A Phase II Randomized Trial Comparing Low-Dose Cytarabine and Venetoclax +/- Midostaurin in Non-Adverse Cytogenetic Risk Acute Myeloid Leukemia: The ALLG AMLM25 Intervene Trial

Background

For fit patients (pts), combining kinase inhibitors (midostaurin, quizartinib) with intensive chemotherapy is standard of care for FLT3mutated AML. New studies also suggest a role for kinase inhibitors (e.g. quizartinib) in modifying disease history in pts with FLT3wild-type AML (Montesinos et al, EHA 2023). Clonal evolution of kinase-activating mutations, including FLT3-ITD is an important mechanism of treatment failure in pts with non-adverse (NON-ADV) cytogenetic risk AML receiving frontline therapy with azacitidine and venetoclax (AZA-VEN) (DiNardo and Tiong et al, Blood 2020). Incorporating kinase inhibitors (e.g. gilteritinib) into less-intensive VEN-based regimens in unfit, older pts has been challenging, with cumulative myelosuppression a dominant issue (Short et al, JCO 2024). Gilteritinib is known to have a long half-life (~113 hrs), which may contribute to its negative impact on marrow recovery when directly combined with chemotherapy and VEN. We hypothesized that myelosuppression could be lessened by using a FLT3 inhibitor with a short half-life (e.g. midostaurin, MIDO) administered sequentially after, rather than concurrently with chemotherapy. The phase 1b/2 INTERVENE study was conducted across 19 sites in Australia and New Zealand to establish the safety and efficacy of a triplet regimen involving low dose cytarabine (LDAC), VEN and MIDO (LVM) in older unfit pts with NON-ADV AML. A phase 1b dose-finding study confirmed safety in 18 pts and determined the recommended phase 2 dose (RP2D) of LVM to be LDAC 20 mg/m² SQ D1-10, VEN D1-28 with dose ramp-up to 600mg D and MIDO 50 mg BD D11-28 (Chua et al, ASH 2022). INTERVENE was designed to seamlessly transition to a randomized phase 2 expansion comparing the RP2D of LVM with LDAC+VEN (LV), the focus of the current report.

Methods

Pts aged ≥60 years (y) unfit for intensive chemotherapy with treatment naïve NON-ADV cytogenetic risk AML (excluding APL or core-binding factor) were randomised 2:1 to receive LVM or LV. Posaconazole antifungal prophylaxis was permitted with dose adjustment of VEN to 50 mg daily and MIDO to 50 mg daily due to increased risk of cardiac toxicities in older populations. The primary endpoint was CR/CRi by end of C4 utilizing dual criteria for proof of concept. Secondary endpoints included other measures of response, duration of response and survival. Exploratory endpoints included FLT3-ITD measurable residual disease (MRD) by PCR-NGS (sensitivity 10^-5). First patient was 10MAR2021, with data cut-off 31MAY2024.

Results

124 pts were randomized with 117 commencing study therapy (76 LVM, 41 LV). Median follow up was 13.4 months. Median age was 74y (range 60-89, 81% ≥70y) and 66% were male. In the LVM vs LV arms, secondary AML was present in 38% vs 26%, prior AZA exposure in 14% vs 3%, adverse European LeukemiaNet 2022 risk in 53% vs 56%, FLT3-ITD in 27% vs 23% and FLT3-TKD in 3% vs 3%, respectively.

Overall, 30-day mortality was 6% (4% LVM vs 10% LV): 4 due to infections, 3 due to progressive AML. In the LVM and LV arms, the most frequent non-hematological adverse events in cycle 1 (C1) were nausea (47% vs 22%, G3+ 1% vs 0%), constipation (29% vs 0%, no G3+) and cardiac symptoms (7% vs 5%, G3+ 4% vs 0%). G3 febrile neutropenia occurred in 28% vs 17%.

Regarding deliverability, the median number of cycles received was 5.5 (range 1-24) for LVM and 4 (range 1-31) for LV. For pts achieving CR/CRi/CRh, the proportion receiving >6 cycles in the LVM and LV arms was 62% vs 56%, including 38% vs 36% receiving >12 cycles, respectively. For pts in remission, the proportion with delayed treatment (commencing next cycle >day 42) in LVM vs LV were 24% vs 7% at the end of C1, declining to 13% vs 13% at the end of C6, and 11% vs 9% at the end of C12, respectively, with no evidence of cumulative delays after successive treatment cycles.

Of 96 patients dosed and completing at least 4 cycles of therapy or off therapy prior to C4, overall response rates (ORR: CR/CRi) were 70% in the LVM arm and 64% in the LV arm. The final analysis of response will be presented at the conference. Among pts with FLT3 mutation, ORR in the LVM and LV arms were 86% vs 50%, respectively. FLT3­-ITD MRD clearance was observed in 9/15 (60%) in the LVM arm and 1/4 (25%) in the LV arm.

Conclusion

In unfit, older pts ≥60 years with newly diagnosed NON-ADV cytogenetics AML, LDAC-VEN-MIDO, with sequential delivery of MIDO after LDAC was well tolerated, with promising clinical and molecular responses in FLT3 mutated AML.