Randomized Double-blind Study Comparing the Efficacy of Gabapentin With Amitriptyline on Diabetic Peripheral Neuropathy Pain FREE

DIABETIC PERIPHERAL neuropathy (DPN) is one of the most common symptomatic, long-term complications in patients with both type 1 and type 2 diabetes mellitus.¹ At initial diagnosis, 7.5% of patients will already experience DPN pain, and approximately 45% will be afflicted with this complication after 25 years.² At this time, however, the exact cause of DPN is not well understood.^3- 4

Two main types of diabetic neuropathy involve the autonomic or somatic nervous systems.^3,5 The most common type of DPN is somatic or sensorimotor neuropathy with peripheral symptoms of burning, shooting, tingling, and allodynia.³ Typically, the sensorimotor neuropathy presents in a distal symmetric pattern, with a glove-and-stocking distribution.

Although DPN pain is prevalent among patients with diabetes, current treatment options, including antidepressants, anticonvulsants, antiarrhythmics, and topical capsaicin, are limited by their variable efficacy and adverse effects.^6- 12 Amitriptyline hydrochloride is effective in approximately 60% to 75% of patients treated for DPN, which is consistently greater than that reported with other agents.^6- 7 Of concern with all available drug therapies are the extensive adverse effects, particularly anticholinergic effects of the tricyclic antidepressants.

In 1993, gabapentin, a γ-aminobutyric acid analog, received Food and Drug Administration approval for adjunctive treatment of partial seizures, with and without secondary generalization in adults with epilepsy.¹³ Despite extensive studies,^14- 18 gabapentin's mechanism of action is unknown. Yet, even without this mechanism of action fully elucidated, there has been an increasing number of case reports of gabapentin's use in neuropathic pain syndromes such as reflex sympathetic dystrophy, postherpetic neuralgia, migraine, trigeminal neuralgia, erythromelalgia, Guillain-Barré syndrome, or other intractable pain states in dosages ranging from 900 to 2400 mg/d.^18- 31 Most recently, a placebo-controlled clinical trial³² in patients with DPN pain has established the efficacy of gabapentin to provide pain relief.

Gabapentin, with its low adverse effect and drug interaction profile, may offer an effective treatment for DPN pain. The primary purpose of this study was to determine the comparative efficacy of gabapentin with amitriptyline, the standard therapy, in the treatment of DPN pain.

From March 1997 to December 1997, patients who were veterans were referred by primary care providers, neurologists, diabetologists, and anesthesiologists at the Veterans Affairs San Diego Healthcare System (VASDHS). Patients were included if they were 18 years or older, had diabetes mellitus with stable glycemic control defined as a hemoglobin A_1c level between 0.043 (4.3%) and 0.079 (7.9%) within 3 months, experienced chronic daily pain for more than 3 months, during which both the quality and location were consistent with DPN pain, as diagnosed by a neurologist (D.F.M. and G.A.S.), and had an estimated creatinine clearance of 0.50 mL/s (30 mL/min) (Table 1 and Table 2).³³

Patients were excluded from the study if they had non-DPN pain more severe than DPN pain; allergy or adverse reaction to gabapentin or amitriptyline; severe depression by diagnosis or as assessed with the Beck Inventory³⁴; were pregnant; were receiving treatment for seizures; had cardiovascular symptoms of postural hypotension, symptomatic coronary artery or peripheral vascular disease; or a creatinine clearance of less than 0.50 mL/s (<30 mL/min). Patients who had received prior treatment with gabapentin or amitriptyline were not excluded, regardless of whether treatment was deemed a success or failure. Patients were excluded, however, if their previous dosage exceeded the study's maximum dosage of either drug.

Use of any medications for DPN pain that patients were taking before the study was discontinued for 2 weeks before entering the study and throughout the study period. All other regular use of analgesics was discontinued; however, patients were allowed up to 4 doses per day of acetaminophen, 325 mg, for severe pain or pain other than DPN. The study was approved by both the University of California at San Diego and VASDHS human subjects committees, and all patients gave informed, written consent.

