Cardiogenic shock in cancer: differences in use of temporary mechanical circulatory support, invasive hemodynamic assessment and patient outcomes

Abstract

Background

Cancer is a growing comorbidity in patients with cardiovascular disease, but implications of cancer for treatment and outcomes of cardiogenic shock (CS) are not well described.

Objectives

This study aimed to identify associations of a cancer diagnosis with use of temporary mechanical circulatory support (tMCS) and survival in young adults hospitalized with CS.

Methods

We identified young adults (age 18–49) with CS in a large, nationwide sample of hospitalizations (years 2016–2020), and stratified cases according to history of cancer. Variables of interest included clinical and hospital characteristics, use of tMCS, and survival. Multivariable logistic regression was performed to evaluate independent associations of cancer with care and outcomes.

Results

Out of 99,335 young adults with CS, 2,705 (2.7%) had cancer. Patients with cancer had a lower burden of cardiovascular risk factors and cardiovascular disease at baseline. STEMI accounted for 13% of CS in patients without cancer compared to 4% in those with cancer. Multivariate analysis adjusted for STEMI, hypertension, diabetes, hyperlipidemia, and other cardiovascular risk factors showed that patients with cancer were less likely to receive tMCS (OR = 0.54, 95% CI 0.40–0.72, p < 0.001) and less likely to undergo pulmonary artery catheterization. They were also more likely to receive palliative care referrals, and more likely to die in the hospital (OR = 1.49, 95% CI 1.23–1.80, p < 0.001).

Conclusions

Among young adults hospitalized with CS in the United States in recent years, patients with cancer were less likely to receive tMCS and more likely to die in the hospital despite less pre-existing cardiovascular disease and fewer comorbidities.

Supplementary Information

The online version contains supplementary material available at 10.1186/s40959-025-00427-1.

Introduction

Cardiogenic shock (CS) is understood to be a complex, deadly syndrome which can arise out of myriad cardiovascular disease processes and clinical contexts [1–3]. Treatment strategies may include pharmacotherapy, coronary revascularization, and temporary mechanical circulatory support (tMCS). Diversity of CS and paucity of randomized evidence have precluded uniform recommendations for management [4], entrusting many critical treatment decisions to clinical judgment and, increasingly, the wisdom of teams [5]. Analysis of patient diagnosis and comorbidity is fundamental to this calculus and may influence treatment decisions and outcomes.

Growing prevalence of cancer and accelerating innovation in cancer therapies call attention to the relevance of cancer as a comorbidity and causal factor in patients with cardiovascular disease and CS. Yet implications of a cancer diagnosis for management of CS, including allocation of tMCS, and outcomes of CS in the modern era are not well known.

A body of literature has explored implications of comorbid disease states for treatment decisions and outcomes in acute cardiovascular illness, perhaps best exemplified by study of acute coronary syndromes in patients with chronic kidney disease [6]. Critical analysis has yielded insights into not only biological rationales for differences in treatment decisions and outcomes, but also potential biases in care and opportunities for improvement and innovation. Recognition of patterns and differences is an essential first step to generate hypotheses for further research. In this spirit, we sought to examine the relationship between a cancer diagnosis and treatment decisions and outcomes of CS in a large, representative sample of hospitalized young adults with CS.

Methods

Cases of CS were identified among hospitalizations included in the National Inpatient Sample between January 1, 2016 and December 31, 2020. Details of the design of the National Impatient Sample have been described previously [7]. To generate national estimates, we analyzed data in a weighted manner with the use of standardized billing codes, including the 10th Revision of the International Classification of Diseases (ICD) and the Current Procedural Terminology (CPT) codes. The present analysis defined a diagnosis of CS by the presence of ICD-10 code R57.0, as has been validated previously [1, 8]. In order to generate a cohort of young adults, inclusion was restricted to patients with an age of 18 to 49 years old. The study cohort was stratified according to the presence or absence of a history of cancer. “Cancer” included either active or prior cancer diagnosis (see Supplementary Table 1), regardless of treatment received.

