Sage Journals: Discover world-class research

Abstract

BACKGROUND:

Breast cancer is the most worldwide commonly found malignancy among women. The evidence for lipidomic studies of breast cancer in the Chinese population is relatively limited.

OBJECTIVE:

Our current study aimed to identify peripheral lipids capable of distinguishing adults with and without malignant breast cancer in a Chinese population and to explore the potential lipid metabolism pathways implicated in breast cancer.

METHODS:

Lipidomics was performed with an Ultimate 3000 UHPLC system coupled with a Q-Exactive HF MS platform by using the serum of 71 female patients with malignant breast cancer and 92 age-matched ( $\pm$ 2 years) healthy women. The data were uploaded to and processed by the specialized online software Metaboanalyst 5.0. Both univariate and multivariate analyses were carried out for potential biomarker screening. Areas under the receiver-operating characteristic (ROC) curves (AUCs) of identified differential lipids were obtained for evaluating their classification capacity.

RESULTS:

A total of 47 significantly different lipids were identified by applying the following criteria: false discovery rate-adjusted P < 0.05, variable importance in projection $\geqslant$ 1.0, and fold change $\geqslant$ 2.0 or $\leqslant$ 0.5. Among them, 13 lipids were identified as diagnostic biomarkers with the area under curve (AUC) greater than 0.7. Multivariate ROC curves indicated that AUCs greater than 0.8 could be achieved with 2–47 lipids.

CONCLUSIONS:

Using an untargeted LC-MS-based metabolic profiling approach, our study provides preliminary evidence that extensive dysregulations of OxPCs, PCs, SMs and TAGs were involved in the pathological processes of breast cancer. We provided clues for furtherly investigating the role of lipid alterations in the pathoetiology of breast cancer.

Keywords

Lipidomics breast cancer biomarker LC-MS metabolism

1. Introduction

Breast cancer is the most worldwide commonly found malignancy among women, which is characterized by complex interactions of genomic and physiological at the tumor site [1, 2]. Due to the lack of effective treatments, aggressive breast cancer is still the main cause of cancer-related deaths among females [3]. With the increasing incidence and mortality both in the clinical setting and in population-based screening during recent decades [4], the burden of breast cancer has grown rapidly [5].

Although the underlying molecular mechanisms of breast cancer have been explored by several publications, the etiology is still not completely understood and causal pathways have not been well delineated [6]. In recent years, with the continuous progress of detection technology, the exploration of the molecular mechanism of breast cancer has been further developed. Among them, metabolites or lipids are the final downstream product of metabolic reactions, which represent the level of small molecules (metabolites and lipids, less than 1200 Da) [7]. Lipidomic analyses, as the endpoint of the ‘omics’ cascade, could provide more global lipid status of a biological specimen compared to the genome, transcriptome, or proteome [8]. Recent evidence showed that abnormalities of lipids may influence the cellular processes, which could result in tumor development [9, 10]. Several studies demonstrated that the biomolecular processes in patients with cancers differ from normal individuals and a variety of cancers share common metabolic features. Thus, deregulated energy metabolism is considered one of the established hallmarks of cancer [11].

Several targeted lipid analyses have explored the association between specific lipids and breast cancer. For example, a genetic study reported that 16 lipid metabolism-related genes were associated with the process of lipid metabolism and fatty acid binding in breast cancer [12]. A vitro experimental study found that lipids palmitic acid (PA), ceramide (Cer), and docosahexaenoic acid (DHA) have anti-tumor efficacies and reduced the proliferative, migratory and invasive properties of breast cancer cells, which indicated that lipid metabolism targeting drugs have anti-tumor potential targeting breast cancer [13, 14]. Meanwhile, several untargeted lipidomic approaches have explored lipid profiles in patients with breast cancer [6, 10]. Previous untargeted lipidomic studies have found that many different lipids are in breast cancer and controls, and the differences between different breast cancer subtypes [15, 16, 17].

However, due to the diversity of detection instruments and technical methods, the results of different studies lack comparability. The evidence for lipidomic studies of breast cancer in the Chinese population is relatively limited. In this current study, an Ultimate 3000 UHPLC system coupled with a Q-Exactive HF MS platform, in conjunction with multivariate statistical analysis, was used to screen and determine potential serum lipidome among patients with breast cancer and compare it with the lipidome of healthy controls. Herein, we aimed to identify peripheral lipids capable of distinguishing adults with and without breast cancer in a Chinese population and to explore the potential lipid metabolism pathways implicated in breast cancer.

2. Methods

2.1 Ethical approval

The research was reviewed and approved by the Peking University Third Hospital in Beijing, China (Approval no. IRB-2018-413-02) and has been performed with the ethical standards as laid down in the 1989 Declaration of Helsinki and its later amendments or comparable ethical standards. All the participants were fully informed of the objectives of the study and signed the informed consent to before their inclusion.

