{"paper_id":"13c72ee7-e6a8-41a1-8eba-fdd3648c906c","body_text":"npj | women's health Article\nhttps://doi.org/10.1038/s44294-025-00116-5\nTranslation of miRNA blood-based\ndiscovery to molecular testing for clinical\ndiagnosis of endometriosis\nCheck for updates\nYanqin Yu1,4, Wing Hing Wong2,4, Yao Hu1, Yirong Shen1, Xiaochun Xu1,S h e nL u1,L i b oZ h u3,\nXinmei Zhang3 & Farideh Z. Bischoff2\nEndometriosis is a common yet often underdiagnosed condition, partly due to the lack of reliable\ndiagnostics. This study examines the clinical feasibility of a blood-based, miRNA-driven test to\ndiagnose endometriosis and address the challenges of translating next-generation sequencing (NGS)\nfindings into clinical use. Serum from 20 patients and 20 controls underwent miRNA sequencing to\nidentify diagnostic biomarkers. A machine learning model built on all NGS-based differentially\nexpressed miRNA biomarkers achieved ≥90% accuracy. Validation by qPCR confirmed some but not\nall findings, underscoring the dif ficulty of adapting NGS discoveries for routine diagnostics.\nNonetheless, serum miRNA biomarkers show strong promise for non-invasive endometriosis\ndetection, with further optimization needed for clinical translation.\nEndometriosis is a chronic and often d ebilitating condition that affects\napproximately 5–10% of women and adolescents within the reproductive\nage range of 15 –49 years1–3. However, among women experiencing infer-\ntility, the prevalence of endometriosis can rise signi ficantly, affecting\napproximately 50% of this population. The disease can manifest as early as\nthe first menstrual period and persist until menopause, causing a wide range\nof symptoms that significantly impact quality of life. Notably, between 50%\nand 80% of women who suffer from chronic pelvic pain are diagnosed with\nendometriosis, highlighting its strong association with this symptom\n1–3.\nDespite its prevalence and impact, endometriosis remains under-\ndiagnosed and poorly understood. Early detection and timely intervention\nare crucial for improving patient outcomes, yet diagnosis is frequently\ndelayed by an average of 6 –11 years\n4,5. This delay is attributed to several\nfactors, including the unknown etiology of the disease, the nonspecifica n d\nvaried nature of its symptoms, and the limitations of current diagnostic\nmethods. Symptoms such as dysmenorrhea, chronic pelvic pain, dyspar-\neunia, and gastrointestinal disturbances can often mimic other gynecologic\nor gastrointestinal conditions, leading to misdiagnosis or delayed recogni-\ntion. This variability, coupled with the lack of reliable diagnostic tools, poses\nsignificant challenges for both clinicians and researchers\n6,7.\nThe current“gold standard” diagnostic endometriosis test involves an\ninvasive laparoscopic surgical procedure, which carries risks of infection,\nbleeding, and damage to surrounding tissues8–10. The procedure is only as\nreliable as the operator ’s surgical experience in locating and identifying\nendometriosis lesions. This challenge is further compounded by the het-\nerogenous presentation of endometriotic lesions. While some lesions are\nmore readily visible, others are deeply infiltrative and located in anatomi-\ncally complex regions of the female reproductive system, such as the rec-\ntovaginal septum in cases of deep en dometriosis. These deeply buried\nlesions are particularly dif ficult to detect using the conventional laparo-\nscopic approach, which lowers diagnostic reliability. Indeed, studies have\nreported that the overall diagnostic accuracy can fall below 60%11–14,w i t h\nsensitivity and specificity in certain cohorts reported to be under 70%10,15.\nFor instance, a study in Canada found that while laparoscopic visualization\nhad a sensitivity of 90%, its speci ficity was only 40% 16. Another study\nreported a sensitivity of 69% and a speci ficity of 83% for laparoscopy in\ndiagnosing endometriosis17.T h e s efindings highlight that, while laparo-\nscopy is a valuable diagnostic tool, it is not infallible. Besides relying heavily\non the operator ’s experience, the heterogeneous appearance of endome-\ntriotic lesions precluded a standardized visual assessment, and the presence\nof deep in filtrating lesions may be missed during the procedure 10,18.I n\naddition, data suggest that some women undergoing the procedure do not\nhave the condition and are therefore unnecessarily exposed to surgical risks.\nFurther diagnostic advances may be expected from ever-evolving and\nindeed less invasive imaging technolo gies; however, as yet, this has not\nachieved the gold standard status.\nGiven these limitations, there is an increasing interest in developing\nnon-invasive diagnostic methods for endometriosis\n19–21. A growing body of\nevidence highlights the significant role of microRNA (miRNA) expression\nchanges in the pathogenesis and progression of various diseases, including\nendometriosis22–24. Numerous research groups have focused on the reg-\nulatory role of miRNAs in key genes associated with endometriosis and have\n1HerAnova Lifesciences, Hangzhou, China. 2HerAnova Lifesciences, Burlington, MA, USA. 3Women’s Hospital, School of Medicine, Zhejiang University,\nHangzhou, China. 4These authors contributed equally: Yanqin Yu, Wing Hing Wong. e-mail: zhangxinm@zju.edu.cn; farideh.bischoff@heranova.com\nnpj Women's Health |            (2025) 3:67 1\n1234567890():,;\n1234567890():,;\n\nexplored their potential as diagnostic biomarkers. For example, studies have\nidentified miR-200and miR-17-5pas promising biomarkers in the plasma of\nendometriosis patients25,26,w h i l emiR-135a has shown diagnostic potential\nin the saliva of these patients 27. Beyond individual candidates, systematic\nreviews have repeatedly highlighted recurrent signals such as miR-17-5p,\nmiR-451a, let-7b, miR-20a-5p, miR-3613-5p, etc., while also emphasizing\nsubstantial heterogeneity in effect sizes and directionality28,29.B l o o d - b a s e d\ninvestigations have reported both single marker and multi-marker panels,\nincluding serum miR-451a with varying accuracies across studies\n30,31.I n\nparallel, extracellular vesicle-derived miRNAs have been proposed as\npotentially more stable analytes than free-circulating miRNAs, with recent\nstudies reporting vesicular miRNA panels that correlate with disease\nseverity, although further large-scale validation is still required\n32,33. However,\nconflicting results have also emerged regarding the diagnostic value of\nmiRNAs. For example, one study found that serum miR-199a was upre-\ngulated in individualswith endometriosis34, while another study reported its\ndownregulation in serum35, highlighting the challenges in interpreting these\nfindings and developing reliable diagnostic tests.\nSeveral factors may contribute to these discrepancies. For instance, the\nmenstrual phase during which samples are collected may influence miRNA\nexpression, as studies have shown that miRNA levels can vary between the\nproliferative and secretory phases of the menstrual cycle36,37.A d d i t i o n a l l y ,\nvariations in study protocols, such as differing inclusion and exclusion\ncriteria, may introduce further inconsistencies, especially considering the\nheterogeneity of endometriosis as a disease. Despite the numerous\nchallenges associated with standardizing miRNA-based diagnostics, such as\nbiological variability, technical inc onsistencies, and the complexity of\nendometriosis as a disease, considerable progress has been made in devel-\noping and implementing diagnostic tools in clinical settings. One notable\nexample is the Endotest developed by Ziwig Inc., a test that utilizes a next-\ngeneration sequencing (NGS) panel comprising over 100 miRNAs to\nevaluate disease status\n38,39. This test has been primarily introduced in Eur-\nopean healthcare systems and represents a significant advancement in non-\ninvasive diagnostics for endometriosis. Although the Ziwig Endotest holds\nsubstantial potential as a laboratory-developed test, its broader imple-\nmentation as a routine in vitro diagnostic (IVD) tool remains limited. This is\nlargely due to the high cost associated with NGS technologies and the\nrequirement for advanced bioinformatics infrastructure and expertise\n40,41,\nwhich may not be readily available in routine clinical settings. As such, while\npromising, the NGS-based test is currently better suited for use in a cen-\ntralized reference laboratory rather than widespread clinical deployment in\nthe form of IVD\n42.