## INTRODUCTION Cardiogenic shock (CS), a clinical syndrome characterized by insufficient cardiac output (CO) and tissue hypoperfusion, is often accompanied by acute multi-organ system dysfunction,1–3 including respiratory, liver, kidney, and neurologic dysfunction. Recent studies have highlighted the clinical heterogeneity of CS, including its distinct etiologic drivers and wide range of clinical severity.2,4,5 In addition, patterns of end-organ injury may be highly variable in CS.6,7 Although the presence and burden of acute non-cardiac organ dysfunction have been shown to correlate with mortality in the context of acute myocardial infarction-related CS (AMI-CS),8 less is known about the epidemiology and prognostic significance of end-organ dysfunction across the spectrum of common CS etiologies, including in heart failure-related CS (HF-CS). Moreover, whether HF chronicity (i.e., acute on-chronic [AoC] vs. *de novo* HF-CS) influences the relationship between acute non-cardiac dysfunction and outcomes in HF-CS is uncertain. These gaps in knowledge hold particular importance as non-cardiac organ function often drives decision-making with respect to candidacy for heart transplant and/or durable mechanical circulatory support (MCS). Understanding the diverse patterns and prognostic relationships of acute non-cardiac organ dysfunction in patients with AoC HF-CS, *de novo* HF-CS, and AMI-CS may provide pathophysiologic insights, improve risk stratification, and inform future investigations of novel therapeutic strategies in these distinct populations. Accordingly, we evaluated the epidemiology and outcomes associated with varying burden and type of acute non-cardiac organ dysfunction across CS subtypes using data from the Critical Care Cardiology Trials Network (CCCTN) Registry. ## METHODS The CCCTN Registry is an investigator-initiated network of predominantly tertiary cardiac intensive care units (CICUs) in the United States and Canada with oversight provided by the TIMI Study Group (Boston, MA). Detailed methods have been reported.9 From 2017 – 2022, participating CICUs contributed data on consecutive medical admissions during annual 2-month data collection periods. In addition, a subset of sites collected data on consecutive admissions year-round. The registry complies with the Declaration of Helsinki, and the protocol and waiver of informed consent were approved by ethics committees at each participating institution. CS was defined by sustained hemodynamic impairment (systolic blood pressure <90 mmHg or requirement for inotropic or vasopressor support) and evidence of end-organ hypoperfusion (e.g., altered mentation, oliguria, acute kidney injury, acute hepatic injury, metabolic acidosis, or serum lactate >2 mmol/L) owing to cardiac insufficiency, consistent with prior studies.2 CS etiology was further classified as AMI-CS, AoC HF-CS, or *de novo* HF-CS based on underlying etiology and HF chronicity.4 For this analysis, admissions for secondary (non-myocardial) CS (e.g., severe valvular disease, pericardial disease, etc) and post-cardiotomy CS were excluded. Definitions of respiratory, kidney, and neurologic dysfunction used in this analysis were informed by previously established thresholds from the Sequential Organ Failure Assessment (SOFA) score10 in conjunction with other relevant clinical data collected in the CCCTN Registry. Liver dysfunction was defined based on alanine transaminase (ALT) elevation, as ALT elevation has been shown to be highly associated with hypoperfusion and short-term mortality in patients with CS, while abnormalities in other liver function tests have not.11 Thresholds reflecting at least moderate organ dysfunction were selected (Table 1). Organ dysfunction was categorized as present if criteria were met at any point during CICU admission. Patients who received mechanical ventilation solely for airway protection to facilitate a procedure, had chronic dependence on mechanical ventilation, or were intubated after cardiac arrest but did not have respiratory failure were not classified as having acute respiratory dysfunction. Individual burden of acute non-cardiac organ dysfunction was quantified as: no non-cardiac organ dysfunction (NOD) when none of the four types of organ dysfunction (meeting the defined thresholds) were present, single organ dysfunction (SOD) when only one type of non-cardiac organ dysfunction was present, and multi-organ dysfunction (MOD) when two or more types of non-cardiac organ dysfunction were present. Descriptive statistics are presented as counts and percentages for categorical