Clinica Chimica Acta 437 (2014) 31–37
Contents lists available at ScienceDirect
Clinica Chimica Acta journal homepage: www.elsevier.com/locate/clinchim
Decreased serum levels of free fatty acids are associated with breast cancer Yaping Zhang a, Lina Song b, Ning Liu c, Chengyan He b,⁎, Zhili Li a,⁎⁎ a b c
Department of Biophysics and Structural Biology, Institute of Basic Medical Sciences, Chinese Academy of Medical Sciences & School of Basic Medicine, Peking Union Medical College, Beijing, PR China Laboratory Medicine Center, China–Japan Union Hospital of Jilin University, Changchun 130033, PR China Central Laboratory, Jilin University Second Hospital, Changchun 130041, PR China
a r t i c l e
i n f o
Article history: Received 23 March 2014 Received in revised form 19 June 2014 Accepted 1 July 2014 Available online 9 July 2014 Keywords: Biomarker Free fatty acids Serum Breast cancer
a b s t r a c t Background: Changes in the levels of lipids are associated with breast cancer (BC). Methods: Disease-specific serum free fatty acids (FFAs) were quantified using chip-based direct-infusion nanoelectrospray ionization-Fourier transform ion cyclotron resonance mass spectrometry (CBDInanoESIFTICR MS) in the negative ion mode. Multiple point internal standard calibration curves between the concentration ratios of fatty acids (i.e., C16:1, C18:3, C18:2, C18:1, C20:4, and C22:6) to internal standards (C17:1 for C16:1, C18:3, C18:2, and C18:1, C21:0 for C20:4 and C22:6) and their corresponding intensity ratios were established with a correlation coefficient of greater than 0.986. Results: Data from 342 serum samples including 202 healthy controls and 140 BC patients indicate that serum concentrations of FFAs in patients with BC were significantly decreased compared with those in healthy controls. A panel of C16:1, C18:3, C18:2, C20:4, and C22:6 showed an excellent diagnostic ability to differentiate the patients with early stage BC from healthy controls, with the area under the receiver operating characteristics (ROC) curve of 0.953, a sensitivity of 83.3%, and a specificity of 87.1%. Conclusion: Our findings suggest that these FFAs may be a valuable biomarker panel for the early-stage detection of BC. © 2014 Elsevier B.V. All rights reserved.
1. Introduction Breast cancer (BC) is the most common malignancy in females in the world [1]. Mammography is a widely used screening tool for this disease based on the ionizing radiation, along with a 10–15% false negative rate [2]. Other gold standards, such as histopathology and blood tests, are rarely used as risk assessment tools due to their high cost and poor sensitivity [3,4]. Cancer antigen 15-3 (CA15-3), HER2, and carcinoembryonic antigen as tumor markers have been used in the patients with BC for predicting BC recurrence and for evaluating BC treatment outcomes, but they are non-specific biomarkers for BC [5,6]. Ideal techniques for the early detection of BC should be low cost and minimally invasive, with a high diagnostic accuracy. Microarray-based gene expression profiling including microRNAs and DNA methylation was performed for the early detection of BC with high sensitivity and high throughput [7,8]. But its ⁎ Correspondence to: C. He, Laboratory Medicine Center, China–Japan Union Hospital of Jilin University, Changchun 130033, PR China. ⁎⁎ Correspondence to: Z. Li, Department of Biophysics and Structural Biology, Institute of Basic Medical Sciences, Chinese Academy of Medical Sciences & School of Basic Medicine, Peking Union Medical College, 5 Dongdan San Tiao, Beijing 100005, PR China. Tel./fax: +86 10 69156479. E-mail addresses: [email protected] (C. He), [email protected] (Z. Li).
http://dx.doi.org/10.1016/j.cca.2014.07.001 0009-8981/© 2014 Elsevier B.V. All rights reserved.
application in clinical routine test involves the complicated extraction of microRNA and DNA. Circulating tumor cells in the peripheral blood are used to diagnose BC, but rare-events detected in the peripheral blood restricted its implementation [9,10]. Downstream of genomics, transcriptomics, and proteomics metabolomics not only provides a novel insight into the changes in the metabolic status of living system but also improves our understanding of the pathogenesis. Free fatty acids (FFAs) play a key role in many metabolic pathways. FFAs act as substrates in energy metabolism and mediators in signal transduction. Changes in blood FFA concentrations are indicators of a healthy status. Several potential biomarkers for diabetes mellitus [11–13], Alzheimer's disease [14], pancreatic cancer [15], and autism [16,17] were found relative to changes in the concentrations of FFAs in the peripheral blood. Recently, serum FFA profiles in BC patients were examined by gas chromatography/mass spectrometry (GC/MS), which demonstrated that changes in FFA concentrations were associated with BC [18]. However, fatty acids must be derivatized prior to GC/MS analysis due to their nonvolatile feature. It should be noted that time-consuming and complicated procedures for sample preparation are disadvantages of GC/MS and liquid chromatography/MS analysis. In this study, we employed chip-based direct infusion nanoelectrospray ionization-Fourier transform ion cyclotron resonance mass
32
Y. Zhang et al. / Clinica Chimica Acta 437 (2014) 31–37
spectrometry (CBDInanoESI-FTICR MS) in the negative ion mode to simultaneously quantify six serum FFAs from the patients with BC and age-matched healthy controls. Mann–Whitney U test was performed to compare healthy controls with the patients. Excellent diagnostic ability with the area under the receiver operating characteristics (ROC) curve (AUC) of 0.953, a sensitivity of 83.3%, and a specificity of 87.1% between healthy controls and patients with early-stage BC may provide a permission of early detection of BC. 2. Materials and methods 2.1. Chemicals and reagents Palmitoleic acid (C16:1), heptadecenoic acid (C17:1), linolenic acid (C18:3), linoleic acid (C18:2), oleic acid (C18:1), arachidonic acid (C20:4), heneicosanoic acid (C21:0), docosahexaenoic acid (C22:6), and ammonium acetate (all purity N 99% except for C22:6 with purity N 98%) were from Sigma-Aldrich Chemicals. Palmitic acid (C16:0, purity N 99%) was from J&K (J&K Scientific Ltd.). HPLC-grade methanol, ethanol, and acetonitrile were from Fisher Scientific. The ultrapure water was supplied by a Milli-Q system (Millipore). 2.2. Participants Characteristics of all participants are summarized in Table 1. 140 patients with BC from China–Japan Union Hospital (Changchun, China) were enrolled in this study, 36 of which were classified as early stage (stage I or II) and 24 as advanced stage (stage III or IV) based on the 7th edition of the Union for International Cancer Control (UICC) tumor– node–metastasis (TNM) classification. Evaluations of hematochemical parameters were also performed in this hospital. 202 healthy controls were also from China–Japan Union Hospital, with no obviously clinical abnormalities. All specimens were remaining sera after clinical laboratory examination. Serum samples from the patients were collected at diagnosis and all participants have given informed consents. This study was approved by the Ethics Review Board at the Institute of Basic Medical Sciences, Chinese Academy of Medical Sciences.
FAs that included C16:1 (342.0 μmol/l), C18:3 (81.7 μmol/l), C18:2 (569.0 μmol/l), C18:1 (836.0 μmol/l), C20:4 (183.0 μmol/l), and C22:6 (56.2 μmol/l) in methanol was further diluted by 500-fold with methanol/acetonitrile/5 mmol/l ammonium acetate in water (42/28/30, v/v/v) as STD1 (684.0 nmol/l C16:1, 163.5 nmol/l C18:3, 1138.0 nmol/l C18:2, 1672.0 nmol/l C18:1, 366.0 nmol/l C20:4, and 112.3 nmol/l C22:6). Then STD1 was further serially diluted by 2-, 10-, 20-, 50-, and 80-fold with methanol/acetonitrile/5 mmol/l ammonium acetate in water (42/28/30, v/v/v), which were named as STD2 (342.0 nmol/l C16:1, 81.8 nmol/l C18:3, 569.0 nmol/l C18:2, 836.0 nmol/l C18:1, 183.0 nmol/l C20:4, and 56.2 nmol/l C22:6), STD3 (68.4 nmol/l C16:1, 16.4 nmol/l C18:3, 113.8 nmol/l C18:2, 167.2 nmol/l C18:1, 36.3 nmol/l C20:4, and 11.2 nmol/l C22:6), STD4 (34.2 nmol/l C16:1, 8.2 nmol/l C18:3, 56.9 nmol/l C18:2, 83.6 nmol/l C18:1, 18.3 nmol/l C20:4, and 5.6 nmol/l C22:6), STD5 (13.7 nmol/l C16:1, 3.3 nmol/l C18:3, 22.8 nmol/l C18:2, 33.4 nmol/l C18:1, 7.3 nmol/l C20:4, and 2.3 nmol/l C22:6), and STD6 (8.5 nmol/l C16:1, 2.0 nmol/l C18:3, 14.2 nmol/l C18:2, 20.9 nmol/l C18:1, 4.6 nmol/l C20:4, and 1.4 nmol/l C22:6), respectively. The above-mentioned standard solutions were spiked with 1 μl of the IS solution with the final concentrations (83.3 nmol/l C17:1 and 33.3 nmol/l C21:0) to establish calibration equations, respectively. STD3 was also used as a quality control (QC) sample to examine the experimental stability. All standard solutions were freshly prepared and stored at 4 °C. 2.5. Sample preparations FFAs were prepared as our own previous study. Briefly, 50 μl of serum was added into 950 μl of methanol/acetonitrile (3/2, v/v) to precipitate serum proteins. The mixture was vortexed for 30 s, and then stored at −20 °C overnight. After centrifuging at 19,000 ×g for 30 min at 4 °C, the supernatant was transferred into a new 1.5 ml tube. An aliquot of 20 μl supernatant was spiked with 1 μl of IS stock solution, and then 500 μl of hexane and 500 μl of water were added followed by vortex-mixing for 30 s. The resulting mixture was then centrifuged at 1500 ×g for 10 min. Organic phase was transferred into another glass vial and then air-dried. One milliliter of methanol/acetonitrile/ 5 mmol/l ammonium acetate in water (42/28/30, v/v/v) was added to redissolve the residues for mass spectrometric analysis.
