The weight-loss drugs keep doing things they weren't designed to do

The drugs people know as Ozempic and Wegovy were built to manage diabetes and weight. This week two large studies tied them to two more things: less cancer, and less addiction. Both findings are association, not proof — but they keep arriving, and they framed a busy week at the world’s largest cancer conference.

The GLP-1 drugs keep widening

GLP-1 drugs are a class of medicines — semaglutide is the best-known — that mimic a gut hormone to curb appetite and steady blood sugar. They were approved for diabetes, then obesity. The news now is what else they seem to touch.

At ASCO, the American Society of Clinical Oncology’s annual meeting — the field’s biggest gathering — researchers presented several studies linking GLP-1 use to lower cancer risk. One analysis found people on the drugs were about 30% less likely to develop breast cancer, the world’s most common cancer, than people not taking them [3][6]. Other work suggested a benefit across several cancer types, and a possible role added to standard breast-cancer treatment [3][6].

Hold the excitement at the right level. These are observational studies — they compare people who happened to take the drugs against people who didn’t [3]. They can show a link; they cannot prove the drug caused it. People on GLP-1 drugs lose weight, and obesity itself raises the risk of many cancers, so some of the effect may simply be the weight loss, not the molecule. The findings point somewhere worth testing properly. They are not a result from a trial built to answer the question.

The same week, a study of 606,434 US military veterans with type 2 diabetes, published in The BMJ, tied GLP-1 drugs to less addiction [2]. Among veterans with no substance-use disorder at the start, those on the drugs had a 14% lower risk of developing one — and lower risk across alcohol, cannabis, cocaine, nicotine and opioids [2]. Among the 81,617 who already had an addiction, the drugs were associated with 40% fewer overdoses and 50% fewer drug-related deaths [2].

Those are large numbers. The same caveat governs them: this is electronic-health-record data, an association, not a controlled trial [2]. Veterans with diabetes are not the general population. And the study reports that the drugs “were associated with” these outcomes — the careful phrasing that means we saw a link, we did not prove a cause [2]. The pattern is striking enough that proper trials are the obvious next step. It is not, yet, a reason the drugs do what these headlines imply.

Cancer’s immune treatments post real wins — with limits attached

Two cancer results this week were stronger than association, because they came from trials where doctors treated patients and watched what happened.

In a UK trial, all 32 patients with stage two or three bowel cancer saw no return of their disease three years later [34]. They shared a specific genetic profile — a faulty DNA-repair system, found in about 10–15% of bowel-cancer cases — that makes tumours easier for the immune system to spot [34]. The patients got the immunotherapy drug pembrolizumab before surgery; it removed the need for chemotherapy [34]. Immunotherapy means a drug that takes the brakes off the body’s own immune cells so they attack the tumour.

The caveat is in the size and the selection. Thirty-two patients is a small trial, and only patients with that one genetic profile were eligible [34]. A 100% result in 32 carefully chosen people is a strong signal, not a settled standard of care. It needs larger trials before it changes how most bowel cancer is treated.

A second result drew the word “unprecedented” from the doctors running it [39]. In an international trial across 11 countries, an injected drug called amivantamab was given to patients whose cancer had spread or returned and had stopped responding to chemotherapy and immunotherapy [39]. It shrank tumours in more than a third of them; in 15 patients the tumours disappeared entirely [39]. For people who had run out of options, that matters.

Read it precisely. “More than a third responded” is the headline number; the complete disappearances were 15 patients, not the whole group [39]. These were people with no remaining standard treatment, which is exactly where a new drug gets its first chance — and exactly why the result, while real, is early.

The week also carried a reminder that efficacy is not the only number. Abivax reported strong results for its drug in a phase 3 trial — the large, final test before approval — for ulcerative colitis, an inflammatory bowel disease [5]. The company’s share price fell anyway, because cancer cases turned up among trial participants [5]. A drug’s safety signal can overshadow its efficacy, and investors price both. AstraZeneca, meanwhile, missed the main goal of a phase 3 trial and is now pitching a smaller patient subgroup to regulators for a rare-disease approval [7]. A “miss” followed by a subgroup pitch is common; it is also how a drug ends up approved on weaker evidence than a clean win would have given.

A near-cure for a virus that infects 250 million people

GSK reported that its hepatitis B drug produced a “functional cure” in about one-fifth of patients in a trial — the company called it a major step [11]. Hepatitis B is a chronic liver infection carried by an estimated 250 million people worldwide; it can lead to liver failure and cancer, and current treatments usually suppress the virus rather than clear it [11].

“Functional cure” is a specific, careful term. It means the virus is controlled to undetectable levels without ongoing treatment — not that every trace is gone forever [11]. One-fifth is a minority of patients, and the durability question — does the control hold for years — is what later, larger trials have to answer. For a disease where the bar has been suppression, not clearance, a one-in-five functional cure is a genuine move. It is a step, as GSK said, not an arrival.

Longevity: the money is loud, the evidence is quiet

Anti-ageing biotech keeps raising large sums. NewLimit, the cellular-rejuvenation company backed by Coinbase founder Brian Armstrong, pulled in $435 million to work on resetting old cells to a younger state [25]. The science it chases — partial cellular reprogramming — is real and active. The honest status is preclinical: this is lab and animal work, not a treatment people can take [25].

That gap between funding and proof is exactly the subject of a new book reviewed in Nature this week. The longevity researcher Saul Newman argues that claims about the limits of human lifespan are riddled with hype, thin data and weak science [18]. His earlier work showed that many of the world’s reported “supercentenarians” — people said to be 110 or older — sit in places with poor birth records and high pension fraud, not unusual longevity [18]. The review’s blunt title: we simply don’t know how long humans can live [18].

The lab evidence underlines the caution. A careful study in Nature Aging this week tested time-restricted feeding — eating only within a set daily window — in 528 mice [27]. It improved several health measures in both sexes and extended lifespan in males [27]. Two things to hold: the lifespan gain was in male mice only, and it was in mice [27]. Most things that work in mice never work in people. A diet result in rodents is a lead for human research, not a human result.

The quieter shift: the FDA tries to speed gene therapies

The story most likely to be missed this week came from the regulator. The FDA, the US drug agency, issued draft guidance meant to speed up gene therapies for rare and life-threatening diseases [8]. The idea is to let developers reuse existing safety and manufacturing knowledge — from related, already-studied therapies — instead of repeating tests from scratch for each new one [8].

This is plumbing, not a headline cure, which is why it travels quietly. It also matters more than most single trials. Gene therapies are slow and expensive to develop, and many rare diseases have too few patients to ever justify the full, redundant testing path [8]. A review in Pediatric Research this week noted the FDA had approved 44 cell and gene therapies as of May 2025, 14 of them for children [1]. Rules that lower the cost of the next 44 shape which diseases get a therapy at all. It is draft guidance — not final, and the details will be argued over. The direction is the news.