Articles Genotype-first approach to identify associations between CDH1 germline variants and cancer phenotypes: a multicentre study by the European Reference Network on Genetic Tumour Risk Syndromes José Garcia-Pelaez, Rita Barbosa-Matos*, Silvana Lobo*, Alexandre Dias*, Luzia Garrido, Sérgio Castedo, Sónia Sousa, Hugo Pinheiro, Liliana Sousa, Rita Monteiro, Joaquin J Maqueda, Susana Fernandes, Fátima Carneiro, Nádia Pinto, Carolina Lemos, Carla Pinto, Manuel R Teixeira, Stefan Aretz, Svetlana Bajalica-Lagercrantz, Judith Balmaña, Ana Blatnik, Patrick R Benusiglio, Maud Blanluet, Vincent Bours, Hilde Brems, Joan Brunet, Daniele Calistri, Gabriel Capellá, Sergio Carrera, Chrystelle Colas, Karin Dahan, Robin de Putter, Camille Desseignés, Elena Domínguez-Garrido, Conceição Egas, D Gareth Evans, Damien Feret, Eleanor Fewings, Rebecca C Fitzgerald, Florence Coulet, María Garcia-Barcina, Maurizio Genuardi, Lisa Golmard, Karl Hackmann, Helen Hanson, Elke Holinski-Feder, Robert Hüneburg, Mateja Krajc, Kristina Lagerstedt-Robinson, Conxi Lázaro, Marjolijn J L Ligtenberg, Cristina Martínez-Bouzas, Sonia Merino, Geneviève Michils, Srdjan Novaković, Ana Patiño-García, Guglielmina Nadia Ranzani, Evelin Schröck, Inês Silva, Catarina Silveira, José L Soto, Isabel Spier, Verena Steinke-Lange, Gianluca Tedaldi, María-Isabel Tejada, Emma R Woodward, Marc Tischkowitz, Nicoline Hoogerbrugge, Carla Oliveira Summary Background Truncating pathogenic or likely pathogenic variants of CDH1 cause hereditary diffuse gastric cancer Lancet Oncol 2023; 24: 91–106 (HDGC), a tumour risk syndrome that predisposes carrier individuals to diffuse gastric and lobular breast cancer. Published Online Rare CDH1 missense variants are often classified as variants of unknown significance. We conducted a genotype– November 24, 2022 https://doi.org/10.1016/ phenotype analysis in families carrying rare CDH1 variants, comparing cancer spectrum in carriers of pathogenic or S1470-2045(22)00643-X likely pathogenic variants (PV/LPV; analysed jointly) or missense variants of unknown significance, assessing the This online publication has frequency of families with lobular breast cancer among PV/LPV carrier families, and testing the performance of been corrected. The corrected lobular breast cancer-expanded criteria for CDH1 testing. version first appeared at thelancet.com/oncology on January 3, 2022 Methods This genotype-first study used retrospective diagnostic and clinical data from 854 carriers of 398 rare CDH1 variants and 1021 relatives, irrespective of HDGC clinical criteria, from 29 institutions in ten member-countries of the *Contributed equally European Reference Network on Tumour Risk Syndromes (ERN GENTURIS). Data were collected from Oct 1, 2018, Instituto de Investigação e Inovação em Saúde (i3S) to Sept 20, 2022. Variants were classified by molecular type and clinical actionability with the American College of (J Garcia-Pelaez MSc, Medical Genetics and Association for Molecular Pathology CDH1 guidelines (version 2). Families were categorised R Barbosa-Matos MSc, by whether they fulfilled the 2015 and 2020 HDGC clinical criteria. Genotype–phenotype associations were analysed S Lobo MSc, A Dias MSc, by Student’s t test, Kruskal-Wallis, χ², and multivariable logistic regression models. Performance of HDGC clinical Prof S Castedo MD PhD, S Sousa BSc, H Pinheiro MD PhD, criteria sets were assessed with an equivalence test and Youden index, and the areas under the receiver operating L Sousa MSc, C Lemos PhD, characteristic curves were compared by Z test. R Monteiro MSc, J J Maqueda MSc, Findings From 1971 phenotypes (contributed by 854 probands and 1021 relatives aged 1–93 years), 460 had gastric and Prof F Carneiro MD PhD, N Pinto PhD, breast cancer histology available. CDH1 truncating PV/LPVs occurred in 176 (21%) of 854 families and missense Prof C Oliveira PhD), Faculty of variants of unknown significance in 169 (20%) families. Multivariable logistic regression comparing phenotypes Medicine (J Garcia-Pelaez, occurring in families carrying PV/LPVs or missense variants of unknown significance showed that lobular breast Prof S Castedo, S Fernandes PhD, cancer had the greatest positive association with the presence of PV/LPVs (odds ratio 12·39 [95% CI 2·66–57·74], Prof F Carneiro, Prof C Oliveira), Institute of Biomedical p=0·0014), followed by diffuse gastric cancer (8·00 [2·18–29·39], p=0·0017) and gastric cancer (7·81 [2·03–29·96], Sciences Abel Salazar p=0·0027). 136 (77%) of 176 families carrying PV/LPVs fulfilled the 2015 HDGC criteria. Of the remaining (R Barbosa-Matos, S Lobo, 40 (23%) families, who did not fulfil the 2015 criteria, 11 fulfilled the 2020 HDGC criteria, and 18 had lobular breast C Lemos PhD, M R Teixeira MD PhD), cancer only or lobular breast cancer and gastric cancer, but did not meet the 2020 criteria. No specific CDH1 variant Center of Mathematics was found to predispose individuals specifically to lobular breast cancer, although 12 (7%) of 176 PV/LPV carrier (N Pinto), and Institute for families had lobular breast cancer only. Addition of three new lobular breast cancer-centred criteria improved testing Molecular and Cell Biology sensitivity while retaining high specificity. The probability of finding CDH1 PV/LPVs in patients fulfilling the lobular (C Lemos), University of Porto, Porto, Portugal; Instituto de breast cancer-expanded criteria, compared with the 2020 criteria, increased significantly (AUC 0·92 vs 0·88; Patologia e Imunologia Z score 3·54; p=0·0004). Molecular da Universidade do Porto, Porto, Portugal Interpretation CDH1 PV/LPVs were positively associated with HDGC-related phenotypes (lobular breast cancer, (J Garcia-Pelaez, R Barbosa-Matos, S Lobo, A Dias, diffuse gastric cancer, and gastric cancer), and no evidence for a positive association with these phenotypes Prof S Castedo, S Sousa, was found for CDH1 missense variants of unknown significance. CDH1 PV/LPVs occurred often in families H Pinheiro, L Sousa, R Monteiro, with lobular breast cancer who did not fulfil the 2020 HDGC criteria, supporting the expansion of lobular breast J J Maqueda, Prof F Carneiro, cancer-centred criteria. N Pinto, Prof C Oliveira); Centro www.thelancet.com/oncology Vol 24 January 2023 91 Articles Hospitalar Universitário Funding European Reference Network on Genetic Tumour Risk Syndromes, European Regional Development Fund, São João, Porto, Portugal Fundação para a Ciência e a Tecnologia (Portugal), Cancer Research UK, and European Union’s Horizon 2020 (L Garrido RN, Prof S Castedo, Prof F Carneiro); Serviço de research and innovation programme. Medicina Interna, Centro Hospitalar Tâmega e Sousa, Copyright © 2022 Elsevier Ltd. All rights reserved. Penafiel, Portugal (H Pinheiro); Escola de Economia e Gestão, Universidade do Minho, Braga, Introduction the 2010 and 2015 guidelines widened testing to isolated Portugal (L Sousa); Bioinf2Bio, Hereditary diffuse gastric cancer (HDGC; OMIM patients with diffuse gastric cancer, aged 40 years or Porto, Portugal (J J Maqueda); number #137215) is an autosomal dominant tumour risk younger, and to individuals or families with both diffuse Department of Laboratory syndrome with increased predisposition to develop early gastric cancer and lobular breast cancer before 50 years of Genetics, Portuguese Oncology Institute of Porto, Porto, onset diffuse gastric cancer and lobular breast cancer.1–3 age.1,7 Phenotype-driven HDGC guidelines were super­ Portugal (C Pinto MSc, HDGC is mainly caused by germline single-nucleotide seded by genotype-driven guidelines in 2020, with clinical M R Teixeira); Porto variants and copy-number variants in the E-cadherin selection criteria being reformulated predominantly on Comprehensive Cancer Center (CDH1) gene (Ensembl gene ID ENSG00000039068; the basis of actionable genetic test results (presence of a Raquel Seruca, Porto, Portugal (J Garcia-Pelaez, RefSeq ID NM_004360.4; Locus Reference Genomic CDH1 pathogenic or likely pathogenic variant).2 The 2020 R Barbosa-Matos, S Lobo, A Dias, ID LRG_301),4,5 which are classified as pathogenic or likely guideline proposed hereditary lobular breast cancer Prof S Castedo, S Sousa, pathogenic variants, according to the American College of (HLBC) as a lobular breast cancer only predisposition H Pinheiro, L Sousa, C Lemos, Medical Genetics and Genomics and Association for syndrome, independent of HDGC (appendix p 1).2,8 Prof F Carneiro, N Pinto, C Pinto, M R Teixeira, Prof C Oliveira); Molecular Pathology CDH1 variant curation guidelines.6 In a large US cohort of probands (tested because Institute of Human Genetics, The International Gastric Cancer Linkage Consortium of HDGC clinical ascertainment or multigene panel Medical Faculty, University of (IGCLC) has published four sets of clinical criteria for testing), individuals carrying CDH1 pathogenic or likely Bonn, Bonn, Germany CDH1 testing to date (appendix p 1).1–3,7 Although pathogenic variants were calculated to have a lifetime risk (Prof S Aretz MD PhD, I Spier MD); Department of guidelines from 1999 supported CDH1 genetic testing only of gastric cancer of 42% for men and 33% for women, Internal Medicine I in families with multiple cases of diffuse gastric cancer, whereas female breast cancer cumulative incidence was (R Hüneburg MD) and National Center for Hereditary Tumor Syndromes (Prof S Aretz, Research in context R Hüneburg, I Spier), University Hospital Bonn, Bonn, Germany; Evidence before this study designing more accurate clinical criteria and for management Department of Oncology- Pathology, Karolinska In 2018, we started a genotype-first study to define the cancer of families with HDGC. Institutet, Stockholm, Sweden landscape in families carrying CDH1 pathogenic or likely Added value of this study (S Bajalica-Lagercrantz MD PhD, pathogenic variants and missense variants of unknown K Lagerstedt-Robinson PhD); To our knowledge, this is the first multicentre significance, assess the frequency of families with lobular breast Department of Clinical genotype–phenotype study and the largest dataset ever Genetics, Cancer Genetic Unit cancer among families carrying pathogenic or likely pathogenic studied of individuals carrying rare CDH1 variants and their (S Bajalica-Lagercrantz, variants of CDH1, and test the performance of different sets of relatives. It highlights specificities in CDH1 variant-type K Lagerstedt-Robinson) and hereditary diffuse gastric cancer (HDGC) clinical criteria driving Cancer Theme related clinical phenotypes and supports the association of CDH1 testing. For the latter, we considered