Aims:
Tiny areas of deep infiltrating carcinoma in colorectal lesions can go unnoticed even
on detailed examination with chromoendoscopy. Therefore, suspicion of this drawback
should be taken into account in decision making. So, it is compulsory needed to determine
predictive factors related to deep hidden infiltrating carcinoma (DHIC).
Methods:
Retrospective analysis with a prospective data source of endoscopic mucosal resection
database. Univariate and multivariate analysis of the factors associated to DHIC were
carried out.
Results:
We assessed 174 lesions (31.72 mm [20 – 120]), 58% located in the proximal colon.
In 10/174 (5.74%) DHIC was detected and the patients had to be referred to surgery.
In the risk analysis, it was found that: size (>/< 40 mm), location (rectum-sigmoid
colon), morphology (0-Is/mixed patterns) and Kudo's Pattern V behaved as independent
factors in a univariate analysis (p < 0.01). However, in a multivariate analysis only
size > 40 mm [OR 11.40 (p = 0.007)], shape 0-Is [OR = 8.94 (P = 0.02)] and rectosigmoid
location [OR 13.50 (P = 0.03)] were noticed as independent risk factors.
Conclusions:
The detection of some or all of these factors might be an effective tool for the prediction
of DHIC and therefore serve as support to tailor the therapeutic decision.
