Pesticide Exposure and Risk of Systemic Lupus Erythematosus in an Urban Population of Predominantly African-American Women

Abstract

Objective:

Past studies have reported associations between pesticide exposure and risk of systemic lupus erythematosus (SLE). Residential pesticide exposure has been less well studied than agricultural exposure. The purpose of this study was to assess SLE risk associated with residential pesticide exposure in an urban population of predominantly African-American women.

Methods:

Adult women with SLE were identified via six hospital databases and community screening in three neighborhoods in Boston, Massachusetts. Controls were adult women volunteers from the same neighborhoods who were screened for absence of connective tissue disease and anti-nuclear antibodies. Subjects were considered exposed to pesticides if they had ever had an exterminator for an ant, cockroach, or termite problem prior to SLE diagnosis or corresponding reference age in controls. Risks associated with pesticide exposure were analyzed using multivariable logistic regression models, adjusted for sociodemographic factors.

Results:

We identified 93 SLE subjects and 170 controls, with similar baseline characteristics. Eighty-three percent were African-American. Pesticide exposure was associated with SLE, after controlling for potential confounders (odds ratio 2.24, 95% confidence interval 1.28–3.93).

Conclusion:

Residential exposure to pesticides in an urban population of predominantly African-American women was associated with increased SLE risk. Additional studies are needed to corroborate these findings.

The etiology of systemic lupus erythematosus (SLE) is not well-understood¹, and certain demographic groups are more affected than others. Among African-Americans, the incidence of SLE is increased by 5- to 9-fold compared to Caucasians.² SLE is also significantly more common among women, as approximately 90% of patients diagnosed with SLE are women.³ As a result, SLE is particularly more common among African-American women, who have a prevalence rate of 211 per 100,000 person-years as compared to 64 per 100,000 person-years in Caucasian women.⁴ The prevalence of SLE is also significantly higher in African-Americans compared to West Africans, suggesting that environmental factors play a key role in SLE pathogenesis.⁵

Environmental factors that have been implicated in the development of SLE include crystalline silica, cigarette smoking, environmental pollutants, infections, and sex hormones.^6-7 Pesticides in particular have been hypothesized to contribute to the development of SLE, and have been shown to have immunologic and hormonal effects in vitro.^7–9 Studies of (NZB x NZW) F1 lupus-prone mice have pointed to a role of organochlorine pesticides in the pathogenesis of SLE^10–12, although the evidence to date in human observational studies has been conflicting.^13–17

The aim of this study was to investigate the association between residential pesticide exposure and risk of developing SLE among women in three urban neighborhoods in Boston, Massachusetts with large African-American populations. Our hypothesis was that residential pesticide exposure was associated with increased risk of SLE in this population.

Patients and Methods

Study population

The study population included women who were enrolled in the Roxbury Lupus Project^18–20, which was initiated in 2002 in response to community concerns about increased SLE incidence in the Boston neighborhoods of Roxbury, Dorchester, and Mattapan, all of which have large African-American populations.²¹ The residents of these three neighborhoods also have median household incomes well below the $53,136 median household income of Boston overall.²² With the assistance of the Massachusetts Department of Public Health, the National Institute of Environmental Health Sciences (NIEHS), and the SLE patient advocacy group Women of Courage, prevalent and incident cases of SLE were identified in these three neighborhoods, healthy controls were recruited from the same neighborhoods, and environmental exposures of all subjects were assessed. This study was conducted in compliance with the Helsinki Declaration.

