Ziconotide and saline diluent were furnished by the Neurex Corporation (now Elan Corporation, South San Francisco, CA). Ziconotide for injection was prepared by aseptic dilution with saline diluent. Solutions were not filtered. Drug dilutions and reconstitutions were made just prior to dosing.

Male beagle dogs (N = 5) were obtained from Harlan Sprague Dawley, Inc. (Ridgeland Farms, Mt. Horeb, WI). These dogs were between 9 and 18 months old and weighed between 12 and 16.5 kg at the start of the study. All animals had free access to dry dog food, with the exception of the nights prior to scheduled clinical pathology, surgery, and sedation. Tap water was available ad libitum. Animals were individually housed in 4- foot by 7-foot inside runs.

After an initial acclimation (3–7 days), dogs underwent surgical implantation of intrathecal injection and sampling catheters to be used for repeated intrathecal dosing and sampling. Catheter placement was performed following strict aseptic precautions. All catheters were fabricated and packaged on site and locally sterilized by e-beam irradiation. Surgery was performed approximately 72 to 96 hr (study day −4 and −3) prior to initiation of dosing. Animals received atropine (0.04 mg/kg, IM) prior to sedation with xylazine (Rompun; 1.5 mg/kg, IM). Tribrissen (15–20 mg/kg, oral, twice daily for 5 days) was given as a prophylactic antibiotic. Anesthesia was mask-induced and the animals were then intubated and maintained under spontaneous ventilation with 1–2% Halothane and 60% N2O and 40%O2. Animals were continuously monitored for oxygen saturation, inspired and end-tidal gas, expired CO2, N2O, O2 and heart and respiratory rates. Surgical areas were shaved and prepped with chlorhexadine scrub and solution. Dexamethasone sodium phosphate (0.25 mg/kg, IM) was given immediately following completion of intrathecal catheter placements.

Using sterile technique, a skin incision was made on the lower back of the neck. The cisterna magna was then exposed by combined blunt and sharp dissection. The dura was then exposed and two small incisions (0.5 mm) were made and the prepared catheters were then individually inserted and passed caudally. The catheters were fabricated of polyethylene PE10 (0.4 mm O.D.), for the infusion catheter, and PE50 (0.97 mm O.D.), for the sampling catheter (Intramedic©, Becton Dickinson). The PE10 catheter was passed approximately 40 cm to the L_2–3 level. The PE50 catheter was passed through an adjacent incision approximately 38 cm. The internal segments of the catheters were then connected to 50 cm PE100 external segments and tunneled subcutaneously and caudally to exit on the upper left and right back at the level of the scapula. Confirmation of the appropriate placement of the catheters in the intrathecal space was based on the free outflow of CSF. The incision was closed in layers using 3–0 Vicryl®. With closure of the incision, anesthesia gases were turned off and the animals allowed to recover under observation. Butorphanol tartrate (Torbugesic; 0.04 mg/kg, IM) was administered to relieve post-operative discomfort. A nylon vest (Alice King Chatham Medical Arts, Hawthorne, CA) was placed on each dog following surgical recovery and a Panomat C-10 infusion pump (Disetronics Medical Systems, Plymouth, MN) was placed in the left vest pocket and connected to the PE50 sampling catheter. The pump was set to deliver approximately 100 µL/hr of 0.9% w/v Sodium Chloride for Injection, USP, to maintain sampling catheter patency.

Each dog received: i) one intrathecal bolus injection of ziconotide (10 µg in 1 mL), ii) chronic intrathecal infusion over two sequential 48-hr intervals of two concentrations of ziconotide (1 µg/hr IT followed by 5 µg/hr IT, respectively, at 100 µl/hr), and iii) lastly, an intravenous bolus injection of ziconotide (0.1 mg/kg). All intrathecal bolus injections were spiked with 1 µCi ³H-inulin and followed by 0.3 mL 0.9% (w/v) Sodium Chloride for Injection, USP, as catheter flush.

Animals were observed for arousal, motor function, and motor strength on a 4-point scale (22). Morbidity/mortality and clinical observations were noted twice daily (morning and late afternoon). On dosing days, the morning clinical observations were conducted prior to dosing. Appetite, presence of stool and urine, general behavior, overt signs of toxicity, and body temperatures were recorded daily.

Heart rate and blood pressures were measured using a tail cuff manometer (Dinamap 8100, Critikon). Respiratory rates were measured by visualization of chest expansion and contraction. These measurements were made periodically before and after each injection and during infusion intervals.

