In B12 deficiency, higher serum is associated with increased total homocysteine and methylmalonic acid concentrations

Jacob Selhub*†, Martha Savaria Morris‡, and Paul F. Jacques‡

*Vitamin Laboratory and ‡Nutritional Epidemiology Program, Jean Mayer United States Department of Agriculture Human Nutrition Research Center on Aging, Tufts University, Boston, MA 02111

Communicated by Nevin S. Scrimshaw, United Nations University, Thornton, NH, October 18, 2007 (received for review June 1, 2007) In a recent study of older participants (age >60 years) in the United States National Health and Nutrition Examination Survey 1999–2002 National Health and Nutrition Examination Survey (NHANES) (11). As expected, low vitamin B12 status was associ- (NHANES), we showed that a combination of high serum folate and ated with all three abnormalities. Furthermore, when vitamin B12 low vitamin B12 status was associated with higher prevalence of status was normal, high serum folate was related to protection from cognitive impairment and anemia than other combinations of cognitive impairment. Among those with low vitamin B12 status, vitamin B12 and folate status. In the present study, we sought to however, high serum folate was directly associated with both determine the joint influence of serum folate and vitamin B12 anemia and cognitive impairment. Consistent with our findings for concentrations on two functional indicators of vitamin B12 status, subjects with normal vitamin B12 status, several cross-sectional total homocysteine (tHcy) and methylmalonic acid (MMA), among studies (12–15), as well as some prospective investigations (16–19), adult participants in phase 2 of the NHANES III (1991–1994) and the have reported direct relations between folate status and cognition. NHANES 1999–2002. Exclusion of subjects who were <20 years old, On the other hand, Morris et al. (20) found that higher folate intake were pregnant, had evidence of kidney or liver dysfunction, or was associated with more rapid cognitive decline in a cohort of reported a history of alcohol abuse or recent anemia therapy left Ϸ4,000 community-dwelling elderly people whose vitamin B12 4,940 NHANES III participants and 5,473 NHANES 1999–2002 par- status was unknown. ticipants for the study. Multivariate analyses controlled for demo- In India, where vitamin B12 deficiency is common, authors of graphic factors, smoking, alcohol use, body mass index, self- the Pune Maternal Nutrition Study found that the combination reported diabetes diagnosis, and serum concentrations of of high blood folate and low vitamin B12 levels during pregnancy creatinine and alanine aminotransferase revealed significant in- was associated with elevated risk of insulin resistance in the teractions between serum folate and serum vitamin B12 in relation children at age six (21). Insulin resistance is not an established to circulating concentrations of both metabolites. In subjects with consequence of vitamin B12 deficiency, but the wide-ranging serum vitamin B12 >148 pmol/liter (L), concentrations of both known effects of the vitamin suggest that it also may have metabolites decreased significantly as serum folate increased. In unknown benefits, and its deficiency may have unknown subjects with lower serum vitamin B12, however, metabolite con- Ϸ risks (22). centrations increased as serum folate increased starting at 20 Our goal in the present study was to determine whether higher nmol/L. These results suggest a worsening of vitamin B12’s enzy- folate status in vitamin B deficiency was associated with higher matic functions as folate status increases in people who are vitamin 12 circulating levels of the functional indicators of low vitamin B B -deficient. 12 12 status, total (t) Hcy, and MMA.

NHANES ͉ cobalamin ͉ methionine synthase ͉ L-methylmalonyl-CoA mutase Results

Subject Characteristics and Serum Vitamin B12. The prevalence of itamin B12 participates as a coenzyme in two reactions, the serum vitamin B12 Ͻ148 pmol/liter (L) was 1.6% in the Vremethylation of homocysteine (Hcy) to methionine and the NHANES III and 2.2% in the NHANES 1999–2002. The isomerization of L-methylmalonyl-CoA to succinyl-CoA (1–3). demographic distribution of the subjects reflected the national The former reaction is catalyzed by methionine synthase in the representation of the survey and our focus on adults of all ages cytosol, and the latter is catalyzed by L-methylmalonyl-CoA (Table 1). Supplement use was much more common during the mutase in the mitochondria. Vitamin B12 deficiency is associated more recent survey, when serum folate was also much higher, with anemia, megaloblastosis, neuropathy, and neuropsychiatric and tHcy was lower. In both surveys, low serum vitamin B12 was disorders (4), all of which could theoretically be linked to the significantly less common among non-Hispanic blacks than it was block in Hcy remethylation or the resulting hyperhomocysteine- in non-Hispanic whites. Also consistent across surveys were mia (5). However, the block in the isomerization reaction and strong associations between low serum vitamin B12 status and resulting buildup of methylmalonic acid (MMA) could also play nonuse of dietary supplements, low folate status, and high tHcy a role (6). and MMA concentrations as well as at least marginally signifi- Early case reports of vitamin B12 deficiency due to pernicious cant associations with higher age, cigarette smoking, and lower anemia described the alleviation of anemia but the precipitation alanine aminotransferase concentration. A notable difference or exacerbation of neurologic or neuropsychiatric sequelae after folic acid administration, an early form of treatment based on the mistaken idea that folate deficiency was the problem (7). How- Author contributions: J.S., M.S.M., and P.F.J. designed research; M.S.M. analyzed data; and ever, a lack of confirmatory data from animal models or human J.S., M.S.M., and P.F.J. wrote the paper. studies has generated controversy about the true consequences The authors declare no conflict of interest. of folic acid administration in vitamin B12 deficiency (8), which Freely available online through the PNAS open access option. is now known to encompass more than pernicious anemia (9, 10). †To whom correspondence should be addressed at: Vitamin Metabolism Laboratory, Jean We recently examined the interaction between serum folate and Mayer U. S. Department of Agriculture Human Nutrition Research Center on Aging, Tufts vitamin B12 status in relation to anemia, macrocytosis, and cognitive University, 711 Washington Street, Boston, MA 02111. E-mail: [email protected]. MEDICAL SCIENCES impairment in older participants (age Ն60 years) in the 1999–2002 © 2007 by The National Academy of Sciences of the USA