\nOne major obstacle in translating NGS-derived miRNA biomarkers\ninto clinically actionable polymerase chain reaction (PCR)-based assays\n(quantitative PCR (qPCR) or droplet digital PCR (ddPCR)) is the challenge\nof selecting appropriate endogenous controls for data normalization\n43,44.\nUnlike NGS, which employs global normalization strategies such as reads\nper kilobase million and transcripts per million, PCR-based approaches\nrequire specific endogenous reference genes for normalization. However,\nmany studies rely on “universal” endogenous controls, such as GAPDH,\nA\nBone marrow\nCervix\nMAPK signaling\nHTLV-1 infection\nFocal adhesion\nCell cycle\nInsulin signaling\nNeurotropin signaling\nERBB signaling\nTGF-beta signaling\n20 30 40 50 60\nNumber of hits\nmiRNA\ntissue\nmiRNA targets\nKEGG\nLog2 fold change\n-1 0 1\n0\n3\n9\n-Log10(P)\nB\nC\nmiR-9-5p\nmiR-532-3p\nmiR-4710\nmiR-4676-3p\n6\n12\nmiR-21-5p miR-497-5p\nUpregulated\nDownregulated\nNot significant\nControlsEndometriosis\nZ-scaled\nexpression\n-4\n-2\n0\n2\n4\n85 Differentially expressed miRNAs\nmiR-9-5p\nmiR-19b-3p\nmiR-4710\nmiR-532-3p\nmiR-4676\nmiR-21-5p\nmiR-24-3p\nlet-7b-3p\nmiR-664a-5p\n-Log10(P-adj)\n9\n7\n5\nNo. of hits\n20\n30\n40\n50\n60\nFig. 1 | Differential expression analysis of miRNAs in endometriosis patients.\nA Heatmap illustrating the expression pro files of miRNAs that were differentially\nexpressed between endometriosis patients and healthy controls. A total of 85\nmiRNAs showed signi ficant expression differences. Notably, several of these miR-\nNAs, including miR-9-5p, miR-21-5p, and let-7b-3p, have been previously associated\nwith the pathogenesis of endometriosis and are highlighted on the plot. B Volcano\nplot depicting the distribution of differentially expressed miRNAs based on fold\nchange and statistical signi ficance. C Pathway enrichment analysis of the differen-\ntially expressed miRNAs. The top panel shows inferred tissue-of-origin patterns,\nsuggesting potential source tissues of the dysregulated miRNAs. The bottom panel\npresents enriched non-cancer-related KEGG pathways derived from the predicted\ntarget genes of these miRNAs, highlighting relevant biological processes potentially\ninvolved in endometriosis pathophysiology.\nhttps://doi.org/10.1038/s44294-025-00116-5 Article\nnpj Women's Health |            (2025) 3:67 2\n\nRNU48,o r miR-16, without assessing their stability in the specificd i s e a s e\ncontexts43,44. This oversight can introduce systematic bias, leading to unre-\nliable and non-reproducible biomarker quantification, thereby limiting their\nclinical utility.\nTo address these challenges, we conducted a proof-of-concept study\nintegrating unbiased miRNA biomarker discovery via miRNA sequencing\n(miRNA-seq) with subsequent experimental validation using qPCR. Our\nprimary objective is to present the development and feasibility of a miRNA-\nbased panel of biomarkers that would enable a non-invasive, simple, and\naccurate blood test in diagnosing endometriosis reliably. To control for\nfemale-specific physiological processes in fluenced by hormonal fluctua-\ntions, such as those occurring during the menstrual cycle, we limited serum\ncollection to women in the secretory phase. Another key innovation of our\nstudy is that we not only identified potential biomarkers but also established\nendometriosis-specific reference endogenous controls for normalization in\nclinical diagnostic settings. These newly identi fied endogenous controls\nwere systematically validated to ensure their stability and suitability for\nqPCR- and ddPCR-based assays. Experimental validation of both the\ncandidate biomarkers and the new ly selected endogenous controls\ndemonstrated strong reproducibility and accuracy, supporting their\npotential clinical application. This approach represents a critical step toward\nthe development of a standardized, rel iable, and clinically translatable\nmiRNA-based diagnostic assay for endometriosis.\nResults\nDifferentially expressed miRNAs in endometriosis patients\nIn the discovery cohort of samples co llected during the secretory phases\n(Supplementary Data 1), we identified a total of 85 differentially expressed\nmiRNAs between the patient group and the control group (Fig. 1Aa n d\nSupplementary Data 2). Among these, 55 miRNAs were signi ficantly\nupregulated in the disease group, while 30 were downregulated, indicating a\ndistinct molecular signature associated with endometriosis. Interestingly,\nsome subjects exhibited higher expression levels of specificm i R N A sc o m -\npared to others within the same group. This variability is likely attributable\nto the heterogeneous nature of endometriosis, which can manifest differ-\nently across individuals in terms of severity, symptom presentation, and\nmolecular profiles. Notably, several miRNAs previously implicated in the\npathogenesis and progression of en dometriosis were found to be highly\ndysregulated in the disease group compared to the control group (Fig.1A,\nB). These included miR-9-5p and miR-21-5p, both of which have been\nreported to play critical roles in regulating in flammatory signaling path-\nways, a key mechanism underlying the development and progression of\nendometriosis. Notably,miR-21-5p has been reported by several studies to\nbe dysregulated in women suffering from endometriosis, underscoring its\nreproducibility as a potential biomarker\n26,45.\nTo further investigate the origin a nd functional relevance of these\ndifferentially expressed miRNAs, we performed a hypergeometric over-\nrepresentation analysis using miRNet46,47. This analysis revealed that the\nmajority of these miRNAs predominantly originated from the bone marrow\nand cervix (Fig.1C). The bone marrow association reflects the serum origin\nof the miRNA data. The cervix, on the other hand, is relevant as it points to\nthe anatomical regions where endometriosis lesions can be found in\nproximity, providing confidence that the observed molecular signals are\nindeed linked to the disease. This suggests that the serum circulating\nmiRNAs may serve as biomarkers for s ystemic changes associated with\nendometriosis.\nAdditionally, we conducted a hypergeometric test using the gene tar-\ngets of the differentially expressed miRNAs against the Kyoto Encyclopedia\nof Genes and Genomes (KEGG) database to identify enriched molecular\npathways\n47. This analysis highlighted that the top non-tumor-speci fic\nmolecular processes included various signaling pathways and focal adhe-\nsion, which are strongly associated with endometriosis (Fig. 1C). For\ninstance, focal adhesion is a critical process in the pathogenicity of endo-\nmetriosis, as it involves cell –matrix interactions that contribute to the\nattachment, survival, and invasion of endometrial cells outside the uterus\n48.\nSimilarly, dysregulation in signaling pathways, such as the mitogen-\nactivated protein kinase (MAPK signaling pathway, has been observed in\nendometrial cells, leading to enhan ced cell proliferation, endometriosis\nlesion establishment, and disease persistence49. Furthermore, MAPK sig-\nnaling is implicated in pain sensitization, suggesting a role in the chronic\npelvic pain experienced by manypatients with endometriosis50.\nAssessment of differentially expressed miRNAs for the predic-\ntion of endometriosis using machine learning\nTo evaluate whether the identified differentially expressed miRNAs possess\npredictive value in distinguishing individuals with endometriosis from\ncontrol subjects, we implemented and assessed three distinct random forest\nmodels, each utilizing a different dataset.