Early prediction of response to Vorinostat inan orthotopic rat glioma modelLi Weia†, Samuel Hongb†, Younghyoun Yoonb, Scott N. Hwangb,Jaekeun C. Parka, Zhaobin Zhangc, Jeffrey J. Olsonc,d, Xiaoping P. Hua andHyunsuk Shimb,d*

Glioblastoma is the most common primary brain tumor and is uniformly fatal despite aggressive surgical andadjuvant therapy. As survival is short, it is critical to determine the value of therapy early on in treatment. Improvedearly predictive assessment would allow neuro-oncologists to personalize and adjust or change treatment sooner tomaximize the use of efficacious therapy. During carcinogenesis, tumor suppressor genes can be silenced by aberranthistone deacetylation. This epigenetic modification has become an important target for tumor therapy. Suberoyla-nilide hydroxamic acid (SAHA, Vorinostat, Zolinza) is an orally active, potent inhibitor of histone deacetylase (HDAC)activity. A major shortcoming of the use of HDAC inhibitors in the treatment of patients with brain tumors is thelack of reliable biomarkers to predict and determine response. Histological evaluation may reflect tumor viabilityfollowing treatment, but is an invasive procedure and impractical for glioblastoma. Another problem is thatresponse to SAHA therapy is associated with tumor redifferentiation and cytostasis rather than tumor size reduction,thus limiting the use of traditional imaging methods. A noninvasive method to assess drug delivery and efficacy isneeded. Here, we investigated whether changes in 1H MRS metabolites could render reliable biomarkers for an earlyresponse to SAHA treatment in an orthotopic animal model for glioma. Untreated tumors exhibited significantlyelevated alanine and lactate levels and reduced inositol, N-acetylaspartate and creatine levels, typical changesreported in glioblastoma relative to normal brain tissues. The 1H MRS-detectable metabolites of SAHA-treatedtumors were restored to those of normal-like brain tissues. In addition, reduced inositol and N-acetylaspartate werefound to be potential biomarkers for mood alteration and depression, which may also be alleviated with SAHAtreatment. Our study suggests that 1H MRS can provide reliable metabolic biomarkers at the earliest stage of SAHAtreatment to predict the therapeutic response. Copyright © 2012 John Wiley & Sons, Ltd.

Keywords: MRS; histone deacetylase inhibitor; glioma; SAHA; Vorinostat; myo-inositol; orthotopic; depression

INTRODUCTION

Glioblastoma (GBM) is the most common and aggressive type ofprimary brain tumor in humans. GBMs involve glial cells andaccount for 52% of all parenchymal brain tumors and 20% ofall intracranial tumors. Despite multimodality treatments, whichconsist of open craniotomy with maximal surgical resection ofthe tumor, followed by concurrent or sequential chemora-diotherapies, anti-angiogenic therapy with bevacizumab andsymptomatic care with corticosteroids, the median survival ofpatients with GBM remains approximately 14months under thebest circumstances (1). Current treatment options for patientswith GBM are limited by both acquired and inherent tumor resis-tance, which include the blood–brain barrier (BBB) and tumorhypoxia (2,3). During carcinogenesis, tumor suppressor genescan be silenced by aberrant histone deacetylation throughhistone deacetylase (HDAC). This epigenetic modification hasbeen recognized with high incidence and has become an impor-tant target for anticancer therapy in GBM (4–6).Histone deacetylation by HDACs is one of the main mecha-

nisms involved in the regulation of gene expression (7). The actionof HDACs on nucleosomal histones leads to condensed coiling ofchromatin, and thus silencing of the expression of various genes,including those implicated in the regulation of cell survival,proliferation, tumor cell differentiation, cell cycle arrest and

* Correspondence to: H. Shim, Department of Radiology and Imaging Sciences,Winship Cancer Institute, Emory University, 1701 Uppergate Drive, C5018,Atlanta, GA 30322, USA. E-mail: hshim@emory.edu

a L. Wei, J. C. Park, X. P. HuDepartment of Biomedical Engineering, Emory University, Atlanta, GA, USA

b S. Hong, Y. Yoon, S. N. Hwang, H. ShimDepartment of Radiology and Imaging Sciences, Emory University, Atlanta,GA, USA

c Z. Zhang, J. J. OlsonDepartment of Neurosurgery, Emory University, Atlanta, GA, USA

d J. J. Olson, H. ShimWinship Cancer Institute, Emory University, Atlanta, GA, USA

† These authors contributed equally to this work.

Abbreviations used: Ala, alanine; Asp, aspartate; BBB, blood–brain barrier; Cho,choline; Cr, creatine; DQC, double quantum coherent; GABA, g-aminobutyrate;GBM, glioblastoma; Glc, glucose; Gln, glutamine; Glu, glutamate; Gly,glycine; GPC, glycerophosphocholine; GSH, glutathione; Gua, guanidoacetate;HDAC, histone deacetylase; HDACi, histone deacetylase inhibitor; Ins, inositol;ISYNA1, inositol-3-phosphate synthase; Lac, lactate; NAA(G), N-acetylaspartate(glutamate); pCho, phosphocholine; pCr, phosphocreatine; PI, phosphati-dylinositol; RT-PCR, reverse transcription-polymerase chain reaction; SAHA,suberoylanilide hydroxamic acid; Sel-DQC, selective-DQC; sI, scyllo-inositol;STEAM, stimulated-echo acquisition mode; Tau, taurine; tCho, total Cho;tCr, total Cr.'

Research Article

Received: 10 August 2011, Revised: 13 December 2011, Accepted: 13 December 2011, Published online in Wiley Online Library: 2012

(wileyonlinelibrary.com) DOI: 10.1002/nbm.2776

NMR Biomed. (2012) Copyright © 2012 John Wiley & Sons, Ltd.

apoptosis (8). HDACs are not limited to involvement in histonedeacetylation, but also act as members of protein complexes toprevent recruitment of transcription factors to the promoterregion of genes, including tumor suppressors, and affect theacetylation status of specific cell cycle regulatory proteins (9).HDAC inhibitors (HDACi) can efficiently suppress HDAC activityand reverse histone deacetylation. In the presence of HDACi,chromatin remains in an open configuration, allowing trans-cription factors to reach DNA promoters and facilitate thetranscription of tumor suppressor genes that check the growthof cancer cells. Over the last 5 years, numerous HDACi have beenevaluated in clinical trials [reviewed in ref. (10)]. They commonlydisplay the ability to hyperacetylate both histone and nonhis-tone targets, resulting in a variety of effects on cancer cells, theirmicroenvironment and immune responses.

