Volume delivered during recruitment maneuver predicts lung stress in acute respiratory distress syndrome

Abstract

Objective

Global lung stress varies considerably with low tidal volume ventilation for acute respiratory distress syndrome (ARDS). High stress despite low tidal volumes may worsen lung injury and increase risk of death. No widely available parameter exists to assess global lung stress. We aimed to determine whether the volume delivered during a recruitment maneuver (V_RM) is inversely associated with lung stress and mortality in ARDS.

Design

Substudy of an ARDS clinical trial on esophageal pressure-guided PEEP titration.

Setting

U.S. academic medical center.

Patients

42 patients with ARDS in whom airflow, airway pressure, and esophageal pressure were recorded during the recruitment maneuver (RM).

Interventions

A single RM was performed before initiating protocol-directed ventilator management. RMs consisted of a 30-second breath hold at 40 cmH₂O airway pressure under heavy sedation or paralysis. V_RM was calculated by integrating the flow-time waveform during the maneuver. End-inspiratory stress was defined as the transpulmonary (airway minus esophageal) pressure during end-inspiratory pause of a tidal breath, and tidal stress as the transpulmonary pressure difference between end-inspiratory and end-expiratory pauses.

Measurements and Main Results

V_RM ranged between 7.4 and 34.7 mL/kg predicted body weight (PBW). Lower V_RM predicted high end-inspiratory and tidal lung stress (end-inspiratory: ß = −0.449, 95% CI −0.664 to −0.234; p < .001; tidal: ß = −0.267, 95% CI −0.423 to −0.111; p = .001). After adjusting for PaO₂/FIO₂ and either driving pressure, tidal volume, or plateau pressure and PEEP, V_RM remained independently associated with both end-inspiratory and tidal stress. In unadjusted analysis, low V_RM predicted increased risk of death (OR 0.85, 95% CI 0.72–1.00; p = .026). V_RM remained significantly associated with mortality after adjusting for study arm (OR 0.84, 95% CI 0.71–1.00; p = .022).

Conclusions

Low V_RM independently predicts high lung stress and may predict risk of death in patients with ARDS.

INTRODUCTION

In patients with acute respiratory distress syndrome (ARDS), the volume of aerated lung is substantially reduced due to alveolar edema and atelectasis (1). Heterogeneous involvement is a key pathologic feature, in which areas of well-aerated lung are adjacent to atelectatic but potentially recruitable lung as well as collapsed, non-recruitable lung (2–4). Aerated lung regions maintain normal specific compliance (compliance per volume of tissue), while in non-aerated regions specific compliance is considerably lower (5). Thus, regional distension may vary considerably with tidal ventilation (6, 7).

High tidal volumes contribute to lung injury in part through cyclic overdistension of well-aerated lung regions, which may constitute a relatively small fraction of the lung in patients with ARDS. This smaller “baby lung,” so-called for its reduced aerated volume available for ventilation, requires smaller tidal volumes than would be needed in healthy lungs to prevent regional overdistension (8, 9). Indeed, the discovery of this baby lung phenomenon in the 1980s (2–4) inspired the landmark clinical trials testing low tidal volume ventilation for ARDS that inform current standard of care (9–12).

Regional overdistension from high tidal volumes relative to baby lung size leads directly to alveolar epithelial and capillary endothelial cell junction breaks, basement membrane detachment, edema formation, and systemic inflammation (13). Due to individual differences in baby lung size, lung stress and strain, and thus risk of ventilator-induced lung injury (VILI), may vary considerably between ARDS patients receiving the same low tidal volume strategy (8, 9, 14).

To date, no routinely available clinical parameter exists to assess reliably global lung stress or baby lung size. Measurement of global lung stress requires esophageal manometry to estimate pleural pressure and calculate transpulmonary pressure (airway minus pleural pressure) (15). Global strain measurements have been performed experimentally with quantitative chest CT or helium dilution to measure lung volumes (1, 14, 16).

Recruitment maneuvers (RMs), typically performed as a continuous positive airway pressure of 30–50 cmH₂O sustained over 30–40 seconds, attempt to open atelectatic but recruitable lung units. While RMs transiently may increase aerated lung volume, improve gas exchange, and minimize lung stress during tidal ventilation, the therapeutic role for RMs is unclear (17). We instead considered RMs for their potential to measure the maximum insufflation volume achievable with clinically prudent continuous airway pressures. We reasoned that this volume increase during a recruitment maneuver (V_RM), when measured beginning from the resting lung volume at PEEP, is analogous to the inspiratory capacity of the baby lung. We hypothesized that V_RM is inversely associated with global lung stress in patients with ARDS, thus identifying patients at heightened risk of VILI and death.

