{"paper_id":"01fc240d-026f-4741-93b9-9f877fbd0e8d","body_text":"Although surgical techniques and therapies\nhave improved over the\npast decades, ovarian cancer (OC) remains the most lethal malignancy\namong gynecological cancers.  The high\ndeath rates due to OC arise from its propensity to misdiagnosis, as\nits symptoms are unspecific in the early stage (FIGO stage I and II)\nand lack accurate early diagnostic biomarkers. A majority of patients\nare diagnosed in late-stage (FIGO stages III and IV) of OC when cancer\nhas already metastasized and treatment options have greatly diminished.\nFurthermore, resistance to chemotherapy and the presence of malignant\nascites among epithelial OC patients have also been associated with\npoor prognosis and early death. \n , \n  In both developed and\nunderdeveloped countries, more than 75% of OC patients are regrettably\ndiagnosed in the third and fourth stages. \n , \n  Surgical\ndebulking followed by chemotherapy remains the current standard front-line\ntreatment for OC. \n , \n  A combination of platinum and\npaclitaxel is the standard first-line chemotherapy treatment approach\nusually administered in cycles of 3 weeks or carboplatin every 3 weeks\nand paclitaxel weekly, for six cycles. Unfortunately, patients may\nexperience intrinsic or acquired resistance to chemotherapy. The most\nprominent mechanisms of platinum resistance include reduced drug uptake\nand increased efflux, enhanced DNA adduct repair, and increased drug\ninactivation by glutathione. \n , \n  Despite the availability\nof new treatment approaches, such as immunotherapy, resistance to\nchemotherapy continues to significantly impact the survival of OC\npatients. Indeed, despite their success in treating several tumor\ntypes including melanoma and lung cancer, the novel therapies arresting\nimmune checkpoint pathways are not recommended for OC patients when\nused as monotherapies because the results of clinical studies have\nso far been disappointing. \n , \n  Cancer Antigen 125 (CA125),\na heavily glycosylated mucin, is the routine biomarker used for OC\ndetection, monitoring response to treatment and detecting relapse. \n , \n  CA125 is also elevated in some physiological states such as pregnancy\nand menses \n , \n  and pathological conditions such as liver\ndiseases, endometriosis, and benign ovarian diseases (BOD), \n , \n  resulting in low sensitivity and specificity. Additionally, CA125\nis not overexpressed in about 20% of the OCs  and does not correlate with the prediction of platinum sensitivity\nor resistance. \n ,\nGlycosylation, the most\nfrequent post-translational modification,\ndoes not only play a crucial role in the functionality and stability\nof proteins  but is also modulated upon\nmalignancy. \n − \n \n \n \n \n \n \n N -Glycan changes previously reported for serum\nglycoproteins in EOC patients include decreases in the abundance of\nhigh-mannose and increases of tri- and tetraantennary sialylated and\nfucosylated  N -glycans. \n , , \n  Previously, our research group profiled permethylated  N -glycans from serum samples of two European EOC cohorts\nby MALDI-TOF-MS and identified 11 altered  N -glycans\nthat were of statistical significance. \n , \n  As a result,\nwe combined the significantly expressed glycan areas into a ratio\nnamed the “GLYCOV” score, which was more performant\nto differentiate EOC from BOD and healthy subjects than CA125 was. \n , \n  In addition, we also reported  N -glycosylation changes\nin EOC patients on acute-phase proteins that included core-fucosylated\nbiantennary  N -glycans on α1-acid glycoprotein,\nincreased antennarity and Lewis X  motif on haptoglobin,\nα1-antitrypsin, and α1-antichymotrypsin as compared with\ncontrols.  Recently, we demonstrated an\nincrease of α2,6-sialylation in OC tumor regions and elevated\nα2,3-sialylation in nontumor regions as well as tumor stroma.  