Oral glucose absorption is enhanced in early metabolic dysfunction-associated steatotic liver disease

Abstract

Aims/hypothesis

Hepatic glucose flux plays a crucial role in the progression of metabolic dysfunction-associated steatotic liver disease (MASLD), promoting de novo lipogenesis, inflammation and fibrosis. This study aimed to evaluate the kinetics of oral glucose absorption and one of its key modulators, gastric emptying, in individuals with early-stage MASLD vs matched control individuals.

Methods

We quantified glucose metabolic fluxes during a 75 g OGTT using stable isotopes in individuals with MASLD without fibrosis and in healthy control individuals. In a separate cohort, we measured the gastric emptying rate using the ¹³C-acetate breath test during an OGTT and estimated hepatic steatosis risk.

Results

Compared with the control group, in the MASLD group the rate of appearance of oral ingested glucose (RaO) normalised to body weight was 34% higher at 1 h post-OGTT (+318±142 µmol/kg, p=0.031), resulting in a 52% increase in total glucose absorption (+6.4±1.8 g, p=0.001). Participants with MASLD exhibited reduced glucose clearance relative to plasma insulin levels but preserved post-load suppression of endogenous glucose production, indicating peripheral rather than hepatic insulin resistance. Among glucose metabolic fluxes, RaO showed the strongest association with prevalent MASLD, with each 1-SD increase in 1 h RaO being associated with fivefold higher odds of MASLD (OR 4.99 [95% CI 1.44, 31.57], p=0.036), independent of potential confounders. Gastric emptying rate was not associated with hepatic steatosis risk.

Conclusions/interpretation

Oral glucose absorption is augmented in individuals with MASLD without fibrosis, apparently unrelated to accelerated gastric emptying. This metabolic alteration may represent an early driver of MASLD pathogenesis, preceding hepatic insulin resistance. Future research should investigate whether modulation of intestinal glucose absorption confers therapeutic benefits in MASLD.

Graphical Abstract

Supplementary Information

The online version contains peer-reviewed but unedited supplementary material available at 10.1007/s00125-026-06730-5.

Introduction

Metabolic dysfunction-associated steatotic liver disease (MASLD) poses a growing health challenge, affecting approximately 40% of adults globally and conferring substantially increased risks of hepatocellular carcinoma and cardiovascular mortality [1]. Formerly termed non-alcoholic fatty liver disease (NAFLD), MASLD exists on a spectrum ranging from simple steatosis to steatohepatitis, fibrosis and cirrhosis [2]. The pathogenesis of MASLD is multifactorial and intrinsically linked to metabolic derangements, with chronic hyperglycaemia, insulin resistance and beta cell dysfunction recognised as key drivers of hepatic lipid accumulation and inflammation [3, 4].

Emerging evidence from both clinical and experimental research has hinted at profound alterations in oral glucose handling in individuals with MASLD, supporting a role for enhanced postprandial glucose flux to the liver in promoting de novo lipogenesis (DNL), inflammation and fibrosis. Oral glucose absorption, quantified in vivo by the rate of appearance of oral ingested glucose (RaO), is a major determinant of postprandial glycaemic and insulinaemic responses, especially in individuals without diabetes [5, 6]. RaO is primarily regulated by gastric emptying, which determines the rate of entry of ingested glucose into the small intestine [7], and by the expression of glucose carriers in the upper small intestine, particularly sodium–glucose cotransporter 1 (SGLT1) [8], whose expression and activity have been associated with MASLD risk [9–11]. On the other hand, gastrointestinal motility and SGLT1 regulate the glucose-induced release of the incretin hormones glucagon-like peptide-1 (GLP-1) and glucose-dependent insulinotropic polypeptide (GIP) [8], which may mitigate hepatic fat accumulation through improved glucose and lipid homeostasis [12, 13]. The role of postprandial hepatic glucose flux in the pathogenesis of MASLD is further corroborated by the observation that 1 h plasma glucose during an OGTT, influenced by both gastric emptying [14, 15] and intestinal glucose absorption [14, 16], is elevated in people with MASLD and correlates directly with hepatic steatosis and fibrosis severity [17, 18].

Nonetheless, previous studies have enrolled predominantly individuals with advanced MASLD and relied on proxies of oral glucose absorption (e.g. expression of glucose transporters or blood glucose profiles during OGTTs). Thus, a substantial knowledge gap persists regarding in vivo dynamics and determinants of glucose flux, particularly in the ‘early’ stage of MASLD. These limitations impede the development of targeted therapeutic strategies, as oral glucose handling (if confirmed as a key pathogenetic factor) could constitute a modifiable contributor to the progression of MASLD.

This study employed the gold-standard dual-tracer technique, involving both i.v. and oral administration of stable glucose isotopes, to quantify intestinal glucose absorption kinetics after an oral glucose load, while simultaneously assessing peripheral glucose disposal and hepatic glucose production. We further explored, in a separate cohort, the relationship between gastric emptying of an oral glucose load assessed by the ¹³C-acetate breath test and the hepatic steatosis index (HSI), a surrogate marker of liver fat. By delineating post-load glucose fluxes and their underlying mechanisms, we tested the hypothesis that oral glucose absorption is higher in individuals with early-stage MASLD than in matched control individuals.

Methods

Study 1: dual-tracer study

Participants

Male and female volunteers aged 18–65 years, with a BMI of 18–40 kg/m², representative of the local population, were recruited at the Section of Dietetics and Clinical Nutrition at the University Hospital of Pisa (Italy) during screening for nutrition and pharmacological trials. Recruitment procedures and study protocols were approved by the local Human Ethics Committee (North-West Wide-Area Ethics Committee, protocol no. 13053_NATALI, 6792_NATALI, 8512015_NATALI). At screening, medical history and medication use were recorded using standardised questionnaires, which also captured current and prior alcohol consumption. Volunteers were excluded if they had acute or chronic disorders, including type 2 diabetes, were taking medications influencing glucose or lipid metabolism, or were pregnant or lactating. The study was conducted in accordance with the principles expressed in the Declaration of Helsinki. All participants provided written informed consent before enrolment.

