Introduction

Malignant peritoneal mesothelioma is rare, with ~300 new patients annually in the United States, and accounts for ~10% of all malignant mesotheliomas (the remainder primarily pleural in origin) [1,2,3]. Peritoneal mesothelioma typically involves men over 50 years of age, women with a wide age range, and rarely adolescents and children [4,5,6,7,8]. The presentation is often diffuse, although exceptional cases of localized peritoneal mesothelioma with no evidence of diffuse serosal spread have been reported [9]. While the prognosis of malignant mesothelioma is generally dismal, with a median survival of 2–4 years despite multimodality treatment [10, 11], some patients show a protracted clinical course [4, 5, 7]. Risk factors for the development of peritoneal mesothelioma include asbestos exposure (albeit with a weaker association than its pleural counterpart) [12], exposure to non-asbestos mineral fibers [13, 14], therapeutic radiation for a prior malignancy, in the setting of chronic inflammatory conditions [15, 16], and in the context of germline BAP1 inactivation syndrome or other germline mutations [17,18,19,20]. Nonetheless, the pathogenesis of peritoneal mesothelioma in many patients, particularly those with no known risk factors, remains unclear.

Recurrent mutations in tumor suppressors and epigenetic regulators, including BAP1, NF2, SETD2, and TP53, have been identified by genomic profiling studies collectively in over 900 pleural mesotheliomas [21,22,23,24,25,26,27,28] and 129 peritoneal mesotheliomas analyzed to date [20, 24, 25, 29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47] (only 61 tumors had clinicopathologic information; Table 1). In particular, recurrent somatic and/or germline BAP1 mutations have been noted in 50–70% of both pleural and peritoneal mesotheliomas [12, 48,49,50,51,52,53,54]. Aside from BAP1, other rare distinct genetic alterations reported in a subset of diffuse malignant peritoneal mesothelioma include EWSR1-YY1 fusion [55], EWSR1-ATF1 and FUS-ATF1 fusions [36], and ALK rearrangements [38, 40, 44, 56, 57]. However, molecular features of diffuse malignant peritoneal mesothelioma, including those lacking BAP1 alterations, and other mesothelioma variants such as localized peritoneal mesothelioma remain poorly understood.

Table 1 Summary of prior published next-generation or other high-throughput sequencing studies of diffuse malignant peritoneal mesotheliomas and this study.

In this study, using targeted next-generation sequencing we examined the genomic characteristics of 26 diffuse malignant peritoneal mesotheliomas (including nine tumors previously published) [40]. In addition to BAP1 mutations, we identified recurrent alterations in other DNA repair, chromatin, and cell cycle regulators. We also identified a diffuse malignant peritoneal mesothelioma with evidence of genomic near-haploidization, highlighting the genetic heterogeneity of peritoneal mesothelioma and implicating these processes in its pathogenesis and potential therapeutic targets. As part of an exploratory analysis, we analyzed one localized peritoneal mesothelioma and one well-differentiated papillary mesothelioma with invasive foci.

Materials and methods

After approval by the Institutional Review Board, cases were retrieved from the surgical pathology and consultation files of Brigham and Women’s Hospital, Boston, MA. From a cohort of 88 patients with primary diffuse malignant peritoneal mesotheliomas diagnosed in 2005–2015, we previously identified three tumors with ALK rearrangements as selected by ALK immunohistochemistry screening and confirmatory sequencing [40] and characterized the genetic alterations of nine tumors. In this study, we identified 17 additional patient samples of diffuse malignant peritoneal mesothelioma with sufficient tissue for targeted next-generation sequencing. In addition, we analyzed a localized peritoneal mesothelioma and a well-differentiated papillary mesothelioma with invasive foci. Tumors were classified into epithelioid, biphasic, and sarcomatoid types according to the WHO criteria [1]. Evaluation for asbestos exposure and the presence of pleural plaques was based on the medical and radiologic records.

