Implanted Drug Delivery Systems for Control of Chronic Pain


The intrathecal delivery of medications has a long history and holds the promise of providing significant analgesia with diminished medication-related side effects. Early work dates to the late 1800s and centers on surgical anesthetics, but the discovery of opioid receptors in the 1970s led to the use of intrathecal medications to provide analgesia. This area of pain medicine has since expanded to include a range of medications from several different drug classes and has greatly benefited from the development and evolution of implantable pain pumps. Overall evidence varies with medication class, and randomized controlled studies are generally sparse. Opioids are well supported for use in malignant pain states, but evidence for their use in treating benign pain is weak and inconsistent. Baclofen use is well supported in pain conditions that include an element of spasticity but is only modestly supported in other neuropathic pain states. Ziconotide is associated with significant dose-dependent central nervous system and psychiatric side-effects, but there is good evidence for its use in treating neuropathic pain. Bupivacaine and clonidine are both well supported as adjuncts in mixed intrathecal infusions but not as sole agents. Corticosteroids are weakly supported for use in postherpetic neuralgia. Potential complications of intrathecal drug delivery can be related to medications or their delivery systems and must be considered when initiating this therapy.


bupivacaine, cancer pain, intrathecal drug delivery, neuropathic pain, opioid, ziconotide



The delivery of medication into the cerebrospinal fluid (CSF) has a long history and can be traced to the spinal anesthetic works of Corning and Bier in the late 1800s. With the discovery of opioid receptors in the central nervous system (CNS) in the 1970s, it was postulated that the CSF could provide a route not only for anesthesia but also for providing analgesia. The use of intrathecal (IT) medications has expanded over the ensuing decades to involve several classes of medications, indications beyond chronic pain, and has benefited from advancements in implantable pump technology.

The primary rationale for instituting intrathecal drug delivery (IDD) is to achieve similar or superior therapeutic effect when compared with oral medications, while avoiding typical dose-dependent side effects. Some medications can safely be delivered long term via implantable pumps, whereas others are primarily indicated for single-shot applications in discrete pain conditions. The decision to proceed with IT therapy is based on several factors, and potential risks and benefits must be weighed carefully, as with any invasive treatment.

Selection criteria, as detailed next, should be used to identify patients that may benefit from IDD:

  • Chronic pain or spasticity refractory to prior treatments

  • Presence of prohibitive side effects with other medication delivery routes

  • Lack of confounding psychological comorbidities

  • Absence of systemic or surgical site infection

  • Absence of coagulopathy

  • Lack of CSF flow obstruction

  • Life expectancy greater than 3 months

  • Demonstrated efficacy with trial implantation

Patients meeting these criteria for therapy with an implantable pump should receive a trial to determine if the selected IT therapy will provide adequate clinical effect, although there are some experts who advocate for proceeding straight to implantation in well-selected patients with cancer pain. Although epidural and IT single-shot trials are described, the preferred method involves placement of an IT catheter to best mirror potential implant conditions. Hospital admission is generally recommended to allow for dose titration while monitoring for adverse effects, mainly because of the possibility of delayed respiratory depression with rostral spread of opioids.


Potential complications of IDD may be procedurally or medication related. Procedural complications can occur after single-shot injection or pump implantation and include superficial and neuraxial infection, peripheral and neuraxial bleeding, postdural puncture headache, and catheter-tip granuloma formation. Infections or bleeding within the neuraxis are rare but carry the potential for catastrophic permanent sequelae. Suspected neuraxial hematoma or infection necessitates prompt imaging and neurosurgical evaluation. Anesthesia risks must also be considered prior to surgical pump implantation, particularly in patients with multiple comorbidities or advanced malignancy.

Patients with implantable pumps are also at risk of complications during required pump refills. Risks involve administration of incorrect medication, pump reprogramming error, and improper refill technique. Misidentification of the pump refill port can result in delivery of medication into the subcutaneous pocket rather than the pump itself, which can lead to catastrophic consequences. An inadvertent “pocket fill” creates a large depot of medication that can be absorbed into the body at a much greater than intended rate. A 2011 report from Medtronic cited 351 cases of pocket fills between 1996 and 2010, eight of which resulted in patient death. Needle placement must be confirmed by adhering to recommended refill procedures, imaging via ultrasound or fluoroscopy should be used when the refill port cannot be palpated, and patients with suspected pocket fill should be closely monitored.