\nFor the first model (Model #1), we constructed a random forest clas-\nsifier using all 85 differentially expressed miRNAs. To rigorously assess its\nperformance, we employed a 30-fold repeated subsampling cross-validation\napproach. Model #1 demonstrated strong predictive capability, achieving an\noverall sensitivity of 0.91, specificity of 0.88, and an area under the curve\n(AUC) of 0.95 (Fig. 2A). To further refine the model and enhance inter-\npretability, we explored feature selection based on importance scores\nderived from the initial model. Speci fically, we developed two additional\nmodels with reduced feature sets: Model #2, which utilized the top 40 most\ninformative miRNAs, and Model #3, which incorporated only the top 20\n(Supplementary Data 3). These refined models were constructed using the\nsame random forest framework and evaluated with the identical repeated\nsubsampling cross-validation procedure. By focusing on the most predictive\nmiRNAs, we aimed to improve classi fication accuracy while eliminating\nnoise introduced by less informative features.\nThe results indicated that both red uced-feature models exhibited\nslightly improved performance compared to Model #1. Model #2, incor-\nporating the top 40 miRNAs, achieved a sensitivity of 0.93, a specificity of\n0.89, and an AUC of 0.97 (Fig.2B). Model #3, which was restricted to the top\n20 miRNAs, further enhanced predictive accuracy, yielding a sensitivity of\n0.95, a speci ficity of 0.90, and an AUC of 0.98 (Fig. 2C). These findings\nsuggest that a substantial proportion of the differentially expressed miRNAs\nmay not meaningfully contribute to disease classi fication and instead\nintroduce noise into the predictive model. The improved performance of the\nreduced-feature models underscores the value of feature selection in\nenhancing both the robustness and interpretability of machine learning-\nbased biomarker discovery in endometriosis.\nSelection of miRNA biomarkers and endogenous controls for\nnon-NGS clinical diagnostics\nThe translation of NGS discovery findings into actionable clinical diag-\nnostics is essential for improving patient care. While NGS provides com-\nprehensive molecular insights, clinical diagnostic applications often rely on\nqPCR due to their faster turnaround time, lower costs, and suitability for\nIVD implementation without requiring the complex bioinformatics infra-\nstructure necessary for NGS analysis\n28–30. Given these advantages, we\nexplored the feasibility oftranslating our NGS-basedfindings into a qPCR-\nbased diagnostic assay.\nOne of the primary considerations in this transition is the limit of\ndetection of qPCR. The 85 differentially expressed miRNAs identified in our\nNGS analysis exhibit a wide range of expression levels across samples,\nspanning from as few as 10 normalized read counts (e.g.,miR-4710)t oo v e r\n1,000,000 (e.g.,miR-21-5p) (Supplementary Data 2). Based on initial qPCR\nassessments, we determined that reliable detection in qPCR is achieved for\nmiRNAs with an average NGS normalized read count exceeding 500.\nApplying this threshold, 38 of the dif ferentially expressed miRNAs fell\nbelow the cutoff and were deemed unreliable for qPCR-based detection (Fig.\n3A). This left 47 differentially expressed miRNAs that met the detection\nthreshold for further investigation (Supplementary Data 4).\nTo assess the predictive potential of these 47 miRNAs in a qPCR\nsetting, we constructed two random forest models. Thefirst model incor-\nporated all 47 differentially expressed miRNAs, while the second utilized a\nhttps://doi.org/10.1038/s44294-025-00116-5 Article\nnpj Women's Health |            (2025) 3:67 3\n\nreduced feature set consisting of the20 most informative miRNAs selected\nfrom this pool. Notably, many of the miRNAs identified as top contributors\nin the previous analysis using all 85 markers—such as miR-21-5p, miR-17-\n5p,a n dmiR-15b-5p—remained among the top-ranked features, suggesting\nthat key predictive biomarkers identified via NGS may indeed be transla-\ntable to a clinical qPCR assay.\nPerformance evaluation of these models revealed that thefirst model\n(using all 47 miRNAs) achieved an AUC of 0.78 (Fig. 3B), whereas the\nsecond model (using the top 20 miRNAs) exhibited improved performance\nw i t ha nA U Co f0 . 8 4( F i g .3C). As observed previously, reducing the feature\nset led to a slight increase in performance, likely due to the removal of\nuninformative or noisy variables. However, it is important to note that these\nqPCR-based models demonstrated lower overall performance compared to\nthe models trained on all 85 differentially expressed miRNAs in the NGS\ndataset. This suggests that some of the most informative miRNAs are\nexpressed at low levels in the samples, such asmiR-9-5p,w h i c hw a sr a n k e d\nas the most predictive but had an average normalized read count of only\n~350. The inability of qPCR to reliably detect miRNAs with low expression\npresents a challenge for clinical translation. Addressing this limitation may\nrequire optimizing primer designs, adopting ddPCR for enhanced sensi-\ntivity, or incorporating target capture methods to improve miRNA\ndetection.\nA second critical factor in transitioning to a qPCR-based assay is the\nselection of a suitable endogenous co ntrol or reference marker for data\nnormalization. Inappropriate reference markers can introduce variability,\nleading to unreliable quantification of expression and potentially skewing\ndiagnostic outcomes. To address th is, we implemented a bioinformatic\npipeline for identifying disease-specific endogenous controls (Ref. “Meth-\nods”). Brie fly, we selected miRNAs that display minimal inter-group\nvariability in expression to be used as endogenous references\n(Supplementary Data 5). One such candidate, miR-92a-3p (Fig. 3D),\ndemonstrated consistent expression levels between individuals with endo-\nmetriosis and control subjects. Notably,miR-92a-3p has been validated in\nprior studies as a reliable endogenous control in blood-based miRNA\nanalyses\n51.\nTo further refine our models, we performed in silico normalization of\nthe 47 differentially expressed miRNAs againstmiR-92a-3p, simulating the\nclinical qPCR diagnostic work flow. Following normalization, we re-\nevaluated the performance of our random forest models. This approach\nled to noticeable improvements in predictive accuracy, with the model using\nall 47 miRNAs achieving an AUC of 0.80, and the reduced model using the\ntop 20 miRNAs attaining an AUC of 0.88 (Fig. 3E, F). These findings\nunderscore the importance of proper normalization strategies in enhancing\nthe robustness of qPCR-based diagnostics and further support the potential\nclinical applicability of our NGS-derived biomarker panel.\nExperimental assessment for diagnostic qPCR assays\nTo evaluate the potential of translating our NGS discoveryfindings into a\nclinically viable qPCR diagnostic assay, we selectedfive candidate miRNAs\n—miR-21-5p, miR-15b-5p, miR-17-5p, miR-19b-3p,a n d miR-23a-3p—for\nexperimental validation using qPCR (Fig.4). These miRNAs were chosen\nbased on their differential expression patterns observed in the NGS dataset,\nas well as their biological relevance to endometriosis. In addition,miR-92a-\n3p w a si n c l u d e da sa ne n d o g e n o u sc o n t r o lf o rn o r m a l i z a t i o n ,g i v e ni t s\ndemonstrated stability across endometriosis and control samples in both\nour dataset and prior studies.