Immunohistochemistry was performed on 4-micron-thick formalin-fixed paraffin-embedded whole-tissue sections for the following antibodies (clone, dilution, antigen retrieval method, vendor): BAP1 (C4, 1:30, citrate buffer pressure cook, Santa Cruz Biotechnology, Santa Cruz, CA), calretinin (polyclonal, 1:200, citrate buffer pressure cook, Life Technologies, Carlsbad, CA), D2-40 (D2-40, 1:100, no retrieval, Biolegend, Dedham, MA), and WT1 (6F-H2, 1:50, citrate buffer pressure cook, Dako, Carpinteria, CA). BAP1 expression was scored as complete loss when the tumor cells showed >90% reduction in nuclear staining. Electron microscopy was performed in the localized peritoneal mesothelioma as previously described [58]. Fluorescence in situ hybridization testing for loss of 9p (CDKN2A) and 22q (NF2) was performed in select cases as previously described [59]. Overall survival was defined as the time of pathologic diagnosis to the time of death from any cause or to the time of last clinical follow-up, at which point the survival data were censored; survival data were analyzed using univariate Cox regression analysis. Statistical analysis of the categorical and numerical data was performed using Fisher’s exact tests and unpaired t tests, respectively, via GraphPad InStat version 3.1 (LaJolla, CA), with a significant p value threshold of < 0.05 and Bonferroni correction applied in multiple statistical comparisons.

For targeted next-generation sequencing, DNA (at least 50 ng/µl) extracted from formalin-fixed paraffin-embedded whole-tissue sections was hybridized to capture probes (Agilent SureSelect; Santa Clara, CA), followed by sequencing using Illumina HiSeq 2500 (San Diego, CA). Over the study period, three versions of the in-house sequencing panel were used, targeting 275 genes covering 757787 base-pairs in 3 (12%) cases (#17–19), 298 genes covering 831033 base-pairs in 6 (23%) cases (#9, #11–12, #23–25), and 447 genes covering 1315708 base-pairs in 17 (65%) cases (#1–8, #10, #13–16, #20–22, and #26) of diffuse malignant peritoneal mesothelioma, one localized peritoneal mesothelioma, and one well-differentiated papillary mesothelioma with invasive foci; the protocol was previously published [60, 61], and the complete gene lists of the 275-, 298-, and 447-gene panels were in Supplementary Table 1. Single-nucleotide variant analysis was restricted to pathogenic alterations in tumor suppressor genes and oncogenes as well as loss-of-function variants (nonsense, frameshift, splice site variants, and structural/rearrangement). Single-nucleotide variants were excluded from analysis if synonymous, harboring a minor variant allele frequency of >0.1% in the Exome Sequencing Project database (University of Washington, Seattle, WA), or suspected germline based on the allele frequencies considering the tumor percentage and associated copy number changes; however, single-nucleotide variants were rescued if listed in the Catalogue of Somatic Mutations in Cancer (COSMIC) database (Wellcome Trust Sanger Institute, United Kingdom). Copy number alterations were interpreted based on the copy number VisCap plots in log2 ratio values relative to the genome baseline.

Results

Clinicopathologic characteristics of peritoneal mesotheliomas

Figure 1 summarizes the clinicopathologic and molecular characteristics of 28 peritoneal mesotheliomas included in the study: 26 diffuse malignant peritoneal mesotheliomas (23 epithelioid and 3 biphasic histologies), one localized peritoneal mesothelioma, and one well-differentiated papillary mesothelioma with invasive foci.

Fig. 1: Landscape of clinicopathologic and molecular features of peritoneal mesotheliomas in this study.
figure 1

Heatmap illustrating the characteristics of 28 peritoneal mesotheliomas, including 26 diffuse malignant peritoneal mesotheliomas (DMPM; patients #1–26), one localized peritoneal mesothelioma (LPM; patient #L) and one well-differentiated papillary mesothelioma with invasive foci (WDPM-IF; patient #W). Patients #9, #11–12, #17–19, and #23–25 were previously published [40]. ND not determined.

The diffuse malignant peritoneal mesothelioma cohort included 21 surgical resections and 5 excisional biopsies from 16 women and 10 men, with a median age of 61 years (range 17–81 years). Two patients had received therapeutic radiation to the abdomen (one for ovarian cancer 28 years prior and one for non-Hodgkin lymphoma 37 years prior to the mesothelioma diagnosis). History of asbestos exposure was documented in eight (31%) patients. Pleural plaques were noted radiographically in 12 (46%) patients, including four with subsequent development of diffuse malignant pleural mesothelioma. Of the 24 diffuse malignant peritoneal mesothelioma patients with known clinical follow-up (0.3 to 12.7 [median 3.0] years), the median overall survival was 4.1 years. This cohort included 23 (88%) epithelioid mesotheliomas (Figs. 2a, 3a–c) and 3 (12%) biphasic mesotheliomas (Fig. 3d). The three patients with biphasic mesothelioma were two women and one man, aged 52, 68, and 73 years, with a median overall survival of 1.7 years (vs 5.9 years in those with epithelioid mesothelioma; p = 0.26). By immunohistochemistry, tumor cells expressed calretinin, WT1, and D2–40 in 22 of 22 (100%), 20 of 20 (100%), and 13 of 17 (76%) cases, respectively. Immunohistochemistry for BAP1 showed complete loss of expression in 13 (50%) tumors (Fig. 2b), including 12 (52%) epithelioid tumors and 1 (33%) biphasic tumor. By cytogenetics and/or fluorescence in situ hybridization testing, loss of chromosomal region 9p (CDKN2A) and 22q (NF2) was noted in 5 of 16 (31%) and 4 of 16 (25%) tumors, respectively, all of epithelioid histotype.

