Mortality in STEMI Patients With Cardiogenic Shock: Results From a Nationwide PCI Registry and Focus on Left Main PCI

Background. The study aims to assess real-life short- and long-term outcomes of patients undergoing primary percutaneous coronary intervention (PCI) for ST-segment elevation myocardial infarction (STEMI) complicated with cardiogenic shock (CS).  Outcome after left main (LM) PCI is of particular interest. Methods. Procedural, 30-day, and >30-day mortality rates were assessed in 2744 CS-STEMI patients enrolled between 2012 and 2019 in a nationwide registry involving 49 centers. Results. Procedural, 30-day, and >30-day mortality rates were 6.9%, 39.8%, and 12.6%, respectively. The mortality rates were significantly higher in the 348 patients (12.7%) who underwent LM-PCI (13.5%, 59.5%, and 18.4%, respectively). LM-PCI, a suboptimal PCI result, and transfemoral access were independent predictors of procedural and 30-day mortality. Operator experience was an independent predictor of procedural mortality, but not 30-day mortality. Conclusions. Mortality remains high in CS-STEMI patients, especially within the first month. Patients undergoing LM-PCI are particularly at risk. Operator experience is predictive of procedural mortality.

J INVASIVE CARDIOL 2022;34(2):E142-E148.

Key words: cardiogenic shock, mortality, myocardial infarction

Cardiogenic shock (CS) occurs in 6%-10% of ST-segment elevation myocardial infarction (STEMI) patients.¹ Urgent revascularization is the main therapeutic strategy that counteracts the downward spiral in CS and lowers mortality.^1,2 Despite advances in medical and interventional treatment, mechanical support, and critical care, in-hospital mortality rates of CS patients have plateaued around 50%.^1,3

If the left main (LM) coronary artery is the infarct-related artery, the situation is particularly critical, as at least 84% of the flow to the left ventricle passes through the LM.⁴ Acute LM occlusion has been reported to occur in only 0.6% of STEMI patients undergoing cardiac catheterization, suggesting that many of these patients die before reaching the cath lab.⁵ The rare early survivors mostly present in CS and in-hospital mortality can be as high as 60% despite primary percutaneous coronary intervention (PCI) and hemodynamic support.^5,6 A subtotal LM occlusion as the culprit lesion is less rare, but in-hospital mortality is similar when patients are in CS.^7–9

Reported series on LM-PCI in CS-STEMI patients remain scarce and mostly reflect the practice in high-volume centers. Analysis of the Nationwide Inpatient Sample database in the United States has shown that mortality is significantly higher in CS patients treated in lower-volume centers.¹⁰ We aimed to evaluate real-world procedural outcome and survival of CS-STEMI patients undergoing primary PCI in general and LM primary PCI in particular based on the Quality-oriented Electronic Registration of Medical Implant Devices (QERMID) registry, which was created in 2012 by the Belgian health authorities.¹¹

Study patients. Between February 1, 2012 and August 29, 2019, all Belgian interventional cardiologists were required to consecutively register baseline and procedural data on every PCI performed in QERMID to obtain reimbursement. Data on 194,369 PCIs in 158,792 patients were captured after obtaining informed consent. In 2020, the Belgian authorities granted access and allowed analysis of the anonymized source data from QERMID and individual survival data from the Belgian national database.¹¹ The authors have conformed to institutional guidelines and those of the American Physiological Society.

A total of 2744 CS patients were enrolled following primary PCI for STEMI. Patients resuscitated for out-of-hospital cardiac arrest or who had previously undergone coronary artery bypass grafting (CABG) were excluded.

Study centers. The number of Belgian centers authorized to perform PCI increased from 31 in 2012 to 49 in 2015. On-site cardiac surgery was not available at 21 centers, including all 18 newly added centers.¹¹

Baseline and procedural characteristics/study definitions. The following baseline characteristics were available in QERMID:  gender, age, body mass index (BMI), diabetic status, access site, and prior thrombolysis (rescue PCI). Mandatory reported procedural characteristics were diseased vessels, treated vessels, post-PCI stenosis, and Thrombolysis in Myocardial Infarction (TIMI) flow. If LM-PCI was performed, it was considered a 2-vessel PCI. PCIs that comprised only the LM, or the LM along with either a proximal segment of the left anterior descending (LAD) artery or left circumflex (LCx) artery were classified as non-complex. All other LM-PCIs were classified as complex. LM equivalent (LMEQ) PCI was defined as PCI of both the proximal LAD and LCx. Complete revascularization was determined when all diseased sites were treated by PCI during the index hospitalization. Optimal PCI result was determined when all treated segments had TIMI 3 flow and a residual diameter stenosis of <20%.

