HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Figures Only Full Text (PDF) All Versions of this Article: 2392021099v1 239/2/361    most recent Submit a response Alert me when this article is cited Alert me when eLetters are posted Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Padhani, A. R. Articles by Husband, J. E. Search for Related Content PubMed PubMed Citation Articles by Padhani, A. R. Articles by Husband, J. E.

Prediction of Clinicopathologic Response of Breast Cancer to Primary Chemotherapy at Contrast-enhanced MR Imaging: Initial Clinical Results¹

¹ From the Cancer Research UK Clinical Magnetic Resonance Research Group (A.R.P., C.H., J.S., M.O.L., J.E.H.) and the Breast Unit (L.A., T.P., A.M.), Institute of Cancer Research and the Royal Marsden NHS Trust, Surrey, England. Received September 2, 2002; revision requested November 4; final revision received May 19, 2005; accepted May 27; final version accepted July 18. Supported by Cancer Research Campaign grant SP1780-0103. Address correspondence to A.R.P., Paul Strickland Scanner Centre, Mount Vernon Hospital, Rickmansworth Road, Northwood, Middlesex HA6 2RN, UK (e-mail: anwar.padhani{at}paulstrickland-scannercentre.org.uk).

	ABSTRACT

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION APPENDIX ADVANCES IN KNOWLEDGE References

 
Purpose: To prospectively document changes in contrast agent kinetics in patients with primary breast cancer treated with systemic chemotherapy after one or two cycles and to determine whether kinetic measures can be used to predict final clinicopathologic response.

Materials and Methods: Institutional committees on clinical research and ethics approval and patient consent were obtained. Dynamic magnetic resonance (MR) examinations were performed in 25 women with primary breast cancer before treatment and after the first (n = 21) and second (n = 15) cycle of neoadjuvant chemotherapy. Kinetic parameters (transfer constant, leakage space, and rate constant) were derived for whole tumor regions of interest. Changes in histogram distributions of pixel data (median value and range) and MR imaging–derived size were correlated with final clinical and histologic response by using nonparametric methods. Receiver operating characteristic (ROC) analysis of tumor size and transfer constant changes were used to identify patients in whom no benefit was gained from chemotherapy.

Results: After the first cycle of treatment, 12 of 14 clinical responders showed decreases in tumor size, and six of seven nonresponders showed increases or no change in tumor size (P < .001). Transfer constant changes did not differ between responders and nonresponders for either clinical or pathologic assessments. After two cycles of treatment, there were tumor size increases in five of six nonresponders compared with decreases in eight of nine responders (P < .001). Reductions in transfer constant range were also observed in responders for both clinical and pathologic assessments (P = .008 and .02, respectively). No other kinetic parameter change predicted response. Size and transfer constant range were equally accurate for predicting the absence of pathologic response after two cycles of treatment (sensitivity, specificity, and area under ROC curve were 100%, 90%, and 0.93, respectively, for size and 100%, 75%, and 0.94, respectively, for transfer constant range).

Conclusion: Reductions in MR imaging–determined size of the primary tumor best predict clinicopathologic response of breast cancer after one cycle of neoadjuvant chemotherapy. Transfer constant and size changes are equally sensitive in the identification of patients who would gain no clinical or pathologic benefit after two cycles of treatment.

© RSNA, 2006

	INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION APPENDIX ADVANCES IN KNOWLEDGE References

 
Neoadjuvant chemotherapy is used prior to surgery in patients with primary breast cancer to reduce surgical requirements (fewer mastectomies) and to treat undetected micrometastases. Results of randomized trials confirm these benefits and show equivalent survival for adjuvant and neoadjuvant chemotherapy in patients with primary operable breast cancer (1–4). A further benefit of neoadjuvant chemotherapy is the opportunity to assess the chemoresponsiveness of the tumor in situ. The overall response rates reported vary between 60% and 100%, with complete clinical responses ranging from 10% to almost 50%, avoiding mastectomy in most cases. Clinical responders have a better prognosis than do nonresponders (1,2,5). The prognostic importance of histopathologic response among patients undergoing neoadjuvant chemotherapy for breast cancer is also recognized (5,6). Patients who have complete pathologic response or pathologic minimal residual disease have a longer disease-free and overall survival compared with patients who have gross residual disease (6). The ability to identify nonresponders early after the start of chemotherapy would be of major benefit because it would enable treatment to be adjusted or enable alternative and possibly more efficacious treatments, such as other types of chemotherapy or early surgery, to be offered as soon as possible (7).

Currently, measurement of breast tumor size (by means of clinical examination, mammography, or ultrasonography [US]) is used to monitor response to treatment and to assess the volume of residual disease after treatment. Results of many studies evaluating neoadjuvant chemotherapy in patients with breast cancer have shown that these measurement techniques are imperfect because tumor size changes often become apparent only after several doses of chemotherapy. Each of these techniques also overestimates the extent of residual disease at the end of treatment. For example, Fiorentino et al (8) showed that clinical assessment on completion of neoadjuvant chemotherapy correlated better with the pathologic volume of residual disease than did US and mammography. However, clinical palpation is an imperfect evaluation method because of the intervening skin and soft tissue, and the examiner is unable to distinguish active disease from postchemotherapy fibrosis. Feldman et al (9) made the observation that there was a significant difference between clinical and pathologic evaluation of residual disease such that the improved patient survival achieved in those with complete pathologic response (P = .002) would not be appreciated if residual disease had been judged clinically (P = .09). Assessments at mammography are hampered by radiographic magnification, and it may be difficult to distinguish residual tumor from postchemotherapy fibrosis. It is also well recognized that some tumors are simply not visualized in patients with radiographically dense breasts. Results of several studies have shown that contrast material–enhanced magnetic resonance (MR) imaging is the most accurate technique for evaluating pathologic response and volume of residual active disease at the end of treatment (10–12).