different sets of (S Bajalica-Lagercrantz), HDGC-related cancers with the presence of CDH1 pathogenic Karolinska University Hospital HDGC criteria developed between 2015 and 2022 (2015, 2020, or likely pathogenic variants, but did not find an association Solna, Stockholm, Sweden; Yale—2022) for comparison with the lobular breast cancer- of HDGC-related cancers with the presence of CDH1 missense Hospital Vall d’Hebron, expanded criteria herein proposed. We searched PubMed for Universitat Autonoma de variants of unknown significance. The study also studies in English on “CDH1 germline variants” and “diffuse Barcelona, Barcelona, Spain demonstrates that families not fulfilling HDGC clinical (J Balmaña MD PhD); gastric cancer” or “lobular breast cancer” or “LBC” or “hereditary criteria, but carrying a CDH1 PV/LPVs, are mainly lobular Department of Molecular diffuse gastric cancer” or “HDGC” from Jan 1, 2015 onwards (the breast cancer-enriched, supporting an expansion of current Diagnostics year of publication of the HDGC policy review and the first rules (Prof S Novaković PhD) and HDGC criteria to include additional lobular breast for variant classification in the clinical context, by the American Department of Clinical Cancer cancer-centred criteria. Genetics (A Blatnik MD, College of Medical Genetics and Genomics and Association for M Krajc MD PhD), Institute of Molecular Pathology). References from relevant articles, HDGC Implications of all the available evidence Oncology Ljubljana, Ljubljana, guidelines papers before 2015, reviews, and previous meta- Given the lack of evidence supporting a positive association Slovenia; Medical Genetics analyses were considered on a case-by-case basis to identify any between the presence of CDH1 germline missense variants and Department, Pitié-Salpêtrière Hospital, AP-HP and Sorbonne additional or relevant study not retrieved by the PubMed HDGC-related cancers, carriers of such variants are unlikely to University, Paris, France search. In 2020, a new HDGC policy review highlighted the lack be predisposed to HDGC. The recurrent identification of (P R Benusiglio MD, of genotype–phenotype correlations in CDH1 variant carriers, pathogenic or likely pathogenic variants in patients and C Desseignés MSc, which still vastly persists today, particularly regarding CDH1- families with lobular breast cancer, with or without gastric F Coulet PharmD PhD); Service de Génétique Oncologique, related predisposition to lobular breast cancer, associations cancer, supports the expansion of clinical criteria for CDH1 Institut Curie, Paris, France between missense CDH1 variants of unknown significance and testing in this setting. The observations herein reported will (L Golmard PharmD PhD, HDGC-related phenotypes, and CDH1 variant-specific cancer contribute to clarify the context for clinical management of M Blanluet PharmD, predisposition. This knowledge gap provides the rationale for families carrying CDH1 germline variants. C Colas MD PhD); Laboratory of 92 www.thelancet.com/oncology Vol 24 January 2023 Articles estimated at 55% at age 80 years.9 These cumulative genotype-first study to identify associations between rare Human Genetics, GIGA incidences are considerably lower for gastric cancer and CDH1 germline variants and cancer phenotypes, in the Institute, University of Liège, Liège, Belgium higher for breast cancer compared with cohorts strictly largest dataset of CDH1 rare variant carriers and their (V Bours MD PhD); Center of fulfilling HDGC criteria and those from other world relatives, the first and most detailed to date, to the best of Genetics, University Hospital, regions.10–13 Given the high risk of cancer in carriers of our knowledge. Liège, Belgium (V Bours); CDH1 pathogenic or likely pathogenic variants, current Department of Human Genetics, University of Leuven, guidelines recommend intensive gastric and breast Methods Leuven, Belgium (H Brems PhD, surveillance, prophylactic gastrectomy in men and Participants and data collection G Michils PhD); Hereditary women, and optional mastectomy in women, after a Data collection started on Oct 1, 2018, and ended Cancer Programme, Catalan CDH1 pathogenic or likely pathogenic variant is found in on Sept 20, 2022. CDH1 variant-related data were Institute of Oncology, Bellvitge Institute for Biomedical a family or individual with relevant history of cancer obtained from diagnostic laboratories across Europe Research and Girona in these organs.2 (29 institutions in ten European countries; figure 1A; Biomedical Research Institute, Many CDH1 pathogenic or likely pathogenic variants, appendix pp 2–4) and were the result of genetic testing Barcelona-Girona, Spain found by multigene panel testing, occur in breast cancer with either CDH1-targeted sequencing, or panel, whole- (J Brunet MD PhD); Laboratorio di Bioscienze, IRCCS Istituto patients not meeting the clinical criteria for HDGC.12,14 exome, or whole-genome sequencing. Only variants in Romagnolo per lo Studio dei Additionally, a 2022 cohort study of patients with cancer the CDH1 locus were collected for the purpose of the Tumori (IRST) “Dino Amadori”, showed that CDH1 germline pathogenic or likely current study. The presence of a rare single-nucleotide Meldola, Italy (D Calistri PhD, G Tedaldi PhD); Hereditary pathogenic variants are enriched in patients with diffuse variant or copy-number variant in CDH1, herein Cancer Program, Catalan gastric cancer and lobular breast cancer.15 The high classified as any variant with frequency below 0·01 in Institute of Oncology, Bellvitge number of patients with breast cancer carrying CDH1 gnomAD, in a proband irrespective of clinical diagnosis Institute for Biomedical pathogenic or likely pathogenic variants in the USA led or HDGC suspicion, drove the collection of aggregated Research, Barcelona, Spain (G Capellá MD PhD, Lerner and colleagues16 to propose the use of clinical molecular and clinical information from carrier C Lázaro PhD); Centro de criteria for hereditary breast and ovarian cancer, in probands and their relatives (unconfirmed carriers). Investigación Biomédica en addition to the modified and less restrictive 2020 criteria The term unconfirmed carrier refers to two different Red de Cáncer (CIBERONC), for HDGC (also known as Yale criteria), to select patients scenarios: affected relatives who were not submitted for Madrid, Spain (G Capellá, C Lázaro); Oncology Service, for CDH1 testing. Although this approach increased the carrier testing, particularly in the case of variants of Biocruces Bizkaia Health sensitivity of the criteria, no data were provided regarding unknown significance and likely benign or benign Research Institute, Cruces their specificity or positive predictive value.16 variants; and affected relatives from families carrying University Hospital, Cruces- CDH1 variants of unknown significance persist as a pathogenic or likely pathogenic variants who were Barakaldo, Bizkaia, Spain (S Carrera MD); Center of clinical challenge because of scarce genotype–phenotype generally submitted for carrier testing, but for whom Human Genetics, IPG, data, and missense variants, the largest class of coding the carrier status is generally unknown to this study. Gosselies, Belgium variants of unknown significance, have been claimed The cohort’s general clinical features (appendix pp 2–3), (K Dahan MD PhD, D Feret MSc); to preferentially predispose individuals to HLBC.8 In information on participating institutions (appendix p 4), Clinical Genetics Department, University Hospital of Ghent, parallel, endoscopic evidence of gastric cancer was found and molecular and clinical information indexed to Ghent, Belgium in more than 90% of families with HLBC, who carry probands from each family (appendix pp 5–46) (R de Putter MD); Molecular CDH1 pathogenic or likely pathogenic variants17, raising were collected. Diagnostics Laboratory, questions about whether HLBC is an HDGC-independent All patients signed an informed consent form for Fundación Rioja Salud, Logroño, Spain entity. Although colorectal cancer has been considered as germline testing, either mentioning future research (E Domínguez-Garrido PhD); a possible CDH1-associated phenotype in some studies,15,18 related to their susceptibility syndrome or allowing CNC—Center for Neuroscience others claim it is rare among carriers of CDH1 pathogenic sample biobanking for future research. In most and Cell Biology, University of or likely pathogenic variants.9 institutions, approval by ethics committees was waived, Coimbra, Coimbra, Portugal (C Egas PhD); Division of Genotype–phenotype association studies, particularly as this is a retrospective study; in the remaining Evolution and Genomic genotype-first studies, are expected to answer some of institutions, local ethics approval was granted. The Sciences, University of the abovementioned challenges. In this study, we aimed whole project received appraisal N19/CECRI/2022 Manchester, Manchester, UK to conduct a genotype–phenotype analysis in carriers from the Committee for Ethical and Responsible (Prof D G Evans MD, E R Woodward MD PhD); of rare CDH1 variants and their relatives, to compare Conduct of Research of the study’s leading institution, Manchester Centre for Genomic the cancer landscape in families carrying pathogenic Instituto de Investigação e Inovação em Saúde, Medicine, Manchester, UK or likely pathogenic variants (considered together for University of Porto, Portugal. (Prof D G Evans, E R Woodward); the purpose of our analyses) or missense variants Department of Medical Genetics, National Institute for of unknown significance, to assess the frequency of Clinical and variant data curation Health Research Cambridge families with lobular breast cancer among families Family history was revisited and compliance with the Biomedical Research Centre carrying pathogenic or likely pathogenic variants, and to 2015 criteria,1 2020 criteria,2 or Yale criteria16 for HDGC (E Fewings PhD, Prof M Tischkowitz MD PhD), test the performance of novel lobular breast cancer- (appendix pp 5–46) was registered. For the purpose of and Early Cancer Institute expanded criteria for the selection of patients for CDH1 comparative analysis, a family with HDGC was defined (Prof R C Fitzgerald MD), testing. We used the collaborative environment of the as any family fulfilling the 2015 clinical criteria for University of Cambridge, European Reference Network on Genetic Tumour Risk HDGC. Phenotypes were collected from probands and Cambridge, UK; Genetics Unit, Biocruces Bizkaia Health Syndromes (ERN GENTURIS) to run a