Does high folic acid intake affect unrecognized cobalamin deficiency, and how will we know it if we see it?

Twelve years of mandatory food fortification with folic acid (FA), during which widespread self-supplementation pushed FA intake even higher, have nurtured questions about risks to persons with cobalamin deficiency but generated little meaningful data until just recently. In 2007 analysis of National Health and Nutrition Examination Study (NHANES) data linked higher folate concentrations with worse methylmalonic acid (MMA) and homocysteine indexes of metabolic deficiency in cobalamindeficient adults (1). A companion analysis of an elderly NHANES cohort indicated that cognition and, curiously, anemia were also worse in that setting, whereas higher folate concentrations were associated with better cognition when cobalamin was normal (2). In this issue of the Journal, Miller et al (3) provide important confirmation of the metabolic association, but not the worse cognition, in the Sacramento Area Latino Study on Aging (SALSA). An intervening report from Oxford found no metabolic, cognitive, or anemia associations (4). Its lack of statistical significance may reflect too few folate values above the reference interval (4.9% in its total study population compared with 26.3% in SALSA) because fortification is not mandatory and self-supplementation is limited in the United Kingdom. In fact, mean MMA and homocysteine values were also considerably higher in the study’s low-cobalamin/high-folate subjects than in those with normal folate. Interestingly, the standard deviations either approached or considerably exceeded their metabolite means in the low-cobalamin/high-folate group, which suggested that the group was metabolically diverse. Similarly wide metabolite distributions characterized the low-cobalamin/high-folate groups in NHANES and SALSA data (1, 3). The apparent skew may hold important clues to the still undetermined meaning of the metabolic associations. The NHANES and SALSA reports hypothesized that FA causes oxidative damage tocobalaminmetabolism.However,considerableprecedent favors a reversed causation, if the association is indeed causal: advanced cobalamin deficiency, with its methyltetrahydrofolate accumulation, typically raises folate concentrations by 20–30% (5). The high red cell folate values (3), which should decline in cobalamin deficiency as red cell methyltetrahydrofolate exits to plasma (5), seemed incompatible but, despite high folate intakes, they were .200 nmol/L lower on average than in normal-cobalamin/highfolate subjects. The wide spread of MMA and homocysteine values that characterized only the low-cobalamin/high-folate group in all 3 studies (1, 3, 4) suggests that this group alone contains subsets, one of which is probably severely cobalamin deficient, whereas another may have no deficiency, as can often occur despite abnormal cobalamin parameters (6). Pernicious anemia, although infrequent, is the predominant cause of clinically expressed cobalamin deficiency (5) and may plausibly explain the severely deficient group, whose 1.4% proportion in SALSA approximates the American frequency of pernicious anemia (7). Focused analyses may help. The 3 studies’ disparate clinical outcomes, however, cannot be reconciled at present. The studies differed in many structural ways, including duration and extent of their populations’ exposure to high folate intake. The studies also had only 1–2 cognitive tests each to analyze, all of which were dissimilar. A quandary in folate-cobalamin exploration is that observational studies cannot prove mechanisms but trials involving FA are subject to ethical constraints in cobalamin-deficient subjects. Alternative ways must be found to extract answers and resolve whether a public health problem requiring intervention exists. Future studies must identify in advance and collect all clinical and test data necessary for reliable characterization; data borrowed from broad multipurpose surveys often lack sufficient depth. The collected information must especially include optimal neurologic as well as cognitive testing. Statistical analyses can be enhanced with more clinically and pathophysiologically appropriate evaluations of group and individual data that may identify obscured subsets and reconfigured answers. Studies may also consider incorporating focused therapeutic trials with cobalamin. All this is a large task, but anything less is unlikely to suffice. The surveys must also be very large because the proportion of informative cases is typically minuscule and can inhibit stratification and subset analyses. The 1535 SALSA subjects, for example, yielded only 100 subjects (6.5%) with low cobalamin concentra-