Complement and coagulation cascade cross-talk in endometriosis and the potential of JAK inhibitors – a network meta-analysis

26

27 Lack of knowledge on the processes driving endometriosis hinders early detection and 28 therapy development. Our purpose was to identify key molecular events involved in lesion 29 formation across diverse populations and to detect transcriptomic changes in eutopic 30 endometrium that accompany endometriosis. 31

32 We searched Gene Expression Omnibus and ArrayExpress and performed differential gene 33 expression analysis and a network meta-analysis on nine qualifying datasets. Those contained 34 transcriptomic data on: 114 ectopic endometrium samples (EL), 138 eutopic endometrium 35 samples from women with endometriosis (EEM) and 79 eutopic endometrium samples from 36 women without endometriosis (EH). Gene ontology and enrichment analysis was performed 37 in DA VID, Metascape and Cytoscape and drug repurposing was done in CMap. 38

39 EEM compared to EH upregulated CCL21 and downregulated BIRC3, CEL and LEFTY1 40 genes (|log2FC|>0.5, p<0.05). EL showed increased expression of complement and serpin 41 genes (EL vs EEM: C7, logFC = 3.38, p <0.0001; C3, logFC = 2.40, p<0.0001; SERPINE1, 42 logFC = 1.02; p<0.05; SERPINE2, logFC = 1.54, p<0.001) and mast cells markers (EL vs 43 EEM: CPA3, logFC = 1.54, p<0.0001, KIT, logFC=0.74, p<0.001). Functional enrichment 44 analysis highlighted complement and coagulation, inflammation, angiogenesis and ECM as 45 drivers of endometriosis. Pharmacogenomic analysis indicated JAK, CDK and topoisomerase 46 inhibitors as therapy targets. 47

48 Our results suggest an interplay between complement and coagulation, mast cells, ECM and 49 JAK/STAT3 pathway in endometriosis. We underscore the significance of complement C3 50 . CC-BY-NC 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted March 25, 2025. ; https://doi.org/10.1101/2025.03.25.25324597doi: medRxiv preprint 3 and propose JAK inhibitors as therapy candidates. Detected expression differences between 51 EEM and EH are important for the development of diagnosis via endometrial biopsy. 52 53

endometriosis, eutopic and ectopic endometrium, network meta-analysis, 54 complement and coagulation, mast cells, JAK inhibitors 55 56 Abbreviations: endometriosis (EM), endometrium from healthy controls (EH), endometrium 57 from women with endometriosis (EEM), endometrial lesions (EL), differentially expressed 58 genes (DEGs), Gene Expression Omnibus (GEO) 59 60 WHAT IS ALREADY KNOWN ON THIS TOPIC 61 • Pathways and genes involved in endometriosis lesions formation are not well 62 characterised. Studies encompassing diverse patients populations are missing. 63 WHAT THIS STUDY ADDS 64 • This study reveals the transcriptomic profile of endometriosis, obtained via integration 65 of nine different datasets spanning various ethnicities and demographics. It 66 demonstrates the importance of complement and coagulation cascades, mast cells and 67 JAK/STAT3 pathway in lesion development. Our meta-analysis identifies 68 transcriptomic differences in eutopic endometrium of women with and without 69 endometriosis which include changes in CCL21, BIRC3, CEL and LEFTY1 70 expression. 71 HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY 72 • Our comprehensive analysis of endometriosis transcriptomic profile highlights genes 73 and pathways that should be explored further as disease biomarkers. JAK inhibitors 74 currently used in clinic in other autoimmune diseases show treatment potential. Gene 75 expression differences between eutopic endometrium of women with and without 76 endometriosis should be further explored as biomarkers in endometrial biopsy. 77 . CC-BY-NC 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted March 25, 2025. ; https://doi.org/10.1101/2025.03.25.25324597doi: medRxiv preprint 4

78 Lack of knowledge on the key processes that drive endometriosis hinders its early detection 79 and therapy development. There is a need to define those molecular events and to understand 80 how they interact to foster lesion implantation and maintenance. 81 Endometriosis is a chronic and complex disease currently showing a median diagnostic delay 82 of 7- 9 years 1,2. There has been a significant progress in the development of endometriosis 83 imaging protocols3, however, laparoscopy remains a gold standard for final diagnosis. There 84 is a need to explore the less invasive endometrial biopsy option. To consider this strategy, the 85 in-depth knowledge on the molecular differences in eutopic endometrium of healthy controls 86 and women with endometriosis is needed. 87 Several attempts have been made at delineating disease biomarkers, but to date this has not 88 yet proven successful. Various omics technologies enabled identification of key genes related 89 to the pathophysiology of endometriosis. However, a consensus has not yet been reached, and 90 we are still missing the focal points on which to concentrate the therapeutic endeavors. A 91 multi-cohort analysis is needed to address the issue in an unbiased and comprehensive 92 manner. 93 In this article we aimed to better understand complex events that underlie endometriotic 94 lesion formation and progression. To achieve this, we systematically reviewed endometriosis 95 data and performed network meta-analysis on chosen datasets. We generated a transcriptomic 96 profile of endometriosis, determined the key pathways involved in lesion formation and 97 explored possible drug candidates for endometriosis therapy. 98 99 . CC-BY-NC 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted March 25, 2025. ; https://doi.org/10.1101/2025.03.25.25324597doi: medRxiv preprint 5

100 Search strategy and study selection 101 Gene Expression Omnibus (GEO) and ArrayExpress were searched using terms 102 “endometriosis” and “Homo sapiens” and filtered with terms "Expression profiling by array" 103 or "Expression profiling by high throughput sequencing". MEDLINE/Pubmed was searched 104 for publications that correspond to those publicly deposited datasets. 105 106 Studies included in the analysis had to contain at least two tissues of interest: ectopic 107 endometrium - endometrial lesion (EL), eutopic endometrium from women without 108 endometriosis (EH) or eutopic endometrium from women with endometriosis (EEM). 109 Inclusion criteria were predefined and stringent to minimize the risk of bias, focusing on 110 datasets using RNA-Seq or microarray technologies, with available raw data. Transcriptomic 111 analysis had to be performed directly on human endometrial tissue, that had not been 112 subjected to any manipulation or cell isolation prior to RNA extraction. Samples had to be 113 taken from patients not on hormonal treatment in three months preceding tissue collection. 114 The presence or absence of endometriosis had to be confirmed with laparoscopy for samples 115 to be included in our study. Only datasets with accompanying publication were considered to 116 ensure all information about samples was available. Datasets with incomplete information 117 were excluded to reduce variability and minimize errors. A full list of inclusion / exclusion 118 criteria together with the PRISMA selection flowchart are summarized in Table S1 and Fig. 1. 119 Two independent reviewers screened datasets for relevance, and any discrepancies were 120 resolved in discussion with a third reviewer. PRISMA guidelines were followed, and study 121 protocol was registered in PROSPERO (ID CRD42024548098). 122 123 . CC-BY-NC 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted March 25, 2025. ; https://doi.org/10.1101/2025.03.25.25324597doi: medRxiv preprint 6 Data extraction and Differential Gene Expression 124 Each dataset was analyzed individually to ensure that data-specific preprocessing and 125 normalization steps were applied appropriately. For microarray datasets, the raw data files 126 were retrieved from the GEO repository using the R package “GEOquery”. The 127 preprocessing of microarray data was conducted following the manufacturer's protocols. 128

