Pilot Study on the Urinary Excretion of the Glyphosate Metabolite Aminomethylphosphonic Acid and Breast Cancer Risk: The Multiethnic Cohort Study

Abstract

Breast cancer is the most commonly diagnosed female cancer and the second leading cause of death in women in the US, including Hawaii. Accumulating evidence suggests that aminomethylphosphonic acid (AMPA), the primary metabolite of the herbicide glyphosate—a probable human carcinogen, may itself be carcinogenic. However, the relationship between urinary AMPA excretion and breast cancer risk in women is unknown. In this pilot study, we investigated the association between pre-diagnostic urinary AMPA excretion and breast cancer risk in a case-control study of 250 predominantly postmenopausal women: 124 cases and 126 healthy controls (individually matched on age, race/ethnicity, urine type, date of urine collection, and fasting status) nested within the Hawaii biospecimen subcohort of the Multiethnic Cohort. AMPA was detected in 90% of cases and 84% of controls. The geometric mean of urinary AMPA excretion was nearly 38% higher among cases vs. controls (0.087 vs 0.063 ng AMPA/mg creatinine) after adjusting for race/ethnicity, age and BMI. A 4.5-fold higher risk of developing breast cancer in the highest vs. lowest quintile of AMPA excretion was observed (OR_{Q5 vs. Q1}: 4.49; 95% CI: 1.46–13.77; p_trend=0.029). To our knowledge, this is the first study to prospectively examine associations between urinary AMPA excretion and breast cancer risk. Our preliminary findings suggest that AMPA exposure may be associated with increased breast cancer risk; however, these results require confirmation in a larger population to increase study power and permit careful examinations of race/ethnicity differences.

Graphical Abstract

Odds ratio of breast cancer as a function of quintiles of urinary aminomethylphosphonic acid excretion (Q1=lowest, Q5=highest quintiles).

Introduction

Breast cancer is the most commonly diagnosed female cancer and the second leading cause of death in women in the US, including Hawaii [1, 2]. Estrogen is a well-known contributor to the development of hormone-dependent breast carcinomas [3, 4]. As such, any factor that interferes with estrogen homeostasis, theoretically, has the ability to influence breast cancer risk and, thus, merits further study.

Endocrine disrupting chemicals (EDCs) are exogenous compounds that interfere with normal hormonal actions in humans [5] and consist of a very heterogenous group of compounds that include, but are not limited to, pesticides (methoxychlor, chlorpyrifos, dichlorodiphenyltrichloroethane) and nonpersistent environmental EDCs such as bisphenol A (BPA; used during the manufacturing of plastics/resins), phthalates (used as plasticizers), triclosan (TCS; antibacterial/antifungal agents), and parabens (used as preservatives) ubiquitously present in modern times [5].

Aminomethylphosphonic acid (AMPA) is the major metabolite resulting from the soil microbial degradation of glyphosate (GLYP) [6, 7], currently the most widely used herbicide worldwide and the subject of ongoing contentious debate regarding its human carcinogenicity among international regulatory bodies [8] with AMPA being more often detected than GLYP [10–12]. AMPA is also the key metabolite of phosphonates found in several household laundry and cleaning detergents [9, 10], which may explain its widespread occurrence in the environment [10–13].

In 2015, the International Agency for Research on Cancer (IARC) cited evidence from in vitro and animal studies showing that AMPA could induce oxidative stress [14]—a characteristic for human carcinogens [15]. In addition, a recent, large study (n>30,000) of farmers’ wives found significant associations between the use of several organophosphate insecticides and elevated breast cancer risk [16]. However, the metabolites of these compounds were not analyzed.

Pesticide metabolites have often been found to be more mobile, persistent, and detrimental than their respective parent compound [17] and, thus, may pose equal or higher risks to humans upon exposure. While most pesticides are ultimately converted to components with less or no toxicity, certain chemical characteristics (i.e., water solubility, vapor pressure) can enable metabolites to become biologically active [17]. AMPA is very water soluble (5.8 g/L near 25°C in water) [18], and has been detected in human urine of non-occupationally exposed humans [19–22] often in a higher percentage of samples than GLYP [19–21].

