Cutaneous Mitochondrial Po₂, but Not Tissue Oxygen Saturation, Is an Early Indicator of the Physiologic Limit of Hemodilution in the Pig

NORMOVOLEMIC hemodilution is a consequence of fluid administration to compensate blood loss and occurs commonly during major surgery, cardiopulmonary bypass interventions, and emergency medicine. One of the risks of hemodilution is tissue hypoxia due to a critical decrease in oxygen supply when hemoglobin levels drop below the individual-dependent threshold. Previous studies have shown serious consequences of insufficient oxygen supply, like stroke,¹  declined cognitive function,²  kidney injury,^3,4  and cardiac complications.^5,6  Oxygen supply is commonly improved by the transfusion of erythrocytes. However, transfusion should be restricted to a minimum due to the risks involved in it, like bacterial^7,8  and viral infection,⁹  transfusion-related acute lung injury,¹⁰  transfusion-associated circulatory overload,¹¹  and febrile nonhemolytic and allergic reactions.^12,13  Despite the incidence of the previously mentioned complications having decreased significantly over the last decades, caution must be exercised due to the potential severity of complications.

The above-mentioned risks clearly indicate the need for a reliable and practical assessment of tissue oxygenation during hemodilution to support the decision for blood transfusion. Hemoglobin level determines the oxygen transport capacity of blood, and therefore is used as a trigger for blood transfusion. However, the level of hemoglobin below which oxygen supply becomes insufficient for the tissue, also known as “critical hemoglobin” or “critical hematocrit,” is not unambiguous.¹⁴  The balance of oxygen supply and consumption depends not only on the amount of erythrocytes, but also on cardiac output, gas exchange, and metabolic demand. As a consequence, critical hemoglobin differs interindividually and even per organ.^15,16  Compensatory mechanisms differ between individuals and so does the tolerance to anemia.¹⁷ 

For this reason, direct measurement of tissue oxygenation, and most preferably intracellularly at the mitochondrial level, could be of advantage next to the measurement of an indirect parameter like hemoglobin. Mitochondrial oxygen tension (mitoPo₂) is a very important parameter for cellular function. With the protoporphyrin IX (PpIX)-triplet state lifetime technique, it is possible to measure mitoPo₂in vivo. This technique has been used successfully in vivo and in vitro in several studies.^18–20 

In particular, monitoring of mitoPo₂ in the skin seems to be of special interest since the skin can be regarded as the canary of the body,²¹  like the gastrointestinal tract.²²  Studies demonstrated that detection of subcutaneous oxygen tension could provide essentially the same information as invasively measured parameters of oxygen metabolism in the gut.^23,24  In addition, the skin can be easily noninvasively monitored with an optical technique in the clinic.

The aim of this study was to investigate the influence of hemodilution on cutaneous mitoPo₂, compared to other common clinical parameters (mean arterial blood pressure [MAP], cardiac output, systemic oxygen consumption, and signs of anaerobic metabolism like lactate). To this end, mitoPo₂ measurements were related to hemoglobin level, mixed venous oxygen saturation, and systemic biochemical and physical signs of hypoxia. In addition to mitoPo₂, we also measured tissue oxygen saturation (StO₂) on the thoracic wall as an already clinically available alternative. Our hypothesis was that ongoing normovolemic hemodilution would cause a decrease in cutaneous mitoPo₂ when the individual limits of compensatory mechanisms would be reached. The limits of these compensatory mechanisms are embodied by the parameters described in this paragraph.

The protocol of the current study was approved by the Animal Research Committee of the Erasmus Medical Center, University Medical Centre Rotterdam, Rotterdam, The Netherlands (EMC protocol no. 129-13-05). Animal care and handling were performed in accordance with the national guidelines for the care of laboratory animals. For the experiments, 26 female Yorkshire pigs with mean body weights of 31.3 ± 0.3 kg (mean ± SD) were used.

The animals were sedated with an intramuscular injection of tilatamine/zolazepam (2.5/2.5 mg/kg; Virbac Laboratories, France) and xylazine (2.25 mg/kg; AST Farma B.V., The Netherlands). After a 15-min induction period, an intravenous access was obtained in the left-ear vein with a 20G Venflon (Becton, Dickinson and Company, USA), and tracheal intubation was performed with a size 6.5 Portex^® endotracheal tube (Smiths Medical International Ltd., United Kingdom). For maintenance of anesthesia, the animals received continuous infusion of ketamine (5 mg kg⁻¹ h⁻¹; Alfasan Nederland B.V., The Netherlands), midazolam (1.5 mg kg⁻¹ h⁻¹; Atavis Group PCT, Iceland), sufentanil (4 μg kg⁻¹ h⁻¹; Janssen-Cilag B.V., The Netherlands), and rocuroniumbromide (4 mg kg⁻¹ h⁻¹; Fresenius Kabi Austria GmbH, Austria). All animals received continuous infusion of crystalloid (Sterofundin^® ISO 10 ml kg⁻¹ h⁻¹; B. Braun, Germany). Each pig received a bolus of magnesium sulphate (500 mg; Pharmachemie, The Netherlands), as arrhythmia prophylaxis, added to the first bag of crystalloid solution. Animals’ temperature, electrocardiogram, and Spo₂ were continuously monitored.

