Influence of ventilatory settings on pendelluft and expiratory muscle activity in hypoxemic patients resuming spontaneous breathing

Abstract

Background

Pendelluft and expiratory muscle activity during spontaneous breathing should be minimized to reduce potential harmful effects. This study aimed to describe pendelluft and expiratory muscle activity in hypoxemic patients recovering spontaneous breathing after ≥ 72 h of lung-protective, fully controlled mechanical ventilation (MV) and assess the effect of pressure support ventilation (PSV) and positive end-expiratory pressure (PEEP).

Methods

A physiological, randomized crossover study was conducted in hypoxemic patients receiving three levels of PSV: 5, 10, and 15 cmH₂O, and two PEEP levels: based on electrical impedance tomography before spontaneous breathing (PEEP_EIT) or according to PEEP-FiO₂ tables (PEEP_ARDS). Pendelluft was defined as the percentage of volume displaced from non-dependent to dependent lung regions during inspiration. Expiratory muscle activity was assessed by the expiratory rise in gastric pressure (ΔPga_EXP), and inspiratory effort was estimated using muscular pressure (Pmus). Statistical analyses included linear mixed-effects models and mediation analyses.

Results

Fifteen patients were enrolled (mean PaO₂/FiO₂ ratio: 262 ± 51 mmHg; median duration of MV: 9 [5–13] days; 6 females). PEEP_EIT was 11 [10–13] cmH₂O and PEEP_ARDS 6 [5–7] cmH₂O. Expiratory muscle activity was observed in 13 patients. Compared to PS 5 cmH₂O, PS to 10 and 15 cmH₂O, adjusted for PEEP, significantly reduced both pendelluft and ΔPga_EXP (p < 0.001). When adjusted for PS, PEEP_EIT was associated with a slight reduction in pendelluft (p = 0.039) but a concomitant increase in ΔPga_EXP (p = 0.007) compared to PEEP_ARDS. The mediation analysis revealed a significant negative mediating effect of ΔPga_EXP on the relationship between PEEP_EIT and pendelluft (p < 0.001). Pmus, which was also significantly associated with pendelluft magnitude (p < 0.001), mediated the effect of PS on reducing pendelluft (p = 0.048), but not that of PEEP (p = 0.46).

Conclusions

In patients with ARDS transitioning to spontaneous breathing, increasing PS reduces pendelluft and expiratory muscle activity. Higher PEEP can decrease pendelluft, but its effect can be counteracted by increased expiratory activity.

Supplementary Information

The online version contains supplementary material available at 10.1186/s13054-025-05784-y.

Background

Experimental and clinical evidence suggests that vigorous spontaneous breathing during mechanical ventilation (MV) may exacerbate lung injury, particularly in patients with acute respiratory distress syndrome (ARDS) [1, 2]. Proposed pathophysiological mechanisms include excessive transpulmonary pressures, increased pulmonary blood flow, decreased lung volume by expiratory activity, patient–ventilator asynchrony, and the phenomenon known as pendelluft [2, 3].

Pendelluft describes an inefficient and potentially harmful pattern of regional lung ventilation, characterized by intrapulmonary air redistribution from non-dependent to dependent lung regions during early inspiration, resulting in localized increases in stress and strain [2–4]. Approximately 40% of patients with ARDS exhibit pendelluft during assisted ventilation [5], and greater pendelluft magnitude has been associated with elevated levels of pro-inflammatory biomarkers, including interleukin (IL)−8, IL-18, and caspase-1 [6]. Clinically, higher pendelluft magnitude correlates with prolonged MV, longer intensive care unit (ICU) stays [5, 7], and an increased likelihood of failure during spontaneous breathing trials [8].

Expiratory muscle activity is also a common feature of spontaneous breathing and has been reported in up to 75% of mechanically ventilated patients [9]. Such activity is linked to high inspiratory load and increased end-expiratory lung volume (EELV), a response that is also present in case of diaphragmatic dysfunction [10–13]. Expiratory muscle activity has been associated with lung injury in experimental studies [14] and weaning failure in hypoxemic patients [15, 16].

