March 10th, 2013
Cangrelor During PCI May Reduce Ischemic Events
Larry Husten, PHD
For more of our ACC.13 coverage of late-breaking clinical trials, interviews with the authors of the most important research, and blogs from our fellows on the most interesting presentations at the meeting, check out our Coverage Headquarters.
In the Cangrelor versus Standard Therapy to Achieve Optimal Management of Platelet Inhibition (CHAMPION PHOENIX) trial, the intravenous platelet inhibitor cangrelor was tested for its effect on ischemic events associated with PCI. Cangrelor is a potent, fast-acting and reversible agent. Results of the trial were presented at the ACC in San Francisco and published simultaneously in the New England Journal of Medicine.
A total of 11,145 PCI patients were randomized to a bolus and infusion of cangrelor or to a loading dose of clopidogrel. A primary endpoint event — death, MI, ischemia-driven revascularization, or stent thrombosis at 48 hours — occurred in 4.7% of the cangrelor group versus 5.9% of the clopidogrel group (adjusted OR, 0.78; 95% CI, 0.66-0.93; P=0.005). The authors calculated that 84 patients would need to be treated with cangrelor instead of clopidogrel to prevent one primary endpoint event.
Cangrelor treatment was not associated with an increased risk for bleeding but was associated with a significant reduction in stent thrombosis:
- Severe bleeding at 48 hours occurred in 0.16% of the cangrelor group and 0.11% of the clopidogrel group, a nonsignificant difference.
- Stent thrombosis occurred in 0.8% of the cangrelor group versus 1.4% of the clopidogrel group (OR, 0.62; 95% CI, 0.43-0.90; P=0.001)
Few adverse events occurred, but the cangrelor group experienced more episodes of transient dyspnea than the clopidogrel group (1.2% vs. 0.3%).
The early benefit was sustained at 30 days, with significant reductions in the primary endpoint (6% vs. 7%; P=0.03) and stent thrombosis (1.3% vs. 1.9%; P=0.01).
“The investigators feel the data are compelling,” said Bhatt in an ACC press release. “The data we’ve shown are clear and consistent across all relevant subgroups or patient populations. This drug has several advantages, and nothing out there right now has quite the same biological properties.”
In an accompanying editorial, Richard Lange and L. David Hillis note that in the clopidogrel group one quarter of the patients received the 300-mg loading dose instead of the more potent 600-mg dose and 37% received clopidogrel during or after PCI. “As a result,” they write, “the antiplatelet effects of clopidogrel were suboptimal at the time of PCI.” In addition, ACS patients, who constituted 45% of the study population, may have been further disadvantaged because they received clopdiogrel instead of the newer agents prasugrel or ticagrelor. They conclude that cangrelor may be beneficial in some PCI patients who can benefit from an intravenous agent, but that “the routine use of this therapy for all patients undergoing PCI is not yet justified.”
4 Responses to “Cangrelor During PCI May Reduce Ischemic Events”
NEJM Journal Watch — Recent Cardiology Articles- Transcatheter Valve Replacement for Severe Tricuspid Regurgitation
- Time in Blood Pressure Target Range and Outcomes in Hypertension: Real-World Data
- When to Start Anticoagulation After Ischemic Stroke in Patients with Atrial Fibrillation
- Aortic-Valve Replacement vs. Clinical Surveillance for Asymptomatic Severe Aortic Stenosis
- Secondary Hypertension in Young Adults
Here’s a little followup on CHAMPION PHOENIX:
At a news conference several experts seemed positive about possible FDA approval of cangrelor. Harrington (co-chair of the trial) noted that with more than 25,000 patients in clinical trials the FDA has plenty of data to evaluate the drug. With the caveat that the FDA has not yet scrutinized the data, he said that “these data look pretty favorable and will be favorably received.” Martin Leon noted that since no major safety concerns about the drug have been raised, “the obvious benefits and flexibility” of cangrelor suggests a favorable prospect.
Responding to a question about the cost implications of cangrelor, Leon said that this was the key remaining question that needs to be answered, assuming the drug gains approval. Bhatt pointed out that there were some potentially compelling cost advantages to cangrelor, including the ability to avoid a lengthy hospital stay for those patients who require CABG surgery.
Hmmm. I have not gone into all the details of the Lancet article on “Effect of cangrelor on preprocedural outcomes in percutaneous coronary interventions: a pooled analysis.” . Here are just a few quick impressions after one read.