MASLD assessment

In accordance with a 2023 multisociety consensus statement [2], steatotic liver disease (SLD) was diagnosed using ultrasound imaging by a trained radiologist, while MASLD was defined as the presence of SLD associated with at least one cardiometabolic risk factor, in the absence of significant alcohol consumption (>140 g/week or >20 g/day for female participants; >210 g/week or >30 g/day for male participants) and other discernible causes of liver disease. Fasting blood samples were collected for measurement of complete blood count, HbA_1c, total and direct bilirubin, and liver enzymes including aspartate aminotransferase (AST), alanine aminotransferase (ALT), γ-glutamyl transferase and alkaline phosphatase. The ratio between the concentrations of AST and ALT (AST/ALT ratio) was calculated. Individuals with no evidence of liver fibrosis of any grade, as assessed using the Fibrosis-4 (FIB-4) index [19], were included in the study. The FIB-4 index was calculated as age (years) × AST (U/l) / (platelet [count in 10⁹/l] × √(ALT [U/l]), and liver fibrosis was excluded based on a cutoff of 1.3 [19]. In six cases with a FIB-4 score ≥1.3 (up to 1.8), fasting liver elastography was performed by a trained hepatologist using a FibroScan device (EchoSens, Paris, France), and the absence of significant fibrosis was confirmed by liver stiffness values ≤8 kPa, as indicated by the current clinical guidelines for MASLD management [20].

Metabolic tests

Each participant underwent a frequently sampled, 75 g OGTT, modified with the use of two stable glucose isotopes to quantify glucose metabolic fluxes. At 08:00 hours, after a standardised dinner and an overnight fast (12 h), participants were admitted to our Clinical Research Unit. A 20-gauge polyethylene cannula was inserted into an antecubital vein for the infusion of all test substances. A second cannula was inserted retrogradely into a wrist vein for blood sampling, and the hand was kept wrapped in a heated blanket to achieve the arterialisation of venous blood. From time −120 min, volunteers received a 5 h primed (28 µmol/kg) continuous (0.28 µmol min⁻¹ kg⁻¹) infusion of 6,6-[²H₂]glucose (Cambridge Isotope Laboratories, Tewksbury, MA, USA). After 2 h (time 0 min), participants consumed a 300 ml oral glucose drink containing 73.5 g dextrose and 1.5 g [U-¹³C]glucose (Cambridge Isotope Laboratories) within 5 min (time 0–5 min). Arterialised blood samples were collected throughout the test at times −120, −15, 0, 15, 30, 45, 60, 90 and 120 min to measure plasma glucose, insulin, C-peptide, glucagon, GLP-1, GIP, NEFA and glucose tracer enrichment (electronic supplementary material [ESM] Fig. 1). Participants were classified as having impaired fasting glucose (IFG) or impaired glucose tolerance (IGT) according to current diagnostic criteria [21].

Analytical procedures

Routine biochemical analyses were performed at the core laboratory of the University Hospital of Pisa, while other analyses were performed at the Laboratory of Nutrition, Metabolism, and Atherosclerosis of the University of Pisa. Plasma glucose concentrations during the OGTT were measured immediately by the glucose-oxidase technique on a GM9 Glucose Analyser (Analox Instruments, Stourbridge, UK). 6,6-[²H₂]glucose and [U-¹³C]glucose were measured by GC-MS [22]. Insulin and C-peptide were assayed by electrochemiluminescence on a COBAS e411 instrument (Roche, Indianapolis, IN, USA). Glucagon was measured by ELISA (Mercodia, Uppsala, Sweden). Total GLP-1 and GIP were measured by ELISA (Merck, Darmstadt, Germany). NEFA were assayed by standard spectrophotometric methods on a Synchron UniCel DxC 600 instrument (Beckman Instruments).

Glucose kinetics

RaO, whole-body glucose clearance (GCl) and endogenous glucose production (EGP) were assessed from the time course of the plasma tracer/tracee ratio of 6,6-[²H₂]glucose and [U-¹³C]glucose using a previously described model [23]. Data were normalised by individual body weight to facilitate comparison with prior studies and to account for group differences in BMI.

Beta cell function modelling

Insulin secretion rate (ISR) was estimated by C-peptide deconvolution [24]. Beta cell function parameters were calculated by mathematical modelling of ISR and glucose concentrations, as previously reported [25, 26]. The relationship between glucose and ISR is described as the sum of two components, with the first component representing the dependence of ISR on absolute glucose concentration. The quasi-linear dose–response function relating the two variables is described by a slope, named beta cell glucose sensitivity (β-GS), and by ISR at a fixed glucose concentration of 5 mmol/l (ISR@5). This function can be modulated by several factors (i.e. gastrointestinal hormones, neurotransmitters, non-glucose substrates), which are collectively modelled as a potentiation factor. The potentiation factor is set to be a positive function of time and to average the value 1 during the OGTT. The potentiation factor ratio of the values at 100–120 min vs 0–20 min is used to express this component. The second component of insulin secretion represents the dynamic dependence of ISR on the rate of change of glucose concentration and is named beta cell rate sensitivity (β-RS).

Insulin clearance

Fasting and post-load endogenous insulin clearance, which refers to the process (mainly hepatic) by which the secreted insulin is removed from the bloodstream [27], were calculated as the ratios between fasting levels or AUC, respectively, of ISR and plasma insulin, allowing comparisons with previous studies [5, 28].

Insulin sensitivity

Whole-body and hepatic insulin sensitivity were quantified using glucose kinetics data, as previously reported [16]. Whole-body insulin sensitivity was measured as the GCl normalised to plasma insulin levels, calculated as the ratio between fasting GCl and plasma insulin (fasting GCl/Ins) or between their AUC during the OGTT (GCl/Ins AUC_0–120). Hepatic insulin resistance was measured as EGP × ISR, the latter being a proxy of portal insulin levels, calculated as fasting EGP multiplied by fasting ISR (fasting EGP × ISR) or by the product of their AUC during the OGTT (EGP × ISR AUC_0–120). Additionally, to facilitate comparison with prior studies, we calculated surrogate indices of whole-body, hepatic and adipose tissue insulin sensitivity. Whole-body insulin sensitivity was estimated using both the HOMA-IR and the Matsuda index [29]. Hepatic insulin resistance was estimated using the hepatic insulin resistance index (HIRI) (calculated as glucose AUC_0–30 [mmol/l × h] × insulin AUC_0–30 [pmol/l × h] [29]). Adipose tissue insulin resistance was estimated using the Adipo-IR index (calculated as fasting NEFA [mmol/l] × fasting insulin [pmol/l] [30]).

