GLP-1 Non-Responders: What to Do at 16 Weeks With <5% Loss

Roughly one in seven to one in five GLP-1 patients fail to clear the FDA “clinically meaningful” 5% body weight threshold by week 16 even at maximum tolerated dose. In STEP-1 (Wilding 2021^[1]) about 13% of semaglutide 2.4 mg patients lost less than 5% by week 68; SURMOUNT-1 (Jastreboff 2022^[2]) reported a similar ~9–10% fraction on tirzepatide 15 mg. Real-world clinics see substantially more, because real-world dose titration is often incomplete and adherence drops off (Gasoyan 2026^[10]). This article is the practical playbook for the week-16 visit: how to distinguish a true non-responder from a slow titrator, what the phenotype and genetic predictors actually say, and the sequential-trial protocol — sema → tirz → retatrutide or bariatric surgery — supported by SURMOUNT-5 (Aronne 2025^[3]), Acosta 2021^[5], and the STAMPEDE 5-year data (Schauer 2017^[9]).

The honest summary

• The 5% / 16-week threshold is the FDA-anchored definition. Anti-obesity-medication labels were built around the “continue if ≥5% by week 16” stopping rule; Fujioka 2016^[7] validated early response as the dominant predictor of 1-year outcome on liraglutide and the pattern holds for semaglutide and tirzepatide.
• True non-response is 10–15% in trials, higher in clinic. STEP-1^[1] ~13% under-5%; SURMOUNT-1^[2] ~9–10% under-5% on tirz 15 mg. T2D patients respond less (STEP-2^[4]): the insulin-resistance backpressure attenuates GLP-1 effect by roughly a third.
• SURMOUNT-5 is the switching evidence. Aronne 2025 NEJM^[3] randomized 751 adults with obesity head-to-head: tirzepatide produced −20.2% body weight vs −13.7% on semaglutide 2.4 mg at week 72 — a 6.5-percentage-point absolute advantage that justifies sema → tirz switching when sema response is inadequate.
• Phenotype-matched prescribing works. Acosta 2021^[5] randomized 312 patients to phenotype-guided versus usual-care AOM selection; the phenotype arm lost −15.9% at 12 months vs −9.0% in usual care — a 1.75× relative gain. The matrix is hungry-brain (phentermine/topiramate), hungry-gut (liraglutide/sema), emotional hunger (naltrexone/bupropion), slow burn (phentermine/topiramate).
• Bariatric surgery remains the high-magnitude option. STAMPEDE 5-year (Schauer 2017^[9]) documented sleeve gastrectomy producing ~−23% body weight at 5 years with durable T2D remission — the appropriate next step for BMI ≥ 40 non-responders who have failed sequential pharmacotherapy.

What counts as a non-responder

The FDA stopping rule for anti-obesity medications — first codified for liraglutide 3.0 mg (Saxenda) and later applied to semaglutide and tirzepatide — is “discontinue if less than 5% body weight loss at week 16 on the maximum tolerated dose.” The rule comes from Fujioka 2016 (Obesity Silver Spring^[7]), which showed that liraglutide early responders (≥5% by week 16) achieved a mean −11.0% at year 1, while early non-responders averaged only −4.0% — the early signal had a positive predictive value above 85%.

Before labeling a patient a non-responder, three conditions must be true. First, the patient must have reached the maximum-tolerated dose — not the starting dose, not the week-4 step, but the highest dose they can sustain without intolerable side effects. Many “non-responders” in clinic are actually slow titrators stuck at 1.0 mg semaglutide or 5 mg tirzepatide because of nausea. Second, the patient must have been on max dose for at least 8 weeks at the assessment point. Third, adherence has to be at or above 80% — one missed weekly injection a month is acceptable; missing every other week is not.

For semaglutide 2.4 mg the max-dose window starts around week 16 (the standard 4-step titration ladder); for tirzepatide the max-tolerated dose is usually identified between week 20 and 24 (5 steps to 15 mg). The realistic non-responder assessment is therefore at week 24–28 for both agents, not week 16, even though the label language anchors on week 16.

Trial vs real-world non-responder rates

STEP-1 (Wilding 2021 NEJM^[1]) randomized 1,961 adults to semaglutide 2.4 mg or placebo over 68 weeks. Mean body weight loss was −14.9% on semaglutide vs −2.4% on placebo. The cumulative-distribution curve showed about 86% of semaglutide patients reached ≥5% weight loss, and roughly 13% did not — the canonical trial-arm non-responder figure for sema.

