Roel Bogie

General discussion

LST subtype. Of the 29.0% possible pairs with discordance, more than 1/3 (10.7%) were between LST-NG-FE and LST-NG-PD and 1/3 (9.6%) between LST-G-H and LST-NG-FE. This indicates that experts disagreed on presence of pseudo-depression in LST-NG and granularity patterns. In another study, the kappa to differentiate LST-NG-PD from LST-NG-FE was much higher among two Japanese experts (0.93). 10 Next, a group of endoscopy trainees was tested, which had lower interobserver agreement initially. An e-learning for training of the Kudo LST classification has been developed that was completed by all endoscopy trainees before they applied the endoscopic classification again in a post-test. Afterwards the trainees performed better, almost matching the results of the expert group (Gwet’s AC1 coefficient: 0.59, 95%CI: 0.53 – 0.65). Analyzing discordant pairs showed that pre training, the trainees had difficulties with differentiating LST-NG-FE from LST-NG-PD and LST-NG FE from LST-G-H similar to the experts. However, differentiating LST-G-NM from LST-G-H was also problematic. Post-training, the disagreement in differentiating LST-G-H from LST-NG-FE and LST-G-H from LST-G-NM improved, but differentiating LST-NG-FE from LST-NG-PD was slightly worse (from 12.5% to 13.4% discordant pairs). This means that the recognition of pseudo-depressions remains difficult and additional training and/or clarification of the definition is needed. Using objective measures when to call a granule dominant or not, could help to improve overall agreement. 11 Morphology of LSTs can also be described using the Paris morphological classification. 12 Among experts, substantial agreement was measured (Gwet’s AC1 of 0.71, 95% CI: 0.65 – 0.78, Chapter 2 ). This was much lower among trainees, both pre- and post-test (Gwet’s AC1 of 0.33 and 0.45 resp.). In another study among both trainees and gastroenterology staff, applying the Paris classification after training showed substantial agreement in LSTs (Gwet’s AC1 0.62, 95% CI: 0.53 – 0.71). No differences between trainees and staff members were found in the agreement coefficients. 13 These results show that as an alternative to the Kudo LST classification, the Paris classification could be used for communication, although information about granularity is missing. Chapter 2 shows that e-learning is an effective tool to familiarize with and learn endoscopic classifications. The website uses cases with questions about morphology and best treatment, alternatedwith videos providing tips and tricks for applying endoscopic classifications and assessing the risk of submucosal invasion. It prepares young endoscopists to use the Kudo LST classification in real life setting. The use of cases in an online environment is also in line with new learning tools outside the field of medicine and fits well with the needs of newer generation endoscopists. 14 LST prevalence In the last two decades, the number of studies reporting LSTs rapidly increased with multiple definitions used. To study worldwide LST prevalence and the specific features, a pooled analysis of prevalence, morphology, colorectal location, and histological outcomes was performed in Chapter 3 . The prevalence was quite low, in only 0.83% (95% CI: 0.62 – 1.07) of all patients undergoing colonoscopy LSTs were found. Among all colorectal neoplasms found, 3.6% were LSTs (95% CI: 2.5 – 4.9). Distal location and larger size were associated with an increased submucosal invasion (SMI) risk of LSTs. LST subtype influenced the risk of SMI, with high risk for non-granular pseudo-depressed and granular mixed-nodular LSTs and low risk for granular homogenous LSTs. The conclusion of the study was that assessment of LST morphology is important for SMI risk estimation before endoscopic treatment. Using the pooled data, multivariable analysis to analyze independent risk factors was not possible. In a large study analyzing risk factors for SMI in large non-pedunculated neoplasms (≥20mm) distal location was an independent risk for SMI. 15 Recently, rectal located LSTs were also found to have a greater malignant potential than LSTs located in the colon. 16, 17 LST-G-NM

12

223