Somewhere between one in twenty and one in ten people carry a gene variant that makes a liver marker come back flagged. It is not a liver problem. On the current evidence it looks like an advantage — and it is the clearest argument I know against the idea that your genetics are a sentence.
Somebody brings me a blood panel with one thing highlighted in red. Bilirubin, slightly above range. Everything else normal — liver enzymes fine, blood count fine, no jaundice they have ever noticed.
They have usually been told it is nothing to worry about, which is true, and left with a vague unease about their liver, which is not useful. Occasionally they have been told it is a sign their detoxification pathways are struggling, which is worse than not useful.
What they most likely have is Gilbert's syndrome, and the interesting thing about it is not that it is harmless. It is that people who have it appear to do better.
Bilirubin comes from the breakdown of old red blood cells. Before it can be excreted, the liver has to conjugate it — attach a glucuronic acid group — using an enzyme called UGT1A1.
In Gilbert's syndrome, a variation in the promoter region of the UGT1A1 gene reduces how much of that enzyme gets made. Conjugation runs at roughly 30% of normal capacity, unconjugated bilirubin backs up slightly, and the number on the report goes mildly above range. Liver enzymes stay normal because there is no liver damage. There is just a slower tap.
It is common. Estimates range from 2% to 20% depending on ancestry, with 5–10% a reasonable figure for European populations, and 30–45% carrying the polymorphism without necessarily showing the raised bilirubin. It becomes visible at puberty because bilirubin production rises then, which is why so many people first see it in their twenties on a routine test.
Gilbert's is normally described as a single variant — the UGT1A1*28 promoter repeat. Ehmer and colleagues showed in Hepatology that it is more complicated than that: 76% of people with Gilbert's were homozygous for a variant haplotype spanning four UGT1A genes, against 9% of blood donors.
That matters, and not only academically. It means the whole glucuronidation family is affected, not just the bilirubin enzyme — and glucuronidation is how the liver clears a great many other things besides bilirubin. Hold that thought; it comes back later and it is the part with real clinical consequences.
For most of the twentieth century bilirubin was understood as rubbish — a by-product of haem breakdown, neurotoxic at high concentrations, of no positive use.
That changed in the late 1980s when bilirubin was shown to be a potent endogenous antioxidant. And unlike most of the antioxidant story, which has been oversold for decades, this one has held up and deepened. Bilirubin is now understood as a signalling molecule affecting multiple intracellular pathways — one recent review in the Journal of Hepatology went as far as calling it "the yellow hormone."
Which raises an obvious question. If bilirubin is protective, what happens to people who run slightly high their whole lives?
Vítek 2002 (Atherosclerosis): ischaemic heart disease was present in 2% of Gilbert's subjects over 40, against 12.1% of the comparable general population.
Maruhashi 2012 (Circulation): 108 young men with Gilbert's versus 108 matched controls. Flow-mediated vasodilation — a direct measure of endothelial function — was 7.2% against 5.9%. Oxidative stress markers were lower, and both correlated with bilirubin concentration.
Wallner 2013: Gilbert's subjects had lower total cholesterol, lower LDL, lower triglycerides, lower ApoB, a better ApoB:ApoA1 ratio and lower IL-6. The advantage was most marked in the older subjects.
Respiratory. Horsfall and colleagues looked at the 1946 British birth cohort — 2,190 people followed for decades. Those with the Gilbert's genotype had better lung function, and the effect was strongest in heavy smokers: mean FEV1 was 409 ml higher at age 53. Odds of respiratory disease were roughly halved.
Athletic performance. A Czech study of 536 elite athletes found Gilbert's syndrome in 22%, against 9.6% of the general population. An Italian study of 1,492 Olympic athletes found 6% — a very different figure, and I would rather show you both than pick the flattering one. Endurance athletes in the Italian cohort did have higher bilirubin, plausibly as an adaptation to oxidative load.
Everything above is observational, which means it cannot establish cause. But there is one study that gets closer.
Atazanavir, an HIV protease inhibitor, raises bilirubin as a side effect — it inhibits the same enzyme — producing something that closely mimics Gilbert's. Dekker and colleagues gave it to sixteen people with type 2 diabetes in a double-blind crossover design. Bilirubin rose from 7 to 64 µmol/L, and endothelial function measurably improved, alongside antioxidant capacity.
Small, short, and in a specific population. But it is the difference between "people with high bilirubin are healthier" and "raising bilirubin does something."
The antioxidant story is not the whole story. Bulmer and colleagues estimated in a meta-analysis that bilirubin accounts for only about 34% of the cardioprotective effect observed in these populations. The rest appears to come from the lipid and inflammatory differences — which may themselves be downstream of bilirubin, or may not be. This is not as clean as "bilirubin is an antioxidant, therefore protection."
