Treatment discontinuation due to toxicity for patients with acute myeloid leukemia receiving intensive chemotherapy Open Access

TO THE EDITOR:

Curative-intent therapy for fit patients with newly diagnosed acute myeloid leukemia (AML) centers on the use of cytarabine as part of an intensive multidrug induction regimen.¹ Treatment-related toxicity may lead patients and their physicians to stop curative-intent treatment early, despite the considerable risks of inadequately treated AML. Previous studies have shown that increased adverse events (AEs) are associated with treatment discontinuation.^2-4 SWOG S1203 was a phase 3 trial of patients with newly diagnosed AML aged 18 to 60 years comparing 7+3, idarubicin with high-dose cytarabine (IA), and IA plus vorinostat.⁵ Rates of complete remission (CR) and survival outcomes were similar in the 2 control chemotherapy arms and have been reported previously; the experimental IA plus vorinostat arm was closed early after crossing an interim futility boundary.⁵ Herein, we evaluated the rates of protocol therapy discontinuation owing to AEs in the 2 chemotherapy arms and time to transplant and the effect of performance status (PS).

We compared the 2 standard chemotherapy arms of the S1203 trial: 7+3 (cytarabine 100 mg/m² continuous infusion on days 1-7; daunorubicin 90 mg/m² on days 1-3) and IA (cytarabine: 1500 mg/m² continuous infusion on days 1-4; idarubicin: 12 mg/m² on days 1-3). Reinduction and postremission therapy were protocol specified.⁵ Eligible patients were 18 to 60 years old with untreated AML. Patients were required to have PS ≤3, an ejection fraction of >44%, a corrected QT interval of ≤500 milliseconds, and no known cardiac disease. There were no exclusion criteria for kidney or liver function. CR and complete response with incomplete hematologic recovery (CRi) were defined per contemporary consensus criteria (CR, absolute neutrophil count of ≥1 × 10³/μL, platelet count of ≥100 × 10³/μL, and <5% bone marrow blasts; CRi, same as CR but either absolute neutrophil count or platelet criteria not met). Toxicities were assessed per the Common Terminology Criteria for Adverse Events version 4.0. Investigators reported reasons for treatment discontinuation, which could include treatment toxicity. Rates of discontinuation during induction therapy and early mortality between the 2 arms were compared using Fisher exact test. Time to transplant was measured from date of randomization to date of transplant in first CR or CRi (CR1) with failure to achieve CR1, relapse from CR1, or death before transplant defined as competing events; associations with time to transplant were evaluated using cause-specific hazard models.

Between April 2013 and November 2015, 522 patients were randomized to the 2 chemotherapy arms (261 in each). Treatment discontinuation during induction owing to either death or toxicity occurred in 8 (3%) vs 22 patients (8%) on 7+3 vs IA, respectively (P = .014; Table 1). During the first cycle of induction, 0 patients on the 7+3 arm discontinued treatment owing to toxicity vs 4 patients on the IA arm (0% vs 2%; P = .12). Toxicities leading to treatment discontinuation during initial induction on the IA arm were congestive heart failure (n = 2, 1 with concurrent arrythmia), acute kidney injury (n = 1), and hypotension complicated by stroke and cardiac arrest (n = 1). Two of the 4 patients who discontinued treatment owing to toxicity on the IA arm had achieved CR after their first cycle. The 30- and 60-day all-cause mortality rates were both significantly lower in the 7+3 arm than the IA arm: 30-day mortality rates were 2.7% vs 6.9%, respectively (P = .025), and 60-day mortality rates were 4.6% vs 9.3%, respectively (P = .039).

The inclusion criteria for S1203 allowed patients with PS up to 3, who would have been excluded on that basis from participation in many clinical trials; however, rates of grade 3 to 5 hematologic and nonhematologic AEs were not increased in those with worse PS in either arm. In the 7+3 arm, 86% of those with PS 0 to 1 had at least 1 grade 3 or higher hematologic AE vs 85% of those with PS 2 to 3 (P > .99), and 54% of those with PS 0 to 1 had at least 1 grade ≥3 nonhematologic AE vs 55% of those with PS 2 to 3 (P > .99). In the IA arm, 86% of those with PS 0 to 1 had grade 3 to 5 hematologic AEs vs 85% of those with PS 2 to 3 (P = .77), and 61% of those with PS 0 to 1 had grade 3 to 5 nonhematologic AEs vs 70% of those with PS 2 to 3 (P = .40). For patients with PS 2 to 3, there was a trend toward a higher rate of grade 5 nonhematologic AEs on the IA arm (3/27 patients [11%]) than on the 7+3 arm (0/40 patients [0%]; P = .065); there were no grade 5 treatment-related hematologic toxicities reported in either arm. Allogeneic hematopoietic cell transplantation was recommended during CR1 for patients with adverse risk disease; there was no significant difference in time to transplant between those randomized to 7+3 and IA (P = .21).

