Immunometabolic profiling of cervicovaginal lavages identifies key signatures associated with adenomyosis

In this study, we investigated immunoproteomic and metabolic differences between patients with and without adenomyosis. We recruited women undergoing hysterectomy for benign conditions (n = 108) and collected non-invasive CVL samples for downstream analysis. The women were stratified according to whether they were diagnosed with adenomyosis (n = 46) or no adenomyosis (n = 62) based on histopathological confirmation post-hysterectomy. Clinical and demographic information for this cohort is reported in Table 1 . The mean age of patients enrolled was 45.6 years, with no significant difference between groups ( p  = 0.65). All demographic information was found to not be significantly different between groups with and without adenomyosis, including: race ( p  = 0.75), ethnicity ( p  = 0.52), body mass index (BMI) ( p  = 0.89). Investigation of socioeconomic factors, education level ( p  = 0.20), income ( p  = 0.43), and employment status ( p  = 0.65) showed no significant differences. Medical history also revealed no significant differences: menopausal status ( p  = 0.62), previous dilation and curettage ( p  = 0.80), co-occurring conditions of endometriosis and fibrosis ( p  = 0.31 and p  = 0.94, respectively), parity ( p  = 0.09), heaviness of periods ( p  = 0.61), history of chronic pelvic pain ( p  = 0.83), history of endometriosis ( p  = 0.11), history of polycystic ovary syndrome (PCOS) ( p  = 0.50). For those patients that provided contraceptive data, there was no significant difference found between hormonal ( p  = 0.24) or non-hormonal ( p  = 0.99) contraceptive use between the groups. The only significant difference found was use of hormone IUD, with the no adenomyosis group more likely to have used it within the past 6 months ( p  = 0.02), however frequency of use was small (n = 8). See Table S1 for more information on contraceptive use within the two patient groups. Table 1 The association of demographics with adenomyosis All (n = 108) Adenomyosis (n = 46) No Adenomyosis (n = 62) p - value Age (mean (SD)) 45.55 (10.01) 45.52 (8.92) 45.58 (10.83) 0.65 Race (n = 107) 0.73 American Indian/Alaskan 5 (4.67) 2 (4.35) 3 (4.92) White/Caucasian 78 (72.90) 34 (73.91) 44 (72.13) Black or African American 11 (10.28) 6 (13.04) 5 (8.20) All Other 13 (12.15) 4 (8.70) 9 (14.75) Ethnicity (n = 108) 0.52 Non-Hispanic 76 (70.37) 34 (73.91) 42 (67.74) Hispanic 32 (29.32) 12 (26.09) 20 (32.26) Education (n = 105) 0.20 Less than high school 3 (2.86) 0 (0.00) 3 (5.08) High school diploma or GED 20 (19.05) 11 (23.91) 9 (15.25) Some college 24 (22.86) 14 (30.43) 10 (16.95) Association degree or Technical certificate 22 (20.95) 8 (17.39) 14 (23.73) Bachelor degree 22 (20.95) 9 (19.57) 13 (22.03) Master/Doctor degree 14 (13.33) 4 (8.70) 10 (16.95) Household income pre-tax ($) (n = 99) 0.43 <10,000 3 (3.03) 2 (4.35) 2 (4.35) 10,000-25,000 10 (10.10) 7 (15.22) 7 (15.22) 25,000-50,000 13 (13.13) 5 (10.87) 5 (10.87) 50,000-75,000 23 (23.23) 10 (21.74) 10 (21.74) 75,000-100,000 15 (15.15) 9 (19.57) 6 (11.32) >100,000 24 (24.24) 8 (17.39) 16 (30.19) Don’t know/refused 11 (11.11) 5 (10.97) 6 (11.32) Employment status (n = 103) 0.65 Yes 75 (72.82) 34 (75.56) 41 (70.69) No 28 (27.18) 11 (24.44) 17 (29.31) Marital status (n = 108) 0.12 Single/Divorced/Widowed 41 (37.96) 21 (45.65) 20 (32.26) Married 60 (55.56) 22 (47.83) 38 (61.29) Cohabitating 5 (4.63) 1 (2.17) 4 (6.45) Other 2 (1.85) 2 (4.35) 0 (0.00) Sexual orientation (n = 100) 0.99 Heterosexual 93 (93.00) 41 (93.18) 53 (92.86) Bisexual 2 (2.00) 1 (2.27) 1 (1.79) Homosexual 5 (5.00) 2 (4.55) 3 (5.36) Alcohol use (current) (n = 101) 0.72 Yes 50 (49.50) 23 (53.49) 27 (46.55) No 47 (46.53) 19 (44.19) 28 (48.28) Quit 4 (3.96) 1 (2.33) 3 (5.17) Tobacco use (within last 6 months) (n = 104) 0.13 Yes 14 (13.46) 7 (15.56) 7 (11.86) No 34 (32.69) 18 (40.00) 16 (27.12) Never 47 (45.19) 19 (42.22) 28 (47.46) Quit 9 (8.65) 1 (2.22) 8 (13.56) Douching (n = 94) 0.99 Yes 15 (15.96) 7 (17.07) 8 (15.38) No 79 (84.04) 34 (82.93) 45 (84.91) BMI (mean (SD) (n = 108)) 30.63 (7.55) 30.36 (6.58) 30.83 (8.23) 0.89 BMI (n = 108) 0.12 <25 23 (21.30) 6 (13.04) 17 (27.42) 25–29 38 (35.19) 21 (45.65) 17 (27.42) 30–34 19 (17.59) 9 (19.57) 10 (16.13) ≥35 28 (25.93) 10 (21.74) 18 (29.03) Menopausal status (n = 108) 0.62 Pre 89 (82.41) 39 (84.65) 50 (80.65) Post 19 (17.59) 7 (15.22) 12 (19.35) Previous dilation and curettage 0.80 Yes 19 (17.59) 9 (19.57) 10 (16.13) No 89 (82.43) 37 (80.43) 52 (83.87) Co-occurring conditions (n = 108) Endometriosis 21 (19.44) 11 (23.09) 10 (16.13) 0.31 Fibroids 70 (64.81) 30 (65.22) 40 (64.52) 0.94 Parity (n = 107) 0.09 0 22 (20.56) 7 (15.22) 15 (24.59) 1 9 (8.41) 2 (4.35) 7 (11.48) 2 23 (21.50) 9 (19.57) 14 (22.95) 3 27 (25.23) 11 (23.91) 16 (26.23) 4+ 26 (24.30) 17 (36.96) 9 (14.75) Heaviness of periods (n = 94) 0.61 Light 5 (5.32) 3 (6.82) 2 (4.00) Moderate 21 (22.34) 8 (18.18) 13 (26.00) Heavy 68 (72.34) 33 (75.00) 35 (70.00) Chronic pelvic pain history (n = 91) 0.83 Yes 50 (54.95) 22 (56.41) 28 (53.85) No 41 (45.05) 17 (43.59) 24 (46.15) Endometriosis history (n = 95) 0.11 Yes 28 (29.47) 16 (39.02) 12 (22.22) No 67 (70.53) 25 (60.98) 42 (77.78) PCOS history (n = 88) 0.50 Yes 10 (11.36) 3 (7.69) 7 (14.29) No 78 (88.64) 36 (92.31) 42 (85.71) Diabetes (n = 108) 0.15 Yes 22 (20.37) 6 (13.04) 16 (25.81) No 86 (79.63) 40 (86.96) 46 (74.19) Hypertension (n = 108) 0.99 Yes 25 (23.15) 11 (23.91) 14 (22.58) No 83 (76.85) 35 (76.09) 48 (77.42) Antibiotics (use within 3 months (n = 95)) 0.23 Yes 23 (24.21) 13 (30.95) 10 (18.87) No 72 (75.79) 29 (69.05) 43 (81.13) Combined contraceptives (use in past 6 months) Hormonal (n = 81) 0.24 Yes 26 (32.10) 9 (25.00) 