Achievement of Measurable Residual Disease (MRD) Negativity Supersedes Treatment Intensity in Predicting the Outcome of Patients with Acute Myeloid Leukemia

Background: Standard therapies for acute myeloid leukemia (AML) include intensive chemotherapy and low-intensity regimens selected based on patient age and/or comorbidities. Both are effective in inducing deep and durable responses. Measurable residual disease (MRD) is a strong, independent prognostic factor for predicting overall survival (OS) and relapse in AML. We hypothesized that patients (pts) achieving MRD negativity may experience similar outcomes irrespective of the treatment regimen applied to achieve the response.

Methods: In this retrospective chart review study, we selected all newly-diagnosed AML pts treated at our institution between 2010 and 2021 who had achieved a first response (CR, CRi, or MLFS) and had undergone MRD testing by flow cytometry at the time of best response. Pts with core-binding factor AML and acute promyelocytic leukemia, as well as pts who had received prior therapy for antecedent myelodysplastic syndrome, were excluded. We divided pts into 2 cohorts: cohort IA consisting of pts receiving intensive chemotherapy based on intermediate to high-dose cytarabine plus an anthracycline (without venetoclax), and cohort LO + VEN consisting of pts treated with a low-dose cytarabine or hypomethylating agent backbone plus venetoclax. The Kaplan Meier method was used to calculate OS and time-to-relapse (TTR). Pts who underwent stem cell transplantation (SCT) were censored at the time of transplant. Multivariate analysis was performed using a Cox proportional hazards model.

Results:
We identified 635 pts meeting inclusion criteria (385 IA, 250 LO + VEN). Compared to pts treated with IA, pts treated with LO + VEN were significantly older (median age 71.6 vs 52.8 yrs), had a lower rate of true CR (76% vs 89%), had a higher rate of CRi/MLFS (24% vs 11%), were less likely to achieve MRD-negativity (60% MRD(-) vs 71%), and were less likely to have undergone SCT (25% vs 50%). In addition, the LO + VEN cohort was enriched for ELN 2017 adverse risk disease (58% vs 34% in the IA cohort) and adverse risk cytogenetic features such as -5/5q- (19% vs 7%), -7/7q- (17% vs 7%), -17/17p- (12% vs 5%), and complex karyotype (29% vs 16%).
The median OS was 51m, 24.6m, 15.1m, and 9.9m in the IA MRD(-), LO + VEN MRD(-), IA MRD(+), and LO + VEN MRD(+) cohorts, respectively (fig. 1A). The median TTR was 59.9m, 55.9m, 8.9m, and 5.9m in the IA MRD(-), LO + VEN MRD(-), IA MRD(+), and LO + VEN MRD(+) cohorts, respectively (fig. 1B). When comparing the IA MRD(-) cohort to the LO + VEN MRD(-) cohort, the IA-treated patients had superior OS (p=0.002, fig. 1A) but similar TTR (p=0.830, fig. 1B). Similar findings were obtained in the MRD(+) pairs (fig. 1A-B). A subset analysis evaluating only pts aged ≥ 60 yrs (n=297) showed no statistically significant differences in OS or TTR within MRD categories (median OS 28.9m in IA MRD(-) vs 24.6m in LO + VEN MRD(-), p=0.09; median OS 15.1m in IA MRD(+) vs 10.6m in LO + VEN MRD(+), p=0.51; median TTR not reached in IA MRD(-) vs 35.7m in LO + VEN MRD(-), p=0.32; median TTR 9.6m in IA MRD(+) vs 6.0m in LO + VEN MRD(+), p=0.43)
Given the imbalances in patient and disease characteristics between the treatment cohorts, we performed a multivariate analysis on the full cohort to better elucidate the relative prognostic impact of MRD and treatment intensity. By univariate analysis, age, treatment cohort (IA vs LO + VEN), best response (CR/CRi/MLFS), MRD status, and ELN risk were significant predictors for OS and were included in the multivariate model. By multivariate analysis, age (HR 1.02, p=0.026), best response (HR 1.41 for CRi, p=0.066; HR 2.69 for MLFS, p<0.001), MRD-status (HR 1.53 for MRD(+), p=0.007), and ELN category (HR 1.34 for intermediate, p=0.161, HR 1.96 for adverse, p=0.001) remained significant predictors of OS. For TTR, best response (HR 2.15 for CRi, p<0.001; HR 3.78 for MLFS, p<0.001), MRD-status (HR 1.94 for MRD(+), p<0.001), and ELN category (HR 1.79 for intermediate, p=0.014, HR 2.89 for adverse, p<0.001) remained significant. Treatment received was not predictive of either OS or TTR by multivariate analysis.

Conclusions: MRD status is a stronger predictor of OS and TTR than the intensity of therapy received. These data suggest achievement of MRD-negativity should be a major objective of AML therapy.