Lipidome atlas of human myometrium reveals distinctive lipid signatures associated with adenomyosis: Combination of high-coverage lipidomics and mass spectrometry imaging

Shuo Liang: Investigation. Jialin Liu: Investigation, Writing – original draft. Maokun Liao: Writing – review & editing. Dandan Liang: Investigation. Yiyi Gong: Methodology. Bo Zhang: Methodology. Nan Zhao: Data curation. Wei Song: Writing – original draft, Funding acquisition, Conceptualization. Honghui Shi: Project administration, Funding acquisition.

Following the UPLC-MS-based lipidomics approach in Fig. 1 A, a repertoire of 317 lipid species spanning 19 major lipid classes was successfully quantified from the 103 myometrial samples ( Table S2 ). The 19 lipid classes belong to five superclasses, including glycerophospholipids ( n = 155), glycerolipids ( n = 83), sphingolipids ( n = 37), fatty acyls ( n = 31), and sterol lipids ( n = 11) ( Fig. 1 B). In terms of content, total glycerophospholipids, mainly comprising PEs and phosphatidylserines (PSs), accounted for 42.5% of the total lipids in the control group and 43.8% in the adenomyosis group. Sphingomyelins (SMs), the predominant sphingolipid, accounted for no less than 32.0%, whereas free fatty acids (FFAs) accounted for no less than 15.4% in both groups ( Fig. 1 C). It is worth noting that among the 19 classes of lipids, 10 showed significant differences in content between the two groups. The contents of bis(monoacylglycero)phosphates (BMPs), lysophosphatidylcholines (LPCs), lysophosphatidylethanolamines (LPEs), lysophosphatidylglycerols (LPGs), ceramides (Cers), hexosylceramides (HexCers), FFAs, and diacylglycerols (DAGs) were appreciably increased, while the contents of cholesterol esters (CEs) and triacylglycerols (TAGs) were decreased in the myometrium of patients with adenomyosis compared to those in the controls ( Fig. 1 D). Multivariate statistical analyses were then performed to further investigate the differential lipid metabolism related to adenomyosis. Fig. 1 Lipidome profiling of human myometrium. (A) Schematic view of the lipidomics analysis. (B) Numbers of lipid species from 19 lipid classes out of five superclasses determined in human myometrial tissues. (C) Multiple pie charts showing the content distributions of different lipid classes in the control and adenomyosis groups. Lipid classes with content percentages no less than 2.0% were labeled. (D) The content level of each lipid class in the control and adenomyosis groups. ∗ P < 0.05, ∗∗ P < 0.01, and ∗∗∗ P < 0.001. ns: no significant difference. LC-MS: liquid chromatography coupled-mass spectrometry; PE: phosphatidylethanolamine; TAG: triacylglycerol; FFA: free fatty acid (FA); DAG: diacylglycerol; PC: phosphatidylcholine; LPE: lysophosphatidylethanolamine; SM: sphingomyelin; CE: cholesterol ester; Cer: ceramide; PS: phosphatidylserine; HexCer: hexosylceramide; LPC: lysophosphatidylcholine; BMP: bismonoacylglycerophosphate; DCer: dihydroceramide; LPG: lysophosphatidylglycerol; LPS: lysophosphatidylserine; MGDG: monogalactosyldiacylglycerol; PG: phosphatidylglycerol; PI: phosphatidylinositol. Fig. 1 Lipidome profiling of human myometrium. (A) Schematic view of the lipidomics analysis. (B) Numbers of lipid species from 19 lipid classes out of five superclasses determined in human myometrial tissues. (C) Multiple pie charts showing the content distributions of different lipid classes in the control and adenomyosis groups. Lipid classes with content percentages no less than 2.0% were labeled. (D) The content level of each lipid class in the control and adenomyosis groups. ∗ P < 0.05, ∗∗ P < 0.01, and ∗∗∗ P < 0.001. ns: no significant difference. LC-MS: liquid chromatography coupled-mass spectrometry; PE: phosphatidylethanolamine; TAG: triacylglycerol; FFA: free fatty acid (FA); DAG: diacylglycerol; PC: phosphatidylcholine; LPE: lysophosphatidylethanolamine; SM: sphingomyelin; CE: cholesterol ester; Cer: ceramide; PS: phosphatidylserine; HexCer: hexosylceramide; LPC: lysophosphatidylcholine; BMP: bismonoacylglycerophosphate; DCer: dihydroceramide; LPG: lysophosphatidylglycerol; LPS: lysophosphatidylserine; MGDG: monogalactosyldiacylglycerol; PG: phosphatidylglycerol; PI: phosphatidylinositol. Principal component analysis (PCA) was first applied to obtain an unsupervised and global view of the lipidome for all myometrial samples. The close clustering of the QC samples indicated the high repeatability and stability of the lipidomics approach ( Fig. 2 A). Samples in the adenomyosis and control groups were readily differentiated, indicating that apparent lipidomic changes appeared in the myometrium upon adenomyosis. Then, supervised OPLS-DA was conducted to identify significant lipids from the lipidome data, and more distinctive differences were observed between the two groups ( Fig. 2 B). The Q 2 from the 200 permutation tests was −0.71, suggesting no overfitting of the OPLS-DA model ( Fig. 2 C). According to the results of Student's t -test, 107 out of the 317 lipid species showed significant differences ( P < 0.05) in contents between the two groups, which involved 13 classes according to their chemical structures ( Fig. 2 D). Using the