To critically examine the evolving evidence on glucagon-like peptide-1 receptor agonists (GLP-1RAs) and the dual glucose-dependent insulinotropic polypeptide (GIP)/GLP-1 receptor agonist tirzepatide as adjuncts to insulin therapy in type 1 diabetes (T1D), with particular emphasis on their benefit–risk balance in individuals with overweight or obesity.
This narrative review examines evidence on liraglutide, semaglutide and tirzepatide in T1D, focusing on effects on body weight and insulin requirements, alongside hypoglycaemia, ketosis, diabetic ketoacidosis and gastrointestinal tolerability.
Early trials with liraglutide demonstrated reductions in body weight and insulin requirements but raised concerns regarding hypoglycaemia and ketosis. More recently, semaglutide and tirzepatide have renewed interest in incretin-based therapy in T1D, particularly among individuals with obesity. Their greater effects on weight and insulin requirements introduce a clinically important safety challenge: insulin must be reduced sufficiently to prevent hypoglycaemia while maintaining adequate insulinization to suppress lipolysis and ketogenesis. Gastrointestinal adverse events may further destabilize this balance through reduced carbohydrate intake and dehydration.
Current evidence supports a phenotype-directed approach centred primarily on adults with T1D and obesity, accompanied by careful patient selection, individualised insulin adjustment, glucose and ketone monitoring, and structured patient education. The clinical question has shifted from whether GLP-1–based therapies are effective in T1D to which patients are most likely to achieve a favourable benefit–risk balance and how these therapies can be used safely.
]]>Accurate non-invasive identification of metabolic dysfunction-associated steatohepatitis (MASH) and at-risk MASH remains a major unmet need in MASLD. We evaluated the diagnostic performance and optimal cutoffs of commonly used non-invasive tests (NITs) in a real-world setting.
In this multicentre prospective study, 328 cases from 226 participants underwent contemporaneous liver biopsy, MRI-PDFF, FibroScan and biochemical assessment, and were classified into MASL, MASH and at-risk MASH based on histopathology. Diagnostic performance was evaluated using area under the receiver operating characteristic curve (AUC), sensitivity and specificity.
CAP at 248 dB/m detected histological steatosis with AUC of 0.80 (sensitivity: 94.7%, specificity: 30.4%); raising the threshold to 304 dB/m increased specificity to 91.3% (sensitivity: 60.3%). The two-step CAP–ALT algorithm achieved a more balanced sensitivity and specificity (78.3% and 82.6%, respectively; AUC: 0.80). MRI-PDFF quantified steatosis accurately (AUC: 0.92; cutoff: 7.78%; sensitivity: 77.1%; specificity: 100%) and differentiated grades, with diagnostic discordance versus histopathology mainly in mild disease. FAST score (AUC: 0.85; cutoff: 0.29; sensitivity: 70.0%; specificity: 83.3%) and MRI-PDFF (AUC: 0.82; cutoff: 17.78%; sensitivity: 56.2%; specificity: 92.6%) identified MASH, while LSM (AUC: 0.92; cutoff: 8.95 kPa; sensitivity: 86.5%; specificity: 84.1%) detected at-risk MASH.
CAP provides an accessible approach for steatosis detection, whereas MRI-PDFF offers quantitative assessment of liver fat and may provide additional information relevant to MASH. FAST may serve as an initial tool for MASH risk stratification, while LSM provides effective identification of at-risk MASH. These findings support a targeted, stepwise non-invasive approach to MASLD assessment and follow-up.
]]>The continuous glucose monitoring (CGM)-derived glucose management indicator (GMI) is used for estimating glycated haemoglobin (HbA1c). This study assessed agreement between GMI—calculated using both original and updated (uGMI) models—and laboratory-measured HbA1c across Dexcom, MiniMed and Abbott sensors.
Real-world data from adults with type 1 diabetes using automated insulin delivery were retrospectively analysed. Δ(GMI-HbA1c) and Δ(uGMI-HbA1c) were evaluated over 90 days preceding HbA1c (n = 383) and over 28 days preceding HbA1c (n = 375). Agreement was measured as directional difference, indicating GMI overestimation or underestimation of HbA1c, and absolute difference, indicating magnitude of agreement between GMI and HbA1c. Agreement was assessed across sensor groups using ANCOVA, adjusting for coefficient of variation and ethnicity.
