Adult (8–10 weeks) male Sprague-Dawley and Wistar rats were used in this study. The animals were group-housed in a climate-controlled room on a 12-h light–dark cycle at the animal facilities of our institution. Experimental protocols were performed in strict accordance with the recommendations of the European Union Criteria for Animal Use in Scientific Experimentation (63/2010 EU) and related Spanish legislation (RD 53/2013). The experimental procedures were approved by the local Animal Welfare Ethics Committee (PROEX 58.7/20).

For the study of intestinal graft quality coming from different donation scenarios, animals were divided into five experimental groups (Five animals per group) (Figure 1A): 1. Conventional LD: i.v. organ perfusion; 2. Ventilated LD (VLD): mechanical ventilation for 2 h, i.v organ perfusion; 3. Donation after 5 min of circulatory death (DCD5): mechanical ventilation for 2 h, 5 min of CD, i.v. organ perfusion 4. Donation after 20 min of circulatory death (CD): mechanical ventilation for 2 h, 20 min of CD, i.v. organ perfusion (10); 5. Brain Death (BD): BD and ventilation for 2 h, i.v organ perfusion (11, 12). Additional information from experimental procedures is available in Supplementary Data S1.

In all groups, samples of the different organs comprising the multivisceral graft (small bowel, stomach, pancreas, colon, and liver) were taken for histological analysis (13–20).

For long-term intestinal graft survival evaluation after allogeneic and heterotopic ITx, two experimental groups was performed (Five ITx per group) (Figure 1B): 1- LD + ITx, and 2: DCD5 + ITx. Additional information is available in Supplementary Data S2.

Samples of transplanted intestine were taken through intestinal ostomy at 0, 30 min, 48 h and 7 days after ITx. Histological and morphometric graft evaluation, Immunofluorescence staining, gene expression analysis and graft functional evaluation was performed (21) (Supplementary Datas S3–S7).

The experimental model developed in this study enabled the replication of crucial phases of Maastricht III DCD (patient maintenance, donor limitation of life support therapy (LLST), diagnosis of CD, hypoperfusion phase, waiting period after cardiac arrest (no-touch period), cannulation time, washing, and multivisceral graft retrieval). The success of the Maastricht III model in rats for visceral transplantation (VT) was 83.3%. Two animals presented complications during the procedure and were excluded from the study (one case of prolonged hypotension and one case of massive hemorrhage through the artery cannula).

Successful donor monitoring was achieved in all experimental groups based on stable body temperature, O₂ saturation, and mean arterial pressure, which were established as inclusion parameters as described in the Materials and Methods section. BDD exhibited a slight increase in mean arterial pressure (MAP) at 30 and 40 min after BD induction, which was attributed to the norepinephrine infusion that was concurrently administered to this group (Figure 2A).

The time elapsed between the LLST and diagnosis of CD was 2.8 ± 1.4 min. No significant differences were observed between DCD5 and DCD20 (Figure 2B). CD temporal heterogeneity was not correlated with MAP at the time of LLST (data not shown). After “no-touch period” (5 or 20 min, according to experimental group), time required for opening, cross-clamp, cannulation and perfusion was 3.4 ± 0.35 min in DCD5 and 3.46 ± 0.35 in DCD20 groups (No significant differences between groups) (Figure 2C). Considering hypoperfusion phase, waiting period after cardiac arrest and cannulation, total warm ischemia time was 8.41 ± 0.35 min and 23.5 ± 0.34 for DCD5 and DCD20, respectively (p < 0.05).

At the time of sampling, the macroscopic appearance of intestinal grafts showed pale appearance in the five experimental groups, without traces of blood in the mesenteric vasculature or small intestinal wall, indicative of correct washing of the graft even in DCD20 (Figure 2D).

The DCD group exhibited time-dependent small bowel histological damage. Prolonged CD generated substantial intestinal mucosal injury (3–5 based on the Park/Chiu score). Significant differences were observed between the DCD20 versus LD and DCD5 groups (p < 0.05, Figure 3A). However, no significant differences were observed in villus/crypt index and mucosal thickness between experimental groups (Figures 3B,C). Similar results were observed in all groups with the exception of DCD20 small bowel samples, in which GC count was considerably lower. A significant difference in Goblet cell (GC) analysis was noted between DCD20 and LD groups (p < 0.05, Figures 4A,B). Conversely, no differences were observed in Paneth cells per crypt count, with an average of approximately 5 Paneth cells in all study groups (Figure 4C). DCD20 group showed major alterations in tight junction proteins ZO-1 and claudin-1 expression (Figure 4E), whereas minor alterations were observed in DCD5 and BD groups.

