Our search yielded less than a hundred studies on this topic. This is comparatively small considering the number of drugs in clinical use worldwide or in current development for clinical use. Thus, the vast majority of drugs routinely administered to humans have not been studied for their effects on Hb oxygen transport. In addition, we found that a significant proportion of the reviewed publications were of poor study quality, with many providing little or no information on statistics, number of subjects, methods, or even test organisms. Other studies suffer from small numbers of participants or test runs that are unlikely to produce statistically significant results. Unfortunately, many studies did not report the distribution of male and female participants, and when they did, males were overrepresented in most cases. This is a notable omission given that the P50 values are significantly different between women and men [13]. In addition, there is a remarkable proportion of studies in which the authors declared a conflict of interest (e.g., paid by the manufacturer of the product), which may also limit comparability of the results. Finally, a large number of studies are outdated and deal with substances that in some countries are no longer used or recommended in clinical use and have been replaced by others. This is the case, for example, for propranolol, a ß-blocker and class II antiarrhythmic used in cardiology with a significant right-ward shift effect on the ODC shown in several studies. This effect is discussed to be due to the release of 2,3-BPG from the erythrocyte membrane and binding to Hb, thereby reducing the oxygen affinity of Hb [14]. This right-ward shift of the ODC might enhance oxygen extraction at the tissue level. However, propranolol has been replaced in many intensive care units for the treatment of arrhythmia by short acting ß-blockers like esmolol or landiolol, for which there are no data on potential effects on oxygen binding to Hb. This is another example of why further studies in this area are urgently needed.

There may be many reasons for this unsatisfying study situation. We speculate that the limited accessibility of appropriate analytical methods necessary to perform an ODC measurement is one of them. For example, the Hemox Analyzer is only available in a few institutions because it is quite expensive for a single-purpose device. Other methods or devices (except blood gas analysis) are not commercially available and would have to be custom-built, which could prevent many researchers from carrying out specific projects. In addition, the vast majority of methods suffer from the limitation that measurements in single cuvette systems, such as the Hemox Analyzer, are cumbersome, time-consuming, and do not provide the ability to run controls side by side. Another reason for the paucity of studies may be that the interest in the ODC has steadily declined since the 1970s and 1980s, as analyzed in Fig. 2. We can only speculate, but perhaps the impact of the ODC in a clinical setting has been overlooked, or people are not really aware of its significance.

The importance of the ODC has already been mentioned in the introduction but should be underscored by the following examples: The working myocardium has an exceptionally high oxygen extraction rate and is very vulnerable in ischemic and hypoxic situations. Right- or left-ward shifted ODCs, resulting in either increased (right-ward shifted) or decreased (left-ward shifted) tissue oxygenation, would have a significant impact on heart functionality, especially in any kind of diseased state when oxygen delivery is generally low. For example, Lucas et al. injected 5-hydroxymethylfurfural (5-HMF) or only vehicle to hamsters [15]. The 5-HMF induced left-ward shift of the ODC resulted in improved cardiac indices, stroke volume, cardiac output, ejection fraction, and stroke work compared to control group when exposed to hypoxia. On the other hand, Watanabe and colleagues showed that mice with severe heart failure and treated with the oxygen affinity modulator RSR13 (which induces a right-ward shift) had improved treadmill running performance due to increased oxygen delivery in skeletal muscle [16]. P50 was increased by 12.5% in their experiments. In humans, a right-ward shift of the ODC by the same amount would increase P50 from a normal value of 26.7 (Fig. 4; black line) to 30.0 mmHg (dashed line). Assuming normoxia and normal pulmonary oxygenation, Hb is almost completely saturated in both cases (SO2 97.3% vs. SO2 96.3%; right vertical solid line), while at the tissue level, assuming a PO2 of 20 mmHg (left vertical solid line), SO2 is lower in the right-shifted ODC and correspondingly more O2 is extracted (SO2 31.4% vs. SO2 25%). In this example, the oxygen extraction rate (O2ER = (SaO2—SvO2) / SaO2); a = arterial, v = venous) is 74.0% compared to 67.7%. Further assuming a Hb concentration of 150 g/l and using a Hüfner number of 1.34 ml O2/g Hb, this would mean that for every liter of cardiac perfusion, an additional ~ 11 ml of pure O2 would be available to the cells. Although we are not aware of a human cohort study demonstrating a direct physiological effect on patients with this additional amount of O2, all cardiologists would probably agree that anything that improves tissue oxygenation and prevents tissue hypoxia should be beneficial to a critically ill patient at some point. In a case report by Al-Qudsi et al., two patients with persistent severe hypoxic respiratory failure were treated with the anti-sickling agent voxelotor [17]. This drug stabilizes Hb in its oxygenated state by inducing a left-ward shift of the ODC. During the treatment, the oxygen saturation to FiO2 ratio increased substantially, reducing the invasiveness of mechanical ventilation. Unfortunately, the change in tissue oxygenation by this left-shifted ODC was not investigated in this study.

