Assessing diagnostic accuracy of circulating microRNAs as biomarkers for ovarian cancer: a systematic review and meta-analysis

Abstract

Ovarian cancer is one of the deadliest gynecological malignancies, mainly because of its silent development and the lack of effective conventional methods for diagnosis. Because of their extraordinary stability in bio fluids and expression signatures that specifically correlate with tumors, circulating microRNAs (miRNAs) have received much attention as non-invasive diagnostic biomarkers. This meta-analysis evaluates the diagnostic value of circulating miRNAs for early ovarian cancer. A systematic literature review was performed in line with PRISMA-DTA guidelines in PubMed, EMBASE, and Web of Science until March 8, 2025. A total of 24 studies were eligible. Pooled diagnostic metrics (sensitivity and specificity, likelihood ratios, and diagnostic odds ratio (DOR)) were calculated based upon the bivariate random effect model. Summary receiver operating characteristic (sROC) curves were generated and subgroup and meta-regression analyses performed to investigate heterogeneity. Sensitivity analyses were performed to assess the strength of the pooled estimates. The combined sensitivity, specificity and DOR was 0.749 (95% CI: 0.702–0.791), 0.748 (95% CI:0.695–0.794) and 8.403 (95% CI: 6.308–11.194) respectively, which means a moderate-to-high diagnostic efficacy. Screening efficacy of serum-based and PBMC-based tests differed significantly at the DOR of 10.55 being maximum amongst serum assays. Biospecimen and regional differences contributed substantially to heterogeneity. The results were robust according to sensitivity analyses, and there was no obvious publication bias based on Deeks’ test (p = 0.08). Circulating miRNAs, especially serum-based profiles, are highly promising as non-invasive diagnosis in ovarian cancer. Standardization of protocols and prospective validation is necessary for clinical application.

Supplementary Information

The online version contains supplementary material available at 10.1186/s13048-026-02067-0.

Simple summary

Ovarian cancer is one of the deadliest cancers affecting women because it develops silently and current diagnostic methods are not effective at detecting it early when treatment is most successful. This research examines whether tiny molecules called microRNAs that circulate in the blood could serve as better diagnostic tools. MicroRNAs are naturally stable in blood and have specific patterns that change when cancer is present, making them promising candidates for a simple blood test. The authors combined data from 24 different studies to determine how accurately these blood-based microRNA tests can identify ovarian cancer patients. Their analysis shows that microRNA tests, particularly those using blood serum samples, can correctly identify about 75% of ovarian cancer cases while avoiding false alarms in healthy individuals at similar rates. This suggests that microRNA blood tests could become valuable tools for earlier ovarian cancer detection, potentially saving lives through improved diagnosis. However, standardized testing procedures and larger validation studies are needed before these tests can be used routinely in medical practice.

Introduction

Ovarian cancer is one of the most aggressive gynecologic cancers known, which may present itself insidiously with nonspecific symptoms and often occurs at advanced stages with a poor prognosis and a relatively high mortality rate [1–3]. The morbidity caused by this disease is staggering, with more than an estimated 300,000 new cases and a similar mortality rate each year [4]. Standard diagnostic and therapeutic modalities used for other cancers are nowhere near adequate when it comes to ovarian cancer; therefore, the need for a new strategy such as using liquid biopsy to make a screening test a reality is paramount [5–7]. Although early detection strongly predicts patient survival, the shortcomings of conventional diagnostic approaches, namely pelvic examination, transvaginal ultrasound and serum biomarkers such as cancer antigen 125 (CA-125) are inadequate, largely because they have insufficient sensitivity and specificity for detecting the disease at an early stage [8, 9].

In past decades, the discovery of the role of microRNAs (miRNAs) as powerful regulatory molecules involved in carcinogenesis has revealed potential ways to improve the diagnosis and early detection of ovarian carcinoma [10, 11]. MiRNAs are small, non-coding RNA molecules usually consisting of 18–25 nucleotides that are known to post-transcriptionally regulate gene expression by binding to complementary sequences in target mRNAs [12]. Collectively via these interactions, miRNAs mediate critical cellular activities such as proliferation, apoptosis, differentiation, and metastasis. Abnormal miRNA expression profiles have been reported in many cancers, and miRNAs are recognized as key players in tumor initiation, progression and chemo resistance [13, 14]. Specific miRNA signatures have been consistently defined in the setting of ovarian cancer supporting their role as non-invasive biomarkers. Distinctive miRNA signatures in ovarian cancer have been repeatedly characterized in different studies, alluding to their potential as non-invasive biomarkers. A number of studies have been conducted highlighting the differential expression of specific miRNAs between patients and healthy controls, demonstrating their potential for diagnostic improvement. Some miRNAs, for example, miR-21, miR-200 family, miR-125b, and miR-145 have been described as having altered expression in ovarian cancer associated with prognosis, subtype classification, and clinical parameters [15–17]. While these results are encouraging, differences in miRNA identified, methods of detection, and source of samples (serum, plasma, tissue and exosomes) and patient populations have made firm conclusions difficult to establish.

