Molecular and Cellular Pathobiology
`
`Cancer
`Research
`
`Gender-Specific Molecular and Clinical Features
`Underlie Malignant Pleural Mesothelioma
`Assunta De Rienzo1, Michael A. Archer1, Beow Y.Yeap2, Nhien Dao1, Daniele Sciaranghella1,
`Antonios C. Sideris1,Yifan Zheng1, Alexander G. Holman3,Yaoyu E.Wang3, Paola S. Dal Cin4,
`Jonathan A. Fletcher4, Renee Rubio3, Larry Croft5, John Quackenbush3,
`Peter E. Sugarbaker1, Kiara J. Munir1, Jesse R. Battilana1, Corinne E. Gustafson1,
`Lucian R. Chirieac4, Soo Meng Ching5, James Wong5, Liang Chung Tay5, Stephen Rudd5,
`Robert Hercus5, David J. Sugarbaker6, William G. Richards1, and Raphael Bueno1
`
`Downloaded from http://aacrjournals.org/cancerres/article-pdf/76/2/319/2741046/319.pdf by guest on 29 September 2023
`
`Abstract
`
`Malignant pleural mesothelioma (MPM) is an aggressive
`cancer that occurs more frequently in men, but is associated
`with longer survival in women. Insight into the survival advan-
`tage of female patients may advance the molecular understand-
`ing of MPM and identify therapeutic interventions that will
`improve the prognosis for all MPM patients. In this study, we
`performed whole-genome sequencing of
`tumor specimens
`from 10 MPM patients and matched control samples to identify
`potential driver mutations underlying MPM. We identified
`molecular differences associated with gender and histology.
`Specifically, single-nucleotide variants of BAP1 were observed
`in 21% of cases, with lower mutation rates observed in sarco-
`matoid MPM (P < 0.001). Chromosome 22q loss was more
`frequently associated with the epithelioid than that nonepithe-
`
`liod histology (P ¼ 0.037), whereas CDKN2A deletions
`occurred more frequently in nonepithelioid subtypes among
`men (P ¼ 0.021) and were correlated with shorter overall
`survival for the entire cohort (P ¼ 0.002) and for men (P ¼
`0.012). Furthermore, women were more likely to harbor TP53
`mutations (P ¼ 0.004). Novel mutations were found in genes
`associated with the integrin-linked kinase pathway, including
`MYH9 and RHOA. Moreover, expression levels of BAP1, MYH9,
`and RHOA were significantly higher in nonepithelioid tumors,
`and were associated with significant reduction in survival of the
`entire cohort and across gender subgroups. Collectively, our
`findings indicate that diverse mechanisms highly related to
`gender and histology appear to drive MPM. Cancer Res; 76(2);
`319–28. Ó2015 AACR.
`
`Introduction
`Malignant pleural mesothelioma (MPM) is a rare, aggressive
`cancer, associated with prior exposure to asbestos (1). Approxi-
`mately 3,200 new cases are diagnosed in United States annually,
`and the incidence worldwide is estimated to rise during the next
`two decades (2). Prognosis is poor for most patients with a
`
`1The Thoracic Surgery Oncology laboratory and the International
`Mesothelioma Program, Division of Thoracic Surgery and the Lung
`Center, Brigham and Women's Hospital, and Harvard Medical School,
`Boston, Massachusetts. 2Department of Medicine, Massachusetts
`General Hospital and Harvard Medical School, Boston, Massachusetts.
`3Center for Cancer Computational Biology, Department of Biostatis-
`tics and Computational Biology, Dana-Farber Cancer Institute, and
`Harvard School of Public Health, Boston, Massachusetts. 4Depart-
`ments of Pathology, Brigham and Women's Hospital, and Harvard
`5Malaysian Genomics
`Medical School, Boston, Massachusetts.
`Resource Centre, Kuala Lumpur, Malaysia. 6Debakey Department of
`Surgery, Baylor College of Medicine, Houston, Texas.
`
`Note: Supplementary data for this article are available at Cancer Research
`Online (http://cancerres.aacrjournals.org/).
`
`M.A. Archer and B.Y. Yeap contributed equally to this article.
`
`Corresponding Author: Assunta De Rienzo, Division of Thoracic Surgery,
`Brigham and Women's Hospital 75 Francis Street, Boston, MA 02115. Phone:
`(617) 732-6526; Fax: (617) 566-3441; E-mail: aderienzo@partners.org.
`
`doi: 10.1158/0008-5472.CAN-15-0751
`Ó2015 American Association for Cancer Research.
