Key Points
-
•
PPM1D mutations dominate t-CH/t-CCUS, show the shortest latency after genotoxic therapy, and are enriched in this subgroup.
-
•
Although the median PPM1D VAF was relatively lower, a higher VAF (≥13%) was linked to inferior overall survival.
Visual Abstract
Abstract
TP53 and PPM1D are key regulators of DNA damage response and repair, and somatic mutations in these genes often co-occur in hematopoietic cells, expanding under genotoxic stress. Unlike TP53 mutations, where mechanisms of progression are defined, pathways underlying clonal fitness and transformation in PPM1D mutant cells remain unclear. In collaboration with 5 academic institutions, we analyzed the clinical and molecular landscape of 337 patients with clonal hematopoiesis (CH) and clonal cytopenia of undetermined significance (CCUS) across 4 genotypes: PPM1Dmt/TP53wt (n = 170 [50%]), PPM1Dmt/TP53mt (n = 25 [7%]), TP53mt/PPM1Dwt (n = 17 [5%]), and TP53wt/PPM1Dwt (n = 125 [38%]). All PPM1D variants were truncating, located in exon 6 of the gene, with a median variant allele frequency (VAF) of 6% (range, 0.3%-64%). The PPM1Dmt/TP53mt genotype was most frequently encountered in therapy-related CH/CCUS (t-CH/t-CCUS; 80%, 66.5%, 76.5%, and 19%; P ≤ .001) and had a shorter time interval to detection from last genotoxic exposure (6.2, 5.9, 11.25, and 24.5 months; P ≤ .001) compared with PPM1Dmt/TP53wt, TP53mt/PPM1Dwt, and TP53wt/PPM1Dwt genotypes, respectively. Acknowledging the short follow-up duration, rates of malignant transformation were lower in the PPM1Dmt/TP53wt (2%) and PPM1Dmt/TP53mt (4%) groups compared with PPM1Dwt/TP53wt (12%) and PPM1Dwt/TP53mt (17%) groups (P ≤ .001), respectively. In summary, PPM1D mutations are frequently observed in t-CH/t-CCUS, with low median VAFs, and are associated with low rates of progression, even when comutated with TP53.
Introduction
Recent studies examining somatic mutations in blood cells of individuals exposed to cytotoxic therapy have highlighted the selective expansion of hematopoietic stem cells harboring alterations in key DNA damage response (DDR) genes. Among these, mutations in PPM1D, TP53, ATM, CHEK2, and SRCAP have emerged as the most prevalent drivers of therapy-related clonal hematopoiesis (t-CH).1, 2, 3, 4, 5 Of these, PPM1D and TP53 are most frequently observed,3,6,7 underscoring their central role in mediating the hematopoietic stem cell response to genotoxic stress.
PPM1D mutations truncate the C-terminal region of the protein, removing proteasomal degradation signals, resulting in elevated intracellular levels of a catalytically active PPM1D. PPM1D, a serine/threonine phosphatase that is transcriptionally activated by p53, functions as a negative regulator of both DDR and p53 signaling pathways by dephosphorylating numerous substrates upstream and downstream of p53, including p53 itself.6 The selective advantage conferred by PPM1D gain-of-function mutations is thought to reflect a partial attenuation of TP53 activity. This concept is supported by murine models in which overexpression of PPM1D produced a tumor spectrum that closely resembled that observed in TP53 loss-of-function models.8 Despite these mechanistic insights, clinical studies have reported no significant difference in overall survival (OS) between patients with therapy-related acute myeloid leukemia (AML)/myelodysplastic syndrome who harbor PPM1D mutations without concurrent TP53 mutations.6,9 Significantly worse outcomes are instead observed in patients with TP53 mutant myeloid neoplasms (especially with multihit states).10 This apparent discrepancy underscores the complexity of clonal evolution under the selective pressure of chemotherapy, and uncovers several critical gaps in our understanding. Although PPM1D mutations are strongly enriched in therapy-exposed individuals, it remains uncertain whether they function as dominant clonal drivers, how their clonal architecture evolves across the spectrum of CH to myeloid neoplasms, and how they interact with other high-risk mutations such as TP53. Moreover, the role of specific therapy exposures such as poly (ADP-ribose) polymerase (PARP) inhibitors or radioligand therapies in shaping PPM1D clonal dynamics has not been fully explored.
Here, we study a large cohort of individuals with PPM1D mutations, including individuals with CH and clonal cytopenia of undetermined significance (CCUS). Our work offers a unique opportunity to characterize the phenotypic spectrum and clonal architecture of PPM1D-mt clones and to evaluate their biological consequences across the spectrum of precursor myeloid neoplasms.
Methods
We conducted an observational study in collaboration with 5 academic institutions. After approval from respective institutional review boards, CH databases were queried, and 337 patients with PPM1Dmt/TP53 wild-type (wt; n = 170 [50%]), PPM1Dmt/TP53mt (n = 25 [7%]), TP53mt/PPM1Dwt (n = 17 [5%]), and TP53wt/PPM1Dwt (n = 125 [38%]) genotypes were identified. The TP53wt/PPM1Dwt genotype group comprised other somatic driver mutations, excluding PPM1D and TP53. We analyzed baseline characteristics, karyotypic changes, somatic comutations, rates of progression to myeloid neoplasms, and survival outcomes for these 4 genotypic groups. We also explored the prevalence of t-CH with respect to different exposures to genotoxic therapies and their impact on survival.
We defined CCUS as persistent unexplained cytopenia(s) associated with known somatic pathogenic variants (variant allele frequency [VAF] ≥2%), with bone marrow dysplasia involving <10% of hematopoietic cell lineages and bone marrow blasts <5%.11 Clinical next-generation sequencing testing was performed on DNA extracted from fresh bone marrow aspirates (319/337 [95%]) or peripheral blood samples (18/337 [5%]), with the panels demonstrating accuracy rates of >99% and reproducibility of 100% for single-base substitutions and insertion/deletion events. Details on the next-generation sequencing panels used by the 5 participating academic institutions, the indications for testing, and the proportion of patients contributed by each institution are provided in the supplemental Material.
