Clinical and regulatory landscape for cardiogenic shock: A report from the Cardiac Safety Research Consortium ThinkTank on cardiogenic shock

Cardiogenic shock (CS) is a leading cause of mortality associated with acute myocardial infarction (AMI).^1,2 Improved prehospital emergency care and timely reperfusion from initially systemic lytics to now percutaneous coronary intervention have decreased the incidence of AMI-CS and associated mortality from 1970 through the early 21st century.³ Unfortunately, between 2003 and 2010, there was a possible near doubling in the incidence of AMI-CS in the United States.^4,5 Results from international randomized controlled trials (RCTs) including the Intraaortic Balloon Support for Myocardial Infarction with Cardiogenic Shock (IABP-SHOCK II)⁶ and PCI Strategies in Patients with Acute Myocardial Infarction and Cardiogenic Shock (CULPRIT SHOCK)⁷ show that 30-day mortality remains at 30%–40% and approaches 50% at 1 year.⁸ This is consistent with clinical practice data from a US national registry (the American College of Cardiology CathPCI Registry), which also showed rising mortality in patients with AMI-CS who are managed with invasive, contemporary therapeutics including timely revascularization and percutaneous mechanical circulatory support.⁹

Clinical trials that evaluate drugs, devices, and best practices for CS have been challenging to conduct and slow to enroll. Since the “Early Revascularization in Acute Myocardial Infarction Complicated by Cardiogenic Shock” (SHOCK)¹⁰ trial opened enrollment in 1993, only 2,500 patients in total have been enrolled in prospective randomized trials worldwide; this represents 0.5% of estimated AMI-CS patients from Europe and North America.¹¹ Moreover, it took nearly 40 years from development of the intra-aortic balloon pump (IABP) in the 1960s to the first appropriately sized RCT studying its efficacy. The “Intraaortic Balloon Pump in Cardiogenic Shock” (IABP-SHOCK II) trial demonstrated no survival benefit or improvement in measures of end-organ function.¹²

The prevalence and persistently high mortality with CS have led to widespread use of novel mechanical circulatory support (MCS) devices for the management of CS despite the paucity of appropriately sized, prospective RCTs to support a beneficial effect on outcome. The Cardiac Safety Research Consortium ThinkTank “Defining the Clinical and Regulatory Landscape for Cardiogenic Shock” was therefore convened in September 2018 in Washington, DC, to attempt to address these issues. Physician experts, US and Canadian regulators, and industry leaders met to discuss current clinical best practices, barriers to generating prospective evidence, areas of CS care that have not been studied (Table I), and future directions for approaches to device and drug development evidence for therapies in CS.

Table I.

Unresolved issues in cardiogenic shock

Revascularization

Revascularization strategy (percutaneous vs surgical bypass)

Access site (radial vs femoral]

Timing of complete revascularization (after treatment of culprit lesion at presentation)

Devices

Ventilation strategies

Use of invasive hemodynamics

Mechanical circulatory support (timing, device selection, patient selection)

Hypothermia

Optimal pharmacological therapies

Inotrope regimens

Vasopressor regimens

Antiplatelet therapies

Anticoagulant therapies

Anti-inflammatory agents

Immunomodulatory agents

Systems of care

Timing and frequency of invasive hemodynamic measurement

Target mean arterial blood pressure

Timing of mechanical circulatory support implantation

The role of specialized shock team

Transfusion strategy

Optimal blood glucose levels

Addressing barriers to generating evidence in cardiogenic shock

Importance of a standardized definition of cardiogenic shock

There are currently multiple ongoing prospective studies of device therapies, pharmacologic therapies, and systems of care interventions for patients with a primary condition of “cardiogenic shock.”¹³ Outcomes of CS may relate to a matrix of critical features, including patient characteristics, severity of CS at time of recognition, clinical comorbidities, coronary anatomy, timing of intervention, support technologies, and postprocedural care. Current definitions of CS used in prospective trials and registries have varied and include combinations of clinical, hemodynamic, and metabolic profiles, as depicted in Table II.

Table II.

