Abstract
Background
Prostate cancer (PC) risk increases with African ancestry and a history of sexually transmitted infections (STIs). Also, SNPs in toll like receptor (TLR) genes influence PC risk. This pilot study explores interactions between STIs and TLR-related SNPs in relation to PC risk among Jamaican men.
Methods
This case-control study evaluates two TLR related SNPs in 356 Jamaican men (194 controls and 162 cases) with or without history of STIs using Stepwise Penalized Logistic Regression (StepPLR) in multivariable analyses.
Results
Age (OR = 1.08, 95% CI: 1.04- 1.12, p < 0.001) and IRF3_rs2304206 GG genotype (OR = 0.47, 95% CI: 0.29- 0.78, p = 0.003) modulated PC risk in people with history of STIs. In the population with no history of STIs, resulting interactions between risk factors did not survive correction for multiple hypothesis testing.
Conclusions
Overall, an interaction between the IFR3_rs2304206 variant and a history of exposure to STIs leads to greater decrease of PC risk than the presence of polymorphic genotype alone. These findings are suggestive and require further validation. Identification of gene variants along with detection of lifestyle behaviors may contribute to identification of men at a greater risk of PC development in the population.
Keywords: Prostate Cancer, Black, Single nucleotide polymorphisms, Sexually Transmitted Infections, Toll-like receptor (TLR), StepPLR
Introduction
Prostate cancer (PC) is the most common cancer among men and the second leading cause of death in the United States 1. While age, family history and race have been identified as significant risk factors for PC 2, recent evidence suggests an increased PC risk associated with prior exposure to sexually transmitted infections (STIs) 3,4. It is suggested that the association between PC risk and STIs may be related to the exposure to pathogens, which causes chronic inflammation within the prostate and ultimately leads to enhanced cell proliferation and carcinogenesis 5,6. In response to exposure to pathogens, the innate immune system serves as the first line of defense. Specifically, a conserved family of pattern recognition receptors (PRRs), called the Toll-like receptor (TLR) family, induces inflammatory cytokine and chemokine genes leading to microbial elimination 7. A more recent report has demonstrated TLRs, specifically TLR subclass 4 (TLR4) and TLR9, may contribute to prostate cancer pathogenesis by stimulating prostate epithelial cell proliferation in response to pathogens 8. Overall, TLR signaling cascades stimulate the expression of inflammatory chemokines, cytokines and interferons, resulting in local inflammation. Genetic variations in these genes may lead to an imbalanced immune response (over exuberant or inadequate), resulting in excessive and harmful inflammation. Several TLR1-TLR4-TLR6-lL10 sequence variants are linked to an increase 9–11 and decrease 12 in PC risk. In addition, several epidemiological studies have found an increased risk of PC in association with history of prostatitis and STIs 13–15. Other aspects of sexual behavior (e.g., number of sexual partners, condom usage, and age at first sex) have also been associated with PC risk 3, as sexual activity may be a marker for higher risk of exposure to STIs which in turn have been linked to PC risk.
As men of African descent are at increased risk for inflammatory diseases 16, genetic variations that lead to increased inflammatory responses in the presence of infectious agents may play an important role in disposing them to PC risk. Since TLRs play an important role in launching an inflammatory response against STIs, TLR associated genetic susceptibility modifies PC risk among men of African descent 17,18. The current study seeks to evaluate interactions between TLR gene variants and STI exposure, in relation to PC risk among Jamaican men. Besides, very few address impact of TLR gene variants or exposure of STIs on PC risk among men of African descent even though the disparity in disease incidence is well known. In our previous study on TLR associated SNPs and PC risk among US black men and Jamaican men, TLR6_rs2381289 and IRF3_rs2304206 were reported to be significantly associated with PC risk in Jamaican men 18. However, these associations did not remain statistically significant after adjustment of multiple testing. Hence, we selected these two genetic markers from our previous study to further evaluate their role in modulating PC risk in Jamaican men with or without the history of STIs and using improved multivariable models with stepwise method based on penalized logistic regression.
Methods
Subjects
The current study included 356 DNA samples (163 cases and 195 controls) from a larger study of 515 Black men (243 cases and 275 controls). Details of the study population (case and control ascertainment, inclusion criteria) have been previously published elsewhere 19. Briefly, patients were recruited from urology clinics at the two main tertiary hospitals and from private practitioners in the Kingston Metropolitan area in Jamaica. Based on clinical practice in Jamaica, the prevalence of prostate cancer with a PSA of 3ng/ml is as high as 15%. A single standard ultrasound-guided biopsy was performed on men attending the clinics for the first time during the recruitment period with abnormal DRE and PSA > 3.0 μg/l. From the biopsy only, histologically-confirmed prostate cancer was graded according to the Gleason scoring system by a local uropathologist. Men who were biopsied and were prostate cancer free were excluded from study (i.e. were not used as controls). No biopsies were performed on the controls. Men on 5-alpha reductase inhibitors were also excluded from the study. Ethics approval was obtained from the University of the West Indies, Mona and Fox Chase Cancer Center Institutional Review Boards. Study participants provided written informed consent to participate in the study.