2.3. Study design
2.6. Mass spectrometry
Serum samples from the patients were randomly classified into two groups: the training set (n = 68) and the validation set (n = 72). Healthy controls were randomly matched by age to the patients in the training set and the validation set, respectively.
All mass spectra were acquired by a 9.4 T Apex-ultra™ hybrid QhFourier transform ion cyclotron resonance mass spectrometer (FTICR MS, Bruker Daltonics) equipped with a NanoMate system including nanoelectrospray ionization (nanoESI) source and a cooling unit (Advion BioSciences). The cooling unit was set at 4 °C to reduce solvent evaporation of the samples. The NanoMate system consists of a 96-well plate, conductive pipette tips, and ESI-Chip with a 20 × 20 array of nozzle. The sample volume was 0.1 μl. A low delivery gas pressure of 0.7 psi and nanoESI voltage of −1.8 kV was applied to the nozzle to generate nanoelectrospray at a flow rate of about 100 nl/min. A mass spectrum was accumulated by 10 full scans in the negative ion mode with time-domain sizes of 1 Mb (resolution: 200,000 at m/z 400). A mixture of C15:0 (molecular weight, 242.22458 Da), C17:0 (270.25588 Da), and C21:0 (326.31848 Da) was employed to calibrate the mass spectrometer over the mass range of 150–400 Da.
2.4. Preparation of stock standard solutions The internal standard (IS) stock solution including 83.3 μmol/l of C17:1 and 33.3 μmol/l of C21:0 was prepared in methanol, and was further diluted to proper concentrations until use. The stock standard solution of
Table 1 Characteristics of the participants for the training and validation set. Training set
Age (years) Mean ± SD Range Stage (male/female) I & II (Early stage) III & IV (advanced stage) BC: breast cancer.
Validation set
Controls
BC
Controls
BC
(n = 70)
(n = 68)
(n = 132)
(n = 72)
53.3 ± 10.9 28–76
52.5 ± 10.8 29–76
52.9 ± 10.3 29–73
50.9 ± 9.2 29–78
36 24
2.7. Data processing Mass spectra were obtained from ApexControl 3.0.0 (Bruker) in expert mode. DataAnalysis 4.0 (Bruker) was used for subsequent data processing. After isotopic deconvolution, peaks were transferred to Microsoft Excel for the following analyses. Serum FFAs were identified based on their observed accurate molecular masses and reliable isotope distributions. Absolute mass error was less than 0.00023 Da, and relative intensity error of observed to theoretical values was b 2%. If FFA
Y. Zhang et al. / Clinica Chimica Acta 437 (2014) 31–37
33
signals were absent, the baseline intensity in each spectrum was adopted for the statistical analysis. 2.8. Statistical analysis Statistical analysis was performed by SPSS software (ver 16.0). The results were expressed as mean ± standard deviation (SD). Comparison of the FFA concentrations between the patients and healthy controls was performed using the Mann–Whitney U test. The ROC curves were generated based on all identified FFAs and their combinations, and the AUC values and cut-off values, along with sensitivity and specificity, were calculated. For the validation set, we used independent samples to re-evaluate the diagnostic accuracy, the sensitivity, and the specificity based on the cut-off values obtained from the training set. For the stagerelated analysis, the sensitivity and specificity were also computed based on the cut-off values from the training set. In all cases, a p b 0.05 was considered statistically significant. 2.9. Method validation
Fig. 1. Representative negative ion mass spectra of serum FFAs of healthy control (a) and one patient with BC (b).
Method validation was performed as described previously. Briefly, STD1, STD2, STD3, STD4, STD5, and STD6 were analyzed for three times, and the results were expressed as mean ± SD. Calibration curves were established between the concentration ratios of fatty acids to ISs (C16:1, C18:3, C18:2, and C18:1 to C17:1, C20:4, and C22:6 to C21:0) and their corresponding intensity ratios. STD3 as a quality control (QC) sample was analyzed once every 10 test serum samples. The relative standard deviations of C16:1, C18:3, C18:2, C18:1, C20:4, and C22:6 from 35 analytical results of QC sample were calculated based on the intensity ratios of FFAs to ISs (C17:1 or C21:0). 3. Results 3.1. Simultaneous qualitative and quantitative analyses of serum FFAs Using CBDInanoESI-FTICR MS platform, 6 serum FFAs were simultaneously quantified. Multiple point internal standard calibration curves were constructed between the concentration ratios of fatty acids to ISs and their corresponding intensity ratios, with the correlation coefficient of more than 0.98. In addition, 35 analytical results of QC sample with the relative standard deviation of less than 19% indicate that this platform could provide a good stability for the analysis of complex biological samples. These data are listed in Table 2. Representative negative ion mass spectra of serum FFAs from one healthy control and one patient are shown in Fig. 1. Identification of FFAs was performed on the basis of observed accurate molecular masses and reliable isotope distributions. In this study, FFAs with the mass error of the observed to the theoretical values of less than 0.00023 and with the relative intensity error of the observed to the theoretical values of less than 2% were considered
to be identified (Supplementary Table S1). The concentrations of FFAs were calculated based on their respective calibration curves listed in Table 2. 3.2. Association of the concentrations of FFAs with BC In the training set, serum concentrations of C16:1, C18:3, C18:2, C20:4, and C22:6, along with the ratios of C18:2/C18:1 and C18:3/C18:1, of the patients were significantly decreased compared with healthy controls (p b 0.05, Fig. 2 and Supplementary Table S2). AUC values, sensitivities, and specificities were calculated on the basis of the above-mentioned variables and their different combinations. As shown in Table 3, panel b (PUFAs, C18:3, C18:2, C20:4, and C22:6) and panel a (C16:1, C18:3, C18:2, C20:4, and C22:6) have high AUC values (0.876 and 0.885, respectively), with the sensitivities of more than 80% and the specificities of more than 82%. The ROC curve for panel a is shown in Fig. 3a. Single variable, such as C16:1, C18:3, C18:2, C18:2/C18:1, or C18:3/C18:1, has high sensitivities of more than 82.9%, with lower specificities of less than 70.0%, whereas C20:4, C22:6, or panel b (C18:2/C18:1 and C18:3/C18:1) has high specificities of more than 80.0%, with low sensitivities of less than 65.0%. In the validation set, decreased concentrations of the abovementioned variables were also significantly associated with BC (p b 0.0001) (Fig. 2 and Supplementary Table S2). The AUC values of C16:1, C20:4, C18:2/C18:1, PUFA, panel a, or panel b are more than 0.8, but only panel a has the sensitivity and specificity of more than 80% (Table 3). The ROC curve for panel a is shown in Fig. 3b. 3.3. Association of the FFA concentrations with BC stages
Table 2 Linear range of concentration, calibration equation, coefficient correlation (R2), and the stability. FAs
C16:1 C18:3 C18:2 C18:1 C20:4 C22:6
Linearity (n = 3)
QC (%)
FA (nM)
Equation
8.5−684.0 2.0−163.5 14.2−1138.0 20.9−1672.0 4.6−366.0 1.4−112.3
Y Y Y Y Y Y
= = = = = =
0.497 (±0.013)X 1.236 (±0.066)X 1.379 (±0.069)X 1.429 (±0.090)X 1.871 (±0.136)X 1.476 (±0.087)X
R2 + − + + + −
0.152 (±0.007) 0.016 (±0.009) 0.326 (±0.016) 1.205 (±0.265) 0.229 (±0.025) 0.019 (±0.014)
0.998 0.997 0.986 0.990 0.997 0.998
b18 b12 b12 b18 b19 b19
X: Concentration ratios of individual FAs to ISs (83.3 nM for C17:1 and 33.3 nM for C21:0); Y: corresponding intensity ratios of FAs to IS
As shown in Fig. 2, the scatter plots of cancer staging versus the concentrations of these variables indicate that decreased concentrations of them were remarkably associated with BC stages. The concentrations of C16:1, C18:3, C18:2, C20:4, C18:2/C18:1, and C18:3/C18:1 show a significant difference in patients with early- or advanced-stage BC compared with healthy controls (p b 0.05, Supplementary Table S2). In addition, C22:6 also has significant differences in patients with early stage BC compared with healthy controls (p = 0.001, Supplementary Table S2). The ROC analysis indicated that panel a has an excellent diagnostic ability to differentiate patients with early-stage BC from healthy controls, with an AUC value of 0.953, a sensitivity of 83.3%, and a specificity of 87.1%, and that panel b also exhibits an excellent diagnostic ability to differentiate patients with advanced-stage BC from healthy controls, with an AUC value of 0.921, a sensitivity of 83.3%, and a specificity of 90.0%
34
Y. Zhang et al. / Clinica Chimica Acta 437 (2014) 31–37
Fig. 2. Scatter plots of the levels of serum FFAs in the training set, the validation set and different stages of BC. *, p b 0.05; **, p b 0.01; ***, p b 0.001.
