Patients
Patients 18 years or older with a diagnosis of PHN and an average Numeric Pain Rating Scale (NPRS) [9] score of 3 to 9 (inclusive) were eligible if at least 3 months had elapsed since vesicle crusting. Patients taking chronic pain medications had to be on a stable dose of those medications for at least 21 days before the day of study patch application and remain on a stable dose throughout the study period. It was prespecified that at least 25% of patients would not be taking concomitant pain medications at entry. Women of childbearing age were required to have a negative pregnancy test and be willing to use an effective method of contraception for 30 days after exposure to study medication.
Exclusion criteria were as follows: use of any topically applied pain medication on the painful area within 21 days before the day of application of the study patch; current use of any investigational drug or class 1 anti-arrhythmic drug; uncontrolled diabetes mellitus or uncontrolled hypertension; significant pain of an etiology other than PHN; painful PHN areas located only on the face, above the scalp hairline, or near mucous membranes; and hypersensitivity to capsaicin, local anesthetics, oxycodone hydrochloride, hydrocodone, or adhesives. As prior use of high-dose opioids could limit the responsiveness to the optional oral opioid rescue analgesics used during the treatment procedure, patients using concomitant opioid medication that were not orally or transdermally administered or exceeded a total dose of 60 mg/day morphine equivalent were excluded.
The study was approved by Institutional Review Boards at all participating sites, and conducted in accordance with the ethical principles of the Declaration of Helsinki, Good Clinical Practice guidelines, and applicable regulatory requirements. Written informed consent was obtained from all participating patients before initiating study-related procedures.
Procedures
A baseline screening period was followed by a treatment day (day 0), a 12-week post-treatment assessment period with clinic visits at weeks 4, 8, and 12. Eligible patients were randomized 2:1 to receive either NGX-4010 (capsaicin 640 μg/cm2, 8%, NeurogesX Inc., San Mateo, CA) or an identically appearing low-concentration capsaicin control patch (capsaicin 3.2 μg/cm2, 0.04%) for 60 minutes according to a randomization scheme prepared by Cardinal Health (Morrisville, NC). The low-concentration capsaicin control patches were used in place of placebo patches to provide effective blinding in the study since topical capsaicin can produce local erythema and a burning sensation. All patients were pre-treated with a topical anesthetic cream (ELA-Max4®, lidocaine 4%; Ferndale Laboratories, Inc., Ferndale, MI) for 60 minutes before the application of the study or control patch(es) which were applied directly to the painful area(s) (up to 1000 cm2). After patch removal, the area was cleansed with a proprietary cleansing gel formulated to remove residual capsaicin. Patients were monitored for 2 hours after patch removal. Local cooling as well as oxycodone hydrochloride oral solution (1 mg/mL) or equivalent could be administered at the onset of treatment-associated discomfort and as needed. Patients could take opioid rescue medication (hydrocodone bitartrate/acetaminophen 5 mg/500 mg) every 8 hours for up to 3 days after patch application for treatment-associated discomfort as needed. Topical pain medications were not permitted during the 12-week study period. Patients were allowed to take acetaminophen up to 2 g/day as needed for aches and pains.
Efficacy Measures and Data Analysis
Efficacy was evaluated with daily Numeric Pain Rating Scale (NPRS) scores throughout the 12-week study period. The NPRS is an 11-point scale (0 to 10) with 0 indicating no pain and 10 indicating the worst possible pain [20]. Patients recorded NPRS scores for "worst pain for the past 24 hours", "average pain for the past 24 hours", and "pain now" in a take home diary beginning on the evening of the Treatment Visit (Day 0) through the evening before the week 12 visit. Patient Global Impression of Change (PGIC; patients reported how they felt compared to baseline on a scale of -3 indicating "very much worse" to +3 indicating "very much improved" with 0 being "no change") and investigator-rated Clinical Global Impression of Change (CGIC) [21] were evaluated at weeks 4, 8 and 12. The modified Brief Pain Inventory (BPI) [22] was collected at screening and at weeks 4, 8 and 12. The Short-Form McGill Pain Questionnaire (SFMPQ) [23] was collected at screening and at weeks 4, 8 and 12.
The primary efficacy endpoint was the percentage change in "average pain for the past 24 hours" NPRS scores from baseline to weeks 2 through 8. To avoid the potential confounding effect of allowed opioid rescue medications during days 0 to 3, week 1 NPRS scores were not included in the primary analysis. Other efficacy measures included: percentage change in NPRS scores from baseline to weeks 2-4 and 2-12; the percentage of patients with a ≥ 30% and ≥ 50% reduction in NPRS score from baseline to weeks 2-4, 2-8 and 2-12; the percentage of patients considered improved (much, or very much) on the PGIC and CGIC at weeks 4, 8 and 12; changes from screening in the BPI questionnaire collected at weeks 4, 8 and 12; and changes from screening in the SFMPQ questionnaire collected at weeks 4, 8 and 12. Weekly changes in NPRS scores were also performed.
