Ultrasound-Guided Percutaneous Cryoneurolysis to Treat Postoperative Pain After Mastectomy

All participants will continue to receive standard and customary postoperative analgesics

of their local treatment center, so there is no risk of participants receiving a lower

degree of analgesia than if they otherwise did not enroll in the study. The

cryoneurolysis procedure will be done in addition to standard local institutional

standard analgesic treatments.

The investigators propose a pragmatic, multicenter, randomized, triple-masked

(investigators, participants, statisticians), sham/placebo-controlled, parallel-arm,

human-subjects clinical trial to determine if cryoneurolysis is an effective non-opioid

treatment for pain following mastectomy. Participants will be individuals undergoing

unilateral or bilateral mastectomy, recruited at 6 centers:

- Walter Reed National Military Medical Center, Bethesda, Maryland

- U.C. San Diego, San Diego, California

- University of Florida, Gainesville, Florida

- Cedars-Sinai, Los Angeles, California

- Cleveland Clinic, Cleveland, Ohio

For women of childbearing age with the possibility of pregnancy, a sample of urine will

be collected before any study interventions to confirm a non-pregnant state. Participants

will have a peripheral intravenous (IV) catheter inserted, standard noninvasive monitors

applied (blood pressure cuff, pulse oximeter, 5-lead ECG), and oxygen administered via a

facemask. Midazolam and fentanyl (IV) will be titrated for patient comfort as needed

throughout the procedure, while ensuring that patients remain responsive to verbal cues.

Peripheral nerve block. Not all enrolling centers provide peripheral blocks as part of

their standard-of-care. Due to the pragmatic design of this trial, study participation

will not alter an institution's current practice. If the local practice includes a

postoperative continuous peripheral nerve block, a perineural catheter will be inserted

2-5 cm beyond the needle tip, the needle withdrawn over the catheter, the catheter

affixed with an occlusive sterile dressing.

Participants will be allocated to one of two treatments:

1. cryoneurolysis

2. sham cryoneurolysis (placebo control)

Randomization will be stratified by enrolling institution, surgical site (unilateral vs.

bilateral), and axillary involvement (none/biopsy vs. dissection) in a 1:1 ratio, and in

randomly chosen block sizes. Randomization lists will be created using Statistical

Analysis Software computer-generated tables by the informatics division of the Department

of Outcomes Research (Cleveland Clinic, Cleveland, OH). Treatment group assignment will

be conveyed to the enrolling sites via the same secure web-based system (REDCap) used to

collect and collate all post-intervention endpoints.

There are multiple types of cryoneurolysis machines cleared by the US FDA, all of which

work on the same principle of a gas being passed through a small annulus, resulting in a

dramatic pressure drop and accompanying temperature drop due to the Joule-Thomson effect.

This study will utilize two different consoles [second option added May 8, 2023]: the

Epimed International (Farmers Branch, TX) and Varian Medical Systems (Palo Alto, CA)

machines. Which machine is used is determined simply by the machine that each enrolling

center has at its disposal.

Varian: This machine uses argon for the freeze cycle and helium to help decrease the thaw

period duration. Sham probes are not available for this type of machine, and therefore an

active probe will be used for all participants. For participants randomized to active

treatment, the probe will be connected to the machine as usual, the argon (and helium)

passed through the probe and then back into the machine, and finally vented out from the

console. For participants randomized to sham treatment, the probe will be connected to an

inactive connector on the back of the machine. For these participants the gas will simply

be ejected directly from the console without ever having passed through the probe. The

investigator administering the study intervention will access the treatment group

assignment using the secure web-based system and attach the probe (active) or leave the

tubing end close to the machine (sham). Therefore, all investigators, participants, and

clinical staff will be masked to treatment group assignment, with the only exception

being the unmasked individual who performs the procedure (and will not have subsequent

contact with the participant).

Epimed: This machine uses nitrous oxide for the freeze cycle with a passive thaw (no gas

flow). Cryoneurolysis probes are available that either (1) pass nitrous oxide to the

distal end inducing freezing temperatures; or (2) vent the nitrous oxide at the proximal

end of the probe so that no gas reaches the distal end, resulting in no temperature

change. The latter is a sham procedure since without the temperature change, no ice ball

forms and therefore the target nerve is not affected. Importantly, these probes are

indistinguishable in appearance and will be differentiated only by an identifying stamp

on the underside of the connector which is not visible during use.

