MR spectroscopy in breast cancer metabolomics

Abstract Breast cancer poses a significant health care challenge worldwide requiring early detection and effective treatment strategies for better patient outcome. A deeper understanding of the breast cancer biology and metabolism may help developing better diagnostic and therapeutic approaches. Metabolomic studies give a comprehensive analysis of small molecule metabolites present in human tissues in vivo. The changes in the level of these metabolites provide information on the complex mechanism of the development of the disease and its progression. Metabolomic approach using analytical techniques such as magnetic resonance spectroscopy (MRS) has evolved as an important tool for identifying clinically relevant metabolic biomarkers. The metabolic characterization of breast lesions using in‐vivo MRS has shown that malignant breast tissues contain elevated levels of choline containing compounds (tCho), suggesting rapid proliferation of cancer cells and alterations in membrane metabolism. Also, tCho has been identified as one of the important biomarkers that help to enhance the diagnostic accuracy of dynamic contrast enhanced magnetic resonance imaging and also for monitoring treatment response. Further, metabolome of malignant tissues can be studied using ex vivo and in vitro MRS at high magnetic fields. This provided the advantage of detection of a large number of compounds that facilitated more comprehensive insight into the altered metabolic pathways associated with the cancer development and progression and also in identification of several metabolites as potential biomarkers. This article briefly reviews the role of MRS based metabolic profiling in the discovery of biomarkers and understanding of the altered metabolism in breast cancer.

The intensity of the MS peak has dependence on ionization efficiency and is not correlated with the concentration of metabolite.
Targeted/Untargeted Can be used for both types of analysis Better for targeted analysis monitoring and detection of breast cancer recurrence. 3Recent breast screening studies have reported sensitivity of DCEMRI to be between 75.2% and 100% and specificity in the range of 83-98.4% in women having hereditary and familial risk of breast cancer. 4,57][8][9] It is known that the malignant transformation of a cell is a resultant of several complex events like alterations in the several regulatory pathways at the molecular level that is manifested as changes in the metabolism of a living system. 10Metabolomics approach is based on the comprehensive measurement of changes in all small molecular weight metabolites in a living system. 6,7The concentrations of metabolites are sensitive to changes in the metabolic pathways therefore, metabolome of an individual represents their phenotype in healthy and diseased states.The alterations in metabolite levels has the potential to provide deeper understanding of the phenotypic changes resulting from genetic alterations, pathological, physiological, environmental, and toxicological influences. 6,7gnetic resonance spectroscopy (MRS) and mass spectrometry (MS) have been the two major techniques to study cancer metabolomics. 8,9Each of these analytical methods has their unique advantages and disadvantages.7][8][9] Further, MRS is nondestructive and has high reproducibility compared to MS analysis.[8][9] However, unlike MRS, MS based analysis requires separation of individual compounds using techniques like liquid chromatography and gas chromatography.Following chromatographic separation, compounds are ionized and then analyzed based on their charge/mass ratio through MS analyzer.2][13] The disadvantage of MRS is its poor sensitivity compared to MS, however, the use of high field magnets (at proton ( 1 H) resonance frequency greater than 1.0 GHz) and cryo probes have markedly enhanced the sensitivity of MR spectroscopic analysis. 14,15It is possible to detect metabolites in picomole concentration with special probes. 14,15A comparison of MS and MRS techniques is presented in the Table 1.

IN VIVO MR SPECTROSCOPY (MRS)
In vivo MRS provides information on the metabolites present in a defined volume of interest (VOI) in any organ such as breast.Most breast MRS studies have been performed using 1 H nuclei due to its high sensitivity and natural abundance in living tissues.The in vivo MRS, is technically referred as image guided localized spectroscopy and it has two variants: namely, (a) single-voxel spectroscopy (SVS), when the spectrum is acquired from a single voxel; and (b) MR spectroscopic imaging (MRSI) or also known as chemical shift imaging, using which spectra from multiple voxels are acquired simultaneously.The placement of voxel is guided using MR images acquired in three orthogonal planes, earlier.The SVS is less sensitive to patient motion and a spectrum of better quality is obtained which is suited for quantitative analysis than MRSI.However, MRSI provides the advantage of simultaneous assessment of the lesion and the normal parenchyma of breast as well as information on the spatial variation of metabolites within a tumor.Additionally, color coded metabolic maps can be generated for visual assessment of metabolite levels which could be developed as more useful tool in clinical settings.[58][59][60][61] Dedicated breast coils either single or bilateral are used for acquisition of breast MRS.1][22][23][24][25][26][27][28][29][30] The MRS of breast is performed at long echo times (TE above 100 ms).][25][26]37

