Pain Medicine 2010; 11:1010–1016Wiley Periodicals, Inc.
Intrathecal Opioid Therapy for Chronic
Nonmalignant Pain:A Retrospective Cohort Study with 3-Year Follow-Up
Aysel Atli, MD,*Brian R. Theodore, PhD,*
Dennis C. Turk, PhD,*and John D. Loeser, MD*†Departments of *Anesthesiologyand Pain Medicine and
†
Neurological Surgery, University of Washington, Seattle, Washington, USA
Reprint requests to:John D. Loeser, MD, Department of Neurological Surgery, University of Washington 356470, Seattle, WA 98195, USA. Tel:206-543-3570; Fax:206-543-8315; E-mail:jdloeser@u.washington.edu. Abstract
Objective. The objective of this study was the analy-sis of outcomes after intrathecal opioid treatment. Design. Retrospective chart review cohort study. Setting. Tertiary care university hospital and clinic. Patients. Adults of both sexes were included. Interventions. The intervention consisted of the implantation of intrathecal catheter and subcutane-ous programmable pump to deliver opioids. Outcome Measures. These included intrathecal and oral opioid consumption, self-reported pain levels, and complications.
Results. We observed reduction of visual analog scale scores, decrease in oral opioid consumption. Stable long-term (3year) pain reports. We also noted gradual increases in intrathecal opioid consump-tion. Pre-implant opioid consumption was inversely correlated with treatment success. The complication rate was approximately 20%.
Conclusions. We conclude that intrathecal opioids without adjunctive intrathecal medications have a favorable outcome. Some patients are able to elimi-nate oral opioids. Results seem stable for prolonged periods, although some increase in intrathecal opioids dosing may be required. 1010
Key Words. Intrathecal; Opioids; Catheter; Pump; Outcomes; Tolerance; Complications Introduction
The recognition of dorsal horn mu receptors led to the development of spinal administration of opioids, in hopes that better pain relief and fewer side effects would result [1].In 1991, the US Food and Drug Administration (FDA)approved a programmable pump for drug infusion (Medtronic,Minneapolis, MN) and this device has led to a rapid expansion in the use of intrathecal opioids since 1995when the FDA granted approval for intrathecal morphine (intrathecaldrug delivery system, IDDS) [2].Although several consensus guidelines have been pub-lished, they are not well-grounded in evidence for out-comes [3].Case series and several systematic reviews of IDDS have been reported [4–6].A deficiencyof long-term treatment outcomes data characterizes the literature on opioids in the management of chronic, noncancer pain syndromes [6–8].
The available literature, however, is limited and as sum-marized in structured reviews, many questions remain unanswered about this therapeutic modality, its efficacy,complications, and long-term dosing issues [5,6].It has also been suggested, without much supporting evi-dence, that IDDS will reduce the need for oral opioids [9].
Proper patient selection strategies, choices of drugs, delivery strategies, catheter locations, and measures to prevent complications of IDDS remain unclear. There is a need for carefully followed case series to establish more rigorous treatment guidelines [5].In this report, we describe a consecutive cohort that was carefully followed and offers information about the effectiveness, safety, adverse effects, and impact of IDDS on oral opioid intake over a period of 3years. Effectiveness rather than efficacycan be evaluated by study of this cohort. The set of a priori hypotheses investigated in this study included:1) signifi-cant and clinically meaningful reductions in pain intensity levels following administration of intrathecal opioid will be observed over the duration of the study; 2) significantreduction in oral opioid intake will be observed following administration of intrathecal opioids; and 3) higher initial dosage of oral opioids will be associated with poorer outcome at 3years, as measured by visual analog scale (VAS)for
pain.
Methods
This retrospective chart review cohort study was approved by the Institutional Review Board of the Univer-sity of Washington. All of the patients were referred to one of the authors (JDL)for advanced interventional pain man-agement. A detailed medical history and physical exami-nation focusing on patients’chief symptoms and the history of treatments received was undertaken with all patients. All had failed oral and/ortransdermal opioid therapies, either because of inadequate pain relief or unmanageable side effects. Imaging and electrodiagnos-tic studies were reviewed and additional studies obtained when necessary to establish a diagnosis and to rule out any surgically treatable disease. Informed consent was obtained by providing patients with a summary of pub-lished literature, a DVD explaining intrathecal therapy, a clinic handout, a sample pump and catheter, and a per-sonal interview with the implanting surgeon. Patients who were appropriate and who elected to proceed with intrath-ecal opioid therapy were advised to try to taper their oral opioids and were scheduled for admission shortly thereafter.