The investigator administering the study intervention will access the treatment group

assignment using the secure web-based system and attach the appropriate probe to the

cryoneurolysis device. Therefore, all investigators, participants, and clinical staff

will be masked to treatment group assignment, with the only exception being the unmasked

individual who performs the procedure (and will not have subsequent contact with the

participant). For both types of machines, it is impossible to mask the individual

performing the cryoneurolysis procedure because the ice ball forming at the distal end of

the probe-with active treatment-is clearly visible by ultrasound; and the lack of an ice

ball for placebo participants is equally clear. It is essential to continuously visualize

the probe and target nerve throughout the freeze/thaw cycle(s) to ensure the entire nerve

diameter is adequately treated and remains relatively motionless. This cannot be achieved

if the ultrasound is turned off during gas administration in an attempt to mask the

provider.

Study intervention. The 2nd-6th thoracic intercostal nerves will be treated on the

ipsilateral surgical side (bilaterally for bilateral surgical procedures). Using a

curved-array transducer, the intercostal nerve will be visualized using ultrasound just

inferior to each treated rib immediately distal to the costotransverse joint. For

participants without a paravertebral block, a skin wheal of lidocaine 1% will be raised

immediately inferior to the transducer to anesthetize the skin. An IV-like hollow-bore

introducer may be inserted through the skin and guided to the target nerve--this is based

on operator personal preference. For participants without anesthetized intercostal nerves

from a peripheral nerve block, lidocaine 1.5-2.0% 2 mL may be injected through the

introducer prior to the cryoneurolysis probe introduction to provide anesthesia during

the cryoneurolysis cycles, although this is not universally applied (they cryoneurolysis

treatment itself induces anesthesia).

A portable cryoneurolysis device (Varian CryoCare or Epimed PainBlocker) will be used.

The probe will be inserted adjacent to the intercostal nerve.

Varian: The cryoneurolysis device will be triggered using 1 cycle of 5.5-minute [all but

4 participants: 064-067] or 3 minutes [only 4 participants 064-067; change based on new

laboratory data, then changed back based on inadequate freeze] argon activation (2000 psi

and 100% power) followed by a 30-second helium defrost.

Epimed: The cryoneurolysis device will be triggered using 2 cycles of 2-minute gas

activation (active or sham) separated by a 1-minute defrost.

The introducer and probe will be withdrawn and this process repeated for each additional

intercostal nerve to be treated. For bilateral mastectomies, the study intervention will

be repeated on the contralateral side with the same probe.

Intraoperative course. Due to the pragmatic nature of this trial, the investigators aim

to change each center's standard practice as little as possible and rather investigate

the results of adding the intervention to current practice. The investigators will

therefore record intraoperative factors such as type of general anesthetic, axillary

dissection, opioid administration, and local anesthetic supplementation; however, the

investigators will not require changes to current standard practice.

Postoperative course. Standard local supplemental analgesics will be used due to the

pragmatic design of this trial. For analysis purposes, all opioids will be converted to

oral oxycodone equivalents. Following a cryoneurolysis treatment, no action is required

by patients regarding this intervention. For example, in contrast to epidural infusions,

there is no infusion pump to manage or anesthetic fluid to replenish. At enrolling

centers that provide a postoperative continuous peripheral nerve block, the ropivacaine

or bupivacaine infusions will be administered per local protocol with the catheters

removed on postoperative day 1 or 2 prior to hospital discharge.

Following study completion, the results will be mailed electronically or by the United

States Postal Service to all enrolled participants in written form using non-technical

(e.g., "layperson") language.

Outcome Measurements (End Points). The investigators have selected outcome measures that

have established reliability and validity, with minimal inter-rater discordance, and are

recommended for pain-related clinical trials by the World Health Organization and the

Initiative on Methods, Measurement, and Pain Assessment in Clinical Trials (IMMPACT)

consensus statement (Table 1, below).5 End points will be evaluated at baseline (prior to

surgery on postoperative day 0), as well as postoperative days 1, 2, 3, 4, 7, 21, 30, 45;

and months 2, 3, 6, 9 and 12).

Statistical Plan and Data Analysis

Primary Specific Aim: To determine the effect of cryoneurolysis on postoperative opioid

requirements and analgesia following mastectomy.

Hypothesis 1: Opioid consumption will be significantly decreased in the first two months

following surgery with cryoneurolysis compared with usual and customary analgesia.

Hypothesis 2: Surgical pain will be significantly decreased within the first two months

following surgery with cryoneurolysis compared with usual and customary analgesia

(measured with a Numeric Rating Scale).