multicentre relatives from families carrying CDH1 germline variants. www.thelancet.com/oncology Vol 24 January 2023 93 Articles Research Institute, Basurto Some individuals contributed more than one phenotype carrier probands and carrier relatives. Because the 2020 University Hospital, Bilbao, to the analysis if they had more than one phenotype. guidelines for HDGC recommend at least 2 years of Bizkaia, Spain (M Garcia-Barcina MD PhD, Information on age of cancer onset and gastric and gastric or breast surveillance, or both, for probands S Merino MD); Sezione di breast cancer histology was collected (appendix pp 2–3). carrying a variant of unknown significance, but not for Medicina Genomica, Diffuse gastric cancer, lobular breast cancer, and gastric relatives, this type of variant was considered non- Dipartimento di Scienze della cancer were considered HDGC-related phenotypes. actionable in this study when assessing the performance Vita e Salute Pubblica, Università Cattolica del Sacro “Other” refers to all phenotypes that differed from of different HDGC clinical criteria. Cuore, Rome, Italy HDGC-related phenotypes, breast cancer of unknown The relative frequency of phenotypes, age of onset, (Prof M Genuardi MD); UOC histotype (herein referred to as “non-specified breast and compliance with HDGC criteria were used in Genetica Medica, Dipartimento cancer” or “breast cancer”), ductal breast cancer, ovarian different comparisons involving families carrying di Scienze di Laboratorio e Infettivologiche, cancer, and colorectal cancer (appendix p 3). different types of CDH1 variants; families bearing Fondazione Policlinico In the context of families, “gastric cancer only” refers to recurrent missense variants of unknown significance or Universitario A Gemelli IRCCS, families enriched in gastric cancer, without breast cancer bearing recurrent truncating PV/LPVs; and families Rome, Italy (Prof M Genuardi); of any type, and could include cancers other than gastric with gastric cancer only, breast cancer only, or gastric Institute for Clinical Genetics, University Hospital Carl Gustav or breast cancer; “breast cancer only” refers to families cancer and breast cancer. Carus and Faculty of Medicine, enriched in breast cancer, without gastric cancer of any The frequency of families carrying PV/LPVs among Technische Universität type, but could include cancers other than gastric or those complying with different HDGC criteria sets was Dresden, Dresden, Germany breast cancer; “gastric cancer and breast cancer” refers to used to evaluate the pick-up rate and performance of (K Hackmann PhD, Prof E Schröck MD); National families with both gastric cancer and breast cancer, and the 2015 and 2020 HDGC criteria, Yale criteria, and the Center for Tumor Diseases, could also include cancers other than gastric or breast lobular breast cancer-expanded criteria. Dresden, Germany: German cancer. “Other cancer” refers to families presenting Cancer Research Center, exclusively cancers other than gastric or breast cancer. Statistical analysis Heidelberg, Germany (K Hackmann, Prof E Schröck); Variant nomenclature was standardised according to the Continuous variables are presented as mean (SD). Helmholtz-Zentrum Dresden- Human Genome Variation Society (HGVS) guidelines.19 Phenotypes without an associated age of onset were Rossendorf, Dresden, Germany Variants were categorised into groups: coding truncating excluded from statistical analyses. Multiple comparisons (K Hackmann, Prof E Schröck); (including nonsense, frameshift, start loss, large deletion, of age of disease onset in carrier probands and their German Cancer Consortium, Dresden, Germany large duplication, and splice-site variants); coding non- relatives and different types of variants (PV/LPV, missense (K Hackmann, Prof E Schröck); truncating (including missense, in-frame insertion, variant of unknown significance, or missense LBV/BV), Max Planck Institute of and in-frame exon skipping variants); non-coding (regu­ were done using a Kruskal-Wallis test with Bonferroni Molecular Cell Biology and latory; including intronic large deletions, 5ʹ untranslated correction. Categorical variables (phenotypes, compliance Genetics, Dresden, Germany (Prof E Schröck); SouthWest region single-nucleotide variants, 3ʹ untranslated region with HDGC criteria, and variant classifi­ cation) are Thames Regional Genetics single-nucleotide variants, and intronic single-nucleotide presented as count (%) and were compared with use of Service, St George’s University variants); and synonymous (referring to single-nucleotide χ² tests. This approach was used to compare variant Hospitals NHS Foundation Trust, London, UK variants unrelated to splicing), according to the Ensembl classes between families fulfilling or not fulfilling clinical (H Hanson MD); Medizinische Variant Effect Predictor (VEP;20 appendix p 47). Variants criteria, and for the analysis of phenotypic distributions Klinik und Poliklinik IV, located until within two bp from the start or end of exons in families carrying different types of variants. Age of Klinikum der Universität and within five bp from the start or end of introns were onset per phenotype was compared with use of an München, Munich, Germany (Prof E Holinski-Feder MD, analysed with NetGene2 version 2.4 splice-site prediction independent-samples Student’s t test. Multivariable V Steinke-Lange MD); software,21 and classified as splice-site variants if a positive logistic regression models using the enter method, Medizinisch Genetisches impact on splicing was predicted. Recurrence was adjusting for phenotypes and age of onset to remove Zentrum, Munich, Germany considered if the same variant occurred in three or more sampling bias in terms of age of onset, were used to (Prof E Holinski-Feder, V Steinke-Lange); Department families. With use of the American College of Medical identify independent predictors for distinct variant types of Human Genetics, Radboud Genetics and Genomics and Association for Molecular (PV/LPVs, and missense variant of unknown significance). University Medical Center, Pathology CDH1 guidelines (version 2), all variants were To evaluate the likelihood of each set of criteria producing Nijmegen, Netherlands classified for their actionability in HDGC as a pathogenic false positive and false negative results, we calculated (Prof M J L Ligtenberg MD, Prof N Hoogerbrugge MD PhD); or likely pathogenic (herein analysed jointly and referred the sensitivity, specificity, positive predictive value, Department of Pathology to in short as PV/LPV), a variant of unknown significance, and negative predictive value, with corresponding (Prof M J L Ligtenberg) and or a likely benign or benign variant (herein analysed 95% CIs, using R (version 4.1.3) with the library epiR Radboud Institute of Molecular jointly and referred to in short as LBV/BV; figure 1A),6 and (version 2.0.50),23 whereas the Youden index (J) used to Life Sciences (Prof M J L Ligtenberg, were confirmed with ClinGen CDH1 Variant Curation estimate the criteria’s discriminating power, and the Prof N Hoogerbrugge), Radboud Expert Panel updates. In case a patient presented corresponding 95% CIs, were computed with the library University Medical Center, two CDH1 germline variants, the most damaging variant ThresholdROC.24 The term “lobular breast cancer- Nijmegen, Netherlands; was considered for genotype–phenotype analysis (appen­ expanded criteria” represents a combination of all the Genetics Service, Biocruces Bizkaia Health Research dix p 48). In the current study, only variants classified as 2020 HDGC criteria plus the new lobular-centred criteria Institute, Cruces University either pathogenic or likely pathogenic were considered herein proposed. To compare performance in terms of Hospital, Cruces-Barakaldo, actionable, as their identification in probands triggers true positive or true negative detection for different sets Bizkaia, Spain cascade testing in relatives and risk reduction measures in of clinical criteria (2015 and 2020 HDGC criteria, Yale 94 www.thelancet.com/oncology Vol 24 January 2023 Articles criteria, and lobular breast cancer-expanded criteria), 182 (21%) families fulfilled the 2015 clinical criteria for (C Martínez-Bouzas PhD, equivalence tests and corresponding 95% CIs were HDGC1 and 672 (79%) did not. The cohort of families M I Tejada PhD); Unidad de Medicina Genómica y Pediatría, computed in R by directly implementing formulae. The fulfilling the 2015 criteria was significantly enriched for Clínica Universidad de Navarra, equivalence limit (δ) was set to 0·12.25 Areas under the PV/LPVs compared with the cohort of families not Programa de Tumores Sólidos, receiver operating characteristic curves (AUCs) from fulfilling the criteria (136 [75%] of 182 vs 40 [6%] of 672; Centro de Investigación different sets of clinical HDGC criteria (2015 and 2020 p<0·0001; figure 1B; appendix p 53). By contrast, mis­sense Médica Aplicada, Instituto de Investigación Sanitaria de HDGC criteria, Yale criteria, and lobular breast cancer- variants of unknown significance were enriched in the Navarra, Pamplona, Navarra, expanded criteria) were compared against each other with cohort of families not fulfilling the criteria (155 [23%] Spain Z tests, using the IBM SPSS Statistics for Windows of 672; these 155 families are part of 159 carrying coding (Prof A Patiño-García PhD); (version 28.0.0). For a 95% CI, a Z score of 1·96 was non-truncating variants in families not fulfilling criteria; Department of Biology and Biotechnology, University of used. Statistical significance was set at p<0·05 for all figure 1B; appendix p 54) compared with the cohort of Pavia, Pavia, Italy statistical tests. families that fulfilled the criteria (14 families carrying (Prof G N Ranzani BSc PhD); coding non-truncating variants [8%] of 182 families GenoMed—Diagnósticos de Role of the funding source fulfilling criteria; p<0·0001; figure 1B; appendix p 54). Medicina Molecular, Lisbon, Portugal (I Silva MSc, The funders of the study had no role in study design, data Early onset diffuse gastric cancer, lobular breast cancer, C Silveira MSc); Molecular collection, data analysis, data interpretation, or writing of and gastric cancer, the classic HDGC-related phenotypes, Genetics Laboratory, Elche the report. were the most frequent among 176 PV/LPV carrier University Hospital, Elche, Spain (J L Soto BSc PhD); families, accounting for 459 (73%) of 631 phenotypes in European Reference Network Results probands and relatives (figure 2A; appendix p 55). on Genetic Tumour Risk 398 different rare