Study 2: gastric emptying and hepatic steatosis

Participants

Male and female participants aged 18–70 years, with a BMI of 18–30 kg/m² and no prior history of diabetes mellitus, representative of the local population, were recruited from the community in Adelaide, Australia, via public advertisements. Participants were screened using a 75 g OGTT with concurrent measurement of gastric emptying by stable isotope breath test for inclusion in a low-energy sweetener trial approved by the Central Adelaide Local Health Network Human Research Ethics Committee (protocol n. 2022/HRE00302). Participants were excluded if they reported habitual use of more than one serve per day of any food or beverage containing a low-energy sweetener during the past 3 months, significant gastrointestinal symptoms, a history of gastrointestinal surgery, or a requirement for medication known to affect gastrointestinal function or appetite. Female participants who used oral contraceptives or were pregnant were also excluded. The study was conducted in accordance with the principles expressed in the Declaration of Helsinki. All participants provided written informed consent.

Measurement of gastric emptying and blood glucose response to a 75 g OGTT

Participants refrained from strenuous physical activity for 24 h before the study and fasted from food and nutrient beverages from 20:00 hours but were allowed to drink water until midnight, before attending our clinical research facility at the University of Adelaide the following day at 08:00 hours. On arrival, an i.v. cannula was inserted into a forearm vein and the arm kept heated to allow sampling of arterialised venous blood. A fasting sample was collected for measurements of HbA_1c, blood glucose and biochemical variables, including serum AST, ALT, alkaline phosphatase and γ-glutamyl transferase. Participants then consumed a 300 ml glucose drink containing 75 g glucose and 150 mg ¹³C-acetate within 5 min (time 0–5 min). Additional venous blood was sampled at 30, 60, 120 and 180 min after the drink. Blood glucose concentrations were assessed using a glucometer (Optium Xceed; Abbott Laboratories, USA) and reported as the mean of duplicate measurements at each time point.

Breath samples were collected immediately before the drink, then every 5 min after the drink for the first 30 min and every 15 min for a further 150 min. ¹³CO₂ in each breath sample was measured by a non-dispersive infrared spectrometer (FANci2; Fischer Analysen Instrumente, Germany). The gastric half-emptying time (T50) and gastric retention at 1 h after the drink were calculated using the Wagner–Nelson method, as previously described [31, 32]. This method has shown comparable accuracy to scintigraphy for the measurement of gastric emptying of both solid and liquid meals [31].

The HSI, a surrogate marker of liver fat, was calculated as follows: 8 × ALT (U/l) /AST (U/l) ratio + BMI (kg/m²) + 2 (if female) + 2 (if diabetes) [33]. An HSI score ≥36 is considered high-risk for MASLD, while MASLD is unlikely with HSI <30 [33].

Statistical analysis

Continuous variables were tested for normality using the Shapiro–Wilk test. Data are reported as mean ± SD or median (IQR) if not normally distributed, unless otherwise stated. AUC was calculated using the trapezoidal rule.

In Study 1, group differences were analysed using unpaired Student t test, Mann–Witney test or Fisher exact test, as appropriate. Repeated measures were analysed by two-way ANOVA followed by multiple pairwise comparisons using Sidak’s tests. Relationships were tested by Pearson correlation. Univariable and multivariable logistic regression analysis was used to identify the postprandial glucose homeostatic mechanisms associated with MASLD, selected between early glucose absorption rate (RaO AUC_0–60), whole-body insulin sensitivity (GCl/Ins AUC_0–120), hepatic insulin resistance (EGP × ISR AUC_0–120), beta cell function (β-GS) and insulin clearance. ORs with 95% CIs per SD change in the predictor variable were obtained from estimating models on standardised variables. To account for the potential impact of ageing, sex, glucose management and obesity on glucose metabolic fluxes or MASLD, multivariable models were performed including age, sex (self-reported) and HbA_1c as covariates; BMI was introduced only later in subsequent models to minimise the risk of overadjustment, given that glucose fluxes were already normalised to body weight. To disentangle whether observed group differences reflected hepatic steatosis per se or overall adiposity, we also performed subset analyses excluding participants with extreme BMI, either >33 kg/m² (n=3, all in the MASLD group) or <20 kg/m² (n=3, all in the control group). The resulting cohort subset and analyses were referred to as BMI-matched.

In Study 2, participants were stratified into three subgroups based on the HSI cut-offs of 30 and 36 [33]. Demographic and clinical variables between subgroups with HSI ≥36 and HSI<30 were compared using Fisher’s exact test, unpaired Student t tests or Mann–Whitney U tests, as appropriate. The T50 and gastric emptying at 1 h post-OGTT were compared in subgroups with HSI ≥36 and HSI <30 using unpaired Student t test, both before and after adjustment for age, sex, HbA_1c and BMI. The blood glucose response to oral glucose in subgroups with HSI ≥36 and HSI <30 was also compared using two-factor repeated measures ANOVA, with group and time as factors. The relationships between HSI and 1 h blood glucose, T50 or gastric emptying at 1 h post-OGTT were evaluated using Pearson correlation analysis.

Statistical analyses were performed using either JMP SE software version 18.2.1 (JMP Statistical Discovery, Cary, NC, USA) or SPSS statistics version 29.0 (IBM, NY, USA). All tests were conducted at a two-sided α level of 0.05.

Results

Study 1

Participants

Forty-two volunteers completed the dual-tracer OGTT (mean age 45±14 years, 25 men and 17 women, BMI 26.7±4.6 kg/m², liver enzymes <1.5 × upper limit of normal in all participants). Participants’ clinical and metabolic characteristics are reported in Table 1. Individuals with MASLD (n=20, 47.6%) had a similar age, sex distribution and BP to individuals without MASLD. The MASLD group had higher BMI (+3.8±1.3 kg/m², p=0.007), HbA_1c (+3.2±1.5 mmol/mol [0.3±0.1%], p=0.036), LDL-cholesterol (+0.5±0.3 mmol/l, p=0.010), triglycerides (+0.8±0.4 mmol/l, p=0.0001), ALT (+8±3 U/l, p=0.021) and γ-glutamyl transferase (+19±8 U/l, p=0.005) levels.

