Customize your JAMA Network experience by selecting one or more topics from the list below.
Callaghan BC, Xia R, Reynolds E, et al. Association Between Metabolic Syndrome Components and Polyneuropathy in an Obese Population. JAMA Neurol. 2016;73(12):1468–1476. doi:10.1001/jamaneurol.2016.3745
Copyright 2016 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.
What are the causes of polyneuropathy in an obese population?
This cross-sectional study found that the prevalence of polyneuropathy is high even in individuals with obesity and normal glucose levels. Prediabetes and obesity are likely important metabolic drivers of polyneuropathy, while diabetes remains the biggest risk factor.
Physicians need to be aware of the potential role of prediabetes and obesity in the development and progression of polyneuropathy.
Past studies have shown an association between metabolic syndrome and polyneuropathy, but the precise components that drive this association remain unclear.
To determine the prevalence of polyneuropathy stratified by glycemic status in well-characterized obese and lean participants and investigate the association of specific components of metabolic syndrome with polyneuropathy.
Design, Setting, and Participants
We performed a cross-sectional, observational study from November 1, 2010, to December 31, 2014, in obese participants (body mass index [calculated as weight in kilograms divided by height in meters squared] of 35 or more with no comorbid conditions or 32 or more with at least 1 comorbid condition) from a weight management program and lean controls from a research website. The prevalence of neuropathy, stratified by glycemic status, was determined, and a Mantel–Haenszel χ2 test was used to investigate for a trend. Logistic regression was used to model the primary outcome of polyneuropathy as a function of the components of metabolic syndrome after adjusting for demographic factors. Participants also completed quantitative sudomotor axon reflex testing, quantitative sensory testing, the neuropathy-specific Quality of Life in Neurological Disorders instrument, and the short-form McGill Pain Questionnaire.
Components of metabolic syndrome (as defined by the National Cholesterol Education Program Adult Treatment Panel III), including glycemic status (as defined by the Expert Committee on the Diagnosis and Classification of Diabetes Mellitus).
Main Outcomes and Measures
Toronto consensus definition of probable polyneuropathy. Secondary outcomes included intraepidermal nerve fiber density and nerve conduction study parameters.
We enrolled 102 obese participants (mean [SD] age, 52.9 [10.2] years; 48 men and 54 women; 45 with normoglycemia [44.1%], 31 with prediabetes [30.4%], and 26 with type 2 diabetes [25.5%]) and 53 lean controls (mean [SD] age, 48.5 [9.9] years; 16 men and 37 women). The prevalence of polyneuropathy was 3.8% in lean controls (n = 2), 11.1% in the obese participants with normoglycemia (n = 5), 29% in the obese participants with prediabetes (n = 9), and 34.6% in the obese participants with diabetes (n = 9) (P < .01 for trend). Age (odds ratio, 1.09; 95% CI, 1.02-1.16), diabetes (odds ratio, 4.90; 95% CI, 1.06-22.63), and waist circumference (odds ratio, 1.24; 95% CI, 1.00-1.55) were significantly associated with neuropathy in multivariable models. Prediabetes (odds ratio, 3.82; 95% CI, 0.95-15.41) was not significantly associated with neuropathy.
Conclusions and Relevance
The prevalence of polyneuropathy is high in obese individuals, even those with normoglycemia. Diabetes, prediabetes, and obesity are the likely metabolic drivers of this neuropathy.
clinicaltrials.gov Identifier: NCT02689661.
Polyneuropathy is a prevalent, disabling condition affecting 2% to 7% of the population1,2; however, a definitive cause is lacking in 30% of patients.3,4 The most common etiologic condition is diabetes, with a similar prevalence in patients with types 1 and 2 diabetes.5-9 In patients with type 1 diabetes, enhanced glucose control reduces the incidence of polyneuropathy substantially, but in those with type 2 diabetes, the effect does not reach statistical significance.10 Therefore, factors other than glucose control are likely to be involved in the development of type 2 diabetic polyneuropathy.11 These same factors may also explain why patients are diagnosed with idiopathic polyneuropathy. Components of metabolic syndrome are potential aspects since these cardiovascular risk factors cluster with hyperglycemia.
