Breast-Feeding and HIV Transmission: The Jury Is Still Out

We present a review of the literature to shed some light on the dilemma facing policy makers, health workers, and HIV-infected women: To breast-feed, or not to breast-feed, that is the question. We searched PubMed for English language publications using the keywords HIV, infection, transmission, and breast-feeding. In addition, we obtained information from abstracts from AIDS and Nutrition conferences since 1997. This review will only consider transmission of HIV-1 (subsequently referred to as HIV) because mother-to-child transmission (MTCT) of HIV-2, through any route, is rare (1). In medicine when any behavior or practice is known to be a risk factor for good health, the obvious intervention is to try to eliminate the risk. In this case, however, removing the risk of breast-feeding consequently introduces another risk factor, viz., commercial infant formula or other replacement feeding with their known risks for child health. The risks for each are evident in crude mortality rates; on one hand, breast-feeding is thought to be responsible for approximately 300,000 HIV infections per year (2), whereas the United Nations International Children's Emergency Fund (UNICEF) estimates that not breast-feeding is responsible for 1.5 million child deaths per year (3). This review attempts to examine the data that are currently available, to present the dilemmas that hinder implementation of programs to reduce MTCT of HIV, and to suggest research that is still needed to allow us to make definitive judgments. BACKGROUND INFORMATION ON BREAST-FEEDING TRANSMISSION OF HIV The report in 1985 of a mother who was infected with HIV via a blood transfusion soon after giving birth, and whose child was infected presumably by breast-feeding was the first to draw attention to the possibility that HIV could be transmitted through breast milk (4). Subsequently, several other reports confirmed this phenomenon in which women who were newly infected transmitted the virus to their infants during the breast-feeding period (5–10). It was generally believed that the risk of HIV transmission from newly infected women was higher than from those breast-feeding women with established HIV infection (11). QUANTIFYING THE RISK OF HIV TRANSMISSION THROUGH BREAST-FEEDING According to Duration of Breast-Feeding Until the advent of the polymerase chain reaction (PCR) test, which amplifies viral genetic material and establishes the presence of HIV infection, the diagnosis of HIV in young infants relied on the enzyme-linked immunosorbent assay (ELISA) which measured antibodies in children at 15 months of age. Therefore, in the era of ELISA testing, investigators were unable to distinguish between in utero, intrapartum, and breast-feeding transmission. Any estimate of breast-feeding transmission relied on comparing overall transmission rates between breast-feeding and non–breast-feeding populations (12), which possibly were not comparable and might have had inherent sampling bias. The meta-analysis by Dunn et al. (13) estimated that breast-feeding by women with established infection increased the absolute rate of transmission by 14% (95% CI, 7–22). These early studies were compromised by small sample sizes and the uneven distribution of breast- and formula-feeding practices. Recently, the results of three studies were published in which both groups had at least 100 infants either formula-fed or breast-fed; two were observational studies (14,15), and only one collected feeding information prospectively (14). The third study was a randomized controlled trial (RCT) (16). The first of these studies was from Brazil (14) and did not use PCR testing; therefore, infection status was only confirmed at age 18 months. Of 168 breast-fed children, the risk of HIV transmission was 21% compared with 13% (P = 0.01) of the 264 formula fed children. The second study was from South Africa (15) and was an observational study in which feeding data were collected 1 week, 6 weeks, and 3 months after delivery and thereafter every 3 months until the infants were 15 months of age. At each visit, blood samples were collected and were tested for HIV by PCR. Samples were tested using ELISA at 15 months or by PCR on earlier samples if the ELISA at 15 months was positive or if the child died or was lost to follow-up before 15 months. Included in the study were 394 breast-fed infants and 157 formula-fed infants. At age 15 months, 31.6% (95% CI, 25.3–38.1) of breast-fed infants were HIV positive compared with 19.4% (95% CI, 13.6–24.0) in the formula-fed group, representing an excess absolute risk of HIV transmission through breast-feeding of 12.2% after 15 months (P = 0.007). The most serious limitation of this study was that the period of recall of infant feeding was 6 weeks during the