Study design

After obtaining institutional review board approval at each participating hospital, adult women with SLE were identified by screening six local hospital databases (Brigham and Women’s Hospital, Boston Medical Center, Massachusetts General Hospital, New England Medical Center, Beth Israel Deaconess Medical Center, and Carney Hospital) for SLE diagnosis codes among women with zip codes within the study area. In addition, adult women residents of the Roxbury, Dorchester, and Mattapan neighborhoods were screened for the presence of SLE at health fairs, neighborhood events, lupus support groups, educational seminars, churches, and community health clinics in the study area, using the Connective Tissue Disease Screening Questionnaire (CSQ)^{18, 23} and a fingerstick blood sample for antinuclear antibody (ANA) testing (titer >1:40 was considered positive). Each subject with possible SLE was reviewed by a rheumatologist and confirmed to have ≥4 American College of Rheumatology (ACR) criteria for SLE.²⁴ Control subjects were adult women volunteers from the same neighborhoods, screened for the absence of connective tissue disease by CSQ and negative fingerstick ANA. Control subjects were identified at health fairs, neighborhood events, educational seminars, churches, and community health clinics in the study area. Each subject provided written informed consent in order to participate in the study.

Exposure assessment

Trained community outreach workers conducted in-person interviews with study participants from April 2002 to August 2003. During these interviews, detailed questionnaire data were collected from study participants regarding lifetime occupational and residential exposure to various chemicals. Type and frequency of pesticide exposure prior to SLE diagnosis or corresponding reference age in subjects without SLE were assessed. Subjects were considered exposed to pesticides if they had ever required an exterminator for an ant, cockroach, or termite problem in their home. Subjects were also asked about the number of times that pesticide exposure had occurred and their age(s) at the time of exposure. Exposure to rodenticides was not included in the analyses in order to decrease heterogeneity of the type of pesticide exposure. SLE subjects and controls were asked about when pesticide exposure had occurred, and only exposures prior to SLE diagnosis in cases or prior to a matched reference age in controls were considered. Occupational exposures to pesticides (for example, work as an exterminator) was not included in the analyses as this only applied to 2 subjects with SLE and 1 control subject. Additionally, other sociodemographic variables such as age, self-reported race/ethnicity (African-American non-Hispanic, Caucasian non-Hispanic, Hispanic, or other), parity (whether ever parous or not), employment status (whether working at least part-time or not), educational attainment (whether completed high school or not), smoking status (current, past, or never smoker prior to age at SLE diagnosis or corresponding reference age), and place of birth (whether born in Boston or not) were ascertained.

Statistical analyses

Baseline characteristics of cases and controls were compared using the Wilcoxon rank-sum test for age and Chi-square tests for categorical variables. The association between exposure to pesticides and SLE was analyzed using multivariable logistic regression models. The first model was adjusted for age and race/ethnicity only, and the second model was adjusted for age, race/ethnicity, parity, employment status, educational attainment, smoking status, and place of birth. The multivariable logistic regression models produced odds ratios (OR) and 95% confidence intervals (CI) to estimate the measure of association between pesticide exposure and SLE. We also investigated whether increasing frequency of pesticide exposure (in quartiles) was associated with increasing risk of SLE, using the Cochran-Armitage test for trend. We set α=0.05 to determine statistical significance, and all p-values were two-sided. Data were analyzed using SAS 9.4 (Cary, NC).

Results

Ninety-three SLE subjects and 170 controls without SLE were identified. Ninety percent of SLE subjects were identified via hospital database screening, with the remaining 10% identified via community screening. Table 1 displays the baseline characteristics of subjects with and without SLE, with no significant differences between the two groups. The study population was 100% female and 83% African-American. Among the 93 subjects with SLE, the mean age at SLE diagnosis was 36 years and the mean age at the time of the study interview was 44 years.

Table 1.

Baseline characteristics of subjects with and without SLE^*

Characteristics	SLE (n=93)	No SLE (n=170)	p-value†
Age, mean ± SD years	44 ± 13	47 ±15	0.11
Race/Ethnicity			0.10
African-American non-Hispanic	71 (76)	147 (86)
Hispanic	7 (8)	7 (4)
Caucasian non-Hispanic	6 (6)	3 (2)
Other	9 (10)	13 (8)
Parity, ever	71 (76)	134 (79)	0.64
Working full-time or part-time	44 (47)	90 (53)	0.38
Completed high school	71 (76)	122 (72)	0.42
Smoking status (≥100 cigarettes per lifetime)‡			0.33
Current (past 30 days smoked ≥1 cigarette)	16 (17)	35 (21)
Past	20 (22)	47 (28)
Never	57 (61)	88 (52)
Born in Boston	36 (39)	81 (48)	0.16

^*Unless indicated otherwise, values are the number (%).