Multiple studies reveal an association between metabolic syndrome and polyneuropathy.12-15 However, investigations of the precise association between individual components of metabolic syndrome and polyneuropathy demonstrate conflicting results.7,16-22 Limitations of these studies include definitions of polyneuropathy based on surrogate tests rather than results of a patient’s neurologic examination and history using rigorous criteria, as well as a focus primarily on patients with diabetes.
We aimed to determine the prevalence of polyneuropathy stratified by glycemic status in obese and lean populations with detailed metabolic and neuropathy phenotyping. We also examined the associations between the individual components of metabolic syndrome and a clinical definition of polyneuropathy (primary outcome) as well as intraepidermal nerve fiber density (IENFD) and nerve conduction study parameters (secondary outcomes).
From November 1, 2010, to December 31, 2014, we recruited obese patients attending the University of Michigan Weight Management Program (before starting a diet and exercise regimen). Inclusion criteria included age 18 years or older and a body mass index (BMI; calculated as weight in kilograms divided by height in meters squared) of 35 or more with no comorbid conditions or 32 or more if they had at least 1 comorbid condition.23 We recruited lean controls with no components of metabolic syndrome (BMI<28 was required rather than a waist circumference cutoff) through a University of Michigan research website (umclinicalstudies.org). Lean controls were excluded if they were taking medications for blood pressure, cholesterol, diabetes, or triglycerides. This study was approved by the University of Michigan Institutional Review Board. All participants provided written informed consent.
Obese and lean patients (except those with a previous diagnosis of diabetes) underwent glucose tolerance testing, and all patients underwent a fasting lipid panel. Patients also had blood pressure, height, weight, waist circumference, and BMI measurements performed.
Participants were classified as having normoglycemia, prediabetes, and type 2 diabetes according to the Expert Committee on the Diagnosis and Classification of Diabetes Mellitus.24 The updated National Cholesterol Education Program Adult Treatment Panel III criteria were used to define metabolic syndrome and its individual components.25
Our primary outcome measure was the Toronto consensus definition of probable polyneuropathy (≥2 of the following: neuropathy symptoms, abnormal results of sensory examination, and abnormal reflexes) as determined by one of 4 neuromuscular specialists (B.C.C. and E.L.F.).26
Our secondary outcome measures were IENFD measured at the distal leg and 3 nerve conduction study parameters (sural, tibial, and ulnar sensory amplitudes). Intraepidermal nerve fiber density was evaluated via bright field immunohistochemistry using an established protocol.27 Nerve conduction studies were performed by a certified nerve conduction study technologist using the Viking (CareFusion) on EDX electrodiagnostic system (Nicolet).
To further characterize peripheral nerve function, we obtained the IENFD at the proximal thigh, and other nerve conduction study parameters, including the sural (peak latency), peroneal (amplitude, distal motor latency, conduction velocity, and F response), tibial (distal motor latency and F response), ulnar sensory (peak latency), median sensory (amplitude and peak latency), and median motor responses (amplitude, distal motor latency, and conduction velocity). Quantitative sudomotor axon reflex testing measurements were performed at the foot, distal leg, proximal leg, and arm using the Q-Sweat Quantitative Sweat Measurement System (WR Medical Electronics Co). Quantitative sensory testing measurements of vibration and cold detection thresholds were performed using the Computer Aided Sensory Evaluator IV (WR Medical Electronics Co). The validated, neuropathy-specific Quality of Life in Neurological Disorders (Neuro-QOL) instrument was used to measure neuropathy-specific quality of life, with higher numbers reflecting a poorer quality of life.28 The validated short-form McGill Pain Questionnaire was used to measure pain with a visual analog scale, a 6-point rating scale of present pain intensity, and a 4-point rating scale of 15 different neuropathic pain descriptors (McGill pain score).29
Descriptive statistics were used to characterize the obese and lean populations in terms of demographics and neuropathic outcome measures. Fisher exact tests or χ2 tests were used to compare the 2 populations in terms of categorical variables, and t tests were used for continuous variables.
