More than 200,000 patients worldwide have had subcutaneous pumps and intrathecal catheters implanted for continuous intrathecal drug delivery (CIDD). Chronic intrathecal therapy is most commonly used to treat chronic pain (benign and malignant) and spasticity (secondary to spinal cord injury, cerebral palsy, stroke, etc.). Infused drugs target the spinal cord parenchyma (e.g., opioids, baclofen, ziconotide) or nerve roots/rootlets (e.g., local anesthetics). In some clinical situations it may be sufficient to target only a very limited length of spinal cord; however, it is more often necessary to target many spinal cord segments. Thus, adequate drug distribution is critical, and an inability to deliver sufficient drug concentration to all affected spinal segments can result in a pharmacokinetically based therapeutic failure. This fact was amply demonstrated by Walker et al.,1 who implanted intrathecal pumps for baclofen delivery in patients only after a trial bolus showed that baclofen (25 to 75 μg) was effective at treating their spasticity. Even though all patients benefited from the small intrathecal baclofen bolus, nearly 40% of patients failed to show clinical improvement when receiving baclofen via chronic infusion, despite doses as high as 1000 μg/day. Thus, the failure of chronic continuous baclofen infusion cannot be explained as a pharmacodynamic failure because all patients were preselected for their positive pharmacodynamic response to baclofen. Rather, this observation is most readily explained as a pharmacokinetic failure, i.e., baclofen failed to reach the necessary spinal cord segments at sufficient concentration when delivered as a chronic continuous infusion.
We propose that the observations of Walker et al. can be explained by marked differences in drug distribution between bolus administration and chronic intrathecal infusion, specifically, that bolus administration results in much greater drug distribution because of the kinetic energy imparted to the injected solution.
Some will find this proposal difficult to accept because of the widely held view that cerebrospinal fluid (CSF) circulates within the subarachnoid space and consequently will widely distribute any drug that is suspended within it. However, this view of the CSF as a “circulatory system” is outdated and incorrect. Numerous contemporary studies using cine-magnetic resonance imaging techniques that permit CSF motion to be seen and quantified in vivo have clearly shown that CSF oscillates in a cephalo–caudad direction with a frequency equal to the heart rate.2 – 6 The frequency is tied to the heart rate because it is the alternating expansion and contraction of the cerebrum as blood is ejected into it that generates the force causing CSF oscillation. Importantly, these studies also document the absence of net CSF motion or “bulk flow.”
Given the absence of appropriate studies examining the pharmacokinetics of drug distribution in CSF and spinal cord during the very slow infusion rates used for chronic drug delivery, we designed this study to quantify morphine distribution within the spinal cord during chronic intrathecal infusion in an ambulatory animal model.
This study was approved by the University of Washington Animal Care and Use Committee and conforms to the guidelines of the Committee for Research and Ethical Issues of the International Association for the Study of Pain and to the guidelines of the Association for Assessment and Accreditation of Laboratory Animal Care.
Farm-bred pigs (Yorkshire crosses) of both sexes weighing 20 to 25 kg were used. Approval was obtained to study 6 animals. One animal was euthanized 4 days after catheter implantation because of hindlimb paraparesis, leaving 5 animals completing the study (subsequent necropsy and histopathology revealed a modest inflammatory infiltrate in the spinal cord near the catheter tip but failed to demonstrate a cause for the paresis. However, we presume that the spinal cord or its vascular supply was accidentally injured during catheter placement or pump implant).
Anesthesia was induced with IM tiletamine (4 mg/kg; Fort Dodge Animal Health, Fort Dodge, Iowa), zolazepam (4 mg/kg; Fort Dodge Animal Health), and xylazine (2 mg/kg; Lloyd Laboratories, Shenandoah, Iowa). Animals were orotracheally intubated and mechanically ventilated to maintain end-tidal CO2 34 to 38 mm Hg (Surgivet model, Smiths Medical PM, Inc., Waukesha, Wisconsin). Anesthesia was maintained with isoflurane (1.5% to 2.5%). Temperature was maintained at 37°C to 39°C via forced air warmer (Bair Hugger model 505, Arizant Healthcare, Inc., Eden Prairie, Minnesota). A 20- or 22-gauge IV catheter was placed in an ear vein for maintenance fluid administration (0.9% NaCl at 100 mL/h).