^†By Chi-square test. Wilcoxon rank-sum test was used for age.

^‡Before age at diagnosis or corresponding reference age.

The prevalence of pesticide exposure was 56% in the overall study population. In unadjusted Chi-squared analyses, subjects with SLE were more likely to have had exposure to pesticides than subjects without SLE (65% vs. 51%, p=0.03). Table 2 displays the results of the multivariable logistic regression analyses. In multivariable logistic regression models adjusted for age and race/ethnicity only, pesticide exposure was significantly associated with increased risk of SLE (OR 1.96, 95% CI 1.14–3.38). This association remained statistically significant when the multivariable logistic regression model was further adjusted for parity, employment status, educational attainment, smoking status, and place of birth in addition to age and race/ethnicity (OR 2.24, 95% CI 1.28–3.93). A dose-response effect for increased frequency of pesticide exposure was not statistically significant (p for trend=0.99, shown in Figure 1).

Table 2.

Multivariable logistic regression analysis examining the association between pesticide exposure and SLE.

Model	OR	95% CI	p-value
Ever exposed to pesticides prior to SLE/index date (Model 1)*	1.96	(1.14, 3.38)	0.02
Ever exposed to pesticides prior to SLE/index date (Model 2)**	2.24	(1.28, 3.93)	0.005

OR=odds ratio. CI=confidence interval.

^*Model 1 was adjusted for age and race/ethnicity.

^**Model 2 was adjusted for age, race/ethnicity, parity, employment status, educational attainment, smoking status, and place of birth.

Figure 1.

Test for trend: risk of SLE with increasing frequency of pesticide exposure

Discussion

In our case-control study, we found that residential exposure to pesticides among an urban population of predominantly African-American women was associated with a greater than two-fold increased risk of SLE, even after controlling for potential confounders. Our findings demonstrate that previous reports of an association between pesticide exposure and SLE^13–15 may be applicable to urban African-American women, a population at increased risk for this disease.

The association between pesticides and SLE has biological plausibility, as pesticide exposure has been shown to lead to altered cytokine expression, increased oxidative stress, increased lymph node follicle development, changes in T-cell subtypes, and binding to estrogen receptors in vitro.^7–9 Data from murine studies also support an association between pesticides and SLE.^10–12 In a study by Sobel et al, ovariectomized female (NZB x NZW) F1 lupus-prone mice who were treated with the organochlorine pesticides chlordecone, methoxychlor, or o,p -dichlorodiphenyltrichloroethane (o,p -DDT) had significantly decreased time to onset of renal impairment and albuminuria. In addition, treatment with chlordecone led to dose-related accelerated formation of anti-double-stranded deoxyribonucleic acid (DNA) antibodies and glomerulonephritis. It was hypothesized that these outcomes were due to estrogenic properties of the organochlorine pesticides, but there was no clear correlation between pesticide treatment and uterine hypertrophy.¹⁰ Follow-up studies in ovary-intact female (NZB x NZW) F1 lupus-prone mice had similar results. Interestingly, treatment with organochlorine pesticides for one year did not lead to SLE in non-lupus-prone mice (BALB/c strain), indicating the importance of genetic susceptibility to SLE for this association.¹¹ In a separate study by Li et al, treatment with o,p -DDT led to accelerated albuminuria in (NZB x NZW) F1 lupus-prone mice.¹² These positive findings in murine studies have led to increased interest in studying the association between pesticide exposure and SLE in humans.

