Telomerase Activation and ATRX Mutations Are Independent Risk Factors for Metastatic Pheochromocytoma and Paraganglioma

Pheochromocytomas and paragangliomas (PPGLs) are neuroendocrine tumors arising from adrenal medulla and paraganglia, respectively. Although the majority of PPGLs never progress, it has been estimated that up to 20% can develop overt metastases (1).

Metastatic PPGLs represent a major clinical challenge due to the limitations in accurate diagnosis and effective treatments. Indeed, reliable tumor biomarkers are still lacking to distinguish, at the time of diagnosis, tumors that will remain benign from those that will progress to metastasis. Consequently, the diagnosis of metastasis remains the only definitive criterion to define malignant PPGLs according to the most recent World Health Organization (WHO) classification (2) and a life-long follow-up is recommended for every patient presenting with PPGLs, despite the fact that most are usually cured after surgery.

PPGLs are characterized by a remarkable genetic determinism with up to 40% of cases explained by germline mutations in 14 susceptibility genes comprising SDHA, SDHB, SDHC, SDHD, SDHAF2 (collectively referred to as SDHx), FH, MDH2, SLC25A11 (3), EPAS1, VHL, NF1, RET, TMEM127, and MAX (4, 5). Importantly, SDHB/FH mutations are associated with high risk of metastatic progression and poor prognosis (6–8), with about half of the mutation carriers presenting synchronous or metachronous metastases. Clinical characteristics such as primary tumor size and extra-adrenal location have been suggested as risk factors as well (9).

Genomics studies revealed that PPGL tumorigenesis is mainly driven by germline or somatic mutations in susceptibility genes, but they failed to identify recurrent genetic alterations linked to metastatic progression (10). Transcriptomic unsupervised classification established 3 main clusters: Krebs cycle cluster C1A, comprising tumors at high risk of metastatic progression (SDHx, FH-, MDH2- and SLC25A11-mutated), the pseudohypoxic cluster C1B mostly specific of VHL-mutated tumors, and the kinase signaling cluster C2 comprising NF1-, RET-, HRAS-, TMEM127-, MAX-mutated tumors (11).

We hypothesized that immortalization in primary PPGLs could be an important risk factor for progression toward metastasis. Immortalization is achieved by reactivation of telomerase in about 85% of human carcinomas (12) or by a recombination-based alternative lengthening of telomeres (ALT) pathway mainly in sarcomas and tumors arising from endocrine and neural tissues (13). The specific contribution of these immortalization mechanisms to PPGL progression remains undetermined.

Telomerase activation can be assessed by evaluation of TERT expression (14). However, given the possible lymphocytic infiltration in tumors, expression of TERT alone should be taken with caution and additional efforts are required to analyze underlying mechanisms such as TERT promoter mutations (15), hypermethylation (16), amplifications (17), and genomic rearrangements (15). On the other hand, the genetic basis of ALT tumors remains poorly defined, although mutations in ATRX have been reported to be associated with this phenotype (18).

Early studies reported telomerase overexpression in malignant pheochromocytomas (19–23), whereas TERT promoter mutations (24, 25) and hypermethylation (26) were recently found in a few SDHx-deficient paragangliomas. The phenotypic assignment of ALT based on telomeric analysis of histological preparations remains limited to 2 discrepant studies reporting a prevalence of 4% (27) and 27% (28) of 75 and 22 PPGLs, respectively. Also, mutations in ATRX, were reported to occur in 12.6% of PPGLs, some associated with ALT and aggressive features (28). However, fewer mutations were detected in exome sequencing studies (29–33). So far, no systematic study has concomitantly analyzed both telomere maintenance mechanisms, which is necessary to determine the actual prevalence of immortalization and its relative contribution for metastatic progression.