Fig. 2
figure 2

Diffuse malignant peritoneal mesothelioma with BAP1 alterations a Peritoneal mesothelioma (patient #7) with BAP1 nonsense mutation showed b complete loss of BAP1 expression by immunohistochemistry. c Schematics illustrating the locations of mutations in BAP1 identified in our study (patients #1–3, #5, #7, #13–15, #18; excluding rearrangements/deletion). UCH: ubiquitin carboxy-terminal hydroxylase domain. BA: BARD1-binding domain. BR: BRCA1-binding domain. H: HCF1-binding domain. NLS: nuclear localization sequence. d Copy number VisCap plot of peritoneal mesothelioma (patient #12) with two-copy loss of BAP1 at 3p21 (red arrow). e Copy number VisCap plot of peritoneal mesothelioma (patient #16) with one-copy loss of BAP1 at 3p21 (red arrow) and two-copy loss of CDKN2A at 9p21 (blue arrow).

Fig. 3
figure 3

Diffuse malignant peritoneal mesothelioma lacking BAP1 alterations a, b Histologic features of BAP1-wild-type peritoneal mesothelioma (patient #22) with TRAF7 activating mutation, characterized by tubules and nests of epithelioid tumor cells. c Histologic features of BAP1-wild-type peritoneal mesothelioma (patient #21) with SUZ12 inactivating mutation and epithelioid histology. d Histologic features of BAP1-wild-type peritoneal mesothelioma (patient #20) with TP53 nonsense mutation and biphasic histology, showing epithelioid (top) and sarcomatoid (bottom) components. e Copy number VisCap plot and f a plot of variant allele fraction in the targeted genome of this biphasic mesothelioma showed extensive loss of heterozygosity in all chromosomes except 5, 7, 16, and 20 (red arrows).

The localized peritoneal mesothelioma involved only the fallopian tube as a solitary pedunculated mass in a woman younger than 40 years, with no other masses identified in the peritoneum radiologically or intraoperatively. The patient was alive with no evidence of disease 13.2 years after the initial diagnosis. Histologically, the tumor was characterized by tubules of epithelioid tumor cells with variably prominent nucleoli (Supplementary Fig. 1A, B), indistinguishable from that of diffuse malignant peritoneal mesothelioma. By immunohistochemistry, tumor cells were diffusely positive for WT1 (Supplementary Fig. 1C) and calretinin (Supplementary Fig. 1D), negative for D2–40, and showed intact BAP1 expression (Supplementary Fig. 1E). By transmission electron microscopy, tumor cells demonstrated prominent intercellular junctions and microvilli with an average thickness of 75 nm and a length-to-width ratio of 13:1, characteristic of mesothelial differentiation (Supplementary Fig. 1F).

The well-differentiated papillary mesothelioma with invasive foci presented in a woman with recurrent well-differentiated papillary mesothelioma and multiple surgical debulking procedures, including most recently at 12.7 years after the initial diagnosis. Histologically, consistent with the description by Churg et al. [62], the well-differentiated papillary mesothelioma with invasive foci was characterized by papillae with central myxoid-to-edematous core covered by a single layer of epithelioid mesothelial cells (Supplementary Fig. 2A); the single-layer arrangement could be obscured by compression of the papillae (Supplementary Fig. 2B). Tumor cells were positive for WT1 (Supplementary Fig. 2C) and calretinin (Supplementary Fig. 2D), with intact BAP1 expression (Supplementary Fig. 2E). Immunohistochemistry for keratin cocktail AE1/AE3 (Supplementary Fig. 2F) highlighted the compressed papillae.

