Elevated liver fat content is independently associated with insulin resistance in people with type 1 diabetes Authors: Jonathan Mertens, Maarten Spinhoven, Rie Braspenning, Floris Vanhevel, Eveline Dirinck, Christophe De Block, Sven Francque Published: 2026 ABSTRACT Background & AimsInsulin resistance is associated with metabolic dysfunction-associated steatotic liver disease (MASLD) in the general population, but its relationship with MASLD in individuals with type 1 diabetes (T1D) remains insufficiently defined.MethodsWe performed a 40 mU/m2/min hyperinsulinaemic-euglycaemic clamp to quantify the whole-body glucose disposal rate (M-value, mg/kg/min). Furthermore, we assessed insulin sensitivity using 13C-glucose breath test and estimated glucose disposal rate. Liver fat content (LFC) was measured by magnetic resonance spectroscopy. We also performed vibration-controlled transient elastography. Participants were stratified 1:1 by MASLD status based on mean LFC >/≤5.56%.ResultsWe studied 39 adults with T1D (diabetes duration 27 ± 13 years; haemoglobin A1c 6.9 [6.5–7.8]%). Mean M-value was 4.41 ± 2.37 mg/kg/min. LFC was higher in individuals with MASLD than in those without MASLD (8.2 [6.5–18.7]% vs 3.0 [2.3–3.6]%, p <0.001). The M-value was significantly lower in the MASLD group (2.92 ± 1.39 vs 6.41 ± 2.46 mg/kg/min, p <0.001). LFC strongly correlated with the M-value (rs −0.72, p <0.001). In linear regression, the M-value independently predicted log-transformed LFC (B −0.22, p <0.001; R2 = 0.47). Lower M-value was associated with MASLD (adjusted odds ratio [aOR] 3.12, 95% CI 1.17–8.37, p = 0.024). Both 13C-glucose breath test (aOR 1.54, 95% CI 1.03–2.33, p = 0.033) and estimated glucose disposal rate (aOR 1.61, 95% CI 1.09–2.38, p = 0.016) were also independently associated with MASLD.ConclusionsInsulin resistance is independently associated with MASLD in adults with T1D.Impact and implicationsThis study addresses the limited understanding of how MASLD relates to insulin resistance in T1D. We show a robust association between LFC and insulin resistance using gold standard and practical non-invasive measures. These findings may help clinicians identify metabolically at-risk individuals with T1D; however, further research is warranted to clarify whether screening and targeted intervention improve outcomes. Introduction Metabolic dysfunction-associated steatotic liver disease (MASLD) is increasingly prevalent, affecting approximately a quarter of the global population.1 The hallmark of MASLD is an increased liver lipid content in the presence of minimally one clinical cardiometabolic risk factor.2 MASLD encompasses a spectrum of pathologies ranging from isolated steatosis (fatty liver) to steatohepatitis (metabolic dysfunction-associated steatohepatitis [MASH]), in the absence or presence of liver fibrosis, which can progress into cirrhosis. The prevalence of MASLD is notably lower in individuals with type 1 diabetes (T1D, 16–20%) than in the general population.[3], [4], [5] Historically, individuals with T1D exhibited lower susceptibility to insulin resistance; however, there is a global increase in a clinical phenotype characterised by the coexistence of autoimmune diabetes and significant insulin resistance, commonly termed as “double diabetes”.6,7 Although the updated nomenclature and definition of MASLD enables clinical determinants of insulin resistance to diagnose MASLD, direct evidence of the link between insulin resistance and liver steatosis in T1D is limited; this is mainly because the quantification of insulin resistance poses considerable challenges. The gold standard, hyperinsulinaemic-euglycaemic clamp (HEC), is invasive and requires an expert setting, making it unsuitable for routine clinical practice. Recently, a simple and sensitive 13C-glucose breath test (13C-GBT) has been reportedly used to directly measure whole-body insulin sensitivity in insulin-dependent glucose metabolism.8 An alternative index, the estimated glucose disposal rate (eGDR), has also been proposed.9 In this study, we aim at describing the relationship between insulin resistance and liver steatosis in individuals with T1D using the established gold standard methods. Secondary objectives are to confirm this correlation using non-invasive alternatives for insulin resistance more suited for routine clinical practice (13C-GBT and eGDR), to explore whether genetic polymorphisms that are commonly linked to MASLD moderate the relationship between insulin resistance and MASLD, and to evaluate the independent association between insulin resistance and the presence of MASLD. Materials and methods Study participants The study population comprised adults diagnosed with T1D attending the outpatient diabetes clinic at the Antwerp University Hospital (Antwerp, Belgium). Diagnosis of T1D was confirmed by documented positivity for beta-cell autoantibodies and low C-peptide levels (<0.2 nmol/L).10 Participants were recruited from two ongoing prospective trials: NCT04623320, which validates the 13C-GBT gainst the HEC; and NCT04664036, which investigates the natural prevalence, characteristics, and genetic analyses of single-nucleotide polymorphisms (SNPs) of MASLD in individuals with T1D.3,8 Individuals who consented to participation in both parent studies were invited to participate in the present analytical sub-study. This study, therefore, represents a predefined sub-analysis of prospectively collected data obtained under ethical approvals 20/40/515 and 18/32/361. Given the exploratory design, recruitment was prospectively planned to include 20 participants with MASLD and 20 without MASLD, based on feasibility considerations and anticipated effect size for the primary metabolic outcome. Eligible individuals attending the outpatient clinic during the recruitment period were screened prospectively. To ensure adequate representation of MASLD cases, individuals with clinical features suggestive of visceral adiposity were more actively approached during recruitment. Final group allocation, however, was based exclusively on magnetic resonance spectroscopy (MRS)-determined liver fat content (LFC). No matching for demographic or metabolic characteristics was performed. Recruitment for this analysis was closed once 20 participants per group with complete datasets were included. All participants were screened for exclusion criteria before metabolic testing. Exclusion criteria were: (1) liver disease other than MASLD; (2) daily alcohol intake exceeding 20 g for women or 30 g for men, assessed using a standardised questionnaire; (3) contraindications to MRS (e.g. implantable devices or claustrophobia); (4) pregnancy or lactation; and (5) brittle diabetes or haemoglobin A1c (HbA1c) >9.0%, to avoid the confounding impact of glucotoxicity on peripheral insulin sensitivity.11 Participants first underwent abdominal ultrasound and vibration-controlled transient elastography (VCTE). Based on the presence or absence of hepatic hyperechogenicity relative to the renal cortex, participants were provisionally categorised as MASLD or non-MASLD. Subsequently, we quantified LFC by MRS. An LFC >5.56% was used to define MASLD.12 In cases of discordance between ultrasound and MRS findings, final allocation was determined solely by MRS results. The study protocol complied with the Declaration of Helsinki and was approved by the local ethics committee of the Antwerp University Hospital (20/40/515 and 18/32/361). All participants provided written informed consent before participation. Baseline characteristics