Abstract
Background:
We previously reported an association between household chemical exposures and an increased risk of pediatric-onset multiple sclerosis.
Methods:
Using a case-control pediatric multiple sclerosis study, gene-environment interaction between exposure to household chemicals and genotypes for risk of pediatric-onset multiple sclerosis was estimated.
Genetic risk factors of interest included the two major HLA multiple sclerosis risk factors, the presence of DRB1*15 and the absence of A*02, and multiple sclerosis risk variants within the metabolic pathways of common household toxic chemicals, including IL-6 (rs2069852), BCL-2 (rs2187163) and NFKB1 (rs7665090).
Results:
490 pediatric-onset multiple sclerosis cases and 716 controls were included in the analyses. Exposures to insect repellent for ticks or mosquitos (OR: 1.47, 95% CI: 1.06–2.04, P = 0.019), weed control products (OR: 2.15, 95% CI: 1.51–3.07, P < 0.001), and plant/tree insect or disease control products (OR: 3.25, 95% CI: 1.92–5.49, P < 0.001) were associated with increased odds of pediatric-onset multiple sclerosis. There was significant additive interaction between exposure to weed control products and NFKB1 SNP GG (attributable proportions (AP) 0.48, 95% CI: 0.10–0.87), and exposure to plant or disease control products and absence of HLA-A*02 (AP 0.56; 95% CI: 0.03– 1.08). There was a multiplicative interaction between exposure to weed control products and NFKB1 SNP GG genotype (OR: 2.30, CI: 1.00–5.30) but not for other exposures and risk variants. No interactions were found with IL-6 and BCL-2 SNP GG genotypes.
Conclusions:
The presence of gene-environment interactions with household toxins supports their possible causal role in pediatric-onset multiple sclerosis
Introduction
While significant progress has been made toward elucidating distinct genetic and environmental risk factors for multiple sclerosis (MS), there is still a large knowledge gap regarding the contributions of environmental exposures in genetically susceptible individuals1.
Our preliminary work reported that exposure to insect repellent for tick or mosquitos, weed control products, and plant/tree insect or disease control products during the perinatal period and childhood was strongly associated with MS risk2. A growing body of literature suggests that exposure to toxic chemicals may increase the risk of other neurologic diseases, including Parkinson’s disease3, Alzheimer’s disease4, autism spectrum disorder5, and attention deficit hyperactivity disorder6. The molecular basis for the contribution of these chemicals to disease susceptibility remains unclear7. Although exposure to a chemical toxin alone may be insufficient to cause neurological diseases such as pediatric-onset multiple sclerosis (POMS), it may be more likely to affect genetically susceptible individuals1.
We sought to examine a role in POMS risk for the potential interactions between chemical exposures and HLA-DRB1*15, HLA-A*02, and MS risk variants within genes in metabolic pathways of common household toxic chemicals, including IL-6 (rs2069852), BCL-2 (rs2187163), and NFKB1 (rs7665090).
Methods
Study population
We used data on POMS cases and healthy controls from the US Network of Pediatric Multiple Sclerosis (R01NS071463, PI Waubant). Participants were recruited at 17 participating pediatric MS clinics in the USA between November 1, 2011, and July 1, 2017. POMS cases with symptom onset at less than 18 years of age were enrolled in this study, as previously described8. Case ascertainment was established based on the diagnosis of at least two pediatric MS specialists from the US pediatric MS network, using the 2010 McDonald Criteria9. The network collects retrospective demographic characteristics and medical history in a centralized database10. Healthy controls were recruited from primary care and non-MS pediatric clinics at the participating institutions during case recruitment and were frequency-matched to cases using age and sex. All controls had no history of neurological or autoimmune disease (apart from headache/migraine), no first-degree biological relatives with MS, and were under 22 years of age.
The institutional review board of each participating institution approved the study protocols. Informed consent was obtained from all participants/legal guardians before enrollment. Blood samples were collected upon enrollment.
Environmental data
Information regarding household chemical exposures was collected using a comprehensive environmental questionnaire (http://www.usnpmsc.org/Documents/EnvironmentalAssessment.pdf) completed by parents/legal guardians at the time of enrollment, including prior exposure and timing of exposure to different household chemicals. Subjects were considered exposed to the toxic chemicals if the exposure was before disease onset. Household chemicals previously found to be associated with the risk of POMS were included in gene-environment interaction (GxE) models: 1) insect repellent for ticks or mosquitos, 2) weed control products, and 3) plant/tree insect or disease control products. Commonly used chemicals previously reported to cause neurodevelopmental and immunological interference were also selected for further investigation, including glyphosate, linuron, chlorpyrifos, diazinon, acephate, malathion, permethrin, bifenthrin, methyl bromide, imidacloprid, and avermectin (5).
Genotyping and definition of genetic risk factors
Study participants were genotyped using the Infinium 660K BeadChip or HumanOmniExpress BeadChip. Stringent quality control measures and comparison of sample genotypes across two Illumina platforms were performed using PLINK version 1.9. The alignment, phasing, imputation, and variant filtering were done as described previously11.
