Liraglutide and Recurrent Stroke by Baseline Insulin Resistance: A Post Hoc Analysis of the LAMP Trial

Abstract

BACKGROUND:

Glucagon-like peptide-1 receptor agonists reduce major adverse cardiovascular events in type 2 diabetes. Although body mass index does not seem to modify these effects, whether insulin resistance influences treatment efficacy remains unclear.

METHODS:

This post hoc analysis of the LAMP trial (Liraglutide in Acute Minor Ischemic Stroke or High-Risk Transient Ischemic Attack Patients With Type 2 Diabetes Mellitus; a multicenter, open-label, randomized controlled trial conducted at 27 hospitals in China between June 25, 2019, and December 27, 2023) included patients with minor ischemic stroke or high-risk transient ischemic attack and type 2 diabetes. Participants were randomized (1:1) to liraglutide plus standard therapy or standard therapy alone. IR was assessed using the homeostasis model assessment of IR, with a cutoff of 2.5 based on prior studies in Asian populations. Treatment-by-IR interactions were evaluated using Cox models. Absolute risk reduction was calculated as the difference in event rates between groups and was based on crude estimates.

RESULTS:

Among 636 enrolled patients, 510 were included in this analysis (mean age, 65 years; 64.7% male; follow-up, 3 months). A significant interaction between treatment and insulin resistance was observed for both stroke recurrence and composite vascular events (P for interaction=0.02 for both). Among patients with homeostasis model assessment of IR ≥2.5, liraglutide reduced stroke recurrence (5.8% versus 18.1%; absolute risk reduction, 12.3% [95% CI, 5.6%–19.0%]; number needed to treat=8) and vascular events (5.8% versus 19.2%; absolute risk reduction, 13.4% [95% CI, 6.6%–20.2%]; number needed to treat=8). No significant benefit was observed in those with homeostasis model assessment of IR <2.5.

CONCLUSIONS:

IR may be an important determinant of the therapeutic efficacy of liraglutide in patients with acute minor ischemic stroke or high-risk transient ischemic attack and type 2 diabetes. IR-based stratification may help optimize the use of glucagon-like peptide-1 receptor agonists in secondary stroke prevention and guide personalized vascular risk management.

REGISTRATION:

URL: https://www.clinicaltrials.gov; Unique identifier: NCT03948347.

Liraglutide is a glucagon-like peptide-1 receptor agonist (GLP-1RA).¹ In the LAMP trial (Liraglutide in Acute Minor Ischemic Stroke or High-Risk Transient Ischemic Attack Patients With Type 2 Diabetes Mellitus), liraglutide initiation within 24 hours of symptom onset, in combination with standard therapy, reduced the 90-day risk of stroke recurrence and improved neurological functional outcomes in patients with type 2 diabetes (T2D) who presented with either acute minor ischemic stroke or high-risk transient ischemic attack (TIA).²

GLP-1RAs demonstrate reductions in major adverse cardiovascular events—defined as cardiovascular death, nonfatal myocardial infarction, or nonfatal stroke—among patients with T2D at high cardiovascular risk.^3–6 However, weight loss does not seem to be the primary mechanism for cardiovascular protection; instead, improvements in insulin resistance (IR) and inflammation may play key roles in vascular protection.^7,8 IR is the central pathophysiological feature of T2D, contributing to the development of cardiovascular disease through chronic low-grade inflammation and endothelial dysfunction, promoting both atherosclerotic changes in large arteries and microvascular injury.^9–11 The IRIS trial demonstrated that improving IR significantly reduced the risk of recurrent stroke and TIA, emphasizing its potential in secondary prevention.¹²

Liraglutide has been shown to improve IR and exert anti-inflammatory effects early in treatment, mechanisms that may underlie its cardiovascular and cerebrovascular protective benefits.^13,14 This study is a post hoc analysis of the LAMP trial, aiming to evaluate the effects of different metabolic states on the efficacy of liraglutide, providing data to support the precision-based therapeutic strategy.

