High-density lipoprotein cholesterol trajectory and new-onset metabolic dysfunction-associated fatty liver disease incidence: a longitudinal study

Abstract

Background

Although high-density lipoprotein cholesterol (HDL-C) exerts a significant influence on the development of metabolic dysfunction-associated fatty liver disease (MAFLD), the association of dynamic changes in HDL-C levels with the risk of MAFLD remains unclear. Thus, the aim of the current study was to explore the association between the changing trajectories of HDL-C and new-onset MAFLD. The findings of this study may provide a theoretical basis for future personalized intervention and prevention targeting MAFLD.

Methods

A total of 1507 participants who met the inclusion criteria were recruited from a community-based physical examination population in Nanjing, China from 2017 to 2021. Group-based trajectory models were constructed to determine the heterogeneous HDL-C trajectories. The incidence of MAFLD in each group in 2022 was followed up, and the Cox proportional hazards regression model was applied to investigate the associations between different HDL-C trajectories and the risk of new-onset MAFLD.

Results

The incidences of MAFLD in the low-stable, moderate-stable, moderate-high-stable, and high-stable groups of HDL-C trajectory were 26.5%, 13.8%, 7.2% and 2.6%, respectively. The incidence rate of MAFLD in the order of the above trajectory groups exhibited a decreasing trend (χ² = 72.55, P_trend<0.001). After adjusting for confounders, the risk of MAFLD onset in HDL-C low-stable group was still 5.421 times (95%CI: 1.303–22.554, P = 0.020) higher than that in the high-stable group. Subgroup analyses of the combined (moderate high-stable and high-stable groups combined), moderate-stable and low-stable groups showed that sex, age, and overweight/obesity did not affect the association between HDL-C trajectory and MAFLD risk.

Conclusions

Persistently low HDL-C level is a risk factor for the onset of MAFLD. Long-term monitoring of HDL-C levels and timely intervention for those experiencing persistent declines are crucial for early prevention of MAFLD.

Background

Metabolic dysfunction-associated fatty liver disease (MAFLD), formerly termed non-alcoholic fatty liver disease (NAFLD) [1], has a wide disease spectrum including simple fatty liver, steatohepatitis, liver fibrosis, cirrhosis, and hepatocellular carcinoma. According to the Global Burden of Disease dataset, MAFLD is the most rapidly increasing cause of cirrhosis, liver failure and liver cancer [2]. It is not merely a liver disease but rather one component of a multi-faceted, multi-organ collection of diseases resulting in a dysfunctional metabolic milieu with diverse effects [3], such as a significantly increased risk of metabolic syndrome, type 2 diabetes, atherosclerosis cardio-renal and peripheral vascular disease [4,5,6]. In recent years, the prevalence of MAFLD has been escalating annually, currently affecting approximately one-third of the global population [7]. In China, the reported prevalence of MAFLD ranges from 21.0–46.7% [8,9,10,11,12]. It is anticipated that by 2030, the total number of MAFLD patients in China will rise to 314.58 million, becoming the country with the largest increase in prevalence of this disease globally [13]. MAFLD deteriorates the quality of life of patients, imposes a heavy economic burden on society, and has become a serious public health issue. Nevertheless, the clinical development of MAFLD is insidious and frequently marked by an asymptomatic onset. Additionally, there is a scarcity of accurate and reliable non-invasive diagnostic tests, as well as tailored treatments specifically designed and approved for use in patients with MAFLD [14]. Thus, further research on predictors of MAFLD and identification of the high-risk population are essential steps for strengthening the prevention, control, and standardized management of this disease.

