4 | DISCUSSION
Several endoscopic features of the gastric mucosa have been proposed for the diagnosis of H. pylori infection, such as RAC, diffuse or spotty redness, mucosal swelling, and nodular changes(1). Our study included 15 relevant studies(17-31) assessing the diagnostic accuracy of RAC for H. pylori, and the results suggest that RAC has a high diagnostic accuracy for H. pylori infection, with a pooled estimate of 0.98 (95% CI 0.95 to 0.99) for sensitivity and 0.75 (95% CI 0.54 to 0.88) for specificity.
Many studies have evaluated the diagnostic accuracy of RAC as a single endoscopic feature for H. pylori, with inconclusive results. A prospective study in a European population enrolling 140 adults found that the presence of RAC in the lesser curvature evaluated with high-definition endoscopy can help identify patients without H. pylori, with a sensitivity of 100% and a specificity of 48.96%(20). A prospective study in Korea involving 617 individuals concluded that the RAC pattern observed using standard endoscopy could predict H. pylori infection status, with a sensitivity of 93.3% and a specificity of 89.1%(19). Again, in a Brazilian cohort of 99 individuals, Machado et al. investigated the diagnostic accuracy of RAC for H. pylori, suggesting that RAC can provide a sensitivity of 96.9% and a specificity of 88.1% for diagnosing H. pylori infection(26).
In a meta-analysis from 2020 including studies published between 2002 and 2019, Li and colleagues concluded that RAC is a valuable endoscopic finding for predicting patients without H. pylori infection(37). However, this meta-analysis has substantial shortcomings in its quantitative data analysis for heterogeneity, with a reported I2 of 0 (95% CI: 0-100%). In contrast, our study found significant heterogeneity between studies, with an I2 of 99.95% (95% CI: 99-100%). Furthermore, we performed a univariable meta-regression and subgroup analysis to identify the potential source of heterogeneity between the studies. The results showed that whether patients with a history of H. pylori infection were excluded and whether consecutive enrollment was used in each study significantly affected the sensitivity of RAC, indicating that these two factors might partly contribute to the heterogeneity between studies.
Likelihood ratios and posttest probabilities are also of importance because they provide information regarding the likelihood that a patient with a positive or negative test has H. pylori infection or not. In our study, a positive likelihood ratio of 3.8 implies that a person with H. pylori infection is 3.8 times more likely to have a positive test result than is a healthy person. Likewise, a negative likelihood ratio of 0.03 indicates that a person without H. pylori infection is 33 times more likely to have a negative test result than is a person with H. pylori infection. Therefore, RAC can serve as a reliable marker for exclusion rather than confirmation of H. pylori infection.
As our results show, RAC is not a perfect endoscopic marker for the diagnosis of H. pylori. Although RAC has a high overall diagnostic accuracy for H. pylori infection, it is not a good predictor for confirming H. pylori infection because of its low positive likelihood ratio. The pathophysiological process of H. pylori infection is complex and can be affected by many factors, which probably contributes to the variations in endoscopic findings(38-41). Therefore, a single endoscopic feature of RAC might be insufficient for confirmation of H. pylori infection because of the relatively low positive likelihood ratio, thus highlighting the need for a combination of multiple endoscopic features to confirm H. pylori infection.
This study has limitations. First, we detected substantial heterogeneity between the studies and found that two of the study characteristics contributed to the observed heterogeneity. However, there are probably additional study characteristics that have impacted study heterogeneity but have not been addressed, such as sample size, geographic area, variation in the methods used for the diagnosis of H. pylori between studies, and variation in the anatomic location where endoscopic physicians observed the RAC, because these factors were challenging to quantitatively analyze. Second, we only included studies written in English, which might have led to selection bias, although the funnel plot did not reveal significant publication bias (p=0.37). Third, this systematic review did not include studies that assessed endoscopic features other than RAC that are possibly associated with H. pylori infection. Despite these limitations, this meta-analysis investigating the diagnostic accuracy of RAC for H. pylori infection is the largest and most comprehensive assessment to date.