Keeping weight off after GLP-1s: what two trials tested

In two placebo-controlled trials published together in Nature Medicine, adults who finished 72 weeks of injectable GLP-1 treatment kept more of their weight loss over the next year after switching to the once-daily oral pill orforglipron than after switching to placebo, and healthy adults who lost weight by dieting regained less with a daily Akkermansia probiotic than with placebo. Neither trial tested stopping all treatment.

What this means

Two randomized, placebo-controlled trials published together in Nature Medicine on May 13, 2026, take a first run at the “what next” after a big weight loss. People who had finished 72 weeks of injectable GLP-1 treatment kept more of their loss a year after switching to the oral pill orforglipron than those switched to placebo — 74.7% versus 49.2% after tirzepatide, 79.3% versus 37.6% after semaglutide — while in a separate trial, healthy adults who had dieted regained less over six months with a daily Akkermansia probiotic (1.2 kg, about 2.6 lb, versus 3.2 kg, about 7.1 lb). Neither trial tested stopping treatment: the pill continues drug therapy in a new form, and the probiotic trial enrolled people who lost weight by dieting. What remains untested is whether the pill matches staying on the injection, and whether either effect lasts beyond the trials’ horizons.

The open question after the injections end

When people stop injectable GLP-1 treatment, most of the lost weight tends to come back: a 2026 meta-analysis in the BMJ put average regain at about 0.4 kg (about 0.9 lb) a month, and about double that after semaglutide or tirzepatide. Our explainer on what happens when people stop weight-loss medications (What Happens When People Stop Weight-Loss Medications) covers that baseline, and it ends on an open problem: no strategy had been shown to prevent the regain. On May 13, 2026, Nature Medicine published two placebo-controlled trials that put “what next” strategies to a randomized test — the first randomized evidence for a switch strategy after injectable GLP-1 treatment, and the first randomized maintenance test of this probiotic strain after dieting. Both are narrower than the headlines suggest, and neither tested stopping treatment.

Trial one: switching from the injection to a pill

ATTAIN-MAINTAIN, a phase 3b trial run at 29 US sites, enrolled 376 adults with overweight or obesity, none with diabetes. All had just finished 72 weeks of injectable tirzepatide (205 people) or semaglutide (171) in the parent trial, SURMOUNT-5, each taken to its maximally tolerated dose, with at least 5% of body weight lost. They were then randomized 3:2 to orforglipron — a once-daily oral GLP-1 medication, started at 12 mg and raised toward 36 mg — or to matching placebo, for 52 weeks. The tirzepatide cohort began at a mean weight of 90.1 kg (about 198 lb), the semaglutide cohort at 94.4 kg (about 208 lb). The trial used an investigational capsule; the drug has since been approved in the USA as Foundayo.

The main measure was how much of the earlier loss was still gone at week 52, as a percentage of the weight lost. Among participants who had reached a weight plateau before the switch — the primary analysis group — the pill group kept 74.7% of the loss versus 49.2% on placebo after tirzepatide (difference 25.5 percentage points, 95% CI 14.5 to 36.5, P<0.001), and 79.3% versus 37.6% after semaglutide (difference 41.7 points, 95% CI 24.4 to 59.0, P<0.001). The same pattern held across all participants. These are model-based estimates. In the plateau group, 43.7% versus 16.4% (after tirzepatide) and 55.0% versus 6.9% (after semaglutide) kept at least 80% of their loss. Cardiometabolic markers were largely preserved — among pill users in the tirzepatide cohort, average HbA1c stood at 5.2% at entry and was still 5.2% at week 52.

The switch had costs. Nausea affected 23.4% of the pill group versus 5.0% of placebo in the tirzepatide cohort, and roughly one in ten in each cohort reported gastrointestinal symptoms in the first four weeks after the switch, mostly mild to moderate. In the tirzepatide cohort, 7.3% stopped the pill because of side effects versus 2.5% on placebo. Among pill users, adjudication confirmed a single case of pancreatitis (tirzepatide cohort) and a single major cardiovascular event (semaglutide cohort); in the semaglutide cohort’s pill arm, one participant died, the cause reported as cardiovascular/stroke, and the trial’s tables do not state whether the event and the death are the same person. Eli Lilly ran and funded the trial; 8 of its 14 authors are Lilly employees and shareholders.

Three limits shape the numbers. First, there was no continued-injection arm: the trial shows that switching beat placebo, and cannot say whether the pill matches staying on the injection. Second, the placebo group stopped being pure at week 24, when a rescue rule let participants who had regained at least half their lost weight cross to the pill — about 65% of those eligible did — so by week 52 only 31.3% (tirzepatide cohort) and 18.2% (semaglutide cohort) of the original placebo participants were still untreated, and late comparisons mix in active-drug experience. Third, the trial ran one year; nothing in it speaks to durability beyond that.

Trial two: a probiotic, tested after dieting

The second trial tested a different next step in different people. Ninety healthy adults aged 20 to 70 with a BMI of 28 to just under 40 — none with diabetes — first went through eight weeks of a low-energy diet (roughly 800 kcal a day), needing to lose at least 8% of their starting weight to continue. Those who did were randomized, in a quadruple-blind design with everyone from participants to outcome assessors unaware, to daily capsules of pasteurized Akkermansia muciniphila MucT or placebo for 24 weeks of unrestricted healthy eating. The trial enrolled people losing weight by diet, not people coming off GLP-1s. About 71 participants (34 to 37 per arm) appear in the reported analyses; only the abstract was available to us at press time — the full paper is paywalled — so the participant-flow details behind that figure could not be checked.

