{"paper_id":"269e5cc9-73e4-46d8-a429-db049340c134","body_text":"Endometriosis (EM)\nis a gynecological disease associated with chronic\npain and affecting women in their reproductive age. 1  It shares many similarities with metastatic cancer 2  and involves the growth of uterine lining (glandular\nand stromal endometrium endometrial tissues) in regions other than\nthe uterus. The frequency of occurrence ranges from 5 to 10% worldwide\nin reproductively active and infertile women. 3  Although it is considered as a benign gynecological condition, it\nis often associated with an increased risk of malignant transformation.\nSeveral factors are involved in the chronic inflammatory process of\nendometriosis, such as hormones, growth or adhesion factors, antigenic\nglycoproteins (e.g., Zn-alpha2-glycoprotein, CA125, and CA19.9), cytokines,\nchemokines, and oxidative stress. 4 − 6  The two most accepted\ntheories for endometriosis, the retrograde mensuration theory 7  and the metaplasia theory, 8  describe the origin of this enigmatic gynecological disease,\ndue to an abnormal defect in the clearance of menstrual efflux in\nreproductively active women. 9  Women suffering\nwith endometriosis disease most often have symptoms of chronic pelvic\npain, fatigue, dysmenorrhea (pain during periods), dyspareunia (pain\nduring intercourse), heavy menstrual bleeding, and bleeding between\nperiods. 10  Moreover, approximately 47%\nof women with chronic pelvic pain and subfertility have chances of\nchronic endometriosis condition. 11  Clinical\ndiagnosis of this condition is being done via familial and physical\nexamination, transvaginal or ultrasonography, noninvasive biomarker\nanalysis, or laparoscopy. 12  So far, laparoscopy\nhas been defined as the gold standard for diagnosis of endometriosis\nand also used for the surgical treatment of it for years. The revised\nAmerican Society of Reproductive Medicine (rASRM) standard has been\nused for classification of endometriosis, which requires description\nof the lesions and their extent to identify the severity of the disease\nand to decrease false positive results by clinicians. 13  As such, EM lesions and symptoms overlap with those of\nmany other uterine/ovarian medical conditions, causing a delay in\ndiagnosis; therefore, new noninvasive biomarkers for differentiating\nEM from other uterine medical conditions are urgently needed.\nMetabolomics represents a useful analytical tool for quantitative\nand comparative analysis of metabolites in biological samples (blood\nplasma, urine, serum, follicular and endometrial fluid etc.) and identifying\nnew promising biomarkers. The metabolic profiling capabilities of\nNMR in combination with statistical analysis tools have successfully\nbeen applied for understanding the pathogenesis of many diseases as\nwell as for identifying metabolic signatures for early diagnosis,\ntherapeutic monitoring, and predicting disease severity, progression,\nand recurrence. 14  The increasing popularity\nof NMR in clinical metabolomics studies is because of its high-throughput\nnature (>100 samples can be analyzed per day), minimal sample preparation\nrequirement, inherently noninvasiveness, nondestructive and nonselective\nnature, and, on top of all this, high level of experimental reproducibility. 14  Despite several advantages, the discoveries\nof NMR based metabolomics studies focusing on disease diagnostics\nstill lack the clinical translations. A major bottleneck for translational\napplications focusing on disease diagnostics is the lack of a suitable\ninternal standard for reliable quantitation of metabolites, especially\nin blood plasma/serum samples. Scientific efforts have already been\nstarted in this direction, and recently, circulatory metabolites (fumaric\nacid and maleic acid) have been reported as reliable internal standards\nfor NMR analysis of protein precipitated plasma and serum samples. 15  Alternatively in our lab, we are using formate\nas an internal calibration standard for NMR analysis of normal plasma\nand serum samples, and the resulting concentration profiles (estimated\nw.r.t. formate) are then used to estimate the metabolic ratios (ratiometric\nparameters). 14 , 16 − 20  As compared to conventionally used normalized spectral\nfeatures, the circulatory metabolic ratios were found to be more reliable\nin reflecting the pathophysiological state of the patients and correlate\nwell with the clinical parameters or disease severity scores. 