Methods

Data Availability

The data that support the findings of this study are available from the corresponding author on reasonable request. Access to the data may require approval from the trial steering committee and relevant institutional review boards due to ethical and privacy considerations.

Study Design and Patients

This study was reported in accordance with the STROBE guidelines (Strengthening the Reporting of Observational Studies in Epidemiology; Supplemental Material). The overall design and primary results of the original trial have been published previously.² Briefly, LAMP was a multicenter, prospective, randomized, controlled, open-label trial with blinded end point evaluation conducted in China, comparing liraglutide plus standard treatment versus standard treatment alone in patients with minor ischemic stroke or high-risk TIA and concomitant T2D enrolled within 24 hours of symptom onset across 27 centers. The prespecified primary end point was recurrent stroke during follow-up. Prespecified secondary end points included a composite vascular outcome (recurrent stroke, myocardial infarction, and vascular death) and a 90-day functional outcome. Due to loss of funding, the trial was terminated early before reaching the prespecified number of events; the final study visits were completed on March 24, 2024. The LAMP trial was approved by the ethics committees of all participating centers (ethical approval number: KY-2019-057), and written informed consent was obtained from all participants or their legal representatives before enrollment. The current analysis evaluating effect modification by IR was a post hoc analysis and was not prespecified in the original trial protocol or statistical analysis plan. However, all variables required to calculate the homeostasis model assessment of IR (HOMA-IR) were prespecified and prospectively collected as part of the original trial assessments. Accordingly, all analyses were considered exploratory and hypothesis-generating. Given this exploratory framework, no formal adjustment for multiple comparisons was applied, and P values should be interpreted with appropriate caution.

To avoid the short-term influence of exogenous insulin on fasting insulin (FINS) levels and the assessment of HOMA-IR, our study excluded patients who had received insulin therapy 1 week before admission. Patients without baseline fasting plasma glucose (FPG) or FINS values were not included in the primary complete-case analysis. All other eligibility criteria were consistent with the main trial.²

Randomization and Masking

Patients were enrolled by investigators at each participating center and were randomly assigned in a 1:1 ratio to the liraglutide or control groups. Block randomization (block size of 4) was implemented using sealed, opaque, and sequentially numbered envelopes. Outcome assessors were blinded to the treatment allocation, and statisticians conducting the final analyses were also masked to group assignments.

Procedures

Treatment of patients in the liraglutide group was initiated within 1 hour after randomization. Liraglutide was administered subcutaneously once daily, starting at 0.6 mg/d, and titrated to 1.2 mg/d and then to a target dose of 1.8 mg/d at weekly intervals. After the occurrence of adverse effects during dose escalation, the dose increase interval was extended, and maintenance doses <1.8 mg/d were considered. Drug tolerability was determined by investigators based on the clinical condition of each participant; patients were considered tolerant if stable and without significant nausea, vomiting, or hypoglycemia. Patients in both groups were managed according to the 2018 American Heart Association/American Stroke Association Guidelines for the Early Management of Patients with Acute Ischemic Stroke, the 2014 China Secondary Prevention Guidelines for Ischemic Stroke and Transient Ischemic Attack, and the 2017 China Diabetes Prevention Guidelines.^15–17 The overall study procedures have been described previously.² In summary, the study drug was dispensed within 1 hour after randomization, and follow-up visits were scheduled at 7±1, 30±3, and 90±7 days after randomization. Study treatment continued until the final study visit. All end points were evaluated in a blinded manner by independent, trained physicians.

Outcomes

The end point of this analysis was consistent with the main LAMP trial.² Based on the LAMP trial, this study evaluated whether IR modifies the therapeutic efficacy of liraglutide, with additional exploratory analyses examining obesity- and inflammation-related subtypes. Functional outcomes at 90 days were defined as excellent functional outcome (modified Rankin Scale [mRS] score 0–1) and good functional outcome (mRS score 0–2).