METHODS

Population

A substudy of the EPVent1 ARDS clinical trial was performed. EPVent1 was a single-center randomized trial of PEEP titration guided by esophageal pressures vs. the ARDSNet PEEP-FIO₂ titration table. The trial’s primary analysis demonstrated significantly lower mortality with esophageal pressure-guided PEEP titration after adjusting for baseline illness severity (APACHE II score). Eligibility criteria and results of the primary trial are published elsewhere (18).

All trial participants with airflow, airway pressure, and esophageal pressure recorded during the protocol-directed RM were eligible for the present study. Patients were excluded if a large air leak occurred during the RM. The hospital institutional review board approved the study under the original trial protocol, for which informed consent was obtained for all participants.

Study procedures and measurements

A single RM was performed on all trial participants before initiating protocol-directed ventilator management. RMs consisted of a 30-second breath hold at 40 cmH₂O airway pressure under heavy sedation or paralysis. Airflow was measured via a Fleisch pneumotachograph placed inline between the endotracheal tube and ventilator circuit. Airway pressure was measured via a separate pressure transducer. Pleural pressure was estimated by measuring esophageal pressure via a thin-walled balloon catheter as previously described (18).

Airflow, airway pressure, and esophageal pressure were recorded continuously during the maneuver. The flow-time waveform was integrated to calculate volume (Figure 1). When the RM was initiated at end-expiration (zero flow), flow was integrated for the duration of the RM to calculate V_RM. When the RM was initiated during tidal insufflation or exhalation (non-zero flow), V_RM was calculated by integrating flow beginning from end-expiration of the preceding breath. In both cases, the preset PEEP level represents the starting airway pressure immediately prior to the RM, which had been set at the clinician’s discretion. To account for between-patient differences in healthy lung size, V_RM was scaled to PBW (9). For comparison, predicted inspiratory capacity was calculated using reference equations recommended by the American Thoracic and European Respiratory Societies (19, 20).

Figure 1.

Depiction of the calculation of V_RM. The RM entailed a 30-second breath hold at 40 cmH₂O airway pressure under heavy sedation or paralysis. The preset PEEP level represents the starting airway pressure, with associated resting end-expiratory lung volume, immediately prior to the RM. Airflow was recorded continuously during the RM via an inline Fleisch pneumotachograph. The flow-time waveform was integrated over the duration of the RM (shaded area under the curve) to calculate the insufflation volume, V_RM.

Determination of global lung stress and tidal volume-to-V_RM ratio

Stress refers to the internal forces per unit area that balance an external load. Transpulmonary pressure is the pertinent distending pressure of the lung (21). It represents global lung stress when considering the load placed on the lung by insufflation. Global lung stress was quantified in two ways, as end-inspiratory stress and tidal stress. End-inspiratory stress was defined as the transpulmonary pressure during end-inspiratory pause of a tidal breath (end-inspiratory stress = end-inspiratory transpulmonary pressure). Tidal stress was defined as the transpulmonary pressure difference between end-inspiratory and end-expiratory pauses during a tidal breath (tidal stress = change in transpulmonary pressure) (14, 15). Unlike tidal stress, end-inspiratory stress additionally accounts for the stress already present before tidal inflation—the stress on the lung from its end-expiratory volume at PEEP—which may differ substantially depending on the preset PEEP, chest wall characteristics, and ARDS severity (15).

Strain refers to the deformation of an object relative to its resting size or shape. The ideal resting conformation of the ARDS lung is unknown. Therefore, to assess relative lung deformation, we instead considered the tidal volume-to-V_RM ratio (V_T/V_RM). V_RM was chosen as the reference volume because it represents the maximum insufflation volume achievable under clinically prudent conditions (maintaining a continuous airway pressure of 40 cmH₂O during the RM) beginning from the resting lung volume at PEEP. Therefore, V_T/V_RM as defined here represents the proportion of maximum insufflation volume delivered during tidal breaths.

