We conducted a single-centre, prospective, randomised, parallel-group, placebo-controlled study with an open-label medical device. Ethical approval (HCP/2013/159) was granted by the Clinical Research Ethics Committee of the Hospital Clinic Barcelona, Barcelona, Spain (Chairperson Dr. X. Carné Cladellas) on 16 September 2013. The study was authorised by the Spanish Agency of Medicines and Medical Devices (AEMPS) on 24 March 2014 and registered at ClinicalTrials.gov (NCT02117908) on 11 April 2014. The study was conducted in accordance with the Declaration of Helsinki, and written informed consent was obtained from all patients.

We considered patients over 18 years of age scheduled for CT-guided RFA of a malignant pulmonary tumour under PSA who were capable of undergoing the tests and explorations required for the study as eligible. The exclusion criteria were any radiological contraindication for RFA, lung nodule biopsy just prior to RFA, intercurrent disease, inability to understand the procedure or intolerance to the CPAP test. We also excluded patients scheduled for additional pulmonary RFA.

We randomised the patients to receive CPAP + 4 or a modified mask for placebo CPAP (Sham-CPAP). We used a computerised randomisation list prepared by an independent investigator for the allocation. The randomisation sequence was generated using the SAS® 9.2 PROC PLAN procedure (SAS Institute, Inc, Cary, North Carolina) with a 1:1 allocation using a random block size of 4. The principal investigator, who was unaware of the randomisation sequence, included cases that met the inclusion criteria in the electronic case report form (eCRF) where the randomisation number appeared indicating CPAP + 4 or Sham-CPAP. We employed a CPAP device (ResMed S9, ResMed Ltd., Bella Vista, New South Wales, Australia) with a fixed pressure of 4 cmH2O and a full-face mask (Ultra MirageTM NV full-face mask, ResMed Ltd., Bella Vista, New South Wales, Australia) in the CPAP + 4 group. The Sham-CPAP group received a placebo treatment described by Farré et al [12]. We launched the CPAP or Sham device when PSA was started. The patients received oxygen at a flow rate of 3 L min−1. The patients and radiologists were blinded to the group allocations.

Figure 1 presents an outline of the study. We established the CPAP + 4 cancellation criterion as any clinically relevant complication attributable to the device.

Study protocol outline. CPAP, continuous positive airway pressure; CT, computed tomography; CT0, CT scan before the start of the procedure and PSA once the patient was positioned; CT1, CT scan at the end of the procedure, electrode inserted, before stopping PSA; CT2, CT scan after removal of the ablation electrode and PSA, patient in supine decubitus, awake, CPAP mask removed; CT3: control CT scan at 24 h prior to discharge; Day 0-, day of the pulmonary RFA; PACU: post-anaesthesia care unit; PSA, procedural sedation and analgesia; RFA, radiofrequency ablation

Respiratory monitoring included airway pressure, respiratory airflow, end-tidal carbon dioxide (ETCO2), respiratory rate (RR) (Fig. 2), and peripheral oxygen saturation (SpO2). All sensors were connected to an analogue-digital converter (DATAQ® Instruments) and recorded and analysed using WindAQ Data Acquisition (DAQ) software. The cut-off points were as follows: T0, basal (patient positioned); TR (1-6), end of each ablation cycle (roll-offs); T1, end of RFA, radiofrequency electrode inserted, PSA in progress, patient in RFA position and T2, RFA completed, radiofrequency electrode extracted, PSA shutdown, CPAP or Sham device removed and patient in the supine position. The mean values of the pressure (cm H2O), flow (L s-1), ETCO2 (kPa) and SpO2 (%) were calculated at 10 s around the cut-off point. The complete record (T0-T2) was analysed for episodes of apnoea (absence of respiratory flow for 10 s or more), hypopnoea (30% reduction in respiratory flow for 10 s or more), hypoxaemia (SpO2 < 90%) or hypercapnia (ETCO2 > 5.3 kPa). The percentage of recording time in which SpO2 was below 90% (CT90), and ETCO2 was higher than 5.3 kPa (40 mmHg) was calculated.

Wiring diagram of the mask for CPAP + 4 or Sham-CPAP. A pressure sensor (Honeywell S&C, Minneapolis, USA) was connected to one of the mask’s inlets, and a pneumotachograph (ResMed Ltd., Bella Vista, New South Wales, Australia) was placed between the CPAP tube and the mask. CPAP pressure and respiratory flow values were measured and recorded to verify nasal pressure and ensure the correct estimation of apnoea and hypopnoea episodes. In the other inlet of the mask, a probe connected to a capnograph (Capnostream® 20 P Oridion, Jerusalem, Israel) was placed to record EtCO2 and respiratory rate. CPAP + 4, continuous positive airway pressure of 4 cm H2O; EtCO2, end-tidal CO2; O2, oxygen; Sham-CPAP, modified mask for placebo CPAP

Patient monitoring also included electrocardiography using a 3-lead system (Philips IntelliVue MP50, Soma Technology, Inc. Bloomfield, USA), continuous non-invasive haemodynamic monitoring (Nexfin®, BMEYE, Amsterdam, The Netherlands), tympanic temperature (tympanic temperature sensor TTS-400, Smiths Medical, Minnesota, USA), Bispectral Index (BIS), (BISTM, Medtronic, formerly Covidien, Minneapolis, USA), Ramsay sedation scale [13] and the visual analogue scale (VAS), (0–10). The monitoring parameters were recorded at each cut-off point.

