Cough Desensitization Treatment: A Randomized, Sham-Controlled Pilot Trial for Patients with Refractory Chronic Cough

Abstract

Background:

The purpose of this study was to determine feasibility of treating refractory chronic cough (RCC) with progressive doses of capsaicin paired with cough suppression.

Methods:

In this sham-controlled, parallel RCT, 14 adults with RCC were randomly assigned to either behavioral cough suppression therapy (BCST) plus 6 treatment sessions involving exposure to nebulized capsaicin in progressively larger concentrations while actively suppressing cough (n=8), or BCST plus 6 sessions of exposure to a single subthreshold dose of capsaicin (sham; n=6). The Leicester Cough Questionnaire (LCQ) was the primary outcome measure. Urge-to-cough (UTC) testing, measuring both UTC and cough frequency, served as secondary outcome measures. Data was analyzed with mixed effects linear regression and follow-up contrasts.

Results:

Results on all measures favored the treatment group; however, there was only strong evidence of a difference in treatment effect on cough frequency during UTC testing. Mean change in LCQ at 3-weeks post treatment was 2.95 and 1.75 (p = .023), in the treatment and sham groups, respectively. Cough frequency during UTC testing reduced by 97% and 56% (p < .0001) at three weeks post, respectively. Within-group comparisons revealed strong evidence of change in the treatment group (p < .001) and moderate evidence of a change in the sham group (p = .08).

Conclusions:

Conclusions from this study are limited due to the very small sample size; however, the study provides feasibility and proof-of-concept evidence to support further investigation of treating RCC with repeated exposure to nebulized capsaicin paired with BCST.

Trial Registration:

NCT04256733.

Introduction

Chronic cough (CC), defined as a cough lasting longer than eight weeks, is an increasingly recognized disorder (1). Patients with CC experience significant psychosocial stress and impairments of quality of life (2–4). Despite extensive testing and medication trials, an estimated 10–20% of patients with CC do not respond to medical treatment and are said to have refractory chronic cough (RCC) (5–8). Distinctive features of RCC include a non-productive cough, abnormal environmental triggers such as smoke, fumes, and dust, or mechanical stimuli, such as talking or laughing (9). These characteristics suggest a heightened sensory response in the airway (6, 9). Enhanced sensitivity has led to the diagnostic term of cough hypersensitivity syndrome (CHS) (1, 6, 10, 11). Although the exact mechanism involved in this hypersensitization is unknown, it appears to be similar to that involved in neuropathic pain (12). As in neuropathic pain, it is hypothesized that most patients with CHS experience initial peripheral sensitization of the airway due to an inflammatory upper respiratory illness; neuroinflammatory receptors then become centralized, leading to long-term cough (1, 6, 12, 13). Evidence exists for both peripheral and central hypersensitivity. Laryngeal biopsy studies have shown an increase in expression of known cough receptors (e.g., TRPV1) expressed on laryngeal c-fibers in patients with RCC (14–16). Functional brain imaging studies have shown heightened activity in the midbrain in individuals with RCC relative to healthy controls (17, 18). Further, patients with RCC are less able to suppress cough than healthy controls and this difference is associated with reduced activity in the dorsomedial prefrontal and anterior mid-cingulate cortices (17) suggesting an impairment in central-mediated inhibitory networks.

Several studies have shown that behavioral therapy, typically provided by a speech-language pathologist, for RCC is efficacious (19–21). Three randomized control trials (19, 20, 22) as well as a number of case series (21, 23–25) have examined the effect of the therapy using various cough outcome measures. All of the studies reported improvement in cough status after treatment, measured by an increase in self-reported quality-of-life, and a decrease in cough symptoms and cough frequency. Many terms are currently used to describe behavioral therapy for cough throughout the literature. The term behavioral cough suppression therapy (BCST) is used in this manuscript because it is descriptive of the therapy and clearly delineates it from medical treatment (5, 26).