www.pnas.org͞cgi͞doi͞10.1073͞pnas.0709487104 PNAS ͉ December 11, 2007 ͉ vol. 104 ͉ no. 50 ͉ 19995–20000 Downloaded by guest on September 26, 2021 Table 1. Characteristics of eligible adult participants in the NHANES III and the NHANES 1999–2002 by serum vitamin B12 concentration* NHANES III NHANES 1999–2002

Serum B12, pmol/L Serum B12, pmol/L

Ն148 Ͻ148 Ն148 Ͻ148 Characteristic n ϭ 4,823 n ϭ 117 OR (95% C.I.) P value n ϭ 5,343 n ϭ 130 OR (95% C.I.) P value

Age, years 45 Ϯ 1.0 53 Ϯ 4.0 1.1 (1.0–1.2)† 0.057 47 Ϯ 0.4 51 Ϯ 1.6 1.1 (1.02–1.13)† 0.016 Female, % 57 64 1.4 (0.7–2.6) 0.640 74 70 1.7 (1.04–2.9) 0.018 Nonhispanic white, % 82 79 1.0 Referent 57 83 1.0 Referent Nonhispanic black, % 8.4 2.7 0.3 (0.2–0.8) 0.011 9.6 3.7 0.4 (0.2–0.8) 0.015 Mexican-American, % 4.3 4.7 1.1 (0.5–2.4) 0.745 6.2 4.6 0.7 (0.4–1.3) 0.236 Cigarette smoker, %‡ 25 32 1.9 (0.9–4.0) 0.096 25 31 1.7 (1.03–2.7) 0.037 Supplement user, %‡ 38 16 0.3 (0.1–0.8) Ͻ0.015 55 49 0.6 (0.3–0.95) 0.044 Alcohol per day, g‡ 10.0 9.2 1.0 (1.0–1.01) 0.711 8.7 14.8 1.0 (1.0–1.01) 0.147 Creatinine, ␮mol/L‡§ 71 Ϯ 0.3 74 Ϯ 1.6 1.0 (1.0–1.03) 1.0 72 Ϯ 0.8 69 Ϯ 1.9 1.0 (1.0–1.01) 0.363 Alanine aminotransferase, 15.2 Ϯ 0.2 12.7 Ϯ 0.8 0.9 (0.9–1.0) 0.057 20.9 Ϯ 0.02 17.8 Ϯ 0.8 0.4 (0.2–0.8) 0.001 units/L‡§ Body mass index¶ 26.4 26.6 1.0 (1.0–1.0) 0.927 27.7 29.1 1.04 (1.02–1.2) 0.002 Serum folate, nmol/L†§ʈ 15.3 Ϯ 1.0 10.6 Ϯ 0.8 0.94 (0.92–0.96) Ͻ0.001 30.8 Ϯ 0.5 25.0 Ϯ 1.2 0.84 (0.8–0.9) Ͻ0.001 Hcy, ␮mol/L§ʈ 8.5 Ϯ 0.1 13.2 Ϯ 2.2 1.1 (1.1–1.2) Ͻ0.001 7.7 Ϯ 0.1 10.7 Ϯ 0.8 1.2 (1.1–1.2) Ͻ0.001 MMA, nmol/L§ʈ** 189 Ϯ 7.5 317 Ϯ 134 1.2 (0.99–1.3) Ͻ0.078 132 Ϯ 1.6 245 Ϯ 47 1.4 (1.3–1.5) Ͻ0.001

