Identification of metabolic biomarkers for diagnosis of epithelial ovarian cancer using internal extraction electrospray ionization mass spectrometry (iEESI-MS)

Abstract

BACKGROUND:

Epithelial ovarian cancer (EOC) is the leading cause of death from gynecologic malignancies. The poor prognosis of EOC is mainly due to its asymptomatic early stage, lack of effective screening methods, and a late diagnosis in the advanced stages of the disease.

OBJECTIVE:

This study investigated metabolomic abnormalities in epithelial ovarian cancers.

METHODS:

Our study developed a novel strategy to rapidly identify the metabolic biomarkers in the plasma of the EOC patients using Internal Extraction Electrospray Ionization Mass Spectrometry (IEESI-MS) and Liquid Chromatography-mass Spectrometry (HPLC-MS), which could distinguish the differential metabolites in between plasma samples collected from 98 patients with epithelial ovarian cancer, including 78 cases with original (P), and 20 cases with self-configuration (ZP), as well as 60 healthy subjects, including 30 cases in the original sample (H), 30 cases in self-configuration (ZH), and 6 cases in a blind sample (B).

RESULTS:

Our study detected 880 metabolites based on criteria variable importance in projection (VIP) $>$ 1, among which 26 metabolites were selected for further identification. They are mainly metabolism-related lipids, amino acids, nucleic acids, and others. The metabolic pathways associated with the differential metabolites were explored by the KEGG analysis, a comprehensive database that integrates genome, chemistry, and system function information. The abnormal metabolites of EOC patients identified by IEESI-MS and HPLC-MS included Lysophosphatidylcholine (16:0) [Lyso PC (16:0)], L-Phenylalanine, L-Leucine, Phenylpyruvic acid, L-Tryptophan, and L-Histidine.

CONCLUSIONS:

Identifying the abnormal metabolites of EOC patients through metabolomics analyses could provide a new strategy to identify valuable potential biomarkers for the screening and early diagnosis of EOC.

Keywords

Epithelial ovarian cancer biomarkers IEESI-MS HPLC-MS metabolites

1. Introduction

Ovarian cancer remains the most common cause of death among women with gynecological malignancies worldwide [1]. According to the American Cancer Society assessment, there were about 21,000 new cases and 14,000 deaths of ovarian cancer in 2021 [2]. Epithelial ovarian cancer (EOC) accounts for more than 95% of ovarian malignancies. Histologically, EOCs are divided into five subtypes: high-grade serous, low-grade serous, clear cell, endometrioid, and mucinous. High-grade serous ovarian cancer is the most common histologic subtype, accounting for more than 70% of EOC [3, 4]. EOC is characterized by the following aspects. Approximately 75% of women with EOC are diagnosed in the advanced stages (IIIc and IV). The five-year survival rate of EOC patients with stage IIIC-IV is less than 30%, with death occurring within the first two years of diagnosis [5]. The standard treatment for EOC includes tumor reduction surgery and platinum-based chemotherapy [6]. Despite a successful initial response to initial treatment, approximately 80% of patients experienced disease relapse, with many receiving multiline therapy after relapse. Each subsequent first-line treatment is characterized by short disease-free intervals, which eventually leads to resistance and death [7]. The low survival rate of ovarian cancer patients is mainly due to the high percentage of late diagnoses in the advanced stages of the disease. Therefore, early diagnosis could be the key to improving the prognosis of ovarian cancer patients because they will gain benefits from screening to have a better clinical outcome. However, currently, there are no effective early screening methods. The most commonly used screening methods for EOC in the clinic are CA125 (cancer antigen 125), HE4 (human epididymis protein 4), and the TVS (transvaginal ultrasonography single or combined inspection). Although positive results could be found using those screening methods in some small studies [8], however, the UK Collaborative Trial of Ovarian Cancer Screening (UKCTOCS) still considered that using those biomarkers for screening EOC patients has not significantly reduced the number of deaths from ovarian and fallopian tube cancers, it is not recommended for the general population [9]. Thus, it is necessary to develop new screening methods for EOC, which could accurately determine the occurrence and progression of the disease and predict the prognosis of patients.

On the other hand, due to the maintenance of the survival of cancer cells, some metabolic changes in cancer patients could be detected prior to the occurrence of symptoms and signs, such as abnormalities in biological fluids, including blood, plasma, urine, and ascites [10]. Metabolomics is an emerging discipline that has been gradually developed and applied in recent years. By analyzing metabolites in patients’ blood, urine, and other body fluids, valuable biomarkers used for clinical diagnosis, prognosis, and monitoring can be identified, which are also applied for screening due to their ability to detect the new molecular therapeutic targets [11, 12]. Since they are products and substrates of metabolism, metabolites also reflect changes in cell functions related to tumorigenesis and cancer development, such as cell proliferation and apoptosis [13]. The metabolomics combined with the multivariate analysis has been applied to diagnose different types of cancer [14, 15, 16]. For example, a simple solid-phase extraction method based on octadecylsilyl (ODS) coupled with internal extractive electrospray ionization mass spectrometry (ODS-SPE-iEESI-MS) was developed for the analysis of blood plasma samples, with features of high sensitivity, high specificity, low sample consumption, and high throughput [17]. Moreover, the iEESI-MS and HPLC-MS techniques can identify the plasma metabolites of the EOC patients, which can directly or indirectly regulate the metabolic function of EOC. This study evaluated the value of the novel biomarkers identified from the plasma metabolites of the EOC patients by iEESI-MS and HPLC-MS in the application of screening, early diagnosis, and prognostic prediction of EOC patients.

2. Materials and methods

2.1 Sample information

2.1.1 The enrolled EOC patients

The participants in this study were divided into two groups: the EOC patients and healthy subjects. All EOC patients were newly diagnosed with EOC, confirmed by surgery and pathology in our hospital’s oncology department. They were treated at the First Hospital of Jilin University during the period from July 2014 to July 2017. A total of 78 patients with EOC were enrolled in this study. Patients enrolled in the study were not allowed to receive chemotherapy or radiation before taking blood samples. Prior to enrolling any patients, all the participating institutions must obtain the approval of the local institutional review board. All the patients had provided written informed consent before collecting their blood samples. Patients with a history of malignancy other than skin basal cell carcinoma were excluded from this study. Patients underwent baseline assessment before entering the study, including physical examination, blood count, CA-125 levels, and preoperative computed tomography. The clinical stages of those patients were determined according to the International Society of Obstetricians and Gynecologists (FIGO) staging standard in 2005. Stage I tumors are confined to the ovaries or fallopian tubes. Stage II tumors involve one or both ovaries, fallopian tubes with pelvic extension (below the pelvic margin), or primary peritoneal carcinoma. Stage III tumors involve one or both ovaries or fallopian tubes or primary peritoneal carcinoma with cytological or histological evidence of spread to the peritoneum outside the pelvis and/or metastasis to retroperitoneal lymph nodes. Stage IV tumors have distant metastases, excluding peritoneal metastasis [18]. The ovarian cancer patients who met the enrollment criteria and agreed to participate in stage I–IV EOC could be enrolled in this study. All of them were admitted to our hospital between July 2014 and July 2017, among which there were 22 cases of stage I, 7 cases of stage II, 46 cases of stage III, and 3 cases of stage IV. Their cell classification included high-grade 39 cases and low-grade 10 cases. Their cell types were listed as serous adenocarcinoma (55 cases), clear cell carcinoma (5 cases), mucinous cystadenocarcinoma (6 cases), endometrial adenocarcinoma (4 cases), and other epithelial carcinomas (8 cases) (Table 1). In addition, 30 healthy women underwent physical examination in our hospital during the same period as controls in our study. All the healthy control group participants provided written informed consent before collecting their blood samples. The investigation of the plasma metabolites of the EOC patients was approved by the Medical Committee in the Hospital Institutional Review Board of the First Hospital of Jilin University, Jilin, P.R. China (2014-337).

Table 1
Clinical information related to patients’ plasma samples

Samples	EOC	Normal	Blind
Number of women	78	30	6
Age
Median/mean	53/50.3	53/53.2	–
Range	22–75	36–75
Staging
Phase I	22	–
Phase II	7
Phase III	46
Phase IV	3
Cell classification
High-grade	39
Low-grade	10
Cell types
Serous adenocarcinoma	55
Mucinous adenocarcinoma	6
Clear cell carcinoma	5
Endometrial adenocarcinoma	4
Other	8

2.1.2 Pretreatment of plasma samples for HPLC-MS

The collected plasma samples were stored in a $-$ 80 ${{}^{\circ}}$ C freezer and thawed to room temperature before measurement, followed by mixing 50 $\mu$ L of plasma samples and 100 $\mu$ L acetonitrile and a vortex for 30 s (800 rpm) at room temperature. After 10 min of equilibrium, these samples were centrifuged at a speed of 12,000 rpm for 15 min to obtain a freeze-dried layer of the sample, which was finally re-dissolved in 50 $\mu$ L deionized water by vortex for 30 s. After 10 min of equilibrium the supernatant (30 $\mu$ L) was transferred to a vial, which was frozen at $-$ 80 ${}^{\circ}$ C before use.

