Steroids to Battle the Perfect Storm? Role of Corticosteroid Treatment in Sepsis Complicated by ARDS

Both sepsis and Acute Respiratory Distress Syndrome (ARDS) are major determinants of morbidity and mortality in the ICU, and sepsis is often complicated by ARDS. We discuss the role of corticosteroids in these patients.

Introduction

Acute respiratory distress syndrome (ARDS) is a heterogeneous syndrome that can be the result of several underlying conditions such as pneumonia, aspiration, trauma, pancreatitis and sepsis. Pneumonia and non-pulmonary sepsis are the main predisposing conditions for ARDS and sepsis also seems to be the primary trigger for pneumonia to progress into ARDS (Sheu et al. 2010). Patients with a sepsis-related ARDS experience worse symptoms and prognosis compared to patients with non-sepsis-related ARDS, including lower PaO₂/FiO₂ ratios, longer time on mechanical ventilation, higher incidence of ICU-acquired weakness, and a mortality reaching up to 50% in some studies (Sheu et al. 2020). This makes patients with sepsis-related ARDS a distinct group that could benefit from specific treatments to improve outcome. Here, we will briefly discuss the pathophysiology underlying sepsis-related ARDS, the role of phenotyping for identifying these hyperinflammatory patients, and the rationale and evidence for corticosteroid treatment to potentially battle this “perfect storm” in daily clinical practice.

Pathophysiology

How the pathophysiological mechanism of sepsis-related ARDS differs from ARDS in non-sepsis patients remains to be fully elucidated, but differences in circulating inflammatory biomarkers have been reported (Sheu et al. 2010; Hu et al. 2020; Brukhorst et al. 1999). Sepsis is characterised by a hyperinflammatory state and progression towards ARDS develops when the innate immune system of the host is not able to clear the pathogen. Inflammatory infiltration (cytokine storm) and high vascular membrane permeability are key factors in this pathogenesis. In their attempt to control the infection, cells of the innate immune system such as endothelial cells are overactivated and produce enormous amounts of proinflammatory cytokines such as IL-1α, IL-6, TNF-α, IFN-γ and GM-CSF, the last being responsible for the proliferation of macrophages, neutrophils, dendritic and T-cells (Hu et al. 2020). In the exudative phase of ARDS, the vascular endothelium of alveoli is damaged by the high levels of proinflammatory cytokines resulting in the leakage of plasma to the alveolar space, directly compromising oxygenation. At the same time, damaged pneumocytes lining the inner alveolar membrane go into sepsis-induced apoptosis, which stimulates the influx of inflammatory cells like neutrophils, macrophages and T-cells. These cells, together with dead pneumocytes, form hyaline membranes. This diffuse alveolar damage (DAD) forms the histological hallmark of the first phase of ARDS. During the second phase, fibroblasts and type-II pneumocytes start to proliferate leading to alveolar wall thickening, thereby further compromising gas exchange. During the exudative phase alveolar oedema with atelectasis prevail, resulting in oxygenation disturbances, while in the proliferative and fibrotic phase hypercapnia forms the major problem caused by the stiff but relatively dry lungs.

Taking this pathophysiology in mind it would be of uttermost importance to control the infection by applying broad spectrum antibiotics and obtaining source control, being restrictive in volume resuscitation, provide oxygen therapy and/or lung-protective mechanical ventilation, and stop the hyperinflammatory state as early as possible via targeted treatment. Regarding the latter, focus should be on identifying those patients that may benefit from corticosteroids.  

The Role of Phenotypes in Identifying Sepsis-Related ARDS and Benefit from Treatment

The complexity of the pathophysiological mechanisms of ARDS is further illustrated by the fact that in one third of patients with clinical criteria of ARDS the pathological hallmark DAD was not found during autopsy (Esteban et al. 2004). Furthermore, looking at the baseline characteristics of studies in ARDS patients, the main cause of ARDS was bilateral pneumonia (Bellani et al. 2016; Mammen et al. 2020). In randomised clinical trials, ARDS has been defined as fulfilling the Berlin-criteria, or in case of older studies, by the ESICM-SCCM consensus definition. These definitions lack aetiological or pathophysiological criteria, making ARDS diagnosis utilising these definitions a sump of different aetiologies and pathophysiological states. Instead of using Berlin criteria that involve physiological subphenotypes based on the PaO₂/FiO₂ ratio (ARDS Definition Task Force 2012), research has focused on identifying subgroups using clinical (e.g., aetiology), radiographic (e.g. focal vs. non-focal (Constantin et al. 2019)) and biological (Alipanah and Calfee 2022) characteristics. The latter aims to distinguish the host response to ARDS via assessment of (combinations of) biomarkers that are related to e.g., alveolar epithelial damage, endothelial damage and proinflammatory cytokines. Considering the pathophysiology of sepsis-related ARDS, using such biological phenotyping could be of particular interest. Via latent class analyses (LCA), Calfee’s group defined two of such phenotypes: a hyperinflammatory and a less-hyperinflammatory subtype, which exhibited heterogeneity of treatment effect (HTE) regarding the level of positive end-expiratory pressure (PEEP) (Calfee et al. 2014), volume resuscitation (Famous et al. 2017) and statins (Sinha et al. 2018), with the hyperinflammatory subtype being associated with higher mortality and (as so) a higher effect of the intervention. However, as LCA is characterised by creating two distinct groups to detect latent heterogeneity in data, we should be very careful with interpreting this as having prognostic value. Before subphenotypes can be leveraged to enrich future clinical trials or guide treatment decisions, external validation of the associated HTE is key. For instance, while the hyperinflammatory subphenotype appeared to derive greater benefit of high PEEP strategies in the ALVEOLI trial (Calfee et al. 2014), this was not replicated in the LOVS trial (Smit et al. preprint). Furthermore, if the objective is to identify patient subgroups that exhibit HTE for specific interventions, alternative predictive methodologies – e.g. risk modelling and effect modelling (Kent et al. 2020) – may be preferable. These approaches directly model HTE and generate individualised treatment effect predictions, rather than simply assigning patients to predefined subgroups.

