IN-VIVO DETECTION AND DIAGNOSIS OF GASTRIC PRENEOPLASTIC LESIONS BY FOURTH-GENERATION ENDOCYTOSCOPY: A PILOT STUDY

Aims Endocytoscopy (EC) provides ultra-high magnification images and enables in-vivo histologic assessment of the mucosa. Aim of our study was to characterize features related to gastric precancerous lesions (GPL) and to assess diagnostic performances of EC both in antrum and corpus compared to standard histopathology.

Methods 80 gastric areas (36, antrum; 44, corpus) of 25 prospectively recruited patients were analyzed. All endocytoscopies were performed by one expert endoscopist (A) by using an Olympus GIF-H290EC-endocytoscope and double staining technique. Histopathology of target biopsies was used as gold standard. EC diagnosis was made as demonstrated in [ Figure 1 ].

Fig. 1

After independent EC-diagnosis 419 high-quality EC-images were selected by endoscopist A and reviewed by 3 endoscopists (B, with minor experience in EC, C and D, with no experience in EC but trained for images assessment), who were blinded to endoscopic findings and histopathology.

Results Gastric areas were histologically classified as shown in Figure 1.

The diagnostic performances of EC for GPL detection are summarized in the [ Table 1 ]. The mean sensitivity, specificity and accuracy for EC diagnosis of GPL (EC2a/2b/2ab+EC3 vs. EC1) were 88%, 93% and 91% among endoscopists A and B and 86%, 68% and 74% among endoscopists C and D, respectively. The interobserver agreement was substantial (κ-value 0.75, p<0.001) between the two EC experts and fair (κ-value 0.29, p=0.006) between the two non-expert endoscopists.

Table 1

EC 1	EC 2a	EC 2b	EC 2a/2b/2ab	EC 2+EC3
Endoscopist A: Sensitivity 98.08% (89.74–99.95); Specificity 96.43% (81.65–99.91); Accuracy 97.50% (91.26–99.70); Positive predictive value 98.08% (88.15–99.71); Negative predictive value 96.43% (79.47–99.47)	Endoscopist A: Sensitivity 64.29% (35.14–87.24); Specificity 100% (94.56–100.00); Accuracy 93.75% (86.01–97.94); Positive predictive value 100.00%; Negative predictive value 92.96% (86.73–96.38)	Endoscopist A: Sensitivity 84.00% (63.92–95.46); Specificity 98.18% (90.28–99.95); Accuracy 93.75% (86.01–97.94); Positive predictive value 95.45% (74.93–99.33); Negative predictive value 93.10% (84.60–97.07)	Endoscopist A: Sensitivity 96.00% (79.65–99.90); Specificity 98.18% (90.28–99.95); Accuracy 97.50% (91.26–99.70); Positive predictive value 96.00% (77.46–99.41); Negative predictive value 98.18% (88.78–99.73)	Endoscopist A: Sensitivity 96.55% (82.24–99.91); Specificity 98.11% (89.93–99.95); Accuracy 97.56% (91.47–99.70); Positive predictive value 96.55% (80.05–99.49); Negative predictive value 98.11% (88.34–99.72)
Endoscopist B: Sensitivity 90.38% (78.97–96.80); Specificity 78.57% (59.05–91.70); Accuracy 86.25% (76.73–92.93); Positive predictive value 88.68% (79.31–94.12); Negative predictive value 81.48% (65.16–91.19)	Endoscopist B: Sensitivity 35.71% (12.76–64.86); Specificity 98.48% (91.84–99.96); Accuracy 87.50% (78.21–93.84); Positive predictive value 83.33% (38.73–97.53); Negative predictive value 87.84% (83.00–91.44)	Endoscopist B: Sensitivity 64.00% (42.52–81.03); Specificity 92.73% (82.41–97.98); Accuracy 83.75% (73.82–91.05); Positive predictive value 80.00% (59.82–91.49); Negative predictive value 85.00% (76.97–90.57)	Endoscopist B: Sensitivity 80.00% (59.30–93.17); Specificity 90.38% (78.97–96.80); Accuracy 87.01% (77.41–93.59); Positive predictive value 80.00% (62.95–90.40); Negative predictive value 90.38% (81.03–95.39)	Endoscopist B: Sensitivity 78.57% (59.05–91.70); Specificity 88.68% (76.97–95.73); Accuracy 85.19% (75.55–92.10); Positive predictive value 78.57% (62.75–88.87); Negative predictive value 88.68% (79.29–94.13)
Endoscopist C: Sensitivity 61.54% (47.02–74.70); Specificity 96.43% (81.65–99.91); Accuracy 73.75% (62.71–82.96); Positive predictive value 96.97% (82.19–99.55); Negative predictive value 57.45% (48.72–65.73)	Endoscopist C: Sensitivity 57.14% (28.86–82.34); Specificity 92.42% (83.20–97.49); Accuracy 86.25% (76.73–92.93); Positive predictive value 61.54% (38.06–80.64); Negative predictive value 91.04% (84.69–94.92)	Endoscopist C: Sensitivity 64.00% (42.52–82.03); Specificity 80.00% (67.03–89.57); Accuracy 75.00% (64.06–84.01); Positive predictive value 59.26% (44.27–72.70); Negative predictive value 83.02% (74.04–89.34)	Endoscopist C: Sensitivity 84.00% (63.92–95.46); Specificity 72.22% (58.36–83.54); Accuracy 75.95% (65.02–84.86); Positive predictive value 58.33% (46.84–68.98); Negative predictive value 90.70% (79.64–96.05)	Endoscopist C: Sensitivity 96.43% (81.65–99.91); Specificity 61.54% (47.02–74.70); Accuracy 73.75% (62.71–82.96); Positive predictive value 57.45% (48.72–65.73); Negative predictive value 96.97% (82.19–99.55)
Endoscopist D: Sensitivity 75.00% (61.05–85.97); Specificity 75.00% (55.13–89.31); Accuracy 75.00% (64.06–84.01); Positive predictive value 84.78% (74.22–91.51); Negative predictive value 61.76% (49.06–73.04)	Endoscopist D: Sensitivity 57.14% (28.86–82.34); Specificity 80.30% (68.68–89.07); Accuracy 76.25% (65.42–85.05); Positive predictive value 38.10% (24.03–54.49); Negative predictive value 89.83% (82.66–94.24)	Endoscopist D: Sensitivity 40.00% (21.13–61.33); Specificity 100.00% (93.51–100.00); Accuracy 81.25% (70.97–89.11); Positive predictive value 100.00%; Negative predictive value 78.57% (72.70–83.47)	Endoscopist D: Sensitivity 68.00% (46.50–85.05); Specificity 78.18% (64.99–88.19); Accuracy 75.00% (64.06–84.01); Positive predictive value 58.62% (44.53–71.43); Negative predictive value 84.31% (74.90–90.64)	Endoscopist D: Sensitivity 75.00% (55.13–89.31); Specificity 75.00% (61.05–85.97); Accuracy 75.00% (64.06–84.01); Positive predictive value 61.76% (49.06–73.04); Negative predictive value 84.78% (74.22–91.51)

Conclusions New-generation EC has a potential to identify GPL with high diagnostic accuracy, high reliability and good reproducibility. Training and experience in performing EC affect substantially the diagnostic performances.

Publication History

Article published online:
14 April 2022