HDL Is the One Number on Your Panel Where Higher Stopped Meaning Better

Most of a standard lipid panel takes some explaining. HDL does not. It is the one line almost everyone believes they already understand: HDL is the good cholesterol, it carries plaque away from the arteries, and a high number is a win. It is the number people volunteer about themselves. It is the number that softens an otherwise mediocre panel.

The observational evidence behind that belief is genuinely strong. Across decades of cohort studies, people with higher HDL cholesterol have had fewer heart attacks, and the relationship is steep. In the Mendelian randomization work described below, a one standard deviation increase in HDL cholesterol was associated with a 38 percent lower odds of myocardial infarction in conventional observational analysis. That is not a weak signal. That is one of the most reproducible associations in cardiovascular epidemiology.

And yet over the past fifteen years, three independent lines of evidence have taken that association apart. Not the association itself, which still holds, but the interpretation of it. The current picture is that HDL cholesterol is a real and useful marker of cardiovascular and metabolic health, and a poor description of a mechanism you can push on. Higher is not reliably better. Very high may be worse. And the part of HDL biology that appears to matter is not the number your lab reports at all.

The genetic experiment that broke the story

The cleanest way to ask whether a biomarker causes a disease, short of a drug trial, is to let genetics run the randomization. People inherit variants at conception, independent of diet, income, exercise, and every other confounder that haunts observational data. If a variant raises HDL for a lifetime and carriers get fewer heart attacks, the association is likely causal. If carriers get exactly the average number of heart attacks, it is not.

Voight and colleagues did this in The Lancet in 2012 (DOI). They used a variant in the endothelial lipase gene, LIPG Asn396Ser, carried by about 2.6 percent of people, which raises HDL cholesterol by 0.14 mmol/L while leaving other lipids and non-lipid risk factors alone. Based on the observational relationship, that lifetime HDL advantage should have translated into roughly a 13 percent lower risk of myocardial infarction. Across 20 studies, 20,913 cases and 95,407 controls, the observed odds ratio was 0.99. No effect at all.

They then repeated the test with a 14-variant genetic score built to move HDL and nothing else. Same result: the observational association predicted protection, the genetic score delivered none. As a positive control they ran the identical method on a 13-variant LDL score, and that one behaved exactly as the observational data predicted. The method works. It just does not find a causal HDL effect.

Four drugs raised it. None of them helped.

Genetics is indirect. Drug trials are the direct test, and the record here is unusually consistent for a field where trials usually disagree.

Torcetrapib, a cholesteryl ester transfer protein inhibitor, raised HDL cholesterol by 72.1 percent and lowered LDL by 24.9 percent. The ILLUMINATE trial enrolled 15,067 high-risk patients and was stopped early, because the torcetrapib group had more cardiovascular events and more deaths, reported in the New England Journal of Medicine (DOI). That result is complicated by off-target effects on blood pressure and aldosterone, so it does not settle the question by itself.

Dalcetrapib, a CETP inhibitor without those off-target problems, raised HDL by 31 to 40 percent in 15,871 patients after a recent acute coronary syndrome. The dal-OUTCOMES trial found no reduction in the primary endpoint (DOI).

Niacin, the oldest HDL-raising drug in the pharmacy, got two large tests on top of statin therapy. AIM-HIGH added extended-release niacin to simvastatin in patients with low HDL and found no incremental benefit (DOI). HPS2-THRIVE tested extended-release niacin with laropiprant in over 25,000 patients and found no significant reduction in major vascular events, alongside a significant excess of serious adverse events (DOI).

Four agents, several mechanisms, tens of thousands of patients, large and verified increases in the number. No reliable clinical benefit. When a marker moves substantially and the outcome does not follow, the marker is usually reporting on something rather than causing it.

Very high HDL is not a trophy

If the story stopped at "raising it does not help," HDL would simply be a passive indicator. The population data are less forgiving than that.

Madsen, Varbo, and Nordestgaard pooled the Copenhagen City Heart Study and the Copenhagen General Population Study, 52,268 men and 64,240 women followed over 745,452 person-years, and published the mortality curve in the European Heart Journal (DOI). The relationship between HDL cholesterol and all-cause mortality was U-shaped in both sexes. The lowest mortality sat at 73 mg/dL in men and 93 mg/dL in women. Above that, risk climbed again. Men with HDL at or above 116 mg/dL had roughly double the all-cause mortality of men at the optimum (hazard ratio 2.06). Women at or above 135 mg/dL had a hazard ratio of 1.68.

