Trends in baseline blood lipid levels in randomized placebo-controlled trials of overweight or obesity from 1990 to 2024

Abstract

BACKGROUND

The global rise in overweight and obesity has reached alarming levels, substantially increasing the risk of metabolic disorders such as dyslipidemia. We outlined the evolving trends in baseline blood lipid levels among patients experiencing overweight or obesity, as observed in placebo-controlled randomized trials, to address the unmet clinical requirements.

AIM

To assess long-term trends in lipid profiles in overweight or obese populations and their association with clinical and treatment factors.

METHODS

EMBASE, PubMed, Cochrane Library, and Web of Science databases were searched up to October 9, 2024. Randomized placebo-controlled trials of participants with overweight or obesity, with reports of baseline lipid levels, were included. The main outcome was a correlation between pooled baseline levels of triglycerides (TG), total cholesterol (TC), high-density lipoprotein cholesterol (HDL-C), and low-density lipoprotein cholesterol (LDL-C) with study year. Subgroup analysis was conducted based on characteristics of the populations and intervention types.

RESULTS

A comprehensive meta-analysis encompassing 866 studies across nearly 60 countries and regions worldwide, involving 3300 participants, revealed significant temporal trends in baseline lipid profiles. The analysis revealed a significant decline in TG (Rs = -0.704, P < 0.001, I² = 98.6%), TC (Rs = -0.884, P < 0.001, I² = 99.6%), and LDL-C (Rs = -0.808, P < 0.001, I² = 96.8%) levels. In contrast, HDL-C (Rs = 0.336, P = 0.041, I² = 99.2%) levels exhibited a progressive increase over the study period. Subgroup analyses revealed that sex, body mass index, blood pressure, diabetes status, and type of intervention influenced the observed trends, especially with patients receiving pharmacological therapies demonstrating more pronounced improvements (TG: Rs = -0.449, P_adj = 0.011; I² = 98.9%; TC: Rs = -0.650, P_adj = 0.001; I² = 99.4%; HDL-C: Rs = 0.650, P_adj = 0.002; I² = 98.6%; LDL-C: Rs = -0.417, P_adj = 0.031; I² = 98.0%).

CONCLUSION

Despite rising obesity rates, lipid control has improved over three decades among individuals with overweight or obesity, reflecting the positive impact of public health efforts and effective dyslipidemia treatment strategies.

Key Words: Overweight; Obesity; Dyslipidemia; Lipid profile; Triglycerides; Cholesterol

Core Tip: This global meta-analysis of 866 randomized trials (n = 3300) reveals significant improvements in lipid profiles among overweight/obese individuals over three decades, with triglycerides, total cholesterol and low-density lipoprotein-cholesterol declining while high-density lipoprotein-cholesterol increased. Notably, pharmacological interventions showed the most pronounced benefits. These findings highlight that despite rising obesity rates, concerted public health efforts and therapeutic advances have successfully mitigated dyslipidemia risks in this high-risk population, offering crucial insights for clinical practice and health policy.

INTRODUCTION

The prevalence of overweight and obesity has risen dramatically over the past three decades, reaching levels considered pandemic worldwide. Projections suggest that if current trends continue, more than half of the global adult population could be affected by 2050[1]. This concerning trend is not merely a cosmetic concern but a significant public health issue that is intricately linked to a range of metabolic disorders, including dyslipidemia[2].

Dyslipidemia is a clinical condition characterized by abnormal blood lipid levels, specifically elevated triglycerides (TG) and total cholesterol (TC), reduced high-density lipoprotein cholesterol (HDL-C), and a predominance of small, dense low-density lipoprotein cholesterol (LDL-C) particles[3]. The relationship between adiposity and dyslipidemia is complex and multifactorial. Mechanistically, several interrelated pathways like insulin resistance[4], chronic inflammation[5], altered adipokine secretion[6], and hepatic lipid metabolism[7] play pivotal roles in exacerbating lipid dysregulation. These lipid abnormalities are known to drive a 2- to 3-fold increased risk of atherosclerotic cardiovascular disease, which includes myocardial infarction and stroke[8].

As the number of placebo-controlled clinical trials focusing on populations with overweight and obesity continues to rise, it is now feasible to assess global patterns of obesity-related lipid disorders through the lens of randomized controlled trials (RCTs). To the best of our knowledge, extensive studies with standardized designs have not yet thoroughly examined the long-term trends in lipid control among individuals with overweight or obesity enrolled in RCTs. Additionally, the unique trajectories of lipid profiles (such as TC, LDL-C, HDL-C, and TG), stratified by factors including sex, age, body mass index (BMI), blood pressure, blood glucose, and types of intervention within this group, remain inadequately characterized.

