- Email: [email protected]
Evaluation of the reproducibility of a protocol for the pharmacokinetic study of breast tumors by dynamic magnetic resonance imaging
Radiología. 2015;57(1):44---49
www.elsevier.es/rx
ORIGINAL REPORT
Evaluation of the reproducibility of a protocol for the pharmacokinetic study of breast tumors by dynamic magnetic resonance imaging夽 J. Etxano a,∗ , A. García-Lallana Valbuena b , I. Antón Ibᘠnez c , A. Elizalde a , a d d L. Pina , J. García-Foncillas , V. Boni a
Departamento de Radiología, Clínica Universidad de Navarra, Pamplona, Navarra, Spain Departamento de Radiología, Hospital San Juan de Dios, Santurce, Vizcaya, Spain c Progenika Biopharma S.A, Derio, Vizcaya, Spain d Departamento de Oncología, Clínica Universidad de Navarra, Pamplona, Navarra, Spain b
Received 4 September 2012; accepted 26 January 2013
KEYWORDS Breast cancer; Magnetic resonance imaging; Pharmacokinetics
Abstract Objective: To evaluate the reproducibility of a protocol for dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) for the pharmacokinetic study of breast tumors. Materials and methods: We carried out this prospective study from October 2009 through December 2009. We studied 12 patients with stages ii--iii invasive breast cancer without prior treatment. Our center’s research ethics committee approved the study. The 12 patients underwent on two consecutive days DCE-MRI with a high temporal resolution protocol (21 acquisitions/minute). The data obtained in an ROI traced around the largest diameter of the tumor (ROI 1) and in another ROI traced around the area of the lesion’s highest Ktrans intensity (ROI 2) were analyzed separately. We used parametric and nonparametric statistical tests to study the reproducibility and concordance of the principal pharmacokinetic variables (Ktrans , Kep , Ve and AUC90 ). Results: The correlations were very high (r > .80; P < .01) for all the variables for ROI 1 and high (r = .70---.80; P < .01) for all the variables for ROI 2, with the exception of Ve both in ROI 1 (r = .44; P = .07) and in ROI 2 (r = .13; P = .235). There were no statistically significant differences between the two studies in the values obtained for Ktrans , Kep and AUC90 (P > .05 for each), but there was a statistically significant difference between the two studies in the values obtained for Ve in ROI 2 (P = .008). Conclusions: The high temporal resolution protocol for DCE-MRI used at our center is highly reproducible for the principal pharmacokinetic constants of breast. © 2012 SERAM. Published by Elsevier España, S.L.U. All rights reserved.
夽 Please cite this article as: Etxano J, García-Lallana Valbuena A, Antón Ibᘠnez I, Elizalde A, Pina L, García-Foncillas J, et al. Evaluación de la reproducibilidad de un protocolo de resonancia magnética dinámica para el estudio farmacocinético de los tumores de mama. Radiología. 2015;57:44---49. ∗ Corresponding author. E-mail address: [email protected] (J. Etxano).
2173-5107/$ – see front matter © 2012 SERAM. Published by Elsevier España, S.L.U. All rights reserved.
Pharmacokinetic study of breast tumors by dynamic magnetic resonance imaging
PALABRAS CLAVE Cáncer de mama; Imagen por resonancia magnética; Farmacocinética
45
Evaluación de la reproducibilidad de un protocolo de resonancia magnética dinámica para el estudio farmacocinético de los tumores de mama Resumen Objetivo: Evaluar la reproducibilidad de un protocolo de resonancia magnética dinámica (RMDC) con contraste para el estudio farmacocinético de los tumores de mama. Material y métodos: Estudio prospectivo realizado entre octubre y diciembre de 2009, que incluyó 12 pacientes con cáncer de mama infiltrante en estadios ii-iii sin tratamiento previo. Este trabajo fue aprobado por el Comité de Ética de Investigación de nuestro centro. A las 12 pacientes se les realizó 2 RM-DC en 2 días consecutivos con un protocolo de alta resolución temporal (21 adquisiciones/minuto). Se analizaron por separado los datos obtenidos en un ROI trazado alrededor del diámetro tumoral mayor (ROI 1) y otro que abarcaba la zona de mayor intensidad de Ktrans de la lesión (ROI 2). Se emplearon pruebas estadísticas paramétricas y no paramétricas para estudiar la reproducibilidad y concordancia de las principales variables farmacocinéticas (Ktrans , Kep , Ve y AUC90 ). Resultados: Las correlaciones fueron muy altas (r > 0,80; p < 0,01) en todas las variables del ROI 1, y altas (r = 0,70---0,80; p < 0,01) en todas las del ROI 2, excepto en Ve tanto en el ROI 1 (r = 0,44; p = 0,07) como en el ROI 2 (r = 0,13; p = 0,235). No hubo diferencias estadísticamente significativas entre los 2 estudios para los valores obtenidos de Ktrans , Kep y AUC90 (p > 0,05 para todas ellas), pero sí que las hubo para Ve en el ROI 2 (p = 0,008). Conclusión: El protocolo de alta resolución temporal de RM-DC de nuestro centro es muy reproducible para las principales constantes farmacocinéticas de los tumores de mama. © 2012 SERAM. Publicado por Elsevier España, S.L.U. Todos los derechos reservados.
Introduction
Materials and methods
The dynamic magnetic resonance imaging (DMR) with IV contrast is one image modality more and more widely used in daily practice. Its indication in high risk patients, the preoperative staging of breast cancer, the hidden breast cancer, the detection of relapses and for the evaluation of the neoadjuvant chemotherapy1---3 are very well known. Because it is based on neoangiogenesis DMR allows the functional assessment of tumors different than conventional modalities---mammography and ultrasound that are only morphological. The actual protocols have a high spatial (with cutting thickness close to 1 mm) and temporal resolution (acquisition in approximately 60 s) that allow us to do semiquantitative dynamic studies based on the curves of signal/time intensity. However if temporal resolution is increased the vascular patency can be determined by quantifying pharmacokinetic parameters that describe the passing of contrast from vascular space to the tumor interstitial compartment.4 To that end it is also necessary to have a specific software capable of estimating these pharmacokinetic parameters. There are many protocols of pharmacokinetic study with DMR reported in the reference (Table 1).5---12 Some of them have even been modified along the study.10 This technical variability can also make the value of pharmacokinetic parameters change---something necessary to study the reproducibility of different protocols. The goal of this study is to evaluate the reproducibility and concordance of data obtained through a DMR protocol optimized for the pharmacokinetic study of stages ii--iii-breast tumors after the administration of neoadjuvant chemotherapy.