2.2 Materials and chemicals

The chemicals used in this study were HPLC-grade methanol, acetonitrile (ACN), and formic acid (FA), all purchased from Merck KGaA (Darmstadt, Germany). Other materials included ultrapure water obtained from the Milli-Q purification system (Millipore; Bedford, MA, USA), nylon 66 pin type filter (pore size 0.22 $\mu$ m) from Tianjin Navigator Lab Instrument Co, Ltd (Tianjin, China), octadecylsilyl (C 18, particle size was 40–75 $\mu$ m. Specific surface area was 300 m ${}^{2}$ /g), and chromatographic column (Accucore aQ C 18, 2.1*100*1.9, Thermo Scientific). All materials were used as received without further purification.

2.3 Preparation of octadecylsilyl using the organic imported nylon 66 needle type filter

Octadecylsilyl (ODS), also called C18, is an organic group. ODS silica gel is the most commonly used packing for reversed-phase chromatography. As shown in Fig. 1A, the mixture of octadecylsilyl and methanol was made in a glass beaker (homogenate concentration of 2 mg/mL), followed by transferring the mixture (0.5 ml) into a syringe (Hongda company, Nanchang, China) (Fig. 1B) for capturing the octadecylsilyl (groups) by an organic imported nylon 66 needle type filter, where ODS (groups) were held. However, methanol waste was allowed to pass through and discharge into a glass beaker (Fig. 1C). Then, ODS captured by the filter was washed with 1 mL ACN/H2O (v/v, 95/5) (Fig. 1D (1)) and 1 mL ACN/H2O (v/v, 5/95) (Fig. 1D (2)) in turn.

Figure 1.

The schematic diagram of the concept of the iEESI-MS method. The Orbitrap Fusion™Tribrid™mass spectrometer (Thermo Scientific, San Jose, CA, USA) was used for all the experiments coupled with a homemade iEESI source. The MS system was performed using the positive-ion (ESI $+$ ). Mass spectra were obtained in the mass range of m/z 100–2000. The ionization source temperature was set at 27 ${}^{\circ}$ C. Meanwhile, the temperature of the ion transfer tube was set at 320 ${}^{\circ}$ C. The capillary voltage was set at 4 kV. The pressure of nitrogen sheath gas was 0.7 MPa. The injection flow rate was set to 60 $\mu$ L/min, and the sample detection time was 5 min. Collision-induced dissociation (CID) experiments were carried out for MS/MS analysis. The system was operated using ESI $+$ , with the mass range (m/z 50–800). The precursor ions were isolated with a widow width of 2.0 Da, and the normalized collision energy was set at 0% and 35%.

2.4 Experimental method of mass spectrometry

2.4.1 SPE based on ODS coupled with iEESI-MS

The solid-phase extraction (SPE) based on the octadecylsilyl coupled with internal extractive electrospray ionization mass spectrometry (ODS-SPE-iEESI-MS) is a novel detection method for the analysis of plasma metabolites (Fig. 1). The Orbitrap Fusion™ Tribrid™mass spectrometer (Thermo Scientific, San Jose, CA, USA) was used for all the experiments coupled with a homemade iEESI source. Briefly, the plasma samples were thawed at room temperature (25 ${{}^{\circ}}$ C). Then, 1 $\mu$ L of the sample was taken using a mini-pipette (10 $\mu$ L volume) for loading. In the absence of pretreatment, the plasma samples will be adsorbed by octadecylsilyl when they are directly added to the filter (Fig. 1E). A syringe containing 1 mL of clean solution ACN/H2O (v/v, 5/95) was used for pretreatment. The clean solution was injected into the filter by a syringe pump (Harvard Apparatus, Holliston, MA, USA). The salt of the plasma samples was eluted with the filter by washing. The waste liquid was discharged into a glass beaker. The flow rate was set at 100 $\mu$ L/min, and the washing time was 1 min. With a syringe containing 1 mL of elution solution ACN/H2O/FA (v/v/v, 25/75/0.1), the elusion process was started with an injection of elusion solution into the filter by a syringe pump (Fig. 1F). Finally, the differential metabolites in plasma samples were analyzed by ODS-SPE-iEESI-MS (Fig. 1G).

2.4.2 Analysis using HPLC-MS technology

Samples were loaded by auto-sampler via the LC system (Thermo Fisher Scientific, Waltham, MA). In this system, the mobile phase A was H ${}_{2}$ O/FA (v/v, 100/0.1), while the mobile phase B was ACN/FA (v/v, 100/0.1). A chromatographic column was used to separate the plasma samples at 30 ${}^{\circ}$ C. The time for sample separation by HPLC-MS was 26 min under the following conditions: (1) 0.1 min, 5% B; (2) 2 min, 5% B; (3) 26 min, 80% B; then enter the post washing, balance post procedure (1) 26.1 min, 90% B: (2) 30 min, 90% B; (3) 30.1 min, 5%B; and (4) 40 min, 5% B. All the changes during the separation period were linear. The flow rate was set at 250 $\mu$ L/min. The MS system was operated under the following conditions: the ESI $+$ and the mass range of m/z 100–2000. The ionization source temperature was set at 350 ${}^{\circ}$ C, and the temperature of the ion transfer tube was set at 320 ${}^{\circ}$ C. The gas flow was set at 0.7 Mpa. The capillary voltage was 4 kV. Furthermore, the sample injection volume was set to 2 $\mu$ L.

2.5 Data preprocessing

Statistical and machine learning methods are essential for analyzing mass spectrometry (MS)-generated complex metabolic data. Construction of the statistical model with calculation algorithms for multivariate analysis and feature selection of metabolomic data requires the initial “training” step and subsequent verification steps. In the “training” step, the orthogonal partial least squares discriminant analysis (OPLS-DA) was used to improve the PLS-DA algorithm for the multivariate analyses of the metabolomic data. The statistical model based on OPLS-DA can discriminate (divide) two or more groups (classes) using multivariate data. Discriminating variables were selected according to variable VIP $>$ 1. Notably, R2Ycum and Q2cum are two critical parameters for evaluating the quality of the models. The values of R2Ycum and Q2cum close to 1 indicate a relatively good prediction for the multivariate models. Cross-validation was also applied to validate the model using permutation testing ( $n=$ 100). Meanwhile, the primary mass spectrum was used to search the database, which was established by metabolomics software (Compound discoverer) for HPLC-MS analysis. In the validation step, our study conducted a collision-induced dissociation (CID) experiment for potential markers, and the structure was further confirmed by MS/MS data. The biomarkers were identified by searching the National Institute of Standards and Technology (NIST) Mass Spectral Database, the Human Metabolome Database (HMDB) (http://www.hmdb.ca), and the Endogenous Metabolites Database 4400 compounds data. The data analysis using OPLS-DA was performed by the supervised multivariate statistical approach (SIMCA) software (version 13.0; Umetrics, Sweden). In addition, the identified metabolites were mapped to the Kyoto Encyclopedia of Genes and Genomes (KEGG) database for comprehensive analysis and pathway enrichment analysis. The results were presented as the bubble map. Each bubble represents a metabolic pathway. The bubble size indicates the number of genes annotated to a KEGG Pathway, while the bubble color represents the enriched Q-value. The darker the color, the smaller the Q-value.

3. Results and discussion

3.1 Evaluation of the capability of the iEESI-MS and the HPLC-MS based on the analysis of metabolomic data using OPLS-DA

The analysis of metabolomic data of all groups was performed using OPLS-DA, including the EOC patients (P group), normal subjects (N group), and blind subjects (B group). The OPLS-DA score plots showed a clear separation trend between the P and N, with satisfactory modeling and predictive abilities, according to the analysis results of the metabolomic data collected from iEESI-MS (R2Ycum $=$ 0.927, Q2cum $=$ 0.796) (Fig. 2A) and the data from the HPLC-MS (R2Ycum $=$ 0.895, Q2cum $=$ 0.811) (Fig. 2E). Furthermore, the partial least squares discriminant analysis (PLS-DA) was used to differentiate the P, N, and B groups. There was a complete separation of the data collected from iEESI-MS: R2Ycum $=$ 0.902 and Q2Ycum $=$ 0.605 in Fig. 2C; R2Ycum $=$ 0.836 and Q2Ycum $=$ 0.738 in Fig. 2G. The validity of our classification model was confirmed because the Q2 regression line (shown in blue) has a negative intercept, and all of the permuted R2-values (shown in green) on the left are lower than the original point of the R2 value on the right (Fig. 2D and H). In the s-plot diagram, the identified biomarkers were loaded far from the origin in the loading plot (Fig. 2B and F). To identify the potential biomarkers, we used their m/z information obtained by positive detection mode to search the established database. The error of the data collected from iEESI-MS was set at 10 ppm.

Figure 2.

The data analysis using the OPLS-DA method to evaluate the capability of the iEESI-MS and the HPLC-MS. A–B: OPLS-DA score plots (A) and S-plots (B) based on the data collected from iEESI-MS (A and B) for distinguishing P and N groups. The variables with VIP $>$ 3.5 were highlighted in red boxes. (C) the PLS-DA score plots based on the data from iEESI-MS for distinguishing P, N, and B groups. (D) Statistical validation of the PLS-DA model by permutation testing ( $n=$ 100). E–F: OPLS-DA score plots (E) and S-plots (F) based on the data from HPLC-MS (A and B) for distinguishing P and N groups. Furthermore, the potential markers were selected. The variables with VIP $>$ 3.5 were highlighted in red boxes. (G) the PLS-DA score plots based on the data from HPLC-MS for distinguishing P, N, and B groups. (H) Statistical validation of the PLS-DA model by permutation testing ( $n=$ 100).