Under general anesthesia, in the lateral decubitus posi-tion, a 3-cm incision was made in the midline over the spinous processes of L2-L4. A #15spinal needle was inserted into the intrathecal space from a paramedian approach and a Medtronic intrathecal catheter was threaded through the catheter under fluoroscopiccontrol until the catheter tip lay in the mid-thoracic region. The needle was then withdrawn around the catheter, and the catheter clamped at its end. The catheter was then anchored to the fascia with an anchoring device. A stab wound was made in the skin of the lower lateral chest wall and a DuPen trochar (Bard,Salt Lake City, UT) with catheter and Pall filter(PallInc., Port Washington, NY) attached was threaded into the back incision. The DuPen catheter was then trimmed to the appropriate length, as was the intrathecal catheter and they were joined. The catheter system was tested with saline for injection to prove its patency. The incision was then closed in layers with interrupted sutures and the patient awakened from anesthesia and transported to the recovery room. An infusion of preservative-free morphine was then started at a rate of 1mg/24h, or, rarely, hydromorphone at 0.2mg/24h or sufentanil at 1mg/24h. The only preservative-free opioids available were morphine, hydro-morphone, and sufentanil. Patients were frequently evalu-ated and infusion rate changed as often as every few hours until satisfactory pain control was obtained. Intoler-ance to the firstdrug utilized resulted in changing to another drug or even a third. If the intrathecal drug resulted in the report of pain relief, oral opioids were tapered as rapidly as tolerated. On the fifthpost-trial day, a decision was made as to whether or not a satisfactory intrathecal drug and dose had been established, by mutual agreement of the physician and the patient. The criteria for making this decision included both adequacy of
Intrathecal Opioids
pain relief (VAS
The patient was then returned to the operating room, where, under general anesthesia, either the catheter system was removed (failedtrial) or a Medtronic Syn-chromed pumpwas implanted. The initial infusion rate was set at the previously determined dosing level. The patient was usually discharged on the firstpostop-erative day. After implantation, all prescribed opioids were managed by our clinic. During follow-up, doses were adjusted and drugs modifiedwhen required. Intrathecal drug was changed in 8patients in the firstyear after implantation, 3patients in the second year, and 4patients in the third year. The decision for drug change or dose modificationwas made on a collaborative basis by the physician and patient. All pump-and catheter-related issues were managed by our service and our records contain complete information on complications, doses, and other medications.
Adjustments in both oral and intrathecal medication could be made at the time of refillingof the pump (3–6month intervals), or at any time that the patient wished to sched-ule an appointment for dose modification.We included in this study all consecutive patients who had their firstpump implanted between January 1, 1999and December 31, 2003. For the purposes of this study, we report 3-year outcome data on this cohort. Associations between diag-nosis groups and demographic factors (sexand age) were tested using the one-way analysis of variance (ANOVA ) for age, and the c2test for sex. Main analyses for outcomes included repeated-measures ANOVA with planned Helmert contrasts to investigate the trend of opioid dosage and pain VAS across time. The independent-groups t -test was utilized for a responder analysis on the association between opioid dose levels and improvements of Ն30%and Ն50%on pain VAS. Finally, correlations between baseline oral opioid dose and intrathecal dose levels across time were tested using Pearson’sproduct–moment correlations. Results
Demographic Data
A total of 57patients received new intrathecal catheters and pumps from 1999to 2003. We have categorized them into fivediagnostic groups:1) failed back surgery syndrome (FBSS)(n=28); 2) neuropathic pain (n=16); 3) visceral pain (n=5); 4) malignancy (n=2); and 5) miscel-laneous (n=6) (Table1). FBSS was definedas a patient who had at least one spinal surgical procedure and per-sisting leg or back pain. We lack the data to describe the exact components of the back or leg pain and neurologic findingsin this group; however, they were all thought not to be candidates for further spinal surgery. Neuropathic pain included patients with complex regional pain syn-drome, radiculopathy independent of spinal surgery, mul-tiple sclerosis, arachnoiditis, spinal cord injury pain, and
1011
Atli et al.
