Over the past decades, there has been a significant increase in the incidence of esophageal cancer (EC), in particular, of esophageal adenocarcinoma (EAC), having increased 7-fold in the last 50 years. Despite advances in the management of Barrett’s esophagus (BE) and endoscopic screening programs, the 5-year survival rate continues to be under 20%.1 EC holds seventh place worldwide in incidence, sixth place in mortality, and is predominant in men, at 70% of cases.2

Given that tumor stage at diagnosis determines prognosis, guidelines recommend selective screening in patients with chronic gastroesophageal reflux and additional factors, such as age above 50 years, male sex, central obesity, smoking, and a family history of EC or BE.3,4 However, more than 90% of patients with EAC have no previous diagnosis of BE and most patients presenting with BE do not develop cancer, raising doubts as to the efficacy and cost-benefit of said programs.5–7

In addition, over half of the cases of EAC are diagnosed at advanced stages, illustrating the limitations of current surveillance strategies.8 The medical literature also reports cases of EC detected a short time after negative endoscopies, a situation similar to that seen in post-colonoscopy colorectal cancer.9–12 To standardize this phenomenon, the terms post-endoscopy esophageal adenocarcinoma (PEEC) and post-endoscopy esophageal neoplasia (PEEN) have been proposed. Undetected high-grade dysplasia and EAC are their primary causes.13–15

Different technical factors may contribute to these diagnostic errors, including insufficient BE segment inspection time, non-compliance with the Seattle biopsy protocol, incomplete mucosa sampling, or difficulty in identifying subtle lesions.14–16 Nevertheless, the information available on PEEC is limited and comes from small case series, lacking clarity on tumor stage at diagnosis.15,17,18 If early-stage esophageal cancers have a long natural history,17,19 then any cancer detected 3 years after endoscopy could have been missed before, as shown in previous studies.15,18,20,21

The present study aims to compare the clinical, endoscopic, and survival characteristics of patients with EC detected during their first endoscopy with those of patients whose EC was diagnosed after three consecutive normal endoscopies, prior to the cancer diagnosis, to provide evidence on the scope and characteristics of PEEC.

The initial cohort of linked data included 509 patients diagnosed with EAC between July 2012 and June 2023, of which only 473 met the inclusion criteria (Fig. 1). Table 1 summarizes the demographic and clinical characteristics of the cohort.

The mean age (± SD) of the patients at EAC diagnosis was 69.3 ± 10.4 years, and 7 out of every 10 patients were men. Three-quarters of the cancers (74%) were located in the lower part of the esophagus. At diagnosis, 24% of the cohort had BE. Of the patients with known TNM stage at diagnosis, 75% presented with cancers in stage III or IV, whereas only 6% were diagnosed with stage 0 (carcinoma in situ) or stage I cancers.

Follow-up and survival analysis

The median follow-up was 5.2 years (interquartile range [IQR], 3.5–8.6 years). There were no significant differences between the two groups in the survival analysis, summarized in Fig. 3.

Figure 3.

Survival curves for the initially diagnosed tumors and the post-endoscopy esophageal tumors. Survival graph (Kaplan-Meier):

DEC: detected esophageal cancer; PEEC: post-endoscopy esophageal adenocarcinoma.

In the present population cohort study, 6.6% of the patients diagnosed with esophageal cancer had undergone one endoscopy between 3 and 36 months before the cancer diagnosis. This results in an estimated overall misdiagnosis rate of 6.1% in the study cohort (the interval cancer rate = number of interval cancers/number of interval cancers + total number of detected cancers x 100). This figure falls within the lower range of estimates from previous studies, which have reported rates from 4.0% to 14.3%.15,18,19,21,23,24 In addition, there was an association between patients having undergone a previous endoscopy and their tumor stage at diagnosis; a negative endoscopy was more frequent in patients with disease stage 0/1. Patients with BE, alarm symptoms, and esophageal cancer in the upper third of the esophagus, also had a higher probability of having undergone an endoscopy before the cancer diagnosis.