Patients Between October and December 2009 we performed one prospective study in our center including 12 treatment naive patients with confirmed histological diagnosis of infiltrating stages ii--iii breast cancer. These 12 patients were part of a phase 2 multicenter clinical trial including 12 hospitals (clinical trial code ML22197/2009-01) sponsored by F. Hoffman---La Roche. This trial enrolled 76 patients and assessed the predictive markers to treatment response with the antiangiogenic drug bevacizumab (Avastin® , F. Hoffman---La Roche, Basel, Switzerland) and neoadjuvant chemotherapy (docetaxel plus adriamicine) in patients with locally advanced breast cancer. In Table 2 the criteria for the inclusion of patients in the study can be seen. The study protocol was approved by the ethical committee of our hospital and all patients gave their informed written consent.
Technique To validate the DMR protocol two daily examinations were performed in two consecutive days with identical technical parameters and the patients did not follow any therapies between both examinations. The MRI protocol can be seen in Table 3. The DMR was performed in one 1.5 T resonance (Magnetom Symphony® , Siemens Healthcare, Erlangen, Germany). T1-weighed standard fl3D echo-spin gradient-enhanced images were modified in an attempt to improve its spatial resolution (150 measurements instead of 6 in a total of 7 min per test with approximately 21 acquisitions per
46 Table 1
J. Etxano et al. DMR protocols in the reference.
Study
Protocol
Ah See et al. (2008) Loo et al. (2011)6
5
Tateishi et al. (2012)7 Rahbar et al. (2012)8 Johansen et al. (2009)9 Partridge et al. (2010)10 El Khouli et al. (2011)11 Baar et al. (2009)12
1 1 2 1 1 2 1 1 2 1 1
View Sagittal Coronal Coronal Sagittal Axial Axial Sagittal Sagittal Sagittal Axial Axial
Voxel size a
0.8 mm 1.2 mm × 1.2 mm × 1.6 mm 1.1 mm × 1.1 mm × 1.2 mm 1.3 mm × 1.3 mm × 0.8 mm 0.9 mm × 1.1 mm × 2.2 mm 0.8 mm × 0.9 mm × 1.6 mm 1.9 mm × 1 mm × 3---4 mmb 1 mm × 1 mm × 2.2 mm 0.8 mm × 0.8 mm × 1.6 mm 1.3 mm × 1.3 mm × 5 mm 2.3 mm × 2.7 mm × 8 mm
Number of acquisitions
Total time
40 5 5 24 5 3 6---7 3 3 14 180
8 min 5 s 7 min 30 s 7 min 30 s 4 min 7 min 30 s 7 min 30 s 5 min 42 s-6 min 39 s 4 min 30 s 6 min 2 min 30 s 24 min
DMR: dynamic magnetic resonance imaging. a This value corresponds to the cutting thickness; the article says nothing about the matrix or the FOV used. b The protocol was applied to twelve (12) study patients with a cutting thickness of 3 mm and to some other 12 with a cutting thickness of 4 mm.
Table 2
Inclusion criteria.*
---------
Women between 18 and 70 years old Breast cancer histologically tested through skin biopsy Stage ii or iii, naïve to all prior therapies. Eastern Cooperative Oncology Group (ECOG) categories 0 to 1 --- Left ventricular ejection fraction (LVEF) ≥ 50% with no signs of cardiac failure
* Patients in the clinical trial of tumor response marker assessment after neoadjuvant therapy combined with one antiangiogenic drug and chemotherapy of stages ii--iiiinfiltrating breast tumors (clinical trial code: ML22197/2009-01).
minute). In turn the spatial resolution decreased (voxel size of 1.25 mm × 1.25 × 4 mm; 22 cuts) and the fat saturation pulse was eliminated. Given that the entire breast could not be studied through this protocol---especially large breasts Table 3 Dynamic magnetic resonance protocol with IV contrast. Magnetic resonance
Sequence View Acquisition Fat saturation TR/TE FoV Cutting thickness Matrix Acceleration Flip angle Number of measurements Number of cuts per block Total acquisition time
1.5 T (Magnetom Symphony® , Siemens Healthcare, Erlangen, Germany) T1-weighed standard fl3D echo gradient Coronal Caudo-cranial Without fat-sat pulse 4.3/1.29 320 mm 4 mm 256 × 256 ×2 GRAPPA 12◦ 150 2 7.08 min
in all patients---the tumor was found through an ultrasound and its location marked on their skin to be able to identify it in the location sequence of the MRI. After the acquisition of three images without IV contrast one bolus of 8 ml gadobutrol (Gadovist® , Bayer-Schering Pharma AG, Berlín, Germany) at 4 ml/s was automatically injected using one automatic injector after which 147 dynamic images could be obtained. Finally subtraction images of lesions were generated. Images were processed with one specific software (TISSUE 4D® , Siemens Healthcare, Erlangen, Germany) capable of evaluating the dynamic uptake of tumor contrast by calculating the pharmacokinetic values in the bicompartmental model of Tofts et al.13 . To assess the concentration of blood contrast or arterial input function (AIF), the software used offers three pre-established categories (quick, intermediate and slow). Given the high temporal resolution of our protocol, the category «quick» was used as the default one in all patients. The following values were analyzed: (1) Ktrans : constant of transendothelial transference. It measures the passing of contrast from the intravascular space to the extracellular space of tumor or in other words it is one measurement for the patency and blood flow of tumor vessels; (2) Kep : constant relating the passing of contrast from the tumor extravascular space to the intravascular one. It is the reverse constant to Ktrans , and is related to tumor washout; (3) Ve : fraction of tumor volume occupied by the extravascular space; and (4) AUC90 : area under the curve of signal intensity through time showing variation in tumor blood volume. It was estimated at 90 s which is the standard value of the software used. These parameters are related through the equation Kep = Ktrans /Ve . In each and every of the two DMR two regions of interest (ROI) were drawn --- one of them around the maximum diameter of lesion in subtraction images (ROI 1) and the other one with the smallest possible standardized size in the region of maximum Ktrans value (ROI 2) that we could identify by analyzing the Ktrans lesion map of colors without software (Fig. 1). In both cases we measured twice to assess the average values of Ktrans , Kep , Ve and AUC90 . We chose the Ktrans map because in the reference this value is
Pharmacokinetic study of breast tumors by dynamic magnetic resonance imaging
47
ROI 1
ROI 2
Roi1 Roi1
0.11
Roi1 Ktrans Kep Ve iAUC
: : : :
0.140 0.475 0.290 16.455 Median
0.136 0.472 0.287 16.044 Mean
0.007 0.011 0.009 0.674 std ktrans
0.03
Figure 1 On the subtracted image obtained in the coronal view (left upper image) the maximum tumor diameter could be identified and the ROI 1 was drawn marking out the area of the lesion. After obtaining the pharmacokinetic parameters the same proceeding was repeated to obtain the average values of the 2 measurements. Similarly using the Ktrans color map (right upper image) the ROI 2 was drawn on the Ktrans area of maximum intensity repeating the proceeding to obtain the average values.
considered to be one of the most relevant ones for the study of changes in neoangiogenesis and the response to neoadjuvant chemotherapy.5 Only one radiologist (LJP) highly experienced in performing DMR did the readings of all the cases.