The average variability explained by principal component analysis was 73% (Figure 4D, principal component 1, 55%; principal component 2, 18%). GC number, mucosal thickness, and villus-crypt index were directly correlated with LD and VLD groups, suggesting that higher GC number, thicker mucosa, and increased crypt-villus index were indicative of small bowel integrity. Park/chiu score was inversely correlated with these variables but was strongly correlated with the DCD20 group. Of note, despite data dispersion, BDD and DCD5 groups were clustered closer to the VD and VLD groups than to the DCD20 group.

In agreement with the trend observed for intestinal grafts, other organs were also affected by longer periods of DCD: DCD20 livers revealed significantly poorer morphology compared to LD (Figure 5A), with mild-to-moderate congestion and cytoplasmic vacuolization as the most relevant microscopic alterations (DCD20 vs. LD group, p < 0.05).

The DCD20 group presented more severe gastric injury, accompanied by glandular damage, epithelial desquamation, and cellular infiltrate of a mild-to-moderate degree (p < 0.05 vs. LD group). Similar but milder injuries were observed in DCD5 and BDD gruops (Figure 5B).

Moderate edema was the most common observation in the pancreas of BDD and DCD20 groups (Figure 5C). A significant difference between DCD20 and LD in pancreas histological damage was observed (p < 0.05).

Slight edema was observed in a subset of DCD and BDD colon samples. No alterations in this organ were detected in LD. No significant differences in the analysis on colon samples were observed between groups.

Considering the results obtained in the graft characterization we decided to perform ITx using a heterotopic model previously described. In order to determine if DCD5 may affect the long-term transplant outcome, in spite of the lack of differences with LD graft. To this aim, two groups were subject to ITx using either DCD5 or LD grafts. Upon vascular anastomosis, DCD5 grafts reperfusion was fast and homogeneous, comparable to LD grafts (Figure 6A). Post-operative recipient recovery was appropriate and weight and clinical evolution was comparable in DCD5 and LD groups. Similar results were observed in survival curve (Figure 6B), only 1 animal from DCD group died in 3rd post-operative day, which was the same animal that showed extensive ischemia/reperfusion damage at 48 h post-transplant (Figure 6C). Anyway, no statistical significant differences in mortality were observed between groups. All animals survived showing no signs of graft rejection until 7 post-operative days, when the protocol was ended. During ITx, serial samples of graft were obtained to evaluate ischemia/reperfusion damage (Figure 6C). Interestingly, although DCD5 donors have a Park/chiu score 2 points higher than LD in the pre-reperfusion sample, showing the impact of CD on intestinal histology as shown in previous experiments. Upon engraftment, the highest damage was observed in the 30 min post-reperfusion sampling, where both DCD5 and LD grafts showed similar damage (median Park/chiu score of 4 in both groups). Recovery from damage was evident in the 48 h post-reperfusion sample of LD and DCD5 groups. Histopathological analysis was performed on intestinal graft samples taken at day 7 post-operative. No signs of rejection were observed in either LD or DCD5 groups (Figure 6D). Furthermore, functional capacity of grafts was assessed at day 7 post-ITx using a glucose absorption test. Either LD or DCD5 grafts showed appropriate absorptive function, indicating that DCD procedure has not impacted in absorptive capacity of the graft (Figure 6E).

To determine comparative effects of DCD and LD on additional pathways, gene expression analysis was performed in graft samples taken at different times: just before the dissection procedure (basal samples) and 30 min, 48 h and 7 days after ITx (Figure 6F). ZO1 and claudin-3 were analyzed as markers of epithelial integrity, and IL6 and TNFα as reflection of pro-inflammatory condition. According to what observed in the histological analysis, no differences were observed between LD and DCD in any of the genes and time-points studied (Figure 6F). Both groups showed a high expression of claudin-3 in basal and 30′ samples which decreased significantly after 48 h (LD: p = 0.002, DCD: p = 0.006), this correlating with the Chiu-Park score normalization in the biopsies. A similar but smoother dynamic was observed in zonulin expression, showing the LD group a significant increase at 30’ vs. basal samples (p = 0.010). Regarding the pro-inflammatory cytokines, TNFα showed the highest levels in basal samples, decreasing progressively afterwards. Nevertheless, IL6, which was almost undetectable at the beginning of the procedure, increased after reperfusion (Figure 6F). Overall, this experiment indicates that selection of appropriate DCD grafts results in a transplantation procedure with comparable outcomes to the LD experimental gold standard scenario.