Upper Endoscopic Surveillance in Lynch Syndrome Detects Gastric and Duodenal Adenocarcinomas

Lynch syndrome is among the most prevalent hereditary cancer predisposition syndromes with a population incidence estimated at 1:279 or greater (1). This autosomal dominant syndrome results from germline disease-causing variants in DNA mismatch repair genes including MLH1, MSH2, MSH6, PMS2, or EPCAM. Individuals with Lynch syndrome have increased lifetime risk of multiple cancers, with the highest gastrointestinal (GI) cancer risk being for colorectal cancer (2, 3). Lynch syndrome–related colorectal cancers develop at an accelerated rate with polyp to colorectal cancer progression within a few years, as opposed to 10 or more years in individuals without Lynch syndrome (4). On the basis of increased neoplastic progression and significant colorectal cancer risk, frequent colonoscopic surveillance has proven to prevent death from colorectal cancer and is widely accepted as standard of care (5–12). Accordingly, colonoscopic surveillance for Lynch syndrome in the United States is recommended at least every 1–2 years (6–8, 11).

Apart from colorectal cancer risk, Lynch syndrome is also associated with increased risk of other GI cancers, including gastric and small intestinal adenocarcinoma. However, unlike colonoscopy surveillance guidelines, the role for upper GI tract surveillance in Lynch syndrome is far less clear. Currently, there is a lack of consensus among guidelines from different groups about the age at which to begin surveillance, the frequency of surveillance, and even the individuals with Lynch syndrome for which upper GI surveillance is recommended (5–11). This leads to widely variable practice patterns and can result in confusion for both patients and providers (3, 6, 13). Finally, as more individuals are now pursuing genetic testing due to improved access and decreasing costs, it is inevitable that Lynch syndrome will be increasingly identified, further emphasizing the need to define the most effective surveillance strategies (14).

Recent studies have begun to identify factors associated with upper GI cancers in Lynch syndrome, suggesting that surveillance could be recommended for individuals at high risk. Among individuals testing positive for LS at a commercial laboratory, factors associated with gastric cancer included male sex, increasing age, family history, and a disease-causing variant in MLH1 or MSH2 (15). The international, multicenter Prospective Lynch Syndrome Database also identified age and disease-causing variants in MLH1 or MSH2 as factors independently associated with upper GI cancers in Lynch syndrome (16). In addition, studies from European cancer registries have identified MLH1 or MSH2 disease-causing variants in particular as being associated with upper GI cancer (17–19). While these studies suggest which particular subgroups of Lynch syndrome should be surveilled, there continue to be knowledge gaps in the efficacy of and optimal intervals for surveillance. Therefore, we present our experience with upper endoscopic surveillance in a Lynch syndrome cohort from a tertiary care referral center, highlighting the discovery of early stage upper GI cancers in patients with Lynch syndrome undergoing surveillance.

This is a retrospective analysis of individuals with Lynch syndrome seen at our tertiary care referral center at the University of Pennsylvania (Philadelphia, PA), approved by the University of Pennsylvania Institutional Review Board. Individuals had a medical record with at least one visit to the University of Pennsylvania Health System as well as a documented pathogenic or likely pathogenic variant in MLH1, MSH2, MSH6, PMS2, or EPCAM, or were a confirmed obligate carrier of a known familial pathogenic/likely pathogenic variant. Information obtained from this cohort included demographics, personal history, family history, and upper endoscopy records. We identified individuals with upper GI cancers detected on surveillance as well as those detected outside of surveillance. Records for data abstraction included medical encounters and endoscopy reports from our center, as well as all available data from outside our institution.

At our institution, we routinely offer upper GI cancer surveillance in Lynch syndrome starting at age 30, and recommend 1–2 year intervals between upper endoscopies, consistent with recommendations from the American Society of Clinical Oncology, the European Society for Medical Oncology, and the European Society of Digestive Oncology, though patients receive care outside of our center as well (5, 7, 9). Patients underwent upper endoscopy with a standard adult endoscope, and biopsy for Helicobacter pylori was performed by some providers, but was not performed universally. Upper GI cancers were defined as esophageal, gastric, or duodenal cancers, but did not include nonduodenal small intestinal cancers (as these would not be expected to be detected on standard endoscopic surveillance). Patients with surveillance detected upper GI cancers were asymptomatic, and apart from Lynch syndrome had no other known significant risk factors for upper GI cancers. Comparative statistical analyses were performed using Stata/IC 15.1.

