O-029 The role of non-coding RNAs in endometriosis diagnosis: A systematic review and meta-analysis Free

Can non-coding RNAs (ncRNAs) serve as sensitive biomarkers to differentially diagnose endometriosis accurately and precisely?

Non-coding RNAs appear to be promising biomarkers for endometriosis diagnosis, providing adequate diagnostic efficiency, especially when co-evaluated as a model and not as single markers.

Endometriosis is a chronic, inflammatory gynecological disease characterized by endometrial-like tissue development outside the uterus. This benign, estrogen-dependent inflammatory disorder affects 6–10% of women of reproductive age and over 40% of patients with endometriosis present with infertility. Despite advances, endometriosis diagnosis remains challenging, and it is mainly relied upon surgical assessment, commonly via laparoscopy. The invasive nature of endometriosis diagnosis could be avoided if sensitive biomarkers for endometriosis deferential diagnosis were available. In the context of developing non-invasive diagnostic methods, the role of ncRNAs has been investigated and this study aims to collectively synthesize and critically analyze existing evidence.

A systematic review was performed in PubMed/Medline and Embase up to April 2022 followed by a meta-analysis. Original, full-text articles in English comparing the diagnostic validity of ncRNAs versus conventional methods for endometriosis diagnosis were included. Study inclusion requirements were the following: presenting detailed information regarding the tissue where ncRNAs were isolated, methods employed for ncRNA detection, and finally inclusion of data on ncRNA diagnostic efficiency, namely sensitivity, specificity, and area under the curve (AUC).

The studied population consisted of women with endometriosis confirmed via conventional diagnostic methods, including laparoscopy and histology. The control group consisted of women without endometriosis. Data indicating ncRNA diagnostic efficiency, including sensitivity, specificity, and AUC were extracted. Following extraction, data were meta-analyzed employing univariate analysis with Clopper-Pearson. Confidence intervals were calculated employing the random-effects model. Bivariate analysis to plot the summary receiver operating characteristic (SROC) curve was also employed. R Programming Language was used.

Twenty-six studies were included and 64 ncRNAs were analyzed. Included studies were categorized into two groups. In the first group the diagnostic efficiency of single ncRNAs was evaluated. In the second group the diagnostic efficiency of models comprising of two or more ncRNAs was assessed. Considering the first group, two ncRNAs, namely miR-200c and miR-199a, differentially diagnosed endometriosis with an AUC of 1.000. Meta-analysis revealed that the total diagnostic efficiency of single ncRNAs is characterized by an AUC of 0.783, sensitivity 78.0% (95%CI:72.3-82.8%), specificity 73.8% (95%CI:67.8-79.1%), Positive Predictive Value (PPV) 81.1% (95%CI:77.1–84.5%), and Negative Predictive Value (NPV) 69.5% (95%CI:62.4–75.9%). Sensitivity analysis presenting the total diagnostic efficiency of single ncRNAs evaluated from two or more individual studies was performed. The total AUC was reported to be 0.824, sensitivity 81.8% (95%CI:72.0–88.7%), specificity 80.0% (95%CI:68.8-87.9%), PPV 87.6% (95%CI:80.3–92.4%), and NPV 72.2% (95%CI:57.4–83.3%). Regarding the second group, the ncRNA model comprised of miR-125b-5p, miR-451a and miR-3613-5p reporting the highest diagnostic efficiency, with an AUC of 1.000. Meta-analysis revealed that the total diagnostic efficiency of ncRNA models is characterized by an AUC of 0.896, sensitivity 86.8% (95%CI:81.9–90.5%), specificity 80.3% (95%CI:68.9-88.2%), PPV 88.3% (95%CI:82.2–92.5%), and NPV 78.0% (95%CI:68.2–85.4%).

The main limitation is the high heterogeneity observed among most of the studied outcomes. This is mainly attributed to the fact that the number of the participants significantly ranged between the studies. Another reason for caution is the high heterogeneity observed regarding the methods employed for ncRNA profiling and validation.

Data presented herein indicate that ncRNAs appear as highly promising non-invasive biomarkers for endometriosis diagnosis, especially when two or more ncRNAs are co-evaluated as a model. Larger scale studies should focus on developing artificial intelligence based diagnostic algorithms incorporating ncRNAs profiling coupled by molecular, biochemical, and imaging diagnostic data.

Not applicable