Quantile-specific heritability of 8-isoprostane and the modulating effects of smoking, alcohol, cardiovascular disease and diabetes on 8-isoprostane-gene interactions

Abstract

Background:

Urinary 8-isoprostane provides a significantly heritable measure of oxidative stress. Prior reports suggest that genetic variants may modulate oxidative stress due to smoking, other environmental factors, and disease. Alternatively, these apparent modulations may reflect a dependence of genetic effects on 8-isoprostane concentrations.

Method:

To test whether genetic effects on 8-isoprostane concentrations are quantile-dependent, quantile-specific offspring-parent (β_OP) and full-sib regression slopes (β_FS) were estimated by applying quantile regression to the age- and sex-adjusted creatinine-standardized urinary 8-isoprostane concentrations of Framingham Heart Study families. Quantile-specific heritabilities were calculated as h²=2β_OP /(1+r_spouse) and h²={(1+8r_spouseβ_FS)^0.5-1}/(2r_spouse)).

Results

Spouse 8-isoprostane concentrations were weakly concordant (r_spouse=0.06). 8-isoprostane heritability (h²±SE) increased significantly with increasing percentiles of its distribution (P_{linear trend}=0.0009, P_{quadratic trend}=0.0007, P_{cubic trend}=0.003) when estimated from β_OP, and when estimated from β_FS (P_{linear trend}=0.005, P_{quadratic trend}=0.09, P_{cubic trend}=0.06). Compared to the 10^th percentile, β_OP-estimated h² was over 22-fold greater at the 90^th percentile (P_difference=9.2×10⁻⁵), and 5.3-fold greater when estimated from β_FS (P_difference=0.004). Significantly higher 8-isoprostane heritability in smokers than nonsmokers (0.352±0.147 vs. 0.061±0.036, P_difference=0.01), and heavier than lighter drinkers (0.449±0.216 vs. 0.078±0.037, P_difference=0.01) were eliminated when corrected for the higher 8-isoprostane concentrations of the smokers and heavier drinkers.

Conclusion:

Heritability of oxidative stress as measured by 8-isoprostane is quantile-dependent, which may contribute to the larger reported effects on oxidative stress by UCP2 −866G>A, IL6 −572C>G and LTA 252A>G polymorphisms in smokers than nonsmokers, by the UCP2 −866G>A polymorphism in coronary heart disease patients, by the ESRRG rs1890552 A>G polymorphism in type 2 diabetics, by the CYBA 242 C>T polymorphism after exercise training, by the PLIN 11482G>A/14995A>T haplotype before weight loss, and by the CYBA −930A>G and GSTP1 I105V haplotypes in patients with pulmonary edema.

Graphical Abstract

Introduction

Oxidative stress occurs when the formation of reactive oxygen species (ROS) exceed antioxidant reserves [1]. Systemic oxidative stress may be measured by urinary concentrations of F₂-isoprostanes, which are generated through non-enzymatic peroxidation of arachidonic acid [2]. Urinary 8-isoprostane (aka 8-iso PGF2α, 8-epi PGF2α) is the most commonly reported F₂-isoprostane oxidative stress measurement [3]. Elevated F₂-isoprostane concentrations have been associated with cancer, type 2 diabetes mellitus (T2DM), and cardiovascular, pulmonary, neurological, renal, liver and other diseases [4,5]. Elevated urinary 8-isoprostane increases the risk for hypertension [6], cardiovascular disease [7], T2DM [8] and all cause mortality [7], prospectively.

Heritability (h²), the proportion of the phenotype variance attributable to additive genetic effects, is generally assumed to be the same throughout a trait’s range of values [9]. “Quantile-dependent expressivity” occurs when the phenotypic expression of genetic variants depend upon whether the trait (e.g., urinary 8-isoprostane) is high or low relative to it distribution [10]. Quantile-dependent heritability has been demonstrated for circulating lipids and lipoproteins [11–14], glucose [15], insulin [15], adiponectin [16], leptin [17], C-relative protein [18], interleukin-6 [19], uric acid [20], and plasminogen activator inhibitor type-1 concentrations [21]. Adiposity [22], postprandial lipemia [23], pulmonary function [24], and coffee and alcohol consumption [25,26] also exhibit quantile-dependent heritability, but not height [10,22] nor dietary intakes of other macronutrients [26].

