Atherogenic index of plasma in stroke: A comprehensive review of its diagnostic, prognostic, and pathophysiological significance

Abstract

The atherogenic index of plasma (AIP), defined as log [triglycerides (TG)/high-density lipoprotein cholesterol], an emerging lipid-based biomarker reflecting circulating TG and high-density lipoprotein cholesterol levels, has been associated with metabolic syndrome, coronary heart disease, and atherosclerosis. Its role in cardiovascular disease has been well established, yet there is growing interest in its application in cerebrovascular conditions, particularly stroke. Stroke is one of the leading causes of death and disability worldwide; hence, there is a need for integrative biomarkers to help improve risk prediction and accuracy of prognostication. Recent studies suggest that elevated AIP is independently associated with stroke incidence, especially among individuals with diabetes, prediabetes, and metabolic syndrome. Higher AIP levels have been associated with worse stroke severity at presentation, a higher risk of early neurological deterioration, and worse short-term outcomes. This is likely a consequence of AIP indicating vascular inflammation, endothelial dysfunction, and intracranial atherosclerosis. In observational studies, AIP has demonstrated comparable or stronger associations than other markers of insulin resistance, such as the TG-glucose index and the Chinese Visceral Adiposity Index, in specific metabolic populations. It is a low-cost and easily available biomarker, making it useful in primary prevention clinics, stroke units, and for managing metabolic syndrome. Given the increasing number of observational studies and population-based data, a comprehensive synthesis is needed to evaluate AIP’s diagnostic, prognostic, and pathophysiological significance in stroke. This narrative review consolidates current findings on AIP’s relevance in ischemic stroke and explores its potential integration into stroke risk stratification. Existing evidence is largely observational in nature, limiting causal interpretation.

Key Words: Atherogenic index of plasma; Ischemic stroke; Intracranial atherosclerosis; Early neurological deterioration; Insulin resistance; Biomarker

Core Tip: The atherogenic index of plasma (AIP), calculated as the logarithmic ratio of triglycerides to high-density lipoprotein cholesterol, is an emerging cardiometabolic biomarker that integrates dyslipidemia and insulin resistance. This review highlights consistent evidence linking elevated AIP with increased ischemic stroke risk, greater stroke severity, early neurological deterioration, and poorer outcomes, particularly in individuals with diabetes and metabolic syndrome. By reflecting endothelial dysfunction, vascular inflammation, and atherosclerotic burden, AIP offers a low-cost, readily available tool with potential to enhance stroke risk stratification and preventive cardiometabolic care.

INTRODUCTION

Stroke is a leading cause of morbidity, mortality, and disability worldwide. The incidence of stroke and stroke-related deaths has significantly increased worldwide between 1990 and 2019[1] and is expected to continue rising due to an aging population and the growing prevalence of modifiable risk factors like hypertension and diabetes[2]. Stroke is classified as either ischemic or hemorrhagic based on underlying pathology, with ischemic stroke accounting for 85% of all cases. The main cause of ischemic stroke is atherosclerosis, and dyslipidemia is a significant risk factor for its development[3]. Dyslipidemia is characterized by elevated triglycerides (TG), total cholesterol, and low-density lipoprotein cholesterol (LDL-C) alongside reduced high-density lipoprotein cholesterol (HDL-C). These lipid parameters are used for vascular risk assessment but have limited predictive value when considered independently[4,5].

New indicators are being explored to understand the pathophysiological mechanisms underlying stroke and to inform both management and prevention strategies (Figure 1). The atherogenic index of plasma (AIP), a biomarker for plasma atherosclerosis, was first introduced by Dobiásová and Frohlich[6] in 2001. It is calculated as log (TG/HDL-C) and reflects the circulating concentrations of TG and HDL-C. Previous studies have shown that elevated AIP is strongly associated with increased cardiovascular disease (CVD) risk[7].

Figure 1 Atherogenic index of plasma: Biomarker for stroke risk and prognosis. AIP: Atherogenic index of plasma; TG: Triglyceride; HDL-C: High density lipoprotein cholesterol; sdLDL: Small dense low density lipoprotein.