Classification Criteria

IR was assessed at baseline using the HOMA-IR, calculated as follows¹⁸:

$${\displaystyle \mathrm{H}\mathrm{O}\mathrm{M}\mathrm{A}\text{-}\mathrm{I}\mathrm{R}=\frac{\mathrm{F}\mathrm{P}\mathrm{G}\left(\mathrm{m}\mathrm{m}\mathrm{o}\mathrm{l}/\mathrm{L}\right)\times \mathrm{F}\mathrm{I}\mathrm{N}\mathrm{S}\left(\mu \mathrm{U}/\mathrm{m}\mathrm{L}\right)}{22.5}}$$

In contrast to the IRIS trial, which defined IR using a HOMA-IR threshold ≥ 3.0,¹² lower cutoff values more appropriate for Asian populations were adopted.^19–21 In our cohort, we adopted 2.5 as the primary cutoff to identify patients at IR and evaluated 2.0 as an alternative threshold in sensitivity analyses. Furthermore, baseline hs-CRP (high-sensitivity C-reactive protein) levels were categorized according to a previous study²²: hs-CRP <2 mg/L was defined as low risk, and hs-CRP ≥ 2 mg/L as high risk. In addition to IR and hs-CRP classification, Body mass index (BMI) was categorized according to the criteria recommended by the Working Group on Obesity in China: BMI <24.0 kg/m² was defined as nonoverweight, 24.0 to 27.9 kg/m² as overweight, and ≥28.0 kg/m² as obesity.²³ BMI was calculated as follows:

$${\displaystyle \mathrm{B}\mathrm{M}\mathrm{I}=\frac{\mathrm{W}\mathrm{e}\mathrm{i}\mathrm{g}\mathrm{h}\mathrm{t}(\mathrm{k}\mathrm{g})}{{[\mathrm{H}\mathrm{e}\mathrm{i}\mathrm{g}\mathrm{h}\mathrm{t}\left(\mathrm{m}\right)]}^2}}$$

Statistical Analyses

Baseline characteristics were compared between treatment groups and across IR strata. Categorical variables were expressed as frequencies (%) and compared using the χ² test or Fisher exact test. Continuous variables were summarized as medians with interquartile ranges and compared using the Kruskal-Wallis test.

During the 90-day follow-up, the incidences of recurrent stroke and vascular events were analyzed using Cox proportional hazards regression models, with hazard ratios (HRs) and 95% CIs reported. The proportional hazards assumption was assessed using Schoenfeld residuals. When multiple events of the same type occurred, only the first event was counted; patients without events were censored at study completion or at the time of nonvascular death. Binary outcomes (mRS score 0–1, mRS score 0–2, hypoglycemia, and gastrointestinal disorders) were analyzed using logistic regression, with ORs and 95% CIs calculated. Given that this substudy included a subset of participants from the main trial, both unadjusted and adjusted models were presented. Adjusted Model 1 included age, sex, BMI, glycated hemoglobin, and random blood glucose, selected based on baseline imbalances. FPG and FINS were not included because they are components of HOMA-IR, to avoid overadjustment and collinearity. Adjusted model 2 further excluded random blood glucose and glycated hemoglobin to minimize potential overadjustment. To assess short-term metabolic and inflammatory responses, between-group differences in HOMA-IR and hs-CRP at 1 week were analyzed using ANCOVA, with treatment group as the main factor and baseline HOMA-IR and hs-CRP as covariates. If the distributions of HOMA-IR or hs-CRP were skewed, log transformation was performed before analysis.

The primary analysis examined whether baseline IR modified the therapeutic effect of liraglutide. Interaction terms (treatment×IR) were included in Cox and logistic regression models to assess differences in treatment effects across IR strata. To further explore potential nonlinear associations between IR and treatment effects, and to reduce the information loss or bias that may arise from categorizing a continuous variable using a fixed cutoff, HOMA-IR was modeled using restricted cubic splines. Exploratory analyses were also conducted for treatment×BMI and treatment×hs-CRP interactions.

The time to clinical benefit for stroke recurrence was estimated by calculating cumulative hazards for successive days after randomization and reestimating daily HRs using Cox models. The clinical benefit onset was defined as the first day on which the liraglutide showed a statistically significant effect versus control (P<0.05) that was subsequently maintained. This analysis was also presented by IR strata.