Figure 3.

iEESI-MS mass spectrograms of plasma samples collected from the EOC patients ( $n=$ 60) and normal subjects ( $n=$ 60). (A): Mass spectrometry of iEESI-MS plasma samples collected from the EOC patients, (B): Mass spectrometry of iEESI-MS plasma samples from normal subjects. (C–E): Score plots based on the analysis of iEESI-MS mass spectrometry data using OPLS-DA. OPLS-D two-dimensional analysis (C), S-plots (D), and the variables with VIP $>$ 3.5, which are highlighted in red boxes and VIP scatter plots (E).

3.2 iEESI-MS mass spectrograms of the plasma samples

The metabolomic data of plasma samples of the EOC patients (60 cases) and normal subjects (60 cases) were collected by IEESI-MS. The data analysis showed that there were significant differences in the m/Z (100–800) mass spectrograms between the EOC patients and normal subjects (Fig. 3A: EOC patients and Fig. 3B: normal subjects). The analysis results of iEESI-MS mass spectrometry data by OPLS-DA using SIMCA software are presented as the score plots in Fig. 3 CE. As shown in the two-dimensional and S-Plot scatter diagram of OPLS-DA (Fig. 3C and D), respectively, there was a significant difference in the ratio of plasma metabolites between the EOC patients and healthy women. S-plot dispersion plots also showed the differential metabolizable product-charge ratios in plasma between the EOC patients and healthy women, with greater differences at points further dispersed. As shown in Fig. 3E, VIP scatter plots were used to distinguish group P from group N the possible markers (VIP $>$ 3.5) were selected. In those plots, the variables (VIP $>$ 3.5) were highlighted in red, suggesting that analysis of the IEESI-MS mass spectrograms of the plasm samples can effectively distinguish the EOC patients from normal subjects.

3.3 PLS-DA recognition score map of the IEESI-MS mass spectrometry data (including self-configured plasma samples)

The EOC patients and healthy subjects could be distinguished in the PLS-DA recognition score map of the IEESI-MS mass spectrometry data of their plasm samples. In addition, all of the blind samples were also distinguished from those of the healthy subjects, suggesting that the analysis results were reliable. All of the analysis results of 98 EOC patients [78 in the original sample (P) and 20 in the self-configured sample (ZP)], normal subjects [30 in the original sample (H) and 30 in the self-configured sample (ZH)], and six in the blind sample (B) were presented in Fig. 4A–C, including (A) PLS-DA diagram of the two-dimensional analysis, (B) PLS-DA diagram of the three-dimensional analysis, and (C) S-plot dispersion (points further apart, showing the difference in metabolites between the EOC patients and healthy subjects). As shown in Fig. 4A–C, the analysis of iEESI-MS mass spectrometry data by the PLS-DA method can also effectively and accurately distinguish the EOC patients and healthy subjects, even under the interference of self-configured plasma samples and blind samples. Further analysis was performed by excluding the self-dispensing and blind samples. Only 78 samples of the EOC patients (P) and 30 samples of the normal subjects (N) were used for analysis. The OPLS-DA recognition score map of IEESI-MS mass spectrometry data was presented in Fig. 4D–F, including (D) the PLS-DA diagram of two-dimensional analysis, (E) the PLS-DA diagram of three-dimensional analysis, and (F) S-plot dispersion. Similar results were obtained to prove our conclusion.

Table 2
Potential biomarkers for distinguishing P and N groups by analysis of data from IEESI-MS

Potential biomarkers	Quasi-molecularion	m/z detected	m/z exact	Chemical formula	Fragment	Mass error (ppm)
6-Dimethylaminopurine	[M $+$ H] $+$	164.09184	164.09307	C7H9N5	148,136,121	7.5
L-Fucose	[M $+$ H] $+$	165.07575	165.07462	C6H12O5	147,137,123	6.8
L-Histidine	[M $+$ H] $+$	156.07539	156.07675	C6H9N3O2	138,110	8.7
L-tryptophan	[M $+$ H] $+$	205.09785	205.09715	C11H12N2O2	187,145	3.4
Phenylpyruvic acid	[M $+$ H] $+$	165.05629	165.05462	C9H8O3	137,105	10
L-Phenylalanine	[M $+$ H] $+$	166.08597	166.08626	C9H11NO2	148,119	1.7
L-Leucine	[M $+$ H] $+$	132.10294	132.10191	C6H13NO2	86	7.8
Lyso PC (16:0)	[M $+$ H] $+$	496.33765	496.33977	C24H50NO7P	478,184	4.3
Lyso PC (18:3)	[M $+$ H] $+$	518.32452	518.32412	C26H48NO7P	459,173,146	0.8
Lyso PC (17:0)	[M $+$ H] $+$	510.33863	510.35542	C25H52NO7P	510,492,451,327,258,184	32
2-Piperidinone	[M $+$ H] $+$	100.075691	100.07582	C5H9NO	100,69,58	1.3

Figure 4.

PLS-DA recognition score map of IEESI-MS mass spectrometry data. A–C are the results of the EOC patients ( $n=$ 98, including the same (P) ( $n=$ 78) and their configuration (ZP) ( $n=$ 20)), normal subjects ( $n=$ 60, including 30 cases of the same (N) and 30 cases of their configuration (useful)), and blind (B) ( $n=$ 6). Figure 4A is the plot of the PLS-DA two-dimensional analysis. Figure 4B is the diagram of the PLS-DA three-dimensional analysis. Figure 4C is the S-plot dispersion. Figure 4D–F are the analysis results of the EOC patients (P) ( $n=$ 78) and normal subjects (N) ( $n=$ 30), in the order as follows: two-dimensional analysis (D), three-dimensional analysis (E), and S-plot dispersion (F).

3.4 Identification of metabolites based on their VIP values of iEESI-MS mass spectrometry data determined by PLS-DA

The VIP value reflects the influence of each variable on classification and the proportion of an individual metabolite in all metabolites. Thus, the VIP value of each metabolite is varied. It is generally considered that the substances (VIP $>$ 1) may be considered to be different metabolites. In this study, 880 metabolites with VIP value $>$ 1 were identified by analyzing iEESI-MS mass spectrometry data using the PLS-DA method. The metabolites with VIP value $>$ 3.5 identified by analysis of iEESI-MS mass spectrometry data (including self-configured plasma samples) are listed in Table S1. After excluding self-configured plasma samples, those with VIP value $>$ 3.5 were listed in Table S2. The molecular structure of metabolites can be preliminarily determined by comparing the CID experimental spectra with those in the existing HMDB database (Table 2). Finally, 11 metabolites (Table 2) were identified, which were significantly associated with EOC development (the error was set at 5 ppm) and were involved in energy metabolism, glucose, lipid, and amino acid metabolism, as well as metabolic reprogramming in cancer and purine nucleotide metabolism, both of which were associated with cancer cell proliferation.

Table 3
Potential biomarkers for distinguishing P and N groups by HPLC-MS

Common name	Formula	RT (min)	Molecular weight exact	Molecular weight detected	Mass error $[\text{ppm}]$	$P$ value
						P versus N	N versus P
Lyso PC (16:0)	C ${}_{24}$ H ${}_{50}$ NO ${}_{7}$ P	23.805	495.33249	495.33171	1.57	1.30E-02	5.70E-02
Lyso PC (18:1(9Z))	C ${}_{26}$ H ${}_{52}$ NO ${}_{7}$ P	22.755	521.34814	521.34764	0.95	1.90E-01	3.20E-01
L-Phenylalanine	C ${}_{9}$ H ${}_{11}$ NO ${}_{2}$	4.207	165.07898	165.07905	$-$ 0.42	2.70E-01	3.90E-01
Lidocaine	C ${}_{14}$ H ${}_{22}$ N ${}_{2}$ O	12.216	234.17321	234.17323	$-$ 0.09	4.40E-01	5.20E-01
Sphinganine	C ${}_{18}$ H ${}_{39}$ NO ${}_{2}$	22.884	301.29808	301.29756	1.71	1.30E-01	2.50E-01
L-Leucine	C ${}_{6}$ H ${}_{13}$ NO ${}_{2}$	1.66	131.09463	131.09459	0.28	2.70E-02	9.40E-02
Decanoylcarnitine	C ${}_{17}$ H ${}_{33}$ NO ${}_{4}$	23.117	315.24096	315.24065	0.97	2.70E-02	9.40E-02
Phenylpyruvic acid	C ${}_{9}$ H ${}_{8}$ O ${}_{3}$	2.072	164.04734	164.04748	$-$ 0.82	1.50E-01	2.70E-01
L-Octanoylcarnitine	C ${}_{15}$ H ${}_{29}$ NO ${}_{4}$	18.664	287.20966	287.20931	1.22	4.80E-01	5.60E-01
L-Tyrosine	C ${}_{9}$ H ${}_{11}$ NO ${}_{3}$	2.103	181.07389	181.07404	$-$ 0.83	1.80E-01	3.10E-01
L-(-)-Methionine	C ${}_{5}$ H ${}_{11}$ NO ${}_{2}$ S	1.224	149.05105	149.05116	$-$ 0.71	2.60E-01	3.90E-01
L-Pipecolate	C ${}_{6}$ H ${}_{11}$ NO ${}_{2}$	1.131	129.07898	129.07914	$-$ 1.22	3.10E-01	4.30E-01
Citric acid	C ${}_{6}$ H ${}_{8}$ O ${}_{7}$	1.559	192.027	192.02745	$-$ 2.32	8.80E-01	9.00E-01
L-Tryptophan	C ${}_{11}$ H ${}_{12}$ N ${}_{2}$ O ${}_{2}$	8.991	204.08988	204.08989	$-$ 0.04	1.40E-01	2.70E-01
Methylacetoacetate	C ${}_{5}$ H ${}_{8}$ O ${}_{3}$	2.243	116.04734	116.04695	3.38	5.40E-01	1.40E-01
L-Histidine	C ${}_{6}$ H ${}_{9}$ N ${}_{3}$ O ${}_{2}$	1.104	155.06948	155.0695	$-$ 0.16	3.70E-01	4.70E-01
1-3-7-Trimethyluric acid	C ${}_{8}$ H ${}_{10}$ N ${}_{4}$ O ${}_{3}$	1.053	210.07529	210.07468	2.89	9.40E-01	9.50E-01
L-Carnitine	C ${}_{7}$ H ${}_{15}$ NO ${}_{3}$	0.817	161.10519	161.10525	$-$ 0.37	4.70E-01	5.50E-01
Hippuric acid	C ${}_{9}$ H ${}_{9}$ NO ${}_{3}$	12.347	179.05824	179.05819	0.29	9.80E-03	4.80E-02
Hexanoylcarnitine	C ${}_{13}$ H ${}_{25}$ NO ${}_{4}$	15.58	259.17836	259.17842	$-$ 0.25	5.00E-02	1.40E-01
Glycocholic acid	C ${}_{26}$ H ${}_{43}$ NO ${}_{6}$	20.608	465.30904	465.31035	$-$ 2.81	1.40E-05	4.60E-04

