Hastrup Review of Science

← All episodes

Deep divefrom Episode 14Heart & blood16 min

Which Cholesterol Drugs Prevent a First Heart Attack

A deep dive into one paper: what was known before, what the researchers did, what the numbers say and what they don't.

The paper: Gnanavelou et al. 2026, eClinicalMedicine: Comparative effectiveness of lipid-lowering therapies for primary cardiovascular prevention (open access)

MP3

Chapters

Transcript

The script the two AI voices read from. The times are worked out from the script, not heard, so a line can start a few seconds early or late.

Read the whole episode

Which pills prevent a first heart attack

Host 1If you have never had a heart attack, which cholesterol-lowering treatments actually cut your chance of a first one, and by how much?

Host 2Pooling twenty-nine trials and more than two hundred and eighteen thousand people, this analysis found that for people at high risk, a strong statin cut major heart events by about a fifth, and adding an injection called evolocumab cut them by about two fifths. Two other popular add-ons, a purified fish oil and the fibrates, showed no clear benefit.

Host 1Welcome to the Hastrup Review of Science, where artificial intelligence reads real science, so you don't have to pretend you understood the paper.

Host 2Today is one paper, published this month in the journal E Clinical Medicine, by a team led by M. Gnanavelou. It tries to put every major cholesterol drug for first-time prevention on one scoreboard.

How cholesterol became a target

Host 1Before we get to the scoreboard, why did doctors start going after cholesterol in people who were not yet sick at all?

Host 2Because the first heart attack is often the event that kills, so waiting for it is a poor strategy. In the wider history of the field, studies through the middle of the last century kept finding that people with more cholesterol in their blood had more heart attacks, and the fatty plaques in arteries turned out to be loaded with it.

Host 1But finding cholesterol in the plaque is not the same as proving that lowering it helps, so how did they get from one to the other?

Host 2With drugs and randomised trials. The statins came out of a Japanese fungus screen in the nineteen seventies, and in the nineteen nineties big trials showed they lowered heart attacks, first in people who already had heart disease and then in healthy people with high cholesterol.

Host 1What exactly is the difference between treating people who already had a heart attack and the people this paper is about?

Host 2The first group is called secondary prevention, and the second, people without a prior event, is called primary prevention. Primary prevention is the harder case, because most of these people would never have had a heart attack anyway, so the benefit per person treated is smaller.

What was already known

Host 1So by the time this paper came along, what did doctors already feel fairly sure about?

Host 2That statins work in primary prevention, and that the more they lower the bad cholesterol, the more they cut risk, which is a principle the field has leaned on for decades. Guidelines broadly offer a statin when someone's estimated ten-year risk passes some threshold.

Host 1And what are the other drugs that have joined the statins since then?

Host 2Several. There are the newer injections that block a liver protein which normally limits how much cholesterol the liver can clear, of which evolocumab is one, a tablet called bempedoic acid often used by people who cannot tolerate statins, purified fish oil in the form of eicosapentaenoic acid, and the older fibrates, which mainly lower other blood fats.

The question nobody had settled

Host 1If all of these already have their own trials, what question was still left open?

Host 2Which one helps most, for whom. Almost every trial compares one drug against a placebo or against a statin alone, so nobody has run the big head-to-head that lines up all of them in the same people.

Host 1So a doctor wondering whether to add an injection or a fish oil capsule to someone's statin has no single trial to point to?

Host 2Correct, and the abstract opens by saying exactly that: comparative effectiveness across these classes in primary prevention remains uncertain. On top of that, results from people at high risk may not carry over to people at low or moderate risk.

How the comparison was built

Host 1So how do you compare drugs that were never actually tested against each other?

Host 2With a method called a network meta-analysis. You collect every relevant trial, draw each drug as a point and each trial as a line between two points, and then you can compare two drugs that never met, through the common partners they each did meet.

Host 1Can you give me an everyday picture of how that indirect comparison works?

Host 2Think of two runners who never raced each other, but both raced the same third runner. If one beat that runner by a minute and the other by two, you can estimate the gap between them, as long as the races were run under similar conditions.

Host 1And that last part sounds important, because what if the races were run on very different days?

Host 2That is the main weakness of the method, and the field calls the required assumption transitivity: the trials have to be similar enough in their patients and settings that the comparison is fair. If they are not, the indirect estimate can be misleading even when the arithmetic is perfect.

Host 1Where did they find the trials, and which ones were allowed in?

Host 2They searched the two big medical databases, MEDLINE and EMBASE, from their start up to the twenty-fifth of July twenty twenty-six. They kept randomised trials of lipid-lowering drugs in adults without a prior coronary event, and the trials they ended up with were published between nineteen eighty-four and twenty twenty-six.

