In this study, we have identified and validated pCLE diagnostic criteria to diagnose SRCC in individuals with HDGC. We have identified 3 pCLE features associated with the histological diagnosis of SRCC, which, when used as a panel, achieved a diagnostic accuracy of 76%.
Endoscopic surveillance of HDGC is recommended as alternative to surgery in patients fulfilling the HDGC criteria without a known germline mutation and in those who refuse or wish to delay risk-reducing surgery due to medical or psyco-social reasons. Although DGC has very poor prognosis and HDGC carries a 56-70% lifetime risk of gastric cancer, endoscopic surveillance with regular time intervals and strict biopsy protocol is safe in expert centres and informs the best timing of surgery [8, 9, 11, 14]. Approximately one third of patients with a CDH1 pathogenic variant will never develop symptomatic cancer and therefore a careful watch and wait strategy is a reasonable option in some individuals especially given the profound impact of a total gastrectomy on the quality of life. [15, 16] Since endoscopic recognition of early SRCC is challenging, an intensive random biopsies protocol is recommended to obviate to the caveat of endoscopic detection. It is estimated that relying on targeted biopsies only, early SRCC can be missed in as many as 60% of patients.
One of the reasons for low sensitivity of endoscopic imaging is the location of the foci of signet ring cells deeper than the surface mucosal plane, which results in a the common clinical observation of significant false negative rate of superficial biopsies even in the presence of linitis plastica [8]. Confocal endomicroscopy has the advantage of allowing deeper scanning of the gastric mucosa, which could reveal cellular and architectural irregularities otherwise missed by conventional and image enhanced endoscopy. However, the pCLE criteria identified in this study do not relate to the signet ring cells per se, but rather reflect indirect effects on the tissue architecture of the SRCC foci. This is expected since the lateral resolution of pCLE does not allow direct and precise visualization of the individual cancer cells. The indirect signs are due to the blurring of crypt contours due to adjacent signet ring cells (glands with attenuated margins), compression and distortion of the architecture of the crypts (glands with spiculated or irregular shape), dislocation of the glands with increased interglandular space due to clusters of signet ring cells (heterogenous granular stroma with sparse glands) and irregularities of the vasculature due to focal compression of superficial vessels with changes in the intravascular flow (enlarged vessels with tortuous shape and turbulent flow attenuated glandular margins). It is intriguing that although the irregularity in the vascular pattern is one of the most reliable features of SRCC on NBI magnification, particularly when associated with pale areas [17], the sensitivity of the criterion D was very low, suggesting that the assessment of the vasculature on pCLE remains challenging. This is in keeping with the fact that pCLE neoplastic criteria based on vessels are not included in the diagnosis of epithelial dysplasia in other organs of the GI tract [18–21].
pCLE has been extensively investigated in the field of intestinal type early gastric cancer [12]. However, there is scarce data on the utility for HDGC. In a previous study, which is preliminarily reported, a systematic approach to identification of diagnostic criteria has not been used [22]. In this work the comparison between the diagnostic accuracy for early SRCC of pCLE on random locations versus the mapping biopsy protocol failed to show a significant benefit of pCLE over random biopsies. Moreover, non-targeted biopsies taken as per Cambridge protocol revealed SRCC in 11.1% of patients (4/36), whereas in-vivo assessment by pCLE showed irregular patterns in 16.7% of cases (6/36).
In our study the video sequences were obtained predominantly from pale areas identified on WLI and NBI, together with one negative control and pCLE was not used for wider interrogation of normal mucosa. The ultimate goal of utilization of pCLE as clinical adjunct would be to screen the normal looking mucosa for areas suspicious for SRCC and inform need of biopsies. For this reason we think that a cut-off of one positive criterion is more appropriate as it is essential to optimize the sensitivity compared to specificity. Future studies will need to assess the diagnostic accuracy and procedural time of an imaging strategy based on more extensive pCLE assessment of gastric mucosa and targeted biopsies based on pCLE assessment versus the random biopsy protocol.
This study has several strengths. We have used a strict multi-stage process to identify and validate the pCLE criteria. The investigators have extensive experience in the field of HDGC and also worked together in the development of pCLE diagnostic criteria in different clinical applications.[18] However, there are also some limitations. Foci of SRCC can be small and can be missed by targeted biopsies, even in the presence of a clearly visible pale areas. Therefore we cannot exclude that some of the areas with no histopathological evidence of SRCC on biopsy material might indeed contain cancer due to sampling error. This might have led to an underestimation of the specificity. The same investigator (MDP) was involved in the endoscopic procedures and off-line assessment of the video sequences; therefore we cannot exclude bias from recollection of the pCLE procedures findings. However, we used a wash out period of at least 8 weeks to minimise this interference and the investigator had no availability in Phase II of clinical information from the procedure making recollection of pCLE patterns extremely unlikely. The criteria were developed based on video sequences mostly derived from pale areas, therefore this criteria might not be necessarily accurate for identification of SRCC foci located in macroscopically normal mucosa. Finally, these criteria have been developed in the context of HDGC and might not be applicable to diagnosis of sporadic diffuse type gastric cancer. In fact, signet ring cell lesions (insitu and pagetoid) have not been described in patients without CDH1 PV or otherwise positive family history of DGC fulfilling clinical criteria for HDGC, therefore it is reasonable to conclude that early cancer in HDGC and sporadic setting might have difference histopathological features.
In conclusion, we have developed and validated off-line diagnostic criteria for early SRCC in the context of HDGC. The criteria identified in this study have 81% sensitivity and 71% specificity. These criteria will need to be validated in prospective studies using this technology for in-vivo diagnosis.