This study was conducted in accordance with the Declaration of Helsinki.
Clonal analysis
Based on established literature examining CH dynamics, we investigated the fitness distributions and temporal acquisition patterns of PPM1D and TP53mt clones. Critically, previous studies have demonstrated that there is no statistically significant difference in fitness distributions between PPM1D and TP53mt clones.12 Furthermore, recent analyses have confirmed that there is no significant difference in the distribution of time of mutation acquisition between these 2 driver mutations.13 Given this equivalence in fitness and timing, predictions of future clonal growth and hierarchies can only be made on the basis of the order of magnitude of the respective mutational VAF. For each participant, we analyzed the maximum observed VAF of mutations in PPM1D and TP53 genes. Of note, individuals with CH/CCUS and suspected or confirmed germ line predisposition (eg, DDX41) and inherited bone marrow failure syndrome–associated mutations, including those linked to telomere biology disorders (eg, TERT), were excluded, as they represent distinct biological entities with different clinical trajectories compared with age- and context-related CH.
To establish dominance of one mutation over the other, we calculated the order of magnitude difference in VAFs using log2 transformation: log2(PPM1D_VAF/TP53_VAF). We created dominance categories based on this log2 difference, where codominance was established when −1 < log2(PPM1D_VAF/TP53_VAF) < 1. PPM1D dominance was defined as log2 ratios >1, and TP53 dominance as log2 ratios <−1. Additionally, we annotated samples with single mutations as “PPM1Dmt/TP53wt,” “PPM1Dwt/TP53mt,” and double wt samples as “wt-wt.” This analytical approach results in a conical region of VAFs defining codominance when plotted on a scatterplot, as the absolute VAF thresholds for codominance scale proportionally with the magnitude of the mutations. Clinical categorization included CCUS, CH, t-CCUS, and t-CH classifications. The complete categorization for all participants can be found in supplemental Tables 1 and 2.
Statistical analysis
Continuous variables were summarized as medians (range), whereas categorical variables were reported as frequencies (percentage). Unadjusted comparisons of patient characteristics and outcomes among the PPM1Dmt/TP53wt, PPM1Dmt/TP53mt, TP53mt/PPM1Dwt, and TP53wt/PPM1Dwt variant groups were made using the Wilcoxon rank sum test (continuous variables) or Fisher exact test (categorical variables). The Kaplan-Meier method was used to estimate progression-free survival (PFS) and OS. The median PFS (mPFS) was calculated from the time of CH/CCUS diagnosis to the time at which progression to an overt myeloid neoplasm, or death from any cause, occurred, and median OS (mOS) was calculated from the time of diagnosis to last follow-up or death. We selected a parsimonious multivariable model to reduce overfitting and improve model stability. Therefore, only variables that meaningfully contributed to the outcome were retained, resulting in a more interpretable and generalizable model. All tests were 2-sided, with P value <.05 considered statistically significant. We analyzed survival outcomes based on PPM1Dmt VAF, comparing patients with VAF <13% vs VAF ≥13%. The VAF cutoff was chosen based on receiver operating characteristic analysis using the Youden index; PPM1D VAF ≥13% was predictive of survival. Comparative frequencies of clonal categories were analyzed across diagnostic subgroups (CCUS vs high-risk myeloid neoplasms), and statistical differences were assessed using χ2 testing. Comutational relationships and relative clonal dominance of PPM1D in the context of TP53 comutation were also systematically evaluated.
Results
Baseline characteristics of patients with CH/CCUS
The median age at diagnosis differed significantly across molecular subgroups. Patients in the PPM1Dwt/TP53wt group were significantly younger than the others (P = .004; Table 1). All PPM1D variants were protein truncating, located in exon 6 of the gene, with a median VAF of 6% (0.3%-64%; supplemental Figure 1). Consistent with the notion that DDR gene mutations are preferentially selected under the pressure of anticancer therapy,6 we observed a marked enrichment of t-CH/t-CCUS: PPM1Dmt/TP53mt (80%), followed by PPM1Dwt/TP53mt (76.5%), PPM1Dmt/TP53wt (66.5%), and PPM1Dwt/TP53wt (19%) groups (P ≤ .001). The median PPM1Dmt VAF was comparable between PPM1Dmt/TP53wt (6% [range, 0.3%-50%] and PPM1Dmt/TP53mt (6% [range, 1%-28%]) groups (P = .96), as was the proportion of individuals with ≥2 PPM1D mutations (19% vs 24%, P = .55). Similarly, the median TP53mt VAF did not differ significantly between PPM1Dmt/TP53mt (4% [range, 0.5%-33%]) and PPM1Dwt/TP53mt (9.5% [range, 5%-42%]) groups (P = .12), respectively. The most common somatic mutations excluding TP53 and PPM1D were TET2 (24%), splicing factor genes (21%), DNMT3A (19%), and ASXL1 (10%). Among them, splicing factor genes (P < .001), TET2 (P < .001), and ASXL1 (P = .03) mutations were significantly more frequent in the PPM1Dwt/TP53wt group (Figure 1; Table 1). We calculated the CH risk scores (CHRS) across 4 groups.14 A high-risk score was observed in 52% of patients with PPM1Dmt/TP53mt, followed by 39% in patients with PPM1Dwt/TP53wt, 36% in patients with PPM1Dmt/TP53wt, and 35% in patients with PPM1Dwt/TP53mt, respectively.
Table 1.
