Pharmacometabonomics analysis reveals serum formate and acetate

Subscriber access provided by READING UNIV

Article

Pharmacometabonomics analysis reveals serum formate and acetate potentially associated with varying response to gemcitabinecarboplatin chemotherapy in metastatic breast cancer patients Limiao Jiang, Soo Chin Lee, and Thian C Ng J. Proteome Res., Just Accepted Manuscript • DOI: 10.1021/acs.jproteome.7b00859 • Publication Date (Web): 01 Feb 2018 Downloaded from http://pubs.acs.org on February 2, 2018

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

Journal of Proteome Research is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 37 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Journal of Proteome Research

Pharmacometabonomics analysis reveals serum formate and acetate potentially associated with varying response to gemcitabine-carboplatin chemotherapy in metastatic breast cancer patients Limiao Jiang,*,† ,‡ Soo Chin Lee,§,# and Thian C Ng‡ †

Department of Epidemiology and Biostatistics, MOE Key Lab of Environment and Health,

School of Public Health, Tongji Medical College, Huazhong University of Science and Technology, 13 Hangkong Road, Wuhan 430030, China. ‡

Department of Diagnostic Radiology, National University of Singapore, 5 Lower Kent

Ridge Road, 119074, Singapore. §

Department of Haematology-Oncology, National University Cancer Institute, National

University Health System, 5 Lower Kent Ridge Road, 119074, Singapore. #

Cancer Science Institute of Singapore, National University of Singapore, 14 Medical

Drive, 117599, Singapore.

*Corresponding author. Address: School of Public Health, Tongji Medical College, Huazhong University of Science and Technology, 13 Hangkong Road, Wuhan 430030, China. Tel: +86-15827530037, E-mail: [email protected]

1

ACS Paragon Plus Environment

Journal of Proteome Research 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 2 of 37

Abstract

Gemcitabine-carboplatin (GC) chemotherapy was efficacious in metastatic breast cancer (MBC) patients probably resistant to anthracyclines and taxanes, but showed significant inter-individual variation in treatment responses. Early prediction of response to treatment is clinically relevant to identify patients who will achieve clinical benefit. In this study, nuclear magnetic resonance (NMR) based pharmacometabonomics was used to non-invasively predict the response to GC chemotherapy of 29 MBC patients with prior exposure to both anthracyclines and taxanes from a phase II study. Formate and acetate levels in the baseline serum collected prior to GC chemotherapy were identified as potential predictive markers to select patients who will achieve clinical benefit and to identify those who should not be treated with the therapy to avoid futile treatment. The significantly lower baseline levels of serum formate and acetate in patients with resistant disease may reflect the higher demand of them as alternate/additional nutritional sources to fuel the accelerated proliferation of breast cancer cells that are biologically more aggressive and/or resistant to therapy. The results suggest that pharmacometabonomics can be a potential useful tool for predicting chemotherapy response in the context of precision medicine. Prospective studies with larger patient cohorts are required for validation of the findings. Keywords:

pharmacometabonomics,

pharmacometabolomics,

gemcitabine-carboplatin, chemotherapy

2


breast

cancer,

Page 3 of 37 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Introduction Breast cancer is the most commonly diagnosed cancer and leading cause of cancer death among females worldwide 1. Substantial efforts have been put into the development of drugs for breast cancer treatment. The anthracyclines and taxanes are the two most active current classes of chemotherapeutic agents in the adjuvant or neoadjuvant settings of breast cancer treatment 2. Therefore, the resistance to the two drugs is increasing in relapsing or progressive patients, which requires the use of other drugs for metastatic breast cancer (MBC) patients. The gemcitabine-carboplatin (GC) combination chemotherapy has been shown to have clinical activity in MBC patients in several phase II/III clinical trials

3-8

. While the

combination was efficacious, there was significant inter-individual variation in treatment responses and substantial hematologic toxicities 3, 4, 9, 10. So, the early prediction of response to such combination therapy is of clinical relevance to reduce unnecessary exposure to an ineffective regimen, to avoid unnecessary side effects, and to save time and cost. Currently a variety of approaches have been utilized for predicting the chemotherapy response of breast cancer patients, including conventional imaging modalities 11, molecular subtyping 12, and tumor- or blood-based biomarkers derived from “omics” approaches (e.g., genomics, proteomics and metabolomics). Although breast cancer subtyping based on the expression of estrogen receptor (ER), progesterone receptor (PR) and human epidermal growth factor receptor 2 (HER2) showed potential for predicting response to neoadjuvant chemotherapy

12

, it currently requires invasive biopsy

13

3


and the suboptimal performance


limits the wide applicability

14

. In comparison, blood-based biomarkers derived from

‘omics’ techniques are less invasive and generally precede changes observed from imaging techniques, and hence can potentially be better markers for the early prediction of treatment responses. Pharmacometabonomics emerged in the past decade as a new paradigm for predicting drug responses 15. It was defined as “the prediction of the outcome (for example, efficacy or toxicity) of a drug or xenobiotic intervention in an individual based on a mathematical model of pre-intervention metabolite signatures”

16

. Because of its sensitivity to both

genomic and environmental factors, pharmacometabonomics shows advantage to pharmacogenomics and plays an essential role in personalized medicine. It has been successfully used in human for the prediction of drug metabolism 17, treatment response 18 , and adverse effect

19

. A few studies have applied pharmacometabonomics for the

identification of potential biomarker candidates that can predict response to treatments of breast cancer

14, 20-22

. The baseline levels of four serum metabolites (i.e., threonine,

glutamine, isoleucine and linolenic acid) could predict the treatment response of locally advanced breast cancer patients to epirubicin-cyclophosphamide-docetaxel-trastuzumab neoadjuvant chemotherapy

14

. Baseline circulating spermidine and tryptophan were found

as potential biomarkers associated with the complete pathological response to trastuzumab-paclitaxel neoadjuvant therapy in locally advanced HER-2 positive breast cancer patients 22.

4


Page 4 of 37

Page 5 of 37 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


In this study, we explored the utility of pharmacometabonomics approach for identifying pre-treatment serum metabolic features associated with varying response to GC chemotherapy in MBC patients with prior exposure to both anthracyclines and taxanes from a phase II study. The aim of the present study was to find metabolites whose basal levels can predict the treatment response of MBC patients to GC chemotherapy. It indicated that the outcome of chemotherapy may be associated with the baseline metabolic profile of patients.

