Cytogenetics and P-glycoprotein (PGP) are independent predictors of treatment outcome in acute myeloid leukemia (AML)

Clinical and biological features have recognized prognostic significance in acute myeloid leukemia (AML). To evaluate the interaction of these variables and weighted effect on treatment outcome, prognostic variables from 96 previously untreated patients were analyzed for association with expression of the MDR1 gene product P-glycoprotein (Pgp), and effect on response to induction chemotherapy, progression-free survival and overall survival. Multivariate relationships were analyzed using six prognostic variables, including age, cytogenetic pattern, gender, CD34⁺ surface phenotype, AML type (de novo versus secondary) and Pgp. Univariate comparisons indicate that Pgp (P = 0.0001), cytogenetic pattern (P = 0.0004) and a CD34⁺ phenotype (P = 0.0005) are predictive of primary treatment failure, whereas Pgp (P = 0.0001) had the greatest predictive value in multivariate analysis. Only cytogenetic pattern retained prognostic significance (P = 0.0143) for response to induction therapy after adjustment for Pgp. Although all variables except gender were associated with Pgp, specimens harboring the favorable karyotypic abnormalities t(15;17), t(8;21) and inv(16) exclusively lacked Pgp expression. In a multivariate model, both Pgp and cytogenetic pattern predicted response and overall survival, whereas secondary AML and cytogenetic pattern influenced remission duration. These findings indicate that cytogenetic pattern has prognostic relevance that is independent of Pgp, and implies the presence of undefined biological mechanisms affecting chemotherapy resistance.