Convective forces increase rostral delivery of intrathecal antisense oligonucleotides in the cynomolgus monkey nervous system

Background The intrathecal (IT) dosing route introduces drugs directly into the CSF to bypass the blood-brain barrier and gain direct access to the CNS. We evaluated the use of convective forces acting on the cerebrospinal fluid as a means for increasing rostral delivery of IT dosed radioactive tracer molecules and antisense oligonucleotides (ASO) in the monkey CNS. We also measured the cerebral spinal fluid (CSF) volume in a group of cynomolgus monkeys. Methods There are three studies presented, in each of which cynomolgus monkeys were injected into the IT space with radioactive tracer molecules and/or ASO by lumbar puncture in either a low or high volume. The first study used the radioactive tracer 64Cu-DOTA and PET imaging to evaluate the effect of the convective forces. The second study combined the injection of the radioactive tracer 99mTc-DTPA and ASO, then used SPECT imaging and ex vivo tissue analysis of the effects of convective forces to bridge between the tracer and the ASO distributions. The third experiment evaluated the effects of different injection volumes on the distribution of an ASO. In the course of performing these studies we also measured the CSF volume in the subject monkeys by Magnetic Resonance Imaging. Results It was consistently found that larger bolus dose volumes produced greater rostral distribution along the neuraxis. Thoracic percussive treatment also increased rostral distribution of low volume injections. There was little added benefit on distribution by combining the thoracic percussive treatment with the high-volume injection. The CSF volume of the monkeys was found to be 11.9 ± 1.6 cm3. Conclusions These results indicate that increasing convective forces after IT injection increases distribution of molecules up the neuraxis. In particular, the use of high IT injection volumes will be useful to increase rostral CNS distribution of therapeutic ASOs for CNS diseases in the clinic.


BACKGROUND
The development of novel CNS treatment modalities can be hampered by the lack of blood brain barrier (BBB) permeability. The intrathecal (IT) dosing route allows for direct delivery of therapeutic molecules to the central nervous system (CNS), bypassing the blood brain barrier. The most practical routine clinical route for IT drug delivery is via lumbar puncture, which introduces drug into the lumbar CSF cistern at the caudal end of the neuraxis. IT administration has been successfully used to deliver the antisense oligonucleotide (ASO) therapeutic Spinraza to patients with Spinal Muscular Atrophy (1). Additionally, several other BBB impermeable therapeutic molecules in development, including proteins, nucleic acids, viral gene therapy vectors, stem cells and exosomes are pursuing use of the IT dosing route to target diseases of the CNS. Since these approaches aim to treat various diseases in both pediatric and adult patient populations that can be associated with significant variability in inter-subject anatomy and CSF volumes, common principles are needed to optimize the IT dosing procedure for broad neuraxial delivery. Approaches that increase neuraxial drug exposure after lumbar IT delivery will be useful as ASO therapies, and other intrathecally delivered modalities, advance into indications that involve more rostral brain structures.
Broad CNS tissue drug delivery following IT dosing requires neuraxial movement within the subarachnoid CSF followed by entry into the CNS parenchyma (2). Drug movement within the subarachnoid space is largely facilitated by several CSF convective mechanisms including CSF turnover, cardiac motion, respiratory thoracic motion, and body movement (3)(4)(5)(6). We recently demonstrated in rodent studies that convective force generated in the CSF by the IT dosing procedure itself can be leveraged to increase the spread of drug along the neuraxis (7). Here, we evaluate the convection enhancing effects of different dosing bolus volumes and externally applied percussive force on the neuraxial distribution and pharmacodynamic effect of ASOs delivered via IT lumbar puncture in cynomolgus monkeys with the goal of enhancing the cranial distribution and widespread CNS action of therapeutic ASOs.
To evaluate the effects of these convection inducing factors on IT drug delivery in a larger species, we conducted a series of experiments in non-human primates following IT delivery of imaging agents and ASOs. First, we quantified neuraxial distribution after IT drug delivery by PET imaging of the small molecule 64 Copper-dodecane tetraacetic acid ( 64 Cu-DOTA) (468 g/mole molecular weight), both with and without the use of convective forces. We then replicated these efforts with an ASO that was co-injected with 99m Tecnicium diethylenetriaminepentaacetic acid ( 99m Tc-DTPA) to allow for SPECT imaging to define the early distribution effects of the convective forces. Then the distribution, pharmacokinetics and pharmacodynamics of the ASO were evaluated and compared with the 99m Tc-DTPA imaging data. Finally, using a therapeutically relevant ASO targeting microtubule associated protein tau (MAPT), with the potential to treat Alzheimer's disease and frontotemporal dementia (8), we confirmed that increasing convective forces can increase rostral CNS delivery of ASOs and imaging agents.

