The oxygen dissociation curve of blood in COVID-19.

COVID-19 hinders oxygen transport to the consuming tissues by at least two mechanisms: In the injured lung, saturation of hemoglobin is compromised, and in the tissues, an associated anemia reduces the volume of delivered oxygen. For the first problem, increased hemoglobin oxygen affinity [left shift of the oxygen dissociation curve (ODC)] is of advantage, for the second, however, the contrary is the case. Indeed a right shift of the ODC has been found in former studies for anemia caused by reduced cell production or hemolysis. This resulted from increased 2,3-bisphosphoglycerate (2,3-BPG) concentration. In three investigations in COVID-19, however, no change of hemoglobin affinity was detected in spite of probably high [2,3-BPG]. The most plausible cause for this finding is formation of methemoglobin (MetHb), which increases the oxygen affinity and thus apparently compensates for the 2,3-BPG effect. However, this "useful effect" is cancelled by the concomitant reduction of functional hemoglobin. In the largest study on COVID-19, even a clear left shift of the ODC was detected when calculated from measurements in fresh blood rather than after equilibration with gases outside the body. This additional "in vivo" left shift possibly results from various factors, e.g., concentration changes of Cl-, 2,3-BPG, ATP, lactate, nitrocompounds, glutathione, glutamate, because of time delay between blood sampling and end of equilibration, or enlarged distribution space including interstitial fluid and is useful for O2 uptake in the lungs. Under discussion for therapy are the affinity-increasing 5-hydroxymethyl-2-furfural (5-HMF), erythropoiesis-stimulating substances like erythropoietin, and methylene blue against MetHb formation.

INTRODUCTION

Infection with the coronavirus SARS-CoV-2 impairs respiratory function, hindering oxygen uptake in the lungs and causing hypoxemia. One should expect compensatory effects on blood oxygen transport function, especially changes of hemoglobin oxygen affinity. Astonishingly, we could find only four short papers with measurements of oxygen affinity in patients with COVID-19 and two commenting letters (PubMed and Google Scholar); an increase in hemoglobin oxygen affinity was detected only in one investigation. Additionally, a review contains only a general description of the oxygen dissociation curve and possible contrary effects (hypocapnia vs. fever), but no data (1).

However, the analysis of the current literature on the disease and related topics shows that counteracting factors may in fact mask a marked influence of COVID-19 on the oxygen dissociation curve.

LUNG INJURY AND RESULTING EFFECTS ON OXYGEN UPTAKE

One of the most obvious effects of COVID-19 infection is the reduction in oxygen loading into blood during the lung passage. There is a variety of causes for this effect.

Intrapulmonary shunting because of edema and atelectasis, loss of perfusion regulation, intravascular microthrombi, in more severe cases reduced lung compliance, increased dead space, increased consolidation and atelectasis (2), lung edema and fibrosis with increase of gas transmission distances (3) result in hindered O2 uptake and in CO2 retention. In the initial phase, however, CO2 retention is moderate, because the diffusion constant is 25 times higher for CO2 than for O2 (4). At rest, moderate hypoxia is not particularly distressing (e.g., combined hypoxia and hypocapnia during sitting at an altitude of 3,500 m) in contrast to hypercapnia (e.g., combined hypoxia and hypercapnia during breathholding); only few subjects can sustain the latter longer than 2 min. Lacking hypercapnia/CO2 retention is probably the cause of the initial “happy hypoxemia” in patients with COVID-19 (5). A further possible cause is infection of O2-sensing glomus cells by SARS-CoV-2 affecting central breathing regulation (6).

