Proton magnetic resonance spectroscopy in oncology: the fingerprints of cancer?

Abstract

Abnormal metabolism is a key tumor hallmark. Proton magnetic resonance spectroscopy (1H-MRS) allows measurement of metabolite concentration that can be utilized to characterize tumor metabolic changes. 1H-MRS measurements of specific metabolites have been implemented in the clinic. This article performs a systematic review of image acquisition and interpretation of 1H-MRS for cancer evaluation, evaluates its strengths and limitations, and correlates metabolite peaks at 1H-MRS with diagnostic and prognostic parameters of cancer in different tumor types.

Diagn Interv Radiol 2016; 22: 7589 Turkish Society of Radiology 2016

ONCOLOGIC IMAGING

REVIEW

Proton magnetic resonance spectroscopy in oncology: the fingerprints of cancer?

Roberto Garca-Figueiras Sandra Baleato-Gonzlez Anwar R PadhaniLaura OleagaJoan C VilanovaAntonio LunaJuan Carlos Cobas Gmez

ABSTRACT

Abnormal metabolism is a key tumor hallmark. Proton magnetic resonance spectroscopy (1H-MRS) allows measurement of metabolite concentration that can be utilized to characterize tumor metabolic changes. 1H-MRS measurements of specic metabolites have been implemented in the clinic. This article performs a systematic review of image acquisition and interpretation of 1H-MRS for cancer evaluation, evaluates its strengths and limitations, and correlates metabolite peaks at 1H-MRS with diagnostic and prognostic parameters of cancer in dierent tumor types.

Magnetic resonance spectroscopy (MRS) is an imaging technique based on the detection of radiofrequency signals generated by spins of magnetic resonance active nuclei (such as 1H, 31P, 13C, and 19F) precessing in an external magnetic eld

(B0). In clinical practice, MRS produces spectra from the patient with an anatomical/spatial reference. MRS is mainly based on 1H, because hydrogen is one of the main elements in human body. In vivo MRS allows the analysis and quantication of metabolites present in a tissue in a noninvasive way (1). MRS is based on the fact that protons in dierent molecules resonate at slightly dierent frequencies. This feature is secondary to the dierences in the local electron cloud, which may shield the nucleus from the main magnetic eld. Dierent metabolites containing the same nucleus exhibit characteristic chemical shifts in resonance frequency. In the oncologic eld, abnormal metabolites may represent emerging tumor biomarkers. MRS allows the characterization of the metabolic changes associated with cancer (2). Up to date, the main diagnostic value of 1H-MRS in tumors has been the detection of elevated levels of choline-containing compounds or total choline at 3.2 ppm, which includes contributions from choline, phosphocholine, and glycerophosphocholine. The most consistent dierence between the majority of normal tissues and tumors is usually found in choline levels. As a general rule, normal tissues display low choline levels, whereas tumors show high choline levels (Fig. 1), although several exceptions must be considered in clinical practice (Fig. 2) (3). Beside this, other metabolic pathways and their metabolites can be assessed using MRS; the signicance/importance of a concrete metabolite is going to depend on dierent features such as the clinical scenario and the organ to be studied (Table 1). Although 1H-MRS has been fundamentally applied clinically to assist in diagnosing and monitoring brain, prostate, and breast cancer (4), it has also been used to investigate other processes in the oncologic eld, including other types of primary tumors and lymph node tumor involvement (26). This article reviews the image acquisition and interpretation of 1H-MRS for cancer evaluation, evaluates its strengths and limitations, and correlates metabolite peaks at 1H-MRS with diagnostic and prognostic parameters of cancer in dierent anatomic areas.

From the Department of Radiology (R.G-F. [email protected], S.B-G.) Hospital Clnico Universitario de Santiago de Compostela, Santiago de Compostela, Spain; Paul Strickland Scanner Centre (A.R.P.), Mount Vernon Cancer Centre, Northwood, Middlesex, United Kingdom; the Department of Radiology (L.O.), Hospital Clnic Barcelona, Barcelona, Spain; the Department of Radiology (J.C.V.), Clnica Girona and Hospital Santa Caterina, Girona, Spain; Clinica Las Nieves (A.L.), Sercosa, Jan, Spain; Department of Radiology (A.L.), Case Western Reserve University, Cleveland, Ohio, USA; the Mestrelab Research (J.C.C.G.), Santiago de Compostela, Spain.

Received 14 January 2015; revision requested 16 February 2015; nal revision received 18 May 2015; accepted 16 June 2015.

Published online 2 November 2015. DOI 10.5152/dir.2015.15009

Spectroscopic imaging: technical requisites

Technically, MRS acquisition is basically very similar to that of magnetic resonance imaging (MRI). However, an optimized preacquisition preparation, adequate spectral acquisition techniques, and advanced methods of analysis are needed for obtaining a clinically useful spectrum with an optimal signal-to-noise ratio (SNR) that allows the separation of the most signicant metabolites. Dierent technical questions need to be considered for generating adequate MRS data.

75

Preacquisition preparation

MRS preparation demands that the magnetic eld (B0) is made as homogeneous as

possible. This process is named shimming (710). Dierent circumstances, including motion artifacts and large air-tissue interfaces, lead to magnetic susceptibility artifacts. Therefore, thorax and abdomen are locations difficult to exam by MRS. Another important part of MRS exams is the suppression of water and fat signals. The water and fat signals are much stronger than the metabolite signals that we are interested in. This feature creates problems with the dynamic range of the magnetic resonance receiver systems, which distorts the baseline of the spectrum, making the other metabolite peaks invisible. Initially, saturation bands must be placed closely around the volume/organ of interest in order to suppress the strong water and fat signals from the surrounding tissue. Frequency selective radiofrequency pulses saturate the water signal (90 pulses or chemical shift selectiveCHESSpulses). On the other hand, lipid signal suppression can be obtained in several ways. Long echo time (TE) sequences attenuate lipid signals. In addition, regions that produce large lipid signals or have unacceptable eld homogeneity can be presaturated. Finally, the use of lipid nulling sequences (such as short tau inversion recoverySTIR) can be considered. However, it must be also considered that frequency selective fat saturation pulses may interfere with the observation of metabolite peaks such as lactate or alanine.

Acquisition parameters

The most adequate MRS technique must be chosen depending on what metabolites and which organ are going to be studied.

Figure 1. A 42-year-old woman with anterior mass (invasive ductal carcinoma) and posterior focal mastopathy area in left breast. Axial 3D contrast-enhanced fat-suppressed image at 2 minutes (top row) shows an anterior mass (arrow, invasive ductal carcinoma already biopsied and classied as a BIRADS6 lesion), and a new posterior enhancing mass (arrowhead) with spiculated margins and curve type3 (BIRADS 5) corresponding to a focal mastopathy area. Single-voxel MRS showed positive choline peak (long white arrow) in anterior invasive ductal carcinoma (bottom left), while it was negative in the posterior focal mastopathy area.

Figure 2. MRS as a tumor biomarker. Although, as a general rule, tumors usually show high choline levels, low-grade malignant tumors may show a negative choline resonance peak at 3.22 ppm. Sagittal 3D contrast-enhanced fat-suppressed image of a 57-year-old woman with pure mucinous carcinoma at 2 minutes (left) shows an enhancing mass with irregular contour and kinetic curve type 3 (not shown). Single-voxel spectrum (right) shows no choline resonance peak in the mass.

Main points

Magnetic resonance spectroscopy (MRS) allows the analysis and quantication of metabolites present in a tissue in a noninvasive way by locating their specic peaks.

MRS allows the characterization of the metabolic changes associated with cancer, which are dependent on tumor type.

MRS interpretation is mainly based on checking the elevation of certain characteristic metabolites or the absence or decrease of normal metabolites.

MRS shows important limitations in the clinical eld: it is technically complex, time-consuming, and requires complex data processing.

