Effect of the GLP-1 receptor agonist exenatide on pro-inflammatory and metabolic biomarkers in individuals with alcohol use disorder: Post hoc results from a randomized, double-blinded, placebo-controlled clinical trial
- Design
- Randomized trial · 124 participants · Mixed outcome
- Match to healthy normal-weight adults aged 55–75
- Partial match[Auto] Participants selected for alcohol use disorder, not for obesity or diabetes (age/BMI not reported in abstract); results apply to that patient group, not to healthy adults.
- 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] Our findings support the well-established link between AUD and inflammation. However, treatment with the GLP-1 receptor agonist exenatide did not impact pro-inflammatory and metabolic biomarkers.
01Findings
What the study reported
- Drugs
- Exenatide
- Dose
- 1.34 mg
- Comparator
- placebo
- Primary outcome
- Not extracted
- Effect
- Not extracted
- 95% confidence interval
- Not extracted
- Follow-up
- 26 weeks
- Adverse events
- Not extracted
- Limitations
- Auto-classified from abstract only; effect estimates, adverse events and limitations not extracted. Requires manual review.
Who was studied
- Metabolic syndrome
- mentioned
- Baseline condition
- alcohol use disorder
- Sample size
- 124
Study quality details
- Study design
- Randomized controlled trial
- Sample size
- 124
- Randomization
- yes
- Blinding
- double-blind
- Comparator
- placebo
- Follow up duration
- 26 weeks
- Outcome type
- mixed
- Replication
- not assessed (auto)
- Consistency with other evidence
- not assessed (auto)
- Population applicability
- PARTIAL
- 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 (mg/L), change from baseline, exenatide vs placebo · 26 weeks (follow-up sample)
High risk of bias · RoB 2 · Two reviewers agreed on every domain
How the overall was reached: D3 is High: more than a third of randomized participants lack the 26-week hsCRP value, the shortfall differs between arms by more than 5 points, the missingness plausibly depends on the true value through discontinuation, and multiple imputation under MAR does not address that. One High domain gives an overall High judgment; D1 and D5 add Some concerns on top.
| Domain | Judgment and reasoning |
|---|---|
| D1 · Randomisation process Some concerns about risk of bias | The trial is described as randomized and double-blinded, but this report gives no method of sequence generation and no method of allocation concealment; the parent trial and protocol are cited but not available in the sources checked. Per the guide's D1 rule, a small academic trial that does not describe its method is NI, not PY. Baseline arm characteristics (age, sex, BMI strata) are similar, but one measured baseline biomarker (GIP) differed significantly between arms, so baseline balance is not clean either. NI on two of three signaling questions with no evidence of a specific problem gives Some concerns rather than High.The serum samples used in this study were derived from a randomized, placebo‐controlled, double‐blinded clinical trial conducted at outpatient alcohol clinics in Copenhagen, Denmark The design and results of the trial have been described in detail previously |
| D2 · Deviations from intended interventions Low risk of bias | Per the collection rule, D2 is judged on discontinuation imbalance and ITT handling, not on unblinding. Trial was double-blinded; discontinuation was substantial in both arms but the paper states time-to-discontinuation did not differ significantly between arms, and the analysis is an assignment-based mixed model over all randomized participants with a baseline sample, using multiple imputation rather than a completers-only comparison. No evidence of differential co-intervention or of an as-treated/per-protocol restriction.While the time‐to‐treatment discontinuation was not significantly different between the exenatide and placebo groups, the discontinuation may have impacted the statistical power to observe an effect of exenatide treatment The change in each outcome measure was compared between the placebo and exenatide groups in generalized linear mixed models. Hence, a total of 25 models were constructed, using an indicator variable, describing whether an individual belonged to the placebo or exenatide group. |
