Background:  Understanding the origin of breast milk virus can have important implications for transmission prevention. The aim of our study was to compare the patterns of resistance mutations of the viral populations in plasma, breast milk, and breast milk cells of HIV-infected women receiving HAART in order to obtain information on the source of the breast milk virus.

Methods:  We studied 40 pregnant women in Mozambique who had received zidovudine, lamivudine, and nevirapine from 28 weeks of gestation until 1 month after delivery for the prevention of mother-to-child transmission. Paired blood and breast milk samples were collected in the first week after delivery. HIV RNA in plasma and whole breast milk and proviral DNA in breast milk cells were quantified and used for the resistance analysis. Genotype was obtained using the Tru Gene assay. Genetic distances were determined by use of Syn-SCAN.

Results:  A total of 16 women had detectable HIV RNA in plasma and breast milk and were included in the analysis. The genotypic resistance patterns differed between the plasma and the cell-free breast milk viruses in 6 of 16 patients (37.5%) and between the cell-free and the cell-associated breast milk viruses in 8 of 16 cases (50.0%). In 2 cases cell-free and cell-associated viruses in breast milk had the same mutational pattern, which differed from that of the plasma virus. Major resistance mutations (K103N+M184V, K103N, K101E+V108I) were present in the plasma viruses of 3 women, 2 of whom had the same pattern in breast milk. In 2 patients, viral strains present in the breast milk compartment (either cell-free or cell-associated) harbored major resistance mutations (M184I and V106A) not present in the plasma virus. No correlation was found between the emergence of major mutations and plasma or breast milk drug concentrations. The median (IQR) of nucleic acid genetic distance within each woman between plasma and cell-free breast milk viruses and between cell-free and cell-associated breast milk viruses were 2.1 (0.9 to 3.9) and 2.1 (1.5 to 5.0) for reverse transcriptase and 1.4 (0.7 to 2.9) and 1.9 (0.4 to 5.2)  for protease, respectively.

Conclusions:  Although a clear correlation between the patterns in the different compartments could not be established, our data suggest that the breast milk virus originates both from passive diffusion from plasma and from local viral production. Post-natal transmission may occur with viral variants that cannot be predicted by those present in plasma.