Skip to Main Content

“Good” cholesterol might be in for a name change.

Raising HDL, widely known as good cholesterol, for years has been thought to protect against heart attack and stroke. But a big new study published Monday found little evidence it does.

The finding upends the advice doctors have been giving millions of patients — and helps explain why the drug industry has failed time and again, despite billions in investment, to develop a drug that cuts deaths from heart disease by boosting HDL levels.


“When you explain [cholesterol levels] to patients, it’s very easy to say one number’s bad and the other number’s good,” said Dr. Dennis Ko, a cardiologist at Canada’s Institute for Clinical Evaluative Sciences and lead author of the study. But it turns out that HDL is associated with poor health generally and does not seem to affect cardiovascular risk.

In the study, Ko and his colleagues looked at years of data from about 630,000 people in Ontario, sorting their HDL scores from low to high. Those with basement-level HDL were more likely to die of cardiovascular complications, but the risk did not drop steadily as good cholesterol levels rose.


Instead, it dipped, then hit a plateau; people with HDL of about 40mg/dL had roughly the same risk as those with about 80 mg/dL. And death risks actually increased for those with extremely high levels of good cholesterol.

Further muddying the picture, people in the low-HDL group were also more likely to die of diseases unrelated to the heart. And they had lower incomes, higher body weights, and poorer diets than others in the study, all of which correlate with increased mortality on their own.

HDL has been thought to lower cardiovascular risk by cleansing the bloodstream of “bad” cholesterol and scrubbing the inner walls of blood vessels, so your levels of HDL were thought to predict your risk of heart attack or stroke. But this new data suggests HDL may just be a fatty substance along for the ride.

“It may be therefore that it’s reflecting other health habits that lead to greater risk, rather than actually being a risk factor itself,” said Dr. Steven Nissen, a Cleveland Clinic cardiologist not involved in the study.

The study, published in the Journal of the American College of Cardiology, will sound familiar to the drug industry, which has repeatedly failed to design a pill that might improve patients’ lives by increasing HDL.

A decade ago, Pfizer spent more than $800 million to get the HDL-boosting medication torcetrapib into late-stage trials, only to find that more patients died on the drug than on placebo. Roche was next to fail when its drug, dalcetrapib, came up short in a 16,000-patient trial in 2012. And last year, Eli Lilly shut down a study testing its evacetrapib on 12,000 patients after discovering that the drug had no effect on heart attack and stroke.

Those drugs, called CETP inhibitors, are meant to interrupt the process that turns HDL into LDL cholesterol, its “bad” relative. And, in a sense, they work — Lilly’s pill boosted HDL by about 130 percent and slashed LDL by more than one-third. But, time after time, changing that ratio of good to bad cholesterol has failed to improve outcomes for patients.

Despite the litany of disappointments, there’s one CETP pill still in the ring. Merck is in the final phase of a 30,000-patient study on anacetrapib, the last of the class in late-stage development. Results are expected early next year, but optimism does not abound.

“I have very little hope that the ongoing trial is going to work,” said Nissen, who chaired Lilly’s failed evacetrapib study. “And if you think about it, if that study also fails, what is the likelihood that anybody will do another large, Phase 3 trial? I think the answer is probably pretty low.”

Still, cardiologists said it may still be useful to keep measuring HDL in blood lipid tests, because it does seem to be correlated with other factors that can affect health, such as diet and exercise habits. Doctors can gather such information with questionnaires, but those are time-consuming and rely on patients accurately recounting their habits.

And HDL may yet redeem itself in the world of drug development. Good cholesterol is complex, with various particle sizes and subspecies floating around in the body. Ko and his colleagues applied a broad brush in their study, looking at HDL as a monolith. It’s possible a deeper look at HDL subtypes could reveal a more nuanced relationship with cardiovascular disease, Dr. Stephen Nicholls and Dr. Peter Psaltis wrote in an editorial accompanying the study.

And it’s possible that drugs to raise HDL may work only in people with certain genetic mutations.

Last year, mining the wreckage of Roche’s failed dalcetrapib study, researchers at the University of Montreal noted that the drug had a marked effect on a subset of patients with mutations to a gene called ADCY9. DalCor Pharmaceuticals, a UK startup, has raised $150 million in venture capital to see if Roche’s drug can have a second life as a targeted therapy.

