Roel Bogie

Chapter 12

are frequently found in the rectum and may contain submucosal invasion, even with a completely benign looking surface pattern. 18, 19 However, multivariable analysis in multiple other LST studies showed an independent effect of LST Kudo subtype and size on risk of SMI, but not of location. 20-23 This may indicate that LSTs behave differently fromother colorectal neoplasms and that assessment of LST morphology and size are most important to determine risks. In the Maastricht UMC+ a higher LST prevalence of 2.3% was found as described in Chapter 4 . This was found among all patients undergoing colonoscopy between 2008 and 2012, before the start of the national CRC screening program. The higher prevalence may be explained by the fact that all staff and endoscopy trainees were trained on the detection of flat neoplasms in 2007, resulting in more skills and awareness. 24 In the national CRC screening setting, the prevalence of LSTs was expected to be even higher. Chapter 5 shows that in the first years of the screening program, 8% of patients had a large non-pedunculated colorectal polyp (LNPCP) of minimal 20mm in diameter. These consist of sessile lesions (Paris Is) and LSTs. About a quarter of the LNPCPs are LSTs, meaning that the prevalence of larger LSTs (minimal 20 mm) is 2% of all persons receiving colonoscopy in the national screening program. Optical diagnosis in LNPCPs In Chapter 3 and 4 the Kudo LST classification was applied to the large flat lesions. As described earlier, LSTs are nowadays more often studied as a subcategory of the LNPCPs. The features characteristic for the Kudo LST subtypes, such as granularity, dominant nodules and (pseudo) depressions, arenowfrequently applied toall LNPCPs. 15 A logistic regressionmodel with stratification with maximal fitting showed that a Kudo V pit pattern, increasing size, distal location, non-granular components, sessile components and depressions were all independent risk factors for submucosal invasion. Some neoplasms have overt signs of submucosal invasion, such as depression and Kudo V pitt patterns. When these signs are missing, non-granular LSTs and non-granular sessile adenomas have the highest risk on covert submucosal invasion. 15 An additional study showed that optical diagnosis has a high sensitivity for submucosal invasion in flat and granular LSTs, but this sensitivity is much lower in LNPCPs located in the distal colon or with sessile components. This information is important in selecting the most suitable method of treatment. A new classification system combining the morphological characteristics with other signs of submucosal invasion in colorectal neoplasms, shows promising results in terms of interobserver agreement and sensitivity for submucosal invasion. 25, 26 LST treatment The most relevant implication of Chapter 3 was that the majority of LSTs were non-invasive at the time of colonoscopic detection, so endoscopic treatment by (piecemeal) endoscopic mucosal resection (EMR) is an option. Pretreatment diagnosis of endoscopic subtype, with specific attention for areas of concern (nodule or depression), enables to identify the LSTs with the highest risk of containing SMI. In these cases en-bloc resection would be beneficial. 6 For large colorectal neoplasms (≥20mm), en-bloc resection can only be achieved by surgery or endoscopic submucosal dissection (ESD). With EMR, mucosal neoplasms are lifted from the submucosa by injection saline beneath the lesion. Then, a snare is placed around the neoplasm and a cut is made with the help of electrocoagulation. 27 Lifting is also the first step of ESD. Here, a knife with electrocoagulation is used to dissect lesions from their margins to the center. Part by part the submucosa is cut apart from the muscularis externa. 28 ESD is more difficult to perform than conventional endoscopic resection,

224