GLP-1 Evidence

Not medical advice. A record of published research and our assessments of it. The limits

Thromboxane-Dependent Platelet Activation in Obese Subjects with Prediabetes or Early Type 2 Diabetes: Effects of Liraglutide- or Lifestyle Changes-Induced Weight Loss

Simeone P, Liani R, Tripaldi R, Di Castelnuovo A et al. · Nutrients · Dec 2, 2018
DOI 10.3390/nu10121872 · PubMed 30513818

Design
Randomized trial · Mixed outcome
Match to healthy normal-weight adults aged 55–75
Different population[Auto] Participants had obesity/overweight (age/BMI not reported in abstract).
Could weight loss explain it?
Possibly[Auto] Participants had obesity and/or type 2 diabetes and the abstract does not separate direct drug effects from weight loss or glycaemic improvement.
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; last sentences of abstract] In obese patients with initial impairment of glucose metabolism, the extent of platelet activation is related to systemic inflammation, isoprostane formation and degree of glycemic control and abdominal SAT. Successful weight loss, achieved with either lifestyle changes or an incretin-based therapy, is associated with a significant reduction in lipid peroxidation and platelet activation.

01Findings

What the study reported

Drugs
Liraglutide
Dose
1.8 mg
Route
subcutaneous
Primary outcome
Not extracted
Effect
Not extracted
95% confidence interval
Not extracted
Follow-up
Not stated
Adverse events
Not extracted
Limitations
Auto-classified from abstract only; effect estimates, adverse events and limitations not extracted. Requires manual review.

Who was studied

Obesity status
obesity/overweight present (all or most)
Diabetes status
diabetes mentioned

Study quality details

Study design
Randomized controlled trial
Sample size
not extracted
Randomization
yes
Blinding
not stated
Comparator
not stated
Follow up duration
not stated
Outcome type
mixed
Replication
not assessed (auto)
Consistency with other evidence
not assessed (auto)
Population applicability
INDIRECT
Statistical precision
not extracted
Funding conflicts
unclear
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.

hs-CRP (serum, high-sensitivity immunoassay), between-arm difference in change, liraglutide 1.8 mg/day vs structured lifestyle counselling, both arms treated to the same -7% initial-body-weight target · At attainment of the -7% weight-loss target; median time-to-weight-loss 4 months (IQR 3-6), no arm difference (4.0 (3.2-6.0) vs 4.0 (3.0-6.0) months); ceiling 15 months

High risk of bias · RoB 2 · Two reviewers agreed on every domain

How the overall was reached: High follows from the algorithm (two High domains, D2 and D3, plus High at D5). It is also the right substantive answer. The design shape is the strongest available for O4 - a randomized, weight-matched, head-to-head comparator with both arms driven to the same -7% target, adjusted for baseline imbalance - but the matching is achieved by conditioning on success: 27 of 62 randomized participants are gone, 10 of them removed and replaced precisely for failing to reach the outcome-defining weight change, and the analysed 35 are no longer randomized groups. A null from that construction is much weaker than a null from an intact randomization. Two further things cap how much the result can bear, and each is recorded rather than folded into a domain: the assessed hs-CRP contrast has no estimate, no confidence interval and no p-value, so its compatibility interval is unknown and "no difference" cannot be distinguished from "underpowered and unreported" (the authors' own first limitation is sample size, and their power calculation was sized for a 1-SD difference in a different marker, 11-dehydro-TXB2, not for hs-CRP); and liraglutide was withheld ~36 h before sampling, so the drug-arm measurement was taken below steady state, biasing any drug-specific increment toward zero. Direction of bias for C-001: every identified problem pushes the liraglutide-vs-lifestyle contrast toward the null, which is the result reported. This study should not be used as evidence against a weight-independent effect; at most it is CONSISTENT_WITH_BUT_DOES_NOT_TEST in the null direction, with high risk of bias.

