The impact of inspired oxygen concentration on tissue oxygenation during progressive haemorrhage
- First Online:
- Cite this article as:
- Dyson, A., Stidwill, R., Taylor, V. et al. Intensive Care Med (2009) 35: 1783. doi:10.1007/s00134-009-1577-2
- 183 Views
Standard resuscitation practice for shock states mandates use of high flow, high concentration oxygen. However, this may induce microvascular constriction and potentially impair regional oxygen delivery. We thus investigated the impact of varying inspired oxygen concentrations in a rat model of progressive haemorrhage.
Tissue oxygen tension (the balance between local O2 supply and demand) was measured in four different organ beds (liver, renal cortex, muscle, bladder), with concurrent assessment of cardiorespiratory function and organ perfusion in a spontaneously breathing, anaesthetised rat model. 10% aliquots of circulating blood volume were removed at 15 min intervals until death. Different oxygen fractions in the gas mixture (0.15–1.0) were administered following 20% blood removal. A control group consisted of normovolaemic animals breathing varying oxygen fractions.
Survival times following progressive haemorrhage were similar in animals breathing room air (98 ± 10 min), 60% O2 (102 ± 6 min) or 100% O2 (90 ± 4 min), but significantly worse in those breathing 15% O2 (52 ± 8 min, P < 0.01). Significant derangements of blood pressure, aortic blood flow and lactataemia were observed in both hypoxaemic and hyperoxaemic groups compared to normoxaemic animals. Breathing 100% O2 increased arterial PO2 sevenfold and tPO2 approximately threefold over baseline values during normovolaemia and mild haemorrhage (20% blood volume removal). However, with progressive haemorrhage, and despite maintained PaO2 values, tissue PO2 fell in line with the decrease in global oxygen delivery.
Hypoxaemia and hyperoxaemia both compromised haemodynamics and biochemical markers of organ perfusion during severe, progressive haemorrhage. This may carry implications for resuscitation practice.
KeywordsHaemorrhageHyperoxiaHypoxaemiaTissue oxygen tensionShockHaemodynamics
High flow, high concentration oxygen therapy is given ubiquitously as part of the standard ‘Airway, Breathing, Circulation’ algorithmic approach to emergency management of the shocked patient. However, the utility of hyperoxaemia as a therapeutic intervention is questionable; across a range of clinical scenarios, both benefit [1, 2] and harm [3, 4] have been reported. As hyperoxaemia can cause microvascular constriction , this may impair local tissue oxygenation and thus compromise organ perfusion.
Tissue oxygen tension (tPO2) represents the balance between local oxygen supply and demand. This varies markedly between organs, being higher in tissues with low metabolic rates such as bladder [6, 7], and lower in tissues that are more metabolically active such as brain  and liver , or in those that possess significant shunts, such as the renal cortex . Decreases in tPO2 have been recorded across numerous organ beds during low oxygen transport states such as haemorrhage [11–13]. Peripheral, readily accessible tissues have been mainly studied [14–18], and assumed to be surrogates for deeper, more vital organs. However, as relatively few studies [11–13, 19] have simultaneously assessed tPO2 in more than one organ bed, inter-organ differences remain largely unknown.
We hypothesized that non-physiological extremes in arterial oxygen pressure would be detrimental in low-flow states such as haemorrhage. We decided to initially study this aspect in isolation of other resuscitation measures such as fluid replacement. Consequently, we constructed a rodent model of severe, progressive haemorrhage in which we studied the impact of concurrent hyperoxaemia and hypoxaemia on global cardiorespiratory variables and tissue oxygenation in peripheral and deep organ beds.
Male Wistar rats of approximately 300 g body weight (n = 6/group) were used in all experiments. Prior to instrumentation, animals were housed in cages of six on an alternating 12 h light–dark cycle with free access to food and water. All experiments were performed according to Home Office (UK) guidelines under the 1986 Scientific Procedures Act with local (University College London) ethics committee approval.