We searched PubMed for papers published before March 31, 2022, with the following terms: “name of each chemical”, “gene”, “expression”, “alteration”, and “regulation”. Seventeen genes were reported with altered expression by exposure to one of the above chemicals (Table 1). Looking at the largest genome-wide association study (GWAS) of MS conducted by the International Multiple Sclerosis Genetics Consortium (IMSGC) three of the seventeen candidate genes, specifically IL-6 (rs2069852, G/A), BCL2 (rs2187163, G/A), and NFKB1 (rs7665090, G/A), were reported to increase MS risk12. The expression of IL-6, BCL2, and NFKB1 were shown to be altered by exposure to glyphosate, which is widely used in weed control products13–15. We utilized these three SNPs to evaluate potential interactions with exposure to weed control products for increased risk of POMS. We further investigated the potential interaction between the presence of HLA-DRB1*15 or absence of HLA-A*02 (as the strongest genetic variants associated with POMS) and exposure to insect repellent for tick or mosquitos, weed control products, or plant/tree insect or disease control products, separately and in different models.
Table 1.
Gene expression alterations by exposure to toxic chemicals.
| Author (Year of publish) |
Toxic chemical | Altered gene |
|---|---|---|
| Gui20 (2012) |
Glyphosate | BECN1 |
| Martinez15 (2020) |
BAX, BCL2, CASP3, CASP9, TP53, NFKB, SNCA, IL-6, TNFα | |
| Tang14 (2020) |
IL-1β, IL-6, TNF-α, NFKB, CASP3 | |
| Wozniak13 (2020) |
TP53, BCL2 | |
| Zheng21 (2021) |
NRF2, NFKB2, IL-1β | |
| Gargouri22 (2018) |
Bifenthrin (pyrethroid) | NFKBp65, TNF-α, NRF2 |
| Ibrahim23 (2021) |
Avermectin | BAX, CASP3, CASP9 |
| Mense24 (2006) |
Chlorpyrifos | IL6R |
| Zhao25 (2019) |
NRF2, IL-1β, IL-18 | |
| Chorfa26 (2013) |
Rotenone, Paraquat, Maneb |
SNCA |
| Wheeler27 (2019) |
Linuron | SIGMAR, XBP1 |
DNA samples of all subjects have been genotyped for DRB1 status, as previously published8. DRB1 status was dichotomized according to the presence of one or more HLA-DRB1*15 alleles versus no carriage. The HLA-A*02 (rs2975033) tagging SNP was imputed. The allele “A” of rs2975033 has been reported to be in strong linkage disequilibrium (r2 = 0.97) with the HLA-A*02 allele16. Participants were categorized according to no carriage of any HLA-A*02 allele (absence of HLA-A*02) versus any carriage. In addition, the subjects were dichotomized according to GG genotype of IL-6 SNP (rs2069852 G/A)17, GG genotype of BCL2 SNP (rs2187163 G/A)12, and GG genotype of NFKB1 SNP (rs7665090 G/A)18 as risk alleles, versus other (GA and AA genotypes).
We used the SNP weighting method to estimate the percentage of genetic ancestry related to four major populations (European, East Asian, West African, and Native American)19.
Statistical analysis
The odds of having POMS associated with each environmental and genetic exposure was assessed separately using logistic regression models adjusted for age, sex, genetic ancestry, mother’s education as a proxy of socioeconomic status (SES), and recruiting site. Environmental exposures included three categories of household chemicals, namely (1) insect repellent for ticks or mosquitos, (2) weed control products, and (3) plant/tree insect or disease control products). The presence of HLA-DRB1*15, absence of HLA-A* 02, and GG genotype of IL-6 SNP, GG genotype of BCL-2 SNP and GG genotype of NFKB1 SNP were also assessed. We tested for additive interaction between each household chemical exposure and genotype status with logistic regression models calculating the relative excess risk due to interaction (RERI) and the attributable proportion (AP) of disease due to interaction28. An interaction term was also added to logistic regression models to assess the presence of multiplicative interaction. In the primary analysis, patients with missing environmental exposures were excluded. Aiming to minimize a possible differential missingness pattern of the environmental exposure, a sensitivity analysis using inverse probability weighting (IPW) was performed, as previously described29. All statistical analyses were done in Stata 17.0 (Stata Corp, College Station, TX). We calculated RERI and AP, their CIs, and p-values according to the method described in VanderWeele and Knol et al30.
Results
Participants characteristics
490 POMS cases and 716 healthy controls were enrolled in this study. Baseline characteristics are shown in (Table 2). Hispanic ethnicity was more frequent in cases than in controls (32.1% vs. 23%, P < 0.001). Mother’s education was higher in controls compared to cases (university degree 46.7% vs. 31.9%, P < 0.001). The mean time interval between clinical onset and MS diagnosis was four months (median months [IQR: 0 – 5.13]).
Table 2.
Characteristics of study participants.
| Characteristics | POMS (n=490) | Healthy controls (n=716) | P-value |
|---|---|---|---|
| Age, median years (IQR) | 16 (3) | 15 (5) | 0.034a |
| Female, n (%) | 316 (64.3) | 420 (59.1) | 0.07b |
| Ratio, F:M | 1.8:1 | 1.44:1 | |
| Self-reported race, n (%) | |||
| White | 312 (68.9) | 439 (66.4) | 0.398b |
| Self-reported ethnicity, n (%) | |||
| Hispanic | 150 (32.1) | 156 (23) | <0.001b |
| Genetic ancestry estimation, mean (SD) | |||
| European ancestry percentage | 65.6 (0.3) | 66.7 (0.3) | 0.57c |
| East Asian ancestry percentage | 3.6 (0.2) | 8 (0.2) | 0.007c |
| Native American ancestry percentage | 10.2 (0.2) | 7.6 (0.2) | 0.006c |
| West African ancestry percentage | 19.5 (0.3) | 17.6 (0.3) | 0.23c |
| Mother’s education, n (%) | |||
| Less than high school | 52 (12.1) | 40 (6.4) | <0.001d |
| High school and college | 240 (56) | 295 (46.9) | |
| University degree | 137 (31.9) | 293 (46.7) |
POMS: Pediatric-onset multiple sclerosis; IQR: interquartile range; SD: standard deviation.
