Reference
Accordino MK, Lee S, Leu CS, et al. Randomized adaptive selection trial of cryotherapy, compression therapy, and placebo to prevent taxane-induced peripheral neuropathy in patients with breast cancer. Breast Cancer Res Treat. 2024;204(1):49-59.
Study Objective
Comparison of cryotherapy, compression, and placebo in prevention of taxane-induced neuropathy in breast cancer patients. The primary goal was to select the best intervention.
Key Takeaway
Compression appears a superior treatment; it produced better results and was better tolerated.
Design
This was a “single-center randomized phase IIB three-arm adaptive sequential selection trial” that compared 2 treatment methods reported to reduce neuropathy resulting from chemotherapy.
Participants
Investigators enrolled 63 female patients, aged 18 years or more with stage I–III breast cancer, at Columbia University Medical Center in New York from April 2019 to April 2021. Median age when signed up was 54.4 years.
All patients received 1 of the taxane chemotherapies (nab-paclitaxel, paclitaxel, or docetaxel) for at least 12 weeks. Patients who had previously received taxane or platinum-based chemotherapy were not eligible. Those with a known history of neuropathy, Raynaud’s phenomenon, peripheral arterial ischemia, or cold intolerance were also ineligible, as were those currently using the antidepressant duloxetine, which may be used off-label to treat neuropathy.
The patients were randomized into triplets, and each of the 3 was randomized to each intervention arm (n=20 cryotherapy, n=22 compression therapy, and n=21 placebo). The majority of patients (60.3%) received docetaxel every 3 weeks.
Interventions
Trained staff applied the study garments to the patients 15 minutes prior to the start of their chemotherapy infusions and removed the garments 15 minutes after completion of treatment.
Patients randomized to cryotherapy as an intervention wore frozen NatraCure flexible socks on their hands and feet. These socks are designed to be frozen and are typically worn to treat plantar fasciitis. Investigators froze the socks at −25 to −30 °C for 3 hours prior to use and inserted 2 frozen gel packs. Patients used 2 sets of socks during each session; as 1 set warmed, it was replaced after about an hour with a fresh set to keep the hands and feet consistently chilled.
Patients randomized to the compression intervention wore Sigvaris Secure Armsleeves compression gloves and gauntlets, which were then wrapped with Sigvaris Compreflex Standard Arm wrap on both arms. These compression garments are designed and used to treat lymphedema. This combination applied 20 to 30 mmHg of compression, which investigators measured with a gauge inserted under the wrapping. In a similar manner, compressive garments were applied to both lower extremities and feet, which provided 20 to 30 mmHg compression on the lower leg and 15 mmHg on the toes and feet.
The placebo group also dressed in layers of Sigvaris garments, but these were only loosely applied and exerted little pressure (<3 mmHg).
Study Parameters Assessed
Investigators used the Functional Assessment of Cancer Therapy Neurotoxicity (FACT-Ntx) to assess changes in symptoms. This FACT-Ntx questionnaire contains 11 questions related to sensory, motor, and auditory neuropathy as well as dysfunction associated with neuropathy.
Secondary end points included comfort/satisfaction and adherence to study garments, quality of life (QOL), vibration and pressure sensation, taxane toxicity, and safety of interventions.
Primary Outcome
Investigators assessed whether the patients found the garments comfortable based on a 4-point scale similar to the 1 used in prior studies of cryotherapy.
They evaluated quality of life in 7 domains (physical function, anxiety, depression, fatigue, sleep disturbance, social functioning, and pain interference) using a 5-point scale via PROMIS-29 v2.1.
Research staff measured sensory perception using a Neuropen to test pressure sensation and a 128-Hz tuning fork to test vibration.
Investigators evaluated taxane toxicity— defined as onycholysis, peripheral and motor sensory neuropathy, dysesthesia, paresthesia, and neuralgia—using the National Cancer Institute’s (NCI) Common Terminology Criteria for Adverse Events (CTCAE).
