Introduction

Diabetic peripheral neuropathy (DPN) is a very common and clinically relevant diabetic complication. In an Italian multicenter study conducted on 8757 patients with type 2 diabetes, the prevalence of DPN was 32.3 %, and its severity increased with aging and duration of diabetes [1]. Distal symmetric polyneuropathy is characterized by sensory and/or motor dysfunctions with a typical symmetrical distribution, and it represents the most frequent clinical form of DPN, affecting more than 90 % of patients with neuropathy [2].

Although chronic hyperglycemia is certainly the major determinant of DPN, several lines of evidence indicate that other nutritional, lifestyle, metabolic, and/or genetic factors may contribute to its occurrence.

In the last two decades, a number of studies have investigated the role of moderate hyperhomocysteinemia as an independent risk factor for both macro- and microvascular diabetic complications, including DPN [3, 4].

It has been shown that elevated levels of total homocysteine (tHcy) significantly correlated with DPN, independent of other risk factors [5–9]. In a cohort of patients with type 2 diabetes, a 23 % increase in the risk of DPN for every increase of 1 μmol of tHcy was demonstrated [6]. Similar results were reported in a Chinese population of 249 diabetic patients with DPN [7], and in the appropriate blood pressure control in diabetes (ABCD) trial [8].

Circulating levels of tHcy are finely regulated by vitamin B12 and folate plasma concentrations, as well as by the common C677T genetic variant in methylenetetrahydrofolate reductase (MTHFR) gene, both in the general [10] and in the diabetic population [11]. Hyperhomocysteinemia may be associated with DPN both directly or as an early marker of B-vitamins deficiency. Thus, folate deficit is involved in neural tube defects [12] and vitamin B12 contributes in the development and maintenance of the neural tube and axonal branching [13]. Furthermore, the role of vitamin B12 deficiency in the development of DPN is supported by several intervention studies demonstrating a beneficial effect of treatment with B-vitamins in DPN [14–18]. Also MTHFR mutation has been associated with a higher DPN risk [19, 20]. In contrast, other Authors have not observed any relationship of MTHFR mutation and/or B-vitamins deficiency with DPN [21].

It is well known that metformin treatment can be associated with a reduction in vitamin B12 levels, as well as with high levels of methylmalonic acid (MMA) and hyperhomocysteinemia [17, 22–25].

Since metformin is the first-line therapy in the management of type 2 diabetes, it is important to ascertain whether this drug is associated with an increased risk of DPN, which could be mediated by the modulation of tHcy, folate, and vitamin B12 levels.

Accordingly, the purposes of this study were (1) to assess the potential associations of DPN with serum levels of homocysteine, vitamin B12, folate, and the common MTHFR C677T polymorphism, (2) to explore the influence of chronic treatment with metformin on homocysteine, vitamin B12, and folate levels, and finally, (3) to evaluate whether, by this influence, metformin is a risk factor for DPN in a group of type 2 diabetic outpatients.

Patients and methods

Patients

This cross-sectional study was conducted in the diabetic outpatient clinic of the Department of Clinical and Experimental Medicine of the University Hospital of Messina and it involved 263 type 2 diabetic subjects who consecutively underwent a specific screening for DPN in the 3 months prior to the study. All subjects were prescribed a low-fat and low-glycemic index diet according to current guidelines [2] and they underwent to regular counseling about dietary habits.

All subjects were on stable hypoglycemic therapy, including metformin, for more than 6 months.

Patients were classified as non-smokers or current smokers, including among the latter those who had quit for less than a year.

Exclusion criteria included: type 1 diabetes, gestational diabetes or other forms of diabetes, pregnancy, active cancer, chronic inflammatory diseases, thyroid diseases, chronic renal failure, nutritional deficiencies, and neurological disorders potentially associated with neuropathic signs and/or symptoms. Subjects who referred a daily intake of alcohol >30 g for men, and >20 g for women were also excluded. Furthermore, pernicious anemia was excluded on the basis of complete blood count (CBC), which is performed as a part of routine laboratory testing at least once a year in the management of all type 2 diabetic outpatients. All subjects with abnormal CBC values are usually addressed to hematological specialist’s appointment and, eventually, second-level blood exams, including the dosage of anti-gastric mucosa Ab.

