Administration of antiretroviral drugs during pregnancy has significantly reduced the mother-to-child transmission (MTCT) rate of HIV-1 [1,2], but potential toxicities from in-utero exposure to combination antiretroviral drugs remain a cause for concern [3,4]. Although some studies have supported the safety to infants from short-term prophylactic perinatal antiretroviral drugs [5,6], other studies conducted in Europe and North America have demonstrated a significant (although not necessarily clinically concerning) effect of maternal antiretroviral combination therapy [highly active antiretroviral therapy (HAART)] on infant hematopoiesis [4,7–11]. No previous studies have examined toxicities associated with postnatal exposure to maternal HAART in breastfeeding infants, however. As access to antiretroviral drugs increases throughout the developing world to regions where breastfeeding is common, and as maternal HAART is studied as a potential approach in preventing breastfeeding-related MTCT, it is critical to study the short-term and long-term infant toxicities associated with exposure to antiretroviral agents both in utero and during breastfeeding. Such toxicities should also be considered within the context of reference values for children from developing countries, recognizing that hematologic values may be lower for African children [12–14].
We performed a nested cohort study within a randomized clinical trial that studied the prevention of mother-to-child transmission (PMTCT) interventions in Botswana to investigate short-term hematologic and hepatic toxicities (up to 7 months of infant age) associated with antenatal and postnatal exposure to maternal HAART.
We performed a nested cohort study to compare hematologic and hepatic toxicities in infants who had or had not been exposed to maternal HAART. All women participated in a randomized clinical trial at four sites in Botswana for the prevention of MTCT, previously described in detail [15,16]. In brief, the Mashi Study (‘Mashi’ means ‘milk’ in Setswana) enrolled 1200 HIV-1-infected pregnant women at 34 weeks' gestation between March 2001 and October 2003. HAART became available to study participants 19 months after the study opened, and was offered to women with a CD4 cell count less than 200 cells/μl or an AIDS-defining illness, and to infants meeting Centers for Disease Control and Prevention (CDC) treatment criteria. Women not on HAART received zidovudine prophylaxis from 34 weeks' gestation until delivery, and all infants received 1 month of zidovudine prophylaxis. We did not include the presence of AIDS-defining illness as a selection criterion for the HAART-unexposed group because nearly all of the women in the Mashi study received HAART based on their CD4 cell counts less than 200. Additionally, using a 2 × 2 factorial design, mother–infant pairs were randomized at enrollment to receive either to single-dose nevirapine or placebo perinatally, and either formula feed with 1 month of infant zidovudine prophylaxis or to exclusively breastfeed for 6 months, with infant zidovudine prophylaxis when breastfeeding. Infant zidovudine was discontinued prematurely among infants with a confirmed grade 3 or higher laboratory abnormality; who were confirmed to be HIV positive; or who were more than 4 weeks of age, assigned to breastfeed, and not receiving breast milk. All infants confirmed to be HIV positive received oral cotrimoxazole prophylaxis.
In August 2002, the first part of the study was modified to provide all infants with single-dose nevirapine. First line HAART to eligible mothers and infants consisted of zidovudine, lamivudine, and nevirapine.
Within the cohort study reported here, all live-born infants of women who had initiated HAART before delivery were selected for analysis. A comparison group was selected, consisting of all live-born infants of women who would have qualified for HAART (CD4+ cell counts < 200 cells/μl) but who did not start HAART before 6 months postpartum due to the delay in its availability within this study. Women in the HAART-exposed group received zidovudine (300 mg twice daily), lamivudine (150 mg twice daily), and nevirapine (200 mg twice daily) and supplemental zidovudine during labor (300 mg every 3 h); women who started HAART after 34 weeks' gestation but before delivery received zidovudine prophylaxis until the time of HAART initiation. Women in the HAART-unexposed group received zidovudine (300 mg twice daily) starting 34 weeks of gestation through delivery, and supplemental zidovudine (300 mg every 3 h) and either single dose nevirapine or placebo during labor. Both HAART-exposed and HAART-unexposed groups were subdivided into two subgroups according to assigned feeding strategy. Thus, all women received zidovudine either as part of their HAART or as chemoprophylaxis during the last 6 weeks of pregnancy.
The study and amendments were approved by ethics committees at the Botswana Health Research Unit and the Harvard School of Public Health. The study was performed in accordance with regulations of the Department of Health and Human Services for the protection of human research participants.
