It is not so long ago that the debate regarding the optimal treatment of Barrett’s high grade dysplasia and mucosal cancer was between surgeons advocating esophagectomy and interventional endoscopists demonstrating their results with endoscopic (piecemeal) mucosal resection (EMR) [1]. There is strong evidence to support the idea that piecemeal EMR has the same cancer-related outcomes as esophagectomy with its perfect histology specimen [1]. As such, endoscopic resection has become the first-choice treatment for high grade dysplasia and early cancer in Barrett’s esophagus, including in cases with submucosal invasion down to 500 µm (sm1) in the absence of high-risk features such as lymphovascular invasion or poor tumor differentiation.

Today’s debate is whether endoscopic resection can be an alternative to esophagectomy in high-risk T1b cases. How remarkable is it that the same interventional endoscopists favor radical resection over piecemeal resection because of the need for a superior histopathological specimen? Evidence to support this approach is largely lacking, as acknowledged in the European Society of Gastrointestinal Endoscopy guideline on endoscopic submucosal dissection (ESD), which presents a weak recommendation for this approach owing to the low quality of evidence [2].

No one will disagree that an en bloc radical resection specimen that is well oriented and has no artifacts is the best specimen to present to the pathologist for histopathological assessment, but is it really necessary? Endoscopic resection ultimately has two objectives: first, to remove the cancer completely and avoid local recurrence; and second, to establish a risk profile for lymph node or distant metastasis based on the resected cancer. The question that, in my opinion, remains unanswered is: do we need an en bloc resection to reach these objectives, or is a piecemeal resection good enough for an expert pathologist to provide these answers?

The study by Gupta et al. in this issue of Endoscopy demonstrates that endoscopists cannot reliably distinguish between T1a and T1b cancers in Barrett’s esophagus using still images [3]. This is a very well designed and executed study, and the results are crystal clear. However, the authors use this outcome to advocate en bloc resection over piecemeal for all T1 lesions because of our shortcomings in endoscopic assessment. Do the results of this study end the debate in favor of ESD? To date, there is still only one randomized trial comparing ESD with EMR [4]. Although this trial was small, the authors showed that R0 resection was achieved more frequently with ESD, but still only in 59% of cases. Importantly, there was no difference between ESD and EMR in complete remission from neoplasia during a follow-up of almost 2 years.

Another issue is that in patient-based endoscopy practice, endoscopists can assess lesions by observing dynamic characteristics such as peristalsis of the esophageal wall. Therefore, image-based assessment alone does not reflect the ideal situation in a real-life setting. A similar study designed to distinguish between T1a and T1b lesions during real-life endoscopy might provide a more definite answer. The same conclusion was drawn in a study in which artificial intelligence (AI) was used to differentiate between T1a and T1b Barrett’s cancers compared with expert assessment [5]. Although the results showed slightly higher accuracy for both the AI system and the experts compared with the current study by Gupta et al. [3], the results were still not within the desired range. It was concluded that performance of the AI system required improvement for use in video-based or real-life endoscopy [5]. This notion is further strengthened by a study from the Barrett’s expert centers in the Netherlands, which showed that endoscopists were able to distinguish between T1 and T2 clinically staged adenocarcinoma in Barrett’s esophagus [6]. Assessment was carried out by an expert endoscopist during patient-based endoscopy. If T1 was suspected, endoscopic resection was performed. Accurate endoscopic assessment was confirmed in 80% of cases.

Indeed, we live in interesting times, with evidence emerging from an ongoing multicenter international trial that the risk of metastasis after endoscopic resection of high-risk T1b Barrett’s cancers is lower than previously presumed [7]. Preliminary results show that metastasis occurred in only 5% during 19 months of intensive follow-up. It is important to add that this is true only for patients who were staged as negative for metastasis at baseline. All patients with a confirmed radical resection at the vertical margin are included in this study, meaning that piecemeal resections can also be included provided the vertical margins are free. The protocol does not mandate free lateral margins.

It is imperative that all resections, certainly in high-risk T1b cases, are assessed by expert pathologists. In this setting, most vertical margins, even in piecemeal resections, can be assessed reliably, with only 16% determined as nonassessable [8].

So, is it time to change our practice and go for en bloc resections in Barrett’s esophagus as the default procedure, as suggested by Gupta et al. [3]? Our apparent poor ability to distinguish T1a from T1b disease might suggest so. However, before we change our practice, we should consider the possibility and the implications of a large proportion of patients being overtreated with a more challenging procedure that is more prone to complications in less experienced hands. Several questions remain unanswered. The first is whether our performance is indeed so poor during live endoscopy assessment. The second unanswered question relates to the necessity for en bloc resection: can we put our trust in good piecemeal resection and expert pathology review, or must we rely on radical resection? The debate continues.

Article published online:
27 November 2024

© 2024. Thieme. All rights reserved.

Georg Thieme Verlag KG
Oswald-Hesse-Straße 50, 70469 Stuttgart, Germany