Interaction of Remission Status and Cause of Death in Acute Myeloid Leukemia

BACKGROUND: Historically, causes of death for patients dying with active acute myeloid leukemia (AML) and those in remission have differed. However, it is unknown how changes in therapies and advances in supportive care have impacted death patterns and whether these differ for patients with active disease or those in morphologic remission (with or without measurable residual disease [MRD]) at the time of death. Here, we investigated the cause of death (COD) in AML patients and studied whether these differed for patients in morphologic remission with or without evidence of MRD.

PATIENTS AND METHODS:We used an institutional database of adults with newly-diagnosed AML treated with intensive induction chemotherapy from 2006-2017 with either "7+3" (N=140) or cladribine, cytarabine, mitoxantrone and G-CSF ("CLAG-M", N=198). Primary COD, time of death (TOD) and clinical variables were abstracted from chart review. Disease status at TOD was determined by the patient's last bone marrow aspirate/biopsy (BMA) prior to death, unless complete blood count or autopsy clearly showed AML. Complete remission (CR), CR with incomplete hematologic recovery (CRi), and relapse were defined per 2017 ELN criteria. MRD was determined by multiparameter flow cytometry (MFC) and cytogenetics on all BMA. All patients treated on clinical studies gave informed consent in accordance with the Declaration of Helsinki and were treated on Institutional Review Board-approved protocols.

RESULTS: Characteristics of the 338 patients are shown in Table 1. With a median follow-up of 928 days, 187/338 (55%) patients died, 2 of which were excluded from subsequent analysis due to incomplete data. The remaining 185 patients were categorized into 3 groups based on disease status at TOD: remission (CR/CRi and MRD negative), MRD (CR/CRi and MRD positive), and relapse. There were 59 (32%) patients in remission, 13 of which had treatment following their last BMA; this treatment consisted of consolidation chemotherapy or transplant in 11/13 patients. There were 17 (9%) MRD patients and in 6/17 the response to last cycle of treatment was unknown. 109 (59%) of patients were in relapse at TOD, and 24/109 were without BMA assessment following last treatment.

36/59 (61%) remission patients underwent hematopoietic cell transplantation, compared to 5/17 (29%) and 42/109 (39%) in the MRD and relapse groups, respectively. Overall, the median survival from the start of initial induction treatment to death was 327 days, with 85/185 (45%) of patients dying within 60 days of their last treatment. Median time between last cycle of treatment and death was 136 days for the remission group, compared to 53 days in MRD and 50 days in relapse groups.

Finally, we examined how COD varied by disease status at time of death (Table 2). Of patients dying with morphologic AML (N=109), AML was the most common COD (45%) followed by infection (33%), other (9%), and unknown (7%). For remission patients (N=59), infection was the predominant COD (56%) followed by unknown (17%) and transplant-related complications (12%). In MRD positive patients (N=17), AML was the COD in 1 patient (6%), infection occurred in 6 (35%), with unknown (29%) or other causes (18%) accounting for the remaining deaths. Notably, only 9 (5%) of patients underwent autopsy and of the 50 patients dying with AML as primary cause, the majority (32/50, 64%) died while on hospice and thus AML as primary cause was presumptive.

CONCLUSIONS: In our cohort, the majority of deaths occurred within one year of starting induction chemotherapy, and most occurred in patients in active relapse. Only a minor subset of patients dying in morphologic remission had MRD, and COD of patients in the MRD group more closely reflected that of patients in remission. A substantial proportion of deaths occurred within 60 days of treatment, demonstrating that active treatment represents a particularly vulnerable period for patients regardless of disease status. Infection accounted for a substantial proportion of deaths for patients in all disease categories, and is likely undercounted in our study given the high proportion of relapse patients dying on hospice. Thus, despite advances in treatment, infection represents the most common cause of death in AML, highlighting the need for advances in diagnosis and treatment of infections to improve overall outcomes.

View largeDownload slide

View largeDownload slide

Close modal