Before incision, IM cetiofur (5 mg/kg; Pfizer Inc., New York, New York) was administered to prevent surgical-site infection. The T10 to 11 interspace was exposed via blunt and sharp dissection and a small hole incised through the ligamentum flavum to expose the spinal cord. A small (approximately 1 mm) hole was cut through the dura and arachnoid membranes and an intrathecal catheter (model 8731SC, Medtronic, Inc., Minneapolis, Minnesota) inserted through the hole into the posterior subarachnoid space a distance of 5 cm cephalad. The catheter was secured to the dura mater at the point it that it entered the subarachnoid space with cyanoacrylate glue, and the intrathecal catheter was sutured to paraspinous muscle using the V-wing anchors provided. To accommodate the implantable pump (Synchromed II, Medtronic, Inc.), we created a subcutaneous pocket on the animal's flank by blunt dissection from the same incision used to access the spine. The pump was filled with 20 mL preservative-free morphine sulfate (1 mg/mL; Hospira, Inc., Lake Forest, Illinois), the catheter connected to the infusion pump, the pump inserted into the subcutaneous pocket and the incision closed in layers with suture. The implanted pump was programmed to first deliver a bolus to fill the catheter deadspace and then infuse morphine at 20 μL/h (0.48 mL/day). Prior to emergence, carbiprofen (4 mg/kg; Pfizer, Inc.) was administered IM to provide postoperative analgesia.
After emergence from anesthesia, animals were tracheally extubated and returned to their cages where they had ad libitum access to water and twice daily feedings of an age-appropriate amount of pig chow. Animals were visited at least twice daily after surgery to monitor their condition, but no research data were collected.
After 14 days of continuous intrathecal morphine infusion, the animals were anesthetized as described above and the spinal cord exposed by laminectomy from the sacrum to the mid-cervical vertebrae. The location of the intrathecal catheter tip was marked on the dura mater, the catheter removed and the animal euthanized by IV saturated KCl injection (10 mL). The spinal cord was excised, cut into 1-cm pieces, weighed, and frozen at −20°C until analyzed for morphine content.
Tissues were assayed as previously described by our laboratory using a gas chromatography/mass spectrometry technique.7
Prospective power analysis was not performed. All statistical tests were performed using Instat software (GraphPad Software, Inc., La Jolla, California). Exponential curve fitting was performed using Kaleidagraph software (Synergy Software, Inc., Reading, Pennsylvania). P values <0.05 were considered statistically significant.
One animal's pump was misprogrammed to deliver morphine at 2 μL/h (0.048 mL/day) instead of the intended 20 μL/h. This animal was omitted from statistical analysis, but the data are presented in Figure 1 for comparison.
A repeated-measures analysis of variance was used to determine whether morphine concentration in spinal cord differed as a function of sampling site (i.e., spinal cord segment). The Dunnet multiple comparison test was used as a post hoc test to perform pairwise comparisons between morphine concentration in the 0-cm spinal cord segment and all other spinal cord segments.
Morphine concentration differed significantly as a function of spinal cord sampling site (P < 0.0001). The average drug concentration in the 0-cm spinal cord segment was significantly larger than that in all other segments (Fig. 2).
Morphine concentration decreased markedly with increasing distance from the drug infusion point (0-cm spinal cord segment). For example, morphine concentration in the spinal cord segments 5-cm caudad and 5-cm cephalad of the 0-cm segment averaged only 24% ± 15% and 20% ± 12% of the peak concentration, respectively. Morphine concentration in the spinal cord segment located 10-cm cephalad of the 0-cm segment averaged 6% ± 5% of the peak concentration, and in the segment 10-cm caudad of the 0-cm segment, morphine concentration averaged 8% ± 5% of the peak concentration. To place this in clinical context, the adult human lumbar vertebrae are approximately 5 cm long.
The shape of the individual animal's concentration versus spinal cord segment data were all similar and mirror the shape of the averaged data (Fig. 2) in Figure 1.
The curves relating average morphine concentration to spinal cord segment (Fig. 3) both cephalad and caudad of the 0-cm segment are well fit by exponential functions (R 2 = 0.94 and 0.91, respectively). Both the y intercepts and the exponents are comparable between the cephalad and caudad curves, indicating that drug distribution was symmetric about the catheter tip.