A Somatom® Emotion Duo CT scanner (Siemens®, Erlangen, Germany) was used in this study. The CT protocol for RFA performed at our hospital remained unchanged. Each patient underwent four CT acquisitions (Fig. 1), with 5 mm thick reconstructions using H80 and H30 filters, as well as 1.5 mm thick reconstructions using an H30 filter. All reconstructions were saved in picture archiving and communication systems (PACS) for further analysis. Data analysis was performed using the Pulmo 3D SyngoVia® software (Siemens®, Erlangen, Germany).

We recorded the duration of the pulmonary RFA procedure, maximum radiofrequency generator power (W), generator impedance (Ohm) and thermal ablation generator time (s). Patients were positioned in the supine, prone, or lateral decubitus position with their arms extended depending on the most appropriate imaging approach. Following asepsis and notching, local anaesthetic infiltration was performed at the puncture site. The radiofrequency generator, which uses a feedback system based on electrical impedance, automatically determines the end of the ablation cycles (roll-offs). Two types of ablation generators and electrodes were used: the RF3000TM radiofrequency generator and the LeVeen CoAccessTM electrode system with a coaxial needle placement system, a self-expanding electrode, and the option to use three sizes (3, 3.5, and 4 cm), (Boston Scientific, Natick, Massachusetts, USA) or the Cool-tipTM RF ablation generator and the Cool-tipTM RF ablation system with a single needle system, cooled tip and the possibility of using a 2 or 3 cm effective tip (Medtronic, formerly Covidien, Minneapolis, USA).

For PSA, we administered a target-controlled infusion (TCI) (Fresenius Kabi Orchestra® Base Primea Bad Homburg, Germany) of remifentanil supplemented with TCI of propofol and an intravenous bolus of ketamine (5–10 mg) as determined by the anaesthesiologist. PSA was initiated after TC0 in conjunction with Sham-CPAP or CPAP + 4 and stopped after TC1. If the patient experienced episodes of hypoxaemia (SpO2 < 90%), the supplemental oxygen flow was increased. Increases in ETCO2 > 5.3 kPa were tolerated as long as there were no clinical repercussions. Adverse respiratory events were managed based on the clinical judgement of the attending physicians. Incremental doses of 5 mg IV urapidil were administered if the mean arterial pressure (MAP) increased above 20% and incremental doses of ephedrine (5 mg) or phenylephrine (50 μg) were administered if MAP fell below 20% of baseline. The infusion time delivered by the TCI pump and anaesthetic drug doses were recorded. After completion of the procedure, patients were monitored in the post-anaesthesia care unit (PACU) and transferred to the hospital ward. Discharge was expected within 24 h of the CT3 scan.

The primary outcomes were the number (%) of subjects reporting at least one serious adverse event (SAE), the Classification for complications from the Cardiovascular and Interventional Radiological Society of Europe (CIRSE) classification system for complications of interventional radiology [14] the Clavien-Dindo classification of surgical complications [15], hospital stays and readmissions. Secondary outcomes included adverse events (AEs), intraoperative episodes of hypopnoea or apnoea, minimum SpO2, CT90, maximum ETCO2, percentage of time ETCO2 > 5.3 kPa, airway interventions, and the local radiological efficacy of RFA (a complete tumour ablation margin, a minimum halo thickness of 5 mm and an increase in tumour size from CT0 to CT3). SAEs and AEs were reported and coded according to the MedDRA (https://www.meddra.org/how-touse/basics/hierarchy) and MDCG 2020-10/1 guidelines (https://ec.europa.eu/health/system/files/2020-09/md_mdcg_2020-10-1_guidance_safety_reporting_en_0.pdf). SAE and AE reporting were conducted from recruitment to one-month follow-up.

No previous studies have applied the proposed anaesthetic technique with the same type of target population. Similar studies [16, 17] have used sample sizes of approximately 20 patients per group. A sample size of 22 patients per group was calculated considering a 10% loss or no consent.

The statistical analysis plan was approved by the authors before the analysis began. Intention-to-treat (ITT) and per-protocol (PP) analyses were performed. Continuous variables were reported as mean ± SD or median (IQR). Categorical variables are presented as number of cases (n) and percentages (%). Comparison of continuous variables was performed using the Student’s t-test or Mann–Whitney test, as appropriate for parametric and non-parametric variables. Categorical variables were compared using Fisher’s exact test. For binary variables, the odds ratio (OR) and their 95% confidence interval (95%CI) were estimated to assess the risk of complications using a logistic regression model. If the risk could not be estimated, differences between treatments were compared using Fisher’s exact test.

Longitudinal continuous variables were analysed using mixed models for repeated measures (MMRM). Statistical analysis was performed using SAS version 9.4 or higher (SAS Institute Inc., Cary, NC, USA) and statistical significance was established at the two-sided 5% level.