All participants underwent anthropometric evaluations (age, height, length, BMI, waist circumference [WC]), and measurement of systolic and diastolic blood pressure, i.e. the lowest value from three consecutive automated measurements following a 5-min supine rest. Blood laboratory tests included HbA1c, creatinine, estimated glomerular filtration rate computed through the Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) equation, liver enzymes (alanine aminotransferase [ALT], aspartate aminotransferase [AST], gamma-glutamyl transferase [GGT]), and lipid profile (triglycerides, total cholesterol, HDL cholesterol [HDL-c], LDL cholesterol [LDL-c]). The total daily dose of insulin (TDI) per kilogram of bodyweight per 24 h was determined by either the sum of daily administered units for injection therapy or by calculating a 30-day average for insulin pump users. Genotyping We performed whole-blood DNA extraction using standard procedures.13 We used predesigned TaqMan Genotyping assays (Thermo Fisher Scientific, Waltham, MA, USA) to genotype the desired SNPs: rs1260326 (glucokinase regulatory protein [GCKR]), rs641738 (O-acyltransferase domain [MBOAT7]), rs738409 (patatin-like phospholipase domain-containing protein 3 [PNPLA3]), and rs58542926 (transmembrane 6 superfamily 2 [TM6SF2]). The TaqMan assays were used as described in the manufacturer’s protocol, on a Bio-Rad CFX384 Real-Time 384-well qPCR Detection System (Bio-Rad, Hercules, CA, USA). Each variant was coded additively as zero, one, or two copies of the risk allele. To evaluate the impact of genetic predisposition to liver steatosis, we exploited the polygenic risk score for hepatic fat content (PRS-HFC) calculated as follows: 0.266 ∗ number of risk alleles for PNPLA3 + 0.274 ∗ number of risk alleles for TM6SF2 + 0.065 ∗ number of risk alleles for GCKR + 0.063 ∗ number of risk alleles for MBOAT7.[14], [15], [16] Given the small sample size and because the PNPLA3 and TM6SF2 variants are more strongly associated with LFC in the general population, a simplified PRS (PRS-HFC2) was also calculated as previously published: 0.266 ∗ number of risk alleles for PNPLA3 + 0.271 ∗ number of risk alleles for TM6SF2.14,15 Controlled attenuation parameter™ All participants underwent VCTE after an overnight fast on the day of the initial ultrasound screening. The performing investigator (JM) had received prior certification by Echosens (Paris, France). We performed VCTE using the FibroScan® 502 device (Echosens). The FibroScan® device had both M and XL probes, and the choice of the probe was determined by the Automatic Probe Selection algorithm based on the skin-liver capsule distance. Following the EASL guideline, we defined liver steatosis as a controlled attenuation parameter (CAP™) value above 275 dB/m, regardless of the probe used, with a sensitivity and positive predictive value exceeding 90%.17 Liver stiffness measurement (LSM) was also performed, and LSM >8.0 kPa was defined as a high risk of clinically significant liver fibrosis. Quantification of liver steatosis We conducted imaging quantification of liver steatosis using a 1.5T MAGNETOM Aera magnetic resonance tomograph (Siemens Healthcare, Erlangen, Germany) equipped with syngo MR E11 software running the clinical application LiverLab. LFC was measured within three regions of interest, and a mean LFC value >5.56% was used to define the presence of liver steatosis.12 A single certified expert radiology operator (FVH) performed all magnetic imaging assessments, and a certified radiologist (MS) interpreted all images and performed the scoring. Both were blinded to all other participant data. Measurement of whole-body insulin sensitivity We conducted HEC after an overnight fast, according to the protocol of DeFronzo et al.18 Two intravenous catheters were inserted into both arms for insulin (Actrapid®, Novo Nordisk, Bagsvaerd, Denmark) and glucose infusion, as well as blood sampling. The insulin infusion started at T = 0 with an 8-min priming dose of insulin (80 mU/m2 body surface area/min) and was then maintained at a rate of 40 mU/m2 body surface area/min for a minimum of 180 min. If blood glucose levels ranged 90–180 mg/dl at the start of HEC, a 40 mU/m2 body surface area/min insulin infusion was given to achieve euglycaemia at the start and to achieve complete suppression of endogenous glucose production (EGP) by the liver. The infusion was then halted, and blood glucose was monitored every 5 min until it reached 90 mg/dl, which was designated as T = 0. On average, this procedure took 30 min. Individuals with starting glycaemia exceeding 180 mg/dl were rescheduled. Blood glucose was clamped at 90 mg/dl using appropriate glucose infusion, guided by blood samples taken every 5 min. Serum glucose and insulin levels were measured at baseline and at 30-min intervals thereafter. The M-value (mg/kg/min) was calculated based on the glucose infusion rate during the final 30 min of the study and represents the glucose disposal rate in hyperinsulinaemic state. A cut-off value of <4.9 mg/kg/min was used to identify individuals with clinically significant insulin resistance, based on Tam et al.19 After HEC, participants were supervised and could only leave when stable or increasing glucose levels were documented. 13C-glucose breath test Participants were instructed how to perform a 13C-GBT, during which they consumed a solution containing 25 mg of universally labelled D-glucose (U-13C6; Cambridge Isotope Laboratories Inc., Andover, MA, USA), dissolved in water and 12 g of orange-flavoured fructose, following previously described procedures.8,20 Participants were instructed to collect two baseline breath samples, followed by one sample every 15 min after ingestion of the glucose mixture up to 150 min after ingestion. Pre-test glycaemic conditions were maintained like those before the HEC test. A patient-tailored bolus of rapid-acting insulin was administered before ingestion to match the 12 g of carbohydrates in the test solution. Baseline breath samples were compared with samples obtained at T = 90 min after ingestion using breath sample tubes that have no gas permeability for up to 90 days after sealing (Campro Scientific GmbH, Berlin, Germany). This absolute increase in 13CO2 compared with baseline (expressed as Δ13CO2 [δ‰]) was used to measure insulin-dependent glucose metabolism. We measured 13CO2 using an isotope ratio mass spectrometer (ABCA2 breath test, Sercon Limited, Crewe, UK). The quality of the breath test samples was evaluated by visualising the cumulative expiration of 13CO2 and calculating the expired fraction of CO2. Breath samples with an expiration of CO2 <2.0% were considered unacceptable and led to the exclusion of the participant. Quality control was conducted by specialised laboratory personnel blinded to all other study results of the participants. Estimated glucose disposal rate We calculated the eGDR using the following formula:[1]eGDR(mg/kg/min)=21.158+(−0.09×WC)+(−3.407×hypertension)+(−0.551×HbA1c%) Hypertension was defined as elevated systolic (>140 mmHg) or diastolic (>90 mmHg) blood pressure, or the use of antihypertensive drugs.9 Statistical analysis Continuous data were reported as mean ± SD or median (IQR), and categorical data were presented as frequencies and percentages. Normality of continuous data was assessed using the Kolmogorov-Smirnov test, and visualising histograms and Q-Q plot. Differences between two groups were analysed using the unpaired t test or Mann-Whitney U test, and comparisons among multiple