variables, and as medians with 25th-75th percentile ranges for continuous variables. Differences in categorical and continuous variables according to burden of organ dysfunction were assessed using Pearson’s chi-square test for trend and the Jonckheere-Terpstra test, respectively. Burden and type of organ dysfunction were compared across CS subtypes. In-hospital mortality rates were compared across CS subtypes and by burden of organ dysfunction within each CS subtype. Univariate logistic regression models were developed to evaluate associations between baseline patient characteristics and development of SOD and MOD. For these analyses, the odds of having SOD or MOD were compared to the odds of NOD for each baseline characteristic. The association between burden of organ dysfunction and in-hospital death was evaluated for the overall cohort and for each CS subtype using logistic regression models adjusted for age, sex, race, body mass index, smoking, diabetes, atherosclerotic cardiovascular disease, severe valve disease, chronic kidney disease, lung disease, liver disease, active cancer, preceding cardiac arrest, and LVEF on presentation. Similar models were developed to evaluate the associations between specific types of organ dysfunction (respiratory, liver, kidney, neurologic) and in-hospital mortality. Sensitivity analyses were performed in which each of the other types of acute non-cardiac organ dysfunction were included in the specific organ dysfunction models. P-value <0.05 was considered significant for all tests, and all P-values were two-sided. Analyses were performed with SAS V9.4 (Cary, NC, USA). ## RESULTS A total of 3,904 admissions for CS from 2017–2022 across 42 sites were included in this analysis. Of those, 1,146 (29.4%) were for AMI-CS, 1,989 (50.9%) were for AoC HF-CS, and 769 (19.7%) were for *de novo* HF-CS (Supplemental Figure 1). Overall, 1,383 (35.4%) CS admissions had NOD (meeting the criteria defined in this analysis), 1,399 (35.8%) had SOD, and 1,122 (28.7%) had MOD. The burden of end-organ dysfunction varied by CS subtype (Figure 1). Admissions for AoC HF-CS had the highest prevalence of NOD (42.6%) and the lowest prevalence of MOD (23.1%), while admissions for *de novo* HF-CS and AMI-CS had similarly high prevalence of MOD (33.9% and 35.0%, respectively) (p<0.01). Presenting SOFA scores were highest among admissions for AMI-CS (median [25th-75th percentile]: 8 [5–11]), followed by *de novo* HF-CS (7 [5–11]) and AoC HF-CS (6 [4–9]) (p<0.01). Characteristics of CS admissions according to burden of acute non-cardiac organ dysfunction are presented in Table 2. The burden of organ dysfunction was strongly correlated with other indicators of illness severity. As burden of organ dysfunction increased, there was higher serum lactate (p<0.01), higher SCAI Shock stage (p<0.01), and greater prevalence of preceding cardiac arrest (p<0.01). Similar patterns were observed within each CS subtype (Supplemental Table 1). By definition, there was also greater use of mechanical ventilation, acute renal replacement therapy, and targeted temperature management with increasing burden of organ dysfunction (Table 3). Intra-aortic balloon pumps (IABPs) were used with approximately equal frequency in patients with NOD, SOD, and MOD, but non-IABP temporary MCS was more commonly used in patients with MOD (p<0.01). The pattern of pulmonary artery catheter (PAC) use in relation to acute non-cardiac organ dysfunction varied by CS subtype. Whereas there was greater use of PACs with increasing burden of organ dysfunction in AMI-CS (p<0.01), PAC use was highest in those with NOD in both AoC HF-CS and *de novo* HF-CS (p_int = 0.01). Rates of successful bridging to heart replacement therapies (heart transplantation or durable MCS implant) were highest in those with lesser burden of organ dysfunction for both AoC HF-CS and *de novo* HF-CS and were very low overall in AMI-CS (Supplemental Figure 2). Respiratory dysfunction was the most common form of acute non-cardiac organ dysfunction across all CS subtypes; however, the prevalence of specific forms of acute non-cardiac organ dysfunction varied by CS subtype (Figure 2). Specifically, patients with AMI-CS and *de* novo HF-CS had higher prevalence of respiratory dysfunction (48% and 42%, respectively, vs. 31%; p<0.01) and neurologic dysfunction (21% and 20% vs. 8%; p<0.01) compared to those with AoC HF-CS. Liver dysfunction was most prevalent in patients with *de novo* HF-CS (30%), followed by AMI-CS (25%) and AoC HF-CS (22%) (p<0.01). By contrast, the prevalence of kidney dysfunction was similar across