Primary end point: In order to claim that cryoneurolysis is superior to usual and

customary analgesia, at least one of Hypotheses 1 and 2 above must be superior while the

other at least noninferior.

Secondary Specific Aims: To determine the effect of cryoneurolysis on physical and

emotional functioning and chronic pain following mastectomy.

Hypothesis 3: Physical and emotional functioning will be significantly improved within

the first two months following surgery with cryoneurolysis as compared with usual and

customary analgesia (measured with the Interference Domain of the Brief Pain Inventory).

Hypothesis 4: The incidence of chronic pain will be significantly decreased 6 and 12

months following surgery with cryoneurolysis as compared with usual and customary

analgesia.

Hypothesis 5: The intensity of chronic pain will be significantly decreased 6 and 12

months following surgery with cryoneurolysis as compared with usual and customary

analgesia (measured with a Numeric Rating Scale).

Balance on baseline covariates will be assessed using absolute standardized difference

(ASD):50 i.e., difference in means or proportions divided by the standard deviation. ASD

> 0.1 will be considered to indicate imbalance, and these variables will be adjusted for

in the statistical analyses. Analyses will be carried out using modified

intention-to-treat (i.e., patients who received any study treatment will be analyzed

according to the group to which they were randomized). The overall type I error rate of

the study will be controlled using a parallel gatekeeping procedure (see Study-wide Type

I error rate control).

Primary Objective (Hypotheses 1 and 2). The investigators will estimate the treatment

effect of cryoneurolysis on opioid consumption (Hypothesis 1) and average pain score

(Hypothesis 2) using a joint hypothesis-testing framework. The investigators will

conclude that cryoneurolysis is more effective compared to the standard of care if it is

noninferior on both opioid consumption and pain score, and superior for at least one of

the outcomes during the first 2 months after surgery. No adjustment for multiple testing

is needed for noninferiority testing since the investigators require noninferiority on

both pain score and opioid consumption outcomes.

Noninferiority Testing.

Hypothesis 1 (Opioid consumption). Cumulative opioid assumption is typically log-normally

distributed. The investigators will therefore assess the treatment effect on the

log-transformed cumulative opioid consumption at 2 months using a linear regression

model. The investigators will test for NI of cryoneurolysis to standard of care using a

1-tailed test assuming alpha of 0.025 and NI delta of 1.2 for the ratio of geometric

means (treatment/placebo). All opioids will be converted to oral oxycodone equivalents.

Hypothesis 2 (Average pain score). The investigators will test for noninferiority (NI) of

cryoneurolysis to standard of care using 1-tailed tests and assuming a 1-tailed alpha of

0.025. The primary pain outcome will be the area under the curve (AUC) of patient

"average" pain scores over the first 2 months (60 days; AUC-60). For this outcome the

noninferiority (NI) delta will be a ratio of geometric means of 1.2 in the AUC-60. Since

AUC-60 is expected to be skewed and to have some proportion of zero values, the

investigators will estimate the treatment effect a 2-sample t-test on the log-transformed

(AUC-60 + 1). Noninferiority will be concluded if the upper 95% confidence interval of

the ratio of geometric means is below the NI delta.

Superiority Testing. If NI is found on both pain and opioid use, the investigators will

test for superiority on each of cumulative opioid consumption and average pain AUC-60

using 1-tailed tests (using the primary analyses specified above) with overall 1-tailed

alpha of 0.025. Since there are 2 tests for superiority the investigators will apply a

Holm-Bonferroni correction and use a significance criterion of 0.025/2 for the smaller

P-value and 0.025 for the larger. Similar tests will be conducted for the sensitivity

analyses for pain score. Cryoneurolysis will be concluded more effective at pain

management than standard care, and the joint null hypothesis rejected, if found superior

on at least pain score or opioid consumption and at least noninferior on both.

Secondary pain outcomes in first 60 days. For each of average, current, least and worst

pain score, as well as the pain with ipsilateral arm raise, the investigators will

conduct all of the analyses describe above for average pain score, as well as 1)

assessing the treatment-by-time interaction in a linear mixed effects model using all

measurements over time, and 2) estimating and reporting the treatment effect at each time

point while controlling type I error across time points within each outcome variable

using the Holm-Bonferroni procedure.