*Subjects Ͻ20 years old, those with high serum creatinine or alanine amino-transferase concentrations, pregnant women, and those who reported heavy alcohol use, a history of diabetes, or recent anemia therapy were excluded. Table values are means (Ϯ SEM), sample-weighted percentages. Odds ratios (OR) for continuously scaled factors are odds of low serum vitamin B12 per unit in the characteristic. †Per five years of age. ‡After multivariate adjustment for age, race/ethnicity, and gender. §Geometric mean (Ϯ SEM). ¶After multivariate adjustment for age, race/ethnicity, gender, cigarette smoking, alcohol intake, and serum concentrations of creatinine and alanine amino-transferase. ʈAfter multivariate adjustment for age, race/ethnicity, gender, cigarette smoking, alcohol intake, and serum concentrations of creatinine, serum alanine amino-transferase, and body mass index. **Per 50-unit increase. MMA was measured in the NHANES III in surplus serum samples from subjects Ն60 years old and from a sample of subjects 30–39 years old.

between the results of the two surveys pertained to body mass served among subjects with serum vitamin B12 Ͼ148 pmol/L index, which was significantly inversely associated with serum (P Ͻ 0.001). Although no significant trends emerged for partic- vitamin B12 in the more recent NHANES only. ipants in the NHANES III, the data for subjects in the two surveys with serum vitamin B12 Ͻ148 pmol/L suggested a Serum Folate and tHcy and MMA. Among participants in both decrease in serum MMA as serum folate increased up to a point, surveys, serum folate interacted significantly with serum vitamin but highly elevated MMA concentrations at serum folate con- Ͻ B12 in relation to tHcy (NHANES III, P 0.002; NHANES centrations Ͼ19.3 nmol/L. 1999–2002, P ϭ 0.005). Specifically, although geometric mean Hcy decreased across increasing serum folate categories among Discussion Ͼ survey participants in both surveys with serum vitamin B12 148 Associations between low vitamin B status and both hyperho- Ͻ 12 pmol/L (P 0.001), a decreasing trend across all serum folate mocysteinemia and elevated MMA concentrations are well categories was not observed for subjects with serum vitamin B12 established, as is an inverse relation between folate status and Ͻ148 pmol/L (Fig. 1). The highest geometric mean tHcy that we estimated was for the NHANES III participants with serum Ͻ vitamin B12 148 pmol/L and serum folate in the low-to- SENAHN II I SENAHN 2-9991 200 deficient range. Furthermore, geometric mean tHcy for 005 1-B 2<148 mp o /l L 05 0 NHANES 1999–2002 subjects with low serum vitamin B12 and 400 400 serum folate Ͻ19.3 nmol/L was in the normal range and close to 03 0 300 B 1- 2 41< 8 lomp L/ that of subjects in the same serum folate category with higher 200 B >21- 841 p L/lom 200 B 21- >1 84 pmol/L MMA (nmol/L) MMA MMA (nmol/L) MMA 001 100 serum vitamin B12 (Table 2). However, the mean for the group 0 0 with low serum vitamin B12 and serum folate Ͼ32.6 nmol/L was 10 0 20 403 0 50 02010 403 050 nearly 12 ␮mol/L, a commonly used cut-off point for hyperho- mureS of etal ( lomn L/ ) Se mur talof e ( /lomn L) Ͻ mocysteinemia. In fact, in subjects with serum B12 148pmol/L, AHN N SE III HN AN SE 1999-20 20 tHcy increased significantly across the four serum folate cate- 02 02 gories represented in the NHANES 1999–2002 (P ϭ 0.003). 51 15 omp 841<21-B omp l L/ B-12 1< 48 p lom /L Serum folate interacted significantly with serum vitamin B12 01 01 mol/L) BL>21- 841 pmo /l mol/L)

μ 21-B > 841 omp /l L μ ( in relation to serum MMA concentrations only for participants 5 ( 5 ϭ Serum homocysteine Serum in the NHANES 1999–2002 (P 0.003). Among participants in 0 homocysteine Serum 0 0 100 2 300 4 50 that survey with serum vitamin B12 Ͻ148 pmol/L, geometric 001 200 304 50 mean serum MMA increased significantly across categories of S)mure of l eta L/lomn( S)re mu f lo ate ( L/lomn increasing serum folate (P ϭ 0.008), while a modest but signif- Fig. 1. Trends in serum MMA and Hcy concentrations with increasing serum icant decrease in MMA with increasing serum folate was ob- folate in the United States population over the age of 20.

19996 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0709487104 Selhub et al. Downloaded by guest on September 26, 2021 Table 2. Least-square geometric mean (95% C.I.) Hcy and MMA concentration by serum vitamin B12 and serum folate category,* NHANES Hcy MMA†

Serum B12, pmol/L Serum B12, pmol/L Serum folate, nmol/L Ͻ148 Ն148 P value Ͻ148 Ն148 P value