Though BCST has been performed by a variety of professionals, speech-language pathologists (SLPs) have become the main advocates for BCST due to their role as specialists in voice and upper airway disorders (5, 19, 27). Participants treated with BCST typically require only 1–4 weekly sessions, and most experience a reduction in cough symptoms within 1–2 weeks (10, 21, 28–30). During BCST, clients are educated on cough hypersensitivity and the rationale for cough suppression, behavioral management of medical conditions that commonly contribute to cough (e.g., reflux and post-nasal drip), vocal hygiene practices, and resonant voice therapy and respiratory retraining, if applicable (e.g., for muscle tension dysphonia or induced laryngeal obstruction). The primary component of BCST is teaching patients various cough suppression strategies to be implemented anytime an urge-to-cough (UTC) exists (30). Although improving control of one’s cough is of benefit to patients, the primary goal of BCST is not to make patients good at suppressing their cough, but rather to stimulate a reduction in cough sensitivity through neuroplasticity. In this way, BCST is designed to capitalize on the lose-it portion of the use-it-or-lose it principle of neuroplasticity (21, 31, 32). It is hypothesized that avoiding cough when an UTC is present results in a reduction in cough-reflex sensitivity, which is consistent with results of studies that show a decrease in cough-reflex sensitivity following BCST (21–23). However, the mechanism by which this change occurs is not yet known. The change may be due to a decrease in the afferent cough signal, through modification of peripheral sensory receptor expression, or by strengthening central inhibitory networks which modify the perception and/or control of such afferent signals. The latter (strengthening central inhibitory networks) would seem to be the most likely in the case of BCST, given the evidence of impaired cough suppression in patients with RCC (17), and training in behavioral cough suppression is the primary component of BCST.

Despite the high success of BCST for many individuals with RCC, the therapy does not improve cough for every patient, and of those who benefit from the therapy, an estimated 11%–16% do not have full resolution of their cough (19, 27). Patients who are unable to suppress their cough, whether due to non-compliance in using cough suppression strategies, or inability to suppress when exposed to strong triggers within their environment, are the most unlikely to benefit from the therapy. We hypothesize these patients may benefit from a modified form of BCST that involves exposing the patient to progressively higher concentrations of a controllable cough stimulant, across repeated sessions, while coaching the patient to actively avoid coughing. The safety, reproducibility, and standardized dilution of aerosolized capsaicin make it an ideal cough stimulant for this purpose. We previously demonstrated proof-of-concept of this hypothesis with five healthy participants who completed six treatment sessions over a two-week period (26), during which they were exposed to aerosolized capsaicin in progressive concentrations via nebulizer and coached to suppress their cough. Every participant demonstrated a significant increase in cough sensitivity thresholds as tested by capsaicin cough challenge testing before and after treatment (26), suggesting a reduction on cough reflex sensitivity. Participants maintained these improvements at three-weeks post-treatment.

The current study investigates this treatment model, which we term cough desensitization treatment (CDT), on patients with RCC who have not had a sufficient response to BCST. We hypothesized that repeated exposure to progressive doses of aerosolized capsaicin while implementing behavioral cough suppression strategies (i.e., CDT) would stimulate a change in cough reflex sensitivity, perceived urge-to-cough (UTC), and cough-related quality of life.

Methodology

Design and Participants

This randomized, parallel-group, sham-controlled trial was approved by both the Food and Drug Administration (FDA; Investigational New Drug (IND) #142148) on January 3, 2019 and the University of Montana Institutional Review Board (IRB #188-18) on April 5, 2019. Fourteen adults with RCC who failed medical treatment and did not have full resolution of cough with traditional BCST completed the study. Participants were recruited from two pools: local speech therapy clinics that provide BCST and from participants who completed a survey study about BCST and scored no greater than 16 on the Leicester Cough Questionnaire following BCST (29), indicating their cough had not fully resolved. In addition to already completing a trial of BCST, inclusion criteria also included normal chest x-ray, normal pulmonary function as measured with spirometry, and normal laryngoscopy. Exclusion criteria were as follows: current smoker, pregnant or trying to become pregnant, diagnosis of a respiratory or pulmonary condition (e.g., asthma, COPD, emphysema, etc.), and not currently or recently on ACE-inhibitor medication or a neuromodulator specifically prescribed for cough. All participants also agreed to avoid any other cough treatments and to inform the investigators if they were prescribed any new medications.

Participants were randomly assigned to either BCST plus an active treatment or BCST plus a sham-treatment. They were told the active treatment under investigation was designed to enhance BCST. They were not told that the active treatment was capsaicin. In order to improve recruitment incentive, participants were informed that if they were randomly assigned to the sham treatment, they would be eligible for the active treatment following their completion of the study. The same data was collected from such participants, during both sham and active treatment, but only their sham data was included in the formal analysis.

