Non-alcoholic fatty liver disease (NAFLD) is characterized by a spectrum of liver injuries resulting from abnormal lipid metabolism and hepatic fat accumulation (1, 2). With the increasing prevalence of metabolic disorders, NAFLD has emerged as the leading cause of chronic liver disease worldwide, with global prevalence reaching 38% (3, 4). Notably, the incidence of NAFLD in non-obese individuals has been steadily rising, now accounting for approximately 40% of NAFLD cases (5). This subgroup presents a particularly complex clinical scenario, with emerging evidence suggesting that non-obese NAFLD patients may experience more aggressive disease progression and potentially more severe clinical outcomes compared to their obese counterparts (6).

The pathogenetic mechanisms underlying NAFLD reveal an intricate interplay of metabolic processes (1, 7). While abnormal blood lipids have long been considered an important pathogenic factor (8, 9), emerging research indicates that many non-obese individuals with NAFLD actually maintain normal lipid levels (10, 11). Given the increasing prevalence and complexity of non-obese NAFLD, continued investigation is essential to identify novel lipid metabolism risk factors for individuals with normal lipid levels, thereby improving more precise preventive, diagnostic, and therapeutic strategies.

Recent research suggests the Lipid Composite Index (LCI) effectively predicts metabolic and cardiovascular risks with greater accuracy than individual lipid measurements alone (12–15). Furthermore, a cross-sectional investigation conducted in a Japanese population demonstrated a significant positive association between LCI and NAFLD, revealed LCI as a more reliable marker for NAFLD identification than traditional lipid parameters (16). However, longitudinal evidence examining the relationship between baseline LCI and NAFLD development remains limited, especially in non-obese Chinese within normal lipids profile. Therefore, this 5-year longitudinal cohort study aims to investigate whether LCI can effectively predict NAFLD incidence in this unique population, potentially providing clinicians with a valuable tool for risk stratification and early preventive strategies in non-obese individuals with normal lipid profiles who may nevertheless be at risk for metabolic complications such as NAFLD.

In this retrospective cohort study, we discovered that elevated LCI was significantly linked with increased risk of NAFLD development in non-obese Chinese population with normative lipid levels. The association between LCI and NAFLD incidence exhibited a non-linear pattern, with critical turning point at 5.514 mmol²/L², where the HR was significantly higher below this threshold (HR = 1.282, 95%CI: 1.162–1.415) compared to above it (HR = 1.063, 95%CI: 1.042–1.084). BMI significantly modified this relationship, with stronger associations observed in normal-weight individuals than lean participants. Additionally, LCI demonstrated a higher predictive capacity for NAFLD than individual lipid parameters with an AUC of 0.717.

This study revealed a cumulative NAFLD occurrence of 8.69% during a median follow-up of 33.10 months, corresponding to an incidence of 31.51 cases per 1000 person-years. This rate varied substantially across LCI quartiles, from 8.68 in the lowest quartile to 66.99 per 1000 person-years in the highest quartile. While broadly consistent with established Asian population studies, our data reveal a nuanced reduction in incidence rates. The meta-analysis conducted by Ye et al. substantiated annual NAFLD prevalence in Asian populations between 35.0 and 52.34 cases per 1000 person-years (5). This relatively lower occurrence rate in our cohort likely reflects our specific inclusion criteria focusing on non-obese individuals with normal lipids profile, which included populations that are often overlooked and typically considered to have relatively low risk (9).

Our Cox regression analysis revealed a significant positive correlation between LCI and NAFLD risk, with each unit increase in LCI corresponding to an 8.5% increased risk of NAFLD after adjusting for confounders. Our results align with the pioneering cross-sectional research by Qiu et al., who initially characterized LCI as a potential NAFLD indicator within a Japanese population (16). However, our study extends these findings by providing longitudinal evidence of this association, establishing the temporal relationship between baseline LCI and subsequent NAFLD development. Additionally, while Qiu et al. focused on NAFLD detection in a general population, our study specifically examined the predictive value of LCI for NAFLD incidence in non-obese individuals with normal blood lipids levels, a population often considered lower risk but increasingly recognized as susceptible to “lean NAFLD”, defined as the presence of hepatic steatosis in individuals with BMI < 25 kg/m² in non-Asian populations or <23 kg/m² in Asian populations without excessive alcohol consumption or other secondary causes of fatty liver (3, 5, 31).

