Melatonin worsened blood sugar, but only in gene carriers

A tightly controlled Diabetes Care trial gave the same 21 adults both 5 mg of melatonin and a placebo. Glucose tolerance worsened in the 10 who carry a common diabetes risk variant, and not in the 11 who do not.

Credit: Placeholder image

Based on a randomized, double-blind, placebo-controlled crossover trial in 21 adults under a controlled 5-day laboratory protocol

Summary

Melatonin use has quintupled in recent decades, and a common gene variant raises diabetes risk.
In carriers of that variant, 5 mg of melatonin worsened glucose tolerance by about 12%.
It suppressed the fast early burst of insulin the pancreas releases, by around 40%.
In non-carriers, none of these effects appeared.
Only 21 people took part, so this explains a mechanism rather than settling advice.

Whether melatonin is bad for blood sugar has had an unsatisfying answer for a decade: sometimes, in some people, by an amount nobody could pin down.

A trial in Diabetes Care went at the question from the other end, and found that the honest answer depends on a gene most people have never been tested for.

Why a sleep aid was ever a metabolic question

Melatonin acts on two receptors, and one of them sits on the insulin-producing cells of the pancreas.

The gene for that receptor is called MTNR1B, and a common variant in it is among the most firmly established genetic risk factors for type 2 diabetes. As the researchers put it, the common melatonin receptor 1B gene (MTNR1B) variant increases risk of type 2 diabetes.

Set that beside how much melatonin people now take. Melatonin use quintupled in recent decades, mostly bought without a prescription and mostly by people who have never considered it a metabolic drug.

The question the trial asked was whether melatonin’s effect on blood sugar depends on which version of that gene you carry.

How the melatonin trial was run

Small, and unusually well controlled, which is the trade-off worth understanding here.

Twenty-one healthy participants of European ancestry took part, 10 carrying the risk variant and 11 not. Each completed a highly controlled 5-day laboratory protocol, during which they received 5 mg oral melatonin or placebo in randomized order on two nonconsecutive days.

Because every participant did both conditions, each person acts as their own comparison. That removes most of the variation between people, which is how a study of 21 can detect anything at all.

Blood sugar handling was measured by injecting glucose directly into a vein and tracking the response, rather than by a meal, which gives a much sharper picture of the pancreas responding.

What happened in the gene-variant carriers

The finding splits cleanly along genotype.

In carriers, melatonin worsened glucose tolerance and reduced early C-peptide responses in risk allele carriers, glucose tolerance worsening by roughly 12% and the early insulin signal dropping by about 19%. C-peptide is released alongside insulin and is used to measure how much the pancreas actually made.

Underneath that sat a bigger number. The impairment stemmed from a 40% suppression of glucose-stimulated first-phase beta-cell responsivity.

First-phase secretion is the fast burst of insulin released within minutes of sugar arriving, drawn from a supply the pancreas keeps ready. It is the body’s quickest metabolic reflex, and losing it is one of the earliest steps toward type 2 diabetes. Melatonin cut it by nearly half in carriers.

A second control also loosened. Carriers showed a slower insulin-induced decline in second-phase insulin secretion rate, meaning the normal braking of insulin output took longer to engage.

In the eleven non-carriers, none of this happened.

The melatonin result that looks like a benefit

One finding runs the opposite way and is worth sitting with, because it shows the same mechanism from behind.

In carriers, melatonin also prevented exogenous insulin-induced hypoglycemia compared with placebo: zero episodes against seven.

A blunted insulin response protects you from a low blood sugar that researchers deliberately induced with injected insulin. That is not a health benefit. It is the same impairment, observed in a situation where impairment happens to help, and it is good evidence that the effect is real rather than a statistical artifact.

What 21 people can establish about melatonin

The number is the obvious limit and it is a serious one. Twenty-one healthy volunteers, all of European ancestry, one dose, two days.

What the design buys in exchange is depth. A crossover design, meaning every participant took both melatonin and placebo on separate days, run in a metabolic ward with glucose delivered intravenously and each person compared against themselves, measures the pancreas far more precisely than a population study can. This is a mechanism study, and it is a good one.

What it is not is guidance. Nobody routinely knows their MTNR1B genotype, no clinical threshold exists, and a single 5 mg dose in a laboratory is not nightly use at home. The authors frame their contribution as identifying the mechanisms linking melatonin signaling to diabetes risk, which is exactly the right size of claim.

For anyone with diabetes or prediabetes who takes melatonin regularly, this is worth raising with a doctor. For everyone else it is a reminder that a hormone sold as a sleep aid is still a hormone.