Standard ODC (black sigmoidal curve) and ODC with a 12.5% increase in P50 (dashed curve). Given a tissue PO2 of 20 mm Hg (left vertical line), O2 extraction (curly brackets) is greater in the right-shifted ODC. For details, see Text

Our search revealed a handful of methods for recording an ODC, and the most common ones are briefly described here. First and foremost is a tonometric approach, where blood samples are successively exposed to ≥ 2 different gas mixtures with predefined PO2, while SO2 is determined by dual wavelength absorption measurements. Continuous methods, in which the oxygen content of the gas phase is constantly decreasing or increasing, require the measurement of PO2 and SO2 in the sample at certain time intervals. Continuous methods are largely based on the method described by Duvelleroy and colleagues, where a known volume of deoxygenated blood is exposed to a known volume of O2 and PO2 and oxygen content are measured [18]. In the mixing technique, blood samples are divided into two aliquots, one exposed to an oxygen-rich gas mixture (SO2 = 100%) and the other to an oxygen-free gas (= SO2 0%). These two aliquots are then mixed in predefined volume fractions to obtain aliquots of known SO2. PO2 is then measured in these aliquots. In vivo studies are usually based on blood gas analysis, where PO2 and SO2 are measured from a blood sample and the P50 is then extrapolated based on a predefined algorithm [19]. Finally, the Hemox Analyzer is a commercially available instrument (TCS Scientific Corp., PA, USA) that has been on the market for more than 40 years. It determines the ODC by exposing 2–50 µl of whole blood or hemolysate, diluted in 5 ml of buffered, anti-foaming solution, to a decreasing partial pressure of oxygen in an optical cuvette under constant agitation. The change in oxygen tension is detected by a Clark oxygen electrode inside the cuvette, while the decrease in oxyhemoglobin fraction (% HbO2) is simultaneously monitored by dual wavelength spectroscopy at 560 nm and 576 nm, respectively. Recently, high-throughput assays have been developed that allow simultaneous measurements of large numbers of samples in microplate reader instruments [20, 21].

Our search has limitations that should be addressed here: First, our scoping review may not be fully replicable by all researchers, as it was limited to publications accessible through our institutional subscriptions. This limitation is important, as different institutions have varying agreements with publishers, and these agreements change over time (see Supplementary Information for the full list of journals accessible through our library). Second, our review does not claim to be exhaustive. Each search mask returns results based on its specific criteria, meaning that papers without the defined keywords may be missed. To avoid this, one would need to define more search forms, which, however, would significantly increase the effort while simultaneously reduce reproducibility. We opted for a strategy we believed would yield the most comprehensive results. Next, we consulted only two electronic databases, PubMed and the Cochrane Library, recognizing that other databases and libraries may include additional studies relevant to this work. However, PubMed was chosen because it is the most popular and likely most relevant database, with 23 million references in 2014 [22]. The Cochrane Library, with 800,000 references in 2014 [22], provides high-quality, peer-reviewed evidence in medical research and other services, mostly in the form of reviews. Google Scholar was not included because it returned almost identical results to PubMed. Furthermore, most of the studies are quite old. More than half of them were published before 1990, and most of them date back to the 1970s. Also, there were also almost no studies from the 1960s and earlier. There could be several reasons for this: English as a scientific language may not have been so common, the databases have not yet fully captured older or very old papers electronically, or the publications are not available in full text (only abstracts), so they did not meet our selection criteria. Another reason could be that the Hemox Analyzer, which greatly simplified the often complex measurements, was not introduced before the 1980s.