Dynamics of Oligoclonal Hematopoiesis in Severe Aplastic Anemia Patients Undergoing Immunosuppressive Treatment: Longitudinal Somatic Mutation Analysis from the EBMT Race Trial

Background

The EBMT RACE trial (NCT02099747) has recently shown the superiority of Eltrombopag given with horse-ATG and cyclosporine (standard IST) (Arm B) in comparison to standard IST (Arm A) alone. Clonal haemopoiesis (CH) has been documented at diagnosis and after immunosuppressive therapy (IST) in idiopathic aplastic anaemia (AA); CH and its dynamics was systematically investigated in the RACE trial to ascertain its clinical significance in the biology of AA.

Methods

The somatic mutational profiles were assessed at diagnosis (Timepoint 0 - TP0) and sequentially, 6 and 24 months (TP6 and TP24) following treatment using a 31-gene ‘core’ (Haloplex HS, TP0, TP6, and TP24 with 170, 151, and 107 patients, respectively) and 291-gene ‘extended’ (Sureselect XT, TP0 with 140 and TP6 with 136 patients) custom targeted panel, with unique molecular identifiers (UMI) allowing identification of low-level clones ( VAF > 0.2 %). 141 patients were analysed at both TP0 and TP6, whilst 89 analysed at all three time points. The repertoire of somatic variants was correlated with treatment arms, age, severity of disease, response to treatment, and other patient-specific variables.

Results

At baseline, 6 and 24 months using the ‘core panel’, 83 mutations were detected in 59 (59/170, 34.7%), 193 mutations in 86 (86/151, 56.9%) and 265 mutations in 84 (84/107, 78.5 %) patients respectively, with significant increase over time (TP0 vs TP6 p= <0.00001, TP0 vs TP24 p= <0.00001, TP6 vs TP24 p= <0.00001). In patients with mutations, the mean mutation rates were 1.4 (TP0), 2.24 (TP6) and 3.15 (TP24). There is also a trend of median VAF increase from TP0 (8.44%) to TP6 (13.16%), with no significant difference in treatment arms. Mutational frequency at TP0 did not differ between treatment arms, and baseline mutations did not correlate with treatment response.

PIGA (TP0 n=23, TP6 n=44, TP24 n=30), DNMT3A (TP0 n=18, TP6 n=33, TP24 n=25), BCOR/BCORL1 (TP0 n=12, TP6 n=35, TP24 n=24), TET2 (TP0 n=10, TP6 n=10, TP24 n=9), ASXL1 (TP0 n=2, TP6 n=22, TP24 n=23), CSMD1(TP0 n=4, TP6 n=15, TP24 n=54) were the most frequently mutated genes. The median age of patients with PIGA, BCOR, and BCORL1 mutations was lower than that of patients with other mutations ((TP0 44 vs 60 years, TP6 49 vs 58 years and TP24 49 vs 56 years). DNMT3A and TET2 mutations occurred more frequently in older patients (median 58 and 61 years at 6 months, 60.5 and 62.5 years at 24 months). Gene interactions, either co-occurrence or exclusivity patterns, were variable across the different time points, with no specific patterns observed. New ASXL1, BCOR and PIGA mutations tend to emerge in between TP0 and TP6, whilst new RAD21, ETV6, CSMD1 and STAG2 mutations mostly appeared between TP6 and TP24. The number of PIGA mutations increased over time with a significant change in Arm A at TP6 (p=0.008) and TP24 (p=0.001) compared to TP0, which also correlated with higher incidence of haemolytic PNH in standard IST arm.

The ‘extended’ gene panel of 291 genes, detected pathogeneic somatic variants in 132 out of 140 patients (94%) at baseline and in all patients at TP6 (100%, 136/136); only 4.3% and 3.8% of the detectable mutations had VAF>2%. The low-level clones (<1% VAF), were also validated by ddPCR (n=3). Mutations (n=36) undetectable by ‘core’ panel at baseline were present at low level (<1%) using the more sensitive ‘extended’ XT panel. The larger panel also uncovered mutations in ATM, NCOR2 and epigenetic pathways regulators.

Conclusions

We observed a high prevalence of CH (94%) in AA patients at diagnosis, which tends to increase after therapy, irrespective of treatment arm, in the absence of progression to myeloid malignancies. These findings suggest CH in AA arises in a stochastic manner before and after immunosuppressive treatment and increase in mutations may represent a bottle neck effect in terms of residual hematopoietic stem cells, but longer-term clinical data is mandatory to ascertain the impact of mutations on malignant transformation