OXYGEN TRANSPORT IN BLOOD: OVERVIEW

Hemoglobin concentration decreases during the course of COVID-19, but the extent is variable. Kumar et al. (7) have communicated a mean value of 13.2 ± 2.0 g/dL (standard deviation is always used) in 1,973 patients. In 1,462 patients with concentrations ranging between 11.8 and 15.2 g/dL, the mean difference between nonsevere and severe cases amounted to ∼0.8 g/dL (2). In a study by Vogel et al. (8), anemia was more pronounced (8.1 ± 1.2 g/dL) in 43 critically ill patients in an intensive care unit. Overall, these publications suggest that total hemoglobin mass is reduced in COVID-19 albeit definite data do not exist.

An essential and frequently discussed [e.g., recently by J. A. Dempsey (9)] factor for loading (in the lungs) and unloading (in the tissues) of oxygen is hemoglobin oxygen affinity. Affinity is an intrinsic property of hemoglobin in red blood cells that can be modified by a variety of factors. For its determination, the oxygen dissociation curve (ODC) has to be measured. This is usually performed in vitro by equilibration of blood with varying Po2.

Characteristic properties of the ODC reviewed by Mairbäurl and Weber (10) are its position (half-saturation pressure P50) and its slope (n in the logarithmic Hill plot), both usually determined in vitro at standard conditions (pH 7.4, Pco2: 40 mmHg, temperature: 37°). A low P50 is favorable for oxygen uptake in the lungs; a high P50 is favorable for oxygen supply to the tissues. In vivo variations in pH, Pco2, and temperature as well as additional, partly unknown factors (reviewed in therapy of anemia and oxygen affinity disturbances) may cause such favorable effects and result in a steeper curve in vivo as compared with in vitro.

Intraerythrocytic factors relate first to the type of hemoglobin, which is HbA in the majority of humans. Changes in red cell metabolism may influence the concentrations of 2,3-bisphosphoglycerate (2,3-BPG) and adenosine triphosphate (ATP), which reduce oxygen affinity by binding to and stabilization of deoxyhemoglobin. In contrast, reduced glutathione (GSH)—important for radical defense—binds to oxy-Hb (11, 12), thus shifting the ODC leftward. The concentration of GSH is, however, smaller than that of 2,3-BPG.

Extraerythrocytic factors relate to changes in plasma composition. Exchanges with neighboring interstitial fluid and tissue cells are relevant for the concentrations of CO2, Cl−, lactic acid, other fixed acids, and water (osmolality) in the erythrocytes. Although osmolality affects the concentration of all solvents, changes in specific solvents may cause specific effects. All acids influence the dissociation state of various Bohr groups of the Hb molecule. CO2, Cl−, and lactic acid additionally bind to end-terminal N (e.g., Ref. 13). Lactic acid is also produced within the erythrocyte but equilibrates slowly with plasma dependent on monocarboxylate-transporter 1 (14–16).

Nitrogen monoxide (produced either from l-arginine or from ingested nitrate) binds to oxy-Hb as well as deoxy-Hb at different sites; the micromolar concentration in vivo is, however, much too low to play an important role in the regulation of oxygen affinity (10, 17). Yet, binding of deoxy-Hb to the anion exchanger (AE) in the cell membrane is important for liberation of nitric oxide (NO) from red blood cells causing vasodilation (18). It can, however, be assumed that this effect, too, does not cause a notable change in Hb-O2 affinity, as the total number of Hb molecules is ∼225 times higher than that of band 3 proteins (10, 19).

In clinical routine, the calculus of J. Severinghaus (20, 21) is often used to describe the standard ODC (plasma pH: 7.4, Pco2: 40 mmHg, temperature: 37°C). These data were originally compiled from various sources and correspond rather well to more recent studies. The standard P50 amounts to 26.6–26.7 mmHg.

For acid and temperature correction, the following formulas are most often used:

Bohr shift by CO2 and fixed acid (e.g., lactic acid): Δlog Po2 = −0.48ΔpH + 0.0013ΔBase Excess

Temperature shift: Δlog Po2 = 0.024ΔT

Notably, the former correction is a simplification, as the Bohr effect is saturation dependent below 20 and above 90% So2 (e.g., Refs. 22, 23).

