The acidosis-induced right shift of the HbO2 dissociation curve is maintained during erythrocyte storage

Background and objectives. In fresh blood, tissue hypoxia increases microcirculatory acidosis, which enhances erythrocyte O2 unloading and increases the amount of available O2. Storage of eryfhrocytes increases the HbO2 affinity and reduces O2 unloading. We examined the development of the affinity change during a period of 5 weeks of storage by present blood bank standards, and investigated to what extent acidosis offsets the affinity change. Materials and methods. Blood from volunteer donors was processed and stored as erythrocyte concentrates (EC). At 2–5 day intervals, EC were drawn from the bags and suspended in plasma and crystalloids to an Hb ≈ 10 g/dL. The suspensions were adjusted to give a pH of 7.40, 7.10, 6.80 or 6.30 and equilibrated with different gas mixtures to SO2 0, 25, 50, 75 and 100%. Measurements of the PO2/SO2 pairs at each pH were used to calculate the position of the HbO2 curve and its P50 value. Results. A significant leftward shift in the HbO2 curve was established after 1 week of storage; after 2.5 weeks only minor further changes were observed. Acidification right-shifted the HbO2 curve, after 2.5 weeks of storage the curve at pH 7.10 was similar to that for fresh blood at pH 7.40. Calculations of extractable O2 showed that the left-shifted HbO2 curve of stored EC could be advantageous at a low arterial PO2. Conclusions. The rightward shift of the HbO2 curve due to acidosis is well maintained in stored eryfhrocytes, a moderate pH decrease offsets the storage-induced increased HbO2 affinity.


Introduction
The O 2 content of arterial blood (CaO 2 ) is mainly determined by the amount of haemoglobin (Hb) and its O 2 saturation (SaO 2 ); the number of O 2 molecules dissolved as gas and measured as the O 2 partial pressure (PaO 2 ) represents under normal circumstances only 1 -2% of the total. The importance of PaO 2 for O 2 delivery to the tissues lies therefore not in the PaO 2 per se , but in its effect on the SaO 2 . The relationship between PO 2 and SO 2 is described by the sigmoid-shaped HbO 2 dissociation curve; at a constant PO 2 a leftward shift in the curve position increases the SO 2 and thus the CaO 2 while a rightward shift decreases the SO 2 . Due to the shape of the HbO 2 curve, the effects of such shifts are usually miniscule at normal or high PO 2 levels but may be crucial at low PO 2 levels.
During blood bank storage of erythrocytes, depletion of their 2,3 diphosphoglycerate (2,3-DPG) content [1] shifts the HbO 2 curve to the left. This increases the CaO 2 , but will also decrease the unloading of O 2 in the tissue microcirculation. Some authors have considered this a major argument against a benefi cial effect of transfusion of stored erythrocytes [2,3], and others have suggested that only fresh erythrocytes should be transfused to intensive care patients [4]. However, others have pointed out that experimental and clinical studies have yielded controversial results on the eventually negative effect of transfusing stored erythrocytes, and that many more factors than the actual position of the HbO 2 dissociation curve infl uence the clinical effect of a red cell transfusion [5].
In fresh blood, tissue acidosis will increase O 2 unloading; hypoxic acidosis will thus automatically Scandinavian Journal of Clinical & Laboratory Investigation, 2011;71: 314-321 ISSN 0036-5513 print/ISSN 1502-7686 online © 2011 Informa Healthcare DOI: 10.3109/00365513.2011.565366 increase tissue O 2 delivery. To what extent acidosis can compensate for the storage-induced leftward shift in erythrocyte concentrates (EC) preserved in SAGMAN solution has not previously been determined. In the massively transfused patient, the circulating blood consists of a mixture of transfused EC and plasma from the blood bank plus infused electrolyte solutions. The O 2 transport capability and tissue unloading of such blood during the acute phase is therefore equivalent to that of stored EC.
Calculations of curve shifts based on standard pH-dependent correction factors do not correct for changes in 2,3-DPG; even in fresh blood such calculations become increasingly inaccurate with mounting acidosis [6]. We therefore measured the storage-induced effect on the HbO 2 curve shifts in EC at 2 -5 day intervals of storage for up to 35 days; on each experimental day, the samples were examined at four different pH values. To simulate in vivo conditions during major bleeding, the stored erythrocytes were resuspended in equal amounts of plasma and crystalloids to an Hb of 9 -11 g/dL. The consequences of the storage dependent HbO 2 curve shifts for the amount of consumable O 2 [7] during threatening tissue hypoxia were also calculated.

