In March 2026, 11 medical societies published a new US dyslipidemia guideline. For the first time it recommended that every adult get a lipoprotein(a) test at least once, and it gave apolipoprotein B (ApoB) an official place in US lipid care (NPR’s coverage is a good plain-language summary). Six months later, the first big trial of a drug built to lower Lp(a) came back negative. That tension is a good reason to ask what each test is for before you order it.
We picked five tests where the biology is clear, the guidance is current, and a result can change a conversation with your doctor. For each one we cover what it measures, how to read the number, and where the evidence stops. We left vitamin D off on purpose. In the NIH-funded VITAL trial of 25,871 adults (median 5.3 years, 2,000 IU a day), cancer rates were 6.1% on vitamin D versus 6.3% on placebo. Testing still makes sense if you suspect a deficiency, but it is not a longevity lever.
ApoB counts particles, not cholesterol 🧬
Every particle that can lodge in an artery wall (LDL, VLDL, remnants, and Lp(a)) carries exactly one ApoB protein. So an ApoB result is a head count of harmful particles. A standard LDL-C result measures the cholesterol cargo those particles carry. When each particle carries little cholesterol, which is common with diabetes or high triglycerides, LDL-C can look fine while the particle count is high.
Does that gap matter in practice? Probably, though the effect is modest. In a UK Biobank analysis of 375,544 adults with no heart disease and no lipid drugs, people with ApoB higher than their LDL-C predicted had a 11% higher risk of major cardiovascular events (HR 1.11, 95% CI 1.06 to 1.15). Those with ApoB lower than predicted had 13% less risk. In 13,015 statin-treated patients followed for a median of 8 years, ApoB and non-HDL cholesterol tracked residual risk better than LDL-C did. Both are observational studies, so they show association, not proof.
The evidence is not unanimous. A small 2019 NHANES analysis of 1,522 adults (162 with cardiovascular disease) found LDL-C associated with disease and ApoB not. It was cross-sectional and tiny, so we weigh it lightly, but it is a reason not to treat ApoB as magic.
The guideline’s position is narrower than the headlines. ApoB is considered reasonable for people who have atherosclerotic disease, type 2 diabetes, cardio-kidney-metabolic syndrome, or high triglycerides, to decide whether to intensify treatment once LDL-C goals are met (expert summary here).
Targets track the LDL-C goals: under 55, 70, or 90 mg/dL depending on your risk category.
Fasting is not required, because ApoB is measured directly rather than calculated.
Discordance (ApoB high while LDL-C looks fine) is the situation where the test earns its fee.
Ask your doctor: “Would ApoB change what we do given my triglycerides or blood sugar?”
Pull up your last lipid report. Is ApoB on it?
Lp(a) is a once-in-a-lifetime test, with an asterisk
Lipoprotein(a) is an LDL-like particle with an extra protein tail called apolipoprotein(a). It is 70 to 90 percent genetically determined, so one measurement in adulthood usually settles it. Roughly one in five people carry an elevated level, and a standard lipid panel does not include it. The 2026 guideline gave one-time testing its strongest recommendation, and the European Atherosclerosis Society had said the same in 2022.
How to read it:
Elevated generally means 125 nmol/L or higher, which is about 50 mg/dL.
Units do not convert cleanly between nmol/L and mg/dL, so check which one your lab used before comparing against any cutoff.
Lifestyle barely moves Lp(a). A high number is inheritance, not a personal failing.
Relatives of someone with a high result each have about a 50 percent chance of sharing it, so cascade testing is worth raising.
Now the asterisk. On September 4, 2026, Novartis (the trial sponsor, with partner Ionis) announced that Lp(a)HORIZON had missed its primary endpoint. The trial enrolled 8,323 patients with established cardiovascular disease and elevated Lp(a). Pelacarsen lowered Lp(a) substantially, but the drop did not significantly reduce heart attacks, strokes, or cardiovascular death in the overall population. Only topline figures are public, and the full data are still to come. Participants were already well treated, with mean LDL-C near 66 mg/dL. Every company, trial, and funding event connected to this story is tracked and searchable in LongevityHub Pro.
Our read is that the test is still worth taking, but a high result is not yet a reason to expect a drug. The National Lipid Association still advises measuring Lp(a), and other outcome trials of lepodisiran, olpasiran, and muvalaplin are running. Until they report, a high Lp(a) means a harder look at LDL-C, ApoB, and blood pressure. Currently, lipoprotein apheresis is the only approved treatment that lowers it directly.