Materials and methods

Patient recruitment and treatment regimen Patient selection Forty-two MBC patients with prior exposure to anthracyclines and taxanes and who had measurable disease were enrolled into a single-centre, open-label, phase II study, following the specific inclusion and exclusion criteria

9, 10

. In this study, all participants were

non-smokers. Patients who had prior treatment with gemcitabine, who received cytotoxic treatments within the preceding 30 days, or had concomitant medical conditions (e.g. active infections, diabetes, and cardiovascular diseases) or active infections were excluded. The study was approved by the institution ethics review board, and all subjects provided written informed consent. Patients were treated with intravenous gemcitabine at 1000 mg/m2 on days 1 and 8, and carboplatin (area under the curve [AUC] of 5 using the Calvert formula23) on day 1 of a 21-day cycle, for a maximum of 6 cycles 5


9, 10

. Treatment response was


evaluated radiologically with CT scans every 2 cycles 9, using the RECIST criteria

Page 6 of 37

24

, as

complete response (CR), partial response (PR), stable disease (SD), or progressive disease (PD). Patients were categorized as having clinical benefit if they achieved CR, PR or SD as their best response. Serum sample collection Blood was drawn before the first cycle of GC chemotherapy and then left to clot at room temperature for 45 minutes. After centrifugation at 1610 g for 10 min at room temperature, the supernatant serum was collected, apportioned and immediately stored at -80°C until NMR spectroscopic analysis. Baseline serum samples from 29 patients were available for metabolic profiling by NMR spectroscopy for this study. Chemicals Analytical grade NaCl, K2HPO4·3H2O and NaH2PO4·2H2O were purchased from Sigma-Aldrich (MO, USA). Deuterium oxide (D2O, 99.9% D) was purchased from Cambridge Isotope Laboratories, Inc. (MA, USA). The phosphate saline buffer for serum NMR analysis was prepared by dissolving K2HPO4 and NaH2PO4 in water (45 mM, pH≈7.40, K2HPO4/NaH2PO4 = 4:1) containing 0.9% NaCl and 50% D2O 25. NMR sample preparation, spectra acquisition and processing After being thawed on ice and vortex, 30 µL of each serum sample was mixed with 30 µL buffer. After vortex and centrifugation (14,489 g, 4°C, 10mins), the supernatant was transferred into a 1.7 mm NMR tube (CortecNet, France) for NMR analysis 26, 27.

6


Page 7 of 37 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


All NMR spectra were acquired at 302 K on a Bruker Avance 800 MHz NMR spectrometer (800.15 MHz for proton frequency) equipped with a 5 mm cryoprobe (Bruker Biospin, Germany). For each serum sample, two one-dimensional (1D) 1H NMR spectra were acquired, using relaxation editing and diffusion editing techniques respectively. The relaxation-edited spectra, selectively highlight signals from small molecule metabolites, were acquired using the Carr-Purcell-Meiboom-Gill (CPMG) pulse sequence

28, 29

(recycle

delay-90°-(τ-180°-τ)n-acquisition), with τ=350 µs, n=70 and water pre-saturation during the recycle delay (2 s). A total of 256 scans were collected into 32,768 data points over a spectral width of 16,025 Hz with a 90° pulse length adjusted to around 11 µs. The diffusion-edited spectra, largely enhance signals from macromolecules (e.g., lipoproteins), were acquired using the bipolar pulse pair longitudinal-eddy-current decay (BPP-LED) pulse sequence 29, 30 (recycle delay–90°–G1–τ–180°–G1*(-1)–τ–90°–G2–τ–△–90°–G1–τ–180°–G1* (-1)–τ–90°–G3–τ–Te–90°–acquisition), with diffusion time = 200 ms, eddy current delay (Te) = 5 ms, and delay for gradient recovery (τ) = 200 µs. Water pre-saturation was performed during recycle delay, △ and Te. G1 is a pulsed-field gradient (1.1 ms), and G2 and G3 are spoil gradients (0.6 ms each), with a ratio of G1 : G2 : G3 = 85%: -17.13%: -13.17%. All other parameters are the same as in the relaxation-edited spectra, except that 128 transients were collected. In order to have the best achievable SNR for the very limited sample volume (i.e. 30 uL), automatic receiver gain calculation was performed during the CPMG experiment for each sample. All 1D 1H NMR spectra were acquired in a random order for all samples. For the assistance of metabolite assignment, a series of two-dimensional (2D) NMR spectra

7



were acquired for pooled samples and processed as described previously

Page 8 of 37

31

., including 1H

J-Resolved Spectroscopy (JRES), 1H-1H Correlation Spectroscopy (COSY), 1H-1H Total Correlation Spectroscopy (TOCSY), 1H-13C Heteronuclear Single Quantum Correlation Spectroscopy (HSQC) and 1H-13C Heteronuclear Multiple Bond Correlation Spectroscopy (HMBC). The acquired 1D spectra were processed as below using Topspin 2.0 (Bruker Biospin, Germany). After Fourier transformation with 1 Hz exponential line broadening and 65,536 data points zero-filling, the spectra were manually corrected for phase and baseline distortions, and referenced to the anomeric proton of α-glucose (δ 5.23). The selected spectral regions (δ 0.50-8.50 for relaxation-edited spectra, δ 0.50-5.50 for diffusion-edited spectra) were segmented into discrete bins (0.003 ppm each in width), with the exclusion of regions containing imperfect water suppression (δ 4.36-5.17) and urea signal (δ 5.50-6.00). For a specific metabolite, from all its peaks, a characteristic peak (Table S1) having least overlap with the peaks from other metabolites was chosen for integration. If the characteristic peak from one metabolite may still have potential overlapping with other signals, its integration was derived from curve fitting, which was done for acetate, acetone, lysine, ornithine, choline, phosphocholine, glycerophosphocholine, and betaine. Spectra segmentation, curve fitting and integration was performed using MestReNova (Version 9.0.1, Mestrelab Research S. L., Spain). The concentration of a certain metabolite was represented by the integration of its least overlapped NMR signal region. For each identified metabolite in CPMG spectra, its concentration was normalized to the total sum of all

8


Page 9 of 37 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


identified metabolites prior to subsequent data analysis. Such normalization was applied for the elimination of receiver gain difference induced variation and for the ease of comparison with closely related studies14,

32, 33

. The fatty acid composition was calculated in the

diffusion-edited spectra according to previous report Unsaturated

Fatty

Acids

(UFA%),

34

, including the percentage of

PolyUnsaturated

Fatty

Acids

(PUFA%),

MonoUnsaturated Fatty Acids (MUFA%), and Saturated Fatty Acids (SFA%) in Total Fatty Acids (TFA), along with their ratios PUFA%/UFA%, PUFA%/MUFA%. The calculation was based on the spectra integrals for TFA (δ 0.85, -CH3), PUFA (δ 2.77, =CH-CH2-CH=), and UFA (δ 5.28, -CH=CH-), along with the consideration of proton numbers and the assumption that MUFA% = UFA%-PUFA%, SFA% = 1-UFA%.

