The Lp(a) Test Everyone Should Get Once, and What It Can't Tell You

Since March, American cardiology has asked something new of every adult: get your lipoprotein(a) measured at least once. The 2026 dyslipidemia guideline from the American College of Cardiology, the American Heart Association and nine partner societies made it a Class I recommendation, the strongest grade a guideline can give, and on September 25 the National Lipid Association repeated the call in a practical guide written for primary care.

The logic is tidy. Lp(a) is a cholesterol carrying particle whose level is set mostly by genes, and the guideline describes it as relatively stable over a lifetime. Measure it once, file the number, move on. But a test you take once has to be a test you can trust once, and the research published this year complicates the tidy version in three ways. The number wobbles, the risk climbs in steps rather than at a cutoff, and the drugs built to lower it are being judged on what they strip from the particle while the outcome that matters is years away.

What the guideline actually says

The recommendation itself is short. Every adult should have Lp(a) measured at least once. A level of 125 nmol/L (50 mg/dL) or more is associated with about 1.4 times the risk of heart attack or stroke over the long run; a level of 250 nmol/L, with at least twice the risk. According to the joint press release, lifestyle changes barely move Lp(a), "so repeat testing is generally not needed." An elevated result is meant to trigger more intensive LDL lowering, closer attention to other risk factors, and testing of close relatives.

What the result cannot trigger yet is a drug aimed at the particle itself. None is approved. The most advanced candidate, pelacarsen, lowered Lp(a) in a trial of 8,323 patients and still missed its primary endpoint, a story I covered when the topline result came out. For now, the test is a sorting tool, which makes the sorting matter.

One test, two answers

Here is where the "once" starts to fray. A registry from Singapore, presented as a conference abstract this summer, followed 271 patients whose first Lp(a) was 70 nmol/L or higher and who had at least two measurements, 1,036 in total over a median of 16 months. Depending on the starting band, between 27.7% (at 200 nmol/L or above) and 56.2% landed in a different severity category at their latest test. Among people who began in the intermediate band, 70 to 119 nmol/L, more than half moved. In the high band, 120 to 199 nmol/L, 47.4% did.

The same registry hints at why. Lp(a) was 15.9% higher at the next outpatient visit than during an acute coronary event. Treatment with a PCSK9 drug was associated with 21.8% lower levels, while statin use was associated with levels 12% to 14% higher. Much of the movement happens in exactly the patients cardiologists see most.

Other data push the other way. A Turkish laboratory study of 148 patients with repeat tests found 92.6% stayed in the same risk category, and no change exceeded the threshold the authors calculated for meaningful variation. But the intermediate group was again the weak spot, with only 68.8% staying put. In 224 children followed at a pediatric lipid clinic, 51% showed swings of more than 20%, and category changes clustered near the decision thresholds. The honest synthesis is narrower than the headline "one in two reclassified." Far from the cutoffs, one test usually holds. Near them, especially during illness or a change in treatment, it may not. The National Lipid Association's new guide now says as much, noting that specific conditions may warrant repeat testing, "especially for individuals with intermediate levels."

Risk climbs in steps, not at a line

The largest analysis of people who already have cardiovascular disease, published in the European Heart Journal using US claims data from 273,770 adults, followed them for a median of 5.4 years. During that time 41,687, or 15%, had another event. Compared with people below 15 nmol/L, the adjusted hazard ratios rose with each band: 1.04 at 15 to 79 nmol/L, 1.15 at 80 to 179, 1.29 at 180 to 299 and 1.45 at 300 or above. There was no plateau.

In absolute terms the gaps are real but not dramatic. Among patients not taking any LDL lowering drug, the rate of first recurrent events went from 23.2 per 1,000 person years in those under 80 nmol/L to 31.1 in those at 180 or above. By my arithmetic, that is roughly eight extra events per 1,000 people each year. The study has the usual limits of billing data: no deaths recorded, smoking and obesity poorly captured. It was funded by the Family Heart Foundation, which campaigns for wider testing. The excess risk looked smaller in people on the most intensive LDL lowering therapy, but the authors flag that comparison as prone to bias.

What lepodisiran strips from the particle

Lp(a) is not dangerous only because it carries cholesterol. It also ferries oxidized phospholipids, inflammatory molecules that many researchers suspect do much of the damage. In a post hoc analysis of the phase 2 ALPACA trial, presented at the European Society of Cardiology congress and published in JACC on August 31, Eli Lilly's lepodisiran cut those molecules sharply. In 213 participants with Lp(a) of 175 nmol/L or more who received two injections six months apart, the highest dose tested lowered oxidized phospholipids carried on apo(a) by 94.2% at day 240, placebo adjusted, and those carried on apoB by 88.5%. A year out, the reductions were still 80.7% and 79.9%.

The falls tracked Lp(a) itself almost perfectly, with a correlation of 0.94 at day 240. They did not track C reactive protein, a general inflammation marker. The authors are careful about what this means: the findings "provide additional biological rationale" for the phase 3 outcomes trial, "but do not establish clinical benefit." That trial, ACCLAIM-Lp(a), has enrolled about 17,300 patients, according to lead investigator Steven Nissen. The study was supported by Lilly.

I think the once in a lifetime rule is right, and I also think it is being sold a little too cleanly. For most people the number will land far from any cutoff and stay there, and one test really is enough. For the person sitting at 110 or 130 nmol/L, measured a week after a hospital stay or right after starting a new cholesterol drug, the label on the lab report is closer to a coin toss than a verdict. The response is also still blunt: a high Lp(a) today mostly means more aggressive LDL lowering, a lever cardiology already pulls. The lepodisiran data make the biology more convincing, but HORIZON already showed that a beautiful lab curve can coexist with a flat line on events. My guess is that the real value of universal testing will only show once a drug clears that bar, and that by then the people who tested near the line will want a second number.