We hypothesized that increasing pressure support (PS) and positive end-expiratory pressure (PEEP) could attenuate pendelluft by reducing inspiratory effort and preventing lung collapse. Given that the effects of ventilatory settings on expiratory muscle activity remain unclear, and that the interaction between pendelluft and expiratory effort has not been previously explored, this study aimed to assess whether PS and PEEP influence pendelluft and expiratory muscle activity in hypoxemic patients transitioning to spontaneous breathing after receiving ≥ 72 h of lung-protective, fully controlled MV. Additionally, the study sought to determine whether the effect of PEEP on pendelluft is modulated by expiratory muscle activity.

Methods

Study population and design

This physiological, randomized, crossover study was conducted in hypoxemic patients receiving pressure support ventilation (PSV) following the transition from fully controlled to spontaneous breathing. Eligible participants had undergone lung-protective, fully controlled MV for ≥ 72 h, including continuous neuromuscular blockade for a minimum of 48 h, during the early phase of moderate-to-severe ARDS [17]. The decision to initiate partial ventilatory support was made by the attending intensivist 24 h prior to study enrollment. During this period, patients received either assisted pressure-controlled (PC) ventilation or a pressure-controlled strategy permitting non-synchronized, unassisted spontaneous ventilation (PC-SV) [18], under moderate-to-light sedation as defined by a Richmond Agitation-Sedation Scale (RASS) score between –2 and –3.

The study received approval from the Institutional Ethics Committee (Approval No. 027/2016). Written informed consent was obtained from each participant’s legal proxy. Exclusion criteria included: age < 18 years, pregnancy, contraindications to either electrical impedance tomography (EIT) or nasogastric catheter insertion, central nervous system injury, chronic neuromuscular diseases, obstructive lung disease, intrinsic PEEP ≥ 3 cmH₂O (measured under deep sedation without inspiratory effort), recent onset of sepsis, moderate-to-severe metabolic acidosis, or clinical evidence of respiratory or hemodynamic instability.

Physiological measurements

Continuous physiological monitoring was obtained using pressure sensors and esophageal-gastric catheter (Neurovent Research Inc®, Canada) with a respiratory monitor (FluxMed, MBMED®) and EIT (Enlight 1800, Timpel®). Correct placement of the catheter was confirmed using standard validation techniques [19].

EIT-derived pendelluft was estimated as the percentage of normalized volume displaced from non-dependent to dependent lung regions during inspiration [6, 20]. Expiratory muscle activity, assessed by measuring the expiratory rise in gastric pressure (ΔPga_EXP), was defined as an increase ≥ 1 cmH₂O in Pga_EXP [9].

Other measurements included airway pressure (Paw), esophageal pressure (Pes), transpulmonary pressure (P_L), tidal volume (V_T), and flow. Dynamic transpulmonary driving pressure (ΔP_L,dyn) was calculated as the difference between Paw and Pes, and transdiaphragmatic pressure (Pdi), as the difference between Pga and Pes during inspiration (Pes_INSP). Muscular pressure (Pmus) was calculated as the difference between chest wall elastic recoil pressure (P_CW) and Pes_INSP (i.e., Pmus = P_CW – Pes_INSP) and dynamic lung compliance (C_Ldyn) as V_T divided by ΔP_L,dyn [20].

Pes_INSP reflects contraction from all inspiratory muscles. In the presence of abdominal muscle activity, the initial drop in Pes_INSP may be due to some extent to this relaxation [21]. To correct for the relaxation of active expiratory muscles when computing Pes_INSP, we applied a correction to account for inter-individual variability in the transmission from Pga to Pes [22–24]. ΔPes_INSP was calculated as: ΔPes_INSP = ΔPes − k⋅ΔPga_EXP, where k represents the ratio of ΔPes to ΔPga during expiration (k = ΔPes_EXP/ΔPga_EXP) derived from multiple breaths (Fig. S1).

ΔPga/ΔPdi ratio during inspiration, or Gilbert index [25, 26], was analyzed to estimate the relative contribution of diaphragm vs. inspiratory accessory muscles to respiratory pressure swings, excluding cases with completely negative Pdi swings or overassistance (Pmus < 5 cmH₂O and respiratory rate < 18 bpm) [27]. The change in end-expiratory lung impedance (ΔEELI) in ml, between PEEP_ARDS and PEEP_EIT, was calculated by multiplying the ΔEELI by the ratio of V_T/ΔZ [28].

