THEMIS: A Role for Ticagrelor in Diabetic Patients With Prior PCI?

September 01, 2019

PARIS — For patients with type 2 diabetes and stable coronary artery disease, the benefits of adding ticagrelor (Brilinta/Brilique, AstraZeneca) to aspirin in terms of reduced ischemic events did not outweigh the bleeding risk in the THEMIS trial.

However, there did appear to be a net clinical benefit of ticagrelor in the subgroup who had previously undergone percutaneous coronary intervention (PCI) — which made up of 58% of the 19,000 patients in the trial.

 "Our conclusion is that careful patient selection would be necessary to determine which patients may benefit from dual antiplatelet therapy with aspirin plus ticagrelor," co-lead investigator Deepak Bhatt, MD, Brigham and Women's Hospital, Boston, Massachusetts, told | Medscape Cardiology.

"Certainly, patients at high bleeding risk would not be considered for ticagrelor treatment. But for patients with diabetes and stable coronary disease who are at low bleeding and high ischemic risk, this could be a new strategy. And the results of the THEMIS-PCI subgroup suggest that patients who have had prior PCI represent the type of patient who could benefit."

The THEMIS trial and the THEMIS PCI substudy were both presented at the European Society of Cardiology (ESC) Congress 2019 and simultaneously published online September 1 in the New England Journal of Medicine (NEJM) and The Lancet, respectively.

"In the group with prior PCI, there was 15% relative risk reduction in the primary ischemic endpoint of CV death/MI/stroke (vs 10% in overall trial), and more importantly, the net clinical benefit prespecified endpoint (death/MI/stroke/fatal bleeding/ICH) was significantly lower in the PCI population compared to the non-PCI population," Bhatt said.

"In our opinion, we think the combination of ticagrelor and aspirin will be useful in this population of diabetic patients with stable coronary artery disease who have had a previous PCI."

He noted that on average PCI had been conducted in these patients more than 3 years before enrollment. "Some patients were even 10 or more years after PCI and the net clinical benefit was still suggested right out to this timepoint, and so we're not talking about just the short term here."  

Bhatt suggested that the favorable results in the prior PCI subgroup could be explained by the selection of patients who had taken dual antiplatelet therapy before without a problem.

"We think the lower bleeding we saw in the prior PCI patients is presumably because they've previously passed a bleeding stress test with dual antiplatelet therapy," he said. "A patient who has received a previous PCI/stent has probably had previous exposure to dual antiplatelet therapy for at least a few months and presumably they would have tolerated it well without a major bleed or their doctor wouldn't have enrolled them into the THEMIS trial. So now when they're re-exposed to ticagrelor and aspirin, they have less bleeding risk than a patient who has not taken dual antiplatelet therapy before but they gain all the anti-ischemic benefits."

"What we found is that there may be role for re-initiating dual therapy with aspirin and ticagrelor in diabetic patients who have previously undergone PCI," he added. "That is quite a large population as diabetic patients will often have had a PCI. But we really want to be careful on patient selection and make sure they are at low bleeding risk. If they have previously suffered a GI bleed or are on chronic NSAIDs, then we wouldn't want to do it. And if they haven't been exposed to dual antiplatelet therapy before, then I would say we probably wouldn't want to start it now several years later."

So can these data be extrapolated to continue dual antiplatelet therapy long-term in diabetic patients recently having undergone PCI?

Bhatt said the study didn't directly address that question, but "if a diabetic patient is on dual antiplatelet therapy after a recent PCI and tolerating it well without any bleeding, then I think they could now be considered for extended-duration treatment."

Another question is how long the dual treatment should be continued. "The net clinical benefit of ticagrelor plus aspirin was suggested right out to 10 years after PCI and a significant benefit out to 6 years," Bhatt said. "So this could be a forever therapy in the right patients unless they develop a new risk of bleeding."

Commenting for | Medscape Cardiology, Shamir Mehta, MD, McMaster University, Hamilton, Ontario, Canada, said, "The benefit observed in the THEMIS PCI subgroup is very consistent with what we know about long-term dual antiplatelet therapy in PCI patients. We know there is benefit, but the problem is how to balance that benefit with the risk of bleeding."

Mehta agreed with Bhatt's interpretation of the probable reason behind the positive result in the THEMIS-PCI population. "Yes, I think the trial probably selected out a group that was high ischemic risk as they had had a PCI before but had already showed that they could tolerate dual antiplatelet therapy."

Mehta said he would now consider this combination of aspirin plus ticagrelor, particularly for higher-risk patients with complex disease with a low risk for bleeding.

A typical example might be a 45- to 50-year-old patient with diabetes and complex disease undergoing PCI, he said. "I would now consider long-term ticagrelor in that setting. For now, I think that is the easiest way of implementing the THEMIS-PCI results. If they haven't bled for the first several months to a year on the combination of aspirin plus ticagrelor, then this data is encouraging for continuing that treatment long term."

"The question is now, Can we further reduce the risk of bleeding by removing aspirin? And we will soon have data on that from the TWILIGHT trial," he added.


The main THEMIS trial randomly assigned 19,220 patients with stable coronary artery disease and type 2 diabetes to ticagrelor plus aspirin or placebo plus aspirin. Patients with previous myocardial infarction (MI) or stroke or those with recent acute coronary syndrome (ACS) were excluded because they could already be eligible for dual antiplatelet therapy.

After a median follow-up of 39.9 months, permanent treatment discontinuation was more frequent with ticagrelor than placebo: 34.5% vs 25.4%.

