Somatic Mosaic Mutations in PPM1D and TP53 in the Blood of Women With Ovarian Carcinoma

Abstract

IMPORTANCE

Somatic mosaic mutations in PPM1D have been reported in patients with breast cancer, lung cancer, and ovarian cancer (OC), but cause or effect has not been established.

OBSERVATIONS

To test the hypothesis that somatic mosaic mutations are associated with chemotherapy exposure, we used massively parallel sequencing to quantitate mutations in peripheral blood mononuclear cells (PBMCs) of 686 women with primary OC (n = 412) or relapsed OC (n = 274). The frequency of somatic mosaic PPM1D mutations in PBMCs was significantly associated with prior chemotherapy (P < .001), and, in patients exposed to chemotherapy, with older age at blood draw (recurrent OC odds ratio [OR], 17.24; 95%CI, 6.80–43.69; and primary OC postchemotherapy OR, 4.82; 95%CI, 1.43–16.18). In contrast, somatic mosaic mutations in TP53 were not significantly associated with chemotherapy or age. In sequential PBMC samples harvested from 13 patients with OC near diagnosis and after a median of 2 different chemotherapy regimens, somatic mosaic PPM1D mutations increased in 11 individuals (84.6%) and TP53 mutations appeared in 2 (15.4%).

CONCLUSIONS AND RELEVANCE

Chemotherapy exposure and age influence the accumulation of PPM1D-mutated PBMC clones. Care should be taken to control for chemotherapy exposure and age at blood draw when testing the association of somatic mosaic mutations in PBMCs with cancer risk.

In 2011, an association between somatic mosaic mutations in the p53-inducible protein phosphatase gene PPM1D in peripheral blood mononuclear cells (PBMCs) in patients with breast cancer and ovarian cancer (OC) was reported.¹ All PPM1D mutations were truncating mutations in the last exon, were thought to be activating, and were absent from matched cancers. Akbari et al² confirmed these findings and suggested that somatic PPM1D mutations reveal an OC predisposition that warrants risk-reducing salpingo-oophorectomy. Somatic mosaicPPM1Dmutations have also been reported in lung cancer.³ Whether PPMID mutations reflect an underlying cancer predisposition or consequence of cancer therapy is unclear. Here we investigated whether PPM1D somatic mosaic mutations reflect a previously unrecognized association with prior chemotherapy and whether similar mutations also occur in other genes associated with OC.

Methods

Blood and tumor tissue were collected from patients with OC who provided written consent, including 412 patients with primary OC from 2 institutional tissue repositories and 274 patients enrolled in a Gynecologic Oncology Group clinical trial for recurrent platinum-resistant OC. All protocols were approved by an institutional review board. Cancer-associated genes in PBMC DNA and tumor DNA were sequenced with BROCA (University of Washington)^4–6 to assess the relationship between prior chemotherapy exposure, age, and somatic mosaic mutations. Targeted genes are listed in eTable 1 in the Supplement. Deleterious mutations present at low frequency in PPM1D, and other genes were individually examined using the Integrative Genomics Viewer program (Broad Institute) to verify local sequence quality. Variants were included if at least 5 high-quality reads were identified. For 13 women with recurrent OC having a PMM1D mutation, DNA from PBMCs obtained close to the time of diagnosis was also sequenced. Validation of low variant calls was performed by resequencing with BROCA at greater depth or, for cases with a greater than 5% variant reads, by Sanger sequencing. Two-tailed P values were generated for contingency tables using Fisher exact test and for age comparisons using the exact Wilcoxon test. Odds ratios were calculated from contingency tables using GraphPad Prism (GraphPad Software).

Results

Targeted sequencing was applied to PBMCs from 686 women with OC. Among 65 genes examined (eTable 1 in the Supplement), somatic mosaic mutations were observed in PPM1D in 69 patients (truncating mutations exclusively in exon 6) and in TP53 in 11 patients (2 frameshift mutations, 8 deleterious missense mutations in the p53DNA binding domain, and 1 missense mutation [p.M44I] predicted to not affect function [http://p53.iarc.fr/TP53GeneVariations.aspx]).

Somatic mosaic PPM1D mutations in women with OC were strongly associated with prior chemotherapy exposure (P < .001) and, for those patients exposed to chemotherapy, with age at blood draw (Table). The presence of PPM1D mutations in patients with relapsed OC was also associated with more previous chemotherapy regimens: PPM1D mutations were detected in 21 of 138 women (15.2%) after 1 regimen vs 35 of 130 women (26.9%) after 2 regimens (P = .02). The occurrence of multiple different PPM1D mutations in the same individual was limited to patients with relapsed OC (14 of 274 women [5.1%] vs 0 of 412; P < .001).

Table .

PPM1D Mutations in PBMCs From Women With Primary and Recurrent Ovarian Carcinoma

Ovarian Carcinoma Subgroup	No.	No. With PPM1D Mutation (Fraction)	OR (95% CI)	Mean Age, y		P Value
				Without PPM1D Mutation	With PPM1D Mutation
Primary, no CT	326	5 (.015)a	NA	61.8	69.0	.30
Primary, CT exposed	86	6 (.070)a	4.82 (1.43–16.18)	55.4	68.2	.02
Recurrent, platinum-resistant	274	58 (.212)a	17.24 (6.80–43.69)	62.1	65.7	.01

Abbreviations: CT, chemotherapy; NA, not applicable; OR, odds ratio; PBMCs, peripheral blood mononuclear cells.

