We examined the cross-sectional relationships between malaria parasitemia and CD4 T cell count and viral load among human immunodeficiency virus (HIV)-infected pregnant women. We then followed women to investigate whether or not baseline parasitemia predicted CD4 T cell counts or viral loads > 90 days post-baseline or predicted time to HIV disease stage 3 or 4 or acquired immune deficiency syndrome (AIDS)-related death (ARD). Parasitemia level was nonlinearly associated with viral load at baseline and among measurements taken > 90 days post-baseline; women with low baseline parasitemia, versus none, had higher viral loads at both time points. Any baseline parasitemia predicted an increased rate of ARD among women with baseline CD4 T cell counts > 500 cells/μL (ratio rate [RR] = 2.6; 95% confidence interval [CI] = 1.1-6.0; P test for heterogeneity = 0.05). Further study is warranted to determine whether or not parasitemia is especially detrimental to individuals with lower levels of immunosuppression or chronic low parasitemia. Copyright © 2010 by The American Society of Tropical Medicine and Hygiene.
The study population consisted of 1,078 HIV-infected pregnant women who enrolled in a double-blind, placebo-controlled randomized trial at a participating antenatal clinic in Dar es Salaam, Tanzania, to examine the effects of daily micronutrient supplements on HIV disease progression and mortality. Trial results and a detailed description of the trial design have been published elsewhere.19,20 Ethical approvals for the trial were obtained from Research and Publications Committee of Muhimbili University College of Health Sciences, the Ethical Committee of the Tanzania National AIDS Control Program, and the Institutional Review Board of the Harvard School of Public Health. Informed consent was obtained from all women. The study enrollment took place from April 1995 through July 1997, a time during which women did not have access to ART in Tanzania. Women were followed until August 2003. Malaria is endemic in Dar es Salaam, and stable transmission occurs all year. The national annual incidence of malaria disease was estimated to be 400–500 per 1,000 in the general population in 2003.21 Bed nets were reported to have been used by all or some children in an estimated 52% of urban Tanzanian households in 1999; however, only a small percentage of these bed nets (10%) had been previously treated with insecticides.22 After randomization, women completed baseline interviews regarding socio-demographic characteristics and medical history, and they were asked to provide blood, stool, and vaginal-fluid specimens for detection of malaria parasites, intestinal parasites, and sexually transmitted infections. Follow-up consisted of monthly clinic visits throughout pregnancy and thereafter for a minimum of 2 years. To diagnose malaria, thick and thin blood films were air-dried and stained with 5% Giemsa at pH 7.2 for 20 minutes. Presence of asexual Plasmodium falciparum malaria parasites was determined by microscopic examination of stained slides.23 A slide was considered negative when no parasites were detected in the process of counting 200 leukocytes on a blood film. Quality control for smear microscopy was ensured through multiple mechanisms. First, known negative and positive control slides were included in every microscopic examination of stained slides. Second, all results from microscopic examination of stained slides were verified by a second testing laboratory technologist, and any discrepant results were resolved by a third senior laboratory technologist. The study laboratory also participated in the World Health Organization (WHO)/National Institute for Communicable Diseases (NICD) proficiency testing program. Women with malaria parasites and other infections received treatment, free-of-charge, in accordance with the Tanzania Ministry of Health treatment guidelines. Chloroquine was the first-line drug for treatment of uncomplicated malaria until August 2001, when it was changed to sulfadoxine-pyrimethamine because of high levels of treatment failure with chloroquine.21,24 A blood specimen was requested at baseline and every 6 months thereafter for the enumeration of CD4 T cell counts using the FACSCount system (Becton Dickinson, San Jose, CA). For a random sample of 415 women, plasma viral load was quantified at a minimum of one time point using the Roche Amplicor HIV-1 monitor version 1.5 assay (Roche Diagnostics Corp., Indianapolis, IN), which has a detection limit of 400 copies/mL. For these analyses, results below this limit (1.6% of all viral load results) were assigned a value of 399. A positive association between parasitemia and viral load has been previously reported in a subset of these women.15 At each follow-up visit, clinicians provided routine clinical care and updated data regarding HIV disease stage, death, and cause of death. HIV disease stage was evaluated in accordance with the WHO criteria on the basis of the woman’s history and physical examination.25 Verbal autopsy techniques were used to approximate the cause of death by conducting standardized interviews with relatives, reviewing medical records, or both. Deaths caused by the following conditions were considered to be because of or related to AIDS: AIDS, tuberculosis (pulmonary or extrapulmonary), anemia, meningitis, stroke, pneumonia, diarrhea, and fever. For women who did not attend the clinic or who traveled out of Dar es Salaam, a home visit was made, and neighbors or relatives were asked about the woman’s vital status. Data analysis consisted of three main parts. The first part consisted of cross-sectional analyses of baseline data to investigate the relationship between parasitemia and continuous CD4 T cell counts and viral loads using generalized linear regression models. Next, we used repeated-measures generalized linear models with an exchangeable correlation structure to investigate the relationship between baseline parasitemia and CD4 T cell count and viral load measurements taken > 90 days after baseline. Stepwise splines were used to model the time from baseline to the CD4 T cell count or viral-load measurement in a nonlinear fashion.26 The goal of this analysis was to examine if any relationship between parasitemia and CD4 T cell count or viral load persisted among measurements taken at least several months after antimalarial treatment. Last, we used Cox proportional-hazards regression models to examine the association between baseline parasitemia and time to progression to HIV disease stage 3 or 4 and time to AIDS-related death.27 Women with a baseline HIV disease stage 3 or 4 were excluded from the time to HIV disease stage 3 or 4 analysis, because they had already experienced the outcome of interest. Women for whom the cause of death was not determined (N = 54; 15.9% of all deaths) or for whom death was deemed unrelated to AIDS (N = 44; 12.9% of all deaths) were censored at the time of death. For all analyses, we examined baseline parasitemia as a binary variable (the presence of any versus none) and as a categorical variable (none; low = 1–999 parasites/µL; medium = 1,000–10,000 parasites/µL; or high > 10,000 parasites/µL). The categorical variable was tested for linearity using a likelihood ratio test with two degrees of freedom. We considered the following baseline characteristics to be potential confounders if they predicted the outcome in a univariable regression model at a P value ≤ 0.20: maternal age, gestational age, body mass index (BMI; kg/m2), mid-upper arm circumference, year of recruitment, WHO HIV disease stage, primiparity, medical antecedents, presence of coprevalent parasitic and sexually transmitted infections, and socio-demographic characteristics (education level, marital status, per person daily food expenditure, and reliance on others for financial support). We did not consider multivitamin use as a potential confounder in this study, because we expected that randomization would have yielded comparable exposure to the multivitamin interventions for women with and without baseline parasitemia. Final multivariable models included all potential confounders that changed the effect estimate by ≥ 10% in either direction as well as other established risk factors for the outcomes. Viral loads were transformed to the log10 scale. The missing-indicator method was used to account for missing covariate data. We excluded extreme outlying values for log10 viral load and CD4 T cell counts as well as values for individuals with outlying within-person standard deviations for these outcomes. Because we hypothesized that the effect of parasitemia on all outcomes could differ according to host immunological status, we stratified each final model by baseline CD4 T cell count group (< 200 cells/µL, 200–499 cells/µL, ≥ 500 cells/µL) and tested for heterogeneity across strata using Cochran's Q test.28
N/A