Air leak is a common complication of lung resection, leading to longer hospital say and increased costs1. Prolonged air leaks have been studied in terms of patient background risks2,3,4, intraoperative repairs5,6, postoperative repairs1,7, and differences among drainage systems8. The water submersion test (W-test) is widely used for detecting intraoperative air leaks but has limitations: [1] subjective judgment; [2] difficulty in detection due to small surgical space in patients with emphysema, obstructive pulmonary disease, or intrathoracic adhesions; [3] potential vagal responses from cold water; [4] poor visibility due to cloudiness when using saline; [5] tissue damage from distilled water9; [6] oversight in insufficient lung expansion10,11; and [7] administered water flowing to the unaffected side in patients with interthoracic communication12. These drawbacks are more pronounced in the video-assisted thoracoscopic surgery (VATS)10,11. However, to the best of our knowledge, no other method currently matches the W-test for detecting intraoperative air leaks. From the experience of smelling gas anaesthetics in air leak cases, we assumed that examining gas components in the thoracic cavity could allow air leak detection and reduce oversight. We investigated the relationship between intraoperative alveolar air leakage by the W-test and the measurement of intrathoracic gas concentrations (G-test). This study's hypotheses were1: all gas concentrations increase in positive patients with a positive W-test, and2 if gas concentrations increase in patients with a negative W-test, it indicates an overlooked air leak.

This single-centre prospective study was presented in accordance with the STROBE Reporting Checklist and registered in the UMIN Registry (UMIN000016006, 19/12/2014). It was conducted in accordance with principles of the Declaration of Helsinki (revised 2013), and was approved by the Ethics Committee of the Joetsu General Hospital (J-77). All patients provided written informed consent prior to study participation.

This study included all patients who underwent elective lung resections at Joetsu General Hospital (Joetsu, Niigata) between November 2013 and May 2015. The exclusion criteria were: local anaesthesia, American Society of Anaesthesiologists Physical Status 4, interstitial pneumonia, thoracotomy, sternotomy, bronchial stump air leak, any thoracic surgery except lung resection, and missing data.

All patients underwent chest X-ray, computed tomography, blood tests, urine tests, electrocardiograms, and pulmonary function tests. Cerebral magnetic resonance imaging and positron emission tomography were performed in patients with lung cancer. Patients aged 80 years or more underwent preoperative echocardiography and exercise stress testing.

Anaesthesia was managed using a combination of general anaesthesia and thoracic epidural anaesthesia. Following placement of the epidural catheter in the awake patient, general anaesthesia was initiated with a bolus administration of propofol (1.0-2.0 mg/kg) and fentanyl (50-100 µg), followed by a continuous infusion of remifentanil (0.25-0.5 µg/kg/min). After a bolus of rocuronium (0.6-1.0 mg/kg) to achieve muscle relaxation, the airway was secured using a double-lumen endotracheal tube (35-37 Fr. Blue line endobronchial tube for left lung, PORTEX, Smiths Medical, USA) for single-lung ventilation. After positioning the patient laterally, bronchoscopy confirmed the endobronchial tube position: the affected side tube was clamped, and auscultation confirmed the position. Single-lung ventilation was maintained at ECO 35-45 mmHg and a maximum airway pressure of 18 cmHO using pressure control mode. Positive end-expiratory pressure was not applied. General anaesthesia was maintained with desflurane (Des), remifentanil, and rocuronium. Des was adjusted between 40 and 60 using the bispectral index (BIS Brain Monitoring System, COVIDIEN, USA).

Partial lung resection was performed using three-port VATS, with an additional 5-mm port for lobectomy and segmentectomy. The ports utilized an AESCULAP Flexibile trocar (B-Braun, Germany). A thoracoscope with a 30-degree, 5- or 10-mm camera was used. All surgeries were performed by the same surgeon.

The W-test and G-test were performed according to the procedure described below (Fig. 1).

