Molecular Changes in Mesothelioma With an Impact on Prognosis and Treatment

Genomic and epigenetic changes of potential interest for MPM histology, diagnosis, and prognosis are described in Table 1.

Genomic alterations in human MPM have been previously reported in numerous studies based on various methods including cytogenetic analysis of standard karyotype, classical comparative genomic hybridization (CGH), CGH array, single nucleotide polymorphism (SNP) array, and representational oligonucleotide microarray analysis (ROMA). Cytogenetic studies first demonstrated that numerous chromosomal abnormalities are associated with MPM, including various structural and numeric changes and recurrent alterations.4,5 These earlier studies1,2 have already been reviewed in detail. Table 2 shows the recurrent regions of chromosomal alterations reported in recent studies with high-throughput analyses. Malignant pleural mesothelioma cell cultures and primary tumors both share similar patterns of chromosomal alterations. However, the frequency of alterations in some particular chromosomal regions is generally higher in cultured cells, most likely due to the presence of normal cells in tumor samples, as mentioned by several authors. Losses of chromosomal regions are always more common than gains. Frequent losses are localized on chromosome arms 1p, 3p, 4q, 6q, 9p, 13q, 14q, and 22q and gains involve chromosome arms 1q, 5p, 7p, 8q, and 17q.6–12 A recent large-scale analysis of gene mutations, based on second-generation sequencing in 1 tumor specimen, confirmed the presence of numerous DNA rearrangements in MPM.13 

Differences in genomic alterations have been described in MPM according to the histologic subtype or the patient's asbestos exposure status. Although recurrent regions of chromosomal alterations are roughly similar between epithelioid and sarcomatoid MPM, significant differences in the frequency of genomic alterations have been observed, such as losses in chromosomal regions 3p14–p21, 8p12-pter, and 17p12–pter or gain in 7q.6 Experimental studies have shown that asbestos fibers induce chromosomal abnormalities in normal human mesothelial cells.14,15 Significant correlations have been described between high contents of asbestos fibers in lung tissue and partial or total losses of chromosomes 1, 4, and 9, and chromosomal rearrangements involving a breakpoint at 1p11–p22.16,17 More recently, comparison between recurrent altered regions in asbestos-exposed and nonexposed patients showed a significant difference in the 14q11.2–q21 region, which was also lost in fiber-induced murine mesothelioma.12 

None of the individual genomic aberrations observed are specific for MPM, as they are also found in other types of tumors. However, some of these genomic aberrations could be used to distinguish benign mesothelial proliferations from MPM. This is the case for the deletion involving the 9p21.3 locus, the site of the cyclin-dependent kinase inhibitor 2A gene (CDKN2A), which is one of the most frequent alterations in MPM and is often homozygous. Detection of CDKN2A deletion by fluorescence in situ hybridization (FISH) has therefore been evaluated for the diagnosis of MPM.18–20 Comparative genomic hybridization analysis has also been used in an attempt to distinguish MPM from adenocarcinoma and large cell anaplastic carcinoma of the lung. The frequency of several genomic alterations can be used to differentiate mesothelioma from lung carcinoma with a sensitivity and specificity of 89% and 63%, respectively.21 It has also been suggested that CGH analysis could be useful to distinguish sarcomatoid MPM from other types of spindle cell tumors of the pleura.22 

Correlations between patient survival and chromosomal imbalance have also been studied. Chromosome copy number and alterations of the short arm of chromosome 7 have been reported to be inversely correlated with survival.16,23 Univariate and multivariate analyses in a larger number of MPMs showed that homozygous CDKN2A deletion, detected by FISH analysis, is a significant independent adverse prognostic factor.24,25 Classification of patients with MPM into 2 groups defined by short-term (less than 12 months) and long-term disease recurrence after surgery also suggested an association between 9p21.3 deletion encompassing the CDKN2A locus and the short-term group.9 In the same ROMA analysis, chromosomal instability corresponding to the number of genomic alterations was shown to be higher in patients with MPM, characterized by a shorter time to relapse.9 In deciduoid MPM, a variant of epithelioid MPM, survival was also found to be longer in patients with a smaller number of losses.26 Interestingly, a correlation was demonstrated between chromosomal instability and tumorigenicity of human mesothelioma xenografts in nude mice.12 These data indicate a correlation between the number of genomic alterations and the aggressive behavior of MPM, and further studies are needed to determine whether chromosomal instability can be used as a prognostic factor.

Data on chromosome imbalance could also be useful to design new treatment strategies: a relevant example targets the methylthioadenosine phosphorylase gene (MTAP). Homozygous codeletion of the MTAP gene and the CDKN2A gene has been observed in most pleural mesotheliomas.24 The MTAP gene encodes a key enzyme in the salvage pathway of adenosine monophospahte synthesis, complementary to the de novo purine biosynthesis pathway. Inhibitors of de novo purine biosynthesis induce selective killing of MTAP-negative cells in culture.27 One clinical trial on MPM using l-alanosine showed that this inhibitor was ineffective at the dose used.28 Further studies are necessary to conclude on the value of this treatment strategy.

Genomic alteration studies have already contributed to our knowledge of the mechanisms of mesothelial carcinogenesis, especially by identifying or confirming the involvement of tumor suppressor genes (TSGs) such as CDKN2A in MPM. They have also identified potential markers for diagnosis, prognosis, and treatment. New genes of interest could be identified by using technologies providing more precise localization of altered chromosomal regions and, especially, by performing integrated mining of genomic data linked with epigenetic, miRNA profiling, and transcriptomic data in the same cultured cells or primary tumors.

Numerous genes have been shown to be down-regulated in mesothelioma cells by epigenetic regulation such as DNA methylation of their transcriptional promoters. These changes dysregulate several signaling pathways, including the Wnt pathway, in which several negative regulators are silenced by hypermethylation.29–32 The global epigenetic profile determined by high-throughput methylation analysis differs between MPM and normal pleura, indicating that MPM, like other cancers, has aberrant CpG island methylation.33,34 Gene profiles of hypermethylation also differ between MPM and other tumors.33–37 These data support the hypothesis that a specific DNA methylation program is induced during mesothelial carcinogenesis.

DNA methylation of gene loci in MPM is dependent on age, ethnic origin, histologic subtype, and asbestos exposure, which could explain discrepancies between the frequencies of DNA methylation in published studies as well as the experimental method used to detect it. Age-dependent changes in DNA methylation have been reported in the literature.38 An age-associated increase of DNA methylation has been reported in patients with MPM.39 The methylation status of the IGFBP2 locus (insulin growth factor–binding protein) and GDF10 locus (bone morphogenetic protein) is significantly higher in MPM for Japanese patients than US patients.40,41 

The frequencies of DNA methylation of TRAIL receptor genes (TNFRSF10C and TNFRSF10D) and of tumor suppressor gene RASSF1 have been reported to be significantly higher in epithelioid MPM than in sarcomatoid MPM histologic subtypes.35,42 These data were not confirmed in another study for RASSF1, but methylation of another gene, MT2A, encoding heavy metal–binding protein, was shown to differ between these 2 histologic subtypes.43 High-throughput methylation analysis showed that epithelioid and sarcomatoid mesotheliomas had differential methylation at 87 CpG loci.44 

A significant association between asbestos exposure and DNA methylation at the MT1A and MT2A gene loci has also been described in MPM.43 Methylation of TSG loci APC, CCND2, CDKN2A, CDKN2B, HPPBP1, and RASSF1 was studied in comparison with asbestos exposure. Only DNA methylation at the RASSF1 locus was correlated with an increased number of asbestos bodies in the patient's lung. A trend toward an increasing number of methylated cell cycle control genes and increasing asbestos body counts was also observed.39 Recently, high-throughput methylation analysis confirmed distinct methylation profiles between MPM from asbestos-exposed and from nonexposed patients and a significant positive association between asbestos fiber burden and methylation status of CDKN2A, CDKN2B, RASSF1, and MT1A in about 100 other loci.33 

