Cookies on this website

We use cookies to ensure that we give you the best experience on our website. If you click 'Accept all cookies' we'll assume that you are happy to receive all cookies and you won't see this message again. If you click 'Reject all non-essential cookies' only necessary cookies providing core functionality such as security, network management, and accessibility will be enabled. Click 'Find out more' for information on how to change your cookie settings.

Abstract Background Within-breath oscillations in arterial oxygen tension (PaO2) can be detected using fast responding intra-arterial oxygen sensors in animal models. These PaO2 signals, which rise in inspiration and fall in expiration, may represent cyclical recruitment/derecruitment and, therefore, a potential clinical monitor to allow titration of ventilator settings in lung injury. However, in hypovolaemia models, these oscillations have the potential to become inverted, such that they decline, rather than rise, in inspiration. This inversion suggests multiple aetiologies may underlie these oscillations. A correct interpretation of the various PaO2 oscillation morphologies is essential to translate this signal into a monitoring tool for clinical practice. We present a pilot study to demonstrate the feasibility of a new analysis method to identify these morphologies. Methods Seven domestic pigs (average weight 31.1 kg) were studied under general anaesthesia with muscle relaxation and mechanical ventilation. Three underwent saline-lavage lung injury and four were uninjured. Variations in PEEP, tidal volume and presence/absence of lung injury were used to induce different morphologies of PaO2 oscillation. Functional principal component analysis and k-means clustering were employed to separate PaO2 oscillations into distinct morphologies, and the cardiorespiratory physiology associated with these PaO2 morphologies was compared. Results PaO2 oscillations from 73 ventilatory conditions were included. Five functional principal components were sufficient to explain ≥ 95% of the variance of the recorded PaO2 signals. From these, five unique morphologies of PaO2 oscillation were identified, ranging from those which increased in inspiration and decreased in expiration, through to those which decreased in inspiration and increased in expiration. This progression was associated with the estimates of the first functional principal component (P < 0.001, R2 = 0.88). Intermediate morphologies demonstrated waveforms with two peaks and troughs per breath. The progression towards inverted oscillations was associated with increased pulse pressure variation (P = 0.03). Conclusions Functional principal component analysis and k-means clustering are appropriate to identify unique morphologies of PaO2 waveform associated with distinct cardiorespiratory physiology. We demonstrated novel intermediate morphologies of PaO2 waveform, which may represent a development of zone 2 physiologies within the lung. Future studies of PaO2 oscillations and modelling should aim to understand the aetiologies of these morphologies.

Original publication




Journal article


Intensive Care Medicine Experimental


Springer Science and Business Media LLC

Publication Date