Analysis To study the reproducibility and concordance of the outcomes obtained parametric statistical tests (interclass correlation coefficient and the Student’s t test for paired samples) and non-parametric tests were used (Spearman’s rho and Wilcoxon rank-sum test) with the software SPSS® v.15.0 (Chicago, Illinois, U.S.A.) Prior to this the normalcy of ROI 1 and ROI 2 variables through the Kolmogorov---Smirnov and Shapiro---Wilk tests was analyzed. P < 0.05 was considered significant.
Results The average maximum diameter of tumors through MRI was 35 mm with a rank between 14 and 100 mm. The four variables showed one normal distribution in ROI 1. In ROI 2 only in Kep and Ve the distribution was normal. The outcomes of the statistical analysis can be seen in Table 4. Very high correlations (r > 0.80; p < 0.01) between the 2 DMR were found in all ROI 1 variables and high correlations (r = 0.70---0.80; p < 0.01) were found in all ROI 2 variables with the exception of Ve both in ROI 1 (r = 0.44; p = 0.07) and ROI 2 (r = 0.13; p = 0.23). The average ROI 1 concordance was higher than that of ROI 2 (r = 0.77 vs r = 0.60). There were no statistically significant differences between both studies for the Ktrans , Kep and AUC90 values obtained (p > 0.05 for all) but there were statistically significant differences for Ve values obtained in ROI 2 (p = 0.008).
Discussion In our study we showed that with our high temporal resolution and reduced spatial resolution DMR protocol the concordance of the main pharmacokinetic variables is really good except for variable Ve . The analysis of angiogenesis through DMR in breast tumors is hot at the moment.7,8 The reduced tumor neovascularization is one of the most widely studied therapeutic targets during the last years and one of the most promising ones in the neoadjuvant therapy of breast cancer.14---16 Hence the importance of the analysis of evolutionary changes occurring in tumor reovascularization for which DMR is the most adequate DMR. In our study we did not find statistically significant differences in any variables between the two consecutive tests except for Ve in ROI 2. Concordance of all variables has been high (ROI 2), or very high (ROI 1), except for Ve values. It is possible that the fact of using a predetermined value of non-adjusted AIF for every study might have influenced the poor concordance found in both cases for Ve . The software used does not offer the possibility of performing a study prior to the contrast bolus for the analysis of real AIF. According to Leach et al.17 it is possible to measure the real AIF before the bolus but this increases the time and complexity of the study. Measuring it requires injecting a small amount of contrast media (around 10%) and one optimized image protocol. Another alternative is measuring AIF in one non-tumor region (e.g. muscular tissue) with known pharmacokinetics properties.17 However to analyze the importance of this study we need to highlight that the most important constants for the analysis of changes in tumor neovascularization are those that have to do with patency and the passing of contrast through the endothelial membrane (Ktrans , Kep ).18 We also performed the segmented analysis of data obtained thanks to one ROI drawn through the tumor
48 Table 4
J. Etxano et al. Statistical results.
Variable
Student’s t test for paired samples
CCI
Median 1
Median 2
P Value
Ratio
P Value
ROI 1 Ktrans Kep Ve AUC90
133.04 291.92 498.83 14.50
134.20 271.80 513.21 13.15
0.928 0.143 0.696 0.196
0.831 0.938 0.443 0.885
0.001 0.001 0.07 0.001
ROI 2 Kep Ve
401.33 610.80
374.00 733.46
0.530 0.008
0.758 0.135
0.001 0.236
Wilcoxon test
Variable
ROI 2 Ktrans AUC90
Spearman’s rho
Median 1
Median 2
P Value
Ratio
P Value
237.5 24.39
242.5 21.56
0.694 0.272
0.748 0.776
0.005 0.003
AUC90 : area under the curve of signal intensity throughout time. Calculated at 90 s; CCI: intraclass correlation coefficient; Kep : constant of contrast passing from the tumor extravascular space toward the intravascular space; Ktrans : constant of transendothelial transport from the intravascular space toward the extracellular tumor space; ROI 1: region of interest drawn around the maximum diameter of lesion in subtraction images; ROI 2: region of interest drawn in the region of maximum value of Ktrans identified through Ktrans color map of the lesion provided by the software; Ve : fraction of tumor volume occupied by extravascular space.
perimeter and another ROI including the area of maximum intensity of Ktrans . It is significant to see how in all cases the concordance of ROI 1 variables is greater than its corresponding ROI 2. This might be due to the fact that while drawing the edges of ROI 2 we rely upon the color map provided by Ktrans and the outlining of edges is more inaccurate. The data obtained in our study seem to indicate that quantification measured through one ROI drawn around the maximum diameter of lesion can be more reproducible than that in one ROI including the greater intensity region of Ktrans map, yet future research is need for confirmatory purposes. Having a DMR protocol highly reproducible is highly recommended since several studies have shown that the assessment of changes in pharmacokinetic constants (Ktrans , Kep , Ve and AUC90 ) can predict tumor response to neoadjuvant chemotherapy. Ah-See et al.5 said that early functional changes of tumor microvessels at the beginning of neoadjuvant chemotherapy can help us find out if we will have tumor response. Padhani et al.19 found that changes in pharmacokinetic constants were as sensitive and specific as changes in the tumor size to know if therapeutic response will ever happen. At present our institution is focused on one multicenter clinical trial aimed at evaluating the efficacy of neoadjuvant therapy with the antiangiogenic drug bevacizumab (Avastin® ) in stages ii--iii breast cancer through a serial study of the pharmacokinetic values obtained through DMR. Our work has some limitations. Firstly the low number of patients and secondly that there was only one radiologist in charge of all the readings---which means that inter-observer variability cannot be analyzed. However in the clinical trial protocol both parameters were agreed upon. Another limitation is the already commented absence of an individual calculation of the AIF to better determine the value of Ve .
