Background: Linking early-onset colorectal cancer and environmental exposure

A study published in Nature Medicine found that a DNA methylation score associated with the herbicide picloram was higher in colorectal cancer patients younger than 50 than in patients aged 70 or older. U.S. regional data also showed an association between picloram use and early-onset colorectal cancer incidence. The findings do not establish that picloram causes colorectal cancer.

As colorectal cancer incidence rises among people under 50, research examining diet, lifestyle, and environmental exposures together is attracting attention. The totality of these exposures is called the exposome. A challenge is that existing cancer cohorts lack sufficient quantitative records of individuals’ past exposures. The researchers used exposure-associated DNA methylation patterns as proxy indicators to analyze differences between early-onset and older patients.

Key findings: Tumor exposure proxies and regional incidence analysis

The main analysis used DNA methylation data from tumor tissue. Using markers reported in previous epigenome-wide association studies, the researchers constructed 63 weighted methylation risk scores (MRSs) corresponding to 29 lifestyle and environmental factors. DNA methylation is an epigenetic modification in which methyl groups attach to specific sites on DNA. Although “risk score” appears in its name, the MRS was used here to estimate exposure indirectly, not to predict an individual’s risk of developing cancer.

In the discovery stage, the researchers compared 31 patients younger than 50 with 100 patients aged 70 or older in colon cancer data from The Cancer Genome Atlas (TCGA). They then conducted a meta-analysis of 83 early-onset and 272 older patients from nine independent colorectal cancer cohorts. Because the study compared two age groups who had already been diagnosed with colorectal cancer, it should be distinguished from research that directly assesses future cancer risk in healthy people.

The picloram-associated MRS was associated with early-onset status in both the discovery and replication analyses. In the replication meta-analysis, the odds ratio was 1.56, with a 95% confidence interval of 1.16–2.09 and a multiple-testing-adjusted P value of 0.0151. This reflects an association between a standardized score and membership in the two patient groups. It cannot be interpreted to mean that actual picloram exposure increased cancer risk by 56%.

Separately, an analysis of data from 94 U.S. counties covering 1992–2012 found an association between the intensity of picloram use and colorectal cancer incidence among people aged 25–49. The association generally persisted in several analyses that accounted for the use of other pesticides and socioeconomic factors, but regional use levels do not indicate the actual exposure of individual residents.

Meaning and outlook: Further testing of causality and actual exposure

Consistent associations across different data sources provide grounds to consider picloram a priority for follow-up research. However, the study did not directly validate how accurately the picloram MRS reflects actual exposure. As a supporting analysis, the researchers compared gene-expression data from pesticide-treated cardiomyocytes derived from induced pluripotent stem cells with tumor MRSs. This analysis did not establish a carcinogenic mechanism for picloram in colon cells.

The number of early-onset patients was small, and residual confounding is possible. The difficulty of accurately assessing the latency between exposure and cancer development is another limitation. The authors identified longitudinal and experimental studies to assess causality and dose-specific risk as priorities for further research. The current findings do not provide sufficient evidence to establish the effectiveness of routine epigenetic testing or specific preventive measures.

The study methods and results are available in the provided publicly accessible full text of the Nature Medicine paper.