Y. Zhang et al. / Clinica Chimica Acta 437 (2014) 31–37
35
Table 3 AUC, sensitivity and specificity of FFAs between different groups based on the same cut-off values. Note: Sens: sensitivity; Spec: specificity; PUFA: polyunsaturated fatty acid. Panel a: C16:1, C18:3, C18:2, C20:4 and C22:6. Panel b: C18:2/C18:1 and C18:3/C18:1. Panel c: C16:1, C18:3, C18:2 and C20:4. FFAs
C16:1 C18:3 C18:2 C20:4 C22:6 C18:2/C18:1 C18:3/C18:1 PUFA Panel a Panel b Panel c
Training set
Validation set
Controls vs. BC
Controls vs. BC
Cut-off
Stage Controls vs. early stage
AUC (95% CI)
Sens (%)
Spec (%)
AUC (95% CI)
Sens (%)
Spec (%)
0.714 (.628–.801) 0.770 (.692–.849) 0.613 (.517–.708) 0.755 (.675–.835) 0.614 (.521–.708) 0.729 (.645–.813) 0.789 (.711–.867) 0.876 (.821–.932) 0.885 (.833–.938) 0.778 (.698–.857)
85.7 84.3 90.0 55.7 41.4 82.9 87.1 80.9 80.8 64.7
55.9 61.8 38.2 88.2 83.8 58.8 67.6 80.0 82.3 88.6
0.847 (.792–.902) 0.675 (.593–.756) 0.703 (.624–.782) 0.858 (.802–.915) 0.730 (.660–.800) 0.871 (.819–.923) 0.758 (.692–.824) 0.850 (.793–.906) 0.908 (.866–.950) 0.871 (.818–.924)
68.2 69.7 81.8 75.8 62.1 72.7 72.0 66.7 83.3 76.4
80.6 56.9 50.0 83.3 81.9 90.3 73.6 87.1 83.3 84.8
29.6 18.7 207.0 153.6 51.1 0.8 0.1 0.5 0.4 0.6 0.4
Controls vs. advanced stage
AUC (95% CI)
Sens (%)
Spec (%)
AUC (95% CI)
Sens (%)
Spec (%)
0.927 (.878–.976) 0.692 (.578–.806) 0.722 (.606–.839) 0.891 (.822–.960) 0.694 (.597–.792) 0.890 (.819–.960) 0.758 (.667–.849) 0.892 (.824–.960) 0.953 (.917–.988) 0.889 (.816–.962)
75.7 68.6 87.1 71.4 41.4 80.0 71.4 72.2 83.3 69.4
91.7 61.1 55.6 88.9 94.4 88.9 77.8 94.3 87.1 94.3
0.788 (.686–.890) 0.656 (.520–.791) 0.718 (.592–.844) 0.818 (.699–.938)
75.7 82.9 87.1 30.0
76.7 50.0 45.8 87.5
0.918 (.851–.986) 0.759 (.656–.862) 0.794 (.673–.915)
80.0 71.4 54.2
91.7 70.8 95.7
0.921 (.848–.993) 0.866 (.778–.954)
83.3 75.0
90.0 91.4
Note: Sens: sensitivity; Spec: specificity; PUFA: polyunsaturated fatty acid. Panel a: C16:1, C18:3, C18:2, C20:4 and C22:6. Panel b: C18:2/C18:1 and C18:3/C18:1. Panel c: C16:1, C18:3, C18:2 and C20:4.
(Table 3). The ROC curves for panels a and b are shown in Fig. 3c and d, respectively. 4. Discussion Serum FFAs are important bioactive molecules and the primary sources of energy in the body. In this study, the decreased concentrations of C16:1, C18:3, C18:2, C20:4, and C22:6, or ratios of C18:2/C18:1 and
C18:3/C18:1 were significantly associated with BC, which may be due to the increasing energy expenditure in cancer development. Previous study suggested that a decreased concentration of C18:3 and an increased concentration of C18:2 were significantly associated with BC, but the concentrations of C16:1, C20:4, and C22:6 were not related to BC [18]. These results are partly consistent with our data obtained in this study. As shown in Fig. 2, change trends of the concentrations of FFAs are observed in the following order: healthy controls N patients with
Fig. 3. The ROC curve analysis for different biomarker panels. (a) Panel a (a combination of C16:1, C18:3, C18:2, C20:4, and C22:6) for differentiating patients with BC from healthy controls in the training set. (b) Panel a for differentiating patients with BC from healthy controls in the validation set. (c) Panel a for differentiating patients with early stage BC from healthy controls. (d) Panel b (a combination of C18:2/C18:1 and C18:3/C18:1) for differentiating patients with advanced stage BC from healthy controls.
36
Y. Zhang et al. / Clinica Chimica Acta 437 (2014) 31–37
advanced-stage BC N patients with early-stage BC, but no statistical differences between patients with early-stage BC and advancedstage BC are observed. These results may suggest that a slight difference in FFA metabolism exists between early-stage BC and advanced-stage BC. Previous studies have found the anti-proliferative effects of n − 3 PUFAs on BC cells in vitro [19]. Emerging studies on the relationship between dietary fatty acids and BC have indicated that n−6 PUFAs have stronger tumor-enhancing effects, whereas n−3 PUFAs have protective effects on mammary cancer [20,21]. However, some studies revealed no association between dietary intake of fatty acids and BC risk [22], or an increased risk of BC [23]. In addition, increased ratio of n−6 PUFAs to n−3 PUFAs may be related to an increased risk of BC [20,24]. It is well known that fatty acids are main components of phospholipids and glycolipids, which were significantly associated with disease status [25–27]. Changes in the concentrations of serum metabolites may reflect heterogeneity in disease status. Monitoring fluctuations of some metabolites in biofluids has become one of vital approaches to detect early-stage BC [28]. Metabolomic studies on BC mainly focused on the metabolites in the urine or the tissue [29–31]. Nuclear magnetic resonance investigation of metabolites in pre- and post-operative serum samples from 44 patients with early-stage BC indicated that some metabolites from pre-operative patients had a good diagnostic accuracy, with a sensitivity of 75%, a specificity of 69%, and an AUC value of 0.72 [32]. A combination of nine miRNAs could discriminate between patients with early-stage BC and healthy controls, with the AUC of 0.665, but the association between miRNA expression and tumor grade, tumor size, menopausal status or lymph node status was not detected [33]. Our previous study has indicated that a combination of C16:1, C18:3, C18:2, C20:4 and C22:6 has a high diagnostic accuracy for differentiating pancreatic cancer from healthy controls, with the AUC of 0.93, the sensitivity of 86%, and the specificity of 85%, and that a combination of the concentration ratios of C 18:2 / C18:1 and C18:3/C18:1, provides an excellent diagnostic accuracy to differentiate early-stage pancreatic cancer from the patients with pancreatitis plus normal controls, with the AUC of 0.912, the sensitivity of 86.7%, and the specificity of 88.6% [15].
5. Conclusions NanoESI-FTICR MS is a high-sensitivity and high-throughput technology that can analyze complex biological samples without cumbersome preparation steps. In the present study, chip-based direct-infusion nanoESI-FTICR MS in the negative mode was carried out to detect FFA concentrations in healthy controls and BC patients. The identified potential biomarkers could be used to distinguish BC patients from healthy controls, especially for early-stage patients. Hence, the measurement of FFAs in biological samples by a high-efficiency technology might be very helpful in the early diagnosis of BC.