Demographics, baseline characteristics including cardiovascular history and risk factors, hospital characteristics, and clinical features of the CS hospitalization were obtained and compared across subgroups. In order to permit comparison of care processes, we measured utilization of percutaneous coronary intervention (PCI), coronary artery bypass graft surgery (CABG), endotracheal intubation, renal replacement therapy, pulmonary artery catheter placement, central venous line placement, palliative care services, tMCS, durable left ventricular assist device (LVAD) implantation and heart transplantation via associated CPT codes (see Supplementary Table 2) and compared rates of use across subgroups. For this analysis, tMCS was defined as use of one or more of the following: intra-aortic balloon pump (IABP), microaxial flow pump (mAFP), and venoarterial extracorporeal membrane oxygenation (VA-ECMO). Mortality included death in the hospital due to any cause.

The primary measure of care processes for this study was allocation of tMCS. The primary outcome measure was mortality. Statistical analysis was performed using Stata version 18 (StataCorp, College Station, TX). Normality was tested for all variables. Continuous variables were compared using t-testing and categorical variables were compared using Chi-square testing. Statistical tests were 2-tailed, and p values less than 0.05 were considered statistically significant. Logistic regression analysis was performed to generate odds ratios (OR) for tMCS allocation and mortality, adjusting for age, cardiovascular risk factors, baseline cardiovascular disease, ST-elevation myocardial infarction (STEMI), and metastasis. Additional adjustments were done for race, ethnicity, socioeconomic, and geographic variables, to calibrate for possible inequities as suggested in a recent study [8]. A total of 36 patients with missing data were excluded from analysis. The institutional review board exempted the study from review.

Results

A total of 99,335 hospitalizations for young adults with CS were identified between the years 2016 and 2020, of whom 2,705 (2.7%) had a history of cancer. Patients with cancer included a higher proportion of female patients (45.6% vs. 35.3%, p < 0.001) and had lower rates of pre-existing hyperlipidemia, diabetes, heart failure, chronic kidney disease, and coronary artery disease (Table 1). Rates of prior cardiac arrest, CABG, and PCI were low in both groups and did not statistically differ.

Table 1.

Baseline characteristics in young cancer and non-cancer patients hospitalized with cardiogenic shock

Variable	Non-Cancer (n = 96,630)	Cancer (n = 2,705)	P Value
Age (years old)	39.3	39.7	0.7186
Female Sex (%)	35.3%	45.6%	< 0.0001
Race (%):
White	52.0%	52.1%	0.4303
African American	27.1%	24.6%
Hispanic	12.2%	12.8%
Asian or Pacific Islander	3.6%	5.3%
Native American	1.1%	1.3%
Other	4.0%	3.6%
Median Household Income* (%):
Quartile 1	35.7%	27.8%	0.0001
Quartile 2	26.5%	24.8%
Quartile 3	21.7%	26.6%
Quartile 4	16.1%	20.8%
Hospital Size †(%):
Small	10.5%	10.2%	0.5377
Medium	20.7%	18.9%
Large	68.8%	71.0%
Country Region (%):
Northeast	14.0%	19.4%	0.0008
Midwest	22.1%	23.5%
South	41.6%	34.6%
West	22.4%	22.6%
Hypertension (%)	12.1%	10.9%	0.4107
Hyperlipidemia (%)	21.5%	11.3%	< 0.001
Diabetes (%)	12.0%	8.0%	0.0037
Heart Failure (%)	47.0%	32.5%	< 0.001
Chronic Kidney Disease (%)	26.0%	19.0%	< 0.001
Coronary Artery Disease (%)	20.1%	12.2%	< 0.001
Peripheral Artery Disease (%)	3.2%	2.03%	0.1226
Atrial Fibrillation (%)	19.7%	12.6%	< 0.001
Tobacco Exposure (%)	24.5%	12.9%	< 0.001
Coagulopathy (%)	20.4%	22.2%	0.3238
History of Cardiac Arrest (%)	2.4%	1.1%	0.0910
History of CABG (%)	1.5%	0.5%	0.0793
History of PCI (%)	0.6%	0.4%	0.5504