2.2 Study population

The clinical data and serum samples of 71 female patients who were histopathological diagnoses of malignant breast cancer and were treated at Peking University Third Hospital in Beijing, China were retrospectively collected. The histopathological diagnoses of the malignant groups were based on ultrasound-guided breast biopsies by trained breast radiologists. The cell morphology was used to determine the malignant and benign cancers. The inclusion criteria of patients were as follows: (I) female breast cancer patients undergoing surgery and chemotherapy in Peking University Third Hospital; (II) aged 18–60 years; (III) with complete clinicopathologic data. At the same time, the clinical data and serum samples of 92 age-matched ( $\pm$ 2 years) healthy women with no evidence of any personal or family story of breast cancer were also collected at the same place. The exclusion criteria of all subjects included the following: (I) with severe liver, kidney, or endocrine diseases or any other severe or unstable medical illness; (II) infection in the past year; (III) with abnormal thyroid function; and (VI) pregnant or lactating women.

2.3 Serum samples preparation and lipid extraction

Fasting blood samples were collected in serum separation tubes at the time of pre-puncture coagulation function and pre-operative immune tests from the forearm vein. Serum samples were separated by centrifugation at 3000 $\times$ g for 10 min at 4 ${}^{\circ}$ C, and then the obtained serum was aliquoted into a labeled 1.5-ml Eppendorf vial and then subsequently stored at $-$ 80 ${}^{\circ}$ C until use. The liquid-liquid extraction method was used to extract lipids from the serum samples. In brief, lipids were extracted from plasma samples using protein precipitation as follows: 50 $\mu$ L of serum was extracted by 300 $\mu$ L of cold methanol. The mixture was vortexed for 5 min at 1200 rpm and stored at $-$ 20 ${}^{\circ}$ C for 1 hour. Then the mixture was centrifuged at 13,000 $\times$ g for 20 min at 4 ${}^{\circ}$ C. The supernatant was collected and evaporated at room temperature under a vacuum. The evaporated samples were stored at $-$ 80 ${}^{\circ}$ C until LC-MS analysis. In addition, equal volumes of the supernatant from healthy control samples were pooled to prepare quality control (QC) samples.

2.4 LC-MS data acquisition

Global metabolomic analysis was performed on an Ultimate 3000 UHPLC system coupled with Q-Exactive HF MS (Thermo Scientific). Chromatographic separation was performed on a phase reversed-phase X-select CSH C18 column (2.1 mm $\times$ 100 mm, 2.5 $\mu$ m, Waters, USA) at 40 ${}^{\circ}$ C. Mobile phase A was 5:95 acetonitrile: water (v/v) with 5 mM ammonium formate and 0.1% formic acid; mobile phase B was 95:5 acetonitrile: water (v/v) with 5 mM ammonium formate and 0.1% formic acid. The gradient was 0 min, 2% B; 1 min, 2% B; 18 min, 100% B; 22 min, 100% B, 22.1 min, 2% B; 26 min, 2% B. The column flow rate was 0.3 mL/min, autosampler temperature was 8 ${}^{\circ}$ C. The samples were suspended with 50 $\mu$ L of acetonitrile: water (1:1, v/v) solution and the injection volume was 10 $\mu$ L. Data-dependent acquisition (DDA) based metabolomic data acquisition was performed using the Q-Exactive HF MS (Thermo Scientific). The acquisition was performed in positive-negative ion switching mode in TOP 10 mode. Samples were analyzed in random order. QC samples were analyzed 10 times at the beginning of the batch analyses and once every 10 samples.

Raw data collected from the DDA-MS were processed on MS-DIAL software (http://prime.psc.riken.jp/ compms/msdial/main.html) according to the user guide. Briefly, the raw MS data were converted from the vendor file format (.raw) into the common file format of Reifycs Inc. (.abf) using the Reifycs ABF converter. After conversion, the MS-DIAL software was used for feature detection, spectra deconvolution, metabolite identification and peak alignment between samples. The MS/MS spectra-based metabolite identification was performed in MS-DIAL by searching the acquired MS/MS spectra against the MassBank database provided by MS-DIAL software, containing MS1 and MS/MS information of metabolites (8068 records in positive ion mode and 4782 records in negative ion mode). The MS/MS spectra-based lipid identification was performed in MS-DIAL by searching the acquired MS/MS spectra against the software’s internal in silico MS/MS spectra database (version: Lipid-DBs-VS23-FiehnO), which includes MS1 and MS/MS information of common lipid species. The tolerance for MS1 and MS/MS search were set to 0.01 Da and 0.05 Da, separately. Other parameters used in MS-DIAL were set as default. Processed data were exported in Excel format for further statistical analysis. The Dataset contains m/z values, retention time, metabolite name, and peak area.

2.5 Statistical analysis

Figure 1.

Metabolic analysis of serum samples from patients with breast cancer and healthy controls. (a) Overview of PLS-DA top 4 components. (b) 2D PLS-DA scores plot of top 2 components. Notes: The values of the abscissa and ordinate reflect the variability of the sample in different principal component directions, while the positive and negative signs represent the positional relationship of the sample in the principal component direction. The explained variances of each component are shown in the cells on the diagonal of Fig. 1(a) and brackets in Fig. 1(b).

