Analgesic Effects of Intravenous Lidocaine and Morphine on Postamputation Pain: A Randomized Double-blind, Active placebo-controlled, Crossover Trial

The protocol was approved by our institutional human subjects review board. The enrollment period began on October 6, 1997 and lasted until February 28, 2001. Inclusion criteria were: age >18 but < 85 yr; presence of persistent postamputation pain for >6 months after amputation of an extremity or portion thereof. Exclusion criteria included: age <18 or >85 yr; history of allergic reaction to any study drug; cardiac conduction defects (second degree or complete heart block), or myocardial infarction within 3 months of enrollment; severe pulmonary disease; current history of alcohol or substance abuse; presence of seizures, dementia, or encephalopathy; pregnancy or breast feeding; chronic hepatic disease or hepatic failure; hematologic disease associated with leukopenia or thrombocytopenia; and any terminal illness with a life expectancy of <6 months. Baseline assessment included a detailed medical history and physical examination. Stump pain was defined as ongoing or stimulus-evoked pain localized to the region of the stump. Phantom pain was defined as pain experienced in the missing part of the limb subsequent to an amputation.

We used a randomized, double-blind, active-placebo-controlled, crossover design. Informed consent and baseline laboratory data (electrocardiogram) were obtained, and a physical examination was performed. Before the administration of any test drug, all subjects were gradually titrated off any opioids, benzodiazepines, antiepileptic drugs, mexiletine, baclofen, or neuroleptic drugs prescribed for pain such that there was a 2 week period where none of these medications were taken by the patients. However, they were allowed to take acetaminophen or nonsteroidal antiinflammatory agents as needed during this period.

Subjects were admitted for a series of three intravenous infusions of test drugs administered in a randomized, double-blind fashion. Before the start of the test, a standardized set of directions was read to the patient to minimize the placebo effect. An intravenous catheter was inserted in an unaffected limb. The subject maintained a supine position throughout the infusion period, and hemodynamic parameters (heart rate, oxygen saturation, respiratory rate, blood pressure, and electrocardiography) were continuously monitored. Subjects rated their ongoing stump pain and phantom pain, and sedation level on a visual analog scale (VAS) using a computer program developed in the authors’ laboratory. A physician was present throughout the infusion period to assess and treat any adverse effects of the infusion.

Each morning (3 days total), the infusion consisted of an initial bolus dose followed by a continuous infusion. After a 30-min baseline period, one of 3 solutions (1 mg/kg of lidocaine, 0.05 mg/kg of morphine, or 10 mg of diphenhydramine) was administered as an intravenous bolus over 2 min, followed by an infusion (4 mg/kg of lidocaine, 0.2 mg/kg of morphine, or 40 mg of diphenhydramine) over 40 min. The maximum infusion dose was 400 mg of lidocaine and 25 mg of morphine. The rate of infusion was decreased if cardiovascular or respiratory depression, central nervous system excitation, or excessive sedation were observed during the drug administration.

The sequence of these infusions was balanced to minimize order effects. The infusions were administered each morning (3 days total), and the interval between the infusions was at least 24 h to minimize carry-over effects. Blood for drug assay was obtained from the patient at the end of the infusion period to determine if therapeutic lidocaine and morphine levels were achieved.

Subjects were randomized to one of six possible combinations of placebo (P), lidocaine (L), morphine (M): L-M-P; L-P-M; M-L-P; M-P-L; P-M-L; and P-L-M. The subjects were randomized in balanced blocks of 12 so that there would be an equal number of subjects who would receive lidocaine, morphine, or placebo as the first drug treatment.

Drugs were prepared by a pharmacist in a way that allowed for administration of an equal volume for each of the three study medications. All study medications were identical in appearance. During the infusion, the investigator administering the study medication was blinded from the outcome assessment (pain and sedation), and the subject and research coordinators were blinded to the exact timing of study medication administration.

Phantom and stump pain ratings and sedation scores were recorded at 5-min intervals using a computerized 0–100 visual analog scale (VAS). These measures were initiated 30 min before drug infusion, continued throughout the drug infusion period, and continued for 30 min after completion of the infusion. Subjective pain relief scores for both stump and phantom pains were also assessed at the end of each infusion by asking the subject to rate the pain relief on a 0–100% numeric scale. Subjects’ overall satisfaction with the treatment was rated at the end of each infusion on a 0–100% numeric scale.