Vorinostat (suberoylanilide hydroxamic acid, SAHA, Zolinza™;Merck & Co., Inc., Whitehouse Station, NJ, USA) is an orally active,potent inhibitor of HDAC activity, which was approved by theFood and Drug Administration in 2006 for the treatment of refrac-tory cutaneous T-cell lymphoma (11). To date, responses with thissingle-agent HDACi have been predominantly observed inadvanced hematologic malignancies, including T-cell lymphoma,Hodgkin lymphoma and myeloid malignancies (12). SAHA mainlyinhibits HDAC 1, 2, 3, 6 and 8, selectively upregulating the expres-sion of pro-apoptotic members of the B-cell lymphoma-2 familyand downregulating the expression of anti-apoptotic genes, suchas cyclin-dependent kinase 2, cyclin-dependent kinase 4, cyclinD1 and cyclin D2 (13). SAHA also increases the expression oftumor necrosis factor (14) and induces the acetylation of thechaperone protein heat shock protein 90, which leads to cellularstress and apoptotic cell death (15). Furthermore, animal studieshave shown that SAHA is capable of crossing the BBB, as itincreases H3 and H4 acetylation in brain tissue (14). SAHA is beingevaluated in many types of cancer, including lung, breast andcolon, and it may also be a welcome addition for the treatmentof GBM (16–18). The antitumor activity of SAHA against malig-nant gliomas has been reported (19); however, there are only asmall number of reports examining the effects of SAHA treatmenton gliomas using noninvasive in vivo studies, and few have exam-ined the changes in cerebral metabolites in tumors.

In vivo 1H MRS has been widely used to study brain tumor me-tabolism in preclinical animal models and in clinical research. Itcan identify specific genetic and metabolic changes that occurin malignant tumors without introducing exogenous variables.Traditionally, N-acetylaspartate (NAA) has been used as a markerfor neurons, total choline (tCho) as a marker of cell membranes,creatine (Cr) as a marker for energy metabolism, inositol (Ins) as amarker for glial cells, and lactate (Lac) and alanine (Ala) asproducts of anaerobic pathways (20,21). As a result, metabolicmarkers, detectable by MRS, not only provide information onbiochemical changes, but also define different metabolic tumorphenotypes.

Ins is a carbocyclic polyol that plays an important role as thestructural basis for a number of secondary messengers in eukary-otic cells, including inositol phosphates, phosphatidylinositol (PI)and phosphatidylinositol phosphate lipids. Ins is synthesized fromglucose-6-phosphate in two steps. First, glucose-6-phosphate isisomerized by inositol-3-phosphate synthase (ISYNA1) to inositol-1-phosphate, which is then dephosphorylated by inositolmonophosphatase to give free Ins (22,23). It has been reportedthat the HDAC RPD3 directly regulates ISYNA1 (also known asINO1 or MIP1), and that a mutation in RPD3 can cause a 32-fold

increase in ISYNA1mRNA level (7). Recent studies have shown thatthe Ins level is low in the cerebrospinal fluid of patients withdepression, and that treatment with Ins can improve significantlytheir behavioral test scores (24,25). Depressive symptoms are alsocommon behavioral changes in patients with cancer and in cancersurvivors, and such changes have been reported in 30–50% ofpatients with brain tumors (26,27). Taken together, the tumor Insprofile, detectable through MRS and/or molecular biology, mayserve as a promising biomarker for the therapeutic activity of SAHAtreatment in malignant glioma.Given that the survival time of patients with GBM is very short,

it is critical to determine the therapeutic activity at the earlieststage of SAHA treatment to enable oncologists to make timelychanges in the treatment of patients for whom therapy hasfailed. In this study, changes in the cerebral metabolites in theorthotopic glioma rat model were measured 3 days after SAHAtreatment. The changes were consistent with improved mood-related behavior and an altered gene expression profile. Ourmain aim was to establish reliable biomarkers at the earlieststage of SAHA treatment to predict the therapeutic response.

MATERIALS AND METHODS

Animal model of gliomas and in vivo SAHA administration

9 L rat glioma cells (50 000 cells in a volume of 5 mL) were stereo-tactically injected into the brains of male Fischer 344 rats atstereotactic coordinates 1mm forward of the frontal zero plane,3mm to the right of the midline and 4.5mm deep (28). SAHAwas resuspended in 10% polyethylene glycol (PEG)-200 and90% of 0.5% methylcellulose. The rats bearing intracranial tumorwere treated with an intraperitoneal injection of the vehicle(control group, n= 5) or SAHA starting from day 9 until day 12after the initial tumor cell injection. The treated groups receiveda dose of 25 (n=5), 50 (n= 5) or 75mg/kg/day (n= 12) of SAHA.The control group and the group treated with 75mg/kg/day ofSAHA were subjected to MRS scans on day 12. The tumors werecollected from all groups to measure ISYNA1 mRNA levels follow-ing the MR examinations. All protocols for animal studies werereviewed and approved by the Institutional Animal Care andUse Committee (IACUC) at Emory University, Atlanta, GA, USA.

MRI/MRS protocol

All MR data were acquired with a 9.4 T/20 cm horizontal boreBruker magnet, interfaced to an Avance console (Bruker, Billerica,MA, USA) and equipped with an actively shielded gradient set(inner diameter, 11.6 cm; maximum gradient strength, 200 mT/m;rise time, 110ms). A two-coil actively decoupled imaging set-upwas used (a 3 cm diameter surface coil for reception and a7.2 cm diameter volume coil for transmission) to achieve max-imal signal-to-noise ratio over the cortical and subcorticalareas of interest. Rats were initially anesthetized with 5%isoflurane in O2, which was reduced to 1.5–2% for mainte-nance. The head of the rat was secured using foam paddingto minimize possible movements. Respiration, electrocardio-gram and blood oxygen level were monitored, and rats werekept normothermic at 37 �C using a circulating water blanket.T2-weighted anatomical reference images were acquired witha multislice, multispin echo sequence for tumor localization.Imaging parameters were as follows: TE = 15, 30, 45, 60,75ms; TR = 2500ms; matrix size, 128� 128; slice thickness,

L. WEI ET AL.

wileyonlinelibrary.com/journal/nbm Copyright © 2012 John Wiley & Sons, Ltd. NMR Biomed. (2012)