Looking more in-depth to the hyperinflammatory subtype, it is characterised by high inflammatory proteins (IL-6, IL-8, TNF-α), more need of vasopressors and lower bicarbonate (Calfee et al. 2014). Clinicians would identify this as typical for septic shock. Indeed, a large proportion of the patients with this subtype have sepsis as their primary identified ARDS risk factor (Calfee et al. 2014). The role of corticosteroids in the treatment of ARDS, and to a lesser extent in septic shock, might still be controversial, though corticosteroids are recommended in the guidelines of both conditions (Chaudhuri et al. 2024). What still misses is a more personalised approach to the use of corticosteroids. It is unknown whether the hyperinflammatory subtype in ARDS (e.g., sepsis-related ARDS) in general has a (higher) benefit of corticosteroid therapy than the less-hyperinflammatory subtype, warranting a prospective study with availability of point-of-care biomarker assessment. During the pandemic, we learned in COVID-19-ARDS patients that there is an association between outcome (increased survival) and corticosteroids in the hyperinflammatory subtype (Chen et al. 2020). We need more studies to look at this relation between subtype and the effect of immunomodulation, including corticosteroids.

Corticosteroids for (Sepsis-Related) ARDS: Knowledge Gaps

Several knowledge gaps exist regarding the use of corticosteroids in ARDS, also beyond the sepsis-related ARDS. Most randomised clinical studies focused on patients with an infectious origin of ARDS. Furthermore, studies used different doses, types and duration of corticosteroids (Yoshihro et al. 2023). Time of initiation of corticosteroids after ARDS diagnosis also varied, warranting early recognition of ARDS to not delay a potentially beneficial treatment (Yoshihro et al. 2023).

Specific patient populations have been excluded from earlier trials (Table 1), precluding recommendations about corticosteroid treatment in these groups. Management of these patient categories is therefore a choice, in which the ‘do-no-harm principle’ should be weighed against the potential benefits of inhibiting a presumed hyperinflammatory response. Among the excluded patients are those with a deficient host immune response, such as HIV, or patients with active malignancy. HIV patients may be less likely to mount a hyperinflammatory response, as shown in a study where HIV patients had lower cytokine levels in response to a COVID-19 infection when compared to patients without HIV, which was particularly apparent in those with a low CD4 count (Maro et al. 2023).

Other reasons for exclusions were the presence of severe infections, for which invasive fungal infection was an exclusion criterion in half of the ARDS trials. In observational and/or retrospective studies in patients with ARDS due to influenza, receipt of corticosteroids was found to be associated with mortality, invasive pulmonary aspergillosis and with longer duration of shedding of the virus (Schauwvlieghe et al. 2018). Guidelines of the WHO and ESICM/SCCM advice against corticosteroids in ARDS due to influenza (WHO 2024; Pastores et al. 2017). However, selection bias is highly likely in these observations. A randomised clinical study evaluating corticosteroids in ARDS due to influenza is currently ongoing in the Netherlands (IMPRINT trial).

Presumably, identification of the (hyperinflammatory) ARDS subtype that benefits from corticosteroids may aid in the decision whether to treat with corticosteroids, also in patients in whom data are currently absent, such as in HIV and cancer patients. It was recently shown in an individual patient data analysis in patients with community-acquired pneumonia that those with a CRP > 204 pg/ml had benefit from corticosteroids, a finding that was irrespective of disease severity, need for mechanical ventilation or the presence of shock (Smit et al. 2025). The same may hold true for ARDS. If so, this simple biomarker may facilitate the decision to administer corticosteroids, or not. 

Conclusion

Patients with sepsis-related ARDS form a complex group with a worse prognosis compared to non-sepsis related ARDS. Although more and more studies suggest that there is a role for corticosteroids in hyperinflammatory ARDS patients, including those with sepsis, it is yet challenging to identify those patients that would benefit most from corticosteroid treatment. Therefore, nowadays guidelines carefully advise to start corticosteroids in all ARDS patients early in the disease course until results from new trials investigating personalised anti-inflammatory treatment become available.

Conflict of Interest

None.