Alavanja, MCR, C Samanic, M Dosemecil, J Lubin, R Tarone, CF Lynch, C Knott, K Thomas, JA Hoppin, J Barker, J Coble, DP Sandler and A Blair. 2003. Use of Agricultural Pesticides and Prostate Cancer Risk in the Agricultural Health Study Cohort. American Journal of Epidemiology 157:800–814.

In a series of previous studies, farming has been the single most consistent occupational risk factor for development of prostate cancer. In this study, Alavanja et al. strengthen that link by documenting a small but significant increase in prostate cancer risk for pesticide applicators, including farmers. Their research found that use of organochlorine pesticides and methyl bromide elevated prostate cancer risk for the entire cohort, and that several other pesticides elevated risk for men with a family history of prostate cancer.

What did they do? Alavanja et al. examined the relationship between exposure and response for 45 important agricultural chemicals and prostate cancer. Their data came from the Agriculture Health Study cohort: 55,332 pesticide applicators (commercial and private) recruited in Iowa and North Carolina 1993-1997.

Participants in the Agriculture Health Study were surveyed to determine uses of 50 different pesticides, methods of application, crops grown, nonfarm occupational exposure, family medical history, plus status of a series of potential confounding variables, such as smoking, alcohol consumption and involvement in engine repair (which results in solvent exposure).

Alavanja et al. used a variety of approaches to characterize each study participant's history of pesticide exposure, ranging from assessment of the frequency and intensity of exposure to each compound one-by-one, to a statistical characterization, called "factor analysis," that identified groups of pesticides that tended to be used together and allowed each study subject to be ranked on the intensity of exposure to these co-occurring groups of pesticides.

Any prostate cancer diagnoses prior to enrollment in the study were excluded from the analysis.

Health condition through 1999 was then determined by examining medical records. Over 98% of the cohort was included in the analysis.

As a baseline for comparison, Alavanja et al. used the overall rate of prostate cancer in men living in Iowa and North Carolina, adjusted for race, age and the period of observation (1994-1998).

What did they find? Not surprisingly, prostate cancer risk was higher in older men and in men from families with a history of prostate cancer.

Comparing pesticide applicators to baselines in North Carolina and Iowa, Alavanja et al. noted a small but significant excess in prostate cancer cases in the pesticide applicators: During the period of observation (1994-1998), 566 new prostate cancer cases were detected within the study population. This is 14% more than would have been expected based on the average overall incidence of prostate cancer (495 cases), or an odds ratio of 1.14. The incidence rate was higher compared to baseline in commercial applicators (41% greater) than among private applicators (13%).

The clearest pattern linking pesticide exposure to prostate cancer risk was for the fumigant methyl bromide. Applicators with the highest levels of exposure to methyl bromide were more likely to develop prostate cancer. For those applicators in the most heavily exposed group (the top 20%), the odds ratio was 3.47.

This association with methyl bromide was observed in an analysis of the total sample, as well in separate analyses of North Carolina and Iowa pesticide applicators and within private and commercial applicators.

In the analysis of groups of co-occurring pesticides (see above), the only significant pattern to emerge was that men who had used an older suite of chlorinated pesticides were at greater risk to prostate cancer.

Analysis also suggested that men from families with a history of prostate cancer were at greater risk following exposure to certain pesticides, including chlorpyrifos, coumaphos, permethrin and butylate. According to the authors, this suggests that familial genes or a shared familial environmental background may increase susceptibility to the effects of pesticides on prostate cancer risk.

What does it mean? Alavanja et al. find a small but significant increase in prostate cancer risk among pesticide applicators in Iowa and North Carolina. This finding is consistent with prior studies on prostate cancer in farmers.

One of the principal strengths of the study was its very large sample size, which allowed confirmation of relatively small effects.

Conversely, an unavoidable weakness was using subject recall of exposures, some of which, in the oldest members of the cohort, would have been based on decades-old memories. In this regard, it is important to note that failures and error in recall would almost certainly hide associations rather than create them, i.e., bias the results toward false negatives and lead to a lowering of the estimate of the odds ratio.

Excluding from analysis those with diagnosed cases of prostate cancer prior to enrollment may also have shifted the odds ratio downward. That would be the case, for example, if one effect of pesticide exposure was to hurry development of the disease in a subset of the population.