Outcomes. We assessed procedural (PCI result and procedural mortality), short-term (CABG and mortality within 30 days), and long-term (mortality beyond 30 days) outcomes in relation to baseline and procedural characteristics.

Statistical analysis. The statistical analysis was performed using IBM SPSS Statistics for Windows, version 27.0 (IBM Corporation) at the Ghent University. An unpaired t-test was used to compare normally distributed continuous variables, while comparison of discrete values was done with the Chi-square test. The level of statistical significance was set at P=.05. Multivariate analysis of the same variables was done with a binary logistic regression with forward conditional selection. To assess variables related to survival beyond 30 days, a univariate survival analysis was performed with a Kaplan-Meier estimate and right censoring assumed to be non-informative. The multivariate survival analysis was performed with Cox regression with forward conditional selection.

Baseline characteristics are presented in Table 1. The vessels treated with PCI are shown in Figure 1; 67.8% of patients underwent 1-vessel PCI, 25.9% underwent 2-vessel PCI, and 6.3% underwent 3-vessel PCI. Complete revascularization was achieved in 61.5% of patients overall, with 100% achieved in 1-vessel disease patients but in only 34.8% of patients with multivessel disease (48.9% in 2-vessel disease and 19% in 3-vessel disease patients; P<.001).

LM-PCI was performed in 348 patients (12.7%). However, of these, only 218 patients (62.6%) had underlying LM disease, while 130 patients (37.4%) did not (Figure 2). In the latter group, it can be assumed that extension of stenting to the LM was considered necessary to adequately treat proximal disease in a daughter vessel. Conversely, of the 286 patients with LM disease, 218 patients (76.2%) underwent LM-PCI, while the remaining 68 patients (23.8%) only had non-LM PCI. While the reasons for this also remain speculative, it can be envisioned that some LM lesions were considered non-culprit or hemodynamically insignificant.

Transfemoral access (TFA) was used in 65.8%, transradial access (TRA) was used in 33.8%, and brachial access was used in 0.4% of patients. TFA was used more in LM-PCI and LMEQ-PCI than in non-LM/LMEQ-PCI (71.6% vs 69.9% vs 65.1%, respectively). TRA was used in only 18.9% of patients in 2012, but accounted for 28.8% of cases in 2015 and 48.5% of cases in 2019 (P<.001). The trend of increasing usage was consistent among the LM-PCI, LMEQ-PCI, and non-LM/LMEQ-PCI subgroups.

In 55.4% of all patients and 58.3% of patients undergoing LM-PCI, PCI was performed with drug-eluting stents only. Drug-eluting stent usage increased from 28.3% in 2012 to 51.2% in 2015 and 82.5% in 2019. In 173 patients (6.3%), the primary PCI was a rescue PCI following failed thrombolysis.

Of the 2744 patients, 2175 (79.3%) underwent the primary PCI at a center with on-site heart surgery. Surgical centers performed an average of 149.9 ± 51.5 primary PCIs per year, while centers without cardiac surgery performed an average of 71.8 ± 51.5 (P<.001).

The procedural, short-term, and long-term outcome results are presented in Table 2.

PCI results. An optimal PCI result was achieved in 73.6% of PCIs overall, but less often in LM-PCIs (66.4%) and LMEQ-PCIs (65.2%) vs non-LM/LMEQ-PCIs (75.1%; P<.001). It was more frequent in younger patients (76.4% if ≤75 years vs 67.8% if >75 years; P<.001), men (74.9% in men vs 70.7% in women; P=.02), TRA (77.0% with TRA vs 71.9% with TFA; P<.001), higher annual primary PCI rate per center and per operator (P<.001 for both), and higher annual LM-PCI rate per center and per operator (P=.05 and P<.01, respectively).

Procedural mortality. Procedural mortality occurred in 6.9% of patients and varied over time (4.2% in 2012, 6.6% in 2013, 6.3% in 2014, 6.9% in 2015, 6.1% in 2016, 4.7% in 2017, 11.7% in 2018, and 9.6% in 2019). It was approximately double in patients undergoing LM-PCIs and LMEQ-PCIs compared with those undergoing non-LM and non-LM/LMEQ-PCIs (13.5% and 11.3% vs 5.9% and 5.7%, respectively; P<.001). Presence of LM disease (independent of LM-PCI) was associated with procedural mortality (13.5% for LM disease vs 6.2% for non-LM disease; P<.001).

Procedural mortality was higher for patients with multivessel PCI compared with 1-vessel PCI (11.6% for 3-vessel PCI, 9.2% for 2-vessel PCI, 5.6% for 1-vessel PCI; P<.001) and was highest (14.3%) for patients who had PCI of both the LM and right coronary artery (RCA). Among multivessel disease patients, incomplete revascularization was not associated with procedural mortality.

