European Journal of Clinical Nutrition (2003) 57, 349–357 ß 2003 Nature Publishing Group All rights reserved 0954–3007/03 $25.00 www.nature.com/ejcn ORIGINAL COMMUNICATION Relationship between methylmalonic acid, homocysteine, vitamin B12 intake and status and socio-economic indices, in a subset of participants in the British National Diet and Nutrition Survey of people aged 65 y and over

CJ Bates1*, J Schneede2, G Mishra1, A Prentice1 and MA Mansoor3

1MRC Human Nutrition Research, Elsie Widdowson Laboratory, Cambridge, UK; 2Department of Pharmacology and Toxicology, University of Bergen, Haukeland Hospital, Bergen, Norway; and 3Department of Clinical Chemistry, Central Hospital in Rogaland, Stavanger, Norway

Objective: Assessment of functional vitamin B12 status in a subset of the respondents in the British National Diet and Nutrition Survey of people aged 65 y and over. Setting: National Diet and Nutrition Survey: a British nationwide cross-sectional sample of people aged 65 y and over, living either in the community or in institutions such as nursing homes, during one calendar year spanning 1994 – 1995. Methods: Methylmalonic acid (MMA) concentrations were measured in plasma samples from 313 subjects (ca 14% of those originally enrolled in the survey). The results were compared with those for serum vitamin B12, vitamin B12 intakes and other status and intake estimates and with socio-demographic indices. Results: Of the NDNS participants overall, 20% had serum vitamin B12 concentrations < 150 pmol=l. In the subset studied here, 24% of free-living and 46% of institution-living participants had MMA >0.5 mmol=l. Geometric mean MMA increased with age, from 0.25 mmol=l in people aged 65 – 74 y to 0.38 mmol=l in people aged 85 þ y. There was little evidence for any gender difference in MMA. It was inversely correlated with serum vitamin B12 and with red blood cell folate; it was positively correlated directly with total homocysteine, but not significantly with serum folate or with vitamin B12 intake. Among respondents with high MMA, a subgroup had normal serum vitamin B12 but higher-than-average plasma urea and creatinine. Socio-demographic co-variates of MMA included receipt of State income benefits, social class of head of household, and educational attainment. These indices were not correlated with serum vitamin B12. Conclusions: The progressive increase in MMA with age is metabolic evidence for increasing risk of functional vitamin B12 deficiency with increasing age in older people. There is evidence that renal function is linked to high MMA in some older people. Age and renal function are thus both important when establishing upper reference limits for MMA. The socio-demographic observations suggest a link between poverty and poor functional vitamin B12 status in older British people. Sponsorship: The Department of Health. European Journal of Clinical Nutrition (2003) 57, 349 – 357. doi:10.1038=sj.ejcn.1601540

Keywords: methylmalonic acid; vitamin B12; biochemical status; elderly; national survey

*Correspondence: CJ Bates, MRC Human Nutrition Research, Elsie Introduction Widdowson Laboratory, Fulbourn Road, Cambridge CB1 9NL, UK. E-mail: [email protected] During the past few decades, a number of studies worldwide Guarantor: Dr CJ Bates. have demonstrated the unique value of serum or plasma Contributors: CJB co-ordinated the analyses, contributed to data-analysis and interpretation and drafted the manuscript; JS and MAM performed methylmalonic acid (MMA) as a specific marker of functional the methylmalonic acid analyses and contributed to the interpretation; vitamin B12 status. Following the early studies that demon- MAM performed the homocysteine analyses; GM provided statistical strated the basic relationship between vitamin B12 and MMA advice and AP directed the project in Cambridge, contributed to the in man (Gompertz et al, 1967; Chanarin et al, 1973; Norman study design and contributed to the interpretation. Received 30 October 2001; revised 15 February 2002; et al, 1982; Carmel, 2000), and which established reliable accepted 28 May 2002 assay techniques for MMA (Stabler et al, 1986), were a Functional vitamin B12 status CJ Bates et al 350 number of key studies showing a surprisingly high preva- September 1995, thereby spanning all four seasons. People lence of raised MMA levels in the plasma of older people, living in institutions such as nursing homes were sampled where it appeared to provide a more reliable indicator of separately. Of an eligible sample of 2172 people living in the

functional vitamin B12 status than serum or plasma vitamin community, 944 (43%) provided a blood sample. Of a sample B12 concentrations per se (Lindenbaum et al, 1988; Rasmus- of 454 people living in institutions 267 (59%) provided a sen et al, 1989; Moelby et al, 1990; Joosten et al, 1993; Allen & blood sample. The MMA assays were performed on a subset Casterline, 1994; Lindenbaum et al, 1994; Koehler of the blood samples, comprising 256 from free-living sub- et al, 1996; Baik & Russell, 1999; Bjorkegren & Svardsudd, jects and 57 from institution-living ones. 1999; Herrmann et al, 2000). Clearly, the ageing process Demographic, socio-economic and lifestyle information in itself, and the increasing prevalence of age-related was obtained by questionnaire, by a trained interviewer in

medical conditions that impair vitamin B12 absorption, the respondents’ homes or institution residence, and the 4 both increase the probability that people will become vita- day weighed diet record was likewise supervised (Finch et al,