There are numerous factors that give rise to an increased risk of having MAFLD, including abnormal levels of blood lipids [15, 16]. Various types of abnormal blood lipid profiles exist, with abnormal cholesterol levels being one of them. Dysregulated cholesterol metabolism is not only a significant contributor to the pathogenesis of MAFLD [17], but also plays an important role in the transition from nonalcoholic steatohepatitis to hepatocellular carcinoma [18]. Cholesterol is transported through lipoproteins in the bloodstream, and high-density lipoprotein (HDL) is one of the most prominent lipoproteins. High-density lipoprotein cholesterol (HDL-C) serves as an indicator for gauging the concentration of cholesterol within HDL. HDL-C possesses the capability to facilitate reverse cholesterol transport (RCT) [19], as HDL absorbs free cholesterol from the cell surface and transports it back to the liver for processing and elimination [16, 20, 21]. Therefore, HDL-C, often regarded as the good cholesterol and known as the “blood vessel scavenger”, lower concentrations of it can be detrimental. The less cholesterol HDL absorbs, the lower the concentration of HDL-C will be, and the more free cholesterol there is, making it more prone to accumulate in the liver, leading to fatty liver [22]. Previous cross-sectional studies have indicated an inverse association between HDL-C and the prevalence of MAFLD [23, 24]. Patients with MAFLD have lower HDL-C levels, especially those with severe MAFLD [25, 26]. In prospective studies, lower baseline HDL-C levels predict the risk of MAFLD during follow-up [27, 28] and are associated with an increased risk of developing hepatocellular carcinoma in patients with liver fibrosis [29]. Based on the evidence above, it can be observed that lower levels of HDL-C are associated with an elevated risk of MAFLD occurrence and disease progression [28, 30,31,32]. However, these studies [23,24,25,26,27,28,29,30,31,32] regarding the relationship between the risk of MAFLD and HDL-C employed only a single measurement of HDL-C levels and did not explore the longitudinal association between changes in HDL-C over time and new-onset of MAFLD. It is widely known that HDL-C levels undergo continuous changes, and long-term monitoring of HDL-C levels is of considerable significance.

Therefore, based on an existing MAFLD population cohort and method of group-based trajectory model (GBTM), the current longitudinal study aimed to investigate the dynamic alternations of HDL-C concentrations and their associations with the risk of MAFLD onset. The findings of the current study may further enhance the understanding of the impact of HDL-C changes on the risk of MAFLD onset and provide more evidence for the identification of appropriate targets for early intervention.

Materials and methods

Study population

A retrospective file review of all individuals aged ≥ 18 years who underwent health examinations in a community in Nanjing, China between 2017 and 2021 was undertaken. The participants were considered as the baseline population when they underwent their initial health examination between 2017 and 2019. A total of 4,283 participants were recorded as having had at least three health examinations over a five-year period. In 2022, they were followed up for one year, and 2,921 participants ultimately attended the health examinations. The exclusion criteria were as follows: (1) participants diagnosed with MAFLD from 2017 to 2021; (2) individuals lacking complete HDL-C data; (3) individuals with autoimmune liver disease, viral hepatitis, cirrhosis, primary liver cancer, etc. As shown in Fig. 1, a total of 1,507 participants were incorporated for analysis. The current study was conducted in accordance with the Declaration of Helsinki, and the study protocol was approved by the Institutional Ethics Review Board of Nanjing Medical University (Nanjing, China). Written informed consent was obtained from all participants.

Fig. 1

Flowchart of the participant inclusion process. Abbreviations: HDL-C, high-density lipoprotein cholesterol; MAFLD, metabolic dysfunction-associated fatty liver disease

Data collection

Demographic information (e.g., age, sex) and anthropometric data (e.g., height, weight, and blood pressure) were collected using a self-designed questionnaire and an electronic medical record system. The surveys were performed by trained investigators. Height, weight, and blood pressure were measured by professional medical staff at the medical examination center using standardized procedures. Body mass index (BMI) was calculated as weight (kg)/height (m)². Blood pressure was measured in the seated position using a mercury sphygmomanometer at 5-minute intervals, and the average of the two readings was recorded as systolic and diastolic blood pressure.