When corrected for creatinine, urinary 8-isoprostane concentrations are significantly heritable [27,28], but it is not known whether genes affecting urinary 8-isoprostane are quantile-dependent, nor whether quantile-dependent genetic effects might explain genetic interactions of smoking, weight loss, exercise training, coronary heart disease, T2DM, and pulmonary edema with urinary 8-isoprostane concentrations. Therefore, quantile regression [29] was applied to a large epidemiologic cohort (Framingham Heart Study [30,31]) to test whether urinary 8-isoprostane concentrations exhibit quantile-dependent heritability in the narrow-sense (h²) when estimated from offspring-parent (β_OP) and full-sib (β_FS) regression slopes [9].

Methods

The Framingham Study data were obtained from the National Institutes of Health FRAMCOHORT, GEN3, FRAMOFFSPRING Research Materials from the National Heart, lung, and Blood (NHLBI) Biologic Specimen and Data Repository Information Coordinating Center. The hypothesis tested is not considered as part of the initial Framingham Study design and is exploratory. Our analyses of these data were approved by Lawrence Berkeley National Laboratory Human Subjects Committee (HSC) for protocol “Gene-environment interaction vs. quantile-dependent penetrance of established SNPs (107H021)”. LBNL holds Office of Human Research Protections Federal wide Assurance number FWA 00006253. Approval number: 107H021–13MR20. All surveys were conducted under the direction of the Framingham Heart Study human use committee guidelines, with signed informed consent from all participants or parent and/or legal guardian if <18 years of age.

The Framingham Heart Study Offspring Cohort included 5,124 adult children of the Original Cohort and their spouses who were initially examined between 1971 and 1975, reexamined eight years later, and then every three to four years thereafter [30]. The Third Generation Cohort consisted of the Offspring Cohort’s children [31]. Subjects used in the current analyses were at least 16 years of age and were self-identified as non-Hispanic White.

Spot urine samples were obtained at examination 7 of the Offspring Cohort and examination 1 of the Third Generation Cohort. Keaney et al. [6] provide detailed description of the laboratory assays and protocol for determining 8-isoprostane concentrations adjusted for creatinine excretion. Briefly, 8-isoprostane levels were measured in duplicate using a commercially available ELISA (Cayman, Ann Arbor, MI) of urine samples stored at −80°C, with an average intra-assay coefficient of variation of 9.7%. Creatinine contents of urine samples stored at −20°C were determined from the reaction of creatinine and alkaline picrate using an Abbott Spectrum CCX and the manufacturer’s instructions, with average intra- and interassay coefficients of variation of 2% and 4%, respectively. Current smokers were defined as using ≥1 cigarette per day, and heavy drinkers were defined as consuming ≥90^th percentile of alcohol intake (170 g/wk) by self-report as previously described [26].

Statistics

The statistical analyses have also been described in detail elsewhere [10–26], and are summarized here only briefly. Offspring-parent regression slopes (β_OP) were computed using parents from examination 7 of the Offspring Cohort and examination 1 of the Third Generation Cohort children. Full-sibling regression slopes (β_FS) were calculated for the sibships identified in the Offspring and Third Generation Cohorts by forming all k_i(k_i-1) sibpair combinations for the k_i siblings in sibship i and assigning equal weight to each sibling [32]. Heritability in the narrow sense (h²) was calculated as h²= 2β_OP/(1+r_spouse) and h²= {(1+8r_spouseβ_FS)^0.5-1}/2r_spouse, where r_spouse is the spouse correlation [9]. Interactions between behavioral variables and heritability were tested by including the effects of the parent’s phenotype, the offspring’s behavior main effect, and the interactive effect of the parent’s phenotype and the offspring’s behavior as independent variables in a standard least squares regression of the offspring’s phenotype.

