PPM1D and DNMT3A Mutations in Myelodysplasia and Clonal Hematopoiesis

Introduction: Age-related clonal hematopoiesis (CH) has been implicated in an increased risk of myeloid neoplasms. While common driver genes mutated in CH largely overlap to those in myeloid neoplasms, a notable exception is protein phosphatase Mg²⁺/Mn²⁺dependent 1D gene (PPM1D), encoding a p53-targeting phosphatase. Although it is known to be involved in DNA damage response pathways and more frequently mutated in therapy-related myeloid neoplasms than in primary ones, its role in CH and myeloid neoplasms has not been fully understood.

Aim: To identify genetic features associated with PPM1D mutations, we examined genetic profiles in the large cohorts of healthy elderly individuals and patients with myelodysplasia.

Methods: We enrolled 10,826 healthy individuals (>60y) and 1,213 cases with myelodysplasia, including myelodysplastic syndromes (MDSs), myelodysplastic/myeloproliferative neoplasms (MDS/MPNs) (n=1,080), and secondary acute myeloid leukemia (sAML) (n=133), of which 567 cases were treated by hematopoietic stem cell transplantation (HSCT) through the Japan Marrow Donor Program just after sampling, and 332 of them underwent any therapy before sampling. Samples from healthy individuals were subjected to multiplex-amplicon sequencing for 22 genes, including PPM1D and other genes, related to CH or myeloid neoplasms. Myelodysplasia samples had previously been sequenced for major myeloid drivers, except for PPM1D, which was newly sequenced in this study.

Results: Frequency of PPM1D mutations in myelodysplasia and healthy individuals was 3.1% and 0.42%, with a median variant allele frequency (VAF) of 0.043 and 0.056, respectively. PPM1D mutations were more frequent in cases with previous treatment (4.8%) than in those without known history of therapy (2.3%) (P=0.038). In MDS and MDS/MPN cases, 59.5% of PPM1D mutations had accompanying mutations, in which DNMT3A mutations were the most frequently identified (16.2%, n=6). These 6 cases were diagnosed with RCUD (n=1), RCMD (n=2), RAEB-2 (n=2), or CMML (n=1). The association between PPM1D and DNMT3A mutations was also seen in 7 of 45 healthy individuals with PPM1D mutations, of which one had a DNMT3A-R882 mutation. In the HSCT cohort, 192 cases harbored ≥2 mutations of the 22 CH-related genes, and the relative temporal order of these mutations was investigated using Bradley-Terry model relying on their tumor cell fractions. The estimate of PPM1D mutations tended to be smaller than that of DNMT3A mutations. To further confirm chronological order of these mutations, VAF values were compared between them in the individuals with concurrent PPM1D and DNMT3A mutations (n=13; 6 myeloid neoplasms and 7 healthy donors). In the combined cohort, the VAFs of PPM1D and DNMT3A mutations were correlated (Spearman; correlation coefficient=0.87, P=1.2x10e-5). In both neoplastic and healthy cohort, the VAFs of DNMT3A-R882 mutations were larger than those of accompanying PPM1D mutations. These findings suggest that these mutations should be acquired in the same cell populations and that DNMT3A mutations might occur prior to PPM1D mutations. With regard to copy number alterations associated with PPM1D-mutated myelodysplasia, del(5q) (16.7%) and complex(-like) karyotypes (13.9%) were among the most frequent chromosomal abnormalities. Approximately 65% of PPM1D-mutated tumor samples had normal karyotype, which was similar to PPM1D-unmutated cases. PPM1D mutations did not significantly influence overall survival, although PPM1D mutations tended to negatively affect clinical outcome among patients who were treated with HSCT (Hazard ratio, 1.61; 95% confidence interval, 0.95 to 2.70).

Conclusion:PPM1D mutations were more enriched in myelodysplasia than in CH, and the median value of VAF in PPM1D mutations in CH was not significantly different from that in myelodysplasia. The size of PPM1D-mutated clones tended to be relatively smaller compared with that of clones with other mutations in myelodysplasia. PPM1D and DNMT3A mutations might be cooperatively associated in the pathogenesis of myelodysplasia and CH.