Outcomes of Patients with Treated Secondary AML: A High Risk Subtype of AML Arising from Previously Treated Myeloid Neoplasms That Warrants an Independent Prognostic Designation

Introduction: Outcomes of patients (pts) with acute myeloid leukemia (AML) are improving, but certain high-risk subsets continue to fare poorly. Outcomes of pts with AML secondary (s-AML) from a prior myeloid neoplasm are usually adverse and often influenced by their underlying genomics. Pts who has had therapy for their prior myeloid neoplasms who then transform to AML have a particularly poor prognosis. These pts with ‘treated secondary’ AML (TS-AML) have been often classified as newly diagnosed AML in clinical trials but warrant an independent prognostic designation and risk stratification. Methods: We retrospectively analyzed adult pts (≥ 18 years) with newly diagnosed AML who had transformed after prior myelodysplastic syndrome (MDS), chronic myelomonocytic leukemia (CMML) or MDS/myeloproliferative neoplasms (MPN) overlap syndrome after therapy with hypomethylating agents (HMA), prior chemotherapy and/or allogeneic stem cell transplantation (HSCT). Pts with prior MPN were excluded. Therapy for antecedent myeloid disorder, genomics at the time of AML diagnosis, AML therapy intensity and use of venetoclax, hematopoietic stem cell transplantation (HSCT) in first remission from AML were captured. CRc (CR+CRi) and overall response rate (ORR; CRc+ MLFS) to frontline AML therapy were tabulated. ELN 2017 guidelines were used for risk stratification. Relapse free survival (RFS) was captured from time of best ORR to relapse/death/last follow-up (LFU) and overall survival (OS) from therapy initiation to death/LFU. Results: From Jan 2012 to Aug 2023, 673 pts with a median age of 70 years (range, 19-94) were included: 536 (80%) of whom transformed from MDS and the remainder from CMML/MDS-MPN. The median lines of therapy received prior to diagnosis of AML was 1 (1-8) and 285 pts (42.3%) had received ≥2 lines of therapy. 149 pts (22%) had also received prior therapy for non-myeloid malignancies. HMA with/without other agents was the most common prior therapy (658 pts, 97.8%); and 87 pts (12.9%) had undergone an HSCT prior to AML transformation. No patient had a core binding factor AML. Amongst evaluable pts (n=497), 106 (21%) had a diploid karyotype and 289 (58%) pts had adverse risk (AR) cytogenetics (including 202 [41%] with complex karyotype). 163/483 (34%) pts had a TP53 mutation and overall, 428/602 (71%) classifiable pts were ELN AR. A majority of pts (452, 67%) received low-intensity therapy for AML and overall, 181 pts (27%) received venetoclax. CRc and OR was attained in 174 (26%) and 246 (37%) pts respectively. Amongst pts treated with low-intensity therapy (LIT), venetoclax increased odds of CRc (OR=2.1, 95%CI 1.4-3.3) and ORR (OR=2.5, 95%CI 1.6-3.7). At a median FU of 43 months, the median RFS and OS were 4.6 and 4.8 months respectively, with med OS 5.2 and 4.5 months for pts <60 years and ≥60 years at AML diagnosis. Median OS for responders was 10.7 months. Median OS for pts with wild type TP53 and non-AR cytogenetics also remained dismal at 7.8 months; stratified by ELN risk OS was 8.2, 6.1 and 4.4 months for favorable, intermediate and AR groups respectively. OS was superior in pts who underwent an HSCT at 13.8 months compared to 10.5 months in a landmarked population of pts <70 years of age who attained an OR but did not undergo HSCT (p=0.006); 1year OS was 55% vs. 37% and 2-year OS 35% vs. 21%, respectively for pts receiving HSCT or not. On multivariate Cox analysis (MVA) for OS, prior therapy for non-myeloid disorder (HR=1.30, 95% CI 1.07-1.59), ≥2 lines of therapy for antecedent myeloid disorder (HR=1.23, 95%CI=1.05-1.23) and ELN AR (HR=1.47, 95% CI 1.22-1.79) increased hazards of death while HSCT (HR=0.50, 95% CI 0.36-0.69) was favorable; Venetoclax was not independently significant. In similar analysis, but including pts only treated with LIT, Venetoclax (HR=0.75, 95% CI0.60-0.94) and HSCT (HR=0.30, 95%CI 0.16-0.47) were favorable while prior therapy for non-myeloid disorder and ELN AR remained detrimental. On gradient boosted regression modelling HSCT strongly predicted survival at 2 years followed by non-AR AML, though only 45 pts (6.7%) attained this survival endpoint. Conclusion TS-AML is associated with dismal survival outcomes irrespective of underlying genomics at AML diagnosis. The inclusion of TS-AML as an independent AR category is critical for proper prognostication and reporting of clinical trials. HSCT improves survival and should be performed in eligible pts.