thresholds of both the variable importance in the projection (VIP) value > 1 from OPLS-DA and P < 0.05, a total of 83 significantly altered lipid markers yielded in patients with adenomyosis (54 increased and 29 decreased) when compared to those in the controls ( Fig. 2 E), the contents of which were further visualized using a clustered heatmap ( Fig. 3 ). The significantly decreased lipids included TAG and CE, while the increased lipids mainly included PE, LPE, LPC, Cers, hexosylceramides, and FFA. Fig. 2 Multivariate statistical analysis of the lipidome data. (A) Score plot of principal component analysis (PCA) to visualize the distribution of all samples. Each dot represents a sample. (B) Score plot of orthogonal partial least squares-discriminant analysis (OPLS-DA). Each dot represents a sample. (C) Results of the permutation test for OPLS-DA. (D) Bubble plot showing the relative differences in the contents of different kinds of lipids between the adenomyosis group and the control group. Each dot represents a lipid species. Relative difference = [(Average in adenomyosis – average in control)/average in control] × 100%. (E) Volcano plot showing differentially expressed lipid metabolites in the adenomyosis group. Each dot represents a lipid species. QC: quality control; PC: phosphatidylcholine; TAG: triacylglycerol; SM: sphingomyelin; PS: phosphatidylserine; PI: phosphatidylinositol; PG: phosphatidylglycerol; PE: phosphatidylethanolamine; PC: phosphatidylcholine; MGDG: monogalactosyldiacylglycerol; LPS: lysophosphatidylserine; LPG: lysophosphatidylglycerol; LPE: lysophosphatidylethanolamine; LPC: lysophosphatidylcholine; HexCer: hexosylceramide; FFA: free fatty acid (FA); DCer: dihydroceramide; DAG: diacylglycerol; Cer: ceramide; CE: cholesterol ester; BMP: bismonoacylglycerophosphate; VIP: variable importance in the projection. Fig. 2 Fig. 3 Heatmap of the significantly altered lipid species between the adenomyosis group and control group. TAG: triacylglycerol; FA: fatty acid; PE: phosphatidylethanolamine; CE: cholesterol ester; LPE: lysophosphatidylethanolamine; LPC: lysophosphatidylcholine; HexCer: hexosylceramide; Cer: ceramide; BMP: bismonoacylglycerophosphate; PG: phosphatidylglycerol; LPG: lysophosphatidylglycerol; FFA: free FA. Fig. 3 Multivariate statistical analysis of the lipidome data. (A) Score plot of principal component analysis (PCA) to visualize the distribution of all samples. Each dot represents a sample. (B) Score plot of orthogonal partial least squares-discriminant analysis (OPLS-DA). Each dot represents a sample. (C) Results of the permutation test for OPLS-DA. (D) Bubble plot showing the relative differences in the contents of different kinds of lipids between the adenomyosis group and the control group. Each dot represents a lipid species. Relative difference = [(Average in adenomyosis – average in control)/average in control] × 100%. (E) Volcano plot showing differentially expressed lipid metabolites in the adenomyosis group. Each dot represents a lipid species. QC: quality control; PC: phosphatidylcholine; TAG: triacylglycerol; SM: sphingomyelin; PS: phosphatidylserine; PI: phosphatidylinositol; PG: phosphatidylglycerol; PE: phosphatidylethanolamine; PC: phosphatidylcholine; MGDG: monogalactosyldiacylglycerol; LPS: lysophosphatidylserine; LPG: lysophosphatidylglycerol; LPE: lysophosphatidylethanolamine; LPC: lysophosphatidylcholine; HexCer: hexosylceramide; FFA: free fatty acid (FA); DCer: dihydroceramide; DAG: diacylglycerol; Cer: ceramide; CE: cholesterol ester; BMP: bismonoacylglycerophosphate; VIP: variable importance in the projection. Heatmap of the significantly altered lipid species between the adenomyosis group and control group. TAG: triacylglycerol; FA: fatty acid; PE: phosphatidylethanolamine; CE: cholesterol ester; LPE: lysophosphatidylethanolamine; LPC: lysophosphatidylcholine; HexCer: hexosylceramide; Cer: ceramide; BMP: bismonoacylglycerophosphate; PG: phosphatidylglycerol; LPG: lysophosphatidylglycerol; FFA: free FA. We then investigated the metabolic pathways of these lipid classes and the proportions of altered lipid species in each class ( Fig. 4 ). Phosphorylation is of vital importance for the biological activities of lipids. One important finding is that several kinds of glycerophospholipids and their derivatives, such as PE, PG, and PC, were markedly more abundant in the myometrium of patients with adenomyosis than in that of controls. Corresponding to the elevated levels of glycerophospholipids, decreased levels of the upstream lipids TAGs were also observed in the adenomyosis group. Cers and their derivatives are another kind of lipid implicated in diverse biological functions, the contents of which were also significantly increased in the adenomyosis group. Most FFAs, especially saturated FFAs, have lipotoxic and pro-inflammatory effects, resulting in oxidative stress and apoptosis in various cell types. The present lipidomics study demonstrated that nearly one-third of the total FFA molecules, especially saturated and monounsaturated FFAs, were obviously overexpressed in patients with adenomyosis ( Figs. 4 and S1 ). Fig. 4 Schematic diagram showing the integrated changes in lipid