For the study of pharmacokinetic parameters of smaller tumors the modification of the protocol with a greater spatial resolution and lower temporal resolution could be considered. In sum the protocol we presented for the assessment of the pharmacokinetic constants of breast tumors is highly reproducible for the values of Ktrans , Kep and AUC90 , with a higher concordance when assessing the ROI of the tumor maximum diameter. However it does not seem to be adequate for the Ve pharmacokinetic constant even though this result needs to be re-analyzed with a larger number of patients.
Ethical responsibilities Protection of humans and animals. Authors confirm that all proceedings and experiments followed relate to the committee of responsible human experimentation ethical rules and regulations in compliance with the World Medical Association and the Declaration of Helsinki. Data confidentiality. Authors confirm that the protocols of their centers have been followed on matters concerning the publishing of data from patients. They also confirm that all patients included in this study have been given enough information and handed over their written informed consent for their participation in this study. Right to privacy and informed consent. Authors confirm that they have obtained the written informed prior consent from patients and/or subjects appearing in this article. This document is in the possession of the corresponding author.
Pharmacokinetic study of breast tumors by dynamic magnetic resonance imaging
Author’s contributions 1. 2. 3. 4. 5. 6. 7. 8. 9.
Manager of the integrity of the study: JE and LP. Original Idea of the Study: JE and LP. Study Design: JE, LP and JGF. Data Mining: JE, LP, JGF and VB. Data Analysis and Interpretation: JE and LP. Statistical Analysis JE, IA and AGL. Reference Search: JE, AGL, AE and LP. Writing: JE, IA, LP, AE, AGL and LP. Manuscript critical review with intellectually relevant contributions: IA, AGL, JGF and VB. 10. Final Version Approval: JE, AGL, IA, AE, JGF, VB and LP.
Conflict of interest The lab Roche Pharma sponsored the MRI studies presented in this work.
References 1. Sardanelli F, Boetes C, Borisch B, Decker T, Federico M, Gilbert FJ, et al. Magnetic resonance imaging of the breast: recommendations from the EUSOMA working group. Eur J Cancer. 2010;46:1296---316. 2. Kuhl CK. Current status of breast imaging. Part 2. Clinical applications. Radiology. 2007;244:672---91. 3. Siegmann KC, Krämer B, Claussen C. Current status and new developments in breast MRI. Breast Care (Basel). 2011;6:87---92. 4. Li J, Yu Y, Zhang Y, Bao S, Wu C, Wang X, et al. A clinically feasible method to estimate pharmacokinetic parameters in breast cancer. Med Phys. 2009;36:3786---94. 5. Ah-See ML, Makris A, Taylor NJ, Harrison M, Richman PI, Burcombe RJ, et al. Early changes in functional dynamic magnetic resonance imaging predict for pathologic response to neoadjuvant chemotherapy in primary breast cancer. Clin Cancer Res. 2008;14:6580---9. 6. Loo CE, Straver ME, Rodenhuis S, Muller SH, Wesseling J, Vrancken Peeters MJ, et al. Magnetic resonance imaging response monitoring of breast cancer during neoadjuvant chemotherapy: relevance of breast cancer subtype. J Clin Oncol. 2011;29:660---6. 7. Tateishi U, Miyake M, Nagaoka T, Terauchi T, Kubota K, Kinoshita T, et al. Neoadjuvant chemotherapy in breast cancer: prediction of pathologic response with PET/CT and dynamic contrast-enhanced MRI imaging-prospective assessment. Radiology. 2012;263:53---63.
49
8. Rahbar H, Partridge SC, Demartini WB, Gutierrez RL, Allison KH, Peacock S, et al. In vivo assessment of ductal carcinoma in situ grade: a model incorporating dynamic contrast-enhanced and diffusion-weighted breast MR imaging parameters. Radiology. 2012;263:374---82. 9. Johansen R, Jensen LR, Rydland J, Goa PE, Kvistad KA, Bathen TF, et al. Predicting survival and early clinical response to primary chemotherapy for patients with locally advanced breast cancer using DCE-MRI. J Magn Reson Imaging. 2009;29:1300---7. 10. Partridge SC, Vanantwerp RK, Doot RK, Chai X, Kurland BF, Eby PR, et al. Association between serial dynamic contrastenhanced MRI and dynamic 18 F-FDG PET measures in patients undergoing neoadjuvant chemotherapy for locally advanced breast cancer. J Magn Reson Imaging. 2010;32:1124---31. 11. El Khouli RH, Macura KJ, Kamel IR, Jacobs MA, Bluemke DA. 3-T dynamic contrast-enhanced MRI of the breast: pharmacokinetic parameters versus conventional kinetic curve analysis. Am J Roentgenol. 2011;197:1498---505. 12. Baar J, Silverman P, Lyons J, Fu P, Abdul-Karim F, Ziats N, et al. A vasculature-targeting regimen of preoperative docetaxel with or without bevacizumab for locally advanced breast cancer: impact on angiogenic biomarkers. Clin Cancer Res. 2009;15:3583---90. 13. Tofts PS, Berkowitz B, Schnall MD. Quantitative analysis of dynamic Gd-DTPA enhancement in breast tumors using a permeability model. Magn Reson Med. 1995;33:564---8. 14. Marinovich ML, Sardanelli F, Ciatto S, Mamounas E, Brennan M, Macaskill P, et al. Early prediction of pathologic response to neoadjuvant therapy in breast cancer: systematic review of the accuracy of MRI. Breast. 2012;21:669---77. 15. Leach MO, Brindle KM, Evelhoch JL, Griffiths JR, Horsman MR, Jackson A, et al. The assessment of antiangiogenic and antivascular therapies in early-stage clinical trials using magnetic resonance imaging: issues and recommendations. Br J Cancer. 2005;92:1599---610. 16. Li SP, Padhani AR. Tumor response assessments with diffusion and perfusion MRI. J Magn Reson Imaging. 2012;35:745---63. 17. Leach MO, Morgan B, Tofts PS, Buckley DL, Huang W, Horsfield MA, et al. Imaging vascular function for early stage clinical trials using dynamic contrast-enhanced magnetic resonance imaging. Eur Radiol. 2012;22:1451---64. 18. Radjenovic A, Dall BJ, Ridgway JP, Smith MA. Measurement of pharmacokinetic parameters in histologically graded invasive breast tumours using dynamic contrast-enhanced MRI. Br J Radiol. 2008;81:120---8. 19. Padhani AR, Hayes C, Assersohn L, Powles T, Makris A, Suckling J, et al. Prediction of clinicopathologic response of breast cancer to primary chemotherapy at contrast-enhanced MR imaging: initial clinical results. Radiology. 2006;239:361---74.
