Mortality in patients with cardiogenic shock (CS) remains high, particularly when complicated by respiratory failure [1, 2]. Strategies for treating patients with CS derives from research conducted on critically ill individuals and those affected by acute respiratory distress syndrome (ARDS) [2, 3]. Lung compliance, the elasticity or stretchability of the lungs, is a known ARDS risk factor associated with poorer outcome [4].

In patients with cardiovascular issues, such as those suffering from heart failure, a decrease in lung compliance occurs regardless of left ventricular failure classification, as defined by the New York Heart Association (NYHA) class or pulmonary vascular pressure [5, 6, 7]. Additionally, patients with cardiopulmonary edema often present with reduced lung compliance [8]. Previous studies, however, focused only on spontaneously breathing individuals, and estimating compliance through measurements of total lung capacity and esophageal pressure [5, 6, 7, 8].

A key hinderance in the field is that limited data exists on lung compliance in CS patients who require invasive mechanical ventilation. A sub analysis of the LUNG SAFE study, which included patients with isolated cardiopulmonary edema, demonstrated that higher peak-, plateau- and driving-pressure were associated with increased hospital mortality, correlating with decreased lung compliance [9]. Thus, lung compliance may serve as a critical prognostic marker for adverse outcomes in ventilated cardiac patients. Furthermore, several studies have shown that higher lung compliance in patients with cardiac arrest correlates with improved short- and long-term survival, as well as better neurological outcome [10, 11, 12]. The present study aims to further elucidate the prognostic significance of lung compliance in patients with CS undergoing invasive mechanical ventilation.

Table 1 outlines the baseline characteristics stratified by quartiles of lung compliance. The groups showed comparable baseline characteristics considering age, body mass index and prior medical history on admission. Patients in the lowest lung compliance quartile demonstrated a higher prevalence of smoking compared to those in higher quartiles (54.3% vs. 36.1% vs. 31.4% vs. 20.0%; p = 0.025). Additionally, these patients exhibited lower systolic blood pressures, escalating progressively across the quartiles to 102 mmHg, 109 mmHg, and 117 mmHg, respectively (p = 0.01). Despite the prevalence of lower systolic blood pressures, most patients maintained pressures above 90 mmHg, facilitated by the common use of vasopressors and dobutamine on admission. Patients in the lower quartiles were less often male compared to those in the higher quartiles (51.4% vs. 41.7% vs. 77.1% vs. 80.0%; p = 0.001). Furthermore, patients in the higher lung compliance quartiles had a lower prevalence of chronic heart failure (37.1% vs. 38.9% vs. 25.7% vs. 11.4%; p = 0.039) and were less often treated with diuretics (48.6% vs. 50.0% vs. 37.1% vs. 14.3%; p = 0.036).

As presented in Table 2, causes of CS and the distribution within the SCAI shock classification were consistent across all groups. Predominantly, patients in this study presented with advanced stages of CS, which was reflected by the significant need for cardiopulmonary resuscitation in 82.3% of cases. This high incidence of SCAI shock stage E was primarily attributed to the frequency of cardiac arrest at admission rather than their hemodynamic condition alone. Regarding lung compliance, the group with values greater than 39.7 mL/cmH₂O exhibited the lowest prevalence of left ventricular ejection fraction 30% (60.0% vs. 63.9% vs. 60.0% vs. 25.7%; p = 0.001) and the smallest average diameter of the inferior vena cava (1.8 cm vs. 2.1 cm vs. 2.0 cm vs. 1.7 cm; p = 0.024). Conversely, patients in the lower quartiles of lung compliance had reduced incidences of shockable rhythms (31.4% vs. 27.8% vs. 60.0% vs. 51.4%; p = 0.032) and higher occurrences of non-shockable rhythms (54.3% vs. 47.2% vs. 28.6% vs. 25.7%; p = 0.032) compared to those in the higher quartiles.

