Introduction

Lipids are hydrophobic or amphiphilic small molecules that play crucial roles in cellular energy storage, structure, and signaling [1]. According to their structures, they can be divided into eight categories, and then into different classes and molecular species [2]. As changes in lipid metabolism and function have important effects on cellular physiological functions and pathological disorders of living organisms, lipidomics has attracted more and more attention in recent years [3]. However, there are still challenges associated with the analytical techniques used in lipidomics because of the extreme structural and content diversity of lipids in real biological samples.

Currently, two main analytical strategies are used in lipidomics; namely, shotgun lipidomics and high-performance liquid chromatography (HPLC)–mass spectrometry (MS) [4]. In general, shotgun lipidomics is considered to be of high throughput, accurate, and highly sensitive. However, discrimination of some lipid isomers is a challenge for this method. Besides, identification of low-abundance or less-ionizable lipids is also difficult because of possible ion suppression [5]. Compared with shotgun lipidomics, HPLC–MS usually provides more comprehensive information about the lipid composition. Because of the chromatographic separation before MS, HPLC–MS can avoid ion suppression to some extent, and it is possible to separate lipid isomers. At present, two main modes of HPLC are used to separate lipids: normal phase liquid chromatography (NPLC) and reversed-phase liquid chromatography (RPLC). NPLC allows the separation of lipids into lipid classes according to their polarity, whereas RPLC is widely used for the separation of individual lipid species according to the carbon number and the double-bond number [6, 7]. Therefore, NP/RP two-dimensional (2D) liquid chromatography (LC)–MS has significant advantages in lipid analysis, as the orthogonal system can enhance the resolving power and provide a higher peak capacity. In our previous work, an online NP/RP 2D LC-MS method was developed on the basis of a solvent evaporation interface for comprehensive lipid profiling, and this could realize the analysis of more than 500 lipid species from 17 lipid classes [8,9,10]. However, the analysis time was rather long (nearly 3 h).

In recent years, supercritical fluid chromatography (SFC) has shown great potential as a comprehensive and high-throughput screening method in different omic fields, including lipidomics. Compared with traditional LC, SFC has met with great favor in lipid profiling because of its high efficiency, low organic solvent consumption, and its unambiguous identification of some isomeric lipids [11]. SFC–MS has been used for targeted lipid profiling, including profiling of phospholipids [12], sphingolipids [13], fatty acids [14], triacylglycerols (TGs) [15], and prenol lipids [16]. As the separation mechanism of SFC using polar stationary phases is similar to that of NPLC [17], the combination of SFC and RPLC could also achieve a high degree of orthogonality and maximize the effective 2D peak capacity. Besides, the analysis time could be significantly reduced compared with that of NP/RP 2D LC–MS method. The potential of RPLC × SFC has been highlighted by Stevenson et al. [18] in off-line mode. Online comprehensive RPLC × SFC was first investigated by Sarrut et al. [19] for bio-oil analysis using traditional sample loops as the interface. More recently, Sun et al. [20] introduced a trapping column interface for the analysis of depolymerized lignin samples. Besides, selective comprehensive 2D RPLC × SFC was also investigated for the analysis of chiral pharmaceutical compounds [21]. According to the classification and structure of lipids, SFC × RPLC is more suitable for the analysis of lipids. The potential of SFC as the first dimension in an online RPLC × SFC system with a trapping column interface was demonstrated by François and Sandra [22]. However, only fatty acids were analyzed.

In the present study, we aimed to develop a novel 2D SFC/RPLC—triple-quadrupole (QQQ) MS method for analysis of lipids, including monoacylglycerol (MG), diacylglycerol (DG), TG, ceramide (Cer), glucosylceramide (GluCer), galactosylceramide (GalCer), lactosylceramide (LacCer), sphingomyelin (SM), lysophosphatidylethanolamine (LPE), phosphatidylcholine (PC), and acylcarnitine (AC), in complex biological samples based on a ten-port, two-position vacuum evaporation interface. The potential of this lipidomic method is demonstrated by the analysis of lipid extract from human plasma.

