Contrast-enhanced magnetic resonance imaging in a spinal epidural tumor model.

Abstract

A spinal epidural tumor model was developed, using the VX-2 adenocarcinoma in rabbits, to assess the strengths and weaknesses of magnetic resonance (MR) as a cross-sectional imaging modality for the evaluation of epidural neoplastic disease. High-resolution MR images were acquired both before and after intravenous gadolinium chelate injection, assessing lesion detectability and efficacy of imaging technique. An adenocarcinoma tumor (VX-2) was produced in the epidural space of six New Zealand White rabbits and subsequently studied on a 1.5 tesla whole body MR scanner. VX-2 tumor tissue was removed from the thigh of a carrier rabbit, minced, and screened. Under fluoroscopic guidance, 0.2 mL of the tumor preparation was then injected into the epidural space of the experimental rabbits. The injection was performed at the L5-6 level using an epidural needle and polyethylene tubing sleeved within the needle. The rabbits were imaged using a circular small parts surface coil 5 to 15 days after the epidural injection. In all six animals, one complete MR exam was obtained within the time frame of days 9 to 11. T1- and T2-weighted axial scans were obtained before contrast injection, with the T1 scans acquired both with and without fat saturation. Postcontrast T1 scans also were obtained, using fat saturation, after the injection of 0.1 and 0.3 (cumulative dose) mmol/kg gadoteridol (Gd HP-DO3A; ProHance) in all animals. The film images were interpreted in a prospective fashion by a single neuroradiologist who was masked to imaging technique and contrast dosing. The digital data was analyzed by region of interest measurement. At the end of the imaging studies, the animals were sacrificed and the epidural lesion confirmed by gross and microscopic exam. On a prospective masked read of the MR films, epidural tumor was depicted best on postcontrast fat saturation T1-weighted scans using a cumulative contrast dose of 0.3 mmol/kg. Substantial contrast enhancement of the tumor was observed in all instances on postcontrast scans. The precontrast T1-weighted scan was least efficacious for lesion identification and differentiation from the compressed spinal cord. Depending on the pulse sequence used, one (T2-weighted) to three (T1-weighted without fat saturation) of the lesions could not be identified prospectively on precontrast scans. Lesion growth with time after implantation was chronicled by MR imaging, accompanied by progression of symptoms. On region of interest analysis, differentiation of epidural tumor from normal cord was greatest (11.6 +/- 6.1) on postcontrast scans using a cumulative contrast dose of 0.3 mmol/kg. The level of differentiation achieved was twice that of postcontrast scans using a contrast dose of 0.1 mmol/kg (5.9 +/- 3.6). These results were superior on statistical analysis to that with all other scan techniques (P = 0.002-0.0005). Cord and tumor could not be differentiated on the basis of signal intensity, with any statistical significance, using precontrast T1 and T2 scans. The lesions were confirmed in each animal by gross and microscopic exam. On inspection of the gross specimen, the tumors were noted to be located in the epidural space and to cause cord compression. On microscopic exam, the tumor was composed of epithelial cells that were moderately pleomorphic. In the New Zealand White rabbit, an epidural tumor could be created consistently using the described percutaneous approach. These lesions are suitable for MR imaging studies, examining lesion detectability and efficacy of imaging technique. The lesions created in the current study could not be diagnosed prospectively in all cases on precontrast T1 and T2 scans images. Postcontrast scans were most efficacious for diagnosis and lesion delineation, with high-dose (0.3 mmol/kg) scans superior to standard dose (0.1 mmol/kg).