End-tidal carbon dioxide tension is a reliable surrogate of arterial carbon dioxide tension across different oxygen, carbon dioxide and barometric pressures

Accurate measurement of arterial carbon dioxide partial pressure (P_aCO2) is critical in emergency medicine as a marker of sufficient alveolar ventilation [1], and in human physiology to understand respiratory and cerebrovascular regulation [2]. The current gold-standard technique to measure P_aCO2 is through arterial catheterisation and analysis of blood gases, although arterialised capillary blood (e.g. earlobe or fingertip) is often used to alleviate the invasiveness and risk associated with arterial sampling [3].

Short abstract

End-tidal CO₂ tension provides an accurate estimation of P_aCO2 in healthy awake individuals over an extensive range of CO₂ pressures induced by 17 environmental conditions combining different O₂, CO₂ and barometric pressures https://bit.ly/3YuKPAY

To the Editor:

Accurate measurement of arterial carbon dioxide partial pressure (P_aCO2) is critical in emergency medicine as a marker of sufficient alveolar ventilation [1], and in human physiology to understand respiratory and cerebrovascular regulation [2]. The current gold-standard technique to measure P_aCO2 is through arterial catheterisation and analysis of blood gases, although arterialised capillary blood (e.g. earlobe or fingertip) is often used to alleviate the invasiveness and risk associated with arterial sampling [3]. In either case, direct measurement of blood gases does not allow for continuous data recording. Several pioneering works [4, 5] and recent studies [3, 6, 7] attempted to investigate noninvasive surrogates of P_aCO2 with greater temporal resolution. End-tidal carbon dioxide partial pressure (P_ETCO2) represents a noninvasive measurement of alveolar ventilation, and is typically considered an adequate substitute for P_aCO2 in healthy adults [7]. However, P_ETCO2 and P_aCO2 values may differ with discrepancies ranging between 1.8 and 4.9 mmHg in awake or anaesthetised healthy individuals [8–10]. We aimed to investigate the P_ETCO2–P_aCO2 relationship over a wide range of environmental conditions (combining different oxygen (O₂), carbon dioxide (CO₂) and barometric pressures (P_B)), to induce a large range of CO₂ pressure variations in resting healthy individuals.

17 healthy males (mean±sd age 21±2 years, body mass index 22.8±1.8 kg·m⁻²) volunteered and gave written informed consent to participate in this study. All participants were not taking any medication and were free from any cardiorespiratory and haematological diseases. Data on normal lung function and diffusion capacity of the lung for carbon monoxide of our participants are available elsewhere [11]. The experimental protocol was pre-registered at ClinicalTrials.gov (NCT04739904), approved by both the University of Ljubljana, Faculty of Sport ethics committee (8/2020–316) and the Aosta Hospital ethical committee (06/05/2021.0038781.I), and performed according to the Declaration of Helsinki.

Participants were instructed to abstain from exercise (>12 h), alcohol and caffeine (>24 h), and avoid heavy meals (>4 h) before testing. Participants were tested under the following 17 environmental conditions while comfortably seated in a quiet and thermoneutral room: 1) normobaric normoxia (NNx); 2) normobaric normoxic hypercapnia (NNx+3%CO₂); 3) hypobaric hypoxia (HHx); 4) hypobaric normoxia (HNx); 5) hypobaric normoxic hypercapnia (HNx+3%CO₂); 6) hypobaric hypoxic isocapnia (with P_ETCO2 clamped at NNx value (HHx+clamp)); 7) normobaric hypoxia (NHx); 8) normobaric hypoxic hypercapnia (NHx+3%CO₂); 9) normobaric hypoxic isocapnia (NHx+clamp); 10) normobaric hyperoxic (97%O₂) hypercapnia (3%CO₂; NHx+3%CO₂); 11) normobaric hyperoxic (94%O₂) hypercapnia (6%CO₂; NHx+6%CO₂); 12) hypobaric hyperoxic (97%O₂) hypercapnia (3%CO₂; HHx+3%CO₂); 13) hypobaric hyperoxic (94%O₂) hypercapnia (6%CO₂; HHx+6%CO₂); and 14–17) normobaric and hypobaric hypocapnia (i.e. voluntary hyperventilation). Normobaric conditions were performed near sea level (295 m; P_B=737±2 mmHg), while hypobaric measurements were carried out at high altitude (3375 m; P_B =503±3 mmHg). NNx+CO₂ and NHx+CO₂ were induced by switching the inspired gas from ambient air to 3%CO₂ (in 20.93%O₂, balance N₂). In HNx and HNx+CO₂, participants breathed supplemental O₂ (inspiratory oxygen fraction (F_IO2) 32%, with 0.03%CO₂, balance N₂ and with 3%CO₂, balance N₂, respectively) to induce the same F_IO2 as in NNx. During conditions 6 and 9, end-tidal clamping was performed using a modified version of a breathing system detailed elsewhere [12]. Briefly, the system delivered high-flow, low-resistance inspired gas with a fixed F_IO2 and a varying inspiratory carbon dioxide fraction. The inspiratory end-point of this system included an open-ended reservoir where room air was mixed with 100%CO₂ compressed gas. The 8-L custom-made reservoir was connected, via a plastic flexible tube, to a two-way nonrebreathing valve (2700 series; Hans Rudolph, Kansas City, MO, USA) attached to a low dead-space face mask (Hans Rudolph mask; 7400 oronasal series; dead space 73 mL). In the normobaric and hypobaric hyperventilation stages, participants were instructed to increase their frequency and/or depth of breathing to reduce their P_ETCO2 by the same magnitude as the increase observed during the corresponding hyperoxic hypercapnic condition. Each condition lasted 4 min. P_ETCO2 was continuously monitored by a calibrated metabolic cart (Ergocard Professional; Medisoft, Sorinnes, Belgium) and the 30-s average at the end of each stage was recorded. Arterialised capillary blood was collected from the earlobe during the last 30 s of each stage, and analysed for P_aCO2 using an arterial blood gas analyser (ABL-90 FLEX; Radiometer, Copenhagen, Denmark).

