Over the past several decades, the decline in ARDS mortality has been largely attributed to lung-protective mechanical ventilation strategies designed to minimize ventilator-induced lung injury (VILI). VILI primarily results from excessive lung stress and strain, manifesting as volutrauma, barotrauma, atelectrauma, and biotrauma (1, 2).

Optimizing positive end-expiratory pressure (PEEP) in ARDS mitigates atelectrauma and prevents VILI (3, 4). Understanding each patient’s unique physiology and adjusting mechanical ventilation settings using advanced monitoring tools may enhance outcomes. In 1975, Suter et al. performed the first study on optimal PEEP (OP), defining it as the level that maximized oxygen transport and respiratory system compliance (C_RS) while minimizing dead space, based on arterial oxygenation, hemodynamics, and respiratory mechanics measurements (3).

Randomized controlled trials (RCTs) have explored various methods for identifying OP, including the ARDS Network PEEP/Fraction of inspired oxygen (FiO₂) tables, compliance-based titration, and recruitment-maneuver (RM)-guided adjustments. However, the absence of consensus and the wide variation in PEEP practices across institutions complicate the interpretation of the overall efficacy of these strategies in ARDS (4).

Recent studies have introduced personalized PEEP-titration strategies for ARDS (5–10). targeting physiological variables such as driving and plateau pressures, transpulmonary-pressure (P_tp) monitoring, and bedside electrical impedance tomography (EIT) (5–10).

Hickling et al. demonstrated that decremental PEEP titration—guided by optimal C_RS—was more effective at opening ARDS lungs than incremental titration (6). This technique is mainly used following the RM. Esophageal balloon catheters guide OP settings by accounting for lung stress and strain through P_tp monitoring (8). Maintaining a direct end-expiratory transpulmonary pressure (P_{tp_ee_direct}) near 0 cm H₂O was associated with improved survival compared with targeting more positive or negative pressures (7). However, pooled mortality did not differ significantly between mechanics-based PEEP strategies and ARDS Network PEEP/FiO₂ tables in a meta-analysis (4).

EIT has recently been proposed for PEEP titration in ARDS (9, 11). This technique has shown benefits in reducing Sequential Organ Failure Assessment (SOFA) scores compared with guidance provided by the ARDS Network PEEP/FiO₂ table (11, 12). Furthermore, EIT can quantify regional overdistension and collapse (10, 13). However, the extent of lung overdistension and collapse at a given low tidal volume and optimal PEEP—when guided by different techniques—remains unclear.

We hypothesized that varying PEEP targets would result in differing degrees of lung overdistension and atelectasis. To test this, we enrolled mechanically ventilated ARDS patients.

Our primary objective was to compare OP values determined by five distinct methods:

Furthermore, we aimed to determine the degree of overdistension and collapse, including lung mechanic parameters. We compared them among five PEEP titration techniques guided by respiratory or lung mechanics, P_tp, or EIT during decremental PEEP titration in ARDS patients.

We conducted a prospective physiological study in ARDS patients admitted to the medical ICUs at Ramathibodi Hospital, Bangkok, between June 2020 and April 2021.

The main objective was to compare the OP chosen using five different methods and assess the extent of overdistention and collapse associated with each method. The OP was determined from the following five methods:

The following criteria characterized the potential risk of VILI at the optimum PEEP:

Regarding the OP levels, we identified them from five distinct techniques. Sample size was calculated a priori in G*Power for a one-way ANOVA with fixed effects (omnibus test). Assuming a large effect size (f = 1.0), α = 0.05, 80% power, and five independent groups, a total of 20 participants was required. Under these parameters, the noncentrality parameter (λ) was 20.00, the critical F-value was 3.06 (df₁ = 4; df₂ = 15), and the achieved power was 0.88.

Data are presented as mean ± standard deviation (SD) or median (interquartile range [IQR]). The Student’s t-test was used to compare two continuous variables, and one-way ANOVA (F statistic) was applied for comparisons involving more than two groups. Repeated-measures ANOVA with Greenhouse–Geisser correction analyzed variables during PEEP titration. Nonparametric data were assessed using the Kruskal–Wallis test. A p-value < 0.05 was considered statistically significant. All analyses were performed with SPSS version 22.0 (IBM, Armonk, NY, USA).

We conducted a prospective observational study in a homogeneous cohort of mechanically ventilated ARDS patients. Strengths of the study included: (1) direct comparison of multiple PEEP-titration methods; (2) evaluation of potential associations with overdistension and collapse; and (3) incorporation of diverse monitoring modalities, including esophageal pressure and EIT.

The main findings are: (1) OP differed across the five titration methods; (2) P_{tp_ee_direct} -guided PEEP (≥0 cm H₂O) was higher than PEEP determined by C_RS or C_L; (3) P_{tp_ee_direct} -guided PEEP exceeded overdistension thresholds (plateau pressure >30 cm H₂O) and produced the highest percentage of overdistension; and (4) C_RS showed a greater potential risk of lung collapse, evidenced by negative end-expiratory P_tp values and a higher EIT-derived collapse percentage.