DNA methylation could be useful for the diagnosis of MPM. Differences in the frequency of DNA methylation for several genes have been described between MPM and lung adenocarcinoma or nonmalignant pulmonary tissues.35,36,43 High-throughput methylation analysis covering several thousand CpG islands confirmed the potential value of DNA methylation profile for distinguishing MPM from these 2 other tissues. Accurate diagnosis could be based on the global methylation profile, but further studies on larger populations are needed before using a limited number of hypermethylated loci.33,34,44 It was recently suggested that DNA methylation at 3 specific loci, TMEM30B, KAZALD1, and MAPK13, could be useful in the differential diagnosis of MPM.34 

DNA methylation status of individual genes, such as those encoding a transcriptional repressor (HIC1), a proapoptotic protein (PYCARD), a tumor suppressor (LZTS1), and a transporter (SLC6A20), has been associated with either a good or poor prognosis.43,45 High-throughput methylation analysis showed that patients with MPM with a low frequency of DNA methylation had significantly longer survival.34 Furthermore, classification based on the methylation profile of patients undergoing surgical resection before any other treatment identified subgroups characterized by different clinical outcomes.33 These data highlight the potential prognostic value of DNA methylation analysis.

In view of the aberrant epigenetic events observed in MPM, the clinical value of histone deacetylase inhibitors (HDACis) has been studied in preclinical models using MPM cell lines and mouse xenograft models. Phase I and II clinical trials for patients with MPM have been conducted with several different HDACis, either alone or in combination with conventional chemotherapy. The encouraging results of these early-phase trials led to a phase III, multicenter, randomized, placebo-controlled study of one of these HDACis for patients with advanced MPM.46 

Like chromosome imbalance studies, epigenetic analyses have identified genes or pathways potentially involved in mesothelial carcinogenesis, such as the Wnt pathway. At the present time, only the global methylation profile appears to be relevant for diagnosis or to evaluate the patient's survival, thereby limiting its clinical applications. Furthermore, epigenetic regulation mechanisms in MPM have been mainly studied in terms of DNA methylation, but insufficient data are available on regulation of histone modifications, despite their crucial role in maintaining chromatin stability. Such data are necessary to support clinical trials based on HDACis.

MicroRNAs are emerging as key players in the control of a multitude of biologic processes and are aberrantly expressed in several tumors including MPM. MicroRNA expression has been shown to differ between MPM tumors and normal pleura47 and between MPM cell lines and immortalized mesothelial cells.48 Malignant pleural mesothelioma histologic subtypes also demonstrate a specific miRNA expression pattern.47,48 Potential targets of these deregulated miRNAs include TSGs, oncogenes, and genes involved in specific signaling pathways.47,49 However, a link between miRNA expression and mesothelial carcinogenesis has been demonstrated by experimental analysis for only miR-31 and miR-29c. MiR-31 is frequently lost in MPM owing to its chromosomal location at 9p21.3, and miR-29c expression is higher in epithelial MPM of patients with a good prognosis (time to disease progression greater than 1 year). Overexpression induced by transfection of these 2 miRNAs decreased in vitro proliferation, migration, invasion, and colony formation of the same 2 MPM cell lines.50,51 

MicroRNAs have been proposed as diagnostic tools. Down-regulation of 7 miRNAs (miR-141, miR-200a, miR-200b, miR-200c, miR-203, miR-205, and miR-429) was shown to be characteristic of MPMs, regardless of their histologic subtypes, and could be used to distinguish MPM from adenocarcinoma.49 Another study52 demonstrated that 3 miRNAs (miR-193, miR-200c, and miR-192) can be selected to distinguish MPM from various carcinomas invading the lung and pleura.

Recent data suggest that miRNA expression also could be used as a prognostic tool, as down-regulation of miR-17 and miR-30c in sarcomatoid MPM and upregulation of miR-29c in epithelioid MPM are significantly associated with better patient survival.48,51 

MicroRNA expression analysis is a promising tool to improve the accuracy of diagnosis and may be complementary to immunohistochemical markers. This analysis also opens up new perspectives for the prognostic assessment of MPM in the near future. However, a better knowledge of miRNA signatures of MPM is still necessary, as certain discrepancies have been observed between miRNA profiling studies. Functional studies in cultured cells and animal models are also needed to determine the precise contribution of miRNAs to mesothelial carcinogenesis and whether or not they can be used as potential targets for anticancer therapy.

Knowledge of gene mutations provides insight into specific mechanistic pathways that can be altered in MPM cells, opening the way for future targeted therapies. A number of genes are known to be recurrently mutated in malignant mesothelioma (MM).

The TP53 gene, a TSG located at 17p13.1 that controls cell cycle and apoptosis, is mutated in many types of human cancers. Its mutation frequency is about 20% in human MPM, a fairly low rate in comparison with other human cancers.3 Point mutations are the main types of alterations in MM. The International Agency for Research on Cancer p53 database indicates 6 point mutations, 5 missense mutations, and 1 stop mutation (http://www.p53.iarc.fr, accessed June 11, 2011). In a study conducted to determine the frequency of simian virus 40 (SV40) in Egyptian patients with MM, altered p53 and pRb expressions were found in 57.5% and 52.5% of patients, respectively, with no p53 mutation.53 These authors assessed the prognostic impact of altered expression of RB1 and TP53 gene status. Univariate analysis showed a significant correlation between overall survival and p53 overexpression (P  =  .05). Although a matter of debate, SV40 has been associated with MM, and is assumed to act as a cofactor of asbestos in carcinogenesis. In some MMs, p53 protein function may be inactivated after binding to the large T (Tag) SV40 protein, but SV40 Tag expression in MM remains controversial.54 In a recent study,55 no expression of SV40-specific miRNA was detected in human malignant pleural mesothelioma (MM) samples.

No relationship has yet been established between TP53 mutation and clinical impact. The uncertainties concerning p53 status in MPM make it difficult to establish a relationship between p53 status and prognosis and/or treatment.

The NF2 (neurofibromatosis 2) TSG, located on 22q12, was one of the first TSGs shown to be inactivated in MPM.56,57 Early conventional cytogenetic studies reported a loss of chromosome 22 in human MM.58,59  NF2 inactivation is frequent, with rates ranging from 20% to 60% depending on the material used, tissue or cells, and the method (classical CGH, DNA sequencing, etc). Various types of lesions have been described, including small and large deletions, homozygous deletions, and nonsense and missense mutations. The role of NF2 in mesothelial carcinogenesis will be described in the section on the Hippo pathway.

A second recurrent gene alteration occurring in human MM consists of inactivation of genes located at the CDKN2A locus. The CDKN2A locus encodes both p16^INK4A and p14^ARF, which share common exons but no common amino acid sequence. Alterations at this locus have been demonstrated by DNA sequencing, FISH, and methylation, as reported above. The most frequent alteration is homozygous deletion in about 70% of cases.3 This alteration is related to asbestos exposure in lung cancer and is also observed in mesotheliomas induced by mineral fibers in mice.60,61 The CDKN2B gene adjacent to CDKN2A is also frequently codeleted in MPM, but at a lower frequency.62 

Fluorescence in situ hybridization detection of CDKN2A deletion has been proposed as a technique to differentiate between reactive and malignant mesothelial cells on paraffin-embedded sections and effusion cytology.19,20,63 Several authors25,64–67 have reported that loss of the encoded protein p16^INK4A, as assessed by immunohistochemistry and FISH analyses and confirmed by gene profiling microarray studies, is associated with lower survival.

