Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Magnetic Resonance in Medicine 59:1183–1189 (2008) Feasibility of Using Limited-Population-Based Arterial Input Function for Pharmacokinetic Modeling of Osteosarcoma Dynamic Contrast-Enhanced MRI Data Ya Wang,1 Wei Huang,1–3* David M. Panicek,2,3 Lawrence H. Schwartz,2,3 and Jason A. Koutcher1– 4 For clinical dynamic contrast-enhanced (DCE) MRI studies, it is often not possible to obtain reliable arterial input function (AIF) in each measurement. Thus, it is important to find a representative AIF for pharmacokinetic modeling of DCE-MRI data when individual AIF (Ind-AIF) measurements are not available. A total of 16 patients with osteosarcomas in the lower extremity (knee region) underwent multislice DCE-MRI. Reliable Ind-AIFs were obtained in five patients with a contrast injection rate of 2 cc/s and another five patients with a 1 cc/s injection rate. Average AIF (Avg-AIF) for each injection rate was constructed from the corresponding five Ind-AIFs. For each injection rate there are no statistically significant differences between pharmacokinetic parameters of the five patients derived with Ind-AIFs and Avg-AIF. There are no statistically significant changes in pharmacokinetic parameters of the 16 patients when the two AvgAIFs were applied in kinetic modeling. The results suggest that it is feasible, as well as practical, to use a limited-populationbased Avg-AIF for pharmacokinetic modeling of osteosarcoma DCE-MRI data. Further validation with a larger population and multiple regions is desirable. Magn Reson Med 59: 1183–1189, 2008. © 2008 Wiley-Liss, Inc. Key words: dynamic contrast-enhanced MRI; arterial input function; osteosarcoma; Ktrans; pharmakinetic modeling There has been increasing interest in the T1-weighted dynamic contrast-enhanced (DCE) MRI method for the study of many different tumor types, using the Gd (III) chelate contrast agents (1). There are generally three approaches for analyzing DCE-MRI signal time courses: 1) qualitative subjective assessment of curve shape, such as wash-out, plateau, and persistent; 2) empirical quantitation, such as maximum slope and percent signal intensity change; and 3) analytical pharmacokinetic modeling. The latter is more sophisticated, and also the more desirable. Unlike the first two approaches, analytical modeling of DCE-MRI data ex- 1Department of Medical Physics, Memorial Sloan-Kettering Cancer Center, New York, New York, USA. 2Department of Radiology, Memorial Sloan-Kettering Cancer Center, New York, New York, USA. 3Department of Radiology, Weill Medical College of Cornell University, New York, New York, USA. 4Department of Medicine, Memorial Sloan-Kettering Cancer Center, New York, New York, USA. Grant sponsor: National Cancer Institute/National Institutes of Health; Grant number: 1 R01 CA104754. Ya Wang and Wei Huang contributed equally to this study. *Correspondence to: Wei Huang, PhD, Department of Medical Physics, Memorial Sloan-Kettering Cancer Center, 1275 York Avenue, New York, NY 10021. E-mail: [email protected] Received 30 January 2007; revised 23 July 2007; accepted 6 September 2007. DOI 10.1002/mrm.21432 Published online in Wiley InterScience (www.interscience.wiley.com). © 2008 Wiley-Liss, Inc. tracts pharmacokinetic parameters that should be independent of data acquisition details, contrast agent dose and injection rate, magnetic field strength, and vendor platform, etc. This improves DCE-MRI study reproducibility and enables meaningful comparison of results from different imaging sites where different DCE-MRI protocols are employed. The extracted pharmacokinetic parameters are usually variants of: Ktrans, a rate constant for contrast agent plasma/interstitium transfer, and ve, the interstitial space volume fraction (the putative contrast agent distribution volume). These parameters have been used for cancer diagnosis (2– 4) and monitoring effects of antiangiogenic therapies (5,6). A characteristic aspect of pharmacokinetic modeling of