Dietary Sources of Vitamin B-12 and Their Association with Vitamin B-12 Status Markers in Healthy Older Adults in the B-PROOF Study

Low vitamin B-12 concentrations are frequently observed among older adults. Malabsorption is hypothesized to be an important cause of vitamin B-12 inadequacy, but serum vitamin B-12 may also be differently affected by vitamin B-12 intake depending on food source. We examined associations between dietary sources of vitamin B-12 (meat, fish and shellfish, eggs, dairy) and serum vitamin B-12, using cross-sectional data of 600 Dutch community-dwelling adults (≥65 years). Dietary intake was assessed with a validated food frequency questionnaire. Vitamin B-12 concentrations were measured in serum. Associations were studied over tertiles of vitamin B-12 intake using P for trend, by calculating prevalence ratios (PRs), and splines. Whereas men had significantly higher vitamin B-12 intakes than women (median (25th–75th percentile): 4.18 (3.29–5.38) versus 3.47 (2.64–4.40) μg/day), serum vitamin B-12 did not differ between the two sexes (mean ± standard deviation (SD): 275 ± 104 pmol/L versus 290 ± 113 pmol/L). Higher intakes of dairy, meat, and fish and shellfish were significantly associated with higher serum vitamin B-12 concentrations, where meat and dairy—predominantly milk were the most potent sources. Egg intake did not significantly contribute to higher serum vitamin B-12 concentrations. Thus, dairy and meat were the most important contributors to serum vitamin B-12, followed by fish and shellfish.


Study Population
This cross-sectional study was performed using baseline data of the B-PROOF study. The B-PROOF study is a randomized, placebo-controlled, double blind intervention study investigating the effect of supplementation of folic acid (400 µg) and vitamin B-12 (500 µg) on the prevention of osteoporotic fractures in 2919 mildly hyperhomocysteinemic older adults ě65 years. This large multi-center project was conducted in The Netherlands by a consortium from Erasmus MC (EMC, Rotterdam), VU University Medical Center (VUmc, Amsterdam), and Wageningen University (WU, Wageningen), with the latter acting as coordinator. Recruitment took place from August 2008 until March 2011. Participants were excluded if they used supplements or injections with high dose of B-vitamins during the past four months, were bedridden, incompetent to make own decisions, diagnosed with cancer (except certain types of skin cancer) in the past five years, had low (<12 µmol/L) or high (>50 µmol/L) plasma homocysteine levels, or renal dysfunction (serum creatinine levels >150 µmol/L). Vitamin B-12 intake from foods was only assessed in the Wageningen study population, where 600 participants provided reliable data on food intake and serum vitamin B-12. Further details on the design and methods of the B-PROOF study can be obtained from the design article [12]. The Medical Ethics Committee of Wageningen UR approved the study protocol and Medical Ethics Committees of VUmc and Erasmus MC confirmed local feasibility. All participants gave written informed consent. RCT trial registration: The B-PROOF study has been registered with ClinicalTrials.gov under identifier NCT00696514 since 9 June 2008.

Dietary Assessment
A Food Frequency Questionnaire (FFQ) containing 190 food items, specifically designed to assess intake of macronutrients, vitamin B-12, folate, vitamin D, and calcium, was used to estimate habitual dietary vitamin B-12 intake. The FFQ was administered once. The FFQ was developed by the dietetics group at the Division of Human Nutrition at Wageningen University and was validated for energy, fat, cholesterol, folate, and vitamin B-12 intake [13,14]. An item was added to the FFQ when it contributed to more than 0.1% of the intake of the nutrients where it was designed for. All items in the FFQ accounted for 90% of the total folate and vitamin B-12 intake according to Dutch National Food Consumption Survey data of 1998 [15]. The Dutch Food Composition database (NEVO) was used to calculate daily vitamin B-12 intake [16].

