Objectives Primary percutaneous coronary intervention (P-PCI) is the preferred reperfusion option in ST-elevation myocardial infarction, but its benefits become attenuated as time to its potential delivery becomes prolonged. Based on the STrategic Reperfusion Early After Myocardial Infarction trial, we assessed the impact of increasing time delay on outcomes in patients randomised to a pharmacoinvasive strategy (PI) or P-PCI.
Methods Thirty-day clinical outcomes were examined according to PCI-related delay (P-RD). Data from hospitals that enrolled >10 randomised patients were used and P-RD categorised as ≤55 min, >55–97 min and >97 min.
Results Composite of death/congestive heart failure/cardiogenic shock/myocardial infarction in PI and P-PCI arms occurred in 10.6% versus 10.3% (≤55 min, p=0.910); 13.9% versus 17.9% (>55–97 min, p=0.148) and 13.5% versus 16.2% (>97 min, p=0.470), respectively. While there was no worsening of outcomes for PI across the P-RD spectrum, this occurred in the P-PCI arm (p(trend)=0.038). For P-RD ≤55 min, fewer events tended to occur with P-PCI than PI. Conversely, as P-RD increased to >55 min, PI-assigned patients had better outcomes than P-PCI, suggesting an event-free advantage with PI as P-RD increased (p(interaction)=0.094). Analysing P-RD continuously showed that for every 10-min increment there was an increasing trend towards benefit among PI-assigned patients (p(interaction)=0.073).
Conclusions As P-RD increased, PI outcomes became superior to P-PCI when P-RD is prolonged and exceeds guideline-mandated times. In such circumstances, a PI strategy may provide an alternative reperfusion option. Adverse time delays for delivery of P-PCI should be considered when evaluating reperfusion strategies.
Trial registration number NCT00623623.
Statistics from Altmetric.com
Primary percutaneous coronary intervention (P-PCI) is the preferred option for ST-elevation myocardial infarction (STEMI) when delivered in a timely manner in an expert 24/7 facility. The original studies supporting this view however compared P-PCI with in-hospital fibrinolysis only.1 ,2 An important subsequent development has been pharmacoinvasive (PI) strategy (early fibrinolysis with rescue PCI for fibrinolytic failures)3 and with subsequent early angiography/PCI following lytic success.4 Some data support PI as being equal or better than P-PCI.5–8 Other studies indicate that there is a direct link between infarct size9 and that mortality rates increase10 the longer it takes to deliver P-PCI. By inference any advantage of P-PCI over fibrinolysis may become attenuated the greater the PCI-related delay (P-RD). Thus, PI may be especially relevant to global reperfusion strategies globally where geography or logistic delays (rural geography or large urban areas with traffic congestion) challenge logistic systems to provide timely P-PCI. Furthermore, since significant patients worldwide do not present to PCI-capable hospitals, early fibrinolysis with appropriate follow-on intervention (PI) may provide an alternative.
The previously published STrategic Reperfusion Early After Myocardial infarction (STREAM) trial11 compared PI with P-PCI in patients with STEMI presenting early after symptom onset. In brief 1892 patients presenting within 3 h after symptom onset, unable to undergo P-PCI within 1 h, were randomised to PI or P-PCI. A nominal but non-significant absolute difference in 30-day outcome of 1.9% favouring PI therapy was demonstrated.
In a prespecified analysis, this study examines the relationship between the 30-day primary outcome and the timeliness of delivery of the two strategies. The analyses were undertaken to provide insights into the timely delivery of reperfusion for a worldwide setting in situations where P-PCI delivery may be challenging and determine whether a PI strategy is particularly desirable with longer time delays to P-PCI. Such a specific time-related prospective analysis has not been previously reported.
The STREAM trial11 compared outcomes in patients treated with PI therapy or P-PCI. The current study cohort comprised 1653 patients at 38 sites. In order to assess timing metrics, without the confounder of small number bias, we chose to use data from the parent study from sites with >10 randomised patients to adequately reflect site-level delay between initiation of tenecteplase (TNK) or treatment with P-PCI. The number of patients per site ranged from a minimum of 11 to a maximum of 334. A sensitivity analysis to confirm the estimation of site-level delay was also undertaken in those sites randomising >20 patients. The P-RD was defined as the difference between the initiation of P-PCI (median time of sheath insertion) and the initiation of PI (median time of initiation of TNK). The time of sheath insertion for P-PCI was chosen as the most reliable indicator of initiation of the (PCI) reperfusion strategy as previously reported in the parent study. For clinical relevance, patients were classified into three categories of P-RD, aiming to distribute them evenly. However, given the site-level nature of P-RD, the first category contained 35.3% of the cohort, the second containing 43% and the third containing the remaining 21.7% of patients.