The thermally evoked nociceptive skin twitch response was measured using a probe with approximately 1 cm surface area maintained at approximately 62.5°C ± 0.5°C with a feedback. (24) The probe was applied to shaven lumbar areas of the back. This stimulus results in a brisk contraction of the local, underlying musculature within 1–3 sec of the probe placement. Upon appearance of this response, the probe was removed and the latency recorded. Failure to respond within 6 sec was cause to remove the probe and assign that value (6 sec) as the latency.

LCSF (approximately 0.3 mL) and blood samples (approximately 1 mL) were collected from each animal before and during dosing intervals. Blood samples were collected on wet ice and prepared as plasma in lavender EDTA-coated test tubes. Cisternal CSF samples (approximately 0.3 mL) were obtained by percutaneous puncture (1.5 inch 22 ga spinal needle) in the lightly anesthetized dog three times: i) just prior to the initiation of 1 µg/hr ziconotide chronic intrathecal infusion, ii) 48 hr later, just prior to the dose change to 5 µg/hr ziconotide infusion, and iii) 48 hr later just prior to termination of chronic intrathecal infusion. All samples were collected onto dry ice and stored frozen at approximately −20°C ± 10°C until assay.

Ziconotide assays were performed using a validated radioimmunoassay (RIA) using ¹²⁵I-ziconotide and a ziconotide-specific antibody. In this assay, the ¹²⁵I-ziconotide-antibody complexes are separated from unbound ¹²⁵I-ziconotide by adsorption to charcoal and the radioactivity in the complexes measured by gamma counting. Ziconotide in a sample inhibits the binding of ¹²⁵I-ziconotide to the antibody and the amount of ziconotide in the sample is calculated by comparison to a standard curve. The lower limit of quantification of the RIA for dog CSF and plasma is 0.07 ng/ml of ziconotide. Standard curves and blanks were established using dog cisternal cerebrospinal fluid and dog plasma. Additional details and characteristics of this assay are presented elsewhere. (25)

After completion of the final dosing interval and sample collection, animals were deeply anesthetized with sodium pentobarbital and intubated, with ventilation being manually maintained. The chest was then opened, the aorta catheterized, and the blood cleared by perfusion at 80 to 160 mmHg of pressure, with approximately 4 L of 0.9% saline. The spinal column was then exposed by laminectomy. Methylene blue dye was injected through both intrathecal catheters to establish their patency and to visualize their position relative to each other.

Behavioral scores (arousal, muscle tone, and coordination), physiological parameters (blood pressures and heart and respiratory rates) and skin twitch response latencies are presented as mean and standard error of the mean (S.E.M.). The pharmacokinetic analyses were done using PK Solutions software (PK Solutions, Version 2.0 SUMMIT Research Services 1374 Hillcrest Drive, Ashland, OH 44805, Copyright 1999). Area under the concentration-time curves (AUC) for plasma and CSF were estimated by the trapezoidal rule up to the last observed concentration (C_last). For concentrations below the limit of detection, a value of one-half the detection limit was used. If the last concentration was detectable, then AUC_0-∞ was estimated as AUC up to the last observed concentration (AUC_0-τ) + C_last/λ_z, where λ_z is the elimination rate constant estimated by linear regression of the terminal portion of the concentration-time curve. Clearance (CL) was calculated as F*D/AUC, where F=bioavailability and D = dose administered. Apparent volume of distribution (V_z) was estimated by CL*λ_z, and relates the amount of drug in the body to the sampled matrix concentration during the terminal phase. Volume of distribution at steady-state (V_ss) was estimated as (D)(AUMC_0-∞)/(AUC_0-∞), where AUMC is the total area under the first moment curve, and relates the amount of drug in the body to the sampled matrix concentration at steady-state. The analysis used curve stripping methods, where drug concentration data in blood or CSF are resolved into a series of rate constant terms corresponding to the absorption, distribution, and elimination phases. Half-lives are then estimated as 0.693/k, where k is the corresponding rate constant for that phase. Two rate constants were estimated, and correspond to either: distribution and elimination (when the concentrations are being measured in the matrix where the dose was administered), or absorption/distribution and elimination (when the concentrations are being measured in a different compartment from where the dose was administered). The following parameters were determined: 1)

