Influence of Pubertal Development on Urinary Oxidative Stress Biomarkers in Adolescent Girls in the New York LEGACY Cohort

Abstract

Context:

Puberty is a time of intense growth and differentiation of breast tissue and a window of susceptibility (WOS) for breast cancer. Although oxidative stress markers have been associated with breast cancer risk, it is unclear whether oxidative stress levels are different during the pubertal WOS, and if so, whether these differences are related to breast cancer susceptibility.

Methods:

We measured urinary biomarkers of whole body oxidative stress (urinary F2-Isoprostanes and 8-oxodeoxyguanosine (8-oxodG)) in 158 girls (ages 6–13 years), 71 with and 87 without a breast cancer family history (BCFH) from a cohort of adolescent girls from the New York site of the LEGACY cohort (Lessons in Epidemiology and Genetics in Adults Cancer from Youth). We compared levels of urinary oxidative stress biomarkers (F2-Isoprostanes and 8-oxodG) across the pubertal window, defined by Tanner Stage (TS) of breast development, both cross-sectionally and longitudinally within girls over an 18-month follow up period.

Results:

Urinary oxidative stress biomarkers were unrelated to pubertal stages in cross-sectional analyses after considering adjustments for body mass index (BMI) and BCFH. In our longitudinal analysis, we found that urinary 8-oxodG levels, but not F2-Isoprostane levels, increased with age in BCFH+ girls (β=6.12, 95%CI=0.08–12.16) compared to BCFH− girls. Higher BMI was associated with higher level of F2-Isoprostane in both cross-sectional (β=0.02, 95%CI=0.0004–0.05) and longitudinal analysis (β=0.02, 95%CI=0.0002–0.05).

Conclusion:

These findings support that higher body mass index increases oxidative stress biomarkers over the pubertal window and that there are changes in 8-oxodG oxidative stress biomarkers in girls with a breast cancer family history compared to girls without a breast cancer family history.

INTRODUCTION

The incidence of non-localized breast cancer is increasing in women under 40 years (1) and at the same time, the age at onset of puberty has been decreasing in the U.S. and globally (2–4). Pubertal milestones, such as onset of breast development and menstruation, play an important role in breast cancer etiology (5–9). A large, prospective cohort of 104,931 women reported that earlier onset of breast development at age ≤10 years was associated with a 20% increased risk of breast cancer, and that this increase was independent of height and age at menarche (10).

Puberty is a time of significant hormone exposure and mitogenic activity in breast tissue (11–13). Oxidative stress, an imbalance in the pro-oxidant/antioxidant mechanisms in the body, has been implicated in tumorigenesis by increasing mutations, and accelerating telomere shortening (14–19). Increasing evidence suggests that oxidative stress may be of particular importance in breast cancer etiology (20–22) and higher levels of oxidative stress have been observed in women diagnosed with breast cancer compared with women without breast cancer (22–26). Oxidative stress is also associated with risk factors for breast cancer, including estrogen levels (27–32). Estrogen quinone metabolites undergo redox cycling, generating superoxide radicals and potentially hydroxyl radicals which can damage DNA (33, 34), suggesting that estrogen-induced reactive oxygen species (ROS) may contribute to the development of breast cancer (35). It is unclear whether oxidative stress levels change with pubertal development.

Using biospecimens and epidemiologic data from 158 adolescent girls from the New York site of the Lessons in Epidemiology and Genetics and Adult Cancer from Youth cohort (LEGACY), we examined whether oxidative stress biomarkers differed across the pubertal window and whether there were differences in levels by breast cancer family history (BCFH). We compared urinary levels of 8-oxo-2’-deoxyguanosine (8-oxodG), a product of oxidative stress on DNA, and 15-F_2t-Isoprostanes (F2-Isoprostanes), a marker of lipid peroxidation, both cross-sectionally at baseline and longitudinally every 6-months for up to 18-months.