Table 1.

Clinical and metabolic characteristics of Study 1 participants

Characteristic	MASLD (n=20)	Control (n=22)	p value
Women, n (%)	7 (35)	10 (46)	0.543
Age, years	50 (37–58)	44 (28–56)	0.252
BMI, kg/m2	28.7±4.4	24.9±4.2	0.007*
Waist circumference, cm	105±8	98±12	0.147
Systolic BP, mmHg	131±8	132±7	0.780
Diastolic BP, mmHg	84±8	80±9	0.433
HbA1c, mmol/mol	39±5	35±4	0.036*
HbA1c, %	5.7±0.5	5.4±0.4	0.036*
Fasting glucose, mmol/l	5.4±0.8	5.2±0.5	0.353
1 h glucose, mmol/l	10.0±1.8	8.4±1.6	0.004**
2 h glucose, mmol/l	8.9±2.5	7.9±2.0	0.178
Glucose peak, mmol/l	10.5±1.9	9.2±1.9	0.048*
Fasting insulin, pmol/l	85 (55–117)	48 (28–69)	0.0004***
IFG/IGT, n (%)	17 (85.0)	14 (63.6)	0.166
Total cholesterol, mmol/l	4.97 (4.48–5.91)	4.40 (3.73–5.10)	0.025*
HDL-cholesterol, mmol/l	1.22 (0.91–1.48)	1.42 (1.04–1.79)	0.144
LDL-cholesterol, mmol/l	3.37 (2.75–4.01)	2.54 (2.02–3.16)	0.010*
Triglycerides, mmol/l	1.38 (0.36–1.85)	0.79 (0.60–0.97)	0.0001***
Creatinine, µmol/l	76.9±10.6	72.5±15.0	0.272
eGFR (CKD-EPI), ml/min per 1.73 m2	93±16	105±23	0.056
AST, U/l	20 (18–26)	17 (15–23)	0.058
ALT, U/l	26 (16–32)	16 (14–22)	0.021*
AST/ALT ratio	0.95±0.31	1.18±0.38	0.057
γ-Glutamyl transferase, U/l	30 (17–64)	14 (10–22)	0.005**
Alkaline phosphatase, U/l	64 (46–73)	67 (47–88)	0.233
Total bilirubin, µmol/l	8.9 (6.2–12.7)	7.0 (6.0–11.6)	0.458
Direct bilirubin, µmol/l	3.3 (2.2–5.0)	2.6 (2.2–5.0)	0.649
Platelets, n/µl	217 (185–251)	255 (223–283)	0.069
HSI	38.7±5.7	32.8±5.3	0.001**
FIB-4	0.9 (0.6–1.4)	0.7 (0.6–0.9)	0.070

Data are presented as n (%), mean ± SD or median (IQR)

^*p<0.05, **p<0.01, ***p<0.001 (analysed by Fisher’s exact test, Student’s t test or Mann–Whitney test, as appropriate)

IFG, impaired fasting glucose; IGT, impaired glucose tolerance

Glucose metabolic fluxes

Plasma glucose concentrations at fasting and 2 h post-OGTT were similar between the two groups (Table 1, Fig. 1a, b). However, blood glucose at 1 h post-OGTT (+1.5±0.5 mmol/l, p=0.004), peak glucose (+1.2±0.6 mmol/l, p=0.048) and glucose AUC_0–120 (+127±56 mmol/l × min, p=0.028) were higher in the MASLD group than in the control group.

Fig. 1.

Plasma glucose and glucose metabolic fluxes. (a–e, g–l) Profiles and AUCs during a 75 g OGTT of plasma glucose (a, b), tracer-derived RaO (c–e), total oral glucose absorbed at 1 h and 2 h (g, h), GCl (i) and EGP (j–l) in individuals with early-stage MASLD (n=20) and control individuals (n=22). Data are mean ± SEM indicated by error bars or shaded areas. (f) Correlation of RaO with 1 h plasma glucose, evaluated using Pearson correlation. Best fit line (continuous line) and 95% CIs (dotted lines) are shown. Group differences were analysed using Student’s t test. Repeated measures were analysed by two-way ANOVA followed by multiple pairwise comparisons using Sidak’s post hoc tests. *p<0.05, **p<0.01, ***p<0.001

The RaO was substantially higher in the MASLD group, particularly in the first hour of the OGTT (Fig. 1c–e), with RaO AUC_0–60 being 33.6% greater in the MASLD group than in the control group (+318±142 µmol/kg, p=0.031) and directly correlating with 1 h plasma glucose (r=0.599, p<0.0001) (Fig. 1f). Group differences in RaO AUC_0–60 and RaO AUC_0–120 were statistically significant after adjustment for age, sex and HbA_1c (p=0.011 and p=0.048, respectively) and after further adjustment for BMI (p=0.003 and p=0.015, respectively). In the MASLD group compared with the control group, the total amount of absorbed oral glucose was 51.6% higher at 1 h during the OGTT (+6.4±1.8 g, p=0.001) and 53.2% higher at 2 h (+10.7±3.6 g, p=0.005) (Fig. 1g, h). Results were materially unchanged in BMI-matched sensitivity analyses (ESM Table 1). After excluding participants with extreme BMI values, group differences in BMI disappeared (p=0.177), while differences remained statistically significant for RaO AUC_0–60 (+446±152 µmol/kg, p=0.006), RaO AUC_0–120 (+737±317 µmol/kg, p=0.028) and absorbed oral glucose at 1 h (+6.9±2.0 g, p=0.001) and 2 h (+11.5±3.8 g, p=0.006) during the OGTT. These differences persisted after adjustment for age, sex and HbA_1c, as well as after further adjustment for BMI (p<0.02 for all).

Whole-body plasma GCl was similar between groups during both fasting and the OGTT (Fig. 1i), without significant differences in GCl AUC_0–60 or GCl AUC_0–120 in either univariable or multivariable analyses (p>0.10).

