Prevalence of Polyneuropathy in Pre-Diabetes and Diabetes Is Associated With Abdominal Obesity and Macroangiopathy

RESEARCH DESIGN AND METHODS

The independent population-based MONICA (Monitoring Trends and Determinants on Cardiovascular Diseases)/KORA (Cooperative Research in the Region of Augsburg) Augsburg surveys were part of the multinational World Health Organization MONICA project (35). The second MONICA Augsburg Survey 1989/1990 (S2) included 4,940 people (participation 76.9%), while the third MONICA Augsburg Survey 1994/1995 (S3) included 4,856 people (participation: 74.9%) aged 25–74 years. The surveys were approved by the local authorities. All participants gave written informed consent. Subjects were classified as having diabetes if they reported a diagnosis of diabetes or if they were taking antidiabetes medication. All diabetic subjects from the S2 and S3 surveys as well as nondiabetic subjects matched (1:1) for age and sex were invited again in March 1997 and assessed until July 1998 for the presence of chronic diabetes complications including polyneuropathy. Included in the present study were cases defined as those who were invited as diabetic and confirmed as being diabetic based on self-reports (n = 201). Among the diabetic subjects, six were excluded due to an incomplete dataset, leaving 195 patients in the final analysis. An oral glucose tolerance test (OGTT) was performed in those who had been invited as nondiabetic control subjects. Age- and sex-matched control subjects were defined as those who were invited as nondiabetic and confirmed in the OGTT as nondiabetic (n = 198). Thus, the entire population studied included 393 subjects, of whom 185 originated from S2 and 208 from S3. In the nondiabetic group, 81 individuals had NGT, 71 had impaired fasting glucose (IFG), and 46 had IGT. Excluded were those who were invited as diabetic but self-confirmed as control subjects (n = 16) as well as those invited as control subjects but confirmed as new diabetic (n = 23) in the OGTT.

Blood pressure, body height, and body weight were determined by trained medical staff (mainly nurses). All measurement procedures have been described in detail elsewhere (36–38). Information concerning sociodemographic variables and cardiovascular risk factors was assessed by standardized personal interviews. BMI was calculated as weight in kilograms divided by the square of height in meters. A regular smoker was defined as a subject who regularly smoked at least one cigarette per day. Alcohol consumption on the previous workday and during the previous weekend was calculated in grams per day. High alcohol intake was defined as ≥40 g/day in men and ≥20 g/day in women. The physical activity level was estimated by means of two separate four-category interview questions asking about the time per week spent on sports activities during leisure time in summer and winter. The winter and summer responses were combined to define one sport variable, whereby a participant was considered physically active if he or she participated in sports in summer and in winter for more than 1 h/week in at least one season. A participant was classified as inactive if he or she was less active during leisure time. Prevalent cardiovascular disease was defined as the need for hospital treatment for myocardial infarction or stroke (38). Total serum cholesterol and HDL and LDL cholesterol levels were measured by enzymatic methods (CHOD-PAP; Boehringer, Mannheim, Germany). Serum creatinine was measured by the para-aminophenazone (PAP) method (Boehringer). Urinary albumin (in milligrams per liter) was determined in a random morning urine sample using an immunoturbidimetric test (Tina-quant; Boehringer).

OGTTs were carried out in the morning (7:00 a.m. to 11:00 a.m.) according to the World Health Organization protocol as previously described (39). Participants were asked to fast for 10 h overnight, to avoid heavy physical activity on the day before examination, and to refrain from smoking before and during the test. Fasting venous blood glucose was sampled, and 75 g anhydrous glucose was given (Dextro OGT; Boehringer). IFG was defined using a cut point for plasma glucose of 100–125 mg/dl according to American Diabetes Association criteria (40).