PEEP titration in ARDS is essential for optimizing oxygenation while minimizing VILI. Common approaches include: (1) selecting the PEEP that yields the highest static C_RS (thereby minimizing driving pressure); (2) targeting a positive P_{tp_ee_direct} to account for variability in lung and chest-wall mechanics, promote alveolar recruitment, and limit end-inspiratory P_tp to avoid excessive lung stress; and (3) Titrating PEEP by visualizing lung aeration and collapse by EIT.

Krebs et al. showed that in 13 patients with moderate-to-severe ARDS, PEEP set either at maximal static C_RS or according to the ARDS Network table failed to prevent negative P_{tp_ee_direct}; in fact, under compliance-guided PEEP, 50% of patients had P_{tp_ee_direct}< 0 cm H₂O (20). Suarez-Sipmann et al. (21) monitored dynamic C_RS and the percentage of non-aerated tissue on chest CT during PEEP titration following RM in an animal model with repeated lung lavages. They found that the proportion of non-aerated tissue increased during the decremental PEEP trial, even at the PEEP level associated with the highest dynamic C_RS. According to each protocol, further analysis of individual PEEP data shows that each patient had different PEEP levels (19). These findings align with our results, which showed that OP differed significantly across the five targets. P_{tp_ee_direct} guided PEEP (≥ 0 cm H₂O) was higher than PEEP determined by the highest C_RS or C_L.

Various protocols have been proposed to set OP in ARDS patients and mitigate VILI, specifically, overdistension and lung collapse (2). We characterized the potential risk of VILI—both overdistension and collapse—across different targeted PEEP levels. Overdistension was evaluated using surrogate markers: plateau pressure > 30 cm H₂O, P_{tp_ei_derived} > 25 cm H₂O, and the EIT-derived percentage of overdistension. Surrogates for collapse (atelectrauma) included negative end-expiratory P_tp values and the EIT-derived percentage of lung collapse.

P_{tp_ee_direct} guided PEEP (≥ 0 cm H₂O) met overdistension criteria—plateau pressure > 30 cm H₂O, P_{tp_ei_derived} approaching 25 cm H₂O, and the highest EIT-derived overdistension percentage. In contrast, C_RS-guided PEEP carried a greater potential risk of collapse, evidenced by negative P_{tp_ee_direct} values and a higher EIT-measured collapse percentage. Additionally, P_{tp_ee_direct} was significantly inversely correlated with the EIT-derived collapse percentage. The underlying reasons for these findings are: (1) this approach yields the highest PEEP among the five methods; and (2) it is selected to maximize recruitment of dependent lung regions, reflected by the lowest EIT-derived collapse percentage. Figure 5 illustrates the relationship between P_{tp_ee_direct} and collapse percentage during PEEP titration; the P_{tp_ee_direct} associated with zero collapse on EIT was 5.15 cm H₂O (1.90, 6.70), a value consistent with Yoshida et al., who reported a minimum P_tp of 4.6 cm H₂O to prevent collapse by EIT assessment (18).

Costa et al. described an EIT-based method for assessing cumulative alveolar collapse and overdistension by analyzing regional pixel compliance during PEEP titration. They reported excellent agreement between recruitable collapse estimated by EIT and the increase in collapse relative to the minimum CT-determined collapse across all PEEP levels. Thus, EIT-guided PEEP appears to balance overdistension and collapse (10). In our study, the percentage of overdistension by EIT correlated significantly with peak airway pressure in PCV mode and with elastance-derived lung stress. However, because driving pressure remained constant in PCV, no correlation was observed between driving pressure and overdistension percentage.

Pavlovsky et al. compared PEEP titration strategies based on EIT, namely, center of ventilation closest to 50% and PEEP from balancing the degree of overdistention and lung collapse, and methods derived from respiratory system mechanics and P_tp monitoring. They showed that the different PEEP titration strategies led to differences in lung mechanics. The OP levels assessed by the crossing point method were higher than P_{tp_ee_direct,} which is contrary to our study. These differences may be explained by the method used to compute lung collapse and overdistension, particularly the comparison between maximal and current compliance at each pixel for a given PEEP level, and may be attributed to the application of PEEP levels during the decremental PEEP trial (from 20 to 0 cm H₂O compared with from 20 cm H₂O to 8 cm H₂O in our study) (22).

In practice, some methods yield higher PEEP levels—risking overdistension—whereas others produce lower PEEP levels with a potential for collapse. The relative harms of atelectrauma versus overdistension remain debated; however, experimental ARDS models suggest that volutrauma from excessive ventilation elicits a more pronounced inflammatory response than atelectrauma (23). On the other hand, in an RCT using a porcine model, EIT-guided strategies that minimized collapse —or explicitly balanced overdistension and collapse—were associated with lower mortality compared with approaches focused solely on preventing overdistension (24).

