Familial uveal melanoma and other tumors in 25 families with monoallelic germline MBD4 variants

Abstract

Background

Monoallelic germline MBD4 pathogenic variants were recently reported to cause a predisposition to uveal melanoma, associated with a specific tumor mutational signature and good response to immunotherapy. Monoallelic tumor pathogenic variants have also been described in brain tumors, breast cancers, and myxofibrosarcomas, whereas biallelic germline MBD4 pathogenic variants have been involved in a recessive hereditary adenomatous polyposis and a specific type of acute myeloid leukemia.

Methods

We analyzed MBD4 for all patients with a diagnosis of uveal melanoma at Institut Curie since July 2021 and in the 3240 consecutive female probands explored at the Institut Curie for suspicion of predisposition to breast cancer between July 2021 and February 2023.

Results

We describe 25 families whose probands carry a monoallelic germline pathogenic variant in MBD4. Eighteen of these families presented with uveal melanoma (including a case patient with multiple uveal melanoma), and 7 families presented with breast cancer. Family histories showed the first familial case of uveal melanoma in monoallelic MBD4 pathogenic variant carriers and other various types of cancers in relatives, especially breast, renal, and colorectal tumors.

Conclusions

Monoallelic MBD4 pathogenic variant may explain some cases of familial and multiple uveal melanoma as well as various cancer types, expanding the tumor spectrum of this predisposition. Further genetic testing in relatives combined with molecular tumor analyses will help define the tumor spectrum and estimate each tumor’s risk.

MBD4 was recently reported to be a predisposition gene for uveal melanoma (UM [Mendelian Inheritance in Man [MIM] No. 606660]) (1-3). MBD4 encodes a glycosylase of the DNA base excision repair system. Monoallelic germline pathogenic variants were found in several patients with uveal melanoma, associated with a specific tumor signature with a high level of CpG>TpG alterations (signature SBS96) caused by the failure to repair G:T mismatches resulting from deamination of 5ʹ-methylcytosine and leading to a high tumor mutation burden (1,2,4,5). Because of this high tumor mutation burden, MBD4 pathogenic variant carriers with advanced uveal melanoma respond better to immunotherapy than patients with uveal melanoma who harbor low tumor mutation burden (1,3,6). The associated tumors also show loss of the wild type MBD4 allele most commonly by monosomy 3, which is a frequent chromosomal copy number alteration in uveal melanoma associated with a high risk of metastasis, following the Knudson 2-hit model for tumor suppressor genes. A relative risk of 9 for uveal melanoma for germline MBD4 pathogenic variant carriers has been reported (2).

Biallelic germline carriers of MDB4 pathogenic variants may develop myelodysplastic syndrome, acute myeloid leukemia, or uveal melanoma (7). Biallelic germline MBD4 pathogenic variants also cause a recessive hereditary adenomatous colorectal polyposis, comparable to polyposis linked to biallelic pathogenic variants of MUTYH or NTHL1, 2 genes also involved in the base excision repair system (8). Digestive lesions and acute myeloid leukemia have been linked to MBD4 deficiency because they harbor a mutational profile similar to the specific SBS96 signature. This syndrome, named MBD4-associated neoplasia syndrome (MANS [MIM No. 619975]), is an extremely rare recessive condition (8).

The SBS96 mutational signature is often present in tumors that carry inherited or acquired MBD4 pathogenic variants (2). It was also found in a glioblastoma described in The Cancer Genome Atlas (TCGA) and, more recently, in breast cancer and myxofibrosarcoma associated with monoallelic germline MBD4 pathogenic variants (1,4). Thus, the tumor spectrum associated with germline pathogenic variants in MBD4 may be broader than initially described and remains mostly unknown.

Here, we describe the first familial case of uveal melanoma, the first case patient with multiple uveal melanoma associated with a monoallelic germline pathogenic variant in MBD4, and the largest series of families with MBD4 pathogenic variants reported so far with 25 monoallelic probands. Eighteen of these families have been identified in a context of uveal melanoma and 7 in a context of familial breast cancer. Cascade genetic testing has been offered to relatives to characterize the tumor spectrum associated with monoallelic germline pathogenic variants in MBD4.

