One of the greatest challenges in developing therapeutic regimens is the inability to rapidly and objectively assess tumor response due to treatment. Moreover, tumor response to therapeutic intervention in many cases is transient, and progressive alterations within the tumor may mask the effectiveness of an initially successful therapy. The ability to detect these changes as they occur would allow timely initiation of alternative approaches, maximizing therapeutic outcome. We investigated the ability of diffusion magnetic resonance imaging (MRI) to provide a sensitive measure of tumor response throughout the course of treatment, possibly identifying changes in sensitivity to the therapy. Orthotopic 9L gliomas were subjected to two separate therapeutic regimens, with one group receiving a single 5-day cycle (1ω) of low-dose 1,3-bis(2-chloroethyl)-1-nitrosourea (BCNU) and a second group receiving two cycles at the same dose, bisected with 2 days of rest (2ω). Apparent diffusion coefficient maps were acquired before and throughout treatment to observe changes in water mobility, and these observations were correlated to standard measures of therapeutic response and outcome. Our results showed that diffusion MRI was indeed able to detect the emergence of a drug-resistant tumor subpopulation subsequent to an initially successful cycle of BCNU therapy, leading to minimal gains from a second cycle. These diffusion MRI findings were highly correlated with tumor growth delay, animal survival, and ex vivo growth inhibition assays showing emerging resistance in excised tumors. Overall, this study highlights the ability of diffusion MRI to provide sensitive dynamic assessment of therapy-induced response, allowing early opportunities for optimization of therapeutic protocols. (Cancer Res 2006; 66(9): 4687-92)

Although advances in therapeutic intervention, such as surgical techniques, radiation therapy, and chemotherapy, have provided some success in managing brain neoplasms, the prognosis for highly aggressive tumors, such as glioblastoma multiforme, is still poor, with a median survival time of 9 months, and only 5% to 10% of patients surviving past 2 years (1). Both neuroresection and radiotherapy have proven to be effective methods in treating gliomas; however, adjuvant chemotherapy has yielded mixed results and is still a subject of controversy (2). Studies have shown that tumors consist of subpopulations, which exhibit variable characteristics, growth rates, and drug sensitivities (35), which may account for variable and unpredictable response to chemotherapy. Even with optimal tumor response, there is a high rate of tumor recurrence of increased malignancy, which is frequently refractory to further treatment. A number of studies suggest that the treatment-refractory nature of these recurrences could result from previous chemotherapy (68). With tumors being such a dynamic entity, optimal therapeutic benefit presumably would require proficient management of the therapeutic regimen to account for these changes. This would require early, sensitive, and progressive assessment of therapeutic response.

Diffusion magnetic resonance imaging (MRI) has emerged as a powerful tool that has garnered attention as a quantitative approach for assessment of tumor response to therapeutic intervention (9). This technique exploits water mobility within tissues as a biomarker that highly correlates with cellularity (1012). Destruction of tumor cells in response to therapy leads to a change in water mobility that is detectable by diffusion MRI. Moreover, these microscopic changes in the diffusibility of water occur well in advance of currently used macroscopic indicators of tumor status, such as changes in volume, performance status, or survival. Our studies, among others, have shown the capability of this approach to provide sensitive and rapid prospective assessment of therapeutic response, which has proven to be invaluable in the evaluation and development of experimental therapies (9, 1218).

In this study, we employed the well-characterized orthotopic 9L glioma model as a system to evaluate therapeutic response to different 1,3-bis(2-chloroethyl)-1-nitrosourea (BCNU) dosing regimens. Historically, this model has been shown to be an effective system for preclinical evaluation of therapeutic strategies, including BCNU. This study focused on the ability of diffusion MRI to provide a quantitative comparison of therapeutic response between subjects given either one 5-day cycle of BCNU given at 2 mg/kg/d (1ω), or two similar cycles of BCNU therapy separated by a 2-day rest period (2ω). Our results showed that not only did diffusion MRI provide an early prediction of therapeutic outcome, but was also sensitive enough to immediately detect the loss of therapeutic response to BCNU. These results were shown to correlate with survival and with ex vivo determination of the emergence of drug resistance in the treated tumors. These findings show that diffusion MRI is an effective tool for dynamic assessment of the dynamic changes in the sensitivity of a tumor to fractionated therapy.

9L animal models. Rat 9L glioma cells were obtained from the Brain Tumor Research Center at the University of California at San Francisco. The cells were grown as monolayers in 10-cm2 sterile plastic flasks in DMEM with 10% fetal bovine serum, 100 IU/mL penicillin and 100 mg/mL streptomycin, and 2 mmol/L l-glutamine in an incubator held at 37°C and 95%/5% air/CO2 atmosphere. Before implantation, cells were harvested by trypsinization, counted, and resuspended in serum-free medium for injection.