Statistical analysis All metabolomics data analysis was performed using MetaboAnalyst 3.0

35

. Based on

common clinical practice and the suggestion of experienced breast cancer clinicians, patients were grouped as achieving clinical benefit (i.e., CR + PR + SD) or having progressive disease (i.e., PD), according to their best treatment response. Firstly, the concentration of identified metabolites in CPMG spectra were auto-scaled and subjected to Partial Least Squares - Discriminant Analysis (PLS-DA) based multivariate analysis using the group information as the Y-matrix. A 10-fold internal cross-validation was performed to evaluate the quality of the PLS-DA model. The model quality was further assessed by a permutation test with 1000 permutations. Three-dimensional (3D) scores plot was used to

9



indicate the overall difference of the metabolome between the two classes. The Variable Importance in Projection (VIP) scores were used to select the potential metabolites that contribute to the separation of the two groups. Then, as a univariate analysis approach, the un-paired t-test was further used to identify differential metabolites between the two groups, taking into account the problem of false discovery rate (FDR) for multiple testing. Taking together, metabolites with VIP scores >1.0 and FDR adjusted p < 0.05 were identified as the differential metabolites. Finally, Receiver Operating Characteristic (ROC) curves analysis on the identified differential metabolites was used to determine their diagnostic power as potential biomarkers (performed to evaluate the effectiveness of potential biomarkers). The area under the curve (AUC) and the cutoff value maximizing sensitivity and specificity was shown in the ROC curves, with the areas under curve (AUC) of ROC curves indicating the overall prediction quality. The difference of fatty acids composition between the two groups were also tested using un-paired t-test with FDR adjustment. For patient demographics, statistical analysis was performed using the SPSS statistical software (V17, SPSS Inc., Armonk, NY, USA). Statistical tests were two-sided unpaired t-test for quantitative data and Chi2 test for qualitative data, with a p value of less than 0.05 considered to be statistically significant.

Results

Patient Demographics

10


Page 10 of 37

Page 11 of 37 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Demographic data for the 29 patients are listed in Table 1. One patient (3.4%) achieved complete response, 13 patients (44.8%) achieved partial response, 8 patients (27.6%) achieved stable disease, and 7 (24.1%) patients had progressive disease to GC chemotherapy. Almost all the patients are from the ethnic groups of Chinese, Malay or Indian. The majority of the patients had visceral metastases. There are no significant difference in the race, age, body mass index (BMI), hormone receptor status, HER2 status, metastases or prior chemotherapy between patients achieving clinical benefit (CR + PR + SD) and those having progressive disease (PD). Metabolite assignment In the acquired serum 1H NMR spectra, a total of 31 metabolites were assigned (Supplementary Table S1) on the basis of 2D NMR experiments, with reference to literature reports

36

and the Human Metabolome Database (HMDB)

37

. Representative

relaxation-edited 1H NMR spectra of baseline serum samples from patients with different responses to chemotherapy were shown in Figure 1. The serum NMR spectra mainly contains signals from a variety of amino acids (isoleucine, leucine, valine, threonine, alanine, glutamine, lysine, ornithine, glycine, tyrosine, phenylalanine, histidine), amino acid derivatives (creatine, creatinine, betaine), different moieties of lipids (CH3, (CH2)n, CH2CO, CH2C=C, C=CCH2C=C, CH=CH), ketone bodies (3-D-hydroxybutyrate, acetone, acetoacetate), choline metabolites (choline, phosphocholine, glycerophosphocholine), carbohydrate metabolism related metabolites (D-glucose, lactate, pyruvate, citrate), N-acetyl glycoproteins, and organic acids (acetate, formate).

11



Potential biomarkers identification From relaxation-edited spectra, clear differences in the metabolic profiles between patients achieving clinical benefit (CR + PR + SD) and those with progressive disease (PD) were shown in the 3D scores plot of supervised PLS-DA (Figure 2), where each point represents the metabolic profile of a patient. The model quality was initially assessed by a 10-fold internal cross-validation, with R2=0.51, Q2=0.45. Then it was further assessed by a permutation test with 1000 permutations (Figure S3). Significant differential metabolites between clinically benefitted and non-benefitted patients were firstly selected by VIP scores from PLS-DA model. Based on the criteria of VIP > 1, 15 metabolites were selected as potential differential metabolites, i.e. Formate, Acetate, Threonine, Glutamine, Citrate, Methionine, Lipid (CH2)n, Lactate, Creatinine, Phenylalanine, Glycerophosphocholine, Acetoacetate, Tyrosine, Phosphocholine, Lipid C=CCH2C=C (Figure 3). Un-paired t-test was further used to identify potential discriminant metabolites between the two groups. Both formate and acetate showed p < 0.05 (Figure 4). Finally, by combining the criteria of VIP > 1 and p < 0.05, serum formate and acetate were selected as differential metabolites between clinically benefitted and non-benefitted patients. Based on diffusion-edited spectra, there were no significant difference in the fatty acids composition between patients achieving clinical benefit (CR + PR + SD) and those with progressive disease (PD), including UFA%, PUFA%, MUFA%, SFA%, PUFA%/UFA%, and PUFA%/MUFA% (Figure S1, Figure S2).

12


Page 12 of 37

Page 13 of 37 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


ROC analysis The performance of the two most differential serum metabolites for the prediction of GC chemotherapy response was evaluated using ROC curves. The baseline serum levels of formate and acetate was used to construct their ROC curves. The AUC [95% confidence interval] was 0.831 [0.669-0.961] for formate and 0.792 [0.565-0.948] for acetate (Figure 5). Based on the ROC curve of formate, it was possible to distinguish patients who achieved clinical benefit from those who had progressive disease with a sensitivity of 0.8 and a specificity of 0.9, by using a threshold of 8.62e-05 for the formate level (Figure 5). Similarly, by using a threshold of 0.00144 for the acetate level, the sensitivity and specificity was also 0.8 and 0.9 respectively (Figure 5). As clearly shown in the box plots to the right side of the ROC curve, there was significantly lower levels of formate and acetate in patients who had progressive disease than those who achieved clinical benefit (Figure 5). However, as shown in Supplementary Table S2, there were no significant difference in the concentration of formate or acetate between different subtypes of breast cancer, i.e. Hormone receptor+ vs Hormone receptor-, HER2+ vs HER2-, both in patients achieving clinical benefit and those having progressive disease.

Discussion Circulating nucleic acids, proteins, metabolites and tumor cells are products of cancer cells, the tumor microenvironment, the host response and their dynamic interactions 38. They can be used as non-invasive blood-based biomarkers for cancer screening and treatment 13



Page 14 of 37

response evaluation 38. Pharmacometabonomics emerged as a new paradigm for predicting drug responses, showed advantage to pharmacogenomics and plays an essential role in personalized medicine. Due to the ease of access and rich metabolite information, blood sample is the most widely used sample type in pharmacometabonomics. In previous pharmacometabonomics studies on breast cancer, the chemotherapy treatments include a combination of taxane or anthracycline chemotherapy with a targeted anti-HER2 treatment, e.g., paclitaxel-trastuzumab neoadjuvant therapy in locally advanced HER-2

positive

breast

cancer

22

patients

,

epirubicin-cyclophosphamide-docetaxel-trastuzumab neoadjuvant chemotherapy in locally advanced breast cancer patients patients

20

.

