Redistribution of pulmonary ventilation after lung surgery detected with electrical impedance tomography

Regional ventilation of the lung can be visualized by pulmonary electrical impedance tomography (EIT). The aim of this study was to examine the post‐operative redistribution of regional ventilation after lung surgery dependent on the side of surgery and its association with forced vital capacity.

having a higher time resolution makes EIT a favorable method. Since forced vital capacity is expected to be reduced after lung surgery, 5,6 pulmonary EIT might show a reduced ventilation on the side of surgery. Because of the anatomic asymmetry of the lungs and the heart, the left lung receives less ventilation and the impedance change during breathing is lower on this side. 17 Thus, an effect of lung surgery on the distribution of ventilation might be dependent on the side of surgery.
The primary aim of this study was to investigate if an ipsi-or contralateral redistribution of the regional ventilation to the side of lung surgery is present according to pulmonary EIT in spontaneously breathing patients. The secondary aim was whether this redistribution is dependent on the side of surgery. We hypothesized that the ventilation always shifts from the side of surgery to the healthy side. The data were correlated with forced vital capacity.

| ME THODS
This monocentric, prospective, observational cohort study examined perioperative redistribution of ventilation in patients with lung surgery. The medical ethics committee of the University Hospital of Würzburg, Germany approved the study protocol (58/16-ge).
All patients provided written informed consent before inclusion.
The study protocol was registered in advance at ClinicalTrials.gov: NCT02779595.

| Patients
Adult patients undergoing open thoracic surgery or video-thoracoscopic procedures were recruited and followed up between June 2016 and June 2017 at the University Hospital of Würzburg. Although the type of anesthesia, airway management, and intraoperative ventilation were not part of the study protocol, procedures were recorded.
In accordance with the standard operating procedures, general anesthesia was induced with fentanyl, propofol, and rocuronium and maintained with an opioid (remifentanil or fentanyl) and an anesthetic (propofol or sevoflurane) completed with or without an additional epidural regional anesthesia with ropivacaine 0.375%. A double lumen tube or a single lumen tube in combination with a bronchus blocker enabled separate ventilation of both lungs. Pressure controlled ventilation was performed. For post-operative analgesia NSAIDs like metamizole, paracetamol or ibuprofen and opioids such as oxycodone or piritramide, and/or patient controlled epidural or intercostal analgesia were used. For epidural analgesia, a combination of ropivacaine 0.2% and sufentanil 5 µg/mL was used whereas intercostal analgesia was achieved with ropivacaine 0.2% alone.
Exclusion criteria were emergency procedures, re-surgery of hospitalized patients, pre-operative pneumothorax, relevant pre-operative pleural effusion, pleural empyema, planned pneumonectomy, expected post-operative mechanical ventilation, expected hospital stay less than 3 days, pregnancy, and contraindications for the application of EIT (eg, pacemaker). Post hoc, patients were excluded if pneumonectomy was performed or the patients were ventilated longer than 3 post-operative days. If re-surgery was performed, patients were only included to follow-up PPCs.

| Study procedures
EIT, spirometry, and pulse oximetry were executed one day before surgery and on the post-operative days 3, 5, and 7 as long as patients were not discharged from hospital. The respiratory rate was recorded by EIT; dyspnea was determined by a questionnaire.
Respiratory infection, respiratory failure, pleural effusion, atelectasis, pneumothorax, bronchospasm, or aspiration pneumonitis that define PPCs 2 were followed up by examining clinically indicated thoracic x-ray or CT, patient visit, and chart review. Pain scores were measured by Numeric Rating Scale (0-10).

| Measurements
The pulse oximeter served to measure peripheral oxygen saturation non-invasively before performing spirometry (combined spirometer and pulse oximeter, Spirodoc ® , Medical International Research).
For EIT, a silicon belt with 16 equidistantly integrated electrodes was applied around the patients' thorax at the height of their third intercostal space determined in the medioclavicular line with a reference electrode on the abdomen. If necessary, wound dressings were removed and were replaced afterward. The patient was in supine position with a 30° elevated upper body and arms resting beside. After a resting period of 2 minutes, the EIT system (Pulmovista 500 ® , Dräger Medical AG) recorded 2 minutes of spontaneous breathing. Then spirometry was performed while EIT recording was continued. The best value of three attempts regarding FVC and FEV1 was used. Percentages of predicted values were calculated according to Knudson (FVC% predicted and FEV1% predicted ).

| Processing of EIT data
The EIT system recorded changes in pulmonary electrical bioimpedance. Data were analyzed with special software (Dräger EIT Data Analysis Tool 6.1, Dräger Medical AG). A low-pass filter (30 bpm)