Genotyping
TLR6_rs2381289 and IRF3_rs2304206 sequence variants were analyzed in this study. Of the 356 samples used the in analyses, genotype data of 216 samples was used as part of a collaborative project that is previously published [18]. Additional 141 samples were genotyped independently for this study. We randomly included several samples from the previously genotyped batch (216 subjects) in our new genotyping batch to cross check any variation which may arise due to change in platforms (batch effect). We obtained same genotypes for the random samples as was reported earlier, verifying that the change of genotyping platform did not lead to any variations in the results. Hence, the data from 216 subjects was merged with recently genotyped 141 subjects to results in a single dataset from 356 subjects. Quality controls, data management of genotype data, testing for Hardy-Weinberg Equilibrium (HWE) are documented elsewhere 18,20. An additional 141 individuals were genotyped for this study using TaqMan SNP genotyping assay. The experiments were performed using Applied Biosystems Step One Plus Real-Time PCR system (ThermoFisher Scientific/Life Technologies, Waltham, MA, USA). The genotype calls were evaluated and analyzed via StepOne Software, version 2.3. SNPs were tested for HWE after addition of new genotype data.
Epidemiologic and Anthropometric Data
A standardized interview administered questionnaire permitted collection of demographic, medical history, dietary habits, lifestyle factors and sexual behavior data. The data on physical activity was collected qualitatively using (i) leisure physical activity and (i) nature of work. Participants were asked if they engaged in leisure physical activity (exercise or walking) for at least 20 minutes that makes them breathe heavier and heart beat faster (responses were no physical activity, fewer than once per week, one or two times per week or at least three or more times a week). Participants were also asked to describe the nature of their work i.e. mainly sitting, not much walking, a lot of walking, but not lifting or carrying heavy things, a lot of walking, often carrying things, or climbing stairs or going uphill and heavy physical work, such as carrying or lifting heavy things, digging and shovelling. Therefore, in our analysis using nature of work and measure of physical activity we categorized as follows: Inactive (no physical activity/ mainly sitting work, not much walking), moderately inactive (fewer than once per week/quite a lot of walking, but not lifting or carrying heavy things), moderately active (one or 2 times per week/lot of walking, often carrying things, or climbing stairs or going uphill), and active (walking at least >3 times a week/heavy physical work, carrying or lifting heavy things, digging, shovelling). Smoking status (yes/no) was taken into account for both current and ex-smokers. Behavioral information included age at sexual intercourse initiation, lifetime sexual partners (number of partners during lifetime and within the past year), personal history of sexually transmitted infections such as syphilis, gonorrhea, and genital sores (yes/no), prostitution (ever had sex with a commercial sex worker or prostitute) and condom use (frequency of use with regular partner and non-regular partners). Prostitution was a measure of whether a study participant ever had sex with a commercial sex worker or prostitute.
Statistical Analyses
Differences in demographic study participant characteristics between cases and controls were assessed using Chi squared statistics for categorical variables and the Wilcoxon Rank-Sum Test for continuous variables using the STATASE (13.0), StataCorp, College Station, TX 21.
The association between PC risk and Toll like receptor SNPs, expressed as odds ratios (ORs) and corresponding 95% confidence intervals (CIs), were estimated using logistic regression (LR) models in STATA SE (13.0), both with and without adjustment for age. The analyses were run separately after stratifying for history of STIs. LR analyses for genetic variants and PC risk were conducted under co-dominant and additive genetic models using major/common genotype as the reference category.
Further, the interactions among genetic variants and risk factors were tested using multivariable models with stepwise method based on penalized logistic regression (StepPLR), as implemented in the StepPLR package in R 22,23. Penalized LR stabilizes the coefficients in a LR model, by adding a penalty term to the likelihood function which results in shrunken coefficients. This method allows for efficient handling of multi-collinearity and sparse cells in multifactor contingency tables, which are typical problems encountered by regular LR when considering higher-order interactions. The regularization parameter λ 22 was chosen by running cross validation (CV) analyses with different values of λ using the StepPLR package, and the value of λ (=1) corresponding to the lowest value of AIC (Akaike Information Criterion) was selected. In all multivariable models the following variables were included: age (as continuous variable), smoking status (yes/no), physical activity (inactive/active), number of sexual partners in a month (none, 1-3/month, >4 /month), and use of condom with non-regular partners (never/always). StepPLR was performed for the total population as well as the study set after stratification by history of STIs. All p-values (except in Table 1) were subjected to multiple hypothesis testing using the Benjamini-Hochberg (BH) false discovery rate (FDR) 24 and statistical significance was determined using FDR<0.05.
