Dyspnoea patterns in patients with advanced diseases: a functional MRI feasibility study protocol

Abstract

Introduction

Dyspnoea is an existentially burdensome symptom in patients with advanced and progressive diseases such as cancer, chronic obstructive pulmonary disease (COPD) and advanced heart failure. Recent studies have highlighted that symptomatic treatment of dyspnoea is often ineffective and may depend on the underlying disease. Immersive virtual reality (IVR) has emerged as a ‘digital therapeutic’ for conditions such as pain, anxiety, and dyspnoea. Brain functional MRI (fMRI) offers the opportunity to identify distinct patterns of dyspnoea. Current findings are mainly limited to healthy volunteers, but clinical data from patients with life-limiting conditions are needed. The aim of this study is to assess the feasibility of identifying dyspnoea patterns in different life-limiting conditions using fMRI and IVR.

Methods and analysis

This is an observational monocentric feasibility study, conducted in a tertiary university centre. Healthy volunteers and patients diagnosed with advanced cancer, COPD, or heart failure and suffering from persistent dyspnoea will undergo an fMRI of the brain using IVR. The primary outcome of feasibility will be evaluated using descriptive statistics. Secondary outcomes include analysis of fMRI patterns of dyspnoea across populations, patient-reported burden of participation, and correlation between dyspnoea and psychological symptoms. These preliminary data will help determine the sample size required for a future study evaluating differences in dyspnoea patterns. Exploratory comparison between the characteristics of all four groups will be assessed with Fisher’s test (for proportions) and either independent Student’s t-test or Mann-Whitney test, depending on distribution. Correlations between variables will be tested using the Pearson’s correlation coefficient. Statistical analysis will be performed using STATA.

Ethics and dissemination

This study protocol received ethical approval on 23 April 2025 from the Commission cantonale d’éthique de la recherche in the Canton of Geneva, Switzerland. The identification number is 2024-02289. Submission to peer-reviewed journals and presentation in international congresses for the dissemination of the study findings are planned.

Trial registration number

Clinical Trials number is NCT07319039; Pre-results.

STRENGTHS AND LIMITATIONS OF THIS STUDY.

• Assessment of preintervention and postintervention patient-reported outcome measures.

• Comparison of the feasibility of intervention in patients with different life-limiting conditions suffering from dyspnoea and healthy volunteers.

• One limitation of this study is its small sample size, as it is designed as a feasibility study.

Introduction

In 2012, the American Thoracic Society (ATS) confirmed an earlier ATS definition of dyspnoea (breathlessness, air hunger, shortness of breath, difficult breathing) as ‘a subjective experience of breathing discomfort that consists of qualitatively distinct sensations that vary in intensity’.¹ Dyspnoea can be continuous, episodic, or both, with episodic dyspnoea being defined as a ‘severe worsening of breathlessness intensity or unpleasantness beyond the usual fluctuations in the patient’s perception’.² Episodes are time limited (seconds to hours) and occur intermittently, with or without underlying continuous dyspnoea. Recently, the updated version of the International Classification of Diseases-11 introduced a classification of dyspnoea based on the duration: ‘acute’ (hours to 3 weeks), ‘subacute’ (3–8 weeks), and ‘chronic’ (>8 weeks) dyspnoea.³ This modification was motivated by suggestions from the palliative care research community and in support of the European Respiratory Society, aimed at avoiding the nihilistic term ‘refractory dyspnoea’.^{4 5} Furthermore, dyspnoea that persists despite optimal pathophysiological treatment is defined as persistent dyspnoea.⁶

Relevance of the condition

The experience of dyspnoea is determined by many physiological, psychological, and social factors and may severely affect the quality of life of the patients, who often suffer from existential anxiety and panic as a result.^{7 8} Moreover, family members and other caregivers may also be adversely affected.⁹ Dyspnoea affects a large proportion of patients with various incurable, advanced and life-limiting (palliative) conditions, such as cancer, chronic obstructive pulmonary disease (COPD) and heart failure, with prevalence increasing with disease progression and severity.^{10 11}

