What do pesticides, farming, and dose effects have to do with the risk of developing connective tissue disease?
Could pesticide exposure in the home or in the fields lead to an increased risk of developing rheumatoid arthritis (RA) or systemic lupus erythematosus (SLE)? It seems that potential environmental exposures that could lead to immunologic derangement and the onset of autoimmunity are almost ubiquitous—cigarette smoke, crystalline silica, asbestos, mercury, lead, viruses and other microbes, pesticides, solvents, petroleum byproducts, and more. Many of these potential risks have been examined in relation to connective tissue disease risk. Not all have been rigorously examined and such studies are challenging.
In this issue of Arthritis Care & Research, Parks et al have investigated whether there is an association between residential or agricultural pesticide exposure and the subsequent development of RA or SLE using data from a large cohort of older women, the Women's Health Initiative Observational Study (WHI-OS) (1). At the start of the multicenter WHI-OS, the 93,676 postmenopausal participants were asked whether they had worked or lived on a farm and for how long (2). At the end of the first year of followup, they were also asked about insecticide exposure since age 21 years. They were asked whether they had “poured, sprayed, or applied insecticides” at home, leisure, or work, but not to report use of insect repellents, herbicides or fungicides, or pet treatments for fleas, ticks, or mites. The questionnaire attempted to have women quantify personal use of insecticides and application by others in duration (up to 20 years) and frequency of use (up to 25 times a year). Over the last few years, there has been an effort to identify and validate new-onset cases of RA and SLE in the WHI by requesting and reviewing a subset of medical records for these women (3). A hybrid strategy combining a woman's self-report of new-onset RA or SLE with the receipt of at least one of a list of disease-modifying antirheumatic drugs (DMARDs) had an acceptable positive predictive value of 62.2% for RA and 40.0% for SLE. In the WHI-OS, 186 incident cases of RA and 35 cases of SLE were then identified during the first 3 years of followup.
A significantly increased risk of developing RA and SLE associated with pesticide use, in particular the personal use of pesticides (but not application by others in the residential or office setting), was detected in this study. Among the women with a history of farming, those who indicated they had been exposed to the frequent application of pesticides by others had a 2.7 times increased risk of RA/SLE. The risks of both RA and SLE appeared to be elevated, although the results were reported for the most part in terms of RA/SLE risk (with the small number of SLE cases, the confidence intervals were wide and the results nonsignificant for SLE alone).
The association between the personal use of pesticides and the increased risk of RA and SLE will first need to be validated. It may be a false-positive; chance, bias, or confounding might explain the association. The study has limitations, although most of these would have resulted in misclassification and muddling, therefore biasing the study toward the null or no finding. The case definition of self-reported RA or SLE and use of certain DMARDs was not very accurate, and there certainly could have been misclassification of both cases and healthy women. The wording on the WHI questionnaire was vague and may have been difficult to understand, leading to misreporting.
While the exposure assessment in this study may have been a bit rough and the outcome assessment may have been a bit coarse, reported use of pesticides was clearly associated with an increased risk of RA and SLE among these older women, with a dose response! Personally mixing or applying pesticides less than once a year was associated with no increased risk, but the risk steadily increased according to the number of times a year and the number of years of personal mixing and pesticide application. Those with the highest amounts and durations of exposure had a 2.2-fold higher risk of developing RA or SLE than those with no exposure.
Epidemiologists love dose responses. They point to underlying biologic mechanisms and perhaps even causation. A biologic gradient was one of the 9 criteria suggesting causation in the epidemiologic associations that Sir Austin Bradford Hill put forward in his presidential address, “The Environment and Disease: Association or Causation,” to the Royal Society of Medicine in 1965 (4). The other criteria that he outlined in that discourse are highlighted below.
Strength of association and consistency of findings across studies
In the current study, a doubling of risk is a hefty increased risk, especially as the women were followed for only a few years at an older age. Farming as an occupation has been associated with increased risk of RA in past studies, although it has not been clear what about farming is responsible for increased risk (5, 6). In the Carolina Lupus Study, a case–control study that collected detailed occupational and exposure histories from African American women with SLE and matched controls, “mixing pesticides for agricultural work” was strongly associated with increased risk of SLE (7). Therefore, while the results are congruent with limited past evidence implicating agricultural pesticide exposure, the association needs confirmation.