Table 1:
Characteristics of study population by case/control status.
| Characteristics | Cases | Controls | p value |
|---|---|---|---|
| Number of Participants | 163 | 195 | |
| Age (years), Median (range) | 69 (49-80) | 63 (40-80) | <0.0001a |
| BMI (kg/m2), Mean (S.D) | 25.28 (4.41) | 25.01 (4.06) | 0.2782 |
| PSA (ng/ml), Median (range) | 33.75 (2.1-10000) | 1.8 (0.1-9.9) | <0.0001a |
| Smoking Status, n (%) | 0.007 | ||
| No | 41 (25.31) | 74 (38.95) | |
| Yes | 121 (74.69) | 116 (61.05) | |
| Family History of PC, n (%) | 0.027 | ||
| No | 132 (80.98) | 174(89.23) | |
| Yes | 31(19.02) | 21 (10.77) | |
| Physical Activity, n (%) | |||
| Inactive | 31 (19.14) | 41 (21.35) | 0.028 |
| Moderately Inactive | 32 (19.75) | 52 (27.08) | |
| Moderately Active | 62 (38.27) | 77(40.10) | |
| Active | 37 (22.84) | 22 (11.46) | |
| Education, n (%) | 0.016 | ||
| Primary or less | 147 (91.3) | 155 (81.15) | |
| Secondary | 8 (4.97) | 26 (13.61) | |
| Tertiary | 6 (3.73) | 10 (5.24) | |
| Marital Status, n (%) | 0.281 | ||
| Married | 100 (61.73) | 125 (65.10) | |
| Previously Married | 32 (19.75) | 25 (13.02) | |
| Others | 30 (18.52) | 42 (21.35) | |
| Prostitution (N=440), n(%) | 0.778 | ||
| no | 126 (82.35) | 147 (83.52) | |
| yes | 27(17.65) | 29 (16.48) | |
| Number of sexual partners (N=447), n(%) | 0.103 | ||
| none | 72 (46.15) | 67 (37.43) | |
| 1-3/month | 49 (31.41) | 54 (30.17) | |
| > 4/month | 35 (22.44) | 58 (32.4) | |
| Condom use with regular partners | 0.468 | ||
| never | 117 (73.13) | 132 (68.75) | |
| always | 43 (26.88) | 59 (30.73) | |
| Condom use with non regular partners | 0.223 | ||
| never | 86 (53.42) | 83 (43.23) | |
| always | 74 (45.96) | 106 (55.21) | |
| STI history (N=499) n (%) | 0.115 | ||
| no | 59 (36.65) | 85 (44.97) | |
| yes | 102 (63.35) | 104 (55.03) | |
| Gleason Score, n (%) | |||
| 3 | 1 (0.61) | ||
| 6 | 65 (39.88) | ||
| 7 | 56 (34.36) | ||
| 8 | 18 (11.04) | ||
| 9 | 16 (9.82) | ||
| 10 | 2 (1.23) | ||
| missing | 5 (3.07) |
Wilcoxon Sum Rank Test
Results
Study population characteristics:
The characteristics for the study population in this analysis (163 cases and 195 controls) are summarized in Table 1. PC cases were older (p < 0.0001), had significantly higher PSA levels (p < 0.0001) and were more likely to have a family history of PC (p = 0.027) than controls. Controls reported higher attainment in secondary and higher education, were less physically active and were more likely to be reported as non-smokers (p = 0.007) than cases. Cases and controls showed similar mean BMI (p =0.2782).. With respect to sexual behavior, usage of condoms with regular (p=0.468) and non-regular partners (p=0.223) was not significantly different between cases and controls. Prostitution, history of STIs, number of sexual partners and marital status were not found to be significantly different between cases and controls.
The minor allele frequency for TLR-associated SNPs, TLR6_ rs2381289 and IRF3_rs2304206, among Jamaican men was found to be 0.23 and 0.32, respectively. Both SNPs followed HWE before and after the addition of samples in this study.
Association between TLR-associated sequence variants and PCA risk:
IRF3_rs2304206 and TLR6_ rs2381289 were associated with PC risk for unadjusted and age-adjusted logistic regression models as shown in Table 2. In particular, the presence of IRF3_rs2304206 GG genotype was associated with a ~60% reduction in PC risk (OR = 0.40, CI: 0.19-0.81, p = 0.012). This association was true for age-adjusted models as well (OR = 0.36, CI: 0.17-0.76, p =0.008). Also presence of TLR6_ rs2381289 GA genotype was associated with 1.7-1.9 fold increase in PC risk under unadjusted (p = 0.014, CI: 1.11-2.67) and age-adjusted (p = 0.009, CI: 1.16-2.96) LR models. Under the additive genetic model the IRF3_rs2304206 sequence variant allele was associated with reduced (~ 70%) PC risk (p = 0.027, CI: 0.52-0.96); whereas, TLR6_ rs2381289 did not remain significant under this model. After stratification of the study population by history of STIs (Table 3), IRF3_rs2304206 (GG) was found to be significant in the population with a history of STI (p= 0.008, CI: 0.10-0.76) and had lower ORs (0.24) when compared to the total population. However, the inheritance of TLR6_ rs2381289 (GA) genotype was associated with elevated PC risk (OR=1.85) but this association did not survive multiple-testing correction. Hence, further interaction analyses were carried out using IRF3_rs2304206 to test gene-environment interactions using StepPLR.