correction and quantile normalization were applied for all array data. To adjust 129 for differences in library size and transcript length, the raw reads count from RNA-Seq 130 datasets were normalized and scaled using the average transcript length for each sample. 131 Following this, library size normalization was performed using the Trimmed Mean of M-132 values (TMM). After preprocessing, group comparisons were conducted on the normalized 133 datasets to identify differentially expressed genes (DEGs) between experimental groups. For 134 this purpose, the “limma” package in R was utilized. The analysis generated log fold change 135 (logFC) values and their corresponding standard error (SE) values which were used for 136 further analysis. 137 138 Network meta-analysis 139 Network meta-analysis on gene expression was performed using “netmeta” package. 140 Although meta-analysis allows for the determination of both direct and indirect effects, in our 141 subsequent analyses, we focused on the combined effect to maximize the quality of the 142 analyzed data and reduce the influence of less reliable direct or indirect effects. For 143 investigated difference measurement we used logFC and its corresponding standard error. 144 These were interpreted as the mean difference and the standard error of the mean difference, 145 respectively, which are widely used metrics in comparative gene expression studies. This 146 standardization ensures that the results are both interpretable and comparable across datasets. 147 . CC-BY-NC 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted March 25, 2025. ; https://doi.org/10.1101/2025.03.25.25324597doi: medRxiv preprint 7 We performed 7664 network meta-analysis for genes that occurred in each of datasets 148 included in the study. Genes with a p-value 0.5 was used to filter genes with biologically meaningful changes in 150 expression. 151 Risk of bias 152 To reduce the risk of bias, we included studies with raw data deposited and results published 153 in peer-review journals. Information from accompanying publication was used to ascertain 154 the quality of the study and to identify if the absence of endometriosis was properly 155 determined and to confirm that tissue did not undergo any manipulation prior to RNA 156 isolation. 157 Heterogeneity was evaluated using I² statistics and Cochran’s Q-test, while sensitivity 158 analyses validated the robustness of findings. Funnel plots were generated to assess bias. 159 These measures ensured a thorough evaluation of potential biases, enhancing the reliability 160 and validity of the meta-analytic findings. 161 162 Gene ontology and pathway analysis 163 The list of DEGs obtained from the network meta-analysis was submitted to DA VID for gene 164 ontology and KEGG and Reactome pathways analysis. For functional clustering, we applied 165 a cut-off enrichment score of >2.5, p<0.05 and medium classification stringency. The same 166 list of DEGs was analyzed in Metascape v3.5.2024.0101 and the most enriched terms were 167 visualized in Cytoscape v3.10.2. 168 169 Computational pharmacogenomics 170 To identify pharmacological compounds likely to reverse endometriosis gene signature, we 171 queried drug repurposing reference database - CMap. We submitted a list of 150 up- and 150 172 . CC-BY-NC 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted March 25, 2025. ; https://doi.org/10.1101/2025.03.25.25324597doi: medRxiv preprint 8 down-regulated genes that had the highest combined logFC for EL vs EH and EL vs EEM 173 comparisons and p<0.05. 174 175 176 177

178 Characteristics of chosen studies 179 Nine datasets met the inclusion criteria (Fig. 1 and Table S1) and were included in the 180 analysis (Table 1). Those contained transcriptomic data on 114 ectopic endometrium samples 181 (EL), 138 eutopic endometrium samples from women with endometriosis (EEM) and 79 182 eutopic endometrium samples from women without endometriosis (EH). The absence of 183 endometriosis in the healthy (EH) group had to be confirmed during laparoscopic procedure. 184 Tissues were collected at three different continents and encompassed all types and stages of 185 endometriosis (clinical data in Table S2). 186 187 GEO Accession Number

Number of detected genes Ectopic endometrium (EL, n=114) Eutopic endometrium from patients with endometriosis (EEM, n=138) Eutopic endometrium from patients without endometriosis/ healthy control (EH, n=79) GSE2327134 high throughput sequencing 17488 - 7 7 GSE153740 & GSE1537395 high throughput sequencing 16840 & 17449 - 4 & 4 4 & 3 GSE1415496 microarrays 19746 79 49 21 . CC-BY-NC 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted March 25, 2025. ; https://doi.org/10.1101/2025.03.25.25324597doi: medRxiv preprint 9 GSE1340567 high throughput sequencing 18885 - 16 22 GSE256288 microarrays 12644 7 9 6 GSE378379 microarrays 21094 18 18 - GSE636410 microarrays 20857 - 21 16 GSE730511 microarrays 20857 10 10 - 188 Table 1. Characteristics of GEO datasets chosen for the network meta-analysis. 189