Key Findings
Between April 2019 and April 2021, 63 patients were randomized: cryotherapy (20); compression (22); placebo (21). Most patients (60.3%) were treated with docetaxel. The stopping criterion was met after the 17th triplet (n=51) was evaluated; success at 12 weeks occurred in 11 (64.7%) on compression therapy, 7 (41.1%) on cryotherapy, and 7 (41.1%) on placebo.
Adherence to the intervention was lowest with cryotherapy (35.0%) compared to compression (72.7%) and placebo (76.2%).
Investigators continued to collect data for a longer period. When this overrun period finished, the study actually ended with 63 patients whose data were evaluable for the primary end point. At that time, 13 out of 22 (59.1%) in the compression arm had a successful outcome; 9 out of 20 (45.0%) had a successful outcome in the cryotherapy arm; and 11 out of 21 (52.4%) had a successful outcome in the placebo arm.
Success of a therapy was defined as less than a 5-point change in FACT-Ntx.
The likelihood ratio at the time of stopping was 47.05, which gives strong evidence that the selected arm (ie, the compression arm) is truly the best as opposed to the apparently second best (ie, the cryotherapy or placebo arm).
Treatment tolerability
As one would predict, patients in the placebo group, who wore the looser-fitting compression garments, ranked them the most comfortable: 93.8% reported they were satisfied/very satisfied with the study garments. Nearly the same percentage (90.5%) of those who wore the snug-fitting compression garments said the same. Only 75.0% treated with cryotherapy reported their garments comfortable.
A better measure of garment tolerability is how often the garments were actually worn during treatment. Only 35% of the cryotherapy group wore the garments for ≥ 80% of the infusions. In contrast, 72.7% of the compression group were compliant in wearing the assigned garments. A similar percentage of the placebo group (76.2%) did as well.
Transparency
The study was funded by the National Institutes of Health’s (NIH) National Cancer Institute and the Thompson Family Foundation Initiative at Columbia University. The authors have no connection with the companies that supplied the garments.
Practice Implications & Limitations
Any discussion of this paper's results must begin by acknowledging that this study used a statistical methodology with which few of us are familiar; it relied on sequential selection procedures. We lack many of the markers that we are accustomed to, so it may be more difficult to gauge the degree to which we can incorporate the findings into clinical practice.
Such sequential selection procedures have been described as a “statistical method that uses accumulating data from a clinical trial to inform and modify its design. Such redesign might include changes in target sample size and even changes in the target population.” The purpose of running a study in this manner is that doing so may reduce the cost and duration of a clinical trial to allow more rapid testing and quicker adoption of new therapies. These are worthy reasons, but this reader feels somewhat lost not seeing the normal signposts such as P values and other routine measures of significance clearly defined.
The more common methods of statistical analysis are employed to design most clinical trials and determine the sample size (number of patients enrolled) by estimating the percentage of people expected to respond to the intervention, as well as the expected placebo response, and then using these estimates in a standard formula to come up with the sample size needed to reach statistical significance. In many situations, there is such limited information before a study is done that the magnitude of therapeutic and placebo responses might only be guessed.
If the actual treatment effect turns out to be different than what was anticipated or guessed to be, this can pose a problem. If the treatment effect is smaller than investigators anticipated when planning the study, it can fail to identify an effect even if it is present, and a potentially useful therapy may be discarded. If the therapeutic effect is larger than anticipated beforehand, the study may enroll more patients than needed to demonstrate the effect, resulting in the study lasting longer and costing more than required.1
While both cryotherapy and compression have been studied in the past, outcomes have varied widely, and assessments of reliability would have been challenging. Choosing to use sequential selection procedures makes sense in this situation.
While we are all well-accustomed to studies where significant findings have a P value of less than 0.05, this study appears to have a wider tolerance for producing inaccurate conclusions. Following the rules set out in the adopted methodology, the authors tell us: “The specified procedure guarantees a probability of at least 80% correct selection [of the truly best intervention].” The logical assumption from this statement is that there is a 20% probability of providing the incorrect selection.
The study ran over its proposed finish line, and additional data were acquired after what should have been the official stopping point. The extra data did not increase the certainty of the outcome but rather decreased it slightly: “Including the overrun data, the observed ORs decreased to 1.77 and 1.59.” where initially at the stopping point, “OR comparing compression therapy to either of the other two groups was 2.62.”