Withdrawal syndromes can also occur with pump malfunction or programming error, and the severity of symptoms varies with the class and dose of medication being delivered. Opioid withdrawal symptoms include increased pain, anxiety, sweating, diarrhea, and vomiting. Withdrawal from opioids, although unpleasant, is generally not life threatening and can be treated with supportive care and delivery of opioids via oral or parenteral routes. A more serious clinical entity is represented by abrupt withdrawal from gamma-aminobutyric acid (GABA) agonists (benzodiazepines or baclofen). Sudden cessation of IT baclofen can result in a constellation of symptoms that includes increased spasticity, hyperthermia, seizures, coma, rhabdomyolysis, multisystem organ failure, and death. Treatment of suspected cases of baclofen withdrawal involves initiation of oral baclofen, which is not always sufficient to prevent withdrawal, and investigation into the cause of interrupted IT delivery. A similar range of symptoms is seen in cases of benzodiazepine withdrawal, but this syndrome is seldom encountered given the relatively infrequent use of benzodiazepines in long-term infusions.

Table 76.1 summarizes the common and serious adverse effects for classes of IT medications.

TABLE 76.1

Class-Specific Adverse Effects

Class of Medication Common Adverse Effects Serious Adverse Effects
Opioids Constipation, sweating, nausea, urinary retention, pruritis Respiratory depression, granuloma with morphine or hydromorphone
Calcium channel blockers Dizziness, nystagmus, confusion, fever Psychosis, seizure, suicidal ideation
Local anesthetics Weakness, edema Paresis, bowel or bladder dysfunction
Gamma-aminobutyric acid agonists Sedation, hypotonia, headache Life-threatening withdrawal syndrome, respiratory depression with overdose
Alpha-2 adrenergic agonists Hypotension, nausea, sedation, bradycardia Hallucination, respiratory depression
Corticosteroids Adrenal suppression with overuse Adhesive arachnoiditis

Medication Classes

Although only morphine, baclofen, clonidine, and ziconotide are currently approved by the US Food and Drug Administration for use in implantable IDD systems, a variety of other medications continue to see use as off-label adjuncts or sole agents delivered by subarachnoid injection. Combinations of several classes of medications are also used when monotherapy fails to provide adequate analgesia, most commonly in the form of mixed opioid and local anesthetic solutions.


Research regarding the analgesic properties of IT opioids began in the 1970s, and they remain the prototypical and most widely used IT analgesic. Originally reserved for treatment of cancer-related pain, IT opioids have increasingly been used for nonmalignant pain in correlation with rising rates of oral opioid use and spine surgery.

Compared with oral opioids, IT opioids can achieve a similar clinical effect at approximately 1/300th of the dose. This dramatic increase in effect is due to the high concentration of opioid receptors in the substantia gelatinosa of the spinal cord, where they synapse with ascending pain tracts. Binding of opioids to the Mu receptor results in presynaptic and postsynaptic effects that inhibit transmission of pain signals from the periphery to the brain. Presynaptic binding leads to decreased release of pronociceptive neurotransmitters, whereas postsynaptic binding leads to decreased neuronal action potential.

IT opioids must remain in the CSF to exert their effect on spinal Mu receptors, and their duration within the spinal fluid is primarily dependent on their degree of hydrophilicity. More hydrophilic medications, such as morphine and dilaudid, remain solubilized in the CSF for extended periods of time, resulting in an extended duration of action as well as spread within the CSF beyond the immediate location of the catheter tip. Conversely, lipophilic opioids, such as fentanyl or sufentanil, are poorly solubilized in the CSF and rapidly diffuse away from the spinal column. Consequently, lipophilic opioids have a shorter duration of spinal action and tend not to distribute within the CSF beyond the point of entry. Table 76.2 summarizes the conversion ratios for commonly used opioids, baclofen, and midazolam.