Accordingly, this study aims to elucidate trends in baseline lipid profiles among populations struggling with obesity and overweight, utilizing data sourced from placebo-controlled randomized trials. This investigation has the potential to offer fresh perspectives for the design of future RCTs aimed at assessing anti-obesity interventions, and it may also encourage advancements in therapeutic strategies and research methods focused on weight management and enhancing metabolic health.

MATERIALS AND METHODS

The meta-analysis protocol was prospectively registered on the PROSPERO platform (ID: CRD420250650197) prior to data extraction. Study implementation rigorously followed the PRISMA guidelines.

Search strategy and study selection

We selected EMBASE, PubMed, Cochrane Library, and Web of Science databases for publications available up to October 9, 2024. Our search utilized a combination of medical terms and free-text search terms, including obesity, overweight, randomized controlled trials, blood lipids, cholesterol, triglycerides, low-density lipoprotein cholesterol, high-density lipoprotein cholesterol, exercise, lifestyle, bariatric surgery, orlistat, bupropion/naltrexone, phentermine/topiramate, and glucagon-like peptide-1 receptor agonists. To ensure the inclusion of relevant studies, we also reviewed the reference lists of pertinent reviews in this field.

The inclusion criteria for this study were as follows: (1) Availability of full text; (2) Inclusion of adult populations aged ≥ 18 years classified as overweight and/or obese; (3) Studies must provide data on baseline blood lipid levels, including TG, TC, HDL-C, and LDL-C; (4) The study design must be a randomized placebo-controlled trial; and (5) Studies must be published in English. The exclusion criteria were: (1) Inability to extract blood lipid data separately; (2) Inclusion of populations with comorbidities such as pregnancy, mental illness, cancer, secondary obesity, or the use of concurrent medications or interventions likely to influence lipid levels; and (3) Studies that excluded populations based on restricted blood lipid levels at baseline. Two investigators (Wang YQ and Xiao QZ) independently browsed the title, abstract, full text, and reference list for potentially eligible trials. Discrepancies were resolved through discussion with a third reviewer (Yang Y).

Data extraction and quality assessment

Two reviewers independently extracted data using predefined forms, including the study characteristics (first author, publication year, country, and sample size), participant characteristics (age, sex, baseline BMI, baseline blood pressure, and blood glucose control condition), and intervention methods (e.g., diet, exercise, pharmacotherapy, and surgery). The quality assessment of selected studies used the Cochrane risk of bias assessment tool.

This meta-analysis aimed to evaluate the pooled effect sizes of baseline blood lipid levels, including TG, TC, HDL-C, and LDL-C, derived from participants in studies that were published in the same year. We utilized a random-effects model for our analysis due to the significant heterogeneity noted among the studies. To facilitate comparisons among studies, we standardized the measurements of blood lipids and glucose levels, with all values expressed in milligrams per deciliter (mg/dL). For studies reporting cholesterol levels (TC, HDL-C, or LDL-C) in mmol/L, we applied a conversion factor of 38.67 to obtain mg/dL. Similarly, triglyceride values reported in mmol/L were converted to mg/dL using a factor of 88.55, and blood glucose levels were standardized to mg/dL by applying a factor of 18. This approach ensured consistent interpretation of estimates across different studies[9,10].

Subgroup analyses were conducted based on participant characteristics, including age group (< 65 or ≥ 65 years old), baseline BMI levels (< 30 kg/m² or ≥ 30 kg/m²), sex (male predominant or female predominant), blood pressure (hypertension or non-hypertension) and blood glucose measures (fasting blood glucose level (> 110 or ≤ 110 mg/dL), glycated hemoglobin A1c (HbA1c) (≥ 6.5% or < 6.5%), diabetes mellitus (DM) complication (DM combined or DM uncombined)); and treatment characteristics, including treatment type (monotherapy or combination therapy), the use of diet, lifestyle, medicine, surgery, and exercise (user or non-users). Hypertension was defined as a baseline mean systolic blood pressure ≥ 140 mmHg and/or a mean diastolic blood pressure ≥ 90 mmHg among study participants. In this meta-analysis, monotherapy refers to the use of a single intervention, whereas combination therapy involves two or more concurrent treatment strategies. Given the focus on overweight and obese populations, subsequent subgroup analyses were stratified by intervention type, including dietary interventions, lifestyle modifications, pharmacologic treatments, bariatric surgery, and structured exercise programs. Changes in blood lipid levels (∆ values) were defined as the difference between pre- and post-intervention values in the intervention group. To reduce the risk of false-positive results due to multiple subgroup comparisons, P values were adjusted using the Benjamini-Hochberg procedure. Adjusted P values (P_adj) are reported for all subgroup analyses.

To compare and analyze the baseline data of TG, TC, HDL-C, and LDL-C in predefined subgroups, the Wilcoxon test was used based on the non-normal distribution of the data. Linearity was assessed through visual inspection of residual-versus-fitted value plots. Subsequently, Spearman correlation and linear regression analyses were performed to evaluate associations. Statistical analyses were conducted in STATA 13.0 and SPSS 26.0. A P value of < 0.05 was considered statistically significant.

