The Effects of Semaglutide on Inflammation and Immune Activation in HIV-associated Lipohypertrophy
- Design
- Randomized trial · 108 participants · Mixed outcome
- Match to healthy normal-weight adults aged 55–75
- Population unclear[Auto] Population characteristics not extractable from abstract (age/BMI not reported in abstract). Needs manual review.
- Could weight loss explain it?
- Unknown[Auto] Not addressed in abstract.
- Study tier
- Study tier 3[Auto] Small or short randomized trial (auto-provisional).
- Assessment
- Version 1 · automatic, not yet reviewed by a person · Sep 13, 2026
Study facts come from the paper. Population match, weight-loss explanation and study tier are our judgments, made against the reference group of healthy normal-weight adults aged 55–75.
[Auto, unreviewed; quoted from abstract conclusions] In this randomized controlled trial of semaglutide in PWH, we report significant decreases in markers of inflammation that are associated with morbidity and mortality in this population. These results add to the growing literature demonstrating the anti-inflammatory effects of semaglutide. Further studies in PWH are warranted.
01Findings
What the study reported
- Drugs
- Semaglutide
- Dose
- 1.0 mg weekly
- Treatment duration
- 32 weeks
- Route
- subcutaneous
- Comparator
- placebo
- Primary outcome
- Not extracted
- Effect
- Not extracted
- 95% confidence interval
- Not extracted
- Follow-up
- 18 years
- Adverse events
- Not extracted
- Limitations
- Auto-classified from abstract only; effect estimates, adverse events and limitations not extracted. Requires manual review.
Who was studied
- Diabetes status
- excluded (no diabetes)
- Sample size
- 108
Study quality details
- Study design
- Randomized controlled trial
- Sample size
- 108
- Randomization
- yes
- Blinding
- double-blind
- Comparator
- placebo
- Follow up duration
- 18 years
- Outcome type
- mixed
- Replication
- not assessed (auto)
- Consistency with other evidence
- not assessed (auto)
- Population applicability
- UNKNOWN
- Statistical precision
- not extracted
- Funding conflicts
- no
- Peer review status
- yes
01bRisk of bias
How much the result can be relied on
Risk of bias is judged per result, not per paper: the same study can report one marker at low risk and another at high. Two reviewers assess each result independently against a written guide, and every domain judgment below carries its reasoning and the sentences it rests on.
These words are not quality scores. On the RoB 2 scale, High risk of bias is the worst rating a result can receive and Low risk of bias is the best — the opposite of how the same words read on some other scales. Levels are always written out in full here for that reason.
hsCRP (high-sensitivity C-reactive protein), change vs placebo · week 32 (32 weeks of treatment)
Some concerns about risk of bias · RoB 2 · Two reviewers agreed on every domain
How the overall was reached: Highest domain judgment is Some concerns, at D1 (self-disclosed baseline imbalance in the outcome variable itself, partially handled by covariate adjustment) and D5 (multiple transformations of the same result, data-driven adjustment set, and a significance claim in tension with the table's own BH footnote). No domain is High and the result is not judged at high risk of bias, but the effect estimate depends on a model chosen after the imbalance was observed.