Table 1Patient demographic characteristics
Age, Mean Diagnosis N Female Male (SD)FBSS
28161251Ϯ12Neuropathic pain 167948Ϯ8Visceral pain 54147Ϯ8Malignancy 20250.5Ϯ19Miscellaneous
6
3
3
50Ϯ14
FBSS =failed back
surgery syndrome,
SD =standard
deviation.
peripheral neuropathy. Abdominal pain included patients with pancreatitis and other visceral pain syndromes. Two patients with malignant diseases (thyroidcarcinoma and mesothelioma) were implanted but not included in the analysis of results because of their shortened life expect-ancies due to their diseases. The miscellaneous category included patients with lipomeningomyelocele, necrotizing fasciitis, idiopathic scoliosis, axial back pain of uncertain etiology, and osteomyelitis. There was no significantdif-ference in the number of males to females included in the study (n=27males, 30females) and the mean age for the total sample was 49.7(standarddeviation [SD]=11.1). There were no associations of age or sex within the fivediagnostic groups.
Seven patients did not complete a 3-year follow-up:the two with cancer mentioned previously, one who was diag-nosed with cancer after his pump was implanted for another diagnosis and rapidly deteriorated from his malig-nant disease, two patients who had early infections and multiple pump and catheter revisions rendering the therapy impossible to evaluate, and two who emigrated from the Seattle area and were lost to follow-up. Finally, one additional patient was excluded from the analysis of pharmacologic and pain level outcomes due to exces-sively high preimplantation oral opioid intake, rapid failure to maintain pain relief with intrathecal opioids, major psy-chiatric difficultiesnot discerned during screening, all of which rendered the patient an extreme outlier. Therefore all opioid and pain analyses are based upon a sample of 49patients. Figure 1provides a Consolidated Standards of Reporting Trials diagram summarizing the flowof the patients through the duration of the study. Pain Measurement
We assessed pain levels using a VAS of 0–10(0=”Nopain,”10=”Worstpain possible”)based on patient self-report at the time of initial assessment (baseline),at the time of firstrefill(usually2–3months postimplant), and at 1, 2, and 3years after implant. Figure 2illustrates the changes in the VAS pain scales over this period of time. Planned contrasts using Helmert coefficientsindicated a statistically significantreduction in pain VAS scores from baseline to all subsequent epochs [F (1,48)=125.0, 1012
P Յ0.001, partial h2=0.72].However, a statistically sig-nificantincrease in pain VAS scores occurred between the firstrefilland all subsequent time periods [F (1,48)=7.1, P =0.011, partial h2=0.13].There was no significantchange in pain VAS scores among years 1, 2, and 3. This result was consistent using both the observed data, as well as imputed data for dropouts/patientswith pumps removed based on the baseline-observation-carried-forward method.
A responder analysis was also conducted on the pain VAS scores to determine the proportion of patients who achieved either 30%or greater or a 50%or greater pain relief relative to baseline levels (Figure3). At the time of the initial refill,33patients (67%)reported Ն30%reduction in their pain VAS relative to baseline. A decreasing trend was observed in the percent maintaining this threshold so that by the third year 18patients (37%of the implanted
Figure 1Consolidated Standards of Reporting Trials diagram of the study cohort. VAS =visual analog
scale.
Figure 2Change in visual analog scale (VAS)pain scores.
Intrathecal Opioids
Figure 3Clinical criterion of improvement based on Ն30%and Ն50%reduction in pain relative to baseline levels. VAS =visual analog
scale.