DCE-MRI signal time course is the requirement for an arterial input function (AIF). The absolute accuracy of the pharmacokinetic parameters, Ktrans and ve, depends on the AIF accuracy (7,8). Ideally, individually measured AIF should be used for kinetic modeling of the corresponding tissue DCE-MRI data in each experiment (3,7,9). However, for patient studies that are conducted in clinical settings, it is often not possible to obtain reliable AIF measurement in each DCE-MRI examination, due to either data acquisition constraints, such as excitation volume coverage and image slice angulation, or lack of a visible artery that is anatomically adjacent to the tissue of interest. One solution to such problem is to generate a population-based average AIF for kinetic modeling of DCE-MRI data when an individual AIF is not obtainable. One recent study (10) shows that use of a population-averaged AIF reduces variability and improves reproducibility of DCE-MRI pharmacokinetic model parameters. Using a semiquantitative approach, we reported that the histogram amplitude of the initial slope of DCE-MRI signal time course correlates significantly with necrosis percentage of osteogenic and Ewing sarcoma, which is an important indicator of the effectiveness of chemotherapy (11). In determination of AIF for absolute quantitation of Ktrans and ve, it is often not possible to perform reliable AIF measurement in each individual osteosarcoma DCE-MRI experiment. Also, the contrast agent injection rate may not be consistent between studies because of variations in location and size of intravenous (IV) catheters that are already in place before patients undergo DCE-MRI studies. In this preliminary study, we sought to assess the feasibility of using an average AIF obtained from a limited population of osteosarcoma patients for kinetic modeling of DCE-MRI data of a larger population, as well as to assess the effects of different injection rates (1 cc/s and 2 cc/s) on determination of pharmacokinetic parameters. 1183 1184 Wang et al. FIG. 1. a: A postcontrast sagittal image extracted from a multislice DCE-MRI acquisition, showing an osteosarcoma in the distal femur and the adjacent femoral artery. The AIF data points were obtained from the red ROI placed within the artery. b: An AIF plot (plasma Gd-DTPA concentration time course). The contrast washout phase was fitted with a biexponential decay function. MATERIALS AND METHODS Patients Prior to definitive surgery, 16 patients (mean age ⫽ 16 years, range ⫽ 10 –29 years) with osteosarcomas in the lower extremity underwent a routine clinical MRI protocol, in which a DCE-MRI scan was added for the purpose of evaluating the efficacy of chemotherapy in inducing tumor necrosis. The DCE-MRI study was conducted under an Internal Review Board–approved protocol, and the written consent was obtained from each patient prior to the DCE-MRI scan. Data Acquisition All the MRI studies were performed with a 1.5T GE Excite system (General Electric Medical Systems, Milwaukee, WI, USA). An extremity knee coil was used for RF transmission and reception. Before the DCE-MRI study was conducted, a standard clinical MRI exam was performed through the tumor. Axial T1-weighted and fat-suppressed fast spin-echo T2-weighted images were obtained with as small a field of view (FOV) as possible. Longitudinal (coronal and/or sagittal) T1-weighted and fat-suppressed fast spin-echo T2-weighted images through the entire bone were also obtained with a small FOV. These were followed by proton density MRI and the T1-weighted DCE-MRI study in the sagittal direction, and then by postcontrast axial fat-suppressed T1-weighted MRI. For DCE-MRI data acquisition, a fast multiplanar spoiled gradient echo sequence was employed with flip angle (␣) ⫽ 30°, TE ⫽ 2.9 ms, TR ⫽ 7.5–9.0 ms, FOV ⫽ 20 –24 cm, and matrix size ⫽ 256 ⫻ 128 zero filled to 256 ⫻ 256 during image reconstruction. The entire tumor was imaged with eight to 11 sagittal slices of 10 –12-mm thickness and zero gap. The total DCE-MRI