3.5 HPLC-MS chromatograms of plasma samples

There were significant differences in HPLC-MS chromatograms of plasma samples or the plasma mass spectrogram between the EOC patients ( $n=$ 78) and healthy subjects ( $n=$ 30). A total of 21 differential molecules between the EOC patients and healthy subjects were identified by the HPLC-MS (Table 3). Those differential molecules were mainly related to energy metabolism, such as L-Carnitine, Citric acid, Decanoylcarnitine, Hexanoylcarnitine, L-Octanoylcarnitine, lipids (including Lyso PC (18:3), Lyso PC (18:1(9Z), and Sphinganine), amino acids, and others, such as Phenylpyruvic acid, L-phenylalanine, L-histidine, L-tryptophan, L-Tyrosine, L-leucine, L-methionine, L-Pipecolate, 1-3-7-Trimethyluric acid, Lidocaine, Methylacetoacetate, Hippuric acid, and Glycocholic acid. 6 identified metabolites by HPLC-MS were consistent with the molecules identified by the IEESI-MS method, including Lyso PC (16:0), L-Phenylalanine, L-Leucine, Phenylpyruvic acid, L-Tryptophan, and L-Histidine.

3.6 Identification of the potential markers for discriminating the groups of the EOC patients and normal subjects

Our study demonstrated that the plasma metabolomics analysis by iEESI-MS and HPLC-MS is a valuable tool for discriminating the groups of the EOC patients and normal subjects ( $n=$ 11) in the analysis of iEESI-MS and HPLC-MS data. In this study, 11 metabolic markers were identified by iEESI-MS (Table 2), all of which had certain biological functions. In addition, 21 metabolic markers were identified by HPLC-MS (Table 3), most of which had relatively clear molecular biological functions. In this study, 5 metabolites were identified by iEESI-MS only; 15 differential metabolites in EOC were identified by HPLC-MS, 6 differential metabolites in EOC were identified by both iEESI-MS and HPLC-MS. Thus, 26 metabolic markers related to EOC were identified in this study (Fig. 5A). These metabolic molecules are mainly involved in glucose, lipid, amino acid, and nucleic acid metabolism (Fig. 5B), which help cancer cells undergo metabolic reprogramming and immune escape and modulate the tumor microenvironment, affecting the occurrence and development of EOC. Six differential metabolites in EOC were identified using iEESI-MS and HPLC-MS methods, including Lyso PC (16:0), L-Phenylalanine, L-Leucine, Phenylpyruvic acid, L-Tryptophan, and L-Histidine. All of them could be used as the essential molecules for screening the EOC patients and monitoring EOC development.

Figure 5.

Potential metabolic markers and hypothesis of mechanisms. A: identification of potential markers for discriminating the EOC patients and normal subjects by the PLS-DA recognition score map using the mass spectrometry data of IEESI-MS and HPLC-MS. B: the possible mechanisms of these metabolic molecules associated with EOC development.

3.7 Analysis of the differential metabolites-related metabolic pathway

In our study, the differential metabolites-related metabolic pathways were explored by the enrichment analysis according to the highest correlation of the differential metabolites with the metabolic pathways. Using the KEGG analysis, which is a comprehensive database that integrates genome, chemistry, and system function information, metabolic pathways associated with the differential metabolites were explored (Fig. 6). The major metabolism-related pathways and molecules included (1) Metabolites involved in Biosynthesis of amino acids, such as L-Leucine, Citric acid, L-Histidine, L-Tyrosine, Phenylpyruvic acid, L-Phenylalanine, and L-Tryptophan; (2) Metabolites involved in Central carbon metabolism in cancer, such as Citric acid, L-Leucine, L-Phenylalanine, L-Histidine, L-Tryptophan, and L-Tyrosine; (3) The molecules involved in 2-Oxocarboxylic acid metabolism, such as L-Leucine, Citric acid, L-Phenylalanine, L-Tryptophan, L-Tyrosine, and Phenylpyruvic acid; (4) The molecules involved in Aminoacyl-tRNA biosynthesis, such as L-Tryptophan, L-Tyrosine, L-Leucine, L-Phenylalanine, and L-Histidine; (5) Metabolic molecules involved in Protein digestion and absorption, such as L-Leucine, L-Phenylalanine, L-Tryptophan, L-Histidine, and L-Tyrosine. Suggesting the involvement of the metabolic network of EOC. The information on the EOC-associated metabolic pathways might provide directions for future studies. The enrichment analysis results of the differential metabolites are presented in Table 4. KEGG enrichment analysis showed that the differentially expressed genes were mainly involved in the cancer-related signaling pathways, such as energy metabolism, cell migration, and transmembrane transporter.

Table 4
The results of the enrichment analysis of the differential metabolites

Pathway	Total	Hits	Raw_p	FDR	Impact
Valine, leucine, and isoleucine biosynthesis	23	1	0.05006	0.08224	0.0434782608696
Histidine metabolism	47	1	0.1001	0.1278	0.0212765957447
Valine, leucine, and isoleucine degradation	42	1	0.08983	0.1215	0.0238095238095
Fructose and mannose metabolism	54	1	0.1142	0.1314	0.0185185185185
Phenylalanine, tyrosine, and tryptophan biosynthesis	35	3	3.885e-05	0.0001489	0.0857142857143
Glycine, serine, and threonine metabolism	50	1	0.1061	0.1285	0.02
Phenylalanine metabolism	60	2	0.006874	0.01757	0.0333333333333
beta-Alanine metabolism	32	1	0.06907	0.1059	0.03125
Aminoacyl-tRNA biosynthesis	52	4	1.783e-06	1.025e-05	0.0769230769231
mTOR signaling pathway	4	1	0.008861	0.02038	0.25
C-type lectin receptor signaling pathway	11	1	0.02421	0.0464	0.0909090909091
2-Oxocarboxylic acid metabolism	134	4	8.047e-05	0.0002644	0.0298507462687
Serotonergic synapse	42	1	0.08983	0.1215	0.0238095238095
Biosynthesis of amino acids	128	5	1.279e-06	9.809e-06	0.0390625
Protein digestion and absorption	47	4	1.179e-06	9.809e-06	0.0851063829787
Mineral absorption	29	3	2.18e-05	0.0001003	0.103448275862
ABC transporters	137	3	0.002281	0.006559	0.021897810219
Shigellosis	16	1	0.03505	0.06202	0.0625
African trypanosomiasis	8	1	0.01766	0.03692	0.125
Central carbon metabolism in cancer	37	4	4.39e-07	9.809e-06	0.108108108108

Figure 6.

Enrichment analysis of the differential metabolites.