TABLE 76.2

Conversion Guidelines Between Routes of Administration for Intrathecal Medications

Agent Oral (mg) Parenteral (mg) Epidural (mg) Intrathecal (mg)
Morphine 300 100 10 1
Hydromorphone 40–60 10 2 0.2
Fentanyl 1 0.1 0.01
Sufentanil 0.1 0.01 0.001
Baclofen 1 0.005
Midazolam 5 1–2

Note: Doses represent relative conversion factors for each agent, not equianalgesic conversion between agents.

The evidence supporting the use of IT opioids in the management of chronic pain is generally positive but is based mostly on uncontrolled prospective and retrospective studies. The lone randomized trial, published by Smith et al. in 2002, compared the effects of IDD plus conventional therapy with conventional therapy alone. A total of 202 patients with refractory cancer pain were randomized to receive either IDD with opioid or conventional medical management, which consisted of all typical medical and interventional therapies excluding IDD and cordotomy. Primary outcome was at least a 20% decrease in pain as measured on a visual analog scale (VAS) coupled with at least a 20% decrease in drug toxicity at 4-week follow-up. Secondary outcomes included several measures, such as differences in individual drug toxicities, quality of life, and health care resource use. The results showed that a higher proportion of the IDD group achieved the primary outcome at 4 weeks (57.7% vs. 37.5%) and a lower percentage of patients achieved neither the threshold 20% reduction in VAS score nor the toxicity end point (11.3% vs. 23.6%). Interestingly, the IDD group also displayed a trend towards increased survival at 6-month follow-up (54% vs. 37%).

Nonrandomized studies have incorporated several types of trial designs but have generally shown good benefit with IT opioids for both malignant and nonmalignant pain. A large retrospective study by Paice et al. ( n = 429) was conducted via surveys completed by patients undergoing IDD and their physicians. Overall results showed a mean pain reduction of 61% with IDD, and 66.7% of patients reported that they were very satisfied with IDD. A high device malfunction rate was also noted (20.1%), which consisted mainly of catheter-related issues such as kinking and withdrawal. A similar study by Winkemüller and Winkemüller ( n = 120) showed mean pain reductions of 67% at 6 months and 58% at 48-month follow-up.

An advent in opioid IDD is the practice of “microdosing,” wherein patients are weaned from all opioids prior to initiation of IT opioid therapy at much lower than typical doses. A 2011 case series reported experience in 20 patients with nonmalignant pain who were weaned from oral opioids and then maintained opioid free for 6 weeks prior to a trial of IT morphine. The dose of morphine was initiated at 25 μg/day on day 1 of the trial and was increased sequentially until satisfactory analgesia was obtained. The average efficacious dose was 140 μg/day. Similar results were shown in the 2016 prospective study by the same researchers that also followed their preimplantation protocol. Seventy-three patients with nonmalignant pain underwent weaning and a trial with microdose opioids, with 60 continuing to pump implantation. Patients were followed for 36 months, and over that time period the opioid dose increased from 221 μg/day of morphine equivalents to 325.4 μg/day, which still represented a substantial dose reduction from a typical IT regimen. The 2012 study by Hamza et al. also used microdosing of opioids in patients with nonmalignant pain, but patients were required to wean from only 50% of their opioids prior to the IT trial. Those patients who had a positive response to trial infusion were then weaned completely from opioids over 3 to 5 weeks and remained opioid-free for 7 to 10 days prior to pump implantation. Significant pain relief was observed at 6-month follow-up, with average pain decreased from a baseline of 7.47 to 2.97. Mean morphine equivalent dose was approximately 1.4 mg/day at 6-month follow-up and had increased to approximately 1.58 mg/day at 36-month follow-up.

Despite these promising results, microdosing remains an area of controversy. It is unclear whether the requirement for pretrial and preimplant weaning leads to selection of patients who inherently require less opioids, or what the magnitude or duration of the wean should be. For example, the 1-week opioid-free period used in the Hamza et al. study is probably not sufficient for any opioid-induced hyperalgesic effect to wear off. The weaning requirement also likely precludes using microdosing in patients with malignant pain. Randomized trials comparing standard trialing methods and regular opioid dosing to microdosing are required.

