Outcomes of Patients With Cancer With Myocardial Infarction-Associated Cardiogenic Shock Managed With Mechanical Circulatory Support

Background Cardiogenic shock (CS) is the leading cause of death among patients with acute myocardial infarction (AMI) and is managed with temporary mechanical circulatory support (tMCS) in advanced cases. Patients with cancer are at high risk of AMI and CS. However, outcomes of patients with cancer and AMI–CS managed with tMCS have not been rigorously studied. Methods Adult patients with AMI–CS managed with tMCS from 2006 to 2018 with and without cancer were identified using the National Inpatient Sample. Propensity score matching (PSM) was performed for variables associated with cancer. Primary outcome was in-hospital death, and secondary outcomes were major bleeding and thrombotic complications. Results After PSM, 1287 patients with cancer were matched with 12,870 patients without cancer. There was an increasing temporal trend for prevalence of cancer among patients admitted with AMI–CS managed with tMCS (Ptrend < .001). After PSM, there was no difference in in-hospital death (odds ratio [OR], 1.00; 95% CI, 0.88-1.13) or thrombotic complications (OR, 1.10; 95% CI, 0.91-1.34) between patients with and without cancer. Patients with cancer had a higher risk of major bleeding (OR, 1.29; 95% CI, 1.15-1.46). Conclusions Among patients with AMI–CS managed with tMCS, cancer is becoming increasingly frequent and associated with increased risk of major bleeding, although there was no difference in in-hospital death. Further studies are needed to further characterize outcomes, and inclusion of patients with cancer in trials of tMCS is needed.


Introduction
Cardiogenic shock (CS) is a common complication of acute myocardial infarction (AMI), occurring in approximately 4% to 12% of patients, and is associated with significant morbidity and mortality. 1,2atients with cancer are a growing population at high risk for cardiovascular events due to improvements in cancer-specific outcomes, shared pathophysiology, cancer therapy-associated cardiotoxicity, and accelerated atherosclerosis. 3Unsurprisingly, there has been a corresponding increase in the prevalence of AMI and AMI-CS in this cohort. 4,5These patients have worse outcomes and lower rates of invasive interventions such as percutaneous coronary intervention (PCI) and coronary angiography. 6,7There are several potential reasons for reduced use of PCI among patients with cancer including anemia, thrombocytopenia, concern for future bleeding risk of dual antiplatelet, and a concern for poor overall prognosis. 8emporary mechanical circulatory support (tMCS), including intraaortic balloon pump (IABP), peripheral left ventricular assist device (pLVAD), and extracorporeal membrane oxygenation (ECMO), is often used to support patients with CS as a bridge to recovery, decision, or advanced therapies (durable MCS or heart transplantation). 9,10Patients with cancer and CS pose a clinical conundrum given their paradoxically increased risk of bleeding and thrombosis, both of which are common complications of tMCS.Cardiologists may also be hesitant to manage patients with cancer with invasive procedures given unknown or perceived poor long-term oncologic prognosis, particularly in patients with metastatic disease. 11Additionally, patients with AMI-CS are often managed with tMCS as a bridge to recovery, which suggests a possible utility of managing patients with limited long-term survival. 12Given the paucity of studies examining outcomes associated with the use of tMCS in cancer patients, we sought to investigate outcomes and trends of these patients admitted with AMI-CS and managed with tMCS using a large, nationwide database. 13Additionally, we sought to identify factors associated with tMCS use among patients with cancer and AMI-CS in order to understand nationwide trends in patient selection for tMCS in this patient population.

Study design and population
Hospitalizations for patients with AMI and CS who were managed with tMCS were identified using the National Inpatient Sample (NIS).
The NIS is the largest inpatient database in the United States, captures approximately 20% of all inpatient admissions, and is part of the Healthcare Cost and Utilization Project.Data in the NIS are derived from administrative billing data submitted by hospitals to statewide data organizations and contain demographic and clinical characteristics.The NIS used International Classifications of Diseases, 9th edition (ICD-9) until September 2015 and subsequently ICD-10.This study was deemed exempt by the New York University Grossman School of Medicine Institutional Review Board given that the NIS is a publicly available and deidentified database.
All hospitalizations with a primary diagnosis of AMI, secondary diagnosis of CS, and any procedure code for tMCS between January 1, 2006 and December 31, 2018 were included.Patients with active cancer were identified using ICD-9 and ICD-10 codes. 14,15Procedures, including IABP, pLVAD, ECMO, left heart catheterization, PCI, vasopressor use, mechanical ventilation, and coronary artery bypass grafting (CABG) were captured using ICD-9 and ICD-10 procedure codes.Comorbidities were captured via ICD-9 and ICD-10 codes and Elixhauser comorbidities. 16The ICD-9 and ICD-10 codes used for this study are listed in Supplemental Tables S1 and S2.