INTRAERYTHROCYTIC CHANGES IN COVID-19

Does the coronavirus damage the functions of red cells? Wenzhong and Hualan (24) performed a computer study considering possible effects of the virus on hemoglobin using structural data of both. They concluded that various components of the virus (Orf1ab, ORF3a, and ORF10) tend to damage the 1-β chain of deoxyhemoglobin and liberate iron-free porphyrins. However, Read (25) as well as DeMartino et al. (26) have heavily contradicted these calculations, stating that methods as well as conclusions were not correct. In any case, the suggested reaction might only occur in erythrocytes with damaged membrane or with Hb molecules dissolved in plasma. Red blood cells do not have ACE2 receptors on their surface, hence, even in cases of viremia (which is extremely rare, e.g., Ref. 27), the virus will not be able to enter the cell interior and get in contact with Hb.

Experimental studies, however, have detected some indications for red blood cell damage. Venter et al. (28) observed “structural pathologies in platelets and erythrocytes” and suggested that red blood cell damage occurred in the vessels. In severe cases, microthrombi especially in the form of amyloid clots are frequent. Similar results were published by Sweeney et al. (29) who observed many schistocytes (fragments of erythrocytes) in blood smears. Gerard et al. (30) described abnormal red cell morphology especially “mushroom-shaped erythrocytes.” In addition, red cells with damaged membranes (fragmentocytes; Fig. 1) have been detected. Obviously, anemia is largely caused by hemolysis and blood clotting. The importance of additional factors (hemodilution by infusions, nutritional deficiency, hemorrhage, and phlebotomy) is of course variable (31, 32).

Figure 1.

Fragmentocytes in a lung capillary from a 58-yr-old patient with COVID-19 (unpublished results used with permission from Konrad Steinestel and Wilhelm Bloch). Lung tissue was immersion-fixed with 4% paraformaldehyde, Epon embedded following standard procedures. Thereafter 70- to 90-nm thick sections were cut with an Ultracut UCT ultramicrotome (Fa. Reichert). The sections were studied with a Zeiss EM 109 electron microscope (Fa. Zeiss). Original magnification: ×4,000, image width: 23.5 µm. (Courtesy of Prof. Dr. Konrad Steinestel, Institute of Pathology and Molecular Pathology, Bundeswehrkrankenhaus Ulm, Oberer Eselsberg 40, 89081 Ulm, Germany.)

Aside of these structural damages, the function of Hb is influenced by formation of methemoglobin and CO-Hb (eventually more than 20%), which cannot bind oxygen reversibly (33). The consequences are described in the oxygen dissociation curve in covid-19.

A very extended and thorough metabolomic, proteomic, and lipidomic analysis of red cells in patients with COVID-19 has been published by Thomas et al. (34). Specifically, the authors analyzed the impact of COVID-19 by comparing erythrocytes from 23 subjects with moderate anemia without exact diagnosis (mean [Hb] 11 g/dL) and 29 patients with COVID-19 (mean [Hb] 12 g/dL). They detected various metabolic and proteomic changes in the latter: increased levels of glycolytic intermediates, oxidation and fragmentation of ankyrin, spectrin β, and the N-terminal cytosolic domain of band 3 (AE1), altered lipid metabolism, especially of saturated fatty acids, acyl-carnitines, and sphingolipids. The demonstrated complex alterations of red blood cells by the disease offer different possible mechanisms, which change oxygen handling in the erythrocytes. Besides direct influences, morphological alteration and changed deformability can cause protein and membrane lipid modification, which indirectly hinder oxygen handling of red blood cells. Such changes could also alter perfusion of lung and peripheral tissues. Thomas et al. (34) suggested from these findings “RBC be less capable of responding to environmental variations in hemoglobin oxygen saturation/oxidant stress when traveling from the lungs to peripheral capillaries and vice versa.”