Patients in the highest quartile for lung compliance demonstrated the least reliance on extracorporeal life support; notably, none of these patients required veno-arterial extracorporeal membrane oxygenation (ECMO) (17.1% vs. 13.9% vs. 28.6% vs. 0.0%; p = 0.009). Regarding respiratory parameters, patients with lung compliance below 23.9 mL/cmH₂O exhibited significantly altered ventilatory settings. These patients had elevated PaCO₂ (52 mmHg vs. 45 mmHg vs. 43 mmHg vs. 43 mmHg; p = 0.045) and were ventilated with higher PEEP (8 cmH₂O vs. 9 cmH₂O vs. 7 cmH₂O vs. 7 cmH₂O; p = 0.003), peak inspiratory pressure (31 cmH₂O vs. 25 cmH₂O vs. 20 cmH₂O vs. 17 cmH₂O; p = 0.001) and dynamic driving pressure (19 cmH₂O vs. 16 cmH₂O vs. 13 cmH₂O vs. 11 cmH₂O; p = 0.001). Conversely, the tidal volumes in this group were lower (371 mL vs. 451 mL vs. 458 mL vs. 580 mL; p = 0.001) compared to the other quartiles. Laboratory parameters showed an even distribution between the two groups, with the exception of pH levels, which varied significantly (7.21 vs. 7.25 vs. 7.27 vs. 7.33; p = 0.024) indicating a trend towards worsening acidosis with decreasing lung compliance.

Patients with lower lung compliance, when stratified by quartiles, had a greater 30-day risk of all-cause mortality compared to patients with higher compliance (log rank p = 0.018), as shown in Fig. 2. Specifically, the mortality rates were 77.1% for patients with lung compliance less than 23.9 mL/cmH₂O, 66.7% for those with lung compliance between 23.9–30.4 mL/cmH₂O, 48.6% for those between 30.5–39.7 mL/cmH₂O, and 51.4% for those with lung compliance above 39.7 mL/cmH₂O. After stratifying by the median, patients with lung compliance values below 30.4 mL/cmH₂O demonstrated a higher mortality rate after 30 days compared to patients with lung compliance at or above 30.4 mL/cmH₂O (50% vs. 71.8%; log rank p = 0.007). These findings remained consistent in a subset analysis that included only CS-patients with cardiac arrest, both when stratified by quartiles (80% vs. 74% vs. 53% vs. 59%; log-rank p = 0.037) and by the median (77.2% vs. 55.9%; log-rank p = 0.008), as shown in Fig. 3.

As detailed in Table 3, univariable Cox regression analysis demonstrated that lung compliance below 30.4 mL/cmH₂O was significantly associated with the primary endpoint of 30-day all-cause mortality in the entire cohort (HR = 1.760; 95% CI = 1.143–2.708; p = 0.032), and in the subgroup of CS-patients with cardiac arrest (HR = 1.783; 95% CI = 1.134–2.804; p = 0.012). This association persisted even after multivariable adjustment, indicating that lower lung compliance under 30.4 mL/cmH₂O remained a significant predictor of all-cause mortality in the entire cohort (HR = 1.698; 95% CI = 1.085–2.659; p = 0.021).

Additionally, the multivariable Cox regression model revealed other significant predictors of the primary endpoint at 30 days. Lactate levels (HR = 1.092; 95% CI = 1.045–1.141; p = 0.001), acute physiology score (HR = 1.051; 95% CI = 1.001–1.104; p = 0.048), and the presence of pneumonia (HR = 1.772; 95% CI = 1.123–2.797; p = 0.014) were associated with increased mortality. However, in the subgroup analysis including only CS-patients with cardiac arrest, the association between lung compliance below 30.4 mL/cmH₂O and all-cause mortality did not reach statistical significance (HR = 1.523; 95% CI = 0.952–2.438; p = 0.080). In contrast, lactate levels (HR = 1.092; 95% CI = 1.043–1.144; p = 0.001), acute physiology score (HR = 1.058; 95% CI = 1.002–1.118; p = 0.042) and existence of pneumonia (HR = 1.824; 95% CI = 1.125–2.959; p = 0.015) remained significant indicators of mortality risk in this specific patient group.

