*3.4. Liver Metabolism During NMP*

During the normothermic phase, the metabolic function of the liver was evaluated to assess the performance of the two different types of perfusate (Table 2). The DO2 of the BLOOD group was higher than in DMEM (DO2: DMEM 0.405 ± 0.015 mL/min vs. BLOOD 1.685 ± 0.090 mL/min, *<sup>p</sup>* <sup>&</sup>lt; 0.001) (Figure 7A). Equally, . *VO*<sup>2</sup> was higher in BLOOD group (*<sup>p</sup>* <sup>=</sup> 0.001). While DMEM . *VO*<sup>2</sup> did not change through the procedure, the BLOOD . *VO*<sup>2</sup> decreased over time (Figure 7B). The energetic load of liver tissue decreased during the 30 min of cold storage (native 1354.2 ± 109.1 pmol/mg vs. SCS 985.9 ± 102.6 pmol/mg, *p* = 0.046). At the end-NMP, ATP levels decreased independently of the presence of OxC (DMEM 327.3 ± 26.6 pmol/mg; BLOOD 565.8 ± 56.6 pmol/mg) compared to both SCS (DMEM vs. SCS, *p* = 0.007 and BLOOD vs. SCS, *p* = 0.012) and native (DMEM vs. native *p* = 0.001 and BLOOD vs. native *p* = 0.003) group, as indicated by a bioluminescent assay of liver homogenates. However, after 120 min of normothermia the grafts perfused with a blood based perfusate showed a statistically significant higher ATP (*p* = 0.031) (Figure 5C). Glucose during NMP was stable in DMEM group, while decreased in BLOOD group (*p* < 0.001), resulting in a greater concentration of glucose in perfusion fluid of DMEM group at the end of the experiment compared to the BLOOD group (*p* = 0.008) (Figure 7C). Glucose uptake ratio was higher in DMEM (*p* = 0.029). Lactates were higher at each time point (*p* = 0.002) and lactates uptake ratio was lower in DMEM (DMEM 0.062 ± 0.359 vs. BLOOD 0.643 ± 0.027, *p* = 0.029) (Figure 7D). Interestingly, during the ischemic phase (DO2 = 0 mL/min),

the estimated glucose uptake ratio was −0.280 ± 0.116, while the lactates uptake ratio was −0.180 ± 0.89. Citrate uptake ratio was 0.228 ± 0.127 in BLOOD group (it was not detectable in DMEM).

**Figure 7.** The use of human red blood cells as oxygen carrier resulted in an increased DO2 in BLOOD group (\* *<sup>p</sup>* <sup>&</sup>lt; 0.05) (**A**) and a consequent improved . *VO*<sup>2</sup> (\* *p* < 0.001) (**B**). Glucose was absorbed during normothermic phase in BLOOD group (\* *p* < 0.001) (**C**) and lactates metabolism was increased during the whole procedure in BLOOD group (# *p* < 0.05) (**D**). These parameters suggest an improved metabolic function of the graft when a higher oxygen content was provided during perfusion. Re, rewarming; DO2, delivery of oxygen; . *VO*2, oxygen consumption. Data are shown as mean ± SEM.

The graft . *VO*<sup>2</sup> (*R*<sup>2</sup> = 0.929; *p* < 0.001) (Figure 8), the end-NMP tissue ATP content (*R*<sup>2</sup> = 0.622; *p* = 0.020) and the ability of the liver to clear lactates (*R*<sup>2</sup> = 0.609; *p* = 0.022) and glucose (*R*<sup>2</sup> = 0.718; *p* = 0.008) were directly related to the amount of oxygen delivered (DO2). Equally, an increase in oxygen consumption ( . *VO*2) led to an increase in lactates clearance (*R*<sup>2</sup> = 0.505; *p* = 0.048) and glucose consumption (*R*<sup>2</sup> = 0.597; *p* = 0.025), while ATP levels showed only a trend toward a significant linear correlation with . *VO*<sup>2</sup> (*R*<sup>2</sup> = 0.454; *p* = 0.067). An increase of 1 pmol/mg of liver ATP produced an increase in lactates uptake ratio of 0.002 <sup>±</sup> 0.001 (*R*<sup>2</sup> = 0.641; *p* = 0.017) and glucose uptake ratio of 0.003 <sup>±</sup> 0.001 (*R*<sup>2</sup> <sup>=</sup> 0.619; *<sup>p</sup>* <sup>=</sup> 0.021).

