Pesticide exposure and prostate cancer: A systematic rewiew and meta-analysis

Abstract

Prostate Cancer is the second most common type of cancer and the fifth cause of cancer mortality in men worldwide. However, its etiology is still very unclear and established risk factors include only advancing age, having a positive family history of prostate cancer and ethnicity. In search for possible modifiable risk factors, many studies have examined the role of various environmental and occupational exposures. Pesticide exposure has received much attention, as experimental studies had suggested carcinogenic potential of many pesticides types. On the other hand, a large number of pesticides have been postulated to possess endocrine disrupting properties, which make them of specific relevance to prostate cancer, being a hormone dependant malignancy. However, in spite of the availability of a large body of epidemiological literature relating farming and pesticide exposure to prostate cancer, results are controversial and inconsistent. Accordingly, we have systematically reviewed available epidemiological literature relating pesticide exposure to prostate cancer in order to examine the hypothesis that farmers are at an increased risk of developing prostate cancer. We have also aimed to evaluate the potential association between different levels of pesticide exposure and prostate cancer, and to explore potential sources of heterogeneity between studies. We searched PubMed, Web of Science and Scopus databases for case-control and cohort studies published from 1986 till 2015. We assessed the quality of the included articles using the Newcastle-Ottawa scale (NOS). Pooled estimates were calculated using the random-effects model. Heterogeneity was explored using sub-set analyses, sensitivity analyses and meta-regression. We have conducted three consecutive meta-analyses, the first included 14 case-control and 11 cohort studies relating farming to prostate cancer. For the cohort studies, pooled estimates showed high heterogeneity. Homogeneity was revealed by sensitivity analysis and the pooled estimate showed no association; 0.99, 95% CI 0.96-1.02, I2 =22.3%, p=0.252. For the case-control studies, there was moderate heterogeneity which was explained by location of the studies, decades of prostate cancer diagnosis, and type of control population. A higher association was observed for studies conducted in the USA, for older studies where data was collected between late 70s and late 80s, as well as for studies that used cancer patients as controls. The repeatedly obtained pooled estimate showed consistently a weak yet statistically significant association. The pooled estimate from the sensitivity analysis was 1.26, 95% CI 1.19-1.33, I2 = 0.00%, p = 0.570. For the second meta-analysis, which included four cohort studies and 21 case-control studies that quantified pesticide exposure, there was no association between low exposure to pesticides and prostate cancer. However, association was weak but significant for high exposure, pooled OR 1.33, 95% CI 1.02-1.63, I2 = 44.8%, p = 0.024. Heterogeneity was explained by a number of variables including the method used to assess pesticide exposure. Pooled OR was weak and non-significant for those having high serum levels of pesticides, 1.12(0.74-1.50), I2 = 0.00%, p=0.966, while a high and significant association was detected for studies that applied grouped non-individualized assessment of exposure, pooled OR was 2.24(1.36-3.11), I2 = 0.00%, p=0.955. On the other hand, for studies that used self-reporting of pesticide exposure, pooled OR was 1.34(0.91-1.77), I2=0.00%, p=0.493. Also, higher pooled estimates were observed for studies conducted in USA/Canada, those where the controls were cancer patients, and those having lower quality according to NOS criteria. Also, studies addressing pesticide exposed farmers and organochlorine pesticides showed higher pooled ORs. Nevertheless, an impact of the exposure assessment methodology and study quality was consistently observed. We have also noted an increased risk of prostate cancer for high exposure to pesticides among individuals with a positive family history of prostate cancer. Pooled OR was 2.23(1.05-3.41), I2 = 0.00%, p=0.646. We have conducted another meta-analysis to further examine the potential association between specific organochlorine pesticides and prostate cancer. Pooled estimates that we obtained for high exposure to DDT, DDE, hexachlorobenzene, oxychlordane and transnonachlor among the general population were weak and insignificant. On the other hand, pooled estimates for high occupational exposure to DDT, heptachlor and lindane showed a positive yet insignificant association. Based on the available epidemiological literature, we conclude that there is no concrete evidence for associating exposure to specific pesticides to an increased prostate cancer risk. Although a weak yet significant association was observed for high exposure to pesticides among farmers, an impact of exposure assessment methodology and the quality of the studies was also observed. There are still gaps in the available epidemiological data that makes the association unclear. These include deficiencies in exposure assessment methods and in selection of control populations, as well as lack of adjusting for important confounders including family history of prostate cancer and PSA testing variability.