Synopsis

Keywords: Quantitative Imaging, Perfusion

Calculation of perfusion parameters from DCE-MRI using pharmacokinetic modeling depends on accurate determination of the dynamic tissue concentration and the dynamic arterial blood concentration (AIF). However, precise AIFs are difficult to measure in small animals. We introduce a novel extracorporeal circulation approach from the external carotid artery for feasible AIF determination. DCE derived AIF and tissue contrast agent concentrations corresponded well with radioactive analogs and mass spectrometry. Perfusion parameters derived by different pharmacokinetic models demonstrated variable agreement with simultaneous radiotracer derived calculations of modeling parameters. Overall, we introduce an accurate and precise framework for quantitative perfusion MRI in small animals.

INTRODUCTION

Calculation of perfusion parameters from DCE-MRI using pharmacokinetic modeling (PKM) depends on the accurate determination of the dynamic tissue concentration and the dynamic arterial blood concentration (arterial input function, AIF). Particularly precise AIFs are difficult to measure in small animals. Extracorporeal AIF measurements from the femoral artery may provide a solution but require delay and dispersion correction and validation remains lacking¹. Here, we introduce a novel access way from the external carotid artery with likely negligible delay and dispersion. We show extensive validation of each step of this DCE MRI and pharmacokinetic modeling (PKM) framework using radioactive agents and mass spectrometry to set a basis for quantitatively robust perfusion MRI in small animals.

METHODS

Overall 25 NMRI U87 intracranial glioma bearing nude mice were measured in a 9.4 T Bruker BioSpec MRI. In 100 µl, either 35 mM Gd-DO3A-butrol (Gadovist) or Gd-DTPA (Magnevist) were mixed with their radioactive analogs ⁶⁸Ga-DO3A-butrol (5.8 ± 2.8 MBq) or ^99mTc-DTPA (24.2 ± 1.3 MBq) or with ^99mTc-labeled red blood cells (13.7 ± 2.7 MBq) or ^99mTc-labeled human serum albumin (13.6 ± 0.4 MBq) and simultaneously injected. An extracorporeal circulation was applied shunting the blood from the external carotid artery to the ipsilateral internal jugular vein. At 1.6 µl volumetric distance to the carotid artery, the circulation featured an MR measurement reservoir . Further downstream, blood radioactivity was assessed using an MR compatible measuring unit (Swisstrace, Twilite). A Golden-angle Radial Sparse Parallel (GRASP) sequence (0.195 x 0.195 x 0.195 mm³, 5 s/frame, TR/TE 4/0.011 ms, FA 13.4°)², combined with a compressed sensing MP2RAGE sequence for T1 mapping (0.195 x 0.195 x 0.195 mm³, TI1/TI2/MP2RAGE TR = 800 ms/2200 ms/6250 ms, TE/TR = 3.253/7 ms, FA 7°)³, was used for DCE-MRI (Figure 1). Contrast agent concentrations were calculated as previously described¹ and were based on relaxation rates of r1 = 3.68 mM^−1s−1 for Gadovist⁴ and r1 = 2.94 mM^−1s−1 for Magnevist⁵. T1 values were individually measured for extracorporeal blood and for tissue. DCE scanning time was approximately 15.5 min. After sacrifice, brains were acutely harvested, sliced and mapped with autoradiography (n=15) and laser ablation-inductively coupled plasma-mass spectrometry (LA-ICP-MS) (n=5). Autoradiography slices were quantified by a simultaneously measured standard⁶. Plasma distribution volume was calculated based either on 99mTc-labeled red blood cells (RBC) or 99mTc-labelled human serum albumin (HSA) and hematocrit⁶. Subsequently, voxel wise pharmacokinetic modeling was performed with the software rocketship⁷ using the extended tofts model (ETM) and the two compartment exchange model (2CXM).

RESULTS

In 6 mice, mean blood c[Gd] extracted from the extracorporeal circulation at the end of the measurement was 0.28 ± 0.08 %ID/g determined by MRI and 0.29 ± 0.08 %ID/g determined by gamma counting together with 0.28 ± 0.08 %ID/g by ICP-MS. AIFs showed a high degree of visual correspondence in consideration of the delay and dispersion resulting from the longer extracorporeal traveling distance to the radiation detector (Figure 2). In 5 mice, calculation of mean tissue c[Gd] for tumor ROIs based on DCE-MRI resulted in values of 0.55 ± 0.18 %ID/g whereas same ROIs showed 0.60 ± 0.16 %ID/g for autoradiography of co-injected radioactive analogs and 0.72 ± 0.16 %ID/g for LA-ICP-MS (Figure 3). Parametric maps of the extravascular extracellular space fractional volume v_e, the volume transfer constant K_trans, and the plasma flow F_P showed plausible values in comparison to previous studies¹ (Figure 4). Only minor differences were found between the parameters v_e and K_trans comparing the ETM and the 2CXM. However, V_p was highly different between the two models and ~5-fold higher in the ETM compared to the 2XCM. Analysis of autoradiography of acute slices of ^99mTc labelled RBC and HSA demonstrated estimations of V_p that were ~2-fold lower for RBCs compared to HSA in tumors. The absolute autoradiography estimates of both radiotracers for V_p more closely matched with the modeling results of the 2XCM compared to the values of the ETM (Figure 5).

DISCUSSION

Dynamic validation of the DCE-MRI AIF with radioactive analogs demonstrated high correspondence indicating quantitative accuracy of our extracorporeal circulation approach. Shunting blood from the external carotid artery as opposed to the previously used femoral artery¹ resulted in a much smaller artificially introduced delay and dispersion, which may obviate the need for correction. End-of-measurement validation of MR CA concentration in tumor and healthy tissue with radioactive analogues and LA-ICP-MS also showed good agreement between DCE and ex-vivo methods. In our preliminary PKM analysis, the encountered realistic values for v_e, K_trans, and F_P underline the appropriateness of measured AIFs and tissue CA concentrations and make a strong impact of dispersion on the extracorporeal AIF unlikely. With autoradiography for determination of V_p, we were able to test the appropriateness of different DCE-models to quantify V_p, which was highly different comparing the 2CXM and ETM. Our results point toward a better match of radiotracer derived Vp with the 2XCM.

CONCLUSION

Stepwise validation of perfusion MRI using contrast agent analogs for AIF determination combined with radiotracers and mass spectrometry for comparison of tissue concentration and parameter maps pointed toward high quantitative accuracy using our small animal DCE-MRI framework employing an external carotid artery shunt for AIF determination without delay and dispersion correction.

Acknowledgements

The Project was supported by the SFB CRC 1450 and the DFG-NAMT application 446292030 of Dr. med Philipp Backhaus. The project critically relied on support of technical staff, especially Roman Priebe, Dirk Reinhardt, Nina Kreienkamp and Irmgard Hoppe. Furthermore, we thank Zhang et al. and Trotier et al. for providing DCE and T1-mapping sequences.