Mapping lung stress to visualize spatial heterogeneity and occult ventilation-induced lung injury risk in acute respiratory distress syndrome

In patients with acute respiratory distress syndrome (ARDS), minimizing lung mechanical stress and strain attenuates ventilation-induced lung injury (VILI), extrapulmonary organ dysfunction, and risk of death.1,2 Global measures of mechanical stress, such as plateau pressure, driving pressure, and transpulmonary pressure; surrogates of strain like tidal volume per predicted body weight; and combined indices like mechanical power, are often targeted in clinical care of ARDS.3 These indices are measured on the ventilator and, when considered with clinical-biological predisposition, may be useful for estimating the risk of VILI. However, because stress and strain exhibit spatial heterogeneity within the lung, some patients may be susceptible to VILI in at-risk lung regions even at acceptable global measures of lung stress and strain.4,5 Identifying which patients may experience occult VILI requires developing tools (1) to identify high regional stress/strain and (2) to evaluate for associated lung injury. In this issue of the Journal, Maezawa et al.tackle the first of those elements: estimating exposure to high occult regional stress/strain.6 They describe the novel approach of lung stress mapping, which estimates regional lung stress by combining esophageal manometry with chest computed tomography (CT). In brief, an esophageal balloon catheter is used to measure the mid-thoracic esophageal pressure, a well-validated surrogate of pleural pressure in the mid chest.7 The relative superimposed lung “hydrostatic” gravitational pressure is estimated from CT as the product of local tissue density (derived from Hounsfield units) and relative vertical height from the esophagus to the lung region of interest (ROI).8 The difference between measured esophageal pressure and the relative superimposed pressure estimates the local pleural pressure at the ROI. Airway pressure is then assumed to be the same throughout the lung during a breath hold. Lung stress for the ROI is calculated as the local transpulmonary pressure, that is, the difference between measured airway pressure and estimated local pleural pressure. Repeating this procedure for each pixel on the CT scan allows for estimation of the local pleural pressure at any pixel ROI. A map of local lung stress can be visualized as an overlay on the CT scan, and pixel-level variables can be integrated at the patient level to quantify the lung stress experienced across heterogeneous areas of the lung, such as maximum and mean lung stress. The authors validate this approach through a series of studies to assess (1) the accuracy of estimated regional pleural pressure; (2) the correlation of regional lung stress with inflammatory lung injury; and (3) the association of lung stress with mortality. To verify regional estimates of pleural pressures within the thorax, a porcine model was used in which pressure sensors were inserted into prespecified anatomic locations within the pleural space. The pleural pressures measured by the sensors were compared to estimated pleural pressures from lung stress mapping. These experiments were repeated across a variety of ventilator settings and lung injury severity. Overall, lung stress differed trivially by an average of 0.6-0.8 cm H2O between lung stress mapping and pressure sensors. To evaluate whether areas with higher regional lung stress experience more lung injury, a surfactant depletion rabbit model of ARDS was used. Correlations of lung stress mapping-derived variables with mRNA expression of inflammatory cytokines interleukin-6 and tumor necrosis factor-α were assessed. Pro-inflammatory gene expression correlated more strongly with lung stress mapping-derived indices than with global indices of lung stress/strain. Finally, to explore clinical significance, lung stress mapping was performed in 20 adults with ARDS. Those who died had higher levels of lung stress mapping-derived variables than survivors (eg, mean lung stress: 12.5 ± 3.9 vs 18.5 ± 3.0, respectively). The authors note important limitations of their work: applicability of preclinical models, utilizing surfactant depletion and a high tidal volume/low PEEP strategy to patients with ARDS, limited confirmatory interrogation for evidence of lung injury, small sample size of the human ARDS cohort, and exclusively supine positioning. The interaction of spatial stress heterogeneity with spontaneous breathing and its implications for VILI is an underexplored area worthy of future interrogation.9,10 Methodologically, proximal airway pressure is not synonymous with alveolar pressure even during breath holds in patients with small airways closure, plugging, or alveolar flooding or collapse.11 Nevertheless, this assumption appears useful as a rough, minimally invasive estimate of regional transpulmonary pressure.12 Finally, mechanics are but one part of the equation for VILI; each patient’s unique predisposition to biophysical damage also influences the extent to which stress-induced lung injury occurs.13 What is the take-home message for clinical practice? Even when global indices of lung stress/strain are reassuring, patients with spatially heterogeneous lung stress may be at heightened risk of VILI. How best to ascertain this risk at the bedside remains a major clinical challenge. Lung stress mapping offers one possibility when quantitative CT and esophageal manometry are available. How to incorporate findings into treatment decisions, however, requires additional investigation. Ideally, a management strategy that minimizes stress heterogeneity in each unique patient could be developed. However, serial CT to guide ventilator titration is impractical in most clinical environments. If spatial stress heterogeneity varies over time or is used for adjudicating treatment responsiveness, the approach may be impractical beyond research protocols. To overcome that issue, future investigations might consider developing predictive algorithms for spatial stress heterogeneity using clinically available data (eg, measures of mechanics and gas exchange, radiographs, molecular markers of lung injury). Such approaches could make the estimation of stress heterogeneity, if not this method of stress mapping, more clinically accessible. In summary, the work by Maezawa et al. highlights the clinically important problem of occult VILI risk in some patients despite the current standard of care. The clinical challenge remains identifying which patients are at heightened risk and then how best to individualize support. Lung stress mapping offers one potential solution to more individualized risk stratification and warrants further development. Clearly, there remains work to be done to tailor respiratory support to each patient’s unique biophysical injury risk, with the hope that this investigative direction will improve outcomes for patients with ARDS. Both authors drafted the manuscript, provided substantive revisions for important content, and approved the final submission. Supplementary material is available at American Journal of ­Respiratory and Critical Care Medicine online. Please see the ICMJE disclosure forms, which have been provided as supplementary material. Funding support for this article was provided by the US National Heart, Lung, and Blood Institute, National Institutes of Health (R01HL168102). No artificial intelligence tools were used in writing this manuscript.