*The quartile classification of the estimated median household income of residents is made according to each patient's ZIP Code. The quartiles are identified by values of 1 to 4, indicating the poorest to wealthiest populations

†Hospital size categories in NIS are based on hospital beds, and are specific to the hospital's location and teaching status. It assesses the number of short-term acute care beds set up and staffed in a hospital

With respect to demographics, patients with cancer were distinguished by greater representation of the highest income quartile and hospitalization in the Northeast region of the United States. No significant differences in race or ethnicity were apparent across groups, and distribution of cases according to hospital size was similar (Table 1).

CS hospitalizations in patients with cancer were less likely to occur in the setting of acute myocardial infarction. This was true for both STEMI (3.9% vs. 13.1%, p < 0.001) and NSTEMI (6.7% vs. 9.9%, p = 0.014). In turn, patients with cancer were less likely to undergo revascularization procedures, including both PCI and CABG. Allocation of endotracheal intubation and renal replacement therapies was similar. Patients with cancer were less likely to undergo placement of a pulmonary artery catheter (17.4% vs. 25.6%, p < 0.001) (Table 2).

Table 2.

Cardiogenic shock profiles by usage of resources, in cancer and non-cancer patients

Variable	Non-Cancer (n = 96,630)	Cancer (n = 2,705)	P Value
STEMI	13.1%	3.9%	< 0.001
NSTEMI	9.9%	6.7%	0.0143
Intubation	47.2%	49.4%	0.3339
Renal Replacement Therapy	10.5%	10.7%	0.8907
Pulmonary Artery Catheter	25.6%	17.4%	< 0.001
PCI	11.2%	2.0%	< 0.001
CABG	4.4%	2.0%	0.0082
Temporary Mechanical Circulatory Support	21.2%	9.8%	< 0.001
IABP	12.0%	5.7%	< 0.001
Impella	5.7%	1.7%	< 0.001
ECMO	6.5%	4.4%	0.0466
LVAD	3.7%	3.0%	0.3793
Transplant	2.3%	0.6%	0.0070
Palliative Care Encounters	12.8%	21.6%	< 0.001
Length of stay (days)	14.0	16.5	0.0016
Mortality	25.8%	39.7%	< 0.001

Patients with cancer were more likely to receive palliative care services (21.6% vs. 12.8%, p < 0.001) (Table 2). This association remained significant after adjustment for age, sex, presence of metastasis, STEMI, hypertension, diabetes, CAD and heart failure (OR = 1.44, 95% CI 1.13–1.82, p = 0.003). The presence of metastasis itself was associated with a significantly higher OR of 3.0 (95% CI 2.00-4.49, p < 0.001) for palliative care services referral (Supplementary Table 3).

Allocation of tMCS was less common among patients with cancer (9.8% vs. 21.2%, p < 0.001). Following multivariable adjustment, a diagnosis of cancer was associated with a nearly 50% reduction in odds of receiving tMCS as part of care for CS (OR = 0.53, 95% CI 0.40–0.71, p < 0.001) (Table 3) (Fig. 1). Reduced odds of tMCS allocation were consistent and significant across all studied tMCS platforms: for IABP, OR = 0.66, p = 0.029; for mAFP, OR = 0.43, p = 0.015; for VA-ECMO, OR = 0.61, p = 0.018. Significant predictors of tMCS allocation included STEMI (OR = 4.40, p < 0.001), coronary artery disease (OR = 1.91, p < 0.001), history of heart failure (OR = 1.21, p < 0.001), and admission to a teaching hospital (OR = 1.43, p < 0.001) (Fig. 2).