metabolic pathways in adenomyosis patients. Red text, increased lipids and blue text, decreased lipids. Red numbers, increased lipid species; blue numbers, decreased lipid species; and black numbers, all determined lipid species from the human myometrium. FFA: free fatty acid (FA); LysoPA: lysophosphatidic acid; PA: phosphatidic acid; CDP: cytidine diphosphate; DAG: diacylglycerol; PI: phosphatidylinositol; LysoPI: lysophosphatidylinositol; PG: phosphatidylglycerol; LysoPG: lysophosphatidylglycerol; BMP: bismonoacylglycerophosphate; PS: phosphatidylserine; LysoPS: lysophosphatidylserine; PE: phosphatidylethanolamine; LysoPE: lysophosphatidylethanolamine; TAG: triacylglycerol; PC: phosphatidylcholine; LysoPC: lysophosphatidylcholine; CE: cholesterol ester; HexCer: hexosylceramide. Fig. 4 Schematic diagram showing the integrated changes in lipid metabolic pathways in adenomyosis patients. Red text, increased lipids and blue text, decreased lipids. Red numbers, increased lipid species; blue numbers, decreased lipid species; and black numbers, all determined lipid species from the human myometrium. FFA: free fatty acid (FA); LysoPA: lysophosphatidic acid; PA: phosphatidic acid; CDP: cytidine diphosphate; DAG: diacylglycerol; PI: phosphatidylinositol; LysoPI: lysophosphatidylinositol; PG: phosphatidylglycerol; LysoPG: lysophosphatidylglycerol; BMP: bismonoacylglycerophosphate; PS: phosphatidylserine; LysoPS: lysophosphatidylserine; PE: phosphatidylethanolamine; LysoPE: lysophosphatidylethanolamine; TAG: triacylglycerol; PC: phosphatidylcholine; LysoPC: lysophosphatidylcholine; CE: cholesterol ester; HexCer: hexosylceramide. To determine potential biomarkers for discriminating patients with adenomyosis from control subjects, univariate ROC analyses were first conducted based on the 83 significantly altered lipid species. The five lipid species with the highest area under the curve (AUC) were CE (18:2), HexCer (18:0/16:0), LPC (18:2), TAG (50:3)_fatty acid (FA) 18:2, and TAG (52:2)_FA 18:1 ( Figs. 5 A and B and Table S3 ; the AUCs ranged from 0.74 to 0.77). Multivariate ROC analyses were then conducted using different combinations of these five lipids ( Fig. S2 ). Using a panel consisting of these five top lipids identified from univariate ROC analyses showed the best discrimination ability, with an AUC value of 0.906 ( Fig. 5 C). Overall, ROC analyses showed that the panel containing five lipid species from different subclasses could accurately distinguish the myometrial tissues of patients with adenomyosis from those of control subjects. Fig. 5 Lipid markers that showed potential in distinguishing patients with adenomyosis from control subjects. (A) The contents of five individual lipid species in myometrium of women from two groups. (B) Receiver operating characteristic (ROC) curves for five individual lipid species. (C) ROC curve for a panel combining the five lipid species. The 95% confidence intervals are shown in brackets. ∗∗ P < 0.01 and ∗∗∗ P < 0.001. HexCer: hexosylceramide; LPC: lysophosphatidylcholine; CE: cholesterol ester; TAG: triacylglycerol; FA: fatty acid; AUC: area under the curve. Fig. 5 Lipid markers that showed potential in distinguishing patients with adenomyosis from control subjects. (A) The contents of five individual lipid species in myometrium of women from two groups. (B) Receiver operating characteristic (ROC) curves for five individual lipid species. (C) ROC curve for a panel combining the five lipid species. The 95% confidence intervals are shown in brackets. ∗∗ P < 0.01 and ∗∗∗ P < 0.001. HexCer: hexosylceramide; LPC: lysophosphatidylcholine; CE: cholesterol ester; TAG: triacylglycerol; FA: fatty acid; AUC: area under the curve. Tissue homogenization before LC-MS analysis resulted in the loss of spatial distribution information of molecules. Although the above lipid makers profoundly altered in normal myometrium between the two groups, their distribution among the adenomyosis lesion and the adjacent surrounding myometrium remains unknown. DESI-MSI was then utilized to explore whether these lipid markers exhibited heterogeneous distributions within the uterus. Histological examination using H&E staining revealed the presence of adenomyotic foci, characterized by ectopic endometrial glands and stroma, scattered in the myometrium in patients diagnosed with adenomyosis ( Fig. 6 ). DESI-MSI was able to visualize all the five lipid species in continuous-slices. Interestingly, HexCer (18:0/16:0) and LPC (18:2) were mainly enriched in ectopic endometrial partitions, while CE (18:2), TAG (50:3)_FA 18:2, and TAG (52:2)_FA 18:1 showed a converse trend. Next, we selected regions of interest (ROIs) within the ectopic endometrium and the adjacent myometrium, respectively, and then compared their intensities between different regions. The mean intensity of ROIs from the ectopic endometrium and the adjacent myometrium were subjected to statistical comparison. The fold change (FC) for each lipid marker was determined by the ratio of its mean intensity in ectopic endometrium to that in the myometrium. As a result, the FCs of HexCer (18:0/16:0) and LPC (18:2) were greater than 1.5, indicating a significant enrichment of them