETECT questionnaire into Brazilian Portuguese. Braz J Anesthesiol (English Edition) 2022; 72: 44–48. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Feldman EL, Stevens MJ, Thomas PK, et al. A practical two‐step quantitative clinical and electrophysiological assessment for the diagnosis and staging of diabetic neuropathy. Diabetes Care 1994; 17: 1281–1289. [DOI] [PubMed] [Google Scholar]
- 5. Tesfaye S, Boulton AJ, Dyck PJ, et al. Diabetic neuropathies: update on definitions, diagnostic criteria, estimation of severity, and treatments. Diabetes Care 2010; 33: 2285–2293. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Ball GH, Hall DJ. A clustering technique for summarizing multivariate data. Behav Sci 1967; 12: 153–155. [DOI] [PubMed] [Google Scholar]
- 7. Freynhagen R, Baron R, Gockel U, et al. painDETECT: a new screening questionnaire to identify neuropathic components in patients with back pain. Curr Med Res Opin 2006; 22: 1911–1920. [DOI] [PubMed] [Google Scholar]
- 8. Von Hehn CA, Baron R, Woolf CJ. Deconstructing the neuropathic pain phenotype to reveal neural mechanisms. Neuron 2012; 73: 638–652. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Rosenberger DC, Blechschmidt V, Timmerman H, et al. Challenges of neuropathic pain: focus on diabetic neuropathy. J Neural Transm (Vienna) 2020; 127: 589–624. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Truini A, Spallone V, Morganti R, et al. A cross‐sectional study investigating frequency and features of definitely diagnosed diabetic painful polyneuropathy. Pain 2018; 159: 2658–2666. [DOI] [PubMed] [Google Scholar]
- 11. Van Acker K, Bouhassira D, De Bacquer D, et al. Prevalence and impact on quality of life of peripheral neuropathy with or without neuropathic pain in type 1 and type 2 diabetic patients attending hospital outpatients clinics. Diabetes Metab 2009; 35: 206–213. [DOI] [PubMed] [Google Scholar]
- 12. Ziegler D, Landgraf R, Lobmann R, et al. Painful and painless neuropathies are distinct and largely undiagnosed entities in subjects participating in an educational initiative (PROTECT study). Diabetes Res Clin Pract 2018; 139: 147–154. [DOI] [PubMed] [Google Scholar]
- 13. Themistocleous AC, Ramirez JD, Shillo PR, et al. The pain in neuropathy study (PiNS): a cross‐sectional observational study determining the somatosensory phenotype of painful and painless diabetic neuropathy. Pain 2016; 157: 1132–1145. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Raputova J, Srotova I, Vlckova E, et al. Sensory phenotype and risk factors for painful diabetic neuropathy: a cross‐sectional observational study. Pain 2017; 158: 2340–2353. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Abbott CA, Malik RA, van Ross ER, et al. Prevalence and characteristics of painful diabetic neuropathy in a large community‐based diabetic population in the U.K. Diabetes Care 2011; 34: 2220–2224. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. Callaghan BC, Cheng HT, Stables CL, et al. Diabetic neuropathy: clinical manifestations and current treatments. Lancet Neurol 2012; 11: 521–534. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Dermanovic Dobrota V, Hrabac P, Skegro D, et al. The impact of neuropathic pain and other comorbidities on the quality of life in patients with diabetes. Health Qual Life Outcomes 2014; 12: 171. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. England JD, Gronseth GS, Franklin G, et al. Distal symmetric polyneuropathy: a definition for clinical research: report of the American Academy of Neurology, the American Association of Electrodiagnostic Medicine, and the American Academy of physical medicine and rehabilitation. Neurology 2005; 64: 199–207. [DOI] [PubMed] [Google Scholar]
- 19. Monteiro‐Soares M, Vaz‐Carneiro A, Sampaio S, et al. Validation and comparison of currently available stratification systems for patients with diabetes by risk of foot ulcer development. Eur J Endocrinol 2012; 167: 401–407. [DOI] [PubMed] [Google Scholar]
- 20. Hicks CW, Selvin E. Epidemiology of peripheral neuropathy and lower extremity disease in diabetes. Curr Diab Rep 2019; 19: 86. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21. Spallone V, Morganti R, D'Amato C, et al. Clinical correlates of painful diabetic neuropathy and relationship of neuropathic pain with sensorimotor and autonomic nerve function. Eur J Pain 2011; 15: 153–160. [DOI] [PubMed] [Google Scholar]
- 22. Baron R, Tölle TR, Gockel U, et al. A cross‐sectional cohort survey in 2100 patients with painful diabetic neuropathy and postherpetic neuralgia: differences in demographic data and sensory symptoms. Pain 2009; 146: 34–40. [DOI] [PubMed] [Google Scholar]
- 23. Won JC, Im YJ, Lee JH, et al. Clinical phenotype of diabetic peripheral neuropathy and relation to symptom patterns: cluster and factor analysis in patients with type 2 diabetes in Korea. J Diabetes Res 2017; 2017: 5751687. [DOI] [PMC free article] [PubMed] [Google Scholar]