“Pao2 is significantly higher in the prone position. How can this be?”
THE physiologic rationale for the prone position in mechanically ventilated subjects was articulated by A.C. Bryan almost 50 yr ago.1 Simply stated, this rationale is as follows: in the supine position, the hydrostatic pressure exerted by abdominal contents curves the dependent portion of the diaphragm cranially, resulting in loss of lung gas volume in dependent caudal regions. In the spontaneously breathing subject, this loss is partially countered by the local mechanical advantage conferred by the smaller radius of curvature of the cranially displaced dependent diaphragm. Because of Laplace’s law, this portion of the diaphragm will generate a higher transpulmonary pressure and hence attract ventilation primarily to the caudal dependent lung. In the passively ventilated subject, this mechanical advantage is absent and ventilation distributes predominantly to nondependent regions, where chest wall elastance and intraabdominal pressure are lower. Turning a subject prone would reverse these gravitational forces and favor expansion and ventilation of dorsocaudal regions. Since then, a significant body of work has supported Dr. Bryan’s intuition, and prone positioning has become a useful tool in the armamentarium of the intensivist treating acute respiratory distress syndrome and one of the few interventions with a demonstrated survival benefit in the syndrome.2
Interestingly, many studies of the physiologic effect of prone positioning have harnessed imaging techniques to measure the spatial distribution of pulmonary functional variables (e.g., ventilation, perfusion, strain). In this issue of the journal, Xin et al.3 add a valuable contribution to this work. Their experimental protocol is straightforward: five anesthetized mechanically ventilated pigs were studied at positive end-expiratory pressures (PEEPs) of 5 and 10 cm H2O in the supine and prone positions. Lung injury was then induced by intratracheal instillation of HCl. After a 60-min stabilization, the PEEP/posture protocol was repeated. End-inspiratory lung computed tomography scans and blood gases were obtained after 10 min at each PEEP/posture combination.
The imaging methodology used by Xin et al.3 is sophisticated and requires a short primer for the general reader. The first challenge one faces in this kind of analysis is that, because the lung is not a solid organ, its shape changes with posture, as does the distribution of tissue inside the lung. Consequently, to track the same piece (“voxel”) of lung tissue between positions one has to register the supine and prone computed tomography images, i.e., establish a topographical correspondence between the two images. This is achieved through a warping function that yields a measure, the Jacobian, of how much each tissue voxel is compressed or expanded in the transition from the supine to the prone position. Once image registration is complete, the second measurement one can make is that of density in the tracked voxel, which is inversely proportional to gas content. Consequently, one can tell, for each tissue voxel, whether it expands or contracts and whether its gas content increases or decreases from supine to prone. This information was then consolidated by cluster analysis to identify which voxels of lung reinflate (i.e., expand and increase their gas content), deflate (i.e., compress and decrease their gas content), or do not change with prone positioning.
Several results confirm previous findings. The prone position resulted in reinflation of dorsal regions, especially in the caudal lung. Despite some degree of deflation of ventral regions, pulmonary densities were overall less prominent in the prone position, especially after injury (fig. 3 of Xin et al.3 ). These results are consistent with the well established lesser gravitational gradient of lung expansion in the prone than supine position and with the effect of cranial diaphragm displacement in the supine position mentioned above. A deeper look reveals, however, some unusual and apparently contradictory results. For example, whole-lung gas content and mean lung density were similar in the supine and prone positions, in contrast to what was previously reported.4 Furthermore, the mass of lung tissue that experienced compression and a decrease in gas content in the prone position (i.e., “deflated”) was actually greater than that which expanded and increased its gas content (i.e., “reinflated”). Nonetheless, Pao2 was significantly higher in the prone position. How can this be?
The first finding could be explained by the fact that the pigs were lying on their belly and anterior chest when prone. Thus, the ventral chest was not free to expand, and the abdominal contents likely pushed the diaphragm cranially. In fact, compliance was not better in the prone than supine position. This is an important difference from prone positioning with unsupported free-floating abdomen, which results in marked improvement of lung gas content.4 The clinical implication of these observations is that if the goal of positioning a given patient prone is to maximize lung aeration at a given airway pressure, then the abdomen and ventral chest should be allowed unrestricted movement. The second finding is more intriguing. Interestingly, even though the mass of lung in the “deflation” ventral cluster was greater than that in the “reinflation” dorsal cluster, there was net lung recruitment in the prone position (fig. 6 of Xin et al.3 ), defined as mass of lung that surpasses an aeration threshold of ~10% (i.e., 10% of volume is gas, and 90% is tissue or edema). Consequently, the prone position was effective in restoring some degree of aeration to completely atelectatic, consolidated, or flooded alveoli, localized predominantly in the dorsocaudal region. Could this offset the effect of deflation of a greater fraction of the lung? Possibly. Answering this question with certainty would require assessment of the regional distribution of perfusion, the lack of which is a significant limitation of the study by Xin et al.3 However, we know from previous studies that: (1) Perfusion is predominantly distributed to dorsocaudal regions in the supine position. Upon turning prone, the extent to which perfusion redistributes toward ventral regions varies among subjects and species,4–6 but most likely, at least in quadrupeds, vascular conductance in dorsal regions still favors dorsal perfusion even in the prone position.7 (2) Richter et al.4 demonstrated that regional gas exchange is preserved down to a gas fraction of ~30% and only below this threshold does regional shunt increase. Consequently, it is reasonable to hypothesize that prone position–induced recruitment of alveoli in the dorsocaudal lung placed them on a more favorable portion of the gas fraction versus shunt fraction relationship.4 This, together with relatively preserved dorsal perfusion, promoted an improvement of arterial oxygenation. The deflation of the ventral cluster, instead, was probably inconsequential in terms of gas exchange as it did not reach the threshold for gas exchange to deteriorate.
These considerations also lead us to the fundamental clinical question: who is the next patient that we should turn prone? Although one must be cautious when drawing clinical conclusions from animal studies, it seems reasonable to infer that patients with hyperdensities predominantly in dorsocaudal regions should benefit the most from prone positioning, as such regions will be recruited into ventilation while still preserving reasonable perfusion. In this respect, it is worth noting that there is substantial intersubject variability in the extent of perfusion redistribution toward ventral regions in prone humans.5 This could explain why not all subjects improve their oxygenation even if their dorsal densities decrease in the prone position.8 The other advantage accrued from recruitment of dorsal regions is that the distribution of regional lung volume and ventilation becomes more homogenous throughout the lung, leading to more uniform tidal strain and hence reduced potential for ventilator-induced lung injury.9
In summary, the findings of Xin et al.3 suggest that the prone position may be beneficial even when it is not associated with an overall increase in lung gas content and that this benefit is predominantly related to the recruitment of the dorsocaudal lung.
Competing Interests
The author is not supported by, nor maintains any financial interest in, any commercial activity that may be associated with the topic of this article.