Materials and methods

Reagents and materials

Chloroform (HPLC grade) was purchased from Mallinckrodt Baker (Phillipsburg, NJ, USA). HPLC grade n-hexane, 2-propanol, and methanol as well as ammonium formate (purity 99.995% or greater), were obtained from Sigma-Aldrich (St Louis, MO, USA). Purified water was acquired from Hangzhou Wahaha Group Co. (Zhejiang, China). All lipid standards, including 1-(10Z-heptadecenoyl)-sn-glycero-3-phosphoethanolamine [LPE (17:1)], 1,2-dimyristelaidoyl-sn-glycero-3-phosphocholine [PC (14:1/14:1)], N-lauroyl-d-erythro-sphingosine [Cer (d18:1/12:0)], d-galactosyl-β-1,1′-N-octanoyl-d-erythro-sphingosine [GalCer (d18:1/8:0)], d-glucosyl-β-1,1′-N-lauroyl-d-erythro-sphingosine [GluCer (d18:1/12:0)], d-lactosyl-β-1,1′-N-octanoyl-d-erythro-sphingosine [LacCer (d18:1/8:0)], N-(heptadecanoyl)-sphing-4-enine-1-phosphocholine [SM (d18:1/17:0)], 1-heptadecanoyl-rac-glycerol [MG (17:0)], 1,2-dilauroyl-sn-glycerol [DG (12:0/12:0)], 1-(9Z-octadecenoyl)-2-tetradecanoyl-3-(10Z-heptadecenoyl)-sn-glycerol [TG (18:1/14:0/17:1)], and palmitoyl d-carnitine [AC (16:0)] were purchased from Avanti Polar Lipids (Alabaster, AL, USA). Stock solutions of individual lipid standard at a concentration of 1 mg/mL were prepared in Folch solution (chloroform–methanol, 2:1 v/v). Lipid standard mixture contained all lipid standards with a concentration of 1 μg/mL. Calibration solutions were prepared by the dilution of lipid standard mixture in a hexane–2-propanol mixture (1:1 v/v). Human plasma samples from ten patients with breast cancer and ten healthy controls were included in this study. For breast tumor, it could be easily recognized by touch. However, we can't distinguish between benign breast tumor and breast cancer. Thus, the tumor should be confirmed by visual inspection of stained tissue by an experienced pathologist. This study was approved by the Ethics Committee of the Fifth Medical Center of the People's Liberation Army General Hospital (previously the Affiliated Hospital of the Academy of Military Medical Sciences). Lipids were extracted from human plasma by a modified Folch extraction method [23]. Before analysis, the dried lipid extracts were redissolved in 500 μL of Folch solution.

Two-dimensional SFC/RPLC–QQQ MS conditions

Two-dimensional SFC/RPLC system

The 2D SFC/RPLC system was based on a ten-port, two-position valve (Agilent Technologies, Santa Clara, CA, USA) vacuum evaporation interface reported previously [24] with some modification (Fig. 1). The mobile phase from the SFC column in the first dimension was evaporated by a vacuum pump when it flowed through the loops. By automatic switching of the valve, the two loops were used to trap analytes and transfer them to the second dimension alternately, and thus a nonstop-flow 2D SFC/RPLC system was realized. These two loops (0.12-mm inner diameter × 50 mm and 0.02 in. × 100 mm stainless steel pipeline for each loop) were maintained at 50 °C by a homemade electrically heated thermostatic oil bath. An RV3 pump (Edwards, Crawley, UK) was used to evaporate the first-dimension solvent and trap analytes in loops.

Fig. 1
figure 1

The two-dimensional supercritical fluid chromatography (SFC)/reversed-phase (RP) liquid chromatography–mass spectrometry system based on a ten-port, two-position valve. BPR backpressure regulator, MS mass spectrometer

Full size image

The first-dimension SFC was performed with a 1260 series SFC system (Agilent Technologies, Santa Clara CA, USA). The final SFC method used the following conditions: two Zorbax RX-SIL columns (2.1 mm × 150 mm, 5 μm; Agilent Technologies, Santa Clara, CA, USA); column temperature, 40 °C; flow rate, 0.4 mL/min; mobile phase A1, supercritical CO2; mobile phase B1, methanol–water mixture (500:10 v/v) containing 20 mM ammonium formate; injection volume, 5 μL; backpressure regulator (BPR) pressure, 160 bar; BPR temperature, 60 °C. The first-dimension SFC instrument was connected with a valve via a commercial kit (Agilent) composed of T-pieces allowing backpressure control and mixing of column effluent with makeup liquid. Methanol containing 5 mM ammonium formate delivered by a quaternary pump (Agilent) was used as the makeup liquid. In the second dimension, a 1290 II series RPLC system consisting of a binary pump and a thermostated column compartment (Agilent Technologies, Santa Clara, CA, USA) was utilized. The second-dimension separation was performed on a Poroshell 120 EC C8 column (2.1 mm × 50 mm, 2.7 μm; Agilent Technologies, Santa Clara, CA, USA). The column temperature was 40 °C and the flow rate was 0.6 mL/min. Mobile phases A2 and B2 were water and methanol, both containing 10 mM ammonium formate. The gradient programs for both dimensions and the valve switching program at the final condition are described in Table 1.