RESULTS

We identified a total of 17457 citations, of which 866 studies involving 3,300 participants were included in our subsequent analysis (Figure 1). The quality assessment conducted using the Cochrane risk-of-bias tool is presented in Supplementary Table 1. The baseline characteristics of the included studies are summarized in Supplementary Table 2. Egger’s funnel plots showed potential publication bias in the analyses of baseline blood lipid levels (Supplementary Figure 1). Our research encompasses trials from nearly 60 countries and regions worldwide, with 213 studies originating from the United States, which accounts for the highest proportion at 24.6% (Figure 2).

Figure 1 Flow diagram showing selection process.

Figure 2 Global distribution by country/region of included trial.

Trends in baseline TG, TC, HDL-C, and LDL-C

A significant decline in baseline TG levels was observed over time (Rs = -0.704, P < 0.001, I² = 98.6%; Figure 3A). The mean baseline TG concentrations demonstrated a pronounced acceleration in the downward trend prior to 2010, followed by a slower decline thereafter, with a peak observed in 1993 and a nadir in 2018. As for the distribution pattern of TG, the proportion of trials in which participants in categories of TG ≥ 200 mg/dL dropped substantially from 66.7% in 1992 to 7.0% in 2024 (Figure 4A).

Figure 3 Trends of baseline blood lipid among patients with overweight or obesity in randomized placebo-controlled trials from 1990 to 2024. A-D: Trends of triglycerides (A), total cholesterol (B), high-density lipoprotein cholesterol (C), and low-density lipoprotein cholesterol (D). TG: Triglycerides; TC: Total cholesterol; HDL-C: High-density lipoprotein cholesterol; LDL-C: Low-density lipoprotein cholesterol.

Figure 4 Trial proportion of different baseline triglycerides, total cholesterol, high-density lipoprotein cholesterol, and low-density lipoprotein cholesterol categories for patients with overweight or obesity in randomized placebo-controlled trials from 1990 to 2024. A-D: Trial proportions of different baseline triglycerides (A), total cholesterol (B), high-density lipoprotein cholesterol (C), and low-density lipoprotein cholesterol (D) categories.

Concurrently, a marked reduction in baseline TC was demonstrated during 1990-2024 (Rs = -0.884, P < 0.001, I² = 99.6%; Figure 3B). Mean TC values exhibited a steady decline from 235 mg/dL in the 1990s to 192 mg/dL by 2024. Regarding distribution patterns, the percentage of studies with participants classified in the low-risk TC category (< 200 mg/dL) increased from 0.0% in 1990 to 67.4% in 2024 (Figure 4B). The proportion of patients with baseline TC exceeding 240 mg/dL remained below 10%, except for 33.3% in 1992 and 50.0% in 1997 (Figure 4B).

In contrast to other lipid parameters, HDL-C levels showed a progressive increase over time (Rs = 0.336, P = 0.041, I² = 99.2%; Figure 3C). Baseline HDL-C levels exhibited greater fluctuations during the 1990s, which can be attributed to the limited number of studies conducted during that period. In contrast, since 2005, HDL-C levels have stabilized at higher values, exceeding 45 mg/dL. Meanwhile, patients with HDL-C below 40 mg/dL remained less than 15.0% in most study years (Figure 4C).

Similarly, baseline LDL-C levels decreased substantially (Rs = -0.808, P < 0.001, I² = 96.8%), dropping by approximately 8.1 mg/dL per decade in patients with overweight or obesity worldwide (Figure 3D). Moreover, patients classified as LDL-C more than 130 mg/dL declined from 50% in 1998 to less than 20% in 2024 (Figure 4D). Supplementary Figure 2 presents the trial number range from 1990 to 2024 of different TG, TC, HDL-C, and LDL-C categories among patients with overweight or obesity in placebo-controlled RCTs.

Trends in baseline TG, TC, HDL-C, and LDL-C categorized by participants’ characteristics

Age: Non-elderly patients (< 65 years) exhibited a marked inverse association between baseline TG levels and study year, whereas elderly patients showed no significant temporal trend (nonelderly individuals: Rs = -0.575, P_adj = 0.002, I² = 98.7%; elderly individuals: Rs = -0.121, P_adj = 0.694, I² = 96.0%, Figure 5A). Notably, inter-subgroup comparisons demonstrated no statistically significant difference in mean TG levels (Z = -0.130, P = 0.897, Supplementary Table 3).