17 (37.78) No 55 (67.90) 27 (75.00) 28 (62.22) Non-hormonal (n = 40) 0.99 Yes 1 (2.50) 1 (5.00) 0 (0.00) No 39 (97.50) 19 (95.00) 20 (100.00) Patients demographics show no significant difference in demographic, socioeconomic or medical history between adenomyosis and no adenomyosis patients. Values are n (%) unless stated as mean (SD). P -values were calculated using Wilcoxon rank-sum test for continuous variables and Fisher exact test for categorical variables. The association of demographics with adenomyosis Patients demographics show no significant difference in demographic, socioeconomic or medical history between adenomyosis and no adenomyosis patients. Values are n (%) unless stated as mean (SD). P -values were calculated using Wilcoxon rank-sum test for continuous variables and Fisher exact test for categorical variables. To study immunoproteomic differences between patients with and without adenomyosis, we investigated the levels of 72 soluble proteins in CVL samples, including cytokines, chemokines, growth factors, circulating cancer biomarkers, and immune checkpoint proteins. Hierarchical clustering analysis was not able to correctly predict adenomyosis based on global immunoproteomic profiles ( Figure S1 ). To identify proteins in the CVL samples that were altered in women diagnosed with adenomyosis we performed two-sample t- tests and fold change analysis. This revealed that eight soluble proteins were significantly ( p <0.05) different in the adenomyosis group compared to the no adenomyosis group ( Figure 1 A). Cell surface antigen CA19-9 ( p  = 0.031) and chemokines GROα ( p  = 0.021) and IP-10 ( p  = 0.015) were all statistically significantly downregulated in the adenomyosis group compared to women without adenomyosis ( Figure 1 B). Significantly upregulated proteins in women with adenomyosis included cell surface antigen CEA ( p  = 0.042) and cytokines: IL-9 ( p  = 0.047), IL-13 ( p  = 0.042), IL-36γ ( p  = 0.017), and TNFβ ( p  = 0.038), however these were not significant after FDR correction ( Figure 1 C). These cytokines, specifically IL-9 and IL-13, play an important role in type II immunity and act as chemoattractants for several immune cells, including mast cells and type 2 innate lymphoid cells (ILC2). In contrast, pro-inflammatory cytokines such as IL-1α, IL-1β, IL-6, IL-8, MIP-1β, RANTES, and TNFα were not significantly increased ( Figure S2 ). Levels of CA125 and HE4 were not significantly different between the adenomyosis and the no adenomyosis groups ( p  = 0.117 and p  = 0.09, respectively) ( Table S2 ). Overall, only a few proteins were altered in the CVL samples, mostly immunoregulatory and those involved in type II immune responses. Figure 1 Immunoproteomic data Immunoregulatory protein levels in the CVL samples were able to distinguish adenomyosis patients from patients with other benign conditions. (A) Volcano plot visualizing the 8 metabolites that were significant up/downregulated in adenomyosis compared to no adenomyosis ( p <0.05). (B) Scatterplots showing the 3 proteins that were significantly downregulated in adenomyosis, the chemokines IP-10 and GROα and the cell surface antigen CA19-9. Line represents the mean. P -values are shown. (C) Scatterplots showing the 5 proteins that were significantly upregulated in adenomyosis, the cell surface antigen CEA and cytokines: IL-36γ, TNFβ, IL-13 and IL-9. Line represents the mean. P -values are shown. Immunoproteomic data Immunoregulatory protein levels in the CVL samples were able to distinguish adenomyosis patients from patients with other benign conditions. (A) Volcano plot visualizing the 8 metabolites that were significant up/downregulated in adenomyosis compared to no adenomyosis ( p <0.05). (B) Scatterplots showing the 3 proteins that were significantly downregulated in adenomyosis, the chemokines IP-10 and GROα and the cell surface antigen CA19-9. Line represents the mean. P -values are shown. (C) Scatterplots showing the 5 proteins that were significantly upregulated in adenomyosis, the cell surface antigen CEA and cytokines: IL-36γ, TNFβ, IL-13 and IL-9. Line represents the mean. P -values are shown. Next, we used liquid chromatography and mass spectroscopy to determine the metabolic profiles of the CVLs collected from women with benign conditions and identify metabolic differences that may be present between women with and without adenomyosis. Global metabolic analysis detected 912 metabolites, 784 fully characterized compounds and 128 partially characterisable or uncharacterized, within the CVL samples ( Figure S3 ). Global metabolic profiles of patients with adenomyosis and no adenomyosis were created by the data reduction method partial least square-discriminant analysis (PLS-DA). To construct the PLS-DA, two components were used, accounting for 11.6 and 11.1% of the variance of the data. Slight separation with some overlap in overall metabolic profiles was observed between the adenomyosis and no adenomyosis groups ( Figure 2 A). Figure 2 Global metabolic profiles Global metabolomic profiles revealed that there is a unique metabolic signature associated with adenomyosis compared to patients with no adenomyosis, made up of mostly amino acids. (A) Partial least-squares discriminant analysis comparing patients with adenomyosis to no adenomyosis, showing separation between the 2 groups revealing a unique metabolomic signature belonging to adenomyosis. (B) Bar chart demonstrating the difference in superpathway distribution of metabolites that were detected overall in CVL samples compared to significantly altered metabolites ( p 2.0) and FDR-corrected significantly altered metabolites ( q <0.1 and q <0.05). (C) Volcano plot visualizing the 82 metabolites that were significantly up/downregulated in adenomyosis compared to no adenomyosis ( p 2.0). 