8mm; 16 slices; field of view, 2.5 cm� 2.5 cm. After localiza-tion, 1H MR spectra were acquired from the tumor side andcontralateral side of each rat using a stimulated-echo acquisi-tion mode (STEAM) sequence with the following parameters:TE = 20ms; TR = 4000ms; TM= 15ms; 2048 complex datapoints; spectral bandwidth, 4000 Hz; voxel size, 5mm� 5mm5mm; 256 averages. The water signal was suppressed by aVAPOR (variable pulse power and optimized relaxationdelays) scheme. The typical linewidth for the water resonanceafter shimming with FASTMAP was 8–10 Hz. All STEAM MRspectral data were analyzed by LC model software (LinearCombination of Model spectra, version 6.2.0), using the watersignal as an internal reference and a 23metabolite basis set includ-ing, but not limited to, glycerophosphocholine (GPC), choline(Cho), phosphocholine (pCho), Cr, phosphocreatine (pCr), gluta-mate (Glu), glutamine (Gln), taurine (Tau), Ins, glycine (Gly), glucose(Glc), N-acetylaspartate(glutamate) [NAA(G)], Ala, g-aminobutyrate(GABA), aspartate (Asp), glutathione (GSH), Lac, guanidoacetate(Gua) and scyllo-inositol (sI). Statistical analysis was performedusing Student’s t-tests.A frequency-selective double quantum coherent (Sel-DQC)

method based on point-resolved spectroscopy (PRESS) (29,30)was implemented to measure the Ala and Lac peaks separately,which overlap the peaks of lipids and macromolecules (Fig. 2a).We used Sel-DQC to selectively detect the methyl protons(�CH3) of Lac at 1.31 ppm, which were coupled with the methy-lene protons (�CH=) of Lac at 4.09 ppm, and the methyl protons(�CH3) of Ala at 1.47 ppm, which were coupled with the methy-lene proton (�CH=) of Ala at 3.8 ppm. For selective Ala detection,Sel-DQC was used to selectively detect the methyl protons(�CH3) of Ala at 1.47 ppm, which were J coupled with the protonof CH of Ala at 3.77 ppm. The DQC path from the excited protoncoupling network of Ala was selected by a gradient combinationwhilst suppressing the other coherence paths (g1 : g2 = 1 : 2, theduration of g1 and g2 were the same). Other parameters were asfollows: TR = 4000ms; TE1 = TE2 = 35ms (TE = 70ms); TM= 14ms;2048 complex data points; spectral bandwidth, 4000Hz; voxelsize, 5mm� 5mm� 5mm; 512 averages. After acquisition, thetime signals were zero filled to 4096 data points and multipliedwith a 3Hz exponential filter.

Cell culture and in vitro SAHA administration

SAHA (NCI/CTEP, Bethesda, MD, USA) was dissolved in dimethylsulfoxide to obtain a 100mM stock solution. The rat glioma cellline 9 L was maintained in Dulbecco’s modified Eagle’s medium(Mediatech, Manassas, MA, USA) supplemented with 10% fetalbovine serum and antibiotics at 37 �C in 5% CO2. 9 L cells wereplated in 100mm cell culture Petri dishes. Cells were then trea-ted 3 days following seeding with fresh medium containingSAHA at concentrations of 0, 0.1, 0.3 and 1mM for 6 h, and werecollected to prepare total RNA. The cells were also incubatedwith 1 mM of SAHA at different incubation time points, and werecollected to assess ISYNA1 mRNA levels. The incubation timeswere 0, 2, 4, 6 and 12 h.

RNA isolation, reverse transcription-polymerase chainreaction (RT-PCR) and real-time RT-PCR

To measure ISYNA1 mRNA levels, total RNA extracted fromin vitro and in vivo samples was prepared with Trizol reagent(Invitrogen, Carlsbad, CA, USA), according to the manufacturer’s

instructions. Total RNA was then reverse transcribed into cDNAin a reaction volume of 20 mL, including 0.5 mg RNA, 200mMdeoxynucleoside triphosphates, 2.5mMMgCl2, 10mMdithiothrei-tol, 8 units RNase inhibitor, 30 units reverse transcriptase and1.25mM random hexamers in 1� RT buffer. A GeneAmp GoldRNA PCR Reagent kit (Applied Biosystems, Foster City, CA, USA)was used for reverse transcription. cDNA synthesis was per-formed at 25 �C for 10min, followed by 42 �C for 45min, andheating the samples to 70 �C for 10min stopped the reaction.Then, the cDNA was stored at 4 �C until use, or immediately usedfor PCR. For the real-time PCR, SYBR Green quantitative PCRamplifications were performed with a multicolor real-time PCRdetection system (Bio-Rad, Hercules, CA, USA). The reactionswere performed in a reaction volume of 20 mL containing10 mL of 2� SYBR Green PCR Master Mix (Applied Biosystems),0.2mM of each forward and reverse primer, and 2 mL of cDNA.The amplifications were initiated at 95 �C for 10min, followedby 40 cycles of 95 �C for 30 s, 54 �C for 30 s and 70 �C for 20 s.The primers for ISYNA1 were 5’-GGAGAGGAGCCAGATCACTG and5’-CAGCACTAGGTCCAGCATGA, and the primers for b-actinwere 5’-GACAGGATGCAGAAGGAGAT and 5’-TGCTTGCTGATCCA-CATCTG (GenBank accession number: NM_001013880).

RESULTS

SAHA restores normal brain tissue-like metabolism inbrain tumors

Tumor localization and size measurements were performed onthe basis of T2-weighted images. Figure 1a shows the T2-weighted images obtained from two representative intracranialtumor-bearing rats. The tumors were typically hyperintense inthe T2-weighted images. After 3 days of treatment, the average

Untreated SAHA-treated

(a)

(b)

InositolC

holine

Creatine

NA

A

Lac/LipidsA

lanine

4.0 3.5 3.0 2.5 2.0 1.5 1.0 ppm

SAHA-treated tumor

SAHA-treatedcontrolateral

Untreated tumor

Figure 1. (a) T2-weighted images of untreated (left) and suberoylanilidehydroxamic acid (SAHA)-treated (right) rats, on which representative regionsof interest used for MR spectroscopy are shown (TR/TE = 2500/45ms).(b) In vivo 1H MRS (stimulated-echo acquisition mode, STEAM) obtainedfrom the tumor side with (red) and without (black) SAHA treatment. Thegray dotted line is the signal obtained from the SAHA-treated contralat-eral side (nontumor side). Lac, lactate; NAA, N-acetylaspartate.