Figure 5 Trends in baseline triglycerides, total cholesterol, high-density lipoprotein cholesterol, and low-density lipoprotein cholesterol categorized by participants’ characteristics. A-D: Baseline triglycerides (TG; A), total cholesterol (TC; B), high-density lipoprotein cholesterol (HDL-C; C), and low-density lipoprotein cholesterol (LDL-C; D) trends categorized by age < 65 years and ≥ 65 years; E-H: Baseline TG (E), TC (F), HDL-C (G), and LDL-C (H) trends categorized by body mass index < 30 kg/m² and ≥ 30 kg/m²; I-L: Baseline TG (I), TC (J), LDL-C (K), HDL-C (L) and trends categorized by male prominent and female prominent; M-P: Baseline TG (M), TC (N), LDL-C (O), and HDL-C (P) trends categorized by hypertension and non-hypertension; Q-T: Baseline TG (Q), TC (R), HDL-C (S), and LDL-C (T) trends categorized by fasting blood glucose (FBG) > 110 mg/dL and FBG ≤ 110 mg/dL; U-X: Baseline TG (U), TC (V), HDL-C (W), and LDL-C (X) trends categorized by HbA1c ≥ 6.5% and HbA1c < 6.5%; Y-AB: Baseline TG (Y), TC (Z), HDL-C (AA), and LDL-C (AB) trends categorized by DM combined and uncombined. TG: Triglycerides; TC: Total cholesterol; HDL-C: High-density lipoprotein cholesterol; LDL-C: Low-density lipoprotein cholesterol; DM: Diabetes mellites; BMI: Body mass index; FBG: Fasting blood glucose, HbA1c: Glycated hemoglobin A1c.

This age-dependent divergence was also observed in TC. Non-elderly individuals displayed a robust negative correlation between TC and study year (nonelderly individuals: Rs = -0.856, P_adj = 0.002, I² = 99.6%; elderly individuals: Rs = -0.323, P_adj = 0.260, I² = 96.5%, Figure 5B). Similarly, mean baseline TC levels remained comparable across subgroups (Z = -0.411, P = 0.681, Supplementary Table 3).

In contrast, HDL-C levels showed no discernible temporal association in either age cohort (non-elderly individuals: Rs = 0.268, P_adj = 0.250, I² = 99.2%; elderly individuals: Rs = -0.022, P_adj = 0.940, I² = 98.2%, Figure 5C). Furthermore, there was no significant difference identified in the average HDL-C levels between the two subgroups (Z = -0.300, P = 0.764, Supplementary Table 3).

In non-elderly patients, a negative relationship was observed between the average baseline LDL-C levels and the year of the study (nonelderly individuals: Rs = -0.755, P_adj = 0.002, I² = 99.0%; elderly individuals: Rs = -0.231, P_adj = 0.448, I² = 98.2%, Figure 5D). Subgroup comparisons again revealed no significant LDL-C disparity (Z = -1.011, P = 0.312, Supplementary Table 3).

BMI: Significant longitudinal trends in baseline lipid profiles were observed among participants with BMI ≥ 30 kg/m² during the 1990-2024 study period. Specifically, TG, TC, and LDL-C levels exhibited statistically significant decreasing trends, while HDL-C did not demonstrate a significant association (TG: Rs = -0.711, P_adj = 0.002, I² = 98.7%; TC: Rs = -0.857,P_adj = 0.002, I² = 99.6%; HDL-C: Rs = 0.394, P_adj = 0.05, I² = 99.2%; LDL-C: Rs = -0.812, P_adj = 0.002, I² = 98.0%). In comparison, participants with BMI < 30 kg/m² showed a significant negative temporal trend only for TC levels, with non-significant correlations observed for other lipid parameters (TG: Rs = -0.171, P_adj = 0.383, I² = 97.9%; TC: Rs = -0.513, P_adj = 0.006, I² = 98.2%; HDL-C: Rs = 0.039, P_adj = 0.852, I² = 99.0%; LDL-C: Rs = -0.329, P_adj = 0.135, I² = 97.9%, Figure 5E-H). Comparative analysis revealed consistently higher baseline TG levels in the BMI ≥ 30 kg/m² patients compared to their lower-BMI counterparts (Z = -3.222, P = 0.001, Supplementary Table 3).

Sex: We observed a reduction in the trends of TG over the years in studies predominantly involving males and females (male predominant: Rs = -0.472, P_adj = 0.011, I² = 98.5%; female predominant: Rs = -0.500, P_adj = 0.006, I² = 98.2%, Figure 5I). There was no significant difference in the magnitude of TG level between males and females (Z = -1.942, P = 0.052, Supplementary Table 3).

In studies predominantly involving females, the overall trends of TC and LDL-C showed a decrease over time (TC: Rs = -0.874, P_adj = 0.002, I² = 98.4%; LDL-C: Rs = -0.696, P_adj = 0.002, I² = 99.1%). However, a sex-stratified analysis did not reveal a notable trend in TC and LDL-C levels in male-predominant studies (TC: Rs = -0.338, P_adj = 0.078, I² = 99.8%; LDL-C: Rs = -0.340, P_adj = 0.089, I² = 98.0%, Figure 5J and K). No significant differences were found in TC and LDL-C levels between these two groups (TC: Z = -1.812, P_adj = 0.070; LDL-C: Z = -0.449, P_adj = 0.653, Supplementary Table 3).

