Weight is now a routine problem in HIV care, and it arrived with the treatment that fixed everything else. Modern antiretroviral therapy stops the virus and, in a large share of people, does not stop there: a review of the field describes weight gain extending well past the old return-to-health phase into overweight and obesity within the first one to two years, with the largest signals attached to starting dolutegravir and tenofovir alafenamide, or coming off tenofovir disoproxil and efavirenz, and with women, lower baseline CD4 counts and higher viral loads all associated with more of it.[1] The obvious next question — whether a GLP-1 drug is the answer — has a literature. It is just much smaller than the conversation around it.
The labels do not mention antiretroviral therapy at all
Read on DailyMed on September 15, 2026, none of the four manufacturer labels in this class — Wegovy, Ozempic, Zepbound, Mounjaro — contains the word “antiretroviral.” None contains a reference to HIV in its own text. The drug interactions sections address insulin and insulin secretagogues, and then oral medications as a general category, with levothyroxine as the one drug singled out by name; the measured absorption effects behind that section are covered in the oral medications article.
That silence is easy to misread in either direction. It is not a clearance, because no dedicated interaction study between these drugs and any antiretroviral has been published. It is also not a warning, because the mechanism most people worry about is a weak one: these peptides show very low potential to inhibit or induce CYP enzymes or drug transporters, which is the pathway most protease-inhibitor and non-nucleoside-reverse-transcriptase interactions run through. What remains is gastric emptying, which reliably lowers the peak concentration of a co-administered oral drug and diminishes with repeated dosing. For an antiretroviral regimen whose effectiveness depends on trough concentrations rather than peaks, that is a theoretical concern that nobody has measured. The nearest analogous work in this corpus is the transplant literature in the immunosuppressants article, where a predicted interaction largely failed to appear.
The one randomized trial
A phase 2b trial at a single US site randomized 108 participants with controlled HIV-1, a body-mass index of 25 or higher and lipohypertrophy but without diabetes, 1:1 to semaglutide or placebo. The schedule was an eight-week titration followed by 24 weeks at 1.0 mg weekly, 32 weeks in total. Eight participants in each arm withdrew early.[2]
The primary outcome was adipose tissue by compartment, and it moved. Abdominal visceral adipose tissue fell by 30.82 cm² (95% CI −50.13 to −11.51), a 30.6% change. Abdominal subcutaneous adipose tissue fell 42.01 cm² (95% CI −75.49 to −8.52), an 11.2% change, and total body fat fell 18.9%. Possibly related or related adverse events did not differ significantly between groups — absolute risk difference 0.1111 (95% CI −0.0727 to 0.2869) — but the semaglutide arm recorded one grade 4 elevated lipase and two cases of cholelithiasis, and the authors state that the potential for serious adverse events deserves scrutiny in larger trials.[2]
Three cases in a 54-person arm is not a rate. It is also the entire randomized safety record in this population, which is the point.
The other study has no control group
The second prospective dataset comes from an open-label, single-arm phase 2b study of semaglutide in people with HIV and fatty liver disease, titrated to 1 mg weekly by week 4 and run for 24 weeks, followed by 24 weeks off treatment. It enrolled around 50 participants — median age 52, median body-mass index 35, 43% women, 33% Black and 39% Hispanic.[3]
Its off-treatment half is the more useful half. After a mean weight loss of 7.8 kg (95% CI 6.1 to 9.5) over 24 weeks, the 24 weeks that followed produced a mean regain of 2.9 kg (95% CI 1.5 to 4.3), with waist circumference up 2.0 cm (0.9 to 3.1) and fasting glucose up 5.1 mg/dL (0.9 to 9.3). Blood pressure, total and LDL cholesterol, triglycerides and metabolic syndrome status did not change significantly over that period.[3] The authors describe the pattern as similar to the general population, where the withdrawal data are in what happens when you stop.
