Optical Assessment of Pulmonary Vascular Leak Using a Model of Ex Vivo Lung Perfusion

Rationale To evaluate the performance of dynamic multi-wavelength in situ optical endomicroscopy in conjunction with intravascular fluorescent probes to detect pulmonary capillary leak during ex vivo lung perfusion Methods Ovine lungs and human (declined for transplantation) lungs (n=4) were retrieved, flushed and cooled for transport. They were initiated on an ex vivo perfusion circuit using an open left atrium and protein rich perfusate. Once physiologically stable , three colour alveolar optical imaging was performed via either a blunt transpleural or transbronchial puncture using a 1.4mm diameter optical fibre before and after intravenous delivery of 2 fluorophores: Indocyanine Green (ICG) and Sulphonated Cy-5. ICG fluoresces in near infra red (NIR) and has a high affinity for plasma protein so is expected to remain intravascularly, whereas Cy-5 fluoresces in red and is expected to move with free water. Localised injury was induced in one lobe using an acid aspiration model (20ml HCl) and imaging was compared with a control lobe after further 45 minutes of perfusion to assess fluorescence intensities. Results In this exploratory study we detected an increase in mean red fluorescence after perfusion, a significant increase (p<0.05) in protein bound NIR signal in acid treated lung regions, but no significant increase in NIR signal in control regions. Conclusion These exploratory results suggest that it is feasible to photonically detect pulmonary capillary leak of protein in situ using fibre endomicroscopy. The increase in NIR signal in the injured lobes reflects movement of intravascular plasma protein due to disruption of the air-blood barrier, this is not seen in control lobes due to intact capillaries. This methodology has potential utility in the assessment of patients in critical care with pulmonary infiltrate to distinguish between hydrostatic pulmonary oedema and the protein rich capillary leak seen in acute respiratory distress syndrome (ARDS). Further work is required to translate this to an in vivo model. The ability to assess discrete lung regions and to reflect capillary integrity offers an advantage over existing methods of lung water assessment.