Exenatide affects circulating cardiovascular risk biomarkers independently of changes in body composition
- Design
- Randomized trial · 69 participants · Mixed outcome
- Match to healthy normal-weight adults aged 55–75
- Different populationINDIRECT: all 69 participants had type 2 diabetes treated with metformin; no age or BMI data are given in the cached text.
- Could weight loss explain it?
- Specifically testedBiomarker differences are reported as body weight change-adjusted least-squares means and stayed significant after further adjustment for total body fat mass change; the authors note a significant univariate correlation between weight change and hs-CRP and limited power. HbA1c was not adjusted for.
Adjusted for: body-weight change, other - Study tier
- Study tier 3LOW: small randomised trial (69 participants) against insulin glargine with body-composition and biomarker outcomes; blinding not stated.
- Assessment
- Version 4 · Two reviewers differed; settled by a recorded ruling · Oct 19, 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.
In 69 adults with type 2 diabetes on metformin, this randomised trial compared one year of exenatide with insulin glargine. Exenatide reduced body weight by 6% and trunk fat by 13%, and lowered the inflammation marker hs-CRP by 61%; the difference from glargine held after allowing for weight change. IL-6, MCP-1 and resistin did not change with either drug, and endothelin-1 fell with glargine but not with exenatide. These are blood markers in a small trial, not heart events.
01Findings
What the study reported
- Drugs
- Exenatide
- Treatment duration
- 1 year
- Comparator
- insulin glargine (added to metformin)
- Primary outcome
- Body composition (DEXA) and circulating cardiovascular risk biomarkers after 1 year
- Effect
- Within the exenatide arm over 1 year: body weight −6%, trunk fat −13%, total adiponectin +12%, hs-CRP −61%; weight-adjusted between-group hs-CRP −52% (P = 0.008); no significant effect on HMW adiponectin, IL-6, MCP-1 or resistin; endothelin-1 fell with insulin glargine but not with exenatide
- 95% confidence interval
- hs-CRP between-group difference −52% (95% CI −71% to −19%)
- P value
- P = 0.008 (hs-CRP between-group, weight-adjusted)
- Follow-up
- 1 year
- Adverse events
- not reported
- Limitations
- Small trial (69 participants) that the authors say may limit statistical power; secondary report of a trial whose design and baseline data are published elsewhere; blinding not stated.
Who was studied
- Diabetes status
- type 2 diabetes required
- Baseline condition
- metformin-treated type 2 diabetes
- Sample size
- 69
- Inclusion exclusion
- metformin-treated type 2 diabetes
Study quality details
- Study design
- Randomized controlled trial
- Sample size
- 69
- Randomization
- yes
- Blinding
- not stated
- Comparator
- insulin glargine (added to metformin)
- Follow up duration
- 1 year
- Outcome type
- mixed
- Replication
- not assessed
- Consistency with other evidence
- not assessed
- Population applicability
- INDIRECT
- Statistical precision
- 95% CIs reported for percentage changes; 36 exenatide and 33 insulin glargine randomised; authors note limited statistical power
- Funding conflicts
- yes
- 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.
Risk of bias is not the same as whether the numbers work. A trial can be well conducted and still print a standard deviation that is arithmetically impossible, or a table whose sign contradicts its own text. Risk of bias has no question for that, so where it was found it is recorded separately below, under Source integrity.
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 percentage change from baseline, exenatide vs insulin glargine (between-treatment group difference) · 1 year (52 weeks)
High risk of bias · RoB 2 · Two reviewers agreed on every domain
How the overall was reached: Worst-domain rule. D5 is High under rob2-guide v1.6 item 11(a): the trial's registry record enumerates its outcomes and names no inflammatory marker, which is affirmative evidence of non-prespecification rather than an absence of evidence about it. No override. (Previously Some concerns at D5 and Some concerns overall, when the registry had not been read for this question.)