The muscle finding, and the reason it is not simple
A substudy of the same trial imaged the psoas muscle in 46 participants. Over 24 weeks at 1 mg weekly, psoas muscle volume fell 9.3% (95% CI −13.4 to −5.2, P < .001), while psoas muscle fat did not change (−0.42%, P = .16).[4]
Physical function went the other way. The ten-time chair rise test improved by 1.27 seconds and gait speed by 0.05 m/s, neither significantly, and the prevalence of slow gait speed below 1 m/s fell from 63% to 46% (P = .029).[4] Function was maintained, and by one measure improved, despite a measurable loss of muscle — a pattern consistent with what the general lean-mass literature shows in the muscle article. The caution attached to it is population-specific: people aging with HIV carry a higher background burden of frailty, and a 9.3% muscle loss at 1 mg says nothing about what 2.4 mg would do over a year, because that has not been run.
Everything above happened at under half the labeled dose
This is the single most important qualifier on the page. The randomized trial capped at 1.0 mg. The single-arm study capped at 1.0 mg. The labeled maintenance dose of semaglutide for weight reduction is 2.4 mg weekly, reached through the schedule in the dose article, and no trial in people with HIV has gone there. A review of the field states the limitation directly: HIV-specific studies have largely used lower doses of semaglutide, and evidence in older populations remains sparse.[5] The same review notes that concerns about lean mass, frailty and bone health exist and that current data, while limited, are generally reassuring on function.
So the honest translation of a 30.6% visceral fat reduction is: at 1 mg, for 32 weeks, at one site, in 54 people, against placebo. Anyone quoting a registration-trial weight figure at a person with HIV is crossing a dose gap and a duration gap that no published study has bridged.
The metabolic penalty is measured; the cardiovascular one is not
The case for treating this weight rests on where it is assumed to lead, and the largest analysis of that question complicates it. A target trial emulation inside a global cardiovascular prevention trial followed 5,114 participants across 99,357 person-trials, comparing people who switched to an integrase strand-transfer inhibitor against people who stayed on a non-integrase regimen, over five years.[6]
Switchers had higher risk of obesity at hazard ratio 1.41 (95% CI 1.22 to 1.59), diabetes at 1.50 (1.24 to 1.81) and hypertension at 1.45 (1.26 to 1.67). Major adverse cardiovascular events did not follow: hazard ratio 1.17 (95% CI 0.87 to 1.57).[6] Three intermediate endpoints moved by roughly half a risk unit each, and the outcome they are supposed to predict did not move at all over five years in a low-to-moderate-risk cohort. That does not make the weight gain harmless; it means the argument from weight gain to heart attacks in this population is an extrapolation, not an observation.
What is being claimed beyond the data
A post hoc analysis of the randomized trial, using blood DNA methylation in 45 semaglutide and 39 placebo participants, reported reductions across several epigenetic aging measures, including PhenoAge at 4.9 years per year (P = 0.004) and a 9% slower pace on DunedinPACE (P = 0.01). Its authors state plainly that epigenetic aging was not a prespecified endpoint and that the post hoc design, modest sample size, HIV-specific cohort and 32-week follow-up limit generalizability.[7] It is a hypothesis with a p-value attached, and it is already being repeated as an anti-aging claim.
What nobody has published
No interaction study between any drug in this class and any antiretroviral. No trial at 2.4 mg or at any tirzepatide dose in people with HIV. No trial longer than 32 weeks. No cardiovascular, bone or frailty outcome trial. No data on whether delayed gastric emptying affects adherence to a once-daily regimen that has to be taken for life, which is the question with the largest downside if the answer is bad.
There is one further gap specific to how these drugs are bought. Every figure above came from branded product supplied inside a trial. Most cash-pay sellers dispense compounded semaglutide or tirzepatide, which is not FDA-approved and is not reviewed by the FDA for safety, efficacy or quality before it is dispensed — the subject of what a compounded vial contains, and the products listed on the semaglutide boards. An online intake that does not ask what else a person takes every day cannot flag a regimen it never saw, and HIV care is the case where continuity between the two prescribers matters most.