Animals
Adult cynomolgus monkeys were used in the experiments described herein. Six were used for the 64 Cu-DOTA experiment, 24 were used for the 99m  animals were used, 5 female and 5 male, their body weights were between 3.0 and 4.5 kg. For all the animals, fluorescent lighting was provided via an automatic timer for 12 hours per day, tap water was supplied ad libitum via an automatic water system and they received an adequate supply of primate chow except during designated fasting periods.

Anesthesia
Each animal was fasted for 4-12 hours prior to induction of anesthesia with ketamine (10-25 mg/kg. IM). Following anesthesia induction, animals were placed on isoflurane (1-2%) in oxygen carrier gas for the lumbar puncture, percussive wrap treatment when indicated, and imaging when indicated. Vital signs were monitored throughout the procedure. Lactated Ringer's Solution was administered in a 5mL subcutaneous (SC) bolus both pre-and post-imaging. Each animal was also given intramuscular (IM) Metacam (0.2mg/kg, IM) and Ceftiofur (2.2mg/kg, IM).

Magnetic Resonance Imaging (MRI)
Whole body MRI was acquired from each animal to measure CSF volume. All animals were imaged on a 1.5T Siemens Symphony MRI scanner (Siemens Medical Systems, Erlangen, Germany).
High-resolution MRI of the spine and cranium were acquired in separate acquisitions within a single imaging session while each animal was under anesthesia. Images were acquired in axial orientation using a 3D T2 fast-spin echo (FSE) sequence with one excitation. Cranium data were acquired with a clinical CP extremity coil, voxel size of 0.7 x 0.7 x 2 mm, acquisition time of 8.05 min, TR of 12.5 ms, and TE of 6.25 ms. Body data were acquired with a clinical body coil, voxel size of 0.35 x 0.35 x 2 mm, acquisition time of 13.97 min, relaxation time of 11.82 ms, and excitation time of 5.91 ms.

MRI data analysis
The cranium and body MRI data were stitched, including alignment, co-registration, and intensity blending in overlapping slices. Coarse, manually defined regions-of-interest (ROIs) were used to segment the brain and spine, including nearby surrounding tissue. Sub-regions with relative homogeneous CSF intensity within the brain and coarse spine ROI regions were selected. Sub-regionspecific intensity thresholds were used to segment the final CSF ROI. In sub-regions with insufficient contrast, the CSF was segmented manually using MRI intensity as a guide.

Lumbar Puncture Procedure
After each animal was anesthetized, an experienced veterinary surgeon injected the test articles to the intrathecal space by lumbar puncture. In the experiments performed at MPI, the animals were placed in left lateral recumbence and a 22-gauge Quinke spinal needle was introduced into the L4/L5 intrathecal space using aseptic technique. In the Mapt experiment performed at NBR, the animals were placed vertically in a seated position and the torso extended over a circular form. A 25-gauge Pencan Paed® pencil-point needle for pediatric use (B Braun) was inserted into the L4/L5 intervertebral space and used to inject the dosing solution. In all cases, the placement of the needle was verified by the presence of CSF at the needle hub pre-and post-injection. Once placement of the needle was verified, test article was injected over ~2 minutes/mL volume, and then the needle was withdrawn.