Overall, the rates of treatment discontinuation for toxicity or any other reason were low in both the 7+3 and IA arms on S1203, suggesting that both regimens are tolerable in younger patients with previously untreated AML. However, more subjects discontinued IA treatment than 7+3 after initial induction. A 1996 SWOG trial sought to answer a similar question of comparing high-dose cytarabine (2 g/m² per day every 12 hours for 12 doses) with standard-dose cytarabine (200 mg/m² per day for 7 days) plus the anthracycline daunorubicin (45 mg/m² per day for 3 days).⁶ In this trial, neither CR rates nor overall survival was improved in the high-dose cytarabine arm; fatal induction toxicity was significantly greater in the high-dose arm (14% vs 5% for age <50 years and 20% vs 12% for age 50-64 years; age-adjusted 2-tailed P = .0033). Since then, numerous trials have demonstrated low rates of early mortality after either 7+3 or high-dose cytarabine induction regimens, so the cytarabine dose alone may not explain the observations in S1203.^5,7-14 One untested hypothesis is that high-dose cytarabine may be more toxic when given by continuous infusion (as done in S1203) rather than via bolus dosing. Furthermore, anthracycline dose may also matter: the recent DaunoDouble trial suggests that a daunorubicin dose of 60 mg/m² (rather than 90 mg/m² as was used in the S1203 trial) may marginally improve safety without impairing efficacy.¹⁵ 

S1203 had broader inclusion criteria than many AML trials, which often exclude those with PS 3 or impaired organ function. Our analysis shows that patients with PS 2 to 3 did not have a significantly higher rate of treatment-related toxicity than those with better PS. A previous analysis combining results from S0106 and S1203 was not able to accurately predict the occurrence of grade 3 to 5 toxicities based on patient baseline characteristics.¹⁶ The low rate of overall protocol discontinuation owing to toxicity suggests that eligibility criteria for other AML clinical trials could be further broadened to improve patient access.

Given the low treatment-related mortality of 7+3 in the setting of relatively broad inclusion criteria, serious consideration should be given to induction chemotherapy when clinicians are evaluating patients with untreated AML. Since the S1203 trial completed enrollment, the most common regimen for patients unfit for intensive chemotherapy has become hypomethylating agents combined with venetoclax. Randomized trials are ongoing to compare induction with 7+3 or hypomethylating agents/venetoclax in both older and younger patients, including a recently published phase 2b trial comparing induction with decitabine/venetoclax and 7+3 in young fit patients.¹⁷ The data herein support efforts for randomized comparisons with early stopping rules by providing expectations that induction death should be well <10%, and severe nonhematologic/noninfectious complications should be rare (<5%).

This is a secondary analysis of SWOG S1203. For SWOG S1203, the institutional review boards of the participating institutions approved all protocols.

This trial was registered at www.ClinicalTrials.gov (identifier: NCT01802333).

Acknowledgments: Research reported in this publication was supported by grants from the National Cancer Institute/National Institutes of Health (CA180888 and CA180819). S.R. received research funding through the National Heart, Lung, and Blood Institute/National Institutes of Health (T32HL007093). J.S.A received research funding through the Scholar Award from the American Society of Hematology.

Contribution: S.R., M.O., M.-E.M.P., and J.S.A. designed the research; M.O. curated and analyzed the data; S.R. and J.S.A. wrote the manuscript; and all authors reviewed and approved the manuscript.

Conflict-of-interest disclosure: M.O. reports consulting fees from Merck and BioSight; and data safety monitoring board fees from Celgene/Bristol Myers Squibb, Grifols, and GlycoMimetics. M.-E.M.P. reports research funding from AbbVie, Ascentage, Astex, BioSight, Bristol Myers Squibb, Cardiff Oncology, GlycoMimetics, Immunogen, Pfizer, Telios, and Vincerx. The remaining authors declare no competing financial interests.

Correspondence: Jacob S. Appelbaum, Fred Hutch Cancer Center, 825 Eastlake Ave E, MS LG-700, Seattle, WA 98109; email: jappelb@fredhutch.org.