| Domain | Judgment and reasoning |
|---|---|
| D1 · Randomisation process Some concerns about risk of bias | Sequence generation is adequately described (independent study statistician, online generator, 1:1, block size 6) and allocation concealment is plausible from a statistician-held central online sequence, though no concealment mechanism (sealed envelopes, IVRS, pharmacy-held list) is described, so concealment is at best PY on inference rather than verified. v1.2 item 6 does not apply: this is a single-site NIH-funded phase IIb trial reporting its own preplanned outcomes, not a secondary analysis inheriting a large industry registration parent, so the presumption is not reached and D1 is judged on what this paper describes. Decisively, the paper self-discloses baseline imbalance in the outcome variables themselves - hsCRP was significantly higher in placebo (P = .01, BH 0.02), as were IL-6 and sICAM-1 - plus imbalance in sex (70% vs 50% male) and current smoking (28% vs 43%). Under plain RoB 2 signalling question 1.3 this is baseline imbalance suggesting a problem with randomization, and under v1.2 it would in any case rebut the presumption. The analysis adjusts for baseline value, age, sex and smoking, which mitigates but does not eliminate the concern. Not High: the imbalance is confined to a handful of skewed biomarkers and two covariates in a 108-participant block-randomized trial, which chance can produce, and the randomization procedure itself is described and unremarkable.Study statistician (A.S.) provided randomization allocation sequences generated from an online software program ( sealedenvelope.com ) with a 1:1 ratio and block size of 6; key personnel, study staff, participants, and participants’ medical providers were masked to treatment assignment. Many of the inflammatory biomarkers differed between the 2 groups (semaglutide vs placebo) at baseline. |
| D2 · Deviations from intended interventions Low risk of bias | Judged on discontinuation imbalance and ITT handling per the collection rule, not on unblinding. Premature withdrawal was exactly balanced (8 of 54, 15%, in each arm), and the analysis is explicitly ITT by randomized assignment using all available data. The trial was double-blinded with matching placebo. No information on per-protocol adherence or on co-interventions differing between arms beyond statin use, which was similar; dietary change did not differ. Nothing cited suggests deviations arising from the trial context that would affect the outcome.All analyses were performed using intention-to-treat principles based on randomized treatment assignment, including all available data. Eight participants in each group withdrew from the study before the 32-week endpoint. |
| D3 · Missing outcome data Low risk of bias | Neither v1.1 prong trips. Missingness is 8/54 = 14.8% per arm, below the 20% threshold, and the between-arm difference is 0 points, below the 5-point threshold. Under calibration ruling 2 the thresholds are decisive both directions, so D3 is Low absent specific contrary evidence, and the absence of an MNAR sensitivity analysis is recorded in limitations rather than as a domain downgrade. No information is given on whether any additional participants lacked an assayed week-32 sample beyond the 8 withdrawals per arm; nothing cited indicates such loss.A total of 108 participants were enrolled and evenly randomized to semaglutide versus placebo. Eight (15%) in each group withdrew prematurely. including all available data |
| D4 · Measurement of the outcome Low risk of bias | hsCRP is an objective biomarker measured by a single commercial ELISA (R&D Systems) in one laboratory, on samples batched and stored at -80 C without prior thaw, with laboratory personnel blinded to arm. Intra- and inter-assay variability are reported. No indication the assay changed partway through or differed between arms. Per the collection rule, central-lab assays of this kind are Low.Plasma was isolated from blood samples collected in EDTA anticoagulant tubes and were stored at −80 °C and batched until processing without a prior thaw. The intra-assay variability ranged between 4% and 8% and inter-assay variability was less than 10% for all markers. All assays were performed in Dr. Funderburg's laboratory at Ohio State University, Columbus, Ohio, and laboratory personnel were blinded to group assignments. |
| D5 · Selection of the reported result Some concerns about risk of bias | The paper states these inflammation outcomes were preplanned, which supports the outcome domain being prespecified, but no protocol, SAP or registry entry was retrieved, so prespecification of the specific analysis is unverifiable. More pointedly, hsCRP is reported in at least four framings of the same data - absolute change, percent change, unadjusted log regression, and covariate-adjusted log/median regression - plus a baseline-hsCRP-median stratification, and the adjustment set was chosen after seeing baseline imbalance, i.e. data-driven model selection. The guide's rule on multiple CRP transformations with one chosen for emphasis lands squarely here. A further selective-emphasis signal: Table 5's own footnote states no P values survived BH correction, while the Results text presents the adjusted hsCRP effect as significant. Not High: the outcome itself is declared preplanned and the hsCRP reduction is directionally consistent across every framing reported, so this is not the affirmatively-absent-prespecification shape of calibration ruling 3.Here, we present the preplanned outcomes of 32 weeks of semaglutide treatment on plasma markers of inflammation and on immune cell populations in PWH with HIV-associated lipohypertrophy. Depending on the distribution of the data, we used either linear regression or quantile (median) regression models to estimate the effect of semaglutide on the outcome variables or biomarkers. Because of the differences in biomarkers at baseline, we controlled for heterogeneities by adding baseline values of the outcome variables as covariates. |
IL-6, change vs placebo · week 32
Some concerns about risk of bias · RoB 2 · Two reviewers agreed on every domain
How the overall was reached: Highest domain is Some concerns, at D1 (baseline imbalance in IL-6 itself, the outcome being assessed) and D5 (divergent framings with the non-randomized within-arm result given headline emphasis over a null between-group estimate). D2, D3 and D4 are Low. The assessed between-group result is itself null, so the risk here is chiefly to how the finding is presented rather than to the estimate carried forward.