16 − 20  The approach has been extended here further for investigating the\ncirculatory levels of proline and glutamine and the proline to glutamine\nratio (PQR) in female endometriosis (EM) patients with respect to\nnormal control (NC) female subjects. The purpose of selecting these\ntwo specific amino acids (i.e., proline and glutamine) is that EM\nis often associated with altered inflammatory and immune processes\nand shares some cancer-like characteristics such as activated glutaminolysis 21 , 22  and increased proline biosynthesis. 23 − 25  Both proline and glutamine\nare interconvertible metabolically, and studies have shown their roles\nin cancer cell metabolic reprogramming, redox homeostasis, occurrence/development\nof cancer (including endometrial carcinoma), and its further progression\ntoward the malignant state. 22 − 31  So based on these pathophysiological hallmarks, we hypothesized\nthat the circulatory proline to glutamine ratio (PQR) would be altered\nin EM and may serve as an indicative biomarker to improve the clinical\ndiagnosis of EM. The metabolic alterations will further underscore\nthe importance of these two nonessential amino acids in cancer metabolism\nand rational design of a new therapeutic strategy for EM.\n\nThe present study involved\n39 endometriosis (EM) patients and 48 normal control (NC) subjects,\nand all were in their reproductive age.  Table  1  represents the physical and clinical features\nof all the subjects (total  N  = 87) involved in this\nstudy. Out of 39 EM patients, 15 were diagnosed with stages I and\nII; however, 24 were diagnosed with stages III and IV, referred to\nhere as moderate EM (MEM) and severe EM (SEM), respectively. Consistent\nwith various previous reports, the circulatory glycoprotein antigen\nCA-125 levels were significantly higher in EM and so were various\nother parameters including BMI and hormonal profiles 32  ( Table  1 ).\nN  = total number\nof subjects, ns = not significant,  p  > 0.05, * p  ≤ 0.05, ** p  ≤ 0.01, *** p  ≤ 0.001. Student’s  t  test\nwas used to compare the mean for the endometriosis and control subjects.\nTwo-tailed  p -values less than 0.05 were considered\nstatistically significant. Abbreviations used: BMI, body mass index;\nkg, kilogram; mg, milligram; μg, microgram; ng, nanogram; pg,\npictogram; mL, milliliter; dL, deciliter; mIU, milli-international\nunits; μIU, micro-international units.\nFigure  1 A shows\nthe stack plot of the 1D  1 H CPMG NMR spectra recorded on\nserum samples obtained from EM and NC subjects. The present study\ninvolves concentration profiling of proline and glutamine in the serum\nsamples of EM patients and NC subjects following a targeted  1 H NMR based metabolomics approach as described previously. 16 − 18 , 20  For this, first we have identified\nthe NMR signals corresponding to proline and glutamine and assigned\nby following a chemical shift and peak pattern matching procedure\nemploying the 800 MHz spectral database of metabolites in the NMR\nsuite of CHENOMX software; the selected spectral regions are shown\nin  Figure  1 B (for glutamine)\nand  Figure  1 C (for\nproline). To minimize the experimental/methodological deviations,\nthe proline and glutamine concentrations were explicitly measured\nusing formate as an internal reference (explicit details are described\nin the  Materials and Method ).\n(A) Stack plot of cumulative\n1D  1 H CPMG NMR spectra\nrecorded on serum samples obtained from 39 endometriosis (EM, blue)\npatients and 48 normal control (NC, red) subjects. The NMR signals\ncorresponding to proline and glutamine were identified and assigned\nby matching their specific chemical shifts and peak patterns employing\nthe NMR suite of CHENOMX software. (B and C) Selected regions from\nthe 1D  1 H CPMG NMR spectra of serum samples displaying\nthe Hδ peaks of (B) glutamine and (C) proline selected in this\nstudy for their targeted concentration profiling. Blue lines in parts\nB and C represent the fitting lines showing pattern matching with\nthe experimental peak pattern (black).\nThe estimated circulatory concentrations of proline and glutamine\nwere then used to estimate the proline to glutamine ratio (PQR; P\nand Q represent proline and glutamine, respectively).  