Statistical Analysis

To compare baseline characteristics, patients were grouped as having V_RM above or below the median V_RM and compared using t-test, Fisher’s exact test, or analysis of variance (ANOVA) as appropriate. PaO₂/FIO₂ was handled as a categorical variable following the Berlin criteria for ARDS severity (mild: 200 < PaO₂/FIO₂ ≤ 300; moderate: 100 < PaO₂/FIO₂ ≤ 200; severe: PaO₂/FIO₂ ≤ 100) (22). The association between V_RM and predicted inspiratory capacity was evaluated using a paired t-test. Pearson correlation was used to inspect the association between compliance and each of V_RM, plateau pressure, and driving pressure (airway plateau pressure minus PEEP) (23). Linear regression models were developed to test the association between lung stress and each of the following: V_RM, plateau pressure, driving pressure, and tidal volume per PBW. Simple linear regression was performed first without adjustment for covariates. Multiple linear regression then was used to determine whether V_RM added predictive value over currently used surrogates for lung stress. V_RM was entered into multivariable models with PaO₂/FIO₂ and either driving pressure, tidal volume per PBW, or plateau pressure and PEEP. Simple linear regression also was used to evaluate the association between V_T/V_RM and lung stress.

Logistic regression models were used to compare the association between V_RM and 28-day mortality. First, univariate logistic regression was performed. Study arm treatment assignment was then forced into the model for face validity. Additional multivariable logistic regression analyses were not performed to avoid overfitting data given the limited sample size and observed event rate. The likelihood ratio test was used for maximum statistical power to compare regression coefficients of nested models. For all analyses, statistical significance was determined using a two-sided p-value threshold of .05.

RESULTS

Sixty-one participants were enrolled in the clinical trial, of which 49 had airflow, airway pressure, and esophageal pressure waveforms recorded during the RM. Seven recordings exhibited evidence of air leak and were excluded. Thus, 42 participants were included in this study, of which 32 (76.2%) survived to day 28. Excluded patients had higher end-inspiratory and end-expiratory transpulmonary pressures (end-inspiratory 11 ± 6 vs. 7 ± 5 cmH₂O, p = .027; end-expiratory 0 ± 5 vs. −3 ± 5 cmH₂O, p = .046), but did not differ significantly by other baseline characteristics nor 28-day mortality (Table E1 in the online data supplement).

Recruitment maneuver characteristics

Mean V_RM was 16.6 ± 6.2 mL/kg PBW (1,097 ± 435 mL), and ranged from a minimum of 7.4 mL/kg PBW (369 mL) to maximum of 34.7 mL/kg PBW (2,160 mL). V_RM was significantly lower than predicted inspiratory capacity (mean difference 29.5 ± 8.6 mL/kg PBW; p < .001) (Figure 2) and did not correlate with predicted inspiratory capacity (r = −0.075; p = .638). Peak transpulmonary pressure during the RM was 21 ± 7 cmH₂O.

Figure 2.

V_RM and predicted inspiratory capacity. Box plots illustrate the median and interquartile range (boxes), mean (diamond), and maximum and minimum values (whiskers). V_RM was significantly lower than predicted inspiratory capacity (mean difference 29.5 ± 8.6 mL/kg PBW; p < .001). When measured beginning from resting lung volume at PEEP, V_RM is analogous to the inspiratory capacity of the ARDS “baby lung.”

Patient characteristics according to V_RM quantile

Patients with higher V_RM (> median) had significantly lower pre-RM PEEP (12 ± 4 vs. 15 ± 5 cmH₂O; p = .034), plateau pressure (28 ± 6 vs. 31 ± 5; p = .049), and end-inspiratory transpulmonary pressure (5 ± 5 vs. 9 ± 4 cmH₂O; p = .002), and higher respiratory system (41 ± 13 vs. 32 ± 7 mL/cmH₂O; p = .009) and lung compliance (65 ± 32 vs 44 ± 12 mL/cmH₂O; p = .008). Patients with higher vs. lower V_RM did not differ significantly by body mass index, APACHE II, PaO₂/FIO₂, tidal volume, driving pressure, or treatment assignment (Table 1).

TABLE 1.

