Telomerase Reverse Transcriptase Promoter Mutations Identify a Genomically Defined and Highly Aggressive Human Pleural Mesothelioma Subgroup

Malignant pleural mesothelioma (MPM) is a rare, highly aggressive tumor arising from mesothelial cells lining the pleural cavity closely linked to asbestos exposure (1, 2). Additional risk factors include certain genetic predispositions like germline BRCA1-associated protein 1 (BAP1) gene mutations. Three histologic subtypes, epithelioid, sarcomatoid, and a mixed biphasic form, have been defined with the sarcomatoid type having the worst prognosis. Current treatment options include surgery, radiotherapy, and chemotherapy; however, median overall survival is only about 6–12 months and the 5-year survival rate is below 5% (1).

Genetically, MPM is characterized primarily by loss of key tumor suppressor and transcription regulator genes (1, 3) including BAP1, cyclin-dependent kinase inhibitor 2A (CDKN2A), neurofibromin 2 (NF2, merlin), tumor suppressor protein 53 (TP53), large tumor suppressor homolog 2 (LATS2), and RNA binding fox-1 homolog 1 (RBFOX1, A2BP1; refs. 2, 4). Gains/amplifications of driver oncogenes occur less frequently, further complicating the search for defined therapeutic targets (5). Despite increasing knowledge concerning molecular factors underlying MPM development (6), effective therapeutic options are still limited and clinically applicable prognostic markers urgently needed (7, 8).

Virtually all MPM are characterized by reactivation of telomerase allowing telomere stabilization and immortalization (9). Besides gene dose changes, epigenetic mechanisms, and the impact of deregulated oncogenic transcription factors (10), activating mutations in the telomerase reverse transcriptase (TERT) gene promoter are a major cause for telomerase rejuvenation in several cancer types (11). These mutations occur at -124 (-124C>T), -146 (-146C>T), or -57 (-57A>C) base pairs upstream of the TERT gene–coding region on chromosome 5p15.33 (also designated C228T, C250T, and A161C in many reports). All three TERT promoter mutations create de novo binding sites for E twenty-six-specific (ETS) transcription factors (12). Besides creating similar ETS-binding sites, TERT promoter mutations have been shown to affect transcription differently and act through different mechanisms (13, 14). These noncoding mutations occur at highly variable frequencies in different cancer types and are widely missing, for example, in breast, colon, and lung cancer (10). Interestingly, TERT promoter mutations have been shown to associate with an unfavorable patient outcome in several cancer types (12, 15). More recently, increased TERT gene transcription and even telomerase activity in promoter-mutated tumors have been observed in most studies (16, 17). However, the exact mechanisms driving worse prognosis associated with TERT promoter mutations are so far unclear.

By analyzing two independent MPM patient cohorts, we aimed to dissect the association of TERT promoter mutation status with clinical parameters and patient prognosis. Moreover, by using an extended panel of MPM primo-cell cultures from surgical specimens, we aimed to unravel cellular and molecular changes associated with activating TERT promoter mutations in MPM. Besides a higher incidence in MPM with nonepithelioid histology, we found strong prognostic power of TERT promoter mutations in both histologic subgroups. Enhanced aggressiveness of TERT promoter mutation–positive tumors was reflected at the level of tumor cell biology by their higher propensity for in vitro immortalization. Besides moderately but significantly enhanced TERT gene expression, TERT promoter–mutated MPM cell models were characterized by reduced levels of chromosomal instability and lack of BAP1 alterations.

In total, 182 patients with MPM were analyzed. The Austrian cohort consisted of 83 patients from the Department of Thoracic Surgery, Medical University of Vienna (Vienna, Austria). The Croatian/Slovenian cohort included 76 Slovenian patients from the Department for Pulmonology, University Clinic Golnik, and 23 Croatian patients from the Department for Respiratory Diseases Jordanovac, University of Zagreb (Zagreb, Croatia). The study was approved by the Ethics Committees at the Medical University of Vienna (#904/2009) and the University Hospital Center Zagreb (#02/21AG). The Institutional Review Board of the University Clinic Golnik granted a waiver for the retrospective analyses. The studies were performed in accordance with the Declaration of Helsinki and all prospectively included patients gave informed written consent. More details are given in the Supplementary Materials and Methods.