The presence or absence of polyneuropathy was determined by the Michigan Neuropathy Screening Instrument (MNSI) using a score cut point >2, as previously described (41). The clinical examination portion of this tool takes into account the inspection of the feet (deformities, dry skin, callus, infection), presence or absence of foot ulceration, ankle reflexes, and vibration perception threshold at the great toe, which was measured by the calibrated Rydel Seiffer tuning fork. In addition, the MNSI questionnaire consisting of 15 questions addressing positive symptoms of polyneuropathy was used.

Peripheral arterial disease (PAD) was assessed using a Mini Dopplex device (HNE Healthcare, Hilden, Germany) and defined by a ankle brachial index <0.9. This cut point has a sensitivity of 95% for the presence of PAD documented by angiography (42).

Statistical analysis

Continuous data were expressed as the mean ± SD or geometric mean ×/÷ standard deviation factor (SDF). For continuous variables satisfying a normal distribution assumption, an ANOVA (F test) for the comparison of the four groups was performed. For log-normal variables, the ANOVA was carried out on the log scale. Binomial proportions were compared using Fisher's exact test. The polyneuropathy score was analyzed nonparametrically by performing the Kruskal-Wallis test. Associations between variables were analyzed both in the entire population studied and in the diabetic group using a stepwise procedure with MNSI >2 as the dependent variable: 1) univariate logistic regression models where age, sex, height, weight, BMI, waist and hip circumference, systolic blood pressure, smoking, physical activity, alcohol consumption, creatinine, albuminuria, myocardial infarction, stroke, PAD, total cholesterol, LDL cholesterol, HDL cholesterol, triglycerides, IGT, diabetes, duration of diabetes, A1C, fasting blood glucose, and 2-h blood glucose in the OGTT were used as independent variables; 2) multivariate logistic regression models; 3) stepwise and backward regression models; and 4) final multivariate logistic regression models including age, sex, height, weight, waist circumference, physical activity, creatinine, albuminuria, duration of diabetes, A1C, IGT, diabetes, and PAD. The level of significance was set at α = 0.05. The SAS statistical package version 8.2 TS2M0 was used for all analyses.

RESULTS

The demographic variables of the subjects with NGT, IFG, IGT, and diabetes are shown in Table 1. There was a significant and steady increase in the sequence from NGT to IFG, IGT, and diabetes in BMI, waist circumference, systolic blood pressure, A1C, albuminuria, and the prevalence of polyneuropathy, whereas HDL cholesterol showed a corresponding decrease (all P < 0.05). Moreover, significant differences between the four groups studied were noted for age, height, the proportions of smokers, persons with PAD, and absent ankle reflexes and in those with high alcohol consumption and low physical activity (all P < 0.05). Fasting and 2-h glucose in the OGTT were significantly different between those with NGT, IFG, and IGT (P < 0.05). No significant differences between the groups were observed for sex, LDL cholesterol, creatinine, and the proportion of individuals with stroke, burning pain, allodynia, and foot ulcers in the lower limbs.

Among the diabetic subjects, 12 and 135 had type 1 and type 2 diabetes, respectively, and 6 had secondary diabetes, and in 42 subjects the diabetes type was not known. Diabetes treatment included oral antidiabetic agents in 86 (44.1%), insulin in 44 (22.6%), oral antidiabetic agents and insulin in 42 (21.5%), and diet only in 23 subjects (11.8%). Cardiovascular medications across the four groups studied included ACE inhibitors in a mean 6.6% (range 4.7–8.0), β-blocking agents in 4.6% (2.1–9.1), calcium channel blockers in 5.2% (3.9–5.8), diuretics in 5.6% (5.2–6.3), and lipid-lowering drugs in 2.9% (2.6–3.1) of the subjects.

According to the above definition, the prevalence (95% CI) of polyneuropathy was 28.0% (21.5–34.5) in the diabetic subjects, 11.3% (5.0–31.0) in those with IFG, 13.0% (4.9–26.3) in those with IGT, and 7.4% (2.8–15.4) in those with NGT. The percentage differences (95% CI) in prevalence adjusted for multiplicity were: diabetes minus IGT, 15% (0–30); diabetes minus IFG, 17% (4–29); diabetes minus NGT, 21% (9–32); IGT minus IFG, 2% (−14 to 18); IGT minus NGT, 6% (−9 to 20); and IFG minus NGT, 4% (−8 to 16).

