P1356: PATIENT-SPECIFIC DIGITAL DROPLET PCR ASSAYS FOR MINIMAL RESIDUAL DISEASE DETECTION OF RUNX1 MUTATIONS IN ACUTE MYELOID LEUKEMIA PATIENTS UNDERGOING ALLOGENEIC STEM CELL TRANSPLANTATION

Abstract

Background: The persistence of non-clonal-hematopoiesis mutations (mut) in remission prior to a consolidating allogeneic stem cell transplantation (HSCT) in acute myeloid leukemia (AML) patients (pts) has been associated with a higher relapse probability. RUNX1 mut are frequent in AML and associate with adverse outcomes, often indicating the need for an allogeneic HSCT. They may also represent suitable targets for measurable residual disease (MRD) detection, but have not yet been evaluated for this purpose separately. Aims: To develop patient-specific digital droplet PCR (ddPCR) assays for sensitive RUNX1 mut detection and evaluate their applicability for predicting relapse in AML pts undergoing allogeneic HSCT. Methods: We retrospectively analyzed 171 AML pts consolidated by allogeneic HSCT (non-myeloablative 64%, 24% reduced-intensity, or myeloablative 12%) for their diagnostic RUNX1 mut status. At diagnosis, cytogenetics, and additional 53 recurrently mutated genes in AML were analyzed. Adequate material at and/or following HSCT was available for 22 RUNX1mut pts (up to 28 days prior to HSCT [n=19], during post-HSCT follow up [61 samples in 11 pts] and at relapse [n=3]). For these, patient-specific ddPCR assays were developed using a competitive probe approach to reach MRD levels of at least 0.01% VAF. Median follow up after HSCT was 1.5 years. Results: At diagnosis, 22 pts had one and 4 pts had two RUNX1 mut (16%, median VAF 36.5, range 3.8-95.1%), 10 were frameshift, 12 were missense, and 8 were nonsense. Most frequent co-mut were found in ASXL1 (31%), DNMT3A (29%), and NRAS (26%). One patient was transplanted with active AML and had a RUNX1 VAF of 30.4% at HSCT. For all 19 measurable pts in morphologic remission at HSCT, RUNX1 mut had decreased compared to VAFs at diagnosis (median VAF at HSCT 2.3, range 0.06-10.9%), but persisted in 11 (58%) of the pts. Noteworthy, 7/8 RUNX1 MRDneg, but only 2/11 RUNX1 MRDpos pts were female (P=.005). There were no other characteristics or mutated genes at diagnosis that associated with RUNX1 MRDpos at HSCT (Figure 1A). RUNX1 MRDpos pts at HSCT had a higher cumulative incidence of relapse (P=.05) and by trend shorter overall survival (P=.07, Figure 1B,C). At day 28 after HSCT, all 11 pts with material available were MRDneg, 3 of them relapsed. In 2 of those pts, RUNX1 MRD predicted relapse at VAFs of 0.09% and 0.04% 10 and 5.5 months earlier, respectively, while in the third patient the last MRD sample taken 4 months prior to relapse remained RUNX1 MRDneg. All 3 pts with relapse material available relapsed with the RUNX1 mut present at diagnosis (at VAFs of 37%, 8.7%, and 36.2%). All 8 pts with material available that remained in remission had no detectable RUNX1 mut at any time during post-HSCT follow up. Image:Summary/Conclusion:RUNX1-mut based patient-specific ddPCR assays could be developed for all RUNX1 mut AML pts. MRD determination by these assays is a feasible tool to determine AML MRD with a relatively short turn-around time during disease course at high sensitivity and relatively low cost. RUNX1 MRDpos at HSCT associated with high relapse risk. Within 28 days after HSCT, all pts became RUNX1 MRDneg, identifying this timepoint as least meaningful for MRD assessment in our cohort. In sequential analyses after HSCT RUNX1-mut based MRD was a stable marker allowing for a reliable and clinically meaningful MRD detection, as they predicted relapse when rising as well as persisting remissions when remaining undetectable in the majority of analyzed AML pts.