Introduction

Quantification of the dynamic arterial input function (AIF) in small animals is challenging in dynamic contrast-enhanced (DCE-)MRI and in radiotracer studies. Only few examples of direct AIF-measurements of MR contrast agent (CA) in mice are published in the literature^1-4, each featuring significant limitations.

Methods

Nine Intracranial tumor-bearing mice (female NMRI Nude, 10 -12 w.o; U87, PSMA-expressing) were anaesthetized (isoflurane, fentanyl & midazolam) and received an extracorporeal shunt from the femoral artery to the tail vein. MRI scanning was performed using a 9.4 T MRI (Bruker BioSpec) with a CryoProbe. The extracorporeal line featured two reservoirs (glass capillaries, 0.94 mm inner diameter) positioned on the skull in the MRI field of view (Figure 1). A MR-compatible measuring chamber featuring a β-microprobe (biospace lab) for β-emitting radiotracers was included in the extracorporeal shunt. Dynamic MRI scanning of the head was performed for 15 minutes using a 3D FLASH sequence (80 x 80 x 8, TR 5.019 ms, TE 1.961 ms, flipangle 15°) with a spatial resolution of 0.175 x 0.175 x 1 mm³ and a temporal resolution of 4.015 s. A 100 µl solution containing CA (Gadovist, 35 µmol/ml) and (for mice with simultaneous injection) 10-20 MBq F-18-PSMA-1007 was injected into the tail vein at 1 ml/min for 6 s. Dispersion correction for MRI CA was performed based on the recorded distinct dispersion effect at the two interspaced reservoirs. In vitro calibration measurements were performed with defined Gadovist concentrations in human blood, which was circulated with defined flow velocities using an injection pump. Calculation of the gadolinium concentration (c_Gd) was performed by the formula c_Gd = (S_post-S_pre) / (S_pre x T₁₀ x r₁)⁵, with S_pre and S_post denoting the pre- and post-contrast signal intensities, respectively, T₁₀ denoting the T1 of pre-contrast blood (2.4 s at 37 °C and 9.4 T, adapted from Dobre et al.⁶) and r₁ the estimated relaxivity at 37 °C and 9.4 T (4.1 s⁻¹·mM⁻¹, extrapolated from Shen et al.⁷). PET scanning was performed using the same animal bed as for MRI scanning in a quadHIDAC small animal PET scanner in list-mode for 35 minutes.Typical transfer delay between DCE-MRI and PET measurement was 5-10 minutes. Blood was immediately withdrawn after the end of DCE-scanning for quantification of c_Gd using mass spectrometry, for calibration of the β-Microprobe measurements by gamma counting and hematocrite measurement.

Results

The CA AIFs of nine recorded mice show little noise and typical AIF curve shapes after dispersion correction (Figure 2, A & C). Eight of nine mice show a close range of peak concentrations (0.55 – 0.85 µmol/ml) and shunt flow velocities (34-58 µl/min) (Figure 3). β-emitting radioactive tracer AIFs can be simultaneously recorded using a MR-compatible β-Microprobe (Figure 2 A) and mice were transferred into the PET-scanner immediately after DCE-MRI (Figure 2 B). Significant inverse correlation between AIF maxima and the delays between the CA-influx into the two reservoirs was observed (r = -0.84) (Figure 3 A). The time constant τ for monoexponential deconvolution was significantly positively correlated with the delay (r = 0.98) (Figure 3 B). The results of mass spectrometry validation show a systematic and consistent underestimation of the image-derived concentrations (4 mice, ratio mass spectrometry-derived / image-derived: 1.57-1.80) (Figure 2 C). However, the MR-based quantification shows good agreement with circulated human blood with defined CA concentrations in the range of expected concentrations and flow velocities (Figure 4).

Discussion

Our method allows for parallel measurements of the AIF of MR CA and PET radiotracer as well as CA dynamics in tissue. AIF curves generated with our method appear to be robust and show low level of noise. The approach should be well transferable to any tissue/region of interest and makes measurements of the AIF independent of specific local constraints. A major limitation of our approach is the need for surgical arterial catheterization which impedes longitudinal studies. The quantitative precision of the MRI-based CA quantification needs further evaluation as discrepancies exist between calibration measurements (Figure 4) and validation by mass spectrometry (Figure 2 C).

Conclusions

We present a novel approach for DCE-measurements of the AIF in mice with conceivable potential compared to so far published methods. Moreover, we present the first dual recordings of AIFs of a MR CA and a PET tracer in mice. This supports evaluation approaches to deduce the CA/PET tracer AIF from one another. Further, it might provide the basis for simultaneous and integrated modeling of PET tracer and CA kinetics in mice, which is of high interest in integrated, simultaneous small animal PET/MRI.