Sensitivity analyses included (1) reanalysis using HOMA-IR≥2.0 as an alternative threshold; (2) repeating primary and interaction analyses after multiple imputation for missing FPG or FINS values (m=20, predictive mean matching), with imputation models incorporating treatment assignment, age, sex, BMI, baseline hemoglobin A1c (HbA1c), baseline National Institutes of Health Stroke Scale score, follow-up time, and outcome status, and passive recalculation of HOMA-IR, with results combined using Rubin’s rules. All statistical tests were 2-sided, and P<0.05 was considered statistically significant. All analyses were performed using R software (version 4.5.1).

Results

Study Patients

After excluding patients who had received insulin 1 week before symptom onset, 522 patients were enrolled, of whom 510 had complete measurements of FINS and FPG (Figure 1). Compared with those who had used insulin before onset, participants included in our analysis had lower prevalences of prior ischemic stroke, diabetes, and coronary heart disease. In addition, fewer patients received concomitant lipid- or glucose-lowering therapy (Table S1).

Figure 1.

Flow diagram. FINS indicates fasting insulin; FPG, fasting plasma glucose; NIHSS, National Institutes of Health Stroke Scale; and TIA, transient ischemic attack.

Among the 510 eligible participants, 348 (68.2%) were classified as having IR. Baseline characteristics are shown in Table 1, with comparisons between IR strata and treatment groups. In our metabolic substudy, the 90-day incidence of recurrent stroke was 7.6% (19/249) in the liraglutide group versus 15.3% (40/261) in the control group (HR, 0.48 [95% CI, 0.28–0.83]; P=0.008). Among patients without prior insulin use, liraglutide reduced stroke recurrence and improved secondary outcomes, consistent with the main trial findings² (Table S2).

Table 1.

Baseline Characteristics According to IR Status and Treatment Allocation

Clinical Outcomes

Stroke recurrence incidence and secondary outcomes varied according to treatment assignment and IR status (Figure 2). In patients with IR, liraglutide was associated with a lower risk of stroke recurrence compared with standard therapy (5.8% versus 18.1%; HR, 0.30 [95% CI, 0.15–0.62]), corresponding to an absolute risk reduction of 12.3% and a number needed to treat of 8 to prevent one recurrent stroke within 90 days. In contrast, no significant benefit was observed among patients with non-IR (11.5% versus 9.5%; HR, 1.22; [95% CI, 0.47–3.16]; absolute risk difference, +2.0%). The treatment×IR interaction was statistically significant (P=0.02). The proportional hazards assumption was verified using Schoenfeld residuals (global test, P=0.28). Additional subgroup-specific analyses stratified by IR status also showed no evidence of violation of the proportional hazards assumption (P=0.80 for the non-IR subgroup and P=0.36 for the IR subgroup). Similar results were obtained for the composite vascular outcome (Table 2). Hypoglycemia incidence and gastrointestinal adverse events did not differ between treatment groups, regardless of IR status (P=0.22 and P=0.35, respectively). Adjusted model results (Tables S3 and S4) were consistent with the unadjusted models. The cumulative risk curves aligned with Cox regression analyses, showing that most events occurred early during follow-up (Figure 3).

Figure 2.

Liraglutide vs control on clinical outcome stratified by insulin resistance (IR) status. Treatment effects are expressed as hazard ratios with 95% CIs for stroke and vascular events, and as odds ratios with 95% CIs for functional outcomes (modified Rankin Scale [mRS] score >1 and mRS score >2). Vascular events were defined as new clinical vascular events, including ischemic stroke, hemorrhagic stroke, myocardial infarction, or vascular death. The size of the data markers reflects the sample size of each subgroup. Effect size estimates were derived from unadjusted models.

Table 2.

Effect of Liraglutide Group Compared With Control Group Outcome Stratified by IR Status

Figure 3.