Regarding HDL-C, we observed positive correlations between HDL-C levels and time in male predominant studies (male predominant: Rs = 0.565, P_adj = 0.004, I² = 98.9%; female predominant: Rs = 0.344, P_adj = 0.050, I² = 99.2%, Figure 5L). Additionally, the average HDL-C level for females was markedly greater than that of males (Z = -2.736, P = 0.006, Supplementary Table 3).

Blood pressure: Among individuals with hypertension and non-hypertension, significant inverse correlations over time were observed for TG (hypertension: Rs = -0.345, P_adj = 0.027, I² = 99.1%; non-hypertension: Rs = -0.154, P_adj = 0.002, I² = 98.7%), TC (hypertension: Rs = -0.513, P_adj = 0.002, I² = 99.2%; non-hypertension: Rs = -0.162, P_adj = 0.002, I² = 99.7%), and LDL-C (hypertension: Rs = -0.401, P_adj = 0.026, I² = 98.9%; non-hypertension: Rs = -0.144, P_adj = 0.030, I² = 99.1%), whereas HDL-C trends remained statistically nonsignificant (hypertension: Rs = -0.142, P_adj = 0.740, I² = 97.3%; non-hypertension: Rs = -0.015, P_adj = 0.747, I² = 99.5%; Figure 5M-P). Between-strata comparisons demonstrated a significant difference in TG trajectory (Z = -2.232, P = 0.026, Supplementary Table 3).

Blood glucose: Fasting blood glucose level: In patients with elevated fasting blood glucose (> 110 mg/dL), a markedly negative temporal association was detected for TG levels (Rs = -2.02, P_adj = 0.004, I² = 98.7%), alongside significant declines in TC (Rs = -0.351, P_adj = 0.002, I² = 97.2%) and LDL-C (Rs = -0.137, P_adj = 0.042, I² = 99.5%), although the HDL-C trend remained statistically insignificant (Rs = 0.097, P_adj = 0.272, I² = 98.8%). Among subgroup of fasting blood glucose ≤ 110 mg/dL, lipid patterns were still showed statistically significant reductions in TG (Rs = -0.093, P_adj = 0.022, I² = 98.6%), TC (Rs = -0.133, P_adj = 0.001, I² = 99.7%), and LDL-C (Rs = -0.122, P_adj = 0.006, I² = 97.4%), with no significant HDL-C trend (Rs = 0.054, P_adj = 0.186, I² = 99.3%; Figure 5Q-T). Between-group comparisons revealed a more pronounced downward trajectory of TG levels in individuals with hyperglycemia (Z = –2.540, P = 0.011, Supplementary Table 3).

HbA1c: In individuals with HbA1c ≥ 6.5%, significantly decreasing trends were observed for TG (Rs = -0.359, P_adj = 0.002, I² = 99.3%), TC (Rs = -0.43, P_adj = 0.002, I² = 96.6%), and LDL-C (Rs = -0.319, P_adj = 0.006, I² = 99.7%), while HDL-C showed a moderate yet statistically significant upward association (Rs = 0.303, P_adj = 0.008, I² = 98.9%). In contrast, participants with HbA1c < 6.5% did not exhibit statistically significant temporal trends in any lipid parameter (TG: Rs = -0.032,P_adj = 0.700, I² = 98.6%; TC: Rs = -0.122, P_adj = 0.142, I² = 98.4%; HDL-C: Rs = -0.096, P_adj = 0.249, I² = 99.1%; LDL-C: Rs = -0.126, P_adj = 0.129, I² = 98.8%; Figure 5U-X). Similarly, the average TG level for the HbA1c ≥ 6.5% subgroup was markedly greater than that of the HbA1c < 6.5% subgroup (Z = -4.667, P < 0.001, Supplementary Table 3).

Combined with DM: Notable between-group differences emerged in lipid trends when stratified by DM status. Patients with comorbid DM demonstrated statistically significant negative temporal trends for TG, TC, and LDL-C, accompanied by a robust positive HDL-C association (TG: Rs = -0.581, P_adj = 0.002, I² = 99.3%; TC: Rs = -0.563, P_adj = 0.004, I² = 98.3%; HDL-C: Rs = 0.608, P_adj = 0.002, I² = 98.6%; LDL-C: Rs = -0.588, P_adj = 0.004, I² = 99.7%). In diabetic-free participants, significant downward trends persisted for TC and LDL-C, while TG and HDL-C showed non-significant associations with temporal progression (TG: Rs = -0.315, P_adj = 0.117, I² = 98.4%; TC: Rs = -0.422, P_adj = 0.032, I² = 99.8%; HDL-C: Rs = 0.060, P_adj = 0.781, I² = 99.4%; LDL-C: Rs = -0.558, P_adj = 0.006, I² = 97.9%, Figure 5Y-AB). Significant between-strata differences were specifically observed for TG trajectories (Z = -5.907, P < 0.001), whereas TC (Z = -0.801, P = 0.423), HDL-C (Z = -1.425, P = 0.154), and LDL-C (Z = -0.738, P = 0.460) patterns remained statistically indistinguishable between groups (Supplementary Table 3).