The data were uploaded to and processed by the specialized online software Metaboanalyst 5.0 (https:// www.metaboanalyst.ca/). Both univariate and multivariate analyses were carried out. The peak area of each lipid was normalized to the total peak areas of all lipids for each sample. Briefly, the raw data were pre-processed by removing features with $>$ 50% missing values, and normalization by median, log transformation, and auto-scaling. After successive alignment and normalization of the matrix data, a two-tailed Mann-Whitney U-test was used to determine the differences between the two groups, and an FDR-adjusted $p$ -value of less than 0.05 was considered significant. The multivariate analysis consisted of the representation of fold change, volcano plot, principal component analysis (PCA), and partial least squares-discriminant analysis (PLS-DA) was also applied to distinguish significant differences in lipids between patients and healthy controls. For pathway enrichment analysis, dysregulated lipids were used as input to search against the sub-chemical class lipid sets, and a $p$ -value of less than 0.05 was considered significant. Moreover, areas under the receiver-operating characteristic (ROC) curves (AUCs) of identified differential lipids were obtained for the evaluation of their classification capacity, which was used to increase the specificity and sensitivity of biomarkers. The molecules identified were ranked according to their sensitivity/specificity. Univariate and multivariate ROC curve analyses were performed using GraphPad Prism 8.0.

3. Results

3.1 Demographic and clinical characteristics of study subjects

Table 1
Basic information of the included participants

Variable	Breast cancer ( $n=$ 71)	HCs ( $n=$ 92)	$p$ -values
Age, year, Mean $\pm$ SD	51.92 $\pm$ 14.33	47.11 $\pm$ 14.37	0.036 ${}^{\text{a}}$
Histologic grade ( $n$ , %)
I	8 (11.3)
II	41 (57.8)
III	22 (30.1)
Menopause status ( $n$ , %)
Yes	37 (52.1)	41 (44.6)	0.339 ${}^{\text{b}}$
No	34 (47.9)	51 (55.4)
Drinker ( $n$ , %)	6 (8.5)	14 (15.2)	0.192 ${}^{\text{b}}$
Smoker ( $n$ , %)	2 (2.8)	3 (3.3)	0.621 ${}^{\text{c}}$
BMI, kg/m ${}^{2}$ , Mean $\pm$ SD	23.78 $\pm$ 2.52	22.95 $\pm$ 3.20	0.076 ${}^{\text{a}}$

${}^{\text{a}}$ Difference between the patients and HCs was determined using a two-tailed independent-samples $t$ -test. ${}^{\text{b}}$ Pearson’s chi-squared test. ${}^{\text{c}}$ Fisher’s exact test.

Figure 2.

ROC analysis revealed candidate lipid biomarkers for breast cancer diagnosis. (a), (b) and (c) ROC curves of PC 56:15, DGCC 50:14, and DGDG 12:0-20:1. The left side of each figure represents the distribution of the ROC curve for differentiating the breast cancer group from the health control group, and the right side is the difference between the features of the two groups. (d) Multivariate ROC curve constructed with 2–47 lipids based on the cross-validation (CV) performance. Each curve represents the potential of top 2, top 3, top 5, top 10, top 20, and total 47 features in differentiating the breast cancer group from the health control group. (e) The predicted class probabilities (average of the cross-validation) for each sample using the 5 features model of lipids. (f) The corresponding predictive accuracy of each PLS-DA model is constructed with different numbers of features. The predictive accuracy of 2 to 47 features is from 65.1% to 83.5%, respectively.

Table 2

Differentiating lipids between breast cancer and healthy control groups identified from the lipidomic data