Blood samples were obtained from the median forearm vein after an 8-hour fasting period. The collected blood samples were promptly analyzed to determine the required biochemical parameters. The main measures included high-density lipoprotein cholesterol (HDL-C), low-density lipoprotein cholesterol (LDL-C), triglyceride (TG), total cholesterol (TC), γ-glutamyl transpeptidase (GGT), alanine aminotransferase (ALT), aspartate transaminase (AST), fasting plasma glucose (FPG), serum uric acid (SUA), total bilirubin (TBil), direct bilirubin (DBil), serum creatinine (Scr), blood urea nitrogen (BUN), and alkaline phosphatase (ALP). All biochemical assays were measured in the same laboratory using an automated biochemical analyzer (Mindray BC-860; Mindray, Shenzhen, China) and standardized methods.

Hepatic steatosis screenings were conducted on the participants using a standardized abdominal ultrasound apparatus (Logiq E9 ultrasound system, General Electric Healthcare, Milwaukee, WI, USA). Examinations were performed by trained sonographers. Two experienced sonographers jointly examined and confirmed the results, observing the near and far-field echogenic changes in the liver and the intrahepatic ductal structures to determine the presence of hepatocellular steatosis.

Definition

According to the diagnostic criteria of MAFLD proposed by the international expert group, MAFLD was defined by the presence of hepatic steatosis (HS) on ultrasound and meeting at least one of three conditions: overweight/obesity (defined as BMI ≥ 23 kg/m² in Asians), presence of type 2 diabetes mellitus (T2DM), or presence of metabolic disorder [1]. Lean/normal-weight individuals with HS but no T2DM were considered to have a metabolic disorder if two or more of the following metabolic risk abnormalities were present:1) waist circumference ≥ 90/80 cm in Asian males and females; 2) blood pressure ≥ 130/85 mmHg or on antihypertensive therapy; 3) plasma triglycerides ≥ 150 mg/dl (≥ 1.70 mmol/L) or on specific drug therapy; 4) plasma HDL-C < 40 mg/dl (< 1.0 mmol/L) for males and < 50 mg/dl (< 1.3 mmol/L) for females or specific drug therapy; 5) prediabetes (i.e., fasting glucose levels 5.6–6.9 mmol/L, or 2-hour post-load glucose levels 7.8–11.0 mmol/L or HbA1c 5.7-6.4%); 6) homeostasis model assessment of insulin resistance score ≥ 2.5; 7) plasma high-sensitivity C-reactive protein level > 2 mg/L.

Trajectory modelling

The trajectory modelling process was completed in STATA software (Version 16.0, STATA Corporation Houston, TX, USA). Group-Based Trajectory Model (GBTM) was constructed using the traj plugin (https://www.andrew.cmu.edu/user/bjones/traj) in STATA to identify the latent trajectory groups of HDL-C among the included participants [33, 34]. This method identified cluster subgroups of events or outcomes that had similar courses over time or age [34, 35]. In our study, time was equal to years from baseline to the last follow-up visit. The nature of the dependent variable HDL-C (normal distribution) prompted us to employ a censored normal model (CNORM) [34]. The probability of group membership was estimated for each individual, and they were assigned to the group with the highest probability. Model fitting was iteratively performed by comparing models with different numbers of subgroup (1 to 6 groups) and different trajectory shapes (linear, quadratic and cubic). Among the statistically significant models, the model with the smallest absolute value of Bayesian Information Criterion (BIC), the average posterior probability (AvePP) of each trajectory group was not less than 0.7 and the membership of each group was not less than 5% was selected as the best-fit model.