A complete blood cell count (CBC) was obtained for all infants at birth and at 1, 4, and 7 months of age; an additional CBC was obtained in infants assigned to breastfeed at 2, 3, 5 and 6 months of age. Aspartate aminotransferase (AST) and alanine aminotransferase (ALT) levels were measured at 1 month of age for all infants, and at 6 months of age in infants assigned to breastfeed.
Toxicities were defined as laboratory abnormality of grade 3 or 4 as specified in the Division of AIDS (DAIDS) Toxicity Table for Grading Severity of Adverse Infant/Pediatric Laboratory Experiences published in 1994 [17,18]. When possible, grade 3 and 4 laboratory toxicities were confirmed with repeat testing. Although values for African children have not been established on the basis of widespread surveys, several studies have suggested differences [12–14].
Statistical analyses were performed using Stata (version 9.0; StataCorp LP, College Station, Texas, USA) and SAS (version 9.1, SAS Institute, Cary, North Carolina, USA). The distributions of continuous maternal and infant baseline characteristics between groups were compared using the Wilcoxon rank-sum test, whereas Fisher's exact test was performed to compare categorical infant baseline characteristics as well as the proportion of infants with abnormal hematologic or hepatic values. The log-rank test was performed to compare hematologic toxicity over the first 7 months of life between groups. Hazard ratio estimation was performed using Cox regression. The following covariates were explored in univariable Cox models: HAART exposure, feeding assignment, maternal age, maternal CD4, maternal plasma HIV-1 RNA viral load, length of antenatal zidovudine or HAART exposure, sex of the infant, infant prematurity, clinic site, year of birth of the infant, infant birth weight, gestational age, and current infant HIV status. Multivariable model building utilized a backwards elimination procedure that retained effect modifiers. P values less than 0.05 were considered to be statistically significant, and no adjustments for multiple comparisons were made. Within the cohort study, assuming a 1% rate of toxicity within the non-HAART group (n = 109), this study's sample size had 83% power to detect (by Fisher's exact test) a 10-percentage point increase in the HAART group (n = 69).
Sixty-nine infants born to women who initiated HAART before delivery were included in analyses; of these, 34 infants were assigned to breastfeed and 35 infants to formula feed (Fig. 1). The comparison group consisted of 109 infants born to women with CD4+ cell counts less than 200 cells/μl at the time of enrollment and naïve to HAART through 6 months postpartum; 50 of these infants were assigned to breastfeed and 59 infants to formula feed. Most factors, including the median maternal baseline CD4+ cell counts and HIV-1 RNA levels, were similar between the HAART-exposed and HAART-unexposed groups (Table 1). The median gestational period at birth, however, was lower in the HAART-unexposed group compared with the HAART-exposed group.
Two (2.9%) infants in the HAART-exposed group and 21 (19.3%) infants from the HAART-unexposed group were infected with HIV-1 by 7 months of age. The MTCT rate of HIV-1 was lower in the HAART-exposed group than the HAART-unexposed group at all of the specified time points (birth, at 1 month, between 2 and 7 months, and cumulatively by 7 months). The difference was statistically significant only between 2 and 7 months and cumulatively by 7 months (P = 0.04 and 0.001, respectively). Six (8.7%) infants (one infected with HIV-1) from the HAART-exposed and 11 (10.1%) infants (four infected with HIV-1) from the HAART-unexposed group died before 7 months of age.
Infant hematologic abnormalities by maternal HAART exposure status and assigned feeding strategy are summarized in Table 2. Proportions with grade 3 or higher hematologic abnormalities were compared between infants exposed and unexposed to maternal HAART. No difference in hematologic abnormalities was observed at birth. However, a significantly higher proportion of infants in the HAART-exposed group experienced neutropenia than those in the HAART-unexposed group at 1 month of age (15.9 vs. 3.7%, respectively; P = 0.006), and this significance persisted when cumulative hematologic toxicities from birth to 1 month of age were compared (P = 0.002). There were no significant differences in the proportions experiencing grade 3 or 4 anemia or thrombocytopenia by HAART exposure status, and no infant experienced more than one type of hematologic toxicity at a time.