first 3 months, and 3 months thereafter. The Kenyan RCT (16) using an intention-to-treat analysis showed that at age 24 months, 36.7% of breast-fed infants were infected compared with 20.5% of the formula-fed infants, representing an excess absolute risk of HIV transmission of 16.2% after 24 months. The most serious limitation of this study was the lack of compliance with the assigned feeding modalities; 30% of the formula-feeding group were known to have breast-fed. The HIV transmission results also suggest that randomization had not produced two similar groups. The rates of HIV infection were 3.1% versus 7.0% at birth and 9.7% versus 19.9% at age 6 weeks in the formula-feeding and breast-feeding arms, respectively (16). Both studies support the earlier 1992 meta-analysis (13) of an approximate 14% breast-feeding transmission for a 2-year breast-feeding period. We now consider the timing of transmission to estimate the risk associated with the duration of breast-feeding. It is assumed that all in utero and intrapartum transmission is detectable by PCR at age 6 weeks and that transmission detected after age 6 weeks is attributable to breast-feeding. Transmission rates in the first 6 weeks have been extrapolated from the rate of new infections between 6 weeks and 6 months in breast-feeding populations. An RCT of breast-feeding versus replacement feeding is required to accurately determine this rate, but as mentioned earlier, the randomization difficulties in the only RCT to have done this limit these data. Table 1 shows data from six recent large studies of breast-feeding women whose infants' HIV infection status was monitored frequently by PCR. The mothers in these studies received no antiretroviral (ARV) drugs, a single ARV for a short period to prevent MTCT of HIV or placebo. The accumulation of risk from 6 weeks to 6 months ranged from 1.3% to 8.1%. In the second 6 months of breast-feeding, the cumulative risk ranged from 0.8% to 4.9% transmission (15–20).TABLE 1: Accumulation of HIV infection in the first 6, 12, and 24 months in breastfed infants * †According to Pattern of Breast-Feeding Most studies attempting to estimate the transmission of HIV attributable to breast-feeding have made no attempt to define the pattern of breast-feeding. Just as it is important to specify the duration of breast-feeding when assigning risk, so too is it important to specify the type of breast-feeding that is practiced. In most studies, insufficient information has been collected to classify infants as exclusively breast-fed (EBF) versus mixed breast-fed. Many researchers have used arbitrary definitions of EBF and not the accepted World Health Organization (WHO) definition (21), which defines EBF as breast milk only with no other solids or liquids. Researchers commonly have allowed consumption of waters, teas, and other nonmilk liquids or solids in their definition of EBF (22). Two studies have used the correct WHO definition of EBF. The first study found that infants who were EBF had a lower risk of HIV transmission; however, the association was not significant (14). A limitation of this study was that feeding data were collected retrospectively, at least 18 months after birth. The second study (15) included HIV-infected pregnant women participating in an RCT of vitamin A (23). Women received counseling antenatally according to the UNAIDS guidelines and chose to either breast-feed or formula feed. Women who chose to breast-feed were encouraged to practice EBF as a possible way of reducing risk of HIV transmission. Of the 551 mother–baby pairs included in the study, 157 (28.5%) infants were never breast-fed, and 394 (71.5%) were breast-fed for some period. Mothers who breast-fed their infants were classified into those who EBF to age 3 months or more (n = 118) and those who mixed breast-fed (n = 276). The women in the three groups were similar in terms of most known risk factors for MTCT of HIV, and transmission rates at birth were similar. The cumulative probability of HIV infection remained similar among those never breast-fed and EBF infants up to 6 months, 19.4% (95% CI, 13.6–26.0) and 19.4% (95% CI, 12.5–27.4), respectively, but it was higher in the mixed breast-fed group, 26.1% (95% CI, 20.5–31.9%). The results from this study suggest that the vertical transmission of HIV through breast milk depends on the pattern of breast-feeding and not simply on breast-feeding per se. A limitation of the study was measuring adherence to the reported feeding practice. That there were only 2 new infections in the formula-feeding group after 6 weeks (1 in a child tested for the first time at age 4 months) would seem to corroborate the reported feeding practices, at least in the formula-feeding group. Measuring adherence will always be difficult, and future studies should incorporate frequent