next to each dataset identifier refer to the original publication. 190 191 192 Transcriptomic profile of eutopic and ectopic endometrium. 193 Differential expression analysis was performed for each of the three comparisons: EL vs 194 EEM, EL vs EH and EEM vs EH. Using p 0.5, we identified 1109 DEGs 195 between EL and EEM, 1267 DEGs between EL and EH and 4 DEGs between EEM and EH 196 (Fig. 2). The heatmap of top 40 up and down regulated genes for all comparisons per dataset 197 is presented in Fig. 2D. The full list of network meta-analysis results is deposited in 198 Supplementary Dataset S1. 199 Meta-analysis revealed that transcriptomic profile of lesions was profoundly different from 200 that of eutopic endometrium (Fig. 2A-B) while the eutopic endometrium from women with 201 (EEM) and without endometriosis (EH) differed in the expression of four genes only (Fig. 202 2C). BIRC3, CEL and LEFTY1 were significantly less expressed in endometrium of women 203 with endometriosis than without (logFC = -0.79, p = 0.0051; logFC = -0.52, p = 0.0051; 204 . CC-BY-NC 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted March 25, 2025. ; https://doi.org/10.1101/2025.03.25.25324597doi: medRxiv preprint 10 logFC = -0.61, p = 0.0099, respectively). CCL21 was significantly higher in EEM versus EH 205 (logFC=0.59, p = 0.0255) and even higher when EL with EEM was contrasted (logFC=1.57, 206 p < 0.0001, Fig. 2C). C-C motif chemokine ligand 21 ( CCL21) is an inflammatory mediator 207 associated with moderate to severe endometriosis 12, however to-date its use as a disease 208 biomarker has failed. Our results showed a directional increase of CCL21 from endometrium 209 of healthy patients through that of endometriosis sufferers to lesions themselves indicating its 210 role in the eutopic endometrium inflammation in patients with endometriosis. 211 212 Pathways contributing to lesion development 213 In further analysis, we selected genes that showed differential expression in both EL vs. EEM 214 and EL vs. EH comparisons (the intersection of the sets, Fig. 2E) and exhibited the same 215 direction of expression. For p0.5 we obtained a list of 989 DEGs: 536 216 upregulated and 453 downregulated, on which we performed functional annotation and 217 enrichment analyses (Fig. 3A and detailed in Table S3A). Results presented below satisfied a 218 p value below 0.0001. Those analyses revealed that most biological processes involved in the 219 formation of endometriotic lesions were linked to cell adhesion (6.6%), inflammatory 220 response (5.5%) and regulation of angiogenesis (2.9%). The gene ontology molecular 221 functions analysis showed that the DEGs were significantly enriched in protein binding 222 (76.7%), identical protein binding (15%) and extracellular matrix structural constituent 223 (2.5%). In the cellular component, DEGs were mainly involved in extracellular exosome 224 (21.2%), extracellular region (21.2%) and extracellular space (17.4%). 225 KEGG analysis showed enrichment in complement and coagulation cascades (2.5%), 226 Staphylococcus aureus infection (2.3%) and cell adhesion molecules (2.7%). The analysis 227 against Reactome database revealed a key role of extracellular matrix organization (5.2%), 228 regulation of complement cascade (1.5%) and complement cascade (1.6%) (Table S3B). 229 . CC-BY-NC 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted March 25, 2025. ; https://doi.org/10.1101/2025.03.25.25324597doi: medRxiv preprint 11 A further pathway enrichment analysis was performed with Metascape (Fig. 3B and Table 230 S4) and visualized in Cytoscape (Fig. 3C). Tube morphogenesis, which relates to vascular 231 development, was the highest ranked result of enrichment analysis (Fig. 3B). Inflammatory 232 and hormonal response as well as locomotion and proliferation were amongst the top 10 most 233 enriched pathways with count value above 75. Functional annotation clustering revealed that 234 the complement cascade was the most enriched with the score of 4.14, followed by platelet 235 activation pathways, DNA remodeling and regulation of transcription and cell/cell-matrix 236 adhesion processes (enrichment score of 3.77, 3.02 and 2.71 respectively (Fig. 3D). 237 238 Altered expression of complement and coagulation pathway genes. 239 Complement and coagulation cascade was the most enriched KEGG pathway for the ectopic 240 versus eutopic endometrium comparison (Fig. 3D). Genes including C3, C2, C3 and SERPIN 241 superfamily genes involved in this pathway were amongst the most differentially expressed in 242 endometrial tissue (Fig. 4 and Dataset S1). 243 Complement genes C1QA (logFC = 1.12; 95%CI = 0.77, 1.47), C3 (logFC = 2.40; 95%CI = 244 1.43, 3.37) and C7 (logFC = 3.36; 95%CI = 2.50, 4.26) were upregulated in endometrial 245 lesions and showed high logFC values (Fig. 4A-C). C7 was the gene that showed the highest 246 level of upregulation among all examined genes. 247 Serpins regulate coagulation fibrinolysis processes13 and were implicated in the development 248 of endometriosis 14–16. Our network meta-analysis showed that serpin genes were 249 differentially expressed between endometrial lesions and eutopic endometrium. In 250 comparison with the above presented complement genes, serpin family genes were 251 characterized by more heterogenous expression between investigated datasets. SERPINE1 252 and SERPINE2 were upregulated (logFC = 1.02; 95%CI = 0.15, 1.90 and logFC = 1.54; 253 95%CI = 0.80, 2.27 respectively) while SERPINA5 was downregulated in lesions (logFC = -254 . CC-BY-NC 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted March 25, 2025. ; https://doi.org/10.1101/2025.03.25.25324597doi: medRxiv preprint 12 0.85; 95%CI = -1.52, -0.16, Fig. 4D-F). Detailed comparisons for each of the subgroups can 255 be found in Dataset S1. 256 257 Mast cells markers 258 Our data showed an upregulation in the expression of mast cells markers including CP A3 259 (logFC = 1.54; 95%CI = 0.96, 2.11), KIT (logFC = 0.74; 95%CI = 0.30, 1.18), MS4A6A 260 (logFC = 0.71; 95%CI = 0.30, 1.11) and markers of mast cells activation FCGR2B (logFC = 261 0.78; 95%CI = 0.20, 1.35) and S100A10 (logFC = 0.87; 95%CI = 0.38, 1.35, Fig. S3A-E). 262 The expression of MS4A4A and MS4A2 was also higher in lesions (Dataset S1). Higher 263 amounts of mast cells and their increased degranulation have been reported in endometrial 264 tissue of animal models and humans 17,18; mast cells colocalized to the vasculature of ovarian 265 endometriomas and they were found to promote endometrial cells migration in in vitro 266 assays19. 267 268 Repurposing JAK and CDK inhibitors for endometriosis therapy. 269 We used CMap drug repurposing software to find most probable connections between 270 therapeutic drugs and our network meta-analysis results. A median tau score value of 90 or 271 above is considered the typical threshold for assessing meaningful drug-induced effects. We 272 applied a median tau score cutoff at 95 and selected the top 15 hits. This analysis indicated 273 that the candidates most likely to reverse the endometriosis mRNA profile were cyclin-274 dependent kinase (CDK) inhibitors, JAK and topoisomerase inhibitors (Fig. 3E). 275 JAK/STAT3 pathway is thought to govern migratory and invasive properties of cells. Its 276 prolonged activation in breast cancer was linked with tumor development and resistance to 277 taxane and platinum therapy 20. Our results showed an increase in the expression of STAT5A 278 (logFC = 0.83; 95%CI = 0.57, 1.08) and STAT5B (logFC = 0.57; 95%CI = 0.37, 0.77) in 279 . CC-BY-NC 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted March 25, 2025. ; https://doi.org/10.1101/2025.03.25.25324597doi: medRxiv preprint 13 lesions compared with control tissue (Fig. S3 G-H). JAK3 significantly increased as well but 280 logFC value was below 0.5 (Dataset S1). 281 JAK inhibitor Ruxolitinib reduced epithelial ovarian cancer cell viability and caused growth 282 inhibition of Tam resistant breast cancer cells in vitro. It was shown to lower mRNA VEGF 283 expression and reduce the number of vessels and overall tumor weight in chorioallantoic 284 assay20. Ruxotinilib is currently tested in combination therapy for endometrial cancer but its 285 use in vitro or in preclinical models of endometriosis has not been reported. Tofacitinib, 286 another JAK inhibitor, showed a decrease in endometrial lesion size in mice and reduced 287 proliferation of endometrial cancer cells in vitro21. 288 289