Especially important for oxygen binding are 2,3-BPG, ATP, Cl−, lactate (La-), and GSH. GSH showed no significant differences between patients with COVID-19 and the “controls.” [2,3-BPG] and [La-], however, were increased.

Messner et al. (35) have shown that changes in blood composition depend on disease severity. However, unfortunately, nothing is communicated in the article of Thomas et al. (34) about the severity of disease in the patients studied for COVID-19. It seems, however, fair to assume that severity was mild to moderate, as no patient was mechanically ventilated and the blood composition was rather normal. No marked alterations of clinical hematological parameters, such as red blood cell (RBC) count, hematocrit, or mean corpuscular hemoglobin were detected. The Hb concentration amounted to ∼12 g/dL, which indicates moderate anemia. Because of technical reasons, 2,3-BPG and ATP concentrations are only presented in arbitrary units. However, as 2,3-BPG is usually increased in anemia (see therapy of anemia and oxygen affinity disturbances), this was probably the case in the control group and even more in COVID-19.

To conclude, additional measurements in more severely ill patients would be of great interest to obtain a better understanding of changes in erythrocyte composition that may affect oxygen binding and transport in COVID-19.

THE OXYGEN DISSOCIATION CURVE IN COVID-19

The deterioration of lung function after infection with SARS-CoV-2 raises the importance of alternative compensation mechanisms that may help to secure oxygen uptake. One possibility to improve O2 loading is increased hemoglobin oxygen affinity visible as a left shift of the ODC. The resulting drawback for oxygen delivery to the consuming organs may be compensated by increased perfusion or by recruitment of additional tissue capillaries. The latter effect reduces the distance to consuming cells and thus the necessary diffusion pressure for oxygen. A further beneficial mechanism might be an enlarged slope of the curve detectable as high n value in the Hill plot. This lowers the Po2 necessary for oxygen loading in the lungs and raises the Po2 at low oxygen saturation in the consuming tissues. In this surrounding also increased Pco2 and temperature support oxygen delivery. Finally, in vivo factors suggested in various papers might influence oxygen flow from alveoli to lung capillaries or from tissue capillaries to cells (reviewed in Refs. 23, 37).

Only few publications have addressed potential changes of the ODC in COVID-19. De Martino (26) measured Po2/So2 pairs in venous blood from 17 patients (So2 range 20%–95%); the measured data points tended to lie to the right of the standard curve, yet without a significant difference. It is, however, unclear whether the measured Po2 values were corrected to standard conditions. The authors concluded “the patients with COVID-19 did not exhibit any hemolytic anemia,” but mean [Hb] was low (∼13 g/dL according to figures, ranging between 7 and 17 g/L), and ferritin as a possible indicator of hemolysis reached partly high values (mean value 500 ng/mL).

Daniel et al. (38) investigated 14 patients with COVID-19 (7 males, 7 females) with marked anemia ([Hb] 9.3 ± 2.3 g/dL) and compared them to 11 subjects (5 males, 6 females) apparently without lung or blood pathologies (no diagnosis communicated, [Hb] 14.3 ± 1.1 g/dL). They, too, did not find a difference in standard P50, which was relatively high in both groups (29.0 ± 2.3 mmHg vs. 28.5 + 1.8 mmHg). The Hill slope was within the normal range (2.72 ± 0.09). Notably, however, the authors used the highly artificial Hemox-Analyzer system for their analyses (39, 40), which determines oxygen dissociation curves of erythrocytes not suspended in plasma, but in a large volume of buffer solution (50 µL blood/5 mL buffer), which contains no Ca2+, bicarbonate, proteins, and lipids, and in the absence of CO2. Accordingly, many substances in the plasma, which interchange with the red cells, are markedly diluted. In vivo effects of altered bicarbonate and lactate (La−) concentrations in the erythrocytes are largely abolished, as these molecules easily cross the red cell membrane and will “disappear” in the buffer solution. A similar criticism has been published by Harutyunyan et al. (41) and was accepted by the authors (42). We have therefore modified this method in our laboratory many years ago in that we used a plasma-like bicarbonate buffer and added CO2 to the equilibration gas [e.g., Böning and Enciso (43)]. Nevertheless, differences to classical methods remained, e.g., possible emigration of La−. Therefore, we returned in later investigations to whole blood equilibration. An important result of the study by Daniel et al. (38) is, however, the probable exclusion of marked damages to many Hb molecules (42) in contrast to previous calculations by Wenzhong and Hualan (24).