Spontaneously breathing animals, anaesthetised by 5% isoflurane in room air, were placed onto a heated mat to maintain rectal temperature at 37°C throughout. Under 2% isoflurane, the left common carotid artery and right internal jugular vein were cannulated using 0.96 mm outside diameter PVC tubing catheter (Biocorp Ltd, Huntingdale, Australia). The arterial line was connected to a pressure transducer (Powerlab, AD Instruments, Chalgrove, UK) for continuous monitoring of mean arterial pressure. The venous line was used for subsequent administration of fluids. A tracheostomy was sited using 2.08 mm external diameter polythene tubing (Portex Ltd, Hythe, UK) to secure and suction the airway. This was connected to a T-piece to maintain anaesthesia and to vary the fraction of inspired oxygen. We acknowledge that the fraction of oxygen in the gas mixture may not precisely reflect the true fraction of inspired oxygen (FiO2) in spontaneously breathing animals, however, for purposes of simplicity, we will use this term henceforth. The bladder was cannulated through a midline laparotomy using 1.57 mm external diameter polythene tubing (Portex Ltd, Hythe, UK) inserted through a small incision at the apex. Anaesthesia was then reduced to 1.5% for the remainder of the experiment. To allow access to the abdominal vasculature, the caecum and small intestine were wrapped in cling film and placed outside the abdominal cavity. The left renal artery and descending aorta were isolated from surrounding tissue by careful blunt dissection. Ultrasonic flow probes (Transonic Systems, Ithaca, NY, USA) of 1 and 2 mm diameter were coated in a water-soluble lubricant and placed around the left renal artery and descending aorta, respectively to measure blood flow. Insertion of large area surface (LAS)™ oxygen sensors (0.7 mm in diameter) connected to an Oxylite™ tissue monitoring system (Oxford Optronix, Oxford, UK) allowed continuous monitoring of tPO2 in muscle, bladder, liver and renal cortex. The oxygen sensors are pre-calibrated within the range 0–26.7 kPa and function by sending short pulses of light (475 nm) along a fibre-optic cable to a platinum-complex fluorophore situated at the tip of the probe. This provides a total measurement surface area of 8 mm2 in contact with the tissue. Upon interaction with oxygen, the fluorophore emits light (600 nm) back to the detection unit, the lifetime of which is inversely proportional to the local PO2 within the tissue of interest. As the fluorescence decay is longer at a lower PO2, highly accurate measurements can be made within the physiological range.
For muscle tPO2, a small incision was made at mid-thigh level, and the sensor inserted into the vastus intermedius muscle to a depth of 10 mm using an 18-gauge cannula. The left kidney was punctured using a 22-gauge needle and the oxygen probe inserted to a depth of 2 mm before later being withdrawn by 1 mm to prevent anomalous measurements resulting from local haematoma. An insertion depth of 1 mm enables measurement of tPO2 within the renal cortex, as previously described . The bladder oxygen probe was sited within the bladder lumen via a catheter which continually drained the bladder, thus ensuring good sensor contact with the epithelial surface. Liver tPO2 was measured by placing the probe directly into the airtight space between two of the liver lobes. Mean arterial pressure, blood flow in left renal artery and descending aorta and all tPO2 measurements were continuously monitored and recorded onto a computer using a 16-channel Powerlab system and Chart 5 acquisition software (AD Instruments, Chalgrove, UK).
Following instrumentation, intravascular volume optimization was achieved by repeated 1.5 ml intravenous fluid challenges given over 10 s every 5 min until blood pressure or aortic blood flow failed to increase >10%. Typically, two fluid challenges were required to ensure all animals were normovolaemic at baseline. As determined from a previous study , a continuous infusion of 0.9% saline was then administered at a rate of 20 ml/kg/h for the duration of the experiment to compensate for evaporative fluid losses. Cling film was also placed over the abdomen to minimise fluid losses. All animals were allowed to stabilise for at least 30 min to achieve stable baseline physiological variables before being divided into groups for two separate studies.
Study 1: Tissue oxygen responses to high inspired oxygen concentrations during normovolaemia
Study 2: Effect of varying inspired oxygen concentration during progressive haemorrhage
Progressive haemorrhage was achieved by removing 10% estimated circulating blood volume (based on 70 ml/kg total) every 15 min (Fig. 1). The FiO2 was changed to either 0.15, 0.6 or 1.0 at 15 min (following 20% estimated circulatory volume removal) and continued at this concentration for the remainder of the experiment (Fig. 1). For comparison, a group of animals breathing 0.21 FiO2 throughout were subjected to continued blood removal as outlined above.
Data are presented as mean ± standard error after confirmation of normality distribution. Survival time was analysed using a log-rank test. All other statistics were performed on raw data using either a repeated measures one- (Study 1) or two-way (Study 2) ANOVA followed by Tukey’s post hoc testing (SigmaStat, Systat Software Inc, San Jose, CA, USA). As Study 2 was a terminal experiment, all data points after 60 min were analyzed using a two-tailed unpaired Student’s t test (MS Excel, SP2). An index of systemic vascular resistance (SVRI) was calculated as blood pressure/aortic blood flow.