AIP has been extensively studied in coronary and metabolic syndromes, but its role in cerebrovascular diseases, particularly stroke, remains incompletely understood. Recent population studies suggest that AIP is associated with increased stroke incidence, particularly in people with diabetes and prediabetes[8,9]. Additionally, there is evidence that increasing AIP levels may indicate the severity of insulin resistance (IR) and its close association with the development of type 2 diabetes mellitus, both important contributors to stroke risk[10]. Given these intersections, there is growing interest in exploring the diagnostic, prognostic, and mechanistic significance of AIP in stroke. In this review, we synthesize current evidence on the role of AIP in ischemic stroke, with particular emphasis on large-artery atherosclerosis and intracranial atherosclerosis (ICAS), which represent the predominant focus of existing studies.

HIGHLIGHTS

AIP is emerging as a robust, low-cost biomarker linking dyslipidemia, IR, and stroke risk. Elevated AIP is associated with both increased incidence and poor outcomes of ischemic stroke, particularly in high-risk metabolic populations. AIP reflects pro-inflammatory lipid profiles and endothelial dysfunction, offering mechanistic insights into stroke pathogenesis. Incorporating AIP into routine clinical assessments may enhance early detection, risk stratification, and personalized stroke prevention strategies.

LITERATURE SEARCH STRATEGY AND STUDY SELECTION

As this is a narrative review, a comprehensive but non-systematic literature search was conducted to identify relevant studies evaluating the association between the AIP and stroke. Electronic databases, including PubMed/MEDLINE, EMBASE, and Web of Science, were searched for articles published between January 2000 and March 2025. Key search terms included combinations of “atherogenic index of plasma”, “AIP”, “stroke”, “ischemic stroke”, “cerebrovascular disease”, “insulin resistance”, and “dyslipidemia”. Original observational studies, cohort studies, cross-sectional analyses, and relevant meta-analyses published in English were included if they reported associations between AIP and stroke incidence, severity, outcomes, or related cerebrovascular pathology. Studies focused exclusively on non-human models, pediatric populations, or unrelated lipid indices without AIP reporting were excluded. Additional relevant articles were identified through manual review of reference lists from key publications. Given the narrative nature of this review, a formal risk-of-bias assessment was not performed.

PATHOPHYSIOLOGICAL BASIS LINKING AIP AND STROKE

Dobiásová and Frohlich[6], in their landmark 2001 study, found that the parameter log (TG/HDL) positively correlates with cholesterol esterification rates in apob-lipoprotein-depleted plasma (FER_HDL) and lipoprotein particle size. The hallmarks of high AIP include elevated TGs and reduced HDL-C, both parameters that independently promote a pro-inflammatory oxidative environment leading to endothelial damage. These result in a reduction in nitric oxide bioavailability, subsequent impairment of vasodilation, and upregulation of leukocyte adhesion molecules (intercellular adhesion molecule-1 and vascular cell adhesion molecule-1), leading to vascular inflammation[11,12]. TG-rich particles are associated with elevated matrix metalloproteinases, which degrade the fibrous cap and promote plaque vulnerability and rupture[13]. Determining the lipoprotein particle size is an important aspect of assessing a patient’s atherogenic profile, and FER_HDL has proven to be an indirect measure of the same. High FER_HDL is associated with smaller LDL and HDL particles, which collectively contribute to a more atherogenic lipid profile[6]. Among these, small dense LDL particles are particularly atherogenic as they are more prone to oxidative modification, have higher arterial wall penetrance, and are poorly cleared from circulation[11].

These particles have a high atherogenic potential and have been associated with the progression of carotid and ICAS. High small dense LDL levels have also been implicated in promoting carotid intimal medial thickness progression, thereby increasing CVD risk[14]. Notably, HDL particle size is also important; smaller HDL3 are less efficient in their anti-atherogenic function and less protective than larger HDL2b particles. Furthermore, lower concentrations of large HDL particles have been associated with an increased risk of stroke[15]. The observed correlation between FER_HDL and AIP suggests that AIP may serve as a potential surrogate marker for determining lipoprotein size and overall vascular risk.

