Dear editor:

We read with great interest the article by Dogani et al. on the potential effect of adding an oxygen mask –without supplemental oxygen– to a high-flow nasal cannula on the improvement of oxygenation, in patients with acute hypoxemic respiratory failure due to COVID-19 [1]. We would like to share some comments considering: (a) the improbability that the oxygenation improvement could be related to a real rise in PaO2 by the use of a mask (although neither PaO2 nor pH data were provided), and (b) the lack of a straightforward physiological mechanism for this observation. We think that the explanation for the observed increase in the arterial oxygen saturation (SaO2) reported was due to a left shift on the hemoglobin dissociation curve, solely under the operation of Bohr and Haldane principles [2, 3].

We propose that an improvement in SaO2 with a stable PaO2 came secondary to a rising in arterial pH according to the Bohr effect, or to a displacement of CO2 bound to hemoglobin (the carbamino compound) by an increase on blood oxygenation (Haldane) (Figs. 1 and 2), while keeping the arterial CO2 content relative stable. In fact, final PaCO2 remained unchanged in the eighteen patients of the study.

Fig. 1
figure 1

Hemoglobin dissociation curves according to acid–base status. Figure shows the arterial oxygen saturation (SaO2) variation (approximate values for the study by Dogani [1]) for a given PaO2 depending on arterial pH. Note: Figures drawn by Ricardo Castro

Full size image
Fig. 2
figure 2

Haldane and Bohr effects

Full size image

If we assume an increase in alveolar ventilation by a more efficient high-flow nasal cannula performance (concurring with the authors on a significant reduction of ambient air entrainment), plasma alkalinization came forth naturally: alongside with a ventilatory-driven PaCO2 decrease, protonated hemoglobin released H + which increased its affinity for O2 (the Bohr effect), increasing SaO2 consequently. On the CO2 metabolic side, this effect shifted the bicarbonate buffer equilibrium towards CO2 formation from carbonic acid, producing the release of this gas from erythrocytes that resulted in a slight rise in PCO2. Furthermore, as hemoglobin carried more oxygen, amino group-bound CO2 became displaced into its dissolved fraction (Haldane effect), all of which induced an increase in PaCO2 which would match the previous ventilatory PaCO2 decrease. As a final result, a slight or no major change in PaCO2 was observed.

Moreover, progressive hypoxemia correction may have produced an adaptive decrease on 2,3-diphosphoglycerate which left-shifted the hemoglobin dissociation curve further [4].

We acknowledge that we are presenting a theoretical model attempting to fill the gaps of an objectively documented observation, despite some lacking data. However, the combination of available data with classic physiologic principles provides another explanation for this interesting and clinically relevant phenomenon.

Authors’ response letter

  • Besarta Dogani,
  • Fredrik Månsson,
  • Fredrik Resman,
  • Hannes Hartman,
  • Johan Tham &
  • Gustav Torisson 

Dear editor,

We thank Dr. Born and Dr. Castro for the input on our research letter describing an increased SaO2 when adding a mask to high-flow nasal cannula (HFNC) in patients with severe COVID-19 [1]. In a well-reasoned argument, Dr. Born and Dr. Castro propose that the increase in SaO2 may be due to a left shift in the hemoglobin dissociation curve, i.e. an increase in SaO2 with a stable PaO2 [5]. To expand the discussion, we would like to provide data regarding PaO2 and pH, from the arterial blood gases taken at baseline and after 30 min with mask.

In all participants, PaO2 increased after 30 min, with a mean difference of 3.2 KPa (95% CI 2.2–4.2), equivalent to ~ 24 mmHg (95% CI 16–32), see Fig. 3. This result is in line with the study by Montiel et al., who applied a surgical mask on top of HFNC [6]. In their study, PaO2 increased in all participants as well, with a mean difference of 20 mmHg (95% CI 13–26). Furthermore, in our study there was a slight decrease in the average pH after 30 min, from 7.480 to 7.471, with a mean difference of − 0.009 (95% CI − 0.018 to − 0.001). Therefore, we consider it unlikely that a left shift in the hemoglobin dissociation curve would fully explain the observed phenomenon.

Fig. 3
figure 3

Partial pressure of oxygen (PaO2) in KPa from arterial blood gases taken at baseline and after 30 min with mask. error bars = 95% confidence interval of the mean. *** = p value < .001 from paired t-test

Full size image

We hypothesised that the mask could minimise entrainment of room air, especially when mouth-breathing, although we have no data to support this hypothesis. To the best of our knowledge, no studies have measured fiO2 in COVID-19 patients with HFNC. Studies in healthy volunteers suggest that fiO2 may decrease upon mouth-breathing or when exercise was used to simulate respiratory distress [7]. Whether this applies in COVID-19 is uncertain.

It was beyond the scope of our research letter to study the underlying mechanism, and several concurrent processes may contribute. Therefore, we are sincerely grateful for all input in this matter, including that of dr Swenson [8]. Hopefully, further research in oxygen delivery in severe COVID-19 will elucidate this matter.