The role of circRNAs in the occurrence and progression of HCC has been confirmed. CircRNAs have been associated with several pathways, such as the PI3K/AKT/mTOR pathway [18], Wnt pathway and JAK/STAT pathway [14, 19]. In addition, several PI3K/AKT/mTOR pathway-related circRNAs have been identified to be abnormally expressed in different cancer progression. Hence, PI3K/AKT/mTOR pathway-related circRNAs may be potential biomarkers for HCC [11]. The current study assessed the role of PI3K/AKT/mTOR pathway-related circRNAs in the diagnosis and prognosis of HCC patients.
4.1. Pooled biomarker efficacy of circRNAs in HCC
We extensively explored the diagnostic and prognostic value of PI3K/AKT/mTOR-associated circRNAs as potential biomarkers for HCC based on 25 articles. We found that the pooled sensitivity of abnormal circRNAs for the diagnosis of HCC was 80%, the specificity was 83%, and the pooled HR for the prediction of the OS of HCC patients was 1.49. The HR of the high expression group was 2.90. These results indicate that PI3K/AKT/mTOR-associated circRNAs not only have high diagnostic efficiency but also good prognostic prediction efficiency and are potential biomarkers for HCC.
4.2. Diagnostic efficacy of circRNAs in HCC
Our study also showed that the pooled sensitivity of all circRNAs was 80% and the pooled specificity was 83%. Currently, AFP is one of the most recommended biomarkers for HCC. According to the 2021 Chinese expert consensus on the early screening strategy for liver cancer, the sensitivity and specificity of AFP in the diagnosis of liver cancer were 25–65% and 80–94%, respectively. Our results suggest that dysregulated circRNAs may have higher sensitivity and specificity than AFP in distinguishing HCC patients, with the sensitivity of HCC diagnosis increasing from a maximum of 65–80%. Considering that there are over 900000 new cases of liver cancer worldwide each year, more than 135000 patients will be correctly diagnosed. At the same time, considering the high mortality rate of HCC, more patients may survive because of the correct diagnosis, indicating that circRNA diagnosis of HCC has a good prospect in clinical application. Compared with previous studies of circRNAs in the diagnosis of HCC [49], the sensitivity of our results was not significantly different (0.82 vs. 0.80), but the specificity was significantly different (0.72 vs. 0.83). This indicates that the selected PI3K/AKT/mTOR pathway-related circRNAs have higher specificity in the diagnosis of HCC, which is one of the important reasons for our selection of pathway-related circRNAs, which has also been confirmed herein. Compared with another circRNA study [50], our study showed significantly higher sensitivity and specificity, further confirming the reasons for their selection.
The expression of circRNAs is tissue-specific, and thus we explored whether there are differences in the diagnostic efficacy of different tissue-derived circRNAs for HCC. Interestingly, there was a difference in the sensitivity and articles between HCC tissue and serum circRNAs. The diagnostic specificity of circRNAs for HCC is not limited by tissue type, and the reason for the difference in sensitivity may involve the influence of threshold setting and different tissue expression levels. Given the false positivity of AFP in some benign lesions and between liver diseases [10], we also explored the diagnostic performance of different controls, and encouragingly, our results show that circRNAs have a higher specificity contributing to a lower false-positive rate when confronted with healthy populations and benign lesions, which also demonstrates the superiority of circRNAs as diagnostic markers for AFP.
4.3. Prognostic efficacy of circRNAs in HCC
In addition to its good diagnostic efficacy, our circRNAs also exhibited a good prognostic effect in HCC patients. The HR of AFP for the OS of HCC was 0.55 [51], while the HR of down-regulated and up-regulated circRNAs for subgroup analysis was 0.53 and 2.90, respectively, both possessing better predictive value than AFP. Our circRNAs also had a better predictive value than that of previous studies (1.51) [52]. Similarly, we conducted a further analysis based on different tissue sources, but due to the limitation of the included articles, we finally only showed that circRNAs in liver cancer tissues had a better predictive value than the summary results, and more studies on circRNAs in HCC serum are needed in the future.