^aFisher exact test of an equal frequency of PPM1D mutations across the 3 chemotherapy groups, P < .001; no CT vs CT exposed (P = .013); and CT exposed vs recurrent platinum-resistant ovarian carcinoma (P < .002).

Somatic mosaic mutations were not confirmed in any other gene on the BROCA panel except TP53. Somatic mosaic TP53 mutations were present in 4 of 326 women (1.2%) with OC not exposed to previous chemotherapy and 7 of 274 women (2.6%) with recurrent, platinum-resistant OC, and were not significantly associated with chemotherapy exposure (P = .24). As chemotherapy was not associated with TP53 mutations, we evaluated the association with age in the entire cohort of 686 patients with OC. The mean age—67.1 years in the 11 patients with TP53 mutations and 61.5 years in those without mutations— was not significantly different (Monte Carlo-based Wilcoxon test, P = .07). Of the 11 blood samples with somatic mosaic TP53 mutations, 5 also had PPM1D mutations.

Paired neoplastic tissue was available for 4 women with somatic TP53 mutations and 13 with PPM1D mutations. In no case was the PBMC mutation identified in tumor DNA.

To further assess whether treatment influenced the presence or frequency of these somatic mosaic mutations, we sequenced DNA at increased depth from pairs of PBMC samples obtained from 13 patients with somatic mosaic mutations near diagnosis and again at relapse (eTable 2 in the Supplement). Overall, 21 PPM1D mutations and 2 TP53 mutations were either undetectable at diagnosis or present at a lower fraction of reads than at recurrence. In 3 patients, PPM1D mutations occurred in a high fraction of reads (30%–37%) at diagnosis and did not increase with chemotherapy exposure.

Key Points.

Question

Does chemotherapy exposure influence the presence of somatic mosaic PPM1D mutations in the blood of women with ovarian carcinoma?

Findings

In 686 women with ovarian carcinoma, the frequency of somatic mosaic PPM1D mutations in peripheral blood mononuclear cells (PBMCs) was significantly associated with prior chemotherapy, and, in patients exposed to chemotherapy, with older age at blood draw.

Meaning

Chemotherapy exposure and age influence the accumulation of PPM1D-mutated PBMC clones, and these factors should be controlled for when testing the association of somatic mosaic mutations in PBMCs with cancer risk.

Discussion

Our results demonstrate that somatic PPM1D mutations in PBMCs are associated with prior chemotherapy and older age. Somatic TP53 mutations were also identified in 11 patients (1.6%) but were not significantly associated with chemotherapy or age. Damaging somatic mosaic mutations in other genes on the BROCA panel were not identified. These results shed new light on recent reports linking somatic mosaic PPM1D mutations with solid tumor risk.

Previous studies have reported that the presence of PPM1D mutations in PBMCs was associated with breast cancer and OC but cause and effect were not established.¹ However, some investigators suggested that identification of such mutations might warrant risk-reducing salpingo-oophorectomy.² Somatic mosaic PPM1D mutations have also been reported in lung cancer.³ Although it has been assumed that the presence of these mutations might reflect an underlying repair defect that contributes to cancer predisposition, our results suggest that this association between PPM1D mutations and solid tumors might have been confounded by prior chemotherapy. In particular, we observe a stepwise increase in the incidence of PPM1D somatic mosaic mutations between patients with OC without prior chemotherapy, patients with newly diagnosed OC but prior chemotherapy exposure, and patients previously treated extensively for OC (Table). Moreover, 11 of 13 pairs of sequential samples show new appearance or increased frequency of PPM1D mutations at 21 of 24 individual mutation sites during the course of treatment (eTable 2 in the Supplement). These observations suggest that PPM1D somatic mosaic mutations reflect prior chemotherapy exposure. Consistent with this suggestion, all 26 patients with lung cancer and OC with a PPM1D mutation previously identified by Akbari et al² and Zajkowicz et al³ also had received chemotherapy before PPM1D sequencing. It is possible that there may also be a rare, second group of younger, chemotherapy-naive patients with a more dominant PPM1D clone (eg, patients 11–13 in eTable 2 in the Supplement), but the clinical significance of that finding requires additional investigation.

We also identified deleterious TP53 mutations in PBMCs from patients with OC. While this work was in progress, Wong et al⁷ detected TP53 mutations at very low allele fractions (0.003%–0.7%) in 9 of 19 elderly individuals without cancer, demonstrated in mouse bone marrow chimeras that TP53^+/− cells are preferentially expanded after chemotherapy, and concluded that TP53 mutations may confer a subtle competitive advantage to nonmalignant clones within the normal marrow, particularly during DNA damaging chemotherapy. In patients with OC, we did not find an association of TP53 mutations with age or chemotherapy exposure, but the number of PBMC TP53 mutations was small (n = 11). Notably, in 2 cases, TP53 mutations emerged after prolonged chemotherapy (eTable 2 in the Supplement), suggesting that there may be an effect of extended chemotherapy on somatic TP53 mutations, although that effect was not detected in the larger cohort. At the time of PBMC harvest, none of the OC patients with a PBMC TP53 mutation had evidence of a therapy-related hematologic malignancy.

Conclusions

Given the relationship between prior chemotherapy, age, and somatic mosaic mutations in PPM1D, long-term prospective studies appear to be required to determine whether PPM1D or TP53 somatic mosaic mutations by themselves actually confer an increased risk of primary or secondary neoplasms.