Table 2:
TLR SNPs and their association with Prostate cancer risk in total population
| Total Population | |||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| SNP | Genotypes | Cases | Controls | OR Unadjusted | CI (95%) | p Unadjusted | FDR | OR Age-Adjusted | CI (95%) | p Age-Adjusted | FDR |
| Irf3_rs2304206 | AA | 85 | 87 | 1 (ref.) | 1 (ref.) | ||||||
| AG | 65 | 75 | 0.88 | 0.56-1.38 | 0.599 | 0.599 | 0.87 | 0.54-1.40 | 0.579 | 0.579 | |
| GG | 13 | 33 | 0.40 | 0.19-0.81 | 0.012 | 0.036 | 0.36 | 0.17-0.76 | 0.008 | 0.024 | |
| GG+AG | 78 | 108 | 0.70 | 0.52-0.96 | 0.027 | 0.041 | 0.67 | 0.49-0.93 | 0.018 | 0.027 | |
| TLR6_rs2381289 | GG | 82 | 121 | 1 (ref.) | 1 (ref.) | ||||||
| GA | 75 | 64 | 1.72 | 1.11-2.67 | 0.014 | 0.042 | 1.86 | 1.16-2.96 | 0.009 | 0.027 | |
| AA | 5 | 10 | 0.73 | 0.24-2.23 | 0.591 | 0.591 | 0.78 | 0.24-2.49 | 0.681 | 0.681 | |
| AA+GA | 80 | 74 | 1.32 | 0.92-1.90 | 0.12 | 0.18 | 1.39 | 0.95-2.05 | 0.088 | 0.132 | |
Statistically significant associations were determined using FDR<0.05 , ref. = reference
Allele counts for Irf3_rs2304206 in total population: A: Cases-235, Controls-249; G: Cases-91, Controls-141
Allele counts for TLR6_rs2381289 in total population: A: Cases-85, Controls-84; G: Cases-239, Controls-306
Table 3:
TLR SNPs and their association with Prostate cancer risk in population stratified by history of STIs
| Pop with STI hist. | Pop with No STI hist. | ||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| SNP | Genotypes | Cases | Controls | OR | CI (95%) | p | FDR | OR | Cases | Controls | CI (95%) | p | FDR |
| Irf3_rs2304206 | AA | 54 | 48 | 1 (ref.) | 1 (ref.) | 30 | 39 | ||||||
| AG | 42 | 37 | 1.03 | 0.56-1.81 | 0.925 | 0.925 | 0.89 | 22 | 32 | 0.43-1.83 | 0.76 | 0.434 | |
| GG | 6 | 19 | 0.24 | 0.10-0.76 | 0.008 | 0.024 | 0.65 | 7 | 14 | 0.23-1.81 | 0.41 | 0.434 | |
| GG+AG | 48 | 56 | 0.63 | 0.41-0.97 | 0.037 | 0.055 | 0.78 | 29 | 46 | 0.47-1.28 | 0.33 | 0.434 | |
| LR6_rs2381289 | GG | 45 | 61 | 1 (ref.) | 1 (ref.) | 36 | 55 | ||||||
| GA | 52 | 38 | 1.85 | 1.05-3.27 | 0.033 | 0.099 | 1.4 | 23 | 25 | 0.69-2.84 | 0.34 | 0.51 | |
| AA | 4 | 5 | 1.08 | 0.27-4.2 | 0.908 | 0.908 | 0.3 | 1 | 5 | 0.34-2.72 | 0.28 | 0.51 | |
| AA+GA | 56 | 43 | 1.49 | 0.92-2.39 | 0.11 | 0.165 | 1.1 | 24 | 30 | 0.56-1.81 | 0.75 | 0.75 | |
Statistically significant associations were determined using FDR<0.05 , ref. = reference
Allele counts for Irf3_rs2304206 in population with a history of STI: A: Cases-150, Controls-133; G: Cases-54, Controls-75
Allele counts for Irf3_rs2304206 in population without a history of STI: A: Cases-82, Controls-110; G: Cases-36, Controls-60
Allele counts for TLR6_rs2381289 in population with a history of STI: A: Cases-60, Controls-48; G: Cases-142, Controls-160
Allele counts for TLR6_rs2381289 in population without a history of STI: A: Cases-25, Controls-35; G: Cases-95, Controls-135
StepPLR Analyses:
The results of multivariable models with age, smoking status, physical activity, condom usage, number of sexual partners and genetic marker IRF3_rs2304206, analyzed using StepPLR for the total population as well as the populations stratified by history of STIs are summarized in Table 4 & 5. This method produced results in terms of main effects and interactive effects of tested SNP locus and other factors in association with PC risk. In the total population, after BH correction, age (OR = 1.04, 95% CI: 1.02-1.06, FDR = 0.005), genetic variant IRF3_rs2304206 (GG) (OR = 0.58, 95% CI: 0.38-0.87, FDR = 0.027) and the interaction of physical activity with age (OR = 1.03, 95% CI: 1.01-1.04, FDR = 0.005) were found to be associated with PC risk. Among individuals with a history of STIs, age (OR=1.08, 95% CI: 1.04-1.12, FDR = 0.012) and possession of the IRF3_rs2304206 (GG) genotype (OR= 0.47, 95% CI: 0.29-0.78, FDR = 0.018) were associated with PC risk. In the population with no history of STIs (Supplemental Table 1), a number of interactions were significantly associated with PC risk, however, these were no longer significant after adjusting for multiple testing.