Percussive wrap and high frequency chest wall oscillation
In the 64 Cu-DOTA and 99m Tc-DTPA/ASO experiments, a percussive wrap that delivers highfrequency chest wall oscillation (HFCWO) therapy (SmartVest ® , Electromed, Inc., New Prague, MN) was applied to each animal after IT injection. The SmartVest ® system is an airway clearance system prescribed for patients with compromised airway clearance (such as in cystic fibrosis or spinal muscular atrophy). The wearable wrap consists of an inflatable bladder connected to an air pulse generating system. A size "S-Small" Single-Patient Use SmartVest Wrap ® with wrap height of 10.5 cm (4-1/4") was used on all monkeys. Treatments were conducted at 5 Hz and 10 psi. Under percussive vest treatment conditions, animals were anaesthetized, test article was injected IT, then the wrap was applied to each animal's torso. HFCWO treatment was applied for 30 minutes where indicated, otherwise the animal remained in the wrap for 30 minutes without activation of the air pulse mechanism. Following percussive wrap treatment, the wrap was removed, and each animal was placed in the PET or SPECT scanner for imaging.
DOTA chelator was purchased from Macrocyclics (Lot M14010001-070713). 64 Cu was supplied in 0.1M HCl by Washington University (St. Louis, MO). 64 Cu-DOTA was prepared prior to each imaging session. Briefly, 0.1M, pH 6 citrate buffer, and 64 Cu were added to a glass vial.
Following the addition of radioactivity, chelator solution (1.0 mg/mL in 0.1M citrate buffer, pH 6) was added to the vial. The vial was vortexed for 1 min then placed in a water bath at 50°C for 30 min.
Following chelation complex formation, two instant thin layer chromatography (ITLC) runs were conducted in parallel to confirm chelation efficiency.
For ITLC, a small aliquot of the 64 Cu-chelator solution was diluted 1:100 in 200 µL. 2 µL were drawn up by pipette and dispensed onto the bottom of the ITLC paper strip and allowed to dry. Once dry, the ITLC strip was placed in a 15 mL conical tube containing 800 µL of ITLC developing buffer (0.1M citrate buffer, pH 6) and chelation efficiency was determined. Chelation efficiency was consistently >99%. Specific activity (SA) was 358.9 ± 166.5 MBq/mg (mean ± SD, n=8 runs).

Cu-DOTA Study design
The tissue distribution of 64 Cu-DOTA radioactivity was evaluated in 6 cynomolgus monkeys after IT bolus delivery of ~18MBq 64 Cu-DOTA under 4 conditions. Each condition consisted of an IT delivery of 64 Cu-DOTA followed by dynamic, whole-body PET imaging. All animals were imaged under each condition in a crossover study design with 1-2 weeks between conditions. The tests on the different experimental days were low (0.36 mL) versus high (1.8 mL) injection volume, low injection volume versus low injection volume with 30-minute percussive wrap treatment, and high injection volume versus high injection volume with 30-minute percussive wrap treatment. There were no significant differences in molar mass of 64 Cu-DOTA injected between conditions. Following the experiments, the animals were returned to the MPI colony.