1991–1994 7 15.5 (13.2–18.1) 10.9 (10.5–11.3) Ͻ0.001 347 (178–676) 202 (180–228) 0.194 11 12.2 (10.4–14.2) 9.2 (8.8–9.2) Ͻ0.001 290 (155–543) 191 (176–206) 0.140 23 11.9 (8.7–16.3) 7.6 (7.5–7.8) 0.009 523 (333–821) 191 (180–202) 0.002 1999–2002 16 9.9 (8.8–11.1) 9.2 (9.0–9.4) 0.222 175 (149–204) 138 (133–145) 0.005 23 10.8 (9.4–12.4) 8.0 (7.8–8.2) Ͻ0.001 265 (218–322) 137 (132–142) Ͻ0.001 30 10.5 (9.1–12.0) 7.5 (7.4–7.7) 0.001 265 (194–363) 132 (127–137) Ͻ0.001 44 11.8 (10.1–13.8) 7.1 (7.0–7.2) Ͻ0.001 314 (234–421) 128 (125–130) Ͻ0.001

Table values were generated by using SUDAAN PROC REGRESS after controlling for age, race/ethnicity, sex, cigarette smoking, alcohol intake, body mass index, self-reported diabetes status, and serum concentrations of creatinine and alanine amino-transferase. *Serum folate values in the table are category medians. †Hcy and MMA were measured in surplus serum samples in phase 2 of the NHANES III. MMA concentrations were measured in surplus serum samples from participants Ն65 years old and from a sample of participants 30–39 years old.

circulating Hcy under normal circumstances. A possible link (2). In the enzyme catalysis, the cob(I)alamin acquires a methyl between folate status and MMA has not been considered, group from bound 5-methyltetrahydrofolate by means of an because folate is not known to participate in pathways affecting oxidation addition mechanism forming a methyl–cob(III)alamin MMA accumulation. In this study of the general United States prosthetic group (24). This bond between the methyl group and population spanning pre- and postfolate fortification eras, we the cobalt ion is then cleaved heterolytically, leaving both found that tHcy does not generally decrease with increasing bonding electrons on the cob(I)alamin and transferring the folate status among people with low serum vitamin B12 concen- methyl group to homocysteine in the form of a carbocation, ϩ trations. On the contrary, in people with low serum B12 con- CH3 (24). Cob(I)alamin is remethylated by 5-methyltetrahy- centrations, high tHcy levels were found at both very low and drofolate. However, the enzyme-bound cob(I)alamin occasion- very high serum folate levels, with lower tHcy levels in between. ally becomes oxidized to cob(II)alamin, and this enzyme form is Furthermore, whereas MMA data for people with low levels of catalytically inactive. The enzyme is reactivated by a reductive both vitamin B12 and folate were limited, the MMA results were methylation in which the methyl group is derived from adeno- consistent with the tHcy results in revealing a narrow range of sylmethionine (25). folate values associated with near-normal metabolite levels and In the mitochondria, the cob(II)alamin undergoes a reduction elevated metabolite levels among those with low and high folate to cob(I)alamin by an electron donor followed by adenosylation concentrations. in a reaction that involves ATP, to form the 5Ј-deoxyadenosyl- In vitamin B12 deficiency, high tHcy reflects impaired methi- cob(III)alamin () (1, 2, 23). This is then onine synthase activity, whereas high MMA indicates impaired associated with the mutase by the action of adenosyltransferase. methylmalonyl-CoA mutase activity. Consequently, our findings A mitochondrial , methylmalonic aciduria cblA type suggest that both pathways of vitamin B12 metabolism are protein (MMAA), associates with the mutase to protect the adversely affected by high serum folate despite the direct mutase from inactivation (26). involvement of folate only in methionine synthase activity. In the MMA–CoA mutase reaction, the bond between the Simultaneous impairment of both pathways is normally caused methylene group of 5Ј-deoxyadenosine and the cobalt ion un- by vitamin B12 deficiency or by the disruption of those early steps dergoes homolytic cleavage, thus resulting in free radicals both in vitamin B12 processing that are common in both pathways (1, on the methylene group and the cobalt ion. The free radical on 23): (i) the intrinsic factor mediated B12 absorption by the the methylene group is transferred to the MMA–CoA molecule intestine; (ii) formation of the B12-transcobalamin II complex in to allow isomerization to succinyl-CoA, which is accompanied by plasma; (iii) transport of the complex into peripheral tissue by a the transfer of the free radical back onto the methylene group for receptor-mediated endocytosis and incorporation of the com- subsequent rebonding with the cobalt ion (1, 2, 23). plex into the lysosomes; (iv) dissociation of the complex and exit Although the reduction of cob(II)alamin to cob(I)alamin of vitamin B12, in the form of acquocob(III)alamin, into the occurs in both compartments, the process is substantially dif- cytosol; and (v) reduction to cob(II)alamin for subsequent ferent for the two. In the cytosol, this reduction takes place while migration either for association with the methionine synthase cob(II)alamin is bound to the enzyme. In the , on (MS) in the cytosol or to enter the mitochondria. the other hand, this reduction precedes the association of the It is difficult to envision how folate could interfere with any mutase. This difference and the fact that these reductions occur of these steps. The alternate possibility is that this interference in separate compartments render it difficult to envision inter- occurs after the diversion of the vitamin to the respective ference by folate. compartments in the course of rendering the vitamin available Another possibility is that folate interferes with the catalytic for interaction with the respective enzyme or during action of MS, and the impairment of the mutase is secondary to catalysis. In the cytosol, the cob(II)alamin associates with the the interaction with the MS. This possibility would be analogous MS to form a cob(II)alamin–MS complex. This complex is then to the observed effects of nitric oxide on B12 metabolism. Nitric reduced by a closely associated enzyme, methionine synthase oxide is a strong oxidizing agent that irreversibly inactivates reductase (MTR) to form a cob(I)alamin–MS that is catalytically methionine synthase by oxidizing enzyme-bound vitamin B12 to active. The MS enzyme contains a binding site for 5-methyltet- form a free radical (27). There is no known direct effect of rahydrofolate, which is more specific for polyglutamyl derivatives nitrous oxide on the mutase. Nevertheless, both humans who MEDICAL SCIENCES

