Biomarker

HDL and LDL cholesterol: the two numbers everyone has

Lowering LDL prevents heart attacks in trial after trial. Raising HDL with drugs has not. That asymmetry is the most useful thing to know about a lipid panel.

15 October 2026 6 min read Editorial policy

Cholesterol is carried in the blood inside lipoproteins. LDL cholesterol is the cholesterol inside low-density lipoproteins, the main particles that deposit cholesterol in artery walls. HDL cholesterol is the cholesterol in high-density lipoproteins, which carry cholesterol away from tissues. Total cholesterol is the sum of all of it. All three are reported in mg/dL in the US and mmol/L elsewhere (divide by 38.7). Many labs calculate LDL from the other values rather than measuring it directly, which becomes inaccurate when triglycerides are high or LDL is very low.

The guideline numbers

Cholesterol categories
MarkerLevelCategorySource
LDLUnder 100 mg/dLOptimalNCEP ATP III
LDL100 to 129Near or above optimalNCEP ATP III
LDL130 to 159Borderline highNCEP ATP III
LDL160 to 189HighNCEP ATP III
LDL190 or moreVery high; high-intensity statin without needing a risk scoreATP III; ACC/AHA 2018
HDLUnder 40 mg/dLLowNCEP ATP III
HDL60 or moreHigh (once called protective)NCEP ATP III
Total cholesterolUnder 200 mg/dLDesirableNCEP ATP III

The 2018 ACC/AHA guideline moved away from fixed targets for most people and towards treating according to 10-year risk, with LDL 70 to 189 mg/dL assessed alongside that risk. European guidelines set LDL goals by risk tier, down to under 55 mg/dL at very high risk. This is one of the few places where "optimal" really is guideline language, from ATP III. But "the higher the HDL the better" is not supported, as below.

What raises and lowers them

  • LDL up: genetics (familial hypercholesterolaemia), saturated and trans fat, an underactive thyroid, kidney disease with protein loss, some drugs, and very-low-carbohydrate high-saturated-fat diets in some people.
  • LDL down: statins, ezetimibe, PCSK9 inhibitors, soluble fibre, replacing saturated with unsaturated fat, weight loss.
  • HDL down: smoking, inactivity, excess weight, high triglycerides, refined carbohydrate.
  • HDL up: exercise, stopping smoking, weight loss over time, and alcohol (not a reason to drink).

What it predicts

For LDL, the trial evidence is about as strong as preventive medicine gets. The Cholesterol Treatment Trialists' 2010 meta-analysis of 26 randomised trials and about 170,000 people found that each 1.0 mmol/L (about 39 mg/dL) reduction in LDL cut the yearly rate of major vascular events by just over a fifth, with no threshold below which the benefit stopped.

HDL behaves differently. In observational data, low HDL goes with higher risk, but drugs that raise HDL have not reduced events. And very high HDL is not protective: in 116,508 people from two Copenhagen cohorts (European Heart Journal, 2017), the link with death from any cause was U-shaped. Compared with the lowest-risk level, men with HDL of 3.0 mmol/L (116 mg/dL) or more had an adjusted hazard ratio of 2.06, and women with 3.5 mmol/L (135 mg/dL) or more had 1.68. Observational, and only a fraction of a percent of people are that high, but it ends the idea that more is always better.

On GLP-1 drugs and weight loss

From the Wegovy and Zepbound labels; secondary endpoints
MeasureSTEP 1, semaglutide 2.4 mgSURMOUNT-1, tirzepatide 15 mgSELECT, semaglutide 2.4 mg
LDL-2.5% vs +1.3% placebo-7.1% vs -1.7%-5.3% vs -3.1%
HDL+5.2% vs +1.4%+8.0% vs -0.7%+4.9% vs +0.6%

So LDL moves a little and HDL rises a little. Neither is large enough to replace a statin for someone who needs one. The triglycerides page covers the bigger change.

How to test

  • Fasting is not routinely needed per the 2016 EAS/EFLM consensus; a normal meal changes LDL and total cholesterol by about 8 mg/dL on average and HDL not at all.
  • Be consistent: same fasting state, same lab, no acute illness, which can lower cholesterol for weeks.
  • Know how LDL was reported. Calculated (Friedewald or Martin-Hopkins) or measured directly; if triglycerides are high, ask for direct LDL or ApoB.

Tracking it in Peptide AI

In Peptide AI, LDL, HDL and total cholesterol sits in the metabolic panel. The app reads it from a lab PDF or manual entry, scores the panel from 0 to 100 alongside an overall grade, flags values that warrant prompt physician consultation, and plots each result against your dose timeline. Log the draw date, whether you fasted, and anything unusual in the days before, such as illness, a hard session or a dose step, so the next result is compared like with like. Bloodwork tracking shows the series next to the dose log, weight and side effects, and the panels page lists every marker the app groups. More markers are in the bloodwork guides.

Questions people ask

What is a good LDL cholesterol level?

NCEP ATP III calls under 100 mg/dL optimal and 190 or more very high. The 2018 ACC/AHA guideline treats LDL by overall risk, with 190 mg/dL or more an indication for high-intensity statin therapy.

Is a very high HDL good?

Not necessarily. In two Copenhagen cohorts, men with HDL of 116 mg/dL or more had about twice the mortality of those at the lowest-risk level. Drugs that raise HDL have not reduced heart attacks.

How much does lowering LDL reduce risk?

In the Cholesterol Treatment Trialists meta-analysis, each 1.0 mmol/L (about 39 mg/dL) lower LDL cut the yearly rate of major vascular events by just over a fifth.

Do GLP-1 drugs lower cholesterol?

Modestly. LDL fell 2.5% on semaglutide 2.4 mg against a 1.3% rise on placebo in STEP 1, and 7.1% on tirzepatide 15 mg against 1.7% in SURMOUNT-1. HDL rose 5 to 8%.

Do I need to fast for a cholesterol test?

Not routinely. The EAS/EFLM consensus found fasting changes LDL and total cholesterol by about 8 mg/dL on average and does not affect HDL.

Free on iOS and Android

Log it once. See it all on one timeline.

Doses, food, training, labs, sleep and scans on one timeline. On iOS and Android.

The PeptideAI protocol screen with a weekly GLP-1 dose on it.