Bolus intrathecal injection: CSF and plasma levels of ziconotide and ³H-inulin versus time. For LCSF and plasma: T_½ of distribution (T_½-α) (LCSF) or absorption/distribution (plasma), T_½ of elimination (T_½-β), peak concentration (C_max), time to peak concentration (T_max), area under the concentration-time curve (AUC_0-∞), ratio of LCSF ziconotide (µg/mL) to LCSF inulin

The canine model has been widely employed in defining the physiological effects of a variety of intrathecal antinociceptive agents, (26) including several opiates, (23, 24), alpha2 adrenergic agonists, (27, 28), neostigmine (22); NSAIDs (ketorolac) (29), and growth factors (21) given as a bolus injection and/or infusion. The use of separate chronically placed sampling and infusion catheters permits ongoing drug delivery and sampling without contamination of the sampling route with the infusate. The use of the cisternal approach reduces the likelihood of changes in local drug distribution due to local leakage at the dural catheterization site. An important limitation of this and any sampling model is that removal of the lumbar CSF (LCSF) serves to diminish the pool of available drug and likely causes an overestimation of the rate of clearance. The impact of this intervention is reduced by limiting the number and volume of samples. Importantly, given the effect of sampling, we believe it is necessary to have concurrent comparisons with diffusion standards, such as inulin, to permit the reliable determination of the pharmacokinetic behavior of the target drug. The inert water-soluble agent, inulin, is widely used as a marker for the extravascular / extracellular space. (See, for example (30) Its large size (approximately 5000 Da) reduces its ability to diffuse though extracellular channels and its clearance from CSF is believed to reflect the CSF clearance through rostral bulk flow and absorption in the arachnoid granules and subsequently into the venous drainage. (30,31,32) Agents cleared as inulin is are, accordingly, believed to be primarily removed by a similar route.

The lumbar CSF space is a poorly stirred volume with limited local fluid movement. (33,34, 35, 36) Accordingly, upon delivery into the intrathecal space, the injectate undergoes an initial local redistribution that depends upon the volume and rate of delivery. Once delivered, the injectate shows an initial fall in concentration that reflects: i) a progressive dilution into the LCSF volume driven by oscillatory pressure waves. (34) A second elimination phase is typically noted which reflects: i) ongoing local dilution in the CSF, ii) movement into the adjacent spinal parenchyma, and iii) movement into and through the adjacent meninges into the epidural space. (35,37). Such biphasic clearances have been observed after bolus delivery in all drugs studies in this model (see below)

Previous work has shown that the principal variables governing the second phase of CSF clearance of an intrathecally delivered agent are its physicochemical properties. Low molecular weight, moderately lipid soluble molecules (e.g., alfentanil) show a rapid movement from the lumbar CSF into tissue and through the meninges and into the vasculature.(35). In contrast, larger, hydrophilic molecules such as ziconotide or inulin show a more delayed clearance. (21) At the extreme, the clearance of a hydrophilic molecule that diffuses poorly into tissue will display a half-life that approaches that of CSF. In the case of the dog, this has been calculated to be on the order of approximately 2 hr. (32) While systematic data are limited, it appears that hydrophilic molecules over several hundred Da display increasingly similar clearance properties with increasing molecular weight. In the present studies, ziconotide (approximately 2500 Da) displayed an elimination half-life after bolus delivery on the order 3 hr. This is longer than the T_1/2 calculated for smaller hydrophilic molecules in this model (e.g. 0.9 hr for ketorolac (29); and 1.2 hr for morphine; Allen and Yaksh, unpublished data). This clearance is slightly faster than that for inulin (3.9 h), and may some reflect binding in parenchyma and/or meninges. Interestingly, IT BDNF, a large growth factor (27 KDa) showed a surprisingly rapid clearance from the CSF (0.7 hrs_(21). Such difference may reflect on other routes of clearance of free agent including binding to meninges. In contrast, comparison of the initial distribution phase for ziconotide and inulin (as well as the other hydrophilic molecules) indicates that they are essentially identical, emphasizing that the initial dilution after bolus delivery represents a local dilution in the CSF. Calculation of the ziconotide: inulin ratios, after bolus delivery, revealed that, consistent with the comparable calculated half-lives, both molecules showed a similar acute redistribution. On the other hand, consistent with the shorter T_1/2ß, ziconotide LCSF concentrations fell more rapidly than those of inulin, such that the ziconotide: inulin ratio fell by a factor of approximately 3 and 12 at 3 hr and 8 hr, respectively. This difference in the decline of lumbar ziconotide concentrations may reflect: i) a slightly more rapid clearance of ziconotide into tissue, ii) adherence of the drug to local sites (i.e., specific and non-specific binding), and/or iii) metabolism of ziconotide. Specific binding has been shown for ziconotide in spinal tissues (11) and local absorption in the meninges is not unlikely. Protein binding in CSF is not likely a factor, as protein concentrations are exceedingly low compared to plasma. As regards metabolism, ziconotide is a peptide and can undergo peptidic hydrolysis in the blood (38), though it appears relatively stable in in vivo studies. (25)