Let’s put these considerations aside and look at the actual topic under discussion, the neuropathies that unfortunately affect many cancer patients after treatment is completed.
Chemotherapy-induced peripheral neuropathy (CIPN), as this condition is properly labeled, is a common and often debilitating side effect of taxane chemotherapy. The taxanes are a class of drugs originally derived from the Pacific yew tree (Taxus brevifolia) that block cell growth by stopping mitosis. They do this by interfering with the microtubule formation (making them antimicrotubule agents). Paclitaxel and docetaxel are probably the most commonly used of drugs in this class.
CIPN is “often described as a ‘glove and stocking’ neuropathy since symptoms are often felt in the hands and feet. It is characterized as (i) a gain of sensory neuronal function, where patients experience sensations of ‘pins and needles’, tingling, and neuropathic pain; (ii) a loss of sensory function, where symptoms are numbness, dulled sensation, and a loss of vibratory sense; or (iii) a combination of both a loss and gain of sensory function. Platinum-based drugs, taxanes, vinca alkaloids, bortezomib, and thalidomide are known to be associated with CIPN development.”2
Some papers suggest that the taxane drugs are more likely to cause neuropathy than other drugs. Klein and Lehmann in a 2021 paper estimate that “peripheral neuropathy is the most common side effect of chemotherapy, affecting up to 60% of all cancer patients receiving chemotherapy. Moreover, paclitaxel induces neuropathy in up to 97% of all gynecological and urological cancer patients. In cancer cells, paclitaxel induces cell death via microtubule stabilization interrupting cell mitosis. However, paclitaxel also affects cells of the central and peripheral nervous system. The main symptoms are pain and numbness in hands and feet due to paclitaxel accumulation in the dorsal root ganglia.” Note that the 97% CIPN rate is with gynecological and urological cancers, not breast cancer. It’s not clear if that matters, as similar dosing schedules are followed in breast cancer.4,5 These estimates of occurrence of acute CIPN are higher than seen in some other studies.
Before accepting this as the final word, however, there are other studies to consider.
In the Accordino study, all the patients had breast cancer and were treated with a taxane drug; the majority (60.3%) were treated with docetaxel and the remainder with paclitaxel.
While neuropathy symptoms vary to some degree between the different taxane drugs, for our purpose, we are interested in only docetaxel and paclitaxel symptoms as the patients in Accordino’s trial did not receive nab-paclitaxel. We have good descriptions of how neuropathy symptoms vary between the different taxane drugs from a 2022 study by Mo et al.6
The majority of nab-paclitaxel patients (83.4%) reported numbness in hands or feet related to sensory symptoms, while 47% of the paclitaxel patients and 44% of the docetaxel patients reported mainly motor symptoms such as weakness in legs. Such weakness implies neurotoxicity in the dorsal root ganglia, which might not be greatly affected by inducing vasoconstriction in the extremities. Patients reported motor symptoms earlier than sensory abnormalities. The risks of patient-reported CIPN were lower in the paclitaxel group (hazard ratio [HR], 0.59 [95% CI, 0.41–0.87]; P=0.008) compared to the docetaxel group (HR, 0.65 [95% CI, 0.45–0.94]; P=0.02).6
These numbers are important to keep in mind when interpreting the Accordino findings; we might wish to invert how we describe the occurrence of neuropathy and think in terms of the risk ratio of not experiencing neuropathy. It is only by doing so that the placebo results become plausible. Expressed this way, 35% of those treated with docetaxel and 41% of those receiving paclitaxel in the Mo study did not report CIPN symptoms.
Can we direct chemo-enriched blood to desired locations?
At first read, it is apparent that the authors believe that compression is the superior treatment for reducing severity of CIPN and that cryotherapy treatments are on par with placebo at preventing symptoms. The placebo group’s success rate seems high unless you recall that about 40% of patients treated will not experience CIPN and so would have been reported as “successes” in Accordino’s study.
