ARTICLE Role of POLE and POLD1 in familial cancer Pilar Mur, PhD1,2,3, Sandra García-Mulero, MS2,4, Jesús del Valle, PhD1,2,3, Lorena Magraner-Pardo, MS5, August Vidal, MD6, Marta Pineda, PhD1,2,3, Giacomo Cinnirella, MD, PhD1,2, Edgar Martín-Ramos, MS7, Tirso Pons, PhD8, Adriana López-Doriga, PhD2,4,9, Sami Belhadj, PhD1,2, Lidia Feliubadaló, PhD1,2,3, Pau M. Munoz-Torres, PhD1,2, Matilde Navarro, MD1,2,3, Elia Grau, BS1,2, Esther Darder, BS10, Gemma Llort, PhD11,12, Judit Sanz, PhD13, Teresa Ramón y Cajal, MD, PhD14, Judith Balmana, MD, PhD15, Joan Brunet, MD, PhD1,2,3,10, Victor Moreno, MD, PhD2,4,9,16, Josep M. Piulats, MD, PhD2,3,17, Xavier Matías-Guiu, MD, PhD2,3,6, Rebeca Sanz-Pamplona, PhD2,4,9, Rosa Aligué, PhD7, Gabriel Capellá, MD, PhD1,2,3, Conxi Lázaro, PhD1,2,3 and Laura Valle, PhD 1,2,3 Purpose: Germline pathogenic variants in the exonuclease domain (POLE p.Met294Arg) was classified as likely pathogenic, four as (ED) of polymerases POLE and POLD1 predispose to adenomatous likely benign, and seven as variants of unknown significance. The polyps, colorectal cancer (CRC), endometrial tumors, and other most commonly associated tumor types were colorectal, endome- malignancies, and exhibit increased mutation rate and highly trial and ovarian cancers. Loss-of-function and outside-ED variants specific associated mutational signatures. The tumor spectrum and are likely not pathogenic for this syndrome. prevalence of POLE and POLD1 variants in hereditary cancer are Conclusions: Polymerase proofreading–associated syndrome con- evaluated in this study. stitutes 0.1–0.4% of familial cancer cases, reaching 0.3–0.7% when Methods: POLE and POLD1 were sequenced in 2813 unrelated only CRC and polyposis are considered. ED variant interpretation probands referred for genetic counseling (2309 hereditary cancer is challenging and should include multiple pieces of evidence. patients subjected to a multigene panel, and 504 patients selected based on phenotypic characteristics). Cosegregation and Genetics in Medicine (2020) 22:2089–2100; https://doi.org/10.1038/s41436- case–control studies, yeast-based functional assays, and tumor 020-0922-2 mutational analyses were performed for variant interpretation. Results: Twelve ED missense variants, 6 loss-of-function, and 23 Keywords: polymerase proofreading–associated polyposis; outside-ED predicted-deleterious missense variants, all with PPAP; exonuclease domain; ultramutated phenotype; hereditary population allele frequencies <1%, were identified. One ED variant colorectal cancer INTRODUCTION breast, ovarian, stomach, pancreas, and skin tumors, among Germline missense pathogenic variants in the exonuclease others.2,6–10 domain (ED) of polymerases epsilon (POLE; a.a. 268–471) Somatic pathogenic variants in the ED of POLE have been and delta (POLD1; a.a. 304–533), which affect the proof- identified in 2–8% of colorectal cancer (CRC),11–14 7–15% of reading capabilities of these polymerases, predispose to endometrial tumors,15–18 and more rarely in other tumor multiple colorectal adenomas and carcinomas, causing the types.19–21 Somatic pathogenic variants affecting POLD1 so-called polymerase proofreading–associated polyposis ED are extremely rare. Tumors with somatic POLE ED (PPAP) (MIM 615083; 612591).1–7 Evidence of extracolonic pathogenic variants and tumors developed in the context of tumors has been reported, including endometrial, brain, PPAP exhibit a dramatically increased mutation rate known as 1 Hereditary Cancer Program, Catalan Institute of Oncology, IDIBELL, Hospitalet de Llobregat, Barcelona, Spain; 2Program in Molecular Mechanisms and Experimental Therapy in Oncology (Oncobell), IDIBELL, Hospitalet de Llobregat, Barcelona, Spain; 3Centro de Investigación Biomédica en Red de Cáncer (CIBERONC), Madrid, Spain; 4Oncology Data Analytics Program (ODAP), Catalan Institute of Oncology, IDIBELL, Hospitalet de Llobregat, Barcelona, Spain; 5Prostate Cancer Clinical Research Unit. Spanish National Cancer Research Center (CNIO), Madrid, Spain; 6Department of Pathology, Bellvitge University Hospital, IDIBELL, Hospitalet de Llobregat, Barcelona, Spain; 7Department of Biomedical Sciences, School of Medicine, University of Barcelona, IDIBAPS, Barcelona, Spain; 8Department of Immunology and Oncology, National Center for Biotechnology (CNB-CSIC), Spanish National Research Council, Madrid, Spain; 9Centro de Investigación Biomédica en Red de Epidemiologia y Salud Pública (CIBERESP), Madrid, Spain; 10Hereditary Cancer Program, Catalan Institute of Oncology, IDIBGi, Girona, Spain; 11Department of Medical Oncology, Parc Taulí, Hospital Universitari Parc Taulí, Sabadell, Barcelona, Spain; 12 Consorci Sanitari de Terrassa, Terrasa, Barcelona, Spain; 13Genetic Counseling Unit, Medical Oncology Department, Althaia Xarxa Assistencial Universitària de Manresa, Manresa, Spain; 14Department of Medical Oncology, Hospital de la Santa Creu i Sant Pau, Barcelona, Spain; 15Department of Medical Oncology, Vall d’Hebron University Hospital, Vall d’Hebron Institute of Oncology (VHIO), Barcelona, Spain; 16Department of Clinical Sciences, Faculty of Medicine, University of Barcelona, Barcelona, Spain; 17Department of Medical Oncology, Bellvitge University Hospital, IDIBELL, Hospitalet de Llobregat, Barcelona, Spain. Correspondence: Laura Valle (lvalle@idibell.cat) Senior co-authorship: Conxi Lázaro, Laura Valle Submitted 21 April 2020; revised 20 July 2020; accepted: 21 July 2020 Published online: 14 August 2020 GENETICS in MEDICINE | Volume 22 | Number 12 | December 2020 2089 ARTICLE MUR et al ultramutated tumor phenotype,17,21–23 characterized by a history of breast cancer (n = 884), ovarian cancer (n = 317), specific mutation signature typified by C:G→A:T transversions and breast and ovarian cancer (n = 267); 247 familial (TCT context) and C:G→T:A transitions (TCG context).12,24 (attenuated and classic) adenomatous polyposis cases; 354 This corresponds to signature 10 in the COSMIC Signatures of Amsterdam- or Bethesda-positive hereditary nonpolyposis Mutational Processes in Human Cancer, or to signatures 14 CRC cases; 15 Li–Fraumeni–suspected cases; and 225 patients and 20 if the POLE or POLD1 pathogenic variant coexists with with suspicion of other minor cancer syndromes. mismatch repair (MMR) deficiency, respectively.12,24,25 Cancer Details of the cohorts are included in Fig. 1 and patients with somatic POLE ED pathogenic variants show Supplementary Table S1. excellent prognosis and good response to immune checkpoint inhibition, probably due to the immune response elicited by Mutational screening these tumors as a result of the large number of neopeptides In the retrospective cohort, direct automated (Sanger) generated as consequence of hypermutation.26 sequencing was used to sequence exons 8–13 (+/−20 b) of To assess the prevalence of POLE and POLD1 pathogenic POLD1 and exons 9–14 (+/−20 b) of POLE, which contain variants in hereditary cancer and refine the tumor spectrum the sequences coding for the ED of each polymerase. Primer of the associated clinical syndrome, we studied a prospective sequences were previously described.7 The hereditary cancer cohort of 2309 unrelated hereditary cancer patients multigene panel applied to the prospective cohort includes subjected to a