Endothelial and inflammatory responses during ex vivo normothermic perfusion of human cardiac grafts

Abstract

Aims

Normothermic ex vivo heart perfusion (NEVHP) allows functional assessment and preservation of donor hearts, but the biological responses occurring during perfusion are not well characterized. This pilot study evaluated the feasibility of sequential biomarker monitoring during human cardiac NEVHP and described inflammatory, endothelial, and haemostatic responses over time.

Methods and results

This single-centre, prospective, observational pilot study included five consecutive donor hearts preserved with the TransMedics Organ Care System (OCS) at the University Hospital of Rennes between March 2021 and January 2023. OCS use was indicated for expected cold ischaemia >4 h or surgical complexity. Perfusate samples were collected after priming (T0), at 10 min (T1), 60 min (T2), and before cooling (T3). Cytokines, chemokines, endothelial markers, and haemostatic factors were quantified by multiplex immunoassay and ELISA. Data were analysed using linear mixed-effects models and expressed as fold-change vs T0. Donor median age was 44 years (IQR 36–50) and 60% male. All grafts were transplanted. Recipient mean age was 53 ± 9 years and 20% female. Two of five (40%) developed severe primary graft dysfunction. Sequential sampling was successful in all perfusions. Inflammatory mediators rose during perfusion: at T3 vs T0, IL-8 increased 11.5-fold (95% CI 6.5–20.1), bFGF 8.4-fold (6.7–10.5), IL-6 4.5-fold (2.4–8.2), and MCP-1/CCL2 4.1-fold (2.6–6.6). IL-10 and TNF-α showed smaller increases (2.0- and 1.6-fold). Leukocyte counts remained stable (0.57 × 10⁹/L at T0; fold-change 0.9). Endothelial markers showed activation without evidence of injury. Angiopoietin-2 increased 1.6-fold (1.2–2.1) and VEGF 2.0-fold (1.2–3.5), while angiopoietin-1, syndecan-1, soluble E-selectin, thrombomodulin, VEGFR2, PlGF, and vWF:Ag showed minimal or inconsistent changes. These trajectories are consistent with endothelial activation in the absence of glycocalyx shedding or structural disruption. Despite high heparin levels (median 6.9 IU/ml), low-grade haemostatic activation occurred. D-dimer increased 1.9-fold (1.3–2.7), fibrin monomer 2.2-fold (1.2–3.9), and soluble P-selectin 1.5-fold (1.1–2.0). Platelet counts declined to 0.8 (0.7–0.9) relative to baseline. Haematocrit decreased slightly (15.5% to 14.6%, fold-change 0.94), consistent with mild haemodilution.

Conclusion

Sequential biomarker monitoring during NEVHP was feasible and demonstrated inflammatory, endothelial, and haemostatic changes. These biological patterns require confirmation in larger cohorts, as potential tools for graft assessment and optimization of perfusion circuits and perfusate composition.

Introduction

Heart transplantation is the treatment of choice for end-stage heart failure, but its use is limited by donor shortage, cold ischaemia, and the risk of primary graft dysfunction (PGD).¹

Normothermic ex vivo heart perfusion (NEVHP) has emerged as a strategy to mitigate ischaemia–reperfusion injury, extend preservation, and allow evaluation of extended-criteria and donation-after-circulatory-death grafts.² However, ex situ perfusion is difficult to maintain for long periods, as extended perfusion can lead to substantial biological and metabolic alterations and has been associated with worse outcomes in donation after circulatory death heart transplantation, which highlights the need for a better understanding of the biological changes that occur during the perfusion window.^3,4 In addition, the increasing use of marginal donors highlights the need for biological markers that could improve graft assessment and provide insight into mechanisms of injury.⁵

Clinical assessment and lactate remain the main parameters to guide organ selection, but both are imperfect predictors of post-transplant outcomes.⁶ Inflammatory, endothelial, and haemostatic responses are key contributors to ischaemia–reperfusion injury and have been associated with PGD, yet their dynamics during human heart perfusion remain poorly characterized.^1,7 We therefore conducted a pilot study to assess the feasibility of monitoring biomarker dynamics during NEVHP and to generate preliminary data on thrombo-inflammatory and endothelial responses.

