{"paper_id":"1d4b4e7f-d3ae-45e3-ba87-66c397ace7c8","body_text":"R E S E A R C H Open Access\nEndometrium metabolomic profiling\nreveals potential biomarkers for diagnosis\nof endometriosis at minimal-mild stages\nJingjie Li 1†, Lihuan Guan 2†, Huizhen Zhang 2, Yue Gao 2, Jiahong Sun 2, Xiao Gong 4, Dongshun Li 2, Pan Chen 3,\nXiaoyan Liang 1, Min Huang 2 and Huichang Bi 2*\nAbstract\nBackground: The sensitivity and specificity of non-invasive diagnostic methods for endometriosis, especially at\nearly stages, are not optimal. The clinical diagnostic indicator cancer antigen 125 (CA125) performs poorly in the\ndiagnosis of minimal endometriosis, with a sensitivity of 24%. Therefore, it is urgent to explore novel diagnostic\nbiomarkers. We evaluated the metabolomic profile variation of the eutopic endometrium between minimal-mild\nendometriosis patients and healthy women by ultra-high-performance liquid chromatography coupled with\nelectrospray ionization high-resolution mass spectrometry (UHPLC-ESI-HRMS).\nMethods: Our study comprised 29 patients with laparoscopically confirmed endometriosis at stages I-II and 37 infertile\nwomen who underwent diagnostic laparoscopy combined with hysteroscopy from January 2014 to January 2015.\nEutopic endometrium samples were collected by pipelle endometrial biopsy. The metabolites were quantified by\nUHPLC-ESI-HRMS. The best combination of biomarkers was then selected by performing step-wise logistic regression\nanalysis with backward elimination.\nResults: Twelve metabolites were identified as endometrios is-associated biomarkers. The eutopic endometrium\nmetabolomic profile of the endometriosis patients was characterized by a significant increase in the concentration of\nhypoxanthine, L-arginine, L-tyrosine, leucine, lysine, inosine, omega-3 arachidonic acid, guanosine, xanthosine,\nlysophosphatidylethanolamine and asparagine. In contrast,the concentration of uric acid was decreased. Metabolites\nwere filtered by step-wise logistic regression with backward elimination, and a model containing uric acid, hypoxanthine,\nand lysophosphatidylethanolamine was constructed. Receiver-operating characteristic (ROC) analysis confirmed the\nprognostic value of these parameters for the diagnosis of minimal/mild endometriosis with a sensitivity of 66.7% and a\nspecificity of 90.0%.\nConclusions:Metabolomics analysis of the eutopic endometrium in endometriosis was effectively characterized by\nUHPLC-ESI-HRMS-based metabolomics. Our study supports the importance of purine and amino acid metabolites in\nthe pathophysiology of endometriosis and provides potential biomarkers for semi-invasive diagnosis of early-stage\nendometriosis.\nKeywords: Endometriosis, Metabolomics, UHPLC-ESI-HRMS, Eutopic endometrium\n* Correspondence: bihchang@mail.sysu.edu.cn\n†Equal contributors\n2School of Pharmaceutical Sciences in Sun Yat-sen University, 132#\nWaihuandong Road, Guangzhou, University City, Guangzhou 510006,\nPeople’s Republic of China\nFull list of author information is available at the end of the article\n© The Author(s). 2018 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0\nInternational License ( http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and\nreproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to\nthe Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver\n(http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.\nLi et al. Reproductive Biology and Endocrinology  (2018) 16:42 \nhttps://doi.org/10.1186/s12958-018-0360-z\n\nBackground\nEndometriosis is a chronic, benign gynaecological dis-\norder characterized by the presence of endometrial cells\nat extrauterine sites and associated with chronic pain\nand infertility. This disease is a highly prevalent disease,\npresenting in 10 –15% of reproductive age women and\napproximately 25 to 50% of infertile women [ 1, 2]. Endo-\nmetriosis has a severe impact on socioeconomics and\nthe quality of life of patients [ 3]. Endometriosis is classi-\nfied into minimal (I), mild (II), moderate (III) and severe\n(IV) stages [ 4]. The incidence of minimal or mild endo-\nmetriosis is more frequent than advanced endometriosis.\nMinimal or mild endometriosis is peritoneal or ovarian\nendometriotic implants and filmy adhesions on the fallo-\npian tubes or ovaries. The presence of early-stage endo-\nmetriosis is associated with poor oocyte quality, lower\nfertilization rate and embryonic developmental compe-\ntence [ 5, 6]. However, no substantial pelvic anatomical\nchanges have been identified. In addition, atypical symp-\ntoms or even no symptoms increase the difficulty of\ndiagnosis in minimal or mild endometriosis, which can\nbe delayed on average by 8 to 11 years [ 7]. Currently,\nthe sensitivity and specificity of non-invasive diagnostic\nmethods for endometriosis, especially early-stage, are not\noptimal. The clinical diagnostic indicator cancer antigen\n125 (CA125) performs poorly in diagnosing minimal\nendometriosis, with a sensitivity of 24% [ 8]. Therefore, it\nis urgent to explore novel diagnostic biomarkers.