| D3 · Missing outcome data High risk of bias | Both D3 prongs trip. Of 127 randomized (65 placebo / 62 exenatide), only 77 follow-up samples were available (37 placebo / 40 exenatide): 43.1% missing in placebo and 35.5% in exenatide, i.e. far above the 20% threshold and a 7.6-point between-arm difference, above the 5-point threshold. Missingness is therefore PN to "data available for nearly all". Missingness could plausibly depend on the true value: it arises from trial discontinuation, which for a GLP-1 RA correlates with drug exposure and tolerability, and inflammatory-marker level tracks ongoing alcohol use, which also drives dropout. The multiple imputation (chained equations, predictive mean matching, BMI plus auxiliary variables correlated >0.4) is a missing-at-random procedure: it borrows strength from observed covariates but does not test or address value-dependent (MNAR) missingness, and the paper reports no MNAR or pattern-mixture sensitivity analysis. Under calibration ruling 2, both prongs tripping with no sensitivity analysis addressing value-dependent missingness gives High.A total of 127 individuals were recruited for the study ( n = 65 in the placebo group and n = 62 in the exenatide group), of which 58 participated in a Week 26 follow‐up ( n = 32 in the placebo group and n = 26 in the exenatide group). A total of 77 samples were available for follow‐up analyses, originating from participants who finished the study at 26 weeks, or discontinued the study after ≥8 weeks but had their final serum sample collected ( n = 37 from the placebo group and n = 40 from the exenatide group). |
| D4 · Measurement of the outcome Low risk of bias | hsCRP is an objective central-laboratory measurement (line immunoassay on a Cobas 8000), identical in both arms, with samples from both arms distributed across plates to control interassay variation, and outcome assessors blinded by the double-blind design. No assay change partway through is reported. The acknowledged pre-centrifugation clotting delay and variable sampling time are non-differential measurement artifacts affecting both arms equally and are recorded under limitations rather than as a domain downgrade.line immunoassays (LIA) of hsCRP, total cholesterol, LDL, HDL, and triglycerides were performed on a Cobas 8000 (Roche Diagnostics) samples from individuals allocated to the exenatide or placebo and individuals without AUD were distributed across the three plates to minimize potential effects of interassay variation |
| D5 · Selection of the reported result Some concerns about risk of bias | The paper labels itself post hoc in its own title, so the biomarker analysis was not a prespecified outcome of the parent trial, and neither the protocol (Antonsen 2018) nor the registry entry (NCT03232112) was available among the sources checked, so prespecification of this specific model, timepoint and transformation cannot be verified. Against selective reporting of a favoured result: all 25 biomarkers are reported in a single table with a single model family and a Bonferroni correction applied uniformly, and the reported result is null, so there is no sign of selection from multiple analyses for emphasis. Calibration ruling 3 (High) is not triggered: it requires a fully searchable protocol and SAP with the specific analyses affirmatively absent, which could not be established here, and this is not a load-bearing derived analysis that the paper claims was prespecified. The guide's "at least Some concerns" floor therefore applies.Post hoc results from a randomized, double‐blinded, placebo‐controlled clinical trial All analyses were performed using SAS Enterprise Guide 8.3, with a significance level of 0.05 after applying the Bonferroni correction for multiple tests. |
IL-6 (pg/mL), change from baseline, exenatide vs placebo · 26 weeks (follow-up sample)
High risk of bias · RoB 2 · Two reviewers agreed on every domain
How the overall was reached: D3 High (both thresholds tripped, no MNAR sensitivity analysis) carries the overall judgment; D1 and D5 Some concerns.