Methods

This pilot observational study was conducted at University Hospital of Rennes with ethical approval (No. 2139, 23 March 2021), in accordance with the Declaration of Helsinki. Recipient data were obtained from clinical records (CNIL No. 1685088) and donor characteristics from the French CRISTAL registry (CNIL No. 363505).

Normothermic ex vivo heart perfusion used the TransMedics Organ Care System (OCS; TransMedics, Andover, MA) per institutional practice. The OCS was used when anticipated cold ischaemia exceeded 4 h or in technically complex cases (redo surgery, long-term mechanical support). Retrieval, perfusion, and cooling followed a standardized protocol (Supplementary Methods).

Five consecutive grafts were included. As a feasibility, hypothesis-generating study, no sample size calculation was performed. Perfusate samples were obtained at baseline after priming (T0), at 10 min (T1), 60 min (T2), and pre-cooling (T3). Biomarkers included complete blood count; inflammatory cytokines/chemokines and growth factors (TNF-α, IFN-γ, IL-1β, IL-6, IL-8, IL-10, MCP-1/CCL2f, bFGF, c-Kit); endothelial markers (angiopoietin-1, angiopoietin-2, VWF:Ag, soluble E-selectin, syndecan-1, thrombomodulin, PlGF, VEGF, VEGFR2); and coagulation/platelet activation markers (D-dimer, fibrin monomer, platelet count, soluble P-selectin). Assay methods and statistical analysis are detailed in Supplementary Methods.

Results

Between March 2021 and January 2023, five donor hearts were preserved using OCS and transplanted (Supplementary Table S1). NEVHP duration ranged from 240 to 344 min, resulting in variable intervals between T2 (60 min) and T3 (pre-cooling) across perfusion runs. Recipient characteristics are reported in Supplementary Table S2. Severe primary graft dysfunction occurred in two of five recipients (40%) according to ISHLT criteria (Supplementary Table S3).

Biomarker dynamics are summarized in a fold-change heatmap (Figure 1), with full trajectories, model-based geometric means and adjusted P values provided in Supplementary Tables S4 and S5.

Figure 1.

Fold-change in plasma biomarker levels relative to baseline (T0). Heatmap showing fold-changes at T1 (10 min), T2 (60 min), and T3 (pre-cooling) relative to baseline (T0, pre-instrumentation after priming) for inflammatory, endothelial, and haemostasis biomarkers during ex vivo normothermic perfusion. Values are estimates from linear mixed-effects models with random intercept for OCS run, back-transformed from the log scale. Darker red shading indicates larger increases relative to T0; values close to 1 indicate no change. Abbreviations: VWF:Ag, von Willebrand factor antigen; sE-selectin, soluble E-selectin; sP-selectin, soluble P-selectin; PlGF, placental growth factor; VEGFR2, vascular endothelial growth factor receptor-2; bFGF, basic fibroblast growth factor; MCP-1, monocyte chemoattractant protein-1; IFN-γ, interferon-γ; TNF-α, tumour necrosis factor-α

Inflammatory mediators increased markedly over time (Supplementary Figures S2 and S5). By pre-cooling (T3), fold-changes vs baseline (T0) were IL-8 11.5 (95% CI, 6.5–20.1), bFGF 8.4 (6.7–10.5), IL-6 4.5 (2.4–8.2), and MCP-1/CCL2 4.1 (2.6–6.6). IL-10 increased to 2.0 (1.2–3.5) and TNF-α to 1.6 (1.1–2.3). IFN-γ and c-Kit showed no consistent trends. Early change was evident for bFGF at 10 min (T1) with 5.6 (4.4–7.0) vs T0. Leukocyte counts were stable across time, with a mean T0 of 0.57 (0.12–2.75) × 10⁹/L and a T3-to-T0 fold-change of 0.90 (0.52–1.58).