290 Understanding the main pathways involved in endometriosis development is necessary for 291 the successful biomarker discovery and improved therapy outcomes. Combining data in 292 meta-analysis, we highlight pathogenetic mechanisms that are critical for lesion formation 293 regardless of endometriosis subtypes and patients’ characteristics. 294 295 Endometrium of women with endometriosis differs from healthy controls 296 We detected differences in gene expression between endometrium of healthy women and 297 those suffering from endometriosis thus showing that endometriosis can also affect eutopic 298 endometrium (Fig. 2C). CCL21 was upregulated whilst BIRC3 , LEFTY1 and CEL were 299 downregulated in EEM versus EH. Increased expression of CCL21 could suggest that this 300 gene takes part in inducing early inflammatory changes in eutopic endometrium in women 301 with endometriosis and that it continues its role in established lesions (Dataset S1). 302 Baculoviral IAP repeat containing 3 ( BIRC3) has not been studied in the context of 303 endometriosis. However, its mutations are often present in endometroid adenocarcinoma and 304 . CC-BY-NC 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted March 25, 2025. ; https://doi.org/10.1101/2025.03.25.25324597doi: medRxiv preprint 14 endometrial cancer 22. In the latter, the lower protein levels of Birc3 correlate with worse 305 patient survival. One could speculate that the decreased B irc3 expression in EEM could 306 contribute to the transformation of endometrium into lesions. Endometrial bleeding 307 associated factor (EBAF/ LEFTY1) partakes in the regulation of cyclical exfoliation of 308 endometrium and in decidualization. Healthy endometrium does not express LEFTY1 during 309 implantation window while endometrium of women suffering from endometriosis as well as 310 infertility showed its expression 23. Our results agree with that finding and suggest that the 311 higher LEFTY1 expression in EEM group could contribute to endometriosis-related 312 infertility. CEL gene encodes carboxyl ester lipase, which partakes in cholesterol and lipid-313 soluble vitamin ester hydrolysis. Its role so far is implicated in diabetes and hereditary 314 pancreatitis and progression of atherosclerosis. The CEL gene has not yet been studied in the 315 context of endometriosis. 316 317 The complement and coagulation cascade in lesion formation. 318 We further focused on delineating the expression profile that can differentiate ectopic 319 endometrium from eutopic endometrium from women with and without endometriosis (Fig. 320 2). Gene ontology analyses highlighted crucial events accompanying lesion formation. Those 321 were immune system activation, angiogenesis, regulation of transcription, response to 322 hormones and cytokines, cell adhesion and ECM – cell surface interactions (Fig. 3). The 323 importance of immune system deregulation in endometriosis has been reported previously; 324 various inflammatory phenotypes have been associated with increased risk of 325 endometriosis24,25. Our result showed that the complement system and platelet coagulation 326 are the two most enriched pathways in endometriosis (Fig. 3D). Both processes are essential 327 in natural endometrium growth and shedding cycle. The fact that both pathways are the most 328 enriched agrees with the current theory that women prone to endometriosis are likely to have 329 . CC-BY-NC 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted March 25, 2025. ; https://doi.org/10.1101/2025.03.25.25324597doi: medRxiv preprint 15 a different, dysregulated peritoneal microenvironment. The complement system is a mediator 330 of tissue growth and regeneration and its activation has for a long time been implicated in the 331 development of autoimmune disease and in promoting tumor growth. Its dysregulation could 332 therefore provide means for immunosurveillance escape and facilitate the implantation of 333 lesions. Its importance in the development of endometriosis has been suspected since the 334 80’s26 and confirmed more recently 27,28. Higher amounts of C1, C3 and C5 have been 335 detected in serum 29 and peritoneal fluid of women with endometriosis 30,31. V arious 336 complement proteins were shown to be present in epithelial cells of endometrial lesions and 337 ovarian cancer tumors. Its local synthesis and deposition has been correlated with progression 338 of various cancer types. 339 Our data revealed an increased mRNA expression of C1q, C2, C6 but especially C3 and C7 in 340 the lesions (Fig. 4A-C, Dataset S1). C7, a complement cascade member responsible for 341 initiation of membrane attack complex, was the most overexpressed gene with the highest 342 fold change in our comparison between diseased and control tissue suggesting its significant 343 role in lesion formation (Fig. 2A-B, Fig. 4C). C7 was found to contribute to inflammation 344 and tissue damage in endometriosis32; it has previously been shown overexpressed in ovarian 345 cancer25 and stromal cells of endometriomas33. 346 C3, a major effector, at which all complement pathways converge, was one of the most 347 differentially expressed genes in endometriosis (Fig. 4B, Figure S2). C3 dysregulation is 348 involved in most if not all inflammatory diseases; it has been found upregulated in cancer, 349 cardiac and neurological diseases, asthma and obesity. Patients with inflammatory bowel 350 disease had a higher expression of C3 in their intestinal tissue and this was thought to 351 contribute to chronic inflammation and tissue injury. A similar situation could occur in 352 endometriosis; increased C3 expression could contribute to inflammation-driven peritoneal 353 tissue injury, which in turn would facilitate lesion implantation. Glandular epithelial cells 354 . CC-BY-NC 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted March 25, 2025. ; https://doi.org/10.1101/2025.03.25.25324597doi: medRxiv preprint 16 found in endometrial lesions were shown to produce C3 locally 34. The activity of both 355 complement cascade members C3 and C4 was higher in serum of women with endometriosis 356 than those without 35. Increased amounts of C1, C3 and C5 were detected in serum 29 and 357 peritoneal fluid of women with endometriosis 30,31. Similarly, in lesion-bearing mice, C3 was 358 increased in their peritoneal fluid. Animals with C3 knockdown formed smaller endometrial 359 cysts and on average less of them34. 360 C3 seems pivotal to endometriosis pathology and given its strong upregulation and presence 361 both in tissue (Fig. 4B) as well as peritoneal fluid of endometriosis-sufferers, it poses an 362 interesting target for early diagnosis and therapy. It has already been proposed as an 363 endometriosis serum biomarker. Since in gastric cancer C3 tissue deposition correlated 364 negatively with plasma levels 36, further research is needed to confirm C3 suitability as 365 endometriosis biomarker. As far as treatment is concerned, C3 inhibitors have entered clinical 366 trials in anti-ovarian cancer therapy37 and treatment against inflammatory bowel disease. Our 367