in ectopic endometrial partitions ( Table 2 and Fig. S3 ). Conversely, the FCs of the other three lipids, CE (18:2), TAG (50:3)_FA 18:2, and TAG (52:2)_FA 18:1, were below 1.0, suggesting an enrichment of these lipids in myometrium. Combined, the converse trends of different lipids further underscored the metabolic heterogeneity between lesion location and myometrium in human uterus and further elucidated the significance of the lipid markers for adenomyosis. Fig. 6 Hematoxylin and eosin (H&E) staining and desorption electrospray ionization ionization-mass spectrometry imaging (DESI-MSI) of the five lipid markers in myometrium of adenomyosis patient. Yellow arrows indicate the ectopic endometrial tissue and grey arrows indicate the myometrium. HexCer: hexosylceramide; LPC: lysophosphatidylcholine; CE: cholesterol ester; TAG: triacylglycerol; FA: fatty acid. Fig. 6 Table 2 Detailed information of the five lipid biomarkers analyzed by desorption electrospray ionization-mass spectrometry (DESI-MS). Table 2 Lipid HRMS ( m / z ) Adduct ion Intensity in ectopic endometrium Intensity in adjacent myometrium FC HexCer (18:0/16:0) 740.5437 [M+K] + 70,087 ± 14,351 43,569 ± 6830 1.61 LPC (18:2) 542.3217 [M+Na] + 17,159 ± 2525 7545 ± 505 2.27 TAG (50:3)_FA 18:2 829.7280 [M+H] + 3951 ± 380 4506 ± 381 0.88 TAG (52:2)_FA 18:1 897.7308 [M+K] + 4777 ± 940 7923 ± 3538 0.60 CE (18:2) 687.5477 [M+K] + 16,193 ± 3478 19,558 ± 7456 0.83 Fold change (FC) = mean intensity in ectopic endometrium/mean intensity in myometrium. Intensity is expressed as mean ± standard deviation (SD) ( n = 3). HRMS: high resolution mass spectrometry; HexCer: hexosylceramide; LPC: lysophosphat-idylcholine; TAG: triacylglycerol; FA: fatty acid; CE: cholesterol ester. Hematoxylin and eosin (H&E) staining and desorption electrospray ionization ionization-mass spectrometry imaging (DESI-MSI) of the five lipid markers in myometrium of adenomyosis patient. Yellow arrows indicate the ectopic endometrial tissue and grey arrows indicate the myometrium. HexCer: hexosylceramide; LPC: lysophosphatidylcholine; CE: cholesterol ester; TAG: triacylglycerol; FA: fatty acid. Detailed information of the five lipid biomarkers analyzed by desorption electrospray ionization-mass spectrometry (DESI-MS). Fold change (FC) = mean intensity in ectopic endometrium/mean intensity in myometrium. Intensity is expressed as mean ± standard deviation (SD) ( n = 3). HRMS: high resolution mass spectrometry; HexCer: hexosylceramide; LPC: lysophosphat-idylcholine; TAG: triacylglycerol; FA: fatty acid; CE: cholesterol ester.

The cross-sectional study was approved by the Ethics Committee of Peking Union Medical College Hospital (PUMCH; Beijing, China) (Approval No.: ZS-2025) in June 2019. A total of 103 female patients who underwent hysterectomy at the general gynecological center of PUMCH from July 2019 to December 2022 were enrolled and divided into two groups. The study group included 38 patients with histologically confirmed adenomyosis, whereas the control group comprised 65 patients without adenomyosis who had undergone hysterectomy for cervical intraepithelial neoplasia III ( n = 39) or uterine leiomyomas ( n = 26). The subject recruitment criteria were primarily aligned with those outlined in our prior study [ 22 ]. Briefly, the diagnosis of adenomyosis was initially determined through magnetic resonance imaging (MRI), followed by confirmation through histologic analysis post-hysterectomy [ 23 , 24 ]. Adenomyosis was defined as the presence of endometrial glands and stroma extending more than 2.5 mm below the endometrial-myometrial junction [ 28 ]. For participants in control group, adenomyosis was excluded through pelvic MRI and surgical examination. All participants reported spontaneous menses in the absence of gonadal steroid treatment. No participant received oral contraception or gonadotrophin-releasing hormone (GnRH) agonists and in one months prior to surgery. Patients with uterine endometriosis or malignant tumors were excluded from this study. Clinical records were reviewed for each participant. Table 1 demonstrated that there was no significant difference in age, body mass index (BMI), menstrual phase, gravidity, or parity between the two cohorts ( P > 0.05). Uteri with and without combined leiomyomata were identified based on MRI findings and recorded in Table 1 . Informed consents were obtained from all participants. Table 1 Clinical characteristics of subjects recruited in this study. Table 1 Characteristics Adenomyosis (−) ( n = 65) Adenomyosis (+) ( n = 38) P value Age (year) 45.44 ± 6.16 45.47 ± 4.65 0.63 BMI (kg/m 2 ) 23.56 ± 3.02 23.89 ± 3.25 0.87 Gravidity 2.12 ± 1.26 2.50 ± 1.45 0.76 Parity 1.22 ± 0.62 0.97 ± 0.72 0.57 Associated leiomyomas 29 (44.62%) 15 (39.47%) / Menstrual phase  Follicular phase 30 (46.15%) 17 (44.74%) /  Luteal phase 28 (43.08%) 16 (42.11%) /  Unknown 7 (10.77%) 5 (13.16%) / Adenomyosis (−), women without adenomyosis; adenomyosis (+), adenomyosis-affected women. Data are expressed as means ± standard deviation (SD) or number (%). /: no data. BMI: body mass index. Clinical characteristics of subjects recruited in this study. Adenomyosis (−), women without adenomyosis; adenomyosis (+), adenomyosis-affected women. Data are expressed as means ± standard deviation (SD) or number (%). /: no data. BMI: body mass index. As described in our previous report, sample collection was performed by senior gynecologist at PUMCH following hysterectomy [ 22 ]. The myometrial tissue was sliced while visual inspection and gross pathological examination were conducted. Tissue fragments containing only normal myometrium, as well as typical tissue fragments containing both the adenomyosis lesion and the adjacent myometrial tissue, were collected and frozen at −80 °C until use, respectively. The former was used for lipidomic profiling, and the latter was used for DESI-MSI analysis. Methanol, acetonitrile, ammonium acetate, formic acid, and isopropanol were of LC-MS grade (Merck, KGaA, Darmstadt, Germany). Dichloromethane (DCM) and methyl tert -butyl ether (MTBE) were of high performance liquid chromatography (HPLC) grade (CNW Technologies, Düsseldorf, Germany). De-ionized water was provided by Milli-Q water purification system (Millipore, Billerica, MA, USA). High-purity nitrogen (99.9%) and helium (99.99%) were from the gas supply center of PUMCH. Internal standards (ISs) for lipidomic analysis were from Sigma-Aldrich (St. Louis, MO, USA) or Avanti Research (Alabaster, AL, USA). The detailed information of each lipid standard was listed in Table S1 . Leucine enkephalin from Waters (Milford, MA, USA) was used to lock mass in DESI-MSI analysis. For lipidomics analysis, the total crude lipids were extracted as follows. After adding 400 μL of water, 20 mg of the myometrial tissue sample was vortexed for 60 s, homogenized at 45 Hz for 4 min, and sonicated for 5 min in an ice-water bath. The homogenization and sonication cycles were repeated three times. Then, 150 μL of homogenate was mixed with 50 μL of water, and 480 μL of extract solution (MTBE:methanol = 5:1, v / v ) containing an IS was added. After vortexing for 60 s, the sample was sonicated for 10 min in an ice-water bath. Then, the sample was centrifuged at 3,000 rpm for 15 min at 4 °C. After that, 250 μL of the supernatant was transferred to a fresh tube. Then, 250 μL of MTBE was added to the rest of the sample, followed by vortexing, sonication, and centrifugation, and another 250 μL of the supernatant was removed. This step was repeated once. The supernatant was combined and dried in a vacuum concentrator at 37 °C. Then, the dried sample was reconstituted in 120 μL of resuspension buffer (DCM:methanol:H 2 O = 60:30:4.5, v / v / v ), and the sample was vortexed for 30 s and sonicated for 10 min in an ice-water bath. The constitution was then centrifuged at 12,000 rpm for 15 min at 4 °C, and 35 μL of the supernatant was transferred to a fresh glass vial for UPLC-MS analysis. Quality control (QC) samples were prepared by mixing equal amounts of supernatant. For DESI-MSI analysis and hematoxylin and eosin (H&E) staining, the frozen tissues that contained the adenomyotic foci were cut into 10-μm thick slices utilizing NX70 cryostat microtome (Thermo Fisher Scientific Inc., Waltham, MA, USA) at −20 °C. Afterwards, the slices were thaw-mounted on Superfrost Plus microscope slides (Thermo Fisher Scientific Inc.) and stored at −80 °C until DESI-MSI analysis. The slices were dried for 2 h before the MSI experiments were conducted. The quantitation of all targeted lipids in human myometrial samples were conducted by UPLC-QqQ-MS. The UPLC separation was carried out using an ACQUITY Premier UPLC System (Waters) coupled with an ACQUITY UPLC HSS T 3 column (1.8 μm, 2.1 × 100 mm, Waters). Mobile phase A consisted of 60% acetonitrile and 40% water with ammonium formate at a final concentration of 10 mmol/L. Mobile phase B consisted of 90% isopropanol and 10% acetonitrile with ammonium formate at a final concentration of 10 mmol/L. Gradient elution was applied within 18 min at a flow rate of 300 μL/min with a column temperature of 45 °C. The injection volume was 2 μL, and the autosampler temperature was 10 °C. A triple Quad™ 6500+ mass spectrometer (AB SCIEX, Framingham, MA, USA) was used for lipid quantitation in multiple reaction monitoring (MRM) mode. The typical ion source parameters were as follows: ion spray voltage, +5500/−4500 V; curtain gas, 40 psi; temperature, 350 °C; ion source gas, 1:50 psi; ion source gas, 2:50 psi; and declustering potential (DP), ±80 V. The MRM transitions and other conditions were referred to previous reports [ 29 , 30 ]. The content of individual lipids corresponding to the IS was calculated using the peak area and actual concentration of the identical lipid class IS. The detailed information of these reference standards used in this study have been listed in Table S1 . The integration of a DESI XS system and a Xevo G2-XS QTOF MS (Waters) was used to perform MSI experiments. Methanol:water (95:5, v / v ) containing 200 pg/μL leucine enkephalin and 0.1% formic acid was selected as the spray solvent in positive ion mode. The flow rate of spray solvent was set at 3 μL/min and the pressure of nebulizing gas (N 2 ) was 0.45 MPa. DESI-MSI analysis was carried out with a scan range of 50–1000 Da. Other mass spectrometer parameters were set as follows: capillary voltage, 0.66 kV; sampling cone, 40 V; source temperature, 150 °C; and heated transfer line temperature, 250 °C. The scan rate was 300 μm/s and spatial resolution was set at 100 μm. All the data are expressed as the means ± standard deviation (SD). The statistical significance of differences between two groups was assessed using unpaired two-tailed Student's t -tests, and differences were considered to be statistically significant at P < 0.05. Orthogonal partial least squares-discriminant analysis (OPLS-DA) was conducted using Simca software (version 16.0.2, Umeå, Sweden). Receiver operating characteristic (ROC) analyses were conducted on MetaboAnalyst 5.0 ( https://www.metaboanalyst.ca ). Other statistical analyses were performed using GraphPad Prism software (V9.3.1, Bethesda, MD, USA) or the online tool ( https://www.bioinformatics.com.cn ).

This study revealed alterations in lipid metabolism in the normal myometrium from patients with adenomyosis, in which glycerophospholipids (especially lysophospholipids), sphingolipids (Cers and their hexosyl products), FFAs, TAGs, and CEs were the main contributors. A diagnostic model combining five lipid species presented robust performance with an AUC above 0.9 in distinguishing the myometrium of patients with adenomyosis from that of control subjects. Furthermore, DESI-MS anlysis displayed the metabolic heterogeneity of these lipid markers across the ectopic endometrium and adjacent myometrium. Lipid phosphorylation has been known to play an important role in regulating the membrane localization of proteins [ 31 ]. Two kinds of glycerophospholipids, PC and PE, account for more than 50% of the total phospholipids and thus play major roles in the structure and function of eukaryotic membranes [ 32 , 33 ]. One previous report that focused on phospholipid metabolism in the uterus revealed that distinct uterine phospholipid profiles were associated with different receptivity to the embryo in cattle [ 34 ]. Our previous study [22] using a metabolomics approach revealed only a few abnormalities in glycerophospholipids in the myometrium of patients with adenomyosis, while the current study provided a global view at the lipidomic scale. Cers, the central intermediates of sphingolipid metabolism, are critically involved in immune system function and affect the migration of various cell lines [ 35 , 36 ]. A novel discovery in our study was the significant elevation of Cer levels in the myometrium of women with adenomyosis. Cer and glucosylceramide have been found to be abnormally accumulated in the serum and peritoneal fluid of endometriosis patients [ 37 ]. In addition, Cer significantly inhibited spontaneous contractions of the uterus in 19-day pregnant rats, probably through its effect on Ca 2+ signaling in the rat myometrium [ 38 ]. Other reports have emphasized the importance of Cer in mediating pro-inflammatory actions and angiogenesis by promoting the production of prostaglandins (PGE 2 and PGF 2α ), interleukin 6 (IL-6), IL-8, and macrophage colony-stimulating factor (M-CSF) in the human endometrium [ 39 , 40 ]. High levels of FFAs, particularly saturated FFAs, can activate inflammatory responses via a variety of mechanisms [ 41 ]. Saturated FFAs also have lipotoxic effects, resulting in oxidative stress and apoptosis [ 42 ]. In an adenomyosis mouse model, increased uterine levels of inflammatory cytokines and reactive oxygen species (ROS) induced endometrial cell apoptosis and tissue damage and subsequently reduced the fertility of adenomyosis patients [ 43 ]. Most FFAs also have deleterious effects on endothelium function, which has been associated with the pathology of adenomyosis [ [44] , [45] , [46] ]. Moreover, FAs, particularly FFAs, are critical constituents in the formation of biomembrane lipids and signaling molecules, and they also function as key substrates in energy metabolism. Aberrant FA metabolism is frequently observed in tumor tissues [ 47 ]. The FAs necessary for the growth and proliferation of malignant tumors are predominantly derived from the de novo synthesis pathway, a specific phenotype commonly observed in malignant tumors. The FFAs synthesized by tumor cells are primarily composed of 16-carbon saturated palmitic acid [ 48 ]. Our lipidomic data demonstrated a significant elevation in the total FFAs content within the myometrium of women diagnosed with adenomyosis, in comparison to those without the condition ( Fig. 1 D). This aberrant increase was predominantly attributed to saturated and monounsaturated FAs ( Fig. S1 ). There was no significant difference in the content of FFAs with a degree of unsaturation of two or more between the two groups. Adenomyosis, although a benign disease, exhibits some characteristics of malignant tumors, such as invasion and metastasis. Our study indicates that the disordered FA metabolism in myometrium of adenomyosis shares some similarities with that in aforementioned tumor tissue. Since targeting FA metabolism has been proved to be a promising therapeutic strategy for tumors, we should further assess the alterations of FA metabolic enzymes in adenomyosis to provide new potential perspectives for its treatment. The etiology