Studies suggest that either some esophageal cancers are not detected at endoscopy or that the natural evolution of a significant number of cancers progresses from an early lesion that is endoscopically invisible to an advanced cancer, in a relatively short period of time. Considering the latter scenario, an alternate interpretation would be that only stage 2–4 cancers diagnosed within 3−36 months following a negative endoscopy are potentially missed, corresponding to a more conservative estimated omission rate of 5.6%. Nevertheless, we believe that the overall figure may be more exact. Even though there are very few studies on the natural history of untreated early esophageal cancer, two small cohort studies suggest that some patients with early cancer survive more than 5 years,25,26 and there are increasingly more confirmations of the slow progression of endoscopically visible dysplasia in patients with BE.27,28

Previous negative endoscopies were more frequent in patients with early cancers, and some of those patients may have been under periodic surveillance. This is supported by our finding that BE was more frequent in those patients (42% vs 58%, p = 0.026).

Several reasons have been proposed for a cancer not being diagnosed during the initial endoscopy.29,30 They include the fact that the endoscopist may not identify a possible lesion, or if identified, its importance is not recognized, and so a biopsy sample is not taken or is taken with an insufficient number of samples.17 The endoscopic appearance of an incipient cancer can be very subtle, and the only sign may be slight changes in the color or contour of the mucosa.4 Thus, these lesions are difficult to detect, reinforcing the need to perform high-resolution white light endoscopy; adding improved imaging techniques (narrow-band imaging [NBI], blue laser imaging [BLI], iScan) may be considered an alternative but their preferred use over conventional white light endoscopy is not recommended in the guidelines, not even in high-risk populations.

The use of proton pump inhibitors before endoscopy may also increase the probability of lesions not being detected,31 by favoring mucosal healing. Ideally, endoscopies should be performed before the prescription of antacid medications. It has been suggested that the middle and upper esophagus may be poorly inspected, given that the endoscope may be rapidly withdrawn during the final phases of the procedure, which reduces the possibility of visualizing subtle lesions.29 In agreement with other studies,30,32 we found a higher number of patients with missed lesions in the upper third of the esophagus, who had undergone an endoscopy one year before EAC diagnosis, than in the patients with lesions in the lower third of the esophagus or the gastroesophageal junction.

More than one-third of the negative endoscopies were performed due to alarm signs, such as dysphagia, and those symptoms were more frequent in patients with DEC (37% versus 63%, p = 0.027). This concurs with the fact that alarm symptoms tend to be associated with advanced disease and with a lower rate of previous endoscopies in patients with stages T3 and T4.

The relation between BE and PEEC deserves special attention. If the number of cases of PEEC were calculated without including the 114 patients with BE, the PEEC rate in our study would have dropped from 6.6% to 5.3%. The percentage of PEEC cases with a history of BE at one year was 63%, compared with 37% of EAC cases without BE, representing an alarmingly high percentage of undetected EAC that should be addressed. Our results highlight the need for stricter compliance with the diagnostic protocols for BE recommended in guidelines, greater attention paid to making diagnoses through advanced imaging and to biopsy protocol, and a more pro-active attitude in the treatment of gastroesophageal reflux disease (GERD). Different case series, when analyzing PEEC, exclude patients with BE18,21,32–34 or with BE and no dysplasia,23,35,36 whereas Yalamarthi et al.37 call attention to the error-inducing role of histologic interpretation (27%) at diagnosis.

The study’s strengths include its multicenter analysis and its assessment of patient characteristics (both clinical and endoscopic), as well as tumor features, such as stage at diagnosis and its association with the performance of previous endoscopies. Endoscopies performed within 3 months prior to the cancer diagnosis were excluded to minimize the likelihood of including patients who were under follow-up after an initial concerning finding.

The retrospective design, a relatively small patient cohort, and medium-term follow-up are some limitations of our study that could potentially bias the results. Additionally, there was no standardization in the performance of the endoscopies or in the interpretation of the histopathologic reports.

