Despite the extensive use of intrathecal morphine infusion for pain, no systematic safety studies exist on its effects in high concentrations. The authors assessed the effects of morphine and clonidine given 28 days intrathecally in dogs.
Beagles with lumbar intrathecal catheters received solutions delivered by a vest-mounted infusion pump. Six groups (n = 3 each) received infusions (40 microl/h) of saline or 1.5, 3, 6, 9, or 12 mg/day of morphine for 28 days. Additional groups received morphine at 40 microl/h (1.5 mg/day) plus clonidine (0.25-1.0 mg/day) or clonidine alone at 100 microg/h (4.8 mg/day).
In animals receiving 9 or 12 mg/day morphine, allodynia was observed shortly after initiation of infusion. A concentration-dependent increase in hind limb dysfunction evolved over the infusion interval. Necropsy revealed minimal reactions in saline animals. At the higher morphine concentrations (all dogs receiving 12 mg/day), there was a local inflammatory mass at the catheter tip that produced significant local tissue compression. All animals with motor dysfunction displayed masses, although all animals with masses did not show motor dysfunction. The mass, arising from the dura-arachnoid layer, consisted of multifocal accumulations of neutrophils, monocytes, macrophages, and plasma cells. Inflammatory cells and endothelial cells displayed significant IL1beta, TNFalpha, iNOS, and eNOS immunoreactivity. No evidence of bacterial or fungal involvement was detected. There were no other changes in spinal morphologic characteristics. In four other groups of dogs, clonidine alone had no effect and in combination with morphine reduced the morphine reaction.
The authors found that high intrathecal morphine concentrations lead to aseptic intrathecal inflammatory masses. The lack of effect of clonidine and the possible suppressive effects of clonidine on the local reaction suggest the utility of such coadministration.
SPINAL delivery of opioids produces a potent antinociception through spinal opioid receptors. 1After demonstration of the clinical efficacy of chronic intrathecal delivery of morphine through implanted pumps, 2this modality evolved as an alternative in the long-term management of chronic pain states. 3–8In spite of its clinical use, there have been limited systematic safety studies on the effects of spinally delivered morphine sulfate. In a dog model, 28-day fixed bolus dose intrathecal delivery studies were undertaken for morphine (1–10 mg/ml) 9but displayed no significant pathology. Although the range of doses that are typically delivered daily has been relatively constant over the past 20 yr (e.g. , 5–15 mg/day), 10,8the concentrations that are routinely employed in the infusion devices have risen, serving to extend refill intervals. Thus, early clinical use typically employed delivery rates of several milliliters per day with concentrations of around 10 mg/ml in the 1980s. Following approval by the US Food and Drug Administration in 1991 of 25 mg/ml morphine for intrathecal infusions (NDA#18–565), concentrations of 25–50 mg/ml have been widely used. This absence of applicable safety data prompted us to examine the effects of continuous lumbar intrathecal infusion of specified concentrations of morphine sulfate in the well-characterized canine intrathecal infusion model. 11In addition, we present data on the effects of clonidine and clonidine–morphine admixtures. Unexpectedly, we found that morphine-initiated formation of intrathecal inflammatory masses, observations that are in accord with clinical case reports that first appeared in the early 1990s. 12
This article is accompanied by an Editorial View. Please see: Follett KA: Intrathecal analgesia and catheter-tip inflammatory masses. Anesthesiology 2003; 99:5–6.
Additional material related to this article can be found on the Anesthesiology Web site. Go to the following address, click on Enhancements Index, and then scroll down to find the appropriate article and link. http://www.anesthesiology.org.
All studies were approved by Institutional Animal Care and Use Committee at the University of California, San Diego. Purpose-bred male beagles (Ridglan Farms Inc., Mt. Horeb, Wisconsin) aged 12–16 months (weight, 12–18 kg) were individually housed in runs with wood shavings, maintained between 62° and 82°F on a 12 h–12 h light–dark cycle and with ad libitum access to fresh food and water. Animals were acclimated for a minimum of 10 days prior to surgery. A nylon vest (Alice King Chatham Medical Arts, Hawthorne, California) was placed on each dog for acclimation approximately 48 h prior to scheduled intrathecal catheter placement surgery.
Antibiotic sulfamethoxazole–trimethoprim (15–25 mg/kg, per os, twice daily) was given 48 h prior to and following surgery. For anesthesia induction, dogs received atropine (0.04 mg/kg, intramuscular) and xylazine (1.5 mg/kg, intramuscular for animals receiving morphine alone) or a combination of ketamine and valium (10/0.5 mg/kg, intravenous, for animals receiving clonidine). Each animal was intubated and anesthesia maintained under spontaneous ventilation with 1.0–2.0% isoflurane and 60% N2O/40% O2. Intraoperatively, animals were continuously monitored for oxygen saturation, heart and respiratory rates, and inspired and end-tidal gases.