\nMetabolomics has emerged as a powerful and reliable\ntool to identify metabolites and biomarkers present in\nthe biological system under a given physiological condi-\ntion. Metabolites not only represent the final products of\nbiological regulatory processes but also act as communi-\ncators between the information-rich genome and the\nfunctional phenotype. In the past few years, several stud-\nies identified a list of potential diagnostic candidates in\nperitoneal fluid, blood and urine from endometriosis\npatients at different stages of disease and menstrual\ncycle [ 9, 10]. However, potential biomarkers from the\neutopic endometrium remain unknown. Therefore, in\nthe current study, ultra-high -performance liquid chro-\nmatography coupled with ele ctrospray ionization high-\nresolution mass spectrometry (UHPLC-ESI-HRMS) was\nused to investigate the metabolomic profile of the euto-\npic endometrium between minimal/mild endometriosis\npatients and controls. Twelve metabolites were identi-\nfied as endometriosis-associated biomarkers. The euto-\npic endometrium metabolomic profile of endometriosis\npatients was characterized by a significant increase in\nthe concentration of hypoxanthine, L-arginine, L-tyrosine,\nleucine, lysine, inosine, omega-3 arachidonic acid, guano-\nsine, xanthosine, lysophosphatidylethanolamine and as-\nparagine. In contrast, the concentration of uric acid was\ndecreased. Our study provides potential biomarkers for\nthe semi-invasive diagnose of endometriosis at minimal-\nmild stages.\nMethods\nSubject selection\nPatient recruitment was carried out at the Sixth Hospital\nof Sun Yat-sen University, and analysis of the endomet-\nrium metabolomic profiles was performed at the School\nof Pharmaceutical Sciences at Sun Yat-sen University.\nEutopic endometrium was collected from 68 volunteers\n(21–38 years old, body mass index less than 30 kg/m 2)\nfrom January 2014 to January 2015 who underwent\ndiagnostic laparoscopy combined with hysteroscopy\nbecause of infertility. Clinical diagnosis and classification\nof subjects were performed through laparoscopic surgery\nto visually confirm the presence of endometriotic\nlesions. Surgery was carried out on the third to fifth day\nafter menstrual cessation. All participants had regular\nmenstrual cycles (between 21 and 35 days) without\nhormonal treatment or use of an intrauterine device in\nthe 3 months before sample collection. Endometrial\ntissues were obtained via Pipelle biopsy during surgery\non the 3rd-5th day after the end of their menstrual\nbleeding. The severity of endometriosis was determined\naccording to the American Society of Reproductive\nMedicine revised system [ 4]. Patients diagnosed with\nendometrial polyp, endometritis, submucous myoma or\nhydrosalpinx should be excluded after confirmation with\nhysteroscopy combined with laparoscopy and further\nconfirmation by histology. Two volunteers diagnosed by\nhysteroscopy with endometrial polyps were excluded.\nNo other pathologies were detected in the 68 volunteers.\nThree volunteers were highly suspicious for endome-\ntrioma on 2 ultrasounds with more than 3 months inter-\nval. The mean sizes of the cysts were 8 mm, 7 mm and\n8 mm, which were confirmed during surgery. Clinical\ninformation associated with each sample group is sum-\nmarized in Table 1. After collection, specimens were im-\nmediately placed into microtubes and preserved in\nliquid nitrogen until analysis. This study received the ap-\nproval of the institutional review board, and all patients\ngave their written informed consent (approval number:\nG2012021).\nSample preparation for metabolomics\nEndometrial tissues were obtained from 37 healthy women\n(Control) and 29 women with endometriosis. Sample prep-\naration was performed according to a previous report with\nslight modifications [ 11]. Briefly, 400 μL of 50% chilled\nmethanol was added to 20 mg of tissue sections in tubes\ncontaining ceramic beads for homogenization by using a\nPrecellys 24 homogenizer (Bertin, France). The supernatant\nwas transferred into a fresh tube, and 800 μL of chilled\n100% acetonitrile was added to precipitate the protein.\nLi et al. Reproductive Biology and Endocrinology  (2018) 16:42 Page 2 of 10\n\nSamples were centrifuged at 18000×g at 4 °C for 15 min. A\ntotal of 500 μL of supernatant was transferred to a fresh\ntube and dried under vacuum. Samples were re-suspended\nin 200 μL of 70% acetonitrile for hydrophobic interaction li-\nquid chromatography (HILIC) mode or in 35% acetonitrile\nfor reversed-phase liquid chromatography (RPLC) mode\nand then centrifuged at 18000×g at 4 °C for 5 min. Finally,\n5 μL of supernatant was transferred to a UPLC vial and\ninjected for UHPLC-ESI-HRMS analysis. The quality con-\ntrol (QC) samples comprised 5 μL of each sample, repre-\nsenting a universal set of metabolites for this study. In\naddition, blank samples were 70% acetonitrile or 35%\nacetonitrile.