Article Text

Download PDFPDF

Original article
Consistency of benefit from an early invasive strategy after fibrinolysis: a patient-level meta-analysis
  1. Husam Abdel-Qadir1,2,3,
  2. Andrew T Yan1,4,
  3. Mary Tan5,
  4. Francesco Borgia6,
  5. Federico Piscione7,
  6. Carlo Di Mario8,
  7. Sigrun Halvorsen9,
  8. Warren J Cantor1,10,
  9. Cynthia M Westerhout11,
  10. Bruno Scheller12,
  11. Michel R Le May13,
  12. Francisco Fernandez-Aviles14,
  13. Pedro L Sánchez15,
  14. Douglas S Lee1,16,
  15. Shaun G Goodman1,4,5,11
  1. 1Department of Medicine, University of Toronto, Toronto, Ontario, Canada
  2. 2Institute of Health Policy, Management and Evaluation, University of Toronto, Ontario, Canada
  3. 3Division of Cardiology, Women's College Hospital, Toronto, Ontario, Canada
  4. 4Division of Cardiology, Terrence Donnelly Heart Centre, St. Michael's Hospital, Toronto, Ontario, Canada
  5. 5Canadian Heart Research Centre, Toronto, Ontario, Canada
  6. 6Federico II University, Naples, Italy
  7. 7University of Salerno, University Hospital San Giovanni di Dio e Ruggi d'Aragona, Salerno, Italy
  8. 8NIHR Cardiovascular BRU, Royal Brompton Hospital & NHLI Imperial College, London, UK
  9. 9Oslo University Hospital Ulleval, Oslo, Norway
  10. 10Southlake Regional Health Centre, Newmarket, Ontario, Canada
  11. 11Canadian VIGOUR Centre, University of Alberta, Edmonton, Alberta, Canada
  12. 12Innere Medizin III, Universitat des Saarlandes, Homburg, Germany
  13. 13University of Ottawa Heart Institute, Ottawa, Ontario, Canada
  14. 14Hospital General Universitario Gregorio Marañón, Madrid, Spain
  15. 15Hospital Universitario Salamanca-IBSAL, Salamanca, Spain
  16. 16Division of Cardiology, University Health Network, Toronto General Hospital, Toronto, Ontario, Canada
  1. Correspondence to Dr Shaun Goodman, St. Michael's Hospital, Toronto, Ontario, Canada M5B 1W8; goodmans{at}smh.ca

Abstract

Background Randomised controlled trials have demonstrated improved outcomes with an early invasive strategy compared with routine care after fibrinolysis among patients with ST-elevation myocardial infarction. However, it remains uncertain whether specific patient subsets derive differential benefit from an early invasive strategy.

Methods Using patient-level data from seven randomised trials, we studied the relationship between treatment assignment (early invasive vs standard care) and adverse cardiovascular events. The outcomes assessed were death/reinfarction at 30 days and at 1 year, as well as death/reinfarction/recurrent ischaemia, major bleeding and stroke at 30 days. The analyses were conducted in strata (age, sex, diabetes, prior infarction, Killip class, anterior infarction and time from symptom onset to fibrinolysis) to assess for an interaction between the stratifying variable and treatment assigned.

Results There were 101 deaths and 115 recurrent infarctions at 30 days in 3010 patients. There were no strata where an invasive strategy conferred a differential treatment effect. With the exception of a marginally significant interaction between Killip class and treatment for death/reinfarction at 30 days and 1 year (p values for interaction 0.044 and 0.038, respectively), no interactions between the stratifying variables and treatment assignment were observed.

Conclusions Benefit from an early invasive strategy after fibrinolysis for ST-elevation myocardial infarction is similar across patient subgroups stratified by these clinical characteristics. Therefore, prediction of risk and benefit from an early invasive strategy after fibrinolysis for ST-elevation myocardial infarction is best achieved by global risk evaluation rather than specific patient characteristics.

Statistics from Altmetric.com

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Introduction

Randomised controlled trial data have demonstrated that patients treated with fibrinolysis for ST-segment elevation myocardial infarction (STEMI) experience better outcomes with an early invasive strategy, involving urgent transfer to a centre with coronary revascularisation capabilities.1–8 Accordingly, North American and European guidelines for management of STEMI recommend the transfer of successfully reperfused patients to a percutaneous coronary intervention (PCI)-capable centre to facilitate early angiography.9 ,10 However, access to PCI-capable centres may be limited in certain geographic settings, particularly in regions where fibrinolysis remains the appropriate first-line reperfusion therapy.