MATERIALS AND METHODS

Study Population and Design

We conducted this study using data and biospecimens from 158 girls (ages 6–13 years) participating in the New York site of the LEGACY Girls Study cohort which is a multicenter prospective study of 1,040 girls and includes 5 sites (New York City, NY; Philadelphia, PA; Salt Lake City, UT; San Francisco Bay Area, CA; and Toronto, Ontario; for details see (36, 37). We collected biospecimens and epidemiological data at baseline and at biennial follow-up visits as well as clinical measures of anthropometry (height and weight) and Tanner Stages (TS). TS was also collected by mother/guardian and daughter. For this paper, we used the mother/guardian report of TS as the primary analysis but also conduct sensitivity analyses based on clinical measurements of TS. Compared with clinical assessment, we previously reported that sensitivity of maternal assessment for breast development was 77.2 and specificity was 94.3 (38). Almost half of the girls have a BCFH, defined as at least one first- or second-degree relative diagnosed with breast cancer. The mean age for girls with and without a BCFH is 9.4±2.6 and 8.8±2.2 (p=0.07). We calculated age-specific body mass index (BMI) centiles based on CDC growth charts (39) and categorized girls in the 85th percentile or above for their age as overweight. We collected first morning urine samples at baseline (F0) and at 6-month (F1), 12-month (F2) and 18-months (F3) follow-up visits after baseline. This study was approved by the Columbia University Medical Center Institutional Review Board. Mothers/guardians provided written informed consent, and girls provided assent based on institutional standards.

Laboratory Assays

We used frozen urine samples from 71 girls with a BCFH (BCFH+) and 87 girls without a BCFH (BCFH−) to measure levels of 8-oxodG and F2-Isoprostanes using well-established ELISAs and assayed all samples in duplicate and blinded to any epidemiological information. Quality control included analysis of a pooled urine sample with each batch of test samples and 5% duplication of samples after relabeling to keep laboratory personnel blinded to sample identity. Urinary biomarker levels were normalized for hydration status by measuring specific gravity, assayed using a handheld refractometer (TS 400, Reichert, Depew, NY). Urinary F2-Isoprostane levels were measured using immunoassay kits from Oxford Biomedical Research (Product Number: EA85 Oxford, MI) according to the manufacturer’s recommendations. For 8-oxodG measurement, urine samples were treated with 20 U/ml urase (Sigma-Aldrich, St. Louis, MO, USA) for 2hrs at 37°C to reduce non-specific binding as recommended (40). Urase treatment was stopped by adding N-ethylmaleimide to a final concentration of 4 mM for 5 min at room temperature. The supernatant was removed after centrifugation and then used for Solid Phase Extraction (SPE) as described by Lam et al (41). The eluted urine samples were then used in an in-house competitive ELISA (42, 43) using anti-8-oxodG N45.1 antibody (Japanese Institute for the Control of Aging, Shizuoka, Japan). As creatinine is known to vary by gender, age, red meat intake and season (44, 45), we adjusted urinary biomarker levels for specific gravity (SG), a measure of hydration status of the urine using the Levine Fahy equation (46): SG-Adjusted Biomarker (nmol/L*SG) = Biomarker (nmol/L) × [(overall mean SG −1)/(sample SG −1)]

We report results as nmol 8-oxodG/L*SG and nmol Isoprostanes /L*SG. To minimize the variability between plate/batch, we ran repeated measures from the same subject on the same SPE and ELISA plate.

Statistical Analysis

Cross-sectional analysis of oxidative stress biomarkers and pubertal development using F0 data

We used Wilcoxon Mann Whitney tests to compare the overall mean difference in urinary 8-oxodG and F2-Isoprostane levels by different puberty stages at baseline (F0). We define pubertal stage by multiple binary indicators for development milestones, which include menarche (yes vs no), prepubertal (Tanner Stage (TS) =1) and thelarche (TS ≥2). We used linear regressions to examine the associations among oxidative stress biomarkers, pubertal development milestones, while adjusting for BCFH (any vs none), BMI (continuous) and age (centered to 9 years). We also tested for additive interaction between BCFH and age.

Longitudinal analysis

With the full longitudinal data, we used linear mixed model to examine the change in oxidative stress levels within girls over time, while accounting for the within-subject variability. Centered age centered, BCFH, BMI as well as breast TS stage and age at menarche are included into the model as covariates. Interaction between BCFH and age are formally tested. All analyses were performed with SAS software 9.4 (SAS Institute, Cary, NC).