Simultaneous quantile regression [29] was performed using the sqreg command of Stata (version. 11, StataCorp, College Station, TX). One thousand bootstrap samples were drawn to estimate the variance-covariance matrix for the 91 quantile regression coefficients between the 5th and 95th percentiles of the offspring’s distribution [33]. Quantile-specific expressivity was assessed by: 1) estimating quantile-specific β-coefficients (±SE) for the 5th, 6th,…, 95th percentiles of the sample distribution; 2) plotting the quantile-specific β coefficient vs. the quantile of the trait distribution; and 3) testing whether the resulting graph was constant, or changed as a linear, quadratic, or cubic functions of the percentile of the trait distribution using orthogonal polynomials [34]. Statistics are reported ± one standard error. “Quantile-specific heritability” refers to the heritability statistic, whereas “quantile-specific expressivity” is the biological phenomenon of the trait expression being quantile-dependent.

The results from other studies were re-interpreted from the perspective of quantile-dependent expressivity using genotype-specific means or medians presented in the original articles [35,36,37,38,39] or by extracting these values from published graphs [40,41,42,43] using the Microsoft Powerpoint formatting palette as previously described [23]. They include studies that measured 8-isoprostane in plasma, which is approximately one hundredth the concentration in urine, but which correlates well with urinary concentrations adjusted for creatinine [44]. Our interpretations of other studies are not necessarily those of the original authors.

Results

Table 1 presents the sample characteristics, showing higher 8-isoprostane levels in women than men, as others have previously reported [6,45].

Table 1.

Sample characteristics

	Offspring Cohort		Third Generation Cohort
	Males	Females	Males	Females
Sample size, N	1044	1265	1772	1994
Age, years	61.04 (9.55)	60.73 (9.30)	40.40 (8.61)	39.88 (8.71)
BMI kg/m2	28.65 (4.58)	27.41 (5.76)	27.98 (4.66)	26.03 (6.12)
Urinary 8-isoprostane ng/mmol creatinine	124.60 (87.24)	147.10 (96.82)	102.68 (95.50)	111.21 (85.78)
Smokers, %	12.84	12.57	17.84	16.61
Alcohol intake, gm/wk			15.15 (18.71)	7.15 (9.72)

Mean (standard deviation)

Classical estimates of heritability

The age and sex-adjusted spouse correlation was r_spouse =0.0598. When adjusted for assortative mating, and ignoring shared environment and dominance in sibs, the classical estimates of age- and sex-adjusted 8-isoprostane heritability (h²) were 0.115±0.038 when estimated from β_OP (P=0.003) and 0.172±0.035 when estimated from β_FS (P=9.5×10⁻⁷). h² tended to be greater in females than males for both the β_OP-based estimates (0.146±0.047 vs. 0.074±0.062), and β_FS-based estimates (0.212±0.061 vs. 0.130±0.064) albeit not significantly so.

Quantile-specific heritability.

Figure 1A presents the offspring-parent regression slopes (β_OP) for selected quantiles of the offspring’s age and sex-adjusted 8-isoprostane concentrations. These quantile-specific regression slopes were included with those of other quantiles to create the quantile-specific heritability function in Figure 1B, i.e., where the offspring-parent slopes (Y-axis) are plotted as a function of the quantile of the offsprings’ sample distribution (X-axis). The figure shows that each ng/mmol increment in the parents’ value was associated with an offsprings’ increase of (slope±SE) 0.010±0.009 ng/mmol creatinine at the 10^th percentile of the offspring’s distribution (NS), 0.019±0.009 increase at the 25^th percentile (P=0.04), 0.026±0.012 increase at the 50^th percentile (P=0.03), 0.071±0.026 increase at the 75^th percentile (P=0.005), and 0.218±0.053 increase at the 90^th percentile (P=4.4×10⁻⁵). These correspond to h²±SEs of 0.019±0.017 at the 10^th, 0.036±0.017 at the 25^th, 0.049±0.022 at the 50^th, 0.135±0.048 at the 75^th, and 0.411±0.101 at the 90^th percentile of 8-isoprotane concentrations. If the offspring-parent slope was the same for all offspring quantiles as traditionally assumed, then Figure 1A would display parallel regression lines, and Figure 1B would present a flat horizontal graph. In fact, the graph shows that the slopes became progressively greater with increasing quantiles above the 70^th percentile, such that on average each 1-percent increase in the offspring’s distribution was associated with a 0.0038±0.0011 unit linear increase in h² (P=0.0009).

Figure 1.