www.elsevier.es/rx
ORIGINAL REPORT
Evaluation of the reproducibility of a protocol for the pharmacokinetic study of breast tumors by dynamic magnetic resonance imaging夽 J. Etxano a,∗ , A. García-Lallana Valbuena b , I. Antón Ibᘠnez c , A. Elizalde a , a d d L. Pina , J. García-Foncillas , V. Boni a
Departamento de Radiología, Clínica Universidad de Navarra, Pamplona, Navarra, Spain Departamento de Radiología, Hospital San Juan de Dios, Santurce, Vizcaya, Spain c Progenika Biopharma S.A, Derio, Vizcaya, Spain d Departamento de Oncología, Clínica Universidad de Navarra, Pamplona, Navarra, Spain b
Received 4 September 2012; accepted 26 January 2013
KEYWORDS Breast cancer; Magnetic resonance imaging; Pharmacokinetics
Abstract Objective: To evaluate the reproducibility of a protocol for dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) for the pharmacokinetic study of breast tumors. Materials and methods: We carried out this prospective study from October 2009 through December 2009. We studied 12 patients with stages ii--iii invasive breast cancer without prior treatment. Our center’s research ethics committee approved the study. The 12 patients underwent on two consecutive days DCE-MRI with a high temporal resolution protocol (21 acquisitions/minute). The data obtained in an ROI traced around the largest diameter of the tumor (ROI 1) and in another ROI traced around the area of the lesion’s highest Ktrans intensity (ROI 2) were analyzed separately. We used parametric and nonparametric statistical tests to study the reproducibility and concordance of the principal pharmacokinetic variables (Ktrans , Kep , Ve and AUC90 ). Results: The correlations were very high (r > .80; P < .01) for all the variables for ROI 1 and high (r = .70---.80; P < .01) for all the variables for ROI 2, with the exception of Ve both in ROI 1 (r = .44; P = .07) and in ROI 2 (r = .13; P = .235). There were no statistically significant differences between the two studies in the values obtained for Ktrans , Kep and AUC90 (P > .05 for each), but there was a statistically significant difference between the two studies in the values obtained for Ve in ROI 2 (P = .008). Conclusions: The high temporal resolution protocol for DCE-MRI used at our center is highly reproducible for the principal pharmacokinetic constants of breast. © 2012 SERAM. Published by Elsevier España, S.L.U. All rights reserved.
夽 Please cite this article as: Etxano J, García-Lallana Valbuena A, Antón Ibᘠnez I, Elizalde A, Pina L, García-Foncillas J, et al. Evaluación de la reproducibilidad de un protocolo de resonancia magnética dinámica para el estudio farmacocinético de los tumores de mama. Radiología. 2015;57:44---49. ∗ Corresponding author. E-mail address: [email protected] (J. Etxano).
2173-5107/$ – see front matter © 2012 SERAM. Published by Elsevier España, S.L.U. All rights reserved.
Pharmacokinetic study of breast tumors by dynamic magnetic resonance imaging
PALABRAS CLAVE Cáncer de mama; Imagen por resonancia magnética; Farmacocinética
45
Evaluación de la reproducibilidad de un protocolo de resonancia magnética dinámica para el estudio farmacocinético de los tumores de mama Resumen Objetivo: Evaluar la reproducibilidad de un protocolo de resonancia magnética dinámica (RMDC) con contraste para el estudio farmacocinético de los tumores de mama. Material y métodos: Estudio prospectivo realizado entre octubre y diciembre de 2009, que incluyó 12 pacientes con cáncer de mama infiltrante en estadios ii-iii sin tratamiento previo. Este trabajo fue aprobado por el Comité de Ética de Investigación de nuestro centro. A las 12 pacientes se les realizó 2 RM-DC en 2 días consecutivos con un protocolo de alta resolución temporal (21 adquisiciones/minuto). Se analizaron por separado los datos obtenidos en un ROI trazado alrededor del diámetro tumoral mayor (ROI 1) y otro que abarcaba la zona de mayor intensidad de Ktrans de la lesión (ROI 2). Se emplearon pruebas estadísticas paramétricas y no paramétricas para estudiar la reproducibilidad y concordancia de las principales variables farmacocinéticas (Ktrans , Kep , Ve y AUC90 ). Resultados: Las correlaciones fueron muy altas (r > 0,80; p < 0,01) en todas las variables del ROI 1, y altas (r = 0,70---0,80; p < 0,01) en todas las del ROI 2, excepto en Ve tanto en el ROI 1 (r = 0,44; p = 0,07) como en el ROI 2 (r = 0,13; p = 0,235). No hubo diferencias estadísticamente significativas entre los 2 estudios para los valores obtenidos de Ktrans , Kep y AUC90 (p > 0,05 para todas ellas), pero sí que las hubo para Ve en el ROI 2 (p = 0,008). Conclusión: El protocolo de alta resolución temporal de RM-DC de nuestro centro es muy reproducible para las principales constantes farmacocinéticas de los tumores de mama. © 2012 SERAM. Publicado por Elsevier España, S.L.U. Todos los derechos reservados.
Introduction
Materials and methods
The dynamic magnetic resonance imaging (DMR) with IV contrast is one image modality more and more widely used in daily practice. Its indication in high risk patients, the preoperative staging of breast cancer, the hidden breast cancer, the detection of relapses and for the evaluation of the neoadjuvant chemotherapy1---3 are very well known. Because it is based on neoangiogenesis DMR allows the functional assessment of tumors different than conventional modalities---mammography and ultrasound that are only morphological. The actual protocols have a high spatial (with cutting thickness close to 1 mm) and temporal resolution (acquisition in approximately 60 s) that allow us to do semiquantitative dynamic studies based on the curves of signal/time intensity. However if temporal resolution is increased the vascular patency can be determined by quantifying pharmacokinetic parameters that describe the passing of contrast from vascular space to the tumor interstitial compartment.4 To that end it is also necessary to have a specific software capable of estimating these pharmacokinetic parameters. There are many protocols of pharmacokinetic study with DMR reported in the reference (Table 1).5---12 Some of them have even been modified along the study.10 This technical variability can also make the value of pharmacokinetic parameters change---something necessary to study the reproducibility of different protocols. The goal of this study is to evaluate the reproducibility and concordance of data obtained through a DMR protocol optimized for the pharmacokinetic study of stages ii--iii-breast tumors after the administration of neoadjuvant chemotherapy.