As the breastfed infants in the HAART-exposed group were exposed to additional antiretroviral drugs through breast milk, the two groups were further stratified according to assigned feeding strategy (Table 2). Among breastfed infants, those in the HAART-exposed group– who had been exposed to maternal HAART through breastfeeding as well as in utero– experienced significantly more neutropenia than the HAART-unexposed, breastfed children at 1 month (23.5% vs. 6.0%, respectively; P = 0.044) as well as cumulatively through 1 month (29.4 vs. 8.0%, respectively; P = 0.012). Among formula-fed infants only, there was no significant difference in proportions experiencing neutropenia between infants in HAART-exposed and HAART-unexposed groups. Within the subgroup of 94 babies assigned to formula feed, assuming a 1% rate of toxicity within the non-HAART group (n = 59), this study's sample size had 79% power to detect (by Fisher's exact test) a 15-percentage point increase in the HAART group (n = 35) (i.e., 16% rate of toxicity). The sample size in the study was not large enough to establish statistical significance. Other than neutropenia, no other significant differences in hematologic enzyme abnormalities were noted between HAART-exposed and HAART-unexposed infants within each of the breastfed and formula-fed strata.
From 2 to 7 months of age, the cumulative hematologic toxicities from 2 to 7 months of age were not significantly associated with exposure to maternal HAART among breastfed infants. No hematologic toxicities were observed in formula-fed infants during the same period (as noted in the Methods section, CBC was performed six times in breastfed infants compared with twice in formula-fed infants, and thus direct comparison between the two feeding groups is not advisable).
Despite the association between infant neutropenia and maternal HAART exposure during the first month of life, absolute neutrophil count (ANC) values in median infant and maternal HAART exposure were not significantly associated during the same period (Fig. 2). The median ANC levels in the HAART-exposed and HAART-unexposed groups were 6300 and 7300 counts at birth (P = 0.43) and 1700 and 1900 counts at 1 month of age (P = 0.14), respectively. Breastfed infants had lower ANC levels than formula-fed infants in both HAART-exposed and HAART-unexposed groups. This result is expected as breastfed infants received a 6-month prophylactic zidovudine therapy compared to 1-month regimen for formula-fed infants, and neutropenia is a well known adverse effect of zidovudine.
Association between maternal HAART exposure and neutropenia was further analyzed using time-to-event methods. HAART exposure, assignment to breastfeed, and higher maternal viral load at enrollment were each statistically significantly associated with neutropenia. Controlling for maternal HIV-1 viral load and feeding assignment (which also controls for the different schedule of CBC between the breastfed and formula-fed infants), exposure to maternal HAART had a hazard ratio of 3.1 for neutropenia compared with those without HAART exposure (95% CI; 1.4–6.9).
The cumulative proportion of infants experiencing neutropenia is shown in Fig. 2. At 7 months of age, a significantly higher proportion of HAART-exposed infants have experienced neutropenia compared with HAART-unexposed infants (P = 0.007). When stratified by assigned feeding strategy, the same pattern was observed (P = 0.042 among breastfed infants, and P = 0.064 among formula-fed infants). In accordance with the results shown in Table 2, the difference in the proportions of infants experiencing neutropenia was concentrated mostly between birth and 1 month of age and did not seem to persist thereafter.
All grade 3 or 4 neutropenias were resolved within a month; no infant experienced neutropenia in two consecutive months. Out of 21 infants who experienced neutropenia by 1 month of age, three infants (one at birth and two at 1 month of age) experienced a grade 4 neutropenia. All three infants had not been exposed to maternal HAART. In addition, clinically significant toxicities were rare in our study. Out of 37 infants in this nested cohort study, who experienced one or more hematologic toxicity, only one infant exhibited clinical symptoms consistent with severe anemia in the breastfed HAART-exposed group. All neutropenic and thrombocytopenic infants as well as two infants with hepatic toxicities were clinically asymptomatic.
Proportions of infants with grade 3 or higher hepatic enzyme abnormalities were compared between infants exposed and unexposed to maternal HAART. No statistically significant difference in hepatic enzyme abnormalities was observed between the groups at 1 month of age. One (1.4%) infant exposed to maternal HAART had an abnormal ALT level, whereas no infant in the HAART-unexposed group experienced an ALT abnormality (P = 0.39). Two (2.9%) infants in the HAART-exposed group and no infant in the HAART-unexposed group experienced AST abnormality (P = 0.15). Of the two infants experiencing hepatic toxicities, one had both ALT and AST levels of grade 4, and the other had a grade 3 AST level. Both infants were infected with HIV, formula-fed, exposed to maternal HAART in utero, and had not initiated infant HAART at the time of sample draw. At 6 months, no infant in the study experienced a hepatic enzyme abnormality.