Kaplan-Meier curves for recurrent stroke and vascular events according to insulin resistance (IR) status. Cumulative incidence of (A) new stroke and (B) vascular events in patients with type 2 diabetes and minor ischemic stroke (MIS) or transient ischemic attack, stratified by treatment group (liraglutide vs control) and IR level (homeostasis model assessment of IR high vs low). Log-rank tests were used for global comparisons of event-free survival across all 4 groups for descriptive purposes. Numbers at risk at each time point are shown below the plots.

In the restricted cubic splines analysis (Figure S1), HOMA-IR demonstrated a possible nonlinear trend in its association with stroke recurrence risk (P=0.07), and the treatment effect varied across HOMA-IR levels (P for interaction=0.07; P for nonlinear interaction=0.02). No significant difference in HOMA-IR levels was observed between groups at week 1 (geometric mean ratio, 1.08 [95% CI, 0.95–1.22]; P=0.26).

Secondary Analysis

A significant reduction in stroke recurrence risk was observed in the liraglutide group starting from day 12 after randomization (HR, 0.511 [95% CI, 0.263–0.995]; P=0.048; Figure S2), indicating an early onset of clinical benefit. The reduction in risk remained consistent, suggesting a sustained treatment effect. Further time-to-event analyses in the overall population demonstrated that the treatment benefit persisted with longer follow-up at 30 days (HR, 0.448 [95% CI, 0.244–0.822]; P=0.0095), 60 days (HR, 0.518 [95% CI, 0.298–0.901]; P=0.0199), and 90 days (HR, 0.478 [95% CI, 0.277–0.826]; P=0.0082). Overall, liraglutide significantly reduced the risk of stroke recurrence from an early stage, and this benefit was maintained throughout the observation period. IR strata results are provided in Figure 4.

Figure 4.

Time to clinical benefit for stroke recurrence. Time to clinical benefit was assessed post hoc. The x axis represents time from randomization, and the y axis represents the time-varying hazard ratio. A, Treatment effect in patients with insulin resistance (IR), with the vertical dashed line (day 5) indicating the time point at which liraglutide treatment first reached and consistently maintained statistical significance (P<0.05). B, Treatment effect in patients with non-IR, with the vertical dashed line (day 49) marking the time point of the minimum P value for the treatment effect, although statistical significance was not achieved at any time. The blue shaded area represents the 95% CI.

Subgroup analyses were conducted within each treatment group to evaluate the association between IR status and clinical outcomes. In the liraglutide group, stroke recurrence occurred in 5.8% of patients with IR and 11.5% of those with non-IR (HR, 0.50 [95% CI, 0.20–1.23]; P=0.13), whereas the incidence of composite vascular events was 5.8% and 12.8%, respectively (HR, 0.45 [95% CI, 0.19–1.08]; P=0.07). In the control group, stroke recurrence occurred in 18.1% of patients with IR and 9.5% of those with non-IR (HR, 2.00 [95% CI, 0.92–4.34]; P=0.08), and functional disability (mRS score >1) was observed in 30.7% and 13.4%, respectively (OR, 2.86 [95% CI, 1.45–6.08]; P=0.004). Detailed results and adverse event data are presented in Table S5.

Sensitivity Analysis

A sensitivity analysis using an alternative cutoff value of 2.0 for HOMA-IR showed results for the primary and vascular outcomes that were consistent with the main analysis, while the interaction for functional outcomes (mRS score >1) was no longer statistically significant (Table S6). Multiple imputation for missing FINS and FPG yielded similar findings. The primary outcome interaction was significant (P=0.025), further supporting the robustness of the results.

Exploratory Analysis

When BMI was dichotomized at 24 kg/m², liraglutide significantly reduced stroke recurrence in patients with BMI<24 kg/m² (HR, 0.30 [95% CI, 0.12–0.74]; P=0.008), but not in those with BMI≥24 kg/m² (HR, 0.55 [95% CI, 0.28–1.11]; P=0.35). The treatment×BMI interaction was not statistically significant in the categorical analysis (P=0.15); however, a significant nonlinear interaction was identified in the restricted cubic splines model (P=0.006; Figure S3), suggesting that BMI may nonlinearly modify the effect of liraglutide on stroke recurrence.

