Reply to Spinelli et al. and to Jha: Continued Vigorous Inspiratory Effort as a Predictor of Noninvasive Ventilation Failure

From the Authors:

We read with interest the letters by Dr. Spinelli and colleagues and by Dr. Jha commenting on our work on esophageal manometry and noninvasive ventilation (NIV) in acute de novo respiratory failure (ARF) (1). Both of them discussed the potential mechanisms behind the different behavior of lung mechanics in patients who failed NIV compared with those who succeeded.

Spinelli and colleagues pointed out that higher values of pressure support (PS) were allowed to fulfill the ventilation need (without increasing the expired Vt [Vte]) in patients who succeeded the NIV trial, whereas significantly lower PS (at comparable ventilation) in the failure group suggested that higher assistance could have produced a harmful rise in Vte.

In his letter, Jha also argued why in the failure group a persistently higher V̇e with a higher inspiratory drive but lower intrapleural pressure could have driven an increased fluid afterload with a reduced Q̇ and/or systemic oxygen delivery.

These points of discussion give us now the opportunity to further discuss the interplay between respiratory effort, lung mechanics (Vte and dynamic transpulmonary pressure [ΔPl]), respiratory drive, and the cardiopulmonary interactions.

Vte at 2 hours was higher than the cutoff limit of 9.5 ml/kg of predicted body weight (2) in both groups of patients and started diverging significantly at 12 hours, with considerable reduction in the success group. This suggests that in patients with ARF, protective ventilation is difficult to achieve soon after NIV application and that Vte alone might be an insufficient marker to identify those patients who may benefit from NIV. Moreover, in our patients, the magnitude of inspiratory effort as assessed by esophageal manometry at the time of NIV start correlated inversely with Vte/ΔPl (a surrogate measure of lung compliance) but not with the baseline Vte (1). Therefore, Vte did not reflect the intensity of the respiratory effort of our patients, introducing the concept of “baby lung assessment” during NIV that surely deserves further investigation. On the other hand, the values of ΔPl increased similarly in both groups within the first 2 hours of NIV. However, this increase was due to the elevated values of esophageal pressure (ΔPes) (with low values of PS to avoid excessive Vte) in patients who failed, whereas it was driven by a higher level of the PS set (associated with an unharmful Vte) in those who succeeded. Overall, an average ΔPl value >30 cm H₂O (as observed in our patients) could be harmful, although this is not a precise marker of the stress applied to the lung parenchyma (3). Because ΔPl seems not to represent the local lung stress and overdistension during spontaneous breathing, the underlying mechanism of patient self-inflicted lung injury would not be similar to that of ventilator-induced lung injury. In particular, patients who present higher ΔPes for a given level of ΔPl, in a heterogeneous “solid-like” injured lung, are those who produce high respiratory effort, with more negative swing in pleural pressure, injurious inflation patterns, more tidal recruitment (Pendelluft phenomenon) in the dependent lung (4), and also might reach a negative alveolar pressure, thus enhancing (inflammatory) alveolar edema (5). Therefore, ΔPes measurement and monitoring would be a more accurate physiological parameter able to discriminate patients at risk for patient self-inflicted lung injury. As a matter of fact, ΔPes correlated with radiographic change on chest X-ray at 2 hours (see online supplement of Reference 1), whereas ΔPl did not (Figure 1).

Figure 1.

Pearson’s correlation coefficient between dynamic transpulmonary pressure (after 2 h of noninvasive ventilation) and radiographic changes on chest X-ray at 24 hours. Each panel corresponds with categories of radiographic change (from left to right: relevant worsening, worsening, mild worsening, unmodified, mild improvement, improvement, and relevant improvement). ΔPl = dynamic transpulmonary pressure; NIV = noninvasive ventilation.

Finally, we agree with Dr. Jha in that the persistence of elevated V̇e in the failure group might indicate a reduced Q̇ of peripheral oxygen delivery, which in turn might increase the respiratory drive, thus giving room for V̇e as a parameter to be monitored. Notwithstanding, it seems more likely that ΔPes is a more accurate and easy parameter to monitor to predict the failure under NIV as compared with V̇e as a surrogate marker of the Q̇. This aspect appears more as an effect rather than the cause, and indeed, monitoring V̇e is clearly inaccurate under NIV, and other methods to monitor hemodynamics are less informative and not easy to set up during ARF, even when this is clinically indicated (6).

When referring to the presence of other nonmechanical determinants of the respiratory drive, we agree with the comments by Spinelli and coworkers. In patients with ARF, the inspiratory effort might be influenced by different stimuli (vagal nerve stimulation, pulmonary stretch receptor inhibition, or pulmonary C-fiber activation) acting on a neuronal-inflammatory basis (7). Because the respiratory drive may be unaffected by the unloading of respiratory muscle through the NIV application, an accurate management of nonmechanical triggers should be welcomed. We strongly agree that a multimodal strategy aimed at taming the respiratory drive is crucial.

Therefore to conclude, we do believe that esophageal manometry may allow researchers to monitor the effects of any “multimodal treatment,” regardless of the contributing causes of an increased respiratory effort.

Quoting a famous movie by Wim Wenders, further studies are needed to establish whether self-inflicted lung injury and ventilator-induced lung injury are really “faraway, so close!” (8).