\nUHPLC-ESI-HRMS measurements of endometrial tissues\nAccording to our previously reported method [ 12], chro-\nmatography was performed using an Ultimate 3000 HPLC\nsystem (Dionex Corporation, Sunnyvale, CA) coupled to a\nQE x a c t i v e™ benchtop Orbitrap high-resolution mass spec-\ntrometer (Thermo Fisher Scientific, San Jose, CA). For the\nHILIC mode, an Atlantis Silica HILIC 3 μmc o l u m n\n(100 mm × 2.1 mm, Waters, Milford, MA, USA), total run\ntime 30 min, was employed. Solvent A was 95% acetonitrile\ncontaining 10 mM ammonium formate and 0.1% formic\nacid, and solvent B was 10 mM ammonium formate and\n0.1% formic acid in 50% acetonitrile. The linear gradient\nused was as follows: holding in 100% A for 0 –1 min,\nincreasing to 100% B linearly for 20 min and washing the\ncolumn for the next 4.9 min, then returning to 100% A\nuntil 30 min for column equilibration with a flow rate of\n0.3 mL/min. For the RPLC mode, samples were injected\nonto an Xterra MS C18 5 μm column (100 mm × 2.1 mm,\nWaters, Milford, MA, USA). The mobile phase consisted\nof 0.1% formic acid in water (A) and 100% acetonitrile (B).\nTable 1 Characteristics of participants for endometriosis\npatients and controls\nEndometriosis\npatients (n = 29)\nControl group\n(n = 37)\nP\nAge (years) 29.69 ± 3.19 29.74 ± 3.43 0.9543\nBMI (kg/m2) 21.04 ± 2.08 21.89 ± 3.22 0.2916\nAMH (ng/ml) 4.35 ± 2.74 6.4 ± 4.68 0.0661\nUric acid(μmol/l) 271.2 ± 67.38 289.1 ± 63.15 0.3018\nThe day of sampling 8.667 ± 0.2108 8.45 ± 0.3033 0.5576\nThe length of menstruation 5.19 ± 0.3209 5 ± 0.3839 0.7044\nEndometriosis stage\nI stage 19 N/A\nII stage 10 N/A\nOvarian endometriomas 3 N/A\nFig. 1 Metabolomic analysis of endometrial tissues from patients with endometriosis ( n = 29, blue diamonds) and healthy controls ( n = 37, red\ndiamonds) under positive ionization mode. Score scatter plots for HILIC ( a) and RPLC ( b) modes and OPLS-DA loadings S-plots for HILIC ( c) and\nRPLC (d) modes. The major ions are labelled in the S-plot\nLi et al. Reproductive Biology and Endocrinology  (2018) 16:42 Page 3 of 10\n\nThe flow rate was kept at 0.3 mL/min during a 22-min\nrun with the following gradient: 100% A for 2 min to 52%\nA at 4 min to 30% A at 11 min to 25% A at 14 min and\nkept at 100% B from 16 min to 17 min and 100% A from\n18 min to 22 min. The column temperature was kept at\n40 °C. Mass spectrometry was performed with an electro-\nspray ionization source both in positive and negative\nionization modes under the following conditions: the\nspray voltage was 3.5 kV. The capillary and aux gas heater\ntemperature were 300 °C and 350 °C, respectively. Nitro-\ngen was used as sheath gas (40 arbitrary) and auxiliary gas\n(10 arbitrary). Data were acquired from 80 to 900 mass-\nto-charge (m/z) for mass scanning, and the step collision\nenergy 15, 30, 45 eV was used for MS/MS fragmentation\nof ions. QC samples were injected intermittently to\naccount for the reproducibility and stability of the\nUHPLC-ESI-HRMS data [13].\nData analysis\nThe mass spectra data were pre-processed by SIEVE 2.2\n(Thermo Fisher Scientific, San Jose, CA) to remove the\nbackground and generate a multivariate data matrix con-\ntaining aligned peak areas with matched m/z and retention\ntimes. Then, SIMCA 13.0 software (Umetrics, Kinnelon,\nNJ) was applied to find the features that were responsible\nfor the discrimination of the groups. An orthogonal partial\nleast squares discriminant analysis (OPLS-DA) was used to\nmaximize the group discrimination. The candidate markers\nwere selected by examining the S-plot based on the variable\nimportance (VIP) value, which was more than 1.0. The\nidentification of the metabolites was confirmed by compari-\nsons of fragmentation spectra and m/z through three main\nonline databases: Metlin (http://metlin.scripps.edu), HMDB\n(http://www.hmdb.ca/ ), and mzcloud ( https://www.\nmzcloud.org/ )[ 14, 15]. To assess the strength of as-\nsociation between individual metabolites and minimal/\nmild endometriosis, a step-wise logistic regression analysis\nwith backward elimination was used to establish a\nmodel and filter crucial metabolites. The receiver op-\nerating characteristic (ROC) curve was plotted, and\nthe area under the curve (AUC) was calculated. The\noptimal point on the ROC curve provided the best\ntrade-off between sensitivity and specificity. Statistical\ntesting was carried out by SPSS 19.0 software (IBM\nAnalytics, USA). Data were assessed for normality of\ndistribution using the Shapiro –Wilk test first. Un-\npaired Student ’s t-test or the non-parametric Mann –\nWhitney U-test was evaluated with a 95% confidence\nlevel for statistical analysis between the two groups.\nFalse discovery rate (FDR) control was performed by\nthe SAS PROC MULTITEST with the FDR option\nFig. 2 Metabolomic analysis of endometrial tissues from patients with endometriosis ( n = 29, blue diamonds) and healthy controls ( n = 37, red\ndiamonds) under negative ionization mode. Score scatter plots for HILIC ( a) and RPLC ( b) modes and OPLS-DA loadings S-plot for HILIC ( c) and\nRPLC (d) modes. The major ions are labelled in the S-plot\nLi et al. Reproductive Biology and Endocrinology  (2018) 16:42 Page 4 of 10\n\n(SAS Inst, Cary, North Carolina, USA). P-values less\nthan 0.05 were considered statistically significant\nwhile controlling FDR at 0.05.