Here, we performed a comprehensive analysis of immortalization in a well-characterized series of 200 PPGLs by combining previous multiomics data (29) with experimental validations (Supplementary Fig. S1). We aimed at identifying telomerase (+) and ALT (+) tumors while ascertaining their association with underlying mechanisms, genetic status, clinical features and outcomes of affected patients. We provide solid evidence that telomerase activation and ATRX mutations help to discriminate metastatic from non-metastatic tumors in high-risk PPGLs, thus suggesting a great clinical potential for diagnostic and eventually for therapeutic purposes.

Two hundred tumor samples from 190 patients with PPGLs, collected by the French “Cortico et Médullosurrénale: les Tumeurs Endocrines” (COMETE) network (34) were analyzed in this study. The study was conducted in accordance with the Declaration of Helsinki. Ethical approval for the study was obtained from the institutional review board [Comité de Protection des Personnes (CPP) Ile de France III, June 2012]. Written informed consent for the sample collection and subsequent analyses was obtained from all patients. Mutation status (germline or somatic) for the main PPGL susceptibility genes and integrative genomic characterization of the cohort were previously reported (29). For this study, we also identified 7/200 (3.5%) MAML3 (+) tumors (Supplementary Fig. S2). Primer sequences for amplification of UBTF–MAML3 fusion transcripts are provided in Supplementary Table S1.

The datasets analyzed during the current study are available in the following repositories: Gene-expression profiling (ArrayExpress entry E-MTAB-733); copy-number alterations (ArrayExpress entry E-MTAB-2817); whole-genome DNA methylation [Gene Expression Omnibus (GEO) entry GSE43298]. Whole-exome sequencing data [European Genome-phenome Archive (entry EGAS00001000933)].

Total RNA was purified with the miRNeasy Mini Kit (Qiagen) and cDNAs prepared with the Superscript III Kit (Thermo fisher scientific). Quantitative PCR was performed with iTaq Universal SYBR Green Supermix (Bio-Rad). Primer sequences are provided in Supplementary Table S1. For absolute quantifications of TERT variants, cDNA amplicons were purified from agarose gels using NucleoSpin Gel and PCR Clean-up Kit (Macherey-Nagel) to be used as standards. Cycling conditions were as reported (35).

Amplicons of 163bp were amplified from tumor DNAs using a set of primers listed in Supplementary Table S1 to target the TERT promoter region (chr5:1295151-1295313). PCR was performed with KAPA HiFi HotStart ReadyMix kit (Kapa Biosystems) using 50-ng genomic DNA. Sequencing reactions in both directions were performed in an Applied Biosystems 3730xl DNA Analyzer. TERT mutations were confirmed to be present only at the somatic level. For analysis of promoter methylation, pyrosequencing of a region of 36 bp containing 5 CpG sites (chr5:1295586-1295621;GRCh37/hg19) was amplified from sodium bisulfite modified genomic DNA and analyzed as described previously (16). Primers are listed in Supplementary Table S1.

To identify TERT rearrangements, PCR-free libraries were prepared with NEBNext Ultra II DNA Library Prep Kit following supplier recommendations. Briefly, 300-ng double-strand gDNA from tumor sample CIT_015 were fragmented using a sonication method to obtain 400-bp average sizes. After ligation with paired-end adaptor oligonucleotides (xGen TS-LT Adapter Duplexes from IDT), fragments were purified for direct sequencing. DNA PCR-free library sequencing was performed on an Illumina HiSeq 4000 instrument with a paired-end read length of 150 nucleotides. Base calling and image analysis were performed using Illumina Real Time Analysis (RTA) Pipeline version 1.12.4.2 with default parameters. The Manta software was used for identification of somatic structural variants nearby the TERT locus.

50 ng of tumor DNAs were spotted onto Biodyne B nylon membranes in alkaline conditions, hybridized to an oligonucleotide labeled with digoxigenin (DIG) and revealed for the DIG signal following Roche instructions. After stripping, membranes were rehybridized with DIG labeled total genomic DNA from HCA2 primary cells. Quantification of telomeric and genomic signals was performed using the MultiGauge software. Telomeric intensities were normalized with the genomic signal to correct for the differences in ploidy.