Lipids	FC	log2(FC)	p.ajusted	VIP	Lipids	FC	log2(FC)	p.ajusted	VIP
Breast cancer $>$ healthy controls					Healthy controls $>$ breast cancer
DAG 22:2e/22:5	3.260	1.705	4.37E-03	1.391	CE 22:6	0.481	$-$ 1.057	3.67E-03	1.431
DGCC 50:14	2.988	1.579	1.01E-13	3.092	DAG 13:1-16:3	0.406	$-$ 1.299	5.90E-05	1.987
DGDG 12:0-20:1	2.330	1.220	5.80E-16	3.290	PC 37:1	0.473	$-$ 1.080	2.66E-03	1.484
OxPC 16:0-18:2+1O	2.006	1.004	1.31E-03	1.597	PC 38:9e	0.328	$-$ 1.608	7.88E-03	1.289
OxPC 16:0-18:2+2O	2.877	1.524	1.35E-02	1.185	PC 39:4	0.477	$-$ 1.067	1.59E-03	1.573
OxPC 16:0-20:3+1O	2.418	1.274	2.22E-03	1.519	PC 39:5	0.453	$-$ 1.143	1.18E-03	1.626
OxPC 16:0-20:4+2O	2.704	1.435	1.26E-03	1.605	PC 41:6	0.427	$-$ 1.228	1.08E-03	1.641
OxPC 18:0-18:2+1O	2.317	1.212	9.40E-04	1.664	PE 17:1e/20:4	0.457	$-$ 1.130	9.61E-04	1.658
OxPC 18:0-18:2+1O	2.034	1.024	1.72E-03	1.559	TAG 16:0e-16:0-16:1	0.337	$-$ 1.569	2.41E-02	1.077
OxPC 18:0-20:3+1O	2.193	1.133	1.68E-03	1.564	TAG 17:0-18:1-22:5	0.459	$-$ 1.124	5.71E-04	1.734
OxPC 18:0-20:4+2O	3.137	1.649	2.73E-04	1.828	TAG 17:0-18:2-22:6	0.470	$-$ 1.089	2.35E-04	1.844
OxPC 18:1-20:4+1O	2.010	1.007	3.67E-03	1.430	TAG 18:1-22:6-22:6	0.389	$-$ 1.362	1.56E-02	1.159
PC 34:1	4.549	2.186	7.40E-03	1.299	TAG 18:1e-16:0-18:0	0.477	$-$ 1.067	9.61E-04	1.658
PC 34:3	2.427	1.279	5.64E-04	1.743	TAG 18:2-22:6-22:6	0.423	$-$ 1.242	2.54E-02	1.066
PC 35:7e	6.954	2.798	2.53E-02	1.067
PC 36:2	2.106	1.074	3.41E-02	1.009
PC 36:3	3.273	1.711	4.63E-03	1.381
PC 37:7	3.402	1.766	1.71E-03	1.560
PC 38:3	2.470	1.305	1.13E-02	1.218
PC 39:6	9.676	3.274	3.05E-02	1.031
PC 39:7	3.883	1.957	3.06E-03	1.461
PC 39:8	2.891	1.532	5.55E-03	1.348
PC 56:15	4.259	2.091	3.55E-09	2.640
PE 34:2	2.858	1.515	2.72E-02	1.054
SM d11:0/34:4	2.995	1.583	4.31E-05	2.017
SM d14:1/21:0	5.198	2.378	4.06E-03	1.406
SM d32:0	27.176	4.764	2.96E-03	1.469
SM d34:0	6.576	2.717	1.26E-03	1.605
SM d34:2	2.270	1.183	2.54E-02	1.065
SM d9:1/38:4	2.567	1.360	1.26E-04	1.907
TAG 12:0-12:0-18:0	2.018	1.013	5.71E-04	1.733
TAG 16:0-16:0-18:1	2.279	1.188	2.98E-04	1.815
TAG 16:0-18:0-18:1	2.294	1.198	6.87E-04	1.704

VIP, variable importance in the projection; FC, fold change.

A total of 71 patients with breast cancer and 92 healthy controls were included in the data analysis. No significant differences in menopause status, smoker, drinker, and BMI were observed between the two groups ( $P>$ 0.05). The age has significant differences between patients with breast cancer and healthy controls (51.92 $\pm$ 14.33 vs. 47.11 $\pm$ 14.37, $P=$ 0.036). Eight (11.3%) of the patients presenting with breast cancer were defined as histologic grade I, 41 (57.8%) were defined as histologic grade II, while the remaining breast cancer subjects ( $n=$ 22, 30.1%) were defined as histologic grade III. The basic information of the included participants is shown in Table 1.

3.2 Serum lipid profiles of patients with breast cancer and healthy controls

There were a total of 1122 lipids detected in positive and negative modes after peak deconvolution, alignment, and picking. We conducted univariate analysis after pre-processing the obtained LC-MS raw data. The independent-sample $t$ -test revealed 514 lipids significantly different between patients with breast cancer and healthy controls (FDR-adjusted $P$ value $<$ 0.05; data not shown).

We combined PCA and OPLS-DA to get a comprehensive view of identifying differences in lipidomic data between patients with breast cancer and healthy controls. According to our PCA scatter plot, the two comparison groups were separated. QC samples were gathered well, proving the reliability of the lipidomic data (Supplemental Fig. 1) The supervised OPLS-DA was applied to obtain a higher level of group separation and a better understanding of the variables responsible for classification. The results of OPLS-DA are shown in Fig. 1. And 470 lipids were identified to distinguish the two comparisons with VIP $\geqslant$ 1.0.

In addition, we also conducted fold change analysis, 62 lipids were reported to be different with fold change $\geqslant$ 2.0 or $\leqslant$ 0.5 (details not shown). Finally, 47 significantly different lipids were identified by applying the following criteria: false discovery rate-adjusted P < 0.05, variable importance in projection $\geqslant$ 1.0, and fold change $\geqslant$ 2.0 or $\leqslant$ 0.5. Among them, 33 lipids were significantly higher, while 14 lipids were significantly lower in patients with breast cancer than in healthy controls (Supplemental Fig. 2). The combined results are shown in Table 2.

