CDH1 mutations in gastric cancers are not influenced by family history

Background: CDH1 mutation is the most frequent genetic alteration in hereditary diffuse gastric cancer (GC) and early onset diffuse GC patients. However, the incidence of CDH1 mutations in sporadic GC with or without family history has not been studied. Methods: This retrospective study includes a total of 993 Korean patients with primary advanced GC who underwent surgery and received palliative chemotherapy . Targeted deep sequencing was performed in all cases and family history of GC was searched with survival analysis. Results: We found CDH1 alterations in 146 of 993 patients (14.7%) and 8 were germline (0.8%). Out of 146 patients with CDH1 mutations, 25 (17.1%) had a family history of GC in one of their first relatives, and 12 patients (8.2%) were diagnosed with familial GC (FGC). All cases with FGC were diffuse type by Lauren classification, and only one harbored a previously reported germline mutation of CDH1 (c.2638G>A) and the remaining 11 harbored known somatic CDH1 mutations. Among all patients with CDH1 mutation, there was no significant survival difference between patients with family history or FGC. In the 847 patients without CDH1 mutation, 189 (22.3 %) had a family history of GC and 92 patients (10.9%) were FGC. CDH1 mutations were more frequent in patients with early onset (<45 years) GC (45.5%) compared with patients with late onset GC (10.9%) (p = 0.001), but were not significantly associated with the family history of GC (p > 0.05). Conclusions: CDH1 mutations are mostly somatic and typically are not associated with family history.

The recent advent of and use of multigene cancer panel testing using next-generation sequencing (NGS) has dramatically improved testing by enabling simultaneous evaluation of many patients for multiple high-risk cancer syndromes as well as moderate-risk cancerassociated genes such as CDH1 [ 13]. While this strategy has resulted in significant increases in identification of individuals at risk, it has also led to clinical conundrums regarding management of nonsyndromic carriers of an unexpected or unanticipated genetic mutation [14], such as patients with CDH1 mutation [15]. With NGS, somatic mutations of CDH1 are found in 9%~36% of GC cases [16][17][18][19][20][21][22] and are highly prevalent in sporadic early-onset diffuse-type GC (EODGC) (53.2% of cases) [17]. Somatic mutations of CDH1 have been suggested as a poor prognostic marker in diffuse-type GC [22,23].
Although a few studies have investigated the frequency and impact of CDH1 mutations in HDGC and EODGC [9,10,17,24] the incidence of CDH1 mutations in sporadic GC with or without family history has not been studied. Although familial aggregation is found in about 10% of GC cases, the prevalence of CDH1 alterations in patients with familial aggregation is unclear [25].
In the present study, we studied CDH1 alterations with deep targeted sequencing and investigated the association with family history in 993 GC patients in a GC-prevalent country.

Patients
This study involved retrospective analysis of patients with GC at Samsung Medical Center in Seoul, South Korea. All patients were diagnosed with GC by pathologic examination and provided written informed consent. The 993 patients included 647 male and 346 female patients, and the patient age ranged from 18 to 80 years (mean 60.8). By Lauren classification, diffuse type was found in 463 cases. Patients < 45 years old were defined as EOGC, and 110 cases in the total patient group were classified as EOCG. Among these 110 cases, 81 were diffuse type and were diagnosed as EODGC.
Patients who had family history of GC in first-or second-degree relatives were considered as GC with family history. Patients with family history were defined as FGC if they met the following criteria: (1) GC in 2 or more first/second-degree relatives, with at least one diagnosis before age 50 or (2) GC in 3 or more first/second-degree relatives, independent of age. The remaining patients were classified as sporadic GC, as previously described [26].

Variant detection
Sequence reads were mapped to the human genome (hg19) using Burrows-Wheeler Aligner (BWA) [27]. Duplicate read removal was performed using Picard and SAMtools [28]. Local alignment was optimized using the Genome Analysis Toolkit (GATK) [29]. To detect single nucleotide variants (SNVs), we integrated the results of three variant callers to increase sensitivity [30][31][32]. For insertion and deletions (indels), Pindel was used [33]. Copy number variations were calculated for targeted regions by dividing read depth per exon by estimated normal reads per exon using an in-house reference.

A n a l y s e s o f C D H 1 v a r i a n t s
Lists of CDH1 variants were assembled from the literature and a search using the ClinVar [34], Human Gene Mutation Database (HGMD) [35], the Genome Aggregation Database  (Table 1). Patients < 45 years were defined as EOGC; a total of 50 cases was classified as EOGC, and 47 of these cases were EODGC. CDH1 mutations were more frequent in patients with EOGC (45.5%) compared with patients with late onset GC (10.9%) (p = < 0.001) ( Table 2). To rule out the possibility of deamination effects of formalin, the nucleotide changes were analyzed, and the prevalence of C > G alterations (2.74%) was not significantly frequent With a cut-off value for variant allele frequency (VAF) of > 50% [36], germline mutations were identified in 8 patients (0.8%) and 2 of them were novel: c.1424T > A in exon 10 and c.1839_1841delCAT in exon 12. The two novel CDH1 mutations were validated as germline mutations by Sanger sequencing of paired normal tissue. Remaining 6 mutations have been reported previously as uncertain significance. The precise mutation profiles of germline CDH1 alteration with clinicopathologic characteristics are described in Table 3.

