Early Response of Prostate Carcinoma Xenografts to Docetaxel Chemotherapy Monitored With Diffusion MRI

Abstract

For many anticancer therapies, it would be desirable to accurately monitor and quantify tumor response early in the treatment regimen. This would allow oncologists to continue effective therapies or discontinue ineffective therapies early in the course of treatment, and hence, reduce morbidity. This is especially true for second-line therapies, which have reduced response rates and increased toxicities. Previous works by others and ourselves have shown that water mobility, measured by diffusion-weighted magnetic resonance imaging (DW-MRI), increases early in tumors destined to respond to therapies. In the current communication, we further characterize the utility of DW-MRI to predict response of prostate cancer xenografts to docetaxel in SCID mice in a preclinical setting. The current data illustrate that tumor volumes and secreted prostate-specific antigen both respond strongly to docetaxel in a dose-responsive manner, and the apparent diffusion coefficient of water (ADC_w) increases significantly by 2 days even at the lowest doses (10 mg/kg). The ADCw data were parsed by histogram analyses. Our results indicate that DW-MRI can be used for early detection of prostate carcinoma xenograft response to docetaxel chemotherapy.

Introduction

Prostate cancer is the most commonly diagnosed cancer and the second leading cause of cancer death in American men. An estimated 39,200 men die of prostate cancer in the US annually [1]. In nearly all patients, growth and progression of cancer occur despite initial androgen ablation therapy [2]. Although the occurrence of prostate cancer is rare before age 40, the prevalence of the disease quickly increases to 80% by the age of 80. In 1941, Huggins et al. [3] showed that bilateral orchiectomy induced acute biochemical and clinical improvement in 80% of men with metastatic prostate cancer. However, this response to androgen ablation therapy was temporary, with a median response duration of 12 to 16 months [4,5]. With increasing life expectancy, hormone-refractory prostate cancer (HRPC) represents the most common form of male cancer in the US and other Western countries [6]. The population of early “geriatric” HRPC patients is rapidly increasing, posing an even greater challenge to oncologists treating this difficult-to-manage disease.

The evolution of early prostate cancer is unpredictable and may extend over many years. Some tumors progress rapidly and become fatal within a few years. Other tumors may advance slowly, or not at all. A commonly held opinion is that chemotherapy plays a limited role in the treatment of prostate cancer because no single agent or combination of agents has been shown to prolong survival in randomized trials. However, this point of view may be changing. Preliminary findings of in vitro experiments have challenged traditional beliefs that prostate cancers are resistant to chemotherapy [7]. Recent clinical trials using taxanes as monotherapy have provided promising results [8,9]. In the treatment of prostate cancer, various interventions that are rapidly evolving combine chemotherapy with traditional therapies (e.g., hormonal and radiation), or use chemotherapy in conjunction with a variety of investigational approaches [10].

Docetaxel belongs to the taxoid family, a class of antineoplastic drugs useful in the treatment of several types of cancers such as breast, ovarian, and non-small cell lung cancer. Docetaxel was derived from the convergent synthesis of paclitaxel, which has been effective in treating breast cancer for more than 4 years. Docetaxel arrests tumor growth via cytostatic and cytotoxic mechanisms. It works by stabilizing microtubules, preventing mitosis, and ultimately inducing apoptosis or necrosis [11–14]. Both paclitaxel and docetaxel bind to specific sites on tubulin subunits to stabilize the polymerized filaments, although docetaxel binds to tubulin with twice the affinity of paclitaxel [15].

Once a treatment regimen has been designed, close monitoring of tumor response to therapy becomes essential. Currently, prostate tumor response to therapy is monitored by frank radiographic changes in tumor morphology and serum levels of prostate-specific antigen (PSA). PSA is an excreted protease that is expressed both by normal and malignant prostatic epithelium. Normal basalar levels are about 4 ng/ml, and in patients with prostate carcinoma, serum PSA levels are generally higher, about 10 ng/ml [16]. Although most studies valuate the cutoff for normal serum PSA levels at about 4 ng/ml, 20% of cancers are not detected (false-negative) and 65% of patients with benign disease are made to undergo further testing (false-positive), making the positive predictive value of a PSA >4.0 ng/ml only about 25% [17,18]. It appears that this low sensitivity/specificity is due to the inability of current immunotechniques to distinguish between early prostate cancer and benign prostatic hyperplasia. Recently, studies have increased the specificity and sensitivity values (90% and 50%, respectively) of PSA testing simply by determining the free PSA (as compared to total PSA or PSA density) in the blood at a cutoff of 30% [19]. In general, changes in tumor morphology and serum PSA are clinically useful, but changes in these parameters may take weeks or months to become evident.

