The Prognostic and Predictive Value of RUNX1 Mutations in Newly Diagnosed Acute Myeloid Leukemia – an International Multicenter Cohort Study

Introduction: RUNX1 mutations (RUNX1m) are defined as acute myeloid leukemia (AML), myelodysplasia-related (AML-MR) by the 2022 ICC but not by the 2022 WHO diagnostic criteria. RUNX1m are also considered adverse risk by the 2022 ELN criteria. However, the independent prognostic impact of RUNX1m is less clear because they frequently co-occur with secondary ontogeny mutations (ASXL1, BCOR, EZH2, SF3B1, SRSF2, STAG2, U2AF1, and ZRSR2). There is also uncertainty about the predictive value of RUNX1m for treatment with intensive chemotherapy (IC) or hypomethylating agents plus venetoclax (HMA+Ven). Hence, we performed a large international multicenter retrospective cohort study in patients (pts) with newly diagnosed (ND) AML with confirmed RUNX1m.

Methods: Pts with ND AML were identified from 4 academic centers (MSKCC, DFCI, Yale, NCI in Lithuania). Pts were treated with IC (either 7+3 or liposomal cytarabine/daunorubicin [CPX-351]) or HMA+Ven between 2013 and 2024, and RUNX1m profiling was performed at participating sites by next-generation sequencing (NGS). Composite complete response (cCR) was defined as complete response (CR) + CR with incomplete count recovery (CRi) per ELN 2022. Overall survival (OS) was compared between groups by log-rank test. Follow-up with formal response assessment was required for inclusion.

Results: We reviewed 1218 ND AML pts, including 220 RUNX1m and 998 RUNX1 wildtype (RUNX1wt). Compared to RUNX1wt, ps with RUNX1m were older (median 67 years (yrs), range: 19-92 yrs vs. 65 yrs, range: 18-93 yrs, p<0.001) and more likely to be male (67% vs. 54%, p <0.001), have previous myelodysplastic syndrome (MDS) or myeloproliferative neoplasms (MPN; 45% vs. 23%, p<0.001), and prior HMA exposure for antecedent MDS or MPN (18% vs. 9.3%, p<0.001).

RUNX1m were highly associated with secondary ontogeny mutations compared to RUNX1wt (69% vs 23%, p<0.001). By contrast, RUNX1m were less frequently associated with mutations in NPM1 (0.9% vs. 22%, p <0.001) or TP53 (3.6% vs. 19%, p<0.001). In the RUNX1m group, 126 pts (57%) received IC vs. 94 (43%) who received HMA+Ven. Allogeneic stem cell transplant (allo-SCT) was pursued in 89 pts (40%) with RUNX1m compared with 437 pts (44%) with RUNX1wt (p=0.41).

In the entire cohort, there was no difference in median OS (mOS) between pts with RUNX1m vs. RUNX1wt (17 months [mo] [95% CI 15-21 mo] vs. 18 mo [95% CI 16-22 mo]; p=0.31). In the RUNX1m group, excluding the 8 pts (3.6%) with concomitant TP53 mutations, RUNX1m pts with additional secondary ontogeny mutations (n=152, 69%) had significantly shorter mOS compared to RUNX1m pts without secondary ontogeny mutations (n=60, 27%; 15 mo [95% CI 13-20 mo] vs. 30 mo [95% CI 19mo-NR]; p=0.007). Moreover, when we analyzed RUNX1m pts without any additional favorable or adverse risk features per ELN 2022 (n=28, 13%), we found that their 24-month OS was comparable to RUNX1wt pts who were intermediate risk by ELN 2022 (n=196, 20%; 56% [95% CI 39-82%] vs. 58% [95% CI 51-66%]; p=0.489).

In unadjusted analysis for all RUNX1m pts, mOS was longer with IC compared to HMA+Ven (27 mo [95% CI 17-51 mo] vs. 15 mo [95% CI 9.2-18 mo]; p<0.001). However, when we evaluated RUNX1m pts aged 60-75 yrs, there was no difference in mOS based on treatment strategy (IC: 15 mo [95% CI 11-32 mo] vs. HMA+Ven 13 mo [8.2-26 mo]; p=0.33). There was also no difference in OS for RUNX1m pts who underwent allo-SCT (24-month OS for IC: 73% [95% CI 63-85%] vs. HMA+Ven: 52% [95% CI 25-100%]; p=0.34). A multivariable analysis including co-mutations and treatment strategy will be presented at ASH.

Conclusion: Our data indicate that in pts with ND AML, RUNX1m itself does not confer worse prognosis, and its predictive value in optimal treatment is limited. Pts with adverse risk AML by ELN 2022 only by virtue of their RUNX1m had comparable 24-month OS to RUNX1wt pts with intermediate risk by ELN 2022. This suggests that RUNX1m alone should not be sufficient to define adverse risk in ND AML.