The study consisted of three phases: (1) baseline testing, (2) treatment sessions twice per week for three weeks, and (3) two post-treatment testing sessions administered at one-week, and three-weeks post-treatment. A flow diagram of the study is shown in Figure 1.

Figure 1.

Study Flow Diagram

CCT = cough challenge testing; LCQ = Leciester Cough Questionnaire; UTC = urge-to-cough testing; PT1 = post-test 1; PT2 = post-test 2.

Capsaicin Quality Control

Pharmaceutical grade capsaicin, in powder form, was purchased from Formosa Laboratories Inc. (Formosa Laboratories, Inc. Taoyuan, Taiwan 338) and diluted in a sterile environment according to standard procedures outlined by the European Respiratory Society (33, 34). The capsaicin was diluted with 95% ethanol to make two stock solutions of concentrations 0.01, and 0.001 Molar. A sample of each stock solution was sent to Toxikon laboratories, which confirmed sterility. Stock solution stability was confirmed with periodic reverse-phase high-performance liquid chromatographic (RP-HPLC) assay (35) across a six-month period. Stock solutions were discarded after six months of use. The stock solutions were diluted with inhalation-safe sodium chloride to the following concentrations: 0.49, 0.98, 1.95, 3.91, 7.8, 15.6, 31.2, 62.5, 125, 250, 500, and 1000 μM. Dilutions were made under sterile conditions (i.e., made in a hood with airflow filtration) within 24 hours of use. Stock solutions and dilutions were protected from UV light and stored in a temperature-controlled, 4° C refrigerator in order to reduce degradation of the solution.

Outcome Measures

Outcome measures were administered at baseline, and one-week and three-weeks post-treatment. Outcome measures included capsaicin cough challenge testing (i.e., to measure cough-reflex sensitivity), the Leicester Cough Questionnaire (LCQ), and urge-to-cough testing (described below).

Cough Challenge Testing.

Capsaicin cough challenge testing (CCT) procedures, as recommended by the European Respiratory Society (34) and approved by the FDA (IND #142148), were followed to measure cough-reflex sensitivity in participants at baseline and both post-treatment assessment points (33, 36, 37). Participants were exposed, one at a time, to doubling doses of capsaicin dilutions, starting with .49 μM. Each dilution was delivered with the Koko Digidoser via a DeVilbliss 646 nebulizer. The output of the nebulizer was controlled by an inspiratory flow regulator. The dosimeter was calibrated weekly. The single inhalation method was used, with a delivery time of .6 second. Every testing session began with a saline trial to ensure the participant was familiar with the testing procedures and to control for any potential startle response. Two additional saline trials were given randomly during testing to control for anticipation effect. Participants were instructed let their body respond naturally without attempting to suppress any UTC. Number of coughs produced within 15 seconds following each exposure were recorded. A minimum of two minutes passed between each exposure. Testing ceased when the capsaicin concentration that caused five or more coughs (C5) was determined, or the maximum dose of 1000 μM was given.

Preliminary data inspection in early 2020 revealed high variability in our CCT results (e.g., coughing four times on a dose and then coughing zero times on the next two doubling doses) and inconsistency with those results relative to our other outcome measures (e.g., a reduction in CCT threshold but a significant improvement in LCQ scores). Upon further investigation and consultation with other experts, we discovered a problem with the set-up of the DeVilbliss nebulizer. Unfortunately, when setting up our procedures, we overlooked the recommendation to weld the straw and baffle of the nebulizer cup together to ensure the exact distance between the straw and the baffle (and hence reliable dosing) across repeated trials. We believe this led to varying amounts of capsaicin released during each trial and, hence, CCT data prior to correcting for this oversight, was deemed invalid. Unfortunately, shortly after discovering this error, the study was halted due to the COVID-19 pandemic and only two new participants (both randomized to sham) enrolled following correction of this error.

Leicester Cough Questionnaire (LCQ).

Participants also completed the LCQ, a 19-item validated patient-report outcome measure of cough-related quality-of-life (38, 39) with psychological, physical, and social domains. All demographic and survey information, including the LCQ, was gathered using the HIPAA-compliant online survey platform, Qualtrics.

Urge-to-Cough (UTC) Testing.

Participants were asked to report their UTC sensation (i.e., sensation of needing to cough) on a scale from 0 (no UTC) to 10 (maximum UTC) when presented with the following stimulants that are common cough triggers in patients with RCC: perfume, bleach, vinegar, laundry soap, fabric softener (each held under the nose for 10 seconds), deep and fast breath in through the mouth, maximum sustained voicing, reading a 55-word passage, and yelling while counting from one to five. Number of coughs during and immediately after each stimulant task was also recorded.