Detection of ALT-associated PML bodies (APBs) was based on colocalization of a telomeric FISH signal with the PML protein. Tissue sections (4 μm) were steamed with citrate buffer and serially incubated with anti-PML antibody (Chemicon, AB1370) diluted 1/1,000, and goat anti-rabbit alexa 488 antibody diluted 1/100. Slides were post fixed with 3.7% formaldehyde and dehydrated. Hybridization with a Cy3-labeled peptide nucleic acid (PNA) telomere probe (Panagene, F1002) was performed overnight. Slides were imaged with a Nikon 80i epifluorescence microscope. The following ALT characteristics were evaluated: (i) dramatic cell-to-cell telomere length heterogeneity, (ii) the presence of large, ultra-bright nuclear foci of telomere FISH signals in tumor cells, and (iii) colocalization of PML protein with telomeric FISH (APBs).

All coding regions of ATRX were amplified using the MASTR Plus technology (Agilent technologies) to allow their analysis using Next-Generation Sequencing (NGS). Amplicon libraries from different samples were further processed by bridge amplification followed by sequencing on the MiSeq Instrument using the MiSeq reagent kit v2, 300 cycles. NGS data were analyzed using the SeqNext module v4.3.1 (JSI medical systems) and the PolyDiag pipeline developed by the Paris Descartes bioinformatic platform. Mutations with a ratio >10% were considered relevant. ATRX mutations were confirmed to be present only at the somatic level.

Associations between molecular and clinical characteristics were evaluated with 2-sided χ² or Fisher-exact tests with significance set to P < 0.05. Differential analysis was determined using moderated T-tests or ANOVA models for multigroup comparison. In case of significant P values for ANOVA tests, we calculated pairwise comparisons with corrections for multiple testing. Local false discovery rate was used to control for multiple testing using the Benjamini and Hochberg method. The significance of a Pearson coefficient computed on 2 quantitative covariates was estimated by a correlation test based on a Student t distribution of the Pearson correlation coefficient. Univariate and multivariate cox models were built to find covariates related to survival. Survival curves were calculated with the Kaplan–Meier method and differences between curves were determined using the log-rank test.

The study cohort consisted of 200 tumors corresponding to 166 (83%) pheochromocytomas, 28 (14%) paragangliomas, and 6 (3%) metastases, collected from 190 patients (115 women and 75 men) with a mean age at diagnosis of 42.53 years (range, 7–82 years). Twenty-three patients presented a metastatic disease at diagnosis or within the follow-up period (median of 7.7 years from the initial diagnosis) with an equal distribution between synchronous and metachronous metastases. The proportion of patients with metastatic disease was consistent with their genetic status: 52% were metastatic in the group at high risk of progression (12/23 SDHx/FH/SLC25A11-mutated, cluster C1A) and 6.6% in other molecular groups (11/167 non-SDHx/FH/SLC25A11-mutated, clusters C1B, C2A/B/C). Twenty-seven samples from metastatic patients correspond to 21 primary tumors and 6 metastases. Using genomic data previously generated for this cohort (29), we first searched which tumors achieved immortalization.

To identify telomerase (+) tumors we analyzed expression levels of TERT using transcriptome data (11). General expression levels of TERT were very low in all PPGLs. Anyhow, we found a significant increase in TERT in transcriptomic cluster C1A enriched in tumors at high risk of metastatic progression (SDHx/FH-mutated), compared with other clusters (Supplementary Fig. S3A). Importantly, SDHx metastatic cases display the highest TERT expression even when compared with nonmetastatic tumors with the same genotype (Supplementary Fig. S3B). Given that full-length TERT can be co-expressed with a splice variant lacking the catalytic domain, the so-called β-deletion (35), we performed absolute quantifications of these variants using RT-qPCR. This validation revealed equivalent abundance of both full-length and β-deletion transcripts (≥1.5 × 10⁻⁴ copies) in 9 of 14 metastatic tumors from the high-risk group (Fig. 1A) and in 3 of 10 metastatic tumors from the other molecular groups (Supplementary Fig. S3C).