In the univariate regression models including the entire population studied, significant differences between those with and without polyneuropathy were noted for the following variables: age, OR 1.08 (95% CI 1.05–1.12); waist circumference, 1.03 (1.01–1.05); low physical activity, 0.40 (0.21–0.78); PAD, 3.26 (1.65–6.45); diabetes, 3.99 (1.99–7.99); fasting glucose, 1.00 (1.00–1.01); A1C, 1.21 (1.06–1.38); log triglycerides, 1.61 (1.09–2.38); log creatinine, 5.78 (2.23–14.97); and log albuminuria, 1.24 (1.09–1.42). No differences were observed for male sex, height, weight, BMI, hip circumference, systolic blood pressure, smoking, alcohol intake, myocardial infarction, stroke, IGT, 2-h glucose, total cholesterol, LDL cholesterol, and HDL cholesterol.

The final mutivariate logistic regression models included age, male sex, height, weight, waist circumference, low physical activity, log creatinine, log albuminuria, A1C, IGT, diabetes, and PAD. The independent variables remaining in the final multiple logistic regression models with polyneuropathy (MNSI >2) as dependent variable are listed in Table 2. In the entire population studied, age, waist circumference, and diabetes were significantly associated with polyneuropathy (all P < 0.05), whereas the relationship with PAD reached borderline significance (P = 0.099). In the diabetic subjects, independent associations with polyneuropathy were noted for age, waist circumference, and PAD (all P < 0.05), whereas duration of diabetes did not reach statistical significance (P = 0.29).

CONCLUSIONS

The results of this study demonstrate that in the general population the prevalence of polyneuropathy is slightly higher in persons with IGT than in those with NGT and more than twofold higher in subjects with diabetes compared with those with IGT. We further show for the first time that the prevalence of polyneuropathy is also slightly higher in individuals with IFG than in those with NGT and only marginally lower than in those with IGT. Moreover, both in the general population and in diabetic patients, apart from age, waist circumference and PAD are independently associated with prevalent polyneuropathy. This is another novel finding suggesting an interplay between polyneuropathy and both cardiovascular risk factors and macroangiopathy in the lower limbs.

The vast majority of previous population-based studies did not assess waist circumference as a potential risk factor of polyneuropathy but did measure BMI or weight (13,16,19,21). However, these studies have not reported any association between BMI or weight and the prevalence of polyneuropathy in diabetic patients. In the U.S. National Health and Examination Survey (NHANES), weight ≥92 kg (4th quartile) was associated with insensate feet as assessed by the 10-g monofilament, yielding an OR of 2.4 (95% CI 1.8–3.1) in the nondiabetic population, but this association was not observed in the diabetic population (19). In the Australian Diabetes Obesity and Lifestyle (AusDiab) study (17), including type 2 diabetic patients, neither BMI nor waist circumference were identified as risk factors for polyneuropathy in univariate analyses. Some studies have not taken measures of obesity into consideration at all when evaluating the possible risk factors of polyneuropathy (10,15,18,23). Moreover, PAD verified by ankle brachial index has not been previously reported as a risk modifier for the prevalence of polyneuropathy in diabetic patients. Thus, the present study is the first to report an independent association of prevalent polyneuropathy with both waist circumference and PAD in the diabetic population. An increase in waist circumference by 1 cm was associated with a 4% increase in the likelihood of polyneuropathy. Due to the cross-sectional nature of this study, it can be concluded that visceral obesity is not a predictor for the development of polyneuropathy and does not play a pathogenetic role, but against the background of the independent association of polyneuropathy with PAD reported herein, it is tempting to speculate that visceral obesity as an important component and macroangiopathy as a frequent sequel of the metabolic syndrome may foster the risk of developing polyneuropathy in diabetic subjects. The metabolic syndrome (visceral obesity, dyslipidemia, hyperglycemia, and hypertension) has become one of the major public health challenges worldwide. There has been growing interest in this constellation of closely related cardiovascular risk factors (43–45). Indeed, central obesity, as assessed by waist circumference, rather than BMI, was agreed as essential to define the metabolic syndrome by different panels because of the strength of the evidence linking waist circumference with cardiovascular disease and the other metabolic syndrome components and the likelihood that central obesity is an early step in the etiological cascade leading to full metabolic syndrome (44,45). However, whether central obesity is a harbinger of diabetic polyneuropathy can only be answered by prospective studies.