Prior

to

the

14

, paclitaxel-lapatinib therapy in metastatic breast cancer

collection

of

baseline

samples,

there

were

no

chemotherapy/hormonal therapy treatment, or there were at least six months separating completion of adjuvant chemotherapy and disease relapse in these patients 14, 20, 22. While in the present study, we focused on predicting the treatment response to non-anthracyclines and non-taxanes based chemotherapy, i.e. gemcitabine-carboplatin, in MBC patients, who had prior exposure to both anthracyclines and taxanes. As these MBC patients received prior treatment by the widely used anthracyclines and taxanes, they might show increased resistance to the two drugs. But probably they will not show resistance to gemcitabine-carboplatin. We used NMR based pharmacometabonomics approach to predict the responses of 29 MBC patients to GC chemotherapy. We observed that, prior to GC treatment, baseline levels of serum formate and acetate could be used to distinguish patients

14


Page 15 of 37 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


who achieved clinical benefit from those who had progressive disease. These two small carboxylates may be used to select patients who should not be treated with GC chemotherapy to avoid unnecessary adverse side effects. Significantly lower baseline levels of serum formate and acetate were observed in breast cancer patients who progress than those who achieve clinical benefit after receiving GC chemotherapy. Similar metabolic changes were observed in two previous breast cancer studies not relating to treatment response prediction, e.g. lower serum formate level in recurrent breast cancer patients compared to those with no clinical evidence of disease after surgical treatment

32

, decreased urinary acetate and formate level in breast cancer patients

compared to healthy volunteers 39. Interestingly, decreased formate and acetate levels were also observed in the blood plasma of lung cancer patients compared to healthy controls 33. Formate, the simplest carboxylate, can be incorporated into RNA, DNA, and protein as a result of the insertion of its carbon via tetrahydrofolates into purine nucleotides, thymidine, methionine, and serine 40, 41. In vitro studies showed that breast cancer patients with disease progression had a more rapid uptake of formate-C14 into breast tumors than those with clinically static or regressing disease

42, 43

. Acetate, a 2-carbon fatty acid, is one of the

smallest nutrients in mammals and can be readily converted to acetyl-CoA by acetyl-CoA synthetases (ACSSs) 44. It is used as a nutritional source by human breast cancer cells in an acetyl-CoA synthetase 2 (ACSS2)-dependent manner, and supplied a significant fraction of the carbon within the fatty acid and phospholipid pools act as alternate fuels for other cancer cells

45

46, 47

. Acetate has also been found to

, e.g., brain tumor

15


48

and liver tumor 49. We


infer that the lower serum formate and acetate levels observed in patients with resistant disease may reflect the higher demand for formate and acetate as an alternate/additional nutritional source to fuel the accelerated proliferation of cancer cells in patients with refractory disease (Figure 6). These possible mechanism on breast cancer cell metabolism may provide useful information for the development of relevant therapeutic drugs. Metabolic profile differences between the molecular subtypes of human breast cancer has been explored

14, 50, 51

. The dependence of metabolic alterations on breast cancer subtypes

was observed in breast cancer tissues

50, 51

, but not in the serum 14. Here we also assessed

the effect of breast cancer subtypes on the performance of the two potential serum biomarkers, i.e. formate and acetate. However, no significant difference in their concentration was found between different subtypes of breast cancer, both for patients achieving clinical benefit and those having progressive disease. It indicated that the content of the two serum metabolites may not be significantly affected by the hormone receptor or HER2 status. There have been less reported studies of metabolomics within metastatic breast cancer 20, which is likely in part due to the greatly increased mutational load and heterogeneity in advanced disease 52. It may explain the relative large inter-subject variation in the content of some serum metabolites observed here, since all patients in this study had metastatic breast cancer. To the best of our knowledge, the present study is the first one for the early prediction of response to GC chemotherapy in breast cancer patients using pharmacometabonomics. It

16


Page 16 of 37

Page 17 of 37 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


suggests that pharmacometabonomics can be a useful tool for the prediction of chemotherapy response, which will provide useful personalized information for precision medicine

53, 54

. These findings reflect how baseline metabolic differences in the peripheral

blood of cancer patients may predict subsequent treatment response. Prospective studies with larger patient cohorts are required for the validation of the findings. Conclusion In the present study, we demonstrated that NMR-based serum pharmacometabonomics can non-invasively predict the response to GC chemotherapy in MBC patients previously exposed to anthracyclines and taxanes. Baseline levels of serum formate and acetate showed high potential for distinguishing patients who achieved clinical benefit from those who had progressive disease. The significantly lower serum levels of formate and acetate in patients with resistant disease may reflect the higher demand of them as alternate/additional nutritional sources to fuel the accelerated proliferation of breast cancer cells that are biologically more aggressive and/or resistant to GC chemotherapy. The findings from relative small sample size require further validation in prospective studies with larger patient cohorts. Nevertheless, it suggests that pharmacometabonomics can be a potential useful tool for predicting chemotherapy response in the context of precision medicine.

Associated content Supporting Information. Table S1: 1H and 13C NMR assignment for metabolites from serum of metastatic breast cancer patients; Table S2: Summary of p values for comparing the 17



concentration of formate and acetate between different subtypes of breast cancer; Figure S1: Fatty acid composition in MBC patients achieving clinical benefit (CR + PR + SD) and those who had progressive disease (PD); Figure S2: The result of univariate un-paired t-test for comparing the fatty acid composition between patients achieving clinical benefit (CR + PR + SD) and those who had progressive disease (PD); Figure S3: Permutation test (1000 permutations) plot for the PLS-DA model. Author information Corresponding Author *(L.J.) E-mail: [email protected]. Author Contributions L.J., S.C.L. and T.C.N designed experiments. S.C.L. collected samples. L.J., S.C.L. and T.C.N analyzed data, and L.J., S.C.L. wrote the paper. Notes The authors declare no competing financial interest. Acknowledgments The authors gratefully acknowledge the participation of all patients. This study was supported by grants from National University of Singapore (R-180-000-016-733), National Medical Research Council of Singapore (NMRC/CSI/0015/2009) and Huazhong University of Science and Technology (513-3004513113).

Abbreviations

18


Page 18 of 37

Page 19 of 37 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


1D, one-dimensional; 2D, two-dimensional; 3D, three-dimensional; ACSS, acetyl-CoA synthetase; ACSS2, acetyl-CoA synthetase 2; AUC, area under the curve; CPMG, Carr-Purcell-Meiboom-Gill; CR, complete response; CT, computed tomography; DNA, deoxyribonucleic acid; ER, estrogen receptor; GC, gemcitabine-carboplatin; HER2, human epidermal growth factor receptor 2; HMDB, human metabolome database; MBC, metastatic breast cancer; NMR, nuclear magnetic resonance; PD, progressive disease; PLS-DA, partial least squares - discriminant analysis; PR, partial response; PR, progesterone receptor; RECIST, response evaluation criteria in solid tumors; RNA, ribonucleic acid; ROC, receiver operating characteristic; SD, stable disease; VIP, variable importance in projection.