Table 3.

Multivariate analysis for MCS allocation in cancer patients, adjusted for cardiovascular risk factors and cardiovascular disease

Predictors	Odds Ratio	95% Confidence Interval	P Value
Cancer	0.53	0.40–0.72	< 0.001
Presence of Metastasis	0.40	0.17–0.94	0.036
Age	0.99	0.98–0.99	< 0.001
Female Sex	0.94	0.87–1.02	0.129
Hypertension	0.95	0.85–1.07	0.438
Diabetes	0.91	0.81–1.02	0.101
Coronary Artery Disease	1.91	1.75–2.09	< 0.001
Heart Disease	1.21	1.11–1.31	< 0.001
STEMI during hospitalization	4.40	3.96–4.89	< 0.001

Fig. 1.

Odds of temporary mechanical circulatory support according to patient and hospital characteristics

Fig. 2.

Central illustration: temporary mechanical circulatory support allocation, pulmonary artery catheter use, palliative care consultation, and mortality of young adults with cardiogenic shock, with and without cancer

In recognition of the potential for presence of metastasis to be a critical driver of patient exclusion from tMCS allocation [9], we conducted a sensitivity analysis examining only those patients with cancer without metastasis. In multivariate logistic regression analysis, patients with cancer remained significantly less likely to receive tMCS, with OR = 0.534 (95% CI 0.39–0.7), p < 0.001. Recognizing the potential influence of palliative care consultation on tMCS decisions, we also conducted a sensitivity analysis excluding patients with a documented palliative care encounter. In this context, a cancer diagnosis remained significantly associated with a lower allocation of tMCS (OR = 0.573 [95% CI = 0.42–0.79], p = 0.001).

Patients with cancer experienced higher rates of in-hospital mortality which remained significant after multivariable adjustment (OR = 1.50, 95% CI = 1.24–1.82, p < 0.001). Cancer was associated with significantly lower allocation of orthotopic heart transplantation (OR = 0.27, 95% CI = 0.09–0.82, p = 0.021) and no patient with presence of metastasis underwent transplantation. LVAD utilization did not differ significantly between patients with or without cancer (3.0% vs. 3.7%, p = 0.38).

Discussion

In this contemporary study of young adults hospitalized with CS in the United States, a diagnosis of cancer predicted lower utilization of invasive hemodynamic monitoring, tMCS and surgical therapies for heart failure; greater utilization of palliative care services; and higher likelihood of in-hospital mortality. Although the cohort of patients with cancer did differ in important respects, including exhibiting a lower burden of causative acute myocardial infarction, the above differences persisted even with multivariable adjustment. Notably, patients with cancer experienced a 50% higher rate of in-hospital death despite substantially lower rates of cardiovascular comorbidities. Taken together, these data serve to support a hypothesis that cancer influences treatment decisions and outcomes in CS.

Patients with cancer hospitalized for CS face a high risk of mortality that aggregates both oncologic and cardiovascular causes of death. Clinicians caring for these patients analyze both of these risks, among others, when evaluating individual candidacy for diagnostic and treatment options, including both cancer and cardiovascular therapeutics. The assessment of a poor oncologic prognosis—perhaps captured best in the current study by the documented presence of metastasis—may demotivate clinicians from offering more advanced cardiovascular therapies for CS such as invasive hemodynamic-guided management and tMCS in favor of earlier allocation of palliative care services.

