The Response of RIF-1 Fibrosarcomas to the Vascular-Disrupting Agent ZD6126 Assessed by In Vivo and Ex Vivo ¹H Magnetic Resonance Spectroscopy

Abstract

The response of radiation-induced fibrosarcoma 1 (RIF-1) tumors treated with the vascular-disrupting agent (VDA) ZD6126 was assessed by in vivo and ex vivo ¹H magnetic resonance spectroscopy (MRS) methods. Tumors treated with 200 mg/kg ZD6126 showed a significant reduction in total choline (tCho) in vivo 24 hours after treatment, whereas control tumors showed a significant increase in tCho. This response was investigated further within both ex vivo unprocessed tumor tissues and tumor tissue metabolite extracts. Ex vivo high-resolution magic angle spinning (HRMAS) and ¹H MRS of metabolite extracts revealed a significant reduction in phosphocholine and glycerophosphocholine in biopsies of ZD6126-treated tumors, confirming in vivo tCho response. ZD6126-induced reduction in choline compounds is consistent with a reduction in cell membrane turnover associated with necrosis and cell death following disruption of the tumor vasculature. In vivo tumor tissue water diffusion and lactate measurements showed no significant changes in response to ZD6126. Spin-spin relaxation times (T₂) of water and metabolites also remained unchanged. Noninvasive ¹H MRS measurement of tCho in vivo provides a potential biomarker of tumor response to VDAs in RIF-1 tumors.

Introduction

A functional vascular network is essential for the survival and growth of solid tumors [1,2]. The dependence of tumor cells on a nutritive blood supply has resulted in the tumor vasculature becoming an attractive therapeutic target. Vascular-disrupting agents (VDAs) have been developed to exploit differences between normal and tumor vascular endothelium, with the aim of selectively destroying tumor blood vessels while leaving normal vessels unaffected [3,4]. N-acetylcolchinol-O-phosphate (ZD6126) is a small-molecule VDA that is shown to have significant antitumor activity against rodent and human tumor model systems [5–9] and against human tumor vasculature [10]. It is a water-soluble phosphate prodrug of the tubulin-binding agent N-acetylcolchinol that inhibits tubulin polymerization, disrupting the microtubule cytoskeleton [11]. Following ZD6126 treatment, disruption of the tubulin cytoskeleton in proliferating endothelial cells leads to endothelial cell detachment, tumor vascular shutdown, and, ultimately, massive central hemorrhagic necrosis [5,6]. A surviving rim of tumor cells is characteristically seen histologically following treatment. As a consequence, VDAs characteristically induce tumor growth delay but not tumor regression, as the tumor regrows from this viable rim. Thus, the clinical development of VDAs will be greatly assisted by the use of biomarkers that are sensitive to disrupted tumor perfusion. To this end, noninvasive imaging methods, such as magnetic resonance imaging (MRI) [9,10] and positron emission tomography [12], are being exploited.

Magnetic resonance spectroscopy (MRS) methods provide noninvasive measurements of tumor biochemistry. Localized in vivo ¹H MRS techniques have been developed to quantify the concentration (in mM) of tumor total choline (tCho; an important metabolic marker that is associated with membrane turnover, malignancy, and tumor grade) [13–15] and lactate (another significant tumor metabolite that is associated with glycolysis, acidosis, and hypoxia) [13–17]. Choline, phosphocholine (PC), and glycerophosphocholine (GPC) all contribute to the tCho trimethylamino resonance detected at 3.23 ppm, but they cannot be individually resolved in vivo due to the typically low magnetic field homogeneity achievable in vivo. To understand the individual contributions of choline compounds to in vivo tCho resonance, both high-resolution magic angle spinning (HRMAS) ¹H MRS, from which well-resolved spectra can be obtained from a few milligrams of unprocessed solid tumor samples [18–20], and ¹H MRS of sequential tumor metabolite extractions have been used [21].

Water diffusion within tumors is another property of tumors that is currently being intensively investigated. Apparent water diffusion coefficient (ADC), measured in vivo by ¹H MRS, is sensitive to the biophysical characteristics of tissues, including the fraction of extracellular space. Several studies have shown the potential of ADC as a biomarker of therapeutic response in tumors [22–25].