\nFor this validation study, we conducted qPCR assays on 90 serum\nsamples, comprising 65 patients with endometriosis and 25 control subjects\nwhose disease status was confirmed via laparoscopic surgery (Supplemen-\ntary Data 6). The goal was to determi ne whether the expression patterns\nA B C\nAUC Sensitivity Specificity\n0.00\n0.25\n0.50\n0.75\n1.00\n0.00\n0.25\n0.50\n0.75\n1.00\nAUC Sensitivity Specificity AUC Sensitivity Specificity\n0.00\n0.25\n0.50\n0.75\n1.000.95 0.91 0.88\n0.97 0.93 0.89\n0.98 0.95 0.90\nCross validation #10 (Representative)\n0.2\n0.4\n0.6\n1.0\n0.8\nTrue positive rate\nFalse positive rate\n0.2 0.4 0.6 0.8 1.0\nModel 1: All DE miRNAs Model 2: Top 40 DE miRNAs Model 2: Top 20 DE miRNAs\n0.2\n0.4\n0.6\n1.0\n0.8\nTrue positive rate\n0.2\n0.4\n0.6\n1.0\n0.8\nTrue positive rate\nFalse positive rate\n0.2 0.4 0.6 0.8 1.0\nFalse positive rate\n0.2 0.4 0.6 0.8 1.0\nAUC = 0.93 AUC = 0.94 AUC = 0.94\nCross validation #12 (Representative) Cross validation #12 (Representative)\nFig. 2 | Diagnostic performance of NGS-identi fied differentially expressed\nmiRNAs using machine learning. The diagnostic utility of differentially expressed\nmiRNAs was assessed through machine learning analysis. Performance metrics\nrepresent the mean of 30 iterations of repeated subsampling cross-validation, with\nerror bars indicating variability across iterations. A Using the full set of differentially\nexpressed miRNAs, the model achieved an AUC of 0.95, with a sensitivity of 0.91 and\nspecificity of 0.88. A representative ROC curve from iteration #10 is shown.\nB Limiting the model to the top 40 most informative miRNAs improved perfor-\nmance, yielding an AUC of 0.97, sensitivity of 0.93, and specificity of 0.89. The ROC\ncurve displayed corresponds to iteration #12. C Further refinement using the top 20\nmost informative miRNAs resulted in the highest diagnostic performance, with an\nAUC of 0.98, sensitivity of 0.95, and speci ficity of 0.90. The representative ROC\ncurve is also from iteration #12.\nhttps://doi.org/10.1038/s44294-025-00116-5 Article\nnpj Women's Health |            (2025) 3:67 4\n\nobserved in the NGS discovery datasetcould be reliably recapitulated using\nqPCR, a crucial step in transitioning towards a clinically deployable assay.\nAnalysis of the qPCR results revealed that two of thefive selected biomarkers\n(miR-21-5pand miR-15b-5p) displayed statistically significant differences in\nnormalized expression between the endometriosis and control groups.\nThese findings are consistent with our NGS discovery dataset, reinforcing\ntheir potential utility as diagnos tic biomarkers. Additionally, miR-17-5p\nexhibited a discernible trend of differential expression between disease and\ncontrol samples, as visualized in the boxplot analysis. However, this dif-\nference did not reach statistical signi ficance, suggesting that, while this\nmarker may hold some biological relevance, additional optimization—such\nas larger sample sizes or re fined qPCR conditions—may be necessary to\nestablish its diagnostic value. In contrast, the remaining two miRNAs,miR-\n19b-3p and miR-23a-3p, did not show clear delineation between patients\nand controls, indicating that their differential expression in the NGS dataset\nmay not translate robustly into a qP CR-based assay. This discrepancy\nunderscores the complexities of biomarker translation and highlights the\nneed for rigorous validation and opti mization at the experimental level.\nFactors such as primer design, ampli fication efficiency, RNA extraction\nvariability, and technical noise could all contribute to differences between\nNGS and qPCR results, emphasizing the importance of careful assay\ndevelopment. Notwithstanding, the data suggest that qPCR-based assays\ncould provide meaningful diagnostic utility with proper refinement.\nDiscussion\nS m a l lR N A ss u c ha sm i R N A sh a v eb e e ns h o w nt op l a yp i v o t a lr o l e si n\nnumerous physiological and pathological processes, influencing gene reg-\nulation in both normal and disease states. Altered miRNA expression\nprofiles have been extensively documented in the plasma and serum of\npatients with various conditions, including diffuse large B cell lymphoma,\novarian cancer, and type 2 diabetes\n52,53.T h e s efindings underscore the\npotential of circulating miRNAs as no n-invasive biomarkers for disease\ndetection and monitoring. In this s tudy, we demonstrate that miRNAs\ncirculating in serum can potentially serve as reliable biomarkers for the\ndiagnosis of endometriosis. The ability to analyze serum miRNA levels in a\nstandardized manner presents a promising approach in disease detection.\nThe fluctuations in specific circulating miRNAs offer a quanti fiable and\nreproducible means of identifying en dometriosis, potentially improving\nearly diagnosis and clinical management.\nUsing serum miRNAs as diagnostic biomarkers offers several key\nadvantages over conventional diagnostic methods in endometriosis. Cur-\nrently, the gold standard for endometriosis diagnosis relies on laparoscopy\nA B C\nD E F\nDetectable \nvia qPCR\nNot detectable \nvia qPCR\nlog2(average expression)\nDE miRNAs\n0\n5\n10\n15\n20\n0.00\n0.25\n0.50\n0.75\n1.00\nAUC Sensitivity Specificity\n0.00\n0.25\n0.50\n0.75\n1.00\nAUC Sensitivity Specificity\n0.00\n0.25\n0.50\n0.75\n1.00\nAUC Sensitivity Specificity AUC Sensitivity Specificity\n0.00\n0.25\n0.50\n0.75\n1.00\nAll qPCR detectable miRNAs Top 20 qPCR detectable miRNAs\nAll qPCR detectable miRNAs\nnormalized with miR-92a-3p\nTop 20 qPCR detectable miRNAs\nnormalized with miR-92a-3p\n0.78\n0.66 0.69\n0.84\n0.68\n0.74\n0.80\n0.70 0.68\n0.88\n0.77\n0.70\nControls Endometriosis\n500000\n750000\n1000000\n1250000Normalized read count\nmiR-92a-3p\nFig. 3 | Diagnostic performance of differentially expressed miRNAs is reliably\nquantifiable by qPCR, assessed using machine learning. Performance metrics\nrepresent the mean of 30 iterations of repeated subsampling cross-validation, with\nerror bars indicating variability across iterations. A Violin plot illustrating the\nexpression distribution of differentially expressed miRNAs based on normalized\nNGS read counts. miRNAs shown in blue represent those with expression levels too\nlow for reliable qPCR detection, while those in red indicate miRNAs with suf ficient\nexpression for reliable quanti fication by qPCR. The dotted red line marks the\nexpression threshold of 500 normalized reads, used to distinguish between the two\ngroups. B Predictive performance of a machine-learning model using all 47 miRNAs\ndeemed reliably detectable by qPCR, resulting in an average AUC of 0.78, with a\nsensitivity of 0.66 and specificity of 0.69.C Model performance using the top 20 most\ninformative miRNAs from the qPCR-detectable set, showing a modest improvement\nwith an average AUC of 0.84, sensitivity of 0.68, and specificity of 0.74. D Boxplot of\nmiR-92a-3p, a potential endogenous control candidate in this disease setting,\ndemonstrating consistent expression with minimal variability between patient and\ncontrol groups. The box ’s lower and upper hinges correspond to the 25th and 75th\npercentiles, respectively, with the median indicated by the line inside the box. The\nwhiskers extend to the most extreme data points within 1.5 times the interquartile\nrange below the 25th percentile and above the 75th percentile. E Predictive per-\nformance using all 47 reliably detectable miRNAs after in silico normalization\nagainst miR-92a-3p, yielding an average AUC of 0.80, with sensitivity of 0.70 and\nspecificity of 0.68. F Model performance using the top 20 most informative miRNAs\nfollowing in silico normalization with miR-92a-3p, showing further improvement\nwith an average AUC of 0.88, sensitivity of 0.77, and speci ficity of 0.68.\nhttps://doi.org/10.1038/s44294-025-00116-5 Article\nnpj Women's Health |            (2025) 3:67 5\n\nwith direct visualization, an invasive surgical procedure. A serum-based\nmiRNA biomarker assay could provide anon-invasive alternative, enabling\ncomprehensive disease assessment without the need for surgery. This is\nparticularly valuable for early detection and for patients who may not have\nimmediate access to specialized surgical evaluation. Secondly, compared to\ninvasive diagnostic procedures, aserum-based miRNA test is significantly\nmore cost-effective. The process involves routine blood collection and\nstandard laboratory processing, making it more accessible for widespread\nclinical implementation. Additiona lly, standardizing miRNA detection\nprotocols could facilitate large-scal e screening efforts, improving early\ndiagnosis and patient outcomes.