Mann Whitney test
Fisher’s exact test
Unpaired t-test
Chi-squared test
Toxic chemical exposures are associated with higher odds of POMS
As shown in (Table 3), logistic regression analysis adjusting for covariates (age, sex, genetic ancestry, mother’s education, and recruiting site), POMS cases had higher odds of exposure to insect repellent for tick or mosquitos (OR 1.47, 95% CI 1.06–2.04, P = 0.019), weed control products (OR 2.15, 95% CI 1.51–3.07, P < 0.001) and plant/tree insect or disease control products (OR 3.25, 95% CI 1.92–5.49, P < 0.001) compared to controls. Sensitivity analyses using IPW resulted in similar results (Tables 4)
Table 3.
Toxic chemical exposure analysis.
| POMS (N = 490 ) | Healthy controls (N = 716 ) | Odds ratio (95% CI) | P-Value | |
|---|---|---|---|---|
| Insect repellent for ticks or mosquitos, x/n (%) | 179/293 (61) | 262/479 (55) | 1.47 (1.06 – 2.04) | 0.019 |
| Weed control products, x/n (%) | 114/309 (37) | 130/498 (26) | 2.15 (1.51 – 3.07) | <0.001 |
| Plant/tree insect or disease control products, x/n (%) | 46/307 (15) | 35/510 (7) | 3.25 (1.92 – 5.49) | <0.001 |
Adjusted for age, sex, genetic ancestry, mother’s highest level of education, recruiting site
Table 4.
Toxic chemical exposure analysis (sensitivity analysis: using inverse probability weighting).
| POMS (N = 490 ) | Healthy controls (N = 716 ) | Odds ratio (95% CI) | P-Value | |
|---|---|---|---|---|
| Insect repellent for ticks or mosquitos, x/n (%) | 179/293 (61) | 262/479 (55) | 1.55 (1.11 – 2.16) | 0.009 |
| Weed control products, x/n (%) | 114/309 (37) | 130/498 (26) | 2.20 (1.54 – 3.15) | <0.001 |
| Plant/tree insect or disease control products, x/n (%) | 46/307 (15) | 35/510 (7) | 3.35 (1.99 – 5.65) | <0.001 |
Adjusted for age, sex, genetic ancestry, mother’s highest level of education, recruiting site
Genetic characteristics
POMS cases were more likely to have at least one HLA-DRB1*15 allele (41% vs. 23.6%, P<0.001). Available imputed genotype data for HLA-A*02 demonstrated a higher proportion of POMS cases than healthy controls without the HLA-A*02 allele (61.5% vs. 56.3%, P= 0.29). In addition, a higher proportion of POMS cases had the GG genotype of NFKB1 risk SNP (rs7665090) (37% vs. 31.5%, P= 0.44).
Assessment of G × E interactions
Results from G × E analyses of environmental toxic chemical exposures and genetic variants of interest are presented in (Table 5). In those without exposure to weed control products, the NFKB1 SNP GG genotype did not significantly increase the odds of POMS (OR 0.77, 95% CI: 0.48–1.21). When both NFKB1 SNP GG genotype and exposure to weed control products were present, the odds of exposure in cases compared to controls increased to 3.32 (95% CI: 1.74–6.36), which is higher than the product of the two respective ratios (0.77 and 1.86). The AP indicated that 50% (AP: 0.50, 95% CI: 0.13–0.88) of the POMS risk in individuals with NFKB1 SNP GG genotype and exposure to weed control products were attributable to their interaction. In addition, greater than 50% (AP: 0.56, 95% CI: 0.06–1.07) of the POMS risk in individuals without HLA-A*02 and exposure to plant/tree insect or disease control products were attributable to the interaction between these risk factors. Without exposure to plant/tree insect or disease control products, the absence of HLA-A*02 did not significantly increase the odds of having POMS. When both absences of HLA-A*02 and exposure to plant/tree insect or disease control products were present, the OR of POMS increased to 4.93 (95% CI: 2.23–10.88), indicating an additive interaction with an AP of 0.56 (95% CI: 0.06–1.07). Sensitivity analyses using IPW resulted in similar results (Tables 6)
Table 5.