Dynamic contrast-enhanced MR imaging techniques allow some of the functional effects of tumor vascularity to be studied in vivo (13). Studies results have shown broad correlations between MR imaging contrast enhancement and histologic markers of angiogenesis in breast cancer—in particular, microvessel density (14,15) and tissue vascular endothelial growth factor (VEGF) staining (16). The ability to quantify dynamic contrast-enhanced MR images enables MR to be used as a functional method for monitoring response to a variety of physical and pharmaceutical treatments. Dynamic contrast-enhanced MR imaging has been evaluated in the clinical setting of neoadjuvant chemotherapy in bladder and breast cancers and bone sarcomas, as well as in radiation therapy, androgen deprivation, and antiangiogenic or vascular targeting agents (see reference 13 for comprehensive review). Results of all these studies have shown that successful treatment causes decreases in the rate and magnitude of contrast enhancement and that poor response results in persistent abnormal contrast enhancement.

The value of dynamic contrast-enhanced MR imaging for an early prediction of the efficacy of neoadjuvant chemotherapy in patients with breast cancer has not been fully assessed. We hypothesized that reductions in MR estimates of microvessel permeability–surface area product (now called transfer constant) could be used to predict breast cancer response shortly after commencing systemic chemotherapy and that an increase or no change in transfer constant would predict nonresponsiveness. Thus, the purpose of this study was to prospectively document changes in contrast agent kinetics in patients with primary breast cancer treated with systemic chemotherapy after one to two cycles and to determine whether kinetic measures can be used to predict final clinicopathologic response.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION APPENDIX ADVANCES IN KNOWLEDGE References

 
Patients, Treatment, and Response Assessment
A prospective, observational clinical trial of 30 consecutive patients with biopsy-proved breast cancer was undertaken. In five patients, results of the first (pretreatment) MR examination were unusable for the following reasons: substantial breast movement (n = 1), no tumor identified in the dynamic section plane (n = 1), extravasation of contrast medium (n = 1), or failure to comply with sequence protocols, resulting in technical failures (n = 2). Thus, 25 patients formed our study cohort. Sixty-one MR examinations were performed before chemotherapy (25 patients) and after one (21 patients) and two (15 patients) cycles of chemotherapy. The examinations were performed between June 1995 and April 1999. Our institutional committees on clinical research and ethics approved the study, and consent was obtained from each patient. The median age of the patients was 49 years (range, 26–65 years). Histologic diagnosis was obtained by means of core or open biopsy (10 patients) or fine-needle aspiration biopsy (with subsequent postchemotherapy pathologic confirmation). Fourteen patients had invasive ductal cancer, three had invasive lobular cancer, six had invasive carcinoma not otherwise specified, one had mixed invasive ductal and lobular cancer, and one had malignant (C5) cytologic findings with no further histologic clarification. The distribution of patients according to tumor size, clinical nodal status, and menopausal status is given in Table 1. Eleven patients had clinical stage IIA, nine had stage IIB, three had stage IIIA, and two had stage IIIB tumors (17).

After primary chemotherapy, 22 patients went on to undergo surgery (wide local excision of the residual tumor in 15 patients and mastectomy in seven patients) and postoperative radiation treatment. Three patients were treated with radical external beam radiation therapy alone after chemotherapy. Fourteen patients also underwent further adjuvant chemotherapy after surgery.

At the end of treatment, clinical response assessments were undertaken by one of two oncologists by using bidimensional measurements from palpation. Response was recorded according to conventional International Union Against Cancer criteria (18): clinical complete response, which represents disappearance of the primary tumor; clinical partial response, which indicates a reduction of 50% or more in the product of the diameters; and clinical stable disease, defined as a reduction of less than 50% or an increase of less than 25%. An increase in size of 25% or more was considered clinical progressive disease. The residual palpable abnormality after a good response that frequently results in a breast irregularity with no evidence of a clinically measurable mass was defined as clinical minimal residual disease. Histopathologic tumor response was assessed in 22 surgical excision specimens according to the scheme proposed by Honkoop et al (19). Two pathologic response categories were defined: macroscopic residual tumors or microscopic extensive infiltration (group A) and microscopic disease (group B). Group A responses were further subclassified by means of clinical response as defined above: clinical stable disease or progressive disease (subgroup A1) and clinical partial response, minimal residual disease, or complete response (subgroup A2). Group B responses comprised tumors with only scattered foci of invasive or noninvasive tumor cells (pathologic minimal residual disease) and those with complete histopathologic response.

MR Imaging
Theory.—This study used a pharmacokinetic model to estimate kinetic parameters from T1-weighted dynamic contrast-enhanced MR imaging data (20,21). These parameters include the transfer constant, K^trans (in min^–1), of gadolinium-based contrast agent between the blood plasma and the extravascular extracellular space (EES), the EES fractional volume, _e, and the rate constant, k_ep, between EES and blood plasma: k_ep = K^trans/_e. The pharmacokinetic model, described in the Appendix, requires as input the tissue concentration of contrast agent as a function of time, C(t), following a bolus intravenous injection.

To quantify tissue contrast agent concentration in vivo by using dynamic contrast-enhanced MR imaging techniques, the changing tissue T1 relaxation rate must be measured. Tissue concentrations of contrast agent may be derived from the following equation:

(1)

where T1(t) is the tissue T1 relaxation rate at time t following contrast agent administration, T1₀ is the tissue T1 relaxation rate before contrast agent administration, and R₁ is the longitudinal relaxivity of protons in vivo owing to a contrast agent. In this study, a value of R₁ = 4.5 sec^–1 · mmol^–1 for the contrast agent gadopentetate dimeglumine (Magnevist; Schering Health Care, Burgess Hill, UK) was taken (22). Tissue T1 relaxation rate values may be obtained from the ratio of two images acquired with different flip angles. Theoretically, the signal intensity, S, from a spoiled fast low-angle shot (FLASH) MR sequence, assuming an echo time (TE) much less than T2*, is given as:

(2)

where S₀ is a constant proportional to the proton density of the sample, is the flip angle, TR is the pulse repetition time, and T1 is the tissue relaxation rate. It is thus possible in theory to calculate the T1 relaxation rate of any voxel from the ratio R obtained from the division of two image data sets acquired with flip angles _A and _B, respectively, and repetition times TR_A and TR_B, respectively, provided that the echo time and the imaging system gain factors are maintained constant:

(3)

In practice, Equation (3) holds true if the excitation profile in the section-select direction is rectangular. This cannot be assumed for the two-dimensional spoiled FLASH acquisitions used in this study. Consequently, T1 relaxation rate reference samples consisting of gels with known T1 relaxation rates were used to calibrate the T1 measurement procedure (23).