Editorial Comment
Regional ventilation in the lung can be redistributed after lung resection surgery. In this study, electrical impedence tomography was used to assess regional ventilation after lung surgery. This imaging modality identified some postoperative ventilation reductions and particularly for one side.
excluded cardiac and perfusion-related impedance changes. The software transformed EIT data into images of an ellipsoid shape that were reconstructed into a 32 × 32 pixel matrix. The EIT image matrix represented the impedance change relative to a baseline set at the expiration of tidal breathing at rest. To analyze breathing at rest, the impedance changes from baseline to inspiration were averaged over 1 minute of normal breathing (analysis at rest, minute image of breathing with tidal volumes at rest). For analyzing forced breathing, the impedance change from baseline to maximal inspiration before forced expiration The calculation of COVx and COVy is described in our previous work. 18 To allow the summation of EIT images independent of absolute impedance and breathing effort, each matrix of the tidal images was normalized to the same global impedance change. Thus, the sum of all 32 × 32 intensity changes was the same for every EIT image. The averaged normalized matrices for each time point visualized the pulmonary ventilation for a whole group by a colored contour line graph (Origin Pro 9.1 G, OriginLab Corporation).

| Statistical analysis
COVx (left/right), measured at forced breathing, was calculated to compare the pre-operative values with those of the post-operative For COVx′ on the pre-operative and post-operative day 3, the requirements for a two-way ANOVA with one between-and one within-subject factor were fulfilled. Since several of the remaining variables did not satisfy requirements for parametric statistics, nonparametric methods were used. The corresponding summary statistics show the median for location and quartiles (25th-75th percentile) for the spread. The Mann-Whitney U-test was used to compare between both groups whereas for intragroup comparisons with three or more time points, Friedman tests were performed. Dunn's method was used for post hoc pair-wise comparisons. The association between changes in COVx′ and FVC% predicted was tested using Spearman's rank correlation. P-values < .05 for two-sided testing were considered to be statistically significant. Sigmaplot for Windows Version 11.0.
(Systat Software Inc) was used for statistical calculations.

| Sample size calculation
The sample size calculation was based on the Geisser-Greenhouse (within-subject factor). Furthermore, the sample size is sufficient to detect a mean difference of 1.5 for patients undergoing right-and left-sided surgery with a power of .80 (between-subject factor). The power to detect an effect of the interaction term twice as much as that of the within-subject factor would be > .99.

| RE SULTS
Pre-operative measurement and examinations on the post-operative day 3 were executed in 31 patients. However, five patients withdraw their consent. Thus, 13 patients with lung surgery on the left or right side, respectively, were recruited. Baseline characteristics ( Table 1), characteristics of surgery, anesthesia, and the ARISCAT risk score for post-operative pulmonary complications 2 ( Table 2) were comparable in all patients. In comparison to baseline, FVC% predicted and FEV1% predicted were reduced at least on the post-operative day 3 (Table 3). Seven patients were discharged from the hospital between days 3 and 5, eight patients between days 5 and 7.
The pre-operative ventilation in the right and left half of the EIT summation images during forced breathing were equally distributed, the right side ventilation markedly decreased in patients with right side surgery on the post-operative day 3 ( Figure 1). COV shifted away from the side of surgery in patients after right-sided surgery (Figure 2A,B). Accordingly, COVx′ indicated a contralateral shift of the COV in those patients ( Figure 2C).

| Results from primary statistical analysis
The influence of the factors side (of surgery) and time (from the preoperative to the post-operative day 3) on COVx′ were tested with a two-way ANOVA with one within-subject factor (time) during forced breathing and one between-subject factor (side). Analysis revealed an effect of the factor time on mean COVx′ (1.29 matrix points decrease, P < .001). There was no statistically significant effect of the factor side (0.92 matrix points lower in right sided surgery; P = .066), but an interaction between side and time (P < .007). Therefore, subgroup analyses had to follow.

| Secondary analysis: subgroup analyses and analysis at fixed time
Post-operatively there was a statistically significant effect of side on COVx′ (1.60 matrix points lower in right sided procedures, P = .029).
Only in right-sided surgery a statistically significant effect of time was detected (1.97 matrix points decrease, P < .001), but not in leftsided surgery (0.61 matrix points decrease, P = .425).
Pre-to post-operative changes of COVx′ and vital capacity (FVC% predicted ) were weakly associated during forced breathing (r = .388; P < .05) and breathing at rest (r = .489; P = .011). Table 4 shows the changes of COVx′ and COVy for the whole study period.
COVy did not change compared to pre-operative values. Breathing at rest showed very similar results ( Abbreviations: FiO 2 , inspiratory oxygen fraction; PEEP, positive endexpiratory pressure; VATS, video-assisted thoracoscopy.