Biochemical Analyses
Blood samples were obtained from participants in the morning at the research center or at external locations in the living area of participants. Participants were in a fasted state, or had taken a light breakfast. To obtain plasma and serum a skilled nurse performed blood sampling of venous blood. For homocysteine determinations, blood collected in EDTA tubes were stored on ice water immediately after sampling, to prevent temperature-and time-dependent increases in plasma homocysteine [17,18]. Plasma homocysteine was measured in batches at the laboratory of Wageningen University using High-performance liquid chromatography (HPLC) (intra assay Coefficient of Variation (CV): 3.1%, inter assay CV: 5.9%) [19]. Serum vitamin B-12, folate, HoloTC, and methylmalonic acid (MMA) analyses were performed at the Laboratory of the Erasmus Medical Centre; all samples were analyzed at once when all samples were collected. Serum concentrations of vitamin B-12 and folate were determined by immunoelectrochemiluminescence assay (Elecsys 2010, Roche GmbH, Mannheim, Germany) (CV vitamin B-12 5.1% at 125 pmol/L and 2.9% at 753 pmol/L; CV folate: 5.9% at 5.7 nmol/L and 2.8% at 23.4 nmol/L) [20]. Holotranscobalamin (HoloTC) was determined by the AxSYM analyzer (Abbott, Wiesbaden, Germany) (CV < 8%) and serum concentrations of MMA were determined by modified gas chromatography-mass spectrometry (LC-MS/MS) (CV < 9%) [21]. Serum creatinine was measured with the enzymatic colorimetric Roche CREA plus assay (CV: 2%).

Covariates
Height was measured at baseline with a stadiometer to the nearest 0.1 cm. Weight was measured to the nearest 0.5 kg with a calibrated analogue scale. Body Mass Index (BMI) was calculated as weight/height 2 . Data on education level (primary, secondary or higher education), smoking status (non-smoker, current smoker, former smoker), physical activity (min/day) [22], and alcohol intake (light, moderate, excessive) were collected by means of questionnaires.