\nIn this study, we investigated serum miRNA expression pro files in\nindividuals with endometriosis and identified a distinct set of circulating\nmiRNAs that may serve as potential biomarkers for disease detection. To\nminimize variability associated with hormonal changes during the men-\nstrual cycle, particularly those impacting female-speci fic physiological\nprocesses, serum samples were collected exclusively during the secretory\nphase. Our results add to the growing evidence supporting the use of serum\nmiRNA signatures as non-invasive diagnostic tools for endometriosis\n19–33.\nIn particular, we experimentally validated miR-21-5p and miR-15b-5p,\nwhich contain significant information delineating endometriosis patients\nfrom the control population. Both miR-21-5p and miR-15-5p have been\npreviously reported to be dysregulated in a variety of pathological condi-\ntions, particularly those characterized by in flammation or aberrant cell\nproliferation. For example,miR-21-5pis widely considered as an‘oncomiR’\nwith roles in cancer, immune activation, and angiogenesis\n54,55. Likewise,\nmiR-15-5p has been associated with tissue fibrosis and immune\nmodulation56,57. The presence of these miRNAs across multiple disease\ncontexts underscores their lack of disease speci ficity. However, their\nreproducible dysregulation in endometriosis is biologically plausible given\nthe in flammatory and fibrotic microenvironment that characterizes this\ndisease, and may serve as important components of a multi-biomarker\npanel, capturing the inflammatory and remodeling milieu of endometriosis\nwhen combined with other biomarkers.\nNonetheless, despite their promise, several key challenges remain\nbefore miRNA-based assays can be translated into clinically reliable diag-\nnostic applications. One of the primary obstacles lies in the choice of\ndetection platform. While NGS provides a comprehensive assessment of the\nmiRNA landscape, its high cost per sample and dependence on complex\nbioinformatics infrastructure render it impractical for routine clinical\ndiagnostics and IVD applications. Consequently, there is a need to transi-\ntion toward more practical methodol o g i e s ,s u c ha sq P C Ro rd d P C R ,w h i c h\noffer lower costs, faster turnaround times, and compatibility with IVD\nrequirements. Another potential challenge in clinical miRNA diagnostics is\nthe selection of appropriate endogenous controls for normalization, as the\ncommonly usedmiR-16-5pcan exhibit instability in blood-based assays\n58.I n\nour study,miR-92a-3pemerged as a more reliable endogenous control based\non empirical comparison betweenthe patient and control groups51,w h e r e a s\nmiR-16-5pshowed greater variability. It should be noted that bothmiR-16-\n5p and miR-92a-3pare linked to inflammatory processes59,60,w i t hmiR-92a-\n3p implicated in neuroin flammation60,61. While our results support the\nsuitability ofmiR-92a-3pas an endogenous control in this context, valida-\ntion in larger, independent cohorts is necessary to con firm its broader\napplicability.\nOur findings suggest that convertingNGS-based miRNA discoveries\ninto clinically applicable assays is achievable, though it necessitates experi-\nmental refinement. Despite using a limited qPCR panel with only five\nmiRNA biomarkers, we demonstrated valu a b l ed i a g n o s t i cp o t e n t i a l ,a l b e i t\nwith a need for further optimization. This highlights both the promise and\nthe technical challenges of implementing miRNA-based diagnostics in\nclinical practice.\nTo improve the reliability and diagnostic accuracy of a qPCR-based\ntest, several key strategies should be c onsidered. (a) Primer optimization:\nrefining primer designs to enhance amplification efficiency and specificity,\nparticularly for low-abundance miRNAs. (b) Adopting ddPCR for\nenhanced sensitivity: ddPCR offers improved precision and sensitivity,\nmaking it a suitable alternative for detecting low-expressed miRNAs that\nmay be missed by qPCR. (c) Expanding sample size: increasing the number\nof clinical samples analyzed will im prove statistical power and ensure\nrobustness across diverse patient populations. (d) Developing a multiplex\nassay: creating a multiplexed qPCR panel would allow for the simultaneous\ndetection of multiple miRNA biomarkers, streamlining work flow and\nimproving diagnostic efficiency.\nWhile initial qPCR validation of selected miRNAs shows promise,\nfurther refinement is necessary and underway to enhance assay reprodu-\ncibility and clinical performance. Future studi es will focus on validating\nthese biomarkers in larger, independent cohorts, optimizing detection\nmethods, and standardizing protocol s to ensure reproducibility across\nclinical testing laboratories. These efforts will be critical in bridging the gap\nbetween high-throughput discovery research and real-world clinical\napplication, ultimately paving the way for a clinically deployable serum\nmiRNA-based diagnostic test for endometriosis.\nMethods\nSpecimen collection\nPeripheral blood samples were collected prospectively from women aged\n18–49 years who presented with mild-to-severe symptoms, including pelvic\npain and/or menstrual bleeding. This is a single-center study, and the par-\nticipants were enrolled under an approved institutional review board (IRB)\nof the Women ’s Hospital of Zhejiang University School of Medicine in\naccordance with the Declaration of Helsinki, with informed consent\nobtained from all individuals. The IRB number is IRB-20240110-R. The\nFig. 4 | Boxplot of qPCR performance for selected\ndifferentially expressed miRNAs. All miRNAs\nwere normalized against miR-92a-3p using qPCR.\nOf the five miRNAs tested, two ( miR-21-5p and\nmiR-15b-5p) showed a signi ficant difference in\nexpression between patients and controls, replicat-\ning the direction of effect observed in the NGS dis-\ncovery dataset. miR-17-5p exhibited a concordant\ntrend, with a p value of 0.065, approaching statistical\nsignificance. The remaining two miRNAs, miR-19b-\n3p and miR-23a-3p, did not show a statistically sig-\nnificant difference between patients and controls.\nThe boxplot’s lower and upper hinges represent the\n25th and 75th percentiles, respectively, with the\nmedian indicated by the line inside the box. The\nwhiskers extend to the most extreme data points\nwithin 1.5 times the interquartile range below the\n25th percentile and above the 75th percentile.\nmiR-21-5p miR-15b-5p miR-17-5p miR-19b-3p miR-23a-3p\n2.7\n3.0\n3.3\n3.6\n-log2(Delta CT)\nmiRNAs normalized with miR-92a-3p\np = 0.029 p = 0.032 p = 0.073 p = 0.729 p = 0.692\nControls\nn = 25\nEndometriosis\nn = 65\n3.9\nhttps://doi.org/10.1038/s44294-025-00116-5 Article\nnpj Women's Health |            (2025) 3:67 6\n\nparticipants included in this study were collected from March 2024 to\nDecember 2024. All participants were clinically suspected of a gynecologic\nabnormal condition and were scheduled to undergo laparoscopy with his-\ntopathological confirmation for endometriosis. A subset of participants in\nour study cohort presented with co-morbidities such as adenomyosis and\nleiomyoma. These conditions frequently co-exist with endometriosis and\nmay present overlapping clinical features, making it dif ficult to fully dis-\nentangle their individual contributions. It is therefore acknowledged that the\npotential influence of these co-morbidities on the observed outcomes can-\nnot be ruled out, and this represents alimitation of the study that should be\nconsidered when interpreting the results. Detailed clinical information for\na l le n r o l l e dp a r t i c i p a n t si sp r o v i d e di nt h eS u p p l e m e n t a r yD a t a .T oa c c o u n t\nfor potential variations in miRNA e xpression due to different menstrual\ncycle phases, blood samples were collected exclusively from women in the\nsecretory phase of their menstrual cycle. The menstrual phase was initially\ndetermined by physicians or surgeons based on self-reported cycle days and\nclinical assessment. However, relying solely on calendar-based timing