1 significantly downregulated and 81 significantly upregulated. (D) Volcano plot visualizing the 39 metabolites that were significantly upregulated in adenomyosis compared to no adenomyosis and passed FDR-correction ( q 2.0). (E) Hierarchical clustering analysis heatmap using Pearson clustering and Ward linkage for metabolites and supervised clustering for patient samples – only showing the top 25 most significant metabolites, as determined by t -test, color coded by superpathway. (F) Scatterplots highlighting the significant upregulation of key metabolites in adenomyosis with FDR-correction, color coded by superpathway (N6-acetyllysine, N-formylmethionine, argininate, pipecolate, 2-hydroxyadipate). Line represents the mean. P - and q -values are shown. Global metabolic profiles Global metabolomic profiles revealed that there is a unique metabolic signature associated with adenomyosis compared to patients with no adenomyosis, made up of mostly amino acids. (A) Partial least-squares discriminant analysis comparing patients with adenomyosis to no adenomyosis, showing separation between the 2 groups revealing a unique metabolomic signature belonging to adenomyosis. (B) Bar chart demonstrating the difference in superpathway distribution of metabolites that were detected overall in CVL samples compared to significantly altered metabolites ( p 2.0) and FDR-corrected significantly altered metabolites ( q <0.1 and q <0.05). (C) Volcano plot visualizing the 82 metabolites that were significantly up/downregulated in adenomyosis compared to no adenomyosis ( p 2.0). 1 significantly downregulated and 81 significantly upregulated. (D) Volcano plot visualizing the 39 metabolites that were significantly upregulated in adenomyosis compared to no adenomyosis and passed FDR-correction ( q 2.0). (E) Hierarchical clustering analysis heatmap using Pearson clustering and Ward linkage for metabolites and supervised clustering for patient samples – only showing the top 25 most significant metabolites, as determined by t -test, color coded by superpathway. (F) Scatterplots highlighting the significant upregulation of key metabolites in adenomyosis with FDR-correction, color coded by superpathway (N6-acetyllysine, N-formylmethionine, argininate, pipecolate, 2-hydroxyadipate). Line represents the mean. P - and q -values are shown. The 912 detected metabolites included amino acids (n = 206, 23%), carbohydrates (n = 34, 4%), cofactors and vitamins (n = 32, 4%), energy (n = 12, 1%), lipids (n = 228, 25%), nucleotides (n = 67, 7%), peptides (n = 40, 4%), xenobiotics (n = 165, 18%), partially characterized metabolites (n = 9, 1%) and uncharacterized metabolites (n = 119, 13%). Analysis of superpathway distribution among the metabolites detected revealed a distribution across all superpathways – with amino acids accounting for only 23% of all metabolites. Next, we performed two sample t -tests to determine which metabolites differed significantly between patients with and without adenomyosis, and fold change analysis to identify metabolites that were up/downregulated. The threshold for fold-change analysis was set at 2.0 or greater to be classified as ‘altered’. A large proportion of significantly ( p <0.05) altered metabolites were amino acids (n = 43) accounting for more than half of those metabolites (52%) that were altered in adenomyosis compared to no adenomyosis. Furthermore, we performed false-discovery rate (FDR) correction of 10 and 5%, this amino acid signature remained dominant, with amino acids accounting for 56% (n = 22) and 69% (n = 9), respectively ( Figure 2 B). Fold change analysis and t -tests were combined to produce volcano plots that revealed 82 metabolites that were significantly altered ( p 2.0) in adenomyosis patients compared to no adenomyosis, with 1 metabolite being downregulated and 81 upregulated ( Figure 2 C and Table S3 ). Statistical analysis with FDR-correction of 10% resulted in 39 significantly upregulated metabolites in adenomyosis ( Figure 2 D). Supervised hierarchical clustering analysis (HCA) was performed to depict levels of metabolites within individual samples and identify whether a clear clustering pattern emerged between patients with adenomyosis and patients without adenomyosis. HCA of the top 25 significant metabolites produced a heatmap with a distinct signature, that reflected an enriched metabolic pattern, for adenomyosis compared to no adenomyosis ( Figure 2 E). N6-acetyllysine ( p <0.0001, q  = 0.018), N-formylmethionine ( p  = 0.0001 and q  = 0.028), argininate ( p  = 0.0002 and q  = 0.029), and pipecolate ( p  = 0.0004 and q  = 0.029) were some of the significantly upregulated amino acids in adenomyosis that contributed to the amino acid signatures. These amino acids were selected as they had the most significant p - values, were still significant after FDR-correction of 5% and were detected in at least 85% of adenomyosis patients ( Figure 2 F). Within the lipid superpathway, 2-hydroxyadipate ( p  = 0.0003 and q  = 0.029) was the only lipid that was significantly upregulated, remained significant after FDR-correction of 5%, and was detected in at least 85% of adenomyosis patients ( Figure 2 F). To summarise, our analyses show that adenomyosis is distinguished by a unique metabolic signature compared to patients with other benign conditions, particularly by an accumulation of amino acids. Next, we performed an enrichment analysis to identify the metabolic pathways that were likely to be altered in patients with adenomyosis compared to patients without adenomyosis based on relative level of metabolites within each group. This revealed that 32 metabolic pathways were significantly ( p <0.05) enriched in adenomyosis patients compared to patients without adenomyosis, the figure shows the top 25 significant pathways ( Figure 3 ). These pathways were associated with nucleotide (n = 1), lipid (n = 6), amino acid (n = 17), or energy metabolism (n = 3), and some were not characterized to a particular superpathway (n = 5). The nucleotide pathway pyrimidine metabolism ( p <0.0001), the lipid pathway carnitine synthesis ( p <0.0001), and the amino acid pathways histidine metabolism ( p <0.0001), and tryptophan metabolism ( p <0.0001) were among the most significantly enriched ( Figures 3 , 4 A, 4B, and 4C). Scatterplots are shown that represent key metabolites from the pathways; these were the most significant ( p <0.05), remained significant after FDR-correction to 10% ( q <0.1), and were detected in at least 85% of adenomyosis patients. 