Endothelial markers showed smaller shifts (Supplementary Figures S3 and S6). At T3, angiopoietin-2 reached 1.6 (1.2–2.1) and VEGF 2.0 (1.2–3.5). VWF:Ag slightly increased at 60 min (T2) with a fold-change of 1.1 (1.0–1.1) without a clear increase at T3. Angiopoietin-1, Syndecan-1, soluble E-selectin, thrombomodulin, VEGFR2, and PlGF showed no consistent time trends.

Despite high unfractionated heparin levels of 6.9 [IQR, 4.9–9.9] IU/ml, haemostasis markers demonstrated modest but significant changes (Supplementary Figures S4 and S7). At T3, D-dimer was 1.9 (1.3–2.7), fibrin monomer 2.2 (1.2–3.9), and soluble P-selectin 1.5 (1.1–2.0); platelet counts decreased to 0.8 (0.7–0.9).

Haematocrit was used to quantify haemodilution during perfusion. At T0, mean haematocrit was 15.47% (10.97–21.82), and the T3-to-T0 fold-change was 0.94 (0.92–0.96), indicating mild haemodilution. Increases in soluble biomarkers therefore occurred against a slightly diluted background.

Discussion

This pilot study evaluated inflammatory, endothelial, and haemostatic responses during normothermic ex vivo heart perfusion of human donor hearts. Sequential sampling from the perfusion circuit was feasible and showed temporal changes in circulating biomarkers. Baseline samples (T0) reflected primed, warmed, anticoagulated, and haemodiluted donor blood. Leukocyte, platelet, and haematocrit levels were low at T0, and baseline biomarker values likely reflected both donor and procurement factors. For this reason, results were expressed as fold-changes from T0 to describe temporal dynamics rather than absolute concentrations.

A clear inflammatory response developed during perfusion despite leukocyte depletion and corticosteroids administration. Pro-inflammatory cytokines and chemokines (IL-8, IL-6, MCP-1/CCL2) increased over time, indicating activation of pathways associated with ischaemia–reperfusion injury. The early rise in bFGF suggests a rapid stress-repair response after ischaemic injury.⁸ The combined rise of CCL2 and IL-8 is consistent with chemotactic signalling that would favour recruitment of monocytes and neutrophils .⁹ Given the low circulating leukocyte content in the perfusate, the coronary endothelium is likely the dominant source of these mediators. Activated endothelial cells are known major producers of IL-8 and important contributors of IL-6, which is involved in myeloid trafficking and activation during reperfusion injury.^7,10

Endothelial markers showed a pattern of activation, with increases in angiopoietin-2 and VEGF without a clear signal of major glycocalyx shedding or structural endothelial disruption over the short perfusion period. Angiopoietin-2, rapidly released from endothelial Weibel-Palade bodies upon activation, increased at T3 .¹¹ In contrast, angiopoietin-1, constitutively secreted by perivascular cells, showed no consistent change. The rise in VEGF also likely reflects initial ischaemia-driven transcriptional upregulation. The concurrent stability of syndecan-1 and thrombomodulin, markers requiring proteolytic cleavage or sustained injury for significant release and of soluble VEGFR2, whose shedding is tonically limited by constitutive receptor endocytosis, supports a pattern of endothelial activation without structural disruption. The absence of significant vWF rise may reflect dilution by donor-derived vWF already present in the perfusate at baseline, which compresses the detectable fold-change. Soluble E-selectin secretion requires de novo protein synthesis with peak expression beyond the present perfusion window, likely explaining the observed stability. This pattern differs from reports of marked glycocalyx degradation in ischaemia–reperfusion injury ,¹² suggesting that rheology during NEVHP may be relatively preserved under current perfusion conditions. However, interpretation is limited by short exposure time, haemodilution, and the restricted coronary endothelial surface exposed to the circuit (<1% of systemic endothelium), which likely attenuates signal amplitude.