All study participants gave their written informed consent and the study was approved by the local Ethical Committee (Comitato Etico Interaziendale della Provincia di Messina).

Methods

Clinical and lifestyle information were collected by a standardized questionnaire. Blood pressure, body mass index (BMI), and waist circumference were measured by standard methods.

In all subjects, blood pressure was measured two times in the seated position after 5 min of rest and mean systolic and diastolic pressure values were recorded. Participants were classified as having hypertension if they reported the use of antihypertensive medications or presented blood pressure levels ≥140 mmHg of systolic blood pressure or ≥90 mmHg of diastolic blood pressure [26]. Dyslipidemia was defined as self-reported treatment with lipid‐lowering medications or having LDL cholesterol levels ≥100 mg/dl, or triglycerides levels >150 mg/dl, or HDL‐C <50 mg/dl in women or <40 mg/dl in men [2].

All participants underwent venipuncture after 12–14 h of fasting for laboratory examination. Glucose and creatinine levels and lipid concentrations were measured with standard automated laboratory methods (Roche Diagnostics, Milan, Italy). Glycated hemoglobin (HbA1c) was measured using an automated high-performance liquid chromatography (HPLC) analyzer (Diamat; Bio-Rad Laboratories, Milan, Italy). LDL cholesterol was calculated using the Friedewald formula [27]. Fasting tHcy plasma concentration was measured by HPLC and fluorimetric detection (Bio-Rad Laboratories, Milan, Italy); the coefficient of variation (CV) of the assay was 2.9 % and the normal laboratory range 5–15 μmol/l. Plasma folate and vitamin B12 concentrations were measured by radioassay (Bio-Rad Laboratories, Milan, Italy); CVs were 3.8 and 5.9 %, respectively.

Definition and diagnosis of diabetic neuropathy (DPN)

All study participants were screened for the presence of DPN in the 3 months prior to recruitment [2, 28]. At the screening visit, DPN was suspected based on the abnormalities of at least one of the following evaluations: accurate foot examination (deformities, dry skin, calluses, muscle wasting, hypertrophic nails, onychomycosis, infection, and cracks), pulse palpation of the dorsalis pedis artery, assessment of Achilles tendon reflex and pressure sensitivity (using Semmes–Weinstein monofilament 10 g), vibratory sensibility (by diapason calibrated to 128 Hz), discrimination (via pin prick on the dorsum of the foot), and tactile sensitivity (by staple cotton applied to the dorsum of the foot). DPN diagnosis was then confirmed by the measurement of conduction velocities of ulnar (motor and sensory), tibial, peroneal, performed by a neurologist.

Definition of other chronic complications

Diabetic micro- and macroangiopathy were screened according to the current Guidelines [2]. Retinopathy was diagnosed or excluded on the basis of a direct ophthalmoscopy (through a dilated pupil) performed by an expert ophthalmologist and/or by fluorescein angiography within 1 year before the start of the study.

Renal function was evaluated by measuring serum creatinine and calculating estimated glomerular filtration rate (eGFR) by the modification of diet in renal disease (MDRD) equation [29]. Microalbuminuria evaluation was available only in a subset of patients and it was therefore not included in the present analysis.

Coronary heart disease (CHD) was defined on the basis of cardiologist medical records and/or hospital discharge as a history of myocardial infarction, coronary artery bypass graft surgery, or coronary angioplasty. As part of the clinical screening, all patients also underwent a standard electrocardiogram evaluation in the 12 months prior to the study, to exclude asymptomatic myocardial infarction or arrhythmia.