Sensitivity analyses for average pain score methodology. In addition to analyzing the

AUC, the investigators will assess the treatment effect on patient "average" pain scores

over time using a linear mixed effects model assuming an autoregressive (AR[1])

correlation structure across scores for the same individual over time. Factors will be

intervention, time (categorical) and baseline average pain score. The investigators will

then test for noninferiority with a 1-tailed t-test in which the numerator is the

estimated treatment effect minus the NI delta of 1 point and the denominator is the

standard error of the estimated treatment effect. In another sensitivity analysis, the

investigators will use a mixed effects proportional odds model with an autoregressive

correlation structure to assess the treatment effect on pain score as an ordinal outcome.

Hypothesis 3 (Physical and emotional functioning). Physical and emotional functioning of

patients will be assessed using: 1) the interference domain of the Brief Pain Inventory

(BPI), and 2) the Patient Health Questionnaire (PHQ-2). For the BPI Interference

subscale, the effect of the intervention will be assessed over the first 2 months as in

Hypotheses 2 and 3 -- using patient AUC as primary analysis and a linear mixed model

adjusting for baseline BPI-Interference domain score as secondary. The investigators will

further analyze the outcome over the entire first 12 months in a linear mixed effects

model and compare the treatment groups at each time point, controlling type I error as

specified in the primary objective under "Secondary pain outcomes in first 60 days". The

effect of the intervention on depression as assessed by the Patient Health Questionnaire

(PHQ-2) at 3-12 months will be analyzed by Wilcoxon rank-sum tests at each time, with

treatment effect estimated as median difference (95% CI).19,20 In addition, a

proportional odds logistic regression analysis adjusting for clinical site will be

conducted for each time point.

Hypotheses 4 and 5 (chronic pain). The effect of the intervention on the maximum/worst

pain (ordinal scale) experienced by patients at each of 6 and 12 months will be assessed

by separate Wilcoxon rank-sum tests at each time point, with treatment effect estimated

as median difference (95% CI). In addition, a proportional odds logistic regression

analysis adjusting for clinical site will be conducted for each time point. Second, the

effect of the intervention on presence of any pain (binary - yes/no) at each of 6 and 12

months will be assessed using chi-square analyses and relative risk (95% CI), as well as

Cochran-Mantel-Haenszel tests stratified by clinical site.

Study-wide Type I error control. The investigators will use a parallel gatekeeping

procedure to control the study-wide type I error at 0.05. For this procedure the

investigators therefore have prioritized (a priori) the study outcomes into 7 ordered

sets (Table 5, following page). Analysis will proceed in that order, and testing will

proceed through each "gate" to the next set if and only if at least one outcome in the

current set reaches significance. The significance level for each set will be 0.05 times

a cumulative penalty for non-significant results in previous sets (i.e., a "rejection

gain factor" equal to the cumulative product of the proportion of significant tests

across the preceding sets). Within a set, a multiple comparison procedure

(Holm-Bonferroni correction) will be used as needed to control the type I error at the

appropriate level. Although the first set represents the 1-tailed joint hypothesis tests

for noninferiority and superiority at alpha=0.025, without modifying the joint hypothesis

test the investigators will use the corresponding 2-tailed alpha level of 0.05 for the

gatekeeping, as all other sets involve 2-tailed tests. Some of the outcomes listed in the

gatekeeping table are overall assessments over repeated measures. As detailed in

statistical methods, treatment effects may also be assessed at individual time points.

Such assessments will proceed according to the gatekeeping framework such that 1) type I

error will be controlled across repeated measurements, and 2) inference will not be made

on outcome variables that are excluded from formal testing/inference due to the

gatekeeping results.

Parallel gatekeeping procedure [revised December 12, 2022, after 7 participants were

enrolled because the study is a pragmatic trial and not all enrolling centers include

peripheral nerve blocks and/or perineurial local anesthetic infusions as part of their

standard care; and therefore anticipated opioid consumption and pain scores will be

higher across all groups relative to the single-center pilot study]

Sets: Time frame... Required to pass to next set

1. H1/H2 - Joint hypothesis - opioids and pain Requires: NI both, superiority on at

least one, 2 months, Reject joint H0 (1 joint test)