| D ISCUSS I ON
In the present study, patients undergoing lung surgery were ex- In this prospective study, FVC and FEV1 were decreased on the post-operative day 3 and did not reach baseline values obtained before surgery within one week. Ergegovac et al also reported decreased FVC and FEV1 that recovered partially within one week in patients after lung resections. In their study, no marked differences in patients with and without respiratory and other complications were present. 19 Decreased vital capacity was found also in patients with cardiopulmonary bypass surgery. In patients with intercostal drainages, the decrease of the FVC was higher than in those with subxyphoid drainages, most likely due to different pain intensities. 20 In our study, expected changes in pulmonary function tests were found in patients serially tested after lung surgery. Pain scores were not different between the groups tested. To our knowledge, this is the first study where an imaging method is used to examine the changes in regional ventilation during the first days after lung surgery. Ventilation scintigraphy Localization of procedure Left (n = 13) Right (n = 13) P

Pain during forced breathing (NRS)
Pre-operative (n = 13/13) 0 (0-0) 0 (0-0) .611 Post-operative day 3 (n = 13/13) 0 (0-4) 2 (0-3) .556 Post-operative day 5 (n = 9/8) 2 (0-2) 2 (0-3) .960 Post-operative day 7 (n = 5/6) 0 (0-0) 0 (0-0 Reasons that may be relevant for more prominent consequences after right-sided procedures observed by EIT measurements are as follows: The anatomic asymmetry of the left and right lung measured by EIT, which is in accordance with data obtained by radionuclide-based imaging of ventilation. 15 Further, operative procedures might not be equal between both groups. Side-dependent differences in the extent of lung traumatization and -resection, effusions and pneumothoraxes have an influence on the measurements and consecutively lead to fewer impedance changes in the EIT. In our study, eight of the patients who underwent left-sided surgery had a pleural effusion post-operatively, compared to six in the other group. The incidence of pneumothorax was different too. Moreover, three patients received bilobectomy on the right side. Distribution of regional ventilation according to pulmonary EIT is depended on body posture and the intercostal level of the EIT belt. 22 In the present study, the belt was placed in the third intercostal space at the  Abbreviation: COV, center of ventilation.

TA B L E 4
Perioperative change of the center of ventilation in patients before and up to seven days after left-and rightsided lung surgery on the right lung than on the left side. These results are in line with the observations recorded by EIT.
To evaluate the regional changes of ventilation evoked by lung surgery, a separate analysis of the impedance changes on either side of the EIT appears suitable. However, in a clinical study on patients with a side-separated ventilation, impedance changes were detected at the contralateral half of the EIT image already before lung surgery. 11 Since radiologically observed shifts of the mediastinum toward the operated side are also described a side-separated evaluation of the regional ventilation is even more difficult. 24 As shown before, 17  Emphysematous bullae, tumor, or COPD cannot be distinguished by EIT since electrical conductivity is similarly reduced in these regions due to a reduced ventilation in all of those cases. 26 The most relevant limitation of EIT in our study is the influence of a pneumothorax. If the lung is not fully inflated because of a residual pneumothorax after extensive sublobar lung resection or lobectomy the lung is not fully attached to the thoracic wall leading to a lack of signal. By EIT, even small pneumothoraxes can be detected according to an experimental swine study and reduction of regional ventilation was confined in the upper quadrant of the affected side in ventral pneumothoraxes. 27 The changes of COVx′ and FVC% predicted showed only a weak association. Besides the explanation by other influencing factors on FVC% predicted , the fixed linear relation between relative changes of thoracic impedance and pulmonary air volume 28 will not necessarily persist after surgery. If the operative procedure modifies the lung tissue, the magnitude of the impedance change per volume might change in the operated lung.
With EIT, the left-right side division was in good agreement with inhaled 81m Kr 15 gas or 99 Tc gas 16,17 radionuclide scans in patients with lung cancer. In those studies, EIT was performed at two electrode planes. The use of EIT in only one plane was done to prevent interference with the surgical wound, but might have affected the results. 22 Other methods such as the hyperpolarized helium-3 gas magnetic resonance imaging of the ventilation 29 are not well-established. In comparison to those methods, EIT is radiation-free, por-  30 Therefore, no true reference method for EIT exists, but the use of an additional imaging method might have eased the interpretation of our results.

| CON CLUS IONS
Pulmonary EIT and spirometry was used to investigate the postoperative redistribution of the regional ventilation of the lung in spontaneously breathing patients after left or right side lung surgery. Only in patients with surgery on the right side, a redistribution of pulmonary ventilation to contralateral side was detected by EIT.
However, forced vital capacity was reduced in both groups and its change was only weakly associated with the shift of ventilation. A global reduction of ventilation in patients after lung surgery might account for some of the changes in forced vital capacity.

CO N FLI C T O F I NTE R E S T S
None.