Baseline characteristics of patients with CH/CCUS (n = 337), median (range), N (%)
| Variable | PPM1Dmt/TP53wt (n = 170) | PPM1Dmt/TP53mt (n = 25) | TP53mt/PPM1Dwt (n = 17) | TP53wt/PPM1Dwt (n = 125) | P value |
|---|---|---|---|---|---|
| Age, y | 72 (31-94) | 71 (52-89) | 65 (52-75) | 69 (20-99) | .004 |
| Sex | |||||
| Male | 106 (62) | 16 (64) | 8 (47) | 83 (66) | |
| Female | 64 (38) | 9 (36) | 9 (53) | 42 (34) | .60 |
| Therapy-related | 113 (66.5) | 20 (80) | 13 (76.5) | 24 (19) | <.001 |
| CH | 43 (25) | 15 (60) | 6 (35) | 6 (5) | <.001 |
| CCUS | 127 (75) | 10 (40) | 11 (65) | 119 (95) | <.001 |
| WBC (×109/L) | 4.3 (0.6-67.2) | 5.2 (1.10-12.4) | 3.90 (1.6-12.4) | 3.5 (0.8-19) | .083 |
| ANC | 2.2 (0.23-8.85) | 4.02 (1.0-10.2) | 1.9 (0.30-9.6) | 1.76 (0.04-13.68) | .009 |
| Hemoglobin (g/dL) | 10.1 (5.6-15.7) | 10.9 (7.80-15.5) | 11.9 (7.2-14.5) | 11.2 (6.90-15.90) | .18 |
| Hemoglobin ≤8 (g/dL) | 14 (12) | 1 (5) | 2 (12) | 8 (6) | .37 |
| Platelet (×109/L) | 138 (4.2-534) | 177.5 (35-421) | 148 (26-407) | 122 (15-595) | <.001 |
| Platelet ≤50 (×109/L) | 16 (14) | 1 (5) | 2 (12) | 11 (8) | .56 |
| PPM1D VAF | 6.0 (0.3-50) | 6.0 (1-28) | – | – | .96 |
| No. of PPM1D mutations ≥2 | 32 (19) | 6 (24) | – | – | .55 |
| PPM1D VAF% ≥20 | 20 (12) | 1 (4) | – | – | .33 |
| Cytogenetic (CG) abnormalities (evaluable n = 252) | 16/104 (15) | 4/10 (40) | 1/16 (6) | 15/120 (12.5) | .25 |
| TP53 VAF% | – | 4 (0.5-33) | 9.5 (5-42) | – | .12 |
| Most common somatic mutation | |||||
| DNMT3A | 37 (22) | 4 (16) | 3 (18) | 19 (15) | .56 |
| TET2 | 20 (12) | 5 (20) | 2 (12) | 54 (43) | <.001 |
| ASXL1 | 12 (7) | 1 (4) | 1 (6) | 21 (17) | .038 |
| Splicing factor | 7 (4) | 4 (16) | 1 (6) | 58 (46) | <.001 |
| BCOR | 6 (3.5) | 1 (4) | 0 | 6 (5) | .95 |
| IDH1 | 4 (2) | 1 (4) | 1 (6) | 8 (6) | .24 |
| CHRS | .06 | ||||
| Low | 8 (5) | 4 (16) | 1 (6) | 16 (13) | |
| Intermediate | 89 (59) | 8 (32) | 10 (59) | 60 (48) | |
| High | 55 (36) | 13 (52) | 6 (35) | 49 (39) | |
| High-risk mutation as per CHRS | 12 (7) | 25 (100) | 17 (100) | 64 (51) | <.001 |
| Median follow-up from the time of diagnosis, mo | 11.21 (0.17-63.5) | 8.8 (0.5-41) | 23.6 (4-60.2) | 34.4 (0.6-152) | <.001 |
| Treatment given for CCUS | 6 (3.5) | 1 (4) | 0 | 2 (2) | .82 |
| Disease progression to MDS/CMML | 4 (2) | 1 (4) | 3 (18) | 15 (12) | <.001 |
| Leukemic transformation | 0 | 0 | 0 | 7 (6) | <.001 |
ANC, absolute neutrophil count; CMML, chronic myelomonocytic leukemia; MDS, myelodysplastic syndrome; WBC, white blood cells.
Figure 1.
The figure depicts the distribution of comutations among patients in 4 distinct clinical groups, including CH, CCUS, and therapy-related myeloid neoplasms.
The follow-up duration from diagnosis differed significantly for these 4 genetic subgroups: 11.2 months for PPM1Dmt/TP53wt, 8.8 months for PPM1Dmt/TP53mt, 23.6 months for PPM1Dwt/TP53mt, and 34.4 months for PPM1Dwt/TP53wt groups, respectively (P ≤ .001). Although a shorter follow-up duration, particularly in the PPM1Dmt groups, may have impacted results, the rates of transformation to myeloid neoplasms were significantly lower in the PPM1Dmt/TP53wt (2%) and PPM1Dmt/TP53mt (4%) groups, in comparison to the PPM1Dwt/TP53mt (18%) and PPM1Dwt/TP53wt (12%) groups (P ≤ .001), respectively. Similarly, AML transformation was only observed in the PPM1Dwt/TP53wt group (6%), acknowledging that this group also had the longest follow-up duration. We analyzed clonal evolution during disease transformation from CH/CCUS to myeloid neoplasms for the entire cohort (supplemental Table 1). Cytogenetic and molecular changes were observed in both the PPM1Dwt/TP53mt and PPM1Dwt/TP53wt groups. Although serial mutational data were not available for all patients with PPM1D mutations who progressed, among those with available data (patients 1 and 5; supplemental Table 1), the PPM1Dmt clone either regressed (VAF decreased from 27% to 23% in patient 1) or was undetectable at progression (patient 5). Moreover, among patients who progressed, the baseline variant persisted at progression, and in some patients an additional mutation emerged during disease evolution.