5,6-dihydrothymine ( p  = 0.0003 and q  = 0.029), thymine ( p  = 0.0017 and q  = 0.064), and N-carbamoylaspartate ( p  = 0.0034 and q  = 0.081) were all key nucleotides belonging to the pyrimidine metabolism pathway ( Figure 4 A). For carnitine synthesis, key metabolites were N6, N6, N6-trimethyllysine ( p <0.0001 and q  = 0.01), succinate ( p  = 0.0024 and q  = 0.069), and deoxycarnitine ( p  = 0.0049 and q  = 0.088) ( Figure 4 B). From histidine metabolism, 4-imidazoleacetate ( p  = 0.0025 and q  = 0.0685), formiminoglutamate ( p  = 0.0051 and q  = 0.0884), and histamine ( p  = 0.0328 and q  = 0.2030) were key metabolites; they remained significant after FDR-correction to 10% ( q <0.1), except histamine, and were detected in at least 70% of adenomyosis patients ( Figure 4 C). In total, enrichment analysis revealed that a number of different pathways are enriched within the adenomyosis patients, with the majority of these belonging to the amino acid superpathway. Pyrimidine metabolism, carnitine synthesis, and histidine metabolism were the most significantly enriched pathways. Figure 3 Enrichment analysis Enrichment analysis revealed that 32 pathways were significantly enriched in adenomyosis vs no adenomyosis. (A) Enrichment analysis of adenomyosis patients vs no adenomyosis patients revealed that 32 pathways were significantly ( p <0.05) enriched – top 25 pathways are shown here color coded by superpathway and p - value. Figure 4 Pathway analysis Pyrimidine metabolism, carnitine synthesis, and histidine metabolism were among the top significantly enriched pathways. (A) Diagram demonstrating a simplified version of the pyrimidine metabolism pathway and the detection of each metabolite in our analyses. Scatterplots of 3 of the significantly altered metabolites in adenomyosis that demonstrate the enrichment of the pyrimidine metabolism pathway (5,6-dihydrothymine, thymine, N-carbamoylaspartate) color coded by superpathway. Line represents the mean. P - and q -values are shown. (B) Diagram demonstrating a simplified version of the carnitine synthesis pathway and the detection of each metabolite in our analyses. Scatterplots of 3 of the significantly altered metabolites in adenomyosis that demonstrate the enrichment of the carnitine synthesis pathway (N6,N6,N6-trimethyllysine, succinate, deoxycarnitine) color coded by superpathway. Line represents the mean. P - and q -values are shown. (C) Diagram demonstrating a simplified version of the histidine metabolism pathway and the detection of each metabolite in our analyses. Scatterplots of 3 of the significantly altered metabolites in adenomyosis that demonstrate the enrichment of the histidine metabolism pathway (4-imidazoleacetate, formiminoglutamate, histamine) color coded by superpathway. Line represents the mean. P - and q -values are shown. Green arrows indicate metabolites that were significantly upregulated ( p <0.05). Bold text represents metabolite shown in scatterplots. Dashed arrows indicate multiple steps in pathway. Enrichment analysis Enrichment analysis revealed that 32 pathways were significantly enriched in adenomyosis vs no adenomyosis. (A) Enrichment analysis of adenomyosis patients vs no adenomyosis patients revealed that 32 pathways were significantly ( p <0.05) enriched – top 25 pathways are shown here color coded by superpathway and p - value. Pathway analysis Pyrimidine metabolism, carnitine synthesis, and histidine metabolism were among the top significantly enriched pathways. (A) Diagram demonstrating a simplified version of the pyrimidine metabolism pathway and the detection of each metabolite in our analyses. Scatterplots of 3 of the significantly altered metabolites in adenomyosis that demonstrate the enrichment of the pyrimidine metabolism pathway (5,6-dihydrothymine, thymine, N-carbamoylaspartate) color coded by superpathway. Line represents the mean. P - and q -values are shown. (B) Diagram demonstrating a simplified version of the carnitine synthesis pathway and the detection of each metabolite in our analyses. Scatterplots of 3 of the significantly altered metabolites in adenomyosis that demonstrate the enrichment of the carnitine synthesis pathway (N6,N6,N6-trimethyllysine, succinate, deoxycarnitine) color coded by superpathway. Line represents the mean. P - and q -values are shown. (C) Diagram demonstrating a simplified version of the histidine metabolism pathway and the detection of each metabolite in our analyses. Scatterplots of 3 of the significantly altered metabolites in adenomyosis that demonstrate the enrichment of the histidine metabolism pathway (4-imidazoleacetate, formiminoglutamate, histamine) color coded by superpathway. Line represents the mean. P - and q -values are shown. Green arrows indicate metabolites that were significantly upregulated ( p <0.05). Bold text represents metabolite shown in scatterplots. Dashed arrows indicate multiple steps in pathway.