MRS signal is inherently low and consequently imaging requires many averages and extremely limited eld of view. SNR and chemical shift separation of metabolite peaks increase approximately linearly with the increase of the magnetic eld. Howev-

er, considering the relatively weak magnetic elds used in clinical practice, a limited chemical shift dispersion and J-coupling can cause spectral overlap and complicate the separation of metabolites (710). Two main sequences are used in clinical practice

76 JanuaryFebruary 2016 Diagnostic and Interventional Radiology Garca-Figueiras et al.

Table 1. The principal metabolites studied in 1H-MRS and their biological signicances

Fingerprints Description ppm Decreased Increased TE

NAA A neuronal marker of density 2.02 Decreased in absence of Canavans leukodystrophy Short/long and viability neurons and axons in mosttumors or white matter

An axonal marker disease such as multiple sclerosis

Choline A metabolic marker of cell 3.22 It is increased due to cell Short/long density and membrane proliferation and breakdown ofintegrity cell membranes

Higher choline levels are shown in higher grade tumors compared with lower grade tumors

Creatine A marker of energy 3.02 Decreased phosphocreatine Short/long metabolism is an inconstant nding in tumors

Lactate Under normal circumstances, Doublet (twin Increased lactate is the eect Short/long lactate is present only in peak) at of the high rate of glycolysis minimum amounts in the 1.33 ppm Usingbrain and is not resolved It accumulates in cystic or intermediate using the normal necrotic areas TEs (135/144 spectroscopic techniques ms), the

Variable projection of the doublet peak is Glycolysis peak at dierent TEs inverted below the baseline

Using very short or very long TE (30 or 288 ms), the doublet peak projects above the baseline

Myoinositol Most important osmolyte 3.56 It is a marker for low-grade Short gliomas; it is only seen at

Glial marker: located short acquisition times exclusively in astrocytes

Glutamine and Glutamate is an excitatory 2.052.5 Glutamate is viewed as an Short glutamate neurotransmitter Complex peak important neurotoxin when its concentration exceeds that needed for neurotransmission

It is also a participant in the redox cycle

Lipids May indicate tumor necrosis 0.9 and 1.3 ppm Tumor necrosis Shortor voxel contamination by usually diploic space fat, scalp, and large broad Membrane subcutaneous tissue peaks lipids have very short relaxation times and are not usually visualized on intermediate or long TE

ppm, parts per million; TE, echo time; NAA, N-acetylaspartate.

for MRS acquisition: PRESS (point-resolved spectroscopy) and STEAM (stimulated echo acquisition mode). The PRESS sequence can be used in either single voxel spectroscopy (SVS) (where a single region is evaluated) or multi-voxel spectroscopy (MVS) (simultaneous spectrum acquisitions in multiple regions). PRESS sequence is most commonly

used in clinical practice because it presents a better SNR and less sensitivity to motion. STEAM is used only in SVS, since it results in better water suppression and shorter TE than PRESS (10). It must be considered that the observed metabolite peaks change depending on TE. Some metabolites like gluta-mine-glutamate (Glx), and myoinositol have

a short relaxation time and as a result they are not visible on a long TE sequence. A short TE acquisition is considered when TE is <40 ms; while a long echo time is considered when TE is >135288 ms. Some authors also consider including the description of intermediate TE for 135 ms. TE selection may be a key point when planning a spectroscopic exam.

Magnetic resonance spectroscopy in oncology 77

a b

Data acquisition

Exponential

AMPLITUDE

4 2 0 -2 -4

2 Hz

6 Hz

4 Hz

Real magnetic resonance spectra

0

0.5 1

SECONDS

Final clinical spectra

Gaussian

AMPLITUDE

8 6 4 2 0 -2 -4

Eddy current compensation

Oset correction

Metabolite peaks

FT

Phase correction

Baseline correction

Zero lling

Apodization

Exponential

0

1 2 3SECONDS

Obtained FID

Frequency (Hz)

Figure 3. a, b. Analysis of the MRS data. Schematic representation of MRS data generation and processing (a) and change in the spectrum shape (b) when dierent values of exponential or Gaussian functions are used in single-voxel MRS of a brain exam. Magnetic resonance spectra usually comprise more than a single frequency (a, top-left). What we actually observe in a magnetic resonance free induction decay (FID) (bottom-left) is their sum, which results in a complex signal that cannot be analyzed in a simple way. The traditional way to determine the dierent resonances present in this complex signal is by means of a mathematical procedure known as Fourier transform. In addition, acquired MRS signals require a preprocessing process to minimize error in the quantitationof metabolites (e.g., eddy current compensation, zero lling). Finally, the result of this processing should be a series of metabolite peaks (right) characterized by their principal descriptors: frequency (chemical shift) and amplitude. Change in the spectrum shape when dierent values of exponential or Gaussian functions are used (b). Manipulating the same FID with dierent window functions (b) (e.g., exponential/Gaussian) will help to increase the resolution, but at the cost of worsening the SNR or introducing artifacts in the spectrum.

For example, if lactate evaluation is required, long TE should be considered because lac-tate is detected with less lipid contamination at long TE. Moreover, TE changes metabolite presentation in spectrum. Lactate should be inverted at TE 140 ms and in phase at TE 280 ms. MVS is generally obtained with long TE due to the difficulties of quantifying overlapping multiplet resonances at short TE. In SVS, the selection of the volume of interest (VOI) is performed by combining three orthogonal slice-selective excitations. The approach of volume preselection eliminates spurious signals (710). SVS is also faster than MVS and can be acquired using both long and short TEs. Finally, an adequate shimming and better water suppression have been shown to be more feasible with this approach. For its part, MVS is useful for the depiction of tumor heterogeneity and its margins. A variety of synonymous terms for MVS have been used in the literature and by scanner vendors, including MRS imaging, chemical shift imaging, and spectroscopic imaging. However, MVS shows some important limitations. It is time consuming and limited in achieving good shimming due to the large volume of tissue examined. Besides, spatial localization is generally not as precise as in SVS, a circumstance that increases partial volume artifacts from adjacent tissues/structures.

Preprocessing of MRS

Preprocessing of MRS signal and data evaluation are the nal steps required for the clinical use of MRS. MRS data become useful in practice when a postprocessing protocol decodes these signals, which can be transformed in a list of spectral components from which metabolite relative concentrations can be determined (Fig. 3). Therefore, acquired MRS signals require a preprocessing protocol for improving the quantitation of metabolites (10). Preprocessing protocols can be divided into two main classes depending on whether it is performed in the time domain or in the frequency domain. The traditional way to determine the dierent resonances present in this complex signal (free induction decayFID) is by means of a mathematical procedure known as Fourier transform, which converts the FID into a frequency domain function (the spectrum) (8, 9). Although a detailed discussion is out of the scope of this manuscript, processing of

1H-MRS data based on the Fourier transform comprises several fundamental operations (712). First, zero lling of the FID is used to increase resolution by inserting additional data points of zero amplitude. Second, a window function is applied to either increase resolution or to improve SNR. When it is used to remove truncation artifacts, this operation is known as apodization. Third, phase correc-

tion is applied in order to have all the resonances of the spectrum in the same phase (i.e., all peaks are pointing upwards). Finally, other typically applied processing operations are baseline correction (spectra are distorted secondary to the presence of intense residual water and/or lipid peaks), eddy current, and eld inhomogeneity corrections, and postacquisition removal of water.

MRS data evaluation

The obtained spectrum represents specific metabolites appearing in certain frequencies due to their specic chemical shifts. The resonance spectrum identies metabolites by locating their peaks. Several peaks can characterize the same compound (i.e., doublet or triplet). Graphic representation of acquired data includes these metabolite peaks (represented on the horizontal axis of the graph) expressed as parts per million (ppm) and their relative signal amplitude in the vertical axis. The area or integral under each peak represents the relative concentration of the detected metabolite (712). Magnetic resonance spectra are evaluated in clinical practice in three ways: qualitative evaluation, semiquantitative evaluation, and absolute quantication. Qualitative evaluation is performed by observing absence, presence, or change of a specic metabolite. Semiquantitative evaluation is

78 JanuaryFebruary 2016 Diagnostic and Interventional Radiology Garca-Figueiras et al.

malignancy) (Fig. 4). Absolute quantication of the concentration of a metabolite can be obtained using a reference standard for calibration.

Clinical value of MRS in

oncology

Although MRS was initially developed for the assessment of brain tumors, metabolic information obtained by MRS can be helpful in diagnosis and monitoring of dierent tumors. Actually, this is an established imaging technique in brain, prostate, and breast cancers (14). MRS interpretation is mainly based on checking the elevation of certain metabolites (such as choline) or the absence or decrease of normal metabolites (e.g., N-acetylaspartateNAAin the brain). A multiparametric imaging assessment of tumors, which may include MRS, represents an attractive approach for mapping the heterogeneity of tumor phenotype. This complex evaluation supposes an important challenge in order to integrate the great volume of information that imaging can oer. Apart from these tumors, there is a growing use of MRS in the evaluation of dierent tumor types, but its value depends on the clinical scenario (e.g., organ, tumor type).

Brain tumors

Diagnosis of intracranial masses based on imaging ndings alone is a challenge for imaging. 1H-MRS oers additional information related to tumor proliferation and metabolism or neuronal damage. In the oncologic eld, there are several well-established indications for 1H-MRS in the brain such as identifying types and grades of central nervous system neoplasms (13), dierentiation between tumors from other lesions, establishment of prognosis, treatment planning with delineation of tumor invasion and denition of the target volume for radiation therapy, monitoring of tumor response, and detection of relapsing tumor (Fig. 5) (24, 14, 15). There is a wide list of metabolites that may be useful in the MRS evaluation of brain tumors, including NAA, choline, lipids, creatine, lactate, ala-nine, myoinositol, and Glx. Nearly all brain tumors have decreased NAA peaks. This critical nding is generally associated to increased levels of choline. NAA decreasing is secondary to the loss of normal neuronal tissue. However, one must proceed with caution when attempting to use the NAA

Magnetic resonance spectroscopy in oncology 79

Figure 4. Multiparametric imaging in prostate cancer imaging. A 65-year-old man with rising PSA values and two previous negative biopsies. T2-weighted and apparent diusion coefficient (ADC) parametric map (top row) and MR spectrum and dynamic contrast-enhanced (DCE) time-signal curve (bottom row) show an anterior prostate mass (arrow). T2 image shows an inltrating mass with extension into the anterior capsule. The mass presents a reduced ADC value (mean ADC=0.66610-3 mm2/s). On multiple-voxel MRS, choline is signicantly elevated compared to citrate and DCE demonstrates a type 2 curve. All these ndings suggest a high probability of malignancy. Biopsy conrmed a Gleason 8 prostate cancer.

