Introduction CDH1 predisposes primarily to diffuse gastric cancer (DGC). Multiple DGC cases in a family, DGC at a young age in an individual or the combination of DGC andlobular breast cancer (LBC) in an individual or a family define the hereditary DGC syndrome (HDGC), and testing for germline CDH1 mutations is warranted in HDGC.
Methods and results We report all index cases from Ile-de-France in which a germline CDH1 mutation has been identified. Out of 18 cases, 7 do not fulfil the HDGC-defining criteria. Three of them are women who presented initially with bilateral LBC below age 50, without personal or family history of DGC, and who subsequently developed symptomatic DGC.
Discussion Our series of CDH1 mutation carriers is the largest to date and demonstrates that LBC might be the first manifestation of HDGC. A personal or family history of multiple LBCs at a young age, even without DGC, should prompt CDH1 mutation screening. It is paramount to identify mutation carriers early, so that they can benefit from prophylactic gastrectomy before they develop symptomatic, highly lethal DGC. We recommend a revision of the HDGC-defining criteria and propose for consideration the name ‘Hereditary Diffuse Gastric and Lobular Breast Cancer’ instead of HDGC.
- Cancer: gastric
- Cancer: breast
- Clinical genetics
Statistics from Altmetric.com
Twin studies, as well as case–control studies comparing gastric cancer (GC) incidence in relatives of patients with GC and in controls, provide evidence that inherited genetic factors contribute to the risk of developing the disease.1 ,2 In 1998, researchers from New Zealand found germline CDH1 mutations in three Maori families with multiple cases of GC of the diffuse type (diffuse GC, DGC).3 CDH1 is located on chromosome 16q22 and codes for the adhesion protein E-cadherin, and gene alterations are inherited dominantly. Experts subsequently defined the hereditary DGC (HDGC) syndrome as any family meeting the following criteria: multiple cases of DGC in first- or second-degree relatives, with one case diagnosed before age 50 if there were only two cases in the family, but regardless of age if there were three cases or more, and they recommended systematic CDH1 testing for HDGC families.4 It has since been shown that CDH1 mutations are also found in individuals with DGC at a young age but no family history of the disease (sporadic DGC).5 Additionally, research groups have reported an excess of lobular breast cancer (LBC) cases in CDH1 families.6 ,7 These findings led to the addition in 2010 of the two following situations to the HDGC criteria: (i) sporadic DGC in an individual below age 40 and (ii) personal or family history of DGC and LBC, with at least one case diagnosed before age 50.8
We report herein the largest series to date of index cases with a germline CDH1 mutation. We advocate a revision of the HDGC criteria, as a substantial proportion of our cases carried a mutation even though they did not fulfil them. We also highlight the importance of identifying carriers early in order to offer them prophylactic gastrectomy before they develop symptomatic, highly lethal DGC. Given the high number of LBC associated with germline CDH1 mutations, we finally propose for consideration the name ‘Hereditary Diffuse Gastric and Lobular Breast Cancer’, instead of HDGC.
We studied retrospectively all index cases from the Ile-de-France region in which testing for a germline CDH1 mutation had been performed between 2006 and 2012. Ile-de-France has a population of twelve million (about a fifth of the total French population) and includes the city of Paris and seven surrounding counties (départements). There is a network of approved, specialised cancer genetics centres in the region, and the total number of medical consultations carried out there annually is close to 9000. All centres took part in this study, and no genetic counselling or testing for CDH1 was carried out outside of these centres. Index cases were seen at least twice by a physician with expertise in cancer genetics, and they signed an informed consent form for gene analysis.
Screening for CDH1 mutations was performed after the indication for testing had been validated by a multidisciplinary team, and it was not a prerequisite for cases to meet the recommended testing criteria in place at the time. The search for point mutations and large rearrangements in CDH1 was done in three different accredited laboratories using Sanger sequencing of exons and their flanking regions and multiplex ligation probe-dependent amplification (MLPA) or qPCR on genomic DNA extracted from peripheral blood leucocytes. Detailed protocols are available on request. Mutations were considered pathogenic if they had been reported and reliably identified as such in the literature, or, for never-reported mutations, if they affected RNA splicing or led to a premature stop codon. Each participating centre first surveyed the medical records for its own patients, and data were anonymised and collected for review.