groups were performed using one-way ANOVA or Kruskal-Wallis test. Correlations were evaluated using Spearman rank or Pearson correlation test according to distribution. Categorical data were compared using the Chi-square or Fisher's exact tests. Linear regression was used to assess the correlation between insulin resistance and LFC as continuous variables. Logistic regression was used to identify factors independently associated with LFC >5.56% and calculate odds ratios (ORs) with 95% CIs. Collinearity of predictors was assessed by calculating the variance inflation factor. This pilot study was designed to generate preliminary data on differences in M-value between individuals with and without MASLD and to explore the relationship between LFC and insulin resistance. Thus, no a priori power calculation was performed. Given its exploratory design, recruitment was prospectively planned to include 20 participants with MASLD and 20 without MASLD based on feasibility of recruitment and estimated effect size for the primary outcome. Based on this sample size, the study has sufficient statistical power to detect large between-group differences in M-value (approximately Cohen’s d ≥0.8) and moderate-to-strong correlations between LFC and M-value (r ≥0.5). Statistical significance was set at p <0.05 for all analyses. Data were analysed using SPSS Statistics version 27 (IBM Corp., Armonk, NY, USA). Graphs were created using GraphPad Prism 8.4 (GraphPad Software, Inc., San Diego, CA, USA). Results Baseline characteristics A total of 39 adults were included in the final analysis, of whom 19 were in the MASLD group. One individual withdrew from the study after completion of the HEC in the MASLD group. All cases of hepatic steatosis identified by ultrasound were confirmed by liver fat quantification using MRS (Fig. 1). All participants with liver steatosis had MASLD according to the diagnostic criteria. Results of 13C-GBT were eligible for analysis in 28 participants. Baseline characteristics of the study participants are summarised in Table 1. Participants with MASLD exhibited considerably higher BMI, greater WC, elevated serum triglycerides, lower HDL levels, and higher systolic blood pressure, all of which are components of the metabolic syndrome. Liver stiffness was higher in the MASLD group, and the presence of liver fibrosis did not differ between groups. Individuals with co-existing MASLD exhibited significantly more insulin resistance, as confirmed by a lower HEC-derived M-value (2.92 ± 1.39 vs 6.41 ± 2.46 mg/kg/min, p <0.001). Individuals with MASLD also had worse glycaemic control, as illustrated by higher HbA1c levels, despite higher TDI administration.Fig. 1Study selection flowchart.All participants who underwent ultrasound screening also underwent vibration-controlled transient elastography using FibroScan® during the same screening session. When the initial screening reached the maximum capacity for each group (n = maximum 20 for each group), any additional eligible participants were excluded from those groups (n = 25). The operators performing magnetic resonance spectroscopy (MRS) were unaware of the prior screening results. There were no instances of false-positive or false-negative ultrasound-based group allocations. To ensure accurate 13C-glucose breath sample collection, a CO2 threshold of ≥2% was established, and samples with CO2 levels below this threshold were excluded from further analysis. ∗One person had withdrawn from the study because of personal reasons. 13C-GBT, 13C-glucose breath test; MASLD, metabolic dysfunction-associated steatotic liver disease, HEC, hyperinsulinaemic-euglycaemic clamp; T1D, type 1 diabetes.Table 1Baseline characteristics of the overall cohort and stratified by the presence or absence of MASLD, as determined by MRS liver fat quantification.Table 1CohortMASLDNo MASLDp valueN391920N/ALFC (%)5.3 (2.9–8.2)8.2 (6.5–18.7)3.0 (2.3–3.6)<0.001Demographic and anthropometric parametersAge (yr)46 ± 1647 ± 2045 ± 120.608Male sex25 (61.4)15 (78.9)10 (50.0)0.096Diabetes duration (yr)27 ± 1327 ± 1326 ± 130.783WC (cm)100 ± 18108 ± 1592 ± 160.002BMI (kg/m2)28.4 ± 5.630.6 ± 5.626.3 ± 4.80.015SBP (mmHg)133 ± 14139 ± 14127 ± 100.004DBP (mmHg)79 ± 879 ± 779 ± 100.909Arterial hypertension25 (64.1)14 (73.7)11 (55.0)0.224Diabetes-related parametersTDI (IU/kg/24 h)0.79 (0.61–1.03)0.93 (0.68–1.35)0.62 (0.58–0.78)0.001HbA1c (%)6.9 (6.5–7.8)7.6 (6.9–7.8)6.6 (6.4–7.3)0.024HbA1c (mmol/mol)52 (48–62)59 (52–62)49 (47–56)0.026Laboratory parametersCreatinine (mg/dl)0.78 (0.65–0.89)0.80 (0.72–0.91)0.72 (0.64–0.79)0.057eGFR (ml/min/1.73 m2)106 (93–122)99 (87–126)107 (98–114)0.055AST (IU/L)23 (21–29)23 (22–29)22 (18–30)0.311ALT, IU/L)25 (19–35)27 (23–44)21 (15–33)0.070GGT (IU/L)21 (15–37)29 (17–37)19 (13–35)0.175TG, (mg/dl)88 (59–150)136 (115–182)53 (43–65)<0.001HDL (mg/dl)49 (43–55)46 (38–50)55 (47–72)<0.001LDL (mg/dl)87 ± 2987 ± 3184 ± 280.759GenotypingPNPLA3 ≥1 risk allele16 (41.0)9 (47.4)7 (35.0)0.433TM6SF2 ≥1 risk allele8 (20.5)7 (36.8)2 (10.0)0.065GCKR ≥1 risk allele22 (65.4)9 (47.4)13 (65.0)0.267MBOAT7 ≥1 risk allele20 (51.3)10 (52.6)11 (55.0)0.882PRS-HFC0.19 (0.07–0.54)0.33 (0.07–0.60)0.13 (0.07–0.38)0.214PRS-HFC20.00 (0.00–0.53)0.27 (0.00–0.54)0.00 (0.00–0.27)0.101Insulin sensitivity indicesM-value (mg/kg/min)4.41 ± 2.372.92 ± 1.396.41 ± 2.46<0.001M-value above threshold†23 (59.0)17 (89.5)6 (30.0)<0.001Δ13CO2 (δ‰)∗5.32 ± 1.934.34 ± 1.666.46 ± 1.610.002eGDR (mg/kg/min)5.34 (4.34–6.45)4.36 (3.99–5.14)6.44 (5.41–9.74)0.004Transient elastography-derived parametersCAP™ (dB/m)278 ± 68327 ± 48229 ± 47<0.001LSM (kPa)5.4 (4.5–6.3)6.0 (5.4–7.5)5.1 (4.3–5.6)0.007LSM >8.0 (kPa)2 (5.6)2 (11.1)0 (0.0)0.486Baseline characteristics are expressed as mean ± SD or median (IQR) according to distribution or frequency (%). Differences between two groups were analysed using the unpaired t test or Mann-Whitney U test, according to distribution. Categorical data were compared using the Chi-square or Fisher's exact tests. Statistical significance was set at p <0.05 for all analyses.Δ13CO2, % of 13CO2 compared with baseline excreted after 90 min; ALT, alanine aminotransferase; AST, aspartate aminotransferase; CAP™, controlled attenuation parameter; DBP, diastolic blood pressure; eGFR, estimated glomerular filtration rate; GGT, gamma-glutamyl transferase; HbA1c, haemoglobin A1c; HDL-c, HDL cholesterol; LDL-c, LDL cholesterol; LFC, liver fat content; LSM, liver stiffness measurement; M-value, glucose disposal rate; MASLD, metabolic dysfunction-associated steatotic liver disease; MRS, magnetic resonance spectroscopy; PRS-HFC, polygenic risk score for hepatic fat content; SBP, systolic blood pressure; TDI, total daily dose of insulin; TG, triglycerides; WC, waist circumference.∗13C-glucose breath test results available in 28 individuals: 15 in the MASLD group and 13 in the no-MASLD group.†Based on M-value <4.9 mg/kg/min. Study selection flowchart. All participants who underwent ultrasound screening also underwent vibration-controlled transient elastography using FibroScan® during the same screening session. When the initial screening reached the maximum capacity for each group (n = maximum 20 for each group), any additional eligible participants were excluded from those groups (n = 25). The operators performing magnetic resonance spectroscopy (MRS) were unaware of the prior screening results. There were no instances of false-positive or false-negative ultrasound-based group allocations. To ensure accurate 13C-glucose breath sample collection, a CO2 threshold of ≥2% was established, and samples with CO2 levels below this threshold were excluded from further analysis. ∗One person had withdrawn from the study because of personal reasons. 