Abbreviations BC breast cancer FFAs free fatty acids CBDInanoESI-FTICR MS chip-based direct infusion nanoelectrospray ionization-Fourier transform ion cyclotron resonance mass spectrometry.
Acknowledgments This study was funded by grant 91029701 from the National Natural Science Foundation of China (to Z. Li).
Appendix A. Supplementary data Supplementary data to this article can be found online at http://dx. doi.org/10.1016/j.cca.2014.07.001. References [1] Gotzsche PC, Jorgensen KJ. Screening for breast cancer with mammography. Cochrane Database Syst Rev 2013;6:CD001877. [2] Smith LF. Palpable cancer of the breast and negative mammography: the ongoing dilemma. J Surg Oncol 2007;96:451–2. [3] Zhang AH, Sun H, Qiu S, Wang XJ. Metabolomics in noninvasive breast cancer. Clin Chim Acta 2013;424:3–7. [4] Vlahou A, Laronga C, Wilson L, Gregory B, Fournier K, McGaughey D, et al. A novel approach toward development of a rapid blood test for breast cancer. Clin Breast Cancer 2003;4:203–9. [5] Pedersen AC, Sorensen PD, Jacobsen EH, Madsen JS, Brandslund I. Sensitivity of CA 15-3, CEA and serum HER2 in the early detection of recurrence of breast cancer. Clin Chem Lab Med 2013;51:1511–9. [6] Agrawal AK, Jelen M, Rudnicki J, Grzebieniak Z, Zysko D, Kielan W, et al. The importance of preoperative elevated serum levels of CEA and CA15-3 in patients with breast cancer in predicting its histological type. Folia Histochem Cytobiol 2010;48:26–9. [7] Ng EK, Li R, Shin VY, Jin HC, Leung CP, Ma ES, et al. Circulating microRNAs as specific biomarkers for breast cancer detection. PLoS One 2013;8:e53141. [8] Zhao H, Shen J, Medico L, Wang D, Ambrosone CB, Liu S. A pilot study of circulating miRNAs as potential biomarkers of early stage breast cancer. PLoS One 2010;5: e13735. [9] Ring A, Smith IE, Dowsett M. Circulating tumour cells in breast cancer. Lancet Oncol 2004;5:79–88. [10] Mostert B, Sleijfer S, Foekens JA, Gratama JW. Circulating tumor cells (CTCs): detection methods and their clinical relevance in breast cancer. Cancer Treat Rev 2009;35:463–74. [11] Liu L, Li Y, Guan C, Li K, Wang C, Feng R, et al. Free fatty acid metabolic profile and biomarkers of isolated post-challenge diabetes and type 2 diabetes mellitus based on GC–MS and multivariate statistical analysis. J Chromatogr B Analyt Technol Biomed Life Sci 2010;878:2817–25. [12] Yi LZ, He J, Liang YZ, Yuan DL, Chau FT. Plasma fatty acid metabolic profiling and biomarkers of type 2 diabetes mellitus based on GC/MS and PLS–LDA. FEBS Lett 2006;580:6837–45. [13] Zhang Y, Wang Y, Guo S, Guo Y, Liu H, Li Z. Ammonia-treated N-(1-naphthyl) ethylenediamine dihydrochloride as a novel matrix for rapid quantitative and qualitative determination of serum free fatty acids by matrix-assisted laser desorption/ ionization-Fourier transform ion cyclotron resonance mass spectrometry. Anal Chim Acta 2013;794:82–9. [14] Wang DC, Sun CH, Liu LY, Sun XH, Jin XW, Song WL, et al. Serum fatty acid profiles using GC–MS and multivariate statistical analysis: potential biomarkers of Alzheimer's disease. Neurobiol Aging 2012;33:1057–66. [15] Zhang Y, Qiu L, Wang Y, Qin X, Li Z. High-throughput and high-sensitivity quantitative analysis of serum unsaturated fatty acids by chip-based nanoelectrospray ionization-Fourier transform ion cyclotron resonance mass spectrometry: early stage diagnostic biomarkers of pancreatic cancer. Analyst 2014;139:1697–706. [16] Wiest MM, German JB, Harvey DJ, Watkins SM, Hertz-Picciotto I. Plasma fatty acid profiles in autism: a case–control study. Prostaglandins Leukot Essent Fatty Acids 2009;80:221–7. [17] El-Ansary AK, Bacha AG, Al-Ayahdi LY. Plasma fatty acids as diagnostic markers in autistic patients from Saudi Arabia. Lipids Health Dis 2011;10:62. [18] Lv W, Yang T. Identification of possible biomarkers for breast cancer from free fatty acid profiles determined by GC–MS and multivariate statistical analysis. Clin Biochem 2012;45:127–33. [19] Bougnoux P, Hajjaji N, Maheo K, Couet C, Chevalier S. Fatty acids and breast cancer: sensitization to treatments and prevention of metastatic re-growth. Prog Lipid Res 2010;49:76–86. [20] Thiebaut AC, Chajes V, Gerber M, Boutron-Ruault MC, Joulin V, Lenoir G, et al. Dietary intakes of omega-6 and omega-3 polyunsaturated fatty acids and the risk of breast cancer. Int J Cancer 2009;124:924–31. [21] Gago-Dominguez M, Yuan JM, Sun CL, Lee HP, Yu MC. Opposing effects of dietary n− 3 and n− 6 fatty acids on mammary carcinogenesis: the Singapore Chinese Health Study. Br J Cancer 2003;89:1686–92. [22] Chajes V, Torres-Mejia G, Biessy C, Ortega-Olvera C, Angeles-Llerenas A, Ferrari P, et al. Omega-3 and omega-6 polyunsaturated fatty acid intakes and the risk of breast cancer in Mexican women: impact of obesity status. Cancer Epidemiol Biomarkers Prev 2012;21:319–26. [23] Stripp Connie, Overvad Kim, Christensen Jane, Thomsen Birthe L, Olsen Anja, Møller Susanne, Tjønneland A. Fish intake is positively associated with breast cancer incidence rate. Nutr Cancer 2003;133:3664–9. [24] Murff HJ, Shu XO, Li H, Yang G, Wu X, Cai H, et al. Dietary polyunsaturated fatty acids and breast cancer risk in Chinese women: a prospective cohort study. Int J Cancer 2011;128:1434–41. [25] Franky Dhaval S, Shilin Nandubhai S, Pankaj Manubhai S, Patel HR, Prabhudas Shankerbhai P. Significance of alterations in plasma lipid profile levels in breast cancer. Integr Cancer Ther 2008;7:33–41. [26] Hilvo M, Denkert C, Lehtinen L, Muller B, Brockmoller S, Seppanen-Laakso T, et al. Novel theranostic opportunities offered by characterization of altered membrane lipid metabolism in breast cancer progression. Cancer Res 2011;71:3236–45.
Y. Zhang et al. / Clinica Chimica Acta 437 (2014) 31–37 [27] Abdelsalam KE, Hassan IK, Sadig IA. The role of developing breast cancer in alteration of serum lipid profile. J Res Med Sci 2012;17:562–5. [28] Tenori L, Oakman C, Claudino WM, Bernini P, Cappadona S, Nepi S, et al. Exploration of serum metabolomic profiles and outcomes in women with metastatic breast cancer: a pilot study. Mol Oncol 2012;6:437–44. [29] Woo HM, Kim KM, Choi MH, Jung BH, Lee J, Kong G, et al. Mass spectrometry based metabolomic approaches in urinary biomarker study of women's cancers. Clin Chim Acta 2009;400:63–9. [30] Chen Y, Zhang R, Song Y, He J, Sun J, Bai J, et al. RRLC–MS/MS-based metabonomics combined with in-depth analysis of metabolic correlation network: finding potential biomarkers for breast cancer. Analyst 2009;134:2003–11.
37
[31] Nam H, Chung BC, Kim Y, Lee K, Lee D. Combining tissue transcriptomics and urine metabolomics for breast cancer biomarker identification. Bioinformatics 2009;25:3151–7. [32] Oakman C, Tenori L, Claudino WM, Cappadona S, Nepi S, Battaglia A, et al. Identification of a serum-detectable metabolomic fingerprint potentially correlated with the presence of micrometastatic disease in early breast cancer patients at varying risks of disease relapse by traditional prognostic methods. Ann Oncol 2011;22:1295–301. [33] Kodahl AR, Lyng MB, Binder H, Cold S, Gravgaard K, Knoop AS, et al. Novel circulating microRNA signature as a potential non-invasive multi-marker test in ER-positive early-stage breast cancer: a case–control study. Mol Oncol 2014;8:874–83.