“Our understanding of the HDL story has changed,” said Dr. Bassem Masri, a cardiovascular disease expert at Weill Cornell Medicine and NewYork-Presbyterian who helped run a slew of CETP trials. “It’s not just [about] increasing HDL; it might be that only a specific population will respond to that increase positively.”

Elsewhere, scientists are studying novel methods of boosting HDL, whether by stimulating other pathways in the body or injecting patients with synthetic good cholesterol. Considering past failures, it looks like an uphill battle. But anything’s possible.

“I’ve been doing medicine long enough that I’ve learned to say you never say never,” Nissen said.

  • Genetic variation in health care re drugs and medical devices. This isn’t rocket science. Where is common sense? We are not one size fits all.

  • Hi i have a question my doctor make. Blood test for me a 2weeks ago and he told that i have 199level of good cholestrole but the problem he didnt explain what i can do to protect this and i dont eat anything have oil like butter or cake thank you

  • One further point – it is only in men with HDL over 90 in men where the association with mortality is really striking.
    According to table one there are 17,944 people with HDL over 90, of whom 86.4% are female and low risk. Out of a total population of 631,762.
    This means that only about 0.3% of the CANHEART population were in this high risk group, so the chances of the rest of us getting there by accident seem slim.

  • The first thing that needs to be said about this article is that no attempt was made to change the HDL cholesterol level in the CANHEART study, so the headline is inappropriate.
    One possible confounder is alcohol use. While the study attempted to adjust for this, such adjustments are hard to do with accuracy. In normal diet epidemiological studies alcohol intake is under-reported by a factor of 2-3. Further, alcoholics do not usually take part in those studies, so we have no idea what their level of under-reporting is, but in the nature of the disease it is likely to be higher. As it was, alcohol correlated with high HDL in CANHEART.
    Alcohol elevates HDL in moderate drinkers and may be associated with benefit, but excessive intake increases the proportion of useless HDL subfractions. Alcohol can also increase the triglyceride fraction, adversely effecting the TG/HDL ratio even when HDL remains high.
    In the lipid lowering trials (from which alcoholics and binge drinkers were presumably excluded) those in the highest quartile for HDL at baseline had the lowest rate of cardiac events and received no extra benefit from LDL- lowering. There was also an inverse relationship with triglycerides. High triglycerides correlated with low HDL in this study.
    The TG/HDL ratio is a fairly accurate proxy for the 2-hour insulin level after an OGTT. The HMG-CoA reductase enzyme which statins inhibit is actually activated by insulin and inhibited by glucagon, and drugs which lower insulin but do not lower LDL (empagiflozin and acarbose) by restricting the retention or appearance of glucose in the blood do reduce cardiovascular events and mortality in people with diabetes.
    People in the low HDL group also had a high cancer death rate, because hyperinsulinaemia is a known promoter of cancer; in the very high HDL group this association is almost certainly due to alcohol consumption.
    It goes without saying that dietary changes, especially those involving carbohydrate restriction, which can improve the TG/HDL ratio and lower fasting and 2-hour insulin, were not tested in the CANHEART study or any of the trials mentioned in this article.
    “It may be therefore that it’s reflecting other health habits that lead to greater risk, rather than actually being a risk factor itself” – exactly. Or, indeed, health habits that lead to reduced risk.
    If you only have a lipid-lowering drug hammer, every problem will look like an LDL nail.

    • Not sure why you’re hung up on TGs, since they have been proven over and over not to be an independent risk factor for cardiovascular disease. For example in Type 2 Diabetic Dyslipidemia, which I have, triglycerides typically run in the 250-300 mg/dl range. My internist is much more concerned with lowering my LDL cholesterol with atorvaststin to lower my CV risk than he is with my trigs.

      Early in my training in lipid metabolism (1970s) we were indeed taught that raising HDLc could be a good thing. Since it was known back then that exercise and alcohol consumption both raised HDLc the recommendation was jokingly to run from bar to bar to bar…

  • This article confirms what has been known for 15 years. Dr Thomas Bersot at the Cardiovascular Research Institute at UCSF has been studying the population of Turkey, where most people have genetically low levels of HDL cholesterol. Dyslipidemia treatment guidelines for Turks do not attempt to raise HDLc. Even though these folks have “normal” LDLc levels, the treatment aims to lower LDLc even further so as to improve the ratio of TC/HDL, as referenced by some below.

