The effect of oleic acid enriched diets on glucose and lipid metabolism: a systematic review and meta-analysis

Abstract

Background

High-oleic diets (HODs), have been proposed to improve lipid and glucose metabolism, yet clinical evidence remains inconsistent. This systematic review and meta-analysis aimed to quantitatively evaluate the effects of HODs compared with low-oleic diets (LODs) on lipid profiles, apolipoproteins, and glucose homeostasis in adults.

Methods

Following PRISMA 2020 guidelines, PubMed, Web of Science, Scopus, and Embase were searched from inception to September 2025. Randomized controlled trials (RCTs) in adults were included if the intervention provided ≥ 70% oleic acid of total fatty acids and differed from control diets by ≥ 5% points in oleic acid content. Data were pooled using random-effects models to estimate weighted mean differences (WMDs) with 95% confidence intervals (CIs).

Results

Twenty-six RCTs (n = 1,244 participants) met the inclusion criteria. Compared with LODs, HODs significantly reduced total cholesterol (WMD: −0.13 mmol/L; 95% CI: −0.24 to − 0.01) and low-density lipoprotein cholesterol (LDL-C) (WMD: −0.11 mmol/L; 95% CI: −0.20 to − 0.01) and modestly increased Apo A1 (WMD: 0.02 mmol/L; 95% CI: 0.00 to 0.03). No significant changes were observed for high-density lipoprotein cholesterol (HDL-C), triglycerides, very low-density lipoprotein cholesterol (VLDL-C), Apo B, and glucose metabolism. Subgroup and meta-regression analyses indicated stronger lipid-lowering effects when ≥ 80% of total fatty acids were oleic acid. Also, low heterogeneity was reported for most variables.

Conclusion

High-oleic dietary interventions modestly improve circulating lipid profiles, particularly total and LDL cholesterol, without adverse effects on glucose metabolism. The proportion of oleic acid intake appears to be a key determinant of metabolic benefit.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12986-026-01110-7.

Introduction

Cardiometabolic disorders, including type 2 diabetes mellitus (T2DM), dyslipidaemia, and metabolic syndrome, continue to impose significant health burdens across the globe [1, 2]. Despite advances in pharmacologic therapies, dietary modification remains a cornerstone of risk-reduction strategies [3]. In this context, the quality, not just the quantity, of dietary fat has emerged as a pivotal determinant of metabolic health [4, 5]. More specifically, monounsaturated fatty acids (MUFAs), and particularly oleic acid (18:1 n-9), have attracted increasing interest for their potential beneficial effects on both lipid and glucose metabolism [6–8].

There is growing epidemiologic and clinical evidence to suggest that oleic acid rich diets or high oleic diets (HODs), either through olive oil, high-oleic vegetable oils, or enriched oils, may improve lipid profiles, reduce atherogenic lipoprotein particles, and favourably modulate insulin sensitivity [9, 10]. For example, a systematic review of high-oleic vegetable oil substitutions reported consistent reductions in total cholesterol (TC) and low-density lipoprotein cholesterol (LDL-C) when saturated or trans fats were replaced by high-oleic oils [11, 12]. In a randomized trial comparing conventional and high-oleic canola oils, improvements in lipid and lipoprotein outcomes were documented [11, 13]. While much of the early focus was on lipids and apolipoproteins, more recent mechanistic data indicate that oleic acid may directly influence glucose homeostasis. For instance, oleic acid has been shown in adipocyte culture to enhance insulin-receptor signalling (via the PI3K/Akt axis) and thereby potentiate insulin-stimulated glucose uptake [14]. Moreover, in pancreatic β-cell models, oleic acid modulated mitochondrial energy metabolism and increased glucose sensitivity, suggesting a plausible mechanism for improved glycaemic control [15]. Even in human observational studies, a higher oleic acid: stearic acid ratio was associated with markers of insulin resistance (e.g., triglyceride/HDL-C, fasting glucose–triglyceride index) in non-obese subjects [16].

Despite this growing body of evidence, important gaps remain. Many of the existing trials vary widely in the proportion of oleic acid provided, baseline characteristics of participants, underlying diets, and duration of intervention, ultimately resulting in conflicting effects reported across studies and significant heterogeneity [17–19]. Furthermore, no meta-analysis has been conducted with this methodology and scope to date.

In light of these gaps, the present meta-analysis aims to provide a rigorous quantitative synthesis of randomized controlled trials comparing HODs versus lower-oleic dietary interventions and to examine their effects not only on lipid profiles and apolipoproteins, but also on glucose metabolism indices. By clarifying not only whether HODs are beneficial but for whom and under what conditions, this work may advance precision nutrition strategies and inform dietary guidelines for cardiometabolic risk reduction.

Methods

Protocol and reporting

This systematic review and meta-analysis was conducted and reported in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA 2020) statement [20]. The study protocol has been previously registered with PROSPERO database (registration number: ID 1181581).