Table 4:
Association and Interactions between SNP Irf3_rs2304206/Life style factors and prostate cancer risk in population with history of STIs and total population
| Population size | Main Effects/Gene | Genotype | Cases | Controls | Interactions | OR | CI (95%) | p | FDR |
|---|---|---|---|---|---|---|---|---|---|
| +Total population | |||||||||
| Age | 1.04 | 1.02-1.06 | <0.001 | 0.005 | |||||
| Irf3_rs2304206 | AA | 85 | 87 | 1.42 | 1.05-1.94 | 0.025 | 0.052 | ||
| Irf3_rs2304206 | AG | 65 | 75 | 1.22 | 0.88-1.68 | 0.227 | 0.284 | ||
| Irf3_rs2304206 | GG | 13 | 33 | 0.58 | 0.38-0.87 | 0.008 | 0.027 | ||
| Physically active: Age | 1.03 | 1.01-1.04 | <0.001 | 0.005 | |||||
| Physically inactive: Age | 1.01 | 1-1.03 | 0.02 | 0.055 | |||||
| No Smoke | 0.82 | 0.65-1.05 | 0.116 | 0.166 | |||||
| Smoke | 1.21 | 0.95-1.55 | 0.116 | 0.166 | |||||
| Physically active | 0.83 | 0.55-1.26 | 0.384 | 0.384 | |||||
| Physically inactive | 1.2 | 0.79-1.83 | 0.384 | 0.384 | |||||
| #Population with History of STIs | |||||||||
| Age | 1.08 | 1.04-1.12 | <0.001 | 0.012 | |||||
| Irf3_rs2304206 | AA | 54 | 48 | 1.57 | 1.07-2.31 | 0.023 | 0.069 | ||
| Irf3_rs2304206 | AG | 42 | 37 | 1.35 | 0.9-2.01 | 0.145 | 0.29 | ||
| Irf3_rs2304206 | GG | 06 | 19 | 0.47 | 0.29-0.78 | 0.003 | 0.018 | ||
| Physically active | 1.07 | 0.85-1.34 | 0.562 | 0.613 | |||||
| Physically inactive | 0.94 | 0.75-1.17 | 0.562 | 0.613 | |||||
| NSP.none:Physically active | 1.06 | 0.67-1.66 | 0.803 | 0.803 | |||||
| NSP.none:Physically inactive | 0.67 | 0.41-1.11 | 0.117 | 0.281 | |||||
| NSPI.1-3/m:Physically active | 1.29 | 0.79-2.1 | 0.307 | 0.432 | |||||
| NSP.1-3/m:Physically inactive | 0.71 | 0.43-1.17 | 0.176 | 0.302 | |||||
| NSP.>4/m:Physically active | 0.78 | 0.48-1.27 | 0.324 | 0.432 | |||||
| NSP.>4/m:Physically inactive | 1.98 | 1.16-3.38 | 0.013 | 0.052 | |||||
Statistically significant associations were determined using FDR<0.05
Allele counts for Irf3_rs2304206 in total population: A:- Cases: 235, Controls: 249; G:- Cases: 91, Controls: 141
Allele counts for Irf3_rs2304206 in population with history of STIs: A:- Cases: 150; Controls: 133; G:- Cases: 54, Controls: 75
Discussion
TLRs play critical roles in the reproductive tract by signaling the presence of infection or other inflammatory stimuli 25, and prostatic inflammation/response to infection may have a direct effect on the onset of prostate cancer. This is the first study to evaluate the effect of interaction between STIs and TLR SNPs on PC risk among men of African descent. StepPLR analyses resulted in many gene-environment interactions; however, the discussion will be limited to only statistically significant results.