Therapeutic dilemma

Once the treatment of the underlying disease and its concomitant pathophysiology does not provide sufficient improvements, the treatment of dyspnoea is called symptomatic. It aims to provide relief from the distressing sensation without modifying the cause of dyspnoea. Unfortunately, these options are limited and often insufficiently effective, leading to chronic, lifelong suffering.^{7 12} Moreover, it has become evident that the use of opioids, which is the key textbook recommendation for the pharmacological symptomatic treatment of dyspnoea, is to be questioned, especially in patients with underlying non-cancer conditions such as COPD and heart failure.^{13 14}

Potential of the study procedure

The reason for the above-mentioned absence of a one-size-fits-all approach for the treatment of dyspnoea can easily be explained by the hypothesis that dyspnoea is conveyed by sensory and central neuronal pathways that differ, for example, due to the underlying disease.¹ Accordingly, the ATS recommends that ‘research is needed in the areas of neuromodulation, neuroimaging and central processing of dyspnoeic sensations’.¹ Yet, since the publication of this recommendation, such neuroimaging research in the field of palliative medicine, aiming at the differentiation of dyspnoea patterns in patients with incurable, advanced and life-limiting conditions has been nearly non-existent. One study involving patients with COPD found that the “cognitive” prefrontal cortex and the ”emotional” limbic system may respond differently when exposed to stimuli related to dyspnoea.¹⁵ Trials recruiting healthy volunteers showed that the provision of morphine or other opioids can relieve the sensation of ‘air hunger’ but not the distress perceived by increased ‘breathing effort’ in dyspnoea.^{16 17} It is known from studies with healthy volunteers that a variety of cerebral structures are involved in the mediation of dyspnoea (ie, anterior insular activation, limbic and paralimbic activations in the cingulate gyrus and amygdala, cerebellar midline regions).¹¹ To our knowledge, no trial aimed at comparing pathways in patients with different life-limiting conditions has been conducted.

Choice of the accompanying immersive virtual reality (IVR) intervention

In recent years, the healthcare industry has seen the development of a ‘digital therapeutic’. IVR is an interactive computer simulation that allows a person to be completely immersed within a three-dimensional virtual environment, creating a realistic sense of presence in the virtual world. Some start-ups have developed IVR therapies offering medical hypnosis sessions, games that respond differently to distract and reorientate the attention, biofeedback, and mindfulness meditation. The main studies and clinical applications have focused on pain and anxiety related to medical or diagnostic procedures, such as functional MRI (fMRI).18,21 IVR as an intervention to reduce pain in palliative care patients is of increasing interest because of its potential to alleviate suffering from pain in these patients.²² Interestingly, cerebral neuronal activation patterns of pain and dyspnoea have been found to show similarities, supporting our choice to combine fMRI and IVR in dyspnoeic patients.^{1 23} Some studies on dyspnoea and virtual reality (VR) interventions in different clinical settings were recently published.^{24 25} In our study protocol, IVR will provide a soothing scenario and the fMRI data will be appraised to identify potential preintervention and postintervention changes in the patterns of activated central structures.

Methods and analysis

As we believe that dyspnoea is experienced and centrally processed in different ways among patients, related to (1) the underlying condition, (2) the triggers for dyspnoeic episodes and (3) the differences in patient-reported multidimensional assessment items (ie, perceived severity vs discomfort vs effort), this study aims to assess the feasibility of identifying different patterns of dyspnoea via fMRI technology. The procedure is augmented by the implementation of an IVR intervention designed to enhance positive emotional aspects of dyspnoea. In the long term, we hope to contribute to a more differentiated and individualised understanding of dyspnoea to facilitate the development of a more effective, tailored pharmacological and non-pharmacological interventions for ‘persistent’ dyspnoea.

Study hypothesis

We hypothesise that fMRI and associated study interventions are feasible for patients suffering from persistent dyspnoea associated with COPD, heart failure or advanced cancer.