The relationship between P-RD and assigned study treatment with the primary composite endpoint at 30 days, which included all-cause death, congestive heart failure (CHF), cardiogenic shock and reinfarction (and the individual components), was assessed. The secondary 30-day endpoint of combined shock or CHF was also examined. Safety endpoints include 30-day ischaemic stroke, intracranial haemorrhage (ICH) and non-ICH major systemic bleeding.
Categorical variables are reported as percentages and as median (25th, 75th centiles) for continuous variables. Differences between groups were tested with χ2 test (or Fisher's exact test, as appropriate) and Wilcoxon rank-sum or Kruskal–Wallis test, respectively. Across categories of P-RD, the Cochran–Armitage test for trend was applied.
P-RD was a site-level variable and was considered in two prespecified manners: (i) three categories of P-RD (ie, ≤55 min, >55–97 min, >97 min) and (ii) continuous. When P-RD was examined as a continuous variable in relation to outcomes, the shape and strength of the association was assessed by using restricted cubic spline functions and 5 knots.
The relative associations with the primary endpoint were assessed via Poisson regression with robust error variance, and relative risk (RR) estimates with corresponding 95% CIs are reported. The interaction between randomised study treatment and P-RD was tested on all outcomes.
Statistical analyses were based on the intention-to-treat principle, and the level of statistical significance was established at p<0.05 and no adjustments for multiple comparisons were made. All analyses were performed with SAS (V.9.3; Cary, North Carolina, USA).
Table 1 shows the population demographics for both the original trial and the current study. Results indicate the current cohort was representative of the original population. Of this cohort 83% were randomised in the ambulance and the rest at the community hospital.
The median age of the patients in the studied cohort was 59 (51–69) years (22% female). There were no differences in infarct site, nor incidence of recognised risk factors between the two groups. The median time delay from the onset of symptoms to first medical contact and to randomisation was similar in the two groups. The median times between symptom onset and reperfusion therapy were 100 min (TNK) and 180 min (P-PCI, p<0.001).
Patient characteristics according to site-level P-RD (classified as ≤55 min (19 sites), >55–97 min (11 sites) and >97 min (8 sites)) and the randomised treatment are shown in table 2. The median time between symptom onset and randomisation (according to increasing P-RD) was 80 min (56–115), 93 min (71–132), and 114 min (76–153), respectively (p<0.001). Importantly, symptom onset to first medical contact and randomisation increased substantially as P-RD increased in each of the randomised group. Especially noteworthy was that time to commencement of reperfusion therapy after randomisation remained short for PI patients but doubled for P-PCI patients at sites with >97 min of delay.
The angiographic findings, metrics of reperfusion and in-hospital revascularisation showed no differences (table 3). However, significantly more patients in the PI than P-PCI groups had Thrombolysis In Myocardial Infarction (TIMI) 3 flow on arrival in the catheterisation lab irrespective of P-RD; after PCI, the percentages were similar.
Associations between PCI-related delay and clinical outcomes
The 30-day primary composite of all-cause death, CHF, cardiogenic shock, or re-infarction were (P-RD ≤55 min) 10.6% (PI) versus 10.3% (P-PCI, p=0.910); (P-RD >55–97 min) 13.9% versus 17.9% (p=0.148); and (P-RD >97 min) 13.5% versus 16.2% (p=0.470) (table 4). The rates for primary composite endpoint and individual components of composite were lowest in first P-RD category and rose in similar proportions in the latter two categories. Although the composite endpoint increased nominally in both treatment groups as P-RD increased, this reached significance only in the P-PCI group (p(trend)=0.038). Within each treatment arm, a significant increase in combined shock and CHF was evident with increasing P-RD (PI: p(trend)=0.035; P-PCI: p(trend)<0.001). The incidence of stroke, ICH, and non-ICH major bleeding was low across all P-RD.
Across each component of the composite clinical endpoint, there was no evidence that P-RD was associated with significant difference in outcomes of randomised study treatment (all tests for interaction p>0.05) (figure 1). However, closer examination of the data highlights some important trends. Among patients enrolled at sites experiencing short P-RD (≤55 min), those randomised to P-PCI had a nominally lower composite event than those randomised to the PI arm. Conversely as P-RD increased PI-assigned patients experience less composite endpoint than those assigned to P-PCI (figure 1). This association was confirmed, and observed to a greater degree, in the sensitivity analysis (sites enrolling more than 20 patients) (p(interaction)=0.061; see online supplementary appendix figure A1).