Patients Between October and December 2009 we performed one prospective study in our center including 12 treatment naive patients with confirmed histological diagnosis of infiltrating stages ii--iii breast cancer. These 12 patients were part of a phase 2 multicenter clinical trial including 12 hospitals (clinical trial code ML22197/2009-01) sponsored by F. Hoffman---La Roche. This trial enrolled 76 patients and assessed the predictive markers to treatment response with the antiangiogenic drug bevacizumab (Avastin® , F. Hoffman---La Roche, Basel, Switzerland) and neoadjuvant chemotherapy (docetaxel plus adriamicine) in patients with locally advanced breast cancer. In Table 2 the criteria for the inclusion of patients in the study can be seen. The study protocol was approved by the ethical committee of our hospital and all patients gave their informed written consent.
Technique To validate the DMR protocol two daily examinations were performed in two consecutive days with identical technical parameters and the patients did not follow any therapies between both examinations. The MRI protocol can be seen in Table 3. The DMR was performed in one 1.5 T resonance (Magnetom Symphony® , Siemens Healthcare, Erlangen, Germany). T1-weighed standard fl3D echo-spin gradient-enhanced images were modified in an attempt to improve its spatial resolution (150 measurements instead of 6 in a total of 7 min per test with approximately 21 acquisitions per
46 Table 1
J. Etxano et al. DMR protocols in the reference.
Study
Protocol
Ah See et al. (2008) Loo et al. (2011)6
5
Tateishi et al. (2012)7 Rahbar et al. (2012)8 Johansen et al. (2009)9 Partridge et al. (2010)10 El Khouli et al. (2011)11 Baar et al. (2009)12
1 1 2 1 1 2 1 1 2 1 1
View Sagittal Coronal Coronal Sagittal Axial Axial Sagittal Sagittal Sagittal Axial Axial
Voxel size a
0.8 mm 1.2 mm × 1.2 mm × 1.6 mm 1.1 mm × 1.1 mm × 1.2 mm 1.3 mm × 1.3 mm × 0.8 mm 0.9 mm × 1.1 mm × 2.2 mm 0.8 mm × 0.9 mm × 1.6 mm 1.9 mm × 1 mm × 3---4 mmb 1 mm × 1 mm × 2.2 mm 0.8 mm × 0.8 mm × 1.6 mm 1.3 mm × 1.3 mm × 5 mm 2.3 mm × 2.7 mm × 8 mm
Number of acquisitions
Total time
40 5 5 24 5 3 6---7 3 3 14 180
8 min 5 s 7 min 30 s 7 min 30 s 4 min 7 min 30 s 7 min 30 s 5 min 42 s-6 min 39 s 4 min 30 s 6 min 2 min 30 s 24 min
DMR: dynamic magnetic resonance imaging. a This value corresponds to the cutting thickness; the article says nothing about the matrix or the FOV used. b The protocol was applied to twelve (12) study patients with a cutting thickness of 3 mm and to some other 12 with a cutting thickness of 4 mm.
Table 2
Inclusion criteria.*
---------
Women between 18 and 70 years old Breast cancer histologically tested through skin biopsy Stage ii or iii, naïve to all prior therapies. Eastern Cooperative Oncology Group (ECOG) categories 0 to 1 --- Left ventricular ejection fraction (LVEF) ≥ 50% with no signs of cardiac failure
* Patients in the clinical trial of tumor response marker assessment after neoadjuvant therapy combined with one antiangiogenic drug and chemotherapy of stages ii--iiiinfiltrating breast tumors (clinical trial code: ML22197/2009-01).
minute). In turn the spatial resolution decreased (voxel size of 1.25 mm × 1.25 × 4 mm; 22 cuts) and the fat saturation pulse was eliminated. Given that the entire breast could not be studied through this protocol---especially large breasts Table 3 Dynamic magnetic resonance protocol with IV contrast. Magnetic resonance
Sequence View Acquisition Fat saturation TR/TE FoV Cutting thickness Matrix Acceleration Flip angle Number of measurements Number of cuts per block Total acquisition time
1.5 T (Magnetom Symphony® , Siemens Healthcare, Erlangen, Germany) T1-weighed standard fl3D echo gradient Coronal Caudo-cranial Without fat-sat pulse 4.3/1.29 320 mm 4 mm 256 × 256 ×2 GRAPPA 12◦ 150 2 7.08 min
in all patients---the tumor was found through an ultrasound and its location marked on their skin to be able to identify it in the location sequence of the MRI. After the acquisition of three images without IV contrast one bolus of 8 ml gadobutrol (Gadovist® , Bayer-Schering Pharma AG, Berlín, Germany) at 4 ml/s was automatically injected using one automatic injector after which 147 dynamic images could be obtained. Finally subtraction images of lesions were generated. Images were processed with one specific software (TISSUE 4D® , Siemens Healthcare, Erlangen, Germany) capable of evaluating the dynamic uptake of tumor contrast by calculating the pharmacokinetic values in the bicompartmental model of Tofts et al.13 . To assess the concentration of blood contrast or arterial input function (AIF), the software used offers three pre-established categories (quick, intermediate and slow). Given the high temporal resolution of our protocol, the category «quick» was used as the default one in all patients. The following values were analyzed: (1) Ktrans : constant of transendothelial transference. It measures the passing of contrast from the intravascular space to the extracellular space of tumor or in other words it is one measurement for the patency and blood flow of tumor vessels; (2) Kep : constant relating the passing of contrast from the tumor extravascular space to the intravascular one. It is the reverse constant to Ktrans , and is related to tumor washout; (3) Ve : fraction of tumor volume occupied by the extravascular space; and (4) AUC90 : area under the curve of signal intensity through time showing variation in tumor blood volume. It was estimated at 90 s which is the standard value of the software used. These parameters are related through the equation Kep = Ktrans /Ve . In each and every of the two DMR two regions of interest (ROI) were drawn --- one of them around the maximum diameter of lesion in subtraction images (ROI 1) and the other one with the smallest possible standardized size in the region of maximum Ktrans value (ROI 2) that we could identify by analyzing the Ktrans lesion map of colors without software (Fig. 1). In both cases we measured twice to assess the average values of Ktrans , Kep , Ve and AUC90 . We chose the Ktrans map because in the reference this value is
Pharmacokinetic study of breast tumors by dynamic magnetic resonance imaging
47
ROI 1
ROI 2
Roi1 Roi1
0.11
Roi1 Ktrans Kep Ve iAUC
: : : :
0.140 0.475 0.290 16.455 Median
0.136 0.472 0.287 16.044 Mean
0.007 0.011 0.009 0.674 std ktrans
0.03
Figure 1 On the subtracted image obtained in the coronal view (left upper image) the maximum tumor diameter could be identified and the ROI 1 was drawn marking out the area of the lesion. After obtaining the pharmacokinetic parameters the same proceeding was repeated to obtain the average values of the 2 measurements. Similarly using the Ktrans color map (right upper image) the ROI 2 was drawn on the Ktrans area of maximum intensity repeating the proceeding to obtain the average values.
considered to be one of the most relevant ones for the study of changes in neoangiogenesis and the response to neoadjuvant chemotherapy.5 Only one radiologist (LJP) highly experienced in performing DMR did the readings of all the cases.