Similar findings of modulated  N -glycosylation features in EOC patients have also been reported by\nother studies from different research groups. \n , , \n  In addition, other glycosylation changes\nwere published for other forms of cancer, as well. For instance, decreased\nexpression of high-mannose and bigalactosylated biantennary  N- glycans was measured for gastric cancer patients, while\nthe nongalactosylated biantennary  N- glycans were\nincreased.  Saldova et al. reported a\nsignificant increase of core-fucosylated biantennary and α2–3-linked\nsialylated  N -glycans in prostate cancer patients\ncompared to the benign prostate hyperplasia subjects.  In pancreatic cancer, branching and antennary fucosylation\nwere increased whereas high-mannosylation was not of statistical relevance.  More importantly,  N -glycosylation\nchanges have been linked to major events in malignancy such as cancer\ngrowth, progression, and metastases. \n , , \n  Kyselova et al. observed elevated sialylated and\nfucosylated glycans in breast cancer, which showed consistency with\ndisease progression.\nOnly a few\nstudies have attempted to investigate the potential\nof  N -glycome changes upon EOC malignancy as markers\nfor monitoring the response to chemotherapy or addressing chemoresistance.\nZahradnikova et al. identified six tissue  N -glycans\nwith characteristic bisecting, tetraantennary structures bearing sialic\nacid and/or fucose residues, which could potentially be used as markers\nof resistance to chemotherapy in OC patients.  In another study, the serum of breast cancer patients changed in\nthe abundance of high-mannose, core fucose, and galactose following\nchemotherapy administration, suggesting a response to chemotherapy.  Recently, Zhao et al. observed different glycosylation\npatterns for Lewis-type biantennary, triantennary trisialylated, and\nLewis-type triantennary glycans that differ between OC chemotherapy\nresponders and nonresponders,  but the\nglycome was not monitored during chemotherapy.\nIn the present\nwork, we investigated the  N -glycome\nfrom African EOC patients, for which samples were collected at the\ntime of the first diagnosis. They were compared with age-matched BOD\nsamples. In addition, serum from African EOC patients was collected\nat different cycles of chemotherapy as cross-sectional samples. The  N -glycans released from these samples were compared with\nthe  N -glycans from primary African EOC subjects.\nTo the best of our knowledge, this is the first time that serum glycome\nhas been reported for a population of African ethnicity.\n\nAll of the chemicals were purchased\nfrom Sigma-Aldrich (St. Louis,\nMO, USA) unless stated otherwise.\nWe recruited an African\ncohort comprising 53 histologically confirmed EOC patients and 46\nBOD patients (Supporting Information,  Table S1 ). Recruitment took place in three Kenyan Hospitals within Nairobi\ncity, namely, Kenyatta National Hospital (KNH), Texas Cancer Centre,\nand St. Mary’s Hospital Lanǵata, as per the ethical\nvote obtained from Kenyatta National Hospital/University of Nairobi\nethics review committee (KNH/UON-ERC) reference no P701/12/2017. All\nthe samples were collected from adult women of 18 years or more after\ngiving their written informed consent. Of the 53 EOC patients, 19\nwere primary EOC patients (pretreatment group), while 34 were chemotherapy\nresponders who had already undergone various sessions of chemotherapy\n(chemo) cycles that comprised carboplatin and paclitaxel (Supporting\nInformation  Figure S1A ). Serum samples\nwere collected into 5 mL vacutainers with a serum clot activator (Becton,\nDickinson GmbH, Heidelberg, Germany). Samples were allowed to stand\nat room temperature between 30 and 120 min before centrifugation at\n1,200 g  for 15 min. The separated serum aliquots were\nthen stored in Eppendorf tubes at −80 °C until their shipment\nto Berlin, Germany. The necessary shipment approvals were obtained\nfrom the respective agencies in Kenya: KNH/UoN-ERC ref no., KNH-ERC/shipment/40,\nMinistry of Health Kenya, ref. No. MOH/F/HRD/01/VOL.11 and the Kenya\nPharmacy and Poisons board export permit ref. no. CD2021000PPB321J0002550623.