Yet imprecision in oncologic prognostication, only further confounded by the forward advance of cancer therapies, risks therapeutic nihilism in acute cardiovascular care and the deprivation of care paradigms associated with improved outcomes of CS. A growing body of literature, for example, has shown the improved likelihood of survival from CS in conjunction with use of invasive hemodynamic monitoring [10]. Randomized studies to date [11–14] have mostly failed to show routine benefit of tMCS in CS, save for a specific trial of mAFP use in acute myocardial infarction-related CS [15]. Nonetheless, the consideration of tMCS is a central element of team-based, protocol-driven multidisciplinary CS programs which have demonstrated significant improvements in CS survival in recent years [16–18]. Data from the present analysis compel us to consider carefully whether our patients with CS and cancer may yet derive benefit from advanced cardiovascular therapies in spite of their oncologic diagnosis. Of note, it is critical to recognize that median survival for many cancers—including even certain metastatic cancers such as breast, renal cell, prostate and neuroendocrine malignancies—has improved substantially, even surpassing 5 years [19–21]. The changing landscape of outcomes in cancer is a matter of subspecialized oncologic expertise that falls outside the scope of most cardiologists practicing in the CS space. To the extent that oncologic prognostication contributes to the calculus of cardiovascular management in patients with CS and cancer, this evolving landscape underscores the essential need for partnership between cardiologists and oncologists in care for these patients. Furthermore, whereas measurable benefits of CS therapies may impact short-term cardiovascular, but not long-term oncologic, prognosis, the true valuation of these benefits depends critically upon the goals and autonomy of a patient.

For this manuscript, we explored avenues to determine frequency of inpatient oncology consultation for our cohort. Whereas ICD-10 codes exist for chemotherapy infusions (Z51.1) and encounters for antineoplastic chemotherapy (Z51.12), there is not a general code for inpatient oncology consultation in the manner that exists for palliative care. Accordingly, it was not feasible for this analysis to reliably ascertain the frequency of inpatient oncologist involvement. Future study capturing formal involvement of oncology is needed to better examine the impact of cardiology-oncology partnership on outcomes of CS in patients with cancer.

Lower rates of acute myocardial infarction and pre-existing heart failure in patients with cancer and CS draw attention to a likely distinct spectrum of etiologic processes driving CS in this cohort. Some, if not all, of these cases may relate directly to cancer therapy-related cardiovascular disease (CTRCD) [22]. Examples of CTRCD potentially relevant to the pathogenesis of CS include well-recognized entities such as anthracycline-induced cardiomyopathy [23, 24], HER2-targeted therapy-related cardiac dysfunction [22, 25, 26], immune check-point inhibitor-associated myocarditis and non-inflammatory heart failure [22, 27, 28], as well as cardiotoxicity associated to more novel cancer treatments such as chimeric antigen receptor T-cell therapy [29, 30] and bispecific T-cell engagers [30, 31]. Some cases of CS may relate directly to the neoplastic disease process itself, as in the examples of takotsubo cardiomyopathy [32, 33], carcinoid heart disease [34, 35], metastatic pericardial disease [36], and primary and metastatic cardiac tumors [37, 38]. Other cases of CS in cancer patients may result from secondary complications of cancer, such as septic shock with cardiomyopathy, pulmonary embolism with cor pulmonale, or pulmonary hypertension, in the case of lung cancer. While limitations of the NIS database preclude fine discrimination of mechanisms of CS in the present analysis, future analysis from other data sources is needed to elucidate the pathogenetic spectrum of CS in cancer.

Advanced modalities of invasive hemodynamic monitoring and support are not appropriate for all patients with CS. This can be the case for some patients with milder variants of CS, in whom rates of survival are expected to be high irrespective of use of a pulmonary artery catheter or tMCS [5]. Conversely, certain patients with CS may have a high risk of near-term mortality regardless of the use of these advanced tools. In such cases, these invasive measures may be non-beneficial or even harmful, inflicting discomfort and imposing added risks of procedural complications. In such cases with a high risk of short-term morbidity and mortality, the early engagement of palliative care services can be appropriate, not only to connect appropriate patients with resources to alleviate symptoms, but also to help patients consider goals of care. In this study, a cancer diagnosis was associated with greater utilization of palliative care services. This observation may speak in part to the established importance of palliative care services in medical oncology, which as a field has been a leader in demonstrating benefits of palliative care [39]. Yet the low rates of palliative care utilization identified in our study relative to observed mortality rates—both in patients with and without cancer—highlight a potential opportunity for improvement. It is possible that the rates ascertained underrepresent true allocation of palliative care [40], which can vary across institutions in both mode of delivery and nature of documentation. It is also likely that the utilization of palliative care services in a context of acute life-threatening illness (such as CS) differs in important ways from utilization in a context of chronic fatal disease, with heterogeneity of customs and practices across institutions [41].