Genetic analysis

Genomic DNA extraction was carried out on whole blood using QIAamp DNA blood Midi kits (Qiagen, Hilden, Germany) according to the manufacturer’s recommended protocol. MTHFR C677T genotyping was performed in all patients as previously described [11]. Briefly, a fragment of 198 bp in the MTHFR gene was amplified by polymerase chain reaction (PCR) using oligonucleotide primers. The PCR products were subjected to restriction enzyme analysis by digestion with HinfI (MspI, MBI Fermentas, Lithuania), and the fragments were separated by electrophoresis on agarose gel, and visualized by UV illumination, after ethidium bromide staining. The resulting fragments were 175 and 23 bp for the mutant T allele and 198 bp for the uncut wild-type C allele.

Statistical analysis

Statistical analysis was performed using SPSS 11.0 for Window package.

Preliminarily, the sample size calculation established a minimum number of 250 subjects to be enrolled in our study as necessary to guarantee a significance level alpha = 0.05.

Numerical data are expressed as mean ± standard deviation (SD) and categorical variables as number and percentage (%). Since the majority of the examined variables showed a normal distribution, as verified by Kolmogorov–Smirnov test, a parametric approach has been used. For each parameter the two-tailed Student’s t test was performed for statistical comparisons; in order to compare numerical variables, the χ 2 test or by Fisher’ s exact test was applied, as appropriate. To compare subjects with and without DPN, Chi-square test (χ 2) for categorical measures was used. In order to assess variables independently associated with DPN, a multivariate logistic regression analysis was performed, including in the model all variables that resulted significant at univariate regression analysis (age, diabetes duration, levels of HbA1c, HDL-C, creatinine, and the presence of CHD). Furthermore, a multivariate regression analysis including all study variables in the model was also performed, but because of multicollinearity, no independent correlations were identified.

Odds ratio and 95 % confidence intervals were calculated by exponentiation of logistic regression coefficients. To compare continuous variables ANOVA was used. All statistical comparisons were two-tailed, and a P value <0.05 was considered statistically significant.

Results

Clinical characteristics of the study population

The clinical characteristics of 263 type 2 diabetic outpatients participating in the study are shown in Table 1.

Table 1 Clinical characteristics of T2DM outpatients
Full size table

Study subjects (151 women and 112 men) had a mean age of 64 years (range 32–86 years) and a mean diabetes duration of 12 years (range 1–46 years). Overall, they were overweight (mean BMI 29 kg/m2, waist circumference 99 cm), with a HbA1c mean value of 7.4 % (57 mmol/mol) and a lipid profile characterized by values of LDL cholesterol (LDL-C) which were, on average, higher than those recommended for diabetic patients (mean LDL-C 129 mg/dl); the values of HDL cholesterol (HDL-C) and plasma triglycerides were within the recommended levels.

Creatinine values were within the normal range, with an eGFR of 77 mL/min/1.73 m2.

Plasma levels of tHcy (13.2 µol/l), folate (4.4 pg/ml), and vitamin B12 (605.6 pg/ml) were within the normal laboratory range, although a moderate hyperhomocysteinemia, as indicated by plasma levels of tHcy ≥ 15 µol/l, was detected in 63 subjects (24 %).

With regard to the treatment of diabetes, 47 % of subjects were treated with metformin (monotherapy or in combination with other hypoglycemic drugs), 55 % with sulfonylureas or repaglinide, 17 % with insulin (alone or in combination with oral hypoglycemic agents), and one subject was treated with glitazones.

Overall, 160 (61 %) study participants had hypertension on pharmacological treatment, and 99 (38 %) were affected by dyslipidemia (61 subjects treated with statins and 9 with fibrates).

Screening of chronic diabetic complications detected 79 subjects (30 %) with DPN, 82 subjects (31 %) with diabetic retinopathy, and 26 (10 %) with documented coronary heart disease (CHD).

