Safety and Efficacy of G-CSF with Intensive Chemotherapy in Newly Diagnosed Acute Myeloid Leukemia: A Subgroup Analysis of the Phase II Trial of Venetoclax in Combination with Cladribine, Idarubicin, and Cytarabine

Introduction

Granulocyte colony-stimulating factor (G-CSF) support is common in the treatment of acute myeloid leukemia (AML) to hasten count recovery, particularly during periods of prolonged neutropenia and infection. However, the safety and efficacy of G-CSF in combination with newer induction strategies — such as intensive chemotherapy with venetoclax — remain unclear.

Methods

This study analyzed 88 patients with newly diagnosed AML treated in the phase II trial of cladribine, idarubicin, and cytarabine in combination with venetoclax (CLIA-venetoclax, NCT02115295). The primary objective was to quantify the impact of G-CSF on the time to count recovery. Secondary objectives included comparing the incidence of grade ≥3 toxicities, overall survival (OS), and relapse-free survival (RFS) of patients treated with and without G-CSF support. An exploratory objective was to determine the molecular predictors of count recovery.

Patients received cladribine 5 mg/m² D1–5, cytarabine 1–1.5 mg/m² D1–5, idarubicin 8–10 mg/m² D1–3, and venetoclax 400 mg D2–8, with dose adjustments for CYP3A inhibitors, for up to 6 cycles. FLT3 and BCR-ABL1 inhibitors were permitted. Antimicrobial prophylaxis was administered with the regimen. The trial allowed treating physicians to use filgrastim 300–480 µg daily for prolonged neutropenia or neutropenic fever; a trial amendment integrated pegfilgrastim at a dose of 6 mg SQ once between D5–9 of each cycle. We defined count recovery as an absolute neutrophil count (ANC) ≥1000 cells/mm³ with platelets ≥100 × 10⁹/L.

Results

In the overall cohort of 88 patients, 85 (97%) had AML and 3 (3%) had MPAL. The median age at diagnosis was 49 years (range, 18–64). Twelve (14%) patients had AML-MRC; 33 (38%) harbored secondary-type mutations defined by ELN 2022 (ASXL1, BCOR, EZH2, RUNX1, SF3B1, SRSF2, STAG2, U2AF1, or ZRSR2). Sixteen (18%) patients had FLT3-ITD — 8 (9%) received gilteritinib and 1 (1%) received midostaurin. Forty-two (47%) patients received G-CSF support in cycle 1, and 64 (73%) received G-CSF support in any cycle.

G-CSF use in the first cycle significantly improved the time to count recovery (25 d. with G-CSF vs 29 d. without, p = 0.028). The median ANC at the best response was significantly better in the G-CSF cohort compared to those treated without (2800 vs 1560 cells/mm³, p = 0.042). G-CSF use in the first cycle was associated with a non-significantly higher proportion of patients that achieved a complete response by day 28 (82% with G-CSF vs 67% without, p = 0.203). MRD negativity by multiparameter flow cytometry at D28 was not significantly different between cohorts (76% with G-CSF vs 70% without, p = 0.783). There was no difference in the proportion of patients refractory to the first cycle (5% with G-CSF vs 14% without, p = 0.269). In a multivariate Cox model analyzing the impact of 9 mutations on the time to count recovery, only FLT3-ITD was associated with delayed count recovery (HR: 0.403, p = 0.004). In contrast, KRAS mutations were associated with swifter count recovery (HR: 3.435, p = 0.005).

G-CSF use significantly reduced the incidence of bacteremia in cycle 1 (5% with G-CSF vs 35% without, p < 0.001). Decreased rates of bacteremia appeared to be attributable, in part, to reduced rates of gram-positive infections (S. mitis/oralis, S. epidermidis, and S. aureus; 2% with G-CSF vs 24% without, p = 0.004). These findings translated to a reduction in neutropenic fever during induction that was nearly significant (64% with G-CSF vs 83% without, p = 0.057).

There were no differences in the ICU admission rate (0% with G-CSF vs 7% without, p = 0.242) or in the duration of hospitalization during cycle 1 (28 d. with G-CSF vs 25 d. without, p = 0.192). G-CSF use during any cycle did not significantly impact the overall survival (not reached [NR] in both cohorts, 2-y. OS: 78% with G-CSF vs 63% without, p = 0.201). G-CSF did not significantly impact relapse rates, with fewer relapses seen with G-CSF (NR in both cohorts, 2-y. relapse incidence: 11% with G-CSF vs 28% without, p = 0.143).

Conclusion

G-CSF support with intensive chemotherapy and venetoclax appeared to be safe and effective — significantly improving the time to count recovery. G-CSF use decreased the incidence of bacteremia, particularly for gram-positive infections. FLT3-ITD AML was associated with delayed count recovery, likely due to concurrent TKI use. G-CSF use had no impact on the incidence of AML relapse.