At our tertiary care referral center, 295 individuals with Lynch syndrome were identified as described in Table 1. Of this cohort, 174 (59.0%) were female, 255 (86.4%) were of reported White race, 5 (1.7%) were of reported Hispanic/Latino ethnicity, and 28 (9.5%) reported an Ashkenazi Jewish ancestry. A pathogenic or likely pathogenic variant was noted in MLH1 in 22.7%, MSH2 in 32.9%, MSH6 in 21.0%, PMS2 in 20.3%, and EPCAM in 3.1%. A personal history of colorectal cancer was present in 71 (24.1%). A family history of colorectal cancer was reported in 223 (75.6%), while a family history of gastric cancer was reported in 55 (18.6%), and of small intestinal cancer in 13 (4.4%).

Of the 295 with Lynch syndrome, 217 (73.6%) underwent at least 1 upper endoscopy. There were a total of 660 upper endoscopies performed with a median of 2 (IQR 1–4) upper endoscopies performed per person. In total, 11 (3.7%) of our entire cohort were diagnosed with an upper GI cancer including 1 with esophageal squamous cell carcinoma, 6 with gastric adenocarcinoma, and 4 with duodenal adenocarcinoma. Table 2 compares the characteristics of those who developed an upper GI cancer (including outside of surveillance) to those who had no upper GI cancer and had undergone at least one upper GI endoscopy. There were no significant differences in terms of gender, race, ethnicity, Ashkenazi Jewish ancestry, Lynch syndrome gene, or family histories between the groups. Individuals who developed cancer were overall older, with median age at first endoscopy of 63.3 years (IQR 54.8–67.9) versus 44.9 years (IQR 34.7–55.7; P < 0.001).

Table 3 describes noncancerous endoscopic findings among individuals with Lynch syndrome who underwent upper endoscopy. Among the 217 individuals who underwent upper endoscopy, precancerous findings were observed, including Barrett's esophagus (7, 3.2%), gastric intestinal metaplasia (18, 8.3%), and duodenal adenomas (4, 1.8%); Helicobacter pylori was identified in 6 (2.8%) individuals. Those with upper GI cancer were more likely to have duodenal adenomas (18.2% vs. 1.0%, P < 0.001), although this should be interpreted with caution given the small numbers. There were no significant differences regarding esophageal or gastric endoscopic findings in those with versus those without upper GI cancers, and overall, Helicobacter pylori infection was rare among this Lynch syndrome cohort.

Of the 11 individuals who developed an upper GI cancer, 5 of these cancers were detected on surveillance. Individuals detected outside of surveillance underwent endoscopy for clinical indications, including anemia, abnormal imaging, and dysphagia. Table 4 describes demographics, personal and family history, and endoscopic details of all of the detected cancers, while Fig. 1 illustrates the chronology of upper GI cancer diagnosis and previous surveillance endoscopies (if they were performed). The endoscopy interval before detection of the upper GI cancer was short: 4 of the 5 individuals (80%) with cancer detected on surveillance had a prior negative upper endoscopy within 2 years of their diagnosis with an upper GI cancer. In addition, 4 of the 5 surveillance-detected upper GI cancers (80%) were detected at stage I, whereas only 2 of the 6 (33.3%) detected outside of surveillance were stage I.

In this retrospective analysis of individuals with Lynch syndrome, we find that among 217 subjects who underwent upper endoscopy, 5 (2.3%) had surveillance-detected gastric or duodenal cancer, and 80% of detected cancers were stage I. To our knowledge, this is the largest study to describe outcomes of endoscopic surveillance in Lynch syndrome in a granular manner. These results support that upper endoscopy surveillance in Lynch syndrome detects upper GI cancers, with the majority being detected at an early stage, and also provides evidence to support consideration of upper endoscopy surveillance in the management of Lynch syndrome. Because upper endoscopy is an overall safe procedure and individuals with Lynch syndrome already undergo sedation for colonoscopies at frequent intervals, the addition of upper endoscopy to Lynch syndrome surveillance would not significantly add to patient or procedural burden, though future studies should evaluate the cost-effectiveness of surveillance for all Lynch syndrome individuals versus those with particular risk factors (11, 20, 21). As recent studies have sought to identify which Lynch syndrome subgroups should be surveilled, knowledge gaps and a paucity of literature regarding outcomes of upper GI surveillance remain. Thus, our study helps to begin filling an important knowledge gap that will directly impact surveillance strategies in Lynch syndrome.