Renoux et al. (44) measured similar standard P50 in seven patients (median [25-75] percentiles: 29.7 [27.2; 31.0] mmHg) also with the Hemox-Analyzer, which were not significantly different from healthy subjects, too. Unfortunately, the median may deviate markedly from the mean value in such a small sample.

The notable advantage of a study by Vogel et al. (8) is that many repeated measurements were performed under approximated in vivo conditions (native arterial or venous blood at 37°, no equilibration) in large numbers of intubated and mechanically ventilated patients (3,518 samples in 43 patients with COVID-19, 15,945 samples in 828 patients with other causes of respiratory failure) during 1 mo. In COVID-19 disease, these nonstandardized Po2/So2 points (eFig. 1 in their paper) were left-shifted compared with the Severinghaus standard curve (P50 26.7 mmHg) except below 50% So2. The latter was obviously caused mainly by venous samples where influences of oxygen-consuming tissues might play a dominant role. Hence, these data points are less relevant for lung function and oxygen uptake. The authors present standard P50 applying the Bohr and temperature coefficients introduced by Severinghaus (20) to their values. The deviations of pH (7.382 ± 0.077), base excess (1.2 ± 4.3 mmol/L), and temperature (36.8 ± 0.8°C) from standard values are small, thus not compromising the calculated P50 values to a relevant extent. These standard P50 were significantly lowered compared with Severinghaus’s value (23.4 ± 3.1 vs. 26.7 mmHg, P < 0.0001). The decrease was rather constant during the whole observation period in most patients.

A drawback of all these studies on the ODC is that the data are not presented separately for men and women. In healthy women before menopause, the average P50 tends to higher values, whereas the Bohr effect is lowered (23, 45, 46). But probably this effect is not large, as the mean age of the patients was above 50 yr.

Nothing is reported about the 2,3-BPG concentration in these patients as well as in the other publications in this section. The probable cause is that test kits are no more available. According to the findings by Thomas et al. (34) as well as in various anemia studies (see the effect of anemia in covid-19), an increase can also be expected during the new disease. Further noteworthy is the fact that none of the studies discussed in this section communicated any information on drugs applied to the patients, which might influence oxygen affinity (mentioned in INTRAERYTHROCYTIC CHANGES IN COVID-19).

Additionally, there are possible influences besides the intraerythrocytic damages mentioned above:

How large is the amount of Hb in plasma because of hemolysis? To our knowledge, nothing has been communicated but hemolysis indicators partly are changed. Although haptoglobin seems to be unaffected (26), ferritin is increased (26, 28, 47). If the amount of Hb in plasma is remarkable, there should be a left shift of the ODC (higher pH, less 2,3-BPG binding). To our knowledge, this has not been measured during COVID-19, but concentrations up to 2 g/dL have been detected after heavy physical exercise (48). One should expect similar effects, if the cell membrane is damaged leading to increased pH and BPG loss.

Is the mean age of the red cells changed as a consequence of RBC damage and perhaps therefore increased erythropoiesis? The half-saturation pressure of young erythrocytes is high because of increased [2,3-BPG], whereas the Bohr coefficients are low (49).

How large is the influence of body temperature? This plays no role in the study of Vogel et al. (8). But most patients suffer from fever, in ∼80% it is higher than 38°C (50, 51). This shifts the ODC rightward: P50 rises by almost 3 mmHg at 39°C (29.8 mmHg).