Fasting tHcy and B-vitamins plasma levels according to MTHFR C677T genotype distribution

The distribution of MTHFR C677T polymorphism did not differ from the frequencies predicted by the Hardy–Weinberg equilibrium (P = NS), with a TT homozygous prevalence of 24.3 % (Table 2).

Table 2 Fasting tHcy, B-vitamins plasma levels, and DPN according to MTHFR C677T genotype distribution in T2DM outpatients
Full size table

Fasting tHcy plasma levels were significantly higher (15.2 ± 8.4 vs 12.2 ± 3.6 μmol/l, P = 0.005) and folate levels were lower (3.9 ± 2.3 vs 4.8 ± 3.5 pg/ml, P = 0.066) in TT subjects as compared to CC subjects, whereas no difference in vitamin B12 concentrations was noted according to genotype (Table 2).

Clinical characteristics of the study population according to the presence of DPN

As shown in Table 3, subjects with DPN (n = 79) were in prevalence men (P = 0.023), older (P = 0.047), with a longer duration of diabetes (P = 0.027); they also showed higher levels of HbA1c (P = 0.017), creatinine (P < 0.001), and lower values of HDL-C (P = 0.027). Moreover, in the group with DPN the use of insulin was more frequent (P = 0.005) and CHD prevalence (P < 0.001) was higher.

Table 3 Clinical characteristics of T2DM outpatients according to the presence of DPN
Full size table

Fasting tHcy and B-vitamins levels were not significantly different in subjects with and without DPN; also the percentage of subjects with the MTHFR mutation (T carriers) was similar in the two groups (66 vs. 72 %, P = NS) (Table 3).

Since age and diabetes duration were significantly different in subjects with and without DPN, an analysis adjusted for these two variables was also performed. However, as shown in Table 3, this adjustment did not change the trend of our results.

Fasting tHcy and B-vitamins levels and DNP prevalence according to treatment with metformin

The distribution of subjects with or without DPN according to metformin therapy is shown in Fig. 1. The percentage of subjects with DPN was 33 % in the group treated with metformin and 27 % in the not-treated group, and this difference was not statistically significant.

Fig. 1
figure 1

Prevalence of subjects with and without DPN according to metformin treatment. DPN diabetic peripheral neuropathy. *P = NS

Full size image

The potential influence of metformin on major study variables is shown in Table 4. Study subjects treated with metformin (either alone or in combination with other hypoglycemic drugs) were younger (P < 0.001) and less frequently male (P = 0.05) compared with those not taking the drug. Fasting vitamin B12 levels were significantly lower in metformin-treated patients (P = 0.001), while there were no significant differences in the plasma levels of tHcy and folate levels and the prevalence of moderate hyperhomocysteinemia between the two groups.

Table 4 Fasting tHcy, B-vitamins plasma levels according to metformin therapy
Full size table

Factors associated with DPN at univariate and multivariate regression analyses

At univariate regression analysis (Table 5), DPN was positively associated with age (P = 0.048), duration of diabetes (P = 0.029), HbA1c (P = 0.019), creatinine levels (P = 0.001), and the presence of CHD (P < 0.001) and negatively with HDL-C concentrations (P = 0.028).

Table 5 Factors associated with DPN at univariate and multivariate regression analyses
Full size table

At multivariate regression analysis, performed including all variables in the model that resulted significant at univariate regression analysis (age, diabetes duration, levels of HbA1c, HDL-C, creatinine, and the presence of CHD), higher creatinine levels (P = 0.06) and low plasma levels of HDL-C levels (P = 0.013) and the presence of CHD (P = 0.001) were the only factors that were associated with DPN (Table 5).

A multivariate regression analysis including all the study variables in the model was also performed, although no independent correlations with DPN were identified because of multicollinearity among the variables (data not shown).