**Figure 8.** Linear regression between DO2 and . *VO*<sup>2</sup> levels during ex-situ normothermic perfusion.

#### **4. Discussion**

The present research demonstrates that an optimized DO2 reduces liver hypoxic damage during NMP. This significant result was achieved using human RBCs as oxygen carriers. Therefore, in addition to the beneficial effects on metabolic parameters, the study points at the use of human RBCs as a strategy to substantially reduce the number of animals employed in accordance with the 3Rs principles.

NMP is increasingly used in clinical settings [28–30] to evaluate, recondition, and preserve liver grafts, and this key strategy could be substantially improved by preclinical research, in particular by rodent models. Further, the use of NMRP, rather than LT to evaluate IRI without the use of a recipient, could reduce the number of animals employed. However, to expand their translational potential, these preclinical models need to be optimized and each experimental step should be carefully implemented [21].

We examined the influence of different perfusion solutions and the use of an oxygen carrier on graft reperfusion and viability. Results show that the addition of human RBCs [12,13] to NMP perfusate increases oxygen delivery and causes a faster and steeper restoration of liver metabolism. Indeed, the increased DO2, achieved through the use of human RBCs, led to a more rapid clearance of lactate and partially counteracted the reduction in ATP content at the end of NMP.

Mischinger and colleagues [31] suggested in 1992 that ex-situ normothermic perfused livers do not need high DO2. Based on this view, NMP set-up could be simplified by removing the OxC; namely, erythrocytes. Indeed, the general idea was that during NMP there was reduced ex-situ metabolic activity and a mitochondrial inhibition due to hypoxia-induced factors [32,33]. Conversely, other authors suggested that an optimized DO2 during perfusion could result in a faster graft recovery after ischemia [34,35]. Autologous rat or human red blood cells can be used as oxygen carriers to increase DO2. If rat blood is used, 3 to 4 donors are required to obtain a 15% hematocrit perfusate. Human blood enriched perfusate was used in ex-situ perfusions [12,13] of rat liver grafts, but the authors reported no clear advantage by using an OxC. Further, some papers showed a detrimental impact of human blood used during experiments on different animals [36].

In the present research, the use of human red blood cells and a dedicated membrane lung led to a higher DO2 and a subsequent higher . *VO*<sup>2</sup> in the BLOOD group. Interestingly, our . *VO*<sup>2</sup> data are markedly higher relative to other reports (Table 3). The importance of <sup>2</sup> and DO2 in our study is further supported by the positive linear regression between the main metabolic parameters and oxygen delivery and consumption. Indeed, early restoration of liver function and increased liver metabolism occurred in the BLOOD group with a better preservation of the energetic pool. The improved glucose and lactate uptake, and the increased glycogen and ATP tissue content, further indicate an upregulated metabolic status in the BLOOD group. The reduced potassium uptake and significant increase of W/D ratio at the end of perfusion (compared to SCS) in DMEM group likely depends on a more severe liver damage [37]. Furthermore, the direct relation between DO2/ . *VO*<sup>2</sup> and metabolic parameters suggests that the total amount of oxygen transported (DO2) results in an increased ability to use oxygen ( . *VO*2) to generate energy (ATP) to sustain hepatocellular metabolism.


**Table 3.** Main metabolic parameters reported in literature during ex-situ normothermic perfusion on rats. Htc, hematocrit; pO2, partial pressure of oxygen in perfusate; . *VO*2, oxygen consumption; RBC, packed red blood cells or centrifugated red blood cells.

These data contradict the assumption of Mischinger and colleagues [31]. Furthermore, we deem that if DO2 is not optimized with an OxC, NMP, or NMRP could result in a suboptimal ex-situ reperfusion procedure. The reason for the disparity between our data and previous observations [32,38] could depend on differences in ex-situ perfusion or other procedural steps.