The underlying mechanisms linking LCI to NAFLD development are multifaceted. As a composite index incorporating multiple lipid parameters, LCI reflects a comprehensive assessment that outperforms individual lipid markers in predicting NAFLD development. This superiority stems from LCI’s ability to simultaneously capture the comprehensive engagement of pro-atherogenic and anti-atherogenic components involved in hepatic lipid accumulation (32). The association between LCI and NAFLD can be explained through three interconnected biological mechanisms that directly address hepatic steatosis development. (1) Lipid Metabolism Dysregulation: Elevated TC, TG and LDL-c levels, which increase LCI, promote hepatic lipid accumulation through increased free fatty acid flux and de novo lipogenesis (33, 34). Each component of LCI reflects specific aspects of disrupted lipid homeostasis. Elevated TC and LDL-c indicate impaired cholesterol clearance and increased hepatic cholesterol content, which activates sterol regulatory element-binding protein-1c (SREBP-1c) and enhances lipogenic gene expression (35). Elevated TG reflects increased de novo lipogenesis and impaired fatty acid oxidation, while reduced HDL-c indicates compromised reverse cholesterol transport and reduced cholesterol efflux from hepatocytes (36). The multiplicative formula of LCI captures the compounding effect when multiple lipid abnormalities occur simultaneously. Reduced HDL-C, also reflected in higher LCI, impairs reverse cholesterol transport and reduces anti-inflammatory protection (37, 38). (2) Liver Fat Accumulation Mechanisms: Additionally, dyslipidemia often coexists with insulin resistance, which further promotes liver lipid accumulation by increasing lipid mobilization in adipose tissue and upregulating lipogenic transcription factors such as SREBP-1c in hepatocytes. Mechanistically, elevated LCI reflects increased substrate availability for hepatic steatosis through two major pathways: enhanced delivery of free fatty acids to the liver and upregulated de novo lipogenesis (39). Insulin resistance serves as a central mediator in this process, promoting adipose tissue lipolysis, impairing very low-density lipoprotein secretion, and activating lipogenic transcription factors such as SREBP-1c in hepatocytes (40). (3) Inflammatory Pathways: The lipid abnormalities captured by LCI trigger hepatic inflammatory responses through multiple pathways. Elevated LDL-c promotes oxidative stress and pro-inflammatory cytokine production in hepatocytes, while reduced HDL-c diminishes anti-inflammatory protection and impairs the liver’s ability to resolve inflammation (41). This inflammatory environment further exacerbates insulin resistance and promotes the progression from simple steatosis to more severe forms of NAFLD (32). As LCI comprehensively captures the complex interplay among insulin resistance, lipid metabolism dysregulation, and inflammatory pathways, it more accurately represents the “multiple-hit” pathogenesis of NAFLD compared to isolated lipid measurements (1, 32). These three mechanisms work synergistically to promote NAFLD development, which may help explain why LCI, as a composite index incorporating multiple lipid parameters, shows enhanced predictive performance compared to individual lipid markers in our study.

Statins, as HMG-CoA reductase inhibitors, directly target the lipids components reflected in elevated LCI scores. By reducing hepatic cholesterol synthesis, statins lower circulating LDL-C and may modulate hepatic lipid metabolism pathways implicated in NAFLD pathogenesis (42). Beyond their lipid-lowering effects, statins can mitigate insulin resistance, oxidative stress and fibrogenesis, potentially interrupting the pathophysiological mechanisms linking elevated LCI to NAFLD development (43, 44). These pleiotropic effects may portend statin therapy’s potential in ameliorating hepatic steatosis, particularly in patients with high LCI scores.

Our subgroup investigations uncovered BMI’s potential role as a significant effect modifier in the LCI and NAFLD risk relationship. The most robust statistical correlations emerged among participants categorized in the intermediate BMI spectrum (BMI ≥ 18.5, < 24 kg/m²). In contrast, individuals with BMI over 24 kg/m² exhibited considerably weaker associations, while participants with BMI below 18.5 kg/m² showed no significant associations. Whereas previous studies demonstrated that lipid parameters mediate the relationship between NAFLD risk and BMI (45), our research is the first to unveil the adjusting effect of BMI on the correlation between LCI and NAFLD incidence. Mechanistically, previous researches have indicated that higher BMI levels are linked to lower NAFLD risk (2), which explains why higher BMI (≥ 24 kg/m²) may partially mask LCI effect. In contrast, lean individuals typically have lower insulin resistance and inflammation levels (31), which may weaken LCI’s predictive value for NAFLD risk. This finding has important clinical implications: it reveals the interaction between LCI and BMI in NAFLD risk, suggesting that LCI has important predictive value for NAFLD incidence in individuals with normal weight and normal blood lipids, providing new insights for individualized NAFLD risk assessment and intervention.