SURMOUNT-1 (Jastreboff 2022 NEJM^[2]) randomized 2,539 adults to tirzepatide 5, 10, or 15 mg or placebo over 72 weeks. Mean loss was −15.0% / −19.5% / −20.9% on the active arms vs −3.1% on placebo. The ≥5% responder fraction was 85% / 89% / 91%, meaning the non-responder fraction was ~9–15% across doses — lower at higher doses.

STEP-2 (Davies 2021 Lancet^[4]) randomized 1,210 adults with obesity and type 2 diabetes to semaglutide 2.4 mg, 1.0 mg, or placebo over 68 weeks. Mean weight loss on the 2.4 mg arm was −9.6%, materially lower than the −14.9% seen in STEP-1 non-diabetic patients. The non-responder fraction in T2D was correspondingly higher — about 22%. The mechanistic explanation is insulin-resistance backpressure on the hypothalamic GLP-1 signal: T2D weight loss responses run roughly two-thirds of non-T2D responses on the same dose.

Real-world (Gasoyan 2026 Diabetes Obesity & Metabolism^[10]) tracked weight outcomes in a Cleveland Clinic cohort after semaglutide or tirzepatide discontinuation. The bigger real-world finding is that only about a third of patients reach the trial-replicated weight loss because real-world max-dose-reach rates are lower than the protocol-mandated titration in the registration trials. Many real-world “non-responders” are actually titration-failures masquerading as non-responders.

Phenotype and genetic predictors of GLP-1 response

The single best-validated predictor of obesity-pharmacotherapy response is phenotype-matching. Acosta 2021 (Obesity Silver Spring)^[5] conducted a pragmatic trial at the Mayo Clinic of 312 adults with obesity, randomized to phenotype-guided AOM selection (using a four-axis classification: hungry brain, hungry gut, emotional hunger, slow burn) versus usual care. At 12 months, the phenotype-guided arm had lost −15.9% body weight vs −9.0% in usual care — a 1.75× gain that held across BMI strata. The Anazco 2025 Int J Obes review^[6] updated the framework and validated it across registration-trial datasets, including STEP-1 and SURMOUNT-1.

The four-phenotype matrix:

• Hungry gut — abnormal satiation; characterized by large meal volume and slow nutrient-induced satiety. GLP-1 agonists are the matched class. This is the plurality phenotype (~30% of obese adults), and the one where sema and tirz produce the published trial outcomes.
• Hungry brain — abnormal satiety; early hunger return between meals. Phentermine/topiramate (Qsymia) is the matched class.
• Emotional hunger — hedonic eating in response to stress or affect dysregulation; high reward-eating score. Naltrexone/bupropion (Contrave) is the matched class; for binge-eating-disorder-overlap patients lisdexamfetamine (Vyvanse) is FDA-approved for BED.
• Slow burn — abnormal energy expenditure (lower-than-predicted resting metabolic rate). Phentermine and stimulant-class agents are the matched class.

Genetic predictors are real but small-effect. Twin and family studies estimate the heritability of weight-loss response to anti-obesity medication at 20–40%; the Anazco 2025 precision-medicine review^[6] summarizes the candidate-gene literature. MC4R loss-of-function variants attenuate response to most agents but respond well to the MC4R agonist setmelanotide (Imcivree, FDA-approved for select rare-disease subgroups). POMC variants behave similarly. Standard clinical genotyping is not yet justified outside research settings.

Behavioral and physiological co-factors that consistently attenuate response: sleep duration < 6 hours per night, untreated obstructive sleep apnea, chronic stress with elevated cortisol, polypharmacy with appetite-stimulating agents (mirtazapine, olanzapine, gabapentin at high dose, insulin), and untreated binge-eating disorder. The practical clinic question at the week-16 visit is whether any of these are modifiable before declaring true non-response.

Magnitude: % body weight loss across the non-responder pathway

SURMOUNT-5: the head-to-head switching evidence

SURMOUNT-5 (Aronne 2025 NEJM^[3]) is the most consequential trial for the sema-non-responder question. The investigators randomized 751 adults with obesity (BMI ≥ 30 without T2D) head-to-head to maximum-tolerated tirzepatide vs maximum-tolerated semaglutide 2.4 mg for 72 weeks. The mean body-weight reduction was −20.2% on tirzepatide vs −13.7% on semaglutide — a 6.5-percentage-point absolute advantage and a roughly 1.5× relative advantage. The ≥25% weight-loss endpoint was achieved by 31.6% of tirzepatide patients vs 16.1% of semaglutide patients.