Specificity of association
Pesticide exposure appears not to be specific for RA or SLE, but may increase the risk of both of these related autoimmune diseases with complex etiologies that primarily strike women. However, RA and SLE are different in many obvious respects and therefore their risk factors may be different. Moreover, as multiple different environmental exposures have been implicated in the etiology of these autoimmune diseases, how specific is pesticide exposure? Although we don't know yet, specific patterns may emerge with continued research. It is possible that pesticides, solvents (8), and persistent organic pollutants could have similar fundamental effects on susceptible genotypes, possibly through the aryl hydrocarbon receptor (9) or toll-like receptors (10), predisposing to the development of autoimmunity in similar ways.
Temporality of exposure and disease
The WHI participants were ages 50–79 years at study baseline when they were asked to recall their pesticide exposures since age 21 years. Is the latency period for the development of RA or SLE up to 58 years long? Unfortunately, information about when these women were mixing and applying pesticides in relation to when they developed disease is not available, so the time window for potential exposure is very wide. Presumably, women at the highest risk would have developed their disease, in particular SLE, when younger than the WHI enrollment age. Therefore, there may have been a “depletion of susceptibles” effect and even an underascertainment of exposure-related cases, and thereby an underestimation of the association. On the other hand, it could be that those who develop disease at younger ages may have a relatively higher load of genetic susceptibility factors, while the effect of cumulative environmental exposures on increasing risk of developing SLE or RA could be more important among those who develop disease at older ages with more moderate underlying genetic risk factor profiles.
Common pesticides include organophosphates, carbamates, and pyrethroid. Organophosphates and carbamate pesticides block acetylcholinesterase and are known neurotoxins (11) and have been associated with increased risk of leukemia (12) and lymphoma (13). Pyrethroids are increasingly used in homes because of low acute toxicity, but have also been implicated in increased risk of multiple myeloma (14). Exposure to pesticides can be transdermal or through inhalation or ingestion. Several possible mechanisms by which these chemicals could affect the risk of autoimmune disease have been proposed, including the generation of reactive oxygen species and oxidant stress, and alteration of lymphocyte and macrophage function, affecting cytokine levels and increasing antibody production (15). These effects have been observed in vitro and in animal models (16–19), although there is no good animal model of pesticide-induced autoimmunity. It is still not clear how pesticides alter immune function or risk of autoimmunity in humans. However, it is highly plausible that they could.
Coherence of data
By this, Sir Hill meant that the “cause-and-effect interpretation of our data should not seriously conflict with the generally known facts of the natural history and biology of the disease.” What we currently know about the natural history and biology of both RA and SLE is that environmental exposures trigger disease in those who are genetically susceptible, and that disease develops insidiously over time; autoantibodies and inflammatory cytokines are detectable years prior to the clinical disease (20, 21). Exposure to pesticides as a triggering factor fits within this framework.
Natural experiment and analogy to other causative associations
Occasionally there is a natural experiment that helps to support causation. For example, Dr. John Snow removed the handle from London's Broad Street pump in 1854 and stymied a cholera outbreak related to its infected water (22). To our knowledge, such a natural experiment does not exist in this case, but we will keep thinking and searching.
The association between pesticides and RA and SLE lends itself to analogy to a host of proven and suspected environmentally-triggered diseases. For example, non-Hodgkin's lymphoma, another disease of altered immunity, has been associated with exposure to pesticides and farm work (23, 24).
Sir Hill concluded his lecture in 1965 by stating, “Finally, in passing from association to causation, I believe in real life we must consider what flows from that decision.” As we accumulate and weigh the evidence concerning environmental exposures and risk of autoimmune diseases, several actions may be necessary. This evidence may lead to new discoveries of the molecular mechanisms of these diseases and to new medications to control or prevent them. Perhaps the evidence will be so strong that certain pesticides or other factors will need to be withdrawn or controlled. Granted, RA and SLE are rare diseases affecting 1–2% of the population, and the absolute risk of developing one of them, even in the face of high pesticide or other triggering exposure, is still likely low. But for some, those with a family history, known genetic variation, or early symptoms, the risk associated with an environmental exposure may be much higher. One goal should be to identify and protect those at risk. We commend Parks and colleagues for their thoroughly thought-provoking analysis and look forward to future research on the topic.
Drs. Costenbader and Laden wrote the article and approved the final version to be published.