Thus, we should not be surprised when the authors write, “At the prespecified selection time, a successful outcome (FACT-NTX change of < 5 points from baseline) occurred in approximately two-thirds of patients treated with compression therapy, and 41.1% of patients in the cryotherapy and placebo group.” If we do not account for the number of patients who undergo chemotherapy and do not develop CIPN, the seemingly high success rate of the placebo group might appear unaccountably elevated.
The varying levels of compliance with the assigned therapies also need to be considered. The cryotherapy treatment may have been too uncomfortable for many patients, as apparently the majority did not complete the treatment intervention with the prescribed adherence. One assumes those cold-retaining socks frozen to subzero temperatures might have been just too uncomfortable to keep on. “Only 35% of the cryotherapy group wore the garments for ≥ 80% of the infusions,” the investigators wrote. In other words, 65% of the cryotherapy group did not wear the garments at the minimum frequency requested. With such poor compliance, it may not be that cryotherapy is an inferior treatment; it may just be harder to get people to do it. Compression may be a superior treatment only because patients are more compliant.
Recall that 59.1% of those in the compression-treated group were judged to have successful outcomes based on their point score. In the placebo group, 52.4% had a successful outcome. If we assume 40% of the study population will not experience CIPN without intervention, we are looking at closer to 19% vs 12% success rates when comparing compression vs placebo.
That is a small difference. The study authors, who undoubtedly understand the statistical underpinnings of this study’s methodology better than I do, write, “The specified procedure guarantees a probability of at least 80% correct selection (of the truly best intervention), if that one is truly superior to the others by a pre-specified amount namely by an odds ratio (OR) greater than or equal to 2.0 comparing the odds on a successful outcome with the truly best treatment to the next best.”
The basic accepted idea is that certain chemotherapy drugs—in particular, the taxanes—injure the neurons in the hands and feet and that this leads to the symptoms. It is blood that carries and distributes these drugs throughout the body, and the simple premise underlying these 2 therapies is that if the blood supply to a specific region is reduced, less of the drug will be carried to that area; consequently, the damage to that region of the body will be proportionally reduced. The thought behind both compression and cryotherapy is that both will cause blood-vessel constriction and reduce the circulation of blood and chemotherapy agents to the specific treated area. This basic idea seems to work well with chemotherapy-induced hair loss. Both cryotherapy and compression have been used to help prevent chemo-induced alopecia.
It is now common to see cryotherapy used in the form of cold caps, worn in chemo-infusion rooms. Apparently, they work. A 2021 meta-analysis of 9 clinical trials compared the risk of chemotherapy-induced alopecia among patients who used scalp-cooling therapy to those who did not and reported a 41% lower risk of alopecia [RR 0.59, 95% CI: 0.53, 0.66].7
Researchers also have investigated compression for preventing alopecia, using tight head bands to decrease the blood supply to the scalp. This method did not become popular, probably because applying enough pressure to alter blood flow to the scalp is uncomfortable, creating a high incidence of headaches and complaints of discomfort.8 Whether it reduced alopecia is unclear from the reports.
Gianpaolo Ronconi et al reviewed these competing techniques of compression and cryotherapy in detail in their March 2024 systematic review of nonpharmacological treatments for CIPN.9 They cite several earlier studies that have investigated cryotherapy for preventing CIPN.