Previous studies evaluating the impact of upper GI cancer surveillance in Lynch syndrome have been markedly smaller. A 2002 Finnish study of 73 subjects with Lynch syndrome identified only 1 duodenal cancer (22). A 2017 Turkish study that surveilled 21 Lynch syndrome individuals found no cancers (23). These studies both concluded that upper GI surveillance was not beneficial. A 2018 German study of 44 subjects, however, found surveillance was associated with earlier stage of detection and suggested benefit to endoscopic surveillance for upper GI cancers (24). Early detection of upper GI cancers improves survival overall; though the data for this is not Lynch syndrome-specific, there is no suggestion that it would be different in Lynch syndrome (25). Although 80% of individuals with surveillance-detected cancers in our cohort had stage I disease, future studies are needed in multicenter cohorts to investigate the mortality benefit of early detection of upper GI cancers in Lynch syndrome.

Another stated knowledge gap relates to the intervals between endoscopies, and recommendations differ widely among societies (3). Individuals in our cohort had 1- to 2-year intervals between upper endoscopies, consistent with recommendations from the American Society of Clinical Oncology, the European Society for Medical Oncology, and the European Society of Digestive Oncology (5, 7, 9). This is more aggressive than American College of Gastroenterology and National Comprehensive Cancer Network recommendations, which recommend longer intervals (8, 11). When it comes to colorectal cancer, it is thought that adenomas progress to adenocarcinoma at a more rapid rate in Lynch syndrome than in non-Lynch syndrome colorectal cancer; thus, a 5- to 10-year surveillance interval between colonoscopies in Lynch syndrome is not sufficient (26). It is unclear whether Lynch syndrome–associated upper GI cancers have a similar rapid progression to adenocarcinoma. While our sample size is not robust enough to allow for definitive recommendations on interval timing, it does appear from our data that upper GI cancers develop relatively rapidly in Lynch syndrome as observed in Fig. 1. In their recently published study identifying clinical factors associated with gastric cancer, Kim and colleagues propose 3-year intervals for endoscopy and suggest more frequent surveillance if a patient has multiple risk factors for gastric cancer (sex, age, pathogenic variant of Lynch syndrome gene, and family history; ref. 15). We argue this proposed interval should be further studied, as even without multiple risk factors, individuals in our study with cancer detected outside surveillance may have benefited from more frequent endoscopy to facilitate earlier detection. While older age has been considered a risk factor, we also demonstrate that in Lynch syndrome, upper GI cancers can occur early, as seen in our Fig. 1, emphasizing that early upper GI surveillance should be considered, as some society guidelines also suggest (5–9, 11). Finally, given that many of the patients who developed upper GI cancers were older, future studies should also evaluate if upper endoscopy surveillance intervals should shorten with increasing age.

Among our Lynch syndrome cohort, upper endoscopy did not find a high prevalence of intestinal metaplasia. Moreover, there was no difference in intestinal metaplasia rates between those who developed cancer and those who did not. This is contrary to another study of a Lynch syndrome cohort, where those who developed gastric cancer had high rates (70+%) of concomitant chronic immune gastritis upon histologic examination of the cancer (27). In that study, individuals who developed gastric cancer were not clearly undergoing endoscopic surveillance. Interestingly, while consensus guidelines for Lynch syndrome are relatively consistent regarding support for Helicobacter pylori infection testing and eradication when upper endoscopy surveillance is performed, in our cohort Helicobacter pylori infection was quite rare. This may be related to the lack of racial diversity of the cohort, as it is known that Helicobacter pylori infection is more prevalent in racial and ethnic minorities (28, 29). This finding, although, is consistent with a study from the Dutch Hereditary Cancer Registry, where Helicobacter pylori was not associated with increased gastric cancer risk among patients with Lynch syndrome, with similar findings in other studies (17, 30).