Sources of blood
Eighteen healthy persons from the donor pool of Oslo University Hospital Blood Bank (OBB) received oral and written information about the investigation and gave written consent to allow their donation on a particular day to be used for research purposes. During preliminary methodological studies, blood from one of the authors (HEH) and samples from outdated blood to be discarded were also used. The investigation was approved by the Regional Committee for Medical research ethics, the Biobank committee and the Committee for Individual rights in research.
The blood to be used in the investigation was collected, processed and stored at 4 ° C by procedures identical to those routinely employed for preparation of erythrocyte concentrates (EC) at the blood bank. Briefl y, 450 mL whole blood was mixed with 63 mL of citrate-phosphate-dextrose-adenine (CPD-A) in a plastic bag (Fenwal Corp., Illinois, USA) and left at room temperature for 1 h. It was then centrifuged at 4 ° C and 4000 g for 20 min, followed by removal of plasma and buffy coat into separate bags by pressure. The remaining erythrocyte concentrate was passed through a leukocyte fi lter (Asahi Kasei Medical Co, Japan) at room temperature, resuspended in SAGMAN solution (Fenwal) and immediately stored at 4 ° C. The procedure until storage typically took 4 -5 h and never exceeded 8 h. EC produced at the OBB show a mean volume ϭ 245 mL, a mean haematocrit ϭ 0.55 and a mean haemoglobin content ϭ 49 g/bag (unpublished quality control data).
The HbO 2 affi nity of EC stored at 4 ° C was examined after different time intervals. On day 0, samples of EC processed for storage the same day were examined before cooling and storage. During the next 3 weeks of storage, samples of EC from the same six donors were examined at time intervals as shown in Figure 1. The necessary volume of EC from each donor bag was transferred aseptically into a smaller bag ( ' Baby bag ' , Fenwal) on each experimental day, after which both bags were shut by welding. Due to the limited volume of the EC bags, the contents of each bag could be followed only for up to 3 weeks. Therefore, blood from additional donors was necessary to obtain data corresponding to a storage time of 3-5 weeks.
In 6 donors, an additional 15 mL of blood was drawn separately into heparinized tubes for estimation of both the normal HbO 2 curve and that representing grave acidosis at approximately pH 6.30 in non-processed blood. The position of the HbO 2 curve at pH 6.30 calculated from these data was in agreement with that previously determined in a detailed investigation in our hospital [6] and served to verify the method for HbO 2 curve determination used in the present investigation. The choice of anticoagulant or the processing of blood to EC had no signifi cant effect on the HbO 2 curve within the normal range of pH values. All laboratory work, including tonometry and blood gas analysis, was carried out in the facilities of the blood bank. All data from 0 -17 days of storage represent blood from the same EC bags. The number of EC bags examined after each storage period is shown in italics above the x-axis. Data for all storage periods longer than 2 days were signifi cantly different from day 0. The p values shown as numbers in the fi gure refer to a difference from the preceding P 50 level by a paired T-test including all pH levels. The asterisks denote a signifi cant ( * p Ͻ 0.05) difference from the preceding value when each individual pH level was analysed separately.