The study consisted of two 6-week treatment periods with a 1-week washout period between treatment arms. The neurologist (D.F.M. and G.A.S.) and clinical pharmacist (C.M.M., S.G.L., and C.P.S.) screened referred patients for study inclusion, and baseline demographics were obtained. Patients receiving treatment for DPN pain began a 2-week washout period before randomization. Patients were given a daily pain diary and were randomized by the VASDHS clinical research pharmacist, the only unblinded investigator for the study, to receive either gabapentin or amitriptyline in a double-blind design per protocol; all other investigators and patients remained blinded until study termination. At the end of each treatment period, patients were seen within 24 hours for a neurologic examination and to administer a global pain assessment rating. The clinical pharmacist (C.M.M., S.G.L., and C.P.S.) conducted pill counts to assess study medication compliance at the end of each treatment period. Following a 1-week washout period, patients were crossed over to the alternative drug therapy. Although 1 week was sufficient for the complete elimination of either study drug, efficacy evaluation was based on the patient's pain ratings recorded during the final week of each treatment, thus allowing an additional 5 weeks for dissipation of drug effects.

During each treatment period, drug dosage was titrated for 2 days to minimize adverse effects, after which the dosage was adjusted based on the patient's clinical response and adverse effects (Table 3 and Table 4). Daily doses of gabapentin ranged from 900 to 1800 mg, and amitriptyline hydrochloride, from 25 to 75 mg. To preserve the double-blind study design, placebo was used to maintain a 3-times-per-day dosage regimen for each study drug (Table 3). Patients received 2 bottles of study medication and were instructed to take the 9 AM and 3 PM doses from the first bottle and the 9 PM dose from the second bottle, which were labeled accordingly.

The clinical pharmacist called patients on days 2, 4, and 6 of the first week and days 1 and 4 of the fourth week of each treatment arm to assess pain control and adverse effects. Patients were interviewed for frequency and severity of the 20 most common adverse effects of the 2 study drugs and any other adverse effects experienced. The purpose of the call was to adjust drug dosage to the maximum tolerated for pain control without intolerable adverse effects. Patients continued to take the maximum tolerated dosage for the remainder of the treatment arm.

The Pain Scale Rating System and Global Rating Scale were used to measure pain relief. Patients rated their pain by completing a daily pain diary in which they chose from a scale of 13 words describing pain intensity, ranging from none to extremely intense.^35- 37 These verbal descriptors were quantified, based on a ratio-scale technique described by Gracely et al,³⁵ which has been shown to be reliable and consistent in previous studies^35- 39 involving human clinical and experimental pain; it distinguished active treatment from control. The pain diary was collected at the end of each treatment period. A neurologist evaluated patients at baseline and at the end of each treatment arm. In this evaluation, the neurologist asked patients to make a global rating of their overall pain relief (complete, a lot, moderate, slight, none, or pain worse) at the end of each treatment arm compared with their baseline pain before entering the study.

For statistical purposes, the verbal descriptors in the pain diary were converted to numerical equivalents using the Pain Scale Rating System as described by Gracely et al.³⁶ The mean pain scores in each final treatment week were compared within patients by paired, 2-tailed t test. Period and sequence effects were examined for scores in the final treatment week by a t test. Global Rating Scale scores were analyzed with a paired, 2-tailed Wilcoxon signed rank test. A 1-sample sign test was used to analyze the frequency and severity of adverse effects between amitriptyline and gabapentin, and the Wilcoxon signed rank test was used to analyze the frequency and severity of adverse effects with time for each medication.

Study drugs were prepared at a Food and Drug Administration–registered repackaging facility. Since this was a double-blind design, all capsules were identical in taste, color, size, and shape.

Twenty-eight patients were eligible for study; 3 withdrew before randomization: 2 because of no neuropathic pain after washout and 1 for unexpected surgery requiring analgesics. Twenty-five patients were enrolled in the study. Because of adverse effects, protocol violation, or voluntary withdrawal, 4 patients withdrew from the study (2 each from the gabapentin and amitriptyline treatment arms). In addition, 3 patients were crossed over early per protocol because of intolerable adverse effects (2) or intolerable pain (1); 1 of whom also dropped out of the study. Thus, whereas 21 patients underwent both treatment arms of the study, 19 completed 6 weeks of treatment with both study drugs (Figure 1). For statistical analyses of mean pain diary scores, all patients were included to prevent bias of study results by omitting those with early crossover due to intolerable pain or adverse effects. Data analysis excluding early crossover patients did not produce statistically significant differences (P=.13).

Of the 25 enrolled patients, 12 were initially randomized to the gabapentin treatment arm, 11 of whom crossed over to amitriptyline. During the gabapentin treatment arm, 1 patient experienced adverse events (diarrhea and ankle edema) and voluntarily withdrew from the study. In addition, 2 patients taking gabapentin were crossed over to amitriptyline early (week 4) because of intolerable adverse effects (sedation and dizziness) or intolerable pain, respectively; 1 of whom eventually withdrew from the study because of adverse events.

