First trial backs blood thinners for atrial fibrillation with one stroke risk factor, on just 17 events

People with atrial fibrillation and a single extra risk factor for stroke sit in a gray zone. Guidelines grade a blood thinner for them one step below a firm recommendation, and that grade has rested on registry studies that disagree with one another. According to the investigators behind a new study, no randomized trial had tested it.

One now has. In a trial of 1,803 such people in South Korea, published in the New England Journal of Medicine, those given a blood thinner had fewer serious events over two years than those given none, most of them strokes, and no clear excess of major bleeding. The catch is the size of the numbers. Across both groups, 17 people had one of the events the trial was counting.

Who counts as intermediate risk in atrial fibrillation

Atrial fibrillation (AF) is an irregular heartbeat in which the heart’s upper chambers quiver instead of beating cleanly. Blood can pool there and form clots, and a clot that reaches the brain causes a stroke. People with the condition are at increased risk of suffering a stroke, as the European Society of Cardiology (ESC) puts it.

The standard defense is an anticoagulant, a medicine commonly called a blood thinner, which slows clotting. The price is a higher chance of bleeding. So the decision turns on how likely a stroke is in the first place.

Doctors estimate that with a points checklist called CHA2DS2-VASc. Points are given for heart failure, high blood pressure, diabetes, a previous stroke, disease of the blood vessels, older age and female sex. The total runs from 0 to 9, with higher scores indicating a greater risk of stroke.

At the top of the scale the advice is settled. European guidelines recommend oral anticoagulants for patients with AF at high risk of stroke, with their strongest grade. Lower down it softens: the same guidelines suggest that oral anticoagulants should be considered in patients at intermediate risk, a second-tier grade.

Intermediate risk has a precise meaning here. It was defined as a CHA2DS2-VASc score of 1 in men or 2 in women. Because being female earns a point on its own, the two amount to the same thing: one risk factor besides sex.

The softer grade reflects the evidence behind it. “Evidence from observational studies with anticoagulants, particularly vitamin K antagonists, remains conflicting in those at intermediate risk,” said Boyoung Joung of Yonsei University in Seoul, the trial’s principal investigator. Vitamin K antagonists are the older class of blood thinner, a group that includes the drug called warfarin.

What the anticoagulation trial did

The study, called SINGLE-AF, was an open-label trial conducted at 18 centers in South Korea. Open-label means that patients and their doctors knew who was taking the drug. No placebo was used.

Patients were eligible if they were between 19 and 80 years old and had atrial fibrillation with the qualifying score. People with serious liver or kidney disease, damage to the heart’s valves or muscle, or active cancer were left out, as were those who needed an anticoagulant because of surgery.

Patients were randomly assigned in a 1:1 ratio to receive either direct oral anticoagulant (DOAC) therapy or no anticoagulation. DOACs are the newer class of blood thinner, developed as alternatives to warfarin. Those in the drug group received apixaban 5 mg twice daily or rivaroxaban 20 mg once daily, with the choice left to the treating doctor.

Of 1,803 patients, 902 were assigned to receive DOAC therapy and 901 were assigned to receive no anticoagulant therapy. The average age was 60.4 years, and 23.7% were women.

The main measure was a composite of stroke, systemic embolism, major bleeding, or death from cardiovascular causes at 24 months. Composite means that any one of those events counted. Systemic embolism means a clot that leaves the heart and blocks an artery somewhere other than the brain.

The measure deliberately mixes benefit and harm. Strokes and clots are what the drug is meant to prevent, and major bleeding is what it can cause. Counting them together asks whether people come out ahead overall.

One safeguard against bias was built in. The design was investigator-initiated, multicenter, open-label, adjudicator-masked. Investigator-initiated means that doctors, not a company, conceived the trial. The last term means that the reviewers who ruled on each suspected event were not told which group the patient was in.

Fewer events with a blood thinner, on small numbers

Over 24 months	Blood thinner (902 patients)	No blood thinner (901 patients)
Any event in the main measure	4 (0.5%)	13 (1.5%)
Stroke	3 (0.3%)	10 (1.1%)
Major bleeding	0.3%	0.5%
Lesser but clinically relevant bleeding	2.5%	1.6%
Death from cardiovascular causes	0	0
Serious adverse events	80 (8.9%)	84 (9.3%)

The percentages are estimates of cumulative incidence, the share of each group affected by two years. They are calculated in a way that allows for people who left the trial early, so they can differ slightly from a simple division of the counts and cannot be turned back into exact head counts. A patient who had more than one kind of event counts once in the top row.

At 24 months, a primary end-point event had occurred in 4 patients in the blood thinner group, or 0.5%, and in 13 in the comparison group, or 1.5%. A primary end-point event is any event in the main measure. That is a difference of 1.0 percentage point, with a plausible range from 2.0 points down to 0.1.

Most of the events were strokes. Stroke occurred in 3 patients taking the drug and in 10 who were not. According to the society, the difference between the groups appeared to be driven by ischemic stroke, meaning the kind caused by a clot blocking an artery in the brain. In an analysis added after the results were in, that kind, together with clots elsewhere in the body, affected 0.1% of the drug group against 1.1% of the comparison group. Set beside three strokes of all kinds in the drug group, that implies most of the three were of another type. The other main type is bleeding in the brain, a known risk of blood thinners.

Looked at on its own, though, the stroke result is less secure than the combined one. Stroke occurred in 0.3% of patients in the DOAC group compared with 1.1% of the control group, the control group being those given no blood thinner. That is a gap of 0.8 percentage points whose plausible range ran from 1.6 points to zero. On strokes alone, in other words, the trial could not quite rule out no difference.