It is precisely the stable presence of miRNAs in such biological fluids coupled with their relative resistance to degradation by endogenous RNases that makes them suitable as diagnostic biomarkers for ovarian cancer, that can be detected and quantified easily in body fluids by minimally invasive liquid biopsy methods [18, 19]. However, a comprehensive summary of existing evidence using strong meta-analytic methods is critical for establishing their clinical utility, determining sensitivity and specificity cutoffs, and informing future research efforts.

A meta-analysis is a powerful statistical method to pool data from multiple independent studies that affords greater statistical power and generalizability through the systematic synthesis and quantification of the aggregate diagnostic efficacy of candidate miRNAs. This method enables assessment of sources of heterogeneity and examination of methodological variability capable of affecting biomarker efficacy. Such robust evaluation can offer valuable perspective on the clinical relevance of miRNAs, allowing for in-depth investigation of the best candidates for both clinical practice adoption and utilization as adjunctive diagnostic tools [20–22].

The present meta-analysis is designed to systematically evaluate and summarize the diagnostic performance of miRNAs as biomarkers for ovarian cancer in order to address the gaps and discrepancies among previous studies. This analysis attempts to integrate published information to define the leading miRNA candidates, diagnostic thresholds, and the clinical significance and reliability of these biomarkers. This complete assessment will finally enhance early detection, facilitate clinical decision-making, and potentially improve survival for women with ovarian cancer, thus hoping to make a significant contribution to clinical oncology.

Methods

Data sources and reporting guidelines

This DTA (diagnostic test accuracy) meta-analysis was performed according to the PRISMA-DTA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses of Diagnostic Test Accuracy Studies) guidelines [23]. The review protocol was registered in the International Prospective Register of Systematic Reviews (PROSPERO; Registration No. CRD420261327816).The registered protocol is available at https://www.crd.york.ac.uk/PROSPERO/view/CRD420261327816. (accessed on 27 February 2026). A systematic and methodologically rigorous search strategy was developed to identify relevant studies up to 8 March 2025. A comprehensive electronic database search was conducted, covering repositories in English language including MEDLINE (via PubMed) and EMBASE (via Ovid) and Web of Science due to their comprehensive coverage of biomedical literature relevant to diagnostic studies. To ensure comprehensive coverage, manual searches were also conducted by screening the reference lists of relevant systematic reviews, meta-analyses, and key primary studies identified during the electronic database search. This supplementary manual search was aimed at identifying any additional eligible studies that may have been missed by the initial automated search process. The search strategy used controlled vocabulary (i.e., MeSH and Emtree terms) and unrestricted keyword searches. These terms were chosen deliberately to encompass studies relevant to the diagnostic accuracy of circulating miRNAs for ovarian cancer. The general search terms were: (“miRNAs” OR “microRNAs” OR “miR*”) AND (“ovarian cancer”) AND (“diagnostics” OR “ROC curve” OR “sensitivity” OR “specificity”). Two independent researchers (MA and FI) iteratively refined the search strategy. For detailed search strategy and lists all databases included in the search and the rationale for their inclusion refer to (Supplementary Table 1).

Review question and PECO framework

The review question was structured according to the PECO framework to ensure methodological clarity and alignment with diagnostic test accuracy review standards. Population (P): Women with suspected or histologically confirmed ovarian cancer.Exposure (E): Circulating microRNA (miRNA) assays measured in serum, plasma, or peripheral blood mononuclear cells (PBMCs).Comparator (C): Healthy controls or individuals with benign gynecological conditions.Outcome (O): Diagnostic accuracy measures including sensitivity, specificity, positive likelihood ratio (PLR), negative likelihood ratio (NLR), diagnostic odds ratio (DOR), and summary receiver operating characteristic (sROC) curve analysis.