We have previously shown that 200 mg/kg ZD6126 induces massive hemorrhagic necrosis in RIF-1 tumors 24 hours after treatment [9]. The aims of this study were to investigate the utility of in vivo ¹H MRS to provide biomarkers of response to ZD6126 in RIF-1 tumors, and to provide any validation of these biomarkers with ex vivo HRMAS and ¹H MRS of tumor extracts following treatment.

Methods

Animals and Tumors

All experiments were performed in accordance with the local ethical review panel, the UK Animals (Scientific Procedures) Act 1986, and the United Kingdom Coordinating Committee on Cancer Research guidelines [26]. Murine radiation-induced fibrosarcoma 1 (RIF-1) was grown subcutaneously in the flanks of C3H female mice (n = 12) [27]. Tumor volume was calculated by measuring length (l), width (w), and depth (d) using calipers and the formula: lwd(κ / 6).

Administration of Anesthesia and Dosage of ZD6126

The mice were anesthetized with a 10-ml/kg intraperito-neal (i.p.) injection of fentanyl citrate (0.315 mg/ml) plus fluanisone (10 mg/ml) (Hypnorm; Janssen Pharmaceutical Ltd., Oxford, UK), midazolam (5 mg/ml) (Hypnovel; Roche, Welwyn Garden City, Herts, UK), and water (1:1:2). ZD6126 was obtained from AstraZeneca (Cheshire, UK) and formulated in 20% of 5% sodium carbonate and 80% phosphate-buffered saline, yielding a clear solution at ca. pH 7.

In Vivo ¹H MRS

Anesthetized mice were placed in the bore of a 4.7-T horizontal magnet fitted with a 10-G/cm 12-cm-bore high-performance auxiliary gradient insert interfaced to a Varian Unity Inova spectrometer (Varian Instruments, Palo Alto, CA), so that the tumors hung on a two-turn ¹H radiofrequency coil of 15 mm diameter. Body temperature was maintained with a water-heated pad placed over the mouse. Voxels were selected from scout gradient echo images, and localized shimming yielded linewidths on the order of 20 to 30 Hz. PRESS localization with water suppression was used to detect choline, with a repetition time of 2 seconds, 64 transients, and echo times of 20, 68, 136, 272, and 408 milliseconds. Water suppression included selective excitation of water signal by 14-millisecond Gaussian pulse followed by dephasing gradients in three orthogonal directions in repetition time delay. For unsuppressed water, 16 transients were acquired with the same acquisition parameters as above, except with a lower receiver gain. A multiple quantum coherence (MQC) editing sequence with a single-shot outer volume-saturated localization [28] and a repetition time of 3 seconds was used to detect lactate signal. A localized STEAM sequence with diffusion-sensitizing gradients in echo-time periods was used to separately obtain diffusion-weighted (DW) ¹H MRS data for each orthogonal (x, y, z) direction. The acquisition parameters for DW-MRS were T_E = 24 milliseconds, T_M = 100 milliseconds, T_R = 3 seconds, d = δ milliseconds, and Δ = 112 milliseconds, with diffusion gradients increasing from 0 to 13 G/cm with a 1-G/cm interval. Corresponding b values were 0, 2731, 10,924, 24,579, 43,696, 68,276, 98,317, 133,820, 174,786, 221,214, 273,104, 330,455, 393,269, and 461,545 sec/cm².

After pretreatment ¹H MRS measurements, saline or 200 mg/kg ZD6126 was administered intraperitoneally, and ¹H MRS was repeated 24 hours later. Immediately after posttreatment in vivo MRS measurements, anesthetized mice were sacrificed by cervical dislocation and the tumors were rapidly excised and either frozen intact for HRMAS MRS or freeze-clamped in liquid nitrogen for high-resolution ¹H MRS of tissue extracts.

In Vivo ¹H MRS Data Analysis

MRUI software [29], using the AMARES [30] algorithm in the time domain, was used for data processing and resonance fitting. The method included metabolite lineshape correction and zero-order phasing using a water reference signal [31], removal of residual water signal in the range of 4.5 to 5.0 ppm using Hankel Lanczos Singular Value Decomposition, and direct current offset calculated from the last 100 data points. The first data point in free induction decay was excluded to minimize the contribution from broad signals underlying the whole spectrum, which, when present, can result in an overestimation of metabolite signals. Gaussian lineshapes were assumed for the estimation of metabolite peak areas. Prior knowledge used to constrain fitting parameters included the application of equal linewidths of tCho and total creatine (tCr; creatine plus phosphocreatine) resonances (with the assumption that shimming dominates linewidth) [17], and also that the chemical shifts of metabolite resonances were limited to 0.05 ppm about their expected values of 3.22 and 3.03. The tCho, tCr, and water peak areas were plotted as a function of echo time, and the intercept (M₀) and metabolite T₂ values were determined by assuming a monoexponential decay function using the Levenberg-Marquardt algorithm.