13C-GBT, 13C-glucose breath test; MASLD, metabolic dysfunction-associated steatotic liver disease, HEC, hyperinsulinaemic-euglycaemic clamp; T1D, type 1 diabetes. Table 1Baseline characteristics of the overall cohort and stratified by the presence or absence of MASLD, as determined by MRS liver fat quantification.Table 1CohortMASLDNo MASLDp valueN391920N/ALFC (%)5.3 (2.9–8.2)8.2 (6.5–18.7)3.0 (2.3–3.6)<0.001Demographic and anthropometric parametersAge (yr)46 ± 1647 ± 2045 ± 120.608Male sex25 (61.4)15 (78.9)10 (50.0)0.096Diabetes duration (yr)27 ± 1327 ± 1326 ± 130.783WC (cm)100 ± 18108 ± 1592 ± 160.002BMI (kg/m2)28.4 ± 5.630.6 ± 5.626.3 ± 4.80.015SBP (mmHg)133 ± 14139 ± 14127 ± 100.004DBP (mmHg)79 ± 879 ± 779 ± 100.909Arterial hypertension25 (64.1)14 (73.7)11 (55.0)0.224Diabetes-related parametersTDI (IU/kg/24 h)0.79 (0.61–1.03)0.93 (0.68–1.35)0.62 (0.58–0.78)0.001HbA1c (%)6.9 (6.5–7.8)7.6 (6.9–7.8)6.6 (6.4–7.3)0.024HbA1c (mmol/mol)52 (48–62)59 (52–62)49 (47–56)0.026Laboratory parametersCreatinine (mg/dl)0.78 (0.65–0.89)0.80 (0.72–0.91)0.72 (0.64–0.79)0.057eGFR (ml/min/1.73 m2)106 (93–122)99 (87–126)107 (98–114)0.055AST (IU/L)23 (21–29)23 (22–29)22 (18–30)0.311ALT, IU/L)25 (19–35)27 (23–44)21 (15–33)0.070GGT (IU/L)21 (15–37)29 (17–37)19 (13–35)0.175TG, (mg/dl)88 (59–150)136 (115–182)53 (43–65)<0.001HDL (mg/dl)49 (43–55)46 (38–50)55 (47–72)<0.001LDL (mg/dl)87 ± 2987 ± 3184 ± 280.759GenotypingPNPLA3 ≥1 risk allele16 (41.0)9 (47.4)7 (35.0)0.433TM6SF2 ≥1 risk allele8 (20.5)7 (36.8)2 (10.0)0.065GCKR ≥1 risk allele22 (65.4)9 (47.4)13 (65.0)0.267MBOAT7 ≥1 risk allele20 (51.3)10 (52.6)11 (55.0)0.882PRS-HFC0.19 (0.07–0.54)0.33 (0.07–0.60)0.13 (0.07–0.38)0.214PRS-HFC20.00 (0.00–0.53)0.27 (0.00–0.54)0.00 (0.00–0.27)0.101Insulin sensitivity indicesM-value (mg/kg/min)4.41 ± 2.372.92 ± 1.396.41 ± 2.46<0.001M-value above threshold†23 (59.0)17 (89.5)6 (30.0)<0.001Δ13CO2 (δ‰)∗5.32 ± 1.934.34 ± 1.666.46 ± 1.610.002eGDR (mg/kg/min)5.34 (4.34–6.45)4.36 (3.99–5.14)6.44 (5.41–9.74)0.004Transient elastography-derived parametersCAP™ (dB/m)278 ± 68327 ± 48229 ± 47<0.001LSM (kPa)5.4 (4.5–6.3)6.0 (5.4–7.5)5.1 (4.3–5.6)0.007LSM >8.0 (kPa)2 (5.6)2 (11.1)0 (0.0)0.486Baseline characteristics are expressed as mean ± SD or median (IQR) according to distribution or frequency (%). Differences between two groups were analysed using the unpaired t test or Mann-Whitney U test, according to distribution. Categorical data were compared using the Chi-square or Fisher's exact tests. Statistical significance was set at p <0.05 for all analyses.Δ13CO2, % of 13CO2 compared with baseline excreted after 90 min; ALT, alanine aminotransferase; AST, aspartate aminotransferase; CAP™, controlled attenuation parameter; DBP, diastolic blood pressure; eGFR, estimated glomerular filtration rate; GGT, gamma-glutamyl transferase; HbA1c, haemoglobin A1c; HDL-c, HDL cholesterol; LDL-c, LDL cholesterol; LFC, liver fat content; LSM, liver stiffness measurement; M-value, glucose disposal rate; MASLD, metabolic dysfunction-associated steatotic liver disease; MRS, magnetic resonance spectroscopy; PRS-HFC, polygenic risk score for hepatic fat content; SBP, systolic blood pressure; TDI, total daily dose of insulin; TG, triglycerides; WC, waist circumference.∗13C-glucose breath test results available in 28 individuals: 15 in the MASLD group and 13 in the no-MASLD group.†Based on M-value <4.9 mg/kg/min. Baseline characteristics of the overall cohort and stratified by the presence or absence of MASLD, as determined by MRS liver fat quantification. Baseline characteristics are expressed as mean ± SD or median (IQR) according to distribution or frequency (%). Differences between two groups were analysed using the unpaired t test or Mann-Whitney U test, according to distribution. Categorical data were compared using the Chi-square or Fisher's exact tests. Statistical significance was set at p <0.05 for all analyses. Δ13CO2, % of 13CO2 compared with baseline excreted after 90 min; ALT, alanine aminotransferase; AST, aspartate aminotransferase; CAP™, controlled attenuation parameter; DBP, diastolic blood pressure; eGFR, estimated glomerular filtration rate; GGT, gamma-glutamyl transferase; HbA1c, haemoglobin A1c; HDL-c, HDL cholesterol; LDL-c, LDL cholesterol; LFC, liver fat content; LSM, liver stiffness measurement; M-value, glucose disposal rate; MASLD, metabolic dysfunction-associated steatotic liver disease; MRS, magnetic resonance spectroscopy; PRS-HFC, polygenic risk score for hepatic fat content; SBP, systolic blood pressure; TDI, total daily dose of insulin; TG, triglycerides; WC, waist circumference. 13C-glucose breath test results available in 28 individuals: 15 in the MASLD group and 13 in the no-MASLD group. Based on M-value <4.9 mg/kg/min. The distribution of MASLD-associated SNP variants in PNPLA3, GCKR, and MBOAT7 was similar between individuals with and without MASLD (Table 1). The TM6SF2 risk allele tended to be more prevalent in the MASLD group (36.8% vs 10.0%); however, this did not reach statistical significance (p = 0.065). Correlation analyses All correlation analyses are summarised in Table 2. We observed a strong inverse correlation between the HEC-derived M-value and LFC, and a moderate correlation between M-value and CAP™. The M-value correlated strongly with Δ13CO2, and less strongly with the eGDR, which is in accordance with our previous work.8Table 2Correlation analysis matrix.Table 2LFCCAP™LSMM-value13C-GBTeGDRLFC1–––––CAP™0.81∗∗∗1––––LSM0.38∗0.41∗1–––M-value-0.72∗∗∗-0.62∗∗∗-0.291––13C-GBT-0.63∗∗∗-0.71∗∗∗-0.51∗0.73∗∗∗1–eGDR-0.54∗∗∗-0.46∗∗∗-0.58∗∗∗0.50∗∗0.62∗∗∗1All p values are indicative of two-tailed analysis. Correlation was calculated using Pearson’s r or Spearman rs depending on normality of distribution.13C-GBT, 13C-glucose breath test (Δ13CO2, δ‰); CAP™, controlled attenuation parameter (dB/m); eGDR, estimated glucose disposal rate (mg/kg/min); LFC, liver fat content (%); LSM, liver stiffness measurement (kPa); M-value, glucose disposal rate (mg/kg/min).∗Correlation significant at the 0.05 level.∗∗Correlation significant at the 0.01 level.∗∗∗Correlation significant at the <0.001 level. Table 2Correlation analysis matrix.Table 2LFCCAP™LSMM-value13C-GBTeGDRLFC1–––––CAP™0.81∗∗∗1––––LSM0.38∗0.41∗1–––M-value-0.72∗∗∗-0.62∗∗∗-0.291––13C-GBT-0.63∗∗∗-0.71∗∗∗-0.51∗0.73∗∗∗1–eGDR-0.54∗∗∗-0.46∗∗∗-0.58∗∗∗0.50∗∗0.62∗∗∗1All p values are indicative of two-tailed analysis. Correlation was calculated using Pearson’s r or Spearman rs depending on normality of distribution.13C-GBT, 13C-glucose breath test (Δ13CO2, δ‰); CAP™, controlled attenuation parameter (dB/m); eGDR, estimated glucose disposal rate (mg/kg/min); LFC, liver fat content (%); LSM, liver stiffness measurement (kPa); M-value, glucose disposal rate (mg/kg/min).∗Correlation significant at the 0.05 level.∗∗Correlation significant at the 0.01 level.∗∗∗Correlation significant at the <0.001 level. Correlation analysis matrix. All p values are indicative of two-tailed analysis. Correlation was calculated using Pearson’s r or Spearman rs depending on normality of distribution. 