RESULTS

Baseline Urinary Oxidative Stress Biomarkers During Adolescence

Table 1 presents the distributions of urinary oxidative stress biomarkers by selected characteristics at baseline (F0). Levels of urinary F2-Isoprostanes were higher in girls who had reached menarche compared with girls who had not; mean levels were 9.11 (SD= 4.17) nmol/L*SG for girls who reached menarche and 7.88 (SD=5.00) nmol/L*SG for girls who did not. Girls who were breast TS 2+ had higher F2-Isoprostanes levels than girls classified as pre-pubescent at TS 1 (9.03± 5.00 vs 7.47± 4.70, nmol/L*SG, p<0.05) (Table 1). The association between F2-Isoprostanes levels and breast TS 2+ was no longer statistically significant after simultaneously adjusting for age, BCFH and BMI (Model 4 in Table 2). The level of urinary 8-oxodG was higher among girls who either reached menarche or breast TS 2+ based on the mother/guardian report, although the associations were not statistically significant.

Table 1.

Levels of urinary oxidative stress biomarkers at baseline (F0) by characteristics among girls at the New York Site of the LEGACY Girls Study

Variable	No, %	8-oxodG	F2-Isoprostanes
Age		Mean (SD), nmol/L*SG	Mean (SD), nmol/L*SG
5–7	49 (31)	119 (50)	7.85 (5.07)
8–10	64 (40)	119 (63)	7.87 (5.11)
11–13	45 (29)	128 (59)	8.71 (4.33)
Race/Ethnicity
Non-Hispanic White	73 (46)	120 (44)	8.04 (3.87)
Hispanic White	54 (34)	122 (71)	8.25 (5.24)
Other	31 (20)	122 (64)	7.96 (6.31)
Breast Cancer Family History
No	87 (55)	122 (58)	7.64 (4.85)
Yes	71 (45)	120 (58)	8.66 (4.87)
Reached Menarche
No	130 (82)	118 (57)	7.88 (5.00)
Yes	28 (18)	134 (64)	9.11 (4.17)
Breast Tanner Stage
Tanner Stage 1	94 (59)	115 (47)	7.47 (4.70)
Tanner Stage 2+	64 (41)	131 (71)	9.03 (5.00)*
Age-specific BMI
<85th percentile	109 (74.7)	121 (58)	7.56 (4.02)
≥85th percentile	37 (25.3)	121 (61)	9.22 (5.62)

P values from Wilcoxon Mann Whitney tests

^*p value<0.05

Table 2.

Association between urinary oxidative stress biomarkers and pubertal development stage and BMI at baseline (F0) at the New York site of the LEGACY Girls Study

8-oxodG	Univariable model		Model 1		Model 2		Model 3		Model 4
	β	CI	β	CI	β	95% CI	β	95% CI	β	95% CI
Age-centered (yrs)	2.56	−1.36, 6,48	2.67	−1.31, 6.65	−1.69	−7.70, 4.31	−1.36	−7.58, 4.86	−6.26	−14.74, 2.22
Breast Cancer Family History	−1.78	−20.30, 16.73	−3.63	−22.31, 15.04	−4.07	−22.60, 14.45	−3.29	−23,69, 17.12	−3.32	−23.60, 16.96
Breast Cancer Family History and age interaction					7.67	−0.29, 15.63	7.32	−1.10, 15.73	7.88	−0.51, 16.27
Age-specific BMI (cont.)	−0.02	−0.34, 0.31					−0.02	−0.35, 0.32	−0.11	−0.48, 0.23
Breast Tanner Stage 2+ vs 1	16.1	−2.49, 34.70							28.5	−5.24, 62.27
Reached Menarche (Yes vs No)	15.74	−8.25, 39.74
F2-Isoprostanes	Univariable model		Model 1		Model 2		Model 3		Model 4
	β	CI	β	CI	β	95% CI	β	95% CI	β	95% CI
Age-centered (yrs)	0.12	−0.21, 0.44	0.09	−0.24, 0.42	0.33	−0.17, 0.84	0.21	−0.26, 0.68	0.02	−0.62, 0.66
Breast Cancer Family History	1.02	−0.51, 2.56	0.96	−0.59, 2.52	0.99	−0.56, 2.54	1.20	−0.34, 2.73	1.19	−0.34, 2.73
Breast Cancer Family History and age interaction					−0.43	−1.10, 0.23	−0.40	−1.03, 0.24	−0.38	−1.01, 0.26
Age-specific BMI (cont.)	0.03	0.002, 0.05*					0.02	0.0004, 0.05**	0.02	−0.01, 0.04
Breast Tanner Stage 2+ vs 1	1.56	0.01, 3.10**							1.11	−1.45, 3.66
Reached Menarche (Yes vs No)	1.22	−0.78, 3.20

