Weekly Issue — 2025-08-24

Non-alcoholic fatty liver disease (NAFLD) is the quiet epidemic running alongside the obesity one. It can sit silently for years, then progress to steatohepatitis (NASH), fibrosis, and eventually cirrhosis. In many countries there is still no approved medical therapy for it.

A 2025 systematic review and meta-analysis in the Journal of the Canadian Association of Gastroenterology pooled six randomized trials covering 478 patients and found that, compared with standard care, GLP-1 receptor agonists were associated with markedly higher odds of NASH resolution on biopsy (odds ratio 4.45, 95% CI 1.92–10.3) and a roughly five-percentage-point reduction in liver steatosis on imaging. Liver stiffness — a proxy for fibrosis — barely moved.

The authors are appropriately measured: GLP-1 therapy probably improves NASH over at least six months. The trials are few, the patient numbers modest, and stiffness, the outcome most tied to long-term liver damage, did not meaningfully change. The takeaway is encouraging without being conclusive.

Most of the cardiovascular headlines around semaglutide have come from the injectable form. But adherence with weekly injections is imperfect, and many patients would prefer a tablet. A 2025 systematic review in the Journal of Clinical Medicine looked specifically at oral semaglutide's effects on cardiovascular risk factors — blood pressure and lipid profile — in people with type 2 diabetes.

The signal, taken across the included studies, points in a familiar direction: oral semaglutide moves cardiovascular risk markers favorably when added to standard care. That matters because, until recently, it was an open question whether the oral formulation would carry the same metabolic halo as the injection. The review does not deliver a definitive head-to-head verdict, and it is built on heterogeneous studies, but it strengthens the case that the route of delivery is not the whole story.

People living with both type 2 diabetes and chronic kidney disease occupy one of medicine's hardest corners. Their cardiovascular risk is high, their medication lists are long, and historically their options for slowing renal decline have been narrow.

A 2025 meta-analysis in Annals of Medicine and Surgery pooled three randomized trials covering 10,013 such patients. Compared with placebo or standard care, semaglutide was associated with a 29% relative reduction in cardiovascular mortality (RR 0.71, 95% CI 0.52–0.97), a 20% reduction in major adverse cardiovascular events, and a 20% drop in kidney-related adverse events. The authors also noted a reduced need for additional cardiovascular medications.

These are relative reductions in a high-risk group, which tends to magnify the absolute benefit a single patient might see — but it also means the results may not generalize neatly to lower-risk readers. And with only three trials in the pool, even an encouraging signal is one or two negative studies away from looking different. Still, for the specific population of type 2 diabetes with CKD, this is among the more coherent stories the GLP-1 class has produced.

The most uncomfortable conversation in this space is not clinical but financial. Anti-obesity medications are expensive, often not reimbursed, and used by a population large enough to shift national drug budgets. The countervailing argument has always been that obesity itself is costly — in medical care, in lost productivity, in years of life lived in poorer health.

A 2025 modeling study in Scientific Reports took a societal-perspective look at this trade-off in Austria. The authors modeled treating 50% of adults with class II and class III obesity (excluding people with diabetes) for 68 weeks with semaglutide alongside lifestyle intervention. In the model, class II obesity prevalence fell from 4% to 2.74% and class III from 1.45% to 0.97%, producing a 12.9% reduction (about €108.7 million per year) in expenses tied to those classes; over the life cycle, dropping an obesity class reduced costs by roughly 40% per patient.

Modeling studies are not real-world evidence. They depend on assumed adherence, assumed durability of weight loss after treatment, and assumed health-system structures. The Austrian figures will not translate cleanly to other countries. But the directional point — that the population-level math can favor treatment in higher-risk groups — is worth taking seriously.

The 2025 evidence does not vindicate the more breathless claims made about GLP-1 drugs, but it does push the conversation beyond the bathroom scale. The liver data are promising and incomplete. The oral cardiovascular data are reassuring and indirect. The kidney data are the strongest of the four, but limited to a specific high-risk group. And the economic data are a model, not a measurement.