DNA was extracted from formalin-fixed, paraffin-embedded samples, frozen tissue, or MPM cell pellets using the respective kits from Qiagen (Qiagen). Mutational status of the TERT core promoter region from position -27 to -286 from ATG start site was determined by PCR and Sanger sequencing as described previously, and data were analyzed by Geneious Pro 5.6.5 software with reference to the sequences from the NCBI gene database (chr5: 1,295,071–1,295,521, hg19 GRCH37; refs. 18, 19).

Establishment of MPM cell cultures from surgical samples and cell line authentication by short tandem repeat analyses were performed as published previously (20, 21). Cell cultures were regularly checked for Mycoplasma contamination using the Mycoplasma PCR detection kit G238 (Applied Biological Materials, Inc). For experimental procedures, cell models ranging between passage 4 and 20 were used. The length of time between thawing and experimental use did not exceed 2 months.

Array CGH analysis was performed as described previously (22) using 4 × 44K whole genome oligonucleotide-based microarrays (Agilent). Data visualization and evaluation is described in the Supplementary Materials and Methods. Data are openly available at ArrayExpress (https://www.ebi.ac.uk/arrayexpress/) under the accession number E-MTAB-8987.

Whole genome gene expression arrays were performed using 4 × 44K microarrays from Agilent as described previously (22). Feature extraction and data analysis were carried out using the Feature Extraction (version 11.5.1.1.) and GeneSpring (version 13.0) softwares, respectively. Data are openly available at ArrayExpress (https://www.ebi.ac.uk/arrayexpress/) under the accession number E-MTAB-8986.

Exome sequencing of MPM cell lines was performed using the Nextera DNA Exome Kit (Illumina TruSeq Rapid Capture Exome 45Mb) on a HiSeq 4000 (2 × 75 bp paired-end) at the Medical University of Vienna Biomedical Sequencing Facility. Data are openly available under the accession number E-MTAB-8989.

For confirmation of the detected mutations in MPM tumor tissue, Ion Torrent sequencing was performed. Details for both methods are provided in the Supplementary Materials and Methods.

qRT-PCR for TERT mRNA expression in MPM cell models was performed as published previously (15) and as described in the Supplementary Materials and Methods.

Genomic DNA of healthy tissues, MPM tumor tissues, and cell lines were analyzed for GSST1 or RBFOX1 deletions as described in detail in the Supplementary Materials and Methods.

Protein activity of telomerase of all MPM cell models was assessed by Telomerase Repeat Amplification Protocol (TRAP assay). In parallel, telomere lengths of the cell lines were analyzed by qPCR and results calculated relative to the ALT-positive osteosarcoma model SA-OS. Both methods are described in detail in the Supplementary Materials and Methods.

TERT promoter activity was analyzed by luciferase reporter assays in a TERT promoter wild-type and a TERT promoter mutated background as published recently (15). Knockdown of GABPA by siRNA (Dharamcon) was performed using ON-TARGETplus SMARTpool siRNA and Lipofectamine RNAiMAX Reagent (Thermo Fisher Scientific) for 48 hours. The respective nontargeting pool served as a negative control.

The impact of the telomerase inhibitor MST-312 (Sigma; ref. 23) and the ETS factor inhibitor YK-4-279 (Selleck Chemicals; ref. 24) was determined by MTT-based viability assay as described previously (15).

Analysis of total protein extracts of MPM cell models by Western blot as well as BAP1 staining by IHC is described in Supplementary Materials and Methods. Antibodies are given there as well.

SPSS and GraphPad Prism software packages were used for statistical analyses. Overall survival was defined as time between MPM diagnosis and death or, in censored patients, diagnosis and last follow-up date. Analysis of survival data and all other statistical analyses are described in the Supplementary Materials and Methods.

