Pharmacokinetics of Intrathecal Oligodeoxynucleotides

Studies were approved by the University of Washington Institutional Animal Care and Use Committee (Seattle, Washington). Twenty-one farm-bred pigs of both sexes, weighing 13.5 ± 2.6 kg, were used. Each of three oligodeoxynucleotides was administered to seven animals.

Anesthesia was induced by intramuscular injection of ketamine, the animal was paralyzed with intramuscular succinylcholine (200 mg), the trachea was intubated, and the animal was mechanically ventilated to maintain end-tidal carbon dioxide at 34–40 mmHg (Datex Airway Gas Analyzer Type GAO; Datex, Helsinki, Finland). Anesthesia was maintained throughout with isoflurane (1.5 – 2.0%) in oxygen. Femoral arterial and venous catheters were placed via  cutdown, and a continuous infusion of normal saline containing 0.05 mg/ml pancuronium bromide was infused through the venous catheter at 4 ml · kg⁻¹· h⁻¹throughout the study.

Laminotomies (approximately 0.5 cm²) were performed in the L4, L1, and T11 vertebrae to expose the spinal cord. At each of these vertebral levels, an incision of approximately 1.0 mm was made through the dura and arachnoid mater to permit insertion of custom-made microdialysis probes into the subarachnoid space (see below for description of microdialysis probes). The microdialysis probe at the L4 level was attached to an epidural catheter (Perifix; B. Braun, Bethlehem, PA) to permit intrathecal oligodeoxynucleotide injection. The incision through the spinal meninges was sealed by means of an epoxy cone plug at the base of the dialysis probe, and cyanoacrylate glue was used to hold the probes in place. A fourth microdialysis probe was inserted into the epidural space opposite the L4 intrathecal probe. The laminotomies were sealed with cyanoacrylate glue to restore the integrity of the epidural space.

Microdialysis probes were perfused with bicarbonate-buffered mock CSF (140 mEq NaCl, 25 mEq NaHCO³, 0.4 mEq MgCl², 3.5 mEq urea, 4.0 mEq glucose, 2.0 mEq CaCl²; pH = 7.38–7.42; 292–298 mOsm). A syringe pump (model 22; Harvard Apparatus Inc., Holliston, MA) was used to pump mock CSF through the dialysis probes at the rate of 10 μl/min.

Microdialysis probe manufacture has been previously described. Briefly, polysulfone dialysis fibers (MicroKros; Spectrum Laboratories, Inc., Rancho Dominguez, CA) with a diameter of 0.5 mm and a pore size of 50 nm were coated with epoxy over all but the center 20 mm, thereby creating a 31.4-mm²dialysis window. A 16.5-μm-diameter tungsten wire (Hamilton Company, Reno, NV) was inserted through the lumen of the dialysis fiber, and the fiber was bent in half to form a dialysis loop. The intraluminal wire prevented kinking of the dialysis probe, thereby maintaining luminal patency. The nondialyzing body of the dialysis probe was coated with a cone-shaped bead of epoxy to fix the shape of the probe and to seal the meningeal hole through which the dialysis probe was inserted. Probes were allowed to “cure” for a minimum of 18 h before use.

Dialysis probes were tested in vitro  to determine whether there was any difference in recovery efficiency for the differently sized oligodeoxynucleotides. Specifically, a dialysis probe was placed in a solution with a known concentration of one of the oligodeoxynucleotides, and the solution was dialyzed for 10 min. The oligodeoxynucleotide concentration in the dialysate was divided by the concentration in the dialyzed solution to determine the “recovery efficiency.” Seven probes were tested against each of the three oligodeoxynucleotides.

The three oligodeoxynucleotides used were complimentary to the portion of the human multiple-drug resistance (MDR-1) mRNA beginning at the start codon and extending 10, 18, or 30 bases into the coding sequence. The specific oligodeoxynucleotide sequences, CAAGATCCAT, GTCCCCTTCAAGATCCAT, and TCCTCCATTGCGGTCCCCTTCAAGATCCAT, were purchased from Gibco BRL (Invitrogen Life Technologies, Carlsbad, CA). The oligodeoxynucleotides were radiolabeled according to the manufacturer’s protocol using the RTS T⁴Kinase Labeling System (Invitrogen Life Technologies) and [γ-thio-³⁵S]adenosine triphosphate (4.63 × 10¹³Bq/mm; NEN Life Science Products, Inc., Boston, MA; specific activity). For the 30-base oligodeoxynucleotide, unincorporated [³⁵S]adenosine triphosphate was removed using the QIAquick Nucleotide Removal Kit (Qiagen Inc., Valencia, CA) according to the manufacturer’s protocol. For the 18- and 10-base-long oligodeoxynucleotides, unincorporated [³⁵S]adenosine triphosphate was removed using Microspin G-25 columns (Amersham Biosciences, Piscataway, NJ) according to the manufacturer’s protocol.

