Abstract
Primary failure of induction chemotherapy or disease recurrence after short remission duration (“therapeutic resistance”) remains the principal problem in adult acute myeloid leukemia (AML). Although cytogenetic and molecular abnormalities have proven useful in the identification of subsets of patients with distinct disease risks, it is unclear to what degree therapeutic resistance can be predicted for individual patients.
We used information on patients with newly diagnosed AML other than acute promyelocytic leukemia receiving curative-intent treatment on trials conducted by the U.K. Medical Research Council/National Cancer Research Institute (MRC/NCRI; 1988-2010; n=2,615), the Dutch-Belgian Cooperative Trial Group for Hematology/Oncology and the Swiss Group for Clinical Cancer Research (HOVON/SAKK; 1987-2008; n=1,098), the U.S. cooperative group SWOG (1987-2009; n=428), and MD Anderson Cancer Center (2000-2011; n=409). Achievement of a complete remission (CR) with the initial 1-2 courses of induction chemotherapy was defined as therapeutic success. Patients who failed to achieve CR were defined as primary refractory for the purpose of this analysis; patients who experienced treatment-related mortality (i.e., death within 28 days of treatment initiation) were excluded from this analysis. We used logistic regression analyses to assess the relationship between individual covariates and various measures of therapeutic resistance. The following pre-treatment covariates were used in the regression modeling: age at diagnosis, gender, white blood cell (WBC) count, platelet count, bone marrow blast percentage, disease type (primary vs. secondary), cytogenetic risk, FLT3/NPM1 status, and treatment site. We then used the area under the receiver operator characteristic curve (AUC) to quantify a model’s ability to predict therapeutic resistance; in this approach, an AUC of 1 indicates perfect prediction while an AUC of 0.5 indicates no prediction; AUC values of 0.6-0.7, 0.7-0.8, and 0.8-0.9 are commonly considered as poor, fair, and good, respectively.
A total of 4,550 patients (median age: 52 years [range: 15-90 years]) were included in this study. A CR to the initial 1-2 courses of induction chemotherapy was achieved in 3,597 (79.1%) of patients, whereas 953 (20.9%) were primary refractory; 1,304/4,497 patients (29.0%) with sufficient follow-up time were either primary refractory or had a relapse-free survival (RFS) of 3 months or less after CR achievement, 1,774/4,445 patients (39.9%) with sufficient follow-up time were either primary refractory or had a RFS of 6 months or less after CR achievement, and 2,523/4,386 patients (57.5%) with sufficient follow-up time were primary refractory or had a RFS of 12 months or less after CR achievement. Increasing age (p<0.001) and WBC (p<0.001), secondary disease (p<0.001), FLT3/NPM1 status (p<0.001), and cytogenetic risk (favorable or intermediate vs. adverse, p<0.001) were independently associated with being primary refractory to induction chemotherapy in a combined analysis of all patients. In the total patient cohort, a bootstrap-corrected multivariate model predicting primary refractoriness yielded an AUC of 0.79; removal of FLT3 and NPM1 from the model minimally, but statistically significantly, decreased the AUC (0.77). Between individual treatment sites, these AUCs varied from 0.82/0.81 to 0.69/0.67. Prediction of therapeutic resistance, as defined as primary refractoriness or relapse after short remission duration, was more difficult. Specifically, when analyzing the entire study cohort, the AUCs for models predicting primary refractory disease or relapse within 3 months were 0.76/0.74 (with/without inclusion of FLT3/NPM1 data) and further decreased to 0.76/0.73 and 0.75/0.71 for models predicting primary refractory disease or relapse within 6 or 12 months, respectively.
Our ability to predict therapeutic resistance based on routinely available clinical covariates, even with inclusion of commonly used molecular data on FLT3 and NPM1, is relatively limited. This finding would support the continued use of randomization to assign patients between standard and investigational therapies, and argues for the integration of early treatment response measures (e.g. minimal residual disease) to optimize prediction of therapeutic resistance.
No relevant conflicts of interest to declare.
Author notes
Asterisk with author names denotes non-ASH members.