Adverse effects with IT opioids include those effects also associated with other routes of opioid administration, including constipation, sweating, nausea, urinary retention, hypothalamic-pituitary axis suppression, and sedation. These effects are generally dose dependent, and most are therefore less prevalent with IT versus other routes of delivery. Side effects that may be more frequent with IT opioids can include urinary retention, pruritis, and lower extremity edema. Respiratory depression may be dose dependent but can also occur at lower doses due to rostral spread of opioid to the respiratory centers in the brainstem. Delayed respiratory depression may occur within 24 hours even after a single bolus dose, but the risk appears to be elevated with doses greater than 0.2 mg or with catheter placement higher in the spinal column.

IT morphine and dilaudid are also associated with the formation of inflammatory masses at their site of delivery into the IT space. These catheter-tip granulomas can result in catastrophic neurologic sequelae due to a mass effect on the spinal cord, and treatment can require surgical removal via laminectomy. Granulomas appear to form due to an inflammatory process at the catheter tip and are composed of multiple types of immune cells to include eosinophils, macrophages, lymphocytes, and fibroblasts. Typical symptoms include those of cord compression, such as new or worsening neuropathic pain, neurologic deficits, and bowel or bladder dysfunction. Mild cases may improve with rotation to another opioid, replacement of the medication with saline, or caudad revision of the IT catheter, but more severe cases typically require neurosurgical consultation for consideration of granuloma resection. Given the time required for granuloma formation, they are more frequently observed in patients with nonmalignant pain, but the primary risk factors for formation are increasing dose and/or concentration of morphine or dilaudid. Lower daily doses and concentration appear to be protective, and the risk has been shown to be highest in patients receiving high daily doses of more concentrated solutions. Table 76.3 summarizes recommended dose ranges for medications delivered via both infusion and single shot.

TABLE 76.3

Dose Ranges for Intrathecal Infusion and Single-Shot Injection

Agent Typical Infusion Dose Typical Starting Dose Typical Bolus Dose
Morphine 1–20 mg/day 0.1–1 mg/day 0.2–1.0 mg
Hydromorphone 0.5–10 mg/day 20–200 μg/day 40–200 μg
Fentanyl 20–300 μg/day 20–75 μg/day 25–75 μg
Sufentanil 2–100 μg/day 1–20 μg/day 5–20 μg
Bupivacaine 4–30 mg/day 1–4 mg/day 0.5–2.5 mg
Baclofen 50–1000 μg/day 50–200 μg/day 50–75 μg
Ziconotide 5–19 μg/day 0.5–2.4 μg/day 1–5 μg
Clonidine 30–1000 μg/day 30–100 μg/day 25–100 μg
Midazolam n/a n/a 1–2 mg
Dexmedetomidine n/a n/a 5–10 μg
Methylprednisolone n/a n/a 60 mg

n/a, Not applicable.

Calcium Channel Blockers

The calcium ion has several pronociceptive effects to include increased excitatory neurotransmitter release and activation of second messenger systems within the spinal cord and brain. Studies have identified the N- and T-type voltage-gated calcium channels as having the predominant influence over these excitatory effects, which has resulted in their targeting as a site for inhibition.

Currently, the only calcium channel blocker approved for IT use is ziconotide, which is marketed under the trade name Prialt. Ziconotide is derived from the venom of a predatory marine snail and acts by blocking the N-type calcium channel in the dorsal horn of the spinal cord. Given its ability to decrease pronociceptive excitation, ziconotide is indicated for use in a range of neuropathic pain and mixed somatic-neuropathic pain conditions. Considering that ziconotide is synthesized from a highly potent neurotoxin, it is not surprising that adverse reactions include many CNS effects. Common reactions include confusion and dizziness, and more serious reactions include hallucinations, new or worsened depression, and suicidal ideation. Ziconotide is contraindicated in patients with a history of psychosis.

Randomized trials have shown favorable results in treating neuropathic pain states. The 2004 trial by Staats et al. randomized 111 patients with chronic pain due to malignancy or acquired immunodeficiency syndrome (AIDS) to receive either IT ziconotide or placebo. Results showed significantly better pain relief in the ziconotide group (52.9% vs. 17.5%) at 2 weeks. However, adverse effects were also significantly higher in the ziconotide group, with a higher percentage reporting any adverse effect (97.2% vs. 72.5%), and more than 3 times the number of control patients experiencing serious adverse effects (30.6% vs. 10.0%), approximately half of which involved the CNS. It should also be noted that the rate of ziconotide initiation was decreased from 0.4 μg/h to 0.1 μg/h or less after the first 48 patients due to an unacceptable rate of adverse effects.

