Subscribe to RSS
DOI: 10.1055/s-0038-1637189
IN VIVO DIAGNOSTIC ACCURACY OF THE NICE CLASSIFICATION FOR PREDICTING DEEP INVASION IN COLONIC LESIONS
Publication History
Publication Date:
27 March 2018 (online)
Aims:
-
The NICE 3 category in the NBI International Colorectal Endoscopic classification was designed to predict deep submucosal invasion in colorectal lesions for which surgery is indicated.
-
The primary aim of this study was to assess the diagnostic accuracy of the NICE classification for predicting deep submucosal invasion in routine clinical practice.
-
The secondary aim was to determine its accuracy combined with the appreciation of other lesion characteristics.
Methods:
-
Multicenter, prospective, observational diagnostic accuracy study conducted by 58 endoscopists at 17 university and community hospitals in their routine clinical practice.
-
All consecutive lesions > 1 cm assessed with NBI were included.
-
The primary outcome was deep invasion according to the index test (NICE classification) and the gold standard (histology).
-
Conditional inference trees were fitted for the analysis of diagnostic accuracy.
Results:
-
Of 2123 lesions analysed, 89 (4.2%) showed deep invasion.
-
Diagnostic accuracy of NICE 3 for predicting deep invasion was: Se = 58.4% (95% CI 47.5 – 68.8), Sp = 96.4% (95% CI 95.5 – 97.2), PPV = 41.6% (95% CI 32.9 – 50.8), and NPV = 98.1% (95% CI 97.5 – 98.7).
-
A conditional inference tree including all the variables showed that the NICE classification was the best predictor of deep invasion (p < 0.001). Further, statistically significant changes in the probability of deep invasion were found in NICE 3 pedunculated polyps (13%, p = 0.007) and NICE 3 non-pedunculated and ulcerated lesions (93%, p = 0.026). By contrast, deep invasion were more frequent in lesions NICE 1/2 if depressed areas (9.7%, p < 0.001) or nodular mixed type morphology (8.6% p < 0.001) was present, but exceptional in the remaining NICE 1/2 lesions (1.1% p < 0.001).
Conclusions:
-
The diagnostic accuracy of optical diagnosis is limited.
-
However, the NICE classification is the best known tool for predicting deep invasion, even without magnification and when used by non-expert endoscopists.
-
This classification could be improved by including certain morphologies and groß morphological malignant features.