1) Like in many CV trials composite endpoints are used and always cloud the data. Typically 3 to 4 endpoints are used like in this pooled analysis. There are discrepancies between the total number of individual events and those reported for the composite end-point. Individual outcomes do not contribute equally to composite measures, so the overall estimate of effect for a composite measure cannot be assumed to apply equally to each of it’s individual outcomes.
2) Odds ratio are used which is not entirely correct because they are not a time-to- event analysis measure. This is why hazard ratios should be used (OR are used in typically used in case control studies ). Relative risk reduction is then derived which is not entirely correct when dealing with hazard ratios. This gets a bit technical and is probably not all that important , but what is important is the baseline risk of an event when interpreting the data. For example in the conclusion it is stated that “stent thrombosis” is reduced by 41% – totally irrelevant when looking at baseline risk which is only 0.8%. What makes the importance of this outcome a little strange is that ST ( at 48hours) and MI were statistically decreased but IDR (ischemia driven revascularization) was not ?? This also brings up the real problems with composite endpoints as a primary outcome: when death is part of a composite end-point/outcome, competing risk will influence the results because patients that die cannot have any further endpoints. So how did ST contribute to death- we don’t know from this analysis. It may not, as IDR has not increased or it may simply because the patient may have died, but that cannot be assumed. This problem can only be addressed by counting only 1 event per patient.
3)Few would argue that death is as important as revascularization ( actually no difference in IDR) , but in the soft endpoint of MI measured by enzymes.
4)The NNT for ST is 333! Ouch. Particularly after reading #6 below, becomes real problematic.
5) Again reporting on RRR of the primary endpoint ( composite end- point) of 19% is very small when this correlates with an ARR of less then 1% and is quite misleading. An ARR of less than 1% for all-cause death would be important. The ARR of the triple composite end-point is even smaller at 0.8% and the ARR increase for , albeit minor bleeding is 5%- in a tightly supervised clinical trial. Like with any of these outcomes real- life outcomes are likely of a lesser magnitude.
6)However, the biggest problem I see- is the way the three trials were delibertalky designed to make Plavix look less favourable. Only going back to 2005 the general concencus was to give a 600-mg loading dose of clopidogrel pre-PCI ( PCI -CLARITY in JAMA 2005), as this dose achieves maximal inhibition of platelets in just two hours. I am quoting the result of the trial: “Clopidogrel pretreatment significantly reduces the incidence of cardiovascular death or ischemic complications both before and after PCI and without a significant increase in major or minor bleeding. These data add further support to the early use of clopidogrel in STEMI and the strategy of routine clopidogrel pretreatment in patients undergoing PCI.”
Looking at the study design of CHAMPION ( PLATFORM, PCI and PHOENIX) , in addition to a placebo arm the dose of Plavix was given at the time of PCI, during or even AFTERWARDS !!! In addition not everybody got 600mg a number of patients got only 300 mg.
The enormous amount of statistical data in this article are misleading and takes the focus away of the design of the actual trials.And of course then there are the survival curves- my hair raises when on the Y-axis the event free % goes from 0 to 2.5% for the full scale , rather then using a larger scale to demonstrate the small differences. But visually it looks impressive.
7)None of the data where compared to ticagrelor or prasugrel.. so a little bit of old news. The bottom line though: Plavix is not given a fair shot in these trials particularly now that it has gone generic.
8) The Medicines Company provided the data for for the analysis. Always the question: who owns the data? This company has only a few drugs in their pipeline…and you know what I am thinking. I attached this reference below. I am absolutely not claiming any falsification of data on the Investigators or Scientist involved in the analysis in this particular case, but I worry far more about the Drug Companies. Who owns the data and what they do with it, particularly when their survival depend on a single drug.
Citation: Fanelli D (2009) How Many Scientists Fabricate and Falsify Research? A Systematic Review and Meta-Analysis of Survey Data. PLoS ONE 4(5): e5738.
doi:10.1371/journal.pone.0005738
In this large survey scientist admitted to either falsification or manipulation of data ranging from 14 to 33% !!!!
9) I still remember criticizing from day # 1 ( I still have the e-mail to CCM in 2008) after I read the article in the NEJM the use of rAPC (Xigris) in sepsis. A huge campaign by Lilly was set up and everybody fell in the trap with widespread use of the drug. Sure enough more recently we find out we were killing patients with the drug and it was rapidly withdrawn from the market. We were killing patients for 5 years with an ineffective drug, with its efficacy that could have been questioned from the very first publication.