Figure 5. Multiparametric MRI of therapy response evaluation. Grade III glioma of the left frontal lobe. Rows: serial images obtained before and after administration of bevacizumab plus temozolamide including postcontrast T1-weighted and T2-weighted images and single-voxel 1H-MRS spectra. Reduced enhancement and decreased tumor size is seen after treatment. However, an increased choline/N-acetylaspartate (NAA) ratio suggests no tumor response. These apparent contradictory ndings maybe secondary to the restoration of the blood-brain barrier as a result of antiangiogenic therapy. This feature explains a lower enhancement on T1-weighted contrast-enhanced image following therapy and a reduction in edema, which may be responsible for the changes in T2 image. Although these imaging ndings may suggest tumor response, MRS ndings do not support it, reinforcing the role of a multiparametric evaluation of the tumor phenotype.

performed by the calculation of amplitude or integral of the metabolite peaks. When using frequency domain methods, the area under the peaks of interest can be obtained using either the traditional running integral

or by deconvolution (Fig. 3b). Metabolite ratios, which are much more reproducible, are commonly used for evaluation (e.g., in prostate, choline+creatine/citrate ratio correlates signicantly with the probability of

Table 2. Main metabolites used in brain MRS

Tumor Metabolites

Astrocytoma Elevated choline, reduced creatine, and signicantly reduced NAA

Elevated choline/creatine ratio in the peritumoral region may suggest high-grade glioma rather than a solitary metastasis

Low-grade glioma typically produces myoinositol

Metastasis Similar features to astrocytomas. High lactate and lipids

Glioblastoma multiforme Same spectral pattern as metastasis. High lipid peak at 1.3 ppm

Meningioma Low creatine and myoinositol, increased choline, and low levels of lipids at 1.3 ppm

There is a characteristic presence of alanine

Radiation necrosis Low choline and NAA

In some cases produces a peak at 2.4 ppm

Choline/creatine and/or choline/NAA ratios are signicantly higher in recurrent tumor (or predominantly tumor) than in radiation injury NAA, N-acetylaspartate; ppm, parts per million.

level to dene the spatial extent of the tumor or to distinguish tumor from other neurologic abnormalities, because NAA is reduced in other neuropathologic entities like multiple sclerosis and Alzheimer disease (15, 16). The increased choline peak in brain tumors indicates an elevated rate of membrane turnover. Elevated choline along with decreased NAA is a diagnostic feature of brain tumors. The choline/NAA ratio increases as the grade increases. In addition, a peak of lipids associated with necrosis or hypoxia is found in high-grade tumors. Although an increased choline peak has been found to correlate well with cellular density, cell proliferation indices (such as Ki-67), and the degree of tumor inltration (14, 15), this is not a specic marker for neoplastic lesions of the central nervous system. Spectra from active plaques in multiple sclerosis show an elevated choline/creatine ratio and normal or reduced NAA/creatine ratio. Chronic multiple sclerosis plaques in white matter show a reduced NAA/creatine ratio and, sometimes, an elevated choline/ creatine ratio, but the ratio is not as high as in tumors. The creatine peak may vary with the tumor type and the grade of glioma. It is thought that the observed decrease in the creatine peak is related to an increased metabolic rate of the tumor, but the specic biochemical mechanisms leading to these changes are not well understood (15). There is often a mild increase of creatine signals in low-grade astrocytomas, followed by progressive depletion with increasing anaplasia. In gliomatosis cerebri, creatine peak may be also elevated. On the contrary, creatine is virtually absent in lymphomas and metastases and low in meningiomas and

oligodendrogliomas. Lipids are also important metabolites. The presence of lipid peaks at 1.3 and 0.9 ppm is a usual nding of glioblastoma, metastases, lymphoma, and abscesses. The biological basis of the increased lipids is presumed to be secondary to necrosis and membrane breakdown. Recent studies have shown that the lipid resonance observed in MRS may be produced during changes in cellular proliferation that occur prior to the onset of necrosis or products of apoptotic processes (17). On the other hand, large amounts of lipids can also be found in areas treated with radiotherapy (18). Another important metabolite peak, lactate, is secondary to anaerobic glycolysis, tumor ischemia, or necrosis. Lactate peak is found mainly in high-grade gliomas, but recent studies evidenced that their presence is not a reliable indicator of tumor grade, as they are found in all pediatric brain tumors regardless of histologic grade (1416, 18). Lactate may also be detected in the necrotic areas of glioblastoma and metastases. A myoinositol peak is typically present in glial tumors even in the absence of increased choline. It is generally higher in low-grade astrocytomas and gliomatosis cerebri (1416). The rapid T2 relaxation of myoinositol requires a short TE MRS sequence for detection. Glx resonances are also most easily detected with short TE sequences, but are difficult to quantify due to the characteristic-rolling baseline of short TE spectra. Despite this difficulty, several studies have reported elevated Glx in meningiomas relative to normal brain and astrocytomas. Finally, alanine is occasionally found in the spectrum of meningiomas and abscesses. Alanine resonates at 1.47

ppm and is a J-coupled, doublet peak that is inverted at TE values between 135 and 144 ms. It may also overlap with lactate to form an apparent triplet peak. As shown, the combination of changes in dierent metabolites is useful in the dierential diagnosis of brain lesions (Table 2). However, lesion variability, heterogeneity, and overlap between dierent tumor types can make characterization difficult. Other important oncologic features can be evaluated using MRS imaging. Survival time appears to be negatively associated in patients with glioma grade IV containing large areas of abnormal metabolism (high lactate and lipid levels) (19). Besides, tumors are commonly quite heterogeneous. The use of magnetic resonance perfusion imaging may help to localize the best area for spectral evaluation (20). This feature may be useful in order to use MRS imaging for selecting representative areas of the tumor for biopsy. In this setting, low-grade areas of a glioma are generally characterized by relatively high NAA/choline ratios; while high-grade areas are usually marked by lactate and lipid peaks (13). Ideally, regions of high metabolic activity should be sampled. Finally, elevation of choline has been recognized as an important surrogate marker of tumor progression and response to therapy (14, 15). 1H-MRS has also been applied to dierentiate radiation-induced tissue injury from relapsing tumor. Increased choline signal is suggestive of tumor recurrence (21).

Breast cancer

Breast MRI shows high sensitivity but limited specicity for cancer detection. Several studies have reported the role of total

80 JanuaryFebruary 2016 Diagnostic and Interventional Radiology Garca-Figueiras et al.

choline as a marker of breast cancer (Fig. 1) (2227). Adding 1H-MRS to breast MRI may improve the specicity of breast cancer detection from 70% up to 92% (2325). Dynamic contrast-enhanced acquisition may help identify enhancing areas in the tumor (usually associated to an increased metabolism) suitable for spectroscopic examination (26). However, several technical constraints must be considered in breast MRS. In lesions less than 2 cm in diameter, MRS may show a reduced SNR of choline resonance. Although any value of TE can be used in breast 1H-MRS, the scientic published literature recommends the use of long TEs (>135 ms). Despite these challenges, several studies have demonstrated that MRS may dierentiate benign and malignant lesions in the breast (2228). Lesions with detectable choline peaks are suspicious for malignancy, with sensitivity and specicity rates reported as 83% and 85%, respectively (29). Choline quantication in a lesion is considered positive when the peak of total choline at 3.2 ppm is two-times above the baseline (30). Another potential use of in vivo spectroscopy is to monitor tumor response to chemotherapy with a diminution of total choline detected in responder breast cancers to neoadjuvant chemotherapy. However, there is still no consensus on the role of MRS for assessing the tumor response (31).