Study 1: Tissue oxygen responses to high inspired oxygen concentrations during normovolaemia
High inspired oxygen concentrations significantly increased arterial PO2 in normovolaemic animals, the extent to which depended on the fraction of inspired oxygen (Fig. 3). Accordingly, the largest increase was seen with rats breathing 100% O2 (from 9.9 ± 0.3 kPa at baseline to 74.9 ± 1.7 kPa on 100% O2; Fig. 3). Although tPO2 increased to different extents in all organ beds studied, proportionality was preserved with a sevenfold increase in arterial PO2 leading to an approximate threefold increase in tPO2 (Fig. 3). The largest increase was observed in bladder followed by muscle then kidney, all being significantly elevated while breathing either 30, 60 or 100% O2. The smallest increment was seen in the liver, and was only significantly raised compared to baseline while breathing 100% O2. While breathing 100% O2 there was a significant increase in mean arterial pressure and a corresponding decrease in descending aortic blood flow (39.3 ± 2.7 ml/min while breathing 21% O2 to 25.1 ± 2.6 ml/min on 100% O2; P < 0.01; Fig. 3). As a result, SVRI increased from 2.6 ± 0.2 at baseline to 5.2 ± 0.5 mmHg min/ml while breathing 100% O2. Although haemoglobin oxygen saturation was increased (P < 0.05), from 85.5 ± 2% (on 21% O2) to 100 ± 0% (on 100% O2), global oxygen delivery was reduced (18.0 ± 1.3 ml/kg/min while breathing 21% O2 vs 14.4 ± 1.4 ml/kg/min on 100% O2), albeit not significantly (P = 0.09; data not shown). Renal blood flow, base excess and lactate levels remained unchanged throughout.
Study 2: Effect of varying inspired oxygen concentration during progressive haemorrhage
Mean arterial pressure decreased during progressive haemorrhage in animals breathing room air, with pre-terminal values averaging 37.3 ± 3.9 mmHg (P < 0.001 vs baseline; Fig. 4). Similar changes were seen in aortic and renal blood flow (P < 0.01). Arterial base excess (ABE) and lactate were maintained within normal limits until approximately 40% of blood volume had been removed, then rapidly deteriorated (P < 0.01, Fig. 4). As we have previously reported , arterial PO2 increased (Fig. 4) and PCO2 decreased (data not shown) as a result of hyperventilation induced by the metabolic acidosis. Tissue oxygen tensions in muscle, bladder and liver were all reduced by a similar proportion following progressive haemorrhage and fell in line with oxygen delivery. Renal cortical tPO2 however increased initially, and only fell below baseline values at the pre-terminal phase (P < 0.05; Fig. 4).
Induction of hyperoxaemia following 20% blood loss significantly increased arterial PO2 (at both 60 and 100% O2; P < 0.01; Fig. 4) which was maintained for the duration of the experiment. Increases in tPO2 in muscle, bladder and renal cortex were initially similar to those seen during normovolaemia, however liver tPO2 did not rise as much. The initial elevation in tPO2 in the organ beds studied decreased with continuing blood loss, with a disproportionately greater fall in liver and renal cortex (Fig. 4). Hyperoxaemia produced an increase in lactate levels (7.1 ± 0.9 mmol/l for 21% O2 vs 13.8 ± 0.9 mmol/l for 100% O2 in the pre-terminal phase; P < 0.01; Fig. 4). Additionally, in those animals breathing 100% O2, the arterial base excess, mean arterial pressure, aortic and renal blood flow all deteriorated significantly below values obtained while breathing 21% O2. The rise in SVR during haemorrhage was greater in the presence of a higher FiO2 (4.7 ± 0.5, 5.3 ± 1.2 and 10.2 ± 3.5 mmHg min/ml for 21, 60 and 100% O2, respectively after 90 min).
An FiO2 of 0.15 caused earlier decreases in tissue oxygen tension, macrovascular blood flows and a greater derangement of arterial base excess and lactate than those breathing room air (Fig. 4), leading to a significantly earlier demise (Fig. 2).
Figure 5 shows the relationship between arterial and tissue oxygen tensions during normovolaemia, and in animals subjected to progressive haemorrhage following 20 and 50% blood loss (mild and severe hypovolaemia, respectively). Hypovolaemia caused a severity-dependent right shift of the curve, thus demonstrating a blunted increase in tissue oxygen tension in comparison with the increase in arterial oxygen tension. This was observed in both peripheral (muscle and bladder) and deep (liver and renal cortex) organ beds.