We then conducted a subgroup analysis in patients with t-CH/t-CCUS (n = 170). The median age in years was numerically higher in the PPM1Dmt/TP53wt group (72 [37-94]) followed by PPM1Dmt/TP53mt (70 [58-89]), PPM1Dwt/TP53mt (67 [59-75]), and PPM1Dwt/TP53wt (67.5 [52-83]) groups, respectively, without reaching statistical significance (P = .12; Table 2). The median time from last cytotoxic therapy to diagnosis of t-CH/t-CCUS was 5.93, 6.21, 11.25, and 24.5 months in PPM1Dmt/TP53wt, PPM1Dmt/TP53mt, PPM1Dwt/TP53mt, and PPM1Dwt/TP53wt groups, respectively (P = .21). A significantly higher proportion of patients in the PPM1Dmt/TP53wt (6%/26%) and PPM1Dmt/TP53mt (24%/24%) groups, in comparison to the PPM1Dwt/TP53mt (0%/0%) and PPM1Dwt/TP53wt (0%/3%) groups, received PARP inhibitors and radioisotope-based cytotoxic therapy, respectively (P = .01/P < .001).
Table 2.
Baseline characteristics of patients with t-CH/t-CCUS (n = 170), median (range), N (%)
| Variable | PPM1Dmt (n = 113) | PPM1Dmt + TP53mt (n = 20) | TP53mt/PPM1Dwt (n = 13) | TP53wt/PPM1Dwt (n = 24) | P value |
|---|---|---|---|---|---|
| Age, y | 72 (37-94) | 70 (58-89) | 67 (59-75) | 67.5 (52-83) | .12 |
| Sex | |||||
| Male | 67 (59) | 14 (70) | 7 (54) | 14 (58) | .79 |
| Female | 46 (41) | 6 (30) | 6 (46) | 10 (42) | |
| t-CH | 33 (29) | 12 (60) | 4 (31) | 4 (17) | .02 |
| t-CCUS | 80 (71) | 8 (40) | 9 (69) | 20 (83) | |
| Time from last cytotoxic Rx to diagnosis, mo | 5.93 (0-352) | 6.21 (0-325.9) | 11.25 (0-160.90) | 24.5 (0-447) | .21 |
| Previous cancer-directed therapy | |||||
| Platinum-based therapy | 17 (14) | 4 (15) | 2 (7) | 3 (11.5) | .53 |
| Alkylating agent | 61 (51) | 13 (62) | 9 (64) | 20 (59) | .70 |
| PARP inhibitor | 7 (6) | 5 (24) | 0 | 0 | .01 |
| Radioisotope-based cytotoxic therapy | 31 (26) | 5 (24) | 0 | 1 (3) | <.001 |
| Radiation therapy | 58 (48) | 14 (67) | 5 (36) | 19 (54) | .27 |
| Autologous stem cell transplant | 13 (11.5) | 3 (15) | 2 (15) | 5 (21) | .56 |
| CAR T-cell therapy | 11 (9) | 3 (14) | 2 (14) | 3 (9) | .68 |
| Nonmyeloid cancer status | |||||
| Active | 51 (45) | 14 (70) | 4 (31) | 7 (29) | |
| In remission/not active | 62 (55) | 6 (30) | 9 (69) | 17 (71) | .04 |
| WBC (109/L) | 4.3 (0.6-67.2) | 5.2 (1.10-12.4) | 3.65 (1.6-10.9) | 3.8 (0.8-13.4) | .11 |
| ANC | 2.2 (0.23-8.85) | 4.0 (2.4-10.2) | 2 (0.48-8.23) | 2.0 (0.12-9.2) | .01 |
| Hemoglobin (g/dL) | 10.1 (6.7-15.7) | 10.5 (7.8-12.6) | 11.9 (7.2-14.2) | 11.7 (8.1-15.7) | .65 |
| Hemoglobin ≤8 (g/dL) | 14 (12) | 1 (5) | 2 (15) | 8 (33) | .18 |
| Platelet (×109/L) | 139 (4.2-534) | 150 (35-320) | 117 (26-238) | 111.5 (34-392) | <.001 |
| Platelet ≤50 (×109/L) | 16 (14) | 1 (5) | 2 (15) | 11 (45) | .84 |
| I PPM1D VAF% | 6.0 (0.3-47) | 6.0 (1-28) | – | – | .92 |
| I PPM1D mutations ≥2 | 30 (26.5) | 6 (30) | – | – | .70 |
| I PPM1D VAF% ≥20 | 15 (14) | 1 (6) | – | – | .46 |
| Cytogenetic (CG) abnormalities (evaluable n = 111) | 8/68 (12) | 2/7 (29) | 0 | 5/23 (22) | .14 |
| TP53 VAF% | – | 4 (0.5-33) | 9.0 (6-41) | – | .17 |
| Most common somatic mutation | |||||
| DNMT3A | 31 (27) | 4 (20) | 2 (15) | 8 (33) | .62 |
| TET2 | 12 (11) | 4 (20) | 2 (15) | 10 (42) | .004 |
| ASXL1 | 8 (7) | 1 (5) | 1 (8) | 3 (12.5) | .72 |
| Splicing factor | 2 (2) | 4 (20) | 0 | 5 (21) | <.001 |
| Median follow-up from the time of diagnosis, mo | 7.8 (0.17-55.9) | 5.7 (0.5-33.5) | 21.4 (4-60.3) | 40.6 (6.1-78.9) | .001 |
| Disease progression to MDS/CMML | 1 (0.8) | 0 | 3 (23) | 3 (12.5) | |
| t-CH | 0 | – | 0 | 1 (25) | |
| t-CCUS | 1 (100) | – | 3 (100) | 2 (75) | <.001 |
| Leukemic transformation | 0 | 0 | 0 | 3 (12.5) | |
| t-CH | – | – | – | 0 | |
| t-CCUS | – | – | – | 3 (100) | .009 |
CAR, chimeric antigen receptor; Rx, medical prescription.
