Why Molecular Subphenotyping Is Needed in Cardiogenic Shock and How to Accomplish This

Cardiogenic shock (CS) represents approximately 15% of admissions to cardiac ICUs and continues to be associated with 30–50% in-hospital mortality (1). CS is defined as a clinical syndrome of hypoperfusion related to cardiac pump failure (2). However, this definition has important limitations. First, its broad nature leads to the “lumping” of potentially very diverse disease states. CS varies in etiology, clinical and macro-hemodynamic profile, and severity (Figure 1A). Additionally, CS may occur concurrently with other critical care syndromes (e.g., septic cardiomyopathy, severe right ventricular dysfunction in acute respiratory distress syndrome [ARDS]), complicating clinical phenotyping. Second, this syndromic definition emphasizes a “cause-and-effect” relationship, overlooking intermediary biological disease drivers within individual or groups of patients. Between inciting etiologies, hemodynamic manifestations, and organ dysfunction, there is likely considerable variability in the “host response” and, subsequently, the resulting molecular pathways driving the disease process and outcomes (Figure 1A). Such pathways include damage-associated molecular patterns, systemic inflammation, immune activity, endothelial dysfunction, micro and macrocirculatory uncoupling, and oxidative stress (3, 4).

Figure 1.

(A) Summary of the traditional (clinical classification), emerging (SCAI shock staging), and future (subphenotyping) approaches for unraveling cardiogenic shock (CS) heterogeneity. The molecular and clinical data representing the multilayered heterogeneity of CS is illustrative and not intended to be exhaustive. (B) Roadmap for evidence generation on CS subphenotyping research. AMICS = acute myocardial infarction CS; ECG = electrocardiogram; MCS = mechanical circulatory support; SCAI = Society for Cardiovascular Angiography and Interventions.

Recent translational research suggests that molecular response profiles vary across CS and its inciting etiologies (5–7). However, in-depth exploration has been constrained by an absence of large CS cohorts with associated biobanks (5, 8, 9). This contrasts with other critical care syndromes such as sepsis and ARDS, where large-scale biosample initiatives have permitted the discovery of clinically important heterogeneity in host response that is linked with prognosis and, possibly, treatment effect (10).

Accordingly, in this viewpoint, we call for an international collaboration to standardize and amalgamate the numerous CS biosample repositories from diverse settings, to create a communal molecular data lake with the needed patient sample sizes supporting this work in CS. In the following text, we outline the unmet need and the steps toward its creation.

CS as a Heterogeneous Critical Care Syndrome

Clinical and Mechanistic Heterogeneity: Current Approaches

CS is currently subclassified on the basis of clinical status, clinical severity stages, and/or physiological and biological subpenotypes (Figure 1A). CS is clinically heterogeneous, prompting a range of proposed clinical subclassification frameworks. These include classification by primary etiology (ischemic or nonischemic); timing (de novo or acute on chronic); hemodynamic profile; severity of organ dysfunction; and the presence of concurrent critical care syndromes such as sepsis, ARDS, acute kidney injury, and cardiac arrest (11). Exploratory data-driven clustering has identified additional clinical subphenotypes with variable prognosis (12, 13). CS is also heterogeneous in terms of clinical severity. The Society for Cardiovascular Angiography and Interventions (SCAI) staging framework stratifies patients with CS into five stages, which are prognostic. Beyond prognosis, there is an initial suggestion that, in the future, this staging framework may be useful in treatment decisions (14, 15).

CS is hemodynamically heterogeneous, comprising left ventricular versus right ventricular versus biventricular or valvular dysfunction, altered organ interactions, and macro- and microcirculatory uncoupling (11, 16). Additionally, a clinically readily evident subphenotype is the 20–30% of patients who have an inappropriately low systemic vascular resistance (“mixed shock”) and concurrent systemic inflammatory response syndrome (17).

Unraveling Molecular Heterogeneity: The Next Step

On the basis of seminal observations in sepsis and ARDS, it is widely hypothesized that CS may also be molecularly heterogeneous. This hypothesis is supported by a very small number of studies of CS and its inciting etiologies (3). For example, in patients who experienced myocardial infarction with preshock (SCAI stage B), as in sepsis, leukocyte transcriptional signatures stratify patients on the basis of hyper- and relatively hypoinflammatory responses, as well as variability in thrombosis activation (5). Understanding such molecular variability would allow the targeting of, for example, immunomodulating therapies to prevent or treat CS (18). Furthermore, recovery subphenotypes among CS survivors—defined by distinct host-response biomarker patterns at ICU discharge (i.e., inflammation, myocardial fibrosis, and endothelial dysfunction)— were similar to those identified in sepsis survivors and were associated with variable long-term mortality and physical disability (7, 19). Accordingly, understanding CS endotypes may support a broader, treatable trait-based lens on diverse critical illness syndromes with an overlap in molecular responses.

Although the findings from these modestly sized studies suggest that such molecular heterogeneity may be present in patients with CS, the ability to rigorously evaluate hypothesis this has been limited by small cohort and biobank sizes to date (5, 8, 9). The integration of multiple biomarkers to identify and validate subphenotypes may also face methodological challenges such as data quality issues, and nonstandardized statistical analysis.

Building a Global Large-Scale CS Molecular Consortium

To assess clinically relevant molecular subphenotypes based on multi-omics biomarkers reflecting CS underlying pathways, large-scale biobanks are needed on the road to develop precision medicine strategies and improve outcomes in CS independently from its etiology and clinical severity (Figure 1B). For this purpose, we have undertaken the development of the Shock and acute Conditions OutcOmes Platform (ShockCO-OP) Consortium (ClinicalTrials.gov ID: NCT06376318). This collaborative group is fully open to participation and membership from any interested investigators and research networks. This Consortium represents both a practical and dynamic forum for data sharing and the amalgamation of publicly available longitudinal molecular data, overcoming the sample size issues that have held the field back, as well as an ideological meeting square for like-minded investigators to allow the standardization of prospective biospecimen collection, biomarker measurements, and unsupervised machine-learning approaches (i.e., model-based clustering). This will ultimately allow an accelerated journey of discovery in collaboration with sepsis and ARDS subphenotyping researchers.

Conclusion

Emerging efforts seek to capture the heterogeneity of the critical care syndrome of CS through a broad range of clinical classifiers, severity stages, and hemodynamic profiles. However, more work on a larger scale is needed to classify CS on the basis of molecular subphenotypes (20). We propose the creation of the ShockCO-OP Consortium to facilitate this critically needed translational research. Such work represents the next step on the road to developing future enriched and mechanistic clinical trials with the inclusion of patients with CS on the basis of a specific molecular subphenotype rather than the usual enrollment of consecutive patients.