Model 1: Multivariable linear regression model with outcome as oxidative stress markers and covariate as age centered and BCFH,

Model 2: Multivariable linear regression model includes age centered, BCFH in Model 1 plus interaction between age centered and BCFH,

Model 3: Multivariable linear regression model includes age centered, BCFH and interaction between age centered and BCFH in Model 2 plus BMI,

Model 4: Multivariable linear regression model includes age centered, BCFH, interaction between age centered and BCFH and BMI in Model 3 plus Breast Tanner Stage,

^*p=0.03

^**p=0.04

In the cross-sectional analyses, there are no differences in levels of urinary oxidative stress biomarkers by family history; the mean levels of 8-oxodG and F2-Isoprostanes were 122 (SD=58) nmol/L*SG and 7.64 (SD=4.85) nmol/L*SG for girls without a BCFH. The mean levels of 8-oxodG and F2-Isoprostanes were 120 (SD=58) nmol/L*SG and 8.66 (SD=4.87) nmol/L*SG for girls with a BCFH (Table 1 and Figure 1).

Figure 1. The distribution of urinary oxidative stress biomarkers by family history at F0 at the New York site of the LEGACY Girls Study.

The boxplot displays the distribution of urinary oxidative stress biomarkers (F2-Isoprostanes and 8-oxodG (nmol/IL)) by family history. The middle bold line represents the median values of urinary oxidative stress markers. Each dot represents the urinary oxidative stress biomarker value of each samples.

A. The distribution of urinary 8-oxodG (nmol/IL) by family history

B. The distribution of urinary F2-Isoprostanes (nmol/IL) by family history

Higher BMI was associated with level of F2-Isoprostanes in multivariable models adjusting for age, BCFH and the interaction between age and BCFH (β=0.02, 95%CI=0.0004–0.05), P=0.04) (Model 3 of Table 2).

The mean levels of 8-oxodG and F2-Isoprostanes were 123 (SD=64) nmol/L*SG and 8.21 (SD=4.04) nmol/L*SG for girls with clinical measurements of TS 2+. The mean levels of 8-oxodG and F2-Isoprostanes were 117 (SD=49) nmol/L*SG and 7.81 (SD=4.77) nmol/L*SG for girls with clinical measurements of TS 1. There is no statistically significant difference between oxidative stress biomarkers and TS based on clinical measurements.

Repeated Measures Analysis of Urinary Oxidative Stress Biomarkers over 18-Months of Follow-Up

Figure 2 shows the distribution of value of urinary oxidative stress biomarkers by age at visit. The dash lines represents the prediction mean value from mixed model. There is a modest increase in the value of urinary 8-oxodG over age at visit (β=2.23, 95%CI=−0.73–5.21, p=0.14) (Figure 2A), while there is no increase in the value of urinary F2-Isoprostanes over age at visit (β=0.13, 95%CI=−0.16–0.43, p=0.36) (Figure 2B).

Figure 2. The levels of urinary oxidative markers by age at visit in the LEGACY Girls Study. A, Levels of urinary 8-oxodG (nmol/L*SG) and (B levels of urinary F2-Isoprostanes (nmol/L*SG).

A * Each dot represents the value of urinary 8-oxodG (nmol/L*SG) at age at visit (centered to the mean age)

** Dash line represents the mean level of predicted value from mixed model

B* Each dot represents the value of urinary F2-Isoprostanes (nmol/L*SG) at age at visit (centered to the mean age).

** Dash line represents the mean level of predicted value from mixed model

The results from the longitudinal analysis using mixed models are consistent with the cross-section analysis (Table 3). We found there was a positive interaction between age and BCFH on the level of 8-oxodG. In the longitudinal analysis using mixed models adjusting for age and BCFH, urinary 8-oxodG levels, but not F2-Isoprostane levels, increased with age at visit in BCFH+ girls (β=6.12, 95%CI=0.08–12.16, p=0.04) compared to BCFH− girls (Mode 3 of Table 3 and Figure 3A), but the association was no longer statistically significant after simultaneously adjusting for BMI (Model 4). Higher BMI was associated with level of F2-Isoprostanes (β=0.02, 95%CI=0.002–0.05), P=0.04) (Model 3).

