Roel Bogie

Chapter 3

The findings of this meta-analysis show that the majority of LSTs are non-invasive at the time of colonoscopicdetection, allowing for removal by (piecemeal) EMR. However, detailedcharacterization with prediction of the risk of containing SMI is necessary to distinguish LSTs that can be resected piecemeal from those in which en-bloc resection would be more favorable. Chromoendoscopy is a useful tool in determining the endoscopic LST subtype as it better delineates nodules and granules. Because of the pooling effect of the dye, pseudo-depressions will also be better visualized. A prediction of SMI in LSTs can be made with higher accuracy by detailed inspection of the pit pattern. Among a highly-experienced group of endoscopists, the accuracy in prediction of deep SMI was 91.7%. 9 The recognition of a large nodule or a depressed area is not complicated, but prediction with high confidence of the depth of SMI using pit patterns is challenging, and magnifying colonoscopes and additional training are required. 71 Application of the endoscopic LST classification is therefore a simple first step to determine the a priori risk of SMI. In general, homogenous granular and flat elevated non-granular LSTs have a low a priori risk of containing SMI (0.5% and 4.9%, respectively). Where en-bloc resection is not feasible or safe, the endoscopist can choose to apply oligo-piecemeal EMR. Nodular mixed granular and pseudo-depressed non-granular LSTs have a higher a priori risk of containing SMI and applying advanced imaging techniques to the areas of interest (dominant nodules, pseudo-depressions) could inform treatment decisions. Training in EMR of large neoplasms remains a critical first step, because most LSTs can be efficiently treated this way. 6, 71 In the absence of experience with diagnosis by advanced imaging, the endoscopist should refer the patient to a centre with experience. As shown in this meta-analysis, the prevalence of pseudo-depressed non granular LSTs is fairly low (5.5% of all LSTs). The additional workload for performing ESDs is not significant, but the oncologic outcome of resection is potentially better. Once the decision is made that an LST has an increased risk of SMI, the treatment modality will vary widely based on local experience, patient preference, endoscopist training, costs, and logistics. In this meta-analysis, we found a higher risk of SMI in LSTs located in the distal colon compared with those in a proximal location (OR 2.50, 95% CI: 1.24 – 5.02). This is supported by the finding that pseudo-depressed non-granular and nodular mixed granular LSTs, which have a higher risk of SMI, are more often located in the distal colon. Endoscopic resection of neoplasms located in the distal colon, especially in the rectum, is technically easier to perform and has a lower risk of perforation than when performed in the proximal colon. 77 Another finding of this meta-analysis was that the pooled rate of LSTs did not appear to change by geographic region or starting year of the study. Most included studies used general colonoscopy populations and there were only a few studies in screening settings. LST prevalence may be higher in screening populations selected with fecal occult blood tests and future studies will be required to investigate the prevalence of LSTs in such high risk populations. In this meta-analysis, the endoscopic Kudo LST classification, instead of the Paris classification, was used to subdivide LSTs. The majority of the papers used the endoscopic Kudo classification. As displayed in Figure 3.1 , homogenous granular and flat elevated non-granular LSTs are often consideredas Paris 0-IIa lesions, whilenodularmixedgranular LSTs are consideredas Paris 0-IIa+Is and pseudo-depressed non-granular LSTs as Paris 0-IIa+IIc, although other combinations are possible. 1 The multicentre ACE study into risk factors for SMI in large (≥20mm) non-pedunculated neoplasms that were resected by EMR used the Paris classification. Along with rectosigmoid location and lesion size, the Paris classification in combination with granularity status was shown to be predictive of SMI. 17 Granularity is not included in the Paris classification and should be mentioned separately. From a clinical, oncologic, and biologic point of view, simply referring to both homogenous granular and flat elevated non-granular LSTs as LST 0-IIa lesions would be a non-differential

66