METHODS

Search strategy

Published studies

With the assistance of an expert scientific librarian, we developed search strategies and applied them to the electronic databases MEDLINE and EMBASE. All searches were run on September 10, 2020, and included all RCTs since 1999. The search strategy included Medical Subject Headings (MeSH) for MEDLINE and keywords related to CS and AMI. An overview of the complete search strategy is available upon request.

Study selection

Eligibility criteria for inclusion in this review required that studies were on human subjects conducted in adults with CS and were full-text articles written in English. We included studies without limitation on sample size. We excluded reviews, commentaries, editorials, letters to the editor, studies where only the abstract was available, case reports, case series, and studies of CS which were not prospective, randomized, controlled trials (RCTs). Furthermore, we excluded studies, in which CS patients were a subgroup of a larger study. References were checked from the included studies and included if relevant.

Each study abstract was assessed independently by two investigators (JJ and JG). We used the software package Covidence, which is an online tool for systematic reviews (www.covidence.org). The software allows upload and evaluation of searches of abstracts and full texts by each investigator blinded to the other evaluator’s decision. Disagreements were marked for later evaluation. We resolved such cases using a third reviewer (MF). After screening abstracts, all the studies that met the inclusion criteria went through full-text screening by JJ and JG.

Data extraction

We then extracted data from each study using a standardized, pilot-tested form. Extracted data included study characteristics (study title, authors, year of publication, study design, number of included patients, inclusion and exclusion criteria, details of the intervention, and endpoints). Inclusion criteria were grouped into basic hemodynamics (heart rate, blood pressure), advanced hemodynamics/cardiac function (left ventricular ejection fraction [LVEF], cardiac index, pulmonic-capillary wedge pressure [PCWP], central venous pressure [CVP]), signs of end-organ hypoperfusion, and biomarkers (lactate). In cases of multiple publications of the same trial, the original publication would be given priority. If data seemed to be missing from a trial, we tried to acquire the data through correspondence with the trial authors. We contacted the corresponding authors of nine trials for missing data, of which five answered [5, 10, 15–17]. Data extraction was conducted by two reviewers (JJ and JG) and independently checked by one further reviewer (MF). Following the extraction of data, careful consideration was given to the suitability of conducting a meta-analysis. As the trials were too heterogeneous, the data were synthesized qualitatively.

Ongoing and future trials

A search for the overall term “Cardiogenic shock” was run on ClinicalTrials.gov on September 10, 2020. A total of 19 trials studying CS are currently registered, of which 8 are relevant for this systematic review. Data was extracted similarly to the method described above and presented in the supplementary.

Ethics

The study was conducted in accordance with the national and institutional ethical guidelines.

Statistical analysis

The results are presented as percentages of the total number of published AMI-related CS RCTs.

RESULTS

The search strategy identified 513 studies and another three were identified through other sources. Of these, 73 were duplicates. After screening of titles and abstracts, 480 studies were excluded for not meeting our inclusion criteria. Two authors (JG and JJ) reviewed the full text of 35 studies, identifying 19 RCTs for inclusion in this review. Details are provided in a Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) flow diagram shown in Figure 1.

Study characteristics

Seven trials investigated mechanical circulatory assist devices such as an intra-aortic balloon pump, TandemHeart®(LivaNova, London, UK) and Impella® (Abiomed, Danvers, MA, USA) [16, 18–22], 8 trials investigated pharmacological treatments [15, 17, 23–28], 3 investigated coronary interventions (PCI) [10, 13, 29], and 1 trial investigated targeted temperature management [16]. The trials were published from 1999 onwards. Two trials included more than 500 patients [11, 13]. Another two trials included 250–500 patients [10, 15]. However, the majority of trials were small and included less than 100 patients. A total of 2674 unique individual patients were randomized in the reviewed trials. Follow-up periods, as well as endpoints of the trials, differed markedly, with mortality being the most common primary endpoint or part hereof in 8 (42%) trials. Study characteristics, sample size, publication year, endpoints, and intervention investigated, are summarized in Table 1.

