PF212 NEXT‐GENERATION SEQUENCING OF CIRCULATING CELL‐FREE DNA PROVIDES COMPLEMENTARY INFORMATION TO GENOMIC PROFILING OF BONE MARROW IN PATIENTS WITH ACUTE LEUKEMIAS

Abstract

Background:Circulating cell‐free DNA (ccfDNA) allows for non‐invasive peripheral blood sampling of cancer‐associated mutations and has established clinical utility in several solid tumors. The ability of ccfDNA to detect and track leukemia‐associated mutations may also be useful in leukemias, both by identifying prognostic or targetable mutations at diagnosis or as a non‐invasive marker of measurable residual disease (MRD).Aims:To evaluate the feasibility of genomic analysis of ccfDNA to detect somatic mutations in patients (pts) with acute myeloid leukemia (AML) or acute lymphoblastic leukemia (ALL).Methods:Plasma ccfDNA was collected and analyzed using a capture‐based next‐generation sequencing (NGS) assay (Qia275) targeting 275 genes. NGS was performed on DNA extracted from the bone marrow (BM) in a CLIA‐certified molecular diagnostics laboratory for the detection of mutations in the coding sequence of 28 genes, all of which were also assessed in the ccfDNA panel. An established bioinformatics pipeline was used to identify somatic variants.Results:A total of 24 pts (AML, n = 22; ALL, n = 2) underwent paired ccfDNA and BM sequencing at diagnosis. ccfDNA was also collected in 11 pts at ≥1 time point during remission. Among baseline samples, ccfDNA was collected a median of 6 days after BM collection (range, 0 to 27 days) and 0.5 days after the start of induction (range, −7 to 7 days). Using the 28 overlapping genes between the two assays, 55 somatic mutations were detected. 12 mutations (22%) were detected in ccfDNA but not BM. The median number of mutations detected in the BM and in ccfDNA per pt were 1.5 (range, 0–5) and 1 (range, 0–4), respectively (P = 0.31). Time from the start of chemotherapy until ccfDNA collection did not impact the concordance with BM mutation analysis. Among mutations detected in baseline ccfDNA samples, the median variant allelic frequency (VAF) was 33.7% (range, 2.7% to 90.8%). ccfDNA VAF was higher in mutations also detected in BM compared to those detected only in ccfDNA (median 38.3% vs 25.4%; P = 0.01), suggesting that these may be small subclonal populations more likely to be missed with BM sampling. Conversely, BM VAF was higher in those also detected in ccfDNA compared to those only found in BM (39.7% vs 21.0%; P < 0.05). Mutations detected with ccfDNA but not BM included FLT3‐ITD in 4 pts, ASXL1 in 2 pts, WT1 in 1 pt, IDH1 in 1 pt, and BRAF and two EGFR mutations in 1 pt. In the 4 pts with FLT3‐ITD, the mutation was detected by PCR of the BM but not by BM sequencing. Among the other 5 pts with mutations detected in ccfDNA but not BM, 2 pts eventually relapsed; in both pts, the discordant mutation (IDH1 and ASXL1) was detected in the relapse BM. Using ccfDNA, the persistence of leukemia‐associated mutations during remission or the emergence of new mutations appeared to herald relapse. Two pts with t(8;21) AML developed new RUNX1 mutations detected by ccfDNA while in remission and relapsed 3 months and 14 months later; in both pts the emergent RUNX1 mutation was also detected in the relapse BM. Another pt with AML in morphological remission on day +30 after stem cell transplant had persistent TP53 and TET2 mutations detected by ccfDNA and relapsed 1 month later.Summary/Conclusion:This study demonstrates the feasibility of ccfDNA monitoring in pts with acute leukemias. ccfDNA can identify clinically significant prognostic or targetable mutations not detected by BM sequencing. The potential use of ccfDNA as a non‐invasive method of MRD assessment needs to be evaluated in larger, prospective cohorts.