Histopathology-led quality evaluation of endoluminal excision specimens – not a bad idea!

Predicting tumor recurrence following endoluminal excision of early cancers of the upper gastrointestinal tract can be vexingly difficult. There are a handful of histopathologic criteria that can predict risk of local and distant tumor recurrence including invasion depth, tumor differentiation grade, and lymphovascular invasion [1]. Recent reporting standards from the International Collaboration on Cancer Reporting (ICCR) have now brought these parameters together in a proforma that can be used to harmonize histopathologic evaluation of endoluminal excision specimens and foster uniform interpretation between sites [2]. These reporting standards feature a panoply of essential criteria, such as those listed previously, as well as optional criteria, including histopathologic tumor type, for which the impact on patient management is currently unclear. However, these risk prediction criteria are crude and some patients are unnecessarily overtreated, whilst others are dangerously undertreated [3] [4]. How can we do better? Are there any relevant reporting parameters that we have so far missed, which may have been hiding in plain sight?

In this issue, Krause and co-workers report their findings from analysis of the quality of submucosal tissue depth in a moderate-sized cohort of 76 endoluminal excision specimens (36 endoscopic mucosal resection [EMR] specimens and 40 endoscopic submucosal dissection [ESD] specimens) [5]. The rationale for this work is straightforward: superficial submucosal excisions complicate assessment of submucosal invasion depth and might increase the risk of incomplete R1 excision. Good quality endoscopic resection (ER) specimens preferably demonstrate a sizeable vertical margin of submucosal tissue along the full horizontal width of the specimen.

“Feedback on specimen quality would allow endoscopists to improve their practice.”

To begin to evaluate this, Krause and co-workers measured the proportional width of endoluminal excision specimens lacking any tissue below the muscularis mucosae, here referred to as a “submucosal defect” (SMD) ([Fig. 1]). The authors compared SMDs between ESD and EMR specimens, as well as between three experienced treatment sites. An interesting aspect of the work is that the authors used objective digital histopathology quantification to arrive at their conclusions. This further leverages the reproducibility of their results.

Fig. 1 Pathologic evaluation of submucosal defects in an endoluminal excision specimen showing: a on macroscopy of a fresh gastric endoscopic submucosal dissection specimen, a depressed lesion at the center of the specimen, which has been removed with wide lateral margins; b on low-power overview of a central slice of the specimen (hematoxylin and eosin [H&E] stained), the lesion (asterisk) corresponding with the depressed area on the fresh specimen, with there being no submucosal tissue below the muscularis mucosae (arrowheads) in the adjoining tissue, indicating a superficial excision, while centrally the muscularis mucosae has been completely obliterated by the invasive lesion, complicating depth assessment.

Their short report drives home two important messages. First, on average about one-third of the horizontal width of ER specimens in their cohort completely lacked any tissue below the muscularis mucosae. Disconcertingly, this was also the case in key regions of the specimen that contained the invasive lesion of interest. Second, there was significant variation between treatment centers, suggesting that this parameter can be addressed through further training and, possibly, histopathology feedback.

What do these results mean? A naïve interpretation might be to dismiss these results on the assumption that such SMDs are the result of suboptimal histopathology work-up. However, this is unlikely to be case. Although tissue retraction during histopathology processing will unavoidably introduce some variation between ER slices and cases, this variation alone cannot explain complete absence of submucosal tissue. It is more likely that submucosal tissue defects, and the variation between treatment sites, are operator-dependent and therefore a consequence of endoscopy technique. A subgroup analysis of T1b cases in the study cohort (n = 9) suggests that SMDs are indeed associated with R1 excision, although this analysis lacked statistical power to draw robust conclusions.

Although this work takes an important step in the direction of histopathology-led quality evaluation, the main caveat to their conclusions is that the authors did not correlate SMDs with the endoscopic appearance or esophageal location of the lesion. This will leave some endoscopists wondering how they might improve their practice if it is unclear what drives this variation between patients and treatment sites. This study in fact excluded bulky lesions and deeply penetrating lesions. In follow-up work, the authors will need to analyze submucosal tissue defects alongside the Paris classification of the lesion and the excision location in a larger and prospective series to understand the impact of these variables.

Histopathology-led quality evaluation is by no means a novel concept, of course, in particular in the surgical realm. Macroscopic evaluation of the mesorectal cuff, for example, is a standard reporting parameter in the pathology work-up of surgical rectal excision specimens and poor excision quality is linked to greater risk of tumor recurrence [6]. There is therefore no need for endoscopists to fear the introduction of such quality parameters, even less so if this could be guided by objective digital histopathology tools, which avoid subjective histopathology interpretation [7]. Feedback on specimen quality would allow endoscopists to improve their practice and understand outcome differences between patients.

Much remains to be done before this parameter can be introduced as a standard reporting parameter for endoluminal excision specimens of the upper gastrointestinal tract though. Krause and co-workers will undoubtedly advance their retrospective analysis, with a prospective analysis that tackles some of the caveats outlined above. Manual annotation of scanned slide images makes the current workflow labor-intensive, which means it is limited to the research setting. However, with the rapidly advancing introduction of digital histopathology and automated computer vision analysis [8], it is not at all unlikely that this parameter will find its way into reporting standards – maybe initially as a desirable reporting parameter and, as more evidence becomes available, as a mandatory reporting parameter.

Publication History

Article published online:
14 December 2021

• References

• 1 Pech O, May A, Rabenstein T. et al. Endoscopic resection of early oesophageal cancer. Gut 2007; 56: 1625-1634
• 2 International Collaboration on Cancer Reporting. Datasets for Digestive Tract. Available from (Accessed: 21.09.2021): https://www.iccr-cancer.org/datasets/published-datasets/digestive-tract
• 3 Schölvinck D, Künzli H, Meijer S. et al. Management of patients with T1b esophageal adenocarcinoma: a retrospective cohort study on patient management and risk of metastatic disease. Surg Endosc 2016; 30: 4102-4113
• 4 Manner H, Wetzka J, May A. et al. Early-stage adenocarcinoma of the esophagus with mid to deep submucosal invasion (pT1b sm2-3): the frequency of lymph-node metastasis depends on macroscopic and histological risk patterns. Dis Esophagus 2017; 30: 1-11
• 5 Krause J, Rösch T, Steurer S. et al. Quantitative analysis of submucosal excision depth in endoscopic resection for early Barrett’s cancer. Endoscopy 2021; 54: 565-570 DOI: 10.1055/a-1659-3514.
• 6 García-Granero E, Faiz O, Munoz E. et al. Macroscopic assessment of mesorectal excision in rectal cancer: a useful tool for improving quality control in a multidisciplinary team. Cancer 2009; 115: 3400-3411
• 7 van der Wel MJ, Coleman HG, Bergman JJGHM. et al. Histopathologist features predictive of diagnostic concordance at expert level among a large international sample of pathologists diagnosing Barrett’s dysplasia using digital pathology. Gut 2020; 69: 811-822
• 8 Janowczyk A, Madabhushi A. Deep learning for digital pathology image analysis: A comprehensive tutorial with selected use cases. J Pathol Inform 2016; 7: 29