To allow the measurement of mitoPo₂, we first shaved 2 × 3 cm of the anterior chest wall bilaterally to remove hair. Second, we applied freshly prepared 5-aminolevulinic acid (ALA) cream (20%, 1 g on each side; 5-aminolevulinic acid hydrochloride, Fargon GmbH & Co. KG, Germany, in Lanette cream, Teva Nederland B.V., The Netherlands) to the shaved skin. The cream was covered by an IV3000 plaster (Smith & Nephew, United Kingdom) and by a layer of aluminum foil.

Pressure-controlled mechanical ventilation (Servo 300; Siemens-Elema, Sweden) was used with an Fio₂ of 0.4. Adjustments were made to maintain normocapnia (ETco₂: 35 to 45 mmHg). Temperature was maintained between 38° and 39°C, as measured rectally, with heating pads underneath the animal.

For continuous measurement of the arterial blood pressure, a 20G catheter (Arterial Leadercath, Vygon^®, France) was placed in the left femoral artery by using the Seldinger technique. For blood withdrawal during hemodilution, a 9Fr introducer sheath (Arrow International Inc., USA) was placed in the right femoral artery. For monitoring of the cardiac output, another sheath in the right jugular vein was used as an introducer for placement of a Swan-Ganz catheter (Edwards Lifesciences, USA) in the pulmonary artery. The side port of this sheath was used for fluid administration. A lower midline laparotomy was performed in order to insert a catheter in the urinary bladder.

On the left lateral part of the chest wall, a near-infrared spectroscopy sensor was attached (InSpectra StO₂ Monitor, Hutchinson Technology, USA) for continuous measurement of tissue oxygenation (StO₂) at three different depths (2.5, 15, and 25 mm). In figure 1, an overview of the preparation is given.

Throughout the entire experiment, MAP, heart rate, continuous cardiac output (CCO), Spo₂, StO₂, and end-tidal carbon dioxide were continuously monitored. At baseline and at each step of the experimental protocol, an arterial and mixed-venous blood sample was obtained for measurement of Po₂, PCO₂, SO₂, lactate, pH, hemoglobin, and hematocrit. These measurements were performed with an ABL 800Flex (Radiometer, Denmark). From CCO, Pao₂, arterial oxygen saturation, mixed venous oxygen tension, and mixed venous oxygen saturation, we calculated the systemic oxygen consumption () by the Fick principle ( = CCO × (Cao₂ − Cvo₂). Oxygen delivery (DO₂) was calculated using the CCO and arterial oxygen content (DO₂ = CCO × Cao₂).

For the measurement of mitoPo₂, we used an optical technique based on the measurement of the delayed fluorescence lifetime of PpIX. A detailed description of the technique can be found elsewhere.^20,25  In short, we applied ALA cream to induce enhanced levels of PpIX in the cutaneous mitochondria (priming of the skin). By using a pulsed laser, we illuminated the primed skin at a wavelength of 510 nm (green) for photoexcitation of PpIX. With a gated photomultiplier tube, we collected the emitted red delayed fluorescence light. Instead of relaxation to the ground state by emission of a photon, excited PpIX can transfer its energy to oxygen (a process known as quenching), making the delayed fluorescence lifetime oxygen dependent. Due to these properties, we could calculate the mitoPo₂ from the lifetime of delayed fluorescence using the Stern–Volmer equation.²⁶ 