may\nnot provide sufficient accuracy. To strengthen the reliability of our sample\nselection, this secretory classification was further validated through serum\nprogesterone measurements using the protein assay from Kangrun Biotech\nCo., Ptd (Guangdong, China), with levels exceeding 1.08 ng/mL serving as a\nbiochemical confirmation of the secretory phase according to the manu-\nfacturer’s protocol. This appr oach minimized the in fluence of hormonal\nfluctuations on biomarker expression, thereby enhancing the reliability of\nour findings. We acknowledge that serum progesterone levels alone may not\nprecisely distinguish between early, mid, and late secretory phases. For\nexample, a progesterone concentration of 2 ng/mL could represent early or\nlate secretory windows due to the cyclical nature of the hormone level, where\ndynamic changes in progesterone signaling, immune cell infiltration, and\nstromal remodeling are well docume nted. Therefore, we recognize that\nresidual heterogeneity due to broad secretory phase classification is a lim-\nitation of the study and may confound interpretation of the data. Given the\nconstraints of patient recruitment a nd sample availability, we adopted a\npragmatic approach that combined calendar-based cycle staging with bio-\nchemical validation using serum progesterone to ensure all participants\nwere indeed in the secretory phase. Future studies with larger cohorts and\nadditional markers of endometrial dating will be needed to minimize this\nsource of heterogeneity. In this study, a total of 40 symptomatic women were\nincluded in the NGS discovery cohort, with 10 mL of blood drawn into\nstandard red-top blood collection tubes prior to the laparoscopic surgery.\nAmong them, 20 women were confirmed to have endometriosis based on\nboth laparoscopic findings and histopathology (disease group), while the\nremaining 20 had no evidence of endometriosis and served as the control\ngroup. Serum was isolated using a two-step centrifugation protocol. First,\nsamples underwent a low-speed centrifugation at 3000 rpm for 10 min at\n4 °C to remove cellular components. This was then followed by a second\nhigh-speed centrifugation at 16,000 ×g for 10 min at 4 °C to ensure com-\nplete removal of debris and platelets. The isolated serum was then aliquoted\nand stored at −80 °C for subsequent RNA extraction and downstream\nprocessing.\nRNA isolation and miRNA-seq\nTotal RNA was isolated from 300 μL of serum using the Norgen RNA\nextraction kit following the manufacturer’s instructions. Total RNAs were\nligated to 3’ adapters by denaturation at 70 °C for 2 min, and then incubated\nat 16 °C for over 8 h using NEB T4 RNA Ligase 2. Afterwards, 5’ adapters\nwere incubated with the previous product using NEB T4 RNA Ligase 1 at\n37 °C for 60 min. Ligated RNA was reverse transcribed in a thermocycler\nusing SuperScript II Reverse Transcriptase from ThermoFisher Inc. under\nthe following conditions: an initial incubation at 50 °C for 60 min, followed\nby a heat inactivation step at 80 °C for 10 min. Following complementary\nDNA (cDNA) synthesis, library preparation was performed using the NEB\nPhusion High-Fidelity DNA Polymerase, adhering strictly to the manu-\nfacturer’s guidelines. The final libraries were then subjected to high-\nthroughput sequencing to profile the miRNA expression by LC Biosciences.\nNGS miRNA differential expression profiling endogenous control\nselection\nThe raw FASTQ data obtained from miRNA sequencing underwent pre-\nprocessing and analysis using the miRge362 software pipeline. Initially, the\nsequences were quality-trimmed, and adapter sequences were removed\nusing a Cutadapt63 wrapper integrated within miRge3. The trimmed reads\nwere then aligned to the miRBase 64 reference database using Bowtie 65\noptimized for short reads. Following alignment, miRge3 generated a com-\nprehensive count table summarizing the abundance of each miRNA across\nthe samples. This count table served as the input for downstream differential\nexpression analysis using the DESeq2\n66 package in R. DESeq2 was employed\nto identify miRNAs that exhi bited statistically significant differences in\nexpression between the patient and con trol groups. After identifying the\ndifferentially expressed miRNAs, hy pergeometric over-representation\nanalyses were conducted using the miRNet platform 46. In this analysis,\ntwo distinct queries were performed: first, the differentially expressed\nmiRNAs were compared against the miRNA-tissue origin database in\nmiRNet to infer potential tissue-speci fic origins or associations of these\nmiRNAs; second, the predicted gene targets of these miRNAs were mapped\nto the KEGG database\n47 to identify statistically enriched biological pathways.\nThis dual-level approach enabled th e contextualization of the miRNA\nexpression patterns in terms of both tissue relevance and functional pathway\ninvolvement.\nEndogenous control selection for in silico normalization and\nqPCR experimental validation\nTo bridge the findings from the next-generation sequencing (NGS) dis-\ncovery cohort into clinically applicable diagnostic tools, we further aimed to\nidentify condition-specific endogenous control miRNAs. These controls are\nessential for normalizing qPCR or ddPCR assays, ensuring accurate and\nreproducible quantification of target miRNA expression. The selection of\nendogenous controls was guided by stringent criteria. Speci fically, we\nevaluated the expression stability of candidate miRNAs by assessing their\ndispersion estimates and imposing constraints on log2 fold-change (|\nlog2FC| < 0.02) between the patient andcontrol groups. This ensured that\nthe selected controls exhibited minimal variability across conditions.\nAdditionally, candidates were filtered based on an adjusted p-value\nthreshold (≥0.8), ensuring that their expression was not influenced by the\nexperimental conditions or disease state.\nDisease prediction model construction using a random forest\nclassifier\nTo assess the predictive capability of the differentially expressed miRNAs\nand determine the extent to which they could accurately classify patients\nwith endometriosis, we constructed a random forest classifier using all the\ndifferentially expressed miRNAs identi fied in the discovery cohort. To\nensure a robust evaluation of the model’s performance, we performed 30\niterations of repeated random subsampling cross-validation, where in each\niteration, the data was split into an 80:20 ratio for training and testing,\nrespectively. This repeated holdout validation approach allowed us to\naccount for variability in model performance due to random data parti-\ntioning and provided a reliable estimate of the model’s predictive accuracy.\nDuring model construction, any missing values were imputed using the\nmedian of the corresponding feature.\nFollowing the initial model construction, we conducted feature selec-\ntion to identify the most informative miRNAs for predicting endometriosis.\nThis was achieved by evaluating the feature importance scores generated by\nthe random forest algorithm. Based on these scores, we created two distinct\nfeature sets: one comprising the top 40 most important miRNAs and\nanother consisting of the top 20 most important miRNAs. These reduced\nfeature sets were then used to generate and assess new models using the\nsame repeated random subsampling cross-validation procedure described\nabove. Reducing the feature set is crucial for the assessment of model per-\nformance and interpretability. A smaller, more informative subset of\nmiRNAs helps prevent overfitting, ensuring the model generalizes well to\nhttps://doi.org/10.1038/s44294-025-00116-5 Article\nnpj Women's Health |            (2025) 3:67 7\n\nnew data. Additionally, focusing on the most important miRNAs reduces\nnoise, leading to more reliable predictions. A reduced feature set also\nfacilitates translation into clinical diagnostics, where the informative miR-\nNAs can be assayed using qPCR and/or ddPCR instead of NGS.