Vitamin B12 deficiency in type 2 diabetic patients on metformin: a cross-sectional study from South-Western part of Ghana

ABSTRACT Introduction: Metformin is the most widely administered anti-diabetic medication among type 2 diabetes mellitus (T2DM) patients. However, metformin induces vitamin B12 malabsorption which may increase the risk of vitamin B12 deficiency among T2DM patients. We determined the prevalence of vitamin B12 deficiency and related risk factors among Ghanaian T2DM patients on metformin therapy. Methods: This cross-sectional study recruited 196 T2DM patients attending the outpatient diabetic clinic at the Effia Nkwanta Regional Hospital, Ghana. Fasting venous blood was collected for biochemical analysis. Vitamin B12 deficiency was defined as serum B12 <100 pg/ml and methylmalonic acid (MMA) ≥ 0.4µmol/L. Results: The prevalence of vitamin B12 deficiency based on serum vitamin B12, MMA, and the combination of both methods were 32.1%, 14.8%, and 14.3%, respectively. Longer duration of metformin use [5-9 years; aOR= 2.83, 95% CI (1.03-7.81), p=0.045 and ≥10 years; aOR= 4.17, 95% CI (1.41-12.33), p=0.010], higher daily dose of metformin [1000-2000 mg/day; aOR= 1.34, 95% CI (0.25-2.74), p=0.038 and >2000 mg/day; aOR= 1.13, 95% CI (0.39-2.97), p=0.047], and very high body fat [aOR= 2.98, 95% CI (1.47-6.05), p=0.020] were significantly associated with increased odds of vitamin B12 deficiency. For daily dose of metformin, a cutoff value of 1500 mg/day presented with a sensitivity, specificity, and AUC of 71.4%, 40.1%, and 0.54 (95% CI, 0.53-0.54), respectively, in predicting vitamin B12 deficiency. A ≥ six (6) years duration of metformin therapy presented with a sensitivity, specificity, and AUC of 70.4%, 62.9%, and 0.66 (95% CI, 0.57-0.75), respectively, in predicting vitamin B12 deficiency. Conclusion: Vitamin B12 deficiency is high among T2DM patients on metformin therapy in Ghana. There is the need for regular monitoring of vitamin B12 levels especially in T2DM patients on metformin daily dose of ≥ 1500 mg for duration of therapy ≥ 6 years.


Introduction
Diabetes mellitus (DM) is a pervasive chronic metabolic disorder and accounts for up to 90% of all diagnosed diabetes [1,2]. Type 2 diabetes mellitus (T2DM) is the most common form of DM, accounting for approximately 90% of DM cases [3]. In Ghana, 6.3% and 6.4% have been quoted as crude and age-adjusted prevalence of T2DM [4,5]. T2DM is associated with several detrimental microvascular and macrovascular complications [6]. As such, effective management of the disease is crucial.
The American Diabetes Association (ADA) and the European Association for the Study of Diabetes (EASD) recommend metformin as the first therapeutic choice for T2DM management with concurrent lifestyle modifications. Reports indicated that metformin improves peripheral insulin sensitivity and reduces the risk of cardiovascular mortality in T2DM [7,8] in addition to its beneficial effects on weight loss and vascular protection [9]. Nonetheless, as with most medications, the use of metformin present with some side effects. Though most of these side effects are mild, reports indicate that metformin use is associated with the diminution of the terminal iliac uptake of vitamin B 12 [10,11]. Other studies also report that long-term use of metformin results in vitamin B 12 malabsorption, with a concomitant reduction in serum vitamin B 12 levels [11][12][13]. Metformin-induced Vitamin B 12 deficiency has been reported to be duration of therapyand dose-dependent [12].
Metformin-induced vitamin B 12 deficiency has also been associated with neuropathy. The neuropathy associated with vitamin B 12 deficiency ranges from paresthesia and attenuated peripheral sensation in response to changes in mental status and proprioception [7] which overlap with diabetic neuropathy. The progression of vitamin B 12 deficiency-induced neurologic damage can, however, be abated through early detection and vitamin B 12 therapy [14]. Nonetheless, if peripheral neuropathy due to deficiency of vitamin B 12 is misconstrued as diabetic peripheral neuropathy, permanent neurological damage may occur [15].
Despite the growing evidence of metformininduced vitamin B 12 deficiency, the prevalence and association between metformin use and vitamin B 12 deficiency in the Ghanaian population is yet to be elucidated. Furthermore, there are no guidelines to address how often Ghanaian T2DM patients on metformin should be screened for the risk of vitamin B 12 deficiency which would inform appropriate prescription of vitamin B 12 supplements probably due to limited studies on the subject matter.
Diagnosis of vitamin B 12 deficiency can, however, be difficult. Reports indicate that high serum vitamin B 12 levels can be accompanied by signs of deficiency, and functional deficiency from tissue uptake defects and action of vitamin B 12 at the cellular level have been implicated in this association [16,17]. Thus, functional vitamin B 12 deficiency can occur regardless elevated serum B 12 levels [17]. As such, a more sensitive screening method is warranted. One such method exploits metabolites that accumulate due to vitamin B 12 deficiency. Vitamin B 12 is involved in the conversion of methylmalonic acid (MMA) to succinyl-CoA as well as homocysteine (HC) to methionine in combination with folic acid [18]. Serum vitamin B 12 , in combination with MMA levels, have been demonstrated to be effective in classifying vitamin B 12 deficiency compared to serum vitamin B 12 alone [16,19].
This study, therefore, employed both serum vitamin B 12 and MMA levels to estimate the prevalence of vitamin B 12 deficiency and identify risk factors associated with vitamin B 12 deficiency among Ghanaian T2DM patients.