indicate that biomarker and therapeutic potential of C3 should be studied in 368 endometriosis in more depth. 369 370 Our results revealed strong enrichment in coagulation cascade and showed a dysregulation of 371 SERPIN superfamily genes in endometrial lesions (Fig. 3D, Fig. 4D-F, Fig. S4) suggesting an 372 imbalance in the coagulation-fibrinolysis processes13. 373 SERPINE1 and SERPINE2 were increased in endometrial lesions (Fig. 4D-E). SERPINE1-374 encoded PAI-1 was found increased in deep infiltrating lesions 38 and correlated with ovarian 375 cancer proliferation and overall poor prognosis 39. PAI-1 inhibition resulted in decreased 376 lesion size40. SERPINE2 was implicated in modulating DNA damage response and favoring 377 cancer cell invasion 41. Its pro-metastatic activity has been linked to extracellular matrix 378 remodeling and increase in matrix metalloproteinase 9 (MMP-9) expression42,43. 379 . CC-BY-NC 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted March 25, 2025. ; https://doi.org/10.1101/2025.03.25.25324597doi: medRxiv preprint 17 Reduction in SERPINA5 expression was linked with an aggressive tumour phenotype and 380 poor prognosis in endometrial and ovarian serous carcinomas, in the latter it was correlated 381 with downstream activation of MMP9 44. Our meta-analysis revealed lower SERPINA5 and 382 higher MMP9 expression in endometrial lesions (Fig. 4F, Fig. S3F). Moreover, ECM 383 interactions were indicated in enrichment analysis (Fig. 3A-B, D). Taken together, our results 384 suggest that the imbalance in coagulation pathway may be affecting extracellular matrix 385 remodeling and contributing to metastatic-like potential of endometriotic cells; thereby 386 promoting lesion formation. 387 JAK/STAT3 pathway inhibition 388 Our search for associations between endometriosis gene signature and CMap reference 389 perturbagens highlighted the role of inhibitors of JAK, CDK and topoisomerase as possible 390 therapy candidates (Fig. 3E). Our meta-analysis revealed an increased expression of both 391 STAT5A and STAT5B in lesions compared with control tissue (Fig. S3 G-H). 392 Interestingly, increased C3 expression was shown to trigger JAK2/STAT3 pathway in gastric 393 cancer, which led to subsequent increase of cell proliferation. C3 inhibition with CR1 394 decreased that activation 36. Our results present a similar picture, complement C3 as well as 395 JAK/STAT3 pathway seems to play a role in the development of endometriosis. This 396 association needs further investigation. JAK inhibitors are already used in clinic for other 397 autoimmune disease therefore their repurposing should be further tested for endometriosis 398 therapy application. 399 400 Proposed pathways crosstalk in endometriosis 401 It has been proposed that both complement system and coagulation pathways are tightly 402 linked; coagulation factors have been reported to cleave and activate complement members 403 C3 and C545. On the other hand, C3 was shown to protect clots from fibrinolysis 46. Increased 404 . CC-BY-NC 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted March 25, 2025. ; https://doi.org/10.1101/2025.03.25.25324597doi: medRxiv preprint 18 amounts of C3 protein were shown to provoke mast cells activation and various mast cell 405 mediators were implicated in the regulation of coagulation and fibrinolysis in anaphylaxis 47. 406 Our meta-analysis revealed that endometrial lesions had a higher expression of mast cell 407 markers including KIT, CP A3 and MS4A6A, FCGR2B and S100A10 (Fig. S3A-E). An 408 increased mast cells burden was detected previously in animal and human endometrial 409 tissue19. Moreover, our results showed that endometrial lesions had a higher level of STAT5A 410 and STAT5B (Fig. S5G-H), members of JAK/STAT3 pathway, which regulate mast cells 48. 411 Targeting mast cells with JAK inhibitors for alleviation of symptoms of endometriosis has 412 been proposed almost two decades ago 49 but not much research has been carried out on the 413 topic since. Our current results fill this gap and suggest the use of JAK inhibitors as 414 immunomodulators in endometriosis. Interestingly, a cooperation between mast cells, 415 complement and coagulation pathways has been reported in an inflammatory disease - 416 chronic spontaneous urticaria50. Our analysis indicates that there exists an interplay between 417 complement and coagulation pathway, mast cells activation, ECM remodeling and 418 JAK/STAT3 pathway (summarized in Fig. 5). To the best of authors knowledge, this 419 relationship has not yet been studied in endometriosis and our results warrant a further in-420 depth look into those processes. 421 422 Strengths and Limitations 423 Our network meta-analysis enabled us to arrive at a consensus endometriosis signature. The 424 use of publicly deposited endometriosis transcriptomic data collected at three different 425 continents, spanning various ethnicities, age groups as well as various types and stages of 426 endometriosis enabled a comprehensive, unbiased and multi-demographic comparison of 427 endometriotic and control tissue. 428 . CC-BY-NC 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted March 25, 2025. ; https://doi.org/10.1101/2025.03.25.25324597doi: medRxiv preprint 19 The following limitations should be considered when interpreting our results. Our meta-429 analysis included only nine datasets because most of the studies lacked a control group, 430 included therapeutic intervention or performed RNA isolation on processed tissue. Secondly, 431 only published studies, where the absence of endometriosis was excluded by laparoscopy, 432 were included in this meta-analysis. Therefore, publication bias may have occurred although 433 none was indicated by the funnel plot. 434 435