Methods

Following our publications of the first case patient of a MBD4 pathogenic variant carrier in 2018 (1) and of another 11 unrelated patients in a uveal melanoma series in 2021 (2), MBD4 was added to the cancer predisposing gene panel of the Institut Curie since July 2021. For all patients with newly diagnosed uveal melanoma after this date, panel testing of BAP1 and MBD4 was offered in a diagnostic context irrespective of patient age and family history.

We also analyzed MBD4 in the 3240 consecutive female probands explored at the Institut Curie for suspicion of predisposition to breast cancer between July 2021 and February 2023. Then, we compared the frequency of MBD4 pathogenic variants found in this cohort with a control cohort from the gnomAD database, version 2.1.1 (non-Finnish European female individuals without cancer and with available whole-exome sequencing data). Only variants passing the quality filter were included in the analysis. The criteria for MBD4 variant selection were the same in breast cancer cases and controls: frameshift insertion and deletion, nonsense substitution, or splicing variant in the consensus splicing site. The comparison of MBD4 pathogenic variants frequency between breast cancer cases and controls was evaluated using Fisher exact tests.

All genetic analyses were performed as part of care, and all involved patients gave their written consent for genetic testing and for use of their data for research purposes and publication. Clinical data were extracted from medical records. The project was approved by the local institutional review board of Institut Curie.

Genomic DNA was extracted from 2-mL blood samples collected on EDTA using the QIAsymphony instrument (QIAGEN, Hilden, Germany), according to the manufacturer’s instructions. The cancer predisposition gene panel analysis was performed by SureSelectQXT enrichment (Agilent, Santa Clara, CA), with custom probes and sequencing on NextSeq 500 (Illumina, Inc, San Diego, CA), according to the manufacturers’ instructions. The bioinformatics pipeline included mapping with the Burrows-Wheeler Aligner, variant calling with VarScan2 and Pindel, and annotation with Annovar. Variant classification was based on the American College of Medical Genetics and Genomics and the Association for Molecular Pathology recommendations (9). Only families with class 5 (pathogenic) or 4 (likely pathogenic) variants were included in the study. Haplotype analyses at the MBD4 locus were performed with 3 microsatellite markers (rs61557048 and rs34121804, located at 15 kilobases and 100 kilobases after MBD4, and rs1553787467, located at 260 kilobases before MBD4) by fragment analysis on the ABI 3500XL Genetic Analyzer (ThermoFisher, Waltham, MA).

Results

Of 289 patients with uveal melanoma seen between July 2021 and April 2023, 6 new MBD4 probands were identified. We describe their personal and family history, along with the family history of the 12 probands previously reported (1,2).

Of 18 probands with uveal melanoma, 9 were male and 9 were female. The mean age at uveal melanoma diagnosis was 56 (range = 20-87) years, which is not statistically significantly different from the mean age at uveal melanoma diagnosis in the whole uveal melanoma series at the Institut Curie (n = 1524) or compared with the subpopulation of patients with uveal melanoma with chromosome 3 monosomy (n = 371) (2-sided Welch 2-sample t test: P = .27 and P = .22, respectively). None had bilateral uveal melanoma, but 1 male patient in family 17 had 2 independent uveal melanomas of the right eye at age 20 years and age 26 years, developed on preexisting naevi, associated with a third choroidal pigmented lesion of the right eye (Figure 1). Four other probands with uveal melanoma had another documented choroidal pigmented lesion, as well. Two patients presented with another cancer: 1 patient from family 3 had a papillary thyroid carcinoma, and 1 patient from family 1 presented with breast in situ carcinoma (2). One patient (family 9) reported colorectal polyps (<5).

Figure 1.