A randomized trial (N=44) by Shigematsu et al, published in 2020, compared cryotherapy treatments vs standard of care in women receiving weekly paclitaxel for 12 weeks and reported that the experimental arm showed a marked decrease in sensory and motor neuropathy symptoms, as measured by changes in FACT-Ntx scores, which were significantly lower in the cryotherapy group than in the control group (41% vs 73%, P=0.03).10
Another study, by Jue et al and published in 2022, described a controlled trial of 48 women on weekly paclitaxel, and it also reported a significant reduction in CIPN: “… patients on standard therapy were three times more likely to develop [CI]PN, with a progression toward severe [CI]PN, when compared to patients who were on cold therapy.”11
Not all past cryotherapy trials have reported significant benefit. In a study conducted in the Netherlands and published in 2020, Beijers et al found a nonsignificant improvement in neuropathy symptoms in the hands of 90 patients wearing frozen gloves during chemotherapy in comparison to an equal-sized control group. Yet, in this case, the authors reported that 1/3 (34%) of the frozen-gloves group discontinued the study before the end of the treatment, complaining of discomfort.12
Another randomized trial by Ng et al, also from 2020, reported no significant difference in CIPN symptom severity in patients using cryotherapy on hands and feet compared to a control group over the short term, though there were slight differences 3 months later. While no significant adverse effects were seen with the cryotherapy treatment, cold tolerance was a problem. The authors admitted that a limitation of their study was that “temporary interruption of cryotherapy occurred in 80.9% of the subjects due to cold intolerance which could have affected the results.” The cryotherapy devices were removed from the hands approximately once per treatment; often the excuse was to use the restroom (58%) or admitted discomfort (37%). The authors report further that 91% of the subjects interrupted the cryotherapy at least once during chemotherapy “due to cold intolerance.”13
Those of us who have had the pleasure of being trained in hydrotherapy will appreciate the significant impact this might have had. While cold applications like frozen gloves may restrict blood flow, warming the hands midway through treatment might have converted the therapy from a simple cold application to an alternating contrast of heat-and-cold applications. The latter might have had a very different effect on blood flow, allowing a reflexive surge in blood circulation and countering the desired effect of reducing drug exposure.
The body’s regulation of blood flow to the hands and feet when exposed to cold is complex, and these past attempts at preventing CIPN with iced gloves or ice-water immersion seem to ignore well-established physiology. Several competing reflexes control blood circulation in the hands and feet, and understanding their actions may be helpful. Whole-body cold stress, for example, falling into a cold lake, will activate the sympathetic nervous system and induce a reflexive vasodilation at the body’s core and vasoconstriction in the extremities; blood will be shunted from peripheral extremities to the body’s core to minimize heat loss and keep vital organs warm. When this occurs, our hands and feet get cold. It is this reflex that the cryotherapy studies have hoped to rely on. The downside of such reflexive peripheral vasoconstriction (as most of us learn early on in childhood when playing outside in the winter) is a sharp decrement in dexterity. Our hands get numb and fumbly. The common strategy to restore the extremities to a functional temperature is by rewarming them (eg, putting them in a warm pocket or putting on mittens or trying to locate them against our companions’ bare flesh).
A second cold-induced reflex, first described in 1930 by Sir Thomas Lewis, is probably relevant. If the hands or feet are exposed to cold for 5 to 10 minutes and the body core is warm, a local thermoregulatory reaction called “cold-induced vasodilation (CIVD)” comes into action. CIVD was initially called the “hunting reaction” because that was where it was first noticed, in hunters, who were able to handle metal guns in cold weather and retain manual dexterity. After 5 to 10 minutes of cold exposure, the arteriovenous anastomoses in the extremities begin to dilate in an oscillatory pattern and provide hands and feet with transient episodes of warm blood.14 Whole-body hypothermia will override this reflex, but until such a life-threatening situation occurs, the hands will get only so cold before the body rescues them. This reflex may be why attempts to limit CIPN by cooling the extremities have not produced better results. The various methods employed trigger this hunting reaction, a localized, cold-induced vasodilation that recirculates the chemotherapy-infused blood into the hands and feet.15 A person’s propensity to elicit this CIVD reflex varies with genetics, ethnicity, age, and physical condition. I would guess it also would vary with temperature exposure. None of these factors seem to be accounted for in these studies, and there may be confounders we are unaware of that explain the discrepancies between results.
It may be that the cryotherapy treatments were too cold. The hunting reaction is triggered by severe cold; perhaps a milder chilling effect would be adequate? A very recent study looked at the potential effects of cooling temperatures on oxaliplatin-induced neuropathy and found benefit. In this new study, hands and feet were cooled to 11 °C (about 52 °F) during infusion, and that apparently worked.16 (Perhaps it isn’t the blood flow but the cellular transport mechanisms for drawing the drugs into cells that shift with temperature?)
Compression therapies have been tested in the recent past without significant benefit.