Trends in baseline TG, TC, HDL-C, and LDL-C categorized by therapeutic regimen

Monotherapy vs combination therapy: We observed a downward trend in baseline TG, TC, and LDL-C levels and upward trends in baseline HDL-C levels with study year in monotherapy (TG: Rs = -0.826, P_adj = 0.002, I² = 98.9%; TC: Rs = -0.833, P_adj = 0.002, I² = 99.7%; HDL-C: Rs = 0.626, P_adj = 0.002, I² = 99.3%; LDL-C: Rs = -0.688, P_adj = 0.002, I² = 99.3%). No correlation was observed between baseline blood lipid levels and time in combination therapy (TG: Rs = -0.102, P_adj = 0.651, I² = 98.5%; TC: Rs = -0.271, P_adj = 0.222, I² = 97.9%; HDL-C: Rs = -0.102, P_adj = 0.651, I² = 96.0%; LDL-C: Rs = -0.012, P_adj = 0.960, I² = 99.0%, Figure 6). Furthermore, no significant difference in baseline lipid levels was detected between the two groups (Supplementary Table 3).

Figure 6 Trends in baseline triglycerides, total cholesterol, high-density lipoprotein cholesterol, and low-density lipoprotein cholesterol categorized by monotherapy and combination therapy. A-D: Baseline triglycerides (A), total cholesterol (B), high-density lipoprotein cholesterol (C), and low-density lipoprotein cholesterol (D) trends categorized by monotherapy and combination therapy. TG: Triglycerides; TC: Total cholesterol; HDL-C: High-density lipoprotein cholesterol; LDL-C: Low-density lipoprotein cholesterol.

Subgroup analysis of therapeutic methods: The mean baseline blood lipid levels (including TG, TC, and LDL-C) showed a significant correlation with study year in patients without diet intervention, while HDL-C did not show a significant correlation (TG: Rs = -0.604, P_adj = 0.002, I² = 98.6%; TC: Rs = -0.836, P_adj = 0.002, I² = 99.3%; HDL-C: Rs = 0.387, P_adj = 0.052, I² = 97.2%; LDL-C: Rs = -0.850, P_adj = 0.002, I² = 99.2%). Furthermore, a negative correlation between the mean baseline levels of TG, TC, and LDL-C and the study year was observed in patients who underwent dietary intervention (TG: Rs = -0.512,P_adj = 0.003, I² = 99.2%; TC: Rs = -0.703, P_adj = 0.003, I² = 98.5%; HDL-C: Rs = 0.177, P_adj = 0.340, I² = 99.0%; LDL-C: Rs = -0.376, P_adj = 0.049, I² = 98.9%, Supplementary Figure 3A-D).

As for exercise, significant temporal trends in baseline lipid profiles were identified among non-exercising patients. Specifically, TG, TC, and LDL-C levels exhibited strong inverse correlations with study year (TG: Rs = -0.739, P_adj = 0.002, I² = 99.8%; TC: Rs = -0.864, P_adj = 0.002, I² = 99.2%; LDL-C: Rs = -0.833, P_adj = 0.002, I² = 98.3%), while HDL-C demonstrated a positive temporal association (Rs = 0.482, P_adj = 0.008, I² = 98.8%). Notably, patients receiving exercise intervention displayed attenuated temporal correlations for blood lipid and time, which were no longer statistically significant (TG: Rs = 0.331, P_adj = 0.098; I² = 99.2%; TC: Rs = -0.348, P_adj = 0.088; I² = 98.5%; HDL-C: Rs = -0.078, P_adj = 0.709; I² = 98.6%; LDL-C: Rs = -0.357, P_adj = 0.086; I² = 99.3%; Supplementary Figure 3E-H).

Similar trends of baseline mean blood lipid profiles with study year were observed in patients with lifestyle interventions (TG: Rs = -0.357, P_adj = 0.191; I² = 98.8%; TC: Rs = -0.284, P_adj = 0.286; I² = 98.5%; HDL-C: Rs = 0.035, P_adj = 0.897; I² = 99.2%; LDL-C: Rs = -0.153, P_adj = 0.572; I² = 98.7%) and without lifestyle interventions like exercise. Except for HDL-C, which was not significant in patients without lifestyle interventions (TG: Rs = -0.711, P_adj = 0.002; I² = 98.6%; TC: Rs = -0.846,P_adj = 0.002; I² = 97.8%; HDL-C: Rs = 0.341, P_adj = 0.096; I² = 96.9%; LDL-C: Rs = -0.819, P_adj = 0.002; I² = 97.2%, Supplementary Figure 3I-L).