13C-GBT, 13C-glucose breath test (Δ13CO2, δ‰); CAP™, controlled attenuation parameter (dB/m); eGDR, estimated glucose disposal rate (mg/kg/min); LFC, liver fat content (%); LSM, liver stiffness measurement (kPa); M-value, glucose disposal rate (mg/kg/min). Correlation significant at the 0.05 level. Correlation significant at the 0.01 level. Correlation significant at the <0.001 level. The 13C-GBT demonstrated a moderate correlation with LFC, and a strong correlation with CAP™. The eGDR demonstrated a moderate correlation with LFC, and only a mild correlation with CAP™. We also examined whether liver stiffness assessed by VCTE correlated with insulin resistance. The M-value did not show any significant correlation with LSM, but 13C-GBT/eGDR demonstrated moderate correlation with LSM. Linear regression As shown in Fig. 2A, of all the 23 participants on cohort level with significant insulin resistance, 17 had liver steatosis (73.9%), whereas 87.5% participants without significant insulin resistance were free of significant liver steatosis. Furthermore, we observed that the distribution of the M-value plotted against LFC showed a logarithmically distributed relationship (Fig. 2B,C). As the values of LFC were right-skewed, not meeting the assumptions of linear regression, we performed a logarithmic transformation of LFC to LnLFC.Fig. 2Scatterplots of M-value vs liver fat content (LFC) and natural logarithm of LFC.(A) Scatterplot of M-value (mg/kg/min) vs LFC (%) with coloured indications of presence or absence of insulin resistance (IR) and metabolic dysfunction-associated steatotic liver disease (MASLD), respectively. A J-shaped association was observed between HEC values and the percentages of liver fat. Among individuals with an M-value exceeding the threshold of 4.9 mg/kg/min, only two of 15 (13.3%) individuals displayed MASLD on MRS, both exhibiting only mild steatosis. Conversely, six of 24 (25%) individuals with HEC-confirmed IR did not exhibit liver steatosis on MRS. (B) Exponential non-linear association between M-value and LFC. Curve: LFC = exp (-0.22 ∗ M-value [mg/kg/min] + 2.69). (C) Linear association between M-value and the natural logarithm (Ln) of LFC. Scatterplots of M-value vs liver fat content (LFC) and natural logarithm of LFC. (A) Scatterplot of M-value (mg/kg/min) vs LFC (%) with coloured indications of presence or absence of insulin resistance (IR) and metabolic dysfunction-associated steatotic liver disease (MASLD), respectively. A J-shaped association was observed between HEC values and the percentages of liver fat. Among individuals with an M-value exceeding the threshold of 4.9 mg/kg/min, only two of 15 (13.3%) individuals displayed MASLD on MRS, both exhibiting only mild steatosis. Conversely, six of 24 (25%) individuals with HEC-confirmed IR did not exhibit liver steatosis on MRS. (B) Exponential non-linear association between M-value and LFC. Curve: LFC = exp (-0.22 ∗ M-value [mg/kg/min] + 2.69). (C) Linear association between M-value and the natural logarithm (Ln) of LFC. Univariate linear regression with M-value as an independent variable and LnLFC as a dependent variable yielded an R2 of 0.47 (B −0.223, 95% CI of B −0.302 to −0.144, p <0.001). Adding triglycerides to the logarithmic model resulted in a significant increase in R2 (R2 0.53, R2 change 0.06, p for significant change = 0.039). R2 further increased to 0.57 by adding PRS-HFC (R2 change 0.04, p for significant change = 0.072). There was no significant increase in R2 of the model by adding age, sex, diabetes duration, HbA1c, HDL, or a sex-by-M-value interaction. Owing to collinearing, BMI or WC were not included simultaneously. In Fig. 3, substituting the M-value with Δ13CO2 yielded similar results (R2 0.42, B −0.261, 95% CI of B −0.385 to −0.138, p <0.001). Adding triglycerides also increased R2 (R2 0.54, R2 change 0.11, p for significant change = 0.007). Adding PRS-HFC did not yield any further significant change to R2. However, when Δ13CO2 was replaced by eGDR, the R2 value decreased (R2 0.21, B −0.131, 95% CI of B −0.215 to −0.047, p = 0.003). Adding triglycerides significantly added to R2 (R2 0.46, R2 change 0.25, p for significant change <0.001). Adding PRS-HFC did not change R2 significantly.Fig. 3Scatterplot panel of Δ13CO2 and eGDR vs liver fat content and CAP™.All scatterplots display indices of insulin resistance on the horizontal axis, Δ13CO2 (δ‰) in the left panels and the estimated glucose disposal rate (eGDR, mg/kg/min) in the right panels, and indices of liver steatosis on the vertical axis, namely the natural logarithm of the liver fat content (Ln(LFC), %) in the top panels and controlled attenuation parameter (CAP™, dB/m) on the bottom panels. Vertical dotted lines indicate ideal cut-off points to identify liver steatosis caused by metabolic-dysfunction-associated steatotic liver disease based on analysis of the AUC curve (see Table 3). Horizontal dotted lines in bottom panels indicate international accepted cut-off point for liver steatosis based on CAP™ (>275 dB/m).17 R2 indicates percentages of variance explained by linear regression analysis. Scatterplot panel of Δ13CO2 and eGDR vs liver fat content and CAP™. All scatterplots display indices of insulin resistance on the horizontal axis, Δ13CO2 (δ‰) in the left panels and the estimated glucose disposal rate (eGDR, mg/kg/min) in the right panels, and indices of liver steatosis on the vertical axis, namely the natural logarithm of the liver fat content (Ln(LFC), %) in the top panels and controlled attenuation parameter (CAP™, dB/m) on the bottom panels. Vertical dotted lines indicate ideal cut-off points to identify liver steatosis caused by metabolic-dysfunction-associated steatotic liver disease based on analysis of the AUC curve (see Table 3). Horizontal dotted lines in bottom panels indicate international accepted cut-off point for liver steatosis based on CAP™ (>275 dB/m).17 R2 indicates percentages of variance explained by linear regression analysis. The M-value showed a moderate linear association with CAP™ (R2 0.38, B −17.312, 95% CI of B −24.957 to −9.66, p <0.001). When comparing Δ13CO2 and eGDR with CAP™, Δ13CO2 exhibited the strongest association (R2 0.51, B −24.092, 95% CI for B −34.147 to −14.037, p <0.001), whereas eGDR showed a weaker association (R2 0.21, B −11.283, 95% CI of B −18.873 to −3.693, p = 0.005). Logistic regression In univariable logistic regression, lower M-value was associated with MASLD (OR 2.42, 95% CI 1.39–4.22, p = 0.002, per decreasing unit of M-value). This association remained significant after adjustment for age, sex, diabetes duration, and BMI (OR 3.12, 95% CI 1.17–8.37, p = 0.024). Swapping BMI for WC yielded similar results (OR 3.24, 95% CI 1.15–9.13, p = 0.026). The association also remained robust in sensitivity analyses additionally adjusting for triglycerides (OR M-value 2.76, 95% CI 1.08–7.04, p = 0.034), TDI/kg (OR M-value 3.14; 95% CI 1.38–7.18, p = 0.007), and PRS-HFC (OR 3.41, 95% CI 1.47–7.90, p = 0.004). PRS-HFC itself was also associated with MASLD. Substituting PRS-HFC for PRS-HFC2 yielded similar results. The PRS-HFC × M-value interaction was not significant. When M-value was replaced by alternative markers of insulin resistance, both Δ13CO2 (OR 2.56, 95% CI 1.20–5.56, p = 0.014) and eGDR (OR 1.41, 95% CI 1.09–1.81, p = 0.008) remained associated with MASLD in adjusted models. Using CAP™-defined steatosis as outcome yielded comparable results for M-value, Δ13CO2, and eGDR. All regression analyses are shown in Table 3.Table 3Logistic regression models for the association between insulin resistance markers and hepatic steatosis.Table 3MarkerAdjustmentsOR95% CIp valueMASLD as outcomeM-valueNone2.42139–4.220.002M-valueAge, sex, diabetes duration, BMI3.121.17–8.370.024M-valueAge, sex, diabetes duration, WC3.241.15–9.130.026M-valueAge, sex, diabetes duration, BMI, triglycerides2.761.08–7.040.034M-valueAge, sex, diabetes duration, BMI, TDI/kg3.141.38–7.180.007M-valueAge, sex, diabetes duration, BMI, PRS-HFC3.411.47–7.900.004M-valueAge, sex, diabetes duration, BMI, PRS-HFC23.621.39–9.390.00813C-GBTAge, sex, diabetes duration, BMI2.561.20–5.560.014eGDRAge, sex, diabetes duration, BMI1.411.09–1.810.008PRS-HFCAge, sex, diabetes duration, BMI, M-value1.061.00–1.130.043PRS-HFC2Age, sex, diabetes duration, BMI, M-value1.091.00–1.170.041CAP defined steatosis as outcomeM-valueAge, sex, diabetes duration, BMI2.861.37–5.880.00513C-GBTAge, sex, diabetes duration, BMI1.751.12–2.780.015eGDRAge, sex, diabetes duration, BMI1.791.12–2.860.015Logistic regression was used to identify factors independently associated with MASLD (LFC >5.56%) or steatosis based on CAP™ (>275 dB) and to calculate ORs with 95% CIs. Statistical significance was set at p <0.05 for all analyses. 