19



References

1.

Jemal, A.; Bray, F.; Center, M. M.; Ferlay, J.; Ward, E.; Forman, D. Global cancer statistics. Ca-Cancer J.

Clin. 2011, 61, 69-90. 2.

Bines, J.; Earl, H.; Buzaid, A. C.; Saad, E. D. Anthracyclines and taxanes in the neo/adjuvant treatment of

breast cancer: does the sequence matter? Ann. Oncol. 2014, 25, 1079-1085. 3.

Loesch, D.; Asmar, L.; McIntyre, K.; Doane, L.; Monticelli, M.; Paul, D.; Vukelja, S.; Orlando, M.;

Vaughn, L. G.; Zhan, F.; Boehm, K. A.; O'Shaughnessy, J. A. Phase II trial of gemcitabine/carboplatin (plus trastuzumab in HER2-positive disease) in patients with metastatic breast cancer. Clin. Breast Cancer 2008, 8, 178-186. 4.

Yardley, D. A.; Burris, H. A.; Simons, L.; Spigel, D. R.; Greco, E. A.; Barton, J. H.; Shipley, D.; Drosick,

D.; Hainsworth, J. D. A Phase II Trial of Gemcitabine/Carboplatin with or Without Trastuzumab in the First-Line Treatment of Patients with Metastatic Breast Cancer. Clin. Breast Cancer 2008, 8, 425-431. 5.

Xu, B. H.; Jiang, Z. F.; Kim, S. B.; Yu, S. Y.; Feng, J. F.; Malzyner, A.; del Giglio, A.; Chung, H. C.; Shen,

L. J.; Pen, D. L. K. Biweekly gemcitabine-paclitaxel, gemcitabine-carboplatin, or gemcitabine-cisplatin as first-line treatment in metastatic breast cancer after anthracycline failure: a phase II randomized selection trial. Beast Cancer 2011, 18, 203-212. 6.

O'Shaughnessy, J.; Schwartzberg, L.; Danso, M. A.; Miller, K. D.; Rugo, H. S.; Neubauer, M.; Robert, N.;

Hellerstedt, B.; Saleh, M.; Richards, P.; Specht, J. M.; Yardley, D. A.; Carlson, R. W.; Finn, R. S.; Charpentier, E.; Garcia-Ribas, I.; Winer, E. P. Phase III Study of Iniparib Plus Gemcitabine and Carboplatin Versus Gemcitabine and Carboplatin in Patients With Metastatic Triple-Negative Breast Cancer. J. Clin. Oncol. 2014, 32, 3840-3847. 7.

Yardley, D. A.; Brufsky, A.; Coleman, R. E.; Conte, P. F.; Cortes, J.; Glueck, S.; Nabholtz, J. M. A.;

O'Shaughnessy, J.; Beck, R. M.; Ko, A.; Renschler, M. F.; Barton, D.; Harbeck, N. Phase II/III weekly nab-paclitaxel plus gemcitabine or carboplatin versus gemcitabine/carboplatin as first-line treatment of patients with metastatic triple-negative breast cancer (the tnAcity study): study protocol for a randomized controlled trial. Trials 2015, 16, 575. 8.

O'Shaughnessy, J.; Osborne, C.; Pippen, J. E.; Yoffe, M.; Patt, D.; Rocha, C.; Koo, I. C.; Sherman, B. M.;

Bradley, C. Iniparib plus Chemotherapy in Metastatic Triple-Negative Breast Cancer. N. Engl. J. Med. 2011, 364, 205-214. 9.

Chan, D.; Yeo, W. L.; Cordero, M. T.; Wong, C. I.; Chuah, B.; Soo, R.; Tan, S. H.; Lim, S. E.; Goh, B. C.;

Lee, S. C. Phase II study of gemcitabine and carboplatin in metastatic breast cancers with prior exposure to anthracyclines and taxanes. Invest. New Drugs 2010, 28, 859-865. 10. Wong, A. L. A.; Yap, H. L.; Yeo, W. L.; Soong, R.; Ng, S. S.; Wang, L. Z.; Cordero, M. T.; Yong, W. P.; Goh, B. C.; Lee, S. C. Gemcitabine and platinum pathway pharmacogenetics in asian breast cancer patients. Cancer Genomics Proteomics 2011, 8, 255-259. 11. Krishna, C. M.; Kurien, J.; Mathew, S.; Rao, L.; Maheedhar, K.; Kumar, K. K.; Chowdary, M. V. P. Raman spectroscopy of breast tissues. Expert Rev. Mol. Diagn. 2008, 8, 149-166. 12. Caudle, A. S.; Yu, T. K.; Tucker, S. L.; Bedrosian, I.; Litton, J. K.; Gonzalez-Angulo, A. M.; Hoffman, K.; Meric-Bernstam, F.; Hunt, K. K.; Buchholz, T. A.; Mittendorf, E. A. Local-regional control according to