\nIn Germany, the Ethical Commission of the Charité-Universitätsmedizin\nBerlin approved the analysis of the samples (approval number EA4/071/19).\nMeasurements of CA125 were done on a\nCobas e 801 immunoassay system (Roche Diagnostics GmbH, Penzberg,\nGermany), a high-throughput fully automated immunochemistry module\ndesigned to carry out electrochemiluminescence sandwich immunoassays.\nThe reagent used was an Elecsys CA125 II reagent (Roche diagnostics\nGmbH, Penzberg, Germany). The normal CA125 cutoff value was set at\n35 kU/L for pre- and postmenopausal women.\nN -Glycans were released from serum samples and\npurified as described previously (Supporting Information  Figure S1B ). \n , , \n  Briefly, 10 μL of serum was diluted in 2 μL\nof 200 mM phosphate buffer (pH 6.5). Serum glycoproteins were then\nreduced by adding 2.5 μL of 200 mM dithioerythritol (DTE) and\nincubated at 60 °C on a shaker for 45 min. Afterward, 10 μL\nof iodoacetamide was added and serum samples were left to alkylate\nfor 1 h at room temperature in darkness. Subsequently, the reaction\nwas stopped by the addition of the excess DTE.  N -Glycans\nwere then enzymatically released from glycoproteins using PNGase F\n(200 mU,  N -Zyme Scientifics, Doylestown, PA, USA)\nfor 16 h at 37 °C. The following day,  N -glycans\nwere isolated and desalted using C18 cartridges and porous graphitized\ncarbon columns, respectively (both purchased from Alltech, Deerfield,\nIL, USA). The eluates were collected in Eppendorf tubes and then dried\nunder a reduced atmosphere by centrifugal evaporation.  N -Glycans were finally permethylated to neutralize sialic acids and\nto improve  N -glycan ionization during MALDI-TOF measurements.\nPermethylated  N -glycans were dissolved in 10 μL of 75% aqueous acetonitrile.\nEqual volumes (0.5 μL) of  N -glycans and the\nsuper 2,5- dihydroxybenzoic acid matrix were spotted in triplicate\non the ground steel target (Bruker Daltonics, Bremen, Germany). A\nglucose ladder was used for calibration, and measurements were made\non an Ultraflex III (Bruker Daltonics, Bremen, Germany) in reflector\npositive ionization mode in the mass range of 1000–5000 Da.\nFor each spectrum generated, at least 4000 laser shorts were recorded;\nbaseline correction and peak picking were carried out using flexAnalysis\n(Bruker Daltonics, Bremen, Germany).  N- Glycan spectra\nwere analyzed using GlycoPeakfinder, and  N -glycan\ncartoons were generated with GlycoWorkbench. \n ,\nN -Glycans were labeled with\n2-aminobenzamide (2-AB), as described previously with small modifications.  A 2AB-glucose ladder, which was prepared in\nparallel, was used as an external standard during HPLC measurements.\nIn short,  N -glycans, released from 5 μL serum,\nwere labeled overnight at 37 °C using a solution containing 1\nM 2AB and 0.24 M 2-picolinborane in acetic acid/methanol (1/9, v/v).\nThe following day, samples were adjusted to 80% acetonitrile and applied\nto self-made cotton microcolumns, preconditioned with 3 × 40\nμL of milli-Q water and then 3 × 40 μL of 80% ACN\ncontaining 0.1% TFA. After applying samples, microcolumns were washed\n3 times with 40 μL of 80% ACN containing 0.1% TFA in order to\nremove the unreacted label. The derivatized  N -glycans\nwere then eluted with 6 × 40 μL of Milli-Q water. Samples\nwere finally dried by centrifugal evaporation and reconstituted in\n30 μL of water, from which 8 μL was taken for HPLC measurements.