This Week in Endocrinology — Sep 2, 2026

Welcome to This Week in Endocrinology. This week we're covering 10 notable papers spanning obesity pharmacotherapy and weight loss as disease modification, diabetes safety and diet, and a cluster of papers across thyroid, adrenal and pituitary disease. Let's dive in.

We'll start with obesity, where two papers from very different genres arrived in the same week. In Annals of Internal Medicine, Moiz and colleagues published an updated systematic review of glucagon-like peptide-1 receptor agonists and co-agonists for weight loss in adults with overweight or obesity but without diabetes [1]. Thirty-eight randomised trials, nearly 26,000 participants, with 14 new trials added since the last iteration. The hierarchy is now fairly clear: placebo-subtracted weight loss of up to about 6 percent for liraglutide, roughly 15 percent for subcutaneous semaglutide, similar for oral semaglutide, about 12 percent for orforglipron, and about 19 percent for tirzepatide. The emerging multiagonists went further still, with numerically greater reductions of around 24 percent for amycretin and 22 percent for retatrutide, though those rest on fewer trials. Gastrointestinal side effects remained the dominant tolerability issue, affecting about three quarters of treated patients versus about 40 percent on placebo, but discontinuation for adverse events was around 11 percent versus 3 percent, serious events were uncommon, and no new safety signals emerged. Head-to-head data confirmed semaglutide beats liraglutide, and tirzepatide and cagrilintide-semaglutide beat semaglutide. The authors flag heterogeneity that precluded pooling and inconsistent safety reporting, so treat the rank order as approximate rather than precise.

The companion piece is a Review in The Lancet Diabetes and Endocrinology from Sattar and colleagues, which asks the harder question: if we can now reliably deliver 14 to 20 percent weight loss, which diseases does that actually modify [2]? Using a triangulation framework across epidemiology, Mendelian randomisation, observational weight-change studies and randomised trials, they argue that robust trial evidence supports weight loss as genuinely disease-modifying in type 2 diabetes and in heart failure with preserved ejection fraction, but that musculoskeletal, respiratory, cardiovascular and mental health conditions remain substantially under-investigated despite strong observational links. Their practical point for the clinic is that the magnitude of weight loss needed likely differs by target condition, and we should not assume that osteoarthritis, sleep apnoea or depression will respond in proportion to the numbers we see on the scale. They also call attention to global access gaps and the absence of trials in diverse populations. Read together, these two papers say we have the drugs and increasingly know how much weight they take off; what we still lack is outcome evidence for most of the multimorbidity that drives us to prescribe them.

Turning to diabetes safety, the standout is a Scandinavian cohort and nested case-control study, also in The Lancet Diabetes and Endocrinology, on ketoacidosis with SGLT2 inhibitors in routine care [3]. Kadesjö and colleagues drew on nationwide registers from Sweden, Denmark and Norway, covering more than 322,000 treatment episodes among about 282,000 patients with type 2 diabetes, with 1,452 ketoacidosis events over a median follow-up of just over a year. The overall incidence was about 2.4 events per 1,000 person-years. Two findings should change how you counsel patients. First, although risk was highest shortly after initiation, it persisted throughout follow-up, so a one-time counselling session at prescribing is not enough. Second, risk is extraordinarily concentrated: a high HbA1c of 83 millimoles per mole or above was associated with roughly a fifteen-fold higher odds of ketoacidosis compared with well-controlled patients, and malnutrition, previous ketoacidosis and a body mass index below 20 each carried roughly a ten-fold increase. Recent hypoglycaemia raised the odds around fivefold. Infection was the most common precipitant, present in about a third of cases versus around 6 percent of controls. Alcohol intoxication, acute renal events, acute abdomen, stroke and major surgery showed the strongest associations, though the authors caution that milder transient exposures may be over-estimated because of under-registration in controls, and exploratory analyses suggested many of these associations are general to type 2 diabetes rather than specific to the drug class. The aftermath matters too: a year after an event, only about a quarter of patients remained on an SGLT2 inhibitor, and insulin use had risen from about a third to nearly three quarters. The practical message is to reassess risk at every visit and give explicit sick-day rules to pause the drug during acute illness, surgery or reduced intake.

Staying with ketones, Diabetes Care published a modelling study from Zhang and colleagues asking how often people with type 1 diabetes actually need to check capillary ketones on well days to predict near-term ketoacidosis risk [4]. Using regression and gradient-boosted tree models on the pooled EASE 2 and EASE 3 trial repository of just over 1,400 participants, they simulated different testing frequencies over a one-month window. Once-weekly well-day testing was the lowest frequency that preserved predictive accuracy compared with twice weekly, with no significant loss in discrimination in either modelling approach. Discrimination was modest in absolute terms, in the range of about 0.7, so this is a risk-stratification tool rather than a diagnostic one. Still, weekly testing using strips already in the cupboard before they expire is a low-burden way to identify who needs closer attention.

Also in Diabetes Care, a secondary analysis of the DASH4D controlled feeding trial from Fang and colleagues looked at glycaemic biomarkers [5]. In 101 adults with type 2 diabetes, mean age 67, predominantly Black participants, each fed four diets in random order for five weeks apiece, the DASH4D diet significantly lowered fructosamine, fasting glucose by about 4.5 milligrams per decilitre, and HbA1c by about 0.09 percentage points versus a typical American diet. These are statistically real but clinically small effects, and the HbA1c change over five weeks partly reflects the short feeding window. It's supportive evidence for recommending a DASH-style diet in diabetes, not a substitute for pharmacotherapy.