The TERT promoter region containing the respective activating mutation sites was sequenced in 182 MPM samples derived from two independent cohorts: one consisting of 83 Austrian and the other of 99 Croatian/Slovenian patients with MPM. TERT promoter mutations were detected in 19 of 182 (10.4%) MPM cases (selected sequencing charts are shown in Supplementary Fig. S1). Patient characteristics of the entire patient collection are given in Table 1 while those of the Austrian and the Croatian/Slovenian cohorts are separately described in Supplementary Tables S1 and S2, respectively. TERT promoter mutations were found at significantly higher frequency in patients with nonepithelioid histology both in the entire patient collective (5.5% vs. 22.2% in epithelioid vs. nonepithelioid histology), but also the two separate cohorts (6.8% vs. 29.2% in the Austrian and 4.4% vs. 16.7% in the Croatian/Slovenian cohort). Moreover, all patients with TERT promoter–mutated tumors and evaluable IMIG stage derived from both cohorts presented at a late disease stage (Table 1). Accordingly, within the Austrian cohort, these patients were less frequently treated in a multimodal therapeutic setting (Supplementary Table S1). With regard to the mutation type, 13/19 mutated MPM harbored the -124C>T (68.4%), 2/19 the -146C>T (10.5%), and 4/19 (21%) the -57A>C genotype. Strikingly, the distribution was different between the histologic subgroups. Comparable with many other tumor types (12, 25), the -124C>T mutation was strongly prevalent in the nonepithelioid cases (10/12 mutated tumors). However, in case of the epithelioid histology, the prevalence of the -124C>T mutation was equal with that of the rare -57A>C mutation (3/7 mutated cases each).

To assess the prognostic value of TERT promoter mutations, we performed Kaplan–Meier survival analyses (Fig. 1). Median overall survival in the entire patient cohort was 262 versus 469 days in TERT promoter mutated as compared with wild-type cases (P < 0.0001; Fig. 1A). Hence, this genomic marker was of higher prognostic value for an aggressive course of disease as compared with the histologic subtype, a known marker of MPM prognosis (353 vs. 459 days for nonepithelioid as compared with epithelioid cases; P = 0.01; Fig. 1B). TERT promoter mutations were significantly enriched in the nonepithelioid histology (Table 1; Supplementary Tables S1 and S2) and only insignificantly enriched in tumors with pleomorphic nuclear features (3/11; 27.3%). With regard to the quality as prognostic marker, the Austrian and Croatian/Slovenian patient collectives were also evaluated separately as test and validation cohorts. In the Austrian test cohort, we found a significantly shorter overall survival in patients harboring TERT promoter mutations (262 vs. 524 days, P = 0.0012; Fig. 1C, left). A comparable and significant difference was observed for the Croatian/Slovenian validation cohort (104 vs. 465 days, P = 0.0024; Fig. 1C, right). This confirms independence of the prognostic quality with regard to regional patient origin even despite differences in country-specific therapy settings. Importantly, even though being more frequent in nonepithelioid tumors, TERT promoter mutations were significantly associated with shorter overall survival in both the epithelioid (Fig. 1D, left; 340 vs. 510 days, P = 0.003) and the nonepithelioid (Fig. 1D, right; 199 vs. 412 days, P = 0.023) subgroup. To test whether presence of a TERT promoter mutation is an independent prognostic factor, we performed univariate as compared with multivariate Cox regression analyses of the entire patient cohort (Supplementary Table S3; Table 2). In univariate analysis, Karnofsky performance status, histology, IMIG stage, EORTC score, and TERT promoter status were significantly associated with patient survival. In the multivariate setting, TERT promoter mutations and nonepithelioid histology remained significant indicators of shorter overall survival (Table 2). Also, when including the EORTC performance score, available for 140 patients, in multivariate analysis, TERT promoter mutation status remained a significant predictor of poor prognosis (Supplementary Table S4). These data prove that TERT promoter mutations are a robust, independent prognostic factor in human MPM.