monitoring, at least weekly, to improve the validity of maternal recall (24). Preliminary results from a study in Kisumu, Kenya (25) also indicated that the introduction of other solids and liquids during the early months of life was a risk factor for breast-feeding transmission. This study included a sample of 220 HIV-infected mothers whose infants' first PCR tests were HIV negative and showed that the incidence rate of HIV infections was greater (P < 0.07) for infants who started mixed feeding before aged 30 days compared with infants who started mixed feedings after aged 30 days. Similarly, incidence of HIV infections was greater (P < 0.05) for infants who started mixed feedings before aged 120 days compared with those who started after aged 120 days. RISK FACTORS FOR BREAST-FEEDING TRANSMISSION Breast Milk Viral Load The majority of children born to HIV-infected women and who are breast-fed do not become infected with HIV (26). There has been considerable interest in factors that may account for this phenomenon. HIV is found in breast milk as cell-associated (DNA) and cell-free (RNA) virus (27). There is evidence that the virus can infect mammary epithelial cells (MEC) (28) and that hormones normally associated with lactation (triiodothyronine, β-estradiol, and prolactin) may stimulate virus replication. Although in vitro HIV replicates in MEC to a lesser extent than in lymphoid cells, MEC-derived HIV may have for the infection of epithelial cells and be for infant or epithelial cells HIV may be transmitted by which milk possibly in with breast milk as In a study of HIV-infected women in the risk of vertical transmission in women with HIV detectable by PCR in their breast milk at 15 days if breast milk antibodies were not present (27). antibodies are the of in breast milk and viral infection by and was detected in breast milk samples from women in this may be for by increased of in the mammary which is in the of HIV-infected In a study of et al. to and in a single sample of breast milk collected approximately 6 weeks was not detected in any breast milk after infants who had evidence of in utero or intrapartum infection were there was no significant in transmission rates from mothers with or detectable HIV PCR in their breast milk at 6 weeks, the of was In a study of et al. found that the of breast milk samples with detectable HIV between at to 4 approximately at 3 and 6 months, and between and at and months. of HIV in breast milk of women in this study that some women shed virus whereas never have detectable viral in breast et al. found that viral during the first 3 months not only time but also between of an at any during the of milk from both and early breast milk a higher of breast milk than milk which may the risk of transmission. et al. breast milk samples from HIV-infected women and found that PCR could HIV in only of all the breast milk the of HIV was higher in milk than in or in milk produced than months (P = and Breast Health is an during which the between mammary epithelial cells There is an in breast milk and cells during This is in one breast may be a of increased breast milk viral In the of the presence of an by an increased breast milk greater than 1 is as A study by et al. using breast milk samples collected 6 weeks found that a breast milk greater than which the of was found in of mothers of HIV-infected infants but in only of mothers of infants. HIV status of infants, however, was estimated when the infant was aged only 6 weeks, which it to any with breast-feeding transmission from these data. Breast milk viral was for a of breast milk viral was in mothers of infected infants versus in mothers of infants, and in women with increased breast milk versus in those with Mothers who had increased breast milk also had lower and higher viral and breast milk In a study of HIV-infected women in South et al. collected breast milk samples from each breast 6, and weeks and found that was and breast milk HIV viral was but at all viral was associated with and maternal between 21% and of the in milk viral during the first blood during and increased were associated with increased milk viral at all There was a for the 6 infants who were infected with the infants who remained to have been to breast milk of higher viral at each time for of HIV infection from breast milk is by a of the of into the infant It has been that MTCT through between HIV and infant breast-feeding, the infant and into with It is that HIV may through in the or through infection of lymphoid cell-free or cell-associated virus is to infect in the by through in the that in epithelial may the risk of transmission. Similarly, if HIV through lymphoid factors that the of lymphoid cells may adherence and transmission of the The introduction of formula may with the of the of infants the first of life In a study