3.8 Biological and cancer-related functions of the potential markers

3.8.1 Lipid metabolism

Lipids are a class of heterogeneous metabolites involved in many vital cellular functions, such as membrane construction, energy storage, and signal transduction [19]. Membrane fluidity can directly affect the motility and behavior of tumor cells at different stages of metastatic spread [20]. Lipid metabolism may be involved in cancer-related cell proliferation, inflammation, drug resistance, tumor immune escape, and apoptosis [21, 22]. In our study, aberrant expression of some lipids was found, including Lyso PC (16:0), Lyso PC (18:3), Lyso PC (17:0), Lyso PC (18:1(9Z)), sphinganine, and intermediate products of lipid metabolism: decanoylcarnitine, hexanoylcarnitine, and glycocholic acid. EOC cells tend to migrate to the omentum, which is mainly composed of adipocytes. Co-culture of adipocytes and EOC cells revealed that adipocytes could directly transfer lipid to ovarian cancer cells, promoting tumor growth in vitro and in vivo [23]. The nutritional lipid environment, and lipid subclasses produced by phospholipid cleavage are closely associated with cancer development [24]. Abnormalities of both lysophospholipid, a kind of product of catabolic metabolism of phospholipid (the basic substance of cell membrane), and lipoprotein A (LPA), a type of Low Density Lipoprotein (LDL) cholesterol, have been detected in blood and ascites of the EOC patients. Both are valuable biomarkers and potential therapeutic targets [25, 26, 27]. In this study, abnormal expression of Lyso PC (16:0), a potential biomarker, is found in the plasma of the EOC patients. Aberrant expression of Lyso PC (16:0) is also found in other solid tumors, such as laryngeal [28] and colorectal cancers [29]. Moreover, PC (16:0/16:1) can promote the proliferation of colorectal cancer cells by activating Akt and ERK pathways. PC (16:0/16:1) was added to the culture media to maintain colorectal cancer and non-cancerous cell lines. The growth rate of the colorectal cancer cell line increased about twofold than that of the non-cancerous cell line. Further studies also found that PC (16:0/16:1) could promote the generation of lysophosphatidic acid, thereby increasing the expression of lysophosphatidic acid receptor and activating Akt and ERK pathways, leading to enhancing cell survival and proliferation [29]. Many metabolic enzymes are abnormally expressed in ovarian cancer cells, affecting lipid synthesis or degradation, resulted in disorder of lipid metabolism which is associated with cancer development. At present, rate-limiting metabolic enzyme inhibitors combined with first-line chemotherapy drugs have become a new strategy in treating ovarian cancer [30, 31].

3.8.2 Amino acid metabolism

Metabolism of Amino acids is important for biological processes because of their involvement in the biosynthesis of nucleotides, antioxidants (glutathione), glucosamine, and polyamines. The metabolites of amino acids are also involved in energy-producing processes, such as the tricarboxylic acid cycle [32]. A total of 20 left-handed amino acids are the cornerstone of the human body. Abnormalities of several amino acids were found in this study, including phenyl pyruvic acid, L-phenylalanine, L-histidine, L-tryptophan, L-Tyrosine, L-leucine, L-methionine, and L-Pipecolate. Methionine, phenylalanine, tryptophan, and leucine are nutritionally essential amino acids; the others are semi-essential amino acids. Tryptophan is an essential amino acid that must be obtained from the diet to support physiological processes, including cell growth and maintenance [33]. The L-tryptophan [34] promotes the inherent malignancy of tumor cells and restricts anti-tumor immunity. For this reason, L-tryptophan might be a potential therapeutic target for cancer immunotherapy. In addition, various roles and functions of the tryptophan catabolic pathway in tumor-related immunosuppression have been explored through mechanism-based studies [35]. For example, the indoleamine 2,3-dioxygenase 1 (IDO1)-induced tryptophan depletion in the tumor microenvironment (TME) can inhibit mammalian rapamycin complex (mTORC) and activate the general control nonderepressible 2 (GCN2), a stress response kinase in tumor-infiltrating T cells, leading to incompetence and apoptosis [36]. In preclinical models, the enhanced kynurenine pathway (KP) has been associated with the failure of anti-tumor immunity and tumor progression, which more likely depends on AhR activation [37]. Therefore, the tryptophan catabolic pathway is considered an essential step for tumor cells to evade the innate and adaptive immune system [38]. Besides tryptophan, arginine and histidine also play a significant role in enhancing immunity [39]. Specifically, as previously reported [40], histidine content was significantly decreased in the cancer patient, compared to that in the normal subjects, including colonic carcinomas, prostate cancer [41], and breast cancer [42]. In addition to the differential diagnosis of malignant diseases, histidine can also be used as a predictor. It has been reported that hypoxanthine and histidine in early treated serum are predictive biomarkers for the response to programmed cell death 1 ligand 1 (PD-1) blockade in patients with advanced non-small-cell lung cancer (NSCLC) [12].

Amino acids are the building blocks of proteins and are metabolized as energy substances or function as signaling molecules. Phenylalanine and phenylpyruvate can inhibit the glycolysis of hepatocytes [43]. Destructive effects of phenylalanine and its metabolites on brain glucose metabolism were also reported. Phenylpyruvate may be responsible for reducing glucose consumption by decreasing hexokinase activity, resulting in decreased ATP production in brain cells [44]. Amino acid metabolism was proven to regulate immune response and cancer development by maintaining tumor proliferation and homeostasis. Targeting amino acid metabolism has emerged as a potential therapeutic strategy for cancer patients.

3.8.3 Cell division and apoptosis

6-dimethylaminopurine (6-DMAP) is a protein phosphorylation inhibitor, which could increase intracellular calcium levels in oocytes and inhibit the extrusion of the second polar body. It has been reported that 6-DMAP can induce apoptosis in S-phase stalled HeLa cells, activate the M-phase promoters, and induce premature mitosis in similarly treated hamster cell lines. However, there was little effect of 6-DMAP on asynchronously grown HeLa cells or lovastatin early G1 stagnate cells [45]. Moreover, treatment with 0.6 mM of 6-DMAP for 6 h can induce cell-cycle arrest at the G2-phase [46]. 6-DMAP also induced apoptosis through a p53-independent pathway in a human lymphoma U937 cell line. 6-DMAP-mediated apoptosis is associated with activating a mitochondrial caspase-dependent pathway and [Ca2 $+$ ]i-dependent pathway [47]. Since EOC is a malignant tumor of the reproductive system with a special secretory function, the expression and function of 6-DMAP may differ from other systems. Therefore, elucidating the exact mechanism underlying the inhibitory effects of 6-DMAP on cancer may be necessary in future clinical applications.

3.9 Substances related to metabolism

3.9.1 Citric acid

Citrate is essential in metabolic energy production and is involved in synthesizing cholesterol, fatty acids, and isoprene compounds. Citrate also links carbohydrates to fatty acid metabolism [48]. Reduced citrate concentrations have been observed in multiple tumor tissues and in the blood of cancer patients, which might be used to indicate cancer aggressiveness and metabolic prognosis [49]. The determination of citric acid in semen may have a crucial impact on improving the diagnosis of prostate cancer in the early stages of the disease [50]. The levels of citric acid and isocitric acid, both of which are the metabolites of tricarboxylic acid (TCA) cycle, were found elevated in A2780, a human EOC cell line, after treatment with paclitaxel but without change in A-PM3 cells, which is a cell line derived from A2780 exposed to paclitaxel [51]. Citrate can regulate the reprogramming of lipid metabolism in tumor cells, control the fate of tumor cells, promote excess lipid biosynthesis, and cause the senescence of tumor cells. Thus, citrate has been used in tumor therapy [52].

3.9.2 Decanoylcarnitine, hexanoylcarnitine, and L-octanoylcarnitine

L-octanoylcarnitine is a combination of 8-carbon fatty acid with carnitine, which is a crucial product of fatty acid catabolism, suggesting that all fatty acids must enter the inner membrane of mitochondria in the form of lipoylcarnitine for $\beta$ oxidation and complete decomposition. L-Octanoylcarnitine has the best performance in diagnosing breast cancer, with a positive predictive value of 100.0%. In addition, the L-octanoylcarnitine level is dependent on tumor size and hormone receptor expression [53]. A previous study showed that the plasma level of L-octanoylcarnitine was significantly reduced in the EOC patients than that in healthy controls [54], suggesting a high demand for carnitine metabolism in the EOC patients. Fatty acid oxidation (FAO) is the primary energy source in the metabolic pathway, in which the cofactor carnitine is crucial because it acts as a “shuttle.” Since the fatty acid acyl parts are allowed to enter the mitochondrial matrix, investigating the expression and function of the carnitine transporters is essential for a better understanding of the molecular mechanisms of cancer therapy [55].

3.9.3 L-fucose

Fucose is a 6-deoxyhexose with a structure similar to glucose. Fucose has a unique L-configuration, whereas all other naturally occurring sugars in mammals are in the D conformation, making fucose a relatively common component of glycan modification on proteins and lipids. Elevated serum level of fucose was detected in cancer patients, especially in advanced and metastatic tumors, such as stomach and ovarian cancer [56]. Increasing fucose glycosylation of tumor cells contributes to abnormal characteristics such as decreased adhesion and uncontrolled tumor growth. Therefore, detecting serum levels of fucose may be a promising method for early diagnosis and prognosis of various types of cancer [57].

3.9.4 2-piperidone

2-piperidone is a derivative of piperidine, which affects the activity and expression of cytochrome P450 (CYP) 2E1, thereby indirectly affecting oxidative stress and metabolic activation of low molecular weight toxins. A decrease in the plasma level of 2-piperidone was found in the EOC patients [58, 59]. A selective piperidone inhibitor (compound 2) was identified as a highly potent murine double minute-2 inhibitor, with remarkable pharmacokinetic properties and in vivo anti-tumor activity in the SJSA-1 osteosarcoma xenograft models [60].