acquisition time was about 5–7 min with 7–10 s temporal resolution and 30 – 60 time course data points. At the beginning of the sixth image set (data point) acquisition, gadopentrate dimeglumine (Gd-DTPA) con- trast agent (Magnevist; Berlex Laboratories, Wayne, NJ, USA) at a dose of 0.1 mmol/kg was administered intravenously with a rate of 1 cc/s or 2 cc/s by a MR-compatible programmable power injector (Spectris; Medrad, Indianola, PA, USA). Besides the MRI examination during the visit, a patient usually also underwent other clinical procedures and often arrived at the MRI suite with the IV catheter already in place. The injection rate was determined according to the location and the size of the IV catheter. Proton density images were acquired for the purpose of determining the longitudinal relaxation rate constant, R1, for each DCE-MRI data point, using the same pulse sequence with ␣ ⫽ 30°, TE ⫽ 2.0 ms, TR ⫽ 350 ms, and DCE-MRI-matching slice number, thickness, and location. DCE-MRI Data Analysis For AIF determination, whenever a femoral artery adjacent to the tumor was clearly visible in the DCE-MRI image series, a manually drawn region of interest (ROI) was placed within the artery and the signal time course was obtained. Figure 1a shows the relative locations of the knee region osteosarcoma and the femoral artery of a patient, as well as the ROI placement (in red) for AIF sampling. Due to the nature of multislice acquisition for DCEMRI and angulation of the sagittal slices, the femoral artery was not always clearly visible, depending on the relative locations of the osteosarcoma and the artery. Reliable ROI signal time courses were obtained in only five patients with 1 cc/s contrast injections and five patients with 2 cc/s injections. To obtain the AIF, the signal time course needs to be converted to plasma contrast concentration time course. Assuming TE ⬍⬍ T2, the signal intensity (S) of a spoiled gradient echo sequence is given by (12): 1 ⫺ exp共 ⫺ TR/T1兲 S ⫽ S 0 sin ␣ , 1 ⫺ cos ␣ exp共 ⫺ TR/T1兲 [1] AIF for Osteosarcoma DCE-MRI 1185 where S0 is a constant proportional to the proton density of the sample. By comparing the S values of the artery ROI from the DCE and proton density images, R1 (⬅ 1/T1) for each time point of the DCE series can be theoretically derived using Eq. [1]. To correct for possible errors in T1 calculation likely caused by imperfect slice profile, a calibration curve of signal intensity ratio of T1-weighted image over proton density image vs. T1 was constructed using a method introduced by Parker et al. (13). A total of 12 agar gel phantoms doped with various concentrations of GdDTPA were imaged with the same pulse sequence and acquisition parameters as those used for DCE and proton density MRI. The T1 values for each phantom were first measured using an inversion-recovery spectroscopy sequence, covering a range of 105 to 2224 ms. The 12 data points were empirically fitted with a biexponential function with offset (13) to generate the calibration curve. The artery ROI R1 values for the DCE image series were obtained from the calibration curve and then converted to Gd-DTPA concentrations using the following linear equation: R 1 ⫽ r1 䡠 Cp共t兲 ⫹ R10 , [2] where Cp(t) is the arterial plasma Gd-DTPA concentration at time t, r1 is the contrast agent relaxivity, which was taken to be 4.1 s–1(mmol/liter)–1 at 1.5T (14), and R10 is the precontrast R1. The derived Cp(t) time course was fitted with a biexponential decay function in the washout phase (15) to generate the AIF. Figure 1b shows an AIF from a DCE-MRI study with a 2 cc/s contrast injection rate. By simple averaging of the five individual AIFs (Ind-AIFs) at each injection rate (1 cc/s and 2 cc/s) with peak height aligned, average AIFs (Avg-AIFs) for the two injection rates were obtained. For the tumor tissue DCE-MRI time course data, an ROI was manually drawn circumscribing the contrast-enhanced tumor for signal intensity measurement. The signal intensity time course was converted to