To summarize the content of this paragraph: at first glance, it is astonishing that oxygen affinity seems not to be hampered by SARS-CoV-2 infection. P50 for standard conditions measured in vitro was not changed. Affinity was even increased in most patients, if measurements were performed in untreated venous or arterial blood (in vivo P50) of patients with severe anemia. Interestingly, the velocity of oxygen unloading was not changed (52).

THE EFFECT OF ANEMIA IN COVID-19

Anemia of varying severity is very distributed in COVID-19 (2, 7, 8, 53). P50 often increases during anemia (54–57) due to a stimulated 2,3-BPG synthesis, which results from lowered venous O2 saturation during tissue passage (reducing free BPG because of improved binding to Hb) or hyperventilation (increased pH). This effect is independent of the type of illness but decreased pH may cancel the P50 change (32, 58). An additional factor in hemolytic anemia or in renal anemia during erythropoietin therapy (54) is the rising proportion of young erythrocytes, which contain high 2,3-BPG concentration (49). The importance of additional factors (hemodilution by infusions, nutritional deficiency, hemorrhage, and phlebotomy) is of course variable (31, 32).

Reduced erythrocyte production ([Hb] 7.3 ± 2.3 g/dL) as well as loss of erythrocytes ([Hb] 7.9 ± 2.8 g/dL) caused a rise by ∼4 mmHg from 26.7 ± 1.7 to 30.5 ± 3.1 and 30.8 ± 3.0 mmHg, respectively, following from an increase of [2,3-BPG] to 18.6 µmol/g Hb (0.7 mmHg per 1 µmol 2,3-BPG//gHb) in two groups of patients with anemia (43). Additionally, the slope n of the ODC was increased especially at high saturation. Interestingly, these patients lived at 1,100 m of altitude, where already a slight decrease of alveolar Po2 occurs (−10 mmHg), i.e., slightly above the threshold for hypoxic reactions (59, p. 42), which mimics the situation in COVID-19.

In contrast, in the investigation of Daniel et al. (38), the patients with COVID-19 were anemic (Hb 9.3 ± 2.3 g/dL) but surprisingly, P50 as well as n were not significantly increased compared with the control group (14.3 ± 1.1 g/dL). This was similar in the small investigation by Renoux et al. (44). The most severe anemia (8.1 ± 1.2 g/dL) in critically ill patients was observed by Vogel et al. (8). They, however, reported even a left shift of the in vivo standard ODC to P50 (23.7 mmHg). The total P50 difference to the non-COVID-19 study by Böning and Enciso (43) amounts to 7 mmHg.

Unfortunately, only the paper by Thomas et al. (34) reports data on 2,3-BPG in COVID-19 showing a significant rise together with moderate anemia. But the amount is not clear because the applied method yields only arbitrary units.

When compared with the usual effects of anemia as well as of hypoxia, namely, a right shift of the ODC by increased 2,3-BPG concentration, COVID-19 obviously erases this effect or causes even a left shift. Without this back shift, the loading of blood with oxygen in the damaged lung would be additionally hindered by lowered oxygen affinity. The problem of survival is thus partly shifted from oxygen uptake in the lung to oxygen transport from capillaries to consuming cells. Increased perfusion as during exercise is necessary.

A probable cause for the left shift of the ODC even in presence of increased [2,3-BPG] is the formation of methemoglobin or COHb. If one hem group in the quaternary Hb molecule is damaged, the affinity of the remaining Hb monomers increases resulting in a left-shifted oxygen dissociation curve. According to Hrinczenko et al. (60), P50 is reduced by 0.2 mmHg/%methemoglobin (MetHb). During COVID-19, COHb remains rather constant, but the change of MetHb is variable [reviewed by Scholkmann et al. (33)]; in a group of 25 patients, it rose to 16.4 ± 9.1% on day 6 of the hospital stay. Thus, 3.3 mmHg decrease in P50 might be expected corresponding to the effect of a pH increase by 0.15 U or a 2,3-BPG reduction by 1 mmol/L red cells or 5 µmol/g Hb, respectively. Suggested causes were drugs like chloroquine and hydrochloroquine (often used in the first half of 2020, when Vogel’s measurements were performed) especially in connection with glucose-6-phosphate dehydrogenase (G6PD) deficiency, increased NO formation due to anemia (necessary for sufficient blood flow), and oxidative stress, finally the age of the patients. Erythrocytes of old subjects are especially sensitive to oxidative stress.