The goal of this study was to characterize morphine distribution in the spinal cord during chronic infusion and to determine whether drug distribution in a chronic model of CIDD is comparable to what we have previously observed in an acute model. 8,9 Because terminal elimination half-life for morphine in the pig's spinal CSF is 1.3 ± 0.7 hours10 (which is comparable to that reported in humans: 1.2 to 1.6 hours11 – 14), the 14-day infusion period used for this study is 253 terminal elimination half-lives and, therefore, clearly constitutes a chronic pharmacokinetic model.
Although the animals were housed in relatively small areas (10 × 10 foot pens), they were free to ambulate at will. When at rest, pigs sleep/recline on their sides, sternally recumbent, and on their backs. In addition, the animals occasionally stand on their hind legs with their forelegs on the fence that forms their pen. Thus, the animals are mobile and they assume a variety of positions that might theoretically affect drug distribution, much as with humans.
The data from this chronic study confirm our earlier findings in an acute model. Specifically, drug distribution in CSF and the spinal cord is quite limited during the very slow intrathecal infusion rates typically used for CIDD. Because our earlier acute studies examined bupivacaine and baclofen distribution, it is not possible to draw quantitative comparisons between these studies and the current chronic study. However, some qualitative comparisons are possible. For example, over the course of an 8-hour infusion at 20 μL/h, the spinal cord distribution of both baclofen (2 mg/mL) and bupivacaine (7.5 mg/mL) were also well described by exponential functions (R 2 averages 0.97 for both drugs). Whether quantitative differences between our acute studies and this chronic study result from differences in the total amount of drug administered, the duration of infusion, or drug-specific pharmacokinetic differences among study drugs cannot be determined from the data.
Because of a programming error, we had 1 animal that received morphine at a rate equal to 10% of that in the remaining 4 animals. Although 1 animal is not sufficient for rigorous comparisons, it is noteworthy that the pattern of drug distribution in this lone animal was comparable to that in the other animals (Fig. 2). In addition the peak morphine concentration in this animal (segment “0”) was approximately what one would expect for a one-tenth dose: 720 ng/g vs. 10,466 ± 5984 ng/g in the full-dose animals.
The data from the current study help to explain clinical studies in which a drug delivered as an intrathecal trial bolus were effective at relieving patient symptoms but, when delivered at the very slow infusion rates used for chronic infusion, it was not effective, i.e., pharmacokinetic failures of chronic intrathecal drug delivery. Specifically, although bolus intrathecal injection produces relatively widespread drug distribution, that is not the case with chronic infusions. Consequently, in some instances, drug distribution during chronic infusion may not be sufficient to produce therapeutic drug concentrations in spinal cord segments distant from the drug infusion point.
Many clinicians may have difficulty accepting our observation that drug distribution is quite limited and nonuniform after 14 days of continuous intrathecal drug infusion. This difficulty presumably results from the commonly held view that CSF “circulates” within the subarachnoid space. Consequently, it has been assumed that given enough time the CSF will widely distribute any molecule that is administered into it. Though this view of the CSF as a “circulatory system” is both widely held and still taught,15 it is incorrect. This outdated view is based on methodologically inappropriate studies in which hypertonic, hyperbaric contrast agents (or large radiolabeled molecules) with very long CSF residence times were incorrectly assumed to behave in CSF as does CSF itself. However, as is noted in the introduction, numerous contemporary studies using magnetic resonance imaging and computed tomography techniques have shown that CSF does not circulate. Rather, it oscillates to and fro in a cephalo–caudad direction with no net movement.2 – 6 This oscillatory motion does result in limited “mixing” and some net drug distribution, but as our data demonstrate, the process is quite slow and incomplete. This fact helps to explain why intrathecal granulomas invariably form at the tip of the intrathecal catheter, i.e., drug concentration remains very high at the catheter tip because there is very little mixing/dilution in CSF.