Molecular characteristics of peritoneal mesotheliomas

Among the 26 diffuse malignant peritoneal mesotheliomas, BAP1 alterations were identified in 18 (69%) tumors (Fig. 1), including 8 (31%) harboring more than one BAP1 alterations each. A total of 26 BAP1 alterations detected included five frameshift mutations, two nonsense mutations, two splice site mutations, one missense mutation (Fig. 2c and Supplementary Table 2), three structural rearrangements/deletion (deletion of 23 base-pairs, inactivating rearrangement with an intronic region in chromosome 9, and inactivating rearrangement with a pseudogene in chromosome 10), and 13 copy number loss (monoallelic in ten, biallelic in three; including six with concurrent BAP1 mutations). Complete loss of BAP1 protein expression was significantly associated with the presence of BAP1 genomic alterations (p = 0.001), and all tumors with complete loss of BAP1 expression by immunohistochemistry harbored BAP1 molecular alterations. The presence of BAP1 genomic alterations (single-nucleotide, copy number, and/or structural alterations) in patients with diffuse malignant peritoneal mesothelioma was associated with older age at the time of diagnosis and worse overall survival in the entire cohort with both epithelioid and biphasic histologies (Supplementary Table 3) and in those with epithelioid histology only (Supplementary Table 4). No significant associations were observed regarding the presence of BAP1 molecular alterations with the presence of pleural plaques or asbestos exposure, history of prior radiation, size of largest tumor nodule, histologic type, chromosomal 9p/22q status by fluorescence in situ hybridization, and treatment received in the entire cohort of diffuse malignant peritoneal mesothelioma patients (Supplementary Table 3) and in those with epithelioid histology only (Supplementary Table 4).

Recurrent alterations in other tumor suppressors and epigenetic regulators were noted in diffuse malignant peritoneal mesotheliomas (Fig. 1). PBRM1 alterations were significantly enriched in the BAP1-altered cohort (p = 0.001; Supplementary Table 5); though this association did not reach significance in tumors with epithelioid histology only (Table 2). Single-nucleotide variants and/or copy number loss of PBRM1, SETD2, NF2, TP53, and TSC1 were identified in 12 (52%), 9 (35%), 5 (19%), 4 (15%), and 3 (12%) tumors, respectively. Two-copy loss of BAP1 (Fig. 2d), CDKN2A/CDKN2B (Fig. 2e), PBRM1, SETD2, and NF2 was found in three (12%), three (12%), two (9%), one (4%), and one (4%) tumors, respectively. Copy number gains were noted involving WT1, KDR, TNFAIP3, NFKBIZ, NTRK1, and MDM4 in 3–13 (12–50%) tumors. Furthermore, DNA ligase 4 (LIG4) nonsense mutation (p.Y698*) and excision repair 6 (ERCC6) frameshift mutation (p.R1318Gfs*12) were identified in one diffuse malignant peritoneal mesothelioma each.

Table 2 Summary of molecular alterations in epithelioid-type diffuse malignant peritoneal mesothelioma.

Of the eight BAP1-wild-type diffuse malignant peritoneal mesotheliomas, one epithelioid mesothelioma (Fig. 3a, b) harbored a TRAF7 activating mutation (p.N520S), and one epithelioid mesothelioma (Fig. 3c) harbored a SUZ12 frameshift mutation (p.F474Ifs*13). Two tumors harbored pathogenic mutations in TP53 (p.R196* and p.H214L), including one with concurrent alterations in both NF2 and TSC1. The remainder TP53-mutant biphasic mesothelioma (Fig. 3d) showed evidence of genomic near-haploidization with extensive loss of heterozygosity except chromosomes 5 and 20 and parts of chromosomes 7 and 16 (Fig. 3e, f). Three BAP1-wild-type diffuse malignant peritoneal mesotheliomas harbored ALK gene rearrangements as previously reported [40]. No reportable variants were noted in the remainder BAP1-wild-type diffuse malignant peritoneal mesothelioma. BAP1-altered and BAP1-wild-type diffuse malignant peritoneal mesotheliomas appeared to be histologically similar (Figs. 2a, 3a–d).

The molecular features of the epithelioid diffuse malignant peritoneal mesotheliomas were summarized in Table 2. The three biphasic diffuse malignant peritoneal mesotheliomas included one with two-copy loss of BAP1 and SETD2 and one-copy loss of PBRM1 and NF2 (patient #11), one TP53-mutant with evidence of genomic near-haploidization as aforementioned (patient #20; Fig. 3d–f), and one with TPM1-ALK fusion as previously published (patient #23) [40].