We explored the hypothesis that high and low inspired oxygen concentrations are both detrimental during progressive haemorrhage. Survival times were similar between all groups studied with the exception of hypoxaemia, however significant derangement of blood pressure, macrovascular blood flows and lactataemia was observed in both hypoxaemic and hyperoxaemic groups over and above that seen in normoxaemic animals. We also found that hyperoxaemia elevated peripheral and deep organ tissue oxygen tensions during normovolaemia and mild hypovolaemia, but not during severe hypovolaemia.
Tissue oxygen tensions
Tissue oxygen tension represents the balance between local oxygen supply and demand. This monitoring modality has been previously used in preclinical studies of blood loss [11–13], and to assess the adequacy of resuscitation from a variety of sites including conjunctiva , subcutaneous tissue  and skeletal muscle . tPO2 values start falling once the tissue’s needs become compromised by reduced oxygen delivery. A well-characterized compensatory response to haemorrhage includes an increase in peripheral vascular resistance and redistribution of blood flow. Both these mechanisms are aimed at maintaining organ perfusion pressure and preserving blood flow to vital organs such as brain and heart . The corollary is that peripheral vasoconstriction would cause early decreases in tissue oxygen in organ beds such as the skin, conjunctiva and skeletal muscle . A number of studies have confirmed decreased tissue oxygen tensions in peripheral sites following haemorrhage [12, 15, 17, 18, 22]. Others have also reported reduced tissue oxygen tensions in deeper organs such as liver [9, 13] and kidney , however inter-organ differences have been rarely compared under the same set of experimental conditions.
tPO2 during haemorrhage
As we previously reported , progressive haemorrhage caused decreases in muscle, bladder epithelial and liver tPO2 that corresponded to the reduction in global oxygen delivery. Similar reductions in liver and peripheral (subcutaneous, transcutaneous, conjunctival) tPO2 were seen in a piglet study of controlled haemorrhage . By contrast, and in line with our previous findings , renal cortical tPO2 was maintained, even following a 75% reduction in macrovascular (renal arterial) blood flow. As renal blood flow predominantly favours the cortex, this likely reflects adaptation of this organ to haemorrhage by a decrease in cellular metabolic demand. Unlike other organ beds, renal oxygen consumption has been shown to vary directly with oxygen delivery, both in the normal physiological range and during haemorrhagic shock . As most of the kidney’s energy expenditure relates to reabsorption of water and electrolytes, a marked fall in glomerular filtration rate would reduce oxygen demand and thus maintain tPO2 homeostasis in low flow states .
Hyperoxic vasoconstriction and oxygen delivery
High inspired oxygen concentrations are routinely used as an initial treatment for acutely unwell patients, to ensure high oxyhaemoglobin levels while efforts are made to stem blood loss, restore cardiac output and correct low haemoglobin levels. It often takes several hours to reduce inspired oxygen concentrations to normoxaemic levels, partly through a non-appreciation of any potential harm from a high PaO2. However, as we found in this study, hyperoxaemia alone can cause haemodynamic effects including vasoconstriction and increased vascular resistance [25, 26]. Mechanisms proposed include a decrease in vasodilatory tone due to diminished release of basal nitric oxide (NO) from the endothelium , oxidative quenching of NO , raised levels of endothelins  or leukotrienes , and decreased synthesis of vasodilating prostaglandins .
The impact of hyperoxaemic vasoconstriction on global oxygen delivery and regional perfusion remains controversial. Some argue that the reduction in blood flow secondary to increased vascular tone is insufficient to offset a higher O2 blood content, thus enhancing global oxygen delivery . Others found that the magnitude of decrease in cerebral blood flow (33%) is more profound than the increase in arterial oxygenation (13%), leading to an overall reduction in cerebral oxygen delivery . During hyperoxaemic normovolaemia, our study is in agreement with the latter finding, with a significant increase in mean arterial pressure and a reduction in cardiac output leading to an overall increase in SVRI and a decrease in global oxygen delivery. As expected, SVRI increased during progressive haemorrhage, however this effect was magnified by hyperoxaemia. The earlier deterioration in blood pressure and macrovascular flows seen during hyperoxaemic haemorrhage may be related to ventricular dysfunction secondary to coronary vasoconstriction-induced myocardial hypoperfusion, as has been described in patients with heart disease . Additionally, peripheral vasoconstriction in some vascular beds may place an excessive load on the poorly perfused heart, thus expediting heart failure and subsequent earlier cardiovascular collapse. However, this explanation remains speculative and requires confirmation.
