In Utero DDT Exposure and Breast Density Before Age 50

Abstract

Prior studies in the Child Health and Development Studies (CHDS) found in utero exposure to the pesticide, dichlorodiphenyltrichloroethane (DDT), increased breast cancer risk by age 52. Mammographic density is considered a primary risk factor for breast cancer. We conducted a study of 309 daughters from the CHDS to examine in utero DDT exposure and mammographic density in midlife. Among daughters with high (>75th percentile) exposure to p,p’-Dichlorodiphenyldichloroethylene (DDE), p,p’-DDT was significantly correlated with increased dense area and percent density regardless of her body mass in midlife. In the subset of women with lower (<75th percentile) p,p-DDE, p,p’-DDT was associated with increased non-dense breast area. This was explained by adjustment for midlife BMI suggesting that p,p’-DDT may be obesogenic. In aggregate our findings indicate that early life p,p’-DDT exposure impacts breast density in a complex way that depends on the hosts biological ability to sequester and process DDT and levels of exposure.

1. Intro

Previously, Child Health and Development Studies (CHDS) daughters were found to have increased risk of breast cancer by age 52 associated with in utero exposure to the pesticide, 2,4-dichlorodiphenyltrichloroethane (o,p’-DDT) [1]. Technical DDT consisted of p,p’-DDT (the primary component), o,p’-DDT (a minor component/contaminant) and its major degradation product is p,p’-dichlorodiphenyldichloroethylene (DDE) While previous in utero exposure to environmental contaminants cannot be changed there might be steps along the life course that can be modified to reduce the health risk caused by exposure. Additionally, future in utero exposures can be prevented using policy that weighs both the benefits and potential long term risks of using DDT in chemical vector control. As stated by Bouwman (2011) there are situations where DDT will provide the best achievable health benefit as a malaria control measure, but maintaining that DDT is safe ignores the results of many studies [2].

Density is considered an important intermediate marker of breast cancer risk [3] and a potentially modifiable risk factor [4, 5]. For example, pharmaceuticals such as tamoxifen have been shown to change breast density [4, 5]. Epidemiologic studies have consistently supported a 4 to 6-fold increase in breast cancer incidence for women with dense breast tissue for up to 10 years following measurement [6–11]. Mammographic density (75% or greater) has also been associated with a nine-fold increase in carcinoma in situ or atypical hyperplasia and a twelve-fold increase in risk of developing hyperplasia [3]. Increased risk associated with mammographic density has been shown for both premenopausal and postmenopausal breast cancer [9]. Density is shaped by cumulative hormonal exposures throughout life, but in particular during the intrauterine, menarche, pregnancy and lactation time periods [12].

Body Mass Index (BMI) has a complex relationship with breast cancer risk and breast density. BMI is associated with increased breast fat [13] and lower percent breast density. BMI before menopause has an inverse relationship with both breast density and breast cancer risk [14, 15]. BMI after menopause is positively associated with postmenopausal breast cancer risk, whereas percent breast density is lower after menopause [14, 16]. Evidence suggests that examining the window of susceptibility when mammary tissue is developing and differentiating is critical in establishing breast cancer risk [16] and in untangling the nexus between breast density and breast cancer. These windows include: in utero, childhood, puberty, before first term pregnancy, during pregnancy and during the perimenopause [16–18]. These periods are characterized by different structure, function and hormonal control mechanisms and thus risk factors for breast cancer during each period might differ. Predictors of breast density and the role of breast density in the etiology of breast cancer may depend on age, biological stage, and underlying vulnerability, and may interact with other risk factors such as genetic susceptibility and exposures during critical windows.

DDTs have been shown to change cancer risk and also act as obesogens [1, 19–21]. In utero exposure to DDT and its break-down product DDE are associated with increased BMI in humans [19, 20]. Animal studies suggest DDE may affect offspring obesity due to its effects on lipid metabolism [20, 22, 23] Thus DDTs may impact breast density via increases in BMI. It is also possible that DDTs may exert an impact on breast density via alternative, BMI-independent pathways. Just as BMI associations with breast cancer can depend on age and menopausal status at the time of diagnosis, DDT associations with breast density may depend on age and menopausal status or other underlying susceptibility features including genetic predisposition. In a companion analysis, McDonald et al. (2019) investigates whether family history of breast cancer modifies associations between in utero of DDT exposure and daughter’s mammographic density women in early menopause [24]. Here, we test whether mammographic density is affected by in utero exposure to environmental DDT compounds in women in their 40s. We also investigate the role, if any, of daughter’s BMI in mediating or modifying this association given that DDTs are obesogens.