Retinophotographs of the proband of family 17, who developed 2 independent uveal melanomas of the right eye at 20 and 26 years of age on preexisting naevi. A) Wide-field retinophotograps of the 3 distinct pigmented choroidal lesions: the first localization of choroidal melanoma, treated in 2015 (widest circle, illustrated in B); the second localization of choroidal melanoma, treated in 2022 (medium circle, illustrated in C); and a flat choroidal pigmented lesion (smallest circle), stable from the first visit but in the field of irradiation of the first localization of choroidal melanoma. B) Photomontage of 2 retinophotographs of the first localization of choroidal melanoma, before treatment. A dome-shaped, pigmented choroidal mass (circle) is situated superotemporal of the macula. C) Upper: retinophotographs of the second localization of choroidal melanoma, in 2020 (left, circle, flat pigmented choroidal lesion before growth) and 2022 (right, dotted circle, documented growth of a dome-shaped choroidal mass corresponding to a choroidal melanoma). Lower: Optical coherence tomography of the choroidal mass. On the left, the flat hyperreflective area in the choroid (arrow) corresponds to the flat, pigmented choroidal lesion (circle). On the right, the thickness of the choroidal lesion increased (dotted arrow), corresponding to the choroidal melanoma (dotted circle).

Family history was collected for 15 of the 18 probands, and the information is summarized in Table 1. Genetic testing was offered to their relatives. One proband had a family history of uveal melanoma: her sister also had uveal melanoma at age 54 years, with normal germline analysis of BAP1. The presence of the MBD4 pathogenic variant was confirmed on a germline sample from her sister (family 15; Figure 2).

Figure 2.

Pedigrees of families 9, 10, 15, 17, 19, and 22.

Table 1.

MBD4 probands’ personal and family history^a

Family ID	Proband’s sex	Proband personal tumor history (age, y)	Pathological variant in MBD4		Family history of other tumors (age, y)
1	Female	In situ ductal carcinoma (74), uveal melanoma (76)	c.1425del	p.(Leu476Trpfs*9)	Leukemia, uterine cancer
2	Female	Uveal melanoma (38)	c.1688G>A	p.(Trp563*)	Breast cancer (70), lung cancer (60), colorectal cancer
3	Female	Papillary thyroid cancer (54), uveal melanoma (54)	c.1544-1G>T	p.?	Breast cancer, pancreatic cancer (83)
4	Male	Uveal melanoma (87)	c.1544-1G>T	p.?	Breast cancer, gastrointestinal cancer, cutaneous melanoma
5	Male	Uveal melanoma (51)	c.1425del	p.(Leu476Trpfs*9)	Head and neck cancer
6	Male	Uveal melanoma (72)	c.1366del	p.(Ala456Leufs*29)	Gastrointestinal cancer (2 occurrences)
7	Female	Uveal melanoma (57)	c.1544-1G>T	p.?	Lung or head and neck cancer (65), gastrointestinal cancer (87)
8	Male	Uveal melanoma (68)	c.541C>T	p.(Arg181*)	Breast cancer (40), head and neck cancer (38), lung cancer (74), metastatic cancer (64), metastatic cancer (65)
9	Male	Colorectal adenomas (40), uveal melanoma (54)	c.1273C>T	p.(Arg425*)	Colorectal cancer (39), metastatic cancer (92), liver cancer
10	Female	Uveal melanoma (35)	c.939dup	p.(Glu314Argfs*13)	Breast cancer (57), breast cancer (43), breast cancer (65), head and neck cancer (42), kidney cancer (58)
11	Female	Uveal melanoma (42)	c.541C>T	p.(Arg181*)	Breast cancer (43), brain tumor (89), uterine cancer (79)
12	Male	Uveal melanoma (55)	c.1688G>A	p.(Trp563*)	N/A
13	Female	Uveal melanoma (41)	c.1384C>T	p.(Arg462Trp)	N/A
14	Male	Uveal melanoma (76)	c.335 + 1G>A	p.?	N/A
15	Female	Uveal melanoma (55)	c.1002_1005del	p.(Lys335Phefs*18)	Uveal melanoma (54), leukemia (45)
16	Female	Uveal melanoma (67)	c.1189dup	p.(Thr397Asnfs*21)	Cutaneous cancer (47)
17	Male	Uveal melanoma (20), uveal melanoma (26)	c.1544-1G>T	p.?	Breast cancer (35), gastrointestinal cancer (83)
18	Male	Uveal melanoma (60)	c.939dup	p.(Glu314Argfs*13)	Breast cancer (50), gastrointestinal cancer (60)
19	Female	Breast cancer (27)	c.942_945del	p.(Glu314Aspfs*3)	Breast cancer (30), breast cancer (44), breast cancer (50), breast cancer (63), dermatofibrosarcoma (31), prostate cancer (65), cutaneous melanoma, prostate cancer (65), brain tumor (30)
20	Female	Breast cancer (50), cutaneous melanoma (52), ovarian cancer (59)	c.1670T>A	p.(Leu557*)	Breast cancer (55), breast cancer (59), bilateral breast cancer (56), liver cancer (58), kidney cancer (42), cutaneous melanoma (70)
21	Female	Breast cancer (36)	c.942_945del	p.(Glu314Aspfs*3)	Cutaneous cancer (100)
22	Female	Breast cancer (63)	c.939del	p.(Glu314Lysfs*4)	Breast cancer (76), breast cancer (42), myxoid liposarcoma (29), cutaneous melanoma (60), prostate cancer (76), lung cancer (91), multiple Lynch syndrome spectrum tumors
23	Female	Breast cancer (35), contralateral breast cancer (38)	c.939dup	p.(Glu314Argfs*13)	Breast cancer (35), breast cancer (65), breast cancer, colorectal cancer (70), lung cancer (90)
24	Female	Breast cancer (60), small-cell lung cancer (68)	c.1617C>G	p.(Tyr539*)	None
25	Female	Breast cancer (46)	c.1670T>A	p.(Leu557*)	None