Table 1 Configuration of the online two dimensional supercritical fluid chromatography (SFC)/reversed-phase liquid chromatography (RPLC)–mass spectrometry system
Full size table

Mass spectrometry

In this 2D SFC/RPLC–MS system, the second-dimension RPLC system was coupled online to an Agilent 6470 QQQ mass spectrometer, with the eluate from the second-dimension column flowing directly into the electrospray ionization source. QQQ MS was performed in positive ion mode, and the instrument parameters were set as follows: sheath gas temperature, 325 °C; sheath gas flow rate, 11 L/min; nebulizer pressure, 45 psi; gas temperature, 325 °C; gas flow rate, 6 L/min; capillary entrance voltage, 3500 V; electron multiplier voltage, 300 V; nozzle voltage, 1500 V. Mass spectrometry data were acquired by multiple reaction monitoring (MRM) scan.

Recovery at the interface

To study the recovery rate of lipids at the interface, we used a six-port, two-position valve on the thermostated column compartment before the interface. During the analysis, one injection was directly analyzed by one-dimensional (1D) SFC without trapping by the loops, and another injection was trapped by the loops and then transferred to the SFC column for analysis (see Fig. S1). Finally, the recovery of each lipid standard was calculated by the ratio of the two peak areas.

Method validation

The linearity, limit of detection (LOD), and repeatability of the 2D SFC/RPLC–QQQ MS method were determined. The calibration curves for this method were constructed for 11 lipid standards, including LPE (17:1), PC (14:1/14:1), Cer (d18:1/12:0), GalCer (d18:1/8:0), GluCer (d18:1/12:0), LacCer (d18:1/8:0), SM (d18:1/17:0), MG (17:0), DG (12:0/12:0), TG (18:1/14:0/17:1), and AC (16:0), with concentrations of 0.0001, 0.0002, 0.0005, 0.001, 0.002, 0.005, 0.01, 0.02, 0.05, 0.1, 0.2, 0.5, and 1 μg/mL. The linearity was determined from calibration curves plotted as peak area versus concentration of lipid standard solutions and calculated by linear regression. Each calibration point was measured in quadruplicate. The LOD was determined from a signal-to-noise ratio of 3. For repeatability, the relative standard deviations (RSDs) of the peak area and retention time of each lipid standard were calculated on the basis of six consecutive injections of the same concentration (1 μg/mL) of 11 lipid standards.

Data processing

MS data were processed by the MassHunter Qualitative Analysis B.07.00 (Agilent Technologies, Santa Clara, CA, USA). All MRM chromatograms were obtained with ±0.5 m/z expansion. Peak areas of each individual lipid species were corrected by the peak area of an internal standard corresponding to the same lipid class or in the same fraction for each sample. Then the corrected peak areas were used for statistical analysis by the Mann–Whitney U test by IBM SPSS Statistics, version 20 (IBM, Armonk, NY, USA). Differences were considered to be significant if p was less than 0.05. Besides, the fold change was calculated from the arithmetic means of the healthy control and patient groups. Origin 2016 (OriginLab, Northampton, MA, USA) was used for visualization of lipid recovery under different interface conditions.

Results and discussion

Optimization of MRM transition settings for lipid standards

Identification of the precursor ions and elucidation of the fragmentation patterns of each lipid standard were performed by direct infusion analysis. Precursor ions of each lipid were clearly observed in the positive ion mode by [M + H]+ or [M + NH4]+. Tandem MS (MS/MS) spectra of the individual lipid precursor ions, which could produce signature product ions specific for each molecular substructure, were acquired by collision-induced dissociation with nitrogen. On the basis of the most abundant adduct precursor ion and the selected product ion for each lipid, we optimized the MS fragmentor voltage and collision energy. The optimized acquisition settings, including precursor–product ion pairs, fragmentor voltage, and collision energy, are presented in Table S1.