Study design ∕ setting
This cross-sectional study was conducted at the Diabetic clinic of the Effia Nkwanta Regional Hospital, Takoradi, Ghana. Effia Nkwanta Hospital is a referral center for peripheral health facilities in the Western Region. The diabetic clinic offers general and specialized care for diabetes patients.

Ethical consideration
The study protocol was reviewed and approved by the Committee for Human Research, Publications and Ethics of SMS/KNUST/KATH and institutional approval was obtained from Effia Nkwanta Regional Hospital. Written informed consent was obtained from participants after the objectives and benefits of the study were explained to them. Participation was entirely voluntary and strict confidentiality of participants' information was maintained throughout the study.

Study population
The sample size for this study was calculated using the Raosoft sample size calculator [20]. At 95% confidence level, 7% margin of error, and a response distribution of 50%, the minimum sample size for the study was 180 T2DM patients. In an effort to enhance the statistical power of the study, a total of 250 T2DM patients were targeted for the study. However, upon excluding subjects who did not fulfill the inclusion criteria, a total of 196 participants were enrolled in the study. Nonprobability sampling technique was employed to recruit 196 consecutive consenting participants who were already diagnosed of T2DM and on treatment. Subjects recruitment occurred during patients' routine clinic visit. Since T2DM is considered a disease of the aged, patients aged 35-65 years old, who had been on metformin, with minimum daily dose of 500 mg, for at least 6 months were included. T2DM patients who were pregnant/lactating, alcoholics or drug abuse, known cases of malabsorption (gastrointestinal surgery, inflammatory bowel diseases and gluten allergy), chronic kidney disease, pernicious anemia, thalassaemia, sickle cell anemia, consumption of vitamin B 12 or calcium supplementation during the last 3 months and receiving antibiotics or any medications known to influence gastrointestinal motility were excluded from the study. All relevant clinical data were retrieved from the hospital's archive. Participants' selection protocol is shown in Figure 1.

Questionnaire administration
Validated questionnaire was used for data collection. The questionnaire gathered information on demographic characteristics, anthropometric variables, medications (type, dose, and duration) and symptoms of neuropathy. Patients information from the hospital's archive were cross-referenced to confirm responses such as medication dose, duration of use, and duration of diabetes.

Blood pressure and anthropometric measurements
Blood pressure was measured from the left upper arm using a mercury sphygmomanometer and a stethoscope by a trained nurse at the diabetic clinic. Participants were asked to rest for at least 5 min before measurement. The average of the two readings taken 5 min apart was recorded as the blood pressure measurement.
Height was measured without wearing shoes using a wall-mounted ruler, to the nearest 0.1 m. Waist and hip circumferences were measured to nearest centimeter using a measuring tape. Waist to hip ratio (WHR) = WC (m)/HC (m), waist to height ratio and (WHtR) = WC (m)/height (m) were calculated [21]. Total body fat, percentage visceral fat (VF) and skeletal muscle were also estimated using the Omron Body Composition Monitor (Omron Corporation, Japan). Body mass index (BMI) was calculated using the equation; [BMI (kg/m 2 ) = weight/height 2 ] [22].