Consequently, there remains a need to identify which patients are most likely to benefit from an early invasive strategy relative to standard care. This may help triage patients and direct most appropriate use of revascularisation facilities for successfully reperfused patients in situations of limited access. Thus, we performed a subgroup analysis of patients enrolled in randomised controlled trials that compared an early invasive strategy with routine care after fibrinolysis for STEMI. Our aim was to identify whether there were any patient characteristics that predicted differential benefit within strata from universal use of early invasive strategies. We therefore evaluated the homogeneity of treatment effect among prognostically important subgroups based upon age, sex, diabetes, Killip class, MI location (ie, anterior vs non-anterior) and time delay from symptom onset to fibrinolysis.

Methods

Patients

The Optimal Timing for Post-Thrombolysis Elective Revascularization (OTTER) meta-analysis (clinicaltrials.gov identifier NCT01014182) pooled patient-level data from seven randomised trials evaluating early invasive versus standard management (Southwest German Interventional Study in Acute Myocardial Infarction (SIAM III),1 Routine invasive strategy within 24 hours of thrombolysis versus ischaemia-guided conservative approach for acute myocardial infarction with ST-segment elevation (GRACIA-1),2 Combined Angioplasty and Pharmacological Intervention Versus Thrombolysis Alone in Acute Myocardial Infarction (CAPITAL AMI),3 Which Early ST-elevation myocardial infarction Therapy (WEST),4 Combined Abciximab REteplase Stent Study in Acute Myocardial Infarction (CARESS-in-AMI),5 Trial of Routine Angioplasty and Stenting after Fibrinolysis to Enhance Reperfusion in Acute Myocardial Infarction (TRANSFER-AMI)6 and NORwegian study on DIstrict treatment of ST-Elevation Myocardial Infarction (NORDISTEMI)7). The meta-analysis excluded all non-randomised trials, as well as randomised studies in which angioplasty was mainly performed without stenting (defined as use of stents in <80% of study population) and those in which non-fibrin-specific lytic agents were used. The details of the meta-analysis and included studies have been previously described.11

The patients analysed within the meta-analysis are reflective of the inclusion and exclusion criteria of the individual studies, which are well detailed in their individual reports.1–7 Broadly, these were patients with STEMI and an expected delay of >60 min between first medical contact and primary PCI. Common exclusion criteria included advanced non-cardiac organ dysfunction, standard contraindications for fibrinolysis or PCI, as well as high-risk features that preclude randomisation to fibrinolysis (such as cardiogenic shock or severe heart failure). Most studies excluded pregnant women and patients over the age of 75 years.

Outcomes and strata assessed

The primary outcome was a composite of death or recurrent MI within 30 days of the index STEMI. We also assessed the risk of death or recurrent MI at 1 year, as well as the 30-day risk of death, recurrent MI or recurrent ischaemia. Of note, the WEST trial collected data on death to 1 year, but only collected data on recurrent MI to 90-day follow-up.4 To study the association of stratifying variables with the impact of an early invasive approach on adverse effects, we also separately studied the outcomes of in-hospital major bleeding and 30-day stroke. These outcomes were assessed in patient strata based on clinical characteristics that might impact the potential benefit–risk balance from an early invasive strategy relative to routine care: age (>70 years vs ≤70 years), sex, diabetes, prior MI, Killip class (1 vs 2–4), infarct location (anterior vs other territories) and time from symptom onset to fibrinolysis (<2 h, 2–3.9 h, 4–6 h and >6 h).