A total of 26 animals were divided into two groups, a hemodilution group (n = 12) and a TC group (n = 14), through randomization. Randomization was performed using “random numbers” in Microsoft Excel (Microsoft, USA). Randomization was not blinded because it is not feasible in the experimental design. Due to animal- or technology-related problems, we had to exclude three animals in the TC and two animals in the hemodilution group. Three hours after surgery and ALA cream application, the aluminum foil, the plaster, and the excess cream were removed. Afterward, we covered the prepared skin with heated (±38°C) infusion bags to prevent it from cooling. After 20 more min, which was necessary for the remaining cream to diffuse into the skin, we performed the first (baseline) set of measurements. One measurement consisted of a set of 10 mitoPo₂ measurements and an arterial and mixed-venous blood sample. In all animals of the hemodilution group, we withdrew 500 ml of blood via the sheath in the right femoral artery after the baseline measurement. Simultaneously, we infused 500 ml of heated (±38°C) colloid solution (Voluven^®; Fresenius Kabi AG, Germany) using a pressure bag (200 mmHg) via the right jugular vein. Each hemodilution step was followed by a 20-min stabilization period. At the end of a stabilization period, a new measurement was performed, followed by a new hemodilution step. A total of eight hemodilution steps and measurements were performed. The animals of the TC group were not subjected to hemodilution. In those animals, a baseline and eight time-related measurements were performed (hemodilution 1, hemodilution 2, etc.). At the end of the experiment, the animals were euthanized with an overdose of potassium chloride (40 mM, Fresenius Kabi AG, Germany). A visual overview of the experimental protocol is given in figure 1.

The number of animals used was based on previous experience in pilot studies. Data are presented as mean ± SD, unless otherwise indicated. When we were interested in group differences and therefore in minimizing the overlap of error bars between the groups, we used mean ± SEM²⁷  to improve interpretation in our figures. Normal distribution was assessed both visually and using Kolmogorov–Smirnov testing. For comparison of animal characteristics and changes between both groups, we used the unpaired Student’s t test. To examine the effect of hemodilution over time on mitoPo₂ and other relevant parameters, linear mixed-effect models were used. The models included subject as random effect and hemodilution, time and hemodilution * time as fixed within-subject effects. Linear mixed-effect models provide unbiased and efficient estimates (assuming data are missing at random) and can accommodate intermittently missing values. For post hoc analysis of the timing of occurrence of significance, we used paired Student’s t test. Correction for repeated measurements was performed with Bonferroni correction. Statistical analysis was performed with SPSS Statistics (Version 23, SPSS, USA). Graphics were produced with GraphPad Prism 6 (GraphPad Software Inc., USA).

All data of the 11 TC and 10 hemodilution animals are presented in our data table (table 1). In the hemodilution group, four animals did not survive all the eight hemodilution steps. We did not exclude these animals because it is likely that the sensitivity toward hemodilution differs per animal. In the TC group, the number of animals is constant (n = 11). Due to the fact that four animals in the hemodilution group did not survive all eight hemodilution steps, the number of animals in this group is lower at hemodilution step 7 (n = 8) and step 8 (n = 6). In the last column, we presented the results at the terminal point for some of the parameters. The terminal point represents the value of the concerning parameter at the point the animal is pre terminal. Because not all of the animals died at the same time point, these values represent the condition of the animal at its most critical hemoglobin level.

In table 1, the course of the hemoglobin level is shown for both the TC and hemodilution groups. As expected, we see a stable hemoglobin level in the TC group. In contrast, in the hemodilution group, we observed a significant decrease in hemoglobin level during hemodilution, with hemoglobin showing an exponential decline in the first hemodilution steps and leveling off around a value of 2 g/l in the final hemodilution steps. Systemic compensation for the loss in oxygen-carrying capacity of the blood is apparent from hemodilution 4 as both heart rate and CCO increase. Nevertheless, DO₂ decreases gradually from the start of hemodilution and becomes significantly lower compared to the baseline at hemodilution 3 (417 ± 85 vs. 312 ± 38; P = 0.0033).

As shown in figure 2, based on linear mixed-effect models, there is no significant time effect in mitoPo₂ in the control group. In the hemodilution group, we found a decreasing trend in mitoPo₂ after a number of hemodilution steps, which reaches significance level starting at hemodilution 4 (23.6 ± 5.3 vs. 14.8 ± 4.3; P = 0.0033). Because four hemodilution animals died before they reached the eighth hemodilution step, the eighth point in this figure does not offer an adequate view on the mitoPo₂ at the point the animal dies. To create this view, we took all the terminal values of the animals that died during the hemodilution and plotted these values as “terminal.”

In figure 3, the mitoPo₂ can be compared to the hemoglobin level and other parameters that are known or expected to be an indicator for reaching the physiologic threshold of hemodilution. The dotted line is placed at the point where mitoPo₂ started to become significantly lower than the baseline. It is evident that this event preceded the trends in increase in lactate, drop in MAP, and drop in by two hemodilution steps. Importantly, none of the other parameters changed significantly unless the animals were pre terminal (table 1). While lactate, MAP, and at least showed a declining trend at the final hemodilution steps, StO₂ did not show a response.