In a hospital-based case-control study of Jamaican black men we observed that tested SNPs in TLR6 and IRF3 genes may contribute to PC risk and this association is modified by previous exposure to STIs status. Under univariate unadjusted LR models, possession of IRF3_rs2304206 GG genotype was inversely associated (OR= 0.40, CI: 0.19-0.81) with PC risk whereas TLR6_ rs2381289 GA genotype increased the risk of developing PC (OR=1.7, CI: 1.11-2.67). To further evaluate the effect of STIs exposure on PC risk the population was stratified by history of ever/never exposure to STIs. Among individuals exposed to STIs, the TLR6_ rs2381289 variant did not remain significant after multiple testing corrections. Hence, we proceeded with IRF3_rs2304206 for further multivariable LR analysis using StepPLR method where we found older age and the IRF3_rs2304206 GG genotype were individually linked to PC risk in both the overall population and in the stratified population with history of STIs exposure. Decreased risk towards PC upon possession of variant allele IRF3_rs2304206 indicates disease protecting nature of the polymorphism in Jamaican men. IFN regulatory factor 3 (IRF3) is a transcriptional factor that plays a crucial role in the activation of innate immunity and inflammation in response to microbial infection and also has been suggested as a potential tumor suppressor gene 26. SNP rs2304206 has been reported as a significant expression quantitative trait loci eQTL [http://www.gtexportal.org/], which means the alternative alleles of rs2304206 are associated with different levels of IRF3 expression. In addition, rs2304206 falls in the regulatory region of the gene 27 and genotype AA has been linked to lower IRF3 expression in blood cells among the Japanese subgroup 28. Inverse relationship with PC in our study puts forward GG genotype of the SNP may play a role in the expression of IRF3 in a way that it aids/ retains tumor suppressing nature. Further, a study looking into the role of IRF3 in glioma invasive properties suggested that in vitro adenovirus mediated IRF3 activation resulted in upregulation of IRF3 thus inhibiting glioma proliferation, migration and invasion 29. This study further concluded that agents that promote IRF3 activation and expression (adenovirus in this case) may serve as potential target therapy against glioma. In our study, we see a statistically significant decrease in the ORs from 0.58 (in the total population, p=0.008, FDR=0.027) to 0.47 (in the sub-population with STI exposure, p=0.003, FDR=0.018) which may suggest (on the basis of the above study) that, in addition to the possession of variant genotype, microbial agents from STIs exposure activate IRF3 gene assisting its tumor suppressing behavior. Additionally, the protective response of IRF3 against viral infections has been shown in other studies, which further supports this account 30,31. In a previous study, after adjustment for multiple hypothesis testing, we did not observe a statistically significant protective association of IFR3_rs2304206 with prostate cancer18. However in this study we now report that a combination of the polymorphic genotype of IFR3_rs2304206 and a history of exposure to STIs provides added shielding from PC as compared to possession of polymorphic genotype alone. This association remained statistically significant after adjustment for multiple hypothesis testing.
In addition to age and variant genotype of IRF3 gene, interaction between age and physical activity was observed as a factor modestly associated with PC risk in the total population. Although physical activity is recommended as primary prevention in PC, studies have reported an inverse relationship between the level of physical activity and PC risk 32,33. Physical activity can have different influences on carcinogenesis, depending on energy supply and age of the subject. This notion is indirectly supported by studies that show intense physical activity may activate cellular stress signaling, accumulation of large amounts of reactive oxygen species (ROS) which can lead to DNA damage, or may support tumorigenesis by triggering pro-inflammatory signaling 34.
LR is the most commonly used tool to find correlation between risk factors and disease risk, but in the case of SNP data with 2 or 3 levels of genotype factors along with other confounding variables, problems of overfitting models may arise especially if data size is small 22. PLR is well suited for modeling a large number of variables and sample size does not limit the number of such variables 35. We realize sufficiently powered studies of interactions need large sample sizes. Hence, step-wise PLR approach was used as quadratic penalization yields a stable fit, even with a large number of parameters and small data size, thus making it possible to use LR in building interaction models. StepPLR has been demonstrated to have excellent power to identify covariate effects as well as superior prediction accuracy compared to competing methods like Classification and Regression Tree Analysis (CART) and Multifactor Dimensionality Reduction (MDR) 35. These results were further filtered through multiple hypothesis testing procedures. Hence the overall conclusion of a protective effect of IRF3_rs2304206 on PC remains unchanged in Jamaican men after addition of ~140 new samples to the study.
It is interesting to note that the Jamaica Health and Lifestyle Study II (JHLS II) which is a survey of a nationally representative sample in Jamaica, reported a STI prevalence in men of 18.2% (95% CI, 15.4-21.3%) 36. The prevalence of a history of STIs in our study was significantly higher than what has been reported in Jamaica and may reflect a bias due to recruitment of the study sample from the hospital and urologist practices. One of our co-authors who evaluates patients in the urology clinic reports that the five most common problems seen in urology clinics are prostate cancer, BPH, stones, urinary tract infections and bladder cancer. They also treat patients with urethral stricture disease which is causally related to STIs (gonococcal and chlamydial urethritis). Therefore, we were not surprised to see a higher prevalence of STI among our controls than in the general population. Furthermore, based on our data on education attainment, there is lower representation of men from higher SES categories (i.e. men with tertiary education) which may also account for this high prevalence of STI infection. Nevertheless, our analyses also speculate the role of STIs in addition to genetic polymorphisms in the susceptibility to PC by modulating immune response. Identification of gene variants as the inherited factor of PC risk along with detection of lifestyle behaviors may contribute to development of better diagnostic strategies and accurate identification of men in the population at a greater risk of PC development. We plan to take the results of this exploratory analysis further using molecular biological studies and bioinformatics analyses to confirm the role of IRF3_ rs2304206 as an eQTL with respect to PC and STIs which could be explored for both PC therapy and identifying target groups at higher risk. In addition, it would be important to investigate whether our study findings are unique to Jamaican men or also observed in non-Jamaican men of African descent or European men. For example, the evaluation of whether innate immune sequence variants combined with sexually transmitted infections may modify prostate cancer could be conducted by leveraging on existing studies, such as the Cancer Genetic Epidemiology Markers (CGEMS) and the Prostate, Lung, Colorectal and Ovarian (PLCO) Cancer Screening Trial (following the necessary approvals). Studies involving Jamaican and non-Jamaican men can also be performed by leveraging on existing collaborative networks within our consortium, the African Caribbean Cancer Consortium and also through a strong alliance with the Prostate Cancer Transatlantic Consortium. These comparative studies are in the planning phase.