The β-catenin status in MPM cells was reported in one study of 2 primary tumors and 8 cell lines in which 1 homozygous deletion was found in 1 cell line.68 A modification of the subcellular localization of β-catenin was reported in another study,69 consistent with activation of β-catenin as transcriptional cofactor.

There is now a general consensus that several TSGs are frequently altered in MPM: NF2, CDKN2A, CDKN2B, and, less frequently, TP53. In contrast, no recurrent oncogene mutation has yet been identified in MPM.

Data from array-based studies indicate deregulation of gene expression in MPM. These studies were conducted to improve histologic classifications and prognosis (Tables 3 through 5). These data were recently reviewed by Gray et al.70 

Early studies were conducted with MPM cell lines and compared with normal pleural mesothelial cells (Table 3). A complementary DNA (cDNA) array including 588 genes showed that 26 genes that play a role in signaling pathways (MAP3K14/NIK [a serine/threonine protein kinase that stimulates NF-κB activity], JAG1/JAGGED1 [a ligand of the notch1 receptor]), in cell cycle (CCND1 [cyclin D1], CCND3, [cyclin D3], CDC25B [CDK phosphatase]), in cell growth (FGF3 and FGF12 [fibroblast growth factor 3 and 12], PDGFRB [platelet-derived growth factor receptor B]), and in DNA damage repair (XRCC5/Ku80) were overexpressed; on the other hand, 13 growth factor–encoding genes such as FGF1 and FGF7 (fibroblast growth factor 1 and 7), CCND2 (a regulatory subunit of cyclin-dependent kinases, involved in cell cycle G1/S transition), KDR/VEGFR2 (vascular endothelial growth factor receptor 2), PDGFRA (platelet-derived growth factor receptor), RARβ (retinoic acid receptor β2), and genes encoding proteins involved in cell adhesion, motility, and invasion were underexpressed.71 

Differentially expressed genes were also related to tumor invasiveness and resistance to anticancer defenses.72 In another study,73 in a series of 14 differentially expressed genes, 8 were upregulated: CFB (complement factor B), FTL (ferritin, light polypeptide), IGFBP7 (insulin-like growth factor–binding protein 7), RARRES1 (retinoic acid receptor responder 1), RARRES2 (retinoic acid receptor responder 2), RBP1 (retinol-binding protein 1), SAT (spermidine/spermine N1-acetyltransferase), and TXN (thioredoxin), while 6 were down-regulated: ALOX5AP (arachidonate 5-lipoxygenase–activating protein), CLNS1A (chloride channel nucleotide-sensitive 1A), EIF4A2 (eukaryotic translation initiation factor 4A2), ELK3 (ETS-domain protein, SRF accessory protein 2), DF2/REQ (apoptosis response zinc finger gene), and SYPL (synaptophysin-like protein).

The expression of 588 cancer-related genes was screened in 16 MPM tumors by using normal mesothelial cell lines and pleural mesothelium as references.74 Eleven genes—COL1A2, COL6A1 (collagen), tPA, MMP9 (protease), CDH3, L1CAM, ITGB4, PLXNA3/PLXN3, KRT14/K14 (cell adhesion or cell surface molecule), SEMA3C (semaphorin), and CXCL10/INP10 (chemokine)—were overexpressed in MPM.74 

Microarray expression data of 40 MM tumor specimens, 4 normal lung specimens, and 5 normal pleural specimens were reported by Gordon et al.75 These authors identified genes that were significantly differentially expressed in tumors compared to normal samples. There were 328 overexpressed genes and 311 underexpressed genes in MM tumors. These authors proposed the existence of 3 novel candidate oncogenes—NME2 (nucleoside diphosphate kinase), EID1/CRI1 (regulator of EP300 and RB1), and PDGFC (platelet-derived growth factor)—and 1 candidate tumor suppressor gene, GSN (cytoskeleton regulator), in MPM.

In another study,76 MM tissue specimens from 16 patients were compared to 4 control pleural tissue samples by using cDNA microarray filters with 4132 clones. Interestingly, upregulation of many genes involved in the glycolysis pathway and the Krebs cycle was observed, in agreement with the ability of cancer cells to rely on aerobic glycolysis, the “Warburg effect.” Other upregulated genes were involved in messenger RNA (mRNA) translation and cytoskeletal reorganization pathways. These authors76 also identified genes encoding gp96 (adenotin, GRP94; HSP90B1), LRP (lung-related resistance protein; MVP), galectin-3 binding protein (LGALS3BP), and Mr 67 000 laminin receptor (RPSA), although RPSA was not expressed on tumor cells, but on infiltrating vessels.

More recently, Crispi et al77 compared MPM tissues from 9 patients to normal pleural tissues from patients undergoing resection for a nonneoplastic disease. Components of the condensin complex (eg, BRRN1, CNAP1, NCAPD3) and members of the kinesin family (eg, KIF14, KIF23, KIFC1) were upregulated. Other upregulated genes were related to cell proliferation and its control, such as cyclin-dependent kinase gene CDK1/CDC2; cyclin genes CCNA2, CCNB1, CCNB2, and CCNL2; the DLG7 component of the mitotic apparatus; the gene encoding the checkpoint kinase involved in response to DNA damage, CHEK1/CHK1; and BUB1 and MAD2L1, components of the spindle checkpoint.

Romagnoli et al78 used a quantitative polymerase chain reaction (PCR)–based, low-density array focusing on genes involved in cell cycle regulation. They studied 45 MPM tumor samples and normal tissue samples obtained by pleural wiping of surgical samples, with no evidence of pleural disease. Several genes were differentially expressed: either down-regulated in cancer cells (UBE1L, CCND2) or upregulated (CHEK1/CHK1, CCNH, CCNB1, p18-CDKN2C, CDC2, FOXM1, CDC6). Overexpression of the cell cycle regulator Chk1 was confirmed in an independent set of 87 MMs by immunohistochemistry with tissue microarrays. In other research, gene expression studies confirmed by reverse transcriptase (RT)–PCR showed down-regulation of the putative TSG FUS1/TUSC2 and the cytokine OSM gene (oncostatin M) when compared to normal samples (matched normal peritoneum specimens).9,79 Down-regulation of FUS1/TUSC2 and PL6/TMEM115 was also observed in comparison with matched normal pleural specimens.9 

Several studies have provided data on MM classification (Table 4). A microarray transcriptional profiling study of 10 MPM cell lines and 4 MPM primary tumor specimens distinguished epithelial, sarcomatoid, and biphasic MPM.80 Upregulated genes included ST14, a gene encoding matriptase, which is a membrane serine protease degrading the extracellular matrix, overexpressed in epithelial MPM. In the comparative study with normal cells quoted above,74 ,SEMA3C, ITGB4, CDH3, and COL6A1 were highly expressed in the epithelioid MPM subtype; L1CAM, K14, and INP10 were overexpressed in the mixed MPM subtype; and MMP9 and PLXN3 were overexpressed in the sarcomatoid MPM subtype. Statistically significant distinct gene expression patterns between epithelial and nonepithelial tumors were reported to be correlated with distinctive subclasses from hierarchical clustering in a series of 40 MPMs.75 In a series of 99 tumors, genes typical of epithelial differentiation and encoding the cell-surface transmembrane proteins uroplakin 1B and 3B (UPK1B and UPK3B) and the protease kallikrein 11 (KLK11) were more highly expressed in epithelioid MM.25 Romagnoli et al78 compared epithelioid and nonepithelioid MPM by using a quantitative PCR-based low-density array. Two genes were overexpressed in epithelioid MPM, namely, the transcription factor gene TFDP2 and the proto-oncogene ABL1, whereas the transcription factor gene TWIST1 was overexpressed in the nonepithelioid group. In an attempt to classify genes according to their correlation with survival, more favorable genes have been associated with epithelioid morphology and unfavorable genes, with sarcomatoid type or epithelioid MM with poor outcome.25 