4.4. Potential clinical value of circRNAs
All the included studies were performed in China and showed good efficacy. Meanwhile, combined with the current situation of a high incidence of HCC and the high disease burden of late diagnosis in China[53], PI3K/AKT/mTOR pathway-associated circRNAs have a good clinical prospect for both diagnosis and prognosis of HCC. Our findings confirmed this proposition in which PI3K/AKT/mTOR pathway-related circRNAs were found to be better biomarkers for HCC. However, whether other pathway-related circRNAs have similar effects warrants further exploration. According to our consideration, the greatest advantage of biomarkers is that they are easy to detect, and the detection samples are particularly easy to collect. Therefore, more studies on circRNAs should be conducted using blood samples to validate our results.
In addition studies have also shown that circRNAs also play a good role in more diseases, for example in silicosis and idiopathic pulmonary fibrosis circ0058493 can also be used as a potential biomarker[54]. Not only as a marker, the involvement of circRNAs in the disease process may provide other options for the treatment of diseases. circ0027791 can regulate the immune escape in hepatocellular carcinoma[55], which may provide other entry directions for the treatment of HCC. the involvement of circRNAs in different processes in different diseases has attracted us to study and discuss circRNAs continuously, For example, Circ-EIF3I regulates the activity of Wnt/β-catenin pathway in lung cancer[56], and there are different findings in liver cancer[57]. In prognosis we also found that the expression level in patients may predict the role of circRNAs in cancer, but the exact mechanism remains to be studied more.
Nonetheless, our methodological quality assessment showed that diagnostic studies have poor methodological quality in patient selection (with a dilemma of whether to avoid case-control studies or include continuous/randomized cases) and index trials (with a dilemma of whether to evaluate the results of the experimental interpretation without knowing the gold standard results and/or determine in advance if the threshold is used), which affects the reliability of reporting and therefore deserves considerable attention.
4.5. Strengths and limitations of the study
The present study has certain strengths. First, we established a well-designed search strategy and inclusion criteria and included articles in all languages and periods with good representation. Secondly, we analyzed both the diagnostic and prognostic efficacies of circRNAs in HCC patients and explored the role of circRNAs in HCC patients in diverse aspects. Third, we targeted PI3K/AKT/mTOR pathway-related circRNAs to screen out circRNAs that are more closely related to HCC for a clearer mechanism, better aggregation and better prospects for clinical transformation. Fourth, we conclude with a QUADAS-2 and NOS-based evaluation of the included studies, with suggestions for common absences in the articles.
However, this study has several limitations. First, the limited number of articles did not allow for comprehensive subgroup analyses based on different factors. Second, all included studies were conducted in China, which may have potential population bias, and it is unclear whether these findings can be generalized to other populations. This seems to be a common phenomenon in the field of circRNA research, with almost all clinical studies coming from China and almost none in European or American populations. For example, in the study by Xiao Zhang et al, 23 papers were included, only one from India and the rest from China[58]. Similarly Xianglin Yang's study included 33 papers, all from China[59]. Additional multicenter studies in diverse populations are required to verify whether these results are equally applicable to other races.Third, there is a large heterogeneity between studies owing to variations in molecular types and sample types of circRNAs. Fourth, all circRNAs expression data in the eligible studies were obtained from serum or tissue, with a paucity of data on exosomal circRNAs, which are crucial biomarkers for multiple cancers, demanding considerable attention in future studies. Fifth, most diagnostic studies use tissues adjacent to the patient’s tumor as controls, rather than being blindly designed, which may lead to reduced applicability of the diagnostic value. Finally, the relatively small number of patients may generate insufficient statistical power; therefore, the correlation between aberrant expression levels of PI3K/AKT/mTOR pathway-related circRNAs and these clinical values should be carefully interpreted and validated in future large-scale studies.