CDH1 germline variants were reported HDGC-related phenotypes were repeatedly observed Syndromes (ERN GENTURIS), in 854 families (figure 1A; appendix pp 5–46). From only among families bearing recurrent PV/LPVs, but not Porto, Portugal (Prof S Castedo, Prof F Carneiro, M R Teixeira, 854 carrier probands and 1021 relatives, age of onset was among families bearing recurrent missense variants of Prof C Oliveira); ERN GENTURIS, available for 1495 patients’ phenotypes and ranged from unknown significance (appendix p 56). Bonn, Germany 1–93 years. Altogether, 1971 phenotypes were collected, of Among 176 families carrying PV/LPVs, 362 (92%) of (Prof S Aretz, R Hüneburg, which 460 were gastric and breast cancers with available 392 diffuse gastric cancer or gastric cancer cases occurred I Spier); ERN GENTURIS, Stockholm, Sweden histology (figure 1A; appendix pp 2–3). in families fulfilling the 2015 criteria for HDGC, whereas (S Bajalica-Lagercrantz, Among the 854 families, coding truncating variants only 32 (48%) of 67 cases of lobular breast cancer occurred K Lagerstedt-Robinson); ERN occurred in 184 (22%), missense variants in 316 (37%), in families fulfilling the criteria (figure 2A; appendix p 55). GENTURIS, Barcelona, Spain and other variant types in 354 families (41%; appen­ Among families carrying missense variants of unknown (J Balmaña, J Brunet, G Capellá, C Lázaro); ERN GENTURIS, dix p 47). In terms of variant clinical actionability,6 significance, 32 (65%) of 49 diffuse gastric cancer or Ljubljana, Slovenia (A Blatnik, 176 (21%) families carried PV/LPVs (100 different gastric cancer cases occurred in families fulfilling the M Krajc); ERN GENTURIS, Liège, truncating variants); 169 (20%) families carried missense HDGC criteria, and among families carrying missense Belgium (V Bours); ERN variants of unknown significance (109 different missense LBV/BVs, this proportion was 18 (38%) of 48. Lobular GENTURIS, Paris, France (C Colas); ERN GENTURIS, variants); 147 (17%) families carried missense LBV/BVs breast cancers were extremely rare among families Ghent, Belgium (R de Putter); (29 different mis­sense variants); 182 (21%) families carried carrying missense variants of unknown significance or ERN GENTURIS, Rome, Italy variants of unknown significance other than missense LBV/BVs (figure 2B; appendix pp 55, 57). (Prof M Genuardi); ERN (122 different variants); and 180 (21%) families carried Among families carrying PV/LPVs, 50 (56%) of GENTURIS, Munich, Germany (Prof E Holinski-Feder, LBV/BVs other than missense (38 different variants) 90 non-specified breast cancer cases (mainly in relatives) V Steinke-Lange); ERN (appendix pp 47, 49–51). Probands carrying missense occurred in families fulfilling the criteria (figure 2A; GENTURIS, Nijmegen, variants of unknown significance or missense LBV/BVs, appendix pp 55, 58). By contrast, among families Netherlands and their relatives, were all significantly older than carrying missense variants of unknown significance, (Prof M J L Ligtenberg, Prof N Hoogerbrugge); ERN probands carrying PV/LPVs and their relatives (p<0·05 in only one (<1%) of 219 cases of non-specified breast GENTURIS, Dresden, Germany all comparisons), whereas probands carrying missense cancer occurred in families fulfilling the HDGC criteria, (Prof E Schröck) variants of unknown significance and missense LBV/BVs and among families carrying missense LBV/BVs, this Correspondence to: and their relatives had similar ages of onset (appendix p 52). proportion was three (2%) of 150 cases (figure 2B; Prof Carla Oliveira, Instituto de CDH1 PV/LPVs included both single-nucleotide variants appendix pp 55, 57–58). Investigação e Inovação em Saúde (i3S), University of Porto, and copy-number variants and were distributed across Overall, in families fulfilling HDGC criteria, the Porto 4200-135, Portugal the whole CDH1 locus, targeting all exons, as were all frequency of either gastric cancers altogether or non- carlaol@i3s.up.pt other variant types (appendix pp 49–51). Only 15 PV/LPVs specified breast cancer in relatives was significantly See Online for appendix (11 single-nucleotide variants and four copy-number higher in carriers of PV/LPVs than in carriers of variants) were recurrent, and were found in 81 (46%) of missense variants of unknown significance or carriers 176 families (appendix p 49). The most frequent PV/LPV of LBV/BVs (p<0·0001 in all comparisons; figure 2A, B; single-nucleotide variant, which occurred in 13 families, appendix pp 55, 57–58). was the c.1901C>T (p.Ala634ProfsTer7) splice-site variant, Multivariate logistic regression was used to evaluate and the most frequent copy-number variant, occurring in the likelihood of occurrence of HDGC-related cancers ten families, was the exon 1–2 deletion NC_000016.10:g. in PV/LPV carrier families as compared with their (?_68737292)_(68738411_?)del (appendix p 49). occurrence in families carrying missense variants www.thelancet.com/oncology Vol 24 January 2023 95 Articles of unknown significance (figure 2C). This analysis in four (2%) of 169 families carrying missense variants highlighted positive associations between the classic of unknown significance) had the greatest positive HDGC-related cancers (lobular breast cancer, diffuse association (odds ratio [OR] 12·39 [95% CI 2·66–57·74], gastric cancer, and gastric cancer) and the presence of p=0·0014), followed by diffuse gastric cancer (8·00 PV/LPVs, in contrast to missense variants of unknown [2·18–29·39], p=0·0017), and gastric cancer (7·81 significance. Lobular breast cancer (a phenotype [2·03–29·96], p=0·0027; figure 2C; appendix p 59). present in 47 (27%) of 176 PV/LPV carrier families and Cancers, such as non-specified breast cancer, ovarian A 101 gastric cancer only 245 breast cancer only 484 individuals with familial history of cancer 98 gastric cancer and breast cancer 40 other cancer 832 families 54 gastric cancer only 29 institutions 10 European countries LBV/BV 196 breast cancer only ACMG–AMP 348 individuals CDH1 guidelines 398 variants with a personal VUS 854 families (version 2) history of cancer 3 gastric cancer and breast cancer PV/LPV 854 probands 95 other cancer 14 individuals 1021 relatives with no cancer (1971 phenotypes*) 22 families 8 individuals with unknown cancer 2015 HDGC criteria fulfilled? status Yes No 182 families fulfilling criteria 672 families not fulfilling criteria 182 probands 672 probands 423 relatives 598 relatives 619 phenotypes* 1352 phenotypes* PV/LPV VUS LBV/BV PV/LPV VUS LBV/BV 298 diffuse 34 lobular 2 ovarian 11 colorectal 48 other‡ 21 diffuse 57 lobular 128 ovarian 67 colorectal 191 other‡ gastric breast cancer cancer gastric breast cancer cancer cancer cancer cancer cancer 164 gastric 59 breast 83 gastric 769 breast cancer cancer† cancer cancer† 3 ductal 36 ductal breast breast cancer cancer (Figure 1 continues on next page) 96 www.thelancet.com/oncology Vol 24 January 2023 Articles cancer, colorectal cancer, and all other cancers together, with lobular breast cancer with or without gastric cancer were not positively associated with the presence of (figure 3B). Detailed analysis of family cancer history a PV/LPV (figure 2C). The likelihood of ovarian cancer revealed that these 18 families presented at least occurring in families carrying missense variants of one confirmed case of lobular breast cancer, either unknown significance was significantly higher than isolated before age 55 years (four of 18), in families with in families carrying PV/LPVs (OR 0·12 [95% CI additional non-specified breast cancers and no gastric 0·02–0·71], p=0·019; figure 2C; appendix p 59). In the involvement (seven out of 18), or in families with history multivariate analysis, age of cancer onset did not of non-specified breast cancer and gastric cancer differ with regard to cancer type or between families (seven out of 18). This analysis resulted in a proposal of carrying PV/LPVs and mis­sense variants of unknown three new lobular breast cancer-centred criteria to be significance (figure 2C; appendix p 59). added to the 2020 HDGC criteria, and herein called We next analysed the evolution of HDGC clinical lobular breast cancer-expanded criteria (figure 3B). criteria, considering the above genotype–phenotype For 11 of 29 families who did not meet the 2020 HDGC associations. From 176 PV/LPV carrier families, criteria, representing 6% of 176 CDH1 PV/LPV carrier 40 (23%) did not meet the 2015 criteria. Of these families, no possible criteria could be envisioned due to families, 11 (6%) fulfilled the 2020 HDGC criteria that very scarce or missing clinical information (figure 3B; considers eligible for CDH1 testing isolated cases of appendix p 61). diffuse gastric cancer in people younger than 50 years We next stratified the full PV/LPV carrier cohort into and those families with at least two cases of lobular family subgroups of gastric cancer only, gastric cancer breast cancer in family members younger than 50 years and breast cancer, and breast cancer only, and plotted the (appendix p 60). From the remaining 29 PV/LPV relative phenotype frequency and distribution in each families who did not meet the 2020 HDGC criteria, only family subgroup. This approach allowed evaluation of the four families did not show breast cancer involvement contribution of the lobular breast cancer-centred criteria (figure 3A; appendix p 61). 18 (62%) of 29 were families in the different subgroups (figure 4; appendix pp 62–68).1,2 B Synonymous (n=77) Non-coding (regulatory) (n=12) Synonymous (n=6) Non-coding (regulatory) (n=85) Coding non-truncating (n=8) p<0·0001 Coding non-truncating (n=139) 100 26 14% 80 20 11% 301 Synonymous (n=37) 45% Coding truncating (n=2) Coding truncating (n=6) Synonymous (n=3) Non-coding (regulatory) (n=1) Relative frequency (%) 60 Coding non-truncating (n=159) Coding non-truncating (n=14) Non-coding (regulatory) (n=129) 40 136 75% 331 49% Coding Coding 20 truncating truncating (n=136) (n=40) PV/LPV VUS LBV/BV 40 0 6% Families fulfilling 2015 HDGC criteria (n=182) Families not fulfilling 2015 HDGC criteria (n=672) Figure 1: Flowchart of the genotype–phenotype study and molecular and clinical features of the cohort of families with rare germline CDH1 variants (A) Flowchart of the genotype–phenotype study. (B) Relative frequency of variant groups and clinical classification in families fulfilling or not fulfilling the 2015 HDGC criteria. A χ² test was used to compare the frequency of variant clinical classifications between families fulfilling and not fulfilling the 2015 HDGC criteria (appendix p 53). 