CS types (24–27%, p=0.33). Several baseline patient characteristics were associated with greater burden of acute non-cardiac organ dysfunction. In the overall cohort, preceding cardiac arrest was the baseline characteristic most strongly associated with burden of organ dysfunction (odds ratio [OR] of SOD vs. NOD 2.29, 95% confidence interval [CI] 1.86–2.81; OR of MOD vs. NOD 3.76, 95% CI 3.06–4.62), a finding that was consistent across CS subtypes (Supplemental Table 2). In addition, peripheral artery disease (OR 1.56, 95% CI 1.19–2.06), CKD (OR 1.45, 95% CI 1.22–1.73), and diabetes (OR 1.35, 95% CI 1.15–1.59) were also each associated with higher odds of MOD vs. NOD. Unadjusted in-hospital mortality was highest in AMI-CS (39.8%), followed by *de novo* HF-CS (30.8%), and AoC HF-CS (25.0%) (p<0.01). In-hospital mortality rate increased as the burden of acute non-cardiac organ dysfunction increased in the overall CS cohort (NOD: 13.3%, SOD: 29.7%, MOD: 52.7%; p<0.01) and within each CS subtype (p<0.01 for all; Figure 3). This pattern was consistent in subsets of CS admissions managed with IABP and non-IABP advanced temporary MCS (Supplemental Table 3). After multivariable adjustment, SOD and MOD remained independently associated with higher odds of in-hospital death compared with NOD in the overall CS cohort. The adjusted associations of SOD and MOD with in-hospital death were similar across CS subtypes (p_int=0.74; Figure 4). Kidney dysfunction was the most prognostically important form of acute non-cardiac organ dysfunction in the overall CS cohort (aOR for in-hospital death 4.13, 95% CI 3.42–4.99) and across each CS subtype (Figure 5). Respiratory dysfunction was associated with the next highest odds of in-hospital mortality in the overall cohort (aOR 2.56, 95% CI 2.17–3.03), with a particularly strong association in patients with AoC HF-CS (aOR 3.01, 95% CI 2.34–3.87). The adverse prognostic significance of liver dysfunction was highest in AMI-CS (Figure 5). Results were consistent when models for each specific form of acute non-cardiac organ dysfunction (e.g., kidney) were further adjusted for the other forms of acute non-cardiac organ dysfunction (e.g., liver, respiratory, and neurologic) (Supplemental Table 4). ## DISCUSSION This analysis of the patterns and prognostic associations of acute non-cardiac organ dysfunction in patients with CS from a multinational registry has several notable findings. First, acute non-cardiac organ dysfunction was common, and a higher burden of acute non-cardiac organ dysfunction was strongly associated with greater risk of in-hospital death, including among patients treated with temporary MCS. This association has been established in patients with AMI-CS,8 and we now extend this observation to patients with HF-CS. Second, compared with AoC HF-CS, AMI-CS and *de novo* HF-CS were associated with a greater burden of acute noncardiac organ dysfunction, most notably respiratory dysfunction. However, acute non-cardiac organ dysfunction was similarly adversely prognostic in AoC HF-CS compared with AMI-CS and *de novo* HF-CS. Lastly, while all forms of acute non-cardiac organ dysfunction were associated with increased risk of in-hospital death, kidney dysfunction was the most prognostically important type of organ dysfunction across CS subtypes. MOD complicated over one-third of admissions in the *de novo* HF-CS and AMI-CS groups, but under one-quarter of admissions for AoC HF-CS. The lower burden of acute non-cardiac organ dysfunction may, at least in part, account for the lower in-hospital mortality rate observed in AoC HF-CS compared with AMI-CS and *de novo* HF-CS in our cohort. Moreover, regional differences in the estimated mortality from AoC HF-CS may relate to variation in the burden of acute non-cardiac organ dysfunction.12–14 The difference in burden of acute non-cardiac organ dysfunction reflects that patients with AoC HF-CS may have chronic compensatory adaptations of non-cardiac organs that are not present in patients with *de novo* HF-CS or AMI-CS. While mechanisms for hypoxic and ischemic conditioning of non-cardiac organs via hypoxia-inducible factors have been explored in preclinical models,15,16 additional investigations are needed in HF and CS patients to better understand compensatory mechanisms that may be present in patients with chronic HF. The specific types of acute non-cardiac organ dysfunction varied by CS subtype. While respiratory dysfunction was the most common type of acute non-cardiac organ dysfunction overall, it was more prevalent in admissions for *de novo* HF-CS and