Inclusion and exclusion criteria for studies

Accordingly, studies were considered eligible if they satisfied the following inclusion criteria: (1) original research investigating the diagnostic accuracy of circulating miRNAs for the detection of ovarian cancer; (2) classification of histological subtypes is given would be taken into account; (3) availability of sufficient quantitative data to construct a 2 × 2 contingency table (TP, FP, FN, TN); and (4) peer-reviewed original studies with available full text. The exclusion criteria included systematic reviews, non-original articles, studies in animals, and meeting abstracts.

Data extraction

A standardized data extraction form was developed a priori. Data extraction was performed independently and in duplicate by two reviewers. Discrepancies were resolved through discussion and consensus, with adjudication by a third reviewer when necessary. Extracted data included the first author, year of publication, country, study design, miRNA detection methods, specimen type, and diagnostic accuracy measures. The accuracy of all extracted data was cross-validated (Supplementary Table 2).

Quality assessment

Included studies were critically appraised using an adapted version of the QUADAS-2 tool, which was designed to assess risk of bias in diagnostic studies; it focuses on four domains: patient selection, index test, reference standard, and flow and timing. Bias risk and applicability concerns were assessed for each domain. Quality assessment was performed by two independent reviewers, and discrepancies were solved by a third reviewer. Reviewers received prior training and had expertise in systematic reviews and diagnostic test accuracy meta-analysis. The QUADAS-2 framework was initially pilot tested to maintain inter-reviewer consistency and methodological coherence prior to formal assessment. Each domain was rated as “low,” “high,” or “unclear” risk of bias and concern for applicability. Visual summary plots were generated using RevMan 5.4.1 (Cochrane Collaboration, Oxford, UK).

Statistical analysis

Statistical analyses were conducted utilizing OpenMetaAnalyst (open-source, cross-platform software for advanced meta-analysis), Stata (v.17.0; StataCorp, USA) and RevMan 5.4.1 (Cochrane Collaboration, Oxford, UK). Using a bivariate random-effects model to account for inter-study variability, pooled estimates were derived for each diagnostic accuracy measure (sensitivity, specificity, positive likelihood ratio (PLR), negative likelihood ratio (NLR), diagnostic odds ratio (DOR)) with corresponding 95% confidence intervals. Cochran’s Q statistic and the I2 metric were used to quantitatively assess heterogeneity. In cases where heterogeneity was substantial (P < 0.10, I2 > 50%), subgroup analyses and meta-regression were employed to elucidate potential sources. Where applicable, sensitivity analyses were performed to assess the robustness of the pooled estimates. Deeks’ funnel plot asymmetry test was systematically used to evaluate publication bias.

Results

Included studies

We identified a total of 752 records from database searches (PubMed = 118; Web of Science = 179; and Embase = 455). A total of 540 records were screened on the basis of the titles and abstracts after removing 212 duplicates. Thus, 100 articles with full text were further assessed for eligibility and finally 24 studies were selected for the meta-analysis [24–47]. Reasons for exclusion at the full text stage were irrelevant study design (n = 31), wrong population (n = 20), no relevant outcome (n = 15), low-quality data (n = 6), and duplicate publication (n = 4). The process of study selection is shown through the PRISMA flow diagram (Fig. 1).

Fig. 1.

PRISMA flow chart

Characteristics and quality assessment of studies

Twenty-four studies covering different methodological designs assessing the diagnostic accuracy of the index tests were included. The QUADAS-2 tool was used to appraise the quality of these studies, in 4 domains, including patient selection, index test, reference standard, and flow and timing.

Risk of bias

The highest risk of bias was identified in the patient selection domain. This was primarily attributable to the frequent use of case–control study designs and non-consecutive sampling approaches. Such designs may artificially inflate diagnostic accuracy because cases and controls are often selected from clearly defined and contrasting groups rather than reflecting real-world clinical populations. In contrast, most studies demonstrated low risk of bias in the index test, reference standard, and flow and timing domains. This indicates that circulating miRNA assays were generally conducted using appropriate procedures, interpreted without knowledge of reference standard results, and applied within a clinically reasonable timeframe. Overall, while laboratory execution appeared methodologically sound, concerns related to patient selection may have contributed to overestimation of pooled diagnostic performance.