The EGP was similar between groups during fasting (p=0.135) and partially suppressed after the oral glucose load (Fig. 1j–l). In the MASLD group, time course profiles displayed lower EGP during the first hour of the OGTT, with EGP AUC_0–60 being 44.2% more suppressed (−148.0±62.4 µmol/kg, p=0.021) compared with the control group. Group differences in EGP AUC_0–60 remained statistically significant after adjustment for age, sex and HbA_1c (p=0.016) and after further adjustment for BMI (p=0.014).

Beta cell function and insulin clearance

Plasma insulin levels were higher in the MASLD group than the control group during fasting (+38±10 pmol/l, p=0.0004) and throughout the OGTT (AUC_0–120 +15±7 nmol/l × min, p=0.031) (Table 1, Fig. 2a, b). Moreover, individuals with MASLD displayed a higher ISR at fasting (+26±12 pmol min⁻¹ m⁻², p=0.039) and at 1 h post-OGTT (+102±48 pmol min⁻¹ m⁻², p=0.038), although C-peptide levels (AUC_0–120 p=0.338), ISR AUC_0–60 (p=0.543) and ISR AUC_0–120 (p=0.432) were similar between groups (Fig. 2c–f). There were no group differences in the glucose-ISR dose–response function (Fig. 2g), nor in model-derived components of beta cell function (Fig. 2h–j), including β-GS (p=0.280), β-RS (p=0.159), potentiation factor ratio (p=0.425) and ISR@5 (p=0.742; data not shown).

Fig. 2.

Beta cell function and insulin clearance. (a–f) Profiles and AUCs during a 75 g OGTT of plasma insulin (a, b), C-peptide (c, d) and C-peptide-derived ISR (e, f) in individuals with early-stage MASLD (n=20) and control individuals (n=22). (g) Dose–response function between the observed plasma glucose concentrations and ISR during the OGTT. (h–j) Model-derived beta cell function parameters, including β-GS (h), potentiation factor (i) and β-RS (j). (k, l) Insulin clearance at fasting (k) and during the OGTT (l) in individuals with early-stage MASLD or control individuals. Data are mean ± SEM indicated by error bars or shaded areas. Group differences were analysed using Student’s t test. Repeated measures were analysed by two-way ANOVA followed by multiple pairwise comparisons using Sidak’s post hoc tests. *p<0.05, **p<0.01, ***p<0.001

Insulin clearance was lower at fasting (−0.8±0.2 l min⁻¹ m⁻², p<0.0001) and during the OGTT (−0.3±0.1 l min⁻¹ m⁻², p=0.005) in individuals with MASLD than in the control group (Fig. 2k, l).

Glucoregulatory hormones and lipid substrates

Plasma glucagon levels were higher in the MASLD group than the control group during fasting (+29±7 pg/ml, p=0.0003) and throughout the OGTT (AUC_0–120 +2.4±0.7 ng/ml × min, p=0.002) (Fig. 3a, b). Similarly, GLP-1 responses were consistently higher during the OGTT in the MASLD group (AUC_0–120 +1.2±0.4 nmol/l × min, p=0.001) (Fig. 3c, d). Plasma GIP and NEFA concentrations were almost identical between groups during fasting and showed similar changes in response to the oral glucose load (AUC_0–120 p=0.819 and p=0.521, respectively) (Fig. 3e–h).

Fig. 3.

Glucoregulatory hormones and lipid substrates. Profiles and AUCs during a 75 g OGTT of plasma glucagon (a, b), GLP-1 (c, d), GIP (e, f) and NEFA (g, h) in individuals with early-stage MASLD (n=20) and control individuals (n=22). Data are mean ± SEM indicated by error bars. Group differences were analysed using Student’s t test. Repeated measures were analysed by two-way ANOVA followed by multiple pairwise comparisons using Sidak’s post hoc tests. *p<0.05, **p<0.01, ***p<0.001

Whole-body and tissue-specific insulin sensitivity

In the fasted state and during the OGTT, participants with MASLD displayed lower whole-body insulin sensitivity than control participants as measured by either stable glucose isotopes (fasting GCl/Ins −37±16 [ml min⁻¹ kg⁻¹] × [nmol min⁻¹ l]⁻¹, p=0.025; GCl/Ins AUC_0–120 −4.5±2.1 [ml/kg] × [nmol/l]⁻¹, p=0.042, respectively), HOMA-IR (+1.4±0.4, p=0.001) or the Matsuda index (−3.6±1.2, p=0.004) (Fig. 4a–d).

Fig. 4.

Whole-body and tissue-specific insulin sensitivity. (a–d) Whole-body insulin sensitivity measured by either tracer-derived glucose clearance normalised to plasma insulin levels (GCl/Ins) at fasting (a) and during a 75 g OGTT (b), or by HOMA-IR (c) and the Matsuda index (d) in individuals with early-stage MASLD (n=20) and control individuals (n=22). (e–g) Hepatic insulin resistance measured by either tracer-derived EGP × ISR, both at fasting (e) and during the OGTT (f), or by the HIRI (g). (h) Adipose tissue insulin resistance measured by the Adipo-IR index. Data are mean ± SEM indicated by error bars. Group differences were analysed using Student’s t test. *p<0.05, **p<0.01, ***p<0.001

There were no group differences in hepatic insulin resistance assessed by either glucose tracers (fasting EGP × ISR +119±92 μmol min⁻¹ kg⁻¹ × pmol min⁻¹ m⁻², p=0.207; EGP × ISR AUC_0–120 μmol/kg × nmol/m² −9.2±8.7, p=0.296) or the surrogate index (HIRI +70±69 [mmol/l×h] × [pmol/l × h], p=0.066) (Fig. 4e–g). Adipose tissue insulin resistance was higher in the MASLD group than in the control group (Adipo-IR +15.7±5.9, p=0.008) (Fig. 4h).

Associations between metabolic variables and prevalent MASLD

In univariable logistic regression analysis, oral glucose absorption was identified as a significant predictor of MASLD (Table 2). For each 1-SD increase in RaO AUC_0–60, the OR for having MASLD was 2.11 (95% CI 1.04, 4.30, p=0.039). The receiver operating characteristic (ROC) analysis for RaO AUC_0–60 in predicting MASLD yielded an AUC of 0.659. No significant associations were observed with whole-body insulin sensitivity (GCl/Ins AUC_0–120) or hepatic insulin resistance (EGP × ISR AUC_0–120).

Table 2.