14 of 14 coding non-truncating variants of unknown significance were missense variants occurring in families fulfilling the 2015 HDGC criteria. 155 of 159 coding non-truncating variants of unknown significance were missense variants occurring in families not fulfilling the 2015 HDGC criteria. Details on variant type and correspondence with clinical criteria and clinical classification are depicted in appendix (pp 47, 54). ACMG–AMP=American College of Medical Genetics and Genomics and Association for Molecular Pathology. HDGC=hereditary diffuse gastric cancer. LBV/BV=likely benign or benign variant. PV/LPV=pathogenic or likely pathogenic variant. VUS=variant of unknown significance. *Probands could contribute more than one phenotype. †Breast cancer of unknown histotype. ‡Other phenotypes not including gastric cancer, breast cancer, colorectal cancer, or ovarian cancer (detailed in appendix p 3). www.thelancet.com/oncology Vol 24 January 2023 97 Articles A Families carrying truncating PV/LPVs (176 families, 631 phenotypes) 35 Families fulfilling 2015 HDGC criteria Families not fulfilling 2015 HDGC criteria 80 47 51 70 30 years 42 years 42 47 (15) (9) Relative frequency of phenotypes (%) years years years 60 Mean age of disease onset (years) 25 (12) (12) (8) 19 41 years 50 20 8 (4) 2 40 15 30 10 132 33 20 5 114 114 18 19 16 47 10 4 5 7 26 17 15 1 2 3 1 2 3 1 11 0 1 1 8 1 0 P R P R P R P R P R P R P R P R Diffuse gastric Lobular breast Gastric cancer Breast cancer* Ovarian cancer Colorectal Ductal breast Other† cancer cancer cancer cancer Number of cases Overall Total 122 116 36 31 3 151 10 80 2 3 0 12 3 2 16 44 631 Fulfilling criteria 114 114 17 15 2 132 3 47 1 1 0 8 0 1 11 26 492 Not fulfilling criteria 8 2 19 16 1 19 7 33 1 2 0 4 3 1 5 18 139 With available age of onset 114 110 33 29 3 122 9 67 2 1 0 10 3 2 6 31 542 B Families carrying missense VUS (169 families, 406 phenotypes) 35 56 years 80 (17) 30 70 45 43 50 years years years Relative frequency of phenotypes(%) (12) (14) 60 Mean age of disease onset (years) 25 (5) 50 20 40 34 years 15 30 101 117 10 20 5 1 10 33 29 14 24 14 13 5 12 2 5 10 11 7 5 0 1 2 0 P R P R P R P R P R P R P R P R Diffuse gastric Lobular breast Gastric cancer Breast cancer* Ovarian cancer Colorectal Ductal breast Other† cancer cancer cancer cancer Number of cases Overall Total 15 13 5 0 2 19 101 118 24 10 11 14 7 0 33 34 406 Fulfilling criteria 14 13 0 0 0 5 0 1 0 0 0 2 0 0 0 5 40 Not fulfilling criteria 1 0 5 0 2 14 101 117 24 10 11 12 7 0 33 29 366 With available age of onset 15 13 5 0 1 15 93 101 19 5 7 2 5 0 17 16 314 C Prevalence of variant among Number of individuals Odds ratio (95% CI) p value individuals with phenotype without phenotype PV/LPV Missense VUS Age (older vs younger; n=856) ·· ·· 1·00 (0·99–1·01) 0·97 Diffuse gastric cancer 224/252 (89%) 28/252 (11%) 604 8·00 (2·18–29·39) 0·0017 Gastric cancer 125/141 (89%) 16/141 (11%) 715 7·81 (2·03–29·96) 0·0027 Lobular breast cancer 62/67 (93%) 5/67 (7%) 789 12·39 (2·66–57·74) 0·0014 Breast cancer 76/270 (28%) 194/270 (72%) 586 0·39 (0·11–1·39) 0·15 Colorectal cancer 10/19 (53%) 9/19 (47%) 837 1·11 (0·24–5·19) 0·90 Ovarian cancer 3/27 (11%) 24/27 (89%) 829 0·12 (0·02–0·71) 0·019 Other 37/70 (53%) 33/70 (47%) 786 1·12 (0·29–4·24) 0·87 01 20 40 60 Positive association of a phenotype with the presence of a PV/LPV vs a missense VUS 98 www.thelancet.com/oncology Vol 24 January 2023 Articles 174 of 176 families fit into the following subgroups: We next plotted a summary of the 854 families, bearing 91 families with gastric cancer only (representing 52% of variants classified according to the American College of all families carrying PV/LPVs) and 255 phenotypes; Medical Genetics and Genomics and Association for 67 families with gastric cancer and breast cancer Molecular Pathology CDH1 variant classification (version (representing 38% of all families carrying PV/LPVs) and 2), against the 2020 HDGC or the lobular breast cancer- 331 phenotypes, and; 16 families with breast cancer only expanded criteria to graphically represent improvement (representing 9% of all families carrying PV/LPVs) and of clinical criteria in identifying PV/LPV carrier families 43 phenotypes (figure 4; appendix pp 62–68). In families (figure 5A). Further, the performance of different sets of with gastric cancer only, 230 (90%)of 255 phenotypes are HDGC criteria (2015, 2020, Yale, lobular breast cancer- diffuse gastric cancer and gastric cancer. Of the 91 of expanded) was tested (figure 5B). For these tests, we families with gastric cancer only, 89 (98%) fulfilled the considered PV/LPVs as actionable, and variants of 2015 or 2020 HDGC criteria (figure 4A; appendix unknown significance or LBV/BVs as non-actionable pp 62–68).2 In families with gastric cancer and breast (appendix p 70). cancer, 162 (49%) of 331 phenotypes were diffuse gastric The lobular breast cancer-expanded set of criteria cancer and gastric cancer, 48 (15%) were lobular breast showed better performance than the 2020 HDGC criteria cancer, and 78 (24%) were non-specified breast cancer. (AUC 0·92 vs 0·88; Z score 3·54; p=0·0004), as well 56 (84%) of 67 families fulfilled either the 2015 or 2020 as better specificity (0·90 [95% CI 0·88–0·93] vs HDGC criteria, and lobular breast cancer-centred criteria 0·34 [0·30–0·37]), and superiority in true negative contributed additional seven (10%) of 67 families detection (p<0·0001) compared with the Yale criteria16 (figure 4B; appendix pp 62–68). From the 16 families with (specificity 0·34 [95% CI 0·30–0·37]; figure 5B, C; breast cancer only (figure 4C), 12 (7% of 176 families appendix pp 70–72), and equivalent true negative carrying PV/LPVs) presented lobular breast cancer only. detection to that of the 2015 and 2020 HDGC criteria In these 16 families, 19 (44%) of 43 phenotypes were (p=0·059 and p=0·21, respectively; figure 5C). The lobular breast cancer, and 12 (28%) unspecified breast lobular breast cancer-expanded criteria also showed cancer. Two (13%) of 16 families fulfilled either the 2015 better sensitivity (0·94 [95% CI 0·89–0·97]) than the 2015 or 2020 HDGC criteria, and lobular breast cancer-centred HDGC criteria (0·77 [0·70-0·83] and the 2020 HDGC criteria contributed additional 11 (69%) of 16 families criteria (0·84 [0·77–0·89]), with superior true positive (figure 4; appendix pp 62–68). detection (p<0·0001 and p=0·0025, respectively), while We next analysed whether specific PV/LPVs were maintaining high specificity (figure 5C; appendix more frequent or exclusive of subgroups of gastric pp 70–72). cancer only, gastric cancer and breast cancer, or breast cancer only families (appendix pp 64–69). A single Discussion variant (c.1565+2dup) was recurrent and consistently Our study shows that CDH1 PV/LPVs are positively associated with gastric cancer only families, and no associated with HDGC-related phenotypes (lobular variant was particularly associated with gastric cancer breast cancer, diffuse gastric cancer, and gastric cancer), and breast cancer families (appendix pp 64–69). Variants and found no evidence of a positive association of these occurring exclusively in families with breast cancer only phenotypes with CDH1 missense variants of unknown occurred in a single family each (appendix pp 64–69). significance. As CDH1 PV/LPVs occurred often in families with lobular breast cancer who did not meet the 2020 HDGC criteria, this study supports the expansion of HDGC criteria towards inclusion of additional lobular Figure 2: Distribution of phenotypes and cancer average age of onset, in breast cancer-centred criteria. probands and relatives from germline CDH1 variant carrier families, The identification of CDH1 pathogenic or likely and multivariable logistic regression models for the association of cancer phenotypes with PV/LPV and missense-variant of unknown significance pathogenic variants in patients with a relevant cancer Relative frequency of phenotypes, phenotype distribution, number of cases, and history results in the application of clinical recom­ average age of onset in probands and relatives in families carrying PVs or LPVs mendations, including prophylactic removal of target molecularly classified as truncating (A) and in families carrying variants of organs in at-risk asymptomatic carriers. Therefore, it is unknown significance molecularly classified as missense (B) (data further detailed in appendix p 55). The left axis shows the relative frequency (%) of each crucial to know which CDH1 variants are actionable and phenotype out of the total number of phenotypes. Mean age of onset (SD) is cause disease, whether variant classification is accurate, shown for HDGC-associated phenotypes. Detailed data on the number of cases which organs are prone to cancer development in and average age of onset for all phenotypes are provided in the appendix (p 55). carriers, and if there are other histopathological cancer (C) Forest plot showing the results of a multivariable logistic regression model for the association of each phenotype with the presence of a PV/LPV vs a features specific to a particular CDH1 variant, besides missense VUS in 856 patients with available age of disease onset and clinical diffuse gastric cancer and lobular breast cancer. Answering phenotype. Probands could contribute more than one phenotype. these questions will improve clinical guidelines and HDGC=hereditary diffuse gastric cancer. PV/LPV=pathogenic or likely pathogenic variant. P=probands. R=relatives. VUS=variants of unknown significance. *Breast ascertain­ment for genetic testing. Given that HDGC is a cancer of unknown histotype. †Other phenotypes not including gastric cancer, very rare tumour risk syndrome, ERN GENTURIS led breast cancer, colorectal cancer, or ovarian cancer (detailed in appendix p 3). a European multicentre study using a genotype-first www.thelancet.com/oncology Vol 24 January 2023 99 Articles A Exon Variant Phenotype Number of Exon Variant Phenotype Number of families families c.208dup Gastric cancer only EX1del Breast cancer only 1 1* Breast cancer only 1 1* 3 c.220C>T Gastric cancer only EX1_16del Breast cancer only 1 1 c.283C>T Breast cancer only 1 CDH1 7 EX7_8del Breast cancer only 4 c.521dup No cancer copy-number 1* 1 variants 5 c.684C>G Breast cancer only 1* c.696_697del Breast cancer only 10 EX10_11del Gastric cancer and breast cancer 1* 2 (1)† 6 c.832+1G>A Gastric cancer and breast cancer 1* Breast cancer only 7 c.1003C>T Other cancer Gastric cancer and breast cancer 1 Gastric cancer only 8 c.1009_1010del Breast cancer only 1* No familial or personal cancer history c.1141A>T Gastric cancer and breast cancer Personal