Although mean directional differences were statistically significant for Δ(GMI-HbA1c) and Δ(uGMI-HbA1c) across all sensor groups (p < 0.001), differences were small and not clinically important, 90-day adjusted means (95% CI)—Δ(GMI-HbA1c): Dexcom 0.17% (0.09, 0.25), MiniMed −0.15% (−0.28, −0.02), Abbott −0.08% (−0.25, 0.09); Δ(uGMI-HbA1c): Dexcom 0.22% (0.14, 0.30), MiniMed −0.10% (−0.23, 0.02), Abbott −0.02% (−0.19, 0.15). Mean absolute difference did not significantly differ across sensor groups—Δ(GMI-HbA1c): 0.42%–0.51% (p = 0.353), Δ(uGMI-HbA1c): 0.41%–0.52% (p = 0.221).
GMI/uGMI performed consistently across sensor groups. Where clinically meaningful discrepancies between GMI/uGMI and HbA1c arise, these warrant interpretation within the context of individual clinical and laboratory factors rather than attribution to sensor manufacturer, reinforcing the need for personalised use of GMI alongside HbA1c.
]]>Reports of diabetic retinopathy after potent glucose-lowering treatment create uncertainty for clinicians, particularly when studies measure different outcomes in patients with different baseline retinal disease. This review examines diabetic retinal outcomes after glucagon-like peptide-1 receptor agonist or tirzepatide initiation.
This critical narrative review uses targeted, non-systematic study selection. It distinguishes disease onset or progression from detection, treatment use and subsequent visual function.
The SUSTAIN-6 retinopathy-complication signal remains clinically relevant; post hoc associations with early glycated haemoglobin reduction do not establish causal mediation. Contemporary cohorts report differing associations across baseline retinal strata, comparators and endpoints. Positive signals should be presented alongside their absolute event proportions and limitations, rather than dismissed because observational adjustment is imperfect. Conversely, non-significant results do not establish equivalence or class-wide safety.
Current guidance supports establishing retinal status and coordinating diabetes and eye care during treatment intensification. The implications differ between patients with recent screening and no known retinopathy, established non-proliferative disease, and active sight-threatening disease or recent retinal treatment. No universal safe rate of glycated haemoglobin decline or retinal-protective titration regimen has been established. Future studies should align eligibility and treatment initiation, distinguish incident disease from progression, and use scheduled retinal assessments. A comparison between two specific medicines can inform treatment choice but cannot exclude risks shared by both.
]]>To evaluate whether cardiovascular effects of glucagon-like peptide-1 receptor agonists (GLP-1RAs) differ by age.
PubMed and Embase were searched to 1 July 2026 for randomized placebo-controlled GLP-1RA trials in adults with type 2 diabetes or overweight/obesity reporting age-stratified major adverse cardiovascular events (MACE; a composite of cardiovascular death, non-fatal myocardial infarction, and non-fatal stroke). Hazard ratios (HRs) for participants aged < 65 and ≥ 65 years were pooled using random-effects models with restricted maximum likelihood estimation and Hartung–Knapp adjustment. Effect modification was assessed by pooling trial-specific ratios of HRs (RHRs) for ≥ 65 versus < 65 years. Where reported, effects were assessed across < 65, 65 to < 75 and ≥ 75 years.
Eight trials (SUSTAIN-6, EXSCEL, Harmony Outcomes, PIONEER 6, REWIND, AMPLITUDE-O, SELECT and SOUL) comprising 71 926 participants were included. GLP-1RAs reduced MACE at < 65 years and ≥ 65 years, with pooled HRs of 0.81 (95% CI 0.69–0.94; I
2
= 64.1%) and 0.84 (95% CI 0.79–0.90; I
2
= 0.0%), respectively. The pooled RHR showed no clear evidence of effect modification (1.03, 95% CI 0.85–1.25), although between-trial heterogeneity was high (I
2
= 59.3%). Among trials reporting three age strata, RHRs showed no clear evidence of effect modification for 65 to < 75 or ≥ 75 years versus < 65 years, but data were available from only four trials, and the ≥ 75-year group was relatively small.