Metabolite concentrations were calculated using tumor water as a reference, assuming tissue water was 80% (88 M protons), according to the equation [13,14]:

$$\left[\text{metabolite}\right]={(\text{metabolite/water})}_T{}_{=}{W}_{\text{c}}(N_{\text{w}}/N_{\text{m}})({\text{AV}}_{\text{n}}{\text{RG}}_{\text{c}})$$

where [metabolite] = metabolite concentration; (metabolite/water)_TE=0 = ratio of metabolite and water intercepts at T_E = 0, as determined from individual T₂ fits; W_c = water concentration (44 M); N_w = number of protons in water (2); N_m= number of protons contributing to metabolite resonance (for tCho, N_m = 9; for tCr, N_m = 3); AV_n = normalization factor for the number of averages; and RG_c = receiver gain correction factor. The corrections for the number of averages and receiver gains were needed because they were different for the acquisition of water and metabolite spectra. Quantitation of lactate was not possible because the signal observed with the MQC editing sequence was acquired at a fixed echo time; thus, an uncorrected lactate/water signal ratio was determined using water intercept values calculated from T₂ curves at T_E = 0.

For analysis of diffusion-weighted data, the plots of normalized water signal (M/M₀) against b values were fitted with a stretched exponential model [M/M₀ = exp{ - (bDDC)^α}], resulting in the estimation of a heterogeneity index (α) and distributed diffusion coefficient (DDC) [32]. The Levenberg-Marquardt algorithm was used for optimization. The average values of DDC and α were estimated by DDC = (DDC_xx + DDC_yy + DDC_zz) / 3 and α= (α_xx + α_yy + α_zz) / 3.

HRMAS ¹H MRS

Frozen tumor tissue (12 ± 3 mg) was transferred to a 4-mm zirconium oxide HRMAS rotor. Deuterium oxide (10 µl) was added for field frequency locking. All HRMAS experiments were performed on a 600-MHz (14.1-T) Bruker AVANCE spectrometer (Bruker Biospin, Coventry, UK) with a spin rate of 4000 ± 2 Hz at 277 K. Spectra were acquired with a CPMG pulse sequence [90 - (τ - 180)_n - τ - acq] as a T₂ filter to remove broad resonances arising from slowly tumbling macromolecules and lipids. The interpulse delay τ was 250 microseconds, and spectra were obtained with 128 transients, 32,000 data points, and 8 kHz spectral width. The 90° pulse was adjusted for individual samples and varied from 13.5 to 14.5 microseconds. T₂ measurements were performed by varying the effective T₂ filter time from 100 to 1000 milliseconds in six increments. Spectral processing included 0.5-Hz apodization, Fourier transformation, and first-order phase correction. A monoexponential decay of metabolite signal with echo time was assumed to estimate equilibrium magnetization (M₀) and T₂ values of metabolites. M₀ was used for quantitation of metabolites. The creatine peak at 3.03 ppm was used as a chemical shift reference [20]. tCr content [tCr = creatine (Cr) + phosphocreatine (PCr)], which was obtained from the total extracted by perchloric acid (PCA), methanol, and chloroform extractions (see later) for each tumor, was used as an internal standard for the quantitation of metabolites by HRMAS [20].