DomainJudgment and reasoning
D1 · Randomisation process
Some concerns about risk of bias		Sequence generation is adequately described (computer-generated, blocks of four, prepared by the trial statistician). Allocation concealment is not described, and the one sentence that touches it - allocation by consecutive random number in order of inclusion - describes assignment order, not a concealment mechanism. Guide D1 rule: small academic trials must describe their method or the answer is NI. Item 6 (v1.2) was considered and NOT applied: the parent trial is identified by full citation (Santilli, Diabetes Care 2017, ref 17), but the presumption in item 6 is the large-industry-registration-trial presumption, and this is a single-centre, investigator-initiated, ministry-funded academic trial (62 randomized, PRIN grant 2010JS3PMZ, no registration number given), so the presumption does not reach it; concealment is left NI rather than presumed. Two baseline imbalances at p < 0.05 in 35 analysed participants (waist, VAT) are consistent with chance in a small trial but are not reassuring, and the authors themselves adjusted the main analysis for them. Not High: sequence generation is sound and the imbalances are modest and adjusted.
The computer-generated random allocation sequence was prepared by the trial statistician in blocks of four participants. The subjects were assigned a consecutive random number based on the order of inclusion in the study, and were then allocated to one of the two treatment groups.
At baseline, the only significant between-arm differences were higher waist ( p = 0.021) and lower VAT ( p = 0.027) in the liraglutide arm than in the lifestyle arm ( Table 1 ), thereby the main analysis was adjusted for these baseline variables.
D2 · Deviations from intended interventions
High risk of bias		This is where the withdraw-and-replace design sits, and it is the central threat to the result. The trial is open-label (daily injection vs nutritionist counselling; no blinding of participants, carers or investigators is claimed anywhere), but per the guide, awareness alone rarely biases a lab biomarker, so D2 turns on discontinuation handling and ITT. Both fail. The protocol made non-attainment of the -7% target within 15 months a withdrawal criterion, withdrawn participants were REPLACED to keep the planned n, and the analysed set is 35 completers out of 62 randomized. That is a deviation from intended intervention arising from the trial context (failure to respond, intolerance) handled by removal-and-substitution, and no appropriate analysis was used to estimate the effect of assignment - no ITT set, no inverse-probability weighting, no sensitivity analysis. Selecting each arm down to its successes is what MANUFACTURES the weight match: the two arms are equal on weight because non-achievers were deleted, and the participants deleted are those in whom drug-specific effects would most plausibly differ from diet-specific effects. The arms are consequently no longer randomized groups. The direction of the resulting bias is toward the null for a drug-versus-diet contrast, which is the direction of the reported result.
Patients not achieving the weight loss goal within 15 months of the initiation of the randomized treatment, as well as those not completing the study for decision of the patient and/or of the investigator, were considered withdrawn from the study, and were replaced, in order to attain the anticipated sample size.
Sixty-two patients were randomized to one of the two treatment arms—31 were allocated to liraglutide and 31 to lifestyle changes—and monitored until achievement of the weight loss goal.
D3 · Missing outcome data
High risk of bias		Both prongs of the guide D3 threshold rule trip, so under calibration ruling 2 this is High. 35 of 62 randomized contribute the outcome: 43.5% missing, far above the 20% prong. The arm-wise split of the 27 losses is never reported - only the analysed arms (18 vs 17) are given - so the 5-point prong cannot be shown not to trip and is NI in the conservative direction. Missingness is value-dependent by construction for an O4 result: 10 participants are missing exactly because they did not reach the weight change that the outcome comparison conditions on, and weight change and inflammatory change are correlated; a further 5 were excluded post hoc for antiplatelet use (3 from baseline, 2 who started aspirin in-trial - the latter a post-randomization, plausibly outcome-related exclusion). No sensitivity analysis addresses value-dependent missingness. The only reassurance offered is baseline-only, unquantified and not shown ("data not shown"), and baseline comparability does not speak to outcome-dependent dropout. This overlaps D2 but is not a duplicate: D2 is that the analysis population was constructed by design, D3 is that the outcome is absent for nearly half the randomized sample and its absence depends on the outcome.
Thus, urinary eicosanoid excretion was evaluated on 35 patients ( Table 1 ).
Five additional patients were excluded from the analysis of the present study because of ongoing antiplatelet treatment (3 since baseline, 2 initiated low-dose aspirin during the study), which is known to influence the urinary levels of 11-dehydro-TXB 2 [ 2 , 3 , 4 ].
D4 · Measurement of the outcome
Low risk of bias		hs-CRP is an objective serum biomarker, measured by a single high-sensitivity immunoassay, identically in both arms, on samples collected by an identical protocol (12-h fast, out-patient, frozen at -20 C). No assay change is reported. Per the guide, central/objective inflammatory assays are Low unless the assay changed or measurement differed by arm; neither applies. The open-label design means assessors could know assignment, but for a machine-read immunoassay that does not plausibly move the value. The 36-hour liraglutide withholding was considered here and placed outside D4: it does not degrade the measurement, it changes what is being measured - the drug's marker effect is assessed at a sub-steady-state plasma concentration. That is an interpretation/limitation issue about the null, and it is recorded in limitations, not as measurement bias.
Serum hs-CRP concentrations were measured using highly sensitive immunoassay.
At admission to the study and after the achievement of the weight loss goal, venous blood samples were collected and frozen at −20 °C for subsequent biochemical measurements. All subjects were studied as out-patients after a 12-h fast
D5 · Selection of the reported result
High risk of bias		Item 7 applies directly and its first bullet is satisfied by an unambiguous author statement: the paper opens its statistical section by declaring itself a post hoc analysis, and states that even its power calculation was performed post hoc. That is affirmative evidence of non-prespecification, not merely an unretrievable protocol, so D5 is High rather than the Some-concerns floor. No protocol, SAP or registry entry was retrievable to check any prespecification claim, and the single phrase in the paper that mentions prespecification ("the two pre-specified intergroup contrasts") attaches to the arm contrasts of the parent design, not to hs-CRP as an outcome of this sub-study; under item 7 bullet 3 a paper's own claim is not verification in any case. Two further features compound it. (a) The assessed result carries no estimate at all: hs-CRP between arms is reported as the two characters "ns" inside a sentence whose subject is 11-dehydro-TXB2, with the reader sent to the parent paper - no least-squares mean difference, no CI, no p-value, despite the mixed model being fully specified in the Methods and having produced exactly those quantities for the primary marker. That is selective reporting of the result at the most granular level: the numeric result exists in the model output and was not reported. (b) The paper declares multiple marker outcomes (hs-CRP, TNF-alpha, leptin, IGF-I) as "secondary outcomes" of a post hoc sub-study with no multiplicity adjustment mentioned anywhere.
This is a post hoc analysis of a published study [ 17 ]. The sample size calculation for the present sub-study has been performed as a post-hoc power assessment.
In the present analysis of the study, the outcome of interest was the change in urinary 11-dehydro-TxB 2 (U-11-dehydro-TXB 2 ) and 8-iso-PGF 2α after achievement of 7% of initial weight loss. Secondary outcomes were changes in metabolic, biochemical and imaging parameters in study.
02Funding