Varying definitions of cardiogenic shock

	Clinical criteria	Hemodynamic criteria
Califf et al (1994)47	SBP < 90 mm Hg or 30 mm Hg below baseline for at least 30 min	AVO2 difference > 5.5 mL/dL Cardiac Index <2.2 L/min/m2 AND PCWP>15mmHg
(S)MASH(1999)48	SBP < 90 mm Hg despite inotropic support AMI < 48 h prior to presentation	PCWP>15mmHg* Thermodilution Cl < 2.2 L/min/m2*
SHOCK (1999)10	SBP < 90 mmHg for at least 30 min OR Need for supportive measures to maintain SBP >/= 90 mm Hg Signs of end-organ dysfunction Cool extremities UOP <30 mL/h AND HR >/= 60 beat/min	Cardiac index </= 2.2 L/min/m2 AND PCWP >/= 15 mm Hg
TRIUMPH (2007)49	<100 mm Hg despite vasopressor therapy LVEF <40%	Elevated left ventricular filling pressures
IABP SHOCK (2010)12	SBP < 90 mm Hg for at least 30 min OR Need for inotropic/vasopressor to maintain SBP >/= 90 mm Hg	Cl < 2.2 L/min/m2
IABP-SHOCK II (2012)50	SBP < 90 mm Hg for at least 30 min OR Infusion of catecholamines to maintain SBP > 90 mm Hg AND Impaired organ perfusion Altered mental status Cold/clammy skin UOP < 30 mL/h Serum lactate >2.0 mmol/L
CULPRIT-SHOCK (2017)7	SBP < 90 mm Hg for at least 30 min OR Infusion of catecholamines to maintain SBP > 90 mm Hg Impaired organ perfusion Clinical signs of pulmonary congestion Altered mental status Cold and clammy skin/limbs UOP (<30 mL/h) Arterial lactate >2.0 mmol/L

SBP, Systolic blood pressure; AVO2, arteriovenous oxygen; PCWP, pulmonary capillary wedge pressure; UOP, urine output; HR, heart rate.

Heterogeneity of inclusion criteria for CS trials reduces external validation and makes it difficult to generalize results to larger patient populations. As was the case with the Bleeding Academic Research Consortium’s (BARC) definition, development and use of an accepted definition of CS for future cardiovascular clinical research (trials and registries) would likely lead to more consistent, efficient, and interpretable evidence for specific devices and drug interventions and for accrual of knowledge in CS overall. In addition, structured classification of the severity of shock may enhance interpretation of results from different data sets.

Role of speed and triage of shock patients

The ThinkTank participants believe that there is a critical time, referred to as “preshock,” during which a potentially reversible hemodynamic state exists, and if not quickly intervened upon, preshock can rapidly progress to systemic shock with multiorgan failure and nearly universal mortality. The consensus of the Think-Tank participants is that such time dependence makes early recognition of shock states critical to their optimal care and that this time dependence has substantial implications for clinical trials that measure outcomes from investigational interventions. As such, the rapid identification of shock patients should constitute a critical metric for advancing both clinical care and research in therapies for shock.

There was strong consensus that once the hemodynamics of shock are identified, a multidisciplinary approach to stratifying different causes of shock with concomitant triage to surgical, cardiac, or other procedures is also of paramount importance in ultimately defining, or improving, outcomes in these patients.¹⁴ In the case of device interventions, the CSRC ThinkTank felt that the differentiation of non-CS from CS, further followed by the identification of CS due to acute ischemic (non–ST-elevation myocardial infarction [non-STEMI or STEMI), might all be relevant to the feasibility of conducting informative clinical trials.

Delineating populations of intended use and outcomes

Regulatory authorities require device labeling to give a clear and clinically relevant definition of the population of intended use. Typically, this population is derived from the characteristics of patients in whom the therapy has been studied investigationally. To date, the vast majority of research in CS has focused on patients with concurrent AMI or other hemodynamic embarrassment potentially due to a reversible coronary mechanism. There are, however, other cohorts of CS patients with similarly poor outcomes who are not as well studied but for whom MCS may have a role (Table III).¹⁵ Despite a paucity of high-quality data including RCTs supporting their use, MCS devices are regularly deployed in these patients.^16–20 In addition to defining the cohorts for future research, discerning the relationship between a given device and its appropriate duration of use is also necessary. Clinical context should be taken into account, including periprocedural use, shorter-term use as a “bridge,” or longer-term support such as definitive “destination therapy” (Table IV).

Table III.