To complement our ongoing investigations on the associations between breast cancer risk and urinary excretion of EDCs (that currently consists of BPA, phthalates, TCS, and parabens), we sought to expand our research in this pilot study to AMPA, using the same urine samples from the same women within the Multiethnic Cohort (MEC).

Methods

Study population and data collection

This study utilized urine samples from a nested case-control study of breast cancer among mostly postmenopausal women participating in the Hawaii biospecimen subcohort of the MEC. The MEC is a prospective study designed to investigate the associations of dietary, lifestyle, and genetic factors with the incidence of cancer. Study details have been described elsewhere [23]. Briefly, between 1993 and 1996, over 215,000 men and women, ages 45–75 years from five racial/ethnic groups (Japanese American, Native Hawaiian, White, African American, Latino) from Hawaii and Los Angeles, California, were enrolled in the cohort. Potential participants were identified through drivers’ license files, voter registration lists, and Medicare files. At cohort entry, participants completed a self-administered, 26-page questionnaire that included queries on demographic characteristics, anthropometric measures, medical history, family history of cancer, physical activity, dietary habits, supplement use, and reproductive history for women.

The prospective MEC biospecimen subcohort was established from 2001 to 2006 among the surviving cohort members who agreed to provide blood and/or urine specimens and to answer a short questionnaire [24]. Biospecimens were prospectively collected from 36,458 mostly postmenopausal women, 32,243 of whom provided a urine sample. Most participants in Hawaii provided an overnight specimen while those in Los Angeles donated mainly a first morning urine specimen; spot urine was requested otherwise. At the time of urine collection, updated information was collected on tobacco use, weight, medication, supplements, and hormone use. The study protocol was approved by the institutional review boards of the University of Hawaii and the University of Southern California. Every participant gave written consent.

Case ascertainment and control selection

Incident invasive breast cancer cases were identified through linkage to the National Cancer Institute’s Surveillance, Epidemiology, and End Results Program registries covering Hawaii and California. Breast cancer diagnoses were classified using the International Classification of Diseases for Oncology, Third Edition codes C50.0-C50.9 and were restricted to invasive malignancies. For this analysis, we selected 124 Hawaii breast cancer cases and 126 matched controls from an ongoing nested case-control study of breast cancer within the MEC with available urine [25], including 123 matched sets. Cases were selected among those diagnosed after urine collection through December 31, 2016. The time between urine collection and breast cancer diagnosis averaged (± standard deviation) 5.5±3.3 years, with minimum of 0.08 and maximum of 12.2 years. One control was selected for each case and was randomly selected from the eligible pool of women who were alive and free of a breast cancer diagnosis at the time of the case’s diagnosis. Controls were matched to cases on birth year (± 1 year), race/ethnicity (Japanese American, Native Hawaiian, White), date of urine collection (± 1 year), time of blood draw (± 2 hours), and hours fasting prior to blood draw (8–10, >10 hours).

Urinary AMPA and creatinine analysis

All assays were performed at the Analytical Biochemistry Shared Resource of the University of Hawaii Cancer Center. The assays were blinded to the analysts regarding case/control status, and samples were analyzed in randomized order, but samples of cases and their respective controls were always kept in the same analysis batch. Urinary AMPA concentrations were measured by a novel ultra-sensitive liquid chromatography high-resolution accurate-mass mass spectrometry (LC/HRAM-MS) method that included derivatization with 9-fluorenylmethoxycarbonyl chloride (Fmoc-Cl) and solid phase extraction (SPE) purification of the resulting adduct [26]. In brief, 0.5 mL of urine were mixed with the internal standard AMPA-¹³C-¹⁵N-d₂ (500 ng/mL in water) and diluted with 1 mL of water followed by reversed phase SPE with polymeric sorbents (Strata-X; Phenomenex, Torrance, CA). The SPE load and wash fractions were mixed with 100 μL of sodium borate buffer (2.6% in water) and 150 μL of Fmoc-Cl (6 mg/mL in acetonitrile) and incubated at 50°C for 20 minutes. After cooling to room temperature, 5 μL of formic acid was added to the reaction mixture and centrifuged. The supernatant was then purified with a SepPak C18 reverse phase SPE cartridge (Waters, Milford, MA), separated by liquid chromatography using a Kinetex EVO C18 analytical column (Phenomenex) and analyzed by LC/HRAM-MS (Q-Exactive, Thermo Scientific, Waltham, MA) with targeted selected ion monitoring by detecting the monoisotopic masses of the protonated analytes (± 5 ppm to account for mass spectrometric inaccuracies). The between-day precision of this assay using pooled urine samples spiked with 0.4 ng/mL AMPA was excellent (CV=5%.) Values below the lower limit of quantitation (LLOQ; 0.001 ng/mL) were assigned a value of 0.0005 ng/mL to allow statistical evaluations. Urinary AMPA concentrations were adjusted for creatinine levels to normalize for urine volume [27]. Urinary creatinine was measured by a Cobas MiraPlus clinical chemistry auto analyzer (Roche Diagnostics) using a commercially available creatinine kit (Randox Laboratories, Crumlin, UK) based on the Jaffé reaction. The between-day precision of this assay using pooled urine samples was 8%. The calibration curve was highly linear (r²>0.995).