multigene hereditary cancer panel, and a the complete coding sequence of POLE and POLD1.27,28 retrospective cohort of 504 unrelated cancer patients— Nonsynonymous variants located within coding exons or hereditary CRC and polyposis patients excluded—selected variants affecting canonical splice sites with a population 1234567890():,; based on previous reports of extracolonic manifestations in minor allele frequency (MAF) <1% were considered. PPAP, which include breast and ovarian cancer, endome- trial, brain, or skin cancer, among other tumors, alone or in In silico predictions combination with other tumor types or colonic polyps. The pathogenicity of the identified missense variants was Cosegregation analyses, yeast-based assays, and tumor analyzed by using the metapredictor REVEL, which combines mutational analyses were performed to facilitate variant pathogenicity predictions and conservation information interpretation. obtained from 18 individual scores, and provides optimal specificity and sensitivity results.29,30 REVEL score ≥0.35 was MATERIALS AND METHODS used for pathogenicity in the current study, a lower cutoff Ethics statement than the one recommended for clinical purposes (≥0.5) and Patients signed consent forms and were assessed at genetic defined based on the scores of known ED pathogenic variants counseling units at Catalan Institute of Oncology, Vall (Supplementary Table S2). d’Hebron Institute of Oncology, and Santa Creu i Sant Pau, Parc Taulí and Manresa Hospitals (Catalonia, Spain). The Functional assessments for variant interpretation and study received the approval of IDIBELL Ethics Committee. classification The exonuclease repair ability of POLE in the presence of ED Patients missense variants was tested in a Schizosaccharomyces pombe Retrospective cohort system. Exome sequencing was performed on DNA extracted Five hundred and four unrelated cancer patients were from the tumors of carriers of POLE/D1 variants to assess the analyzed, including 192 high risk breast and/or ovarian presence of hyper or ultramutation (10 or 100 Mut/Mb families, 178 patients with personal or familial history of respectively) and/or the mutational signature(s) associated different tumor types previously associated with PPAP with the presence of an ED pathogenic variant.24,31 POLE and (combinations include CRC and associated tumors, breast/ POLD1 variant frequencies in CRC and controls were ovarian cancer, skin cancer or brain), 30 patients with obtained from a Spanish population-based case–control study aggregation of other multiple tumors, either in the patient or (MCC-Spain, www.mccspain.org). ED variants were classified the family, and finally, 104 patients fulfilling the criteria for by applying the American College of Medical Genetics and TP53 genetic testing. None of the patients carried germline Genomics/Association for Molecular Pathology (ACMG/ pathogenic variants in the known high-penetrance cancer AMP) guidelines32 (Supplementary Table S3). Additional genes associated with the patient and/or family’s phenotype details are in Supplementary Methods. (Supplementary Table S1). RESULTS Prospective cohort Of the 2309 unrelated probands included in the prospective Two thousand three hundred and nine unrelated familial/ cohort, 374 (16.2%) carried (likely) pathogenic germline early-onset cancer patients, prospectively recruited from 2015 variants in known cancer-predisposing genes (Fig. 1; Supple- to 2018 in a hereditary cancer clinic–based context, were mentary Table S4). subjected to a multigene hereditary cancer panel.27,28 The A total of 12 novel or rare (MAFgnomAD<1%) missense cohort comprised patients with personal and/or familial variants located within the EDs of POLE (n = 7) or POLD1 2090 Volume 22 | Number 12 | December 2020 | GENETICS in MEDICINE MUR et al ARTICLE STUDY COHORTS PROSPECTIVE COHORT RETROSPECTIVE COHORT (N=2,309) (N=504) REFERRAL Breast and/or CRC and/or Suspected Li-Fraumeni SELECTED PHENOTYPES ovarian cancer polyposis and other phenotypes PHENOTYPES Non-BRCA1/2 HBOC Multiple primary (n=1,468) (n=601) (n=240) (n=192) malignancies BRCA1 MSH6 FLCN (n=30) BRCA2 FH MUTYH CDKN2A ATM MSH2 BRCA2 PALB2 TSC2 CRC, EC or GC Brain cancer & other BRIP1 MLH1 MULTI-GENE STK11 & BC or OC malignancies CHEK2 APC ATM PANEL MSH6 PMS2 PTEN (n=122) (n=22) GERMLINE MUTYH TP53 POLE TESTING RAD51C BAP1 ATM MLH1 RB1 MEN1 SDHD BC, EC BrC or Melanoma Non-TP53 suspected CDH1 RAD51D SMAD4 TSC1 & >5 polyps Li-Fraumeni SDHB PTEN SMAD4 (n=349) (n=104) TP53 CDH1 STK11 MUTYH CDKN2A RAD51D MSH6 FLCN BRIP1 APC CHEK2 PALB2 MSH2 BRCA2 BRCA1 0 50 100 0 20 40 60 0 2 4 6 8 % CARRIERS OF HBOC: 21% (55 / 267) HNPCC: 27% (95 / 354) Suspected LF: 13% (2 / 15) Selection performed excluding cases with germline (L)P variants (L)P VARIANTS IN HBC: 4% (106 / 884) Polyposis: 13% (31 / 247) Others: 18% (41 / 225) in genes associated with clinical phenotype KNOWN GENES HOC: 14% (44 / 317) POLE / POLD1 POLE p.K425R (HBOC) POLE p.G380C (Polyposis) POLE p.I307V (Melanoma) POLE p.G380C (Multiple tumors) ED VUS POLD1 p.R525W (HBC) POLE p.A426V (HNPCC) POLD1 p.I307M (HOC) POLD1 p.I325V (HBC) POLE / POLD1 POLE p.M294R: (2 HNPCC) ED (L)P VARIANTS PPAP 0 - 0.27% 0.33 - 0.67% 0 - 0.42% PREVALENCE (0 - 4 / 1,468) (2 - 4 / 601) (0 - 1 / 240) POLE p.M294R carriers TUMOR • CRC 67 SPECTRUM • CRC 37 FOR (L)P • CRC 57, 59, 69 VARIANTS • CRC 42, EC 55 • OC 36, EC 36 Fig. 1 Schematic representation of the characteristics of the cohorts analyzed in this study, prevalence of (likely) pathogenic variants in cancer-predisposing genes, and results of the current study for exonuclease domain (ED) missense variants in POLE and POLD1. BC breast cancer, BrC brain tumor, CRC colorectal cancer, EC endometrial cancer, GC gastric cancer, HBC hereditary breast cancer, HBOC hereditary breast and ovarian cancer, HOC hereditary ovarian cancer, HNPCC hereditary nonpolyposis colorectal cancer, LF Li–Fraumeni syndrome, LP likely pathogenic, OC ovarian cancer, PPAP polymerase proofreading–associated polyposis, VUS variant of unknown significance. (n = 5) were identified in 15 families (Table 1). Five loss-of- and other tumors, including breast cancer, squamous cell function (LoF) variants were found in five families, and 23 carcinoma, and non-Hodgkin lymphoma,10,33 and in a patient predicted-deleterious POLE (n = 19) and POLD1 (n = 4) with a Lynch syndrome–associated MLH1-deficient colorectal variants outside the ED were identified in 29 probands tumor (Supplementary Table S5).34 Of the 13 studied cancer- (Supplementary Fig. S1). affected individuals (in this and previous reports) only 7 Mutational screening of the region coding the ED of POLE carried POLE c.861T>A. No association with cancer was and POLD1 in the retrospective cohort (n = 504 unrelated detected for this variant in a Spanish case–control study, probands) identified a missense POLE ED variant, c.1138G>T either for CRC or breast cancer (source: MCC-Spain; (p.Gly380Cys), and a frameshift variant, POLE c.1185_1188del Supplementary Table S6). (p.Glu396Thrfs*15). The novel, predicted pathogenic POLE c.881T>G (p.Met294Arg) variant, which affects a highly conserved Phenotypic characteristics of carriers of ED variants amino acid, was identified in two families. One proband was a Twelve rare or