20


Page 20 of 37

Page 21 of 37 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


surrogate markers of breast cancer subtypes and response to neoadjuvant chemotherapy in breast cancer patients undergoing breast conserving therapy. Breast Cancer Res. 2012, 14, R83. 13. Lloyd, S. M.; Arnold, J.; Sreekumar, A. Metabolomic profiling of hormone-dependent cancers: a bird's eye view. Trends Endocrinol. Metab. 2015, 26, 477-485. 14. Wei, S. W.; Liu, L. Y.; Zhang, J.; Bowers, J.; Gowda, G. A. N.; Seeger, H.; Fehm, T.; Neubauer, H. J.; Vogel, U.; Clare, S. E.; Raftery, D. Metabolomics approach for predicting response to neoadjuvant chemotherapy for breast cancer. Mol. Oncol. 2013, 7, 297-307. 15. Everett, J. R. From Metabonomics to Pharmacometabonomics: The Role of Metabolic Profiling in Personalized Medicine. Front. Pharmacol. 2016, 7, 297. 16. Clayton, T. A.; Lindon, J. C.; Cloarec, O.; Antti, H.; Charuel, C.; Hanton, G.; Provost, J. P.; Le Net, J. L.; Baker, D.; Walley, R. J.; Everett, J. R.; Nicholson, J. K. Pharmaco-metabonomic phenotyping and personalized drug treatment. Nature 2006, 440, 1073-1077. 17. Clayton, T. A.; Baker, D.; Lindon, J. C.; Everett, J. R.; Nicholson, J. K. Pharmacometabonomic identification of a significant host-microbiome metabolic interaction affecting human drug metabolism. Proc. Natl. Acad. Sci. U. S. A. 2009, 106, 14728-14733. 18. Nam, H. W.; Karpyak, V. M.; Hinton, D. J.; Geske, J. R.; Ho, A. M. C.; Prieto, M. L.; Biernacka, J. M.; Frye, M. A.; Weinshilboum, R. M.; Choi, D. S. Elevated baseline serum glutamate as a pharmacometabolomic biomarker for acamprosate treatment outcome in alcohol-dependent subjects. Transl. Psychiatr. 2015, 5, e621. 19. Backshall, A.; Sharma, R.; Clarke, S. J.; Keun, H. C. Pharmacometabonomic Profiling as a Predictor of Toxicity in Patients with Inoperable Colorectal Cancer Treated with Capecitabine. Clin. Cancer Res. 2011, 17, 3019-3028. 20. Tenori, L.; Oakman, C.; Claudino, W. M.; Bernini, P.; Cappadona, S.; Nepi, S.; Biganzoli, L.; Arbushites, M. C.; Luchinat, C.; Bertini, I.; Di Leo, A. Exploration of serum metabolomic profiles and outcomes in women with metastatic breast cancer: A pilot study. Mol. Oncol. 2012, 6, 437-444. 21. Tenori, L.; Oakman, C.; Morris, P. G.; Gralka, E.; Turner, N.; Cappadona, S.; Fornier, M.; Hudis, C.; Norton, L.; Luchinat, C.; Di Leo, A. Serum metabolomic profiles evaluated after surgery may identify patients with oestrogen receptor negative early breast cancer at increased risk of disease recurrence. Results from a retrospective study. Mol. Oncol. 2015, 9, 128-139. 22. Miolo, G.; Muraro, E.; Caruso, D.; Crivellari, D.; Ash, A.; Scalone, S.; Lombardi, D.; Rizzolio, F.; Giordano, A.; Corona, G. Phamacometabolomics study identifies circulating spermidine and tryptophan as potential biomarkers associated with the complete pathological response to trastuzumab-paclitaxel neoadjuvant therapy in HER-2 positive breast cancer. Oncotarget 2016, 7, 39809-39822. 23. Calvert, A. H.; Newell, D. R.; Gumbrell, L. A.; O'Reilly, S.; Burnell, M.; Boxall, F. E.; Siddik, Z. H.; Judson, I. R.; Gore, M. E.; Wiltshaw, E. Carboplatin dosage: prospective evaluation of a simple formula based on renal function. J. Clin. Oncol. 1989, 7, 1748-56. 24. Therasse, P.; Arbuck, S. G.; Eisenhauer, E. A.; Wanders, J.; Kaplan, R. S.; Rubinstein, L.; Verweij, J.; Van Glabbeke, M.; van Oosterom, A. T.; Christian, M. C.; Gwyther, S. G. New guidelines to evaluate the response to treatment in solid Tumors. J. Natl. Cancer Inst. 2000, 92, 205-216. 25. Jiang, L. M.; Zhao, X. J.; Huang, C. Y.; Lei, H. H.; Tang, H. R.; Wang, Y. L. Dynamic changes in metabolic profiles of rats subchronically exposed to mequindox. Mol. Biosyst. 2014, 10, 2914-2922. 26. Huang, C. Y.; Lei, H. H.; Zhao, X. J.; Tang, H. R.; Wang, Y. L. Metabolic influence of acute cyadox exposure on Kunming mice. J. Proteome Res. 2013, 12, 537-545.

21



27. Wu, J. F.; Holmes, E.; Xue, J.; Xiao, S. H.; Singer, B. H.; Tang, H. R.; Utzinger, J.; Wang, Y. L. Metabolic alterations in the hamster co-infected with Schistosoma japonicum and Necator americanus. Int. J. Parasit. 2010, 40, 695-703. 28. Meiboom, S.; Gill, D. Modified spin-echo method for measuring nuclear relaxation times. Rev. Sci. Instrum. 1958, 29, 688-691. 29. Beckonert, O.; Keun, H. C.; Ebbels, T. M. D.; Bundy, J. G.; Holmes, E.; Lindon, J. C.; Nicholson, J. K. Metabolic profiling, metabolomic and metabonomic procedures for NMR spectroscopy of urine, plasma, serum and tissue extracts. Nat. Protoc. 2007, 2, 2692-2703. 30. Wu, D. H.; Chen, A. D.; Johnson, C. S. An improved diffusion-ordered spectroscopy experiment incorporating bipolar-gradient pulses. J. Magn. Reson., Ser. A 1995, 115, 260-264. 31. Jiang, L. M.; Huang, J.; Wang, Y. L.; Tang, H. R. Metabonomic analysis reveals the CCl4-induced systems alterations for multiple rat organs. J. Proteome Res. 2012, 11, 3848-3859. 32. Asiago, V. M.; Alvarado, L. Z.; Shanaiah, N.; Gowda, G. A. N.; Owusu-Sarfo, K.; Ballas, R. A.; Raftery, D. Early detection of recurrent breast cancer using metabolite profiling. Cancer Res. 2010, 70, 8309-8318. 33. Rocha, C. M.; Carrola, J.; Barros, A. S.; Gil, A. M.; Goodfellow, B. J.; Carreira, I. M.; Bernardo, J.; Gomes, A.; Sousa, V.; Carvalho, L.; Duarte, I. F. Metabolic signatures of lung cancer in biofluids: NMR-based metabonomics of blood plasma. J. Proteome Res. 2011, 10, 4314-4324. 34. Xu, W. X.; Wu, J. F.; An, Y. P.; Xiao, C. N.; Hao, F. H.; Liu, H. B.; Wang, Y. L.; Tang, H. R. Streptozotocin-induced dynamic metabonomic changes in rat biofluids. J. Proteome Res. 2012, 11, 3423-3435. 35. Xia, J. G.; Sinelnikov, I. V.; Han, B.; Wishart, D. S. MetaboAnalyst 3.0-making metabolomics more meaningful. Nucleic Acids Res. 2015, 43, W251-W257. 36. Nicholson, J. K.; Foxall, P. J. D.; Spraul, M.; Farrant, R. D.; Lindon, J. C. 750 MHZ 1H and 1H-13C NMR spectroscopy of human blood plasma. Anal. Chem. 1995, 67, 793-811. 37. Wishart, D. S.; Tzur, D.; Knox, C.; Eisner, R.; Guo, A. C.; Young, N.; Cheng, D.; Jewell, K.; Arndt, D.; Sawhney, S.; Fung, C.; Nikolai, L.; Lewis, M.; Coutouly, M. A.; Forsythe, I.; Tang, P.; Shrivastava, S.; Jeroncic, K.; Stothard, P.; Amegbey, G.; Block, D.; Hau, D. D.; Wagner, J.; Miniaci, J.; Clements, M.; Gebremedhin, M.; Guo, N.; Zhang, Y.; Duggan, G. E.; MacInnis, G. D.; Weljie, A. M.; Dowlatabadi, R.; Bamforth, F.; Clive, D.; Greiner, R.; Li, L.; Marrie, T.; Sykes, B. D.; Vogel, H. J.; Querengesser, L. HMDB: the human metabolome database. Nucleic Acids Res. 2007, 35, D521-D526. 38. Hanash, S. M.; Baik, C. S.; Kallioniemi, O. Emerging molecular biomarkers-blood-based strategies to detect and monitor cancer. Nat. Rev. Clin. Oncol. 2011, 8, 142-150. 39. Slupsky, C. M.; Steed, H.; Wells, T. H.; Dabbs, K.; Schepansky, A.; Capstick, V.; Faught, W.; Sawyer, M. B. Urine metabolite analysis offers potential early diagnosis of ovarian and breast cancers. Clin. Cancer Res. 2010, 16, 5835-5841. 40. Bowen, D.; Folsch, E.; Guernsey, L. A. The interaction between fluoropyrimidines and methotrexate, and formate-C14 incorporation into nucleic acids and protein. Cancer Chemother. Pharmacol. 1980, 4, 111-116. 41. Stokstad, E. L.; Koch, J. Folic acid metabolism. Physiol. Rev. 1967, 47, 83-116. 42. Skypeck, H. H.; Kofman, S.; Taylor, S. G.; Winzler, R. J. The incorporation of formate-C14 into human breast cancer in vitro. Cancer Res. 1960, 20, 125-131. 43. Kofman, S.; Skypeck, H. H.; Perlia, C. P.; Economou, S. G.; Winzler, R. J.; Taylor, S. G. A correlation between the incorporation of formate-C14 in tumors and the clinical course of patients with disseminated breast cancer. Cancer 1960, 13, 425-431.