Discussion

Type 2 diabetes is associated with microvascular complications that contribute to its increased morbidity and mortality risk, including DPN. In our study, DPN was diagnosed in 30 % of subjects undergoing clinical screening, in line with the prevalence previously reported in a large Italian cohort [1].

Beyond the undisputed role of chronic hyperglycemia in the pathogenesis of this debilitating complication, other factors could contribute to its incidence and/or progression. Among these, several data suggest the potential role of mild hyperhomocysteinemia, MTHFR C677T mutation, and/or vitamin B12 deficiency [5–9, 14–20].

We therefore evaluated the relationships between these variables and DPN in a group of type 2 diabetic outpatients, assessing the potential influence of metformin therapy.

In our cohort, fasting tHcy, vitamin B12, and folate levels were generally within the normal range and comparable between subjects with and without DPN.

Studies conducted to date have not provided unequivocal results about the association of hyperhomocysteinemia and/or its common genetic and vitamin determinants with several long-term diabetic complications, including DPN. However, these observations are still very sparse and inconclusive. In line with some of these studies [11, 21, 30], our results do not support a significant role of hyperhomocysteinemia or MTHFR mutation on the development of DPN, since in our cohort, tHcy levels and MTHFR genotype distribution were comparable between subjects with or without DNP.

According to these results, it seems that hyperhomocysteinemia does not have a causal role on DPN, but it may rather be a surrogate marker for vitamin B12 deficiency. It is well known that chronic vitamin B12 deficiency is associated with pernicious anemia and several neurological deficits and, consequently, it could also be involved in the development of DPN. Furthermore, the hypothesis that low levels of vitamin B12 are a risk factor for DPN is supported by several observations, including a limited number of intervention studies. Thus, in a population of type 2 diabetic subjects with DPN, the administration of methylcobalamin, the active form of vitamin B12, together with folate and pyridoxal 5′-phosphate reduced the frequency and intensity of symptoms and improved the density of epidermal nerve fibers in calf muscle biopsies [14]. Furthermore, in 20 type 2 diabetic patients with DPN, treatment with the same micronutrients for 1 year significantly improved the loss of skin sensitivity [15], and in another small group of patients supplemented for 4 months with multivitamins including vitamin B12, a significant improvement in the symptoms of neuropathy was observed [16].

Although all our study subjects were regularly counseled on diet according to current diabetes guidelines [2], dietary information was not available for this analysis. Since the mean age of our participants was >60 years, we cannot rule out that an age-related impairment of nutrition may have influenced vitamin levels, as reported in elderly subjects [31].

However, subjects recruited in our study had vitamin B12 levels within the normal laboratory range and comparable to those reported in other studies of the same background [10, 11, 17]. Furthermore, in order to limit our observation to subjects commonly attending diabetes outpatient clinics, vitamin B 12 deficiency-associated conditions, such as alcohol abuse or pernicious anemia, were carefully excluded in our study.

Beyond the nutritional or age-related causes of vitamin B12 deficit, other factors, such as hypoglycemic therapies, may impair vitamin B levels in diabetic subjects. The potential deleterious effects of vitamin B12 deficiency on DPN occurrence has attracted clinicians’ attention after noticing that the most widely used drug for the treatment of diabetes, i.e., metformin, may also reduce vitamin B12 levels [25, 32]. In line with other reports [24, 25, 32–34], our study confirmed that metformin treatment is associated with a reduction of vitamin B12 serum concentration, although in our population this reduction was mild and still within the reference range. The proposed mechanisms include alterations in small bowel motility and bacterial overgrowth, the competitive inhibitions in intrinsic factor levels, and interaction with the cubilin endocytic receptor [35]; metformin has also been shown to inhibit the calcium-dependent adsorption of the vitamin B12-IF complex at the terminal ileum [36].

Despite this effect on vitamin B12, treatment with metformin had no clinical consequences on DPN risk in our study subjects, accordingly to what reported by some Authors [32]; on the contrary, others have observed a dose-dependent DPN risk associated with this drug [34].