Analysis of the secondary endpoint as presented in Fig. 4, revealed a significant correlation between lung compliance and ventilator-free days. Patients with lower lung compliance experienced fewer ventilator-free days (p = 0.007). This trend was robust and remained statistically significant even when patients were stratified by the median lung compliance value of 30.4 mL/cmH₂O (p = 0.003).

The correlation between lung compliance and laboratory and clinical data is shown in Table 4. Notably, there were moderate inverse correlations with sex (r = –0.222; p = 0.001), acute physiology score (r = –0.144; p = 0.014), PaCO₂ (r = –0.159; p = 0.006). This inverse correlation also held true for mechanical ventilation settings including PEEP (r = –0.195; p = 0.002) and peak inspiratory pressure (r = –0.583; p = 0.001).

The primary objective of this study was to explore the prognostic influence of lung compliance at the time of ICU admission in patients with CS who required invasive mechanical ventilation. Our findings indicate that lower lung compliance is significantly associated with increased 30-day all-cause mortality. After stratifying by the median, we determined that patients with lung compliance below the median value of 30.4 mL/cmH₂O experienced higher mortality rates than those with higher lung compliance. These findings persisted even after adjusting for multiple variables. Furthermore, patients with low lung compliance also had fewer ventilator-free days, suggesting a prolonged need for mechanical ventilation. These observations underscore the potential of lung compliance, a readily measurable parameter in ventilated CS patients, as a valuable tool for risk stratification.

Lung compliance is a critical parameter in the management of patients with ARDS and serves both diagnostic and prognostic purposes [4, 20, 21]. While it is well-established in ARDS, its implications in heart failure have predominantly been explored in spontaneously breathing patients. Studies in this group have consistently shown that lung compliance decreases with the manifestation of left ventricular failure regardless of the underlying NYHA class or pulmonary vascular pressure [5, 6, 7]. Observational data indicate a direct correlation between the degree of reduced lung compliance and higher pulmonary capillary wedge pressure [6]. In patients with CS, elevated pulmonary capillary wedge pressures are common, often leading to pulmonary edema and, consequently, reduced lung compliance [9, 22]. However, specific details on how varying degrees of reduced lung compliance affect CS patients remain sparse. Insights from a subanalysis of the LUNG SAFE study, focusing on patients with isolated cardiogenic pulmonary edema undergoing non-invasive or invasive ventilation, reveal that higher peak, plateau and driving pressures—markers of more severe disease—were associated with increased mortality [9]. This was accompanied by a reduction in lung compliance [9], which could explain the higher pressures needed for ventilation. Lung compliance in the high-pressure group was around 30 mL/cmH₂O, correlating closely with results from the current study where a significant differentiation in Kaplan-Meier analysis was observed for lung compliance below 30.4 mL/cmH₂O.

In the present study the majority of the included CS-patients suffered from cardiac arrest. Previous research has consistently demonstrated that lung compliance decreases after cardiac arrest from different causes, potentially exacerbating the outcome [10, 11, 12, 23]. In our analysis, CS patients with accompanied cardiac arrest demonstrated higher mortality rates when lung compliance was in the lower quartiles. This supports the hypothesis that post-cardiac arrest syndrome—characterized by complications such as chest compression, inflammation and aspiration pneumonia—could induce lung edema, thus impacting lung compliance and influencing study outcomes [24, 25]. However, when stratified by median lung compliance and adjusted for potential cofactors using multivariable Cox regression, the association of reduced lung compliance (30.4 mL/cmH₂O) with 30-day all-cause mortality did not reach the level of significance in the subgroup of patients with cardiac arrest. This suggests that while lung compliance is a critical factor in the context of post-cardiac arrest syndrome, its influence extends beyond this subgroup. Indeed, in the broader cohort of CS patients, including those without cardiac arrest, reduced lung compliance remained significantly associated with increased mortality even after controlling for multiple variables. This highlights the importance of lung compliance not only in the immediate aftermath of cardiac arrest but also as a general prognostic factor in cardiogenic shock management.