22


Page 22 of 37

Page 23 of 37 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


44. Boroughs, L. K.; DeBerardinis, R. J. Metabolic pathways promoting cancer cell survival and growth. Nat. Cell Biol. 2015, 17, 351-359. 45. Schug, Z. T.; Peck, B.; Jones, D. T.; Zhang, Q. F.; Grosskurth, S.; Alam, I. S.; Goodwin, L. M.; Smethurst, E.; Mason, S.; Blyth, K.; McGarry, L.; James, D.; Shanks, E.; Kalna, G.; Saunders, R. E.; Jiang, M.; Howell, M.; Lassailly, F.; Thin, M. Z.; Spencer-Dene, B.; Stamp, G.; van den Broek, N. J. F.; Mackay, G.; Bulusu, V.; Kamphorst, J. J.; Tardito, S.; Strachan, D.; Harris, A. L.; Aboagye, E. O.; Critchlow, S. E.; Wakelam, M. J. O.; Schulze, A.; Gottlieb, E. Acetyl-CoA synthetase 2 promotes acetate utilization and maintains cancer cell growth under metabolic stress. Cancer Cell 2015, 27, 57-71. 46. Alderton, G. K. Metabolism: acetate nourishes stressed tumour cells. Nat. Rev. Cancer 2015, 15, 67-67. 47. Keenan, M. M.; Chi, J. T. Alternative fuels for cancer cells. Cancer J. 2015, 21, 49-55. 48. Mashimo, T.; Pichumani, K.; Vemireddy, V.; Hatanpaa, K. J.; Singh, D. K.; Sirasanagandla, S.; Nannepaga, S.; Piccirillo, S. G.; Kovacs, Z.; Foong, C.; Huang, Z. G.; Barnett, S.; Mickey, B. E.; DeBerardinis, R. J.; Tu, B. P.; Maher, E. A.; Bachoo, R. M. Acetate is a bioenergetic substrate for human glioblastoma and brain metastases. Cell 2014, 159, 1603-1614. 49. Comerford, S. A.; Huang, Z.; Du, X.; Wang, Y.; Cai, L.; Witkiewicz, A. K.; Walters, H.; Tantawy, M. N.; Fu, A.; Manning, H. C.; Horton, J. D.; Hammer, R. E.; McKnight, S. L.; Tu, B. P. Acetate dependence of tumors. Cell 2014, 159, 1591-1602. 50. Budczies, J.; Brockmoller, S. F.; Muller, B. M.; Barupal, D. K.; Richter-Ehrenstein, C.; Kleine-Tebbe, A.; Griffin, J. L.; Oresic, M.; Dietel, M.; Denkert, C.; Fiehn, O. Comparative metabolomics of estrogen receptor positive and estrogen receptor negative breast cancer: alterations in glutamine and beta-alanine metabolism. J. Proteomics 2013, 94, 279-288. 51. Tang, X. H.; Lin, C. C.; Spasojevic, I.; Iversen, E. S.; Chi, J. T.; Marks, J. R. A joint analysis of metabolomics and genetics of breast cancer. Breast Cancer Res. 2014, 16, 415. 52. Hart, D. C.; Tenori, L.; Luchinat, C.; Di Leo, A., Metabolomics in Breast Cancer: Current Status and Perspectives. In Novel Biomarkers in the Continuum of Breast Cancer, Stearns, V., Ed. Springer International Publishing: Cham, 2016; pp 217-234. 53. Beger, R. D.; Dunn, W.; Schmidt, M. A.; Gross, S. S.; Kirwan, J. A.; Cascante, M.; Brennan, L.; Wishart, D. S.; Oresic, M.; Hankemeier, T.; Broadhurst, D. I.; Lane, A. N.; Suhre, K.; Kastenmuller, G.; Sumner, S. J.; Thiele, I.; Fiehn, O.; Kaddurah-Daouk, R.; Precision Med, P. Metabolomics enables precision medicine: "A White Paper, Community Perspective". Metabolomics 2016, 12, 149. 54. Everett, J. R. Pharmacometabonomics in humans: a new tool for personalized medicine. Pharmacogenomics 2015, 16, 737-754.

23



List of Tables Table 1. Clinical demographics for the patients in this study (n=29).

List of Figures Figure 1. Representative 800 MHz relaxation-edited 1H NMR spectra of baseline serum samples from patients with varying responses to subsequent chemotherapy. The regions in dash line were vertically expanded 32, 8 and 4 times respectively. CR, complete response; PR, partial response; SD, stable disease; PD, progressive disease; PC, phosphocholine; GPC, glycerophosphocholine. Figure 2. PLS-DA 3D scores plot for the analysis of 1H CPMG NMR spectra from the baseline serum of metastatic breast cancer patients with varying responses to subsequent chemotherapy. Each point represents the metabolic profile of a patient. CPMG, Carr-Purcell-Meiboom-Gill; CR, complete response; PR, partial response; SD, stable disease; PD, progressive disease. Figure 3. VIP scores plot derived from PLS-DA model showing potential metabolites (VIP>1) that significantly contribute to the discrimination between patients achieving clinical benefit (CR + PR + SD) and those who had progressive disease (PD). The colored boxes on the right indicate the relative concentrations of the corresponding metabolite in each group. CR, complete response; PR, partial response; SD, stable disease; PD, progressive disease.