Haemostatic markers showed low-grade activity despite high levels of unfractionated heparin. D-dimer, fibrin monomer, and soluble P-selectin increased, with a mild decline in platelet counts, suggesting ongoing thrombin generation and platelet activation. This pattern is consistent with contact activation and shear-mediated haemostatic activation described in extracorporeal circuits.¹³ Additional thrombo-inflammatory mechanisms, not evaluated in this study, are likely involved. Interactions between platelets and leukocytes are recognized drivers of coagulation, inflammation, and microvascular dysfunction.¹⁴ In addition, neutrophil extracellular traps (NETs) can promote thrombosis, induce endothelial activation, and amplify inflammatory signalling during reperfusion injury.¹⁵ We did not assess NET-related biomarkers or platelet–leukocyte aggregates. These measurements could be included in future studies to better characterize thrombo-inflammatory processes during NEVHP.

Despite dilution, the short observation window, and exclusive coronary exposure, sequential biomarker assessment during NEVHP was feasible and showed temporal biomarker changes during perfusion. The clinical relevance of inter-graft variability requires evaluation in larger cohorts. These findings support the feasibility of future studies integrating biological profiles with perfusion parameters, graft selection, and post-transplant outcomes.

This pilot study has limitations. The sample size was small, perfusion duration was short, and the biomarker panel was limited. The timing of the final sampling point varied between perfusion runs. Haemolysis was not measured, although it may contribute to endothelial activation during ex situ perfusion. Metabolic profiling was not performed and could have provided complementary information. Tissue or imaging correlates were not available. Because of these limitations, mechanistic conclusions cannot be drawn and associations with post-transplant outcomes cannot be established. Future studies should improve biological profiling, including assessment of NET-related markers, platelet–leukocyte aggregates, and more detailed endothelial measurements, and evaluate modifications of circuit design and perfusate composition. These approaches could improve characterization of biological processes during NEVHP and inform preservation strategies.

Conclusion

Sequential biomarker monitoring during NEVHP was feasible and demonstrated inflammatory, endothelial, and haemostatic changes. These biological patterns require confirmation in larger cohorts, as potential tools for graft assessment and optimization of perfusion circuits and perfusate composition.

Supplementary data

Supplementary data are available at ESC Heart Failure online.

Author contributions

A.M.: Conceived the study, elaborated the analysis plan, contributed to the investigation and data collection, performed the statistical analysis, analysed and interpreted the data, wrote the first draft of the manuscript, designed the tables and figures, revised the manuscript, approved the final work and agree to be accountable for all aspects of the research and for the accuracy and integrity of the work. N.P.P.: Conceived the study, elaborated the analysis plan, contributed to the investigation and data collection, performed the statistical analysis, analysed and interpreted the data, wrote the first draft of the manuscript, designed the tables and figures, revised the manuscript, approved the final work and agree to be accountable for all aspects of the research and for the accuracy and integrity of the work. J.F.: Contributed to the investigation and data collection, provided critical revisions to the manuscript, approved the final work and agree to be accountable for all aspects of the research and for the accuracy and integrity of the work. I.G.: Contributed to the investigation and data collection, provided critical revisions to the manuscript, approved the final work and agree to be accountable for all aspects of the research and for the accuracy and integrity of the work. E.F.: Contributed to the investigation and data collection, provided critical revisions to the manuscript, approved the final work and agree to be accountable for all aspects of the research and for the accuracy and integrity of the work. N.N.: Conceived the study, elaborated the analysis plan, contributed to the investigation and data collection, performed the statistical analysis, analysed and interpreted the data, wrote the first draft of the manuscript, designed the tables and figures, revised the manuscript, approved the final work and agree to be accountable for all aspects of the research and for the accuracy and integrity of the work

Declarations

Disclosure of Interest

A.M. received payments made to his institution from i-SEP for consulting fees, and from LFB, Aguettant, Viatris, and Pfizer for lecture fees. The other authors declare no competing interests.

Data Availability

No data were generated or analysed for this manuscript.

Funding

This work was supported by grants from Association Ouest-Transplant (France) and Fédération Hospitalo-Universitaire SUPORT (France).