\nResults\nCharacteristics of participants with endometriosis and\ncontrols\nClinical information associated with each sample group\nis summarized in Table 1. A total of 66 volunteers were\nrecruited in this study. Twenty-nine patients had laparo-\nscopically confirmed endometriosis, staged as minimal\n(n =1 9 )a n dm i l d(n = 10). Three patients had a laparoscop-\nically documented presence of ovarian endometrioma. All\nof the endometriomas were histologically confirmed. The\nmean sizes of all the cysts were less than 1 cm. Age, BMI,\nmenstrual cycle, AMH and uric acid in serum were com-\nparable between the two groups ( P > 0.05). Both the mean\nday of sampling and the length of menstruation were not\nsignificantly different between the endometriosis patients\nand the control group. No volunteer had a history of smok-\ning in this study.\nMultivariate statistical analysis of difference between the\nendometriosis and control groups\nThe alignment of all the features in all samples gener-\nated a data matrix by SIEVE 2.2 software with an abun-\ndance of 5388 features under HILIC mode and 3424\nfeatures under RPLC mode. To compare the overall vari-\nation of metabolic profiles between the endometriosis\npatients and healthy controls, a classification model was\nbuilt by the supervised OPLS-DA, which revealed a clear\nseparation between the two groups (Figs. 1a, b and 2a, b ).\nThe model also showed that samples from humans had\ngreat individual differences. An OPLS-DA loadings S-plot\nwas performed to highlight significantly different variables\nin the two groups (Figs. 1c, d and 2c, d). Each point repre-\nsented a detected ion (variables). The further away from\nthe plot origin an ion point lies, the more the ion contrib-\nutes to the difference between the two study groups.\nTherefore, variables plotted at the top or bottom were\nchanged most significantly. Metabolite features of interest\nwere selected by a VIP value > 1.0. With such a strategy,\n450 variables from the HILIC mode results and 469\nFig. 3 Identified metabolites with increasing contributions to the difference in metabolomic profiles between the two groups based on VIP scores\nLi et al. Reproductive Biology and Endocrinology  (2018) 16:42 Page 5 of 10\n\nvariables from the RPLC mode results were considered to\nhave impact on the model.\nIdentification of detected metabolites\nThe m/z of the selected variables and their MS/MS frag-\nmentation spectra were used for comparison with com-\npounds annotated in the online databases. Finally, 27\nmetabolites from positive and negative ionization modes\nwere uniquely identified on the basis of exact mass and\nretention time. Among them, levels of 12 metabolites cor-\nresponding to high variable importance (VIP > 1) (Fig. 3)\nwere different between the endometriosis and control\ngroups ( P < 0.05). In addition, their detailed information\nis summarized in Table 2 and labelled in the S-plot\n(Figs. 1c, d and 2c, d ). Obviously, levels of hypoxan-\nthine, L-arginine, L-tyrosine, leucine, lysine, inosine,\nomega-3 arachidonic acid, guanosine, xanthosine, lyso-\nphosphatidylethanolamine and asparagine were higher\nin the endometriosis group than in the control group,\nwhereas the level of uric acid was higher in the control\ngroup (Fig. 4). It is noteworthy that the xanthosine level\nin the endometriosis group was 2.53-fold higher than\nthat of the control group, while the amount of uric acid\nwas decreased by half, indicating that purine metabol-\nism was disturbed in endometriosis patients. After\nusing step-wise multivariate logistic regression analysis\nwith backward elimination, a model with three predic-\ntors was established, including uric acid, hypoxanthine\nand lysophosphatidylethanolamine, with a sensitivity of\n66.7% (95% CI: 0.417 –0.875) and a specificity of 90.0%\n(95% CI: 0.600 –1.000). The receiver operating charac-\nteristic (ROC) curve shows improved effects of adding\nseparate variables to the model. The apparent AUC\no ft h eR O Cc u r v ef o rt h em o d e lp r e d i c t i n ge n d o m e t -\nriosis at the minimal/mild stages was 0.868 (95% CI:\n0.774–0.963) (Fig. 5). The combination of three vari-\nables led to a curve with significantly better perform-\nance and allows very good discrimination between\nendometriosis patients at early stages and controls.\nDiscussion\nIn the current study, we applied a UHPLC-ESI-HRMS-\nbased metabolome profiling approach to investigate meta-\nbolic changes in the eutopic endometrium samples from\nendometriosis patients and identified metabolites for early-\ndiagnosed endometriosis. In this regard, 11 metabolites\nincluding hypoxanthine, L-arginine, L-tyrosine, leucine,\nlysine, inosine, omega-3 arachidonic acid, guanosine,\nxanthosine, lysophosphatidylethanolamine and aspara-\ngine were significantly increased in the endometriosis\ngroup, whereas the uric acid level was decreased. The\nglobal metabolomics and subsequent multivariate analysis\nclearly distinguished metabolic changes in the endometri-\nosis patients from those in the matched controls. A com-\nbination of three predictors (uric acid, hypoxanthine and\nlysophosphatidylethanolamine) shows a very good poten-\ntial for use in diagnosing endometriosis at early stages.