2. Methods

2.1. Study Population

The Child Health and Development Studies (CHDS) is a pregnancy cohort designed to examine the association between prenatal exposures and health and development over the life course for parents and children. The CHDS recruited women residing in the Oakland, California area who were members of the Kaiser Permanente Foundation Health Plan and who sought obstetric care for pregnancies between 1959 and 1966 (with deliveries extending into 1967) [25]. More than 98% of all eligible women enrolled. The CHDS founding mothers voluntarily participated in an in-person interview and gave permission to access their own medical records and those of their children to researchers. The mothers’ blood specimens were collected at several times through pregnancy and 1–3 days after delivery. The CHDS regularly links to the California Cancer Registry (CCR). Maternal breast cancer cases were identified using linkage to CCR records to identify cancer diagnoses, including year and age at diagnosis. Daughter breast cancer cases were identified by both linkage to the California Cancer Registry and by self-report during a computer-assisted telephone interview in an adult follow-up study of CHDS daughters from 2010 to 2013[1].

Between 2005 −2007, 567 daughters born into the CHDS participated in an adult follow-up study described by Terry et al. (2011) [26]. A 45 minute telephone interview was conducted to collect personal health history and medication use, family history of cancer, sociodemographic factors, alcohol and tobacco use, detailed reproductive history and anthropometric measures. Yearly mammographic density screening was recommended as standard practice beginning at age 40 from 1997 to 2015 [27] which encompasses the time frame of the present study. We collected and were able to read 348 digital screening mammograms (61%) of the 567. Of the 348, 12 were missing on menopausal status, 9 women were previously diagnosed with breast cancer, 8 were missing weight at interview, 7 were missing maternal weight, two were missing lipid measures, and age at menopause was unknown for one woman. Our analysis includes 309 CHDS daughters who had never been diagnosed with breast cancer, had adult mammographic density data and maternal perinatal serum measurements of DDT and lipids. Of the 309 women, 192 had no siblings in the study and 117 had at least one sibling. There were 33 women, 11% of our sample, with mothers who had a breast cancer diagnosis. The present study was reviewed and approved by the Institutional Review Boards of the Public Health Institute and Columbia University.

2.2. Samples and Assays

Blood samples were collected from pregnant women in each trimester and in the early post-partum period between 1959 and 1967, processed to isolate serum, and stored since then at −20 Celsius. DDT-related compounds (p,p’-DDT and o,p’-DDT, and p,p’-DDE) were examined for a number of different studies and assay methods are described in previous publications [1, 28]. Briefly, samples were measured using gas chromatography with electron capture detection (GC/ECD) with a capillary column (10%) or GC/ECD with dual columns (86%) or using gas chromatography triple quadrupole mass spectrometry (4%). Sholtz et al. 2011 [28] showed DDT congeners measured in different labs and using different techniques could be combined without risking meaningful misclassification due to variation. This allows us to conserve irreplaceable archived serum samples.

Total cholesterol and triglycerides were measured enzymatically in a lab certified by the Centers for Disease Control and Prevention/National Heart, Lung, and Blood Institute Lipid Standardization Program.

2.3. Mammographic Density Data

Collection and measurement of screening mammograms are described by Terry et al. (2011) [26]. Briefly, researchers requested films from radiological facilities of the participants, digitized the films using a Kodak Lumisys Film Digitizer (Kodak LS85), and promptly returned the films to the facilities. We assessed mammographic density using Cumulus, a computer-assisted thresholding program [29] in which the reader outlines the total breast area and dense area, and the software measures the size by identifying the number of pixels within the outlined areas. We calculated absolute breast area and dense area by converting the measure from pixels to cm². Non-dense breast area is calculated by taking absolute breast area and subtracting dense area. Percent mammographic density was calculated as dense area divided by total breast area multiplied by 100. If both left and right images were available, we only read left breast images. We used the film taken closest to the date of interview in our analyses.