In this study, we investigated the immune protein and metabolic profiles of CVLs from women with benign gynecologic conditions, grouped according to presence (n = 46) or absence (n = 62) of adenomyotic lesions and we identified unique immunometabolic signatures associated with adenomyosis. The evaluation of immunometabolic profiles associated with adenomyosis can help us to better understand this enigmatic condition, as well as identify potential biomarker candidates for diagnostic development. Our previous research has shown that non-invasive CVL sampling coupled with immunometabolomic analysis is successful in allowing us to better understand the pathophysiology of conditions such as HPV, cervical dysplasia, and cancer. 22 , 23 , 24 , 25 , 26 To our knowledge, this is the first study investigating soluble proteins present in CVL samples from patients with and without adenomyosis. Out of 72 tested proteins, we identified eight potential protein biomarkers associated with adenomyosis. This included downregulation of two chemokines, IP-10 and GROα. IP-10 is a pro-inflammatory chemokine which has particular involvement in induction of chemotaxis and apoptosis. 27 , 28 GROα is an pro-oncogenic chemokine that plays a role in immune cell trafficking and regulation, 29 and has been shown to drive metastatic growth in cancer. 30 The downregulation of these two proteins elucidates that there is a state of immune dysregulation that may allow the excess cellular proliferation within the myometrium associated with adenomyosis. We also found downregulation of CA19-9, a mucin that has been heavily investigated as a biomarker for various cancers. 31 , 32 , 33 There is no existing research that adequately explores the levels of CA19-9 in adenomyosis patients. One case study found CA19-9 to be elevated in the serum of one adenomyosis patient, 34 which was in contrast to our findings in lavage samples in this cohort, however that was a case study that investigated a single sample and thus may have limited generalisability. Carcinoembryonic antigen (CEA) is another cell-membrane antigen that is overexpressed in various malignancies including mucinous ovarian carcinoma. 35 Excess expression of CEA may be involved in reducing local immune response through inhibition of various immune cells. 36 Our findings reveal that CEA was increased in CVL samples of adenomyosis patients, suggesting that it may have the same role within adenomyotic lesions for dampening immune responses. Another study investigating levels of CEA in cervicovaginal fluids from patients with cervical condylomas and cervical intraepithelial neoplasia, found that CEA levels measured locally were more indicative than in serum, and were significantly different from healthy controls. 37 Although this provides evidence that CEA is a locally detected biomarker in women with adenomyosis, it may not have great clinical value as it is not specific for only adenomyosis. Several cytokines were also elevated in the samples from adenomyosis patients compared to patients without adenomyosis; this included IL-9 and IL-13. These cytokines are involved in promoting a type II immune response and act as chemoattractants for mast cells, 38 leading to the stimulation of tissue repairs and reduction in tissue inflammation. 39 In addition, IL-9 and IL-13 are produced by type 2 innate lymphoid cells (ILC2) which are also cells involved in the type II immune response. 40 ILC2s require IL-9 to increase their survival, and to amplify the ILC2 functions within type II immunity, particularly tissue repair. 41 Previous studies have shown that dysregulation of these immune responses can lead to fibrotic tissue growth. 39 This mechanism of altered tissue repair may be occurring in adenomyosis resulting in uncontrolled tissue growth/fibrosis that promotes the formation of adenomyotic lesions. 42 Furthermore, IL-36γ was increased, which conversely is associated with chronic inflammation. 43 It also has an immunoregulatory role by acting as a chemoattractant for neutrophils and other immune cells. 44 We have previously found IL-36γ to be elevated in cervicovaginal samples from patients with invasive cervical carcinoma 22 and bacterial vaginosis. 45 Finally, TNFβ was upregulated. This cytokine has previously been studied in endometriosis and was proposed as a factor for causing disruption to immune responses by promotion of inflammation and cellular proliferation. 46 TNFβ also has the ability to induce apoptosis within tumor cells while protecting normal cells. 47 The levels of CA125 and HE4, both previously investigated as potential biomarkers for adenomyosis, were not significantly different within the CVL samples in our cohort. This highlights the difference that can be observed in detection of these proteins systemically vs. locally. Although minimally invasive blood draws enable measuring these biomarkers in the serum, previous research has showed that these targets do not have high predictive accuracy for adenomyosis systemically. 48 These markers are not discriminatory for adenomyosis when measured in CVL samples. In total, alterations in cervicovaginal protein levels in adenomyosis patients reflect the dysregulation of immune responses. This is induced by attraction of mast cells and induction of ILC2s that promote type II immune responses, leading to the imbalance of tissue repair and inflammatory responses, allowing a state of excess proliferation and tissue growth to occur leading to adenomyotic lesion development within the myometrium. These results show that CVLs are a non-invasive method that can be utilized to quantify immune proteins with potential diagnostic value. However, the proteins investigated were not sensitive or specific enough alone to stratify patients with and without adenomyosis. Yet, these results show that future investigations in validation cohorts is warranted; combining protein markers in combination with each other or metabolites could lead to a robust diagnostic tool. Studies assessing the metabolic profiles of adenomyosis patients are also limited with only two small pilot reports published to date. One study used serum samples to investigate systemic metabolic alterations, 20 and the other investigated myometrial tissue samples for a local metabolic approach. 21 To our knowledge, we are the first to investigate metabolites associated with adenomyosis in CVL samples, which, because of the anatomical continuity of the female reproductive tract, allows successful investigation of the local microenvironment non-invasively. Our results reveal significant upregulation of metabolites in adenomyosis patients, with a large proportion of these being amino acids, revealing a robust amino acid signature associated with the condition. Previous research has linked amino acid signatures to cancer. Wang et al. showed that amino acid profiles from serum samples distinguished epithelial ovarian cancer patients from healthy controls. 49 Song et al., also demonstrated that the metabolic results from myometrial investigation in adenomyosis led to the discovery of significantly altered amino acids and a signature for adenomyosis, 21 which is in accordance with our findings in CVL. Compared to previous reports, our study resulted in a greater detection of metabolites (n = 912) and greater number of upregulated amino acids identified, likely because of a more advanced metabolomics platform being utilized. Dysregulation of metabolism has been widely recognized as a hallmark of cancer. 50 The uptake and metabolism of amino acids are aberrantly upregulated in many malignancies that display a greater need for amino acids. 51 Amino acids facilitate survival and proliferation of cancer cells under genotoxic, oxidative, and nutritional stress. 51 This suggests that the hallmarks shared by adenomyosis and cancer, particularly excess cellular proliferation, may be governed by upregulation of amino acids, creating a metabolic signature that can be detected both locally and systemically. We identified pathways, which may be dysregulated in adenomyosis and therefore provide insights into the pathophysiology of this complex condition. The most significantly enriched pathway in adenomyosis was pyrimidine metabolism. Pyrimidine metabolism is a nucleotide pathway associated with synthesis of nucleic acids, 52 a process upregulated during cellular proliferation. A study utilizing a mouse model of adenomyosis found that reduction in enzymatic activity related to the pyrimidine metabolism pathway, which decreases DNA synthesis and reduced incidence of adenomyosis in mice. 53 This preclinical report compliments our clinical findings and strongly supports that upregulation of pyrimidine metabolism is a key metabolic signature of adenomyosis. In addition, another enriched pathway was carnitine synthesis, which is essential for transportation of fatty acids into the mitochondria for fatty acid oxidation. 