Environmental Pollution 277 (2021) 116848

Contents lists available at ScienceDirect

Environmental Pollution
journal homepage: www.elsevier.com/locate/envpol

Pilot study on the urinary excretion of the glyphosate metabolite

aminomethylphosphonic acid and breast cancer risk: The Multiethnic
Cohort study*
Adrian A. Franke a, *, Xingnan Li a, Yurii B. Shvetsov b, Jennifer F. Lai a
a
University of Hawaii Cancer Center, Cancer Biology Program, USA
b
University of Hawaii Cancer Center, Epidemiology Program, USA

a r t i c l e i n f o a b s t r a c t

Article history: Breast cancer is the most commonly diagnosed female cancer and the second leading cause of death in
Received 23 November 2020 women in the US, including Hawaii. Accumulating evidence suggests that aminomethylphosphonic acid
Received in revised form (AMPA), the primary metabolite of the herbicide glyphosateda probable human carcinogen, may itself
24 February 2021
be carcinogenic. However, the relationship between urinary AMPA excretion and breast cancer risk in
Accepted 25 February 2021
Available online 1 March 2021
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
Keywords:
Urine
of urine collection, and fasting status) nested within the Hawaii biospecimen subcohort of the Multi-
Aminomethyphosphonic acid ethnic Cohort. AMPA was detected in 90% of cases and 84% of controls. The geometric mean of urinary
Glyphosate AMPA excretion was nearly 38% higher among cases vs. controls (0.087 vs 0.063 ng AMPA/mg creatinine)
Breast cancer after adjusting for race/ethnicity, age and BMI. A 4.5-fold higher risk of developing breast cancer in the
Multiethnic Cohort highest vs. lowest quintile of AMPA excretion was observed (ORQ5 vs. Q1: 4.49; 95% CI: 1.46e13.77;
ptrend ¼ 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.
© 2021 Elsevier Ltd. All rights reserved.

1. Introduction pesticides (methoxychlor, chlorpyrifos, dichlorodiphenyltrichloro-

ethane) and non-persistent environmental EDCs such as bisphenol
Breast cancer is the most commonly diagnosed female cancer A (BPA; used during the manufacturing of plastics/resins), phtha-
and the second leading cause of death in women in the US, lates (used as plasticizers), triclosan (TCS; antibacterial/antifungal
including Hawaii (Kohler et al., 2015; Hawaii Tumor Registry d U, agents), and parabens (used as preservatives) ubiquitously present
2016). Estrogen is a well-known contributor to the development in modern times (Diamanti-Kandarakis et al., 2009).
of hormone-dependent breast carcinomas (Vihko and Apter, 1989; Aminomethylphosphonic acid (AMPA) is the major metabolite
Thomas, 1984). As such, any factor that interferes with estrogen resulting from the soil microbial degradation of glyphosate (GLYP)
homeostasis, theoretically, has the ability to influence breast cancer (Duke, 2011; Bai and Ogbourne, 2016), currently the most widely
risk and, thus, merits further study. used herbicide worldwide and the subject of ongoing contentious
Endocrine disrupting chemicals (EDCs) are exogenous com- debate regarding its human carcinogenicity among international
pounds that interfere with normal hormonal actions in humans regulatory bodies (Benbrook, 2019) with AMPA being more often
(Diamanti-Kandarakis et al., 2009) and consist of a very heteroge- detected than GLYP (Battaglin et al., 2014; Kolpin et al., 2006;
nous group of compounds that include, but are not limited to, Medalie et al., 2020). AMPA is also the key metabolite of

*
This paper has been recommended for acceptance by Payam Dadvand.
* Corresponding author. University of Hawaii Cancer Center, 701 Ilalo Street, Honolulu, Hawaii, 96813, USA.
E-mail address: [email protected] (A.A. Franke).

https://doi.org/10.1016/j.envpol.2021.116848
0269-7491/© 2021 Elsevier Ltd. All rights reserved.
A.A. Franke, X. Li, Y.B. Shvetsov et al. Environmental Pollution 277 (2021) 116848