of adenomyosis is thought to be multifactorial, and to date, there is no sole theory that can explain the overall pathogenesis of adenomyosis. For an ideal framework of pathological study, it is perhaps more profitable to conduct a navigational scan of all, if possible, distributing factors related to this disease. Fortunately, the global development of omics technology has assisted in obtaining a conclusion that is closer to the full truth. As a continuation of our metabolomics study on adenomyosis [ 22 ], we conducted the present study to further reveal the metabolic disturbances in the uterus of patients with adenomyosis at the lipidomic level. To the best of our knowledge, this is the first lipidomics study of uterine tissues. The lipid composition of human myometrium with and without adenomyosis was dissected systematically, which laid the foundation for further pathologic studies of adenomyosis and, perhaps, other uterine diseases. Meanwhile, DESI-MSI provided precise spatial localization information of the lipid markers to exhibit their heterogeneous distribution between lesion location and adjacent myometrium. This integrated approach could be utilized as a database that provides inspiration for exploring potential therapeutic targets and could serve as a reference for further clinical examination of adenomyosis. At present, there is still an unmet need to further clarify the pathogenesis of adenomyosis so that novel diagnostic methods or therapies can be developed. There are several novel findings in these lipidome data that warrant subsequent research, such as the quantitative abnormalities of Cers and FFAs in the myometrium of patients with adenomyosis. These lipid disorders collectively indicate inflammatory and immune disorders, which have also been mentioned in some previous studies on adenomyosis [ 49 , 50 ]. In contrast to previous reports, the importance of the inflammatory microenvironment inside the myometrium, rather than the endometrium alone, was further highlighted in our research. Although various pathological conditions have been associated with altered metabolism of Cers or saturated FFAs, this is the first report in adenomyosis. These results bring us one step closer by identifying relevant lipid metabolites directly in the human uterus and, therefore, could lead to further exploration of potential targets and even novel therapies for adenomyosis. Functional evaluations of the identified lipids and the key metabolic enzymes involved in adenomyosis are still warranted. Although it is currently premature to recommend an anti-inflammatory diet or specific supplementation (such as omega-3 FAs) as way to alleviate adenomyosis, these results suggest a potential mechanism by which alleviating lipid metabolism to reduce inflammation may be associated with the relief of discomfort caused by adenomyosis. More studies on targeting specific metabolic pathways that have direct clinical implications are needed. Our study suggests that lipids could be potential indicators for discriminating myometrial tissues from patients with or without adenomyosis. Since targeted lipid species could be conveniently tested by shotgun or MSI approaches, lipid markers showed potential as alternatives in the histological diagnosis or rapid intraoperative diagnosis of adenomyosis, for which the diagnostic certainty should be further validated in independent cohorts and multiple centers. Based on the present research, it is necessary to investigate the lipidome characteristics of patients with adenomyosis in different samples (such as endometrium and serum) and different populations. Additionally, the MS-based imaging technique employed in this study has demonstrated its efficacy in elucidating the spatial distribution of specific lipid molecules within the uterus affected by adenomyosis. In the near future, we intend to enhance this imaging method by expanding the mass range and improving its spatial resolution. This will be complemented by unbiased screening strategies to further identify differentially expressed lipid species and systematically explore the correlations among their spatial distributions. One advantage of this study is the integrative profiling of myometrial samples using a high-coverage lipidomics approach and ambient MSI, which yielded quantitative, reproducible, and visualized results, allowing the expansion of this research to other subjects and other populations. Another strength of the present study lies in the novelty of the determined tissues, which are adjacent to lesional sites and should be more responsive to the metabolic changes associated with adenomyosis. One limitation of the study was that although we investigated the lipid phenotypes of the myometrium in patients with and without adenomyosis, it is unclear whether lipid dysfunction contributes to the pathogenesis of adenomyosis or is only a byproduct of disease progression. Moreover, further studies are needed to elucidate whether the present findings are applicable in early adenomyosis, considering that most patients who undergo hysterectomy have severe adenomyosis. In addition, further studies in larger cohorts are required to validate our findings.