Study protocol

Following a 15–20 min period of clinical stability and tolerance during PSV with a PS of 12 cmH₂O, each participant was subjected to three randomly assigned PS levels (5, 10, and 15 cmH₂O), applied for 5 min each, and two PEEP levels administered in randomized sequence. A 5-min washout period was implemented between PEEP transitions. A 5-min duration was selected because volume, mechanics and effort variables typically stabilize within a few breaths after changes in PS or PEEP. Consistent with prior physiological studies, steady-state sampling is usually performed after approximately 2–5 min [29, 30].

The two PEEP strategies were defined: (i) PEEP_ARDS, i.e. PEEP guided by oxygenation following the ARDS-Net low PEEP table at the day of the study [31], and (ii) PEEP according to EIT monitoring (PEEP_EIT) defined as the level that minimized the combined occurrence of collapse and overdistension [32, 33]. PEEP_EIT was determined the day before the protocol under passive ventilation (no spontaneous respiratory effort). On the same day, we measured quasi-static respiratory mechanics, including P_CW.

All study procedures were performed with patients in a semi-recumbent position. The head-of-bed angle was set to approximately 30–35° using the bed’s built-in inclinometer and was kept constant across all randomized settings. Because we were interested in assessing the effect of two different PEEP levels, patients with a difference between PEEP_EIT and PEEP_ARDS ≤ 2 cm H₂O were excluded.

Physiological data were analyzed over the final 2 min at each PS level, without applying washout periods between PS transitions. For each condition, sampling included 20–30 representative breaths, defined as respiratory cycles with reliable esophageal and gastric signals. Pendelluft magnitude, expiratory muscle activity, and the other measurements, were simultaneously assessed within the same respiratory cycles to facilitate intra-condition comparisons. The correction factor k, used to account for inter-individual variability in the transmission from Pga to Pes, was derived from these 20–30 breaths. No inspiratory or expiratory occlusion maneuvers were employed. A timeline Figure illustrating the transition from controlled ventilation to assisted PCV/PC–SV, leading up to the study protocol performed on PSV, is provided in the Additional file 1 (Fig. S2).

Statistical analysis

We recruited a convenience sample of 15 patients, in line with previous research in the field [34–36]. Data are expressed as mean ± standard deviation or median [interquartile range]. Repeated measures ANOVA, followed by Dunnett's post-hoc test was performed to compare variables at different levels of pressure support (PS5, PS10 and PS15 cmH₂O) with the same PEEP level, and paired t-test was used to compare variables between both levels of PEEP at each pressure support level. Linear mixed-effects models (LMM) with patients as random intercepts were performed to associate pendelluft magnitude or expiratory muscle activity with PS and PEEP (independently). Directed Acyclic Graphs were used to identify and address confounding variables in LMM (Figure S3). A complementary mediation analysis was used to assess the effect of ΔPga_EXP in the relation between PEEP_EIT and pendelluft. Analyses were performed in R Studio v2024.12.1 and GraphPad Prism v10.4.2

Results

Fifteen patients were prospectively enrolled (mean age: 56 ± 15 years; 6 females). During the early phase of ARDS, the lowest PaO₂/FiO₂ ratio was 124 ± 31 mmHg, and five patients required prone positioning. Quasi-static respiratory system compliance was 39 ± 11 mL/cmH₂O one day before the protocol. At study inclusion, the median duration of MV was 9 [5–13] days, the PaO₂/FiO₂ ratio was 262 ± 51 mmHg, V_T was 8 ± 1 mL/kg of predicted body weight and respiratory rate was 23 ± 7 breaths per minute (Additional demographic and clinical details are provided in Table 1). PEEP_ARDS was 6 [5–7] cmH₂O, while PEEP_EIT was 11 [10–13] cmH₂O. Compared with PEEP_ARDS, the application of PEEP_EIT resulted in an increase in EELV of 143 [47–342] mL.

Table 1.