Sample collection, preparation and biochemical assays
Five millimeters (5 ml) of venous blood sample was collected after an overnight fast. One (1) milliliter of blood was dispensed into tubes containing fluoride oxalate, another 1 ml was dispensed into tubes containing EDTA, and the remaining 3 ml of blood was dispensed into gel separator tubes. The sample in the EDTA tube was used for estimation of hemoglobin (Hb) and glycated hemoglobin (HbA1c) while the sample in the fluoride oxalate was used for glucose estimation (FPS). The gel separator tubes were placed in a centrifuge and span at 3000 rpm for 5 min to obtain the serum. Hb, HbA1c, and FPS were measured immediately and the serum for the measurement of other biochemical variables were stored at −20°C until analysis. FPS, HbA1c, total cholesterol (TC), highdensity lipoprotein cholesterol (HDL-C), triglyceride (TG), low-density lipoprotein cholesterol (LDL-C), and calcium were estimated using an automated chemistry analyzer (Selectra Pro S System, Elitech Group, France). The estimation of serum vitamin B 12 was based on solid-phase Enzyme Linked Immunosorbent Assay (ELISA) method (standardized with an intra-assay %CVs = 4.4-5.6% and inter-assay %CVs = 6.6-7.9%) (MyBioSource, Inc. San Diego, CA, USA) according to the manufacturer's instructions. Briefly, 100 μL of standards, samples and controls were pipetted into appropriate microtitre wells followed by 100 μL of enzyme conjugate reagent, mixed thoroughly, covered with an adhesive strip, and incubated at 37°C for 60 min. The incubation mixture was aspirated from the wells followed by five washes with the wash solution (400 μL). Residual water droplets were removed by striking the wells onto absorbent paper. A 50 μL of Substrate A and 50 μL of Substrate B were pipetted into each well, mixed gently and incubated at 37°C for 15 min.
The estimation of serum methylmalonic acid was based on the competitive binding enzyme immunoassay technique (standardized with an intra-assay %CVs <6.1% and inter-assay %CVs <10.2%) (MyBioSource, Inc. San Diego, CA, USA) according to the manufacturer's instructions. Briefly, 50 μL of standards, controls, and samples were pipetted into appropriate microtitre wells which were pre-coated with an antibody specific to MMA. Fifty microliters (50 μL) of Detection A working solution was added to each well, mixed thoroughly, covered with an adhesive strip, and incubated at 37°C for 60 min. The incubation mixture was aspirated from the wells followed by five washes with the wash solution (400 μL). Residual water droplets were removed by striking the wells onto absorbent paper. A 100 μL of Detection A working solution was pipetted into each well, mixed gently and incubated at 37°C for 45 min.
Both reactions were stopped by adding 50 μL of Stop Solution to each well and gently mixed for 30 s. The absorbance of the final colored product was measured spectrophotometrically at 450 nm using Thermo Electron Multiskan EX plate reader (Shanghai, China). The mean absorbance value (OD 450 ) for each set of reference standards, controls and samples were calculated. The calculated mean OD 450 obtained for each reference standard were used to construct a standard curve and the concentrations of samples and controls determined from the standard curve. Vitamin B 12 deficiency was defined by B 12 levels <100 pg/mL based on measurement of serum vitamin B 12 and ≥0.4 µmol/L based on MMA levels [23,24]. Daily calibration and maintenance of analyzer was performed according to the manufacturer's instructions as previously described [25]. Quality control (QC) was assessed using quality control materials provided by the manufacturer.

Statistical analysis
Categorical data were presented as frequencies and percentages, and Chi-square and Fisher exact test were used to test for significance of associations where applicable. Parametric and non-parametric continuous data were presented as means ± SD and medians (IQR), respectively, and Independent t-test and Mann Whitney U test were used to test significance of associations where applicable. Multivariate logistic regression model was used to identify risk factors associated with vitamin B 12 deficiency after adjusting for age, sex, overweight, anemia, insulin and sulfonylurea use, DM and medication duration. The covariates selected for the multivariate regression were known or surmised factors that could influence vitamin B 12 status. The receiver operating characteristic (ROC) curve was used to determine the cutoff value of metformin dose and duration of metformin therapy associated with vitamin B 12 deficiency. The optimum cutoff value was determined based on the highest Youden index [J = (sensitivity + specificity) -1]. A ρ value <0.05 was considered statistically significant. Data processing was done using Microsoft Excel 2016. Statistical analysis and graphical presentation were performed using the R Language for Statistical Computing version 3.5.2 (R Core Team, Vienna, Austria) [26].

