Roel Bogie

Endoscopic subtypes of colorectal laterally spreading tumors and risk of submucosal invasion: A meta-analysis

categorization. Valuable information regarding the risk of SMI would be missing as the risks of SMI are different (0.5 vs 4.9%). Furthermore, the Paris classification does not distinguish between real depression (sharply demarcated and deep based, deeper than the healthy mucosa) and pseudo- depression (less clear demarcation and shallow). 15 The distinction between‘depression’and‘pseudo- depression’ could aid in the differentiation between deep and superficial SMI, which has therapeutic consequences. Studies will be required to compare the interobserver agreement in both endoscopic Kudo LST classification and Paris classification of LSTs. The strengths of this systematic review reside in the inclusion of a substantial number of studies, reflecting theworldwide experience over approximately twodecades. Our study is the first topresent the global experience on the risk of SMI in LSTs stratified by endoscopic subtype, and lesion size and location to provide a more solid basis for the treatment strategy. Several limitations to our study should be acknowledged. First, there is variation in the definition of LSTs among studies. In order to capture all relevant studies, we expanded the definition to include ‘non-polypoid lesion ≥10mm’, as a surrogate for LST. Some studies included serrated polyps, while others did not. Furthermore, there is wide variation among studies with respect to study design, inclusion criteria and endoscopists’ experience in the diagnosis and treatment of LSTs, which is reflected in the high heterogeneity index in some analyses. To mitigate any potential bias, we performed sensitivity analyses, which showed similar results. For prevalence estimates, only data from population-based studies were used. However, the design and goals of these studies were different. There were two outliers in the prevalence analysis that involved smaller studies from experienced centers. 31, 39 For estimates of the risk of containing SMI in LSTs, both population-based and consecutive lesion studies were included, with even higher heterogeneity. Because the a priori risk of superficial SMI is an indication for en-bloc resection, studies using only specimens that were en-bloc resected could bias the outcomes and were excluded from the analysis. Bias among the studies that reported SMI rates could also be caused by differences in tissue processing. 9, 56 Resected specimens were sectioned with different sampling intervals 47, 48, 62 and the use of specific stains for the muscularis mucosae 78 varied among studies. Piecemeal resection may lead to the underestimation of SMI. 30 Outliers in the SMI analyses were all studies in small-sized populations, which limited their effect on the pooled outcome. Publication bias could lead to an overestimation of the prevalence of LSTs and of the proportion of LSTs with SMI. The statistically significant outcome of the Egger’s test could also be the effect of smaller studies and/or less solid methodology. 79 Furthermore, the location analysis also showed multiple outliers. In the study by Yamada et al., 9 only one proximally located granular LST was included; in the study by Chiu et al., 36 granular LSTs were predominantly located in the proximal colon. A number of studies examined consecutive LSTs resected in endoscopy centers with expertise in ESD, while others were performed in screening colonoscopy practice. Within fecal immunochemical test (FIT) positive populations, the sensitivity for detection of proximal and non-polypoid neoplasms is relatively low. 80 Because of differences in the total area of the epithelial mucosal surface, nodular mixed granular LSTs and non-granular LSTs may differ in their bleeding risk, which could affect the sensitivity of the FIT-test. It remains to be determined whether these factors underlie the large differences in results between studies. In conclusion, this meta-analysis summarizes worldwide data on the risk of SMI in LSTs. Although the vast majority of LSTs are non-invasive and can be treated with (piecemeal) EMR, non-granular LSTs are at higher risk of SMI. Optical diagnosis of LSTs with accurate image interpretation highlights areas of concern (dominant nodule, depression) where en-bloc resection is the preferred therapy.

3

67