Treatment procedures

Cough Suppression Training.

During the baseline testing session (following collection of all measures), participants were instructed in relaxed-throat breathing as a cough suppression strategy. This two-part technique takes advantage of natural reflexes within the airway to keep the vocal folds open (40). It consists of a sharp inhale through the nose, followed by a long exhale through pursed lips. The sharp inhale elicits the abductor laryngeal reflex to abduct the vocal folds. Exhaling through pursed lips increases intra-oral and intra-laryngeal pressures, thus resisting closure of the vocal folds to resist coughing. Participants were asked to complete approximately 10 relaxed-throat breaths, four times per day before the first treatment session (typically 3–5 days) and throughout the treatment phase, when not feeling an UTC, in order to create a strong motor pattern. They were also instructed to use the breathing technique to attempt to suppress cough when feeling an UTC in and outside of therapy sessions.

While participants were instructed in relaxed-throat breathing, cough suppression was the goal, so participants were encouraged to modify the breathing technique to maximize cough suppression success. Common modifications included slowing the rate of nasal inhalation, adapting the exhalation to include using a small diameter straw, and variations of inspiratory and expiratory volume. Because the method varied across individuals, it will now be referred to as cough-control breathing (CCB).

Active Treatment.

Participants attended treatment sessions twice per week, for three weeks, with a minimum of 72 hours between each session (required by the FDA). Spirometry was performed before and after each session to ensure participants were safe to begin treatment and pulmonary function remained unchanged following treatment. The goal of each session was to repeatedly elicit a strong UTC that could be successfully suppressed. UTC was elicited by exposing the patient to multiple doses of diluted aerosolized capsaicin via the Koko Digidoser. The same procedures were used as during CCT except participants were coached to suppress their cough. The initial dose given during the first session was the capsaicin concentration that first caused the participant to cough during baseline testing. After receiving the capsaicin via a single inhalation, participants were instructed to complete their breath in, and then immediately begin CCB to suppress the cough. Participants were also allowed to take a sip of water if they found it helpful for suppressing the cough, but only after at least 2–3 CCB cycles. They were asked not to speak for at least 15 seconds after each exposure, to ensure any UTC was elicited by the capsaicin and not from talking. Fifteen seconds after each dose (or longer if active cough suppression was still occurring), participants were asked to indicate their maximum UTC as well as their discomfort on a 0 (no discomfort) to 10 (maximum discomfort) visual-analog scale. If the participant was unable to successfully suppress cough, the capsaicin concentration was reduced for the next exposure. If the participant appeared to suppress cough without apparent significant effort, the capsaicin concentration was increased by half or a full doubling dose for the next exposure. The decision to increase the concentration by half or a full doubling dose was determined by the investigator running the session, based on appearance of effort needed to suppress the cough and the participant’s reported UTC. Generally, the concentration was doubled if the UTC was rated a 3 or less and increased by half if the UTC was higher than 3. The goal of each session was to gradually increase the capsaicin concentration throughout the session, without the participant coughing. Each treatment session started at the capsaicin concentration one level below the maximum suppressed concentration in the prior session.

We initially limited each session to six individual capsaicin exposures. We chose six exposures because, during the proof-of-concept study, healthy participants were generally able to tolerate a doubling increase of capsaicin concentration after just 1–2 exposures at a single dose. However, we soon recognized participants with RCC were not able to progress as quickly. Most required 3–4 exposures at a single dose before they were able to move to a higher dose. We submitted an IRB amendment to increase the exposure rate to 12 concentrations per session after the initial eight subjects; however, by the time we received authorization to increase the exposure rate, only one active treatment participant (T17), received 12 exposures per session rather than 6.

Sham Treatment.

The sham group underwent the same procedures, including number of sessions, spirometry, inhalation of aerosolized substance, and relaxed throat-breathing; however, sham participants received a sub-threshold concentration of capsaicin (i.e., a concentration that did not elicit cough during baseline testing) and the concentration never increased. Two exposures of saline were also given randomly during each session. The sham participants were asked to complete five trials of CCB immediately following each exposure, followed by a sip of water. The nebulizer was rinsed in between each application to avoid participants suspecting that they were in the sham group.