Results
A total of 196 T2DM patients with mean age of 50.4 (±6.8) years were included in this study. The median duration of T2DM and metformin therapy was 6.0 (3.0-10.0) years and 5.0 (3.0-8.0) years, respectively. There were more females (69.9%) than males (30.1%). The prevalence of anemia based on hemoglobin level was 45.9% and the average vitamin B 12 and MMA levels among the entire study population were 205.4 pg/ml and 0.07 µmol/L, respectively ( Table 1).
The prevalence of vitamin B 12 deficiency based on serum vitamin B 12 (sVB 12  There were no statistically significant differences between the demographic, anthropometric, biochemical parameters and vitamin B 12 status (Table 2).
In the multivariate logistic regression, after adjusting for possible confounding variables, longer duration of metformin use [5-9 years (Table 4).

Discussion
This study reports a high prevalence of vitamin B 12 deficiency among Ghanaian T2DM on metformin.
The prevalence of vitamin B 12 deficiency based on serum vitamin B 12 , MMA, and the combination of both methods was 32.1%, 14.8%, and 14.3%, respectively. Vitamin B 12 deficient subjects presented with significantly longer duration of T2DM, duration of metformin use and used higher metformin doses compared to non-deficient subjects. Using multivariate logistic regression models, after controlling for multiple potential covariates observed in other  studies, we found longer duration of metformin use (≥5 years), higher daily dose of metformin (≥1000 mg/day) and very high body fat to be significantly associated with increased risk of vitamin B 12 deficiency. In determining the optimal cutoff value for daily dose of metformin and duration of metformin therapy in predicting vitamin B 12 deficiency, we employed the ROC curve analysis. We found metformin dose of 1500 mg/day and ≥six (6) years duration of metformin therapy to be predictive of vitamin B 12 deficiency among the study population.
Varying prevalence rates of vitamin B 12 deficiency among T2DM patients on metformin have been reported in diverse populations. Report from an early clinical observational study by Tomkin et al. [27] among Irish diabetic patients on long-term metformin therapy with concomitant dietary management indicated a prevalence of 30% for vitamin B 12 malabsorption. DeFronzo et al. also reported a prevalence of 29% among American T2DM patients using metformin [28]. Another study by Sparre Hermann et al. reported a 26.7% prevalence of vitamin B 12 deficiency in metformin-treated T2DM patients in Sweden [29]. Additionally, a study by Damião et al. [30] and Pflipsen et al. [23] found a 22.5% and 22% prevalence of B 12 deficiency among Brazilian and American T2DM patients using metformin. Wulffele   [33]. Compared to the recent studies, the alarmingly high prevalence of vitamin B 12 deficiency given the conservative cutoff points utilized in this study is striking. This underscores the need for apt guidelines to address how often T2DM patients on metformin should be screened for the risk of vitamin B 12 deficiency which would inform appropriate prescription of vitamin B 12 supplements. That aside, the discrepancies in prevalence rates may be attributed to disparities in the cutoff values for B 12 deficiency, influence of geographical variations and dietary characteristics. Though the mechanisms underpinning metformin-induced vitamin B 12 deficiency have not been fully elucidated, bacterial overgrowth in the small intestine due to DM, changes in the bacterial flora, alterations in small bowel motility, the inactivation of vitamin B 12 absorption, and the effect of calcium on cell membranes have been proposed [10,12,27,31]. Clinically, vitamin B 12 deficiency has been associated with macrocytic anemia, neuropathy, and mental changes [7,15,34,35]. Thus, though anemia observed in this study may have multifactorial causes, vitamin B 12 deficiency-associated macrocytic anemia may be the most probable [34,35]. Importantly also, the neurologic damage can present as peripheral neuropathy and may be misconstrued for diabetic neuropathy in patients on metformin therapy [7,15]. Thus, since vitamin B 12 -associated neuropathy is reversible, prompt detection and treatment of vitamin B 12 deficiency among T2DM patients on metformin is crucial to avert permanent neurological damage. Nonetheless, it is worthy of note that we did not observe significant association between the manifestations of neuropathy and anemia with vitamin B 12 deficiency. A possible reason for this finding may be due to the direct neuroprotective effect of metformin through its glucose-lowering effect and antihyperglycemic-independent, direct anti-neuropathic impact on neurons including inhibition of oxidative stressrelated apoptotic cell death [36,37]. Studies by Ahmed et al. in South Africa and Russo et al. in Italy [38,39] also found no significant difference in the presence of neuropathy between subjects with normal vitamin B 12 levels and those with vitamin B 12 deficiency. Additionally, though classical B 12 deficiency is associated with clinical symptoms such as anemia and peripheral neuropathy, these symptoms are usually absent in those with only biochemical vitamin B 12 deficiency [33]. This may be the reason for the non-significant association between the vitamin B 12 deficiency and anemia observed in this study.