Baseline clinical characteristics and respiratory variables of the study population

Variable	N = 15
General characteristics
Age	56 ± 15 years
Female	6 (40%)
BMI (kg/m2)	28 ± 6
Comorbidities
Type 2 Diabetes	7 (47%)
Arterial hypertension	4 (27%)
Chronic heart failure	2 (13%)
Chronic liver failure	4 (27%)
ARDS cause
Pulmonary	11 (73%)
Extrapulmonary	4 (27%)
Early ARDS phase
Highest SOFA	12 ± 4
Highest blood lactate (mmol/L)	2.3 [0.9–6.0]
Lowest PaO₂/FiO₂ (mmHg)	124 ± 31
Adjunctive therapy
Prolonged prone positioning	5 (33%)
Vasopressors	9 (60%)
Steroids	10 (66%)
At study inclusion
Duration of MV (days)	9 [5–13]
SOFA	8 ± 3
PaO₂/FiO₂ (mmHg)	262 ± 51
Peak pressure (cmH2O)	20 ± 5
Tidal volume (mL/kg of PBW)	8 ± 1
Respiratory rate (bpm)	23 ± 7
PEEP (cmH2O)	11 [10–13]

BMI: body mass index; ARDS: acute respiratory distress syndrome; SOFA: Sequential Organ Failure Assessment Score; PaO₂/FiO₂: ratio of arterial oxygen partial pressure to fractional inspired oxygen; CRRT: continuous renal replacement therapy; MV mechanical ventilation; PEEP: positive end-expiratory pressure

The patients’ variables obtained with PS5, PS10 and PS15 are shown in Table 2. Increasing PS was associated with higher V_T and ΔP_L,dyn, while it significantly reduced respiratory rate, Pmus, as well as ΔPga_EXP and pendelluft magnitude at a given PEEP level. In addition, increasing PS was associated with higher end-expiratory transpulmonary pressure (P_{L end-exp}), but this effect was observed only at PEEP_EIT (Table 2). Expiratory muscle activity was observed in 13 patients at PS5 but disappeared in 6 patients at PS15; nine patients met overassistance criteria at PS15 at both PEEP levels [27]. At all PS levels, no significant differences were detected between PEEP_EIT and PEEP_ARDS across the measured variables, except for respiratory rate at PS5 and P_{L end-exp}, which was consistently lower at PEEP_EIT (Table 2).

Table 2.

Physiological variables at 5, 10 and 15 cm H₂O of pressure support under individualized PEEP (PEEP_EIT) and standardized PEEP (PEEP_ARDS)

	PS 5	PS 10	PS 15	p-value
Tidal volume, ml, mean ± SD
PEEPARDS	434 ± 89	494 ± 102	572 ± 106	< 0.001
PEEPIEIT	466 ± 124	508 ± 120	577 ± 109	< 0.001
Respiratory rate, bpm, mean ± SD
PEEPARDS	23 ± 7	20 ± 6	19 ± 7	0.002
PEEPEIT	21 ± 7*	19 ± 6	17 ± 5	< 0.001
Pmus, cmH2O, mean ± SD
PEEPARDS	8.9 ± 6.3	6.7 ± 5.9	4.9 ± 4.5	< 0.001
PEEPEIT	8.8 ± 5.4	5.9 ± 4.7	4.6 ± 4.4	< 0.001
ΔPgaEXP, cmH2O, mean ± SD
PEEPARDS	2.1 ± 1.5	1.4 ± 1.2	0.7 ± 1.1	< 0.001
PEEPEIT Pendelluft magnitude (%), mean ± SD	2.8 ± 1.8	1.8 ± 1.3	1.1 ± 1.3	< 0.001
PEEPARDS	16.5 ± 15.6	12.0 ± 13.5	10.7 ± 13.1	< 0.001
PEEPEIT	14.8 ± 14.5	10.2 ± 12.6	8.2 ± 11.9	< 0.001
ΔPL,dyn, cmH2O, mean ± SD
PEEPARDS	11.6 ± 5.6	14.1 ± 5.4	17.2 ± 4.4	< 0.001
PEEPEIT	11.3 ± 4.6	13.3 ± 4.0	16.6 ± 3.7	< 0.001
PL end-exp, cmH2O, mean ± SD
PEEPARDS	− 10.2 ± 6.1	− 9.4 ± 5.9	− 8.8 ± 6.1	0.143
PEEPEIT	− 5.9 ± 5.3*	− 5.0 ± 5.0*	− 4.8 ± 4.9*	< 0.001