Applicability concerns

Applicability concerns evaluate whether the included studies are sufficiently aligned with the intended clinical use of circulating miRNAs as diagnostic biomarkers for ovarian cancer. Overall, concerns regarding applicability were judged to be low across most domains. The index test domain demonstrated low concern in all studies, indicating that circulating miRNA assays were relevant to the review question and reflective of clinical diagnostic approaches. Similarly, the reference standard domain showed low concern, as most studies used appropriate diagnostic confirmation methods such as histopathological verification.However, minor concerns were observed in the patient selection domain. Several studies employed case–control designs or included highly selected populations (e.g., clearly defined cancer cases versus healthy controls), which may not fully represent real-world clinical diagnostic settings where patients often present with non-specific symptoms. These differences may affect the generalizability of findings to routine clinical practice. A summary of overall risk of bias and applicability concerns is presented in Supplementary Fig. 1.

Diagnostic performance of circulating miRNAs

The reliability of circulating miRNAs as diagnostic biomarkers for case-control discrimination was moderate to high in the meta-analysis. The summary estimation of sensitivity and specificity were 0.749 (95% CI: 0.702–0.791) and 0.748 (95% CI: 0.695–0.794), respectively, corresponding to a balanced diagnostic performance. In addition, the positive likelihood ratio (LR⁺) and negative likelihood ratio (LR⁻) for circulating miRNAs were 2.635 (95% CI…) and 0.335 (95% CI…), respectively, further indicating that circulating miRNAs were of moderate rule-in and rule-out power. The diagnostic OR (DOR) of 8.403 (95% CI 6.308–11.194) also attests to the overall discriminating ability. The pooled diagnostic estimates or Combined Diagnostic Meta-Analysis Results are shown in (Supplementary Table 3). The plots of the LR⁺, LR⁻ and DOR (Figs. 2 and 3) demonstrate the distribution and consistency of effect sizes among the studies incorporated in the meta-analysis. Statistical heterogeneity was marked however, with I² values of 80.1% for LR⁺, 85.6% for LR⁻ and 81.9% for DOR, all (p < 0.001), indicating strong between-study variability. This negative correlation between sensitivity and specificity (r = − 0.2709) suggests threshold effects, which are often observed with diagnostic test accuracy studies. The summary receiver operating characteristic (sROC) curve (Fig. 4) showed a concentration of studies in the upper-left quadrant, indicating a good trade-off between sensitivity and specificity. The sROC curve also highlights the diagnostic capability of circulating miRNAs in all different study designs.

Fig. 2.

NLR and PLR forest plot

Fig. 3.

Diagnostic odds ratio forest plot

Fig. 4.

Summary Receiver Operating Characteristic curve (sROC)

Subgroup analysis and heterogeneity

In order to explore the possible effect of biospecimen type on the diagnostic value, we performed subgroup analysis on studies using serum, plasma and PBMCs as biological matrix for detection of circulating miRNAs (Supplementary Fig. 2 and Table 4). The pooled DOR were the largest with studies based on serum (DOR = 10.55, 95% CI: 7.51–14.82), indicating a highly effective diagnostic ability in this subgroup. AD: Studies conducted in plasma also showed a good discriminatory capability (DOR = 7.91, 95% CI: 5.24–11.92), while studies performed with PBMCs exhibited a relatively modest diagnostic performance (DOR = 2.37, 95% CI: 1.64–3.42). All subgroup effects were statistically significant (p < 0.001), indicating biospecimen type could be an effect modifier.

Heterogeneity was still observed in every subgroup, and serum studies demonstrated the highest heterogeneity (I² = 76.83%) compared with PBMCs (I² = 59.37%) or plasma (I² = 55.34%). The heterogeneity of the overall analysis was also found to be significant (I² = 81.94%, p < 0.001), suggesting that there was substantial between-study variation. These findings suggest that the diagnostic performance of circulating miRNAs is significantly influenced by the type of biospecimen used. Serum appears to offer the most reliable matrix, followed by plasma, while PBMCs show more limited diagnostic utility. This subgroup analysis underscores the importance of biospecimen selection in miRNA-based diagnostics and may help explain part of the observed heterogeneity in the overall meta-analysis. A stage-specific subgroup analysis was not performed because most included studies reported aggregated diagnostic accuracy data across mixed FIGO stages without providing stage-stratified 2 × 2 contingency tables. Accordingly, the pooled estimates represent overall diagnostic performance across combined-stage populations.