10) Careful analysis of trials is necessary to keep it all in perspective and not get caught up in the hype.
And that’s exactly the time we dont have anymore !
I am referring to the recent Online Publication of :Effect of cangrelor on periprocedural outcomes in percutaneous coronary interventions: a pooled analysis of patient-level data Original Research Article
In Press, Corrected Proof, Available online 3 September 2013
Philippe Gabriel Steg, Deepak L Bhatt, Christian W Hamm, Gregg W Stone, C Michael Gibson, Kenneth W Mahaffey, Sergio Leonardi, Tiepu Liu, Simona Skerjanec, Jonathan R Day, Robert S Iwaoka, Thomas D Stuckey, Harinder S Gogia, Luis Gruberg, William J French, Harvey D White, Robert A Harrington, for the CHAMPION Investigators
Composite outcomes are useful because they allow testing of treatments or therapies with more statistical power than using single outcomes, particularly when the outcome is mortality, which is rare. This is important for several reasons: first, mortality, although the ultimate outcome, does not fully reflect the potential impact of a given treatment on health status. There are many other nonfatal outcomes which are important and impact morbidity or quality of life. Thus, using a composite outcome often allows a more comprehensive assessment of the impact of novel therapies. Yet, composite outcomes may create difficulties in interpretation because components are often of unequal clinical importance and may not necessarily track with each other. Therefore, it is important to report the effect of any treatment on the composite outcome, but also on its individual components, as was done in the paper. Note that contrary to the assertion of the letter writer, there is no discrepancy between number of individual endpoints and composite outcomes. Differences are expected, since a single patient can experience multiple outcomes.
As authors, we never argued that death was as important as revascularization (or rather the converse) but reported individual as well as composite outcomes of potential interest. Of note, the benefit of cangrelor over clopidogrel or placebo was seen to affect all MIs, but also, importantly, Q-wave MIs, events which very few would question as being clinically important.
NNTs should be interpreted in the context of totality of events, not one single component of the composite outcome. Certain events are relatively rare, others may be more common. Selecting one rare event and highlighting a high NNT is meaningless. It is also important to remember that PCI is a widely used technique worldwide and that any reduction in its complications, even if this represents a relatively small absolute reduction, is potentially important for public health, given the large number of procedures done.
The letter writer’s points regarding type of control are not novel and the fact that clopidogrel was used in the control arm and therefore there were no data on prasugrel or ticagrelor was already part of our own discussion of the results. As also discussed in the paper, the role of preloading with clopidogrel prior to procedures remains uncertain, as was exemplified by the recent meta-analysis (Bellemain Appaix et al, JAMA 2013) showing uncertain benefit of preloading in PCI (particularly elective PCI), and, in the context of ACS, with the recent results of the ACCOAST trial (Montalescot et al NEJM 2013) which suggest that preloading with oral ADP receptor blockers prior to PCI in ACS may be associated with increased bleeding but no reduction in ischemic outcomes. Note that the subgroup analyses provided in the paper show consistent effects of cangrelor, regardless of whether clopidogrel was given before or after the start of PCI and regardless of whether a loading dose of 300 or 600 mg was used.
Finally, we strongly object to a couple of statements made by the letter writer:
First, “the enormous amount of statistical data in this article is misleading and takes the focus away of the design”: Contrary to what this language suggests, we feel strongly that providing extremely detailed and comprehensive data is important to assess fully all outcomes. It is very unusual to be accused of disclosing too many results. We find the idea that this is done with the intent to take focus away from the design both surprising and illogical, particularly when Figure 1 of the article is actually a color schematic design of the trials.
Second, the mention of a reference about fabrication and manipulation of data is at the very least inappropriate, if not offensive (even though there is “disclaimer” that this is not claiming fabrication of data by the scientists or academics involved). We will point out to the acknowledgment section of our article, which states that the data were analyzed in parallel by the sponsor’s statisticians and by those from the Duke Clinical Research institute and that the pooled statistical analysis was also verified by academic statisticians from Harvard Clinical Research Institute. We hope that statistical verification of the data by two academic statistical groups will be sufficient to reassure the letter writer about the validity of the results reported.
Ph.Gabriel Steg
Deepak L Bhatt
Robert A Harrington