The sample size was based on pilot data and a standard biostatistical formula 13:

n = (z)²/(Δ)², where Δ is fixed by using the formula:‖μ−μ0 ‖/ς.

Δ was estimated to be 11/22 = 0.5, and the sample size was estimated to be: (1.96 + 0.84)²/(0.5)²≈ 32 patients (α= 0.05, two-tailed; β= 0.20, one-tailed). This sample size allowed us to determine a 20% change in pain from baseline values. Pain and sedation scores were analyzed using two-way repeated measures analysis of variance (ANOVA). Between-group comparisons of pain and sedation scores were then performed using paired t  tests. Pain relief, verbally reported by patients for the three medications, was analyzed using one-way ANOVA. Bonferroni multiple comparisons test was used to conduct post hoc  analysis.

As a separate measure of drug effects, the number needed to treat (NNT) 14was obtained for lidocaine and morphine. In recent reviews on drugs used for neuropathic pain, NNT has been used as a clinically relevant measure to quantify treatment effects. 15,16It is defined as the number of subjects that are required to be treated to obtain one patient with the desired outcome. Analysis was based on the number of subjects in each drug group who obtained at least 30% pain reduction during the infusions. Statistical analysis was performed using Stata 6.0 software (Stata Corporation; College Station, Texas). P < 0.05 was considered statistically significant.

Thirty-two eligible subjects were enrolled in the protocol. Of the subjects who were enrolled, one dropped out from the study because of absence of pain before the initiation of the infusion. Of the remaining 31 subjects, 11 subjects had stump pain alone, 9 had phantom pain alone, and 11 had both stump and phantom pains. Data could not be collected on four infusion days because of technical difficulties with intravenous access (n = 2) and because of the cancellation of the infusion as a result of nausea and vomiting from the previous day's infusion (n = 2). Demographic data obtained at the time of en-rollment in the study are shown in table 1.

Stump pain was commonly described by the patients as a burning, throbbing, or stabbing sensation. The most common descriptors for phantom pain were burning or crushing. On the days of infusion, pain was rated as mild, moderate, and severe by 25%, 66%, and 9% of the subjects, respectively. Three fourths of the subjects reported use of prosthesis. Physical stress (55%), changes in weather (52%), stump spasm (52%), excessive use of prosthesis (42%), stump problems (39%), back pain (30%), and psychological stress (23%), were some of the factors that altered the level of pain in the subjects.

The effects of study medications on stump and phantom pain scores are shown in figure 1. Compared with placebo, morphine significantly reduced both stump and phantom pains (P < 0.01 and P < 0.001, respectively). In contrast, lidocaine significantly reduced stump but not phantom pain (P < 0.01 and P > 0.05, respectively) when compared with placebo. Placebo infusions did not result in significant reduction in phantom or stump pains.

Self-reported stump and phantom pain relief scores and satisfaction scores for the three study medications are shown in table 2. Both lidocaine and morphine infusions provided significantly greater self-reported pain relief when compared with the placebo. There were no differences in self-reported pain relief of stump and phantom pain between the lidocaine and morphine groups. Similarly, satisfaction scores were significantly higher for both lidocaine and morphine when compared with placebo, but were not different between the lidocaine and morphine groups. Significant correlation was observed between pain reduction measures based on VAS and patients’ self-reports of pain relief (P < 0.05 for lidocaine and P < 0.01 for morphine). There were no significant differences between groups with regard to sedation scores (table 3).

The NNT with lidocaine for at least 30% reduction in stump pain was 2.5 (95% confidence interval [CI]: 1.5 to 7.4), while NNT with morphine to obtain similar relief was 2.1 (95% CI: 1.4 to 5.2). For a 30% reduction in phantom pain, the NNT with lidocaine was 3.8 (95% CI: 1.9 to 16.6) and the NNT with morphine was 1.9 (95% CI: 1.3 to 3.7).

Eleven subjects had both stump and phantom pains on the days of infusion. In this subset of subjects, no significant correlation (Pearson correlation coefficient, r = 0.13, P = 0.49) between phantom and stump pain scores was found at baseline; however, a significant correlation (r = 0.62, P < 0.01) was observed after the infusions, as shown in figure 2. Plotting the relationships between change in phantom and stump pains from baseline, at, and 30 min after the end of infusion suggests that lidocaine may have a significant but transient effect in reducing postamputation pain, unlike morphine, which may be associated with a more persistent analgesia (fig. 3). The mean ± SD plasma lidocaine and morphine levels were 2.1 ± 1.5 μg/ml and 36.2 ± 40.7 ng/ml, respectively.