This study also used a saturation-recovery turbo FLASH sequence to measure T1 relaxation rate. This sequence consists of a saturation scheme followed by a recovery time, , after which an image is acquired by using a spoiled FLASH technique with centric-ordered phase encoding (24). To allow T1 relaxation rate values to be inferred by using this sequence, an approach similar to that described for the spoiled FLASH protocol was used except that the acquisition parameters, apart from the recovery time, are the same for the two image data sets, A and B. Thus, for this sequence, and for _B much greater than T1, T1 can be determined directly from the signal intensity ratio, R:

(4)

Data acquisition.—All examinations were performed with a 1.5-T MR system (Vision; Siemens Medical Solutions, Erlangen, Germany) by using a standard bilateral breast coil. The imaging protocol comprised precontrast coronal three-dimensional T1-weighted gradient-echo and transverse two-dimensional T2-weighted turbo spin-echo image acquisitions of both breasts for the purposes of tumor localization. The imaging parameters for the coronal three-dimensional gradient-echo T1-weighted sequence were as follows: repetition time (TR) of 12 msec, TE of 5 msec, 2.5-mm-thick contiguous sections, one signal acquired, matrix size of 512 x 190, field of view of 34 cm, and acquisition time of 2 minutes 42 seconds. For the transverse T2-weighted turbo spin-echo sequence, the imaging parameters were as follows: TR of 3.5–4.1 seconds, TE of 90 msec, echo train length of seven, 4–6-mm-thick contiguous sections, one signal acquired, matrix size of 256 x 256, field of view of 35 cm, and an acquisition time of 3 minutes 50 seconds. These images were inspected by a radiologist with 3 years of breast MR imaging experience for the presence of an abnormality consistent with cancer.

To determine contrast agent concentration by using Equation (1), the tissue T1 values were calculated from the signal intensity ratio of a T1-weighted gradient-echo image and an intermediate-weighted gradient-echo image. The signal intensity of the intermediate-weighted image was assumed to be invariant with the administration of contrast agent; therefore, this image was acquired just once prior to the injection. The T1-weighted images were acquired in a sequential fashion before and after contrast agent administration. The acquisition time of the T1-weighted image was 9–10 seconds, depending on the protocol, and a total of 42 acquisitions were performed. Gadopentetate dimeglumine (Magnevist; Schering Health Care) was manually injected intravenously as a bolus through a peripherally placed cannula after the third baseline data point by a single radiologist (0.1 mmol per kilogram of body weight injected within 10 seconds, followed by a 20-mL flush of normal saline).

The MR images were acquired in the sagittal plane by using either a single-section two-dimensional spoiled FLASH sequence (in 28 examinations) or a five-section two-dimensional saturation recovery turbo FLASH sequence (in 33 examinations). The parameters of the intermediate-weighted two-dimensional FLASH sequence were TR of 350 msec, TE of 5 msec, and flip angle of 20°, and the parameters of the T1-weighted two-dimensional FLASH sequence were TR of 35.1 msec, TE of 5 msec, and flip angle of 70°. The imaging matrix was 256 x 198 (read-phase), field of view was 20 cm, and section thickness was 10 mm for both types of image. The parameters of the intermediate-weighted two-dimensional saturation-recovery turbo FLASH sequence were TR of 11.7 msec, TE of 4.4 msec, flip angle of 20°, recovery time of 10 000 msec, and acquisition time of 58 seconds, and the parameters of the T1-weighted two-dimensional saturation recovery turbo FLASH sequence were TR of 11.7 msec, TE of 4.4 msec, flip angle of 20°, recovery time of 150 msec, and acquisition time of 9 seconds. The imaging matrix was 128 x 128 (read-phase), field of view was 20 cm, and section thickness was 8 mm for both types of image. Machine-dependent signal intensity scaling factors were kept constant throughout the examinations.

Image Review and Analysis
Two observers (the radiologist mentioned previously and a breast MR physicist) working in consensus, who were unaware of the patients' eventual response to treatment, undertook evaluation of the images. Morphologic and kinetic parameter assessments were made on an independent workstation by using original specialist software (MR Imaging Workbench; Institute of Cancer Research, London, UK) designed to quantitatively display and analyze dynamic contrast-enhanced data sets (25). All analyses were performed at the same session (in individual patients) to minimize intraobserver variability. Each dynamic data set took approximately 30–40 minutes to evaluate.

Size assessments.—The size of the tumor at each MR imaging examination was estimated by using the product of the bidimensional diameters of the enhancing tumor at its center (measured on the 90–120-second subtraction image). When multifocal enhancing lesions were seen, the largest measurable lesion was chosen as the index lesion. Changes in enhanced tumor size were categorized as increased (>10% increase from baseline), unchanged (±10%), or decreased (>10% reduction in size). When no tumor was visible, a complete response (100%) was assumed.

Kinetic parameter assessments.—Regions of interest were manually placed around the enhancing edge of the tumor on subtraction images (see above) with an appropriate window to display a clear enhancing tumor margin. Care was taken to avoid areas of nonenhancement around individual lesions. All sections containing tumor were analyzed. Regions of interest were placed at similar locations on pre- and posttreatment examination images as far as was possible. One patient had no visible tumor after two cycles of treatment, and thus kinetic parameters of the "tumor bed" were recorded.

The MR Imaging Workbench (Institute of Cancer Research) software calculates the relative signal intensity change due to contrast medium accumulation in each pixel and converts this to changes in T1 relaxation rate by using methods described above. Contrast agent concentration is then calculated by using Equation (1). The kinetic parameters, transfer constant, and EES fractional volume were estimated by using the pharmacokinetic model described in the Appendix. The rate constant between the EES and blood plasma was also estimated. The assumptions and limitations of the model are also discussed in the Appendix. It should be noted that the interpretation of the transfer constant is dependent on underlying physiologic characteristics of tumor. If the delivery of the contrast medium to a tissue is insufficient (flow-limited situations or where vascular permeability is greater than inflow), then blood perfusion will be the dominant factor determining contrast agent kinetics and transfer constant approximates to tissue blood flow per unit volume. If tissue perfusion is sufficient and transport out of the vasculature does not deplete intravascular contrast medium concentration (non–flow limited situations), then the transfer constant equals the permeability surface area product (20).