Several magnetic resonance imaging (MRI) modalities have been employed for the early detection of therapy response in various cancer models. Clinical trials have evinced the utility of T2-weighted imaging to monitor therapeutic response in a pancreatic tumor model [20], and T1(rho) imaging (in conjunction with other modalities) has been used to assess treatment planning in brain tumors [21]. Dynamic contrast-enhanced (DCE) MRI measures the rate of enhancement on a pixel-by-pixel basis using T1-weighted imaging following a bolus injection of gadolinium-based contrast agents, and DCE-MRI has been used both clinically and preclinically to monitor response to cytotoxic and cytostatic therapies [22]. Additionally, ¹H magnetic resonance spectroscopy has been used to detect changes in tumor lactate levels as an early indicator of radiosensitive RIF-1 tumor response to radiation therapy [23].

A novel indicator of tumor response to therapy is the apparent diffusion coefficient of water (ADC_w) as measured by diffusion-weighted (DW) MRI. In preclinical models, DW-MRI can detect tumor response to chemotherapy quantitatively, sensitively, and early in the treatment regimen [24–31]. Although the exact mechanisms underlying changes in ADC_w following therapy are unknown, it is hypothesized that the increased ADC_w is reflecting a decreased intracellular volume fraction caused by apoptosis-associated cell shrinkage, necrosis, or vasogenic edema. Because water is not as diffusionally restricted in the extracellular space, compared to the intracellular, a decrease in cell volume fraction will result in an overall increase in the ADC_w [28]. In the current study, we investigated the response of the tumor ADC_w to a novel drug, docetaxel, in a novel tumor type, LnCaP. Docetaxel was chosen because it is cytostatic at low doses and cytotoxic at higher doses (unpublished observations). The LnCaP cell line was chosen because it secretes PSA. Serum PSA levels and tumor volumes were used to noninvasively measure tumor response. Although there are additional and more specific markers of prostate cancer, such as DD3 and membrane-associated PSA [32], serum PSA was used in this case because repeated and noninvasive measurements could be performed on the same mouse.

Materials and Methods

Cell Lines and Tumors

LnCaP, a tumorigenic cell line originally derived from a lymph node metastasis of a human prostatic adenocarcinoma, was obtained from the American Type Culture Collection (CRL 1740; Rockville, MD). Passages were made twice weekly with a 1:2 split and cultured in DMEM:F12 supplemented with 10% FBS (HyClone, Ft. Collins, CO). For inoculation, approximately 4x10⁶ cells in 2.6 ml of media were suspended in 2.4 ml of Matrigel (Becton Dickinson Labware, Bedford, MA) and injected subcutaneously into the right flanks of SCID mice. SCID mice are severely deficient in T and B cells and fail to reject allogeneic grafts or produce antibodies to common antigens, including PSA. Several human tumor xenografts have been grown in SCID mice and effectively treated with chemotherapy [33,34]. Male SCID mice of ages 5 to 6 weeks were obtained from the Arizona Cancer Center Experimental Mouse Shared Services.

Measurement of Tumor Response

Most mice developed palpable tumors within 2 to 3 weeks of inoculation. At the time of chemotherapeutic treatment and imaging, tumor volumes ranged from 75 to 975 mm³. Tumor volumes were measured using standard calipers and calculated as (lengthxwidth²)/2, with the length and width defined as the long and short axes, respectively. Blood samples of 100 to 200 µL were obtained from each mouse from the retro-orbital plexus using capillary tubes: samples were collected twice monthly until tumor volume was measurable, once per week thereafter, and at 0, 2, 4, 6, 8, and 15 days following commencement of treatment. The whole blood samples were stored at -80°C pending analysis. The samples were analyzed via enzyme-linked immunosorbent assay (ELISA; C. Hamlin, CWRU, Cleveland, OH) to determine serum PSA levels.

MRI

MRI experiments were initiated on a mouse when the tumor size was between 220 and 800 mm³. The chemotherapy was fractionated and synchronized with imaging as follows. Both imaging and therapy commenced with half the cumulative dose on day 0. Mice were imaged again on days 2 and 4. On days 4 and 6, mice were treated with one fourth the cumulative dose of chemotherapy. All MRI experiments were carried out on a 40-cm horizontal bore Bruker Biospec Avance 4.7-T imaging spectrometer (Bruker, Karlsuhe, Germany) equipped with an actively shielded gradient coil capable of 150 mT/m. Mice were anesthesized with 1.5% isoflurane delivered in O₂, at 1.0 l/min, and placed in a homebuilt probe that allowed positioning of the tumor within a 1-cm-diameter, two-turn solenoid coil. Animal temperatures were continuously monitored using a rectal Luxtron fluoroptic thermometer (Luxtron, Santa Clara, CA). A recirculating water-filled heating pad was used to maintain core body temperature between 36°C and 37°C during the course of the imaging experiments.