Contents lists available at ScienceDirect
Clinica Chimica Acta journal homepage: www.elsevier.com/locate/clinchim
Decreased serum levels of free fatty acids are associated with breast cancer Yaping Zhang a, Lina Song b, Ning Liu c, Chengyan He b,⁎, Zhili Li a,⁎⁎ a b c
Department of Biophysics and Structural Biology, Institute of Basic Medical Sciences, Chinese Academy of Medical Sciences & School of Basic Medicine, Peking Union Medical College, Beijing, PR China Laboratory Medicine Center, China–Japan Union Hospital of Jilin University, Changchun 130033, PR China Central Laboratory, Jilin University Second Hospital, Changchun 130041, PR China
a r t i c l e
i n f o
Article history: Received 23 March 2014 Received in revised form 19 June 2014 Accepted 1 July 2014 Available online 9 July 2014 Keywords: Biomarker Free fatty acids Serum Breast cancer
a b s t r a c t Background: Changes in the levels of lipids are associated with breast cancer (BC). Methods: Disease-specific serum free fatty acids (FFAs) were quantified using chip-based direct-infusion nanoelectrospray ionization-Fourier transform ion cyclotron resonance mass spectrometry (CBDInanoESIFTICR MS) in the negative ion mode. Multiple point internal standard calibration curves between the concentration ratios of fatty acids (i.e., C16:1, C18:3, C18:2, C18:1, C20:4, and C22:6) to internal standards (C17:1 for C16:1, C18:3, C18:2, and C18:1, C21:0 for C20:4 and C22:6) and their corresponding intensity ratios were established with a correlation coefficient of greater than 0.986. Results: Data from 342 serum samples including 202 healthy controls and 140 BC patients indicate that serum concentrations of FFAs in patients with BC were significantly decreased compared with those in healthy controls. A panel of C16:1, C18:3, C18:2, C20:4, and C22:6 showed an excellent diagnostic ability to differentiate the patients with early stage BC from healthy controls, with the area under the receiver operating characteristics (ROC) curve of 0.953, a sensitivity of 83.3%, and a specificity of 87.1%. Conclusion: Our findings suggest that these FFAs may be a valuable biomarker panel for the early-stage detection of BC. © 2014 Elsevier B.V. All rights reserved.
1. Introduction Breast cancer (BC) is the most common malignancy in females in the world [1]. Mammography is a widely used screening tool for this disease based on the ionizing radiation, along with a 10–15% false negative rate [2]. Other gold standards, such as histopathology and blood tests, are rarely used as risk assessment tools due to their high cost and poor sensitivity [3,4]. Cancer antigen 15-3 (CA15-3), HER2, and carcinoembryonic antigen as tumor markers have been used in the patients with BC for predicting BC recurrence and for evaluating BC treatment outcomes, but they are non-specific biomarkers for BC [5,6]. Ideal techniques for the early detection of BC should be low cost and minimally invasive, with a high diagnostic accuracy. Microarray-based gene expression profiling including microRNAs and DNA methylation was performed for the early detection of BC with high sensitivity and high throughput [7,8]. But its ⁎ Correspondence to: C. He, Laboratory Medicine Center, China–Japan Union Hospital of Jilin University, Changchun 130033, PR China. ⁎⁎ Correspondence to: Z. Li, Department of Biophysics and Structural Biology, Institute of Basic Medical Sciences, Chinese Academy of Medical Sciences & School of Basic Medicine, Peking Union Medical College, 5 Dongdan San Tiao, Beijing 100005, PR China. Tel./fax: +86 10 69156479. E-mail addresses: [email protected] (C. He), [email protected] (Z. Li).
http://dx.doi.org/10.1016/j.cca.2014.07.001 0009-8981/© 2014 Elsevier B.V. All rights reserved.
application in clinical routine test involves the complicated extraction of microRNA and DNA. Circulating tumor cells in the peripheral blood are used to diagnose BC, but rare-events detected in the peripheral blood restricted its implementation [9,10]. Downstream of genomics, transcriptomics, and proteomics metabolomics not only provides a novel insight into the changes in the metabolic status of living system but also improves our understanding of the pathogenesis. Free fatty acids (FFAs) play a key role in many metabolic pathways. FFAs act as substrates in energy metabolism and mediators in signal transduction. Changes in blood FFA concentrations are indicators of a healthy status. Several potential biomarkers for diabetes mellitus [11–13], Alzheimer's disease [14], pancreatic cancer [15], and autism [16,17] were found relative to changes in the concentrations of FFAs in the peripheral blood. Recently, serum FFA profiles in BC patients were examined by gas chromatography/mass spectrometry (GC/MS), which demonstrated that changes in FFA concentrations were associated with BC [18]. However, fatty acids must be derivatized prior to GC/MS analysis due to their nonvolatile feature. It should be noted that time-consuming and complicated procedures for sample preparation are disadvantages of GC/MS and liquid chromatography/MS analysis. In this study, we employed chip-based direct infusion nanoelectrospray ionization-Fourier transform ion cyclotron resonance mass
32
Y. Zhang et al. / Clinica Chimica Acta 437 (2014) 31–37
spectrometry (CBDInanoESI-FTICR MS) in the negative ion mode to simultaneously quantify six serum FFAs from the patients with BC and age-matched healthy controls. Mann–Whitney U test was performed to compare healthy controls with the patients. Excellent diagnostic ability with the area under the receiver operating characteristics (ROC) curve (AUC) of 0.953, a sensitivity of 83.3%, and a specificity of 87.1% between healthy controls and patients with early-stage BC may provide a permission of early detection of BC. 2. Materials and methods 2.1. Chemicals and reagents Palmitoleic acid (C16:1), heptadecenoic acid (C17:1), linolenic acid (C18:3), linoleic acid (C18:2), oleic acid (C18:1), arachidonic acid (C20:4), heneicosanoic acid (C21:0), docosahexaenoic acid (C22:6), and ammonium acetate (all purity N 99% except for C22:6 with purity N 98%) were from Sigma-Aldrich Chemicals. Palmitic acid (C16:0, purity N 99%) was from J&K (J&K Scientific Ltd.). HPLC-grade methanol, ethanol, and acetonitrile were from Fisher Scientific. The ultrapure water was supplied by a Milli-Q system (Millipore). 2.2. Participants Characteristics of all participants are summarized in Table 1. 140 patients with BC from China–Japan Union Hospital (Changchun, China) were enrolled in this study, 36 of which were classified as early stage (stage I or II) and 24 as advanced stage (stage III or IV) based on the 7th edition of the Union for International Cancer Control (UICC) tumor– node–metastasis (TNM) classification. Evaluations of hematochemical parameters were also performed in this hospital. 202 healthy controls were also from China–Japan Union Hospital, with no obviously clinical abnormalities. All specimens were remaining sera after clinical laboratory examination. Serum samples from the patients were collected at diagnosis and all participants have given informed consents. This study was approved by the Ethics Review Board at the Institute of Basic Medical Sciences, Chinese Academy of Medical Sciences.
FAs that included C16:1 (342.0 μmol/l), C18:3 (81.7 μmol/l), C18:2 (569.0 μmol/l), C18:1 (836.0 μmol/l), C20:4 (183.0 μmol/l), and C22:6 (56.2 μmol/l) in methanol was further diluted by 500-fold with methanol/acetonitrile/5 mmol/l ammonium acetate in water (42/28/30, v/v/v) as STD1 (684.0 nmol/l C16:1, 163.5 nmol/l C18:3, 1138.0 nmol/l C18:2, 1672.0 nmol/l C18:1, 366.0 nmol/l C20:4, and 112.3 nmol/l C22:6). Then STD1 was further serially diluted by 2-, 10-, 20-, 50-, and 80-fold with methanol/acetonitrile/5 mmol/l ammonium acetate in water (42/28/30, v/v/v), which were named as STD2 (342.0 nmol/l C16:1, 81.8 nmol/l C18:3, 569.0 nmol/l C18:2, 836.0 nmol/l C18:1, 183.0 nmol/l C20:4, and 56.2 nmol/l C22:6), STD3 (68.4 nmol/l C16:1, 16.4 nmol/l C18:3, 113.8 nmol/l C18:2, 167.2 nmol/l C18:1, 36.3 nmol/l C20:4, and 11.2 nmol/l C22:6), STD4 (34.2 nmol/l C16:1, 8.2 nmol/l C18:3, 56.9 nmol/l C18:2, 83.6 nmol/l C18:1, 18.3 nmol/l C20:4, and 5.6 nmol/l C22:6), STD5 (13.7 nmol/l C16:1, 3.3 nmol/l C18:3, 22.8 nmol/l C18:2, 33.4 nmol/l C18:1, 7.3 nmol/l C20:4, and 2.3 nmol/l C22:6), and STD6 (8.5 nmol/l C16:1, 2.0 nmol/l C18:3, 14.2 nmol/l C18:2, 20.9 nmol/l C18:1, 4.6 nmol/l C20:4, and 1.4 nmol/l C22:6), respectively. The above-mentioned standard solutions were spiked with 1 μl of the IS solution with the final concentrations (83.3 nmol/l C17:1 and 33.3 nmol/l C21:0) to establish calibration equations, respectively. STD3 was also used as a quality control (QC) sample to examine the experimental stability. All standard solutions were freshly prepared and stored at 4 °C. 2.5. Sample preparations FFAs were prepared as our own previous study. Briefly, 50 μl of serum was added into 950 μl of methanol/acetonitrile (3/2, v/v) to precipitate serum proteins. The mixture was vortexed for 30 s, and then stored at −20 °C overnight. After centrifuging at 19,000 ×g for 30 min at 4 °C, the supernatant was transferred into a new 1.5 ml tube. An aliquot of 20 μl supernatant was spiked with 1 μl of IS stock solution, and then 500 μl of hexane and 500 μl of water were added followed by vortex-mixing for 30 s. The resulting mixture was then centrifuged at 1500 ×g for 10 min. Organic phase was transferred into another glass vial and then air-dried. One milliliter of methanol/acetonitrile/ 5 mmol/l ammonium acetate in water (42/28/30, v/v/v) was added to redissolve the residues for mass spectrometric analysis.