Assessment of elimination after bolus IV delivery revealed a half life of 1.6 hrs, faster than previously reported in rats and dogs after steady state infusion.(25) The elimination T_1/2 of ziconotide in CSF following IV injection was longer than that in plasma (1.64 hr vs 0.75 hr), but was comparable to the elimination T_1/2 in CSF when ziconotide was given by IT bolus (1.2 hr). This suggests that transfer of ziconotide out of the CSF is the rate-limiting for its spinal action.

From the above reported observations and analyses, it can be concluded that intrathecally delivered ziconotide is distributed in the lumbar and cisternal CSF as a large molecule in a manner suggesting bulk redistribution in the spinal intrathecal space. The molecule displays linear kinetics that are not altered by continuous infusion, coming to a steady state in LCSF with continuous infusion within an 8-hr interval in this model -- a value consistent with the half-life determined from clearance following bolus delivery or clearance from steady state. Of particular interest in both the humans and dog study was the apparent delay in the onset of the behavioral effects. In the dog study, skin twitch was not blocked at 8 hrs, although the CSF steady state was apparently maximal at this time. In the human work, it was evident that a lag time exists between the onset of analgesia in these pain patients and the LCSF-PK. The source of this delay likely reflects upon the time required for the distribution from the site of delivery and the spinal site of action. In the case of the N-type calcium channel blocker this action lies within the superficial layer of the spinal dorsal horn. (5) Accordingly, after IT delivery, the drug must reach the spinal levels mediating that pain state (e.g., lumbar spinal cord, not roots). After bolus delivery, based on this dog study, this initial redistribution may be relatively brief. However, with the use of infusion, it is likely that the interval will be longer. At any given level, the agent must then penetrate to the site of spinal action. While there are few systematic data, previous work has emphasized that the time to onset of inhibition of dorsal horn neurons firing correlates with the time of onset of analgesia. Agents that are lipid soluble will penetrate rapidly and show a correspondingly more rapid onset than that observed with polar molecules. (47, 48) Moreover, consistent with spinal cord size, the time of onset of effect after a particular IT drug such as morphine is shortest in small species (e.g., mouse) and longest in larger species (e.g., dog and human). Accordingly, it is not surprising that there is a definable hysteresis between LCSF steady state and onset of action of large, relatively poorly diffusible molecules such as ziconotide. These comments reflect upon the likely need to increment the dosing of intrathecal ziconotide to achieve optimal dosing. Based on CSF levels in the dog we would estimate that CSF steady state will require around 8 hrs. One would then consider that time to diffuse into tissue would require an additional interval. It is not possible from these studies to estimate the additional interval, but given the time to analgesia observed in the dog infusion studies, this interval would occur inside of 24 hrs. From a clinical perspective, dose incrementation should therefore not occur at intervals of less than 24 hrs. As non-spinal side effects increase as a function of dose, rapid incrementation would tend to overshoot the minimum therapeutic levels and increase the likelihood of untoward effects. In humans, confusion and other mental changes, as well as autonomic effects, have been reported to occur at roughly 24–48 hr (20, 49) or 48–72 hr (15) intervals after incrementation of intrathecal ziconotide and probably reflect a supraspinal action. The present studies indicate that increasing plasma versus LCSF levels continue to rise through 48 hr. Though systemic ziconotide displays limited systemic bioavailability, the delayed onset of side effects may reflect a role for systemic redistribution. We did not examine cisternal ziconotide concentrations at shorter time intervals than 48 hrs, but it seems likely that supraspinal steady state levels will occur after the LCSF C_max levels that were observed between 8 and 24 hr. Importantly, the stability and linearity of the kinetics after either bolus or continuous infusion observed in these studies provides assurance that drug effects may be predictably titrated by slower incrementation than used in initial clinical studies of ziconotide. (15, 20, 17, 50)