| Domain | Judgment and reasoning |
|---|---|
| D1 · Randomisation process Some concerns about risk of bias | Same randomization and same rebutting baseline imbalance as O1, and here the imbalance is in IL-6 itself (2.52 vs 3.33 pg/mL, P = .02, BH 0.04, higher in placebo). Sequence generation described and adequate; concealment inferred, not verified; v1.2 item 6 not reached (single-site self-reporting phase IIb trial, no parent trial). Baseline imbalance in the assessed outcome is a plain RoB 2 signalling-question-1.3 concern, mitigated by adjustment for the baseline value but not removed.Study statistician (A.S.) provided randomization allocation sequences generated from an online software program ( sealedenvelope.com ) with a 1:1 ratio and block size of 6 IL-6, pg/mL 2.52 (1.46–3.76) 3.33 (2.16–5.02) .02 |
| D2 · Deviations from intended interventions Low risk of bias | Identical to O1: balanced 15% withdrawal in each arm, explicit ITT analysis by randomized assignment on all available data, double-blind with matching placebo, no arm difference in statin use or dietary change.All analyses were performed using intention-to-treat principles based on randomized treatment assignment, including all available data. Eight (15%) in each group withdrew prematurely. |
| D3 · Missing outcome data Low risk of bias | Identical to O1: 14.8% missing per arm (below 20%) with a 0-point between-arm difference (below 5 points). Both v1.1 prongs untripped, so Low is decisive; the absent MNAR sensitivity analysis is a limitations note.A total of 108 participants were enrolled and evenly randomized to semaglutide versus placebo. Eight (15%) in each group withdrew prematurely. |
| D4 · Measurement of the outcome Low risk of bias | IL-6 measured by the same single-laboratory R&D Systems ELISA on batched, never-thawed plasma, with laboratory personnel blinded to arm and reported assay variability under 10%. Objective marker, no assay change, no between-arm difference in measurement.using an enzyme-linked immunosorbent assay, we measured the following plasma biomarkers: sCD14, soluble CD163 (sCD163), tumor necrosis factor receptors I and II, vascular and intercellular cell adhesion molecule 1 (VCAM-1 and ICAM-1), hsCRP, and IL-6, all from R & D Systems, Minneapolis, Minnesota. laboratory personnel were blinded to group assignments |
| D5 · Selection of the reported result Some concerns about risk of bias | The IL-6 outcome is declared preplanned, but no protocol/SAP/registry was retrieved to verify the analysis, and IL-6 is reported in five framings whose conclusions diverge: within-arm change is significant (P = .02), the between-group absolute change is not (P = .43), the between-group percent change is not (P = .56), the adjusted regression is a non-significant trend (P = .07), and a baseline-hsCRP-median subgroup is significant (P = .01). The abstract and conclusions foreground the within-arm result ("reduced baseline levels of C-reactive protein, interleukin-6 ... all P < .02"), which is not a randomized comparison, while the between-group estimate does not reach significance. That is selection of the reported result for emphasis. Not High: the outcome is stated to be preplanned and the full set of framings is reported in the tables rather than suppressed, so a reader can recover the null between-group result.Thirty-two weeks of semaglutide treatment reduced baseline levels of C-reactive protein, interleukin-6, and soluble CD163 (all P < .02) IL-6, pg/mL −0.39 (−1.62–0.30) .02 d −0.25 (−1.66 to 0.74) .25 .43 0.06 |
Inflammatory marker change not explained by weight change - correlation of change in hsCRP with change in weight and with change in abdominal visceral adipose tissue (VAT), reported as absent; with a VAT-IL-6 association found overall but not within the semaglutide arm · week 32 (change from baseline to week 32)
High risk of bias · RoB 2 · Two reviewers differed; settled by a recorded ruling