There have been two studies examining the association between pesticide exposure and ANA positivity in humans, and they have had conflicting results. Rosenberg et al reported that, in a cross-sectional study of 322 adults in rural Canada in 1999, lifetime exposure to pyrethroid insecticides was associated with a positive ANA after adjusting for age, sex, and other insecticide exposures.²⁵ In contrast, a cross-sectional study of 137 African-American male farmers in rural North Carolina from 2004 found no association between plasma concentrations of the DDT metabolite dichlorodiphenyldichloroethylene (DDE) and ANA positivity, after adjusting for potential confounders include age, cigarette smoking, and years of pesticide use.⁸

In addition, several human observational studies have examined pesticide exposure and risk of SLE specifically. A 2004 study of 265 SLE cases and 355 controls in the southeastern United States (U.S.) found that a history of mixing pesticides for agricultural work was associated with SLE (OR 7.4, 95% CI 1.440.0).¹³ A case-control study conducted by Parks et al within the NIEHS Sister Study cohort (2016) found that SLE was associated with childhood residential pesticide use at least once monthly (OR 2.3, 95% CI 1.34.1) and with personal use of agricultural pesticides (OR 4.2, 95% CI 2.4–7.7).¹⁴ In addition, a prospective cohort study of 76,861 post-menopausal women in 2011 found that a prior history of applying insecticides was associated with a 50% increased risk of developing SLE over 3 years of follow-up.¹⁵ The findings of these three studies are contrasted with null findings by Gold et al in 2007 and Cooper et al in 2010.^16–17 Gold’s group found that in a retrospective study of U.S. death certificates, there was no significant association between deaths from SLE and occupational pesticide exposure. However, as occupational pesticide exposure was ascertained from information on the death certificate, there was a possibility of information bias in that study. Cooper et al conducted a 2010 study of 258 SLE cases and 263 controls in Canada and also found no association between occupational exposure to pesticides and SLE (OR 1.1, 95% CI 0.43–3.0). Given these conflicting results, we sought to further elucidate the association between pesticide exposure and SLE in a majority-minority population in which residential pesticide exposure is significantly more common than agricultural or occupational exposure.

Strengths of our study include a focus on residential pesticide exposure which has been less well-studied than agricultural and occupational pesticide exposure, and a study population that is at increased risk for SLE but is less likely to be included in research studies (urban, predominantly African-American women). Limitations of our study include a relatively small study population (N=263), lack of detailed data regarding specific types of pesticide exposures, lack of data regarding other potential confounders (including socioeconomic status (SES), body mass index, alcohol use, and oral contraceptive use) which may have led to residual confounding, and potential for recall bias given the case-control study design. Recall bias may have particularly affected our data regarding frequency of pesticide exposure, as subjects may have had limited ability to correctly recall the number of prior pesticide exposures. We hypothesize that this, and small sample size, may have affected our test for trend analyses, which were non-significant for increasing frequency of pesticide exposure. Additionally, recall bias may have led to overestimation of the effect of pesticide exposure on SLE risk, as it is probable that subjects with SLE are more likely to report their prior exposures than controls given public concern that environmental factors contribute to SLE pathogenesis. We attempted to minimize our lack of SES data by adjusting for employment status and educational attainment, and by recruiting all subjects from neighborhoods with low median household incomes.

In conclusion, we found that residential exposure to pesticides was associated with a greater than two-fold increased risk of SLE in an urban, predominantly African-American population of adult women. Since 51% of control subjects also had a history of residential exposure to pesticides, our results suggest that genetic susceptibility to SLE may be necessary for the increased risk conferred by pesticide exposure, as has been demonstrated in previous murine studies.^10–12 As our results may be limited by recall bias and/or residual confounding, additional research is needed to determine whether pesticide exposure is implicated in SLE pathogenesis, or potentially is serving as a surrogate for a related exposure such as pest burden or poor living conditions. Additionally, larger, prospective studies with more detailed data on type, frequency, and duration of pesticide exposure should be conducted to confirm our findings. We also advocate for further studies examining pesticide exposure and SLE risk in other populations at increased risk for SLE, including populations of Asian and Hispanic descent.