`
`reported median survival between 4 and 12 months because
`there are few effective systemic treatments for this malignancy
`(3). Three major histologic types are recognized in MPM: epithe-
`lioid, sarcomatoid, and biphasic (4). The incidence of MPM is
`higher in men than women, likely because of the more common
`occupational asbestos exposure in men, whereas women are
`usually subject to secondary exposure through spouses' clothing,
`low-level environmental exposure, and other sources (5, 6). A
`recent study suggests that female patients with MPM live signif-
`icantly longer compared with men even after adjustment for
`potential confounders such as age, stage, and treatment (5).
`High-throughput sequencing techniques have revealed that
`each cancer genome appears to have a unique genetic profile
`acquired through cumulative genetic alterations, including driver
`mutations, that may confer advantageous survival phenotypes on
`the cancerous cells and represent precise therapeutic targets (7).
`To date, the genetic landscape of MPM has been primarily
`described in terms of chromosomal rearrangement events inves-
`tigated mostly in epithelioid and biphasic human specimens
`and cell lines (8). The cyclin-dependent kinase inhibitor 2A
`(CDKN2A),
`the neurofibromatosis type 2 (NF2), and the
`BRCA1-associated protein-1 (BAP1) genes have been shown to
`be the most frequently mutated tumor suppressor genes in MPM
`(9–12). Recently, two independent investigations have identified
`BAP1 as frequently mutated in the chromosomal region 3p21.1 in
`sporadic and familial MPM, indicating the importance of genetic
`factors in MPM susceptibility (10, 12, 13). Although BAP1
`
`www.aacrjournals.org
`
`319
`
`Personalis EX2157
`
`

`

`Downloaded from http://aacrjournals.org/cancerres/article-pdf/76/2/319/2741046/319.pdf by guest on 29 September 2023
`
`De Rienzo et al.
`
`mutations have been associated with age (10) and with the
`epithelioid subtype (14), no other significant correlation of these
`genes to demographic, clinical, or pathologic variables has so far
`been identified.
`In the current study, a comprehensive genome-wide approach
`was applied to discover novel genetic alterations in MPM. Whole-
`genome sequencing data of tumor specimens from 10 MPM
`patients and their matched normal tissue were analyzed to iden-
`tify novel somatic point mutations. A bioinformatic approach was
`employed to prioritize genes to further analyze by focused
`sequencing in close to three hundred additional MPM tumor
`samples.
`
`Materials and Methods
`Specimens and cell lines
`All tumor and normal lung specimens were collected at surgery
`with patient consent, freshly frozen, stored, and annotated by the
`institutional tumor bank with Institutional Review Board approv-
`al at Brigham and Women's Hospital (BWH; Boston, MA). High
`quality, tumor enriched (15) samples were prepared as previously
`described (16). Tumor cell enrichment (>83% for whole-genome
`sequencing experiments; >40% for targeted resequencing experi-
`ments) was confirmed by reviewing hematoxylin and eosin
`(H&E)-stained frozen sections. Eleven mesothelioma cell lines
`(H2052, MSTO-211H, H2452, H28, and the epithelial virus
`transformed MET5A from the (ATCC), JMN and JMN1B from
`Cell Culture Core (BWH), and MESO257, MESO428, MESO589,
`MS924 from the laboratory of J.A. Fletcher] were cultured in RPMI
`(Invitrogen Corporation) containing 10% FBS (ATCC) and main-
`
`tained at 37
`C in 5% CO2. Cells were collected at 70% confluence
`and frozen for DNA extraction or placed in RNA extraction
`buffer (TRIzol reagent, Invitrogen) for RNA extraction. DNA was
`isolated using the DNeasy Tissue Kit (Qiagen) and RNA using the
`TRIzol (Thermo Fisher) method in combination with RNeasy kit
`(Qiagen). RNase or DNase I (Qiagen) treatments were conducted
`according to the manufacturer's instructions. Matched normal
`DNA was prepared from peripheral blood or lung tissue. Nucleic
`acids were quantified using an ND-1000 spectrophotometer
`(Thermo Fisher). The integrity of the DNA and RNA were deter-
`mined using Qubit 2.0 Fluorometer (Thermo Fisher) and Agilent
`2100 Bioanalyzer (Agilent), respectively.
`
`Whole-genome sequencing
`Whole-genome sequencing of 10 tumor and 10 matched
`normal genomic DNA (obtained from blood) samples (3 mg)
`was accomplished using a Complete Genomics platform (17).