Testing for CDH1 germline mutations was carried out in 165 unrelated Ile-de-France index cases (table 1). Eighteen (11%) were mutation carriers (table 2). Of these, 11 had a personal or family history that fulfilled the 2010 HDGC criteria: one came from a family with three cases of DGC; five came from families with two cases of DGC, one of them below age 50; three had sporadic DGC below age 40; and two had a personal or family history of LBC and DGC, with one of the cancers diagnosed before age 50. The seven remaining cases did not meet the HDGC criteria: three had sporadic bilateral LBC below age 50; three came from families with two cases of DGC after age 50; and one had sporadic DGC after age 40. There were 16 different CDH1 gene mutations in the 18 index cases. Fourteen were point mutations or small insertions/deletions detected by sequencing, and two were large deletions detected by MLPA/qPCR. Nine of the 16 gene mutations have never been reported in the literature.
Three of our cases, IC1, IC2 and VJ3, were women with a personal history that consisted initially of bilateral LBC below age 50. Family history was negative for DGC in first- and second-degree relatives, and they therefore did not meet the HDGC criteria at this stage. They all tested negative for mutations in BRCA1 and BRCA2. Cases IC1 and IC2 were only tested for CDH1 after subsequently developing symptomatic DGC and, as a result, meeting these criteria. Conversely, VJ3 had CDH1 testing shortly after the physician in charge had received the negative BRCA1/2 results. Upper gastrointestinal endoscopy with multiple biopsies followed, but no foci of cancer were found. She was nevertheless offered prophylactic gastrectomy, and the pathologist diagnosed invasive pT1N0 DGC.
To our knowledge, our series of 18 index cases with a germline CDH1 mutation is the largest reported to date. Oliveira et al9 did report 61 CDH1 families, but they came from a pooled analysis of 15 previously published studies. Our cases, added to those reported before, will improve awareness within the cancer genetics community of phenotypes associated with germline CDH1 mutations.
We describe three female cases who presented with bilateral LBC below age 50 but no personal or family history of DGC. Only two other groups have reported CDH1 mutations in cases or families with multiple, histologically proven early-onset LBC, but no DGC. Masciari et al10 described the case of a woman carrier with LBC at age 42, whose mother had been diagnosed with the same condition at age 28. Xie et al11 reported a CDH1 family with bilateral LBC in a woman at age 42 and unilateral LBC in her mother at age 62. Collectively, these cases demonstrate that early-onset LBC might be the first manifestation of HDGC and that a personal or family history of multiple LBCs at a young age, even in the absence of DGC, should prompt cancer geneticists to test for CDH1 germline mutations.
The penetrance of CDH1 mutations for DGC is approaching 100%, as foci of cancerous signet-ring cells (the hallmark of DGC), often spreading beyond the basement membrane, are almost always observed on prophylactic gastrectomy specimens, even in young adults.12–14 Of note, these foci are most often undetectable at screening endoscopy. The prognosis of GC, regardless of histology, is dismal.15 It is therefore paramount to identify mutation carriers before they develop lethal, symptomatic DGC. In our case series, the earlier identification of a CDH1 mutation in IC1 and IC2 would have allowed for prophylactic surgery (both ended up developing symptomatic DGC), while VJ3 did benefit from early CDH1 testing and subsequently from prophylactic gastrectomy.
According to the 2010 HDGC criteria, the presence of at least one case of DGC is mandatory for CDH1 genetic testing.8 We consider that these criteria are too strict. Indeed, solid evidence has since emerged of CDH1 cases and families with multiple early-onset LBCs, but no DGC. The potentially dire consequences for a carrier of a missed germline mutation in CDH1 should be emphasised, as illustrated by our study. As a result, we believe that these testing criteria should be expanded. We would therefore recommend adding ‘personal or family history of two histologically proven LBCs below age 50, after exclusion of a germline mutation in BRCA1 and BRCA2’ to the existing criteria. Given that LBC is now an integral part of HDGC, we also propose for consideration the name ‘Hereditary Diffuse Gastric and Lobular Breast Cancer’, instead of HDGC, to define this cancer susceptibility syndrome.
