Type 2 diabetes mellitus (T2DM) is recognized as a significant and independent risk factor for coronary heart disease [1, 2]. Vascular events remain the primary cause of both mortality and morbidity in individuals with T2DM [3]. According to the guidelines issued by the American Diabetes Association (ADA) [4], acetylsalicylic acid (ASA) is recommended as a secondary prevention measure for T2DM patients with a prior history of atherosclerotic cardiovascular disease (CVD). ASA may also be contemplated as a primary prevention strategy for diabetic individuals at heightened cardiovascular risk [4]. Despite the recommended use of aspirin therapy in patients with T2DM, the efficacy of aspirin therapy in primary CVD prevention is currently suboptimal [5]. Specifically, as a secondary prevention strategy, the reduction in cardiovascular event risk in T2DM patients following ASA was less than 10%, which contrasts with a greater than 20% decrease observed in non-diabetic patients [6]. The diminished efficacy of aspirin use as an antiplatelet agent is attributed to aspirin resistance (AR), characterized by high platelet reactivity (HPR) and poorly inhibited thromboxane synthesis in vivo despite administering the recommended dose of aspirin. Some studies showed that AR occurred more frequently in patients with diabetes [7, 8]. The clinical implications of this inadequate platelet suppression could be significant, as AR has been associated with an elevated risk of adverse cardiovascular events in individuals with a prior history of myocardial infarction, as well as a more than threefold increase in the risk of adverse primary outcomes in patients with chronic coronary syndromes [9, 10].

An essential mechanism through which aspirin exerts its antiplatelet effect is by inhibiting the cyclooxygenase-1-related (COX-1) pathway [11]. In T2DM patients, prolonged hyperglycemia, resulting from insulin resistance and metabolic disorders, mediates the accumulation of oxidative stress and triggers the damage of endothelium by creating an imbalance between vasodilators and vasoconstrictors [12]. Endothelial dysfunction and an associated chronic low-grade inflammation state accelerate the generation of proinflammatory cytokines, acute phase proteins, adipokines, and chemokines [13]. Accordingly, there is an increase in platelet turnover and a higher count of young, reticulated platelets [13, 14]. Although unacetylated COX-1 and COX-2 from newly-formed platelets are believed to play a pivotal role in AR, a broadly applicable and widely recognized comprehensive mechanism remains elusive. The complicated interactions between platelet activation and the pathogenic events occurring in patients with high platelet reactivity make it challenging to interpret current mechanistic information as clinically formative indicators [15]. If statistically verified by available evidence, such clinical predictors of AR could assist in providing personalized therapies, leading to more favorable clinical outcomes. However, to our knowledge, no available clinical trial study is currently attempting to determine clinical predictors of AR in patients with T2DM. Some observational studies suggested that demographic characteristics, such as age and body mass index (BMI), are potential predictive factors [16, 17]. Conversely, other studies have proposed diverse laboratory parameters, including glycemic levels and serum lipid profiles, as possible markers of AR in T2DM patients [18, 19]. A study of diabetic patients taking ASA daily found that AR is linked to lipid dysfunction and a history of smoking [20]. However, current findings are inconsistent and mainly based on observed evidence. Moreover, there is a shortage of comprehensive reports synthesizing evidence of clinical predictors of AR within patients with T2DM.

Therefore, this study aimed to review the current literature on clinical predictors of AR in diabetic patients, to inform decision-making regarding suitable interventions for preventing diabetes-related complications, and to improve the likelihood of more favorable clinical outcomes.

This systematic review and meta-analysis was conducted according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) 2020 guidelines [21]. The review protocol was registered in PROSPERO (ID: CRD42023388170).

Our systematic review and meta-analysis focused on comparing the clinical characteristics of AR versus non-AR among diabetic patients receiving aspirin treatment. The main findings can be summarized as follows: (1) AR is associated with younger patients, whereas there were no significant differences in gender distribution, BMI, and smoking status between the two groups. (2) Non-AR patients exhibit similar profiles of coexisting conditions and concurrent medications compared to AR patients. (3) Regarding laboratory results examined in this meta-analysis, all lipid control parameters (HDL, LDL, TG, and TC levels) and two diabetic parameters (fasting glucose and HbA1c) demonstrated significant correlations with AR.

