All animals received standard care in accordance with French law for animal research. The protocol was approved by the Marie Lannelongue Hospital institutional review board and by the French committee for animal research (APAFIS). The principles of replacement, refinement, and reduction were followed. This manuscript was written in compliance with the ARRIVE 2.0 guidelines for animal research (3).

Figure 1 shows the experimental protocol. The lungs and whole blood were retrieved from 12 Mangaliza pigs. The animals were allocated in a 1:1 ratio to a cold ischemia time (CIT), with static preservation at 4°C, of 1 h or 18 h. The 1-hour period allowed the operator sufficient time to prepare the next phase of the experiment, while the 18-hour period was chosen based on previous reports indicating that it maximizes IRI in pig lungs (4).

Ex-vivo lung perfusion (EVLP) with the whole blood of the donor animal was then performed for 6 h in both groups.

The pigs were anesthetized with repeated intravenous injection of propofol (3 mg/kg), cisatracurium (0.3 mg/kg), and sufentanil (0.2 mg/kg). A Swan-Ganz catheter was inserted into the pulmonary artery via percutaneous puncture of the superior vena cava. Correct catheter placement was confirmed by radiography. Pulmonary artery pressures and cardiac output (CO) were measured three times, and the mean value was recorded.

A median sternotomy was performed. After heparinization (30,000 IU), the pulmonary artery was cannulated and connected to a flushing line. The inferior vena cava was cannulated (Ultraflex 28Fr, Medtronic, Dublin, Ireland) through the right atrium and connected to blood-collection bags (CPDA1 blood storage bag, Macopharma, Mouvaux, France) (Figure 2). The left hemi-azygos vein was ligated. The pig was placed in the Trendelenburg position, the atriocaval cannula clamp was released, and 1.5 L of blood was collected, within less than 5 min. The vena cava was then ligated, and the left atrium was opened.

The lungs were flushed through the pulmonary artery with 2 L of 4°C Perfadex Plus® (XVIVO Perfusion, Mölndal, Sweden) and 1,000 µg of epoprostenol (Flolan®, GSK, London, UK). The lungs were recovered, inflated with 50% FiO₂, and stored at 4°C in bags after a retrograde pneumoplegia with 1l of Perfadex Plus® for either the short period of 1 h (S-CIT group) or the long period of 18 h (L-CIT group).

After 1 h or 18 h of cold static preservation, the lungs were connected to a closed circuit (EVLP circuit kit without the leukocyte filter, XVIVO) consisting of a centrifugal pump and a blood reservoir. The left atrium and pulmonary artery were cannulated, and the trachea was intubated. The circuit was primed with donor whole blood and 10,000 IU of heparin (Figure 3).

The perfusion pressures of the pulmonary artery (20 mmHg) and left atrium (5 mmHg) were chosen to simulate the conditions experienced by lung grafts at the time of clamp release in human clinical practice. The perfusion flow rate was continuously measured and constantly adjusted to maintain the pulmonary arterial pressure (PAP) at 20 mmHg. The level of the reservoir was adjusted to maintain the left atrial pressure at 5 mmHg throughout the procedure.

The temperature was gradually increased to 32°C after 20 min of perfusion and maintained at 37.5°C from 30 min onward. Ventilation was started once the graft reached 32°C, with the following parameters: tidal volume, 7 ml/kg/min; positive end-expiratory pressure (PEEP), 5 mmHg; FiO₂, 21%, and respiratory rate, 7 breaths/min. Every hour, the lungs were challenged by increasing the tidal volume to 10 ml/kg/min, FiO₂ to 100%, and the respiratory rate to 10 breaths/min, for 10 min, to allow assessments. The gas flow of the EVLP system was initiated with ventilation and adjusted to maintain an inflow PaCO₂ between 35 and 45 mmHg.

