Author + information
- Duane S. Pinto, MD, MPH∗ ( and )
- E. Wilson Grandin, MD, MPH, MEd
- Division of Cardiology, Richard A. and Susan F. Smith Center for Cardiovascular Outcomes Research, Beth Israel Deaconess Medical Center, Harvard Medical School, Boston, Massachusetts
- ↵∗Address for correspondence:
Dr. Duane S. Pinto, Division of Cardiology, Interventional Section, Beth Israel Deaconess Medical Center, 185 Pilgrim Road, Boston, Massachusetts 02115.
Cardiogenic shock (CS) complicates acute myocardial infarction (AMI) in 5% to 9% of cases. Despite advances, mortality rates in AMI CS remain at 40% to 50% (1,2). Nearly 2 decades ago, the SHOCK (Should We Emergently Revascularize Occluded Coronaries for Cardiogenic Shock) trial demonstrated a benefit for early revascularization in patients presenting with AMI and CS (3), but subsequent pharmacological and device trial results have been disheartening. Risk stratification at presentation is challenging, and few studies have analyzed risk factors for death in AMI CS. Only 3 risk scores in the era of percutaneous coronary intervention (PCI) exist (2,4,5).
In this issue of the Journal, Pöss et al. (6) contribute significantly to the field by developing a risk score for 30-day mortality in patients with AMI CS. By using clinical features at presentation among patients in the IABP-SHOCK II (Intraaortic Balloon Pump in Cardiogenic Shock II) trial, the score’s performance was validated internally in the trial’s own registry and externally in the European CardShock cohort. Six variables were associated with 30-day mortality: age; prior stroke; creatinine; serum glucose; arterial lactate; and post-PCI TIMI (Thrombolysis In Myocardial Infarction) flow grade <3. A total of 480 patients were ultimately included in the derivation cohort, with 49.0% classified as low risk, 37.7% as intermediate risk, and 13.3% as high risk. The 30-day mortality rates for these patients were 23.8%, 49.2%, and 76.6%, respectively, with similar performance in the 2 validation cohorts.
We applaud Pöss et al. (6) for creating a simple risk score that can be easily derived. For patients with CS that is complicating AMI, therapy must commence rapidly and often with incomplete knowledge of the patient’s medical history. Prior stroke stands out as the only variable that may be challenging to obtain and that lacks an obvious pathophysiological link to early death. Evidence of cerebral hypoperfusion has consistently emerged as a risk factor for death in AMI CS (2,5), and a history of stroke may identify patients who are more vulnerable to decreased cerebral oxygen delivery. Alternatively, prior stroke may correspond to other conditions, such as atrial fibrillation, which could influence the ability to tolerate AMI CS.
Modeling risk is challenging, and variables may be retained because they are consistently obtained, capture multiple domains of risk, or provide orthogonal data to other important risk factors. In this trial, prior stroke status was available for nearly all subjects, thus potentially contributing to its retention over other factors. Future work should probe the relationship between other factors and survival in AMI CS, to search for causal factors that are modifiable and easily obtained.
Although risk scores stratify populations effectively, tradeoffs exist whenever a risk score is developed, and simplicity can come at the expense of accurate application to the individual. Pöss et al. (6) wisely elected to dichotomize continuous variables by using a well-established methodology, the Youden index, to establish optimal discriminatory cutpoints for 30-day survival. Although this approach maximizes the predictive value when dichotomizing a variable, it does imply that a threshold exists for the outcome among these continuous variables.
On the basis of clinical experience and previously published data, none of the continuous variables in the IABP-SHOCK II risk score have a threshold associated with survival. In the TRIUMPH (Tilarginine Acetate in Patients With Acute Myocardial Infarction and Cardiogenic Shock) trial, each 10 ml/min increase in creatinine clearance was associated with a 23% reduction in the odds of 30-day mortality (7), with similar findings from the CardShock registry (2). In the CardShock risk score, every 1 mmol/l increase in serum lactate portended 40% higher odds of in-hospital death (2). Older age has been linked with higher mortality rates in every published study of AMI CS. Each additional year consistently confers approximately 4% to 5% increased odds of death (2,5,8). Admission serum glucose concentration appears to have a U-shaped relationship with mortality rates in CS, with the highest risk in patients with hypoglycemia and severe hyperglycemia (9). Dichotomization of serum glucose obscures the profound risk associated with hypoglycemia and dilutes the gradient of risk observed as serum glucose concentrations become extremely elevated.