Search strategy

We performed a comprehensive literature search of PubMed/MEDLINE, Embase, Web of Science, and Scopus from database inception to September, 2025, without language or publication-date restrictions. In addition, gray literature sources were systematically searched, including ClinicalTrials.gov and the WHO International Clinical Trials Registry Platform (ICTRP), to identify ongoing or unpublished trials. Conference abstracts and theses were screened when accessible. Gray literature records were included only if sufficient quantitative data were available and all predefined eligibility criteria were met. Records lacking complete outcome data or detailed fatty acid composition were excluded from quantitative synthesis. The electronic search strategy combined controlled vocabulary (MeSH and Emtree) and free-text terms for oleic acid and relevant outcomes. Example search terms included: (“oleic acid” OR “high-oleic” OR “monounsaturated fatty acid” OR “MUFA” OR “high-oleic sunflower oil” OR “high-oleic canola” OR “olive oil”) AND (“randomized controlled trial” OR “randomised” OR “cross-over” OR “clinical trial”) AND (“cholesterol” OR “LDL” OR “HDL” OR “triglyceride” OR “apolipoprotein” OR “fasting glucose” OR “insulin” OR “HOMA-IR”).

Study selection and screening

After removal of duplicates, two reviewers (Reviewer A and Reviewer B) independently screened titles and abstracts for potentially relevant articles. Full texts of potentially eligible records were retrieved and assessed independently by the same reviewers against pre-specified inclusion and exclusion criteria. Disagreements were resolved by consensus or by consultation with a third reviewer (Reviewer C). We recorded reasons for exclusion at the full-text stage and will present a PRISMA flow diagram showing numbers screened, excluded, and included.

Eligibility criteria (PICOS)

Studies were eligible if they met all of the following criteria:

• Population: Adults (≥ 18 years), any health status (healthy, overweight/obese, metabolic syndrome, type 2 diabetes, hyperlipidaemia).

• Intervention: A dietary intervention was classified as high-oleic exposure if it provided ≥ 70% of total fatty acids as oleic acid, corresponding approximately to ≥ 45–50% of total fat as MUFA or ≥ 15–20% of total energy from oleic acid (≈ 10 g/day or more). This threshold was defined a priori based on international compositional standards (e.g., FAO/WHO and Codex definitions of high-oleic oils) [21, 22], where oils containing ≥ 70% oleic acid are recognized as high-oleic varieties. The use of an absolute threshold ensured that the intervention arm represented a biologically meaningful dominance of oleic acid, rather than modest compositional differences. In addition, a minimum absolute between-group difference of ≥ 5% points in oleic acid content was required to ensure sufficient exposure contrast.

• Comparator: A diet with lower oleic acid content such that there was an absolute difference ≥ 5% points in oleic acid (C18:1) as % of total fatty acids between intervention and control. This dual criterion (absolute ≥ 70% threshold plus ≥ 5%-point contrast) was intended to avoid inclusion of trials with trivial compositional differences and to ensure meaningful exposure separation between arms.

• Outcomes: At least one of the prespecified metabolic outcomes reported: TC, LDL-C, High-density lipoprotein cholesterol (HDL-C), triglycerides (TG), very low-density lipoprotein cholesterol (VLDL-C), apolipoprotein A1 (Apo A1), apolipoprotein B (Apo B), fasting blood sugar (FBS), fasting insulin, or Homeostatic Model Assessment of Insulin Resistance (HOMA-IR).

• Study design: Randomized controlled trials (parallel or crossover). Minimum intervention duration of 1 weeks (shorter dietary exposures were excluded).

• Other: Studies providing sufficient numerical data (means and SDs, or change scores and SDs, or SE/CI from which SDs could be derived) were required.

Exclusion criteria: Only randomized controlled trials employing isoenergetic substitution designs were eligible. Intervention and control diets were required to be matched for total energy intake and overall fat percentage of total energy, ensuring that differences between groups were primarily attributable to fatty acid composition rather than total fat quantity. When reported, carbohydrate and protein distributions were also examined to confirm similar macronutrient profiles between arms. Studies involving additional dietary modifications (e.g., caloric restriction, fiber supplementation, weight-loss interventions, or concurrent medication changes) were excluded to minimize confounding. This approach allowed isolation of the specific effect of oleic acid enrichment on metabolic outcomes. Non-randomized studies, animal or in-vitro studies, trials combining the high-oleic intervention with other co-interventions that could not be isolated (e.g., simultaneous medication changes), studies without suitable control arms or without adequate fatty-acid composition data, duplicate reports (only the most complete/longest follow-up was retained), and conference abstracts with insufficient data after attempts to contact authors. Only studies in which intervention and control diets were designed to be isoenergetic and with comparable total fat content were included, ensuring that observed effects could be attributed primarily to differences in fatty acid composition rather than energy or fat quantity.

Data extraction

Two reviewers independently extracted data using a piloted standardized form. Extracted items included: study ID (first author, year), country, study design (parallel/crossover), sample size (total and per arm), participant characteristics (age, sex, baseline BMI, metabolic status), details of the intervention and control diets (oil source, % total fatty acids as oleic acid, total fat % of energy, energy intake), intervention duration (weeks), outcome measures (mean ± SD at baseline and endpoint or mean change ± SD), and whether intention-to-treat or per-protocol analyses were reported. When necessary data were missing, study authors were contacted twice by email; if no response was obtained, data were imputed as described below. Specifically, when within-subject standard deviations were not reported, crossover trials were analyzed as parallel-group studies without assuming a correlation coefficient, in order to avoid artificial reduction of variance and inflation of statistical significance.