\nSeparations were performed using an Ultimate 3000 (Dionex, Germany)\nequipped with a RS fluorescence detector (Dionex, Germany) with excitation\nand emission wavelengths of 330 and 420 nm, respectively.  N- Glycan profiling was achieved at a flow rate of 0.75 mL/min\nwith a XBridge Premier Glycan BEH Amide column (2.5 μm; 4.6\nmm × 100 mm; Waters; USA) and a temperature of 20 °C. The\nlinear gradient consisting of 100 mM ammonium formate (pH 4.5) (solvent\nA) and 100% ACN (solvent B) was set as follows: samples were injected\nin 22% A, and then the proportion of A was increased to 64% in 45\nmin. After maintaining 64% A for 10 min, the initial conditions were\nrestored and maintained for 8 min.  N -Glycans were\nassigned as previously  using the GlycoBase\ndatabase.\nData analysis was performed using\nSPSS version 28 (SPSS Inc., Chicago, IL, USA). Mann–Whitney  U  test was used to compare the expression patterns of the\ndetected  N -glycans between the different patient\ngroups, which were then presented as medians, range, and p-values.\nThe  N -glycan index GLYCOV previously established\nwas calculated as follows: (sum of relative areas of  m / z  3776.9, 3951.0, 4226.1, 4400.2, 4587.3, 4761.4,\nand 4935.5)/7*4/(sum of relative areas of  m / z  1579.8, 1783.9, 1988.0, and 2192.3).  Receiver Operating Curves (ROC) were built for the  N -glycans that were of statistical significance. The corresponding\nvalues of the area under the curve (AUC, 95% C.I) were used to describe\nthe accuracy levels of assessing EOC diagnosis and of monitoring response\nto chemotherapy. AUC values of ROC curve >0.9 indicated a “high\naccuracy” outcome, values of 0.8 to 0.9 meant “good\naccuracy”, 0.7 to 0.8 indicated moderate accuracy, while values\n0.5 and 0.7 were interpreted as “uninformative”. Box\nplots were generated to describe the distribution of the  N -glycan index and CA125 values in the patients’ three clusters\nof chemotherapy intake.\n\nWe present the first evaluation of the  N -glycome\nprofile in an African EOC cohort, testing its clinical utility in\nmonitoring the chemotherapy response. We also sought to validate our\nprevious findings of early EOC diagnostic  N -glycan\nsignatures from Caucasian cohorts in African patients. \n ,\nIntra- and interday reproducibility\nwere verified by analyzing the same serum sample in triplicate on\na single day and on three consecutive days (Supporting Information  Figure S2 ). Four low-abundance  N -glycan signals corresponding to high-mannose  N -glycans\n( m / z  1579.8, 1783.9, 1988.0, and\n2192.1) were selected for this evaluation because they had previously\nbeen described by our research group as biomarker signatures of EOC\namong Caucasians. The mean coefficients of variation are below 10%,\nindicating the good reproducibility of our analytical workflow.\nWe cleaved, isolated, and analyzed  N -glycans from an African cohort consisting of 53 EOC patients\nand 46 BOD subjects by MALDI-TOF-MS. A total of 46 signals were detected\nand assigned to permethylated  N -glycans (Supporting\nInformation  Table S2 ).  Figure  \n  shows a representative MALDI-TOF\nspectrum of  N -glycans from a control subject against\nan EOC patient of African ethnicity. A nonparametric Mann–Whitney  U  test was applied to the relative intensities of the  N -glycan peaks to describe the differences in their expression\nbetween primary EOC patients and BOD subjects.\nRepresentative MALDI-TOF\nmass spectra of the permethylated  N -glycans from\n(A) a BOD subject and (B) a primary EOC patient.\nMeasurements were carried out in positive-ion mode and molecular ions\nare present in their [M + Na] +  form. The monosaccharides\nare depicted as follows: Man, green circle; Gal, yellow circle; GlcNAc,\nblue square; Fuc, pink triangle; Neu5Ac, pink diamond.