M u t a t i o n s o f C D H 1 i n p a t i e n t s w i t h a n d w i t h o u t f a m i l y h i s t o r y
In the 146 patients with CDH1 mutations, 25 (17.2%) had a familial history of GC in one of their first relatives, and 12 patients (8.2%) were diagnosed with FGC with CDH1 mutation ( Unexpectedly, we found germline mutations of CDH1 in 6 patients without a family history of GC. Among all patients with CDH1 mutation, there was no significant survival difference between patients with family history or FGC (Fig. 4). In the 847 patients without CDH1 mutation, 189 (22.3%) had a family history of GC in one of their first relatives, and 92 patients (10.9%) were diagnosed as FGC (Table 2). We could not find any statistical significance in CDH1 mutations with family history of GC (p > 0.05).

Discussion
Although familial aggregation is found in about 10% of GC cases [25], the incidence of CDH1 alterations in patients with or without family history of GC is unclear. We explored CDH1 alterations with NGS and investigated the association with family history in 993 GC patients. We found that occurrence of CDH1 mutation was not significantly associated with family history.
After the first report of germline CDH1 mutation in GC [37], various types of CDH1 mutations have been reported in HDGC or diffuse type GC by polymerase chain reaction and sequencing [3][4][5][6][7] aided with multiplex ligation-dependent probe amplification [8][9][10][11] or high-resolution melting analyses [30] from patients with FGC, HDGC, or EOGC. CDH1 germline mutations were identified in 10% of FGC cases [10,11], 0 ~ 19% of HDGC cases [3][4][5]8] and 0 ~ 8.9% [5][6][7]9] of EOGC cases [6], as well as in 1.7% of patients with sporadic GC [12]. With NGS, somatic mutations of CDH1 are found in 9%~36% of GC cases [16][17][18][19][20][21][22] and are highly prevalent in sporadic EODGC (53.2% of cases). Somatic CDH1 mutation has also been suggested as a poor prognostic marker in diffuse-type GC [17,22,23]. The previously reported germline and somatic CDH1 mutations detected in patients with gastric cancer by various technologies are summarized in Table 4. Although a few studies have investigated the frequency and impact of CDH1 mutations in HDGC and EODGC [9,10,17,24], the incidence of CDH1 mutations in sporadic GC patients with or without family history is unclear [25]. To the best of our knowledge, this is the first study exploring CDH1 mutations in association with family history. Our results showed that CDH1 mutations are not influenced by family history. in sporadic GC patients [12]. In previous studies in EODGC patients, CDH1 germline mutation rates were higher in Korean (8%, 2/25) [9] and Italian populations (7.2%, 19/264) [6]. In a meta-analyses study, GC occurring in a high-prevalence area harbored less frequent CDH1 germline mutations than that in a low-incidence area [2]. Based on these observations, our low incidence of CDH1 germline mutations may be due to patient age (more frequent CDH1 germline mutations in younger patients) and because the present study was performed in a GC-prevalent area.
In GC-prevalent areas, the ratio of missense mutations was higher than that of nonmissense (deletion, insertion, truncating, and nonsense) mutations [2]. In addition, nonmissense mutations were reported to be likely pathogenic compared with missense mutations. We also found that the frequency of missense mutation was higher (67.8%) compared with that of non-missense mutation, and this result is similar to the previous analysis. A recent comprehensive genomic study demonstrated that RhoGAP domaincontaining fusions or PPAPDC1A fusions, which are mostly somatic mutations, are common in GC of diffuse-type [38]. In addition, germline mutations in genes such as CTNNA1 were observed at a similar frequency as CDH1 germline mutation [39]. These findings suggest that GC might be frequently caused by somatic mutations rather than germline alterations of CDH1 in GC-prevalent areas such as Korea, as previously suggested [2].
Unlike previous studies that focused exclusively on HDGC patients, we compared the distribution of CDH1 alterations in patients with and without family history of GC. As a result, most CDH1 alterations in both groups were somatic mutations. Thus, CDH1 somatic mutation appears to play a more important role than germline mutation regardless of family history; The limitation of this study is that the patient's family history was investigated only through electronic medical records. Therefore, the age of onset and histological information of GC in the patient's family could not be accurately identified. In addition, there is a possibility of missed records. Further studies might be required to analyze the family information about GC more in detail with pedigree analysis.

Conclusions
In conclusion, we found 146 CDH1 mutations in 14.7% of GC patients. CDH1 mutations were not associated with family history. Rare germline mutations in sporadic and familial GC patients suggest that somatic CDH1 mutation may play an important role in pathogenesis of GC in GC-prevalent areas.

Availability of data and materials
All data generated or analyzed during this study are included in this published article.

Ethics approval and consent to participate
The Institutional Review Board of Samsung Medical Center approved this study and waived informed consent.

Consent for publication
Not applicable Figure 1 Flow charts of CDH1 variants filtered by public databases