history of other cancer 9 1* CDH1 c.1488_1494del Gastric cancer and breast cancer single-nucleotide 1* variants Gastric cancer and breast cancer 10 c.1565+1G>T 1 Breast cancer only 2* c.1565+1G>A Gastric cancer and breast cancer 1* 11 c.1612del Breast cancer only 1 c.1795del Breast cancer only 1* 12 Breast cancer only 1* c.1901C>T Gastric cancer and breast cancer 1* c.2062del Gastric cancer and breast cancer 1 13 c.2164+2T>A Gastric cancer and breast cancer 1 c.2195G>A Gastric cancer and breast cancer 1* 14 c.2275G>T Breast cancer only 1* B 2022 Families or individuals fulfilling new lobular breast cancer-centred criteria 2020 Testing criteria Probands, n (%) Families or individuals fulfilling 2020 lobular breast cancer-centred criteria Testing criteria Probands, n (%) 1. ≥2 cases of lobular breast cancer in family members aged <50 years 4 (2%) 3. ≥2 cases of lobular breast cancer 4 (2%) 2. Bilateral lobular breast cancer in individuals aged <70 years 0 in family members aged <50 years Re-thinking Lobular breast 3. ≥2 cases of breast cancer, 1 case of confirmed lobular breast cancer 7 (4%) 8. Bilateral lobular breast cancer 0 lobular breast cancer-centred aged <70 years cancer criteria 4. History of gastric cancer and breast cancer, 1 confirmed lobular breast cancer regardless of age 7 (4%) 176 families carrying PV/LPVs 5. Isolated lobular breast cancer in individuals aged <55 years 4 (2%) Families or individuals not 29 (16%) of 176 Families or individuals not fulfilling 11 (6%) of 176 fulfilling 2020 HDGC criteria lobular breast cancer-expanded criteria 18 (62%) of 29 families resolved with lobular breast cancer-expanded criteria Figure 3: CDH1 PV/LPV carrier families and lobular breast cancer-related criteria (A) CDH1 single-nucleotide and copy-number variants occurring in PV/LPV carrier families not fulfilling the 2020 HDGC criteria. PV/LPVs found in the cohort are shown by CDH1 exon location. The size of each circle is proportional to the number of families found to carry a particular variant. (B) Proposed lobular breast cancer-expanded criteria and proportion of the 176 PV/LPV carrier families explained by these criteria. Lobular breast cancer-centred criteria 3, 4, and 5 are proposed as an addition to the 2020 HDGC criteria. dup=duplication. EX=exon. del=deletion. HDGC=hereditary diffuse gastric cancer. PV/LPV=pathogenic or likely pathogenic variant. *Families resolved by lobular breast cancer-centred criteria. †One of two families is resolved by the lobular breast cancer-centred criteria. 100 www.thelancet.com/oncology Vol 24 January 2023 Articles approach to address these questions, using the largest truncation.11,29 We should not, however, discard the series to date (to our knowledge) of clinically characterised possibility that some purely missense variants can cause rare CDH1 variant carriers and their relatives, who were HDGC. Large databases and increasingly robust clinical tested irrespective of fulfilling clinical criteria. and functional studies are needed to show whether some Our data support germline truncating pathogenic or missense variants of unknown significance could likely pathogenic variants as being major risk factors for cause HDGC.6 HDGC-related cancers, and do not provide evidence of Our results consolidate the spectrum of cancers such an association for missense variants of unknown associated with CDH1 PV/LPVs, which is enriched for significance. These data confirm previous findings,2,6,9 early onset, histologically confirmed, diffuse gastric while challenging the role of most CDH1 missense cancer and lobular breast cancer in probands, and variants of unknown significance as being causative for gastric cancer in relatives, and occurs mainly in families HDGC.8,27,28 Indeed, the limited number of missense fulfilling the 2020 HDGC clinical criteria.2 Furthermore, variants, that were initially deposited in VEP and ClinVar to the best of our knowledge, we showed for the as pathogenic or likely pathogenic variants, have later first time that lobular breast cancer had the greatest been shown to affect splicing, generating premature positive association with CDH1 PV/LPVs, followed by A Families with gastric cancer only (91 families, 255 phenotypes) Families fulfilling 2015 HDGC criteria 35 90 Additional cases included by proposed lobular breast 1 cancer-centred criteria 4 80 30 Relative frequency of phenotypes (%) 1 Mean age of disease onset (years) Additional cases included by 2020 HDGC criteria 41 70 Families not fulfilling criteria 25 40 73 years years 60 (12) (11) 20 50 15 45 40 years 30 10 (14) 79 72 20 5 3 1 10 1 10 1 3 6 0 0 P R P R P R P R P R P R P R P R Diffuse Lobular Gastric Breast Ovarian Colorectal Ductal Other† gastric cancer breast cancer cancer cancer* cancer cancer breast cancer Number of cases Overall Total 84 72 0 0 0 74 0 0 0 1 0 4 0 0 7 13 255 2015 HDGC criteria fulfilled 79 72 NA NA 0 73 NA NA 0 1 0 3 NA NA 6 10 244 Additional cases included by 2020 HDGC criteria 4 0 NA NA 0 0 NA NA 0 0 0 0 NA NA 0 0 4 Additional cases included by lobular breast cancer-centred criteria NA NA NA NA 0 0 NA NA 0 0 0 0 NA NA 0 0 0 No criteria fulfilled 1 0 NA NA 0 1 NA NA 0 0 0 1 NA NA 1 3 7 With available age of onset 80 67 0 0 0 57 0 0 0 1 0 3 0 0 2 8 218 B Families with gastric cancer and breast cancer (67 families, 331 phenotypes) 35 90 49 80 30 Relative frequency of phenotypes (%) 44 50 years Mean age of disease onset (years) 44 47 70 years years (16) 25 years years (14) (9) 41 60 (12) (7) 2 years 9 20 8 50 (4) 7 6 15 8 40 1 1 1 2 30 10 7 3 59 5 47 20 42 1 5 35 1 6 1 1 2 10 16 15 1 16 1 2 4 1 1 1 5 1 1 1 1 1 5 0 3 0 P R P R P R P R P R P R P R P R Diffuse Lobular Gastric Breast Ovarian Colorectal Ductal Other† gastric cancer breast cancer cancer cancer* cancer cancer breast cancer Number of cases Overall Total 38 44 24 24 3 77 9 69 1 1 0 8 2 2 6 23 331 2015 HDGC criteria fulfilled 35 42 16 15 2 59 3 47 1 0 0 5 0 1 5 16 247 Additional cases included by 2020 HDGC criteria 2 1 1 6 0 7 4 8 0 0 0 0 0 0 1 1 31 Additional cases included by lobular breast cancer-centred criteria 0 0 7 3 0 9 1 6 0 0 0 1 1 0 0 1 29 No criteria fulfilled 1 1 0 0 1 2 1 8 0 1 0 2 1 1 0 5 24 With available age of onset 34 43 22 22 3 65 9 59 1 0 0 7 2 2 4 15 288 (Figure 4 continues on next page) www.thelancet.com/oncology Vol 24 January 2023 101 Articles Families fulfilling 2015 HDGC criteria C Families with breast cancer only (16 families, 43 phenotypes) Additional cases included by proposed lobular breast cancer-centred criteria 35 90 Additional cases included by 2020 HDGC criteria 54 80 Relative frequency of phenotypes (%) 30 Families not fulfilling criteria 10 Mean age of disease onset (years) years 45 (10) 70 25 years 3 60 (9) 20 50 15 3 40 4 30 10 8 2 20 5 1 3 1 3 10 0 1 1 1 1 1 0 P R P R P R P R P R P R P R P R Diffuse Lobular Gastric Breast Ovarian Colorectal Ductal Other† gastric cancer breast cancer cancer cancer* cancer cancer breast cancer Number of cases Overall Total 0 0 12 7 0 0 1 11 0 1 0 0 1 0 2 8 43 2015 HDGC criteria fulfilled NA NA 1 0 NA NA 0 0 0 0 0 0 0 0 0 0 1 Additional cases included by 2020 HDGC criteria NA NA 1 3 NA NA 0 0 0 0 0 0 0 0 0 3 7 Additional cases included by lobular breast cancer-centred criteria NA NA 10 4 NA NA 0 8 0 1 0 0 0 0 1 2 26 No criteria fulfilled NA NA 0 0 NA NA 1 3 0 0 0 0 1 0 1 3 9 With available age of onset 0 0 11 7 0 0 0 8 0 0 0 0 1 0 0 8 35 Figure 4: Distribution of phenotypes in probands and relatives from families carrying germline CDH1 PV/LPVs and having gastric cancer only, gastric cancer and breast cancer, and breast cancer only, with reference to HDGC clinical criteria Relative frequency of phenotypes, phenotype distribution, number of cases, and average age of onset of probands and relatives in families carrying PV/LPVs and having gastric cancer only, gastric cancer and breast cancer, or breast cancer only (data further detailed in appendix pp 62–63). The left axis shows the relative frequency (%) of each phenotype out of the total number of phenotypes. Mean age of onset (SD) is shown for HDGC-associated phenotypes. Detailed data on number of cases and average age of onset for other phenotypes are provided in appendix (pp 62–63). Probands could contribute more than one phenotype. HDGC=hereditary diffuse gastric cancer. PV/LPV=pathogenic or likely pathogenic variant. P=probands. R=relatives. VUS=variants of unknown significance. *Breast cancer of unknown histotype. †Other phenotypes not including gastric cancer, breast cancer, colorectal cancer, or ovarian cancer (detailed in appendix p 3). diffuse gastric cancer and gastric cancer. These findings from the USA, which showed occult diffuse gastric contribute to our understanding of the relationship cancer in risk-reduction gastrectomy specimens from between CDH1 PV/LPVs and diffuse gastric cancer and 93·8% of individuals with CDH1 pathogenic or likely lobular breast cancer in probands, and gastric cancer pathogenic variants in presumed HLBC families.17 in relatives.9,12,30 Other cancer types—including ductal 27% of CDH1 PV/LPV carrier families, herein described, breast cancer, ovarian cancer, and colorectal cancer— had at least one histologically confirmed case of lobular were extremely rare in PV/LPV carrier families, showing breast cancer, with or without gastric cancer (sometimes no evidence of association with the presence of PV/LPVs. unconfirmed diffuse gastric cancer), highlighting the This finding supports the exclusion of these cancer need (as reinforced by others)32 to widen the 2020 HDGC types from the spectrum of cancers associated with criteria.2 To maximise the diagnosis of CDH1 PV/LPV CDH1 PV/LPVs.31 Altogether, these genotype–phenotype carriers, we propose three novel lobular breast cancer- data support clinical recommendations for intensive centred criteria to drive CDH1 testing in patients with surveillance and prophylactic surgery of stomach and histologically confirmed lobular breast cancer with or breast only in carriers of CDH1 pathogenic or likely without a history of gastric cancer, representing most pathogenic variants.2 The apparent association between clinical contexts among carriers of CDH1 PV/LPVs who the occurrence of ovarian cancer in families carrying do not meet the 2020 HDGC criteria.2 This widening missense variants of unknown significance merits needs to assure that the criteria remain sensitive enough further research, which is beyond the scope of the to identify as many pathogenic or likely pathogenic current study. variant carriers as possible, with genetic testing being This European cohort had a