  • It’s about time the truth comes out. I myself stopped taking taking Lipitor several years ago after examining the evidence. I’m certainly not a genius – I’m not even in the health field – but there was plenty of evidence that cast serious doubt on claims. In fact, a Newsday (Long Island) investigative journalist spent 6 months made a very strong case it not only had harmful side effects which pharma knew about, but was of very doubtful efficacy. I myself realized after just a few months that my athletic performance, which had plateaued for several years, improved by 70%!!!! And several of my fellow athletes had exactly the same experience.

    And here’s the most interesting part: the rest of the media failed to follow up on all the exactly the same experience. I suspect the reason was the public was already too deeply committed (habituated?) to this new modern version of The Fountain of Youth.

    • Lipitor is for decreasing LDL, which is not in doubt!

      While I certainly understand that you might find the side effects to be worse than the increased risk, there is not doubt that decreasing LDL reduces risk. Try to reach a low cholesterol level by diet instead!

  • Years ago my primary physician told me that I had low HDL’s but that it was the ratio it had to LDL’s and triglycerides which he looked at. As I take a Statin and my ratio of LDL to HDL is about 3-1 he felt I was doing ok. Before the Statin it was 7-1. At 68 with no sign of heart disease I can no longer worry about my low HDL- thanks!

  • The article’s title is misleading: the study it reports on wasn’t an experiment in which HDL-levels were raised — either through lifestyle changes or meds — and the effects measured. And the jump to “HDL has been thought to lower cardiovascular risk by cleansing the bloodstream of “bad” cholesterol and scrubbing the inner walls of blood vessels, so your levels of HDL were thought to predict your risk of heart attack or stroke. But this new data suggests HDL may just be a fatty substance along for the ride.” doesn’t come from the study; it’s an extrapolation based on the study. Presumably there was some bio-chemical basis for the idea that HDL (or one of its subtypes) scrubbed blood vessels?

    In summary, not very good reporting, in my opinion.

  • I’m under the impression that HDL that one makes oneself is beneficial, but that artificially boosting HDL appears not to be helpful. So sadly for the drug industry, it appears to be lifestyle changes that benefit people the most in the cholesterol department. Speaking anecdotally as an athletic type, I see my HDL go up when I’m able to do longer distance workouts and I weigh less, and my HDL goes down during years when I work out less and gain weight, so it certainly seems that my higher HDL is reflecting better fitness. I have to doubt that there’s any way a pill or injection can duplicate that.

    I’m especially intrigued by the assertion here that “[D]eath risks actually increased for those with extremely high levels of good cholesterol.” What did those people die of? Heart disease, strokes, Type II diabetes? Or “all causes”?

    • Good question, but I suspect the answer in broad terms is pretty straightforward: with an appropriate large control group where it can be assumed that the sheer size of the study cancels out – if you will – other effects, the med is the probable cause. But beyond that, a large body of smaller studies points to the lack of efficacy. But it’s also juicy targets for the pharms to claim the usual “more study is needed”. But even then, of course, the loss of huge amount of revenue guarantees they won’t give up easily. Hence the desperate claim the drug might eventually prove its claimed worth.

  • Raising HDL is, apparently, not useful in most people. What about lowering LDL? Is that also not useful, or do these studies not bear at all on that issue?

    • Lowering LDL cholesterol in order to lower cardiovascular events is probably the MOST tested hypothesis in scientific history. See analysis of data from 174,000 people treated with statins in 27 clinical trials: . Sure, some people have side effects from statins, like all drugs, but lowering LDL works across race, gender, and age, and reduces not only CV mortality, but all-causes mortality, probably related to undetected events (e.g., subclinical stroke –> car accident) that are avoided while on statins. The HDL piece hadn’t been tested until the CETP inhibitors, and 3/4 have failed. So while there are questions about which kind of LDL, it’s very clear that reducing it with statins is useful, every single time they’ve been tested — across 7 different statins. Reassuringly in the sense that they’re not magic, they failed in trials to treat heart failure, which wouldn’t be expected to respond either to LDL reduction, or the various non-LDL-related mechanisms of statins.

Comments are closed.