Statistical analysis

All statistical analyses were processed using SPSS (Version 21.0, SPSS Inc., Chicago, IL, USA) and R software (Version 4.3.1). A two-tailed P < 0.05 was regarded as statistically significant. Continuous variables with normal and skewed distributions were described by mean ± standard deviation (SD) and median and interquartile range (IQR), respectively. Categorical variables were described by frequencies and constituent ratios. According to distribution, the Kruskal-Wallis test or one-way analysis of variance was used for comparisons of continuous variables and the chi-square test was conducted to analyze categorical variables. The trend between multiple groups was analyzed by the χ² test for linear trend. Bonferroni correction was used to control the family-wise type I error rate in all multiple comparisons, and the P-value threshold for significance after correction was 0.05/6 = 0.008. Cox proportional hazard regression was used to estimate the hazard ratio (HR) and 95% confidence interval (95% CI) for quantifying the associations of different HDL-C trajectory groups with the risk of new-onset MAFLD. Kaplan-Meier analysis was used for a graphical presentation of the time to the development of MAFLD, and the Log rank test was used to assess differences among groups. Finally, combined groups showed no statistically significant difference in prevalence. Subgroups and interaction analyses were conducted in accordance with sex, age and BMI because those grouping variables have been reported to be associated with MAFLD risk [36, 37]. Age was classified as < 38 and ≥ 38 years which were determined based on the mean values. The BMI groups were defined in line with the definition of overweight/obesity in the MAFLD diagnostic criteria.

Result

The trajectory of HDL-C

In this study, GBTM identified four distinct trajectory patterns based on the HDL-C change from 2017 to 2021, as illustrated in Fig. 2. The best-fit patterns exhibited a BIC of 817.53 and AvePP of 88.13%, 83.80%, 90.38%, and 92.82% (AvePP > 70.0%), respectively. From low to high, they were named low-stable group (n = 480, 31.85%), moderate-stable group (n = 545, 36.16%), moderate high-stable group (n = 405, 26.88%) and high-stable group (n = 77, 5.11%). The HDL-C moderate-stable group demonstrated a gradual and steady increase. The trends of the low-stable group, moderate high-stable group, and high-stable group were basically the same, with an increasing trend followed by a decreasing trend. Figure 3 showed that the mean HDL-C ranges were 1.06–1.30,1.36–1.66, 1.60–2.10, and 2.04–2.70 in the low-stability, moderate-stable, moderate high-stable and high-stable groups, respectively.

Fig. 2

HDL-C trajectory during 2017–2021. Abbreviations: HDL-C, high-density lipoprotein cholesterol

Fig. 3

HDL-C mean during 2017–2021. Abbreviations: HDL-C, high-density lipoprotein cholesterol

Baseline characteristics

The demographic and clinical baseline characteristics of the participants, stratified by HDL-C trajectories, were presented in Table 1. A total of 1507 participants (956 males and 551 females) with an average age of 42.44 ± 8.51 years were enrolled in this study. There were 399 (83.1%) males and 81 (16.9%) females in low-stable group; 363 (66.6%) males and 182 (33.4%) females in moderate-stable group; 178 (44.0%) males and 227 (56.0%) females in moderate high-stable group; 16 (25.7%) males and 61 (79.2%) females in high-stable group. Significant statistical differences were observed in age, sex, BMI, SBP, DBP, ALT, AST, Scr, BUN, SUA, ALP, FPG, GGT, LDL-C, HDL-C, TG, and TC among different HDL-C trajectories (all P values < 0.05). However, no significant differences were detected between the groups in DBIL and TBIL (all P values > 0.05).

Table 1 Baseline characteristics in different HDL-C trajectory groups

Incidence of MAFLD in different trajectory groups

The findings of the study in Fig. 4 demonstrated that during an average follow-up period of 4.49 ± 0.62 years, a total of 233 (15.5%, 233/1507) incident cases of MAFLD were detected. The incidence of MAFLD by the end of the final follow-up was 26.5%, 13.8%, 7.2%, and 2.6% in the low-stable, moderate-stable, moderate high-stable, and high-stable groups, respectively. In the order of the above groups, the incidence rate of MAFLD showed a decreasing trend (χ² = 72.55, P_trend<0.001). The results from multiple comparisons revealed that the incidence of MAFLD in the low-stable group was significantly higher than that in the other three groups (all P for Bonferroni correction < 0.008), and the moderate-stable group exhibited a higher incidence than both the moderate high-stable and high-stable groups (all P values for Bonferroni correction < 0.008). However, no significant difference was observed in the incidence of MAFLD between the high-stable and moderate high-stable groups (P for Bonferroni correction = 0.135).