a

N/A = not available.

Various other types of cancers were reported in probands’ relatives, including 10 breast cancers. Cascade genetic testing was performed in 6 uveal melanoma families (families 4, 7, 9, 10, 15, and 16). In total, 11 relatives (8 women, 3 men) were tested, and MBD4 pathogenic variants were found in 3 of them, including 2 women. All carrier relatives had a personal history of tumor: colorectal cancer at age 39 years with fewer than 5 adenomas (family 9), renal clear cell carcinoma at age 58 years (family 10), and uveal melanoma at age 54 years (family 15). Of the 8 relatives tested who did not carry the familial MBD4 pathogenic variant, 7 had no personal history of tumor, and 1 (family 10) had breast cancer at age 65 years (grade 1 hormone receptor–positive, HER2-negative invasive ductal carcinoma). Pedigrees of families 9, 10, 15, and 17 are presented in Figure 2. The other uveal melanoma pedigrees are available on request.

During this period, we also found a monoallelic germline pathogenic variant in MBD4 in a family that underwent whole-genome sequencing because of a severe, unexplained family history of early-onset breast cancer (family 19). The proband was a woman who developed breast cancer (intraductal carcinoma) at age 27 years. She had 2 sisters who had had breast cancer: 1 at age 44 years (grade 2 hormone receptor–positive, HER2-negative invasive ductal carcinoma), and the other at age 30 years (grade 3 triple-negative invasive ductal carcinoma). The latter sister also had dermatofibrosarcoma at age 31 years, with normal TP53 germline analysis. No pathogenic variant was identified in the breast cancer–predisposing genes (BRCA1, BRCA2, PALB2, TP53, CDH1, PTEN, RAD51C, RAD51D, MLH1, MSH2, MSH6, PMS2 [exons 6-8 and exon 10]). The only common pathogenic variant found in the 3 sisters by whole-genome sequencing was the pathogenic variant in MBD4, which was inherited from the father, who had had cutaneous melanoma and prostate cancer at age 65 years. Of note, the mother and a maternal aunt also had postmenopausal breast cancer, at ages 63 and 50 years, respectively. Therefore, the MBD4 pathogenic variant does not explain the maternal family history of breast cancer.