Inadequate evaluation of premalignant or focal lesions, inadequate quality of the endoscopy and biopsy, and poor decision-making regarding surveillance or follow-up plans have been identified as the most common explanations of PEEC.30 The unadjusted PEEC rate was 6.6% in our study, which is within the goal of <10% proposed in a position statement on quality norms in endoscopy in the United Kingdom.38Table 4 presents a more comprehensive overview of the problem by compiling different case series reviewed and published on PEEC, highlighting their heterogeneity.

Artificial intelligence (AI) has emerged as a powerful tool in gastroenterology, especially in the early detection of neoplastic and pre-neoplastic lesions. A multicenter study conducted in China shows the potential of AI for improving the detection of superficial squamous cell carcinomas of the esophagus and precancerous lesions in areas where conventional methods, such as white light endoscopy and NBI, may have limitations. Their trial showed that AI reduced the undetected lesion rate, with a failure rate per lesion of 1.7% in the AI-assisted group, compared with 6.7% in the routine group. Although the difference was not statistically significant, the trend suggests that AI may have a potential clinical benefit, particularly in terms of diagnostic sensitivity. In addition, AI could aid in reducing interpretation variability among endoscopists, a well-known challenge in the detection of subtle lesions. However, considering the long-term evaluation of cost-benefit and effectiveness in real-world clinical settings is crucial. The integration of AI into daily practice should be carefully weighed against additional costs and the need to train personnel.39

Additional studies are needed to determine whether endoscopic detection improved through advanced imaging techniques (NBI, BLI, iScan), a longer inspection time, a higher number of biopsy samples, and the performance of repeat endoscopy within defined time limits would result in the surveillance of BE having a real impact on reducing the incidence and prevalence of EAC. In addition, cost-effectiveness analyses of the proposal should be supported by the implications of the findings.

Based on our findings and the present medical literature, we suggest implementing the following strategies that could contribute to reducing the rate of PEEC: 1) Strict compliance with the biopsy protocol. It is essential to rigorously adhere to the established protocols, such as the Seattle biopsy protocol, for patients with BE, guaranteeing adequate sampling of the mucosa for the early detection of dysplasia or neoplasia, 2) Use of advanced imaging techniques and photo-documentation. The incorporation of improved imaging techniques, such as NBI, BLI, iScan, or chromoendoscopy, increases sensitivity in the detection of subtle lesions, especially in the proximal esophagus, where said lesion may go unnoticed; photo-documentation with at least 4 images of the esophagus (upper, middle, cardia, and retrovision) is also recommendable, 3) Increased inspection time and sedation use. It is essential to spend adequate time on the endoscopic inspection, suggesting a minimum of 7 min for the endoscopy, or one minute for each centimeter of BE, to minimize the risk of missing neoplastic lesions; sedation can be a valuable tool for ensuring patient collaboration and reducing discomfort during the procedure, 4) Proactive follow-up in high-risk patients. Patients with tumors of the head and neck, BE, or a family history of EAC should have regular endoscopic surveillance, with a proactive focus on the early detection of malignant changes, and 5) Negative endoscopy reevaluation. In cases of previously negative endoscopies, especially in patients who are asymptomatic, at high risk, and with equivocal findings, early reevaluation utilizing advanced techniques should be considered, as well as possibly repeating the endoscopy to ensure that no important lesions have been missed.

Our study results showed that 6.6% of esophageal cancers were diagnosed after a previously negative endoscopy, a diagnostic omission that was significantly more common in patients with BE, proximal tumors, asymptomatic disease, and early disease stages. These results underline the need for optimizing endoscopic practice, focusing on a meticulous inspection of the proximal esophagus and strict adherence to biopsy protocols in groups identified as high-risk, to reduce the rate of PEEC.

This study was financed using the authors own resources.

Artificial intelligence was not utilized at any stage of this scientific study, including the design, data collection, analysis, or interpretation of results. All the procedures and analyses were carried out manually by the research team, guaranteeing that the conclusions are based solely on traditional methods of scientific research and medical judgement.