Baseline Characteristics of Patients

	Overall (n=42)	Lower VRM (n=21)*	Higher VRM (n=21)*	P value
Age (years)	53 ± 20	53 ± 22	54 ± 18	.880
Female	15 (35.7%)	8 (38.1%)	7 (33.3%)	1.000
Non-white race	5 (11.9%)	1 (4.8%)	4 (19.0%)	.343
Height (cm)	173 ± 11	172 ± 12	174 ± 11	.478
Actual body weight (kg)	92.0 ± 29.7	91.2 ± 33.8	92.7 ± 25.7	.869
Predicted body weight (kg)	66.4 ± 10.8	65.2 ± 10.8	67.7 ± 10.9	.466
Body mass index (kg/m2)	30.8 ± 9.8	30.8 ± 10.6	30.7 ± 9.1	.965
Clinical Characteristics
APACHE II	27 ± 6	27 ± 6	27 ± 6	.853
Primary cause of lung injury
Pulmonary	7 (16.7%)	4 (19.0%)	3 (14.3%)
Abdominal	17 (40.5%)	8 (38.1%)	9 (42.9%)
Trauma	11 (26.2%)	6 (28.6%)	5 (23.8%)	1.000
Sepsis	5 (11.9%)	2 (9.5%)	3 (14.3%)
Other	2 (4.8%)	1 (4.8%)	1 (4.8%)
Organ failure at baseline
Cardiac	15 (35.7%)	7 (33.3%)	8 (38.1%)	1.000
Renal	22 (52.4%)	11 (52.4%)	11 (52.4%)	1.000
Hepatic	15 (35.7%)	8 (38.1%)	7 (33.3%)	1.000
Hematologic	8 (19.0%)	3 (14.3%)	5 (23.8%)	.697
Respiratory Characteristics
Randomized to Esophageal Pressure-Guided Study Arm	20 (47.6%)	10 (47.6%)	10 (47.6%)	1.000
Tidal volume (mL)	497 ± 107	469 ± 83	525 ± 122	.087
Tidal volume (mL/kg PBW)	7.6 ± 1.4	7.3 ± 1.3	7.8 ± 1.5	.274
PEEP	13 ± 5	15 ± 5	12 ± 4	.034
FIO2	70 ± 18	68 ± 17	73 ± 19	.400
PaO2/FIO2	146 ± 54	147 ± 57	145 ± 52	.891
Berlin ARDS Severity
Mild	8 (19.0%)	5 (23.8%)	3 (14.3%)
Moderate	25 (59.5%)	11 (52.4%)	14 (66.7%)	.693
Severe	9 (21.4%)	5 (23.8%)	4 (19.0%)
Plateau pressure (cmH2O)	29 ± 5	31 ± 5	28 ± 6	.049
Driving pressure (cmH2O)	16 ± 4	16 ± 4	16 ± 4	.842
End-inspiratory transpulmonary pressure (cmH2O)	7 ± 5	9 ± 4	5 ± 5	.002
End-expiratory transpulmonary pressure (cmH2O)	−3 ± 5	−2 ± 4	−4 ± 6	.095
Respiratory system compliance (mL/cmH2O)	37 ± 11	32 ± 7	41 ± 13	.009
Lung compliance (mL/cmH2O)	54 ± 26	44 ± 12	65 ± 32	.008
Chest wall compliance (mL/cmH2O)	190 ± 144	205 ± 151	174 ± 139	.493
Dead-space fraction	0.66 ± 0.10	0.66 ± 0.12	0.66 ± 0.06	.896

^*Patients grouped as having V_RM either higher or lower than the median V_RM (15.7 mL/kg PBW).

V_RM and compliance

V_RM correlated strongly with lung compliance (r = 0.572; p < .001) and moderately with respiratory system compliance (r = 0.469; p = .002). V_RM was not significantly correlated with chest wall compliance (r = −0.190; p = .233).

Plateau pressure correlated weakly with lung compliance (r = −0.359; p = .020) and moderately with respiratory system compliance (r = −0.469; p = .002) but was not associated with chest wall compliance (r = 0.047; p = .772). In contrast, driving pressure correlated neither with lung (r = −0.182; p = .247) nor chest wall compliance (r = 0.035; p = .827), though it was significantly correlated with respiratory system compliance (r = −0.393; p = .010).