Statistical Analysis

Breast cancer cases and controls were compared with respect to several demographic characteristics and potential risk factors of interest. Adjusted and unadjusted geometric means of AMPA excretion values were compared between cases and controls, by race/ethnicity, age groups (≤64 y, 65–74 y, 75 y and older), and body mass index (BMI; normal: 18.5–24.9, overweight: 25–29.9, obese: ≥30) at urine collection, using general linear models and a pairwise t-test.

The association of urinary AMPA excretion with the risk of breast cancer was examined using conditional logistic regression with matched sets as strata. AMPA excretion was categorized into quintiles based on the controls. Odds ratios (OR) and 95% confidence intervals (CI) were computed for each quintile, using the lowest quintile as a reference. Models were adjusted for potential confounders of postmenopausal breast cancer risk, including BMI at urine collection (continuous), family history of breast cancer (yes, no), age at menarche (<13, 13–14, ≥15 or missing), age at first live birth (≤20 or missing, 21–25, 26–30, ≥31 or nulliparous), menopausal status (pre-, postmenopausal), hormone use (ever, never), tobacco smoking (never, former, current smoker), daily alcohol intake (0, <12 g/d, ≥12 g/d, missing), education (less than college degree, college degree or higher), mammography screening (ever, never), moderate to vigorous physical activity (hours/day, continuous). Dietary factors such as total daily energy intake, healthy eating index-2010 diet score [28], and daily caffeine intake were also considered, but not included in the final model as their inclusion did not change the risk estimates by more than 10% [29]. Linear trend was evaluated by fitting a model with the nominal trend variable. Heterogeneity in associations by race/ethnicity was tested using the Wald test of an AMPA-race/ethnicity cross-product term in the logistic model.

Analyses were conducted using SAS version 9.4 (SAS Institute, Inc., Cary, North Carolina). All tests were two sided, and P < 0.05 was considered statistically significant.

Results

Study characteristics of breast cancer cases and controls are shown in Table 1. The largest racial/ethnic group among both cases and controls was Japanese Americans (n=65 cases, n=68 controls), followed by Whites (n=40 cases, n=39 controls), and Native Hawaiians (n=19 cases, n=19 controls). Cases were less likely than controls to have started menarche before 13 years old or to be parous; however, these differences were not statistically significant. The majority of participants were postmenopausal at baseline (81%) and reported no history of breast cancer among first degree female family members (86% cases, 83% controls). Almost all participants (97% cases, 96% controls) reported ever having a mammogram.

Table 1.

Characteristics of the Hawaii participants of the breast cancer nested case-control study in the Multiethnic Cohort^a