In addition to being a surrogate marker of lipoprotein particle size, AIP has also been established to reflect IR, frequently observed in individuals with obesity, metabolic syndrome, and type 2 diabetes. An IR state worsens hypertriglyceridemia by increasing hepatic very LDL production, and impaired catabolism leads to a reduction in HDL-C, thereby increasing AIP[16,17]. This metabolic environment promotes systemic inflammation marked by elevated interleukin-6 and tumor necrosis factor-alpha, which causes further damage to vascular integrity[16]. IR also adversely modifies macrophages and enhances apoptosis, promoting the formation of a necrotic core leading to subsequent thrombosis[18]. The consequent lipid alterations further promote atherosclerotic plaque formation and vascular remodeling, particularly in cerebral arteries vulnerable to metabolic stress.

AIP has also emerged as a potential predictor of carotid and coronary plaque vulnerability, with previous studies linking it to the development of large artery atherosclerosis[5,19-21]. Recent studies have also linked AIP with the development of cerebral small vessel atherosclerosis, evidenced by the presence of white matter hyperintensities and lacunes[22]. These results suggest that elevated AIP may contribute to widespread endothelial dysfunction involving both large and small cerebral arteries. Further research is warranted to establish AIP thresholds that may differentiate between various stroke subtypes.

In general, oxidative stress and pro-inflammatory lipids are implicated in creating a vicious cycle promoting blood-brain barrier disruption, a hallmark of ischemic stroke. HDL-C has been shown to play an important role in maintaining the integrity of the blood-brain barrier by improving reactive oxygen species clearance and reducing local inflammation[23]. This disruption may also contribute to poorer post-stroke outcomes in patients with elevated AIP[5]. The study by Keshk et al[24] revealed that AIP correlates with inflammatory and lipid parameters in ischemic stroke patients, reinforcing its value as a potential endothelial stress marker in addition to informing atherogenic burden. Furthermore, studies have shown a stronger association between AIP and stroke risk among patients with elevated fasting glucose levels, suggesting a potential synergistic effect of metabolic dysfunction and dyslipidemia in stroke[25]. Collectively, these findings support the biological plausibility and clinical relevance of AIP as an integrative marker of lipid metabolism, glycemic status, and vascular health.

MEASUREMENT, STANDARDIZATION, AND CLINICAL INTERPRETATION OF AIP

AIP is the logarithmic transformation of the widely used TGs to HDL cholesterol ratio - log₁₀ (TGs/HDL-cholesterol). It has emerged as a reliable marker for predicting CVD risk by reflecting the balance between atherogenic and protective lipoproteins. In their study, Dobiásová and Frohlich et al[6] observed that while the existing TG/HDL-C ratio correlates well with FER_HDL (correlation coefficient = 0.67), they observed extreme skewness. The logarithmic transformation of this ratio yielded a better approximation of the normal distribution with a higher correlation with FER_HDL and a closer linear relationship (correlation coefficient = 0.803)[6]. AIP is a cost-effective, non-invasive index derived from standard lipid profiles and requires fasting blood samples for accurate measurement. It has shown consistent predictive value and is particularly useful when traditional lipid indices fail to identify subclinical risk. Thus, AIP represents a valuable addition to the cardiovascular risk assessment toolkit, especially in resource-limited settings where advanced lipid testing may not be feasible.