Funding and conflicts

Funding
Ministero dell'Istruzione, dell'Università e della Ricerca
Industry funded
Unclear
Manufacturer
Novo Nordisk, Eli Lilly, Lilly, AstraZeneca, Sanofi
Sponsor role
not reported in abstract
Author conflicts
Consoli received lecture fees and fees for serving on advisory boards from Novo Nordisk, Eli Lilly, AstraZeneca, Sanofi Aventis, Merck Sharp & Dohme, and Takeda and grant support to his institution from Novo Nordisk. No other potential conflicts of interest relevant to this article were reported.
Independent replication
unknown
Notes
Authors declare relationships with the drug's manufacturer: Novo Nordisk

Funding is shown on every study and never used to score it.

03Claims

Claims this study bears on

04Source

The source, as retrieved

Abstract

Thromboxane (TX)-dependent platelet activation and lipid peroxidation, as reflected in vivo by the urinary excretion of 11-dehydro-TXB₂ and 8-iso-prostaglandin (PG)F2α, play a key role in atherothrombosis in obesity and type 2 diabetes mellitus (T2DM) since the earlier stages. Thirty-five metformin-treated obese subjects with prediabetes or newly-diagnosed T2DM were randomized to the glucagon-like peptide receptor agonist (GLP-RA) liraglutide (1.8 mg/day) or lifestyle counseling until achieving a comparable weight loss (-7% of initial body weight), to assess whether changes in subcutaneous (SAT) and visceral (VAT) adipose tissue distribution (MRI), insulin sensitivity (Matsuda Index) and beta-cell performance (multiple sampling OGTT beta-index), with either intervention, might affect TX-dependent platelet activation, lipid peroxidation and inflammation. At baseline, Ln-8-iso-PGF2α (Beta = 0.31, p = 0.0088), glycosylated hemoglobin (HbA1c) (Beta = 2.64, p = 0.0011) Ln-TNF-α (Beta = 0.58, p = 0.0075) and SAT (Beta = 0.14, p = 0.044) were significant independent predictors of 11-dehydro-TXB₂. After achievement of the weight loss target, a comparable reduction in U-11-dehydro-TXB₂ (between-group p = 0.679) and 8-iso-PGF-2α (p = 0.985) was observed in both arms in parallel with a comparable improvement in glycemic control, insulin sensitivity, SAT, high-sensitivity C-reactive protein (hs-CRP). In obese patients with initial impairment of glucose metabolism, the extent of platelet activation is related to systemic inflammation, isoprostane formation and degree of glycemic control and abdominal SAT. Successful weight loss, achieved with either lifestyle changes or an incretin-based therapy, is associated with a significant reduction in lipid peroxidation and platelet activation.

Where this record came from

SourceRetrievedIdentifier
pubmedSep 13, 202630513818
first ingestion