R1 time course using the signal ratio of the T1-weighted DCE image over the proton density image and the T1 calibration curve, and subsequently converted to tumor tissue– contrast agent concentration, Ct(t), time course with Eq. [2] by substituting Cp(t) with Ct(t). An in-house IDL (version 6.0; Research Systems, Boulder, CO, USA) program based on the Toft’s (16) model was used to fit the Ct(t) time course for the extraction of the Ktrans and ve parameters, as shown in the following Kety-Schmidt type of rate law equation: FIG. 2. Average AIFs for 1 cc/s and 2 cc/s contrast injection rates obtained from five individually sampled AIFs of the corresponding injection rates, respectively. data. Exclusion of the vp parameter may cause significant errors, however, when there is less contrast extravasation, such as when Ktrans ⬍ 0.001 min–1. For each DCE MRI data set of this study, since the arrival of AIF peak amplitude preceded the apparent rise of tumor tissue signal intensity, defined as signal intensity rising more than one standard deviation (SD) of the signal intensities of the five precontrast injection baseline data points, only the biexponential function-fitted AIF washout phase was used for Ct(t) curve fitting with time zero in Eq. [3] set as the time of AIF peak amplitude. Both Ind-AIFs (on the 10 corresponding patient data sets) and Avg-AIFs (on all 16 patient data sets) were used for the pharmacokinetic modeling analyses of the tumor tissue time course data, which were performed for the ROI, as well as each image pixel within the ROI. For the latter approach, histogram analyses (11) of the pixel Ktrans and ve were performed and the median values of these parameters were calculated. For the purpose of this study, only the image slice that included the center portion of the tumor was used for data analysis. Student’s paired t-test was used to evaluate differences in pharmacokinetic parameters resulted from the use of Ind- and Avg-AIFs of the same contrast injection rate, as well as differences resulted from the use of Avg-AIFs of the two injection rates. RESULTS t 冕 C t共t兲 ⫽ Ktrans Cp共t⬘兲exp共 ⫺ Ktransve⫺1 共t ⫺ t⬘兲兲dt⬘. [3] 0 The addition of the term that includes plasma volume fraction (vp), vpCp(t), is ignored on the right side of this equation. Li et al. (17) have recently shown that when there is sufficient contrast agent extravasation from plasma to interstitium, such as in tumor tissue, the Ktrans and ve parameters are adequate for kinetic modeling of DCE-MRI Figure 2 shows the biexponential function-fitted Avg-AIFs for 2 cc/s (solid line) and 1 cc/s (dotted line) contrast injection rates, respectively. The two Avg-AIFs have almost the same maximum Gd-DTPA concentration, with minor wash-out shape mismatch. Figure 3 shows scatter plots of tumor tissue ROI Ktrans parameters (Fig. 3a) and median values of Ktrans histograms (Fig. 3b) obtained from kinetic modeling with the Ind-AIFs and the Avg-AIF for the five patients who had a 2 cc/s contrast injection rate. The straight lines connect the data points from the same patient. There are no statistically significant differences between Ktrans parameters derived with Ind-AIFs and those 1186 Wang et al. FIG. 3. Scatter plots of Ktrans parameters obtained from single-image slice pharmacokinetic modeling analyses of five DCE-MRI studies (five patients) with 2 cc/s contrast injection rate. Kinetic analyses using the individually measured AIF (Ind-AIF) and the average AIF (Avg-AIF) were performed for each study: (a) tumor tissue ROI analysis; (b) median value of histogram analysis of pixel Ktrans parameters within the ROI. The straight lines connect data points from the same study. derived with the Avg-AIF (P ⫽ 0.45 for ROI Ktrans and 0.37 for histogram median Ktrans, paired t-test). The comparisons between the Ktrans and ve parameters obtained with the Ind-AIFs and those obtained with the Avg-AIFs for both injection rates are summarized in Table 1. At either injection rate, no significant changes in Ktrans and