| Domain | Judgment and reasoning |
|---|---|
| D1 · Randomisation process Some concerns about risk of bias | As O1 — trial-level judgment; no sequence generation or concealment method in any available source.The serum samples used in this study were derived from a randomized, placebo‐controlled, double‐blinded clinical trial conducted at outpatient alcohol clinics in Copenhagen, Denmark GIP, pg/mL 63.1 (54.9) |
| D2 · Deviations from intended interventions Low risk of bias | As O1 — discontinuation not significantly different between arms, assignment-based mixed model over all randomized with a baseline sample.While the time‐to‐treatment discontinuation was not significantly different between the exenatide and placebo groups, the discontinuation may have impacted the statistical power to observe an effect of exenatide treatment The change in each outcome measure was compared between the placebo and exenatide groups in generalized linear mixed models. |
| D3 · Missing outcome data High risk of bias | Same missing-data structure as O1 (the follow-up serum sample carries all 25 biomarkers): 43.1% missing placebo vs 35.5% exenatide of those randomized, both prongs tripped, multiple imputation under MAR only, no MNAR sensitivity analysis.A total of 77 samples were available for follow‐up analyses, originating from participants who finished the study at 26 weeks, or discontinued the study after ≥8 weeks but had their final serum sample collected ( n = 37 from the placebo group and n = 40 from the exenatide group). A relatively high proportion of missing data was observed for the outcome measures, which may have affected the risk estimates. |
| D4 · Measurement of the outcome Low risk of bias | IL-6 measured by validated central multiplex immunoassay (MSD V-PLEX Pro-inflammatory Panel 1) on the same platform for both arms, with arms distributed across plates and assessors blinded. Values below the limit of detection were imputed by a stated rule applied identically in both arms, so any resulting error is non-differential.Serum samples ( n = 234) were analyzed using multiplexed immunoassays (Meso Scale Discovery©): V‐Plex Pro‐inflammatory Panel 1 (human), V‐Plex Metabolic Panel 1 (human) concentrations of samples below the lower limit of detection were imputed from a uniform distribution between zero and the lower limit of detection |
| D5 · Selection of the reported result Some concerns about risk of bias | As O1 — self-declared post hoc analysis, protocol and registry entry not available for verification; mitigated by uniform reporting of all 25 markers with a single prespecified-in-paper model family and Bonferroni correction.Post hoc results from a randomized, double‐blinded, placebo‐controlled clinical trial At 26 weeks of follow‐up, no significant differences were observed in any of the 25 biomarkers after Bonferroni corrections (Table 3 ). |
TNF-alpha (pg/mL), change from baseline, exenatide vs placebo · 26 weeks (follow-up sample)
High risk of bias · RoB 2 · Two reviewers agreed on every domain
How the overall was reached: D3 High carries the overall judgment; D1 and D5 Some concerns.
| Domain | Judgment and reasoning |
|---|---|
| D1 · Randomisation process Some concerns about risk of bias | As O1 — trial-level judgment; randomization asserted but no method described in the available sources.The serum samples used in this study were derived from a randomized, placebo‐controlled, double‐blinded clinical trial conducted at outpatient alcohol clinics in Copenhagen, Denmark GIP, pg/mL 63.1 (54.9) |
| D2 · Deviations from intended interventions Low risk of bias | As O1.While the time‐to‐treatment discontinuation was not significantly different between the exenatide and placebo groups, the discontinuation may have impacted the statistical power to observe an effect of exenatide treatment The change in each outcome measure was compared between the placebo and exenatide groups in generalized linear mixed models. |
| D3 · Missing outcome data High risk of bias | Same missing-data structure as O1 and O2; both thresholds tripped; MAR-based multiple imputation only, with no analysis addressing value-dependent missingness.A total of 77 samples were available for follow‐up analyses, originating from participants who finished the study at 26 weeks, or discontinued the study after ≥8 weeks but had their final serum sample collected ( n = 37 from the placebo group and n = 40 from the exenatide group). A relatively high proportion of missing data was observed for the outcome measures, which may have affected the risk estimates. |