Selhub et al. PNAS ͉ December 11, 2007 ͉ vol. 104 ͉ no. 50 ͉ 19997 Downloaded by guest on September 26, 2021 inhaled the gas and laboratory animals subjected to prolonged cognitive impairment (11) by demonstrating that, among people exposure had elevated plasma and spinal fluid levels of not only with low serum vitamin B12 concentrations, high plasma folate tHcy but also MMA (28–32). is associated with higher concentrations of the two functional Using a methionine-independent human glioma cell line indicators of impaired B12 status, Hcy and MMA. These obser- (P60H), Reidel et al. (33) showed that exposure to nitrous oxide vations provide a possible biochemical explanation for high folic resulted in cellular depletion of methylcobalamin and, after a lag acid intake’s exacerbation of the clinical manifestations of period, decreased adenosylcobalamin and conversion of the vitamin B12 deficiency. mutase holoenzyme to its apoenzyme. These data suggest a coordinate distribution of cobalamin cofactors between the Methods mitochondria and the cytosol compartments. Study Population. The NHANES, which is currently implemented High serum folate as encountered in this study was attribut- as a continuous annual survey, monitors the nation’s health and able to high intake of folic acid from supplements and fortified nutritional status by using a complex multistage probability design food, which undoubtedly resulted in the appearance of unme- to select a representative sample of the noninstitutionalized United tabolized folic acid in plasma (34, 35). In the Framingham Study, States civilian population. To increase the precision of estimates the percent of folate consumed as folic acid increased sharply as derived from the survey, adolescents, the elderly, Mexican Amer- plasma folate increased, from 50% among those with plasma icans, and blacks are oversampled. The protocols for conducting the folate Յ30 nmol/L to 80% among those with plasma folate Ͼ50 NHANES were approved by the institutional review board of the nmol/L (S. Choumenkovitch, P.F.J., and J.S., unpublished work). National Center for Health Statistics, Centers for Disease Control A high intake of folic acid in vitamin B12 deficiency would and Prevention, and informed consent was obtained from all change the concentration and character of cellular folate, par- participants (44). For this investigation, we used data from phase 2 ticularly tetrahydrofolate (36). A central hypothesis to account (1991–1994) of the third NHANES (NHANES III) and the for the observed effect of folic acid intake in the remission of the NHANES conducted in 1999–2000 and 2000–2002. Consistent hematological complications in pernicious anemia is that the flux with NHANES analytic guidelines, we combined data from these of unmethylated folate into the bone marrow cells circumvents latter two surveys into a single dataset (45). the inhibition of DNA synthesis caused by the trapping of folate Trained interviewers used a computer-assisted personal in- coenzymes as methyltetrahydrofolate (37). Folate coenzymes terview system to interview participants in their homes. The participating in DNA synthesis include formylated tetrahydro- participants were also asked to report to a mobile examination folate, which is used for purine synthesis, and 5,10- methylenetet- center (MEC) to provide further interview data and undergo a rahydrofolate, which is used for thymidylate synthesis. Unsub- physical examination that included phlebotomy. A detailed stituted tetrahydrofolate and dihydrofolate are the respective description of blood collection and processing can be found in products of these syntheses. the NHANES Phlebotomy Manual (46). Although the survey A common denominator for some of these is that they included people of all ages, we focused our attention on adults can serve as electron acceptors or oxidation agents. Formylated (i.e., those aged Ͼ19 years). tetrahydrofolate can be reduced to 5,10- methylenetetrahydro- Besides participants Ͻ20 years old, we excluded pregnant folate; 5,10- methylenetetrahydrofolate can be reduced to women and men and women who reported a history of heavy 5-methyltetrahydrofolate, whereas folic acid and dihydrofolate alcohol use or recent anemia therapy as well as those with serum acquire electrons for conversion to tetrahydrofolate. Based on creatinine concentration (based on Jaffe reaction) indicative of these properties, it is tempting to suggest that folic acid or its renal dysfunction (i.e., men, Ͼ131 ␮mol/L; women, Ͼ115 partially reduced form, dihydrofolate, as well as formylated and ␮mol/L) or alanine aminotransferase Ͼ40 units/L, which sug- methylene tetrahydrofolate could act as electron acceptors and gests liver disease (47). Of the 8,588 adult participants in phase thereby accelerate the rate of oxidation of MS cob(I)alamin to 2 of the NHANES III, 4,940 were included in this study (Fig. 2). MS cob(II)alamin. The enzyme activity could be even more Of the 10,291 adult participants in the NHANES 1999–2002, compromised because of the lower S-adenosylmethionine 5,473 were included in this study. (SAM) availability and lower enzyme synthesis due to B12 deficiency (38, 39). Biochemical Measurements. Priority analyses were carried out at This proposed interaction between folate derivatives (folic the Inorganic Toxicology and Nutrition Branch or the Division acid, dihydrofolate, etc.) and a heterologous protein (i.e., MS) is of Laboratory Sciences, National Center for Environmental not unprecedented: 5-methyltetrahdrofolate is an activator of Health, or at central laboratories that varied with survey years. cystathionine ␥-synthase in Neurospora crassa (40), an inhibitor Serum concentrations of folate and vitamin B12 were measured of glycine methyltransferase in mammalian systems (41), and by using the Quantaphase II Radioassay Kit (Bio-Rad). Serum binds to cytosolic serine hydroxymethylase (42). Binding to alanine aminotransferase was measured by using an autoana- serine hydroxymethylase is associated with diminished synthesis lyzer (NHANES III, Hitachi; NHANES 1999–2001, Hitachi; of methionine and diversion of the flow of one carbon units NHANES 2002, Beckman; Synchron LX20). In all survey years, toward thymidylate synthesis. Dihydrofolate is a competitive serum creatinine concentration was based on the Jaffe reaction, inhibitor of pig liver methylenetetrahydrofolate reductase but values varied greatly among the NHANES III, NHANES (MTHFR) (43). In all cases, these associations are enhanced 1999–2000, and NHANES 2001 and 2002. Also, only the values when the respective folates are polyglutamated, which is con- reported for the last 2-year period were consistent with results sistent with the idea that the great majority of folate enzymes using a gold-standard method (46). After Coresh et al. (48), we have a common binding site for the glutamic acid side chain. As subtracted 20 ␮mol/L (0.23 mg/dl) from the NHANES III values. pointed out above, MS also contains a binding site for methyl- We also added 11.5 ␮mol/L (0.13 mg/dl) to the NHANES tetrahydrofolate and its product, tetrahydrofolate (2). 1999–2000 values as recommended by NHANES documenta- Based on this paradigm, we propose that the high serum MMA tion (46). associated with the combination of low B12 status and high Concentrations of MMA and tHcy were measured in plasma folate status stems from the disruption of B12 hemostasis NHANES III as phase 2 (1991–1994) surplus serum projects in the mitochondria, analogous to the effect of nitric oxide. carried out at the United States Department of Agriculture In conclusion, we have extended the findings of our recently Human Nutrition Research Center on Aging after approval by published study of the interaction between vitamin B12 status the New England Medical Center Human Investigations Review and folate status in relation to anemia, macrocytosis, and Committee and the Surplus Sera Bank Steering Committee.