Kotani et al reported in 2021on a trial (N=49) in which women undergoing weekly paclitaxel treatments for breast cancer served as both control and experimental “arms” of the experiment. The women wore surgical gloves on both hands starting 30 minutes before infusions; they did not remove the gloves until 30 minutes after. They wore normal-fitting gloves on the control hand and gloves a size too small on the experimental side to create compression. The patients and physicians were “blinded” as to which hand was the “study” side, though one might think the different glove sizes would make it obvious. The compression created by the tighter gloves made no significant difference in symptom outcome.17
This result seems especially relevant in light of the Accordino study’s conclusion that compression provides benefit.
Just recently, Dongxue et al published the results from a study conducted in China in which 80 patients with breast cancer were divided into 2 groups of 40. One group served as a control and received routine care, while the second group was treated with compression therapy in addition to routine care. Thirty minutes prior to receiving their infusion therapy, these patients donned undersized surgical gloves and elastic socks and wore them until 30 minutes after the infusion was completed. The investigators compared incidence of CIPN, anxiety, depression, and sleep scores between both groups before and after compression therapy during chemotherapy cycles 2, 4, and 6. In this most recent study, compression using tight surgical gloves and elastic socks was associated with significantly less CIPN (P<0.05). The compression treatments were also associated with significantly lower anxiety and depression and with improved sleep.18 It is unclear if and how this study differed from the earlier study that did not find benefit.
Although the statistical methodology used in Accordino et al is unfamiliar, this reader is willing to accept their conclusions that compression therapy reduced CIPN, whereas cryotherapy was no more effective than placebo at doing so. It may be that these benefits can be elicited using simpler, less-expensive compression garments than employed in the Accordino protocol—for example, by doubling up too-small surgical gloves on the hands as Dongxue et al did. While many practitioners may wish to wait until these methods are better defined and better proven, our patients may be eager to try anything that might lessen their chance of experiencing debilitating discomfort. At this point, there is no reason to think that it will hurt to try some form of compression during infusion therapies with taxane drugs.
This study is worth our attention, in no small part for pointing out the potential benefits of compression during chemo, but I think more importantly for the suggestion that chemotherapy drugs may be directed toward or away from different areas of the body. In the 2 examples investigated, the goal was to keep the drugs at lower concentrations in sensitive regions. There may be other ways to employ this general concept.
As discussed, the body’s most basic response to sudden cold is to vasoconstrict blood vessels in the peripheral extremities and to bring blood to the body’s core. Perhaps chilling the body’s core might be a more direct approach than these frozen-garment strategies? Infusion patients complain how chilled they get during therapy because the drugs are “cold.” If the drugs were to be purposefully chilled, could they be used to direct blood flow? Not just away from the peripheral extremities but, in situations where tumors are located in the core, toward the areas where we want the most chemo action to occur? It has become common practice to offer infusion patients heated blankets to prevent them from becoming chilled during treatment. Do we know whether this is really a good strategy, or could the added warmth shunt more blood to their peripheral areas and away from their cores? * Hyperthermia using infrared treatments is 1 method being explored for superficial and deep-tissue tumors. Our hydrotherapy instructors certainly instilled in us the idea that alternating applications of hot and cold will increase blood flow to a specific targeted area. While alternating hot/cold seems rather simple both in concept and practice, I’ve not seen it done in practice with the intent of affecting blood flow. Various light applications, both laser and light-emitting diode (LED), are common therapeutics used to treat injuries, because they trigger the release of nitric oxide and increase blood flow to areas exposed to these lights.19,20 Might these simple lights be used during infusion to direct chemo action?
Two decades ago, it was reported that the increased blood flow triggered by sildenafil (which increases nitric oxide [NO] and results in vasodilation) enhances chemo action,21 and such studies continue to be published suggesting that vasodilation and increasing blood flow is a useful adjunctive treatment.22,23 Rather than just preventing chemotherapy drugs from reaching nerve cells, perhaps we can direct those drugs and increase exposure of tumors to the cytotoxicity of the chemotherapy using hot and cold and, by doing so, improve treatment outcomes.
Conflict of Interest Disclosure
The author reports no conflict of interest.