2.4. Statistical Analysis

Dense area, non-dense area, percent density, o,p’-DDT and p,p’-DDT were natural log transformed. p,p’-DDE was categorized according to quartiles based on the distribution among the 567 women who participated in the adult follow-up study. Quartile cutpoints for the 567 were nearly identical to those for the 309 in the analysis sample. For analysis, DDE was dichotomized as ≤75^th percentile (≤58 ng/mL) versus >75^th percentile (>58 ng/mL). Associations were estimated in Proc Mixed, SAS using log-log models to adjust for sibling clustering using SAS 9.3 (SAS Institute Inc., Cary, NC). Our starting models were models that included variables found to be related to mammographic density and breast cancer risk from previous CHDS studies and in the literature. Using backward selection, we generated the most parsimonious models. Models 1 and 2 were adjusted for mother’s age during pregnancy (years), and mother’s weight (dichotomized as body mass index,BMI ≥25kg/m², vs. BMI <25kg/m²), o,p’-DDT (continuous natural log transformed), mother’s total cholesterol (continuous, mg/dL), mother’s total triglycerides (continuous, mg/dL), daughter’s age at mammogram (continuous, years). Model 2 was further adjusted for daughter’s weight at interview (BMI ≥ 25kg/m², vs. <25kg/m²). Further models were run adjusting for maternal breast cancer history and menopausal status, but did not change results (results not shown). Model 1 and model 2 were also run using continuous BMI and results did not change (Supplemental Table 1).

3. Results

The maternal and daughter characteristics of the analysis subset are presented in Table 1. Mothers were a median age of 26 years when they were pregnant with their daughters. Daughters were a median age of 44 years at study interview and similarly had a median age of 43 years at the time of their mammogram. At time of study interview, most daughters had not entered menopause (72%).Table 1 also shows distributions for maternal perinatal serum levels of DDTs. It has been established that organochlorine levels are consistent across all trimesters of pregnancy and soon after delivery [30], thus these levels represent proxy in utero exposures for offspring daughters.

Table 1.

Maternal and daughter characteristics of the analysis subset (n=309)

F0	Min	25%	Median	75%	Max
Height (m)	1.42	1.6	1.63	1.68	1.78
Weight (kg) *	44	55.03	60.69	69.33	102.58
BMI†*	16.86	21.26	22.72	25.42	44.17
p,p’-DDT (ng/ml)	0.60	6.00	9.12	14.10	84.90
p,p’-DDE (ng/ml)	11	32	46	61	163
o,p’-DDT (ng/ml)	0.0043	0.21	0.36	0.64	3.59
Total Cholesterol (mg/dL)	75	200	240	290	560
Triglycerides (mg/dL)	45	152	190	245	56 ‘
Age at birth of daughter	16	23	26	31	‘3
	N (%)
BMI ≥ 25 kg/m2	89 (29)
Breast Cancer	33 (11)
F1	Min	25%	Median	75%	Max
Height (m)Δ	1.49	1.6	1.65	1.73	1.88
Weight (kg)Δ	41.40	63.00	72.00	89.10	154.80
BMI†	14.01	23.11	26.24	31.37	62.42
Age at interview	39	42	44	44	47
Age at mammogram	34	42	43	45	47
Dense area	1.41	17.99	30.43	45.46	127.71
Non-dense area	10.12	44.75	84.33	137.73	431.35
Percent Density	1.26	13.29	29.71	46.38	79.09
	N (%)
BMI ≥ 25 kg/m2	181 (59)
Menopausal statusΔ
Premenopausal	224 (72)
Transition	49 (16)
Post-menopausal	36 (12)
Race
White	229 (74)
Black	43 (14)
Hispanic	19 (6)
Asian	10 (3)
Mixed	6 (2)
Other	2 (1)

^*Adjusted weight for 101 days of pregnancy

^†BMI = weight (kg)/[height(m)]²

^Δat interview

High (>75th percentile) co-exposure to p,p’-DDT and p,p’-DDE in utero predicted higher dense area (Table 2 and Figure 1a) and higher percent density (Table 2 and Figure 1c). Among daughters who had both high p,p’-DDE and high p,p’-DDT exposures in utero, p,p’-DDT was significantly positively correlated with dense area and percent density regardless of her BMI in midlife. Model 2 shows that each 10 percent increase in in utero p,p’-DDT in women with high p,p’-DDE, there was a predicted 2.5 percent increase in dense area of the breast and a 2.9 percent increase in percent density. Further, we estimated the change in percent density associated with the mid-points of the interquartile range for the natural log of p,p’-DDT within the 4th quartile of natural log p,p’-DDE. Among women who were highly exposed to p,p’-DDT (median of 4th quartile vs. median of 1st quartile) within the highest exposure of p,p’-DDE (4th quartile), we observed 6% increase in percent density (95% CI=1.5% to 10.8%, p-value=0.01).

Table 2.