Both the localized peritoneal mesothelioma and the well-differentiated papillary mesothelioma lacked alterations typical of diffuse malignant peritoneal mesothelioma (Fig. 1). The localized peritoneal mesothelioma (Supplementary Fig. 1) harbored a CHEK2 nonsense mutation (p.R137*) (Supplementary Table 2). The well-differentiated papillary mesothelioma with invasive foci (Supplementary Fig. 2) harbored one-copy loss of PRDM1 with a low-copy gain of WT1 and TNFAIP3, but otherwise demonstrated no reportable single-nucleotide variants in this study.

Discussion

Our findings underscored the genetic heterogeneity of diffuse malignant peritoneal mesotheliomas and implicated DNA repair, epigenetics, and cell cycle regulation in the pathogenesis of malignant peritoneal mesothelioma. Diffuse malignant peritoneal mesotheliomas harbored recurrent alterations, most commonly BAP1 mutations, followed by mutations or copy number loss in other DNA repair, chromatin, and cell cycle regulators: PBRM1, SETD2, NF2, ARID1B, PRDM1, and CDKN2A, among others. In addition to the BAP1-mutant diffuse malignant peritoneal mesotheliomas, we uncovered the genomic alterations of BAP1-wild-type tumors, including TP53 mutations, TRAF7 activating mutation, SUZ12 inactivating mutation, ALK rearrangements we previously described [40], and one tumor with genomic near-haploidization. Furthermore, while the limited number of biphasic diffuse malignant peritoneal mesotheliomas examined herein precludes definitive evaluation of their molecular differences from epithelioid peritoneal mesotheliomas, this study illustrates the genetic heterogeneity of biphasic peritoneal mesotheliomas, with two-copy loss of BAP1 and SETD2, TP53 mutation with genomic near-haploidization, and ALK rearrangement described in one tumor each. The case of localized peritoneal mesothelioma harbored a nonsense CHEK2 mutation, whereas the case of well-differentiated papillary mesothelioma with invasive foci harbored no reportable variants; nonetheless, given the exploratory nature of our analysis using single cases, definitive conclusions for molecular differences between diffuse malignant peritoneal mesothelioma and other mesothelial lesions require evaluation using larger cohorts.

This single institutional study represents the largest next-generation sequencing series of diffuse malignant peritoneal mesothelioma to date (Table 1). Overall, certain genetic alterations have been noted in both pleural and peritoneal mesotheliomas with similar prevalence, such as BAP1 mutations in 50–70% of cases and NF2 loss in 20–40% of cases [12, 21, 23, 45, 50,51,52,53,54, 63], though other genetic alterations differ between peritoneal and pleural mesotheliomas. Loss of CDKN2A appears less frequent in peritoneal mesotheliomas (20–50%) as compared with pleural mesotheliomas (60–70%) [12, 53, 63,64,65,66]; this may be related to the stronger association between CDKN2A loss and sarcomatoid mesothelioma, preferentially seen in the pleura [23, 59], while sarcomatoid peritoneal mesothelioma is extremely rare [67]. Furthermore, rare diffuse malignant mesotheliomas with EWSR1-ATF1 and FUS-ATF1 fusions are primarily peritoneal in location [36]. ALK rearrangements characteristic of a small subset of peritoneal mesotheliomas [38, 40, 44, 56, 57] have not been identified in pleural mesotheliomas examined to date [23, 40]. The discovery of ALK rearrangements raises the possibility of treatment with ALK inhibitors [44].

Alterations in BAP1 are a consistent finding across studies of diffuse malignant peritoneal mesothelioma, including our own. BAP1 mutations, copy number loss, and expression loss were noted in 69%, 77%, and 85% of tumors, respectively, in Joseph et al. [37]; 32%, 42%, and 57% of tumors, respectively, in Leblay et al. [54]; and 28%, 44%, and 44% of tumors, respectively, in Shrestha et al. [45]. Our identification of BAP1 single-nucleotide/structural variants, copy number loss, and complete expression loss in 42%, 50%, and 50% of cases appeared more in line with the studies by Leblay et al. and Shrestha et al. Patients with diffuse malignant peritoneal mesothelioma that harbored BAP1 mutation(s) were noted to show better outcome in Leblay et al. [54] but worse overall survival in our study. This discrepancy between Leblay et al. and our study may be due to differences in the cohort baseline characteristics, such as the percentage of patients with BAP1-altered tumors being women (21% vs 50%) and the age of patients with BAP1-wild-type tumors (median > 60 vs 52). Differences in the prevalence of the somatic vs germline BAP1 mutations may also contribute, as prolonged survival had been described in mesothelioma patients with germline BAP1 mutations [68]. Germline BAP1 mutation status was not available in our study and may explain the different results from Leblay et al. [54].