Effects of hyperoxaemia on tissue oxygenation
Despite a modest decrease in convective oxygen delivery with hyperoxaemia, the driving force for oxygen diffusion into tissues is determined by the local partial pressure. We found that hyperoxaemia increased tissue oxygen tensions in all monitored organ beds, albeit to differing extents. The magnitude of this increase during mild hypovolaemia was similar to that observed during normovolaemia. However, this increment diminished rapidly with continued blood loss, falling to the same (or even lower) levels than those seen in animals managed in room air. Measuring the downstream increase in tPO2 following an increase in PaO2 has been coined the ‘oxygen challenge test’. This was recently demonstrated to be a useful marker for the early detection of shock and an early prognosticator in septic patients [35, 36]. While the increment in peripheral tPO2 under normal physiological conditions correlates with the fraction of inspired oxygen [16, 37], this is compromised during low flow states. This relates to oxygen supply dependency where the reduction in oxygen delivery exceeds the increased oxygen gradient generated by a high PaO2. Hyperoxaemia (and its impact on diffusive oxygen delivery) may thus be beneficial only in the presence of an adequate blood flow (convective oxygen delivery), suggesting a utility for supplemental oxygen only after correction of hypovolaemia.
We observed a discrepancy between the increases in arterial and tissue PO2 in the different organ beds studied. During haemorrhage this could be attributed to regional alterations in blood flow, but also occurred while the animal was normovolaemic and may thus represent a protective role, preventing excessive exposure of tissues to potentially toxic partial pressures of oxygen. In the kidney this could be achieved by increasing arterio-venous shunting , while in other organ beds protection may be implemented by increasing vascular tone.
Hyperoxaemia and hyperlactataemia
The blood lactate level represents the balance between cellular production (from increased glycolysis and/or decreased mitochondrial uptake of pyruvate) and metabolism, notably by the liver and, to a lesser extent, the heart, gut and brain. We observed that hyperlactataemia was most profound in animals breathing 100% O2, and that liver tPO2 was the most affected of the organs studied, at least in terms of tissue oxygen tensions. The liver has been implicated as the organ most vulnerable to insults such as haemorrhage as its normal oxygen extraction ratio operates close to maximum capacity . The significant reduction in hepatic oxygen consumption related to decreased delivery could compromise its capacity to metabolise lactate . During hyperoxaemia, hepatic metabolic demand may be further increased as the liver is responsible for synthesis of glutathione and other anti-oxidant defences activated during oxidative stress. Consequently, this may further restrict the liver’s capacity to metabolise lactate.
Limitations of the study
We focused on a scenario of ongoing haemorrhage in which inspired oxygen concentrations were varied. We fully acknowledge that many patients with active bleeding do respond promptly to resuscitative measures, however we wished to mimic a particularly severe and parlous situation in which concurrent hyperoxaemia appears to be detrimental. We are performing ongoing studies examining the impact of variable FiO2 on controlled haemorrhage followed by resuscitation. Our blood gas analyser delivers co-oximetry-measured values of oxyhaemoglobin saturation that, despite calibration, are lower than would be expected for the measured arterial oxygen pressure. The manufacturer is unable to resolve this discrepancy that may be related to spectroscopic differences between human and rat blood. Although absolute values of oxygen delivery may underestimate by ~7%, trend following is likely to be reasonably accurate.
Clinical utility of varying inspired oxygen concentrations
From a clinical perspective, the utility of high concentration, high flow oxygen therapy still remains controversial. Both benefit and harm have been reported across a wide range of clinical and experimental pathologies. During haemorrhagic shock, the evidence is equally diverse; while some studies showed a reduction in lethality [1, 40], others did not [41, 42]. Likewise, the benefit conferred by hypoxaemic resuscitation is evident in some, but not all studies [4, 43]. We found no difference between time to death, nor in the quantity of blood that could be removed prior to death between groups treated with or without oxygen supplementation. Nevertheless, the deleterious effects on blood pressure and cardiac output with increased hyperlactataemia, especially in those given higher concentrations of oxygen, suggest caution should be employed. The accelerated deterioration and time to death seen in haemorrhaged animals with concurrent hypoxaemia indicates that an optimal therapeutic approach in the shocked patient could be restoration of normoxaemia with avoidance of extremes. As recently editorialised , supplemental oxygen administration only provides benefit to those who would otherwise become hypoxaemic breathing room air.
This work was funded by the UK Medical Research Council and Integrative Pharmacological Fund. This work was undertaken at UCLH/UCL who receive a proportion of funding from the UK Department of Health’s NIHR Biomedical Research Centre’s funding scheme. We thank Oxford Optronix for kindly providing the tPO2 probes.