In utero p,p’-DDT and adult breast density before age 50 by in utero p,p’-DDE levels

	ln p,p’-DDTf Correlations with Breast Density Parameters
	In Dense Areag			In Non-Dense Areag			In Percent Densityg
		95% CI			95% CI			95% CI
	coefficient	lower	upper	coefficient	lower	upper	coefficient	lower	upper
Model 1a
p,p’-DDE
Lowd	−0.13	−0.31	0.05	0.25	0.07	0.42	−0.26	−0.48	−0.04
Highe	0.29	0.05	0.53	−0.17	−0.41	0.06	0.39	0.10	0.68
p-value for interactionc	<0.005			<0.005			<0.001
Model 2b
p,p’-DDE
Lowd	−0.09	−0.27	0.10	0.09	−0.06	0.24	−0.12	−0.32	0.08
Highe	0.26	0.03	0.50	−0.06	−0.26	0.13	0.30	0.04	0.57
p-value for interaction	<0.02			NS			<0.02

Bold coefficients indicates significant finding (p<.05); ln, natural log

^a.adjusted for age at mammogram, o,p’-DDT ng/ml (natural log transformed), maternal age at pregnancy, maternal weight adjusted weight for 101 days of pregnancy (Body Mass Index (BMI) ≥ 25 kg/m² vs. BMI <25 kg/m²), maternal total serum cholesterol and total serum triglycerides.

^b.additional adjustment of daughters weight at interview defined as BMI ≥ 25 kg/m² or BMI <25 kg/m².

^c.The interaction is tested by a product term of the p,p’-DDT and p,p’-DDE variables.

^d.High p.p’-DDE is defined by >75^th percentile, >58 ng/mL

^e.Low p.p’-DDE is defined by <75^th percentile, ≤58 ng/mL

^f.Natural log transformed; ng/ml

^g.Natural log transformed; Dense Area and Non-Dense area, cm²

Figure 1.

In Utero p,p’-DDT and Adult Breast Density Parameters Before Age 50 by in Utero p,p’-DDE quartile

Panel A. Dense Area

Panel B, Non-dense Area

Panel C. Percent Density

High p,p’-DDT in daughters with lower perinatal p,p’-DDE (<75^th percentile) predicted higher non-dense breast area (Table 2, model 1 and Figure 2b). However, adjusting for BMI ≥ 25kg/m² accounted for this association (Table 2, Model 2). In the subset of women with lower p,p-DDE, the associations between in utero exposures to p,p’-DDT and non-dense breast area in adult women support evidence that DDT acts as an obesogen and impact non-dense breast tissue by increasing fat in the breast. The coefficient for daughters’ BMI in Model 2, for the non-dense area outcome provides a quantitative estimate of the impact of daughter’s BMI on non-dense area: β = 0.89, 95% CI= 0.7446 to 1.04, p-value < .0001)

Interactions between p,p’-DDT and p,p’-DDE were statistically significant for both dense area (p<0.02) and percent density (p<0.02). We did not see an association between o,p-DDT exposure and breast density. It is worth noting however, when we used the threshold for low o,p-DDT exposure based on a previous breast cancer study in daughters [1], we did see a significant association only with non-dense area of the breast in this sample (β=0.27, 95% CI= 0.06 to 0.48, p-value=0.01; Supplemental Table 1).

Findings were not explained by menopausal status or maternal age, maternal serum lipids or maternal body mass. Adjusting for maternal breast cancer history, menopausal status at interview or race did not change results. Adjusting for continuous maternal or daughters BMI did not change the significance or interpretation of our results (Supplemental Table 2).

4. Discussion

While many epidemiologic studies have not reported associations between DDT exposure and breast cancer risk [31, 32], many of those studies fail to measure DDT congeners [32, 33] and fail to look at windows of susceptibility [17]. Both are essential to understanding the relationship of environmental exposures to breast cancer risk and intermediate markers of breast cancer risk. The CHDS provides a unique opportunity to study both. Previous CHDS research has already shown a relation between in utero DDT exposure and risk of breast cancer in daughters [1]. Examining intermediate risk factors at different biological time points during periods of breast development and proliferation when the cells are more susceptible to programming errors [34], offers the possibility of understanding the path to breast cancer and also an opportunity to find an intermediate point for intervening to modify risk before disease onset.

Of the women who were highly exposed to p,p’-DDT (median of 4th quartile vs. median of 1st quartile) within the highest exposure of p,p’-DDE (4th quartile), we observed 6% increase in percent density. This difference in breast density associated with high exposure is important given that breast density is a strong intermediate marker for breast cancer [35, 36]. In a recent examination of the association between percent breast density and the long-term excess risk of breast cancer, Rebolj et al. (2019) found a 3% increase in breast cancer risk per percentage point increase in percent density [37]. Applying this risk estimate to our findings would suggest that women highly exposed to DDTs in utero could have in excess of an 18% increased risk of breast cancer.