2.3. Study design
2.6. Mass spectrometry
Serum samples from the patients were randomly classified into two groups: the training set (n = 68) and the validation set (n = 72). Healthy controls were randomly matched by age to the patients in the training set and the validation set, respectively.
All mass spectra were acquired by a 9.4 T Apex-ultra™ hybrid QhFourier transform ion cyclotron resonance mass spectrometer (FTICR MS, Bruker Daltonics) equipped with a NanoMate system including nanoelectrospray ionization (nanoESI) source and a cooling unit (Advion BioSciences). The cooling unit was set at 4 °C to reduce solvent evaporation of the samples. The NanoMate system consists of a 96-well plate, conductive pipette tips, and ESI-Chip with a 20 × 20 array of nozzle. The sample volume was 0.1 μl. A low delivery gas pressure of 0.7 psi and nanoESI voltage of −1.8 kV was applied to the nozzle to generate nanoelectrospray at a flow rate of about 100 nl/min. A mass spectrum was accumulated by 10 full scans in the negative ion mode with time-domain sizes of 1 Mb (resolution: 200,000 at m/z 400). A mixture of C15:0 (molecular weight, 242.22458 Da), C17:0 (270.25588 Da), and C21:0 (326.31848 Da) was employed to calibrate the mass spectrometer over the mass range of 150–400 Da.
2.4. Preparation of stock standard solutions The internal standard (IS) stock solution including 83.3 μmol/l of C17:1 and 33.3 μmol/l of C21:0 was prepared in methanol, and was further diluted to proper concentrations until use. The stock standard solution of
Table 1 Characteristics of the participants for the training and validation set. Training set
Age (years) Mean ± SD Range Stage (male/female) I & II (Early stage) III & IV (advanced stage) BC: breast cancer.
Validation set
Controls
BC
Controls
BC
(n = 70)
(n = 68)
(n = 132)
(n = 72)
53.3 ± 10.9 28–76
52.5 ± 10.8 29–76
52.9 ± 10.3 29–73
50.9 ± 9.2 29–78
36 24
2.7. Data processing Mass spectra were obtained from ApexControl 3.0.0 (Bruker) in expert mode. DataAnalysis 4.0 (Bruker) was used for subsequent data processing. After isotopic deconvolution, peaks were transferred to Microsoft Excel for the following analyses. Serum FFAs were identified based on their observed accurate molecular masses and reliable isotope distributions. Absolute mass error was less than 0.00023 Da, and relative intensity error of observed to theoretical values was b 2%. If FFA
Y. Zhang et al. / Clinica Chimica Acta 437 (2014) 31–37
33
signals were absent, the baseline intensity in each spectrum was adopted for the statistical analysis. 2.8. Statistical analysis Statistical analysis was performed by SPSS software (ver 16.0). The results were expressed as mean ± standard deviation (SD). Comparison of the FFA concentrations between the patients and healthy controls was performed using the Mann–Whitney U test. The ROC curves were generated based on all identified FFAs and their combinations, and the AUC values and cut-off values, along with sensitivity and specificity, were calculated. For the validation set, we used independent samples to re-evaluate the diagnostic accuracy, the sensitivity, and the specificity based on the cut-off values obtained from the training set. For the stagerelated analysis, the sensitivity and specificity were also computed based on the cut-off values from the training set. In all cases, a p b 0.05 was considered statistically significant. 2.9. Method validation
Fig. 1. Representative negative ion mass spectra of serum FFAs of healthy control (a) and one patient with BC (b).
Method validation was performed as described previously. Briefly, STD1, STD2, STD3, STD4, STD5, and STD6 were analyzed for three times, and the results were expressed as mean ± SD. Calibration curves were established between the concentration ratios of fatty acids to ISs (C16:1, C18:3, C18:2, and C18:1 to C17:1, C20:4, and C22:6 to C21:0) and their corresponding intensity ratios. STD3 as a quality control (QC) sample was analyzed once every 10 test serum samples. The relative standard deviations of C16:1, C18:3, C18:2, C18:1, C20:4, and C22:6 from 35 analytical results of QC sample were calculated based on the intensity ratios of FFAs to ISs (C17:1 or C21:0). 3. Results 3.1. Simultaneous qualitative and quantitative analyses of serum FFAs Using CBDInanoESI-FTICR MS platform, 6 serum FFAs were simultaneously quantified. Multiple point internal standard calibration curves were constructed between the concentration ratios of fatty acids to ISs and their corresponding intensity ratios, with the correlation coefficient of more than 0.98. In addition, 35 analytical results of QC sample with the relative standard deviation of less than 19% indicate that this platform could provide a good stability for the analysis of complex biological samples. These data are listed in Table 2. Representative negative ion mass spectra of serum FFAs from one healthy control and one patient are shown in Fig. 1. Identification of FFAs was performed on the basis of observed accurate molecular masses and reliable isotope distributions. In this study, FFAs with the mass error of the observed to the theoretical values of less than 0.00023 and with the relative intensity error of the observed to the theoretical values of less than 2% were considered
to be identified (Supplementary Table S1). The concentrations of FFAs were calculated based on their respective calibration curves listed in Table 2. 3.2. Association of the concentrations of FFAs with BC In the training set, serum concentrations of C16:1, C18:3, C18:2, C20:4, and C22:6, along with the ratios of C18:2/C18:1 and C18:3/C18:1, of the patients were significantly decreased compared with healthy controls (p b 0.05, Fig. 2 and Supplementary Table S2). AUC values, sensitivities, and specificities were calculated on the basis of the above-mentioned variables and their different combinations. As shown in Table 3, panel b (PUFAs, C18:3, C18:2, C20:4, and C22:6) and panel a (C16:1, C18:3, C18:2, C20:4, and C22:6) have high AUC values (0.876 and 0.885, respectively), with the sensitivities of more than 80% and the specificities of more than 82%. The ROC curve for panel a is shown in Fig. 3a. Single variable, such as C16:1, C18:3, C18:2, C18:2/C18:1, or C18:3/C18:1, has high sensitivities of more than 82.9%, with lower specificities of less than 70.0%, whereas C20:4, C22:6, or panel b (C18:2/C18:1 and C18:3/C18:1) has high specificities of more than 80.0%, with low sensitivities of less than 65.0%. In the validation set, decreased concentrations of the abovementioned variables were also significantly associated with BC (p b 0.0001) (Fig. 2 and Supplementary Table S2). The AUC values of C16:1, C20:4, C18:2/C18:1, PUFA, panel a, or panel b are more than 0.8, but only panel a has the sensitivity and specificity of more than 80% (Table 3). The ROC curve for panel a is shown in Fig. 3b. 3.3. Association of the FFA concentrations with BC stages
Table 2 Linear range of concentration, calibration equation, coefficient correlation (R2), and the stability. FAs
C16:1 C18:3 C18:2 C18:1 C20:4 C22:6
Linearity (n = 3)
QC (%)
FA (nM)
Equation
8.5−684.0 2.0−163.5 14.2−1138.0 20.9−1672.0 4.6−366.0 1.4−112.3
Y Y Y Y Y Y
= = = = = =
0.497 (±0.013)X 1.236 (±0.066)X 1.379 (±0.069)X 1.429 (±0.090)X 1.871 (±0.136)X 1.476 (±0.087)X
R2 + − + + + −
0.152 (±0.007) 0.016 (±0.009) 0.326 (±0.016) 1.205 (±0.265) 0.229 (±0.025) 0.019 (±0.014)
0.998 0.997 0.986 0.990 0.997 0.998
b18 b12 b12 b18 b19 b19
X: Concentration ratios of individual FAs to ISs (83.3 nM for C17:1 and 33.3 nM for C21:0); Y: corresponding intensity ratios of FAs to IS
As shown in Fig. 2, the scatter plots of cancer staging versus the concentrations of these variables indicate that decreased concentrations of them were remarkably associated with BC stages. The concentrations of C16:1, C18:3, C18:2, C20:4, C18:2/C18:1, and C18:3/C18:1 show a significant difference in patients with early- or advanced-stage BC compared with healthy controls (p b 0.05, Supplementary Table S2). In addition, C22:6 also has significant differences in patients with early stage BC compared with healthy controls (p = 0.001, Supplementary Table S2). The ROC analysis indicated that panel a has an excellent diagnostic ability to differentiate patients with early-stage BC from healthy controls, with an AUC value of 0.953, a sensitivity of 83.3%, and a specificity of 87.1%, and that panel b also exhibits an excellent diagnostic ability to differentiate patients with advanced-stage BC from healthy controls, with an AUC value of 0.921, a sensitivity of 83.3%, and a specificity of 90.0%
34
Y. Zhang et al. / Clinica Chimica Acta 437 (2014) 31–37
Fig. 2. Scatter plots of the levels of serum FFAs in the training set, the validation set and different stages of BC. *, p b 0.05; **, p b 0.01; ***, p b 0.001.