Daily Homework

All participants logged their adherence to daily practice of CCB (prescribed four times per day) as well as an estimate of the percentage of success they had with suppressing their cough each day. This paper log was turned in at each treatment session.

Statistical Analysis

Statistical analysis was performed with the statistical software R (R Core Team, 2021) with mixed models estimated using the lme4 package (41), p-values generated using lmerTest (42), and contrasts estimated using emmeans (43). Linear or generalized linear mixed models were used with a random subject effect to account for three repeated measurements (baseline, PT1, and PT2) on each participant, with fixed effects for the three time points and group (treatment, placebo) and their interaction, incorporated. First, the interaction was tested for. Then, follow-up contrasts were used to compare the differences in the change from baseline to each follow-up between the treatment and placebo groups. The Mann-Whitney U test was used to determine between-group differences on demographic data.

Results

Nineteen participants enrolled in the study. Three were unable to complete the study due to onset of illness unrelated to the study. Two dropped due to the COVID-19 pandemic and chose to not re-enroll when the study resumed. Zero adverse events were reported throughout the study. Participant demographics are shown in Table 1. Mean age for each group was similar at 62 and 66, for treatment and sham groups, respectfully (Mann-Whitney U test (U=24, z=0, p=1.0)). All but three participants reported a cough duration of over two years. Evidence of a difference in cough duration between the two groups was very weak (U=15.00, z=−1.62, p=0.282). Every sham participant, and 7 of 8 treatment participants reported over 90% compliance with daily home practice, an insignificant difference. All six participants from the sham group opted to receive the active treatment following full participation of the sham treatment and post-test data collection; however, one participant was not able to complete the treatment due to the study being halted secondary to the COVID-19 pandemic.

Table 1.

Participant demographics.

	Participant ID	Gender	Age (years)	Cough duration
Sham	S2	M	89	> 2 years
	S7	F	63	> 2 years
	S10	F	75	> 2 years
	S15	F	60	> 2 years
	S20	F	57	> 2 years
	S21	F	52	> 2 years
Treatment	T1	F	72	> 2 years
	T3	F	71	> 2 years
	T4	F	66	> 2 years
	T5	F	62	> 2 years
	T8	F	72	18–24 months
	T11	F	76	> 2 years
	T12	F	28	12–18 months
	T17	F	51	4–6 months

M=male; F=female

Leicester Cough Questionnaire (LCQ).

At one-week post-treatment (PT1), 6/8 (75%) treatment participants and 2/6 (33%) sham participants achieved a clinically meaningful improvement in LCQ, evidenced by a change in total LCQ score of at least 1.3(44). Results at three-weeks post-treatment (PT2) were similar with the addition of one sham participant reaching the clinically meaningful threshold, relative to baseline. A mixed effects linear regression model revealed very weak evidence of a difference in LCQ over time between the two groups (F(2,24)=1.58, p-value=0.23); however, the differences were in the direction of higher mean LCQ in the treatment group. Follow-up contrasts were conducted to estimate the change vs. baseline in each group. There was strong evidence of change in the treatment group at PT1 (t(24) = 4.24, p-value < .001) and PT2 (t(24) = 4.60, p-value < .001), with estimated mean increases of 2.95 and 3.20 LCQ points compared to baseline, respectively. There was very weak evidence of change in the sham group at PT1 (t(24) = 1.47, p-value = .310) but moderate evidence of a change at PT2 (t(24) = 2.17, p-value = .08), with increases of 1.18 and 1.75 points, relative to baseline, respectively. LCQ scores per group for each participant are shown in Figure 2.

Figure 2:

LCQ scores at baseline, 1-week post-treatment and 3-weeks post-treatment per group.

Urge-to-Cough (UTC) Testing: Cough Frequency.

Mean change in the sum of coughs during UTC testing at PT1 and PT2, relative to baseline, was −11.86 and −14.00 in the treatment group, respectively, and −6.67 and −6.33 in the sham group, respectively. A Poisson mixed effects generalized linear regression model revealed strong evidence of a difference over time between the treatment and sham in total coughs during UTC testing (Chi-square(2) = 18.0, p-value = 0.0001). Follow-up contrasts were conducted to estimate the change vs. baseline in each group. There was strong evidence of a change in mean cough counts during UTC testing for both groups at PT1 and PT2 (p-value <.0001 for all contrasts), with an 83% and 97% reduction in the treatment group, and 59% and 56% reduction in the sham group, respectively. Contrasts comparing the estimates of change between the two groups revealed moderate evidence of a different change in the treatment group at PT1 (Z = −1.62, p-value =.11) and very strong evidence of a different change in the treatment group at PT2 (Z = −4.1, p-value <.0001), with both changes in the direction of lower cough rates in the treatment group. See Figure 3.