Other authors have tried to improve prognostic assessment by using gene expression analyses (transcriptome and/or quantitative RT-PCR) of MM primary tumors or cell lines (Table 5). In a study investigating mesothelioma surgical specimens,81 calculation of 3 gene expression ratios, KIAA0977/GDIA1, L6/CTHBP, and L6/GDIA1, was found to be a good predictor of surgical treatment–related outcome. Samples with geometric means greater than 1 and less than 1 were assigned to good-outcome and poor-outcome groups, respectively. In a cohort of 39 patients undergoing surgery, Gordon et al82 validated previous findings and identified new sets of gene expression ratios, CD9/KIAA1199, CD9/THBD, DLG5/KIAA1199, and DLG5/THBD, allowing classification of tumors according to patient outcome. In a more recent study, Gordon et al83 investigated 120 consecutive patients with malignant pleural mesothelioma treated by surgery. None of the patients had received preoperative neoadjuvant chemotherapy or radiation therapy. By analyzing data for 4 genes, they defined 3 ratios of gene expression (TM4SF1/PKM2, TM4SF1/ARHDDIA, COBLL1/ARHDDIA), which, associated with other prognostic factors, were able to discriminate high-risk from low-risk patients.

A cohort of 1153 samples from patients diagnosed with 11 distinct types of cancer, including 17 patients with mesothelioma from the study by Gordon et al,81 was investigated by microarray analysis for molecular signatures based on the polycomb group BMI-1–associated gene expression pathway, a pathway essential for self-renewal of hematopoietic and neural stem cells.84 Expression of the 11-gene signature was a powerful predictor of poor prognosis in cancer patients. These 11 genes were Gbx2, KI67, CCNB1, BUB1, KNTC2, USP22, HCFC1, RNF2, ANK3, FGFR2, and CES1.84 

Several studies have identified specific genes associated with patient outcome. In a study comparing MM samples from patients with short-term recurrence after surgery (STR) and patients with longer time to relapse,9 the cadherin gene CDH2 was upregulated especially in the STR group. In contrast, the gene for chaperone protein, DNAJA1, showed reduced expression in the STR cohort. In addition, the authors9 noted no discrimination between epithelial and biphasic histologic types.

Aurora kinases A and B (AURKA and AURKB) are serine/threonine kinases that play an important role in chromosome alignment, segregation, and cytokinesis during mitosis. They were found to be overexpressed in a study of 99 MPMs.25 The expression of aurora kinases and genes participating in cell division and mitotic control was further investigated in 29 MPMs.85 Expressions of AURKA and AURKB and related genes were correlated, and overexpression of AURKB, determined by immunohistochemistry, was significantly correlated with poor outcome.85 

A correlation between metalloproteinase MMP14 expression and overall survival was reported in 1 study of 9 patients with MPM treated by standard thoracotomy for therapeutic purposes, compared to 4 normal pleural samples.77 High MMP14 expression was associated with lower survival. This gene has been proposed as a potential MPM biomarker. Upregulation of MELK (maternal embryonic leucine zipper kinase) was associated with poor survival, confirming previous findings by Lopez-Rios et al,25 but BTG2, which plays a role in regulation of G1/S transition, was associated with different outcomes in these 2 studies. Other genes including BIRC5, an inhibitor of apoptosis; KIF4A, an ATP (adenosine triphosphate)–dependent microtubule-based motor protein; and SEPT9, a member of the septin family involved in cytokinesis and cell cycle control, were upregulated and associated with poor prognosis.77 In this study, a favorable survival was associated with down-regulation of transcription factor gene WT1 in contrast with a previous study,25 which associated long-term survival with upregulation of WT1.

Microarray analysis discriminated between normal and MM samples in a comparative study of 8 normal peritoneum specimens and 7 stage I MMs,86 subsequently validated on a large set of matched normal/MM samples by RT-PCR. Intense overexpression of HAPLN1 (hyaluronan and proteoglycan link protein 1, a protein of the extracellular matrix) was observed in MM samples. Immunostaining with anti-HAPLN1 antibodies demonstrated that all MPM types (epithelial, mixed, and sarcomatoid), as well as reactive mesothelium, expressed this gene. Moreover, HAPLN1 expression was negatively correlated with time to progression and survival.86 Functional studies using transfection assays revealed that MM cells overexpressing full-length HAPLN1 or its functional domains strongly supported the protumorigenic role of HAPLN1.

A meta-analysis was performed from published data on microarray analysis of gene expression profiles in mesothelioma, glioma, and prostate cancer.87 Mesothelioma data were derived from the study by Gordon et al.81 Malignant mesothelioma cases consisted of 8 good responders who survived more than 17 months, with 10 patients in the poor-responder group surviving less than 6 months. A list of genes generated according to patient outcome showed similarities between the 3 types of cancers.87 Thirteen highly expressed genes and 1 gene expressed at low levels were identified as being equally related to poor survival in the 3 types of cancers. These genes encode proteins of the extracellular matrix and regulators of extracellular matrix assembly, and include angiogenesis genes.87 These results are consistent with a more aggressive state of malignant cells and a more deleterious tumor microenvironment. These results may be of interest for combining tumor-specific and more global therapies.

An analysis of 6 MPMs, compared to normal visceral and parietal pleural tissues, has focused on differential gene expression and identification of pathways that could be related to the drug and irradiation resistance of pleural MM.88 Several genes encoding proteins known to control DNA replication, cell cycle regulation, and DNA repair were identified as overexpressed or underexpressed in MPM and could account for MPM resistance mechanism to chemotherapies.88 

These studies show changes in the expression of genes involved in several regulatory pathways. Discrimination between epithelioid and nonepithelioid MPM was reported in several studies, without apparent benefit for classification of MPM subtypes in comparison with classical histologic analysis. Other studies developed a gene ratio approach to predict outcome in patients having undergone surgery. No extrapolation can be made to other therapeutic settings, such as chemotherapy, at the present time. Several specific genes were identified as potential predictors of patient outcome. Although providing a number of candidate areas to kill cancer cells or abolish their growth, these results need to be confirmed on a larger number of cases before proceeding to clinical applications. An important issue is to determine the most pertinent individual approach in relation to the various biologic features of MM cells.

Membrane receptor tyrosine kinases (RTKs) drive downstream cell signaling of cell proliferation and cell cycle control, survival, and differentiation.89 Networks downstream from RTKs can be activated by RTK mutation or sustained signaling by autocrine or paracrine mechanisms, providing a useful context to therapeutically counter the effects of RTK activation.

Epidermal growth factor receptor (EGFR) is generally not mutated in human MPM. However, in an immunohistochemical study, EGFR was expressed in 44% of MPM cases.90 The EGFR protein status was statistically significantly associated with a favorable prognosis, but was not an independent prognostic factor, when compared to clinicopathologic status.90 A tissue array study91 was performed on epithelioid tissue samples from 48 MPM cases for comparison between long-term survival and short-term survival, in association with expression of other proteins involved in the corresponding pathway. A relationship was found between EGFR expression and long-term survival, whereas platelet-derived growth factor receptor (PDGFR) signaling was more strongly associated with short-term survival.91 In contrast, no relationship was found between survival and EGFR protein or mRNA expression.92 

Epidermal growth factor receptor alteration cannot be considered to be critical in MPM at the present time, which may explain why, despite high EGFR expression in MPM, the EGFR inhibitors gefitinib and erlotinib did not induce any significant tumor response in phase II studies of patients with MPM.93 Response rates were between 0% and 4% and median overall survival was between 4.6 and 13.1 months in phase II trials of patients with either failure of first-line chemotherapy or no previous treatment.94–96 

KIT/CD117 encodes a stem cell factor receptor. In MPM, KIT expression has mostly been studied by immunohistochemistry, showing a low percentage of positive tumors.97 No expression was detected by RT-PCR in a study of 37 MPMs.98  KIT has not been shown to be characteristic of MPM at the present time.

