For decades, colorectal cancer looked like a disease of getting old. Around 90 percent of cases show up after age 50. That story has been fraying. Registries around the world keep reporting more tumors in people in their thirties and forties, and the reasons have stayed stubbornly vague. Something in modern life is doing this. The trick is naming it.
A team led by Silvana C. E. Maas and Jose A. Seoane at the Vall d'Hebron Institute of Oncology in Barcelona tried a sideways approach, published in Nature Medicine. Instead of asking patients what they ate or breathed or sprayed, which people forget or never knew, they read the record left inside the tumors themselves. Their standout suspect is a herbicide most people have never heard of: picloram.
Reading exposure off the genome
The body keeps a chemical diary. Environmental exposures nudge patterns of DNA methylation, small tags that sit on the genome and shift how genes behave. Those tags linger. The researchers turned published studies linking specific exposures to specific methylation marks into what they call methylation risk scores. Each score is a proxy. A high one suggests a person carried more of that exposure, even when no one measured it directly.
They built 63 such scores covering 29 lifestyle and environmental factors, including 14 pesticides, then compared tumors from patients under 50 against tumors from patients 70 and older. The discovery set was small, 31 early-onset cases versus 100 late-onset, drawn from The Cancer Genome Atlas. A meta-analysis across nine more cohorts added 83 early cases and 272 late ones.
The scores recovered things researchers already believed. Younger patients carried marks pointing to lower Mediterranean diet adherence, less education, and more smoking exposure. That the method reproduced known signals gave the authors some confidence it could spot new ones. Picloram was the surprise. Its score ran higher in early-onset tumors, with an odds ratio of 1.56 in the combined analysis. Glyphosate, already labeled "probably carcinogenic" by the World Health Organization's cancer agency, and atrazine also showed up.
A second look at the map
Methylation scores are clever but indirect, so the team went looking for corroboration in the ground truth of where pesticides actually get used. They matched county-level herbicide application data to cancer incidence across 94 US counties spanning seven states over 21 years. Of 225 pesticides with enough data, 62 tracked with early-onset colorectal rates. After adjusting for income, poverty, education, and other pesticides, 27 held up. Picloram was the most durable of the lot, surviving nearly every adjustment.
Picloram works by mimicking auxins, the hormones that drive plant growth, pushing tissues into abnormal, unregulated development until the plant dies. It was registered in the US in 1964 and shows up as residue in some grain and meat byproducts. Its long-term effects on human health have basically never been described. The authors float a plausible timing argument. Older patients were children before this chemistry saturated agriculture. Younger ones grew up steeped in it.
There is a further twist. Age at diagnosis is a blunt stand-in for how old a tumor really is, since cancers can smolder for years before detection. The team used a mutational signature called SBS1, which accumulates with cell division, as a molecular clock. Tumors that looked biologically "young" by that clock also carried the picloram signal, hinting the association is about tumor biology, not just birth year.
What this does not prove
Correlation is doing a lot of work here, and the authors say so. No one has directly measured how picloram changes human DNA methylation, so the picloram score was validated against gene-expression changes rather than real exposure data. The patient numbers are modest, especially once colon and rectal tumors are split apart. County pesticide records only start in 1992, which leaves a blind spot around childhood exposure and makes any latency period impossible to pin down. This is a signal worth chasing, not a verdict on a chemical. Establishing cause would take long-term birth cohorts and animal work the authors openly call for.
Still, the framing is what makes it interesting. Rather than hunting for a single guilty molecule, the study treats the tumor as an archive of a lifetime's exposures and reads it back out. If that approach holds up, the rise of cancer in the young might turn out to be legible after all, written in a code we are only starting to decode.
Comments