This study is the first to demonstrate that lung compliance may impact in risk prediction in patients with CS. This aligns with the growing body of evidence indicating that mechanical ventilation in patients with CS significantly impact mortality outcomes [26, 27]. Recent research, including a study by Povlsen et al. [26], found that non-survivors of CS due to acute myocardial infarction had notably higher ventilator pressure settings [26]. Our findings extend this observation, showing that patients with lower lung compliance not only require higher ventilatory pressure settings but also exhibit higher 30-day all-cause mortality.

The large retrospective cohort study by Povlsen et al. [26] showed that such correlations were predominantly observed in patients who had not experienced out-of-hospital cardiac arrest. This reinforces the notion that the effects of CS on patient outcomes extend beyond those typically seen in post-cardiac arrest syndrome. It suggests that the management strategies for invasively ventilated cardiac patients should consider the distinct impact of CS itself, independent of complications arising from cardiac arrest.

The underlying mechanisms contributing to the observed relationships in this study are likely multifactorial. First, the direct impact of positive pressure mechanical ventilation on lung tissue must be considered. Patients with CS who exhibit low lung compliance typically require higher positive pressure to ensure adequate ventilation. This necessity arises particularly because pulmonary capillary wedge pressure is generally increased in cases of left ventricular failure, commonly leading to pulmonary edema in these patients [9, 22]. With increasing lung fluid, ventilated lung volume decreases and lung compliance reduces, mirroring the “baby lung” concept observed in patients with ARDS [4, 28]. Consequently, areas of the lung filled with fluid need a higher pressure of mechanical ventilation, resulting in higher shear and lung stress, which is known to induce ventilator induced lung injury (VILI) [29, 30, 31, 32, 33, 34, 35]. This hypothesis is supported by our data where patients with lower lung compliance were initially ventilated with median driving pressure of 19 cmH₂O, surpassing the threshold for lung-protective ventilation strategies [31]. Such high pressures can induce barotrauma, potentially leading to VILI, which can be prevented by applying low positive pressure ventilation [31]. Furthermore, adopting low pressure ventilation strategies might reduce the incidence of ARDS and pneumonia, further supporting the need for careful management of ventilatory pressures in this patient population [36].

In addition to the direct impacts on lung tissue, the need for higher pressure settings might also lead to an acceleration of systemic inflammation. Low pressure ventilation in ARDS patients was associated with lower levels of interleukin 6, which is an inflammatory marker [31]. This reduction in systemic inflammation could critically influence the progression of CS, as ongoing systemic inflammation exacerbates the shock condition and can accelerate patient deterioration [2, 37, 38]. Finally, the high-pressure ventilation caused by low lung compliance might have direct negative cardiovascular effects. An analysis by Rali et al. [27] of ventilator parameters in 2226 CS patients on extracorporeal life support suggested a cardioprotective effect of low-pressure ventilation. The study found that CS-patients managed with low pressure ventilation had the lowest mortality rates [27]. This observation is particularly significant as mortality in patients on extracorporeal life support is often not primarily due to respiratory failure. The authors suggest that low-pressure ventilation may provide cardio-protective benefits by influencing transmural pressures and the mechanical loading of the right and left ventricles. This hypothesis introduces an additional dimension to the benefits of low-pressure ventilation, suggesting that it could have systemic cardiovascular advantages beyond the respiratory system.