13C-GBT, 13C-glucose breath test (Δ13CO2, δ‰); CAP™, controlled attenuation parameter (dB/m); eGDR, estimated glucose disposal rate (mg/kg/min); M-value, glucose disposal rate (mg/kg/min); MASLD, metabolic dysfunction-associated steatotic liver disease; OR, odds ratio; PRS-HFC, polygenic risk score for hepatic fat content, composed of four single nuclear polymorphisms; PRS2, polygenic risk score composed of two single nuclear polymorphisms; TDI, total daily dose of insulin. Table 3Logistic regression models for the association between insulin resistance markers and hepatic steatosis.Table 3MarkerAdjustmentsOR95% CIp valueMASLD as outcomeM-valueNone2.42139–4.220.002M-valueAge, sex, diabetes duration, BMI3.121.17–8.370.024M-valueAge, sex, diabetes duration, WC3.241.15–9.130.026M-valueAge, sex, diabetes duration, BMI, triglycerides2.761.08–7.040.034M-valueAge, sex, diabetes duration, BMI, TDI/kg3.141.38–7.180.007M-valueAge, sex, diabetes duration, BMI, PRS-HFC3.411.47–7.900.004M-valueAge, sex, diabetes duration, BMI, PRS-HFC23.621.39–9.390.00813C-GBTAge, sex, diabetes duration, BMI2.561.20–5.560.014eGDRAge, sex, diabetes duration, BMI1.411.09–1.810.008PRS-HFCAge, sex, diabetes duration, BMI, M-value1.061.00–1.130.043PRS-HFC2Age, sex, diabetes duration, BMI, M-value1.091.00–1.170.041CAP defined steatosis as outcomeM-valueAge, sex, diabetes duration, BMI2.861.37–5.880.00513C-GBTAge, sex, diabetes duration, BMI1.751.12–2.780.015eGDRAge, sex, diabetes duration, BMI1.791.12–2.860.015Logistic regression was used to identify factors independently associated with MASLD (LFC >5.56%) or steatosis based on CAP™ (>275 dB) and to calculate ORs with 95% CIs. Statistical significance was set at p <0.05 for all analyses. 13C-GBT, 13C-glucose breath test (Δ13CO2, δ‰); CAP™, controlled attenuation parameter (dB/m); eGDR, estimated glucose disposal rate (mg/kg/min); M-value, glucose disposal rate (mg/kg/min); MASLD, metabolic dysfunction-associated steatotic liver disease; OR, odds ratio; PRS-HFC, polygenic risk score for hepatic fat content, composed of four single nuclear polymorphisms; PRS2, polygenic risk score composed of two single nuclear polymorphisms; TDI, total daily dose of insulin. Logistic regression models for the association between insulin resistance markers and hepatic steatosis. Logistic regression was used to identify factors independently associated with MASLD (LFC >5.56%) or steatosis based on CAP™ (>275 dB) and to calculate ORs with 95% CIs. Statistical significance was set at p <0.05 for all analyses. 13C-GBT, 13C-glucose breath test (Δ13CO2, δ‰); CAP™, controlled attenuation parameter (dB/m); eGDR, estimated glucose disposal rate (mg/kg/min); M-value, glucose disposal rate (mg/kg/min); MASLD, metabolic dysfunction-associated steatotic liver disease; OR, odds ratio; PRS-HFC, polygenic risk score for hepatic fat content, composed of four single nuclear polymorphisms; PRS2, polygenic risk score composed of two single nuclear polymorphisms; TDI, total daily dose of insulin. Discussion In this study, we examined that insulin resistance correlated with liver steatosis in individuals with T1D. We observed that the glucose disposal rate (M-value), i.e. the gold standard index of insulin sensitivity derived from HEC, strongly and independently correlated with the liver lipid content as assessed by MRS. Moreover, the M-value was significantly lower in individuals with MASLD than in those without MASLD. Nearly all participants with hepatic steatosis exhibited marked insulin resistance, defined by an M-value <4.9 mg/kg/min. The severity of insulin resistance correlated positively with LFC, indicating that higher degrees of insulin resistance are associated with more advanced hepatic steatosis. Performing HEC is time consuming, invasive, less desired in clinical practice, and requires highly trained personnel. We demonstrated that less cumbersome indices of insulin resistance designed specifically for people with T1D, i.e. 13C-GBT and eGDR, were also significantly associated with LFC and identified MASLD with comparable accuracy. This association held true regardless of whether MRS or CAP™ was used to determine steatosis, which reinforces our findings and offers alternatives for routine clinical practice. The latter argument is fortified even further by the complete overlap between magnetic resonance- and CAP™-based diagnosis of MASLD in our study population. Evidently, individuals with MASLD displayed greater WC, higher BMI, higher systolic blood pressure, elevated triglycerides, and decreased HDL levels. All these findings correlate with the clinical phenotype of significant insulin resistance. Importantly, the correlation between insulin resistance and LFC remained significant, even when adjusted for BMI. This independent correlation between insulin resistance and LFC aligns our findings with previous epidemiological reports in populations without T1D and might further substantiate the potential role of insulin resistance in liver steatosis development in T1D.3,4,7 Interestingly, the combination of any index of insulin resistance and triglycerides increased the R2 value in linear regression, indicating that these combined parameters increased the explained variability in LFC. Given that triglycerides reflect hepatic lipid metabolism, the increases in R2 after adding triglycerides suggest that circulating triglyceride levels provide additional information on hepatic lipid metabolism beyond insulin resistance alone. Conversely, triglycerides and insulin resistance are metabolically connected; compared with four other criteria of the metabolic syndrome, triglycerides are more subject to day-to-day variability. This may explain why R2 increased proportionally more when using eGDR as a proxy for insulin resistance, as eGDR is composed of more static components. In a previous study, Llaurado et al.21 determined HEC values and LFC in 32 overweight adults with T1D and controls without diabetes who were matched for age, sex, and BMI. They used HEC in combination with infusion of D-[3-3H] glucose to assess whole-body insulin resistance vs hepatic insulin resistance. The authors concluded that LFC is lower in individuals with T1D. In this group, bodyweight and triglycerides best predicted LFC, whereas peripheral lipolysis determined LFC in their matched control group. They did not correlate LFC with insulin resistance in this study, as it was not their study objective. Furthermore, the average BMI of the individuals in that study was 28.9 ± 0.7 kg/m2; however, because the authors failed to specify the number of obese participants, MASLD-positive