Human in silico trials for parametric computational fluid dynamics investigation of cerebrospinal fluid drug delivery: impact of injection location, injection protocol, and physiology.

BACKGROUND: Intrathecal drug delivery has a significant role in pain management and central nervous system (CNS) disease therapeutics. A fluid-physics based tool to assist clinicians in choosing specific drug doses to the spine or brain may help improve treatment schedules. METHODS: This study applied computational fluid dynamics (CFD) and in vitro model verification to assess intrathecal drug delivery in an anatomically idealized model of the human CSF system with key anatomic features of the CNS. Key parameters analyzed included the role of (a) injection location including lumbar puncture (LP), cisterna magna (CM) and intracerebroventricular (ICV), (b) LP injection rate, injection volume, and flush volume, (c) physiologic factors including cardiac-induced and deep respiration-induced CSF stroke volume increase. Simulations were conducted for 3-h post-injection and used to quantify spatial-temporal tracer concentration, regional area under the curve (AUC), time to maximum concentration (Tmax), and maximum concentration (Cmax), for each case. RESULTS: CM and ICV increased AUC to brain regions by ~ 2 logs compared to all other simulations. A 3X increase in bolus volume and addition of a 5 mL flush both increased intracranial AUC to the brain up to 2X compared to a baseline 5 mL LP injection. In contrast, a 5X increase in bolus rate (25 mL/min) did not improve tracer exposure to the brain. An increase in cardiac and respiratory CSF movement improved tracer spread to the brain, basal cistern, and cerebellum up to ~ 2 logs compared to the baseline LP injection. CONCLUSION: The computational modeling approach provides ability to conduct in silico trials representative of CSF injection protocols. Taken together, the findings indicate a strong potential for delivery protocols to be optimized to reach a target region(s) of the spine and/or brain with a needed therapeutic dose. Parametric modification of bolus rate/volume and flush volume was found to have impact on tracer distribution; albeit to a smaller degree than injection location, with CM and ICV injections resulting in greater therapeutic dose to brain regions compared to LP. CSF stroke volume and frequency both played an important role and may potentially have a greater impact than the modest changes in LP injection protocols analyzed such as bolus rate, volume, and flush.

Investigation of Human Intrathecal Solute Transport Dynamics Using a Novel in vitro Cerebrospinal Fluid System Analog.

Background: Understanding the relationship between cerebrospinal fluid (CSF) dynamics and intrathecal drug delivery (ITDD) injection parameters is essential to improve treatment of central nervous system (CNS) disorders. Methods: An anatomically detailed in vitro model of the complete CSF system was constructed. Patient-specific cardiac- and respiratory-induced CSF oscillations were input to the model in the subarachnoid space and within the ventricles. CSF production was input at the lateral ventricles and CSF absorption at the superior sagittal sinus. A model small molecule simulated drug product containing fluorescein was imaged within the system over a period of 3-h post-lumbar ITDD injections and used to quantify the impact of (a) bolus injection volume and rate, (b) post-injection flush volume, rate, and timing, (c) injection location, and (d) type of injection device. For each experiment, neuraxial distribution of fluorescein in terms of spatial temporal concentration, area-under-the-curve (AUC), and percent of injected dose (%ID) to the brain was quantified at a time point 3-h post-injection. Results: For all experiments conducted with ITDD administration in the lumbar spine, %ID to the brain did not exceed 11.6% at a time point 3-h post-injection. Addition of a 12 mL flush slightly increased solute transport to the brain up to +3.9%ID compared to without a flush (p < 0.01). Implantation of a lumbar catheter with the tip at an equivalent location to the lumbar placed needle, but with rostral tip orientation, resulted in a small improvement of 1.5%ID to the brain (p < 0.05). An increase of bolus volume from 5 to 20 mL improved solute transport to the brain from 5.0 to 6.3%ID, but this improvement was not statistically significant. Increasing bolus injection rate from 5 to 13.3 mL/min lacked improvement of solute transport to the brain, with a value of 6.3 compared to 5.7%ID. Conclusion: The in vitro modeling approach allowed precisely controlled and repeatable parametric investigation of ITDD injection protocols and devices. In combination, the results predict that parametric changes in lumbar spine ITDD-injection related parameters and devices can alter %ID to the brain and be tuned to optimize therapeutic benefit to CNS targets.