Obviously a large rise of MetHb can nearly equalize the expected P50 increase caused by the high [2,3-BPG] during severe anemia (+4 mmHg). Yet compared with 23.7 mmHg for standard conditions in anemic patients with COVID-19 (8), this difference rises even to 7 mmHg. This is probably caused by additional in vivo effects, which have been observed in normal subjects as well as in patients with cystic fibrosis (23) and during acute or chronic altitude sojourns above 3,300 m (61, 62). Surprisingly when venous Po2 was corrected to pH 7.4, the values were significantly lower than expected from the individual standard curve at saturations above 50% (Fig. 2). Additionally, Hill’s coefficient amounted to 3.4 at altitude. The findings are supported by 15,495 additional measurements in 828 critically ill patients (Hb 9.4 g/dL) in the article of Vogel et al. (8). In spite of low Hb (9.4 g/dL), mean in vivo P50 amounted to only 24.6 mmHg. In contrast, in vivo right shifts have been described at low oxygen saturation especially during exercise (reviewed in Ref. 63). We have discussed a variety of possible mechanisms (23, 37): concentration changes of Cl−, 2,3-BPG, ATP, La−, nitrocompounds, glutathione, glutamate are possible, partly because of the delay between blood sampling and end of equilibration and partly because of the contact with the tissues. For example, La− is steadily produced by glycolysis in the red cell. La- as well as Cl− partly bind to deoxy-Hb (e.g., Ref. 13) and reduce oxygen affinity, thus shifting the ODC rightward. With increasing saturation, they are liberated and may leave the red cell supported by a changed Donnan equilibrium. In vitro, the additional distribution space is limited to plasma. In vivo, the ions may emigrate to the three times larger interstitial volume or even the tissue cells. Their concentration and thus their influence on oxygen affinity decreases.

Figure 2.

Deviation of in vivo Po2 (heparinized blood measured immediately after sampling) corrected to pH 7.4 from the corresponding individual in vitro standard ODC between 40% and 90% So2 in 14 controls and 14 patients with cystic fibrosis. The in vivo left shift facilitates oxygen loading in the lungs. [From Böning et al. (23) with permission from PLoS One.]

A synopsis of all possible effects is shown in Fig. 3.

Figure 3.

Half-saturation pressures (pH 7.4, Pco2, 37°). Standard: in vitro standard according to Severinghaus (20); anemia: in vitro according to Böning and Enciso (43); Methemoglobin (MetHb): effect on anemia calculated according to Scholkmann et al. (33); and COVID-19 in vivo: untreated blood samples according to Vogel et al. (8).

Of importance is the effect on the volume of oxygen bound in arterial blood. The amount depends on the upper part of the dissociation curve and can be calculated using data assembled by Sharan et al. (64). At 80% arterial saturation, a plausible value for patients with COVID-19, 100 mL blood with normal Hb concentration (15 g/dL) and normal P50 (26.6 mmHg) can bind:

15 g/dL × 1.34 mL/g × 0.80 = 16.1 mL/dL oxygen

1.34 O2 bound by 1 g of Hb (Hüfner’s number)

With only 10 g Hb/dL, this decreases to 10.7 mL/dL. For a right-shifted ODC (P50 30.8 mmHg), we obtain only 9.8 mL/dL. If this curve is again left shifted by a P50 decrease of 7 mmHg, So2 rises to 90% and O2 content to 11.8 mL/dL. The effect is double: loading of O2 is facilitated and the amount is increased by 2 mL/dL or 20%. On the other hand, this completely compensates the drawback of 16% MetHb.