\nA total of 39  N -glycans were differentially\nexpressed\nin primary EOC patients as compared to their control counterparts\n( p  < 0.05) (Supporting Information  Table S2 ). In the mass range  m / z  < 3600, 27 signals, corresponding mainly to\nhigh-mannose, monoantennary, and biantennary  N -glycans,\nhad decreased intensities in EOC patients. On the opposite, in the\nmass range  m / z  > 3600, 12 signals,\nassigned to complex-type tri- or tetraantennary  N -glycans being sialylated and fucosylated, had increased intensities\nin EOC patients as compared to control subjects (Supporting Information  Table S2 ). Interestingly, the 11  N -glycans constituting the glycan index GLYCOV \n , \n  that was previously developed by our research group using Caucasian\ncohorts are part of the statistically significant  N -glycans in the present African cohort. The 11  N -glycans constituting the glycan index GLYCOV included four high-mannose  N -glycans ( m / z  1579.8,\n1783.9, 1988.0, and 2192.1), which were decreased in primary EOC,\nand seven increased complex-type  N -glycans ( m / z  3776.9, 3951.0, 4226.1, 4400.2, 4587.3,\n4761.4, and 4935.5). We set a cutoff value at <1.82 to denote a\nnormal  N -glycan index value, as obtained from the\ncoordinates of the ROC curve. From the foregoing, the  N -glycan index was highly accurate in discriminating EOC patients\nfrom the control subjects (AUC 0.94, [C.I 0.880–1.00], SP,\n98%) compared to CA125 (AUC 0.88, [C.I 0.801–0.961, SP, 67%])\n(Supporting Information  Figure S3 ). No\nadditional discriminatory advantage was observed when the  N -glycan index was used in combination with CA125 (AUC 0.94,\nC.I [0.880–1.000]).\nThe total serum  N -glycome of EOC patients undergoing chemotherapy was then analyzed\nto determine its potential utility in monitoring patients’\nresponse to chemotherapy. To this end, EOC patients were subdivided\ninto three categories based on the number of chemotherapy cycles received\nat enrollment into the study: pretreatment, 1–3 chemo cycles,\nand 4–6 chemo cycles.  Figure  \n  shows a representative MALDI-TOF mass spectrum of\nthe  N -glycans from each patient category.\nRepresentative\nMALDI-TOF mass spectra of the permethylated  N -glycans\nfrom EOC patients: (A) pretreatment; (B) 1–3\nchemo cycles; and (C) 4–6 chemo cycles. The  N -glycans of  m / z  < 3600, which\ninclude the high-mannose-type, were increased after 1–3 and\n4–6 chemo cycles (B,C) as compared to the pretreatment group\n(A). Conversely, the  N -glycans of  m / z  > 3600, which include complex-type fucosylated\nand sialylated  N -glycans, were decreased with the\nincreasing number of chemo cycles (B,C). Measurements were performed\nin positive ionization mode and molecular ions are present in their\n[M + Na] +  form. Man, green circle; Gal, yellow circle;\nGlcNAc, blue square; Fuc, pink triangle; Neu5Ac, pink diamond.\nSample-independent Kruskal–Wallis test was\nused to compare\nmedians across the groups, and ROC curves were built for each of the\n27 statistically significant  N -glycans to determine\nthe accuracy of differentiating the patients in the three categories.\nSeventeen  N -glycans (of  m / z  < 3600) showed a statistically significant increase\n( p  < 0.05) and AUC> 0.7 as the number of chemo\ncycles increased (Supporting Information  Table S3 ). They comprised the  N -glycans of  m / z  1579.8, 1783.9, 1988.0, 1620.8, 1416.7,\n1982.0, 2390.2, 2285.2, 2315.2, 2489.3, 2850.4, 1661.8, 2070.0, 2519.3,\n2693.4, 2880.4, and 3054.5, boldly highlighted by their respective\nAUCs. The 17  N -glycans signatures (of  m / z  < 3600) demonstrated response to chemotherapy\nagents within the first three cycles of chemotherapy when compared\nto the primary EOC baseline group. The