|Home | About | Journals | Submit | Contact Us | Français|
In observational studies, adequate selenium status has been associated with better pregnancy outcomes and slowed HIV disease progression.
We investigated the effects of daily selenium supplements on CD4 cell counts, viral load, pregnancy outcomes, and maternal and infant mortality among 913 HIV-infected pregnant women.
In this randomized, double-blind, placebo-controlled trial, eligible women between 12 and 27 wk of gestation were given daily selenium (200 μg as selenomethionine) or placebo as supplements from recruitment until 6 mo after delivery. All women received prenatal iron, folic acid, and multivitamin supplements irrespective of experimental assignment.
The selenium regimen had no significant effect on maternal CD4 cell counts or viral load. Selenium was marginally associated with a reduced risk of low birth weight [relative risk (RR) = 0.71; 95% CI: 0.49, 1.05; P = 0.09] and increased risk of fetal death (RR = 1.58; 95% CI = 0.95, 2.63; P = 0.08), but had no effect on risk of prematurity or small-for-gestational age birth. The regimen had no significant effect on maternal mortality (RR = 1.02; 95% CI = 0.51, 2.04; P = 0.96). There was no significant effect on neonatal or overall child mortality, but selenium reduced the risk of child mortality after 6 wk (RR = 0.43; 95% CI = 0.19, 0.99; P = 0.048).
Among HIV-infected women from Dar es Salaam, Tanzania, selenium supplements given during and after pregnancy did not improve HIV disease progression or pregnancy outcomes, but may improve child survival. This trial was registered at clinical-trials.gov as NCT00197561.
HIV infection is a serious global public health problem, and >33 million persons are currently living with the infection (1). Sub-Saharan Africa is the geographic region most heavily affected by the HIV epidemic and accounts for about two-thirds of the global total of infections. Despite the progress toward increasing access to highly active antiretroviral therapy (HAART) in sub-Saharan Africa, only one-quarter of those in need receive it (2). Therefore, treatment modalities that prolong the need for HAART may be of great benefit in sub-Saharan Africa.
The trace element selenium has been proposed as a key nutrient among persons living with HIV (3). Biochemical selenium deficiency has been associated with increased mortality among those infected with HIV (4–7) and with accelerated HIV disease progression through increased viral load (8). Selenium’s role in antioxidant defense (9) and immunity (10) may be the underlying mechanism. Because advanced HIV disease is a risk factor for adverse pregnancy outcomes (11, 12), maternal selenium deficiency may also worsen pregnancy outcomes (13).
The interpretation of observational studies that used plasma selenium concentrations to determine selenium deficiency may be limited by confounding due to predictors of adverse outcomes that may also affect selenium concentrations, such as the acute-phase response to infection (14). The role of selenium status, especially among those experiencing infections, should thus preferably be assessed by using randomized, placebo-controlled supplementation trials. Trials conducted among HIV-infected populations from the United States showed that selenium supplements decreased hospital admissions (15) and suppressed viral load (8). The effect of selenium supplements among HIV-infected populations in sub-Saharan Africa is unclear.
We therefore conducted a randomized, double-blind, placebo-controlled trial of selenium among HIV-infected women enrolled during pregnancy to examine the effect of selenium supplementation on maternal HIV disease progression, pregnancy outcomes, and maternal and child survival.
Pregnant women seeking care at antenatal clinics in Dar es Salaam, Tanzania, were offered HIV-1 testing as part of the prevalent standard of care. Women who tested positive for HIV were referred to a study clinic located at Muhimbili National Hospital, where HIV status was confirmed. Those who were pregnant at between 12 and 27 wk of gestation, who were residents of Dar es Salaam at the time of random assignment, and who intended to stay in the city until at least 1 y after delivery were invited to give consent for participation in the trial.
At baseline, participants were randomly assigned to receive a daily tablet of 200 μg elemental selenium (as selenomethionine) or placebo throughout the antenatal period and after delivery until 6 mo postpartum. The proposed dose is one-half of the tolerable upper intake level of 400 μg/d for pregnant women and can thus be considered safe (16). Active and placebo tablets were indistinguishable in shape, size, and color and were packed in identical coded bottles. The study statistician (DS) based in Boston, MA, prepared a computer-generated randomization list in blocks of 20. Onsite study pharmacists stored the coded randomization list in a locked file cabinet and concealed allocation by covering the numeric regimen code on each bottle with stickers. After informed consent was obtained at the baseline visit, pharmacists assigned eligible woman to the active or placebo regimen on the basis of the randomization list. At every monthly clinic visit, a new bottle containing 45 tablets was given to each woman, and pills remaining in used bottles were counted to assess compliance. Study physicians, research nurses, and participants were unaware of the assignment groups. Tishcon Corp, Salisbury, MD, prepared the tablets but was not involved in the study design, implementation, analysis, or reporting of findings. The study protocol was approved by the Institutional Review Boards at Muhimbili University of Health and Allied Sciences and the Harvard School of Public Health.