Optimization of the 2D SFC/RPLC system interface

As mentioned before, a ten-port, two-position valve with two loops was used as the interface of the 2D SFC/RPLC system. Compared with the NP/RP 2D LC system, the combination of SFC and RPLC using a vacuum solvent evaporation interface should be better. As the mobile phase from the first-dimension SFC contains a large amount of supercritical CO2, which will change to CO2 gas after backpressure regulation, the viscosity of the solvent vapor is lower and the pressure at the solution front is higher, leading to a faster evaporation rate. Before the interface was used to trap the first-dimension eluate and transfer it to the second dimension, the conditions for the interface were optimized. The temperature of oil bath, in which the loops were maintained at constant temperature, was studied under three conditions (30, 50, and 70 °C). Besides, different flow rates of the mobile phase (0.2, 0.3 , 0.4, 0.5, and 0.6 mL/min) were also investigated to achieve the greatest recovery of lipids.

According to a previous report [25], raising the temperature and increasing the flow rate of the mobile phase can increase the evaporation rate. A higher evaporation rate is good for the evaporation of the mobile phase, which can increase the recovery at the interface. However, when the evaporation rate is too high, it will cause loss of samples, resulting in a decrease of recovery at the interface. Therefore, the interface recovery initially increases with increase of temperature and flow rate of the mobile phase, but decreases with further increase of them. As shown in Fig. 2, the results of the optimization experiment were consistent with our expectation.

Fig. 2
figure 2

Recovery of diacylglycerol (12:0/12:0) under different interface conditions

Full size image

After optimization, 50 °C and 0.4 mL/min were chosen as the temperature of the oil bath and the flow rate of first-dimension mobile phase, respectively. Under these conditions, the recoveries of 11 lipid standards were all more than 88%, as shown in Table 2.

Table 2 Recovery of 11 lipid standards when the oil bath temperature was 50 °C and the flow rate of first-dimension mobile phase was 0.4 mL/min
Full size table

Two-dimensional SFC/RPLC–MS method development

We aimed to develop a nonstop-flow 2D SFC/RPLC–MS method, in which the first-dimension SFC is used to separate lipid classes by the difference in their polarity, and the second-dimension RPLC is used to separate the lipid molecular species according to their different acyl chains, for the analysis of lipids, including MG, DG, TG, Cer, GluCer, GalCer, LacCer, SM, LPE, PC, and AC, in human plasma. For this purpose, a column with silica stationary phase was selected for lipid class separation in the first-dimension SFC and a short Poroshell 120 EC C8 column was selected for fast separation of lipid molecular species in the second-dimension RPLC. As lipid class separation in the first dimension is particularly important for the 2D method, individual parameters of the SFC separation were carefully optimized, including BPR pressure, column temperature, and flow rates of the mobile phase and the makeup liquid, with a lipid standard mixture.

We found that the retention times of the lipids increased with increase of BPR pressure (Fig. S2). Besides, the influence of column temperature was not consistent for different lipids. As shown in Fig. S3, for lipids with a retention time of less than 12 min, the retention time increased with increase of the column temperature. But for lipids with a retention time of more than 22 min, the retention time decreased with increase of the column temperature. Because of the large increase of the organic phase concentration during analysis, the mobile phase changed from the supercritical fluid to subcritical fluid and then to organic fluid. For conventional LC with an organic liquid as the mobile phase, the retention time of analytes decreases with higher temperature. But for supercritical fluid, higher temperature induces a lower density of the mobile phase, leading to an increase of the retention time [26]. The retention times of the lipids decreased with increase of flow rate of the mobile phase. Besides, the peak width also decreased (Fig. S4). The flow rate of the makeup liquid did not affect the retention times of the lipids, but it had a great influence on the peak shapes. As shown in Fig. S5, the peaks became smoother with a higher flow rate of the makeup liquid. The peak areas were increased with a higher flow rate of the makeup liquid. Because there should be sufficient time left for the second-dimension separation, the optimized conditions for the first-dimension SFC were as follows: BPR pressure, 160 bar; column temperature, 40 °C; flow rate of mobile phase, 0.4 mL/min.

In 1D SFC–MS analysis, SFC effluent is mixed with makeup fluid to ensure the ionization of analytes. As the SFC effluent flows to the interface in the 2D SFC/RPLC system, we studied whether a makeup fluid is necessary with the 2D SFC/RPLC system. Under the optimized first-dimension conditions, the flow rate of the makeup liquid was set as 0.25 and 0 mL/min. As shown in Fig. S6, LPE (17:1), AC (16:0), PC (14:1/14:1), and SM (d18:1/17:0) disappeared when the flow rate of the makeup liquid was 0.25 mL/min. Besides, the peak shape of DG (12:0/12:0) and TG (18:1/14:0/17:1) would be influenced without makeup liquid. Thus, makeup liquid was necessary for the first several fractions, and should be removed in the last several fractions. The optimized flow rate gradient of the makeup liquid is shown in Table 1, under which condition good responses were obtained for all the lipid standards.