Age >75 years was associated with procedural mortality. There was a trend toward higher mortality in female and diabetic patients. Within the subgroup of LM-PCI, age, gender, and diabetic status were not associated with significantly different mortality rates.

TFA and a suboptimal PCI result were predictors of procedural mortality in a univariate analysis. Procedural mortality was higher in low-volume centers (13.0% for <36 primary PCIs/year vs 6.6% for ±36 primary PCIs/year; P<.01) and for low-volume operators (12.1% for <12 primary PCIs/year vs 6.4% for ±36 primary PCIs/year; P<.001). If the primary PCI involved the LM or LMEQ, procedural mortality was much lower if the PCI was performed by an operator performing >52 primary PCIs/year (11.4% vs 28.6%, respectively; P<.01). Overall, procedural mortality was also higher in centers without on-site cardiac surgery (10.3% vs 6.0% in centers with on-site cardiac surgery; P<.01). This was mainly driven by higher mortality rates in 2017 and 2018 (10.3% vs 2.7% in 2017 [P<.01] and 17.5% vs 9.6% in 2018 [P=.03]).

Thirty-day mortality.  Thirty-day mortality (Figure 3) occurred in 39.8% of patients and did not change significantly over time. The mere presence of disease in the LM was associated with higher 30-day mortality. Of the 348 patients who underwent LM-PCI, 207 (59.5%) died within the first 30 days. This 59.5% 30-day mortality rate was comparable to that following LMEQ-PCI (53.0%; P=non-significant) but higher than after non-LM and non-LM/LMEQ-PCI (37.0% and 36.2%, respectively; P<.001). Multivessel PCI was also associated with 30-day mortality. PCI from both LM and RCA had a mortality of 66.1% at 30 days. In patients with multivessel disease, completeness of revascularization did not affect 30-day mortality rates.

Thirty-day mortality was higher in patients >75 years old and in female patients, and there was a trend toward higher mortality in diabetic patients. TFA and a suboptimal PCI result were associated with higher 30-day mortality in a univariate analysis (P<.001 for both). Within the LM/LMEQ-PCI subgroup, use of TRA (39.1% vs 65.9% with TFA; P<.001) and an optimal PCI result (52.9% vs 67.5% if non-optimal; P<.01) were both associated with lower 30-day mortality. Center and operator experience in primary PCI and availability of on-site cardiac surgery were not associated with 30-day mortality.

Mortality beyond 30 days. As 89% of deaths occurred within the first month, we separately assessed mortality beyond the first 30 days among the 1651 30-day survivors. The mean follow-up was 71.9 ± 0.7 months. Mortality beyond 30 days occurred in 207 (12.6%) of the 30-day survivors. The mere presence of LM disease remained associated with higher mortality (17.4% for LM disease vs. 9.9% for non-LM disease; P<.01). Of the 141 patients alive at 1 month post LM-PCI, 26 (18.4%) died during follow-up, representing a higher mortality rate compared with patients after LMEQ-PCI (n = 4/54 [7.4%]; P<.001), and non-LM/LMEQ-PCI (n = 177/1455 [12.2%]; P<.001).

Mortality was also higher in patients who had 2- or 3-vessel PCI than in patients after 1-vessel PCI (15.8% and 16.5% vs 11.3%; P=.02) and was 26.3% in patients who had both LM and RCA-PCI. Mortality remained higher in those >75 years old (21.2% vs 9.7% in those ≤75 years old; P<.001) and in diabetics (21.5% vs 10.3% in non-diabetics; P<.001), while gender and BMI did not impact mortality. Access site, final stent result, and volume of primary PCIs per center and per operator were no longer associated with survival.

Univariate and multivariate predictors of mortality are presented in Figure 4.

Analysis of the QERMID data, along with survival data from the national database, allowed us to assess invasive treatment and short- and long-term survival of 2744 unselected STEMI patients with CS. A total of 348 patients (12.7%) underwent LM-PCI, making our series one of the largest on LM-PCI in STEMI with CS.

Overall, procedural mortality was 6.9%. Thirty-day mortality was approximately 40% and represented 89% of the total observed mortality. It did not change significantly over time. This number represents the real-life outcome of CS-STEMI patients undergoing primary PCI in a Western European country with a large and easily accessible network of primary PCI-capable hospitals. Early revascularization by primary PCI is recommended in all STEMI patients with suitable coronary anatomy and patients in CS benefit most when performed as soon as possible.^1,2,12,13 Shortening time from symptom onset to PCI, or thrombolysis when PCI cannot be performed immediately, may reduce the occurrence of CS and lower mortality if CS has already developed.^1,3,14 However, given the multitude of primary PCI-capable hospitals in Belgium, expertise varies between centers.¹¹ Shaefi et al showed an association between lower expertise in primary PCI in CS and higher in-hospital mortality.¹⁰ In our cohort, lower expertise was independently predictive of procedural mortality, but not 30-day mortality. The same was true for unavailability of on-site surgery. As non-surgical centers performed fewer primary PCIs per year, their higher procedural mortality may be partly explained by lower experience. Although procedural mortality was always higher in centers without heart surgery, the difference was only significant in 2017 and 2018, shortly after these new centers opened. Primary PCI expertise and patient management may have improved since then.