Every 30 min, we recorded the flow rate required to maintain a pulmonary artery pressure of 20 mmHg, as a percentage of the donor's CO. Pulmonary vascular resistance (PVR, dyne·s·cm⁻⁵) was calculated as [(PAP-LAP)/Flow rate] × 80, where PAP is the pulmonary artery pressure and LAP the left atrial pressure.

After one hour then every 30 min, we recorded the ventilatory peak pressure (Ppeak) and mean pressure (Pmean). Pulmonary compliance (ml/mmHg) was calculated as Tidal volume/(Ppeak-PEEP).

Blood gas analyses were performed simultaneously in the inflow and outflow circuits after each lung challenge. Graft oxygenation capacity was expressed as PaO₂ outflow - PaO₂ inflow (ΔPaO₂).

Hourly, 5 ml of perfusate was taken for analysis and 5 ml of blood was centrifuged at 3,000 rpm for 10 min and the supernatant stored at −80°C for further analysis.

Perfusate levels of interleukin (IL)-2, IL-4, IL-6, IL-8, IL-10, IL-12, IL-18, and TNFα were measured using a Bio-Plex Pro Human Cytokine Assay kit (Bio-Rad Laboratories, Mississauga, Canada), following the manufacturer's protocol. The analytes were read with a Bioplex200 analyzer (Bio-Rad Laboratories), and the data were analyzed using Bio-Plex Manager 6.0 (Bio-Rad Laboratories).

Lung samples from the left lower lobe were taken hourly for a descriptive histological evaluation. After 6 h, lung samples from the left superior lobe were taken for a comparative histological assessment. Specimens were fixed in buffered 4.5% formaldehyde for standard microscopic examination. Sections were colored with hematoxylin eosin saffron. Two slides per animal were assessed by the same lung pathologist who was blinded to data on the animals.

Based on our previous observations, we developed a semi-quantitative descriptive histological score for IR injury, based on four items, each with subcategories: edema (subpleural, septal, peribronchiolar, perivascular), lymphangiectasis (septal, peribronchiolar), inflammatory infiltrate (subpleural, septal, peribronchiolar, perivascular, intra-alveolar) and vascular congestion. Each subcategory was rated 0 (absent), 1 (present, <25% of section), 2 (present, 25%–50% of section), or 3 (present, >50% of section). The score could thus range from 0 (no lesions) to 36 (most severe lesions).

To evaluate lung edema formation, at the end of EVLP, the right superior lobe was dissected, weighed, and heated at 80°C for 7 days to allow calculation of the wet/dry lung weight ratio.

The variables are described as mean ± SD. The Shapiro–Wilk test was applied to assess variable distribution. Comparisons were with Student's t test for normally distributed variables and with the non-parametric test Mann–Whitney test for skewed variables. Repeatedly measured variables were evaluated using two-way analysis of variance (ANOVA) and linear regression with or without an interactive term (time or groups) depending on the significance of the interaction verified by the ANOVA. Time was handled as a categorical variable in the regression analyses. The linear regression results were expressed as the estimated mean differences with their 95% confidence intervals (95% CI).

All the statistical analyses were performed with the R program (v4.1.2, open source, http://cran.r-project.org/) and graphs were drawn with Prism® V8.0 (GraphPad, La Jolla, CA). All tests were two-sided and P values <0.05 were considered significant.

With a constant pulmonary pressure of 20 mmHg, the perfusion flow rate (Figure 3A) increased over 2 h in the S-CIT group from 42% ± 32% to 64 ± 15% of CO and over 3.5 h in the L-CIT group, but only from 16% ± 24% to 40% ± 21% of CO. During the remainder of the procedure, the flow rate had to be decreased in both groups to prevent an increase in pulmonary pressure. The maximum achievable flow was significantly lower in the L-CIT group during the first 3 h of perfusion (estimated mean difference: −39% of CO; 95% CI, −66 to −12; P = 0.005).