2. H3 - BPI interference subscale, 2 months, Significance on this outcome

3. H4/H5 - Chronic pain: (1) incidence and (2) worst pain, 6 months, Significance on

either outcome H4/H5

4. H4/H5 - Chronic pain: (1) incidence and (2) worst pain, 12 months, Significance on

either outcome H4/H5

5. H2 - Percentage of each group that required <=3 opioid tablets from recovery room

discharge through 2 mos

6. H1 - Percentage of each group that experienced no more than moderate (NRS < 7) pain

at all time points 2 months, Significance on either outcome 5 or 6

7. H3 - Depression screen PHQ-2 6 & 12 months Significance on either outcome

Interim Analyses. The investigators will use a group sequential design with a non-binding

futility boundary and conduct an interim analysis at 50% of the maximum planned

enrollment to assess the efficacy/futility of the intervention. Specifically, the

investigators will maintain the overall alpha level (monitoring efficacy) at 0.025 using

gamma parameter of -4 and power at 90% (monitoring beta, type II error) using gamma

parameter of -4. Under the alternative hypothesis, the cumulative probability of crossing

an efficacy (and futility in parentheses) boundary at the 1st and 2nd analyses will be

0.33 (0.01), and 0.90 (0.10). Under the joint hypothesis testing framework, the

investigators aim to have 90% power to detect NI on both outcomes and superiority on any

one outcome.

Sample Size Justification and Power Analyses. Sample size calculations and power analyses

for the full study were informed by estimates from the pilot trial. The investigators

plan to have 90% power for rejecting the joint hypothesis test for the primary aim.

Opioids. In the pilot study (N=30), the median [quartiles] of cumulative opioid

consumption over 60 days was 91 [15, 146] in the control group and 3 [0, 15] in the

treatment group. The ratio of geometric means [95% CI] was 0.13 [0.03, 0.54] indicating

an 87% estimated relative percent reduction in cumulative opioid consumption at 60 days.

The investigators observed a coefficient of variation (CV) of 1.4. Noninferiority:

Assuming a geometric mean ratio of 0.70, a NI delta of 1.2, alpha of 0.025 and CV of 1.4,

the investigators would have 96.2% power to reject the null hypothesis. Superiority: A

sample size of 108 patients in each group would yield 95.0% power to detect a geometric

mean ratio of 0.57 (treatment/ placebo), assuming a CV of 1.4, a 1-tailed alpha of

0.0125, and after adjusting for interim analyses.

Pain. In the pilot study (N=30), the median [quartiles] of AUC-60 was 29 [16, 67] in the

control group and 1 [0, 4] in the treatment group. The ratio of geometric means [95% CI]

was 0.10 [0.04, 0.27] indicating a 90% estimated relative percent reduction in pain score

at 60 days in treatment versus placebo. The investigators observed a coefficient of

variation (CV) of 1.4. Noninferiority: Assuming a true ratio of geometric means in AUC-60

of 0.71, a NI delta of 1.2, a 1-tailed alpha of 0.025 and CV of 1.4, a sample size of 108

patients per group would yield 95.3% power to detect noninferiority of cryoanalgesia

versus control after adjusting for interim analyses. Superiority: A sample size of 108

patients in each group would yield 95.0% power to detect a geometric mean ratio of 0.57

(treatment/placebo), assuming a CV of 1.4, a 1-tailed alpha of 0.0125, and after

adjusting for interim analyses. With the same sample size, the investigators would have

92.4% power to detect a decrease of 1 point in the pain score assuming alpha = 0.0125,

standard deviation of 2.5, intraclass correlation coefficient (ICC) of 0.55 and an

average cluster size (measurements per participant) of 7.

Power for Joint Hypothesis test. The investigators will have 90% power to reject the

joint null hypothesis, e.g., 95% power for superiority on pain score times 96% power for

noninferiority on opioids = 91%, assuming independence between the two outcomes.

Loss to follow-up. Within our pilot study (n=30) there were no lost patients for the full

1-year follow-up period. However, this might be a unique subset of patients which may

differ from a future cohort at the same or other enrolling centers. Based on previous

multicenter clinical trials, the investigators estimate that at most 7% of participants

in each group are expected to drop out of the study before reaching the 2-months primary

outcome assessment. For those missing data the investigators will use intent-to-treat and

multiple imputation (multiple imputation for chained equations (MICE)) using data on all

observed baseline and outcome data.

Sample size re-estimation. At the first interim analysis (50% of maximum enrollment), the

investigators will estimate variance and ICC of the pain scores, and CV for opioids, and

re-estimate the required sample size. All analyses will either adjust for clinical site

(e.g., in a regression model) or consider it as a stratification variable (e.g., in a

Cochran-Mantel-Haenszel relative risk analysis). Statistical Analytic Software (Carey,

North Carolina), R programming language (The R Project for Statistical Computing) and

East 5.3 software (Cytel Inc.) will be used for all analyses.