Figure 3:

Number of coughs elicited during urge-to-cough (UTC) testing at baseline, 1-week post-treatment, and 3-weeks post-treatment per group.

UTC Testing: UTC.

The mean change in the total UTC rating at PT1 and PT2, relative to baseline, was −12.31 and −13.75 in the treatment group, respectively, and −9 and 8.83 in the control group, respectively. A linear mixed model was used to account for repeated measures on the subjects that found little evidence of a Time by Group interaction on the UTC responses (F(2,24)=0.57, p-value=0.57). Follow-up tests compared the differences in changes from PT1 or PT2 to baseline in the treatment and control groups, but, not unexpectedly, little evidence was found of differences in the mean change between the groups (t(24)=−0.71, p-value=0.49 for PT1 vs Baseline and t(24)=−1.05, p-value=0.30 for PT2 vs Baseline). The reduction was larger for the treatment group by 3.3 and 4.92 points, respectively, but not enough to be detected, given the sample size and variability of this scale. All data LCQ and UTC data per participant are included in Table 2.

Table 2.

Participant outcomes for the Leicester Cough Questionnaire and urge-to-cough testing

	Participant	Leicester Cough Questionnaire: Total score			Urge-to-Cough Testing: Sum Total Coughs			Urge-to-Cough (UTC) Test: Sum Total UTC
		Baseline	PT1	PT2	Baseline	PT1	PT2	Baseline	PT1	PT2
Sham	S2	16.98	18.13	17.18	0	0	0	0	0	1.5
	S7	12.48	15.11	16.48	0	0	0	6.5	9	6.5
	S10	11.39	9.0	10.07	2	12	0	30	18.5	3.5
	S15	12.8	18.05	18.95	28	1	4	19	1.5	5
	S20	12.27	11.93	12.29	20	5	16	24	2.5	15.5
	S21	8.66	9.46	10.11	8	0	0	8.5	2.5	3
	Mean (SD)	12.43 (2.69)	13.61 (4.09)	14.18 (3.85)	9.67 (11.76)	3.00 (4.82)	3.33 (6.41)	14.67 (11.50)	5.67 (7.00)	5.83 (5.04)
Treatment	T1	14.68	16.91	18.61	0	0	1	3	.5	1.5
	T3	14.88	17.96	16.75	9	0	0	12.5	0	0.5
	T4	6.68	7.68	7.7	8	0	0	18	.5	0
	T5	11.21	16.82	17.04	2	1	1	18.5	2.5	2
	T8	11.64	14.66	15.95	3	0	0	11	0	3
	T11	15.20	16.07	16.2	0	0	0	0	0	0
	T12	11.04	17.43	15.68	45	8	1	43.5	25.5	14
	T17	15.91	17.32	18.93	48	11	0	39	18	14.5
	Mean (SD)	12.65 (3.11)	15.61 (3.36)	15.85 (3.50)	14.38 (20.12)	2.50 (4.41)	0.38 (0.52)	18.19 (15.67)	5.88 (10.03)	4.44 (6.14)

PT1=post-test 1 (1 week post-treatment); PT2=post-test 2 (3 weeks post-treatment); SD=standard deviation; UTC=urge-to-cough

Cough challenge testing (CCT).

As described above, only CCT data from the final two enrollees (both of which were randomized to sham treatment) was considered valid. Both of these participants showed a reduction in the C5 threshold from baseline to PT1 of one doubling dose and no change between PT1 and PT2, which suggests no considerable change in cough-reflex sensitivity.

Treatment results following sham treatment.

Five participants who completed the sham treatment completed the active treatment following full participation in the sham treatment. All five participants completed the LCQ at one-week (PT1) and three-weeks post-treatment (PT2). UTC testing was completed on all five participants at PT1; however, two participants were unable to return to the clinic for UTC testing at PT2. Two of the five (40%) participants showed dramatic improvements on the LCQ and UTC testing at PT1. One additional participant reached the minimum clinically relevant change in LCQ at PT2. Results are shown in Table 3.

Table 3.

Outcomes following active treatment for participants randomized to sham treatment who then elected to receive the active treatment.

