The Dark Side of Spontaneous Breathing during Noninvasive Ventilation. From Hypothesis to Theory

Breathing by our own respiratory muscles is physiologically natural. The diaphragmatic contraction with spontaneous breathing (vs. muscle paralysis) tends to distribute ventilation into dorsal, well-perfused lung regions, the benefit of which was first observed in healthy subjects or anesthetized patients (1). Subsequent to these classical studies, the role of spontaneous breathing in critically ill patients has been vigorously examined (2); now it is well known that spontaneous breathing during mechanical ventilation brings various benefits to ICU patients (e.g., better gas exchange, maintenance of peripheral muscles, and diaphragm function) (2, 3). Of course, because liberation from the ventilator has been a major goal among ICU patients, all patients with this goal will need to transition to spontaneous breathing effort. Therefore, facilitation of spontaneous breathing during mechanical ventilation has a central place in the ICU.

In the 1980s, however, there were some studies (i.e., case reports and animal studies) alerting physicians to the risk of spontaneous breathing in acute respiratory failure (4, 5), concluding that “a trial of paralysis should be considered in patients with Adult Respiratory Distress Syndrome who exhibit vigorous activity of the respiratory muscles when maintenance of arterial oxygenation is a life-threatening problem” (4). About two decades later, a randomized clinical trial regarding muscle paralysis was performed and it was found that early muscle paralysis improved 90-day mortality in severe acute respiratory distress syndrome (ARDS) (6). This clinical trial indirectly supports the risk of spontaneous effort in severe ARDS. Since then, the risk of spontaneous effort and its potential mechanisms have been extensively discussed (7, 8). So far, accumulating evidence indicates that spontaneous effort during mechanical ventilation may worsen lung injury, especially when spontaneous effort is vigorous and lung injury is severe (7, 8). In 2017, this concept of effort-dependent lung injury was applied to nonintubated patients with high respiratory drive in acute hypoxemic respiratory failure, with the assumption that a process similar to effort-dependent lung injury might be involved, and the term “patient self-inflicted lung injury” (P-SILI) was coined (9). Thus, P-SILI is a new hypothesis without enough direct evidence; most of the evidence has been extrapolated from clinical and experimental data derived from mechanically ventilated subjects with spontaneous-effort (i.e., effort-dependent) lung injury (9).

In this issue of the Journal, the elegant clinical study by Tonelli and coworkers (pp. 558–567) may evolve the concept of P-SILI from a hypothesis to a theory substantiated by clinical evidence (10). Tonelli and coworkers (10) carefully estimated the intensity of spontaneous breathing effort in 30 patients with acute hypoxic de novo respiratory failure (Pa_O2/Fi_O2 ≈ 125 mm Hg) by using esophageal balloon manometry during the first 24 hours of noninvasive mechanical ventilation (NIV) and tested the hypothesis that vigorous spontaneous effort may worsen lung injury (estimated by chest X-ray), resulting in NIV failure in severe acute respiratory failure. Several intriguing findings were revealed by Tonelli and coworkers (10).

First, Tonelli and coworkers (10) found that vigorous spontaneous effort (i.e., a negative swing in esophageal pressure [∆Pes] ≈ −34 cm H₂O) was present in patients with severe acute respiratory failure before starting NIV and that reduced lung volume seems to proportionally increase the strength of spontaneous effort because the negative correlation between the intensity of spontaneous effort (estimated by ∆Pes) and lung compliance (estimated by dividing Vt by the change in transpulmonary pressure) was clearly observed (see Figure E4 of Reference 10). Of note, reduced lung volume has important effects on the force–length relationship and curvature of the diaphragm (11). Thus, the lower the lung volume is, the more force the diaphragm can generate (10, 12). This finding revealed by Tonelli and coworkers (10) may highlight the importance of an adequate amount of positive end-expiratory pressure (PEEP) to restore lung volume and thus decrease spontaneous effort in severe acute respiratory failure. In a randomized clinical study comparing the delivery of NIV with a helmet and the delivery of NIV with a face mask in patients with ARDS, NIV with a helmet could deliver higher PEEP, resulting in less spontaneous effort (suggested by a lower respiratory rate), a lower intubation rate, and better survival (13). Importantly, less spontaneous effort was observed in the helmet group (vs. the face-mask group) despite less pressure support (13).

Second, Tonelli and coworkers (10) found that persistent vigorous spontaneous effort after the induction of NIV was associated with worsening lung injury (estimated by chest X-ray in Figure E6 of Reference 10) and was the earliest and most accurate parameter to predict NIV failure (Figures 4 and E1A of Reference 10; Table 3 of Reference 10). To estimate how much spontaneous effort was involved to distend the lung, they further investigated the ratio of the ∆Pes to the change in transpulmonary pressure during NIV (Figure E1C of Reference 10). Interestingly, Tonelli and coworkers (10) found that a higher proportion of spontaneous effort in comparison with total lung-distending pressure caused a higher incidence of NIV failure, despite the similar total lung-distending pressure. These are the most striking clinical data to support P-SILI in severe acute respiratory failure. Under the same amount of total lung-distending pressure, whether generated by their own spontaneous breathing (i.e., active condition) or by mechanical ventilation (i.e., passive condition), lung injury should theoretically be the same (14), but the authors claimed this was not true. As the authors described, vigorous spontaneous effort can cause additional lung damage by local overdistension in dorsal lung regions associated with pendelluft and increased lung perfusion (8, 15).

We cannot say that the clinical study by Tonelli and coworkers (10) provides enough scientific evidence to confirm the concept of P-SILI, as this was still a physiological, exploratory study with 30 patients. However, Tonelli and coworkers’ (10) observations are very important and clinically relevant. These new clinical data will reassure us about our current clinical practice of avoiding excessive spontaneous effort and delayed intubation during NIV in acute respiratory failure. While waiting for a large, appropriately designed clinical trial, we should carefully monitor spontaneous activity (e.g., physical examination, airway occlusion pressure, and ∆Pes) in patients with acute respiratory failure under NIV. If vigorous spontaneous effort is persistent after the induction of NIV, such effort should probably be manipulated by treatment of acidosis, careful use of sedation and analgesia, adjusting the amount of PEEP, or early intubation.