| Aims: To investigate the diagnostic value of our independently developed autof-luorescence endoscopy in gastrointestinal malignancy and precancerous lesions; Toassess the real-time diagnostic value of magnifying endoscopy with narrow-band im-aging (ME-NBI) and confocal laser endomicroscopy (CLE) for gastric neoplastic le-sions using histopathology as a gold standard and propose an applicable clinicalstrategy for diagnoses; To evaluate the effect of a training program on the diagnosis ofesophageal neoplastic lesions by magnification endoscopy with narrow band imagingamong endoscopists of varying experience.Methods:1. Atotal of170patients’operation specimens in vitro and30patientsin vivo with definite or suspicious gastrointestinal malignancy from June2010toMarch2013in Renji Hospital, Shanghai Jiaotong University School of Medicine wereenrolled. Autofluorescence endoscopy was used to detect the autofluorescence imag-ing. Histopathological examination was considered as a golden standard. The sensi-tivity, specificity, accuracy, positive predictive value (PPV) and negative predictivevalue (NPV) were assessed.2. White-light endoscopy (WLE), ME-NBI and CLE were performed for thosepatients with suspected gastric superficial neoplastic lesions from January2013toJanuary2014in Renji Hospital, Shanghai Jiaotong University School of Medicine.The reading results by ME-NBI and by CLE were assessed for sensitivity, specificity,accuracy, PPV and NPV, as well as interobserver agreement using histopathology as agolden standard.3. We divided endoscopists into3groups based on different experience (NEE: non-experienced endoscopist; LEE: less-experienced endoscopist; HEE: high-ly-experience endoscopist). Then a2-hour training program about ME-NBI and in-tra-epithelial papillary capillary loop (IPCL) patterns was given to NEEs and LEEs.They evaluated the test images and made diagnosis both before and after training.Diagnostic accuracy and interobserver agreement of three modalities were estimated.Results:1. The overall accuracy, sensitivity, specificity, positive predictive value(PPV) and negative predictive value (NPV) of autofluorescence endoscopy in diag-nosing gastrointestinal malignancy and precancerous lesions were94.0%,94.6%,93.5%,92.6%and95.3%, respectively. The agreement between autofluorescence en-doscopy and histopathological examination was excellent (κ=0.88). In vitro, no sig-nificant difference in detection rate was found between early cancer (including pre-cancerous lesions) and advanced cancer (88.2%vs.95.8%, p=0.238), significant dif-ference in detection rate was found between differentiated cancer and undifferentiatedcancer (97.9%vs.86.8%, p=0.009). In vivo, no significant difference in detection ratewas found between differentiated cancer and undifferentiated cancer (94.1%vs.80%,p=0.373).2. Finally,82patients with86gastric lesions were enrolled. Overall,40neoplas-tic lesions were identified. The accuracy, sensitivity and specificity for ME-NBI were93.75%,91.67%and95.45%, compared with91.86%,90%and93.48%, respectively,for CLE (all p＞0.05). When ME-NBI in conjunction with CLE, the sensitivity couldbe improved to94.44%. For undifferentiated gastric carcinoma, CLE had a higheraccuracy than ME-NBI (81.25%vs.73.33%, p＞0.05). Both of the interobserveragreements were almost perfect (ME-NBI, κ=0.87; CLE, κ=0.84).3. The diagnostic accuracies of ME-NBI for differentiating esophageal neoplasticlesions were significantly improved in non-experienced endoscopist groups (84.2%vs.47.4%) and less-experienced (92.8%vs.55.9%), and kappa (κ) values for both groupsachieved good agreement after training (NEE,0.68vs.0.24; LEE,0.76vs.0.43), allp<0.05. As for LEE, the diagnostic accuracy could be increased to be equivalent tothat of HEE (93.4%).Conclusion:1. Autofluorescence endoscopy can be used to diagnose gastroin- testinal malignancy and precancerous lesions with high value. It may become a prom-ising diagnostic modality for the early detection of gastrointestinal malignancy.2. ME-NBI had a similar diagnostic yield with CLE for gastric neoplastic lesionswith high validity and reliability. In combination these two methods will be useful fordiagnosing gastric indeterminate or uninterpretable lesions.3. ME-NBI could be learnt easily and rapidly by beginners in short time. For di-agnosis of esophageal neoplastic lesions, it could improve the diagnostic skill ofless-experienced endoscopist to be equivalent to that of highly-experienced endos-copist, in other words, ME-NBI has a short learning curve for diagnosing esophagealneoplastic lesions using IPCL patterns classification. |
PDF Full Text Request