Statistical Analysis
Kinetic parameters (transfer constant, rate constant_, EES fractional volume, and maximum contrast medium concentration) for each pixel from the whole tumor region of interest were analyzed by using a statistical software package (StatsDirect, Cambridge, UK). Multisection data were combined and analysis was performed on enhancing pixels with transfer constant values ranging from 0.001 to 5.0 min^–1. Pixels with transfer constant values lower than 0.001 min^–1 were assumed to represent nonviable (necrotic) areas, and when transfer constant values were above 5.0 min^–1, the pathophysiologic correlates were uncertain. The numbers of nonenhancing pixels and enhancing pixels where our model did not fit the data that we observed were counted (modeling failures). Descriptive statistics were obtained on the histogram distributions of each kinetic parameter (26). MR imaging size changes and the median value and range of the histogram for each kinetic parameter were compared by using nonparametric methods (Mann-Whitney U test for two independent random samples and Kruskal-Wallis test for multiple independent random samples). Parameter changes on follow-up studies after the first and second cycle of chemotherapy were compared with the baseline study and correlated with clinical and histopathologic findings. The level was set at 5% (two tailed). Although this was an observational study, receiver operating characteristic (ROC) analysis of tumor size and transfer constant changes were used to try to identify patients in whom no benefit was gained from chemotherapy (ie, patients with macroscopic residual tumor or extensive microscopic infiltrating cancer [pathologic subgroup A1 responses]).

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION APPENDIX ADVANCES IN KNOWLEDGE References

 
Clinical and Histopathologic Responses
There were no patient characteristics, including patient age, menstrual state, estrogen-receptor staining, or tumor stage, that predicted response. A summary of clinical and pathologic responses is given in Table 2. After neoadjuvant chemotherapy had been completed, in four patients there was clinical progression with treatment (ie, clinically progressive disease), and six patients were considered to have clinically stable disease (both groups were classified as clinical nonresponders). There were 15 clinical responders (three with complete response, two with minimal residual disease, and 10 with partial response). Pathologic response assessments were not available in three patients because they did not undergo surgical excision. For pragmatic reasons, one patient (patient 9 in Table 3) was subsequently classified into group A1 because of clinically stable disease; two patients (patients 22 and 25 in Table 3) were place in group B on the basis of clinically complete response in one and minimal residual disease in the other.

Evaluation after One Cycle of Chemotherapy
Three patients defaulted from the study, giving no reasons, and one patient was too ill to attend. Table 4 and Figure 1 summarize clinical and pathologic responses by size (in 21 patients) and transfer constant changes (in 15 patients). Kinetic parameters could not be estimated in six patients because of the following: poor bolus injection of contrast medium (patient 11), MR examination stopped due to discomfort (patient 19), or technical failures (patients 6, 14, 21, and 22).

View larger version (14K): [in this window] [in a new window] [Download PPT slide]   Figure 1a: Box plots show changes in (a) size and (b) transfer constant range values during chemotherapy correlated with pathologic response. White and gray boxes show change after one and two cycles of chemotherapy, respectively. All values are normalized to baseline. The boundaries of the box show 25th and 75th percentiles, and a line within the box marks the median value. Whiskers show 90th and 10th percentiles. Outliers () are also shown. Subgroup A1 = gross residual disease at pathologic evaluation with clinical progression or stable disease, subgroup A2 = gross residual disease at pathologic evaluation with clinical partial or complete response or minimal residual disease, group B = pathologic microscopic disease or complete response.

View larger version (17K): [in this window] [in a new window] [Download PPT slide]   Figure 1b: Box plots show changes in (a) size and (b) transfer constant range values during chemotherapy correlated with pathologic response. White and gray boxes show change after one and two cycles of chemotherapy, respectively. All values are normalized to baseline. The boundaries of the box show 25th and 75th percentiles, and a line within the box marks the median value. Whiskers show 90th and 10th percentiles. Outliers () are also shown. Subgroup A1 = gross residual disease at pathologic evaluation with clinical progression or stable disease, subgroup A2 = gross residual disease at pathologic evaluation with clinical partial or complete response or minimal residual disease, group B = pathologic microscopic disease or complete response.

View larger version (77K): [in this window] [in a new window] [Download PPT slide]   Figure 2: Images show transfer constant changes in 48-year-old postmenopausal woman with grade 2 infiltrating ductal-lobular carcinoma of the right breast responding to mitoxantrone and methotrexate chemotherapy. Columns show anatomic subtraction images (obtained by subtracting MR image acquired at 100 seconds after contrast agent administration from baseline image), corresponding transfer constant maps (color map displays range of 0–1 min–1), and histograms from pixel data. Rows show data before treatment and after one and two cycles of mitoxantrone and methotrexate chemotherapy, respectively. After one cycle of treatment (middle row), a decrease in the transfer constant median and range is seen (47% and 45%, respectively), compared with a 28% decrease in tumor size. After two treatments (bottom row), further decrease in transfer constant median and range is seen (80% and 75%, respectively) on the transfer constant histogram, compared with a 49% decrease in tumor size.

View larger version (69K): [in this window] [in a new window] [Download PPT slide]   Figure 3: Images show changes in transfer constant in 44-year-old perimenopausal woman with a grade 3 infiltrating ductal carcinoma of the left breast not responding to mitoxantrone and methotrexate chemotherapy. Columns show anatomic subtraction images (obtained by subtracting MR image acquired at 100 seconds after contrast agent administration from baseline image), corresponding transfer constant maps (color map displays range of 0–1 min–1), and histograms from pixel data. Rows show data before treatment and after one and two cycles of mitoxantrone and methotrexate chemotherapy, respectively. After one cycle of treatment (middle row), an increase in the transfer constant median and range is seen (57% and 34%, respectively), compared with a 10% decrease in tumor size. After two treatments (bottom row), a further increase in the transfer constant median and range is seen (186% and 181%, respectively) on the transfer constant histogram, compared with a 11% increase in tumor size.

View larger version (22K): [in this window] [in a new window] [Download PPT slide]   Figure 4: Scatter plot of correlation of pretreatment transfer constant values with transfer constant change after one cycle of treatment. Scatter plot and linear regression analysis demonstrate that changes in transfer constant values after one treatment are significantly related to transfer constant before treatment begins (r2 = 0.55, P = .001).