Analysis To study the reproducibility and concordance of the outcomes obtained parametric statistical tests (interclass correlation coefficient and the Student’s t test for paired samples) and non-parametric tests were used (Spearman’s rho and Wilcoxon rank-sum test) with the software SPSS® v.15.0 (Chicago, Illinois, U.S.A.) Prior to this the normalcy of ROI 1 and ROI 2 variables through the Kolmogorov---Smirnov and Shapiro---Wilk tests was analyzed. P < 0.05 was considered significant.
Results The average maximum diameter of tumors through MRI was 35 mm with a rank between 14 and 100 mm. The four variables showed one normal distribution in ROI 1. In ROI 2 only in Kep and Ve the distribution was normal. The outcomes of the statistical analysis can be seen in Table 4. Very high correlations (r > 0.80; p < 0.01) between the 2 DMR were found in all ROI 1 variables and high correlations (r = 0.70---0.80; p < 0.01) were found in all ROI 2 variables with the exception of Ve both in ROI 1 (r = 0.44; p = 0.07) and ROI 2 (r = 0.13; p = 0.23). The average ROI 1 concordance was higher than that of ROI 2 (r = 0.77 vs r = 0.60). There were no statistically significant differences between both studies for the Ktrans , Kep and AUC90 values obtained (p > 0.05 for all) but there were statistically significant differences for Ve values obtained in ROI 2 (p = 0.008).
Discussion In our study we showed that with our high temporal resolution and reduced spatial resolution DMR protocol the concordance of the main pharmacokinetic variables is really good except for variable Ve . The analysis of angiogenesis through DMR in breast tumors is hot at the moment.7,8 The reduced tumor neovascularization is one of the most widely studied therapeutic targets during the last years and one of the most promising ones in the neoadjuvant therapy of breast cancer.14---16 Hence the importance of the analysis of evolutionary changes occurring in tumor reovascularization for which DMR is the most adequate DMR. In our study we did not find statistically significant differences in any variables between the two consecutive tests except for Ve in ROI 2. Concordance of all variables has been high (ROI 2), or very high (ROI 1), except for Ve values. It is possible that the fact of using a predetermined value of non-adjusted AIF for every study might have influenced the poor concordance found in both cases for Ve . The software used does not offer the possibility of performing a study prior to the contrast bolus for the analysis of real AIF. According to Leach et al.17 it is possible to measure the real AIF before the bolus but this increases the time and complexity of the study. Measuring it requires injecting a small amount of contrast media (around 10%) and one optimized image protocol. Another alternative is measuring AIF in one non-tumor region (e.g. muscular tissue) with known pharmacokinetics properties.17 However to analyze the importance of this study we need to highlight that the most important constants for the analysis of changes in tumor neovascularization are those that have to do with patency and the passing of contrast through the endothelial membrane (Ktrans , Kep ).18 We also performed the segmented analysis of data obtained thanks to one ROI drawn through the tumor
48 Table 4
J. Etxano et al. Statistical results.
Variable
Student’s t test for paired samples
CCI
Median 1
Median 2
P Value
Ratio
P Value
ROI 1 Ktrans Kep Ve AUC90
133.04 291.92 498.83 14.50
134.20 271.80 513.21 13.15
0.928 0.143 0.696 0.196
0.831 0.938 0.443 0.885
0.001 0.001 0.07 0.001
ROI 2 Kep Ve
401.33 610.80
374.00 733.46
0.530 0.008
0.758 0.135
0.001 0.236
Wilcoxon test
Variable
ROI 2 Ktrans AUC90
Spearman’s rho
Median 1
Median 2
P Value
Ratio
P Value
237.5 24.39
242.5 21.56
0.694 0.272
0.748 0.776
0.005 0.003
AUC90 : area under the curve of signal intensity throughout time. Calculated at 90 s; CCI: intraclass correlation coefficient; Kep : constant of contrast passing from the tumor extravascular space toward the intravascular space; Ktrans : constant of transendothelial transport from the intravascular space toward the extracellular tumor space; ROI 1: region of interest drawn around the maximum diameter of lesion in subtraction images; ROI 2: region of interest drawn in the region of maximum value of Ktrans identified through Ktrans color map of the lesion provided by the software; Ve : fraction of tumor volume occupied by extravascular space.
perimeter and another ROI including the area of maximum intensity of Ktrans . It is significant to see how in all cases the concordance of ROI 1 variables is greater than its corresponding ROI 2. This might be due to the fact that while drawing the edges of ROI 2 we rely upon the color map provided by Ktrans and the outlining of edges is more inaccurate. The data obtained in our study seem to indicate that quantification measured through one ROI drawn around the maximum diameter of lesion can be more reproducible than that in one ROI including the greater intensity region of Ktrans map, yet future research is need for confirmatory purposes. Having a DMR protocol highly reproducible is highly recommended since several studies have shown that the assessment of changes in pharmacokinetic constants (Ktrans , Kep , Ve and AUC90 ) can predict tumor response to neoadjuvant chemotherapy. Ah-See et al.5 said that early functional changes of tumor microvessels at the beginning of neoadjuvant chemotherapy can help us find out if we will have tumor response. Padhani et al.19 found that changes in pharmacokinetic constants were as sensitive and specific as changes in the tumor size to know if therapeutic response will ever happen. At present our institution is focused on one multicenter clinical trial aimed at evaluating the efficacy of neoadjuvant therapy with the antiangiogenic drug bevacizumab (Avastin® ) in stages ii--iii breast cancer through a serial study of the pharmacokinetic values obtained through DMR. Our work has some limitations. Firstly the low number of patients and secondly that there was only one radiologist in charge of all the readings---which means that inter-observer variability cannot be analyzed. However in the clinical trial protocol both parameters were agreed upon. Another limitation is the already commented absence of an individual calculation of the AIF to better determine the value of Ve .