BIOMARKERS FOR VORINOSTAT RESPONSE

NMR Biomed. (2012) Copyright © 2012 John Wiley & Sons, Ltd. wileyonlinelibrary.com/journal/nbm

tumor size was smaller in SAHA-treated rats (90.3� 18.7mm3)than in untreated rats (119.8� 21.5mm3).

1H MR spectra obtained from the tumor side and contralateralside of each rat were localized on the basis of the T2-weightedimages. Figure 1b shows the STEAM MRS results obtained fromtwo representative rats (SAHA-treated, red; untreated, black). Incontrast with the contralateral side, all tumor spectra demon-strated decreased levels of Cr, NAA and Ins with significantlyincreased levels of Cho, Lac and Ala (p< 0.05). After 3 days oftreatment, the spectra obtained from the tumors of SAHA-treatedrats showed increased levels of NAA, Cr and Ins, and decreasedlevels of Ala and Lac, in comparison with untreated animals.The concentrations of each metabolite in the spectra are shownin Figure 2b. A decrease in Ala and Lac levels in the tumorafter SAHA treatment was confirmed by the Sel-DQC spectral

results Metabolite concentrations (mean� standard deviation)are summarized in Table 1.

SAHA improves mood-related behavior in rats

Daily behavioral notes for SAHA-treated and untreated tumor-bearing rats were recorded until the endpoint of each rat. Ingeneral, SAHA-treated rats demonstrated increased movementrelative to untreated rats, as measured by a marked increase inthe number of grooming, swaying, sniffing and climbing-likebehaviors. Moreover, untreated rats showed a significantlydecreased amount of activity in general and exhibited a generallack of response to opening of the cage lid. These observationsindicate that SAHA may improve depression-induced behaviorin rats bearing gliomas. Interestingly, the behavioral difference

LacAla

(a)

(b)

RF

Gradient Gz

ACQ90°x

Gy Gx

Echo

TE1/2

90°x90°x

TM

Frequency selectiveRF pulse

2g1g

TE2/2TE2/2Time TE1/2

3.0 2.0 1.52.5 PPM 2.0 1.52.5 PPM

180°y 180°y

Figure 2. (a) Frequency-selective double quantum coherent (Sel-DQC) sequence for alanine (Ala) and lactate (Lac) detection. RF, radiofrequency. (b) Invivo 1H Sel-DQC spectrum of Ala and Lac obtained from tumors with (red) and without (black) suberoylanilide hydroxamic acid (SAHA) treatment (TR/TE/TM=3000/70/14ms).

Table 1. Metabolite concentrations (mM) as mean� standard deviation

Metabolite

Cho NAA+NAAG Cr Ins Lac AlaSAHA contralateral (n= 8) 1.42� 0.67 9.34� 0.49 11.25� 0.7 6.36� 0.72 1.39� 0.73 0.2� 0.16SAHA tumor 2.34� 1.01 5.47� 0.8* 9.07� 0.85** 5.58� 0.95* 3.45� 0.81** 1.26� 0.54**Untreated contralateral (n= 6) 1.24� 0.74 7.89� 0.59 9.92� 0.7 5.54� 0.73 1.16� 0.33 0.17� 0.16Untreated tumor 2.28� 1.43 2.38� 1.14 4.96� 2.02 3.80� 0.37 6.05� 2.22 3.48� 0.85

Ala, alanine; Cho, choline; Cr, creatine; Ins, inositol; Lac, lactate; NAA(G), N-acetylaspartate(glutamate).Statistical analysis was performed with Student’s t-tests.*p< 0.05 compared with ‘Untreated tumor’.**p< 0.01 compared with ‘Untreated tumor’.

L. WEI ET AL.

wileyonlinelibrary.com/journal/nbm Copyright © 2012 John Wiley & Sons, Ltd. NMR Biomed. (2012)

between the treated and untreated groups correlated with thedifference in the MRS Ins levels between these groups.

SAHA increases the tumor ISYNA1 mRNA level in vivo

Implanted tumors from the group treated with different dosesof SAHA (25, 50 and 75mg/kg/day) and from the untreatedgroup (vehicle) were collected for real-time RT-PCR analysisimmediately after MRS assessment. Figure 3 shows the relativefold change of the ISYNA1 mRNA level in tumors from bothgroups. The ISYNA mRNA level showed a dose-dependentincrease in response to treatment, in which an approximatelyfive-fold increase in the average mRNA level was induced by75mg/kg/day of SAHA treatment; 25 and 50mg/kg/day ofSAHA treatment produced an approximately three-fold increasein the ISYNA mRNA level. The increased ISYNA mRNA level inSAHA-treated tumors correlates with the increased MRS Inslevels seen in the tumors of the treated group.

SAHA increases 9 L cell ISYNA1 mRNA level in vitro

To assess the effect of SAHA on ISYNA1 mRNA levels in vitro, 9 Lcells grown in complete medium were exposed to four differentconcentrations of SAHA for 6 h. Figure 4a shows the RT-PCR

results of the ISYNA1 mRNA level in cells treated with variousconcentrations of SAHA. The ISYNA mRNA levels increased in adose-dependent manner up to 1 mM.

The 9 L rat glioma cell line was also treated with 1 mM of SAHAfor different incubation times. Figure 4b shows the relative foldchange of the ISYNA mRNA level in cells treated with 1mM SAHAand incubated for 0, 2, 4, 6 and 12 h. The fold change wasgreatest between 2 and 4h post-treatment, reaching an approxi-mately six-fold increase at 4 h.

DISCUSSION

SAHA, also known as Vorinostat, is an orally active, potent inhib-itor of HDAC1, 2, 3 and 6 activities (31). SAHA not only has adirect impact on gene transcription via the inhibition of HDACfunction on histone tails, but also targets gene transcription byindirect mechanisms via the inhibition of HDAC interactions withnonhistone proteins. At high concentration, SAHA consequentlyresults in cellular stress and apoptotic cell death (15).