Data Analyses
Participant characteristics are reported as mean with standard deviation (SD), or percentages. Medians with interquartile range (25th-75th percentile) were used to report skewed variables. Differences between groups were analyzed using ANOVA in case of normally distributed continuous variables, Kruskal-Wallis in case of skewed continuous variables, and chi-square test in case of categorical variables. In order to compare serum vitamin B-12 concentrations across tertiles of vitamin B-12 intake from food items, analysis of covariance (ANCOVA) was performed in order to obtain adjusted means with 95% confidence intervals (95% CI). Compared food items included total vitamin B-12 intake, vitamin B-12 intake from meat, vitamin B-12 intake from fish and shellfish, vitamin B-12 intake from eggs, and vitamin B-12 intake from dairy. Furthermore, also subclasses of total dairy intake were examined, including vitamin B-12 intake from milk, vitamin B-12 intake from yogurt, and vitamin B-12 intake from cheese. p-for trend across tertiles of vitamin B-12 intake was calculated to assess a potentially statistical significant dose-response relationship between vitamin B-12 intake and serum vitamin B-12. Cox proportional hazards analysis with robust error variance was conducted to calculate Prevalence Ratios (PRs) for adequate serum vitamin B-12 concentrations (ě200 pmol/L) and normal serum vitamin B-12 and MMA concentrations (serum vitamin B-12 ě200 pmol/L and methylmalonic acid <0.27 µmol/L) [11] by tertiles of vitamin B-12 intake from different dietary sources, using the lowest tertile as the reference group. By assigning a constant risk period to all participants in the study, the obtained hazard ratio can be considered a prevalence ratio (PR) [23]. This PR corresponds to the probability of having adequate serum vitamin B-12 concentrations or normal serum vitamin B-12 and MMA concentrations in participants in the middle and upper vitamin B-12 intake tertile, compared to participants in the lowest vitamin B-12 intake tertile. Analyses were adjusted for age, sex (model 1), BMI, education, alcohol intake, physical activity level, smoking, creatinine (model 2), total energy intake, and intake of vitamin B-12-containing food items other than the exposure for the specific model (model 3). Again, P-for trend across tertiles of vitamin B-12 intake was calculated to assess a potential statistically significant dose-response relationship of vitamin B-12 intake with the probability of having an adequate serum vitamin B-12 concentration or normal serum vitamin B-12 and MMA concentrations. Associations of vitamin B-12 intake (continuously) with the probability of having an adequate serum vitamin B-12 concentration (dichotomously) were furthermore investigated, using restricted cubic spline regression conducted with three knots located at 1st decile of intake, 5th decile of intake, and 9th decile of intake. All analyses were performed using the statistical package SAS, version 9.2 (SAS Institute Inc., Cary, NC, USA). Unless stated otherwise, a two-sided p-value of ď0.05 was considered to be statistically significant.
Results in Table 4 confirm the findings in Table 3. Namely, the probability that participants in the upper tertile of vitamin B-12 intake as obtained by meat intake (ě1.52 µg/day) have an adequate serum vitamin B-12 concentration (PR, 95% CI: 1.22, 1.08-1.37) is comparable to the probability that participants in the upper tertile of vitamin B-12 intake as obtained by dairy intake (ě1.49 µg/day) have an adequate serum vitamin B-12 concentration (PR, 95% CI: 1.24, 1.10-1.39). For fish and shellfish intake, the probability of having an adequate serum vitamin B-12 concentration was 16% in the upper intake tertile (ě0.77 µg/day) when compared to the lowest tertile (PR, 95%: 1.16, 1.04-1.30). The borderline non-significant results for milk in Table 3, did appear to be statistically significant when investigated in relation to having an adequate serum vitamin B-12 concentration (PR, 95% CI for T1:1.0; T2: 1.12, 1.00-1.25; T3: 1.19, 1.06-1.33; P for trend: 0.004). For all food groups under study, differences in results between model 2 and model 3 were predominantly driven by additional adjustment for total energy intake.   The dose-response association was further examined by visualizing the associations between vitamin B-12 intake from different dietary sources and the probability of having adequate an serum vitamin B-12 concentration by spline regression analyses, where vitamin B-12 intake was modelled continuously (Figure 1). In addition to previous results, these graphs furthermore suggest that-at an intake level of 1 µg/day-dairy and meat seem to be somewhat more potent in reducing the probability of having an inadequate serum vitamin B-12 concentration than fish and shellfish in this population.
The dose-response association was further examined by visualizing the associations between vitamin B-12 intake from different dietary sources and the probability of having adequate an serum vitamin B-12 concentration by spline regression analyses, where vitamin B-12 intake was modelled continuously (Figure 1). In addition to previous results, these graphs furthermore suggest that-at an intake level of 1 μg/day-dairy and meat seem to be somewhat more potent in reducing the probability of having an inadequate serum vitamin B-12 concentration than fish and shellfish in this population.

Figure 1.
Associations between vitamin B-12-intake from different food sources with serum vitamin B-12-concentations ≥ 200 pmol/L, obtained by spline regression with three knots located at 1st, 5th, and 9th decile. Dotted lines represent 95% confidence intervals. Prevalence ratios were adjusted for age, sex, BMI, education, alcohol intake, physical activity level, smoking, creatinine, total energy intake, and intake of other vitamin B-12-containing food items. P for non-linearity was 0.003 for total vitamin B-12 intake; 0.002 for vitamin B-12 intake from meat; 0.16 for vitamin B-12 intake from fish and shellfish; and 0.34 for vitamin B-12 intake from dairy. P for linearity was 0.19 for vitamin B-12 intake from eggs; 0.06 for vitamin B-12 intake from milk; 0.94 for vitamin B-12 intake from yogurt; 0.44 for vitamin B-12 intake from cheese (data not shown).