Fig. 4

Incidence of MAFLD among different HDL-C trajectory groups. Abbreviations: MAFLD, metabolic dysfunction-associated fatty liver disease

Univariate analysis of predictive factors for MAFLD

The presence of MAFLD at the follow-up health examination (no = 0, yes = 1) was employed as the dependent variable, while independent variables such as sex, age, and BMI were considered predictive factors. A univariate analysis of potential predictive factors for MAFLD was conducted using the Cox hazard regression method, with the results presented in Table 2. The findings indicated that sex, BMI, SBP, DBP, ALT, ASL, DBIL, TBIL, Scr, SUA, ALP, GGT, LDL-C, and TG were influencing factors for the incidence of MAFLD (all P values < 0.05).

Table 2 Univariate Cox proportional hazard associations of MAFLD risk in the study population

Relationship between HDL-C trajectory and MAFLD risk

The results of the Kaplan-Meier analysis across the 4 trajectory groups of HDL-C level are depicted in Fig. 5. The MAFLD risk was significantly different between the four trajectory groups (log rank test χ² = 84.22, P < 0.001). Further multiple comparisons showed that the MAFLD risk was lower in the moderate-stable, moderate high-stable, and high-stable groups than in the low-stable group, and lower in both the moderate high-stable and high-stable groups than in the moderate high-stable group.

Fig. 5

Kaplan-Meier analysis of MAFLD risk according to HDL-C trajectories

A Cox proportional risk regression model was conducted to assess the association between HDL-C trajectories and MAFLD risk. The variables that exhibited a significant association with the new-onset of MAFLD through univariate Cox proportional hazards regression analysis were included as confounding factors and adjusted for in this analysis. The results are shown in Table 3. In the unadjusted model (Model 1), the risk of MAFLD in the low-stable and moderate-stable groups were 13.459 (95% CI: 3.328–54.438) and 6.677 (95% CI: 1.639–27.204) times higher than that in the high-stable group, respectively. After adjusting for sex, BMI, SBP, and DBP (Model 2), the risk of MAFLD in the low-stable and moderate-stable groups were 7.835 (95% CI: 1.924–31.907) and 4.838 (95% CI: 1.183–19.780) times higher than that in the high-stable group. In the fully adjusted model (Model 3), the risk of MAFLD in the low-stable group was 5.421 (95% CI: 1.303–22.554) times higher than that in the high-stable group.

Table 3 Relationship between HDL-C trajectory and MAFLD risk

Subgroup and interaction analyses

In light of the absence of statistically significant disparities in incidence between the moderate high-stable and high-stable groups, these two groups were consolidated and designated as the combined group. Subgroup analyses were performed according to sex, age, and BMI. As shown in Table 4, the association between HDL-C trajectories and MAFLD risk was evident in all subgroups (all P_trend<0.05). Interaction analyses revealed no significant differences in the MAFLD risk among trajectory groups across sex, age, and BMI subgroups (all P_interaction>0.05).

Table 4 Subgroup analyses for the association between HDL-C trajectories and MAFLD risk

Discussion

In this longitudinal study examining participants without MAFLD at baseline, 15.5% developed MAFLD over an average follow-up of 4.51 ± 0.61 years. We conducted a longitudinal analysis of HDL-C changing trajectories from 2017 to 2021 in a community cohort, employing GBTM analysis. Four distinct HDL-C trajectory patterns for 5 years were identified. Compared with those in the moderate-stable group, the moderate high-stable group, and the high-stable group, participants with a persistently low HDL-C level demonstrated a significantly increased risk of MAFLD during the follow-up period.