novel missense variants located within the EDs woman diagnosed with ovarian and endometrial tumors at of POLE (n = 7) and POLD1 (n = 5) were identified in 16 age 36. Her mother, diagnosed with CRC and endometrial unrelated individuals of the 2813 cancer patients studied cancer at ages 42 and 55 respectively, and her CRC-affected (Table 1; pedigrees in Supplementary Figs. S2 and S3). uncle, were also carriers. POLE c.881T>G was also identified POLE c.861T>A (p.Asp287Glu) (MAFgnomAD_NFE = 0.17%) in a patient diagnosed with an MMR-proficient CRC at age was identified in two unrelated breast cancer patients, one of 37, and in his maternal uncle, affected with three metachro- whom was also diagnosed with an MMR-deficient CRC nous CRCs. The variant has been recently reported in a (Table 1). This variant had been previously reported in three patient affected with multiple colonic adenomas, breast unrelated families diagnosed with melanoma at ages 22–73 cancer, and endometrial cancer, diagnosed at 48, 50, and 55 GENETICS in MEDICINE | Volume 22 | Number 12 | December 2020 2091 Table 1 POLE and POLD1 exonuclease domain germline variants identified in the study (population MAF <1%). 2092 Genetic variant dbSNP rs # Evolutionary Proband’s phenotype Mutator Tumor sequencing analysis ACMG/AMP variant (motif/DNA binding population conservation and cosegregation phenotype (mutational burden and signature classificationf42 cleft) (Suppl. Fig. S1 MAFNFE (%)b (PhyloP/ data (Suppl. Figs. S2 in yeast contribution) (Fig. 3 and Suppl. (InterVar/gInterVar-adjusted) and Fig. 2)a PhastCons)c and S3) (Fig. 2)e Tables S6 and S7) ARTICLE In silico prediction (REVEL)d POLE c.861T>A; rs139075637 Moderately Fam A: BC 40 Yes (+++) Fam A–II.3 (BC): 185.7 Mut/Mb; Likely benign (PM1, PM2, BS2 p.Asp287Glu (Flanking (0.1710)h conserved Noncarrier: Mother 16.3% MMR-d sig and BP4)/Likely benign (PM1, Exo I/Yes) (1.189/1) (OC 53) BP4, PS3_moderate, BS1, Benign (0.286) Fam B: BC 59; CRC 59 BS4_supporting) Cancer-affected carriers vs. total cancer-affected individuals:i 7/13 (2/6 in one same family) c.881T>G; n.a. Conserved Fam C: OC 36; EC 36 Yes (++) Fam C–II.2 (OC): 249 Mut/Mb; Uncertain significance (PM1, p.Met294Arg (Flanking (9.279/1) Carriers: mother 15.3% MMR-d + POLE sig (SBS14) PM2)/Pathogenic Exo I/Yes) Damaging (CRC 42; Fam C–II.2 (EC): 300 Mut/Mb; 95.4% (PM1, PM2, PP1, PP3, (0.816) EC 55), uncle (CRC 67) MMR-d sigs PS3_supporting, PP4_strong) Fam D: CRC 37 Fam C–I.4 (CRC): 11 Mut/Mb; 23.1% Carriers: uncle (CRC 57, POLE sig (SBS10) 59, 69) Total meiosis: 3 across 1 family c.919A>G; p.Ile307Val n.a. Conserved Fam E: Melanoma Yes Fam E–II.1 (Melanoma): 32 Mut/Mb; Uncertain significance (outside Exo/No) (7.979/1) (×4) 81 21.6% MMR-d sigs (PM1, PM2)/(PM2) Benign (0.337) Carriers: sister (BC 83), sister (melanoma 55) c.1007A>G; rs5744760 Conserved Fam F: CRC 38 No effect40 n.a. Benign (PM1, BS1, BS2 and BP6)/ p.Asn336Ser (outside (0.0197)j (7.979/1) Fam G: Melanoma 48 Likely benign (BS1, Exo/No) Damaging BS3_supporting, PP3) (0.425) c.1138G>T; rs199746481 Conserved Fam H: BC (×2) 61 Yes (++) Fam H–II.4 (BC): 0.7 Mut/Mb Uncertain significance p.Gly380Cys (outside (0.0051) (7.867/0.878) Fam I: 10–20 Polyps 72 (PM1, PM2, PP3)/(PP3, Exo/No) Damaging PS3_supporting) (0.531) c.1274A>G; rs757186755 Conserved Fam J: BC 46; OC n.p. n.a. Uncertain significance (PM1 and p.Lys425Arg (0.0025) (7.979/1) 48, EC 48 PM2)/(PM1, PP3) (Exo IV/Yes) Damaging Total reported (0.457) carriers:k 8 c.1277C>T; rs374920539 Conserved Fam K: Renal cancer 49; Yes (++) Fam K–II.1 (CRC): 41 Mut/Mb; Uncertain significance (PM1 and p.Ala426Val (0.0034) (9.953/1) CRC 44; adenomas. 16.3% MMR-d sigs PM2)/(PM1, PP3, PS3_supporting, (Exo IV/Yes) Damaging Carriers: paternal uncle PP1) Volume 22 | Number 12 | December 2020 (0.392) (CRC 73), father (CRC 60) Total meiosis: 3 across 1 family | GENETICS in MEDICINE MUR et al Table 1 continued MUR et al Genetic variant dbSNP rs # Evolutionary Proband’s phenotype Mutator Tumor sequencing analysis ACMG/AMP variant (motif/DNA binding population conservation and cosegregation phenotype (mutational burden and signature classificationf42 cleft) (Suppl. Fig. S1 MAFNFE (%)b (PhyloP/ data (Suppl. Figs. S2 in yeast contribution) (Fig. 3 and Suppl. (InterVar/gInterVar-adjusted) and Fig. 2)a PhastCons)c and S3) (Fig. 2)e Tables S6 and S7) In silico prediction (REVEL)d POLD1 c.921T>G; p.Ile307Met n.a. Nonconserved Fam L: OC 40 Not n.a. Uncertain significance (outside Exo/No) (−2.516/0.00) conserved in (PM1, PM2 and BP4)/(PM2 Benign (0.268) S. pombe and BP4) c.973A>G; p.Ile325Val rs558345043 Moderately Fam M: BC 33 Not n.a. Uncertain significance (Exo I/Yes) (0.0017) conserved conserved in (PM1, PM2, and BS2)/(PM1, BP4) (2.440/0.984) S. pombe Benign (0.041) c.1054C>T; rs762330164 Conserved Fam N: 5 n.p. TCGA (CRC): 35 Mut/Mb, 71% MMR-d Uncertain significance p.Arg352Cys (outside (0.0032) (4.348/1) adenomatous polyps sigs. COSMIC (Lymph): 24.7 Mut/Mb; (PM1 and PM2)/Likely benign Exo/No) Benign (0.299) non-POLE/D1, non-MMR-d sigs (BP4, BP5) c.1562G>A; rs143076166 Conserved Fam O: CRC (×2) 34 n.p. Fam-6 (CRC)7,43: 39 Mut/Mb; 14.5% Uncertain significance GENETICS in MEDICINE | Volume 22 | Number 12 | December 2020 p.Arg521Gln (outside (0.0238)l (7.044/1) MMR-d sigs (14.51%) (PM1 and PM2)/Likely benign Exo/Yes) Benign (0.278) (PM1, BS2_supporting and BP4) c.1573C>T; rs201804732 Moderately Fam P: BC 35, 49, 53; Not TCGA (GC): 13.7 Mut/Mb; non-POLE/ Uncertain significance (PM1,PM2 p.Arg525Trp (outside (0.0010) conserved CRC 69 conserved in D1, non-MMR-d sigs (other DNA repair and BS2)/(PM1, PM2, BP4) Exo/Yes) (3.030/1) S. pombe defects) Benign (0.267) ACMG/AMP American College of Medical Genetics and Genomics/Association for Molecular Pathology, CRC colorectal cancer, EC endometrial cancer, GC gastric cancer, gnomAD Genome Aggregation Database, Lymph lymphoid neoplasm, MAF minor allele frequency, MMR-d mismatch repair deficiency, n.a. not available; NFE non-Finnish Europeans, n.p. not performed, OC ovarian cancer, SBS single base signature, sig(s) signature(s). a RefSeq GRCh37: POLE (NM_006231.2; NP_006222.2) and POLD1 (NM_001256849.1; NP_001243778.1). b Population MAF obtained from gnomAD non-Finnish, noncancer Europeans (gnomAD v.2.1.1). c PhyloP/PhastCons values were obtained from alignments of 100 vertebrate sequences. The higher the score, the more conserved the site is (PhyloP range score: −20 to +10; PhastCons: 0 to 1). d In silico prediction according to REVEL ≥0.35. e Yeast variant rate increase (+, p = 0.01; ++, p = 0.01–0.001; +++, p < 0.001 indicate the differences with the wildtype control, calculated using the Mann–Whitney nonparametric test). f According to the standard guidelines for the clinical interpretation of genetic variants (ACMG/AMP recommendations) using the automatic InterVar software. g ACMG/AMP–based manual curation (new evidence considered is underlined). Details are shown in Supplementary Methods and Supplementary Table S3: PM2, absent or <1/100,000 in general population (gnomAD non- cancer data set); PP1, cosegregation with disease in multiple affected family members (at least three meiosis across one family); PP3, in silico prediction indicates that the variant could be damaging (REVEL ≥ 0.35); PP4, the patient’s