min B12-deficient as they grow older. Functional deficiency 1998). An early morning, usually fasting, blood sample was may occur even in people whose serum vitamin B12 taken by a trained nurse. Part was sent to a haematology concentrations are in the normal (ie reference) range laboratory; the remainder (with lithium heparin anticoagu- (Joosten et al, 1993). lant) was taken in a cool-box to a local hospital laboratory The relationships between MMA, tHcy and cysteine are for immediate separation and storage of the blood fractions potentially important because both tHcy (Ueland et al, 1992) at 7 40C for up to 3 months and then at 7 80C, for up to and cysteine (Mansoor et al, 1995; El-Khairy et al, 1999) can 5 y. A wide range of status analyses was carried out at the independently predict vascular disease risk, tHcy is increased Micronutrient Status Laboratory, formerly of the MRC Dunn

in vitamin B12 deficiency as one among several determi- Nutrition Unit and now part of MRC Human Nutrition nants, and the combination of serum vitamin B12 and Research. Plasma tHcy and cysteine were measured in MMA estimates can therefore be used to distinguish the Dr MA Mansoor’s laboratory in Stavanger, Norway (Mansoor

vitamin B12-deficiency-effects on tHcy from those of other et al, 1992; Bates et al, 1997). nutrient deficiencies and influences. MMA was measured by capillary electrophoresis The purpose of the present study was to examine vitamin (Schneede & Ueland, 1995). The local reference range of

B12, MMA and their correlates in a sample of older British the assay for non-elderly adults is 0.05 – 0.26 mmol=l and people who participated in a recent National Survey of diet the CV is below 10% for MMA concentration within the and nutrition (NDNS). normal range. About half of the MMA analyses were carried out by Dr J Schneede in Bergen, Norway while the rest were Subjects and methods performed by Dr Mansoor using an identical assay. Those As part of a series of government-commissioned surveys of performed in Bergen included all those with the lowest 3.5%

several age-groups of the British population in the final and highest 3.5% of serum vitamin B12 values; those ana- decade of the twentieth century, a survey of people aged lysed in Stavanger were randomly selected from the remain- 65 y and over was carried out in 1994 – 1995 (Finch et al, der. The mean CV% for QC samples with values between 1998; Bates et al, 1999). Estimates of mean daily nutrient 0.05 and 0.51 mmol=l was 7.1% in Stavanger. Good agree- intakes (from a 4 day weighed diet record), a socio- ment was obtained by comparison with a mass spectrometric demographic questionnaire and a fasting blood sample for method performed at the Centers for Disease Control (CDC) biochemical and other status measurements were included. Atlanta, Georgia, USA (CDC-value ¼ 0.013 þ 1.142Â The blood measurements included, inter alia, serum vitamin Stavanger-value, n ¼ 9, r2 ¼ 0.999), and good agreement B12 and serum and red cell folate concentrations, and a with a mass-spectrometric assay was also achieved in separate study examined plasma total homocysteine (tHcy) Bergen. The inter-assay CV in Bergen was 8% at 0.18 and and plasma cysteine concentrations (Bates et al, 1997). For 0.36 mmol=l MMA levels, and the intra-assay CV was < 5%. the present study, measurements of plasma MMA concentra- Participation in an external QA scheme (Moller et al, 1999) tions were performed in two laboratories, on a subset of the provided further evidence of validity and accuracy. To remaining unused plasma samples. These measurements achieve inter-laboratory harmonization between the Stavan- have enabled a comparison to be made with key socio- ger and the Bergen data-sets, 15 of the samples were analysed demographic indices as well as with other biochemical (blind) in both laboratories. These exhibited a log – linear indices of micronutrient status. relationship (r ¼ 0.89) with an intercept indistinguishable The survey plan and procedures have been described in from zero, and a slope which differed significantly (by detail elsewhere (Finch et al, 1998), therefore only a brief 14%) from unity. Therefore the two sets of results, on the summary is included here. A representative cross-sectional log scale, were harmonized to a common mean slope. population sample of people aged 65 y and over, living in Because the MMA data-set was positively skewed, all data mainland Britain, was obtained by random selection from 80 analyses were performed after log-transformation, and trans- randomly selected postcode sectors. These were allocated to formation was also needed for most of the other biochemical four sequential 3 month ‘waves’ between October 1994 and indices and intake estimates.

European Journal of Clinical Nutrition Functional vitamin B12 status CJ Bates et al 351 Data were analysed by a DataDesk (Data Descriptions Inc., of the corresponding MMA values. Since the key socio- Ithaca, NY, USA) statistical package, using a general linear demographic characteristics of the selected subset of partici- model for multivariate linear regression calculations, with pants were not identical to the census population of Great continuous or discrete variables as appropriate. To construct Britain, for the data shown in Table 1, an individual weighting

Figure 1a and b, serum vitamin B12 and tHcy were subdivided factor was applied to the datasets from each respondent, to into fifths of their distribution and the geometric mean correct the category proportions (based on age, gender, value for each fifth was plotted against the geometric mean socio-demographic status etc) to those of the census popula- tion (Finch et al, 1998). Permission for the survey procedures was given by the individual Local Research Ethics Committees representing each of the 80 postcode sectors included, and by the MRC Dunn Nutrition Unit’s Ethics Committee.

Results Table 1 shows the summary statistics by age-group for MMA

and the corresponding serum vitamin B12, tHcy, plasma creatinine and urea data for free-living respondents, and for a combined group of subjects in institutions (since numbers in each age-group in the institutions were too small for separate analysis). In the free-living group, the

loge(MMA) value increased progressively with age (P < 0.0001). The slope of the regression line for loge(MMA) against age was 0.019. After adjustment for age, there was no

significant difference in loge(MMA) between the two gen- ders. The high mean MMA in the institution group was consistent with the fact that 73% of them were aged 80 y or over.

