A viral Bryan Johnson post framed cancer and Alzheimer's as a 'tug of war.' The peer reviewed evidence is real; the molecular mechanism is stranger and still out of therapeutic reach.
A growing body of cohort research shows that people who develop cancer are significantly less likely to also develop Alzheimer's, and the reverse holds too. The numbers look like one disease shielding against the other. The mechanism is stranger: the same molecular switches that let tumor cells escape growth control appear, in a different configuration, to hold off the neurodegeneration that defines Alzheimer's. (npj Aging, 2026)
The framing went viral this week when Bryan Johnson, the entrepreneur behind the "Don't Die" longevity protocol, posted on X that "If you get cancer, your risk of Alzheimer's drops by about a third. If you get Alzheimer's, your risk of cancer drops by about half," and called the body "playing tug of war with itself." The figures are loosely drawn from real epidemiology, and the tension he flagged is a recognized scientific paradox rather than a fringe claim. What he left out is that "tug of war" is a metaphor covering several distinct molecular stories, each on shakier ground than the cohort numbers suggest.
The cleanest epidemiological anchor is the Framingham Heart Study follow-up published in 2012 by Driver and colleagues. Tracking participants over decades, the team found that incident cancer was associated with a statistically significant reduction in the risk of subsequently developing Alzheimer's disease, and that the reciprocal direction also held. (Driver et al., 2012) The inverse pattern has since been reproduced across multiple population datasets and consolidated in short reviews that treat it as a recurring feature of aging cohorts rather than a one-study artifact. (Inverse Correlation Between Alzheimer's Disease and Cancer)
Both directions of the effect, however, are vulnerable to well-known methodological distortions. Patients who die of cancer cannot go on to develop dementia, a textbook competing-risk problem that inflates the apparent protective effect. Cancer treatment itself, including chemotherapy and surveillance imaging, can impair cognition in ways that complicate any dementia diagnosis. Conversely, an Alzheimer's diagnosis usually means fewer cancer screenings are performed, mechanically reducing cancer incidence in the dementia group. The hazard ratios that produce "about a third" and "about half" can move substantially once these biases are modeled, and Johnson's rounded numbers should be read as a rough direction across broad populations rather than as magnitudes that hold cancer-by-cancer and dementia subtype by dementia subtype. (Molecular crosstalk review)
The mechanistic literature is where the paradox becomes genuinely interesting. Two cellular hubs sit at the center of the trade-off. p53, the tumor-suppressor protein often called the "guardian of the genome," triggers cell death or senescence when DNA damage accumulates; in cancer, mutations that disable p53 let damaged cells keep dividing. In neurons, the same protein's failure to clear damaged cells has been linked to the build-up of pathology that drives Alzheimer's. Glycogen synthase kinase 3 (GSK3), a signaling enzyme, plays an opposite kind of dual role: it promotes both tau phosphorylation in Alzheimer's and proliferation in several cancers. Early reviews proposed GSK3 and p53 as candidate shared hubs whose dysregulation can push cellular fate in opposite directions. (GSK3 and p53 in Alzheimer's disease)
A more specific candidate has emerged in the past decade: PIN1, an enzyme that helps fold other proteins into their working shapes by acting on phosphorylated serine-threonine bonds. PIN1 is required for healthy cell division, which is exactly why many cancers overexpress it. In neurons, low PIN1 activity is associated with tau and amyloid pathology. The same protein, dysregulated in opposite directions, sits at the intersection of both diseases. (Common molecular and cellular pathway review)
Direct cellular evidence has accumulated. A 2018 study of patients who later developed Alzheimer's showed that a prior cancer history left an imprint on their lymphocytes, with elevated PARP-1, an enzyme that repairs DNA damage, and a stronger p53-dependent resistance to oxidative stress. (Frontiers in Neuroscience, 2018) The mechanistic literature has grown enough that the npj Aging framing of "the cancer-Alzheimer's disease paradox" treats the inverse risk as a defined research question deserving its own mechanistic program rather than a statistical curiosity.
None of this has translated into clinical tools. No drug has been approved to exploit the trade-off, and trials of cancer-relevant kinase inhibitors in neurodegeneration have struggled to show benefit at tolerable doses. The honest read of the peer-reviewed literature is that the inverse association is well established at the population level, biologically plausible through several candidate mechanisms, and almost entirely out of therapeutic reach. Johnson's post is useful as a hook; the underlying science is the more interesting story, and it is one the field itself is still writing.