Additive interaction between household chemical exposure and genotype. Stratified analysis assessing the interactions between household chemical exposure and presence of HLA-DRB1*15 alleles, absence of HLA-A*02, IL-6 SNP GG genotype, BCL2 SNP GG genotype and NFKB1 SNP GG genotype
| Household chemical exposure | Genetic exposure | N Case/ N Control | OR (95% CI) | RERI (95% CI) | AP (95% CI) |
|---|---|---|---|---|---|
| Exposure to Insect repellent | Presence of HLA-DRB1*15 | ||||
| − | − | 67/172 | reference | ||
| − | + | 44/45 | 2.92 (1.69 – 5.03)*** | ||
| + | − | 102/187 | 1.69 (1.12 – 2.53)* | ||
| + | + | 74/74 | 2.91 (1.82 – 4.66)*** | -0.69 (−2.47 – 1.09) | -0.23 (−0.88 – 0.40) |
| Exposure to Insect repellent | Absence of HLA-A*02 | ||||
| − | − | 35/73 | reference | ||
| − | + | 64/99 | 1.31 (0.76 – 2.24) | ||
| + | − | 64/92 | 1.50 (0.86 – 2.61) | ||
| + | + | 91/112 | 1.85 (1.09 – 3.12)* | 0.03 (−0.91 – 0.98) | 0.01 (−0.49 – 0.53) |
| Exposure to weed control products | Presence of HLA-DRB1*15 | ||||
| − | − | 109/280 | reference | ||
| − | + | 77/87 | 2.62 (1.74 – 3.93)*** | ||
| + | − | 61/94 | 2.42 (1.56 – 3.76)*** | ||
| + | + | 47/33 | 4.79 (2.72 – 8.43)*** | 0.74 (−1.89 – 3.38) | 0.15 (−0.32 – 0.64) |
| Exposure to weed control products | Absence of HLA-A*02 | ||||
| − | − | 74/130 | reference | ||
| − | + | 93/164 | 0.99 (0.65 – 1.49) | ||
| + | − | 32/37 | 1.80 (0.98 – 3.30) | ||
| + | + | 64/54 | 2.65 (1.60 – 4.39)*** | 0.86 (−0.54 – 2.26) | 0.32 (−0.13 – 0.78) |
| Exposure to weed control products | Presence of IL-6 SNP,GG |
||||
| − | − | 32/47 | reference | ||
| − | + | 129/229 | 0.77 (0.42 – 1.42) | ||
| + | − | 11/9 | 2.16 (0.69 – 6.74) | ||
| + | + | 84/81 | 1.84 (0.92 – 3.66) | -0.09 (−2.46 – 2.27) | -0.05 (−1.34 – 1.23) |
| Exposure to weed control products | Presence of BCL2 SNP, GG | ||||
| − | − | 63/104 | reference | ||
| + | − | 35/29 | 3.46 (1.78 – 6.71) | ||
| + | + | 57/60 | 2.36 (1.34 – 4.17)** | -1.38 (−3.64 – 0.87) | -0.58 (−1.59 – 0.42) |
| Exposure to weed control products | Presence of NFKB1 SNP, GG | ||||
| − | − | 114/199 | reference | ||
| − | + | 52/92 | 0.77 (0.48 – 1.21) | ||
| + | − | 62/69 | 1.86 (1.18 – 2.93)** | ||
| + | + | 34/21 | 3.32 (1.74 – 6.36)*** | 1.68 (−0.44 – 3.82) | 0.50 (0.13 – 0.88)** |
| Exposure to plant/tree insect control products | Presence of HLA-DRB1*15 | ||||
| − | − | 147/360 | reference | ||
| − | + | 110/115 | 2.33 (1.64 – 3.32)*** | ||
| + | − | 25/22 | 3.93 (2.01 – 7.71)*** | ||
| + | + | 19/13 | 5.08(2.23 – 11.57)*** | -0.19 (−4.95 – 4.57) | -0.03 (−0.99 – 0.92) |
| Exposure to plant/tree insect control products | Absence of HLA-A*02 | ||||
| − | − | 94/159 | reference | ||
| − | + | 130/212 | 1.03 (0.72 – 1.47) | ||
| + | − | 12/11 | 2.10 (0.81 – 5.42) | ||
| + | + | 26/13 | 4.93(2.23 – 10.88)*** | 2.79 (−1.34 – 6.94) | 0.56 (0.06 – 1.07) |
ref: reference group; RERI: Relative Excess Risk due to Interaction; AP: Attributable Proportion of disease; SI: Synergic Index; CI: Confidence Intervals
: P-value <0.05
: P-value <0.01
: P-value <0.00
Table 6.