Those are unusual values, not everyday ones. But they are values that get congratulated in ordinary clinical encounters, because the reference range on the report has no upper bound in the direction most people care about. A result flagged as excellent by the lab may sit on the far side of the curve.

What low HDL is actually telling you

The largest single reappraisal came from the CANHEART cohort: 631,762 Ontario adults aged 40 and older with no prior cardiovascular disease, linked across 17 data sources, followed for a mean of 4.9 years, published in the Journal of the American College of Cardiology (DOI).

The finding that matters is not that low HDL predicted cardiovascular death. It did. The finding is that low HDL also predicted non-cardiovascular death, including cancer and other causes, and that people in the lowest HDL groups carried a heavier burden of unhealthy behaviors and comorbid conditions generally. A marker that predicts death from causes with no plausible connection to reverse cholesterol transport is behaving like a barometer of overall physiological state, not like a specific arterial protectant.

This reframes a low HDL usefully. It is a prompt to look for what tends to travel with it: insulin resistance, elevated triglycerides, visceral adiposity, inactivity, smoking, chronic inflammation. Those are the things doing the damage, and several of them are measurable on the same draw.

The part of HDL that does seem to matter

There is a version of the HDL hypothesis that has survived, and it is not about concentration.

Cholesterol efflux capacity measures function: how well a person's HDL actually accepts cholesterol from macrophages, the first step of reverse cholesterol transport. Rohatgi and colleagues measured it in 2,924 adults free of cardiovascular disease in the Dallas Heart Study and followed them for a median of 9.4 years, in the New England Journal of Medicine (DOI).

Two results sit side by side in that paper. HDL cholesterol level was not associated with cardiovascular events in adjusted analysis (hazard ratio 1.08, confidence interval crossing 1). Efflux capacity was: the highest quartile carried a 67 percent lower risk than the lowest (hazard ratio 0.33), after adjustment for traditional risk factors, HDL cholesterol level, and HDL particle concentration. Notably, efflux capacity had minimal correlation with the metabolic variables that track with HDL cholesterol, which is why it survives adjustment.

The catch is practical: efflux capacity is a research assay. It is not orderable at a commercial reference laboratory, and it is not coming to an annual physical soon. The value of knowing it exists is interpretive. It explains why the concentration behaves like a proxy, and why a proxy stopped predicting once the trials pushed on it directly.

Reading your own HDL

None of this argues for ignoring the line. It argues for demoting it.

Treat HDL as a status light rather than a lever. A low HDL is a legitimate signal, and the useful response is to read it alongside the markers that describe the same underlying problem more directly: triglycerides, fasting insulin, hemoglobin A1c, waist circumference, hs-CRP. The triglyceride to HDL ratio carries more metabolic information than either number alone.

Keep the causal work on the causal markers. The evidence that survived every one of the tests above is the evidence for apolipoprotein B and LDL particle burden. When a panel shows a reassuring HDL next to an elevated ApoB, the ApoB is the finding and the HDL is decoration.

And do not read a very high HDL as a margin of safety. Values above roughly 90 mg/dL in men or 100 mg/dL in women deserve context rather than congratulation, particularly alongside heavy alcohol intake, thyroid abnormality, or a family pattern suggesting a genetic CETP variant.

Whether any individual result warrants further testing, and what to do about it, is a judgement for a licensed clinician reading the whole panel and the whole history. The general lesson is narrower and firmer than any single recommendation: an association can be strong, reproducible, and biologically plausible, and still fail to describe a mechanism you can act on. HDL is the best-documented example of that failure currently sitting on a routine blood test.

The Peptide Dispatch publishes research summaries for educational purposes. This material describes what the published literature reports about laboratory markers and physiology. It is not medical advice, it is not a diagnosis, and it does not recommend or endorse any treatment, product, or provider. Laboratory results require interpretation by a licensed clinician in the context of your individual history. Always discuss testing and treatment decisions with a qualified healthcare professional.

Research summaries in this article are based on articles retrieved from PubMed and Europe PMC.