The incidence of MAFLD has been widely investigated. A global meta-analysis, encompassing 17 countries, revealed an overall estimated incidence of 46.9 cases per 1000 person-years for MAFLD [38]. A meta-analysis focused on MAFLD in Asia reported an incidence of 50.9 per 1,000 person-years, with the highest incidence of 63 per 1,000 person-years in mainland China and the lowest incidence of 29 per 1,000 person-years in Japan [39]. During a Korean cohort study, with a follow-up duration of 348,193.5 person-years, 10,340 participants developed MAFLD at an incidence rate of 29.7 per 1,000 person-years [40]. In addition, a study of 565 participants from Hong Kong with a 3–5 year follow-up estimated that MAFLD incidence was 13.5% [41]. The incidence rate of MAFLD in this study was slightly higher than that in the Hong Kong area, confirming the higher incidence of MAFLD in mainland China.

In order to identify 5-year trajectory groupings and trends in HDL-C, this study employed GBTM for trajectory construction and finalized four different trajectory patterns. In a study involving 330 participants and exploring the effect of lifestyle and biochemical measurements on HDL-C trajectories, the trajectory of HDL-C change was categorized into 3 trajectories [42]. The trajectory trends of Group 1 and Group 2 exhibited a pattern similar to those of the low-stable group, moderate high-stable group and high-stable group in this study, which was characterized by an initial increase and then a subsequent decrease. The trajectory of Group 3 displayed a decreasing trend followed by an increasing trend, and no trajectories similar to it were found in this study. The different number of trajectory groups might be attributed to the different number of study participants.

The above study [42] also reported significant differences in sex, BMI, SBP, DBP, FPG, HDL-C, TG, and TC among the three groups at the baseline, which was consistent with the results in our study. However, in this report, no significant difference in age and LDL-C was observed among different groups, which was inconsistent with our study. Additionally, in our study, it was found that ALT, AST, Scr, BUN, SUA, ALP, and GGT were significantly different at baseline. Previous prospective studies have demonstrated that factors such as sex, BMI, SBP, DBP, ALT, Scr, SUA, GGT, DBIL, TG, and LDL-C contribute to the risk of MAFLD [43, 44]. These observations were consistent with this study. Furthermore, our study also discovered that AST, ALP, and TBIL were also risk factors for the risk of MAFLD. The elevation of AST indicated liver damage, and related studies showed that in instances where FibroScan was unavailable, the AST to platelet ratio index (APRI) was one of the best indicators for evaluating liver fibrosis in MAFLD patients [45]. ALP and TBIL were also crucial indicators reflecting liver function. However, the specific mechanisms through which these parameters contributed to the development of MAFLD remain unclear and necessitate further exploration.

In further analysis, after adjusting for sex, BMI, SBP, and DBP, the risk of MAFLD in the low-stable group of HDL-C trajectory was 7.835 times higher than that in the high-stable group. On this basis, after adjusting for ALT, AST, Scr, SUA, ALP, GGT, LDL-C and TG, the risk of MAFLD in the low-stable group of HDL-C trajectory was 5.421 times higher than that in the high-stable group. Therefore, a persistent reduction in HDL-C level could be an important clue to the development of MAFLD, and the risk of MAFLD was the highest when HDL-C was 1.06–1.30 mmol/L in the low-stable group. In a cohort study with a 6-year follow-up, baseline HDL-C remained significantly and independently associated with MAFLD incidence [27]. In another health screening cohort study, it was shown that the incidence of MAFLD decreased as HDL-C quartiles rose [28]. This finding was consistent with the results of our study. MAFLD was regarded as a manifestation of the metabolic syndrome in the liver [44], and a decreased HDL level was itself a feature of the metabolic syndrome [46]. Many studies showed that the imbalance of cholesterol efflux activity in HDL-C function might be the main cause of metabolic abnormality [26, 47, 48], which might support the results of this study to a certain extent.