Lipid and water composition in the normal and malignant breast tissue
Normal breast is a glandular organ containing high amount of lipid and low water content.Figure 1A shows the voxel position in the T1 weighted image, while Figure 1B shows the typical 1 H MR spectrum of the normal breast tissue of a healthy volunteer acquired without suppression of water and fat. 37,38The spectrum shows an intense peak at 1.3 ppm pertaining to [-(CH 2 ) n -] protons of the lipids and the peak at 0.9 ppm is due to the CH 3 protons of glycerides.α-methylene protons of the glyceride chain are assigned at 2.2 ppm while diallylic CH 2 protons resonate at 2.7 ppm.The peak observed at 5.2 ppm is due to CH of glycerol backbone and olefinic protons.The water resonance is observed at 4.7 ppm.The ratio of the integral values of the major lipid peaks (at 1.3 ppm and 0.9 ppm) and the water peak provides the measure of the W-F ratio.Our group investigated the changes in W-F ratio of the normal breast parenchyma of healthy female volunteers during the five phases of the menstrual cycle in three anatomical regions of the breast, namely; upper and lower quadrants and paraareolar region (see Figure 2). 62The W-F ratio was found to be higher in para-areolar region compared to upper and lower quadrants at all the phases of the menstrual cycle. 62Cyclic changes were seen in the W-F ratio in the para-areolar region during the menstrual cycle. 62The W-F value was high (0.90 ± 0.41) in the proliferative phase, which showed a reduction (0.46 ± 0.21) in follicular and luteal phases (0.45 ± 0.25), while it increased during secretory (0.76 ± 0.61) and menstrual phases (0.87 ± 0.37). 62These findings suggested the heterogeneity of lipid composition within the breast.Further it suggested that the physiological factors like menstrual cycle affect the water and lipid compositions of the normal breast tissue.This also suggested the use of W-F ratio for assessment of breast pathology requires a careful consideration of the time of menstruation and the location of the lesion within the breast. 62e T2-weighted fat suppressed MR image of a breast cancer patient with histological type, infiltrating ductal carcinoma (IDC), is shown in Figure 3A, while the water and the fat unsuppressed spectrum from the VOI shown in the image is presented in Figure 3B.
The MR spectrum from a malignant tumor showed a predominant water peak and a reduced lipid peak.Malignant breast tissues, thus found to have high W-F value compared to the control and the normal unaffected breast tissues and from the contralateral breast tissues of patients. 25,31,32This indicated that the water content is considerably increased during malignancy. 25,31,32,63Also, alterations in lipid content have been reported with the tumor progression. 64The W-F parameter has also been used to monitor the response to neo-adjuvant chemotherapy (NACT) in patients with locally advanced breast car-cinoma (LABC).Data from our laboratory showed a reduction in W-F ratio in patients receiving chemotherapy and the value was found to be associated with the reduction in the primary tumor size indicating that it can be used as a non-invasive biomarker of tumor response. 31,32wever, the value of W-F ratio showed significant overlap between benign and malignant tissues, suggesting its limited utility in the diagnosis of breast lesions. 65Recently, Wang et al suggested that composition of water and lipid and density in breast tissue is associated with the risk of breast cancer and it can be used for screening high-risk women. 35sults of a recent study from our laboratory showed that malignant breast tissues are characterized with a lower fat fraction compared to normal breast tissues and benign lesions. 33Out of 68 breast cancer patients, 22 had ER+/PR+ hormone receptor status while 33 patients belonged to ER−/PR− status.Higher fat fraction was observed in patients with estrogen receptor negative (ER-)/progesterone receptor (PR-) status as compared to ER+/PR+ patients.Forty-one patients showed Human epidermal growth factor 2 neu (HER2neu+), whereas 24 had HER2neu-status.The HER2neu+ tumors showed significantly higher fat fraction (median 0.14; range 0.02-0.69)compared to HER2neu-(median 0.08; range 0.01-0.25)breast tumors.A cut-off value of 0.21 was calculated for fat fraction that provided 76% sensitivity and a 74.5% specificity in differentiating malignant and normal breast tissues while a cut off value of 0.18 was obtained for differentiation of malignant and benign lesions with a sensitivity of 75% and a specificity of 68.6%. 33In a recent study, using 1 H MRS at 1.5 T, the levels of six lipid metabolites were determined and the differences in the composition of lipids were documented in benign and malignant lesions and among luminal A/B and other subtypes of breast cancer. 66