Risk of bias assessment and certainty of evidence (GRADE)

Two reviewers independently assessed the risk of bias of each included RCT using the Cochrane Risk of Bias 2 (RoB 2) tool covering the domains [23]: randomization process, deviations from intended interventions, missing outcome data, measurement of the outcome, and selection of reported results. Each domain and the overall judgment were categorized as “low”, “some concerns”, or “high” risk of bias. Disagreements were resolved by discussion or third-party adjudication. We rated the overall certainty for each primary outcome using the GRADE approach considering risk of bias, inconsistency (heterogeneity), indirectness, imprecision, and publication bias [24]. Evidence was graded as high, moderate, low, or very low, with reasons for downgrading/upgrading documented.

Statistical analysis and data synthesis

All statistical analyses were performed using STATA version 12.0 (StataCorp, College Station, TX, USA). The primary outcomes of this meta-analysis were TC and LDL-C, as these biomarkers represent the most clinically relevant lipid endpoints and were central to the study hypothesis. Secondary outcomes included HDL-C, TG, VLDL-C, Apo A1, Apo B, FBS, fasting insulin, and HOMA-IR. For each outcome, weighted mean differences (WMDs) with corresponding 95% confidence intervals (CIs) were calculated. When SDs were not directly reported, they were derived from standard errors, confidence intervals, or p-values using established methods [25, 26]. In cases where only baseline and endpoint means were available, the SD of the change was estimated using the standard formula incorporating baseline and final SDs and an assumed correlation coefficient. An assumed within-subject correlation coefficient of r = 0.5 was used, in line with recommendations from the Cochrane Handbook. Sensitivity analyses using r = 0.25 and r = 0.75 were performed to evaluate the impact of this assumption on pooled estimates. Standard errors of the mean were converted to SD by multiplying by the square root of the sample size. A random-effects model (DerSimonian and Laird method) was used to pool effect sizes, accounting for between-study variability. Between-study heterogeneity was assessed using Cochran’s Q test and quantified with the I² statistic with thresholds interpreted as: 0–25% (low), 26–50% (moderate), 51–75% (substantial), and > 75% (considerable) [27]. Pre-specified subgroup analyses (based on a priori hypotheses) included: Intervention duration: ≤4 weeks vs. > 4 weeks, Baseline BMI: ≤25 kg/m² vs. > 25 kg/m², and Oleic acid proportion: <80% vs. ≥ 80% of total fatty acids. Univariate meta-regressions were run first; variables with p < 0.10 were entered into multivariable models when sufficient studies (rule of thumb: ≥10 studies per covariate) were available. Sensitivity analyses were performed using the leave-one-out approach, whereby the meta-analysis was repeated after sequential omission of each study. Potential publication bias was evaluated visually through funnel plots and formally using Egger’s regression test [28]. Where asymmetry was suggested, the trim-and-fill method was applied to estimate the effect of potentially missing studies on pooled results.

Results

Study selection

As shown in Fig. 1, a total of 3198 records were identified through database searching. After removing duplicates, 2414 records remained for title and abstract screening, of which 2356 were excluded for not meeting the eligibility criteria. Fifty-eight full-text articles were then assessed in detail, and 32 articles were excluded due to reasons such as insufficient data (n = 6), incompatible study design (n = 16), irrelevant outcomes (n = 7), absence of an appropriate control group (n = 1), or combining HODs with other interventions (n = 3). Ultimately, 26 RCTs were included in the final systematic review and meta-analysis.

Fig. 1.

Flow chart of study selection process

Characteristics of Included Studies (Based on Table 1)

Table 1.