In light of this finding, of the high number of breast cancers reported in the families of patients with uveal melanoma and a pathogenic variant in MBD4 and of the recent publication of MBD4 pathogenic variants in breast cancers (4), we analyzed MBD4 in the 3240 consecutive female probands explored at the Institut Curie for suspicion of predisposition to breast cancer between July 2021 and February 2023.

Seven MBD4 carriers were identified, including 6 new probands and 1 already identified because she also had uveal melanoma (family 1). None had any other pathogenic variant in breast cancer–predisposing genes. To note, 1 patient (family 22) belonged to a known Lynch syndrome family resulting from a germline heterozygous PMS2 pathogenic variant [PMS2 pathogenic variant: c.861_864del, p.(Arg287Serfs*19)] (MIM No. 614337).

They all presented with breast cancer of different subtypes (invasive ductal carcinoma or invasive lobular carcinoma, hormone receptor positive or triple negative), with a median age of 50 years (range = 35-74). One had bilateral breast cancer at age 35 years and at age 38 years (family 23). One also had ovarian cancer at age 59 years (clear cell carcinoma) and cutaneous melanoma at age 52 years (family 20).

We compared the frequency of MBD4 pathogenic variants in the 3233 consecutive female probands seen at Institut Curie for suspicion of predisposition to breast cancer (7/3233 [0.216%]) with the frequency of MBD4 pathogenic variants in 22 131 female controls from the gnomAD database (11/22 120 [0.00049%]). We found a positive association between breast cancer and MBD4 pathogenic variants (P = .0048).

Family history was collected for all probands, and the information is summarized in Table 1. Cascade genetic testing was performed in family 22, the known Lynch syndrome family with a germline heterozygous PMS2 pathogenic variant (pedigree presented in Figure 2). Four relatives were tested, including 3 women, and the MBD4 pathogenic variant was found in all of them. They all presented with a personal history of tumor: 2 women had breast cancer at age 55 and 76 years, respectively; 1 had myxoid liposarcoma at age 29 years, with normal TP53 germline analysis; and the male patient had prostate cancer and cutaneous melanoma, similar to the case patient in family 19. It is interesting to note that the MBD4 pathogenic variant co-segregates with breast cancers and sarcoma, which are not tumors included in the Lynch syndrome spectrum. Two women with breast cancer and 2 other relatives carried both MBD4 and PMS2 pathogenic variants. One woman with breast cancer was an MBD4 carrier and did not carry the PMS2 pathogenic variant. No tumor material was available to look for the SBS96 signature in breast cancer or in sarcoma to support the role of the MBD4 pathogenic variant in the oncogenesis of these tumors. Pedigrees of families 19 and 22 are presented in Figure 2. The other pedigrees are available on request.

In these 25 families, all germline pathogenic variants in MBD4 were stop mutations, frameshifts, or splicing variants except 1 (family 13), which was a missense: c.1384C>T; p.(Arg462Trp). It was classified as pathogenic based on functional testing (absence of glycosylase activity of the recombinant protein carrying the variant) (2). Five pathogenic variants were identified in several unrelated families: c.541C>T, p. (Arg181*) in 2 families, c.1425del, p. (Leu476Trpfs*9) in 2 families, c.1544-1G>T, p.? in 4 families, c.1670T>A, p.(Leu557*) in 2 families, and c.1688G>A, p.(Trp563*) in 2 families. Moreover, 3 pathogenic variants found in 6 different families were located in the same homopolymeric stretch region [c.939dup, p.(Glu314Argfs*13); c.942_945del, p.(Glu314Aspfs*3); and c.939del, p.(Glu314Lysfs*4)]. Distribution of variants is presented in Figure 3, along with previously published pathogenic variants in monoallelic carriers (1,2).

Figure 3.

Reported germline monoallelic pathogenic variants in MBD4. HBOC = Breast cancer families.

A haplotype analysis based on 3 microsatellites close to MBD4 (rs61557048, rs34121804, and rs1553787467) was performed on all pathogenic variants found in more than 1 family. Indeed, a specific haplotype was associated with each MBD4 pathogenic variant, including c.541C>T, c.942_945del, c.1544-1G>T, c.1670T>A, and c.1688G>A, respectively, in all tested MBD4 pathogenic variant carriers. Only the 2 closest microsatellite markers were common in families with the pathogenic variants c.939dup and c.1425del. The haplotype analyses are presented in Supplementary Figure 1 (available online).