Populations other than AMI in whom acute MCS may have a role

Acute on chronic left ventricular /biventricular heart failure

Bridge to durable mechanical circulatory support

Bridge to orthotopic heart transplantation

Bridge to bridge

Postcard iotomy

Acute myocarditis

Primary graft dysfunction following orthotopic heart transplantation

Acute pulmonary embolism

Valvular cardiomyopathy

Left ventricular outflow obstruction

Peripartum cardiomyopathy

Refractory arrhythmia

Takotsubo syndrome

Acute on chronic right ventricular failure

Table IV.

Duration of MCS for CS

	Hyperacute	Acute	Temporary	Durable
Time	≤6h	> 6 h, ≤ 14 d	> 14 d, <1 y	Indefinite
Clinical scenarios	Periprocedural High-risk PCI Failure to wean from CPB	Failure to wean from CPB Acute on chronic HF Awaiting recovery Bridge to decision	Acute on chronic HF Awaiting recovery Bridge to decision Bridge to durable VAD or transplant	“Destination therapy” VAD

PCI, Percutaneous coronary intervention; CPB, cardiopulmonary bypass; HF, heart failure; VAD, ventricular assist device.

Following deployment of temporary MCS for CS, approximately one third of patients survive as a result of receiving a durable left ventricular assist device or cardiac transplant.²¹ Unfortunately, strokes remain a common complication of temporary MCS,^22,23 with longer duration of support associated with higher stroke risk.²³

Accordingly, temporary MCS, regardless of indication, should have clearly established goals for improvement prior to device placement. In addition to improving survival, prioritizing outcomes that are important to patients is necessary and has been a goal of international professional societies and regulatory agencies.²⁴

Universal access to research in cardiogenic shock: informed consent in emergency research settings

In the United States, marketing approval of devices such as those for use in CS requires a demonstration of safety and effectiveness, whereby valid scientific evidence indicates both that probable benefits outweigh probable risks and that a significant portion of the target population will garner clinically meaningful results.²⁵ Generating the necessary scientific evidence for devices associated with significant risk is complex. One barrier particularly relevant to CS studies involves the consent of CS patients. Obtaining informed consent in emergency situations is often challenging and can result in delays, potentially making therapies less effective. Federal regulations have allowed for exceptions from standard informed consent requirements for use in “emergency research” scenarios (21 CFR 50.24), which include life-threatening situations when patients are unable to provide consent and obtaining consent from next of kin or legally authorized representatives is not feasible. The community where the research is to occur must agree to the investigational approach, and it is mandatory that subjects have the prospect of receiving direct benefit from the planned intervention.²⁵ The ThinkTank attendees advocate that relevant stakeholders should consider CS research to be a form of “emergency research.” However, under current interpretation of these regulations, because of the heterogeneous presentation, it is clear that not all CS research qualifies for exception of informed consent under 21 CFR 50.24 by default.

Thus, a paradox exists regarding informed consent for devices being studied in CS settings. To use an investigational therapy, investigators must obtain informed consent from the patient. Although there are exceptions in certain situations, the spectrum of CS is such that not all patients may qualify to have consent waived, and thus, therapies potentially may be delayed or withheld. Under these circumstances and compounded by inherent selection bias by providers, trials will continue to enroll poorly and evidence will continue to be slowly generated; this scenario constitutes an important barrier to overcome in approval of newer, safer therapies for CS.

Requirements of informed consent vary by country, further magnifying this barrier while highlighting the potential value of future international collaborations for CS research. Attendees of the ThinkTank widely endorsed the creation and empowerment of an international standard to direct emergency research, as this could substantially aid local institutional review boards with the conflicts involving current guidelines and regulations.

Clinical trial design considerations for cardiogenic shock therapies

Registry-based clinical trials

RCTs form the cornerstone of evidence-based medicine and are required to determine treatment effects of medical interventions.²⁶ Operationally, RCTs are challenging, particularly in identifying patients who both meet inclusion/exclusion criteria and who are willing to consent to a randomized treatment assignment. RCTs in CS have been exceedingly slow and inefficient. More efficient and inclusive approaches to developing high-quality outcomes evidence in CS are a shared priority of providers, regulators, and the device industry.

Logistically, the success of a randomized trial in CS is also highly dependent on the willingness of providers to randomize unstable patients. There may exist a perceived ethical dilemma regarding clinical equipoise for CS. Investigator bias may significantly impact selection of candidates for either trials comparing experimental therapy to standard of care or trials comparing 2 experimental therapies. Lack of equipoise may contribute to poor enrollment, enrollment of lower-risk patients unlikely to benefit from MCS, or significant crossover between study arms.