The mean serum vitamin B12 concentration in the survey subset in which both serum B12 and MMA were measured was 211 pmol=l, which was significantly lower (by 13%) than the mean of 242 pmol=l in the subset in whom serum

vitamin B12, but not MMA, was measured. Additional com- parisons were made between the biochemical, haematologi- cal and anthropometric status indices of the ‘MMA’ and the ‘non-MMA’ subsets (not shown). Folate status and tHcy concentrations did not differ between the ‘MMA’ and the ‘non-MMA’ subsets. Only eight of 61 indices exhibited sig- nificant differences between the two subsets. The adjustment from observed to census-adjusted index values so as to mirror the socio-demographic patterns of the entire UK census population (Table 1) made little difference to any of the values; therefore the selected (‘MMA’) subset can be considered representative of the British population as a whole, except insofar that they had slightly lower mean

serum vitamin B12 levels. Eleven of the subjects whose MMA was measured were

users of vitamin B12 supplements. The amount ranged from 1 Figure 1 Serum vitamin B12 and plasma tHcy were divided into fifths of their distribution. They and the corresponding MMA values were loge to 5.75 mg=day, with a median of 2 mg=day. Although their transformed, and the means were adjusted (by multiple regression) for mean MMA concentration was lower than that of the sub- age and plasma creatinine, and were plotted, after back-transformation, jects not using B supplements (0.30 vs 0.41 mmol=l), this as fifths of their distribution, against MMA as the dependent variable. 12 There were n ¼ 58 – 61 pairs of values for each of the points in the upper was not significant. Fifteen subjects were users of folic acid figure and 53 – 57 pairs for those in the lower one. Error bars are s.e.m. supplements (daily intake 88 – 5000 mg=day, median Linear regressions between loge (serum vitamin B12)and loge (MMA) and 300 mg=day), but there was no significant difference in their between loge (tHcy) and loge (MMA) were both significant at MMA concentrations. P < 0.0001.

European Journal of Clinical Nutrition Functional vitamin B12 status CJ Bates et al 352

Table 1 Summary statistics for plasma MMA and serum vitamin B12

Free-living participants Institution participants Description 65 – 74 y 75 – 84 y 85 þ y All All

Methylmalonic acid (mmol=l) n 93 107 56 256 57 Median 0.243 0.302 0.368 0.299 0.349 Geometric mean (adjusted):a 0.249 0.320 0.376 0.283 0.369 (male-adjusted)a 0.271 0.318 0.415 0.292 0.364 (female-adjusted)a 0.233 0.321 0.365 0.278 0.371 percentage >0.5 mmol=l1221271932

Serum vitamin B12 (pmol=l) n 90 105 55 250 56 Geometric mean (adjusted):a 185 179 159 181 191 (male-adjusted)a 179 168 186 176 188 (female-adjusted)a 191 186 152 184 197 percentage < 130 pmol=l2239353227

Plasma total homocysteine (mmol=l) n 85 94 50 229 53 Geometric mean (adjusted):a 13.6 16.1 17.0 14.7 18.3 (male-adjusted)a 15.4 17.2 15.5 16.0 17.9 (female-adjusted)a 12.2 15.5 17.5 13.9 18.4

Plasma creatinine (mmol=l) n 88 105 54 247 55 Geometric mean adjusted:a 77.3 86.4 84.2 81.2 83.3 (male-adjusted)a 79.4 95.2 100.8 85.5 103.4 (female-adjusted)a 75.9 81.7 79.8 78.6 78.0

Plasma urea (mmol=l) n 92 106 54 252 55 Geometric mean adjusted:a 5.28 6.01 6.30 5.62 6.29 (male-adjusted)a 5.18 6.24 7.36 5.60 6.33 (female-adjusted)a 5.37 5.89 6.02 5.63 6.25

aMeans have been adjusted by a weighting factor, designed to correct for demographic differences between the sample and the entire British census population: see Subjects and Methods. Linear regression of the log-transformed variables against age (continuous), for the free-living group only, revealed the following

relationships: for MMA, direct, P < 0.0001; for serum vitamin B12, P ¼ 0.14, NS; for plasma tHcy, direct, P ¼ 0.0005; for plasma creatinine, P ¼ 0.16, NS; for plasma urea, direct, P ¼ 0.0001. Working normal ranges intended to define major biochemical deficiency=functional abnormality for these indices (Finch et al, 1998: Bates et al, 1997; Joosten et al, 1993) were: for MMA, < 0.5 mmol=l; for tHcy, < 20 mmol=l; for serum folate, >7 nmol=l;

for red cell folate, >345 nmol=l; for serum vitamin B12, >150 pmol=l; for plasma creatinine, < 160 mmol=l, and for plasma urea, < 15 mmol=l.

Table 2 shows the slopes and significance of the linear Table 2 Correlations between MMA and selected indices of vitamin relationships between loge(MMA) and key status indices and status and vitamin intakes nutrient intakes. MMA was inversely correlated with serum

vitamin B12 and with red cell folate, and directly correlated All subjects with plasma tHcy. However, there was no significant rela- tionship between MMA and serum folate or plasma cysteine. Index correlated with MMA df Coefficient (s.e.) P The overall correlation between MMA and vitamin B 12 Serum vitamin B12 298 70.440 (0.062) < 0.001 intake was of only borderline significance, but the partici- Plasma total 275 0.525 (0.090) < 0.0001 homocysteine (tHcy) pants with the lowest vitamin B12 intakes (n ¼ 82 with intakes < 3 mg=day) had MMA concentrations significantly Red cell folate 301 70.019 (0.005) 0.0009 Serum folate 297 70.052 (0.052) 0.3 higher than those with vitamin B12 intakes  3 mg=day (0.35 Plasma cysteine 264 0.146 (0.313) 0.6