\nTable 2 Summary of the data from the 12 features found in positive and negative ionization modes contributing to the\ndiscrimination of endometrial tissues between endometriosis patients and healthy controls\nm/za tR (min)b Metabolite Molecular formula Adduct Fold change c P valued Adj P valuee\nHILIC mode\n131.0462 10.793 Asparagine C 4H8N2O3 M-H 1.44 0.013 0.0173\n132.1019 8.833 Leucine C6H13NO2 M + H 1.68 0.002 0.0120\n137.0455 4.431 Hypoxanthine C5H4N4O M + H 1.64 0.033 0.0360\n167.0209 5.283 Uric acid C5H4N4O3 M-H 0.54 0.000 0.0053\n267.0737 4.539 Inosine C10H12N4O5 M-H 1.58 0.037 0.0370\n282.0844 6.017 Guanosine C10H13N5O5 M-H 1.55 0.023 0.0276\n283.0685 4.771 Xanthosine C10H12N4O6 M-H 2.53 0.008 0.0137\nRPLC mode\n147.1125 0.696 Lysine C6H14N2O2 M + H 1.54 0.008 0.0137\n175.1186 0.703 L-Arginine C6H14N4O2 M + H 1.41 0.004 0.0137\n182.0808 1.091 L-Tyrosine C9H11NO3 M + H 1.47 0.006 0.0137\n303.2329 15.245 Omega-3 Arachidonic acid C 20H32O2 M-H 1.57 0.005 0.0137\n478.2935 10.770 LPE (18:1(9Z)/0:0) C23H46NO7P M-H 1.20 0.013 0.0173\nam/z is the detected mass to charge ratio from LC-MS/MS runs\nbRetention time in minutes\ncThe fold change of the endometriosis group vs the control group (a higher ratio indicates a higher level of expression of a compound in the EMS group)\ndP value is the significance level of the difference between the two groups\neP values were adjusted for false discovery rate correction at the significance level of 5%\nLPE lysophosphatidylethanolamine, PC phosphatidylcholine\nLi et al. Reproductive Biology and Endocrinology  (2018) 16:42 Page 6 of 10\n\nHowever, a study with a larger sample size is needed to\nobtain stronger evidence and avoid wide confidence inter-\nvals in the future.\nEndometriosis is a disease characterized by the pres-\nence of endometrial glands and stroma at ectopic sites.\nThis gynaecological disease occurs in approximately 10%\nof women of reproductive age, who present symptoms\nincluding dyspareunia, dysmenorrhoea, chronic pelvic\npain and subfertility [16]. Laparoscopy is the gold standard\nfor the diagnosis of endometriosis. However, laparoscopy\nis an invasive operation with several limitations, such as\nsurgery-associated risks and financial burden [ 17]. So far,\nit has not been able to accurately predict the presence of\nendometriosis based on non-invasive way. Ultrasound\ncould efficiently detect the presence of ovarian endome-\ntriomas, but it is inadequate for the diagnosis of peritoneal\nendometriosis, deep endometriosis and endometriosis-as-\nsociated adhesions. CA125 is the most frequently\nFig. 4 Scatter diagram of 12 selected metabolites. Data are expressed as the mean ± SD. * P < 0.05, **P < 0.01, ***P < 0.001, endometriosis patients\n(EMS, n = 29) vs healthy controls (Control, n = 37)\nLi et al. Reproductive Biology and Endocrinology  (2018) 16:42 Page 7 of 10\n\nstudied biomarker for endometriosis [ 18]. However, it\nmay be more beneficial for diagnosing advanced stages\n(III–IV) compared to early stages (I - II) [ 19]. Hirsch et\nal. showed that CA 125 performs poorly in diagnosing\nminimal endometriosis, with a sensitivity of 24% [ 8]. At\npresent, over 100 potential biomarkers of endometriosis\nhave been reported; however , few markers were useful\nfor the detection of minimal –mild endometriosis [ 20].\nThe diagnosis of endometriosis can be delayed, on aver-\nage, by 8 to 11 years, which leads to significant symp-\ntoms [ 7]. Thus, the cost-effectiveness of endometriosis\nd i a g n o s i sa n dt h e r a p ys h o u l db eu r g e n t l yi m p r o v e d .\nIncreasing evidence shows that metabolomics using\neasily accessible human biosamples has become an ef-\nfective tool to explore diagnostic biomarkers and investi-\ngate disease progression [ 21–24]. Metabolomics analysis\nin endometriosis has been performed in peripheral\nblood, peritoneal fluid, follicular fluid and urine [ 25–28].