Additive interaction between household chemical exposure and genotype sensitivity analysis: using inverse probability weighting). Stratified analysis assessing the interactions between household chemical exposure and presence of HLA-DRB1*15 alleles, absence of HLA-A*02, IL-6 SNP GG genotype, BCL2 SNP GG genotype and NFKB1 SNP GG genotype
| Exposureto Insect repellent | Presence of HLA-DRB1*15 | Case/ Control | OR (95% CI) | RERI (95% CI) | AP (95% CI) |
|---|---|---|---|---|---|
| − | − | 67/172 | reference | ||
| − | + | 44/45 | 2.94 (1.69 – 5.13)*** | ||
| + | − | 102/187 | 1.75 (1.15 – 2.65)* | ||
| + | + | 74/74 | 3.13 (1.93 – 5.05)*** | -0.56 (−2.40 – 1.27) | -0.18 (−0.79 – 0.43) |
| Exposure to Insect repellent | Absence of HLA-A*02 | ||||
| − | − | 35/73 | reference | ||
| − | + | 64/99 | 1.24 (0.71 – 2.17) | ||
| + | − | 64/92 | 1.48 (0.84 – 2.62) | ||
| + | + | 91/112 | 1.90 (1.10 – 3.27)* | 0.17 (−0.76 – 1.10) | 0.09 (−0.40 – 0.58) |
| Exposure to weed control products | Presence of HLA-DRB1*15 | ||||
| − | − | 109/280 | reference | ||
| − | + | 77/87 | 2.67 (1.77 – 4.02)*** | ||
| + | − | 61/94 | 2.50 (1.61 – 3.88)*** | ||
| + | + | 47/33 | 4.86 (2.72 – 8.67)*** | 0.68 (−2.03 – 3.40) | 0.14 (−0.35 – 0.63) |
| Exposure to weed control products | Absence of HLA-A*02 | ||||
| − | − | 74/130 | reference | ||
| − | + | 93/164 | 0.97 (0.63–1.47) | ||
| + | − | 32/37 | 1.76 (0.97–3.21) | ||
| + | + | 64/54 | 2.70 (1.61–4.51)*** | 0.96 (−0.43–2.36) | 0.35 (−0.07 – 0.79) |
| Exposure to weed control products | Presence of IL-6 SNP,GG |
||||
| − | − | 32/47 | reference | ||
| − | + | 129/229 | 0.76 (0.41–1.39) | ||
| + | − | 11/9 | 2.47 (0.73–8.30) | ||
| + | + | 84/81 | 1.83 (0.93–3.59) | -0.39 (−3.27 – 2.47) | -0.21 (−1.78 – 1.34) |
| Exposure to weed control products | Presence of BCL2 SNP, GG | ||||
| − | − | 63/104 | reference | ||
| − | + | 101/182 | 1.24 (0.78–1.97) | ||
| + | − | 35/29 | 3.40 (1.75–6.62)*** | ||
| + | + | 57/60 | 2.38 (1.34–4.22)** | -1.27 (−3.49 – 0.94) | -0.53 (−1.51 – 0.44) |
| Exposure to weed control products | Presence of NFKB1 SNP, GG | ||||
| − | − | 114/199 | reference | ||
| − | + | 52/92 | 0.75 (0.46 – 1.20) | ||
| + | − | 62/69 | 1.92 (1.21 – 3.05)** | ||
| + | + | 34/21 | 3.27 (1.74 – 6.13)*** | 1.60 (−0.44 – 3.64) | 0.48 (0.10 – 0.87)* |
| Exposure to plant/tree insect control products | Presence of HLA-DRB1*15 | ||||
| − | − | 147/360 | reference | ||
| − | + | 110/115 | 2.36 (1.65 – 3.36)*** | ||
| + | − | 25/22 | 4.06 (2.08 – 7.93)** | ||
| + | + | 19/13 | 5.33 (2.36 −12.02)*** | -0.09 (−5.03 – 4.84) | -0.02 (−0.95 – 0.92) |
| Exposure to plant/tree insect control products | Absence of HLA-A*02 | ||||
| − | − | 94/159 | reference | ||
| − | + | 130/212 | 1.02 (0.71 – 1.47) | ||
| + | − | 12/11 | 2.22 (0.83 – 5.98) | ||
| + | + | 26/13 | 5.08(2.36 – 10.94)*** | 2.82 (−1.41 – 7.06) | 0.56 (0.03 – 1.08) |
ref: reference group; RERI: Relative Excess Risk due to Interaction; AP: Attributable Proportion of disease; SI: Synergic Index; CI: Confidence Intervals
: P-value <0.05
: P-value <0.01
: P-value <0.0
We also detected an increased odds of exposure to insect repellent for ticks or mosquitos and the absence of HLA-A*02 (OR: 1.85, CI: 1.09–3.12), although the AP and RERI were not significant. Similarly, a significant increase in OR was observed for exposure to weed control products and the absence of HLA-A*02 (OR: 2.65, CI: 1.60–4.39), although the AP and RERI were not significant. There was no interaction of DRB1 with environmental exposures.
(Table 7) shows multiplicative interaction analysis results for toxic chemical exposures with genetic risk factors in various logistic regression models adjusted for age, sex, genetic ancestry, and mother’s education. There was evidence of a multiplicative interaction between exposure to weed control products and NFKB1 SNP GG genotype (OR: 2.30, CI: 1.00–5.30) but not for other exposures and risk variants. Sensitivity analyses using IPW resulted in similar results (Table 8).
Table 7.