phenotype is highly specific for a disease with a single genetic etiology (moderate: when hyper/ultramutation and POLE/D1-associated mutational signatures 14 or 20 were detected in the tumor; strong: when hyper/ultramutation and the POLE/D1-associated mutational signature 10 were detected in the tumor); PS3, functional studies are supportive of a damaging effect on the gene (supporting: when a mutator phenotype in S. pombe was observed with p = 0.01–0.001; moderate: when a mutator phenotype in S. pombe was observed with p < 0.001). BS1, allele frequency is greater than expected for disorder (considering MAF > 0.1% in any gnomAD v2.1.1 noncancer subpopulation); BS2_supporting, observed in ≥10 healthy adult individuals (above 60 years of age) for a dominant (heterozygous) disorder (not applied when BS1 was considered); BS3, func- tional studies showed no damaging effect on protein function (supporting: when a mutator phenotype in S. pombe was not observed); BS4_supporting, lack of segregation in more than three cancer-affected members (same or related phenotype) of a family; BP4, in silico prediction of pathogenicity estimated by REVEL suggests no impact (REVEL < 0.30). BP5, variant found in a case with an alternate molecular basis for disease (for tumors with somatic POLE/D1 variants: when an MMR-proficient tumor was not hyper/ultramutated or when an DNA repair–proficient tumor was hypermutated but mutational signatures 10, 14, or 20 were not identified. For tumors from carrier families: same criteria as defined in the previous sentence for tumors with somatic variants but applied to two different tumors to avoid a false classification due to the presence of a phenocopy). h MAF in European non-Finnish, noncancer population (gnomAD): 202/118,154 individuals (60 above 60 years of age). i Characteristics of POLE c.861T>A (p.Asp287Glu) carrier families are shown in Supplementary Table S4. j MAF in African, noncancer population (gnomAD): 623/23,530 individuals (2.6%); 10 homozygous carriers. k POLE c.1274A>G (p.Lys425Arg) carriers from our study and previously reported.3,10,35,36 l MAF in European non-Finnish, noncancer population (gnomAD): 28/117,648 individuals (10 above 60 years of age). Families carrying the corresponding POLE/POLD1 variant, tumors studied for the analysis of mutational burden and signatures, and ACMG/AMP variant classification results, are highlighted in bold. Cosegregation data (carriers, non-carriers, meiosis) are shown in italics. 2093 ARTICLE ARTICLE MUR et al years old respectively, and in her sister, with seven colonic p.Arg525Arg and ATM p.Ser2407* developed a very aggres- adenomas at age 58.35 sive phenotype, including three metachronous early-onset POLE c.919A>G (p.Ile307Val), affecting a conserved breast tumors (ages 35–53) and a CRC at age 69, supporting a residue and not reported in public databases, was identified cumulative effect of the two variants. in an individual with four melanomas diagnosed at age 81 and in two of his sisters, diagnosed with melanoma at 55 and Functional assessment of ED variants: exonuclease repair breast cancer at 83. yeast-based assay and tumor analysis POLE c.1007A>G (p.Asn336Ser) (MAFgnomAD_NFE = The exonuclease repair ability of POLE in the presence of 0.02%) was identified in two families. One of the probands, missense ED variants was tested in a yeast system. The of Moroccan origin, was diagnosed with rectal cancer at age number of revertant colonies was significantly higher for p. 38, and the other one with melanoma at 48. This variant is Leu424Val (positive control), p.Asp287Glu, p.Met294Arg, p. relatively frequent in African populations (MAF = 2.6%), Gly380Cys, and p.Ala426Val, compared with the wildtype where a total of 10 homozygous carriers have been identified (fold change increase of 7–13) (Fig. 2). No significant among 11,975 Africans (gnomAD v.2.1.1). differences were observed for POLE p.Ile307Val, and a POLE c.1138G>T (p.Gly380Cys) (MAFgnomAD_NFE = previous study had shown no effect for p.Asn336Ser.40 0.005%), predicted deleterious, was identified in a woman Notably, POLE p.Asp287Glu, p.Met294Arg, p.Leu424Val, diagnosed with two synchronic breast tumors, and in an and p.Ala426Val, the ones with the highest number of unrelated individual with multiple (10–20) adenomatous revertant colonies, were located within the DNA binding polyps and no familial cancer history. pocket structure (Fig. 3). All POLD1 ED variants identified in POLE c.1274A>G (p.Lys425Arg) (MAFgnomAD_NFE = the study were predicted benign, and only two, p.Arg352Cys 0.0025%) was identified in a woman affected with breast, and p.Arg521Gln, affected conserved amino acids in yeast. ovarian, and endometrial tumors before age 50. This Since all gathered evidence for those two was enough to predicted pathogenic variant affects a highly conserved classify them as likely benign (Table 1), we did not perform residue located within the Exo IV motif active site. The the yeast-based functional assay. variant had been previously reported in six families, including Exome sequencing was performed in tumors developed by a melanoma patient,10 a patient with early-onset CRC,3 an POLE p.Asp287Glu, p.Met294Arg, p.Ile307Val, p.Gly380Cys, individual with colonic polyps and family history of CRC,36 a p.Ala426Val, and POLD1 p.Arg521Gln carriers, including, as 63-year-old CRC patient,35 a woman diagnosed with breast positive controls, tumors developed by POLE p.Leu424Val, cancer at age 31,35 and two first-degree relatives, who also POLD1 p.Asp316Gly, and POLD1 p.Asp316His carriers2,7 carried the MSH2 c.942+3A>T pathogenic variant, diagnosed (Fig. 4; Supplementary Table S7). All positive controls showed with three synchronic CRCs at age 30 and one CRC at age 33 hyper- or ultramutation (range: 14–478 Mut/Mb). POLE/D1 respectively.35 ED-associated signatures were detectable in the POLD1 POLE c.1277C>T (p.Ala426Val) (MAFgnomAD_NFE = 0.0034%), p.Asp316His carrier’s CRC tumor (signature 10 contribution: affecting a highly conserved residue within the Exo IV motif 20.66%), but barely identifiable in the POLD1 p.Asp316Gly active site, was identified in a male diagnosed with CRC and carrier’s liver metastasis (signature 10: 0.75%; signature multiple adenomas at age 44 and with renal cancer at 49, and in 20 [POLD1 + MMR deficiency]: 2.65%) or in the POLE his CRC-affected paternal uncle. The proband’s father, an p.Leu424Val carrier’s oligodendroglioma (signature 14 obligate carrier, was diagnosed with CRC at age 60. All three [POLE + MMR deficiency]: 4%), where MMR deficiency relatives also carried a likely pathogenic variant in CHEK2, hoarded most of the mutational signatures’ contribution. c.593–1G>T (rs786203229), a gene that confers a moderate risk Three primary tumors—one MMR-deficient ovarian can- of breast cancer, and possibly other tumor types, including cer, one MMR-deficient endometrial cancer and one MMR- CRC.37,38 In this family, the carriers of the two variants did not proficient CRC—from two carriers of POLE p.Met294Arg, develop particularly aggressive phenotypes, two of them having were analyzed. All three displayed hyper/ultramutation developed CRC late in life (ages 60 and 73). (10.96–300.85 Mut/Mb). Signature 10 was identified in the The five ED variants identified