V_RM and global lung stress

In univariate analysis, lower V_RM per PBW was associated with higher end-inspiratory lung stress (ß = −0.449, 95% CI −0.664 to −0.234; p < .001) (Figure 3). Higher plateau pressure also was associated with higher end-inspiratory stress in univariate analysis (ß = 0.565, 95% CI 0.325 to 0.805; p < .001). Neither tidal volume per PBW, driving pressure, nor PaO₂/FIO₂ was significantly associated with end-inspiratory stress (tidal volume ß = 0.253, 95% CI −0.880 to 1.387; p = .654; driving pressure ß = 0.394, 95% CI −0.019 to 0.807; p = .061; PaO₂/FIO₂ ANOVA p = .444).

Figure 3.

Prediction of global lung stress. Left: End-inspiratory lung stress, calculated as the transpulmonary pressure during end-inspiratory pause of a tidal breath. Right: Tidal lung stress, calculated as the transpulmonary pressure difference between end-inspiratory and end-expiratory pauses during a tidal breath. Unlike tidal stress, end-inspiratory stress additionally accounts for the stress already present before tidal inflation—the stress on the lung from its end-expiratory volume at PEEP—which may differ substantially depending on the preset PEEP, chest wall characteristics, and PaO₂/FIO₂. Regression coefficients represent unadjusted association between each predictor and lung stress.

In a multivariable model including V_RM, driving pressure, and PaO₂/FIO₂, V_RM remained significantly associated with end-inspiratory stress (ß = −0.428, 95% CI −0.657 to −0.199; p < .001), while driving pressure was not significantly associated with end-inspiratory stress. In a similar model replacing driving pressure with plateau airway pressure and PEEP as separate terms, both V_RM and plateau pressure were significantly associated with end-inspiratory stress (V_RM ß = −0.294, 95% CI −0.525 to −0.063; p = .014; plateau pressure ß = 0.385, 95% CI 0.032 to 0.738; p = .033), while PEEP was not significantly associated with end-inspiratory stress. Similarly, after replacing driving pressure with tidal volume per PBW in the model, the association between V_RM and end-inspiratory stress remained significant (ß = −0.481, 95% CI −0.708 to −0.254; p < .001) while tidal volume per PBW was not significantly associated with end-inspiratory stress.

Similar findings were observed with tidal lung stress (Figure 3). In univariate analysis, lower V_RM per PBW was associated with higher tidal stress (ß = −0.267, 95% CI −0.423 to −0.111; p = .001). Plateau pressure and driving pressure also were associated with tidal stress (plateau pressure ß = 0.337, 95% CI 0.161 to 0.513; driving pressure ß = 0.542, 95% CI 0.301 to 0.783; p < .001 for both). Neither tidal volume per PBW nor PaO₂/FIO₂ was significantly associated with tidal stress (tidal volume ß = −0.119, 95% CI −0.900 to 0.661; p = .759; PaO₂/FIO₂ ANOVA p = .409).

In a multivariable model including V_RM, driving pressure, and PaO₂/FIO₂, V_RM again remained significantly associated with tidal stress (ß = −0.202, 95% CI −0.340 to −0.064; p = .005). Driving pressure also was significantly associated with tidal stress in the multivariable model (ß = 0.495, 95% CI 0.273 to 0.717; p < .001), while PaO₂/FIO₂ did not reach statistical significance. In a similar model replacing driving pressure with plateau airway pressure and PEEP as separate terms, V_RM remained significantly associated with tidal stress (ß = −0.180, 95% CI −0.332 to −0.027; p = .023), while plateau pressure and PEEP also achieved statistical significance (plateau pressure ß = 0.518, 95% CI 0.285 to 0.752; p < .001; PEEP ß = −0.447, 95% CI −0.707 to −0.186; p = .001). In a similar model replacing driving pressure with tidal volume per PBW, V_RM again was significantly associated with tidal stress (ß = −0.260, 95% CI −0.431 to −0.089; p = .004), while tidal volume per PBW was not significantly associated with tidal stress.

V_T/V_RM and global lung stress

V_T/V_RM, which represents the degree of tidal distension relative to maximum insufflation volume achievable under clinically prudent conditions, was on average 51.5 ± 19.5 %. End-inspiratory lung stress increased linearly with increasing V_T/V_RM as expected (ß = 1.170, 95% CI 0.437 to 1.903 per 10 % increase in V_T/V_RM; p = .003) (Figure 4). To determine whether this relationship was explained by driving pressure, a second model was developed including both V_T/V_RM and driving pressure as covariates. In this multivariable model, V_T/V_RM remained significantly predictive of end-inspiratory lung stress (ß = 1.048, 95% CI 0.299 to 1.798 per 10% increase in V_T/V_RM; p = .007), while driving pressure was not independently associated with end-inspiratory lung stress (ß = 0.259, 95% CI −0.135 to 0.652 per 1 cmH₂O increase in driving pressure; p = .191).