It has been reported that SAHA has antitumor activitiesagainst malignant glioma cell lines in vitro and ex vivo (19). Ithas also been reported that intracranial administration of SAHAusing an orthotopic glioma mouse model increased the survivaltime from 22 to 42 days (32). In this short-term study, histologicalanalysis showed an 80% reduction in tumor volume in the SAHA-treated group. This reduction in tumor volume was associatedwith a significant increase in the apoptosis rate and a decreasein proliferation and angiogenesis. In another animal study,in vivo murine experiments demonstrated that SAHA (10mg/kgintravenously or 100mg/kg intraperitoneally) could cross theBBB, as shown by significantly increased levels of acetyl-H3 andacetyl-H4 in the brain tissue (14). Furthermore, SAHA significantly(p< 0.05) inhibited the proliferation of GL26 glioma cells grow-ing in the mouse brain and increased their survival. Eyupogluet al. (19) reported that a single intratumoral injection of SAHAin vivo, 7 days after orthotopic implantations of glioma cells insyngeneic rats, doubled their survival time. Galanis et al. (33)reported that SAHA was well tolerated in patients with recurrentGBM, and nine of the 52 patients were progression free at6months after 14 days of SAHA treatment. These observationssupport SAHA as a possible and promising pharmacotherapyfor malignant gliomas.

Conventional MRI of high-grade gliomas characteristicallyexhibits vasogenic edema and contrast enhancement. Highsignal intensity is typically displayed in T2-weighted and fluid-attenuated inversion recovery images (FLAIR), whereas low inten-sity is displayed in T1-weighted images (34). For nonenhancinggliomas, the response to therapy cannot be gauged by MRIalone. A further difficulty is that the response in many cases isassociated with tumor stasis, rather than tumor shrinkage (35),limiting the utility of traditional imaging methods. SAHA canmodulate numerous genes and proteins involved in cell survival,proliferation, differentiation and apoptosis, which may be associ-ated with changes in visible MRS metabolites within the tumor,such as NAA, Cr, Ins and Lac, etc.

1H MRS is an important tool for the detection of metabolitelevels in many diseases. It is also currently used as a noninvasivemeans of classifying and grading brain tumors (36,37). Weiset al. (37) reported that the concentration of tCr may be apotential biomarker for the grading of malignant gliomas. In

ISY

NA

1 m

RN

A fo

ld c

hang

e

Control 25 50 75 (mg/kg/day)

6

4

3

2

1

0

5

Figure 3. Dose-dependent relative fold change of inositol-3-phosphatesynthase (ISYNA1) mRNA level in tumors from the suberoylanilide hydro-xamic acid (SAHA)-treated and untreated (vehicle) groups.

1 0.3 0.1 DMSO 0 NC SAHA (µM)

ISYNA1

Actin

0

1

2

3

4

5

6

7

8

0 2 4 6 12 (hours)

ISY

NA

1 m

RN

A fo

ld c

hang

e

(a)

(b)

Figure 4. (a) Inositol-3-phosphate synthase (ISYNA1) mRNA level in 9 L cellswith different doses of suberoylanilide hydroxamic acid (SAHA) adminis-tration. DMSO, dimethyl sulfoxide; NC, negative control. (b) Time-dependentrelative fold change of ISYNA1 mRNA level in 9L cells after administrationof 1mM of SAHA.

BIOMARKERS FOR VORINOSTAT RESPONSE

NMR Biomed. (2012) Copyright © 2012 John Wiley & Sons, Ltd. wileyonlinelibrary.com/journal/nbm

astrocytomas, several studies have suggested that the tCho con-centration positively correlates with tumor grade. However, GBMsusually have lower tCho concentrations relative to grade II and IIIastrocytomas. This observation could be explained by tumor ne-crosis, which is associated with a decrease in all metabolites andan increase in Lac/lipid levels (38). The concentration of tCho istherefore affected by the tumor prognosis and amount of necroticcore in the tumor (39). Recent investigations have suggested that1H MRS may also be an effective prognostic tool for monitoringthe patient response to treatment, as the technique can localizeregions of heterogeneous tumor regression during and/or follow-ing treatment. The treatment response data may, in turn, help toidentify those patients who would probably benefit from specifictherapies. In a study of 39 patients with malignant gliomas, Li et al.(20) observed that, prior to surgical resection, large volumes ofcontrast enhancement, high levels of Cho and Lac, and low levelsof NAA and Cr were associated with a poor prognosis. In a studyusing 1H MRS to monitor 16 patients with malignant gliomatreated with tamoxifen, Sankar et al. (40) found that, prior totreatment, responders had higher levels of Cr and NAA relativeto nonresponders, but had lower levels of Lac. Levels of Cho didnot differ between the two groups. Thus, the association of meta-bolites with the treatment response also appears to depend onthe treatment modality.

The purpose of the current work was to establish reliablemetabolic MRS biomarkers at the earliest stage of SAHA treat-ment to predict the therapeutic response of GBMs. Althoughthe 9 L rat orthotopic brain tumor model is not completely anal-ogous to human gliomas, as it grows as a well-defined solid massrather than as an ill-defined infiltrating tumor, it provides amodel system in which to develop the concept of SAHA MRS.However, the observations described in the current work mustbe confirmed in human tumors. In comparison with the contra-lateral normal-appearing hemisphere, all tumor spectra exhib-ited lower levels of NAA, tCr, Ins and Gln/Glu, with higher levelsof Cho, Lac and Ala, which is consistent with previous reportsin clinical studies of GBMs (38,41). In comparison with MRS ofthe untreated group, MRS obtained from tumors of the SAHA-treated group showed higher levels of NAA, Cr and Ins, with sig-nificantly lower levels of Ala and Lac. There were no significantchanges in Cho concentration. These observations suggest thatthere is an improvement in the metabolic profile within thetumors only 3 days after SAHA treatment.

NAA is an important metabolite mainly stored within neurons,and the intensity of the NAA peak at 2.02 ppm is considered as aspectroscopic neuronal marker for intact axons. Normally, NAAhas been reported to be dramatically reduced in high-gradegliomas (41). The decrease in NAA level is widely interpretedas the loss, dysfunction or displacement of normal neuronaltissues (42). In the current work, NAA concentrations were higherwithin the tumors of SAHA-treated rats than in those ofuntreated rats. This observation suggests that there are moreviable neurons in the tumor as a result of treatment.

The total Cr peak at approximately 3.02 ppm is reduced inmalignant gliomas. Several studies have found that tCr may be asensitive biomarker for tumor response to therapies (40,43). Cr isa nitrogenous organic acid that occurs naturally in vertebratesand helps to supply energy to all cells in the body. It is thoughtto be low in an active tumor because of energy consumed bydividing neoplastic cells. Consequently, the observed increase inthe concentration of tCr in the tumors of SAHA-treated rats mayindicate a decrease in rapid tumor cellular division.