Statistical significance in the above analyses did not change when HIV-infected infants were excluded; therefore, all infants were included in the analyses.
To our knowledge, this is the first study to investigate infant toxicities potentially associated with exposure to maternal HAART during breastfeeding. Our results suggest that hematologic and hepatic toxicities associated with antenatal and postnatal exposure to maternal HAART are minimal even in the context of prolonged infant zidovudine prophylaxis, with the exception of increased early neutropenias. The significant association between early neutropenias and breastfeeding suggests that postnatal exposure to antiretroviral drugs during breastfeeding is a plausible mechanism for infant toxicity, and may exacerbate toxicity caused by in-utero HAART and infant prophylactic zidovudine exposure.
Our findings were similar to previous studies of short-term effects of in-utero exposure to antiretroviral drugs on hematologic measures and toxicities. In the Women and Infant Transmission Study, conducted in the United States and Puerto Rico, a statistically significant difference in neutrophil, lymphocyte, CD4+ cell count, and CD8+ cell count was found between infants exposed to antiretroviral combination therapy and those exposed to antiretroviral monotherapy up to 2 months of age . With the exception of CD8+ cell counts, these differences did not persist in infants 6–24 months age. In the French Perinatal Study , neutrophil, lymphocyte, and platelet counts in infants with perinatal exposure to antiretroviral combination therapy were significantly lower than those with exposure to antiretroviral monotherapy only up to 15 months of age. In another study  conducted in Germany, a significantly higher proportion of infants exposed to maternal HAART experienced grade 2 or higher anemia and neutropenia than those who had been exposed to a mono or dual therapy at 1 month of age. However, this difference did not persist beyond 1 month of age. A study  conducted in Latin America demonstrated that proportion of neutropenic infants at birth, the incidents of neutropenia, exposure to maternal HAART containing a protease inhibitor were associated with a lower risk of abnormal hemoglobin values than exposure to mono or dual therapy within 6 days of birth. However, as in other studies, this association was not observed by 6 months of age. Infants in all of the studies cited here were formula-fed.
In the European Collaborative Study (ECS), a long-term effect of antiretroviral drug exposure to neutrophil counts was investigated. The study showed that those with perinatal exposure to prophylactic antiretroviral drugs had consistently lower neutrophil counts up to 8 years of age than those with no exposure to antiretroviral drugs, although the clinical significance of this association is unclear . These results suggest that exposure to maternal HAART may be associated with an increased risk of long-term hematologic abnormalities compared with exposure to prophylactic antiretroviral drugs for PMTCT, and a further study is needed to investigate this possibility in both laboratory and clinical contexts.
Unlike previous studies, our analysis includes breastfeeding as a potential risk factor for infant toxicities. The increased early neutropenias among breastfed infants as compared with formula-fed infants in our study is of interest. It suggests a potential role for direct antiretroviral drug toxicity by breastfeeding, which had only been a hypothetical risk previously. Zidovudine is the most likely agent to be associated with infant neutropenias, although an association with nevirapine has been described in infants as well . Zidovudine achieves concentrations in breast milk that are approximately three-fold those in serum , although the total amount of zidovudine ingested through breastfeeding is expected to be less than 1% of the standard oral dose. Although a pharmacokinetic substudy was conducted among infants of breastfeeding women receiving HAART in the Mashi cohort , these infants were all receiving oral zidovudine. Therefore, the contribution of breastfeeding to their serum zidovudine levels could not be measured, and warrants further investigation.
All toxicities in this study were defined by the DAIDS toxicity tables, as specified by the Mashi study protocol. However, recent studies have shown that hematologic measures of African children are generally lower than those of Whites, and thus the DAIDS toxicity tables may not be applicable for African children. A study of ANC in children born to HIV-uninfected mothers in Zimbabwe, a country neighboring Botswana, has demonstrated that mean ANC for Zimbabwean children were less than half of the mean ANC for American children . Another study showed that ANC were substantially and significantly lower among Ugandan children compared with White children . These data suggest a high proportion of grade 3 or 4 neutropenia (11.7%) observed in this study and lack of its clinical correlation may be explained in part by this differential immunohematologic profile of African children compared with that of White children.
