Perfusion circuit description
A previously described ex-vivo isolated and perfused rabbit lung model of VILI was used. However, this model has been modified by our research group [15, 16] in order to allow more accurate and real-time monitoring of the oxygenation and acid-base status of the perfusion fluid.
Briefly, the model consisted of a perfusion circuit incorporating the heart-lungs preparation. The perfusion medium consisted of a custom-made perfusate  and autologous blood. The circuit's venous pool consisted of a bottle containing the perfusion medium which was directed towards the pulmonary circulation under the drive of a mechanical pump (Masterflex 07550; Cole-Parmer Instrument Co, IL, USA). An "efferent" tube was inserted in the left atrium, and ended up over the venous pool bottle "closing" the circuit. By voluntarily adjusting the height of the end of the "efferent" tube, the afterload of the pulmonary circulation could be manipulated.
Blood pressure monitoring catheters were connected to a calibrated monitoring system (Direc/NEP 201 B Physiologic Recording System, Raytech Instruments Inc, Vancouver, BC, Canada). Additionally, real time blood gas measurement was carried out via a catheter inserted to the "afferent" tube and connected to the respective equipment (Trend Care, 7+™, TCM 7000, Diametrics Medical Inc., Minnesota, USA). A gas mixture of CO2 (35-45%) and room air (65-55%) was insufflated into the venous pool bottle achieving normocapnic conditions and oxygen partial pressure at the 120-170 mmHg range. Surgical extraction of the heart-lungs block and incorporation of the preparation into the perfusion circuit was performed according to a previously described technique . According to the Greek legislation, approval from the Veterinary Directorate of the Prefecture of Athens in conformance to the 160/1991 Council Directive of the European Union was obtained before the beginning of the study. New Zealand white rabbits were utilized. After extraction of the heart-lungs block and through a small transverse incision at the right ventricle, an "afferent" tube was inserted into the pulmonary artery trunk and secured with sutures. Respectively, a small transverse incision was made at the apex of the heart and the "efferent" tube was advanced into the left atrium and secured with sutures and a cotton tape. Finally, the preparation was suspended from an electronic balance (BG 025, Mark-10, NY, USA) and connected to a ventilator (T-Bird AVS III, Thermo Respiratory Group, CA, USA).
Measurements - Indices of lung injury
Lung injury was assessed using hemodynamic parameters as well as histology indices.
Weight gain represented solely the mass of the perfusate leaking through the alveolar space-blood barrier, i.e. pulmonary edema. Measurements of weight gain were made during the 1-hour period of pressure controlled ventilation (PCV) ventilation.
Ultrafiltration coefficient (Kf)
The integrity of the endothelium was assessed via its Kf using the double occlusion technique, previously described elsewhere [17, 18].
Pulmonary artery pressure (PAP) changes
PAP measurements were recorded during the PCV interval, at set time points (20 and 40 min of ventilation).
Histology was performed on the left lung. Three tissue slices were taken from the lung apex, hilum and base. Four indices of histologic damage were evaluated similarly to other studies . Intra-alveolar as well as interstitial hemorrhage, accumulation of inflammatory cells, degree of alveolar remodelling and interstitial edema were evaluated by a pathologist blinded to specimen group allocation. After evaluation of each of these parameters, a cumulative lung injury score ranging from 0 to 3 was assigned to the specimen.
Ventilation - perfusion protocol
All preparations were initially perfused in normocapnic conditions (pCO2 = 40 mmHg) and maintained at a continuous positive airway pressure (CPAP) of 5 cmH2O. Initial flow was 30 ml/min and was progressively increased to a steady value of 300 ml/min within 20 min. A 5-min period of compensation was allowed (under CPAP = 5 cmH20) and an initial measurement of the ultrafiltration coefficient - Kf (Kfbase) took place.
Within the next 30-min period under CPAP = 5 cmH2O, preparations were randomized to: (i) the control (C) group (pCO2 = 40 mmHg, pH = 7.4), (ii) the hypercapnic, respiratory acidosis (RA) group (pCO2 = 100-130 mmHg, pH = 6.9-7.2) or (iii) the normocapnic, metabolic acidosis (MA) group (pCO2 = 40 mmHg, pH = 6.9-7.2). Hypercapnic acidosis was achieved by increasing the fraction of the CO2 insufflated into the venous bottle from 35-45 to 60% while metabolic acidosis was achieved by slowly infusing into the perfusate medium approximately 2 ml of HCL solution 0.5 N.
Once the desirable metabolic conditions were reached and a second Kf (Kf1) measurement was obtained, the second stage of randomization took place. This included ventilation of the preparations of each metabolic group (C, RA, MA) with high pressure (HP; PCV = 22 cm H2O, PEEP = 3 cm H2O) or low pressure (LP; PCV = 12 cm H2O, PEEP = 3 cm H2O) (a total of 6 groups). All preparations were then ventilated for 1 hour. At the end of this period, a third Kf (Kf2) measurement was performed and the ventilation-perfusion was terminated. In addition, the preparation's weight was measured in order to calculate the initial lung weight. The latter was calculated by subtracting the heart, connective tissue and tracheal tube weight from the initial preparation weight, after removal of the lungs at the end of the protocol.
In order to adjust for technical failures during the delicate surgical maneuvers, heart-lungs preparations macroscopically damaged during their surgical extraction, exhibiting vascular failure during the 5-min compensation period or the first 5 minutes of PCV ventilation were excluded from the analysis. In addition, if - during the 60-min period of PCV ventilation - the heart-lungs preparations exhibited visible edema overflow from the tracheal tube due to obvious failure of the lungs' vascular beds, the ventilation-perfusion was terminated. In that case, the last recorded readings of weight and PAP were included in the analysis but Kf2 could not be performed nor change in ultrafiltration coefficient (dKf) could be calculated.
Two-way (analysis for two independent variables and their interaction) ANOVA was used for multiple comparisons of physiologic indices of lung injury (weight gain, Kf changes as well as injury scores) for between group comparisons with least significance difference (LSD) test for within-group comparisons.