The treatment of phantom and stump pain has been disappointing, in part, because of the uncertain nature of the mechanisms of postamputation pain and the lack of well-controlled clinical studies. We performed a randomized, double-blind, active-placebo-controlled, crossover trial comparing the effect of intravenous morphine and lidocaine on phantom and stump pains. We demonstrated that stump pain was attenuated by both morphine and lidocaine, while phantom pain was attenuated by morphine but not lidocaine. These data suggest that different therapeutic sensitivities of stump and phantom pain to opioids and sodium channel blockers exist, and that the mechanisms of stump and phantom pain may differ.

Two common types of postamputation pain states have been recognized: phantom pain (reported in the missing limb) and stump pain (reported in the residual part of the extremity). 17,18Although the mechanisms of phantom and stump pains are uncertain, they are considered to be interconnected. Studies have reported a high incidence of concomitant occurrence of both stump and phantom pains. 2,19 

The mechanism of stump pain appears to be primarily peripheral and may include nerve injury and neuroma formation. Animal studies have demonstrated that peripheral nerve injury may lead to spontaneous ectopic discharges from injured axons in the nerve, neuromas, or dorsal root ganglia. 20–22In addition, more recent evidence suggests that there are several tetrodotoxin-resistant (TTX-R) sodium channels (subtypes; PN3/SNS and NaN/SNS2) that are expressed in injured peripheral neurons and may be involved in the peripheral mechanism of chronic neurogenic pain. 6,7Sodium channel blockers, such as lidocaine, may attenuate the ectopic activity from these neuromas and hence diminish stump pain. 12Although the mechanism of ectopic discharge following peripheral nerve injury is currently believed to primarily involve abnormal activation of sodium channels, other mechanisms (e.g. , increased sensitivity of α-adrenergic receptors) may also exist. 23 

The mechanism of phantom pain is considered to be primarily central, although some studies suggest an important role for input from the periphery in maintaining the altered central state. 24The central mechanism of phantom pain is most likely the cortical reorganization that occurs as a result of amputation, reflecting the plasticity of the somatosensory cortex. 24,25Although it is not clear whether cortical reorganization is the result or the cause of phantom pain, the topographic representation of the lost extremity may be taken over by sensory input from other areas of the body, resulting in perceptual remapping. 8,26The extent of cortical reorganization may correlate with the intensity of phantom limb pain. 25Although the mechanisms of cortical reorganization are not clear, long-term potentiation of synapses mediated by N -methyl-D-aspartate (NMDA) receptors or reduction of inhibitory neurotransmitters (e.g. , GABA) may be implicated. 8Another possible central mechanism of phantom pain is the NMDA receptor-mediated sensitization of dorsal horn neurons in the spinal cord which results in spontaneous neuronal activity, increased response to afferent input and expansion of peripheral receptive fields. 4Peripheral input from stump pain (neuroma and nerve injury) may potentially contribute to the maintenance of centrally-mediated pain. 4,9Hence, despite the fact that stump and phantom pains may have different underlying mechanisms, the two entities may be interdependent. It is possible that continued peripheral activity of stump pain may maintain the central hyperexcitable state of phantom pain.

The therapeutic modalities currently being used clinically for the treatment of postamputation pain syndromes are far from being satisfactory, with success rates rarely exceeding placebo response rates. 26,27Many therapies have been used empirically, and there are methodological concerns (e.g. , small sample sizes, heterogeneous populations, lack of blinding, lack of controls, and short follow-up periods) with published studies that currently make it difficult to determine whether opioids or local anesthetics would be effective for neuropathic pain states such as postamputation pain.

Neuropathic pain has been traditionally viewed as being resistant to the analgesic effects of opioids, although the route of opioid delivery may affect efficacy. 28Although the exact mechanisms for such apparent ineffectiveness have not been clearly delineated, some cellular phenomena involved in neuropathic pain are implicated in the development of tolerance to morphine. 29Peripheral axotomy has been shown to result in a reorganization and temporary down-regulation of opioid receptors in the dorsal horn of the spinal cord, which may lead to a decreased effectiveness of opioids in relieving postamputation pain. 30–32Certainly, there are various reasons (e.g. , use of multiple definitions of neuropathic pain, different animal and human models of neuropathic pain, methodological issues in available studies, different methods of pain assessment, and inconsistent duration of follow-up) that have contributed to this controversy. 20 