ve parameters occurred when the Avg-AIF was used for kinetic modeling. Figure 4 displays the representative graphs for a patient who had a 2 cc/s contrast injection, showing pixel Ktrans (Fig. 4a) and ve (Fig. 4b) values within the tumor tissue ROI obtained with the Avg-AIF plotted against those obtained with the Ind-AIF. There were a total of 1940 pixels within the ROI. Both plots demonstrate significant linear correlations (P ⬍ 0.0001) with the slope values close to one (1.048 for Ktrans and 1.064 for ve). Similar results were obtained from the other nine patients with reliable Ind-AIF measurements at either injection rate. This indicates that the use of Avg-AIF works equally well for both ROI and pixel-by-pixel data analyses. Figure 5 shows scatter plots of tumor ROI Ktrans parameters (Fig. 5a) and median values of Ktrans histograms (Fig. 5b) obtained from kinetic modeling with the two Avg-AIFs for all 16 patients. There are no statistically significant differences between the two sets of Ktrans parameters (P ⫽ 0.92 for ROI Ktrans and 0.86 for histogram median Ktrans) and ve parameters (plots not shown here. P ⫽ 0.74 for ROI ve and 0.79 for histogram median ve). DISCUSSION AND CONCLUSIONS Previous studies (3,7,9) suggest that AIF should be individually monitored if accurate kinetic modeling of DCEMRI time course data is desired. However, for clinical DCE-MRI studies, because of data acquisition constraints or lack of a visible artery adjacent to the tissue of interest (TOI), it is not practical to measure individual AIF for each DCE-MRI scan. Therefore, it is important to find a reasonable AIF substitute for pharmacokinetic modeling when individual AIF measurement is not achievable. Works by Yankeelov et al. (18,19) and Walker-Samuel et al. (20,21) show that if a reliable AIF is not available, a reference region model is a reasonable alternative for measuring DCE-MRI kinetics. In this model, however, literature values of Ktrans and ve have to be assigned to a reference tissue and the assumption that the TOI and the reference tissue share the same AIF has to be made. The TOI Ktrans and ve parameters can be extracted by comparing the curve shapes of the TOI and the reference tissue. Another study by Yang et al. (22) proposed a similar approach using a double-reference method, in which two reference tissue Table 1 Comparisons of Ktrans and ve Obtained with Ind-AIF and Avg-AIF* Contrast injection rate 2 cc/s Ktrans (min⫺1) ROI Histogram median Ktrans (min⫺1) ROI ve Histogram median ve 1 cc/s Ind-AIF Avg-AIF Ind-AIF Avg-AIF 0.87 ⫾ 0.55 0.57 ⫾ 0.31 0.58 ⫾ 0.17 0.59 ⫾ 0.19 0.76 ⫾ 0.61 ⫾ 0.35b 0.59 ⫾ 0.14e 0.61 ⫾ 0.17f 0.61 ⫾ 0.56 0.55 ⫾ 0.53 0.50 ⫾ 0.24 0.50 ⫾ 0.23 0.66 ⫾ 0.66c 0.52 ⫾ 0.48d 0.49 ⫾ 0.26g 0.50 ⫾ 0.27h 0.46a *Mean ⫾ SD; Student’s paired t-test for Ktrans and ve values obtained with Ind-AIF and Avg-AIF at each injection rate: aP ⫽ 0.45, bP ⫽ 0.37, cP ⫽ 0.56, dP ⫽ 0.65, eP ⫽ 0.93, fP ⫽ 0.48, gP ⫽ 0.69, hP ⫽ 0.88. AIF for Osteosarcoma DCE-MRI 1187 FIG. 4. Scatter plots of pixel (a) Ktrans and (b) ve parameters obtained from single-image slice pharmacokinetic modeling analyses of the DCE-MRI data from one patient with a 2 cc/s contrast injection. The Ktrans and ve parameters extracted with the average AIF (Avg-AIF) are plotted against those extracted with the individually measured AIF (Ind-AIF). The analyzed pixels (N ⫽ 1940) are within the tumor tissue ROI. The solid straight lines represent linear correlations, while the dashed ones are lines of identity. regions are assumed to have the same pharmacokinetic parameters and share the same AIF with the TOI. The AIF is extracted by comparing the curve shapes of the two reference tissues and then used for kinetic modeling of the TOI. The advantage of these two reference methods is that a direct AIF measurement is not required. However, both methods require assumptions that do not necessarily reflect the actual tissue pharmacokinetic characteristics, and may introduce errors in kinetic modeling. Through direct