Figure 4.

Global lung stress and V_T/V_RM. The ratio of the tidal volume (V_T) to the volume delivered during a recruitment maneuver (V_RM) defines the degree of tidal distension relative to maximum insufflation volume. V_RM was measured beginning from the resting lung volume at end-expiration. Left: End-inspiratory lung stress, calculated as the transpulmonary pressure during end-inspiratory pause of a tidal breath. Right: Tidal lung stress, calculated as the transpulmonary pressure difference between end-inspiratory and end-expiratory pauses of a tidal breath.

Tidal lung stress also increased linearly with increasing V_T/V_RM (ß = 0.743, 95% CI 0.229 to 1.256 per 10% increase in V_T/V_RM; p = .006) (Figure 4). After adding driving pressure to the model, both V_T/V_RM and driving pressure were significantly predictive of tidal stress (V_T/V_RM: ß = 0.520, 95% CI 0.069 to 0.970; p = .025; driving pressure: ß = 0.474, 95% CI 0.238 to 0.711; p < .001).

V_RM and mortality

V_RM per PBW was significantly higher among survivors compared to non-survivors (17.7 ± 6.3 vs. 13.1 ± 4.9 mL/kg PBW; p = .042). In simple logistic regression, V_RM per PBW was inversely associated with risk of death (OR 0.85, 95% CI 0.72 to 1.00 per 1 mL/kg PBW increase in V_RM; p = .026). This association remained significant and qualitatively unchanged after adjusting for study arm assignment (V_RM: OR 0.84, 95% CI 0.71 to 1.00; p = .022).

End-inspiratory stress also was significantly associated with mortality in univariate analysis (OR 1.18, 95% CI 1.00 to 1.40 per 1 cmH₂O increase; p = .036). The univariate models of V_RM and end-inspiratory stress did not differ significantly in their ability to predict mortality (AUC difference = −0.009; p = .932). When V_RM and end-inspiratory stress were entered into a model together, neither was predictive of mortality due to high collinearity, nor did mortality prediction improve significantly (AUC difference 0.016; p = .799). By contrast, tidal stress was not associated with mortality in univariate analysis (OR 1.09, 95% CI 0.89 to 1.34 per 1 cmH₂O increase; p = 0.398), and V_RM remained significantly associated with mortality and qualitatively unchanged after adjusting for tidal stress (OR 0.85, 95% CI 0.71 to 1.01; p = .039).

DISCUSSION

The present study considers RMs for their value in sizing the ARDS baby lung. The central finding of this study is that V_RM predicts both tidal and end-inspiratory lung stress—biomechanical markers of lung injury risk. V_RM was unique in our cohort, compared to alternative clinical measures, in its ability to predict independently the latter. End-inspiratory stress predicted 28-day mortality in our cohort. Similarly, V_RM predicted 28-day mortality even after accounting for treatment effect of the clinical trial’s randomly assigned study intervention. Our findings suggest the association of V_RM with mortality is explained by its prediction of end-inspiratory lung stress.

Traditionally, RMs have been performed to open atelectatic but recruitable lung units, aiming to improve oxygenation and perhaps decrease VILI (17). Human studies of RMs have found improvements in oxygenation and mechanics are variable and transient (24, 25). Adverse events, namely hypotension and desaturation, occur infrequently, but these effects too are transient (17). While animal models have suggested RMs may elicit a proinflammatory response in the lung (26), similar findings have not been replicated in humans (27). Taken together, RMs appear to be well-tolerated without contributing to serious adverse events, though their therapeutic utility is questionable.

A key rationale that spurred use of low tidal volumes for ARDS was the discovery using CT that only part of the ARDS lung is well-aerated and available for ventilation (2–4). However, use of the “baby lung” concept to individualize therapy in research and clinical practice has been limited by lack of a simple bedside measure that can be performed without specialized equipment and expertise. V_RM represents a fast and direct measurement of the maximum insufflation volume achievable under clinically prudent conditions (maintaining a continuous airway pressure of 40 cmH₂O during the RM). When measured beginning from resting lung volume at PEEP, V_RM quantifies the maximum baby lung volume available for tidal ventilation, analogous to the baby lung inspiratory capacity.