patients that had taken 4–6\nchemo cycles had the highest peak increases, affirming a positive\nresponse of patients to chemotherapy. On the other hand, nine complex-type  N -glycans (of  m / z  >\n3600)\nwhose intensities were elevated in pretreatment EOC patients, were\nsignificantly decreased after 1–3 and 4–6 chemo cycles\n( p -value <0.05 and an AUC value >0.70). They\ncomprised\nthe  N -glycans of  m / z  3415.7, 3776.9, 3864.9, 3951.0, 4226.1, 4400.2, 4587.3, 4761.4,\nand 4935.5 (Supporting Information  Table S4 ). It should be noted that these nine  N -glycans\ncould not differentiate between primary EOC patients and EOC patients\nwho received 1–3 chemo cycles. MALDI-TOF data were verified\nby 2AB-HPLC. To this end, released  N -glycans were\nlabeled with 2-AB and then measured by HPLC equipped with fluorescence\ndetection (Supporting Information  Figure S4 ). An increase of high-mannose  N -glycans and a decrease\nof sialylated  N -glycans were observed as the number\nof chemo cycles increased.\nAs the 11  N -glycans\nthat are part of the glycan\nindex GLYCOV previously established by our research group for EOC\nin Caucasian cohorts \n , \n  were of statistical relevance,\nGLYCOV was computed and evaluated for its performance in distinguishing\npatients across the three chemotherapy categories. The log2-transformed\nvalues of the  N -glycan index and CA125 for each patient\ncategory were plotted in a box plot, and the differences between them\nwere statistically tested ( Figure  \n ). The  N -glycan index showed statistically\nsignificant decreases between the three patient categories, the pretreatment\ngroup having the highest values and 4–6 chemo cycles having\nthe lowest values ( Figure  \n A). On the other hand, CA125 could differentiate pretreatment\nfrom 4–6 chemo cycles but not pretreatment from 1 to 3 chemo\ncycles ( Figure  \n B).\nBox plots\nof log2-transformed values of (A) GLYCOV, the  N -glycan\nindex and (B) CA125 analyzed by the Mann–Whitney  U  test (* p  < 0.05, ** p  < 0.01,\nand *** p  < 0.001).\nNext, we built ROC curves for both cancer biomarkers;\nthe  N -glycan index demonstrated improved accuracy\nin distinguishing\nEOC patients across the three treatment categories compared to CA125\n( Figure  \n ). Indeed,\nAUC values were 0.77 for the  N -glycan index versus\n0.54 for CA125 when pretreatment patients were compared to patients\nundergoing 1–3 chemo cycles ( Figure  \n A). When pretreatment was compared with 4–6\nchemo cycles, AUC of the  N -glycan index was as high\nas 0.89 versus 0.8 for CA125 ( Figure  \n B).\nROC curves comparing CA125 and the glycan index to assess\nresponse\nto chemotherapy: (A) pretreatment versus 1–3 chemo cycles,\nand (B) pretreatment versus 4–6 chemo cycles.\n\nIn the past years, various independent research\ngroups including\nour laboratory have studied extensively EOC-related changes of the  N -glycome, \n , , , , − \n \n \n  but mostly using Caucasian and Asian populations. The data on neoplastic\nglycosylation changes of EOC in African populations had not been documented\nso far, in contrast to the Caucasian and Asian ethnicities. Therefore,\nfor the first time, we provided in this work data on the EOC-mediated\nalterations of the total  N -glycome in an African\npopulation to identify signature biomarkers for primary diagnosis\nof EOC as well as for monitoring patients’ response to chemotherapy.\nThe subjects recruited into this study were largely stemming from\nurban and peri-urban settings and hence nearly of relatively similar\nsocial and economic demographics to that of the Caucasian cohorts\nused for the comparison analysis. The similarities in the variables\nof the two comparative groups reinforce the resultant findings thereof.