Impact of minimal residual disease and next generation sequencing discordance in AML following induction with hypomethylating agent + venetoclax.

Abstract

Abstract Introduction: Response assessment after induction is a key milestone in the treatment and care of acute myeloid leukemia (AML). Increasing use of novel assays such multiparameter flow cytometry (MFC) and polymerase chain reaction (PCR) based next generation sequencing (NGS) have expanded testing, leading to increased emphasis on minimal/measurable residual disease (MRD) detection as a valuable disease assessment tool. Previous work described the prognostic value of persistent somatic mutations at time of morphologic complete remission (CR) in AML treated with high-intensity induction chemotherapy (Jongen-Lavrencic, N. Engl. J. Med., 2018). The utility of concordance between MRD and non-MRD assessment in patients treated with hypomethylating agents (HMA) + venetoclax (Ven) has not been adequately described. We report outcomes of patients with AML who achieved a CR with HMA + Ven induction based on concordant or discordant MRD and non-MRD (NGS-based) testing. Methods: Retrospective single center study. Patients treated with HMA + Ven for newly diagnosed AML and achieved a CR. MRD MFC testing completed by outside laboratory (Hematologics, Seattle WA) with sensitivity of 0.02%. NGS via in house assay with reported 5% sensitivity for somatic mutations. Patients stratified by MRD(-)/NGS(-), MRD(-)/NGS(+), MRD(+)/NGS(-), and MRD(+)/NGS(+) for overall survival (OS) and relapse free survival (RFS). MRD(-)/NGS(+) patients were classified as either DTA(+) (presence of only DNMT3A, TET2, and/or ASXL1 mutations) or DTA(-) (other known pathogenic mutations). We analyzed the association between MRD/NGS groups and OS/PFS using a Cox proportional hazards (PH) model adjusted for transplant as a time-dependent covariate. Results: Forty-five patients met inclusion criteria. Median age at diagnosis was 69 years and 26 (58%) were male. Thirty-one (69%) were white and 11 (24%) were black. Twelve (27%) patients had complex cytogenetics and 39 (87%) had de novo AML. The most common somatic mutations were IDH1/2 (20%) DNMT3A (16%), SRSF2 (16%), ASXL1 (13%) and TET2 (13%). Median follow-up from time of CR was 20 months. Sixteen received hematopoietic stem cell transplant (HSCT). Number of patients (N) within each group were: MRD(-)/NGS(-): 17, MRD(-)/NGS(+): 18, MRD(+)/NGS(+): 7, MRD(+)/NGS(-): 3. MRD(-)/NGS(+) included 3 DTA(+) patients and 15 DTA(-) patients. Median OS was 909 days in the MRD(-)/NGS(+) group, 158 days within MRD(+)/NGS(+) group and not reached in MRD(-)/NGS(-) and MRD(+)/NGS(-). Hazard ratio (HR) for OS (compared to MRD(-)/NGS(-)) for MRD(-)/NGS(+): 1.59 (95% CI: 0.46 - 5.45, p=0.5), and MRD(+)/NGS(+): 2.02 (95% CI: 0.45 - 9.1, p=0.4). RFS (compared to MRD(-)/NGS(-)) HR for MRD(-)/NGS(+): 2.50 (95% CI: 0.48 - 12.9, p=0.3), MRD(+)/NGS(-): 2.64 (95% CI: 0.24 - 29.5, p=0.4), and MRD(+)/NGS(+): 5.99 (95% CI: 1.09 - 33.0, p=0.04). Median RFS was not reached in the MRD(-)/NGS(-) group. In the MRD(-)/NGS(+) group, OS (compared to MRD(-)/NGS(-)) HR for DTA(+): 1.23 (95% CI: 0.14 - 11.1, p=0.9) and DTA(-): 1.71 (95% CI: 0.48 - 6.14, p=0.4). RFS HR for DTA (-) was 3.11 (95% CI: 0.60 – 16.1, p=0.2) and there were no relapses within the DTA (+) group. Six patients within the DTA (-) group underwent HSCT and 2 in the DTA (+) group. Conclusion: MRD(-)/NGS(+) showed a trend towards inferior OS and RFS when compared to MRD(-)/NGS(-) among patients who achieved CR with HMA + Ven induction, despite MRD testing being >2 log more sensitive. NGS+ consisting only of DTA mutations trended towards being less deleterious compared to NGS+ with non-DTA mutations. Our findings support prognostic impact of strict clearance of disease at CR and support previous findings in patients treated with intensive therapy. Our results suggest that MFC MRD negativity may not adequately prognosticate relapse alone – either due to intrinsic issues with flow-based disease assessments or possibly that NGS+ occurred more often in secondary/higher-risk AML with more complex clonal architecture. Median OS was not reached in MRD(+)/NGS(-) patients – likely due to small sample size, though it suggests the importance of mutational clearance at time of CR. Limitations of this study include small sample size and a single-center population limiting demographic diversity. Future directions will include site expansion, increased patients and refining genetic subgroups to further evaluate their impact on relapse and survival.