This prospective clinical study was designed to investigate the effects of prone position on gas exchange and respiratory mechanics in patients with early versus persistent ARDS. We demonstrate that the effects of prone position on gas exchange depends on the timing since ARDS onset. To summarize, (1) prone position decreased V/Q mismatch in early ARDS, but increased V/Q mismatch in persistent ARDS; (2) proning increased shunt in patients with persistent ARDS, without affecting shunt in early ARDS (3) proning decreased dead space in early ARDS, but not in persistent ARDS; and finally (4) proning improved oxygenation in early ARDS, but not in persistent ARDS.

In early ARDS, lung edema, mediastinal weight, and intra-abdominal pressure may contribute to increased ventral-dorsal pleural pressure gradient, facilitating development of atelectasis and such reduce ventilation in the dependent lung regions [19]. In transitioning to prone position, the pleural pressure gradient from nondependent to dependent regions is reduced [20]. As a result, dorsal alveoli may be recruited improving ventilation. Indeed, a previous study using CT scanning in supine and prone position demonstrated dorsal lung recruitment after proning in unselected ARDS patients [5].

In persistent ARDS, prone position may decrease alveolar overdistension in the ventral region due to the decrease of chest wall compliance, and further facilitate distribution of ventilation from ventral to dorsal region. In a pilot study, Xin et al. evaluated the change in regional ventilation between supine and prone position in persistent ARDS (N = 2), and they found EIT-measured compliance did not improve in the dorsal region but worsened in the ventral region after proning, suggesting reduced lung distension in the ventral region after proning [21]. In the current study, evaluating many more patients, we found that both in early and persistent ARDS, a significant shift in tidal volume distribution from ventral to dorsal lung regions occurred after transition to prone position.

Pulmonary blood flow to different portions of the lung is regulated by several physiological mechanisms, including airspace compression of vessels, lung/heart geometry, and hypoxic pulmonary vasoconstriction [2]. On the other hand, gravity has limited impact on local pulmonary perfusion. Hence, a lack of perfusion redistribution has been demonstrated in patients with early ARDS after prone position [17, 22, 23]. This is in apparent contrast with our study, showing that the percentage of pulmonary dorsal perfusion significantly increased both in early and persistent ARDS, when turning from supine to prone position. Monitoring pulmonary perfusion in our study was however performed at 12 h after prone position initiation, which may help to explain the apparent discrepancy. Indeed, Wang et al. showed that compared to supine position, pulmonary dorsal perfusion was significantly increased after an average of 15.5 h in prone position [18]. The reduction in hypoxic vasoconstriction, less heart superimposed pressure, and less airspace compression of vessels in the dorsal region, to some extent, may contribute to the increase in dorsal perfusion after proning. Furthermore, a previous study reported marked interindividual variability in dorsal perfusion response after proning [24]. Hence, individualized dynamic lung perfusion monitoring may be warranted to assess the perfusion response of positional interventions.

Although it is well-known that oxygenation improves in hypoxemic patients in prone position, we found this is restricted to patients with early ARDS, while proning did not affect oxygenation in patients with persistent ARDS. This may be explained by the difference in effect of prone position on V/Q matching in these two groups. Previous studies showed that an improvement in the V/Q matching, rather than lung recruitment per se, explains improved oxygenation after prone position in the early ARDS [17, 18]. Consistently, our study demonstrates the increase in V/Q matching after prone position was restricted to patients with early ARDS, while V/Q distribution deteriorated in persistent ARDS following prone position. Accordingly, timing since ARDS onset appears an important factor in the effect of proning on V/Q distribution. This may be explained by reduced ability of alveolar recruitment and increased distributing of perfusion in the dorsal regions in persistent ARDS.

The observation that global shunt in patients with early ARDS was unmodified after proning is consistent with previous observations in patients with CARDS [17, 23]. Following prone position, functional stiffening of chest wall leads to a decrease in the total chest wall compliance. Consequently, ventilation in the ventral region decreased, resulting in the increase in ventral shunt in these patients. In patients with persistent ARDS, shunt significantly increased both in the ventral and dorsal regions. For this group of patients, low lung recruitability was confirmed with lower R/I ratio [14]. It is possible that prone position may merely inflate the aerated alveoli, rather than recruit nonaerated alveoli in the dorsal region due to low lung recruitability. Minimal alveolar recruitment, with increased perfusion, resulted in an increase in dorsal shunt. This finding is consistent with evolution toward a “fibrosis-like” pattern in lung pathology [9]. In our study, unmodified shunt and improved oxygenation after prone position in patients with early ARDS is consistent with previous studies [17, 23]. Of note, EIT can only identify pure shunt, which differs from “physiological shunt” as calculated by Berggren shunt equation in that it excludes the contribution from low V/Q areas [25]. Hence, we hypothesize that improved oxygenation after proning can be explained by reduction in regions with low V/Q ratio, as previously described [23].

Prone position decreased the ventral dead space since it decreased ventilation in this region which tended to cause alveolar overdistension in patients with early ARDS. For patients with persistent ARDS, dead space did not alter despite of the potential collapse of ventral region. Interestingly, PaCO2 and ventilatory ratio were significantly reduced when turning from supine to prone position in the early ARDS, but not in the persistent ARDS. This discrepancy may be explained by the marked decrease in ventral perfusion after proning in patients with persistent ARDS.

This study has some limitations: (1) Considering the effect of prone position on the chest wall elastance and lung elastance, esophageal pressure can further determine the change of lung compliance between supine position and prone position. However, these data were not obtained in our study. (2) EIT cannot provide images of the whole lung, and it only provides a validated projection of a three-dimensional distribution of ventilation and perfusion on a two-dimensional axial plane, so it is not enough for assessment of the entire spectrum of V/Q matching in the lung. (3) Due to lack of cardiac output, pixel-level V/Q ratios measured by EIT were relative. Additionally, the percentage of low V/Q units and high V/Q units was not obtained. (4) The method for EIT perfusion requires further clinical validation and we did not directly assess the effects of proning on hemodynamics. (5) Our finding regarding V/Q distribution were derived from EIT acquisition. The results should be generalized with caution due to different principles of different techniques used for monitoring V/Q distribution. (6) Patients classified as “persistent ARDS” may be comprise a rather heterogeneous of patients. This cohort may include patients with slowly worsening ARDS and patients that already met indication for prone position at an earlier time point. However, proning was not standard of care in some of the center referring patients to our center.

This study shows that the effects of prone position on oxygenation are limited in persistent ARDS, in contrast to early ARDS. In line with an original study and a recent guideline, prone position should be initiated early in the course of ARDS [1, 26]. Worth mentioning, the benefit of prone position in improved survival is not attributed to improved oxygenation [27, 28]. However, VILI was not assessed in this study. Hence, despite the limited response of proning on oxygenation and V/Q matching in persistent ARDS, the effect on mortality in these patients remains to be investigated.