Over the 24 maintenance weeks, the probiotic group regained 1.2 kg (about 2.6 lb) on average versus 3.2 kg (about 7.1 lb) with placebo (P=0.012) — a between-group difference of 2.0 kg (about 4.4 lb). Measured from the pre-diet starting weight, the probiotic group finished the trial 3.1 kg (about 6.8 lb) ahead of the placebo group (P=0.009); that net figure is a separate comparison, not a restatement of the regain result. No serious adverse events related to the treatment were observed. Both results are reported as single P values, with no adjustment for the paper’s multiple analyses mentioned in the accessible abstract, so each reads as a single-test result.

The conflicts here are unusually dense. The Akkermansia Company makes the product and funded the study: one author (W. M. de Vos) co-founded the company and is its chief technology officer; another (P. Suenaert) is its chief medical officer; one author is a company employee; and senior author E. E. Blaak is a scientific advisor. Four authors are inventors on a patent application related to the study. The design leaves mechanism open — with no deactivated-bacteria comparator, whether an effect comes from the pasteurized bacteria, other capsule components, or something else is untested. The strain’s prior human evidence was mixed: a 2019 proof-of-concept improved insulin sensitivity without significantly moving weight.

What neither trial can say

Nothing in either trial shows that a pill or a probiotic prevents regain after stopping GLP-1s, and press coverage that framed the pair that way overreached. The orforglipron arm changed treatment from an injection to a pill; the people in it stayed on an active GLP-1 medication. The probiotic trial’s participants lost weight by dieting, so it cannot say whether the probiotic does anything for someone coming off injectables. Durability past one year for the pill, and past six months for the probiotic, is untested. And no trial has compared the oral switch with staying on the injectable: the nearest evidence is SURMOUNT-4, in which adults who kept taking tirzepatide after losing weight lost a further 5.5% over a year while those switched to placebo regained 14% of their body weight. The tested ground is narrow — continuing treatment by another route, and a sponsor’s probiotic in healthy dieters — and what happens when treatment stops completely remains the field’s open problem.

Terms explained

Percent of weight loss maintained — how much of an earlier weight loss is still gone at a later time, as a percentage of the weight lost; 74.7% means about three-quarters of the lost weight stayed off.
Rescue therapy — a built-in escape for participants faring poorly on their assigned treatment; here, placebo participants who regained at least half their lost weight could cross to the active pill from week 24.
Pasteurized probiotic — bacteria heat-treated so they are no longer alive; the trial tested heat-treated Akkermansia, so no living microbes are involved.
Model-based estimate — a result estimated through statistical adjustment rather than a plain average of everyone’s weights.

Sources

Primary source (trial A): Aronne LJ, Horn DB, le Roux CW, et al. “Orforglipron for maintenance of body weight reduction: the double-blind, randomized phase 3b ATTAIN-MAINTAIN trial.” Nature Medicine 32:2679-2687, published online May 13, 2026 (issue July 2026; open access CC-BY 4.0). https://doi.org/10.1038/s41591-026-04386-7 · PMID 42120723 · PMCID PMC13375559. Registry: ClinicalTrials.gov NCT06584916, https://clinicaltrials.gov/study/NCT06584916. Full-text access.
Primary source (trial B): Mount S, Canfora EE, Jocken JW, et al. “Pasteurized Akkermansia muciniphila MucT for weight loss maintenance in people with overweight and obesity: a controlled randomized trial.” Nature Medicine 32:2107-2116, published online May 13, 2026 (issue June 2026). https://doi.org/10.1038/s41591-026-04394-7 · PMID 42120725. Registry: ClinicalTrials.gov NCT05417360, https://clinicaltrials.gov/study/NCT05417360. Only the abstract was available to us at press time; full text paywalled.
Context — parent trial of trial A: SURMOUNT-5 (tirzepatide versus semaglutide), New England Journal of Medicine, published online May 11, 2025. https://doi.org/10.1056/NEJMoa2416394 (PMID: 40353578; NCT05822830). Abstract-level access; cited for trial A’s design and population only.
Context — regain baseline: 2026 BMJ meta-analysis of weight regain after stopping weight-loss treatment. https://doi.org/10.1136/bmj-2025-085304 (PMID: 41500720). Full access in Nutritious News’s prior published explainer packet; cited for the regain rates only.
Context — continuation data: SURMOUNT-4 (tirzepatide continuation after weight loss), JAMA, 2024. https://doi.org/10.1001/jama.2023.24945 (PMID: 38078870). Full access in the prior explainer packet; cited for the continue-versus-switch comparison only.
Context — strain history: pasteurized Akkermansia muciniphila human proof-of-concept (Depommier, Cani, de Vos), Nature Medicine, 2019. https://doi.org/10.1038/s41591-019-0495-2 (PMID: 31263284). Abstract-level access; cited for prior human evidence only.
Context — approval note: orforglipron (Foundayo) drug review, Shirley, Drugs, 2026 (PMID: 42479349). Cited for the approval note (first approval, USA, April 2026).
News signal (coverage framing, corrected against the primary sources): “Quitting weight-loss drugs — or a diet — can cause weight regain. Two strategies could help prevent that,” Scientific American, May 13, 2026. https://www.scientificamerican.com/article/quitting-weight-loss-drugs-or-a-diet-can-cause-weight-regain-two-strategies-could-help-prevent-that/ — cited for press framing only; every trial number in this draft follows the primary papers where the two differ.