| Domain | Judgment and reasoning |
|---|---|
| D1 · Randomisation process Some concerns about risk of bias | As O4 D1: randomization sequence and allocation concealment not described in this report, small (N=69) trial so the v1.2 large-trial presumption does not reach it, appendix baseline table not retrieved, and baseline hs-CRP is higher in the exenatide arm. NI on both mechanism questions, no contrary evidence -> Some concerns.Details on study design were reported previously ( 5 ). Patients were randomized to exenatide ( n = 36) or insulin glargine ( n = 33) added to their ongoing metformin therapy hs-CRP (mg/l) Insulin glargine 29 1.42 ± 0.27 ... Exenatide 30 1.81 ± 0.25 |
| D2 · Deviations from intended interventions Some concerns about risk of bias | As O4 D2: no blinding stated and the two regimens are self-evidently distinguishable; the biomarker itself is insensitive to awareness, but the analysis is completers-only with no ITT estimate of the assignment effect and no information on co-interventions.All outcome measures are compared between the two treatment groups using an ANCOVA model including factors for treatment, investigative site, and baseline A1C stratum (≤8.5% or >8.5%), and baseline values of corresponding outcome measure as a covariate |
| D3 · Missing outcome data Low risk of bias | hs-CRP available for 30/36 (16.7% missing) and 29/33 (12.1% missing); neither the 20% nor the 5-point prong trips, so Low by ruling 2. Reasons for missingness and any sensitivity analysis are absent - a limitations note.hs-CRP (mg/l) Insulin glargine 29 1.42 ± 0.27 1.38 ± 0.35 −20% (−50% to +27%) Exenatide 30 1.81 ± 0.25 1.30 ± 0.22 −61% (−74% to −42%) |
| D4 · Measurement of the outcome Low risk of bias | Objective central-lab ELISA, all samples in a single batch in one laboratory; no assay change and no between-arm difference in measurement.All serum samples were analyzed in the Lundberg Laboratory for Diabetes Research using a single batch. |
| D5 · Selection of the reported result High risk of bias | High, under rob2-guide v1.6 item 11(a). The assessed result is the 52-week hs-CRP between-treatment group difference of −52% (95% CI −71% to −19%), P = 0.008, in Table 1. NCT00097500 was retrieved and it enumerates the trial's outcomes: one primary ('Beta-cell Function After 52 Weeks of Therapy') and eight secondary (beta-cell function at week 56, first- and second-phase C-peptide release, HbA1c, fasting plasma glucose, seven-point SMBG, body weight, and M-value). hs-CRP is not in that list, and the strings 'CRP' and 'C-reactive' do not occur anywhere in the record. The paper nonetheless reports it: 'Exenatide 30 1.81 ± 0.25 1.30 ± 0.22 −61% (−74% to −42%) −52% (−71% to −19%) 0.008'. A source that set out what this study's outcomes were, and omits the marker being assessed, is affirmative evidence of non-prespecification in the same way a paper's own outcome list is. Item 9's three limits, as item 11 carries them: (1) this is not a stub - nine outcomes, each with a described derivation and time frame, with results posted for all nine plus participant flow, baseline characteristics and adverse events; (2) nothing fuller or later overrides it - no protocol or SAP is in the cache, the paper states no outcome list and claims no prespecification ('Herein we present additional data...'), and the registry record was last updated in 2015, five years after the paper, still enumerating no biomarker, so it is both the fuller and the later source; (3) both halves are quoted above. Independently of item 11, 5.3 is PY: hs-CRP is reported in three framings (unadjusted −61%, weight-adjusted −52%, fat-mass-adjusted −48%) with the largest carried into the abstract, which the guide's D5 domain rule already treats as at least Some concerns. Recorded countervailing evidence, which does not lift the judgment: the registry's official title names 'Cardiovascular Risk Markers' as a study objective, so the trial was not blind to this territory. A title names a domain; it does not enumerate a marker, a timepoint or an analysis, and the nine outcomes that are enumerated contain no marker of any kind. The result is High, not Some concerns, because the evidence here is affirmative absence from a real enumeration, not a protocol that merely could not be retrieved (item 7's floor case).Beta-cell Function After 52 Weeks of Therapy ... Beta-cell Function 4 Weeks After Cessation of Therapy ... Change in First Phase C-peptide Release ... Change in Second Phase C-peptide Release ... Change in Glycosylated Hemoglobin (HbA1c) ... Change in Fasting Plasma Glucose ... Seven Point Self Monitored Blood Glucose (SMBG) Measurements ... Change in Body Weight ... M-value at Baseline, Week 52 and Week 56 All serum samples were analyzed in the Lundberg Laboratory for Diabetes Research using a single batch. Total adiponectin, high molecular weight (HMW) adiponectin, resistin, leptin, high-sensitive C-reactive protein (hs-CRP), interleukin (IL)-6, monocyte chemotactic protein (MCP)-1, and endothelin-1 were determined by commercial ELISAs (R&D Systems, Abingdon, U.K.). |
IL-6 percentage change from baseline, exenatide vs insulin glargine (between-treatment group difference) · 1 year (52 weeks)
High risk of bias · RoB 2 · Two reviewers agreed on every domain
How the overall was reached: Worst-domain rule. D5 is High under rob2-guide v1.6 item 11(a): the trial's registry record enumerates its outcomes and names no inflammatory marker, which is affirmative evidence of non-prespecification rather than an absence of evidence about it. No override. (Previously Some concerns at D5 and Some concerns overall, when the registry had not been read for this question.)
