Background: Intrathecal drug delivery is an efficient method to administer therapeutic molecules to the central nervous system. However, even with identical drug dosage and administration mode, the extent of drug distribution in vivo is highly variable and difficult to control. Different cerebrospinal fluid (CSF) pulsatility from patient to patient may lead to different drug distribution. Medical image-based computational fluid dynamics (miCFD) is used to construct a patient-specific model to quantify drug transport as a function of a spectrum of physiological CSF pulsations.
Methods: Magnetic resonance imaging (MRI) and CINE MRI were performed to capture the patient's central nervous system anatomy and CSF pulsatile flow velocities. An miCFD model was reconstructed from these MRIs and the patient's CSF flow velocities were computed. The effect of CSF pulsatility (frequency and stroke volume) was investigated for a bolus injection of a model drug at the L2 vertebral level. Drug distribution profiles along the entire spine were computed for different heart rates: 43, 60, and 120 bpm, and varied CSF stroke volumes: 1, 2, and 3 mL. To assess toxicity risk for patients with different physiological variables, therapeutic and toxic concentration thresholds for a common anesthetic were derived from experimental studies. Toxicity risk analysis was performed for an injection of a spinal anesthetic for patients with different heart rates and CSF stroke volumes.
Results: Both heart rate and CSF stroke volume of the patient strongly influence drug distribution administered intrathecally. Doubling the heart rate (from 60 to 120 bpm) caused a 26.4% decrease in peak concentration in CSF after injection. Doubling the CSF stroke volume diminished the peak concentration after injection by 38.1%. Computations show that potentially toxic peak concentrations due to injection can be avoided by changing the infusion rate. Using slower infusion rates could avoid high peak concentrations in CSF while maintaining drug concentrations above the therapeutic threshold.
Conclusions: Our computations identify key variables for patient to patient variability in drug distribution in the spine observed clinically. The speed of drug transport is strongly affected by the frequency and magnitude of CSF pulsations. Toxicity risks associated with an injection can be reduced for a particular patient by adjusting the infusion variables with our rigorous miCFD model.