HbA1c and fasting insulin show different stages of the same problem 🧪
HbA1c measures the share of your hemoglobin coated in glucose, which reflects average blood sugar over roughly two to three months. The American Diabetes Association’s prediabetes range is 5.7 to 6.4 percent, and fasting glucose of 100 to 125 mg/dL falls in the same zone. Most doctors order both without prompting.
Fasting insulin is the less common request, and the mechanism explains why people want it. When cells respond poorly to insulin, the pancreas compensates by making more of it. Glucose stays normal for years while insulin quietly climbs. Dividing the product of fasting insulin (µIU/mL) and fasting glucose (mmol/L) by 22.5 gives HOMA-IR, a rough index of that resistance.
The evidence is supportive but thin:
A 2010 study of 965 patients (29.7% had prediabetes) found that fasting insulin above 9.0 µIU/mL identified about 80% of them. It was cross-sectional, and the group was mostly white, obese, and female.
A 2025 diagnostic study in Lima found HOMA-IR had an AUC of 0.843 for prediabetes, which is good, not perfect.
Cutoffs are not standardized. Studies use values from about 2.0 in adults to 3.4 in a pediatric cohort, and insulin assays differ between labs.
We are more skeptical here than with the lipid tests. Fasting insulin correlates with prediabetes, but we did not find a trial showing that acting on a high insulin alongside a normal glucose improves outcomes. It is a reasonable add-on if your family history, waist circumference, or triglycerides worry you. It is not a verdict. Does your doctor know your triglyceride-to-HDL ratio? That cheap number often points the same direction.
hs-CRP is a decent risk marker and an unsettled target
High-sensitivity C-reactive protein is a protein your liver makes when inflammatory signals, mainly interleukin-6, rise. Inflammation drives plaque growth and rupture, so the idea of measuring it is sound. In the Physicians’ Health Study, among 1,086 healthy men, the top quartile of hs-CRP had a 2.9-fold higher heart attack risk than the bottom quartile.
Then came the intervention test. In CANTOS, 10,061 people with a prior heart attack and hs-CRP of 2 mg/L or higher received canakinumab, which blocks interleukin-1β, or placebo. Over a median 3.7 years it cut major cardiovascular events by a relative 15 percent without changing LDL-C. That was the first phase 3 proof that targeting inflammation can help. Novartis funded it, and canakinumab is not a routine heart drug.
The follow-up has been rougher. TCTMD reported that Novo Nordisk’s ZEUS trial of ziltivekimab, which targets IL-6, posted negative topline results in August 2026, with full data still pending. So hs-CRP is a reasonable marker, but whether lowering it helps most people remains open.
2 mg/L was the CANTOS entry threshold, and many labs flag higher values as elevated risk.
Above 10 mg/L usually means infection or injury, not artery risk. Retest two or more weeks after you recover.
Chronic inflammatory conditions such as lupus or rheumatoid arthritis are listed as risk enhancers in the 2026 guideline.
Best use: when your 10-year risk sits in a borderline zone and you are deciding whether to treat.
TSH has a normal range that people argue about 🧠
Thyroid-stimulating hormone comes from your pituitary gland. When thyroid hormone dips, TSH rises to push the thyroid harder, so TSH usually moves before free T4 does. A raised TSH with normal free T4 is called subclinical hypothyroidism, found in roughly 5 to 10 percent of people.
Where “normal” ends is contested. Hamilton and colleagues studied 1,861 people from the Hanford Thyroid Disease Study and screened out anyone with signs of thyroid disease, including antibodies and ultrasound findings. In the strictest group of 766, the upper reference limit came out near 4.0 to 4.1 mU/L. The European Thyroid Association guideline uses 0.4 to 4.0 and splits mild elevation (4.0 to 10) from marked elevation (above 10). Reference ranges also drift upward with age, so a TSH of 6 in an 85-year-old may be unremarkable.
What the guidelines suggest after a high result:
Repeat TSH with free T4 and TPO antibodies after two to three months, since single values wobble.
Treat at TSH of 10 or above on two occasions three months apart, per NICE guidance.
Trial levothyroxine for six months only if you are under 65, have symptoms, and sit below 10.
Stop if symptoms do not improve once TSH is back in range.
Be careful with expectations here. Evidence that treating mild elevation improves symptoms or cardiovascular risk is weak, and fatigue has a dozen other causes. The test is worth ordering if you have symptoms or a family history. It is not worth chasing a TSH of 3.5 because a wellness account called it “suboptimal.”
Insurance coverage for ApoB, Lp(a), and fasting insulin varies a lot, so ask about cost before the draw. Which of these five has your doctor never mentioned, and what answer would you want from the result?


