Identification of predictive, treatment-specific Charlson comorbidity score thresholds in acute myeloid leukemia.

Abstract

e18887 Background: Acute myeloid leukemia (AML) is an aggressive hematological malignancy with a 5-year overall survival of 24.0%. Outcomes of AML stratified by treatment regimen exhibit considerable heterogeneity between clinical trials and the real-world setting. Therefore, the selection of treatment approaches for AML patients needs improvement. As clinical trials frequently account for comorbidities in inclusion criteria, we assessed if the Charlson Comorbidity Index (CCI) score could be used as an adjunctive measure to predict survival in patients with AML treated outside of clinical trials, stratifying by the chemotherapy regimen. Methods: We retrospectively analyzed 522 patients with AML across 1,023 phases of treatment at VCU Massey Cancer Center from January 2013 to January 2022. We separated patients into four cohorts stratified by treatment phase and strategy: upfront anthracycline and cytarabine (conventional 7+3), upfront and salvage venetoclax and a hypomethylating agent (HMA), and salvage FLAG-based strategies. Patients that did not have a CCI score available for retrospective review were excluded. We analyzed cohorts starting at a Charlson Comorbidity Index (CCI) score threshold of 2 and incrementally increasing until a significant difference was discovered between groups. We analyzed survival by the Kaplan-Meier method and compared groups with the Mantel-Cox test. The date of death was used to calculate overall survival; patients were censored at the date of the last contact. Results: In the 7+3 cohort, 67 patients with a CCI score of 2 had significantly improved overall survival compared to 255 patients with CCI scores of greater than 2 (not reached at a median follow-up time of 4.8 y. versus 16.6 m., p = 0.0004). In the upfront venetoclax cohort, a CCI score threshold of less than or equal to 6 identified a cohort of 50 patients with significantly improved survival compared to 29 patients with scores greater than 6 (8.4 m. versus 3.8 m., p = 0.035). In the relapsed venetoclax cohort, 36 patients with a CCI score of less than or equal to 5 had significantly superior survival compared to 21 patients with scores of greater than 5 (9.2 m. vs 4.4 m., p = 0.018). In the FLAG cohort, 78 patients with a CCI score threshold of 4 or lower had superior survival that approached significance compared to 24 patients with CCI scores greater than 4 (14.8 m. versus 8.8 m., p = 0.126). Conclusions: The Charlson Comorbidity Index score can be used as an adjunctive measure to predict survival in patients with AML undergoing chemotherapy. We identified clinically relevant CCI score thresholds for conventional 7+3, venetoclax + HMA, and FLAG-based regimens in the upfront and relapsed or refractory setting that may improve the selection of treatment candidates in AML outside of clinical trials.