Several immunohistochemical studies have demonstrated an enhanced expression of vascular endothelial growth factor (VEGF) in a large proportion of MPMs in comparison with nonneoplastic specimens.99 Contradictory results were found regarding the correlation between VEGF expression and survival. Vascular endothelial growth factor was not identified as a prognostic factor in studies of 52 MPM specimens100 and 37 MPM specimens.101 In contrast, in a study of 40 MPM tissues,102 vascular endothelial growth factor showed significant correlation with short survival and was an independent prognostic factor. Malignant pleural mesothelioma cells express both VEGF and vascular endothelial growth factor receptors (VEGFRs) (fms-related tyrosine kinases; FLT1 and FLT4) and fetal liver kinase (KDR/FLK1).103–106 An autocrine role of VEGF has been suggested, since neutralizing antibodies against VEGF or VEGFR, or antisense oligonucleotides against VEGF, significantly reduce MM cellular proliferation.105,107 Vascular endothelial growth factor expression can be regulated by lipoxygenases. Human MPM cells, but not normal mesothelial cells, express a catalytically active arachidonate 5-lipoxygenase (5-LO). A 5-LO antisense oligonucleotide potently and time dependently reduced VEGF mRNA and constitutive VEGF accumulation in the conditioned media of MPM cells.108 These results indicate that VEGF may have multiple effects, as a key regulator of MM growth via activation of its tyrosine kinase receptors, and as promoter of tumor angiogenesis.

Despite unsuccessful early trials of anti-VEGF therapy, numerous clinical trials are testing the benefit of VEGF inhibitors in combination with chemotherapy.95,109 

Malignant mesothelioma cell growth may be linked to autocrine or paracrine stimulation by platelet-derived growth factor (PDGF), and the regulation by PDGF appears to be complex in MM cells. Normal human mesothelial cells express low levels of PDGF-A mRNA chain, and PDGF-B mRNA was not detectable.110 These cells express PDGFR-A mRNA and protein and have weak to undetectable levels of the PDGFR-B mRNA and protein.111 In contrast, human MM cells express high levels of PDGF-A and PDGF-B, as well as PDGFR-B.110,111 However, expression of PDGFR-B is controversial, and weak to undetectable levels have been reported.110–113 Nevertheless, an autocrine proliferation can be suggested in MM, as it may occur via binding of homodimer of PDGF-B chains.114 Platelet-derived growth factor has been suggested as a regulatory factor for proliferation of MM cells, either directly or indirectly, via the hyaluronan/CD44 pathway. Hyaluronan is an important constituent of the extracellular matrix. PDGF-BB–stimulated normal human mesothelial cells express both hyaluronan synthase and hyaluronan.115,116 

PDGF-A–stimulated autocrine loop does not seem to play a positive role in mesothelioma proliferation in vitro, but nude mice injected with MM cells that overexpress PDGF-A showed increased tumor incidence and reduced latency period to tumor formation.117,118 These data suggest that PDGF-A could contribute to tumor formation via a paracrine mechanism to generate favorable environmental conditions, for example, by stimulating angiogenesis for tumor proliferation.118 

Like EGFR-targeted therapy, the PDGFR inhibitor imatinib mesylate was ineffective in clinical trials.119,120 

Human MM cells express insulin growth factor (IGF) and insulin growth factor receptor (IGFR).121 IGF-1 appears to function as an autocrine growth stimulus in human mesothelial cells.122 When activated, IGFR phosphorylates multiple classes of signal transduction adaptor molecules, including insulin receptor substrates. Insulin receptor substrate 1 was found to induce cell proliferation in response to IGF-1, whereas cell migration was induced by insulin receptor substrate 2.123 In addition, various members of the insulin-like growth factor–binding protein (IGFBP) family have been investigated in MPM. IGFBPs form a complex with IGFR subunit and IGF and have been shown to either inhibit or stimulate the growth-promoting effect of IGF. IGFBPs can be either expressed or unexpressed in MM, modulating the aggressiveness of the MM phenotype.121,124,125 

MET is a proto-oncogene whose mutation appears to be uncommon in MPM. No mutation was reported in a study of 20 cell lines,126 but 5 point mutations and 1 deletion were identified in a series of 43 primary tumors and 7 cell lines.127 The encoded protein is involved in pathways regulating development, cell growth and survival, motility and invasion. It is expressed in most MPMs and in reactive mesothelium but not in normal mesothelial cells.128,129 Hepatocyte growth factor/scattering factor (HGF/SF), the related Met ligand, is also expressed in some but not all MPM cells. In vitro stimulation of MPM cells by HGF/SF increases spreading, motility, and/or invasiveness, but these effects are dependent on the cell line.127,130,131 Experimental studies with cultured MPM cells demonstrated that inhibition of MET by RNA interference or protein kinase inhibitor resulted in G1/S arrest and reduction of the activity of Akt and Erk1/2 signaling in some cell lines.127,131 However, no correlation was found between levels of MET and ERK1/2 phosphorylation.126 In light of these results showing a tumor-dependent activation of HGF/MET signaling, HGF/MET status may define various MPM subclasses.

The activation status of MET and other RTKs (EGFR family [Erb1, Erb2, Erb3], and PDGFR-B) was investigated in 20 MPM cell lines and 23 primary specimens of MPM, and the effect of MET-specific inhibitors (MET-shRNA interference vector and RTK inhibitors) was investigated on cell lines.126 The results showed that inhibition of a single RTK was not sufficient to obtain a tumor suppressor effect but that inhibition of multiple RTKs should be considered.126 

As several RTK receptors are tyrosine phosphorylated in some MMs, downstream activation of the mitogen-activated protein kinase (MAPK) proliferation-associated signaling pathway is likely. Several studies have investigated phosphorylation of proteins of the MAPK cascade, extracellular-regulated kinases (ERKs), Jun amino-terminal kinases/stress-activated kinases (JNKs/SAPKs), and p38 MAPK. Other studies have tried to modulate MAPK pathways to inhibit cell survival and induce apoptosis.

Phospho-ERK expression was studied by immunohistochemistry in 50 biopsy specimens including non–small cell lung cancer and normal lung, and pleural tissue comprising 10 MPMs (6 epithelioid, 1 sarcomatoid, and 3 biphasic).132 Malignant pleural mesothelioma showed significant ERK phosphorylation, compared to lung cancer and normal tissues.132 Activation of ERK, JNK, and p38 MAPK was investigated in 28 MPMs, 8 peritoneal MMs (32 effusions and 4 biopsy specimens), and 14 samples of reactive mesothelium by assessing the expression of phosphorylated proteins by immunohistochemistry and Western blot. MAPK activation did not differentiate between benign and malignant mesothelial cells.133 The authors argued against a major role for this pathway in the malignant transformation of mesothelial cells. They also noted that MAPK expression and phosphorylation were better predictive factors of outcome, in agreement with data obtained in ovarian cancer.133 Arsenic trioxide (As₂O₃) is a chemical compound that inhibits cell proliferation and induces apoptosis in tumor cells via the MAPK pathways. As₂O₃ was shown to inhibit proliferation and induce apoptosis in 1 mesothelioma cell line.134 As₂O₃ did not alter phosphorylation of either Akt or Src, while ERK1/2 and JNK1/2, but not p38 MAPK, were markedly phosphorylated after As₂O₃ treatment, indicating the involvement of the JNK-dependent, ERK-dependent pathway in the cell response.134 However, p38 MAPK appears to be involved in the response to transforming growth factor β. In 6 human MM cell lines, migration and invasion linked to the production of metalloproteinases were stimulated by transforming growth factor β 1 via phosphorylation of p38 MAP kinase. The authors113 suggested that this pathway could be targeted to reduce mesothelioma progression. Ou et al135 determined the relative levels of tyrosine phosphorylation of 42 distinct RTKs in mesothelioma cell lines established from surgical specimens and found coordinated activation of RTKs EGFR, ERBB3, AXL, and MET. As MAPK can be activated by heat shock proteins (HSPs), these authors studied the effect of HSP90 inhibition on ERK1/2 activation. HSP90 inhibition reduced tyrosine kinase phosphorylation and induced apoptosis.135 The effect of other HSPs was also investigated in the context of the possible use of hyperthermic chemotherapy.136 HSP40 was upregulated in response to heat stress, associated with activation of the ERK1/2 and p38 pathways in a study of 3 MPM cell lines, thereby suggesting that treatment could be more effective by blocking these pathways.136 HSP90 overexpression has been reported in MPM,88 and DNAJA1, a member of the HSP40 family, showed decreased expression in MPM with short-term recurrence of the disease.9 