Our experimental observations in this chronic study and our earlier acute studies have led us to propose the following qualitative description of drug distribution during chronic intrathecal infusion. We suggest that the primary limitation on how far a drug molecule can be distributed by means of CSF motion is the molecule's residence time within CSF. That is, a drug molecule can only be distributed while it remains suspended in the CSF. Once cleared into the spinal cord or epidural space, it no longer distributes within the CSF.16 Thus, the drug distribution pattern we observed in both our acute and chronic studies can be interpreted as a probability distribution. That is, the probability that any individual drug molecule will remain in CSF long enough to be distributed a given distance along the spinal cord decreases as the distance increases. Alternatively, the greater the distance from the infusion point, the lower the probability that any given drug molecule will remain in CSF long enough to be distributed that distance.
Assumptions underlying this hypothesis are that neither a drug's CSF distribution rate nor CSF residence time changes with increasing duration of infusion. That is, drug accumulation in spinal cord and epidural space does not appreciably change the concentration gradient driving drug clearance from CSF and, therefore, the drug's clearance rate does not change with time. If a drug does accumulate in spinal cord and epidural space sufficiently to decrease CSF clearance rates, then the drug's CSF residence time will increase and the probability that it will remain in CSF long enough to distribute a given distance will increase correspondingly.
This hypothesis also suggests that drug distribution distance can be increased by increasing the mass (concentration) of drug administered. That is, increasing the mass of drug increases the number of individual drug molecules that have a CSF residence time significantly longer than the average drug molecule, thereby increasing the probability that an individual molecule will persist in CSF long enough to reach a given distance. Unfortunately, increasing drug concentration also increases the risk of neurotoxicity.17,18
Our hypothesis also suggests that use of drugs with long CSF residence times can also result in increased drug distribution. An interesting study by Kroin et al.19 is informative in this regard. This study, which is inaccurately titled “The Distribution of Medication Along the Spinal Canal After Chronic Intrathecal Administration,” took patients receiving chronic intrathecal drug infusions and filled their implanted pumps with 111In-labeled diethylenetriamine penta-acetic acid (111In-DTPA) and infused it at 42 to 62 μL/h for 72 hours and then used a γ camera to measure 111In-DTPA distribution (N.B.: the study did not characterize “drug distribution” as suggested by the title; it characterized 111In-DTPA distribution). Importantly, diethylene triamine penta-acetic acid (DTPA) has been reported to have a half-life as long as 5.5 days in humans,20 or approximately 110 times longer than morphine. Despite the very long half-life, Kroin et al.19 still observed a large decrease in 111In-DTPA concentration with distance along the spinal canal, specifically, a 57% decrease between T2 and T12. This decrease is not negligible, but is much less than what we observed with morphine in the current study, a fact that we attribute to the very long CSF residence time of 111In-DTPA.
In summary, this is the first study to measure morphine distribution in the spinal cord after chronic intrathecal infusion in an ambulatory animal. As in our earlier acute studies,8,9 we found that drug distribution is quite limited with a marked decrease in spinal cord drug concentration over distances as short as 5 cm. This finding is consistent with what is now known about the lack of CSF circulation and may help explain pharmacokinetic failures of chronic intrathecal drug delivery.
Name: Sean H. Flack, MBChB, FCA
Contribution: This author helped design the study, conduct the study, analyze the data, and write the manuscript.
Attestation: Sean H. Flack has seen the original study data, reviewed the analysis of the data, approved the final manuscript, and is the author responsible for archiving the study files.
Conflict of interest: Sean H. Flack reported no conflicts of interest.
Name: Christine M. Anderson, MD
Contribution: This author helped design the study, conduct the study, and write the manuscript.
Attestation: Christine M. Anderson has seen the original study data and approved the final manuscript.
Conflict of interest: Christine M. Anderson reported no conflicts of interest.
Name: Christopher Bernards, MD
Contribution: This author helped design the study, conduct the study, analyze the data, and write the manuscript.
Attestation: Christopher Bernards has seen the original study data, reviewed the analysis of the data, approved the final manuscript, and is the author responsible for archiving the study files.
Conflict of interest: Dr. Bernards consults for Arsenal Medical and Isis Pharmaceuticals and has received research funding from Medtronic Inc. Medtronic Inc. did not provide direct financial support for the current study but did loan the investigators 6 Synchromed II pumps.