The strong and histology-independent negative prognostic impact of TERT promoter mutations suggests a major role of these noncoding, genomic alterations in the aggressive phenotype of human MPM. To clarify whether this is a consequence of altered cancer cell biology and dissect the underlying molecular mechanisms, we set up primo-cell cultures from 45 patient samples of the Austrian cohort (termed “original panel” in Supplementary Table S5). Immortalized cell lines were successfully generated from 22 (48.9%) patients. Cell line characteristics are summarized in Supplementary Table S5. The TERT promoter mutation status of the cell lines reflected the ones of the original tumors in all cases available for parallel analysis (n = 10) with exception of Meso71. This MPM cell line was derived from a recurrence after chemotherapy that harbored a -57A>C mutation neither detectable in the patient's blood nor in tumor tissue of the primary lesion before chemotherapy. Five of nine TERT promoter–mutated cell models harbored also wild-type alleles in accordance with the described monoallelic TERT expression from the mutated alleles (26), while 4 of 9 showed homozygous mutations, indicating LOH of the wild-type allele. Cell line generation was achieved in all nine (100%) mutated but only 13 of 36 (36.1%) wild-type cases, demonstrating a significant association of in vitro immortalization with TERT promoter status (Fig. 2B), not seen with histology or disease stage (Fig. 2C and D). Successful cell line propagation in vitro, however, obviously represents another strong negative prognostic factor itself. The survival differences between patients with cell line–forming and nonforming tumors remained statistically significant (Fig. 2A; median overall survival 268.5 vs. 607 days, P = 0.0008), even after exclusion of the nine TERT promoter–mutant patients (not shown; median overall survival 313.7 vs. 577.3 days, P = 0.014). Interestingly, the presence of pleomorphic features was associated with successful cell line establishment (6/7; 85.7%; Fig. 2E). The relations between TERT promoter status, cell line formation capacity, histology, pleomorphic features, and stage at time of cell culture are opposed to overall survival data on single patient level in Fig. 2E. Moreover, C-reactive protein (CRP) levels, another marker of poor MPM prognosis (27), were significantly higher in patients with TERT promoter mutated as compared with wild-type tumors (7.3 ± 2.2 mg/dL vs. 3.8 ± 0.6 mg/dL; P < 0.05; Supplementary Fig. S2). Together, these observations suggest that the worse prognosis of patients with MPM with TERT promoter mutations is, at least in part, attributable to a more aggressive cancer cell biology.

Reactivation of TERT expression has been repeatedly found in virtually all MPM tumor samples analyzed (9, 28). Hence, we tested whether TERT promoter mutations are associated with altered TERT gene expression, telomerase activity, and telomere lengths. All MPM cell models expressed detectable levels of TERT mRNA and telomerase activity. Despite a significant correlation between these two parameters (linear regression; r = 0.44, P = 0.04), only the mRNA transcript level but not telomerase activity was significantly higher in the TERT promoter–mutated cell lines (P = 0.006; Supplementary Fig. S3A; P = 0.07; Supplementary Fig. S3B, respectively). Accordingly, TERT promoter status was not associated with altered telomere length, although all rare cases with very long telomeres were TERT promoter wild-type (Supplementary Fig. S3C). To estimate whether TERT promoter mutations indeed drive enhanced TERT mRNA expression, luciferase experiments with wild-type and mutated promoter constructs were performed in MPM cell models with endogenous wild-type (VMC48) or -124C>T–mutated (VMC20) TERT promoter background. Presence of either activating mutation -124C>T or -146C>T significantly increased promoter activity both in the wild-type and the -124C>T–mutated background. Interestingly, the effect was more robust using the -124C>T promoter construct in the -124C>T–mutated MPM background while both mutations were equally effective in the wild-type model (Supplementary Fig. S4A). Knockdown of the ETS factor GABPA, a key factor for activation of the mutated TERT promoter (29), resulted in massive downregulation of TERT promoter activity when using the two mutated but not the wild-type constructs. The effect was, however, significantly stronger for the -124C>T but not the -146C>T construct when transfected into the MPM cell model with mutated (VMC20, -124C>T) as compared with the wild-type background (VMC48; Supplementary Fig. S4B). This suggests that cellular parameters like the transcription factor expression pattern might be adapted for ideal utilization of the endogenous mutated TERT promoter sequence. Accordingly, expression of several ETS transcription factor genes at the mRNA level, including GABPA, ETS1, and FLI1, tended to be higher in mutated as compared with wild-type models (Supplementary Fig. S4C).