Primary objective

To determine the feasibility and acceptability of the study procedure (fMRI and IVR).

Secondary objectives

To identify distinct fMRI signals of dyspnoea according to the underlying condition and to describe different types of dyspnoea patterns through fMRI, IVR and clinical parameters, including patient-reported outcome measures (PROMs).

Study design

This is an observational monocentric feasibility study, conducted in a tertiary university centre. The study started on 1 August 2025, with a planned duration of 18 months and is scheduled to conclude on 1 February 2027.

Participants

We aim to recruit a minimum of 16 participants: 4 patients with heart failure, 4 patients with COPD, 4 patients with cancer and 4 healthy volunteers. The inclusion and exclusion criteria are listed in table 1.

Table 1. Inclusion and exclusion criteria for patients.

Inclusion criteria	Exclusion criteria
Age≥18 years	Diagnosed neurological disorders, dementia (frontotemporal dementia, Alzheimer’s disease, etc), brain pathology (tumour, stroke, Parkinson’s disease), epilepsy
Outpatient at the palliative care clinic of the Geneva University Hospitals, Switzerland	Diagnosed psychiatric disorder (severe depression, severe anxiety, psychosis, others) or receiving antipsychotic treatment that is deemed to be a contraindication for fMRI
Underlying condition: Heart failure stage NYHA III–IV COPD with dyspnoea mMRC Scale grade 3–4 Cancer with either primary or secondary pulmonary location	Presence of claustrophobia, acrophobia, photophobia, severe hearing loss and/or severe visual deficit
Stable clinical condition (absence of acute cardiac, respiratory and/or neurological failure leading to hospitalisation in the previous 4 weeks)	Contraindication to fMRI and/or IVR (migraines, photosensitive epilepsy, vertigo)
Persistent dyspnoea (>3 weeks despite adequate and maximal medication according to the underlying condition)	Inability to lie down flat (supine position)
Dyspnoea at rest >2 and <8 on the NRS 0–10	Dyspnoea at rest ≥8 on the NRS 0–10

COPD, chronic obstructive pulmonary disease; fMRI, functional MRI; IVR, immersive virtual reality; mMRC, modified Medical Research Council; NRS, Numeric Rating Scale; NYHA, New York Heart Association.

Healthy volunteers, who will act as a control group, will be recruited if aged ≥18 years, without respiratory symptoms (dyspnoea, cough, wheezing), and with a smoking history of <10 pack-years. Other exclusion criteria will be similar to those defined for patients eligible for recruitment.

Patients will be recruited at the outpatient palliative care clinic of the Geneva University Hospitals (single centre) through a patient medical record screening and a pre-verification of the inclusion and exclusion criteria. We included patients with advanced disease, defined as those with an modified Medical Research Council (mMRC) Score of 3–4 or New York Heart Association (NYHA) class III–IV. To ensure that participants experienced dyspnoea at rest during the fMRI, eligibility was restricted to individuals with a resting dyspnoea score>2 on the 0–10 Numeric Rating Scale (NRS). Patients with an NRS score≥8 were excluded, as clinical experience suggests that these individuals would likely find the fMRI intervention excessively burdensome.

The healthy controls will be recruited from a convenience sampling meeting the inclusion criteria. The age group will match that of the recruited patients. No hierarchical influence from any member of the research team will be exerted.

All eligible participants will be given an information document and a consent form describing the research project. The project leader or her designee will explain to each participant the nature of the research project, its purpose, the procedures involved, the expected duration, the potential risks, and any discomfort that may result. Each participant will be informed that participation in the research project is voluntary and that withdrawal from the research project is possible at any time. Withdrawal of consent will not affect subsequent medical assistance and treatment. The participants will be informed that they should also feel free to ask any questions. Participants should read and understand the information, and they will be given enough time, with a minimum of 48 hours, to voluntarily agree before signing and dating the informed consent form.

