Hesitate between confocal laser endomicroscopy and narrow-band imaging: how to choose a better method in the detection of focal precancerous state of gastric cancer.

Background: With a high incidence globally, deaths form gastric cancer (GC) are not rare. Early diagnosis is crucial to ameliorate its prognosis. Confocal laser endomicroscopy (CLE) and narrow band imaging (NBI) have been extensively applied in gastroscopy, particularly when it comes to the detection and management of premalignant gastric lesion. Our meta-analysis intends to appraise the diagnostic capability and compare the efficacy of NBI and CLE for focal precancerous state of gastric cancer.

Methods: We performed a literature search up to November 5, 2020 in online databases and major conferences. Two investigators assessed the methodological bias by QUADAS-2, followed by sophisticated study selection and data exaction to make a comparison between sensitivity, specificity, positive and negative likelihood values, and diagnostic odds ratio. A symmetric summary receiver-operating curve (sROC) and its area under the curve (AUC) were used to estimate threshold effect. Additionally, we evaluated the publication bias by Deeks' asymmetry test.

Results And Conclusions: Four studies involved 248 patients and 526 lesions. In analysis drawn from every lesion, the NBI's pooled sensitivity and specificity were 87% (95% CI: 0.80-0.92) and 85% (95% CI: 0.75-0.91), and those of CLE were 90% (95% CI: 0.85-0.91) and 87% (95% CI: 0.83-0.91). CLE illustrated that the pooled two were slightly higher than NBI when compared at the level of every lesion. The AUC for NBI and CLE was 0.92 (0.90-0.94) and 0.95 (0.92-0.96), and there might be a threshold effect, according to the shoulder-like distribution of scatter points in the sROC. We did not find obvious publication bias in our meta-analysis.