Intracerebral tumor implantation was done as previously described (19). Male Fischer 344 (Harlan Sprague-Dawley, Indianapolis, IN) rats weighing between 125 and 150 g were anesthetized with a ketamine/xylazine mixture given i.p. A small skin incision was made over the right hemisphere, and a 1-mm-diameter burr hole was drilled through the skull. A 5-μL suspension containing 1 × 105 9L cells was injected through the burr hole to a depth of 3 mm. The surgical field was then cleaned with 70% ethanol, and the burr hole was filled with bone wax to prevent extracerebral extension of the tumor. Surgical staples were then applied to the incision to seal the open incision.

For treatment, BCNU was dissolved in absolute ethanol and diluted in saline at the time of treatment. Animals were subdivided into three groups. The 1ω group (n = 5) received a total of 10 mg/kg BCNU fractionated into 2 mg/kg doses given by daily i.p. injections for 5 days. The 2ω group (n = 17) received a total of 20 mg/kg BCNU fractionated into 2 mg/kg daily doses for 5 days, given 2 days of rest, and then five more 2 mg/kg daily i.p. injections. Lastly, control animals with 9L tumor implants (n = 5) received injections of only vehicle (10% ethanol in saline) at volumes equivalent to treated groups for a complete 5-day cycle.

Diffusion MRI. Imaging was done every other day beginning 10 days after tumor cell implantation on a Varian Unity Inova system equipped with a 7.0-T, 18.3-cm horizontal bore magnet and a quadrature rat head coil (USA Instruments, Aurora, OH). For MRI examination, rats were anesthetized with an isofluorane/air mixture and maintained at 37°C inside the magnet using a heated animal bed. A single-slice sagittal spin-echo sequence was used to facilitate proper positioning of the animal. An isotropic, diffusion-weighed sequence (20) was employed with two interleaved b factors (Δb = 1,148 s/mm2) and the following acquisition variables: TR/TE = 3,500/60 ms, 128 × 128 matrix, and a 3-cm field of view. Thirteen 1-mm-thick slices separated by a 0.5-mm gap were used to cover the whole rat brain. The low b factor images (b = 100 s/mm2) were essentially T2-weighed to allow tumor volume measurements, as previously described (19). Images were acquired before treatment and at 2-day intervals thereafter. Isotropic apparent diffusion coefficient (ADC) maps were calculated for each image set.

Functional diffusion map analysis. Magnetic resonance images following initiation of therapy were coregistered to the initial pretreatment diffusion map with an automated mutual information module (21). After coregistration, brain tumors were then manually contoured to define the region of interest. The diffusion maps within the tumor at days 4 and 8 were compared with the pretreatment values. Shrinkage or growth of the tumor during the time between scans may have occurred; therefore, only voxels that were present in both the pre-therapy and post-therapy tumor volumes were included. The tumor was further segmented into three different categories, yielding the functional diffusion maps (fDM) in which the red voxels represent areas within the tumor where ADC increased (>20 × 10−3 mm2/s), the blue voxels represent a decreased ADC (<20 × 10−3 mm2/s), and the green voxels represent regions within the tumor that were within these thresholds. Moreover, the percentages of red voxels (>20 × 10−5 mm2/s) were calculated for each group.

Tumor growth delay. Tumor volumes were calculated from the low b factor images, as previously described, and were obtained every 2 days until cessation of the experiment. The mean time for the control group tumors to reach eight times the initial volume was determined. Subsequently, the mean time for tumors from each treatment group, 1ω and 2ω, to reach eight times the initial pretreatment volume was also determined and subtracted by the mean time of the control group to obtain the tumor growth delay value. Tumor growth delay represents the difference in days for treated groups to reach eight times the initial volume compared with the control group.

In vitro cytotoxicity assays. Growth inhibition was evaluated using the sulforhodamine B assay as previously described (22). Briefly, 9L cells were seeded at a density of 4,000 per well in 96-well plates. After 24 hours, the cells were treated with serial dilutions of BCNU ranging from 200 to 0.1 μmol/L or vehicle in complete medium. Cells were allowed to grow until untreated cells reached confluence, at which time the cells were fixed and stained with sulforhodamine B (Sigma, St. Louis, MO). A microtiter plate reader was used to measure the absorbance at 490 nm (Vmax; Molecular Devices, Sunnyvale, CA). Data plotted represent the mean and SD of 2ω BCNU (n = 8) and vehicle controls (n = 2) with each cell line done in quadruplicate.