Our thyroid and adrenal cluster brings three shorter but clinically interesting reports. In European Journal of Endocrinology, Son and colleagues used the Korean national insurance database to follow more than 311,000 adults aged 40 and over, and found that longer cumulative antibiotic exposure in the preceding five years was associated with higher subsequent risk of Graves' disease [6]. Compared with non-users, 15 days or more of antibiotics was associated with roughly a 40 percent higher relative risk, and exposure to four or more antibiotic classes with about a 50 percent increase, with a clear dose-response trend. But note the absolute numbers: incidence rose from about 4.6 to about 8 per 10,000 person-years. This is a plausible microbiome signal, not a reason to change prescribing, and the authors are explicit that causality is not established.

In Thyroid, Furumura and colleagues addressed a metabolic puzzle in patients with papillary thyroid microcarcinoma under active surveillance who receive mild TSH-lowering levothyroxine [7]. Comparing over 300 treated patients with matched untreated controls, free thyroxine rose as expected, but free triiodothyronine was unchanged, while reverse triiodothyronine and the reverse-T3-to-T3 ratio were significantly higher and the free-T3-to-free-T4 ratio lower. Reverse T3 correlated with free thyroxine but not with TSH or T3. The interpretation is a compensatory buffering mechanism shunting excess T4 down the inactivating pathway, which is reassuring if you use mild TSH suppression in this setting, and a reminder not to read a rising free T4 as tissue-level over-replacement.

Also in European Journal of Endocrinology, the MAPA proof-of-concept study from Rossi and colleagues tested whether a single oral dose of roxithromycin could serve as a functional marker of KCNJ5-mutated aldosterone-producing adenoma [8]. Among 373 hypertensive patients challenged, 18 had KCNJ5-mutated adenomas, and aldosterone fell selectively in that group only, not in wild-type adenomas or in patients without primary aldosteronism. Importantly, blood pressure did not fall in the mutated group, and the modest blood pressure reduction seen in non-primary-aldosteronism hypertensives appears aldosterone-independent, attributable in mouse work to attenuated angiotensin two signalling and nitric-oxide-dependent vasodilation. With 18 mutated cases this is hypothesis-generating, but it points toward a non-invasive route to precision subtyping.

Rounding out the week, European Journal of Endocrinology also published a longitudinal quality-of-life analysis from the PROMPT study of metyrapone in 49 adults with endogenous Cushing's syndrome [9]. Over 36 weeks, CushingQoL improved by about 10 points, with close to half OF PATIENTS achieving at least a 10-point gain, and the Tübingen CD-25 also improved. The pattern is instructive: eating behaviour and depressive symptoms improved early by week 12, while sexual, social and environmental domains recovered later at weeks 24 to 36, and bodily restrictions and cognition did not significantly improve at all. Health anxiety and appearance concerns remained elevated throughout despite cortisol normalisation. Counsel patients that biochemical control is the beginning, not the end, of recovery. Finally, the Growth Hormone Research Society published an updated consensus statement on long-acting growth hormone in the same journal, drawing on up to seven years of data in over 8,000 individuals [10]. Short- to mid-term efficacy appears comparable to daily somatropin provided dosing is individualised and IGF-1 monitored, with no new major safety signals, injection-site reactions the commonest adverse event, and anti-drug antibodies of apparently limited clinical impact. Key gaps remain in transition care, BMI trajectories, metabolically high-risk groups, and cancer survivors, where recurrence data are simply not yet available.

If you only have time for one paper this week, make it the Scandinavian SGLT2 inhibitor ketoacidosis study [3]. Almost every endocrinologist prescribes these drugs weekly, and it gives you a concrete, actionable risk profile plus the crucial insight that risk does not disappear after the first few months.

Here are the key takeaways from this week in Endocrinology. First, ketoacidosis risk with SGLT2 inhibitors is heavily concentrated in patients with very high HbA1c, low body mass index, malnutrition or prior ketoacidosis, and it persists beyond initiation, so sick-day rules need repeating. Second, the obesity pharmacotherapy hierarchy is now reasonably well defined, with tirzepatide ahead of semaglutide ahead of liraglutide, and multiagonists pushing further, but outcome evidence for disease modification exists mainly for type 2 diabetes and heart failure with preserved ejection fraction. Third, once-weekly well-day capillary ketone testing appears sufficient for one-month ketoacidosis risk stratification in type 1 diabetes. Fourth, a DASH-style diet tailored for diabetes produces small but real glycaemic improvements. And fifth, in Cushing's syndrome, expect mood and appetite to improve early but social, sexual, cognitive and appearance-related burden to lag well behind cortisol normalisation.

That's your roundup for This Week in Endocrinology. The full transcript and references are available on the episode page in your AudioScholar library. This is an AI-curated summary — for clinical decisions, always consult primary sources and current guidelines. See you next week.

If this weekly briefing is useful, follow the show in your podcast app so new episodes arrive automatically. And if it saves you time, a quick rating in your podcast app helps other physicians find the show.

This is an automated summary generated by artificial intelligence, which can make mistakes. Always review the original source materials.