Grasso et al. used P_{tp_ei_derived} to adjust PEEP in severe H1N1-associated ARDS, increasing it until end-inspiratory Ptp reached 25 cm H₂O. This strategy significantly improved oxygenation and, in some patients, obviated the need for extracorporeal membrane oxygenation (25). In our study, P_{tp_ei_derived} guided titration yielded an optimal PEEP lower than that targeted by P_{tp_ee_direct}, although the difference was not statistically significant. Repeated-measures ANOVA showed that P_{tp_ei_direct} values were consistently lower than P_{tp_ei_derived} (Supplementary Figure 2). Prior studies in porcine models and human cadavers have demonstrated that P_{tp_ei_derived} closely approximates inspiratory pleural pressure in non-dependent lung regions (18). The P_{tp_ei_derived} represents the highest level of inspiratory lung stress and can be considered a clinical target for minimizing VILI. This approach is consistent with the stress–strain concept, suggesting that P_{tp_ei_derived} guided PEEP selection or tidal volume reduction may lower the risk of volutrauma. Meanwhile, maintaining a P_{tp_ee_direct} ≥ 0 cm H₂O helps prevent atelectrauma by reducing lung collapse (26).

An ideal PEEP strategy should: (1) ensure adequate gas exchange; (2) maintain lung patency by keeping P_{tp_ee_direct} ≥ 0 cm H₂O; (3) prevent overdistension by limiting P_{tp_ei_derived} to ≤ 25 cm H₂O or by selecting the EIT-derived balance point between overdistension and collapse; and (4) preserve hemodynamic stability (27).

During PEEP titration, all EIT-derived overdistension percentages were analyzed. The median overdistension (147 measurements from 21 patients) was 12% (2, 23). Patients were categorized into two groups: those with EIT-derived overdistension ≥ 12% were defined as having a higher percentage of overdistension, and those with < 12% were expressed as a lower percentage. We further analyzed the best threshold value using discriminant analysis. The threshold value of peak airway pressure that best distinguishes these two groups of patients was 30 cm H₂O, and the best threshold value of elastance-derived lung stress was 25 cm H₂O. Likewise, we subgrouped patients according to the threshold value of peak airway pressure by 30 cm H₂O and elastance-derived lung stress by 25 cm H₂O. We found a similar best threshold value of the overdistension at 12 percent. As a result, we may use 12 percent of overdistension by EIT to identify the higher percentage or the lower percentage of overdistension.

This study has limitations. The sample size was small and predominantly comprised patients with intrapulmonary ARDS. In ARDS with normal chest-wall mechanics, Ecw contributes roughly 15–20% of Ers; our median Ecw/Ers was 0.18 (0.11, 0.27), so the findings may not extrapolate to patients with elevated chest-wall elastance. All measurements were obtained in the supine position; prone positioning might yield different results.

In our study, PEEP was reduced from 20 cm H₂O to 8 cm H₂O in 2 cm H₂O decrements at one-minute intervals, which may not allow full stabilization of respiratory system mechanics and may explain the absence of hemodynamic instability. During PEEP titration, the interval at each PEEP level is commonly about 2 min, but protocols vary: some extend to 3 or even 10 min, whereas others shorten the interval to approximately 30 s (19, 28, 29). A decremental PEEP trial generally requires a shorter equilibration time than an incremental trial. In a cohort of 44 ARDS patients, when PEEP was reduced (e.g., from 15 to 10 or from 15 to 5 cm H₂O), oxygenation variables stabilized within 5 min, while respiratory system compliance declined only slowly and modestly over 60 min (29). However, experimental and clinical observations indicate that a dwell time as short as ~40 s can still yield a reasonably accurate estimate of changes in compliance, because any airway closure during a decremental PEEP trial occurs very rapidly (30, 31). Finally, this study was not designed to assess clinical outcomes, and final PEEP settings were determined by the attending physicians. Further research is warranted to evaluate whether optimized PEEP-titration strategies that reduce VILI risk translate into improved clinical outcomes.

In 21 patients with moderate–severe ARDS, OP varied significantly across five titration methods, with median values as follows: 8.0 cm H₂O (8.0, 13.9) by C_RS, 9.8 cm H₂O (8.0, 14.0) by C_L, 14.0 cm H₂O (11.9, 17.9) when targeting P_{tp_ee_direct} ≥ 0 cm H₂O, 12.0 cm H₂O (10.0, 13.9) when limiting P_{tp_ei_derived} ≤ 25 cm H₂O, and 13.01 cm H₂O (9.88, 14.78) by the EIT crossing-point method. P_{tp_ee_direct} guided PEEP produced the greatest overdistension (median plateau pressure 31.3 cm H₂O [27.7, 33.3]; EIT overdistension 17% [6, 22]), whereas C_RS-guided PEEP carried the highest collapse risk [median P_{tp_ee_direct} −2.3 cm H₂O (−4.2–0.8); EIT collapse 13% (6, 21)]. These findings underscore that OP is method-dependent and that effective titration must balance recruitment against the risks of overdistension and collapse.