Prostate cancer

MRS may be used for detection, localization, staging, tumor aggressiveness evaluation, and tumor response assessment of prostate cancer (3242); however, its value in some of these indications has been subject to discussion (39, 41, 43). The use of endorectal coil in prostate MRI may be particularly valuable for inherently lower SNR sequences, such as MRS (39). The use of a higher magnetic eld strength (3.0 T) along with endorectal coil results in higher SNR and improved spectral resolution. Main peaks observed in MRS spectra of the prostate are citrate, creatine, and choline compounds. However, dierent anatomic zones of the healthy prostate show dierent amplitudes for these metabolites as well as dierent (choline+creatine)/citrate integral ratios (33). Citrate is produced in the epithelial cells as an intermediate product in the Krebs cycle. It accumulates in the luminal space of the prostate. The lower citrate peak in prostate cancer is secondary to altered metabolism and reduction of luminal space. Apart from its diagnostic value, MRS oers possibilities for a noninvasive assess-

ment of prostate cancer aggressiveness in vivo. Compared with normal peripheral or benign prostate hyperplasia (BPH) tissues, citrate signals are reduced and those of choline compounds are often increased in prostate cancer (Fig. 4). Combinations of dierent metabolite ratios have been evaluated for detecting aggressive tumors (32, 34, 35). The maximum choline+creatine/ citrate ratio and the maximum choline/ creatine ratio correlated signicantly with aggressiveness. However, owing to the presence of BPH, cancer in central gland is more difficult to discern (36). A commonly used system for the evaluation of MRS in prostate was developed by Jung et al. (37), which reported a standardized scoring system for the evaluation of the spectral data of the peripheral and central gland zones. This scoring system uses a visual classication system and a threshold metabolite approach corresponding to the (choline+creatine)/citrate integral ratio. The accuracy of the scoring system improved when at least three adjacent voxels showed abnormal ndings (metabolite peaks are greater than ve times the standard deviation of noise level). This classication showed good accuracy in dierentiating benign from malignant lesions and excellent interobserver agreement. A multiparametric imaging assessment of prostate lesions based on the Prostate Imaging-Reporting and Data System (PI-RADS) also included a classication of MRS ndings for lesion evaluation (Fig. 4) (38). However, it must be remarked that the new version of this scoring system (PI-RADS v2) does not include the use of MRS, perhaps due to its technical complexity (39). MRS may also be useful in image-guided focal therapy (40) and to evaluate prostate cancer response to dierent therapies (41). In case of androgen deprivation therapy, metabolic evaluation might be challenged. The secondary glandular atrophy causes a reduction of citrate peaks in both tumor and in normal glandular areas and only a slow reduction of choline and creatine peaks in tumor. On the other hand, persistent elevation of choline levels can also indicate ongoing active disease in the prostate gland. Finally, in the evaluation of prostate MRS ndings, sources of false positive/negative ndings must be considered. False positive may be secondary to areas/ lesions that show either reduced citrate levels (i.e., the anterior bromuscular stroma or stromal BPH nodules) or elevated choline levels, such as in the vicinity of seminal vesicles or in the periurethral zone (due to el-

evated levels of glycerophosphocholine in the seminal uid) or in areas of prostatitis. On the other hand, false negative ndings can occur with small or inltrating lesions.

Head and neck cancer

Main clinical applications of MRS in the head and neck area include characterization of the head and neck masses, prediction of treatment response to therapy, and monitoring patients with head and neck cancer undergoing therapy (44). In this setting, previous articles evidenced higher cho-line/creatine ratios in squamous cell carcinoma compared with normal tissues (i.e., muscle). On its part, lymphomas showed higher ratios than that of the carcinomas, which was attributed to high cell density of the lymphomas (44, 45). Concerning tumor response evaluation, King et al. (46) reported that the presence of a choline peak in a post-treatment mass might be a marker of residual cancer. Future applications may include characterization of the lymph nodes. The metastatic lymph nodes showed a signicantly higher choline/creatine ratio compared with benign lymphoid hyperplasia (5).

Hepatobiliary system

There is limited use of MRS in the hepatobiliary system (Table 3). Main limitation of MRS in the hepatobiliary system and gastrointestinal tract is motion. The MRS acquisition and processing protocol can be improved by introducing a control of respiratory motion using breath-hold ac-

Figure 6. A 68-year-old man with hepatocellular carcinoma. Axial contrast-enhanced T1-weighted gradient-echo image in the delayed phase demonstrates a big encapsulated mass (arrows). Single-voxel MR spectrum depicts a choline peak within this mass.

Magnetic resonance spectroscopy in oncology 81

Table 3. Published literature about hepatobiliary tumors

Technical Unsaturated Organ parameter Purpose Choline Lipids fatty acids

Xu et al. (51) Liver 3.0 T To investigate the Lipid accumulation can

PRESS normal hepatic MRS result from the increased TE: 30 ndings of choline/lipids fat in the body depending on age and BMI

Lipids can mask the resonance signal of choline

Fishbach et al. (50) Liver 3.0 T To dierentiate liver No signicant dierences

PRESS parenchyma from were observed between TE: 35 neoplastic lesions the contents of choline in VOI: 222 cm using localized MRS malignant liver tumors and normal liver

parenchyma

Li et al. (48) Liver 3.0 T The quantication of The choline concentrations in choline containing HCCs are substantially compounds in hepatic higher than thosetumors obtained from healthy

volunteers

Kuo et al. (47) Liver 3.0 T The value of in vivo Malignant tumors have

PRESS MRS in the assessment elevated total choline TE: 30 ms of large focal hepatic resonances compared VOI: 322 cm lesion with uninvolved liver or benign tumors

Yao et al. (54) Pancreas 3.0 T To identify metabolic Choline/unsuppressed 1.3 Pancreatic

Respiration- features of pancreatic water ratio in normal Fatty acids/lipids ratio carcinoma has triggered carcinoma pancreas was statistically in normal pancreas was a higher fatty No supressed greater than that in statistically lower than acids/lipids water pancreatic cancer that in pancreatic cancer ratio

(P = 0.006)

Su et al. (52) Pancreas 3.0 T To characterize normal

Compare breath- pancreas metabolismholding and free-breathing

Ma et al. (53) Pancreas No supressed water To analyze the metabolic Lipids may potentially be features and distribution sensitive biomarkersof normal pancreas and for pancreatic cancer

pancreatic adenocarcinoma

PRESS, point-resolved spectroscopy; TE, echo time; MRS, magnetic resonance spectroscopy; BMI, body mass index; VOI, volume of interest; HCC, hepatocellular carcinoma.

quisitions and an abdominal compression belt. Pre- and postprocessing including automatic phase and frequency correction may remove potential distortions introduced mainly by motion. To our knowledge, up to date there is no added value in using MRS in these elds. Application of

1H-MRS studies in the liver aim to characterize the hepatic mass or monitor hepatocellular carcinoma (HCC) treated with chemoembolization. In this setting using a short TE, various groups found that malignant liver tumors present higher levels of choline compared with uninvolved liver or benign tumors (4749) (Fig. 6). Kuo et al. (47) reported signicant decrease of the total choline in the HCC after transcatheter arterial chemoembolization, while lipid and water signals were increased. On the

contrary, Fischbach et al. (50) did not ob-serve any signicant dierence between malignant liver tumors and normal liver parenchyma for the total choline. Due to these divergences, recent studies evaluated the possibility of quantication of choline in the liver. Xu et al. (51) concluded that lipids could mask the resonance signal of choline. However, the ability to reliably distinguish benign and malignant tumors from normal liver parenchyma has yet to be established.

In the pancreas, the main challenge for the radiologist is diagnosing pancreatic cancer. Unfortunately, choline, a classic marker of cancer, is also present in normal parenchyma (52). Several authors indicate lipids as potential markers of pancreatic cancer (53, 54).

Gastrointestinal tumors

A few studies evaluated the value of

1H-MRS in gastrointestinal tumors. Kim et al. (55) proposed using 1H-MRS to diagnose rectal cancer and monitor treatment response after chemoradiotherapy. They found that after treatment, choline peak disappeared, resulting in only the lipid peak at 1.3 ppm in 97% of patients (Fig. 7). Previously, Dzik-Jurazs et al. (56) had detected the same metabolites (choline and lipids) in rectal cancers. Mun et al. (57) used 1H-MRS to determine the characteristics of gastric cancers and found that cancer lesions showed increased choline peaks, decreased lipid levels, and signicant lactate doublet peaks in short TE compared with noncancerous gastric tissue.