Evaluation after Two Cycles of Chemotherapy
Nine patients defaulted from the study, giving no reasons, and one patient was removed from the study because the patient became ill. Table 5 and Figure 1 summarize clinical and pathologic responses by size (in 15 patients) and transfer constant changes (in 13 patients) after two cycles of treatment. Kinetic parameters could not be estimated in two patients because of technical failures (patients 14 and 21).

Correlation with final pathologic response.—In regard to pathologic response, there were nine nonresponders (subgroups A1 and A2) and six responders (group B). Statistically significant changes in tumor size changes were noted between these groups (Kruskal-Wallis test, P = .02). Results in four of five patients in subgroup A1 showed no change or an increase in tumor size compared with results in three of four patients in subgroup A2, which showed a decrease in size. In five of six patients in group B, there was a decrease in tumor size (Fig 1). Changes in the transfer constant range were statistically different between the pathologic response categories (P = .04). Again, changes in leakage space, maximum contrast medium accumulation, and rate constant did not correlate with pathologic tumor response.

Identification of Nonresponders
ROC analysis was used to determine the degree of size and transfer constant range change that would identify patients in whom no benefit was gained from neoadjuvant chemotherapy (ie, pathologic subgroup A1 patients) (Fig 5). An increase, no change, or a reduction of less than 15% in size after one cycle of treatment would have had 100% sensitivity in the identification of these patients but would have included two of 14 patients who would have benefited from neoadjuvant chemotherapy (specificity, 85%; area under ROC curve, 0.90). After two cycles of treatment, this size cutoff value had similar accuracy (sensitivity, 100%; specificity, 90%; area under ROC curve, 0.93). An increase, no change, or a reduction of less than 11% in transfer constant range after one treatment would have enabled identification of four of five evaluable patients in subgroup A1 (sensitivity, 80%) but would have included three of 10 patients who would have benefited from neoadjuvant chemotherapy (specificity, 70%; area under ROC curve, 0.76). After two cycles of treatment, this transfer constant range cutoff value had an improved accuracy (sensitivity, 100%; specificity, 75%; area under ROC curve, 0.94). Thus, transfer constant range and size changes are equally accurate (equal area under ROC curves) in helping identify patients in whom no clinical or pathologic benefit would be gained after two cycles of treatment.

View larger version (21K): [in this window] [in a new window] [Download PPT slide]   Figure 5a: ROC analysis for predicting patients who would not benefit from neoadjuvant chemotherapy. (a) For size change analysis, after one cycle of treatment (area, 0.90), an increase, no change, or a decrease in size of less than 15% results in 100% sensitivity for identifying subgroup A1 patients (specificity, 85%). After two cycles of treatment (area, 0.93), this cutoff value had similar accuracy (sensitivity, 100%; specificity, 90%). (b) For transfer constant range analysis, after one cycle of treatment (area, 0.76), an increase, no change, or a decrease of less than 11% results in a 80% sensitivity for identifying subgroup A1 patients (specificity, 70%). After two cycles of treatment (area, 0.94), this cutoff value had 100% sensitivity and 75% specificity.

View larger version (21K): [in this window] [in a new window] [Download PPT slide]   Figure 5b: ROC analysis for predicting patients who would not benefit from neoadjuvant chemotherapy. (a) For size change analysis, after one cycle of treatment (area, 0.90), an increase, no change, or a decrease in size of less than 15% results in 100% sensitivity for identifying subgroup A1 patients (specificity, 85%). After two cycles of treatment (area, 0.93), this cutoff value had similar accuracy (sensitivity, 100%; specificity, 90%). (b) For transfer constant range analysis, after one cycle of treatment (area, 0.76), an increase, no change, or a decrease of less than 11% results in a 80% sensitivity for identifying subgroup A1 patients (specificity, 70%). After two cycles of treatment (area, 0.94), this cutoff value had 100% sensitivity and 75% specificity.

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION APPENDIX ADVANCES IN KNOWLEDGE References

The effects of chemotherapy on breast cancer angiogenesis, microvascular blood flow, and permeability are poorly documented in the literature. There is evidence from preclinical studies that chemotherapeutic drugs cause acute increases as well as decreases in tumor blood flow. Paclitaxel has been reported to have antiangiogenic tumor activity, but it increases human vessel permeability (27). Chemotherapeutic drugs have also been noted to have antiangiogenic activity when used in low-dose continuous schedules (28). Makris et al (29) recently reported that fewer tumor microvessels were seen in breast cancer patients treated with chemoendocrine therapy compared with untreated patients. However, they reported no differences in microvessel density counts between responders and nonresponders.

The causes of the changes in transfer constant that we observed are likely to be multifactorial, relating to both microvessel density and function. A possible explanation is that successful chemotherapy causes reductions in tumor angiogenic factors, including VEGF. VEGF is a strong stimulus of tumor neoangiogenesis and a potent tissue permeability factor in breast cancer (30). With respect to the vasculature, it is clear that VEGF is also required for vascular homeostasis and maintains the high fraction of immature vessels within most tumors. Immature vessels (those without investing pericytes and/or smooth muscle cells) are highly dependent on exogenous survival factors, including VEGF (31), and these immature vessels are by definition the most permeable. Cytotoxic tumor cell death would result in a reduction of tissue VEGF production and hence apoptosis of endothelial cells in immature vessels. Endothelial cell apoptosis in immature vessels as a response to VEGF withdrawal has been noted in a number of treatments, including antiangiogenic and hormonal treatments (32–34). The initial selective decreases in the highest transfer constant values can be explained by the hypothesis that tissue drug delivery is best to areas of highest microvessel permeability. With high drug concentration, tumor cell death would result in VEGF withdrawal and apoptosis of immature vessels, leading to a decrease in transfer constant.

On the other hand, tumor resistance to chemotherapy would result in ongoing production of angiogenic factors that maintain or increase the proportion of immature vessels. The latter may explain the lack of change or increase in transfer constant values we observed in nonresponders; other authors have also noted that persistent abnormal enhancement is also associated with the presence of active, nonresponsive disease (35–38). Tamoxifen is also known to have an antiangiogenic effect in xenografts (39), but our patient sample was too limited to analyze its antivascular effects in human breast cancer.