At baseline, trained research nurses collected data on clients’ socioeconomic status, education, obstetric history, anthropometry, and current health status. They also obtained blood samples to measure immunologic, virologic, and hematologic parameters. Study physicians performed clinical examinations. During the women’s pregnancies, nurse interviews and physician examinations were conducted monthly.
Immediately after delivery, research midwives measured the weight of the neonates to the nearest 50 g with a standard mechanical scale. Midwives measured infant length with a length board and head circumference with a nonstretchable tape, both to the nearest 0.1 cm. Gestational age was based on the date of the last menstrual period, which was obtained at the time of random assignment.
At 6 wk postpartum, research nurses obtained data on child anthropometry and morbidity history, and study physicians examined the child. Child vital status was tracked until the discharge visit at 6 mo postpartum through monthly clinic visits. Data on child HIV status was not available during the study, but HIV testing became available outside of the study through our HIV Care and Treatment Program. Maternal blood samples were obtained at 6 wk and 6 mo postpartum, and maternal nurse interviews and physician examinations were conducted monthly.
We originally proposed to enroll 300 participants to examine the regimen effect on 4 primary endpoints: 1) HIV-1 viral load, 2) CD4 cell counts, 3) lower genital shedding of HIV-1 infected cells, and 4) risk of mastitis. Here, we present findings on viral load and T cell counts. In March 2004 we increased our sample size to 800 to examine with adequate statistical power the regimen effect on birth weight and pregnancy outcomes. Statistical power calculations were based on the two-sided t test with a Type I error rate of 5%, adjusted for a 10% dropout rate. With a sample size of n = 300, we had >99% power to detect a 0.25 change in log10-transformed viral load and 91% power to detect a 20-cell/mm3 change in CD4 cell counts (17–19). With n = 800 and SD = 522 g, we had 89% power to detect a 125-g difference in birth weight (20). We increased the sample size to 915 to further stabilize statistical power. Given the excellent statistical power for viral load with n = 300, we analyzed viral load only on a random subsample of participants with data on viral load, lower genital shedding of HIV-1 infected cells, and risk of mastitis. The sample size for each of these 3 endpoints was higher than the 300 sets with results on all 3 primary endpoints because additional unmatched results were also available. We hypothesized that the supplements would lower HIV-1 viral load, improve CD4 cell counts, improve pregnancy outcomes, and lower the risks of maternal and child mortality.
Participants were randomly assigned between September 2003 and July 2005 and were followed up until August 2006. During the study, a Data Safety Monitoring Board reviewed the results of the study 5 times regarding efficacy and safety of the primary endpoints. The Peto stopping boundary was used for early stopping with nominal P = 0.001 for efficacy endpoints and P = 0.05 for safety endpoints (21).
Study participants had access to standard prenatal and postnatal care, including nutritional counseling and support. Prenatal care included daily doses of ferrous sulfate (200 mg, equivalent to 60 mg ferrous Fe) and folic acid (0.25 mg) and malaria prophylaxis in the form of sulfadoxine-pyrimethamine tablets (Fansidar; Roche Pharmaceuticals, Nutley, NJ) at 20 and 30 wk of gestation. On the basis of earlier findings on the benefits of prenatal multivitamins among HIV-positive women (20), the participants received 20 mg thiamine, 20 mg riboflavin, 25 mg vitamin B-6, 100 mg niacin, 50 μg vitamin B-12, 500 mg vitamin C, 30 mg vitamin E, and 0.8 mg folic acid from the time of enrollment until delivery. These doses correspond with multiples of the recommended dietary allowances (RDAs). A subset of the 913 participants (n = 314; 156 in the selenium group and 158 in the placebo group) were co-enrolled in an efficacy trial comparing these vitamin doses with lower doses in line with the RDAs. To prevent transmission of HIV from mother to child, a 200-mg single dose of nevirapine was given orally to all women during labor, and a dose of 2 mg nevirapine/kg syrup was given to infants within 72 h of delivery (22). Access to HAART became increasingly available during the study period. A total of 31 women were initiated on HAART at some point during their follow-up; of these, 17 were in the selenium group and 14 in the placebo group (P = 0.53).