Surgical areas were shaved and prepared with chlorhexidine. Catheters were implanted as previously described. 13In brief, by use of sterile technique the cisterna magna was exposed. Through a small incision, the PE-10 (OD = 0.61 mm) intrathecal catheter (fabricated of polyethylene tubing and sterilized by E-beam irradiation) was inserted and passed caudally a distance of 40 cm to the lumbar segment (L2–3). Dexamethasone sodium phosphate (0.25 mg/kg, intramuscular) was administered just after catheter placement. The external catheter was then tunneled subcutaneously and caudally to exit at the left scapular region. The incision was closed in layers. During recovery, butorphanol (0.04 mg/kg, intramuscular) was administered as necessary to relieve postoperative discomfort. Following anesthetic recovery, the nylon vest was placed on the animal and an infusion pump ((PANOMAT C-10; Disetronic Medical Systems, Saint Paul, Minnesota) secured in the vest pocket where it was connected to the intrathecal catheter.
Morphine sulfate (Infumorph® 500, 25 mg/ml, Lot # 070092; Elkins-Sinn, Cherry Hill, New Jersey) and clonidine HCl ((2 mg/ml, Lot # CTM-723; Medtronic, Inc. 710 Medtronic Parkway, Minneapolis, Minnesota) were diluted as required in sterile saline (0.9% weight/volume; Sodium Chloride Injection, USP, Abbott Laboratories, North Chicago, Illinois) to prepare the individual infusion preparations. Osmolarity across differing drug concentrations was adjusted to be approximately 310–320 mOsm by changes in sodium chloride. For each pump cartridge change, drug samples were randomly collected during the filling of the cartridges and analyzed for test article concentration to confirm dosing.
Dogs were randomly assigned to receive lumbar intrathecal infusion of the vehicle, morphine alone, morphine plus clonidine, or clonidine as set forth in table 1. Infusions were undertaken for a minimum of 28 days.
A physical examination of each dog was conducted prior to its inclusion in the study. Animals were observed for health status at least twice daily. Pump function and catheters were checked at least once daily. Signs of ill health or reaction to treatment were recorded as they were observed. A summary of observations and schedules of testing are presented in table 2.
Each animal underwent neurologic examinations postsurgery but prior to initiation of infusions, and postinfusions but prior to necropsy. Examiners were blinded with respect to treatment.
After surgery, each dog's rectal temperature was recorded daily. Specific behavioral indices chosen to assess state of arousal, muscle tone, and motor coordination were observed and recorded twice daily throughout the in-life phase of the study. Scores of 0–3 were assigned for each parameter, with 0 being normal function and 3 being severely abnormal, as described in detail elsewhere. 13Body weights, heart rate, and blood pressure measurements were recorded on prior to initiation of drug or vehicle, at four-day intervals, and on the day of necropsy.
Respiration rate was measured by observation of chest movement. Heart rate and blood pressure measurements for each dog were made by use of a tail cuff manometer (Dinamap 8100, Criticon, Inc. Tampa, Florida).
Laboratory Chemical Study.
Blood for clinical chemical study and test article quantification were collected from each dog by cephalic venipuncture. Cisternal cerebrospinal fluid (CSF) samples (approximately 1–2 ml) for clinical chemistries and test article measurements were obtained by percutaneous puncture of the cisterna magna.
Clonidine was assayed by capillary gas chromatography after quantitative extraction of drug and internal standard (MK801) from the matrix with solid-phase cartridge chromatography. The limit of detection for clonidine measurement was 1 ng/ml and relative errors for accuracy and precision within the dynamic range (1–1,000 ng/ml) were both less than ± 10%.
Morphine was assayed by either high-pressure liquid chromatography or monoclonal enzyme-linked immunosorbent assay. Samples assayed by high-pressure liquid chromatography were quantitatively extracted by solid-phase cartridge chromatography (after addition of codeine as an internal standard) prior to analysis. The limit of detection for morphine measurement by high-pressure liquid chromatography was 5 ng/ml and the relative errors for accuracy and precision within the dynamic range (5–50,000 ng/ml) were both less than ± 10%. The limit of detection for morphine measurement by enzyme-linked immunosorbent assay was 0.1 ng/ml and accuracy and precision within the dynamic range (0.1–1,000 ng/ml) were both less than ± 15%.
Following collection of blood samples, animals were anesthetized with an intravenous dose of sodium pentobarbital (35 mg/kg or to effect). CSF was then obtained via percutaneous sampling from the cisterna magna. The aorta was cannulated and the dog perfused with four liters of 0.9% (weight/volume) saline at 80–160 mmHg of pressure, followed by four liters of neutral buffered formalin,10%.