A) Offspring-parent regression slopes (β_OP) for selected quantiles of the offspring’s creatinine-standardized urinary 8-isoprostane concentrations from 4288 offspring-parent pairs, with corresponding estimates of heritability (h²=2β_OP/(1+r_spouse) [9]), where the correlation between spouses was r_spouse=0.0598. The slopes were relatively constant through the 50^th percentile and became progressively greater at the 75^th and 90^th percentiles. B) The selected quantile-specific regression slopes were included with those of other quantiles to create a quantile-specific heritability function. Significance of the linear, quadratic and cubic trends and the 95% confidence intervals (shaded region) determined from 1000 bootstrap samples. C) Quantile-specific full-sib regression slopes (β_FS) from 4595 full-sibs in 1665 sibships, with corresponding estimates of heritability as calculated by h²={(8r_spouseβ_FS+1)^0.5-1}/(2r_spouse) [9].

Heritability was over 22-fold greater at the 90^th than at the 10^th percentile of offspring’s 8-isoprostane distribution. The h² difference between the least and greatest oxidative stress (i.e., 90^th −10^th percentile) was 0.393±0.100 (P=9.2×10⁻⁵). Figure 1C shows that the quantile-specific heritability for 8-isoprostane was also evident from the regression slopes between siblings. Specifically, the β_FS-estimated heritability: 1) increased 0.0032±0.0012 (P=0.005) for each percentile increase in the distribution of the ratio, and 2) was 5.3-fold greater at the 90^th (0.373±0.107) than the 10^th percentile (0.071±0.012) of the sibs’ distribution.

Gene-environment interactions

Average 8-isoprostane concentrations were significantly higher in smokers than nonsmokers (156.72±6.32 vs. 99.07±1.70 ng/mmol creatinine, P=10⁻¹⁶) and significantly higher in heavier than lighter drinkers (144.11±8.95 vs. 103.41±1.75 ng/mmol creatinine, P=1.7×10⁻¹¹), where heavy drinkers were defined as ≥90^th percentile of consumption (≥170.65 g/wk). Standard least squares regression showed significantly greater 8-isoprostane heritability in smokers than nonsmokers (0.352±0.147 vs. 0.061±0.036, P_difference=0.01) and in heavier than lighter drinkers (0.449±0.216 vs. 0.078±0.037, P_difference=0.01). Figure 2A present the smokers’ and nonsmokers’ quantile-specific heritability plots when matched by percentiles of their separate distributions, showing that their heritabilities were indistinguishable below the 50^th percentile and then diverged sharply at higher percentiles. The problem with this comparison is that h² is being compared at different 8-isoprostane concentrations, i.e., at 63.5 ng/mmol in smokers to 42.3 ng/mmol in nonsmokers at the 10th percentile, 86.9 ng/mmol in smokers to 58.3 ng/mmol in nonsmokers at the 25th percentile, 126.7 ng/mmol in smokers to 83.2 ng/mmol in nonsmokers at the 50th percentile, 182.7 ng/mmol in smokers to 115.2 ng/mmol in nonsmokers at the 75th percentile, and 270 ng/mmol in smokers to 163.6 ng/mmol in nonsmokers at the 90th percentile. Specifically, because the 8-isoprostane heritability doesn’t really increase with increasing 8-isoprostane concentrations until ≥140 ng/mmol creatinine, the smokers and nonsmokers h² will be indistinguishable so long as both concentrations are <140 ng/mmol. However, once the smokers’ 8-isoprostane concentration exceed 140 ng/mmol their h² rapidly diverge from that of the nonsmokers. Heritability will also accelerate upwards in the nonsmokers once their 8-isoprostane concentrations exceed 140 ng/mmol, but this doesn’t occur until their 80^th percentile.

Figure 2.

Offspring-parent heritability plots: A) in smoking and nonsmoking offspring showing their heritability difference when compared at their corresponding quantiles (smoking offspring’s vs. nonsmoking offspring’s h² compared at the 5th percentile of their separate distributions, the 6th percentile of their separate distributions, …, 95th percentile of their separate distributions); B) showing the smoker vs. nonsmoker difference is eliminated when the heritabilities are compared at their corresponding 8-isoprostane concentrations (i.e. probability-probability (P-P) plots [46] used to re-plot their heritabilities at their matching concentrations); C) in heavier vs. lighter drinking offspring showing their heritability difference when compared at their corresponding quantiles; D) showing the heritability difference between heavier vs. lighter drinkers is eliminated when their heritabilities are compared at their corresponding 8-isoprostane concentrations.

