Ethics

The present study was approved by the Central Regional Committee on Health Research Ethics, Denmark (1–10-72–251−20) and was registered at the Danish Data Protection Agency and ClinicalTrials.gov (NCT04656236). Written informed consent was obtained from all participants before participation in the study.

Participants

Ten participants diagnosed with type 1 diabetes and ten age-, BMI- and weight-matched control participants from the Central Denmark Region were included in the study (Fig. 1). Participants with type 1 diabetes were recruited between March 2021 and February 2022 from the outpatient clinic at Steno Diabetes Center, Aarhus, and through social media; healthy control participants were recruited only through social media. Inclusion criteria were male sex, age >18 years, and BMI between 19 and 26 kg/m2. Individuals with type 1 diabetes were eligible for inclusion if they had a diagnosis of type 1 diabetes and were C-peptide negative (<0.2 nmol/l), while exclusion criteria were severe comorbidity, daily use of medication other than insulin or use of very long-acting insulin analogues (duration of action >24 h). Control participants were healthy male individuals without chronic or acute illness or any regular use of medication. A normal screening examination and blood tests (sodium, potassium, albumin, creatinine, C-reactive protein, leucocytes, alanine aminotransferase, bilirubin, thyroid stimulating hormone, haemoglobin, thrombocytes, international normalised ratio) were required for all participants before they were enrolled in the study. One potential control participant was assessed as unsuitable for participation due to very limited abdominal subcutaneous adipose tissue, making adipose tissue biopsy collection unfeasible. HOMA-IR was calculated for participants as fasting insulin (pmol/l) × fasting glucose (mmol/l)/135. Sex and ethnicity were self-reported. Because of the small sample size of ten participants per group, we chose to include only men, to minimise outcome variation from potential sex differences in metabolism.

Fig. 1

CONSORT flow diagram. T1DM, participants with type 1 diabetes; CTR, healthy control participants

Study design

The study was designed as a randomised, single-blind, crossover trial. All participants were examined on one study day while receiving 3 h i.v. infusions of (i) sodium-d/l-3-OHB (3-OHB, 75 g/l, Sigma-Aldrich, MO, USA; a racemic mixture of d- and l-3-OHB) and (ii) saline (NaCl, 34 g/l; Capital Region Pharmacy, Herlev, Denmark) in random order and separated by a 1 h washout period. The two interventions were matched for sodium content. The 3-OHB infusion was administered as a priming bolus (30 g/h for 15 min), followed by a continuous infusion (10.5 g/h). NaCl was administered in equivalent volumes. A 1 h washout period was considered sufficient to avoid significant carryover, based on the results of previous studies examining the effects of intravenous 3-OHB administration [11].

Five participants were randomised to receive 3-OHB followed by NaCl and five to receive NaCl followed by 3-OHB within both groups. The resource at www.randomizer.org was used for the randomisation process. All participants were blinded to the intervention order. The enrolment of participants, the generation of intervention sequence allocation and the intervention assignment for each participant were conducted by the primary investigator/first author.

Study day

Examinations were conducted at the Medical/Steno Aarhus Research Laboratory (Aarhus University Hospital, Aarhus, Denmark). All participants were studied after an overnight fast (10 h) and were instructed to avoid strenuous physical activity, alcohol and coffee intake, and to follow the national dietary recommendations (approximately 30% fat, 20% protein and 50% carbohydrate) for 48 h before the study day.

Participants arrived at the research facilities at 07:00 hours and were placed in a bed shortly after. Venous catheters were inserted in a cubital vein for intravenous infusions and in a dorsal hand vein heated by a heating pad for arterialised blood sampling [12]. At approximately 08:00 hours, the first intervention period (3-OHB/NaCl) was initiated; the second intervention period (3-OHB/NaCl) began at approximately 12:00 hours. The timing of examinations, blood sampling and adipose tissue biopsy collection were similar in the two interventional periods, and are schematically shown in the study flow chart (Fig. 2).