Thirteen of the 25 patients were initially randomized to the amitriptyline treatment arm, 11 of whom crossed over to gabapentin. During the amitriptyline treatment arm, 1 patient experienced adverse events (bilateral ankle edema and dizziness), which resulted in discontinuation of drug use, and 1 voluntarily withdrew from the study after crossover because of increasing pain from arthritis that exceeded DPN pain. One patient taking amitriptyline crossed over to gabapentin early (week 4) because of intolerable adverse effects (sedation and constipation).

The clinical characteristics of the patients enrolled in the study were consistent with the general diabetic population, with the exception of a higher male-to-female ratio (Table 1). Study patients experienced typical features of DPN pain (Table 2), for which they had received prior treatment, including amitriptyline (14), nonsteroidal anti-inflammatory drugs (3), nortriptyline hydrochloride (1), gabapentin (1), carbamazepine (1), opioid analgesics (1), and capsaicin (1). Only 1 patient had prior treatment with 2 medications and 1 with 4 medications. At the time of study enrollment, only 1 patient was receiving gabapentin and 9 receiving amitriptyline required washout before study entry.

After excluding data of patients who did not fully complete both treatment arms, the weekly mean pain scores of the remaining 19 patients are shown in Figure 2. In patients treated with gabapentin followed by amitriptyline, pain scores steadily declined during the first 2 weeks of dosage titration, followed by a plateau effect of pain relief. The pain returned during the 1-week washout period and then declined during the second treatment arm. A similar response was seen in patients treated with amitriptyline followed by gabapentin.

Comparing the baseline pain scores in 21 patients who underwent both treatment arms to the end-of-treatment pain scores, there was a statistically significant difference in pain score reduction in patients treated with both gabapentin (P<.001) and amitriptyline (P<.001) by 2-tailed, paired Student t test. However, the mean difference between drugs in pain intensity scores during the final week of treatment favored amitriptyline by 0.091 unit (95% confidence interval, −0.074 to 0.256), in which 0.35 units was the difference between moderate and mild pain.³⁵ Thus, there was no statistically significant difference in pain intensity scores between gabapentin and amitriptyline by the end of treatment (P=.26).

Based on the global description of pain relief for patients who underwent each treatment arm (Table 5), moderate or greater pain relief was experienced in 11 (52%) of 21 patients during gabapentin treatment and 14 (67%) of 21 patients during amitriptyline treatment. There was no statistically significant difference in pain relief between groups (P>.1 by 2-tailed, paired Wilcoxon signed rank test), significant period or carryover effects were not detected, and there was no correlation between global score and maximum dosage achieved.

Figure 3 shows the mean changes in pain intensity during the first 6 weeks of each study drug for the 19 patients who fully completed their first treatment arm. Mean (±SE) pain diary scores decreased by 0.31±0.064 unit when patients were treated with gabapentin and 0.44±0.089 unit with amitriptyline. Although the mean pain score change from baseline favored amitriptyline, there was no statistically significant difference by paired t test (P=.3).

Symptoms possibly related to study drugs caused 3 patients to withdraw from the study (1 receiving gabapentin and 2 receiving amitriptyline) and 2 patients (1 each receiving gabapentin and amitriptyline) to cross over early to the alternative treatment arm. Seventeen patients receiving amitriptyline and 18 patients receiving gabapentin experienced adverse effects (Table 6). With the exception of weight gain with amitriptyline, there was no statistically significant difference in occurrence of adverse effects between drugs (P>.05). Prevalent adverse effects included sedation, dry mouth, dizziness, postural hypotension, weight gain, ataxia, and lethargy.

Comparing the frequency and severity of adverse effects at study weeks 1 and 4 showed that, with amitriptyline, dry mouth worsened with time (P<.005) and pruritus was worse than with gabapentin at week 1 (P<.03) but was not statistically significant from gabapentin at week 4. The frequency and severity of dizziness with gabapentin diminished with time (P=.02).

After dosage titration based on individual response, the mean dosages of gabapentin and amitriptyline hydrochloride were 1565 and 59 mg, respectively. Of patients treated with gabapentin, 65% reached 1800 mg/d, 26% reached 1200 mg/d, and 9% were maintained with 900 mg/d. With amitriptyline, 54%, 29%, and 17% of patients reached 75, 50, and 25 mg/d, respectively. Medication compliance, defined as the total percentage of doses taken, was 94.8% with gabapentin and 96.4% with amitriptyline.

