Regional pulmonary perfusion, blood volume, and their relationship change in experimental early ARDS.

Regional pulmonary perfusion (Q) has been investigated using blood volume (Fb) imaging as an easier-to-measure surrogate. However, it is unclear if changing pulmonary conditions could affect their relationship. We hypothesized that vascular changes in early acute respiratory distress syndrome (ARDS) affect Q and Fb differently. Five sheep were anesthetized and received lung protective mechanical ventilation for 20 h while endotoxin was continuously infused. Using dynamic 18F-FDG and 13NN Positron Emission Tomography (PET), regional Fb and Q were analysed in 30 regions of interest (ROIs) and normalized by tissue content (Fbn and Qn, respectively). After 20 h, the lung injury showed characteristics of early ARDS, including gas exchange and lung mechanics. PET images of Fbn and Qn showed substantial differences between baseline and lung injury. Lung injury caused a significant change in the Fbn-Qn relationship compared to baseline (p < 0.001). The best models at baseline and lung injury were Fbn = 0.32 + 0.690Qn and Fbn = 1.684Qn-0.538Qn2, respectively. Endotoxine-associated early ARDS changed the relationship between Fb and Q, shifting from linear to curvilinear. Effects of endotoxin exposure on the vasoactive blood flow regulation were most likely the key factor for this change limiting the quantitative accuracy of Fb imaging as a surrogate for regional Q.

Mechanical ventilation guided by driving pressure optimizes local pulmonary biomechanics in an ovine model.

Mechanical ventilation exposes the lung to injurious stresses and strains that can negatively affect clinical outcomes in acute respiratory distress syndrome or cause pulmonary complications after general anesthesia. Excess global lung strain, estimated as increased respiratory system driving pressure, is associated with mortality related to mechanical ventilation. The role of small-dimension biomechanical factors underlying this association and their spatial heterogeneity within the lung are currently unknown. Using four-dimensional computed tomography with a voxel resolution of 2.4 cubic millimeters and a multiresolution convolutional neural network for whole-lung image segmentation, we dynamically measured voxel-wise lung inflation and tidal parenchymal strains. Healthy or injured ovine lungs were evaluated as the mechanical ventilation positive end-expiratory pressure (PEEP) was titrated from 20 to 2 centimeters of water. The PEEP of minimal driving pressure (PEEPDP) optimized local lung biomechanics. We observed a greater rate of change in nonaerated lung mass with respect to PEEP below PEEPDP compared with PEEP values above this threshold. PEEPDP similarly characterized a breaking point in the relationships between PEEP and SD of local tidal parenchymal strain, the 95th percentile of local strains, and the magnitude of tidal overdistension. These findings advance the understanding of lung collapse, tidal overdistension, and strain heterogeneity as local triggers of ventilator-induced lung injury in large-animal lungs similar to those of humans and could inform the clinical management of mechanical ventilation to improve local lung biomechanics.