We also noted that the change in transfer constant values correlated inversely with the pretreatment value (r² = 0.55, P = .001). This was more marked among responders (r² = 0.56, P = .02) than among nonresponders (r² = 0.65, P = .05). This finding is similar to an observation made by Reddick et al (38), who also showed a similar finding in the rate constant in patients with soft-tissue and bone sarcomas treated with neoadjuvant chemotherapy. They did not distinguish between responders and nonresponders but noted that both the initial rate constant values and the change in rate constant values significantly predicted both pathologic response and disease-free survival. These findings may also be explained on the basis of increased drug access to the tumor interstitium in areas of high microvessel permeability (represented by the transfer constant in our patients with breast cancer or rate constant in patients with bone sarcoma). Thus, reductions in transfer constant may be greater in patients with tumors that are chemosensitive.

Size Changes and Tumor Response
In six (86%) of seven nonresponders, no change or an increase in tumor size was seen after one cycle of chemotherapy, and in five (83%) of six patients, an increase in size was seen after two cycles. A decrease in tumor size was seen in 12 (86%) of 14 responders after one cycle of treatment compared with eight (89%) of nine patients after two cycles of treatment. Thus, decrease in tumor size is the best early predictor of eventual tumor response. We noted that concordance between tumor size and transfer constant range change was best in responders. Interestingly, an increase in transfer constant range can predict nonresponsiveness to chemotherapy after one treatment cycle even in the absence of size increase, and we have shown that size and transfer constant range changes were equally accurate in the identification of patients in whom no clinical or pathologic benefit would be neoadjuvant chemotherapy (sensitivity, specificity, and area under ROC curve: 100%, 90%, and 0.93, respectively, for size change and 100%, 75%, and 0.94, respectively, for transfer constant range change) after two cycles of treatment.

Study Limitations
No direct histologic validation of angiogenic characteristics was obtained during neoadjuvant chemotherapy. Obtaining such data would require acquisition of core biopsy specimens, which is not normal practice in an observational study and would have been burdensome for patients. Core biopsy is also subject to sampling errors and would also have incited healing (angiogenic) responses that would compromise follow-up dynamic contrast-enhanced MR imaging. Histologic correlation of final response was obtained in 22 of 25 patients at the end of treatment. We were also unable to exactly coregister pretreatment studies with those obtained during treatment. This occurred because tumors changed in size with treatment, and automated registration software is not available for pliable organs such as the breast. A radiologist with 3 years of breast MR imaging experience chose the plane of the dynamic contrast-enhance MR image acquisition at each examination to minimize such errors. Errors relating to region of interest placement were minimized by evaluating anatomic images at consensus review, by performing both pre- and posttreatment kinetic analysis at the same session (thus reducing intraobserver variability), and by recording the exact sites of the regions of interest. The limitations and uncertainties of our modeling techniques are discussed in the Appendix. This was an observational study that was not specifically designed to have adequate power to assess nonresponsiveness. This was because there were very limited data on the early dynamic contrast-enhanced MR imaging changes in breast cancer treated by means of neoadjuvant chemotherapy (35). Nevertheless, it was interesting to note the possibility that dynamic contrast-enhanced MR imaging may be able to be used to predict those patients in whom no clinical benefit would be gained from neoadjuvant chemotherapy. An appropriately powered follow-up study is currently underway to specifically address the utility of dynamic contrast-enhance MR imaging in this regard in a more uniformly treated population.

It is to be noted that the number of patients undergoing MR imaging decreased after successive cycles of chemotherapy. As this was a voluntary observational study, there was no requirement for patients to undergo MR examinations. This was mainly because the patients believed that the examination was uncomfortable and placed an additional burden in excess of what they were prepared to undergo while receiving chemotherapy. Statistically, this was handled by performing group analyses. With such a small observational data set, we did not have power in this study to assess all the possible associations between MR imaging and patient outcomes taking into account variables such as age, cancer type, histologic grade, and treatments performed.

Low-molecular-weight extracellular MR contrast agent kinetics can be used to quantify the early microvascular effects of chemoendocrine therapy on breast cancer. Our data show that transfer constant changes mirrored tumor response, and this parameter may thus inform on pathophysiologic mechanisms underlying the angiogenic response of tumors to chemotherapy. From the clinical perspective, after one or two cycles of neoadjuvant chemotherapy, tumor size is best for helping predict eventual tumor regression, but changes in transfer constant are equally accurate in helping identify patients in whom no clinical or pathologic benefit would be gained after two cycles of treatment. Dynamic contrast-enhanced MR imaging thus provides information on functional angiogenic activity that is complementary to conventional methods of assessing breast cancer response to neoadjuvant chemotherapy.

	APPENDIX

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION APPENDIX ADVANCES IN KNOWLEDGE References

 
A compartmental model was used to model the concentration of gadopentetate dimeglumine (Magnevist; Schering Health Care) in tissue with time following a bolus intravenous injection (21). This model allows estimation of the transfer constant, K^trans (in min^–1), of gadopentetate dimeglumine between the blood plasma and the EES, the EES fractional volume, _e, and the rate constant, k_ep, between EES and blood plasma: k_ep = K^trans/_e. The contrast agent diffuses freely from the blood plasma into the EES through microscopic defects in the capillary walls. The blood plasma compartment has a volume V_p and a contrast agent concentration C_p, and the EES (also known as the interstitial space) has a volume V_e and contrast agent concentration C_e. The rate equation, which relates tissue concentration of contrast agent, C_t, to arterial plasma contrast agent concentration, C_p, assuming that the plasma volume, V_p, is small, is given as:

(A1)

The model assumes a short bolus injection time, instant mixing, and fast exchange of all mobile protons within the tissue. It also assumes that transport of the contrast agent out of the vasculature is a negligible perturbation on the blood pool concentration. In this study, we assumed that the time cycle of contrast agent concentration in the plasma is a biexponential decay and that it equals that as measured by Weinmann et al (40) for all subjects:

(A2)

where D is the dose of gadopentetate dimeglumine (in millimoles per liter per kilogram of body weight), a₁ = 3.99 kg/L and a₂ = 4.78 kg/L are amplitudes, and m₁ = 0.144 min^–1 and m₂ = 0.0111 min^–1 are rate constants.

By using an original software package (25) (Magnetic Resonance Imaging Workbench), image intensities were converted into T1 values by using the methods described. Contrast agent concentration is calculated by using Equation (A1). From the resultant concentration-time curve, K^trans and _e are calculated for each pixel by fitting the theoretic model to each pixel's concentration-time data by using a routine based on the Levenburg-Marquardt algorithm. The results may be viewed as color-coded overlays on the anatomic images.