Discussion

We confirm that monoallelic germline MBD4 pathogenic variants are associated with an increased risk of developing uveal melanoma (2). The previously described relative risk  of 9 based on 8 cases (2) may be underestimated, given the high number of MBD4 carriers found out of 289 probands with uveal melanoma (twice as many as BAP1 carriers) and the description of a familial case and of a case patient with multiple uveal melanoma. Monoallelic germline MBD4 pathogenic variants could explain a more significant part in uveal melanoma than first expected. Identifying patients with pathogenic variants in MBD4 among patients with uveal melanoma is difficult, however, because there are no obvious selection criteria to propose a germline genetic analysis (age of onset and family history are clearly insufficient to identify the patients in this series). Tumor analysis could help in selecting patients, but some patients have conservative treatment and no available tumor samples. Given the therapeutic stakes, we believe that germline genetic analysis to look for MBD4 pathogenic variants must be carried out at least for all patients with metastatic uveal melanoma. Moreover, identifying these patients is essential for genetic counseling and could be important for their monitoring. Germline MBD4 pathogenic variant carriers should be screened repeatedly and for life for the presence of suspicious lesions in the uveal tract (choroid, ciliary body, iris), including naevi that may be at greater risk of malignant transformation in this context.

Suspecting an association between breast cancer and MBD4, we also identified for the first time monoallelic germline MBD4 pathogenic variants in families with breast cancer and sarcoma. No specific histology of breast cancer stood out. The co-segregation of the MBD4 pathogenic variants among the breast cancers supports the hypothesis that germline MBD4 pathogenic variants have a role in their oncogenesis. Moreover, looking into gnomAD, 1 of the largest available control cohorts (despite its limitations) (10), we concluded that there was an association between breast cancer and MBD4 pathogenic variants. More studies are required, however, to explore the relationship between germline MBD4 pathogenic variants and breast cancer. It is also interesting to note the association with sarcomas in light of the 2 recently published myxofibrosarcomas harboring signature SBS96 (11) and the young age at diagnosis for these 2 sarcomas (which was not a selection criterion for breast cancer families). Finding the SBS96 signature in new breast cancer or sarcoma would be an additional argument supporting the role of the MBD4 pathogenic variants in the oncogenesis of these tumors.

Other types of tumors were found in MBD4 carriers, including 3 cases of cutaneous melanoma and 1 case patient with clear cell renal carcinoma. No formal link can be established between the oncogenesis of these tumors and germline MBD4 pathogenic variants, but it should be noted that the only other predisposing gene for uveal melanoma, BAP1, is also linked to an increased risk of kidney cancer and cutaneous melanoma (12).

Moreover, none of the monoallelic MBD4 carriers we describe here had an adenomatous polyposis, but the proband in family 9 had few colorectal adenomas, and his sister, also a carrier of the MBD4 pathogenic variant, developed colorectal cancer at a young age and had multiple adenomas. As biallelic pathogenic variant in MBD4 predispose carriers to attenuated colorectal polyposis (8), it is tempting to speculate that monoallelic germline MBD4 pathogenic variants could be associated with a moderately increased risk of colorectal cancers, as it has been described in MUTYH heterozygous carriers (MIM No. 608456) (13). There was also 1 reported monoallelic MBD4 germline pathogenic variant carrier with a colonic oligopolyposis and early-stage rectosigmoid cancer, as described in biallelic germline MBD4 pathogenic variant cases (14), although a recent study could not demonstrate any association between monoallelic germline MBD4 pathogenic variants and gastrointestinal tumor predisposition (15).

