Cholesterol and the lipid system

One route, not several categories

The liver loads fat and cholesterol into a large particle called VLDL, very low density lipoprotein, and releases it into the blood. As that particle travels, tissues take the fat out of it for fuel or storage. It shrinks, and because the fat is leaving faster than the cholesterol, what remains becomes proportionally more and more cholesterol.

That is the whole story of the names. The large starting particle, the half-emptied one, and the small cholesterol-rich one at the end are the same vehicle at 3 points in one delivery run. It leaves as VLDL. As the fat comes out it becomes IDL, intermediate density. What is left at the end is LDL. They are not 3 different substances. And every one of them carries exactly one ApoB molecule, from the moment it leaves the liver to the moment it is taken back up. That is why counting ApoB counts the entire convoy regardless of where each vehicle has got to.

HDL runs the other way. HDL particles are released small and largely empty, and their job is collection, taking cholesterol out of tissues, including out of the foam cells in an artery wall, and carrying it back to the liver for disposal. That return journey is the reason HDL is called the good one.

What the profile looks at

Total cholesterol is everything added together, which makes it the least informative number on the panel. HDL cholesterol and LDL cholesterol split it by the type of particle carrying it. Non-HDL cholesterol is total minus HDL, which gathers everything on the outbound route into one figure.

In the UK, non-HDL is the number that matters most, and that is a deliberate choice. Three reasons. It captures every particle that can get into an artery wall rather than just the LDL ones, so remnants and lipoprotein(a) are counted too. It does not need a fasting sample. And unlike LDL it is a straight subtraction rather than a calculation that breaks when triglycerides are high, which is explained below.

National guidance uses non-HDL as the main target rather than LDL. For someone starting treatment to prevent a first heart attack or stroke, the aim is a reduction of more than 40% in non-HDL. For someone who has already had one, the target is an LDL of 2.0 or below, or a non-HDL of 2.6 or below. A total cholesterol above 9.0, or a non-HDL above 7.5, prompts specialist assessment for familial hypercholesterolaemia rather than ordinary treatment. That is an inherited condition where the liver cannot clear LDL properly, so levels run high from birth.

European and American guidance builds its targets around LDL instead. Same biology, different chosen number.

The trade-off is straightforward. LDL has almost all of the trial evidence behind it. Decades of drug trials were designed and reported around LDL targets, so that is the number the literature speaks in. Non-HDL is the more robust measurement, since it counts everything atherogenic, needs no fasting, and cannot be broken by a calculation failing. The UK picked the number that is harder to get wrong. Europe and America picked the number the evidence was built on. Bear that in mind before comparing a British report against advice written elsewhere.

Why LDL is calculated, and what the newer equations fixed

Direct LDL assays do exist, but they cost more, they are not standardised between labs, and UK guidance advises against using them when triglycerides are high, which is the situation people assume they are for. It suggests measuring ApoB instead. It is worked out by taking total cholesterol, subtracting HDL, and subtracting an estimate of the cholesterol still riding on the triglyceride-carrying particles. The original equation estimated that last part by dividing the triglyceride value by a fixed number.

The fixed number is the flaw. The real relationship between triglycerides and the cholesterol those particles carry is not constant, and the equation fails in two specific places. It fails when triglycerides are high. And it fails when LDL itself is low, which is precisely the situation of somebody on a statin, or a statin plus ezetimibe, being measured against a target. In both cases it underestimates, which means it can make a result look better than it is.

Two newer equations are more accurate at exactly the points where the old one breaks. They need no extra sample, no extra cost and no new assay, just different math. But many labs still use the original.

There is a noticeable consequence. On the older equation, labs stop reporting an LDL at all once triglycerides pass a certain point, because the result is no longer trustworthy. On a newer one, that upper limit is roughly twice as high. If an LDL is simply missing from a report with high triglycerides, that is usually why.

Triglycerides

Triglycerides are a different kind of fat and the most volatile thing on the panel. They are very much affected by what you ate and drank in the last few hours.