\nRNA extraction and qPCR experimental validation\nTotal RNA was extracted from 200μL serum using the miRNeasy Serum/\nPlasma Advanced Kit from Qiagen, following the manufacturer’s recom-\nmended protocol. Subsequently, targeted miRNAs were reverse transcribed\ninto cDNA using the FastKing RT Kit II from TianGen Inc. The reverse\ntranscription reaction was carried out in a thermocycler under the following\nconditions: an initial incubation at 42 °C for 15 min to facilitate cDNA\nsynthesis, followed by a heat inactivation step at 95 °C for 1 min to terminate\nthe reactions. The synthesized cDNA was subjected to qPCR analysis of the\nfive miRNA markers. The PCR reaction mixtures were prepared by miR-\nCURY LNA miRNA SYBR Green PCR Kit from Qiagen. qPCR ampli fi-\ncation was performed in the QuantStudio qPCR system following an initial\ndenaturation at 95°C for 2 min, followed by 40 cycles with denaturation at\n95 °C for 10 s and annealing at 56 °C for 60 s.\nStatistical analyses\nDifferential expression analysis was performed using DESeq2, which\nmodels NGS count data with a negative binomial distribution to assess\nstatistical differences between patients and controls. For machine learning-\nbased predictions, sensitivity, specificity, and AUC were evaluated using\nPython’s scikit-learn package. Pairwise expression comparisons between\npatients and controls were assessed using the Wilcoxon rank-sum test.\nStudy approval\nThe participants were enrolled under an approved institutional review\nboard protocol (IRB-20240110-R) at the Women ’sH o s p i t a lo fZ h e j i a n g\nUniversity School of Medicine, with i nformed consent obtained from all\nindividuals.\nData availability\nSequencing data were deposited into Genome Sequence Archive under the\naccession number HRA011242 and can be accessed viahttps://ngdc.cncb.\nac.cn/search/specific?db=hra&q=HRA011242.\nCode availability\nScripts used for feature selection, random forest model construction, and\nfigure plotting have been deposited in the GitHub repository (https://github.\ncom/Heranova-Lifesciences/endometriosis_miRNAseq_manuscript).\nReceived: 4 May 2025; Accepted: 13 November 2025;\nReferences\n1. Zondervan, K. T. et al. Endometriosis. N. Engl. J. Med. 382,\n1244–1256 (2020).\n2. Bonavina, G. & Taylor, H. S. Endometriosis-associated infertility: from\npathophysiology to tailored treatment.Front. Endocrinol. 13, 1020827\n(2022).\n3. Parasar, P., Ozcan, P. & Terry, K. L. Endometriosis: epidemiology,\ndiagnosis and clinical management. Curr. Obstet. Gynecol. Rep. 6,\n34–41 (2017).\n4. Fryer, J., Mason-Jones, A. J. & Woodward, A. Understanding\ndiagnostic delay for endometriosis: a scoping review using the social-\necological framework. Health Care Women Int. 46, 335–351 (2025).\n5. Kirk, U. B. et al. Understanding endometriosis underfunding and its\ndetrimental impact on awareness and research. npj Women’s Health\n2, 45 (2024).\n6. Davenport, S., Smith, D. & Green, J. D. Barriers to a timely diagnosis of\nendometriosis: a qualitative systematic review. Obstet. Gynecol. 142,\n571–583 (2023).\n7. Pascoal, E. et al. Strengths and limitations of diagnostic tools for\nendometriosis and relevance in diagnostic test accuracy research.\nUltrasound Obstet. Gynecol. 60, 309–327 (2022).\n8. Leyland, N., Casper, R., Philippe, L. & Singh, S. S. & SOGC.\nEndometriosis: diagnosis and management. J. Obstet. Gynaecol.\nCan. 32,S 1–S32 (2010).\n9. Dunselman, G. A. et al. ESHRE guideline: management of women with\nendometriosis. Hum. Reprod. 29, 400–412 (2014).\n10. Taylor, H. S. et al. An evidence-based approach to assessing surgical\nversus clinical diagnosis of symptomatic endometriosis. Int. J.\nGynecol. Obstet. 142, 131–142 (2018).\n11. Mak, J., Leonardi, M. & Condous, G. ‘Seeing is believing’: arguing for\ndiagnostic laparoscopy as diagnostic test for endometriosis. Reprod.\nFertil. 3, C23–C28 (2022).\n12. Agarwal, S. K. et al. Clinical diagnosis of endometriosis: a call to\naction. Am. J. Obstet. Gynecol. 220, 354.e1–354.e12 (2019).\n13. Gerges, B., Li, W., Leonardi, M., Mol, B. W. & Condous, G. Meta-\nanalysis and systematic review to determine the optimal imaging\nmodality for detection of uterosacral ligaments/torus uterinus,\nrectovaginal septum and vaginal deep endometriosis. Hum. Reprod.\nOpen 2021, hoab041 (2021).\n14. Dantkale, K. S. & Agrawal, M. A comprehensive review of the\ndiagnostic landscape of endometriosis: assessing tools, uncovering\nstrengths, and acknowledging limitations. Cureus 16\n, e56978 (2024).\n15. Albee, R. B. Jr., Sinervo, K. & Fisher, D. T. Laparoscopic excision of\nlesions suggestive of endometriosis or otherwise atypical in\nappearance: relationship between visual findings and final histologic\ndiagnosis. J. Minim. Invasive Gynecol. 15,3 2–37 (2008).\n16. Gratton, S. et al. Diagnosis of endometriosis at laparoscopy: a\nvalidation study comparing surgeon visualization with histologic\nfindings. J. Obstet. Gynaecol. Can. 44, 135–141 (2022).\n17. Sutton, C. J., Ewen, S. P., Whitelaw, N. & Haines, P. Prospective,\nrandomized, double-blind, controlled trial of laser laparoscopy in the\ntreatment of pelvic pain associated with minimal, mild, and moderate\nendometriosis. Fertil. Steril. 62, 696–700 (1994).\n18. Afors, K. et al. Employing laparoscopic surgery for endometriosis.\nWomen’s Health 10, 431–443 (2014).\n19. Vanhie, A. et al. Plasma miRNAs as biomarkers for endometriosis.\nHum. Reprod. 34, 1650–1660 (2019).\n20. Avery, J. C. et al. Noninvasive diagnostic imaging for endometriosis part\n1: a systematic review of recent developments in ultrasound, combination\nimaging, and artificial intelligence.Fertil. Steril.121, 164–188 (2024).\n21. Oskouei, B. S., Asadi, Z. & Esfahlan, R. J. Non-invasive blood tests for\nearlier diagnosis and treatment of endometriosis. J. Reprod. Immun.\n169, 104521 (2025).\n22. Bjorkman, S. & Taylor, H. S. MicroRNAs in endometriosis: biological\nfunction and emerging biomarker candidates. Biol. Reprod. 100,\n1135–1146 (2019).\n23. Bendifallah, S. et al. MicroRNome analysis generates a blood-based\nsignature for endometriosis. Sci. Rep. 12, 4051 (2022).\n24. Nazri, H. M. et al. The role of small extracellular vesicle-miRNAs in\nendometriosis. Hum. Reprod. 38, 2296–2311 (2023).\n25. Rekker, K. et al. Circulating miR-200-family micro-RNAs have altered\nplasma levels in patients with endometriosis and vary with blood\ncollection time. Fertil. Steril. 104, 938–946 (2015).\n26. Papari, E., Noruzinia, M., Kashnani, L. & Foster, W. G. Identi fication of\ncandidate microRNA markers of endometriosis with the use of next-\ngeneration sequencing and quantitative real-time polymerase chain\nreaction. Fertil. Steril. 113, 1232–1241 (2020).\n27. Perricos, A. et al. Hsa-mir-135a shows potential as a putative\ndiagnostic biomarker in saliva and plasma for endometriosis.\nBiomolecules 12, 1144 (2022).\n28. Vanhie, A. et al. Circulating microRNAs as non-invasive biomarkers in\nendometriosis diagnosis –\na systematic review. Biomedicines 12, 888\n(2024).\nhttps://doi.org/10.1038/s44294-025-00116-5 Article\nnpj Women's Health |            (2025) 3:67 8\n\n29. Leonova, A., Turpin, V. E., Agarwal, S. K., Leonardi, M. & Foster, W. G.\nA critical appraisal of the circulating levels of differentially expressed\nmicroRNA in endometriosis. Biol. Reprod. 105, 1075–1085 (2021).\n30. Zafari, N. et al. microRNAs as novel diagnostic biomarkers in\nendometriosis patients: a systematic review and meta-analysis.\nExpert Rev. Mol. Diagn. 22, 479–495 (2022).\n31. Nothnick, W. B., Falcone, T., Joshi, N., Fazleabas, A. T. & Graham, A.\nSerum miR-451a levels are signi ficantly elevated in women with\nendometriosis and recapitulated in baboons ( Papio anubis) with\nexperimentally-induced disease. Reprod. Sci. 24, 1195–1202 (2017).\n32. Zhang, L. et al. Serum exosomal MicroRNAs as potential circulating\nbiomarkers for endometriosis. Dis. Markers 2020, 2456340 (2020).\n33. Chu, X. et al. Extracellular vesicles in endometriosis: role and\npotential. Front. Endocrinol. 15, 1365327 (2024).