To investigate whether the TERT promoter status might predict sensitivity against telomerase inhibition, MPM cell models with either wild-type or mutated TERT promoter (9 each) were analyzed for responsiveness to the telomerase inhibitor MST-312 (23) and the ETS factor inhibitor YK-4-279 (24). The mean IC₅₀ values for the MPM models with mutant TERT promoter were moderately but significantly lower as compared with the wild-type subgroup in case of MST-312, while for YK-4-279 only an insignificant trend in this direction was detected (Supplementary Fig. S5).

As mentioned above, the enhanced TERT mRNA expression in TERT promoter–mutant mesothelioma cell models did not fully translate into significantly augmented telomerase activity (compare Supplementary Fig. S3). Moreover, TERT mRNA expression positively but only weakly correlated with shorter patient survival in the TCGA database in silico analysis (Supplementary Fig. S6). Clearly, this effect was minor as compared with the predictive power of the TERT promoter mutations in our patient cohorts (compare Fig. 1). Therefore, we hypothesized that the level of TERT expression and telomerase activity at the time of surgery might not represent the sole factor driving aggressiveness in TERT promoter–mutated MPM. Array CGH analyses performed for the original MPM cell line panel (n = 22, Supplementary Table S5) revealed that, unexpectedly, TERT promoter wild-type cell lines (n = 13) exhibited distinctly and significantly (P = 0.016) enhanced chromosomal alterations (35.3% genome affected by gains/losses) as compared with the panel of mutated models (n = 9; 16.6% genome affected). While the amount of chromosomal alterations was rather comparable regarding mutated models, a clearly higher variability was detected within wild-type models (Fig. 3A and B). No differences were detected with regard to chromosome 5p harboring the TERT gene locus, where both groups showed gains in more than 50% of cases (Fig. 3B). Interestingly, the higher variability of the wild-type cell models involved genomic gains as well as losses (Fig. 3A). Moreover, differences were not randomly distributed but concerned distinct genomic regions including chromosomes 1, 5q, 9p, 7, 14, and 20 (Fig. 3B, top). Three affected chromosomes (chromosomes 1, 9, and 14) are enlarged in Fig. 3B (bottom). Quantification of aberrations at the single probe–based level for chromosome 1, 9p, and 14 is given in Fig. 3C. While at chromosome 1 the gains observed for the TERT promoter wild-type models were widely missing in the mutated cell lines, much stronger losses were found on chromosome 14 in the mutated cells. The characteristic loss at chromosome 9p including the CDKN2A locus was narrow and focused in case of the TERT promoter–mutant cells but highly variable and much broader in the wild-type panel, affecting frequently multiple adjacent genes. This suggests a higher chromosomal stability of MPM cells harboring TERT promoter mutations.

An extended MPM cell line panel (N = 26; Supplementary Table S5) was analyzed by exome sequencing (sequencing depth >100×) to explore a potential relationship of TERT promoter mutations with known genomic alterations of MPM including mutations/loss of BAP1 and NF2 (Fig. 4; for mutations of BAP1, CDKN2A, and NF2 compare Supplementary Table S6). Generally, although a tendency towards a lower mutational burden in TERT promoter–mutated samples was detected, this difference did not reach statistical significance (see Supplementary Fig. S7 for the coding variants). The general mutational pattern in MPM cell lines was equally detected in the original tumor tissues using Ion Torrent sequencing (Fig. 4B). Interestingly, neither loss/mutations of the CDKN2A locus nor of NF2 or TP53 showed any association with TERT promoter status or MPM histology (Fig. 4A and B). In contrast, BAP1 mutations (5/26; 19.2%) or deletions (3/26; 11.5%) were exclusively found in MPM cell models lacking TERT promoter mutations (8/17; 47.1%; Fig. 4A and B). BAP1 protein expression by Western blot was completely absent in seven cell extracts, of which four harbored a BAP1 mutation and three a gene deletion, and only one sample with a BAP1 point mutation (Meso221) showed detectable protein levels (Supplementary Fig. S8A). Accordingly, both BAP1 mRNA and protein expression levels were lower in the TERT promoter wild-type models (Supplementary Fig. S8B, top) and inversely correlated with overall survival (Supplementary Fig. S8B, bottom). Cell line data widely correlated with BAP1 expression in the corresponding tumor specimens detected by IHC (Fig. 4B; Supplementary Fig. S9). Loss of nuclear BAP1 in tumor samples (n = 75, Austrian cohort) was clearly associated with epithelioid histology (Supplementary Tables S7 and S8) and wild-type TERT promoter status (Supplementary Table S9).