Diffusion-weighed MRI detects cellular response to fractionated BCNU therapy. Representative serial ADC maps (color overlaid on anatomic image) of controls (vehicle only), 1ω- and 2ω-treated animals are shown in Fig. 1. Pretreatment ADC maps were similar, indicating that the treatment groups were well matched for ADC values. However, whereas tumor diffusion of the control animals remained essentially unchanged over time, an appreciable increase in ADC was observable in the BCNU-treated animals over the first 6 days. Moreover, a progressive decrease in diffusion (shift towards blue) was observed in both 1ω and 2ω groups that started on day 8, which subsequently reached and remained at baseline until the end of the experiment. Interestingly, despite receiving a second cycle of BCNU therapy, the 2ω group similarly decreased in diffusion as with the 1ω group through qualitatively comparing representative images from each group. Visual comparison of the serial MRI images also qualitatively revealed a modest retardation of tumor growth rate in BCNU-treated animals compared with the vehicle control group.

Figure 1.

Anatomic images with color ADC maps from representative tumors over course of therapy. Normalized quantitative color ADC maps overlays of representative vehicle control tumor (left), 1ω group (middle), and 2ω group (right) from days 0 to 14 of therapy. Color intensities are directly related to ADC values (ADC units of 1 × 10−3 mm2/s), with blue representing low diffusion and red representing high diffusion. The ADC value of the tumor in the control animal remained relatively constant throughout the study, whereas both 1ω and 2ω BCNU-treated tumors showed a much greater increase in tumor diffusion throughout treatment.

Figure 1.

Anatomic images with color ADC maps from representative tumors over course of therapy. Normalized quantitative color ADC maps overlays of representative vehicle control tumor (left), 1ω group (middle), and 2ω group (right) from days 0 to 14 of therapy. Color intensities are directly related to ADC values (ADC units of 1 × 10−3 mm2/s), with blue representing low diffusion and red representing high diffusion. The ADC value of the tumor in the control animal remained relatively constant throughout the study, whereas both 1ω and 2ω BCNU-treated tumors showed a much greater increase in tumor diffusion throughout treatment.

Close modal

Figure 2 presents quantitative analysis of the entire imaging data set. These data show a significant increase in the mean ADC beginning on day 2 after initiation of BCNU administration and peaking on day 6 in the 1ω and 2ω groups with values of 18% and 15% increase, respectively. Comparatively, control animals displayed a slight 3% increase in ADC on day 2, which subsequently subsided to baseline values and remained fairly constant throughout the rest of the experiment. Observed differences between control and treated groups were confirmed to be statistically significant as early as day 4 (1ω, P < 0.04; 2ω, P < 0.04), and statistical significance was maintained on day 6 (1ω, P < 0.01; 2ω, P < 0.01) and day 8 (1ω, P < 0.03; 2ω, P < 0.02). As expected, there was no significant difference in mean ADC change between the 1ω- and 2ω-treated groups over the first 8 days (P > 0.2, days 2-8). The mean ADC values of the 1ω group started to decrease at day 7 probably due to completion of therapy and perhaps repopulation of tumor cells. Interestingly, although the second phase of BCNU therapy for the 2ω group was initiated, a progressive drop in ADC was also observed starting on day 7 and progressed throughout the second cycle of dosing (days 7-11). This indicated that a decrease in extracellular space was occurring over time, which would be consistent with repopulation of the tumor mass, possibly due to the emergence of BCNU resistance.

Figure 2.

Mean ADC changes from serial acquisitions throughout therapy. Normalized ADC values from each group were plotted as a function of time. Arrows, 2 mg/kg BCNU injections, where the first cycle was given from days 0 to 4, and animals receiving a second cycle were injected from days 7 to 11. Changes in mean ADC of control groups remained fairly constant, where as both BCNU-treated groups showed a significant rise in ADC after the initial cycle of BCNU therapy. Note that animals receiving a second cycle of BCNU therapy showed a progressive decrease in ADC throughout the second course of BCNU therapy.

Figure 2.

Mean ADC changes from serial acquisitions throughout therapy. Normalized ADC values from each group were plotted as a function of time. Arrows, 2 mg/kg BCNU injections, where the first cycle was given from days 0 to 4, and animals receiving a second cycle were injected from days 7 to 11. Changes in mean ADC of control groups remained fairly constant, where as both BCNU-treated groups showed a significant rise in ADC after the initial cycle of BCNU therapy. Note that animals receiving a second cycle of BCNU therapy showed a progressive decrease in ADC throughout the second course of BCNU therapy.