Under the optimized conditions, 11 classes of lipids were separated into seven different fractions after first-dimension SFC (Fig. 3). The different classes of sphingolipids, including Cer, GluCer, GalCer, LacCer, and SM, were separated into different fractions, which avoided the disturbance of identification between different head groups of lipids. Although there was still coelution of different classes in the first dimension, such as DG and TG, they could be easily separated by second-dimension RPLC according to significantly different acyl chains. Thus, these results indicate the feasibility of online combination of SFC and RPLC. Compared with the NP/RP 2D LC–MS system, the analysis time is much shorter. Besides, the peak width of lipids after RPLC were much narrower than those after 1D SFC. Lee et al. [13] reported that methylation by trimethylsilyl-diazomethane could suppress peak tailing and increase the detection sensitivity of SFC–MS for diverse polar lipids, which could also be achieved by an additional dimension of RPLC separation.

Fig. 3
figure 3

Multiple reaction monitoring chromatograms of lipid standard mixture investigated by the one-dimensional supercritical fluid chromatography (SFC)–mass spectrometry (MS) method (top) and the two-dimensional SFC/reversed-phase liquid chromatography–MS method (bottom). AC acylcarnitine, Cer ceramide, DG diacylglycerol, GalCer galactosylceramide, GluCer glucosylceramide, LacCer lactosylceramide, LPE lysophosphatidylethanolamine, MG monoacylglycerol, PC phosphatidylcholine, SM sphingomyelin, TG triacylglycerol

Full size image

Validation of the 2D SFC/RPLC–MS method

We validated the method using lipid standard mixture. The results of the analytical validation are summarized in Table 3. From these results, the correlation coefficients (R2) were greater than 0.99 for all 11 lipid standards. Besides, the LODs ranged from 0.05 to 10 ng/mL. By our 2D SFC/RPLC–MS method, the LODs for TG (18:1/14:0/17:1), MG (17:0), and Cer (d18:1/12:0) were 0.1, 0.05, and 0.05 ng/mL, respectively. These sensitivities were 50-fold, tenfold, and tenfold higher, respectively, than those in the NP/RP 2D LC–MS method [8]. For repeatability, the RSDs of the retention time and peak area ranged from 0.00% to 0.10% and from 1.54% to 19.85%, respectively. Among these, the RSDs of the peak area for SM (d18:1/17:0), PC (14:1/14:1), and AC (16:0) were relatively high. But they were still within acceptable limits. Thus, this 2D SFC/RPLC–MS method could be applied for lipid profiling analysis in different biological or clinical samples.

Table 3 Analytical validation of the two-dimensional supercritical fluid chromatography/reversed-phase liquid chromatography–mass spectrometry method by the determination of lipid standards
Full size table

Analysis of human plasma

Lipids in human plasma were separated and identified with use of the 2D SFC/RPLC–MS system. According to the MS/MS fragment patterns of each lipid class and major fatty acid composition, we optimized the MRM transition settings for each lipid species. To be specific, Cer, GluCer, GalCer, and LacCer have a similar backbone, which will yield specific fragment ions, including ions with m/z 266.3, 264.3, and 262.4 for d18:0, d18:1, and d18:2, respectively. PC and SM have the same polar head group, which could generate ions with m/z 184.1. Besides, the specific fragment ion for AC has m/z 85.1. Thus, lipids in these classes were identified by precursor ion scan. For DG and TG, loss of RCOO- will occur during MS/MS. Thus, we conducted a product ion scan to select the MRM transitions for lipids in these two lipid classes. LPE was monitored by the specific loss of an ion with m/z 141.0 in the neutral loss scan. Apart from this, lipid molecular species were also identified by retention times. All the retention times of the lipid species in one class were in the same fraction. In this method, a pair of isomers, GluCer and GalCer, were separated into different fractions, the retention times of which were very useful for determining the lipid species in these two classes. Besides, the relative retention time of lipid species in the same lipid class was an effective support for identification. As shown in Fig. 4a, SMs having the same double-bond number in two fatty acid chains were eluted in order of increasing total carbon number. For eight SMs with one double bond, they were eluted in the order SM (34:1), SM (36:1), SM (37:1), SM (38:1), SM (39:1), SM (40:1), SM (41:1), and SM (42:1). Besides, DGs having the same total carbon number in two fatty acid chains were eluted in the order of decreasing double-bond number. For five DGs with a carbon number of 36, they were eluted in the order DG (36:5), DG (36:4), DG (36:3), DG (36:2), and DG (36:1) (Fig. 4b). In addition, separation of isomers, for example, GluCer (d18:1/23:0) and GluCer (d18:0/23:1), was also observed (Fig. 4c).