Characteristics of eligible studies

First author et al.	Year	Country	Population	Mean Age year	Sex (Male %)	Sample Size Study		Follow up of intervention (Weeks)	Type of RCTs	Intervention group	Control group	Baseline of BMI (kg/m2)
						Intervention	Control
1. González‑Rámila	2023	Spain	Healthy subjects and at-risk (hypercholesterolemic) subjects.	44	56.6 or 38.2	64	64	4	Crossover	High-oleic acid sunflower oil (HOSO) (45 gr Rich in MUFA, with an oleic acid (C18:1n9) content of 76.5%)	Olive pomace oil (OPO) (45 gr Rich in MUFA, with an oleic acid (C18:1n9) content of 71.0%)	24
2. Loganathan	2022	Malaysia	Healthy volunteers	33.68	15	40	40	4	Crossover	Extra virgin olive oil (EVOO)(oleic acid (C18:1n9) content of 71%)	Cocoa butter (CB)(oleic acid (C18:1n9) content of 33%)	21.67
3. González‑Rámila	2022	Spain	Healthy subjects and at-risk (hypercholesterolemic) subjects.	30	25.8 or 51.3	68	68	4	Crossover	Olive pomace oil (OPO) (45 gr Rich in MUFA, with an oleic acid (C18:1n9) content of 74.32%)	Sunflower oil (SO) (45 gr Rich in MUFA, with an oleic acid (C18:1n9) content of 29.76%)	23
4. Baer	2021	USA	Healthy volunteers with moderately elevated LDLc	55.1	NR	48	50	4	Crossover	High-Oleic Soybean Oil(oleic acid (C18:1n9) content of 70.9%)	Soybean oil (SBO)(oleic acid (C18:1n9) content of 21.9%)	29.5
5. Bowen	2019	Canada	Adults with Central Adiposity	44	37	119	119	6	Crossover	High-oleic acid canola oil (HOCO)(oleic acid (C18:1n9) content of 70%)	Conventional canola oil(oleic acid (C18:1n9) content of 60%)	31.7
6. Liu	2018	Canada	Population with or at Risk for Metabolic Syndrome	49.5	50	101	101	4	Crossover	High-oleic acid canola oil (HOCO)(oleic acid (C18:1n9) content of 72%)	Conventional canola oil(oleic acid (C18:1n9) content of 62%)	29.4
7. Harris	2017	USA	Postmenopausal Women	58.8	0	20	20	4	Crossover	High-oleic acid sunflower oil (HOSO) (30 ml, oleic acid (C18:1n9) content of 80%)	Virgin Coconut Oil (30 ml, oleic acid (C18:1n9) content of 3%)	26.4
8. Pu	2016	Canada	Participants with cardiovascular disease risk	45.6	41.6	84	84	4	Crossover	High-oleic acid canola oil (HOCO)(60 gr oleic acid (C18:1n9) content of 72%)	Conventional canola oil(60 gr oleic acid (C18:1n9) content of 60%)	29.6
9. Jones	2015	Multicenter	Individuals with abdominal obesity	45.8	50.9	50	50	4	Crossover	High oleic canola oil(oleic acid (C18:1n9) content of 71.5%)	Blend of corn/safflower oil (oleic acid (C18:1n9) content of 17.6%)	30.4
10. Barbour	2015	Australia	Healthy subjects	65	47.5	61	61	12	Crossover	High Oleic Peanut(≈ 75–80% oleic acid of total fatty acids)	Regular peanuts (≈ 50% oleic acid).	31
11. Alves	2014	Brazil	Overweight/obese men	27.4	100	21	22	4	Parallel	High Oleic Peanut(≈ 81.5% oleic acid of total fatty acids)	Regular peanuts (≈ 51% oleic acid).	29.8
12. Jones	2014	Multicenter	Adult with at least one of the cardiovascular risk factors were	46.4	46.1	130	130	4	Crossover	High oleic canola oil(oleic acid (C18:1n9) content of 71.5%)	Regular canola oil(oleic acid (C18:1n9) content of 58.6%)	29.8
13. Hlais	2013	Lebanon	Healthy subjects	NR	100	20	22	12	Parallel	High-oleic acid sunflower oil (HOSO) (oleic acid (C18:1n9) content of 81.6%)	Fish oil (oleic acid (C18:1n9) content of 15–20%)	26.49
14. Gillingham	2011	Canada	Hypercholesterolemic subjects	47.49	36.1	36	36	4	Crossover	High-oleic acid canola oil (HOCO)(oleic acid (C18:1n9) content of 73.7%)	Typical Western diet (WD)(oleic acid (C18:1n9) content of 46.5%)	28.56
15. Tholstrup	2004	Denmark	Healthy subjects	23.4	100	17	17	3	Crossover	High-oleic acid sunflower oil (HOSO) (oleic acid (C18:1n9) content of 89.4%)	MCT oil (oleic acid (C18:1n9) content of 1–2%)	22.4
16. Zambon	1999	Italy	Mildly obese women	30	0	9	11	24	Parallel	Olive-oil-enriched (HiMUFA) hypocaloric diet (oleic acid (C18:1n9) content of 75.5%)	Carbohydrate enriched (HiCarbo) hypocaloric diet (oleic acid (C18:1n9) content of 10–20%)	31
17. Nicolaı¨ew	1998	France	Healthy subjects	32	100	10	10	3	Crossover	High-oleic acid sunflower oil (HOSO) (oleic acid (C18:1n9) content of 87%)	Extra virgin olive oil (oleic acid (C18:1n9) content of 75.5%)	22.4
18. Choudhury	1997	Australia	Healthy young adults	38.97	57.1	42	42	4	Crossover	High-oleic acid sunflower oil (HOSO) (oleic acid (C18:1n9) content of79.88%)	Palmolein (oleic acid (C18:1n9) content of42.2%)	24.5
19. Cater	1997	USA	Middle-aged men with mild hypercholesterolemia	66	100	9	9	3	Crossover	High-oleic acid sunflower oil (HOSO) (oleic acid (C18:1n9) content of87%)	Palmolein (oleic acid (C18:1n9) content of35%)	27
20. Reaven	1994	USA	Healthy subjects	NR	50	6	6	6	Parallel	High-oleic acid sunflower oil (HOSO) (oleic acid (C18:1n9) content of80%)	Standard diets (oleic acid (C18:1n9) content of 10–20%)	NR
21. Reaven	1991	USA	Healthy subjects	NR	NR	5	4	5	Parallel	High-oleic acid sunflower oil (HOSO) (oleic acid (C18:1n9) content of85%)	Regular Sunflower oil (oleic acid (C18:1n9) content of 10–20%)	NR
22. Wardlaw	1990	USA	Healthy subjects or At-risk (hypercholesterolemic) subjects.	NR	100	86	86	5	Crossover	High-oleic acid sunflower oil (HOSO) (oleic acid (C18:1n9) content of85%)	Corn oil(oleic acid (C18:1n9) content of 20–30%)	NR
23. Allman-Farinelli	2005	Australia	Healthy subjects	46.5	33/3	15	15	10	Crossover	High-oleic acid sunflower oil (HOSO) (oleic acid (C18:1n9) content of85%)	Regular Sunflower oil (oleic acid (C18:1n9) content of 66%)	24.7
24. Kien	2014	USA	Healthy subjects	77.5	50	18	18	3	Crossover	Low–palmitic acid and high–oleic acid (HOA) diet(oleic acid (C18:1n9) content of74.8%)	High–palmitic acid (HPA)(oleic acid (C18:1n9) content of39.9%)	29/5
25. Lambert	2007	South Africa	Regularly exercising individuals	32	50	32	32	12	Parallel	High-oleic acid sunflower oil (HOSO) (oleic acid (C18:1n9) content of85%)	Conjugated linoleic acid (CLA) (oleic acid (C18:1n9) content of24.7%)	22.5
26. Tholstrup	2011	Denmark	Healthy subjects	29.6	100	32	32	3	Crossover	Olive oil(oleic acid (C18:1n9) content of 73.3%)	Palm olein(oleic acid (C18:1n9) content of 46%)	22.9