predominance of families done preferentially in individuals who are most likely to with gastric cancer only and gastric cancer and breast carry such variants. Lerner and colleagues16 followed a cancer over families with breast cancer only, which similar rationale and proposed testing for CDH1 variants represented 9% of all CDH1 PV/LPV carrier families. By in everyone fulfilling the modified 2020 HDGC criteria contrast, similar cohorts in the USA reported breast and the Hereditary Breast and Ovarian Cancer criteria cancer only (presumably HLBC) in more than 35% of (Yale criteria), reporting an increased sensitivity and families carrying CDH1 pathogenic or likely pathogenic simplified system. Although a high sensitivity was variants.9,12 Our observation that a large fraction (38%) of attained, it was accompanied by a substantial decrease in PV/LPV carrier families have both gastric cancer and specificity and Youden index (appendix p 70). Using the breast cancer are further supported by additional data Yale criteria to select patients for CDH1 testing will lead to 102 www.thelancet.com/oncology Vol 24 January 2023 Articles A B 2020 HDGC Lobular breast criteria cancer-expanded 1·0 criteria (n) (n) Criteria fulfilled (n=199) Criteria fulfilled (n=230) 0·8 147 PV/LPV PV/LPV 165 18 22 VUS 0·6 USS 28 VUS Sensitivity 6 30 LBV/BV LBV/BV 37 PV/LPV 0·4 29 7 PV/LPV 11 0·2 329 VUS VUS 323 Criteria not fulfilled (n=624) Criteria not fullfilled (n=655) 0 0 0·2 0·4 0·6 0·8 1·0 1–specificity Source of curve Lobular breast cancer-expanded criteria AUC 0·92 (95% CI 0·90–0·94) 2020 HDGC criteria AUC 0·88 (95% CI 0·85–0·91) 297 LBV/BV LBV/BV 290 2015 HDGC criteria AUC 0·85 (95% CI 0·82–0·89) Yale criteria AUC 0·65 (95% CI 0·62–0·67) Reference C Difference (95% CI) p value Lobular breast cancer-expanded criteria 0·57 (0·52 to 0·61) <0·0001 vs Yale criteria –0·01 (–0·064 to 0·030) 0·48 Lobular breast cancer-expanded criteria –0·02 (–0·049 to 0·011) 0·21 vs 2020 HDGC criteria 0·10 (0·037 to 0·17) 0·0025 Lobular breast cancer-expanded criteria –0·03 (–0·057 to 0·001) 0·059 vs 2015 HDGC criteria 0·17 (0·093 to 0·24) <0·0001 -δ 0 +δ 0·2 0·4 0·6 δ=0·12 Lobular breast Lobular breast True negative detection cancer-expanded cancer-expanded True positive detection criteria inferior criteria superior Figure 5: CDH1 variant classification and distribution according to compliance with 2020 HDGC clinical criteria, and performance evaluation of HDGC clinical criteria sets (A) Alluvial plot (created with the visualisation platform developed by Mauri and colleagues)26 displaying the number of families carrying CDH1 variants and fulfilling 2020 HDGC criteria or lobular breast cancer-expanded criteria. Variants are categorised according to the American College of Medical Genetics and Genomics and Association for Molecular Pathology (version 2). (B) Receiver operating characteristic analysis comparing the AUCs for four different clinical criteria sets (lobular breast cancer-expanded vs 2020 HDGC criteria Z=3·54, AUC difference 0·042 [95% CI 0·02–0·06], p=0·0004). (C) Forest plot representing the lobular breast cancer- expanded clinical criteria equivalence test. The equivalence limit (δ) was set to 0·12. HDGC=hereditary diffuse gastric cancer. LBV/BV=likely benign or benign variant. PV/LPV=pathogenic or likely pathogenic variant. VUS=variants of unknown significance. an excess of patients with breast cancer being tested with that specific CDH1 PV/LPVs predispose individuals a very low pickup rate, representing a waste of resources. differentially to HDGC (families with gastric cancer only Contrarily, the lobular breast cancer-expanded criteria or gastric cancer and breast cancer) or to HLBC (families proposed in this Article showed superior performance to with lobular breast cancer only), as suggested previously, that of previous criteria, and are expected to identify a particularly in the HLBC context.8,17,27,28 These data are larger number of carriers of CDH1 pathogenic or likely supported by Gamble and colleagues, who studied pathogenic variants with high specificity and sensitivity. 31 families tested because of suspicion of HLBC, and Despite the established CDH1-related causality in carrying 19 different CDH1 pathogenic or likely lobular breast cancer, HLBC as an entity remains pathogenic variants.17 These authors conclude that a controversial since its definition in 2020.2,8,12,17 Our proportion of families with HLBC who initially did not genotype–phenotype analysis does not provide evidence report history of gastric cancer exhibited high rates of www.thelancet.com/oncology Vol 24 January 2023 103 Articles occult signet ring cell gastric cancer in a prospective In summary, this is the first multicentre genotype- study. Additionally, our literature review showed that first study to identify associations between rare CDH1 most CDH1 PV/LPVs reported in families with breast germline variants and cancer phenotypes, using the cancer only have been reported in families with HDGC largest dataset of rare CDH1 variant carriers and their elsewhere (appendix pp 73–77).17 Alternatively, HLBC and relatives, to the best of our knowledge, completed under HDGC might differentially occur in pathogenic or likely the umbrella of ERN GENTURIS. The study highlights pathogenic variant carrier families, if they originate from associations between a specific category of rare CDH1 particular geographical regions and due to exposure to germline variants and HDGC specific phenotypes, and particular environmental factors or the presence of proposes novel clinical criteria to trigger CDH1 germline modifier genetic events,33–35 which might favour clinical testing. Altogether, these findings will contribute to expression of diffuse gastric cancer only or lobular breast clarify the context for clinical management in families cancer only, or both (appendix pp 73–77). Additionally, carrying CDH1 germline variants. these modifier genes or environmental factors might Contributors prevent the progression of intramucosal gastric signet JG-P and CO conceptualised the study and did the formal analysis. JG-P, ring cell carcinoma,35 leading to familial presentations of RB-M, SL, AD, and CO contributed to investigation and methodology. RB-M, JG-P, SL, AD, and CO contributed to figure design. JG-P and SL lobular breast cancer only, explaining HLBC in some contributed to classification of variants using American College of instances. In such cases, CDH1-associated HLBC might Medical Genetics and Genomics and Association for Molecular exist as a HDGC-independent clinical entity, according to Pathology CDH1 guidelines. AD, NP, and CLe contributed to the the 2020 guidelines paper.2 In any case, as available data statistical analysis. LS contributed to data management. RM contributed to the Variant Effect Predictor analysis. JJ-M contributed to the CDH1 indicate that CDH1 pathogenic or likely pathogenic gene models. LGa, SCas, SS, HP, SF, FC, CP, MRT, SA, SB-L, JBa, AB, variants predispose individuals to both diffuse gastric PRB, MB, VB, HB, JBr, DC, GC, SCar, CC, KD, RdP, CD, ED-G, CE, cancer and lobular breast cancer, clinical management DGE, DF, EF, RCF, CF, MG-B, MG, LGo, KH, HH, EH-F, RH, MK, should address both organs. KL-R, CLa, MJLL, CM-B, SM, GM, SN, AP-G, GNR, ES, ISi, CS, JLS, ISp, VS-L, GT, M-IT, ERW, MT, NH, and CO contributed to data An additional factor that might constitute a confounder collection. JG-P and CO wrote the original draft of the manuscript. is variant classification. Six CDH1 missense variants All authors contributed to reviewing and editing the manuscript. earlier described as related to HLBC8,17,27,28 were revisited CO contributed to supervision, project administration, and funding and reclassified in this study as LBV/BVs or variants of acquisition. JG-P, RB-M, SL, AD, LGa, and CO directly accessed and verified the underlying raw data in all research articles needed for this unknown significance, in accordance with American submission. All authors had access to the data in the study, critically College of Medical Genetics and Genomics and Association revised the manuscript for important intellectual content, and have for Molecular Pathology CDH1 guidelines (appendix agreed to the submitted version of the results in accordance with the pp 73–77). This example highlights the need of accurate underlying data. variant classification for correct evaluation of disease- Declaration of interests DGE declares fees from Astrazeneca and Recursion. ERW declares grants causal factors, and adequate disease management. Clinical from International Alliance for Cancer Early Detection, for which they classification is challenging, mainly for newly discovered are codirector of the research domain. GNR declares receipt of funding variants with missing clinical data.6,32 This is one of the for study materials, medical writing, and article processing charges from reasons that led us to share as many clinical details of Italian Ministry of Education (GNR). MJLL declares consulting fees (via the Radboud University Medical Center) from Merck Sharp & Dohme the families analysed as possible (appendix pp 5–46). (MSD), AstraZeneca, Lilly, Janssen-Cilag, Illumina, GlaxoSmithKline. A limitation of this study is the fact that Sweden, PRB declares fees from AstraZeneca, MSD, and Bristol Myers Squibb; the Netherlands, and the UK contributed 83 (10%) of and is a scientific committee member for the Geneticancer patients 854 families tested specifically for CDH1 due to clinical association. JBa declares fees from AstraZeneca, Lilly, and Pfizer. SA is a member of APC subVCEP of the InSiGHT/ClinGen Hereditary suspicion of HDGC, as opposed to all other countries that Colorectal Cancer/Polyposis Variant Curation Expert Panel; is an unpaid contributed data irrespective of HDGC clinical suspicion, member of the German Gene Diagnostics Commission and speaker of as requested in the frame of the study design. Additionally, the Centre for Hereditary Tumour Syndromes of the University of Bonn. as a multicentre study, we recognise a possible information RH declares grants from SLA Pharma and Janssen Pharmaceuticals; consulting fees from Janssen and One Two Therapeutics; equipment bias inherent in the use of retrospective data from multiple from Fujifilm; is the head of German Consortium for Familial insti­tutions, with restricted and non-harmonised pheno­ Gastrointestinal Cancer; and is an unpaid advisory board member of the typic descriptions. Furthermore, assumption of CDH1- Lynch Syndrome advocacy Group and the Familial Polyposis Group. positive carrier status in clinically affected relatives, and ES declares grants from NCT/DKTK Master. ES declares honoraria for presentations from AstraZeneca, Georg Thieme Verlag KG, and payment lack of confirmed histology in many cases of gastric cancer for expert testimony from Illumina; is a member of the board of directors and breast cancer among relatives, are limitations. We of Deutsche Gesellschaft für HumanGenetik; an advisor for Dresden- also recognise the possibility of overfitting in the present concept Genome Center; and is board of directors president (paid) for study, and, to our knowledge, a cohort with the same level LNS laboratoire National de Santé. RdP declares support for presentations (via his institution) from MSD and AstraZeneca. of detail is currently not available for cross-validation GC declares to receive funding for study materials, medical writing, purposes. This is the most detailed description of clinical article processing charges from the Spanish Ministry of Science and data from rare CDH1 variant carriers published to date, to Innovation, the Instituto de Salud Carlos III CIBERONC, and the the best of our knowledge, which we believe minimises Government of Catalonia.; consulting fees from VCN Biosciences Synthetic Biologics; is the chair of the Council of the International the study’s limitations. 