Close modal

Comparison of heterogeneity in tumor response using fDM analysis. In addition to monitoring mean ADC changes, evaluation of tumor treatment-induced heterogeneity in ADC values was accomplished for the 1ω and 2ω groups. The fDM approach provided a means to quantitatively analyze the heterogeneity of tumor response by tracking regional variations in diffusion (23). By comparing day 4 and 8 ADC maps to pretreatment images, representative scatter plots from vehicle controls (Fig. 3A and D), 1ω group (Fig. 3B and E), and 2ω group (Fig. 3C and F) were generated to represent changes in ADC within specific voxels. The red, green, and blue data points represent the three different tumor regions based on fDM analysis of ADC change. Qualitatively, Fig. 1A and D shows that there was minimal increase in ADC for vehicle controls, whereas a larger fractional increase in ADC, as depicted by a more pronounced red region, was observed in both 1ω and 2ω groups. The percentage of the tumor falling into this region was then calculated to provide a quantitative measure for comparison between groups. In the control group, 11.7 ± 6.4% and 13.1 ± 1.5% of the tumor voxels were found to have an increase of at least 20 ADC units at days 4 and 8, respectively (Table 1). However, the 1ω and 2ω groups exhibited large increases with a mean of 37.6 ± 3.4% and 35.0 ± 4.1%, respectively, at day 4, which were both statistically significant versus controls (1ω, P < 0.05; 2ω, P < 0.05) but were not statistically significant to each other (P > 0.3). At day 8, significant regions of tumor response were also apparent in the 1ω (33.5 ± 14.5%, P < 0.05 versus control) and 2ω (26.1 ± 2.4%, P < 0.05 versus control) groups. More importantly, although the 2ω group received a second cycle of BCNU therapy, this also failed to produce significant additional response over the 1ω group based on fDM analysis (P > 0.2). This is consistent with the earlier finding, which showed minimal differences in mean ADC values at the same time point. Additionally, comparison of days 8 to 4 indicates a slight decrease in mean values for both 1ω and 2ω groups, correlating with the dynamic shift in ADC means. This decrease in ADC values suggests a decline in tumor responsiveness of the 2ω group to BCNU, although daily treatments were still being given.

Figure 3.

Scatter plots of tumor response using fDM analysis. Scatter plots were generated for representative animals from control (A and D), 1ω group (B and E), and 2ω group (C and F) at day 4 (A-C) and day 8 (D-F) to show the distribution of ADC (ADC units of 1 × 10−3 mm2/s) changes for the entire tumor volume. The red pixels indicate areas of increased diffusion, whereas the blue and green pixels indicate regions of decreased and unchanged ADC, respectively. For the control group, a minimal percentage of tumor voxels were found to have an increase in ADC at either day 4 (A) or day 8 (D). The 1ω (B and E) and 2ω (C and F) groups exhibited much larger increases that were statistically significant versus controls at day 4 (1ω, P < 0.05; 2ω, P < 0.05) and day 8 (1ω, P < 0.05; 2ω, P < 0.05). However, statistical significance was not achieved between the 1ω and 2ω groups at either day 4 (P > 0.3) or day 8 (P > 0.02).

Figure 3.

Scatter plots of tumor response using fDM analysis. Scatter plots were generated for representative animals from control (A and D), 1ω group (B and E), and 2ω group (C and F) at day 4 (A-C) and day 8 (D-F) to show the distribution of ADC (ADC units of 1 × 10−3 mm2/s) changes for the entire tumor volume. The red pixels indicate areas of increased diffusion, whereas the blue and green pixels indicate regions of decreased and unchanged ADC, respectively. For the control group, a minimal percentage of tumor voxels were found to have an increase in ADC at either day 4 (A) or day 8 (D). The 1ω (B and E) and 2ω (C and F) groups exhibited much larger increases that were statistically significant versus controls at day 4 (1ω, P < 0.05; 2ω, P < 0.05) and day 8 (1ω, P < 0.05; 2ω, P < 0.05). However, statistical significance was not achieved between the 1ω and 2ω groups at either day 4 (P > 0.3) or day 8 (P > 0.02).

Close modal
Table 1.