Risk stratification thresholds have been proposed: AIP < 0.11 indicates low risk, 0.11-0.21 indicates intermediate risk, and > 0.21 is associated with high CVD and metabolic risk. In addition, studies have shown a higher prevalence of obesity, hypertension, and uric acid levels among the high AIP cohort[26,27]. In patients with acute coronary syndrome, specific AIP thresholds of 0.186 and 0.305 were independently predictive of thin-cap fibroatheroma and plaque rupture, respectively[21]. There have also been provisional AIP cutoffs for risk stratification in ischemic stroke patients. AIP values in the range of 0.11-0.12 have consistently predicted adverse short-term outcomes in acute ischemic stroke, including death, disability, and early neurological deterioration (END)[28,29]. Another study showed that patients with cumulative AIP ≥ 0.28 had a 1.45-fold higher risk of ischemic stroke over 11 years, suggesting this level may serve as a chronic risk indicator[9]. Despite its clinical advantage, standardization remains challenging due to population heterogeneity, differences in assay methods, and demographic variables such as age, gender, and ethnicity.

KEY RESULTS ACROSS STUDIES

AIP and stroke incidence

Multiple large cohort studies have consistently demonstrated that elevated AIP levels are significantly associated with an increased risk of both first-ever and new-onset ischemic strokes. Qu et al[8] in 2024 demonstrated a strong association between higher AIP levels and stroke risk in individuals with prediabetes or diabetes, highlighting the importance of AIP monitoring in these high-risk populations. Similarly, Zheng et al[9] in 2023 reported that long-term exposure to elevated AIP significantly increases the risk of ischemic stroke over an 11-year follow-up period in a dose-dependent manner (Table 1), underscoring the need for sustained lipid monitoring and proactive management. Zhang et al[30] in 2024 further validated this association in the Chinese population, supporting the value of AIP as a marker for stroke risk stratification. Zhai et al[31] also in 2024 noted that AIP mediated approximately 30% of the relationship between obesity and stroke, suggesting that lipid management could be a pivotal strategy for stroke prevention in individuals with obesity. Additionally, studies by Wang et al[32,33] in 2020 and 2024 observed that elevated AIP levels are significantly associated with increased ischemic stroke risk, particularly in hypertensive individuals.

Table 1 Association of atherogenic index of plasma and other insulin resistance surrogate markers with ischemic stroke risk and outcomes.

Ref.	IR surrogate	Effect estimate (OR/HR)	Associated stroke subtype	Population studied
Jiang et al[38], 2025; Zheng et al[9], 2023; Nam et al[34], 2024	Atherogenic index of plasma (AIP)	OR = 1.193; HR = 1.45; aOR = 3.60	General ischemic stroke, ICAS, LAA	Abnormal glucose metabolism, ICAS, general
Jiang et al[38], 2025; Wang et al[32], 2020; Nam et al[34], 2024	Triglyceride-glucose index (TyG)	OR = 1.246; HR = 1.32; aOR = 1.60	General ischemic stroke, ICAS	General population, LAA/ICAS
Jiang et al[38], 2025; Liu et al[42], 2025	Chinese visceral adiposity index (CVAI)	HR = 1.50 (moderate-high); HR = 2.15 (high-increasing)	Ischemic stroke (long-term CVAI trajectory)	Chinese adults with diabetes/prediabetes
Weng et al[45], 2023	Metabolic score for insulin resistance (METS-IR)	OR = 1.01 per unit increase	General ischemic stroke	United States adults, stronger in age < 60
Jiang et al[38], 2025; Zabala et al[44], 2021	Estimated glucose disposal rate (eGDR)	HR = 0.77 (4-6 mg/kg/minutes); HR = 0.68 (6-8 mg/kg/minutes); HR = 0.60 (≥ 8 mg/kg/minutes)	General ischemic stroke, T2DM	Chinese adults with diabetes, Swedish cohort with T2DM
Jiang et al[38], 2025	TyG-BMI	OR = 1.186 per SD increase	Not subtype-specific	Abnormal glucose metabolism

OR: Odds ratio; HR: Hazard ratio; aOR: Adjusted odds ratio; ICAS: Intracranial atherosclerosis; LAA: Large artery atherosclerosis; T2DM: Type 2 diabetes mellitus; BMI: Body mass index; CVAI: Chinese visceral adiposity index.

