To our knowledge, this is the first study using EIT imaging to examine the real-time effects of high vs. low FIO2 during BLVR with EBV. We purposely selected a pig model that is CV- [29] because it simulates the ideal conditions for BLVR with one-way valves. In addition, the pig has a severe hypoxic pulmonary vasoconstriction (HPV) reflex that helps in its ventilation-perfusion (V̇/Q̇) balance [30]. Under such conditions, high FIO2 significantly amplifies the acute dynamics of lung volume reduction.
The foundations of BLVR explored the sustained lobar atelectasis to promote lung volume reduction without surgery [31]. In a landmark elegant study in a sheep model of emphysema produced by inhalation of papain, the authors demonstrated alveolar instability induced by FIO2 1.0 for 15 minutes before occlusion and wash out of the surfactant within the target segment. There are similarities between that study and our present study regarding the underlying mechanisms and interventions. Ingenito and colleagues [31] found that the obliteration of dysfunctional regions of the hyperinflated lung could be achieved nonoperatively by a bronchoscopic production of sustained collapse by the following procedures: a) filling target regions with oxygen, benefiting from its high absorption coefficient, which promotes reabsorption atelectasis; b) using an airway occlusion method; in our case with either our inflatable balloon or EBV whereas in their case by the injection of a biocompatible fibrin-based glue acting as a sealant. In the last statement of their seminal paper, the authors encouraged a more highly refined approach to BLVR. Our investigation moves one step further in this direction.
Alveolar collapse has been associated with hypoventilation and airway closure during general anesthesia. The prevention of atelectasis has been attempted with the use of a titrated PEEP [32] or by lower FIO2, particularly before the induction of anesthesia [14–16, 33]. Nitrogen is a break for the gas absorption within the air pocket generated by airway occlusion. It increases the time to achieve atelectasis from as low as 4 to 6 minutes with FIO2 1.0, to 6 to 8 hours with FIO2 0.21 [33, 34]. The driving force of gas absorption rate is a function of the pressure gradient between alveolar and mixed-venous gas concentrations, the absorption coefficient of each gas, and local blood flow [33–35]. Our data showed a larger EELI decrease with FIO2 1.0. Notwithstanding, during balloon occlusion, approximately 90% of the lobar volume reduction occurred within the first 5 minutes in both FIO2 regimens (Fig. 1). It reflects the initial higher absorption rate of oxygen that decelerates proportionally as the concentration gradient decreases.
The CT images and data supported the EIT findings. It shows that the steepest fall in the regional CT-derived gas content achieved with FIO2 1.0 is likely derived from the exhaustion of all gas content within the occluded segment. Conversely, under FIO2 0.5 the reduction of the gas content does not necessarily result in atelectasis in the short term because nitrogen still supports the alveoli (Fig. 2). The regional CT-derived gas content and regional EELI findings data presented a notable matching (Fig. 2). Likewise, the local pocket pressure and EELI data showed an apparent matched behavior over time. The FIO2 1.0 yielded a steeper decay in volumes and pressures when compared to FIO2 0.5, reaching sub atmospheric pressures. These negative pressures likely indicate negative pleural/interstitial pressures created by interdependence forces generated by extreme volume reduction of the target lobe as compared to the surroundings (Fig. 5). This situation mimics the final result of a successful valve much more than the results at FIO2 0.5.
Concerning such dynamics of volume reduction resulting from the gas absorption following lobar occlusion, our findings altogether reinforce that FIO2 1.0 yields a faster and more significant volume reduction than FIO2 0.5, which relies on nitrogen to slow down the atelectasis induction once oxygen is absorbed. While EELI detected the dynamics of the occluded area, TIV expressed the behavior of the non-occluded lung regions. Coherently, our results also showed a dissociation between the two, where changes in EELI were FIO2 and time-dependent, while the TIV changes were immediate and independent from FIO2 or its effects on EELI within the region. Of note, after balloon or valve occlusion, regional ventilation of the non-occluded areas increased immediately.
The decrease in perfusion distribution found with EIT and CT perfusion measurements after 15 minutes of occlusion of the LLL, regardless of FIO2, corroborates the hypothesis of complete oxygen consumption followed by regional hypoxia within occluded areas. A strong HPV reflex was described previously in piglets to be triggered by local PaO2 between 25 and 50 mmHg [36]. Perfusion distribution in emphysema varies widely, considering the increase of CV as an adaptive resource for V̇/Q̇ adjustments [37], in addition to the inverse relation between CV and HPV reflex [38]. In patients submitted to BLVR with EBV, V̇/Q̇ scintigraphy evaluations pre- and post-intervention showed a proportional reduction of perfusion in the treated lobe and a contralateral compensatory increase [39]. Thomsen and colleagues [40] showed a positive correlation between high non-target ipsilateral lobe perfusion and 6 minutes-walk distance test.