	Cases (n=124)	Controls (n=126)	P-valueb
	n (%)c	n (%)c
Age at urine collection, y
≤64	58 (47)	55 (44)
65–74	46 (37)	54 (43)
≥75	20 (16)	17 (14)	0.62
Ethnicity
Japanese American	65 (52)	68 (54)
White	40 (32)	39 (31)
Native Hawaiian	19 (15)	19 (15)	0.97
Education
≤High school graduate	47 (38)	57 (45)
Some college	30 (24)	24 (19)
College graduate	21 (17)	23 (18)
Graduate school	26 (21)	22 (18)	0.73
Family history of breast cancer
No	107 (86)	105 (83)
Yes	16 (13)	15 (12)
Unknown	1 (1)	6 (5)	0.16
BMI at urine collectiond
18.5–24.9	74 (59)	68 (55)
25–29.9	31 (25)	36 (29)
30 or higher	21 (17)	20 (16)	0.82
Age at menarche, y
<13	59 (48)	70 (56)
13–14	49 (40)	43 (34)
>14	16 (13)	12 (10)
Unknown	0 (0)	1 (1)	0.41
Menopausal status
Pre-menopausal	24 (19)	24 (19)
Post-menopausal	100 (81)	102 (81)	0.95
Age at first live birth, y
missing	0 (0)	3 (2)
Nulliparous	20 (16)	13 (10)
≤20	21 (17)	25 (20)
21–25	44 (36)	45 (36)
26–30	29 (24)	35 (28)
≥31	10 (8)	5 (4)	0.22
Number of children
0	20 (16)	13 (10)
1	16 (13)	15 (12)
2–3	70 (57)	65 (52)
≥4	17 (14)	32 (25)
missing	1 (1)	1 (1)	0.18
Hormone replacement therapy use
Never used	45 (36)	53 (42)
Ever used	79 (64)	73 (58)	0.35
Smoking status
Never smoker	73 (59)	78 (62)
Former smoker	35 (28)	38 (30)
Current smoker	16 (13)	10 (8)	0.44
Alcohol intake, g/day
None	67 (54)	76 (60)
<12	37 (30)	35 (28)
≥12	19 (15)	15 (12)
missing	1 (1)	0 (0)	0.56
Mammography screening
Never had a mammogram	4 (3)	5 (4)
Ever had a mammogram	120 (97)	121 (96)	0.75

^aCharacteristics at baseline questionnaire, unless specified.

^bP-values are based on the chi-square test for categorical variables and on the t-test for continuous variables.

^cpercentages may exceed 100 due to rounding

^dBody Mass Index, kg/m².

The percentage of values above LLOQ was 90% and 84% among breast cancer cases and controls, respectively (Table 2). Geometric mean urinary AMPA excretion after creatinine-adjustment was nearly 40% higher among cases than controls (0.087 vs 0.063 ng AMPA/mg creatinine) after adjusting for race/ethnicity, age, and BMI, and there was higher AMPA excretion among the 65–74 y age group (0.086) compared to ≤64 y and ≥75 y age groups, (0.068 and 0.070, respectively, but these differences were not statistically significant. Overweight women (BMI 25.0–29.9) had non-significantly higher AMPA excretion (0.092), followed by normal-weight (BMI 18.5–24.9) and obese women (BMI≥30), (0.075 and 0.057, respectively). There were no differences in AMPA excretion by race/ethnicity after adjusting for age, BMI and case/control status.

Table 2.

Aminomethylphosphonic acid (AMPA) excretion among Hawaii participants of the breast cancer nested case-control study in the Multiethnic Cohort

Characteristic	n	Above LLOQa (%)	Unadjusted geometric mean (ng AMPA/mg CREAb)	Adjusted geometric meanc (ng AMPA/mg CREAb)	95% confidence interval (ng AMPA/mg CREAb)	p-valued
All study participants	250	87	0.074
Breast cancer cases	124	90	0.089	0.087	0.055 – 0.119	0.21
Controls	126	84	0.065	0.063	0.032 – 0.095	reference
Age group:
64 y or younger	113	85	0.068	0.068	0.039 – 0.098	reference
65–74 y	100	87	0.087	0.086	0.054 – 0.120	0.38
75 y or older	37	95	0.074	0.070	0.019 – 0.123	0.95
Race/ethnicity:
Japanese American	133	84	0.075	0.072	0.041 – 0.104	reference
Native Hawaiian	38	92	0.079	0.076	0.027 – 0.127	0.91
White	79	90	0.078	0.076	0.041 – 0.111	0.88
BMIe at urine collection
Under 25	142	87	0.075	0.075	0.045 – 0.105	reference
25 – 29.9	67	93	0.094	0.092	0.054 – 0.131	0.45
30 or higher	41	78	0.057	0.057	0.010 – 0.107	0.53

^alower limit of quantitation (0.001 ng/mL).

^bcreatinine.

^ccovariate-adjusted means, adjusted for all other factors in the table.

^dbased on pairwise t-test comparing adjusted means for each category to that for the reference category.

