Application of narrow-band imaging without magnifying in diagnosis of colorectal lesions by NICE classification / 上海交通大学学报（医学版）

ABSTRACT

Objective: To evaluate narrow-band imaging (NBI) without magnifying in the diagnosis of colorectal lesions by NBI International Colorectal Endoscopic Criteria (NICE classification), and analyze the safety and practicability of "do-not-resect" and "resect and discard" policies in clinical practice. Methods: The patients undergoing screening or surveillance colonoscopy, who were found colorectal lesions in the examination, from May to December in 2017 were enrolled. All the patients were examined by NBI without magnifying by any of the designated two physicians. NICE classification was used to diagnose colorectal lesions, and the diagnostic confidence of each lesion was recorded. The results of endoscopy were compared with those of pathology, and the accuracy rate and the confidence rate of diagnosis were calculated. The sensitivity, specificity, positive predictive value and negative predictive value of the diagnostic method for differentiating superficial tumors from non-tumors were also calculated. Finally, the feasibility, safety and cost savings of using "do-not-resect" and "resect and discard" policies in clinic were analyzed. Results: A total of 764 lesions were detected in the 636 enrolled patients. The overall accuracy of NICE classification was 84.95% and the diagnostic confidence rate was 81.68%. The sensitivity, specificity, positive predictive value and negative predictive value for differentiating tumors from non-tumors were 91.77%, 67.68%, 88.69%, and 74.86%, respectively. The diagnostic accuracy of diminutive colorectal lesions (≤5 mm) with high confidence was 94.98%, and the negative predictive value of diminutive rectosigmoid lesions (≤5 mm) with high confidence was 96.25%. They achieved the criteria of "resect and discard" and "do-not-resect" policies. If "do-not-resect" and "resect and discard" policies had been executed in clinical practice, ¥165 490 could have been saved and the omission diagnostic rates of "do-not-resect" and "resect and discard" policies would have been 3.75% and 0, respectively, in this study. Conclusion: It is feasible to use NBI without magnifying in differentiating tumors from non-tumors. The diminutive colorectal lesions and rectosigmoid lesions with high diagnostic confidence may achieve the criteria of "resect and discard" and "do-not-resect" policies, respectively.