Complete blood cell counts were evaluated among women at baseline and postpartum at 6 wk and 6 mo. T-cell subsets counts (CD4, CD3, and CD8) were quantified with use of the FAC-Scount system (Becton-Dickinson, San Jose, CA), viral load with use of the Roche Amplicor version 1.5 assay (Roche Diagnostics Corporation, Indianapolis, IN), and hemoglobin concentrations with use of the CBC5 Coulter Counter (Coulter Corporation, Miami, FL).
The intention-to-treat principle was used for statistical analyses. For pregnancy outcomes, generalized estimating equations were used to account for the correlations owing to twinning (n = 22 pairs of twins) (23). Binary endpoints were modeled with a log link and binomial variance function to yield relative risks, whereas the identity link and the Gaussian variance function were used for continuous endpoints to yield comparisons of means (24). Mean differences in maternal T-cell counts and viral load between treatment groups were estimated with general linear models for repeated measurements (PROC MIXED; SAS Institute Inc, Cary, NC). In these models, a compound symmetry working correlation matrix with the empirical variance was used (23). Point estimates of postrandomization change in values and 95% CIs directly modeled the difference between repeated measures. P values were obtained through a parallel groups analysis, adjusted for baseline. For infant (singletons only) and adult mortality endpoints, Cox proportional hazards models were used to obtain hazard ratios and 95% CIs.
Compliance with the study regimen was calculated as the number of tablets absent from the returned bottles divided by the total number of tablets the participant should have taken. Breast-feeding status was determined at each postpartum visit by interview. All P values reported are two-sided; statistical significance was defined as P < 0.05. No adjustments for multiple comparisons or interim analyses were done. Statistical analyses were carried out by using SAS system version 9.1 (SAS Institute Inc). Unless otherwise noted, values are means ± SDs or percentages.
Of the 915 women randomly assigned, 1 woman had a gestational age at entry outside the allowable range of 12–27 wk and 1 woman was found to not be pregnant; both of these women were excluded from the final analyses (Figure 1). A total of 3 women died before delivery, and data on birth outcomes (miscarriage, stillbirth, or live birth) were not available for 18 women (13 in the selenium group and 5 in the placebo group; P = 0.06). Among women with live births, birth weight was unavailable for 14 women (7 in the selenium group and 7 in the placebo group; P > 0.99).
At the time of random assignment, women had a mean gestational age of 21.6 ± 3.4 wk and were on average 27.5 ± 4.9 y old. The intervals from the time of random assignment to delivery (4.0 ± 1.0 mo) and to 6 wk postpartum (5.4 ± 1.0 mo) did not differ significantly between treatment groups (P = 0.72 and P = 0.78, respectively). The proportion of participants with randomly selected viral load measurements was similar in each treatment group (P = 0.92). There were no relevant differences in baseline characteristics (Table 1). Mean compliance with the study regimen by 6 mo in the selenium (91.2 ± 6.3%) and placebo (91.7 ± 6.3%) groups did not differ significantly between groups (P = 0.10).
Selenium had no significant effect on absolute CD4, CD8, and CD3 cell counts or on viral load over the follow-up period (Table 2). The mean (± SD) birth weight in the selenium (3008 ± 543 g) and placebo (2982 ± 588 g) groups did not differ significantly (P = 0.76; Table 3). Selenium supplements had no significant effect on risks of low birth weight (P = 0.09), preterm birth (P = 0.93), and small for gestational age (P = 0.54).
Among the 913 women, 32 (3.5%) died during follow-up. Mortality rates were not significantly different between treatment groups (P = 0.96). There were 58 fetal deaths in the study; among them, there were 50 stillbirths and 8 miscarriages (Table 4). Selenium supplements tended to increase the risk of fetal death, but the results were not statistically significant (P = 0.08). Among singleton live births, 50/815 (6.1%) died during follow-up. The risk of infant death did not differ significantly by treatment group (RR = 0.64; 95% CI = 0.36, 1.13; P = 0.12). However, selenium supplements resulted in a lower risk of infant death after 6 wk (RR = 0.43; 95% CI = 0.19, 0.99; P = 0.048). The prevalence of breastfeeding in this cohort was 90%, and the median duration among those who breastfed was 16 wk (inter-quartile range: 12–23). These estimates did not differ significantly by treatment group.
In this randomized, double-blind, placebo-controlled trial, selenium supplements given during the antenatal and postpartum periods to HIV-infected Tanzanian women did not have significant effects on HIV-1 viral load, CD4 cell counts, pregnancy outcomes, or maternal and overall infant mortality. Several lines of laboratory evidence suggested beneficial effects of selenium in HIV infection. Selenium status may influence both the humoral and cell-mediated arms of immune function (25). Supplemental selenium increases expression of the T-cell high-affinity interleukin-2 receptor (26), up-regulates the activity of natural killer and cytotoxic T-cells (27), and increases interferon-γ production and T-helper cell counts (28).
