Elevation of fetal hemoglobin is a biomarker for response to decitabine in AML: Results of the AML21 randomized phase III trial of the EORTC leukemia group, gimema and german MDS study group Free

Introduction

DNA-hypomethylating agents (HMA) have become standard treatment for patients (pts) with acute myeloid leukemia (AML) who are unable or unwilling to receive intensive chemotherapy. Lack of response or delayed response to therapy are major challenges in the use of HMA. Therefore, outcome predictors are needed. One potential biomarker for response to HMA treatment is fetal hemoglobin (HbF). Results from retrospective, single-center studies suggest that HbF induction following decitabine (DEC) or azacitidine treatment is associated with superior outcome in pts with AML or myelodysplastic neoplasms (Stomper et al., Haematologica 2019, Saga et al., Intern. Med. 2024). To validate these findings in a large, prospective, multi-center study, we investigated the association between HbF and outcome in the EORTC Leukemia Group trial AML21, a randomized phase 3 study, which included 606 fit, older AML pts who received 10-day DEC or 3+7 chemotherapy as induction, with similar overall survival but better tolerance and health-related quality of life in the DEC arm (Lübbert et al., Lancet Haematol. 2023). We hypothesized that HbF levels measured at the end of DEC infusions (i.e., at day 10) and at baseline are associated with outcome.

Methods

The EORTC trial AML21 was conducted across 9 European countries and enrolled pts aged ≥ 60 years with newly diagnosed AML who were eligible for intensive chemotherapy. Pts in the DEC arm received DEC (20 mg/m² for 10 days in the first cycle and for 5 or 10 days every 28 days in subsequent cycles). Pts in the 3+7 chemotherapy arm received daunorubicin (60 mg/m² days 1-3) and cytarabine (200 mg/m² days 1-7), followed by 1-3 additional chemotherapy cycles. HbF was measured in EDTA-anticoagulated blood by high-performance liquid chromatography at baseline and, in the DEC arm, also at day 10 of treatment. Elevated HbF was defined as HbF >1.0%.

Results

A total of 606 pts with AML underwent randomization. A total of 303 pts each were randomly assigned to the DEC arm and to the 3+7 chemotherapy arm. HbF was measured in 155 pts in the DEC arm and 97 pts in the 3+7 arm. 135 pts in the DEC arm had HbF measured both at baseline and day 10 of treatment. Pts in the 3+7 arm had HbF measured only at baseline. In the DEC arm, the median day 10 HbF level was the same as median baseline HbF (0.5%). 46 pts in the DEC arm had an increase in HbF between baseline and day 10; 89 pts had no increase. The ratio of day 10 to baseline HbF was higher in pts who reached complete remission (CR) or CR with incomplete count recovery (CRi) than in those who did not (median ratio of 1 and 0.86, respectively, p=0.037). HbF increase from baseline to day 10 in the DEC arm was associated with increased odds of reaching CR/CRi in both unadjusted analysis (odds ratio [OR] 2.40, 95% confidence interval [CI] 1.15-5.02, p=0.020) and after adjusting for age, white blood cell count at diagnosis, and sex (OR 2.51, 95% CI 1.17-5.38, p=0.018). The hazard ratio (HR) for death comparing pts with an increase in HbF between baseline and day 10 to those without was 0.83 (95% CI 0.55-1.28, p=0.40).

Baseline HbF in the DEC arm was comparable between pts who reached and who did not reach CR/CRi (median HbF = 0.5% and 0.45%, respectively; p=0.69). The associations between baseline HbF and CR/CRi (OR 0.89, 95% CI 0.45-1.75, p=0.73) or overall survival (HR 0.86, 95% CI 0.58-1.28, p=0.47), when comparing pts with elevated versus normal HbF, were not statistically significant but showed a direction that was consistent with previous reports.

In the 3+7 arm, baseline HbF was also similar between pts who reached and who did not reach CR/CRi (median HbF = 0.4% and 0.25%, p=0.42) and showed no evidence of an association with CR/CRi (OR 0.89, 95% CI 0.32-2.51, p=0.83) or overall survival (HR 0.93, 95% CI 0.52-1.68, p=0.81) when comparing pts with elevated versus normal HbF.

Conclusions

This is the first large prospective study evaluating the potential predictive value of HbF as an easy-to-measure early biomarker for response to HMA treatment. The statistically significant association between HbF increase from baseline, already at the early timepoint of day 10, and CR/CRi indicates that HbF may indeed be a useful biomarker, warranting further research. In the prospective, randomized phase 3 trial DECIDER-2 (DEC/venetoclax +/- ATRA in newly diagnosed AML pts), the predictive value of serial HbF measurements is currently being studied.