In this study, we present the inaugural integrated lipidome profiling of human myometrium, accomplished through the combined application of high-coverage targeted lipidomics and ambient MSI. This strategy enabled us to furnish comprehensive and visual evidences of disordered lipid metabolism in the myometrium from women diagnosed with adenomyosis. These lipid aberrations implicated various metabolic pathways and highlighted processes such as inflammation, cell migration, and immune dysregulation in the context of adenomyosis. Furthermore, a panel comprising five specific lipid species demonstrated high efficacy in distinguishing myometrial tissues from women with and without adenomyosis. Distinct visualization of these lipid markers using DESI-MSI further revealed their heterogeneous metabolic characteristics within adenomyotic lesions and adjacent myometrial tissues. These findings contribute to a deeper understanding of the pathophysiology of adenomyosis and provide valuable insights into the development of biomarkers and therapeutic strategies for managing this condition.

Adenomyosis is a common gynecological disease characterized by the aberrant location of endometrium-like epithelial and stromal tissues in the myometrium of the uterus [ [1] , [2] , [3] ]. It is associated with a spectrum of symptoms that can severely impact female quality of life, including dysmenorrhea, menorrhagia, chronic pelvic pain, and diminished fertility [ 4 , 5 ]. Despite its high prevalence, the molecular mechanisms underlying the pathogenesis of adenomyosis are still poorly understood [ 6 , 7 ]. Although it is likely that multiple pathogenic pathways are involved in adenomyosis, either in cooperation or by acting independently, previous studies regarding the initiation and progression of adenomyosis have focused on the endometrium itself [ [8] , [9] , [10] ]. Nevertheless, the myometrium is the site at which ectopic endometrial tissue invades adenomyosis tissue. One of the most important characteristics of adenomyosis is the presence of hypertrophic, disorganized myometrial layer within the uterus [ 11 , 12 ]. Several animal studies have suggested that adenomyosis might primarily be caused by defects during the formation of myometrium [ [13] , [14] , [15] , [16] ]. These experimental evidences support the hypothesis that myometrial dysfunction may play an important role in the development of adenomyosis. It has been postulated that dysfunctions of the myometrium may facilitate the invasion of endometrial cells [ [15] , [16] , [17] , [18] ]. As such, integrated investigations focused on the myometrium may lead to new insights into the pathogenesis of adenomyosis and identify potential targets for preventing or managing such disorders. To date, there is still a profound lack of knowledge about the uterus, especially the myometrium, from patients with adenomyosis. Therefore, it is highly important to elucidate the myometrial characteristics and underlying mechanisms at the molecular level. Lipids are vital components of biological membranes and play important roles in energy storage and signal transduction in cellular physiology [ 19 , 20 ]. It has been reported that disturbed lipid metabolism extensively affects membrane functions to promote abnormal cell proliferation, metastasis, and invasion in various diseases [ 20 , 21 ]. We hypothesize that myometrial dysfunctions, such as loosening and degradation, are, at least in part, related to abnormal lipid metabolism, which cannot be ignored in the progression of endometrial invasion upon adenomyosis. Our previous study using an untargeted metabolomics approach preliminarily revealed that the levels of several kinds of lipids, such as phosphatidylethanolamine (PE) and phosphatidylcholine (PC), were significantly increased in the myometrium of adenomyosis patients [ 22 ]. Nevertheless, lipids represent a distinctive class of metabolites with low polarity and structural complexity. The coverage of routine metabolomic analysis of lipids is extremely limited due to their special physicochemical properties, thus leaving a great knowledge gap about lipid signatures in the full metabolic spectrum. Despite the resurgence of biochemical and physiological studies on lipids, metabolic disturbances at the lipidome level in adenomyosis patients remain poorly understood. Lipidomics is a major branch of metabolomics that has gradually developed into a separate discipline focused on the systematic characterization and quantification of lipid species [ 19 , 20 ]. It has greatly facilitated the identification of lipid features and biomarkers involved in disease pathogenesis [ 23 , 24 ]. The point is that the homogenizing process during sample preparation before lipidomics analysis using liquid chromatography-tandem mass spectrometry (LC-MS) leads to the loss of lipids’ spatial information. Due to the presence of ectopic endometrium, the uterus affected by adenomyosis exhibits greater spatial heterogeneity compared to the healthy uterus [ 7 , 10 ]. Recently, MS imaging (MSI) has become a powerful tool for directly detecting and localizing thousands of molecules in complex biological tissue [ 25 ]. Desorption electrospray ionization-MSI (DESI-MSI), as one of the most widely used ambient MSI techniques, enables easy detection of lipids with no spatial information loss under atmospheric conditions [ 26 ]. In view of the above advantages, DESI-MSI has been increasingly used to investigate tissue metabolic heterogeneity [ 26 , 27 ]. In this study, we utilized a high-coverage lipidomics approach based on ultra-performance LC (UPLC) coupled with triple quadrupole MS (UPLC-QqQ-MS) to construct the first lipidome atlas of human myometrium. On this basis, the lipidome changes in the myometrium of patients with adenomyosis were comprehensively investigated, revealing dysregulated lipid pathway and potential lipid markers that could identify whether the myometrium invaded by adenomyosis. Subsequently, DESI-MSI was employed to further elucidate the spatial distribution characteristics of these lipid markers in the adenomyosis lesions and surrounding myometrial tissues. The current study may provide a lipid-centric biological insight that is important for clarifying the pathophysiology, as well as identifying potential biomarkers of adenomyosis.

The authors declare that there are no conflicts of interest.