Quantitation of tumor response using fDM analysis

Day 4
Day 8
ControlControl
Mean (SE), % 11.7 (6.4) 37.6 (3.4) 35 (4.1) 13.1 (1.5) 33.5 (14.5) 26.1 (2.4) 
P vs control — <0.05 <0.05 — <0.05 <0.05 
P vs 1ω — — >0.3 — — >0.2 
Day 4
Day 8
ControlControl
Mean (SE), % 11.7 (6.4) 37.6 (3.4) 35 (4.1) 13.1 (1.5) 33.5 (14.5) 26.1 (2.4) 
P vs control — <0.05 <0.05 — <0.05 <0.05 
P vs 1ω — — >0.3 — — >0.2 

NOTE: Animals from control, 1ω group, and 2ω group were analyzed using the fDM approach to determine the average percentage of tumor voxels exhibiting a significant increase (red pixels) in ADC (>20 × 10−3 mm2/s).

Minimal therapeutic gain with 2ω as compared with 1ω. Therapeutic benefit of 1ω and 2ω dosing regimens were assessed using tumor volume measurements and animal survival. By using the low b factor images to measure tumor volumes (as described in Materials and Methods), changes in growth kinetics due to therapeutic intervention were calculated and represented as tumor growth delays. Comparing the treatment groups to control, 1ω and 2ω groups exhibited significant tumor growth delays of 4.05 ± 0.25 days (P < 0.001 versus control) and 3.55 ± 0.52 days (P < 0.001 versus control), respectively. However, significant differences in tumor growth delay between the 1ω and 2ω groups was not achieved (P > 0.3). Additionally, a Kaplan-Meier plot of animal survival probabilities, shown in Fig. 4, was generated by measuring the time differential between initiation of therapy to animal death due to tumor burden. The median survival time of control, 1ω group, and 2ω group was 9, 12.5, and 13 days, respectively. A Mantel-Cox log rank test was done, indicating statistical significance between survival times of control versus 1ω group (P < 0.04) and control versus 2ω group (P < 0.001). However, the 2ω group did not show improved survival compared with animals treated with a single cycle of BCNU (1ω, P > 0.4). This finding indicates that the inclusion of a second cycle of BCNU therapy had no discernable benefit although presumably still adding to systemic toxicity. Although therapeutic outcome is dependent on many factors, the results presented here show that diffusion MRI is predictive of therapeutic outcome by providing early detection of a loss in tumor response during the second treatment cycle as evidenced by changes in mean ADC values and fDM analysis.

Figure 4.

Kaplan-Meier survival curves. Kaplan-Meier survival curves for animal survival were determined as the time of survival from onset of therapy until animal death due to tumor burden. As shown in the curve, there was a significant survival advantage (P < 0.05) for both 1ω (12.5 days) and 2ω (13 days) fractionated BCNU-treated groups versus the vehicle control group (9 days). However, there was no significant survival advantage for animals receiving a second cycle of therapy (2ω) versus animals receiving only a single cycle (1ω; P > 0.4).

Figure 4.

Kaplan-Meier survival curves. Kaplan-Meier survival curves for animal survival were determined as the time of survival from onset of therapy until animal death due to tumor burden. As shown in the curve, there was a significant survival advantage (P < 0.05) for both 1ω (12.5 days) and 2ω (13 days) fractionated BCNU-treated groups versus the vehicle control group (9 days). However, there was no significant survival advantage for animals receiving a second cycle of therapy (2ω) versus animals receiving only a single cycle (1ω; P > 0.4).

Close modal

Ex vivo BCNU sensitivity assays reveal acquired drug resistance in BCNU-treated tumors. As previously described, ADC values in 2ω-treated animals failed to increase during the second cycle of therapy and were shown to correlate with a lack of additional therapeutic benefit. To investigate whether the lack of response in mean ADC during the second cycle of BCNU in the 2ω group was due to the emergence of a BCNU-resistant population, tumor cells were isolated from vehicle control tumors as well as tumors that had undergone two cycles of BCNU treatment. The sensitivity of these cells to BCNU was then evaluated using the sulforhodamine B growth inhibition assay to determine possible alterations in BCNU sensitivity of treated tumors compared with controls. As shown in Fig. 5, cell lines derived from 2ω-treated animals exhibited resistance to BCNU compared with control tumors. Calculated IC50s showed a 2-fold decrease in sensitivity of the BCNU-treated tumors compared with vehicle controls (150.4 ± 1.5 and 73.3 ± 4.9 μmol/L, respectively). This observed increase in IC50 of BCNU-treated tumors is comparable with previous reports, where 9L tumors have been shown to acquire resistance to chemotherapy following in vivo BCNU therapy (24).

Figure 5.