Significant temporal correlations were observed in baseline TG, TC, HDL-C, and LDL-C levels in medicine intervention patients (TG: Rs = -0.449, P_adj = 0.011; I² = 98.9%; TC: Rs = -0.650, P_adj = 0.001; I² = 99.4%; HDL-C: Rs = 0.650, P_adj = 0.002; I² = 98.6%; LDL-C: Rs = -0.417, P_adj = 0.031; I² = 98.0%). In non-medicine intervention patients, baseline TG, TC, and LDL-C exhibited negative correlations with study year (TG: Rs = -0.480, P_adj = 0.010, I² = 97.6%; TC: Rs = -0.795, P_adj = 0.001, I² = 97.8%; LDL-C: Rs = -0.743, P_adj = 0.002, I² = 98.8%), while HDL-C demonstrated no temporal association in this group (Rs = 0.022, P_adj = 0.904, I² = 98.5%; Supplementary Figure 3M-P).

Regarding surgery, a significant negative correlation was observed in baseline mean TG, TC, and LDL-C levels in participants without surgical intervention, except for HDL-C, which not indicated a significant correlation (TG: Rs = -0.720,P_adj = 0.002; I² = 98.6%; TC: Rs = -0.878, P_adj = 0.002; I² = 99.3%; HDL-C: Rs = 0.343, P_adj = 0.094; I² = 98.1%; LDL-C: Rs = -0.801, P_adj = 0.002; I² = 96.4%). In contrast, all the blood lipid profiles showed no significant association with the study year in surgical populations. (TG: Rs = -0.307, P_adj = 0.265; I² = 98.3%; TC: Rs = 0.214, P_adj = 0.443; I² = 98.8%; HDL-C: Rs = 0.161, P_adj = 0.567; I² = 99.0%; LDL-C: Rs = 0.064, P_adj = 0.829; I² = 99.1%, Supplementary Figure 3Q-T).

The compared subgroup analysis of different therapeutic methods is detailed in Supplementary Table 3.

Trends in ∆TG, ∆TC, ∆HDL-C, and ∆LDL-C categorized by therapeutic regimen

The change in value of blood lipid (including ∆TG, ∆TC, ∆HDL-C, and ∆LDL-C) did not present a significant correlation with time in different subgroups of the therapeutic regimen. Compared to the subgroup without surgery, the patients with surgery exhibited a greater reduction in ∆TG, ∆TC, and ∆LDL-C, as well as a more significant increase in ∆HDL-C (TG: Z = -6.301, P < 0.001; TC: Z = -5.322, P < 0.001; LDL-C: Z = -5.287, P < 0.001; HDL-C: Z = -4.804, P < 0.001). Moreover, ∆TG, ∆TC, and ∆HDL-C levels in patients with diet intervention were much lower than in patients without diet intervention (TG: Z = -3.143, P = 0.002; TC: Z = -2.756, P = 0.006; HDL-C: Z = -3.474, P = 0.001). Detailed results of subgroup analysis were reported in Supplementary Table 4.

DISCUSSION

In our analysis, among participants with overweight or obesity in placebo-controlled randomized trials, baseline blood lipid levels have improved over the past 34 years (1990–2024). Specifically, baseline mean TG, TC, and LDL-C levels have decreased, while HDL-C levels have increased. These findings indicate that blood lipid control has improved in patients with obesity despite the emergence and development of the obesity pandemic[11].

Individuals with obesity exhibit significantly elevated risks of cardiovascular disease[12], diabetes[13], cancer[14], as well as an increased risk of all-cause mortality compared to the general population[15]. Dysregulation of lipid metabolism is multifactorial, linked to genetic predisposition, hypertension, insulin resistance, chronic inflammatory states, dietary imbalance (e.g., high intake of saturated fatty acids), and sedentary lifestyle[16]. While the pathophysiological interactions remain complex, scientific consensus confirms that obesity can significantly elevate LDL-C levels through multiple pathophysiological pathways (including but not limited to the spillover effect of free fatty acids induced by visceral fat accumulation[17,18], increased synthesis of very low-density lipoprotein in the liver[19], and inhibition of lipoprotein lipase activity[20]), while simultaneously reducing HDL-C concentration and its reverse cholesterol transport function[21,22]. Current evidence offers limited insight into whether weight loss or LDL-C reduction is more effective in mitigating cardiovascular risk. However, combined intensive lipid-control and weight-control strategies in patients with obesity have been shown to lower cardiovascular risk substantially[23]. Based on the results of the Look AHEAD study, moderate weight loss (5%-10%) can improve cardiovascular risk factors, with greater cardiovascular benefits associated with higher degrees of weight loss, specifically manifested as reductions in HbA1C and blood pressure. Furthermore, when weight loss exceeds 15%, blood TG levels decrease by more than 60 mg/dL, while blood HDL-C levels increase by 4-5 mg/dL; however, there is no statistically significant change in LDL-C observed[24]. Moreover, considering the limited LDL-C reduction achieved by most obesity treatments and the cumulative impact of dyslipidemia on cardiovascular risk, aggressive lipid management is of significant clinical importance[25]. This may, in part, explain the overall improvements in blood lipid profiles.