BMI: body mass index, NR: Not reported

A total of 26 randomized controlled trials (RCTs) published between 1990 and 2023 were included in this meta-analysis. These studies encompassed a wide geographical distribution, representing 13 countries, including Spain, Malaysia, Canada, the United States, Australia, Brazil, Denmark, Italy, France, Lebanon, South Africa, and multicenter collaborations involving multiple regions. The populations investigated were diverse, spanning healthy adults, hypercholesterolemic or at-risk individuals, overweight or obese participants, postmenopausal women, and adults with central adiposity or metabolic syndrome. Mean participant ages ranged from 23.4 years in healthy young adults (Tholstrup 2004) to 77.5 years in older healthy adults (Kien 2014). The proportion of male participants varied widely, from studies including only women (e.g., Zambon 1999) to those enrolling exclusively men (e.g., Cater 1997, Alves 2014). Sample sizes ranged from 9 to 130 per group, with total study populations spanning 14 to 260 participants. Most studies (n = 19) adopted a crossover design, while 7 used parallel-group designs (e.g., Alves 2014; Hlais 2013; Zambon 1999; Reaven 1994; Lambert 2007). The duration of interventions ranged between 3 and 24 weeks, with the majority lasting 4 to 12 weeks. Across all included RCTs, dietary interventions compared high-oleic acid oils—such as high-oleic sunflower, canola, peanut, soybean, or olive oils—with corresponding conventional or low-oleic counterparts, or with other dietary fats (e.g., palmolein, fish oil, cocoa butter). The oleic acid (C18:1n9) content in intervention oils typically ranged from 70% to 89%, compared with 15% to 60% in control fats. Baseline BMI values across studies ranged from approximately 21.7 kg/m² in healthy volunteers (Loganathan 2022) to 31.7 kg/m² in centrally obese adults (Bowen 2019), reflecting a broad metabolic spectrum from normal-weight to obese participants. The included trials therefore represent a well-balanced dataset for evaluating the cardiometabolic effects of high-oleic acid interventions across diverse populations, age groups, and health statuses.

Risk of bias assessment (Based on Table 2)

Table 2.

Risk of bias assessment according to the Cochrane collaboration’s risk of bias assessment tool

Overall, the methodological quality of the included RCTs was moderate to high. Most trials reported adequate random sequence generation, with the majority judged as low risk for selection bias. Allocation concealment was generally satisfactory but not always explicitly described, resulting in several trials rated as “unclear risk.” Blinding of participants and personnel showed considerable variability, primarily due to the sensory differences between dietary oils, which make complete blinding difficult. Roughly half of the studies were judged to be high or unclear risk for performance bias. In contrast, blinding of outcome assessment was well maintained in most cases, particularly in those employing laboratory-based biochemical endpoints, leading to a predominantly low risk for detection bias. Incomplete outcome data and selective reporting were each rated as low risk in most studies, although a few lacked complete reporting of all outcomes measured. Overall, approximately nine trials were rated low risk of bias, fifteen as unclear, and two as high risk, primarily due to missing data or unblinded intervention administration. In summary, while most studies demonstrated robust randomization procedures and sound methodological quality, certain limitations, especially regarding participant blinding, should be considered when interpreting the pooled effects of high-oleic dietary interventions on cardiometabolic health outcomes.

According to the GRADE framework, the certainty of evidence was rated as moderate for the primary outcomes (TC and LDL-C), mainly due to substantial heterogeneity across studies (inconsistency). For secondary lipid outcomes (HDL-C, TG, Apo A1, Apo B), certainty ranged from moderate to low, primarily due to imprecision and heterogeneity. For glucose-related outcomes (FBS, insulin, HOMA-IR), the certainty of evidence was rated as low, owing to imprecision, limited number of trials, and variability in study populations.

Meta-analysis

Lipid profiles

With TC and LDL-C defined as the primary outcomes, the findings of the meta-analysis indicated that HODs intake significantly decreased serum of TC (WMD: −0.13 mmol/l; 95% CI: −0.24 to −0.01; P-value = 0.036) with considerable heterogeneity across studies (I² = 77.8%) and LDL-C (WMD: −0.11 mmol/l; 95% CI: −0.20 to −0.01; P-value = 0.033; I² = 62.1%) as well as significantly increased serum of Apo A1(WMD: 0.02 mmol/l; 95% CI: 0.00 to 0.03; P-value = 0.047; I² = 0.0%), compared with low-oleic diets (LODs) intake. In contrast, no significant changes were observed for other outcomes included: HDL-C(WMD: −0.01 mmol/l; 95% CI: −0.03 to 0.02; P-value = 0.665; I² = 0.0%), TG(WMD: −0.02 mmol/l; 95% CI: −0.07 to 0.02; P-value = 0.274; I² = 0.5%), and VLDL-C(WMD: −0.02 mmol/l; 95% CI: −0.11 to 0.07; P-value = 0.660; I² = 0.0%). Likewise, Apo B followed a similar trend, with WMDs 0.0 mmol/l(P-value = 0.599 for Apo B). Low heterogeneity was also observed for Apo B (I² = 5.5%) (Fig. 2).