^ebody mass index (kg/m²).

Non-creatinine normalized urinary AMPA concentrations of our samples (<LLOQ to 3,698 ng/L) were within range found by others (Table 3).

Table 3.

Aminomethylphosphonic acid found in human urine

n, Population	Country, sampling years	LLODa, LLOQb (ng/L)	Urine Concentration (ng/L)	% samples >LLOD, LLOQ	Ref
399 healthy adults	Germany, 2001–2005	100b	<LLOQ-1,880	15–60%d,e	[21]
182, unknown population	18 European countries, 2013	150b	<LLOQ-2,630	36%	[19]
40 lactating women	USA, 2014–2015	30a, 100b	<LLOD-1,330	73%	[22]
100 elderly healthy adults	USA, 1993–1996, 2014–2016	40a	cyears 1993–1996 114–222 years 1999–2000 205–384 years 2001–2002 185–339 years 2004–2005 164–290 years 2014–2016 319–482	5–71 %e	[20]
250 mostly postmenopausal women	Hawaii, USA, 2001–2006	15a, 40b	<LLOQ-3,698	87%	current study

^alower limit of detection

^blower limit of quantitation

^ctaking into account only the amount of participants with levels above lower limit of detection

^dat or above LLOQ

^evaries by year

Compared to the lowest quintile of AMPA excretion, women in the second highest and the highest quintile had a statistically significantly higher risk of breast cancer (OR: 3.03; 95% CI: 1.02–9.03 and OR: 4.49; 95% CI: 1.46–13.77, respectively; p for trend=0.029; Table 4). There was no evidence of heterogeneity in association by race/ethnicity (p_{heterogeneity}=0.55; data not shown).

Table 4.

Association between urinary aminomethylphosphonic acid (AMPA) excretion and breast cancer risk among Hawaii breast cancer cases and their controls nested in the Multiethnic Cohort^a

AMPA quintileb	Cases	Controls	OR	95% CI		P-value
AMPA (continuous)c	123	123	3.49	0.35	34.51	0.286
Quintile 1 [0.0005–0.0045]	16	25	1 (ref)
Quintile 2 (0.0045–0.0276]	27	24	2.59	0.93	7.17	0.068
Quintile 3 (0.0276–0.0427]	18	26	1.69	0.55	5.14	0.358
Quintile 4 (0.0427–0.0758]	25	23	3.03	1.02	9.03	0.047
Quintile 5 (0.0758–3.698]	37	25	4.49	1.46	13.77	0.009
P for trendd						0.029

^aModeled using conditional logistic regression with adjustment for body mass index at urine collection and for the following baseline factors: family history of breast cancer, age at menarche, age at first live birth, menopausal status, hormone replacement therapy use, smoking status, alcohol use, education, moderate to vigorous physical activity, ever having had a mammogram.

^bquintiles based on the controls.

^cLog-transformed AMPA levels. The number of cases and controls correspond to the complete matched sets (1 case, 1 control).

^dP-value for trend is based on the quintile medians.

Note: OR, odds ratio; CI, confidence interval

Discussion

To our knowledge, this is the first study to prospectively examine urinary excretion of AMPA, (a metabolite of the herbicide GLYP as well as of phosphonate-based detergents [9, 10]), and breast cancer risk in a population of predominantly postmenopausal women. While previous reports investigated breast cancer risk and pesticides exposure based entirely on self-reported information of pesticide exposure [30, 16, 31], our study objectively measured urinary AMPA concentrations by a state-of-the-art liquid chromatography mass spectrometry-based assay [26]. The preliminary results presented here suggest that increased AMPA exposure, as reflected by urinary AMPA excretion in overnight urine, may be associated with increased breast cancer risk.

In humans, GLYP is not known to be metabolized to AMPA or other products [15]. Thus, AMPA measured in urine samples of our population is likely due to direct AMPA exposure. We speculate exposure to have occurred through the diet and/or by drinking water contaminated with AMPA. This is a plausible theory owing to several studies that have detected GLYP and AMPA residues in numerous food and water sources (Figure 1) [32–36, 10, 37, 13].

Figure 1.