Fig. 2

Study flow diagram. Ten men with type 1 diabetes and ten healthy men were examined during 3 h of i.v. infusion with 3-OHB and saline (NaCl) in a randomised crossover design. The timing of blood sampling, adipose tissue biopsy collection and tracer infusions is shown. Participants with type 1 diabetes (T1DM) received a stable insulin infusion throughout the study day. Created in BioRender. Møller, N. (2025) https://BioRender.comd/z88k119

Insulin treatment of participants with type 1 diabetes

Long-acting insulin analogue therapy was replaced with subcutaneous rapid-acting insulin 24 h before the study day. Participants with type 1 diabetes were hospitalised at 22:00 hours the evening before the study day to substitute their subcutaneous insulin administration with a continuous intravenous human insulin infusion (Actrapid, Novo Nordisk). Insulin infusion rates were titrated overnight to maintain a blood glucose level of approximately 7–10 mmol/l, as previously described [13]. A glucose sensor was used to measure blood glucose levels once every hour (Freestyle Libre, Abbott, Abbott Park, IL, USA). On the morning of the study day, participants with type 1 diabetes transferred themselves to the research facilities. A continuous insulin infusion rate for the study day was chosen based on the individual’s titrated overnight insulin infusion rate (Fig. 3). The overnight glucose concentrations and insulin infusion rates are provided in the electronic supplementary material (ESM Fig. 1).

Fig. 3

Insulin treatment for participants with type 1 diabetes. Created in BioRender. Møller, N. (2025) https://BioRender.com/i74j440

Metabolic fluxes measured using radioactive isotopes

To quantify the lipolytic rate (palmitate flux), a [9,10-3H]palmitate tracer was infused (0.666 MBq/h) at time = 120–180 min in each intervention period. Glucose kinetics were estimated by a primed infusion of [3-3H]glucose tracer (priming bolus 0.444 MBq, continuous infusion 0.2664 MBq/h) starting at time = 0 min in the first period and continuing throughout the study day. Blood samples were drawn in triplicate during the last 20 min of each intervention period at the time points 160, 170 and 180 min. Plasma palmitate concentration and specific activity were determined by HPLC, while plasma [3-3H]glucose specific activity was measured using a liquid scintillation counter. Whole-body palmitate flux (µmol/min) was calculated as palmitate tracer infusion rate (dpm/min) divided by the steady-state specific activity (dpm/μmol) for palmitate as previously described [14]. The glucose rate of appearance (endogenous glucose production, EGP) and rate of disappearance (Rd) were calculated using Steele’s non-steady-state equation [15].

Adipose tissue biopsies and western blotting

Abdominal subcutaneous adipose tissue biopsies were obtained 90 min after initiating each intervention period by needle aspiration under local anaesthesia (8 ml lidocaine ‘SAD’, 10 mg/ml; Amgros, Copenhagen, Denmark). The biopsies were immediately cleaned for blood contamination and snap-frozen in liquid nitrogen before being stored at −80°C. To extract protein, samples were homogenised at 4°C in PI3 buffer (50 mmol/l HEPES, 137 mmol/l NaCl, 10 mmol/l Na4P2O7, 20 mmol/l NaF, 5 mmol/l EDTA, 1 mmol/l MgCl2, 1 mmol/l CaCl2, 2 mmol/l Na3VO4, 5 mmol/l nicotinamide, 1% Halt protease inhibitor cocktail (100×, Thermo Fisher), 1% NP-40, 10% glycerol in demineralised water), and centrifuged for 20 min (16,060 g) to separate the protein-containing liquid (infranatant) from the lipid layer and pellet. The protein concentration in each sample was measured using a Pierce BCA protein assay (Thermo Fisher). Laemmli buffer (40 ml glycerol, 8.2 g SDS, 25 ml Tris base/HCl, 0.03 g bromophenol blue, 6.2 g dithiothreitol, 35 ml demineralised water) was added, and protein concentrations were standardised for all samples (1 µg/1 µl) by dilution with demineralised water.