Statistical analysis

Baseline characteristics and clinical events were assessed among pooled patients according to randomised treatment strategy. Continuous variables are presented as medians, with IQRs; the statistical significance of differences between the treatment arms were assessed using the Wilcoxon rank sum test. Categorical variables are presented as counts (with percentages), and the χ2 test (or Fisher's exact test, where appropriate) was used for the corresponding statistical test. We conducted two analyses using Cox proportional hazards models stratified by trials in which we regressed treatment strategy (early invasive vs routine care) against time to death or recurrent myocardial infarction at (1) 30 days and (2) 1 year. We also used trial-stratified logistic regression to assess the association between treatment strategy and (1) death/MI/ischaemia at 30 days, (2) in-hospital major bleeding and (3) stroke at 30 days. Logistic regression was used for the latter three outcomes due to the unavailability of event times for a substantial number of patients.

Since patients were randomised to the two treatment arms within each trial, we expected that potential confounders were equally balanced between the two groups; thus, other baseline variables were not included in the regression models to maintain parsimony. The statistical significance of an interaction between the stratifying variable and treatment benefit was determined by including an interaction term between treatment assignment and the stratifying variable of interest in the regression model. After completing these analyses, we conducted post hoc analyses that examined rates of cardiac catheterisation in patients with Killip class 1 compared with those in higher Killip classes to explore some of our findings (see ‘Results’ section for details).

All analyses were performed using the SAS software, V.9.3 (SAS Institute, Cary, North Carolina, USA). Statistical significance level was defined by a two-tailed p value <0.05.

Results

There were 3011 patients in the combined database, of which 1502 were randomised to routine care and 1509 to an early invasive strategy. In total, 1059 patients (35.2%) were from the TRANSFER-AMI trial, 598 (19.9%) from the CARESS trial, 500 (16.6%) from GRACIA1, 266 (8.8%) from NORDISTEMI, 221 (7.3%) from WEST, 197 (6.5%) from SIAM III and 170 (5.7%) from CAPITAL-AMI. The median age was 59 years (25th, 75th percentiles 51, 68), and 79.5% were men. The characteristics of the patients in the two treatment arms are summarised in table 1.

Table 1

Baseline characteristics of patients pooled from the seven randomised controlled trials

Univariate analyses demonstrated that patients in the two groups were well matched, with no significant differences in measured baseline variables, with the exception of small differences in heart rate (median value of 72  bpm in the early invasive arm vs 74  bpm in the routine care arm; p=0.036).

Death or recurrent infarction by 30 days

By 30 days, there were 46 deaths among 1508 patients available for follow-up in the early invasive group (3.1%) compared with 55 deaths out of 1502 patients in the routine care arm (3.7%). There were also 42 recurrent infarctions (2.8%) in the early invasive group compared with 73 (4.9%) in the routine care group. The early invasive strategy was associated with a lower risk of death or recurrent infarction at 30 days with an HR of 0.65 (95% CI 0.50 to 0.86, p=0.0024).

The HRs for 30-day death/recurrent MI associated with treatment strategy in the stratified analyses are illustrated in figure 1. There was no statistically significant interaction between treatment assignment and most stratifying characteristics, with the exception of Killip class (p=0.044), such that the HR was 1.08 for patients with Killip class >1 (95% CI 0.61 to 1.89, p=0.79) in contrast to a beneficial effect for patients in Killip class I (HR=0.56, 95% CI 0.40 to 0.78, p<0.001).

Figure 1

The risk of death or recurrent infarction at 30 days, as evaluated by Cox regression. The model was evaluated in the overall sample population and in pre-specified subgroups of interest. The number of patients within each stratum is listed in parentheses. EI%, percentage of patients in early invasive arm with death or MI; LCL, lower confidence limit; p, p value for interaction between stratifying variable and treatment strategy (early invasive vs routine care); reMI, recurrent myocardial infarction; Std%, percentage of patients in standard treatment arm with death or MI; UCL, upper confidence limit.

Death or recurrent infarction by 1 year

By 1 year, there were 74 deaths among 1488 patients available for follow-up in the early invasive group (5.0%) compared with 81 deaths out of 1483 patients in the routine care arm (5.5%). There were also 62 recurrent infarctions (4.2%) in the early invasive group compared with 101 (6.8%) in the routine care group. Overall, the early invasive strategy was associated with a lower risk of death or recurrent infarction at 1 year with an HR of 0.72 (95% CI 0.58 to 0.91, p=0.0047). The HRs at 1 year for death/recurrent MI associated with treatment strategy in the stratified analyses are illustrated in figure 2. The only statistically significant interaction with treatment assignment was observed for Killip class (p=0.038), with an HR that approached unity for patients with Killip class >1 (1.14, 95% CI 0.71 to 1.84, p=0.59).