While no significant associations with the TERT promoter status were found for the frequencies of other mutations/deletions for MPM described previously (3), two distinct deletions concerning the splicing regulator RBFOX1 and the glutathione S transferase GSTT1 were strongly related to both TERT promoter status as well as tumor histology (Fig. 4A and B). Deletion of the GSTT1 locus is a well-known germline polymorphism, which has been shown to increase the risk for severe fibrotic changes upon asbestos exposure (30). In contrast, RBFOX1 deletions at 16p13.3 represent a cancer phenomenon and have been reported for example in glioblastoma, where RBFOX1 is crucial for terminal cell differentiation (31). Concerning our mesothelioma cases, RBFOX1 deletions were solely found in the malignant, but never the corresponding nonmalignant tissues. In case of GSTT1 deletions, only 2 of 10 affected patients also harbored a germline deletion (Fig. 4B). RBFOX1 deletions occurred either in TERT promoter–mutated (7/9) or nonepithelioid (3/17) samples, but were completely absent in TERT promoter wild-type, epithelioid cell lines, an association also seen for the GSTT1 deletion.

In this translational research study, we describe a strong and independent negative prognostic impact of activating TERT promoter mutations in human MPM. Besides a clear-cut enrichment in nonepithelioid specimens, the prognostic quality remained significant also in patients harboring tumors of epithelioid histology, despite the relatively low number of affected cases. This negative impact on overall survival was equally found in two independent patient cohorts from Austria and Croatia/Slovenia, despite differences in stages of disease and treatment modalities. This indicates that TERT promoter mutations might represent a novel marker for worse prognosis of human MPM, as recently suggested for several other cancer types including gliomas (32). Our data are in accordance with a very recent report concerning the Inserm series, also identifying TERT promoter mutations in a comparable percentage of MPM cases. Mutations were significantly associated with nonepithelioid histology and poor patient prognosis mainly within a three gene mutation signature (28, 33).

Concerning the TERT promoter mutation types in MPM, preferentially the -124C>T mutation, predominant in most cancers (10), was detected. However, in contrast to the French study (33), we also found the less frequent -146C>T mutation in two cases of our MPM cohort. More prevalent was, surprisingly, the very rare -57A>C mutation, with three affected cases concerning the rarely mutated epithelioid histology. Originally, -57A>C has been described as a causal high-penetrance germline mutation in a melanoma-prone family (11) and only occasionally as a somatic event in melanoma and bladder cancer (34, 35). Thus, relatively high prevalence of this mutation type might be specific for epithelioid MPM.

To investigate the impact of TERT promoter mutations on MPM cell characteristics, we set out to establish permanent cell cultures from part of the Austrian MPM sample collection. Interestingly, TERT promoter mutations were strongly enriched in samples with in vitro immortalization potential. Corresponding associations of TERT promoter mutations and more efficient in vitro immortalization have been reported for mixed brain tumors (36), thyroid cancer (37), and by our group for meningioma (15). In addition, our data demonstrate that effective in vitro cell immortalization itself might represent a marker for high MPM aggressiveness, even after exclusion of the TERT promoter–mutated cases. One underlying factor might be the comparably high prevalence of samples with pleomorphic nuclear features, known to be associated with unfavorable prognosis (38), in the TERT promoter wild-type cell line–forming (3/13; 23%) as compared with non-cell line–forming (1/23; 4.3%) cases.

An association of TERT promoter mutations with worse clinical outcome has been observed in multiple cancer types (12). However, the underlying factors are not entirely clear so far. Besides telomerase-mediated replicative immortality, also noncanonical functions of TERT (10) in cancer cell motility, stemness and therapy resistance have been suggested as underlying mechanisms (39). In our study, MPM cell models with TERT promoter mutations were significantly more responsive to telomerase inhibition than wild-type cases, suggesting an enhanced dependency on telomerase activity in this MPM subgroup. In addition, ETS factor expression profiles and luciferase reporter data indicated that TERT promoter–mutated MPM cells might be ideally adapted for optimal promoter activation, for example by binding of the ETS factor GABPA (29).