Sensitivity analysis, publication bias, and meta-regression analysis

Sensitivity analysis supported the reliability of the pooled diagnostic performance estimate, as reanalysis by omitting studies one by one at a time showed that the diagnostic odds ratio (DOR) of the combined estimate remained tightly ranged around the original pooled estimate (8.403, 95% CI: 6.308–11.194), and p < 0.001 in all included studies. The low level of between-study variation in DORs and SEs (Supplementary Fig. 3A) highlights the robustness and trustworthiness of the meta-analyses results. Embracing Deeks’ funnel plot, no strong indication of publication bias was found as the slope of the regression line was insignificantly strong (p = 0.08), however caution is warranted due to the borderline (Supplementary Fig. 3B). A meta-regression analysis was conducted in order to investigate sources of heterogeneity, showing that diagnostic performance was significantly impacted by biospecimen type and geographic region. In particular, PBMC-based analyses had a significantly poorer diagnostic performance compared to serum (coefficient = − 1.560, p = 0.030), whereas plasma showed no significant difference. Regarding study region, studies from Australia had a significantly lower DORs in comparison to those from China (coefficient = − 1.344, p = 0.010; for Germany and Japan the results were close to the significance (p = 0.055 and 0.082, respectively). There was no statistically significant difference between any of the epithelial or histological subtypes of cancer. The meta-regression showed statistical significance (p < 0.001), implying that the covariates included explained part of the detected heterogeneity (total I² = 81.94%) (Supplementary Table 5).

Discussion

This comprehensive diagnostic test accuracy meta-analysis, encompassing 24 studies across diverse populations and methodological approaches, demonstrates the considerable potential of circulating miRNAs as non-invasive diagnostic biomarkers for ovarian cancer. The aggregate diagnostic performance metrics (sensitivity, 0.749; specificity, 0.748; diagnostic odds ratio, 8.403) substantiate moderate-to-high discriminative capability, particularly evident in serum-based detection platforms. From a clinical perspective, the pooled sensitivity (0.749) and specificity (0.748) observed in this meta-analysis indicate moderate diagnostic accuracy. While these values are comparable to CA125 in certain settings particularly in advanced-stage disease, they may not be sufficient, to support the use of circulating miRNAs as standalone diagnostic tools, especially for population-level screening. Current diagnostic algorithms typically rely on combinations of serum CA125 measurement and transvaginal ultrasonography, which provide established clinical benchmarks. Therefore, circulating miRNAs are more appropriately positioned as adjunctive biomarkers that may enhance risk stratification or complement existing modalities rather than replace them. Future research should focus on multimodal diagnostic strategies integrating miRNA signatures with CA125 and imaging-based approaches to improve overall diagnostic performance and clinical utility.

Several recent studies, including those conducted by well-established research groups, have evaluated multi-miRNA classifiers or machine learning–based panels demonstrating high discriminative performance (AUC > 0.9), particularly for early-stage ovarian cancer detection. However, many of these studies did not provide sufficient 2 × 2 contingency data required for inclusion in diagnostic test accuracy meta-analysis. Therefore, such classifier-based approaches were not pooled in the present analysis. Nonetheless, these findings suggest that integrative multi-marker models may enhance diagnostic performance beyond single-miRNA assessments and warrant further prospective validation.

The robustness of our pooled estimates, corroborated through comprehensive sensitivity analyses, underscores both the methodological rigor of included investigations and the stability of the observed diagnostic effect. Notably, no individual study exerted disproportionate influence on the combined results, indicating substantial internal consistency across the evidence base. The absence of significant publication bias, as evidenced by Deeks’ funnel plot analysis (p = 0.08), further strengthens the validity of our conclusions, though the marginally significant p-value necessitates cautious interpretation regarding potential small-study effects.

Subgroup and meta-regression analyses revealed critical sources of heterogeneity that provide clinically actionable insights. Biospecimen type emerged as a pivotal moderating factor, with serum-based assays demonstrating superior diagnostic performance compared to peripheral blood mononuclear cell-based approaches (DOR = 10.545), as substantiated by meta-regression analysis (β = −1.560, p = 0.030). This finding emphasizes the fundamental importance of biospecimen selection in relation to miRNA extraction efficiency, analytical stability, and assay reproducibility considerations essential for standardizing clinically viable diagnostic platforms.