Our o,p’-DDT findings for non-dense area suggest that there are different effects for the different DDTs: high co-exposure to p,p’-DDT and DDE were associated with percent density and dense area, but higher o,p’-DDT exposure was associated with non-dense area. Additional evidence of differential effects for the different DDT compounds is provided in our companion paper in this issue showing an o,p’-DDT association with dense area but only in daughters with a maternal history of breast cancer [24]. Taken together these findings suggest a complex role for DDT, depending on timing of exposure, target tissue and outcome measure.

This study highlights the importance of measuring all DDT related compounds as they have varying effects at different concentrations on different aspects of breast density in women. The underlying explanation for a differential relationship between levels of p,p-DDT and p,p’-DDE are unknown but could result from one or more underlying mechanism including: 1) an indicator of chronic and current exposure to the technical pesticde [38, 39]; 2) biological ability to metabolize p,p’-DDT to p,p’-DDE; and/or 3) the ability to sequester p,p’-DDT in fat before metabolizing it to p,p’-DDE.

In this study percent density of women was affected by in utero exposures to DDTs in two ways: increased dense area from high co exposure to p,p’-DDT and p,p’-DDE; and increased non-dense area with increased p,p’-DDT levels in women with low DDE. Adjusting for BMI accounted for the p,p’-DDT association with non-dense area (fat and stroma). Our findings provide evidence that high levels of p,p’-DDT may act as an obesogen, increasing BMI and non-dense area. Previous studies have found different results for obesity looking at DDT congeners. A meta-analysis of prospective human studies across the world demonstrates a consistent positive association between prenatal exposure to the DDT pesticide metabolite DDE and childhood obesity [20, 40]. A study in young children recently exposed to DDTs, found a positive association with p,p’ DDT and weight and body composition in girls [41]. The CHDS cohort found that o,p’-DDT was associated with higher BMI and waist circumference in midlife [42]. Possible pathways for DDT associations with obesity include impaired adipocyte differentiation and dysfunction [41, 43] and changes in the epigenome [44]. We find DDT increases non-dense area of the breast via increased BMI, lending supportive evidence for the obesogenic effects of DDT compounds. Both higher postmenopausal BMI and dense breasts are considered breast cancer risk factors. DDTs may be playing dual and corresponding roles in breast cancer risk by increasing both breast density (dense area and percent density) and late life BMI. DDTs potentially can be increasing breast cancer risk both in the premenopause as indicated by the DDT-associated risk with dense area and percent density as seen in this study; and potentially in the postmenopause through increased BMI. To further complicate the narrative, it has been shown that higher pre-menopausal and adolescent BMI is associated with reduced risk of all breast cancer [13, 15, 45]. Additionally there is evidence of a strong association for increased non-dense area with lower breast cancer risk [13], suggesting that fat at younger ages may play a protective role against breast cancer.

High levels of both p,p’-DDT and p,p’-DDE could be reflective of chronic and/or current exposure [38, 39]. Even so, animals, including humans, have developed defenses against natural and synthetic carcinogenic toxins [46, 47]. Since DDTs are lipophilic, obesity and higher body fat might be the body’s own protection against possible toxic DDT effects. Women who do not store DDTs in their fat as effectively might be the women we see with the highest serum levels of p,p’-DDT and p,p’-DDE. Women who are unable to utilize fat to store DDTs in pregnancy would theoretically have higher circulating DDT levels and would expose their daughters to higher levels in utero. The differences in p,p’-DDT effects on breast density parameters by p,p’-DDE level might also be an indicator of susceptible maternal metabolism. Relative levels of DDT and DDE resulting from differential metabolism could be used to create a metric for identifying women at higher risk of breast cancer. Future studies, particularly metabolomics, would need to be done to disentangle the genetic, epigenetic and exposure interactions that lead to differences in breast density parameters related to in-utero exposure to DDTs.

Highlights.

• DDTs may alter breast cancer risk through DDT-related changes in breast density.

• In utero DDT may affect breast cancer depending on co-exposure to DDT compounds.

• Findings support the hypothesis that DDTs can act as obesogens.

Abbreviations:

CHDS

Child Health and Development Studies

DDT

dichlorodiphenyltrichloroethane

DDE

Dichlorodiphenyldichloroethylene

BMI

Body Mass Index