These results show that regulation of mesothelioma cells via MAPK pathways is complex. Targeting these pathways to abolish cell proliferation could be proposed, but the treatment strategy would be difficult to define at the present time. MAPK activation is important for cell survival and can also be linked to apoptosis events. More specific investigations taking into account specific tumor characteristics and microenvironment must be conducted in order to trigger cell growth inhibition and apoptosis.

Constitutive activation of RTKs in MM results in downstream signaling cascades including phosphatidylinositol 3–kinase (PI3K-AKT), a cascade regulating cell growth processes, cell migration, and apoptosis. Phosphorylation of AKT protein, the active form of the protein, has been demonstrated in MM cells. Immunohistochemical analysis has revealed elevated levels of phospho-AKT in nearly two-thirds of human primary MPMs. A strong association with elevated phospho-mTOR positivity in the same tumors confirmed activation of the Akt pathway.137 Activation of AKT triggers antiapoptotic mechanisms. However, while the PI3K-Akt signaling pathway was activated in adherent MPM cells, loss of anchorage resulted in inactivation of this pathway and failed to restore apoptosis.138 Inactivation of phosphatase and tensin homolog (PTEN, deleted from chromosome 10), a TSG and negative regulator of the PI3K-AKT pathway, could account for PI3K-AKT activation. PTEN homozygous deletion has been reported in a small subset of MPM cell lines.139,140 A tissue microarray–based study conducted on 206 tumor tissues141 demonstrated that loss of PTEN expression was observed in 62% of cases. In this study, PTEN expression was correlated with better survival from data available for 129 patients. PTEN was an independent prognostic biomarker in patients with mesothelioma.141 

The Wnt signaling pathway regulates developmental processes, cell proliferation, and cell polarity. It is driven by membrane protein activation involving low-density lipoprotein receptor–related protein (LRP) and Frizzled, and G protein–coupled receptors. Activation of the Wnt signaling pathway prevents β-catenin phosphorylation and its subsequent ubiquitination and degradation. β-Catenin plays a central role in the Wnt pathway activity, as β-catenin can act as a coactivator of transcription, allowing the expression of a variety of genes exerting pleiotropic effects.142 While no recurrent mutation of β-catenin has been described in MPM, the Wnt pathway could be altered as a result of promoter hypermethylation of regulatory genes.29,31,32 Apart from this canonical Wnt/β-catenin pathway, a noncanonical β-catenin–independent Wnt pathway can also transduce signals in MPM cells. This was demonstrated in β-catenin–deficient MPM cells in which inhibition of Wnt signaling produced growth reduction and apoptosis.30,143 

Gene expression profiling of MM cell lines, primary MPM tumors, and normal pleural tissue has been studied by using a custom array designed to profile the expression of genes involved in the Wnt signaling pathway and downstream of Wnt signaling.144 In the 16 matched samples (malignant tissue and normal adjacent pleura) investigated, numerous Wnt genes (WNT1, WNT2, WNT5) and Wnt-related genes (MYC, CCND1, JUN) were upregulated. WNT2 was most frequently upregulated. In contrast, WNT8A and some WNT antagonists (DKK1, SFRP2, and SFRP4) were down-regulated. A role for WNT2 in cell survival was demonstrated with anti-Wnt2 antibody and Wnt2 small interfering RNA, associated with inhibition of the downstream effectors of the Wnt pathway.144 Wnt signaling inhibition is dependent on several factors including the Dickkopf (DKK) gene family. One member, REIC/Dickkopf-3, is down-regulated in numerous human cancers.145 In 4 human MM cell lines, REIC/Dickkopf-3 expression was lower than in normal tissue, and overexpression by transduction in 1 cell line induced apoptosis via a JNK-dependent pathway.145 Moreover, a preclinical study consisting of orthotopic inoculation of REIC/Dickkopf-3–deficient, luciferase-labeled MM cells, followed by intrapleural injection of recombinant REIC/Dickkopf-3-adenovirus, resulted in a strong antitumor effect.145 These results suggest that deregulation of the Wnt signaling pathway can be involved in mesothelial carcinogenesis and that identification of key targets could be of interest to suppress tumor development.

Merlin, the protein encoded by NF2, regulates cell growth by signaling via the Hippo pathway to inhibit the function of the transcriptional coactivator and candidate oncogene YAP1 via its phosphorylation. Overexpression of YAP1 was found in 1 MM cell line.146 Moreover, Yap1 protein physically and functionally interacted with merlin; in NF2-transfected cells, merlin expression was accompanied by reduction of nuclear localization of Yap1, suggesting that merlin can inhibit Yap1 function by sequestration.146 Inactivating homozygous deletions or mutations of LATS2 were recently demonstrated by CGH and DNA sequencing analyses in about 22% of MPMs including 20 cell lines and 25 primary tumors.147 Disruption of NF2 signaling plays a major role in the development of MPM because of the high rate of mutations in this tumor. Despite a wild-type status for NF2, merlin also appears to be present in an inactivated phosphorylated form in MPM cells.148 Recent data suggest that the Hippo pathway involving merlin could be targeted for treatment strategies. There is now consensus concerning inactivation of the Hippo pathway in MPM. To the best of our knowledge, NF2 expression has not been associated with any specific MPM subtype or specific characteristics and has not been linked to prognosis. Investigation of merlin function in MPM could be useful to develop new therapies. Some examples have been published in the literature. Using NF2-negative MM cell lines transduced with a recombinant NF2 adenovirus (AdNF2), cDNA microarray analyses have revealed differences in gene expression profiles characterized by a decrease in cyclin D1 (CCND1) expression—a gene upregulated in MPM—in cells transduced with AdNF2 compared to those transduced with the control adenovirus. In parallel, CDK4, the catalytic partner of cyclin D1, was inactivated and pRb was dephosphorylated, in agreement with efficient control of the G1/S transition in NF2-expressing cells. G1 cell cycle arrest was confirmed by cell cycle analysis.149 In this study, the authors found that the effect of NF2 was related to repression of cyclin D1 promoter activity via PAK1 inhibition.149 ,NF2 function could also be related to regulation of motility and invasiveness in MM cells, as demonstrated by down-regulation of focal adhesion kinase, and inhibition of motility and invasiveness after NF2 transfection and overexpression of focal adhesion kinase in 2 NF2-deficient mesothelioma cell lines.150 A relationship between NF2 expression and apoptosis in MM cells has been reported in other studies. In a study on the role of integrin-specific signaling in the control of apoptosis factors, NF2 was shown to have an inactivating role on integrin-dependent mTORC1 signaling.151 In this study, 11 MM cell lines were analyzed (of which 4 did not express merlin, while 7 did) for their activity in mTORC1, ERK, and AKT. While activation of ERK or AKT was not correlated with the loss of merlin or activation of mTORC1, inactivation of merlin promoted mTORC1 signaling independently of AKT or ERK.151 