We determined the prevalence of polyneuropathy stratified by glycemic status. We then applied the Mantel-Haenszel χ2 test to investigate for a trend in the prevalence of polyneuropathy in the 4 groups based on obesity and glycemic status.
For the primary analysis, univariable and multivariable logistic regression were used to model the primary outcome of polyneuropathy as a function of the components of metabolic syndrome after adjusting for demographic factors. Sensitivity analyses were performed using symptomatic polyneuropathy as the primary outcome after excluding patients with conditions known to be associated with neuropathy (n = 9), a first-degree relative with idiopathic neuropathy (n = 3), or those without results of a glucose tolerance test. For the secondary outcomes, we fitted multivariable linear regression to analyze them as a function of the components of metabolic syndrome, adjusting for the same demographic factors. Regression analyses were restricted to the obese population with complete metabolic data. Lean controls were not included in these analyses. All analyses were performed with SAS, version 9.3 (SAS Institute Inc).
During the recruitment period, the University of Michigan Weight Management Program enrolled 532 patients. Of these patients, 394 consented to be contacted about research studies; 102 enrolled in our study. Of 79 lean control participants who consented to participate, 53 met all inclusion and exclusion criteria after metabolic phenotyping and participated in the study. Screening failures among the lean control participants consisted of 6 individuals with elevated blood pressure, 6 with elevated fasting or 2-hour glucose levels, 5 with elevated high-density lipoprotein (HDL) cholesterol or triglyceride levels, and 2 with a BMI greater than 28. Two participants failed to meet multiple criteria. In addition, 4 patients did not complete testing for neuropathy, and 5 withdrew consent.
The obese group was older (mean [SD] age, 52.9 [10.2] vs 48.5 [9.9] years; P = .01), and differences were observed between the obese and lean populations in terms of race and marital status as well as metabolic variables (Table 1). Ten obese patients did not receive a glucose tolerance test, but all had a fasting glucose level and/or a hemoblogin A1C level to allow assessment of glycemic status. Similarly, 4 obese patients did not undergo a lipid panel, and 1 did not have a waist circumference measurement.
The prevalence of polyneuropathy was 3.8% (2 of 53) in the lean control group, 11.1% (5 of 45) in the obese participants with normoglycemia, 29% (9 of 31) in the obese participants with prediabetes, and 34.6% (9 of 26) in the obese participants with diabetes (P < .01 for trend).
Comparing neuropathy measures between the obese population with and without neuropathy revealed that the mean (SD) IENFD at the leg (2.2 [2.3] vs 4.5 [2.2] fibers per millimeter; P < .01) and thigh (7.2 [4.9] vs 9.5 [4.1] fibers per millimeter; P = .03) were reduced in those with neuropathy (Table 2). Similarly, all 3 sensory amplitudes were reduced in those with neuropathy, but no significant changes were observed for the corresponding peak latencies. Six of the 7 lower extremity motor nerve conduction study parameters were worse in those with neuropathy, with the exception of the tibial distal motor latency. In contrast, the only upper extremity motor nerve conduction study parameter that was different between the groups was the median motor conduction velocity. Quantitative sensory testing for vibration (mean [SD] just-noticeable difference, 21.9 [3.3] vs 19.0 [4.3]; P < .01) and cold (just-noticeable difference, 16.6 [4.5] vs 13.8 [4.1]; P < .01) demonstrated higher thresholds in those with neuropathy than those without. In contrast, none of the 4 quantitative sudomotor axon reflex testing parameters were significantly different between the groups.
The total mean (SD) Neuro-QOL score was higher, indicating a poorer quality of life, in the obese population with neuropathy compared with those without neuropathy (3.4 [2.7] vs 1.9 [1.3]; P = .02) (Table 3). All 5 mean (SD) subdomain scores were also higher in the group with neuropathy, but only the pain (3.5 [2.3] vs 1.9 [1.2]; P < .01) and reduced sensation (4.0 [4.0] vs 1.4 [0.9]; P < .01) scores were statistically significant. The mean (SD) McGill pain score (8.2 [8.6] vs 2.8 [3.3]; P < .01) and mean (SD) visual analog scale score (27.2 [26.9] vs 14.8 [20.1] mm; P = .02) were higher in those with neuropathy. No significant difference was observed between groups in the present pain intensity score.