Sagittal scout images were obtained to determine the position and extent of tumors. DW-MRI was then carried out using 2-mm-thick coronal slices covering the entire tumor volume. A diffusion-weighted radial acquisition of data (DIFRAD) method was employed [35]. Typical acquisition parameters were: TR=1300 to 2000 ms (depending on the number of slices), TE=50.00 ms, FOV=3.84x3.84 cm, matrix size=128x128, slice thickness=2.00 mm (contiguous), Δ=20.00 ms, ∂=7.00 ms, where ∂ and Δ represent duration and separation of diffusion gradients, respectively. Three orthogonal diffusion gradient directions were employed. At each slice location, images were obtained at three b values [b=γ²G_d²∂²(Δ-∂/3)], where G_d is the strength of the diffusion weighting gradient and γ is the gyromagnetic ratio for protons (42.58 MHz/T).Images were reconstructed from the radial data using a magnitude filtered back projection algorithm, which minimizes artifacts due to motion [35]. Apparent diffusion coefficient (ADC_w) maps were generated by fitting the signal intensity of each pixel to a single exponential decay:

$$S=S_0{e}^{-bADCw}$$

where S₀ is the signal intensity in the absence of diffusion weighting, and S is the signal intensity with diffusion weighting. ADC_w maps were analyzed using programs written in Interactive Data Language (Research Systems, Boulder, CO). Regions of interest (ROIs) corresponding to tumor were drawn on the ADC_w maps, and ADC_w distribution histograms were obtained for each tumor from each imaging session (days 0, 2 and 4). The same person defined all ROIs in order to minimize interoperator error, and hence, maximize precision.

Chemotherapy

Docetaxel was obtained from spent clinical doses, stored at 25°C for up to 2 weeks prior to use. Dosage limits were determined in control non-tumor-bearing SCID mice treated with single doses of 40, 60, 80, and 100 mg/kg docetaxel. The highest dosages, 80 and 100 mg/kg, were lethal. Hence, the maximum tolerated single dose was established at 60 mg/kg. Total doses administered ranged from 0 to 60 mg/kg, delivered in three fractions of 0.5, 0.25, and 0.25 total dose on days 0, 4, and 6, respectively. This dosing scheme was chosen to avoid redosing animals during the imaging schedule, which occurred on days 0, 2 and 4.

Results

Tumor Response

Independent measurements of tumor volume and PSA levels were used to characterize tumor response following chemotherapy. A conventional indicator of response, tumor growth delay (TGD), measures the time it takes for tumors to regrow to pretreatment volumes following chemotherapy [36]. Figure 1a shows an example of a typical tumor volume plot along with calculation of TGD. Changes in serum PSA levels for individual mice were plotted against time, as shown in Figure 1b, demonstrating that serum PSA levels provided an adequate biomarker for the progression and response of prostate tumors. Figure 2a illustrates the strong correlation between the PSA levels and tumor volumes before treatment (r²=0.90). Figure 2b demonstrates that within the range of 6 to 24 days following commencement of therapy, this correlation was reduced, and levels of PSA no longer correlated with caliper-measured tumor volumes. These data can be interpreted to indicate that, following therapy, there is a reduction in PSA production without a concomitant decrease in tumor volumes.

Figure 1.

Tumor volumes and serum PSA levels following docetaxel therapy. (a) Tumor volumes were measured by palpation using calipers (the description of calculations can be found in Materials and Methods section). The graph is an example of an individual tumor volume plot, which follows the progression, as well as regression, of the tumor over time. Line (1) shows the least squares fit of pretreatment volume data points, and line (2) shows the least squares fit of volume data points during regrowth. The x-axis projections of the linear regressions during initial growth and regrowth were used to determine the TGD. (Arrows) Represent days 32, 36, and 38 on which animals received treatment with docetaxel of the indicated doses in milligrams per kilogram. (b) This representative plot follows the serum PSA levels over time. Blood serum samples were collected from the orbital plexus at each recorded data point, and levels of the serum PSA were analyzed by ELISA (Materials and Methods section). (Arrows) Represent the days 32, 36, and 38 on which animals received treatment with docetaxel of the indicated doses in units of milligrams per kilogram. Note the rapid decline in levels of PSA following the initial dose, providing illustration that PSA levels are a more sensitive indication of change than tumor volume measurements in (a) of same mouse.

Figure 2.