Differences in the expression of genes involved in the ubiquitin/proteasome pathway have been observed between MM and normal tissue or according to histologic subtype. Several genes encoding proteasome complex subunits were upregulated in MPM tumors compared to normal parietal pleura.88 Other proteins involved in the ubiquitin/proteasome pathway, such as the FAS-associated factor (FAF1), which inhibits protein degradation of ubiquitinylated proteins, were recurrently altered at the genomic level in MM of p19^ARF (+/−) mice and were down-regulated in human MM.152,153 In peritoneal MM, several genes involved in the ubiquitin-proteasome pathway were upregulated in biphasic tumors compared to epithelioid tumors.154 In pleural MM, subunits of the proteasome complex (PSME3, PSMA3, and PSMA4) and ubiquitin-conjugating enzyme (UBE2S) were upregulated in the epithelioid phenotype variant compared to the sarcomatoid phenotype variant of the same MPM cell lines.155 

Several studies have analyzed the impact of proteasome inhibitors on MPM malignancy in preclinical models. Bortezomib (PS-341 or Velcade, Millennium Pharmaceuticals Inc, Cambridge, Massachusetts), a specific inhibitor of 20S proteasome activity, induces in vitro apoptosis and in vivo tumor growth inhibition in mice of 1 MPM cell line.156 Other proteasome inhibitors, PSI (A.G. Scientific Inc, San Diego, California) or MG-132 (EMD-CalBiochem, San Diego, California), were also shown to induce apoptosis in some MPM cell lines.157,158 With MPM cell lines in monolayer culture, bortezomib was shown to increase the cytotoxicity of chemotherapeutic agents.159 However, MM cell lines, when grown as multicellular spheroids, acquired resistance to apoptosis, induced by a combination of the proteasome inhibitor MG-132 and other apoptotic stimuli.160 Results of ongoing phase II clinical trials using bortezomib combined with cisplatin will indicate the efficacy of proteasome inhibitors in the management of MM (ClinicalTrials.gov identifier: NCT00458913).

Alteration of genes located at the INK4 locus, encompassing CDKN2A and CDKN2B, is a feature of human MM. Inactivation of these genes allows uncontrolled cell proliferation. While some MMs do not show mutation or methylation of these genes, another level of regulation could occur via deregulation of miRNA expression (see “MicroRNA Expression”). Several authors161–163 have developed experimental studies to try to restore cell cycle control in MM by adenovirus-mediated expression of p16^INK4A and p14^ARF in human MM cells and have found effects on both cell cycle progression and reduction of tumor growth in immunocompromised mice.

Cell cycle control can be affected in MM cells by the loss of other negative regulators, cyclin-dependent kinase (CDK) inhibitors, or by the overexpression of CDKs and cyclins (CCNs), and regulators of the mitotic checkpoints.85,88 The expression profile of 60 genes involved in cell cycle has been investigated in 45 MM tumor samples and normal pleural tissue.78 Among genes overexpressed in MM, several were involved in cell cycle checkpoints, such as CDK1/CDC2 (cyclin-dependent kinase 1), CDC6 (cell division cycle 6, a regulator of replication), CDKN2C (cyclin-dependent kinase inhibitor 2C, p18), CCNH (cyclin H), CCNB1 (cyclin B1, controlling the cell cycle at the G2/M transition), CHEK1 (Chk1 is required for checkpoint-mediated cell cycle arrest in response to DNA damage), and FOXM1 (forkhead transcription factor, a regulator of gene expression in the G2 phase). In contrast, CCND2 (cyclin D2, a regulator of Cdk4 and Cdk6, which controls the cell cycle at the G1/S transition) was underexpressed.78 Aurora kinases are involved in microtubule formation and are important regulators of the mitotic spindle checkpoint system, controlling progression of mitosis until all chromosomes are properly aligned during metaphase. An overexpression of aurora kinases has been reported in different studies.25,85 Aurora B levels increase after γ-irradiation, and MM cells arrest at the G2/M checkpoint of the cell cycle to repair DNA damage before proceeding through mitosis.164 Stathmin is also important for the evolution of mitosis, as it is involved in the regulation of microtubule dynamics by inhibiting the formation of microtubules and/or promoting their depolymerization. Kim et al165 identified potential genes involved in pathogenesis of MPM. They investigated 7 MM cell lines, fresh mesothelioma tissues, and adjacent normal pleural tissues by using cDNA microarray chips. Multiple genes were overexpressed in MM cell lines, compared to the human mesothelial cell strain LP-9 derived from the ascitic fluid of a patient with an ovarian carcinoma, and stathmin was one of the most strongly overexpressed genes.165,166 Protein expression of stathmin was observed in MPM tissues but not in matched normal pleural samples.165 

Because of these different alterations, response to DNA damage can be impaired in MPM cells entailing chromosomal instability. Well-controlled cell cycle progression is necessary for cells to respond to both endogenous and exogenous DNA damage. Although MPM cell cycle may be arrested in response to DNA-damaging agents, it may be assumed that MPM cells recover, most likely due to their inability to trigger the apoptotic mechanism. Moreover, heterogeneity exists between different tumors. After exposure to γ-radiation, human MPM cells were arrested either in 1 or more phases of the cell cycle, demonstrating heterogeneity in cell cycle control. G1 arrest was p21WAF1/CIP1- and p53-dependent.167 As mentioned in “Gene Mutations,” p53 can be inactivated in MPM, and its inactivation will facilitate chromosomal instability, in relation to loss of cell cycle control, especially in response to DNA damage. Regulation of p53 function occurs via posttranslational mechanisms and interaction with several proteins. MDM4 was recently shown to control p53 function in a human MM cell line.168 

Overall, these studies demonstrate that cell cycle dysregulation occurs in all phases, at the level of checkpoint control and related factors, thereby encouraging the search for stimulation of death pathways in MPM cells.

Malignant MM responds poorly to standard therapy.169 Mesothelioma tissue usually has a lower apoptotic index than that of other carcinomas,170 suggesting major defects in the apoptotic machinery. Apoptosis is mediated by 2 signaling pathways, the extrinsic and intrinsic pathways. The extrinsic pathway is initiated by death receptors, while the intrinsic pathway is triggered by internal apoptotic signals and involves the release of cytochrome c from the mitochondrial intermembrane space. These 2 pathways merge and share mechanisms of the caspase cascades.171 In the extrinsic pathway, the death receptor agonist TRAIL can induce apoptosis with a high specificity toward tumor cells and is currently being tested in clinical trials in a variety of human cancers. In mesothelioma, TRAIL has been shown to enhance the chemosensitivity of tumor cells to various therapeutic agents, such as doxorubicin, gemcitabine, cisplatinum, or etoposide. However, most MM cells are resistant to apoptosis induced by TRAIL alone.172 This resistance can be explained notably by overexpression of the caspase-8 inhibitor (FLIP/CFLAR) and by methylation of TRAIL receptors in MM cells.173 Several multimodal approaches have subsequently been applied to sensitize MM cells to TRAIL. Heat stress, as well as subtoxic doses of α-tocopheryl succinate or anisomycin can sensitize MM cells to TRAIL and induce apoptosis in vitro, via Bid-dependent mitochondrial amplification of the apoptotic signal.174–176 Inversely, the multikinase inhibitor sorafenib showed synergistic effects with TRAIL in cells resistant to TRAIL, independently of caspase activation.177 Interestingly, in contrast with mesothelioma cell monolayers, tumor fragment spheroids exhibit higher resistance to apoptosis and notably to TRAIL-combined treatments, and this resistance is mediated by the mTor/S6K pathway.160,178 In the intrinsic pathway, the mitochondrial membrane potential and permeability are regulated by the Bcl-2 family of proteins. Members of this family include both proapoptotic proteins such as Bax, Bak, Bad, Bid, or Bim and antiapoptotic proteins, such as Bcl-2, Bcl-xL, and Mcl-1. Bcl-2 is rarely expressed in mesothelioma,170 while high levels of Bcl-xL are commonly observed.179 Several studies have shown that down-regulation of Bcl-xL decreases baseline tumor cell viability and improves sensitivity to chemotherapeutic agents, both in vitro and in vivo.180–182 Mcl-1 has also been implicated in the apoptotic resistance of mesothelioma cells.158,179 Recently, Varin et al183 showed that Bcl-xL and Mcl-1 cooperate to protect mesothelioma cells from cell death and that their concomitant targeting is sufficient to induce apoptosis. Most members of the proapoptotic Bcl-2 family appear to be expressed in mesothelioma with functional integrity, suggesting that the loss of their apoptosis-inducing properties is due to sequestration by Bcl-xL or Mcl-1.184 In particular, functional inhibition of Bim contributes to survival in the spheroid model of mesothelioma cells.138 