Fig. 2.

Forest plots from the meta-analysis of clinical trials investigating the effects High-Oleic vs. Low- Oleic Diets on (a) TC (mmol/l), (b) LDL-C (mmol/l), (c) HDL-C(mmol/l), (d) TG(mmol/l), (e)VLDL-C(mmol/l), (f)Apo A1(mmol/l), and (g) Apo B(mmol/l). WMD: weighted mean

Glucose metabolism

The impact of HODs compared with LODs on glucose parameters was investigated and no statistically significant effect was observed for FBS (WMD: − 0.02 mmol/l; 95% CI: − 0.09 to 0.05; P-value = 0.626; I² = 20.9%), fasting insulin(WMD: − 0.42 mIU/l; 95% CI: − 1.21 to 0.37; P-value = 0.299; I² = 0.0%), and HOMA-IR(WMD: 0.0; 95% CI: − 0.19 to 0.19; P-value = 0.998; I² = 0.0%) (Fig. 3).

Fig. 3.

Forest plots from the meta-analysis of clinical trials investigating the effects High-Oleic vs. Low- Oleic Diets on (a) FBS (mmol/l), (b) Insulin (mIU/L), (c)HOMA-IR. WMD: weighted mean

Subgroups analysis

Based on the subgroup analysis results presented in Supplementary Table 1, several statistically significant effects of HODs compared with LODs on lipid outcomes were identified. Specifically, TC concentrations significantly decreased among participants consuming diets with ≥ 80% of total fatty acids from oleic acid (WMD: −0.29 mmol/l; 95% CI: −0.52 to − 0.06), whereas no significant effects were observed in subgroups categorized by intervention duration (≤ 4 weeks: WMD: −0.07 mmol/l; 95% CI: −0.17 to 0.04; >4 weeks: WMD: −0.15 mmol/l; 95% CI: −0.46 to 0.15) or body mass index (BMI ≤ 25 kg/m²: WMD: −0.11 mmol/l; 95% CI: −0.36 to 0.14; BMI > 25 kg/m²: WMD: −0.11 mmol/l; 95% CI: −0.24 to 0.01).

Similarly, LDL-C levels were significantly reduced in the subgroup with ≥ 80% oleic acid intake (WMD: −0.28 mmol/L; 95% CI: −0.44 to − 0.11), while no significant changes were found in subgroups based on BMI or duration. In contrast, HDL-C and TG did not show any statistically significant differences across all subgroup comparisons, as the confidence intervals included the null value.

Regarding apolipoproteins, Apo A1 concentrations showed a modest but statistically significant increase in studies where oleic acid contributed < 80% of total fatty acid intake (WMD: 0.02 mmol/L; 95% CI: 0.00 to 0.04), whereas Apo B remained unchanged across all subgroups. Furthermore, FBS and insulin levels were unaffected by dietary oleic content regardless of study duration, BMI, or oleic acid proportion.

These significant effects were observed across subgroups with varying levels of heterogeneity, with TC and LDL-C showing substantial variability (I² > 60%), while heterogeneity was notably lower for Apo A1 (I² = 6.5%). Collectively, these findings suggest that the proportion of oleic acid in total fatty acid intake represents a key determinant of the metabolic response, potentially contributing to the observed heterogeneity across included studies.

Meta-regression

Meta-regression analyses were conducted to examine the association between HODs intake and absolute mean differences in lipid profile and glucose metabolism, based on both the percentage of oleic intake from total fatty acid (%) and the duration of intervention (weeks). However, no significant relationships were identified for these variables except for TG and LDL-C. The findings demonstrated a linear association between TG (Coef = −0.0100154, P = 0.024) and LDL-C (Coef = −0.0180325, P = 0.027) alterations and percentage of oleic intake from total fatty acid (%) (Supplemental Figs. 1–4).

Sensitivity analysis

The leave-one-out method was applied to assess the influence of each individual study on the pooled effect size. The findings remained robust after sequential elimination of studies (Supplemental Figs. 5–6).

Publication bias

A visual review of the funnel plots indicated no apparent asymmetry, suggesting a low likelihood of publication bias concerning the effects of HODs intake across the evaluated outcomes. This visual observation was further corroborated by Egger’s regression test, which revealed no statistically significant evidence of publication bias for any of the parameters analyzed. The P-values were as follows: FBS (P = 0.681), insulin (P = 0.586), HOMA-IR (P = 0.453), TC (P = 0.206), LDL-C (P = 0.056), HDL-C (P = 0.881), TG(P = 0.653), VLDL-C (P = 0.881), Apo A1 (P = 0.602), and Apo B (P = 0.477). Supplementary Figs. 7 through 8 present the corresponding funnel plots. Additionally, the trim-and-fill method revealed no missing studies, further supporting the robustness of the publication bias analysis.