to the of infant feeding on HIV infection, was measured in infants of HIV-infected South women at 6, and weeks At each visit, information infant and was and in a of infants, was measured as an of who were HIV infected by age weeks had higher (P < 0.01) at 6 and weeks and higher < at all when compared to infants. At age 1 week, infants no breast milk had higher (P < 0.05) than infants breast milk exclusively or with other This was not present at time feeding practice did not of infants fed breast milk other was never increased to that of EBF infants. also had no on infant HIV infection was to in and possibly before but the of feeding on infant did not a possible of EBF on MTCT of The of to and the of of of lymphoid make it to determine the and of transmission. from and of for other may to this of OF BREAST-FEEDING women not to breast-feed in to prevent transmission of HIV is not because this intervention The of any to prevent MTCT be to overall child children of to this is the risk of and of breast-feeding versus not breast-feeding and will the and of formula milk or other replacement to HIV-infected women have on the feeding of Breast milk the for the first 6 months of life and a of up to age 2 and is the most and of infant feeding and is important in the and may the the from the it the mother and child and The of breast-feeding to the infant is risk of infection, and The of breast-feeding are and have been by a recent meta-analysis In this which included studies from and the breast-feeding was to child mortality during the early months of life and for infants aged and months, The for the from and infections during the first 6 months of life were and The of this study that will be difficult, if not to breast milk to children from The of breast-feeding in the of mortality from infections is to be as important in the risks of feeding can be in however, breast-feeding may as and infant important have from the Kenyan RCT (16). In the formula-feeding group which the mothers were and also had to and frequent support of health the risk of HIV infection may have been by the risk of and mortality from other during the first 6 months. the first 3 months of infants in the formula had an increased risk of and infections Similarly, the cumulative mortality at 6 weeks and 3 months was higher in the formula-fed group compared with the breast-fed group and was most obvious in the infants. mortality of infants in the first 6 months of life was in formula-fed infants compared with breast-fed of the of the risks associated with breast-feeding and formula feeding in of HIV several groups have to the have to the risks of HIV transmission via breast-feeding the increased risks of mortality associated with not breast-feeding. reported that when the infant mortality rate is more than per the risk of infant deaths from from replacement feeding would the risk of of HIV by breast-feeding This did not a possible of EBF for HIV transmission. a more to estimate the of feeding in the of breast-feeding on all child mortality until age This that the of ARV and replacement of breast milk increased with increased HIV of counseling and testing, compliance with and rates for HIV-infected children. the of formula viz., the of the non–breast-feeding had on mortality rates in et al. estimated that breast-feeding infants up to age 6 months 6 months, as mortality rates and the risk of from feeding the infant feeding was to use an to breast-feeding. and UNAIDS the replacement feeding is and of all breast-feeding by HIV-infected mothers is EBF is during the first months of on health to children of status To HIV transmission risk, breast-feeding should be as soon as into account the and the risks of replacement feeding infections other than HIV and HIV-infected mothers not to breast-feed from birth or breast-feeding should be with and support for at least the first 2 of the life to replacement should to improve that will make replacement feeding for HIV-infected mothers and HIV-infected mothers who breast-feed should be with and support when breast-feeding to and and to breast health. These an and on the and with delivery of MTCT At a large and on the of and one of the in South Africa a MTCT was of the health could estimate the risk of HIV through breast-feeding Many reported to HIV-infected 1 of the for the MTCT was to define breast-feeding, and 1 had never of the on but important breast-feeding difficulties was HIV-infected mothers were that should not breast-feed, and reported the risks of not breast-feeding. of breast milk or milk with to be and women to make an infant feeding is not simply a of and the risks and feeding Health to to is most and for is required to the factors that breast-feeding transmission or from replacement and to to reduce these It also a of the of the the to on risk in a way that is by