and Clinical Implications 436 We highlight the role of complement and coagulation cascade in endometriosis and propose 437 an interplay between both those processes and mast cells, ECM interaction and JAK/STAT3 438 pathway that need further investigation. We underscore the significance of C3 and call for 439 further research into its diagnostic and therapeutic potential. Furthermore, we propose JAK 440 inhibitors discovered in drug repurposing analysis and validated in vitro, as potential therapy 441 candidates. 442 Our results show differences in expression in eutopic endometrium from patients with and 443 without endometriosis. Those should be further explored to understand if they contribute to 444 endometrial seeding. Detected gene differences may be potential biomarkers that could be 445 used in the less invasive endometriosis biopsy and should be further studied. 446 447

448 This research is part of the project No. 2022/47/P/NZ5/02484 co-funded by the National 449 Science Centre and the European Union Framework Programme for Research and Innovation 450 Horizon 2020 under the Marie Skłodowska-Curie grant agreement No. 945339. For the 451 purpose of Open Access, the author has applied a CC-BY public copyright licence to any 452 Author Accepted Manuscript (AAM) version arising from this submission;”. 453 . CC-BY-NC 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted March 25, 2025. ; https://doi.org/10.1101/2025.03.25.25324597doi: medRxiv preprint 20 454 Author contributions 455 M.A.G, W.M.F and conceived and supervised the study, A.R., M.S. performed GEO database 456 search, M.A.G., A.R. and M.S. performed study selection, K.S. contributed to database 457 search, A.R., J. C. performed differential genes expression, A.R., K.S. performed network 458 meta-analysis, M.A.G performed GO and enrichment analysis, M.A.G performed 459 pharmacogenomic analysis and cell culture in vitro experiments, M.A.G. and AR conducted 460 quality control of the data, M.A.G. obtained study funding, M.A.G. and A.R drafted the 461 manuscript, and W.M.F, M.A.G. and A.R revised the manuscript. 462 463 464 Conflict of Interests 465 Authors declare no conflict of interests. 466 467 468

