The central insight of the review is that hypoxaemia and impaired carbon dioxide clearance in ARDS are not two versions of the same problem. Low arterial oxygen is driven predominantly by intrapulmonary shunt, meaning blood that perfuses alveoli which are no longer ventilated and therefore returns to the circulation without ever picking up oxygen. Carbon dioxide retention, by contrast, is a consequence of increased dead space ventilation, where air moves in and out of the lungs but never reaches perfused alveoli where gas exchange could occur. Both problems sit on a continuum of disturbed ventilation-perfusion relationships, but they demand different therapeutic logic, and conflating them is one of the most common interpretive errors at the bedside.

The authors begin with the alveolar gas equation, the foundational tool that sets the theoretical ceiling for arterial oxygenation. In a homogeneous lung, the alveolar oxygen pressure is determined by three variables: the composition of inspired gas, alveolar ventilation as reflected by alveolar carbon dioxide tension, and the respiratory exchange ratio, which is the ratio of carbon dioxide production to oxygen consumption. Notably, the equation shows that alveolar oxygen pressure is independent of mixed venous oxygenation, a point with major clinical consequences later in the analysis. Because carbon dioxide diffuses rapidly and almost completely across the alveolar-capillary membrane, and because the carbon dioxide dissociation curve is nearly linear, arterial carbon dioxide closely tracks its ideal alveolar counterpart, making it a reliable surrogate in the equation.

From this foundation flows the alveolar-arterial oxygen difference, normally a modest 10 to 15 millimetres of mercury. This gradient remains normal in isolated hypoventilation or when breathing low-oxygen mixtures, but widens whenever ventilation-perfusion mismatch, shunt, or diffusion limitation is present. The review cautions, however, that the gradient is not a fixed property of the lung: because alveolar oxygen pressure rises more steeply than arterial oxygen pressure as inspired oxygen increases, the difference grows with the fraction of inspired oxygen, limiting its reliability at high inspired oxygen concentrations. A patient’s gradient on 100 percent oxygen cannot be directly compared with the same patient’s gradient on 40 percent oxygen, a subtlety that matters when tracking disease progression.

The conceptual heart of the review is the classical three-compartment model introduced by Riley in 1949, which simplifies the continuous spectrum of ventilation-perfusion ratios into ideal units, shunt units with a ratio of zero, and dead space units with an effectively infinite ratio. ARDS is characterized by a pathological mixture of all three: shunt, regions of very low ventilation-perfusion ratio, and expanded dead space. Diffusion limitation, often invoked intuitively, plays a minimal role in practice, because the high inspired oxygen concentrations used in these patients and the rapid equilibration of carbon dioxide make diffusion barriers almost irrelevant in ventilated units. At a normal ratio of roughly 0.8, alveolar oxygen and carbon dioxide pressures approximate 100 and 40 millimetres of mercury respectively; as the ratio falls, alveolar gas composition drifts toward mixed venous values, and as it rises, it approaches inspired gas.

Quantifying shunt is where physiology meets haemodynamics, and this is the section with perhaps the most surprising bedside implications. The Berggren equation estimates the physiological shunt fraction from the oxygen contents of ideal end-capillary, arterial, and mixed venous blood. But arterial oxygenation depends not only on the shunt fraction; it depends equally on mixed venous oxygenation, which the Fick principle ties directly to cardiac output. For a fixed shunt fraction and metabolic rate, a fall in cardiac output lowers mixed venous oxygen content, and the desaturated venous blood passing through shunted units drags arterial oxygen down with it, even though the lung itself has not changed at all. The authors cite a striking numerical example: on 100 percent oxygen with a fixed 20 percent shunt, arterial oxygen pressure may exceed 400 millimetres of mercury at normal cardiac output but fall below 150 when output is low. PaO2, in other words, is an unreliable index of lung function whenever the circulation is unstable.

The feedback loop runs in both directions. The shunt fraction itself is not fixed, because a rise in cardiac output raises venous oxygen tension, which attenuates hypoxic pulmonary vasoconstriction, the reflex that redirects blood away from poorly ventilated lung regions. Higher cardiac output can therefore simultaneously improve mixed venous oxygenation and worsen venous admixture by redistributing perfusion toward poorly ventilated units. The net effect on arterial oxygen depends on which mechanism dominates in a given patient, which explains why interventions that boost cardiac output sometimes improve oxygenation and sometimes worsen it, and why the review warns against reading oxygen numbers without simultaneously reading the circulation.

These caveats strike directly at the most widely used severity metric in critical care: the PaO2/FiO2 ratio, which underpins the Berlin definition of ARDS. The ratio is intended as a surrogate for shunt fraction, but it is sensitive to inspired oxygen concentration, positive end-expiratory pressure, and cardiac output. Because of the non-linear interaction between inspired oxygen, haemoglobin saturation, and venous admixture, simply changing the FiO2 can alter the ratio without any change in the underlying shunt. The same patient can be classified into different severity categories under different ventilator settings, and the review notes that prognostic accuracy improves when the ratio is measured under standardized PEEP and FiO2 conditions.

On the carbon dioxide side, the review walks through the family of dead space measurements, each capturing a different physiological slice. The Bohr equation, computed with mean alveolar carbon dioxide from volumetric capnography, isolates true dead space, comprising the conducting airways and high ventilation-perfusion units, and is unaffected by shunt. The Enghoff modification, which substitutes arterial carbon dioxide, reflects global gas exchange impairment rather than true dead space alone, and a high shunt fraction can inflate the Enghoff value even in the absence of genuine alveolar dead space. The Fowler method measures anatomical dead space by single-breath washout, and because this volume is relatively constant, tidal volume strongly influences the dead space fraction: the same 150 millilitres of anatomical dead space represents a fraction of 0.25 at a tidal volume of 600 millilitres but 0.50 at 300 millilitres, which is why the authors recommend reporting dead space in absolute millilitres per breath as well as a fraction.

The clinical payoff is prognostic. The Enghoff dead space fraction independently predicts mortality in ARDS, reflecting the degree of functional lung impairment and ventilation-perfusion heterogeneity, and the end-tidal to arterial carbon dioxide ratio offers a bedside surrogate that integrates dead space, shunt, and ventilation-perfusion mismatch, falling progressively as disease severity increases. For units without capnography, the ventilatory ratio, calculated from measured minute ventilation and arterial carbon dioxide against predicted values, requires no expired gas analysis and also independently predicts mortality. The authors close with a message that is both a warning and an invitation: shunt is the principal determinant of hypoxaemia, dead space reflects structural heterogeneity and carries independent prognostic weight, and every bedside index, from PaO2/FiO2 to end-tidal carbon dioxide, must be interpreted through the physiology it actually measures rather than the number it displays.

Subject of Research: Physiological principles of gas exchange and bedside indices in acute respiratory distress syndrome

Article Title: Principles of gas exchange in ARDS: a bedside physiological approach

Article References: Collino, F., Camporota, L., & Pinsky, M. R. (2026). Principles of gas exchange in ARDS: a bedside physiological approach. Intensive Care Medicine. https://doi.org/10.1007/s00134-026-08601-4

Image Credits: AI Generated

DOI: 10.1007/s00134-026-08601-4

Keywords: ARDS, gas exchange, intrapulmonary shunt, dead space, ventilation-perfusion mismatch, alveolar gas equation, PaO2/FiO2 ratio, cardiac output, hypoxic pulmonary vasoconstriction, volumetric capnography, ventilatory ratio, critical care