In absolute terms the benefit is modest. A one-point difference works out at roughly 100 people taking a blood thinner for two years to spare one of them a stroke, serious bleed or similar event. At the edges of the plausible range, that number could be as low as 50 or as high as 1,000.

Joung drew a firm conclusion. “We now have evidence from a randomized trial that patients with AF at intermediate stroke risk benefit from DOAC therapy, without an increase in major bleeding,” Joung said. Joung added that the data “may be used to inform future guideline recommendations and reimbursement policies”.

Bleeding and other risks of anticoagulation

Bleeding is the reason anyone hesitates over these drugs, and here the trial was reassuring within its limits. Major bleeding generally means bleeding that is fatal, affects a critical site such as the brain, or needs a transfusion. The society’s summary was that there was no difference in major bleeding: 0.3% with the drug and 0.5% without, figures too small to tell apart. The paper’s own wording is careful. The incidence of systemic embolism and major bleeding appeared to be similar in the two trial groups.

Less serious bleeding ran the other way. Clinically relevant nonmajor bleeding occurred in 2.5% of the drug group and 1.6% of the comparison group. That category covers bleeding that needs medical attention without counting as major. That gap was within the range of chance, and with so few events the trial could have missed a real one. Taken together, major or clinically relevant nonmajor bleeding was 2.7% versus 2.0%.

How the trial fits with earlier evidence on stroke risk

Before this trial, the argument rested on what happened to untreated patients in national registries. Two studies published in early 2015 in the Journal of the American College of Cardiology reached different answers.

The first was a retrospective study of 140,420 patients with AF in Swedish nationwide health registries, retrospective meaning that it looked back at existing records. It found that the stroke rate attached to a score of 1 depended heavily on what was counted as a stroke. For men with that score, the annual rate was between 0.5% and 0.7%, depending on which event definition was used. Its authors concluded that the risk of ischemic stroke in patients with AF and a CHA2DS2-VASc score of 1 seems to be lower than previously reported, and their title called a benefit from anticoagulation unlikely.

The second drew on Danish registers and came out more worried. Among untreated patients, one additional risk factor increased the stroke rate at 1 year to 1.55 per 100 person-years. That is roughly 1.55 strokes a year for every 100 such patients, and about three times the rate in patients with no risk factor. Bleeding and deaths were also more frequent than in patients with no risk factor.

The new trial offers a check on both. In its untreated group, 1.1% of patients had a stroke over two years, or a little over half a percent a year. That is close to the Swedish estimate and about a third of the Danish one. Even at that lower rate, the group given a blood thinner had fewer events.

What the trial adds to the older drug trials is the comparison. A meta-analysis, meaning a pooled analysis of separate trials, was published in The Lancet in 2013. It covered all 71,683 participants included in the four main trials of the newer anticoagulants. The newer drugs reduced stroke or systemic embolic events by 19% compared with warfarin. Major bleeding was similar, but increased gastrointestinal bleeding was seen, meaning bleeding in the stomach or gut. Those trials set one blood thinner against another. None had a group that took nothing.

What other cardiologists made of the blood thinner trial

Jonathan Piccini of Duke University School of Medicine, who was not among the trial’s authors, welcomed it in a commentary for Healio, a medical news site. “This is a question faced by many clinicians and patients and the trial really helps provide guidance,” Piccini wrote.

“I suspect that more patients with a single risk factor will be offered oral anticoagulation,” Piccini added. Piccini reports receiving research grants from several organizations, among them Bayer and Bristol Myers Squibb, companies that market the two drugs used in the trial.

The editorial that accompanied the paper was more guarded. Its author, Paulus Kirchhof, wrote that anticoagulation for patients with one risk factor may be encouraged in light of the results. Kirchhof added that “in patients with a low AFib burden, no therapy may be preferred”. AFib is a common short form of atrial fibrillation, and burden means how much of the time the heart spends in the abnormal rhythm.

The investigators attached their own warning. They stress that the results should be interpreted with caution because fewer events occurred than they had expected and the trial was small.

Limits of the trial of blood thinners at intermediate risk

Few events. Seventeen participants out of 1,803 had an event in the main measure. With numbers that small, a few events falling differently could have changed the verdict, and one end of the plausible range sits at a difference of 0.1 percentage points.

No blinding. Patients and doctors knew who was on the drug. The reviewers who judged events did not, which protects the counting. Knowing can still shape who reports symptoms and who is sent for a scan.

One country, selected patients. The study included only patients from South Korea, and the balance of stroke and bleeding risk may differ elsewhere. People over 80 and those with serious kidney, liver or structural heart disease were not enrolled. Fewer than one in four participants was a woman.

One risk factor is not one kind of patient. A 68-year-old man with no other conditions and a 45-year-old man with diabetes both qualify. Results looked consistent across the subgroups the investigators had planned to examine, but with 17 events such comparisons carry little weight.

Two years. The trial stopped counting at 24 months. A person who starts a blood thinner at 60 may take it for decades, and the trial cannot show whether the benefit holds, grows or is overtaken by bleeding over that span.

Funding. The trial was funded by the Ministry of Health and Welfare, South Korea, and others. The money came mainly from the ministry and in part by Hanmi and Samjin, two South Korean drug companies. Joung reports research grants from both, and from four other companies.

Whether a blood thinner makes sense for a particular person with atrial fibrillation depends on their own risks of stroke and of bleeding, which is assessed by their doctor. Do not stop or change a prescribed medicine without talking to your doctor.

In what its investigators call the first randomized trial in people with atrial fibrillation and one other stroke risk factor, 0.5% of those given a blood thinner had a stroke, clot, major bleed or cardiovascular death within two years, against 1.5% of those given none, a result that rests on 17 events in an unblinded study from one country.