PET Image Acquisition
Whole-body continuous bed motion positron emission tomography (PET) and computed tomography (CT) data were acquired on a Focus220 microPET (Siemens Medical Systems, Knoxville, TN) and CereTom CT (NeuroLogica Corp, Danvers, MA), respectively. All animals were imaged in the head-first prone position. Radioactive fiducial markers were placed on the bed at three different positions for each scan to support co-registration of the PET time frames and as a standard of radioactivity. Dynamic PET data were acquired for either 0-120 minutes in scans without percussive treatment or 30-120 minutes in scans with percussive treatment PET data were acquired in 3D listmode and re-binned into 2D sinograms. PET images were reconstructed by a 2D Ordered Subset Expectation Maximization (OSEM2D) algorithm with 14 subsets and 4 iterations. Corrections were made for detector normalization, decay, dead-time, random coincidences, and attenuation into images with 256x256x693 pixels and 0.95x0.95x0.80 mm voxels.
After each PET scan, a CT was acquired for anatomical registration. All animals were imaged on a specially design bed, which was transferred from the PET to the CT to allow consistent positioning of the animal between the two modalities. CT acquisition time was 6 minutes and based on an axial range of 450 mm, a tube peak voltage of 120 kVp, with 720 projections per rotation, and 4 seconds per projection.
Estimation of tissue uptake of 64 Cu-DOTA ROIs were defined for the cranium, heart, liver, and kidneys by fitting ellipsoids of fixed volume to each organ based on CT. The cranium ROI contained both parenchyma and CSF within and surrounding the brain. The bladder ROI was defined through automated thresholding of the PET signal in the bladder or drawn by hand based on anatomy (in cases of low bladder signal). Spinal column ROIs were defined by applying a combination of manual and automated segmentation thresholds to the CT. The CSF ROI included both the spinal cord and CSF within the IT space. The CSF ROI was divided into cervical, thoracic, and lumbar sub-regions based on vertebral level.
Quantitative data from these ROIs were extracted and radioactivity concentration in units of percent injected dose per gram (%ID/g) at each time point for each ROI was plotted (assuming 1 cm 3 is equivalent to 1 g of tissue). The area-under-the-curve (AUC) was calculated for each ROI. For AUC comparisons from conditions with and without percussive wrap treatment, AUCs for all regions were calculated from 30-120 min of data post-injection. Plots of radioactivity concentration (%ID/g) over the length of the spine were also generated for each ROI. All analyses were performed in VivoQuant™

SPECT Image Acquisition
Immediately after IT 99m Tc-DTPA delivery, the animals were maintained under anesthesia and SPECT data were acquired using a SPECT/CT scanner (NanoSPECT/CT, Bioscan, Inc.) for 10 minutes (time 0 scan). Following this scan, each animal was placed in the percussive wrap for a 30 minute percussive wrap treatment (either activated or not) then imaged again with the SPECT scanner for 30 minutes (time 40 minute scan). The animals were allowed to recover from the anesthesia and then were re-anesthetized for a final 30 minute SPECT scan at 6 hours post IT injection. The animals were again allowed to recover from anesthesia. All SPECT scans were followed immediately by a CT scan.
The anatomical scan range was head to mid torso using a UHR parallel-hole collimator with 96 projections and an energy window of 126.5 -154.6 keV. SPECT projection images were reconstructed in the software ReSPECT v 2.5 (SciVis, Germany) using a maximum likelihood estimation method (MLEM) with 6 iterations. Object background threshold was set to 5 and an attention coefficient of 0.12/cm was used. No smoothing was used in reconstruction and data were reconstructed into 1 mm isotropic voxels.
CT scans were performed from head to mid torso region on a CereTom CT (NeuroLogica Corp, Danvers, MA). Acquisition time was 604 seconds with tube peak voltage of 120 kVp, current set to 4 mA, 288 projections per rotation, and 6 seconds per projection.

SPECT Image Processing and Image Generation
A single quantification factor (QF) was calculated and applied to each SPECT image. Three fiducial markers, consisting of vials containing a known amount of activity were included in each SPECT scan. Each fiducial marker was manually segmented from the unprocessed reconstructed SPECT image, and the given activity in that fiducial marker (decay-corrected to the scan time) was divided by the counts from the segmentation to obtain a QF for that fiducial. The QFs for the three fiducials were then averaged to obtain the QF for the image.
SPECT and CT images at each time point were co-registered. Maximum intensity projection (MIP) images of co-registered SPECT and CT were generated with SPECT data scaled from 0 to 1% ID/g. Image processing and MIP generation was performed in VivoQuant 2.0. Full quantitative analysis of the image data was not performed.

