What GIP Does on Its Own

Tirzepatide acts on two receptors, and its advantage over drugs acting on one is usually credited to the second: GIP. Researchers infused native GIP for six weeks, alone and on top of semaglutide, raising blood levels roughly thirtyfold. Blood sugar did not improve.^[1] That is the second study to find GIP doing very little for glucose in people, and it leaves an open question about why tirzepatide works.

What was done

Sixty-one people with type 2 diabetes were randomized into four groups — placebo alone, GIP alone, semaglutide alone, and semaglutide plus GIP — for six weeks of continuous subcutaneous infusion, double-blind.

The exposure achieved was not marginal. Fasting bioactive GIP rose from around 7 pmol/L on placebo to 45 on GIP alone and 85 alongside semaglutide; total GIP went from 14 to 375 and 527 respectively. Whatever this study failed to find, it did not fail for want of drug.

The trial had set a target of a 1.50 mmol/L improvement, and neither comparison came close. The second row is as flat a result as this kind of analysis produces — a P value of 1.00 and an interval sitting symmetrically around zero.

Thirty times the hormone, and the glucose line did not move.

The second study to say this

We covered a phase 1 trial of a long-acting GIP receptor agonist, which produced dose-dependent weight loss — and whose effect on fasting glucose was transient and no longer significantly different from placebo by day 29. That is in the drugs that are not here yet.

That is worth stating clearly because GIP is an incretin — a hormone named for its ability to stimulate insulin release after eating. Its glucose-lowering role is the reason anyone was interested in it. In people with type 2 diabetes, it appears not to deliver that.

So what is tirzepatide's GIP component doing?

This is the question the result raises and does not answer. Tirzepatide beats semaglutide substantially on weight — 20.2% against 13.7% in the head-to-head, covered in the head-to-head — and the dual mechanism is the usual explanation. If GIP does nothing, the explanation needs revisiting.

• It may act on weight rather than glucose. The long-acting agonist study found exactly that pattern: weight loss without a durable glycemic effect. This trial measured glucose.
• A long-acting agonist is not the native hormone. Engineered molecules can differ from natural ligands in how strongly they bind, how long they stay, and how the receptor responds — differences that change the biology.
• A continuous infusion is not a physiological signal. GIP is released in pulses after meals. Flooding the system steadily may not reproduce what pulses do.
• The receptor may be desensitized in type 2 diabetes, which is a long-standing hypothesis and would explain a null in exactly this population.

None of those is established, and the honest position is that the GIP contribution to tirzepatide is less well understood than the confident explanations suggest.

What limits the trial

• Sixty-one people at a single center over six weeks, with 16% discontinuing — and the authors state that the dropouts constrain what they can conclude.
• Every participant self-reported as White, which is the systematic pattern set out in who was actually in the trials.
• It measured glucose, not weight. A null on one endpoint is not a null on the other.
• Median diabetes duration was 6.3 years with a median HbA1c of 54 mmol/mol — reasonably controlled disease, where there is less room to improve.

Injection site reactions were the commonest adverse event at 36%, which is what a six-week continuous infusion produces and not a property of the hormone.