`Mean coverage of tumor and normal genomes were 187 and
`188, respectively (Supplementary Table S1). All the pair-end
`reads were filtered on the basis of average read quality score  20.
`Candidate somatic mutations were detected using the SynaAlign
`program (Synamatix). Briefly, sequencing reads were first aligned
`to the human reference genome (hg19) using short read mapper
`SXmapper (Synamatix) with default parameter settings. The
`numbers of reads containing single nucleotide variants (SNV)
`and indels in both tumor and germline samples were counted and
`the quality score of each base and the read direction information
`were collected. This information and the mapping statistics for
`each read across the genome (reads in repeat regions etc.) were
`used to call SNVs and indels. The null hypothesis of equal allele
`frequencies in tumor and germline samples was tested using the
`
`two-tailed Fisher exact test on quality-filtered SNVs and indels.
`Data were also collected for each sample on larger scale structural
`variations, loss of heterozygosity, and copy number variations.
`
`Validation of SNV by Sanger sequencing
`Selected candidate SNVs were further characterized using PCR
`and conventional Sanger sequencing to confirm tumor-specific
`single-nucleotide mutations. Gene-specific primer pairs were used
`to PCR amplify genomic DNA from tumor and normal specimens
`(Supplementary Table S2) as previously described (16). For TP53
`analysis, primers were chosen in the flanking intronic region of
`each exon. Twenty-two samples were analyzed using both rese-
`quencing and Sanger methods. Discordant results were found in
`six cases, probably due to tissue heterogeneity. In discordant cases,
`the mutation was called as present for each sample.
`
`Karyotype and FISH analysis
`Metaphase chromosome spreads from diagnostic MPM sam-
`ples were prepared and G-banded according to standard proce-
`dures as part of the routine BWH pathologic evaluation of all
`MPM cases. Additional FISH analyses were performed in those
`samples either with no metaphases or normal karyotype in the
`context of pathologic mesothelioma positive report. FISH analysis
`was performed using the Vysis LSICDKN2A/CEP9 Dual Color
`Probe Set (Abbott Molecular) for the CDKN2A locus at 9p21,
`CEP9 at the chromosome 9 centromeric region, and Vysis TUPLE
`1/LSI ARSA Dual Color Probe Set (Abbott) for TUPLE 1 at 22q11.2
`and ARSA at 22q13. Observed aberration in  2% of observed
`nuclei (minimum 50 per case) was reported as consistent with a
`clonal, neoplastic, and cellular proliferation.
`
`Additional dual-color FISH analysis
`For a subset of cases, dual-color FISH analysis was performed
`on frozen sections (5 mm) to assess two distinct regions of
`chromosome 22. The sections were fixed at room temperature
`for two hours in freshly prepared 3:1 methanol:acetic acid, then
`air-dried.
`Bacterial artificial chromosome (BAC) DNA probes (CHORI,
`BAC PAC Resource) containing most or all of the coding sequence
`for NF2 (RP11-155B12, 159 kb) and MYH9 (RP11-105I18,
`161 kb) were selected and labeled by nick translation (Abbott),
`NF2 with Orange dUTP, and MYH9 with Green dUTP. Slides were
`pretreated with pepsin (Digest-All 3, Invitrogen) 5 to 15 minutes
`
`at 37
`C, rinsed in PBS for 5 minutes at room temperature, post-
`fixed 1 minute in 10% formalin, rinsed in PBS, dehydrated in
`ethanol series, and air-dried. FISH procedure followed standard
`protocols. Scoring was performed by two technologists, each
`scoring Orange (O, NF2) and Green (G, MYH9) signals present
`on 50 nuclei, for 100 nuclei total per specimen in an area
`determined by a pathologist in parallel H&E-stained sections to
`have a minimum tumor involvement of 40%. Negative controls
`were a normal lymphoblast line with metaphases to confirm
`localization, and sections from four tumors with no histologically
`apparent tumor in the scored areas. Each deletion signal pattern
`(1O1G, 1O2G, 2O1G) was determined to be present if its fre-
`quency exceeded the mean plus 3 SDs of the same pattern in the
`normal sections.
`
`Target capture, sequencing, and data analysis
`Library preparation, custom exome capture, and next-genera-
`tion sequencing were performed at
`the Center for Cancer
`
`320
`
`Cancer Res; 76(2) January 15, 2016
`
`Cancer Research
`
`Personalis EX2157
`
`

`

`Downloaded from http://aacrjournals.org/cancerres/article-pdf/76/2/319/2741046/319.pdf by guest on 29 September 2023
`
`Computational Biology. Library construction and custom exome
`capture were performed using the SureSelect XT2 for Illumina
`protocol (Agilent) according to the manufacturer's instructions
`Libraries were combined to 3 library pools, diluted to a 2 nmol/L
`working stock and sequenced at a final concentration of 12 pmol/
`L on a paired end flowcell with 50 cycles in each direction.