Figure  2 A–C shows the box plots\ncomparing the circulatory concentrations of proline and glutamine\nand the PQR between the EM and NC groups. As evident, the circulatory\nproline level and the PQR were significantly increased in EM patients\n(with their mean values Pro = 38.99 ± 2.86 μM and PQR =\n1.24 ± 0.14 μM) compared to NC subjects (Pro = 23.01 ±\n1.36 μM and PQR = 0.53 ± 0.05) with  p -value\n< 0.0001, respectively ( Figure  2 A and C). On the other hand, the circulating glutamine\nlevels were decreased in EM patients (Gln = 37.72 ± 2.77 μM)\ncompared to NC subjects (Gln = 51.83 ± 4.19 μM) with nearly\nsignificant  p -value equal to 0.056 ( Figure  2 B). The decreased circulatory\nglutamine levels might be related to active glutaminolysis in EM patients.\nIt is important to mention here that both the proline and glutamine\nlevels have been estimated with respect to formate (as an internal\nreference), and possibly, this may have impacted the quantitative\nprofiles of these metabolites, rendering a nearly significant change\nin the glutamine levels between the EM and NC groups ( p -value < 0.056,  Figure  2 B). In order to check this possibility, we additionally performed\na small exercise if formate levels change in EM with respect to NC\nsubjects. For this, the reference concentration of TSP was calibrated\nto 100 μM (in the software program CHENOMX) and the concentration\nlevels of formate, glutamine, and proline were estimated and compared\nbetween NC subjects and EM patients (the results are presented in  Figure S2 ). Compared to NC subjects, the serum\nproline levels were significantly increased in EM patients (66.88\n± 4.07 μM in EM vs 37.88 ± 1.86 μM in NC with  p -value < 0.0001), whereas the changes in the serum levels\nof glutamine (decreased in EM) and formate (increased in NC) were\nfound to be insignificant between EM and NC patients (see the values\nin  Figure S2 ). The insignificant changes\nin formate levels are also evident from  Figure S3  comparing the cumulative NMR signal of formate between EM\nand NC subjects with respect to TSP.\n(A–C) Box-cum-whisker plot comparing\nthe circulatory proline\nand glutamine levels and the proline to glutamine ratio (PQR) between\nEM and NC subjects. (D–F) Plots representing the receiver operating\ncharacteristic (ROC) curve analysis performed for evaluating the diagnostic\npotential of these metabolic features in differentiating EM patients\nfrom NC subjects. The AUROC values (area under ROC curve) with standard\nerror and 95% confidence interval (CI) are shown in blue for each\nROC plot.\nTo be mentioned here is that the\nformate levels have been reported\nto be decreased in endometrial tissue, 33  and if we consider this decrease in EM patients here as well, the\nestimated glutamine level (technically the glutamine to formate ratio)\nmay not change between the study groups. Indeed metabolomics studies\nhave shown that the circulatory formate level decreases in lung cancer\nand breast cancer patients relative to healthy controls. 34  Contrary to glutamine, the circulatory proline\nlevel (technically proline to formate ratio) and PQR were found to\nbe elevated in EM patients, which might be related to disturbed proline\nmetabolism in EM patients.\nThese metabolic features were then\ntested for their diagnostic\npotential using ROC curve analysis, and the results are shown in  Figure  2 D–F. Of the\nthree, the circulatory PQR demonstrated the highest diagnostic potential\nwith area under ROC (AUROC) curve values equal to 0.88 ± 0.04\n[95% CI = 0.81–0.95] ( Figure  2 F). The AUROC values for proline and glutamine were\n0.82 ± 0.04 [95% CI = 0.74–0.91] and 0.62 ± 0.06\n[95% CI = 0.50–0.74], respectively ( Figure  2 D and E). These observations were found in\nwell concordance with various metabolic studies on endometriosis patients,\nsuggesting the potential of these metabolites, particularly of their\nPQRs, to serve as noninvasive diagnostic and prognostic markers. 35 − 38\nFurther, we compared the circulatory proline and glutamine\nlevels\nand PQR between moderate EM (MEM) and severe EM (SEM) patients w.r.t.