Geographic variation represented another significant source of heterogeneity, with Australian studies exhibiting markedly lower diagnostic performance relative to Chinese investigations (β = −1.344, p = 0.010). These disparities likely reflect regional differences in sample handling protocols, population-specific disease prevalence, and potentially distinct miRNA expression profiles, collectively highlighting the imperative for internationally harmonized methodological guidelines. Conversely, ovarian cancer subtype classification demonstrated no appreciable impact on diagnostic performance, suggesting broad applicability of circulating miRNA signatures across diverse tumor histologies within the ovarian cancer spectrum.

The clinical translation potential of circulating miRNAs as supplementary diagnostic indicators is compelling, given their non-invasive nature and demonstrated moderate-to-high discriminative capability. These characteristics position miRNA-based assays as promising candidates for early detection screening, diagnostic triage, and clinical decision-making adjuncts. However, successful implementation necessitates standardized biospecimen handling protocols, prospective validation studies, and comprehensive external validation across diverse populations and disease subtypes. Furthermore, integration of multivariate biomarker panels with advanced analytical approaches, including machine learning-based classifiers, may substantially enhance the diagnostic accuracy and clinical applicability of miRNA-based testing platforms.

Limitations

Several methodological limitations warrant consideration when interpreting these findings. The QUADAS-2 assessment revealed persistent high risk of bias in the patient selection domain, predominantly attributable to retrospective study designs and case-control methodologies employed across included investigations. While most studies demonstrated acceptable risk profiles regarding index test execution, reference standard application, and temporal flow considerations, the prevalence of non-randomized patient selection strategies likely contributed to overestimation of diagnostic accuracy parameters.

Substantial between-study heterogeneity (I² > 80% for multiple diagnostic metrics) represents a significant analytical challenge, despite our comprehensive efforts to identify and address contributing factors through subgroup and meta-regression analyses. This heterogeneity reflects methodological diversity in miRNA extraction protocols, detection platforms, normalization strategies, and patient population characteristics, potentially limiting the generalizability of pooled estimates.

Significant clinical and analytical heterogeneity was observed among the included studies, particularly with respect to histological subtype distribution, FIGO stage stratification, control group characterization, endogenous reference gene selection, and RT-qPCR detection platforms. Such methodological and population-level variability likely contributed to the substantial between-study heterogeneity in pooled diagnostic estimates and may limit direct comparability and external validity.

The predominance of case-control study designs inherently limits the applicability of our findings to real-world clinical screening scenarios, where disease prevalence and patient presentation patterns differ substantially from highly selected research cohorts. Additionally, the absence of standardized miRNA detection methodologies across included studies introduces technical variability that may not reflect the performance characteristics of optimized, clinically validated assay platforms.

Geographic concentration of included studies, with substantial representation from specific regions, may limit the generalizability of findings to diverse global populations with varying genetic backgrounds, environmental exposures, and healthcare delivery systems. Furthermore, the focus on aggregate miRNA performance, while methodologically appropriate, precludes detailed assessment of individual biomarker contributions to diagnostic accuracy, potentially obscuring opportunities for biomarker panel optimization.

Conclusion

This meta-analysis offers strong evidence that circulating microRNAs are potential diagnostic biomarkers, especially for those detected in serum. The overall diagnostic performance of the assay is robust, reproducible, and consistent among various studies. Although there was heterogeneity in the included studies, largely related to the type of biospecimen, and by region, important opportunities for methodological improvement and standardization are underscored. Hence, an international, harmonized approach with ongoing prospective validation is mandatory to realize the true translational potential of circulating miRNAs and to incorporate them into evidence-based diagnostic pathways.

Authors’ contributions

M.O.I designed the study. M.O.I, A.A., and A.A.S.B searched the database and selected the studies and Y.G and J.C. extracted the data and evaluated the studies’ quality. B.Z and M.O.I did the analysis, wrote the manuscript and performed the editing. All authors have read and approved the content.

Funding

This work was supported by Postdoctoral Innovation Project of Shandong Province (SDCX- ZG-202400087), the project of Qingdao Postdoctoral Science Foundation funded (QDBSH20240102010), the program of Qingdao Marine Science and Technology Centre (No. 2022QNLM030003-4, No. 8 − 01) and the project of Qingdao University of Science and Technology talent introduction research start-up fund.

Data availability

The original contributions presented in the study are included in the article. Further inquiries can be directed to the corresponding author.

Declarations

Ethics approval and consent to participate

Not applicable (Meta Analysis).

Consent for publication

Not applicable (Meta Analysis).

Competing interests

The authors declare no competing interests.