Taking these potential effects into account, we can hypothesize implications for daily clinical care. Patients with low lung compliance, who represent a high-risk subgroup within the CS population, require special attention. Pulmonary edema is a critical factor in the development of low lung compliance, suggesting that reducing lung fluid through diuretic therapy or renal replacement therapy may be beneficial [39]. Furthermore, low lung compliance may obstruct the application of low pressure ventilation. Therefore, in some patients, lung protective ventilation could be solely provided in prone positioning. However, it can be dangerous to place patients in prone position when they are experiencing CS since they are hemodynamically unstable. This challenge is reflected in our study’s real-world data, where a significant portion of the cohort suffered from cardiac arrest and was classified as SCAI shock stage E, resulting in high mortality rates. Consequently, prone positioning was only feasible for a small subset of patients. Despite these concerns, a study by Ruste et al. [40] has indicated that prone positioning was shown to increase cardiac index in patients with ARDS. Considering this, when the hemodynamic condition permit, it may be beneficial to place CS-patients in the prone position to improve survival. However, to corroborate this hypothesis, randomized controlled trials are needed.

This study contains a few limitations. Although we used multivariable Cox regression to account for potential cofounders, results may still be influenced by measurable or unmeasured cofounding factors due to the single-center and observational design of the study. Consequently, these findings should be considered as hypothesis generating rather than confirmatory because a direct causal association could not be demonstrated. While lung compliance might be a helpful risk predictor in CS, it should not be used in isolation. Rather, it should be integrated with additional parameter to enhance its predictive accuracy. Another limitation is the relatively small sample size of the study, which may have compromised its statistical power and influenced the outcomes. Therefore, the current findings require a reassessment as part of a more extensive investigation in larger trials. Additionally, due to the small sample size in the subgroup of patients without cardiac arrest on admission, survival analyses could not be performed, limiting the generalizability of our findings to the broader CS patient population without cardiac arrest. Additionally, the analysis did not include echocardiographic and invasive measurements which were inconsistent or only available in a small portion of the cohort. This included the measurement of the systolic pulmonary artery pressure, cardiac index, cardiac power output, or pulmonary capillary wedge pressure, before or after the implementation of invasive mechanical ventilation. Finally, Kaplan-Meier curves of the third and the fourth quartile of lung compliance showed crossing phenomena, which could be explained by specific clinical interventions and outcomes within these groups. Patients in the third quartile, who more frequently required extracorporeal life support and exhibited the highest rate of shock stage E, showed higher mortality in the initial days compared to the fourth quartile. Conversely, patients in the fourth quartile, who had higher applied tidal volume, may have succumbed to VILI, resulting in higher mortality after several days of ventilation. This suggests differential impacts on survival based on the management strategies and clinical conditions within these quartiles.

The present study demonstrated that patients with CS who exhibit low lung compliance experience a significantly higher 30-day all-cause mortality and fewer ventilator-free days. These findings provide new insights and call for more thorough research with prospective trials to better understand the impact of mechanical ventilation, including prone positioning, in patients with CS and cardiac arrest.

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

Conceptualization, JR, KW, IA, MAki and MB; methodology, JR; software, TS and AS; validation, JR, TS and KW; formal analysis, AS, PT, KM, MAyo and TS; investigation, SE-W and MR; resources, KM and MAyo; data curation, SE-W and MR; writing—original draft preparation, JR; writing—review and editing, JR, MB, IA, SE-W and TS; visualization, AS; supervision, MB, KM, MAyo and PT; project administration, IA and MB; revision of the manuscript, KW, MR, AS, MAki, PT, KM, MAyo. All authors read and approved the final manuscript. All authors have participated sufficiently in the work and agreed to be accountable for all aspects of the work.

The medical ethics committee II of the Medical Faculty Mannheim, University of Heidelberg, Germany, authorized the registry (study number 2019-695N). The Medical Faculty Mannheim’s Medical Ethics Committee II waived the requirement for study-specific informed consent.

This research received no external funding.

The authors declare no conflict of interest.