Deep divefrom Episode 5Kids & family healthHeart & blood7 October 202616 min

Adult Metabolic Disease, Already Visible in Kids' Blood?

A deep dive into one paper: what was known before, what the researchers did, what the numbers say and what they don't.

The paper: Landman, Highland, Perry, Howard et al. — Paediatric proteomic signatures of cardiometabolic disease-associated traits predict adult disease outcomes (Nature Metabolism)

MP3

Chapters

Transcript

The script the two AI voices read from. The times are worked out from the script, not heard, so a line can start a few seconds early or late.

Read the whole episode

Cold open

Host 1You heard the trailer — this is the whole story. Welcome to the Hastrup Review of Science — where artificial intelligence reads real science, so you don't have to pretend you understood the paper. One paper today: can a blood sample from a thirteen-year-old already carry the molecular fingerprint of the heart, liver and metabolic disease that adults get decades later?

Host 2The paper is by Joshua Landman, Heather Highland, Andrew Perry, Annie Green Howard and colleagues, published in Nature Metabolism on the eleventh of September twenty twenty-six. One honest note up front: we could not retrieve the full text for this episode, so everything we say about the paper itself comes from its abstract. Anything beyond that is established background from the field, and I will flag it as such.

Host 1So fewer numbers than usual.

Host 2Fewer numbers, and some methods questions we will have to leave open rather than guess at. The headline from the abstract is that protein patterns linked to cardiometabolic traits in children look very much like the protein patterns linked to cardiometabolic disease in adults.

Why this field exists

Host 1Start with the name. The abstract talks about C K M D. What is that?

Host 2It stands for cardiovascular-kidney-metabolic disease. As background, the American Heart Association formalised the idea of a cardiovascular-kidney-metabolic syndrome in a presidential advisory in twenty twenty-three. The point was that obesity, type two diabetes, fatty liver, chronic kidney disease and heart disease are not separate problems that happen to cluster, but one connected system that fails together.

Host 1Why lump them together?

Host 2Because they share drivers: excess and dysfunctional fat tissue, insulin resistance, chronic low-grade inflammation and disordered cholesterol handling. Once one organ starts to struggle, it tends to push the others. Treating them as one syndrome lets clinicians stage risk across organs instead of waiting for each one to fail on its own.

Host 1And why children? These are diseases of middle age.

Host 2The clinical events are, but the biology starts much earlier. As background, autopsy studies such as the Bogalusa Heart Study and the P D A Y study, which examined young people who died of accidents and other causes, found fatty streaks and early plaques in the arteries of adolescents and young adults. The abstract makes the same point: the earliest pathological signs appear before age twenty.

Host 1And the obesity trend makes that more urgent.

Host 2Yes. As background, in the United States roughly one in five children and adolescents now meets the definition of obesity, and the rate is higher among Hispanic and Latino children. Fatty liver disease in children, now called metabolic dysfunction-associated steatotic liver disease, is also more common in Hispanic youth, partly because a liver-fat risk variant in a gene called P N P L A three is more frequent in that population.

What was known before

Host 1So what did doctors already have for finding at-risk kids?

Host 2Mostly the adult toolkit, scaled down. Body mass index percentiles, blood pressure, fasting glucose, cholesterol, and liver enzymes such as A L T. The abstract's argument is that adult-based risk thresholds miss a substantial fraction of children who are in fact at high risk.

Host 1Why would adult thresholds fail in children?

Host 2Partly because children are not small adults. Puberty transiently increases insulin resistance, body composition changes fast, and the normal ranges for many lab tests shift with age and sex. A well-known background example is liver enzymes: work from the SAFETY study suggested healthy children have A L T values well below the upper limits many labs use for adults, so a child can have fatty liver while their test reads as normal.

Host 1Where does proteomics come in?

Host 2Proteomics means measuring thousands of proteins in a blood sample at once. As background, two platforms dominate large studies: one uses aptamers, short strands of D N A that bind specific proteins, and the other uses pairs of antibodies whose attached D N A tags are read out by sequencing. In adults, this has already produced protein-based risk scores for heart disease, diabetes and many other conditions, including large analyses in U K Biobank.

Host 1So the adult side is fairly mature.

Host 2Comparatively, yes. The paediatric side is much thinner, largely because big proteomics studies have been run on adult cohorts with decades of follow-up. That is the gap this paper targets.

The open question

Host 1Put the open question in one sentence.