For readers already on a GLP-1, or considering one, the most useful posture is neither evangelism nor suspicion. It is curiosity, paired with a clinician who knows your numbers. The questions worth asking are specific: what does my liver look like; what is my kidney function; what is my cardiovascular risk; what would a year, three years, or longer on this class plausibly do for each of them — and what would stopping look like?

The class is doing more than we thought it would. It is also being asked to do more than the evidence yet supports. Holding both ideas at once is, for now, the honest read.

The most consequential data point of the year sits inside a meta-analysis published in the Journal of Diabetes that pools 29 randomized controlled trials, nine GLP-1RA-based drugs, and 37,348 non-diabetic adults with overweight or obesity. Compared with placebo, the pooled therapies were associated with a 19% relative reduction in total cardiovascular events and a 20% reduction in major adverse cardiovascular events, alongside a 28% drop in myocardial infarction and a 19% drop in all-cause mortality. Cardiovascular death and stroke, notably, did not reach statistical significance — a reminder that the headline is a signal, not a sweep.

For a community used to parsing forest plots, the texture matters as much as the totals. The analysis singles out orforglipron for systolic blood pressure reduction (mean difference −7.10 mmHg) and tirzepatide for the largest cardiometabolic shifts, hinting that the class is not monolithic. Different molecules pull different levers — weight, pressure, lipids, glycemia — and the cardiovascular benefit appears to ride on the composite, not on any single biomarker.

The trials feeding this meta-analysis were designed primarily around weight and metabolic endpoints, not hard cardiovascular outcomes. That makes the pooled effect estimates suggestive rather than definitive: the confidence intervals are tight, but the events are comparatively few, and the absence of a significant stroke or cardiovascular-death effect is consistent with a class whose benefit may be largely mediated by weight loss, blood pressure, and inflammation rather than a direct vascular mechanism. The authors themselves frame the findings as evidence that cardioprotection in non-diabetic obesity is plausible and worth dedicated outcome trials, not as a finished case.

For the n-of-1 crowd tracking apoB, resting heart rate, and overnight glucose on a continuous monitor, the practical read is narrower than the press releases imply. The peptides appear to move several knobs at once in the right direction; whether that translates into your personal event risk depends on baseline cardiometabolic load that no wearable fully captures.

A second 2025 review, in Current Opinion in Obstetrics & Gynecology, tracks the class into territory diabetes trialists rarely touch. In polycystic ovary syndrome — a condition tangled with insulin resistance and metabolic dysfunction — the authors report that GLP-1 receptor agonists may improve weight loss, metabolic markers, and menstrual regularity, with early signals that combined GLP-1RA and metformin therapy before IVF is associated with higher spontaneous conception and pregnancy rates. The evidence base is preliminary, but the mechanistic logic — restore insulin sensitivity, restore ovulation — is familiar.

The same review flags two practical issues that anyone on a peptide should know about. First, delayed gastric emptying — a core mechanism of GLP-1RA satiety — has implications for perioperative management and for the efficacy of oral contraception in patients taking tirzepatide. Second, research into endometrial hyperplasia and infertility is active but not yet conclusive. None of this is a contraindication; all of it is a conversation to have with a clinician before optimizing a stack around these drugs.

Further upstream, the receptor logic is being remixed. A preclinical paper in International Immunopharmacology reports that a novel dual CCK/GLP-1 receptor agonist improved cognitive deficits, reduced amyloid-beta accumulation, and alleviated mitochondrial damage in 5×FAD mice by inducing PINK1/Parkin-mediated mitophagy, outperforming the GLP-1 analogue liraglutide on several measures. That is a mouse model of Alzheimer's disease, not a human trial, and the gap between rodent mitophagy and clinical cognition is famously wide. But it illustrates where the chemistry is heading: combinatorial peptides aimed at organ systems well beyond the pancreas.