54 Previous research has linked fatty acid oxidation to the development of various cancers, via increased ATP production driving tumor growth. 55 Similar to our study, both pyrimidine metabolism and carnitine synthesis were highlighted as pathways that may play pivotal roles in adenomyosis in a previous study utilizing myometrial tissues, 21 thus supporting the cervicovaginal signatures identified in our study. Finally, we also identified histidine metabolism as a significantly enriched pathway within adenomyosis. Histamine synthesis from histidine can occur in mast cells, 56 which our immunoproteomic results show are likely recruited to adenomyotic lesions because of IL-9 and IL-13 production. Previous studies have shown that histamine produced by mast cells is linked to uterine contraction. 57 Excess release of histamine is also known to cause histamine intolerance within the female reproductive organs, and, in consequence, a key symptom of adenomyosis--dysmenorrhea, 58 likely because of uterine contractility. Herein, we have tested the largest number of protein and metabolic signatures in patients with adenomyosis to date, by utilising non-invasive local CVL sampling. Compared to other studies, we examined a relatively large cohort comprising 108 benign hysterectomy patients, 46 with adenomyosis and 62 without adenomyosis. Although adenomyosis is a condition highly co-occurring with other benign disorders, there were no significant differences in the co-occurrence of fibroids or endometriosis among the groups; nor any significant differences between demographics, socioeconomic background, or medical history (except hormone IUD use, p  = 0.02). These factors allowed us to identify a robust, unique signature associated with adenomyosis that warrants further research. In summary, through immunometabolic profiling of CVL samples, we identified key signatures associated with adenomyosis. The immunoproteomic data was not able to stratify patient groups, however we observed significant increases in cytokines and growth factors related to type II immunity and other immunoregulatory processes and corresponding metabolic pathways, thereby providing pathophysiological insights into adenomyosis (see graphical abstract). The metabolites and identified metabolic pathways were predictive of disease and have potential diagnostic value. Overall, the immunometabolic pattern validated a novel role for mast cells, ILC2s, proliferation, and the symptomology of adenomyosis. The immunometabolic pattern in CVL samples reflected pathophysiological changes in the upper female reproductive tract, thereby demonstrating the utility of CVL for detection of adenomyosis. Although we had a relatively large cohort size for a preliminary study, further research in large and diverse cohorts is required to validate our findings and improve generalisability. Another limitation of our study is the overlap between co-occurring benign gynecologic conditions which influences both our exposed group with adenomyosis and our comparison group; however, we did not observe significant differences in these conditions between our groups. Because of the high prevalence of other benign gynecologic conditions co-occurring with adenomyosis it is not feasible to investigate an exposure group that only has adenomyosis. These other gynecologic conditions may modify the outcomes we investigated. However, our study population is clinically representative of the patients who would be receiving an adenomyosis diagnosis and therefore, appropriate for this analysis. Future analysis that considers other pathologies that may alter the immunometabolic microenvironment are required in order to confirm the signature identified is diagnostic for adenomyosis. Our study also had several strengths, including the global untargeted investigation of metabolites. We detected 912 metabolites, with 784 of those being of known identity and 128 uncharacterized or partially characterized molecules. There is potential for further investigation of the uncharacterized metabolites in the future as they become characterized. To our knowledge this is the largest study to date resulting in the largest metabolomics analysis on adenomyosis.

Adenomyosis is a benign gynecologic condition characterized by the ectopic growth of endometrial tissue in the myometrium. It is a highly prevalent condition, with an estimated occurrence of 30–35% in symptomatic patients. 1 Symptoms include dysmenorrhea, chronic pelvic pain, and abnormal uterine bleeding 2 ; however, it is estimated that a third of cases are asymptomatic. 3 Adenomyosis is also associated with infertility, 4 as well as a number of negative obstetric outcomes, including: pre-eclampsia, pre-term birth, fetal malpresentation, and post-partum hemorrhage. 5 Current definitive diagnostic tools for adenomyosis rely on histopathology testing after hysterectomy. 3 Imaging techniques are currently being explored for diagnosis; however, this approach requires further investigation and optimization for reliable results. 6 Reliance on hysterectomy for diagnosis may introduce a bias, as women willing to undergo hysterectomy are more often older and have higher parity 2 ; therefore, adenomyosis prevalence is likely to be higher. Adenomyosis has a high co-occurrence rate with other gynecologic conditions, particularly endometriosis and fibroids. 7 The combination of bothersome symptoms, comorbidity, and the lack of accurate diagnostic tools leads to a poor quality of life for these patients. 7 Thus, there is an urgent need for improved understanding of the pathophysiology of the condition, and the development of robust and accessible diagnostic tools to detect adenomyosis. There is currently very limited research into potential biomarkers for adenomyosis, particularly those detectable through non-invasive methods. The majority of previous studies focused on the serum protein biomarkers, such as cancer antigen 125 (CA125) or human epididymis protein 4 (HE4), 8 , 9 , 10 , 11 but these biomarkers have not been adapted for clinical use for adenomyosis. Serum levels of CA125 have been shown to be increased in adenomyosis patients 12 , 13 , 14 ; however, this biomarker is typically elevated in women with other conditions, such as endometriosis, 15 , 16 or endometrial, 17 , 18 and ovarian cancers, 19 which limits its diagnostic value. 11 Similarly, there is limited research investigating the metabolic biomarkers/signatures of patients with adenomyosis. Only two pilot studies have been published on metabolic alterations in adenomyosis patients and neither used non-invasive cervicovaginal lavage (CVL) sampling. 20 , 21 One small study employed nuclear magnetic resonance spectroscopy and revealed differences in serum metabolic profiles between adenomyosis cases (n = 32) and controls (n = 45), as well as differences between disease phenotypes (focal vs. diffuse) despite the small sample size. 20 Another pilot study from 2022 investigated metabolites in the myometrial tissues collected from patients undergoing hysterectomy for adenomyosis (n = 17) or other conditions (leiomyomas/fibroids or cervical neoplasia, n = 25). 21 By utilizing gas/liquid chromatography-mass spectroscopy, they found significant upregulation of metabolites implicated in inflammation, oxidative stress, energy metabolism (acylcarnitines), cell proliferation (pyrimidines and purines) and apoptosis (glycerophospholipids) in the myometrium of adenomyosis patients. 21 Although these studies provide some insights on metabolism in the context of adenomyotic lesions, there is still an unmet and urgent need for further research in large and ethnically diverse cohorts to evaluate novel, non-invasive sampling for biomarker discovery and identification of biological mechanisms underlying this enigmatic condition. In this study, we utilize multiplex immunoassay and global metabolomics platforms coupled with non-invasive CVL sampling, to identify novel protein and metabolic biomarkers for adenomyosis detection. In addition, we aimed to better understand the pathophysiological processes behind adenomyosis. Through this acquired knowledge, there is potential for improved detection and treatment of adenomyosis.