Across the included trial populations, GLP-1RAs lowered MACE in the age groups examined, with no clear evidence that relative cardiovascular effects differed by age.
]]>Type 1 diabetes (T1D) in adults is characterised by a less aggressive disease course than in children. Most patients retain preserved C-peptide (> 0. $.$2 nmol/L) and some of them benefit from a protracted phase without the need for insulin. Here, we propose a hypothesis-generating therapeutic paradigm with once-weekly basal insulin combined with an SGLT2 inhibitor (SGLT2is) as initial therapy in T1D patients with preserved C-peptide. Once-weekly basal insulin, approved in Europe and now also in the US, confirm comparable HbA1c reductions to insulin degludec or glargine, albeit always alongside prandial insulin. SGLT2is, approved as adjuncts in T1D in select regions, can control prandial peaks without boluses. No trials have tested the combination of weekly insulin with SGLT2is in adult T1D patients. Nonetheless, the rationale is robust especially in adults with C-peptide > 0·2 nmol/L. In this respect, we urge proof-of-concept trials of 12-months randomised studies in adults ≥ 18 years, to evaluate this combination.
]]>Diabetes mellitus (DM) is a common comorbidity of psoriatic arthritis (PsA); however, its global prevalence and the impact of treatment exposure remain unclear. This study aimed to assess the global prevalence of DM in patients with PsA using different subgroup analyses.
A systematic search of PubMed, Embase, the Cochrane Library and the Web of Science was performed from inception until February 2025 for observational studies reporting the prevalence or incidence of DM in PsA patients. Random-effects models were applied to pool estimates, and subgroup, meta-regression, and sensitivity analyses were conducted to explore heterogeneity. Publication bias was evaluated using the Egger’s test and the trim-and-fill method.
Sixty-three studies, including 293,582 patients with PsA, were analysed. Patients with PsA included in this study were divided into treatment-disclosed and non-treatment-disclosed groups. DM overall prevalence in patients with PsA was 13.84% in the non-treatment-disclosed population and 13.50% in the treatment-disclosed cohort. Geographically, the comorbidity prevalence rates were higher in South America (37.98%) and Asia (16.76%) than in Europe (12.81%) and North America (12.38%) in the non-treatment-disclosed population. In the treatment-disclosed population, prevalence was also highest in South America (28.13%), followed by Asia (18.30%). Meta-regression analysis identified age, type of DM, region, and study design as contributors to heterogeneity. Evidence of small-study effects was detected in the T2DM subgroup and cohorts receiving systemic therapy plus biologics.
Subgroup analysis shows that the prevalence of DM is associated with the type of study and the region in which the patients were located. These findings support routine metabolic screening for PsA and highlight the need for longitudinal studies to track disease activity, treatment patterns, and metabolic outcomes.
]]>CagriSema combines cagrilintide with semaglutide and may improve glycaemic control and body weight in Type 2 diabetes. We evaluated its dose- and comparator-specific efficacy and safety.
PubMed, Embase, CENTRAL and ClinicalTrials.gov were searched through 30 June 2026 for randomized controlled trials. Analyses were stratified by CagriSema dose and comparator. For HbA1c and body weight, efficacy/trial-product estimands were aligned where sufficiently reported, with treatment-policy estimates examined in sensitivity analyses. Bayesian random-effects models generated posterior medians and 95% credible intervals (CrIs).
Six trials representing 4554 potentially eligible participants with Type 2 diabetes were included. Compared with placebo, CagriSema 2.4 mg produced between-group differences of −1.87 percentage points in HbA1c (95% CrI, −2.08 to −1.62) and −12.69 percentage points in percentage body-weight change (95% CrI, −13.78 to −11.61). Versus semaglutide 2.4 mg, the additional HbA1c reduction was modest and uncertain (−0.21 percentage points; 95% CrI, −0.62 to 0.11), whereas additional body-weight reduction remained probable (−5.84 percentage points; 95% CrI, −10.08 to −2.15). Gastrointestinal adverse events and adverse-event-related discontinuation were increased versus placebo (RR, 2.27 [95% CrI, 1.81 – 3.00] and 3.16 [1.50 to 6.87], respectively), while estimates versus semaglutide were less precise.