While enhancing simplicity, dichotomization of continuous variables in a risk score can lead to overestimation or underestimation of risk, particularly for patients with values hovering around the selected cutpoints. Consider the following 2 hypothetical patients:
Patient A: No prior stroke; TIMI flow = 2 after PCI; age 72 years; glucose 190 mg/dl; creatinine 1.4 mg/dl; lactate 4 mmol/l
Patient B: No prior stroke; TIMI flow = 2 after PCI; age 74 years; glucose 200 mg/dl; creatinine 1.6 mg/dl; lactate 6 mmol/l
Despite similar clinical profiles, the score for patient A is 2 (low risk) and for patient B is 7 (high risk), thereby implying a roughly 50% increased mortality risk for patient B. Dichotomization generates a false sense of “full risk” or “no risk” around these cutpoints when there is probably a complex continuum of risk across the range of these variables.
As Pöss et al. (6) suggested, the greatest utility of the IABP-SHOCK II score may be for informing patient selection for future trials in AMI CS. More granular risk stratification may help to identify the “sweet spot” for therapeutic intervention in AMI CS. Although the IABP-SHOCK II trial failed to show a benefit of IABP use in AMI CS, the overall mortality rate in the trial was lower than expected, and median lactate was only modestly elevated (∼4 mmol/l), thus leading critics to question whether patients were sick enough to derive benefit. Conversely, a trial of mechanically ventilated patients with AMI and severe CS (median lactate ∼8 mmol/l) demonstrated no survival benefit for a percutaneous mechanical circulatory support device (Impella CP, Abiomed, Danvers, Massachusetts) compared with IABP (10). Most patients had cardiac arrest (92%), return of spontaneous circulation was frequently delayed >20 min (48%), and death was predominantly caused by anoxic brain injury (46%); these findings raise concern that many patients might have been too sick to benefit from any intervention.
These studies highlight challenges in assessing the many domains of risk in AMI CS. Pöss et al. (6) did not assess for a differential impact of IABP across the risk profiles, and although such an analysis would be hypothesis generating, stratification could help identify groups to target. Focusing on intermediate- and high-risk patients, as has been done in acute coronary syndromes, might select salvageable patients with sufficient complexity and circulatory derangement to derive benefit from invasive interventions. A post hoc SHOCK trial analysis suggested that early revascularization had the greatest benefit for patients identified as intermediate or high risk according their risk score (5). Additionally, challenges remain among high-risk patients to determine when intervention is futile or when a more aggressive circulatory support platform is necessary. The IABP SHOCK II risk score will be useful to clinicians in guiding early discussions about prognosis, but until further data validating and refining the score become available, we caution against using this score to withhold therapeutic interventions or generalizing the findings to patients who are not undergoing PCI or who have CS from other causes.
Pöss et al. (6) have advanced care for patients with AMI complicated by CS. The score appears to provide reproducible qualitative risk stratification for early death among patients with AMI CS who are undergoing PCI and may help triage for future studies. Striving to reduce the unacceptable mortality rates associated with this challenging clinical syndrome, we must endeavor to avoid the trial and error method favored by Goldilocks when assigning risk to CS patients. We hope the IABP-SHOCK II risk score will help differentiate “too sick” from “too well,” to allow application of novel and beneficial therapeutics to patients who are “just right.”
↵∗ Editorials published in the Journal of the American College of Cardiology reflect the views of the author and do not necessarily represent the views of JACC or the American College of Cardiology.
Both authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- 2017 American College of Cardiology Foundation
- Harjola V.P.,
- Lassus J.,
- Sionis A.,
- et al.
- Pöss J.,
- Köster J.,
- Fuernau G.,
- et al.
- Zeymer U.,
- Vogt A.,
- Zahn R.,
- et al.
- Kataja A.,
- Tarvasmaki T.,
- Lassus J.,
- et al.
- Ouweneel D.M.,
- Eriksen E.,
- Sjauw K.D.,
- et al.