p-value: significance of repeated measures ANOVA

^* p < 0.05 compared with PEEP_ARDS

PEEP_ARDS: PEEP level guided by oxygenation following the ARDS-Net low PEEP table; PEEP_EIT: PEEP level guided by the lowest combination of collapse and overdistension according to electrical impedance tomography monitoring; Pmus: muscular pressure; ΔPga_EXP: expiratory rise in gastric pressure; ΔP_L,dyn: Dynamic transpulmonary driving pressure; P_{L end-exp}: end-expiratory transpulmonary pressure

Effects of settings on Pendelluft

Across all patient-condition combinations, increasing PS was associated with a significantly lower magnitude of pendelluft compared to PS5 after adjustment for PEEP (p < 0.001 for both PS10 and PS15, Fig. 1A). Similarly, application of PEEP_EIT, adjusting for PS, significantly reduced pendelluft compared to PEEP_ARDS (p = 0.039; Fig. 1B).

Fig. 1.

Effects of pressure support and PEEP on pendelluft and expiratory muscle activation. Magnitude of pendelluft (A and B) and expiratory muscle activation (C and D), quantified as the expiratory rise in gastric pressure (∆Pga_exp), across the different levels of pressure support and positive end-expiratory pressure (PEEP) evaluated in the study. Data are presented as least squares means ± standard error. PEEP_ARDS: PEEP level guided by the ARDS-Net low PEEP table; PEEP_EIT: PEEP set considering the lowest combination of collapse and overdistension according to electrical impedance tomography; PS₅: pressure support of 5 cmH₂O; PS₁₀: pressure support of 10 cmH₂O; PS₁₅: pressure support of 15 cmH₂O

Effects of settings on expiratory muscle activity

Higher PS levels—adjusted for PEEP—significantly reduced ΔPga_EXP relative to PS5 (p < 0.001, Fig. 1C). In contrast, higher PEEP (PEEP_EIT) increased ΔPga_EXP compared to PEEP_ARDS (p = 0.007), after adjusting for PS (Fig. 1D).

Interplay between Pendelluft and expiratory muscle activity

Given the opposing effects of PEEP_EIT on pendelluft and ΔPga_EXP, we conducted a mediation analysis to investigate whether expiratory muscle activity could attenuate the influence of PEEP on pendelluft. The analysis demonstrated that ΔPga_EXP negatively mediated the association between PEEP_EIT and pendelluft (Figure S4), indicating that greater expiratory gastric pressure changes weakened (and/or reversed) the relationship by which PEEP_EIT would otherwise be expected to reduce pendelluft. ΔPga_EXP remained an independent predictor of pendelluft after adjusting for PEEP, PS and C_Ldyn (p = 0.0045; Fig. 2).

Fig. 2.

Expiratory muscle activity independently predicts the magnitude of pendelluft. Scatterplot showing the association between expiratory muscle activation quantified as the expiratory rise in gastric pressure (∆Pga_EXP), and pendelluft magnitude across varying levels of pressure support (PS) and positive end-expiratory pressure (PEEP). Each data point represents an individual measurement with each color representing a different patient. The solid black line indicates the fitted regression from the linear mixed-effects model, with the shaded area representing the 95% confidence interval

Representative tracings of patients enrolled in the study showing the different effect of the different ventilatory settings on pendelluft and expiratory muscle activation are shown in Fig. 3.

Fig. 3.

Representative tracings illustrating the effect of ventilatory settings on pendelluft and expiratory muscle activation. Physiological waveforms recorded in a 77-year-old female patient with ARDS secondary to sepsis (lowest PaO₂/FiO₂ of 134), evaluated on day 10 of invasive mechanical ventilation. Tracings (top to bottom) include airway pressure, flow, gastric pressure (Pga), muscular pressure (Pmus) and regional impedance in non-dependent (Z ND) and dependent (Z D) lung regions. Inspiratory phase is shown in light gray. Compared with PEEP_ARDS and PS of 5 cmH₂O (left panel), a decrease in pendelluft and an increase in expiratory muscle activity is observed with PEEP_EIT (middle panel), with a further decrease of both phenomena with the increase of PS to 15 cmH₂O (right panel). Solid arrows indicate impedance changes in non-dependent and dependent lung regions during early inspiration. Notably, at PS5 – PEEP_ARDS, impedance decreases in the non-dependent region while increasing in the dependent region, consistent with pendelluft phenomenon