24


Page 24 of 37

Page 25 of 37 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Figure 4. The result of univariate un-paired t-test for comparing the metabolite level difference between patients achieving clinical benefit (CR + PR + SD) and those who had progressive disease (PD). The x-axis indicates the compound ID, and the y-axis indicates –log10(p). The pink dots represent metabolites with FDR adjusted p < 0.05. Compound ID: 1, Formate; 2, Phenylalanine; 3, Tyrosine; 4, Histidine; 5, Unsaturated Lipid (-CH=CH-); 6, α-D-Glucose; 7, D-Mannose; 8, Threonine; 9, Lactate; 10, Creatinine; 11, Creatine; 12, Glycine; 13, Betaine; 14, Glycerophosphocholine; 15, Phosphocholine; 16, Choline; 17, Ornithine; 18, Lysine; 19, Polyunsaturated Lipid (C=CCH2C=C); 20, Methionine; 21, Citrate; 22, Glutamine; 23, Pyruvate; 24, Acetoacetate; 25, Acetone; 26, N-acetyl Glycoproteins (NAG); 27, Acetate; 28, Lipid CH2CH2CO; 29, Alanine; 30, Lipid (CH2)n; 31, 3-D-Hydroxybutyrate; 32, Dihydrothymine; 33, Valine; 34, Isoleucine; 35, Leucine; 36, Lipid CH3. Figure 5. Areas under the Receiving operating curve (ROC) for discriminating patients achieving clinical benefit (CR + PR + SD) and those who had progressive disease (PD). (A) formate, (B) acetate. AUC indicates the area under the ROC curve, with 95% confidence interval in brackets. The red dot in the AUC plot refers to the cutoff value maximizing sensitivity and specificity. The Box plot describes the distribution, median and interquartile ranges of metabolite concentration. CR: complete response, PR: partial response, SD: stable disease, PD: progressive disease. Figure 6. Metabolic pathways related to the metabolism of formate and acetate. ACSS2, acetyl-CoA synthetase short chain family member 2; AMP, adenosine monophosphate; ATP,

25



adenosine triphosphate; CoA, Coenzyme A; CR, complete response; CS, citrate synthase; LDH, L-lactate dehydrogenase; PD, progressive disease; PFL, pyruvate formate lyase (i.e., formate C-acetyltransferase); PR, partial response; SD, stable disease; *, p=4

6

2

1

3

5

6

2

2

Prior adjuvant/neoadjuvant

0.706

chemotherapy Yes

0

9

5

27


5


CR: complete response, PR: partial response, SD: stable disease, PD: progressive disease, HER2: human epidermal growth factor receptor 2. a

p-value was calculated using two-sided unpaired t-test for age and BMI, and using Chi2 test for all other data.

28


Page 28 of 37

Page 29 of 37 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Figure 1

Figure 1. Representative 800 MHz relaxation-edited 1H NMR spectra of baseline serum samples from patients with varying responses to subsequent chemotherapy. The regions in dash line were vertically expanded 32, 8 and 4 times respectively. CR, complete response; PR, partial response; SD, stable disease; PD, progressive disease; PC, phosphocholine; GPC, glycerophosphocholine.

29



Figure 2

Figure 2. PLS-DA 3D scores plot for the analysis of 1H CPMG NMR spectra from the baseline serum of metastatic breast cancer patients with varying responses to subsequent chemotherapy. Each point represents the metabolic profile of a patient. CPMG, Carr-Purcell-Meiboom-Gill; CR, complete response; PR, partial response; SD, stable disease; PD, progressive disease.

30


Page 30 of 37

Page 31 of 37 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Figure 3

Figure 3. VIP scores plot derived from PLS-DA model showing potential metabolites (VIP>1) that significantly contribute to the discrimination between patients achieving clinical benefit (CR + PR + SD) and those who had progressive disease (PD). The colored boxes on the right indicate the relative concentrations of the corresponding metabolite in each group. CR, complete response; PR, partial response; SD, stable disease; PD, progressive disease.

31



Page 32 of 37

Figure 4 Formate

Acetate

Figure 4. The result of univariate un-paired t-test for comparing the metabolite level difference between patients achieving clinical benefit (CR + PR + SD) and those who had progressive disease (PD). The x-axis indicates the compound ID, and the y-axis indicates –log10(p). The pink dots represent metabolites with FDR adjusted p < 0.05. Compound ID: 1, Formate; 2, Phenylalanine; 3, Tyrosine; 4, Histidine; 5, Unsaturated Lipid (-CH=CH-); 6, α-D-Glucose; 7, D-Mannose; 8, Threonine; 9, Lactate; 10, Creatinine; 11, Creatine; 12, Glycine; 13, Betaine; 14, Glycerophosphocholine; 15, Phosphocholine; 16, Choline; 17, Ornithine; 18, Lysine; 19, Polyunsaturated Lipid (C=CCH2C=C); 20, Methionine; 21, Citrate; 22, Glutamine; 23, Pyruvate; 24, Acetoacetate; 25, Acetone; 26, N-acetyl Glycoproteins (NAG); 27, Acetate; 28, Lipid CH2CH2CO; 29, Alanine; 30, Lipid (CH2)n; 31,

32


Page 33 of 37 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


3-D-Hydroxybutyrate; 32, Dihydrothymine; 33, Valine; 34, Isoleucine; 35, Leucine; 36, Lipid CH3.

33



Figure 5

34


Page 34 of 37

Page 35 of 37 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Figure 5. Areas under the Receiving operating curve (ROC) for discriminating patients achieving clinical benefit (CR + PR + SD) and those who had progressive disease (PD). (A) formate, (B) acetate. AUC indicates the area under the ROC curve, with 95% confidence interval in brackets. The red dot in the AUC plot refers to the cutoff value maximizing sensitivity and specificity. The Box plot describes the distribution, median and interquartile ranges of metabolite concentration. CR: complete response, PR: partial response, SD: stable disease, PD: progressive disease.

35



Figure 6

Figure 6. Metabolic pathways related to the metabolism of formate and acetate. ACSS2, acetyl-CoA synthetase short chain family member 2; AMP, adenosine monophosphate; ATP, adenosine triphosphate; CoA, Coenzyme A; CR, complete response; CS, citrate synthase; LDH, L-lactate dehydrogenase; PD, progressive disease; PFL, pyruvate formate lyase (i.e., formate C-acetyltransferase); PR, partial response; SD, stable disease; *, p

Pharmacometabonomics analysis reveals serum formate and acetate

Recommend Documents