Endoscopic predictors of deep submucosal invasion in colorectal laterally spreading tumors.

BACKGROUND AND STUDY AIMS

The depth of invasion of the bowel wall influences the treatment of colorectal laterally spreading tumors (LSTs). The aim of this study was to evaluate the risk factors and patterns of submucosal invasion in a large series of LSTs that were removed en bloc.

PATIENTS AND METHODS

Prospectively collected endoscopic and pathological data on a total of 822 LSTs, ≥ 10 mm in size and removed en block by endoscopic submucosal dissection (n = 670) or surgery (n = 152), were retrospectively analyzed.

RESULTS

In 414 LSTs of the granular type, submucosal invasion was detected in 80 cases (19 %; 95 % confidence interval [CI] 16 - 23) and was deep (≥ 1000 μm) in 79 % of cases. The invasion site was under a large (≥ 10 mm) nodule (56 %), depression (28 %), or was multifocal (16 %). Risk factors for deep submucosal invasion on multivariate analysis were the presence of a large nodule (odds ratio [OR] 12, 95 %CI 2 - 59), depression (OR 59, 95 %CI 9 - 387), and invasive pit pattern (OR 33, 95 %CI 12 - 88). The sensitivity and specificity of invasive pit pattern for detection of deep submucosal invasion were 52 % (95 %CI 40 % - 64 %) and 98 % (95 %CI 96 % - 99 %), respectively.In 408 LSTs of the nongranular type, submucosal invasion was detected in 159 cases (39 %; 95 %CI 34 - 44) and was deep in 54 % of cases. The invasion site was under a submucosal mass-like elevation (10 %), depression (45 %), or was multifocal (45 %). Risk factors for deep submucosal invasion were the presence of a submucosal mass-like elevation (OR 8, 95 %CI 1 - 61), depression (OR 28, 95 %CI 8 - 97), and invasive pit pattern (OR 79, 95 %CI 25 - 256).

CONCLUSIONS

Because of a substantial risk of submucosal invasion and multifocal invasion, granular type LSTs with a large nodule or depression and nongranular type LSTs should be endoscopically removed en bloc.