\nAccording to the widely accepted theory of retrograde\nmenstruation, the endometrium is the source of ectopic\nendometriotic foci. Previous studies showed that the\neutopic endometrium contributed to the pathogenesis\nof endometriosis due to the increase of proliferation,\nmigration and invasion of ectopic endometrium [29–33]. In\nthis study, we did not detect a significant difference in pa-\ntients’ uric acid level in serum. However, uric acid level was\nreduced by about half in the eutopic endometrium of pa-\ntients. The differential expression of uric acid in the serum\nand endometrium indicated that the eutopic endometrium\nwas more representative, stable and similar to ectopic le-\nsions compared to other samples. In addition, we utilized a\nsemi-invasive way of sampling. The pipelle endometrial\nbiopsy can be used without cervical dilatation in the out-\npatient department and causes minimal discomfort. Thus,\nmetabolomics analysis via pipelle endometrial biopsy is a\nviable method to explore molecular markers of endometri-\nosis. All the samples were obtained strictly on the third to\nfifth day after menstrual cessation because we tried to\nexamine samples in the early follicle phase. Although theor-\netically we should sample on the same day of the menstrual\ncycle, each patient’s menstrual period and speed of follicle\ngrowth varies. We chose this time to collect samples based\non hysteroscopic surgical requirements and patient compli-\nance. Unfortunately, we did not collect enough data on pa-\ntients with advanced endometriosis to analyse because few\npatients in stages III-IV in our centre had not been exposed\nto hormonal drugs within 3 months.\nPurine metabolites, including inosine, xanthosine, guano-\nsine and hypoxanthine, were significantly upregulated in\nthe eutopic endometrium, whereas uric acid, as the end\nproduct of purine metabolism, was remarkably downregu-\nlated. This observation indicates that local purine salvage is\nFig. 5 Receiver operating characteristic curves for the model of endometriosis at minimal/mild stages\nLi et al. Reproductive Biology and Endocrinology  (2018) 16:42 Page 8 of 10\n\npotentially impaired. Multiple enzymes participate in the\npurine metabolism process. Among them, purine nucleo-\nside phosphorylase (PNP) is one of the essential enzymes\nmediating the generation of uric acid from purines. High\nlevels of expression of this enzyme are postulated to reflect\nextensive programmed cell dea th during the implantation\nprocess [34, 35]. In addition, pharmacological inhibition of\nPNP has been demonstrated to be embryo-lethal or terato-\ngenic [ 34]. Our data indicated that the accumulation of\nthese purine metabolites and decrease in uric acid level in\nthe eutopic endometrium may be due to suppressed PNP\nexpression. A previous study applied parallel gene expres-\nsion profiling using high-density oligonucleotide microarrays\nto investigate the regulationof gene expression in the endo-\nmetrium [36], and reduction of PNP expression was found\nin endometriosis patients, which supports our hypothesis.\nEndometriosis has been known to exhibit similar features\nof malignancy [ 37, 38]. Clinical and microscopic examin-\nation proved that endometriosis exhibited cancer-like char-\nacteristics, demonstrated by uncontrolled growth, cell\ninvasion, neovascularization and apoptosis [39]. Almost all\nof the amino acids have been reported to be upregulated in\ncarcinoma tissues in previous studies [40]. In cancer cells, a\nhigh energy demand leads to the alteration of biochemistry\nincluding citric acid cycle dysfunction [ 41]. Therefore, al-\nternative routes of carbon backbone delivery are required.\nThe increased ectopic endometrium levels of L-arginine,\nL-tyrosine, leucine, lysine and asparagine observed in the\npresent study might be caused by the alteration of energy\nmetabolism and high turnover of structural protein. These\nobservations are in agreement with a study carried out on\nserum samples of endometriosis [ 42] and metabolic alter-\nations observed in oesophageal cancer patients [ 43, 44].\nConclusion\nMetabolomics provides a powerful approach to explore\ndiagnostic biomarkers by analysing changes in metabolic\nprofiles. Overall, this study is the first to demonstrate a\ncomprehensive analysis of metabolic changes in the euto-\npic endometrium in endometriosis at early stages. Metabo-\nlites involved in purine, amino acid and arachidonic acid\nmetabolic pathways could be potential biomarkers for early\ndiagnosis of endometriosis. These findings provide poten-\ntial biomarkers for semi-invasive diagnosis of endometri-\nosis at minimal-mild stages in clinical practice. The\nimplications of these individual metabolites in the patho-\nphysiology and analysis of metabolites in all stages of endo-\nmetriosis have now to be further studied.\nFunding\nThe authors would like to acknowledge the support from the National\nNatural Science Foundation of China (No. 81601347, 81503156, 81320108027),\nNatural Science Foundation of Guangdong Province (No. 2014A030310096) and\nPublic Welfare Research and Capacity Building Fund of Guangdong\n(No. 2016A020218006).\nAvailability of data and materials\nThe data for this study are available from the corresponding author upon\nreasonable request.\nAuthors’ contributions\nJJL conceived of the study, wrote the manuscript and supervised patient\nrecruitment. LHG, HZZ, YG and DSL contributed to the study execution and\nanalysis and interpretation of the data. XG performed data analysis and\ninterpretation. JHS and PC reviewed the manuscript. HCB, MH and XYL\nsupervised patient recruitment, collected and evaluated data, and drafted,\nedited and approved the final version of this paper for submission. All\nauthors read and approved the final manuscript.