Conversely, PVR (Figure 3B) decreased initially (S-CIT: from 1,253 ± 1,021 dyne·s·cm⁻⁵ at 1 h to 478 ± 129 dyne·s·cm⁻⁵ at 2 h: L-CIT: from 3,799 ± 1,862 dyne·s·cm⁻⁵ at 1 h to 848 ± 313 dyne·s·cm⁻⁵ at 3.5 h) then rose sharply. Overall, PVR remained significantly higher in the L-CIT group (estimated mean difference, 1,077 dyne·s·cm⁻⁵; 95%CI, 685–1,469; P < 0.001).

Pulmonary compliance (Figure 3C) exhibited a sawtooth pattern in both groups due to the increase in ventilation pressures during the hourly challenges. The range was 60 ± 23–75 ± 11 mmHg in the S-CIT group and 48 ± 11–66 ± 20 mmHg in the L-CIT group. Overall, compliance was significantly lower in the L-CIT group (estimated mean difference, −11.17 mmHg; 95% CI, −17 to −5; P < 0.001).

After the first hour of perfusion, water evaporation in the closed circuit led to a gradual increase in hemoglobin levels (Figure 4A) (S-CIT: 8 ± 0.5–9.2 ± 0.5; L-CIT: 8.2 ± 0.8–9.3 ± 1; P = 0.49).

Rewarming resulted in the resumption of aerobic metabolism in the lungs. Thus, glycemia (Figure 4B) decreased linearly (S-CIT: 22 ± 3–16 ± 4 mmol/L; L-CIT: 21 ± 3–16 ± 4 mmol/L; P = 0.48). Lactate levels (Figure 4C) increased steadily over six hours (S-CIT: 5.7 ± 1–14.4 ± 5 mmol/L; L-CIT: 5.4 ± 0.8–14.8 ± 3.5 mmol/L; P = 0.51).

ΔPaO₂ (Figure 4D), a marker of lung gas-exchange capacity, remained high in both groups throughout the procedure (S-CIT: 307 ± 87–410 ± 52 mmHg; L-CIT: 327 ± 83–450 ± 61 mmHg; P = 0.189).

Blood leukocyte counts decreased gradually during EVLP. Lymphocyte counts (Figure 4E) declined moderately over the 6 h (S-CIT: 5,457 ± 1,614–3,066 ± 1,526; L-CIT: 4,852 ± 2,434–3,062 ± 1,002 cells/mm³; P = 0.42). The most notable change was in neutrophil counts (Figure 4F) (H0, H3, and H6: S-CIT, 4,559 ± 1,111, 1,388 ± 1,738, and 884 ± 988 cells/mm³; L-CIT: 4,252 ± 2,452, 699 ± 549, and 434 ± 283 cells/mm³). This decrease was significantly greater in the L-CIT group (estimated mean difference, −744.02 cells/mm³; 95% CI, −1,343.11 to −144.92; P = 0.017).

Eosinophil counts (Figure 4G) also decreased from H0 to H6 (S-CIT: 127 ± 36–38 ± 56 cells/mm³; L-CIT: 92 ± 32–14 ± 18 cells/mm³; P = 0.32). Monocyte (Figure 4H) counts showed little change throughout EVLP (S-CIT: 82 ± 28–67 ± 20 cells/mm³; L-CIT: 77 ± 31–54 ± 29 cells/mm³; P = 0.18). Basophil counts (Figure 4I) from H0 to H6 also remained fairly stable (S-CIT: 41 ± 22–46 ± 26 cells/mm³; L-CIT: 51 ± 26–41 ± 19 cells/mm³; P = 0.32).