High-Resolution ¹H MRS of Tissue Extracts

Tissue metabolites from frozen tissue were extracted using PCA and methanol/chloroform methods [21]. Frozen tumor tissue was ground to a fine powder using mortar and pestle. One part of tissue plus four parts of 6% PCA were homogenized and centrifuged, and the resulting supernatant was neutralized. After lyophilization, the sample was used for high-resolution ¹H MRS. Methanol and chloroform (2:1; 3 ml/g) were added to the residual pellet (after PCA extraction) and homogenized. The mixture was left on ice for 15 minutes, and then chloroform and distilled water (1:1; 1 ml/g) were added to form an emulsion. Centrifuging the sample resulted in a clear separation of the upper (methanol and water) phase from the lower organic (chloroform) phase, with the pellet in the middle. Both phases were carefully separated and dried at room temperature. All dried samples were then reconstituted in 1 ml of deuterium oxide, except for chloroform-extracted metabolites that were taken up in 1 ml of CDCl₃, and 0.5 ml of this sample was placed ina5-mm standard nuclear magnetic resonance tube. Sodium 3-trimethylsilyl-2,2,3,3-tetradeuteropropionate (50 µl, 5 mM) was added as an external standard for chemical shift calibration and quantitation of metabolites. Immediately before MRS analysis, the pH of the samples was readjusted to 7 with PCA or potassium hydroxide. All high-resolution ¹H MRS data were acquired at 600 MHz (14.1 T) using a Bruker AVANCE spectrometer. ¹H MRS acquisition included presaturation of water using gated irradiation centered on water frequency, a repetition time of 5 seconds, and 128 averages.

Results

Irrespective of treatment, the tumor volume of all RIF-1 tumors significantly increased over the 24-hour period of study (Table 1). Localized PRESS ¹H MRS yielded a well-resolved tCho resonance at 3.23 ppm and a tCr resonance at 3.03 ppm in all RIF-1 tumors in vivo. Representative spectra from one RIF-1 tumor before and after treatment with 200 mg/kg ZD6126 are shown in Figure 1. Lactate signals were also detected in both pretreatment and post-treatment MQC ¹H spectra.

Table 1.

Summary of In Vivo ¹H MRS Data Obtained from RIF-1 Tumors before and 24 Hours after Treatment with Either Saline (n = 6) or 200 mg/kg ZD6126 (n = 6).

Variable	Saline		ZD6126
	Pretreatment	Posttreatment	Pretreatment	Posttreatment
Volume (mm3)	458 ± 65	589 ± 85*	518 ± 52	671 ± 57†
tCho (mM)	4.43 ± 0.20	5.09 ± 0.17*	5.73 ± 0.46	2.40 ± 0.51*
tCr (mM)	4.81 ± 0.38	5.14 ± 0.76	4.56 ± 0.73	3.14 ± 0.57
Water T2 (milliseconds)	53.3 ± 1	52.8 ± 2	51.4 ± 1	46.9 ± 4
tCho T2 (milliseconds)	190 ± 13	157 ± 17	162 ± 14	142 ± 12
tCr T2 (milliseconds)	142 ± 12	133 ± 16	138 ± 12	116 ± 13
Lactate/water (x10-3)	3.79 ± 0.80	3.08 ± 0.95	3.06 ± 0.82	1.54 ± 0.68
DDC(x10-6 cm2/sec)	5.39 ± 0.45	5.17 ± 0.48	4.72 ± 0.28	5.63 ± 1.30
Heterogeneity index (α)	0.77 ± 0.03	0.77 ± 0.01	0.75 ± 0.02	0.75 ± 0.02

Values are mean ± SEM.

^*P < .02, Student's paired t test between pretreatment and posttreatment values.

^†P < .05, Student's paired ttest between pretreatment and posttreatment values.

Figure 1.

In vivo PRESS-localized ¹H MRS spectra (T_E = 136 milliseconds) acquired from one RIF-1 tumor (a) before and (b) 24 hours posttreatment with 200 mg/kg ZD6126.

Individual tCho concentrations for all RIF-1 tumors before and 24 hours after treatment with either saline or 200 mg/kg ZD6126 are shown in Figure 2. All saline-treated tumors showed an increase in tCho, whereas tCho decreased in all ZD6126-treated tumors. The individual tumor tCho concentrations before and after treatment with either saline or ZD6126, plotted as a function of tumor volume, are shown in Figure 3. The in vivo ¹H MRS data obtained from RIF-1 tumors are summarized in Table 1. A significant (P = .009) reduction in the tCho of RIF-1 tumors was determined 24 hours after treatment with ZD6126, whereas tCho significantly increased over 24 hours in saline-treated control tumors. There were no significant changes in water T₂, tCho T₂, tCr T₂, and lactate/water ratio in either control or treated cohorts. In vivo water DDC and the heterogeneity index α showed no change in either saline-treated or ZD6126-treated cohorts.

Figure 2.