CSRC discussions of the need for more efficient models of prospective RCTs for CS included the “registry-based RCT” developed in previous CSRC ThinkTanks leading to the SAFE-PCI trial for Women.^27–29 The registry-based RCT uses existing infrastructure and data capture to streamline the process of site activation and to reduce the site-coordinator work. Patients are consented and randomized in real time. Established mechanisms that are already sending demographics, comorbidities, medications, and index hospitalization outcomes to the registry are “dual purposed” to autopopulate case report forms. In SAFE PCI for Women, this reduction in site coordinator work translated into significantly accelerated enrollment for the trial overall, enrolling 1,787 patients in less than 2 years. Furthermore, in addition to conducting a randomized trial, registry data can also be used to describe practice patterns and temporal trends in outcomes, identify outliers, and improve performance measures as a complementary context for outcomes evidence from the actually randomized cohort. Participants at the ThinkTank also acknowledged, however, that registry-based trials of CS devices are also likely to have certain limitations that need to be recognized and addressed for the collected data to allow for proper inferences.

Factorial study designs

Factorial clinical trial designs test the effect of more than 1 treatment and also test potential interactions among treatments. Such designs may accommodate evidence development for both a drug and a device in the same study, as was reported for the Bivalirudin During Primary PCI in Acute Myocardial Infarction (HORIZONS-AMI) trial.³⁰ This trial included a nested factorial design in which 3,602 patients with STEMI were initially randomized prior to angiography either to unfractionated heparin plus GP IIb/IIIa inhibitor or to bivalirudin (+/− provisional GP IIb/IIIa inhibitor). For the 3,006 patients who were to undergo primary PCI, patients were then randomized a second time to receive paclitaxel-eluting TAXUS stent or bare metal stent.

This approach reduces costs and timelines for evidence generation compared to conducting 2 independent clinical trials. Acknowledging the potential for interaction between simultaneous or iterative therapies, this approach could be highly relevant to adding efficiency to very high quality, prospective CS RCT designs. Although a larger sample size might be required, use of a registry-based trial as described above might alleviate enrollment issues historically encountered with previous shock trials.

Statistical methods to study cardiogenic shock

Adaptive clinical trial designs

In this challenging patient population, it is important to consider advancements in clinical trial design. The Thrombectomy 6 to 24 Hours After Stroke with a Mismatch between Deficit and Infarct (DAWN) Trial³¹ in ischemic stroke provides a template for an innovative trial design that provides advantages for some of the specific challenges to carrying out RCTs in the CS population. Briefly, in DAWN, patients with an ischemic stroke (last known to be well 6–24 hours earlier) were randomized to receive late endovascular thrombectomy or standard therapy. After 206 patients were enrolled, the trial was stopped for efficacy, with Bayesian posterior probabilities of >0.999 suggesting strong evidence in favor of late endovascular thrombectomy. There are 2 design features that could be of particular utility in a CS population.

First, this design allows frequent interim analyses without compromising validity of final results. With a trial design based on Bayesian posterior probabilities of success, the DAWN team planned to conduct interim analyses beginning after enrollment of 150 patients and again every 50 patients up to a maximum enrollment of 500 patients. The trial was ultimately stopped at the first interim analysis that allowed stopping for efficacy, after enrolling 206 patients (with Bayesian posterior probability >0.999 for superiority of late endovascular thrombectomy). Had there been a smaller benefit that was not yet conclusive at the interim analysis, the trial would have continued until sufficient data accrued to conclude that (1) the data strongly supported benefit of late endovascular thrombectomy or (2) the trial was very unlikely to demonstrate benefit. Although the specific number of patients required and interim strategy deployed for a trial in CS patients would be dependent on other operating characteristics, the key takeaway is that this design enrolls just as many patients as needed to answer the question but no more. This provides a practical advantage and mitigates ethical concerns, as patient exposure is minimized.

Second, the adaptive-enrichment strategy allows fine tuning the patient population at interim analyses. The DAWN trial prespecified 5 patient subpopulations based on infarct size. At each planned interim analysis, if the highest currently open group had less than 40% probability of demonstrating an average positive treatment effect, enrollment of patients in that group would be suspended, concluding that the experimental treatment was “futile” in that population and that there was nothing to be gained by continuing to enroll those patients, but would remain open in the other groups. This feature was not ultimately triggered in DAWN, as the treatment turned out to be efficacious across all infarct sizes, but it was in place if needed.