Host 2When a child has early signs of liver fat, excess or poorly functioning fat tissue, stiffer blood vessels or creeping blood sugar, is the protein signature in their blood the same one that marks real disease in adults? If it is, the adult disease process is already running in childhood, and blood proteins could flag risk earlier and more sensitively than adult cut-offs.

Host 1And if it is not?

Host 2Then childhood traits might reflect growth or puberty rather than an early version of adult disease, and borrowing adult markers would be misleading. Both outcomes would be informative.

Methods: the children

Host 1Who did they study?

Host 2Two hundred and seventy-three Hispanic or Latino children and adolescents. Their average age was thirteen point one years, give or take two point seven, and fifty-three percent were female. The abstract does not name the cohort, so I will not guess which one it is.

Host 1And what did they measure in them?

Host 2Twenty-five cardiovascular-kidney-metabolic phenotypes, in four families: liver traits, fat-tissue traits, vascular traits and dysglycaemia, meaning abnormal blood sugar control. Each of those traits was then linked to the circulating proteome, the full panel of proteins measured in the blood.

Host 1Do we know which proteomics platform, or how many proteins?

Host 2Not from the abstract. We also do not know exactly how liver fat or vascular health were measured, for example imaging versus blood tests. Those details matter for how far you trust each signature, and they are in the paper we could not read.

Host 1What does linking traits to the proteome actually involve?

Host 2In studies of this kind, as general background, you test each protein against each trait, usually adjusting for age, sex and often puberty and body size, and then correct for the enormous number of tests. With twenty-five traits and thousands of proteins, that is tens of thousands of comparisons, so false positives are the main hazard. What you end up with is, for each trait, a list of proteins that go up or down with it and the strength of each association.

Methods: the adult comparison

Host 1Then they went to adults.

Host 2Two adult groups. Six hundred and eighty-five adults from the same community as the children, and twenty-eight thousand two hundred and fifty-six adults from U K Biobank. The question was whether the same proteins line up with the same traits in the same direction across ages.

Host 1Why use both?

Host 2The local adults share the children's ancestry, environment and probably the same lab methods, so they are the cleanest comparison. U K Biobank is mostly white British, much older and very large, so if the patterns hold there as well, they are not just a quirk of one community. As background, U K Biobank also links participants to hospital records, which is how proteins can be tied to actual diagnoses rather than just risk factors.

Host 1And the drug part?

Host 2They asked whether proteins linked to childhood cardiometabolic traits change when adults take a G L P one receptor agonist, the drug class that includes semaglutide. As background, large proteomic analyses from semaglutide trials have been published, but the abstract does not say which data set this paper used.

Host 1And genetics?

Host 2They also tested whether those proteins are linked to genetic liability to cardiometabolic disease. The abstract does not say whether that was done with polygenic scores, with Mendelian randomisation, or both. The general idea is that genes are fixed at conception, so a genetic link argues the protein sits on a causal pathway rather than simply reflecting lifestyle.

Results: concordance

Host 1So what did they find?

Host 2The main finding is high concordance between the children's trait-protein relationships and the adults' ones. Proteins that rose with, say, liver traits in children tended to rise with the corresponding traits in adults, in both the local cohort and U K Biobank. The abstract states that childhood proteome signatures were strongly associated with clinical outcomes in adults.

Host 1Do we get the actual strength of that concordance?

Host 2Not in the abstract. We do not have the correlation coefficients, the hazard ratios for adult outcomes, or how much prediction improved over standard risk factors. So I would call the direction of the finding clear and its size unknown to us.

Results: the biology

Host 1Which biology came up?

Host 2Four themes: pancreatic beta-cell health and insulin sensitivity, liver homeostasis, inflammation, and cholesterol metabolism. Beta cells are the insulin-producing cells, so that theme fits the dysglycaemia traits. The other three are the classic adult cardiometabolic pathways.

Host 1So nothing exotic.

Host 2No, and that is reassuring rather than disappointing. If thirteen-year-olds showed completely novel pathways, you would suspect noise or puberty effects. Seeing the familiar adult pathways is what you would expect if this is the same disease process at an earlier stage.

A word from our sponsor

Host 1A brief interruption.

Host 1Are you tired of finding out about problems only when it is far too late? Then you are in luck. This part of the Hastrup Review of Science is brought to you by menstrual cups.

Host 2Medical-grade silicone, reusable for years, and they hold up to three times more than a regular tampon. Early detection is not usually a design concern.

Host 1Menstrual cups come in several sizes and firmness levels, and you can find them on Amazon in Germany at very reasonable prices. Personally, I like that the volume markings give you a clean, quantitative readout.