Probability-probability plots [46] were used to re-plot the smokers and nonsmokers heritability estimates at their corresponding 8-isoprostane concentrations, which shifted the smoker’s heritability estimates to the right. The result, displayed in Figure 2B, shows no difference in the smokers and nonsmokers h² when plotted at the same 8-isoprostane concentration. This means that the difference in smoker and nonsmoker heritability in Figure 2A was entirely attributable to quantile-dependent heritability and the higher 8-isoprostane concentrations of the smokers. Similarly, the heritability difference between heavier and lighter drinkers when plotted by their respective percentiles in Figure 2C was entirely eliminated when the heritability of the heavier and lighter drinkers were re-plotted at their matching 8-isoprostane concentrations (Figure 2D).

Logarithmically transformed 8-isoprostane concentrations.

Estimated heritability for log 8-isoprostane concentrations was 0.1495±0.0337 when estimated from β_OP, and 0.2572±0.0381 when estimated from β_FS, using an r_spouse=0.1034. Heritability of log-transformed 8-isoprostane concentrations was quantile-dependent when estimated from β_OP (P_linear=0.05, P_quadratic=0.06) albeit not when estimated from β_FS.

Discussion

These analyses of Framingham Heart Study families suggest that 8-isoprostane heritability is quantile dependent. Offspring-parent estimated heritability increased significantly with increasing 8-isoprostane concentrations (P_{linear trend} =0.0009), accelerating sharply at the highest concentrations (P_quadratic=0.0007, P_cubic=0.003), such that heritability at the 90^th percentile was over 22-fold greater than at the 10^th percentile of its distribution (P_difference=9.2×10⁻⁵, Figure 1B). Heritability estimated from full sibs was generally consistent with these results, showing a significant linear trend with increasing concentrations (P_linear=0.005) and 5.3-fold difference in heritability at the 90^th vs. the 10^th 8-isoprostane percentiles (Figure 1C). Quantile-dependent heritability fully explained the significant interactions of smoking and heavy drinking on 8-isoprostane heritability (Figure 2).

Gene-environment interactions

Cigarette smoke, a potent stimulator of ROS [47], contains multiple oxidizing species that can give rise to lipid peroxidation [48]. Smokers also have lower circulating vitamin C concentrations (an important antioxidant) [49]. Compared to nonsmokers, smokers have been reported to have 40% greater plasma-free isoprostane [47], and 65% greater urinary 8-isoprostane concentrations [6]. Excessive intake of alcohol is associated with increased inflammation and increase oxidative stress [50]. Urinary 8-isoprostane has been shown to increase in a dose-dependent manner with the administration of alcohol in healthy volunteers [51], and to decrease significantly in moderate-to-heavy drinkers when they decreased their intake by nearly 90% [52].

The Framingham sample showed significantly higher urinary 8-isoprostane concentrations in offspring who reported smoking or heavier alcohol consumption. Correspondingly, 8-isoprostane h² was greater in the drinkers and smoker. However, quantile regression showed that the heritability difference was restricted to the higher percentages of the 8-isoprostane distribution, and that when matched for 8-isoprostane concentrations rather than the percentiles of their corresponding distributions, these heritability differences disappeared. Thus, these apparent gene-environment interactions were entirely attributed to quantile-dependent heritability. This approach of using probability-probability plots to compare genetic effects when matched by the phenotype rather than their corresponding percentile distributions has been previously shown to explain sex-differences in the heritability of plasma plasminogen activator inhibitor type-1 concentrations [21], plasma leptin concentrations [17], and plasma adiponectin concentrations [16].

Relevance to previously published results.