Definition of cardiogenic shock

A persisting systolic blood pressure (SBP) value <90 mm Hg and/or vasopressor requirements were the most frequent inclusion criteria and were used in 13 (68%) trials [10, 11, 13, 17–20, 22–24, 26, 29, 30]. Three (16%) trials used an SBP value <100 mm Hg [15, 25, 27] and 1 (5.3%) trial used SBP <80 mm Hg [21]. No trials used mean arterial or diastolic blood pressure, while 3 trials included heart rate [10, 17, 25]. A total of 12 (63%) trials required signs of impaired end-organ perfusion [10, 11, 15, 17, 19–23, 26, 30], with clinical assessment being a central part of which 10 trials required at least 1 of several possible symptoms; 11 trials (58%) included low urine output defined either as a urine output <30 ml/h or more unspecific as “oliguria” [10, 11, 13, 15, 19–23, 26, 30]. Altered mental status was used in 6 (32%) trials [11, 13, 22, 23, 26, 30] and clinically assessed cold and/or clammy skin and limbs was used in 12 trials (63%) [10, 11, 13, 15, 17, 19–23, 26, 30]. Elevated arterial lactate as a biochemical sign of hypoperfusion was used as an inclusion criterion in 5 (26%) trials [11, 13, 22, 23, 30]. A value of >2 mmol/l was used in all 5 trials. In total 10 (53%) trials mandated signs of circulatory impairment or reduced cardiac function [10, 15, 19, 20, 22, 24–27, 29], while the remaining trials did not use any criteria of hypoperfusion for inclusion. Low cardiac index (9 [47%]) and reduced LVEF (3 [16%]) were used as estimates of cardiac function. Cardiac index of <2.2 l/min/m2 was the most frequent limit for inclusion, however one trial used a limit of <1.8 l/min/m2 [24] and another trial used <2.5 l/min/m2 [26]. Pulmonary capillary wedge pressure (PCWP) was used in 9 (47%) trials [10, 15, 20, 22–24, 26, 27, 29], and a cut-off value of 15 mm Hg was used in 6 (32%) trials whereas 18 mm Hg was used in 3 (16%) trials [23, 24, 26]. One trial had a pragmatic selection process, which included STEMI with PCI within 24 hours, inotropic drugs, and an intra-aortic balloon pump [16]. Only 6 (32%) trials mandated both criteria for impaired cardiac function/low cardiac index and reduced end-organ perfusion [10, 15, 19, 20, 22, 26]. Overview of inclusion and exclusion criteria for each trial is presented in Table 2.

Resuscitated out-of-hospital cardiac arrest in cardiogenic shock trials

A total of 10 (53%) trials included patients resuscitated from a cardiac arrest [10, 11, 13, 17, 18, 20, 22, 24, 26, 29], 6 (32%) trials did not report whereas only 3 (16%) trials specifically excluded cardiac arrest patients [16, 23, 30]. Of the trials including cardiac arrest, the proportion of cardiac arrest patients were large ranging from 28%–91% [10, 18].

Outcomes

Eight (42%) trials used mortality as endpoints either as all-cause mortality rate alone [10, 11, 15, 18, 24, 26] or a combined endpoint where mortality was part of other variables [13, 17]. Six (32%) trials used measures of cardiac function at follow-up as the primary endpoint [16, 20, 21, 28–30], with cardiac index and cardiac power index being most frequent. Two (11%) trials used mean arterial blood pressure as an endpoint [22, 27]. One (5.3%) trial used the change in left ventricular preload, measured as the PCWP [23]. One trial used NT-proB-type Natriuretic Peptide [25] and 1 trial used the change in Acute Physiologic Assessment and Chronic Health Evaluation (APACHE)-score [19].

Ongoing trials

The CS definitions used in future trials show a similar heterogeneity as the already published trials (Supplementary material, Tables S1–S2).

DISCUSSION

We described and summarized the definitions, inclusion criteria, and outcomes used in RCTs of CS. Furthermore, we assessed differences and similarities of trials regarding inclusion criteria, which relates to how researchers define CS. We report that inconsistencies of inclusion criteria across different trials exist. Furthermore, a similar pattern is seen in ongoing trials. This could increase the heterogeneity of trial populations between studies, thus making interstudy comparisons, including meta-analyses, difficult in addition to limiting external validity. This and a low level of evidence for CS treatment may partly explain the wide variation of care delivered to CS patients [30]. It would be advantageous to have uniform criteria for inclusion in future CS trials as well as having comparable outcomes.

Blood pressure

The most frequent inclusion criterion identified in this review was low systolic blood pressure, followed by different signs of peripheral hypoperfusion and reduced cardiac function (Figure 2).

It is remarkable that blood pressure is the most frequent inclusion criterion in CS trials, since blood pressure is not part of the definition of CS [7, 8] in contemporary recommendations. On the contrary, patients can have hypoperfusion and reduced cardiac output without hypotension [32]. However, hypoperfusion is often associated with low blood pressure and blood pressure is an easily obtainable parameter that can be measured by non-physicians already prehospitally without the need for invasive catheters. Furthermore, low blood pressure has been shown to be a prognostic factor in CS, with lower blood pressure being associated with worse outcome [9]. Consequently, the Acute Cardiovascular Care Association’s (ACVC) position statement for AMI-related CS has included hypotension (systolic blood pressure <90 mm Hg for >30 min or use of vasopressors to maintain pressure >90 mm Hg during systole) as one of the criteria for CS [8]. It seems reasonable to use blood pressure as an inclusion criterion in CS trials, but research is needed to assess whether systolic, diastolic, or mean arterial blood pressure best reflects cardiovascular function. Low diastolic blood pressure can potentially be harmful and compromise coronary perfusion even with adequate systolic pressure [33].