When stratified by hs-CRP levels, liraglutide significantly reduced the risk of stroke recurrence in patients with low inflammatory status (hs-CRP <2 mg/L; HR, 0.38 [95% CI, 0.16–0.90]; P=0.028), whereas no significant benefit was observed in those with high inflammation (hs-CRP ≥2 mg/L HR, 0.61 [95% CI, 0.30–1.25]; P=0.179). The treatment×hs-CRP interaction was not statistically significant (P=0.40). In the restricted cubic splines model, the overall association between hs-CRP and stroke recurrence was statistically significant (P=0.015), whereas the nonlinear component was not significant (P=0.073). No significant interaction between hs-CRP and liraglutide was observed (P for interaction, 0.15; P for nonlinear interaction, 0.06; Figure S4). No significant difference in hs-CRP levels was observed between groups at week 1 (geometric mean ratio, 1.18 [95% CI, 0.94–1.49]; P=0.15).

Discussion

This post hoc analysis of the LAMP trial suggests that baseline IR may modify the therapeutic efficacy of liraglutide in patients with acute minor ischemic stroke or high-risk TIA and T2D. A greater apparent benefit was observed among patients with marked IR, whereas little or no benefit was seen among those with relatively preserved insulin sensitivity. Notably, ≈31.8% of patients with T2D were classified as having relatively preserved insulin sensitivity based on HOMA-IR, which may be partly explained by the fact that approximately two-thirds of these patients had received glucose-lowering therapy before the index event. In this setting, fasting indices may seem normalized despite persistent systemic metabolic dysfunction, leading to partial misclassification of IR and attenuation of observed treatment-effect heterogeneity.

IR represents a systemic metabolic dysfunction characterized by impaired insulin signaling, ectopic lipid accumulation, chronic low-grade inflammation, and endothelial impairment, which collectively increase vascular vulnerability following ischemic injury.^24–26 GLP-1RAs act on multiple nodes within this pathophysiological network, enhancing insulin sensitivity and glucose utilization, reducing metabolic overload, improving endothelial function, and suppressing inflammatory and oxidative cascades.^27–29 They also decrease visceral and hepatic fat deposition, ameliorate neuroinflammation, and promote cerebrovascular repair by improving cerebral hemodynamics.³⁰ These pleiotropic effects may be particularly relevant in patients with pronounced IR, in whom metabolic and vascular derangements are more severe and potentially more amenable to pharmacological modulation, whereas individuals with relatively preserved insulin sensitivity exhibit less metabolic stress and therefore limited scope for additional therapeutic benefit. However, no significant changes in HOMA-IR or hs-CRP levels were observed during the first week of treatment. This may be attributable to the relatively low initial dose of liraglutide administered during the acute phase and the limited availability of longitudinal biomarker measurements. Despite the absence of short-term changes in these biomarkers, liraglutide significantly reduced the risk of recurrent stroke early after randomization, particularly among patients with IR. These findings suggest that early clinical benefits may not rely solely on measurable metabolic or inflammatory improvements but may also involve rapid, nonmetabolic vascular or neuroprotective mechanisms, such as enhanced endothelial NO bioavailability, improved vasodilation and microvascular perfusion, and suppression of oxidative stress and platelet activation.^31–34 However, these mechanistic interpretations remain speculative and warrant further validation in future studies incorporating more refined biomarkers and dedicated mechanistic investigations.

