Paper Details

ABSTRACT

Introduction:

KMT2A alterations are common in acute myeloid leukemia (AML) and define biologically distinct subsets with either rearrangements (KMT2Ar) or partial tandem duplications (KMT2A-PTD). Pre-analytical factors and treatment may impact KMT2A transcript levels, limiting reliable measurable residual disease (MRD) quantification. Given the relevance of these targets in drug development and clinical management, as part of the FNIH Biomarkers Consortium for MRD in AML, we developed a tumor-informed approach to design personalized DNA-based six-channel multiplex droplet digital PCR (ddPCR) assays that detect and track patient-specific KMT2A alteration alongside up to four co-occurring leukemia-associated variants. We then applied this method to samples collected on a prospective clinical protocol, MEASURE (NCT05224661), conducted at 18 major US cancer centers for adults with AML in complete remission (CR) undergoing first allogeneic transplant (HCT).

Methods:

We developed and analytically validated six-channel multiplex ddPCR assays in KMT2Ar AML cell lines and patient samples, establishing limits of blank (LOB) and detection (LOD) per CLSI EP17-A2 guidelines. Structural variant (SV) assays used breakpoint-flanking primer/probe strategies; small variants (SNV/indel) used dark-probe designs.

Whole genome sequencing of diagnostic material (Dx, 59-300X) and paired CR blood (24-89X) identified MEASURE participants with KMT2Ar or KMT2A-PTD. Beyond KMT2A, up to 4 targets were selected per patient, prioritizing oncogenic over passenger variants and SV/InDel over SNV; the sixth channel was reserved for the reference AGO1. To maximize genome sampling, 1ug of genomic DNA from CR blood collected 5-79 days before HCT was distributed across five 200ng wells for MRD testing. Each patient-specific ddPCR test included positive controls (two Dx dilutions 0.1%, 0.05%), one healthy donor (HD) negative control and no template control.

Oncogenic variants were positive (pos) when supported by ≥3 pos droplets above the assay-specific LOB; passenger variants were pos only when signals exceeded the lowest Dx dilution. Surrogate VAF was the ratio of target to reference concentrations. For relapsed but MRD-neg patients, longitudinal samples were analyzed when available.

Results:

After initial development using 7 KMT2Ar AML cell lines, validation was expanded to serial dilutions of 5 AML patient samples using optimized probe design and expanded dark probe use. LOB was 0; median LODs were 0.03% for SVs (range 0.02-0.05%) and 0.06% for small variants (range 0.01-0.09%).

Personalized ddPCR assays were successfully designed for all 27 patients with KMT2Ar and 21 of 34 patients with KMT2A-PTD (limited by repetitive region complexity). Targets per patient ranged 3-5 (median 5); 57 (25%) SVs, 168 (75%) small variants. All Dx dilutions were pos with VAF ranging 0.002%-2.5% (median 0.14%); all HD controls were neg.

The 48 patients tested had median age 56yr (19-77yr); median follow-up time post-HCT was 18mo (11-36mo). 9 relapsed post-HCT. Pre-HCT multiparameter flow cytometry (MFC) as clinical standard of care for MRD testing was available for 47 (98%) patients.

Of 27 patients with KMT2Ar, 4 relapsed (15%); pre-HCT MFC was neg in all. Pre-HCT ddPCR for KMT2Ar was pos for 9 patients (33%) and identified all 3 who relapsed within 1yr post-HCT with sensitivity (sens) 100%, specificity (spec) 75%, positive predictive value (PPV) 33%, negative predictive value (NPV) 100% for 1yr relapse. A fourth patient relapsed at 13mo; serial post-HCT ddPCR detected conversion to pos 5 weeks prior to relapse.

Of 21 patients with KMT2A-PTD, 5 relapsed (24%), all within 1yr; pre-HCT MFC was pos in 2 patients and they both relapsed. Pre-HCT ddPCR for KMT2A-PTD was pos for 17 patients (81%) and identified 4/5 relapsed patients with sens 80%, spec 19%, PPV 24%, NPV 75%. Among the 13 KMT2A-PTD MRD-pos patients who did not relapse, all had at least one co-detected AML-associated variant, suggesting true clonal persistence rather than technical artifact.

Conclusions:

Detection of KMT2Ar persistence in pre-HCT remission blood by personalized DNA-based multiplex ddPCR was associated with post-HCT relapse. KMT2A-PTD, while more commonly detected in pre-HCT CR blood, may represent a lower value AML MRD marker. Co-detected oncogenic/passenger variants and serial post-HCT monitoring data will be presented at the meeting.