`Sequencing was done on a HiSeq 2000 (Illumina) according to
`the manufacturer's protocols. Sequencing reads were demulti-
`plexed using the Illumina CASAVA package and aligned to the
`hg19 assembly of the human genome using BWA aln with default
`parameters and a mismatch penalty of 1 (18). SNV and indel
`discovery was performed using HaplotypeCaller within the GATK
`package with default parameters (19). Variants with a minimum
`phred-scaled confidence threshold under 10 were excluded, and
`those under 30 were flagged as low quality. Resulting variant calls
`were annotated for potential genetic impact using the SnpEff
`package (20).
`
`Microarray analysis
`Expression levels of specific genes were explored using micro-
`array data from a parallel project. Epithelioid samples (N ¼ 129)
`were selected from the International Mesothelioma Program
`Tumor Bank (Supplementary Fig. S1). This sample set consisted
`of male and female samples matched by age and nodal status
`(Coleman MH, and Bueno R; personal communication). In
`addition, 19 sarcomatoid and 3 biphasic MPM samples were
`included in the analysis. To determine the levels of transcripts in
`each sample, 0.25 mg of total RNA was amplified using the
`Ambion WT Expression Kit (Thermo Fisher). The cRNA was
`hybridized to Affymetrix Human Gene 1.1 ST Array (Thermo
`Fisher), subsequently labeled with GeneChip WT Terminal Label-
`ing Kit (Thermo Fisher), and scanned with a GeneAtlas Worksta-
`tion (Thermo Fisher). Hybridization, washing, staining, and
`scanner procedures were performed per manufacturer's recom-
`mendations. For quality control across platforms, two MAQC
`samples were included in the analysis (21). A blind control was
`also added to check the variability of the expression values across
`the chips. The probe intensity distribution was examined for
`quality control and outliers were removed. Expression levels for
`BAP1, NF2, TP53, MYH9, MYH6, MYH10, PIK3C2A, RHOA, and
`TNFRSF1A for
`this cohort are included in Supplementary
`Table S3.
`
`Statistical analysis of mutation status and gene expression
`versus clinicopathologic variables
`Asbestos exposure was analyzed as a binary covariate with
`background exposure defined as asbestos body counts in lung
`
`Novel Molecular and Gender-Specific Features in Mesothelioma
`
`tissue  50 fibers per gram of wet lung tissue. Fisher exact test
`was used to compare histology, gender, and asbestos exposure
`between mutation and wild-type groups, whereas Wilcoxon
`rank sum test was used to analyze age as a continuous covariate.
`Overall survival was defined from the date of definitive surgery
`until the date of death or was censored at the date of last follow-
`up for patients who had not died at their latest contact. Overall
`survival was estimated by the Kaplan–Meier method, with
`group differences assessed by the log-rank test. The normalized
`expression level of each candidate gene was initially grouped
`into quartiles for exploratory survival analysis. Patient sub-
`groups with comparable survival were combined for further
`analysis to estimate the meaningful differences. Proportional
`hazards regression was used to estimate the HR for reporting a
`survival difference, including adjusting for the gender effect. All
`P values are based on a two-sided hypothesis, with P < 0.01
`used to identify potentially significant results to be conservative
`on false positives. Data analysis was performed using SAS 9.4
`(SAS Institute).
`
`Results
`Experimental subjects
`Ten MPM tumor and matched normal DNA samples were
`analyzed by whole-genome sequencing (Table 1). Matched MPM
`and normal DNA from 283 additional patients was subjected
`to focused sequencing (Table 2).
`
`Tumor somatic DNA variants
`Point mutations, small (<30 bp) structural variations, and
`copy number alterations were discovered in 10 pairs of whole-
`genome sequencing MPM tumor and germline DNA samples
`using computational approaches. Single nucleotide variants
`(N ¼ 146; SNVs) mapping in the amino acid coding regions
`of annotated exons and generating nonsynonymous amino
`acid changes were further resequenced in the tumor and normal
`DNA by Sanger methodology. Eighty-five were confirmed to
`be tumor-specific (Supplementary Table S2). Each patient's
`sample pair displayed a unique small number of point muta-
`tions ranging from three to 20 per tumor. Seventy-eight SNVs
`were missense and seven were nonsense mutations. TP53 was
`found mutated in 2 of 10 tumors (20%). In both cases, the
`tumors came from female patients with asbestos body counts in
`the background range as measured in resected lung tissue (data
`not shown). Three of the mutated genes (RNF43, TP53 and
`MYH9) were listed in the Catalogue Of Somatic Mutation In
`Cancer (COSMIC, Release v69) database (22).