Fig. 4
figure 4

Multiple reaction monitoring chromatograms of selected lipid molecular species detected by the two-dimensional supercritical fluid chromatography/reversed-phase liquid chromatography–triple-quadrupole mass spectrometry method in human plasma. DG diacylglycerol, GluCer glucosylceramide, SM sphingomyelin

Full size image

Finally, 370 lipid species from ten lipid classes were determined in human plasma by this 2D SFC/RPLC–QQQ MS method. The optimized acquisition settings for each lipid species, including precursor–product ion pairs, fragmentor voltage, and collision energy, are presented in Table S2. Among these lipids, some low-abundance species were successfully identified, such as GluCer (d18:1/15:1) and DG (25:4/17:0) (Fig. S7). Figure 5 shows the distribution of detected lipids in each lipid class. As we can see, the highest number of lipid species was identified for TG (81), SM (57), DG (54), PC (54), and Cer (43). Compared with 1D SFC–MS for the analysis of TGs, the number of identified TGs was similar [15]. Although more lipid classes were detected by 1D ultrahigh-performance SFC–MS [27], more lipid species in each lipid class were identified by the 2D SFC/RPLC–MS method. Thus, the 2D SFC/RPLC-MS method offered a much higher peak capacity, reduced ion suppression effects among lipid molecular species, and facilitated the identification of the lipid molecular species.

Fig. 5
figure 5

Detected lipids in each lipid class in human plasma. AC acylcarnitine, Cer ceramide, DG diacylglycerol, GalCer galactosylceramide, GluCer glucosylceramide, LacCer lactosylceramide, LPE lysophosphatidylethanolamine, PC phosphatidylcholine, SM sphingomyelin, TG triacylglycerol

Full size image

Comparison of patients with breast cancer and healthy controls

With use of this novel 2D SFC/RPLC–MS method, the differences in lipids in human plasma between patients with breast cancer (n = 10) and healthy controls (n = 10) were explored. Among the detected lipid species, 15 TGs, 3 Cers, 1 DG, and 1 PC showed significant differences (p < 0.05) between the two groups. Their average corrected peak area in each group, p values, and fold change are summarized in Table 4. As we can see, TGs were upregulated in the patient group, whereas Cer, DG, and PC were downregulated. They were proposed as potential lipid biomarkers for breast cancer. Among these lipids, TGs are present in the greatest number. Adipose triglyceride lipase is the rate-limiting enzyme in the TG hydrolysis cascade. A previous report showed that low levels of Adipose triglyceride lipase messenger RNA correlated with significantly reduced survival in patients with breast cancer [28], meaning TG hydrolysis was inhibited in patients with breast cancer. Consequently, TGs were upregulated, which was consistent with our results.

Table 4 Detailed information of potential lipid biomarkers for breast cancer
Full size table

Conclusions

In this work, online coupling of SFC and RPLC with QQQ MS was realized for lipid profiling in human plasma for the first time. As lipid classes could be separated in the first-dimension SFC by the differences in their polarity, and lipid species could be further separated in the second-dimension RPLC according to different acyl chains of lipids in each lipid class, 370 lipid species from ten lipid classes were successfully identified. Besides, the sample peak broadening in the first-dimension column was suppressed, leading to high sensitivity of the method. The analysis time was significantly reduced compared with that for our previous NP/RP 2D LC–MS method (170 min). In addition, the LOD, linearity, and repeatability were satisfactory for most of the lipid standards, indicating the high sensitivity and good repeatability of this method. Finally, the method was used for the analysis of human plasma from patients with breast cancer and healthy controls. Twenty potential lipid biomarkers for breast cancer were found. From the findings taken together, the 2D SFC/RPLC–QQQ MS method shows high potential as an MS-based method in the field of lipidomics and clinical research.