Effect of measurable (‘minimal’) residual disease (MRD) information on prediction of relapse and survival in adult acute myeloid leukemia

Background: With the gradual improvement in treatment-related mortality over the last several decades, therapeutic resistance remains the primary challenge in adult acute myeloid leukemia (AML). Accurate prediction of treatment outcomes, including relapse, would facilitate decision-making, e.g. regarding assignment of patients to risk-stratified investigational or standard therapies. Yet, our previous studies have indicated that our ability to predict therapeutic resistance based on detailed pre-treatment information, including cytogenetic and molecular profiling data, remains relatively limited. Here, we investigated to what degree the prediction of relapse and survival can be improved for individual patients by inclusion of post-treatment data, in particular minimal residual disease (MRD) after the initiation cycle of induction chemotherapy. Patients and Methods: We used data on adults aged 18-60 years with newly diagnosed de novo AML who achieved a morphologic complete remission (CR) on a recent phase 3 intergroup trial (S0106) that investigated the value of gemtuzumab ozogamicin (GO) when added to standard induction chemotherapy (NCT00085709). MRD was prospectively assessed by multiparameter flow cytometry in bone marrow specimens obtained at CR following ANC and platelet recovery; any level of MRD was considered MRDpos. The flow cytometric method used was an early generation, 3 tube, 10-color assay performed on an LSRII and relied on identification of progenitor populations that differ from normal by visual inspection. We used Cox regression analyses to assess the association between the outcomes relapse-free survival (RFS) and overall survival (OS) and the covariates: age, gender, performance status, white blood cell (WBC) count, platelet count, bone marrow blast percentage, cytogenetic risk, FLT3-ITD and NPM1 mutational status, and MRD (present vs. absent). We then used the C-statistic to quantify a model9s ability to predict outcomes; in this approach, a C-statistics of 1 indicates perfect prediction while a C-statistic of 0.5 indicates no prediction; C-statistics values of 0.6-0.7, 0.7-0.8, and 0.8-0.9 are commonly considered as poor, fair, and good, respectively. Results: Among 595 eligible patient randomly assigned to receive induction chemotherapy with or without GO, 416 patients achieved a CR; of these, MRD data was available for 170 patients: 38 (22%) were MRDpos, and 132 (78%) were MRDneg. Consistent with previous studies by others, on univariate analysis MRD status was statistically significantly associated with RFS (hazard ratio [HR]=2.28 [95% confidence interval: 1.45-3.60], p Conclusion: Although MRD status after induction chemotherapy is statistically significantly associated with RFS and OS and the most important predictor of both outcomes on multivariable analysis, the accuracy of multivariate models predicting these outcomes is only minimally increased when MRD information is included and remains limited. Support: NIH/NCI/NCTN grants CA180944 and CA180819, NCI grant CA182010, and in part by Wyeth (Pfizer) Pharmaceuticals Disclosures Erba: GlycoMimetics; Janssen: Other: Data Safety u0026 Monitoring Committees; Sunesis; Pfizer; Daiichi Sankyo; Ariad: Consultancy; Millennium/Takeda; Celator; Astellas: Research Funding; Seattle Genetics; Amgen: Consultancy, Research Funding; Novartis; Incyte; Celgene: Consultancy, Patents u0026 Royalties. Walter: Covagen AG: Consultancy; AstraZeneca, Inc.: Consultancy; Pfizer, Inc.: Consultancy; Seattle Genetics, Inc.: Research Funding; Amgen, Inc.: Research Funding; Amphivena Therapeutics, Inc.: Consultancy, Research Funding.