\nInterestingly, African BOD subjects had  N -glycan\nprofiles that were very comparable with our previously reported European\ncontrol profiles. \n , \n  Surprisingly, although Kenya\nborders Ethiopia, control subjects from both countries seem to have\ndifferent  N -glycosylation patterns.  Gebrehiwot and co-workers reported increased intensities\nfor high-mannose, core-fucosylated, multiantennary, and multisialylated\nglycans in Ethiopian subjects when compared with US, Indian, and Japanese\nhealthy subjects, which we did not observe in this work.  It should be noted that the method used in that\narticle prior and posterior to  N -glycan release was\ndifferent from ours, preventing a direct comparison of the data. It\nis also worth noting that the total serum glycome, unlike the IgG\nglycome, is not significantly influenced by age.  Pongracz and co-workers recently showed that the  N -glycans that are of relevance in this present research\nwork, namely, sialylated tri- and tetraantennary and high-mannose  N -glycans, are not markedly associated with age.  In addition, the  N -glycome\nis temporally stable in single individuals.\nThe current analysis showed that  N -glycome\nmodulations resulting from the malignant EOC growth in individuals\nof African ethnicity were comparable to the ones previously found\nin European cohorts. \n , , , , − \n \n  We identified 39  N -glycan biomarker signatures\nthat successfully discriminated primary EOC patients from BOD subjects.\nThe reported  N -glycosylation changes include a decrease\nof high-mannosylation and an increase of antennarity, sialylation,\nand fucosylation. Using nano-HPLC-chip-TOF-MS, Hua and colleagues\npreviously identified 26 differentially expressed  N -glycans between the EOC patients and the healthy controls, highlighting\ndecreased abundances of high-mannose and hybrid  N -glycans in EOC while also reporting increases in peaks of complex-type  N -glycans.  Modulated expression\nof high-mannose is a key feature of many cancers, as corroborated\nby many studies including the present one. \n , , , , , , \n  In addition, elevated levels of free mannose were recently measured\nin the serum of EOC patients;  however,\nthe exact role played by high-mannosylation in tumor etiology is yet\nto be overtly elucidated. Alley et al. also reported increased abundances\nof tri- and tetraantennary complex-type  N -glycans\namong EOC patients, while bisecting  N -glycans were\nreduced.  Furthermore, increased sialylation\nwas also observed in the serum glycome from a mouse model of ovarian\ncancer.  A decrease in high-mannose structures\nwas similarly observed in gastric cancer patients,  while fucosylated, sialylated, and branched  N -glycans structures were increased in prostate, lung, pancreatic,\nand colorectal cancer. \n , − \n \n  On the contrary\nto gastric and breast cancer, \n , \n  an increase of agalactosylated\nbiantennary  N -glycans is not observed in ovarian\ncancer, demonstrating the cancer specificity of glycome changes. Indeed,\nthese glycan features have also been associated with the progression,\ntumor evasion, and metastasis of cancers including OC and with the\noverexpression of the corresponding glycosyltransferases, namely,  N -acetylglucosaminyltransferase IV and V, the sialyltransferases\nST3Gal I and ST6Gal I, and alpha-1,3-fucosyltransferase type VII. \n , \n  Moreover, glycosylation may also be influenced by nutrition, as\nalready shown  in vitro : adipose conditioning medium\nwas previously found to increase α2,6- and α 2,3-linked\nsialylation of ovarian cancer cells.  