Aspirin resistance is a prevalent clinical phenomenon and is empirically defined as a condition where the conventional dose of aspirin fails to exhibit consistent antiplatelet effects [32]. This ambiguous definition leads to inconsistencies in various aspects of research concerning AR. Firstly, there is a lack of standardized aspirin dosage; a daily dose of aspirin, recommended by the American Society for Vascular Surgery, is between 75 and 325 mg as a secondary prevention strategy against adverse cardiovascular complications [33]. In this meta-analysis, patients from the included studies were administered varying doses of aspirin, ranging from 75 to 325 mg daily. Secondly, AR is currently verified by various platelet function tests [34]. Certain assays have been adopted in clinical practice based on considerations such as sensitivity in test results, availability of resources, and simplicity of use [34]. Six AR assays were identified in the included papers. Several heterogeneity tests were performed to assess the potential influence of heterogeneity in combined results caused by aspirin doses and AR detections. The I² test, Cochran’s Q test, and Galbraith test all suggested that the heterogeneity of the included studies was generally low. After sensitivity analysis, two AR tests (STI and VNS) were doubtful, necessitating further investigation. Subgroup analysis by aspirin dose showed that the significant differences in three laboratory parameters (TG, HDL, and fasting glucose) between AR and no-AR groups became non-significant in patients receiving various doses of aspirin (between 75 to 325 mg). Conversely, the meta-regression analysis results concluded that aspirin dose and AR assays were not the primary source of heterogeneity. Roller et al. [35] investigated the impact of aspirin dose on AR and reported that increasing the aspirin dose did not convert aspirin non-responders into responders. Aspirin non-responders were defined as possessing a collagen and epinephrine closure time (CEPI-CT) exceeding 165 s. Their study confirmed AR individuals using the PFA-100 value among participants treated with 100 mg ASA daily for 7 days. Five identified ASA non-responders were re-examined after taking 300 mg ASA daily for three weeks, and none exhibited a transition to aspirin responders. Clinical evidence also suggested that higher doses do not enhance the cardioprotective effects of aspirin [36]. On the other hand, aspirin formulations (plain aspirin or enteric-coated) also had no significant impact on aspirin responsiveness [37, 38]. In summary, our study suggests minimal heterogeneity attributed to aspirin doses. While each laboratory assay for AR has inherent limitations [34], our subgroup analysis and meta-regression results demonstrated that AR assays are not a significant source of heterogeneity. This conclusion is consistent with findings from other meta-analysis reports [39, 40], which combined results focusing on clinical outcomes among AR patients. The argument is that since each individual assay reveals a certain level of rationality and sensitivity, any genuine clinical effects on screened AR individuals should also be observable [39].