Western blot analyses were performed using the Bio-Rad Criterion system (4–15% Criterion XTBis-Tris gels; Bio-Rad, Hercules, CA, USA) to measure the relative contents of specific protein targets. Gels were placed in running buffer (30.2 g Tris base, 114.2 g glycine, 10 g SDS in 1000 ml demineralised water) during protein loading and the western blot procedure. Membranes with transferred proteins were rinsed and stored in 1× TBS-T buffer (10× includes 2.2 g Tris base, 13 g Tris base/HCl, 87.7 g NaCl, 5 ml Tween-20 in 1000 ml water, adjusted to pH 7.4). The membranes were blocked in 1% BSA in TBS-T before adding the primary antibody. The following primary antibodies were used: hormone-sensitive lipase phosphorylated at Ser563, Ser565 or Ser660 (pHSL) (4139, 4137 and 4126; Cell Signaling Technology, MA, USA), HSL (PA5-17196; Cell Signaling Technology), phospho-PKA substrate (9624; Cell Signaling Technology), perilipin (PLIN1; 9349; Cell Signaling Technology), Akt phosphorylated at Ser473 or Thr308 (4060 and 9275; Cell Signaling Technology), pan Akt (ma5-14916; Thermo Fisher) and the phosphatase and tensin homologue (PTEN; 9188; Cell Signaling Technology). They were diluted in 1% BSA in TBS-T with 0.002% sodium azide. All antibodies were validated prior to use. For the secondary antibody, we used goat anti-rabbit IgG secondary antibody (31460; Thermo Fisher) diluted in 1% BSA in TBS-T. The same western blot membrane was exposed to both phospho-PKA substrate and PLIN1 antibody to identify the band representing PKA phosphorylation of PLIN1 as a measure of PKA activity. Target bands were visualised using the ChemiDoc MP imaging system (Bio-Rad), and quantified using Image Lab 5.0 (Bio-Rad). Total protein normalisation was performed to account for variability in protein loading and transfer using stain-free technology [16]. Western blot data are presented as median ratios of phosphorylated protein/total protein or phosphorylated/non-phosphorylated protein unless otherwise specified, and are plotted as ratio change relative to the median value for control participants during NaCl infusion.

Blood sample analysis

Blood samples were drawn at baseline and at intervals throughout the study day. Plasma glucose and lactate were immediately measured using immobilised enzyme biosensor technology (YSI 2300 model Stat Plus; Bie & Berntsen, Herlev, Denmark). Blood d-3-OHB was measured using Freestyle Precision ketone test strips (Abbott). All other samples were centrifuged and stored at −20°C for batch analyses after study completion. Commercially available kits were used for quantification of plasma/serum concentrations of insulin (Insulin ELISA; Mercodia, Uppsala, Sweden), C-peptide (C-peptide ELISA; Mercodia), NEFA (FujiFilm/Wako Chemicals Europe, Germany) and glucagon (Glucagon ELISA; Mercodia) in accordance with the manufacturers’ guidelines. Blood total d/l-3-OHB was measured by hydrophilic interaction LC tandem MS [17].

Endpoints

The primary endpoint of this study was the difference in palmitate flux between groups after 3 h of 3-OHB infusion (time = 180 min). Secondary endpoints were differences in circulating concentrations of NEFA, insulin, glucagon, glucose and lactate, glucose kinetics (time = 180 min), and the relative activity of enzymes involved in lipolytic signalling (time = 90 min).

Power calculation and statistics

Based on previous work from our laboratory examining the effect of 3-OHB on lipolytic rate in healthy men, a power calculation was performed using α = 0.05, β = 0.8 and an expected reduction in lipolytic rate for participants with type 1 diabetes, which was 30% less than for control participants with an assumed SD of σ = 0.17. This resulted in a necessary sample size of n=10 in each group.

All statistical analyses were performed using R software version 4.4.1 (R Foundation for Statistical Computing, Vienna, Austria). Repeated measures were analysed using a mixed model, with time, intervention, group, intervention sequence (order) and period of the day (first or second) as fixed effects, and participants and visit within participants as random effects, followed by pairwise comparisons of estimated means. Baseline data were compared using an unpaired Student’s t test and are presented as medians (ranges). Model validation was performed by inspecting QQ plots and histograms of residuals and predicted vs fitted residuals. In the absence of normal distribution, the mixed-model analysis was performed on log-transformed data.

The results were graphically visualised using GraphPad Prism version 10.0.0 for Windows (GraphPad Software, MA, USA). Blood sample and tracer data are presented as curves showing means ± SEM or bar plots showing means ± SD. Western blot results are shown as dot plots with medians. The result estimates in the text are presented as means ± SD. Effect estimates are presented as mean differences or median ratios with 95% CI and p value. A p value <0.05 was considered statistically significant.