Your Blood Sugar Keeps Time: The Clock Genes Behind the Midnight Snack

It is eleven at night, and there is ramen. You already know how this goes: the same bowl that would have been lunch is, at this hour, somehow a different event. Every household has a rule about it — nothing after eight, nothing after nine — and everyone has wondered whether the rule is biology or just somebody's grandmother.

It turns out to be at least partly biology, and this year someone found a clean way to show it. Not by feeding people at midnight, but by looking at the clock on the wall when their blood was drawn.

Half a million blood samples, each with a timestamp

When a biobank draws your blood, it writes down the time. Sinnott-Armstrong et al. 2026 used that: in about 420,000 UK Biobank participants, they asked whether the effect of a genetic variant on blood sugar changes with the hour the sample was taken. Then they checked the answer in four more biobanks — about 100,000 people in Estonia, 500,000 in Finland, 160,000 US veterans and 52,000 more in Boston.

It does change. Glucose levels, the study found, are under diurnal genetic control — control that runs on a daily cycle. Circadian rhythms were already known to regulate glucose metabolism as well as sleep; what had never been mapped was which variants do the timing.

The melatonin receptor, of all things

Two positions stood out. One is in MTNR1B, the gene for a receptor that melatonin — the hormone your brain releases as it gets dark — binds to. The other is in CRY2, one of the handful of genes that make up the core circadian clock itself. In other words, the two strongest time-of-day signals for blood sugar sit in the machinery that tells your body what time it is.

The study went further in three directions. Sleep itself changed glucose levels. The variants acted independently of sleep, so this is not just "people who sleep badly have worse blood sugar". And the same variants independently changed the risk of type 2 diabetes. A separate study this year, of gestational diabetes, found that MTNR1B's effect is amplified during pregnancy — the same gene, turning up louder in a different setting.

What it does not say

The study also reports that use of sleep medications, including melatonin, was associated with type 2 diabetes. That sentence is easy to turn into a headline, so here is what it is: an association in population data. People who take something to sleep differ from people who do not in many ways, and an association cannot say which way the arrow points, or whether there is an arrow at all. The study does not say melatonin causes diabetes, and neither does this page.

What it does say is smaller and more interesting: your body handles sugar on a schedule, part of that schedule is written in your genes, and the genes are the ones that keep time.

Where this touches this site

The two headline positions are not in this catalogue yet. 21 other positions from the study are, and they are a tour of the blood sugar genes this site already holds: rs560887 in G6PC2, rs34872471 in TCF7L2, rs3802177 in SLC30A8 and rs11708067 in ADCY5. The fasting blood glucose page and the blood sugar and sleep topics collect the rest.

So the grandmother's rule has something behind it after all — not a number on a clock, but a clock in you. What time it says is, a little, inherited.