AMI-CS compared with AoC HF-CS. This observation may be related to increases in thoracic lymphatic capacity that occurs over time and results in less overt pulmonary edema in patients with chronic HF and chronically elevated pulmonary artery wedge pressures.17 Additionally, chronic pulmonary venous hypertension can lead to cellular and morphologic alterations of the pulmonary capillary basement membrane that lower the capillary filtration coefficient and increase hydrostatic thresholds for the development of pulmonary edema.18,19 In contrast, patients with AMI-CS or *de novo* HF-CS have more limited lymphatic reserve, making them prone to develop overt pulmonary edema precipitously. Even though pre-existing CKD was far more prevalent in the AoC HF-CS group, rates of acute kidney dysfunction were similar across CS subtypes, again suggesting the possibility of chronic compensatory adaptations in patients with chronic HF. Despite AoC HF-CS patients having a lower burden of SOD and MOD, acute non-cardiac organ dysfunction was strongly associated with risk of in-hospital death in this group, similar to AMI-CS and *de novo* HF-CS. This may reflect that when significant organ dysfunction develops in AoC HF-CS, chronic compensatory mechanisms are overwhelmed and irreversible organ injury ensues. Moreover, patients with AoC HF-CS may no longer be candidates for durable MCS or isolated heart transplant after irreversible acute non-cardiac organ injury develops. Whether a framework of severity of acute non-cardiac organ dysfunction may be used to help select appropriate and exclude inappropriate candidates for temporary MCS will be an important area for ongoing investigation. Differences in the associations between specific types of organ dysfunction and in-hospital death were also observed. Kidney dysfunction was associated with ~4-fold increased risk of in-hospital death in the overall cohort, and this association was strongest in the AoC HF-CS group. This finding raises the possibility that mitigating the incidence and impact of kidney dysfunction may be important for improving CS outcomes and could be considered in the design of future clinical trials. Respiratory dysfunction was also most strongly associated with increased mortality risk in the AoC HF-CS group, possibly indicating higher levels of illness severity when chronic lymphatic adaptations are overwhelmed in patients with pre-existing HF and CS. This study should be interpreted in the context of several limitations. First, the definitions of acute non-cardiac organ dysfunction were derived from the SOFA score, which was originally developed in the context of septic shock,10 and further modified based on data collected in the CCCTN Registry. Results and point estimates could vary if differing definitions of acute non-cardiac organ dysfunction were used. Moreover, the NOD group included patients with acute changes in organ function that were not severe enough to meet the criteria used in this analysis. Second, while hemodynamic and laboratory parameters were collected in the CCCTN Registry, the timing of these parameters relative to the onset of end-organ dysfunction was not. Thus, multivariable models could not reliably be adjusted for such covariates. Third, the duration of pre-existing HF in the AoC HF-CS group was not known, and the association between pre-existing HF duration and outcomes in CS should be further investigated. Fourth, post-hospital outcomes were not available for analysis. Fifth, abstracting GCS from electronic health record data may not be as reliable as bedside evaluation, and whether neurologic dysfunction was due to malperfusion versus toxic-metabolic encephalopathy was unclear. Sixth, since the CCCTN Registry captures individual admissions rather than individual patients, the results do not take into account repeated admissions for the same patient. Lastly, residual unmeasured confounding may be present in adjusted models given the observational design of this study. ## CONCLUSIONS In this analysis of admissions for CS to North American CICUs, admissions for AoC HF-CS had a lower burden of acute non-cardiac organ dysfunction and in-hospital mortality compared with *de novo* HF-CS and AMI-CS. However, end-organ dysfunction was similarly associated with risk of in-hospital death in AoC HF-CS compared with other groups. Kidney dysfunction was most strongly associated with adverse prognosis across CS subtypes. These findings may help to further improve risk stratification and prognostication in patients admitted with AMI-CS, *de novo* HF-CS, or acute-on-chronic HF-CS.