Our study has several limitations, including that it is a single institution, retrospective study at a referral center that limits generalizability. However, our cohort includes good representation among all known Lynch syndrome–causing genes, suggesting our results are applicable to Lynch syndrome in general. Our relatively small sample size limits the ability to detect differences among patients who developed and did not develop upper GI cancer and to identify with accuracy optimal surveillance intervals. Our small sample size also limits the ability to stratify by pathogenic variant. We further did not have MMR status available on the majority of the cancers, limiting our ability to determine if the tumors were Lynch syndrome related. There is also selection bias in that those who elect to undergo endoscopic surveillance may be more motivated due to personal/family history or other factors. For example, Fig. 1 demonstrates that those who underwent upper GI surveillance and developed cancer were generally older than those who developed an upper GI cancer but did not undergo surveillance. Thus age, or some other factor not captured in our data, may help explain differences in decisions to undergo surveillance. However, our overall rates of upper GI cancer are generally similar to another cohort describing patients with Lynch syndrome, providing support regarding the external validity of our results (27). Finally, our subset of patients with cancers within 1–2 years of a prior endoscopy raises the question of endoscopy quality. It is possible that patients had incomplete endoscopies or missed lesions, leading to an apparent short interval between prior surveillance endoscopy and cancer diagnosis. However, it is also possible that, similar to colorectal cancer, Lynch syndrome–associated upper GI cancers may develop quickly with an accelerated progression, which may be the case as 80% of the upper GI cancers discovered were stage I despite recent normal upper endoscopy. However, at this time, it remains unclear whether there is an accelerated carcinoma sequence in Lynch syndrome–related upper GI cancers, similar to colorectal cancer, or if these cancers develop via other mechanisms. Our study provides evidence in an area with a marked paucity of literature, which has been highlighted as an area of critical need to improve Lynch syndrome medical management. As such, despite the aforementioned limitations and our small sample size (of both Lynch syndrome individuals and detected upper GI cancers), this study begins to lay the groundwork to establish concrete upper endoscopic surveillance guidelines for Lynch syndrome, with future studies needed to focus on validating these results in larger and ethnically diverse cohorts.

The utility of endoscopic surveillance for upper GI cancers in Lynch syndrome continues to be an area of uncertainty. Using the largest cohort of surveilled Lynch syndrome subjects to date, we find that among 217 individuals who underwent upper endoscopy, 5 (2.3%) were identified to have a surveillance-detected gastric or duodenal cancer. These surveillance-detected upper GI cancers were primarily detected at an early stage, and were discovered with a median surveillance interval of 1 year, underscoring that, similar to Lynch syndrome–associated colorectal cancers, Lynch syndrome–associated upper GI cancers may develop quickly. Taken together, this study supports upper GI surveillance as part of a Lynch syndrome management program, and that if upper GI surveillance is performed, providers should consider using a shorter surveillance interval of 1–2 years, although this interval should be further studied. Future prospective studies should aim to identify persons with Lynch syndrome most likely to benefit from surveillance, and to identify the optimal surveillance modalities and intervals for these individuals.

S. Kumar reports other from Boston Scientific (travel for educational conference) and other from Olympus (travel for educational conference) outside the submitted work. B.W. Katona reports grants from NIH/NIDDK during the conduct of the study; personal fees from Exact Sciences; and other from Janssen (paid travel related to a clinical trial) outside the submitted work. No potential conflicts of interest were disclosed by the other authors.

S. Kumar: Conceptualization, software, formal analysis, validation, investigation, methodology, writing-original draft, writing-review and editing. C.M. Dudzik: Resources, data curation, investigation, writing-review and editing. M. Reed: Resources, data curation, project administration, writing-review and editing. J.M. Long: Resources, data curation, writing-review and editing. K.J. Wangensteen: Resources, project administration, writing-review and editing. B.W. Katona: Conceptualization, resources, data curation, supervision, validation, investigation, methodology, writing-original draft, project administration, writing-review and editing.

This work was supported by NIH/NIDDK grants T32DK774022 (to S. Kumar), K08DK106489 (to B.W. Katona), and R03DK120946 (to B.W. Katona); and The Jason and Julie Borrelli Lynch Syndrome Research Fund (to B.W. Katona). We would like to acknowledge Dr. E. John Wherry at the University of Pennsylvania Perelman School of Medicine for his help, guidance, and advice related to this project.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.