Will GLP-1s Work for You? Predicting Response

Average weight-loss figures for GLP-1 drugs hide a wide spread. In the pivotal STEP 1 trial of semaglutide 2.4 mg (NEJM, 2021), the mean change in body weight at 68 weeks was −14.9%, versus −2.4% on placebo. But “mean” is doing a lot of work: that headline is the center of a distribution, and individual results scatter around it. If you are considering one of these drugs, the honest version of “will it work?” is really “where might I land in that distribution?” — and the science can narrow the guess only partway.

What variation looks like

STEP 1 randomized 1,961 adults with obesity (without diabetes) in a double-blind design. While the mean loss was large, the trial reported that 86.4% of the semaglutide group lost at least 5% of body weight — meaning roughly one in seven did not even clear that modest threshold. A minority of people respond poorly to an effective drug at an adequate dose. This is normal biology for any medication, but it gets flattened in marketing.

The drug and dose also matter. SURMOUNT-5 (NEJM, 2025), the first head-to-head trial, randomized 751 adults with obesity to maximum tolerated doses of the dual GIP/GLP-1 agonist tirzepatide or to semaglutide. At week 72, mean weight reduction was greater with tirzepatide.

Trial	Drug	Mean weight loss	Source
STEP 1 (n=1,961)	Semaglutide 2.4 mg	−14.9% at 68 wk	NEJM 2021
SURMOUNT-5 (n=751)	Tirzepatide	−20.2% at 72 wk	NEJM 2025
SURMOUNT-5 (n=751)	Semaglutide	−13.7% at 72 wk	NEJM 2025

On average, tirzepatide outperformed semaglutide head-to-head — but a group average does not tell any single person where they will land. Individual response still varies inside each arm.

Can early response predict the rest?

This is the most useful, and most misunderstood, question. A post-hoc analysis of the STEP 4 trial (Journal of the Endocrine Society, 2021) tested whether weight loss by week 20 forecasts the eventual result. The finding cuts against the intuitive story: early non-response was a poor predictor of ultimate failure. Many participants who had not lost much by week 20 still went on to reach clinically meaningful (≥5%) weight loss by week 68 if they stayed on treatment. Early response had reasonable positive predictive value but weak negative predictive value.

The practical implication is the opposite of what you might expect: a slow start is not a reliable reason to quit.

What doesn’t predict well

It is worth saying plainly what we cannot do yet. There is no validated, widely available genetic test or biomarker that tells an individual in advance whether they will be a strong or weak responder. Research into predictors is active, but it has not produced a clinical tool you should be paying for today. The most consistent levers remain unglamorous: staying on an effective dose, since side effects that force dose reduction or discontinuation blunt results, and the surrounding behaviors — diet quality, sleep, and activity — that do not disappear because a drug is involved.

The takeaway

GLP-1s produce substantial average weight loss, and on the current evidence tirzepatide produces more than semaglutide head-to-head. But averages are not destiny: a meaningful minority respond poorly, and we cannot reliably identify them before treatment — nor, per the STEP 4 analysis, can we confidently write someone off after a slow first few months. The honest approach is empirical and supervised: give the drug a fair trial at an appropriate dose, and decide with a clinician based on your actual trajectory rather than an average from a press release.

Sources

• Once-Weekly Semaglutide in Adults with Overweight or Obesity (STEP 1, NEJM 2021)
• STEP 1 Trial Summary — American College of Cardiology
• SURMOUNT-5: Tirzepatide vs Semaglutide — American College of Cardiology
• Clinically-Relevant Weight Loss Achieved Independently of Early Weight Loss Response (STEP 4 post-hoc, J Endocr Soc 2021)