Supplementary Material
Acknowledgements
This work was supported in part by the following: National Health Fund [HSF19], CHASE Fund, the Planning Institute of Jamaica to M.J., the National Cancer Institute (R13 CA130596) to C.R. and in part by grant numbers CA138017 and CA006927 from the National Cancer Institute and an appropriation from the Commonwealth of Pennsylvania to Fox Chase Cancer Center, Clinical Translational Science Pilot Grant to LRK; the JGBCC Bucks for Brains “Our Highest Potential” in Cancer Research Endowment to LRK and P20-MD000175, U54 MD012392 NIH NIMHD to KSK.
Footnotes
CONFLICT OF INTEREST: The authors declare no conflict of interest.
References
- 1.Society AC. Cancer Facts & Figures 2019. American Cancer Society; 2019. [Google Scholar]
- 2.Gann PH. Risk factors for prostate cancer. Rev Urol. 2002;4 Suppl 5:S3–S10. [PMC free article] [PubMed] [Google Scholar]
- 3.Sarma AV, McLaughlin JC, Wallner LP, et al. Sexual behavior, sexually transmitted diseases and prostatitis: the risk of prostate cancer in black men. The Journal of urology. 2006;176(3):1108–1113. [DOI] [PubMed] [Google Scholar]
- 4.Taylor ML, Mainous AG 3rd, Wells BJ. Prostate cancer and sexually transmitted diseases: a meta-analysis. Family medicine. 2005;37(7):506–512. [PubMed] [Google Scholar]
- 5.Coussens LM, Werb Z. Inflammation and cancer. Nature. 2002;420(6917):860–867. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.De Marzo AM, Platz EA, Sutcliffe S, et al. Inflammation in prostate carcinogenesis. Nature reviews Cancer. 2007;7(4):256–269. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Sonnex C Toll-like receptors and genital tract infection. International journal of STD & AIDS. 2010;21(3):153–157. [DOI] [PubMed] [Google Scholar]
- 8.Kundu SD, Lee C, Billips BK, et al. The toll-like receptor pathway: a novel mechanism of infection-induced carcinogenesis of prostate epithelial cells. The Prostate. 2008;68(2):223–229. [DOI] [PubMed] [Google Scholar]
- 9.Cheng I, Plummer SJ, Casey G, Witte JS. Toll-like receptor 4 genetic variation and advanced prostate cancer risk. Cancer epidemiology, biomarkers & prevention : a publication of the American Association for Cancer Research, cosponsored by the American Society of Preventive Oncology. 2007;16(2):352–355. [DOI] [PubMed] [Google Scholar]
- 10.Song J, Kim DY, Kim CS, et al. The association between Toll-like receptor 4 (TLR4) polymorphisms and the risk of prostate cancer in Korean men. Cancer genetics and cytogenetics. 2009;190(2):88–92. [DOI] [PubMed] [Google Scholar]
- 11.Wang MH, Helzlsouer KJ, Smith MW, et al. Association of IL10 and other immune response- and obesity-related genes with prostate cancer in CLUE II. The Prostate. 2009;69(8):874–885. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Chen YC, Giovannucci E, Lazarus R, Kraft P, Ketkar S, Hunter DJ. Sequence variants of Toll-like receptor 4 and susceptibility to prostate cancer. Cancer research. 2005;65(24):11771–11778. [DOI] [PubMed] [Google Scholar]
- 13.Dennis LK, Lynch CF, Torner JC. Epidemiologic association between prostatitis and prostate cancer. Urology. 2002;60(1):78–83. [DOI] [PubMed] [Google Scholar]
- 14.Sutcliffe S, Giovannucci E, De Marzo AM, Leitzmann MF, Willett WC, Platz EA. Gonorrhea, syphilis, clinical prostatitis, and the risk of prostate cancer. Cancer epidemiology, biomarkers & prevention : a publication of the American Association for Cancer Research, cosponsored by the American Society of Preventive Oncology. 2006;15(11):2160–2166. [DOI] [PubMed] [Google Scholar]