According to data from the Non-communicable Disease Risk Factor Collaboration, the global age-standardized mean levels of TC and non-HDL-C exhibited little change between 1980 and 2018. Specifically, among the 200 countries and regions included in the study, the median age-standardized mean non-HDL-C levels in men increased slightly from 3.36 mmol/L in 1980 to 3.37 mmol/L in 2018, while in women, it decreased slightly from 3.44 mmol/L in 1980 to 3.34 mmol/L in 2018[26]. Notably, our analysis shows that nearly half of the data originates from the United States, with African randomized placebo-controlled trials contributing less than 1% of the total data. This suggests that our results are more representative of the health management standards in developed regions. To facilitate more accurate conclusions, systematic collection of additional data from underrepresented regions remains imperative.

Currently, there is a lack of globally unified epidemiological data regarding lipid levels and their distribution characteristics in overweight or obese populations. In addition to documenting global trends in the average lipid levels among overweight or obese patients, our findings also illustrate the unique distribution of all lipid categories within this group. Over the past 30 years, in placebo-controlled randomized trials, the proportion of overweight or obese patients with LDL-C levels exceeding 160 mg/dL has ranged from 20% to 40%, while the proportion with HDL-C levels below 40 mg/dL has consistently remained around 10%. This provides a foundation for future clinical research tailored to specific population characteristics.

Our analysis revealed no significant differences in baseline lipid trends between younger and older participants. Notably, the findings indicate that non-elderly patients exhibit a notable inverse relationship between baseline TG, TC, LDL-C, and study year. In contrast, elderly patients did not present significant temporal trends for TG, TC, HDL-C, and LDL-C, highlighting a possible age-related resistance to changes in lipid profiles over time[27].

According to existing research, obesity management has shifted from traditional, generalized approaches to more individualized treatment strategies[28]. However, current weight loss interventions for individuals with obesity remain relatively limited, with a predominant reliance on lifestyle modifications[29]. In the included studies, fewer than 10% employed pharmacological interventions. Nevertheless, these agents demonstrated significant improvements in lipid profiles[30]. Our analysis of randomized placebo-controlled studies involving pharmacological interventions demonstrated that the baseline average levels of TG, TC, and LDL-C in overweight or obese populations decreased significantly over time, while HDL-C levels increased notably. This suggests that, in addition to lifestyle modifications, the appropriate application of novel drugs capable of significantly enhancing obesity management should be encouraged based on individual assessments.

Nevertheless, our research does have certain limitations. First, the individuals enrolled in the included RCTs were drawn from relatively specific populations under strict eligibility criteria, which limits the external validity of baseline findings. Although we excluded trials with lipid-level-based inclusion criteria, the baseline lipid profiles of RCT participants may not be generalizable to the broader real-world population. This distinction is critical, as trends in RCT-derived baseline values may reflect shifts in trial design philosophies or population characteristics rather than true epidemiological changes in the general population. Second, Egger’s test suggested a possible publication bias in the evaluation of baseline blood lipid. Therefore, the findings must be considered carefully. Third, there was significant variation in the publication years, study designs, treatment regimens, and research methodologies of the studies included, which resulted in considerable heterogeneity among them. While a random-effects model was employed for the analysis, and subgroup analyses were performed to investigate the origins of the pronounced heterogeneity, the substantial level of heterogeneity in this meta-analysis remains a limitation. This heterogeneity may compromise the precision and robustness of the pooled estimates, suggesting the results should be interpreted with certain settings. Fourth, there was a marked geographic imbalance in the distribution of included studies, with nearly half of the data derived from the United States and less than 1% from Africa. This uneven geographic representation may introduce region-specific biases and limit the generalizability of our findings to global populations, particularly to low- and middle-income countries. Consequently, the observed trends may better reflect patterns in developed nations or regions, and caution is warranted when extrapolating these results to populations in other regions. Future research efforts should aim to include more data from underrepresented regions to enhance global applicability. Fifth, only a limited number of the studies included reported clinical endpoint events, such as all-cause mortality or major adverse cardiovascular events. Due to the scarcity of such data, we were unable to perform a quantitative analysis on these outcomes. Future studies that report clinical endpoints will offer profiles that facilitate comprehensive evaluations.

CONCLUSION

Our findings indicate a significant decline in baseline TG, TC, and LDL-C levels over time, coupled with an increase in HDL-C levels. These trends reflect the positive impact of public health initiatives and effective treatment strategies aimed at managing dyslipidemia.