Preparation of blood samples
Five mL of EC was mixed with 3 mL of thawed Octaplas ® (pooled, standardized, solvent/detergent virus inactivated whole plasma) of blood group AB, the pH, base excess (BE) and other blood gas parameters of this mixture were then measured in an ABL 700 blood gas apparatus (Radiometer Medical, Br ø nsh ø j, Denmark) equipped with a co-oximeter. Subsequently, 3 mL of an electrolyte mixture was added, giving an Hb of 9 -11 g/dL in the fi nal solution. The mixture consisted of sterile H 2 O (Fresenius Kabi, Halden, Norway) and lactic acid (Sigma Aldrich, St Louis, MO, USA) or NaHCO 3 0.5 mol/L (Braun Melsungen, Melsungen, Germany). The amount of lactic acid or NaHCO 3 was calculated from the initial pH and BE values to give pH levels of approximately 7.40, 7.10, 6.80 or 6.30 in the fi nal EC-PE mixture. In addition, concentrated NaCl (4 mol/L, Addex NaCl, Fresenius Kabi, Halden, Norway) was added in amounts calculated to give a fi nal volume of 3 mL and a Na ϩ of 140 mmol/L in the electrolyte mixture. The accuracy of the calculations was verifi ed by the electrolyte analysis in the fi nal EC-plasma-electrolyte (EC-PE) mixture given by the blood gas apparatus. In experiments involving heparinized blood no plasma was added; the effect of acidifi cation to pH 6.30 was examined after adjusting the pH by addition of 3 mL of electrolyte mixture by a procedure identical to that described above.
Both processing and storage acidify the EC. Preliminary experiments showed that additions of acid or NaHCO 3 aimed to give pH values of 7.40 and 7.10 could be calculated from the Siggaard-Andersen acid-base nomogram [8] with reasonable accuracy. The nomogram proved inaccurate, however, when the goal was pH 6.80, and a pH of 6.30 was out of its range. Separate preliminary experiments were therefore necessary to establish empirical addition factors. Obtaining the exact designated pH value in the lower pH range proved diffi cult and time consuming; the median pH value deviation from the designated value for each group was, however, always less than 0.1 pH units. For simplicity, the designated pH values are shown in fi gures and text.

Tonometry and measurements
The EC-PE mixture was divided into two 5 mL plastic syringes (3.5 mL each) and placed in a tonometer (RNA medical equilibrator model 300, RNA Medical, Massachusetts, USA) preheated to 37 ° C. The syringes, which were pretreated with a foaminhibiting coating by the tonometer manufacturer, were equilibrated with a gas containing 5% CO 2 and 20% O 2 in nitrogen or 5% CO 2 in nitrogen, respectively, for 20 min. The EC-PE mixture from the two syringes had an SO 2 of 90 -100% (depending on the pH) or close to 0%, respectively, with a pCO 2 of approximately 5 kPa. The contents of the two syringes were then transferred anaerobically into 2 mL syringes containing a glass bead for mixing purposes, in amounts calculated to give the fi nal solutions an SO 2 of approximately 25%, 50% and 75%, respectively. The fi nal mixtures, as well as the remaining ≈ 0% and ≈ 100% blood, were then analysed in the blood gas apparatus. The fi ve sets of corresponding SO 2 -PO 2 values at each pH were then used for determination of the position of the HbO 2 curve. The PO 2 corresponding to half saturation of the Hb with O 2 , the P 50 value, were calculated from our data (see below) and used to express changes in the HbO 2 curve position.

Estimation of storage and pH effects on the HbO 2 curve and P 50 values
Often used equations (Siggaard-Andersen, Severinghaus) for calculating changes in the HbO 2 curve induced by variations in pH overestimate the rightward shift of the HbO 2 curve in grave acidosis [6]. A dedicated data program using LabView software was developed locally by one of the authors, a research data engineer (TAS). The program fi tted a sigmoid shaped curve to the position of the fi ve PO 2 -SO 2 measurement pairs at each pH; this curve was then compared to a population of HbO 2 curves calculated by the equations proposed by Severinghaus [9] and by Kellman [10] for determination of the effect of pH changes on the HbO 2 curve position. The curve representing the best fi t to the measured data was chosen as the ' true ' HbO 2 curve. Curves calculated according to the Kellman method gave the best visual fi t to the observed PO 2 -SO 2 data, the pH of which the curves were calculated also deviated least from the actual pH of the samples. This routine was therefore utilized for determination of the position of the curve as defi ned by its P 50 value. The median P 50 value for heparinized blood at a pH of 6.30 calculated by this method (8.9 kPa) was very close to that previously determined in our institution (9.1 kPa) by multiple direct measurements in heparinized blood at the same pH [6], thus verifying our method for determining the position of the HbO 2 curve even in grave acidosis.