3.9.5 The differential metabolites related to energy metabolism

Metabolic substrates or the substrates of metabolic pathways are the products that can affect metabolism and participate in various signaling pathways as regulatory factors. The enrichment analysis results of the differential metabolites are presented in Table 4 and Fig. 6 (KEGG enrichment analysis). The enrichment pathways for the abnormal metabolite citric acid included the following aspects: citrate cycle (TCA cycle), glyoxylate and dicarboxylate metabolism, glyoxylate and dicarboxylate metabolism, 2-oxocarboxylic acid metabolism, taste transduction, carbon metabolism, glucagon signaling pathway, biosynthesis of amino acids, and central carbon metabolism in cancer. Moreover, 2-piperidone indirectly affected oxidative stress and metabolic activation by regulating the activity and expression of cytochrome P450 (CYP) 2E1. A decreased plasma level of 2-piperidone in EOC patients has been reported [58, 59]. Our study also obtained similar results. Although we could not identify the pathways or biological functions associated with 2-piperidone, L-fucose, which was abnormally expressed in this study, plays an important role in cell biology. L-fucose and fucosylation are essential for maintaining the immune system and organ development processes, which play an important role in tumor biology and immunology [61]. Therefore, honing and exploiting fucosylation might provide potential therapeutic opportunities in EOC. This study found that the abnormal generation of decanoylcarnitine, l-octanoylcarnitine, and hexanoylcarnitine in patients with EOC may be related to the abnormal fat metabolism of ovarian cancer cells [58, 59]. Fatty acid $\beta$ oxidation requires the presence of carnitine, which is the mitochondrial transporter of the acyl part of fat [62].

The balance between nutrient absorption and energy production, macromolecular biosynthesis, and oxidation-reduction reaction is needed in rapidly proliferating cancer cells. In advanced and metastatic cancer, metabolic reprogramming supports metabolic plasticity and high energy production. Therefore, the therapy targeting metabolic pathways has become the focus of cancer treatment [13].

3.10 Comparison between our study and other studies

Compared to previous studies which used HPLC-MS or/and iEESI-MS methods for diagnosis of EOC, we made a summary including the following aspects: (1) Abnormal generation of methionine, histidine, tryptophan, leucine, L-phenylalanine, and 2-Pipecolate in ovarian cancer [58, 63, 64, 65], suggesting the metabolic characteristics of the EOC patients, and implying high amino acid uptake and consumption in tumor tissue; (2) Abnormal generation of lipid metabolites, such as lyso PC (16:0), lyso PC (17:0) [65], and sphinganine/C16 sphinganine [66] in ovarian cancer; (3) HPLC-MS [66] analysis showed that the metabolic characteristics between the normal control group, platinum-sensitive group, and platinum-resistant group were significantly different. These findings may provide theoretical clues for predicting prognosis and chemotherapy response, reversing drug resistance, and discovering new targets for therapeutic intervention. Although different studies found different results, we attributed those differences to the histological and staging characteristics of the ovarian cancer tissue samples and the metabolome analysis techniques.

3.11 The clinical application of our findings in future

The biomarkers found in this study, such as lyso PC (16:0), l-phenylalanine, l-leucine, phenylpyruvic acid, l-tryptophan, and l-histidine may be used for clinical application in various ways, such as screening of the ECO patients by distinguishing benign tumors and ECO, as well as excluding the diseases which are easily mixed with EOC, like pelvic inflammatory disease and tuberculosis. Further research, particularly on the relationship between the levels of those biomarkers and the long-term prognosis of ECO, is required to predict the pathological type, clinical stage, and curative effect of therapeutic approaches on ECO.

4. Limitations and future continuation of this study

First, only a few potential EOC biomarkers were identified in this study, performed in a single center. Therefore, it is required to conduct further multi-center studies which have larger sample sizes for validation and quantitative testing.

Second, changes in plasma metabolomics in EOC patients reflect the disease progression. Therefore, to explore the potential biomarkers by screening, the differences in metabolomics between early (stage I/II) and late (stage III/IV) EOC should be investigated in future studies. It is worth noting that chemotherapy resistance is the main cause of disease progression. Therefore, comparing the changes in plasma metabolomics before and after treatment with platinum-resistant drugs between the platinum-sensitive and platinum-resistant groups is necessary. The comparison results may provide useful clues for predicting chemotherapy response, reversing drug resistance, and even discovering new targets for therapeutic intervention.

5. Conclusion

Our study provides a comprehensive understanding of the metabolic changes in plasma between the EOC patients and normal control and the significance of the differential metabolites as screening markers for early diagnosis of EOC. This study confirmed that the iEESI-MS is a reliable screening method for EOC patients because it is simple to carry out analysis, with fast and easy-to-use sample pretreatment. This method is also helpful for discovering new potential biomarkers. Metabolomic screening biomarkers might provide reliable indicators for screening and diagnosing EOC and cancer prevention.

Authors’ contributions

Songling Zhang: Put forward the conception; Jiajia Li, Zhenpeng Wang: Interpretation or analysis of data; Zhentong Wei, Wenjie Liu: Preparation of the manuscript; Dongzhen Liu, Linsheng Tan: Supervision; Songling Zhang, Zhentong Wei, Jiajia Li, and Yunhe Yu: Revision for important intellectual content.

Supplementary data

The supplementary files are available to download from http://dx.doi.org/10.3233/CBM-220250.

sj-docx-1-cbm-10.3233_CBM-220250.docx - Supplemental material

Supplemental material, sj-docx-1-cbm-10.3233_CBM-220250.docx

Footnotes

Acknowledgments

The authors appreciate the International Cooperative Joint Laboratory of Nanoarchitecture Chemistry. The work was supported by the Jilin Municipal Department of Finance – Application of metabolomics based on mass spectrometry in early diagnosis and recurrence monitoring of ovarian cancer (No. 20190303143SF), The First Hospital of Jilin University – Application of metabolomics based on mass spectrometry in the diagnosis of ovarian cancer (No. JDYYJCHX003), and the Jilin Provincial Finance Department – Ovarian cancer cells up-regulated NRF2 signaling to adapt to autophagy inhibition (No. JLSWSRCZX2020-088); Special health project of the Department of Finance of Jilin Province (no. JLSCZD2019-078).

Conflict of interest

The authors declare that the research was conducted without any commercial or financial relationships that could be construed as a potential conflict of interest.

References

Jayson

G.C.

Kohn

E.C.

Kitchener

H.C.

and Ledermann

J.A.

, Ovarian cancer, Lancet (London, England) 384 (2014), 1376–1388.

American Cancer Society: American Cancer Society, Cancer Facts & Figures 2021, Atlanta: American Cancer Society, 2021, 1–72.

Torre

L.A.

Trabert

DeSantis

C.E.

Miller

K.D.

Samimi

Runowicz

C.D.

Gaudet

M.M.

Jemal

and Siegel

R.L.

, Ovarian cancer statistics, 2018, CA: A Cancer Journal for Clinicians 68 (2018), 284–296.

Prat

, Ovarian carcinomas: Five distinct diseases with different origins, genetic alterations, and clinicopathological features, Virchows Archiv: An International Journal of Pathology 460 (2012), 237–249.

Merritt

M.A.

Rice

M.S.

Barnard

M.E.

Hankinson

S.E.

Matulonis

U.A.

Poole

E.M.

and Tworoger

S.S.

, Pre-diagnosis and post-diagnosis use of common analgesics and ovarian cancer prognosis (NHS/NHSII): A cohort study, The Lancet. Oncology 19 (2018), 1107–1116.

Armstrong

D.K.

Alvarez

R.D.

Bakkum-Gamez

J.N.

Barroilhet

Behbakht

Berchuck

Chen

L.M.

Cristea

DeRosa

Eisenhauer

E.L.

Gershenson

D.M.

Gray

H.J.

Grisham

Hakam

Jain

Karam

Konecny

G.E.

Leath

C.A.

Liu

Mahdi

Martin

Matei

McHale

McLean

Miller

D.S.

O’Malley

D.M.

Percac-Lima

Ratner

Remmenga

S.W.

Vargas

Werner

T.L.

Zsiros

Burns

J.L.

and Engh

A.M.

, Ovarian Cancer, Version 2.2020, NCCN Clinical Practice Guidelines in Oncology, Journal of the National Comprehensive Cancer Network: JNCCN 19 (2021), 191–226.

Son

J.H.

Lee

Yum

S.H.

Kim

Kong

T.W.

Chang

S.J.

and Ryu

H.S.

, Simplified selection criteria for secondary cytoreductive surgery in recurrent ovarian cancer, Cancers 14 (2022).

Gentry-Maharaj

Blyuss

Ryan

Burnell

Karpinskyj

Gunu

Kalsi

J.K.

Dawnay

Marino

I.P.

Manchanda

Yang

W.L.

Timms

J.F.

Parmar

Skates

S.J.

Bast

R.C.

, Jr. Jacobs

I.J.

Zaikin

and Menon

, Multi-Marker Longitudinal Algorithms Incorporating HE4 and CA125 in Ovarian Cancer Screening of Postmenopausal Women, Cancers 12 (2020), 1931.

Menon

Gentry-Maharaj

Burnell

Singh

Ryan

Karpinskyj

Carlino

Taylor

Massingham

S.K.

Raikou

Kalsi

J.K.

Woolas

Manchanda

Arora

Casey

Dawnay

Dobbs

Leeson

Mould

Seif

M.W.

Sharma

Williamson

Liu

Fallowfield

McGuire

A.J.

Campbell

Skates

S.J.

Jacobs

I.J.

and Parmar

, Ovarian cancer population screening and mortality after long-term follow-up in the UK Collaborative Trial of Ovarian Cancer Screening (UKCTOCS): A randomized controlled trial, Lancet (London, England) 397 (2021), 2182–2193.

10.

Wang

Liu

and Yang

, Application of metabolomics in cancer research: As a powerful tool to screen biomarker for diagnosis, monitoring and prognosis of cancer, Biomark J 1 (2018), 2472–1646.

11.

Schmidt

D.R.

Patel

Kirsch

D.G.

Lewis

C.A.