	Leicester Cough Questionnaire: Total Score			Urge-to-Cough-Testing: Sum Total Coughs			Urge-to-Cough Testing: Sum Total UTC
	Baseline	PT1	PT2	Baseline	PT1	PT2	Baseline	PT1	PT2
S2T	16.61	17.09	17.34	0	0	NT	0	0	NT
S10T	11.36	11.82	13.09	20	0	0	14.0	1.5	4.5
S15T	13.5	19.57	16.5	46	8	7	31.0	8.5	14
S20T	12.29	11.7	11.71	16	13	5	15.50	14	7.5
S21T	10.11	13.55	15	0	0	NT	3.0	.5	NT
Mean (SD)	12.77 (2.48)	14.75 (3.47)	14.78 (2.34)	16.40 (18.89)	4.20 (6.02)	4.00 (3.61)	12.70 (12.25)	4.90 (6.14)	8.67 (4.86)

NT=not tested; PT1=post-test 1 (1 week post-treatment); PT2=post-test 2 (3 weeks post-treatment); UTC=urge-to-cough

Discussion

The purpose of this study was to determine if a novel treatment, which we call cough desensitization treatment (CDT), may be effective for patients with refractory chronic cough (RCC) who fail standard medical treatment and behavioral cough suppression therapy (BCST). CDT combines principles of programmatic desensitization with behavioral cough suppression to stimulate a reduction in cough sensitivity, which evidence suggests is the underlying cause of RCC(12, 13, 45–47). We hypothesize this clinically controlled treatment may be particularly beneficial for patients whose cough hypersensitivity is so severe they cannot suppress their cough when exposed to environmental stimuli, reducing the effectiveness of BCST.

The treatment consisted of training in cough-suppression strategies including cough-control breathing strategies (e.g., quick nasal inhale followed by extended exhale through either pursed-lips, hissing, or blowing through a cocktail straw) and sipping water with an effortful swallow. The two groups studied here received the exact same treatment aside from the level of capsaicin exposure given during treatment sessions. The active treatment consisted of progressive doses of capsaicin vapor at a suprathreshold level, which elicited an UTC followed by active cough-suppression. The sham treatment consisted of repeated exposures to a single subthreshold dose of capsaicin vapor that did not elicit an UTC, followed by five cycles of relaxed-throat breathing.

Although the conclusions that can be drawn from the study results are limited by the small sample size, which was primarily the result of the COVID-19 pandemic, the results provide sufficient evidence of feasibility and proof-of-concept to warrant further study. Seven of eight participants who received the active treatment achieved a clinically meaningful improvement in the primary outcome measure at both post-treatment assessment points, evidenced by at least 1.3 increase in the LCQ(44). Zero treatment participants worsened on the LCQ. Of the seven with a clinically relevant improvement, four achieved a change in LCQ of over 3.0 points, suggesting a fairly large improvement. In contrast, only two of six sham participants improved by greater than 3.0 points at both post-treatment time points, and two participants worsened. Given the small sample size, it is not surprising the statistical evidence for a difference between the groups on the primary outcome measure (LCQ) was weak. However, if the pattern of results holds true in a larger sample, it should be easy to demonstrate statistical significance.

There was strong evidence of a difference between the groups on cough frequency during UTC testing, that favored the treatment group; however, we acknowledge this is an unvalidated measure. Additionally, one participant in the sham group (blue line in Figure 3), whose cough frequency dropped considerably at PT1 but went up at PT2, likely contributed heavily to the difference found in this measure. It is interesting, however, to note that cough frequency did not rise at PT2 in any of the treatment participants and, in fact, in two participants it declined between PT1 and PT2. This may suggest better stability of the treatment response with the active treatment, but additional study is needed to make this determination.