\nthe NC subjects ( Figure  3 ). The circulating levels of proline were evidently and significantly\nincreased in MEM and SEM with their mean values equal to 32.75 ±\n2.58 μM and 42.89 ± 4.20 μM, respectively, compared\nto NC subjects (23.01 ± 1.36 μM) ( Figure  3 A). The  p -values for NC\nvs MEM and for NC vs SEM comparisons were found to be <0.05 and\n<0.0001, respectively ( Figure  3 A). Inversely, the circulatory levels of glutamine\nwere decreased in MEM and SEM patients with their mean values equal\nto 33.57 ± 3.93 μM and 40.74 ± 3.74 μM, respectively,\ncompared to NC subjects (51.83 ± 4.19 μM) ( Figure  3 B). The  p -value\nanalysis, however, revealed the statistically significant difference\nbetween NC and MEM, whereas significant differences were found neither\nbetween NC and SEM nor between MEM and SEM ( Figure  3 B). Finally, the circulatory PQR levels were\ncompared and showed a progressive and significant increase from NC\nto MEM (with  p -value < 0.05) and from MEM to SEM\n(with  p -value < 0.0001) with their mean values\nequal to 0.52 ± 0.05 μM for NC, 0.99 ± 0.13 μM\nfor MEM, and 1.39 ± 0.22 μM for SEM ( Figure  3 C). The ROC curve analysis was further used\nto evaluate the diagnostic potential of these metabolic features,\nand the results are summarized in  Figure  2 D–I. It is clearly evident that the\ncirculatory PQR levels exhibited the highest AUROC values equal to\n0.89 ± 0.04 [95% CI = 0.82–0.96] for comparison between\nNC and SEM ( Figure  3 F) and 0.87 ± 0.04 [95% CI = 0.79–0.96] for comparison\nbetween NC and MEM ( Figure  3 I). Following the PQR, the circulatory proline levels also\nshowed strong diagnostic potential AUROC values equal to 0.86 ±\n0.04 [95% CI = 0.77–0.94] for comparison between NC and SEM\n( Figure  3 G) and 0.81\n± 0.06 [95% CI = 0.69–0.93] for comparison between NC\nand MEM ( Figure  3 I).\nThese findings were further cross-validated using the NMR spectral\ndata set of a previously reported plasma metabolomics study. 39  The results summarized in  Figure  4  reveal that the circulatory PQRs are significantly\nelevated in EM (0.38 ± 0.02) compared to NC (0.29 ± 0.02)\nwith a statistical  p -value equal to 0.0006 ( Figure  4 C). The circulatory\nPQRs were also found significantly elevated (with  p -value < 0.05) in stage IV endometriosis patients (E4, 0.46 ±\n0.21) compared to stage III endometriosis patients (E3, 0.32 ±\n0.07) and NC (0.29 ± 0.10) ( Figure  4 C). However, the circulatory proline and\nglutamine levels failed to show a statistically significant difference\nbetween different study groups ( Figure  4 A and B).\nBox plots showing comparison of the circulatory\nlevels of (A) proline\nand (B) glutamine and (C) the proline to glutamine ratio in moderate\nendometriosis (MEM) and severe endometriosis (SEM) patients compared\nto normal control (NC) subjects. The symbols * and **** represent\nthe  p -values <0.05 and <0.0001, respectively,\nderived from the unpaired statistical  t  test for\neach metabolic comparison. The plots in parts D–F and G–I\nrepresent the receiver operating characteristic (ROC) curve analysis\nperformed for evaluating the diagnostic potential in moderate endometriosis\nand severe endometriosis patients compared to NC subjects, respectively.\nThe AUROC values (area under ROC curve) with standard error and 95%\nconfidence interval (CI) are shown in blue for each ROC plot.\n(A–C) Box plots showing a comparison of the circulatory\n(A) proline and (B) glutamine levels and (C) proline to glutamine\nratio (PQR) between endometriosis (EM) patients and normal control\n(NC) subjects. (D) Mean values (with standard error of mean (SEM))\nof these circulatory parameters (i.e., proline, glutamine, and PQR)\nin EM patients and NC subjects shown in tabulated form. (E–G)\nBox plots showing a comparison of the circulatory (E) proline and\n(F) glutamine levels and (G) proline to glutamine (PQR) ratio between\nstage II, III, and IV endometriosis patients (E2, E3, and E4, respectively)\nwith respect to normal control (NC) subjects. (H) Mean values of these\ncirculatory parameters (i.e., proline, glutamine, and PQR) in NC subjects\nand E2, E3, and E4 patients shown in tabulated form. For each box\nplot, the boxes denote interquartile ranges, the horizontal line inside\nthe box denotes the median, and the bottom and top boundaries of the\nboxes are the 25th and 75th percentiles, respectively. Lower and upper\nwhiskers are 5th and 95th percentiles, respectively. Note: the NMR\nspectral data used for cross-validation purpose corresponds to that\nused in the previous plasma based clinical metabolomics study by Vicente-Muñoz\net al. 39\n\nA noninvasive test for endometriosis (EM) would\nbe useful for nearly\nall women with minimal to mild stage EM and/or subfertility with normal\nultrasound. 