in POLD1, p.Ile307Met, CRC (contribution: 23%), and signature 14, linked to the co- p.Ile325Val, p.Arg352Cys, p.Arg521Gln, and p.Arg525Trp, occurrence of MMR deficiency and POLE ED pathogenic were predicted to be neutral (Table 1; pedigrees in variant in the ovarian tumor (contribution: 15.32%). MMR- Supplementary Fig. S3). POLD1 c.1562G>A (p.Arg521Gln) deficient associated signatures were the major contributors in had been previously identified by our group in a 48-year-old the ovarian and endometrial tumors (Fig. 4; Supplementary CRC patient.7 POLD1 c.1573C>T (p.Arg525Trp) co-occurred Table S7). with ATM c.7220C>A (p.Ser2407*; rs1555122149) in a While the breast tumor from the POLE p.Gly380Cys carrier woman affected with multiple tumors. ATM pathogenic showed no hypermutation (0.67 Mut/Mb), a breast tumor variants confer a moderate risk of breast cancer (two to from a POLE p.Asp287Glu carrier, a melanoma from a POLE fivefold), but are not associated with bilateral breast p.Ile307Val carrier, the MMR-proficient CRC from the POLE tumors.37,38 Available studies have been unable to quantify p.Ala426Val and CHEK2 c.593–1G>T carrier, and a CRC the postulated increased risk to CRC.39 The carrier of POLD1 from a POLD1 p.Arg521Gln carrier showed hypermutation 2094 Volume 22 | Number 12 | December 2020 | GENETICS in MEDICINE MUR et al ARTICLE a POLE_Human 1 69 Pol2_Yeast 1 79 POLE_Human 70 147 Pol2_Yeast 80 158 p.Y224C POLE_Human 148 227 Pol2_Yeast 159 230 p.D287E p.M294R p.I307V POLE_Human 228 307 Pol2_Yeast 231 308 p.N336S p.G380C POLE_Human 308 387 Pol2_Yeast 309 388 p.K425R p.A426V POLE_Human 388 467 Pol2_Yeast 389 468 p.L424V POLE_Human 468 547 Pol2_Yeast 469 547 b 1000 14 Number of revertants (colonies) Fold rate increase (mutated/wt) 800 12 10 600 8 400 6 4 200 2 0 0 +) 4C 0C 4R +) 4C 0C 6V 7E 4R V T 6V 7E V T W (C 07 W (C 07 22 38 29 42 28 22 38 29 42 28 I3 I3 4V 4V Y G M A D Y G M A D p. p. p. p. p. p. p. p. p. p. p. p. 42 42 L L p. p. Fig. 2 Functional analysis carried out in Schizosaccharomyces pombe for variants located within (or close to) the POLE exonuclease domain (ED). (a) Alignment of human POLE and their homolog in yeast (Pol2). The identified variants are highlighted in yellow (conserved residues) and the POLE p.Leu424Val, used as positive control, in green. ED is depicted in red (human: residues 268–471, yeast: residues 98–428) and its sequence motifs44 are shaded in gray. (b) Left panel: box plots showing mutation rates of ade6-485 S. pombe (number of colonies) for pol2 wildtype (WT, negative control), ED mutation-positive control (pol2-Leu425Val; C+) and the corresponding variants. Right panel: fold rate increase relative to the median number of revertants in the WT. Data obtained from two independent experiments performed in triplicate. *p value = 0.01, **p value = 0.01–0.001, and ***p value < 0.001 indicate the differences with the WT clone, and were calculated using the Mann–Whitney nonparametric test. but none of the POLE/D1-associated mutational signatures ED variant classification (Fig. 4; Supplementary Table S7). Sixteen (16/2813; 0.57%) probands carried novel or rare ED To help characterize the identified variants, we used the missense variants. Taking into account several lines of TCGA and COSMIC tumor sequencing data to see if those evidence (functional data, tumor mutation burden and variants, when occurring somatically and in absence of signatures, in silico predictors, cosegregation, and additional somatic POLE/D1 ED variants, caused a hyper/ case–control data), and applying the ACMG/AMP guidelines ultramutated phenotype and the POLE/D1 ED-associated for variant classification, POLE p.Met294Arg was classified as signatures. Considering the ED variants identified in the pathogenic; POLE p.Asp287Glu, POLE p.Asn336Ser, POLD1 study, only two, POLD1 p.Arg352Cys and p.Arg525Trp, p.Arg352Cys, and POLD1 p.Arg521Gln as likely benign; and fulfilled the abovementioned conditions. A CRC and a POLE p.Ile307Val, p.Gly380Cys, p.Lys425Arg, and p.Ala426- lymphoid neoplasm harboring POLD1 p.Arg352Cys, as well Val, and POLD1 p.Ile307Met, p.Ile325Val, and p.Arg525Trp as a stomach carcinoma with POLD1 p.Arg525Trp, displayed as variants of unknown significance (VUS) (Table 1). Details hypermutation but not the POLE/D1 ED-associated signa- of the criteria considered for the classification of each variant tures (Supplementary Table S8). In the case of the CRC and are shown in Supplementary Table S3. the stomach carcinoma, the hypermutation detected was due to the presence of microsatellite instability (MSI) (MMR Germline loss-of-function POLE and POLD1 variants deficiency signatures’ contribution: 77%), and to the presence Six LoF heterozygous variants were identified in the retro- of BRCA2 p.Pro1088Leufs*16 (BRCA1/2 signature contribu- spective (n = 1) and prospective (n = 6) cohorts: POLE tion [3]: 52%), respectively. In the POLD1 p.Arg352Cys c.1185_1188delGGAG (p.Glu396Thrfs*15), c.4480C>T hematologic tumor, in absence of DNA repair alterations, no (p.Gln1494*), and c.2297_2298insA (p.Tyr766*); and POLD1 POLE/D1 ED-associated signatures were identified. c.230delC (p.Pro77Leufs*92), c.1195C>T (p.Gln399*), and GENETICS in MEDICINE | Volume 22 | Number 12 | December 2020 2095 ARTICLE MUR et al a N336S b G380C M294R I325V I307V D287E R352C R521Q K425R A426V R525W I307M Fig. 3 Structural representation of human POLE and POLD1 and location of the ED variants identified in the current study. (a) 3D structure of POLE. (b) 3D structure of POLD1. Single-stranded DNA from the aligned bacteriophage T4 polymerase complex (PDB ID: 1NOY) is shown in yellow. Variants in the DNA binding pocket are highlighted in red. c.3305delC (p.Pro1102Leufs*22) (Supplementary Table S9). observed in cancer-free controls (0.22% [5/2309] in cases vs. The most prevalent phenotype in the carrier families was 0.20% [119/59,095] in controls, p value = 0.81), thus agreeing breast cancer (also the most represented tumor type in the with their lack of association with cancer predisposition41 studied cohorts), diagnosed in eight carriers (five families) (LoF variants present in the European population data (pedigrees: Supplementary Fig. S4). The POLD1 p.Gln399* set [gnomAD] are detailed in Supplementary Table S10). carrier also carried a BRCA1 pathogenic variant (exon 8–13 Moreover, tumor sequencing demonstrated no increased deletion). mutation burden, supporting intact DNA repair. POLE c.1185_1188delGGAG (p.Glu396Thrfs*15), novel and disrupting the ED, perfectly segregated with breast cancer Variants identified outside the ED in the carrier family: the variant was present in the proband, We identified 142 missense variants outside the ED (POLE n her mother, and two maternal aunts, all four of whom were = 92 and POLD1 n = 50) in 354 patients, with no relative diagnosed with breast cancer (ages at diagnosis: 28, 71, 64, overrepresentation among any of the phenotype-based and 67, respectively). Two primary breast tumors from two groups. Four POLD1 and 19 POLE outside-ED variants, carriers of POLE p.Glu396Thrfs*15 were analyzed by exome identified in 29 individuals, were predicted deleterious sequencing. None of the tumors showed hypermutation (Supplementary Table S11). (1.09–1.70 Mut/Mb) or a POLE-associated signature (Fig. 