HN Debrief

Semaglutide linked to lower predicted dementia risk

  • Public Health
  • Biotech
  • Medicine
  • Obesity

The paper is a post hoc analysis of the SELECT trial. It looked at blood proteins that feed a dementia risk score and found that semaglutide slowed worsening of that score over two years in older adults with overweight or obesity and cardiovascular disease who did not have diabetes. It did not measure whether people actually developed dementia less often or whether memory and cognition improved. That limitation drove the reaction. People were not especially shocked that a Novo Nordisk paper made semaglutide look good. The stronger point was that this is exactly the kind of result that can be technically true and still clinically thin. Several comments stressed that the headline claim rests on a proteomic risk signature, not patient outcomes, and that the study design gives the company room to imply more than it proved.

Do not treat this as evidence that semaglutide prevents dementia. If GLP-1s are already relevant to your health or business, watch for outcome-based trials and real-world incidence data, because biomarker wins are not the same thing as cognitive benefit.

  • alz-journals.onlinelibrary.wiley.com
  • Discuss on HN

Discussion mood

Interested but skeptical. Most readers think GLP-1s are doing something real and broadly beneficial for metabolic health, but they disliked how a company-funded biomarker analysis can be read as if it showed protection against actual dementia.

Key insights

  1. 01

    Weight loss can bias dementia comparisons

    In older adults, losing weight without trying can itself be an early sign of dementia. That means comparing semaglutide users to placebo while adjusting on BMI change can make the drug look better than it is, because weight loss in the placebo group may partly identify people already headed toward decline rather than people getting healthier. The same comment also points out that near-term dementia risk scores are heavily driven by inflammation and metabolic markers, which are exactly the things weight loss changes fastest.

    Be careful with any analysis that treats weight change as a clean mediator in aging populations. If you are evaluating GLP-1 outcome claims, look for studies that measure actual dementia incidence and cognition, not models built on biomarker shifts.

      Attribution:
    • jtrn #1
  2. 02

    BMI adjustment does not settle mechanism

    The paper’s sensitivity analysis reported that most of the 20-year biomarker effect remained after adjusting for BMI change, which sounds like evidence for a direct drug effect. The catch is that BMI is a crude proxy. It does not capture where fat was lost, how diet changed, whether inflammation dropped, or whether people became more active, so surviving a BMI adjustment is suggestive but nowhere near proof that semaglutide has a brain-specific mechanism.

    Do not overread residual effect after BMI adjustment as mechanistic evidence. If mechanism matters to your work, wait for studies that track visceral fat, inflammatory markers, activity, and cognition together.

      Attribution:
    • Muskwalker #1
    • bitexploder #1
  3. 03

    Dose and tolerability shape real outcomes

    Several people using GLP-1s described a pattern that matters more than the abstract trial result. Maximum or standard escalation doses can produce low energy, nausea, bloating, and a generally miserable calorie deficit, while lower or slower dosing often preserves much of the benefit with fewer side effects. That is a useful reminder that the commercial story around these drugs often tracks best-case efficacy, while real long-term adherence depends on finding a maintenance dose people can live with.

    If you are thinking about GLP-1 deployment in care programs or employee benefits, plan for individualized titration rather than assuming everyone should march to the labeled maximum dose. Persistence and function may matter more than peak weight-loss numbers.

      Attribution:
    • jaggederest #1 #2
    • cj #1
    • 01100011 #1
    • turbocon #1
    • mchusma #1
  4. 04

    Users report benefits before major weight loss

    A recurring firsthand claim was that blood work, liver markers, pain, and mobility can improve even when body weight barely moves. One account described dramatic lab improvement on a tiny dose with little weight loss. Another described major gains in mobility despite minimal scale change. None of that proves the dementia claim, but it does support the broader idea that GLP-1 effects are not reducible to pounds lost alone.

    When assessing GLP-1 value, track metabolic and functional measures beyond weight. If you only watch the scale, you will miss part of what users and clinicians think they are buying.

      Attribution:
    • jaggederest #1
    • alyandon #1
    • wincy #1
  5. 05

    Older GLP-1 data are hard to generalize

    People noting that GLP-1s have been around for decades also acknowledged the catch. Most of the long history comes from diabetes populations, where the disease itself and related dietary changes muddy attempts to infer effects on dementia risk in everyone else. This paper was notable mainly because it used a non-diabetic cohort, even though it still stopped short of clinical dementia outcomes.

    Do not assume a long drug history automatically answers adjacent questions in a new population. For strategy or investing, separate evidence in diabetics from evidence in obese non-diabetics.

      Attribution:
    • akiselev #1
    • toast0 #1
    • Muskwalker #1
  6. 06

    Vision loss risk looks rare and conditional

    One useful safety note cut through the alarm. The reported semaglutide link to non-arteritic anterior ischemic optic neuropathy is a very rare risk in adults with type 2 diabetes, and commenters noted that the condition already clusters with the same metabolic problems semaglutide treats. One commenter also said anatomical predisposition such as a crowded optic disc appears to matter, which makes the headline risk less universal than it first sounds.

    If you are weighing GLP-1 adoption, do not ignore safety headlines but do inspect base rates and risk concentration. Rare adverse events matter most when they change screening or patient selection, not when they are quoted as context-free relative risk.

      Attribution:
    • mapontosevenths #1
    • adam_arthur #1
    • tptacek #1

Against the grain

  1. 01

    Corporate authorship was not hidden

    Some pushed back on the outrage over Novo Nordisk funding. The paper openly identifies company involvement, and one comment noted that this was not an arms-length grant to outside academics but a study done by company employees as part of their work. The useful point is not that skepticism is unnecessary. It is that readers should distinguish undisclosed conflict from plainly disclosed corporate research and then judge the methods on top of that.

    Treat disclosed company research as biased but legible, not as a scandal by default. Put your effort into interrogating endpoints and study design rather than acting as if the funding itself was a surprise.

      Attribution:
    • cowsandmilk #1
    • tokai #1
  2. 02

    There may still be a real effect

    A few comments argued that dismissing the paper entirely goes too far because semaglutide already has observational correlations with lower dementia rates and a biologically plausible path through metabolic improvement. They framed this paper as one weak piece in a larger pile rather than a standalone proof. That does not rescue the headline, but it does argue against writing off the whole line of inquiry just because this one study used biomarkers.

    Keep two ideas separate. This study is weak evidence for dementia prevention, and the broader hypothesis that GLP-1s could reduce dementia risk remains worth following.

      Attribution:
    • cowsandmilk #1
    • danielmarkbruce #1

In plain english

biomarker ↩
A measurable biological indicator, such as blood glucose or triglycerides, used to track health or disease risk.
BMI ↩
Broadcast Music, Inc., a U.S. performance rights organization.
GLP-1 ↩
Glucagon-like peptide-1, a class of drugs used for diabetes and weight loss.
post hoc analysis ↩
An analysis done after a study is completed using data that were not the main original endpoint of the trial.
SELECT trial ↩
A large randomized clinical trial that studied semaglutide in people with overweight or obesity and cardiovascular disease.
semaglutide ↩
A drug used for diabetes and weight loss that helps control blood sugar and appetite.
visceral fat ↩
Fat stored around internal organs, which is more strongly linked to metabolic disease than fat stored under the skin.

Reference links

Study and safety references

Metabolism and exercise references

Diet and sugar references

Weight loss physiology references