After calculating the (fig. 4A), we found only a significant difference at the terminal point. However, the DO₂ (fig. 4B) shows a significant decrease with ongoing hemodilution from the fourth hemodilution step onward. When we made the comparison between and DO₂ (fig. 4C), we found that the tends to remain stable during decrease in DO₂ in the first hemodilution steps. At the sixth hemodilution step, the tends to decrease during progression of subsequent hemodilution.

The previous data presentation clearly shows the dependency of mitoPo₂ on hemoglobin after hemoglobin dropped below a certain threshold. While mitoPo₂ appears to decline gradually with ongoing hemodilution, on the individual level, mitoPo₂ typically dropped much more acute. Figure 5A shows an example in an animal in which mitoPo₂ already showed a steep decline from 24 mmHg to 8 mmHg in one hemodilution step, relatively early in the protocol (between hemodilution 3 and 4). This acute drop preceded hemodynamic instability in the next hemodilution steps, and the animal did not survive after hemodilution 6. Therefore, to create a somewhat better visual presentation of the real course of mitoPo₂ during hemodilution, we aligned the individual curves of the experimental animals on their deflection point. In figure 5, we presented the mean ± SEM of these realigned curves.

Our data support our main hypothesis that a decrease in mitoPo₂ is a direct and early indicator for the development of tissue hypoxia due to failing compensation mechanisms during ongoing hemodilution. The fact that cutaneous mitoPo₂ decreased abruptly at a certain stage of hemodilution instead of showing a more gradual decline was a surprising finding. This could prove a very valuable behavior of mitoPo₂ when aiming to use it as physiologic transfusion trigger. The unexpected nonresponse of StO₂ shows that measuring quantitative Po₂ at the intracellular level provides different and complementary data.

The measurement of mitoPo₂ during hemodilution has never been performed before. The significant decrease in mitoPo₂ with ongoing hemodilution is in line with previous microvascular Po₂ (μPo₂) measurements by van Bommel et al.¹⁶  Measuring μPo₂ has the disadvantage that it is not suitable for clinical use because the oxygen sensor dye Pd-porphyrin is exogenous and potentially nephrotoxic. Furthermore, after realignment of the mitoPo₂ data of the individual hemodilution animals, we found an even more convincing abrupt decrease in mitoPo₂ with decreasing hemoglobin levels. This suggests a strong relationship between the hemoglobin level and cutaneous mitoPo₂ in the pig.

When mitoPo₂ is compared with other parameters proposed to be the indicators of reaching the physiologic limit of hemodilution, it is found that mitoPo₂ values are decreasing in an earlier phase of the hemodilution. When MAP, lactate level, , and DO₂ start to decrease, there is already anaerobic metabolism as indicated by increased lactate levels. Directly measuring the detrimental effects of hemodilution on cutaneous mitochondrial oxygenation provides a means to detect the individual limit without systemic signs of decompensation. Before the experiments were performed, we assumed that the StO₂ measured by near-infrared spectroscopy could be an early indicator as well. However, during the experiments, the StO₂ remained unaltered until the animals became hemodynamically unstable. So the decrease in StO₂ appears more related to the overall hemodynamic collapse than the reduced hemoglobin level per se.

Looking at the and DO₂ data, we found results comparable with the results published by van Bommel et al.²⁸  Pearce et al.²¹  presented an illustrative picture of the relationship between DO₂ and in which the remains stable during hemodilution while the DO₂ decreases. When DO₂ decreases further, it will eventually compromise . In our experiments, we have found a similar, but less obvious, trend in and DO₂, with only becoming significantly reduced in the nonsurviving animals. In a continuous and relatively fast hemodilution protocol in pigs, a critical hemoglobin (Hb_crit) value of 2.7 g/l has been reported by Lauscher et al.¹⁵  Although it is difficult to directly compare our results to this study, from table 1, it appears that the trend in total body starts to decline somewhere between hemodilution 5 and 6. At this point, the trend in lactate also starts to increase. This would correspond with a Hb_crit around 2.4 to 2.5 g/l, which is not far from the previously reported value of 2.7 g/l. The difference might be attributed to the contrasting hemodilution protocols (stepwise vs. continuous) and, probably even more important, the different Fio₂ values (0.4 and 0.21, respectively). The fact that we used a stepwise form of hemodilution could be a limitation of our study design. A continuous form of hemodilution, as for example, used by Lauscher et al.,¹⁵  is more comparable with clinical practice. In this study, we chose a stepwise form to be able to perform measurements in-between the successive hemodilution steps.