The median PPM1Dmt VAF (6% [0.3%-47%] vs 6.0% [range, 1%-28%], P = .92) and proportion of patients with ≥2 PPM1Dmt (26.5% vs 30%, P = .70) were not significantly different between the PPM1Dmt/TP53wt and PPM1Dmt/TP53mt groups, respectively. The proportions of patients with TET2 and splicing factor gene mutations were significantly higher in the PPM1Dwt/TP53wt group (42% and 21%), followed by PPM1Dmt/TP53mt (20% and 20%), PPM1Dmt/TP53wt (11% and 2%), and TP53mt/PPM1Dwt groups (15% and 0%), with P values of .004 and <.001, respectively. No statistically significant differences in CHRS were observed across the 4 genotypic groups (Table 1).14 However, high-risk mutations, as defined by the CHRS, were present in 12 (7%), 25 (100%), 17 (100%), and 64 (51%) patients with PPM1Dmt/TP53wt, TP53mt/PPM1Dmt, TP53mt/PPM1Dwt, and TP53wt/PPM1Dwt genotypes, respectively (P ≤ .001).
The median follow-up durations were 11.2, 8.8, 23.6, and 34.3 months from the time of diagnosis in PPM1D mt/TP53wt, PPM1Dmt/TP53mt, PPM1Dwt/TP53mt, and PPM1Dwt/TP53wt groups, respectively (P = .001). Although follow-up duration was shorter in PPM1Dmt/TP53wt and PPM1Dmt/TP53mt groups, the rate of myeloid neoplasm transformation was negligible in PPM1Dmt/TP53wt (0.8%) and PPM1Dmt/TP53mt (0%) groups, compared with PPM1Dwt/TP53mt (23%) and PPM1Dwt/TP53wt groups (12.5%; P ≤ .001), respectively. Similarly, only patients in the PPM1Dwt/TP53wt group had documented AML transformation (12.5%; P = .009; Table 2).
Distribution of mutational dominance patterns
In participants with comutated PPM1D and TP53 (n = 23), the distribution across categorical differences in dominance was well spread: codominant (n = 9), TP53-dominant (n = 8), and PPM1D-dominant (n = 6; Figure 2A). The codominant clones spanned a substantial range of PPM1D VAFs between 3% and 28%, demonstrating that codominance can occur across different clonal sizes. Marginal density distributions showed that both PPM1Dmt and TP53mt exhibited similar VAF ranges, consistent with the literature-supported equivalent fitness characteristics of these mutations.
Figure 2.
Mutational dominance analysis of PPM1D and TP53 co-occurrence. (A) Scatterplot showing the relationship between TP53 VAF (x-axis) and PPM1D VAF (y-axis) in individual samples. Points are colored by dominance category: codominant (orange), PPM1D-dominant (cyan), and TP53-dominant (magenta). The shaded orange region indicates the codominance zone where −1 < log2(PPM1D_VAF/TP53_VAF) < 1, creating the characteristic conical shape due to the order-of-magnitude–based classification. Marginal density plots show the distribution of VAFs for each mutation type. The diagonal dashed line represents equal VAF between PPM1D and TP53 mutations. (B) Stacked bar chart displaying the relative proportions of different dominance categories across clinical categories (CCUS, CHIP, t-CCUS, and t-CHIP). Colors represent: PPM1D mutant with TP53 wild-type (blue), PPM1D-dominant (cyan), codominant (orange), TP53-dominant (magenta), PPM1D wild-type with TP53 mutant (red), and wild-type for both genes (green). The chart demonstrates the shift toward increased codominant mutations in therapy-induced conditions and the predominance of single mutations or wild-type status in CCUS. CHIP, CH of indeterminate potential.
Clinical distribution and therapy-associated changes
The relative proportions analysis revealed that there is a small proportion of comutated individuals within the CCUS category, which is dominated by either PPM1Dwt/TP53wt or PPM1Dmt/TP53wt genotypes (Figure 2B). Notably, we observed that the proportion of individuals with codominant mutations increased substantially in patients with t-CH of indeterminate potential and t-CCUS compared with their nontherapy-associated counterparts. CCUS samples showed the highest proportion of wt individuals, whereas the therapy-induced categories (t-CCUS and t-CH of indeterminate potential) demonstrated increased complexity in mutational patterns. The increased prevalence of codominant mutations in therapy-induced conditions suggests that treatment-related selective pressures may favor the emergence of multiple competing clones with equivalent fitness advantages.
Survival analysis
We then analyzed survival outcomes among patients with CH and CCUS. In the CH subgroup, the mPFS (P = .10) and mOS (P = .10) were not significantly different in PPM1Dmt/TP53wt (not reached [NR]; 68.4% at 2 years, each), PPM1Dmt/TP53mt (20.4 months and NR [77.7% at 2 years]), PPM1Dwt/TP53mt (NR; 100% at 2 years, each), and PPM1Dwt/TP53wt (60.67 months, each) groups, respectively (Figure 3A-B). In the CCUS subgroup, although the mPFS (P = .17) was not significantly different, the mOS (P = .02) was significantly different among the 4 groups. The mOS and mPFS were NR (62% and 65.2% at 2 years) in PPM1Dmt/TP53wt, NR (67.5% at 2 years, each) in PPM1Dmt/TP53mt, 16.7 and 23.6 months in PPM1Dwt/TP53mt, and NR (69% and 78.5% at 2 years) in PPM1Dwt/TP53wt groups, respectively (Figure 3C-D).
Figure 3.
Survival curves in CH/CCUS. Kaplan-Meier survival curves depicting (A) progression-free survival and (B) OS in CH group; and (C) progression-free survival and (D) OS in CCUS group across 4 genotypes (PPM1Dmt/TP53wt, PPM1Dmt/TP53mt, PPM1Dwt/TP53wt, and PPM1D wt/TP53mt).