Almost all of it is observational. People with Gilbert's differ from people without it in ways beyond bilirubin, including a tendency to lower body mass. Association is not causation, and the interventional evidence amounts to one small crossover study.
And "diseases of civilisation" is doing a lot of work. You will see Gilbert's described as protective against cardiovascular disease, diabetes, several cancers, autoimmune and neurodegenerative disease. The cardiovascular and respiratory evidence is genuinely reasonable. The rest thins out considerably, and the further down that list you go the more you are reading hypothesis rather than finding.
Here is where the haplotype point comes back, and it is the most clinically actionable thing in this article.
Glucuronidation does not only clear bilirubin. It is a major route by which the liver clears drugs. If UGT1A1 activity is running at a third of normal — and if, as Ehmer's work suggests, neighbouring glucuronidation genes are affected too — then some medications will clear more slowly than the prescriber expects.
Irinotecan, a chemotherapy agent, is the clearest example. Gilbert's genotype substantially increases the risk of severe toxicity, and this is recognised on the drug labelling. Protease inhibitors used in HIV treatment are the other well-documented group. Paracetamol is also glucuronidated, though at normal doses this is not generally considered a problem.
Ehmer's conclusion was blunt: Gilbert's syndrome should be considered a potential risk factor for drug toxicity. So if you have it, it is worth mentioning to any doctor prescribing something new — not as a worry, but as information they may not have. That is a more useful thing to do with the knowledge than feeling pleased about your cardiovascular risk.
Gilbert's has been drawn into the chronic Lyme conversation through something described as the “Gitlin Triad” — the combination of Hashimoto's thyroiditis, a low white cell count, and raised bilirubin, presented in some circles as a near-diagnostic signature of Lyme disease, with predictive figures quoted in the nineties of a percent.
I went looking for the source. What exists is a description that Dr Gitlin has observed this pattern and thinks it may be a hallmark of Lyme disease. That is a clinician's impression, reported second-hand. There is no published validation, no cohort, no sensitivity or specificity, and certainly nothing supporting a predictive value of 92–100%. Those numbers appear to have attached themselves to the observation somewhere in transmission.
Which is a shame, because the underlying observation might be worth studying. Three findings clustering more often than chance would be interesting. But an untested clinical impression is not a diagnostic triad, and given that roughly one in ten people has Gilbert's anyway, any triad containing it will co-occur with a great many things.
I would not use it, and I would be cautious about anyone who does.
Most people reading this will not have Gilbert's. The more common finding is the opposite one, and it gets no attention at all because it never gets flagged.
Bilirubin at the very bottom of the reference range. Laboratories typically report something like 3–21 µmol/L, and a result of 4 sits comfortably inside it. Nobody comments. But if the epidemiology above is pointing at anything real, the interesting position is not "inside the range" — it is where inside the range, because the association with cardiovascular risk runs across the whole span and does not stop at the cut-off.
The functional range we work to is 5–14 µmol/L, derived from the epidemiology rather than from a laboratory's distribution — the UK Biobank analysis of 467,519 participants being the largest contributor. A bilirubin of 4 or 5 is not a disease. It is a position worth noticing, particularly alongside other markers of oxidative demand.
What I would not do is chase it. There is no established intervention to raise bilirubin, and the things that transiently raise it — fasting, dehydration, illness, intense exercise — are not a strategy. And I would be actively wary of the opposite instinct, which crops up in functional practice: treating a Gilbert's-range bilirubin as a detoxification problem to be corrected downwards. On the evidence above, that would be precisely backwards.
Almost everyone who brings me a genetic result brings it as bad news. A variant found, a risk identified, a sense of a sentence handed down. MTHFR, APOE, COMT — the conversation is nearly always about what is wrong.
Gilbert's is the same category of finding and it points the other way. A gene variant, present in up to one in ten people, that reduces the function of an enzyme — and appears on balance to be an advantage. It is not the only one. Variants that were once described as defects have a habit of turning out to be trade-offs, and which side of the trade you are on depends on the environment you are living in.
A gene variant is not a verdict. It is a parameter — and its meaning depends entirely on the context it is operating in, which is the part you can actually influence.
The same logic applies in the other direction, and it is the more useful half. If a variant only expresses itself under particular conditions — low folate, high oxidative load, a specific drug — then the variant is fixed and the conditions are not.
That is not optimism. It is just where the leverage is.
Educational content, not medical advice. A raised bilirubin should be interpreted by a doctor in the first instance — Gilbert's syndrome is a diagnosis of exclusion, and other causes of unconjugated hyperbilirubinaemia, including haemolysis and liver disease, need ruling out before it can be assumed. If you have Gilbert's, mention it when any new medication is prescribed.
A flagged result and a normal result can both be misread. Position within the range is often more informative than which side of the line it falls.
Ask AIdan → See testing options