104 www.thelancet.com/oncology Vol 24 January 2023 Articles Society of Hereditary Gastrointestinal Tumours and the FUREGA 6 Lee K, Krempely K, Roberts ME, et al. Specifications of the Fundació Recerca en Gastroenterologia; and stock in Synthetic Biologics. ACMG/AMP variant curation guidelines for the analysis of CLa declares consulting fees and honoraria from AstraZeneca and MSD, germline CDH1 sequence variants. Hum Mutat 2018; and is a paid advisory board member for Illumina. 39: 1553–68. 7 Fitzgerald RC, Hardwick R, Huntsman D, et al. Hereditary diffuse Data sharing gastric cancer: updated consensus guidelines for clinical Data, aggregated and pseudo-anonymised at the source, on CDH1 management and directions for future research. J Med Genet 2010; germline variant molecular and clinical classification, disease 47: 436–44. spectrum, family history, and age of disease onset, will be made 8 Corso G, Figueiredo J, La Vecchia C, et al. Hereditary lobular breast available through a bulk submission to the ClinVar database and in the cancer with an emphasis on E-cadherin genetic defect. J Med Genet For ClinVar see https://www. appendix (pp 5–46) of this Article. All participating institutions have 2018; 55: 431–41. ncbi.nlm.nih.gov/clinvar/ confirmed their agreement in making the information available 9 Roberts ME, Ranola JMO, Marshall ML, et al. Comparison of and public. CDH1 penetrance estimates in clinically ascertained families vs families ascertained for multiple gastric cancers. JAMA Oncol Acknowledgments 2019; 5: 1325–31. This research was supported by ERN GENTURIS (project ID 739547); 10 Pharoah PD, Guilford P, Caldas C. Incidence of gastric cancer and ERN GENTURIS is partly cofunded by the EU within the framework of breast cancer in CDH1 (E-cadherin) mutation carriers from the Third Health Programme ERN-2016—Framework Partnership hereditary diffuse gastric cancer families. Gastroenterology 2001; Agreement 2017–2021. This work has received funding from the 121: 1348–53. European Union’s Horizon 2020 research and innovation programme 11 Kaurah P, MacMillan A, Boyd N, et al. Founder and recurrent (grant number 779257, Solve-RD: H2020-SC1–2017-Single-Stage-RTD) CDH1 mutations in families with hereditary diffuse gastric cancer. via a PhD studentship awarded to JG-P. The work has also been JAMA 2007; 297: 2360–72. supported by the European Regional Development Fund through the 12 Xicola RM, Li S, Rodriguez N, et al. Clinical features and cancer COMPETE 2020-Operacional Programme for Competitiveness and risk in families with pathogenic CDH1 variants irrespective of Internationalisation, Portugal 2020, Programa Operacional Regional do clinical criteria. J Med Genet 2019; 56: 838–43. Norte under the project “The Porto Comprehensive Cancer Center” 13 Hansford S, Kaurah P, Li-Chang H, et al. Hereditary diffuse gastric with the reference NORTE-01-0145-FEDER-072678—Consórcio PORTO. cancer syndrome: CDH1 mutations and beyond. JAMA Oncol 2015; CCC—Porto.Comprehensive Cancer Center; the Portuguese national 1: 23–32. funds through Fundação para a Ciência e Tecnologia, Ministério da 14 Lowstuter K, Espenschied CR, Sturgeon D, et al. Unexpected Ciência, Tecnologia e Inovação (project references PTDC/BTM- CDH1 mutations identified on multigene panels pose clinical TEC/30164/2017 [3DChrome] and PTDC/BTM-TEC/6706/2020 management challenges. JCO Precis Oncol 2017; 1: 1–12. [LEGOH]); the University of Porto through the Doctoral Programme in 15 Adib E, El Zarif T, Nassar AH, et al. CDH1 germline variants are Biomedicine (Faculty of Medicine; awarded to JG-P), Doctoral enriched in patients with colorectal cancer, gastric cancer, and Programme in Molecular and Cellular Biotechnology Applied to Health breast cancer. Br J Cancer 2022; 126: 797–803. Sciences (ICBAS; awarded to RB-M and SL), Doctoral Programme In 16 Lerner BA, Xicola RM, Rodriguez NJ, Karam R, Llor X. Simplified Molecular And Cell Biology (ICBAS; awarded to AD); the Manchester and more sensitive criteria for identifying individuals with pathogenic CDH1 variants. J Med Genet 2022; published online National Institute for Health and Care Research (NIHR) Biomedical Jan 25. https://doi.org/10.1136/jmedgenet-2021-108169. Research Centre (IS-BRC-1215-20007), providing support to DGE and 17 Gamble LA, Rossi A, Fasaye GA, et al. Association between ERW; the NIHR Cambridge Biomedical Research Centre (BRC-1215- hereditary lobular breast cancer due to CDH1 variants and gastric 20014), providing support to MT; Medical Research Council (RG84369) cancer risk. JAMA Surg 2022; 157: 18–22. and Cancer Research UK (C14478/A12088; RG80420/A21047/A29225), 18 Oliveira C, Bordin MC, Grehan N, et al. Screening E-cadherin in providing support to RF; the Carlos III National Health Institute and gastric cancer families reveals germline mutations only in hereditary Ministerio de Ciencia e Innovación, funded by the European Regional diffuse gastric cancer kindred. Hum Mutat 2002; 19: 510–17. Development Fund (PI19/00553; PID2019-111254RB-I00) and 19 den Dunnen JT, Dalgleish R, Maglott DR, et al. HGVS CIBERONC (CB16/12/00234), providing support to CLa, GC, and JBr; recommendations for the description of sequence variants: 2016 the Government of Catalonia (Pla estratègic de recerca i innovació en update. Hum Mutat 2016; 37: 564–69. salut [PERIS_MedPerCan and URDCat projects], 2017SGR1282 and 20 McLaren W, Gil L, Hunt SE, et al. The ensembl variant effect 2017SGR496); and the Fundação para a Ciência e Tecnologia predictor. Genome Biol 2016; 17: 122. (UIDB/04539/2020, UIDP/04539/2020, LA/P/0058/2020, and POCI-01- 21 Hebsgaard SM, Korning PG, Tolstrup N, Engelbrecht J, Rouzé P, 0145-FEDER-022184) providing support to CE. We are grateful to the Brunak S. Splice site prediction in Arabidopsis thaliana pre-mRNA ERN GENTURIS National Coordinators for promoting data collection by combining local and global sequence information. at the country level for the current project; Rui Magalhães (ICBAS, Nucleic Acids Res 1996; 24: 3439–52. University of Porto, and Centro Hospitalar Universitário do Porto) for 23 Stevenson M, Sergeant E. EpiR: tools for the analysis of consultancy in statistics; and Víctor Treviño (Escuela de Medicina y epidemiological data. https://cran.r-project.org/web/packages/epiR/ Ciencias de la Salud del Tecnológico de Monterrey, Monterrey, Mexico) index.html (accessed Aug 25, 2022). for the R script for gene models. 24 Perez-Jaume S, Skaltsa K, Pallarès N, Carrasco JL. ThresholdROC: optimum threshold estimation tools for continuous diagnostic tests References in R. J Stat Softw 2017; 82: 1–21. 1 van der Post RS, Vogelaar IP, Carneiro F, et al. Hereditary diffuse 25 Walker E, Nowacki AS. Understanding equivalence and gastric cancer: updated clinical guidelines with an emphasis noninferiority testing. J Gen Intern Med 2011; 26: 192–96. on germline CDH1 mutation carriers. J Med Genet 2015; 52: 361–74. 26 Mauri M, Elli T, Caviglia G, Uboldi G, Azzi M. RAWGraphs: a visualisation platform to create open outputs. 12th Biannual 2 Blair VR, McLeod M, Carneiro F, et al. Hereditary diffuse gastric Conference on Italian SIGCHI Chapter; Sept 18–20, 2017 (abstr 28). cancer: updated clinical practice guidelines. Lancet Oncol 2020; 21: e386–97. 27 Girardi A, Magnoni F, Vicini E, et al. CDH1 germline mutations in families with hereditary lobular breast cancer. 3 Caldas C, Carneiro F, Lynch HT, et al. Familial gastric cancer: Eur J Cancer Prev 2022; 31: 274–78. overview and guidelines for management. J Med Genet 1999; 36: 873–80. 28 Corso G, Intra M, Trentin C, Veronesi P, Galimberti V. CDH1 germline mutations and hereditary lobular breast cancer. 4 Guilford P, Hopkins J, Harraway J, et al. E-cadherin germline Fam Cancer 2016; 15: 215–19. mutations in familial gastric cancer. Nature 1998; 392: 402–05. 29 Barbosa-Matos R, Leal Silva R, Garrido L, et al. The CDH1 c.1901C>T 5 Oliveira C, Senz J, Kaurah P, et al. Germline CDH1 deletions in variant: a founder variant in the Portuguese population with severe hereditary diffuse gastric cancer families. Hum Mol Genet 2009; impact in mRNA splicing. Cancers (Basel) 2021; 13: 4464. 18: 1545–55. www.thelancet.com/oncology Vol 24 January 2023 105 Articles 30 Lo W, Zhu B, Sabesan A, et al. Associations of CDH1 germline 34 Tanikawa C, Kamatani Y, Toyoshima O, et al. Genome-wide variant location and cancer phenotype in families with hereditary association study identifies gastric cancer susceptibility loci at diffuse gastric cancer (HDGC). J Med Genet 2019; 56: 370–79. 12q24.11-12 and 20q11.21. Cancer Sci 2018; 109: 4015–24. 31 Massari G, Magnoni F, Favia G, et al. Frequency of CDH1 35 Ellison-Loschmann L, Sporle A, Corbin M, et al. Risk of stomach germline mutations in non-gastric cancers. Cancers (Basel) 2021; cancer in Aotearoa/New Zealand: a Māori population based 13: 2321. case-control study. PLoS One 2017; 12: e0181581. 32 Katona BW, Clark DF, Domchek SM. CDH1 on multigene panel testing: look before you leap. J Natl Cancer Inst 2020; 112: 330–34. 33 Zhang H, Ahearn TU, Lecarpentier J, et al. Genome-wide association study identifies 32 novel breast cancer susceptibility loci from overall and subtype-specific analyses. Nat Genet 2020; 52: 572–81. 106 www.thelancet.com/oncology Vol 24 January 2023