Discussion

This meta-analysis assessed randomized controlled trials comparing HODs versus LODs to determine effects on lipid and glucose metabolism. Our pooled results indicate modest but statistically significant reductions in total cholesterol and LDL-C, a small rise in Apo A1, and no clear overall effect on fasting glucose or insulin. The following discussion examines each outcome in detail, compares findings to prior evidence, explores plausible mechanisms, and interprets subgroup and meta-regression results in a clinical context.

Our pooled analyses demonstrated small but significant reductions in TC (WMD ≈ − 0.13 mmol/l) and LDL-C (WMD ≈ − 0.11 mmol/l). Although the absolute reductions in TC and LDL-C observed in this meta-analysis are modest (approximately 0.1 mmol/L), such changes should not be dismissed as clinically irrelevant. Large-scale epidemiological studies and statin trials suggest that each 0.1 mmol/L reduction in LDL-C is associated with an estimated 2–3% lower risk of major cardiovascular events over time. From a public health perspective, even small shifts in population lipid distributions may translate into meaningful reductions in cardiovascular burden. Nevertheless, these effect sizes are insufficient to replace pharmacological therapy in high-risk individuals and should be viewed as complementary benefits of dietary modification rather than stand-alone therapeutic effects. These findings are consistent with prior feeding trials and systematic reviews showing that replacing saturated or PUFA-dominated fats with high-oleic (MUFA-rich) fats tends to lower LDL-C and TC. A focused systematic review of high-oleic vegetable-oil substitutions reported similar reductions in TC and LDL-C when SFA or other oils were replaced by HO oils [12]. Mechanistically, oleic acid modulates hepatic lipid handling by promoting LDL-receptor–mediated clearance and altering lipoprotein composition (e.g., increasing cholesteryl-oleate content in LDL) [29, 30], which reduces LDL atherogenicity and circulating concentrations [31].

We observed no consistent change in HDL-C or TG across pooled studies. Prior literature is heterogeneous: some MUFA interventions report increases in HDL or shifting to larger HDL subfractions, but many controlled trials find little change in total HDL concentration, especially in short trials or among normolipidemic individuals. One reason is that MUFA intake often influences HDL particle functionality or subpopulation distribution rather than raising total HDL mass; another reason is that TG lowering typically requires calorie deficit, weight loss, or marked carbohydrate lowering, conditions not consistently present in the included trials [32, 33]. Studies that showed TG reductions often involved subjects with baseline hypertriglyceridemia or longer interventions [34, 35].

Apo A1 increased slightly and significantly (WMD ≈ 0.02 mmol/l), whereas Apo B was unchanged. The increase in Apo A1 aligns with mechanistic data indicating that MUFAs can up-regulate Apo A1 expression (via nuclear receptor and SREBP pathways) and/or alter HDL particle remodeling [36, 37]. Clinically, a modest Apo A1 rise could reflect improved HDL functionality even when total HDL-C remains static [38, 39]. The null Apo B result suggests that particle number did not change substantially beyond LDL-C concentration shifts, or that assay variability and study heterogeneity attenuated detectability.

We found no statistically significant pooled effects on fasting glucose, insulin, or HOMA-IR. This neutral result mirrors several feeding-trial meta-analyses which conclude that MUFA substitution does not consistently improve fasting glycaemia across mixed populations; PUFAs may show more consistent glycaemic benefits in some analyses [40]. Several potential reasons could explain the lack of significant results for variables related to glucose metabolism [1]. baseline metabolic health; [2] short durations, insulin signalling and whole-body insulin sensitivity can require longer exposures or weight change to show measurable fasting differences; [3] variation in background carbohydrate quality/amount, glycaemic effects of MUFAs are often modulated by concurrent carbohydrate reduction or fiber intake; [4] heterogeneous endpoints, postprandial insulin/glucose or clamp studies may reveal benefits not captured by fasting indices [41, 42].

Taken together, the findings of this meta-analysis suggest that high-oleic diets exert selective metabolic effects, primarily targeting lipid homeostasis rather than glucose regulation. The consistent reductions in TC and LDL-C, alongside modest increases in Apo A1, indicate improved lipoprotein metabolism, likely mediated through hepatic mechanisms such as enhanced LDL receptor activity, altered cholesteryl ester composition, and reduced LDL proteoglycan binding. In contrast, the absence of significant effects on fasting glucose and insulin suggests that oleic acid alone may be insufficient to modify systemic insulin sensitivity in the absence of concurrent weight loss, carbohydrate reduction, or longer intervention durations. This divergence between lipid and glycaemic responses is biologically plausible, as lipid metabolism is more directly responsive to fatty acid quality, whereas glucose homeostasis is regulated by multifactorial pathways involving energy balance, adiposity, and pancreatic β-cell function.

Blinding represents a recognized challenge in dietary intervention trials, particularly when comparing different oils with distinct sensory properties. Although approximately half of the included studies exhibited high or unclear risk of performance bias, the main outcomes assessed in this meta-analysis were objective biochemical measures, which are inherently less vulnerable to expectation or placebo effects. In addition, laboratory analyses were commonly conducted using automated assays and, in many trials, by personnel unaware of group allocation. Consequently, while lack of blinding may have influenced subjective factors such as adherence or dietary behavior, its impact on the validity of the primary lipid and glucose outcomes is likely limited.