`
`Table 1. Clinical and molecular characteristics of the 10 MPM samples included in the whole-genome sequencing analysis
`Tumor sample
`Histology
`Gender
`Asbestos body counts
`CG3
`Biphasic
`M
`N/A
`CG5
`Biphasic
`M
`87
`CG4
`Epithelial
`M
`25
`CG10
`Biphasic
`M
`N/A
`CG9
`Epithelial
`M
`934
`CG12
`Biphasic
`M
`1,382
`CG14
`Epithelial
`F
`6
`CG16
`Epithelial
`F
`3
`CG18
`Epithelial
`F
`0
`CG20
`Epithelial
`M
`25
`Abbreviation: N/A, not available.
`
`9p (FISH)
`N/A
`Abnormal
`Abnormal
`Normal
`Normal
`Normal
`Abnormal
`Abnormal
`Normal
`Normal
`
`22q (FISH)
`N/A
`Abnormal
`N/A
`Normal
`Abnormal
`Abnormal
`Abnormal
`Abnormal
`N/A
`Abnormal
`
`www.aacrjournals.org
`
`Cancer Res; 76(2) January 15, 2016
`
`321
`
`Personalis EX2157
`
`

`

`De Rienzo et al.
`
`Table 2. Clinical and histopathologic characteristics of patients in mutational
`and expression analyses
`
`Mutational
`283
`34 (12%)
`
`45 (2–106)
`
`61 (17–84)
`
`195 (69%)
`88 (31%)
`
`Evaluable for analysis
`Alive at last follow-up
`Follow-up from surgery, months
`Median (range)
`Age, years
`Median (range)
`Gender
`Male
`Female
`Histologic subtype
`205 (72%)
`Epithelioid
`52 (18%)
`Biphasic
`25 (9%)
`Sarcomatoid
`1 (<1%)
`Desmoplastic
`Asbestos body counts per gram wet lung tissuea
`50 fibers
`31 (37%)
`67 (45%)
`>50 fibers
`53 (63%)
`81 (55%)
`aAsbestos data missing for 135 patients in mutational analysis and for 67 patients
`in expression analysis.
`
`Expression
`151
`6 (4%)
`
`59 (4–99)
`
`59 (17–75)
`
`85 (56%)
`66 (44%)
`
`129 (85%)
`3 (2%)
`19 (13%)
`0
`
`Downloaded from http://aacrjournals.org/cancerres/article-pdf/76/2/319/2741046/319.pdf by guest on 29 September 2023
`
`correlation was found between mutation in BAP1 and gender,
`exposure to asbestos, or survival. BAP1 mutations were associated
`with marginally older age compared with wild-type tumors
`(median 67 vs. 64 years; P ¼ 0.050). BAP1 mutations were also
`observed in cell lines H2452 and H28. Table 3 describes the cases
`carrying BAP1 germline mutations.
`NF2 point mutations occurred in 21 (14%) of 147 analyzed
`samples (Supplementary Table S5A). In addition, karyotype, and/
`or FISH analyses available for 133 of these 147 samples showed
`that 75 samples (56%) had loss of chromosome 22q. Eleven of
`133 (8%) had both NF2 point mutation and 22q deletion.
`Compared with the cases with two copies of 22q, the deletion
`of 22q was more frequently associated with the epithelioid than
`nonepithelioid histology (P ¼ 0.037). The difference was greater
`in women but did not reach strict statistical significance due to low
`power (P¼ 0.141). Mutations in NF2 were not correlated with any
`of the other clinical or pathologic variables examined. NF2
`mutation was found in H2052.
`Karyotyping and/or FISH analyses of CDKN2A were available
`for 133 of 147 samples. Eighty-two (62%) of 133 samples had
`deletion of the CDKN2A region on 9p (determined by karyotyp-
`ing) and/or confirmed CDKN2A deletion (as determined by FISH
`with a CDKN2A probe). Forty-one of 74 (55%) epithelioid, 29 of
`41 (71%) biphasic, 11 of 17 (65%) sarcomatoid, and one of one
`desmoplastic samples had CDKN2A deletion. Compared with the
`cases with two copies of CDKN2A, cases with CDKN2A deletion
`appeared to be more frequently associated with the nonepithe-
`lioid histology (P ¼ 0.117). When this analysis was repeated for
`gender, the association between histology and CDKN2A loss was
`statistically significant only for the men (P ¼ 0.021). Interestingly,
`deletions in CDKN2A were correlated with shorter overall survival
`in the whole cohort (P¼ 0.002), but the difference was statistically
`significant only in the men (P ¼ 0.012).