Total choline (tCho) in normal, malignant, and benign breast tissues
The 1 H MR spectrum of a breast cancer patient suffering from IDC with water and lipid suppression shows a peak at 3.2 ppm due to choline-containing compounds (tCho) (see Figure 3C) 38 .In our laboratory, we reported the absolute concentration of tCho in patients with early breast cancer (n = 31), LABC (n = 120), normal breast of healthy volunteers (n = 31), and benign breast lesions (n = 38), and also an association of tCho concentration with ER, PR, and HER2neu status.The concentration of tCho ranged from 0.76 to 21.2 mmol/kg in malignant breast tissues that was significantly higher compared to the benign lesions (0.04 to 2.70 mmol/kg) and normal breast tissues of healthy volunteers (0.1 to 1 mmol/kg). 30The concentration of tCho was found to be higher in patients with early breast cancer (5.4 ± 3.7 mmol/kg) compared to LABC patients (3.8 ± 2.0 mmol/kg). 30is may be attributed to the presence of necrosis in advanced stage tumors due to insufficient supply of nutrients.The tCho levels were higher in triple positive and non-triple positive patients compared to triple negative patients.The cut-off value for discriminating malignant and benign lesions was found to be 2.54 mmol/kg.Our results demonstrated that the quantitative assessment of tCho may be used in the clinical settings for diagnosis of breast lesions. 30A metaanalysis of pooled MRS data from five initial studies that included the data from our laboratory as well, showed that the use of detection of tCho signal has a sensitivity of 83% and a specificity of 85% in differentiating breast cancer from benign lesions. 262][23] The tCho peak has been found to have contribution from several Cho containing compounds such as free choline (Cho), phosphocholine (PC), and glycerophosphocholine (GPC).
8][69] A closer look at the biosynthetic pathway of two important phospholipids, phosphotidylcholine (PtdCho) and phosphotidylethanolamine (PtdEtn) provides an understanding of higher tCho seen in malignant tumors (Figure 4).These phospholipids are important constituents of adipocytes.8][69]   This enhances the levels of PC that leads to increase in the biosynthesis of PtdCho. 70Further, it has been reported that cancer cells have faster transport of Cho in a study comparing transport of Cho in MCF-7 breast cancer cell lines and normal mammary epithelial cells. 71ltzer and Dietzel in 2013 performed a systematic review and meta-analysis for estimating diagnostic performance of breast MRS in distinguishing between malignant and benign lesions.They included 19 breast MRS studies performed at 1.5 T and 3.0 T in a total of 1183 patients with 1198 lesions (malignant n = 773, benign n = 452).
They reported a pooled sensitivity and a specificity of 73% and 88%, respectively noting no significant influences of higher magnetic field strength, post-contrast acquisition and quantitative versus qualitative MRS evaluation. 54A meta-analysis reported by Cen and Xu in 2014 included data of 18 SVS MRS studies of breast cancer patients and patients with benign diseases.This analysis included 750 malignant and 419 benign lesions. 55The pooled sensitivity obtained was 71% and the specificity was 85% for breast MRS.This analysis pointed toward standardization of the protocol for MRS acquisition among centers and also recommended a multicenter trial. 55

tCho in lactating and normal breast tissues
With the technological advances in magnets, breast coils and pulse sequences, a number of studies also reported the tCho signal in normal and lactating breast tissues. 21,24,29This raised a question on the use of tCho as a marker of breast malignancy.We performed a systematic breast MRS study on patients with LABC and normal healthy lactating women volunteers. 72The concentration of tCho was determined and spectral characteristics were compared between the two groups.The tCho peak was observed in all patients with LABC and the concentration was 3.51 ± 1.72 mmol/kg.In lactating women, tCho peak was detected in 10 out of 12 volunteers, with a concentration value of 3.52 ± 1.70 mmol/kg, which is similar to that observed in malignant lesions (Figure 5). 72Higher tCho levels in lactating breast tissues is due to the lactogenesis process.In neonates the choline obtained through mother's milk and is required for normal growth. 73However, a peak corresponding to lactose sugar was also observed in these volunteers, suggesting that this spectral characteristic is unique to lactating breast tissues. 72Further, it was documented through careful referencing by Stanwell et al that tCho peak observed in lactating breast is mainly due to GPC and not due to PC as observed in malignant lesions. 29Additionally, another MR parameter namely the apparent diffusion coefficient determined by diffusion MRI, could be used for differentiating lactating breast tissues from malignant tissues. 72The value of apparent diffusion coefficient was higher in lactating breast than in malignant breast lesions.