| Domain | Judgment and reasoning |
|---|---|
| D1 · Randomisation process Some concerns about risk of bias | As O1/O4 D1. Baseline IL-6 is well balanced (2.11 vs 1.96 pg/ml), but the randomization mechanism is still undescribed in a small trial whose parent report and registry entry were not retrieved -> NI -> Some concerns.IL-6 (pg/ml) Insulin glargine 29 1.96 ± 0.21 ... Exenatide 30 2.11 ± 0.22 Details on study design were reported previously ( 5 ). Patients were randomized to exenatide ( n = 36) or insulin glargine ( n = 33) |
| D2 · Deviations from intended interventions Some concerns about risk of bias | As O1 D2: unblinded regimens, completers-only analysis with no ITT estimate, no information on co-interventions or adherence.All outcome measures are compared between the two treatment groups using an ANCOVA model including factors for treatment, investigative site, and baseline A1C stratum (≤8.5% or >8.5%), and baseline values of corresponding outcome measure as a covariate |
| D3 · Missing outcome data Low risk of bias | Same denominators as hs-CRP: 30/36 and 29/33; neither D3 prong trips -> Low by ruling 2.IL-6 (pg/ml) Insulin glargine 29 1.96 ± 0.21 2.17 ± 0.20 −4% (−26% to +25%) Exenatide 30 2.11 ± 0.22 2.10 ± 0.25 −10% (−28% to +14%) |
| D4 · Measurement of the outcome Low risk of bias | Objective central-lab ELISA, single batch, one laboratory.All serum samples were analyzed in the Lundberg Laboratory for Diabetes Research using a single batch. |
| D5 · Selection of the reported result High risk of bias | High, under rob2-guide v1.6 item 11(a), on the same enumeration and the same three limits as O1. The assessed result is the 52-week IL-6 between-treatment group difference of −6% (95% CI −30% to +26%), P = 0.670, in Table 1. IL-6 is absent from NCT00097500's nine enumerated outcomes and the strings 'IL-6' and 'interleukin' do not occur anywhere in the record, while the paper reports the marker: 'Exenatide 30 2.11 ± 0.22 2.10 ± 0.25 −10% (−28% to +14%) −6% (−30% to +26%) 0.670'. The registry is not a stub and nothing fuller or later overrides it. Two things are worth recording without changing the judgment. First, the direction of concern that D5 usually carries is weak here: the result is null, the paper reports it in Table 1 and again in the narrative ('No statistically significant effect of either treatment on HMW adiponectin, IL-6, MCP-1, and resistin was observed'), and the whole ELISA panel named in Methods is reported, so no marker from that panel was measured and then buried - 5.2 and 5.3 are both PN. Second, item 11(a) is nevertheless categorical and does not make its finding conditional on the reported result's direction: what it establishes is that the outcome and its analysis were not prespecified in the best source available, and that is as true of the null IL-6 result as of the significant hs-CRP one. The reader of an unprespecified panel cannot know how many markers were assayed and not shown; the eight that are shown are the paper's own account, not an external check. High.Beta-cell Function After 52 Weeks of Therapy ... Beta-cell Function 4 Weeks After Cessation of Therapy ... Change in First Phase C-peptide Release ... Change in Second Phase C-peptide Release ... Change in Glycosylated Hemoglobin (HbA1c) ... Change in Fasting Plasma Glucose ... Seven Point Self Monitored Blood Glucose (SMBG) Measurements ... Change in Body Weight ... M-value at Baseline, Week 52 and Week 56 All serum samples were analyzed in the Lundberg Laboratory for Diabetes Research using a single batch. Total adiponectin, high molecular weight (HMW) adiponectin, resistin, leptin, high-sensitive C-reactive protein (hs-CRP), interleukin (IL)-6, monocyte chemotactic protein (MCP)-1, and endothelin-1 were determined by commercial ELISAs (R&D Systems, Abingdon, U.K.). |
hs-CRP percentage change from baseline, adjusted for body weight change, exenatide vs insulin glargine · 1 year (52 weeks)
High risk of bias · RoB 2 · Two reviewers agreed on every domain
How the overall was reached: D5 is High, which sets the floor at High under the RoB 2 algorithm. Two further reasons are recorded in writing rather than folded into a domain. First, the assessed contrast conditions on body weight change, a post-randomization variable, so it is not protected by randomization: the adjusted comparison is a derived, effectively non-randomized contrast with all the confounding exposure that implies (per v1.1 ruling 4 this is carried here and at D5, not at D1). Second, the paper's central weight-independence claim is in tension with its own data: hs-CRP change correlates strongly with weight change within the exenatide arm (r = -0.590, P = 0.001), and the authors themselves concede the study may be underpowered to separate the two, while the abstract states the changes were "statistically independent of the change in total body fat mass and body weight". With no unadjusted estimate published, the size of the weight-independent component cannot be bounded from this paper.