Why the overall is not simply the worst domain: rob2-guide v1.3 item 7 limit: the weakness is that the weight-independence analysis barely exists — non-randomized, unreported n, unprespecified — which is carried in the overall rather than in D5. data/assessments/Q-020-D5-resolution-2026-09-14.md
| Domain | Judgment and reasoning |
|---|---|
| D1 · Randomisation process Some concerns about risk of bias | Same trial-level randomization judgment as O1/O2 - adequate described sequence generation, concealment inferred rather than verified, v1.2 item 6 not reached - with the same self-disclosed baseline imbalance in hsCRP, IL-6, sICAM-1, sex and smoking. Per calibration ruling 4, D1 stays trial-level: the non-randomized character of a correlation/mediation contrast is carried at D5 and in the overall judgment, not written into D1.Study statistician (A.S.) provided randomization allocation sequences generated from an online software program ( sealedenvelope.com ) with a 1:1 ratio and block size of 6 Many of the inflammatory biomarkers differed between the 2 groups (semaglutide vs placebo) at baseline. |
| D2 · Deviations from intended interventions Low risk of bias | Trial-level deviations and ITT handling are as for O1: balanced 15% withdrawal per arm, ITT by randomized assignment on all available data. No information suggests the weight/VAT correlation analyses departed from the randomized population in a way driven by deviations; the analytic concern belongs at D5.All analyses were performed using intention-to-treat principles based on randomized treatment assignment, including all available data. Eight (15%) in each group withdrew prematurely. |
| D3 · Missing outcome data Some concerns about risk of bias | Trial-level missingness is 14.8% per arm with no between-arm difference, so neither v1.1 prong trips for the biomarker itself. But this derived result requires paired biomarker AND paired weight/VAT (DXA/CT imaging) measurements, and the paper gives no denominator for the correlation analyses - the supplementary tables that hold them were not retrieved. Data availability for the assessed result is therefore NI rather than demonstrated, and imaging-based VAT is more prone to additional loss than a plasma assay. Some concerns, not Low, because the assessed result's own denominator is unknown; not High, because nothing cited indicates substantial or differential loss.There were no significant correlations between changes in hsCRP and changes in weight or changes in abdominal visceral adipose tissue (VAT) ( Supplementary Tables 3A and 3B ) |
| D4 · Measurement of the outcome Low risk of bias | Both components are objectively measured: the inflammatory markers by blinded single-laboratory ELISA, and weight/VAT by the anthropometric and imaging measures reported in the parent body-composition paper. Per the collection rule, central-lab assays are Low, and nothing indicates measurement differed between arms or changed during the trial.All assays were performed in Dr. Funderburg's laboratory at Ohio State University, Columbus, Ohio, and laboratory personnel were blinded to group assignments. We have previously reported that semaglutide reduced total body fat, trunk/abdominal fat, limb fat, and weight in this study population |
| D5 · Selection of the reported result Some concerns about risk of bias | This is the affirmatively-absent-prespecification shape. The Statistical Analyses section is complete and specific - regression model choice, baseline-value adjustment, age/sex/smoking covariates, ITT, between-group tests, software - and it contains no mention whatsoever of correlating biomarker change with weight or VAT change, of a within-arm subgroup analysis of that correlation, or of mediation analysis. Those appear for the first time in the Results narrative. The paper's "preplanned outcomes" claim covers the marker outcomes, not these derived weight-independence analyses. The analyses are load-bearing: they are the