individuals were underrepresented. Indeed, our study group previously determined that obesity is a major risk factor to have MASLD in T1D.3,4,22 The current study adds to previous prevalence studies that obesity, although confirmed as a risk factor for MASLD, is a clinical proxy for the underlying insulin resistance.3,5 Notably, peripheral lipolysis plays a crucial role in triglyceride flux towards the liver, and impaired suppression of insulin-induced lipolysis can lead to liver lipid accumulation. Interestingly, Llaurado et al.21 found greater insulin-mediated suppression of EGP and lipolysis in individuals with T1D compared with controls; however, these individuals are underrepresented, as mentioned earlier. In another study, the authors examined intrahepatic lipid content in individuals with T1D compared with controls without diabetes but did not establish a significant association with insulin sensitivity.23 The studies by Llaurado et al.21 and Perseghin et al.23 had some limitations: (1) they evaluated only 32 and 19 participants, respectively; (2) they included only overweight or lean individuals; (3) they compared their findings with healthy controls without diabetes instead of assessing individuals with varying degrees of insulin resistance within the T1D population; and (4) they omitted participants with increased LFC, thus limiting the possibility of linear correlation. Our study had a more comprehensive approach, including patients with a wider range of M-value (0.88–10.12 mg/kg/min, LFC (1.9–27.6%), and BMI (17.5–44.6 kg/m2). Consequently, insulin resistance demonstrated a significant and strong correlation with LFC in loglinear and logistic regression with robust statistical power. This correlation persisted even when substituting the gold standard methods with their non-invasive equivalents, further strengthening our findings. Based on literature, individuals with T1D experience impaired insulin sensitivity compared with controls without diabetes.24 The combination of autoimmune insulin deficiency and clinically significant insulin resistance is often referred to as “double diabetes.”6,25,26 Increasing evidence suggests that the presence of double diabetes elevates the risk of long-term complications, including microvascular and macrovascular complications.[27], [28], [29] Insulin resistance is strongly associated with MASLD and serves as an independent cardiovascular risk factor; furthermore, MASLD is also increasingly suspected to be an independent risk factor and insulin resistance. Therefore, our data underscore the relevance of screening for MASLD in individuals with T1D.4,[30], [31], [32], [33] The question remains whether MASLD and insulin resistance are independently associated with an enhanced risk profile, or if they are collinear in their effect on cardiovascular disease. Because MASLD is a silent disease associated with an enhanced risk for adverse hepatic or cardiovascular outcomes in the general population, and potentially in those with T1D, physicians should screen individuals with T1D for MASLD whenever marked insulin resistance is suspected. The mechanisms underlying the relatively low prevalence of MASLD in individuals with T1D remain unclear. Although obesity is a major risk factor, not all individuals with T1D and obesity develop MASLD, suggesting disease-specific protective factors.[3], [4], [5] One potential mechanism involves altered portal insulin dynamics. In individuals with T1D, the absence of endogenous insulin production considerably alters insulin concentration in the portal system compared with those without T1D. Normally, upon secretion by pancreatic beta-cells, insulin undergoes a first-pass effect in the liver by traveling through the portal vein. However, individuals with T1D administer insulin peripherally, bypassing this first-pass effect and modifying the peripheral-portal insulin gradient.21,34 Fat partitioning, which is defined as the tissue distribution of fat storage, reflects the balance between lipogenesis and fat oxidation. In T1D, this is likely affected by the abovementioned altered peripheral-portal insulin distribution. This results in reduced inhibition of hepatic lipolysis. Systemic insulin resistance, especially in the general population intertwined with hyperinsulinaemia, further promotes steatotic liver disease through several additional pathways. Insulin resistance in adipose tissue increases lipolysis, leading to enhanced release of free fatty acids into the circulation, which are subsequently taken up and stored by the liver.35 In addition, altered ectopic fat partitioning and impaired storage capacity of subcutaneous adipose tissue may favour lipid deposition in the liver. Hepatic insulin resistance can also impair the suppression of gluconeogenesis while simultaneously maintaining lipogenic signalling pathways, thereby promoting hepatic lipid synthesis.35,36 In hepatocytes, insulin stimulates the PI3K signalling pathway, leading to activation of sterol regulatory element-binding protein-1c (SREBP-1c), a central regulator of genes involved in fatty acid and triglyceride synthesis. SREBP-1c, together with carbohydrate response element-binding protein (ChREBP), enhances the expression of key lipogenic enzymes such as acetyl-CoA carboxylase and fatty acid synthase, thereby promoting de novo lipogenesis.7,37,38 Hyperglycaemia can further stimulate this pathway independently of insulin, suggesting that not only hyperinsulinaemia (exogenously administered) but also hyperglycaemia contributes to hepatic lipid synthesis in T1D.39 A recent meta-analysis demonstrated that individuals with MASLD and T1D exhibit both a higher HbA1c and an increased TDI; this elevated TDI indicates whole-body insulin resistance, possibly aligning with the abovementioned pathways.5 However, these hypotheses require further dedicated studies in this particular population. In addition, we explored the potential contribution of genetic susceptibility to hepatic steatosis using established MASLD-associated SNPs and a PRS. Although the distribution of individual variants did not significantly differ between groups, we observed a trend towards a higher prevalence of the TM6SF2 risk allele in individuals with MASLD. Moreover, the PRS-LFC was independently associated with MASLD, supporting a role for genetic predisposition in hepatic fat accumulation in T1D. Importantly, no interaction was observed between genetic risk and insulin resistance, suggesting that the relationship between insulin resistance and hepatic lipid accumulation is largely independent of genetic susceptibility in this cohort. These findings indicate that genetic factors may contribute to interindividual variability in LFC; however, given the limited sample size, these results should be interpreted cautiously and warrant confirmation in larger cohorts. From a clinical perspective, assessment of insulin resistance may help identify individuals within the T1D population who could benefit from targeted case finding for MASLD. Current European guidelines do not recommend routine MASLD screening in individuals with T1D, largely owing to conflicting data regarding disease severity and the limited diagnostic accuracy of commonly used screening tools.2 Moreover, both the overall prevalence of MASLD and the natural history of the disease appear to be less pronounced in individuals with T1D than in those with type 2 diabetes, but follow-up studies describing the natural history of MASLD are currently lacking.40,41 However, individuals with T1D who