After a period of reflection that the participant deems necessary to make an informed decision, written consent will be sought during a subsequent visit to the outpatient palliative care consultation. If the patient accepts, the consent form will be signed and dated by both the project leader or her designee and the patient. The MRI Safety Questionnaire will also be completed at that time. Formal consent, using the approved consent form, will be obtained before enrolment in the research project.

Sample size

This feasibility study aims to test the practical aspects of the research design, to optimise data acquisition methods and identify potential challenges in participant recruitment or adherence. We estimate that a small sample size will be sufficient to provide preliminary insights into the operational feasibility and patient burden associated with the fMRI and IVR procedure.

As no study has yet been conducted on this subject, the sample size required to detect differences in dyspnoea patterns across different patient populations cannot be estimated at this point. This study will allow us to gather important preliminary data on this matter.

We aim to include a minimum of 16 participants (4 patients per underlying disease and 4 healthy volunteers) within the 18-month timeframe for this feasibility study, based pragmatically on the number of patients followed at the outpatient palliative care clinic of the University Hospitals of Geneva.

Intervention

On the day of the study procedures, participants will be welcomed by the research associate in a separate room, close to the fMRI. The intervention will be re-explained in detail and any further questions will be answered. Informed consent will be confirmed. Vital parameters and the first set of PROMs will be recorded at this point (figure 1). Participants will be led to the MRI examination room, where they will lie down on the scanner bed.

Figure 1. fMRI and IVR procedures.Vital parameters: saturation of oxygen, respiratory rate, transcutaneous CO₂ (not during MRI). fMRI, functional MRI; IVR, immersive virtual reality; NRS, Numeric Rating Scale; PROMS, patient-reported outcome measures; VRS, Verbal Rating Scale.

Donning the IVR headset

Before the first set of fMRI, patients will be given MRI-compatible headphones. The IVR head-mounted display by Nordic Neurolab (Bergen, Norway) is then attached to a standard head coil (figure 2). This device will be used within the scope of its European Conformity labelling. The setup takes about 5 min.

Figure 2. Detailed functional MRI and immersive virtual reality (IVR) procedures. (A) Participants lie down on the scanner bed and are equipped with the standard head coil. The Nordic Neurolab (https://nordicneurolab.com) head-mounted display (HMD) is attached to the head coil and adjusted for each participant. This HMD has been used at the Geneva University Hospitals without any adverse indications in previous protocols, such as 2020-02553. (B) We have a virtual replica of the MRI room, allowing the patient to view the same room while immersed in IVR, but without the scanner-bore overhead.

IVR positive distraction

In collaboration with HypnoVR (HypnoVR, Strasbourg, France), we will provide patients with visual immersion in a virtual world and a non-narrative audio. HypnoVR’s solution is certified as a medical device and has been studied in various medical situations. The investigators and the patients can select from a predefined set of virtual environments.

Based on this setup, we can adapt VR immersion for each patient, using a default scenario of immersion into a forest with relaxing, non-narrative audio. If patients do not feel comfortable with the forest scenario, another non-stressful scenario may be chosen (beach or space).

Participants will enter the fMRI while the IVR displays a simulation of the scanning room with an open MRI environment, a feature designed to reduce stress and discomfort associated with entering a confined space. Once the participant has been correctly positioned in the fMRI, a brief acclimation period of several minutes will be provided to allow the participant to become familiar with the environment. Following this adjustment phase, the participant will be asked to indicate readiness, at which point the IVR display will be switched off and the first fMRI sequence will begin. A second fMRI sequence will then be conducted with the IVR system activated, during which participants will be presented with their chosen non-stressful scenario. Throughout the procedure, patients will retain the ability to communicate with the medical staff at all times, as is standard practice during routine examinations. If at any point the patient experiences discomfort due to dyspnoea, both the fMRI and IVR will be immediately stopped, and the patient will be offered the opportunity to take his or her usual rescue medication. Should either the investigators or the patient determine that continuation of the fMRI is no longer possible, the patient will be safely guided out of the scanner and excluded from the study, according to local standard procedures (attached as emergency flowchart in the (online supplemental file 1). Following the fMRI, participants will undergo final data collection.