Additional determinants of Pendelluft and expiratory muscle activity

Pmus was significantly associated with pendelluft magnitude (Fig. S5); specifically, each 1 cmH₂O increase in Pmus corresponded to a 1.3% rise in pendelluft. Moreover, the reduction in Pmus observed with increasing PS was accompanied by parallel decreases in pendelluft (p = 0.02). Changes in Pmus between the two PEEP levels were also positively associated with changes in pendelluft (p < 0.001). Finally, additional analyses indicated that Pmus mediated the effects of PS on pendelluft (ACME: average causal mediation effect − 3.434; p = 0.048), but not those of PEEP (ACME – 0.196; p = 0.46).

To investigate additional mechanisms underlying the variable PEEP-associated reduction in pendelluft, we analyzed changes in C_Ldyn and pendelluft across the two PEEP levels. Greater improvements in C_Ldyn were associated with more pronounced reductions in pendelluft magnitude (Fig. S6). Notably, changes in C_Ldyn also correlated with changes in Pmus: each 1.2% increase in C_Ldyn was associated with a 1.0% reduction in Pmus (p = 0.01). In absolute values, every 4.8 ml/cmH₂O increase in C_Ldyn corresponded to a 1 cmH₂O decrease in Pmus. On the other hand, no associations were found between C_Ldyn and ΔPga_EXP.

Expiratory muscle activity emerged as a determinant of P_{L end-exp}: for every 1 cmH₂O increase in ΔPga_EXP, there was a corresponding 2 cmH₂O decrease in P_{L end-exp} (p = 0.025) suggesting a concomitant decrease in lung volume.

Because calculation of ∆Pga/∆Pdi ratio required excluding cases with negative Pdi swings or overassistance, we dichotomized the variable at the median to preserve model stability. A lower ratio, indicating greater accessory muscles recruitment, was independently associated with increased pendelluft (Fig. 4) and higher ΔPga_EXP (p = 0.001). In contrast, higher PS correlated with higher ΔPga/ΔPdi (p = 0.01), indicative of diminished inspiratory accessory muscle recruitment, whereas PEEP_EIT showed no association (p = 0.58).

Fig. 4.

Association between extradiaphragmatic inspiratory muscle activity and pendelluft magnitude. Pendelluft magnitude plotted according to the degree of extradiaphragmatic inspiratory muscle activity, assessed by the ratio of the inspiratory gastric pressure swing (∆Pga) to transdiaphragmatic pressure (∆Pdi). The median value of ∆Pga/∆Pdi of the overall sample was 0.15. A lower ∆Pga/∆Pdi ratio reflects higher extradiaphragmatic inspiratory muscles recruitment. Data are presented as least squares means ± standard error

Discussion

In ARDS patients transitioning to PSV after prolonged protective controlled MV, increasing PS was associated with a reduction in the magnitude of pendelluft. The effect of PEEP was more balanced. While higher PS levels attenuated expiratory muscle activity, higher PEEP levels tended to enhance it. Since expiratory muscle activity is linked to increased pendelluft, the beneficial effect of higher PEEP in reducing pendelluft can be counteracted by its simultaneous stimulation of expiratory activity.

Reductions in PS have been associated with increased pendelluft, and this effect was shown to be more pronounced in patients with higher initial levels of pendelluft [7]. As in a recent experimental study [37], we show that higher levels of PS decrease both inspiratory accessory muscles recruitment and expiratory muscle activity, explaining the systematic reduction in both pendelluft and expiratory muscle recruitment observed at increasing PS in the current study.