We also analyzed survival outcomes among patients with t-CH and t-CCUS. In the t-CH subgroup, the mPFS (P = .12) and mOS (P = .22) were not significantly different in PPM1Dmt/TP53wt (NR; 60.7% and 64.5% at 2 years), PPM1Dmt/TP53mt (NR; 75% at 2 years, each), PPM1Dwt/TP53mt (NR; 100% at 2 years, each), and PPM1Dwt/TP53wt (60.67 months, each) groups, respectively (Figure 4A-B). Similarly, in the t-CCUS subgroup, the mPFS (P = .90) and mOS (P = .29) were not significantly different in PPM1Dmt/TP53wt (NR; 51% at 2 years and 31.5 months), PPM1Dmt/TP53mt (NR; 58.3% at 2 years, each), PPM1Dwt/TP53mt (14.3 months and 53.3% at 2 years), and PPM1Dwt/TP53wt (38.08 months, each) groups, respectively (Figure 4C-D).
Figure 4.
Survival curves in t-CH/t-CCUS. Kaplan-Meier survival curves depicting (A) progression-free survival and (B) OS in t-CH group; and (C) progression-free survival and (D) OS in t-CCUS group across 4 genotypes (PPM1Dmt/TP53wt, PPM1Dmt/TP53mt, PPM1Dwt/TP53wt, and PPM1D wt/TP53mt).
We analyzed survival outcomes based on PPM1Dmt VAF, comparing patients with VAF <13% vs VAF ≥13%. The VAF cutoff was chosen based on receiver operating characteristic analysis using the Youden index; PPM1D VAF ≥13% was predictive of survival. We observed a significant difference in PFS (NR; 73% at 2 years vs 21.6 months; P = .004) and OS (NR; 73% at 2 years vs 22.9 months; P = .01). We subsequently developed a parsimonious multivariable model for OS and PFS, including variables that meaningfully contributed to outcomes, ensuring a more interpretable and generalizable model (supplemental Table 2). Active nonmyeloid cancer (hazard ratio [HR], 4.61 [95% confidence interval (CI), 2.26-9.41]; P < .001/HR, 4.79 [95% CI, 2.32-9.80]; P < .001), age ≥70 years (HR, 2.55 [95% CI, 1.21-5.34]; P = .01/HR, 2.64 [95% CI, 1.25-5.56]; P = .01) and PPM1D-mutant VAF ≥13% (HR, 2.36 [95% CI, 1.22-4.54]; P = .01/HR, 2.28 [95% CI, 1.18-4.37]; P = .01) remained independently associated with inferior PFS and OS, respectively. Interestingly, multiple PPM1D mutations did not impact OS or PFS.
Discussion
In our comprehensive study assessing the clonal landscape of patients with PPM1Dmt CH/CCUS, we demonstrate that patients with PPM1Dmt/TP53wt and PPM1Dmt/TP53mt genotypes were significantly enriched in the therapy-related subgroup (39%), indicating a strong correlation of these mutations with genotoxic therapy exposures. Additionally, the median time interval from last genotoxic therapy exposure to diagnosis of t-CH/t-CCUS was shorter (≤6 months) in the PPM1Dmt/TP53wt and PPM1Dmt/TP53mt groups, in comparison to >1 year in the PPM1Dwt/TP53mt and PPM1Dwt/TP53wt groups, respectively. The therapy-related PPM1Dmt CH group, with or without TP53 mutations, had significantly higher exposures to PARP inhibitors (30%) and radioligand therapies (50%). Although PPM1D mutations were frequently identified in t-CH/t-CCUS in the context of low VAF (6% [range, 0.3%-50%]), within the limitations of a short follow-up (7.13 months [range, 0.17-63.5]) in our cohort, they were associated with a low rate of myeloid clonal evolution and transformation, even in the presence of concomitant TP53 mutations (21/146 [14%]). In multivariable analysis, t-CH/t-CCUS and PPM1D-mt VAF ≥13% remained independently associated with inferior PFS and OS, respectively.
PPM1D is part of a regulatory feedback loop with p53, in which activated p53 induces expression of PPM1D, which then dephosphorylates p53, resulting in downregulation of p53-mediated apoptosis.6 PPM1D has been found to be mutated in several human cancers, including breast, lung, and colorectal cancers.6,15, 16, 17, 18 Although the mechanism of leukemogenesis in TP53mt precursor disease states and myeloid neoplasms is well known,10,19,20 mechanisms remain to be elucidated for PPM1Dmt disease.1 In patients with CH, in general, VAFs of ≥20% are strongly associated with hematological dysfunction (CCUS); however, this is not the case with PPM1D mutations, where lower VAFs (<10%) are associated with cytopenias. These data are in line with previous observations that showed that clonal fitness can substantially differ by gene category,12 and that PPM1D mutations, even at lower VAF, can influence clonal selection.
