Improving Postamputation Functioning by Decreasing Phantom Pain With Perioperative Continuous Peripheral Nerve Blocks: A Department of Defense Funded Multicenter Study

When a limb is amputated, pain perceived in the part of the body that no longer exists often develops, called "phantom limb" pain. The exact reason that phantom limb pain occurs is unclear, but when a nerve is cut-as happens with an amputation-changes occur in the brain and spinal cord that are associated with persistent pain. The negative feedback-loop between the injured limb and the brain can be stopped by putting local anesthetic-called a "nerve block"-on the injured nerve, effectively keeping any "bad signals" from reaching the brain. A "continuous peripheral nerve block" (CPNB) is a technique providing pain relief that involves inserting a tiny tube-smaller than a piece of spaghetti-through the skin and next to the target nerve. Local anesthetic is then introduced through the tiny tube, which bathes the nerve in the numbing medicine. This provides a multiple-day block that provides opioid-free pain control with no systemic side effects, and may prevent the destructive feedback loop that results in phantom limb pain following an amputation. We propose a multicenter, randomized, triple-masked (investigators, subjects, statisticians), placebo-controlled, parallel arm, human-subjects clinical trial to determine if a prolonged, high-concentration (dense), perioperative CPNB improves post-amputation physical and emotional functioning while decreasing opioid consumption, primarily by preventing chronic phantom limb pain.

Subjects will receive 2 perineural catheters: a femoral and sciatic. With a high-frequency linear array ultrasound transducer, the target nerves will be identified in a transverse cross-sectional view: the sciatic nerve proximal to the anticipated level of amputation and the femoral nerve at the inguinal crease. For each insertion, normal saline will be injected via the needle to open the perineural space allowing subsequent insertion of a perineural catheter 3-5 cm beyond the needle tip. The needle will be removed over the catheter, and the catheter affixed to the skin and covered with occlusive dressings. Local anesthetic (20 mL, lidocaine 2% with epinephrine 2.5 µg/ml) will be injected via each catheter.

The nerve blocks will be evaluated 20 minutes following local anesthetic injection and the catheter insertions considered successful when subjects have a decreased sensation to cold temperature (alcohol swabs) in the appropriate cutaneous distribution for each target nerve. Catheters may be re-bolused with lidocaine 2% with epinephrine 2.5 µg/ml, at the discretion of the Site Director.

Treatment group assignment (randomization). Subjects will be randomized to one of two groups:

1. Experimental: bupivacaine 0.3% (or ropivacaine 0.5%) for a total of 7 days
2. Control: bupivacaine 0.1% (or ropivacaine 0.2%) for 1 day followed by normal saline for 6 additional days

Randomization will be stratified by institution in randomly chosen block sizes of 2 or 4. Investigational pharmacists at each institution will prepare all study solutions as determined by the randomization lists. Unmasking will not occur until statistical analysis is complete (termed "triple masking").

In addition to the perineural infusions, subjects will receive standard-of-care oral and intravenous postoperative analgesics. The perineural infusions will be initiated within the operating or recovery room. The continuous basal infusion rate for lower extremity catheters will be determined by amputation location: below-knee [femoral 3 mL/h; sciatic 5 mL/h] and above-knee [femoral 5 mL/h; sciatic 3 mL/h]. No patient-controlled bolus dose will be included.

The morning of postoperative day 1, the local anesthetic reservoirs will be replaced with new reservoirs containing study infusate: additional bupivacaine 0.3% for the Experimental group [alternative: ropivacaine 0.5%]; normal saline for the Control Group. Subjects will be discharged home with portable infusion pumps when medically ready, without restriction due to study participation and telephoned daily. If premature dislodgement occurs prior to 72 hours, the subject may opt to have the catheter replaced as soon as can be arranged with the investigators (a minimum of 72 hours will be considered a successful treatment application).

Seven days following catheter insertion, subjects or their caretakers will remove the perineural catheters with instructions given by an investigator via telephone.

Outcome measurements (end points). The primary end point will be mobility at 9 months following surgery (measured with the Locomotor Capabilities Index). The primary analyses will compare the two treatment groups. End points will be evaluated at baseline (postoperative day 0), during the initial infusion of either bupivacaine 0.3% or 0.1% [alternative: ropivacaine 0.5% or 0.2%] (morning of postoperative day 1); during the second infusion of either bupivacaine 0.3% [alternative: ropivacaine 0.5%] or placebo (postoperative day 2), following the completion of the infusion (postoperative day 8), and the follow-up period (postoperative months 1, 3, 6, 9, and 12).

The questionnaires will differentiate among multiple dimensions of limb pain:

Residual limb ("stump") pain: painful sensations localized to the portion of limb still physically present Phantom limb sensations: non-painful sensations referred to the lost body part Phantom limb pain: painful sensations referred to the lost body part

Medical history will include the indication for amputation, history of the surgical limb (e.g. previous surgeries), preoperative limb pain levels (e.g., least, average, worst and current), comorbidities, medications, and amputation level. In addition, we will apply the PTSD Checklist (PCL-C) at baseline.

Hypothesis 1: Mobility will be significantly increased within the 12 months following a surgical amputation with a 7-day high-concentration perioperative CPNB compared with usual and customary analgesia (as measured with the Locomotor Capabilities Index-5). The primary outcome measure of the proposed trial will occur at the 9 month time point.

Hypothesis 2: General physical and emotional disability will be significantly lower within the 12 months following a surgical amputation with a 7-day high-concentration perioperative CPNB compared with usual and customary analgesia (as measured with the World Health Organization Disability Assessment Schedule 2.0).

Hypothesis 3: Depression will be significantly decreased within the 12 months following a surgical amputation with a 7-day high-concentration perioperative CPNB compared with usual and customary analgesia (as measured with the Beck Depression Inventory).