Sulforhodamine B growth inhibition assay for BCNU sensitivity. Cell lines were developed from excised 2ω and vehicle control tumors and assayed for BCNU sensitivity. Using a previously described assay (Materials and Methods), cell lines were grown in varying BCNU concentrations ranging from 1 to 200 μmol/L and were then fixed to determine differences in total protein. A 2-fold shift in IC50 was observed in tumors from the 2ω group (150.4 ± 1.9 and 73.8 ± 4.9 μmol/L, respectively).

Figure 5.

Sulforhodamine B growth inhibition assay for BCNU sensitivity. Cell lines were developed from excised 2ω and vehicle control tumors and assayed for BCNU sensitivity. Using a previously described assay (Materials and Methods), cell lines were grown in varying BCNU concentrations ranging from 1 to 200 μmol/L and were then fixed to determine differences in total protein. A 2-fold shift in IC50 was observed in tumors from the 2ω group (150.4 ± 1.9 and 73.8 ± 4.9 μmol/L, respectively).

Close modal

Current methods for gauging tumor response to therapy rely either on volumetric change over the course of therapeutic intervention or alterations in survival as indicators of efficacy. Our previous studies have validated the use of diffusion MRI as a noninvasive biomarker for evaluating anticancer efficacy (12, 13, 25). This approach has been shown to be universally applicable to a wide variety of models (11, 14, 16, 17, 26, 27) and is highly sensitive because microscopic alterations in tissue structure and physiology result in significant changes to water movement within the microenvironment (28, 29). Moreover, these have shown that the changes detected by diffusion MRI precede detectable alterations in tumor size or growth kinetics. Although these studies have validated the use of diffusion MRI for preclinical studies, only recently have studies provided direct clinical evidence that diffusion MRI is predictive of therapeutic outcome in patients with primary brain tumors (12, 30, 31). In our present study, we show the use of diffusion MRI for allowing dynamic access into understanding therapeutic response by serially observing changes in ADC. The implications of this work provide a new direction for dynamic management of therapy where the ability to rapidly assess therapeutic efficacy during treatment could provide an opportunity to optimize initial regimens.

To test the use of diffusion MRI as a noninvasive, sensitive readout for optimization of dose and schedule of cancer therapy, we evaluated two separate schedules of BCNU therapy in an orthotopic 9L tumor model. In this study, diffusion MRI predicted initial therapeutic response, but more importantly, also detected a loss of tumor response during ongoing therapy. As shown in Fig. 2, there was a statistical difference in mean ADC change between BCNU-treated groups and control. However, comparison of the two treatment groups showed only slight differences in ADC profiles. The 1ω group did have a more pronounced decline in ADC at the end of therapy compared with the 2ω group, suggesting perhaps that a second cycle of therapy produced some level of tumor response. However, statistical significance of mean ADC change between the treated groups was never achieved throughout the course of the experiment. Nonetheless, this mirroring of ADC profiles between the 1ω and 2ω groups correctly predicted comparable tumor response. This provided the first indication that although the 2ω group received a much higher total dose of BCNU, the more aggressive therapy did not lead to greater tumor cell death compared with the 1ω group. Moreover, the decrease of ADC values in the 2ω group before and throughout the second cycle of therapy indicated that the initial cycle of BCNU therapy altered tumor responsiveness to BCNU cytotoxicity during the second cycle.

fDM analysis of ADC change has been validated as an invaluable clinical tool for stratifying tumor response and predicting patient outcome in cases of primary brain neoplasms (29, 30). More importantly, the fDM approach provides an avenue to account for heterogeneity inherent of solid tumors and their response to therapy (23). Using this methodology, a quantitative assessment of therapeutic tumor response is obtained by measurements of the changes in regional fluctuations in ADC values within the whole tumor. In this study, we showed (Fig. 3; Table 1) that treated tumors displayed a significant tumor response versus controls, as evidenced by an increase of the tumor fraction with elevated ADC values. However, no statistical difference in tumor response, based on fDM analysis, was observed between the 1ω and 2ω groups at day 4 (P > 0.3). Moreover, at day 8, whereas therapy was still in progress for the 2ω group, similar tumor response was observed between the 1ω and 2ω groups, suggesting that the second cycle of therapy provided minimal therapeutic benefit over tumors receiving only a single cycle.