AIP and stroke prognosis

Elevated AIP levels have also been predictive of unfavorable stroke outcomes. Liu et al[28] in 2021 found that higher AIP levels were independently associated with poor recovery in acute ischemic stroke, particularly in large-artery atherosclerosis. Wang et al[29] in 2023 and Nam et al[34] in 2025 also demonstrated that both AIP and the triglyceride-glucose (TyG) index were significantly associated with END in patients with large-artery atherosclerosis stroke, supporting their utility in early risk identification and tailored intervention planning.

AIP and post-stroke depression

Kong and Zou[35] in 2024 showed a linear relationship between elevated AIP and the development of post-stroke depression, with the association being stronger among female patients. This finding underscores the broader psychological implications of dyslipidemia post-stroke and suggests AIP as a potential marker for post-stroke depression risk.

AIP in specific stroke populations

AIP has also demonstrated prognostic significance in specific patient groups. Ma et al[36] in 2024 reported that higher AIP, along with other lipid-derived indices such as Non-HDL-C and Lipoprotein Combine Index, was associated with an increased risk and worse short-term outcomes of acute ischemic stroke in patients on hemodialysis. Similarly, Ahn et al[37] in 2020 found that an AIP of 0.11 or higher can predict stroke risk in patients with antineutrophil cytoplasmic antibodies-associated vasculitis, indicating its potential utility in identifying high-risk individuals within this group.

Comparative and mechanistic insights

Jiang et al[38] in 2025 compared six IR markers and found that lower Estimated Glucose Disposal Rate (eGDR) was linked to reduced stroke risk, while higher Chinese Visceral Adiposity Index (CVAI), TyG, TyG-body mass index (BMI), Metabolic Score for IR (METS-IR), and AIP levels were linked to increased risk- especially in older adults with abnormal glucose metabolism. Nam et al[22] in 2024 demonstrated that AIP and TyG index were strongly related to ICAS, likely reflecting deeper metabolic issues beyond just TG levels. Li et al[25] in 2024 also found that high AIP, total cholesterol, LDL cholesterol, and fasting glucose levels increased stroke risk in ICAS patients, highlighting the importance of managing overall metabolic health (Table 1).

CLINICAL IMPLICATIONS AND INTEGRATION INTO PRACTICE

AIP is a practical, low-cost, and reproducible biomarker that can be seamlessly incorporated into routine lipid panels without additional laboratory testing, although its clinical utility is currently supported primarily by observational evidence. In primary care settings, AIP can be a helpful tool for early detection of CVD and cerebrovascular risk, particularly in patients with subclinical dyslipidemia or prediabetes. Studies suggest that AIP performs well in prediabetic individuals as a risk-associated marker for new-onset stroke[8]. In patients with acute ischemic stroke, AIP can predict the severity of neurological impairment at presentation with higher National Institutes of Health Stroke Scale scores[28] and END[29]. Beyond the acute phase, AIP can also predict short-term functional outcomes and may help guide early rehabilitation planning and prognostic discussions[28].

AIP can be integrated into diabetes clinics to help stratify vascular risk owing to its strong correlation with IR[10]. Incorporating AIP into routine screening can help clinicians detect residual cardiovascular risk even in patients with controlled LDL-C levels. It performs well in identifying high-risk individuals with prediabetes or early-stage metabolic dysfunction[8,39] and informs clinical decisions regarding intensification of lipid-lowering therapy or early lifestyle intervention in patients with overlapping metabolic disorders.

AIP can be incorporated at multiple decision points in stroke care and metabolic risk management. Elevated AIP in patients with prediabetes or metabolic syndrome could trigger screening for IR, vascular imaging, or closer follow-up schedules[39]. In stroke units, high AIP levels may signal a higher risk of END or poor functional recovery, justifying enhanced monitoring or early rehabilitation strategies[28,29]. In secondary prevention, sustained elevation of AIP over time may support intensifying dual therapy, especially in patients with diabetes or prior stroke[9]. Currently, widely used stroke prediction tools such as the Framingham Stroke Risk Profile and Atherosclerotic Cardiovascular Disease risk calculators do not incorporate AIP. However, studies have shown that AIP adds predictive value over traditional lipids and may improve the assessment of risk in patients with cardiometabolic disease[40]. AIP can be incorporated into electronic health records-based decision support tools, as it requires only routine fasting lipid values. Future stroke risk prediction models may consider AIP as a core component to enhance vascular risk detection.

