The molecular profile of clonal hematopoiesis in patients (pts) with idiopathic aplastic anaemia (iAA) has been shown to predict response to immunosuppressive therapy (IST). Moreover, clonal evolution is associated with transformation of iAA to aggressive myeloid malignancies such as myelodysplastic syndrome (MDS) and acute myeloid leukaemia (AML). Detection of somatic mutations may help identify pts more likely to respond to IST and also those at increased risk of transformation however the predictive value of detecting mutations at a single timepoint is limited. Longitudinal detection of changes in mutation profile over time may improve prediction of transformation risk, however due to bone marrow hypocellularity DNA quality and quantity from pts with iAA is frequently suboptimal. Serial mutation analysis of cell-free DNA (cfDNA) is one approach that may overcome this problem as the majority of cfDNA originates from hematopoietic cells and mutations detected in this compartment have been shown to closely resemble those detected from peripheral blood (PB) / bone marrow (BM) analysis in other haematological malignancies. We aimed to characterise longitudinal genomic changes in pts with iAA and contrast mutation detection in the cellular compartment with those detected in cfDNA.
Fifteen pts (10 female and 5 male) with a diagnosis of iAA were included (median age 35 years (range 17 - 78)); eleven were newly diagnosed (within six months of diagnosis) and four were two or more years post diagnosis. Cellular DNA (PB or BM aspirate) and cfDNA were collected upon study enrolment and cfDNA samples were collected longitudinally for each pt. Somatic mutations in cellular samples were assessed using custom targeted next generation sequencing panels with an analytical sensitivity of ~1%. cfDNA samples were assessed using a custom anchored multiplex PCR panel with molecular barcodes with an analytical sensitivity of ~0.5%. Twenty-nine genes were assessed in both cellular and cfDNA.
Somatic mutations were detected in 13 of 15 (87%) pts at the first sampled timepoint, with two or more mutations detected in seven pts. Overall 27 mutations were detected in seven genes including PIGA (n=12), DNMT3A (n=5), ASXL1 (n=5), BCOR (n=2), BCORL1, NPM1 and RUNX1 (n=1 each) with a higher proportion detected in cfDNA (25 of 27) compared to cellular DNA (22 of 27). Mutations unique to cfDNA were in PIGA (n=3), ASXL1 (n=1) and DNMT3A (n=1) and those unique to cellular DNA were in NPM1 (n=1) and PIGA (n=1). The variant allele frequency (VAF) was <10% for 26 of 27 mutations (median cellular and cfDNA VAF 1.78% and 1.19% respectively), highlighting the need for sensitive mutation detection approaches in the setting of iAA regardless of the compartment tested.
We next assessed the mutation dynamics in cfDNA over time by analysing a median of 3 (range 2 - 4) sequential samples per pt (median follow-up interval 670 days, range 249 - 923). 20 of 25 cfDNA mutations were detected in longitudinal cfDNA at similar VAF, including PIGA (n=9), BCOR (n=2), DNMT3A (n=5), ASXL1 (n=3) and RUNX1 (n=1). Five mutations in PIGA (n=2), ASXL1 (n=2) and BCORL1 (n=1) became undetectable over time. Importantly, acquisition of new mutations during follow-up was observed in four pts (total of 13 new cfDNA mutations detected). Three out of these four pts experienced disease transformation to hematological malignancy (MDS n=2, AML n=1), with acquired mutations in SETBP1, NPM1, NRAS, KRAS, WT1 and KIT. Importantly, these mutations often went from undetectable to relatively high VAF within a short interval (5 - 13 months) which is important to consider when designing molecular screening schedules to potentially pre-empt transformation.
In summary, our data shows that cfDNA mutation profiling in pts with iAA is technically feasible and is able to detect additional mutations not detectable in the cellular compartment. Importantly, we demonstrate the limitations of analysis of both compartments in the context of low VAF mutations, suggesting that concurrent testing of both cellular and cfDNA may increase the sensitivity for detecting low frequency events. Our data also suggests that longitudinal changes in mutation profile may improve identification of pts at high risk of transformation to hematological malignancy. The optimal frequency of monitoring and the magnitude of increased risk requires validation in prospective trials.
GLR and LCF contributed equally to this work
Szer:Alexion Pharmaceuticals Inc.: Consultancy, Honoraria, Membership on an entity's Board of Directors or advisory committees, Speakers Bureau; Apellis: Consultancy; Pfizer: Honoraria, Speakers Bureau; Takeda: Honoraria, Speakers Bureau; Novartis: Consultancy, Honoraria, Speakers Bureau; Prevail Therapeutics: Honoraria, Membership on an entity's Board of Directors or advisory committees. Bajel:Astellas: Honoraria; Novartis: Honoraria; Amgen: Honoraria, Speakers Bureau; Pfizer: Honoraria; Abbvie: Honoraria. Blombery:Novartis: Consultancy; Invivoscribe: Honoraria; Amgen: Consultancy; Janssen: Honoraria.
Asterisk with author names denotes non-ASH members.