After completion of the surgical preparation, baseline dialysate samples were collected. The study oligodeoxynucleotide was diluted in mock CSF to a final volume of 250 μl and injected into the CSF over 1 min. The doses administered averaged 263 nmol for the 10-base oligodeoxynucleotide, 79 nmol for the 18-base oligodeoxynucleotide, and 50 nmol for the 30-base oligodeoxynucleotide. Differences in the molar dose injected reflect the fact that a constant mass of oligodeoxynucleotide was injected.

Dialysate samples were collected into sterile 1.5-ml microcentrifuge tubes over 5-min intervals for 20 min (50-μl samples), over 10-min intervals until 60 min (100-μl samples), and over 20-min intervals until 180 min (200-μl samples).

Ten milliliters of scintillation fluid (Ecoscint; National Diagnostics, Atlanta, GA) was added to each aliquot of the dialysate sample, and the amount of ³⁵S-labeled oligodeoxynucleotide present was quantified in a Tri-Carb 2002 (Packard Instruments, Downer’s Grove, IL) liquid scintillation counter. All samples were counted for 5 min or until the SD of the disintegrations per minute was less than 2%. All samples were corrected for background disintegrations per minute.

To determine whether significant metabolism of the oligodeoxynucleotides occurred in the CSF, aliquots of representative sets of microdialysate samples for each of the three oligodeoxynucleotides were analyzed using polyacrylamide–urea gel electrophoresis. As a control, the “parent” radiolabeled oligodeoxynucleotide was also loaded on each gel. After removing urea by soaking gels in 10% aqueous acetic acid, they were dried in vacuo  at 80°C and autoradiographed.

Noncompartmental analysis of the oligodeoxynucleotide concentration–time data was performed using PK Solutions 2.0 software (Summit Research Services, Montrose, CO). PK Solutions 2.0 uses the trapezoidal rule to determine the area under the concentration-versus -time data and from the area calculates area under the curve (AUC), mean residence time, volume of distribution (V^d), volume of distribution at steady state (V^ss), and clearance. For all data relying on AUC, we used the AUC extrapolated to infinity. However, the extrapolation was small and accounted for less than 5% of the total area in nearly all cases. Terminal elimination half-lives and the initial volume of the central compartment (V^c) were obtained by stripping the concentration–time plots and fitting an exponential curve using the method of least squares in the PK Solutions software.

Statview 5 software (SAS Institute Inc., Cary, NC) was used for all statistical analyses. Differences in pharmacokinetic parameters were assessed for statistical significance using two-way analysis of variance with oligodeoxynucleotide and sampling site as the independent variables. The Fisher protected least significant difference test was used for post hoc  testing. A P  value of 0.05 was considered statistically significant.

The in vitro  studies performed to determine whether there was a difference in the ability of the dialysis probes to recover the different-length oligodeoxynucleotides showed that the dialysate concentration of the short oligodeoxynucleotide averaged 22 ± 7% of the concentration in the dialyzed solution. For the intermediate-length oligodeoxynucleotide, the dialysate concentration averaged 18 ± 8% of the concentration in the dialyzed solution, and for the long-length oligodeoxynucleotide, the dialysate concentration averaged 21 ± 10% of the concentration in the dialyzed solution. These differences were not statistically significant; therefore, oligodeoxynucleotide concentration was not corrected for differences in recovery.

Polyacrylamide–urea gel electrophoresis of representative samples of all three oligodeoxynucleotides showed a single, discrete band of the same molecular weight as the parent radiolabeled oligodeoxynucleotide (control lane) at each time point seen (fig. 1). The single band of the same molecular weight as the starting oligodeoxynucleotide indicates the absence of autoradiographic evidence of oligodeoxynucleotide degradation or metabolism within the CSF.

Figures 2–5show both the individual and the average concentration-versus -time data for each oligodeoxynucleotide at the L4, L1, and T11 intrathecal and the L4 epidural sampling sites, respectively. Table 1lists the descriptive pharmacokinetic parameters that were derived from the individual concentration-versus -time plots.

There were significant differences in dose-normalized AUC among the three oligodeoxynucleotides at all sample sites. At the L4 and L1 sites, there was a significant and strong inverse linear relation between oligodeoxynucleotide length and AUC (fig. 6). At the more rostral T11 and the epidural sampling sites, there was not a simple relation between oligodeoxynucleotide length and AUC.

At the intrathecal injection site (L4), there were significant differences in V^c, V^ss, V^d, and clearance among the three oligodeoxynucleotides. There was also a strong and statistically significant linear relation between oligodeoxynucleotide length and V^cand V^ss(fig. 7) and an exponential relation between V^d(fig. 8) and clearance (fig. 9).