| Domain | Judgment and reasoning |
|---|---|
| D1 · Randomisation process Some concerns about risk of bias | No randomization sequence generation or allocation concealment method is described in this report; the parent report (ref 5) and the registry entry were not retrieved. The v1.2 ruling-6 presumption is NOT applied: it is written for large industry registration trials with central IVRS/IWRS randomization, and the parent here is a 69-patient three-site trial - the guide's D1 rule requires small trials to describe their method, or the answer is NI. Baseline hs-CRP is also somewhat higher in the exenatide arm (1.81 vs 1.42 mg/l geometric means), which does not on its own indicate a broken sequence at this sample size but is not reassuring, and the baseline-characteristics table sits in an appendix that was not retrieved. NI on both mechanism questions with no contrary evidence gives Some concerns, not High. Per ruling 4, the derived / post-randomization-adjusted character of this contrast is deliberately NOT written into D1.Metformin-treated patients with type 2 diabetes (N = 69) were randomized to exenatide or insulin glargine and treated for 1 year. Details on study design were reported previously ( 5 ). Patients were randomized to exenatide ( n = 36) or insulin glargine ( n = 33) added to their ongoing metformin therapy |
| D2 · Deviations from intended interventions Some concerns about risk of bias | No blinding is claimed anywhere; comparing subcutaneous exenatide twice daily with insulin glargine titration makes participants and clinicians aware of assignment, and the parent trial was open-label in design. For a central-lab biomarker this awareness rarely biases the measurement (guide, D2 rule), so the judgment rests on discontinuation handling: the analysis is completers-only (30/36 and 29/33), there is no ITT analysis of the assignment effect, and no statement about co-interventions or adherence. Loss is modest and roughly balanced, so Some concerns rather than High. No information on deviations from intended intervention is reported.All outcome measures are compared between the two treatment groups using an ANCOVA model including factors for treatment, investigative site, and baseline A1C stratum (≤8.5% or >8.5%), and baseline values of corresponding outcome measure as a covariate hs-CRP (mg/l) Insulin glargine 29 ... Exenatide 30 |
| D3 · Missing outcome data Low risk of bias | 59 of 69 randomized participants (85.5%) contribute hs-CRP data: 30/36 exenatide (16.7% missing) and 29/33 glargine (12.1% missing). Neither prong of the guide's D3 threshold trips - missingness is under 20% overall and the between-arm difference is 4.6 points, under 5 - so by ruling 2 (thresholds are decisive) D3 is Low absent specific contrary evidence, and there is none in the retrieved sources. Recorded against that: the paper gives no reasons for missingness and performs no sensitivity analysis, and the disposition figure (supplemental Fig. 1) was not retrieved, so the balance is inferred from the Table 1 n column rather than read from a flow diagram. That is a limitations note, not a downgrade.Patients were randomized to exenatide ( n = 36) or insulin glargine ( n = 33) added to their ongoing metformin therapy (baseline characteristics and patient disposition are shown in supplemental Fig. 1 in the online appendix hs-CRP (mg/l) Insulin glargine 29 1.42 ± 0.27 1.38 ± 0.35 −20% (−50% to +27%) Exenatide 30 1.81 ± 0.25 1.30 ± 0.22 −61% (−74% to −42%) |
| D4 · Measurement of the outcome Low risk of bias | hs-CRP is an objective central-laboratory measure. All samples were frozen at -80 C and run in a single batch in one laboratory with a commercial ELISA, so the assay did not change partway through and measurement did not differ between arms. Outcome assessors' awareness of assignment cannot influence an automated immunoassay result. Guide D4 rule gives Low.All serum samples were analyzed in the Lundberg Laboratory for Diabetes Research using a single batch. high-sensitive C-reactive protein (hs-CRP), interleukin (IL)-6, monocyte chemotactic protein (MCP)-1, and endothelin-1 were determined by commercial ELISAs (R&D Systems, Abingdon, U.K.). |
