Biological characteristics and prognostic value of in vitro three-drug resistance to prednisolone, L-asparaginase, and vincristine in childhood acute lymphoblastic leukemia.

The purpose of this study was to investigate the biological characteristics and prognostic value of in vitro three-drug resistance to prednisolone, L-asparaginase, and vincristine in childhood acute lymphoblastic leukemia (ALL). We carried out in vitro tests with a 4-day culture and a methyl-thiazol-tetrazolium assay on bone marrow samples from 209 children newly diagnosed with ALL. After testing the resistance of leukemic cells to 14 drugs, we classified the patients into two groups according to their sensitivity to three drugs (prednisolone, L-asparaginase, and vincristine) used in remission induction therapy. The three-drug resistant group (RR: sensitive to no drugs or to one drug) correlated with both short-term and long-term treatment failure. Three-year event-free survival (95% confidence interval) for the sensitive group (SS: sensitive to two or three drugs) was 0.813 (0.773-0.853) and that of the RR group was 0.616 (0.569-0.669) (P = 0.0001). Univariate analysis showed that Philadelphia-chromosome (Ph1) positivity and immunophenotype of mixed lineage were also prognostic factors in the 209 patients. The prognosis of the SS/RR drug resistance profile within 14 Ph1 patients was marginally significant (P = 0.062). Multivariate Cox regression analysis showed that Ph1 was an overwhelmingly adverse factor in event-free survival, with a relative hazard of 5.37 (2.57-11.21, P < 0.0001), followed by RR, with a relative hazard of 2.98 (1.69-5.25, P = 0.0001). Furthermore, we clarified the characteristics of the RR group by examination of the pattern of drug resistance to other drugs in comparison with the SS group. The leukemic cells of RR patients were more resistant than those of SS patients (P < 0.0001) to all the drugs tested, with resistance ratios of 1.6 to 13.1 (mean 3.4). In conclusion, in vitro three-drug resistance at the initial stage is an important independent predictor of treatment failure for both induction response and long-term outcome in childhood ALL.