1. Walker RH, Danisi FO, Swope DM, Goodman RR, Germano IM, Brin MF. Intrathecal baclofen for dystonia: benefits and complications during six years of experience. Mov Disord 2000;15:1242–7
2. Enzmann DR, Pelc NJ. Cerebrospinal fluid flow measured by phase-contrast cine MR. Am J Neuroradiol 1993;14:1301–7
3. Henry-Feugeas MC, Idy-Peretti I, Baledent O, Poncelet-Didon A, Zannoli G, Bittoun J, Schouman-Claeys E. Origin of subarachnoid cerebrospinal fluid pulsations: a phase-contrast MR analysis. Magn Reson Imaging 2000;18:387–95
4. Wayte SC, Redpath TW. Cine magnetic resonance imaging of pulsatile cerebrospinal fluid flow using CSPAMM. Br J Radiol 1994;67:1088–95
5. Wayte SC, Redpath TW, Beale DJ. Magnetic resonance imaging of pulsatile cerebrospinal fluid flow by spatial modulation of magnetization. Br J Radiol 1992;65:495–501
6. Wayte SC, Redpath TW, Beale DJ. A cine–SPAMM sequence for NMR imaging of pulsatile cerebrospinal fluid flow. Phys Med Biol 1993;38:455–63
7. Thompson SJ, Koszdin K, Bernards CM. Opiate-induced analgesia is increased and prolonged in mice lacking P-glycoprotein. Anesthesiology 2000;92:1392–9
8. Bernards CM. Cerebrospinal fluid and spinal cord distribution of baclofen and bupivacaine during slow intrathecal infusion in pigs. Anesthesiology 2006;105:169–78
9. Flack S, Bernards C. Cerebrospinal fluid and spinal cord distribution of hyperbaric bupivacaine and baclofen during slow intrathecal infusion in pigs. Anesthesiology 2010; 110:165–73
10. Bernards CM, Shen DD, Sterling ES, Adkins JE, Risler L, Phillips B, Ummenhofer W. Epidural, cerebrospinal fluid, and plasma pharmacokinetics of epidural opioids (part 1): differences among opioids. Anesthesiology 2003;99:455–65
11. Moulin DE, Inturrisi CE, Foley KM. Cerebrospinal fluid pharmacokinetics of intrathecal morphine sulfate and D-Ala2-D-Leu5-enkephalin. Ann Neurol 1986;20:218–22
12. Sjostrom S, Tamsen A, Persson MP, Hartvig P. Pharmacokinetics of intrathecal morphine and meperidine in humans. Anesthesiology 1987;67:889–95
13. Kotob HI, Hand CW, Moore RA, Evans PJ, Wells J, Rubin AP, McQuay HJ. Intrathecal morphine and heroin in humans: six-hour drug levels in spinal fluid and plasma. Anesth Analg 1986;65:718–22
14. Eisenach JC, Hood DD, Curry R, Shafer SL. Cephalad movement of morphine and fentanyl in humans after intrathecal injection. Anesthesiology 2003;99:166–73
15. Carr D, Cousins M. Spinal route of analgesia: opioids and future options for spinal analgesia chemotherapy. In: Cousins M, Carr D, Horlocker T, Bridenbaugh P, eds. Neural Blockade in Clinical Anesthesia and Pain Medicine. Philadelphia: Wolters Kluwer/Lippincott Williams and Wilkins, 2009
16. Ummenhofer WC, Arends RH, Shen DD, Bernards CM. Comparative spinal distribution and clearance kinetics of intrathecally administered morphine, fentanyl, alfentanil, and sufentanil. Anesthesiology 2000;92:739–53
17. Yaksh TL, Hassenbusch S, Burchiel K, Hildebrand KR, Page LM, Coffey RJ. Inflammatory masses associated with intrathecal drug infusion: a review of preclinical evidence and human data. Pain Med 2002;3:300–12
18. Yaksh TL, Horais KA, Tozier NA, Allen JW, Rathbun M, Rossi SS, Sommer C, Meschter C, Richter PJ, Hildebrand KR. Chronically infused intrathecal morphine in dogs. Anesthesiology 2003;99:174–87
19. Kroin JS, Ali A, York M, Penn RD. The distribution of medication along the spinal canal after chronic intrathecal administration. Neurosurgery 1993;33:226–30
20. Nau R, Emrich D, Prange HW. Inverse correlation between disappearance of intrathecally injected 111
In-DTPA from CSF with CSF protein content and CSF-to-serum albumin ratio. J Neurol Sci 1993;115:102–4