Outcomes
In-hospital outcomes were evaluated for patients with cancer and compared to patients without cancer.Our primary outcome was inhospital death.Secondary outcomes of interest included major bleeding composite (procedure-related bleeding, intracranial bleeding, gastrointestinal bleeding, or transfusion of blood products), thrombotic complication composite (stroke, arterial thromboembolism, or venous thromboembolism [VTE] 17 ), sepsis or catheter-related infections, and their components.Outcomes were abstracted using ICD-9 and ICD-10 codes (Supplemental Table S1).

Statistical analysis
Admissions with and without cancer were compared, and standardized mean difference (SMD) was calculated for variables before and after PSM.Continuous variables are presented using the mean and SD, and categorical variables are presented as counts and percentages.Imbalances between groups were considered to be significant if the SMD for a given covariable was !0.10.A propensity score (the predicted probability of cancer) was calculated using a nonparsimonious multivariable logistic regression.We included age, sex, race, prior myocardial infarction, prior PCI, prior CABG, heart failure, atrial fibrillation, anemia, chronic lung disease, smoking, diabetes, obesity, hypertension, liver disease, peripheral vascular disease, chronic kidney disease (CKD), prior stroke, prior VTE, thrombocytopenia, ST-segment myocardial infarction (STEMI) presentation, cardiac arrest, mechanical complication, tMCS type, PCI, CABG, invasive hemodynamic monitoring, mechanical ventilation, vasopressor use, palliative care consult, do-not-resuscitate (DNR) status, year of admission, and insurance type as covariables.We used a greedy algorithm to propensity score match 1 patient with cancer to 10 patients without cancer using 0.1 SD caliper width.All analyses between patients with cancer and no cancer were performed using a propensity score matched (PSM) cohort.Patients with cancer and no cancer were compared using logistic regression analysis with results presented as the odds ratio (OR) with 95% CI.An analysis of cancer vs no cancer was also performed and stratified by type of cancer (solid and hematologic cancer).
Data in the NIS can be weighted to provide national estimates of the entire United States hospitalized population using the Agency for Healthcare Research and Quality sampling and weighting method.Unweighted counts were used for all statistical analyses with the exception of trend assessments, for which national weighted estimates were used.Temporal trends in number of weighted AMI-CS admissions managed with tMCS with cancer and outcomes were examined using the Mann-Kendall trend test.
To identify risk factors for in-hospital death, major bleeding, and thrombotic complications in patients with cancer who were hospitalized with AMI-CS and managed with tMCS, we compared the characteristics of patients with cancer who had these events with those who did not.Characteristics that differed between groups (P <.15) were included in a multivariable logistic regression, with age, sex, and race being covariables in all models.
To identify factors associated with tMCS use among admissions with cancer and AMI-CS, characteristics of those managed with tMCS were compared with those not managed with MCS.Characteristics that differed between groups (P < .15)were included in a multivariable logistic regression, with age, sex, and race as covariables.
Analyses were conducted using SPSS version 29.0 (IBM) and Stata version 15 (STATA Corp).A 2-tailed P value of <.05 was considered significant.
After stratifying by cancer type, admissions with solid cancer and hematologic cancer similarly were associated with increased rates of major bleeding, procedure-related bleeding, and transfusion, without significant difference in in-hospital death, thrombotic complications, sepsis, or catheter-related infections compared with those without cancer (Supplemental Table S5).However, after adjusting for variables that differed (SMD !0.10) between hematologic and noncancer admissions (management with CABG, invasive hemodynamic monitoring, and insurance status), patients with hematologic cancer did not have an increased rate of VTE (adjusted OR, 1.03; 95% CI, 0.63-1.66).Conversely, admissions with solid cancer did have an increased rate of VTE (OR, 1.62; 95% CI, 1.17-2.stratification by cancer type and PSM is summarized in Supplemental Table S6.

Trends in cancer patients admitted for AMI-CS and managed with MCS and types of MCS use
From 2006 to 2018, the number of weighted admissions for AMI-CS managed with tMCS increased from 13,466 to 19,135 among admissions without cancer and 312 to 625 among those with cancer (P trend < .001for both).The proportion of admissions with cancer increased from 2.27% in 2006 to 3.16% in 2018 (P trend < .001).Among admissions with cancer with AMI-CS and managed with MCS, the proportion of admissions managed with IABP decreased from 100% in 2006 to 72.8% in 2018 (P trend < .001).Admissions managed with pLVAD increased from 0% in 2006 to 32.0% in 2018 (P trend < .001).Similarly, admissions managed with ECMO increased from 0% in 2006 to 4.0% in 2018 (P trend ¼ .001).Temporal trends of number of admissions admitted with AMI-CS managed with MCS, proportion of admissions with cancer, and types of MCS used in patients with cancer are shown in Figure 3.