Table 3.

Longitudinal analysis of the association between urinary oxidative stress biomarkers and pubertal development stage and BMI at the New York site of the LEGACY Girls Study

8-oxodG	Model 1		Model 2		Model 3		Model 4		Model 5
	β	CI	β	95% CI	β	95% CI	β	95% CI	β	95% CI
Age-centered (yrs)	2.23	−0.73, 5.21	2.67	−1.27, 6.61	−1.39	−5.97, 3.19	−0.30	−5.11, 4,52	−0.77	−6.16, 4.62
Breast Cancer Family History			5.04	−11.0–21.1	4.44	−11.55, 20.42	4.24	−12.79, 21.27	5.73	−11.11, 22.58
Breast Cancer Family History and age interaction					6.12	0.08, 12.16*	5.49	−0.89, 11.87	5.30	−1.04, 11.64
Age-specific BMI (cont.)							−0.05	−0.31, 0.21	−0.06	−0.32, 0.20
Breast Tanner Stage 2+ vs 1									2.89	−12.00, 17.78
F2-Isoprostanes	Model 1		Model 2		Model 3		Model 4		Model 5
	β	CI	β	95% CI	β	95% CI	β	95% CI	β	95% CI
Age-centered (yrs)	0.13	−0.16, 0.43	0.10	−0.19, 0.40	0.41	−0.03, 0.85	0.21	−0.20, 0.62	0.03	−0.49, 0.54
Breast Cancer Family History			1.06	−0.34, 2.47	0.93	−0.47, 2.33	1.14	−0.23, 2.51	1.08	−0.29, 2.45
Breast Cancer Family History and age interaction					−0.54	−1.13, 0.05	−0.46	−1.02, 0.09	−0.45	−1.00, 0.11
Age-specific BMI (cont.)							0.02	0.002, 0.05	0.02	−0.002, 0.004
Breast Tanner Stage 2+ vs 1									1.13	−0.77, 3.04

Model 1: Univariable mixed model with outcome as oxidative stress markers and covariate as age centered,

Model 2: Multivariable mixed model includes age centered in Model 1 plus BCFH,

Model 3: Multivariable mixed model includes age centered, and BCFH in model 2 plus interaction between age centered and BCFH,

Model 4: Multivariable mixed model includes age centered, BCFH, and interaction between age centered and BCFH in Model 3 plus BMI,

Model 5: Multivariable mixed model includes age centered, BCFH, interaction between age centered and BCFH, and BMI in model 4 plus Breast Tanner Stage,

^*p=0.047

^**p=0.03

Figure 3. The levels of urinary oxidative markers at age at visit in girls with and without a breast cancer family history at F0 in the LEGACY Girls Study. A, Levels of urinary 8-oxodG (nmol/L*SG) and (B) Levels of urinary F2-Isoprostanes (nmol/L*SG) at age at visit by BCFH.

A * Each dot represents the value of urinary 8-oxodG (nmol/L*SG) at age at visit (centered to the mean age)

** Dash line represents the mean level of predicted value from mixed model

B * Each dot represents the value of urinary 8-oxodG (nmol/L*SG) at age at visit (centered to the mean age)

** Dash line represents the mean level of predicted value from mixed model

DISCUSSION

Our results indicate that higher body mass index increases oxidative stress biomarkers over the pubertal window and that there are changes in 8-oxodG oxidative stress biomarkers in girls with a breast cancer family history compared to girls without a breast cancer family history. During pubertal development, under the influences of hormones, the breast undergoes proliferation and development (47–49). Human breast epithelial cell lines treated with 17 β-estradiol (E2) have been shown to result in increased oxidative DNA damage (50). Animal studies demonstrated that excess androgen receptor activation produces systemic oxidative stress (51, 52) and mice treated with testosterone had higher lipid peroxidation compared with control mice (53). Measuring endogenous reproductive hormones and plasma F2-Isoprostane in women aged 18–44, Schisterman et al. (54) reported F2-Isoprostanes levels were positively associated with estradiol after adjusting for other covariates. In addition, F2-Isoprostanes levels varied during the menstrual cycle (54). Overweight girls are more likely to start breast development at an early age compared to non-overweight girls (55). Our study also suggest that higher BMI is associated with higher oxidative stress. 8-Isoprostane, a prostaglandin (PG)-F2-like compound belonging to the F2 isoprostane family, was significantly associated with BMI in an obese pediatric population (56).