AIF measurement, our preliminary results suggest that it is reasonable, as well as practical, to use a limited-population-based Avg-AIF for pharmacokinetic modeling of osteosarcoma DCE-MRI data from a large population when it is not possible to measure Ind-AIF in each patient. This finding is consistent with that of a recent study (10), which demonstrates improved reproducibility in DCE-MRI kinetic modeling using a population-averaged AIF. Limitedpopulation-averaged AIF was also used in a breast DCEMRI study (4) for data analysis of each acquisition, though comparison of extracted pharmacokinetic parameters resulted from the use of Avg-AIF and Ind-AIFs were not described. The Avg-AIF approach probably works the best for the extremity region that is far away from the heart, with the dispersion effects reducing intersubject differences in Ind-AIF. For anatomical regions closer to the heart, such as head, neck, and breast, the use of Avg-AIF may cause errors in kinetic analysis due to large intersubject differences in Ind-AIF. It is important to note that the results of this study were obtained under the conditions where the osteosarcomas were located in the knee area and FIG. 5. Scatter plots of Ktrans parameters of all 16 patients obtained from pharmacokinetic modeling analyses using the average AIFs (Avg-AIFs) for 1 cc/s and 2 cc/s injection rates: (a) tumor tissue ROI analysis; (b) median value of histogram analysis of pixel Ktrans parameters within the ROI. 1188 the AIF data samplings were all performed in the femoral artery. Our experience has shown that AIFs sampled from the femoral artery for DCE-MRI data analysis of tumors located in the knee area are not similar to those sampled from different arteries adjacent to osteosarcomas in other anatomical regions, such as the ankle (Huang W, Wang Y, Koutcher JA., unpublished results). Therefore, an Avg-AIF used for knee area osteosarcoma DCE-MRI data analysis should not be used for the same purpose for ankle area osteosarcoma or tumors in other anatomical regions. We are currently evaluating AIFs in different anatomic regions (Dave A, Lee N, Stambuk H, Wang Y, Huang W, Koutcher JA., unpublished results). In Ind-AIF measurements from a major artery, partial volume effects are largely avoided by selecting the image slice that contains the central portion of the artery and placing the ROI for AIF sampling well within the artery. Despite the use of short TR values employed in the spoiled gradient echo type sequence, the in-flow effects are minimized because the image slice for AIF measurement is usually located near the central position of the multislice acquisition volume and the ROI for AIF sampling is placed as far away from the edge of the FOV as possible. Clinical requirement for diagnosis often dictates large imaging spatial coverage of the TOI and high image inplane resolution, which result in poorer temporal resolution for DCE-MRI acquisition. It is reported by Roberts et al. (23) that insufficient data sampling (poor temporal resolution) of the single-bolus AIF may lead to large errors in kinetic modeling of DCE-MRI data. Our approach of using limited-population-based Avg-AIF provides opportunity to improve accuracy of quantitative analysis of osteosarcoma DCE-MRI data that were acquired with poorer temporal resolution, but otherwise with the same data acquisition scheme and contrast injection setup as the current study. The Avg-AIFs of 1 cc/s and 2 cc/s contrast injection rates in the femoral artery in the knee area have similar peak amplitude and decay curve shape, and the use of each for kinetic modeling of all 16 patient data sets does not seem to cause significant changes in derived pharmacokinetic parameters. This may be due to the small difference in the injection rate. Following bolus contrast administration with 2 cc/s injection rate (mostly in a peripheral vein) or with 1 cc/s injection rate (mostly in a central vein), the dispersion effects may be to such an extent, that by the time the Gd-DTPA