tCho and molecular markers
There has been an increasing interest in understanding the molecular mechanism of elevation of tCho seen in breast cancer patients.The molecular mechanism underlying tCho elevation in malignant tissues of breast cancer patients was recently reported by our group.
An association of tCho with the expressions of β-catenin and cyclin D1 proteins was documented (Figure 6). 74The Wnt/ β-catenin pathway is a signal transduction pathway that regulates several cellular processes. 75It has been reported that activation of this pathway stimulate cell proliferation and associated with poor survival. 76In our study, the β-catenin expressions was higher both in cytosolic and nuclear fractions in malignant tissues compared to benign and non-involved tissues.In malignant tissues, a positive correlation was found between tCho and cytosolic and nuclear expressions of β-catenin and cyclin D1.
Patients with PR negative status showed higher cytosolic β-catenin expression than positive patients.The findings may explain the molecular mechanism of elevated tCho and indication for β-catenin pathway as a therapeutic target. 74The tCho level also showed correlation with nuclear grade and ER and PR status. 77A recent study showed a correlation between tCho levels and the expression of calcium-sensing receptors. 78

1 H MRS in evaluating therapeutic response
0][81][82] In an earlier study, we found that before therapy 10/14 cases showed tCho, while post-therapy tCho signal was detected only in 7 patients indicating a positive response to NACT. 20We also evaluated the use of tCho signal-to-noise ratio (ChoSNR) in monitoring the tumor response.It was shown that in the patients who showed positive response to NACT, the level of ChoSNR and the tumor size reduced after three cycles of NACT compared to their pretherapy value.While in non-responders, the values of the tumor size and the ChoSNR remained similar or increased compared to pre-therapy value after NACT. 80cently our group reported the role of multi-parametric approach using three parameters namely, tCho, apparent diffusion coefficient determined from diffusion MRI and the tumor volume in monitoring both clinical and pathological responses of patients with LABC (n = 42) undergoing NACT (Figure 7). 81Our study showed that both tCho and ADC reduced as early as first cycle of NACT in both the clinical and the pathological responders while the tumor volume reduced only after second cycle of chemotherapy indicating that these two parameters can serve as an indicator of early response. 81Recently Drisis et al reported the value of tCho signal as early response predictor after NACT in 39 LABC patients. 83After first cycle of NACT, tCho level was found useful in predicting early response and pathological response.At the end of NACT, changes in tumor diameter, volume along with tCho levels predicted tumor response, while change in K trans could predict only pathological response.The quantification of tCho was more sensitive for predicting pathological response in triple negative tumors. 83lan et al reported a multicenter analysis of ability of tCho as pathological response to NACT in the patients with LABC.MRS measurements were carried out before and after 20-96 h of first cycle of NACT. 84The study enrolled 119 subjects, however, usable data could be obtained only for 29 cases.The decrease in tCho levels after 20-96 h of first cycle of NACT showed poor ability to predict pathological response. 84Zhou et al recently evaluated the tumor response after the second and fourth cycles of NACT in breast cancer patients having nonconcentric shrinkage pattern. 85Significant change in tCho integral after fourth cycle was seen while there was no change in tumor size in responder and non-responder groups.The sensitivity to detect response was 93.75%, with a positive predictive value of 78.9% and the AUC as 0.747 for tCho integral. 85Cho et al measured the changes in tCho by MRS and maximum and peak standardized uptake values and total lesion glycolysis by 18 F-fluorodeoxyglucose positron emission tomography and evaluated their ability to predict response to NACT in the 35 patients with LABC. 86Of the 35 patients, six had pathologic response while 29 were non-responders.Mean percentage change in tCho, standardized uptake values and total lesion glycolysis of the patients with pathological response were larger than those of the non-responders.The diagnostic accuracy for tCho was found to be comparable with other parameters. 86

EX VIVO 1 H MRS OF INTACT BREAST TISSUE
Ex vivo HRMAS proton MR spectroscopy has also been widely used to study metabolic processes in breast cancer tissue a non-destructive manner.Most studies focused on assessing the diagnostic biomarkers, correlations of metabolite levels with prognosis and monitoring therapeutic response. 40,41The advantage with HRMAS is that tissue integrity is maintained and same tissue can be used for histopathological analysis. 40,41Chae et al recently utilized HRMAS based metabolic profiling of breast tissue in distinguishing ductal carcinoma in situ lesions with and without invasive components. 87The GPC/PC ratio as well as the concentration of succinate and myo-inositol was higher in the pure ductal carcinoma in situ group than in the ductal carcinoma in situ accompanying invasive carcinoma group.The orthogonal partial least square discriminant analysis models built with metabolic profiles clearly discriminated the two groups. 87Haukaas et al classified the metabolic profiles obtained using HRMAS MR spectroscopy in three different metabolic clusters and combined them with gene and protein expression data. 88Their results showed three metabolic clusters; cluster 1 having the highest levels of GPC and PC, cluster 2 was characterized with highest level of glucose, while cluster 3 showed the highest levels of alanine and lactate. 88Pathway analysis of metabolites and gene expression data indicated differences in glycerophospholipid glycolysis and gluconeogenesis metabolic pathways between the clusters.Genes related to extracellular matrix and collagens were down-regulated in metabolic cluster 1 and unregulated in cluster 2 and 3, implying the differences in protein subtypes within the metabolic clusters. 88From India, recently, Paul et al. evaluated the differences in lipid profile of malignant breast tissues, lymph nodes and benign breast tissues. 89Their study reported reduced lipid content along with a higher fraction of free fatty acids in malignant breast tissues. 89skeødegård et al. investigated the relationship of metabolic profiles of breast cancer tissue with 5 year survival of breast cancer patient. 90e metabolic profiling of excised tissue was performed using highresolution MR spectroscopy.In a subgroup of ER positive patients, higher levels of lactate and glycine showed association with lower survival rates.These metabolites were suggested for predicting prognosis in breast cancer. 90