Y. Zhang et al. / Clinica Chimica Acta 437 (2014) 31–37
35
Table 3 AUC, sensitivity and specificity of FFAs between different groups based on the same cut-off values. Note: Sens: sensitivity; Spec: specificity; PUFA: polyunsaturated fatty acid. Panel a: C16:1, C18:3, C18:2, C20:4 and C22:6. Panel b: C18:2/C18:1 and C18:3/C18:1. Panel c: C16:1, C18:3, C18:2 and C20:4. FFAs
C16:1 C18:3 C18:2 C20:4 C22:6 C18:2/C18:1 C18:3/C18:1 PUFA Panel a Panel b Panel c
Training set
Validation set
Controls vs. BC
Controls vs. BC
Cut-off
Stage Controls vs. early stage
AUC (95% CI)
Sens (%)
Spec (%)
AUC (95% CI)
Sens (%)
Spec (%)
0.714 (.628–.801) 0.770 (.692–.849) 0.613 (.517–.708) 0.755 (.675–.835) 0.614 (.521–.708) 0.729 (.645–.813) 0.789 (.711–.867) 0.876 (.821–.932) 0.885 (.833–.938) 0.778 (.698–.857)
85.7 84.3 90.0 55.7 41.4 82.9 87.1 80.9 80.8 64.7
55.9 61.8 38.2 88.2 83.8 58.8 67.6 80.0 82.3 88.6
0.847 (.792–.902) 0.675 (.593–.756) 0.703 (.624–.782) 0.858 (.802–.915) 0.730 (.660–.800) 0.871 (.819–.923) 0.758 (.692–.824) 0.850 (.793–.906) 0.908 (.866–.950) 0.871 (.818–.924)
68.2 69.7 81.8 75.8 62.1 72.7 72.0 66.7 83.3 76.4
80.6 56.9 50.0 83.3 81.9 90.3 73.6 87.1 83.3 84.8
29.6 18.7 207.0 153.6 51.1 0.8 0.1 0.5 0.4 0.6 0.4
Controls vs. advanced stage
AUC (95% CI)
Sens (%)
Spec (%)
AUC (95% CI)
Sens (%)
Spec (%)
0.927 (.878–.976) 0.692 (.578–.806) 0.722 (.606–.839) 0.891 (.822–.960) 0.694 (.597–.792) 0.890 (.819–.960) 0.758 (.667–.849) 0.892 (.824–.960) 0.953 (.917–.988) 0.889 (.816–.962)
75.7 68.6 87.1 71.4 41.4 80.0 71.4 72.2 83.3 69.4
91.7 61.1 55.6 88.9 94.4 88.9 77.8 94.3 87.1 94.3
0.788 (.686–.890) 0.656 (.520–.791) 0.718 (.592–.844) 0.818 (.699–.938)
75.7 82.9 87.1 30.0
76.7 50.0 45.8 87.5
0.918 (.851–.986) 0.759 (.656–.862) 0.794 (.673–.915)
80.0 71.4 54.2
91.7 70.8 95.7
0.921 (.848–.993) 0.866 (.778–.954)
83.3 75.0
90.0 91.4
Note: Sens: sensitivity; Spec: specificity; PUFA: polyunsaturated fatty acid. Panel a: C16:1, C18:3, C18:2, C20:4 and C22:6. Panel b: C18:2/C18:1 and C18:3/C18:1. Panel c: C16:1, C18:3, C18:2 and C20:4.
(Table 3). The ROC curves for panels a and b are shown in Fig. 3c and d, respectively. 4. Discussion Serum FFAs are important bioactive molecules and the primary sources of energy in the body. In this study, the decreased concentrations of C16:1, C18:3, C18:2, C20:4, and C22:6, or ratios of C18:2/C18:1 and
C18:3/C18:1 were significantly associated with BC, which may be due to the increasing energy expenditure in cancer development. Previous study suggested that a decreased concentration of C18:3 and an increased concentration of C18:2 were significantly associated with BC, but the concentrations of C16:1, C20:4, and C22:6 were not related to BC [18]. These results are partly consistent with our data obtained in this study. As shown in Fig. 2, change trends of the concentrations of FFAs are observed in the following order: healthy controls N patients with
Fig. 3. The ROC curve analysis for different biomarker panels. (a) Panel a (a combination of C16:1, C18:3, C18:2, C20:4, and C22:6) for differentiating patients with BC from healthy controls in the training set. (b) Panel a for differentiating patients with BC from healthy controls in the validation set. (c) Panel a for differentiating patients with early stage BC from healthy controls. (d) Panel b (a combination of C18:2/C18:1 and C18:3/C18:1) for differentiating patients with advanced stage BC from healthy controls.
36
Y. Zhang et al. / Clinica Chimica Acta 437 (2014) 31–37
advanced-stage BC N patients with early-stage BC, but no statistical differences between patients with early-stage BC and advancedstage BC are observed. These results may suggest that a slight difference in FFA metabolism exists between early-stage BC and advanced-stage BC. Previous studies have found the anti-proliferative effects of n − 3 PUFAs on BC cells in vitro [19]. Emerging studies on the relationship between dietary fatty acids and BC have indicated that n−6 PUFAs have stronger tumor-enhancing effects, whereas n−3 PUFAs have protective effects on mammary cancer [20,21]. However, some studies revealed no association between dietary intake of fatty acids and BC risk [22], or an increased risk of BC [23]. In addition, increased ratio of n−6 PUFAs to n−3 PUFAs may be related to an increased risk of BC [20,24]. It is well known that fatty acids are main components of phospholipids and glycolipids, which were significantly associated with disease status [25–27]. Changes in the concentrations of serum metabolites may reflect heterogeneity in disease status. Monitoring fluctuations of some metabolites in biofluids has become one of vital approaches to detect early-stage BC [28]. Metabolomic studies on BC mainly focused on the metabolites in the urine or the tissue [29–31]. Nuclear magnetic resonance investigation of metabolites in pre- and post-operative serum samples from 44 patients with early-stage BC indicated that some metabolites from pre-operative patients had a good diagnostic accuracy, with a sensitivity of 75%, a specificity of 69%, and an AUC value of 0.72 [32]. A combination of nine miRNAs could discriminate between patients with early-stage BC and healthy controls, with the AUC of 0.665, but the association between miRNA expression and tumor grade, tumor size, menopausal status or lymph node status was not detected [33]. Our previous study has indicated that a combination of C16:1, C18:3, C18:2, C20:4 and C22:6 has a high diagnostic accuracy for differentiating pancreatic cancer from healthy controls, with the AUC of 0.93, the sensitivity of 86%, and the specificity of 85%, and that a combination of the concentration ratios of C 18:2 / C18:1 and C18:3/C18:1, provides an excellent diagnostic accuracy to differentiate early-stage pancreatic cancer from the patients with pancreatitis plus normal controls, with the AUC of 0.912, the sensitivity of 86.7%, and the specificity of 88.6% [15].
5. Conclusions NanoESI-FTICR MS is a high-sensitivity and high-throughput technology that can analyze complex biological samples without cumbersome preparation steps. In the present study, chip-based direct-infusion nanoESI-FTICR MS in the negative mode was carried out to detect FFA concentrations in healthy controls and BC patients. The identified potential biomarkers could be used to distinguish BC patients from healthy controls, especially for early-stage patients. Hence, the measurement of FFAs in biological samples by a high-efficiency technology might be very helpful in the early diagnosis of BC.