The findings from the subgroup analysis demonstrated that the association between HDL-C trajectory and the risk of MAFLD was unaffected by sex, age, and overweight/obesity. In fact, the impact of sex on MAFLD remained inconclusive, and the mechanisms underlying sex differences were still unclear [49]. Previous research reports had indicated that overweight/obesity was a significant risk factor for MAFLD [50]. However, a meta-analysis indicated that MAFLD might also occur in non-obese populations [51]. Our study, for the first time, found a correlation between HDL-C changing trajectory and the incidence of MAFLD, independent of sex and overweight/obesity. These results and the involved mechanisms warrant further investigation.

Predicting the progression of MAFLD using biochemical indicators has been a prominent research topic in recent years. Owing to the insufficiency of a single indicator test, liver function indices such as ALT, AST, and GGT are subject to individual reactivity, resulting in limited sensitivity and specificity. Furthermore, serum lipid indices TC and TG are susceptible to interferon levels within the patient’s body, leading to fluctuations in serum concentrations. Recent studies have instead concentrated on the combination of HDL-C with other biochemical indices. A cross-sectional study indicated that GGT/HDL-C ratio could serve as a predictive factor for the prevalence of MAFLD [26]. A secondary prospective cohort study conducted among non-obese Chinese adults revealed a positive and non-linear relationship between the GGT/HDL-C ratio and the risk of MAFLD [52]. TG/HDL-C was a potential risk factor for MAFLD and could predict the incident fatty liver [53,54,55]. SUA/HDL-C was inexpensive and easy-to-assess, thereby being regarded as a useful tool in diagnosing hepatic steatosis [56]. In addition, research has indicated that small dense low-density lipoprotein cholesterol (sdLDL-C) levels were specifically elevated in patients with diabetes and MAFLD [57]. The elevated ratio of sdLDL-C to HDL‐C (SHR) is independently linked to an increased risk of MAFLD in patients with T2DM, and may serve as practical indicators for assessing the risk of MAFLD in T2DM patients [58]. However, most studies only considered HDL-C levels at baseline or at a point in time, ignoring the continuity of HDL-C. In China, researchers explored the relationship between HDL-C trajectory and the new-onset of diabetes mellitus (DM) and clarified the role of HDL-C changes in the process of DM [59]. This showed that HDL-C indicators might be closely associated with metabolic-related diseases. Therefore, in this study, GBTM was employed to develop a dynamic HDL-C trajectory for non-MAFLD patients. To the best of our knowledge, this is the first exploration of the interrelation between diverse HDL-C alteration trajectories and the progression of MAFLD.

The main strengths of the study lied in the longitudinal study design and the application of GBTM trajectory analysis, which provided robust epidemiological evidence for the dynamic change of HDL-C level and its relationship with the incidence of MAFLD in Chinese population. The high-risk population of MAFLD can be screened early by monitoring the change of HDL-C. However, this study had some limitations. Firstly, we were incapable of employing liver tissue biopsy as a diagnostic criterion for hepatic steatosis. Given the difficulties associated with liver biopsy in general population screening, abdominal ultrasound was utilized to identify steatosis. Secondly, our study was deficient in relevant information regarding lifestyle factors, such as exercise and diet, that may affect HDL-C levels. Thirdly, our sample was confined to Chinese adults in a community, and it remains ambiguous whether the study findings can be extrapolated to other populations. Consequently, further research is required to adjust for possible confounding factors and corroborate these results.

In summary, our findings revealed a significant correlation between HDL-C changing trajectory and the risk of developing MAFLD. The incidence of MAFLD in the low-stable HDL-C dynamic level was significantly higher than that in the other 3 trajectory groups. This suggested that monitoring longitudinal patterns for HDL-C change and lipid management may provide an inspiration for targeted intervention and prevention of MAFLD.