Because all participants had already completed a trial of BCST with minimal benefit, we did not anticipate an improvement with the sham group. While it is possible this improvement was due to a placebo-effect, it is also possible the daily tracking log of adherence with breathing practice and success with cough-suppression, turned in at each session, increased patient motivation and adherence with BCST strategies, which contributed to an improvement in outcomes. Repeated capsaicin exposure, albeit at a subthreshold dose, may have also resulted in some level of desensitization. Current literature describes tachyphylaxis in single-inhalation cough challenges lasting no more than a few minutes (48–50), which lead us to hypothesize that repeated exposure alone would not result in long-term desensitization; however, desensitization as a result of repeated exposure to subthreshold doses has not been investigated, making the possibility of long-term desensitization in this scenario unknown. Similarly, the treatment effect observed in patients who received the active treatment may also have been due solely to desensitization of airway c-fibers from repeated capsaicin exposure. In other words, active cough suppression may have provided no benefit to the treatment. A similar desensitizing effect has been shown in studies looking at the effect of nasal capsaicin spray for non-allergic rhinitis (51, 52), as well as topical capsaicin for neuropathic pain (53). Similarly, Ternesten-Hasseus et al. showed a reduction in cough-reflex sensitivity and cough symptoms in patients with RCC when taking pure capsaicin via oral capsule daily for four weeks (54). A study contrasting capsaicin exposure alone to capsaicin exposure paired with cough suppression is needed to answer this question. Even if capsaicin is found to be solely responsible for the treatment effect, it is likely that cough suppression will continue to be a valuable part of the treatment because it will improve tolerability of the treatment, given repeated coughing from capsaicin exposure may be considered intolerable.

Limitations

We acknowledge several limitations in this study. The most obvious is the small sample size, which resulted in minimal power in our statistical analyses. Our sample size goal was 25 participants but this was prevented by the COVID-19 pandemic, which halted the study from March to August 2020. Although we were able to resume the study in August 2020, with significant mitigation efforts, most potential participants were reluctant to enroll. Rather than waiting to gather a larger sample size, we chose to analyze our data to determine if we should wait to collect additional data or refine the protocol for a second pilot study. Given the data provided sufficient proof-of-concept and feasibility to justify continued study, but also revealed the need to revise the protocol in a second pilot study before proceeding with a Phase II clinical trial, we chose to close the study.

The limited treatment course (i.e., 6 sessions) as well as a change in treatment intensity per session (i.e., doubling the exposures per session) for the final treatment participant are also limitations of the study. The participants who improved with the treatment began reporting a change in daily cough severity at approximately the 4^th or 5^th session and progressed further by the 6^th session. Given their cough was not fully resolved at that point, it is likely these patients would have made further progress with additional sessions.

Potential unblinding of the active treatment participants is also possible, given they likely perceived the gradual change in capsaicin concentration across the treatment, and also likely felt transient throat irritation (37). This may have contributed to a greater placebo-effect in the treatment group over the sham group.

Lastly, UTC testing, the only measure which revealed strong evidence of a greater improvement in the active treatment group than the sham group, has not been validated or tested for reliability, which poses an additional limitation.

Future Directions

While strong evidence of a difference between the two treatments was only found on one non-validated measure, both groups did show significant improvement in self-report measures compared to baseline measures. Given every participant in the study had exhausted all conventional evidence-based treatments currently available, this is a remarkable finding that justifies further investigation. To specifically explore the impact of repeated capsaicin exposure coupled with cough suppression, potential changes to the research design include adding a “true” placebo group that receives saline instead of a sub- threshold dose of capsaicin, a capsaicin-only group (without cough suppression), and extending the treatment length. It is likely that a one-size-fits-all approach, as investigated in this study, is not the most optimal approach to CDT. Rather, we suspect that optimal treatment intensity will be dictated by patient response.

Conclusions

Cough desensitization treatment (CDT), examined in this study, had clear benefits for the majority of patients with RCC in our study (6/8). Although there are several limitations to this study, the results provide sufficient evidence of feasibility and proof-of-concept to warrant further study.

Figure 4.

Sum of urge-to-cough (UTC) scores at baseline, 1-week post-treatment, and 3-weeks post-treatment per group.

Highlights:

• Patients with refractory chronic cough were treated with repeated exposure to aerosolized capsaicin or a sham treatment

• Exposure to repeated inhalations of aerosolized capsaicin via nebulizer reduced cough symptoms on patients with refractory chronic cough

• Exposure to repeated inhalation of aerosolized capsaicin may prove to be an effective treatment for refractory chronic cough. Additional research is needed.

List of Abbreviations:

BCST

behavioral cough suppression therapy

CDT

cough desensitization treatment

CCT

cough challenge testing

CCB

cough-control breathing

IRB

institutional review board

LCQ

Leicester Cough Questionnaire

PT

post-test

RCC

refractory chronic cough

RP-HPLC

reverse-phase high-performance liquid chromatographic

TRPV1

receptor potential vanilloid type 1

UTC

urge-to-cough

Availability of data and material:

The first author holds all data and materials, which can be made available upon request.