exhibit marked insulin resistance may represent a metabolically vulnerable subgroup with an increased risk of hepatic steatosis and a higher cardiovascular risk profile.4,5,42 Insulin resistance assessment in this context may, therefore, have important clinical implications, as it could prompt more thorough measures in several treatment areas: (1) in weight management, where incretin mimetics are reportedly effective in T1D;22,[43], [44], [45] (2) intensification of physical activity can help reduce insulin resistance; and (3) the traditional targets for reduction of cardiovascular risk (smoking cessation, lipid and blood pressure control) should not be overlooked. Our study has several strengths, but also some limitations. Firstly, we studied a sizable sample of 39 adults with diverse insulin sensitivity levels, ensuring comprehensive evaluation of metabolic control. Inclusion of participants with and without liver steatosis distinguishes our study from studies solely comparing with control groups lacking individuals with steatosis. Although the study was not powered a priori, the observed difference in M-value between MASLD and control participants was large (Cohen’s d = 1.7). For this effect size, the available sample (20 per group) provides >99% statistical power, indicating that the group difference in insulin resistance was robustly detectable despite the pilot design. Secondly, we used gold standard methods (MRS for LFC and HEC for insulin sensitivity) to minimise measurement errors. Practical alternatives, CAP™ and 13C-GBT/eGDR, were also considered and validated owing to cost considerations and technical challenges in clinical practice. Tracer-enhanced HECs were not performed, preventing further correlation analyses of tissue-specific insulin resistance with LFC. Importantly, the balanced 1:1 ratio of individuals with and without MASLD does not reflect natural prevalence which could inflate effect estimates compared with unselected populations with T1D. Consequently, we did not calculate the positive and negative predictive values because of their dependency on natural prevalence. Therefore, further longitudinal studies of the predictive accuracy of indices of insulin resistance to diagnose MASLD or predict MASLD progression are required to validate our findings in cohorts with natural prevalence rates. Thirdly, our study encompasses exploratory data on SNPs associated with MASLD, revealing that the association between insulin resistance and MASLD remained significant after correcting for PRS-HFC. A fourth remark is the relatively low insulin infusion rate during HEC, which was chosen to minimise post-HEC hypoglycaemia risk. A recent review suggested that people with severe insulin resistance, especially hepatic insulin resistance, require higher insulin infusion rates to completely suppress EGP.46 However, to ensure complete suppression of EGP, HEC duration was prolonged in cases of initial hyperglycaemia, primarily in individuals with severe insulin resistance.47 Lastly, it is important to acknowledge that we were unable to establish causality because of the cross-sectional nature of the study. Further longitudinal studies are needed to confirm the hypothesis that insulin resistance is a pathophysiological driver of MASLD in T1D, and to examine whether improving insulin sensitivity reduces LFC in T1D. Conclusions Insulin resistance and liver steatosis are associated in individuals with T1D, independent of BMI, age, diabetes duration, and sex. These findings hold significant clinical implications, as individuals with the double diabetes phenotype are at an increased risk of developing MASLD. Abbreviations 13C-GBT, 13C-glucose breath test; ALT, alanine aminotransferase; AST, aspartate aminotransferase; CAP™, controlled attenuation parameter; ChREBP, carbohydrate response element-binding protein; CKD-EPI, Chronic Kidney Disease Epidemiology Collaboration; eGDR, estimated glucose disposal rate; eGFR, estimated glomerular filtration rate; EGP, endogenous glucose production; GGT, gamma-glutamyl transferase; HbA1c, haemoglobin A1c; HDL-c, high-density lipoprotein cholesterol; HEC, hyperinsulinaemic-euglycaemic clamp; LDL-c, low-density lipoprotein cholesterol; LFC, liver fat content; LSM, liver stiffness measurement; MASLD, metabolic dysfunction-associated steatotic liver disease; MASH, metabolic dysfunction-associated steatohepatitis; MRS, magnetic resonance spectroscopy; PRS-HFC, polygenic risk score for hepatic fat content; SNP, single-nucleotide polymorphism; SREBP-1c, sterol regulatory element-binding protein-1c; T1D, type 1 diabetes; TDI, total daily dose of insulin; VCTE, vibration-controlled transient elastography; WC, waist circumference. Authors' contributions JM, SF, and CDB conceptualised the study. JM interpreted the analyses, visualised the results, searched the literature, and wrote the manuscript. JM and MS coordinated and supervised the data collection. JM, SF, and CDB acquired funding for the analysis. JM designed the statistical analyses. JM, MS, and FV performed and supervised all magnetic imaging. All the authors added to the study plan, interpreted the data, and critically revised the scientific content of the manuscript. All the authors had access to all the data of the study, approved the final manuscript as submitted, and agreed to be accountable for all aspects of the work. SF and CDB are the guarantors of this work and, as such, had full access to all the data in the study, verified the data, and take responsibility for the integrity of the data and the accuracy of the data analysis. JM, SF, and CDB had final responsibility for the decision to submit for publication. Data availability Owing to patient privacy and the extent of patient consent, no individual patient data can be shared with outside investigators. However, collaborative research proposals are welcome. Aggregated data are available upon reasonable request via e-mail to JM (jonathan.mertens@uza.be). Funding JM received a doctoral grant from the 10.13039/501100007660University of Antwerp (BOF, Antigoon ID: 44430) and a research grant from the Belgian Association for the study of the Liver (BASL). SF holds a senior clinical investigator fellowship from the Research Foundation Flanders (FWO) (1802154N). Conflicts of interest ED has served as a consultant for Novo Nordisk, Ely Lilly, and Boehringer Ingelheim. CDB reports consulting fees and honoraria for speaking for Abbott, AstraZeneca, Boehringer Ingelheim, A. Menarini Diagnostics, Eli Lilly, Medtronic, Novo Nordisk, and Roche, and research support from AstraZeneca, Boehringer Ingelheim, Indigo Diabetes, and Novo Nordisk. SF has received grants from Astellas, Falk Pharma, Genfit, Gilead Sciences, GlympsBio, Janssens Pharmaceutica, Inventiva, Merck Sharp & Dome, Pfizer, and Roche. He has acted as a consultant for Abbvie, Actelion, Aelin Therapeutics, Aligos Therapeutics, Allergan, Astellas, Astra Zeneca, Bayer, Boehringer Ingelheim, Bristol-Meyers Squibb, CSL Behring, Coherus, Echosens, Eisai, Enyo, Galapagos, Galmed, Genetech, Genfit, Gilead Sciences, Intercept, Inventiva, Janssens Pharmaceutica, Julius Clinical, Madrigal, Medimmune, Merck Sharp & Dome, NGM Bio, Novartis, Novo Nordisk, Promethera, and Roche. He has been a lecturer for Abbvie, Allergan, Bayer, Eisai, Genfit, Gilead Sciences, Janssens Cilag, Intercept, Inventiva, Merck Sharp & Dome, Novo Nordisk, and Promethera. All other authors declare no conflicts of interest. Please refer to the accompanying ICMJE disclosure forms for further details.