IN VITRO 1 H MRS
In vitro MRS studies of tissue extracts, axillary nodes and fine needle aspirates have also been reported to find out biomarkers for diagnosis and in evaluating the biochemistry of breast tumors.[44][45][46][47][48][49][50][51][52] Further, the quantification of metabolites is comparatively easier using in vitro MRS than in vivo. 48,49

MRS of breast tissues, axillary nodes, and blood sera
The nuclear magnetic resonance (NMR) spectroscopic metabolic profiling of breast tissues obtained either by surgery or biopsy requires extraction of tissue metabolites.In general, the extraction of water soluble metabolites is carried out using perchloric acid extraction procedure. 91,92Extraction of both water soluble and lipid soluble metabolites is performed using chloroform-methanol-water extraction procedure.The details of methodologies have been described elsewhere. 91,92Sample collection and storage are important steps in metabolomics studies, requiring standardized protocols to be followed, since concentration of metabolites like lactate may change due to degradation of sample on exposure to room temperature during storage or sample processing. 93,94Inappropriate handling of samples can result in high variability and inaccuracy in metabolite levels. 93,94ibbestad et al reported first in-vitro high-resolution proton MRS study of normal (non-involved) and malignant (involved) breast tissues following perchloric acid extraction. 42The non-involved tissues showed predominant signals from glucose and other carbohydrates, while these compounds were in low levels in involved tissues.Tumor extracts were characterized with high concentrations of taurine,