Finally, several MBD4 pathogenic variants were identified in different families, raising the question of the existence of recurrent variants in this gene, especially in the homopolymeric stretch region around c.939 (5 families in this series, 6 families out of 8 in Palles et al.) (8). For other variants, the hypothesis could be a founder effect, with families from the same French region (eg, families 8 and 11) and shared haplotypes. It is interesting to note that some pathogenic variants are common to monoallelic and biallelic patients. No genotype-phenotype correlation is observed between breast cancer families and uveal melanoma families in monoallelic germline MBD4 pathogenic variant carriers.

To conclude, we describe the largest series of monoallelic MBD4 families with the first familial case of uveal melanoma and the first case patient with multiple uveal melanoma in this context, suggesting that monoallelic germline MBD4 pathogenic variants could explain a more significant part in uveal melanoma than first expected. Our report is strongly in favor of MBD4 being a multitumor-predisposing gene, including in the monoallelic state, with an overlapping but possibly distinct cancer spectrum from the biallelic state. Therefore, we suggest distinguishing monoallelic MBD4-associated neoplasia syndrome (MMANS) from biallelic MBD4-associated neoplasia syndrome (BMANS). Epidemiological-genetic approaches combined with molecular tumor data will help confirm associations and estimate risks of each tumor type associated with pathogenic variants in this gene, which is crucial for genetic counselling. MBD4 should be added to diagnostic panels, especially for patients with uveal melanoma.

Data availability

The data that support the findings of this study are available on reasonable request from the corresponding author. The data are not publicly available because of ethical restrictions.

Author Contributions

Marie-Charlotte Villy, MD (Data curation; Investigation; Visualization; Writing—Original draft), Manuel Rodrigues, MD, PhD (Data curation; Funding acquisition; Writing—review & editing), Marc-Henri Stern, MD, PhD (Conceptualization; Funding acquisition; Writing—review & editing), Nathalie Cassoux, MD, PhD (Data curation; Writing—review & editing), Dominique Stoppa-Lyonnet, MD, PhD (Conceptualization; Writing—review & editing), Mathias Cavaillé, MD, PhD (Formal analysis; Writing—review & editing), Amal Ait Omar, PhD (Data curation; Writing—review & editing), Alain Lortholary, MD (Data curation; Writing—review & editing), Virginie Bubien, MD (Data curation; Writing—review & editing), Denis Malaise, MD (Data curation; Writing—review & editing), Lisa Golmard, PharmD, PhD (Formal analysis; Writing—review & editing), Alexandre Matet, MD, PhD (Data curation; Writing—review & editing), Mathias Schwartz, MD (Formal analysis; Writing—review & editing), Catherine Dubois d'Enghien, BTEC (Formal analysis; Writing—review & editing), Anne Vincent-Salomon, MD, PhD (Conceptualization; Writing—review & editing), Sophie Vacher, MSc (Formal analysis; Writing—review & editing), Ivan Bièche, PharmD, PhD (Formal analysis; Writing—review & editing), Alexandre Houy, MSc (Formal analysis; Writing—review & editing), Julien Masliah-Planchon, PharmD, PhD (Investigation; Methodology; Writing—review & editing), Marine Le Mentec, MSc (Conceptualization; Data curation; Resources; Writing—Original draft; Writing—review & editing), Anaïs Le Ven, MSc (Data curation; Visualization; Writing—Original draft; Writing—review & editing), Sophie Piperno-Neumann, MD (Data curation; Writing—review & editing), Chrystelle Colas, MD, PhD (Conceptualization; Writing—Original draft; Writing—review & editing).

Funding

This work was supported by the Programme de Recherche Translationnelle en Cancérologie (grant No. PRT-K19-51), Institut National du Cancer—Direction générale de l’Offre de soins (INCa-DGOS).

Conflicts of interest

The authors declare no competing interests. They were not involved in the editorial review or decision to publish this manuscript.

Acknowledgements

The authors would like to thank the patients and their families for providing written consent for publication. The funder had no role in the design of the study; the collection, analysis, or interpretation of the data; or the writing of the manuscript and decision to submit it for publication.

Web resources: The Cancer Genome Atlas: https://portal.gdc.cancer.gov/. gnomAD: https://gnomad.broadinstitute.org/.