ApoB vs LDL cholesterol: which matters more?

Because LDL particles vary in how much cholesterol each carries, the same LDL-C can sit on very different particle counts [5]. ApoB also counts VLDL, IDL and Lp(a), which LDL-C partly misses [5].

Most labs still calculate LDL-C rather than measure it, from total cholesterol, HDL-C and triglycerides. Formulas such as Friedewald assume a fixed ratio of triglyceride to cholesterol in VLDL, which fails when triglycerides are high or LDL-C is very low [5]. ApoB is measured directly, so it does not have that weakness.

• A meta-analysis of 12 studies with 233,455 people found relative risks per standard deviation of 1.43 for ApoB, 1.34 for non-HDL-C and 1.25 for LDL-C [1].
• In UK Biobank, only ApoB stayed linked to heart attack when ApoB, non-HDL-C and triglycerides were tested together [6].
• Genetic analyses tie the benefit of lowering lipids to the fall in ApoB, not to LDL-C or triglycerides on their own [7] [8].
• But another UK Biobank analysis found similar risk per standard deviation (1.23 for ApoB, 1.20 for LDL-C) and almost no gain in prediction from adding ApoB [4].

The older cohort data agree on direction. In the Swedish AMORIS study, ApoB was a stronger predictor of fatal heart attack than LDL-C in both men and women [9]. The practical reading: ApoB is the more faithful measure of the cause, while LDL-C remains a reasonable stand-in for most people most of the time.

These conditions produce smaller, cholesterol-depleted LDL particles, so LDL-C falls while ApoB stays high [5]. The NLA's regression shows how wide the spread is: in untreated people with non-HDL-C of 100 mg/dL, 95 percent of ApoB values ran from 52 to 78 mg/dL [5].

On treatment, the gap widens. An LDL-C of 70 mg/dL corresponds on average to an ApoB of 73 mg/dL in treated people but 63 mg/dL in untreated people [5]. When discordance occurs, risk follows ApoB: on statins, ApoB above the median with LDL-C below it carried a hazard ratio of 1.49 for heart attack [2].

In a pooled analysis of 17,035 people on all common statins and doses, average reductions were 42.1 percent for LDL-C, 39.6 percent for non-HDL-C and 33.1 percent for ApoB [3]. Expressed as population percentiles, treatment brought LDL-C down to the 21st percentile on average but ApoB only to the 55th [3]. A patient can therefore look well controlled on LDL-C while still carrying an average particle count.

Newer drugs behave similarly: PCSK9 antibodies lowered LDL-C by 52 percent and ApoB by 43 percent when used alone [10]. This gap is why European guidelines treat ApoB as a secondary target in people with diabetes, obesity, high triglycerides or very low LDL-C [11].

Non-HDL-C measures the cholesterol content of atherogenic particles, while ApoB estimates their number [11]. In statin-treated UK Biobank participants, adding ApoB to a model that already had non-HDL-C did not improve prediction [4]. Canada treats ApoB and non-HDL-C as interchangeable alternatives when triglycerides are high [12].

For the population as a whole, ApoB barely changes risk prediction [13] [4]. For an individual with discordant markers, it can change the picture [2]. The ApoB test page lists who guidelines suggest should test, and the LDL cholesterol hub covers LDL-C in depth.