Heritability (h²) measures the cumulative genetic contribution across multiple loci, but lacks the specificity of genetic variants measured directly. Others have described the effects of specific polymorphisms under conditions of high and low 8-isoprostane concentrations. They include studies that sought to identify genetic markers for identifying patients susceptible to oxidative stress (a precision medicine perspective, e.g. the histograms of Figures 3–5). Alternatively, quantile-dependent expressivity postulates that the effect size of a genetic variant may differ when the phenotype is high or low (e.g., the line graphs of Figures 3–5). When this occurs, the genotype-specific changes in the phenotype cannot move in parallel, i.e., the changes in the phenotype will be genotype specific [13,14]. In this case, the genetic marker may simply track the change in heritability associated with higher vis-à-vis lower phenotype values. A potential consequence of quantile-dependent expressivity is that sampling by characteristics that distinguish high- vs. low-valued phenotypes may produce genetic differences traditionally ascribed to gene-environment, gene-drug and gene-diet interactions when a simpler explanation of quantile-dependence might suffice [12,15].

Figure 3.

Precision medicine perspective of genotype-specific F₂- and 8-isoprostane differences (histogram inserts) vs. a quantile-dependent expressivity perspective (line graphs showing larger genetic effect size when average F₂- or 8-isoprostane concentrations were high) for the data presented in: A) Stephens et al. [35] 2008 report on smokers vs. nonsmokers by the mitochondrial uncoupling protein-2 (UCP2) −866G>A polymorphism; B) Jang et al.’s [40] 2007 report on smokers vs. nonsmokers by the lymphotoxin-α (LTA) 252A>G genotypes; C) Shin et al.’s [41] 2007 report on smokers vs. nonsmokers by the interleukin-6 (IL-6) −572C>G genotypes (P_interaction=0.02); D) Dhamrait et al.’s [36] 2004 report on diabetic men with vs. without coronary heart disease (CHD) by UCP2 −866G>A genotypes; E) Feairheller et al.’s [37] 2009 report on the exercise-induced changes in 8-isoprostane by p22phox (CYBA) 242C>T genotypes; F) Jang et al. [42] 2006 report on dieting-induced changes in 8-isoprostane by 11482G>A/14995A>T haplotypes of the perilipin (PLIN) locus. *per mg or mmol creatinine.

Increased oxidative stress due to high levels of free radicals and oxidants produced by smoking toxicities may explain the linkage between smoking and accelerated aging [53]. Several studies report genetic interactions between 8-isoprostane and smoking, where genetic variants might modify 8-isoprostane heritability by affecting ROS generation directly [35] or as a consequence of their effect on TNF-α and IL-6 inflammatory responses [40,41]. For example, the mitochondrial uncoupling protein-2 (UCP2) uncouples ATP synthesis from electron transport in mitochondria, resulting in decreased ROS production [54]. UCP2-expression is impeded by the A-allele of the −866G>A polymorphism in the UCP2 promoter region, causing enhanced mitochondrial ROS production and oxidative stress to increase [55,56]. Stephens et al. [35] reported significantly higher plasma total F₂-isoprostanes in smokers than nonsmokers (693.9 vs. 446.2 pg/ml), which the histogram in Figure 3A suggests primarily reflects the effect of smoking in UCP2 −866AA homozygotes. Correspondingly, the line graph of Figure 3A shows a significant difference between genotypes (P=0.04) at the higher concentration of the smokers but not at the lower concentration of the nonsmokers (P=0.74). This may be because UCP2 gene activity and expression are increased by oxidative stress [57,58], i.e., the polymorphism’s affect on oxidative stress being accentuated by smoking-induced ROS generation.

The lymphotoxin-α (LTA) 252G allele has higher TNF-α secretary capacity, and is associated with higher circulating TNF-α concentrations than 252A allele. Smoking-induced inflammatory response and oxidative stress are interrelated, i.e., oxidative stress up-regulates proinflammatory gene expression while inflammation induces oxidative stress in the lungs [59]. Jang et al. [40] reported a significant interaction between the 252A>G polymorphism and smoking in their effects on both TNF-α and IL-6 circulating concentrations. Correspondingly, the smokers had higher urinary 8-isoprostane excretion than nonsmokers (927±35 vs. 861±29 pg/mg creatinine, P=0.04), and Figure 3B shows there was a significant interaction between the 252A>G polymorphism and smoking status on 8-isoprostane excretion (P<0.001), with the smoking effect being greater in GG-homozygotes than A-allele carriers (187 vs. 64 pg/mg creatinine, histogram). The line graph shows that the difference between GG-homozygotes than A-allele carriers was greater at the higher mean concentrations of the smokers than at the lower concentrations of the nonsmokers (343 vs. 220 pg/mg creatinine), paralleling similar results for circulating TNF-α (not displayed).