The inhibitor of apoptosis protein (IAP) survivin, encoded by the BIRC5 gene, was highly expressed in all MM primary tumors (12 samples) and cell lines (7 of 8) compared with normal pleura.185 Survivin expression in 34 MM tumors was confirmed by immunohistochemistry and was linked to an apoptotic defect.170 Down-regulation of survivin with antisurvivin oligonucleotides induced apoptosis when tested in 1 cell line.185 Inhibition of survivin expression has been shown to decrease tumor cell growth and enhance drug response.186 XIAP is also frequently expressed in malignant mesothelioma and is notably upregulated in mesothelioma effusions and peritoneal mesothelioma.187 Moreover, XIAP inhibition has been shown to increase the sensitivity of mesothelioma cells to TRAIL-induced apoptosis.188 Together, these results suggest that combined approaches, triggering the extrinsic and intrinsic pathways or the caspase cascade, are promising for the treatment of mesothelioma.

Human telomeres progressively shorten during cell division, and critical shortening is believed to limit the cellular life span and is involved in conferring growth-promoting properties to tumor cells. Telomere lengthening is due to telomerase (TERT) activity, which was found in a large proportion of the 22 primary pleural MMs and the 4 MM cell lines, in comparison with mesothelial cells from normal pleura, with the telomeric repeat amplification protocol.189 These findings were confirmed in a more recent study190 carried out with peritoneal MM; another mechanism, alternative lengthening of telomeres, was also demonstrated to maintain telomere length. Interestingly, in their series of 44 MM peritoneal lesions from 38 patients, these authors found that telomerase activity was a significant prognostic factor for 4-year relapse and disease-free survival. Telomerase activity was reduced in MM cell lines in comparison with normal cells by inhibition of MetAP2 (methionine aminopeptidase) with angiostatic agents fumagillin and ovalicin. This enzyme is overexpressed in MM cells.191 

Molecular studies have identified somatic genetic and epigenetic alterations in MPM cells, associated with altered expression and activation or inactivation of critical genes in oncogenesis. Deregulation of signaling pathways related to differentiation, survival, proliferation, apoptosis, cell cycle control, metabolism, migration, and invasion has been demonstrated in complementary studies. These changes were found by investigating individual gene status in genomic and transcriptomic studies, and were supported by immunohistologic studies. Malignant pleural mesothelioma cells show a large spectrum of abnormalities shared with other malignancies, or more specific alterations such as those of the NF2 gene. Comparative studies of series of MPMs have usually demonstrated that both alterations in a given gene and combined genetic and epigenetic alterations are present in MPM subsets, consistent with interindividual variations of molecular alterations. There are therefore at least 2 levels of heterogeneity, at the genome level and at the gene level, suggesting that identification of patient subgroups would be essential to develop more specific therapies. Moreover, the tumor microenvironment, consisting of a large number of different cell types, adds another level of complexity in identifying the best strategy to improve the outcome of this disease. This tumor heterogeneity could explain differences in patient survival and response to treatments.

This review provides insight into a limited number of genes known to be frequently altered in MPM, INK4 locus and NF2, and a larger number of candidates that may play a role in MPM carcinogenesis, especially those involved in various signaling pathways. Further studies should define the clustering of these genes in specific MPM subsets. These findings have already been the basis for several studies testing various therapeutic approaches targeting specific RTKs, but mostly with limited success. Demonstration of the multiple alterations present in the tumor should encourage research into combined or more global therapies. Other studies have emphasized deregulation of signaling pathways, but no pathway seems to be specific or a particularly relevant target, as discrepancies have been observed in the response of MPM cells to specific inhibitors, and key regulatory players in one pathway may interact with another pathway. Focusing on apoptosis is probably an interesting strategy to counteract or trigger the activity of several of these pathways. More recent data have indicated the presence of alterations that could be targeted at a global level (methylation). Studies are ongoing to take advantage of these abnormalities for MPM treatment.

Prediction of a positive response in MPM would avoid a rapidly unfavorable course and avoid wasting time and resources with inappropriate treatments. The critical issue concerning targeted therapy is to focus on the most relevant target(s). Some molecules, pathways, and/or epigenetic changes should be selected, provided they are key factors in MPM. This is not an easy task, with the interplay between the various regulatory pathways and the diversity of genomic alterations. Molecular studies must be developed to identify and classify genomic alterations in MPM cells and to correlate these alterations with disease outcome in order to avoid random testing of therapies already used in other cancers, but with unknown relevance in MPM. In recent years, several studies have been designed to evaluate the predictive role of microarray data for MM outcome. Various authors have developed predictors of survival, but in some studies the accuracy was lower than that of prognosis based on the usual methods comprising clinicopathologic variables and morphology. Other authors have proposed innovative predictors based on gene expression ratios. These procedures are of great interest and deserve further validation.

Our improved understanding of MPM development and treatment is partly based on well-designed preclinical studies. Numerous in vitro investigations are currently underway to suppress MPM cell growth and/or induce apoptosis by interacting with proteins regulating proliferation and survival, or by silencing gene expression (RNA interference). These methods benefit from the data derived from molecular analyses providing preclinical proof of concept for the feasibility of such strategies. However, these studies have been carried out in MPM specimens that do not necessarily present the same genomic status as the tumors of patients selected for the relevant therapy. In the context of preclinical investigations, animal models must be combined with studies before translation to the human context. An important point to be emphasized here is the paramount importance of frozen and paraffin MPM tissue banks to allow better characterization and annotation of MPM, as well as panels for diagnostic certification. Databases and panels are already available, such as the Mesothelioma Virtual Bank192 (http://www.mesotissue.org, accessed June 11, 2011) or the International Mesothelioma Excellence Center (IM@EC).

In recent years, considerable methodologic progress has been made in the field of molecular approaches to study cancer biology, and this progress has been applied to MPM. Improvements are still ongoing. Other methodologies have not yet been applied to MPM, such as proteomics, cell imaging, and integrative biology and will most likely be useful in the future, to identify MPM biomarkers, exposure markers, and MPM subgroups.

Various clinical studies have shown that future treatment strategies must not be based on monotherapy, but must comprise multisite and multimodal treatment. As this disease is particularly aggressive, it requires a specific treatment strategy. Investigation of the tumor genome and related pathophysiologic events has therefore become a key step to a better understanding and possible cure of this dreadful incurable cancer.