Lac and Ala are byproducts of anaerobic metabolism and areusually elevated in high-grade gliomas because of their constitu-tive use of anaerobic pathways as an energy source. In addition,the intratumoral accumulation of Lac is attributed to theimpeded clearance of necrotic tissue within gliomas (44). Thedecrease in the concentration of Lac in the tumors of SAHA-treated rats may suggest the amelioration of necrosis and energyconsumption as a result of a decrease in tumor cell division.The most interesting observation in this study is that the

tumors of SAHA-treated rats demonstrate a higher concentrationof Ins. An increase in the Ins level is usually observed in lowergrade malignant gliomas and is explained by the infiltration orproliferation of active malignant glial cells. In this case, thedecrease in NAA may be associated with an increase in Ins (25).Ins can also be used as a biomarker for the grading of malignantgliomas (24) as it is increased in low-grade astrocytomas, butdecreased in GBMs. In our study, the visible MRS Ins level inthe tumor increased significantly after 3 days of SAHA treatment.Recent studies have shown that the Ins level is low in the cere-brospinal fluid of patients with depression, and that treatingthem with Ins could significantly improve their behavioral testscores (24,25). ISYNA1 is the bottleneck enzyme in the produc-tion of myo-inositol. Previously, ISYNA1 (also known as INO1)has been reported to be regulated directly by an HDAC RPD3(7). Consistent with this previous report, both our in vivo andin vitro SAHA administration results showed that there was adose-dependent increase in the level of ISYNA1 mRNA, demon-strating that SAHA could increase Ins production in tumor cells.Thus, the ISYNA1 mRNA level, together with the MRS Ins profile,may be utilized as an independent biomarker obtainablethrough biopsy. The recovery of NAA and Ins within a tumor sug-gests that SAHA may induce redifferentiation of brain tumors, inwhich the redifferentiated cells regain normal function and thusproduce an elevated amount of Ins and NAA (45). The potentialredifferentiation effect of SAHA sustained by the restoration ofthe Ins level may also be the underlying etiology of the observedmood stabilization. Mood-stabilizing actions of HDAC inhibitorshave also been reported recently by multi-institutional preclini-cal studies (46). Ins is a simple isomer of Glc and is a keymetabolic precursor in the PI cycle. Unlike L-DOPA (L-3,4-dihydroxyphenylalanine) and tryptophan, which are aminoacid precursors of monoamine neurotransmitters and have beenreported to have antidepressant properties, Ins is a precursorof an intracellular second messenger system. The PI cycle is asecond messenger system for numerous neurotransmitters (47).Barkai (48) reported that depressed patients, both unipolar andbipolar, had markedly reduced levels of Ins in cerebrospinal fluid.In a study of 11 depressed patients who had been resistant toprevious antidepressant treatments, Ins treatment led to adecline in the mean Hamilton Depression Scale from 31.766 to16.269. It was shown that a 12 g daily intake of Ins raised cere-brospinal fluid Ins levels by 70% (49). Levine (50), in a reviewarticle, made a comparison of eight controlled studies of Instreatment in different psychiatric disorders, and suggested thatIns had therapeutic effects in the spectrum of illnesses responsiveto serotonin selective reuptake inhibitors, including depression,panic and serotonin selective reuptake inhibitors obsessive–compulsive disorder, but was not beneficial in schizophrenia,dementia, Alzheimer’s disease, attention deficit disorder, autismor electroconvulsive therapy-induced cognitive impairment.Depression, an imbalance in mood stabilization, is a common

and important complication of primary cerebral glioma. The

L. WEI ET AL.

wileyonlinelibrary.com/journal/nbm Copyright © 2012 John Wiley & Sons, Ltd. NMR Biomed. (2012)

symptoms of depression may be a consequence of tumor,surgery and/or adjuvant treatments. It has been reported thatmore than 30% of patients with glioma have depression and thatmost patients with glioma exhibit symptoms of depression aftersurgery (26,27). In our study, the mood of tumor-bearing rats wasevaluated. SAHA-treated rats bearing tumors showed improvedmood-related behavior relative to untreated rats, including signsof increased appetite, social interactions and locomotive activi-ties. Combined with the increased levels of NAA and Ins, theimproved mood-related behavior observed in SAHA-treated ratsmay indicate that SAHA is effective in treating both high gradegliomas and glioma-related depression.In conclusion, we have established reliable metabolic biomar-

kers that predict the therapeutic response at the earliest stage ofSAHA treatment. After only 3 days of SAHA treatment, the con-centrations of NAA, Cr and Ins increased, and the concentrationsof Lac and Ala decreased, as noninvasively measured by 1H MRS.ISYNA1 mRNA levels from in vivo and in vitro samples confirmedthe dose-dependent responses to SAHA treatment. Mood-related behavior in rats with brain tumors was improved bySAHA treatment. Our results indicate that SAHA inhibits tumorgrowth and induces normal brain tissue-like metabolism in braintumors. SAHA, as an HDAC inhibitor, is potentially an excitingantitumor agent for GBM treatment that induces redifferentia-tion, rather than cell death. The establishment of reliable MRSbiomarkers, as well as molecular pathological biomarkers, forthe assessment of an early response would clearly be of valuein personalizing the management of patients with GBM byaiding clinicians to adjust the dose of SAHA treatment or toswitch to alternative treatment. Importantly, the translation ofour MRS-based tool will assess the restoration of normal brainmetabolism and indirectly monitor the subject’s quality of lifein the clinic.

Acknowledgements

This work was supported by National Institutes of Health (NIH)grants CA109366 (HS) and CA128301 (HS, JJO and XPH). This workwas supported in part by the Biomedical Imaging TechnologyCenter. We thank Ms Jessica Paulishen for careful reading of themanuscript and helpful remarks. We thank Drs Thaddeus Paceand Andrew Miller for their help with the interpretation of depres-sive behavior. We thank the National Cancer Institute/CancerTherapy Evaluation Program (NCI/CTEP) for supplying a preclinicalbatch of Vorinostat.

REFERENCES1. Stupp R, Mason WP, van den Bent MJ, Weller M, Fisher B, Taphoorn

MJ, Belanger K, Brandes AA, Marosi C, Bogdahn U, Curschmann J,Janzer RC, Ludwin SK, Gorlia T, Allgeier A, Lacombe D, CairncrossJG, Eisenhauer E, Mirimanoff RO. Radiotherapy plus concomitantand adjuvant temozolomide for glioblastoma. N. Engl. J. Med.2005; 352(10): 987–996.

2. Scott AW, Tyler BM, Masi BC, Upadhyay UM, Patta YR, Grossman R,Basaldella L, Langer RS, Brem H, Cima MJ. Intracranial microcapsuledrug delivery device for the treatment of an experimental gliosar-coma model. Biomaterials 2011; 32(10): 2532–2539.