Questionnaires and measurements

1. Pre-fMRI intervention:

   • Revalidation of the MRI Safety Questionnaire and signature (online supplemental file 2).

   • Dyspnea-12 Questionnaire (0–36) (online supplemental file 3).

   • NRS dyspnoea average, worst and on exertion over a period of 24 hours (0–10).

   • Number of acute dyspnoeic episodes per day (open).

   • Potential triggers of dyspnoea (open).

   • Edmonton Symptom Assessment Scale (ESAS) Tool (0–100) (online supplemental file 4).

   • Hospital Anxiety and Depression Scale (HADS) for Depression Questionnaire (0–21) and HADS for Anxiety Questionnaire (0–21) (online supplemental file 5).

   • Charlson Comorbidity Index (online supplemental file 6).

   • Chronic Respiratory Disease Questionnaire (CRQ) (online supplemental file 7).

   • Oxygen saturation (SpO₂ in %), transcutaneous CO₂ (ptCO₂ in kPa), respiratory rate over 1 min (RR in n/min) and heart rate (HR in bpm).

2. During fMRI (between the two fMRI recordings):

   • NRS dyspnoea ‘right now’ (0–10).

   • Continuous SpO₂ (in %), RR in n/min and HR (in bpm).

3. Post-fMRI (15 min after the end of the second fMRI) (online supplemental file 8):

   • NRS dyspnoea ‘right now’ (0–10).

   • Study burden rated by Verbal Rating Scale (1–5).

   • Claustrophobia questionnaire (online supplemental file 9) post-MRI questionnaire.

   • SpO2 (in %), ptCO₂ (in kPa) and RR in n/min and HR (in bpm).

Outcome measures

Primary outcome

Feasibility will be assessed based on recruitment and retention rate.

• Recruitment rate: proportion of eligible patients who consent to participate, calculated as the number of participants enrolled divided by the total number of eligible patients approached during the recruitment period. A recruitment rate≥50% will be considered feasible.

• Retention rate: proportion of enrolled participants who complete the planned intervention (fMRI, vital parameters and study-associated questionnaires). A retention rate≥70% will be considered acceptable.

These thresholds are based on previous research. A recruitment rate of between 40% and 50% is considered standard in palliative care. Furthermore, an fMRI study reported attrition rates between 15% and 30% in a vulnerable population, providing a conservative estimate.^{26 27} If both thresholds are met, progression to a larger observational study will be considered. All patients with the indicated pathologies who are followed by the outpatient palliative care clinic at the Geneva University Hospitals will undergo prescreening by investigators, and the total number of screened patients will be recorded. Patients who might meet the inclusion criteria will be invited to participate in the study. All patients who sign the informed consent and meet the inclusion criteria will be scheduled for the study.

Data will be collected on the number of participants who withdraw from the study, along with the reasons for withdrawal. The number of adverse events occurring during the conduct of the study and during the MRI examinations will also be documented, including whether such events resulted in a premature exit from the study or an interruption of the MRI procedure that could subsequently be resumed. In cases where the MRI is resumed, detailed information regarding the time required and the procedures undertaken to restart the examination will be recorded. The study aims to achieve completion of all procedures by at least four patients for each group, and investigators will therefore continue recruiting participants within the same pathology group until this threshold is reached within the planned 18-month study period.

Secondary outcomes

Safety measures corresponding to adverse event description, time and duration of event, study phase—questionnaire filling, before, during or after MRI—and the need for rescue medication administration will be registered. In case of need for rescue medication during the study, with or without premature exit, the patient’s health data, the type and the dosage of the rescue medication will also be noted.

The acceptability for the patients is based on the intervention’s perceived burden evaluated on a 5-point Likert scale (1 meaning no burden and 5 the maximum possible burden) registered in the 15 min after the intervention, with a cut-off≤3, meaning that the study was moderately burdensome.