Conversely, the effect of PEEP on pendelluft and inspiratory effort appears to be highly variable and context-dependent across studies. In a randomized crossover trial by Bello et al. [36], increasing PEEP did not consistently reduce esophageal pressure swings or pendelluft. Instead, its impact was modulated by changes in respiratory system compliance. Similarly, Bassi et al. [34] reported that higher PEEP did not systematically reduce the prevalence or volume of pendelluft in patients with acute hypoxemic respiratory failure. Morais et al. [38], using experimental ARDS models, demonstrated that high PEEP could reduce pendelluft and spontaneous effort by recruiting dependent lung regions and homogenizing pleural pressure distribution. The authors suggested that PEEP may attenuate pendelluft by modulating diaphragmatic neuromuscular efficiency.

These divergent findings suggest that while PEEP has the potential to mitigate pendelluft, its effectiveness likely depends on patient-specific factors such as lung recruitability, baseline compliance, changes in respiratory system mechanics when increasing PEEP, and the intensity of spontaneous effort. A novel finding of the present study is that expiratory muscle recruitment can itself contribute to increased pendelluft, potentially counterbalancing any beneficial effects of PEEP. This mechanism may help explain previous inconsistencies in the literature. Moreover, we identified a significant negative mediating effect of expiratory muscle activity on the relationship between PEEP_EIT and pendelluft, likely reflecting interindividual heterogeneity in response to PEEP.

The transition from fully controlled to partially assisted ventilation represents a vulnerable period in patients with ARDS [39, 40]. During this phase, the diaphragm is often functionally impaired [41, 42]. As a result, when the inspiratory load exceeds the capacity of the respiratory pump, expiratory muscles are recruited in a stereotyped and hierarchical fashion [10–13]. The presence of expiratory muscle activity in 87% of our cohort likely reflects this physiological compensatory response. The recruitment of expiratory muscles may enhance inspiratory efficiency by cranially displacing the diaphragm, thereby optimizing its mechanical advantage. However, its complex physiological role contributes to reductions in both lung volume and P_L. In the context of ARDS or atelectasis, expiratory muscle activity can increase pleural pressure during expiration and also promote the occurrence of pendelluft. Elevated pleural pressure lead to negative P_L, promoting cyclic alveolar collapse, airway closure, or expiratory flow limitation, all of which may contribute to small airway and alveolar injury [13]. In our study, increases in ΔPga_EXP were associated with lower end-expiratory P_L.

Pendelluft is attributed to vigorous diaphragmatic effort in the setting of a solid-like lung behavior in ARDS [2, 3]. Our findings suggest that pendelluft can be driven by a breathing pattern involving expiratory and inspiratory accessory muscles with a non-predominant diaphragmatic activity as suggested by low values of the ΔPga/ΔPdi ratio. Expiratory muscles tend to relax immediately before the onset of inspiration, coinciding with prominent activation of inspiratory muscles [12]. This pattern may lead to early inspiratory expansion of dorsal lung regions with concurrent deflation of the ventral areas, promoting intrapulmonary gas redistribution and pendelluft.

Our observations should be interpreted with caution due to several limitations: (1) the relatively small sample size; (2) the limited duration of assessment at each PS level; (3) the fact that EIT-based PEEP titration was performed 24 h prior to study inclusion, and (4) the absence of serial total PEEP measurements during assisted ventilation.

The protocol was designed to ensure both physiological plausibility and consistency with routine clinical practice. Respiratory mechanics and pressure-derived variables adapted within a few breaths, consistent with the findings of Viale et al. [29]. In this classic study, the authors demonstrated that ventilatory responses to inspiratory unloading during PS are essentially complete within 6–8 breaths, with tidal volume, respiratory rate, and inspiratory effort reaching a new steady state rapidly and showing minimal drift thereafter [41]. Accordingly, the short stabilization periods adopted in our protocol were deemed appropriate for assessing our primary physiological outcomes—pendelluft, inspiratory effort, and expiratory muscle activity—which typically reach stable values within a few breaths after changes in ventilatory settings. In contrast, if the objective had been to evaluate gas exchange variables, substantially longer equilibration times would have been necessary, as changes in oxygenation and PaCO₂ evolve more slowly and can further influence respiratory drive and, consequently, inspiratory effort. After a PEEP reduction, PaO₂/FiO₂ generally achieve a new steady state within approximately five minutes, whereas PaCO₂ may require a longer period to stabilize. Conversely, following a PEEP increase, oxygenation tends to equilibrate more gradually and with greater variability [43].