We found a positive interaction between age and BCFH in urinary 8-oxodG levels in both cross-sectional and longitudinal analysis. We previously reported that BCFH+ girls have an earlier onset of breast development than BCFH− girls (37). Compared with BCFH− girls, BCFH+ girls have higher concentrations of androgen including total testosterone, free testosterone and androstenedione (57). These findings suggest that in addition to androgen levels, BCFH+ girls might also differ on other biomarkers including higher oxidative stress during the period of the pubertal development.

The biomarkers we measured examined two different forms of oxidative stress that may be subject to different sources of ROS. 8-oxodG is the most commonly used marker of oxidative DNA damage (58). 8-oxodG is stable in urine but in DNA is known to lead to point mutations (59). In addition to oxidative stress, urinary 8-oxodG levels are also influenced by DNA repair rates (60, 61). It is possible that rapid cell division during breast development limits the repair mechanisms that excise 8-oxodG from DNA and ultimately leads to its excretion in urine. We previously reported that 8-oxodG levels were higher in postmenopausal women (62). Other studies also reported positive associations between age and 8-oxodG levels (63, 64). As urinary 8-oxodG is thought to be largely a result of DNA repair, increased levels of urinary 8-oxodG might be due to increase in oxidative stress as aging. Urinary levels of 8-oxodG are not significantly influenced by dietary sources or cell death (65).

F2-Isoprostanes are produced solely through the free radical oxidation of arachidonic acid, a ubiquitous polyunsaturated fatty acid. They are a sensitive and specific markers of lipid peroxidation (58). In contrast to oxidative DNA damage, level of F2-Isoprostanes is not influenced by DNA repair capacity (66). Urinary F2-Isprostanes have an advantage over plasma measurements in that there is no artificial production of F2-Isoprostanes during handling and storage of urine samples, and the levels were stable even when urine was maintained at room temperature for as much as 10 days (67). In a test of various biomarkers of oxidative stress after exposure to carbon tetrachloride only isoprostanes were consistently elevated and in a dose-dependent manner (68). We did not adjust for diet or physical activity which may be related to urinary F2-isoprostanes. (69)

Little is known about oxidative stress levels in young girls during the pubertal window. Comparing changes in oxidative stress levels within girls during follow-up permitted examination of within-individual changes over time which advances our understanding beyond cross sectional studies that use single spot urine samples. Further, because we were able to use prospective measures of pubertal development using Tanner Staging, both by maternal/guardian and clinician, we were able to temporally compare the biomarkers with pubertal development. Urinary, as opposed to blood biomarkers, provides a robust way to collect information on oxidative stress that may be more acceptable to children. We had a very high response rates for urine donation (98% at baseline) compared to only half of the girls who donated blood samples.(36) The decrease in the number of participants with each follow up visit is not due to loss-to-follow-up, but the fact that urine samples were analyzed while follow-up was ongoing; baseline enrollment (F0) was complete, and the majority of girls had reached the 6-month follow-up visit (F1) but only the girls who had joined the study early had reached the 12-month (F2) or 18-month (F3) follow-up visit.

Our study found that selected markers of oxidative stress increase over the pubertal window and are associated with higher body size. We also found higher levels of 8-oxodG in girls with a BCFH compared to girls without a BCFR. Differences between the oxidative stress markers may be due to other factors that influence these markers, particularly as we did not consider further adjustment for diet and physical activity. Larger prospective studies of the factors associated with oxidative stress trajectories during adolescence may be relevant to understanding the mechanisms contributing to the familial clustering of breast cancer.

Abbreviations:

8-oxodG

8-oxo-7 8-dihydro-2’-deoxyguanosine

BCFH

breast cancer family history

BMI

body mass index

LEGACY

Lessons in Epidemiology and Genetics of Adult Cancer from Youth

ROS

reactive oxygen species

SG

specific gravity

TS

Tanner Stage

WOS

window of susceptibility