Tracking accuracy of T2- and diffusion-weighted magnetic resonance imaging for infusate distribution by convection-enhanced delivery

Abstract

Object: Because convection-enhanced delivery relies on bulk flow of fluid in the interstitial spaces, MR imaging techniques that detect extracellular fluid and fluid movement may be useful for tracking convective drug distribution. To determine the tracking accuracy of T2-weighted and diffusion-weighted MR imaging sequences, the authors followed convective distribution of radiolabeled compounds using these imaging sequences in nonhuman primates.

Methods: Three nonhuman primates underwent thalamic convective infusions (5 infusions) with (14)C-sucrose (MW 342 D) or (14)C-dextran (MW 70,000 D) during serial MR imaging (T2- and diffusion-weighted imaging). Imaging, histological, and autoradiographic findings were analyzed.

Results: Real-time T2- and diffusion-weighted imaging clearly demonstrated the region of infusion, and serial images revealed progressive filling of the bilateral thalami during infusion. Imaging analysis for T2- and diffusion-weighted sequences revealed that the tissue volume of distribution (Vd) increased linearly with volume of infusion (Vi; R(2) = 0.94, R(2) = 0.91). Magnetic resonance imaging analysis demonstrated that the mean ± SD Vd/Vi ratios for T2-weighted (3.6 ± 0.5) and diffusion-weighted (3.3 ± 0.4) imaging were similar (p = 0.5). While (14)C-sucrose and (14)C-dextran were homogeneously distributed over the infused region, autoradiographic analysis revealed that T2-weighted and diffusion-weighted imaging significantly underestimated the Vd of both (14)C-sucrose (mean differences 51.3% and 52.3%, respectively; p = 0.02) and (14)C-dextran (mean differences 49.3% and 59.6%; respectively, p = 0.001).

Conclusions: Real-time T2- and diffusion-weighted MR imaging significantly underestimate tissue Vd during convection-enhanced delivery over a wide range of molecular sizes. Application of these imaging modalities may lead to inaccurate estimation of convective drug distribution.

Fig. 1

Coronal T2-weighted MR images obtained in a primate brain after infusion of 15 μl (A), 30 μl (B), and 60 μl (C) of ¹⁴C-dextran. Infusion of ¹⁴C-dextran solution provides a distinct region of T2-weighted hyperintensity in each thalamus compared with surrounding noninfused parenchyma.

Fig. 2

Coronal diffusion-weighted MR images obtained in a primate brain after infusion of 15 μl (A), 30 μl (B), and 60 μl (C) of ¹⁴C-dextran. Infusion of ¹⁴C-dextran solution provides a distinct region of diffusion-weighted hyperintensity in each thalamus compared with surrounding noninfused parenchyma.

Fig. 3

Graph demonstrating a linear relationship (R² = 0.94) between Vd with T2-weighted MR imaging and Vi in primates infused with ¹⁴C-sucrose and ¹⁴C-dextran. The mean Vd/Vi ratio was 3.6 ± 0.5 (mean ± SD).

Fig. 4

Graph demonstrating a linear relationship (R² = 0.91) between Vd with diffusion-weighted MR imaging and Vi in primates infused with ¹⁴C-sucrose and ¹⁴C-dextran. The mean Vd/Vi ratio was 3.3 ± 0.4 (mean ± SD).

Fig. 5

Coronal T2-weighted (A) and diffusion-weighted (B) MR images obtained after CED of 88 μl of radiolabeled ¹⁴C-dextran into the right thalamus. The insets show the corresponding autoradiograms of ¹⁴C-dextran distribution and demonstrates the spatial inaccuracy and underestimation of T2-weighted and diffusion-weighted MR imaging volumes of distribution compared with drug distribution determined by autoradiography.

Fig. 6

Autoradiographic analysis of ¹⁴C-dextran concentration across the infused region. The concentration profile, based on specific activity, was square-shaped with a steep drop-off at the margins indicative of a relatively uniform concentration over the infused region.