40  This would also benefit women\nwith moderate to severe stage EM without clearly visible ovarian endometrioma. 40  Metabolomics analysis of blood plasma/serum\ncan provide noninvasive biomarkers for diagnostic and therapeutic\nmonitoring and also an emerging approach for gaining better understanding\nof the pathophysiology of this disease. 35 , 41  Considering\nthe relevance of this omics approach, several metabolomics studies\nhave been carried out in the recent past showing alterations in the\nblood plasma/serum levels of various amino acids in EM patients (with different disease stages) compared\nto healthy controls. 33 , 35 , 37 , 42\nRecent studies suggested that epigenetic\nmodifications (such as\nDNA methylation, histone modifications, and noncoding microRNAs (miRNAs))\nare involved in the malignant transformation of EM in ovarian tumorigenesis. 43  The microRNA miR-23b* was reported to be overexpressed\nduring tumor progression and markedly suppress the expression of POX\nenzyme. 31  Besides its role as tumor suppressor,\nPOX enzyme has an important role in the proline cycle where it converts\nproline into delta1-pyrroline-5-carboxylate (P5C), an intermediate\nwhich spontaneously changes into glutamic-gamma-semialdehyde (GSA). 31  This GSA produces glutamate with the help of\nP5C dehydrogenase enzyme and enters in the TCA cycle for energy production.\nHowever, due to decreased expression of POX in EM, this pathway remains\nsuppressed, causing increased availability of circulatory proline\nin the endometrial tissue of EM patients. In this targeted  1 H NMR based metabolomics study, we performed concentration profiling\nof proline and glutamine in serum samples of normal healthy women\nand women with different stages of endometriosis (moderate, stages\nI and II; severe, stages III and IV). The comparison (based on univariate\nstatistical analysis) revealed that the circulatory proline levels\nwere significantly elevated in EM patients whereas the circulatory\nglutamine levels were found to be decreased. The reduced glutamine\nconcentration in the serum of patients with endometriosis reflected\nactivated glutaminolysis in EM to fulfill the higher energy requirements\nof endometriotic cells which share similar proliferative capacity\nto cancer cells. 21 , 22  The increased proline 38  and reduced glutamine 35  levels were well supported with previous serum/plasma metabolomics\ncharacteristics of EM patients, suggesting an amplified utilization\nof glutamine  via  glutaminolysis and suppressed proline\ncycle probably due to increased expression of an mRNA, i.e. MiR-23b. 31\nFigure  5 A represents\ninterconnected glutaminolysis and the proline cycle within the cells\nof endometrial tissue, and it is shown here that glutamine is converted\ninto glutamate, which is finally utilized in energy production  via  the TCA cycle in cells with endometriosis. 22  Briefly, glutamine (Gln) is transported into\nthe cell and converted into glutamate, which fulfills the high energy\ndemands of proliferating tumor (malignant) cells via the TCA cycle.\nAdditionally, the glutamate participates into the proline cycle, where\nit is reduced into glutamic-gamma-semialdehyde (GSA) by P5C synthase\nand spontaneously converted into delta-pyrroline-5-carboxylate (P5C),\nwhich is further reduced by P5C reductase (also known as PYCR) into\nproline ( Figure  5 A).\n(A) Schematic\nshowing interconnected proline (Pro) and glutamine\n(Gln) metabolism in a cell with severe endometriosis; both are linked\nthrough an intermediate glutamic-γ-semialdehyde (GSA) as depicted.\nMessenger RNA 23b (MiR-23b) is reported to be overexpressed during\ntumor progression, which markedly suppresses the expression of POX\nenzyme. (B) Pearson  r  correlation analysis performed\nbetween circulatory proline and glutamine levels measured in endometriosis\n(EM) patients and normal control (NC) subjects (including the cross-validation\ncohort).\nConsidering that glutamine and\nproline are biosynthetically linked,\ntheir circulatory levels are expected to correlate positively if the\nmetabolic pathway is operating within the cells of endometrial tissue.