After perfusion, the dura and the spinal cord was exposed by laminectomy of the spinal canal and the lower brainstem. Dye was injected through the intrathecal catheter to determine the patency of the catheter, the location of the tip, and spread of the dye around the catheter tip. The condition of the spinal cord was noted, and a photographic (digital) record of the spinal cord–catheter relationship was made. The cords were divided (taking care to keep the dura intact) into four sections—cervical, thoracic, high lumbar (catheter tip region), and lower lumbar—for microscopic analysis. Peripheral necropsies were not performed.
Tissues were embedded in paraffin, sectioned at 3–5 μm, and stained with hematoxylin and eosin. In addition, special stains were performed on the spinal cord in the block that included the catheter tip region. These were immunohistochemical staining for glial fibrillary acidic protein (Dako, Carpenteria, California; Kit 20334) and histochemical stains that included Gram stain (McDonald's Gram Stain, American Master, Lodi, California; Kit # KTMGS) for bacteria and Grolcott methenamine-silver stain ((Grolcott's Methenamine Silver Stain, American Master, Lodi, California; Kit # KTMGSMIC) for fungi. To assess the presence of acid fast bacilli, Ziehl-Neelsen stain (AccuStain Carbol-Fuchsin solution, Sigma Diagnostics, St. Louis, Missouri) was employed.
On completion of the histologic examination, sections from each of the spinal cord sections were examined by light microscopy by a board-certified veterinary pathologist who was blinded to the treatment. Aside from the description of the tissue, a score from 0–4 was assigned (0 = none, 4 = severe pathology). After completion of the initial scoring, blocks from the catheter tip region were rank-ordered as to the severity of the pathology (from least to worst) within each of the two groups of dogs (saline or morphine and clonidine plus morphine).
To investigate the nature of the intrathecal masses, spinal cords were harvested in two separate dogs after whole body perfusion with saline and snap-frozen in isopropanol on dry ice. Cryostat sections were taken at 10 μ and stained for: CD68 (macrophages, Dako, Hamburg, Germany), Leu4 (T-cells, Becton Dickinson, San Jose, California), HLA-DR (antigen-presenting cells, Dako), NOS2–inducible nitric oxide synthase (iNOS, Chemicon, Hofheim, Germany), NOS3 (endothelial nitric oxide synthase, eNOS, Santa Cruz, California), Interleukin-1β (Genzyme, Rüsselsheim, Germany), tumor necrosis factor-α (R&D, Wiesbaden, Germany), and MMP-9 (matrix metalloprotease 9, R&D). Immunoreactivity was visualized by 3,3′-diaminobenzidine tetrahydrochloride (DAB). Controls were routinely performed by omission of primary antibodies.
Baseline heart rate and blood pressure in saline-infused animals were 116 ± 4 beats/min and 106 ± 5 mmHg (μ± SEM). No treatment groups were different from saline control. Morphine- and clonidine-infused dogs tended to show a mild dose-dependent hypotension and bradycardia, as did dogs receiving clonidine alone (4.8 mg/day). In general, with regard to body temperature or respiratory rate, no clinically significant trends were noted in any treatment group.
Two out of three animals in the 9 mg/day and all dogs in the 12 mg/day morphine treatment groups displayed episodes of local tactile allodynia (light tactile stimuli applied to the dorsal lumbar region evoking a stimulus-mediated agitation) during the first week after the initiation of infusion and were observed periodically though the time of sacrifice. Such allodynia was not typically observed in animals receiving lower doses or vehicles.
None of the animals in the saline control or 1.5, 3, 6, or 12 mg/day morphine groups displayed significant effects on arousal during the infusion interval. One animal in the 9 mg/day morphine group (dog REE-9) displayed intermittent mild sedation over the first eight days of morphine infusion. None of the animals in the clonidine alone or in the clonidine–morphine admixture groups displayed any systematic effects on arousal over the infusion interval.
None of the animals in the saline control group showed any effects on motor function (motor tone or coordination) over the infusion interval. There was, however, a dose-dependent increase in the severity of the motor dysfunction in morphine-treated animals as assessed by the cumulative motor coordination scores over the final week of morphine infusion (fig. 1). The time of onset of changes (from baseline observations) in motor dysfunction was inversely correlated with dose (fig. 2). Following the initial infusion, none of the clonidine or clonidine–morphine admixture animals displayed any change in motor function over the 28-day interval.
Clinical Laboratory Measurements.