patients and 54.5%of those meeting the Ն30criterion at firstrefill)maintained a Ն30%reduction in pain VAS. For the Ն50%reduction in pain VAS relative to baseline, 23patients (47%)met this criterion at the time of the firstrefill.Once again, a decreasing trend was identifiedso that by the third year follow-up nine patients (18%of the total sample of implanted patients, 39.1%of those who meet the Ն50%criterion at the firstrefill)maintained the 50%reduction criterion relative to baseline. Oral Opioid Dosages
All opioids were converted into the equivalent doses of morphine using standard conversion tables [10].The oral opioids consumed by the implanted patients at baseline and at years 1, 2, and 3are depicted in Figure 4. We found that the mean baseline oral opioid dosage in mor-phine or its equivalent was 184mg/24h. Planned con-trasts using Helmert coefficientsindicated a significantreduction in oral opioid intake between baseline and all subsequent intervals [F (1,41)=16.6, P
3years of treatment. These changes in oral opioid con-sumption were statistically significantwhen compared with baseline. There were no significantdifferences between years 1and 3indicating that the substantial reduction in oral opioids following implantation was main-tained, amounting to a 69%reduction from baseline level. At the 3-year follow-up, 12of the 49patients (24%)in the outcomes analysis had completely ceased all oral opioid use. There is no question that our interactions with the patient and the desire to reduce oral opioids played a significantrole in postimplantation oral medication consumption.
In addition, a significantcorrelation of r =0.49(P =0.001) was observed only between baseline oral opioid dose and 3-year pain VAS scores. This relationship was especially prominent when considering patients who achieved Ն50%reduction in pain levels at 3-year follow-up (meanbaseline oral dose =68.1, SD =64.0) vs those who did not (meanbaseline oral dose =215.6, SD =256.8). This difference between the responder groups was statistically significant[t (40)=2.98, P =0.005, Cohen’sd =0.79]after adjusting for heterogeneous variances in baseline oral opioid dosage levels between the two groups. Hence, higher baseline oral dose was associated with lower like-lihood of long-term pain relief with intrathecal opioids. Intrathecal Opioid Dosages
For the analysis of intrathecal opioid dosages, all other drugs were converted into morphine equivalents using standard conversion tables [10].The intrathecal doses of opioids received by the patients are depicted in Figure 5. The starting pump dose was determined by the dose required to achieve adequate pain relief during the trial. We are aware of the fact that this is above the dose on the package label for Infumorph (Baxter,Deerfield,IL). FDA-mandated labeling often does not reflectactual usage patterns. No adjunctive intrathecal medications were administered. The mean dose of morphine or its equiva-lent at discharge from the hospital was 6.5mg/24h, this increased to 9.3mg/24h at 1year, 10.0mg/24h at 2
1013
Figure 4Change in oral opioid dose.
Atli et al.
Figure 5Change in intrathecal opioid dose.
years, and 12.2mg/24h at 3years, an 88%increase over the 3-year duration. Planned contrasts using Helmert coefficientsindicated that there was a statistically signifi-cant increase in intrathecal opioids dosage between dis-charge and all subsequent time periods [F (1,38)=16.0, P
No significantassociations were observed between initial intrathecal dose level and pain intensity at subsequent follow-up periods (i.e.,all correlation coefficientsbetween initial dose level and all pain levels were not significantlygreater than 0). However, significantpositive correlations of large magnitude (r >0.50) were observed between baseline oral opioid dose at each of the four intrathecal dose levels across the duration of the study:r =0.60(P
Fourteen complications occurred in 10of the 57patients for whom we had data (Table2). These included 5patients with infections, 3with catheter revisions, 2with a seroma at the pump site, and 2who developed an intrathecal granuloma. By the end of the study period, 10patients had their pumps and catheters removed (20%),either because of infection or loss of efficacyand 2patients were 1014
not using their pumps although they had not been explanted. By any standard, these 12patients must be considered failures, leading to a treatment failure rate of 24%.This rate is consistent with that previously reported by others [11,12].The infection rate is in the range of those previously published. The infections were manifest at the pump insertion site and are therefore not likely to be due to the method of trialing intrathecal drug administra-tion. No patient developed meningitis. Discussion
The primary objectives of this study were to determine the long-term efficacyand safety of intrathecal opioid therapy in patients with chronic, noncancer pain. As depicted in Figure 2, intrathecal opioids were effective in reducing pain severity for at least 3years after the initiation of therapy. As we could not classify our patients as to neu-ropathic or nociceptive pain at baseline, we cannot deter-mine if one type of pain is more or less likely to respond to intrathecal opioids.