vs 0.55 mmol=l). This was not attenuated by adjustment for Vitamin B12 intake 290 70.131 (0.065) 0.05 Folate intake 290 70.088 (0.075) 0.2 age, gender or domicile. Therefore, vitamin B12 intake, as well as efficiency of absorption, influences MMA concentra- Status indices and vitamin intake estimates were loge-transformed, and the tions here. models adjusted for age and plasma creatinine and, in the case of nutrient Figure 1 shows the relationships between MMA (as depen- intakes, also for energy and protein intakes. If combined in a multivariate model with backward elimination, significance (P < 0.0001) was retained for dent variable) and serum vitamin B12 and tHcy (as indepen- both serum vitamin B12 and tHcy.

European Journal of Clinical Nutrition Functional vitamin B12 status CJ Bates et al 353 dent variables) calculated by fifths of the distribution of the independent variables. Whereas the relation between MMA and tHcy was essentially linear, that between MMA and serum vitamin B12 was clearly non-linear. When the lowest values of serum vitamin B12 were progressively removed

% Age (y) 22), serum vitamin from the calculation, the correlation between MMA and iron – Plasma

saturation serum vitamin B12 became non-significant (P >0.05) when 21 ¼ the cut-off reached 170 pmol=l. n a

l)

Table 3 shows the result of subdividing the respondents =

g

m

( into four groups on the basis of their serum vitamin B12 and Serum

ferritin MMA values, using cut-off values of 150 pmol=l for vitamin l; group 3 ( = B12 (cf. Sauberlich, 1999), and 0.5 mmol=l for MMA (the mol observed 80th centile of MMA and the upper limit in healthy m )

fl

( elderly people, as observed by Joosten et al, 1993). There 0.5 Mean < emerged two subgroups of subjects, both with elevated MMA cell volume values, but with characteristically different biochemical and haematological characteristics. One of these, group 3, had a

l)

= normal serum vitamin B but raised plasma urea and crea- l and MMA 12 crit (l =

tinine concentrations, by comparison with group 1. This Haemato- tests showed that for tHcy, groups 1 and 3 and groups 2 and 3 could group had normal red cell folate and iron status indices.

The other high-MMA group, (group 4), had low serum 150 pmol  dl)

= post-hoc vitamin B12 and red cell folate and raised MMA, but had 12

(g globin

Haemo- l.

normal plasma urea and creatinine. tHcy was lowest in group =

1, highest in group 4, and intermediate in groups 2 and 3. mol

l) m ). Scheffe

=

None of the groups differed significantly from each other b 0.5 and methylmalonic acid values with regard to the basic haematology indices: blood haemo-  per se urea Plasma globin, haematocrit or mean red cell volume. 12 (mmol

Table 4 illustrates the socio-demographic covariates of 69), serum vitamin B – b MMA. High MMA was significantly correlated with receipt l)

= 61 l and MMA = of State income benefit (a marker of low income), with ¼ mol n

Plasma m manual social class of the head of household, and with ( creatinine lower educational attainment (certificate of secondary edu- 150 pmol  cation or above, versus none). People receiving State income l) l group 2 ( b = 12 benefit, those of manual social class and those with less = mol

folate Red cell cantly from each other. For plasma urea, only group 3 differed from all other groups. For serum ferritin, only group 1 differed educational attainment had the higher MMA concentra- (nmol m fi

tions. In a multivariate model containing age, sex and all 0.5 <

l)

the above socio-demographic indices, only age, receipt of b = State benefits and social class of head of household remained

Serum folate significantly correlated with MMA after backward elimina- (nmol 42), serum vitamin B –

tion. A combination of age, receipt of State benefits, social l and MMA =

l) 34

= class of head of household, log serum vitamin B and log b

e 12 e ¼

n tHcy together explained 31% of the variation in MMA. This mol tHcy Plasma m

( did not change significantly if the subjects with the poorest 150 pmol >

renal function (highest 10% of plasma creatinine and plasma 12

l)

b = l; group 4 (

urea) were excluded. None of the socio-demographic indices =

mol were strongly associated with serum vitamin B . 0.0001 0.0007 0.13 0.035 0.035 0.0003 0.5 0.6 1.0 0.0004 0.6 0.03 m

12 Plasma mol MMA ( < m 0.5 

l)

= b

12 97 0.827 22.6 11.3 326 86 5.7 13.4 0.41 90.6 43 23.6 81.0 257103 0.238193 0.283 13.9 0.770 18.2 12.8 19.0 9.9 464 10.0 366 435 88 77 98 6.1 5.4 13.5 7.7 13.6 0.41 13.1 0.41 90.4 0.40 90.7 74 90.1 45 25.0 80 24.8 77.4 22.5 78.8 80.0

B 0.0001

Serum 180), serum vitamin B vitamin (pmol < Discussion – l and MMA = The progressive increase in MMA concentrations with age 166 ¼

n (Table 1) is even more striking than the decrease in serum Biochemical and haematological characteristics of four groups based on serum vitamin B 12 12 12 12 vitamin B12 concentrations with age (cf. Finch et al, 1998). cance of differences between groups by ANOVA: models were all adjusted for age (except that for age 150 pmol fi

Judging from the entire NDNS dataset (Finch et al, 1998) > low MMA) low MMA) high MMA) high MMA) 12 Log-transformed for the analysis and back-transformed for reporting. Group 1 ( 2 (low B 3 (high B 4 (low B ANOVA-P Table 3 1 (high B a b B Signi 20% of all the survey respondents had serum vitamin B12 Group not be distinguished, nor couldfrom groups both 2 2 and and 4, 4. but Age all others differed differed between signi groups 1 and 4.