The primary efficacy endpoint — a composite of cardiovascular (CV) death, myocardial infarction (MI), and stroke — was lower in the ticagrelor group than the placebo group, but TIMI major bleeding was higher in the ticagrelor group, as was intracranial hemorrhage. Fatal bleeding was not significantly different.

An exploratory outcome of irreversible harm (death from any cause, MI, stroke, fatal bleeding, or intracranial hemorrhage [ICH]) — a measure of net clinical benefit — was similar in the two groups.

Table 1. THEMIS: Main Results


Endpoint Ticagrelor (%) Placebo (%) Hazard Ratio (95% CI) P Value
Primary efficacy endpoint (CV death/MI/stroke) 7.7 8.5 0.90 (0.81 - 0.99) .04
TIMI major bleeding 2.2 1.0 2.32 (1.82 - 2.94) <.001
ICH 0.7 0.5 1.71 (1.18 - 2.48) .005
Fatal bleeding 0.2 0.1 1.90 (0.87 - 4.15) .11
Net clinical outcome (death/MI/stroke/fatal bleeding/ICH) 10.1 10.8 0.93 (0.86 - 1.02)


In an accompanying editorial in the NEJM,  Eric Bates, MD, University of Michigan, Ann Arbor, notes that THEMIS adds to a long list of other studies that have investigated ticagrelor in various patient populations. 

While ticagrelor was superior to clopidogrel in patients with ACS, results in patients with cerebrovascular disease or peripheral artery disease as well as the long-term treatment of MI have not shown net clinical benefit because of bleeding risk.

In terms of the current trial, he concludes,  "It may be possible to identify individual patients who have an increased risk of thrombotic events and a reduced risk of bleeding for whom this trade-off is advantageous. But for most patients with type 2 diabetes and known coronary disease who fit the THEMIS enrollment criteria, the addition of ticagrelor to aspirin is not recommended."


The THEMIS-PCI subgroup involved 11,154 patients with a history of previous PCI (58% of the overall THEMIS trial). Median follow-up was 3.3 years.

In this group, the primary efficacy outcome (CV death/MI/stroke) was lower in the ticagrelor group. TIMI major bleeding was increased, but fatal bleeding and ICH were similar in the two groups, and ticagrelor was associated with an improved net clinical benefit:

Table 2. THEMIS-PCI Results

Endpoint Ticagrelor (%) Placebo (%) Hazard Ratio (95% CI) P Value
Primary efficacy endpoint (CV death/MI/stroke) 7.3 8.6 0.85 (0.74 - 0.97) .013
TIMI major bleeding 2.0 1.1 2.03 (1.48 - 2.76) <.0001
ICH 0.6 0.6 1.21 (0.74 - 1.97) .45
Fatal bleeding 0.1 0.1 1.13 (0.36 - 3.50) .83
Net clinical outcome (death/MI/stroke/fatal bleeding/ICH) 9.3 11.0 0.85 (0.75 - 0.95) .005


In a Comment accompanying the Lancet publication of the THEMIS-PCI subgroup, Marco Valgimigli, MD, and Negar Manavifar, MD, Bern University Hospital,  Switzerland, point out that a different definition of net clinical benefit with the primary efficacy endpoint (cardiovascular death; MI or stroke; and the primary safety endpoint, TIMI major bleeding) did not show a significant benefit of ticagrelor.

They note that an additional post hoc net clinical benefit endpoint, including bleeding score of 3 or more as defined by the BARC (Bleeding Academic Research Consortium) criteria, with MI and stroke, has not been provided "but would be of interest, on the basis of a previous comparative analysis showing an independent association with mortality."

The editorialists conclude that clinicians would now need to consider the combination of ticagrelor and aspirin as a new treatment option for patients with stable coronary artery disease and diabetes undergoing PCI.

"However, optimal patient selection, timing of treatment initiation, and treatment duration remain unclear. Whether previously validated bleeding risk scores, with or without concomitant assessment of the ischaemic risk, will help to maximise the benefits over the risks warrants further investigation."

THEMIS was funded by AstraZeneca. Bhatt reports receiving research grants from AstraZeneca during the conduct of this study. Steg reports personal fees and nonfinancial support from AstraZeneca, during the conduct of the study; grants and personal fees from Bayer/Janssen and Servier, Merck, Sanofi, and Amarin; and personal fees from Amgen, Bristol-Myers Squibb, Boehringer Ingelheim, Pfizer, Novartis, Regeneron, Lilly, AstraZeneca, Novo Nordisk, and Idorsia, outside the submitted work.

European Society of Cardiology (ESC) Congress 2019.  Presented September 1, 2019.

THEMIS: N Engl J Med. Published online September 1, 2019. Abstract, Editorial

THEMIS-PCI: Lancet.  Published online September 1, 2019. Abstract, Comment   

For more from | Medscape Cardiology, join us on Twitter and Facebook

Discussant Unimpressed

Designated discussant of the THEMIS trial at the ESC hotline session, Colin Baigent, FRCP,  University of Oxford, United Kingdom, was not impressed with the focus on the PCI subgroup.

He pointed out that there was no significant treatment interaction for the primary efficacy outcome in the PCI group.  "So why pick out this group when the drug probably doesn't work any better in these patients?" he asked. 

He said he thought it would be very difficult to find any group who may truly safely benefit from this treatment when the benefits and hazards were so finely balanced. 

"The same factors predict increased ischemic risk and increased bleeding risk, so patients at increased bleeding risk and reduced bleeding risk are very few and far between," he commented. 


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.