4; Cosegregation data suggested that POLE c.671A>G Supplementary Table S7). Similar results were observed when (p.Tyr224Cys), the closest to the ED (44 amino acids upstream), analyzing two CRC tumors with a somatic variant in that was not associated with the increased breast cancer risk same residue (Supplementary Table S8). No additional POLE observed in the carrier family (Supplementary Table S11); or POLD1 somatic pathogenic variants were detected in the observation supported by a yeast functional assay that revealed tumors, discarding a somatic second hit. no mutator effect (Fig. 2). No cosegregation analyses could be Tumor exome sequencing performed in the MMR-deficient performed for the remaining outside-ED predicted pathogenic colon cancer developed by the POLD1 c.230delC (p.Pro77- variants. Tumors harboring the same somatic outside-ED Leufs*92) carrier revealed that ~85% of the signatures’ missense variants in absence of other (likely) pathogenic POLE contribution corresponded to MMR deficiency and no variants did not have ED-associated mutational signatures. representation of POLE/D1-associated signatures (Fig. 4; Sup- Eight of the 23 predicted pathogenic variants (41/142 of all plementary Table S7). Sequencing data from a breast tumor variants) could be analyzed in different tumor types with this harboring a somatic frameshift variant in the same residue as approach (Supplementary Table S8). The frequency of outside- the germline POLD1 c.3305delC (p.Pro1102Leufs*22) showed ED predicted pathogenic variants in familial cancer (and all the neither hypermutation nor the POLE/D1-associated mutational subtypes considered separately) does not differ from that signature (Supplementary Table S8). observed in cancer-free controls (1.26% [29/2309] in cases vs. The frequency of LoF variants in familial cancer (and all the 0.91% [539/59,095] in controls, p value = 0.10) (Supplementary subtypes considered separately) is almost identical to that Table S12). 2096 Volume 22 | Number 12 | December 2020 | GENETICS in MEDICINE MUR et al ARTICLE a Variants Family – Individual Tumor type Age at dx MSI statusc MMR expression MMR deficient Mut/Mb p.L424V (C+) (Valle 2014; Bellido 2016) Oligodendroglioma 30 n.a. n.a. n.a. 477.60 p.D287E Fam A –II.3 Breast cancer 40 MSS n.a. No 185.74 p.M294R (1) Fam C –II.2 Ovarian cancera 36 n.a. MSH6- Yes 248.98 p.M294R (2) Fam C –II.2 Endometrial cancer 36 n.a. MLH1/MSH2/MSH6- Yes 300.85 POLE p.M294R (3) Fam C –I.4 CRC 42 MSS Normal No 10.96 p.I307V Fam E –II.1 Melanoma 81 MSS n.a. No 32.35 p.G380C Fam H –II.4 Breast cancer 61 n.a. n.a. n.a. 0.67 p.A426V Fam K –II.1 CRC 44 MSS Normal No 41.18 p.E396Wfs*15 (1) Fam Q –I.2 Breast cancer 64 MSS n.a. n.a. 1.09 p.E396Wfs*15 (2) Fam Q –I.3 Breast cancer 67 MSS n.a. n.a. 1.70 p.D316G (C+) Fam 2 –III.2 (Bellido 2016) Liver metastasisb n.a. MSS n.a. n.a. 13.66 p.D316H (C+) Fam 1 –III.8 (Bellido 2016) CRC 58 MSS n.a. No 62.15 POLD1 p.P77Lfs* 92 Fam T –III.2 CRC 40 MSI MLH1/PMS2- Yes 37.42 p.V295M Fam 5 –III.1 (Bellido 2016) CRC 48 MSS Normal No 4.89 p.R521Q Fam 6 –II.1 (Bellido 2016) CRC 48 MSS Normal No 38.54 a High-grade serous ovarian carcinoma b Intstinal origin c MSI classification using Bethessa panel b POLE variants POLD1 variants 100 90 MMR deficiency (6, 15 and 26) 80 POLD1 pathogenic variant + MMR deficiency (20) 70 Signature contribution (%) POLE pathogenic variant + MMR deficiency (14) 60 POLE/POLD1 pathogenic variant (10) 50 non-MMR and non-POLE/D1 defective DNA repair 40 Other signatures 30 20 10 0 p. p. p. p. p. p. p. p. p. p. p. D A P D L D R M I3 M M 42 42 77 31 31 28 52 29 07 29 29 4V 6V Lf 6G 6H 7E 1Q 4R V 4R 4R s* (C (C (C 92 (1 (2 (3 +) ) +) +) ) ) Fig. 4 Somatic analysis performed in tumors from POLE/POLD1 variant carriers. (a) Tumor features including mismatch repair (MMR) deficiency status and mutational burden (hypermutation was considered when the tumor had more than 10 exonic Mut/Mb). aHigh-grade serous ovarian cancer. b Intestinal origin. cMicrosatellite instability (MSI) classification using Bethesda panel. C+, positive controls, i.e., tumors from carriers of variants affecting the POLE/POLD1 ED previously classified as pathogenic. (b) Mutational signature contribution (DeconstructSigs) for hyper/ultramutated tumors (>10 Mut/Mb). CRC colorectal cancer. Tumor spectrum and prevalence of germline POLE and cancer cohort, reached 0.33% (2/601) when restricted to CRC POLD1 ED variants in hereditary cancer and polyposis phenotypes (Fig. 1; Supplementary Table S13). Two hereditary cancer families (2/2813; 0.07%) carried a The frequency of rare ED missense variants—including pathogenic variant in POLE (p.Met294Arg), making a total of pathogenic and VUS—in POLE and POLD1 in hereditary 5 carriers. The phenotypic spectrum of the carriers included cancer was 0.39% (9/2309). Almost half of the pathogenic (1) CRC in absence of adenomatous polyposis, diagnosed in 4 variants and VUS (4/9; 44.44%) were found in patients with a of the 5 carriers (80%) (mean age at CRC diagnosis 51; range referral of hereditary nonpolyposis CRC or polyposis, who 37–57), one of whom had developed three metachronous accounted for 26% (601/2309) of the total number of patients colorectal tumors; (2) endometrial cancer, diagnosed in two tested. In this group of patients, the prevalence of ED variants carriers (40%) (mean age at diagnosis 46; range 36–55); and increased to 0.67%. With regard to the most represented (3) ovarian cancer, diagnosed in one carrier (20%) (age at group of patients, i.e., those referred for genetic testing as diagnosis 36). Three ED variant carriers (60%) developed ≥2 hereditary breast and/or ovarian cancer, the frequency of primary tumors. The prevalence of (likely) pathogenic ED ED variants was 0.27% (4/1468) (Fig. 1; Supplementary variants, being of 0.09% (2/2309) in the prospective familial Table S13). GENETICS in MEDICINE | Volume 22 | Number 12 | December 2020 2097 ARTICLE MUR et al DISCUSSION cleft shows a mutator effect in yeast, even if it is not POLE and POLD1 ED variant interpretation is challenging pathogenic, as occurs with p.Asp287Glu. Based on this and and the inclusion of multiple pieces of evidence is highly the high variability of the yeast assay among replicates and advisable. Population and cosegregation data, in silico experiments, also observed in other studies,35,40 we have used prediction of pathogenicity, location within the DNA binding its results with caution. We posit that yeast assay observations cleft, results of the functional repair assay, and tumor may be considered supporting or, at most, moderate mutational data were taken into consideration for variant (not strong) evidence for ACMG/AMP variant classification, classification. Based the relevance of each piece of evidence, depending on the level of significance when compared with we adapted the ACMG/AMP variant classification guidelines the wildtype clone. to the particular conditions of POLE and POLD1 ED variants In our study, two families carrying POLE and POLD1 VUS and to the relevance of the functional information gathered, also carried pathogenic variants in ATM and CHEK2, two i.e., tumor mutational signatures, yeast repair assay results, genes that confer a moderate risk of breast cancer.37,38 The and alteration of the DNA