Key takeaways
- HDL cholesterol measures cargo inside the particles, not particle number and not particle function.
- CETP inhibitors and niacin raised the level substantially and did not reduce cardiovascular events; one increased mortality.
- Mendelian randomisation shows genetic variants that raise HDL cholesterol from birth confer no protection, while the same method confirms LDL causality.
- Very high HDL cholesterol is associated with higher all-cause mortality, probably because extreme values often reflect dysfunctional particles.
- Low HDL matters as a flag for insulin resistance and high triglycerides; the target to act on is ApoB.
HDL is the clearest example in modern medicine of a marker being mistaken for a mechanism. The epidemiology was solid, reproducible and pointed in one direction for decades: higher HDL cholesterol, fewer cardiovascular events. Drugs were built on that observation. They raised HDL cholesterol substantially and did not prevent heart attacks — and one of them killed people. The failure was not a failure of the data. It was a failure to notice that HDL cholesterol measures the wrong thing.
What the number on your panel actually is
Start with what is being measured, because this is where the confusion begins. HDL cholesterol is not a measurement of HDL particles. It is a measurement of how much cholesterol is being carried inside them — cargo, not vehicles, and not vehicle performance.
HDL is not a single entity either. It is a heterogeneous population of particles varying in size, density, protein composition and function, continuously remodelled in circulation by enzymes and transfer proteins. Small, dense, newly formed particles behave differently from large, mature, cholesterol-loaded ones. A single cholesterol concentration averages all of that into one figure and discards everything that distinguishes them.
So two people with an identical HDL-C of 55 may have entirely different numbers of particles, entirely different particle composition, and entirely different functional capacity. The number cannot tell them apart.
What HDL is supposed to do
The protective functions attributed to HDL are specific processes, and each depends on particle quality rather than cholesterol content.
- Cholesterol efflux. HDL particles accept cholesterol from peripheral cells — including the macrophages inside an arterial plaque — and carry it to the liver for disposal. This is reverse cholesterol transport, and it is the mechanism the whole HDL hypothesis rests on.
- Antioxidant activity. Enzymes carried on the particle, notably paraoxonase-1, limit oxidation of LDL. Oxidised LDL is considerably more atherogenic than the unmodified form.
- Anti-inflammatory signalling within the vessel wall.
- Endothelial support, including effects on nitric oxide availability.
Note that the first of these is a rate — how effectively particles pick cholesterol up — and a static concentration cannot report a rate. A large pool of cholesterol sitting in HDL particles could mean efflux is working well, or it could mean cholesterol is arriving and not being delivered onward. Those are opposite situations with the same laboratory value.
Three separate lines of evidence, all pointing the same way
The drug trials. CETP inhibitors raised HDL cholesterol dramatically — in the region of 30-100% — and did not reduce cardiovascular events. Torcetrapib increased mortality and was abandoned (Barter et al., N Engl J Med 2007). Niacin raised HDL cholesterol and did not reduce events when added to statin therapy (Boden et al., N Engl J Med 2011). Raising the number changed nothing about the disease.
Human genetics. Mendelian randomisation provides a natural experiment: people carrying genetic variants that raise HDL cholesterol from birth should, if the relationship were causal, have lower lifetime cardiovascular risk. They do not. The same analytical approach applied to LDL-related variants shows exactly the causal relationship you would expect, which confirms the method works and that HDL is the outlier (Voight et al., Lancet 2012).
The upper tail. Very high HDL cholesterol is not associated with progressively lower mortality. Population data describe a U-shaped relationship, with elevated all-cause mortality at the extreme high end (Madsen et al., Eur Heart J 2017). The likeliest explanation is that extreme values often arise from genetic defects in HDL metabolism which leave particles cholesterol-loaded and functionally impaired — high cargo, broken vehicle.
Three independent methods — pharmacological, genetic and epidemiological — converging on the same conclusion is about as clear as this kind of evidence gets. HDL cholesterol is a marker of underlying metabolic health, not a lever.
The measure that does track outcomes
When cholesterol efflux capacity is measured directly rather than inferred, it predicts cardiovascular events independently of HDL cholesterol concentration (Rohatgi et al., N Engl J Med 2014). That is the finding that closes the argument: the functional property predicts, the concentration does not, and they can move in opposite directions in the same person.
Efflux capacity is not a routinely available clinical test. It remains a research assay. So the practical position is not "measure function instead" — it is "stop treating the concentration as a target you can act on."
Why low HDL still means something
None of this makes a low HDL cholesterol irrelevant. Values below roughly 40 mg/dL in men and 50 mg/dL in women do mark elevated risk — they simply do not mark it causally. Low HDL travels with insulin resistance, high triglycerides, visceral adiposity, smoking and inactivity, and it is those things that generate the risk.
The mechanism linking them is worth knowing because it explains why the two markers move together. When triglyceride-rich lipoproteins are abundant, cholesteryl ester transfer protein swaps triglyceride into HDL particles in exchange for cholesterol. The resulting triglyceride-enriched HDL is cleared faster, so HDL cholesterol falls. Low HDL is therefore, in most people, a downstream readout of triglyceride metabolism — which is why the triglyceride-to-HDL ratio carries more information than either component alone, and why it works as a rough surrogate for insulin resistance.
Treat low HDL as a flag pointing at metabolic health. Fix what it is pointing at.
What to use instead
| Marker | What it tells you | Actionable? |
|---|---|---|
| HDL-C | Metabolic health, indirectly | No — not as a target |
| ApoB | Number of atherogenic particles | Yes — the primary target |
| Triglyceride/HDL ratio | Insulin resistance and lipoprotein pattern | Yes — via metabolic change |
| Lp(a) | Inherited risk, measured once | Changes risk stratification |
ApoB counts particles, and each atherogenic particle carries exactly one ApoB molecule, so it measures the thing that actually enters and damages the arterial wall — the case is made in ApoB, the real cardiovascular number and ApoB versus LDL cholesterol. Lp(a) belongs on the panel once in a lifetime.
If you want to move it anyway
The interventions that raise HDL cholesterol as a by-product — aerobic exercise, loss of visceral fat, stopping smoking, reducing refined carbohydrate — are all worth doing on their own merits, and they improve particle function rather than just cholesterol content. That is the difference between a number rising because the system improved and a number rising because a drug forced it.
Alcohol raises HDL cholesterol too, and is the clearest illustration of why that is not a reason to do anything. It moves the marker without any evidence of moving the outcome favourably, and it brings its own risks with it.
The clinical pearl: the era of raising HDL is over, and it ended on unusually strong evidence — failed drug trials, negative genetics, and a mortality curve that turns upward at the top. Read HDL as a barometer of metabolic health, act on ApoB and the triglyceride-to-HDL ratio, and treat any product or protocol marketed on its ability to raise your HDL number as a red flag.
Bottom line
HDL cholesterol measures cargo, not particle count and not particle function, and it is the functional capacity that predicts outcomes. Drugs that raised the concentration failed; genetic variants that raise it from birth confer no protection; and the very top of the distribution carries higher, not lower, mortality. Low HDL remains a useful signal because of what it travels with — insulin resistance, high triglycerides, visceral fat — so the correct response is to address those rather than the number. For cardiovascular risk, use ApoB as the primary target and the triglyceride-to-HDL ratio as the metabolic read, and leave HDL alone.
Educational content, not medical advice. Laboratory interpretation and any treatment decision are made by a licensed physician after individual evaluation. Individual results vary.