Tumor growth delays and a Kaplan-Meier plot of animal survival were generated to ascertain the therapeutic value obtained from two cycles of BCNU therapy (2ω) versus a single cycle (1ω). As previously described, although treated groups showed a significant tumor growth delay compared with control tumors (1ω, 4.05 ± 0.25 days; 2ω, 3.55 ± 0.52 days), no significant difference was discernable between the 1ω and 2ω groups (P > 0.3). Moreover, as shown in Fig. 4, the 2ω group exhibited negligible survival advantage compared with the 1ω group, where median survival times were 13 and 12.5 days, respectively. This correlated with the diffusion data where a progressive decrease in mean ADC change during the second phase of therapy showed a decrease in cellular response to BCNU. Moreover, fDM analysis of treated tumors showed that the 1ω and 2ω groups had minimal differences in voxels of increased ADC at both days 4 and 8, thereby indicating that a second cycle of BCNU therapy did not provide an appreciable gain in tumor response. These findings reveal that diffusion MRI is effective as an early predictive biomarker for detection of diminished tumor response to a second cycle of BCNU therapy. This was confirmed by a lack in survival benefit of the 2ω group over the 1ω group.

With both diffusion MRI and animal survival studies indicating diminishing response of 2ω tumors to BCNU therapy, we investigated whether BCNU-treated tumors did in fact acquire resistance to BCNU. Representative tumors from the 2ω and vehicle control groups were excised, and cell lines were derived to establish BCNU sensitivity. Using a growth inhibition assay (22), we showed that BCNU-treated tumors were 2-fold more resistant to BCNU than control tumors with an IC50 of 150.4 ± 1.9 versus 73.8 ± 4.9 μmol/L, respectively. These findings parallel previous reports documenting the development of BCNU-resistant 9L tumors induced by in vivo BCNU therapy (23, 31) and confirms that diffusion MRI predicted loss of tumor response over time, which corresponded to diminished therapeutic benefit.

In summary, this report shows that diffusion MRI can provide a rapid assessment of therapy-induced tumor response to fractionated BCNU as well as the dynamic detection of the emergence of resistance. The findings of this study provide an example of the potential for diffusion MRI to play a major role in the preclinical development of therapeutic regimens with dynamic assessment of therapeutic effect driving efficient optimization of treatment regimens within a single experiment, dramatically increasing the efficiency of preclinical experimentation. Moreover, the data presented also clearly points to the potential application of diffusion MRI technologies for dynamic optimization of therapeutic outcomes of individual patients in the clinic.