Aspirin achieves its primary antithrombotic effect by acetylating the serine-529 residue of COX-1 irreversibly [41]. COX-1 induces the conversion of arachidonic acid into thromboxane A2 (TXA2, which is a potent platelet activator that binds to the TXA2 receptor (TP) expressed on platelet membranes, thereby initiating the TP-mediated platelet aggregation pathway [41, 42]. Briefly, aspirin suppresses platelet activation by blocking the synthesis of COX-1-dependent TXA2. While the inhibition of COX-1 by aspirin is rapid, irreversible, saturable at low doses, and sustained throughout the lifespan of a platelet (7–10 days) [43], the prevalence of AR in patients with T2DM could be up to 60% depending on the measurements used [44]. In our pooled analysis, AR was mainly associated with increased insulin resistance and poor lipid control indicators. The predictors found in this study play crucial roles in the mechanisms of AR in diabetes. Firstly, hyperglycemia and hypercholesterolemia are thought to cause endothelial dysfunction through elevated oxidative stress and impaired nitric oxide (NO) biosynthesis and transportation [12, 45]. Endothelial dysfunction mediates platelet activation and adhesion to endothelial cells, causing rapid platelet generation and heavy platelet consumption [12]. The resulting enhanced platelet turnover leads to the production of immature platelets, which are rich in mRNA and could generate unacetylated COX-1 and COX-2 [11]. Given the short half-life of aspirin, the newly formed COX-1 may not be adequately inhibited. Consequently, activating the COX-1-dependent TXA2 pathway leads to the aspirin treatment failing. Subsequently, the glycated platelet membrane proteins are structurally altered and become less accessible for acetylation, making aspirin less effective [11]. Secondly, COX-2, the second isoform of COX, is typically found in less than 10% of resting platelets; however, its expression is upregulated in inflammatory conditions by accelerated platelet turnover. Notably, COX-2 is not sensitive to low-dose aspirin and induces the production of TXA2 in a COX-1-independent manner [46]. Platelets can also be activated by another COX-1-independent pathway, which entails the oxidation of arachidonic acid and subsequent generation of isoprostanes—aspirin-insensitive agonists that bind the TXA2 receptor and activate platelets [41]. Interestingly, platelet aggregation still occurs in the presence of isoprostanes, even though TXA2 levels are significantly decreased by aspirin [14, 41]. In conclusion, hyperglycemia and dyslipidemia play significant roles in both the COX-1-dependent and COX-1-independent platelet activation pathways.

In addition to the previously discussed results, further observations from the current study are briefly outlined. An additional association was observed between AR and younger patients. This finding could be elucidated from a pharmacokinetic standpoint, as aspirin esterase activity diminishes in older individuals, particularly those with heightened inflammatory conditions [16, 47]. Given that most participants in this study were administered a relatively low aspirin dose (no more than 100 mg per day), the presence of AR in younger patients could potentially be linked to a heightened rate of aspirin hydrolysis. Intriguingly, despite insulin levels and HOMA-IR being considered reliable indicators of insulin resistance, they did not show an association with AR in our analysis. This discrepancy could be partly attributed to the limited number of studies (only three) that investigated insulin levels and HOMA-IR, leading to notable heterogeneity in the pooled results (p 0.01) and thereby casting doubt on the findings. Furthermore, the absence of a robust correlation between AR and coexisting vascular diseases may be rationalized by the shared underlying mechanisms of AR in both diabetes and vascular disorders [48]. Therefore, it is plausible that prior vascular events may not exert a significant impact on the development of AR in diabetic patients, especially considering their existing chronic inflammatory state and elevated oxidative stress levels attributed to inadequate glucose management and hyperlipidemia. Conversely, our findings indicate that BMI, although identified as a determinant of AR in certain studies [18, 29], may not be reliable as a marker of dyslipidemia and hyperglycemia.

Our study has several limitations. First, regarding the AR study, most available randomized clinical trials (RCTs) and meta-analyses focused on the efficacy and safety of aspirin treatment as a first or second prevention strategy for vascular events. Notably, no RCTs specifically investigating the clinical predictors of AR in diabetic patients were identified. As a result, all publications included in this meta-analysis were observational studies shadowed by limited sample sizes and methodological issues, such as selection bias and confounding and controversial causal claims. Although the selection bias attributed to the non-randomized selection of intervention and control groups is unavoidable, all included studies adopted logistic regressions to reduce confounding. Notably, propensity score matching was not conducted in any of the selected studies, and sample size constraints might be a major concern. To evaluate the publication bias in our analysis, a funnel plot and Egger’s test were used, and the test results suggested that the conclusions of our meta-analysis were not skewed by publication bias. Second, possible sources of heterogeneity might be derived from variations in AR laboratory detection, ASA dosage, duration of treatment, and clinical characteristics of enrolled patients. Thus, Cochran’s Q test, I² test, and Galbraith plot were conducted to assess the overall heterogeneity in our findings, while subgroup tests, sensitivity analyses, and univariable regression were also used to examine the distinctive differentials. Despite substantial variations, our study revealed minimal evidence of significant heterogeneity, underscoring the clinical clarity and specificity of our results. Finally, the factors and underlying mechanisms discussed above may not be sufficient to understand AR. Indeed, a wide diversity of aspirin pharmacodynamics and pharmacokinetics is equally important, if not more critical, in regulating aspirin metabolism, which includes processes such as absorption, bioavailability, and the excretion of this antiplatelet agent [15]. This makes us believe that AR is personalized and has multiple causes. To delve deeper into this issue, population-based longitudinal studies are urged to resolve meaningful questions. These questions include whether AR could be categorized as genetically determined (permanent) and acquired (temporary), elucidating the inheritance patterns of AR, investigating the duration of transient AR, and exploring the potential reversibility of AR status under specific conditions (health status and lifestyle). We conjecture that enhanced glucose and lipids levels may contribute to AR, although further evidence is required before determining any conclusions.