REAGENT or RESOURCE SOURCE IDENTIFIER Biological samples Human cervicovaginal lavage samples Banner University Medical Center Phoenix Dignity Health Chandler Medical Center N/A Chemicals, peptides and recombinant proteins 0.9% saline solution Teknova, Hollister, CA Cat#S0699 Critical commercial assays Milliplex MAP Magnetic Bead Immunoassay: Human Cytokine Chemokine Panel 1 Millipore, Billerica, MA Cat# HCYTOMAG-60K Milliplex MAP Magnetic Bead Immunoassay: Human Circulating Cancer Biomarker Panel 1 Millipore, Billerica, MA Cat# HCCBP1MAG8-58K Milliplex MAP Magnetic Bead Immunoassay: Human Immuno-Oncology Checkpoint Protein Panel 1 Millipore, Billerica, MA Cat# HCKP1-11K Human IL-36γ ELISA kit RayBiotech, Norcross, GA Cat#ELH-IL1F9-1 Softwares and algorithms Bio-Plex Manager 5.0 software Bio-Rad, Hercules, CA bio-rad.com Other Bio-Plex 200 instrument Bio-Rad, Hercules, CA Cat#171000201 Metabolon, Inc (global metabolomics platform) Metabolon, Inc, Durhamm, NC metabolon.com MetaboAnalyst 5.0 MetaboAnalyst metaboanalyst.ca Further information and requests should be directed to and will be fulfilled by the lead contact, Dr Herbst-Kralovetz ( [email protected] ). This study did not generate any new unique reagents. This study was approved by the Institutional Review Board at the University of Arizona (IRB no. 1708726047). All participants provided written informed consent and the study was performed in accordance with the Declaration of Helsinki and federal guidelines. One hundred and eight participants undergoing hysterectomy for benign conditions were recruited at two clinical sites in the Phoenix (AZ, USA) metropolitan area: Banner University Medical Center – Phoenix and Dignity Health Chandler Medical Center. Histopathology of biopsy samples collected from the surgery were used to stratify into two groups: adenomyosis (n = 46) and no adenomyosis (n = 62). Women were excluded from the study if they were currently menstruating; currently lactating; currently (or within the past 3 months) on antifungals, antivirals, or topical steroids; currently (or within the past 3 months) had vaginal, vulvar, urinary tract, or sexually transmitted infections; used any douching products, vaginal medications or suppositories, feminine deodorant sprays, wipes, or lubricants within the past 48 h; used any depilatory treatments in the genital area in the past 72 h; had any skin condition in the genital area; had sexual intercourse in the past 48 h; were bathing or swimming in the past 4 h; were smoking or consuming nicotine-containing products in the past 2 h; had hepatitis; were HIV-positive. Inclusion criteria included any women, 18 years of age and older, of any race and ethnicity who were undergoing hysterectomy for benign conditions. Demographic, socioeconomic, and medical history data were collected from surveys and/or medical records. CVL samples were collected by a surgeon in the operating room during the standard-of-care hysterectomy procedure. Samples were obtained after induction of anesthesia and prior to sterile preparation. CVLs were collected using a non-lubricated speculum and 10 mL of sterile 0.9% saline solution (Teknova, Hollister, CA) Following collection, samples were immediately placed on ice and frozen at −80°C within an hour. Prior to analyses, the samples were thawed on ice; centrifuged (700 × g for 10 min at 4°C); aliquoted, to prevent multiple freeze-thaw cycles; and stored at −80°C. Protein concentrations in CVL samples were measured using the Milliplex MAP Magnetic Bead Immunoassays: Human Cytokine Chemokine Panel 1, Human Circulating Cancer Biomarker Panel 1, and Human Immuno-Oncology Checkpoint Protein Panel 1 (Millipore, Billerica, MA) according to the manufacturer’s protocol. Levels of 71 proteins (AFP, BTLA, CA15-3, CA19-9, CA125, CD27, CD28, CD40, CD80, CD86, CEA, CYFRA21-1, EGF, eotaxin/CCL11, Flt-3L, FGF-2, fractalkine/CXC3CL1, G-CSF, GITRL, GROα/CXCL1, GM-CSF, HE4, HGF, HVEM, ICOS, IFNα2, IFNγ, IL-1α, IL-1β, IL-2, IL-4, IL-5, IL-6, IL-7, IL-8/CXCL8, IL-9, IL-10, IL-12 p40, IL-12 p70, IL-13, IL-15, iIL-17A, IP-10/CXCL10, LAG3, leptin, MCP-1/CCl2, MCP-3/CCL7, MDC/CCL22, MIF, MIP-1α/CCL3, MIP-1β/CCL4, OPN, PD-1, PD-L1, PD-L2, PDGF-AA, PDGF-AB/BB, prolactin, PSA total, RANTES/CCL5, SCF, sCD40L, sFas, sFasL, TGF-α, TIM-3, TLR2, TNFα, TNFβ, TRAIL, VEGF) were quantified using a Bio-Plex 200 instrument and Bio-Plex Manager 5.0 software (Bio-Rad, Hercules, CA). Levels of IL-36γ (IL-1F9) were measured in the samples by enzyme-linked immunosorbent assay using Human IL-36γ ELISA kit (RayBiotech, Norcross, GA) in accordance with the manufacturer’s instructions. All samples were analyzed in duplicate. Concentrations were determined using a five-parameter logistic regression curve fit. If the concentrations measured were below the detection limit, the value was substituted with 0.5 of the minimum detectable concentration provided in the manufacturer’s instructions. Data was normalised using the log 10 transformation. Soluble metabolites in the CVL samples were determined using a global metabolomics platform at Metabolon, Inc. (Durham, NC). Samples were prepared using with Micro-Lab STAR® system (Hamilton, Reno, NV). Recovery standards were added for quality control purposes. To recover metabolites and remove protein, the samples were precipitated with methanol under shaking for 2 min (Glen Mills GenoGrinder 2000) followed by centrifugation. Samples were then placed on a TurboVap® (Zymark) to remove the organic solvent. Samples