For the measurement of intrathoracic gas concentrations, the inhaled gases were fixed (oxygen (O) 5 L/min, Des 5.0%, FiO equivalent to 95%) and airway pressure maintained at 15 cmHO for at least 5 min before measurement. Thoracoscopic ports were placed, the thoracic cavity was observed, and the affected lung was confirmed to be sufficiently collapsed and unventilated. Intrathoracic gas components were measured before lung parenchyma operation. The sampling tube of the multi-gas channel of the anaesthesia gas monitor (Lifescope, NIHON KOHDEN, Japan) was removed from the anaesthesia machine. An extension tube (50 cm, Suffed, TERUMO, Japan) of a 12 Fr. suction catheter (NIPRO, Japan) and a tracheal suction kit (MD-33050, Sumitomo Bakelite, Japan) for water trapping were connected to the sampling tube. A 12 Fr. suction catheter was inserted into the thoracic cavity from one port, and intrathoracic gases were continuously collected and measured (Fig. 2A). Baseline levels of Des. (%), O (%), and carbon dioxide (CO) (mmHg) were recorded 1 min after starting intrathoracic gas sampling (Fig. 2B). During gas collection, the surgeon manually blocked other ports to prevent air inflow into the thoracic cavity.

Warm distilled water (1 L) was administered to the thoracic cavity after specimen removal. After re-inflating the operated lung (peak pressure 15 cmHO), the location of any fistula was identified. Bronchial stump air leaks were excluded in this study. Alveolar air leaks were rated as follows: 0, none; 1, mild, characterized by non-coalescent single bubbles; 2, moderate, characterized by intermittent coalescent single bubbles; and 3, severe air leak with coalescent bubbles or multiple air leaks. Rating was based on surgeon-anaesthesiologist agreement. After the W-test, a G-test was performed without fistula repair.

After aspirating distilled water, both lungs were continuously re-pressurized at 15 cmHO. Intrathoracic gas concentrations were measured then as in the G-control after 1 min of pressurization. We assumed a G-test positive if their G-test values were higher than the G-control value at least two gases. If the initial G-test was positive but the W-test was negative, only the W-test was repeated for the second test to minimize potential lung damage from further inflation.

For mild leaks, the fistula was covered with a polyglycolic acid sheet (Neoveil, Gunze, Japan) and fibrin glue. For severe leaks, the fistula was sutured with 4 - 0 Prolene (Ethicon, USA) and covered with a polyglycolic acid sheet and fibrin glue.

A 20 Fr. chest tube was used with a Thopaz (Medela, Switzerland) and controlled at - 10 cmH0. The tube was removed when drainage was < 200 mL/day, and air leak was < 10 mL/min for over 12 h.

The primary endpoint was the association between intraoperative alveolar air leakage assessed by the W-test and intrathoracic gas concentrations evaluated by the G-test.

Patient and surgical characteristics, as well as follow-up parameters, were recorded from the preoperative period to 3 months after surgery. This included age, sex, past medical history, smoking history, body mass index, estimated glomerular filtration rate, spirometry-measured respiratory function, diagnosis, diseased side, emphysema, procedure type (partial resection, segmentectomy, lobectomy), intraoperative bleeding, W-test, G-test, operative time, chest tube duration, and complications (including prolonged air leak ≥ 5 days). Complications were defined as any deviation from the expected postoperative course and graded according to the Clavien-Dindo classification. Diagnoses of interstitial lung diseases were confirmed based on a combination of clinical and radiological findings following the 2011 guidelines of the American Thoracic Society.

This prospective study investigated data collected from the Joetsu General Hospital. Surgical procedures, anaesthesia management, and data measurements were performed by the surgeon and four anaesthesiologists.

The sample size was calculated based on previous studies and our experience with the frequency of prolonged air leak. We expected to include 25% of patients with intraoperative alveolar air leaks in this study. Considering a statistical power of 80% and significance level of 5% (one-arm), the sample size was estimated to be 95. Expecting a dropout rate of 20%, we aimed to recruit 114 patients.

Continuous variables are presented as means with interquartile ranges (IQRs) for normally and non-normally distributed data, respectively. Categorical variables are presented as numbers (percentages). Air leaks that occurred after returning to the ward were not included in the analysis. The relationship between the number of bubbles in the W-test and the intrathoracic gas concentration in the G-test was evaluated using the correlation coefficient.

The prognostic value of the intraoperative air leak test was evaluated using its sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and diagnostic accuracy to predict the outcome in the dataset. These analyses were performed based on the results of the first and second W-test and each intrathoracic gas measurement of the G-test. In this analysis, the G-test was considered positive only when all three gas types were elevated. Air leaks were defined as present if the W-test yielded a positive result or if air leaks were observed immediately after wound closure before the patient left the operating room. Air leaks that occurred after the patient returned to the ward were not included from this analysis. Based on these results, we calculated the sensitivity, specificity, PPV, and NPV of each test.

All statistical analyses were performed using JMP version 16.0 (SAS Institute Inc., Cary, NC, USA).