Blood clinical chemistries for individual dogs remained within normal ranges (data not shown). A summary of CSF clinical chemistry results is presented in table 3. Cisternal CSF total protein, glucose, and cell counts were within the range of normal values in samples obtained at the time of surgery. In cisternal samples obtained by percutaneous sampling at the time of death, protein levels and white blood cell counts were elevated in all groups, including those receiving saline. CSF protein levels at necropsy were significantly increased in dogs with masses as compared with those without. Elevated levels of total protein and white cells reflect a proinflammatory stimulus likely induced by the surgery and long-term catheter placement. Glucose levels were unchanged, suggesting an absence of bacterial infection.
Cisternal CSF collected at necropsy from animals displaying significant motor dysfunction underwent aerobic and anaerobic cultures. Cultures were negative with only one exception: RYW-9 CSF showed a rare α-hemolytic Streptococcus that was neither S. pneumoniae nor an enterococcus species. Samples of morphine dilutions and morphine–clonidine dilutions for infusions were cultured for bacteria and fungi with no samples displaying positive results.
Histologic sections containing inflammatory masses were stained for gram-negative bacteria and acid-fast bacteria. None were observed to be positive. Observation of such sections under polarized light failed to reveal any particulate matter.
In general, the spinal cord neuropil was intact in all dose groups. In both treated and control groups there was no evidence of reactive gliosis or large-scale cell, indicating that the treatment did not directly damage the neural parenchyma (figs. 3 and 4).
In animals receiving morphine, the prominent observation was the appearance of well-organized masses that lay in the vicinity of the catheter tip. The masses were typically as much as 1 cm in length centered around the catheter tip, with girths sufficient to produce distortion of the adjacent cord parenchyma (fig. 3). These cellular collections were characterized by dense collections of lymphocytes, monocytes, macrophages, and plasma cells that formed well-defined masses in the intrathecal space. In this material, many of the macrophages displayed a foamy appearance. A few large macrophages were also noted in the inflammatory regions that could be considered to be epithelioid-like macrophages, but these cells did not appear to form in sheets or clusters. In some animals, fibrosis and neovascularization associated with the inflammation were observed (fig. 4).
These masses were found in all groups receiving morphine alone but not those receiving saline, and were most reliably expressed at higher morphine concentrations.
Further histochemical examination of the masses was carried out specifically in two additional dogs after exposure to 12 mg/day morphine (fig. 5;table 6). All animals displayed similar intrathecal pericatheter tip masses. These animals were deeply anesthetized and underwent saline perfusion, with quick freezing of the spinal tissues. In these animals, cells in the intrathecal masses typically revealed extensive positive immunoreactivity for macrophages, as indicated by large numbers of cells showing CD68-ir (ir = immunoreactive) and relatively few T cells (Leu4). Many of these cells were characterized as HLA-DR–positive antigen-presenting cells. Examination of NOS-ir showed many iNOS-positive cells, likely macrophages. In contrast, eNOS was extensively observed in the endothelium of vessels present in the mass as well as in some cells in the mass. MMP-9–ir (matrix metalloprotease 9) showed relatively few positive cells that were immunoreactive. Modest levels of the cytokines IL1β and TNFα were also noted in the masses (not shown). Importantly, macrophage immunoreactivity was not observed in cervical or thoracic sections (i.e. , distant from the infusion site) (table 6) nor in the spinal parenchyma adjacent to the mass.
The forced ranking of severity of pathology emphasizes the tendency for the severity to be greatest in the highest morphine concentrations. Figures 6 and 7present the individual pathology ranking by morphine-dose groups.
Correlation between Neurologic Findings and Spinal Pathology.
Importantly, there was a close correlation (r = 0.7, P < 0.005; Spearman Rank Correlation) between incidence of motor incoordination score and pathology rank. For the tactile allodynia, as noted above, two of three animals in the 9 mg/day morphine treatment group and all dogs in the 12 mg/day morphine treatment group displayed episodes of local tactile allodynia through the time of kills. All of these dogs displayed masses. In dogs developing masses at the lower doses, allodynia was typically not observed.
As indicated in figure 8, addition of clonidine in doses ranging from 0.25 to 1.0 mg/day added to 1.5 mg/day morphine resulted in a dose-dependent diminution in the pathology scores. A single animal in the lowest dose clonidine group displayed a small inflammatory mass, but no such observations were made in any other group receiving 1.5 mg/day morphine in combination with higher doses of clonidine. Clonidine alone (4.8 mg/day) had no evident effect on spinal cord morphology and displayed no incidence of reaction. Interestingly, the local catheter reaction was less in the clonidine-treated animals (trace-to-none) than in animals treated with saline (mild).
Plasma and CSF drug concentrations collected at the time of necropsy (prior to termination of the infusion pump) are presented for morphine and clonidine–morphine admixtures in tables 7 and 8, respectively. Cisternal CSF and plasma concentrations and CSF-to-plasma ratios observed at necropsy are plotted in figure 9. With the clonidine–morphine admixtures, morphine concentrations in plasma did not vary across clonidine doses and were comparable with levels observed when morphine (1.5 mg/day) was delivered alone (table 7).
