What Is the Optimal Treatment Modality in Molecularly Defined Secondary AML? a Multicenter Cohort Study

Introduction

Secondary ontogeny acute myeloid leukemia (AML), defined by the presence of mutations in ASXL1, BCOR, EZH2, SF2B1, SRSF2, STAG2, U2AF1 or ZRSR2 was recently integrated into both the ICC and WHO diagnostic criteria, as well as into the ELN 2022 risk classification as adverse risk. However, it is unknown whether cytarabine + anthracycline (7+3), liposomal cytarabine and daunorubicin (CPX-351) or hypomethylating agent + venetoclax (HMA+VEN) is the optimal frontline treatment for these patients (pts). We performed the largest multicenter retrospective cohort study to date, dedicated to pts with molecular-defined secondary ontogeny AML, comparing treatment modalities and assessing the impact of clinical and molecular characteristics on response and survival.

Methods

We included pts with newly diagnosed (ND) AML who were treated at 4 large academical centers with 7+3, CPX-351 or HMA+VEN. Secondary ontogeny was defined as absence of TP53 mutation and presence of at least one of the aforementioned mutations. Composite complete response (cCR) was defined as complete response (CR) + CR with incomplete count recovery (CRi). Logistic regression was fitted to evaluate odds ratio (OR) for cCR. Overall survival (OS) was compared between groups by log-rank test. COX regression for OS were fitted to evaluate effect of secondary/co-mutations within each treatment group, as well as the effect of treatment and allogeneic stem cell transplantation (SCT) as a time-varying covariate.

Results

Out of 1132 ND AML pts in the entire cohort, 401 (35%) were molecularly defined as secondary ontogeny and included in the final analysis. Initial treatment was 7+3 in 173 (43%), HMA+VEN in 162 (40%) and CPX-351 in 66 pts (17%). Pts in the HMA+VEN group were older (median age 74 years[yrs]; CPX351 66 yrs; 7+3 63 yrs, p<0.001), while pts in the CPX351 group were more likely to have prior myeloid neoplasm (n=52 [79%]; HMA+VEN n=72 [44%]; 7+3 n=48 [28%], p<0.001) and prior HMA exposure (n=27 [41%]; HMA+VEN n=23 [14%]; 7+3 n=28 [16%], p<0.001).

The cCR rates were 56% (97/173), 56% (90/162) and 44% (29/66) in 7+3, HMA+VEN and CPX-351, respectively (p=0.21). In the 7+3 group, the presence of RUNX1 co-mutation was associated with lower cCR rate (OR 0.8, 95% confidence interval [CI] 0.7-0.95, p=0.01), while PTPN11 was associated with higher cCR rate (OR 1.4, 95% CI 1.1-1.8, p=0.011). In the HMA+VEN group, monosomal karyotype (OR 0.7, 95% CI 0.5-0.9, p=0.002), NRAS co-mutation (OR 0.6, 95% CI 0.5-0.8, p<.001) and SF3B1 mutation (OR 0.76, 95% CI 0.6-0.9, p=0.008) were associated with lower cCR rate and FLT3-TKD with higher cCR rate (OR 1.5, 95% CI 1.1-2.1, p=0.01). In the CPX-351 group, NRAS co-mutation was associated with lower cCR rate (OR 0.65, 95% CI 0.5-0.9, p=0.005).

As most younger pts (aged ≤ 60 yrs, n=85) were treated with 7+3 (n=64) or CPX-351 (n=13) and almost exclusively older pts (age >75, n=78) were treated with HMA+VEN (n=74), we performed age-group based analyses. In pts aged > 60 yrs (n=318), the median OS (mOS) was comparable between groups (7+3: 16 months [mo], 95% CI 13-27; HMA+VEN 15 mo, 95% CI 13-19; CPX-351 11 mo, 95% CI 9-28 p=0.57). Similar results were seen in pts 60-75 yrs (n=238): 7+3 mOS 16 mo, 95% CI 13-27; HMA+VEN mOS 16 mo, 95% CI 13-26; CPX-351 mOS 10 mo, 95% CI 8-28, p=0.74.

The effect of secondary/co-mutations on survival differ between groups: For 7+3 splicing mutations were associated with worse OS, whereas IDH1/2 were associated with improved OS ( Figure). Conversely, in HMA+VEN and CPX351 groups K/NRAS co-mutations were associated with worse survival. NPM1 co-mutation did not affect survival in any treatment group.

In multivariable analysis, age, prior myeloid disease, K/NRAS co-mutations and monosomal karyotype were associated with worse OS ( Table). Conversely, BCOR mutation, IDH1/2 co-mutation and SCT were associated with improved OS. Treatment with HMA+VEN or CPX351 were not statistically different vs 7+3, although there was a trend for better OS with HMA+VEN vs 7+3 (HR 0.68, 95% CI 0.44-1.03, p=0.069).

Conclusion

In pts with ND secondary ontogeny AML the OS with HMA+VEN was at least comparable to 7+3 or CPX351 when adjusted for multiple covariates. The effect of co-mutations on response and OS differed between treatment modalities and may aid in optimal treatment selection in this population. Prospective trials should evaluate the potential superiority of HMA+VEN and effect of co-mutations in secondary ontogeny AML.