OBJECTIVE

We examined the association between mother-to-child human T cell lymphotropic virus type I (HTLV-I) transmission and human leukocyte antigen (HLA) class I types.

METHODS

In 1989, children born to HTLV-I-infected mothers in Jamaica were enrolled and prospectively evaluated for HTLV-I infection. HLA class I types in mothers and children were determined by DNA-based polymerase chain reaction methods. Associations between HLA class I types and transmission of HTLV-I were analyzed using proportional-hazards regression models adjusted for the duration of breast-feeding. Transmission risk in children still breast-feeding at 12 months was determined using actuarial methods.

RESULTS

Of 162 children, 28 (17%) became infected. After Bonferroni's adjustment for multiple comparisons, the transmission risk was not influenced by any specific HLA class type or the A2 supertype. However, compared with children who shared 3 HLA class I types with their mothers (the minimum number possible), the transmission risk increased 1.8-fold with 4 shared types and 3.0-fold with 5 or 6 shared types (Ptrend = .039; 1.75-fold increase for each additional concordant HLA type). This association was independent of maternal HTLV-I proviral level, antibody titer, and household income.

CONCLUSIONS

We found a significant dose-response relationship between HTLV-I transmission via breast-feeding and mother-child HLA class I type concordance. Immunological interactions between a child's cells and maternal cells may influence the risk of HTLV-I infection by breast-feeding, perhaps because antigens on maternal cells are seen by the child as being "self."