There are 3 major genetic/epigenetic alterations that can lead to transcriptional activation of TERT in tumors: enhancing promoter mutations (15), hypermethylation of the promoter region that includes the so so-called THOR region (TERT hypermethylated Oncological Region; ref. 16), and amplification of the locus (17). To determine the mechanisms underlying telomerase activation, we sequenced the TERT promoter region in tumor DNAs of the 200 PPGLs. We found 1 SDHC-mutated benign tumor carrying a non-enhancing C228A mutation (15) and 6 SDHx-mutated metastatic tumors carrying an enhancing C228T mutation (Fig. 1B). Interestingly, analysis of 3 samples from 1 of the individuals revealed that the C228T mutation was not present in the primary tumor, whereas it appeared in 1 out of the 2 metachronous metastases. Further analysis of methylation levels at the TERT promoter region using methylome data (36) revealed hypermethylation in 5 SDHB-mutated metastatic cases, including the aforementioned metachronous metastases (Supplementary Fig. S4A). Hypermethylation was significantly correlated with expression of TERT (Fig. 1C) and covered CpG islands spanning 4 Kb, including the so-called THOR region (ref. 16; Supplementary Fig. S4B), a finding confirmed using pyrosequencing (Supplementary Fig. S4C). Next, by examining the SNP array data, we identified copy-number gains involving the TERT locus (5p15.33) in 2 SDHB-mutated metastatic tumors (Fig. 1D). Interestingly, these samples also exhibit hypermethylation at the THOR region and presented the highest expression levels of TERT. Finally, no chromosomal rearrangements within 100 kb nearby the TERT locus were identified by whole-genome sequencing in the metastatic sporadic tumor CIT_015.

Taking into account a high co-expression of TERT splice variants, which was coupled to an underlying mechanism in most cases, we assigned a telomerase (+) status to 12 PPGLs, all of which were metastatic. Interestingly, 9 of 12 were tumors classified in the group at high risk of progression (P = 2.42e−09).

ALT (+) tumors are characterized by displaying long and heterogeneous telomeres. We first estimated the telomere length of each tumor by applying slot blot analysis, and we focused on the 48 tumors exhibiting long telomeres, defined as those that fall above the third quartile (Fig. 2A). Next, samples for which material was available (n = 22 with long telomeres and n = 7 with intermediate/short telomeres used as controls) were assayed for telomere heterogeneity using APB assays (Fig. 2B; ref. 13). Twelve out of the 22 tumors with long telomeres were classified as ALT (+), whereas samples with intermediate/short telomeres were negative for the presence of APBs (Supplementary Table S2). We found that tumor samples with long telomeres harbor the lowest ATRX mRNA expression (Fig. 2C). Therefore, we used this criterion to assign the ALT (+) status to 9 additional tumors in which we could not perform APB assays.

Finally, we sought to determine the prevalence of ATRX mutations using a next-generation sequencing approach. In addition to the tumor carrying a frameshift mutation in ATRX that we have described previously (29), we identified 7 PPGLs, with damaging mutations (Table 1). Among them, expression of ATRX was absent at the protein level as confirmed by immunohistochemistry in available tissues (Supplementary Fig. S5). ALT characteristics were present in 6 of 8 (75%) tumors harboring ATRX mutations (Table 1), of which 2 had already been identified in the APBs screening, and 4 were thus considered as new ALT (+) cases.

Altogether, we assigned the ALT (+) status to 25 out of the 200 (12.5%) PPGLs of the cohort. In support of an ALT phenotype, we found that these tumors exhibit, as expected, higher levels of telomere repeat-containing RNA (TERRA) than telomerase (+) samples (Fig. 2D). Of note, although the ALT mechanism was activated irrespective of metastatic status or tumor subtype (P = 0.169), ATRX mutations were more frequent in tumors at high risk of metastatic progression (P = 0.0014).