3.3 ROC analysis revealed candidate lipid biomarkers for breast cancer

To estimate the diagnostic value of the dysregulated lipids, ROC curves and binary logistic regression models were used to screen markers by examining the AUCs of the biomarkers in the breast cancer group and healthy control group (Fig. 2) According to the works of literature, the biomarkers can be employed as diagnostic biomarkers if the AUC is greater than 0.7. Our ROC analysis revealed 13 lipids with an area under the curve (AUC) greater than 0.7. Multivariate ROC curves constructed with 247 lipids are shown in Fig. 2d, AUCs of greater than 0.8 could be achieved, and the corresponding predictive accuracy of each PLS-DA model constructed with different numbers of features are shown in Fig. 2e and f.

4. Discussions

Using an untargeted LC-MS-based metabolic profiling approach, our study explored the distribution of lipids in patients with breast cancer, and several lipids, including but not limited to PCs, OxPCs, SMs, and TAGs, were found dysregulated between breast cancer and healthy controls, which were considered as potential abnormal biomarkers of the disease.

The most differential lipids reported in our study were PCs. Sixteen PCs were reported dysregulated eleven were up- and 5 were down-regulated in patients with breast cancer. Univariate ROC analysis indicated that the AUC of PC 56:15 was 0.879, which can be considered one of the good diagnostic biomarkers The association between PCs and breast cancer has been reported previously [6, 18]. A targeted lipidomics analysis reported that the total concentration of PCs and PEs of the patient with breast cancer was higher than controls but they decreased significantly following surgery [19] Similar to our finding, Buentzel et al.’s metabolomic profiling study also found several PCs, such as PC ae C40:6, PC aa C38:5, PC ae C34:2, PC ae C32:2, and PC ae C38:3, has good classification potential between breast cancer and healthy controls [20]. Moreover, another study reported that PC 32:0, PC 34:1, and PC 36:2 were overexpressed in HER2-positive breast cancer compared to adjacent normal tissue [21]. The distribution of PCs in breast cancer patients with different pathological characteristics deserves further investigation.

Various OxPC molecules are formed by the oxidation of PC molecules at unsaturated sites and have different biological activities from the parent molecules [22]. Recent evidence suggested that OxPC molecules can be considered end-product markers of oxidative stress and a novel class of bioactive lipids in human pathophysiology [23, 24, 25]. Our current study identified 9 OxPCs were significantly higher in patients with breast cancer than in healthy controls. However, we did not find much evidence to explain our current findings. A recently published study demonstrated that OxPCs could induce autophagic flux in cancer cells and increase the metastatic potential of cancer cells [26]. An OxPC named 1-palmitoyl-2-(5-oxovaleroyl)-sn-glycerol-3-phosphocholine (POVPC) was found to increase the migration and invasion of human hepatocellular carcinoma cells (HepG2) and human breast cancer cells (MCF7) by inducing epithelial-mesenchymal transition (EMT) [26]. Moreover, increased OxPCs were also reported in other diseases, such as neurodegeneration [27]. Thus, the abnormal alteration and potential pathways of PCs and OxPCs in breast cancer are worthy of further exploration.

It has been reported that SMs are the most abundant sphingolipid and important constituents of lipid rafts in normal cells [28], they are involved in the regulation of numerous signaling pathways [29]. There are conflicting reports on the association between SMs and malignancy. Reduced levels of SMs were observed in colon cancer tissues, A549 and U118 cancer cell lines [30, 31]. Consistent with our findings, Nagahashi and colleagues reported higher levels in the human breast cancer tissue compared to the normal breast tissue [32]. Bhadwal et al. did not find significant alteration of SM (18:2/22:0), SM (40:1) and SM (18:0/24:2) in breast tumor tissues [33]. The consistent increase of the six SMs found in our breast cancer samples is complementary to previous studies. In addition, based on the important roles of SMs in various signaling pathways, our findings are also worthy of further targeted validation

The abnormal changes in glycerolipids were also reported in breast cancer. Among them, 9 TAGs and 2 DAGs showed inconsistently significant changes. TAG is a necessary energetic substrate for cell membrane formation. DAG is a precursor to TAG [34]. Both of them are crucial to the metabolism of structural and signaling lipids [35]. Similar to the inconsistent alterations in our study, the changes of TAG in breast cancer also did not consistently in previous publications [36, 37, 38]. Interestingly, Dai et. al. found that triple-negative breast cancer patients with a high TAG level had shorter overall survival and disease-free survival [39]. Mika and colleagues found that colorectal cancer tissues had significantly lower TAG content than normal tissue, and attributed it to the result of enhanced oxidation [40]. Therefore, future research can focus on investigating the effect of different TAGs on the prognosis of breast cancer, potential pathway and mechanisms. Additionally, our lipidomics study also reported the difference between DGCC 50:14 and DGDG 12:0-20:1 between breast cancer and healthy controls, and both of them had good classification performance for the two groups. DGDG has been pointed out to have anti-inflammatory and cancer benefits [41]. However, the previous evidence is too limited, and our current findings of non-targeted lipidomics need to be furtherly verified.