Vander Heiden

M.G.

and Locasale

J.W.

, Metabolomics in cancer research and emerging applications in clinical oncology, CA: A Cancer Journal for Clinicians 71 (2021), 333–358.

12.

Nie

Xia

Gao

Liu

Yang

Chen

Duan

Yao

Chen

Wang

and Yang

, Serum Metabolite Biomarkers Predictive of Response to PD-1 Blockade Therapy in Non-Small Cell Lung Cancer, Frontiers in Molecular Biosciences 8 (2021), 678753.

13.

Zanotelli

M.R.

Zhang

and Reinhart-King

C.A.

, Mechanoresponsive metabolism in cancer cell migration and metastasis, Cell Metabolism 33 (2021), 1307–1321.

14.

Kwon

H.N.

Lee

Park

J.W.

Kim

Y.H.

Park

and Kim

J.J.

, Screening for early gastric cancer using a noninvasive urine metabolomics approach, Cancers 12 (2020), 2904.

15.

Troisi

Raffone

Travaglino

Belli

Anand

Giugliano

Cavallo

Scala

Symes

Richards

Adair

Fasano

Bottigliero

and Guida

, Development and validation of a serum metabolomic signature for endometrial cancer screening in postmenopausal women, JAMA Network Open 3 (2020), e2018327.

16.

Plumb

R.S.

Gethings

and Peck

, Diagnostic screen for lung cancer using high throughput metabolomics screen, Cancer Research 79 (2019).

17.

Zhang

Huang

Wei

Liu

Chingin

and Chen

, Selective detection of phospholipids in human blood plasma and single cells for cancer differentiation using dispersed solid-phase microextraction combined with extractive electrospray ionization mass spectrometry, The Analyst 145 (2020), 7330–7339.

18.

Prat

, Abridged republication of FIGO’s staging classification for cancer of the ovary, fallopian tube, and peritoneum, Cancer 121 (2015), 3452–3454.

19.

Agarwala

P.K.

Aneja

and Kapoor

, Lipidomic landscape in cancer: Actionable insights for membrane-based therapy and diagnoses, Medicinal Research Reviews 42 (2022), 983–1018.

20.

Shinitzky

, Membrane fluidity in malignancy. Adversative and recuperative, Biochimica et Biophysica Acta 738 (1984), 251–261.

21.

J.L.

Wang

Lan

X.L.

Zeng

Z.C.

Liang

Y.S.

Yan

Y.R.

Song

F.Y.

Wang

F.F.

Zhu

X.H.

Liao

W.J.

Liao

W.T.

Ding

Y.Q.

and Liang

, CAFs secreted exosomes promote metastasis and chemotherapy resistance by enhancing cell stemness and epithelial-mesenchymal transition in colorectal cancer, Molecular Cancer 18 (2019), 91.

22.

Liu

Han

Wang

Fang

and Yin

, Cancer-associated fibroblasts: An emerging target of anti-cancer immunotherapy, Journal of Hematology & Oncology 12 (2019), 86.

23.

Zhao

Cheng

Shi

Yun

and Yang

, MIEF2 reprograms lipid metabolism to drive progression of ovarian cancer through ROS/AKT/mTOR signaling pathway, Cell Death & Disease 12 (2021), 18.

24.

Bian

Liu

Meng

Xing

and Lu

, Lipid metabolism and cancer, The Journal of Experimental Medicine 218 (2021), e20201606.

25.

Xiao

Y.J.

Schwartz

Washington

Kennedy

Webster

Belinson

and Xu

, Electrospray ionization mass spectrometry analysis of lysophospholipids in human ascitic fluids: comparison of the lysophospholipid contents in malignant vs nonmalignant ascitic fluids, Analytical Biochemistry 290 (2001), 302–313.

26.

Chen

and Hu

, Diagnostic value of total plasma lysophosphatidic acid in ovarian cancer: a meta-analysis, International Journal of Gynecological Cancer: Official Journal of the International Gynecological Cancer Society 25 (2015), 18–23.

27.

Zhang

Y.J.

Cao

L.Y.

Z.Z.

Wang

Y.J.

Wang

G.X.

and Gu

, Clinical significance of plasma lysophosphatidic acid levels in the differential diagnosis of ovarian cancer, Journal of Cancer Research and Therapeutics 11 (2015), 375–380.

28.

and Wang

, Lipidomics Identified Lyso-Phosphatidylcholine and Phosphatidylethanolamine as Potential Biomarkers for Diagnosis of Laryngeal Cancer, Frontiers in Oncology 11 (2021), 646779.

29.

Kurabe

Nakamura

Kurachi

Kahyo

and Sugimura

, Abstract 1601: Human colorectal cancer marker PC (16:0/16:1) induces cell growth by activating Akt and Erk pathways, Cancer Research 74 (2014), 1601–1601.

30.

Shen

Feng

Kong

Shao

Meng

Zhang

and Yang

, Deregulation of lipid metabolism: The critical factors in ovarian cancer, Frontiers in Oncology 10 (2020), 593017.

31.

Chen

R.R.

Yung

M.M.H.

Xuan

Zhan

Leung

L.L.

Liang

R.R.

Leung

T.H.Y.

Yang

Sharma

Chan

K.K.L.

Ngu

S.F.

Ngan

H.Y.S.

and Chan

D.W.

, Targeting of lipid metabolism with a metabolic inhibitor cocktail eradicates peritoneal metastases in ovarian cancer cells, Communications Biology 2 (2019), 281.

32.

Lukey

M.J.

Katt

W.P.

and Cerione

R.A.

, Targeting amino acid metabolism for cancer therapy, Drug Discovery Today 22 (2017), 796–804.

33.

Badawy

A.A.

, Kynurenine pathway of tryptophan metabolism: Regulatory and functional aspects, International Journal of Tryptophan Research: IJTR 10 (2017), 1178646917691938.

34.

Kim

and Tomek

, Tryptophan: A Rheostat of Cancer Immune Escape Mediated by Immunosuppressive Enzymes IDO1 and TDO, Frontiers in Immunology 12 (2021), 636081.

35.

Cheong

J.E.

and Sun

, Targeting the IDO1/TDO2-KYN-AhR Pathway for Cancer Immunotherapy – Challenges and Opportunities, Trends in Pharmacological Sciences 39 (2018), 307–325.

36.

Munn

D.H.

Shafizadeh

Attwood

J.T.

Bondarev

Pashine

and Mellor

A.L.

, Inhibition of T cell proliferation by macrophage tryptophan catabolism, The Journal of Experimental Medicine 189 (1999), 1363–1372.

37.

Bessede

Gargaro

Pallotta

M.T.

Matino

Servillo

Brunacci

Bicciato

Mazza

E.M.

Macchiarulo

Vacca

Iannitti

Tissi

Volpi

Belladonna

M.L.

Orabona

Bianchi

Lanz

T.V.

Platten

Della Fazia

M.A.

Piobbico

Zelante

Funakoshi

Nakamura

Gilot

Denison

M.S.

Guillemin

G.J.

DuHadaway

J.B.

Prendergast

G.C.

Metz

Geffard

Boon

Pirro

Iorio

Veyret

Romani

Grohmann

Fallarino

and Puccetti

, Aryl hydrocarbon receptor control of a disease tolerance defence pathway, Nature 511 (2014), 184–190.

38.

Munn

D.H.

and Mellor

A.L.

, IDO in the tumor microenvironment: Inflammation, counter-regulation, and tolerance, Trends in Immunology 37 (2016), 193–207.

39.

Ahmed, A., Tantawy, Deyala, M. and Naguib, Arginine, histidine and tryptophan: A new hope for cancer immunotherapy – ScienceDirect, PharmaNutrition on SciVerse ScienceDirect 8, 100149–100149.

40.

Ahmed-Salim

Galazis

Bracewell-Milnes

Phelps

D.L.

Jones

B.P.

Chan

Munoz-Gonzales

M.D.

Matsuzono

Smith

J.R.

Yazbek

Krell

Ghaem-Maghami

and Saso

, The application of metabolomics in ovarian cancer management: A systematic review, International Journal of Gynecological Cancer: Official Journal of the International Gynecological Cancer Society 31 (2021), 754–774.

41.

Lécuyer

Victor Bala

Demidem

Rossary

Bouchemal

Triba

M.N.

Galan

Hercberg

Partula

Srour

Latino-Martel

Kesse-Guyot

Druesne-Pecollo

Vasson

M.P.

Deschasaux-Tanguy

Savarin

and Touvier

, NMR metabolomic profiles associated with long-term risk of prostate cancer, Metabolomics: Official Journal of the Metabolomic Society 17 (2021), 32.

42.

Hong

Zhang

Yao

Zhang

Sheng

and Guo

, Tumor microenvironment-related multigene prognostic prediction model for breast cancer, Aging 14 (2022), 845–868.

43.

Suzuki

Sato

Fukaya

Suzuki

Yoshizawa

and Sato

, Energy metabolism profile of the effects of amino acid treatment on hepatocytes: Phenylalanine and phenylpyruvate inhibit glycolysis of hepatocytes, Nutrition (Burbank, Los Angeles County, Calif.) 82 (2021), 111042.

44.

Rosa

A.P.

Jacques

C.E.

Moraes

T.B.

Wannmacher

C.M.

Dutra Ade

and Dutra-Filho

C.S.

, Phenylpyruvic acid decreases glucose-6-phosphate dehydrogenase activity in rat brain, Cellular and Molecular Neurobiology 32 (2012), 1113–1118.