\nEthics approval and consent to participate\nThis study received approval from the Sixth Affiliated Hospital of Sun Yat-sen\nUniversity Research Ethics Committee (approval number: G2012021).\nCompeting interests\nThe authors declare that they have no competing interests.\nPublisher’sN o t e\nSpringer Nature remains neutral with regard to jurisdictional claims in\npublished maps and institutional affiliations.\nAuthor details\n1Center of Reproductive Medicine, the Sixth Affiliated Hospital, Sun Yat-sen\nUniversity, Guangzhou, China. 2School of Pharmaceutical Sciences in Sun\nYat-sen University, 132# Waihuandong Road, Guangzhou, University City,\nGuangzhou 510006, People ’s Republic of China. 3Pharmacy Department, the\nFirst Affiliated Hospital, Sun Yat-sen University, Guangzhou, China. 4School of\nPublic Health, Guangdong Pharmaceutical University, Guangzhou, China.\nReceived: 15 January 2018 Accepted: 24 April 2018\nReferences\n1. Olive DL, Pritts EA. Treatment of endometriosis. N Engl J Med. 2001;345:\n266–75.\n2. Counseller VS, Crenshaw JL Jr. A clinical and surgical review of endometriosis.\nAm J Obstet Gynecol. 1951;62:930–42.\n3. Simoens S, Dunselman G, Dirksen C, Hummelshoj L, Bokor A, Brandes I,\nBrodszky V, Canis M, Colombo GL, DeLeire T, et al. The burden of\nendometriosis: costs and quality of life of women with endometriosis and\ntreated in referral centres. Hum Reprod. 2012;27:1292 –9.\n4. Revised American Society for Reproductive Medicine classification of\nendometriosis. Fertil Steril. 1996;1997(67):817 –21.\n5. Barnhart K, Dunsmoor-Su R, Coutifaris C. Effect of endometriosis on in vitro\nfertilization. Fertil Steril. 2002;77:1148 –55.\n6. Bergendal A, Naffah S, Nagy C, Bergqvist A, Sjoblom P, Hillensjo T. Outcome\nof IVF in patients with endometriosis in comparison with tubal-factor\ninfertility. J Assist Reprod Genet. 1998;15:530 –4.\n7. Hadfield R, Mardon H, Barlow D, Kennedy S. Delay in the diagnosis of\nendometriosis: a survey of women from the USA and the UK. Hum Reprod.\n1996;11:878–80.\n8. Hirsch M, Duffy J, Davis CJ, Nieves Plana M, Khan KS. International collaboration\nto harmonise O, measures for E: diagnostic accuracy of cancer antigen 125 for\nendometriosis: a systematic review and meta-analysis. BJOG. 2016;123:1761–8.\n9. Ahn SH, Singh V, Tayade C. Biomarkers in endometriosis: challenges and\nopportunities. Fertil Steril. 2017;107:523 –32.\n10. Marianna S, Alessia P, Susan C, Francesca C, Angela S, Francesca C, Antonella N,\nPatrizia I, Nicola C, Emilio C. Metabolomic profiling and biochemical evaluation\nof the follicular fluid of endometriosis patients. Mol BioSyst. 2017;13:1213–22.\n11. Cheema AK, Pathak R, Zandkarimi F, Kaur P, Alkhalil L, Singh R, Zhong X,\nGhosh S, Aykin-Burns N, Hauer-Jensen M. Liver metabolomics reveals\nincreased oxidative stress and fibrogenic potential in gfrp transgenic mice\nin response to ionizing radiation. J Proteome Res. 2014;13:3065 –74.\n12. Bi H, Krausz KW, Manna SK, Li F, Johnson CH, Gonzalez FJ. Optimization of\nharvesting, extraction, and analytical protocols for UPLC-ESI-MS-based\nmetabolomic analysis of adherent mammalian cancer cells. Anal Bioanal\nChem. 2013;405:5279–89.\nLi et al. Reproductive Biology and Endocrinology  (2018) 16:42 Page 9 of 10\n\n13. Want EJ, Masson P, Michopoulos F, Wilson ID, Theodoridis G, Plumb RS,\nShockcor J, Loftus N, Holmes E, Nicholson JK. Global metabolic profiling of\nanimal and human tissues via UPLC-MS. Nat Protoc. 2013;8:17 –32.\n14. Yu T, Wang Y, Zhang H, Johnson CH, Jiang Y, Li X, Wu Z, Liu T, Krausz KW,\nYu A, et al. Metabolomics reveals mycoplasma contamination interferes\nwith the metabolism of PANC-1 cells. Anal Bioanal Chem. 2016;408:4267 –73.\n15. Zhang H, Jiang Y, Wu J, Zheng C, Ran X, Li D, Huang M, Bi H. Metabolic mapping\nof Schisandra sphenanthera extract and its active lignans using a metabolomic\napproach based on ultra high performance liquid chromatography with high-\nresolution mass spectrometry. J Sep Sci. 2017;40:574–86.\n16. Dunselman GA, Vermeulen N, Becker C, Calhaz-Jorge C, D'Hooghe T, De Bie\nB, Heikinheimo O, Horne AW, Kiesel L, Nap A, et al. ESHRE guideline:\nmanagement of women with endometriosis. Hum Reprod. 2014;29:400 –12.\n17. Slack A, Child T, Lindsey I, Kennedy S, Cunningham C, Mortensen N,\nKoninckx P, McVeigh E. Urological and colorectal complications following\nsurgery for rectovaginal endometriosis. BJOG. 2007;114:1278 –82.\n18. Check JH. CA-125 as a biomarker for malignant transformation of\nendometriosis. Fertil Steril. 2009;91:e35. author reply e36\n19. Mol BW, Bayram N, Lijmer JG, Wiegerinck MA, Bongers MY, van der Veen F,\nBossuyt PM. The performance of CA-125 measurement in the detection of\nendometriosis: a meta-analysis. Fertil Steril. 1998;70:1101 –8.\n20. May KE, Conduit-Hulbert SA, Villar J, Kirtley S, Kennedy SH, Becker CM.\nPeripheral biomarkers of endometriosis: a systematic review. Hum Reprod\nUpdate. 2010;16:651 –74.