binding pocket (Table S3). aggressive phenotype developed by the POLD1 p.Arg525Arg We consider tumor sequence analysis of utmost importance and ATM (p.Ser2407*) carrier supports a cumulative effect for a proper classification of POLE and POLD1 ED variants. of the two variants, while that may not be the case for the Mutational signature 10, in the presence of hyper/ultramuta- carriers of POLE p.Ala426Val (VUS) and CHEK2 tion, is unequivocally linked to the presence of an ED c.593–1G>T, where two of the three carriers developed late- pathogenic variant—a priori independently of the tumor type onset CRC. Hamzaoui et al. identified two relatives who analyzed—however, one must be certain of the inexistence of carried both POLE p.K425R (VUS) and MSH2 c.942+3A>T a somatic ED pathogenic variant that could be causing the (pathogenic), in a family with multiple CRC-affected presence of the POLE/D1-associated tumor molecular phe- members diagnosed at extremely young ages (range: 19–33), notype. Moreover, a challenging situation occurs when the suggesting a cumulative effect of the two variants.35 Whether tumor we analyze has other DNA repair defects, e.g., MMR the aggregation of cancer in these families is explained only by deficiency (MSI). In this situation, if the mutational signature the pathogenic variants in ATM, CHEK2, or MSH2, or the analysis does not reveal any of the POLE/D1-associated combined effect with the POLE/D1 variants (all of them signatures, i.e., signatures 14 or 20 when the germline variant classified as VUS until now), remains to be elucidated. affects the ED of POLE or ED respectively, it is highly The role of LoF and outside-ED POLE and POLD1 advisable to analyze a second tumor, ideally MMR proficient, heterozygous variants has been a matter of controversy since developed in the same family or harboring the same variant the description of PPAP in 2013. Our data suggest that these but in a somatic context, verifying that no other somatic ED variants are most likely nonpathogenic, at least not associated variant is present. with PPAP—this being relevant to avoid misdiagnoses in Due to the high homology of the ED of POLE and POLD1 the clinical setting. A particular case is that of POLE in human and yeast, this model has been widely used for p.Glu396Thrfs*15, identified in four breast cancer–affected variant functional assessment. In our study, unequivocal women of the same family in absence of additional pathogenic repair impairment (p < 0.001) was observed for POLE variants in known or candidate hereditary cancer genes p.Leu424Val (positive control), p.Met294Arg (likely patho- (exome sequencing; data not shown). Despite the absence of genic), and p.Ala426Val (uncertain significance), but also for tumor hypermutation and POLE-associated mutational sig- p.Asp287Glu, a variant classified as likely benign based on natures, whether the variant is the cause of the cancer cosegregation data, frequency in cases and controls, and aggregation in the family remains to be elucidated, and if so, absence of an ED-associated mutational signature. It seems what is the molecular mechanism underlying its potential evident that, in addition to the (likely) pathogenic variants causal effect. directly affecting the exonuclease catalytic residues, such as In conclusion, our findings indicate that PPAP constitutes POLD1 p.Asp316Gly, p.Asp316His, ED variants affecting 0.1–0.4% of familial cancer cases referred to a hereditary residues in direct contact or in very close proximity to the cancer clinic, 0.3–0.7% when considering nonpolyposis CRC DNA fragment, i.e., in the DNA binding pocket, have a very and polyposis patients, with the most commonly associated high chance of being pathogenic. In fact, the vast majority of tumor types being colorectal, endometrial, and ovarian POLE and POLD1 (likely) pathogenic variants reported to cancers. ED variant interpretation is challenging and should date are localized in the DNA binding cleft, including POLE include multiple pieces of evidence. LoF and outside-ED p.Thr278Lys, p.Met294Arg (this study), p.P286L, p.N363K, variants are not associated with PPAP, a cancer-predisposing p. D368N, p.Leu424Val, and p.Pro436Ser, as well as POLD1 syndrome characterized by a defect in the proofreading p.Leu474Pro and p.Ser478Asn (Supplementary Table S2). activity of polymerases epsilon and delta. Hamzaoui et al. observed that variants interfering with DNA binding (p.P286L and p.N363K) produce a higher mutagenic SUPPLEMENTARY INFORMATION effect in yeast than variants disrupting ion metal coordination The online version of this article (https://doi.org/10.1038/s41436- at the exonuclease site.35 In line with Barbari et al.,15,40 our 020-0922-2) contains supplementary material, which is available results indicate that any variant located in the DNA binding to authorized users. 2098 Volume 22 | Number 12 | December 2020 | GENETICS in MEDICINE MUR et al ARTICLE ACKNOWLEDGEMENTS 15. Barbari SR, Shcherbakova PV. Replicative DNA polymerase defects in We thank Gemma Aiza, Mireia Menéndez, Sara González, and human cancers: consequences, mechanisms, and implications for therapy. DNA Repair (Amst). 2017;56:16–25. Xavier Muñoz for support, and Biobanc HUB-ICO-IDIBELL, 16. Kandoth C, Schultz N, Cherniack AD, et al. Integrated genomic integrated in the Spanish Platform Biobanks Network and funded characterization of endometrial carcinoma. Nature. 2013;497:67–73. by Instituto de Salud Carlos III (PT17/0015/0024) and Xarxa Bancs 17. Church DN, Briggs SE, Palles C, et al. DNA polymerase ε and δ exonuclease domain mutations in endometrial cancer. Hum Mol Genet. de Tumors de Catalunya sponsored by Pla Director d’Oncologia 2013;22:2820–2828. de Catalunya. This work was funded by the Spanish Ministry of 18. Church DN, Stelloo E, Nout RA, et al. Prognostic significance of POLE Science and Innovation, cofunded by FEDER funds (SAF2016- proofreading mutations in endometrial cancer. J Natl Cancer Inst. 2015; 107:402. 80888-R, SAF2015-68016-R, Severo Ochoa SVP-2014-068895 19. Shinbrot E, Henninger EE, Weinhold N, et al. Exonuclease mutations in contract [L.M.-P.]); Instituto de Salud Carlos III (PI16/00563, PI16/ DNA polymerase epsilon reveal replication strand specific mutation 00588, PI14/00613, PI19/00553, CIBERONC CB16/12/00234, patterns and human origins of replication. Genome Res. 2014;24: 1740–1750. CIBERESP, Sara Borrell contract [P.M.]); Government of Catalonia 20. Zou Y, Liu FY, Liu H, et al. Frequent POLE1 p.S297F mutation in [AGAUR 2017SGR1282, CERCA Program]; and Fundación Olga Chinese patients with ovarian endometrioid carcinoma. Mutat Res. 2014; Torres. This study was facilitated by COST Action CA17118, 761:49–52. 21. Erson-Omay EZ, Çağlayan AO, Schultz N, et al. Somatic POLE mutations supported by COST (European Cooperation in Science and cause an ultramutated giant cell high-grade glioma subtype with better Technology). prognosis. Neuro Oncol. 2015;17:1356–1364. 22. Campbell BB, Light N, Fabrizio D, et al. Comprehensive Analysis of Hypermutation in Human Cancer. Cell. 2017;171:1042–.e1010. 23. Briggs S, Tomlinson I. Germline and somatic polymerase ε and δ mutations DISCLOSURE define a new class of hypermutated colorectal and endometrial cancers. The authors declare no conflicts of interest. J Pathol. 