The exhaustive analysis of immortalization enabled us to determine that only 37 of 200 (18.5%) PPGLs activated a telomere maintenance mechanism (Fig. 3A), probably explaining why the great majority of these tumors never progress. Interestingly, telomerase activation appears to be more frequent in paraganglioma than in pheochromocytoma and the opposite was found for ALT (+) tumors (Fig. 3A). Furthermore, telomerase (+) but not ALT (+) tumors had a larger size than nonimmortalized tumors (P < 0.05; Fig. 3B).

It has been previously shown that tumor size (>5 cm), extra-adrenal location and SDHB-mutated status are clinical factors associated with metastatic progression and decreased overall survival (OS; ref. 9). Given that these variables appeared highly correlated with a telomerase (+) status, and given the tight association of ATRX mutations and the ALT phenotype in some metastatic tumors, we sought to identify which of these variables impact the most on the clinical outcome of affected patients.

Univariate Cox regression models showed a strong impact of all of these covariates except tumor size and ALT phenotype, on both metastasis-free survival (MFS) and OS. Remarkably, multivariate analysis performed on the remaining covariates revealed that only telomerase activation and ATRX mutations were independent risk factors: MFS (hazard ratio, 48.2 and 33.1; P = 6.50E−07 and 1.90E−07, respectively); OS (hazard ratio, 33.1 and 12.9; P = 2.60E−03 and 2.20E−03, respectively; Table 2). In fact, 18 out of 27 tumor samples (67%) from the 23 patients with confirmed metastatic status harbor 1 of these alterations regardless the tumor subtype. Of note, these alterations were present not only in 7 primary tumors from patients with synchronous metastases, but also in 7 primary tumors from 6 patients with metachronous metastases (Supplementary Table S3). When combined, detection of telomerase activation and ATRX mutations reaches the best sensitivity (0.70) and specificity (0.99; Supplementary Table S4), thus performing more accurately (AUC 0.84) than the previously suggested risk factors on the ability to discriminate metastatic from nonmetastatic PPGLs.

To examine further the relationship between these alterations and prognosis, we performed survival analyses of the entire cohort. We found that patients with ATRX-mutated or telomerase (+) tumors exhibited a significantly shorter MFS and OS than patients without such alterations (log-rank tests P < 0.001; Fig. 3C and D). Similar results were obtained after excluding 6 metastatic samples from the analysis, of which 4 harbored these alterations (Supplementary Fig. S6). No statistically significant differences in MFS and OS between telomerase (+) and ATRX-mutated PPGLs were identified.

Strikingly, although the presence of telomerase activation and ATRX mutations greatly improved the stratification of patients at high risk of metastatic progression (SDHx/FH-mutated, cluster C1A; Fig. 4A–C), these alterations were present only in 5 out of the 13 (38.4%) metastatic tumors from the other molecular groups (clusters C1B, C2A/B/C; Fig. 4B and C). Therefore, we concluded that assessment of telomerase activation mechanisms and screening of ATRX mutations could be used to identify the metastatic potential of PPGLs, especially in tumors at high risk of progression.

We here assigned a telomere maintenance mechanism to most tumor samples of the well-characterized cohort of 200 PPGLs collected by the COMETE network. We identified 12 telomerase (+) PPGLs, all of which having a metastatic status, and showed that both isoforms, full-length TERT and β-deletion, were overexpressed in these tumors. Although full-length is required for telomere maintenance, the catalytically dead β-deletion has also oncogenic functions (35, 37). Therefore, expression of both isoforms may promote cellular growth and progression to metastasis in PPGL tumors.

Mechanistically, hotspot mutations in the TERT promoter, C250T and C228T activate telomerase expression in human cancers (15) by creating a de novo binding motif for the transcription factor GABPA (38) and by driving an epigenetic switch (39, 40). We found 6 C228T hotspot promoter mutations exclusively in SDHx/FH-mutated metastatic tumors. This is consistent with previous studies, suggesting that although this mutation is not frequent in PPGLs, it can be found in SDHx-deficient tumors (24, 25, 41). Our study is the first to show that this recurrent mutation is associated with overt metastatic disease.