Abbreviations BC breast cancer FFAs free fatty acids CBDInanoESI-FTICR MS chip-based direct infusion nanoelectrospray ionization-Fourier transform ion cyclotron resonance mass spectrometry.
Acknowledgments This study was funded by grant 91029701 from the National Natural Science Foundation of China (to Z. Li).
Appendix A. Supplementary data Supplementary data to this article can be found online at http://dx. doi.org/10.1016/j.cca.2014.07.001. References [1] Gotzsche PC, Jorgensen KJ. Screening for breast cancer with mammography. Cochrane Database Syst Rev 2013;6:CD001877. [2] Smith LF. Palpable cancer of the breast and negative mammography: the ongoing dilemma. J Surg Oncol 2007;96:451–2. [3] Zhang AH, Sun H, Qiu S, Wang XJ. Metabolomics in noninvasive breast cancer. Clin Chim Acta 2013;424:3–7. [4] Vlahou A, Laronga C, Wilson L, Gregory B, Fournier K, McGaughey D, et al. A novel approach toward development of a rapid blood test for breast cancer. Clin Breast Cancer 2003;4:203–9. [5] Pedersen AC, Sorensen PD, Jacobsen EH, Madsen JS, Brandslund I. Sensitivity of CA 15-3, CEA and serum HER2 in the early detection of recurrence of breast cancer. Clin Chem Lab Med 2013;51:1511–9. [6] Agrawal AK, Jelen M, Rudnicki J, Grzebieniak Z, Zysko D, Kielan W, et al. The importance of preoperative elevated serum levels of CEA and CA15-3 in patients with breast cancer in predicting its histological type. Folia Histochem Cytobiol 2010;48:26–9. [7] Ng EK, Li R, Shin VY, Jin HC, Leung CP, Ma ES, et al. Circulating microRNAs as specific biomarkers for breast cancer detection. PLoS One 2013;8:e53141. [8] Zhao H, Shen J, Medico L, Wang D, Ambrosone CB, Liu S. A pilot study of circulating miRNAs as potential biomarkers of early stage breast cancer. PLoS One 2010;5: e13735. [9] Ring A, Smith IE, Dowsett M. Circulating tumour cells in breast cancer. Lancet Oncol 2004;5:79–88. [10] Mostert B, Sleijfer S, Foekens JA, Gratama JW. Circulating tumor cells (CTCs): detection methods and their clinical relevance in breast cancer. Cancer Treat Rev 2009;35:463–74. [11] Liu L, Li Y, Guan C, Li K, Wang C, Feng R, et al. Free fatty acid metabolic profile and biomarkers of isolated post-challenge diabetes and type 2 diabetes mellitus based on GC–MS and multivariate statistical analysis. J Chromatogr B Analyt Technol Biomed Life Sci 2010;878:2817–25. [12] Yi LZ, He J, Liang YZ, Yuan DL, Chau FT. Plasma fatty acid metabolic profiling and biomarkers of type 2 diabetes mellitus based on GC/MS and PLS–LDA. FEBS Lett 2006;580:6837–45. [13] Zhang Y, Wang Y, Guo S, Guo Y, Liu H, Li Z. Ammonia-treated N-(1-naphthyl) ethylenediamine dihydrochloride as a novel matrix for rapid quantitative and qualitative determination of serum free fatty acids by matrix-assisted laser desorption/ ionization-Fourier transform ion cyclotron resonance mass spectrometry. Anal Chim Acta 2013;794:82–9. [14] Wang DC, Sun CH, Liu LY, Sun XH, Jin XW, Song WL, et al. Serum fatty acid profiles using GC–MS and multivariate statistical analysis: potential biomarkers of Alzheimer's disease. Neurobiol Aging 2012;33:1057–66. [15] Zhang Y, Qiu L, Wang Y, Qin X, Li Z. High-throughput and high-sensitivity quantitative analysis of serum unsaturated fatty acids by chip-based nanoelectrospray ionization-Fourier transform ion cyclotron resonance mass spectrometry: early stage diagnostic biomarkers of pancreatic cancer. Analyst 2014;139:1697–706. [16] Wiest MM, German JB, Harvey DJ, Watkins SM, Hertz-Picciotto I. Plasma fatty acid profiles in autism: a case–control study. Prostaglandins Leukot Essent Fatty Acids 2009;80:221–7. [17] El-Ansary AK, Bacha AG, Al-Ayahdi LY. Plasma fatty acids as diagnostic markers in autistic patients from Saudi Arabia. Lipids Health Dis 2011;10:62. [18] Lv W, Yang T. Identification of possible biomarkers for breast cancer from free fatty acid profiles determined by GC–MS and multivariate statistical analysis. Clin Biochem 2012;45:127–33. [19] Bougnoux P, Hajjaji N, Maheo K, Couet C, Chevalier S. Fatty acids and breast cancer: sensitization to treatments and prevention of metastatic re-growth. Prog Lipid Res 2010;49:76–86. [20] Thiebaut AC, Chajes V, Gerber M, Boutron-Ruault MC, Joulin V, Lenoir G, et al. Dietary intakes of omega-6 and omega-3 polyunsaturated fatty acids and the risk of breast cancer. Int J Cancer 2009;124:924–31. [21] Gago-Dominguez M, Yuan JM, Sun CL, Lee HP, Yu MC. Opposing effects of dietary n− 3 and n− 6 fatty acids on mammary carcinogenesis: the Singapore Chinese Health Study. Br J Cancer 2003;89:1686–92. [22] Chajes V, Torres-Mejia G, Biessy C, Ortega-Olvera C, Angeles-Llerenas A, Ferrari P, et al. Omega-3 and omega-6 polyunsaturated fatty acid intakes and the risk of breast cancer in Mexican women: impact of obesity status. Cancer Epidemiol Biomarkers Prev 2012;21:319–26. [23] Stripp Connie, Overvad Kim, Christensen Jane, Thomsen Birthe L, Olsen Anja, Møller Susanne, Tjønneland A. Fish intake is positively associated with breast cancer incidence rate. Nutr Cancer 2003;133:3664–9. [24] Murff HJ, Shu XO, Li H, Yang G, Wu X, Cai H, et al. Dietary polyunsaturated fatty acids and breast cancer risk in Chinese women: a prospective cohort study. Int J Cancer 2011;128:1434–41. [25] Franky Dhaval S, Shilin Nandubhai S, Pankaj Manubhai S, Patel HR, Prabhudas Shankerbhai P. Significance of alterations in plasma lipid profile levels in breast cancer. Integr Cancer Ther 2008;7:33–41. [26] Hilvo M, Denkert C, Lehtinen L, Muller B, Brockmoller S, Seppanen-Laakso T, et al. Novel theranostic opportunities offered by characterization of altered membrane lipid metabolism in breast cancer progression. Cancer Res 2011;71:3236–45.
Y. Zhang et al. / Clinica Chimica Acta 437 (2014) 31–37 [27] Abdelsalam KE, Hassan IK, Sadig IA. The role of developing breast cancer in alteration of serum lipid profile. J Res Med Sci 2012;17:562–5. [28] Tenori L, Oakman C, Claudino WM, Bernini P, Cappadona S, Nepi S, et al. Exploration of serum metabolomic profiles and outcomes in women with metastatic breast cancer: a pilot study. Mol Oncol 2012;6:437–44. [29] Woo HM, Kim KM, Choi MH, Jung BH, Lee J, Kong G, et al. Mass spectrometry based metabolomic approaches in urinary biomarker study of women's cancers. Clin Chim Acta 2009;400:63–9. [30] Chen Y, Zhang R, Song Y, He J, Sun J, Bai J, et al. RRLC–MS/MS-based metabonomics combined with in-depth analysis of metabolic correlation network: finding potential biomarkers for breast cancer. Analyst 2009;134:2003–11.
37
[31] Nam H, Chung BC, Kim Y, Lee K, Lee D. Combining tissue transcriptomics and urine metabolomics for breast cancer biomarker identification. Bioinformatics 2009;25:3151–7. [32] Oakman C, Tenori L, Claudino WM, Cappadona S, Nepi S, Battaglia A, et al. Identification of a serum-detectable metabolomic fingerprint potentially correlated with the presence of micrometastatic disease in early breast cancer patients at varying risks of disease relapse by traditional prognostic methods. Ann Oncol 2011;22:1295–301. [33] Kodahl AR, Lyng MB, Binder H, Cold S, Gravgaard K, Knoop AS, et al. Novel circulating microRNA signature as a potential non-invasive multi-marker test in ER-positive early-stage breast cancer: a case–control study. Mol Oncol 2014;8:874–83.