Notably, patients enrolled in our study were on stable hypoglycemic therapy and they were treated with metformin for more than 6 months at the maximum tolerated dosage, with 80 % of them taking ≥2 g/day (data not shown). Despite this relatively long treatment with therapeutic dosage of metformin, vitamin B12 levels were overall in the normal range, with only 20 % of subjects showing levels <300 pg/ml, suggesting that at least in a population of Mediterranean origin, like ours, the nutritional intake of B-vitamins is adequate and overcomes the effect of the drug.

Taken together, our data indicate that hyperhomocysteinemia, MTHFR mutation, and B-vitamin deficiency are not major risk factors for the development of neuropathic damage. Conversely, age, duration of diabetes, and HbA1c levels were positively associated with DPN in our population, thus confirming the role of glycemic control in its pathogenesis [37]. However, HDL-C, CHD, and creatinine levels, were the only factors independently associated with DPN at multivariate regression analysis, although with different degrees of significance.

The clinical meaning of the association of higher creatinine levels with DPN occurrence, also reported in other studies [38, 39], may be related to the concomitant presence of different microangiopathic complications in the same patients. However, retinopathy did not show the same trend, suggesting that beyond chronic hyperglycemia, nephropathy, and neuropathy may share the same underlying risk factors, for example low HDL-C levels.

Although still limited, the number of observations supporting the role of diabetic dyslipidemia, i.e., hypertriglyceridemia, higher sdLDL concentrations, and low HDL-C levels, in diabetic microangiopaty is increasing. Some Authors have reported that diabetic dyslipidemia including low HDL-C levels correlated with DPN and was an independent risk factor for its development [40–42]. Furthermore, several clinical studies have associated poor lipid profile with DPN development and progression, independent of glycemic control [43, 44].

The mechanisms by which circulating lipid levels may contribute to pathogenesis of DPN are still unknown, although several hypotheses have been proposed, including underlying insulin resistance or the generation of oxidized LDL particles induced by low HDL-C levels [42, 45, 46]. All these clinical and experimental evidences support an important role of diabetic dyslipidemia also on microvascular complications and in particular on DPN, and our results demonstrating that low HDL-C levels are an independent risk factor for DPN are well in line with this hypothesis.

Furthermore, our data show an association between CHD occurrence and DPN. This association between macro- and microvascular complications can be interpreted in the view of common risk factors, such as endothelial dysfunction, oxidative stress, and/or subclinical inflammation that may be elicited by high glucose or lipid levels in these patients. On the other hand, in non-diabetic subjects, neuropathy has been also associated with vascular factors such age, waist circumference, comorbidities, and peripheral arterial disease [47], and this could easily apply also to our diabetic patients.

Our study has some limitations, firstly related to its cross-sectional design, that does not allow us to fully explore the potential temporal and causal relationships between metformin treatment, vitamin B12 deficiency, and the onset of DPN. Furthermore, the sample size is relatively small, although the number of subjects included in our analysis was sufficient to detect meaningful differences. Moreover, our sample size was larger than that reported in the other few studies measuring vitamin B levels in addition to genotype analysis [19–21] and, unfortunately, in none of them metformin treatment was taken into account.

Also the specialty-based setting (tertiary referral center of diabetes) does not allow us to generalize our findings to all the people with type 2 diabetes. Finally, microalbuminuria evaluation was not included in the present analysis since it was available only in a subset of patients.

In conclusion, our results indicate that the presence of CHD and low levels of HDL-C, and to a lower extent higher creatinine concentrations are independently associated with DPN, while tHcy, folate, and vitamin B12 plasma levels and the common MTHFR C677T mutation do not play a significant role, even in metformin-treated subjects. Larger studies are needed to confirm the utility of treating diabetic dyslipidemia, and particularly low levels of HDL-C, in order to reduce not only the cardiovascular but also the neuropathic risk in diabetic subjects.