Obesity is a well-established risk factor for atherosclerotic cardiovascular disease.³⁵ Although GLP-1RAs induce substantial weight loss, emerging evidence suggests that improvements in conventional metabolic parameters (such as BMI and weight loss) cannot fully account for their vascular protective effects.^7,8 Consistent with previous findings, baseline BMI appeared to have little impact on liraglutide efficacy in our cohort.^36,37 Notably, this trial showed a U-shaped association between BMI and stroke recurrence in the control group. At lower BMI levels, recurrence risk was higher, consistent with previous meta-analyses and the obesity paradox, suggesting that underweight status may reflect frailty, malnutrition, or insufficient metabolic reserve.³⁸ However, unlike a previous study showing a continuously decreasing recurrence risk with increasing BMI,³⁸ our findings revealed an increased risk at higher BMI levels. This discrepancy may be attributable to population heterogeneity, as all participants in our cohort had T2D, in whom excessive adiposity may exacerbate metabolic syndrome and further elevate vascular risk. In addition, it should be noted that the so-called obesity paradox may also arise from methodological biases, including selection bias and collider stratification bias.³⁹ Collectively, BMI represents an external anthropometric phenotype, whereas IR directly reflects intrinsic metabolic dysfunction. Compared with BMI, IR may serve as a more accurate indicator of individual metabolic status and differential therapeutic response. Stratification based on IR rather than BMI may offer a more precise framework for identifying high-risk patients and optimizing GLP-1RA–based interventions.

The potential modifying role of inflammation, proxied by baseline hs-CRP, was evaluated. Consistent with previous studies, higher hs-CRP was associated with an increased risk of stroke recurrence.⁴⁰ Previous analyses of semaglutide cardiovascular outcome trials suggested that baseline inflammation does not substantially modify the efficacy of GLP-1RAs.⁴¹ Similarly, no significant interaction was observed between hs-CRP and liraglutide treatment in our study. This null finding may relate to the widespread postevent use of statins in our cohort. The JUPITER trial (Rosuvastatin to Prevent Vascular Events in Men and Women With Elevated C-Reactive Protein) demonstrated that statin therapy reduces major cardiovascular events among individuals with elevated hs-CRP, which could attenuate any detectable differential treatment effect across hs-CRP strata.⁴²

This study has some limitations. First, although the primary end points of the LAMP trial were prespecified, this was a secondary exploration analysis with a limited sample size and number of events, which may have reduced statistical power. Patients who had received insulin within 1 week before the index event were excluded, which may have introduced selection bias and limited generalizability, as these individuals may represent a population with more advanced diabetes or different vascular risk profiles. Second, the open-label design of the trial may have introduced potential surveillance bias. Third, HOMA-IR primarily reflects hepatic IR and may not fully represent systemic insulin sensitivity, particularly in peripheral tissues such as skeletal muscle. Therefore, it may underestimate or misclassify whole-body IR. Fourth, HOMA-IR was derived from fasting glucose and insulin measurements obtained across multiple centers without central laboratory standardization or external validation against alternative indices (eg, triglyceride-glucose index). Intercenter variability in assay performance may have introduced random measurement errors and partial misclassification of IR strata. Fifth, the study population consisted exclusively of Chinese patients, which may limit generalizability to other ethnicities or nondiabetic high-risk populations. Sixth, although multiple end points were analyzed, no correction for multiple comparisons was performed, potentially increasing the risk of type I error. Finally, the absence of standardized HOMA-IR cutoffs introduces uncertainty, although sensitivity analyses supported the robustness of our findings.

Conclusions

IR may influence the vascular efficacy of liraglutide in patients with acute minor ischemic stroke or high-risk TIA and T2D. Individuals with greater IR appear to obtain enhanced vascular protection, highlighting its potential role as a mechanistic biomarker for precision metabolic intervention. However, these findings are hypothesis-generating and should be interpreted with caution. Future prospective and stratified studies are warranted to confirm these observations and to clarify the clinical value of IR-guided therapeutic strategies in secondary stroke prevention.

ARTICLE INFORMATION

Acknowledgments

The authors thank the participants who took part in the LAMP trial (Liraglutide in Acute Minor Ischemic Stroke or High-Risk Transient Ischemic Attack Patients With Type 2 Diabetes Mellitus).

Disclosures

None

Supplemental Material

Tables S1–S6

Figures S1–S4

STROBE Checklist

Funding Statement

National Natural Science Foundation of China (81901200); Science and Technology Program of Guangzhou (202201020070, 2023A03J1023, 2023A03J1022, and 2023A03J0578); Open Research Project of the National Key Laboratory of Bioactive Molecules and Drugability Optimization (SKLBMDA-2025101).