Efficacy analyses were based on the intent-to-treat population that consisted of all patients who received any study treatment and had at least 3 days of available NPRS scores during the baseline period. The NGX-4010 group was compared to the control group using a gender-stratified ANCOVA model with baseline pain as the only covariate. The same method was used to analyze the differences between the NGX-4010 groups and the control group in NPRS scores for weeks 2-4 and 2-12. Logistic regression with gender and baseline as covariates tested the difference in the proportion of patients with a ≥30% and a ≥50% mean decrease from baseline in NPRS scores during weeks 2-4, 2-8 and 2-12. For the Short-Form McGill Pain Questionnaire and BPI, a t-test was used to test for differences in change from screening to week 8 between treatment groups. For PGIC and CGIC, Fisher's exact test was used to test for differences between treatment groups in the percentage of patients considered improved (much, or very much). To assess the potential impact of enrolling patients with disease of less than 6 months duration, post hoc analyses were performed on the subgroup of patients with PHN duration of at least 6 months (180 days). Treatment comparisons were performed using gender-stratified ANCOVA to test for a difference between the NGX-4010 and control groups, with baseline pain, pre-anesthetic pain score, and percent change in NPRS score from pre-anesthetic to pre-patch application time point as covariates. The percent change in pain following application of the topical anesthetic and the pain score reported immediately before application of the topical anesthetic were found to be significant covariates in these post hoc analyses and included in the model.
Missing post-treatment NPRS scores were imputed using a modified last-observation-carried-forward approach. If the NPRS score was missing on days 0-8, the baseline score was imputed for that day. If the NPRS score was missing for any day past day 8, then the latest available non-missing score collected before that day was imputed for that missing value. If NPRS scores were missing for all post treatment study days (including day 0), then the baseline score was imputed for all missing scores. No imputation was used for the calculation of weekly scores. For the calculation of NPRS baseline scores, all available screening scores which were not biased by pain medication changes were used. For changes in non-SSRI antidepressant or anticonvulsant medications, pain scores up to 14 days after the medication change were considered biased. For changes in other pain medications, pain scores up to the day of medication change were considered biased. Changes in minor OTC analgesics (acetaminophen, aspirin) were ignored.
It was estimated that to achieve 90% power at the 0.05 significance level, a total of 150 patients, with 100 designated for NGX-4010 and 50 designated for control treatment, were required to detect a difference of 15% in change from baseline in NPRS scores between the NGX-4010 and control group.
Safety Measures and Data Analysis
Safety was assessed by continuous monitoring of adverse events and periodic assessments of clinical laboratory parameters, vital signs, physical examinations, dermal assessments (0- to 7-point severity score) [24], pain experienced during and after the patch application by using NPRS scores on the day of treatment (prior to topical local anesthetic application, 5 minutes prior to patch application, 5 minutes prior to patch removal and 1 hour after patch removal), and rescue medication and concomitant medication usage. Treatment associated erythema, discomfort and pain on the day of treatment were not captured as adverse events but reported as dermal assessment scores or "Pain Now" NPRS scores. Standardized neurosensory examinations (allodynia, light brush, pinprick, warmth and vibration) were performed at screening, week 4, week 8 and 12. Changes in neurosensory assessments from screening to each assessment time point were categorized using prespecified algorithms. On the day of treatment, the proportions of patients reporting each level of dermal response, the "Pain Now" NPRS score and the change in "Pain Now" NPRS score from the pre-local anesthetic time point, vital signs (systolic blood pressure, diastolic blood pressure, heart rate, and respiratory rate), change in vital signs from the pre-local anesthetic time point and the number and proportion of patients who had shorter patch application times (< 90% of intended patch application duration) were summarized. The proportion of patients classified as "same or no increase", " < 33% increase", or " > 33% increase" in maximum pain score from baseline during the first 48 hours were also summarized.
Adverse events were coded using the Medical Dictionary for Regulatory Activities (MedDRA; Version 7.0). Rescue medication use from days 0 to 5, the number of patients completing the intended patch duration, demographics, and baseline clinical characteristics were compared between groups with Fisher exact tests or t-tests, as appropriate. The proportions of patients reporting each level of maximum dermal response on the day of treatment were compared between the active group and the control group using Fisher's exact test. The results of the allodynia assessments at each visit were compared using a t-test while the rest of the neurologic sensory exams were compared using a Cochran-Mantel-Haenszel test for trend.