contrast agent reaches the part of femoral artery in the knee area, the differences between 1-cc/s and 2 cc/s injection rates may not be detectable in AIF data sampling. It is quite possible that a very different AIF curve shape will be generated if a significantly higher contrast agent injection rate is employed. For this osteosarcoma DCE-MRI study, only the 1 cc/s and 2 cc/s injection rates are employed clinically depending on the location and size of IV catheter. The findings in this work indicate that Avg-AIFs derived from 1 cc/s and 2 cc/s contrast injection rates may be used interchangeably for kinetic modeling of DCE-MRI data when individual AIF measurement is not feasible. For quality control purposes, the contrast agent delivery process should be standardized for the DCE MRI study, such as the use of a power injector, the timing of injection, the volume of saline flush, etc. Wang et al. With the limited study population of 16 patients, the AvgAIF of each injection rate was obtained from only five Ind-AIFs. Further validation of the Avg-AIF approach with a larger population and in multiple regions is desirable. In this study, a biexponential decay function was used to fit the AIF data points. In a DCE-MRI study of the knee in children with juvenile rheumatoid arthritis (24), the AIF data points were sampled from a popliteal artery and fitted with three different functions: triexponential, gamma-variate plus biexponential, and biexponential. Statistically significant differences in extracted pharmacokinetic parameters were found between the biexponential fitting and the other two fittings, suggesting that choice of curve fitting function for AIF data points can cause systemic errors. However, the ultimate goal of our osteosarcoma DCE-MRI study is to assess the efficacy of chemotherapy as a longitudinal study prior to definitive surgery. As long as the biexponential function is consistently used for AIF curve fitting, any systematic error that might be introduced by the selection of such function should not impede the longitudinal comparisons of changes in pharmacokinetic parameters that are caused by chemotherapy treatment. ACKNOWLEDGMENTS We thank Drs. Amita Dave and Yousef Mazaheri Tehrani for providing the data for the T1 calibration curve, and Ms. Melissa Potuzak for managing the patient data base and the IRB protocol. REFERENCES 1. Leach MO, Brindle KM, Evelhoch JL, Griffiths JR, Horsman MR, Jackson A, Jayson GC, Judson IR, Knopp MV, Maxwell RJ, McIntyre D, Padhani AR, Price P, Rathbone R, Rustin GJ, Tofts PS, Tozer GM, Vennart W, Waterton JC, Williams SR, Workman P. 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 –1610. 2. Knopp MV, Weiss E, Sinn HP, Mattern J, Junkermann H, Radeleff J, Magener A, Brix G, Delorme S, Zuna I, van Kaick G. Pathophysiologic basis of contrast enhancement in breast tumors. J Magn Reson Imaging 1999;10:260 –266. 3. Buckley DL, Roberts C, Parker GJM, Logue JP, Hutchinson CE. Prostate cancer: evaluation of vascular characteristics with dynamic contrastenhanced T1-weighted MR imaging—initial experience. Radiology 2004;233:709 –715. 4. Li X, Huang W, Yankeelov TE, Tudorica A, Rooney WD, Springer CS. Shutter-speed analysis of contrast reagent bolus-tracking data: preliminary observations in benign and malignant breast disease. Magn Reson Med 2005;53:724 –729. 5. Galbraith SM, Rustin GJ, Lodge MA, Taylor NJ, Stirling JJ, Jameson M, Thompson P, Hough D, Gumbrell L, Padhani AR. Effects of 5,6-dimethylxanthenone-4-acetic acid on human tumor microcirculation assessed by dynamic contrast-enhanced magnetic resonance imaging. J Clin Oncol 2002;20:3826 –3840. 6. Morgan B, Thomas AL, Drevs J, Hennig J, Buchert M, Jivan A, Horsfield MA, Mross K, Ball HA, Lee L, Mietlowski W, Fuxius S, Unger C, O’Byrne K, Henry A, Cherryman GR, Laurent D, Dugan M, Marme D, Steward WP. Dynamic contrast-enhanced magnetic resonance imaging as a biomarker for the pharmacological response of PTK787/ZK 222584, an inhibitor of the vascular endothelial growth factor receptor tyrosine kinases, in patients with advanced colorectal cancer and liver metastases: results from two phase I studies. J Clin Oncol 2003;21: 3955–3964. 