only basis on which this study speaks to C-001 at all. Two further selection signals: the mediation analyses actually performed target glucose, HbA1c and insulin, not weight, so the weight-independence inference rests on absence of a correlation rather than on a prespecified test of it; and the one association that was found (VAT with IL-6 change) is immediately qualified as "insignificant in the semaglutide group in a subgroup analysis", a post hoc subgroup used to set aside an inconvenient result. Under calibration ruling 3, "at least Some concerns" is a floor and this clears it: High.After rigorous data quality checks using descriptive analyses and graphs, we plotted biomarkers for visualization over the study period. Depending on the distribution of the data, we used either linear regression or quantile (median) regression models to estimate the effect of semaglutide on the outcome variables or biomarkers. There were no significant correlations between changes in hsCRP and changes in weight or changes in abdominal visceral adipose tissue (VAT) ( Supplementary Tables 3A and 3B ), however, we did find that changes in VAT were associated with changes in IL-6 (β = –0.02; 95% CI, −0.03 to −0.004; P = .012), but this association was insignificant in the semaglutide group in a subgroup analysis. |
02Funding
Funding and conflicts
- Funding
- NIDDK NIH HHS; NCATS NIH HHS
- Industry funded
- No
- Manufacturer
- None identified
- Sponsor role
- no manufacturer funding identified
- Author conflicts
- Potential conflicts of interests. N.F. has received research funding from Gilead unrelated to this project. G.A.M. served as a consultant for Merck, Gilead, and ViiV/GSK. A.R.E. served as an advisor for Gilead Sciences and Theratechnologies. All other authors report no potential conflicts.
- Independent replication
- unknown
Funding is shown on every study and never used to score it.
04Source
The source, as retrieved
Abstract
[BACKGROUND] Cardiovascular and metabolic comorbidities are common in people with HIV (PWH) and are linked to chronic inflammation and immune activation. We assessed the effects of semaglutide on plasma markers of immune activation/inflammation that are known to be increased in PWH and are associated with morbidity and mortality in this population. [METHODS] We conducted a single-site, randomized, double-blinded, placebo-controlled trial of virologically suppressed, nondiabetic PWH ≥18 years of age on stable antiretroviral therapy with body mass index ≥ 25 kg/m2, increased waist circumference/waist-to-hip ratio, and subjective increased abdominal girth after antiretroviral therapy initiation (clinicaltrials.gov: NCT04019197). Participants were randomized 1:1 to 32 weeks of semaglutide (8-week titration + 24 weeks of 1.0 mg weekly subcutaneous injection) or matching placebo. Signed-rank tests were used to determine changes over 32 weeks in soluble markers and cellular phenotypes of inflammation/immune activation within groups; semaglutide effects were assessed using linear or quantile regression analyses. [RESULTS] A total of 108 participants were enrolled and evenly randomized to semaglutide versus placebo. Eight (15%) in each group withdrew prematurely. Thirty-two weeks of semaglutide treatment reduced baseline levels of C-reactive protein, interleukin-6, and soluble CD163 (all P < .02) and trended to reduce levels of sCD14 (P = .08). Circulating monocyte proportions and T-cell phenotypes were not altered by semaglutide. [CONCLUSIONS] In this randomized controlled trial of semaglutide in PWH, we report significant decreases in markers of inflammation that are associated with morbidity and mortality in this population. These results add to the growing literature demonstrating the anti-inflammatory effects of semaglutide. Further studies in PWH are warranted.
Where this record came from
| Source | Retrieved | Identifier |
|---|---|---|
| pubmed | Sep 13, 2026 | 40160348 first ingestion |