469 1. Ghai, V., Jan, H., Shakir, F., Haines, P. & Kent, A. Diagnostic delay for superficial 470 and deep endometriosis in the United Kingdom. J. Obstet. Gynaecol. 40, 83–89 471 (2020). 472 2. Requadt, E., Nahlik, A. J., Jacobsen, A. & Ross, W. T. Patient experiences of 473 endometriosis diagnosis: A mixed methods approach. BJOG An Int. J. Obstet. 474 Gynaecol. 131, 941–951 (2024). 475 3. Quesada, J. et al. Endometriosis: A multimodal imaging review. European Journal of 476 Radiology vol. 158 (2023). 477 4. Li, X. yan et al. Combined proteomics and transcriptomics identifies serpin family C 478 . CC-BY-NC 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted March 25, 2025. ; https://doi.org/10.1101/2025.03.25.25324597doi: medRxiv preprint 21 member 1 associated protein as a biomarker of endometriosis. Ann. Med. 55, (2023). 479 5. Bane, K. et al. Endometrial DNA damage response is modulated in endometriosis. 480 Hum. Reprod. 36, 160–174 (2021). 481 6. Gabriel, M. et al. A relational database to identify differentially expressed genes in the 482 endometrium and endometriosis lesions. Sci. data 7, (2020). 483 7. Akter, S. et al. Machine Learning Classifiers for Endometriosis Using Transcriptomics 484 and Methylomics Data. Front. Genet. 10, (2019). 485 8. Crispi, S. et al. Transcriptional profiling of endometriosis tissues identifies genes 486 related to organogenesis defects. J. Cell. Physiol. 228, 1927–1934 (2013). 487 9. Khan, M. A., Sengupta, J., Mittal, S. & Ghosh, D. Genome-wide expressions in 488 autologous eutopic and ectopic endometrium of fertile women with endometriosis. 489 Reprod. Biol. Endocrinol. 10, (2012). 490 10. Burney, R. O. et al. Gene expression analysis of endometrium reveals progesterone 491 resistance and candidate susceptibility genes in women with endometriosis. 492 Endocrinology 148, 3814–3826 (2007). 493 11. Hever, A. et al. Human endometriosis is associated with plasma cells and 494 overexpression of B lymphocyte stimulator. Proc. Natl. Acad. Sci. U. S. A. 104, 495 12451–12456 (2007). 496 12. Sundqvist, J. et al. Endometriosis and autoimmune disease: association of 497 susceptibility to moderate/ severe endometriosis with CCL21 and HLA-DRB1. Fertil. 498 Steril. 95, 437–440 (2011). 499 13. Bianchini, E. P., Auditeau, C., Razanakolona, M., Vasse, M. & Borgel, D. Serpins in 500 Hemostasis as Therapeutic Targets for Bleeding or Thrombotic Disorders. Front. 501 Cardiovasc. Med. 7, 622778 (2020). 502 14. Li, Y. et al. The effects of coagulation factors on the risk of endometriosis: a 503 . CC-BY-NC 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted March 25, 2025. ; https://doi.org/10.1101/2025.03.25.25324597doi: medRxiv preprint 22 Mendelian randomization study. BMC Med. 21, (2023). 504 15. Wang, L. et al. The coagulation status in women of endometriosis with stage IV. BMC 505 Womens. Health 24, 1–6 (2024). 506 16. Casalechi, M. et al. The link between inflammation and hemostasis in endometriosis: a 507 call for research. J. Endometr. Uterine Disord. 3, 100040 (2023). 508 17. Paula, R. et al. The intricate role of mast cell proteases and the annexin A1-FPR1 509 system in abdominal wall endometriosis. J. Mol. Histol. 46, 33–43 (2015). 510 18. McCallion, A. et al. Estrogen mediates inflammatory role of mast cells in 511 endometriosis pathophysiology. Front. Immunol. 13, (2022). 512 19. Li, T. et al. Possible involvement of crosstalk between endometrial cells and mast cells 513 in the development of endometriosis via CCL8/CCR1. Biomed. Pharmacother. 129, 514 110476 (2020). 515 20. Kim, J. W., Gautam, J., Kim, J. E., Kim, J. A. & Kang, K. W. Inhibition of tumor 516 growth and angiogenesis of tamoxifen-resistant breast cancer cells by ruxolitinib, a 517 selective JAK2 inhibitor. Oncol. Lett. 17, 3981 (2019). 518 21. Kotlyar, A. M., Mamillapalli, R., Flores, V. A. & Taylor, H. S. Tofacitinib alters 519 STAT3 signaling and leads to endometriosis lesion regression. Mol. Hum. Reprod. 27, 520 (2021). 521 22. BIRC3 Mutation - My Cancer Genome. 522 https://www.mycancergenome.org/content/alteration/birc3-mutation/. 523 23. Tabibzadeh, S. et al. Dysregulated expression of ebaf, a novel molecular defect in the 524 endometria of patients with infertility. J. Clin. Endocrinol. Metab. 85, 2526–2536 525 (2000). 526 24. Vallvé-Juanico, J., Houshdaran, S. & Giudice, L. C. The endometrial immune 527 environment of women with endometriosis. Hum. Reprod. Update 25, 565–592 528 . CC-BY-NC 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted March 25, 2025. ; https://doi.org/10.1101/2025.03.25.25324597doi: medRxiv preprint 23 (2019). 529 25. Suryawanshi, S. et al. Complement pathway is frequently altered in endometriosis and 530 endometriosis-associated ovarian cancer. Clin. Cancer Res. 20, 6163–6174 (2014). 531 26. Bartosik, D., Damjanov, I., Viscarello, R. R. & Riley, J. A. Immunoproteins in the 532 endometrium: clinical correlates of the presence of complement fractions C3 and C4. 533 Am. J. Obstet. Gynecol. 156, 11–15 (1987). 534 27. Rahal, D., Andrade, F. & Nisihara, R. Insights into the role of complement system in 535 the pathophysiology of endometriosis. Immunol. Lett. 231, 43–48 (2021). 536 28. Agostinis, C. et al. Immunological Basis of the Endometriosis: The Complement 537 System as a Potential Therapeutic Target. Front. Immunol. 11, (2021). 538 29. Karadadas, E. et al. Evaluation of complement system proteins C3a, C5a and C6 in 539 patients of endometriosis. Clin. Biochem. 81, 15–19 (2020). 540 30. Sikora, J. et al. The role of complement components C1q, MBL and C1 inhibitor in 541 pathogenesis of endometriosis. Arch. Gynecol. Obstet. 297, 1495–1501 (2018). 542 31. Aslan, C. et al. Overexpression of complement C5 in endometriosis. Clin. Biochem. 543 47, 496–498 (2014). 544 32. Bae, S. J. et al. Identification and analysis of novel endometriosis biomarkers via 545 integrative bioinformatics. Front. Endocrinol. (Lausanne). 13, 1–16 (2022). 546 33. Rekker, K. et al. High-throughput mRNA sequencing of stromal cells from 547 endometriomas and endometrium. Reproduction 154, 93–100 (2017). 548 34. Agostinis, C. et al. The Inflammatory Feed-Forward Loop Triggered by the 549 Complement Component C3 as a Potential Target in Endometriosis. Front. Immunol. 550 12, (2021). 551 35. Kamer-Bartosi ń ska, A., Szyłło, K., Tchórzewski, H., Lewkowicz, P. & Lewy, J. 552 Evaluation of peripheral blood neutrophils activity in women with endometriosis. 553 . CC-BY-NC 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted March 25, 2025. ; https://doi.org/10.1101/2025.03.25.25324597doi: medRxiv preprint 24 Ginekol. Pol. 72, 431–436 (2001). 554 36. Yuan, K. et al. Complement C3 overexpression activates JAK2/STAT3 pathway and 555 correlates with gastric cancer progression. J. Exp. Clin. Cancer Res. 39, (2020). 556 37. Our Science - Modulating the Complement Cascade to Develop Novel Therapeutics | 557 Apellis Pharmaceuticals. https://apellis.com/our-science/science-of-c3/. 558 38. Alotaibi, F. T. et al. Plasminogen activator inhibitor-1 (PAI-1) expression in 559 endometriosis. PLoS One 14, (2019). 560 39. Mashiko, S. et al. Inhibition of plasminogen activator inhibitor-1 is a potential 561 therapeutic strategy in ovarian cancer. Cancer Biol. Ther. 16, 253–260 (2015). 562 40. Gómez, R. et al. Effects of hyperprolactinemia treatment with the dopamine agonist 563 quinagolide on endometriotic lesions in patients with endometriosis-associated 564 hyperprolactinemia. Fertil. Steril. 95, 882-888.e1 (2011). 