Outcomes of older or unfit AML patients admitted to intensive care unit in the venetoclax era

Abstract

Abstract Background Venetoclax (VEN) has transformed the management of patients (pts) with acute myeloid leukemia (AML) ineligible for intensive chemotherapy. When combined with azacitidine (AZA), approximately 2/3 of pts achieve complete remission (CR) and median overall survival (OS) is 14 months. Although the benefit of early intensive care unit (ICU) admission has been shown in young AML pts, there is limited data on non-intensively treated pts admitted to the ICU. However, this issue is becoming increasingly important in routine practice, as some pts can now expect prolonged long-term survival. Methods In this multicenter retrospective study, we collected data on adult pts treated with VEN-AZA for newly diagnosed AML, admitted to the ICU between Jan 2018 and Feb 2025, in 14 French centers from the FILO group. We describe the outcome of pts admitted to the ICU before VEN-AZA initiation (“diagnosis” group) or after starting VEN-AZA for a complication (“post VEN-AZA” group). Post ICU OS was calculated from the date of ICU admission to death or last FU. Results We included 183 pts treated upfront with VEN-AZA (n=151, 82.5%) or VEN-AZA in alternance with cladribine/low dose ARAC (n=11, 6%) or triplets including targeted therapies or immunotherapies (n=21, 11.5%). Median age was 73 years (range, 19-85) and 123 pts were male (67.2%), 45 (24.6%) had a prior cancer, 43 (23.5%) a prior MPN and 24 (13%) a prior MDS including 8 (4.4%) previously treated with AZA. ELN 2024 risk was favorable (fav.) in 70 (38.3%), intermediate (int.) in 36 (19.7%), poor in 39 (21.3%) and missing for 38 (20.8%) pts. Seventy-four (40.4%) were admitted to the ICU in the “diagnosis” group, and 109 (59.6%) in the “post VEN-AZA” group, including 69 (37.7%) during cycle 1, 19 (19.4%) during cycle 2 and the 21 remaining pts (11.5%) after cycle 2. Among the 142 pts evaluable for response, best response was CR/CRi in 66 pts (46.5%). Main reasons for ICU admission were acute respiratory failure (37.7%), shock (20%), hyperleukocytosis and/or tumor lysis syndrome prophylaxis (TLS, 15.8%), sepsis (12%), acute kidney injury (7.7%) and other causes (6%). Median Sequential Organ Failure Assessment (SOFA) at admission was 6 (range, 1-18). During ICU stay, 141 pts (77%) needed respiratory assistance, including 40 (21.9%) requiring mechanical ventilation and 49 (26.8%) non-invasive ventilation and/or high flow oxygen. Seventy-four (40.4%) required vasopressors and 23 (12.6%) renal replacement therapy. Median duration of ICU stay was 5 days (range, 0-33). Sixty-four pts (34.9%) died within 30 days after ICU admission. Median post ICU OS was 3.4 months (1.7-5.1) and 12-month post ICU OS was 28.5%. In multivariate analyses for post ICU OS, taking into consideration gender, age, ELN 2024 risk, disease response and SOFA score, the factors that remained significant were the absence of response (HR=5.28, IC95%, 1.54-18.1, p=0.008) and a high SOFA score (HR=1.23, IC95%, 1.05-1.44, p=0.011). We compared “diagnosis” vs “post VEN-AZA” subgroups. Pts of the “diagnosis” group were less often male (52.7% vs 77.1%). ELN 2024 classification was fav., int. and poor in 48.3%, 36.2% and 15.5% of patients vs 48.3%, 17.2% and 34.5% of pts, respectively. SOFA score was 6 (range, 1-13) vs 6.5 (range, 1-18) in the two groups. In the “diagnosis” group, more pts were admitted to the ICU because of hyperleukocytosis/TLS (=37.8% vs 1%), or shock (29.4% vs 8.1%) compared to the “post VEN-AZA” group. Organ replacement was different as 59.3% vs 9.5% needed amines in and 31.2% vs 8.1% had mechanical ventilation in the diagnosis and the post VEN-AZA group, respectively. In the post VEN-AZA group, 37/65 (56.9%) evaluable pts were in response at time of ICU admission. ”Diagnosis” pts had a Day-30 post ICU OS of 89% (vs 48%). Median post ICU OS was 7.3 months (IC95%, 2.2-12.7) in the “diag” group, and 42% and 22% of pts were still alive at 1-y and 2-y, respectively. By contrast, pts from the “post VEN-AZA” group had a median post ICU OS of 0.8 month (IQR 0-2.4), and only 19% were alive at 1-y and 0% at 2-y. Conclusion Early ICU management, prior to initiating VEN-AZA, is associated with prolonged OS. By contrast, older AML pts admitted to the ICU after the beginning of VEN-AZA have a dramatically poor outcome (<1 month). This may help ICU admission discussion conducted between intensivists, hematologists and palliative care practionners, which also takes pts and families choices in consideration.