Cisternal CSF morphine concentrations showed wide variations at the time of sacrifice across infusion concentrations, which is evidenced both by observation of the absolute levels and by consideration of the ratio of cisternal CSF-to-plasma morphine concentrations at the time of death. These values are in the range of approximately 100:1 to approximately 1:1 (see fig. 9).
Inspection of the cisternal CSF concentrations and correlation with pathology revealed the unexpected inverse correlation between the presence of a mass and lower cisternal morphine concentrations. This correlation is emphasized in figure 9.
This study was conducted to determine the effects of continuous lumbar intrathecal infusion for 28 days of specified concentrations of morphine or clonidine and morphine admixtures on behavior, motor function, and spinal histopathology. The work revealed that intrathecal infusions of morphine produced a time- and dose-dependent increase in motor dysfunction and intrathecal mass formation. Several properties of this phenomenon will be considered.
Behavioral and Neurologic Observations
Two sets of neurologic events were noted. First, shortly after initiating infusion of high concentrations of morphine, a tactile allodynia corresponding to the lumbar dermatomes was noted. The early onset of the effect and its correlation with high concentrations parallel reports in which high concentrations of morphine were observed to produce allodynia in animals. 14,15Although the spinal mechanisms of this allodynia are not certain, previous work indicated that these effects are not naloxone reversible and have been hypothesized to be related to the formation of morphine-3-glucuronide. A similar role for glucuronide metabolites is hypothesized in humans. 16
Second, over intervals of several weeks, increased hind limb motor tone (i.e. , spasticity) and dysfunction were noted to evolve, with the time of onset typically shortest in those dogs receiving the highest daily morphine doses (fig. 1). The motor effects are believed to reflect upper motor neuron lesions secondary to the spinal compressive effects associated with the masses observed in dogs that received morphine. Thus, although all dogs with intrathecal masses showed no motor impairment, all animals with motor impairment had notable intrathecal masses. The lack of perfect correlation between compression and motor effect is anticipated, given variations in location and extent of mass as well as its progressive development.
Spinal Histopathology with Morphine Infusion
The prominent observation was a local inflammatory mass at the catheter tip that produced local compression of the spinal cord proximal to the catheter tip in dogs that received morphine. The cellular constitution of the mass typically consisted of multifocal accumulations of a variety of inflammatory cells, reflecting a “chronic” or “chronic–active” inflammation. 17We note that although we consider these pericatheter masses to be granulomas, controversy can exist regarding this appellation. Thus, for example, collections of inflammatory cells classified as granulomas characteristically consist of epithelioid-like macrophages, which form sheets or clusters. 18,19Such organization was not particularly evident in this material. On the other hand, we note that if the inflammatory effect is secondary to the drug, the reaction could be extensive, reflecting the local distribution of the drug, as opposed to a single nidus induced by the physical presence of an irritant or infection.
The presence of immunoreactivity for the proinflammatory cytokines IL-1β and TNF-α, as well as for the HLA-DR epitopes and MMP-9, indicates the presence of activated cells. The NO-synthases eNOS and iNOS have been implicated in angiogenesis 20,21and in the macrophage response to activation. 22The strong presence of iNOS and eNOS in the intrathecal masses is consistent with an ongoing neovascularization.
The inflammatory cells forming the mass are considered to be vascular in origin. Systematic examination of serial sections of the masses emphasizes that the cellular collection arises from the dura-arachnoid as opposed to the pia or the parenchyma itself. Accordingly, based on the extensive capillary network in the dura mater as compared with the arachnoid 23and the observed origin of the mass, we suggest that the observed cells likely arise from this dural microvasculature that lies in proximity to the catheter tip. Importantly, dural microvessels are morphologically and functionally distinct from pial vessels. The latter possess fenestrations, are permeable to large tracers molecules, 24and show enhanced permeability in the presence of a variety of chemical stimuli. 25
As in previous studies with this chronic model, a mild local pericatheter reaction was noted in all dogs, including control (saline) animals, a finding consistent with a local foreign body reaction. 26In dogs displaying mass formation, examination of the cervical, thoracic, and even upper lumbar cords showed only a limited reaction around the catheter, despite the presence of a prominent reaction in the vicinity of the catheter tip. Aside from compression of adjacent spinal tissue, there were no other notable changes in spinal morphology. Absence of demyelination with little change in glial fibrillary acidic protein immunoreactivity adjacent to the masses suggest that even high persistent concentrations of morphine had no direct effect on spinal myelinated axons or neurons.