It is interesting to note that pain severity scores were lower at the time of the firstrefillof the pump than they were at annual intervals thereafter, suggesting that some degree of tolerance to intrathecal opioid administration did occur, although many other nonspecifictreatment effects could also have played a role. However, the statistically signifi-cant reduction in pain severity scores was maintained for 3years. The clinical significanceof a 2-point reduction (from7.7/10to 5.7/10)is less clear, but it is probably meaningful as well [13].This outcome is consistent with those reported by Anderson and Burchiel [9].However, their follow-up was only 2years and we are unaware of any reports with follow-up of 3-year duration, as in our study.
One of the arguments in favor of intrathecal opioids has been the reduction and, hopefully, elimination of oral opioids. We recognize that opioid consumption by the oral route is a function both of the patient’sdesire for the drug and the physician’swillingness to prescribe it. All of these patients had been told that successful intrathecal therapy would be definedto some degree by reduction and, hope-fully, cessation of oral medications and all were asked in the postimplantation epochs if they could tolerate a
reduc-
Table 2Complications of catheter and pump implantation
No. of Percentage Complications
Patients (%)Wound infection
5/578.8Intrathecal granuloma 2/573.5Seroma
2/573.5Catheter migration/fracture3/575.3Pump malposition
2/57
3.5
tion in oral medications. We recognize that the potency of intrathecal opioids is much greater than oral opioids and functional opioid dosage may not be reduced even though total intake is diminished. The question of the develop-ment of tolerance is an important issue in the use of intrathecal opioids. As shown in Figure 5, there were sig-nificantincreases in the intrathecal opioid dosage from discharge from the hospital at the time of pump and catheter implantation to the firstyear follow-up and at each yearly interval thereafter. The mean daily morphine (orequivalent) dose was 6.5mg/24h at discharge and increased to 12.2mg/24h at 3-year follow-up, an 88%increase. We do not know if continued therapy thereafter would show progressive dose increments or if the dose would stabilize; however, the increasing dose trend was just as marked in year 2–3as in prior years. Similar data suggesting the progressive increase of intrathecal dosing over time has been reported in many other case series [4,9,11,14–16].
Another potential confounder in our cohort was the strat-egy of surgically implanting the catheter at the initial trial. Patients may have desired to have the pump implanted and not assessed their pain relief accurately. On the other hand, we believed that this strategy provided the best test of the intrathecal route as the drug to be administered was the test drug and it was within the cerebrospinal fluid(CSF).Unless the patient had a prior history of morphine intolerance, this was the initial drug for all patients. All patients who did not tolerate morphine intrathecally were trialed on hydromorphone and, if necessary, sufentanil. Long-term outcome was not a function of the drugs that were utilized in the trial or the initial pump filling.One of the unique aspects of our study was that the patients received only mu agonist opioids via their pumps and catheters. No adjunctive medications were adminis-tered intrathecally. This contrasts with most of the other reported series, in which a variety of non-opioid drugs were coadministered, making it difficultto describe the effects of intrathecal opioids per se [2,4,15,16].Our long-term results are comparable with those reported by others, suggesting that the benefitsof adding medications to the infusion are minimal, at best, although differences in the populations of patients reported in various studies makes comparisons difficult[14,16].Whether or not a dose-sparing effect occurs such that the inclusion of adju-vant medication permits a lower dose of opioids is unknown. The use of compounded drugs for intrathecal administration certainly increases costs and risks of errors in medication dosages; it also appears to add to the risk of developing a granuloma at the catheter tip [17].We also assessed the side effects of intrathecal opioid therapy over 3years (Table3). Of 54patients for whom we had data on side effects, 48reported none (89%),5patients reported significantedema, 1had myoclonic jerking, 3had urinary retention, and 2reported nausea and vomit-ing. These were the only reported side effects. Another limitation of this study was that the occurrence of side effects was based on patients’spontaneous reports to open-ended query. It is reasonable to expect that addi-
Intrathecal Opioids
Table 3Intrathecal opioid side effects
No. of Percentage Side Effects
Patients (%)Fluid retention, edema 5/578.8Urinary retention 3/575.3Myoclonic jerks 1/571.8Nausea/vomiting
2/57
3.5
tional problems could have been ascertained by more systematic and in-depth interrogation either by interview or a standard side effect checklist. Conclusions
The results of this cohort study show that intrathecal opioid therapy does produce long-term reductions in pain severity scores and oral opioids consumption. The signifi-cant complication rate is 24%.The failure rate for this treatment strategy is almost 25%and intrathecal opioid therapy is expensive when contrasted with other modes of drug delivery. Limitations of this study include the absence of a control group, its relatively small size, the lack of data more than 3years after implantation, and the lack of documentation of functional status either preimplantation or during follow-up. Despite these limitations, several advantages inherent in the present study are that:1) a single surgeon provided all clinical care including implan-tation of the IDDS; 2) detailed education of patients occurred prior to the decision to conduct a trial; 3) inclu-sion of all consecutive patients who were implanted with a pump and catheter in the epoch; 4) long-term follow-up of at least 3years postimplantation; and 5) the exclusive intrathecal use of pure mu agonists. The decision to seek pain relief with an intrathecal catheter and pump should be shared with the patient, as there are both potential long-term benefitsand risks.