MALAT1 RNA PCR Analysis
Tissue punches (~12 mm 3 ) were taken from fresh frozen brain and spinal cord slices and homogenized with sterile ceramic beads in guanidinium thiocyanate buffer containing 8% 2mercaptoethanol using a bead homogenizer. Total RNA was prepared from tissue lysates using a RNeasy 96 kit (Qiagen). The prepared RNA was assayed for MALAT1 and cyclophilin A levels using primer/TaqMan probe sets with an EXPRESS One-Step SuperScript quantitative reverse transcriptase polymerase chain reaction (qRT-PCR) kit (Invitrogen). qRT-PCR plates were run on StepOnePlus Real-Time PCR machines (Applied Biosystems) and data was initially analyzed using StepOne Software (Applied Biosystems). MALAT1 RNA expression level was normalized to cyclophilin A mRNA expression to correct for the amount of RNA in the reaction. Malat1 RNA expression level was further normalized as percent naïve control.
Malat1 primers/probe sequences: Control tissue homogenate for curves was made by weighing control monkey brain and adding homogenization buffer at a 9 to 1 ratio. Five hundred microliter aliquots were pipetted into a 96-well plate and appropriate amounts of calibration standards were spiked in the corresponding wells. Wells containing samples and calibration standards then had internal standard (IS) and approximately 0.25cm³ granite beads added. The plates were then extracted via a liquid-liquid extraction with ammonium hydroxide and phenol:chloroform:isoamyl alcohol (25:24:1). The aqueous layer was then further processed via solid phase extraction on a Strata X plate (Phenomonex Inc., CA). Eluates had a final pass through a protein precipitation plate before they were dried under nitrogen at 50°C.
Dried samples were reconstituted in 140µL water containing 100µM EDTA. An Agilent liquid chromatograph with mass spectrometry detection (LC-MS/MS) instrument consisting of a 1290 binary pump, a column oven, an auto sampler, and a 6460 triple quadrupole mass spectrometer was used for analysis (Agilent, Wilmington, DE, USA).
The tissue extracts were injected onto a Kinetex analytical column (Phenomenex, 100 x 2.1 mm, 2.6 µm particle size; 100 Å) which was equilibrated with 15% methanol in 5mM triethyalamine (TEA) and 400mM hexo-fluoro-isopropanol (HFIP) and then maintained at 55ºC and with a flow rate of 0.3 mL/min throughout the analysis. A gradient from 15 to 50% methanol over 7 minutes, increased to 80% over 0.5 minutes, and maintained at 80% for 0.5 minutes was used to separate the Malat1 ASO and internal standard (IS) from background peaks. A re-equilibration time of 2 minutes at the starting conditions was used between samples.
All mass measurements were made on-line with the scan time set from 2 to 8 minutes.
During that time window, the mass spectrometer was set to scan for MRM transitions for the full length ASO and IS (-8m/z, 893.1->94.8; and -4m/z, 1160.7->94.8, respectively). Mass spectra were obtained using a spray voltage of -1500 V, a nebulizer gas flow of 25 psig, a sheath gas flow rate of 12 L/min at 350ºC, a drying gas flow rate of 5 L/min at 250 ºC, and a capillary voltage of -3750 V. Chromatograms were analyzed using Agilent Mass Hunter software.
Tissue calibration curves were constructed using peak area ratios of the ASO to the IS and applying a weighted (1/x) Quadratic regression. All tissues had a calibration range for the ASO from 0.036 -178.82 µg/g in 50 mg monkey brain homogenate. A minimum signal-to-noise ratio of 5:1 was used to distinguish ASO peaks from background. Acceptance criteria for the calibration curve were set to 85-115% of nominal values. All samples were stored at -70°C ± 10°C, upon receipt.

RESULTS: MRI determination of CSF volume in Cynomolgus monkeys
To support the selection of IT dose volumes, total CSF volumes were measured in cynomolgus monkeys. The CSF volume in 20 animals (10 males and 10 females, 3.6 + 0.4 kg body weight, no significant differences between male and female body weights), was 11.6 ± 1.5 cm 3 for the entire neuraxis including the ventricles (Table 1). Total CSF volume was 12.1 ± 1.6 cm 3 in males and 11.0 females in CSF volume (Table 1).
To aid in translation of this work, we chose the experimental dosing volumes in NHP to be a similar percentage of total CSF volume as dose volumes used previously in human patients. High intrathecal dose volumes used in humans have corresponded to approximately 14% of total adult human CSF volume and volumes of up to 33% of the total CSF volume have previously been given safely in humans (9-11). We therefore evaluated the effect of IT dose volumes ranging from ~3%-