Calculation of consumable oxygen
At an end venous PO 2 below 2.7 kPa ( ϭ 20 mmHg, see Conversions section below), cells may become dysfunctional and lactic acid production can be induced [6,11]. As more than 98% of the O 2 in normal arterial blood is bound to Hb, the amount of consumable O 2 calculated for a particular HbO 2 dissociation curve can, for simplicity, be expressed as the difference between the SO 2 corresponding to that of a given arterial PO 2 level and that at a PO 2 of 2.7 kPa. For the purpose of such calculations, are depicted in Figure 2a. There was no detectable change in P 50 values during the fi rst 2 days of storage; from 7 days and onwards, the P 50 decreased (i.e. HbO 2 affi nity increased) for all pH levels ( p Ͻ 0.001 for all pH levels pooled, p Ͻ 0.05 for analysis of each pH level separately). Analysis of changes in P 50 values from the previous observation during storage showed a signifi cant (see Figure 1) further decrease during the second week of storage. After 16 -17 days, however, the P 50 at all pH values was stabilized (see Figure 2b for actual HbO 2 curves on days 16 -17 of storage). The minor changes for the pH 6.80 level at the end of the storage period could not be detected for the other pH levels, and probably represent only random effects. PaO 2 levels corresponding to grave hypoxemia (5.3 kPa, 40 mm Hg), normoxia (13.3 kPa, 100 mm Hg) and supernormal O 2 levels (20 kPa, 150 mmHg) were used. These calculations assume a similar pH in arterial and capillary blood. Under extreme conditions, however, major differences between arterial and microcirculatory pH levels may exist (see discussion). Therefore, calculations of consumable O 2 , assuming a HbO 2 curve position corresponding to pH 7.40 in arterial blood and to pH 6.80 locally in the microcirculation of hypoxic tissue, were also carried out.

Conversions
To facilitate comparisons of P 50 values between our results and those of other authors, data originally given in mmHg or Torr is converted to kPa by the factor 1 kPa ϭ 7.5 mmHg or Torr.
As analysis of raw data showed them to be normally distributed, parametric methods were utilized. A one-way ANOVA analysis was initially carried out for all storage times and each pH level. As all data from the fi rst 16 -17 days were obtained from samples originating from the same EC bags, a paired T-test was used for comparison of P 50 values for the different storage periods during this period. Independent samples T-test was utilized for comparison of the longer storage periods, where the variations in donors on different storage days were greater. Regression analysis was used to analyse the relation between the pH and P 50 values at each week of storage, differences in regression lines for the pH-P 50 plots were based on differences between the 95% confi dence intervals for each line. All analyses were carried out by means of the SPSS statistical software package, version 15.