European Journal of Clinical Nutrition Functional vitamin B12 status CJ Bates et al 354

Table 4 Variation of MMA and serum vitamin B12 with socio-demographic (SD) and related variables

Regression Regression vs Regression vs

vs ln(MMA) ln(serum vitamin B12) ln(vitamin B12 intake) Socio-demographic or related variable PP P

Region (four standard regions) 0.16a 0.2 0.04 Receipt of State benefits 0.0003b 0.4 0.2 Social class, head of household 0.0001c 0.7 0.15 Income group 0.2 0.7 0.8 Educational attainment 0.0005d 0.9 0.3 Self-reported health 0.13 0.11 0.11 Self-reported physical activity 0.7 0.6 0.9 Self-reported smoking habit 0.6 0.6 0.2

All respondents with MMA values were included in these analyses: dF 219 – 312. Models were adjusted for age and

plasma creatinine and, in the case of vitamin B12 intake, also for energy and protein intakes. If combined together in a multivariate model with backward elimination, ln(MMA) remained significantly correlated with receipt of state benefits and social class of head of household. aThe four standard regions are: (1) Scotland; (2) northern England; (3) central and southern England and Wales; (4) London and south-east England. bMMA was higher in those receiving State income benefits. cMMA was higher in those where the head of household was, or had been, in a manual occupation. d MMA was higher, and vitamin B12 intake was lower, in those who had not received an education up to at least the equivalent of Certificate of Secondary Education (CSE).

concentrations < 150 pmol=l, and are thus biochemically None of the three groups with low serum vitamin B12 or deficient as defined in the present study. Results from a high plasma MMA or both, in Table 3, had abnormal hae- recent survey of young people living in Britain (Gregory et moglobin, haematocrit or mean red cell volume. Mean red al, 2000; Bates et al, 2002) have confirmed that the risk of cell volume may be reduced (microcytosis) by iron defi-

biochemical vitamin B12 deficiency is considerably less likely ciency, which could mask any increase (macrocytosis) result- in young than in elderly people. ing from vitamin B12 deficiency. However, when all subjects The results shown in Table 3 of the present study indicate with low ( < 15%) transferrin saturation and=or low that some elderly respondents with raised MMA levels also ( < 20 mg=l) serum ferritin were excluded from our data-set, have abnormal plasma urea and creatinine concentration, which reduced the number of subjects from 303 to 231, the and therefore impaired kidney function. Many other studies linear regression between MCV and MMA remained non- have also detected a link between raised blood MMA levels significant (P ¼ 0.6, NS). It is, however, of interest that, in the and impaired kidney function (Stabler et al, 1986; Rasmussen entire NDNS data-set with 1079 subjects, there was a sig-

et al, 1990a, b; Joosten et al, 1993; Lindenbaum et al, 1994; nificant inverse relationship between serum vitamin B12 and Koehler et al, 1996; Norman, 1998, 1999; van Asselt et al, mean red cell volume (P ¼ 0.025), which became even more 1998; Herrmann et al, 2000; Hvas et al, 2000). According to highly significant (P ¼ 0.015) if subjects with poor iron status Whitehead et al (1994), the 97.5th centile of plasma creati- were omitted. This suggests that some of the respondents

nine in adults aged 25 – 55 y is approximately 122 mmol=l with low serum vitamin B12 may have macrocytosis. (males) and 100 mmol=l (females), and for plasma urea it is Bjorkegren & Svardsudd (1999), reported that, although 7.6 mmol=l (males) and 6.8 mmol=l (females). If all subjects MMA increases with age, it is not necessarily correlated with with plasma creatinine and=or urea concentrations above any abnormality of the classical haematological indices. these cut-off points were eliminated from the present Excessive reliance on the evidence of haematological indices dataset, there remained 107 men with apparently normal may not be an efficient way of detecting incipient vitamin

renal function, of whom 20% still had MMA con- B12 deficiency in older people, and better ways of monitoring centrations >0.5 mmol=l, and 112 women with apparently functional vitamin B12 deficiency in populations are clearly normal kidney function, of whom 15% had MMA con- needed (Savage et al, 1994; Nilsson-Ehle, 1998).

centrations >0.5 mmol=l. Therefore, a substantial proportion Respondents with low serum vitamin B12 but normal of the high MMA concentrations in the present study are MMA (group 2 in Table 3) appeared to be intermediate unlikely to be attributable to impaired kidney function, and between group 1 and group 4 in their biochemical character-

are thus more likely to be due to poor vitamin B12 status. An istics. Their red cell folate and serum ferritin levels were MMA elevation above 0.70 mmol=l is unlikely to be explained relatively low, yet they had normal MMA levels. Therefore,

by moderately impaired renal function alone (Rasmussen et al, they appeared to have a milder vitamin B12 deficiency than 1990a,b; Moelby et al, 2000). More work is clearly needed to those of group 4, who had clearly raised MMA levels. Their establish reasonable MMA reference limits in older people moderately raised tHcy was presumably the result of their

with partially impaired renal function. poor folate and vitamin B12 status.