Exclusion of Animals from data analysis
In some cases, in the 64 Cu-DOTA and the 99m Tc-DTPA/ASO experiments it was determined that there were IT doses missed due to technical failures. These animals were removed from further analysis. The following are the criteria used for the exclusion of these animals.
In the 64 Cu-DOTA experiment, the surgeon assessed all IT injections to be successful, except for the injection for one subject under the low volume plus wrap condition. Despite the apparent success of the injections, the image data suggested that at least some injections were multicompartmental (e.g. IT and epidural space) or only partially in the IT space ( Figure 1A). Basic in-life tolerability assessments were included to confirm the tolerability of the manipulations, and that they did not alter vital signs that may contribute to CSF dynamics. The test article, 64 Cu-DOTA, was well-tolerated. Percussion treatments, delivered using a percussive wrap and HFCWO, were also well-tolerated. Vital signs were within normal ranges for cynomolgus monkeys: heart rate was 132 ± 23 BPM, respiratory rate was 31 ± 8 respirations/min, and body temperature was 96.9 ± 1.9 ºF.
The AUC of the 64 Cu-DOTA was observed in the lumbar spine near the injection site with the concentration decreasing along a caudal to rostral gradient. Comparing the low and high-volume IT injection conditions, there were no significant differences between the 64 Cu-DOTA AUC values in the lumbar spinal cord, the site nearest to the injection, of the two volumes (Figure 2A). The AUC values in the cranium and the thoracic spine were significantly higher in the high-volume condition than the low volume condition. There was also a nonsignificant trend towards higher cervical spinal concentrations with the higher volume compared to the low volume injection (P = 0.0624).
To determine if percussive wrap treatment could further influence distribution, we compared low and high volumes of 64 Cu-DOTA with percussive wrap treatment ( Figure 2B and 2C After IT co-injecting 99m Tc-DTPA with MALAT1 ASO the animals were imaged using SPECT at 0 min (prior to percussive wrap treatment), 40 min (after percussive wrap treatment) and 6 hours post-injection. The images were limited to the cranial, cervical, and thoracic spinal cord due to limitations of the SPECT system's field of view. In the low injection volume with no percussive wrap group ( Figure 3A) the radioisotope progressed up the spinal column. Addition of percussive wrap treatment to the low dose volume (figure 3B) increased the rostral distribution of the radioisotope signal into the cervical spine and brainstem area. Increasing the dose volume from low to high with no percussive wrap treatment ( Figure 3C) resulted in greater rostral distribution in the base of the brain and into the transverse fissure between cerebellum and cerebrum and the lateral fissure between the temporal and frontal lobes. Addition of percussive wrap treatment to the high dose volume did not appreciably affect the distribution of the radioisotope over high volume alone ( Figure 3D).
To determine the distribution of ASO into tissues, samples from the lumbar, thoracic, cervical spinal cord, and frontal cortex were harvested 7 days after injection and analyzed for concentrations of ASO. Tissue concentrations across the CNS are lower in the low volume group ( Figure 4A) than those in the tissues of the animals dosed with the high dose volume ( Figure 4C).