Correlations between tumor volume changes and serum PSA levels pre- and post-therapy. (a) Illustrates a significant correlation between caliper-measured tumor volume data and serum PSA levels before treatment (n=6). (b) Demonstrates that following chemotherapy, the correlation between the tumor volume and PSA levels is reduced (n=5).

Dose Response to Docetaxel

Figure 3a shows a strong dose response of the average TGD. In some cases, TGD information was not available. Several mice expired after imaging, but before tumor regrowth occurred. Consequently, the percent tumor volume remaining by day 25 was used as an alternative measure of response. As shown in Figure 3b, the ratio of the average tumor volume remaining on day 25 relative to day 0 (i.e., 25 days posttreatment) decreased as the dose increased. The average tumor volume remaining at 0, 10, 30, and 60 mg/kg docetaxel was 1.0, 0.84 (P=.08), 0.54 (P<.05), and 0.22 (P≪0.001), respectively. Because PSA production failed to recover in all tumors, the percent level of PSA remaining by day 18 relative to day 0 was also used as a measure of response. At 60 mg/kg docetaxel, serum PSA levels dropped to 24.5±8.1% of the levels recorded on day 0 (P<.05). In untreated mice, average PSA levels increased by greater than 20-fold (P<.001) over this time (18d).

Figure 3.

Demonstration of the changes in parameters following treatment. (a) Illustrates the average TGD calculated for mice at corresponding treatments with docetaxel. (b) Illustrates the ratio of the average tumor volume remaining on day 25 relative to day 0 (i.e., 25 days post-treatment) at corresponding doses of docetaxel.

Diffusion Response

Preliminary diffusion tensor imaging showed no appreciable diffusion anisotropy, and it was reasonably assumed that diffusion within the LnCaP tumor model is isotropic (data not shown). Nonetheless, three diffusion gradient directions were used to acquire data, and averaged to obtain isotropic diffusion coefficients. DW-MRI were obtained at three b values of 200, 400, and 800 sec/mm², as shown in Figure 4a – c. ADC_w maps were generated from these images as described above and shown in Figures 4d and 5. Reproducibility of these ADC_w measurements was determined by the acquisition of repeated data sets from the same animal in the same session, wherein the animal was removed from the magnet and repositioned between acquisitions. From these measurements, we determined that the reproducibility of mean ADC_w values were ±0.88%, ±4.28%, and ±2.7% on days 0, 2, and 4, respectively. Hence, changes in the ADC_w greater than 5.16% and 3.58% on days 2 and 4 (relative to day 0) are significant (i.e., greater than 2 SD). Figure 5 also illustrates the effect of docetaxel on the ADC_w of a responding tumor. As treatment is initiated, the majority of pixels in an ADC_w map have low ADC_w values (approximately 0.2 to 0.6x10^-3 mm²/sec, shown in green). As treatment progresses, there is a shift to higher ADC_w values (0.8 to 1.7x10^-3 mm²/sec, shown in red and yellow). During imaging, variable tumor volumes necessitated a variable number of acquisition slices (usually four to six slices). As mentioned in Materials and Methods section, raw data acquired from the ADC_w map of each tumor volume slice were analyzed and assembled into histograms that plotted the pixel frequency at different ADC_w values. Histograms from each slice were compiled to generate one inclusive histogram, which represented the total tumor volume. Example histograms before and after treatment are shown in Figure 6b. In addition, there were significant linear correlations between the total number of pixels in each total tumor volume histogram and the caliper measurements of their corresponding tumor volumes before and after treatment (r²=0.85 and 0.83, respectively).

Figure 4.

Calculation of the ADC_w from a series of diffusion-weighted images. Four diffusion-weighted images were obtained at different b values: (a) b=200, (b) b=400, and (c) b=800 sec/mm². For illustrative purposes, a ROI of 25 pixels was defined on each diffusion-weighted image, and the mean signal intensity of this region for each b value is fitted to the single exponential decay shown in the graph. A similar fit was performed on a pixel-by-pixel basis to generate the ADC_w map shown in (d).

Figure 5.

Tumor ADC_w maps obtained from mouse treated at 60 mg/kg docetaxel. Images were acquired and processed as described in text. The ADC_w maps were obtained on different days, and although these data were approximately of the same imaging slice, registration of these images is not implied. Reproducibility was ±0.88%, ±4.28%, and ±2.72% on days 0, 2, and 4, respectively.

Figure 6.