A novel finding in this research is the non-linear relationship between LCI and NAFLD risk, with an inflection point at 5.514 mmol²/L². Below this threshold, the association between LCI and NAFLD was substantially stronger compared to values above this point. While no previous studies have specifically examined non-linearity in the LCI-NAFLD relationship, our results are consistent with the work of Chen et al., which demonstrated non-linear relationships between traditional lipid parameters and NAFLD (46). The identified inflection point at LCI = 5.514 mmol²/L² represents a critical metabolic threshold with important clinical implications. Below this threshold, LCI demonstrates substantially stronger association with NAFLD development (HR = 1.282, 95%CI: 1.162–1.415) compared to values above this point (HR = 1.063, 95%CI: 1.042–1.084), potentially suggesting distinct metabolic phenotypes (47). Supplementary Table 2 reveals that participants with an LCI of ≤5.514 mmol²/L² had a smaller proportion of females and exhibited lower levels of BMI, SBP, DBP, AST, ALP, GGT, UA, and LDL-C. These variables are strongly linked to NAFLD. In this metabolically healthier population, the relative absence of conventional risk factors amplifies the predictive impact of subtle lipid abnormalities captured by LCI (48). This phenomenon aligns with the concept that early metabolic dysfunction may be more readily detected through composite lipid indices before overt clinical manifestations appear (49). When LCI exceeded 5.514 mmol²/L², its influence on NAFLD was small. In contrast, below 5.514 mmol²/L², these risk contributors for NAFLD were diminished and had a lesser impact, making the LCI effect more pronounced. This curvilinear relationship between LCI and NAFLD has important clinical meaning. This threshold enables enhanced risk stratification, particularly in seemingly low-risk populations. For individuals with LCI below 5.514 mmol²/L², targeted lipid-modifying interventions may yield greater NAFLD prevention benefits due to the steeper risk gradient in this range (50). The stronger predictive value in metabolically healthier individuals supports LCI’s utility as a screening tool for populations who might be overlooked by traditional risk assessment strategies. By precisely intervening to reduce LCI below 5.514 mmol²/L² in non-obese populations, we can significantly lower the risk of NAFLD. When LCI below this critical point, the incidence of NAFLD events demonstrates a more rapid decline. This research offers an innovative risk assessment and prevention decision-making tool for clinicians.

The AUC value of 0.717 for LCI indicates acceptable discriminatory ability for NAFLD screening (51). A study by Qiu et al. using the NAGALA cohort reported a higher AUC of 0.8118 for LCI in the general population (16). The difference likely reflects distinct study designs (longitudinal cohort vs. cross-sectional) and population characteristics: our study focused on non-obese individuals with normal lipid profiles, while NAGALA included the general population. NAFLD prediction becomes particularly challenging in non-obese, normal-lipid populations where traditional risk factors show limited discriminatory power. Importantly, DeLong test results in both studies confirmed that LCI significantly outperformed individual lipid markers (all P < 0.05), validating the superior diagnostic value of composite indices across different populations. For screening applications, moderate discriminatory ability remains clinically meaningful when identifying individuals requiring further evaluation (52). The value of LCI lies in its practical advantages as a cost-effective, non-invasive tool that enhances NAFLD risk stratification without additional laboratory costs.

This study has several notable advantages: (1) We employed a substantial sample size and ensured adequate follow-up, establishing the temporal relationship between baseline LCI and subsequent NAFLD development. (2) Significantly, this study is the first to thoroughly investigate the predictive capacity of LCI for NAFLD in non-obese individuals who have normal blood lipids, an often-overlooked population. (3) Rigorous statistical adjustments were implemented to reduce possible confounding factors. (4) Our comprehensive statistical approach, including non-linear modeling, provided nuanced insights into the LCI-NAFLD relationship. (5) The robustness of our findings was assessed using a range of stratified analyses and sensitivity testing approaches.

Despite these strengths, several limitations should be acknowledged. First, the diagnosis of NAFLD relied on ultrasonography instead of liver biopsy, limiting assessment of disease severity. Second, our study included exclusively Chinese individuals, potentially limiting generalizability to other populations. Third, we only assessed baseline LCI without accounting for changes over time. Finally, our study used NAFLD rather than the recently proposed metabolic dysfunction-associated fatty liver disease (MAFLD) classification, as our design predated this change. Nevertheless, epidemiological data suggest more than 95% of NAFLD patients fulfill the criteria for MAFLD, supporting that our findings can likely be extrapolated to the MAFLD framework (53).

In conclusion, our study demonstrates a positive and non-linear association between LCI and NAFLD risk in non-obese Chinese individuals with normal lipids. While LCI showed a relative advantage over individual lipid parameters in NAFLD prediction, the moderate discriminatory ability (AUC = 0.717) suggests its utility may be primarily as a practical screening marker in normal-weight individuals. Our analysis revealed this relationship exhibits a threshold effect, where NAFLD risk increases significantly when the LCI falls below 5.514 mmol²/L². These findings suggest that LCI assessment may serve as a complementary screening approach for identifying individuals at risk in seemingly low-risk populations with normal lipid profiles. However, given the moderate predictive performance, clinical implementation should be considered cautiously and in conjunction with other risk factors. Future studies should validate these findings across diverse populations and investigate whether LCI-guided screening strategies can effectively contribute to early NAFLD detection and prevention efforts.

As previously reported, the study was conducted in strict adherence to the ethical guidelines of the Declaration of Helsinki (10). Each participant provided oral informed consent prior to study involvement. Personal identifiable information was anonymized and replaced with a medical examination identifier. The Ethics Committee of Wenzhou People’s Hospital reviewed and approved the research protocol (10).