`TP53 was mutated in 22 cases (15%) in this sample set
`(Supplementary Table S5B). Three mutations (D49N, C176F,
`A86T) were present in both the tumor and the normal DNA
`(Table 3). Notably, one sample showed only one germline TP53
`mutation (D49N) and no mutation in any of the other genes
`investigated herein. None of the cell lines examined exhibited
`TP53 mutations.
`No difference was observed in overall survival between the
`group of 62 patients with point mutation in at least one of the
`established MPM-related genes (BAP1, NF2, TP53) and the group
`wild-type for these genes (P ¼ 0.646).
`Forty SNVs in 31 samples (21%) occurred in ILK pathway–
`associated genes (TNFRSF1A, PIK3C2A, MYH10, MYH6, MYH9,
`and RHOA; Supplementary Table S5C), of which 35 (88%) were
`germline. Fifteen patients (43%) carrying a germline mutation in
`one or more of these genes had familial history of cancer. Only
`one SNV, in TNFRSF1A, is known to occur frequently (>1%–2%)
`in the general population (23). This SNV occurred in 5 patients
`and was predicted to have a "probably damaging" effect
`(PolyPhen: http://genetics.bwh.harvard.edu/pph2/) on the
`corresponding protein. Seven germline mutations were identified
`in PIK3C2A. Five were predicted to be benign and two "possibly
`damaging", one of which resides in the PI3K C2 domain. Twenty-
`six SNVs resided in three myosin heavy chain genes. Two samples
`displayed germline mutation of MYH10. Although functional
`analyses predict a benign effect, these mutations occur near
`mutations identified in other cancers (24). Nine germline muta-
`tions were detected in MYH6, two of which had a predicted
`
`Functional analysis of point mutations
`To identify clusters of genes associated with known pathways,
`functional enrichments of genes affected by point mutations were
`performed utilizing Ingenuity Pathway Analysis (Qiagen) using
`Biological Process Gene Ontology terms and canonical pathways.
`Among 216 canonical pathways identified, 51 were significantly
`enriched (P < 0.05) for the mutated genes (Supplementary
`Table S4). Mutations affecting genes involved in the integrin-
`linked kinase (ILK) pathway were the most significantly (P ¼
`4.9e5) enriched. Specifically, 5 of the 10 sequenced MPM
`samples showed point mutations in at least one of 6 identified
`genes associated with this pathway (MYH9, MYH6, MYH10,
`PIK3C2A, RHOA, and TNFRSF1A; Fig. 1A).
`
`Validation of the candidate driver genes in an independent
`cohort
`To further assess the prevalence of mutations observed herein
`and in previous studies (9–12), a panel of 147 MPM tumors
`(including 81 epithelioid, 41 biphasic, 24 sarcomatoid, and 1
`desmoplastic) and 11 previously established MPM cell lines were
`further analyzed using targeted resequencing of BAP1, NF2, TP53,
`MYH9, MYH6, MYH10, PIK3C2A, RHOA, and TNFRSF1A. This
`cohort included tumor samples from women (n ¼ 30) to evaluate
`the frequency of mutations with respect to gender. In addition,
`clinical karyotype analysis and FISH data for CDKN2A and
`22q were available for most of the patients. Targeted resequen-
`cing revealed 116 nonsynonymous SNVs and/or INDELs (Sup-
`plementary Table S5A–S5C). Seven of these were confirmed to
`be single-nucleotide polymorphisms occurring at high frequen-
`cy (>1%–2%) in the normal population. The remaining 109
`occurring in 86 samples were further examined in normal and
`tumor DNA pairs by Sanger sequencing. One or more somatic
`mutations were confirmed in 81 of 147 samples (55%).