\n21. Zhang A, Sun H, Wang X. Serum metabolomics as a novel diagnostic\napproach for disease: a systematic review. Anal Bioanal Chem. 2012;404:\n1239–45.\n22. Patel NR, McPhail MJ, Shariff MI, Keun HC, Taylor-Robinson SD. Biofluid\nmetabonomics using (1)H NMR spectroscopy: the road to biomarker\ndiscovery in gastroenterology and hepatology. Expert Rev Gastroenterol\nHepatol. 2012;6:239 –51.\n23. Weiss RH, Kim K. Metabolomics in the study of kidney diseases. Nat Rev\nNephrol. 2011;8:22 –33.\n24. Rhee EP, Gerszten RE. Metabolomics and cardiovascular biomarker\ndiscovery. Clin Chem. 2012;58:139 –47.\n25. Vicente-Munoz S, Morcillo I, Puchades-Carrasco L, Paya V, Pellicer A, Pineda-\nLucena A. Pathophysiologic processes have an impact on the plasma\nmetabolomic signature of endometriosis patients. Fertil Steril. 2016;106:\n1733–41. e1731\n26. Vicente-Munoz S, Morcillo I, Puchades-Carrasco L, Paya V, Pellicer A, Pineda-\nLucena A. Nuclear magnetic resonance metabolomic profiling of urine\nprovides a noninvasive alternative to the identification of biomarkers\nassociated with endometriosis. Fertil Steril. 2015;104:1202 –9.\n27. Cordeiro FB, Cataldi TR, Perkel KJ, Do Vale Teixeira da Costa L, Rochetti RC,\nStevanato J, Eberlin MN, Zylbersztejn DS, Cedenho AP, Turco EG. Lipidomics\nanalysis of follicular fluid by ESI-MS reveals potential biomarkers for ovarian\nendometriosis. J Assist Reprod Genet. 2015;32:1817 –25.\n28. Rizner TL. Diagnostic potential of peritoneal fluid biomarkers of\nendometriosis. Expert Rev Mol Diagn. 2015;15:557 –80.\n29. Joshi NR, Su RW, Chandramouli GV, Khoo SK, Jeong JW, Young SL, Lessey\nBA, Fazleabas AT. Altered expression of microRNA-451 in eutopic\nendometrium of baboons (Papio anubis) with endometriosis. Hum Reprod.\n2015;30:2881–91.\n30. Laudanski P, Charkiewicz R, Tolwinska A, Szamatowicz J, Charkiewicz A, Niklinski J.\nProfiling of selected MicroRNAs in proliferative Eutopic endometrium of women\nwith ovarian endometriosis. Biomed Res Int. 2015;2015:760698.\n31. Goteri G, Altobelli E, Tossetta G, Zizzi A, Avellini C, Licini C, Lorenzi T,\nCastellucci M, Ciavattini A, Marzioni D. High temperature requirement A1,\ntransforming growth factor beta1, phosphoSmad2 and Ki67 in eutopic and\nectopic endometrium of women with endometriosis. Eur J Histochem.\n2015;59:2570.\n32. Chelariu-Raicu A, Wilke C, Brand M, Starzinski-Powitz A, Kiesel L, Schuring\nAN, Gotte M. Syndecan-4 expression is upregulated in endometriosis and\ncontributes to an invasive phenotype. Fertil Steril. 2016;106:378 –85.\n33. Li Y, An D, Guan YX, Kang S. Aberrant methylation of the E-cadherin gene\npromoter region in endometrium and ovarian Endometriotic cysts of\npatients with ovarian endometriosis. Gynecol Obstet Investig. 2017;82:78 –85.\n34. Witte DP, Wiginton DA, Hutton JJ, Aronow BJ. Coordinate developmental\nregulation of purine catabolic enzyme expression in gastrointestinal and\npostimplantation reproductive tracts. J Cell Biol. 1991;115:179 –90.\n35. Hong L, Mulholland J, Chinsky JM, Knudsen TB, Kellems RE, Glasser SR.\nDevelopmental expression of adenosine deaminase during decidualization\nin the rat uterus. Biol Reprod. 1991;44:83 –93.\n36. Kao LC, Germeyer A, Tulac S, Lobo S, Yang JP, Taylor RN, Osteen K, Lessey\nBA, Giudice LC. Expression profiling of endometrium from women with\nendometriosis reveals candidate genes for disease-based implantation\nfailure and infertility. Endocrinology. 2003;144:2870 –81.\n37. Karanjgaokar VC, Murphy DJ, Samra JS, Mann CH. Malignant transformation\nof residual endometriosis after hysterectomy: a case series. Fertil Steril.\n2009;2037(92):e2019 –21.\n38. Lin J, Zhang X, Chen Y. Mutagen sensitivity as a susceptibility marker for\nendometriosis. Hum Reprod. 2003;18:2052 –7.\n39. Munksgaard PS, Blaakaer J. The association between endometriosis and\ngynecological cancers and breast cancer: a review of epidemiological data.\nGynecol Oncol. 2011;123:157 –63.\n40. Denkert C, Budczies J, Weichert W, Wohlgemuth G, Scholz M, Kind T,\nNiesporek S, Noske A, Buckendahl A, Dietel M, Fiehn O. Metabolite profiling\nof human colon carcinoma –deregulation of TCA cycle and amino acid\nturnover. Mol Cancer. 2008;7:72.\n41. Chen JQ, Russo J. Dysregulation of glucose transport, glycolysis, TCA cycle\nand glutaminolysis by oncogenes and tumor suppressors in cancer cells.\nBiochim Biophys Acta. 2012;1826:370 –84.\n42. Dutta M, Joshi M, Srivastava S, Lodh I, Chakravarty B, Chaudhury K. A\nmetabonomics approach as a means for identification of potential\nbiomarkers for early diagnosis of endometriosis. Mol BioSyst. 2012;8:3281 –7.\n43. Zhang J, Liu L, Wei S, Nagana Gowda GA, Hammoud Z, Kesler KA, Raftery D.\nMetabolomics study of esophageal adenocarcinoma. J Thorac Cardiovasc\nSurg. 2011;141:469 –75. 475 e461 –464\n44. Zhang J, Bowers J, Liu L, Wei S, Gowda GA, Hammoud Z, Raftery D.\nEsophageal cancer metabolite biomarkers detected by LC-MS and NMR\nmethods. PLoS One. 2012;7:e30181.\nLi et al. Reproductive Biology and Endocrinology  (2018) 16:42 Page 10 of 10","source_license":"CC0","license_restricted":false}