Spinal Histopathology with Clonidine Infusion
Clonidine alone at the highest dose (2 mg/ml at 100 μl/h) displayed no evidence of intrathecal pathology. These results are consistent with the absence of pathology reported with the continuous epidural delivery of clonidine in dogs. 27Interestingly, in the dogs receiving clonidine and morphine admixtures, the severity of the pathology in the lumbar sections was dose-dependently decreased by clonidine. These observations suggest that clonidine has a modest suppressive action on the inflammatory effects observed with morphine.
Time-Course of Inflammatory Mass Development
The time-course of mass development cannot be directly defined by these studies. At the concentrations examined, prominent masses could clearly occur by 4 weeks. As neurologic dysfunction corresponded to the presence of pathology, the time of detectable changes in motor function might be used to infer conservatively the development of a critical mass. Accordingly, based on the data shown in figure 2, a steady-state concentration of morphine produced by the infusion of 9–12 mg/day resulted in a mass of clinically significant size over an interval of a few weeks. Whether longer term infusion of lower concentrations would result in an inflammatory mass is not known.
Cisternal CSF Concentration
Cisternal CSF drug concentrations are taken as a marker of the degree to which the molecule delivered in the lumbar CSF is redistributed by bulk redistribution up the intrathecal space. For drugs that are not metabolized in the central nervous system, the plasma levels reflect the net clearance of the drug into the peripheral compartment. In dogs without intrathecal masses, the cisternal levels of morphine were noted to be higher than in those dogs in which a mass was present. Similarly, examination of the plasma levels revealed lower levels than would have been anticipated in the higher morphine dose groups. Nevertheless, there was a greater reduction in the rostral redistribution than systemic redistribution, as indicated by the cisternal CSF/plasma morphine ratios. In dogs with masses, there are considerably lower ratios than in animals without intrathecal masses. This association with mass formation and altered rostral redistribution may have several components. Lower cisternal levels might reflect a maldistribution that results from the mass at the catheter tip. In patients undergoing intrathecal chemotherapy for meningeo-carcinomatosis, hindered rostral redistribution has been reported. 28,29However, the overall reduction in cisternal concentrations with chronic lumbar infusion would also require an enhanced removal of drug from the lumbar space, either by clearance and/or by metabolism. The capacity of CSF to be cleared at the spinal level has been demonstrated. 30The inflammatory masses typically display vascularization, which could result in an enhanced local clearance. We cannot exclude that the inflammation may lead to enhanced absorption into the adjacent dural microcirculation. Movement from the intrathecal space through the dura has been shown to be a significant route of clearance even in the normal tissues. 31Interestingly, dural permeability is enhanced by local inflammation. 32The overall reduction in plasma levels, however, cannot be the result of a reduced spinal redistribution and may reflect an enhanced metabolism by the inflammatory tissue, a possibility consistent with the presence of glucuronyltransferasein macrophage populations. 33
Origin of Inflammatory Masses
The origin of the intrathecal mass is not certain and several issues need to be addressed.
Role of Infectious Processes.
The mass formation was an aseptic process. Cultures of cisternal CSF were, with one exception (RYW-9), negative. As the CSF cisternal fluid is sampled by a percutaneous puncture, the possibility of inadvertent contamination cannot be ruled out in this animal. In addition, no staining for fungal or bacterial components was seen in any histologic section. CSF glucose levels were within normal ranges, and there was no pyrexia. Particulate matter that could serve as an alternate nidus of irritation was not present.
Drug Delivery System.
The effect is not due to simple catheter placement or infusion. Intrathecal masses were not noted in saline or clonidine control animals and were most reliably evident at the higher infusion concentrations of morphine. Bolus delivery of morphine at concentrations up to 10 mg/ml repeated daily for 28 days in this dog model did not induce intrathecal mass formation, 9which emphasized the importance of continuous drug exposure to induce mass formation. Other agents delivered for 28 days at the maximum available concentrations have not been shown to cause such intrathecal masses in this dog model, even when delivered at rates up to 100 μl/h (as compared with 40 μl/h here), including baclofen (2 mg/ml), 34clonidine (2 mg/ml, present study), adenosine (5 mg/ml), 35neostigmine (2 mg/ml), 36and brain-derived nerve growth factor (2 mg/ml). 13Finally, the effect is not limited to the canine model. Comparable results have also recently been observed in sheep 37(see Web Enhancement at Anesthesiology Web site at http://www.anesthesiology.org).
An important question relates to the mechanisms by which continuous morphine exposure may lead to the observed effects.
(1) The overall incidence of intrathecal masses was evidently dependent on the dose of morphine in the infusate, with 12 mg/day resulting in masses which appeared in all animals that received this dose reported in this study (i.e. , 5 of 5). As the drug delivery volume was fixed, increasing doses reflect increasing concentrations. Accordingly, it is not clear at this time whether the effects are related to total dose or to concentration. Based on the role of local concentration in drug toxicity, we hypothesize that a given dose (e.g. , 12 mg/day) delivered in a larger volume will reduce the incidence of mass formation.