With respect to PEEP titration, there is no consensus on the optimal method to titrate PEEP during spontaneous breathing. In this study, we used two approaches: (i) oxygenation-guided PEEP per the ARDSNet low-PEEP table on the study day, and (ii) EIT-guided PEEP defined as the level that minimizes the combined occurrence of collapse and overdistension, determined the day before during fully controlled ventilation. Returning to deep sedation—and often neuromuscular blockade—to optimize titration using EIT would represent a step backward in the progression toward spontaneous ventilation [32, 33]. Conversely, alternative strategies, such as PEEP titration based on expiratory transpulmonary pressure [44] or the combined use of EIT and esophageal pressure monitoring (45), proved difficult in our cohort due to the marked abdominal muscle activation. Although PaO₂/FiO₂ was moderately preserved at enrollment, the end-expiratory transpulmonary pressure at PEEP_EIT was negative, and no hemodynamic instability occurred during the protocol, findings that argue against excessive PEEP and lung overdistension.

Finally, it is important to emphasize that our findings are restricted to a complex subset of patients transitioning to partial ventilatory support and should not be generalized to more severely hypoxemic patients with shorter exposure to controlled ventilation.

Conclusions

In hypoxemic ARDS patients transitioning to PSV after prolonged controlled ventilation, both increased PS and PEEP may contribute to reducing pendelluft. While higher levels of PS are associated with attenuation of expiratory muscle activity, higher PEEP tends to enhance it. The reduction in pendelluft with increasing PS appears to be primarily mediated by a decrease in overall respiratory muscle activation. In contrast, the effect of PEEP on pendelluft seems to occur through improvements in dynamic lung compliance, which secondarily reduces inspiratory effort. However, this effect can be counteracted by a concomitant increase in expiratory muscle activity.

Abbreviations

MV

Mechanical ventilation

ARDS

Acute respiratory distress syndrome

IL

Interleukin

ICU

Intensive care unit

EELV

End-expiratory lung volume

PS

Pressure support

PEEP

Positive end-expiratory pressure

PSV

Pressure support ventilation

PC-SV

Pressure-controlled unassisted spontaneous ventilation

RASS

Richmond Agitation-Sedation Scale

EIT

Electrical impedance tomography

Pga

Gastric pressure

Paw

Airway pressure

Pes

Esophageal pressure

P_L

Transpulmonary pressure

V_T

Tidal ventilation

ΔP_L,dyn

Dynamic transpulmonary driving pressure

Pdi

Transdiaphragmatic pressure

Pes_INSP

Pes during inspiration

Pmus

Muscular pressure

P_CW

Chest wall elastic recoil pressure

C_Ldyn

Dynamic lung compliance

EELI

End-expiratory lung impedance

ΔZ

Tidal change in electrical impedance

LMM

Linear mixed-effects models

PaO₂/FiO₂ ratio

Ratio of arterial partial pressure of oxygen to inspired oxygen fraction

P_{L end-exp}

End-expiratory transpulmonary pressure

Author contributions

Authorship credit was based on **1.1** Substantial contributions to conception and design: RC, RB. **1.2** Acquisition of data: RC, RB, CR, DA, ML, AF, EP, VR, MB, MM, LM, JM. **1.3** Analysis and interpretation of data: RC, RB, CM, AB, JB, MA, LB, NE. **2** Drafting the article or revising it critically for important intellectual content: RC, RB, CR, CM, DA, ML, AF, EP, VR, MB, MM, LM, JM, AB, JB, MA, LB, NE. **3** Final approval of the version to be published: RC, RB, CR, CM, DA, ML, AF, EP, VR, MB, MM, LM, JM, AB, JB, MA, LB, NE.

Funding

Grant FONDECYT Nº1161510 and Nº1221829 awarded to Rodrigo Cornejo. The funding bodies had no role in the design of the study, or the collection, analysis, or interpretation of data or the manuscript preparation.

Data availability

The datasets and materials used and/or analyzed during the current study are available from the corresponding author on reasonable request.

Declarations

Ethics approval and consent to participate

The Institutional Review Board reviewed and approved the study (approval number N.027/2016, Comité Ético Científico Hospital Clínico Universidad de Chile). Informed consent was obtained from the patient’s next of kin.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.