The association between study treatment and 30-day composite endpoint according to the P-RD as a continuous variable is shown figure 2. The RR for events was less for P-PCI in sites with shorter delay until P-RD approached 50 min. Thereafter, there was a clear linear trend towards benefit in primary endpoint among PI-assigned patients (p(interaction)=0.073).
It is generally accepted that P-PCI is the preferred reperfusion option for patients with STEMI but also appreciated that the benefit becomes attenuated the longer it takes to deliver P-PCI following symptom onset.8 ,12–14 Furthermore, historical comparisons of P-PCI with fibrinolysis have in general been compared with in-hospital lytic alone1 ,2 rather than a PI strategy.
The previously published STREAM 30-day results showed that the PI strategy was a useful alternative option11 producing similar clinical outcomes in patients who could not receive timely PCI (within prespecified 60 min from when fibrinolysis could be delivered). This paper specifically examines the relationships between time and outcomes and presents analyses of time-related delay not available in the original paper.
Much has been written on the importance of timing to initiation of reperfusion. Animal data12 demonstrated irreversible cell death after about 40 min. A derived time construct15 indicates that during the first 2–3 h after symptom onset, time to treatment critically determines reduction in mortality: thereafter, a lesser benefit occurs. In the current analysis, patients randomised to the PI strategy received treatment within 2.4 h (median 1.7 h) of symptom onset compared with 3.8 h (median 3 h) for those randomised to P-PCI.
Guideline standards have been set for optimal time delay metrics.15 ,16 Attaining within-guideline door-to-balloon times17 ,18 may be the easiest parameter to achieve, especially driven by in-hospital audits. Overall ischaemic time may have more importance. In a recent publication,13 the door-to-balloon times fell significantly from median 83 min (2005–2006) to median 67 min (p<0.001) in 2009–2010. However, the in-hospital 30-day mortality remained similar (4.8%) (p=0.64). This can be explained by recognising that the door-to-balloon delay is only one component of the patient's overall ischaemia time, and shortening the overall symptom to reperfusion time (the system delay)14 is the key to improving outcomes. Terkelsen19 has clearly demonstrated the impact of system delay, with excess mortality evident over a 7-year follow-up (p<0.001), especially when the system delays were >180 min. Guidelines promote networks to reduce system delays.16 However, despite worthy efforts, insurmountable system delays continue in many parts of the world (due to geography, hospital-to-patient population density ratios, climatic issues and traffic congestion in large urban areas). These and other circumstances indicate that system delays may not always readily be overcome. In such global circumstances, initiating early fibrinolysis with timely follow-on angiography (including provision for urgent rescue PCI) may be preferable in order to achieve best outcomes.
It is the relationship of time delay to outcomes for the two strategies that was the specific focus of this study. It should be noted that the design of the current study selected a STEMI patient population presenting within 3 h of symptom onset who were judged unable to receive P-PCI within 1 h. Against a background of loss of benefit of PCI over PI therapy when the site-related delay is >1 h, our results (figure 2) emphasise the potential overall benefit of PI as this time delay progressively increases. Although historical trials of fibrinolysis have shown excess intracranial bleeding compared with P-PCI, in the parent STREAM trial, this was not the case once the protocol amendment to reduce TNK by 50% in the over 75 year old had been implemented.
In this prespecified substudy, the 30-day STREAM outcomes were analysed according to P-RD recognising that our study design mandated at least a 60 min P-RD in accordance with the European Society of Cardiology guidelines.20 The most important finding was that in the PI arm there was no increased hazard across P-RD (p(trend)=0.292), whereas there was an increased hazard in the P-PCI arm (p(trend)=0.038). For the 30-day composite of death/CHF/cardiogenic shock/re-MI, the test for interaction between P-RD and study treatment approached statistical significance (p(interaction)=0.094). This potential effect on metrics of LV function is of interest and in agreement with the findings from the CAPTIM and WEST studies.8
A substantial proportion of STREAM patients had a P-RD <60 min despite the protocol specifying that patients were only to be included if they could not undergo PCI within 60 min of first medical contact (table 2 and figure 2). Their incidental inclusion confirms such patients appear well served by P-PCI. This relationship was also observed in the sensitivity analysis, which restricted the cohort to 20 patients per site (p(interaction)=0.049). Thus, while this exploratory analysis suggests that when the site-related delay is short (≤55 min), patients do slightly better with P-PCI, as P-RD increased, the advantage swings towards PI. Despite the strong trend towards an interaction between P-RD and treatment strategy on clinical outcomes, the directional nature of our findings should be interpreted with caution given the modest size of the sample. Notwithstanding this caveat, the outcome data in this very early treated STEMI population provide new insights into the importance of timely reperfusion and add support for employing the PI strategy in parts of the world where (guideline) timely P-PCI cannot be achieved.