The 2006 trial by Rauck et al. was performed not only to further detail the analgesic efficacy of ziconotide, but also to determine whether a slower titration schedule would decrease the incidence of adverse effects. The study randomized 220 patients with poorly controlled pain despite medical management or nonziconotide IDD to receive either ziconotide or placebo for 3 weeks. Titration was started at 0.1 μg/h, and the target dose was reached slowly over a 3-week period to a mean dose of 0.29 μg/h. Analgesic efficacy was again superior in the ziconotide group, which demonstrated a greater decrease in VAS pain score (14.7% vs. 7.2%), and a higher percentage of patients reporting complete or “a lot” of satisfaction (28.4% vs. 12.1%). Although total reported adverse effects were higher with administration of ziconotide (92.9% vs. 82.4%), serious adverse effects occurred less frequently than in the previous study (11.6%) and were similar to those reported with placebo (9.3%). This study suggests that a lower starting dose and slower upwards titration may ameliorate some serious adverse effects. A retrospective report in 15 patients who received ziconotide therapy initiated at 0.05 μg/h yielded no serious adverse effects, which provides further evidence that slow initiation and titration are key in providing tolerable analgesia with this medication.

The long-term effects of ziconotide are unclear given that it is a recent addition to the field of IT analgesia. The longest-term retrospective study followed 104 patients with malignant and benign pain enrolled in the Italian Registry of Ziconotide for up to 12 months. The study’s data revealed that only 43% of patients continued therapy with ziconotide for 12 months, but this subset of patients reported an average 36.9% decrease in pain scores and a stability of dose that suggests tolerance is not as prominent a factor as it is with opioids.

Local Anesthetics

Local anesthetics have the longest history of IT use but were not used for treatment of chronic pain until the 1990s. The analgesic effect of local anesthetics is accomplished through blockade of sodium channels, which inhibit nerve transmission by preventing development of an action potential. Local anesthetics are generally used as adjuncts in mixed IT infusions and require administration close to the desired spinal level(s). Potential side effects include dose-dependent local anesthetic effects, such as weakness, numbness, and bowel or bladder dysfunction.

Studies supporting the use of IT local anesthetics have generally focused on combination infusions that included bupivacaine and have shown particular benefit in neuropathic pain conditions. Several prospective studies have also demonstrated an opioid-sparing effect when combination infusions were compared with monotherapy with opioids.

The 1999 Van Dongen et al. study in 20 patients with chronic pain due to terminal cancer compared IT morphine with IT morphine plus bupivacaine. The results showed a similar degree of pain relief in both groups, but those in the combination group required a significantly lower morphine dose. In addition, five patients in the morphine group switched to the combination group due to inadequate analgesia with morphine alone, and the only patient who required oral morphine supplementation was also allocated to morphine monotherapy. Similar results were demonstrated in the 2002 Deer et al. retrospective study that compared analgesia in 109 patients who were transitioned from morphine infusions to morphine-bupivacaine infusions. Pain relief in these patients was significantly improved with the addition of bupivacaine, and secondary measures were also improved to include dose of oral opioids, number of pain clinic visits, and overall patient satisfaction.

Gamma-Aminobutyric Acid Agonists

GABA is the primary inhibitory neurotransmitter in the CNS and exerts its actions at two receptor subtypes, GABA A and GABA B . The GABA A receptor is a ligand-gated chloride channel, whereas the GABA B receptor is a G-protein–linked complex, but activation of either results in neuronal hyperpolarization and decreased excitability.

GABA A agonists include substances from multiple classes to include ethanol, barbiturates, benzodiazepines, zolpidem, and esczopiclone. Among these disparate agonists, only midazolam has been advocated as an effective IT medication for treatment of pain.