2013;230:148–153. 24. Alexandrov LB, Nik-Zainal S, Wedge DC, et al. Signatures of mutational processes in human cancer. Nature. 2013;500:415–421. Publisher’s note Springer Nature remains neutral with regard 25. Haradhvala NJ, Kim J, Maruvka YE, et al. Distinct mutational signatures to jurisdictional claims in published maps and institutional characterize concurrent loss of polymerase proofreading and mismatch affiliations. repair. Nat Commun. 2018;9:1746. 26. Bourdais R, Rousseau B, Pujals A, et al. Polymerase proofreading domain mutations: new opportunities for immunotherapy in hypermutated REFERENCES colorectal cancer beyond MMR deficiency. Crit Rev Oncol Hematol. 2017; 1. Palles C, Cazier JB, Howarth KM, et al. Germline mutations affecting the 113:242–248. proofreading domains of POLE and POLD1 predispose to colorectal 27. Feliubadaló L, Tonda R, Gausachs M, et al. Benchmarking of whole adenomas and carcinomas. Nat Genet. 2013;45:136–144. exome sequencing and ad hoc designed panels for genetic testing of 2. Valle L, Hernández-Illán E, Bellido F, et al. New insights into POLE and hereditary cancer. Sci Rep. 2017;7:37984. POLD1 germline mutations in familial colorectal cancer and polyposis. 28. Castellanos E, Gel B, Rosas I, et al. A comprehensive custom panel design Hum Mol Genet. 2014;23:3506–3512. for routine hereditary cancer testing: preserving control, improving 3. Rohlin A, Eiengård F, Lundstam U, et al. GREM1 and POLE variants in diagnostics and revealing a complex variation landscape. Sci Rep. 2017; hereditary colorectal cancer syndromes. Genes Chromosomes Cancer. 7:39348. 2016;55:95–106. 29. Ioannidis NM, Rothstein JH, Pejaver V, et al. REVEL: an ensemble method 4. Elsayed FA, Kets CM, Ruano D, et al. Germline variants in POLE are for predicting the pathogenicity of rare missense variants. Am J Hum associated with early onset mismatch repair deficient colorectal cancer. Genet. 2016;99:877–885. Eur J Hum Genet. 2015;23:1080–1084. 30. Ghosh R, Oak N, Plon SE. Evaluation of in silico algorithms for use 5. Spier I, Holzapfel S, Altmüller J, et al. Frequency and phenotypic spectrum with ACMG/AMP clinical variant interpretation guidelines. Genome Biol. of germline mutations in POLE and seven other polymerase genes in 266 2017;18:225. patients with colorectal adenomas and carcinomas. Int J Cancer. 2015; 31. Alexandrov LB, Kim J, Haradhvala NJ, et al. The repertoire of mutational 137:320–331. signatures in human cancer. Nature. 2020;578:94–101. 6. Chubb D, Broderick P, Frampton M, et al. Genetic diagnosis of 32. Richards S, Aziz N, Bale S, et al. Standards and guidelines for the high-penetrance susceptibility for colorectal cancer (CRC) is achievable interpretation of sequence variants: a joint consensus recommendation for a high proportion of familial CRC by exome sequencing. J Clin Oncol. of the American College of Medical Genetics and Genomics and the 2015;33:426–432. Association for Molecular Pathology. Genet Med. 2015;17:405–424. 7. Bellido F, Pineda M, Aiza G, et al. POLE and POLD1 mutations in 529 33. Heitzer E, Tomlinson I. Replicative DNA polymerase mutations in cancer. kindred with familial colorectal cancer and/or polyposis: review of reported Curr Opin Genet Dev. 2014;24:107–113. cases and recommendations for genetic testing and surveillance. Genet 34. Jansen AM, van Wezel T, van den Akker BE, et al. Combined mismatch Med. 2016;18:325–332. repair and POLE/POLD1 defects explain unresolved suspected Lynch 8. Hansen MF, Johansen J, Bjørnevoll I, et al. A novel POLE mutation syndrome cancers. Eur J Hum Genet. 2016;24:1089–1092. associated with cancers of colon, pancreas, ovaries and small intestine. 35. Hamzaoui N, Alarcon F, Leulliot N, et al. Genetic, structural, and Fam Cancer. 2015;14:437–448. functional characterization of POLE polymerase proofreading variants 9. Rohlin A, Zagoras T, Nilsson S, et al. A mutation in POLE predisposing to a allows cancer risk prediction. Genet Med. 2020 May 19; https://doi.org/ multi-tumour phenotype. Int J Oncol. 2014;45:77–81. 10.1038/s41436-020-0828-z [Epub ahead of print]. 10. Aoude LG, Heitzer E, Johansson P, et al. POLE mutations in families 36. Shirts BH, Casadei S, Jacobson AL, et al. Improving performance of predisposed to cutaneous melanoma. Fam Cancer. 2015;14:621–628. multigene panels for genomic analysis of cancer predisposition. Genet 11. Brannon AR, Vakiani E, Sylvester BE, et al. Comparative sequencing Med. 2016;18:974–981. analysis reveals high genomic concordance between matched primary 37. Couch FJ, Shimelis H, Hu C, et al. Associations between cancer and metastatic colorectal cancer lesions. Genome Biol. 2014;15:454. predisposition testing panel genes and breast cancer. JAMA Oncol. 2017; 12. Network CGA. Comprehensive molecular characterization of human 3:1190–1196. colon and rectal cancer. Nature. 2012;487:330–337. 38. Easton DF, Pharoah PD, Antoniou AC, et al. Gene-panel sequencing and 13. Seshagiri S, Stawiski EW, Durinck S, et al. Recurrent R-spondin fusions in the prediction of breast-cancer risk. N Engl J Med. 2015;372:2243–2257. colon cancer. Nature. 2012;488:660–664. 39. West AH, Blazer KR, Stoll J, et al. Clinical interpretation of pathogenic 14. Giannakis M, Mu XJ, Shukla SA, et al. Genomic correlates of immune-cell ATM and CHEK2 variants on multigene panel tests: navigating moderate infiltrates in colorectal carcinoma. Cell Rep. 2016;17:1206. risk. Fam Cancer. 2018;17:495–505. GENETICS in MEDICINE | Volume 22 | Number 12 | December 2020 2099 ARTICLE MUR et al 40. Barbari SR, Kane DP, Moore EA, Shcherbakova PV. Functional analysis Open Access This article is licensed under a Creative Commons of cancer-associated DNA polymerase ε variants in Saccharomyces Attribution 4.0 International License, which permits use, sharing, cerevisiae. G3 (Bethesda). 2018;8:1019–1029. adaptation, distribution and reproduction in any medium or format, as long as you 41. Mur P, Palles C, Tomlinson I, Valle L. Reply to: “Development of an MSI- give appropriate credit to the original author(s) and the source, provide a link to positive colon tumor with aberrant DNA methylation in a PPAP patient”. the Creative Commons license, and indicate if changes were made. The images or J Hum Genet. 2020;65:513–514. other third party material in this article are included in the article’s Creative 42. Li Q, Wang K. InterVar: clinical Interpretation of genetic variants by the Commons license, unless indicated otherwise in a credit line to the material. If 2015 ACMG-AMP guidelines. Am J Hum Genet. 2017;100:267–280. material is not included in the article’s Creative Commons license and your 43. Bellido F, Pineda M, Aiza G, et al. POLE and POLD1 mutations in 529 intended use is not permitted by statutory regulation or exceeds the permitted kindred with familial colorectal cancer and/or polyposis: review of reported use, you will need to obtain permission directly from the copyright holder. To view cases and recommendations for genetic testing and surveillance. Genet a copy of this license, visit http://creativecommons.org/licenses/by/4.0/. Med. 2015;18:325–332. 44. Shevelev IV, Hübscher U. The 3’ 5’ exonucleases. Nat Rev Mol Cell Biol. 2002;3:364–376. © The Author(s) 2020 2100 Volume 22 | Number 12 | December 2020 | GENETICS in MEDICINE