women who do not have of these and the to support women in a that their children, and the of their a counseling in MTCT programs a to one feeding than for each This may mothers to a feeding practice that be or which may child at risk of mixed feeding and at an increased risk of of HIV At a it may to a of support for to breast-feeding, in and for some women to breast-feeding if do not their HIV BREAST-FEEDING An early report of an association between breast-feeding and maternal mortality has and The investigators reported that breast-feeding women in the study were more to than mothers who formula fed their infants The data should be with because of which are in the by Although the of the women in the two randomization groups were similar at viral was higher in those randomized to breast-feeding, and information on was The rate of in utero transmission was increased in the breast-feeding group, which could suggest more in these to follow-up review earlier in the formula-feeding group than in the breast-feeding therefore, deaths in the formula-feeding group may be The results of this study are to the results of the study which to an in either mortality or in the breast-feeding group. the of these two studies, the WHO an and that there was insufficient evidence to suggest that breast-feeding by HIV-infected women their mortality of the of breast-feeding on maternal mortality and is BREAST-FEEDING TRANSMISSION Until data are to mothers of these can be done to breast-feeding transmission and child of health be to mothers make that are to women who or to breast-feed, support should be to good breast-feeding to breast HIV viral and to the of breast using a that is and at can be used to all cell-free HIV This may also be as a to EBF when the mother is or from those who replacement support should be to and of commercial infant formula and other replacement to risks of and Women can be to limit breast-feeding to 6 months when it is and to other replacement and are too to support in the be in of breast-feeding because the child will be Health to improve breast-feeding and feeding counseling to be to be of mothers with HIV infection and the to infant feeding that may There are still that we to before we can make any which and HIV The of these is the risk of HIV transmission if HIV-infected women breast-feed exclusively for 6 months and to prevent breast is the of ARV for the or both during is the of lactation and on and therefore, on risk of breast-feeding is the of breast-feeding replacement to and women is the of breast-feeding replacement on infant and is the of early of breast-feeding on HIV transmission and and is the of and for infants during is the of cell-associated and cell-free virus for transmission and in for early versus more is the of breast-feeding on maternal health and is the of maternal health and on risk for HIV is the of breast milk in the of infected children fed breast milk to AIDS more is the for of breast-feeding to the least to mother and OF It is that any groups will attempt an RCT of feeding because the Kenyan trial showed that it is to a behavior breast-feeding, which is so of a it is now generally that it would be to a large studies in South and are currently in to examine more the of EBF on the risk of breast-feeding transmission. The study in addition, comparing the between of breast-feeding at 4 months compared with breast milk 2 The trial in is a trial of versus with to infants. These mothers will according to the A trial of is also for will HIV during the period are breast-fed. A has up a and HIV International Transmission The study will data from randomized of at reducing MTCT of HIV in populations breast-feeding is This will an into the of HIV infection during the first months of of the timing of breast-feeding transmission and of of breast-feeding transmission. in South and Kenya are the of or as to infants to prevent breast-feeding transmission. some of the of MTCT research is to the HIV transmission risk according to infant feeding practice and to and mortality risks according to infant feeding It is that breast-feeding greater risk of MTCT of HIV than replacement this risk is by duration and pattern of breast-feeding, maternal breast health, and maternal in replacement feeding an increased risk for infant and suggest that 6 months of breast-feeding may be than replacement feeding for HIV-infected We the results of research on the of EBF in reducing breast-feeding transmission. In the we to a of the risks of breast-feeding and replacement feeding and according to to make an and support the that child the of and HIV This should not breast-feeding in women who are HIV or of the is that commercial infant most HIV-infected women will breast-feeding because of and Therefore, the is on health and policy to to make breast-feeding or replacement feeding