(2) We do not know if the effects of morphine represent an opioid receptor-mediated effect. Further studies are required to address this issue. The possible contribution of intrathecal morphine glucuronide metabolites on mass formation is not known.
(3) Acute morphine treatment will suppress a variety of immune cell-related functions, including natural killer cell activity, T cell and B cell function, macrophages, and polymorphonuclear leukocytes. 38These immune cells in general possess μ-opioid receptors that are comparable to those in neurons (e.g. , naloxone-sensitive). 39Morphine has been shown to increase the production and release of nitric oxide and has been shown to induce apoptosis in macrophage populations by a p53-dependent mechanism that is naloxone sensitive. 40
(4) Although μ opioids directly suppress chemokine-mediated migration of neutrophils and monocytes, 41–43morphine can also serve as a mitogen, increasing activity of lymphocyte populations and activating mitogen-activated protein kinase cascades in lymphocyte cultures. 44Moreover, in vivo studies have demonstrated that the administration of morphine sulfate to opioid-naive rhesus monkeys may activate quiescent lymphocytes for proliferation and cause enhanced release of interleukin-2. 45These effects have been reported to be naloxone sensitive.
(5) The local pericatheter environment is complex and includes nervous tissue, leptomeningeal cells, and the local vessels in the meninges. In ex-vivo studies, morphine alone had no effect on monocyte migrations, but in the presence of mesangial cells the migration of monocytes across the filter was markedly enhanced. This effect was prevented by a free radical scavenger. 46Opiates can initiate release of nitric oxide in human endothelial cells, granulocytes, and monocytes. 47,48
(6) Morphine will initiate mast cell degranulation. 49Degranulation results in endothelial cells lining microvessels to express an antigen necessary for endothelial-leukocyte adhesion. This interaction was mediated by a TNFα-dependent mechanism. 50Such an adhesion would be the first step in promoting cellular extravasation.
These observations suggest several working hypotheses. First, these effects may reflect an action mediated though dural endothelial cells. Here morphine leads acutely to activation of eNOS and an acute suppression of immunocyte activity. With persistent exposure, there is a cascade that serves to sensitize granulocytes and monocytes to ongoing cytokines and proinflammatory stimuli. We recognize that a modest ongoing stimulus is provided by the catheter that might lead to moderate levels of cytokines that then become sufficient to initiate a local aseptic inflammatory reaction. Continued exposure of immunocytes to morphine in vitro lead to an exaggerated response of monocytes to other proinflammatory stimuli. 51This scenario has been proposed for the glomerulosclerosis observed in opiate-addicted patients. 46Consistent with this mechanism, we showed a marked enhancement of both inducible NOS (iNOS) and endothelial cell NOS (eNOS). Second, morphine in high, sustained concentrations leads to a degranulation of dural mast cells that serve to promote local dural permeability and leukocyte adhesion, resulting in vascular cell migration.
As noted, we cannot determine from this work whether these morphine effects are related to the activation of an opioid receptor. Should that be the case, we speculate that the same phenomena may be observed with other opioid agents and the doses required to induce the intrathecal mass will be some ratio of their relative activity as opioid analgesics. Should this prove not to be an opioid effect, more potent opioids even with similar structures may confer a therapeutic advantage as compared with morphine.
Effects of Clonidine.
In the present work, clonidine at the highest concentration examined was without evident toxicity. The observation that clonidine seemed to have a “suppressive” effect on the pathologic findings otherwise associated with the low dose of morphine was unexpected. Although specific kinetics were not done, cisternal and plasma concentrations of morphine were similar to those noted in the morphine-alone group. We thus do not believe the effects were secondary to an abnormal clearance of morphine. A possible alternative is that clonidine modifies the immunocyte response initiated by morphine. Clonidine has been reported to act by an α-2 receptor to suppress release of cytokines (TNF) 52and at high concentrations to inhibit neutrophil chemotaxis. 53Interestingly, α-1 receptor activity will suppress nitric oxide formation in smooth muscle. 54Clonidine, though considered to be an α-2 agonist, also has measurable α-1 effects. 55Structurally, clonidine resembles the hydrophobic heterocyclic NOS inhibitors, and has been shown to block neuronal-dependent NOS (nNOS) without affecting eNOS activity. 56Finally, clonidine inhibits histamine release from basophils and mast cells, an effect prevented by an H2-antagonist. 57These results are jointly consistent with a possible suppressive effect on activity in immunocytes and on mast cells otherwise activated by μ-opioid agonists.