Comparing neuropathy measures between the obese participants without neuropathy and lean controls demonstrated that mean (SD) IENFD at the leg (4.5 [2.2] vs 6.4 [3.8] fibers per millimeter; P < .01) and thigh (9.5 [4.1] vs 12.3 [6.5] fibers per millimeter; P < .01) were reduced in the obese participants without neuropathy (Table 2). In contrast, the only lower extremity nerve conduction study parameter that was significantly worse in the obese participants without neuropathy was mean (SD) tibial amplitude (9.4 [4.7] vs 11.1 [4.8] mV; P = .046). However, mean (SD) peroneal amplitude was higher (6.0 [2.7] vs 5.0 [2.4] mV; P = .04) and the mean (SD) peroneal distal motor latency was shorter (4.7 [0.6] vs 5.0 [0.8] milliseconds; P < .01) in the obese participants without neuropathy compared with the lean controls. The total mean (SD) Neuro-QOL score (1.9 [1.3] vs 1.4 [0.4]; P < .01), McGill pain score (2.8 [3.3] vs 1.0 [1.9]; P < .01), and visual analog scale score (14.8 [20.1] vs 6.0 [13.3] mm; P < .01) were higher in the obese participants without neuropathy compared with the lean controls (Table 3). None of the 4 quantitative sudomotor axon reflex testing parameters or 2 quantitative sensory testing thresholds was significantly different between these 2 groups.
In a univariable logistic regression model investigating the individual components of metabolic syndrome, age (odds ratio [OR], 1.08; 95% CI, 1.02-1.14) and diabetes (OR, 4.24; 95% CI, 1.24-14.51) were significantly associated with the primary outcome (Table 4). Prediabetes (OR, 3.27; 95% CI, 0.98-10.98), waist circumference (OR, 1.11; 95% CI, 0.97-1.26), and height (OR, 1.23; 95% CI, 0.95-1.58) were not significantly associated with the primary outcome. Based on multivariable logistic regression, age (OR, 1.09; 95% CI, 1.02-1.16), diabetes (OR, 4.90; 95% CI, 1.06-22.63), and waist circumference (OR, 1.24; 95% CI, 1.00-1.55) were significantly associated with neuropathy. Systolic blood pressure, triglyceride level, and HDL cholesterol level were not associated with neuropathy. In the sensitivity analysis from which 11 patients were removed because of a history of conditions known to be associated with neuropathy or a first-degree relative with neuropathy of unknown cause, the ORs for diabetes and prediabetes increased to 7.37 (95% CI, 1.16-46.86) and 6.29 (95% CI, 1.21-32.78), respectively. Sensitivity analyses using the primary outcome of symptomatic polyneuropathy or excluding those patients yielded similar results.
When investigating the association of the components of metabolic syndrome and the 4 secondary neuropathy outcomes in multivariable models, diabetes was associated with a significant reduction in IENFD (parameter estimate, –1.39; 95% CI, –2.61 to –0.17) and nonsignificant reductions in the 3 nerve conduction study parameters (Table 5). Waist circumference was associated with a significant reduction in the tibial amplitude (parameter estimate, –0.51; 95% CI, –0.92 to –0.10) and nonsignificant reductions in the sural and ulnar sensory amplitudes. Higher HDL cholesterol levels were associated with significant reductions in the IENFD at the leg (parameter estimate, –0.56; 95% CI, –1.03 to –0.10) and sural amplitude (parameter estimate, –1.97; 95% CI, –3.66 to –0.28) and nonsignificant reductions in the tibial and ulnar sensory amplitudes. Prediabetes, systolic blood pressure, and triglyceride levels were not associated with the secondary outcomes.
In obese and lean control populations who received comprehensive metabolic and neuropathy phenotyping, we found a higher prevalence of neuropathy in obese patients with normoglycemia compared with lean controls. The prevalence of neuropathy continued to increase in obese patients with prediabetes and diabetes. Diabetes, waist circumference, and likely prediabetes were the main metabolic factors associated with neuropathy. In contrast, systolic blood pressure, triglyceride levels, and HDL cholesterol levels were not associated with neuropathy. Future intervention studies are needed to confirm a causal association between these metabolic factors and neuropathy.