F I G U R E 8
In vitro 1 H magnetic resonance (MR) spectrum from the aliphatic region of the perchloric acid extracted from involved breast cancer tissue recorded at 400 MHz nuclear magnetic resonance (NMR).Abbreviations: Ala, alanine; Ace, acetate; Arg, arginine; Asp, aspartate; Cho, choline; Cr, creatine; Glc, glucose; Glu, glutamate; Gln, glutamine; GPC, glycerophosphocholine; Gly, glycine; Iso, isoleucine; KG, ketogultarate Lac, lactate; Leu, leucine; Lys, lysine; mI, myo-inositol; PCr, phosphocreatine; PCho, phosphocholine; Pyr, pyruvate; Suc, succinate; Tau, taurine; Val, valine (Figure as  We also reported the metabolic profile of extracts of involved and non-involved breast tissues using high-resolution in vitro proton MRS (Figure 8). 43,95Significantly higher concentration of several metabolites like lysine, alanine, glutamine, glutamic acid, phosphocreatine+creatine, Cho, GPC+PC, acetate lactate, and myo-inositol was seen in involved tissues compared to the noninvolved tissues.These metabolites showed five-to tenfold increase in malignancy. 43No significant differences in the levels of phenylalanine, tyrosine, formate, and glucose between cancerous and non-involved tissues was observed.Beckonert et al using in-vitro 1 H MRS of tissue extracts in combination with pattern recognition approach reported that the metabolic profile of tumors is related to tumor grade. 44ey reported significant differences in uridine di-phosphate-hexose, phosphoethanolamine and PC between tumors with different grades.
Malignant tumors had higher levels of taurine and lipid metabolites, while glucose and myoinositol were in higher level in controls.
These findings indicated changes in metabolism associated with breast tumor formation.It is known that in normal cells the main energy substrate is glucose that is utilized through glycolytic pathway followed by tricarboxylic acid cycle and oxidative phosphorylation for energy generation in the presence of oxygen. 96However, in cancer cells, the rate of glycolysis is increased and pyruvate, a product of glycolysis is converted to lactate, thus enhancing the level of this metabolite even in the presence of sufficient oxygen levels. 97,98It has been reported that several intermediate compounds of glycolytic pathway are used as substrates for synthesis of other molecules through various biosynthetic pathways like pentose phosphate pathways.For example, ribose-phosphate is an intermediate of glycolytic pathway is utilized for the synthesis of nucleic acids through pentose phosphate pathway.
Since, tumor cells have rapid rate of cell division, there is requirement of larger pool of substrates for biosynthesis of other molecules like nucleic acid.][100] The higher concentration of several amino acids like alanine, glutamine, lysine, and glutamate in tumor cells. 95Amino acids have been found to have various roles in cell metabolism.They can be utilized as energy substrate, as anti-inflammatory agents and act as regulators as required for maintenance and cellular growth.2][103] An important antioxidant glutathione is also synthesized using glutamate as substrate.Amino group of glutamate is also used in the synthesis non-essential amino acids such as alanine, aspartate, glycine, and serine. 101,102illary node metastasis is an important prognostic factor in breast cancer management.Generally, axillary node dissection is the standard part of surgery for breast cancer patients.However, this can be avoided if there are no micro-metastases in nodes.Therefore, detection of micro-metastases may have an important role in the prognosis of breast cancer.We investigated the metabolite composition of axillary lymph nodes using various one-and two-dimensional in vitro NMR methods in patients with breast cancer. 48The comparison of metabolic profile of involved and non-involved nodes showed significantly higher concentration of GPC+PC in involved nodes.This indicated the increased membrane synthesis in rapidly dividing cancer cells.The level of lactate was also higher in the involved nodes, indicating a higher rate of glycolysis in tumor cells. 104Our study further showed that nodes with metastases have increased ratio of metabolites [(GPC+PC)/threonine.The use of this as a marker of malignancy detected axillary node metastases with a sensitivity of 88% and a specificity of 91%. 49A higher lactate level has also been suggested as an indicator of the presence of malignant cells in a study that used MRS of lymph nodes excised from nude mice. 103Mountford et al reported the MRS study of nodes from tumor bearing rats for detecting micrometastases and reported that MRS detected micro-metastases that were missed by conventional histopathology. 103study by Singh et al reported the metabolomics of serum of breast cancer patients. 50They reported that patients with high expression of inositol 1,4,5 trisphosphate receptor were having higher levels of alanine, lactate, lysine, and lipoprotein content compared to healthy subjects.The levels of glucose and pyruvate were lower in sera of breast cancer patients compared to healthy subjects. 50Few studies have also reported differences in the metabolic profile of serum of breast cancer patients in early stage and metastatic breast cancer. 51,105Significant differences were seen in metabolites such as glutamate, lysine, betahydroxybutyrate, glucose, lactate, and N-acetyl glycoprotein in early and late stage cancer indicating the role of these metabolites in cancer progression. 51chad et al investigated the metabolome of blood plasma of 50 early breast cancers and 15 metastatic breast cancer patients by 1 H-NMR spectroscopy. 106Using multivariate analysis, levels of several plasma metabolites including glucose, pyruvate, lactate, alanine, acetate, β-hydroxy-butyrate,acetoacetate, isoleucine, leucine, glutamate, glutamine, lysine, valine, threonine, glycine, phenylalanine, tyrosine, urea, creatine, and creatinine were found to be modulated between patients with early and metastatic breast cancer. 106Additionally, lactate levels were inversely correlated with the tumor size.It was suggested that metabolism is altered even at early stages of breast cancer. 106Flote et al characterized serum lipid profiles of breast cancer patients using NMR spectroscopy. 107They documented inverse associations between HDL phospholipids and Ki67, specifically in between HDL1's contents of phospholipids, cholesterol, apolipoproteins-A1, A2, and Ki67. 107

MRS of fine needle aspirates
6][47] Usually, two aspirations from a lesion are taken after palpation of the tumor to obtain appropriate quantity of the sample for in-vitro high-resolution NMR studies of FNAC samples.In a study from our laboratory the metabolic profile of FNAC sample of patients with IDC showed increased levels of many metabolites in addition to choline compared to benign aspirates (Figure 9). 43This is in agreement with the observation of elevated tCho levels seen using in-vivo MRS in breast cancer patients.Significant differences in the metabolic profile were also seen in other breast cytopathologies. 43Use of MRS of FNAC samples has been reported in discriminating invasive cancer, normal, and benign tissues on the basis of tCho levels. 46Mackinnon et al performed 1 H MRS of FNAC samples from benign and malignant lesions and documented that choline to creatine ratio (Cho/Cr) has a 95% sensitivity and a 96% specificity for discriminating malignant from benign lesions. 45Similarly, a three-stage statistical classification strategy analysis has been reported for diagnosis and prognosis of breast cancer. 47It distinguished malignancy with a 93% accuracy from the benign disease. 47