82 JanuaryFebruary 2016 Diagnostic and Interventional Radiology Garca-Figueiras et al.

Table 4. Published literature about genitourinary tumors (except prostate)

Technical

Organ parameter Purpose Choline Lipids Lactate Specicity Sensitivity

Takeuchi et al. (62) Uterus 3.0 T The clinical High lipid peak is 94% 100%Single 144 signicance of suggestive ofVoxel 8 mL the lipid peak in uterine sarcomas distinguishing uterine sarcomas Lipid peak is observed from benign in both viable and leiomyomas necrotic areas in

sarcomas

Takeuchi et al. (61) Uterus 3.0 T Distinguishing Malignant Cuto: 7 mM Cuto: 7 mMSingle 144 malignant from (9.212.21 mM) 83% 93% Voxel 8 mL benign lesions

Benign(4.592.22 mM)

Okada et al. (59) Uterus 1.5 T Evaluation of Peak in solid High peak in Present in Single 135 female intrapelvic tumors in dermoid cyst anaerobic Voxel 827 tumors by clinical benign and glycolysis mL proton MRS malignant

tumors

Celik et al. (60) Uterus 1.5 T Clinical utility in Endometrial Endometrial Proliferative Single 136 endometrial carcinoma carcinoma endometrium Voxel 18 mL lesions

Endometrial Secretory hyperplasia endometrium

Payne et al. (63) Cervix 1.5 T To establish Present inSingle 135 dierences normal andVoxel 2518 between cervical tumoral tissue15 mm tumors and total choline No dierencebetween any tumoral types

Mahon et al. (64) Cervix 1.5 T To compare In vivo studies The measured lipid Single 135 in vivo 1H-MRS detected choline levels were more3.4 mL spectra of in normal, cervical than double in preinvasive and intraepithelial malignant cervical invasive cervical neoplasia, and tissue compared lesions with ex vivo cancer patients with benign

MAS spectra of with no cervical tissue intact biopsy signicant dierences in

levels

Booth et al. (58) Cervix 3.0 T To characterize No statisticallySingle 72 the spectra of signicant Voxel a variety of dierence 5.381.3 mm3 benign and between choline malignant levels in various gynecologic lesion types

lesions (P = 0.735) or between benign and malignant disease

Lee et al. (65) Cervix 1.5 T To diagnose Present in Peak at 1.3 ppm isEndovaginal cervical carcinoma adenocarcinoma present in surface coil and categorize the and squamous squamous cell Single spectrum cell carcinoma carcinoma 135 according to

Voxel 13 mL histologic type Peak at 2 ppm (tryglicerides) ispresent in all adenocarcinomas

Takeuchi et al. (70) Adnexa 3.0 T To retrospectively High lipid peak in 92% 100%Single evaluate the thecomas/144 TE signicance of lipid brothecomas with 222 cm peak in in vivo MRS 100% sensitivity,(8 mL) for the diagnosis 92% specicity,of ovarian 88% PPV, and thecomas/ 100% NPV brothecomas

Magnetic resonance spectroscopy in oncology 83

Table 4. (Continued)

McLean et al. (68) Adnexa 3.0 T To characterize Choline was detected

Single 144 primary and in 10/12 primary metastatic ovarian tumors and 5/11 cancer by 1H-MRS metastatic lesions

in vivo

Stanwell et al. (69) Adnexa 3.0 T To provide Choline/creatine

Single potentially integral ratio >3voxel/135 diagnostic was foundbiochemical to indicateinformation that malignancymay aid in the characterization Choline/creatineof ovarian integral ratio <1.5neoplasms in benigndetected during massesclinical MRI

Okada et al. (59) Adnexa 1.5 T / Single Evaluation of The choline peak High peak in High lactate signals

135 / Voxel female intrapelvic was detected in dermoid cyst were detected in 827 mL tumors by clinical the solid part on cystadenocarcinoma proton MRS ovary tumors but not in

cystadenoma

Firat el al. (72) Testicle Univoxel To determine the Choline/lipids Choline/lipids ratio

PRESS pre- and ratio was higher was higher in theTE 31 ms postpubertal 1H in the postpubertalTE 136 ms MRS characteristics postpubertal period101010 or of the normal period 151515 testis The decrease inmm3 Increase choline the lipid peak mayrepresents the represent the eect presence of of testosterone onspermatogenesis testicular tissue or may be dueto histochemical changes initiated by puberty

Aaranson et al. (73) Testicle 1H high- To identify Choline resolution metabolic concentrations MAS signatures are signicantly spectroscopy associated with higher in testes various with histological spermatogenesis states of

spermatogenesis in infertile men

MRS, magnetic resonance spectroscopy; MAS, magic angle spinning; ppm, parts per million; TE, echo time; PPV, positive predictive value; NPV, negative predictive value; PRESS, point-resolved spectroscopy.

Genitourinary tumors (excluding prostate)

There is limited experience with the use of MRS in the evaluation of female pelvic lesions due to the wide range of pathological types of neoplasms with dierent behaviors (Table 4). In general, there was no statistically signicant dierence between choline levels in various types of gynecologic tumors or between benign and malignant lesions (Fig. 8) (58). However, there are some contradictory data. In studies regarding the uterine tumors, Okada et al. (59) and Celik et al. (60) found that choline was present in benign and malignant lesions, while Takeuchi et al. (61) reported that malignant lesions have higher levels of choline, and a cuto of 7 mmol might distinguish between them

with 83% sensitivity and 93% specicity. Lipid peaks also showed promising results for distinguishing uterine sarcomas from benign leiomyomas (62). In the case of tumors of the cervix, unfortunately choline is present in normal and tumor tissue without any detectable dierences between them (58, 6367). Concerning ovarian tumors, dierent authors found higher levels of cho-line in malignant lesions, in tumors as well as metastatic lesions (68). Stanwell et al. (69) determined that the choline/creatine ratio could distinguish between benign and malignant ovary lesions, thus a choline/creatine integral ratio >3 indicated a malignant tumor, whereas a choline/creatine integral ratio less than 1.5 indicated a benign nature.

Takeuchi el al. (70) studied the signicance of lipid peak in patients with solid gyneco-logic tumors with areas of low signal intensity on T2-weighted images. They demonstrated that the presence of high lipid peak might distinguish thecomas/brothecomas from other ovarian brotic neoplastic lesions. Okada et al. (59) investigated the metabolic prole of various gynecologic tumors. They found high levels of choline in the solid part on ovarian tumors, elevated lipid peaks in dermoid cyst and high levels of lactate in the cystadenocarcinomas.

The literature related to the value of

1H-MRS in the testes is rather scarce. Spermatogenesis is a complex process, in which

84 JanuaryFebruary 2016 Diagnostic and Interventional Radiology Garca-Figueiras et al.

a

b

Figure 7. A 74-year-old man with an advanced rectosigmoid malignant tumor. Axial T2-weighted image, constant transfer (Ktrans) parametric map, and time-signal intensity curve (top row) and ADC color-scaled parametric map and single-voxel MR spectrum (bottom row) demonstrate a bulky rectosigmoid tumor (arrows) with areas of increased Ktrans, a type 2 curve (arrowhead), and low ADC values. MRS evidences a lipid peak in the tumor.

Figure 8. A 47-year-old woman with a malignant tumor of the cervix. Sagittal diusion-weighted image (b value=800) (right) demonstrates a big tumor in the uterine cervix with restricted diusion (arrowheads). T2-weighted images in dierent planes (top left) show the position of the MRS voxel. Single-voxel MRS depicts the presence of a choline peak in the tumor.

Figure 9. a, b. A 39-year-old man with a seminoma. Multiparametric MRI evaluations of the tumor (a) and the contralateral healthy testicle (b). Tumor evaluation (a) shows a small lesion on axial plane (white arrows) in the right testicle. T2-weighted image and constant transfer (Ktrans)

parametric map (top row) demonstrate a small lesion with low signal on T2 and increased Ktrans

values which correspond to a seminoma. T2-weighted images in dierent planes (second row) show the position of the MRS voxel. Single-voxel MRS shows a low choline peak (white arrow) in the tumor. MRS of the contralateral healthy testicle (b) in the same patient shows a higher choline peak (white arrow) in the normal parenchyma of this testicle compared with the tumor. (Note: same scale has been used for visual comparison).

the formation of spermatozoa constantly requires large amounts of choline for membrane synthesis (71). Firat et al. (72) revealed two signicant dierences between pre- and postpubertal 1H-MRS pattern: rst, an increase in the choline peak after puberty due to initiation of spermatogenesis and second, a decrease of the lipid peak secondary to increased testosterone synthesis in the testicular tissue, initiated by puberty. They found statistically signicant dierences between choline/lipid ratios of pre- and postpubertal males. In this setting, Aaronson et al. (73) studied three histologic patterns in 27 snap-frozen testicular tissues using 1H-MRS: normal spermatogenesis, maturation arrest, and Sertoli-cell-isolated histology. They found that choline con-

centrations were higher in patients with normal spermatogenesis compared with those having Sertoli-cell-isolated histology. Preliminary results also reported that normal testis presents naturally high levels of choline; however, several causes of infertility (such as varicoceles or testicular tumors) decreases these levels due to a failure of spermatogenesis (Fig. 9) (74).

Soft-tissue tumors

1H-MRS may help dierentiate benign and malignant soft tissue lesions. Absence of choline peak is highly predictive of benign soft tissue lesions as shown by Russo et al. (75) using SVS and long TE. Previously, Subhawong et al. (76) revealed that a discrete choline peak had 88% sensitivity and

68% specicity in detection of malignant musculoskeletal lesions (Fig. 10). Similarly, Doganay et al. (77) reported that choline had 72.2% sensitivity and 83.3% specicity in detecting malignant bone and soft tissue tumors.