\nThe statistical correlation plot between the circulatory glutamine\nand proline levels in EM patients was generated using the Pearson  r  method and is shown in  Figure  5 B. As evident from the figure, there is a\nstatistically significant and positive correlation of proline with\nglutamine; therefore, the conjecture that proline and glutamine are\nbiosynthetically linked is well supported. 24  The Pearson  r  correlation value was found to be\ndecreased in EM patients ( r  = 0.41 (95% CI: 0.10\nto 0.64),  p -value = 0.0105) compared to NC subjects\n( r  = 0.52 (95% CI: 0.26 to 0.69),  p -value = 0.0002), suggesting disturbed utilization of proline or\nglutamine in EM. For the validation cohort as well, the Pearson  r  correlation value was found to be decreased in EM patients\n( r  = 0.63 (95% CI: 0.42 to 0.77),  p -value <0.0001) compared to NC subjects ( r  =\n0.78 (95% CI: 0.54 to 0.90),  p -value < 0.0001)\n( Figure  5 A).\nThe nutritional and hormonal regulation of amino acid homeostasis\nis a well established physiological phenomenon and is achieved through\nexchange of essential amino acids with nonessential amino acids and\nthe transfer of amino groups from oxidized amino acids to amino acid\nbiosynthesis. 28 , 44 , 45  Both proline and glutamine are nonessential amino acids and interconvertible\nmetabolically. 24  Various studies in the\nrecent past have shown that increased proline biosynthesis plays an\nimportant role in cancer cell metabolic reprogramming, the occurrence/development\nof cancer, and its further progression toward the malignant state. 23 , 25  Of the two, glutamine (the most abundant amino acid in the circulatory\nsystem) is known to serve as an anaplerotic substrate to replenish\ntricarboxylic acid (TCA) cycle intermediates during growth of cancer\ncells and tumor progression. 27 , 46  The glutamate produced\nfrom glutamine is further utilized in part for proline biosynthesis\nto support anaplerosis, ATP production, protein and nucleotide synthesis,\nand redox homeostasis in cancer cells. 24  As proline is a major and essential constituent in collagen, the\nelevated PQR levels in EM might be related to increased proline biosynthesis\nand its further utilization in collagen synthesis as support. 30  There are some previous studies suggesting that\nthe collagen concentration is significantly higher in chronic endometriosis\nassociated fibrosis and that it may have a major role in endometriosis. 47  On the other hand, the cellular processes mediating\ntumor progression in EM are utilizing glutamine at a higher rate,\nrendering its decreased circulatory levels in the sera of EM patients\n( Figures  2 B and  3 B). The result of this metabolic derangement is\nthe elevated proline to glutamine ratio (PQR) which has been demonstrated\nin this paper. The results are well in line with previous clinical/preclinical\nstudies suggesting mitochondrial dysfunction and decreased energy\nproduction in uterine endometriosis tissue, 48  as proline catabolism in mitochondria serves as an important source\nof energy production and previous transcriptomics studies have demonstrated\nthat there is reduced expression of proline oxidase (POX, a mitochondrial\ninner-membrane flavoenzyme involved in the catabolic degradation of\nthe proline) due to overexpression of microRNA (known as MiR-23b). 31 , 49  Epigenetics studies on clinical samples have demonstrated that miRNA-23b\nexhibits regulatory roles, especially in the development of cancer, 50 , 51  and its overexpression is negatively correlated with the expression\nof tumor suppressor gene TUSC7 in endometrial carcinoma. 52  Studies have shown that miRNA-23b directly binds\nthe POX mRNA 3′-untranslated region; thus, the overexpression\nof miRNA-23b is directly correlated with downregulation of mitochondrial\nproline oxidase (POX), as schematically depicted in ( Figure  5 A). 53  Alternatively, the elevated circulatory proline levels can be because\nof increased expression of PYCR (i.e., P5C reductase), as proposed\nin some of the previous studies. 