A new study of adults receiving obesity treatment found substantial metabolic improvements among those who lost more weight, alongside an encouraging brain-scan signal. But the October 30 paper in Obesity Science & Practice does not show that weight loss prevented Alzheimer’s disease or improved cognition. Cognitive performance did not significantly change over one year, and the main between-group brain-volume signal did not survive correction for multiple statistical tests.
That distinction is the most useful part of this preliminary report. Changes in a scan, changes on cognitive tests and fewer future dementia diagnoses are different outcomes. The study gives researchers a reason to investigate further, while leaving the question readers care about—whether treatment protects memory over time—unanswered.
The 40% threshold was not total body-weight loss
The New York University cohort included 84 adults with baseline and follow-up data. Eligibility required ages 33–59, a BMI between 30 and 50, and type 2 diabetes, prediabetes or insulin resistance. Researchers classified 54 participants as having lost at least 40% of their excess weight and 30 as having lost less. These groups were defined after treatment, rather than randomly assigned.
“Excess weight” meant weight above the amount corresponding to a BMI of 25. Losing 40% of that excess is not losing 40% of starting body weight. Table 1 reports average total body-weight losses of 23.62% in the greater-loss group and 3.38% in the lower-loss group. The cutoff was a study grouping choice, not a target shown to protect the brain or a recommendation for an individual.
Participants received bariatric surgery or medical and behavioral treatment; most in the greater-loss group underwent surgery. Some used GLP-1 medicines alongside their care. This was not a randomized comparison of surgery against medication, and it cannot identify the effect of a particular drug. Our guide to randomized and observational evidence explains why grouping people by their eventual response leaves room for other differences to influence outcomes.
Metabolic findings were clearer than the brain findings
In the paper’s longitudinal analyses, the greater-loss group had larger improvements in several measures, including HbA1c, fasting glucose, insulin-resistance measures and C-reactive protein, an inflammation marker. Several metabolic comparisons remained statistically significant after correction for multiple testing. Blood-pressure changes did not clearly differ between groups.
The imaging evidence was less robust. The change in subcortical gray-matter volume differed between groups with an uncorrected P value of 0.016. Researchers accounted for age and the interval between scans, and an additional MRI tissue-measurement adjustment did not remove the signal. However, the authors explicitly report that it did not survive false-discovery-rate correction.
When researchers test many outcomes, some can appear positive by chance. A correction helps account for that problem. Losing statistical significance after correction does not prove the signal is false, but it makes a confident claim of brain protection premature. The clinical-trial reading guide offers a broader framework for checking which outcomes support a headline.
“Preserved” regions did not establish a treatment advantage
The authors also examined a combined measure of regions vulnerable in Alzheimer’s disease. Within the lower-loss group, that measure declined; within the greater-loss group, it did not significantly change. Those separate observations can sound like evidence of preservation.
Yet the direct comparison of changes between groups was not statistically significant. A significant change in one group and a nonsignificant change in another do not, by themselves, establish a significant difference between them. The paper’s discussion acknowledges that these within-group observations are preliminary.
Cognitive outcomes provide another check on the interpretation: no significant improvement was detected over the year. A longer study might find a benefit, or it might not. The present results cannot settle that question, and an MRI volume measurement is not a diagnosis of Alzheimer’s disease or a count of dementia cases prevented.
An older randomized study shows why the distinction matters
A separate 2016 Look AHEAD brain-imaging report studied 319 volunteers with type 2 diabetes and overweight or obesity, scanned 10–12 years after random assignment to intensive lifestyle intervention or diabetes support and education. Total brain volume, its primary imaging outcome, did not differ significantly. Some other imaging measures favored the intervention, but cognitive tests showed no consistent benefit.
This older study provides context, not a replication of the NYU cohort. It involved a different population, intervention and duration, and lacked baseline MRI scans. Its diet-and-activity program also cannot isolate weight loss from the rest of the intervention. Together, the reports illustrate why promising imaging findings need separate evidence linking them to better thinking or memory.
What would make the new evidence more convincing?
The NYU analysis excluded 45 people without the required repeat evaluations or imaging data, and two further participants were excluded from structural analyses because of scan artifacts. Missing follow-up can matter even when measured characteristics appear similar. The sample was predominantly female, and the nonrandomized, mixed-treatment design limits generalization and causal claims. Longer follow-up, larger samples and clearly specified outcomes would help test whether the imaging signal persists and predicts meaningful cognitive change.
For someone discussing obesity care, a useful question is which benefits are established for their condition and which remain research possibilities. This paper supplies no basis for choosing surgery over a medicine, changing a prescription or seeking MRI monitoring of weight loss. Our guides to BMI and health and setting useful health goals explain why progress should be interpreted across relevant health measures, rather than through a promise of future brain protection.