Changes in HbO 2 dissociation curve during storage
The median P 50 in the EC-plasma-electrolyte mixture on the day of blood donation was 3.47 kPa at pH 7.40, 4.75 kPa at 7.10, 5.96 kPa at pH 6.80 and 8.45 kPa at pH 6.30 (Figure 1). The actual HbO 2 dissociation curves corresponding to the P 50 values The magnitude of the P 50 change, defi ned as the difference between the P 50 values at day 0 and at day 16 -17, was more marked in acidotic blood than at normal pH (Figure 1), as could be expected from graphic presentations of pH-dependent curve shifts. The HbO 2 leftward shift measured at pH ≈ 7.40 corresponded to a P 50 change of approximately 0.68 kPa, with a curve position equal to that of fresh blood at a pH of ≈ 7.72 (Figure 2b). The leftward shift of P 50 at pH ≈7.10 was 1.50 kPa, corresponding to normal blood at pH ≈ 7.40. At pH ≈ 6.80 the P 50 shift was 1.45 kPa, corresponding to normal blood at pH » 7.20. There was no statistical difference between the shifts at pH 7.10 and 6.80, but both were larger than at pH 7.40 ( p Ͻ 0.001). At pH ≈ 6.30 the P 50 shift was approximately 2.3 kPa, which was greater than that for the other pH values examined ( p Ͻ 0.001 compared to pH 7.40, p Ͻ 0.05 compared to pH 7.10 and 6.80). The shift at this pH gave a P 50 value corresponding to that of fresh blood at pH ≈ 6.85.

Consequences of erythrocyte storage for consumable oxygen
The amount of consumable O 2 , expressed as the difference in SO 2 between that corresponding to the PO 2 of arterial blood and a microcirculatory endvenous PO 2 of 2.7 kPa ( Δ SO 2 ), was calculated from the HbO 2 dissociation curves on storage day 0 (Figure 2a and insert) and after the affi nity change had stabilized at 16 -17 days (Figure 2b and insert). On day 0, the amounts of consumable O 2 at high (20.0 kPa, 150 mm Hg) and normal (13.3 kPa, 100 mm Hg) PO 2 levels increased with mounting acidosis from pH 7.40 to pH 6.80, but started to decline when pH fell to 6.30. At low PO 2 (5.3 kPa, 40 mm Hg), corresponding to grave clinical hypoxemia, the consumable O 2 decreased continuously with increasing acidosis. After storage for 16 -17 days, the amount of consumable O 2 at high and normal PO 2 followed the same pattern as on day 0, but was reduced by almost 20% at pH 7.40 and about 10% at pH 7.10. At extreme acidosis of pH 6.30, however, the amount of consumable O 2 was higher after storage, by about 2% at high and by 14% at normal PaO 2 . At the low PO 2 level, the amount of consumable O 2 at pH 7.40 was slightly smaller after storage than on day 0; the increased HbO 2 affi nity of storage proved, however, progressively benefi cial compared to day 0 with mounting acidosis ( Δ SO 2 at pH 7.10: 42.0% vs 37.5%, at pH 6.80: 39.9% vs 27.3% and at pH 6.30: 29.1% vs 12.1%, respectively).
Similar calculations, but assuming that the arterial blood was maintained at pH 7.40 while that in poorly oxygenated tissue was acidifi ed to pH 6.80 (see methods), are shown in Figure 3. Again, storage reduces consumable O 2 at high and normal PaO 2 values, but to a lower degree than that calculated for a constant pH. Storage will have a slight benefi cial effect at very low PO 2 ( Δ SO 2 63.6 % vs . 62.1 %).

P 50 as a function of pH
The relationship between pH and P 50 was almost linear within the acidotic pH range compatible with survival, both initially and after storage (Figure 4). Analysis of regression with pH as the independent variable showed a strong correlation before storage (r 2 ϭ 0.981, p Ͻ 0.001), as well as after 1 week, (r 2 ϭ 0.928, p Ͻ 0.001), 2 weeks (r 2 ϭ 0.974, p Ͻ 0.001), 3 weeks (r 2 ϭ 0.981, p Ͻ 0.001), 4 weeks (r 2 ϭ 0.978, p Ͻ 0.001) and 5 weeks (r 2 ϭ 0.961, p Ͻ 0.001). The rise rate was steeper on day 0 before storage, with essentially similar slopes at 1 week of storage and for longer storage periods. The 95% confi dence interval for the pH-P 50 regression line on day 0 was different from those for 2 weeks and onward ( Ϫ 4.146, -3.681 vs. Ϫ 3.384, Ϫ 2.887 at 2 weeks, -3.228, Ϫ 2.788 at 3 weeks, Ϫ 3.458, Ϫ 3.060 at 4 weeks and -3.006, -2.546 at 5 weeks of storage). The difference between the lines at day 0 and 1 week of storage ( -3.774, -2.946) was borderline.