In a recent study by Fandrei et al,21 112 patients with PPM1D mutations across various myeloid neoplasms were analyzed, with DNMT3A (29%) and TP53 (25%) being the most frequently comutated genes. In their study, among 44% of patients with PPM1D-mutated AML, the mutation was identified in the founder clone with rare TP53 comutations, indicating its earlier presence in disease progression. Moreover, single-cell DNA and surface protein analysis in 7 patients confirmed that PPM1D mutations can arise in the founding clone and are associated with the expression of leukemic markers. We investigated the fitness distributions and temporal acquisition patterns of PPM1Dmt and TP53mt clones. We observed that codominant clones spanned a substantial range of PPM1D VAFs between 3% and 28%, demonstrating that codominance can occur across different clonal sizes. Furthermore, marginal density distributions showed that both PPM1Dmt and TP53mt exhibited similar VAF ranges, consistent with earlier observations that PPM1Dmt and TP53mt have equivalent fitness characteristics.12 Furthermore, the clonal status of PPM1D mutations differs markedly across disease states and previous cytotoxic therapy. PPM1Dmt was more likely to be dominant in CH than CCUS, and especially following cytotoxic therapy. These findings emphasize that the clonal positioning of PPM1D mutations is determined by disease and previous cytotoxic therapy. The enrichment of subclonal PPM1D mutations in therapy-related myeloid neoplasms, often subordinate to mutant TP53, highlights a potential model in which PPM1Dmt clones emerge early under selective pressure, cause significant cytopenia, but are outcompeted or superseded by TP53 mutations during malignant progression. These data collectively point to a stage- and therapy-dependent role for PPM1D in clonal evolution, with implications for early detection, risk stratification, and potential intervention in precursor states.21,22
Interestingly, among the t-CH/t-CCUS subgroups, in terms of previous cytotoxic therapy, PPM1D mutations with or without TP53 mutations were predominantly observed among patients with previous PARP inhibitors or radioisotope therapy exposures, as has been reported previously.23, 24, 25 Although we cannot be certain that these mutations were present in all patients before these therapies, it is probable that PPM1D and TP53 mutations had a selective advantage compared with other genes on exposure to these genotoxic stressors. In a recently conducted study analyzing patients with Hodgkin lymphoma treated with autologous transplant (hematopoietic cell transplant), targeted DNA sequencing was performed on peripheral blood stem cell products to identify mutations associated with CH and MN.4 CH was detected in 46 of 321 (14.3%) patients; DNMT3A (54%), PPM1D (15.2%), TET2 (15.2%), and TP53 (10.8%) were most common. The presence of TP53 and/or PPM1D mutations was associated with a 7.29-fold higher hazard of therapy-related myeloid neoplasm than individuals without CH mutations. These mutations were also associated with 4.17-fold higher hazard of nonrelapse mortality. In our cohort, 23 of 170 (13.5%) patients with t-CH/t-CCUS had undergone an autologous hematopoietic cell transplant, with relatively equal distributions across the 4 genotype groups.
We acknowledge important limitations of our study, including shorter follow-up durations in the PPM1Dmt/TP53wt and PPM1Dmt/TP53mt groups, heterogeneity in genomic sequencing techniques between the 5 academic institutions that participated in this study, and the fact that the time of mutation detection may have lagged behind the actual onset of a clonal event. Nevertheless, to our knowledge, this work represents the largest series reporting on the spectrum of precancerous PPM1D mutations in the context of evolving cancer therapies. In conclusion, although PPM1D mutations were frequently identified in t-CH/t-CCUS and were associated with unexplained cytopenias in the context of low VAF, within the limitations of a short follow-up in our cohort, they were associated with a low rate of myeloid clonal evolution, even in the presence of mutant TP53. Longer-term follow-up is planned to better assess impacts on PFS and OS.
Conflict-of-interest disclosure: T.B. reports receiving research funding from Takeda; and served on advisory boards for Takeda, MorphoSys, Pfizer, Syndax, and Amgen. M.M.P. received research funding from Kura Oncology, Stemline, Polaris, Epigenetix, and Solu Therapeutics; and served on the advisory boards for Sobi/CTI, GlaxoSmithKline, and AstraZeneca. E.L. received institutional support from the Centres de Recerca de Catalunya Program/Generalitat de Catalunya. O.J. served on advisory boards for Ascentage, Sobi, Maat, and TERN. A.P. received research funding from Kronos Bio, Sumitomo, Servier, Incyte, and Pfizer; and honoraria for advisory boards from Sobi and AbbVie. C.F. served on advisory board for Stemline; and received honoraria from the Binaytara Foundation. C.L. served on advisory boards for ADC Therapeutics and Autolus; and provided consulting services for Rigel. The remaining authors declare no competing financial interests.
Acknowledgments
The authors thank the patients and families for contributing to this study. M.M.P. would like to thank the Mayo Clinic Center for Individualized Medicine and Mayo Clinic Comprehensive Cancer Center for sponsoring the CH/CCUS Clinic. The authors also acknowledge Mahesh Kumar for his valuable efforts in data collection for this project. This research was partially carried out at the Barcelona Collaboratorium for Modelling and Predictive Biology.
E.L. was funded by the Spanish Research Agency through the Severo Ochoa (CEX2020-001049-S, MCIN/AEI/10.13039/501100011033) and Maria de Maeztu Program for Centers and Units of Excellence in R&D (CEX2020-001084-M).
Authorship
Contribution: T.B., L.M., E.L., and M.M.P. designed the study, performed the analysis, and wrote the manuscript; M.E.K., B.J.M., and K.J.S. helped in collecting data; T.L., F.D.O., C.L., Y.K., O.J., K.D., A.C., C.F., J.F., M.K.-D., Y.-S.F., L.J., R.H., M.T., A.P., D.V., M.V.S., A.S., A.M., A.A.-K., N.G., and M.L. contributed patients; and all authors reviewed and approved the final draft of the manuscript.
Footnotes
Individual patient data have been deidentified to protect participant privacy. Data supporting the findings of this study are available from the corresponding author, Talha Badar (badar.talha@mayo.edu), upon reasonable request.
The full-text version of this article contains a data supplement.