25\nIn order to find clinical correlates of circulatory PQR, we performed\nstatistical correlation analysis (employing the Spearman method) for\nPQR with various clinical parameters estimated for EM patients. The\nresults are summarized in  Table  2  and show that circulatory PQR levels do not correlate\nsignificantly with most of the clinical parameters; rather, the PQR\nshows significant correlations with proline ( r  =\n0.43, 95% CI: 0.12 to 0.66,  P  = 0.0069) and glutamine\n( r  = −0.53, 95% CI: −0.73 to −0.25,  P  = 0.0005). In an example clinical study, the diagnostic\npotential of Zn-alpha2-glycoprotein has been compared with the other\nantigenic glycoproteins CA125 and CA19.9. The sensitivities have been\nreported to be 69.4%, 33%, and 13%, respectively. 54  Compared to all these, the cumulative sensitivity and specificity\nof circulatory PQR (at the cutoff value 0.74) have been found to be\n74.36% and 85.42% ( Figure  2 F), suggesting this metabolic ratio has apposite potential\nto serve as a surrogate marker to improve the clinical diagnosis of\nEM.\nAbbreviations\nused: mg, milligram;\nμg, microgram; ng, nanogram; mL, milliliter; dL, deciliter;\nμIU, micro-international units.\n\nEM is a chronic, hormone-dependent gynecologic\ndisease which, though\nconsidered benign, is associated with an increased risk of malignant\ntransformation and involves various mechanisms of disease progression\nand development. 33 , 42 , 43 , 55  Putative biomarkers such as antigenic glycoproteins\n(e.g., Zn-alpha2-glycoprotein, CA125, and CA19.9), growth or adhesion\nfactors, hormones, or proteins related to immunology or angiogenesis\nhave failed to successfully diagnose the disease. 40  So far, there is no single biomarker or panel of biomarkers\nin the blood that has been validated as a clinical test for the diagnosis\nof endometriosis or definition of its stages. 40  In this targeted metabolomics study, an attempt has been made to\nvalidate the previously reported hallmark of endometriosis; that is,\nthe microRNA named miR-23b is overexpressed during EM tumor progression, 49  which markedly suppresses the expression of\nmitochondrial POX enzyme (a novel tumor suppressor), rendering decreased\nutilization of proline in the generation of reactive oxygen species\n(ROS, critical for regulation of cell growth and apoptosis). 53  The availability of proline, on the other hand,\ninduces collagen synthesis—a process essential for tumor growth\nand progression. 26  The present study will\nserve as a proof of the principal demonstrating that epigenetic/transcriptomic\nchanges altering the metabolic profiles can be evaluated or assessed\nusing metabolomics approaches. The elevated PQRs in EM have also been\ncross-validated on NMR spectral data recorded on another cohort of\nEM patients. 39  Overall, the present study\nwill serve as a basis for future studies aiming to develop diagnostic/prognostic\ntests based on circulatory PQRs. Further, the reprogramming of circulatory\nPQRs will serve as an indicator for the therapeutic efficacy against\nEM.\n\nAll patients ( N  = 39) were recruited from the outpatient\ndepartment of\nthe Sir Sunderlal Hospital, Department of Obstetrics and Gynecology,\nInstitute of Medical Sciences, Banaras Hindu University (BHU), Varanasi,\nIndia, during March 2017 to March 2020, and belonging to the province\nof Eastern Uttar Pradesh, India. Participants were included after\nfilling out the informed consent, proforma, and questionnaire. Demographic\nparameters, clinical symptoms, and physical examination findings of\npatients suffering with endometriosis were recorded. The ethical approval\nwas granted by the Institutional Ethical Committee (ref No: I. Sc./ECM-IX/2016-17/04).\nFemale participants ( N  = 48) included in the control\ngroup were having healthy medical examinations, normal reproductive\ncycles, and ≥2 pregnancies without history of pregnancy-related\ncomplications. And in the case group, only those who have been diagnosed\nfor endometriosis by ultrasonography, had no endocrinal radiation\nor chemical therapy or had not taken oral contraceptives 3 months\nprior to their admission, or not having any other disease history\nwere included.\nFurther, diagnosis of endometriosis patients\nwas done via laparoscopic inspections of the pelvis, preferably with\nhistological biopsy confirmation, and stages were defined based on\nrASRM. From each subject (EM patient or normal control), approximately\n2.0 mL of blood sample was drawn from the medial cubital vein, and\nthe collected blood was kept at room temperature for 30–40\nmin for coagulation and then centrifuged at 3000 rpm, 15 min, at 4\n°C. Supernatant serum was isolated and stored at −80 °C\nin an ultradeep refrigerator freezer.