45.

Meikrantz

Gisselbrecht

Tam

S.W.

and Schlegel

, Activation of cyclin A-dependent protein kinases during apoptosis, Proceedings of the National Academy of Sciences of the United States of America 91 (1994), 3754–3758.

46.

Grau

González-Picazo

Soler

Escribá

Vendrell

and Viloria

, Effect of different 6-dimethylaminopurine (6-DMAP) treatments on reversible arresting of mono- and tripronuclear embryos at the pronuclear stage, Fertility and Sterility 112 (2019), e274–e275.

47.

Nakashima

Cui

Z.G.

Tabuchi

Takasaki

Zhao

Q.L.

Kondo

D.Y.

Shoji

Honda

Kitagawa

Kojima

and Nakano

, Apoptosis induced by an alkylated purine, 6-dimethylaminopurine, and changes in gene expression in human lymphoma U937 cells, Anticancer Research 28 (2008), 609–620.

48.

Choi

Son

and Baek

J.H.

, Tricarboxylic Acid (TCA) Cycle Intermediates: Regulators of Immune Responses, Life (Basel, Switzerland) 11 (2021), 69.

49.

Huang

Wang

and Peng

, Targeting citrate as a novel therapeutic strategy in cancer treatment, Biochimica et Biophysica Acta. Reviews on Cancer 1873 (2020), 188332.

50.

Buszewska-Forajta

Monedeiro

Gołȩbiowski

Adamczyk

and Buszewski

, Citric acid as a potential prostate cancer biomarker determined in various biological samples, Metabolites 12 (2022), 268.

51.

Seino

Ohta

Suzuki

Sakaki

Sudo

and Nagase

, Metabolomic analysis of ovarian cancer cells treated with paclitaxel, International Journal of Gynecological Cancer: Official Journal of the International Gynecological Cancer Society 28 (2018), 831.

52.

Zhao

Liu

Huang

Gao

Lin

Hoft

D.F.

Shao

and Peng

, Citrate promotes excessive lipid biosynthesis and senescence in tumor cells for tumor therapy, Advanced science (Weinheim, Baden-Wurttemberg, Germany) 9 (2022), e2101553.

53.

Park

Shin

Kim

T.H.

Kim

D.H.

and Lee

, Plasma metabolites as possible biomarkers for diagnosis of breast cancer, PloS One 14 (2019), e0225129.

54.

Nieman

K.M.

Kenny

H.A.

Penicka

C.V.

Ladanyi

Buell-Gutbrod

Zillhardt

M.R.

Romero

I.L.

Carey

M.S.

Mills

G.B.

Hotamisligil

G.S.

Yamada

S.D.

Peter

M.E.

Gwin

and Lengyel

, Adipocytes promote ovarian cancer metastasis and provide energy for rapid tumor growth, Nature Medicine 17 (2011), 1498–1503.

55.

Juraszek

Czarnecka-Herok

and Nałȩcz

K.A.

, Glioma cells survival depends both on fatty acid oxidation and on functional carnitine transport by SLC22A5, Journal of Neurochemistry 156 (2021), 642–657.

56.

Yazawa

Madiyalakan

Izawa

Asao

Furukawa

and Matta

K.L.

, Cancer-associated elevation of alpha(1–3)-L-fucosyltransferase activity in human serum, Cancer 62 (1988), 516–520.

57.

Keeley

T.S.

Yang

and Lau

, The diverse contributions of fucose linkages in cancer, Cancers 11 (2019), 1241.

58.

Zhang

Yin

Fan

Zhang

Zhao

Zhang

Hou

Zhou

Lou

and Li

, Discrimination between malignant and benign ovarian tumors by plasma metabolomic profiling using ultra performance liquid chromatography/mass spectrometry, Clinica Chimica Acta; International Journal of Clinical Chemistry 413 (2012), 861–868.

59.

Yang

Jiang

Sun

Hou

Sun

Zhong

Wang

and Sun

, Identification of Potential Biomarkers and Metabolic Profiling of Serum in Ovarian Cancer Patients Using UPLC/Q-TOF MS, Cellular Physiology and Biochemistry: International Journal of Experimental Cellular Physiology, Biochemistry, and Pharmacology 51 (2018), 1134–1148.

60.

Sun

Rew

Gribble

Bartberger

M.D.

Beck

H.P.

Canon

Chen

Chow

Deignan

Duquette

Eksterowicz

Fisher

Fox

B.M.

Gonzalez

A.Z.

Gonzalez-Lopez De Turiso

Houze

J.B.

Huang

Jiang

Jin

Kayser

Liu

J.J.

M.C.

Long

A.M.

Lucas

McGee

L.R.

McIntosh

Mihalic

Oliner

J.D.

Osgood

Peterson

M.L.

Roveto

Saiki

A.Y.

Shaffer

Toteva

Wang

Y.C.

Wortman

Yakowec

Yan

Zhao

Zhou

Zhu

Olson

S.H.

and Medina

J.C.

, Discovery of AMG 232, a potent, selective, and orally bioavailable MDM2-p53 inhibitor in clinical development, Journal of Medicinal Chemistry 57 (2014), 1454–1472.

61.

Adhikari

Liu

Burton

Mockabee-Macias

Lester

D.K.

and Lau

, L-fucose, a sugary regulator of antitumor immunity and immunotherapies, Molecular Carcinogenesis 61 (2022), 439–453.

62.

Knottnerus

S.J.G.

Bleeker

J.C.

Wüst

R.C.I.

Ferdinandusse

I.J.

Wijburg

F.A.

Wanders

R.J.A.

Visser

and Houtkooper

R.H.

, Disorders of mitochondrial long-chain fatty acid oxidation and the carnitine shuttle, Reviews in Endocrine & Metabolic Disorders 19 (2018), 93–106.

63.

Bachmayr-Heyda

Aust

Auer

Meier

S.M.

Schmetterer

K.G.

Dekan

Gerner

and Pils

, Integrative systemic and local metabolomics with impact on survival in high-grade serous ovarian cancer, Clinical Cancer Research: An Official Journal of the American Association for Cancer Research 23 (2017), 2081–2092.

64.

Horala

Plewa

Derezinski

Klupczynska

Matysiak

Nowak-Markwitz

and Kokot

Z.J.

, Serum free amino acid profiling in differential diagnosis of ovarian tumors – a comparative study with review of the literature, International Journal of Environmental Research and Public Health 18 (2021).

65.

Hishinuma

Shimada

Matsukawa

Saigusa

Kudo

Tsuji

Shigeta

Tokunaga

Kumada

Komine

Shirota

Aoki

Motoike

I.N.

Yasuda

Kinoshita

Yamamoto

Koshiba

and Yaegashi

, Wide-targeted metabolome analysis identifies potential biomarkers for prognosis prediction of epithelial ovarian cancer, Toxins 13 (2021).

66.

Wang

Yin

Yang

Zhang

Gabra

and Li

, Identification of proteomic and metabolic signatures associated with chemoresistance of human epithelial ovarian cancer, International Journal of Oncology 49 (2016), 1651–1665.

Identification of metabolic biomarkers for diagnosis of epithelial ovarian cancer using internal extraction electrospray ionization mass spectrometry (iEESI-MS)

Abstract

BACKGROUND:

OBJECTIVE:

METHODS:

RESULTS:

CONCLUSIONS:

Keywords

1. Introduction

2. Materials and methods

2.1 Sample information

2.1.1 The enrolled EOC patients

Table 1 Clinical information related to patients’ plasma samples

2.2 Materials and chemicals

2.3 Preparation of octadecylsilyl using the organic imported nylon 66 needle type filter

2.4.1 SPE based on ODS coupled with iEESI-MS

2.4.2 Analysis using HPLC-MS technology

2.5 Data preprocessing

3. Results and discussion

3.1 Evaluation of the capability of the iEESI-MS and the HPLC-MS based on the analysis of metabolomic data using OPLS-DA

3.3 PLS-DA recognition score map of the IEESI-MS mass spectrometry data (including self-configured plasma samples)

Table 2 Potential biomarkers for distinguishing P and N groups by analysis of data from IEESI-MS

Table 3 Potential biomarkers for distinguishing P and N groups by HPLC-MS

3.6 Identification of the potential markers for discriminating the groups of the EOC patients and normal subjects

Table 4 The results of the enrichment analysis of the differential metabolites

3.8.1 Lipid metabolism

3.8.2 Amino acid metabolism

3.8.3 Cell division and apoptosis

3.9 Substances related to metabolism

3.9.1 Citric acid

3.9.2 Decanoylcarnitine, hexanoylcarnitine, and L-octanoylcarnitine

3.9.3 L-fucose

3.9.4 2-piperidone

3.9.5 The differential metabolites related to energy metabolism

3.10 Comparison between our study and other studies

3.11 The clinical application of our findings in future

4. Limitations and future continuation of this study

5. Conclusion

Authors’ contributions

Supplementary data

sj-docx-1-cbm-10.3233_CBM-220250.docx - Supplemental material

Footnotes

Acknowledgments

Conflict of interest

References

Table 1
Clinical information related to patients’ plasma samples

Table 2
Potential biomarkers for distinguishing P and N groups by analysis of data from IEESI-MS

Table 3
Potential biomarkers for distinguishing P and N groups by HPLC-MS

Table 4
The results of the enrichment analysis of the differential metabolites