MRS of cell lines
The metabolomics studies of breast cancer cell lines have also been found as a valuable platform for discovery of biomarkers and understanding the molecular mechanism of cancer progression and response to therapy. 8,9Mori et al have studied the choline and lipid metabolism in breast cancer cell lines. 108An increase of PC and tCho as well as alterations in lipids have been consistently shown in cancer cells and tissue.They have compared the metabolic profile and protein expression of enzymes regulating Cho and lipid metabolism in breast cancer and prostate cancer cell lines. 108Significant differences in lipid and Cho metabolism and expression of proteins were reported in both the breast and prostate cancer cells.In another study, authors used targeted silencing of two glycerophosphodiesterase genes, GDPD5 and GDPD6 by small interfering RNA (siRNA) for studying choline and phospholipid metabolism in MDA-MB-23 and MCF-7 cell lines of breast cancer. 109It was reported that silencing of GDPD6 increased the GPC levels more than GDPD5 silencing, suggesting this as a potential treatment strategy for breast cancer. 109Gowda et al using NMR metabolomics approach studied the effect of inhibition of glutaminase by using the inhibitor BPTES (bis-2-(5-phenylacetamido-1,3,4-thiadiazol-2-yl)ethyl sulfide) in two breast cancer cell lines MCF7 and MDA-MB231, cancer proliferation is found to be associated with glutamine addiction. 108The inhibition of glutaminase was found to affect several metabolic pathways including glycolysis, Kreb's cycle, amino acid and nucleotide metabolism, the metabolic changes were more pronounced in MCF7 cells with alterations in 14 metabolites compared to seven for MDA-MB231. 110Armiñán et al investigated the changes in metabolomic profiles in response to doxorubicin and N-(2-hydroxypropyl) methacrylamide (HPMA) copolymer-conjugated form in an in-vitro cell culture model and also in an in vivo orthotopic breast cancer model along with protein expression. 111They found that polymer conjugation led to increased apoptosis, reduced phospholipids and glycolysis compared to the free doxorubicin. 111Bhute et al investigated the effect of veliparb and radiation on three breast cancer cell lines and reported that therapy induced metabolic changes are cell line dependent. 112Nitrogen metabolism, aminoacyl-tRNA biosynthesis, taurine and hypotaurine metabolism, glycine, serine, and threonine metabolism were found to be enriched by pathway enrichment and topology analysis after therapy in all three breast cancer cell lines.These metabolic changes were cell line-dependent, indicating the importance of different metabolic responses in different cancer sub-types. 112hese studies have shown that various metabolic pathways, including glycolysis, amino acid metabolism, and membrane metabolism have been altered in the breast cancer.The potential of metabolomics approach using in vitro NMR studies of various biological matrices has provided wide coverage of metabolites.Further, these studies have provided a more comprehensive understanding of the metabolic profile of the tumor tissues.However, there are only limited numbers of studies in the literature.There is a need to perform studies in a large cohort of patients, including various histological and molecular subtypes of malignant and benign breast lesions.This will provide a better understanding of the mechanisms of malignant transformation, identifying new therapeutic targets, better biomarkers for early diagnosis, and response assessment.Additionally, various in vivo MRS studies have documented altered lipid composition, water-fat ratio, fat fraction, and tCho levels determined using in vivo MRS as promising biomarkers for diagnosis and for monitoring early response to therapy.

SUMMARY, OUTLOOK, AND FUTURE DIRECTIONS
Early diagnosis of the disease and the prediction of early response of tumor is important in better management of the disease, enabling surgical options, and open the possibility of alternative therapy for non-responding patients.Studies have investigated the molecular mechanism of tCho elevation and its association with the molecular heterogeneity of breast lesions.There is need to perform large scale multi-center studies, which would also help in comparing the data of different centers and discovery of robust biomarkers.Moreover, analysis of various biological matrices from breast cancer patients using high-throughput analytical platforms MS and MRS have identified a large number of metabolites as potential candidates to serve as biomarkers, however, these state-of-the-art methodologies are still confined to research laboratories.There is a significant scope to offer simplistic and reproducible markers that may be used in complementing existing diagnostic techniques in clinics for providing personalized health care.Nevertheless, there is a need to perform systematically designed metabolomics studies of biological matrices like blood sera/plasma and urine to discover robust non-invasive biomarkers for concerning issues of early diagnosis, stratifying high-risk patients, monitoring therapy, understanding drug-resistance, recognizing new therapeutic targets and developing better therapeutic approaches.
in a prone position in the gantry of the magnet while the breast is fitted in the cup of the breast coil with additional cushions inside to reduce motion artifacts.The MR spectrum of breast acquired without the suppression of water and the lipid resonances provides information on the various lipid resonances and the water content in tumors.While the signal from choline containing metabolites (tCho) is observed when signals from both the water and the lipid resonances are suppressed.

