In this cross-sectional analysis of an elderly population-based sample from northern Germany comprising n = 788 individuals, we assessed metabolic and lifestyle correlates of serum holo-TC, the biologically active form of vitamin B12, as well as its relation to intake of animal derived food products. We found serum holo-TC levels to be statistically significantly associated with total cholesterol, a parameter of liver function (namely ALAT), diastolic blood pressure, and alcohol consumption, as well as to be higher in individuals using vitamin B complex supplementation and in females. Furthermore, in a subsample excluding users of vitamin B complex supplementation, higher serum holo-TC was seen in males and females who consumed higher amounts of dairy products and in males who consumed higher amounts of fish and seafood, whereas intakes of eggs and meat and meat products were not associated with serum holo-TC.

Holo-TC is considered a good biomarker for monitoring the vitamin B12 status of populations [9] and to have better diagnostic accuracy in detecting vitamin B12 deficiency than circulating vitamin B12 [10, 20, 36, 37]. According to Hermann and Obeid [7], vitamin B12 deficiency is unlikely when serum holo-TC levels exceed 50 pmol/L. In our study sample, only 6.6% of participants fell below this threshold, whereas the middle 50% (interquartile range) of participants had serum holo-TC levels between 66.9 and 115.2 pmol/L. This indicates an overall good vitamin B12 status in our cohort. However, to conclusively detect functional vitamin B12 deficiency, biomarkers of cellular vitamin B12 status, namely homocysteine and methylmalonic acid, should be additionally considered [7, 9, 38], but were, unfortunately, not available in our cohort. Not surprisingly, regular use of supplements that contain vitamin B12 is associated with higher circulating holo-TC [4, 5, 15, 39, 40], an association that was also observed in our sample.

The only natural source of vitamin B12 is animal derived food products. In general, observational studies reported that intake of such food groups correlated well with blood levels of biomarkers of vitamin B12 status, namely total cobalamin (often referred to as vitamin B12), holo-TC, and methylmalonic acid [4,5,6, 22, 41,42,43]. As a marker of cellular vitamin B12 deficiency, methylmalonic acid generally correlated inversely with intake of animal derived food products [22], while circulating vitamin B12 and holo-TC mostly showed direct associations [5, 6, 22, 42]. However, the extent to which different animal derived food products correlate with these biomarkers varies across food groups and is not entirely consistent in the published literature.

In our sample, we observed (1) a direct association of intake of both dairy products (in males and females) and fish and seafood (in males) with serum holo-TC, and (2) that consumption of eggs and meat and meat products was not associated with circulating holo-TC. The direct association of vitamin B12 status with intake of dairy products is in line with most previous reports [4,5,6, 22, 41,42,43]. In one study (n = 728), however, milk consumption in non-vegetarians was directly associated with serum vitamin B12, but not with serum holo-TC [4]. By contrast, in n = 1266 pregnant women, plasma vitamin B12 as well as holo-TC were directly associated with intake of dairy products [22]. This highlights possible differences when relating food group intake to either circulating vitamin B12 or holo-TC, an observation that cannot be explained by the published literature thus far. However, this should be considered when comparing our results with those previously published as most available data is based on circulating vitamin B12.

With regard to fish and seafood consumption, two studies [6, 22] reported a direct association with vitamin B12 status, consistent with the observations in our sample. However, others showed a direct association with serum vitamin B12 only in individuals with low vitamin B12 status (n = 1230) [41] or reported an inverse association of fish intake and serum holo-TC (n = 728) [4].

Contradictory results on the association of circulating biomarkers of vitamin B12 status were also reported for the consumption of meat, the food group with the highest amounts of bioavailable vitamin B12 [44]. In line with our results, some studies showed no association with circulating vitamin B12 or holo-TC [6, 41], whereas others reported a direct association [4, 22, 42]. Potential explanations for the lack of association between meat consumption and vitamin B12 biomarkers [6, 41] include that (1) meal preparation, e.g. cooking and frying, might lead to destruction and loss of bioavailable vitamin B12 due to severe heat influence and water loss [45, 46] and (2) the relative absorption rate of vitamin B12 decreases with higher vitamin concentrations in the meal consumed due to saturation of vitamin B12 receptors [44]. In our cohort, absolute intake of meat and meat products was considerably larger compared to fish and seafood intake. Therefore, a higher relative uptake of vitamin B12 from fish in small portions than from meat in larger portions could help to explain the different associations between these food groups and serum holo-TC in our sample. However, as the accuracy of estimated daily intake amounts by use of a FFQ is limited [47], this point remains speculative. Interestingly, the direct impact of fish consumption on vitamin B12 status underlines the importance of this dietary component with regard to human health outcomes, especially when compared to a rather detrimental effect of high meat consumption [48].

In contrast to the lack of association between egg consumption and serum holo-TC observed in our cohort, two prior studies reported direct associations of egg consumption with holo-TC levels [4, 22]. However, it has been reported that vitamin B12 from eggs is generally poorly absorbed in comparison to other food groups [3, 49].

The new food-based dietary guidelines recently published by the German Nutrition Society emphasize the regular consumption of fish (1–2 portions/week) and dairy products (2 portions/day), while concomitantly recommending a reduced consumption of meat (max. 300 g/week) and eggs (1 egg/week) [50]. As nutrient goals were one of the key aspects incorporated into these recommendations [51], these food-based dietary guidelines support the results of our analyses, namely that we observed no association between serum holo-TC and consumption of meat or eggs, but a direct association with dairy products and fish and seafood intake.