Diabetes is a well-established risk factor for neuropathy, and our data support diabetes as the largest risk factor for polyneuropathy.10 However, previous studies have shown conflicting results regarding prediabetes and neuropathy. Two separate groups have shown a higher prevalence of prediabetes in patients with idiopathic neuropathy compared with literature-based controls.18,19 In addition, patients with impaired glucose tolerance and neuropathy had an increase in IENFD after an extensive diet and exercise intervention (no control group), which is in contrast to historical controls.30 Furthermore, 3 independent population-based studies (Monitoring Trends and Determinants on Cardiovascular Diseases/Cooperative Research in the Region of Augsburg [MONICA/KORA], the San Luis Valley Diabetes study, and Prospective Metabolism and Islet Cell Evaluation [PROMISE]) demonstrated a higher prevalence of neuropathy in patients with impaired glucose tolerance compared with normoglycemic individuals.7,31,32 In contrast, Hughes et al33 did not find a significant association between impaired glucose tolerance and neuropathy in a case-control study. Similarly, Dyck et al34 found no difference in the prevalence of neuropathy in patients with impaired glucose tolerance compared with matched controls in a population-based study in Olmsted County. Two other groups also failed to find a higher prevalence of neuropathy in those with prediabetes compared with those without.35,36 However, only 1 of the population-based studies used a rigorous definition of polyneuropathy that incorporated results of the neurologic examination, and this study required abnormalities in results of nerve conduction studies to meet the definition.34 Since results of nerve conduction studies are often normal in small fiber–predominant polyneuropathies, this definition may have improperly categorized those with early polyneuropathy. Furthermore, not all studies required the oral glucose tolerance test to classify glycemic status.36 Our study, which used a definition of neuropathy based on a neurologic examination and history, lends further support to prediabetes as a cause of polyneuropathy. The neuropathy prevalence of 29% among the obese participants with prediabetes approached that seen in the obese participants with diabetes. Patients with prediabetes also had a large increased odds of polyneuropathy in univariable and multivariable logistic regression models. As a result, our data indicate that prediabetes is a likely cause of polyneuropathy; therefore, testing for this common condition should be performed in those with a new diagnosis of polyneuropathy of unknown cause.37 Whether treatment of prediabetes improves or prevents neuropathy remains to be determined.
In addition to diabetes and prediabetes, the other metabolic component associated with polyneuropathy was obesity. Most previous studies investigating obesity and polyneuropathy have also demonstrated a significant association. Specifically, 3 of 4 cross-sectional studies and 1 longitudinal study support this association.17,20-22,35 In contrast to these studies, our study did not reveal a significant association between polyneuropathy and other metabolic components, such as hypertension, HDL cholesterol, and triglycerides. One important difference in our study is that our population consisted of severely obese patients with a mean BMI of 41.1. Furthermore, our study is the first, to our knowledge, that did not focus solely on patients with diabetes and used a rigorous definition of polyneuropathy, including a neurologic examination. Our data support obesity, more than other metabolic factors, as one of the main metabolic drivers of polyneuropathy in addition to hyperglycemia and provide support for targeting this component in intervention trials designed to prevent or improve polyneuropathy. Specifically, weight loss interventions may be more likely to be successful at preventing or improving polyneuropathy than efforts to improve hyperlipidemia and hypertension.
Not only is the prevalence of neuropathy high in the obese population but neuropathy also results in a significant effect on patient-oriented outcomes, such as neuropathy-specific quality of life and pain. Although many previous studies have demonstrated an effect of neuropathy on these 2 important domains,38-41 our data show that the neuropathy observed in an obese population has a similar effect even when including patients with normoglycemia and prediabetes. We also observed that obese participants with neuropathy had abnormalities on other neuropathy outcome measures such as IENFD, nerve conduction studies, and quantitative sensory testing, which gives further evidence to support the high prevalence of significant neuropathy in those with obesity.