Line series of tCho concentration (mM) determined by in vivo ¹H MRS for each RIF-1 tumor (a) before and 24 hours after saline, and (b) before and 24 hours after 200 mg/kg ZD6126, i.p.

Figure 3.

tCho concentration plotted as a function of tumor volume for each individual RIF-1 tumor investigated by in vivo ¹H MRS.

The metabolite concentrations determined from unprocessed tumor tissue by HRMAS ¹H MRS and from high-resolution ¹H MRS of tissue extracts are summarized in Table 2. Both methods revealed a significant reduction in PC and GPC, but no change in free choline concentrations, in ZD6126-treated RIF-1 tumors. The sum of free choline, PC, and GPC (the metabolites that contribute to tCho resonance detected in vivo) was calculated for each MRS method for comparison of in vivo and ex vivo tCho data. A highly significant reduction in tCho was found in ZD6126-treated unprocessed tumor tissue by HRMAS, but not in tissue extracts determined by high-resolution ¹H MRS. Both ex vivo MRS methods revealed a significant reduction in tCr and taurine following treatment with ZD6126. Alanine was also found to be significantly reduced, as determined by HRMAS ¹H MRS only. There were no significant changes in the concentration of lactate, glycine, or acetate.

Table 2.

Tumor Metabolite Concentrations (µmol/kg) Determined by ¹H HRMAS of Unprocessed Tissues and by High-Resolution ¹H MRS of Tissue Extracts of RIF-1 Tumors 24 Hours after Treatment with Saline (n = 6) or 200 mg/kg ZD6126 (n = 6).

Metabolite	HRMAS		Extracts
	Saline	ZD6126	Saline	ZD6126
Choline	1.94 ± 0.3	1.11 ± 0.4	3.23 ± 0.8	2.96 ± 0.8
PC	2.34 ± 0.3	0.65 ± 0.3*	1.03 ± 0.2	0.36 ± 0.03*
GPC	1.80 ± 0.2	0.52 ± 0.1*	1.34 ± 0.2	0.74 ± 0.11†
tCho	6.08 ± 0.8	2.28 ± 0.5*	5.60 ± 0.7	4.07 ± 0.7
PC/GPC	1.32 ± 0.07	1.32 ± 0.42	0.82 ± 0.14	0.58 ± 0.15
Lactate	34.2 ± 11	14.02 ± 2.5	6.24 ± 0.92	7.70 ± 0.73
Alanine	4.34 ± 0.5	2.15 ± 0.46*	2.77 ± 0.3	2.17 ± 0.3
tCr	2.94 ± 0.3	1.55 ± 0.4*	2.94 ± 0.3	1.55 ± 0.4*
Taurine	16 ± 1.8	4.7 ± 1.4*	7.8 ± 0.3	3.07 ± 0.98*
Glycine	3.48 ± 0.31	2.53 ± 0.7	2.35 ± 0.18	1.51 ± 0.41
Acetate	0.73 ± 0.09	0.55 ± 0.15	0.94 ± 0.17	1.21 ± 0.10

Values are mean ± SEM.

^*P < .02, Student's t test.

^†P < .05, Student's t test.

A comparison of the tCho concentrations of RIF-1 tumors determined 24 hours after saline or ZD6126 by in vivo ¹H MRS, HRMAS ¹H MRS, and high-resolution ¹H MRS of tissue extracts is shown in Figure 4. The ex vivo data measured in micromoles per gram per wet weight were converted to millimolars by dividing by a factor of 0.80 (mM = µmol/g per wet weight per 0.8) to allow for a comparison with the in vivo ¹H MRS data measured in millimolars [18,33]. Both in vivo and HRMAS ¹H MRS, but not high-resolution ¹H MRS of tumor extracts, revealed a significant reduction in tCho concentration in ZD6126-treated tumors. Figure 5 summarizes the tCho/tCr ratios for saline-treated and ZD6126-treated tumors, as determined by in vivo ¹H MRS, HRMAS ¹H MRS, and high-resolution ¹H MRS of tissue extracts. There was no significant difference in the tCho/tCr ratio in response to ZD6126 measured by the three different ¹H MRS methods.

Figure 4.

tCho concentrations (mM) determined by in vivo ¹H MRS of unprocessed tissues, by HRMAS ¹H MRS of extracts, and by HR ¹H MRS of tissue extracts of RIF-1 tumors 24 hours after treatment with saline or 200 mg/kg ZD6126. tCho determined in vivo and by HRMAS was significantly reduced (**P < .02).

