Is there a shift of the oxygen-hemoglobin dissociation curve in COVID-19?

to the editor: We read with great interest the review by Böning et al. (1), which provided insightful perspectives about hemoglobin-oxygen affinity in COVID-19. We recently published a study comparing 253 blood gas samples from 100 patients with COVID-19 with 221 samples from 100 non-COVID-19 controls with homogenous age and sex ratio distribution (2). After standardization for normal conditions (body temperature = 37°C; pH = 7.4; Pco2 = 40 mmHg), we computed P50 in approximated in vivo conditions from samples with hemoglobin saturation <97%, in accordance with the manufacturer’s instructions (3). As this P50 calculation is not as accurate as that used by Böning et al. (1) and Böning and Enciso (4), we also compared standardized measured oxyhemoglobin in each sample with predicted oxyhemoglobin given by the standard oxygen-hemoglobin dissociation curve (ODC), in relation to Po2. Neither method showed any significant difference between COVID-19 and non-COVID-19 groups at any given Po2 and regardless of disease severity (28% of our patients with COVID-19 required intensive care at some point), although we were able to identify a shift of the ODC in two “positive control” groups for abnormal affinity (with carboxyhemoglobin ≥8% and with sickle cell disease).

To further discuss the potential role of sex and methemoglobin pointed out by Böning et al. (1), we reanalyzed our data and found that, in the COVID-19 group, median P50 was 26.1 mmHg [25.4–27.1] in men, 26.1 mmHg [25.3 − 27.3] in women under 50 yr, and 25.7 mmHg [24.5 − 26.5] in women 50 yr or older (P = 0.21). As one would expect, our cohort only comprised 8 women aged <50 yr (extreme values: 33–48) and 22 women aged ≥50 yr (51–85). Moreover, median methemoglobin was 1.5% [1.2–1.8] in patients with COVID-19 having received hydroxychloroquine versus 1.1% [1–1.3] in the absence of hydroxychloroquine (P < 0.0001), with a maximal value of 2.5% in a critically ill 67-yr-old man. These results are in line with those recently communicated by Vogel et al. (5). However, when considering the previously reported left shift of the ODC curve (6), it may be hypothesized that it could have been overestimated by overrepresented samples from a few patients in the analyses (i.e., 3,518 samples from only 43 patients), compared with an unmatched historical cohort, or the presence of samples with a too high Po2 to accurately extrapolate P50 (7). The question about a potentially abnormal ODC in COVID-19 remains open, especially as there is conflicting data on possible impaired oxygen transport at a cellular level (8, 9).

Finally, we fully agree that the measurement of 2,3-bisphosphoglycerate concentration could be of great interest in patients with COVID-19, who can present with chronic hypoxia, respiratory alkalosis, and anemia. However, in our cohort, we found no significant correlation between hemoglobin concentration and P50 value. Nevertheless, only 24% of them displayed anemia at some point: none of them exhibited obvious hemolysis; the review of blood smears and patient files by a hemobiologist revealed that inflammation was the most common cause. Of course, these results do not preclude that some patients with COVID-19 with more severe anemia, potentially related to SARS-CoV-2 infection of erythroid progenitors (10), may harbor alteration in hemoglobin-oxygen affinity.

DISCLOSURES

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

AUTHOR CONTRIBUTIONS

T.G. drafted manuscript; T.G., L.S., E.A., J.-F.B., J.-P.R., and C.P. edited and revised manuscript; T.G., L.S., E.A., J.-F.B., J.-P.R., and C.P. approved final version of manuscript.