| D5 · Selection of the reported result High risk of bias | Prespecification is unverifiable (no protocol, SAP or registry entry retrieved), which alone would leave the Some concerns floor under v1.3 ruling 7 - the paper makes no explicit "exploratory" or "post hoc" statement, so the affirmative- evidence branch of ruling 7 is not triggered. High comes instead from the selection-from-multiple-analyses limb of D5, on evidence inside the paper: (i) two different multivariate adjustment sets for the same contrast are reported for hs-CRP (weight-change-adjusted -52%, fat-mass-change-adjusted -48%), plus log-transformed and percentage-change framings, which is the guide's "multiple CRP transformations with one chosen for emphasis" situation in a sharper form; (ii) the unadjusted between-group hs-CRP contrast - the natural reference against which any weight-adjusted estimate is read - is not reported at all, in a table whose every cell is adjusted, so the reader cannot see the analysis the adjusted one was selected over; (iii) nine biomarkers are tested with no multiplicity adjustment and no statement of a primary biomarker endpoint, and hs-CRP is one of four with P < 0.05. Together these are affirmative signs of selection from multiple eligible analyses of the assessed result, not merely an unretrievable protocol.Data are means ± SEM (body composition measures) or geometric means ± SEM (cardiovascular biomarkers) and body weight change–adjusted least-squares mean percentage change (95% CI) from baseline. LS, least-squares. The crude between–treatment group differences remained statistically significant after additional multivariate adjustment for total body fat mass change: total adiponectin +16% (95% CI: +5% to +28%), P = 0.004; leptin −20% (−34% to −2%), P = 0.028; hs-CRP −48% (−69% to −13%), P = 0.015; and body weight change ( Table 1 ): ... hs-CRP −52% (−71% to −19%), P = 0.008. |
02Funding
Funding and conflicts
- Funding
- Amylin Pharmaceuticals, Inc., and Eli Lilly and Company
- Industry funded
- Yes
- Manufacturer
- Amylin Pharmaceuticals, Inc., and Eli Lilly and Company (named as sponsors)
- Sponsor role
- investigators initiated and designed the study, had full data access and controlled the statistical analysis and interpretation
- Author conflicts
- Two authors are Eli Lilly employees (one a stockholder), several are speakers, advisers or consultants for Lilly or Amylin, and the authors' institutions received research grants from both companies.
- Independent replication
- unknown
- Notes
- Authors declare relationships with the drug's manufacturer: Eli Lilly and Company, Amylin Pharmaceuticals
Funding is shown on every study and never used to score it.
03Claims
Claims this study bears on
- GLP-1 receptor agonist treatment reduces systemic inflammatory markers beyond what can be explained by treatment-associated weight loss.Supports
In 69 metformin-treated adults with type 2 diabetes, exenatide vs insulin glargine for 1 year lowered hs-CRP by -52% (95% CI -71% to -19%) after adjustment for body weight change, but IL-6 and MCP-1 were not significantly changed. Active comparator; only within-arm unadjusted changes are printed.
04Source
The source, as retrieved
Abstract
[OBJECTIVE] To study the effect of exenatide on body composition and circulating cardiovascular risk biomarkers. [RESEARCH DESIGN AND METHODS] Metformin-treated patients with type 2 diabetes (N = 69) were randomized to exenatide or insulin glargine and treated for 1 year. Body composition was evaluated by dual-energy X-ray absorptiometry. Additionally, body weight, waist circumference, and cardiovascular biomarkers were measured. [RESULTS] Treatment with exenatide for 1 year significantly reduced body weight, waist circumference, and total body and trunkal fat mass by 6, 5, 11, and 13%, respectively. In addition, exenatide increased total adiponectin by 12% and reduced high-sensitivity C-reactive protein by 61%. Insulin glargine significantly reduced endothelin-1 by 7%. These changes were statistically independent of the change in total body fat mass and body weight. [CONCLUSIONS] Exenatide treatment for 1 year reduced body fat mass and improved the profile of circulating biomarkers of cardiovascular risk. No significant changes were seen with insulin glargine except a trend for reduced endothelin-1 levels.