Models assessing multiplicative interactions for odds of POMS
| Model | Interaction term | Exposure to toxic chemical, OR (95% CI) | Secondary risk factors (DRB1*15 +, A*2 -, IL-6 GG+, BCL2 GG+, NFKB1 GG+), OR (95% CI) | Interaction, OR (95% CI) |
|---|---|---|---|---|
| 1 | Exposure to Insect repellent for ticks or mosquitos X DRB1*15 | 1.69 (1.122.53)* | 2.92 (1.695.03)*** | 0.59(0.20 – 1.18) |
| 2 | Exposure to Insect repellent for ticks or mosquitos X A*02 | 1.50 (0.86–2.61) | 1.31 (0.76–2.24) | 0.93(0.46–1.88) |
| 3 | Exposure to weed control products X DRB1*15 | 2.42(1.56–3.76) | 2.62(1.74–3.93) | 0.75(0.36–1.54) |
| 4 | Exposure to weed control products X A*02 | 1.80 (0.98–3.30) | 0.85 (0.65–1.49) | 1.48(0.69–3.17) |
| 5 | Exposure to weed control products X IL-6 GG + | 2.16 (0.60–6.74) | 0.77 (0.42–1.42) | 1.09(0.33–3.65) |
| 6 | Exposure to weed control products X BCL2 GG + | 3.46(1.78–6.71) | 1.29 (0.82–2.02) | 0.52(0.23–1.16) |
| 7 | Exposure to weed control products X NFKB1 GG + | 1.86(1.18–2.93)** | 0.77 (0.48–1.21) | 2.30(1.00–5.30)* |
| 8 | Exposure to plant/tree insect or disease control products X DRB1*15 | 3.93(2.01–7.71) | 2.33 (1.63–3.32) | 0.55(0.18–1.60) |
| 9 | Exposure to plant/tree insect or disease control products X A*02 | 1.03 (0.72–1.47) | 2.10 (0.81–5.42) | 2.27(0.67–7.67) |
The logistic regression models were adjusted for age, sex, genetic ancestry, and mother’s education, recruiting site.
p-Value
Table 8.
Models assessing multiplicative interactions for odds of POMS (sensitivity analysis: using inverse probability weighting)
| Model | Interaction term | Exposure to toxic chemical, OR (95% CI) | Secondary risk factors (DRB1*15 +, A*2 -, IL-6 GG+, BCL2 GG+, NFKB1 GG+), OR (95% CI) |
Interaction, OR (95% CI) |
|---|---|---|---|---|
| 1 | Exposure to Insect repellent for ticks or mosquitos X DRB1*15 | 1.75 (1.15–2.65)* | 2.94 (1.69–5.11)*** | 0.60 (0.30–1.22) |
| 2 | Exposure to Insect repellent for ticks or mosquitos X A*02 | 1.48 (0.84–2.62) | 1.24 (0.71–2.17) | 1.02 (0.50–2.09) |
| 3 | Exposure to weed control products X DRB1*15 | 2.50 (1.61–3.88)*** | 2.67 (1.77–4.02)*** | 0.72 (0.35–1.49) |
| 4 | Exposure to weed control products X A*02 | 1.76 (0.97–3.21) | 0.96 (0.63–1.47) | 1.58 (0.74–3.36) |
| 5 | Exposure to weed control products X IL-6 GG + | 2.47 (0.73–8.30) | 0.76 (0.41–1.39) | 0.97 (0.27–3.46) |
| 6 | Exposure to weed control products X BCL2 GG + | 3.40 (1.75–6.62)*** | 1.24 (0.78–1.97) | 0.56 (0.25–1.24) |
| 7 | Exposure to weed control products X NFKB1 GG + | 1.92 (1.21–3.05) | 0.75 (0.46–1.20) | 2.26 (0.99–5.17) |
| 8 | Exposure to plant/tree insect or disease control products X DRB1*15 | 4.06 (2.08–7.93)*** | 2.36 (1.65–3.36)*** | 0.55 (0.19–1.60) |
| 9 | Exposure to plant/tree insect or disease control products X A*02 | 1.02 (0.71–1.47) | 2.22 (0.83–5.98) | 2.21 (0.64–7.65) |
The logistic regression models were adjusted for age, sex, genetic ancestry, and mother’s education, recruiting site.
p-Value
Discussion
In this study, we expanded on results from a previous study finding evidence of associations between household toxic chemical exposures and risk of POMS using GxE interactions (2). More specifically, we identified evidence of additive and multiplicative GxE interaction between MS risk variant NF-κB1 SNP GG genotype (rs7665090) and exposure to glyphosate.
The mechanisms through which toxic chemical exposure may contribute to MS onset remains are not completely understood. Exposure to weed control agents can alter immune response, potentially resulting in inflammatory diseases, including rheumatoid arthritis and systematic lupus erythematosus31,32. Glyphosate (RoundUp®) can induce the nuclear factor-kappa B (NF-κB) signaling pathway in several human cell types14,15,21 inducing gene expression of proinflammatory cytokines like TNF- α, IL-1, and IL-6. Activation of NF-κB in multiple cell types in the CNS and peripheral blood mononuclear cells of MS patients has been reported33. Furthermore, genome-wide studies have identified the contribution of NF-κB to MS risk34.
MS risk variant rs7665090G, located near NF-κB, has been associated with the upregulation of NF-κB in human astrocytes18. Here, we report evidence for both additive and multiplicative interactions between the NF-κB SNP and exposure to weed control products influencing POMS risk. The presence of additive and specifically multiplicative interaction suggests the potential role for synergically induced NF-κB signaling pathway through environmental exposure in genetically susceptible individuals.
Studies of GxE interactions in MS have been limited by the sample size and have mainly focused on the two strongest MS risk variants, HLA-DRB1*15 and the absence of HLA-A021. While the interaction between HLA-DRB1*15 and EBV, HSV, smoking, and adolescent obesity has been reported in whites, the interaction of HLA-DRB1*15 and EBV has also been reported in blacks1,28,35. HLA-DRB1*15 and the absence of HLA-A*02 may also interact with occupational exposure to inhaled organic solvents36. We did not find any additive or multiplicative interaction between HLA-DRB1*15 and the absence of HLA-A*02 and the studied environmental exposures. The gene-environment interactions in MS may be attributed to the various epigenetic factors that affect the downstream biological effects due to molecular modifications1. This could explain why we did not observe any interaction between our candidate environmental risk factors and HLA-DRB1*15 and the absence of HLA-A*02.