Similarly, Shin et al. [41] reported interactions between smoking and the −572C>G polymorphism in the IL-6 promoter region on 8-isoprostane concentrations reflective of their effects on inflammatory cytokines. Specifically the −572C>G polymorphism was associated with greater differences in IL-6, fibrinogen, C-reactive protein, and oxidized low-density lipoprotein concentrations in Korean smokers than nonsmoker. Correspondingly, Figure 3C shows urinary 8-isoprostane concentrations were higher in smokers than nonsmokers (965±46 vs. 901±32 pg/mg creatinine, P=0.06), and were significantly higher in GG homozygotes than C-allele carriers of the IL6 −572C>G promoter polymorphism in smokers (P=0.05) but not nonsmokers (P=0.16).

Reactive oxygen species play an important role in the pathogenesis of atherosclerosis [60], including the formation of oxidized LDL. Dhamrait et al. [36] reported that plasma F₂-isoprostane concentrations were higher in men with coronary heart disease (CHD) than those without (168.3 vs. 111.7 pg/ml, P=0.05), particularly in AA homozygotes of the UCP2 −866G>A polymorphism (P_interaction=0.014, Figure 3D histogram). The associated line graph shows a significant difference between genotypes at the higher concentrations of the CHD patients (P=0.005) but not at the lower concentrations of those free from manifest CHD (P=0.58), consistent with quantile-dependent expressivity.

The NADPH oxidase enzyme complex is the primary source of the vascular ROS, superoxide (O₂⁻) [61]. Although low intensity (25% VO_2max) and moderate intensity aerobic exercise training (50% VO_2max) reduce NADPH oxidase expression, high-intensity aerobic exercise training (70% VO_2max) increases plasma oxidative stress levels [62]. The T-allele of the 242C>T (rs4673) polymorphism of the p22phox (CYBA) subunit gene is associated with reduced NADPH oxidase activity [63]. Feairheller et al. [37] reported that 6 months of high-intensity aerobic exercise training significantly increased in urinary 8-isoprostane from 0.32±0.01 to 0.42± 0.03 nmol/mmol creatinine (P = 0.002), which the histogram of Figure 3E suggests increased in relation to the number of C-alleles. Correspondingly, consistent with quantile-dependent expressivity, the associated line graph shows a greater divergence between genotypes at the higher post-training than at the lower pre-training mean 8-isoprostane concentrations.

Perilipins, located principally in adipocytes, are required for triglyceride lipolysis during fasting and exercise [64]. Genetic variations at the perilipin PLIN locus have been associated with body weight and may modulate circulating free fatty acids [65]. Higher circulating FFA levels may increase of ROS, urinary 8-isoprostane, oxidized-LDL, and oxidative stress [66,67].

Jang et al. [42] reported that caloric restriction produced significantly greater urinary 8-isoprostane reductions in PLIN GA/GA haplotypes (Figure 3F histogram, P<0.05). Consistent with quantile-dependent expressivity, the difference between haplotypes was significant at baseline when average 8-isoprostane concentrations were higher, but not after weight loss when average concentrations were reduced (Figure 3F, line graph).

Mishra et al. [39] reported significantly higher plasma 8-isoprostane concentrations in patients experiencing pulmonary edema on high altitude sojourns, particularly in G-allele carriers of the CYBA −930A>G or GSTP1 I105V (A>G) polymorphism (Fig. 4 histogram). Again, the haplotype difference was greater for the higher average plasma concentrations of the pulmonary edema patients than for the lower concentrations of the unaffected patients (Figure. 4 line graph).

Figure 4.

Precision medicine perspective of genotype-specific 8-isoprostane difference between patients experiencing and not experiencing pulmonary edema during a high altitude sojourns (histogram insert [39]) by G-allele carriers of the CYBA −930A>G and GSTP1 I105V haplotypes vs. a quantile-dependent expressivity perspective showing larger genetic effect size when average 8-isoprostane concentrations were high (line graph).