We agree that our suggestion to expand the CDH1 testing criteria is based on a small number of cases, three included here and two published previously by others.10 ,11 They nevertheless likely represent a tiny fraction of actual numbers, since patients with a personal or family history of multiple LBCs at a young age but no DGC are very rarely, if ever, tested for CDH1 mutations. In our study, only five such cases were tested, and three (60%) were mutation carriers, suggesting that a substantial proportion indeed carries a germline alteration in CDH1. No research group has studied the prevalence of CDH1 mutations in patients meeting our proposed testing criterion. Masciari et al10 and Schrader et al16 found mutations or potentially causal variants in, respectively, 1/23 and 4/318 women with LBC either before age 45 or regardless of age if there was a family history of any type of breast cancer. However, it is not known how many women actually had a family history of breast cancer of the lobular type, as histology in relatives was not specified and as they did not specifically recruit patients with multiple LBCs. Furthermore, BRCA1/2 mutations had been excluded in most (77%), but not all cases reported by Schrader et al.16
We have based our proposed 50-year-old age cut-off for two LBCs on the observation that in our study, no family with multiple LBCs and a CDH1 mutation had cancer after age 50. In an online reply to the Schrader's paper, Newman et al did report a CDH1 family with four cases of LBC ages 49, 50, 51 and 53.16 It does not exactly meet our proposed testing criterion, as only one case was diagnosed below age 50. The four cases were nevertheless all very close to 50, and as a result, we do not think this single family warrants a change in or a suppression of the age cut-off. We would trust the team in charge of a future, potential family similar to this one, to use their clinical judgment and conclude that it is suggestive of a CDH1 mutation.
Among cases with two DGCs (≥ age 50) in the family, 3/7 were carriers. Only once before has a mutation been reported in such a family.6 Whether many CDH1 mutations are missed because of the 50-year-old DGC age cut-off requires further exploration, but we think that the test might be justified if written evidence of diffuse histology is available for both GC s. The high proportion of carriers in this population contrasts with what we have observed among individuals with three DGCs in the family, regardless of age (one carrier among 16 index cases). The difference is surprising and could be due to chance, or more likely, to the fact that, in families with three DGCs, evidence of diffuse histology was rarely available in relatives, and these relatives could actually have had intestinal-type GC or another cancer of the digestive tract. Readers should be reminded that the 2010 CDH1 testing criteria only require confirmation of diffuse histology for one relative with GC.
Fifty-seven index cases had CDH1 testing even though they had a personal history of GC or BC of non-diffuse, non-lobular or unknown histology. Unsurprisingly, no mutations were found. Finally, we found only one mutation among individuals with sporadic DGC after age 40, out of 14 tested. This low proportion of positive tests was expected.17
We have stated that carriers of CDH1 mutations must be identified early and benefit from prophylactic gastrectomy as their risk of developing DGC is almost 100%. Strictly speaking, it is the histopathological penetrance that approaches 100%. The clinical penetrance is somewhat lower ranging from 40% to 83%, suggesting that some DGCs remain indolent for decades.6 ,18 However, even the most conservative penetrance estimates are high for a disease that kills the majority of affected patients and equals or exceeds, for example, ovarian cancer penetrance for individuals with BRCA1/2 mutations in which prophylactic salpingo-oophorectomy is strongly recommended.19 ,20
As always in cancer genetics, we must not neglect the psychological consequences of germline mutation testing and of the prophylactic surgery that may follow. If new CDH1 testing criteria were to be adopted, all tested patients would then require easy access to psychologists or psychiatrists knowledgeable in the field.
We pay tribute to Liliane Demange whose contribution to this work was substantial, but who sadly passed away before the manuscript was finalised.
DM and ER have contributed equally to this work.
Collaborator Liliane Demange.
Contributors PRB, DM, ER and OC designed the study with critical input from the other authors. PRB, DM, ER and OC gathered and analysed data collected previously by ER, ADP, BB, EF, CC, MW, SG, DMo, CT and OC. PRB, DM, ER, ADP, BB, CN, EF, CC, MW, SG, AS, PLP, DMo, MDM, VB, SD and OC saw the patients in the clinic and prescribed genetic testing. ER, CC, FC and MB carried out genetic testing in the laboratory. PRB wrote the manuscript with support from DM, ER and OC. All authors were involved in the final interpretation of the global data, revised the manuscript critically for important intellectual content and approved the final version. PRB is responsible for the overall content as guarantor.
Funding PR Benusiglio has received support from the Swiss National Foundation, Grant Number: 139844.
Competing interests None.
Ethics approval All patients signed an informed consent form before routine genetic testing was performed, and the form was in accordance with French law and National guidelines edited by Health Authorities.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.