The identification of mutations in DNA Ligase 4 (LIG4) and excision repair 6 (ERCC6) in diffuse malignant peritoneal mesothelioma and CHEK2 mutation in a localized peritoneal mesothelioma implicates defective DNA repair in their pathogenesis. Some of these alterations harbor allelic frequencies close to 0.5, raising a possibility that these represent germline alterations; however, definitive interpretation was precluded by our current assay that analyzed tumor samples only. In a study using blood samples from 198 patients with diffuse malignant mesothelioma, germline mutations involving BAP1, BRCA2, CHEK2, ATM, VHL, and others were identified in 7% of pleural mesothelioma patients and 25% of peritoneal mesothelioma patients [20]. Malignant peritoneal mesothelioma had been reported in an infant with germline ATM mutation [69]. Diffuse malignant peritoneal mesothelioma had also been described in a young patient with neurofibromatosis harboring germline NF2 mutation, with tumor emergence after a “second-hit” somatic inactivation [47]. Collectively, these findings implicate germline and/or somatic mutations in the DNA repair pathway in the development of peritoneal mesotheliomas, including some cases of the BAP1 wild-type tumors. Identification of DNA repair alterations raises the possibility of treatment with poly ADP-ribose polymerase inhibitors: for example, a clinical trial of using niraparib in tumors including mesothelioma with aberrant DNA repair is ongoing (NCT03207347).

In BAP1-wild-type peritoneal mesotheliomas, we identified mutations in SUZ12, TRAF7, and evidence of genomic near-haploidization in one tumor each. Given the small number of cases, we could not address the question if these alterations were mutually exclusive of BAP1 alterations in diffuse malignant peritoneal mesothelioma. Recurrent mutations in SUZ12, a component of the Polycomb Repressive Complex 2 (PRC2), lead to loss of downstream Histone 3 lysine 27 trimethylation (H3K27Me3) [70]. Identification of the SUZ12 mutation in a diffuse malignant peritoneal mesothelioma suggests the importance of chromatin regulation, though seemingly contrary to the prevailing model of targeting PRC2 complex as a therapeutic strategy: BAP1 expression loss with EZH2 overexpression has been noted in 50–70% of malignant mesotheliomas [71, 72], and BAP1 mutations have been implicated in EZH2-dependent transformation in malignant mesothelioma [73]. In contrast, our finding of SUZ12 mutation in a BAP1-wild-type peritoneal mesothelioma implies that perturbed PRC2 function in either direction—and not necessarily PRC2 activation alone—may be oncogenic in the development of diffuse malignant peritoneal mesothelioma, consistent with published observations in other tumor types [74].

Recurrent activating mutations in TRAF7, a component of the nuclear factor kappa B (NF-kB) signaling pathway, have been described in adenomatoid tumors of genital type [75, 76], a subset of well-differentiated papillary mesothelioma of the peritoneum [77], rare diffuse malignant pleural [23] and peritoneal mesotheliomas [25], and a subset of localized pleural mesothelioma [78]. In particular, the TRAF7 p.N520S mutation noted in one diffuse malignant peritoneal mesothelioma herein has been reported previously in two well-differentiated papillary mesotheliomas [77]. The presence of TRAF7 mutation alone is thus not entirely specific in the distinction between benign and malignant tumors of mesothelial origin. To date, TRAF7-mutant malignant pleural and peritoneal mesotheliomas [23, 25] including in this study all lack BAP1 mutations, suggesting TRAF7 activation as one of the BAP1-independent mechanisms for the pathogenesis of mesotheliomas and related lesions.

Genomic near-haploidization, characterized by extensive loss of heterozygosity (or copy-number-neutral uniparental disomy secondary to genome endo-reduplication), has been described in 3% of diffuse malignant pleural mesothelioma in the TCGA/ICGC cohorts [26], one localized pleural mesothelioma [78], and one diffuse malignant peritoneal mesothelioma [79]. Notably, chromosomes 5 and 7 were often retained in these near-haploid mesothelial tumors [26, 78, 79], including the TP53-mutant biphasic peritoneal mesothelioma described herein; the reason for such preferential chromosomal retention remained unknown.

In conclusion, diffuse malignant peritoneal mesotheliomas harbor recurrent alterations in BAP1 and other DNA repair, chromatin, and cell cycle regulators, with one tumor showing genomic near-haploidization. Our findings illustrate the genetic diversity of diffuse malignant peritoneal mesothelioma, with implications on their diagnosis and selection of potential therapeutic targets.