Comparative Study On The Diagnostic Efficacy Of The Japan Narrow-band-imaging Expert Team And Pit Pattern Classifications For Colorectal Neoplastic Lesions

Objective:To evaluate and compare the diagnostic efficacy of the Japan Narrow-band-imaging Expert Team（JNET）classification and Pit pattern classification in the diagnosis of colorectal neoplastic lesions under magnifying endoscopy with narrow-band imaging（ME-NBI）,so as to better guide the endoscopic diagnosis and treatment.Methods:This study was a single central retrospective study.A total of 233 lesions in 217patients who underwent colonoscopy in magnifying endoscopy with narrow-band imaging（ME-NBI）mode and pathological examination from January 1,2019 to March 31,2022were collected,The endoscopic diagnosis results based on JNET classification and Pit pattern classification were compared with actual pathological results.The data were statistically analyzed using SPSS26.0 software.The measurement data of normal distribution were expressed as（?）±s,while the counting data were expressed by percentage and frequency.The reliability of the screening test was compared by Mc Nemar test,and the influencing factors of the diagnostic accuracy were analyzed by chi-square test.P<0.05was considered statistically significant.Results:1.A total of 233 lesions in 217 patients were included in this study.Among the 217patients,141 were male and 76 were female,aged 29 to 84 years,with a median age of 59years and a mean age of（59.54±10.11）years old.The locations were left hemicolon（descending colon to sigmoid colon）in 77 lesions,right hemicolectomy（cecum to transverse colon）in 77 lesions,and rectum in 79 lesions;the mean diameter of lesions was（13.79±9.88）mm（1 lesion was half a cycle）.There were 90 lesions with polyps<10mm and 143 lesions with polyps≥10mm.There were 65 single lesions and 168 multiple lesions.There were 187 lesions as protruded,6 lesions as flat,1 lesion as depressed and 22 lesions as lateral spreading tumor.The histological characteristics were non-neoplastic in 35lesions and neoplastic in 198 lesions.2.The accuracy of JNET classification in the diagnosis of neoplastic lesions and non-neoplastic lesions was 96.57%under ME-NBI,and there was no statistical significance in the effect of different gender,age,lesion diameter,lesion site,lesion morphology and lesion growth type on the accuracy of JNET classification in the diagnosis of neoplastic lesions（P>0.05）.3.In the diagnosis of neoplastic lesions and non-neoplastic lesions by JNET classification and Pit pattern classification,the sensitivity,specificity,accuracy,positive predictive value（PPV）and negative predictive value（NPV）were 99.49%,80.00%,96.57%,96.57%,96.55%and 97.98%,77.14%,95.28%and 96.04%,90.00%,respectively.The sensitivity,specificity,accuracy differences were not statistically significant.Different gender,lesion morphology（protruded,non-protruded）,lesion location（left hemicolon,right hemicolon,rectum）,lesion diameter（<10mm,≥10mm）and lesion growth type（single,multiple）had no statistical significance on the accuracy of diagnosis of neoplastic lesions and non-neoplastic lesions by JNET classification and Pit pattern classification（P>0.05）.4.Under ME-NBI,the total accuracy of each type of JNET classification and each type of Pit pattern classification was 73.82%and 87.55%,respectively,and the difference was statistically significant（P<0.01）.The sensitivity,specificity,accuracy,positive predictive value and negative predictive value of JNET type 2A and Pit pattern typeⅢ_L+Ⅳin the diagnosis of adenoma and low-grade intraepithelial neoplasia（LGIN）were 79.70%、82.00%、80.69%、85.48%、75.23%and 87.96%,84.00%,86.27%,87.97%,84.00%,respectively;The sensitivity,specificity,accuracy,positive predictive value and negative predictive value of JNET type 2B and Pit pattern typeⅤ_I-Lin the diagnosis of high-grade intraepithelial neoplasia（HGIN）and superficial submucosal invasive（SM-s）cancer were64.00%,77.05%,74.25%,43.23%,88.68%and 80.00%,91.80%,89.27%,72,73%,94.38%respectively;The sensitivity,specificity,accuracy,positive predictive value and negative predictive value of JNET type 3 and Pit pattern typeⅤ_I-H+Ⅴ_N in the diagnosis of superficial submucosal invasive（SM-d）cancer were 20.00%,98.62%,93.56%,50.00%,94.71%and73.33%,98.17%,96.57%,73.33%,98.17%,respectively.There were statistically significant differences in the specificity and accuracy of the diagnosis of HGIN/SM-s cancer between Pit pattern typeⅤ_I-L and JNET type 2B（P<0.05,P<0.05）,and in the sensitivity between Pit pattern typeⅤ_I-H+Ⅴ_N and JNET type 3 in the diagnosis of SM-d cancer（P<0.01）,while no statistically significant differences in the rest（P>0.05）.Conclusion:1.Under ME-NBI,JNET classification and Pit pattern classification have considerable and reliable diagnostic value in differentiating colorectal neoplastic lesions from non-neoplastic lesions,and in differentiating HP,adenoma,LGIN,and SM-d cancer.Moreover,JNET classification has a more direct guiding treatment strategy,which is worthy of popularization and application.2.For HGIN/SM-s cancer,the diagnostic value of JNET type 2B under ME-NBI is lower than that of Pit pattern type V_I-L.The quality of diagnosis and treatment should be further improved by means of endoscopic ultrasound.