Discussion
In this study, higher dietary vitamin B-12 intake was significantly associated with higher serum vitamin B-12 concentrations. Dairy and meat were the most important suppliers of dietary vitamin B-12 Figure 1. Associations between vitamin B-12-intake from different food sources with serum vitamin B-12-concentations ě 200 pmol/L, obtained by spline regression with three knots located at 1st, 5th, and 9th decile. Dotted lines represent 95% confidence intervals. Prevalence ratios were adjusted for age, sex, BMI, education, alcohol intake, physical activity level, smoking, creatinine, total energy intake, and intake of other vitamin B-12-containing food items. P for non-linearity was 0.003 for total vitamin B-12 intake; 0.002 for vitamin B-12 intake from meat; 0.16 for vitamin B-12 intake from fish and shellfish; and 0.34 for vitamin B-12 intake from dairy. P for linearity was 0.19 for vitamin B-12 intake from eggs; 0.06 for vitamin B-12 intake from milk; 0.94 for vitamin B-12 intake from yogurt; 0.44 for vitamin B-12 intake from cheese (data not shown).

Discussion
In this study, higher dietary vitamin B-12 intake was significantly associated with higher serum vitamin B-12 concentrations. Dairy and meat were the most important suppliers of dietary vitamin B-12 in this population, followed by fish and shellfish. Vitamin B-12 intake resulting from egg consumption was relatively low and did not significantly associate with increasing serum vitamin B-12 concentrations. When exploring the intake of dairy products in more detail, specifically by studying milk, yogurt, and cheese intake individually, milk intake was associated with the highest serum vitamin B-12 concentrations.
Before extending these findings to the results of previous studies and speculating on the potential practical application of these findings, first some limitations and strengths need to be discussed. First of all, sample size of the current study was relatively small compared to the studies by Tucker and colleagues (2000) [10] and Vogiatzoglou and colleagues (2009) [11]. However, based on the significant associations observed in our study, power does not seem to be an issue and therefore results of this study do contribute to the current scientific evidence in this field as data on bioavailability of vitamin B-12 from different food sources is very limited. Secondly, we explored differences in bioavailability from different vitamin B-12 food sources. It should be kept in mind though, that whereas vitamin B-12 intakes obtained from dairy and meat were in the same range, vitamin B-12 intakes from fish and shellfish, eggs, and individual dairy subgroups were lower. This limits comparability of these food sources to for instance total dairy and meat intakes. Thirdly, this study was conducted in a population with mildly elevated homocysteine concentrations, which may have resulted from lower vitamin B-12 intakes in this population when compared to a general population of older adults. When comparing our vitamin B-12 intake data with vitamin B-12 intake data as obtained by the recent Dutch Food Consumption Survey 2010-2012, median vitamin B-12 intake in the B-PROOF population indeed appears to be slightly lower than vitamin B-12 intake in the general older population. To be more specific, according to the Dutch Food Consumption Survey, which assessed vitamin B-12 intake in a population that is considered representative for the total older Dutch population, median vitamin B-12 intake from foods in men was 5.0 (25th-75th percentile: 3.9-6.4) µg/day and in women this was 4.3 (3.2-5.2) µg/day. In our population median (25th-75th percentile) vitamin B-12 intake was 4.5 (3.2-5.2) µg/day in men and 3.4 (2.6-4.3) µg/day in women. Fourthly, although diabetes has been associated with lower serum vitamin B-12 concentrations, diabetes patients were not excluded in this study. However, self-reported diabetes was assessed in 377 participants of the B-PROOF population, where no distinction was made between type I and type II diabetes. Additional analyses in this subgroup did not indicate that adjustment for diabetes resulted in substantially different outcomes; nor did sensitivity analyses in which participants reporting diabetes (n = 27 of n = 377) were excluded. Strengths of this study include the use of an FFQ that was validated for macronutrients as well as vitamin B-12 and folate intake, and the possibility to include important potential relevant covariates.