Predictors of tMCS use among admissions with cancer
Among admissions with AMI-CS and cancer, those managed with tMCS were younger (mean 69.5 vs 73.3 years, P < .001)and less likely to be female (28.2% vs 45.7%, P < .001).Admissions managed with MCS were also less likely to have solid cancer, metastatic disease, nonluminal gastrointestinal cancer, palliative care consult, .024 Atrial fibrillation ( .006 Chronic lung disease ( DNR status, and Medicare or Medicaid insurance (Table 3).After multivariable logistic regression, age, female sex, nonluminal gastrointestinal cancer, chronic lung disease, peripheral vascular disease, CKD, prior stroke, and DNR status were associated with decreased MCS use.Mechanical complication, STEMI, PCI, CABG, invasive hemodynamic monitoring, and mechanical ventilation were associated with increased MCS use (Table 4).

Discussion
Cardiovascular disease, including AMI and AMI-CS, are increasingly recognized complications of cancer and cancer therapy.Patients with AMI-CS are often managed with tMCS, including patients with cancer.Our study suggests that among admissions with AMI-CS managed with tMCS, cancer was not associated with higher risk of in-hospital mortality.However, cancer was associated with increased risk of major bleeding, procedure-related bleeding, transfusion, and VTE.Additionally, our study identified potential risk factors for in-hospital death (STEMI, non-IABP MCS, cardiac arrest), major bleeding (upper gastrointestinal cancer, peripheral vascular disease, ECMO, VTE), and thrombotic complications (brain primary tumor or metastasis) among patients with cancer and AMI-CS managed with MCS.Our study also demonstrated an increasing prevalence of patients with AMI-CS managed with MCS with cancer, stressing the importance of investigating outcomes and risk factors in this patient population (Central Illustration).We also examined patient factors associated with tMCS use in patients with cancer, and our findings suggest that age, female sex, DNR status, and nonluminal gastrointestinal cancer were associated with decreased use of tMCS and STEMI, while mechanical complications and revascularization were associated with increased use.

Central Illustration.
In-hospital outcomes and trends among patients with cancer with AMI-CS managed with tMCS.(A) Among patients admitted for AMI-CS managed with MCS, patients with cancer were associated with increased risk of major bleeding but not in-hospital death or thrombotic complications.After stratifying for solid cancer (B) and hematologic cancer (C), patients with cancer were associated with increased risk of major bleeding but not death, thrombotic complications, sepsis, or catheter-related infections.(D) There was and increased temporal trend of prevalence of patients with cancer among patients admitted for AMI-CS managed with MCS.(E) Among patients with cancer, there was a temporal trend of decreased IABP use and increase in pLVAD and ECMO use.AMI, acute myocardial infarction; CS, cardiogenic shock; ECMO, extracorporeal membrane oxygenation; IABP, intra-aortic balloon pump; MCS, mechanical circulatory support; OR, odds ratio; tMCS, temporary mechanical circulatory support.
9][20][21] However, the most common cause of CS in patients with cancer is AMI, and data on outcomes of patients with cancer with AMI-CS managed with MCS are sparse. 22ur study findings suggest that patients with cancer have similar in-hospital mortality compared with patients without cancer.Nonetheless, a substantial proportion of patients with cancer in this analysis had metastatic disease (21.6%).This suggests that cancer, even metastatic cancer, should not be considered an absolute contraindication for MCS in AMI-CS.Additionally, infection is a feared complication among patients with cancer given immunosuppression due to cancer-related therapy, immune and bone marrow dysfunction due to cancer itself, and indwelling catheters.However, among patients with AMI-CS managed with tMCS, cancer was not associated with increased risk of sepsis or catheter-related infections.
In our study, cancer was associated with increased risk of major bleeding including postprocedural bleeding and transfusion, consistent with prior studies of patients managed with percutaneous MCS 23 and durable LVAD. 249][30][31] Further investigation is needed to mitigate bleeding risk in this patient population.
Our study identified potential risk factors for bleeding complications among patients with cancer and AMI-CS managed with MCS.Upper gastrointestinal cancers were associated with increased risk of bleeding in our cohort.This is consistent with prior studies suggesting luminal gastrointestinal cancers are associated with increased risk of bleeding among patients managed with MCS and receiving anticoagulation. 23,32dditionally, ECMO and pLVAD were associated with increased bleeding compared with IABP.This has also previously been demonstrated and may be due to increased platelet shearing from device-related hemolysis and larger bore vascular access. 33,34Peripheral arterial disease was another risk factor for bleeding among patients with cancer that has also been noted among the general population managed with MCS. 357][38] In our study, there was no association between cancer and thrombotic complications, aside from an increased risk of VTE.Additionally, our study identified potential risk factors for thrombosis.Cancer-specific risk factors for thrombotic complications in our study were the presence of primary brain tumor or metastasis.Patients with primary brain tumors have been noted to be at high risk of VTE due to increased production of tissue factor, a potent procoagulant.Prior studies have noted a prevalence of VTE of upwards of 30% among patients with primary brain tumors and 20% in patients with brain metastasis. 39Interestingly, our study did not find an association between brain tumors and intracranial bleeding.This is in line with prior studies suggesting that anticoagulation in patients with brain tumors does not increase risk of intracranial hemorrhage. 40ur study examined factors associated with tMCS use in patients with cancer, and our findings suggest that age, female sex, peripheral vascular disease, CKD, prior stroke, DNR status, and nonluminal gastrointestinal cancer were associated with lower tMCS use.Prior studies in the general population have also suggested that older patients and female patients with CS (both AMI-CS and non-AMI-CS) are less likely to be managed with tMCS. 41,42Similarly, peripheral vascular disease and CKD have been associated with lower rates of tMCS use and may be related to increased vascular complications and in-hospital mortality. 43The association between nonluminal gastrointestinal cancer (including pancreatic and hepatobiliary) may be due to increased short-term mortality in this population compared to other cancer types, though further investigation is needed. 44