19998 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0709487104 Selhub et al. Downloaded by guest on September 26, 2021 Fig. 2. Creation of study datasets from NHANES datasets.

Surplus serum samples were stored at Ϫ70°C and underwent one et al. (53) attributed the different MMA values obtained in the to four freeze–thaw cycles before tHcy and MMA analyses. Total two surveys to differences in the laboratories conducting the Hcy and MMA concentrations have been shown to remain stable analyses and the matrices and methods used. in long-frozen samples (49, 50). In 1994, Hcy concentrations were measured in serum samples from participants Ն12 years old Statistical Analyses. Data analyses were performed by using by the HPLC method of Araki and Sako (51). In 1998, MMA SUDAAN release 9.0 (Research Triangle Institute) with appro- concentrations were measured in surplus serum samples from priate sampling weights and masked variance pseudostratum and 65-year-old participants (n ϭ 1,145) and, for comparison pur- pseudoPSU variables available from the NHANES website to poses, in a random sample of available serum samples from 30- account for the survey’s complex sampling design (44). P Ͻ 0.05 to 39-year-old participants (n ϭ 1,026). Serum MMA concen- was considered statistically significant for all tests. trations were measured by gas chromatography/mass spectrom- We first used SUDAAN PROC REGRESS, SUDAAN etry using the solid extraction method and derivatization with PROC CROSSTAB, and SUDAAN PROC RLOGIST to de- cyclohexanol as described by Rasmussen (52). scribe the qualifying participants via least-squares means and In the NHANES 1999–2002, measurement of plasma tHcy proportions and to perform comparisons between subjects with and MMA were priority analyses. tHcy was measured by using serum vitamin B12 below the conventional cut-off point for a fully automated fluorescence polarization immunoassay (Ab- deficiency of 148 pmol/L and the remaining subjects. bott Laboratories). Plasma levels were calculated by using a For our main data analyses, we used SUDAAN PROC machine- stored calibration curve (Abbott IMx analyzer, 1999– REGRESS to estimate least-square geometric mean (95% C.I.) 2001; Abbott AxSYManalyzer, 2002). Plasma MMA was mea- circulating tHcy and MMA for serum folate categories after sured by gas chromatography/mass spectrometry with cyclohexa- multivariate adjustment for age, gender, race/ethnicity, body nol derivatization (53). It is important to note that MMA mass index, smoking status, alcohol intake, self-reported diabe- measurements for the NHANES 1999–2002 were much lower tes diagnosis, and serum concentrations of creatinine and ala- than the measurements recorded in the NHANES III. Pfeiffer nine aminotransferase. These analyses were performed stratified

