(introduction...)

Breast-milk transmission of HIV has been well documented. The first reports indicating the possibility of HIV-1 transmission through breast milk were in breastfed infants of women who were infected postnatally through blood transfusion or through heterosexual exposure (Palasanthiran et al., 1993; Van de Perre et al., 1991; Stiehm and Vink, 1991; Hira et al., 1990; Colebunders et al., 1988; Lepage et al., 1987; Ziegler et al., 1985;). There were also reports of infants, with no other known exposure to HIV, who were infected through wet-nursing and through pooled breast milk (Nduati et al., 1994; Colebunders et al., 1988;).

Generally, higher rates of mother-to-child transmission of HIV are observed where most infants are breastfed rather than where fewer infants are breastfed. However, other reasons for variations in transmission rates, such as maternal nutritional status, stage of HIV disease and possible differences in transmission of HIV subtypes cannot be excluded. Additional evidence is provided by results from prospective studies which indicate that among infants born to HIV-infected mothers, those who are breastfed are more likely to be infected than those who are formula-fed, even allowing for other factors known to be associated with mother-to-child transmission of HIV (European Collaborative Study, 1992; Ryder, 1991; Blanche et al., 1989; Tovo et al., 1988; Tess et al., 1998a).

Mechanisms of breast-milk transmission

Although HIV has been detected in breast milk, (Nduati, 1995; Ruff, 1994; Van de Perre et al., 1993) mechanisms of breast-milk transmission are not yet fully understood. The respective roles of cell-free and cell-associated virus in breast-milk transmission are not known, nor is the association between plasma and milk virus levels understood. The portal of entry for the virus via the infant mucosa also merits further investigation.

Animal models have been used to explore potential mechanisms of transmission. It is possible to infect neonatal rhesus monkeys with simian immunodeficiency virus (Baba et al., 1994) and kittens with feline immunodeficiency virus (Sellon et al., 1994) by applying cell-free virus on the mucosa. This suggests that cell-free HIV in breast milk could infect cells of the intestinal mucosa. M-cells, which are specialized epithelial cells found in the Peyer's patches of the intestinal mucosa, may be a mechanism allowing infectious agents such as HIV to cross the intact mucosa. M-cells engulf and transport the pathogen and present it to macrophages that indent the serosal surface of the M-cell (Featherstone, 1997). Results from in vitro studies on rabbit M cells suggest that HIV-1 particles could use M cells to cross the intestinal barrier (Amerongen et al., 1991). A recent in vitro study indicated that HIV-infected cells themselves may also play an important role by stimulating ordinary enterocytes to engulf HIV particles presented by HIV-infected cells in the intestinal lumen (Bomsel, 1997). Moreover, HIV RNA has been detected in the oropharyngeal and gastric aspirates of a substantial proportion of infants born to HIV-infected mothers (Nielsen et al., 1996, Ait-Khaled et al., 1998).

Quantifying the risk of breast-milk transmission

Early studies investigating the frequency of breast-milk transmission and associated factors were limited by small numbers as well as by the predominance of one method of infant feeding in any one cohort (European Collaborative Study, 1992; Ryder, 1991; Blanche et al., 1989; Tovo et al., 1988).

In 1992, a meta-analysis was carried out using data from four studies reporting on 42 recently infected women and six studies reporting on 1772 women with established infection. The majority of the women had breastfed for 2-4 weeks, and 106 women had breastfed for longer than six months. The estimated additional risk of transmission from breast milk, above that occurring during pregnancy and delivery, among women with established HIV infection, was approximately 15% (95% Confidence Interval 7-22%) (Dunn et al., 1992). However, 15% may be an under-estimation among women who breastfeed for longer periods of time.

The risk of transmission through breast milk among women with recent infection (HIV infection acquired in the postpartum period) was nearly twice as high (29% (95% CI 16-42%)).

Insufficient information is available to estimate the exact association between duration of breastfeeding and the risk of transmission. However, there is strong evidence for a gradual and continued increase in transmission risk as long as the child is breastfed (Taha et al., 1998, Leroy et al., 1998).

Timing of HIV transmission during breastfeeding

Transmission of HIV through breast milk can take place at any point during lactation. The persistence of maternal antibodies and the presence of a "window period" during which infection is undetectable using currently available technology, make it impossible to determine whether an infant has been infected during delivery (intrapartum) or through breastfeeding in the period following birth. Therefore, when seropositive women breastfeed their infants, it is not possible to differentiate between HIV transmission attributable to delivery and that resulting from breastfeeding from birth. (Newell, 1998; Bobat et al., 1997; Mandelbrot et al., 1996; Bertolli et al., 1996; Simonon et al., 1994; Datta et al., 1994).

Later postnatal transmission through breastfeeding can be determined using currently available diagnostic tools. Studies of infants found to be negative by PCR testing at 2-6 months of age, but who subsequently showed evidence of infection, have provided estimates of the risk of late postnatal transmission (after 3-6 months of age) ranging from 4-12% (Ekpini et al., 1997; Karlsson et al., 1997; Bertolli et al., 1996; Simonon et al., 1994).

Table 2. Studies of the risk of late postnatal transmissions

Colostrum and mature milk

Cell-free and cell-associated HIV-1 have been detected in both colostrum and mature breast milk of women with established HIV infection. In a study in Haiti, HIV DNA (cell-associated virus) was detected in 70% of 47 colostrum samples and about 50% of breast-milk samples obtained at 6 (n=30) and 12 (n=15) months postpartum (Ruff et al., 1994). HIV DNA was detected in 47% of 129 samples of breast milk collected 15 days after delivery, and in 20% of 96 samples collected six months after delivery (Van de Perre et al., 1993). Both studies suggest a higher level of cell associated HIV in early milk compared to later, which would reflect the relatively high level of cells in colostrum compared to mature milk.

Somewhat differently, in a study in Kenya (Nduati et al., 1995) a higher proportion of samples of milk collected between seven days and six months had HIV DNA (65% of 108 breast milk samples) than did colostrum (51% of 77 samples) (p=0. 05). Among positive samples, the proportion of infected cells ranged from less than 1 in 10 000 cells to 1 in 3. High concentrations of HIV-infected cells were more common during the period 8-90 days after delivery than in samples taken either earlier or later. A second study quantified HIV-1 RNA (measuring cell-free virus) from breast-milk supernatants collected from the same group of women at the same times (Lewis et al., 1998). The range of viral concentration in the breast-milk supernatants was very wide and, although the difference was not statistically significant, viral load tended to be higher in milk collected more than 8 days after delivery than in milk samples taken earlier (p=0.10).

The potential effect of various factors makes it difficult to draw any conclusions about the relative risk of transmission through colostrum and mature breast milk. First, colostrum and mature breast milk contain different types of cells and different levels of immune modulating components (e.g. vitamin A, immunoglobulins and lactoferrin). Second, the total volume of colostrum ingested by the infant is much smaller than that of mature breast milk. Third, the infant's immune system is less well developed during the first few days of lactation than in later lactation, while younger infants have an increased blood concentration of maternal antibodies. In the study by Tess et al. (1998b), vertical transmission was not associated with a history of colostrum intake in 148 breastfed children.