P138 Results of dysplasia surveillance programme with chromoendoscopy for inflammatory bowel disease: do we have to biopsy all lesions?
J. F. Muñoz*1, M. García-Alvarado2, B. Sicilia3, M. Sierra4, N. Fernandez4, L. Arias3, J. Barrio5, B. Velayos6, L. Hernández-Villalba7
1Complejo Asistencial Universitario de Salamanca, Gastroenterology, Salamanca, Spain, 2Hospital Virgen de la Concha, Gastroenterology, Zamora, Spain, 3Complejo Asistencial de Burgos, Gastroenterology, Burgos, Spain, 4Complejo Asistencial Universitario de León, Gastroenterology, León, Spain, 5Hospital del Río Hortega, Gastroenterology, Valladolid, Spain, 6Hospital Clínico Universitario, Gastroenterology, Valladolid, Spain, 7Hospital Santos Reyes, Gastroenterology, Aranda de Duero, Spain
Chromoendoscopy with targeted biopsy detects any dysplasia 9 more times than conventional colonoscopy does with random biopsies, and it is the technique recommended in main guidelines and consensus.
We analysed the results of the screening programme by chromoendoscopy in 6 hospitals in Castilla y León (Spain) and assessed whether is it possible to identify low-risk lesions of dysplasia by the endoscopic appearance and thereby avoid histological analysis. Retrospectively, we collected the screening colonoscopies with chromoendoscopy between October 2012 and September 2015. The technique was performed with indigo carmine (0.2%–0.4%) managed with catheter spray (Olympus PW-205V) and high-definition endoscopes. We present a descriptive analysis of the findings and the diagnostic yield of pit pattern to predict dysplasia. We also calculate the Kappa coefficient for Kudo’s pit pattern and dysplasia agreement.
A total of 243 patients were included in the study (46.5% female and 53.5% male). All of them met the ECCO criteria for inclusion in a screening programme for CRC. The median duration of the disease was 15.8 years (SD 6.86); 72.7% had ulcerative colitis (56.8% extensive). The duration of the colonoscopy was 36.6 (SD 10.36) minutes, and preparation was adequate (Boston ≥ 6) in 73.3%. Endoscopic activity was observed in 29.63% of patients (14% Mayo 1 and 6.58% Mayo 2), and histological activity was in 28%. In total, 953 lesions (3.9 lesions per patient) were found; most of them were located in the left colon (65.28%), followed by the rectum (15.3%). Further, 87.2% had no neoplastic potential (40.3% normal mucosa, 34.5% hyperplastic lesions, and 11.9% inflammatory polyps), 10.6% had low-grade dysplasia (DBG), and 3 cases with high-grade dysplasia (0.3%) were found. Correlation between the presence of dysplasia and Kudo’s pit patterns predictors of dysplasia (≥ III) was low (Kappa 0.235; p < 0.001) with an area under the curve (AUC) of 0.65 (sensitivity: 40.7% specificity: 89.4%, PPV 25.3%, and NPV:94.5%). A learning curve is possible because the correlation in the last 30 colonoscopies was higher than in the first 30 (Kappa: 0.256; p = 0.005 with an AUC of 0.698 vs kappa 0.1; p = 0.404 with AUC of 0.557). In the multivariate analysis, only the colon cleansing was associated with a greater number of found lesions.
In a screening programme with chromoendoscopy, many lesions are detected but most of them have not neoplastic potential; only about 10% are premalignant. Kudo’s pit pattern classification is not an accurate diagnostic method to detect dysplasia in these patients. However, these results improve with experience, but histological analysis of all lesions remains neccessary.