A significant strength of our study is the large POMS cohort, with cases and controls enrolled nationally at the same sites. We adjusted the analyses for multiple potential confounders, including genetic ancestry estimations, mother’s education as a proxy of SES, age, sex, and recruiting site. There were several limitations to our study. The questionnaire did not address the specific time, dose, and brand name of chemical exposures. We considered that all the participants had been exposed to the commonly used chemicals as we did not have the specific type of chemicals to which each subject was exposed. Furthermore, this is a retrospective study of exposures; however, focusing on pediatric rather than adult onset MS improves recall quality due to the much shorter time between exposure and outcome. Our results will need to be replicated in a larger study. We considered repeating similar analyses in the Ausimmune and the Swedish cohorts, but the age range, type of exposure, and questionnaires were too different.
Taken together, exposure to herbicides may be a risk factor for POMS, especially among individuals with genetic susceptibility to the disease. G×E interactions between weed control products and NF-κB SNP (rs7665090, GG genotype) suggest a possible role of NF-κB signaling pathway in susceptibility to POMS. To fill the gap in the underlying biological pathways, international collaborative studies with diverse genetic and environmental backgrounds can be helpful.
Key messages:
Identifying gene-environment interactions contributing to the risk of MS is critical in advancing the understanding of molecular processes at play.
We report a novel gene-environment interaction association with household toxic chemical exposures increasing the risk of pediatric-onset MS.
We identified both additive and multiplicative interactions with MS risk variant NF-κB1 SNP GG genotype (rs7665090) and likely exposure to glyphosate as one of the commonly used herbicides.
We take one step forward toward unanswered research questions for the concurrent role of genetic and environmental factors in susceptibility to MS
Disclosures/Acknowledgements:
Funding for this work includes R01NS071463 (PI Waubant), NMSS HC-1509-06233 (PI Casper), and MSIF (PI Nasr). None of the authors have any conflict of interest with this study. We thank Dr. Robyn Lucas for her assistance in considering duplication of the results in Aussies cohort. We thank the patients and families who participated in the study.
Footnotes
Potential Conflicts of Interest
E.W has current support from the NIH, NMSS, PCORI, CMSC, and Race to Erase MS.
References:
- 1.Waubant E, Lucas R, Mowry E, et al. Environmental and genetic risk factors for MS: an integrated review. Ann Clin Transl Neurol 2019;6:1905–22. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Mar S, Liang S, Waltz M, et al. Several household chemical exposures are associated with pediatric-onset multiple sclerosis. Ann Clin Transl Neurol 2018;5:1513–21. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Eriguchi M, Iida K, Ikeda S, et al. Parkinsonism Relating to Intoxication with Glyphosate. Intern Med 2019;58:1935–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Yan D, Zhang Y, Liu L, et al. Pesticide exposure and risk of Alzheimer’s disease: a systematic review and meta-analysis. Sci Rep 2016;6:32222. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.von Ehrenstein OS, Ling C, Cui X, et al. Prenatal and infant exposure to ambient pesticides and autism spectrum disorder in children: population based case-control study. BMJ 2019;364:l962. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Roberts JR, Dawley EH, Reigart JR. Children’s low-level pesticide exposure and associations with autism and ADHD: a review. Pediatr Res 2019;85:234–41. [DOI] [PubMed] [Google Scholar]
- 7.Madani FZ, Hafida M, Merzouk SA, et al. Hemostatic, inflammatory, and oxidative markers in pesticide user farmers. Biomarkers 2016;21:138–45. [DOI] [PubMed] [Google Scholar]
- 8.Ziaei A, Lavery AM, Shao XM, et al. Gene-environment interactions increase the risk of pediatric-onset multiple sclerosis associated with ozone pollution. Mult Scler 2022;28:1330–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Polman CH, Reingold SC, Banwell B, et al. Diagnostic criteria for multiple sclerosis: 2010 revisions to the McDonald criteria. Ann Neurol 2011;69:292–302. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Belman AL, Krupp LB, Olsen CS, et al. Characteristics of Children and Adolescents With Multiple Sclerosis. Pediatrics 2016;138:e20160120. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Gianfrancesco MA, Stridh P, Shao X, et al. Genetic risk factors for pediatric-onset multiple sclerosis. Mult Scler 2018;24:1825–34. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Patsopoulos NA, Baranzini SE, Santaniello A, et al. Multiple sclerosis genomic map implicates peripheral immune cells and microglia in susceptibility. Science 2019;365:eaav7188. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Woźniak E, Reszka E, Jabłońska E, et al. Glyphosate affects methylation in the promoter regions of selected tumor suppressors as well as expression of major cell cycle and apoptosis drivers in PBMCs (in vitro study). Toxicol In Vitro 2020;63:104736. [DOI] [PubMed] [Google Scholar]