The use and efficacy of opioids in treating neuropathic pain syndromes, including postamputation pain, remain controversial, although it appears that opioids may be beneficial in treating some types of neuropathic pain. 33Our data indicate that opioids would be effective in diminishing both stump and phantom pain. Despite the fact that some animal and clinical studies suggest that opioids may have a poor analgesic effect on neuropathic pain, 30,34there are other data that corroborate our findings and show that neuropathic pain can be successfully treated with opioid therapy. 32,35,36A randomized, double-blind, active-placebo-controlled, crossover trial demonstrated that intravenous fentanyl was effective in relieving neuropathic pain and that this was not the result of any sedative or mood-altering effect of the fentanyl. 11Other randomized, double-blind, controlled trials have also suggested that intravenous opioids may diminish neuropathic pain. 37,38In addition, oral administration of opioids has been shown to be effective in diminishing neuropathic pain (e.g. , postherpetic neuralgia). 10 

Although our study was not designed to distinguish between central and peripheral mechanisms, our observations that lidocaine may be more effective in relieving stump than phantom pain suggests that phantom pain may have a predominant central mechanism. If there were a significant peripheral contribution into maintenance of phantom pain, we would have expected that there would be a reduction in both stump and phantom pains with lidocaine infusion; however, this was not the case. Presumably, the sodium channel binding property of lidocaine is an important mechanism of action in decreasing ectopic neural activity peripherally and hence reducing stump pain. 12Intravenous lidocaine infusion has been used to treat various types of neuropathic pain, and prolonged relief of neuropathic pain after systemic local anesthetic administration has been reported. 35,39,40Despite the fact that our data do not suggest and that of others do not demonstrate a benefit of local anesthetics in relieving central pain, 38,41there are some data to suggest that local anesthetics may reduce spontaneous and evoked spinal (central) neuronal activity. 42–44 

Under physiologic circumstances, activity in central pain signaling neurons is influenced by input from peripheral nociceptors. The lack of correlation at baseline between stump pain, considered to be predominantly peripherally mediated, and phantom pain in the subset of patients who had both types of pain suggests alterations in central pain modulatory mechanisms after amputations. A reduction in pain by lidocaine and morphine is associated with a change in the relationship between stump and phantom pains so that the two pain states are positively correlated. In keeping with the pharmacokinetics of lidocaine and morphine, the change in the relation between stump and phantom pains is short-lived with lidocaine, but longer lasting with morphine. These observations need to be verified with additional studies with larger sample sizes.

There are possible limitations of our study. With the crossover design of our study, patients may have had carry-over effects of study medications from one infusion to the next. Although carry-over effects are possible, baseline scores for pain and sedation did not differ significantly between the 3 days of infusion, and with the relatively short duration of action of study medications, it is unlikely that the carry-over effect was significant. Use of an active placebo has been recommended to prevent unmasking of the double-blind design, and diphenhydramine is a good active placebo as it has no analgesic properties and mimics the side effects of the other test drugs. 32,45Despite the fact that intravenous lidocaine therapy has been used to treat neuropathic pain (such as postamputation pain), there are a number of controversial issues regarding the systemic use of lidocaine for neuropathic pain, including the effective dose range, predictive value, identification of the specific symptoms relieved, and dosage, duration, and end-point of a positive lidocaine test. 11Resolution of these problems is important, since successful treatment of neuropathic pain with intravenous lidocaine infusion may predict subsequent efficacy of oral congeners such as mexiletine. 12,46The results of our intravenous infusions do not allow for conclusions regarding long-term opioid or local anesthetic therapy for the treatment of postamputation pain; however, we are currently undertaking such a study.

In conclusion, this randomized, double blind, active-placebo-controlled, crossover trial demonstrates that stump pain was diminished both by morphine and lidocaine, while phantom pain was diminished only by morphine. Our observations suggest that the mechanisms and pharmacological sensitivity of phantom and stump pains differ. Stump pain may be predominantly peripherally mediated via  a mechanism involving sodium channels, while phantom pain may involve both peripheral and central mechanisms. Despite the observed efficacy, the drugs tested did not eliminate pain completely, suggesting that these patients may require multimodal therapy. Future analgesic studies in this area might include comparing other treatments (e.g. , neuraxial opioids, anticonvulsant, and antidepressants) to the currently tested drugs.

The authors thank The Perioperative Clinical Investigation Unit (Department of Anesthesiology and Critical Care Medicine, The Johns Hopkins University, Baltimore, Maryland) for their assistance in statistical analysis.