COMPARATIVE PERFORMANCE OF AIP AND OTHER IR SURROGATES

In addition to AIP, many surrogate markers of IR resistance have been studied for their utility in predicting stroke risk, especially among patients with metabolic syndrome. These include the TyG Index, the CVAI, the METS-IR, and the eGDR. These non-insulin-based markers have demonstrated promising clinical utility; however, their relative predictive power varies based on the population characteristics and stroke subtypes. The TyG index, calculated as log (fasting TG × glucose/2), has been widely validated in predicting diabetes and CVD risk[41]. This index uses readily available parameters and correlates strongly with the homeostasis model assessment for IR index, thus reflecting its role as a robust IR surrogate marker[41]. CVAI is a reliable index based on sex-specific anthropomorphic formulas accounting for waist circumference, BMI, TG, and HDL-C in the Chinese population[42]. METS-IR is another promising screening tool that combines BMI, TG, fasting glucose, and HDL-C, offering a broader outlook of cardiometabolic dysfunction[43]. eGDR, a composite metric using hemoglobin A1c, blood pressure, and BMI, has been validated against the existing standard using the euglycemic-hyperinsulinemic clamp. It has shown strong potential in predicting cardiovascular events among patients with metabolic syndrome[38,44].

In individuals with abnormal glucose metabolism, eGDR showed the most significant overall predictive power, with the other indices also showing a significant association in predicting ischemic stroke[38]. However, the role of eGDR may be limited by the need for hemoglobin A1c measurements, which may not be available in routine clinical settings, whereas AIP and TyG offer more practical alternatives. AIP also holds promise in predicting ICAS, particularly in low HDL-C cohorts[22]. In contrast, CVAI and METS-IR provide a broader view into metabolic dysfunction and visceral adiposity[38,45]. Most of these indices have been predominantly studied in Asian populations; thus, future research is required to evaluate their generalizability across diverse global cohorts.

FUTURE DIRECTIONS AND RESEARCH GAPS

Although AIP has shown strong associations with stroke risk, most current evidence comes from observational studies, limiting causal inference between AIP and the risk of stroke. More robust evidence is needed to confirm a causal relationship between AIP and stroke-specific pathophysiology before it can be integrated into clinical practice. Large multicentric prospective trials can confirm whether AIP-based risk models can improve stroke prediction in clinical settings. Mendelian randomization studies use genetic variants to test causality and can be considered to explore causal associations. Additionally, most available studies originate from East Asian populations, particularly Chinese cohorts, limiting generalizability to Western and multi-ethnic populations and underscoring potential ethnic differences in lipid metabolism and AIP distribution. Consequently, the currently proposed AIP thresholds should be interpreted cautiously until validated across diverse ethnic, age, and sex-specific populations. No interventional trials have yet assessed whether reducing AIP levels through lifestyle or pharmacological interventions improves stroke outcomes. Interventions such as dietary changes and statins can be used to evaluate the impact of lowering AIP on clinical outcomes. AIP-guided risk-scoring tools could enhance predictive accuracy beyond traditional lipids by incorporating other metabolic markers like glucose, blood pressure, and waist circumference to form the basis for a novel stroke risk prediction model. Evidence supporting a role of AIP in hemorrhagic stroke, cardioembolic stroke, or lacunar infarction remains limited or absent, representing an important area for future investigation.

CONCLUSION

AIP is a cost-effective and accessible marker that may aid in the identification of individuals at an elevated stroke risk. It reflects both dyslipidemia and IR, two key drivers of vascular damage. Studies demonstrate that elevated AIP levels are associated with both increased stroke risk and worse outcomes post-stroke. Because AIP can be easily calculated from routine lipid panels, it has great potential for broad clinical integration in both primary prevention and post-stroke care.