However, based on our data, it is questionable whether enhanced TERT promoter activity at the time of diagnosis is the sole mechanism underlying the dismal prognosis of patients harboring TERT promoter mutated tumors. Generally, all MPM are telomerase positive and several wild-type MPM models in this study displayed comparable TERT mRNA expression and telomerase activity levels as the mutated ones. In addition, the enhanced TERT mRNA expression in TERT promoter–mutated cases did not translate into pronounced augmentation of telomerase activity and no correlation with telomere lengths was observed. In search for additional oncogenic driver mechanisms, we analyzed whether defined chromosomal alteration or mutation patterns might be associated with TERT promoter mutations in human MPM cell cultures. Generally, our array CGH and whole-exome sequencing data are well in agreement with previous studies, with losses at chromosomes 4q, 9p, 13, 14, and 22; gains in chromosomes 5p and 7p; as well as mutations/deletions of CDKN2A, NF2, BAP1 and TP53 as predominant genomic alterations in human MPM (2, 3, 40, 41). Interestingly, we found distinct differences in the general pattern of chromosomal alterations and at defined gene loci associated with TERT promoter mutations. Most strikingly, TERT promoter–mutated MPM samples showed a distinctly lower level of chromosomal instability. These results were rather unexpected, as Ivanov and colleagues reported association of enhanced chromosomal instability with poor survival of 22 patients with MPM (40). The underlying mechanisms are unknown so far, but it has to be considered that the TERT promoter–mutated MPM subgroup is rather minor and as such might have been missed in smaller patient collectives. Moreover, our MPM cell model approach clearly selected for TERT promoter–mutated MPM cases, significantly prone to immortalize with higher efficacy. In addition, the differences in chromosomal aberration patterns suggest that TERT promoter mutations might be associated with an alternative malignant transformation process. Accordingly, it has been shown that TERT promoter mutations time-dependently supported immortalization especially by stabilizing short telomeres at a critical length, hence modifying telomere shortening–associated crisis (42). It might be hypothesized that early acquisition of a TERT promoter mutation in a small subgroup of MPM may circumvent critical telomere shortening during malignant transformation and hence avoid massive chromosomal rearrangements. Accordingly, a study about clonal evolution of glioblastoma cells revealed TERT promoter mutations as an early event during malignant progression shared by all tumor cell subclones (43).

While several well-known MPM alterations (CDKN2A mutations/deletions, NF2 and TP53 mutations) did not associate with TERT promoter status, BAP1 mutations/deletions were exclusively found in TERT promoter wild-type MPM models. This association was confirmed in a large series of surgical specimens by IHC BAP1 detection. Accordingly, BAP1 mRNA expression was lower in the TERT promoter wild-type cells, and samples completely missing BAP1 protein expression were only found within this genotype. Interestingly, the deubiquitylase BAP1 has been shown to prevent chromosomal instability by several mechanisms (44). These include cell death induction following DNA damage (45), deubiquitination of γ-tubulin (46), the centrosome protein MCRS1 (47), and the chromatin remodeling molecule Ino80 in several human cancer cell models including mesothelioma (48). Thus, exclusiveness of BAP1 loss with TERT promoter mutations might be another explanation for the lower chromosomal instability observed in the latter subgroup. Accordingly, BAP1 loss might support telomerase activation by other mechanisms than promoter mutations based on accelerated chromosomal instability. In agreement with our results, mutual exclusivity between BAP1 and TERT promoter mutations was observed within the clinical Inserm MPM series (33), while, in a MPM cell culture collection from the same research group, 3 of 12 cell models harbored both alterations (28). The reasons for this discrepancy are currently unclear. Interestingly, in urothelial carcinoma the very frequently occurring TERT promoter mutations did not show any association with the BAP1 mutation status (49), while in liver cancer with biliary phenotype, as in our study in MPM cell models, exclusiveness of these two mutations was found (50). It might be hypothesized that the better prognosis of BAP1-mutated mesothelioma cases (2) might be, at least partly, based on the lack or reduced frequency of BAP1 alterations in the small but very aggressive subgroup of patients with TERT promoter–mutated MPM.