To the best of our knowledge, this is the first study to report metabolic and lifestyle correlates of serum holo-TC in a population-based sample with adequate vitamin B12 status. We identified total cholesterol, a liver function parameter (ALAT), sex, alcohol consumption, and vitamin B complex supplementation as key correlates of serum holo-TC concentrations. Further, diastolic blood pressure was inversely associated with serum holo-TC in the overall analyses, yet was not identified as a correlate in the sensitivity analysis excluding individuals with holo-TC concentrations > 150 pmol/L, which calls into question the reliability of this finding. Moreover, in a population-based sample from the Netherlands, plasma vitamin B12 concentrations were not associated with diastolic or systolic blood pressure [16]. Age, an essential covariate in epidemiologic analyses, was not associated with serum holo-TC in our sample. This contrasts the results of prior studies. Some studies report a higher vitamin B12 status with increasing age [10, 52], while others report this association to be inverse [53, 54].

Females from our analytical sample showed higher serum holo-TC levels than males, which is in accordance with observations in participants aged 60–79 years from a Swiss study [55]. Likewise, in a subsample from NHANES, elderly men had lower serum vitamin B12 and a higher prevalence of vitamin B12 deficiency than women [56]. Contrastingly, healthy female blood donors had lower holo-TC than men, although this association was restricted to individuals ≤ 45 years [57].

Vitamin B12 is involved in lipid metabolism with a low status of this vitamin being associated with dyslipidemia, mostly manifesting itself in elevated total and LDL cholesterol [58]. In a hepatic cell culture model, low vitamin B12 conditions increased fatty acid synthesis leading to higher intracellular fatty acid concentrations [59]. Consistently, in women, a low vitamin B12 status was associated with higher LDL and total cholesterol [60]. However, other studies could not find such associations [18, 19]. We observed a direct association between total cholesterol and holo-TC, which is in line with some of these reports, while other biomarkers of the lipid metabolism were not statistically significantly associated with serum holo-TC in our sample.

We also observed a direct association between serum holo-TC and ALAT, a biomarker of liver function. Similarly, in a population-based sample from the Netherlands (n = 5571), ALAT showed a direct correlation with plasma vitamin B12 [16]. By contrast, in a small (n = 118) cross-sectional study of adults with cardiometabolic risk factors, ALAT was not associated with vitamin B12 concentrations [17]. In clinical samples of patients with alcohol induced hepatocellular damage, higher levels of liver enzymes, including ALAT, were associated with generally higher serum vitamin B12 as it leaked out of the damaged liver tissue [61]. Furthermore, alcoholism was associated with very high levels of serum vitamin B12 in hospital-treated adults [62]. In contrast, in an overall healthy sample from NHANES, only minimal changes in vitamin B12 concentrations with higher alcohol consumption were seen [15], which is in line with the slight inverse association between serum holo-TC and alcohol consumption in our sample. By contrast, in a population-based sample from Denmark, no association between serum vitamin B12 and alcohol consumption was observed [63]. Based on these observations it could be inferred that high alcohol consumption might have an impact on the vitamin B12 metabolism, while alcohol consumption in moderation, as seen in our sample, may have a negligible association with the vitamin B12 status.

Impaired kidney function as indicated by higher creatinine levels, was identified to increase holo-TC levels [10, 20], possibly due to reduced renal holo-TC filtration [20]. In our analytical sample, serum creatinine was not identified as a correlate of serum holo-TC and, therefore, not included in the final linear regression model. However, creatinine levels were within the normal range in our study sample, which might have impeded a statistically significant association between serum holo-TC and this kidney function parameter. Likewise, in an elderly cohort (n = 1209) with low incidence of increased creatinine, holo-TC was not associated with creatinine [64]. Furthermore, creatinine within the reference range was not suggested as a determinant of holo-TC in healthy adults (n = 500) [57]. Contrastingly, in elderly Swedish men (n = 1010), normal-range creatinine levels showed a weak direct correlation with serum holo-TC, while elevated creatinine showed no association with holo-TC [65].

In general, heterogeneity in study samples and the predominant use of circulating vitamin B12 instead of holo-TC in the literature hinders comparability of our results. Furthermore, metabolic and lifestyle correlates explained only a moderate fraction (11.9%) in the interindividual variation of serum holo-TC levels, indicating that other factors might have a considerable impact on holo-TC levels, including, e.g., diet (in particular dairy products and fish and seafood), genetic variation [66], and the gut microbiota [67].

Strengths of our analyses include the population-based and well-characterized study sample allowing for consideration of a broad variety of metabolic and lifestyle factors. A limitation of the holo-TC measurement (conducted in a professional laboratory) was that the upper detection limit was at 150 pmol/L, which is in contrast to other studies reporting holo-TC values well above this threshold [4, 36, 68]. Setting all individuals above this limit uniformly to 151 pmol/L possibly weakens the results reported here as we could not account for higher serum holo-TC indicating an even better vitamin B12 status. Still, excluding these individuals in sensitivity analyses did hardly change the presented results. Furthermore, in our analyses we only included serum holo-TC, the biologically active form of vitamin B12, as a biomarker of vitamin B12 status and did not measure biomarkers of cellular vitamin B12 deficiency, namely methylmalonic acid or homocysteine [7, 9, 38], and did not take into account other metabolites closely linked to the vitamin B12 metabolism, such as folate [8], which constitutes an important limitation. Another possible limitation is the ascertainment of dietary intake using an FFQ. An FFQ is an established tool in epidemiological studies to ascertain dietary intake over a longer period of time. However, as such, it is prone to misreporting and the daily dietary intake can only be calculated approximately [47]. In general, the backward selection process applied in our study is a data-driven exploratory approach, and as such, chance findings cannot be ruled out. Finally, the study sample of this cross-sectional study comprised elderly individuals of a specific geographical location in northern Germany, which limits the generalizability of our findings.