CagriSema provides substantial glycaemic and body-weight benefits versus placebo. Its incremental advantage over dose-matched semaglutide is more evident for body-weight reduction, while comparative gastrointestinal tolerability and discontinuation risk remain uncertain.
]]>Metformin is increasingly used to treat gestational diabetes mellitus and other insulin-resistant conditions during pregnancy. Because it crosses the placenta, concerns remain regarding its potential effects on offspring growth and metabolic programming. This systematic review evaluated the effects of prenatal metformin exposure on infant nutritional status and adiposity during the first year of life.
PubMed and Scopus were searched from inception through May 2026 for randomized controlled trials comparing metformin with placebo or standard care during pregnancy. Eligible studies reported infant nutritional status or adiposity-related outcomes from birth to 12 months. Risk of bias was assessed, and results were synthesized narratively.
Eight publications from six randomized controlled trials were included. Birthweight and birth size classifications were the most frequently reported outcomes. Metformin was not consistently associated with differences in birthweight or other anthropometric measures. One trial reported lower birthweight, while two reported fewer large-for-gestational-age or macrosomic infants. No study assessed direct measures of infant adiposity.
Current evidence does not consistently support an effect of prenatal metformin exposure on early infant anthropometry. Limited postnatal data and absent adiposity measures limit conclusions regarding postnatal growth and early metabolic programming and support standardized longitudinal body composition assessment.
]]>Although closely correlated, apolipoprotein B (apoB) reflects atherogenic particle number, whereas low-density lipoprotein cholesterol (LDL-C) measures cholesterol carried by LDL particles. We examined whether their associations with incident Type 2 diabetes (T2D) diverged after accounting for each other.
We included UK Biobank participants recruited in 2006–2010 without identified baseline diabetes. ApoB and LDL-C were assessed in baseline blood samples, and incident T2D was ascertained through 19 December 2022. Cause-specific Cox models estimated hazard ratios (HRs) per 10 mg/dL higher concentration. Primary models used the same participants and nonlipid covariates, with each biomarker entered individually and both entered in one model. Additional analyses assessed combined metabolic adjustment and time-varying associations.
Among 402 144 participants followed for 5.33 million person-years, 16 537 developed T2D. ApoB and LDL-C were highly correlated (r = 0.958). When modelled individually, HRs per 10 mg/dL were 1.059 (95% CI: 1.052–1.066) for apoB and 1.018 (95% CI: 1.013–1.023) for LDL-C. When included together, summary HRs over the full follow-up were 1.502 (95% CI: 1.468–1.537) for apoB and 0.759 (95% CI: 0.745–0.772) for LDL-C. Combined cardiometabolic and HbA1c adjustment attenuated both conditional associations, with HRs of 1.151 (95% CI: 1.122–1.181) for apoB and 0.890 (0.873–0.908) for LDL-C. Time-varying HRs at Years 1 and 10 were 1.658 and 1.464 for apoB, and 0.691 and 0.777 for LDL-C, respectively.
Higher apoB at a given LDL-C was associated with a higher hazard of incident T2D, whereas higher LDL-C at a given apoB was associated with a lower hazard.
]]>EDENT1FI is a European initiative aiming to implement a roadmap for screening, monitoring and prevention of childhood Type 1 diabetes (T1D). The Betty project is a nationwide T1D screening program recently established in Czechia within the framework of EDENT1FI. Here we report its results after having tested 1% of the Czech paediatric population.
The screening targets children aged 2–17.9 years, regardless of whether they have a family history of T1D. It follows a two-step process for detecting islet-specific autoantibodies (anti-GAD, anti-IA2, anti-ZnT8 and IAA). If the capillary screening sample is positive in the first-level test (screening ELISA yielding a summary signal of three autoantibodies), the same sample is retested at the EDENT1FI reference laboratory using a battery of 4 second-level tests for individual autoantibodies (LIPS or RBA). When any autoantibody is detected, a venous sample is requested for confirmation. If multiple autoantibodies are confirmed from this venous sample, the child is diagnosed with early-stage T1D and offered follow-up tailored to their individual estimated rate of progressing to clinical diabetes. Children with only a single confirmed autoantibody are offered a less intensive follow-up outside the EDENT1FI protocol.