Discussion
The P 50 value in fresh blood at pH 7.40 is usually given as 3.47 -3.87 kPa (26)(27)(28)(29) [12,13]. In our study, the P 50 in the EC suspension after the processing procedure (anticoagulation and separation) for EC employed in a modern blood bank corresponded to the lower value, and did not change after cooling and storage for 2 days. A leftward shift of the HbO 2 curve became evident after 7 days of storage and became stabilized after 2.5 weeks. A storageinduced leftward shift of the HbO 2 curve in whole blood was fi rst quantifi ed by Valtis and Kennedy in Some authors have considered the reduced O 2 unloading ability of blood with a leftward curve shift as a major argument against a benefi cial effect of bank blood transfusions [2,3]. In our study, adding H ϩ to the erythrocyte solution could abolish the HbO 2 affi nity change induced by blood bank storage. In clinical settings involving massive transfusions (e.g. major trauma), arterial and/or tissue acidosis can be expected; the transfused blood would then have an O 2 unloading capability similar to that of a person ' s own blood at normal pH.
Whether an HbO 2 curve shift has negative or positive consequences for tissue O 2 delivery depends on the mechanisms underlying threatening or manifest hypoxia in each particular patient. A rightward shift of the HbO 2 curve increases O 2 unloading in the microcirculation; at extreme, but clinically relevant acidosis [18], the microcirculatory SO 2 may fall to 7 -8% before the PO 2 decreases below 2.7 kPa (20 mmHg) [6]. This makes almost all O 2 in the blood available to the tissue cells. On the other hand, it decreases the SO 2 corresponding to a given PO 2 in the blood leaving the lungs. This effect is of minor signifi cance if the acidosis is moderate and the PaO 2 levels are in the normal or supranormal range. At an extreme pH of 6.30, however, a normal PaO 2 of 13.3 kPa (100 mmHg) results in an SaO 2 of only 76 -77%, and a PaO 2 of approximately 30 kPa (225 mmHg) is necessary to obtain a normal SaO 2 [6]. The effect of acidosis-induced rightward shift on SaO 2 is accentuated if the patient also has respiratory insuffi ciency. A PaO 2 of 7.0 kPa (52.5 mmHg), which at normal pH would give a SaO 2 of 87 %, resulted in a SaO 2 of 45% in a patient with a pH of 6.66 [19].
Under normal conditions, the CO 2 released by tissue cells induces a slight acidifi cation and thus a minor rightward shift of the HbO 2 curve as arterial 1954 [14] and later confi rmed by others [15,16]. In blood processed and stored by the methods common at that time (ACD blood), the P 50 after 15 -20 days of storage, when calculated back to conditions at pH 7.40, was found to be lower than that of EC in our investigation. The magnitude of these differences are moderate, from 0.3 kPa (2 mmHg) [14] to 0.7 kPa (5 mmHg) [16]. In contrast to our data, two of the previous studies [14,15] found that most of the change in the HbO 2 curve occurred within the fi rst 7 days of storage, while another [16], analogous to our data, found that the bulk of the P 50 change occurred within the fi rst 15 days. The differences in the magnitude of change and the time pattern may, at least in part, be attributed to the different routines for erythrocyte preservation in use during the time when the investigations were carried out.
The main cause of the affi nity change has been identifi ed as a reduction in erythrocyte 2, 3-diphosphoglycerate (2,3-DPG) [1] during blood bank storage. A previous investigation in our blood bank, utilizing blood processed and stored by the same routine as in our experiments, showed that the erythrocyte 2,3-DPG levels were well maintained during the fi rst 24 h, but declined to 60 -80% during the fi rst week of storage and further reduced to 30 -40% of the initial value after 2 weeks of storage.
A further reduction to about 20% was seen after 25 days [17]. As the leftward shift of the HbO 2 curve seems to have stabilised after 2 -2.5 weeks in our study, the last 20% reduction of 2,3-DPG seems to have little impact on the P 50 of the stored blood. Incubation of erythrocytes with phosphate mixtures may increase the erythrocyte content of 2,3 DPG. In a clinical setting involving urgent need for massive transfusions, such pre-treatment of stored erythrocytes would be impractical. tissue PO 2 levels vary considerably [33 -35]. Our calculations indicate, however, that the left-shifted HbO 2 curve of stored EC could be advantageous at a low arterial PO 2 . Since the pH-dependent change in Hb affi nity for O 2 is largely conserved in stored blood, the leftward shift of the HbO 2 curve in stored EC probably has no deleterious effect in most patients.