European Journal of Clinical Nutrition Functional vitamin B12 status CJ Bates et al 355 It is possible that some subjects with apparently normal correlation between serum vitamin B12 and vitamin B12 serum vitamin B12 had altered levels of active vitamin B12 intake is relatively weak (Finch et al, 1998; Bates et al, 1999). (holo-TCII). This normally constitutes about 10 – 30% of The lack of any correlation between MMA and plasma total vitamin B12 in serum and is filtered in the glomerulus cysteine (Table 2) supports the idea that circulating cysteine and actively re-absorbed in the proximal tubule. An altera- is controlled by factors that are different from those which tion in holo-TCII may not be detected by conventional control tHcy. This observation helps to distinguish those vitamin B12 assays. metabolic factors which control the sulphur amino-acids as As expected, in the present study a strong inverse correla- a group, and those which control individual members of the tion between MMA and serum vitamin B12 was observed. group. There was also a strong direct correlation between MMA and The socio-demographic correlates of MMA (Table 4) sug- tHcy and both of these correlations still remained significant gest that there may be important environmental factors that after adjustment for age (and creatinine; Table 2 and Figure predict functional vitamin B12 status in older people 1). However, at serum vitamin B12 concentrations above in Britain. There was, however, little evidence of socio- 170 pmol=l the relation between MMA and serum B12 demographic influences on serum vitamin B12. The under- became non-significant. Others have estimated the mini- lying mechanisms controlling these relationships are not mum serum B12 for a plateau of MMA as ca 250 pmol=l well understood, and deserve further investigation. (Lindenbaum et al, 1994). This indicates the existence of a In summary, the observations of the present study have stable low plateau of MMA (Rasmussen et al, 1996) when the extended our knowledge of some of the factors correlated supply of vitamin B12 is optimal. with vitamin B12 status and its functional index, MMA, in An inverse correlation between MMA and red cell folate, older British people. The observations of the present study, but not between MMA and serum folate, was observed (Table when combined with those of van Asselt et al, (1998) and 2). This is consistent with literature evidence that vitamin Bjorkegren and Svardsudd (2001), suggest that older people

B12 inadequacy is associated with failure of maintenance of living in Europe may be at risk of functional vitamin B12 intracellular folate polyglutamates, followed by leakage of deficiency. Functional B12 deficiency in North America has monoglutamate folates out of the cells (Perry et al, 1976; also been highlighted (Pennypacker et al, 1992; Lindenbaum Chanarin, 1979). In the NDNS survey, 15% of the free-living et al, 1994; Carmel et al, 1999; Stabler et al, 1999), especially participants and 39% of institution-living ones had serum in Caucasians (Carmel et al, 1999; Stabler et al, 1999). It is folate concentrations < 7 nmol=l, and similar proportions (8 unlikely to be due to a simple lack of vitamin B12 in the diet, and 16%, respectively) had red cell folate concentrations and thus is more likely to relate to impaired absorption or below 230 nmol=l (Finch et al, 1998). retention of the vitamin, by mechanisms that remain poorly It is of interest to estimate, from the survey dataset (Finch understood (Howard et al, 1998; van Asselt et al 1998). A et al, 1998), what proportion of subjects with raised plasma recent study has demonstrated a correlation between sever- tHcy may have increases that are specifically due to low ity depression and MMA, but not serum vitamin B12 in older vitamin B12 status. Using a lower cut-off of 150 pmol=l for American women (Penninx et al, 2000), thus raising the serum vitamin B12 and an upper cut-off of 20 mmol=l for possibility of subtle functional (neurological) effects of plasma total homocysteine, 83=972 (8.5%) of subjects had early vitamin B12 deficiency. A high prevalence of subtle both low serum B12 and high tHcy. If one then excludes vitamin B12 deficiency among the elderly in populations all those participants with either a low serum folate newly exposed to folate fortification of flour and other ( < 7 nmol=l) or a low red cell folate ( < 230 nmol=l), then staple foods (Jacques et al, 1999) could pose new public 16 still have raised tHcy, and this falls to 11 after elimination health problems, and MMA as a sensitive marker of func- of people with moderately raised plasma creatinine and=or tional cobalamin status is likely to become an important urea levels, as defined earlier. In this subgroup, at least, all component in their investigation. Despite the substantially other likely causes of raised tHcy having been excluded, this increased dietary requirement for vitamin B12 in people with leaves low serum B12 as the most likely remaining explana- impaired absorption efficiency, it seems likely that a sub- tion. Clearly at least some of the raised tHcy levels seen in stantial number of older people living in Britain could the NDNS respondents are likely to be attributable solely or benefit from moderate increases in intake of this vitamin, principally to poor vitamin B12 status. which could be provided, for instance, by fortification of The rather weak correlation between MMA and estimated cereals. vitamin B12 intake (Table 2) may be explained by the fact that much of the variance of serum vitamin B12 between individuals arises from differences in efficiency of vitamin Acknowledgements

B12 absorption, rather than individual differences in dietary The survey was commissioned jointly by the Department of supply, except when the dietary intake is very low. In addi- Health and the Ministry of Agriculture, Fisheries and Food, tion, it is likely that vitamin B12 status is dependent on long- whose responsibility has since been transferred to the Food term, as distinct from short-term variations in vitamin B12 Standards Agency. It was carried out by Social and Commu- intakes. This is also suggested by the observation that the nity Planning Research (now the National Centre for Social