In STEMI patients without CS, complete revascularization is superior to culprit-only PCI in reducing cardiovascular death or myocardial infarction.^13,15 However, when patients present in CS, the most recent guidelines recommend against routine revascularization of non-culprit lesions during index procedure following the results of the CULPRIT-SHOCK trial.^3,13,16 In QERMID, only 34.8% of multivessel disease patients were completely revascularized at the end of hospitalization and incomplete revascularization was not an independent predictor of higher procedural and 30-day mortality.

LM-PCI patients had a significantly higher procedural (13.5%) and 30-day mortality rate (59.5%) compared with non-LM PCI patients (37.0%). The 30-day mortality rate is comparable to the reported rates in meta-analyses by Vis et al and Yeoh et al (57% and 66.7%, respectively).^7,17 In contrast to our study, these studies also included out-of-hospital cardiac arrest survivors. Importantly, patients who underwent LM-PCI and survived the first month continued to have a significantly higher mortality compared with patients who had non-LM/LMEQ PCI.

The high mortality after LM primary PCI in CS may be explained by the extent of myocardium with impaired or threatened perfusion, but the procedure itself is particularly delicate and requires mastery of complex techniques and imaging to achieve the best results. Not all interventional cardiologists have LM-PCI experience. Even at the end of the registry, about 30% of operators did not perform at least 1 LM-PCI per year, and those who performed LM-PCI did so on average only 5 times per year.¹¹ We found that procedural mortality after LM primary PCI was much lower if the operator had extensive experience with primary PCI. Overall, primary PCI resulted in TIMI 3 flow and <20% residual stenosis in 73.6% of our patients, but if PCI of the LM was performed, this number dropped significantly to 66.4%. A suboptimal result was an independent predictor of procedural and 30-day mortality, as already indicated in the SHOCK trial.¹⁸

TFA was a strong and independent predictor of higher procedural and 30-day mortality, confirming previous findings.¹⁹ Although TRA was overall used in 1 in 3 CS-STEMI patients, it increased to 1 in 2 patients at the end. TFA may have been used more frequently in the most critically ill patients (eg, patients with weak pulses where rapid TRA was unlikely and patients in whom the operator was considering subsequent placement of a hemodynamic support device).

After a patient survived the first month, age >75 years old, insulin-dependent diabetes mellitus, and presence of multivessel disease were predictive of mortality, but procedural aspects, such as vascular access route and treatment of LM, were no longer predictive of mortality. This may be related to the fact that most deaths occur early and that procedural aspects are particularly important to the short-term outcome.

Study limitations. The analysis of the QERMID registry has important limitations. Details on coronary anatomy (eg, culprit lesion, degree of stenosis or occlusion before primary PCI, SYNTAX score), left ventricular function, procedural information (stent technique, contrast use, and hemodynamic support), and comorbidities were not systematically reported. Completeness of revascularization after 30 days was not assessed because data on CABG during follow-up were not available. Systematic monitoring and core lab assessment were not performed and may have reduced the number of patients with true CS. However, given the population size, we assume a limited effect of possible errors on our conclusions.

Mortality following primary PCI remains high in CS-STEMI patients, especially in the first month where it approaches 40%, but also beyond 30 days. LM-PCI was an independent predictor of procedural and 30-day mortality, and patients undergoing LM-PCI had approximately 60% 30-day mortality. TFA and a suboptimal PCI result were independent predictors of procedural and 30-day mortality. Patients treated by experienced operators had a lower risk of dying during the procedure. The current high mortality rate underscores the need to develop more efficient preventive and therapeutic strategies for CS and can serve as a historical reference to evaluate their effect.

• Predictors of Hemodynamic Response to Intra-Aortic Balloon Pump Therapy in Patients With Acute Decompensated Heart Failure and Cardiogenic Shock
• Initiation of Extracorporeal Membrane Oxygenation in the Cardiac Catheterization Laboratory: The Mayo Clinic Experience
• Percutaneous Mitral Valve Repair With MitraClip in Inoperable Patients With Severe Mitral Regurgitation Complicated by Cardiogenic Shock
• Right Atrial, Right Ventricular, Superior Vena Cava Dissection Caused by Swan-Ganz Catheter Placement