Pneumothorax after BLVR is a feared complication following lobar atelectasis. This patient population has a high relative risk and an incidence between 18 and 34% in clinical trials [41]. Approximately 86% occur within the first 72 hours post-BLVR [42], and it is a potentially life-threatening situation [43]. A randomized clinical trial with EBV Zephyr valves (Liberate Trial) [10] reported a 26.6% incidence of pneumothorax. Post-BLVR atelectasis is a surrogate for better functional outcomes, and pneumothorax does not influence long-term survival [44]. The predictors of pneumothorax post-BLVR are based on static CT imaging data, including the presence of pleural-pulmonary adhesions, large volume of the ipsilateral untreated lobe over the volume of the hemithorax (whose effects on it are real-time detectable by EELI monitoring from EIT), emphysema type, fissure integrity, poor lung function, and exercise testing. Such factors yield a high probability of post-BLVR pneumothorax (84%) [12].
Egenod and Born [45, 46] proposed and performed a sequential placement of EBV-Zephyr valves in the target lobe by splitting the procedure into two stages one month apart. All but one endobronchial valve are placed in the first stage, and the last and more proximal valve is placed four weeks later. This strategy decreased the incidence of post-BLVR pneumothorax from 25–9%. Nevertheless, this BLVR with staged valve placement procedure has not yet reached a consensus due to the increased risk of bronchoscopy-related complications [47], and awaits a proper randomized controlled trial.
Lentz and colleagues hypothesized that a lower FIO2 during BLVR with EBV could slow absorption atelectasis by preventing nitrogen wash-out of the treated lung and reducing the incidence of pneumothorax [48]. The authors demonstrated that a low FIO2, compared with high FIO2 during BLVR with EBV was associated with a marked reduction in the incidence of post-procedure pneumothorax [48]. Furthermore, approximately 22% of the pneumothoraces in the high FIO2 group occurred during the first hour after the procedure thus supporting that atelectasis drives pneumothorax in high FIO2 patients. Our study may challenge this perspective for two reasons: first, the low FIO2 may decrease the chances of pneumothorax but, at the same time, they might decrease the efficiency and impact of each valve, as suggested in Fig. 1; and secondly, instead of promoting a slower reabsorption, we might propose a faster one, mimicking the full, long-term effect of a single valve, but monitoring the impact in real-time. This could result in fewer valves in place, but each of them with a maximized impact.
Based on the EIT's ability to detect regional lung volume changes and ventilation shifts in real-time, we hypothesized that its use during the BLVR procedure could show the ventilation shifts that occur immediately after the procedure and potentially anticipate post-procedural pneumothorax, as shown in the two patient observations described.
Study Limitations
The EIT-based regional analysis are both bidimensional (data are provided by electrodes placed in a single plane representing the axial view), and tridimensional (the finite elements mesh model assumes a 3-D propagation of current and voltage field). This results in an EIT cross-sectional slice with approximately 15cm of thickness, although this may vary with the size and shape of the animal [49]. This property could potentially limit the detection of localized changes in lung volume occurring within the apical zones of the lung.
Despite the amount of data and the study's consistent findings, data came from a few animals with healthy lungs. We were expecting more pronounced effects in the contralateral EELI (i.e. a contralateral increase), as observed in two animals, but not in all. The lack of consistent effects in the contralateral EELI, as opposed to ipsilateral non-target lung regions, can derive from normal lung compliance and less parenchymal heterogeneity, as opposed to an emphysematous lung. Of note, although the EIT data did not show a contralateral volume increase, the CT images show a considerable expansion of the accessory lobe, displaced towards the left hemithorax.
Lastly, the right and left hemithoraces (ROIs) segmentation used a division based on the mediastinum midline. The invasion of the left ROI following the volumetric increase of the accessory lobe may have decreased the left ROI sensitivity. Such limitation derives from the EIT spatial reconstruction algorithms and the challenging anatomical segmentation of individual lobes.
Clinical Implications
The development of lung volume reduction procedures identified several physiological mechanisms [31, 50]. In patients with severe emphysema eligible to BLVR with EBV valves, the best results are related to the absence of significant CV between the target and ipsilateral lobe(s), resulting in lobar atelectasis and, thereby, an effective lung volume reduction [4]. On the other hand, as illustrated by the current study, the onset of complete lobar atelectasis of the treated lobe triggers a fast and immediate intrathoracic negative pressure and non-target lobe hyperdistention that increases the chance of pneumothorax. We tested the impact of different FIO2 during this procedure as a proof-of-concept study in a well-controlled experimental setting. It showed that faster volume reduction can be achieved with higher FIO2, enhancing the final impact of EBV valves. Although the chances of pneumothorax could be potentially increased, the real-time monitoring with EIT would allow us to stop the procedure once a pre-specified volume reduction and ventilation redistribution was achieved. This procedure would also favor the maximal impact of each valve in place.
We acknowledge that the impact of other variables such as PEEP and MV mode, or the decision to maintain or not the patient's respiratory drive during mechanical ventilation [4] must be considered. The study results are the basis for future clinical studies addressing EIT imaging to monitor and manage these complex, challenging, and variable patients in real-time.
EIT emerges as a non-invasive technology able to provide meaningful information that can contribute to the safety of BLVR as a peri-procedural monitoring tool, particularly for intraoperative decision-making and early detection of complications such as pneumothorax.