THERAPY OF ANEMIA AND OXYGEN AFFINITY DISTURBANCES

Notably, Woyke et al. (65) have suggested to use 5-hydroxymethyl-2-furfural (5-HMF) for treatment in COVID-19 disease. This substance binds to 1α valine of the hemoglobin molecule and increases oxygen affinity. A similar effect is caused by Voxelotor (66). Whether this additional left shift of the ODC is useful, may be questioned. The substance can only be beneficial, if at the tissue level sufficient O2 delivery is secured by increased perfusion or local mechanisms (67, 68). As means against anemia, erythropoietin may be considered. Hadadi et al. (69) communicated in a case report “miraculous improvement in his symptoms and hemoglobin level” (from 6.7 to 9.3 g/dL) after 1 wk of treatment. Various authors (e.g., Refs. 70–72) suggest application of erythropoietin not only because of its stimulation of erythropoiesis but also due to suggested additional effects (improved respiration, counteracting inflammation, neuroprotection, and neuroregeneration) similar as against high-altitude disease. High-altitude physiologists (73), however, heavily contradict to this “mistaken analogy” and mention that it is “associated with an increased risk of thrombotic cardiovascular events.” Similarly, clinicians warn against the application of erythropoietin because of this danger (74). Additionally, erythropoiesis-stimulating agents (ESAs) are considered as possibly dangerous, too (75), instead blood transfusion is recommended. One might discuss to use blood stored until 2,3-BPG content is lowered. Therapeutic use of NO, as proposed by Kobayashi and Murata (76), potentially causes additional methemoglobin as well as Hb[FeNO] formation depending on the dosage (10). Yet inhaled NO may optimize ventilation/perfusion matching, thereby increasing the total O2 load. The applicability is still investigated (77). A drug which inhibits methemoglobin formation is methylene blue (78, 79). Probably this does not change oxygen uptake in the lungs, as the reduced left shift of the ODC is compensated by increased concentration of functional Hb. But the rise of normal Hb with changing affinity according to metabolic necessities is surely important for oxygen delivery to the consuming cells. A drawback is, however, the inhibition of soluble guanylyl cyclase and thus vasodilation by NO (80).

CONCLUSIONS

Based on the currently available data, it seems that the mean standard P50 in SARS-CoV-2 infection is not increased as would be expected under hypoxic and anemic conditions in humans. In the largest study by Vogel et al. (8), it is even decreased to 23.4 mmHg compared with 26.7 mmHg in healthy subjects. Patients in this study suffered from not only lung injury but also anemia, which may cause a P50 increase by 4 mmHg following from 2,3-BPG synthesis (e.g., Ref. 43). Thus, the real P50 difference between COVID-19 and known types of anemia might rise to 7 mmHg. At low pulmonary Po2 in the injured lung, this amounts to 2 mL O2/100 mL blood more than in “normal” anemia (+20%). This pronounced left shift in patients with COVID-19 compared with “regular anemics” is probably caused partly by methemoglobin formation, which increases oxygen affinity of the remaining undamaged Hb monomers. Additionally, still unknown unstable in vivo factors probably increase oxygen affinity. These changes are helpful for oxygen loading in the compromised lung during COVID-19.

For further clarification, improved measurements of in vitro and in vivo oxygen affinity, 2,3 BPG concentrations, and Hb mass should be performed.

NOTE ADDED IN PROOF

Recently, an article was published stating that there is no effect of COVID-19 on the oxygen dissociation curve (81). However, in this study only nonsevere cases mostly without anemia were investigated.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

W.B. prepared figures; D.B., W.M.K., and W.B. drafted manuscript; D.B., W.M.K., and W.B. edited and revised manuscript; D.B., W.M.K., and W.B. approved final version of manuscript.