Arterio-venous anastomoses in mice affect perfusion measurements with dynamic contrast enhanced CT

Accurate measurement of perfusion with dynamic contrast enhanced CT requires an arterial input curve (AIC) uncontaminated by venous sources. Arterio-venous anastomoses (AVAs) are sources of contamination if contrast is injected intravenously. We seek to identify AVAs in mice and associated errors in perfusion measurements. Six transgenic mice with spontaneous prostate tumor were scanned with a micro-CT scanner (GE Healthcare (GE)) using a high resolution anatomical and a lower resolution perfusion protocol. For the anatomical protocol, a CT scan was performed during injection of an iodinated contrast agent (Hypaque) into a tail vein. Images covering the thoracic, abdominal and pelvic regions at an isotropic resolution of 175 microm were reconstructed and rendered in 3D to show the arterial and venous tree (Advantage Window, GE). For the perfusion protocol, each mouse was continuously scanned for 40 s and the contrast agent (Hypaque) was injected via a tail vein 5 s into scanning. Tumor images were reconstructed every second. Tumor blood flow (BF) and volume (BV) maps were calculated with CT perfusion software (GE) using AIC measured either from abdominal aorta (AA) or tail (caudal) artery (TA). In all mice, there was an AVA from the bifurcation of the inferior vena cava to the tail artery shunting venous blood and portion of the contrast agent injected into the tail vein into the TA. Contrast arrival time at the TA preceded that at the AA by 3.3 +/- 0.5 s (P < 0.05). Mean tumor BV and BF values calculated with AA versus TA were 10.0 +/- 1.8 versus 4.8 +/- 2.1 ml (100 g)(-1) (P < 0.05) and 108.8 +/- 26.5 versus 33.0 +/- 8.5 ml min(-1) 100 g(-1) (P < 0.05), respectively. AVA in the murine pelvic region can result in inaccurate and more variable measurements of pelvic organ/tissue perfusion when the tail artery is used as the AIC.