The present study had limitations. First, though matched by baseline characteristics, HAART was administered on the basis of need for treatment and therefore its assignment was not randomized. In addition, the small sample size, comprising 69 mother—infant pairs in the HAART-exposed group and 109 pairs in the HAART-unexposed group, could hinder interpretation by leading to an underestimation of toxicities events. The differential schedule of laboratory testing between formula-fed and breastfed infants after 1 month of age curtailed our ability to compare laboratory toxicities by feeding group at these later time points. The median duration of antenatal HAART exposure in our study was short (34 days), and may underestimate hematologic toxicities potentially associated with a longer duration of antiretroviral drug exposure in utero. Finally, all the infants in this study were receiving a zidovudine prophylaxis, and therefore our findings may not be applicable to those infants only receiving a single-dose nevirapine. These limitations are important, as undetected infant toxicities could pose a real danger to HAART-exposed infants, and ongoing monitoring for these toxicities is warranted.
In conclusion, we demonstrated that infants exposed to maternal HAART in utero and who receive 1 month of infant zidovudine prophylaxis may be at a higher risk for transient, asymptomatic neutropenia up to 1 month of age. Although excess neutropenias from HAART exposure did not persist beyond 1 month, we found that exposure to maternal HAART during the first month of breastfeeding was a potential mechanism for infant toxicity that requires further study. No clinical associations with early neutropenias were detected, and excess anemias related to either in utero or breastfeeding HAART exposure were not detected. These findings provide reassurance for women receiving HAART who might expose their infants to antiretroviral drugs in utero and through breastfeeding, and support larger clinical trials to further evaluate the safety of maternal antenatal and postnatal HAART among exposed infants.
We would like to thank the patients who participated in this study, the entire study team, the members of the Data Safety Monitoring Board, the Botswana Ministry of Health (antiretroviral medications and monitoring for participants receiving antiretroviral treatment), Boehringer Ingelheim (single-dose nevirapine), GlaxoSmithKline (zidovudine prophylaxis), UNICEF, and the Eunice Kennedy Shriver National Institute of Child Health and Human Development (R01 HD37793 and K23 HD049292).
Author contributions: W.H.B., C.W., R.L.S., S.L., I.T., and M.E. designed the study. W.H.B., C.W., and K.O. collected data. W.H.B. and L.M.S. carried out statistical analysis. W.H.B. and C.W. drafted the paper. All authors commented on the manuscript and approved the final version.
The present study was funded by the Eunice Kennedy Shriver National Institute of Child Health and Human Development (R01 HD37793 and K23 HD049292).
1. Connor EM, Sperling RS, Gelber R, Kiselev P, Scott G, O'Sullivan MJ, et al
. Reduction of maternal-infant transmission of human immunodeficiency virus type 1 with zidovudine treatment. Pediatric AIDS Clinical Trials Group Protocol 076 Study Group. N Engl J Med 1994; 331:1173–1180.
2. Cooper ER, Charurat M, Mofenson L, Hanson IC, Pitt J, Diaz C, et al
. Combination antiretroviral strategies for the treatment of pregnant HIV-1-infected women and prevention of perinatal HIV-1 transmission. J Acquir Immune Defic Syndr 2002; 29:484–494.
3. Brogly SB, Ylitalo N, Mofenson LM, Oleske J, Van Dyke R, Crain MJ, et al
. In utero nucleoside reverse transcriptase inhibitor exposure and signs of possible mitochondrial dysfunction in HIV-uninfected children. AIDS 2007; 21:929–938.
4. Pacheco SE, McIntosh K, Lu M, Mofenson LM, Diaz C, Foca M, et al
. Effect of perinatal antiretroviral drug exposure on hematologic values in HIV-uninfected children: An analysis of the women and infants transmission study. J Infect Dis 2006; 194:1089–1097.
5. Culnane M, Fowler M, Lee SS, McSherry G, Brady M, O'Donnell K, et al
. Lack of long-term effects of in utero exposure to zidovudine among uninfected children born to HIV-infected women. Pediatric AIDS Clinical Trials Group Protocol 219/076 Teams. JAMA 1999; 281:151–157.
6. Mofenson LM, Munderi P. Safety of antiretroviral prophylaxis of perinatal transmission for HIV-infected pregnant women and their infants. J Acquir Immune Defic Syndr 2002; 30:200–215.