Host 1Forty-two years of trials is a very long stretch, so how many did they end up with in total?

Host 2Twenty-nine trials, with two hundred and eighteen thousand, six hundred and fifty-seven unique participants. Unique means each person is counted once, even if the same trial shows up in several comparisons.

Host 1You said the answer might differ for high-risk and low-risk people, so did they mix everyone together?

Host 2No, and that is one of the better design choices. They split the analysis by baseline risk of atherosclerotic cardiovascular disease, which is the artery-clogging kind, so high-risk people were analysed separately from people at low to moderate risk.

Host 1What did they do with the statins, given that there are many brands and many doses?

Host 2They grouped statin regimens by intensity, meaning how hard they push cholesterol down, so a high-intensity statin is a strong drug at a strong dose and a moderate-intensity statin is a milder one. Everything was measured against a common baseline of no lipid-lowering therapy at all.

Host 1The summary mentions a random-effects model, so what does that mean in plain words?

Host 2It means they assumed each trial measures a slightly different true effect, because populations and methods differ, rather than all trials chasing one single number. It is the more cautious choice and gives wider, more honest uncertainty ranges.

Host 1And what counted as the drug working, since people do not simply have one kind of heart event?

Host 2The main yardstick was three-point MACE, short for major adverse cardiovascular events. In the field it usually means a heart attack, a stroke or a death from cardiovascular causes, whichever comes first, and the fibrate results used a four-point version that adds one more kind of event.

What the numbers show

Host 1Before the numbers, how should I read a hazard ratio when I hear one?

Host 2A hazard ratio of one means no difference from the comparison group. Below one means fewer events, so zero point eight means roughly twenty percent fewer, and every estimate comes with a ninety-five percent confidence interval, the range of values the data comfortably allow.

Host 1So if that range crosses one, the drug might be doing nothing at all?

Host 2Yes, that is the practical rule. An interval running from below one to above one means the data cannot rule out no effect.

Host 1Then start with the high-risk group: what did a strong statin do on its own?

Host 2Against no lipid-lowering therapy, high-intensity statin had a hazard ratio of zero point eight, with an interval from zero point six nine to zero point nine two. That is about a twenty percent cut in major events, and the whole range stays below one.

Host 1And the injection, evolocumab, added on top of a statin?

Host 2Evolocumab plus statin came out at zero point five nine, with an interval from zero point four nine to zero point seven one, so about forty-one percent fewer events than no treatment. Compared directly with its statin comparator, evolocumab still added benefit, at zero point seven four, interval zero point six five to zero point eight three.

Host 1So if I have this right, the injection is not just riding on the statin, it adds roughly a quarter fewer events on top of it?

Host 2That is a fair reading of the zero point seven four, with one precision: it is relative to the statin regimen used in those trials. In the wider picture, a large evolocumab trial in people without a prior heart attack or stroke was reported in twenty twenty-five, and results like this one probably lean heavily on it.

Host 1What about bempedoic acid, the tablet for people who cannot take statins?

Host 2Against no lipid-lowering therapy, it came out at zero point six four, interval zero point four eight to zero point eight five. That is a real benefit, but the interval is wide, which tells you the evidence base is thinner.

Host 1Wait, zero point six four is a smaller number than the statin's zero point eight, so does that mean bempedoic acid beats a strong statin?

Host 2No, the paper's abstract does not claim that, and the numbers do not support it. The two intervals overlap a lot, the bempedoic acid estimate rests on far fewer trials, and the people in those trials were often quite different, so ranking them on the central number alone would be a mistake.

Host 1Then what happened with the purified fish oil?

Host 2The purified fish oil against its statin comparator came out at zero point eight one, but the interval runs from zero point six two to one point zero six. Because it crosses one, the paper counts it as not retaining a clear benefit.

Host 1And the fibrates, which have been around the longest of the add-ons?

Host 2For the fibrates, measured with four-point MACE in people at low to moderate risk, the hazard ratio was one point zero four, interval zero point eight to one point three six. That is essentially no signal in either direction.

Host 1Now the lower-risk people, which is probably most of the people listening, so what did statins do for them?

Host 2The benefit rose with statin intensity. High-intensity statin came out at zero point five three, interval zero point three seven to zero point seven five, and moderate intensity at zero point seven six, interval zero point six four to zero point nine one.

Host 1That is odd, because the strong statin looks better in low-risk people than in high-risk people, so how can that be?

Host 2Two things to keep in mind. These are relative reductions, and in the wider field it is well known that a big relative cut in a small risk can still be a small number of events saved, and the abstract we have does not give the absolute numbers.

Host 1You said there were two things to keep in mind, so what is the second one?