Multivariable associations between metabolic variables and prevalent MASLD

Variable	Univariable analysis	Multivariable model 1a	Multivariable model 2b	Multivariable model 3c
RaO AUC0–60	2.11 (1.04, 4.30)*	2.61 (1.10, 7.79)*	4.73 (1.53, 24.26)*	4.99 (1.44, 31.57)*
GCl/Ins AUC0–120	0.34 (0.10, 1.11)	0.17 (0.01, 1.04)	0.53 (0.04, 2.16)	1.32 (0.07, 18.33)
EGP × ISR AUC0–120	0.71 (0.37, 1.35)	0.43 (0.12, 1.21)	0.20 (0.03, 0.73)*	0.23 (0.02, 1.13)
Age	1.42 (0.75, 2.66)	1.01 (0.33, 3.07)	0.83 (0.21, 2.78)	1.30 (0.23, 8.07)
Sex, male	1.55 (0.45, 5.37)	4.61 (0.82, 36.23)	10.99 (0.81, 148.59)	12.52 (0.61, 257.97)
HbA1c	2.07 (1.03, 4.14)	1.10 (0.32, 4.17)	0.60 (0.14, 2.31)	0.33 (0.05, 1.53)
BMI	2.61 (1.22, 5.57)*	–	9.26 (1.89, 87.16)*	12.22 (1.92, 190.30)*
β-GS	0.70 (0.37, 1.34)	–	–	0.68 (0.19, 2.07)
Insulin clearance	0.27 (0.09, 0.79)*	–	–	0.22 (0.02, 1.42)

Data are presented as OR (95% CI) for MASLD for each 1-SD increase in the independent variable

^aModel 1: adjusted for age, sex and HbA_1c

^bModel 2: as for model 1, with further adjustment for BMI

^cModel 3: as for model 2, with additional inclusion of β-GS and insulin clearance as covariates

^*p<0.05

The relationship between glucose absorption and MASLD remained statistically significant in multivariable models adjusted for age, sex and HbA_1c (OR 2.61 [1.10, 7.79], p=0.048), after further adjustment for BMI (OR 4.73 [1.53, 24.26], p=0.022), and after additional inclusion of β-GS and insulin clearance as covariates (OR 4.99 [1.44, 31.57], p=0.036) (Table 2). Furthermore, the association between RaO and prevalent MASLD was confirmed in BMI-matched subset analyses (ESM Table 2).

Study 2

Participants

Ninety-one participants completed the OGTT with concurrent measurement of gastric emptying and were included in the analysis (48 men and 43 women, age 30 [23–40] years, BMI 24.6±3.4 kg/m², HbA_1c 34.9±3.1 mmol/mol (5.3±0.3%), T50 85.6±28.8 min and HSI 33 [29–36]) (ESM Table 3). Compared with Study 1 participants, this cohort were of younger age and had different racial/ethnic composition, better glucose tolerance and lower BMI and HSI. Twenty-seven Study 2 participants had HSI ≥36, 38 participants had HSI between 30 and 36, and 26 participants had HSI <30. Their clinical and metabolic characteristics are reported in Table 3. Compared with participants with HSI <30, those with HSI ≥36 were older and had higher BMI, ALT and γ-glutamyl transferase levels. Sex distribution, fasting blood glucose, HbA_1c, AST and alkaline phosphatase were comparable between groups.

Table 3.

Clinical and metabolic characteristics of Study 2 participants

Characteristic	HSI <30 (n=26)	30≤ HSI <36 (n=38)	HSI ≥36 (n=27)	p valuea
Women, n (%)	10 (38.5)	22 (57.9)	11 (40.7)	0.951
Age, years	27 (20–36)	30 (24–42)	36 (27–50)	0.022*
BMI, kg/m2	20.8±1.8	24.7±2.2	28.0±2.0	0.0001***
HbA1c, mmol/mol	34.5±2.5	34.8±3.0	35.2±3.8	0.413
HbA1c, %	5.3±0.2	5.3±0.3	5.4±0.4	0.418
Fasting blood glucose, mmol/l	5.1±0.5	5.1±0.4	5.3±0.5	0.296
1 h blood glucose, mmol/l	8.7±1.5	8.5±2.2	10.0±2.2	0.023*
2 h blood glucose, mmol/l	6.2±1.3	6.7±1.0	7.3±1.5	0.013*
AST, U/l	24 (21–30)	22 (20–25)	26 (22–30)	0.545
ALT, U/l	18 (13–22)	20 (16–27)	29 (23–47)	0.0001***
Alkaline phosphatase, U/l	75 (56–82)	68 (59–78)	78 (62–92)	0.400
γ-Glutamyl transferase, U/l	14 (11–19)	15 (11–19)	23 (15–36)	0.0001***
HSI	26.9±2.0	32.9±1.5	38.9±2.6	0.0001***
T50, min	86.5±32.3	89.8±23.5	78.7±31.6	0.378
Gastric retention 1 h post-OGTT, %	58.7±13.2	60.9±7.7	56.2±11.2	0.351

Data are presented as n (%), mean ± SD or median (IQR)

^aDifference between subgroups with HSI≥36 and HSI<30

^*p<0.05, ***p<0.001 (analysed by Fisher’s exact test, Student’s t test or Mann–Whitney test, as appropriate)

Glucose tolerance and gastric emptying

After the glucose drink, the blood glucose response was higher in the group with HSI ≥36 (group effect: p=0.010) (Fig. 5a); this group showed higher blood glucose levels at 1 h and 2 h post-OGTT than the group with HSI <30 (Table 3).

Fig. 5.

Relationships between glucose tolerance, gastric emptying and estimated hepatic steatosis. (a–c) Comparison of blood glucose responses to a 75 g OGTT (a), gastric half-emptying time (T50) (b) and gastric emptying at 1 h post-OGTT (c) between groups with HSI ≥36 (n=27), 30≤ HSI <36 (n=38) and HSI <30 (n=26). (d–f) Correlations of HSI with 1 h blood glucose (d), T50 (e) and gastric emptying at 1 h post-OGTT (f). Data are mean ± SEM. Group differences in blood glucose were analysed by two-way ANOVA. The differences in T50 and gastric emptying at 1 h post-OGTT between subgroups with HSI≥36 and HSI<30 were analysed using Student’s t test, with adjustment for age, sex, BMI and HbA_1c. The relationships were evaluated using Pearson correlation. *p<0.05

Between these two groups, there were no differences in T50 or gastric retention (expressed as % of baseline) at 1 h post-OGTT, either before (p=0.378 and p=0.351, respectively) or after (p=0.760 and p=0.733) adjustment for age, sex, HbA_1c, and BMI (Fig. 5b, c).