3. Tredan O, Galmarini CM, Patel K, Tannock IF. Drug resistance and thesolid tumor microenvironment. J. Natl. Cancer Inst. 2007; 99(19):1441–1454.

4. Nagarajan RP, Costello JF. Molecular epigenetics and genetics inneuro-oncology. Neurotherapeutics 2009a; 6(3): 436–446.

5. Nagarajan RP, Costello JF. Epigenetic mechanisms in glioblastomamultiforme. Semin. Cancer Biol. 2009b; 19(3): 188–197.

6. Natsume A, Kondo Y, Ito M, Motomura K, Wakabayashi T, Yoshida J.Epigenetic aberrations and therapeutic implications in gliomas.Cancer Sci. 2010; 101(6): 1331–1336.

7. Rundlett SE, Carmen AA, Suka N, Turner BM, Grunstein M. Transcrip-tional repression by UME6 involves deacetylation of lysine 5 ofhistone H4 by RPD3. Nature 1998; 392(6678): 831–835.

8. Jones PA, Baylin SB. The fundamental role of epigenetic events incancer. Nat. Rev. Genet. 2002; 3(6): 415–428.

9. Ellis L, Pili R. Histone deacetylase inhibitors: advancing therapeuticstrategies in hematological and solid malignancies. Pharmaceuticals(Basel) 2010; 3(8): 2411–2469.

10. Boumber Y, Issa JP. Epigenetics in cancer: what’s the future? Oncology(Williston Park) 2011; 25(3): 220–226, 228.

11. Piekarz RL, Robey RW, Zhan Z, Kayastha G, Sayah A, Abdeldaim AH,Torrico S, Bates SE. T-cell lymphoma as a model for the use of histonedeacetylase inhibitors in cancer therapy: impact of depsipeptide onmolecular markers, therapeutic targets, and mechanisms of resis-tance. Blood 2004; 103(12): 4636–4643.

12. Prince HM, Bishton MJ, Harrison SJ. Clinical studies of histone deace-tylase inhibitors. Clin. Cancer Res. 2009; 15(12): 3958–3969.

13. Fandy TE, Shankar S, Ross DD, Sausville E, Srivastava RK. Interactiveeffects of HDAC inhibitors and TRAIL on apoptosis are associatedwith changes in mitochondrial functions and expressions of cellcycle regulatory genes in multiple myeloma. Neoplasia 2005; 7(7):646–657.

14. Yin D, Ong JM, Hu J, Desmond JC, Kawamata N, Konda BM, BlackKL, Koeffler HP. Suberoylanilide hydroxamic acid, a histonedeacetylase inhibitor: effects on gene expression and growth ofglioma cells in vitro and in vivo. Clin. Cancer Res. 2007; 13(3):1045–1052.

15. Rodriguez-Gonzalez A, Lin T, Ikeda AK, Simms-Waldrip T, Fu C,Sakamoto KM. Role of the aggresome pathway in cancer: targetinghistone deacetylase 6-dependent protein degradation. Cancer Res.2008; 68(8): 2557–2560.

16. Komatsu N, Kawamata N, Takeuchi S, Yin D, Chien W, Miller CW,Koeffler HP. SAHA, a HDAC inhibitor, has profound anti-growthactivity against non-small cell lung cancer cells. Oncol. Rep. 2006;15(1): 187–191.

17. Cohen LA, Marks PA, Rifkind RA, Amin S, Desai D, Pittman B, RichonVM. Suberoylanilide hydroxamic acid (SAHA), a histone deacetylaseinhibitor, suppresses the growth of carcinogen-induced mammarytumors. Anticancer Res. 2002; 22(3): 1497–1504.

18. Hsi LC, Xi X, Lotan R, Shureiqi I, Lippman SM. The histone deacetylaseinhibitor suberoylanilide hydroxamic acid induces apoptosis viainduction of 15-lipoxygenase-1 in colorectal cancer cells. CancerRes. 2004; 64(23): 8778–8781.

19. Eyupoglu IY, Hahnen E, Buslei R, Siebzehnrubl FA, Savaskan NE,Luders M, Trankle C, Wick W, Weller M, Fahlbusch R, Blumcke I.Suberoylanilide hydroxamic acid (SAHA) has potent anti-gliomaproperties in vitro, ex vivo and in vivo. J. Neurochem. 2005; 93(4): 992–999.

20. Li X, Jin H, Lu Y, Oh J, Chang S, Nelson SJ. Identification of MRI and 1HMRSI parameters that may predict survival for patients with malig-nant gliomas. NMR Biomed. 2004; 17(1): 10–20.

21. Oh J, Henry RG, Pirzkall A, Lu Y, Li X, Catalaa I, Chang S, Dillon WP,Nelson SJ. Survival analysis in patients with glioblastoma multiforme:predictive value of choline-to-N-acetylaspartate index, apparentdiffusion coefficient, and relative cerebral blood volume. J. Magn.Reson. Imaging 2004; 19(5): 546–554.

22. Salway JG.Metabolism at a Glance. Blackwell Publishing: Malden, MA;2004.

23. Lehninger AL, Nelson DL, Cox MM. Lehninger Principles of Biochemis-try, 4th edn. WH Freeman: New York; 2005.

24. Castillo M, Smith JK, Kwock L. Correlation of myo-inositol levels andgrading of cerebral astrocytomas. Am. J. Neuroradiol. 2000; 21(9):1645–1649.

25. Hattingen E, Raab P, Franz K, Zanella FE, Lanfermann H, Pilatus U.Myo-inositol: a marker of reactive astrogliosis in glial tumors? NMRBiomed. 2008; 21(3): 233–241.

26. Talacchi A, Santini B, Savazzi S, Gerosa M. Cognitive effects of tumourand surgical treatment in glioma patients. J. Neuro-oncol. 2010;103(3): 541–549.

27. Rooney AG, Carson A, Grant R. Depression in cerebral gliomapatients: a systematic review of observational studies. J. Natl. CancerInst. 2011; 103(1): 61–76.

BIOMARKERS FOR VORINOSTAT RESPONSE

NMR Biomed. (2012) Copyright © 2012 John Wiley & Sons, Ltd. wileyonlinelibrary.com/journal/nbm

28. Kang SH, Cho HT, Devi S, Zhang Z, Escuin D, Liang Z, Mao H, Brat DJ,Olson JJ, Simons JW, Lavallee TM, Giannakakou P, Van Meir EG, ShimH. Antitumor effect of 2-methoxyestradiol in a rat orthotopic braintumor model. Cancer Res. 2006; 66(24): 11 991–11 997.