Neither mean residence time nor terminal elimination half-life differed significantly among the three oligodeoxynucleotides at any of the four sampling sites, although the power for these analyses was relatively low (0.09–0.48).

The purpose of this study was to investigate the pharmacokinetics of intrathecally administered oligodeoxynucleotides in anticipation that these compounds are likely to be administered spinally in humans in the not-too-distant future. Evidence that intrathecally administered oligodeoxynucleotides will someday play a role in human medicine comes from multiple sources. The National Institute of Neurologic Disorders and Stroke recently solicited Research Funding Proposals for “Gene Expression Profiling in the Nervous System Following Traumatic Spinal Cord Injury” (BAA-RFP-NIH-NINDS-01-03). The stated objective of this Research Funding Proposal is to identify patterns of gene expression that have both negative and positive influences on outcome after spinal cord injury, with the expectation that future targets for oligodeoxynucleotide-based intervention will be identified. Another example of the future role of antisense oligodeoxynucleotide in treating CNS disease comes from the recent introduction of the term molecular neurosurgery  to refer to the use of genetic techniques to treat surgical diseases, such as CNS cancer. 12–18In addition, animal models suggest that antisense oligodeoxynucleotides may offer a means of treating human CNS diseases, such as multiple sclerosis. 9 

Perhaps the most striking finding from this study is the degree to which several pharmacokinetic variables are related to oligodeoxynucleotide length. Because nucleotide bases differ somewhat in size and molecular weight, the number of bases is not a perfect predictor of either size or weight. However, the fact that oligodeoxynucleotide length produced such strong correlations with AUC, apparent V^d, V^c, V^ssand clearance (r = 0.97–1.0) suggests that oligodeoxynucleotide length is likely a better predictor of these pharmacokinetic parameters than is molecular weight or molecular volume.

Why length should be such a strong predictor of these pharmacokinetic parameters is not clear from the data. It may simply be that the rate at which the clearance process proceeds is related to the length of the oligodeoxynucleotide. Alternatively, the clearance mechanism may differ for oligodeoxynucleotides of different lengths, and the underlying processes may proceed at different rates. Because there was no evidence of oligodeoxynucleotide metabolism, size-related differences in metabolic rate cannot be invoked to explain differences in AUC, clearance, or apparent volumes of distribution. It is also unlikely that differences in diffusion rates explain the relation with oligodeoxynucleotide length because diffusion rate is inversely proportional to the square root of the molecular weight of a molecule. If differences in diffusion rate played a major role in oligodeoxynucleotide pharmacokinetics, one would expect variables such as clearance to be inversely related to oligodeoxynucleotide length, which was not the case.

Although the AUC was significantly greater for the shorter oligodeoxynucleotides compared with the long oligodeoxynucleotide at the more distant T11 sampling site, the relation between AUC and oligodeoxynucleotide length was not as strong as at the sampling sites closer to the site of oligodeoxynucleotide injection. The reason for this is not entirely clear but may reflect the fact that oligodeoxynucleotide concentration at this site is more dependent on the rate of oligodeoxynucleotide movement from the injection site to the sampling site than is the case with the L1 and L4 sampling sites. Regardless of the explanation, the important observation is still that the longest oligodeoxynucleotide has the lowest AUC at all sampling sites.

More striking was that in the epidural space the AUC was so much less for the short oligodeoxynucleotide than for the two longer oligodeoxynucleotides. Whether this reflects more rapid clearance of the smaller oligodeoxynucleotide from the epidural space or reduced penetration of the spinal meninges is unclear.

The potential clinical significance of the strong dependence of some pharmacokinetic variables on oligodeoxynucleotide length is that it offers the potential of manipulating oligodeoxynucleotide pharmacokinetics for clinical advantage. For example, given that shorter oligodeoxynucleotide lengths persisted longer in the CSF, the use of a shorter oligodeoxynucleotide would be expected to result in greater distribution in the CSF, with correspondingly greater exposure of the underlying spinal cord to the oligodeoxynucleotide. Conversely, longer oligodeoxynucleotide sequences would be expected to have a more limited distribution. Of course, the antisense efficacy of oligodeoxynucleotides is related to their length; therefore there are limits to the extent that the length of any oligodeoxynucleotide can be manipulated without losing antisense efficacy. However, within those limits, which vary among oligodeoxynucleotides, changes in length may be an effective means to influence intrathecal distribution.

In summary, this is the first study to examine the pharmacokinetic behavior of oligodeoxynucleotides in the intrathecal space. The principal finding was that intrathecal but not epidural pharmacokinetics are largely determined by oligodeoxynucleotide length. This may be of significance in the future when designing oligodeoxynucleotides for clinical use.