IL-6 levels (Figure 5A) were similar at baseline between groups and increased significantly after the first two hours of perfusion in both groups (S-CIT: 0.53 ± 0.15–2.79 ± 0.62 pg/ml; L-CIT: 0.48 ± 0.15–3.39 ± 0.86 pg/ml). The increase was significantly greater in the L-CIT group (estimated mean difference, 1.1 pg/ml; 95% CI, 0.39–1.8; P = 0.003). IL-8 concentrations (Figure 5B) increased more markedly than IL-6 (S-CIT: 0.11 ± 0.16–1.32 ± 2.10 pg/ml; L-CIT: 0.49 ± 0.55–30.62 ± 20.04 pg/ml; estimated mean difference, 29.31 pg/ml; 95% CI, 16.00–42.61; P < 0.001).

IL-2 (Figure 5C) and IL-4 (Figure 5D) levels were significantly higher from H0 to H6 in the L-CIT group (IL-2: 0.55 ± 0.8 and 0.51 ± 0.69 pg/ml; IL-4: 1.16 ± 1.8 and 1.09 ± 1.61 pg/ml) than in the S-CIT group (IL-2: 0.15 ± 0.14 and 0.17 ± 0.19 pg/ml; IL-4: 0.21 ± 0.22 and 0.25 ± 0.3 pg/ml). The estimated mean difference for IL-2 was 0.33 pg/ml (95% CI, 0.11–0.55; P = 0.004). Corresponding values for IL-4 were 0.78 pg/ml (95% CI, 0.3–1.25; P = 0.002). IL-18 concentrations (Figure 5E) also increased more in the L-CIT group (from 1.16 ± 1.36 to 3.32 ± 1.00 pg/ml) than in the S-CIT group (from 0.45 ± 0.23 to 1.02 ± 0.31 pg/ml), with an estimated mean difference of 1.7 pg/ml (95% CI, 1.36–2.05; P < 0.001).

Levels of the anti-inflammatory cytokine IL-10 (Figure 5F) were significantly higher in the L-CIT group, increasing from 0.9 ± 1.42 to 2.08 ± 1.1 pg/ml vs. 0.18 ± 0.17 to 0.85 ± 0.49 pg/ml in the S-CIT group (estimated mean difference, 0.79 pg/ml; 95% CI, 0.43–1.15; P < 0.001). Nonetheless, the IL-8/IL-10 ratio (Figure 5G) indicated significantly worse inflammation in the L-CIT group at H6 (estimated mean difference, 15.95 pg/ml; 95% CI, 5.36–26.53; P = 0.005). TNF-α concentrations (Figure 5H) followed a different pattern compared to other cytokines, increasing until H3 (to 0.76 ± 0.97 and 1.41 ± 1.18 pg/ml in the S-CIT and L-CIT groups, respectively) then decreasing, with no significant between-group difference (P = 0.083). IL-12 concentrations (Figure 5I) were similar in both groups (at H0 and H6: S-CIT group, 0.37 ± 0.06 and 0.52 ± 0.1 pg/ml; L-CIT group: 0.39 ± 0.14 and 0.55 ± 0.19 pg/ml; P = 0.42).

Figure 6 shows the histological findings. At baseline, the lung parenchyma was normal in both groups. During EVLP, microvasculopathy developed gradually, with leukocyte adhesion to the endothelium initially then progressively worsening neutrophilic capillaritis. These lesions did not affect the large vessels. After 3 h, tissue infiltrates, edema, and lymphatic-vessel dilation developed.

At H6 (Figure 7), the most prevalent lesions in the S-CIT group were interseptal lymphangiectasia (mean score, 2.3 ± 0.2), peribronchiolar infiltrates (1.8 ± 0.2), and peribronchiolar lymphangiectasia (1.7 ± 0.2). The most severe lesions in the L-CIT group were interseptal lymphangiectasia (2.5 ± 0.2), septal edema (2.2 ± 0.2), and vascular congestion (1.5 ± 0.2). The total histological score was not significantly different between groups (S-CIT: 14 ± 1.1; L-CIT: 14.3 ± 0.; P = 0.79). The wet/dry ratio was similarly elevated in both groups (S-CIT: 4.9 ± 0.19; L-CIT: 5.4 ± 0.22, P = 0.13).