End-organ hypoperfusion

Assessment of cardiac output and cardiac function in the CS-definition

Per definition, true CS represents an extreme degree of acute heart failure whereby cardiac output is insufficient to meet basic metabolic requirements. These organ-level perturbations may occur at different levels of cardiac output, which varies from individual to individual [37]. As such, no universal cardiac output cut-off can be used to define CS from a theoretical point of view. Furthermore, it has been proven difficult to relate low cardiac output with poor outcomes in critically ill patients [40, 41]. In this review, eight trials used cardiac index to include patients, and values varied from <1.8 l/min/m2 [24] to <2.5 l/min/m2 [26]. When including patients with hyperacute, severe illness in trials it may, however, be difficult to obtain a valid measure of cardiac output prior to randomization if an invasive procedure is needed [40].

An alternative measure of cardiac performance is an echocardiographic assessment. Three (16%) trials in this review used LVEF to include patients. Two trials used LVEF <40% [15, 25] and one trial used LVEF <30% [20]. Echocardiography has several advantages, being easily accessible, if trained personnel are present, and not having the adverse effects of an invasive procedure. Echocardiography should be performed in all patients with suspected CS and AMI for rapid assessment of differential diagnoses and rule-out of mechanical complications [8]. LV dysfunction is a marker of CS, whereas small heart cavities with normal function may suggest hypovolemic shock. It may, therefore, be problematic to consider a patient for a CS trial without echocardiographic evidence of reduced cardiac function. However, the optimal cut-off value of LVEF is unknown.

There are several limitations to LVEF and cardiac index as estimates of cardiac function. In dilated hearts, stroke volume and cardiac output may be normal despite reduced LVEF, and further LVEF is often unprecise, especially when estimating LVEF in bedridden, intubated patients. Only 6 (32%) trials had defined requirements for both inadequate tissue perfusion and low cardiac output/impaired cardiac function. Lastly, CS can be caused by predominantly RV failure, which has previously been shown to be the primary cause of CS of approximately 7% of the patients [41]. In such cases echocardiographic assessment of RV function plays a key role, as the typical sign of LV failure usually is absent.

Pulmonary capillary wedge pressure

The cardiac index can be reduced because of insufficient preload when filling pressures are low despite preserved cardiac function. Thus, hypovolemia is a differential diagnosis to CS. Therefore, adequate filling of the failing ventricle is required for the diagnosis of CS. If PCWP is low, CS is unlikely unless caused by right heart failure [42]. When PCWP is elevated, it is an estimate of backward LV failure, thus giving an index of poor LV function. In this review, 9 trials used elevated PCWP (>15 mm Hg or >18 mm Hg) as an inclusion criterion, and 1 trial used low/normal PCWP (<18 mm Hg) as an exclusion criterion. A PCWP >15 mm Hg in the supine position is used by guidelines as a diagnostic for left-sided heart failure. This corresponds to a value of 2 mm Hg higher than the 98th percentile of the healthy [43]. Yet, PCWP in heart failure has a poor negative predictive value (52%), because PCWP is often elevated only during exercise in chronic heart failure [44]. In this context, it is meaningful to use low PCWP to exclude CS. However, high PCWP is not necessarily associated with CS and is a highly invasive technique.

Outcomes

The outcomes of patients included in trials as well as the outcomes definitions and endpoints differ significantly between trials. Most frequently, all-cause mortality was used, which is a robust endpoint but limited to larger trials with sufficient power. Since CS patients are rare compared to for example myocardial infarction patients, it is to be expected that some trials use other outcome measures than mortality, which enables smaller sample sizes. Underlining the point, that CS trials lacks comparability, it is striking that multiple different hemodynamic endpoints have been chosen in different trials. Cardiac index, cardiac power index, as well as blood pressure, change in ventricular preload, PCWP, and biomarkers have all been used as endpoints. A discussion and consensus among experts regarding relevant endpoints should be published for the guidance of future trialists. So far no discussion of outcomes has been made in published consensus documents [8].

What to do with CS patients with cardiac arrest?