29. Allen PS, Thompson RB, Wilman AH. Metabolite-specific NMR spec-troscopy in vivo. NMR Biomed. 1997; 10(8): 435–444.

30. Zhao T, Heberlein K, Jonas C, Jones DP, Hu X. New double quantumcoherence filter for localized detection of glutathione in vivo. Magn.Reson. Med. 2006; 55(3): 676–680.

31. Bradner JE, West N, Grachan ML, Greenberg EF, Haggarty SJ, WarnowT, Mazitschek R. Chemical phylogenetics of histone deacetylases.Nat. Chem. Biol. 2010; 6(3): 238–243.

32. Ugur HC, Ramakrishna N, Bello L, Menon LG, Kim SK, Black PM,Carroll RS. Continuous intracranial administration of suberoylanilidehydroxamic acid (SAHA) inhibits tumor growth in an orthotopicglioma model. J. Neuro-oncol. 2007; 83(3): 267–275.

33. Galanis E, Jaeckle KA, MaurerMJ, Reid JM, AmesMM, Hardwick JS, ReillyJF, Loboda A, Nebozhyn M, Fantin VR, Richon VM, Scheithauer B,Giannini C, Flynn PJ, Moore DF Jr, Zwiebel J, Buckner JC. Phase II trialof vorinostat in recurrent glioblastoma multiforme: a north centralcancer treatment group study. J. Clin. Oncol. 2009; 27(12): 2052–2058.

34. Panigrahy A, Bluml S. Neuroimaging of pediatric brain tumors: frombasic to advanced magnetic resonance imaging (MRI). J. ChildNeurol. 2009; 24(11): 1343–1365.

35. Kelly WK, Richon VM, O’Connor O, Curley T, MacGregor-Curtelli B, TongW, Klang M, Schwartz L, Richardson S, Rosa E, Drobnjak M, Cordon-Cordo C, Chiao JH, Rifkind R, Marks PA, Scher H. Phase I clinical trial ofhistone deacetylase inhibitor: suberoylanilide hydroxamic acid admin-istered intravenously. Clin. Cancer Res. 2003; 9(10 Pt 1): 3578–3588.

36. Opstad KS, Ladroue C, Bell BA, Griffiths JR, Howe FA. Linear discrim-inant analysis of brain tumour (1)H MR spectra: a comparison ofclassification using whole spectra versus metabolite quantification.NMR Biomed. 2007; 20(8): 763–770.

37. Weis J, Ring P, Olofsson T, Ortiz-Nieto F, Wikstrom J. Short echo timeMR spectroscopy of brain tumors: grading of cerebral gliomas bycorrelation analysis of normalized spectral amplitudes. J. Magn.Reson. Imaging 2010; 31(1): 39–45.

38. Preul MC, Leblanc R, Caramanos Z, Kasrai R, Narayanan S, Arnold DL.Magnetic resonance spectroscopy guided brain tumor resection:differentiation between recurrent glioma and radiation change intwo diagnostically difficult cases. Can. J. Neurol. Sci./J. Can. Sci. Neurol.1998; 25(1): 13–22.

39. Lai PH, Weng HH, Chen CY, Hsu SS, Ding S, Ko CW, Fu JH, Liang HL,Chen KH. In vivo differentiation of aerobic brain abscesses and ne-crotic glioblastomas multiforme using proton MR spectroscopic im-aging. Am. J. Neuroradiol. 2008; 29(8): 1511–1518.

40. Sankar T, Caramanos Z, Assina R, Villemure JG, Leblanc R, LanglebenA, Arnold DL, Preul MC. Prospective serial proton MR spectroscopicassessment of response to tamoxifen for recurrent malignant glioma.J. Neuro-oncol. 2008; 90(1): 63–76.

41. Howe FA, Barton SJ, Cudlip SA, Stubbs M, Saunders DE, Murphy M,Wilkins P, Opstad KS, Doyle VL, McLean MA, Bell BA, Griffiths JR.Metabolic profiles of human brain tumors using quantitativein vivo 1H magnetic resonance spectroscopy. Magn. Reson. Med.2003; 49(2): 223–232.

42. Barker PB. N-Acetyl aspartate – a neuronal marker? Ann. Neurol.2001; 49(4): 423–424.

43. Tzika AA, Zurakowski D, Poussaint TY, Goumnerova L, AstrakasLG, Barnes PD, Anthony DC, Billett AL, Tarbell NJ, Scott RM, BlackPM. Proton magnetic spectroscopic imaging of the child’s brain:the response of tumors to treatment. Neuroradiology 2001; 43(2): 169–177.

44. Semenza GL. Tumor metabolism: cancer cells give and take lactate. J.Clin. Invest. 2008; 118(12): 3835–3837.

45. Munster PN, Troso-Sandoval T, Rosen N, Rifkind R, Marks PA, RichonVM. The histone deacetylase inhibitor suberoylanilide hydroxamicacid induces differentiation of human breast cancer cells. CancerRes. 2001; 61(23): 8492–8497.

46. Covington HE 3rd, Maze I, LaPlant QC, Vialou VF, Ohnishi YN, BertonO, Fass DM, Renthal W, Rush AJ 3rd, Wu EY, Ghose S, Krishnan V,Russo SJ, Tamminga C, Haggarty SJ, Nestler EJ. Antidepressantactions of histone deacetylase inhibitors. J. Neurosci. 2009; 29(37):11 451–11 460.

47. Baraban JM, Worley PF, Snyder SH. Second messenger systems andpsychoactive drug action: focus on the phosphoinositide systemand lithium. Am. J. Psychiatry 1989; 146(10): 1251–1260.

48. Barkai AI. Dopamine turnover in the intact rabbit brain: effect ofpentobarbital or haloperidol. J. Pharmacol. Exp. Ther. 1978; 205(1): 133–140.

49. Levine J, Rapaport A, Lev L, Bersudsky Y, Kofman O, Belmaker RH,Shapiro J, Agam G. Inositol treatment raises CSF inositol levels. BrainRes. 1993; 627(1): 168–170.

50. Levine J. Controlled trials of inositol in psychiatry. Eur. Neuropsycho-pharmacol. 1997; 7(2): 147–155.

L. WEI ET AL.

wileyonlinelibrary.com/journal/nbm Copyright © 2012 John Wiley & Sons, Ltd. NMR Biomed. (2012)