- 14.Tang Q, Tang J, Ren X, et al. Glyphosate exposure induces inflammatory responses in the small intestine and alters gut microbial composition in rats. Environ Pollut 2020;261:114129. [DOI] [PubMed] [Google Scholar]
- 15.Martínez MA, Rodríguez JL, Lopez-Torres B, et al. Use of human neuroblastoma SH-SY5Y cells to evaluate glyphosate-induced effects on oxidative stress, neuronal development and cell death signaling pathways. Environ Int 2020;135:105414. [DOI] [PubMed] [Google Scholar]
- 16.CorteA, PuliL, LeJ, et al. Major histocompatibility complex associations of ankylosing spondylitis are complex and involve further epistasis with ERAP1. Nat Commun 2015;6:7146. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Han L, Lee CK, Pang H, et al. Genetic predisposition to lung adenocarcinoma among never-smoking Chinese with different epidermal growth factor receptor mutation status. Lung Cancer 2017;114:79–89. [DOI] [PubMed] [Google Scholar]
- 18.Ponath G, Lincoln MR, Levine-Ritterman M, et al. Enhanced astrocyte responses are driven by a genetic risk allele associated with multiple sclerosis. Nat Commun 2018;9:5337. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Chen CY, Pollack S, Hunter DJ, et al. Improved ancestry inference using weights from external reference panels. Bioinformatics 2013;29:1399–406. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Gui YX, Fan XN, Wang HM, et al. Glyphosate induced cell death through apoptotic and autophagic mechanisms. Neurotoxicol Teratol 2012;34:344–9. [DOI] [PubMed] [Google Scholar]
- 21.Zheng T, Jia R, Cao L, et al. Effects of chronic glyphosate exposure on antioxdative status, metabolism and immune response in tilapia (GIFT, Oreochromis niloticus). Comp Biochem Physiol C Toxicol Pharmacol 2021;239:108878. [DOI] [PubMed] [Google Scholar]
- 22.Gargouri B, Bhatia HS, Bouchard M, et al. Inflammatory and oxidative mechanisms potentiate bifenthrin-induced neurological alterations and anxiety-like behavior in adult rats. Toxicol Lett 2018;294:73–86. [DOI] [PubMed] [Google Scholar]
- 23.Ibrahim KA, Eleyan M, Khwanes SA, et al. Alpha-mangostin attenuates the apoptotic pathway of abamectin in the fetal rats’ brain by targeting pro-oxidant stimulus, catecholaminergic neurotransmitters, and transcriptional regulation of reelin and nestin. Drug Chem Toxicol 2022;45:2496–508. [DOI] [PubMed] [Google Scholar]
- 24.Mense SM, Sengupta A, Lan C, et al. The common insecticides cyfluthrin and chlorpyrifos alter the expression of a subset of genes with diverse functions in primary human astrocytes. Toxicol Sci 2006;93:125–35. [DOI] [PubMed] [Google Scholar]
- 25.Zhao MW, Yang P, Zhao LL Chlorpyrifos activates cell pyroptosis and increases susceptibility on oxidative stress-induced toxicity by miR-181/SIRT1/PGC-1α/Nrf2 signaling pathway in human neuroblastoma SH-SY5Y cells: Implication for association between chlorpyrifos and Parkinson’s disease. Environ Toxicol 2019;34:699–707. [DOI] [PubMed] [Google Scholar]
- 26.Chorfa A, Bétemps D, Morignat E, et al. Specific pesticide-dependent increases in α-synuclein levels in human neuroblastoma (SH-SY5Y) and melanoma (SK-MEL-2) cell lines. Toxicol Sci 2013;133:289–97. [DOI] [PubMed] [Google Scholar]
- 27.Wheeler MA, Jaronen M, Covacu R, et al. Environmental Control of Astrocyte Pathogenic Activities in CNS Inflammation. Cell 2019;176:581–596.e18. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Lavery AM, Collins BN, Waldman AT, et al. The contribution of secondhand tobacco smoke exposure to pediatric multiple sclerosis risk. Mult Scler 2019;25:515–22. [DOI] [PubMed] [Google Scholar]
- 29.Seaman SR, White IR. Review of inverse probability weighting for dealing with missing data. Stat Methods Med Res 2013;22:278–95. [DOI] [PubMed] [Google Scholar]
- 30.VanderWeele TJ and Knol MJ. A tutorial on interaction. Epidemiol Method 2014; 3(1): 33–72. [Google Scholar]
- 31.Parks CG, Costenbader KH, Long S, et al. Pesticide use and risk of systemic autoimmune diseases in the Agricultural Health Study. Environ Res 2022;209:112862. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Chittrakul J, Sapbamrer R, Sirikul W. Pesticide Exposure and Risk of Rheumatoid Arthritis: A Systematic Review and Meta-Analysis. Toxics 2022;10:207. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Zhou Y, Cui C, Ma X, et al. Nuclear Factor κB (NF-κB)-Mediated Inflammation in Multiple Sclerosis. Front Immunol 2020;11:391. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Beecham AH, Patsopoulos NA, Xifara DK, et al. Analysis of immune-related loci identifies 48 new susceptibility variants for multiple sclerosis. Nat Genet 2013;45:1353–60. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Xiao D, Ye X, Zhang N, et al. A meta-analysis of interaction between Epstein-Barr virus and HLA-DRB1*1501 on risk of multiple sclerosis. Sci Rep 2015;5:18083. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Barragán-Martínez C, Speck-Hernández CA, Montoya-Ortiz G, et al. Organic solvents as risk factor for autoimmune diseases: a systematic review and meta-analysis. PLoS One 2012;7:e51506. [DOI] [PMC free article] [PubMed] [Google Scholar]