`BAP1 was mutated in 31 of 147 (21%) samples (Supplemen-
`tary Table S5A). Twenty-six tumor-specific variants and 5 germline
`variants (present in tumor and normal matching DNA) were
`observed. In six samples, multiple BAP1 mutations were identi-
`fied. The mutations occurred mostly in epithelioid (23%) and
`biphasic (24%) samples, whereas only two of 24 sarcomatoid
`(8%) samples had tumor-specific BAP1 mutations. No significant
`
`322
`
`Cancer Res; 76(2) January 15, 2016
`
`Cancer Research
`
`Personalis EX2157
`
`

`

`Novel Molecular and Gender-Specific Features in Mesothelioma
`
`Downloaded from http://aacrjournals.org/cancerres/article-pdf/76/2/319/2741046/319.pdf by guest on 29 September 2023
`
`Figure 1.
`A, Ingenuity Pathway Analysis was used to connect a subset of six genes (MYH9, MYH6, MYH10, PIK3C2A, RHOA, and TNFRSF1A), with tumor specific mutations in
`five MPM samples, mapping in the integrin-linked kinase pathway. Four genes (UBC, MDM2, CLTC, and SRC) were added to show the indirect relationship. The
`connectivity map was generated from available published data (Ingenuity Systems). Each symbol for gene denotes the function of the interacting protein.
`Symbols representing specific categories of cellular molecules as well as interactive relationships are depicted in the legend. B, schematic representation of the SNVs
`identified in MYH9, MYH10, MYH6, and RHOA. Nucleotide-binding domains and protein domains are included. C, dual color FISH analysis of MYH9 and
`NF2 in MPM samples. Negative control (normal lymphoblast line with metaphases to confirm localization) is in the top left. The orange probe is specific for NF2, the
`green for MYH9.
`
`"probably damaging" effect and one caused a reading frame
`shift. Twelve samples exhibited germline mutations in MYH9;
`seven had a predicted "possibly or probably damaging" effect.
`Although many mutations of MYH9 have been identified in
`several other cancers, only R1576Q identified in our series has
`been previously described (24).
`In 4 patients, tumor-specific mutations of ILK pathway–asso-
`
`ciated genes were identified: 3 in MYH9 (1.4%; Q738
`, K737N,
`A1197fs) and 2 in RHOA (1.4%; Y66N, A161V). One patient had
`in MYH9 one germline and two tumor-specific mutations. No
`significant correlation was found between the mutations in the
`ILK pathway–associated genes and clinical or pathologic vari-
`ables. No cell line exhibited mutations in these genes. Figure 1B
`depicts the mutations found in the myosin heavy chain genes and
`RHOA. Aberrations identified in BAP1, NF2, TP53, MYH9, MYH6,
`
`MYH10, PIK3C2A, RHOA, TNFRSF1A, 9p/CDKN2A, and 22q are
`illustrated in Supplementary Fig. S2.
`
`Assessment of NF2 and MYH9 copy number status
`Given the proximity of NF2 and MYH9 on the chromosome,
`22q copy number status was explored. Although allelic loss in this
`region is common in MPM, the probes used for standard clinical
`FISH analysis do not distinguish copy number independently
`for these two genes. Dual-color FISH analysis using MYH9- and
`NF2-specific probes was undertaken for the 13 cases carrying
`MYH9 mutation. Of the 13 primary tumors in which the two
`markers could be assessed, 2 (15%) exhibited no aberrations
`involving NF2 or MYH9; 1 (8%) was abnormal by losing NF2
`only; 7 (54%) had loss of both markers; and 3 (23%) samples
`exhibited extra copies of both markers likely due to polyploidy
`
`www.aacrjournals.org
`
`Cancer Res; 76(2) January 15, 2016
`
`323
`
`Personalis EX2157
`
`

`

`Downloaded from http://aacrjournals.org/cancerres/article-pdf/76/2/319/2741046/319.pdf by guest on 29 September 2023
`
`De Rienzo et al.
`
`Table 3. Protein alteration and clinical information of the cases carrying BAP1 and TP53 germline mutations
`Functional classification
`Gene
`Protein alteration
`Gender
`Other cancer (self)
`W196
`BAP1
`Male
`No
`BAP1
`W202C
`Male
`No
`V171
`BAP1
`Male
`No
`
`Probably damaging (score 1)a
`
`BAP1
`BAP1
`
`TP53
`
`TP53
`TP53
`
`TP53
`
`G41S
`Splice site acceptor
`
`Benign (score 0.078)a
`
`R248W
`
`Deleteriousb
`
`D49N
`C176F
`
`A86T
`
`Neutralb
`Deleteriousb
`
`Neutralb
`
`aPolyphen (http://genetics.bwh.harvard.edu/pph2/).
`bhttp://p53.iarc.fr/.