Limitations and considerations
This study has several limitations to consider.The first limitation is the retrospective, cross-sectional nature of our study, which makes it prone to residual, unmeasured confounding.The data in the NIS are abstracted from ICD-9 and ICD-10 billing codes, which are prone to misclassification.Data on cancer activity, treatment, laboratory values including blood counts, duration of disease, and genetic testing are not reported and may affect outcomes in this patient population.Additionally, patients with and without cancer were selected by clinicians to be managed with tMCS, and thus, there may be selection and treatment bias.The NIS does not provide the temporal relationship of diagnoses, therefore thrombotic and bleeding outcomes in our cohort may have occurred prior to hospitalization or implantation of MCS.Further, the NIS is restricted to the hospitalization episode, so no postdischarge data is available.Granular data on details of revascularization (ie, disease severity and vessels revascularized) and antithrombotic management are not reported in the database.Additionally, we did not correct for multiple comparisons, and therefore, there is an increased risk of type I error.

Conclusions
Among patients with AMI-CS managed with tMCS, there was no association between cancer and in-hospital mortality after propensity matching.Cancer was associated with increased risk of major bleeding and VTE.Our study suggests that among patients with cancer, MCS may provide benefit, with risks that are similar to the general population.Therefore, clinicians managing AMI-CS should not consider cancer a contraindication for management with tMCS even in the presence of metastatic disease.However, clinicians should be aware of increased thrombotic risk among patients with cancer with brain tumors, obesity, and peripheral vascular disease and of increased bleeding risk among patients with upper gastrointestinal cancer.Our results are hypothesisgenerating given the retrospective nature of the study.Future prospective studies and trials of tMCS in AMI-CS should include patients with cancer in order to confirm our findings and better delineate risks and benefits of tMCS in this patient population.

Declaration of competing interest
Eric Yang reports research funding from CSL Behring, Boehringer Ingelheim/Eli Lilly, Bristol Myers Squibb, and consulting fees from Pfizer and Edwards Lifesciences.The remaining authors declare no potential competing interests.

Funding sources
This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

Figure 1 .
Figure 1.Outcomes of patients hospitalized for acute myocardial infarction and cardiogenic shock (AMI-CS) managed with mechanical circulatory support (MCS) with vs without cancer.Patient outcomes after propensity score matching of patients with vs without cancer hospitalized for AMI-CS managed with MCS.Logistic regression modeling used to estimate odds ratio (OR) and 95% CI of outcomes in patients with cancer compared with patients without cancer.ATE, aortic thromboembolism; VTE, venous thromboembolism.

Figure 3 .
Figure 3. Temporal trends of patients with cancer and AMI-CS managed with MCS.(A) Bar graph depicting number of admissions of AMI-CS managed with MCS with and without cancer.(B) There has been a temporal increase in the proportion of patients with AMI-CS managed with MCS with cancer.(C) Among patients with cancer, there has a been a temporal trend in decreased IABP use and increased pLVAD and ECMO use.AMI, acute myocardial infarction; CS, cardiogenic shock; ECMO, extracorporeal membrane oxygenation; IABP, intra-aortic balloon pump; MCS, mechanical circulatory support; pLVAD, peripheral left ventricular assist device; tMCS, temporary mechanical circulatory support.

Table 1 .
Baseline and hospitalization characteristics after propensity score matching.

Table 2 .
Comparison of unweighted admissions with death, bleeding, or thrombosis with patients without complications.

Table 3 .
Characteristics of cancer admissions with AMI-CS managed with vs without MCS.

Table 4 .
Predictors of MCS use among admissions with cancer and AMI-CS.