The association of duration of metformin use and vitamin B 12 deficiency found in this study is expected as similar findings have been reported by previous studies. A study by Bauman et al. found that 12 out of 14 T2DM patients presented with reduced serum total vitamin B 12 levels after 3 months of metformin therapy [10]. A study by Ting et al. also found increased duration of metformin use to be associated with more than two-fold increased risk of developing vitamin B 12 deficiency [12]. Another study by De Jager et al. found that the absolute risk of vitamin B 12 deficiency after a period of 4 years was 7.2% higher in T2DM patients on the metformin [40]. A similar finding has also been reported by Wile and Toth [13], Wulffelé et al. [31], and Andrès et al. [41]. Despite the coherence with previous studies, metformin-induced vitamin B 12 deficiency in this study appeared to be duration of use-and dose-dependent similar to the findings of Beulens et al. [42]. Ko et al. [32] found that T2DM patients who had taken metformin >1,000 mg/day were approximately 10 times as likely to have vitamin B 12 deficiency as patients taking metformin ≤1,000 mg/day which is in harmony with this present study. However, Ko et al. also found that this dose (>1,000 mg/day) was associated with vitamin B 12 deficiency after ≥45.5 months (≈12 years) of therapy which is a longer period compared to the cutoff predictive of vitamin B 12 deficiency in this study (≥6 years). The longer duration of metformin use may be due to dissimilarities in the cutoff values of B 12 used. Ko et al. defined vitamin B 12 deficiency as serum levels ≤300 pg/mL while we defined vitamin B 12 deficiency as serum levels of both vitamin B 12 < 100 pg/mL and MMA ≥ 0.4µmol/L. The higher cutoff level in their study may have allowed for the inclusion of more vitamin B 12 deficient subjects with longer duration of metformin therapy, which may have been missed in this study, consequently influencing risk associations. Additionally, the interaction of duration of T2DM and duration of metformin therapy may be partly involved.
Also of note is the association between vitamin B 12 deficiency and very high body fat. A study by Baltaci et al. [43] reported that obesity based on bioelectric fat analysis (total body fat %) was significantly associated with vitamin B 12 deficiency which is in harmony with our study finding. Also consistent with our finding is a study by Pinhas-Hamiel et al. [44] who reported a greater than fourfold increased risk of reduced vitamin B 12 status in Israeli obese compared to normal-weight children and adolescents. Though the mechanisms underpinning the influence of obesity in vitamin B 12 deficiency is still under investigation, proposed mechanisms include obesity-induced B 12 malabsorption, poor dietary content, repeated short-term restrictive diets and increased requirements frequently seen in obese people [44][45][46]. Clinically, however, metformin is frequently prescribed at higher doses and for longer duration among obese T2DM patients. Thus, our finding suggests that, a more frequent assessment of vitamin B 12 status for obese T2DM patients using metformin may have immense positive influence in abating the high prevalence rate of metformin-induced vitamin B 12 deficiency.
Despite interpreting our results with caution, some limitation of the study should be acknowledged. One of which is the cross-sectional design of this study which precluded the establishment of the causal relation between metformin and vitamin B 12 deficiency. Thus, case-control or longitudinal studies are required to prove any causality in this association. The small sample size is also a limiting factor. Therefore, the findings may not be generalizable to the general population. The issue of external validity is also of concern because the study was conducted in a single urban center and significant disparities in prevalence rates may exist from in different the regions. A follow up using a larger sample size is, thus, warranted.

Conclusion
It was evident from this study that vitamin B 12 deficiency is frequent in Ghanaian T2DM patients on metformin. The study also confirms findings from other countries that daily dose and duration of metformin are risk factors for the development vitamin B 12 deficiency. However, these findings should not be considered as a basis for discontinuing of the use of metformin in the treatment of T2DM because metformin has been proven to produce a better glycemic control in T2DM. Rather, there is the need for regular monitoring of vitamin B 12 levels especially in patients on ≥1500 mg daily dose of metformin for a duration of six or more years.
Paul Nsiah is Senior Lecturer of Chemical Pathology at the