In our study, survival to hospital discharge after a diagnosis of CS was nearly 75% in most patients and more than 60% in patients with cancer. These optimistic outcomes are in line with the best observed outcomes of contemporary series of CS [15, 16, 40] and exceed historical averages for survival from CS [14, 42–45]. In interpreting the significance of survival rates in CS, it is of paramount importance to understand the case mix of cardiovascular diagnoses, including proportions of acute myocardial infarction versus heart failure-related CS [46], contributions of ischemic versus non-ischemic cardiomyopathy [47], and CS severity [48]. In this analysis of young adults with CS, the overwhelming majority of cases were attributable to non-acute myocardial infarction related CS, with a low burden of underlying coronary artery disease. The architecture of the NIS, at present, does not readily permit stratification of CS cases according to Society of Cardiovascular Angiography and Intervention (SCAI) severity classification.

Findings of our study must be interpreted with caution given the inherent limitations of retrospective analysis of the NIS, a claims-based database reliant upon ICD-10 and CPT coding and lacking elements of hemodynamic data, laboratory data and care processes (such as use of vasoactive drug therapies) valuable to delineating the diagnosis and severity of CS. The potential for inter-institution variation in reporting and unmeasured confounders preclude inferences of causality. Furthermore, the static representation of data found in an administrative database fails to capture the dynamic nature of CS, impeding insight into intra-hospital trajectories of illness that can influence care decisions, such as patient decision to decline invasive treatment. Interpretation of our findings should be limited to our stated goal of recognizing patterns and differences in care to generate hypotheses for future research.

Conclusion

In this analysis of young adults hospitalized with CS in the United States between 2016 and 2020, a diagnosis of cancer predicted a lower use of invasive hemodynamic monitoring and tMCS, a greater use of palliative care services, and a higher likelihood of in-hospital mortality. Patients with cancer and CS had a lower burden of pre-existing cardiovascular disease and traditional cardiovascular risk factors. Future research is needed to understand factors influencing disparate decision-making in patients with cancer and CS, to develop standardized risk assessment models to offer timely tMCS to well-selected patients with reasonable oncological prognosis, and to identify opportunities to improve cardiovascular survival in this cohort of young cancer patients presenting with CS.

Abbreviations

CS

Cardiogenic shock

tMCS

Temporary mechanical circulatory support

ICD

International Classification of Diseases

CPT

Current Procedural Terminology

PCI

Percutaneous coronary interventions

CABG

Coronary artery bypass graft surgery

IABP

Intraaortic balloon pump

mAFP

Microaxial flow pump

VA-ECMO

Venoarterial extracorporeal membrane oxygenation

CTRCD

Cancer therapy-related cardiovascular disease

Authors’ contributions

SDS contributed with conceptualization, data analysis, as well as writing the first draft of the manuscript. MIT and GS contributed with conceptualization, methodology, and writing of the second draft of the manuscript. VT, MG, ST, PA, MN, KA, BS, SP all contributed to writing the final main manuscript text with figures and tables. All authors reviewed the manuscript prior to submission.

Funding

None.

Data availability

Any data generated is provided within the manuscript or supplementary information files. The data used for the production of this manuscript is readily available as part of the National Nationwide Inpatient Sample (NIS) Database Documentation from the Healthcare Cost & Utilization Project https://hcup-us.ahrq.gov/db/nation/nis/nisdbdocumentation.jsp.

Declarations

Competing interests

The authors declare no competing interests.