For the study of pharmacokinetic parameters of smaller tumors the modification of the protocol with a greater spatial resolution and lower temporal resolution could be considered. In sum the protocol we presented for the assessment of the pharmacokinetic constants of breast tumors is highly reproducible for the values of Ktrans , Kep and AUC90 , with a higher concordance when assessing the ROI of the tumor maximum diameter. However it does not seem to be adequate for the Ve pharmacokinetic constant even though this result needs to be re-analyzed with a larger number of patients.
Ethical responsibilities Protection of humans and animals. Authors confirm that all proceedings and experiments followed relate to the committee of responsible human experimentation ethical rules and regulations in compliance with the World Medical Association and the Declaration of Helsinki. Data confidentiality. Authors confirm that the protocols of their centers have been followed on matters concerning the publishing of data from patients. They also confirm that all patients included in this study have been given enough information and handed over their written informed consent for their participation in this study. Right to privacy and informed consent. Authors confirm that they have obtained the written informed prior consent from patients and/or subjects appearing in this article. This document is in the possession of the corresponding author.
Pharmacokinetic study of breast tumors by dynamic magnetic resonance imaging
Author’s contributions 1. 2. 3. 4. 5. 6. 7. 8. 9.
Manager of the integrity of the study: JE and LP. Original Idea of the Study: JE and LP. Study Design: JE, LP and JGF. Data Mining: JE, LP, JGF and VB. Data Analysis and Interpretation: JE and LP. Statistical Analysis JE, IA and AGL. Reference Search: JE, AGL, AE and LP. Writing: JE, IA, LP, AE, AGL and LP. Manuscript critical review with intellectually relevant contributions: IA, AGL, JGF and VB. 10. Final Version Approval: JE, AGL, IA, AE, JGF, VB and LP.
Conflict of interest The lab Roche Pharma sponsored the MRI studies presented in this work.
References 1. Sardanelli F, Boetes C, Borisch B, Decker T, Federico M, Gilbert FJ, et al. Magnetic resonance imaging of the breast: recommendations from the EUSOMA working group. Eur J Cancer. 2010;46:1296---316. 2. Kuhl CK. Current status of breast imaging. Part 2. Clinical applications. Radiology. 2007;244:672---91. 3. Siegmann KC, Krämer B, Claussen C. Current status and new developments in breast MRI. Breast Care (Basel). 2011;6:87---92. 4. Li J, Yu Y, Zhang Y, Bao S, Wu C, Wang X, et al. A clinically feasible method to estimate pharmacokinetic parameters in breast cancer. Med Phys. 2009;36:3786---94. 5. Ah-See ML, Makris A, Taylor NJ, Harrison M, Richman PI, Burcombe RJ, et al. Early changes in functional dynamic magnetic resonance imaging predict for pathologic response to neoadjuvant chemotherapy in primary breast cancer. Clin Cancer Res. 2008;14:6580---9. 6. Loo CE, Straver ME, Rodenhuis S, Muller SH, Wesseling J, Vrancken Peeters MJ, et al. Magnetic resonance imaging response monitoring of breast cancer during neoadjuvant chemotherapy: relevance of breast cancer subtype. J Clin Oncol. 2011;29:660---6. 7. Tateishi U, Miyake M, Nagaoka T, Terauchi T, Kubota K, Kinoshita T, et al. Neoadjuvant chemotherapy in breast cancer: prediction of pathologic response with PET/CT and dynamic contrast-enhanced MRI imaging-prospective assessment. Radiology. 2012;263:53---63.
49
8. Rahbar H, Partridge SC, Demartini WB, Gutierrez RL, Allison KH, Peacock S, et al. In vivo assessment of ductal carcinoma in situ grade: a model incorporating dynamic contrast-enhanced and diffusion-weighted breast MR imaging parameters. Radiology. 2012;263:374---82. 9. Johansen R, Jensen LR, Rydland J, Goa PE, Kvistad KA, Bathen TF, et al. Predicting survival and early clinical response to primary chemotherapy for patients with locally advanced breast cancer using DCE-MRI. J Magn Reson Imaging. 2009;29:1300---7. 10. Partridge SC, Vanantwerp RK, Doot RK, Chai X, Kurland BF, Eby PR, et al. Association between serial dynamic contrastenhanced MRI and dynamic 18 F-FDG PET measures in patients undergoing neoadjuvant chemotherapy for locally advanced breast cancer. J Magn Reson Imaging. 2010;32:1124---31. 11. El Khouli RH, Macura KJ, Kamel IR, Jacobs MA, Bluemke DA. 3-T dynamic contrast-enhanced MRI of the breast: pharmacokinetic parameters versus conventional kinetic curve analysis. Am J Roentgenol. 2011;197:1498---505. 12. Baar J, Silverman P, Lyons J, Fu P, Abdul-Karim F, Ziats N, et al. A vasculature-targeting regimen of preoperative docetaxel with or without bevacizumab for locally advanced breast cancer: impact on angiogenic biomarkers. Clin Cancer Res. 2009;15:3583---90. 13. Tofts PS, Berkowitz B, Schnall MD. Quantitative analysis of dynamic Gd-DTPA enhancement in breast tumors using a permeability model. Magn Reson Med. 1995;33:564---8. 14. Marinovich ML, Sardanelli F, Ciatto S, Mamounas E, Brennan M, Macaskill P, et al. Early prediction of pathologic response to neoadjuvant therapy in breast cancer: systematic review of the accuracy of MRI. Breast. 2012;21:669---77. 15. Leach MO, Brindle KM, Evelhoch JL, Griffiths JR, Horsman MR, Jackson A, et al. The assessment of antiangiogenic and antivascular therapies in early-stage clinical trials using magnetic resonance imaging: issues and recommendations. Br J Cancer. 2005;92:1599---610. 16. Li SP, Padhani AR. Tumor response assessments with diffusion and perfusion MRI. J Magn Reson Imaging. 2012;35:745---63. 17. Leach MO, Morgan B, Tofts PS, Buckley DL, Huang W, Horsfield MA, et al. Imaging vascular function for early stage clinical trials using dynamic contrast-enhanced magnetic resonance imaging. Eur Radiol. 2012;22:1451---64. 18. Radjenovic A, Dall BJ, Ridgway JP, Smith MA. Measurement of pharmacokinetic parameters in histologically graded invasive breast tumours using dynamic contrast-enhanced MRI. Br J Radiol. 2008;81:120---8. 19. Padhani AR, Hayes C, Assersohn L, Powles T, Makris A, Suckling J, et al. Prediction of clinicopathologic response of breast cancer to primary chemotherapy at contrast-enhanced MR imaging: initial clinical results. Radiology. 2006;239:361---74.