MRS challenges

Nowadays most clinical MRI scanners have routine sequences for 1H-MRS measurements, providing a wide range of metabolic information integrated with complementary anatomical or functional MRI sequences. However, MRS shows considerable technical complexity, is time consuming, oers lower sensitivities, and requires complex data processing (Table 5). More-

Magnetic resonance spectroscopy in oncology 85

Table 5. Technical challenges of MRS and possible solutions

Technical problems Possible solutions

Low signal strength of the metabolites Larger voxel sizes

MRI instruments at a higher eld strength

Surface phase-array coils

Poor spatial resolution Increase the signal-to-noise ratio (e.g., MRI systems with a higher eld strength and surface phase-array coils)

Increase acquisition time

Spatial ltering in the reconstruction process

Long acquisition time MRI instruments with a higher eld strength

Parallel imaging

Surface phase-array coils

Constrain the acquired matrix size

Echo-planar or spiral phase-encoding techniques

Spherical or elliptical k-space

Weighted averaging strategies (i.e., collection of fewer averages at peripheral k-space points)

Motion artifacts Motion correction techniques based on navigator signals

Decrease acquisition time

STEAM is more susceptible to the eects of motion than PRESS

Magnetic eld inhomogeneity, which Avoid anatomical regions having strong magnetic eld inhomogeneity (tissue-bone or tissue-air interfaces) introduces regionally varying spectralline broadening Automated or manual shimming

Single-voxel MRS

Artifacts related to water and lipid signals Presaturate water and fat signals

Postacquisition water removal

Use intermediate-to-long echo time due to short lipid signal T2 values

Baseline correction(MR spectra usually exhibit baseline distortions caused by the corruption of the rst few data points in free induction decay or due to the superposition of broad lines arising from lipids or macromolecules. Water sup- pression also leads to baseline imperfections)

J-modulation of multiplet resonances MRI instruments with a higher eld strength(Some metabolite signals exhibit complexmultiplet structures due to the eect of Improve localization performance and try to shorten the TEJ-coupling) (The short TE stimulated echo acquisition mode or STEAM is less sensitive to J-coupling)

Homodecoupled or pure-shift experiments (where the indirect scalar couplings are removed)

Gibbs ringing at tissue boundaries Apply k-space lters in image reconstruction that suppress the outer regions of k-space (At the cost of reduced spatial resolution)

Metabile peak separation MRI systems with a higher eld strength

Two-dimensional MRS at higher magnetic eld strengths to separate the overlapping peaks in an orthogonal dimension

Homodecoupled or pure-shift experiments (where the indirect scalar couplings are removed)

Quantication (MRI scanners are generally not designed to measure absolute signal levels)Metabolite ratios show good correlation with malignancy and can be used to identify suspicious areas

Absolute concentration measurements can be obtained by calibration of measured metabolite signals against a reference signal produced by a material having a known concentration (e.g., an external phantom or an internal tissue)

Improve quantication by correcting T1-related signal saturation and T2 relaxation of the metabolite signals and the reference water signal

Improved data processing and quantitation algorithmsMRS, magnetic resonance spectroscopy; MRI, magnetic resonance imaging; STEAM, stimulated echo acquisition mode; PRESS, point-resolved spectroscopy; TE, echo time.

86 JanuaryFebruary 2016 Diagnostic and Interventional Radiology Garca-Figueiras et al.

Figure 10. A 39-year-old woman with a malignant soft tissue mass corresponding to a dermatobrosarcoma protuberans (bottom left image clinical picture). Coronal and axial T2-weighted images and wash-in parametric map derived from a dynamic contrast-enhanced acquisition (top right) demonstrate a well-vascularized exophytic tumor in the left groin (white arrows). Single-voxel MRS shows a choline peak (bottom right).

References

1. Hajek M, Dezortova M. Introduction to clinical in vivo MR spectroscopy. Eur J Radiol 2008; 67:185193. http://dx.doi.org/10.1016/j.ejrad.2008.03.002

Web End =[CrossRef]

2. Glunde K, Bhujwalla ZM. Metabolic tumor imaging using magnetic resonance spectroscopy. Semin Oncol 2011; 38:2641. http://dx.doi.org/10.1053/j.seminoncol.2010.11.001

Web End =[CrossRef]

3. Glunde K, Bhujwalla ZM, Ronen SM. Choline metabolism in malignant transformation. Nat Rev Cancer 2011; 11:835848. http://dx.doi.org/10.1038/nrc3162

Web End =[CrossRef]

4. Kwock L, Smith JK, Castillo M, et al. Clinical role of proton magnetic resonance spectroscopy in oncology: brain, breast, and prostate cancer. Lancet Oncol 2006; 7:859868. http://dx.doi.org/10.1016/S1470-2045(06)70905-6

Web End =[CrossRef]

5. Jansen JF, Carlson DL, Lu Y, et al. Correlation of a prior DCE-MRI and (1)H-MRS data with molecular markers in neck nodal metastases: Initial analysis. Oral Oncol 2012; 48:717722. http://dx.doi.org/10.1016/j.oraloncology.2012.02.001

Web End =[CrossRef]

6. Yeung DK, Yang WT, Tse GM. Breast cancer: in vivo proton MR spectroscopy in the characterization of histopathologic subtypes and preliminary observations in axillary node metastases. Radiology 2002; 225:190197. http://dx.doi.org/10.1148/radiol.2243011519

Web End =[CrossRef]

7. Skoch A, Jiru F, Bunke J. Spectroscopic imaging: basic principles. Eur J Radiol 2008; 67:230239. http://dx.doi.org/10.1016/j.ejrad.2008.03.003

Web End =[CrossRef]

8. Klose U. Measurement sequences for single voxel proton MR spectroscopy. Eur J Radiol 2008; 67:194201. http://dx.doi.org/10.1016/j.ejrad.2008.03.023

Web End =[CrossRef]

9. Pinker K, Stadlbauer A, Bogner W, Gruber S, Helbich TH. Molecular imaging of cancer: MR spectroscopy and beyond. Eur J Radiol 2012; 81:566577. http://dx.doi.org/10.1016/j.ejrad.2010.04.028

Web End =[CrossRef]

10. Posse S, Otazo R, Dager SR, Alger J. MR spectroscopic imaging: principles and recent advances. J Magn Reson Imaging 2013; 37:13011325. http://dx.doi.org/10.1002/jmri.23945

Web End =[CrossRef]

11. Poullet J-B, Sima DM, Van Huel S. MRS signal quantitation: A review of time- and frequency-domain methods. J Magn Reson 2008; 195:134144. http://dx.doi.org/10.1016/j.jmr.2008.09.005

Web End =[CrossRef]

12. Mandal PK. In vivo proton magnetic resonance spectroscopic signal processing for the absolute quantitation of brain metabolites. Eur J Radiol 2012; 81:e653e664. http://dx.doi.org/10.1016/j.ejrad.2011.03.076

Web End =[CrossRef]

13. Bulik M, Jancalek R, Vanicek J, Skoch A, Mechl M. Potential of MR spectroscopy for assessment of glioma grading. Clin Neurol Neurosurg 2013; 115:146153. http://dx.doi.org/10.1016/j.clineuro.2012.11.002

Web End =[CrossRef] 14. Horsk A, Barker PB. Imaging of brain tumors: MR spectroscopy and metabolic imaging. Neuroim-aging Clin N Am 2010; 20:293310. http://dx.doi.org/10.1016/j.nic.2010.04.003

Web End =[CrossRef]

15. McKnight TR. Proton magnetic resonance spectroscopic evaluation of brain tumor metabolism. Semin Oncol 2004; 31:605617. http://dx.doi.org/10.1053/j.seminoncol.2004.07.003

Web End =[CrossRef]

16. Oz G, Alger JR, Barker PB, et al. Clinical proton MR spectroscopy in central nervous system disorders. Radiology 2014; 270:658679. http://dx.doi.org/10.1148/radiol.13130531

Web End =[CrossRef]

17. Griffin JL, Bollard M, Nicholson JK, et al: Spectral proles of cultured neuronal and glial cells derived from HRMAS (1)H NMR spectroscopy. NMR Biomed 2002; 15:375384. http://dx.doi.org/10.1002/nbm.792

Web End =[CrossRef]

18. Brando L, Castillo M. Adult brain tumors: clinical applications of magnetic resonance spectroscopy. Neuroimaging Clin N Am 2013; 23:527555. http://dx.doi.org/10.1016/j.nic.2013.03.002

Web End =[CrossRef]

19. Crawford FW, Khayal IS, McGue C, et al. Relationship of pre-surgery metabolic and physiological MR imaging parameters to survival for patients with untreated GBM. J Neurooncol 2009; 91:337351. http://dx.doi.org/10.1007/s11060-008-9719-x

Web End =[CrossRef]

20. Chawla S, Wang S, Wolf RL, et al. Arterial spin-labeling and MR spectroscopy in the dierentiation of gliomas. AJNR Am J Neuroradiol 2007; 28:16831689. http://dx.doi.org/10.3174/ajnr.A0673

Web End =[CrossRef]

21. Chernov MF, Hayashi M, Izawa M, et al. Multivoxel proton MRS for dierentiation of radiation-induced necrosis and tumor recurrence after gamma knife radiosurgery for brain metastases. Brain Tumor Pathol 2006; 23:1927. http://dx.doi.org/10.1007/s10014-006-0194-9

Web End =[CrossRef]

22. Suppiah S, Rahmat K, Mohd-Shah MN, et al. Improved diagnostic accuracy in dierentiating malignant and benign lesions using single-voxel proton MRS of the breast at 3T MRI. Clin Radiol 2013; 68:502510. http://dx.doi.org/10.1016/j.crad.2013.04.002

Web End =[CrossRef]

23. Begley JKP, Redpath TW, Patrick J, Bolan PJ, Gil-bert FJ. In vivo proton magnetic resonance spectroscopy of breast cancer: a review of the literature. Breast Cancer Res 2012, 14:207. http://dx.doi.org/10.1186/bcr3132

Web End =[CrossRef]

24. Baara I, rg , Cokun T. Single voxel in vivo proton magnetic resonance spectroscopy of breast lesions: experience in 77 cases. Diagn Interv Radiol 2013; 19:221226.