Supplementary Material
References
- 1.Kartal-Kaess M, Bochtler T, Kraft B, et al. PPM1D mutations are rare in de novo and therapy-related acute myeloid leukemia. Blood. 2018;132(suppl 1):1472. [Google Scholar]
- 2.McNerney ME, Godley LA, Le Beau MM. Therapy-related myeloid neoplasms: when genetics and environment collide. Nat Rev Cancer. 2017;17(9):513–527. doi: 10.1038/nrc.2017.60. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Wong TN, Ramsingh G, Young AL, et al. Role of TP53 mutations in the origin and evolution of therapy-related acute myeloid leukaemia. Nature. 2015;518(7540):552–555. doi: 10.1038/nature13968. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Yan C, Richard MA, Gibson CJ, et al. Clonal hematopoiesis and therapy-related myeloid neoplasms after autologous transplant for Hodgkin lymphoma. J Clin Oncol. 2024;42(20):2415–2424. doi: 10.1200/JCO.23.02547. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Bartek J, Lukas J. Chk1 and Chk2 kinases in checkpoint control and cancer. Cancer Cell. 2003;3(5):421–429. doi: 10.1016/s1535-6108(03)00110-7. [DOI] [PubMed] [Google Scholar]
- 6.Hsu JI, Dayaram T, Tovy A, et al. PPM1D mutations drive clonal hematopoiesis in response to cytotoxic chemotherapy. Cell Stem Cell. 2018;23(5):700–713.e6. doi: 10.1016/j.stem.2018.10.004. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Bondar T, Medzhitov R. p53-mediated hematopoietic stem and progenitor cell competition. Cell Stem Cell. 2010;6(4):309–322. doi: 10.1016/j.stem.2010.03.002. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Milosevic J, Fransson S, Gulyas M, et al. High expression of PPM1D induces tumors phenotypically similar to TP53 loss-of-function mutations in mice. Cancers. 2021;13(21):5493. doi: 10.3390/cancers13215493. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Badar T, El Kettani M, Shah KJ, et al. PPM1D somatic mutations in myeloid neoplasms: clinical profile, clonal characteristics and impact of concurrent somatic TP53 mutations. Blood. 2024;144(suppl 1):3218. [Google Scholar]
- 10.Shahzad M, Amin MK, Daver NG, et al. What have we learned about TP53-mutated acute myeloid leukemia? Blood Cancer J. 2024;14(1):202. doi: 10.1038/s41408-024-01186-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Xie Z, Komrokji R, Al Ali N, et al. Risk prediction for clonal cytopenia: multicenter real-world evidence. Blood. 2024;144(19):2033–2044. doi: 10.1182/blood.2024024756. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Robertson NA, Latorre-Crespo E, Terradas-Terradas M, et al. Longitudinal dynamics of clonal hematopoiesis identifies gene-specific fitness effects. Nat Med. 2022;28(7):1439–1446. doi: 10.1038/s41591-022-01883-3. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Latorre-Crespo E, Robertson NA, Kosebent EG, et al. Clinical progression of clonal hematopoiesis is determined by a combination of mutation timing, fitness, and clonal structure. bioRxiv. Preprint posted online 3 March 2025 doi: 10.1101/2025.02.28.640879. [DOI] [Google Scholar]
- 14.Weeks LD, Niroula A, Neuberg D, et al. Prediction of risk for myeloid malignancy in clonal hematopoiesis. NEJM Evid. 2023;2(5) doi: 10.1056/evidoa2200310. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Deng W, Li J, Dorrah K, et al. The role of PPM1D in cancer and advances in studies of its inhibitors. Biomed Pharmacother. 2020;125 doi: 10.1016/j.biopha.2020.109956. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Lambros MB, Natrajan R, Geyer FC, et al. PPM1D gene amplification and overexpression in breast cancer: a qRT-PCR and chromogenic in situ hybridization study. Mod Pathol. 2010;23(10):1334–1345. doi: 10.1038/modpathol.2010.121. [DOI] [PubMed] [Google Scholar]
- 17.Peng T-S, He Y-H, Nie T, et al. PPM1D is a prognostic marker and therapeutic target in colorectal cancer. Exp Ther Med. 2014;8(2):430–434. doi: 10.3892/etm.2014.1762. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Yang H, Gao X-Y, Li P, Jiang T-S. PPM1D overexpression predicts poor prognosis in non-small cell lung cancer. Tumour Biol. 2015;36(3):2179–2184. doi: 10.1007/s13277-014-2828-6. [DOI] [PubMed] [Google Scholar]
- 19.Bernard E, Nannya Y, Hasserjian RP, et al. Implications of TP53 allelic state for genome stability, clinical presentation and outcomes in myelodysplastic syndromes. Nat Med. 2020;26(10):1549–1556. doi: 10.1038/s41591-020-1008-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Badar T, Nanaa A, Atallah E, et al. Prognostic impact of ‘multi-hit' versus ‘single-hit' TP53 alteration in patients with acute myeloid leukemia: results from the Consortium on Myeloid Malignancies and Neoplastic Diseases. Haematologica. 2024;109(11):3533–3542. doi: 10.3324/haematol.2024.285000. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Fandrei D, Pegliasco J, Pasquier F, et al. Clonal evolution of PPM1D mutations in the spectrum of myeloid disorders. Clin Cancer Res. 2025;31(11):2241–2253. doi: 10.1158/1078-0432.CCR-24-3683. [DOI] [PubMed] [Google Scholar]
- 22.Miller PG, Sperling AS, Mayerhofer C, et al. PPM1D modulates hematopoietic cell fitness and response to DNA damage and is a therapeutic target in myeloid malignancy. Blood. 2023;142(24):2079–2091. doi: 10.1182/blood.2023020331. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Marmouset V, Decroocq J, Garciaz S, et al. Therapy-related myeloid neoplasms following PARP inhibitors: real-life experience. Clin Cancer Res. 2022;28(23):5211–5220. doi: 10.1158/1078-0432.CCR-22-1622. [DOI] [PubMed] [Google Scholar]
- 24.Kusne Y, Lasho T, Finke C, et al. Clonal hematopoiesis in patients with neuroendocrine tumor treated with Lutetium-177 and the risk of thrombocytopenia: a prospective study. JCO Precis Oncol. 2024;8 doi: 10.1200/PO.24.00143. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Kusne Y, Patnaik MM, Halfdanarson TR, Sonbol MB. Therapy-related myeloid neoplasms in 177Lu-DOTATATE treated neuroendocrine tumor patients: how great is the risk? Endocr Relat Cancer. 2025;32(6) doi: 10.1530/ERC-25-0025. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.