References
1Yaksh TL, Rudy TA. Analgesia mediated by a direct spinal action of narcotics. Science 1976;192(4246):1357–8.2Wallace M, Yaksh TL. Long-term spinal analgesic delivery:A review of the preclinical and clinical litera-ture. Reg Anesth Pain Med 2000;25(2):117–57.3Hassenbusch SJ, Portenoy RK, Cousins M, et al. Polyanalgesic Consensus Conference 2003:An update on the management of pain by intraspinal drug delivery—Reportof an expert panel. J Pain Symptom Manage 2004;27(6):540–63.4Smith HS, Deer TR, Staats PS, et al. Intrathecal drug delivery. Pain Physician 2008;11(2Suppl):S89–S104.5Turner JA, Sears JM, Loeser JD. Programmable intrathecal opioid delivery systems for chronic noncan-1015
Atli et al.
cer pain:A systematic review of effectiveness and complications. Clin J Pain 2007;23(2):180–95.6Noble M, Tregear SJ, Treadwell JR, Schoelles K. Long-term opioid therapy for chronic noncancer pain:A systematic review and meta-analysis of efficacyand safety. J Pain Symptom Manage 2008;35(2):214–28.7Chou R, Fanciullo GJ, Fine PG, et al. Clinical guide-lines for the use of chronic opioid therapy in chronic noncancer pain. J Pain 2009;10(2):113–30.8Abs R, Verhelst J, Maeyaert J, et al. Endocrine con-sequences of long-term intrathecal administration of opioids. J Clin Endocrinol Metab 2000;85(6):2215–22.9Anderson VC, Burchiel KJ. A prospective study of long-term intrathecal morphine in the management of chronic nonmalignant pain. Neurosurgery 1999;44(2):289–301.10Souter KJ, Fitzgibbon D. Equianalgesic dose guide-lines for long-term opioid use—Theoreticaland prac-tical considerations. Semin Anesth 2004;23:271–80.11Angel IF, Gould HJ Jr, Carey ME. Intrathecal morphine
pump as a treatment option in chronic pain of nonma-lignant origin. Surg Neurol 1998;49(1):92–8;discus-sion 98–9.
1016
12Follett KA, Burchiel K, Deer T, et al. Prevention of
intrathecal drug delivery catheter-related complica-tions. Neuromodulation 2003;6(1):32–41.13Farrar JT, Young JP Jr, LaMoreaux L, et al. Clinical
importance of changes in chronic pain intensity mea-sured on an 11-point numerical pain rating scale. Pain 2001;94(2):149–58.14Kumar K, Kelly M, Pirlot T. Continuous intrathecal
morphine treatment for chronic pain of nonmalignant etiology:Long-term benefitsand efficacy.Surg Neurol 2001;55(2):79–88.15Krames ES. Intraspinal opioid therapy for chronic non-malignant pain:Current practice and clinical guide-lines. J Pain Symptom Manage 1996;11(6):333–52.16Mironer Y, Haasis J, Chapple I, et al. Efficacyand
safety of intrathecal opioid/bupivacainemixture in chronic nonmalignant pain:A double blind, random-ized, crossover, multicenter study by the National Forum of Independent Pain Clinicians (NFIPC).Neuro-modulation 2002;5(4):208–13.17Coffey RJ, Burchiel K. Inflammatorymass lesions
associated with intrathecal drug infusion catheters:Report and observations on 41patients. Neurosurgery 2002;50(1):78–86;discussion 86–7.