At the high end they stop being a cardiovascular question and become a pancreatic one. Pancreatic lipase breaks down the flood of triglyceride-rich particles and dumps a large quantity of free fatty acids into the tissue itself, which are toxic to the pancreatic cells and to the small vessels feeding them. The sheer number of particles also thickens the blood enough to sludge in those capillaries. Local damage plus reduced blood supply is pancreatitis.

National guidance treats a triglyceride level above 20 mmol/L as grounds for urgent specialist review, as long as it is not down to heavy drinking or poorly controlled blood sugar, and a level that stays above 10 on a repeat test as grounds for seeking specialist advice. The guidance does not spell out why, but at those levels the worry is the pancreas rather than the arteries. That is the same connection the Pancreas section makes from the other end, where very high triglycerides sit behind a minority of pancreatitis cases.

Counting particles rather than weighing cargo

ApoB is the reason for the diagram above. There is exactly one apolipoprotein B molecule on each particle travelling from the liver out to the tissues, so measuring ApoB counts vehicles rather than weighing what is inside them. Two people can carry identical LDL cholesterol in very different numbers of particles, and it is the particle number the artery wall responds to. ApoA1 is the matching protein on HDL particles, but it does not count them in the same way. A HDL particle carries anywhere from 2 to 5 copies of it, so ApoA1 measures how much collecting protein is in circulation rather than how many collectors there are.

Lipoprotein(a), and one story that has not held up

Lipoprotein(a) is an ordinary LDL particle with an extra protein bolted onto it. The extra protein is between 75% and 100% identical in sequence to plasminogen, the protein your body uses to dissolve clots. The resemblance is not a coincidence of shape; they are related proteins. But the working end of it, the part that would do the cutting, is inactive.

The obvious conclusion from that has not survived testing. In the lab, the protein does compete with plasminogen and does slow clot breakdown, and for years that was taken as the explanation for why lipoprotein(a) raises risk. In people it has not held up. Lowering lipoprotein(a) sharply does not measurably change how fast clots dissolve, and genetic evidence does not support it causing venous clots.

What is established is that it causes artery disease and calcification of the aortic valve, and the better-supported route is a different one. Lipoprotein(a) carries a disproportionate share of the oxidised fats circulating in blood, and those are among the strongest drivers of the inflammatory response described in the section above. So a striking structural resemblance turns out to be the less important half of the story.

It is also almost entirely inherited, set by the number of repeats in a single gene, and stable across a lifetime. Which means that unlike everything else on this panel, measuring it once usually tells you what you need to know for good.

What a raised one actually changes

UK guidance grades the risk across four bands rather than drawing one line, and those bands are in the Biomarker Guide along with the caveats that belong to them. They were derived largely from one northern European population, and guidance elsewhere draws its lines in different places.

That limitation is not a technicality if you are not of northern European background. Lipoprotein(a) levels vary substantially between populations. They run highest in people of African descent, higher again in South Asian populations than European, and lower in East Asian. The bands were set using a Danish population. So somebody of African or South Asian background is being measured against numbers that were never derived from people like them, and the organisation that published them says openly that more work is needed. The measurement is still worth having. Just know that the band it lands in is less certain than it looks.

The part that belongs here is what that guidance says to do about a raised one, because it is not what people expect. The recommended response is not to attack the lipoprotein(a) itself. It is to reduce everything else that can be reduced, particularly non-HDL cholesterol. It is a marker that tells you how hard to work on the rest of the panel rather than a number to drive down. That is the same structural point the Inflammation section makes about hs-CRP, and for the same reason.

One practical note. A raised lipoprotein(a) is occasionally secondary to something else, with kidney disease and an underactive thyroid the usual candidates.

Two caveats

HDL is routinely described as good cholesterol, and the picture is messier than that. The collection route is real, but trials that successfully raised HDL have not reliably produced the benefit that description implies, and a high HDL is not a credit that offsets the rest of the panel.

And the HDL as a percentage of total cholesterol figure on our profile is a ratio worked out from two other numbers, not a separate measurement.

What changes when this system is under strain

Nothing. Lipoprotein particles accumulating in an artery wall produce no sensation whatsoever, sometimes for decades. The first symptom of the process can be the event it causes.

There is no version of this you can feel your way to. It is measured or it is unknown.