Two gene deletions, namely RBFOX1 (also designated A2BP1) and GSTT1, were significantly enriched in TERT promoter–mutated MPM cell models. Moreover, the only two cell models with tumor necrosis factor (TNF) receptor–associated factor 7 (TRAF7) mutations (compare Fig. 4), previously detected in about 2% of MPM (3), both harbored the -57A>C mutation and were of epithelioid histology. Together, these data might reflect an altered propensity in the TERT promoter–mutated MPM subset to undergo gene deletions and to associate with specific genomic alterations.

Summarizing, we report that activating TERT promoter mutations identify a specific subgroup of aggressive human MPM characterized by reduced chromosomal instability and lack of BAP1 mutations. TERT promoter mutations represent an independent prognostic marker for dismal patient survival. After validation in prospective studies, this marker might support better estimation of mesothelioma patient prognosis.

M. Jakopovic reports receiving speakers bureau honoraria from AstraZeneca, Boehringer Ingelheim, Roche, Pfizer, and MSD. M. Samarzija reports receiving speakers bureau honoraria from Roche, AstraZeneca, Boehringer Ingelheim, Pfizer, MSD, and Novartis. K. Mohorcic reports receiving speakers bureau honoraria from MSD, Pfizer, Roche, AstraZeneca, Boerhinger Ingelheim, Bristol-Myers Squibb, and Abbvie. No potential conflicts of interest were disclosed by the other authors.

Conception and design: C. Pirker, M. Grusch, R. Kumar, B. Hegedus, W. Berger

Development of methodology: C. Pirker, B. Heidenreich, D. Lötsch-Gojo, S. Spiegl-Kreinecker, M.A. Hoda, L. Müllauer, R. Kumar, W. Berger

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): C. Pirker, A. Bilecz, B. Heidenreich, V. Laszlo, P. Stockhammer, D. Lötsch-Gojo, S. Spiegl-Kreinecker, A. Steindl, T. Klikovits, M.A. Hoda, M. Jakopovic, M. Samarzija, K. Mohorcic, I. Kern, B. Kiesel, L. Brcic, F. Oberndorfer, L. Müllauer, R. Kumar, B. Hegedus, W. Berger

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): C. Pirker, A. Bilecz, T. Mohr, P. Stockhammer, J. Gojo, L. Gabler, T. Klikovits, M. Jakopovic, M. Samarzija, K. Mohorcic, F. Oberndorfer, L. Müllauer, W. Klepetko, W.M. Schmidt, R. Kumar, B. Hegedus, W. Berger

Writing, review, and/or revision of the manuscript: C. Pirker, A. Bilecz, M. Grusch, V. Laszlo, P. Stockhammer, J. Gojo, L. Gabler, B. Dome, A. Steindl, T. Klikovits, M.A. Hoda, M. Jakopovic, M. Samarzija, I. Kern, L. Brcic, W.M. Schmidt, R. Kumar, B. Hegedus, W. Berger

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): C. Pirker, A. Bilecz, T. Klikovits, M.A. Hoda, M. Jakopovic, M. Samarzija, B. Kiesel, L. Brcic, F. Oberndorfer, B. Hegedus

Study supervision: C. Pirker, R. Kumar, B. Hegedus, W. Berger

This work was supported by the Austrian Science Fund (FWF, I2872, to B. Hegedus; T906-B28, to D. Lötsch-Gojo; and I3522-B33, to V. Laszlo), the Hungarian National Research, Development and Innovation Office (K109626, K108465, KNN121510, and SNN114490, to B. Dome), the Vienna Fund for Innovative Interdisciplinary Cancer Research (to B. Dome), the Medical Scientific Fund of the Mayor of the City of Vienna (17028, to T. Klikovits), and the EFOP (3.6.3-VEKOP-16-2017-00009 fund, to A. Bilecz). We thank Petra Vician, Gerald Timelthaler, Jennifer Hsu, Mirjana Stojanovic, Barbara Dekan, Andreas Wagner, and Violetta Piurko for competent technical assistance and cell culture establishment (https://doi.org/10.13039/501100002428).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.