Hypotheses 4 & 5: Phantom and residual limb pain will be significantly decreased within the 12 months following a surgical amputation with a 7-day high-concentration perioperative CPNB compared with usual and customary analgesia (as measured with Brief Pain Inventory questions 1-4). Current/present, worst, least, and average phantom pain will be assessed using a Numeric Rating Scale (NRS) included in the Brief Pain Inventory (short form). These same measures will be included for residual limb pain as well. Additional pain-related data. Frequency and average duration of non-painful phantom sensations, phantom limb pain, and residual limb pain will also be assessed. In addition, other pain locations/severity will be evaluated using the NRS. Lastly, to investigate masking adequacy, subjects will be queried the day following catheter removal on the infusion type they believe they received (active drug vs. placebo).

Hypothesis 6: Opioid consumption will be significantly decreased within the 12 months following a surgical amputation with a 7-day high-concentration perioperative CPNB compared with usual and customary analgesia (measured in oral morphine equivalents).

Supplemental analgesic use will be recorded at all time points.

Data collection. Subject demographic, surgical and CPNB administration data will be uploaded from each enrolling center via the Internet to a secure, password-protected, encrypted central server (RedCap, Department of Outcomes Research, Cleveland Clinic, Cleveland, Ohio). The questionnaires for all subjects-regardless of enrolling center-will be administered by telephone from the University of California San Diego. Staff masked to treatment group assignment will perform all assessments.

Statistical Plan and Data Analysis Primary analyses will be modified intent-to-treat, such that all randomized patients who receive at least some of the study intervention will be included in the analyses. All patients will be analyzed in the group to which they were randomized.

Aim 1 - Mobility (H1). We will assess the treatment effect of continuous peripheral nerve blocks (CPNB) versus standard therapy on the primary outcome of the Locomotor Capabilities Index (LCI) at 9 months using a proportional odds logistic regression model, as long as the proportional odds assumption holds (i.e., P-value > 0.05 assessing proportional odds). In the event that the proportional odds assumption does not hold, we will compare groups using an extension of the Wilcoxon rank-sum test to allow covariate adjustment. The primary analysis will use the total LCI score (range 0-56), while the basic and advanced LCI subscales-comprised of 7 questions each-will be considered as secondary. Additional secondary analyses will compare groups at each of 6 and 12 months on the same outcome. We do not use repeated measures or longitudinal data models because we do not expect the treatment effect to be consistent across 6, 9 and 12 months.

Aim 1 - General physical and emotional disability (H2). We will assess the effect of CPBN on general physical and emotional disability as measured by the World Health Organization Disability Assessment Schedule 2 (WHODAS2) questionnaire across time points 1, 3, 6, 9 and 12 months post-randomization using a linear mixed effects model to account for the within-subject correlation over time (assuming an auto-regressive correlation structure). Baseline WHODAS2 scores will be adjusted for in order to gain precision in the treatment effect estimates. The WHODAS2 overall score across the 36 items and the 6 subscale scores will be calculated for each exam using the complex scoring algorithm (called "item-response-theory" (IRT) based scoring) which takes into account multiple levels of difficulty for each WHODAS 2.0 item. Scores will then be converted to a 0 (no disability) to 100 (full disability) scale for ease of interpretation.

Aim 1 - Depression (H3). We will assess the effect of CPBN on depression at 12 months post randomization as measured by the Beck Depression Inventory (BDI-II) using a multiple linear regression model to adjust for baseline BDI scores and clinical site, as well as to assess the treatment group-by-site interaction.

Aim 2 - Phantom limb pain (H4), Residual limb pain (H5), opioid consumption (H6). We will assess the effect of CPBN on each of phantom limb pain, residual limb pain and opioid consumption over time using linear mixed effects models as specified above for analysis of the WHODAS2 score. In these models we will assess the treatment-group by time interaction over all times (postoperative days 1, 2, 8, months 1, 3, 6, 9, 12) and also the interaction between treatment group and early times (postoperative days 1, 2, 8) versus late times (postoperative months 1, 3, 6, 9, 12).

Opioid consumption will be log-transformed (or other appropriate transformation) to achieve normality in the observed data before modeling and will be analyzed as the amount of opioids being consumed at each time point. Throughout these models we will adjust for baseline pain score or opioid consumption as a covariate to improve precision of the treatment effect estimates.

Interim analyses. We will conduct interim analyses for efficacy and futility at each 25% of planned enrollment using a group sequential design and gamma spending function for alpha (efficacy, parameter -4) and beta (futility, gamma parameter -2).

Sample size considerations. We powered this trial to be able to detect a clinically important difference between groups on the primary outcome of the LCI-5 at 9 months post-randomization with 90% power at the 0.05 significance level. While the minimal clinically important difference (MCID) for the LCI remains undetermined, current literature suggests that the MCID for similar instruments is about 10% of the possible range of the scores, or else, similarly, about half of a standard deviation of the score (please see following paragraph for details on the MCID). For our study we therefore power the study to detect a difference of 5.6 on the 56-point LCI total score, and we assume a SD of 11 for the total score (Larsson, 2009 observed SD of 9 and 11 for two different time points). With the given assumptions we would require 89 subjects per group, or a total of 178 for a 2-sample t-test. While the analysis will utilize a proportional odds logistic regression, with a range of 56 points the outcome variable might be close to normally distributed and the power for the planned test is expected be close to that of a 2-sample t-test on the same data. Incorporating the planned interim analyses at each 25% of total enrollment, the study requires a maximum total of 203 subjects. Expecting a maximum of 7-8% of patients lost-to-follow-up or withdrawn by 9 months, we will enroll a maximum total of 218 patients to achieve the goal of N=203 for analysis.