CS frequently occur together with cardiac arrest due to shared etiology, being acute or chronic myocardial injury. Comatose patients successfully resuscitated from a cardiac arrest often require vasopressors to maintain blood pressure, have decreased urine output (acute kidney injury), have cold skin/extremities (targeted temperature management), have markedly elevated lactate levels due to cessation of circulation, and have reduced myocardial function with low cardiac output and reduced LVEF (myocardial stunning) [38, 45]. Therefore, a large proportion of resuscitated cardiac arrest patients fulfill the criteria used in CS trials [12] and this may explain the large proportion of cardiac arrest patients in the CS trials, which we found in this review.

However, low blood pressure in cardiac arrest is often a consequence of vasoplegia due to systemic inflammation, and not always a consequence of low cardiac output [46]. Cardiac arrest and CS also often differ regarding sequelae (i.e., anoxic brain injury vs multisystem organ failure) and anoxic brain injury is unlikely to improve with cardiovascular therapies.

Another important issue to consider when including cardiac arrest-patient in CS trials, is whether the patient is expected to recover neurologically. Therefore, cardiac arrest patients with early signs of poor prognosis, such as a long time to ROSC or non-shockable primary rhythm, should be considered excluded for CS trials [47].

Where to go from here?

Future CS trials should use uniform inclusion criteria based on consensus by a relevant international group of experts in the field, including relevant measures of acute LV and RV dysfunction and tissue hypoperfusion and particularly consensus of whether to include patients resuscitated from cardiac arrest, would be of value. Furthermore, a relevant measure to exclude patients in hypovolemic shock should be determined. Assessing PCWP with an invasive approach is one possibility, however, research in the acute setting cannot wait for the insertion of a pulmonary artery catheter, which is time-consuming and has not been shown to improve outcomes in shock [48]. Signs of pulmonary congestion, such as rales or a chest X-ray with pulmonary edema could be alternatives, as well as echocardiographic measures of elevated LV preload. Tissue hypoperfusion with at least one of the following: altered mental status, cold, clammy skin, or low urine output combined with increased arterial lactate (possibly >2.5 mmol/l). Low blood pressure can be used to raise suspicion of CS, but the optimal cut-off has not been determined. Specific cut-off values should be discussed internationally, and consensus guidelines should be published in order to increase comparability of future CS trials. The number of cardiac arrest patients included should be reported, and subgroup analyses on this group performed.

In septic shock, which is equally difficult to define, a data-driven approach has been undertaken and a possible septic shock definition with clinical criteria was generated through meetings, Delphi processes, and voting, followed by feedback from international professional societies [49]. Inspired by the current septic shock definition, a more pragmatic approach in CS RCTs could leave out invasive signs of reduced cardiac function and include all patients with AMI and signs of hypoperfusion (vasopressor requirements and elevated lactate). In this case, it is assumed that AMI-patients with shock most frequently have cardiogenic shock, but this approach will likely result in the inclusion of non-cardiogenic shock patients. However, a brief and focused transthoracic echocardiographic examination could quickly rule in or rule out a cardiac etiology and should be mandatory in every case where acute heart failure is suspected.

CONCLUSIONS

In conclusion, significant heterogeneity of inclusion criteria exists between CS trials. Differences are mainly related to objective measures of cardiac function such as LVEF or cardiac index and whether to include comatose patients after cardiac arrest. Uniform inclusion criteria in the future would be beneficial for interstudy comparisons, and we suggest an international consensus of overall CS enrolment criteria for future trials.

Supplementary material

Supplementary material is available at https://journals.viamedica.pl/kardiologia_polska.

Article information

Conflict of interest: AP receives grant support from the Novo Nordisk Foundation and Pfizer. JEM receives grant support from Abiomed and speakers fee Orion Pharma and Abiomed. The remaining authors report no conflicts of interest.

Funding: The study was funded by The Danish Heart Foundation (18-R125-A8472-22085) and the Lundbeck Foundation (R186-2015-2132). None of the funders had any influence on any aspects of the current study.

Open access: This article is available in open access under Creative Common Attribution-Non-Commercial-No Derivatives 4.0 International (CC BY-NC-ND 4.0) license, allowing to download articles and share them with others as long as they credit the authors and the publisher, but without permission to change them in any way or use them commercially. For commercial use, please contact the journal office at kardiologiapolska@ptkardio.pl.

How to cite: Josiassen J, Frydland M, Hassager C, et al. Randomized clinical trials of patients with acute myocardial infarction-related cardiogenic shock: a systematic review of used cardiogenic shock definitions and outcomes. Kardiol Pol. 2021; 79(9): 1003–1015, doi: 10.33963/KP.a2021.0072.

REFERENCES