Endoscopic screening for identification of signet ring cell gastric cancer foci in carriers of germline pathogenic variants in CDH1

The discrepancy between published literature reporting DGC in 33–42% of patients with CDH1 PVs by the age of 80 but gastrectomy specimens detecting early stage SRC DGC in up to 97% of carriers needs to be reconciled. Endoscopic diagnosis of SRC is challenging due to the lack of a visible dominant lesion. Prior studies have shown that in order to have a 90% detection rate of SRC, thousands of endoscopic biopsies are required, which is not feasible in clinical practice [16]. Multiple surveillance protocols with different numbers of targeted and random biopsies have been described with limited detection rates from less 12–63% [9, 10, 17‐19]. We altered our endoscopic surveillance approach to increase the number of biopsies to enhance the detection of preoperative SRC and describe the sensitivity of our novel protocol.

Increasing the number of random biopsies to 77 and 11 locations in the stomach resulted in a preoperative accuracy of detection of SRC to 81%. This rate is higher than prior studies that have used the IGCLC guideline protocol for preoperative detection of SRC (12–63%).^9,10,17–19 Our findings that more biopsies are associated with greater SRC detection align with Curtin et al., who recently demonstrated that evolving from the IGCLC guideline protocol to 88 random biopsies increased the preoperative detection of SRC per endoscopy from 15–36% [20]. Importantly, in our center, increasing the number of biopsies was not associated with endoscopic complications and supports the feasibility and safety of our protocol. A variety of other imaging techniques and methods such as endoscopic ultrasound, autofluorescence imaging, confocal endomicroscopy, and chromoendoscopy have been assessed and found to be of no utility for detection of SRC [10, 17, 20‐24].

Although SRC were distributed throughout the stomach, they were most found in the proximal stomach, as previously described by authors applying the IGCLC and the Bethesda protocols [17, 20]. This finding suggests that particular attention should be paid to target this area when performing EGD surveillance in this patient population.

Our findings reinforce the concept that more endoscopic biopsies increase the sensitivity of SRC detection but should not discount the utility of targeted biopsies of mucosal abnormalities such as pale patches, nodules, and ulcers [17]. In our study we found that 11% (4/36) of pale patches biopsied on preoperative EGD had SRC foci. Three of the patches that harbored SRC were found on the first endoscopy, and the other one on the second surveillance EGD. The yield of pale patches for SRC detection in the literature varies. Our findings are similar to a recent study reporting 12% of pale areas yielded SRC foci [7]. Other studies have reported SRC detection in pale mucosal areas (specificity 91.7% and sensitivity 57.1%) [6]. There is mixed data about the role of mucosal findings and their association with the presence of SRC and their subsequent development into advanced lesions. Some studies have not found any association between other endoscopic gastric findings, such as polyps, ulcers and erosion, and SRC detection [6, 7], while others have shown that thickened rugal folds, poor distensibility, disturbed vascular patterns, focal ulceration and erosions could identify more advanced SRC DGC [12]. For instance, in a study of 270 CDH1 PV carries, 120 patients continued endoscopic surveillance and 2 developed clean-based ulcers that harbored SRC on their initial and third EGD. On gastrectomy pT3N0M0 DGC was confirmed [12].

Nearly half of our patients opted to have EGD surveillance instead of pTG when no SRC were detected, while all patients who had preoperative biopsies positive for SRC underwent TG. Most patients (88%) who underwent TG were diagnosed with T1a stage cancer which corroborates the early-stage cancer observed in other asymptomatic cohorts undergoing EGD surveillance [7, 12, 14‐16]. None of the patients without SRC in EGD surveillance progressed to advanced gastric cancer during the nearly 3-year follow-up period. It is important to highlight that 50% of these 24 patients in surveillance have a family history of DGC. As 94% of our patients with SRC were detected within the first 3 annual EGDs, the lack of SRC detection on 3 consecutive EGDs in a high detection program may identify a subgroup of patients with low or no disease burden in whom pTG could be deferred. Other centers corroborate our findings that the highest SRC detection occurs on initial endoscopies and no or rare advanced DGC occurs in short term follow up [5, 11]. In our cohort, asymptomatic patients who did not have any SRC found during 3 consecutive annual exams, were not found to have any SRC in subsequent EGDs. More long-term data is needed to establish the safety of lengthening the frequency of exams in those with multiple consecutive normal EGDs.

The natural history of asymptomatic SRC detected on EGD is not established. In our cohort, all patients found to have preoperative SRC went for TG, so we could not assess the behavior, natural history nor risk factors for progression from early to advanced SRC cancer. It appears that early stage SRC cancer foci could have an indolent behavior [25]. Emerging data suggests most SRC found in patients on random biopsies represent dormant lesions not necessarily tending to progress into advanced cancer [7, 12]. The ability of endoscopic biopsies to discern between T1a and more advanced lesions is not established. Emerging but limited data of surveillance in patients with SRC suggests the progression to advanced DCG is uncommon. In one series of 22 patients with CDH1 PV, 1 patient with SRC who did not undergo TG within one year of diagnosis continued at least annual surveillance and developed stage IV DCG 40 months after SRC were first detected [17]. In another study of 42 patients with CDH1 PV, of 2 patients with SRC who did not have TG, one did not progress to advanced DGC after 7 EGDs over 5 years while the other patient declined surveillance [11]. In a series of 270 CDH1 PV carriers with a follow up time of 31.1 months, 97% with a family history of gastric cancer, 58 patients were diagnosed with SRC in the absence of a macroscopic abnormality and continued 6 monthly endoscopic surveillance without progression to advanced DGC [16]. Whether endoscopic surveillance could be a widely utilized alternative to TG will require larger studies and longer follow-up demonstrating the absence of progression to advanced cancer. We recommend that surveillance endoscopies for patients with CDH1 PV is done in centers with expert endoscopists who can recognize subtle early SRC lesions. Performance of high-quality endoscopy in CDH1 carriers requires scrutinous visual inspection of meticulously cleansed mucosa. Identification of subtle early SRC lesions is very important and requires significant expertise. This does not discount the importance of random biopsies. A recent study by Lee showed the value of random biopsies in diagnosing advanced disease. Omitting random biopsies would have led to underdiagnosis in their historical cohort of 42%. Additionally, the presence of SRC on random biopsies requires a thorough discussion with the patient of risks and potential treatment strategies. It is paramount to have these discussions and follow-up in centers with expert gastroenterologists and foregut surgeons. The timeframe of progression from early to advanced SRC cancer and the relationship of genotypic factors, histo-morphologic and molecular features, and patient related characteristics, requires better understanding to allow a personalized approach to the need for and timing of both pTG and TG.

In conclusion, our study shows EGD surveillance in our center by an expert endoscopist with use of the Weiss Center protocol was not associated with endoscopic complications and demonstrated a high preoperative detection of SRC. Over annual surveillance in 24 patients without SRC, including 54% with a family history of DGC, none had progression to DGC over nearly 3 years of surveillance. Over one in ten of our patients who underwent pTG did not have SRC on the gastrectomy specimen. The data our study provides to the accumulating literature allows a more comprehensive management discussion with patients and may reassure patients without SRCs on endoscopic biopsy who are managed in a center with high endoscopic detection of SRC that pTG can be delayed or even avoided. The data to support delaying TG in patients with SRC is limited and the safety of these approaches requires further study.