After having checked for normality by Shapiro–Wilk test, linear regression and correlation analyses between P_ETCO2 and P_aCO2 were performed by the least-squares residual method (Prism v.6.0; GraphPad Software, La Jolla, CA, USA). Residual plot analysis was used to determine the linear fitting of our data. Moreover, a linear regression of all the individual residuals versus the average CO₂ response was tested to determine that there was not a systematic difference throughout the range of values. Bland–Altman analysis calculating the difference versus the mean was used to compare paired readings of P_ETCO2 and P_aCO2; 95% confidence intervals were also calculated. All p-values were two-tailed, and statistical significance was defined a priori at p<0.05.

P_ETCO2 and P_aCO2 showed a strong to very strong correlation for each of the 17 environmental conditions, separately (r>0.60, p<0.044), as well as when all conditions were pooled (figure 1a). Bland–Altman analysis indicated that, when all conditions were pooled, the bias of the P_ETCO2 was −2.43 mmHg (95% CI −6.08–1.23 mmHg; figure 1b). Taken separately, the bias ranged from −3.99 (95% CI −7.13– −0.85 mmHg) in NNx to −0.18 (95% CI −1.84–1.47 mmHg) in NHx+3%CO₂. The linear regression analysis of the residuals showed that there was not a systematic difference throughout the range of values (slope= −0.331, p=0.798; figure 1c).

FIGURE 1.

Pooled correlation and linear regression analyses of a) end-tidal carbon dioxide tension (P_ETCO2) versus arterial carbon dioxide tension (P_aCO2) measured during 17 experimental conditions and b) Bland–Altman plot of the actual difference versus the mean of P_ETCO2 and P_aCO2; c) the residuals plot. In b) dotted lines represent the bias and the 95% confidence intervals. CO₂: carbon dioxide; O₂: oxygen.

P_ETCO2 represents an attractive, noninvasive alternative for P_aCO2 measurement. We observed a strong correlation between P_ETCO2 and P_aCO2 over a wide range of inspired CO₂ partial pressures in young healthy adults. Previous studies investigating the P_ETCO2–P_aCO2 relationship in both healthy and mechanically ventilated individuals concluded that P_ETCO2 may [8, 9] or may not [13, 14] represent a surrogate of P_aCO2 in different population and/or experimental settings. There are several conditions where P_ETCO2 does not accurately reflect P_aCO2, such as exercise, ageing and body position, as well as in patients with lung diseases [7]. Moreover, respiration and dead space undoubtedly influence both P_ETCO2 and P_aCO2, although recent work reported a moderate-to-strong correlation between P_ETCO2 and P_aCO2 across a wide range of dead space to tidal volume ratios [15]. However, in this study we only focused on understanding the influence of different environmental conditions (i.e. hypobaric versus normobaric, normocapnic versus hypercapnic, normoxia versus hypoxia) on the P_ETCO2–P_aCO2 relationship in healthy individuals, and demonstrated that the P_ETCO2–P_aCO2 relationship remains valid across numerous environmental conditions combining different levels of O₂, CO₂ and barometric pressures.

In conclusion, these novel findings suggest that P_ETCO2 measurement provides an accurate estimation of P_aCO2 in healthy awake individuals over an extensive range of CO₂ pressures induced by various environmental conditions combining different O₂, CO₂ and barometric pressures. Therefore, our results support the use of P_ETCO2 as an alternative to invasive monitoring and/or repeated arterial blood gas analyses in applied environmental physiology research. However, the present findings can be only used to draw conclusions in healthy adults, since the P_ETCO2–P_aCO2 relationship does not persist in patients with alveolar ventilation/perfusion abnormalities [7], leading to a significant underestimation of P_aCO2 [6]. Nonetheless, in healthy individuals, P_ETCO2 can be easily measured breath-by-breath or continuously, which is particularly useful in a variety of experimental and applied contexts.