There are uncertainties with regard to the reliability of kinetic parameter estimates derived from the application of tracer kinetic models to dynamic contrast-enhanced MR data (41–43). These derive from assumptions implicit in kinetic models and those for the measurement of tissue contrast agent concentration. Our application of the Tofts model used a standard description of the time varying blood concentration of contrast agent, also called the input function (40), and assumed that the supply of contrast medium is not perfusion limited and that tissue blood volume contributes negligible signal compared with that arising from contrast medium in the interstitial space. Reliable methods for measuring arterial input function for routine dynamic contrast-enhanced MR imaging studies are only now emerging and were not widely available at the time when our study was conducted. Buckley (44) has suggested that the application of commonly accepted models and their respective model-based assumptions to dynamic contrast-enhanced MR data leads to systematic overestimation of transfer constant in tumors. However, it is difficult to be certain about how accurately model-based kinetic parameter estimates compare with the physiologic parameter that they purport to measure, particularly as there is no reliable clinical reference standard. Nevertheless, a large body of scientific literature supports the clinical utility of dynamic contrast-enhanced MR imaging techniques in this regard (13), and quantitative kinetic parameters thus derived can provide insights into underlying tissue pathophysiologic processes as illustrated in this report.

	ADVANCES IN KNOWLEDGE

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION APPENDIX ADVANCES IN KNOWLEDGE References

 

• Low-molecular-weight extracellular MR contrast agent kinetics can be used to quantify early microvascular effects of neoadjuvant chemoendocrine therapy on locally advanced breast cancer.
  
• Substantial changes in lesion size appear to precede changes in kinetic parameters.
  
• After two cycles of treatment, size and transfer constant changes appear equally powerful at predicting nonresponsiveness to therapy.
  

	ACKNOWLEDGMENTS

 
We thank Andy Norman, PhD, Royal Marsden NHS Trust, London, UK, for his expert statistical advice.

	FOOTNOTES

 

Abbreviations: EES = extravascular extracellular space • FLASH = fast low-angle shot • ROC = receiver operating characteristic • TE = echo time • TR = repetition time • VEGF = vascular endothelial growth factor

² Current address: Department of Oncology, Mount Vernon Hospital, Middlesex, United Kingdom.

³ Current address: Clinical Age Research Unit, King's College Hospital, School of Medicine and Dentistry, London, United Kingdom.

Author contributions: Guarantor of integrity of entire study, J.E.H.; study concepts, A.R.P., J.E.H., C.H., T.P., M.O.L.; study design, A.R.P., J.E.H., T.P., J.S., M.O.L.; literature research, A.R.P., C.H., L.A., A.M.; clinical studies, A.R.P., L.A., T.P.; experimental studies, J.S., C.H.; data acquisition, J.S., L.A.; data analysis/interpretation, C.H., A.R.P.; statistical analysis, A.R.P., C.H.; manuscript preparation, A.R.P., C.H., A.M.; manuscript definition of intellectual content, A.R.P., M.O.L., J.E.H.; manuscript editing, A.R.P., C.H., L.A., T.P., A.M., M.O.L., J.E.H.; manuscript revision/review, C.H., L.A., A.M., M.O.L., J.E.H.; manuscript final version approval, M.O.L., J.E.H.

Authors stated no financial relationship to disclose.

	References

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION APPENDIX ADVANCES IN KNOWLEDGE References

 