The Betty project collected 19 044 capillary screening samples (1% of the paediatric population) between April 2024 and February 2026. Samples were collected across 270 screening sites mostly at primary care and via home sampling (4.1% samples). The majority of participants (89.7%) were from the general population having no first-degree relative (FDR) with T1D. We found a prevalence of multiple autoantibodies of 0.39% (95% CI: 0.30%–0.51%) in the general population and 2.8% (95% CI: 2.1%–3.7%) prevalence among FDRs of subjects with T1D. At the time of writing, of 201 children with non-negative capillary screening sample, 102 were diagnosed with early-stage T1D by a confirmatory venous sample and 83 had their T1D stage assessed: 77 (93%) were in Stage 1, 4 (4.8%) in Stage 2 and 2 (2.4%) in Stage 3. Further seven children, who were identified by capillary screening, manifested with Stage 3 T1D before their staging could be completed. All families who came for staging and education visits consented to long-term follow-up.
This nationwide T1D screening program successfully reached approximately 1% of the Czech paediatric population, demonstrating the feasibility of implementing population-based islet autoantibody screening based on the common EDENT1FI protocol. These findings provide an important foundation for the potential future implementation of routine population-based T1D screening within the Czech national healthcare system and offer valuable insights for other countries considering similar screening programmes.
]]>Bariatric surgery is highly effective for inducing remission of Type 2 diabetes mellitus (T2DM), yet the frequency and timing of diabetes medication re-initiation after achieving medication-free status are poorly characterized.
To estimate diabetes medication re-initiation after a postoperative medication-free interval and determine whether longitudinal postoperative weight and HbA1c trajectories provide incremental prognostic information beyond baseline diabetes severity.
We performed a retrospective cohort study using electronic health record data mapped to the OMOP Common Data Model. Adults with preoperative T2DM undergoing sleeve gastrectomy or Roux-en-Y gastric bypass between 2013 and 2026 were included. Patients entered the re-initiation risk set after achieving a postoperative medication-free interval of at least 90 days. Time to first subsequent diabetes medication restart was analysed using Cox models incorporating baseline clinical factors and longitudinal postoperative weight and HbA1c trajectories.
Among 8012 unique patients undergoing bariatric surgery, 2197 (27.4%) had documented T2DM before the index operation. Of these, 2056 (93.6%) achieved a postoperative medication-free interval of at least 90 days and entered the re-initiation risk set; 882 (42.9%) subsequently re-initiated diabetes medication. Among patients who re-initiated therapy, median time to re-initiation from risk-set entry was 1.04 years (IQR, 0.30–2.84). In the primary Cox model, diabetes duration (HR: 1.219 per SD, 95% CI: 1.152–1.289), preoperative insulin use (HR: 1.627, 95% CI: 1.403–1.886) and baseline HbA1c (HR: 1.089 per SD, 95% CI: 1.030–1.151) were associated with increased medication re-initiation, whereas Roux-en-Y gastric bypass (RYGB) was associated with lower risk (HR: 0.778, 95% CI: 0.684–0.883). Age was not independently associated with re-initiation. Postoperative weight slope had a small association with re-initiation (HR: 1.042 per SD, 95% CI: 1.028–1.056; C-index = 0.6237). In the 1251-patient longitudinal HbA1c cohort, addition of HbA1c slope improved model fit (likelihood-ratio p = 0.0075) but provided little improvement in discrimination.
Diabetes medication re-initiation following bariatric surgery was common and occurred relatively early after achievement of medication-free status. More than 40% of patients re-initiated therapy during follow-up. This frequency exceeds rates reported in much of the previous literature on biochemical diabetes recurrence, emphasizing that medication re-initiation and biochemical relapse represent distinct postoperative outcomes. Baseline diabetes severity was associated with medication re-initiation, whereas longitudinal postoperative weight and HbA1c trajectories provided limited additional prognostic information.
]]>