SURMOUNT-5 was not a switching trial — patients started each arm de novo — but the head-to-head data anchor the clinical inference. If a patient has lost <5% on maximum-tolerated semaglutide at week 16–24, switching to tirzepatide carries a population-level expected gain of ~6–7 percentage points of body weight, putting the median sema non-responder squarely into the FDA ≥5% responder category on tirz. Real-world switching observational data are still maturing; the SURMOUNT-5 head-to-head is the best randomized evidence available to a 2026 clinician.

The practical sequential protocol

1. Week 16 visit (or week 24 if titration was slow): measure %TBWL. If ≥5%, continue on max tolerated dose. If <5%, proceed to step 2.
2. Confirm max-dose-reach + adherence + behavioral fundamentals. Was the patient actually at the max tolerated dose for ≥8 weeks at the assessment point? Is adherence ≥80%? Are sleep, OSA, stress, and appetite-stimulating polypharmacy ruled out? If any of these are modifiable, fix and reassess at week 24–28.
3. If sema non-responder: switch to tirzepatide. Start tirz 2.5 mg weekly the same day sema is stopped (no washout required given overlapping mechanism). Titrate by standard ladder. Reassess at week 16 of tirz exposure (~week 32–36 of total therapy).
4. If tirz non-responder: consider retatrutide once FDA-approved, OR refer for bariatric evaluation. Retatrutide is the GLP-1/GIP/glucagon triple agonist with Phase 2 data showing −24% at week 48 in the highest-dose arm. As of 2026 it remains investigational; Phase 3 TRIUMPH-1/2 data are anticipated 2026–2027. For BMI ≥ 40 patients who have failed sequential pharmacotherapy, bariatric referral is appropriate now; STAMPEDE 5-year data (Schauer 2017^[9]) document sleeve gastrectomy producing ~−23% body weight at 5 years.
5. Adjunctive therapy by phenotype. For emotional-hunger overlap with BED: add lisdexamfetamine. For hungry-brain overlap: add topiramate (off-label) or consider switch to Qsymia. For sleep and circadian dysregulation: CBT-I, melatonin, OSA treatment.
6. Do not discontinue the responsive agent. SURMOUNT-4 (Aronne 2024 JAMA^[8]) randomized responders to continued tirzepatide vs withdrawal at week 36; the withdrawal arm regained 14 percentage points of body weight by week 88. Once a working agent is identified, it is a chronic-therapy commitment, not a course.

Insurance and prior-authorization considerations

Most commercial payers require documented ≥5% body weight loss at week 12–16 to continue coverage of any anti-obesity GLP-1, and the documentation requirement is the primary reason the week-16 visit matters administratively. Patients who do not meet the threshold are typically denied continuation on the same agent but can re-PA onto a different GLP-1 with fresh documentation of clinical need. Switching from semaglutide to tirzepatide usually requires a new PA but does not require a new diagnosis or new step-therapy attempt in most commercial plans — the prior on-formulary GLP-1 attempt satisfies most step-therapy requirements. State Medicaid varies widely.

For the small subset of patients whose insurance denies continuation entirely, the practical cost-pivot is to compounded semaglutide or compounded tirzepatide through a 503A pharmacy with a valid clinical indication, or to cash-pay through LillyDirect (tirzepatide vials) or NovoCare (semaglutide). The cost-pivot is not a treatment failure; it is an access pathway.

Related research and tools

• GLP-1 maintenance and de-escalation protocol — what to do once a responsive dose is established and when (if ever) to step down
• Mounjaro vs Zepbound: switching evidence — same-molecule branded-switch logistics and PA considerations
• Retatrutide and the TRIUMPH program — the Phase 3 timeline and where retatrutide fits after a tirzepatide non-response
• GLP-1 to bariatric surgery: the decision framework — when sequential pharmacotherapy failure should prompt referral
• Switching between GLP-1 medications — the practical washout, titration, and side-effect-management mechanics
• Qsymia for GLP-1 non-responders — the hungry-brain phenotype add-on option
• Stress, cortisol, and GLP-1 response — modifiable behavioral co-factors before declaring non-response

Important disclaimer. This article is educational and does not constitute medical advice. The 5% / 16-week threshold is a regulatory and payer convention, not an absolute clinical cutoff; individual response trajectories vary and the decision to switch, continue, escalate, or refer should be made jointly by the patient and a qualified obesity-medicine clinician based on full history, comorbidities, side-effect tolerability, and shared decision-making. Bariatric surgery referral involves separate eligibility criteria, multi-disciplinary evaluation, and long-term nutritional follow-up. Retatrutide remains investigational as of May 2026. PMIDs were verified live against the PubMed E-utilities API on 2026-05-29.

Last verified: 2026-05-29. Next review: every 6 months, or sooner when SURMOUNT-5 secondary endpoints or TRIUMPH Phase 3 retatrutide data are published.