7. Le Chenadec J, Mayaux MJ, Guihenneuc-Jouyaux C, Blanche S. Perinatal antiretroviral treatment and hematopoiesis in HIV-uninfected infants. AIDS 2003; 17:2053–2061.
8. Bunders MJ, Bekker V, Scherpbier HJ, Boer K, Godfried M, Kuijpers TW. Haematological parameters of HIV-1-uninfected infants born to HIV-1-infected mothers. Acta Paediatr 2005; 94:1571–1577.
9. Feiterna-Sperling C, Weizsaecker K, Buhrer C, Casteleyn S, Loui A, Schmitz T, et al
. Hematologic effects of maternal antiretroviral therapy and transmission prophylaxis in HIV-1-exposed uninfected newborn infants. J Acquir Immune Defic Syndr 2007; 45:43–51.
10. European Collaborative Study. Levels and patterns of neutrophil cell counts over the first 8 years of life in children of HIV-1-infected mothers
11. Thorne C, Newell ML. Safety of agents used to prevent mother-to-child transmission of HIV: is there any cause for concern? Drug Saf 2007; 30:203–213.
12. Bunders M, Lugada E, Mermin J, Downing R, Were W, Thorne C, Newell ML. Within and between race differences in lymphocyte, CD4+, CD8+ and neutrophil levels in HIV-uninfected children with or without HIV exposure in Europe and Uganda. Ann Trop Paediatr 2006; 26:169–179.
13. Wells J, Shetty AK, Stranix L, Falkovitz-Halpern MS, Chipato T, Nyoni N, et al
. Range of normal neutrophil counts in healthy Zimbabwean infants: implications for monitoring antiretroviral drug toxicity
. J Acquir Immune Defic Syndr 2006; 42:460–463.
14. Bunders M, Cortina-Borja M, Newell ML. Age-related standards for total lymphocyte, CD4+ and CD8+ T cell counts in children born in Europe. Pediatr Infect Dis J 2005; 24:595–600.
15. Shapiro RL, Thior I, Gilbert PB, Lockman S, Wester C, Smeaton LM, et al
. Maternal single-dose nevirapine versus placebo as part of an antiretroviral strategy to prevent mother-to-child HIV transmission in Botswana. AIDS 2006; 20:1281–1288.
16. Thior I, Lockman S, Smeaton LM, Shapiro RL, Wester C, Heymann SJ, et al
. Breastfeeding plus infant zidovudine prophylaxis for 6 months vs formula feeding plus infant zidovudine for 1 month to reduce mother-to-child HIV transmission in Botswana: a randomized trial: the Mashi Study. JAMA 2006; 296:794–805.
17. Division of AIDS (DAIDS) toxicity
table for grading severity of pediatric (≤3 months of age) adverse experiences. http://rcc.tech-res.com/DAIDS%20RCC%20Forms/ToxicityTables_Pediatric_Under3MonthsOld_VPRP_v02.pdf
. [Accessed 18 February 2008].
18. Division of AIDS (DAIDS) toxicity
table for grading severity of pediatric (>3 months of age) adverse experiences. http://rcc.tech-res.com/DAIDS%20RCC%20Forms/ToxicityTables_Pediatric_Over3MonthsOld_VPRP_v03.pdf
. [Accessed 18 February 2008].
19. Mussi-Pinhata MM, Rego MA, Freimanis L, Kakehasi FM, Machado DM, Cardoso EM, Read JS. Maternal antiretrovirals and hepatic enzyme, hematologic abnormalities among human immunodeficiency virus type 1-uninfected infants: the NISDI perinatal study. Pediatr Infect Dis J 2007; 26:1032–1037.
20. Shetty AK, Coovadia HM, Mirochnick MM, Maldonado Y, Mofenson LM, Eshleman SH, et al
. Safety and trough concentrations of nevirapine prophylaxis given daily, twice weekly, or weekly in breast-feeding infants from birth to 6 months. J Acquir Immune Defic Syndr 2003; 34:482–490.
21. Shapiro RL, Holland DT, Capparelli E, Lockman S, Thior I, Wester C, et al
. Antiretroviral concentrations in breast-feeding infants of women in Botswana receiving antiretroviral treatment. J Infect Dis 2005; 192:720–727.