4.1 Limitations

Some limitations should be noted in the present study. Firstly, this study is based on a single-institutional database, there may be selection bias in the study subjects. Secondly, due to the limited evidence, some of our findings have not found a reasonable explanation The detailed mechanisms and the biological significance of lipids in patients with breast cancer require further theoretical and experimental exploration. Thirdly, untargeted metabolomics does not enable the quantification of lipids. The putative lipid biomarkers identified in this study should be further validated using targeted metabolomics methods and verified in larger prospective studies.

5. Conclusions

In summary, using an untargeted LC-MS-based metabolic profiling approach, the present study provides preliminary evidence that extensive dysregulation of OxPCs, PCs, SMs and TAGs was involved in the pathological processes of breast cancer. Our current findings provided clues for furtherly investigating the role of lipid alterations in the pathoetiology of breast cancer, as well as several potential biomarker panels for breast cancer diagnosis or prognosis in clinical practice.

Funding

This work was sponsored by the National Natural Science Foundation of China (No. 81701067). The supporting foundation had no involvement in study design, the collection, analysis, or interpretation of data, writing of the report; or in the decision to submit the article for publication.

Availability of data and material

The datasets generated during and/or analyzed during the current study are available from the corresponding author upon reasonable request.

Author contributions

Conception: Li N and Yan LL.

Interpretation or analysis of data: Cao B, and Yang SY.

Preparation of the manuscript: Cao B, Yan LL, and Li N.

Revision for important intellectual content: Yan LL and Yang SY.

Supervision: Li N and Yan LL.

Ethics approval

The study protocol was reviewed and approved by the research ethics committee of the Peking University Third Hospital in Beijing, China (IRB-2018-413-02).

Consent to participate

All subjects enrolled in this study were from this hospital and were provided written informed consent before their inclusion.

Consent for publication

All authors agreed to the publication of this manuscript.

Supplementary data

The supplementary files are available to download from http://dx.doi.org/10.3233/CBM-220462.

sj-docx-1-cbm-10.3233_CBM-220462.docx - Supplemental material

Supplemental material, sj-docx-1-cbm-10.3233_CBM-220462.docx

Footnotes

Conflict of interest

The authors declare that there is no conflict of interest.

References

Altinoz

Al Ameri

Qureshi

Boush

Nair

S.C.

and Abdel-Aziz

, Clinicopathological characteristics of gene-positive breast cancer in the United Arab Emirates, Breast 53 (2020), 119–124.

Tsai

C.H.

Hsieh

H.F.

Lai

T.W.

Kung

P.T.

Kuo

W.Y.

and Tsai

W.C.

, Effect of multidisciplinary team care on the risk of recurrence in breast cancer patients: A national matched cohort study, Breast 53 (2020), 68–76.

Pandey

Kumar Arya

Kumar Ramar

Chidambaram

and Rajinikanth

P.S.

, Engineered nanomaterials as an effective tool for HER2+ breast cancer therapy, Drug Discov Today (2022).

Louro

Roman

Posso

Comerma

Vidal

Saladie

Alcantara

Sanchez

Quintana

M.J.

Del Riego

Ferrer

Penalva

Bargallo

Prieto

Sala

and Castells

, Differences in breast cancer risk after benign breast disease by type of screening diagnosis, Breast 54 (2020), 343–348.

Liu

Feng

Wang

Huang

Dai

Zhang

Song

Wang

Zhang

Zheng

Song

and Chen

, Tumor markers CA15-3, CA125, CEA and breast cancer survival by molecular subtype: a cohort study, Breast Cancer 27 (2020), 621–630.

Giallourou

Urbaniak

Puebla-Barragan

Vorkas

P.A.

Swann

J.R.

and Reid

, Characterizing the breast cancer lipidome and its interaction with the tissue microbiota, Commun Biol 4 (2021), 1229.

Dunn

W.B.

Bailey

N.J.

and Johnson

H.E.

, Measuring the metabolome: current analytical technologies, Analyst 130 (2005), 606–25.

Kus

Kij

Zakrzewska

Jasztal

Stojak

Walczak

and Chlopicki

, Alterations in arginine and energy metabolism, structural and signalling lipids in metastatic breast cancer in mice detected in plasma by targeted metabolomics and lipidomics, Breast Cancer Res 20 (2018), 148.

Hahnefeld

Gruber

Schomel

Fischer

Mattjus

Gurke

Beretta

Ferreiros

Geisslinger

and Wegner

M.S.

, Ether lipid and sphingolipid expression patterns are estrogen receptor-dependently altered in breast cancer cells, Int J Biochem Cell Biol 127 (2020), 105834.

10.

Ward

A.V.

Anderson

S.M.

and Sartorius

C.A.

, Advances in Analyzing the Breast Cancer Lipidome and Its Relevance to Disease Progression and Treatment, J Mammary Gland Biol Neoplasia 26 (2021), 399–417.

11.

Pavlova

N.N.

and Thompson

C.B.

, The Emerging Hallmarks of Cancer Metabolism, Cell Metab 23 (2016), 27–47.