Our results differ from other published data because of important differences in median treatment times as expected given our narrow 3 h randomisation inclusion window from symptom onset. In our total population, P-RD was greater largely due to a much shorter fibrinolytic initiation time of 78 min compared with ≥160 min DANAMI-2 trial highlighting the ‘time advantage’ of fibrinolysis in STREAM.21 For the same reasons, our findings also contrast with the PCAT-2 Trialists where the median time from symptom onset to in-hospital fibrinolysis was 162 min.22
The parent STREAM study was designed as a proof-of-concept study. All statistical tests were of an exploratory nature, and the small numbers do not allow for robust analytical assessment of clinical endpoints, especially mortality. Thus, the results of this presented subgroup analysis should also be considered exploratory and hypothesis generating. We also acknowledge that there may be unmeasured confounders related to delay, treatment and outcomes. However, within this randomised trial we are able to provide insights into how time delays affect both treatment strategies. The timely performance of rescue PCI and excellent protocol mandated ancillary pharmacological therapy likely contributed to our positive PI results and may be challenging to achieve in some clinical domains. Finally, our results do not challenge the assertion that P-PCI is the preferred option for STEMI when delivered in an expedient and timely manner in an expert 24/7 facility.
This study underscores the importance of total ischaemic time and highlights the importance of system delays in influencing outcomes after STEMI. An association between P-RD and clinical outcome has been demonstrated for the first time in a prospective study comparing directly PI with P-PCI. The results indicate that when ambulance systems and community hospitals face a P-RD of >60 min, a PI strategy as used in STREAM should be least considered. The PI strategy may be applicable as the best reperfusion option in the many parts of the world when P-PCI cannot be delivered expeditiously.
What is already known on this subject?
Primary percutaneous coronary intervention (P-PCI) is the preferred option in ST-elevation myocardial infarction (STEMI). However, the benefits of P-PCI become attenuated as time to its delivery increase. A pharmacoinvasive (PI) strategy may be an option.
What might this study add?
Within the context of a randomised clinical trial comparing reperfusion with a PI strategy versus P-PCI in early presenting patients with STEMI, as PCI-related delay increased, outcomes with the PI strategy became superior to P-PCI.
How might this impact on clinical practice?
Globally, in circumstances where PCI-related delay exceeds guideline-mandated times, a PI strategy provides an acceptable alternative reperfusion option to P-PCI.
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Files in this Data Supplement:
- Data supplement 1 - Online supplement
Contributors The paper has been contributed to by all the coauthors, and all have agreed the contents of this final submitted draft.
Funding Funding for this trial was provided by Boehringer Ingelheim.
Competing interests AHG discloses grant support and honoraria from Boehringer Ingelheim and Advisory Board honoraria from AstraZeneca and Daiichi Sankyo; PWA discloses grant support and honoraria from Boehringer Ingelheim. PWA’s financial activities outside the submitted work are posted and routinely updated through http://www.vigour.ualberta.ca/en/About/ConflictofInterest.aspx; KH discloses lecture fees from Boehringer Ingelheim; SH discloses lecture fees from Boehringer Ingelheim; PGS discloses research grant (to INSERM U698): NYU School of Medicine, Sanofi and Servier; speaking or consulting: Amarin, AstraZeneca, Bayer, Boehringer-Ingelheim, Bristol-Myers-Squibb, Daiichi-Sankyo, GSK, Iroko Cardio, Lilly, Medtronic, Otsuka, Pfizer, Roche, Sanofi, Servier, The Medicines Company and Vivus. Stockholding: Aterovax; PG discloses speakers bureau support from AstraZeneca and Bayer; honoraria from sanofi, Boehringer Ingelheim, Eli Lilly and Medicines Company; ACC discloses consulting fees from Boehringer Ingelheim, AstraZeneca, Daiichi Sankyo; FVdW discloses research grant, other support from Boehringer Ingelheim; RGW has no conflicts to declare.
Patient consent Obtained.
Ethics approval The study protocol was approved by national regulatory authorities as well as the local ethics committee at each study centre.
Provenance and peer review Not commissioned; externally peer reviewed.
Data sharing statement All requests for further data from this study should be addressed to the corresponding author.
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.