F
I G U R E 1 (A) T1-weighted MR image of the normal breast from a volunteer showing the voxel position from which a single-voxel 1 H MR in vivo spectrum (B) was obtained without water and fat suppression (Reprinted with permission from John Wiley & Sons, Inc. from references # 37 and 38) F I G U R E 2 In vivo proton MR spectra acquired at TE = 135 ms from three different voxel (8 ml) locations within the normal breast of a 31 year old normal female volunteer.(A) Upper quadrant, (B) Para-areolar region, (C) Lower quadrant (Reprinted with permission from Elsevier from reference # 62) MRI of a patient with locally advanced breast cancer showing the voxel position from which the single-voxel 1 H MR in-vivo spectrum was obtained without water and fat (lipid) suppression (B) and with the suppression of water+fat (lipid) resonances (C) (Reprinted with permission from John Wiley & Sons, Inc. from reference # 38)

F I G U R E 4
The role of metabolic reprogramming in breast cancer cells, their role in cancer and the induced co-adaptive mechanism (Reprinted with permission from John Wiley & Sons, Inc. from reference # 27) activation of CK that leads to enhanced phosphorylation of Cho to PC.

F I G U R E 5 F I G U R E 6
T2-weighted fat suppressed axial image (A) from the normal breast tissue of a lactating women volunteer showing a voxel location of size 20×20×20 mm 3 (B) corresponding 1 H MR spectrum obtained without water and lipid suppression showing the water and lipid peaks.(C) 1 H MR spectrum obtained with water and lipid suppression showing the residual water and lipid along with the tCho and the lactose peaks (Reprinted with permission from John Wiley & Sons, Inc. from reference # 72) Schematic representation of the link between the choline synthesis and the Wnt-mediated β-catenin pathway.(A) Normal breast tissue: In a normal breast tissue Wnt (green polygon) signaling is absent and β-catenin (brown rectangles) level is maintained low in the cell cytosol due to phosphorylation (red stars) and degradation of β-catenin.Also, cyclin D1 (red circles) translocates to the nucleus at a reduced rate to participate in G1 to S phase transition.The cell division rate is regulated.Phosphatidyl choline (PtdCho; blue rectangle) which is a membrane phospholipid, converts into phosphocholine (PCho; brown triangle) in the cytosol by the activity of phospholipase D (PLD; grey circle).(B) Malignant breast tissue: During malignancy, Wnt signaling is active and hence β-catenin increases in the cytosol that can translocate to the nucleus and bind to cyclin D1 (brown rectangle + red circle), to increase the rate of cellular transcription.With the increase in cellular proliferation, membrane requirement for PtdCho increases that in turn leads to increased PLD activity (double grey circle).PCho increases in the cell cytosol, thereby increasing the tCho levels.Increased cytosolic β-catenin levels also increase the activity of PLD (Reprinted with permission from John Wiley & Sons, Inc. from reference # 74)

F I G U R E 9
High-resolution 400 MHz proton MR spectrum of fine-needle aspirate (FNAC-ex vivo) from (A) malignant and (B) benign breast tissues.Abbreviations: Leu: leucine; Val: valine; Thr: threonine; Lac: lactate; Ala: alanine; Lys: lysine; Ac: acetate; Glu: glutamic acid; Gln: glutamine; PCr: phosphocreatine; Cr: creatine; PEtn: phosphoethanol amine; Etn: ethanol amine; PCho: phosphocholine; Cho: choline; Tau: taurine; Gly: glycine; Asp: aspartate; CH 3 , CH 2 CO, CO-CH 2 -CH 2 -: groups of lipids (Reprinted with permission from Bentham Science Publishers from reference # 43) MR spectroscopy and MS play animportant and complementary role by measuring large number of metabolites in various biological matrices, providing valuable information for diagnostic applications, and for translational research.Compared to MS, the high reproducibility, minimal sample preparation, non-selective and non-destructive nature, and the ability to identify the unknown metabolites have made MRS an important tool for various metabolomics applications.This review briefly discussed the use of MRS based metabolomics approach in identifying biomarkers for diagnosis and therapeutic outcome in breast cancer, especially focusing on the various works carried out in India.The usefulness of ex vivo HRMAS MRS of intact tissue, in vitro MRS of tissue extracts, axillary nodes, and FNAC samples in the biochemical characterization of breast tissues was presented.