Host 2Which is more than we can say for this abstract.

Host 1Menstrual cups. Our referral link is in the show notes.

Results: drugs and genes

Host 1Back to the proteins. What about the G L P one result?

Host 2Many of the proteins linked to childhood cardiometabolic traits shifted when adults were treated with a G L P one receptor agonist. The abstract says many, not how many, and it does not say whether they moved in the healthy direction, though that is the implication.

Host 1And the genetic link?

Host 2The proteins were linked to genetic liability to cardiometabolic disease. Together with the drug result, the authors argue this liability is built in early, written into the proteome in childhood, and at least partly modifiable. Their phrase is precision prevention: act long before clinical disease appears.

Why it matters

Host 1So what is the principle here?

Host 2That cardiometabolic disease is not something that switches on in middle age. If the same protein patterns mark early traits in children and real outcomes in adults, the disease process is plausibly continuous across the lifespan, and the child's blood already reflects it. That is consistent with what autopsy studies hinted at decades ago, but now at the molecular level and in living children.

Host 1Who does this change things for?

Host 2In the near term, mainly researchers. It gives paediatric studies a set of candidate markers with adult validation behind them, and it gives trial designers a possible way to measure benefit in children without waiting forty years for heart attacks. In the longer term, if a test were built and validated, it could help paediatricians decide which children with similar weight and normal labs actually need closer follow-up.

Host 1And the population matters too.

Host 2It does. Hispanic and Latino children carry a high burden of obesity and fatty liver, but they are underrepresented in large biomarker studies, which have been dominated by adults of European ancestry. Starting from this population and checking outward against U K Biobank is the reverse of the usual direction.

Host 1What does it not mean?

Host 2It does not mean there is a blood test your child should get. It does not show that putting children on G L P one drugs would prevent adult disease, because the drug data came from adults. And it does not show that any of these proteins cause disease; genetic links strengthen that case without proving it.

Honest caveats

Host 1Now the honest caveats.

Host 2The biggest one is in the title. As far as the abstract tells us, the same children were not followed into adulthood. The prediction is inferred: childhood trait signatures resemble signatures that, in other adults, go with disease outcomes.

Host 1So predict is doing a lot of work.

Host 2It is doing an inference's work. That inference is reasonable, but it is not the same as measuring proteins at age thirteen and counting heart attacks at fifty. That second kind of study is the true test, and it takes decades or an old cohort with stored childhood blood.

Host 1What else?

Host 2Size. Two hundred and seventy-three children is respectable for paediatric proteomics but small for testing thousands of proteins against twenty-five traits, so the adult replication carries a lot of weight. Also, the child data are probably cross-sectional, one blood draw per child, so we cannot see how signatures change through puberty.

Host 1Generalisability?

Host 2One community of Hispanic or Latino children. That is a strength for representation, but the findings need to be checked in children of other backgrounds and other countries. The U K Biobank concordance helps for adults, but it says nothing about non-Hispanic children.

Host 1And the technical side.

Host 2If the child and U K Biobank samples were run on different proteomic platforms, concordance is harder to measure, because the two technologies do not always agree on the same protein. We cannot tell from the abstract whether that applies here. And the abstract reports no clinical performance figures, such as how many extra at-risk children a protein score would catch beyond standard tests, which is the number that would matter for any screening use.

Host 1And the drug angle?

Host 2That a protein moves under a drug in adults tells you it is responsive, not that moving it in children improves anything. G L P one drugs are now approved for adolescents with obesity in several countries, but whether earlier treatment changes adult outcomes is still untested.

What to watch next

Host 1What should we watch for next?

Host 2First, longitudinal tests: cohorts with stored childhood blood and adult outcome data, where proteins measured in youth can be checked against what actually happened. Second, replication in children of other ancestries.

Host 1And third?

Host 2A head-to-head comparison: does a small, cheap protein panel identify more at-risk children than B M I, A L T and glucose alone? If that answer turns out to be yes, and modestly so, it would be worth pursuing. If it is not, this remains good biology rather than a clinical tool.

Sign-off

Host 1That is the deep dive: Paediatric proteomic signatures of cardiometabolic disease-associated traits predict adult disease outcomes, by Landman, Highland, Perry, Howard and colleagues, in Nature Metabolism. The link is in the show notes.

Host 2Every Friday this same feed carries the roundup episode: five new papers as trailers, and you pick which become deep dives. This has been the Hastrup Review of Science.