Figure 2

The risk of death or recurrent infarction at 1 year, as evaluated by Cox regression. The model was evaluated in the overall sample population and in pre-specified subgroups of interest. The number of patients within each stratum is listed in parentheses. Patients from the WEST trial were censored at 90 days (see text for details). EI%, percentage of patients in early invasive arm with death or MI; LCL, lower confidence limit; p, p value for interaction between stratifying variable and treatment strategy (early invasive vs routine care); reMI, recurrent myocardial infarction; Std%, percentage of patients in standard treatment arm with death or MI; UCL, upper confidence limit.

Death, recurrent infarction or recurrent ischaemia by 30 days

By 30 days, there were an additional 68 recurrent ischaemia events among 1508 assessable patients in the early invasive arm (4.5%) and 147 events among 1499 assessable patients (9.8%) in the routine care arm. Thus, there were 136 instances of the composite outcome of death or recurrent infarction/ischaemia in the early invasive arm (9.0%) compared with 245 events in the routine care arm (16.3%), with an OR of 0.49 (95% CI 0.39 to 0.62, p<0.0001) in favour of the early invasive arm. The ORs for 30-day death or recurrent MI/ischaemia associated with treatment strategy in the stratified analyses are listed in table 2. We did not observe any patient characteristics that demonstrated a significant interaction with treatment assignment.

Table 2

Subgroup analysis using logistic regression models assessing the relationship between treatment strategy and the outcome of death, recurrent myocardial infarction or recurrent ischaemia at 30 days

Major bleeding and stroke

There were 139 in-hospital major bleeding events, of which 69 occurred in the early invasive group (4.6%) and 70 (4.7%) occurred in the routine care group. There were also 18 strokes in each of the two treatment groups, corresponding to an incidence of 1.2%. We could not identify any groups that were associated with a differential risk of major bleeding or stroke. The ORs for either outcome that were associated with an early invasive strategy relative to routine care in the evaluated strata are listed in table 3 for major bleeding and in table 4 for stroke.

Table 3

Subgroup analysis using logistic regression models assessing the relationship between treatment strategy and in-hospital major bleeding

Table 4

Subgroup analysis using logistic regression models assessing the relationship between treatment strategy and stroke at 30-day follow-up

Post hoc analysis of coronary angiography rates

Due to the borderline statistically significant interaction between Killip class and benefit from an early invasive strategy, we compared coronary angiography rates among patients in the routine care arm based on their Killip class. We observed that 865 (69.5%) of 1245 patients with Killip class 1 who were randomised to the routine care arm received coronary angiography compared with 133 (56.8%) of 234 patients with higher Killip class.

Discussion

This stratified analysis of patient-level data from seven randomised controlled trials demonstrates that the benefit from an early invasive strategy after fibrinolysis for STEMI was relatively uniform and unaffected by most evaluated patient characteristics. We observed a possible interaction with Killip class, potentially suggesting that only patients presenting without heart failure gain benefit from routine early transfer to a revascularisation-capable hospital. However, this finding was of marginal statistical significance and does not take into account the impact of multiple comparisons. We also assessed major adverse events by studying the interaction of patient characteristics with the risk of major bleeding or stroke, and found that there were no patient groups that were disproportionately harmed by an early invasive strategy.

This stratified patient-level analysis builds upon prior attempts to establish which patients would be most likely to benefit from an early invasive strategy after fibrinolysis for STEMI. A previous meta-analysis of these seven trials explored the relationship between treatment strategy, patient risk profile in each trial (without using individual patient-level data) and clinical outcomes. This demonstrated a trend towards a reduction of death and reinfarction in higher risk patients with the early invasive strategy.8 In contrast, an analysis of the TRANSFER-AMI trial suggested potential harm with an early invasive strategy among patients with a high GRACE12 score, with benefit seen only among those with a low or intermediate GRACE score.13 Finally, a more recent analysis using pooled patient-level data from these seven trials concluded that an early invasive strategy was associated with clear benefit in the low and intermediate risk groups, but not in the high-risk stratum.11