Preliminary exploratory data will be collected about potential differences in dyspnoea activation patterns as displayed by fMRI signals in different groups and how these patterns can be influenced by IVR will be collected through different fMRI parameters. The per cent signal change and latency of response in the blood-oxygen-level-dependent (BOLD) signal change will be collected in relevant brain regions (left precentral gyrus, left middle frontal gyrus, left insula, left amygdala, left hippocampus, left frontal medial orbital frontal cortex, left and right brainstem). Furthermore, the peak amplitude (maximum change in BOLD signal during trigger) will also be measured.

Statistical analysis plan

Statistical analysis will be performed using STATA V.16 (StataCorp, Texas, USA). The primary outcome, feasibility, will be presented using descriptive statistics (mean, SD). Given the small group sizes (n=4 per group), the statistical analysis for secondary endpoints will primarily rely on descriptive statistics and exploratory inferential methods, acknowledging the limited statistical power for detecting significant differences. Comparison between the baseline characteristics of all four groups will be assessed with Fisher’s test (for proportion) and independent Student’s t-test or Mann-Whitney test, depending on the distribution. Correlations between variables will be tested using the Pearson’s correlation coefficient. Continuous variables, including fMRI parameters such as BOLD signal per cent change, latency of response, peak amplitude, functional connectivity and cerebral blood flow maps, will be reported as medians and IQRs to account for likely non-normal distributions. Means and SDs may also be presented for comparison. Categorical variables, such as the presence or absence of significant activation in predefined brain regions, will be summarised using absolute frequencies (n) and percentages (%). Data visualisation techniques such as boxplots, violin plots and heatmaps will be used to illustrate variability within and between groups.

Inferential statistics will be performed exploratorily and with caution, as the small sample size limits statistical power. Non-parametric tests will be prioritised due to the expected non-normality of fMRI data. Comparisons of fMRI parameters between the different groups (COPD, heart failure, cancer and healthy volunteers) will be conducted using the Kruskal-Wallis test. If significant differences are observed, pairwise comparisons will be explored using Mann-Whitney U tests, applying a cautious approach to multiple comparisons. For categorical variables, Fisher’s exact test will be used to assess differences in activation patterns across brain regions.

To evaluate the effect of IVR on fMRI parameters, Wilcoxon signed-rank tests will be applied to compare pre-IVR and post-IVR changes within each group. Additionally, effect sizes (Cohen’s d or rank-biserial correlation) will be reported to provide a meaningful interpretation of observed differences. The relationship between fMRI parameters and patient-reported outcomes (Dyspnea-12, ESAS, NRS dyspnoea scores, anxiety and depression) will be explored using Spearman’s rank correlation coefficient, considering the expected non-linear relationships between subjective symptom burden and neuroimaging measures. Functional connectivity patterns, including the connectivity of the left and right anterior insula with the anterior cingulate cortex, will be assessed descriptively using network visualisation methods, with principal component analysis applied to identify dominant connectivity patterns across disease groups.

For multiple comparisons, the focus will be on effect size estimation and clinical relevance rather than strict p value significance thresholds. While multiple comparison corrections (eg, Bonferroni adjustment) may be overly conservative in this setting, they will be considered where appropriate. However, results will primarily be interpreted with an emphasis on biological and clinical plausibility rather than statistical significance alone.

Data will be accessible directly through the research team, which is reachable via email.

Ethics and dissemination

This research project will be conducted in accordance with the protocol, the Declaration of Helsinki, the principles of Good Clinical Practice, the Human Research Act and the Human Research Ordinance, as well as other locally relevant regulations.^{28 29} The Project Leader acknowledges her responsibilities. This study was approved by the Commission cantonale d’éthique de la recherche in the Canton of Geneva on 23 April 2025 (2024-02289). Study results will be communicated to participants and healthcare professionals in charge of the patients.

Study results will be communicated to participants and healthcare professionals in charge of the patients. Submission to peer-review journals and presentation in international congresses for dissemination of the study findings are planned.