Grant support: NIH/National Cancer Institute grants P01CA85878, R24CA83099, and P50CA093990.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1
Behin A, Hoang-Xuan K, Carpentier AF, et al. Primary brain tumours in adults.
Lancet
2003
;
361
:
323
–31.
2
DeAngelis LM. Brain tumors.
N Engl J Med
2001
;
344
:
114
–23.
3
Dexter DL, Calabresi P. Intraneoplastic diversity.
Biochim Biophys Acta
1982
;
695
:
97
–112.
4
Fidler IJ, Hart IR. Biological diversity in metastatic neoplasms: origins and implications.
Science
1982
;
217
:
998
–1003.
5
Heppner GH. Tumor cell societies.
J Natl Cancer Inst
1989
;
81
:
648
–9.
6
Goldie JH, Coldman AJ. The genetic origin of drug resistance in neoplasms: implications for systemic therapy.
Cancer Res
1984
;
44
:
3643
–53.
7
Heros DO, Renkens K, Kasdon DL, et al. Patterns of recurrence in glioma patients after interstitial irradiation and chemotherapy: report of three cases.
Neurosurgery
1988
;
22
:
474
–8.
8
Rudoler S, Corn BW, Werner-Wasik M, et al. Patterns of tumor progression after radiotherapy for low-grade gliomas: analysis from the computed tomography/magnetic resonance imaging era.
Am J Clin Oncol
1998
;
21
:
23
–7.
9
Ross BD, Chenevert TL, Kim B, et al. Magnetic resonance imaging and spectroscopy: application to experimental neuro-oncology.
Q Magn Reson Biol Med
1994
;
1
:
89
–106.
10
Lyng H, Haraldseth O, Rofstad EK. Measurement of cell density and necrotic fraction in human melanoma xenografts by diffusion weighted magnetic resonance imaging.
Magn Reson Med
2000
;
43
:
828
–36.
11
Sugahara T, Korogi Y, Kochi M, et al. Usefulness of diffusion-weighted MRI with echo-planar technique in the evaluation of cellularity in gliomas.
J Magn Reson Imaging
1999
;
9
:
53
–60.
12
Chenevert TL, Stegman LD, Taylor JM, et al. Diffusion magnetic resonance imaging: an early surrogate marker of therapeutic efficacy in brain tumors.
J Natl Cancer Inst
2000
;
92
:
2029
–36.
13
Chenevert TL, McKeever PE, Ross BD. Monitoring early response of experimental brain tumors to therapy using diffusion magnetic resonance imaging.
Clin Cancer Res
1997
;
3
:
1457
–66.
14
Hakumaki JM, Poptani H, Puumalainen AM, et al. Quantitative 1H nuclear magnetic resonance diffusion spectroscopy of BT4C rat glioma during thymidine kinase-mediated gene therapy in vivo: identification of apoptotic response.
Cancer Res
1998
;
58
:
3791
–9.
15
Galons JP, Altbach MI, Paine-Murrieta GD, et al. Early increases in breast tumor xenograft water mobility in response to paclitaxel therapy detected by non-invasive diffusion magnetic resonance imaging.
Neoplasia
1999
;
1
:
113
–7.
16
Stegman LD, Rehemtulla A, Hamstra DA, et al. Diffusion MRI detects early events in the response of a glioma model to the yeast cytosine deaminase gene therapy strategy.
Gene Ther
2000
;
7
:
1005
–10.
17
Hamstra DA, Lee KC, Tychewicz JM, et al. The use of 19F spectroscopy and diffusion-weighted MRI to evaluate differences in gene-dependent enzyme prodrug therapies.
Mol Ther
2004
;
10
:
916
–28.
18
Hall DE, Moffat BA, Stojanovska J, et al. Therapeutic efficacy of DTI-015 using diffusion magnetic resonance imaging as an early surrogate marker.
Clin Cancer Res
2004
;
10
:
7852
–9.
19
Ross BD, Zhao YJ, Neal ER, et al. Contributions of cell kill and posttreatment tumor growth rates to the repopulation of intracerebral 9L tumors after chemotherapy: an MRI study.
Proc Natl Acad Sci U S A
1998
;
95
:
7012
–7.
20
Wong EC, Cox RW, Song AW. Optimized isotropic diffusion weighting.
Magn Reson Med
1995
;
34
:
139
–43.
21
Meyer CR, Boes JL, Kim B, et al. Demonstration of accuracy and clinical versatility of mutual information for automatic multimodality image fusion using affine and thin-plate spline warped geometric deformations.
Med Image Anal
1997
;
1
:
195
–206.
22
Skehan P, Storeng R, Scudiero D, et al. New colorimetric cytotoxicity assay for anticancer-drug screening.
J Natl Cancer Inst
1990
;
82
:
1107
–12.
23
Moffat BA, Chenevert TL, Meyer CR, et al. The functional diffusion map: an imaging biomarker for early prediction of cancer treatment outcome.
Neoplasia
2006
;
8
:
259
–66.
24
Barker M, Hoshino T, Gurcay O, et al. Development of an animal brain tumor model and its response to therapy with 1,3-bis(2-chloroethyl)-1-nitrosourea.
Cancer Res
1973
;
33
:
976
–86.
25
Chenevert TL, Meyer CR, Moffat BA, et al. Diffusion MRI: a new strategy for assessment of cancer therapeutic efficacy.
Mol Imaging
2002
;
1
:
336
–43.
26
Poptani H, Puumalainen AM, Grohn OH, et al. Monitoring thymidine kinase and ganciclovir-induced changes in rat malignant glioma in vivo by nuclear magnetic resonance imaging.
Cancer Gene Ther
1998
;
5
:
101
–9.
27
Zhao M, Pipe JG, Bonnett J, et al. Early detection of treatment response by diffusion-weighted 1H-NMR spectroscopy in a murine tumour in vivo.
Br J Cancer
1996
;
73
:
61
–4.
28
Evelhoch JL, Gillies RJ, Karczmar GS, et al. Applications of magnetic resonance in model systems: cancer therapeutics.
Neoplasia
2000
;
2
:
152
–65.
29
Rowley HA, Grant PE, Roberts TP. Diffusion MR imaging. Theory and applications.
Neuroimaging Clin N Am
1999
;
9
:
343
–61.
30
Moffat BA, Chenevert TL, Lawrence TS, et al. Functional diffusion map: a noninvasive MRI biomarker for early stratification of clinical brain tumor response.
Proc Natl Acad Sci U S A
2005
;
102
:
5524
–9.
31
Hamstra DA, Chenevert TL, Moffat BA, et al. Evaluation of the functional diffusion map as an early biomarker of time-to-progression and overall survival in high-grade glioma.
Proc Natl Acad Sci U S A
2005
;
102
:
16759
–64.
32
Saito R, Bringas J, Mirek H, et al. Invasive phenotype observed in 1,3-bis(2-chloroethyl)-1-nitrosourea-resistant sublines of 9L rat glioma cells: a tumor model mimicking a recurrent malignant glioma.
J Neurosurg
2004
;
101
:
826
–31.