12.

Chang

and Xing

, Identification of a novel lipid metabolism-related gene signature within the tumour immune microenvironment for breast cancer, Lipids Health Dis 21 (2022), 43.

13.

Lucht

S.A.

Eliassen

A.H.

Bertrand

K.A.

Ahern

T.P.

Borgquist

Rosner

Hankinson

S.E.

and Tamimi

R.M.

, Circulating lipids, mammographic density, and risk of breast cancer in the Nurses’ Health Study and Nurses’ Health Study II, Cancer Causes Control 30 (2019), 943–953.

14.

Jiang

Zhang

Pan

Yan

Zhang

Weng

Wang

Chen

and Yang

, Potential plasma lipid biomarkers in early-stage breast cancer, Biotechnol Lett 39 (2017), 1657–1666.

15.

Eiriksson

F.F.

Nohr

M.K.

Costa

Bodvarsdottir

S.K.

Ogmundsdottir

H.M.

and Thorsteinsdottir

, Lipidomic study of cell lines reveals differences between breast cancer subtypes, PLoS One 15 (2020), e0231289.

16.

Eghlimi

Shi

Hrovat

and Gu

, Triple Negative Breast Cancer Detection Using LC-MS/MS Lipidomic Profiling, J Proteome Res 19 (2020), 2367–2378.

17.

Purwaha

Piyarathna

D.W.B.

Rajendiran

Ravindran

Omilian

A.R.

Jiralerspong

Das

Morrison

Ambrosone

Coarfa

Putluri

and Sreekumar

, Unbiased Lipidomic Profiling of Triple-Negative Breast Cancer Tissues Reveals the Association of Sphingomyelin Levels with Patient Disease-Free Survival, Metabolites 8 (2018).

18.

His

Viallon

Dossus

Schmidt

J.A.

Travis

R.C.

Gunter

M.J.

Overvad

Kyro

Tjonneland

Lecuyer

Rothwell

J.A.

Severi

Johnson

Katzke

Schulze

M.B.

Masala

Sieri

Panico

Tumino

Macciotta

Boer

J.M.A.

Monninkhof

E.M.

Olsen

K.S.

Nost

T.H.

Sandanger

T.M.

Agudo

Sanchez

M.J.

Amiano

Colorado-Yohar

S.M.

Ardanaz

Vidman

Winkvist

Heath

A.K.

Weiderpass

Huybrechts

and Rinaldi

, Lifestyle correlates of eight breast cancer-related metabolites: a cross-sectional study within the EPIC cohort, BMC Med 19 (2021), 312.

19.

Min

H.K.

Kong

and Moon

M.H.

, Quantitative analysis of urinary phospholipids found in patients with breast cancer by nanoflow liquid chromatography-tandem mass spectrometry: II. Negative ion mode analysis of four phospholipid classes, Anal Bioanal Chem 396 (2010), 1273–80.

20.

Buentzel

Klemp

H.G.

Kraetzner

Schulz

Dihazi

G.H.

Streit

Bleckmann

Menck

Wlochowitz

and Binder

, Metabolomic Profiling of Blood-Derived Microvesicles in Breast Cancer Patients, Int J Mol Sci 22 (2021).

21.

Kim

I.C.

Lee

J.H.

Bang

Choi

S.H.

Kim

Y.H.

Kim

K.P.

Kim

H.K.

and Ro

, Lipid profiles for HER2-positive breast cancer, Anticancer Res 33 (2013), 2467–72.

22.

Solati

and Ravandi

, Lipidomics of Bioactive Lipids in Acute Coronary Syndromes, Int J Mol Sci 20 (2019).

23.

Stamenkovic

Pierce

G.N.

and Ravandi

, Oxidized lipids: not just another brick in the wall (1), Can J Physiol Pharmacol 97 (2019), 473–485.

24.

Solati

Surendran

Edel

Roznik

Allen

and Ravandi

, Increase in Plasma Oxidized Phosphatidylcholines (OxPCs) in Patients Presenting With ST-Elevation Myocardial Infarction (STEMI), Front Med (Lausanne) 8 (2021), 716944.

25.

Dong

D’Mello

Pinsky

Lozinski

B.M.

Kaushik

D.K.

Ghorbani

Moezzi

Brown

Melo

F.C.

Zandee

Prat

Whitehead

S.N.

and Yong

V.W.

, Oxidized phosphatidylcholines found in multiple sclerosis lesions mediate neurodegeneration and are neutralized by microglia, Nat Neurosci 24 (2021), 489–503.

26.

Seok

J.K.

Hong

E.H.

Yang

Lee

H.E.

Kim

S.E.

Liu

K.H.

Kang

H.C.

Cho

Y.Y.

Lee

H.S.

and Lee

J.Y.

, Oxidized Phospholipids in Tumor Microenvironment Stimulate Tumor Metastasis via Regulation of Autophagy, Cells 10 (2021).

27.

Dong

and Yong

V.W.

, Oxidized phospholipids as novel mediators of neurodegeneration, Trends Neurosci 45 (2022), 419–429.