565 41. Smirnova, T. et al. Serpin E2 promotes breast cancer metastasis by remodeling the 566 tumor matrix and polarizing tumor associated macrophages. Oncotarget 7, 82289–567 82304 (2016). 568 42. Chen, W. jin et al. Single-cell RNA-seq integrated with multi-omics reveals 569 SERPINE2 as a target for metastasis in advanced renal cell carcinoma. Cell Death Dis. 570 2023 141 14, 1–13 (2023). 571 43. Fayard, B. et al. The serine protease inhibitor protease nexin-1 controls mammary 572 cancer metastasis through LRP-1-mediated MMP-9 expression. Cancer Res. 69, 5690–573 5698 (2009). 574 44. Bijsmans, I. T. G. W. et al. Loss of SerpinA5 protein expression is associated with 575 advanced-stage serous ovarian tumors. Mod. Pathol. 2011 243 24, 463–470 (2010). 576 45. Amara, U. et al. Molecular Intercommunication between the Complement and 577 Coagulation Systems. J. Immunol. 185, 5628–5636 (2010). 578 . CC-BY-NC 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted March 25, 2025. ; https://doi.org/10.1101/2025.03.25.25324597doi: medRxiv preprint 25 46. Pryzdial, E. L. G., Leatherdale, A. & Conway, E. M. Coagulation and complement: 579 Key innate defense participants in a seamless web. Front. Immunol. 13, 918775 580 (2022). 581 47. Guilarte, M., Sala-Cunill, A., Luengo, O., Labrador-Horrillo, M. & Cardona, V. The 582 Mast Cell, Contact, and Coagulation System Connection in Anaphylaxis. Front. 583 Immunol. 8, 846 (2017). 584 48. Morales, J. K., Falanga, Y. T., Depcrynski, A., Fernando, J. & Ryan, J. J. Mast cell 585 homeostasis and the JAK–STAT pathway. Genes Immun. 11, 599 (2010). 586 49. D’cruz, O. J. & Uckun, F. M. Targeting mast cells in endometriosis with janus kinase 587 3 inhibitor, JANEX-1. Am. J. Reprod. Immunol. 58, 75–97 (2007). 588 50. Yanase, Y., Takahagi, S., Ozawa, K. & Hide, M. The Role of Coagulation and 589 Complement Factors for Mast Cell Activation in the Pathogenesis of Chronic 590 Spontaneous Urticaria. Cells 10, (2021). 591 592 593 FIGURES and LEGENDS 594 595 Figure 1. PRISMA flow diagram of Gene Expression Omnibus search for transcriptomic data 596 comparing eutopic and ectopic endometrial tissue. All datasets from ArrayExpress were also 597 deposited in Gene Expression Omnibus thus they were not further considered in the selection 598 process. 599 600 Figure 2. Differentially expressed genes identified by network meta-analysis. Vo l c a n o 601 plots showing differentially expressed genes for the following comparisons (A) endometriotic 602 lesions versus endometrium from women with endometriosis, (B) endometriotic lesions 603 . CC-BY-NC 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted March 25, 2025. ; https://doi.org/10.1101/2025.03.25.25324597doi: medRxiv preprint 26 versus endometrium from women without endometriosis, (C) endometrium from women with 604 and without endometriosis. For graphs A to C, red points signify genes with logFC less than -605 0.5 or more than 0.5 and p-value less than 0.05, blue points signifiy genes with logFC 606 belonging to -0.5 to 0.5 range and p-value less than 0.05, green points signify genes with 607 logFC less than -0.5 or more than 0.5 and p-value more than 0.05 and grey dots signify genes 608 with with logFC belonging to -0.5 to 0.5 range and p-value more than 0.05. Genes with p-609 values|2| (Fig. 2A-B) and p-values|0.5|(Fig. 2C) are 610 labelled. Heatmap with top 40 most differentially expressed genes per comparison per dataset 611 (D). Expression pattern of differentially expressed genes per comparison type (E). EL – 612 endometrial lesion, EEM - eutopic endometrium from women with endometriosis, EH – 613 eutopic endometrium from women without endometriosis. 614 Figure 3. Enriched pathways analysis, functional clustering and computational 615 pharmacogenomics of DEGs between endometriosis lesions and eutopic endometrium. 616 Gene ontology analysis using DA VID (A) reveals the importance of inflammation, cell 617 adhesion, angiogenesis and ECM remodeling. Metascape enrichment analysis (B) and 618 relationship network of enriched terms visualised in Cytoscape (C) show key events that 619 contribute to endometriosis development. Those include inflammatory and hormonal 620 response and proliferation and locomotion. Functional annotation clustering reports the 621 highest enrichment score for complement and coagulation cascade, platelet activation, DNA 622 remodeling and integrin mediated signaling respectively (D). Top 15 drug candidates 623 identified using a drug repurposing reference database – CMap. and showing median tau 624 value above 95. JAK, CDK and topoisomerase inhibitors are identified as potential 625 pharmacological targets for endometriosis therapy (E). 626 627 628 . CC-BY-NC 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted March 25, 2025. ; https://doi.org/10.1101/2025.03.25.25324597doi: medRxiv preprint 27 Figure 4. Differential gene expression across studies for selected genes from the 629 complement and coagulation pathway for the comparison between EL vs EEM. Forest 630 plot showing the expression of C1QA (complement C1q A chain) - A, C3 (complement C3) - 631 B, C7 (complement C7) - C, SERPINE1 (serpin family E member 1) - D, SERPINE2 (serpin 632 family E member 2) - E, SERPINA5 (serpin family A member 5) - F. Direct and indirect 633 comparisons from meta-analysis are presented in row number five and six. The indirect 634 comparisons had a low impact on the combined comparison outcome due to the analyses 635 being performed on datasets containing comparisons between EL and EEM (Table 2). 636 637 Figure 5. A schematic showing key molecular processes contributing to the development 638 of lesions. An interplay between complement and coagulation pathway further influences 639 mast cells activation, ECM remodelling and JAK/STAT3 pathway. JAK inhibitors carry 640 potential for endometriosis therapy. Genes names in green signify differentially expressed 641 genes for the comparison between EL and EEM. Created in BioRender.com. 642 . CC-BY-NC 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted March 25, 2025. ; https://doi.org/10.1101/2025.03.25.25324597doi: medRxiv preprint Databases identified through GEO searching (n = 122) Databases identified for screening (n = 104) Original datasets (n = 92) Datasets assessed for eligibility (n = 9) Duplicate databases (n = 18) Databases excluded due to lack of abstracts and/or articles (n = 12) Animal datasets (n = 7) Lack of group with endometriosis (n = 18) Only adenomiosis (n = 7) Without control group (n = 6) Single cell (n = 6) Without raw data (n = 5) Without mRNA (n = 4) Only cells/organoids (n = 13) Serum (n = 1) IdentificationScreeningIncluded . CC-BY-NC 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted March 25, 2025. ; https://doi.org/10.1101/2025.03.25.25324597doi: medRxiv preprint ED CBA . CC-BY-NC 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted March 25, 2025. ; https://doi.org/10.1101/2025.03.25.25324597doi: medRxiv preprint A B C D E . CC-BY-NC 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted March 25, 2025. ; https://doi.org/10.1101/2025.03.25.25324597doi: medRxiv preprint A B C D E F . CC-BY-NC 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted March 25, 2025. ; https://doi.org/10.1101/2025.03.25.25324597doi: medRxiv preprint Coagulation cascade Serpin superfamliy SERPINE1, SERPINE2, SERPINA3, SEPINA5 ECM remodelling MMP-9 Complement cascade C3, C1QA, C2, C6, C7 Mast cells KIT ,MS4A6A, CPA3, FCGR2B, S100A10 JAK/STAT3 pathway STAT5A STAT5B P P JAK inhibitors C3 . CC-BY-NC 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted March 25, 2025. ; https://doi.org/10.1101/2025.03.25.25324597doi: medRxiv preprint