European Journal of Clinical Nutrition Functional vitamin B12 status CJ Bates et al 356 Research), in conjunction with the Micronutrient Status Jacques PF, Selhub J, Bostom AG, Wilson PWF & Rosenberg IH (1999): The effect of folic acid fortification on plasma folate and total Laboratory of the MRC Dunn Nutrition Unit (now part of homocysteine concentrations. New Engl. J. Med. 19, 1449 – 1454. MRC Human Nutrition Research). Funding for the measure- Joosten E, van den Berg A, Riezler R, Naurath HJ, Lindenbaum J, ment of MMA was provided by the Department of Health. Stabler SP & Allen RH (1993): Metabolic evidence that deficiencies of vitamin B-12 (cobalamin), folate and vitamin B-6 occur com- monly in elderly people. Am. J. Clin. Nutr. 58, 468 – 476. References Koehler KM, Romero LJ, Stauber PM, Pareo-Tubbeh SL, Liang HC, Allen LH & Casterline J (1994): Vitamin B12 deficiency in elderly Baumgartner RN, Garry PJ, Allen RH & Stabler SP (1996): Vitamin individuals: diagnosis and requirements. Am. J. Clin. Nutr. 60,12– supplementation and other variables affecting serum homocys- 14. teine and methylmalonic acid concentrations in elderly men and Baik HW & Russell RM (1999): Vitamin B12 deficiency in the elderly. women. J. Am. Coll. Nutr. 15,364– 376. A. Rev. Nutr. 19, 357 – 377. Lindenbaum J, Healton EB, Savage DG, Brust JC, Garrett TJ et al Bates CJ, Mansoor MA, van der Pols J, Prentice A, Cole TJ & Finch S (1988): Neuropsychiatric disorders caused by cobalamin deficiency (1997): Plasma total homocysteine in a representative sample of in the absence of anemia or macrocytosis. New Engl. J. Med. 318, 972 British men and women aged 65 and over. Eur. J. Clin. Nutr. 1720 – 1728. 51,1– 7. Lindenbaum J, Rosenberg IH, Wilson PWF, Stabler SP & Allen RH Bates CJ, Prentice A, Cole TJ, van der Pols JC, Doyle W, Finch S, (1994): Prevalence of cobalamin deficiency in the Framingham Smithers G & Clarke PC (1999): Micronutrients — highlights and elderly population. Am. J. Clin. Nutr. 60,2– 11. research challenges from the 1994=5 National Diet and Nutrition Mansoor M, Svardal A & Ueland P (1992): Determination of the in Survey of People Aged 65 Years and Over. Br. J. Nutr. 82,7– 15. vivo redox status of cysteine, cysteinylglycine, homocysteine and Bates CJ, Mansoor MA, Gregory J, Pentieva K & Prentice A (2002): glutathione in human plasma. Anal. Biochem. 200, 218 – 229. Correlates of plasma homocysteine, cysteine and cysteinyl-glycine Mansoor MA, Bergmark C, Svardal AM, Lonning PB & Ueland PM in respondents in the British National Diet and Nutrition Survey (1995): Redox status and protein binding of plasma homocysteine of Young People Aged 4 – 18 Years, and a comparison with the and other aminothiols in patients with early-onset peripheral Survey of People Aged 65 Years and Over. Br. J. Nutr. 87,71– 79. vascular disease. Arterioscler. Thromb. Vasc. Biol. 15, 232 – 240. Bjorkegren K & Svardsudd K (1999): Elevated serum levels of methyl- Moelby L, Rasmussen H, Jensen MK & Pedersen KO (1990): The malonic acid and homocysteine in elderly people. A population- relationship between clinically confirmed cobalamin deficiency based intervention study. J. Intern. Med. 246, 317 – 324. and serum methylmalonic acid. J. Intern. Med. 228, 373 – 378. Bjorkegren K & Svardsudd K (2001): Serum cobalamin, folate, Moelby L, Rasmussen K, Ring T & Nielsen G (2000): Relationship methylmalonic acid and total homocysteine as vitamin B12 and between methylmalonic acid and cobalamin in uremia. Kidney Int. folate deficiency markers amongst elderly Swedes — a popula- 57, 265 – 273. tion-based study. J. Intern. Med. 249, 423 – 432. Moller J, Rasmussen K & Christensen L (1999) External quality Carmel R (2000): Current concepts in cobalamin deficiency. A. Rev. assessment of methylmalonic acid and total homocysteine. Clin. Med. 51, 357 – 375. Chem. 45, 1536 – 1542. Carmel R, Green R, Jacobsen DW, Rasmussen K, Florea M & Azen C Nilsson-Ehle H (1998): Age-related changes in cobalamin (vitamin (1999): Serum cobalamin, homocysteine, and methylmalonic acid B12) handling. Implications for therapy. Drugs Aging 12, 277 – 292. concentrations in a multiethnic elderly population: ethnic and Norman EJ (1998): Urinary methylmalonic acid=creatinine defines sex differences in cobalamin and metabolite abnormalities. Am. J. true cobalamin deficiency. Br. J. Haematol. 100, 617 – 618. Clin. Nutr. 70, 904 – 910. Norman EJ (1999): Urinary methylmalonic acid=creatinine ratio: a Chanarin I (1979): The Megaloblastic Anaemias, Chap. 14. Oxford: gold standard test for tissue vitamin B12 deficiency. J. Am. Geriatr. Blackwell Scientific Publications. Soc. 47, 1158 – 1159. Chanarin I, England JM, Mollin C & Perry J (1973): Methylmalonic Norman EJ, Martelo OJ & Denton MD (1982): Cobalamin (vitamin acid excretion studies. Br. J. Haematol. 25,45– 53. B12) deficiency detection by urinary methylmalonic acid quanti- El-Khairy L, Ueland P, Nygard O, Refsum H & Vollset S (1999): tation. Blood 59, 1128 – 1131. Lifestyle and cardiovascular disease risk factors as determinants Penninx BWJH, Guralnik JM, Ferrucci L, Fried LP, Allen RH & Stabler of total cysteine in plasma: the Hordaland Homocysteine Study. SP (2000): Vitamin B12 deficiency and depression in physically Am. J. Clin. Nutr. 70, 1016 – 1024. disabled older women: epidemiololgic evidence from the women’s Finch S, Doyle W, Lowe S, Bates CJ, Prentice A, Smithers G & Clarke P health and aging study. Am. J. Psych. 157, 715 – 721. (1998): National Diet and Nutrition Survey: People Aged 65 Years or Pennypacker LC, Allen RH, Kelly JP, Matthews LM, Grigsby J, Kaye K, Over. Volume 1. Report of the Diet and Nutrition Survey. London: Lindenbaum J & Stabler SP (1992): High prevalence of cobalamin The Stationery Office. deficiency in elderly outpatients. J. Am. Geriatr. Soc. 40, Gompertz D, Jones JH & Knowles JP (1967): Metabolic precursors of 1197 – 1204. methylmalonic acid in vitamin B12 deficiency. Clin. Chim. Acta Perry J, Lumb M Laundy M, Reynolds EH & Chanarin I (1976): Role 18, 197 – 204. of vitamin B12 in folate coenzyme synthesis. Br. J. Haematol. 32, Gregory J, Lowe S, Bates C, Prentice A, Jackson L & Smithers G 243 – 248. (2000): National Diet and Nutrition Survey: Young People Aged 4 to 18 Rasmussen K, Moelby L & Jensen MK (1989): Studies on methylma- Years. Volume 1. Report of the Diet and Nutrition Survey. London: lonic acid in humans. II. Relationship between concentrations in The Stationery Office. serum and urinary excretion, and the correlation between serum Herrmann W, Schorr H, Bodis M, Knapp JP, Muller A, Stein G & cobalamin and accumulation of methylmalonic acid. Clin. Chem. Geisel J (2000): Role of homocysteine, cystathionine and methyl- 35, 2277 – 2280. malonic acid measurement for diagnosis of vitamin deficiency in Rasmussen K, Moller J, Ostergaard K, Ostergaard Kristensen M & high-aged subjects. Eur. J. Clin. Invest. 30, 1083 – 1089. Jensen J (1990a): Methylmalonic acid concentrations in serum of Howard JM, Azen C, Jacobsen DW, Green R & Carmel R (1998): normal subjects: biological variability and effect of oral L-isoleu- Dietary intake of cobalamin in elderly people who have abnormal cine loads before and after intramuscular administration of coba- serum cobalamin, methylmalonic acid and homocysteine levels. lamin. Clin. Chem. 36, 1295 – 1299. Eur. J. Clin. Nutr. 52,582– 587. Rasmussen K, Vyberg B, Pedersen KO & Brochner-Mortensen J Hvas AM, Juul S, Gerdes LU & Nexo E (2000): The marker of (1990b): Methylmalonic acid in renal insufficiency: evidence of cobalamin deficiency, plasma methylmalonic acid, correlates to accumulation and implications for diagnosis of cobalamin plasma creatinine. J. Intern. Med. 247, 507 – 512. deficiency. Clin. Chem. 36, 1523 – 1524.