Oxidative stress is associated with insulin resistance, β-cell dysfunction, impaired fasting glucose, impaired glucose tolerance, and T2DM [6]. Kim et al. [38] examined the association between 8-isoprostane concentrations and a genetic risk score (GRS) consisting of nine nominally significant diabetes-related SNPs that were weighted to predict impaired fasting glucose and T2DM status. The histogram in Figure 5A shows that the 8-isoprostane difference between patients (impaired fasting glucose or T2DM) and controls (normal fasting glucose) was greater in those with the highest GRS than those with low or intermediate GRS. The associated line graph shows that the 8-isoprostane difference between GRS genotypes was also greater in patients (r = 0.197, p < 0.001) than normal glucose controls (P=NS), consistent with the higher average 8-isoprostane concentrations of the patients (1761.4±35.5 vs. 1497.3±24.0 pg/mg creatinine, P<0.001) and quantile dependent expressivity.

Figure 5.

Precision medicine perspective of genotype-specific 8-isoprostane differences (histogram inserts) vs. a quantile-dependent expressivity perspective (line graphs showing larger genetic effect size when average 8-isoprostane concentrations were high) for the data presented Kim et al.’s 2017 and 2018 reports on patients (T2DM and impaired fasting glucose) vs. controls (normal fasting glucose) by A) high and low-to-intermediate valued diabetes-related genetic risk score (GRS) [38]; B) estrogen-related receptor gamma (ESRRG) rs1890552 A>G polymorphism [43]. * per mg creatinine.

Our final example involves estrogen-related receptor gamma (ESRRG), belonging to the orphan nuclear receptor superfamily, that is a novel candidate gene for T2DM [68]. Kim et al. reported that the GG homozygote of the ESRRG rs1890552 A>G polymorphism was associated with increased risk for both impaired fasting glucose and T2DM [43]. They also reported significantly higher 8-isoprostane concentration in patients (impaired fasting glucose and T2DM) than controls (normal fasting glucose, 1625.1±36.5 vs. 1434.4±21.9 ng/mg creatinine, P<0.001). The patient-control difference was greatest in GG homozygotes, intermediate in AG heterozygotes, and smallest in AA homozygotes (Figure 5B histogram). The line graph shows the 8-isoprostane genotype difference was approximately 3-fold greater at the higher 8-isoprostane concentrations of the patients than at the lower concentrations of the normal controls.

Limitations

Several limitations warrant consideration. 8-isoprostane concentrations were determined for spot rather than 24-hour urine collection and were analyzed using the less precise ELISA rather than mass spectrometry coupled to gas chromatography (GC/MS) methodology. 8-isoprostane can also be generated by prostaglandin endoperoxide synthase enzymes induced during inflammation and therefore may not be specific to oxidative stress [69]. The Framingham Heart Study cohort was generally non-Hispanic White and therefore these results may not be generalizable to other ethnicities. Moreover, Falconer’s formula for estimating heritability may not adequately reflect the true complexity of genetic effects on 8-isoprostane concentrations.

Quantile-regression does not depend upon statistical normality and therefore has the advantage of assessing urinary 8-isoprostane heritability as originally measured. Nearly all of the gene-specific effects of smoking and disease are presented as untransformed 8-isoprostane concentrations. Our analyses of smoking and drinking interactions are limited by a lack of information on their history of use. Finally, we acknowledge that quantile-dependent expressivity of 8-isoprostane concentrations, and its ability to explain genetic interactions between 8-isoprostane concentration and smoking and drinking require replication in other populations.

With regards to the cited examples of gene-environment interaction published by others, it is important to acknowledge that SNPs selected in these studies are not necessarily directly related to in vivo formation of 8-isoprostanes. Moreover, no genomewide association study has yet been published for 8-isoprostanes, and the findings on selected SNPs and 8-isoprostane have been rarely replicated.

Conclusion

In conclusion, heritability of oxidative stress as measured by 8-isoprostane is quantile-specific, which may contribute to the larger reported effects on oxidative stress by UCP2 −866G>A, IL6 −572C>G and LTA 252A>G polymorphisms in smokers than nonsmokers, by the UCP2 −866G>A polymorphism in coronary heart disease patients, by the ESRRG rs1890552 A>G polymorphism in type 2 diabetics, by the CYBA 242 C>T polymorphism after exercise training, by the PLIN 11482G>A/14995A>T haplotype before weight loss, and by the CYBA −930A>G and GSTP1 I105V haplotypes in patients with pulmonary edema.