Comparison with Human Original Investigations
Our observations are consistent with clinical case reports that describe patients receiving continuous opioid infusion who presented with a progressive motor or sensory dysfunction secondary to a local compressive lesion. 58–64,12Coffey and Burchiel reviewed the previously reported cases plus an additional 25 new patients who received high dose opioid infusion between 1990 and 2000. 65In humans, where the masses were resected, the masses consisted, as in the dog, of aggregations of macrophages, neutrophils, and monocytes, with a necrotic center, and there was no evidence of an infectious process. It should be stressed that the humans in whom intrathecal masses have been positively identified are those who have presented with neurologic symptoms.
With regard to doses, we found intrathecal masses in all dogs receiving 12 mg/day of morphine sulfate. Our current hypothesis, however, is not that the pathology arises from the actual doses infused, but rather from the local concentration at the catheter site to which the spinal cord is exposed. Accordingly, the interpretation of these preclinical pathology studies requires consideration of the actual local concentrations in the vicinity of the catheter tip. At present there are no data with chronic low rate delivery of morphine in humans with which we can compare these values. Several parameters may impact the local concentrations to which the local tissue is exposed. These include (1) whether the catheter tip is in the lumbar sac or whether it overlies the lumbar cord proper (where local dilutional volumes of CSF are smaller); (2) the local kinetics around a catheter tip as defined by the complexity of the local structures 66; and (3) the local volume of the intrathecal sac. 67
The time required for mass development in humans is not known. The preclinical data suggest that with fixed dosing, the time of onset is shortest at highest concentrations and can occur within an interval of several weeks. In humans, drug exposure typically involves progressive increment of concentrations and doses over an extended period and the combined use of adjuvant agents such as local anesthetics. Nevertheless, in the Coffey and Burchiel review, 65only four patients had received infusion for less than 6 months. The onset of the neurologic syndromes in 23 patients was characterized as sudden in 6, sudden with prodromal symptoms in 2, and slowly evolving in 15. Whether the effect is limited to morphine is not known. However, human data suggest that chronic intrathecal hydromorphone infusion can also be result in intrathecal masses. 65In preliminary studies, we have shown that hydromorphone delivered at a dose of 3 mg · ml−1· day in the dog would yield a prominent intrathecal mass (Jeff Allen, Ph.D., Department of Anesthesiology, University of California, San Diego, unpublished observation, June 2002).
Recently a consensus conference was held regarding the continuous infusion of spinal opioids for chronic pain management. 68Several points were emphasized: (1) the preponderance of patients receive morphine initially; (2) with regard to morphine, the suggestion was that doses up to 20 mg/day were acceptable and that concentrations should be adjusted to allow as long an interval between refills as possible; and (3) although many patients start with morphine, 50–65% of patients will receive either other opioids alone, including hydromorphone, methadone, fentanyl, or meperidine, or admixtures of morphine and bupivacaine or clonidine. 68We note that over 60% of patients employed infusates formulated by hospital pharmacies. Many of these patients were likely receiving morphine concentrations that exceed what is commercially available (e.g. , 25 mg/ml). Though infrequently discussed in the published literature, clearly current practices often employ concentrations of morphine that are at or near the absolute solubility of morphine sulfate (e.g. , 50–55 mg/ml). It is our premise that such high concentration may account in part for an increased risk of developing a catheter-tip mass. From a historical perspective, the phenomenon of intrathecal masses appears to be relatively recent, 12which may reflect two phenomena. First the early use of implanted pumps was largely restricted to cancer patients with a relatively limited life span. Second, there has been a progressive trend towards the use of higher morphine concentrations. This trend has been driven by the desire to increase the duration between pump refills. The highest concentration previously examined in a large intrathecal animal (dog) model was 10 mg/ml given as a daily bolus for 28 days. 34Hence, it is clear that until the preclinical studies reported here, the actual formulation being delivered in humans was never examined systematically with continuous intrathecal infusion. Presumably, a large number of patients receive other opioids or admixtures of morphine. For these formulations, there are also no systematic investigations of intrathecal safety.
In conclusion, these studies confirm the lack of a direct toxicity to spinal parenchyma in animals exposed for 28 days to morphine sulfate infusion concentrations up to 12.5 mg/ml (12 mg/day). However, such concentrations led to an increased prevalence of aseptic granulomatous masses. These effects were not observed at the doses of clonidine examined, and it appeared that clonidine might in fact suppress the observed pro-inflammatory reaction of morphine. These studies emphasize the importance of careful attention to increased drug requirements and/or changes in neurologic function that arise in the course of such therapy. Although increases in intrathecal drug requirement may signal a change in the pain state and neurologic signs can signal abscess, tumor, aneurysm, or bony defects, an additional suspicion must now be an aseptic process that reflects the formation of an intrathecal inflammatory mass.