A recent Endocrinology digest,
as sent to subscribers

Study Type: Retrospective cohort study (propensity score-matched, real-world claims data analysis)

Key Question: Does empagliflozin's cardiovascular benefit in older adults with type 2 diabetes vary by sociodemographic factors or baseline cardiovascular disease status?

Key Findings:

• Versus DPP4is, empagliflozin reduced MACE (HR 0.77, 99.9% CI 0.69–0.85), HHF (HR 0.72, 0.67–0.79), and ACM (HR 0.66, 0.57–0.76); versus GLP-1RAs, HHF was reduced (HR 0.88, 0.81–0.95).
• Absolute risk reductions (IRDs) were consistently larger in patients with prevalent CVD, though relative risk (HR) heterogeneity by CVD status was seen only for HHF.
• No effect modification by age, sex, race/ethnicity, or socioeconomic status, though a numeric trend favoured greater absolute benefit in those ≥75 years.

Clinical Relevance: Supports equitable SGLT2 inhibitor benefit across diverse elderly NHS patient populations, reinforcing prioritisation in those with existing cardiovascular disease for greatest absolute risk reduction.

Limitations: Observational design using Medicare claims (US, ≥65 years) limits causal inference and generalisability to younger or non-US populations.

Study Type: Retrospective nationwide cohort study

Key Question: Does age at type 1 diabetes diagnosis influence long-term mortality, cardiovascular, and kidney outcomes?

Key Findings:

• Each 1-year older age at diagnosis was associated with 1.7% lower all-cause mortality risk (HR 0.98, 95% CI 0.98–0.99), equating to ~8.2% lower risk per 5-year increase.
• Cardiovascular mortality showed a stronger inverse association (HR 0.97, 95% CI 0.97–0.98) than non-cardiovascular mortality (HR 0.99, 95% CI 0.98–0.99).
• Excess risk versus the general population was highest for diagnosis at age 0–10 years, with HRs of 3.23 (all-cause mortality), 5.73 (cardiovascular mortality), and 3.01 (non-cardiovascular mortality).

Clinical Relevance: Reinforces the need for intensive, early cardiometabolic risk surveillance and aggressive management in paediatric-onset type 1 diabetes within NHS paediatric-to-adult transition pathways, given disproportionately high long-term cardiovascular mortality risk in this group.

Limitations: Observational Swedish registry data limit causal inference and generalisability to UK populations with differing diabetes care models.

Study Type: International position statement / expert consensus guideline (not original research).

Key Question: How should CGM, insulin pumps, and automated insulin delivery (AID) systems be practically and safely used in non-critically ill hospitalised adults with diabetes?

Key Findings:

• Provides selection criteria for continuing or initiating diabetes technology use during admission, including non-critical care settings.
• Defines CGM-derived glycaemic metrics/targets, alarm thresholds for hypo-/hyperglycaemia, and insulin titration guidance for pump/AID users.
• Addresses practical workflow issues: perioperative management, imaging-related device considerations, EHR data integration, staff training, and discharge planning.

Clinical Relevance: Offers UK endocrinology teams a structured framework to standardise inpatient diabetes technology use, supporting NHS efforts (e.g., Joint British Diabetes Societies guidance) to reduce inpatient hypoglycaemia/hyperglycaemia and safely manage the growing number of patients admitted with pumps/AID/CGM.

Limitations: As a consensus statement rather than primary data, recommendations rely on expert opinion and existing evidence base rather than new controlled outcome data.

Study Type: Commentary/call to action (non-systematic, position statement)

Key Question: Should automated insulin delivery (AID) systems be adapted and studied specifically for use in pregnant women with type 1 diabetes (T1D)?

Key Findings:

• Current FDA-approved AID systems are largely not labelled or designed for pregnancy, and most reduce insulin delivery when CGM values fall within ranges considered optimal for fetal well-being, conflicting with tighter pregnancy glycemic targets.
• Pregnant women have been systematically excluded from AID clinical trials, leaving minimal evidence to guide device use in this population despite AID being standard of care for non-pregnant adults with T1D.

Clinical Relevance: UK endocrinologists managing pregnant women with T1D should recognise that existing AID algorithms may be poorly suited to pregnancy targets, informing counselling, off-label use decisions, and advocacy for inclusion in trials and NHS-led real-world evaluations.

Limitations: As a commentary, it presents no original data or systematic evidence review, relying on expert opinion and identified evidence gaps.

Study Type: Prespecified secondary analysis of a randomised controlled trial (SELECT)

Key Question: Can routinely available baseline clinical characteristics predict the magnitude of peak weight loss achieved with long-term semaglutide use?