Sources leading to human exposure of aminomethylphosphonic acid

Following application, GLYP is absorbed by foliage then metabolized to AMPA within the plant or degraded mainly (70%) to AMPA by soil microbes [6, 7]. Our study participants were not explicitly asked about their current or previous residential GLYP application. Thus, we can only surmise that general GLYP use may have been a source for AMPA exposure in our population and that runoff from GLYP-treated areas reached surface or groundwater.

AMPA is also a key metabolite of phosphonate-based household laundry and cleaning detergents [9, 10]. Thus, contamination of the environment by these detergents could also be a source of AMPA exposure (Figure 1) if domestic wastewater entering wastewater treatment plants is re-used/recycled into drinking water. However, domestic wastewater in the State of Hawaii was not reused or recycled for drinking purposes between December 2001and May 2006, the period our samples were collected (personal communication with Sina Pruder, Engineering Program Manager, Hawaii Department of Health, Environmental Management Division Wastewater Branch). Thus, AMPA exposure from this source for our population is most likely negligible.

AMPA is a small compound (111 Daltons) [18], does not contain a phenolic structure, and is highly water soluble (5.8 g/L near 25°C) [18], which would suggest their non-persistence in humans. Thus, the fact that AMPA excretion was measurable (above the LLOQ), albeit at low levels, suggests that other compounds in the body may have additively or synergistically enhanced the bioaccumulation or persistence of AMPA, a common occurrence with EDCs and other environmental pollutants [5]. To date, studies investigating the endocrine-disrupting properties of GLYP are diverging and ongoing between independent investigators [38–40] and the European Food Safety Authority [41], while the endocrine-disrupting abilities of AMPA are unknown.

Unfortunately, the cross-sectional nature of and limited information available for this pilot study makes it difficult to ascertain the exact source of AMPA exposure in our participants. Moreover, the lag time between exposure and disease occurrence (in our breast cancer cases) creates further challenges in establishing causal relationships. Nonetheless, regardless of the source of AMPA, which remains inconsequential for our study purposes, the role of AMPA as a possible EDC and/or human carcinogen merits further study.

The urinary AMPA concentrations in our study, when expressed as ng/L, are within range of those found in the literature among supposedly non-occupationally exposed adults (Table 3). However, it is important to note that urinary analyte concentrations can vary widely depending on urine volume when urine specimens other than timed samples are utilized (only Conrad et al. [21] collected 24-hour urines). To adjust for dilution, urine samples are typically normalized by a urinalysis parameter. Creatinine is a non-protein bound breakdown product of muscle and protein metabolism, and the most commonly used normalization parameter because it is produced in the body at a fairly constant rate [42]. Aside from the fact that three of the studies listed in Table 3 were spot urine samples, a considerable percentage of samples were reported to contain AMPA (Table 3). It remains to be determined whether the extremely high percentage of detectable AMPA in our samples (87% overall; Table 3) is a result of the higher analytical sensitivity of our analysis method, higher exposure to AMPA in our multiethnic study population, other unknown factors, or a combination of these factors.

The relatively small sample size of our current study prohibits us from further speculation. We intend to analyze approximately 1800 more MEC samples to increase study power. With these additional 900 breast cancer cases and 900 controls, we will have 80% power to detect as significant the OR=1.4 or higher in the entire sample and OR=1.8 or higher within races/ethnicities. Thus, we will be able to examine racial/ethnic differences, along with other environmental exposures including BPA, phthalates, TCS, and paraben, and other risk factors.

Another limitation is the imputation of a fixed value for AMPA excretion below LLOQ; however, this did not affect our quintile model association results. Nonetheless, these limitations should not detract from the main findings of this study, which need to be further explored in a larger cohort study in the future.

Conclusion

To our knowledge, this is the first study to prospectively examine urinary AMPA excretion and breast cancer risk in a population of predominantly postmenopausal women. Our preliminary results suggest a potential association between AMPA exposure, as reflected by urinary AMPA excretion in overnight urine, and breast cancer risk. In our study, women with higher urinary AMPA excretion had a significantly increased risk of breast cancer compared to women with lower AMPA excretion. Analysis in a larger cohort is warranted to confirm our preliminary findings.

Highlights.

• AMPA was detected in urine of 90% of cases, 84% of controls.

• Breast cancer risk was 4.5-fold higher in the highest vs. the lowest AMPA quintile.

• AMPA exposure may be associated with increased breast cancer risk.