ADC_w histogram plot. From the diffusion maps, ADC_w histograms were generated by plotting the pixel density against a corresponding ADC_w value. From these histograms, the mean and median values were calculated. It was expected that a right shift in the ADC_w mean value (as indicated by the solid line) would be greater at post-treatment, concomitant with a greater diffusion within the tumor following induction of cell death. The above histogram demonstrates a typical right shift in the ADC_w value of total tumor volume following chemotherapy.

Changes in the mean ADC_w values following treatment are shown in Figure 7. The ratio of the mean ADC_w measured on days 2 and 4 with the mean ADC_w measured before treatment (day 0) is plotted versus dose. For animals receiving no treatment (0 mg/kg), the ADC_w decreased by day 2, with a greater decrease by day 4, indicating that a decrease in ADC_w accompanies stable tumor growth (P<.1). However, no significant correlation was observed between tumor size and ADC_w values (data not shown). For doses of 10 to 60 mg/kg docetaxel, the mean ADC_w increased on days 2 and 4 (P=0.03 and P<0.01, respectively). From these data, it is concluded that an early increase in the mean ADC_w was indicative of tumor response to chemotherapy.

Figure 7.

Demonstration of changes in ADC_w following treatment. The ratio of the mean ADC_w values on day 2 (dashed line) and day 4 (solid line) relative to day 0 is shown at specified doses. The 95% confidence belts are shown as dotted lines. Hence, the ADC_w of all untreated controls decreases significantly on days 2 and 4, whereas treated animals show significant increases only on day 4. The ADC_w values for all treated animals were significantly higher than untreated controls on both days 2 and 4.

Discussion

These data illustrate that docetaxel is effective in treating xenografts of prostate adenocarcinoma, and that the changes in the ADC_w as measured by DW-MRI are an earlier indicator of response than tumor volume or serum PSA level. Of the mice enrolled in this study, only one animal was nonresponsive to the docetaxel chemotherapy. Significant growth delays and reductions in serum PSA were observed at all doses >10 mg/kg. Serum PSA levels also declined early in the treatment regimen indicating that, in this model, the decline in serum PSA may be used as an early marker for therapeutic response. Because LnCaP is one of few PSA-secreting prostate cancer cell lines available, this phenomenon is unique to this tumor xenograft system. In humans, PSA is secreted by stromal, as well as tumor, cells. Hence, the use of PSA as an early response indicator in a clinical setting is not robust (vide supra). Another interesting observation was the discordance between the serum PSA levels and tumor volumes posttherapy, as shown in Figure 2b. These data can be interpreted to indicate that, in response to therapy, PSA production, which presumably requires healthy and well-energized cells, is a more sensitive indicator of response than is tumor volume. Moreover, the relative lack of reduction in tumor volume might be due to lack of macrophage clearance of cell debris. In other systems, it has been shown that tumor volume measurements systematically underestimate cell kill and that this underestimation may be due to an accumulation of dead cells within the residual tumor volume following treatment [37]. It has also been hypothesized that such disparities between cell kill and tumor volume estimations may be linked to cellular swelling or extracellular edema [38]. Because tumors were not imaged by DW-MRI at later time points, it is unclear whether there are large edematous or necrotic regions that occupy volume but do not produce PSA.

Increases in the ADC_w were observed at all doses of docetaxel. According to the current treatment protocol, animals receiving docetaxel doses of 10 mg/kg actually received 5 mg/kg, or half the total dose, prior to imaging. At these doses, there are significant increases in the mean ADC_w within 2 days following commencement of therapy. This same group had less significant decreases in PSA and tumor volumes by 18 to 25 days posttreatment. This suggests that, although the changes in ADC_w of tumors are presumably reporting changes in tumor physiology and microenvironment, these changes may not be intimately coupled to therapeutic response. A current hypothesis states that these changes are reporting damage to the tumor physiology from which the cells may recover. Hence, these changes may be used to optimize timing of fractionated chemotherapy such that subsequent doses may be applied when the tumor is most vulnerable (i.e., at its highest ADC_w value).

In a clinical setting, the utility of an early response indicator is limited to its strength in predicting the response of individual tumors in individual patients. Clinical trials are ongoing to assess the applicability of ADC_w as an early response predictor in brain tumors [39,40], metastatic lymphomas (G. Gonzales, Massachusetts General Hospital, personal communication), and metastatic breast cancer lesions in the liver [unpublished observation]. In the current project, the lack of correlation between the TGD and PSA data precludes testing of the hypothesis that early ADC_w changes can predict response in individual tumors (Gillies et al., in preparation).

Abbreviations

ADCw

the apparent diffusion coefficient of water measured by MRI

DW-MRI

diffusion-weighted magnetic resonance imaging

HPRC

hormone-refractory prostate cancer

PSA

prostate-specific antigen

TGD

tumor growth delay