Hypermethylation has been linked to TERT overexpression in numerous tumor types (16, 42), sometimes associated with worse prognosis (43, 44). A recent report also suggested this hypermethylation to be restricted to metastatic paraganglioma (26). Here, we reinforce this observation and highlight its relevance from the standpoint of diagnosis because even though SDHx/FH-mutated tumors display a hypermethylator phenotype (36), this is the first time that a specific alteration in methylation is associated directly with the metastatic status. In addition, we observed that hypermethylation is not mutually exclusive with TERT promoter mutations and copy-number gains. In fact, it has been shown that genomic rearrangements involving the TERT locus also coexist with hypermethylation of the THOR region in high-risk neuroblastomas (45). These observations suggest that distinct genomic alterations cooperate in driving the transcriptional activation of TERT in PPGLs and that analysis of TERT promoter mutations and methylation could be useful in the clinical routine to identify metastatic PPGLs in high-risk tumors.

Regarding the ALT immortalization mechanism, discrepant prevalences of 4% PPGLs (27) and 27% PPGLs (28) were reported based only histological data. To clarify this issue, we used a combined analysis of APB assays and low expression of ATRX in tumors with long telomeres, which enabled us to estimate that 25 of 200 (12.5%) tumors, the great majority benign pheochromocytomas, activated the ALT mechanism.

We also find ATRX mutations in 8 of 200 PPGLs (4%), which contrasts with the original estimated prevalence of 12.6% (28). Accordingly, when exome-sequencing studies for PPGLs were revisited (Supplementary Table S5), the prevalence for ATRX mutations is 29 of 593 (4.8%). Of these, half have been associated with clinically aggressive behavior mostly in SDHx-mutated cases. Given that ATRX mutations are more frequent in this tumor subtype, this might explain why cohorts enriched for SDHx patients present a higher prevalence of ATRX mutations (28). We further found that only 6 of 25 (24%) ALT (+) PPGLs were linked to ATRX mutations and that these mutations had a stronger impact on poor prognosis than the ALT (+) status alone, particularly in PPGLs at high risk of metastatic progression (SDHx/FH-mutated, cluster C1A).

Our findings highlight a prominent role of immortalization-related mechanisms for the progression of neural-crest derived tumors, as noticed recently in neuroblastomas in which telomerase activation, ATRX mutations and MYCN amplifications define 3 nonoverlapping high-risk subgroups (46), or in gliomas in which mutations in the isocitrate dehydrogenase (IDH) and TERT/ATRX are also concurrent events that guide molecular classification and diagnosis (47).

Importantly, we could not assign any immortalization mechanism to 8 metastatic tumors, 6 of which from molecular groups C1B, C2A/B/C (Fig. 4C). This observation supports recent reports indicating that a subset of metastatic tumors, including melanomas and neuroblastomas could have progressed towards metastasis without activating a telomere maintenance mechanism (48, 49). On the contrary, our most striking result is that telomerase activation and ATRX mutations do impact the prognosis, particularly in the group at high risk of progression (cluster C1A). In fact, these alterations were present not only in metastatic samples and primary tumors from patients with synchronous metastases, but also in primary tumors long before the first metastasis appeared.

Interestingly, we found 1 patient with an SDHB mutation who developed 2 telomerase (+) metastases, even though his primary tumor operated on 7 years earlier was telomerase-negative (CIT_087). In addition, 1 patient presented with 2 primary tumors (CIT_073 and CIT_074), harboring an ATRX mutation and telomerase overexpression, respectively, 4 years before the first metastasis appeared. Thus, our results point these somatic alterations as key drivers of metastatic PPGLs. Nevertheless, we acknowledge caution with regard to the extent of heterogeneity within primary tumors of metastatic cases without evidence of telomerase activation or ATRX mutations. Given that extensive analyses of the whole tumor is impractical in the clinical routine, detection of these alterations in liquid biopsies would be of utmost importance to capture this tumor heterogeneity.