`
`Male
`Female
`
`Female
`
`Male
`Female
`
`Male
`
`No
`No
`
`No
`
`Skin cancer
`Myelodysplasia, Acute myeloid
`leukemia
`No
`
`Other cancer in the family
`No
`Nasal (father), Breast (sister)
`Renal and gastric (mother), lung
`(grandfather)
`Lung (father)
`Prostate (father), Meningioma
`(mother), Bladder (sister)
`Prostate (father),
`Rhabdomyosarcoma
`(grandmother)
`No
`No
`
`Unknown (two sisters and
`grandparents)
`
`(Supplementary Table S5D; ref. 25). Two of the 7 cases displaying
`loss of both markers had tumor-specific mutation in both MYH9
`and NF2, whereas 5 had tumor-specific mutation of MYH9 only.
`Furthermore, the case exhibiting NF2 loss only presented two
`tumor-specific mutations in addition to one germline mutation in
`MYH9. Examples of FISH analysis are shown in Fig. 1C.
`
`Mutual exclusivity
`In MPM, tumor-specific mutations in RHOA, MYH9 and TP53
`were mutually exclusive events that occurred more frequently in
`women (30%) than in men (12%) (P ¼ 0.023). We examined the
`relationship between RHOA and MYH9 in other cancers using the
`cBioPortal for Cancer Genomics (accessed July 13, 2015; ref. 24).
`Notably, in 26 of 31 (84%) tumor types that had mutations in
`both genes, RHOA and MYH9 mutations had a strong tendency
`toward mutual exclusivity (Supplementary Table S6).
`
`Analysis of TP53 in additional MPM cases
`Given that mutations of TP53 were found more commonly in
`women, and that women with MPM generally have lower levels of
`asbestos exposure than men, we expanded our validation cohort
`to power exploration of these associations. Mutational status of
`TP53 was analyzed by Sanger sequencing in 136 additional MPM
`samples. Asbestos counts were available for 148 of 283 cases in the
`combined cohort, including tumors from 88 women and 67
`patients with asbestos body counts in lung tissue consistent with
`background exposure ( 50 fibers/g lung tissue; ref. 26). Thirty-
`eight of 283 patients (13%) had TP53 point mutations, including
`18 men (9%) and 20 women (23%), (P ¼ 0.004; Supplementary
`Table S5B). Expanding the cohort added one case (total 4/283)
`with germline TP53 mutation (R248W; Table 3). Although TP53
`mutations were found more frequently in tumors with back-
`ground versus higher asbestos exposure, the analysis did not have
`sufficient power to exclude exposure level as a risk factor in either
`the whole cohort (P ¼ 0.068) or among women (P ¼ 0.498). No
`significant correlation of mutations in TP53 was found to either
`survival, or histologic subtype.
`
`Expression analysis of the candidate genes
`A comparative exploratory analysis of MPM transcriptomic
`profiles was conducted to determine whether histologic subtype,
`survival, or gender are correlated with specific expression patterns
`in BAP1, NF2, TP53, MYH9, MYH6, MYH10, PIK3C2A, RHOA,
`
`CDKN2A, and TNFRSF1A (Supplementary Fig. S1; Supplemen-
`tary Table S3). This cohort comprised 85 men and 66 women. Ten
`samples were also included in the sequencing analysis; these were
`pure epithelioid tumors showing no mutation in BAP1, MYH9, or
`RHOA.
`Nonepithelioid (versus epithelioid) tumors expressed signif-
`icantly higher levels of BAP1 (P < 0.001), MYH9 (P < 0.001),
`RHOA (P < 0.001), and MYH10 (P ¼ 0.001). RHOA was more
`highly expressed in men than women (P ¼ 0.001). Expression
`levels of other genes examined were not correlated with gender
`or histology.
`Patients with lower BAP1 expression had improved overall
`survival (P < 0.001; Fig. 2A). The highest quartiles of BAP1
`expression were associated with twice the risk of death (HR ¼
`2.31) compared with the lowest quartile, after adjusting for the
`independent effect of gender. The survival difference was similar
`across gender subgroups (male: HR¼ 2.20; P¼ 0.005; and female:
`HR ¼ 2.38; P ¼ 0.004; Fig. 2B). Similarly, the highest quartile of
`MYH9 expression was associated with increased gender-adjusted
`risk of death (HR ¼ 2.23) compared with the lower three quartiles
`(P < 0.001; Fig. 2C). MYH9 expression was associated with twice
`the risk of death in both males (HR¼ 2.26; P¼ 0.001) and females
`(HR ¼ 2.17; P ¼ 0.022; Fig. 2D). In the entire cohort, lower RHOA
`express