Our study adds to these analyses by establishing that there are no single baseline patient characteristics that influence the likelihood of benefit from an early invasive strategy after fibrinolysis. The totality of the evidence indicates that there are no universal, dichotomous indicators of who is more likely to gain benefit or harm from an early invasive strategy. Thus, patient decisions in this setting have to be made based on an individualised global assessment of expected risk and benefit, paying close attention to which components contributing to patient risk are potentially modifiable (and which are not). Since the benefit from an early invasive strategy has been demonstrated in multiple randomised trials in different geographic settings and time periods, the totality of the evidence is in favour of routine use of an early invasive strategy. Thus, unless evidence of harm is clearly established in particular subgroups, we believe that all eligible patients should be considered for a strategy of routine transfer to a revascularisation-capable centre after fibrinolysis for STEMI. We were able to show that there are no patient characteristics that would be associated with major bleeds or stroke, two of the most dreaded complications among patients undergoing coronary angiography and revascularisation after fibrinolysis.

Previous studies have identified that age, sex, diabetes, anterior infarct location and higher Killip class are powerful prognostic factors in the setting of MI.12 ,14–19 It has also been suggested that women are less likely to benefit from revascularisation in the setting of MI, which may be related to higher rates of bleeding and other revascularisation-related adverse events20–23 or inappropriate perception of risk by treating physicians.24 Thus, these characteristics might be expected to identify patients who would disproportionately benefit from an early invasive strategy. The lack of differential benefit in patients stratified on these characteristics may reflect that the factors leading to worse prognosis are not directly modifiable by revascularisation. Moreover, the subgroup analyses may have been underpowered to detect differences in spite of pooling data from seven randomised clinical trials.

We considered the possibility that the lack of benefit in the expected subgroups may reflect the ‘real-world’ setting in which these trials were performed. It is plausible that a large proportion of patients with diabetes, heart failure, anterior infarction location and prior infarcts who were randomised to the routine strategy would have been more urgently referred for urgent revascularisation in the setting of failed reperfusion because of recognition of their high-risk status. This could theoretically negate the benefit provided by a routine strategy of early transfer. Since we observed a possible interaction between Killip class and benefit from an early invasive strategy, we evaluated this hypothesis in a post hoc analysis of angiography rates in the routine care arm based on Killip class. We did not observe a higher rate of coronary angiography among patients in higher Killip classes, indicating that if a beneficial impact from the two strategies existed, it was due to other explanations. One possibility is that the time to revascularisation was longer in patients with higher Killip class, who would need to be stabilised before undergoing angiography. An alternative explanation is that a higher Killip class is a surrogate for a large degree of myocardial damage after extensive infarction that is not amenable to being salvaged by revascularisation, thus lowering the benefit–risk ratio from an early invasive strategy.

Limitations

This study has several limitations. First, this is a subgroup analysis of patients from several randomised controlled trials, and therefore may be liable to bias. In particular, there is the aforementioned concern about confounding by indication that could have underestimated potential benefit with an early invasive strategy among patients with diabetes and prior myocardial infarction. Second, the exclusive use of randomised trials raises concerns about the external validity of our findings to patients who are typically under-represented in clinical trials. In particular, almost 80% of patients were men and almost 80% were younger than 70 years. This limits the ability to extrapolate our findings to patients over the age of 75 years, which is a growing patient demographic that is often at higher risk of bleeding and stroke in the setting of an invasive strategy. Third, the studies spanned a long period of time when there were numerous advancements in the treatment of acute myocardial infarction, leading to potential heterogeneity between these trials. In particular, the time to coronary angiography and revascularisation varied considerably across the trials. Nevertheless, since our focus is on the randomised comparisons of two management strategies in an intention-to-treat analysis, we did not examine for interactions between timing of revascularisation (a post-randomisation variable) and the subgroups of interest. Furthermore, patients from the WEST trial only had follow-up on recurrent MI to 90 days. In addition, the seven trials used different definitions of recurrent infarction, although we would not expect systematic bias between the two arms within each trial. Moreover, in spite of pooling patient-level data from 3011 individuals from seven trials, our analysis may be underpowered to detect benefit from an early invasive strategy among subgroups. However, our results are consistent with the overall pooled analysis of all seven trials that stratified patients according to a prognostic risk index incorporating multiple clinical characteristics.8 Also, this is the largest study of randomised controlled trial data available, and it is unlikely that further data will emerge given the totality of the evidence indicating a benefit of an early invasive strategy following fibrinolysis.