Sialic\nacid and fucose residues cap galactose residues, preventing the clearance\nof glycoproteins with the asialoglycoprotein receptor and allowing\na longer circulation of acute-phase proteins in the bloodstream. \n −\nThe findings on the use of our  N -glycan score,\nconsisting of 11  N -glycan signatures, for primary\ndiagnosing EOC in an African cohort, were concordant with our former\nanalysis on Caucasian cohorts. \n , \n  Compared with CA125,\nthe  N -glycan index demonstrated excellent diagnostic\naccuracy and specificity. The specificity improved even further to\n98% when the  N -glycan index was used in conjunction\nwith CA125. Glycosylation has previously been found to improve existing\nroutine biomarkers, such as prostate-specific antigen (PSA) and CA125.\nIndeed, increased PSA fucosylation, as measured by binding to the\nlectin Aleuria Aurantia Lectin, is able to refine the diagnosis of\nprostate cancer.  In addition, Llop and\nco-workers used PSA-sialylation to discriminate aggressive from nonaggressive\nprostate cancer.  CA125 glycoforms containing\nthe Sialyl-Thomsen-Friedenreich antigen were also used to predict\nthe progression-free survival and relapse of high-grade serous EOC\npatients.\nThe current\nanalysis identified potential 27  N -glycans for monitoring\nchemoresponse in patients that had undergone 1–3 and 4–6\nchemo cycles. The 27  N -glycans were originally differentially\nexpressed in primary EOC, and they included the 11  N -glycans that constitute our GLYCOV score. Of importance to note\nis that the 27  N -glycans showed the efficacy of chemotherapy,\nwhereby the high-mannose  N -glycans were statistically\nincreased within the period of administration of 1–3 chemo\ncycles whereas the complex-type, sialylated, and fucosylated  N -glycans only statistically decreased from the fourth to\nsixth chemo cycles. Interestingly, our glycan index was able to differentiate\nthe pretreatment group from the group of 1–3 chemo cycles,\nwhereas CA125 was not able to do so. Both biomarkers were able to\ndiscriminate between the pretreament group and the group of 4–6\nchemo cycles. As we used cross-sectional samples from responder patients\nto address treatment monitoring in this work, it is likely that false-negative\nsamples are underestimated. Future studies should include a larger\nnumber of patients and treatment monitoring of the same patients over\ntime to confirm the data. Although data on glycomic-based biomarkers\nfor monitoring chemotherapy response in malignant patients is limited,\nour data is in line previous research. \n , \n  Evidence from\na previous study by Miyahara et al. identified the high-mannose HexNAc 2 Hex 9  as a potential biomarker for the efficacy\nof gemcitabine monotherapy treatment in unresectable advanced pancreatic\ncancer patients as well as for predicting patient survival.  In another study, Zhao et al. collected and\nanalyzed serum from EOC patients to predict resistance to chemotherapy.  They identified the Lewis type HexNAc 4 Hex 5 dHex 1 Neu5Ac 2 , HexNAc 5 Hex 6 dHex 1 Neu5Ac 3  and the trisialylated\nHexNAc 5 Hex 6 dHex 1 Neu5Ac 3  as predictive markers for chemoresistance, α2–3 sialylation\nbeing increased whereas α2–6 sialylation was decreased.  At the tissue level, Zahradnikova et al. had\nassociated eight  N -glycans with resistance to chemotherapy;\nthey were complex-type mono- and biantennary  N -glycans.  Besides, similarly to the current analysis,\nSaldova et al. also found an increase of the otherwise decreased high-mannose\nHexNAc 2 Man 6  in breast cancer patients upon initiation\nof chemotherapy.\n\nOur glycan-based biomarker was more efficient\nthan CA125 to diagnose\nprimary EOC and to monitor the response to therapy in an African cohort.\nWhile larger multicenter studies are needed to confirm the present\ndata without concern for ethnic variations in glycosylation, the current\nfindings justify the use of a common  N -glycan biomarker\napproach for diagnosing EOC in a mixed ethnic population of African\nand Caucasian patients.","source_license":"CC-BY-4.0","license_restricted":false}