Real-world AML survival paradox: Early escalation vs. sustained remission

Abstract The prognostic value of measurable residual disease (MRD) is well established in acute myeloid leukemia (AML), with strong evidence suggesting that MRD analysis can guide personalized treatment decisions. Failure to achieve MRD negativity after initial therapy represents a strong adverse prognostic factor in AML patients. In this retrospective real-world study, we utilized early treatment escalation (defined as additional therapy within 6 months after first-line treatment) as a proxy for MRD positivity, while no escalation within this timeframe served as a proxy for sustained remission (MRD negativity). Our study aims to measure the prognostic impact of early treatment escalation—a real-world event highly correlated with refractory disease or early relapse often indicated by MRD positivity—on overall survival (OS) rates in AML patients. This study evaluated adult AML patients who received standard induction ± consolidation therapy from the TriNetX federated electronic health record (EHR) network. Patients were stratified into two cohorts: Cohort A (early escalation, n=1,460), comprising patients who received either second-line chemotherapy (including fludarabine, cladribine, gilteritinib, enasidinib, revumenib, olutasidenib, etoposide, or clofarabine) or hematopoietic stem cell transplant (HSCT) as salvage therapy within 6 months of completing initial therapy; and Cohort B (no escalation, n=10,540), consisting of patients who received no additional therapy during this same period. Overall survival (OS) was defined as the time from the index date (end of initial therapy) to death. We compared 1-year, 3-year, and 5-year OS between cohorts using Hazard Ratio (HR) and 95% confidence interval (CI), with 1:1 propensity score matching employed to balance age and demographic variables between groups. Early escalation was associated with superior overall survival (OS). In the propensity-matched analysis (n=1,360 each), 5-year OS was significantly higher for Cohort A (early escalation) at 62.5 % compared to 54.6 % for Cohort B (no early escalation) (Log Rank Test: χ2 =14.591, p< 0.001) (HR=0.76, 95%CI: 0.663,0.877) ). One-year OS was 71.5 % vs 66.4 %, and 3-year OS 65.4 % vs 57.3 %, consistently favoring the early escalation group. Median OS was not reached in either cohort. Key baseline differences were observed prior to matching, with Cohort A patients being approximately 6 years younger on average than Cohort B (mean 53 vs 59.1, p<0.001). Patients who required early salvage therapy demonstrated better long-term survival than those who did not , presenting a paradoxical finding. This likely reflects both selection and intervention effects: younger, fitter patients with early relapse received effective salvage therapy, leading to prolonged survival, whereas some MRD-negative patients in sustained remission later relapsed or died without receiving salvage treatment. These results may also reflect the impact of newer targeted therapies for relapsed/refractory disease, allowing for improved outcomes compared to those achieving complete remission with standard-of-care regimens. Our findings underscore that MRD status must be interpreted in the context of subsequent treatment options. While early relapse) typically portends poor prognosis if left untreated, timely and effective therapeutic intervention can substantially improve survival outcomes. This analysis suggests that early aggressive intervention mitigated some of the adverse biology of early relapse. These results have important clinical implications. First, formal risk stratification in AML should consider patient age in addition to fitness. With available salvage options and the advancement of targeted therapies an early relapse is not invariably fatal. Second, these findings support investigating the use of targeted therapies in the up-front setting. Our findings highlight the complex interplay between disease biology and treatment interventions: achieving MRD-negativity remains critical, but when MRD-positivity occurs, subsequent treatment decisions based on disease genetics prove critical. Prospective studies incorporating molecular MRD monitoring are needed to confirm these findings and further optimize post-remission therapy approaches.