The HSI was related directly to 1 h blood glucose (r=0.220, p=0.034) but was unrelated to the T50 (p=0.509) or to the gastric retention at 1 h post-OGTT (p=0.589) (Fig. 5d–f).

Discussion

This study is the first to map postprandial glucose fluxes in individuals with early-stage MASLD. We demonstrated that, compared with healthy control individuals, individuals with MASLD and mild metabolic alterations, but without fibrosis, exhibited over 50% greater oral glucose absorption, resulting in higher plasma glucose and insulin responses to an oral glucose load. Moreover, early-stage MASLD was associated with impaired peripheral insulin sensitivity, marked by reduced GCl normalised to insulin and higher Adipo-IR, while hepatic insulin sensitivity and beta cell function remained largely preserved. Although gastric emptying is a key determinant of oral glucose absorption, we observed that there was no association between gastric emptying and estimated liver fat content in a second cohort of young, non-diabetic participants. Collectively, these observations support the concept that enhanced oral glucose absorption, likely at the intestinal level rather than due to rapid gastric emptying, represents an early feature of MASLD. Thus, glucose absorption appears to be a biologically plausible and actionable target for the prevention and management of MASLD.

Alterations in oral glucose handling marked by enhanced RaO may contribute to the onset and progression of MASLD through several interconnected mechanisms, as supported by preliminary clinical evidence [17, 18] and experimental studies [34]. Glucose absorbed from the gut is preferentially delivered to the liver via the portal vein, where excess glucose influx can stimulate triglyceride accumulation and promote MASLD progression from simple steatosis to steatohepatitis and fibrosis. Additionally, increased oral glucose absorption translates into higher plasma glucose excursions and sustained hyperinsulinaemia, which may further stimulate DNL and impair lipid export from the liver [34]. The molecular mechanisms may involve upregulation of key transcription factors such as carbohydrate-responsive element-binding protein (ChREBP) and sterol regulatory element-binding protein-1c (SREBP-1c), which upregulate the expression of lipogenic enzymes such as acetyl-CoA carboxylase (ACC) and fatty acid synthase (FAS) [35]. Rapid postprandial glucose elevation can also promote the formation of AGEs and reactive oxygen species (ROS), and hence stimulate the release of inflammatory cytokines [9, 34, 36]. Moreover, the persistent activation of DNL may generate lipotoxic intermediates such as diacylglycerols and ceramides, further activating inflammatory signalling pathways [34]. Oral glucose absorption can be modulated by a variety of nutritional [37–39], pharmacological [11, 15, 40] and surgical interventions [41], aimed at interfering with gastric emptying, carbohydrate breakdown and transporter expression or activity. A more comprehensive understanding of MASLD pathogenesis in relation to alterations in RaO will be crucial for informing the design of future targeted clinical trials.

The RaO is regulated by three interrelated mechanisms, namely gastric emptying, intestinal glucose absorption and splanchnic glucose uptake, which operate sequentially to modulate postprandial glucose flux. The finding that T50, as well as gastric retention at 1 h post-OGTT, was similar in individuals with low and high HSI suggests that gastric emptying is not a primary driver of liver fat accumulation. This narrows the focus towards alternative plausible mechanisms of elevated RaO in MASLD, particularly enhanced intestinal glucose absorption. Individuals with MASLD often consume high-carbohydrate, obesogenic diets, which may upregulate duodenal GLUT expression [42] and drive mucosal hyperplasia [43]. Supporting this view, prior clinical evidence shows that increased duodenal SGLT1 expression correlates with hepatic steatosis and fibrosis [9], whereas genetically reduced SGLT1 activity is linked to lower risk of MASLD [10]. Moreover, pharmacological SGLT1 inhibition prevents diet-induced steatosis in mice [11]. Splanchnic glucose uptake, representing first-pass glucose extraction by the gut and liver, has not been explored due to the difficulty of in vivo measurement [44]. Although reduced hepatic glucose uptake during hyperglycaemic clamp studies has been negatively associated with hepatic fat content [45], the specific contribution of splanchnic glucose uptake to oral glucose absorption in individuals with MASLD remains to be determined.

Early defects in insulin sensitivity are central to the pathogenesis of MASLD [3, 34]. However, emerging evidence challenges the notion that hepatic insulin resistance is the primary upstream event responsible for initiating intrahepatic fat accumulation. We observed that whole-body insulin actions on tracer-derived GCl were substantially impaired in early-stage MASLD, while hepatic responses to fasting and post-load insulin levels on EGP remained relatively intact, indicating peripheral rather than hepatic insulin resistance. Consistently, Sabatini et al [46] reported that tracer-measured EGP increases with biopsy-proven liver fibrosis and inflammation, driven by hepatic and extrahepatic insulin resistance, but not with the degree of steatosis. Likewise, Basset-Sagarminaga et al [45] did not find a significant association between insulin-induced EGP suppression and hepatic fat content. Differences in post-load EGP suppression in our study can be explained by higher plasma insulin levels in the MASLD group, resulting from reduced insulin clearance despite similarly stimulated insulin secretion, in line with previous evidence [4, 47]. This occurred alongside comparable declines in glucagon levels, which nevertheless remained consistently higher in the MASLD group. Hyperglucagonaemia has been consistently associated with hepatic steatosis and may reflect a disruption in the liver–alpha cell axis mediated by amino acids (i.e. glucagon resistance) [48]. On the contrary, identical NEFA levels throughout the OGTT exclude a substantial role for lipolytic control on gluconeogenesis in this setting, which is noteworthy given that EGP is largely influenced by NEFA flux to the liver [49].