Conclusions

This analysis of patient-level data from seven randomised controlled trials of care strategies after fibrinolysis for STEMI indicates that there is no single patient-level factor that substantially modifies the benefit from an early invasive strategy. Thus, a global evaluation of risk and benefit taking into account characteristics of the individual patient is necessary. In the absence of strong contraindications, our data suggest that most patients should be considered for an early invasive strategy after receiving fibrinolysis for STEMI.

Key messages

What is already known on this subject?

  • Randomised controlled trials have demonstrated that patients treated with fibrinolysis after ST-elevation myocardial infarction benefit from an ‘early invasive’ strategy of routine transfer to a percutaneous coronary intervention centre.

What might this study add?

  • This study assesses the relative benefit and risk in patient subgroups within seven randomised controlled trials comparing an early invasive strategy with routine care. The subgroups were divided based on age (>70 years), sex, diabetes, heart failure at presentation, anterior infarct location and time delay from symptom onset to fibrinolysis. We observed that treatment benefit and harm did not differ significantly across these subgroups.

How might this impact on clinical practice?

  • There are no discrete patient characteristics that identify differential benefit from an early invasive strategy after fibrinolysis. This suggests that an early invasive strategy should be universally considered after fibrinolysis. Prediction of treatment risk and benefit may be best facilitated by a global risk assessment.

References

Footnotes

  • Contributors HA-Q contributed to data analysis and interpretation, and drafted and revised the manuscript. ATY, MT, FB, FP, CDM, SH, WJC, CMW, BS, MRLM, FF-A, PLS and DSL contributed to data analysis, interpretation and manuscript revision. SG contributed to conception, design, data analysis, interpretation and manuscript revision.

  • Funding HA-Q is is supported by a Fellowship from the Canadian Institutes of Health Research. SGG is supported by the Heart and Stroke Foundation of Ontario in his role as Heart and Stroke Foundation (Polo) Chair at the University of Toronto. The TRANSFER-AMI Study was supported by a grant from the Canadian Institutes of Health Research, and an unrestricted grant from Hoffman La Roche, Canada; stents in TRANSFER-AMI were provided by Abbott Vascular Canada. The CARESS-in-AMI trial was sponsored by the Italian Society of Interventional Cardiology (GISE) via an unrestricted grant provided by Eli Lilly Critical Care Europe, Switzerland. The GRACIA-1 trial (GRupo de Analisis de la Cardiopatia Isquemica Aguda) was supported in part by the Spanish Network for Cardiovascular Research RECAVA, from the Instituto de Salud Carlos III, and from the Spanish Ministry of Science and Innovation; stents in GRACIA-1 were provided by Abbott Vascular Spain. The West trial was supported by unrestricted research grants from Hoffman-La Roche and sanofi aventis Canada and also Eli Lilly Canada. The CAPITAL AMI study was supported by a peer-reviewed grant from the Canadian Institutes of Health Research and a CIHR Industry-Partnered Program with Hoffmann La-Roche Limited, Canada, and Guidant Corporation Canada. NORDISTEMI was funded by grants from the Scientific Board of the Eastern Norway Regional Health Authority, Hamar, Norway; Ada and Hagbarth Waage's Humanitære og Veldedige Stiftelse, Oslo, Norway; and the Innlandet Hospital Trust, Hamar, Norway.

  • Competing interests Consulting fees, honoraria and/or research: ATY, Sanofi Aventis; CDM, Eli Lilly; SH, Astra Zeneca, Eli Lilly, Bristol Myers Squibb, Pfizer, Sanofi, Bayer, Boehringer Ingelheim, MSD; WJC, Roche, Sanofi Aventis; SG: Roche, Sanofi Aventis.

  • Ethics approval Review board approval was obtained by all participating hospitals in all seven trials.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Data sharing statement The principal investigators of the seven trials have shared specific data fields with the Canadian Heart Research Centre, which coordinated the analyses. Questions regarding unpublished data from the study should be directed to the corresponding author.

Linked Articles