Table 3. Category definitions and sample sizes* NHANES III† NHANES 1999–2002

Hcy MMA Hcy/MMA

Serum folate, nmol/L Ͻ9.1 9.1–13.5 Ն13.6 Յ9.7 9.7–13.5 Ն13.6 Ͻ19.3 19.3–25.9 26–32.5 Ն32.6

Serum B12 Ն 148 pmol/L, n 1,581 1,271 1,970 501 309 655 1,032 1,152 892 2,267 Serum B12 Ͻ 148 pmol/L, n 39 41 37 15 13 16 32 33 32 33

*Sample sizes given are after exclusions and for subjects with complete information for multivariate modeling. †In the NHANES III, Hcy and MMA were measured in surplus serum collected in the last 3 years of the 6-year survey, and MMA was measured in surplus serum MEDICAL SCIENCES samples measured from subjects Ն65 years old and from a sample of subjects 30–39 years old.

Selhub et al. PNAS ͉ December 11, 2007 ͉ vol. 104 ͉ no. 50 ͉ 19999 Downloaded by guest on September 26, 2021 by vitamin B12 status after interactions between vitamin B12 from the NHANES III with data from the NHANES 1999–2002. status and folate status were tested. Because of the much smaller We divided participants in the 3-year second phase of the number of subjects in the group with lower serum vitamin B12 as NHANES III into three folate categories, and we divided compared with the number with higher serum vitamin B12 participants in the 4-year NHANES 1999–2002 into four folate values, we used the distribution of the former group to determine categories (Table 3). the serum folate category cut-off points. The availability of tHcy and MMA values for NHANES III This material is supported by the United States Department of Agri- participants allowed us to examine interrelations among folate, culture under agreement no.1950-51520-008-00D and United States vitamin B12, and the metabolites across a broader range of folate Department of Agriculture Grant 2006-35200-17198. Any opinions, levels than the NHANES 1999–2002 dataset afforded given the findings, conclusions, or recommendations expressed in this publication institution of government- mandated food folic acid fortification are those of the authors and do not necessarily reflect the view of the between the two surveys. We made no attempt to combine data United States Department of Agriculture.