With respect to Hb_crit and the critical hematocrit (cHct), it is important to realize that reported values in animal studies differ somewhat per species, with values for cHct ranging from 10 to 15%.^14,29–31  Pigs are clearly on the lower end of the spectrum, with recent studies reporting cHct and Hb_crit around 10%¹⁴  and 2.7 g/dl,¹⁵  respectively. Likewise, in our study, mitoPo₂ dropped significantly between hemodilution steps 3 and 4, the point where hematocrit decreased below 11%. Thereafter, the animals became hemodynamically unstable during ongoing hemodilution. Due to interspecies differences, care should be taken when translating results of animal experiments to humans. Pigs are relatively anemic (normal hemoglobin in the range 8 to9 g/dl), and in humans, the reference range for hemoglobin is twice as high. Hb_crit in humans is also different. For example, in healthy volunteers, a slight reversible cognitive dysfunction was seen during acute anemia at hemoglobin concentrations as high as 5 to 6 g/dl.³² 

Following the concept of previous experimental studies (e.g., Lauscher et al.¹⁵ ; Meier et al.²⁹ ) in our study design, a decrease in total body oxygen consumption was interpreted as the absolute limit of hemodilution. MitoPo₂ reacts abruptly well before reaching this point, and therefore, it might be a valuable parameter for indicating an individual’s transfusion need. This tempting idea needs thorough further evaluation, especially considering the fact that vital organs show different dependencies on hemoglobin level.^15,16,28  Future studies should, therefore, next to incorporation of transfusion groups, include the evaluation of a set of clinically relevant endpoints regarding organ function and damage.

The different sensitivities of organs to anemia pose a challenge when considering a local mitoPo₂ measurement as a potential transfusion trigger. One has to look at an organ that is one of the first to get into trouble and one of the last to normalize. Based on previous findings, the skin certainly qualifies as such an organ. Using phosphorescence-quenching technology, simultaneous microvascular Po₂ (μPo₂) measurements on different organs have been performed in the past,¹⁶  showing intestinal μPo₂ to decrease earlier during hemodilution compared to cardiac μPo₂ in rats. In pigs, it was demonstrated that during hemodilution intestinal serosal μPo₂ became impaired at an earlier stage than cerebral μPo₂.²⁸  Since oxygenation measurements in the skin can provide similar information to intestinal measurements,^23,24  the skin is hypothesized to be a valuable early indicator (“canary”).

However, whether the above assumption of mitoPo₂ in the skin being a valuable canary holds for all organs, especially the kidney, remains to be seen. In previous studies in rats, μPo₂ dropped in very early stages of hemodilution in which hematocrit level did not even decrease clinically significantly.^16,33  In pigs, Konrad et al.³  showed that normovolemic hemodilution to a hematocrit of 15% with a crystalloid significantly impaired renal function. Importantly, in the latter study, renal function loss seemed a result of tissue edema formation and not of anemia per se since hemodilution with a colloid preserved renal function. Also, biases due to other factors influencing skin perfusion and oxygenation (like temperature and centralization of the circulation) might be potential limitations for the use of cutaneous mitoPo₂ as a transfusion trigger in clinical practice.

In conclusion, looking at individual experiments, there is an abrupt decrease in mitoPo₂ with ongoing hemodilution. Furthermore, mitoPo₂ precedes changes in other parameters, which could make mitoPo₂ a clinically useful physiologic trigger for the need for erythrocyte transfusion. The concept of using cutaneous mitoPo₂ as a potential individual blood transfusion trigger is promising but obviously needs further research.

The authors would like to thank Wilhelmina J. Voorbeijtel, B.Sc., and Patricia A. C. Specht, B.Sc., Laboratory of Experimental Anesthesiology, Erasmus University Medical Center Rotterdam, Rotterdam, The Netherlands, for their skillful technical assistance, which was very important for the success of the experiments. Furthermore, the authors would like to thank Ankie W. M. M. Koopman-van Gemert, M.D., Ph.D., Albert Schweitzer Hospital, Dordrecht, The Netherlands, for reading and commenting the manuscript.

Supported by the Eurostars Programme as Eurostars Project E!7294 COMET (Eurostars, Brussels, Belgium).

Dr. Mik is founder and shareholder of Photonics Healthcare B.V. (Utrecht, The Netherlands), a company aimed at developing a clinical monitoring device based on the delayed fluorescence lifetime technology for measuring mitochondrial oxygen. Photonics Healthcare B.V. holds the exclusive licenses to several patents related to this technology, filed and owned by the Academic Medical Center in Amsterdam and the Erasmus Medical Center in Rotterdam, The Netherlands. The other authors declare no competing interests.