Magnetic resonance spectroscopy in oncology 87

over, this technique needs radiologists expertise and clinicians are unfamiliar with it. Finally, in many cases, it is not clear how the inclusion of MRS imaging might aect clinical decision-making and outcomes. All these factors continue to limit the application of MRS in the clinical setting (2, 4, 9, 10).

Conclusion

In conclusion, 1H-MRS technique can aid in the management of cancer patients, serving as a noninvasive biomarker of metabolism in tumors. 1H-MRS has achieved great strides as a molecular imaging technique since its introduction, and its scope in many clinical scenarios and research settings is rising. However, MRS needs expertise and is time consuming, which limit its clinical applicability. In this setting, spectra analysis needs to be simplied. Future work should also be concentrated on the evaluation of changes in the spectral pattern as an indicator of response during treatment of malignant disease.

Conict of interest disclosure

A.R. Padhani serves on the advisory board of Siemens Healthcare, speakers bureau of Siemens Healthcare and Johnson & Johnson. He is a researcher of Siemens Healthcare.

Juan Carlos Cobas Gmez is the chief of research of Mestrelab, a company devoted to the development of software applications for chemistry investigation and industry.

25. Tse GM, Cheung HS, Pang LM, et al. Characterization of lesions of the breast with proton MR spectroscopy: comparison of carcinomas, benign lesions, and phyllodes tumors. AJR 2003; 181:12671272 http://dx.doi.org/10.2214/ajr.181.5.1811267

Web End =[CrossRef]

26. Kousi E, Tsougos I, Vasiou E, et al. Magnetic resonance spectroscopy of the breast at 3T: preand post-contrast evaluation for breast lesion characterization. Sci World J 2012; 2012:19. http://dx.doi.org/10.1100/2012/754380

Web End =[CrossRef]

27. Baltzer P, Dietzel M. Breast lesions: diagnosis by using proton MR spectroscopy at 1.5 and 3.0 T- systematic review and meta-analysis. Radiology 2013; 267:735746.http://dx.doi.org/10.1148/radiol.13121856

Web End = [CrossRef]

28. Dorrius MD, Pijnappel RM, Jansen-van der Weide MC, et al. Determination of choline concentrations in breast lesions: quantitative multivoxel proton MR spectroscopy as a promising noninvasive assessment tool to exclude benign lesions. Radiology 2011; 259:695703.http://dx.doi.org/10.1148/radiol.11101855

Web End = [CrossRef]

29. Katz-Brull R, Lavin PT, Lenkinski RE. Clinical utility of proton magnetic resonance spectroscopy in characterizing breast lesions. J Natl Cancer Inst 2002; 94: 11971203. http://dx.doi.org/10.1093/jnci/94.16.1197

Web End =[CrossRef]

30. Jacobs MA, Barker PB, Argani P, et al. Combined dynamic contrast enhanced breast MR and proton spectroscopic imaging: a feasibility study. J Magn Reson Imaging 2005; 21:2328. http://dx.doi.org/10.1002/jmri.20239

Web End =[CrossRef]

31. Le-Petross HC, Hylton N. Role of breast MR imaging in neoadjuvant chemotherapy. Magn Reson Imaging Clin N Am 2010; 18:249258. http://dx.doi.org/10.1016/j.mric.2010.02.008

Web End =[CrossRef]

32. Hoeks CM, Barentsz JO, Hambrock T, et al. Prostate cancer: multiparametric MR imaging for detection, localization, and staging. Radiology 2011; 261:4666. http://dx.doi.org/10.1148/radiol.11091822

Web End =[CrossRef]

33. Verma S, Rajesh A, Ftterer JJ, et al. Prostate MRI and 3D MR spectroscopy: how we do it. AJR Am J Roentgenol 2010; 194:14141426. http://dx.doi.org/10.2214/AJR.10.4312

Web End =[CrossRef]

34. Zakian KL, Sircar K, Hricak H, et al. Correlation of proton MR spectroscopic imaging with gleason score based on step-section pathologic analysis after radical prostatectomy. Radiology 2005; 234:804814. http://dx.doi.org/10.1148/radiol.2343040363

Web End =[CrossRef]

35. Kobus T, Vos PC, Hambrock T, et al. Prostate cancer aggressiveness: in vivo assessment of MR spectroscopy and diusion-weighted imaging at 3T. Radiology 2012; 265:457467. http://dx.doi.org/10.1148/radiol.12111744

Web End =[CrossRef]

36. Zakian KL, Eberhardt S, Hricak H, et al. Transition zone prostate cancer: metabolic characteristics at 1H MR spectroscopic imaginginitial results. Radiology 2003; 229:241247.http://dx.doi.org/10.1148/radiol.2291021383

Web End = [CrossRef]

37. Jung JA, Coakley FV, Vigneron DB, et al. Prostate depiction at endorectal MR spectroscopic imaging: investigation of a standardized evaluation system. Radiology 2004; 233:701708. http://dx.doi.org/10.1148/radiol.2333030672

Web End =[CrossRef]

38. Barentsz JO, Richenberg J, Clements R, et al. ESUR prostate MR guidelines 2012. Eur Radiol 2012; 22:746757. http://dx.doi.org/10.1007/s00330-011-2377-y

Web End =[CrossRef]

39. PI-RADS v2. Prostate Imaging and Reporting and Data System: Version 2. Available at: http:// www.acr.org/~/media/ACR/Documents/PDF/ QualitySafety/Resources/PIRADS/PIRADS%20 V2 Accessed January 9, 2015.

40. Sankineni S, Wood BJ, Rais-Bahrami S, et al. Image-guided focal therapy for prostate cancer. Diagn Interv Radiol 2014; 20:492497. http://dx.doi.org/10.5152/dir.2014.14134

Web End =[CrossRef]

41. Vargas HA, Wassberg C, Akin O, Hricak H. MR imaging of treated prostate cancer. Radiology 2012; 262:2642.http://dx.doi.org/10.1148/radiol.11101996

Web End = [CrossRef]

42. Yu KK , Scheidler J , Hricak H, et al. Prostate cancer: prediction of extracapsular extension with endorectal MR imaging and three-dimensional proton MR spectroscopic imaging. Radiology 1999; 213:481488. http://dx.doi.org/10.1148/radiology.213.2.r99nv26481

Web End =[CrossRef]

43. Platzek I, Borkowetz A, Toma M, et al. Multiparametric prostate magnetic resonance imaging at 3 T: failure of magnetic resonance spectroscopy to provide added value. J Comput Assist Tomogr. 2015 May 1. [Epub ahead of print]. http://dx.doi.org/10.1097/RCT.0000000000000261

Web End =[CrossRef]

44. Abdel Razek AA, Poptani H. MR spectroscopy of head and neck cancer. Eur J Radiol 2013; 82:982989. http://dx.doi.org/10.1016/j.ejrad.2013.01.025

Web End =[CrossRef]

45. Bisdas S, Fetscher S, Feller A, et al. Primary B cell lymphoma of the sphenoid sinus: CT and MRI characteristics with correlation to perfusion and spectroscopic imaging features. Eur Arch Otorhinolaryngol 2007; 264:12071213.http://dx.doi.org/10.1007/s00405-007-0322-0

Web End = [CrossRef]

46. King A, Yeung D, Yu K, et al. Monitoring of treatment response after chemoradiotherapy for head and neck cancer using in vivo 1H MR spectroscopy. Eur Radiol 2010; 20:165172. http://dx.doi.org/10.1007/s00330-009-1531-2

Web End =[CrossRef]

47. Kuo YT, Li CW, Chen CY, Jao J, Wu DK, Liu GC. In vivo proton magnetic resonance spectroscopy of large focal hepatic lesions and metabolite change of hepatocellular carcinoma before and after transcatheter arterial chemoembolization using 3.0-T MR scanner. J Magn Reson Imaging 2004; 19:598604. http://dx.doi.org/10.1002/jmri.20046

Web End =[CrossRef]

48. Li CW, Kuo YC, Chen CY, et al. Quantication of choline compounds in human hepatic tumors by proton MR spectroscopy at 3 T. Magn Reson Med 2005; 53:770776.http://dx.doi.org/10.1002/mrm.20412

Web End = [CrossRef]