Figure 5.

tCho/tCr ratios of RIF-1 tumors determined by in vivo ¹H MRS, HRMAS ¹H MRS, and high-resolution ¹H MRS of tissue extracts 24 hours after treatment with either saline or 200 mg/kg ZD6126.

Discussion

The aim of this study was to investigate the use of in vivo ¹H MRS to assess the metabolic status of RIF-1 tumors before and 24 hours after treatment with 200 mg/kg ZD6126, a dose and timing regime previously shown to induce massive hemorrhagic tumor necrosis in this murine model [9]. Validation of in vivo response was sought using HRMAS ¹H MRS and high-resolution ¹H MRS of tissue extracts. In addition, the biophysical properties of tumor water T₂, diffusion, and their response to ZD6126 were also investigated.

Metabolite Response to ZD6126

A common observation in MRS studies of cancer is elevation in the concentration of choline metabolites, although the precise reasons for this are still unclear [34]. It has been traditionally attributed to an increased biosynthesis of membranes by rapidly proliferating tumor cells [35], and the significant increase in the tCho concentration of control RIF-1 tumors over 24 hours reported herein is consistent with this. Recent evidence suggests that, in addition to biosynthetic pathways, mitogen-induced and oncogene-induced activation of phosphatidylcholine and phosphatidylethanolamine metabolism can increase the accumulation of choline metabolites in tumor cells [36]. Upregulation of choline kinase, which catalyzes the phosphorylation of choline to yield PC, may also contribute [37].

The significant decrease in tCho concentration of RIF-1 tumors in vivo 24 hours after treatment with ZD6126, which was observed despite continued tumor growth (Figure 3), is consistent with reduced cell membrane turnover associated with necrosis following disruption of the tumor vasculature. This in vivo response biomarker was further validated by a significant ZD6126-induced decrease in both PC and GPC (key metabolites associated with membrane biosynthesis that contribute to tCho resonance detected in vivo), as measured by HRMAS ¹H MRS and high-resolution ¹H MRS of tissue extracts. Quantitative ¹H MRS revealed a similar response pattern of HT29 tumors after treatment with the VDA 5,6-dimethylxanthenone acetic acid (DMXAA) [13]. A significant reduction in tCho concentration was found 24 hours after treatment with 21 mg/kg DMXAA. Metabolite ratios, derived from in vivo ¹H MRS, showed a significant reduction in tCho/water of C3H mammary carcinomas 24 hours after treatment with 250 mg/kg combretastatin A4 phosphate (CA4P), another VDA [38,39].

The difficulty of acquiring quantitative metabolite concentrations using volume-selective in vivo ¹H MRS has resulted in metabolite ratios being reported as surrogate markers of tumor metabolism and response. Intriguingly, the tCho/tCr ratio determined in vivo by HRMAS ¹H MRS and high-resolution ¹H MRS of tumor extracts showed no significant change following treatment with ZD6126. One possible explanation is that ZD6126-induced necrosis and cell death reduce cell density, resulting in the reduction of both tCho and tCr concentrations (hence, no change in their ratio). This highlights the need for caution when using metabolite ratios as biomarkers of response to therapy.

We initially hypothesized that the tumor lactate/water ratio would increase as a result of hypoxia/acidosis induced by ZD6126-induced vascular shutdown. However, there was no significant change in this ratio—a finding similar to that reported for C3H mammary carcinomas 24 hours after treatment with 250 mg/kg CA4P [38]. Experiments using ¹⁴C-labeled ZD6126 have shown that the VDA itself is cleared from tumors within 24 hours of administration (A. J. Ryan, personal communication). Lactate is a molecule smaller than ZD6126 and is readily extruded from tumor cells [40]; thus, it is reasonable to assume that any lactate produced following treatment with ZD6126 is largely cleared from the tumor within 24 hours, hence the absence of any significant changes in the tumor lactate/water ratio.

Both HRMAS ¹H MRS and high-resolution ¹H MRS of tissue extracts showed a significant reduction in the taurine concentration of ZD6126-treated RIF-1 tumors. Taurine is an important organic osmolyte in mammalian cells and is abundant in skeletal muscles [41]. A recent study showed that ZD6126 treatment reduces tumor interstitial fluid pressure [42]; thus, reduction in taurine may reflect a change in pressure gradients and reduced fluid movement through the tumor interstitium associated with the induction of cell death and necrosis.