were split into five aliquots, one for each of the analyses and one spare. A pooled matrix was generated by mixing a small volume of each sample to serve as a technical replicate. Extracted water samples were utilised as process blanks and a mix of quality control standards were selected and added to each sample to monitor instrument performance and aid chromatographic alignment. All methods utilised a Waters ACQUITY ultra-performance liquid chromatography (UPLC) and a Thermo Scientific Q-exactive high resolution/accurate mass spectrometer interfaced with a heated electrospray ionisation (HESI-II) source and Orbitrap mass analyser operated at 35000 mass resolution. The sample extract was dried and then resuspended in solvents compatible with each of the four methods listed below. The resuspension solvents all contained a series of standards at fixed concentrations for chromatographic consistency. One aliquot was analyzed using acidic positive ion conditions, optimised for hydrophilic compounds. The extract was gradient eluted from a C18 column (Waters UPLC BEH C18-2.1 × 100 mm, 1.7 μm) using water and methanol, containing 0.05% perfluoropentanoic acid, and 0.1% formic acid. Another aliquot was analyzed using acidic positive ion conditions, optimised for hydrophobic compounds. The extract was gradient eluted from the same C18 column using methanol, acetonitrile, water, 0.05% perfluoropentanoic acid, and 0.01% formic acid and was operated at a higher organic content. The third aliquot was analyzed using basic negative ion optimised conditions using a separate C18 column. The basic extracts were gradient eluted using methanol and water, and 6.5 mM ammonium bicarbonate at pH 8. The fourth aliquot was analyzed via negative ionisation following elution from a HILIC column (Water UPLC BEH Amide 2.1 × 150 mm, 1.7 μm) using a gradient consisting of water and acetronitrile with 10 mM ammonium formate, pH 10.8. Peak analysis and quality control processing were performed by the Metabolon Laboratory Information System for compound identification. Metabolon’s library is able to match compounds to more than 3300 purified standards. In addition, recurrent unknown entities were also reported. Peaks were quantified using area-under-the-curve for relative intensity. Data was normalised by registering the medians of each compound to equal one and normalising each data point proportionately. Data was transformed using the log 10 transformation and autoscaled (mean-centred and divided by the standard deviation of each variable). Percentage fill value data was determined by calculating the percentage of samples that a particular metabolite was detected in for each group (adenomyosis and no adenomyosis). Partial least squares-discriminant analysis (PLS-DA) was performed using MetaboAnalyst 5.0 59 to visualise the separation of the two patient groups: adenomyosis (n = 46) and no adenomyosis (n = 62). PLS-DA is a supervised regression method that aims to plot the greatest separation between groups by finding the maximum covariance between the data and the assigned group. A two-sample t -test was performed with a significance value threshold of 0.05 pvalue. Fold change analysis compared the absolute value of change between the means of each metabolite/protein between the two groups. The fold-change analysis utilises the dataprior todata transformation and scaling. Data from fold change and t -test analysis were combined to produce volcano plots that depict the significantly up-/down regulated metabolites and immune proteins in the women with adenomyosis compared to the women without adenomyosis. The comparison of direction was adenomyosis vs. no adenomyosis and the fold change threshold was 2.0. The analysis was performed using MetaboAnalyst 5.0. 59 Partially supervised hierarchical clustering analysis was performed on metabolite and immunoprotein datasets, individually, using MetaboAnalyst 5.0 59 to produce heatmaps. Metabolites/immune proteins were autoscaled and then Pearson distance measure and Ward linkage was applied to the metabolites/immune proteins. Samples were analyzed both with and without clustering; for those analyzed without clustering, the order of samples remained in the supervised order inputted, which was categorised based on adenomyosis status. Enrichment analysis was completed in MetaboAnalyst 5.0 59 by comparing metabolite data to the Small Molecule Pathway Database metabolite set based on normal human metabolic pathways. Enrichment ratio and significance of enrichment of metabolic pathways were calculated based on the number of metabolites detected within a specific pathway relative to number of known metabolites in that pathway. The algorithm also considered the relative intensity of the metabolites in adenomyosis compared to no adenomyosis. The data was pre-processed by normalisation, log 10 transformation and auto-scaled therefore it was appropriate to measure the statistical differences between the mean intensities of metabolites among the groups using a t -test. p-values were corrected using the false discovery rate (FDR) method and q -values have been reported. p-values<0.05 were considered statistically significant. Statistical analyses were performed using MetaboAnalyst 5.0. 59 Differences in demographic, socioeconomic and other patient-related variables between disease groups (adenomyosis vs. no adenomyosis) were tested using Wilcoxon rank-sum test for continuous variables and Fisher exact test for categorical variables.