This study does not confirm some previous population-based studies indicating that IGT is associated with an increased prevalence of polyneuropathy (10,30). While the point estimate indicates an increased prevalence, the difference did not reach statistical significance, possibly due to the relatively low sample size. On the other hand, it is conceivable that higher age and waist circumference may contribute to a higher prevalence of polyneuropathy in individuals with IGT compared with those with NGT, as these risk factors were associated with polyneuropathy in the entire population studied. In the San Luis Valley Diabetes Study (10) the prevalence of polyneuropathy was 3.9, 11.2, and 25.8% in subjects with NGT, IGT, and diabetes, respectively. The OR (95% CI) for the presence of polyneuropathy in individuals with IGT (n = 89) was 3.5 (1.5–7.9) compared with those with NGT (n = 488) (10). In the Hoorn Study (30) only the risk of bilateral absence of ankle reflexes (OR 1.7 [95% CI 1.1–2.8]), but not knee reflexes (1.2 [0.4–4.1]) or vibration sensation at the big toe (0.8 [0.5–1.3]) or at the medial malleoli (0.9 [0.4–2.2]), was associated with IGT as compared with NGT. Other studies have found no association between IGT and prevalent polyneuropathy (9,13,31,46). In a large sample of individuals with IGT or IFG, the AusDiab Study (47) recently reported a markedly lower prevalence of polyneuropathy, as compared with our study reaching only 3.9% when diagnosed by the Neuropathy Disability Score and 6.1% when diagnosed by an overall neuropathy score. However, the corresponding rates of polyneuropathy in a population with NGT were not reported (47). Thus, the results of the present study are compatible with the notion that the available evidence does not generally suggest a significantly elevated prevalence of polyneuropathy in individuals with IGT.

An interesting aspect in the context of a presumable “pre-diabetic neuropathy” (27,33) is the role of IGT in CIAP. Several uncontrolled observational studies have recently reported an increased prevalence of IGT in patients with CIAP (27,32,33). In the only controlled study hitherto available, 32% of patients with CIAP and 14% of the control subjects had IGT or fasting hyperglycaemia, but after adjusting for age and sex the difference was not significant, even in the painful neuropathy subgroup (34). A recent review has concluded that despite extensive studies, it is unclear whether IFG or IGT may cause diabetic polyneuropathy or CIAP, as some studies suggest that pre-diabetes is a common and important cause of CIAP, whereas others do not. It was judged that a considerable degree of this disparity may relate to differences in selection of patients, choice of control subjects, assessment of chronic glycemic exposure and of diabetes complications, and statistical power (29). There is general agreement that prospective controlled studies are required to definitively answer the question whether polyneuropathy develops more frequently and more severely in individuals with pre-diabetes compared with those with NGT (26,28,29).

In conclusion, at the population level the prevalence of polyneuropathy in individuals with IGT is slightly higher than in those with NGT. To establish whether this is a true difference, larger samples are required. Apart from age, an important risk factor associated with polyneuropathy in diabetic patients is waist circumference, whereas PAD is a relevant associated disorder. Thus, abdominal obesity and peripheral macrovascular disease may represent important targets to prevent diabetic polyneuropathy.