\nBefore starting NMR experiments,\nthe stored serum samples were withdrawn from −80 °C and\nthawed at room temperature. Each NMR sample was prepared by mixing\n300 μL of sodium phosphate buffer (0.9% saline, buffer strength\n50 mM prepared in 100% D 2 O, pH 7.4) with serum samples\n(300 μL in each case) and centrifuged at 16,278 g  for 5 min. After that, 450 μL of each prepared sample was\ntransferred to a 5 mm NMR tube (Wilmad Glass, USA). A coaxial NMR\ntube containing 1.0 mM TSP (sodium salt of 3-trimethylsilyl-(2,2,3,3)-propionic\nacid- d 4 ) dissolved in D 2 O was\ninserted separately that served as an external reference (offering\na final apparent concentration approximately equal to 0.1 mM). Deuterium\noxide (D 2 O) and the sodium salt of trimethylsilylpropionic\nacid- d 4  (TSP) used for NMR spectroscopy\nwere purchased from Sigma–Aldrich (St. Louis, MO, USA).\nNMR spectra of the prepared serum\nsamples, collected from healthy normal control and EM patients, were\nacquired using an 800 MHz Bruker Avance III NMR spectrometer equipped\nwith a TCI cryogenic probe. A one dimensional Bruker standard CPMG\n(Carr–Purcell–Maiboom–Gill) spin–echo\npulse sequence with water presaturation and a T2 filter for suppressing\nbroad signals of protein and other macromolecules was used to record\nthe spectra at 300k. 56 , 57  The total time of the T2 filter\n(i.e., [τ–180°−τ] n ) used\nwas ∼80 ms with the spin echo time (2τ) used equal to\n600 μs, 180° RF pulse equal to 25 μs, and loop counter\n(n) equal to 128. The other acquisition parameters used were as follows:\n128 transients with 64k data points, relaxation delay of 5 s, and\nspectral width of 20 ppm with an acquisition time per scan of 15 min.\nEach spectrum was then manually phased and baseline corrected using\nTopSpin 3.6.1. Afterward, the spectrum was opened in the PROCESSOR\nmodule of CHENOMX NMR Suite 8.6 software and further better corrected\nfor baseline and calibrated with respect to the formate peak at 8.43\nppm. For concentration profiling, formate was also used as an internal\nreference and the concentration was set to 10 μM (i.e., nearly\nclose to the detection limit of metabolites in the CPMG NMR spectra\nof serum samples recorded at 800 MHz NMR spectrometer). 58 , 59  The advantage of selecting formate as an internal reference has\nalready been demonstrated in previous methodological studies 60 , 61  including recent metabolomics studies from our lab. 16 − 18 , 20  Studies have shown that, unlike\nTSP, formate does not interact with serum proteins/macromolecules 60 − 62  and, therefore, has legitimate potential to serve as an internal\nreference for quantitative profiling of metabolites from blood serum\n(and eventually other biological fluids) in normal and diseased conditions\nnot involving disorders of endogenous formate metabolism. After data\nprocessing, the spectrum was imported to the PROFILER-Module of CHENOMX\nand the concentrations of selected metabolites (i.e., proline and\nglutamine) were estimated in all the serum samples of EM patients\nand NC subjects. It is to be mentioned here that the present study\nis the very first part of the ongoing clinical metabolomics study\non EM patients; the NMR spectra recorded using TSP as an external\nreference signal will be used for calibration in future studies.\nThe statistical analysis was performed\nusing the software program GraphPad Prism v6.01. The circulatory metabolic\nconcentrations between the study groups were compared using the unpaired\nstudent  t  test analysis method, and the change was\nconsidered statistically significant if the test  p -value was <0.05. The differences in the levels of significantly\naltered metabolites were visualized using box plots. The diagnostic\npotential of circulatory metabolites was evaluated using receiver\noperating characteristic (ROC) curve analysis and the area under the\nROC curve (AUROC), with value ≥0.85 considered as the criterion\nfor diagnostic significance. The correlations of circulatory metabolites\nwith various clinical parameters were evaluated based on the Spearman\ncorrelation coefficient ( r ). The categorical variables\nwere expressed as percentage and continuous variables as mean ±\nSD.","source_license":"CC0","license_restricted":false}