- 15.Huang WY, Hayes R, Pfeiffer R, et al. Sexually transmissible infections and prostate cancer risk. Cancer epidemiology, biomarkers & prevention : a publication of the American Association for Cancer Research, cosponsored by the American Society of Preventive Oncology. 2008;17(9):2374–2381. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Taylor BD, Darville T, Ferrell RE, Ness RB, Haggerty CL. Racial variation in toll-like receptor variants among women with pelvic inflammatory disease. The Journal of infectious diseases. 2013;207(6):940–946. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Tindall EA, Hayes VM, Petersen DC. Inflammatory genetic markers of prostate cancer risk. Cancers. 2010;2(2):1198–1220. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Rogers EN, Jones DZ, Kidd NC, et al. Toll-like receptor-associated sequence variants and prostate cancer risk among men of African descent. Genes and immunity. 2013;14(6):347–355. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Jackson MD, Walker SP, Simpson-Smith CM, et al. Associations of whole-blood fatty acids and dietary intakes with prostate cancer in Jamaica. Cancer causes & control : CCC. 2012;23(1):23–33. [DOI] [PubMed] [Google Scholar]
- 20.Jones DZ, Ragin C, Kidd NC, et al. The impact of genetic variants in inflammatory-related genes on prostate cancer risk among men of African Descent: a case control study. Hereditary cancer in clinical practice. 2013;11(1):19. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Stata Statistical Software: Release 13 [computer program]. College Station, TX: Stata Corp LP; 2013. [Google Scholar]
- 22.Park MY, Hastie T. Penalized logistic regression for detecting gene interactions. Biostatistics. 2008;9(1):30–50. [DOI] [PubMed] [Google Scholar]
- 23.R: A language and environment for statistical computing. [computer program]. Vienna, Austria: R Foundation for Statistical Computing; 2008. [Google Scholar]
- 24.Benjamini Y, Hochberg Y. Controlling the False Discovery Rate: A Practical and Powerful Approach to Multiple Testing. Journal of the Royal Statistical Society Series B (Methodological). 1995;57(1):289–300. [Google Scholar]
- 25.Girling JE, Hedger MP. Toll-like receptors in the gonads and reproductive tract: emerging roles in reproductive physiology and pathology. Immunology and cell biology. 2007;85(6):481–489. [DOI] [PubMed] [Google Scholar]
- 26.Kim TK, Lee JS, Oh SY, et al. Direct transcriptional activation of promyelocytic leukemia protein by IFN regulatory factor 3 induces the p53-dependent growth inhibition of cancer cells. Cancer research. 2007;67(23):11133–11140. [DOI] [PubMed] [Google Scholar]
- 27.Consortium EP. An integrated encyclopedia of DNA elements in the human genome. Nature. 2012;489(7414):57–74. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Akahoshi M, Nakashima H, Sadanaga A, et al. Promoter polymorphisms in the IRF3 gene confer protection against systemic lupus erythematosus. Lupus. 2008;17(6):568–574. [DOI] [PubMed] [Google Scholar]
- 29.Tarassishin L, Lee SC. Interferon regulatory factor 3 alters glioma inflammatory and invasive properties. Journal of neuro-oncology. 2013;113(2):185–194. [DOI] [PubMed] [Google Scholar]
- 30.Tamura T, Yanai H, Savitsky D, Taniguchi T. The IRF family transcription factors in immunity and oncogenesis. Annu Rev Immunol. 2008;26:535–584. [DOI] [PubMed] [Google Scholar]
- 31.Menachery VD, Pasieka TJ, Leib DA. Interferon regulatory factor 3-dependent pathways are critical for control of herpes simplex virus type 1 central nervous system infection. J Virol. 2010;84(19):9685–9694. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Friedenreich CM, Thune I. A review of physical activity and prostate cancer risk. Cancer causes & control : CCC. 2001;12(5):461–475. [DOI] [PubMed] [Google Scholar]
- 33.Oliveria SA, Lee IM. Is exercise beneficial in the prevention of prostate cancer? Sports medicine. 1997;23(5):271–278. [DOI] [PubMed] [Google Scholar]
- 34.Na HK, Oliynyk S. Effects of physical activity on cancer prevention. Annals of the New York Academy of Sciences. 2011;1229:176–183. [DOI] [PubMed] [Google Scholar]
- 35.Wang M, Mehta A, Block TM, Marrero J, Di Bisceglie AM, Devarajan K. A comparison of statistical methods for the detection of hepatocellular carcinoma based on serum biomarkers and clinical variables. BMC medical genomics. 2013;6 Suppl 3:S9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Morgan ND, Ferguson TS, Younger NO, et al. Prevalence of high-risk sexual behaviour in Jamaican adults and its relationship to sociodemographic and religious factors: findings from the Jamaica Health and Lifestyle Survey 2007-2008. West Indian Med J. 2012;61(9):873–880. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.