Consistent with the SPECT imaging (Figure 3), the rostral neuraxial distribution appeared to be increased by increasing the dose volume, where the percussive wrap treatment was beneficial in the low dose volume group ( Figure 4B versus 4A), but in this experiment, not in the high dose volume groups ( Figure 4D versus 4C). When the ASO concentrations in the different CNS tissues were analyzed across treatments, no significant differences were detected, however there was a trend towards higher tissue concentrations in the Frontal cortex when low volume versus high volume were compared (Table 2). Tissue samples adjacent to those used for the ASO tissue concentration determinations were processed and analyzed for MALAT1 RNA expression by qRT-PCR. As the MALAT1 ASO is designed to suppression MALAT1 RNA, this is a measure of pharmacodynamic effect. Consistent with the drug concentrations, MALAT1 RNA expression is significantly suppressed from naïve expression levels in the spinal cord, but not in the frontal cortex in the low volume and no wrap conditions ( Figure 4E, Table 3). However, when the percussive wrap treatment was added to the low volume injection, RNA suppression occurs throughout the CNS structures, including the frontal cortex which is distal from the injection site ( Figure 4F, Table 3). With high dose volume alone ( Figure 4G) and with high dose volume and percussive wrap treatment ( Figure 4H), more uniform suppression of expression of the target RNA can be observed across the CNS with significant reductions of expression compared to naïve animals (Table 3). This pattern of RNA suppression is similar between the two different high dose volume groups ( Figures 4G and H, Table 3) suggesting that the addition of percussive forces does not add any distribution benefit to the high dose volume for the distribution of the ASO to rostral structures. The MALAT1 RNA suppression in the frontal cortex in both of the high dose volume treatments are similar to the low dose volume with percussive wrap treatment ( Figures 4G, H and F) and dissimilar to the low dose volume without percussive wrap treatment ( Figure 4E) suggesting that the addition of percussive wrap increases the rostral distribution of the ASO with low dose volume equal to the increase seen with the high dose volume injection. Taken together, either increasing dose volume, or adding percussive wrap treatment to the low injection volume, could increase rostral distribution of ASOs.  Broad pharmacology following high volume delivery of a MAPT-ASO Suppression of human MAPT has been proposed as a potential therapy for tauopathies, including Alzheimer's disease and frontotemporal dementia (8). Indeed, ASOs suppressing Mapt in rodent models of disease can reverse pathology and ameliorate phenotype (8). Since many of the potential indications for a MAPT ASO likely require cortical suppression, we evaluated whether we could increase distribution and pharmacology of a MAPT ASO that targets monkey MAPT mRNA by using a high volume (2 mL) compared to a low dose volume (0.8 mL). We found a trend toward increased ASO concentrations in the more rostral structures in the high-volume group compared to the low volume group ( Figure 6A). Consistent with the ASO concentration data, increasing the dose volume significantly increased the pharmacological action of the ASO, with greater reduction in MAPT mRNA in the high-volume group compared to the low volume group ( Figure 6B). Taken together, using a therapeutically relevant modality, increasing the convective forces within the CSF following an IT drug delivery can increase ASO concentration and pharmacological action throughout the brain.