Furthermore, selenium functions in oxidative defense, such as in the glutathione peroxidase (GSH-Px) enzyme system (9). Poor selenium status lowers GSH-Px activity (29, 30), which may lead to oxidative stress followed by apoptosis of T lymphocytes (27) and increased HIV replication rates (31, 32). During embryonic and fetal development, oxidative stress may damage DNA and cell membranes (33, 34).
Selenium supplements had no significant effects on pregnancy outcomes, yet marginally decreased the risk of low birth weight and marginally increased the risk of fetal death. The role of selenium status for pregnancy outcomes has only been examined in a few epidemiologic studies and remains unclear. In those studies, low maternal serum selenium concentrations were related to adverse pregnancy outcomes such as fetal death (13), neural tube defects (35), miscarriage (33), and poor weight gain during pregnancy (36).
Among HIV-infected adults, low plasma selenium concentrations are related to decreased survival (4, 5, 7). However, there is concern that these associations were confounded by the acute phase response to infection (14, 37, 38). A randomized, placebo-controlled trial from Miami, FL, examined the effect of 200 μg selenium given daily for 9 mo in primarily symptomatic HIV-infected adult men and women (8). Elevations in serum selenium concentrations related to selenium supplementation were associated with decreases in HIV-1 viral load, which in turn were related to improved CD4 cell counts. Selenium responders (defined as those with mean serum selenium >3 SDs above the mean change in the placebo group) had lower HIV-1 viral loads and higher CD4 counts than did the selenium nonresponder or placebo groups. However, interpretation of these findings is difficult, because extraneous factors that are related to both supplement use or response and outcome may have biased these findings. For instance, the beneficial effects of improvements in serum selenium concentrations, indicative of high supplement adherence, may have been confounded by higher than normal adherence to antiretroviral therapy (39, 40). In another trial from Miami, FL, 200 μg selenium decreased the number of hospitalizations among 186 HIV-infected participants over a 12-mo follow-up period (25). The supplements had no effect on mean CD4 cell counts over time, but reduced the risk of a CD4 cell count decline >50 cells/mm3.
Similar to evidence from adults, low plasma selenium concentrations are related to poor survival among HIV-infected children (6, 41). In this study, selenium reduced the risk of child mortality after 6 wk. This reduction may be due to improved child selenium status as a result of increased supply through the placenta, umbilical cord (42), and breast milk (43).
Several explanations are possible for the lack of effect of selenium on maternal and pregnancy outcomes. All participants received supplements containing B-complex vitamins, vitamin C, and vitamin E at multiples of the RDAs, which may have limited the effect of selenium. The intervention may only be effective among patients with advanced HIV disease or those receiving HAART; in our study, participants were primarily asymptomatic at baseline and HAART use was uncommon. Findings from a previous study in Dar es Salaam among HIV-infected pregnant women (13) showed that only 2% of participants had low plasma selenium concentrations (<85 μg/L). Despite uncertainty about selenium requirements (44), selenium deficiency is thus likely to be uncommon in our setting. This may be due to adequate intake of selenium-rich foods, such as plant foods grown in selenium-containing soils, or animal foods such as seafood (16).
Our study is the largest selenium supplementation trial conducted to date and included rigorous design and analysis methods. Even though there may be some benefits for children, we did not find benefits of selenium supplements on maternal disease progression and mortality. Therefore, there is no support for providing selenium supplements to HIV-infected populations naïve to HAART, who receive high-dose multivitamin supplements, and who live in areas where selenium deficiency is likely to be rare.
We thank the mothers and children and the members of the research team, including physicians, nurses, midwives, supervisors, laboratory staff, and administrative staff, who made the study possible. We greatly appreciate the input of the following colleagues: Illuminata Ballonzi, Juliana Mghamba, Megan O’Brien, Antje Hebestreit, Paul Petraro, Ellen Hertzmark, and Heavengton Mshiu. We thank the Permanent Secretary, Ministry of Health, and the officials at Muhimbili University of Health and Allied Sciences, Muhimbili National Hospital, the City of Dar es Salaam Regional Health Authority, and the National AIDS Control Program for their support.
2Supported by the National Institute of Child Health and Human Development (NICHD R24 043555-05).
The contributions of the authors were as follows—RK: analyzed and interpreted the data and wrote the initial draft of the manuscript; FM and WWF: are principal investigators of the trial and contributed to the study design and its implementation; GIM: assisted in the trial design and with JLF in data interpretation; SA: supervised the laboratory analysis; DS: provided statistical guidance in the design of the study and in data analyses. All coauthors participated in the manuscript preparation. The authors did not have a conflict of interest.