Currently, the Dutch Health Council recommends older adults meet a vitamin B-12 intake of 2.8 µg/day [24], which is met by the majority of the participants of the B-PROOF study. Spline regression suggested that the impact of total vitamin B-12 intake from foods to ensure adequate serum vitamin B-12 concentrations in our population levelled off somewhere around an intake of 6 µg/day. In a population of 98 Danish postmenopausal women, curves for several vitamin B-12 related variables-including circulating vitamin B-12 concentrations, HoloTC, MMA, and total homocysteine-also levelled off at a daily vitamin B-12 intake of 6 µg/day [25]. In the Framingham Offspring study [10] and Hordaland Homocysteine Study [11] this plateau was located around 10 µg/day.
When studying specific vitamin B-12 rich foods, we observed that dairy and meat product intake were significantly associated with higher serum vitamin B-12 concentrations. Dose-response analyses by means of spline regression subsequently suggested that both food groups possessed an about equal potency to raise vitamin B-12 concentrations. More specifically, at an intake level of 1 µg vitamin B-12/day as obtained by meat, participants were 30% more likely to have adequate serum vitamin B-12 concentrations compared to those consuming 0 µg vitamin B-12/day as obtained by meat. For dairy, this probability was 20% higher for those consuming 1 µg vitamin B-12/day. Of the dairy subclasses, milk seemed to be the most important component of dairy intake in order to raise serum vitamin B-12 concentrations. Our findings with respect to meat are in contrast to the findings in the Hordaland Homocysteine Study and the Framingham Offspring Study [10,11], which did not observe a significant impact of meat on circulating vitamin B-12 concentrations while median vitamin B-12 intake by meat intake was comparable (i.e.,˘1.3 µg/day). It may be speculated that this discrepancy with respect to the impact of meat intake in raising vitamin B-12 concentrations between our study and Hordaland Homocysteine Study and the Framingham Offspring Study relates to the overall higher vitamin B-12 intakes and higher intakes of fish and shellfish in the Hordaland Homocysteine Study and the Framingham Offspring Study compared to our study and the suggested maximal load of vitamin B-12 intestinal absorption system per meal (i.e.,˘1.5-2.0 µg/meal). With respect to dairy intake, our findings are in line with these previous studies, where the Framingham Offspring Study also suggested a significant impact of dairy-particularly milk as did the Hordaland Homocysteine Study [10,11]. Although the impact of fish and shellfish on vitamin B-12 concentrations in our population was smaller than the impact of meat and dairy, higher fish and shellfish intake was significantly associated with higher vitamin B-12 concentrations, indicating that persons with a vitamin B-12 intake of 1 µg/day as obtained by fish and shellfish had a 13% higher probability of having an adequate serum vitamin B-12 concentration compared to those not consuming any fish or shellfish. The fact that the impact of fish and shellfish intake and serum vitamin B-12 in our population was less strong than the impact of dairy and meat, while this was not the case in for instance the Hordaland Homocysteine Study, may relate to the fact that the overall vitamin B-12 intake from fish and shellfish in our population (e.g., for men, median (25th-75th percentile): 0.61 (0.26-1.20) µg/day) was substantially lower than the overall fish and shellfish intake in the Hordaland Homocysteine Study (e.g., for men aged 71-74 years old, median (25th-75th percentile): 3.4 (2.0-5.3) µg/day) [11]. Finally, the null findings in this study with respect to vitamin B-12 as obtained by egg intake are in line with the findings in the Hordaland Homocysteine Study [11]; in both studies vitamin B-12 intake as obtained by eggs was low and the range was narrow, which may explain the low impact on circulating vitamin B-12 concentrations of this product group.

Conclusions
Our data suggest that vitamin B-12 as obtained from dairy, meat, and fish and shellfish substantially contribute to raise vitamin B-12 concentrations in a Dutch population of older adults, where dairy-particularly milk and meat seem to be the most potent sources in order to prevent inadequate serum vitamin B-12 concentrations. Next to determining-and if possible treating-the cause of the vitamin B-12 deficiency, targeted recommendations focusing on these food groups may help to counteract low vitamin B-12 concentrations, especially in cases of modest vitamin B-12 deficiency. In cases of more severe vitamin B-12 deficiency other strategies may be warranted, including for instance recommending high dose oral vitamin B-12 supplements or vitamin B-12 injections.