49. Ter Voert E, Heijmen L, van Laarhoven H, Heerschap A. ln vivo magnetic resonance spectroscopy of liver tumors and metastases. World J Gastroenterol 2011; 17:51335149. http://dx.doi.org/10.3748/wjg.v17.i47.5133

Web End =[CrossRef]

50. Fischbach F, Schirmer T, Thormann M, Freund T, Ricke J, Bruhn H. Quantitative proton magnetic resonance spectroscopy of the normal liver and malignant hepatic lesions at 3.0 Tesla. Eur Radiol 2008; 18:25492558. http://dx.doi.org/10.1007/s00330-008-1040-8

Web End =[CrossRef]

51. Xu L, Liu B, Huang Y, et al. 3.0 T proton magnetic resonance spectroscopy of the liver: quantication of choline. World J Gastroenterol 2013; 19:14721477. http://dx.doi.org/10.3748/wjg.v19.i9.1472

Web End =[CrossRef]

52. Su TH, Jin EH, Shen H, Zhang Y, He W. In vivo proton MRS of normal pancreas metabolites during breath-holding and free-breathing. Clin Radiol 2012; 67:633637. http://dx.doi.org/10.1016/j.crad.2011.05.018

Web End =[CrossRef]

53. Ma X, Zhao X, Ouyang H, et al. The metabolic features of normal pancreas and pancreatic adenocarcinoma: preliminary result of in vivo proton magnetic resonance spectroscopy at 3.0 T. J Comput Assist Tomogr 2011; 35:539 543. http://dx.doi.org/10.1097/RCT.0b013e318227a545

Web End =[CrossRef]

54. Yao X, Zeng M, Wang H, Fei S, Rao S, Ji Y. Metabolite detection of pancreatic carcinoma by in vivo proton MR spectroscopy at 3T: initial results. Radiol Med 2012; 117:780788.http://dx.doi.org/10.1007/s11547-011-0757-7

Web End = [CrossRef]

55. Kim MJ, Lee SJ, Lee JH, et al. Detection of rectal cancer and response to concurrent chemoradiotherapy by proton magnetic resonance spectroscopy. Magn Reson Imaging 2012; 30:848853. http://dx.doi.org/10.1016/j.mri.2012.02.013

Web End =[CrossRef]

56. Dzik-Jurasz AS, Murphy PS, George M, et al. Human rectal adenocarcinoma: demonstration of 1H-MR spectra in vivo at 1.5 T. Magn Reson Med 2002; 47:809811. http://dx.doi.org/10.1002/mrm.10108

Web End =[CrossRef]

57. Mun CW, Cho JY, Shin WJ, et al. Ex vivo proton MR spectroscopy (1H-MRS) for evaluation of human gastric carcinoma. Magn Reson Imaging 2004; 22:861870. http://dx.doi.org/10.1016/j.mri.2004.01.045

Web End =[CrossRef]

58. Booth SJ, Pickles MD, Turnbull LW. In vivo magnetic resonance spectroscopy of gynaecological tumours at 3.0 Tesla. BJOG 2009; 116:300303. http://dx.doi.org/10.1111/j.1471-0528.2008.02007.x

Web End =[CrossRef]

59. Okada T, Harada M, Matsuzaki K, Nishitani H, Aono T. Evaluation of female intrapelvic tumors by clinical proton MR spectroscopy. J Magn Reson Imaging 2001; 13:912917. http://dx.doi.org/10.1002/jmri.1130

Web End =[CrossRef]

60. Celik O, Sarac K, Hascalik S, Alkan A, Mizrak B, Yologlu S. Magnetic resonance spectroscopy features of uterine leiomyomas. Gynecol Obstet Invest 2004; 58:194201.http://dx.doi.org/10.1159/000080020

Web End = [CrossRef]

61. Takeuchi M, Matsuzaki K, Harada M. Dierentiation of benign and malignant uterine corpus tumors by using proton MR spectroscopy at 3T: preliminary study. Eur Radiol 2011; 21:850 856. http://dx.doi.org/10.1007/s00330-010-1974-5

Web End =[CrossRef]

62. Takeuchi M, Matsuzaki K, Harada M. Preliminary observations and clinical value of lipid peak in high-grade uterine sarcomas using in vivo proton MR spectroscopy. Eur Radiol 2013; 23:23582363. http://dx.doi.org/10.1007/s00330-013-2856-4

Web End =[CrossRef]

63. Payne GS, Schmidt M, Morgan VA, et al. Evaluation of magnetic resonance diusion and spectroscopy measurements as predictive biomarkers in stage 1 cervical cancer. Gynecol Oncol 2010; 116:246252. http://dx.doi.org/10.1016/j.ygyno.2009.09.044

Web End =[CrossRef]

64. Mahon MM, Williams AD, Soutter WP, et al. 1H magnetic resonance spectroscopy of invasive cervical cancer: an in vivo study with ex vivo corroboration. NMR Biomed 2004; 17:19. http://dx.doi.org/10.1002/nbm.869

Web End =[CrossRef]

65. Lee JH, Cho KS, Kim YM, et al. Localized in vivo 1H nuclear MR spectroscopy fo evaluation of human uterine cervical carcinoma. AJR Am J Roentgenol 1998; 170:12791282. http://dx.doi.org/10.2214/ajr.170.5.9574601

Web End =[CrossRef]

66. Allen JR, Prost RW, Griffith OW, Erickson SJ, Erickson BA. In vivo proton (H1) magnetic resonance spectroscopy for cervical carcinoma. Am J Clin Oncol 2001; 24:522529. http://dx.doi.org/10.1097/00000421-200110000-00021

Web End =[CrossRef]

67. Delikatny EJ, Russell P, Hunter JC, et al. Proton MR and human cervical neoplasia: ex vivo spectroscopy allows distinction of invasive carcinoma of the cervix from carcinoma in situ and other preinvasive lesions. Radiology 1993; 188:791796. http://dx.doi.org/10.1148/radiology.188.3.8351349

Web End =[CrossRef]

88 JanuaryFebruary 2016 Diagnostic and Interventional Radiology Garca-Figueiras et al.

68. McLean MA, Priest AN, Joubert I, et al. Metabolic characterization of primary and metastatic ovarian cancer by 1H-MRS in vivo at 3T. Magn Reson Med 2009; 62:855861. http://dx.doi.org/10.1002/mrm.22067

Web End =[CrossRef]

69. Stanwell P, Russell P, Carter J, Pather S, Heintze S, Mountford C. Evaluation of ovarian tumors by proton magnetic resonance spectroscopy at three Tesla. Invest Radiol 2008; 43:745751. http://dx.doi.org/10.1097/RLI.0b013e31817e9104

Web End =[CrossRef]

70. Takeuchi M, Matsuzaki K, Harada M. Preliminary observations and diagnostic value of lipid peak in ovarian thecomas/brothecomas using in vivo proton MR spectroscopy at 3T. J Magn Reson Imaging 2012; 36:907911. http://dx.doi.org/10.1002/jmri.23711

Web End =[CrossRef]

71. Sharpe RM. Regulation of spermatogenesis. In: Knobil E, Neill JD, Eds. The physiology of reproduction. New York: Raven Press, 1994; 13631434.

72. Firat AK, Ura M, Karaka HM, et al. 1H magnetic resonance spectroscopy of the normal testis: preliminary ndings. Magn Reson Imaging 2008; 26: 215220.http://dx.doi.org/10.1016/j.mri.2007.06.008

Web End = [CrossRef]

73. Aaronson DS, Iman R, Walsh TJ, Kurhanewicz J, Turek PJ. A novel application of 1H magnetic resonance spectroscopy: non-invasive identication of spermatogenesis in men with non-obstructive azoospermia. Hum Reprod 2010; 25: 847852.http://dx.doi.org/10.1093/humrep/dep475

Web End = [CrossRef]

74. Baleato-Gonzlez S, Len-Mateos L, Prez-Santiago MI, C. Vilanova JC. Scrotum. In: Luna A, Vilanova, JC, Hygino Da Cruz Jr. LC, Rossi SE (eds.), Functional imaging in oncology, clinical applications - Volume 2. Berlin Heidelberg: Springer-Verlag, 2014; 12091232.

75. Russo F, Mazzetti S, Grignani G, et al. In vivo characterisation of soft tissue tumours by 1.5-T proton MR spectroscopy. Eur Radiol 2012; 22:11311139. http://dx.doi.org/10.1007/s00330-011-2350-9

Web End =[CrossRef]

76. Subhawong TK, Wang X, Durand DJ, et al. Proton MR spectroscopy in metabolic assessment of musculoskeletal lesions. AJR Am J Roentgenol 2012; 198:162172. http://dx.doi.org/10.2214/AJR.11.6505

Web End =[CrossRef]

77. Doganay S, Altinok T, Alkan A, Kahraman B, Karakas HM. The role of MRS in the dierentiation of benign and malignant soft tissue and bone tumors. Eur J Radiol 2011; 79:3337. http://dx.doi.org/10.1016/j.ejrad.2010.12.089

Web End =[CrossRef]

Magnetic resonance spectroscopy in oncology 89