Increasing chimerism after allogeneic stem cell transplantation is associated with longer survival time

Donor chimerism after allogeneic stem cell transplantation (allo-SCT) is commonly used to predict overall survival (OS) and disease-free survival (DFS). Because chimerism is observed at 1 or more times after allo-SCT and not at baseline, if chimerism is in fact associated with OS or DFS, then the occurrence of either disease progression or death informatively censors (terminates) the observed chimerism process. This violates the assumptions underlying standard statistical regression methods for survival analysis, which may lead to biased conclusions. To assess the association between the longitudinal post-allo-SCT donor chimerism process and OS or DFS, we analyzed data from 195 patients with acute myelogenous leukemia (n = 157) or myelodysplastic syndrome (n = 38) who achieved complete remission after allo-SCT following a reduced-toxicity conditioning regimen of fludarabine/intravenous busulfan. Median follow-up was 31 months (range, 1.1 to 105 months). Fitted joint longitudinal-survival time models showed that a binary indicator of complete (100%) donor chimerism and increasing percent of donor T cells were significantly associated with longer OS, whereas decreasing percent of donor T cells was highly significantly associated with shorter OS. Our analyses illustrate the usefulness of modeling repeated post-allo-SCT chimerism measurements as individual longitudinal processes jointly with OS and DFS to estimate their relationships.