Considerations for the use of transient tests of the peripheral chemoreflex in humans: the utility is in the question and the context

We wish to thank Dr Swenson for his thoughtful Viewpoint (Swenson, 2016) on our recently published manuscript entitled ‘Comparing and characterizing transient and steady-state tests of the peripheral chemoreflex in humans’ (Pfoh et al. 2016) in a recent issue of Experimental Physiology. We are grateful for his positive comments, as well as his thoughts about expanded utility and possible future directions using transient tests. We are looking forward to using the transient tests in a variety of applied contexts. With regard to his commentary, we wish to clarify a few points. First, as he stated, we correlated the magnitude of the hypoxic ventilatory response (HVR) following a transient hypoxia test (TT-N2) with that of a steady-state isocapnic and poikilocapnic hypoxic test (SS-ISO and SS-POI, respectively), within individuals. It is true that the TT-N2 test using peripheral pulse oximetry (peripheral arterial O2 saturation; ) as the stimulus index was similar in magnitude to the SS-ISO. However, the SS-ISO and all other comparisons were made using calculated oxygen saturation () as an index (see Fig. 5 of Pfoh et al. 2016). Using , our TT-N2 test had a much lower HVR than the SS-ISO test but similar magnitude to that of the SS-POI test (Fig. 5 of Pfoh et al. 2016). We feel that is a more appropriate index for these tests, in part for the following reasons: (i) there is a temporal delay of the peripheral pulse oximeter (e.g. Trivedi et al. 1997); (ii) the is inaccurate as a direct measure of arterial oxygen saturation () during transient hypoxia (e.g. Carter et al. 1998); and (iii) calculated using the Severinghaus transform has been validated against in hypoxic conditions (Severinghaus, 1979). In our study, underestimated the hypoxic stimulus measured by , resulting in a smaller change (Δ) from baseline, making the calculated HVR response magnitude (/Δ% oxygen saturation) larger with ∆% than with ∆% (see Fig. 5 of Pfoh et al. 2016). In light of these results, we are left wondering what the ‘real’ stimulus is with respect to a change in oxygenation using hypoxic tests. We would much rather use end-tidal , because that is a more proximal stimulus for the carotid body (Lahiri et al. 2006). However, the well-known curvilinear response (e.g. Weil et al. 1970; Weil & Zwillich, 1976) makes that approach inappropriate for a single data point (i.e. we are on the non-linear portion on the curve). As a result of the linear relationship (e.g. Rebuck & Campbell, 1974), we are left with three possibilities if we wish to calculate the HVR as a change in ventilation for a given change in oxygen saturation, as follows: (i) use (easiest; likely to be inaccurate; temporal delay; see Fig. 1A of Pfoh et al. 2016); (ii) use (invasive; expensive; accurate; no temporal resolution); or (iii) use (instantaneous calculation from end-tidal stimulus breath using Severinghaus transform). As we have shown in our study that and are different during dynamic changes in O2, our data serve as a caution to those making HVR measurements, because the calculated HVR magnitude (Δlitres per minute per Δ%) is different between and . Second, contrary to Dr Swenson's summary of our results, we found that the magnitudes of the TT-N2 test () were of similar magnitude to the SS-POI (; see Fig. 5 of Pfoh et al. 2016). However, these two tests were not well correlated within individuals (Fig. 6B of Pfoh et al. 2016). Conversely, we found that the magnitude of the TT-N2 test () was was much smaller than that of the SS-ISO (; Fig. 5 of Pfoh et al. 2016), but that these tests were moderately correlated within individual (Fig. 6A of Pfoh et al. 2016). This is important, because critics of the TT-N2 will see that the resulting response is better correlated with the SS-ISO than the SS-POI test, probably because the hypoxic stimulus in both cases is isocapnic (i.e. hypocapnia occurred after the stimulus, as a result of the HVR), even if the full magnitude of the HVR is not expressed with TT-N2 (e.g. Weil & Zwillich, 1976; Teppema & Dahan, 2010; Powell, 2012). The remaining within-individual variability between the TT-N2 and SS-ISO is likely to be attributed to the well-known cardiovascular, cerebrovascular and sympathetic responses with steady-state tests (e.g. Steinback & Poulin, 2008; Steinback et al. 2009). These many confounders aside, the utility of the standard steady-state hypoxia test as a test that aims to isolate the peripherally mediated HVR should be further called into question now that central hypoxia sensors that drive breathing have been demonstrated (e.g., Angelova et al. 2015). Third, we wish to respond to Dr Swenson's comments about the transient CO2 test (TT-CO2) of the peripheral chemoreflex (PCR) and his query about why we did not compare the CO2 responses between the TT-CO2 test and a ‘conventional hyperoxic hypercapnic test’. The transient tests target both the temporal domain and, in the case of hypoxia, the stimulus specificity of the peripheral chemoreceptors. We made direct, within-individual comparisons of the HVR responses resulting from steady-state tests and the transient test, because the steady-state tests are widely accepted as tests of the HVR (e.g. Steinback & Poulin, 2007; Teppema & Dahan, 2010). However, because both central and peripheral chemoreceptors detect increases in CO2, the stimulus specificity consideration is not useful in this context. A hyperoxic hypercapnic test is most certainly a test of the central chemoreflex, given the likely silencing effect of hyperoxia on the carotid body responsiveness to CO2 (e.g. Lahiri & DeLaney, 1975). In other words, no steady-state CO2 test of the PCR exists, so the transient CO2 test, in our view, is one of the few ways to ‘get at’ the CO2 response magnitude of the PCR in humans. This is why there are no comparisons between tests for CO2 in our study and why we focused only on characterizing the TT-CO2 test for cardiovascular and cerebrovascular responses, something that to date had not been performed. We regret that we did not make these points more explicitly in the manuscript. Dr Swenson's comments have brought to mind an additional consideration. It might have been useful for us to compare the TT-CO2 with rebreathing techniques, where investigators perform two consecutive ‘Duffin’ rebreathing tests (Duffin, 2011), one hyperoxic and one normoxic, and subtract out the normoxic PCR CO2 response for each individual (e.g. Slessarev et al. 2010). Comparing this ‘Duffin’ rebreathing technique and the TT-CO2 might be a useful addition to these characterizations. However, as with the steady-state tests, rebreathing tests have large cardiovascular and cerebrovascular effects, and performing two consecutive rebreathing tests would be more uncomfortable for the participant (both the prior hyperventilation and incremental, sustained hypercapnia) and would take longer to administer. As with all respiratory chemoreflex methodologies, the utility of each test is in the specific research question and the context. All methods have caveats, and there is no definitive respiratory chemoreflex test in humans, in our view. We do think that our comparisons and characterizations have made explicit the potential utility of transient chemoreflex tests of the PCR in humans, increasing the diversity of our collective toolbox. We wish to thank Dr Swenson again for his positive comments and for helping to raise awareness of our manuscript, as well as the journal for allowing us to respond.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.