Aspirin is affordable, widely accessible, and a commonly used antiplatelet drug. Moreover, aspirin therapy is endorsed by the ADA as a secondary prevention measure for T2DM patients with a history of atherosclerotic cardiovascular disease [4]. This study and follow-up research could potentially positively impact clinical practices. While prescribing aspirin to T2DM patients for continuous therapy against adverse cardiovascular outcomes, we suggest that patients should be fully informed about the risk of poor lipids and glucose control. Since the risks extend beyond diabetes-related syndrome, the weakened antiplatelet capacity of aspirin makes patients vulnerable to unfavorable cardiovascular complications. This risk might be remarkably reduced by following the best practices in lipid and glucose control and regularly taking glucose and lipid profile blood tests, even though the prescription of aspirin therapy remains unchanged. AR assays may be less effective in accessing a dynamic internal environment with fluctuating glucose levels, lipids, and numerous macromolecules in the long term. Fortunately, advancements in understanding and managing these intricate dynamics offer opportunities for regulating and leveraging their potential benefits.

The current meta-analysis demonstrates that glucose levels and dyslipidemia markers effectively predict aspirin resistance in individuals diagnosed with T2DM. Further studies are needed to deepen this understanding, and the findings of our analysis and subsequent research may positively impact aspirin therapy among diabetic patients.

ACE, angiotensin-converting enzyme; ADA, American Diabetes Association; AR, aspirin resistance; ASA, acetylsalicylic acid; BMI, body mass index; CEPI-CT, collagen and epinephrine closure time; CI, confidence interval; COX-1, cyclooxygenase-1; COX-2, cyclooxygenase-2; CVD, cardiovascular disease; eGFR, estimated glomerular filtration rate; GRADE, Grading of Recommendations, Assessment, Development, and Evaluations; HbA1c, glycated hemoglobin; HDL, high-density lipoprotein; HOMA-IR, Homeostasis Model Assessment-Insulin Resistance; HPR, high platelet reactivity; IQR, interquartile range; IV, inverse variance method; LDL, low-density lipoprotein; LTA, light transmission aggregometry; MD, mean difference; MPV, mean platelet volume; MI, myocardial infarction; MH, Mantel–Haenszel model; MPA, multiplate analyzer; NO, nitric oxide; NOS, Newcastle–Ottawa scale; PFA, platelet function analyzer; PLT, platelet count; RCTs, randomized clinical trials; STI, serum thromboxane B2 immunoassay; TC, total cholesterol; TEG, thromboelastography; TGs, triglycerides; TI, TXB2 immunoassay; TP, TXA2 receptor; TXA2, thromboxane A2; TXB2, thromboxane B2; T2DM, type 2 diabetes mellitus; VNS, VerifyNow system.

The datasets used in our study are available from the corresponding author on reasonable request.

HZ and FZ planned and designed the study. FZ conducted the literature search. HZ and FZ independently performed literature screening and data extraction. HZ conducted data analysis. HZ and FZ drafted the manuscript. All authors reviewed and approved the final manuscript. All authors have participated sufficiently in the work and agreed to be accountable for all aspects of the work.

Not applicable.

This research received no external funding.

The authors declare no conflict of interest.

Supplementary material associated with this article can be found, in the online version, at https://doi.org/10.31083/RCM26009.