Subgroup analyses and meta-regression

Our pre-specified subgroup results revealed stronger TC/LDL-C lowering when the oleic proportion in total fatty acids was ≥ 80% and when the absolute difference between arms exceeded the pre-defined ≥ 5% points, consistent with a dose-response relation. These findings suggest a dose–response relationship between oleic acid proportion and lipid outcomes, indicating that while ≥ 70% oleic acid is sufficient to define high-oleic exposure, higher thresholds may be required to achieve clinically detectable metabolic effects. Quantitatively, meta-regression indicated that each 1% increment in oleic acid content corresponded to approximately 0.018 mmol/L lower LDL-C, supporting a biologically plausible gradient of effect. These findings indicate that magnitude of exposure contrast (not merely labeling as “high-oleic”) is a key determinant of effect size, and help explain between-study heterogeneity. Duration and baseline BMI had weaker or inconsistent moderating effects, suggesting that oleic dose is a primary driver within the studied ranges.

Clinical implications

For clinicians, substituting saturated fats or linoleic-rich oils with high-oleic alternatives (provided the oleic contrast is substantial) can produce modest TC and LDL-lowering and Apo A1 benefits, a pragmatic dietary strategy complementary to other lifestyle measures. For glycaemic control, HODs alone should not be relied upon as a primary intervention in patients with overt insulin resistance; combining with carbohydrate quality improvement or weight loss may be necessary. From a practical standpoint, achieving very high oleic acid intake (≥ 80% of total fatty acids) is feasible through the use of specific culinary oils. Extra-virgin olive oil, high-oleic sunflower oil, high-oleic canola oil, and high-oleic safflower oil typically contain 75–85% oleic acid and represent realistic dietary sources. Replacing saturated fats (e.g., butter, palm oil) or conventional polyunsaturated vegetable oils with these high-oleic oils may therefore be a pragmatic strategy to enhance cardiometabolic health.

Limitations and strengths

Key limitations include heterogeneity in oil sources/compositions, relatively short durations in many trials, variable reporting of fatty-acid analysis (some studies lacked measured % oleic and relied on formulation), and limited numbers of high-risk populations. The included studies encompassed heterogeneous populations, ranging from healthy adults to individuals with obesity, dyslipidaemia, or metabolic syndrome. Although the overall direction of lipid effects was broadly consistent, it is likely that absolute benefits are greater among individuals with higher baseline cardiometabolic risk. Unfortunately, the limited number of trials within specific clinical subgroups prevented formal stratified analyses. Nevertheless, inspection of individual trial results suggests that the direction of lipid-modulating effects was broadly consistent across populations, with greater absolute benefits likely among individuals with higher baseline cardiometabolic risk. Imputation of some SDs and combining crossover and parallel designs (with approximations) add uncertainty, although sensitivity analyses were performed. The substantial heterogeneity observed for TC and LDL-C likely reflects multiple sources beyond oleic acid proportion alone. Differences in oil sources (e.g., olive oil vs. high-oleic sunflower or canola), baseline lipid status, overall dietary background, and study design (crossover vs. parallel) may all contribute to between-study variability. However, formal exploration of these factors was limited by insufficient reporting and the small number of studies within each category, precluding robust meta-regression analyses. Consequently, residual heterogeneity remains and should be considered when interpreting the pooled effect estimates. Using the GRADE approach, the main reason for downgrading evidence certainty was inconsistency (heterogeneity) for lipid outcomes and imprecision for glucose-related outcomes, reflecting the relatively small sample sizes and short duration of several trials. In crossover trials, the adequacy of washout periods represents an important methodological consideration. Most included studies reported washout intervals of 2–6 weeks, which are generally sufficient for stabilization of circulating lipid biomarkers following dietary fat interventions. In studies where washout was not clearly specified, this was considered in the risk-of-bias assessment. Notably, any residual carryover effect would be expected to attenuate between-group differences and bias results toward the null, suggesting that the observed lipid-lowering effects are unlikely to be artefactual. Although sensitivity analyses based on exclusion of high-risk-of-bias studies are clinically informative, only two included trials were classified as having an overall high risk of bias. Excluding these studies did not materially alter the direction or magnitude of effect estimates in exploratory checks, and leave-one-out analyses similarly indicated that no single trial disproportionately influenced pooled results. This suggests that the main findings are robust to study quality. Strengths are strict inclusion criteria (a priori ≥ 70% oleic and ≥ 5%-point absolute difference), rigorous PRISMA-based methodology, comprehensive subgroup/meta-regression exploration, and focus on both lipid and glucose domains with mechanistic context. A major strength of the included trials is that most employed isoenergetic dietary substitution designs, in which high-oleic fats replaced comparator fats without altering total caloric intake or overall fat percentage. This design minimizes confounding by energy balance and supports the interpretation that observed metabolic effects are primarily attributable to fatty acid quality rather than differences in total fat consumption.

Conclusion

High-oleic dietary interventions modestly improve circulating lipid profiles, particularly total and LDL cholesterol, with stronger effects observed at higher proportions of oleic acid intake. The magnitude of benefit appears to follow a dose–response pattern, with very high-oleic diets (≥ 80% of total fatty acids) yielding the most consistent lipid improvements. Future longer and larger RCTs in insulin-resistant populations with carefully quantified fatty-acid exposure are needed to clarify potential glycaemic benefits and long-term cardiovascular implications.

Author contributions

XH, XX, SMH: conception, design, statistical analysis, data collection, writing-original draft, supervision.SMH, XX: data collection and writing-original draft.All authors approved the final version of the manuscript.

Funding

No funding.

Data availability

The datasets used and/or analysed during the current study available from the corresponding author on reasonable request.

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.