The importance of adequate DO2 is further highlighted in the rewarming phase. The BLOOD group showed an early . *VO*<sup>2</sup> peak with early lactate normalization. A higher DO2 could help to overcome the oxygen debt accumulated during ischemia [17]. Although high O2 levels during reperfusion could potentially lead to oxidative tissue injury causing a more pronounced tissue damage [39], we did not observe any change in this direction.

The safety of human RBCs used in this model is demonstrated by the absence of endothelial damage, sinusoid obstruction, extravascular hemorrhage, or parenchymal damage. An adequate wash-out of rat blood during perfusion and the virtual absence of human plasma can avoid incompatibility reactions and, therefore, reduce species-specific antibody reactions.

Furthermore, in our study, histological data confirm the preserved graft viability indicated by the metabolic evaluation. As recently suggested [40], we used a combination of different biomarkers to assess graft quality, and some of them (glucose and potassium uptake ratio, citrate clearance, hyaluronic acid clearance) are described in this research for the first time. Their interaction could be used in clinical or preclinical settings to evaluate graft viability.

Some limitations of the study deserve comment. First of all, the duration of the NMP was relatively short if compared to a clinical setting. While accelerated clearance and sinusoidal trapping of human RBCs were observed in rat transfusion models [41], we did not observe significant hemolysis due to human blood use, and no red blood cells were trapped within sinusoidal spaces. However, the exploitation of these types of perfusate should be tested in prolonged perfusion to confirm absence of blood related liver damage. Non-cellular hemoglobin was recently used in clinical NMP [18], and it should be tested in this setting.

In conclusion, this study provides a detailed description of a small animal model of NMP characterized by optimized liver function, liver metabolism, and absence of injury. Results suggest that OxC may be adopted with NMP or NMRP, in particular with expanded criteria grafts, to avoid the detrimental impact of low DO2. In compliance with the 3R principle, human RBCs can be safely used to improve DO2, but additional studies are needed to confirm the safety, effectiveness, and optimal administration protocol.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2077-0383/8/11/1918/s1, Table S1: brief review of ex-situ normothermic liver rat perfusion; Table S2: summary of the main characteristics of normothermic ex-situ perfusion protocols on rats published in literature; Table S3: reagents used during experiments; Table S4: instruments employed during experiments; Table S5: characteristics of the perfusion fluid in the two experimental groups; Figure S1: Schematic representation of experimental workflow; Figure S2: modified Fick equation used to calculate . *VO*<sup>2</sup> and DO2 in our experiments; Figure S3: hepatocellular damage markers during normothermic machine perfusion.

**Author Contributions:** Conceptualization, D.D., C.L. and S.G.; Methodology, C.L., D.D., S.G., A.Z., R.M., A.S.; Formal Analysis, C.L., R.M., M.M., M.B., L.V., O.B., S.V.; Investigation, D.D., S.G., A.S., C.L., R.M., A.V.; Data Curation, A.S., D.D., C.L., R.M., V.L.; Writing—Original Draft Preparation, D.D., C.L.; Writing—Review & Editing, S.G., A.Z.; Supervision, S.G., A.Z.

**Funding:** The present study was supported by Fondazione Ca' Granda Ospedale Maggiore Policlinico, Milan, with the following grants: 5 × 1000–2014 "Ottimizzazione della procedura di perfusione di organi isolati a scopo di trapianto mediante modulazione del liquido di perfusione—Improvement of perfusion of isolated organs before transplantation through perfusion fluid optimization", 5 × 1000–2014 "Il benessere animale nella ricerca preclinica: strategie di implementazione—Implementation of strategies for laboratory animal care".

**Acknowledgments:** We thank "Associazione Italiana Copev per la prevenzione e cura dell'epatite virale Beatrice Vitiello—ONLUS" and "MILTA". Authors would like to thank Samata Oldoni for her valuable help.

**Conflicts of Interest:** The authors declare no conflict of interest.