Why is this useful in a CS trial? One of the emergent themes from the clinical community is the heterogeneity of clinical presentation in CS patients and concerns about recruiting the right set of patients for a given therapy while avoiding enrollment of patients who may not benefit from the therapy (as brilliantly illustrated in satire by Yeh et al in the PARACHUTE trial³²). An adaptive-enrichment design could be used to determine if a particular therapy (ie, percutaneous mechanical circulatory support device) is beneficial for all CS patients or only for patients with certain prespecified features. Perhaps an agreed-upon classification system (such as Stage I, Stage II, Stage III, Stage IV cardiogenic shock) could be created for this trial, and the trial could be designed in such a way that interim analyses could test for safety/efficacy in the respective subgroups; at the interim analyses, ultimately “closing” enrollment in subgroups that had accrued enough patients to demonstrate clear efficacy or futility while keeping other subgroups open until an answer could be determined for them as well.

More ambitiously, it may be possible to combine several of the principles (large registry of CS centers, factorial designs, and adaptive-enrichment principles) into a single trial. For example, the Randomized, Embedded, Multifactorial Adaptive Platform Trial for Community Acquired Pneumonia (REMAP-CAP; https://clinicaltrials.gov/ct2/show/NCT02735707) is currently enrolling subjects in a trial that evaluates multiple treatment combinations, drops those that are shown to be less effective, and can even add new treatments during the course of the study. This would require much careful planning and investment, but the ultimate payoff could be a functional, dynamic trial that would be embedded into the clinical care of CS patients at the participating centers and could give deeper insight into the most effective treatment combinations for specific profiles of CS patients than a standard parallel-group design.

Delivery of systems-based shock care and reliable reporting of data

Operationalized systems of care, hospital, and physician volumes positively correlate with survival in high-risk patients such as in cardiovascular surgery and primary percutaneous coronary intervention for STEMI.^33–38 As such, professional societies including the American Heart Association, American College of Cardiology, and Society for Cardiac Angiography and Interventions have recommended minimum procedural volumes for hospitals and operators and have supported organization of STEMI networks to direct patients efficiently to facilities capable of providing care.^34,39–41

For CS, a similar volume-to-outcome association has been demonstrated by Shaefi et al.⁴² Observational data from the National Cardiogenic Shock Initiative and retrospective data from the Inova Heart and Vascular Institute show that operationalized care for early detection and management of patients with CS is associated with a reduction in mortality.^43–46 Importantly, these programs have multidisciplinary care teams with high-volume operators available 24 hours a day, 7 days a week. Although limited in scope, these studies serve as justification for a larger initiative, leveraging systems of care already in place, to organize formally and improve care of patients with CS.

Although the attendees of this ThinkTank generally agreed that establishment of “centers of excellence in cardiogenic shock” is paramount to successful CS research, concerns were expressed that the paucity of data on what constitutes “best practices” for the treatment of shock makes it difficult to specify fully the therapeutic capabilities or performance criteria for a “center of excellence” designation. Therefore, consensus focused on defining centers of excellence as a function of the speed and accuracy of shock diagnosis, the correct stratification of etiology, and the implementation of appropriate triage strategies. Such centers have the demonstrated ability to recruit suitable patients on a timely basis and consistently collect high-quality data using agreed metrics. Such centers could be benchmarked by the establishment of fully staffed multidisciplinary “shock teams” with 24/7 availability. Early and accurate identification of patients in shock and stratification of patients with CS could greatly enhance both CS studies’ enrollment and their general data collection. Participation of such centers in both ongoing registries, enhancing real-world evidence data, as well as device, drug, or factorial prospective RCTs, could enhance both national and international efforts to generate evidence for regulatory decisions and best practice recommendations for CS patients.

Conclusions

Cardiogenic shock remains a leading cause of inhospital mortality among patients with AMI. In addition, CS also complicates progressive chronic heart failure with reduced ejection fraction. Mechanical support devices are used in many of these patients without a solid base of randomized trial data. This gap in evidence is partly due to varying definitions of CS, challenges with the early identification and triage of shock and CS patients, difficulties with the conduct of randomized trials in a very sick patient population, and lack of perceived equipoise among clinicians for RCTs. This CSRC ThinkTank outlined areas of potential collaboration and delineated strategies for more efficient and informative evidence generation to further guide care and improve outcomes in this extremely high-risk population.