European Journal of Clinical Nutrition Functional vitamin B12 status CJ Bates et al 357 Rasmussen K, Moller J, Lyngbak M, Pedersen AM & Dybkjaer L Stabler SP, Allen RH, Fried LP, Pahor M, Kittner SJ, Penninx BWJH & (1996): Age- and gender-specific reference intervals for total Guralnik JM (1999): Racial differences in prevalence of cobalamin homocysteine and methylmalonic acid in plasma before and and folate deficiencies in disabled elderly women. Am. J. Clin. after vitamin supplementation. Clin. Chem. 42, 630 – 636. Nutr. 70, 911 – 919. Sauberlich HE (1999): Laboratory Tests for the Assessment of Nutritional Ueland P, Refsum H and Brattstrom L (1992): Plasma homocysteine Status, 2nd edn. Vitamin B-12 (Cyanocobalamin; Corrinoids),pp and cardiovascular disease. In: Atherosclerotic Cardiovascular Dis- 135 – 159. Boca Raton, FL: CRC Press. ease, Hemostasis and Endothelial Function. ed. Francis RBJ, pp. 183 – Savage DG, Lindenbaum J, Stabler SP & Allen RH (1994): Sensitivity 236. New York: Marcel Dekker. of serum methylmalonic acid and total homocysteine determina- van Asselt DZB, de Groot LCPGM, van Staveren WA, Blom HJ, Wevers tions for diagnosing cobalamin and folate deficiencies. Am. J. Med. RA, Biemond I & Hoefnagels WHL (1998): Role of cobalamin 96, 239 – 246. intake and atrophic gastritis in mild cobalamin deficiency in Schneede J & Ueland PM (1995): Application of capillary electro- older Dutch subjects. Am. J. Clin. Nutr. 68, 328 – 334. phoresis with laser-induced fluorescence detection for routine Whitehead, TP, Robinson D, Hale AC & Bailey AR (1994) Clinical determination of methylmalonic acid in human serum. Anal. Chemistry and Haematology: Adult Reference Values. London: BUPA Chem. 34, 812 – 819. Medical Research and Development. Stabler SP, Marcell PD, Podell ER, Allen RH and Lindenbaum J (1986): Assay of methylmalonic acid in the serum of patients with cobalamin deficiency using capillary gas chromatography-mass- spectrometry. J. Clin. Invest. 77, 1606 – 1612.

European Journal of Clinical Nutrition