7. Port RE, Knopp MV, Brix G. Dynamic contrast-enhanced MRI using Gd-DTPA: interindividual variability of the arterial input function and consequences for the assessment of kinetics in tumors. Magn Reson Med 2001;45:1030 –1038. AIF for Osteosarcoma DCE-MRI 8. Yankeelov TE, Rooney WD, Li X, Springer CS. Variation of the relaxographic “shutter-speed” for transcytolemmal water exchange affects the CR bolus-tracking curve shape. Magn Reson Med 2003;50:1151–1169. 9. Rijpkema M, Kaanders JH, Joosten FB, van der Kogel AJ, Heerschap A. Method for quantitative mapping of dynamic MRI contrast agent uptake in human tumors. J Magn Reson Imaging 2001;14:457– 463. 10. Parker GJ, Roberts C, Macdonald A, Buonaccorsi GA, Cheung S, Buckley DL, Jackson A, Watson Y, Davies K, Jayson GC. Experimentallyderived functional form for a population-averaged high-temporal-resolution arterial input function for dynamic contrast-enhanced MRI. Magn Reson Med 2006;56:993–1000. 11. Dyke JP, Panicek DM, Healey JH, Meyers PA, Huvos AG, Schwartz LH, Thaler HT, Tofts PS, Gorlick R, Koutcher JA, Ballon D. Osteogenic and Ewing sarcomas: estimation of necrotic fraction during induction chemotherapy with dynamic contrast-enhanced MR imaging. Radiology 2003;228:271–278. 12. Haase A. Snapshot FLASH MRI. Application to T1, T2, and chemicalshift imaging. Magn Reson Med 1990;13:77– 89. 13. Parker GJM, Suckling J, Tanner SF, Padhani AR, Revell PB, Husband JE, Leach MO. Probing tumor microvascularity by measurement, analysis and display of contrast agent uptake kinetics. J Magn Reson Imaging 1997;7:564 –574. 14. Yankeelov TE, Rooney WD, Huang W, Dyke JP, Li X, Tudorica A, Lee J-H, Koutcher JA, Springer CS. Evidence for shutter-speed variation in CR bolus-tracking studies of human pathology. NMR Biomed 2005;18: 173–185. 15. Weinmann HJ, Laniado M, Mutzel W. Pharmacokinetics of GdDTPA dimeglumine after intravenous injection into healthy volunteers. Physiol Chem Phys Med NMR 1984;16:167–172. 16. Tofts PS. Modeling tracer kinetics in dynamic Gd-DTPA MR imaging. J Magn Reson Imaging 1997;7:91–101. 1189 17. Li X, Rooney WD, Springer CS. A unified magnetic resonance imaging pharmacokinetic theory: intravascular and extracellular contrast reagents. Magn Reson Med 2005;54:1351–1359. 18. Yankeelov TE, Debusk LM, Billheimer DD, Luci JJ, Lin PC, Price RR, Gore JC. Repeatability of a reference region model for analysis of murine DCE-MRI data at 7T. J Magn Reson Imaging 2006;24:1140 – 1147. 19. Yankeelov TE, Luci JJ, Lepage M, Li R, Debusk L, Lin PC, Price RR, Gore JC. Quantitative pharmacokinetic analysis of DCE-MRI data without an arterial input function: a reference region model. Magn Reson Imaging 2005;23:519 –529. 20. Walker-Samuel S, Parker CC, Leach MO, Collins DJ. Reproducibility of reference tissue quantification of dynamic contrast-enhanced data: comparison with a fixed vascular input function. Phys Med Biol 2007; 52:75– 89. 21. Walker-Samuel S, Leach MO, Collins DJ. Reference tissue quantification of DCE-MRI data without a contrast agent calibration. Phys Med Biol 2007;52:589 – 601. 22. Yang C, Karczmar GS, Medved M, Stadler WM. Estimating the arterial input function using two reference tissues in dynamic contrast-enhanced MRI studies: fundamental concepts and simulations. Magn Reson Med 2004;52:1110 –1117. 23. Roberts C, Buckley DL, Parker GJ. Comparison of errors associated with single- and multi-bolus injection protocols in low-temporal-resolution dynamic contrast-enhanced tracer kinetic analysis. Magn Reson Med 2006;56:611– 619. 24. Workie DW, Dardzinski BJ. Quantifying dynamic contrast-enhanced MRI of the knee in children with juvenile rheumatoid arthritis using an arterial input function (AIF) extracted from popliteal artery enhancement, and the effect of the choice of the AIF on the kinetic parameters. Magn Reson Med 2005;54:560 –568.