Some important study limitations are worth noting. Use of end-inspiratory transpulmonary pressure (herein termed “end-inspiratory stress”) as a surrogate for lung injury risk is controversial because it requires esophageal manometry to estimate pleural pressure (14, 15). Transpulmonary pressure is the pertinent distending pressure of the lung (21). Esophageal manometry, as an estimate of pleural pressure, has been shown previously to yield findings consistent with known respiratory physiology in healthy individuals and patients with ARDS (15, 28). Moreover, the present study found higher end-inspiratory stress was associated with increased risk of death, further suggesting our measure of transpulmonary pressure has clinical relevance.

V_RM may include some degree of hyperinflation depending on respiratory system compliance and the transpulmonary pressure achieved during the RM. However, the contribution of hyperinflation to V_RM appears to be negligible. A prior human ARDS study using CT performed during RMs at 30–50 cmH₂O airway pressure found hyperinflation accounted for only 2.9 ± 4.0 % of total lung volume (29). Moreover, the transpulmonary pressures achieved during RMs in our study were comparable to those observed at total lung capacity using the same measurement techniques in healthy individuals (30), further supporting use of V_RM as a measure of inspiratory capacity.

V_RM also may include potentially recruitable lung, i.e. lung units that are collapsed at the current set PEEP but could be recruited with higher airway pressures. If the RM transiently recruits lung that otherwise remains collapsed during tidal ventilation, V_T/V_RM may underestimate relative lung deformation, limiting use of V_T/V_RM as a surrogate for lung strain. In the present study, average PEEP was 13 ± 5 cmH₂O, which is considerably higher than in most other clinical trials of ARDS and likely reduced the proportion of potentially recruitable lung. Furthermore, V_RM correlated strongly with lung compliance, and prior CT studies have shown compliance to predict baby lung size (5, 31). This association between V_RM and lung compliance would be weakened had a large volume of collapsed lung been recruited variably during the RM. In future studies, PEEP could be set using a mechanics-based approach (11, 18, 32) prior to measuring V_RM in effort to standardize lung volumes between patients and minimize contribution of transient lung recruitment to V_RM.

PEEP immediately before the RM had been set at the clinician’s discretion, and participants with lower V_RM had slightly higher PEEP. Higher PEEP may increase end-expiratory lung volume and bias toward smaller V_RM. However, patients with lower V_RM also had worse lung and respiratory system compliance, such that higher PEEP likely was required to maintain comparable transpulmonary pressure and lung recruitment at end-expiration. Indeed, end-expiratory transpulmonary pressure did not differ according to higher vs. lower V_RM. Moreover, statistical adjustment for PEEP did not affect the association between V_RM and either end-inspiratory or tidal stress. Future studies might consider standardizing PEEP according to respiratory mechanics as above before the RM is performed.

Additionally, while low V_RM appears to predict increased risk of death, results are not definitive. V_RM was inversely associated with 28-day mortality in unadjusted analysis. V_RM remained significantly predictive of 28-day mortality after adjusting for study arm in the primary clinical trial. Additional multivariable regression models were not performed to minimize risk of overfitting data due to our limited sample size. A larger validation cohort is required to confirm an independent association between V_RM and mortality and to provide reliable effect estimates.

Finally, the implications of V_RM as a bedside measure of baby lung relative inspiratory capacity require further investigation. Based on current understanding of biomechanical mechanisms, lower V_RM should predict higher risk of VILI when ventilator settings are standardized to healthy lung size (e.g. 6 mL/kg PBW). However, the association between V_RM and VILI was not tested directly. We speculate that scaling tidal volume to each patient’s baby lung size (e.g. targeting a specified percentage of V_RM) rather than healthy lung size (mL/kg PBW) might confer additional lung protection. The observed association between V_RM and mortality supports this possibility, but validation of our findings and ultimately prospective interventional trials will be needed.

CONCLUSIONS

V_RM represents a novel, widely available measure to determine maximum insufflation volume of the ARDS “baby lung,” analogous to the inspiratory capacity relative to resting volume at PEEP. V_RM independently predicts end-inspiratory and tidal lung stress, can be used to quantify the degree of tidal distension relative to maximum insufflation volume, and may be associated inversely with increased risk of death.