Tumor Biophysical Response to ZD6126

The majority of anticancer chemotherapeutic drugs kill cells by inducing either apoptosis or necrosis. In either process, the tissue microenvironment is altered by a reduction in intracellular space and a subsequent increase in extracellular space. Quantitation of the spin-spin relaxation time T₂ of water, tCho, and tCr revealed no significant change irrespective of treatment determined in vivo (Table 1) or by HRMAS ¹H MRS (data not shown). A small (ca. 20%) yet significant increase in the T₂ of water in rat GH3 prolactinomas, as quantified by MRI 24 hours after treatment with ZD6126, was recently reported and attributed to edema [43]. No significant change in T₂-weighted image intensity of C3H mammary carcinomas was reported 24 hours post-treatment with CA4P [38]. We also hypothesized that ZD6126-induced hemorrhagic necrosis would result in an increase in extracellular space and, thus, in water diffusion biomarkers. Instead, we found no significant changes in DDC or tumor heterogeneity index (α) in vivo, suggesting no significant changes in the biophysical characteristics of the tumor tissue 24 hours after treatment. A similar observation was recently reported in murine RIF-1 tumors treated with 200 mg/kg ZD6126, in which there was no significant change in the apparent diffusion coefficient (ADC) of tumors 24 hours posttreatment, but subsequent significant increases in ADC at 48 and 72 hours posttreatment [44]. A significant increase in the ADC of rat rhabdomyosarcomas 48 hours after treatment with CA4P has also been reported [25]. Together, these data suggest that any VDA-induced changes in the biophysical structure of the tumor tissue occur relatively slowly and are preceded by changes in tumor metabolism.

Utility of MRS Biomarkers of Tumor Response

As a consequence of their inability to induce tumor regressions, the development of VDAs has necessitated the identification of sensitive biomarkers of tumor vascular disruption. In this regard, ¹H MRI methods, with their high spatial and temporal resolutions, are attractive, and the biomarkers K^trans (the transfer constant describing the transendothelial transport of a low-molecular-weight contrast agent into the extravascular extracellular space) and IAUGC (the initial area under the gadolinium concentration-time curve), derived from dynamic contrast-enhanced MRI, are particularly suited for VDA evaluation in both preclinical and human tumors [45]. Preclinical in vivo MRS methods have also been applied to assess tumor response to VDAs. Both ³¹P MRS and ¹H MRS have shown dramatic reductions in tumor energetics and cell membrane proliferation in response to VDAs, and have highlighted the potential of the βNTP/Pi ratio and tCho as early biomarkers of tumor response to this class of drug [13,38,39,46,47]. Similar MRS methods have also been used for the preclinical assessment of other novel targeted cancer therapeutics, and it is clear from these studies that the determination of tCho by ¹H MRS in vivo also provides a robust biomarker of tumor response [14,48], in contrast to those derived by ³¹P MRS where evidence of objective tumor response is unclear [14,49]. Recently, ¹H MRS has been successfully used in the clinic to quantify tCho concentrations of locally advanced breast cancer before and after neoadjuvant chemotherapy to predict response [15]. Significant reductions in tCho were reported post-treatment, further highlighting the potential utility of tCho as a biomarker of tumor response.

In conclusion, we have shown that quantitation of tCho by ¹H MRS in vivo is a sensitive biomarker of both the progression and the response of RIF-1 tumors to the VDA ZD6126. Tumor tCho concentration was significantly reduced 24 hours after treatment with 200 mg/kg ZD6126 and appeared to precede any changes in the biophysical properties of the tissue. HRMAS ¹H MRS and high-resolution ¹H MRS of tissue extracts showed significant reductions in the concentration of PC and GPC (constituents of the tCho resonance detected in vivo), thereby validating tCho response. Quantitation of tCho (mM) by in vivo ¹H MRS may provide a noninvasive biomarker of early tumor response to VDAs.

Abbreviations

ADC

apparent water diffusion coefficient

DDC

distributed diffusion coefficient

GPC

glycerophosphocholine

HRMAS

high-resolution magic angle spinning

i.p.

intraperitoneal

MRS

magnetic resonance spectroscopy

PC

phosphocholine

tCho

total choline

tCr

total creatine

VDA

vascular-disrupting agent