Our data also suggest that a significant proportion of obese participants without clinical neuropathy, based on results of a neurologic examination and history, likely have asymptomatic neuropathy that is small fiber predominant. This finding is supported by the lower IENFD measurements (small fibers) in this population compared with lean controls without significant changes in nerve conduction studies (large fibers). Another possible explanation is that obese participants have lower IENFD values based solely on larger skin areas. However, this explanation would not account for the worse neuropathy-specific quality of life measures and the higher visual analog scale and McGill pain scores observed in the obese participants without neuropathy compared with lean controls. Further evidence against obesity itself reducing IENFD is that the published normative data did not reveal a BMI effect.42 Previous studies have shown that the neuropathy associated with prediabetes and early diabetes is a small fiber–predominant neuropathy.43,44 Our results support a similar small fiber–predominant neuropathy early in the course of metabolic neuropathy.
Limitations of our study include a small sample size, which limits our ability to detect small effects of the components of metabolic syndrome. However, we did find statistically significant associations between components of metabolic syndrome and neuropathy. We recruited obese participants attending a weight management program at a tertiary referral center, less than 20% of these patients agreed to participate, and the population was primarily non-Hispanic white. Therefore, how these results generalize to the entire clinic population and other populations is unknown. Lean controls were recruited primarily from a university research website, which may have introduced selection bias. The age difference between obese and lean participants is a potential confounder.
Obese patients have a higher prevalence of neuropathy than do lean controls, even in those without diabetes or prediabetes. Furthermore, the prevalence of neuropathy in those with prediabetes is only slightly lower than the prevalence in those with diabetes. The neuropathy in this population is associated with lower quality of life and higher pain scores, indicating that the neuropathy is clinically important. Current clinical practice concentrates on the management of diabetes in those with neuropathy. However, our data also support management of obesity and prediabetes more than other metabolic factors, such as hyperlipidemia and hypertension. Future studies are needed to determine the best interventions to prevent and improve neuropathy in the obese population.
Accepted for Publication: July 28, 2016.
Corresponding Author: Brian C. Callaghan, MD, MS, Department of Neurology, University of Michigan, 109 Zina Pitcher Pl, 4021 Biomedical Science Research Bldg, Ann Arbor, MI 48104 (email@example.com).
Published Online: October 31, 2016. doi:10.1001/jamaneurol.2016.3745
Author Contributions: Dr Callaghan had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Callaghan, Xia, Burant, Villegas-Umana, Feldman.
Acquisition, analysis, or interpretation of data: Callaghan, Xia, Reynolds, Banerjee, Rothberg, Burant, Villegas-Umana, Pop-Busui.
Drafting of the manuscript: Callaghan.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Callaghan, Xia, Reynolds, Banerjee.
Obtained funding: Callaghan, Burant, Feldman.
Administrative, technical, or material support: Rothberg, Burant, Villegas-Umana.
Study supervision: Callaghan, Banerjee, Burant, Pop-Busui, Feldman.
Conflict of Interest Disclosures: Dr Callaghan reported receiving research support from Impeto Medical Inc, performing medical consultations for Advance Medical, performing medical legal consultations, and consulting for a Patient-Centered Outcomes Research Institute grant. Dr Rothberg reported involvement in a research trial for Optifast (Nestle Inc). Dr Pop-Busui reported receiving research support from Impeto Medical Inc. No other disclosures were reported.
Funding/Support: Drs Callaghan, Burant, and Feldman are supported by the Taubman Medical Institute. Dr Callaghan is supported by K23 grant NS079417 from the National Institutes of Health. Drs Burant, Pop-Busui, and Feldman are supported by grant DP3DK094292 from the National Institutes of Health. Dr Pop-Busui is supported by grants R01DK-107956 and R01HL102334 from the National Institutes of Health. Dr Feldman is supported by grant R24 082841 from the National Institutes of Health. This study was supported by grant UL1TR000433 from the Michigan Institute for Clinical & Health Research.
Role of the Funder/Sponsor: The funding sources had no role in the design and conduct of the study; collection, management, analysis, or interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Create a personal account or sign in to: