Human and Clinical Nutrition

Low Serum Vitamin B-12 Concentrations Are Associated With Faster Human Immunodeciency Virus Type 1 (HIV-1) Disease Progression1,2
Alice M. Tang,*3 Neil M. H. Graham,* Ranjit K. Chandra and Alfred J. Saah*
*Department of Epidemiology, Johns Hopkins School of Hygiene and Public Health, Baltimore, MD 21205; Department of Medicine, Johns Hopkins School of Medicine, Baltimore, MD 21205; and Janeway Child Health Centre, St. Johns, Newfoundland, A1A 1R8, Canada
ABSTRACT We conducted a nonconcurrent prospective cohort study to examine associations between serum concentrations of vitamin B-6, vitamin B-12 and folate and the risk of progression to rst acquired immunodeciency syndrome (AIDS) diagnosis and CD4/ cell decline to 2 1 108 cells/L. The study population was drawn from a cohort of homosexual and bisexual men in the Baltimore-Washington, DC, area. Eligible subjects were human immunodeciency virus type 1 (HIV-1)-seropositive at study entry and had serum available in the serum repository from their 1984 baseline study visit. Serum micronutrient levels were assessed in 310 subjects. The follow-up period (April 1984 through December 1993) was approximately 9 y. In Kaplan-Meier analyses, participants with low serum vitamin B-12 concentrations ( 120 pmol/L) had signicantly shorter AIDS-free time than those with adequate vitamin B-12 concentrations (median AIDS-free time 4 vs. 8 y, respectively, P 0.004). This effect persisted in Cox proportional hazards models after adjusting for HIV-1related symptoms, CD4/ cell count, age, serum albumin, use of antiretroviral therapy before AIDS, frequency of alcohol consumption and serum folate concentration [relative hazard (RH) 1.89, 95% condence interval (CI) 1.153.10). To further explore the temporal relation between low serum vitamin B-12 concentrations and disease progression, additional analyses were performed excluding subjects with more advanced disease at baseline. In these analyses, the increase in risk of progression to AIDS for those with low serum vitamin B-12 concentrations remained signicant (RH 2.21, 95% CI 1.134.34), providing further evidence that low vitamin B-12 concentrations preceded disease progression. In contrast, low serum concentrations of vitamin B-6 and folate were not associated with either progression to AIDS or decline in CD4/ lymphocyte count. Intervention studies are needed to determine whether correction of low serum vitamin B-12 concentrations in early HIV-1 infection will inuence the natural history of disease progression. J. Nutr. 127: 345351, 1997. KEY WORDS: AIDS folate HIV-1 infection vitamin B-6 vitamin B-12 humans

Downloaded from jn.nutrition.org by guest on March 18, 2013

Many B-group vitamins (vitamin B-6, vitamin B-12 and folate, in particular) play crucial roles in protein and nucleic acid synthesis. Deciencies of these micronutrients can lead to dramatic reductions in host resistance and lymphocyte function (Beisel 1982, Chandra and Newberne 1977, Rall and Meydani 1993). Abnormally high prevalences of low serum vitamin B-6 and vitamin B-12 have been reported in populations of humans infected with human immunodeciency virus type 1 (HIV-1),4 whereas the evidence for folate deciency has been less consistent (Baum et al. 1991b, Beach et al. 1992a, Bogden et al. 1990, Burkes et al. 1987, Coodley et al. 1993, Herbert et al. 1989, Mantero-Atienza et al. 1991). Low serum
1 Funded by the National Institute of Allergy and Infectious Diseases grant U01-AI-35042-02 and the National Institutes of Health Outpatient General Clinical Research Center grant RR007222. 2 The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 USC section 1734 solely to indicate this fact. 3 To whom correspondence and reprint requests should be addressed: 615 N. Wolfe St. E6007, Baltimore, MD 21205. 4 Abbreviations used: AIDS, acquired immunodeciency syndrome; ARC, AIDS-related complex; CI, condence interval; HIV-1, human immunodeciency virus type 1; RDA, recommended dietary allowance; RH, relative hazard.

concentrations of these micronutrients, although not necessarily indicators of overt deciency states, have been associated with poor outcomes in both acquired immunodeciency syndrome (AIDS) patients and asymptomatic HIV-1infected individuals (Baum et al. 1991b and 1995, Herzlich et al. 1990, Kieburtz et al. 1991, Remacha et al. 1991). Vitamin B-12 has been the most extensively studied of the three micronutrients in the context of HIV-1 infection. Low serum vitamin B-12 concentrations have been associated with peripheral neuropathy and myelopathy in HIV-1infected patients (Kieburtz et al. 1991), diminished performance on specic measures of information processing speed and visuospatial problem-solving skills (Beach et al. 1992b), CD4/ cell decline (Baum et al. 1995, Remacha et al. 1990), increased mortality (Remacha et al. 1991) and increased zidovudine-related bone marrow toxicity (Herzlich et al. 1990). Few studies have examined the consequences of low serum vitamin B-6 and folate concentrations in HIV-1 infection. Low vitamin B-6 status in HIV-1seropositive subjects was signicantly associated with a reduced response of peripheral blood lymphocytes to mitogens and decreased natural killer cell cytotoxicity (Baum et al., 1991b). In a more recent study, development of low vitamin B-

0022-3166/97 $3.00 1997 American Society for Nutritional Sciences. Manuscript received 28 May 1996. Initial review completed 29 July 1996. Revision accepted 1 November 1996. 345

/ 4w13$$0022

01-24-97 09:42:36

nutra

LP: J Nut

346

TANG ET AL.

6 status over 18 mo was associated with a signicant decline in surrogate markers of HIV disease stage (Baum et al. 1995). Folate deciency, like vitamin B-12 deciency, may have adverse effects on the immune system and neurological functions (Smith et al. 1987) in HIV-1 infection. We measured serum concentrations of vitamin B-6, vitamin B-12 and folate in a cohort of HIV-1 seropositive homosexual and bisexual men using sera collected and stored in 1984. These individuals have been followed semiannually for more than 10 y, providing us with a unique opportunity to examine the effect of these serum micronutrients on the natural history of HIV-1 infection using a longitudinal study design. SUBJECTS AND METHODS
Study population. Between April and November 1984, 1153 homosexual and bisexual men were enrolled into the Baltimore-Washington, DC, site of the Multicenter AIDS Cohort Study, a prospective study of the natural history of HIV-1 infection in homosexual and bisexual men. Eligibility criteria for this study were men over the age of 18, who had a history of sexual activity with male partners and no AIDS-dening illness. At baseline and subsequent semi-annual follow-up visits, participants provided information about health and medical status, therapeutic and illicit drug use, and sexual practices. All subjects were examined by a medical provider, and blood was collected for HIV-1 serology, laboratory studies, and repository storage. Serum and plasma from all subjects were processed, and aliquots derived, according to standard protocols (Kaslow et al. 1987). The HIV-1 serostatus was determined by enzyme immunoassay (Genetic Systems, Seattle, WA) and a Western blot (Du Pont, Co. Wilmington, DE, and Bio-Rad, Hercules, CA) with interpretation using standard criteria. Details on the objectives, design and recruitment protocol of this study have been described elsewhere (Kaslow et al. 1987). Of the 1153 men who enrolled in the study at baseline, 812 (70%) were HIV-1 negative and 341 (30%) were HIV-1 positive. All subjects who were HIV-1 seropositive upon study entry and had serum available in the repository from their 1984 baseline study visit were eligible for the present study. The HIV-1 seropositive men who had serum available in the repository (n 312) did not differ signicantly from those who did not (n 29) with respect to age, race, education, cigarette smoking, CD4/ T cell count, or weight (data not shown). This study was reviewed and approved by the Committee on Human Research of The Johns Hopkins University, School of Hygiene and Public Health. Serum micronutrient concentrations. Concentrations of vitamin B-6, vitamin B-12 and folate in the serum were obtained for 310 of the 312 subjects. Vitamin B-6 concentrations were determined using HPLC method (Sampson and OConnor 1989). Normal range for serum B-6 concentrations was dened as 88 to 390 nmol/L based on laboratory standards. The AccessR (Sano Pasteur, Chicago, IL) competitive-binding immunoenzymatic assay was used to determine serum vitamin B-12 and serum folate concentrations. The normal ranges for vitamin B-12 and folate, as determined by laboratory standards, were dened as 120 to 720 pmol/L and 3.4 and 4.5 nmol/L, respectively. Serum albumin and C-reactive protein concentrations were also determined. Serum albumin concentrations were assessed by a timeendpoint colorimetric method using the Beckman Synchron CXR (Beckman Instruments, Brea, CA) system, with normal range between 35 and 50 g/L (Heymseld et al. 1994). Because serum albumin concentrations are inuenced by factors such as physiologic stress and specic disease conditions, we also assessed C-reactive protein concentrations, an indicator of acute physiologic stress. C-reactive protein concentrations were determined using the RapitexR (Rapitex, London, UK) turbidimetric assay based on immunochemical reaction between C-reactive protein and antibodies to C-reactive protein bound to latex particles; there is a correlation between C-reactive protein concentrations and extent of agglutination of the latex particles. Levels above 8.0 mg/L (as determined by the laboratory) were considered to be elevated, indicating the presence of an acute, underlying infection.

Assessment of dietary intake. In 1984, usual dietary intake was assessed through a self-administered, semi-quantitative food frequency questionnaire consisting of 116 food items (Willett et al. 1988). Subjects were asked to estimate their frequency of consumption of each food item over the previous 12 mo, in terms of a typical serving size. Nine frequency categories were available to choose from, ranging from never or less than once per month to 6/ per day. Also included on the form were sections for recording the brand name, frequency and amount of any nutritional supplements taken, a writein section for foods not listed, and a series of questions on the brands and types of fat used for frying, cooking and baking. Completed questionnaires were reviewed and coded by a trained researcher and entered into computer data les. Estimates of food and supplemental intake of vitamin B-6, vitamin B-12 and folate were determined using the nutrient analysis program developed specically for the questionnaire (Willett et al. 1988). Independent covariates. Most of the independent variables included in this study were based on data collected at the baseline clinic visit. Frequency of alcohol consumption was based on the question how often do you drink now? asked at the baseline study visit. This variable was originally coded into eight categories ranging from one to ve times per year to at least once per day. For this study, frequency of alcohol consumption was collapsed into two categories: 2 times per week and 2 times per week. Twentythree subjects were missing information on this variable from their baseline visit. To avoid excluding these individuals from the analyses, values from their next study visit (approximately 6 mo later) were substituted. Information on cigarette smoking was based on the question do you smoke cigarettes now? asked on the baseline questionnaire. Subjects were categorized either as nonsmokers or as regular or occasional smokers according to their response. Body mass index was calculated as weight divided by height squared (kg/m2). The HIV-1 related symptoms variable was dened as the presence of one or more of the following most commonly reported symptoms lasting two or more weeks: persistent or recurring fever 37.8C, persistent diarrhea, oral thrush, persistent fatigue, or unintentional weight loss of more than 4.5 kg. Categories of CD4/ T-lymphocyte count were dened as 5 1 108, 5 1 108 to 7.5 1 108, and 7.5 1 108 cells/L in order to give approximately equal numbers of men in each category. CD4/ T cell counts were not available at the baseline visit for 32 (10.3%) of the subjects. These subjects were assigned to one of the three categories based on data from subsequent adjacent visits. Because the mean decline of CD4/ cell counts in this cohort was only approximately 3 1 107 cells/L per 6 mo, we believe it was possible to estimate the appropriate CD4/ cell category for these individuals with a high degree of accuracy. Information on the use of antiretroviral drugs (zidovudine, didanosine and zalcitubine) and Pneumocystis carinii pneumonia prophylaxis (aerosolized pentamidine, trimethoprim-sulfamethoxazole and dapsone) was obtained from data collected over the entire follow-up period. Two binary variables were created to indicate ever or never use of each of these types of drugs during the follow-up period but before the onset of AIDS. Outcomes for survival analyses. The cutoff date for the present study was December 31, 1993, slightly more than 9 y after the beginning of the Baltimore Multicenter AIDS Cohort Study. Outcomes examined in this study included time to rst AIDS diagnosis and time to CD4/ cell decline to 2 1 108 cells/L. The AIDS diagnoses were based on the 1987 revision of the Centers for Disease Control surveillance case denition (Centers for Disease Control 1987). For all analyses in which AIDS was the outcome variable, the following censoring rules were used. For subjects who were diagnosed with AIDS during the follow-up period (n 163), follow-up time was calculated as the time from the rst visit date to the date of rst AIDS diagnosis. Subjects who developed AIDS after December 31, 1993, (n 4) or remained AIDS-free and were not lost to follow-up (n 78) were censored at the study cutoff date. Subjects whose rst AIDS diagnosis was at death (n 17) or who were lost to follow-up (n 38) were censored at the date they were last seen (AIDS-free) in the clinic. Finally, subjects who died during the follow-up period of causes other than AIDS (n 12) were censored at their date of death. For each subject, CD4/ cell counts were measured at every follow-

Downloaded from jn.nutrition.org by guest on March 18, 2013

/ 4w13$$0022

01-24-97 09:42:36

nutra

LP: J Nut

B-GROUP VITAMINS AND HIV-1 DISEASE PROGRESSION

347

up visit. Time to the rst visit where CD4 count was 2 1 108 cells/ L was calculated for each subject whose CD4/ cell count reached this level before the study cutoff date (n 180). Subjects whose CD4/ cell count never fell below 2 1 108 cells/L (n 100) were censored at the date of their last CD4/ cell count measurement if this date was prior to the study cutoff date. Otherwise, these subjects were censored at the study cutoff date (n 29). Subjects whose CD4/ cell counts declined to 2 1 108 cells/L after the study cutoff (n 3) were also censored at the cutoff date. Statistical methods. All data analyses were performed using STATA (version 4.0, Computing Resource Center, Santa Monica, CA) and SAS (version 6.09, SAS Institute, Cary, NC). For each of the three micronutrients, the proportion of subjects with low serum concentrations was determined. All independent covariates were examined for their associations with serum micronutrient concentrations using t tests for continuous variables and chi-square tests for categorical covariates. Univariate associations between serum micronutrient concentrations and the development of AIDS or reaching a CD4/ cell count of 2 1 108 cells/L were assessed using the Kaplan-Meier product-limit method (Cox and Oakes 1984). Median AIDS-free times were estimated based on the Kaplan-Meier curves. The logrank test was used to assess the association of the micronutrients and independent variables with both outcomes. To examine the independent effect of serum nutrient concentrations on progression to clinical AIDS or CD4/ cell count 2 1 108 cells/L, Cox proportional hazards models (Cox and Oakes 1984) were used to adjust for covariates that were associated with either serum micronutrient concentrations, disease progression or both (P 0.15) in the univariate analyses. A step-down procedure was used to eliminate variables that did not produce signicant changes in the model estimates. Separate Cox models were t for each B-group vitamin. Serum nutrient concentrations were entered into the models as binary variables (indicating adequate vs. low levels) and as quartiles. All independent covariates were entered into the models as categorical variables. The relation between serum micronutrient concentrations and food and supplemental intake of vitamin B-6, vitamin B-12 and folate was examined using nonparametric methods due to the skewed distribution of the intake data. Spearman rank-order correlations (Conover 1980) were used to examine the association between serum nutrient concentrations and nutrient intake. The Wilcoxon rank-sum test (Snedecor and Cochran 1989) was used to compare median daily nutrient intakes between subjects with adequate vs. low serum nutrient concentrations and to compare median serum nutrient concentrations between current and past or never users of multivitamin and single-vitamin supplements.

TABLE 1
Characteristics of the study population1
Characteristic Frequency n (%)

Demographics (at baseline) Age, y 30 3040 40 Race and ethnicity White, non-Hispanic Other Education level College degree No college degree Alcohol consumption (frequency) 2 times/week 2 times/week Cigarette smoking Nonsmoker Regular or occasional smoker Health status (at baseline) HIV-related symptoms2 No symptoms Symptomatic CD4/ T cells/L 5 1 108 5 1 108 to 7.5 1 108 7.5 1 108 Body mass index (kg/m2, quartiles) 21.5 21.522.7 22.824.4 24.5 Serum albumin, g/L 35 35 C-reactive protein, mg/L 8.0 8.0 Treatment Antiretroviral therapy3 Not used before AIDS Used before AIDS PCP prophylaxis4 Not used before AIDS Used before AIDS

67 (22%) 194 (62%) 51 (16%) 269 (86%) 43 (14%) 188 (60%) 124 (40%) 171 (55%) 141 (45%) 194 (62%) 118 (38%)

254 (81%) 58 (19%)

Downloaded from jn.nutrition.org by guest on March 18, 2013

118 (38%) 97 (31%) 97 (31%) 80 78 76 78 (26%) (25%) (24%) (25%)

71 (23%) 241 (77%) 141 (45%) 169 (55%)

165 (53%) 147 (47%) 236 (76%) 76 (24%)

RESULTS Population characteristics. Table 1 shows the characteristics of the 312 men in this study. The mean age of the study population at baseline was 34 y (range 30 to 65 y), and the majority of men (86%) reported their race as white, nonHispanic. The mean CD4/ cell count was 6.4 1 108 cells/L, and only 11 men (4%) had CD4/ cell counts 2 1 108 cells/ L. Although most men (81%) were still in the asymptomatic stage of infection, 55% had elevated serum C-reactive protein concentrations (8.0 mg/L), and 71 (23%) had low serum albumin concentrations (35 g/L). Over the 9-y follow-up period, 163 men (52%) progressed to AIDS, and 180 men (58%) reached a CD4/ cell count of 2 1 108 cells/L. First AIDS diagnoses were predominantly due to P. carinii pneumonia (35%), Kaposis sarcoma (25%), Candida albicans (8%) and Mycobacterium avium-intracellulare complex (6%). Wasting syndrome and dementia were the rst AIDS diagnoses for seven (4.3%) and ve (3.1%) of the subjects, respectively. Table 2 summarizes the serum concentrations and dietary intake levels of the three micronutrients. For all three micronutrients, mean and median serum concentrations fell within
1 Subjects were 312 HIV-1seropositive homosexual and bisexual men from the Baltimore-Washington, DC, site of the Multicenter AIDS Cohort Study, 1984. AIDS acquired immunodeciency syndrome, PCP Pneumocystis carinii pneumonia. 2 Presence of fever 37.8C, diarrhea, persistent fatigue, thrush or unintentional weight loss 4.5 kg. 3 Zidovudine, didanosine or zalcitabine use at any time during follow-up period. 4 Aerosolized pentamidine, trimethoprim-sulfamethoxazole or dapsone used during follow-up period.

the normal range. Thirty-four subjects (11%) had low serum vitamin B-6, 37 (12%) had low serum vitamin B-12, and 24 (8%) had low serum folate. Seventy-seven men (25%) had at least one serum micronutrient concentration in the low range, and 17 (5%) had at least two low nutrient concentrations. Only one subject had low serum concentrations of all three vitamins. Mean and median intakes from food and supplements combined were above the U.S. Recommended Dietary Allowance (RDA) for all three micronutrients. Although only two subjects had vitamin B-12 intakes below the RDA, 21% and 12% of subjects were consuming vitamin B-6 and folate, respectively, at levels below the RDA.

/ 4w13$$0022

01-24-97 09:42:36

nutra

LP: J Nut

348

TANG ET AL.

TABLE 2
Summary of serum micronutrient levels and total nutrient intake (from food and supplements) in 310 HIV-1seropositive homosexual and bisexual men1,2
Mean { SD Median (IQR)3 Proportion inadequate Cutoff for inadequacy

% below normal Serum nutrient concentration Vitamin B-6, nmol/L Vitamin B-12, pmol/L Folate, nmol/L Total nutrient intake4 Vitamin B-6, mg/d Vitamin B-12, mg/d Folate, mg/d 223 { 98 243 { 121 10.1 { 6.7 240 (160290) 220 (155305) 8.6 (5.512.8) 11 12 8 % below RDA5 10.4 { 30.9 25.0 { 93.3 508 { 284 3.3 (2.26.0) 13.9 (7.420.6) 455 (292669) 21 0.7 12

lower limit of normal 88 120 3.4 RDA 2 2 200

1 Subjects were participants in the Baltimore Multicenter AIDS Cohort Study, 1984. AIDS acquired immunodeciency syndrome. 2 Serum concentrations were estimated in 1994 from sera frozen in 19841985. 3 IQR interquartile range. 4 Data available for 272 subjects. 5 RDA Recommended Dietary Allowances, based on the 1989 revised version.

Subjects with low serum vitamin B-6 concentrations (88 nmol/L) were more likely to be over the age of 40 y (chisquare test P 0.08), drink more frequently (chi-square test, P 0.02), have lower CD4/ cell counts (chi-square test P 0.12), lower body mass index (t test, P 0.04), lower serum albumin concentrations (t test, P 0.001) and higher serum C-reactive protein concentrations (chi-square test, P 0.006) compared with those with adequate serum vitamin B-6 (88 nmol/L). Subjects with low serum vitamin B-12 concentrations (120 pmol/L) were slightly older (chi-square test, P 0.12), tended to drink more frequently (chi-square test, P 0.13), had lower CD4/ cell counts (chi-square test, P 0.14), lower serum albumin concentrations (t test, P 0.001) and were less likely to have started antiretroviral therapy before AIDS (chi-square test, P 0.12). Low serum folate concentrations (6.8 nmol/L) were associated with younger age (chi-square test, P 0.04), lower education level (chi-square test, P 0.13), lower body mass index (t test, P 0.05) and lower serum albumin concentrations (t test, P 0.001). Univariate analyses. In univariate Kaplan-Meier analyses using time to rst AIDS diagnosis as the outcome, subjects with low serum vitamin B-12 concentrations had a signicantly lower probability of remaining AIDS-free over the follow-up period than subjects with adequate serum vitamin B12 concentrations (Fig. 1). The estimate of median AIDSfree time for those with low serum vitamin B-12 concentrations (120 pmol/L) was 4.4 y, whereas the median AIDSfree time for those with adequate vitamin B-12 concentrations (120 pmol/L) was 8.4 y. The logrank test of the difference between these two curves was highly signicant (P 0.004). No signicant differences were found between low vs. adequate serum concentrations for any of the three micronutrients in Kaplan-Meier analyses using CD4/ cell decline to 2 1 108 cells/L as the outcome (logrank P 0.46, 0.24 and 0.70 for vitamin B-6, vitamin B-12 and folate, respectively). Multivariate analyses. To examine the independent effect of serum nutrient concentrations on HIV-1 disease progression, Cox proportional hazards models were used to adjust for potential confounders. Variables that were signicantly associated with AIDS or CD4/ cell decline (P 0.15) were entered as covariates in the Cox models. These included HIV-related symptoms, CD4/ cell count, age, serum albumin concentration, serum C-reactive protein concentrations, antiretroviral

therapy before AIDS, P. carinii prophylaxis before AIDS and frequency of alcohol consumption. Serum C-reactive protein concentration and use of P. carinii prophylaxis before AIDS were subsequently dropped from the multivariate Cox models because they did not produce signicant changes in the model estimates. In the baseline Cox models, older age and the presence of HIV-1related symptoms were strongly and independently associated with an increased risk of progression to AIDS and CD4/ cell decline to 2 1 108 cells/L. Higher baseline CD4/ cell counts, more frequent alcohol consumption and the use of antiretroviral therapy before the onset of AIDS were all signicantly and independently associated with a decreased risk of progression to both outcomes. Serum albumin concentration was not a statistically signicant predictor for either outcome but was included in the Cox models because of its signicant effects on the model estimates. Table 3 shows the crude and adjusted hazard ratios for each of the three micronutrients, using progression to clinical AIDS

Downloaded from jn.nutrition.org by guest on March 18, 2013

FIGURE 1 Kaplan-Meier plot of the proportions of subjects surviving AIDS-free, stratied by low vs. adequate serum vitamin B-12 levels. Subjects were 310 HIV-1seropositive men enrolled in the BaltimoreWashington, DC, site of the Multicenter AIDS Cohort Study (October 1984 to December 31, 1993).

/ 4w13$$0022

01-24-97 09:42:36

nutra

LP: J Nut

B-GROUP VITAMINS AND HIV-1 DISEASE PROGRESSION

349

TABLE 3
Relative risk of AIDS progression in 310 HIV-1seropositive homosexual and bisexual men1,2
Serum concentrations No. events/total Crude relative risk (95% CI3) Adjusted relative risk4 (95% CI)

Low vs. adequate Vitamin B-6 88 nmol/L 88 nmol/L Vitamin B-125 120 pmol/L 120 pmol/L Folate 3.4 nmol/L 3.4 nmol/L Quartiles Vitamin B-12 (pmol/L)5 Q1: B12 155 Q2: 155 B12 220 Q3: 220 B12 305 Q4: B12 305

141/276 21/34 137/273 25/37 148/286 14/24

1.00 1.20 (0.761.89) 1.00 1.87 (1.222.87)* 1.00 0.92 (0.531.60)

1.00 1.30 (0.802.10) 1.00 1.89 (1.153.10)* 1.00 0.85 (0.471.54)

51/77 34/74 39/81 38/78

1.00 0.59 (0.380.91) 0.61 (0.400.93) 0.60 (0.390.92)

1.00 0.73 (0.461.16) 0.55 (0.350.85)5 0.55 (0.350.86)5

1 Subjects were participants in the Baltimore Multicenter AIDS Cohort Study, 19841994. AIDS acquired immunodeciency syndrome. HIV human immunodeciency virus. 2 Outcome is time to rst AIDS diagnosis (in days). 3 CI condence interval. 4 All models adjusted for the following covariates: HIV-related symptoms, CD4/ cell count, age at baseline, serum albumin level, use of antiretroviral therapy before AIDS, and frequency of alcohol consumption. 5 Multivariate model includes category of serum folate level (low vs. adequate). * Statistically signicant (P 0.05).

Downloaded from jn.nutrition.org by guest on March 18, 2013

as the outcome. Because low serum folate concentrations can lead to falsely low serum vitamin B-12 concentrations, these nutrients were placed together in the same multivariate model. In this way, an estimate of the relative hazard for serum vitamin B-12 adjusted for serum folate could be obtained. In crude analyses, subjects with low serum vitamin B-12 concentrations (120 pmol/L) had an 87% increase in risk of AIDS progression [relative hazard (RH) 1.87, 95% condence interval (CI) 1.22 2.87] compared with those with adequate serum vitamin B12 concentrations (120 pmol/L). After adjustment for serum folate and several other covariates, the RH increased slightly to 1.89 (95% CI 1.153.10). When serum vitamin B-12 concentrations were entered into the Cox models as quartiles, a threshold effect was observed where subjects in quartiles 2 through 4 showed a signicant decrease in risk of progression to AIDS compared with those in the lowest quartile. There was little evidence for changing risk after quartile 2 in the crude model or quartile 3 in the adjusted model. Vitamin B-6 and folate were not associated with AIDS progression in these analyses (Table 3). Furthermore, in multivariate Cox models no signicant associations were found between any of the three micronutrients and progression to CD4/ cell count 2 1 108 cells/L [RH for low vs. adequate levels 1.23 for vitamin B-6 (95% CI 0.771.95), 1.06 for vitamin B-12 (95% CI 0.661.68) and 0.83 for folate (95% CI 0.461.47)]. Association between dietary intake and serum nutrient concentrations. Data on food and supplemental intake of vitamin B-6, vitamin B-12 and folate were available for 272 of the 310 subjects. Those who completed and returned the selfadministered food frequency questionnaire were more likely to be nonsmokers (chi-square test, P 0.001), to have a college degree (chi-square test, P 0.002) and to be older (t test, P 0.02) than those who did not complete the questionnaire. There were no differences between those who did and did not complete the food frequency questionnaire with respect to the

other independent variables, AIDS progression, or CD4/ cell decline to 2 1 108 cells/L. Spearmans correlations between serum concentrations and food intake were low and nonsignicant for vitamins B-6 and B-12. A low but statistically signicant correlation was found, however, between serum folate concentrations and folate intake from foods (r 0.17, P 0.006). The correlation between serum concentration and total intake (from food and vitamin supplements) was not signicant for vitamin B-6 but was signicant for vitamin B-12 and folate (r 0.12, P 0.02 for vitamin B-12; r 0.41, P 0.0001 for folate). Those who reported current use of multi- or B-complex vitamin supplements (n 168) had signicantly higher median serum concentrations of vitamin B-12 and folate compared with past or never users (n 100), but there was little difference in serum vitamin B-6 concentrations between the two groups (Table 4).

TABLE 4
Median serum micronutrient concentrations by use of oral vitamin supplements in 272 HIV-1seropositive homosexual or bisexual men1
Vitamin supplements

Serum nutrient

Current use2

Past use or never used

Wilcoxon P value

Vitamin B-6, nmol/L Vitamin B-12, pmol/L Folate, nmol/L

210 235 10.4

240 190 6.5

0.41 0.0009 0.0001

1 Subjects were participants in the Baltimore Multicenter AIDS Cohort Study, 1984. 2 Current use of multivitamin and/or B-complex vitamin supplements.

/ 4w13$$0022

01-24-97 09:42:36

nutra

LP: J Nut

350

TANG ET AL.

DISCUSSION In this study, we examined the associations between low serum concentrations of specic B-group vitamins and two key outcomes in the natural history of HIV-1 disease progression: rst AIDS diagnosis and CD4/ cell decline to 2 1 108 cells/ L. In Kaplan-Meier analyses of the association between serum B-12 concentrations and AIDS progression, we found that median AIDS-free time was 4 y shorter in the low serum vitamin B-12 group than in the normal serum vitamin B-12 group. In multivariate Cox proportional hazards models, we found that low serum concentrations of vitamin B-12 remained signicantly associated with an increased risk of progression to AIDS compared with adequate vitamin B-12 levels. Low serum concentrations of vitamin B-6 and folate, however, were not associated with either of the two outcomes presented here. Our results are consistent with those from other studies conducted in HIV-1infected patients (Baum et al. 1995, Remacha et al. 1991, Rule et al. 1994). In a cross-sectional study, Remacha et al. (1991) found that HIV-1infected patients that had lower serum vitamin B12 concentrations had lower hemoglobin, leukocytes, CD4/ lymphocytes and CD4//CD8/ lymphocyte ratios than HIV1infected patients with normal serum vitamin B-12 concentrations. Ninety percent of the patients with low serum vitamin B-12 concentrations had AIDS compared with only 66% of patients with adequate vitamin B-12 concentrations. In a 30-mo follow-up study, Rule et al. (1994) reported that of nine subjects whose disease progressed to AIDS or AIDS-related complex (ARC), seven had falling serum vitamin B-12 concentrations before the onset of AIDS or ARC, and the remaining two experienced drops in serum vitamin B-12 concentration just after their rst AIDS diagnosis. These ndings were corroborated in a larger study by Baum et al. (1995), who found that development of a low serum vitamin B-12 concentration over an 18-mo follow-up period was associated with more rapid disease progression, as indicated by declines in CD4/ cell counts and an AIDS index (composite measurement of CD4/ cell count and b2-microglobulin). Subjects in the lowest tertile of plasma vitamin B-12 concentration (195 pmol/L) showed a signicantly greater decline in CD4/ cell count over time than those in the highest tertile (349 pmol/ L). We found that the associations of low serum vitamin B12 concentration with short-term changes in immune variables continued to hold true for clinical outcomes over the long term, as evidenced by an 89% increase in risk of progression to clinical AIDS in our study population over an approximate 10-y follow-up period. With observational studies such as this one, an association that seems to be causal may in fact be a reection of more advanced disease progression. Although we adjusted for category of CD4/ cell count at baseline as a marker of disease progression, there may have been some residual confounding with this variable. To more clearly determine whether low serum vitamin B-12 concentrations were a cause rather than a result of disease progression, we repeated the analyses after excluding subjects with more advanced disease at baseline. These included men with HIV-1related symptoms, CD4/ cell counts 2 1 108 cells/L at baseline, and those who progressed to AIDS within 3 y after study entry. We found that in the remaining 217 asymptomatic HIV-1infected subjects, those with low serum vitamin B-12 concentrations still had a signicant increase in risk of progression to AIDS compared with those with adequate serum vitamin B-12 concentrations (RH 2.21, 95% CI 1.134.34). Furthermore, among 43 subjects with normal C-reactive protein concentrations and

CD4/ cell counts 7.5 1 108 cells/L, the risk of AIDS progression for those with low serum vitamin B-12 concentrations was 3.4 times greater than for subjects with adequate vitamin B-12 concentrations (95% CI 0.7615.0). The fact that the association remained strong in asymptomatic subjects with levels measured at least 3 y prior to the onset of AIDS, and in those with normal C-reactive protein concentrations and CD4/ cell counts, provides us with stronger evidence for causality. Because an increase in risk of AIDS progression was clearly observed only in subjects with low serum vitamin B-12 concentrations, this would indicate that megadoses of vitamin B-12 supplements may not be necessary for a trial to assess the effects of vitamin B-12 on HIV-1 disease progression. Thus far, we have seen little evidence of additional therapeutic effects with vitamin B-12 supplementation beyond the correction of low serum concentrations. The sera used in this study were stored for approximately 10 y before nutrient analysis. Therefore, a foremost concern in this study was the amount of degradation that might have occurred over time. Few studies have documented the stability of vitamin B-6, vitamin B-12 and folate in frozen sera. Howard et al. (1984) reported that plasma vitamin B-6 concentrations were highly stable in plasma frozen at 020C for at least 2 y, with a rate of decline of only about 2.2% per year. Ocke et al. (1995) noted no decline in serum vitamin B-6 and B-12 concentrations after 4 y of storage at 020C. Because serum nutrient concentrations in our population were not analyzed when the sera were rst collected, we could not document the actual amount of degradation that occurred. However, mean serum and plasma concentrations of these vitamins have been reported in many studies of HIV-1infected individuals, at various stages of infection and from various parts of the world (Baum et al. 1991a and 1995, Beach et al. 1992a, Bogden et al. 1990, Boudes et al. 1990, Coodley et al. 1993, Dowling et al. 1993, Harriman et al. 1989, Herzlich et al. 1992, ManteroAtienza et al. 1991, Remacha et al. 1993, Revell et al. 1991, Robertson et al. 1993, Rule et al. 1994, Veilleux et al. 1995). Mean serum vitamin B-6 and B-12 concentrations in our study fell within the ranges reported in these other studies. The mean folate concentration in our subjects was slightly lower than those reported in the above populations; however, the prevalence of low serum folate concentrations in our population was comparable to the prevalences reported in these studies. Because all the sera used in our study underwent the same freezing, storage and thawing processes, however, conclusions drawn from internal comparisons should not be affected by the possibility of slight folate degradation during the followup period. We conducted this longitudinal study in a well-established cohort of HIV-1seropositive homosexual and bisexual men who donated serum more than 10 y ago. A caveat with respect to our data is that the blood concentrations of vitamin B-6, vitamin B-12 and folate were measured at one point in time. This is of particular concern for folate, because serum folate concentrations tend to more accurately reect acute rather than chronic folate status. However, the relatively good agreement between serum folate concentrations and the average daily intake of folate reported over the previous 12 mo supports the fact that serum folate concentrations in our subjects were a good reection of folate status over a substantial period of time. The same is true for serum vitamin B-12 concentrations. Serum vitamin B-6 concentrations remain fairly constant with changes in dietary intake, taking 3 to 4 wk for concentrations to reach a new steady-state level (Gibson 1990). Therefore, we are condent that vitamin B-6 intake on the day of blood

Downloaded from jn.nutrition.org by guest on March 18, 2013

/ 4w13$$0022

01-24-97 09:42:36

nutra

LP: J Nut

B-GROUP VITAMINS AND HIV-1 DISEASE PROGRESSION

351

sampling did not greatly affect serum vitamin B-6 concentrations. Another potential caveat in our study is that intake, absorption and metabolism of these B-group vitamins is likely to have changed during the course of disease progression for many individuals. Examining changes in both serum concentrations and dietary intakes of these vitamins over the course of disease progression may give more insight into their role in HIV-1 infection rather than measuring absolute levels at one point in time. This is supported by the ndings of Rule et al. (1994) and Baum et al. (1995) that declining serum vitamin B-12 concentrations were associated with declining CD4/ cell counts. This may explain the lack of association with longterm decline in CD4/ cell count that we observed with only a single measurement of serum vitamin B-12 concentrations. This is the rst longitudinal study to link low serum vitamin B-12 concentration with an increased risk of progression to AIDS over a follow-up period of over 9 y. We have found that between 8 and 12% of the men in our study had a subnormal serum concentration of vitamin B-6, vitamin B-12 or folate at study entry. Low baseline serum vitamin B-12 concentrations (120 pmol/L) were associated with a nearly twofold increase in risk of progression to AIDS after adjusting for HIV-1 related symptoms, CD4/ cell count, age, serum albumin, use of antiretroviral therapy before AIDS, serum folate concentration and frequency of alcohol consumption. This relation remained strong in asymptomatic subjects who started the study with CD4/ cell counts 2 1 108 cells/L and who started the study at least 3 y before their rst AIDS diagnosis. Therefore, serum vitamin B-12 concentrations seem to be an early and independent marker of HIV-1 disease progression. The effectiveness of vitamin B-12 replacement therapy in slowing disease progression, however, is still unknown and should be the focus of further research. LITERATURE CITED
Baum, M. K., Javier, J. J., Mantero-Atienza, E., Beach, R. S., Fletcher, M. A., Sauberlich, H. E., Feaster, D. & Shor-Posner G. (1991a) Zidovudine-associated adverse reactions in a longitudinal study of asymptomatic HIV-1infected homosexual males. J. Acquired Immune Dec. Syndr. 4: 12181226. Baum, M. K., Mantero-Atienza, E., Shor-Posner, G., Fletcher, M. A., Morgan, R., Eisdorfer, C., Sauberlich, H. E., Cornwell, P. E. & Beach, R. S. (1991b) Association of vitamin B6 status with parameters of immune function in early HIV1 infection. J. Acquired Immune Dec. Syndr. 4: 11221132. Baum, M. K., Shor-Posner, G., Lu, Y., Rosner, B., Sauberlich, H. E., Fletcher, M. A., Szapocznik, J., Eisdorfer, C., Buring, J. E. & Hennekens, C. H. (1995) Micronutrients and HIV-1 disease progression. AIDS 9: 10511056. Beach, R. S., Mantero-Atienza, E., Shor-Posner, G., Javier, J. J., Szapocznik, J., Morgan, R., Sauberlich, H. E., Cornwell, P. E., Eisdorfer, C. & Baum, M. K. (1992a) Specic nutrient abnormalities in asymptomatic HIV-1 infection. AIDS 6: 701708. Beach, R. S., Morgan, R., Wilkie, F., Mantero-Atienza, E., Blaney, N., Shor-Posner, G., Lu, Y., Eisdorfer, C. & Baum, M. K. (1992b) Plasma vitamin B12 level as a potential cofactor in studies of human immunodeciency virus type 1 related cognitive changes. Arch. Neurol. 49: 501506. Beisel, W. R. (1982) Single nutrients and immunity. Am. J. Clin. Nutr. 35 (Feb. suppl.): 417468. Bogden, J. D., Baker, H., Frank, O., Perez, G., Kemp, F., Bruening, K. & Louria, D. (1990) Micronutrient status and human immunodeciency virus (HIV) infection. In: Micronutrients and Immune Functions (Bendich, A. & Chandra, R. K., eds.), vol. 587, pp. 189195. New York Academy of Sciences, New York, NY. Boudes, P., Zittoun, J. & Sobel, A. (1990) Folate, vitamin B12, and HIV infection - letter to the editor. Lancet 335: 14011402. Burkes, R. L., Cohen, H., Krailo, M., Sinow, R. M. & Carmel, R. (1987) Low

serum cobalamin levels occur frequently in the Acquired Immune Deciency Syndrome and related disorders. Eur. J. Haematol. 38: 141147. Centers for Disease Control (1987) Revision of CDC surveillance case denition for acquired immunodeciency syndrome. MMWR 36 (suppl. 1): 1S15S. Chandra, R. K. & Newberne, P. M. (1977) Nutrition Immunity and Infection. Mechanisms of Interactions. Plenum Press, New York, NY. Conover, W. J. (1980) Practical Nonparametric Statistics, 2nd ed. John Wiley & Sons, New York, NY. Coodley, G. O., Coodley, M. K., Nelson, H. D. & Loveless, M. O. (1993) Micronutrient concentrations in the HIV wasting syndrome. AIDS 7: 15951600. Cox, D. R. & Oakes, D. (1984) Analysis of Survival Data. Chapman and Hall, London, U.K. Dowling, S., Lambe, J. & Mulcahy, F. (1993) Vitamin B12 and folate status in human immunodeciency virus infection. Eur. J. Clin. Nutr. 47: 803807. Gibson, R. S. (1990) Principles of Nutritional Assessment, pp. 397404. Oxford University Press, New York, NY. Harriman, G. R., Smith, P. D., Horne, M. K., Fox, C. H., Koenig, S., Lack, E. E., Lane, C. & Fauci, A. S. (1989) Vitamin B12 malabsorption in patients with acquired immunodeciency syndrome. Arch. Intern. Med. 149: 20392041. Herbert, V., Jacobson, J., Shevchuk, O., Fong, W., Stopler, T., Castellar, L. & Tsougranis, M. (1989) Vitamin B12, folate and lithium in AIDS. Clin. Res. 37: 594A (abs.). Herzlich, B. C., Ranginwala, M., Nawabi, I. & Herbert, V. (1990) Synergy of inhibition of DNA synthesis in human bone marrow by azidothymidine plus deciency of folate and/or vitamin B12? Am. J. Hematol. 33: 177183. Herzlich, B. C., Schiano, T. D., Moussa, Z., Zimbalist, E., Panagopoulos, G. & Nawabi, I. (1992) Decreased intrinsic factor secretion in AIDS: relation to parietal cell acid secretory capacity and vitamin B12 malabsorption. Am. J. Gastroenterol. 87: 17811788. Heymseld, S. B., Tighe, A. & Wang, Z. M. (1994) Nutritional assessment by anthropometric and biochemical methods. In: Modern Nutrition in Health and Disease (Shils, M. E., Olson, J. A. & Shike, M., eds.), vol. 2, pp. 812841. Lea & Febiger, Philadelphia, PA. Howard, M. P., Andon, M. A. & Reynolds, R. D. (1984) Long-term stability of pyridoxal phosphate in frozen human plasma. Fed. Proc. 4: 486. Kaslow, R. A., Ostrow, D. G., Detels, R., Phair, J. P., Polk, B. F. & Rinaldo, C. R., Jr. (1987) The multicenter AIDS cohort study: rationale, organization, and selected characteristics of the participants. Am. J. Epidemiol. 126: 310318. Kieburtz, K. D., Giang, D. W., Schiffer, R. B. & Vakil, N. (1991) Abnormal vitamin B12 metabolism in human immunodeciency virus infection: association with neurological dysfunction. Arch. Neurol. 48: 312314. Mantero-Atienza, E., Baum, M. K., Morgan, R., Wilkie, F., Shor-Posner, G., Fletcher, M. A., Eisdorfer, C. & Beach, R. S. (1991) Vitamin B12 in early human immunodeciency virus-1 infection. Arch. Intern. Med. 151: 1019 1020. Ocke, M. C., Schrijver, J., Obermann-deBoer, G. L., Bloemberg, B. P., Haenen, G. R. & Kromhout, D. (1995) Stability of blood (pro)vitamins during four years of storage at 020 degrees C: consequences for epidemiologic research. J. Clin. Epidemiol. 48: 10771085. Rall, L. C. & Meydani, S. N. (1993) Vitamin B-6 and immune competence. Nutr. Rev. 51: 217225. Remacha, A. F., Montagud, M., Cadafalch, J., Riera, A., Martino, R. & Gimferrer, E. (1993) Vitamin B12 transport proteins in patients with HIV-1 infection and AIDS. Haematologica 78: 8488. Remacha, A. F., Riera, A., Cadafalch, J. & Gimferrer, E. (1991) Vitamin B12 abnormalities in HIV infected patients. Eur. J. Haematol. 47: 6064. Revell, P., ODoherty, M. J., Tang, A. & Savidge, G. F. (1991) Folic acid absorption in patients infected with the human immunodeciency virus. J. Intern. Med. 230: 227231. Robertson, K. R., Stern, R. A., Hall, C. D., Perkins, D. O., Wilkins, J. W., Gortner, D. T., Donovan, M. K., Messenheimer, J. A., Whaley, R. & Evans, D. L. (1993) Vitamin B12 deciency and nervous system disease in HIV infection. Arch. Neurol. 50: 807811. Rule, S.A.J., Hooker, M., Costello, C., Luck, W. & Hoffbrand, A. V. (1994) Serum vitamin B12 and transcobalamin levels in early HIV disease. Am. J. Hematol. 47: 167171. Sampson, D. A. & OConnor, D. K. (1989) Analysis of B-6 vitamers and pyridoxic acid in plasma, tissues and urine using high performance liquid chromatography. Nutr. Res. 9: 259272. Smith, I., Howells, D. W., Kendall, B., Levinsky, R. & Hyland, K. (1987) Folate deciency and demyelination in AIDS (letter to the editor). Lancet 2: 215. Snedecor, G. W. & Cochran, W. G. (1989) Statistical Methods, 8th ed. Iowa State University Press, Ames, IA. Veilleux, M., Paltiel, O. & Falutz, J. (1995) Sensorimotor neuropathy and abnormal vitamin B12 metabolism in early HIV infection. Can. J. Neurol. Sci. 22: 4346. Willett, W. C., Sampson, L. & Stampfer, M. J. (1988) The use of a self-administered questionnaire to assess diet four years in the past. Am. J. Epidemiol. 127: 188199.

Downloaded from jn.nutrition.org by guest on March 18, 2013

/ 4w13$$0022

01-24-97 09:42:36

nutra

LP: J Nut