1. Banerjee R (2006) ACS Chem Biol 1:149–159. 30. Kondo H, Osborne ML, Kolhouse JF, Binder MJ, Podell ER, Utley CS, Abrams 2. Ludwig ML, Matthews RG (1997) Annu Rev Biochem 66:269–313. RS, Allen RH (1981) J Clin Invest 67:1270–1283. 3. Savage DG, Lindenbaum J (1995) in Folate in Health and Disease, ed Bayley 31. Rask H, Olesen AS, Mortensen JZ, Freund LG (1983) Scand J Haematol LB (Marcel Decker, NY), pp 237–285. 31:45–48. 4. Baik HW, Russell RM (1999) Annu Rev Nutr 19:357–377. 32. Vieira-Makings E, Chetty N, Reavis SC, Metz J (1991) Biochem J 275(Pt 5. Scott JM (1999) Proc Nutr Soc 58:441–448. 3):585–590. 6. Allen RH, Stabler SP, Lindenbaum J (1998) Eur J Pediatr 157(Suppl 2):S122– 33. Riedel B, Fiskerstrand T, Refsum H, Ueland PM (1999) Biochem J 341(Pt S126. 1):133–138. 7. Chodos RB, Ross JF (1951) Blood 6:1213–1233. 34. Sweeney MR, McPartlin J, Weir DG, Daly L, Scott JM (2006) Br J Nutr 8. Reynolds E (2006) Lancet Neurol 5:949–960. 95:145–151. 9. Carmel R (1990) Am J Hematol 34:108–114. 35. Troen AM, Mitchell B, Sorensen B, Wener MH, Johnston A, Wood B, Selhub 10. Miller JW (2002) Nutr Rev 60:142–144. J, McTiernan A, Yasui Y, Oral E, et al. (2006) J Nutr 136:189–194. 11. Morris MS, Jacques PF, Rosenberg I, Selhub J (2007) Am J Clin Nutr 36. Stokstad EL, Reisenauer A, Kusano G, Keating JN (1988) Arch Biochem 85:193–200. Biophys 265:407–414. 12. Duthie SJ, Whalley LJ, Collins AR, Leaper S, Berger K, Deary IJ (2002) Am J 37. Scott JM, Weir DG (1981) Lancet 2:337–340. Clin Nutr 75:908–913. 38. Oltean S, Banerjee R (2003) J Biol Chem 278:20778–20784. 13. Lindeman RD, Romero LJ, Koehler KM, Liang HC, LaRue A, Baumgartner 39. Oltean S, Banerjee R (2005) J Biol Chem 280:32662–32668. J Am Coll Nutr RN, Garry PJ (2000) 19:68–76. 40. Selhub J, Savin MA, Sakami W, Flavin M (1971) Proc Natl Acad Sci USA 14. Ramos MI, Allen LH, Mungas DM, Jagust WJ, Haan MN, Green R, Miller JW 68:312–314. (2005) Am J Clin Nutr 82:1346–1352. 41. Cook RJ, Wagner C (1984) Proc Natl Acad Sci USA 81:3631–3634. 15. Wahlin A, Hill RD, Winblad B, Backman L (1996) Psychol Aging 11:487–496. 42. Herbig K, Chiang EP, Lee LR, Hills J, Shane B, Stover PJ (2002) J Biol Chem 16. Tucker KL, Qiao N, Scott T, Rosenberg I, Spiro A, III (2005) Am J Clin Nutr 277:38381–38389. 82:627–635. 43. Matthews RG, Baugh CM (1980) Biochemistry 19:2040–2045. 17. Durga J, van Boxtel MP, Schouten EG, Kok FJ, Jolles J, Katan MB, Verhoef 44. National Center for Health Statistics (2005) National Health and Nutrition P (2007) Lancet 369:208–216. Examination Survey. Available at www.cdc.gov/nchs/data/nhanes/gendoc.pdf. 18. Nurk E, Refsum H, Tell GS, Engedal K, Vollset SE, Ueland PM, Nygaard HA, Smith AD (2005) Ann Neurol 58:847–857. Accessed September 1, 2005. 19. Kado DM, Karlamangla AS, Huang MH, Troen A, Rowe JW, Selhub J, Seeman 45. National Center for Health Statistics (2005) National Health and Nutrition TE (2005) Am J Med 118:161–167. Examination Survey: Analytic Reporting Guidelines. Available at www.cdc.gov/ ࿝ ࿝ ࿝ ࿝ ࿝ ࿝ 20. Morris MC, Evans DA, Bienias JL, Tangney CC, Hebert LE, Scherr PA, nchs/data/nhanes 03 04/nhanes analytic guidelines dec 2005.pdf. Accessed Schneider JA (2005) Arch Neurol 62:641–645. December 31, 2005. 21. Yajnik C (2006) Nutr Rev 64:S50–S51. 46. National Center for Health Statistics (2007) National Health and Nutrition 22. Volkov I, Press Y, Rudoy I (2006) J Nippon Med Sch 73:65–69. Examination Survey: Lab 18, Biochemistry Profile. Available at www.cdc.gov/ 23. Rosenblatt DS, Cooper BA (1987) Blood Rev 1:177–182. nchs/data/nhanes/frequency/phdoc.pdf. Accessed September 1, 2005. 24. Smith AE, Matthews RG (2000) Biochemistry 39:13880–13890. 47. Ioannou GN, Boyko EJ, Lee SP (2006) Am J Gastroenterol 101:76–82. 25. Jarrett JT, Hoover DM, Ludwig ML, Matthews RG (1998) Biochemistry 48. Coresh J, Eknoyan G, Levey AS (2002) J Am Soc Nephrol 13:2811–2812. 37:12649–12658. 49. Israelsson B, Brattstrom L, Refsum H (1993) Scand J Clin Lab Invest 26. Lerner-Ellis JP, Dobson CM, Wai T, Watkins D, Tirone JC, Leclerc D, Dore 53:465–469. C, Lepage P, Gravel RA, Rosenblatt DS (2004) Hum Mutat 24:509–516. 50. Stabler SP, Marcell PD, Podell ER, Allen RH, Savage DG, Lindenbaum J 27. Ng J, Frith R (2002) Lancet 360:384. (1988) J Clin Invest 81:466–474. 28. Waclawik AJ, Luzzio CC, Juhasz-Pocsine K, Hamilton V (2003) Wis Med J 51. Araki A, Sako Y (1987) J Chromatogr 422:43–52. 102:43–45. 52. Rasmussen K (1989) Clin Chem 35:260–264. 29. Drummond JT, Matthews RG (1994) Biochemistry 33:3732–3741. 53. Pfeiffer CM, Gunter EW (1999) Clin Chem 45:A166 (abstr 593).

20000 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0709487104 Selhub et al. Downloaded by guest on September 26, 2021