An unexpected observation was the selective increase in total GLP-1 responses in MASLD, whereas GIP responses were comparable between groups. This dissociation argues against generalised incretin hypersecretion and instead suggests altered nutrient sensing or L cell-specific regulation. Given that GLP-1 and GLP-2 are co-secreted in equimolar amounts from enteroendocrine L-cells, enhanced GLP-1 secretion likely indicates concomitant augmentation of GLP-2 release. GLP-2 exerts well-established intestinotrophic effects and has been shown to increase mRNA expression of SGLT1 (also known as SLC5A1) and glucose transport capacity, potentially via enhanced epithelial growth and enterocyte maturation along the crypt–villus axis [50]. Whether the amplified GLP-1/2 axis represents a compensatory adaptation to increased intestinal glucose absorption or contributes causally to altered intestinal–hepatic crosstalk in MASLD remains to be determined.

This study has some limitations that should be acknowledged. First, the cross-sectional design precludes establishing causality and temporal direction of the relationships between enhanced RaO and MASLD. Nonetheless, prior mechanistic evidence and current results provide a valuable framework for future longitudinal and interventional studies aimed at confirming and expanding these associations. Second, liver steatosis and fibrosis were assessed through methods that are either operator-dependent or indirect, and therefore subject to potential misclassification. To minimise variability and measurement bias, all ultrasound examinations were performed by experienced sonographers and surrogate indices were selected from those with well-established validation and strong support in clinical guidelines. Furthermore, hepatic steatosis was assessed using ultrasound in Study 1 and estimated using the HSI in Study 2. The use of different methods introduces heterogeneity that should be considered when interpreting cross-cohort observations. Third, the sample size was relatively small. However, it was adequate to detect clear-cut differences between groups, and the observed effect size supports the robustness of the main findings. Fourth, the study includes two independent cohorts with differing characteristics, especially with respect to BMI, preventing cross-cohort observations and the ability to draw conclusions about the relationship between RaO and gastric emptying in MASLD. Fifth, although associations persisted after BMI adjustment and matching, residual confounding by adiposity cannot be fully excluded. Finally, RaO AUC_0–120 did not account for the entire 75 g oral glucose load, the remaining fraction likely reflecting ongoing absorption beyond 120 min, splanchnic extraction, temporary glycogen storage or metabolic conversion prior to systemic appearance, which have not been quantified. Thus, we cannot exclude that glucose uptake beyond the observation window ultimately contributes directly or indirectly to hepatic substrate availability.

In conclusion, our studies provide the first direct clinical evidence of enhanced RaO in people with early-stage MASLD, using a quasi-physiological dual-tracer method. The lack of apparent correlation between the gastric half-emptying time and estimated hepatic steatosis suggests that this phenomenon is likely to occur at the intestinal level, in concordance with prior histological evidence. These results uncover a novel pathogenetic mechanism and suggest a potential actionable target for early prevention and treatment of MASLD.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

β-GS

Beta cell glucose sensitivity

β-RS

Beta cell rate sensitivity

ALT

Alanine aminotransferase

AST

Aspartate aminotransferase

DNL

De novo lipogenesis

EGP

Endogenous glucose production

FIB-4

Fibrosis-4 index

GCl

Glucose clearance

GCl/Ins

Fasting GCl/plasma insulin ratio

GIP

Glucose-dependent insulinotropic polypeptide

GLP-1

Glucagon-like peptide-1

HIRI

Hepatic insulin resistance index

HSI

Hepatic steatosis index

ISR

Insulin secretion rate

ISR@5

ISR at a fixed glucose concentration of 5 mmol/l

MASLD

Metabolic dysfunction-associated steatotic liver disease

RaO

Rate of appearance of oral ingested glucose

SGLT1

Sodium–glucose cotransporter 1

SLD

Steatotic liver disease

T50

Gastric half-emptying time

Acknowledgements

We gratefully acknowledge the contributions of all study volunteers, whose involvement made this research possible. We also thank A. Mengozzi and G. Nesti at the University of Pisa (Pisa, Italy), M. Borg and Y. Sun at the University of Adelaide (Adelaide, Australia), and all clinical staff, nurses and research coordinators at the two institutions for their assistance with participant recruitment, data collection and study implementation. Preliminary findings from this study were previously presented at the EASD Annual Meeting, held in Vienna, Austria, in September 2025, and the Italian Society of Internal Medicine (SIMI) Annual Meeting, held in Rimini, Italy, in October 2025. The current manuscript provides a comprehensive and updated analysis of the full dataset, incorporating additional results and extended discussion not included in the earlier presentations.

Data availability

The data that support the study findings are available from the corresponding authors on reasonable request starting from the date of publication and will remain accessible for a minimum of 10 years, in line with institutional policies. Access will be granted to qualified researchers to address research questions that are consistent with the scope of the original study and meet ethical and legal requirements.

Funding

Open Access funding enabled and organized by CAUL and its Member Institutions. We acknowledge financial support under the National Recovery and Resilience Plan (PNRR), Mission 4, Component 2, Investment 1.1, Call for tender no. 104 published on 2 February 2022 by the Italian Ministry of University and Research (MUR), funded by the European Union – NextGenerationEU – Project Title ‘Exploring the role of SGLT-1 driven intestinal glucose absorption and altered intestinal mucosa integrity in the pathogenesis of non-alcoholic fatty liver disease’ – CUP I53D23005180001, F53D23007130006 and B53D23021760006 – grant assignment decree no. 2022LWXFM4 dated 20 July 2023. Data obtained at the University of Adelaide were funded by an Australian National Health and Medical Research Council (NHMRC) Ideas Grant awarded to CKR (ID 2020454). CX has been supported by an Australian Diabetes Society Skip-Martin Early Career Fellowship. TW has been supported by the Australian Medical Research Future Fund (MRFF) (ID: MRFCDDM000009). CKR is supported by the Michell Bequest Foundation.

Authors’ relationships and activities

The authors declare that there are no relationships or activities that might bias, or be perceived to bias, their work.

Contribution statement

The study was conceived by DT and designed by DT, TW, AM, CKR and AN. Funding was obtained by DT, TW, TVF, RB, CKR and AN. All authors contributed to the acquisition, analysis or interpretation of the data. Statistical analysis and mathematical modelling were performed by DT, CX and AM. DT drafted the manuscript, and all authors critically revised it for important intellectual content. CKR and AN supervised the work. All authors approved the final version of the manuscript. DT and AN take responsibility for the integrity of the work as a whole.