This study was designed to evaluate the efficacy and safety of gabapentin compared with amitriptyline, the current standard of therapy for DPN pain. Study results indicate that, although both gabapentin and amitriptyline provide statistically significant pain relief by the end of treatment, there was no statistically significant difference between the 2 drugs, as measured by daily pain diary scores and global ratings of pain relief. As expected, amitriptyline provided moderate or greater pain relief in 67% of patients, which is consistent with previously reported efficacy rates.^6- 7 Gabapentin provided moderate or greater pain relief in 52% of patients and reduced pain diary scores by 0.31 unit, indicating a less-than-moderate impact on pain relief compared with a 0.44-unit reduction with amitriptyline (Figure 3). This is similar to the recently reported efficacy of 60% for gabapentin compared with 33% for placebo.³¹

The limited number of patients enrolled in our study introduces the probability of a type II (β) error.⁴⁰ Sample size could not be determined since there were no published studies describing gabapentin's efficacy in DPN pain at study initiation. Post hoc analysis revealed that a sample size of approximately 260 patients per paired crossover study would be necessary to provide 80% power to detect a mean difference between treatments of approximately one third of the difference between mild and moderate pain at a .05 significance level. Larger controlled trials are needed to determine if there is any detectable difference in efficacy between amitriptyline and gabapentin.

Accepting patients with well-controlled diabetes might have limited the study size, yet hyperglycemia can decrease the pain threshold, thereby interfering with study results. Our patients maintained a stable hemoglobin A_1c level throughout the study, reducing the potential of hyperglycemia interference. Allowing patients who had received prior treatment with either gabapentin or amitriptyline into the study may have potentially introduced bias, since those who were taking one of these drugs and in whom DPN pain had been successfully controlled would not likely volunteer to discontinue use of the drug, whereas those who entered the study were more likely to experience treatment failure.

Dosage ranges in this study were based on current literature and practice standards at the VASDHS. Review of amitriptyline for neuropathic pain within our institution indicated that 25 to 75 mg/d was the most prevalent dosage range. Gabapentin dosage remained within the manufacturer's guidelines for seizure treatment. Higher dosage ranges may have achieved greater efficacy and should be evaluated in future studies. To measure the full potential of each drug, it might be useful to titrate the dosage upward, either to complete pain relief or to the maximum tolerated dosage.

All study patients were initiated with a 2-day dosage titration to diminish the occurrence of benign effects. Although gabapentin is often cited as having a relatively safe adverse effect profile, both gabapentin and amitriptyline had a similar rate of adverse effects. The frequency and severity of reported adverse effects did not appear to subside with time, with the exception of worsening dry mouth with amitriptyline and resolving dizziness with gabapentin. Similarly, dizziness and somnolence were predominant adverse events reported by Backonja et al³²; however, this was attributed to a forced 4-week dosage titration to 3600 mg/d. This information may be useful for patients when initiating therapy with either agent, since a slower dosage titration may decrease the frequency and severity of adverse effects.

Both gabapentin and amitriptyline in the dosages used in this study appear to provide pain relief in DPN pain; however, mean pain diary and global pain relief score data indicate no statistical difference between them. With the exception of weight gain, dry mouth, and dizziness, the frequency and severity of adverse effects were similar with each drug.

Although gabapentin provides pain relief in patients with DPN pain, it should be reserved as an alternative to patients in whom a less costly agent fails, such as amitriptyline, or for whom tricyclic antidepressants are contraindicated. Our results suggest that larger controlled trials are necessary to further define gabapentin's place in the treatment of DPN pain.

Accepted for publication January 25, 1999.

We acknowledge Stephen D. Funk, PharmD, VASDHS research clinical pharmacist, who maintained the integrity of the double-blind design as the unblinded investigator; Veterans Affairs Clinical Research Pharmacy Coordinating Center at Albuquerque, NM, Mike R. Sather, RPh, MS, Frank Lueddeke, and Roy W. Fetter, who aided in the preparation of study drug; Christopher S. Morello, PhD, University of California at San Diego, who performed statistical analysis of the data; Patricia Hlavin, MD, independent physician reviewer of adverse events for the study duration; and Philip O. Anderson, PharmD, manuscript reviewer.

Corresponding author: Candis M. Morello, PharmD, Veterans Affairs San Diego Healthcare System, Pharmacy Service (119), 3350 La Jolla Village Dr, San Diego, CA 92161.