| D4 · Measurement of the outcome Low risk of bias | TNF-α measured on the same validated central multiplex platform in both arms, with plate balancing across allocation and blinded assessment. Note only that TNF-α values sit near the low end of the assay range (means 1.2-1.4 pg/mL) and below-LOD imputation applies, but identically in both arms, so the error is non-differential.Serum samples ( n = 234) were analyzed using multiplexed immunoassays (Meso Scale Discovery©): V‐Plex Pro‐inflammatory Panel 1 (human) samples from individuals allocated to the exenatide or placebo and individuals without AUD were distributed across the three plates to minimize potential effects of interassay variation |
| D5 · Selection of the reported result Some concerns about risk of bias | As O1 — post hoc by the paper's own label, prespecification unverifiable from the available sources; uniform reporting of all 25 markers mitigates selection among results.Post hoc results from a randomized, double‐blinded, placebo‐controlled clinical trial All analyses were performed using SAS Enterprise Guide 8.3, with a significance level of 0.05 after applying the Bonferroni correction for multiple tests. |
02Funding
Funding and conflicts
- Funding
- P.A. Messerschmidt og Hustrus Fond; Lundbeck Foundation; Slagtermester Max Worzner og hustru Inger Worzners mindelegat; Grosserer L.F. Foghts Fond; The A.P. Moeller Foundation; The Augustinus Foundation; The Research Foundation, Capital Region of Denmark; The Ivan Nielsen Foundation; The Novavi Foundation; The Research Foundation, Mental Health Services, Capital Region of Denmark; Hartmann Foundation; Aase and Ejnar Danielsen Foundation
- Industry funded
- No
- Manufacturer
- Novo Nordisk, Eli Lilly, Lilly, AstraZeneca, Boehringer Ingelheim, Amgen
- Sponsor role
- no manufacturer funding identified
- Author conflicts
- TV has been part of speaker's bureaus, served on scientific advisory panels, served as a consultant to and/or received research support from Amgen, Boehringer Ingelheim, Eli Lilly, Gilead, AstraZeneca, Mundipharma, MSD/Merck, Novo Nordisk, and Sun Pharmaceuticals. AF‐J has received an unrestricted research grant from Novo Nordisk to investigate the effects of GLP‐1 receptor stimulation on metabolic disturbances in antipsychotic‐treated patients with a diagnosis of schizophrenia and serves on an advisory panel for Novo Nordisk (no honorarium).
- Independent replication
- unknown
- Notes
- Authors declare relationships with the drug's manufacturer: AstraZeneca (originally Amylin/Eli Lilly)
Funding is shown on every study and never used to score it.
04Source
The source, as retrieved
Abstract
[BACKGROUND] Alcohol use disorder (AUD) has been associated with inflammation, metabolic syndrome, and increased risk of all-cause mortality. This study aimed to compare the pro-inflammatory and metabolic biomarker profiles in individuals with AUD with individuals without AUD, and to evaluate the effect of exenatide on these biomarkers in individuals with AUD. [METHODS] Serum concentrations of 25 biomarkers (interferon-γ [IFN-γ], tumor necrosis factor-α [TNF-α], interleukin (IL)-1β, IL-2, IL-4, IL-6, IL-8, IL-10, IL-12p70, IL-13, monocyte chemoattractant protein-1 [MCP-1], C-peptide, gastric inhibitory polypeptide [GIP], glucagon-like peptide [GLP-1], glucagon, insulin, leptin, pancreatic polypeptide [PP], adiponectin, high sensitivity C-reactive protein [hsCRP], fibroblast growth factor 21 [FGF-21], total cholesterol [CHOL], high-density lipoprotein [HDL], low-density lipoprotein [LDL], and triglycerides [TG]) from individuals with AUD were measured at baseline and after 26 weeks of treatment with the GLP-1 receptor agonist (GLP-1RA) exenatide once-weekly or placebo, using multiplexed immunoassays, enzyme-linked immunosorbent assay (ELISA), and line immunoassays. Serum samples from 23 individuals with no record of AUD or treatment with a GLP-1RA were measured once for comparison with individuals with AUD. [RESULTS] IL-6 (1.56 vs. 0.62 pg/mL), hsCRP (3.30 vs. 1.34 mg/L), and FGF-21 (1794.97 vs. 306.11 pg/mL) were significantly higher, whereas GIP (63.06 vs. 111.07 pg/mL) was significantly lower in individuals with AUD (n = 124) than in those without AUD (n = 23). No significant changes in biomarker levels were observed after treatment with exenatide (n = 40) compared with treatment with placebo (n = 37). [CONCLUSION] Our findings support the well-established link between AUD and inflammation. However, treatment with the GLP-1 receptor agonist exenatide did not impact pro-inflammatory and metabolic biomarkers.
Where this record came from
| Source | Retrieved | Identifier |
|---|---|---|
| pubmed | Sep 13, 2026 | 40630018 first ingestion |