Key Findings:

• Among 7,594 participants treated ≥1 year, median peak weight loss was 12.5% (IQR 7.9–17.9%), reached at a median of 19 months.
• The full predictive model explained <10% of variability in peak weight loss; female sex alone accounted for ~8% (85% of the model's explanatory power).
• Higher maximum dose was associated with greater weight loss (6.0% at 0.24 mg vs 13.3% at 2.4 mg); women on the highest dose achieved the greatest response (16.6% vs 12.4% in men).

Clinical Relevance: For UK clinicians prescribing semaglutide within NHS obesity/cardiometabolic pathways, individual weight-loss response cannot be reliably forecast from baseline characteristics, reinforcing the need for dose titration to maximum tolerated dose and ongoing monitoring rather than pre-treatment risk stratification.

Limitations: The model's low explanatory power suggests unmeasured biological, behavioural, or genetic factors drive treatment response, limiting clinical applicability of the predictors examined.

Study Type: Basic/translational laboratory study (human pancreatic islet experiments)

Key Question: Does harmine's ability to enhance beta cell identity and function occur solely through DYRK1A inhibition, or via a distinct second target?

Key Findings:

• All tested DYRK1A inhibitors (harmine, 2-2c, 5-IT, INDY, leucettine, CC-401, GNF4877) induced beta cell proliferation via DYRK1A inhibition, but DYRK1A gene silencing did not affect identity markers, showing proliferation and identity effects are mechanistically separate.
• Only harmine, 2-2c and 5-IT enhanced beta cell identity/phenotypic markers (e.g., PDX1, MAFA, NKX6.1), acting through activation of a novel, as-yet unidentified "Target 2" that engages the PKA pathway.
• This PKA-mediated effect is independent of DYRK1A inhibition, indicating not all DYRK1A inhibitors are functionally equivalent.

Clinical Relevance: For NHS-based regenerative beta cell therapy research in diabetes, these findings highlight that future drug selection/development should prioritise agents combining proliferative and functional-maturation effects, rather than assuming class-wide equivalence among DYRK1A inhibitors.

Limitations: Findings are based on in vitro human islet studies with an unidentified molecular target, limiting immediate clinical translatability.

Study Type: Pooled analysis of three prospective observational studies using ecological momentary assessment and multilevel network modelling.

Key Question: How do daily diabetes-specific emotional burdens interact and accumulate over time to influence subsequent emotional well-being in type 1 and type 2 diabetes?

Key Findings:

• In type 1 diabetes, guilt about self-management predicted next-day increases in loneliness (β=0.09), feeling overwhelmed (β=0.05), restricted (β=0.05), and hyperglycaemia burden (β=0.05, all p<0.05); in type 2 diabetes, guilt was instead predicted by prior loneliness and hypoglycaemia burden.
• Greater overall network density (stronger day-to-day burden interconnection) predicted incident elevated diabetes distress (OR 1.50) and depressive symptoms (OR 2.08, p<0.001), with guilt and feeling restricted as key drivers.

Clinical Relevance: Identifies guilt around self-management as a modifiable, central emotional trigger that could be targeted in structured diabetes education and psychological support pathways within NHS diabetes services to prevent escalation to distress or depression.

Limitations: Observational design with self-report EMA data limits causal inference despite temporal modelling.

Study Type: Genome-wide association study meta-analysis with integrated functional/epigenetic characterisation

Key Question: Does non-HLA genetic risk for type 1 diabetes overlap mechanistically with genetic determinants of insulin secretion and insulin resistance?

Key Findings:

• Meta-analysis (26,198 cases, 36,733 controls) identified 69 non-HLA risk loci, including three novel signals (AHI1, TRAF1, ADCY7 regions; ORs 1.07–1.09)
• 26/69 variants associated with insulin secretion or resistance indices; fine-mapping prioritised 136 functional variants enriched in pancreatic islet, liver, muscle and adipose regulatory elements
• ADCY7 (16q12.1 locus) identified as a probable causal gene: cytokine exposure downregulated Adcy7 in beta cells, affecting insulin secretion and apoptosis, while adipocyte knockdown improved insulin sensitivity

Clinical Relevance: Suggests type 1 diabetes pathogenesis involves shared genetic pathways with metabolic dysfunction (beyond autoimmunity alone), relevant to UK precision-medicine efforts (e.g., MODY/T1D reclassification, biomarker refinement) and future risk stratification or therapeutic targeting.

Limitations: Findings are largely computational/functional-genomic with limited in vivo human validation, so causal clinical relevance remains preliminary.