Literature supporting the use of IT midazolam is sparse and conflicting. Animal models have shown both benefit and possible neurotoxicity, whereas human data are gleaned mainly from studies using single-shot techniques in the perioperative period. A 1992 trial randomized 28 patients with axial low back pain to receive either epidural steroid or IT midazolam but did not include a placebo group. There was improvement in both the steroid and midazolam groups, but the data are difficult to interpret given that steroid was suspended in 10 mL of normal saline and was also accompanied by IT injection of dextrose, whereas the IT midazolam injection was diluted in dextrose but was accompanied by epidural injection of 10 mL of normal saline. The 2010 Dureja et al. trial randomized 150 patients with chronic pain due to postherpetic neuralgia (PHN) to receive either (1) epidural methylprednisolone, (2) IT midazolam, or (3) both epidural methylprednisolone and IT midazolam. Their results showed that pain relief was improved in both groups that received IT midazolam, but the effects were more pronounced when combined with steroid. Multiple trials have shown favorable results with the addition of midazolam to spinal anesthetics, but its role in sustained infusions via implantable pump remains unclear.

GABA B receptor agonists are less common and include several experimental compounds, the rare amino acid isovaline, and baclofen. Of these, only baclofen is used clinically, and is FDA approved for IT treatment of spasticity. Activation of the GABA B receptor at the spinal level results in decreased release of excitatory neurotransmitters, which in turn yields muscular relaxation. Although its precise mechanism for providing analgesia is unclear, several studies have shown an analgesic effect for baclofen in treatment of complex regional pain syndrome and central pain syndromes.

The 2000 van Hilten et al. trial was designed primarily to assess the effects of IT baclofen on dystonia in patients with CRPS. All seven patients had IT catheters placed and received randomized daily injections of varying doses of baclofen or saline, after which they graded both their dystonia and level of pain. Higher doses of baclofen (50 or 75 μg) were associated with reductions in both dystonia and pain, and this effect persisted at 6-month follow-up. This result is supported by the 2002 Zuniga et al. case series, which showed significant improvement in two patients with pain related to CRPS who did not have features of spasticity or dystonia. The results from a small randomized study that compared the effects of IT baclofen versus saline in nine patients with central pain showed significant improvement in dysesthetic pain with baclofen, but the improvement seemed to coincide with improvements in muscle spasm and dystonia. Similar results were reported in the 1995 case series by Taira et al., who found significant decreases in allodynia and hyperalgesia in 9 of 14 patients with central pain with central pain due to stroke or spinal cord injury.

Baclofen use is well supported for conditions that include an element of spasticity or dystonia, but its use in other pain conditions is less well established. A 2000 case series by Zuniga et al. describes the use of IT baclofen in five patients with chronic pain due to causes that included residual limb pain, postlaminectomy syndrome, and cerebral palsy. Three patients had persistent pain despite receiving IT morphine via an implantable pump, one had pain despite implantation of a spinal cord stimulator, and one had previously undergone both spinal cord stimulation and IDD with morphine. All patients experienced significant improvement with transition to IT baclofen. Four of the patients were maintained on baclofen monotherapy, whereas one required a combination morphine-baclofen infusion. The 2010 randomized trial by Schechtmann et al. compared IT baclofen, clonidine, and saline infusions in patients with neuropathic pain and inadequate analgesia despite use of spinal cord stimulation. Ten patients received daily injections of clonidine (25, 50, or 75 μg), baclofen (25, 50, or 75 μg), or saline, and pain was measured via VAS every 30 minutes until it returned to baseline. Results showed significant improvement with either baclofen or clonidine versus placebo, but the improvement with baclofen increased linearly with increasing dose, whereas relief with clonidine was best at 50 μg. Parameters further improved with the combination of IDD and the patient’s existing spinal cord stimulator.

The risk of granuloma formulation with baclofen is unclear but likely low. Two cases of apparent granuloma formation in patients receiving baclofen monotherapy were reported in 2007, both of whom improved with catheter revisions. Subsequent reevaluation of these cases suggested that precipitation of baclofen was the more likely cause of these catheter-tip masses given the reported precipitation of baclofen noted in in vitro studies.

Table 76.4 summarizes the outcomes of randomized studies regarding the use of IT medications for chronic pain.

Sep 21, 2019 | Posted by in PAIN MEDICINE | Comments Off on Implanted Drug Delivery Systems for Control of Chronic Pain
Premium Wordpress Themes by UFO Themes