Paper Details

ABSTRACT

Introduction: Measurable residual disease (MRD) testing is increasingly important in the care of patients with cancer, exemplified first for hematological malignancies. For patients with acute myeloid leukemia (AML), multiparametric flow cytometry (MFC) is the most broadly applicable MRD method but has suboptimal sensitivity and standardization. Molecular MRD tests offer superior performance with fewer false negatives but are validated only in select AML subsets. We report here the prospective multicenter validation of whole genome sequencing (WGS)-informed AML MRD testing as a universal approach.

Methods: MEASURE (NCT05224661) was a prospective CIBMTR facilitated, NMDP-sponsored clinical protocol at 18 US cancer centers. Adults with AML in complete remission (CR) undergoing first allogeneic transplant (alloHCT) Sep2022-Dec2024 with stored diagnostic material (Dx) available were enrolled. WGS was performed on Dx (59-300X) and paired CR blood (24-89X) to identify MRD targets. Patient-specific capture MRD panels (≤500 probes) were designed prioritizing coverage of known oncogenic (onc) small (SNV/InDel) or structural variants (SV) plus a common region (FLT3, NPM1, KIT, KRAS, NRAS, PTPN11). Personalized MRD (pMRD) testing, via targeted error-corrected consensus clustered (single-strand 12nt UMI) sequencing, was performed on at least 2 replicates of up to 500ng DNA from CR blood collected at a median of 14 (IQR 8-24) days pre-alloHCT. Detection of any pMRD target (excluding DNMT3A, TET2, ASXL1, IDH1 [DTAI]) was considered positive (pos) and assessed blind to outcomes. The statistical analysis plan was pre-registered with primary endpoint cumulative incidence of relapse (CIR) with non-relapse mortality as competing risk.

Results: 254 participants (pts) were included (median age 59yr, 19-79), 53% male, 69% non-Hispanic White. Median follow-up was 18mo (IQR 12-24). 1yr CIR was 14%.

Most pts (n=251, 99%) received MFC testing pre-alloHCT as standard of care, with 1yr CIR 40% in MFC pos (n=28) vs 11% in MFC negative (neg, n=223). Of 35 relapses occurring within 1yr, only 11 (31%) were pos by MFC pre-alloHCT.

In pts with an appropriate Dx sample (89%), a median of 325 (range 47-422) targets/pt was included for pMRD testing, with fewer targets (median 4, range 1-14) when only FFPE Dx was available (n=22, 11%). Ultra-deep pMRD sequencing (median 38,325 UMI depth, range 18,327-39,359) was performed on CR pre-alloHCT samples. Of onc targets (excluding DTAI), 252/687 (37%) SNV/InDel and 84/232 (36%) SV were pos. 8 onc non-Dx common region variants were detected. Of presumed passenger targets, 75/128 (59%) SV were pos and 70/232 (30%) patients had SNV/InDel median VAF≥0.1%.

For the primary endpoint, pts testing pMRD neg (n=101, 40%) had significantly lower CIR than those testing pos (n=153, HR=7.5, 95% CI 2.8-20.6, p<0.001), meeting the pre-specified criterion. Most relapses (90%) occurred in the pMRD pos group, with 6-month CIR 15% vs 0%. The pMRD neg group had significantly higher overall survival (OS, HR=4.3, 95% CI 1.9-9.7, p<0.001, 18mo 94% vs 75%) and relapse-free survival (RFS, HR=4.3, 95% CI 2.1-8.7, p<0.001, 18mo 91% vs 68%) compared to the pos group.

The only relapse in the ELN 2022 Dx favorable group (n=41) was pMRD pos. Among intermediate/adverse risk pts, pre-HCT pMRD status stratified outcomes, with pos (n=132) having higher relapse risk than neg (n=81) pts (HR=6.9, 95% CI 2.5-18.9, p<0.001; 1yr CIR 25% vs 1%).

Adjusting for baseline genetics, multivariable analysis confirmed pMRD status as an independent predictor of relapse (HR=6.1, p<0.001), even after incorporating MFC status (HR=5.2, p=0.002).

Exploratory analyses assessed the value of MRD target subsets. After excluding pts with a target covered by an ELN-recommended test (n=88, FLT3-ITD, NPM1, CBF fusions), pMRD stayed prognostic in the remaining pts (n=166, CIR HR=6.3 p=0.002, OS HR=3.3 p=0.01, RFS HR=3.8 p=0.002).

All 254 pts had ≥1 onc target (median 3/patient, range 1-10) and restricting to onc targets yielded consistent findings, with the pMRD neg group (n=110) having reduced CIR (HR: 5.5, p<0.001) and improved OS (HR: 4.3, p<0.001) relative to the pMRD pos group (n=144). No relapses were observed within 6 months of alloHCT in the pMRD neg group.

Conclusions: The MEASURE study is the first prospective validation of personalized WGS-informed testing as a universal and high-performance AML MRD approach. Post-alloHCT testing will also be presented at the meeting.