Conclusion
As the leftward HbO 2 curve shift occurring during storage does not abolish the rightward shift induced by acidosis, the increased HbO 2 affi nity of EC stored for 2 weeks or more does not represent a major tissue oxygenation problem in most transfused patients. blood enters the microcirculation (the ' Bohr shift ' ). The shift is reversed when CO 2 is excreted by the lungs. During episodes of threatening or manifest tissue hypoxia, the optimal condition for transport and utilization of O 2 would be a left-shifted HbO 2 curve position in the blood during passage through the lungs (ensuring optimal CaO 2 ), and a major shift to the right in the microcirculation (ensuring optimal tissue PO 2 ). During circulatory failure or maximal exercise, a local tissue H ϩ concentration up to ten times higher than that in normal arterial blood may develop [20,21]. If the amount of the locally generated excess acid metabolites is within the elimination capability of the organism, this acidosis will not necessarily be refl ected in arterial blood [21,22]. As illustrated by Figure 3, a situation in which arterial pH is 7.40 and that in the most vulnerable tissues is 6.80, the amount of consumable O 2 at the very low PaO 2 of 5.3 kPa may be maintained at a level close to that calculated for a normal PaO 2 at a pH of 7.40.
Under conditions where the PaO 2 is very low, a leftward shift of the HbO 2 curve (see Figures 2a and b) may be necessary for survival. During vaginal delivery the baby ' s PaO 2 may decrease to values in the 2.9 -3.33 kPa (22)(23)(24)(25) range [23], a leftward HbO 2 curve shift due to fetal Hb makes a SaO 2 in the 60 -70% range possible. A marked leftward shift, induced by arterial blood alkalosis, is also a prerequisite for maintaining an acceptable SaO 2 during high altitude ascents without supplementary O 2 [24,25]. Theoretically, a left shifted HbO 2 curve could therefore also be advantageous in catastrophic lung failure with very low PaO 2 values.
The increased affi nity for O 2 in erythrocytes stored for up to 35 days in our study was reversed by acidifi cation of the suspension medium. The range of pH values chosen for examination in our study is clinically relevant. Most clinicians will treat a metabolic acidosis before it reaches pH 7.10 [26 -28], especially in unstable patients, a pH of 6.80 is often cited as the lower limit for expected survival [29,30]. Nevertheless, many clinicians have successfully treated patients with even lower values; ultimate survival and restitution with a pH value of 6.33 in arterial blood has been reported [18]. Equal, or even lower pH values may exist locally in hypoxic tissue [20,21]. In our study, acidifi cation right-shifted the HbO 2 curve of stored EC close to that of fresh blood; after 2.5 weeks of storage the curve at pH 7.10 was similar to that of fresh blood at pH 7.40.
The term consumable (or extractable) O 2 has been used to denote the amount of oxygen that can be extracted from the blood before tissue cells becomes dysfunctional because of hypoxia [31,32], calculations of this parameter give an indication of the consequences of HbO 2 curve shifts for O 2 tissue supply. The results of such calculations are not exact, as the use of a lower borderline value of 2.7 kPa (20 mmHg) is somewhat arbitrary and direct measurements of