DISCUSSION
We embarked on this series of experiments to evaluate how manipulation of convective forces affected the rostral distribution of drug after lumbar intrathecal administration in the monkey. The objective of this work is to incorporate the lessons learned here in clinical practice to potentially increase the cranial distribution of IT delivered therapeutics. Our earlier work in rodents demonstrated that adjusting the volume of injection can increase neuraxial exposure distal to the lumbar IT injection site (7). There is a small body of literature that describes the effects of several convective mechanisms in the CNS space on CSF motion including CSF turnover, cardiac motion, respiratory thoracic motion, and body movement (3)(4)(5)(6). We chose injection volume and exogenous forces applied with a percussive wrap as potential modifiers of ASO distribution, as these would be easily incorporated into the clinical environment.
Although increasing dose volume has not been systematically taken advantage of in IT drug therapy, injected volumes of up to 50 mL (approximately 33% of the total CSF volume) have previously been given safely in humans. Moreover, myelography contrast injection guidelines call for up to 17 mL IT injection of CT contrast media in adults (9, 10).
In our first experiment, 64 Cu-DOTA was successfully injected IT and imaged in cynomolgus monkeys in high (1.8 mL) or low (0.4 mL) bolus volumes, with or without exogenous force applied to the thorax via a percussive wrap. There was an increase of rostral radioactivity distribution distal to the injection site with the larger bolus dose volume compared to the lower dose volume. Rostral radioactivity distribution was also increased with the addition of the percussive wrap in both the low and high dosing volumes.
These results then led to our next experiment dosing a pharmacological dose of an unlabeled ASO to the non-protein coding RNA MALAT1 spiked with 99m Tc-DTPA to bridge the previous radioactive imaging results with ASO distribution using dose volume and percussive wrap. We again found that the larger injection volume increased rostral distribution of both the radioactivity and the MALAT1 ASO distal from the lumbar injection site. We also found that percussive wrap treatment increased the rostral distribution of the radioactivity and MALAT1 ASO above that seen with the low dose volume. However, percussive wrap did not seem to increase rostral distribution of the ASO or radioactivity above that seen with the high dose volume.
We then utilized the lessons learned from these two experiments and evaluated the distribution of an ASO against MAPT after lumbar intrathecal administration dosed in a low (0.8 mL) or high (2 mL) volume. We again found an increase in rostral distribution of the MAPT ASO distal from the lumbar injection site with the larger volume when compared to the smaller volume. The rostral distribution of the MAPT ASO to the cranium is important because this target is important in frontotemporal dementia and Alzheimer's disease, both neurodegenerative diseases of the brain.
These convective forces effects were also evident in the increase in the action of the ASO in the large dose volume animals compared to the low dose volume ones.
It is likely that both increased dose volume and mechanical percussion involve convection enhancement. In addition to increasing radioactive tracer concentrations in the more cranial regions of the neuraxis, the percussive wrap reduced tracer concentration in the lumbar region in the low volume condition. The use of increased dose volume and exogenous mechanical force may thus be useful for increasing IT therapeutic dose to cranial regions and reducing exposure at the lumbar site of delivery when required. Conversely, for IT dosed drugs targeted for local lumbar spinal cord or nerve root action where neuraxial spread is not desired, it may be useful to minimize convective force by using lower dose volumes.
We evaluated the CSF volume in a large group of cynomolgus monkeys using T2-weighted volumetric MRI methods. We found that the cynomolgus monkey has a total CSF volume of 11.9 + 1.6 mL (mean + standard deviation). This value was used to guide selection of IT bolus injection volumes to model published high and low intrathecal injection volumes delivered in the clinic.
Total CSF volume in humans is generally quoted as approximately 130 mL (12 They found a volume of 263 + 55 mL (mean + standard deviation) in a group of 15 normal human subjects, which is much larger than the traditional 130 mL volume. When we re-evaluate our injection volume percentages of total CSF volume using these new human data, they change from 3-7% to 1.4-3% for the low volume doses and from 15-20% to 7-10% for the high-volume doses. When allometrically scaling intrathecal dosing between monkey to human, what was previously a 130/11.9 mL, or an approximately 10-fold higher CSF volume in the human, now becomes a 263/11.9 mL or approximately 20-fold higher CSF volume in the human than in the monkey.
Clear assessment of the variability in the quality of IT 64 Cu-DOTA delivery by injection as revealed by PET imaging also illustrates the value of molecular imaging in assessing efficiency of IT therapeutics delivery. IT nuclear imaging, which has long been used to evaluate CSF leaks, may provide great value in understanding the efficiency and pharmacodynamics of IT dosed therapeutics.
Experiments in humans are required to assess the translatability of the volume principles demonstrated by the experiments within this manuscript. The assumption that rostral distribution as a function of percentage of CSF volume injected is consistent from NHP to human has yet to be tested.
Nuclear imaging will allow us to test these volume principles in humans. It may also be crucial in determining differences in intrathecal dynamics in different disease states, for instance where cerebral ventricular volumes are increased, or where cortical atrophy increased subarachnoid CSF space.    injection site (frontal cortex) on the right. Graph A has data from the animals treated with low volume (0.8 mL) with no percussive wrap treatment in green; graph B has data from the animals treated with low volume with percussive wrap treatment in blue; graph C has data from animals treated with high volume (2.4 mL) with no percussive wrap treatment in red; graph D has data from animals treated with high volume with percussive wrap treatment in orange; graph E has data from the animals treated with low volume (0.8 mL) with no percussive wrap treatment in green; graph F has data from the animals treated with low volume with percussive wrap treatment in blue; graph G has data from animals treated with high volume (2.4 mL) with no percussive wrap treatment in red; graph H has data from animals treated with high volume with percussive wrap treatment in orange.
Data in black is the MALAT1 RNA expression from the naïve animals.