1. Fisher B, Bryant J, Wolmark N, et al. Effect of preoperative chemotherapy on the outcome of women with operable breast cancer. J Clin Oncol 1998;16:2672–2685.[Abstract]
2. Makris A, Powles TJ, Ashley SE, et al. A reduction in the requirements for mastectomy in a randomized trial of neoadjuvant chemoendocrine therapy in primary breast cancer. Ann Oncol 1998;9:1179–1184.[Abstract/Free Full Text]
3. Powles TJ, Hickish TF, Makris A, et al. Randomized trial of chemoendocrine therapy started before or after surgery for treatment of primary breast cancer. J Clin Oncol 1995;13:547–552.[Abstract/Free Full Text]
4. Mauriac L, Durand M, Avril A, Dilhuydy JM. Effects of primary chemotherapy in conservative treatment of breast cancer patients with operable tumors larger than 3 cm: results of a randomized trial in a single centre. Ann Oncol 1991;2:347–354.[Abstract/Free Full Text]
5. Machiavelli MR, Romero AO, Perez JE, et al. Prognostic significance of pathological response of primary tumor and metastatic axillary lymph nodes after neoadjuvant chemotherapy for locally advanced breast carcinoma. Cancer J Sci Am 1998;4:125–131.[Medline]
6. Chollet P, Amat S, Cure H, et al. Prognostic significance of a complete pathological response after induction chemotherapy in operable breast cancer. Br J Cancer 2002;86:1041–1046.[CrossRef][Medline]
7. Heys SD, Eremin JM, Sarkar TK, Hutcheon AW, Ah-See A, Eremin O. Role of multimodality therapy in the management of locally advanced carcinoma of the breast. J Am Coll Surg 1994;179:493–504.[Medline]
8. Fiorentino C, Berruti A, Bottini A, et al. Accuracy of mammography and echography versus clinical palpation in the assessment of response to primary chemotherapy in breast cancer patients with operable disease. Breast Cancer Res Treat 2001;69:143–151.[CrossRef][Medline]
9. Feldman LD, Hortobagyi GN, Buzdar AU, Ames FC, Blumenschein GR. Pathological assessment of response to induction chemotherapy in breast cancer. Cancer Res 1986;46:2578–2581.[Abstract/Free Full Text]
10. Drew PJ, Kerin MJ, Mahapatra T, et al. Evaluation of response to neoadjuvant chemoradiotherapy for locally advanced breast cancer with dynamic contrast-enhanced MRI of the breast. Eur J Surg Oncol 2001;27:617–620.[CrossRef][Medline]
11. Abraham DC, Jones RC, Jones SE, et al. Evaluation of neoadjuvant chemotherapeutic response of locally advanced breast cancer by magnetic resonance imaging. Cancer 1996;78:91–100.[CrossRef][Medline]
12. Rieber A, Brambs HJ, Gabelmann A, Heilmann V, Kreienberg R, Kuhn T. Breast MRI for monitoring response of primary breast cancer to neo-adjuvant chemotherapy. Eur Radiol 2002;12:1711–1719.[CrossRef][Medline]
13. Padhani AR. MRI for assessing antivascular cancer treatments. Br J Radiol 2003;76(suppl 1):S60–S80.[Abstract/Free Full Text]
14. Stomper PC, Winston JS, Herman S, Klippenstein DL, Arredondo MA, Blumenson LE. Angiogenesis and dynamic MR imaging gadolinium enhancement of malignant and benign breast lesions. Breast Cancer Res Treat 1997;45:39–46.[CrossRef][Medline]
15. Buckley DL, Drew PJ, Mussurakis S, Monson JR, Horsman A. Microvessel density of invasive breast cancer assessed by dynamic Gd-DTPA enhanced MRI. J Magn Reson Imaging 1997;7:461–464.[Medline]
16. Knopp MV, Weiss E, Sinn HP, et al. Pathophysiologic basis of contrast enhancement in breast tumors. J Magn Reson Imaging 1999;10:260–266.[CrossRef][Medline]
17. Fleming ID, Cooper JS, Henson DE, et al. AJCC cancer staging manual. Philadelphia, Pa: Lippincott-Raven, 1997.
18. Hayward JL, Carbone PP, Heuson JC, Kumaoka S, Segaloff A, Rubens RD. Assessment of response to therapy in advanced breast cancer: a project of the Program on Clinical Oncology of the International Union Against Cancer, Geneva, Switzerland. Cancer 1977;39:1289–1294.[CrossRef][Medline]
19. Honkoop AH, van Diest PJ, de Jong JS, et al. Prognostic role of clinical, pathological and biological characteristics in patients with locally advanced breast cancer. Br J Cancer 1998;77:621–626.[Medline]
20. Tofts PS. Modeling tracer kinetics in dynamic Gd-DTPA MR imaging. J Magn Reson Imaging 1997;7:91–101.[Medline]
21. Tofts PS, Brix G, Buckley DL, et al. Estimating kinetic parameters from dynamic contrast-enhanced T(1)-weighted MRI of a diffusable tracer: standardized quantities and symbols. J Magn Reson Imaging 1999;10:223–232.[CrossRef][Medline]
22. Donahue KM, Burstein D, Manning WJ, Gray ML. Studies of Gd-DTPA relaxivity and proton exchange rates in tissue. Magn Reson Med 1994;32:66–76.[Medline]
23. Parker GJ, Suckling J, Tanner SF, et al. Probing tumor microvascularity by measurement, analysis and display of contrast agent uptake kinetics. J Magn Reson Imaging 1997;7:564–574.[Medline]
24. Parker GJ, Baustert I, Tanner SF, Leach MO. Improving image quality and T(1) measurements using saturation recovery turboFLASH with an approximate K-space normalisation filter. J Magn Reson Imaging 2000;18:157–167.
25. Parker GJ, Suckling J, Tanner SF, Padhani AR, Husband JE, Leach MO. MRIW: parametric analysis software for contrast-enhanced dynamic MR imaging in cancer. RadioGraphics 1998;18:497–506.[Abstract]
26. Hayes C, Padhani AR, Leach MO. Assessing changes in tumour vascular function using dynamic contrast-enhanced magnetic resonance imaging. NMR Biomed 2002;15:154–163.[CrossRef][Medline]
27. Belotti D, Vergani V, Drudis T, et al. The microtubule-affecting drug paclitaxel has antiangiogenic activity. Clin Cancer Res 1996;2:1843–1849.[Abstract]
28. Browder T, Butterfield CE, Kraling BM, et al. Antiangiogenic scheduling of chemotherapy improves efficacy against experimental drug-resistant cancer. Cancer Res 2000;60:1878–1886.[Abstract/Free Full Text]
29. Makris A, Powles TJ, Kakolyris S, Dowsett M, Ashley SE, Harris AL. Reduction in angiogenesis after neoadjuvant chemoendocrine therapy in patients with operable breast carcinoma. Cancer 1999;85:1996–2000.[Medline]
30. Dvorak HF, Brown LF, Detmar M, Dvorak AM. Vascular permeability factor/vascular endothelial growth factor, microvascular hyperpermeability, and angiogenesis. Am J Pathol 1995;146:1029–1039.[Abstract]
31. Darland DC, D'Amore PA. Blood vessel maturation: vascular development comes of age. J Clin Invest 1999;103:157–158.[Medline]
32. Benjamin LE, Golijanin D, Itin A, Pode D, Keshet E. Selective ablation of immature blood vessels in established human tumors follows vascular endothelial growth factor withdrawal. J Clin Invest 1999;103:159–165.[Medline]
33. Jain RK, Safabakhsh N, Sckell A, et al. Endothelial cell death, angiogenesis, and microvascular function after castration in an androgen-dependent tumor: role of vascular endothelial growth factor. Proc Natl Acad Sci U S A 1998;95:10820–10825.[Abstract/Free Full Text]
34. Shaheen RM, Davis DW, Liu W, et al. Antiangiogenic therapy targeting the tyrosine kinase receptor for vascular endothelial growth factor receptor inhibits the growth of colon cancer liver metastasis and induces tumor and endothelial cell apoptosis. Cancer Res 1999;59:5412–5416.[Abstract/Free Full Text]
35. Knopp MV, Brix G, Junkermann HJ, Sinn HP. MR mammography with pharmacokinetic mapping for monitoring of breast cancer treatment during neoadjuvant therapy. Magn Reson Imaging Clin N Am 1994;2:633–658.[Medline]
36. van der Woude HJ, Bloem JL, Verstraete KL, Taminiau AH, Nooy MA, Hogendoorn PC. Osteosarcoma and Ewing's sarcoma after neoadjuvant chemotherapy: value of dynamic MR imaging in detecting viable tumor before surgery. AJR Am J Roentgenol 1995;165:593–598.[Abstract/Free Full Text]
37. Mayr NA, Yuh WT, Arnholt JC, et al. Pixel analysis of MR perfusion imaging in predicting radiation therapy outcome in cervical cancer. J Magn Reson Imaging 2000;12:1027–1033.