A key question in PPGL research is the identification of biomarkers able to distinguish between potentially metastatic and nonmetastatic tumors, which is crucial for diagnosis, treatment and follow-up. Although the present study is limited to the retrospective analysis of a modest sample of metastatic cases per molecular groups, our findings suggest that assessment of telomerase activation mechanisms (TERT promoter mutation/hypermethylation/copy-number gain or chromosomal rearrangement) and screening of ATRX mutations can identify potentially metastatic PPGLs, particularly in tumors carrying SDHx/FH mutations that are currently considered at high risk of progression. In addition, discrimination of immortalization mechanisms may become relevant to identify patients that would benefit from therapies targeting either telomerase or ALT (50). Prospective large multicenter studies will be required to address these issues, and to ascertain whether patients with SDHx-/FH–mutated PPGLs but without TERT activation or ATRX mutations would be henceforth considered as “low-risk” hereditary PPGLs for their surveillance.

No potential conflicts of interest were disclosed.

Conception and design: J. Favier, L.J. Castro-Vega, A.-P. Gimenez-Roqueplo

Development of methodology: I. Draskovic, L.J. Castro-Vega, A.-P. Gimenez-Roqueplo

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): I. Draskovic, A. Buffet, J. Cros, C. Lépine, T. Meatchi, M. Sibony, L. Amar, J. Bertherat, A. Londoño-Vallejo, J. Favier, L.J. Castro-Vega, A.-P. Gimenez-Roqueplo

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): S. Job, I. Draskovic, N. Burnichon, A. Buffet, J. Cros, A. Venisse, V. Verkarre, A. de Reyniès, J. Favier, L.J. Castro-Vega

Writing, review, and/or revision of the manuscript: S. Job, I. Draskovic, N. Burnichon, J. Cros, L. Amar, J. Bertherat, A. de Reyniès, A. Londoño-Vallejo, J. Favier, L.J. Castro-Vega, A.-P. Gimenez-Roqueplo

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): C. Lépine, A. Venisse, E. Robidel, J. Favier, A.-P. Gimenez-Roqueplo

Study supervision: J. Favier, L.J. Castro-Vega, A.-P. Gimenez-Roqueplo

A.-P. Gimenez-Roqueplo was supported by European Commission grants FP7 Research and Innovation funding program for 2007-2013 (n° 259735) and Horizon 2020 (n° 633983), and from Institut National du Cancer and Direction Générale de l'Offre de Soins (DGOS), Programme de Recherche Translationnelle en cancérologie (PRT-K 2014, COMETE-TACTIC, INCa_DGOS_8663). J. Favier received funding from Agence Nationale de la Recherche (ANR-2011-JCJC-00701 MODEOMAPP) and the Alliance nationale pour les sciences de la vie et de la santé (AVIESAN), Plan Cancer: Appel à projets Epigénétique et Cancer 2013 (EPIG201303 METABEPIC). N. Burnichon is funded by the Cancer Research for Personalized Medicine—CARPEM project (Site de Recherche Intégré sur le Cancer—SIRIC). The group is supported by the Ligue Nationale contre le Cancer (Equipe Labellisée). This work is part of the "Cartes d'Identité des Tumeurs (CIT) program" funded and developed by the Ligue Nationale contre le Cancer. Work in the Telomere and Cancer lab is supported by grants from La Ligue Nationale contre le Cancer (Equipe Labellisée) and Fondation ARC pour la recherche sur le cancer. We thank Profs Pierre-François Plouin and Xavier Bertagna for making this work possible through the COMETE Network, and to all members of the Genetics Department, Biological Resources Center and Tumor Bank Platform, Hôpital européen Georges Pompidou (BB-0033-00063), and of the Nikon Imaging Centre @ Institut Curie-CNRS (ANR-10-INSB-04) for technical support. We also thank Dr. Pedro Castelo-Branco and Dr. Joana Apolónio for providing technical assistance with pyrosequencing assay of the THOR region.

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.