The honest read on all three papers, taken together, is that GLP-1 receptor agonists are looking less like a diabetes drug that happens to cause weight loss and more like a cardiometabolic platform with credible — if uneven — extensions into reproductive endocrinology and speculative ones into neurodegeneration. The evidence is moderate, the enthusiasm is high, and the gap between the two is where careful readers live.

The intellectual engine behind this shift is a technique called drug-target Mendelian randomization. The logic is elegant: because we inherit our genes at random, people who happen to carry variants that nudge a specific protein up or down spend a lifetime in a kind of natural experiment. If those nudges track with lower rates of heart disease, the protein becomes a credible drug target — one with decades of human safety data already written into population biology. PCSK9 inhibitors were the proof of concept. The same playbook is now pointing at two new addresses on the lipoprotein map.

Lipoprotein(a), or Lp(a), is an LDL-like particle with an extra protein appendage that makes it unusually atherogenic and unusually stubborn. Levels are roughly 80 to 90 percent genetically determined, largely fixed from birth, and largely unmoved by diet, exercise, or statins. For years, clinicians could measure Lp(a), warn patients about it, and then do essentially nothing. A 2025 review in Current Opinion in Lipidology describes how genetic evidence has put Lp(a) inhibitors — a class of RNA-targeting therapies designed to switch off hepatic Lp(a) production — squarely in late-stage development, with the pharmacologic lipid effects closely mirroring what naturally occurring genetic variants predicted.

That recapitulation matters. When a drug's effect on a biomarker tracks the genetic signal, it raises the prior that the downstream cardiovascular benefit will also follow. It is not proof. Phase 3 outcomes trials remain the arbiter, and at the time of writing those readouts are still pending. But the design of the bet is sounder than any cardiology has placed before.

The second target sits further down the lipid cascade. Lipoprotein lipase is the enzyme that clears triglyceride-rich lipoproteins from circulation, and its activity is throttled by a set of inhibitor proteins — ANGPTL3, ANGPTL4, and apoC-III among them. People who inherit loss-of-function variants in those inhibitors run lower triglycerides, lower remnant cholesterol, and, importantly, lower rates of coronary disease over a lifetime. That genetic pattern has motivated a wave of lipoprotein-lipase pathway activators, including antisense and siRNA therapies designed to dial down the brakes on the enzyme.

The same review notes that beyond their lipid effects, robust genetic evidence supports a broader cardiometabolic benefit from this class — a reasonable expectation given that elevated triglyceride-rich remnants are implicated in pancreatitis, fatty liver, and metabolic syndrome as well as atherosclerosis. Again: genetic evidence is suggestive, not conclusive. The outcomes data will need to land.

This is where the language has to slow down. The genetic case for both targets is strong, and the early pharmacology is encouraging. What is not yet in hand is the thing that ultimately matters to patients: large, randomized cardiovascular outcomes trials demonstrating that lowering Lp(a) or activating lipoprotein lipase reduces heart attacks and strokes in real people over real time. Until those trials report, the appropriate posture is interested, not converted.

It is also worth saying what these therapies are not. They are not replacements for LDL lowering, which remains the foundation of prevention with the deepest evidence base. They are candidates for layering on top of LDL therapy in patients whose residual risk is driven by something else — a high Lp(a) inherited from a parent, say, or persistently elevated remnants despite a normal LDL. The promise, as the reviewers frame it, is a more tailored preventive medicine, where treatment maps to a patient's specific lipoprotein-lipid profile rather than a single number on a lab report.

The history of cardiovascular medicine is littered with biomarkers that looked promising in observational data and disappointed in trials — HDL-raising therapies being the most painful recent example. What is different this time is the discipline of the underlying method: the targets here were not chosen because they correlated with disease in a cohort, but because nature has been running the randomized trial in human populations for millennia. That does not guarantee success. It does mean the bets are better designed than they have ever been, and the next few years of outcomes data will tell us whether the era of tailored lipid lowering has truly arrived.