\n34. Wang, W., Zhao, Y., Han, B., Hong, S. & Chen, Y. Circulating\nmicroRNAs identified in a genome-wide serum microRNA expression\nanalysis as noninvasive biomarkers for endometriosis. J. Clin.\nEndocrinol. Metab. 98, 281–289 (2013).\n35. Hsu, C. Y. et al. miRNA-199a-5p regulates VEGFA in endometrial\nmesenchymal stem cells and contributes to the pathogenesis of\nendometriosis. J. Pathol. 232, 330–343 (2014).\n36. Petracco, R. et al. Evaluation of miR-135a/b expression in\nendometriosis lesions. Biomed. Rep. 11, 181–187 (2019).\n37. Lamon, S. et al. The effect of the menstrual cycle on the circulating\nmicroRNA pool in human plasma: a pilot study. Hum. Reprod. 38,\n46–56 (2022).\n38. Bendifallah, S. et al. Machine learning algorithms as new screening\napproach for patients with endometriosis. Sci. Rep. 12, 639 (2022).\n39. Ferrier, C. et al. Saliva microRNA signature to diagnose\nendometriosis: a cost-effectiveness evaluation of the Endotest.BJOG\n130, 396–406 (2023).\n40. Zalis, M. et al. Next-generation sequencing impact on cancer care:\napplications, challenges, and future directions. Front. Genet. 15,\n1420190 (2024).\n41. Desai, K. et al. Real-world trends in costs of next-generation sequencing\n(NGS) testing in U.S. setting.J. Clin. Oncol. 39, e18824 (2021)\n42. Karlovich, C. A. & Williams, P. M. Clinical applications of next-\ngeneration sequencing in precision oncology. Cancer J. 25, 264–271\n(2019).\n43. Chapman, J. R. & Waldenstrom, J. With reference to reference genes:\na systematic review of endogenous controls in gene expression\nstudies. PLoS ONE 10, e0141853 (2015).\n44. Taylor, S. C. et al. The ultimate qPCR experiment: producing\npublication quality, reproducible data the first time. Trends\nBiotechnol. 37, 761–774 (2019).\n45. Kupec, T. et al. Diagnostic potential of serum circulating miRNAs for\nendometriosis in patients with chronic pelvic pain. J. Clin. Med. 14,\n5154 (2025).\n46. Chang, L., Zhou, G., Soufan, O. & Xia, J. miRNet 2.0: network-based\nvisual analytics for miRNA functional analysis and systems biology.\nNucleic Acids Res. 48, W244–W251 (2020).\n47. Kanehisa, M. & Goto, S. KEGG: Kyoto Encyclopedia of Genes and\nGenomes. Nucleic Acids Res. 28,2 7–30 (2000).\n48. Nagai, T. et al. Focal adhesion kinase-mediated sequences, including\ncell adhesion, inflammatory response, and fibrosis, as a therapeutic\ntarget in endometriosis. Reprod. Sci. 27, 1400–1410 (2020).\n49. Bora, G. & Yaba, A. The role of mitogen-activated protein kinase\nsignaling pathway in endometriosis. J. Obstet. Gynaecol. Res. 47,\n1610–1623 (2021).\n50. Uimari, O. et al. Genome-wide genetic analyses highlight mitogen-\nactivated protein kinase (MAPK) signaling in the pathogenesis of\nendometriosis. Hum. Reprod. 32, 780–793 (2017).\n51. Solayman, M. H. M. et al. Identi fication of suitable endogenous\nnormalizers for qRT-PCR analysis of plasma microRNA expression in\nessential hypertension. Mol. Biotechnol. 58, 179–187 (2016).\n52. Leva, G. D. & Croce, C. M. miRNA pro filing of cancer. Curr. Opin.\nGenet. Dev. 23,3 –11 (2013).\n53. Vasu, S. et al. MicroRNA signatures as future biomarkers for diagnosis\nof diabetes states. Cells 8, 1533 (2019).\n54. Yan, L. et al. miR-21-5p induces cell proliferation by targeting TGFBI in\nnon-small cell lung cancer cells. Exp. Ther. Med. 16, 4655–4663\n(2018).\n55. Park, S., Yang, H. D., Seo, J., Nam, J. & Nam, S. hnRNPC induces\nisoform shifts in miR-21-5p leading to cancer development. Exp. Mol.\nMed. 54, 812–\n824 (2022).\n56. González-López, P. et al. Role of miR-15a-5p and miR-199a-3p in the\ninflammatory pathway regulated by NF-kB in experimental and human\natherosclerosis. Clin. Transl. Med. 13, e1362 (2023).\n57. Liu, D. & Guan, Y. Mechanism of action of miR-15a-5p and miR-152-\n3p in paraquat-induced pulmonary fibrosis through Wnt/β-catenin\nsignaling mediation. PeerJ 12, e17662 (2024).\n58. Pritchard, C. C. et al. Blood cell origin of circulating microRNAs: a\ncautionary note for cancer biomarker studies. Cancer Prev. Res. 5,\n492–497 (2012).\n59. Eyileten, C. et al. Thrombosis-related circulating miR-16-5p is\nassociated with disease severity in patients hospitalized for COVID-\n19. RNA Biol. 19, 963–979 (2022).\n60. Geng, X., Guo, X., Wang, T., Xu, J. & Jiang, L. miR-92a-3p regulates\nneuropathic pain and neuroinflammation by regulating the expression\nof WNT5A. J. Neuroimmunol. 407, 578695 (2025).\n61. Fujiwara, M. et al. microRNA-92a promotes CNS autoimmunity by\nmodulating the regulatory and in flammatory T cell balance. J. Clin.\nInvest. 132, e155693 (2022).\n62. Patil, A. H. & Halushka, M. K. miRge3.0: a comprehensive microRNA\nand tRF sequencing analysis pipeline. NAR Genom. Bioinform. 3,\nlqab068 (2021).\n63. Marcel, M. Cutadapt removes adapter sequences from high-\nthroughput sequencing reads. EMBnet.journal 17,1 0–12 (2011).\n64. Kozomara, A., Birgaoanu, M. & Grif fiths-Jones, S. miRbase: from\nmicroRNA sequences to function. Nucleic Acid Res. 47, D155–D162\n(2019).\n65. Langmead, B., Trapnell, C., Pop, M. & Salzberg, S. L. Ultrafast and\nmemory-efficient alignment of short DNA sequences to the human\ngenome. Genome Biol. 10, R25 (2009).\n66. Love, M. I., Huber, W. & Anders, S. Moderated estimation of fold\nchange and dispersion for RNA-seq with DESeq2. Genome Biol. 15,\n550 (2014).\nAcknowledgements\nThe authors wish to express their gratitude to Jonathan Zhao and Frank\nZhang (both from Heranova Lifesciences) for their valuable input on the study\ndesign and contributions to the development of the manuscript. This study\nwas funded by the internal R&D budget of Heranova Lifesciences.\nAuthor contributions\nY. Yu and W.H. Wong led the research group, analyzed the data, and wrote\nthe manuscript. Y. Yu, W.H. Wong, X. Zhang, and F.Z. Bischoff\nconceptualized the study. X. Zhang and L. Zhu provided the samples while S.\nLu coordinated sample collection between the hospital and the laboratory.\nW.H. Wong and Y. Hu performed random forest analysis. Y. Yu, Y. Shen, and\nX. Xu performed molecular experiments.\nCompeting interests\nAll authors, except X.Z. and L.Z., are employees of Heranova Lifesciences, a\ncompany engaged in the commercial development of a non-invasive test for\nendometriosis. F.Z.B. holds stock options of Heranova Lifesciences. The\nauthors declare no other conflicts of interest, financial or otherwise.\nhttps://doi.org/10.1038/s44294-025-00116-5 Article\nnpj Women's Health |            (2025) 3:67 9\n\nAdditional information\nSupplementary informationThe online version contains\nsupplementary material available at\nhttps://doi.org/10.1038/s44294-025-00116-5\n.\nCorrespondenceand requests for materials should be addressed to\nXinmei Zhang or Farideh Z. Bischoff.\nReprints and permissions informationis available at\nhttp://www.nature.com/reprints\nPublisher’s note Springer Nature remains neutral with regard to\njurisdictional claims in published maps and institutional affiliations.\nOpen Access This article is licensed under a Creative Commons\nAttribution-NonCommercial-NoDerivatives 4.0 International License,\nwhich permits any non-commercial use, sharing, distribution and\nreproduction in any medium or format, as long as you give appropriate\ncredit to the original author(s) and the source, provide a link to the Creative\nCommons licence, and indicate if you modi fied the licensed material. You\ndo not have permission under this licence to share adapted material\nderived from this article or parts of it. The images or other third party\nmaterial in this article are included in the article ’s Creative Commons\nlicence, unless indicated otherwise in a credit line to the material. If material\nis not included in the article’s Creative Commons licence and your intended\nuse is not permitted by statutory regulation or exceeds the permitted use,\nyou will need to obtain permission directly from the copyright holder. To\nview a copy of this licence, visit http://creativecommons.org/licenses/by-\nnc-nd/4.0/\n.\n© The Author(s) 2025\nhttps://doi.org/10.1038/s44294-025-00116-5 Article\nnpj Women's Health |            (2025) 3:67 10","source_license":"CC0","license_restricted":false}