28.

Mullen

T.D.

Spassieva

Jenkins

R.W.

Kitatani

Bielawski

Hannun

Y.A.

and Obeid

L.M.

, Selective knockdown of ceramide synthases reveals complex interregulation of sphingolipid metabolism, J Lipid Res 52 (2011), 68–77.

29.

Barcelo-Coblijn

Martin

M.L.

de Almeida

R.F.

Noguera-Salva

M.A.

Marcilla-Etxenike

Guardiola-Serrano

Luth

Kleuser

Halver

J.E.

and Escriba

P.V.

, Sphingomyelin and sphingomyelin synthase (SMS) in the malignant transformation of glioma cells and in 2-hydroxyoleic acid therapy, Proc Natl Acad Sci U S A 108 (2011), 19569–74.

30.

Lou

Liu

Hou

Kabir

Liu

Ding

Dong

Chen

Bui

H.H.

Roth

Cao

and Jiang

X.C.

, 2-Hydroxy-oleic acid does not activate sphingomyelin synthase activity, J Biol Chem 293 (2018), 18328–18336.

31.

Selzner

Bielawska

Morse

M.A.

Rudiger

H.A.

Sindram

Hannun

Y.A.

and Clavien

P.A.

, Induction of apoptotic cell death and prevention of tumor growth by ceramide analogues in metastatic human colon cancer, Cancer Res 61 (2001), 1233–40.

32.

Nagahashi

Tsuchida

Moro

Hasegawa

Tatsuda

Woelfel

I.A.

Takabe

and Wakai

, High levels of sphingolipids in human breast cancer, J Surg Res 204 (2016), 435–444.

33.

Bhadwal

Dahiya

Shinde

Vaiphei

Math

R.G.H.

Randhawa

and Agnihotri

, LC-HRMS based approach to identify novel sphingolipid biomarkers in breast cancer patients, Sci Rep 10 (2020), 4668.

34.

Zhang

and Li

, Effect of diacylglycerol on postprandial serum triacylglycerol concentration: a meta-analysis, Lipids 44 (2009), 161–8.

35.

Heier

and Kuhnlein

R.P.

, Triacylglycerol Metabolism in Drosophila melanogaster, Genetics 210 (2018), 1163–1184.

36.

Fan

Ding

Wang

Yuan

Zhang

and Xu

, Decreased serum HDL at initial diagnosis correlates with worse outcomes for triple-negative breast cancer but not non-TNBCs, Int J Biol Markers 30 (2015), e200–7.

37.

Liu

and Gao

, Serum Lipids and Breast Cancer Risk: A Meta-Analysis of Prospective Cohort Studies, PLoS One 10 (2015), e0142669.

38.

Berrino

Villarini

Traina

Bonanni

Panico

Mano

M.P.

Mercandino

Galasso

Barbero

Simeoni

Bassi

M.C.

Consolaro

Johansson

Zarcone

Bruno

Gargano

Venturelli

and Pasanisi

, Metabolic syndrome and breast cancer prognosis, Breast Cancer Res Treat 147 (2014), 159–65.

39.

Dai

Chen

Wang

Tang

Zhao

Xie

and Wei

, Pretreatment TG/HDL-C Ratio Is Superior to Triacylglycerol Level as an Independent Prognostic Factor for the Survival of Triple Negative Breast Cancer Patients, J Cancer 7 (2016), 1747–1754.

40.

Mika

Pakiet

Czumaj

Kaczynski

Liakh

Kobiela

Perdyan

Adrych

Makarewicz

and Sledzinski

, Decreased Triacylglycerol Content and Elevated Contents of Cell Membrane Lipids in Colorectal Cancer Tissue: A Lipidomic Study, J Clin Med 9 (2020).

41.

Gao

Q.M.

Xia

Shine

M.B.

Wang

Navarre

Kachroo

and Kachroo

, Mono- and digalactosyldiacylglycerol lipids function nonredundantly to regulate systemic acquired resistance in plants, Cell Rep 9 (2014), 1681–1691.

Comprehensive serum lipidomic analyses reveal potential biomarkers for malignant breast cancer: A case-control study 1

Abstract

BACKGROUND:

OBJECTIVE:

METHODS:

RESULTS:

CONCLUSIONS:

Keywords

1. Introduction

2. Methods

2.1 Ethical approval

2.2 Study population

2.3 Serum samples preparation and lipid extraction

2.4 LC-MS data acquisition

2.5 Statistical analysis

3.1 Demographic and clinical characteristics of study subjects

Table 1 Basic information of the included participants

3.3 ROC analysis revealed candidate lipid biomarkers for breast cancer

4. Discussions

4.1 Limitations

5. Conclusions

Funding

Availability of data and material

Author contributions

Ethics approval

Consent to participate

Consent for publication

Supplementary data

sj-docx-1-cbm-10.3233_CBM-220462.docx - Supplemental material

Footnotes

Conflict of interest

References

Table 1
Basic information of the included participants