Host 2The low-risk estimates rest on a different and smaller set of trials, which shows in the wider interval of zero point three seven to zero point seven five. The two risk groups are better read as separate analyses than as a contest between them.

Host 1This is a meta-analysis, so let me check one thing: it does not include any new patients at all?

Host 2Correct. It re-analyses trials that already exist, so its value is in lining them up, not in new data.

A word from our sponsor

Host 1This episode is brought to you by the Tefal non-stick frying pan, which its makers are, in the gentlest possible way, recalling.

Host 2Tefal has not paid for this, has not been asked, and has, as far as we know, never heard of us.

Host 1In a head-to-head comparison against no frying pan at all, the Tefal reduced sticking by forty-one percent, with a confidence interval that never touches the hob.

Host 2Mine was stratified by baseline risk. At high risk nothing stuck, and at low risk the eggs left of their own accord.

Host 1Owners should return the pan if any of the following no longer stick to it: butter, Mondays, their own surname, or the floor.

Host 2Mine was returned last week. It slid out of the post office and is now under the Atlantic, still not sticking.

Why it matters

Host 1So what is the main lesson here, beyond a list of hazard ratios?

Host 2That in primary prevention, the drugs that lower the bad cholesterol strongly, the statins at higher intensity and the injection on top, carried the clearest benefit. The add-ons that work mainly on other blood fats, the fish oil and the fibrates, did not show a clear benefit against what people were already taking.

Host 1So for an ordinary person on a statin who wonders about a fish oil capsule or a fibrate, does this say not to bother?

Host 2It says the trial evidence for adding them to prevent a first major event is not convincing, at least in this analysis. It is not a personal prescription, and anyone already taking something should talk to their doctor before changing it.

Host 1And who might this actually change things for?

Host 2Mainly guideline writers and doctors choosing what to add when a statin is not enough or not tolerated. The evolocumab and bempedoic acid results give some support to using them in high-risk people who have not yet had an event, which has long been the less tested territory for those drugs.

Host 1What can we not conclude from it, even if we wanted to?

Host 2We cannot conclude which single drug is best for any one person, or that a cheaper pill is beaten by an expensive injection for everyone. And nothing in the abstract tells us about side effects or costs, which matter a great deal when choosing a lifelong treatment for healthy people.

The honest caveats

Host 1How much should I trust a pooled result that stretches from nineteen eighty-four to this year?

Host 2With some care. Background care, smoking rates and blood pressure treatment changed a lot over those decades, so a placebo group in nineteen eighty-four is not the same as one today, which strains the transitivity assumption we talked about.

Host 1Are there other weak spots that a careful listener should keep in mind?

Host 2Several. Some drugs rest on one or very few trials, so their place in the network is fragile, and risk was classed at the level of whole trials rather than individual patients, so people inside one trial could vary a lot.

Host 1And what about the fact that the comparators were not all the same?

Host 2That matters too. Evolocumab and the purified fish oil were judged against their own statin comparators, while others were judged against no therapy, and a statin comparator in one trial is not identical to one in another.

Host 1One more thing, how much of this are you reading from the full paper?

Host 2Our numbers come from the published abstract, which is cut off before the end of the low-risk results. So we have not covered absolute risk reductions, safety outcomes, or how the authors rated the certainty of each comparison, and those deserve a look in the full text.

What to watch next

Host 1So what should we keep an eye on from here?

Host 2Whether guideline panels pick this up when they next revise primary-prevention advice, especially for the injections in high-risk people. In the wider field, several newer drugs that lower cholesterol in other ways, including a twice-yearly injection and a tablet version of the approach behind evolocumab, are in large outcome trials, and when those report, a network like this one can simply add them.

Host 1And for the drugs that came out empty-handed here, is that the end of the story?

Host 2Not necessarily, but the bar is clear. A fish oil or a fibrate would need new trials in primary prevention showing a benefit on top of modern statin treatment, and so far this analysis does not find one.

Sign-off

Host 1So the short version is that cholesterol lowering by the right drugs prevents first heart events, the strong statin and the injection have the best evidence, and two common add-ons do not show a clear benefit?

Host 2That is a fair summary, with the caveat that it rests on indirect comparisons across four decades of trials. It is a careful map, not a verdict on any one patient.

Host 1That is the Hastrup Review of Science, and every weekend this same feed carries the roundup, the week's new papers in one short episode, for people who would like their science in smaller doses.

More from this week

Get the next one

A roundup of new papers every week and a deep dive into one paper most nights. Free, no account needed: paste this address into any podcast app that can add a show by its address, or find “Hastrup Review of Science” by name in an app that searches Podcast Index, like Fountain or Podverse.

https://reviewofscience.org/podcast.xml

Only want the weekly roundup, or only the deep dives? There are two smaller feeds: see Listen and subscribe.