phosphonates found in several household laundry and cleaning the institutional review boards of the University of Hawaii and the
detergents (Jaworska et al., 2002; Battaglin et al., 2014), which may University of Southern California. Every participant gave written
explain its widespread occurrence in the environment (Battaglin consent.
et al., 2014; Kolpin et al., 2006; Medalie et al., 2020; Aparicio
et al., 2013).
2.2. Case ascertainment and control selection
In 2015, the International Agency for Research on Cancer (IARC)
cited evidence from in vitro and animal studies showing that AMPA
Incident invasive breast cancer cases were identified through
could induce oxidative stress (Benbrook, 2016)da characteristic for
linkage to the National Cancer Institute’s Surveillance, Epidemi-
human carcinogens (monographs on, 2017). In addition, a recent,
ology, and End Results Program registries covering Hawaii and
large study (n > 30,000) of farmers’ wives found significant asso-
California. Breast cancer diagnoses were classified using the Inter-
ciations between the use of several organophosphate insecticides
national Classification of Diseases for Oncology, Third Edition codes
and elevated breast cancer risk (Engel et al., 2017). However, the
C50.0-C50.9 and were restricted to invasive malignancies.
metabolites of these compounds were not analyzed.
For this analysis, we selected 124 Hawaii breast cancer cases and
Pesticide metabolites have often been found to be more mobile,
126 matched controls from an ongoing nested case-control study of
persistent, and detrimental than their respective parent compound
breast cancer within the MEC with available urine (Wu et al., 2021),
(Somasundaram and Coats, 1991) and, thus, may pose equal or
including 123 matched sets. Cases were selected among those
higher risks to humans upon exposure. While most pesticides are
diagnosed after urine collection through December 31, 2016. The
ultimately converted to components with less or no toxicity, certain
time between urine collection and breast cancer diagnosis aver-
chemical characteristics (i.e., water solubility, vapor pressure) can
aged (±standard deviation) 5.5 ± 3.3 years, with a minimum of 0.08
enable metabolites to become biologically active (Somasundaram
and a maximum of 12.2 years. One control was selected for each
and Coats, 1991). AMPA is very water soluble (5.8 g/L near 25  C
case and was randomly selected from the eligible pool of women
in water) (Grunewald et al., 2001), and has been detected in human
who were alive and free of a breast cancer diagnosis at the time of
urine of non-occupationally exposed humans (Hoppe, 2013; Mills
the case’s diagnosis. Controls were matched to cases on birth year
et al., 2017; Conrad et al., 2017; McGuire et al., 2016) often in a
(±1 year), race/ethnicity (Japanese American, Native Hawaiian,
higher percentage of samples than GLYP (Hoppe, 2013; Mills et al.,
White), date of urine collection (±1 year), time of blood draw
2017; Conrad et al., 2017).
(±2 h), and hours fasting prior to blood draw (8e10, >10 h).
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 2.3. Urinary AMPA and creatinine analysis
to expand our research in this pilot study to AMPA, using the same
urine samples from the same women within the Multiethnic Cohort All assays were performed at the Analytical Biochemistry Shared
(MEC). Resource of the University of Hawaii Cancer Center. The assays were
blinded to the analysts regarding case/control status, and samples
2. Methods were analyzed in randomized order, but samples of cases and their
respective controls were always kept in the same analysis batch.
2.1. Study population and data collection Urinary AMPA concentrations were measured by a novel ultra-
sensitive liquid chromatography high-resolution accurate-mass
This study utilized urine samples from a nested case-control mass spectrometry (LC/HRAM-MS) method that included deriva-
study of breast cancer among mostly postmenopausal women tization with 9-fluorenylmethoxycarbonyl chloride (Fmoc-Cl) and
participating in the Hawaii biospecimen subcohort of the MEC. The solid phase extraction (SPE) purification of the resulting adduct
MEC is a prospective study designed to investigate the associations (Franke et al., 2020). In brief, 0.5 mL of urine were mixed with the
of dietary, lifestyle, and genetic factors with the incidence of cancer. internal standard AMPA-13Ce15N-d2 (500 ng/mL in water) and
Study details have been described elsewhere (Kolonel et al., 2000). diluted with 1 mL of water followed by reversed phase SPE with
Briefly, between 1993 and 1996, over 215,000 men and women, polymeric sorbents (Strata-X; Phenomenex, Torrance, CA). The SPE
ages 45e75 years from five racial/ethnic groups (Japanese Amer- load and wash fractions were mixed with 100 mL of sodium borate
ican, Native Hawaiian, White, African American, Latino) from buffer (2.6% in water) and 150 mL of Fmoc-Cl (6 mg/mL in aceto-
Hawaii and Los Angeles, California, were enrolled in the cohort. nitrile) and incubated at 50  C for 20 min. After cooling to room
Potential participants were identified through drivers’ license files, temperature, 5 mL of formic acid was added to the reaction mixture
voter registration lists, and Medicare files. At cohort entry, partic- and centrifuged. The supernatant was then purified with a SepPak
ipants completed a self-administered, 26-page questionnaire that C18 reverse phase SPE cartridge (Waters, Milford, MA), separated
included queries on demographic characteristics, anthropometric by liquid chromatography using a Kinetex EVO C18 analytical col-
measures, medical history, family history of cancer, physical activ- umn (Phenomenex) and analyzed by LC/HRAM-MS (Q-Exactive,
ity, dietary habits, supplement use, and reproductive history for Thermo Scientific, Waltham, MA) with targeted selected ion
women. monitoring by detecting the monoisotopic masses of the proton-
The prospective MEC biospecimen subcohort was established ated analytes (±5 ppm to account for mass spectrometric inaccur-
from 2001 to 2006 among the surviving cohort members who acies). The between-day precision of this assay using pooled urine
agreed to provide blood and/or urine specimens and to answer a samples spiked with 0.4 ng/mL AMPA was excellent (CV ¼ 5%.)
short questionnaire (Park et al., 2009). Biospecimens were pro- Values below the lower limit of quantitation (LLOQ; 40 ng/L) were
spectively collected from 36,458 mostly postmenopausal women, assigned a value of 20 ng/L to allow statistical evaluations. Urinary
32,243 of whom provided a urine sample. Most participants in AMPA concentrations were adjusted for creatinine levels to
Hawaii provided an overnight specimen while those in Los Angeles normalize for urine volume (Franke et al., 2008). Urinary creatinine
donated mainly a first morning urine specimen; spot urine was was measured by a Cobas MiraPlus clinical chemistry auto analyzer
requested otherwise. At the time of urine collection, updated in- (Roche Diagnostics) using a commercially available creatinine kit
formation was collected on tobacco use, weight, medication, sup- (Randox Laboratories, Crumlin, UK) based on the Jaffe  reaction. The
plements, and hormone use. The study protocol was approved by between-day precision of this assay using pooled urine samples
2
A.A. Franke, X. Li, Y.B. Shvetsov et al. Environmental Pollution 277 (2021) 116848

was 8%. The calibration curve was highly linear (r2 > 0.995). Table 1
Characteristics of the Hawaii participants of the breast cancer nested case-control
study in the Multiethnic Cohort.a
2.4. Statistical analysis
Cases (n¼124) n Controls (n¼126) n P-
Breast cancer cases and controls were compared with respect to (%)c (%)c valueb

several demographic characteristics and potential risk factors of Age at urine collection, y
interest. Adjusted and unadjusted geometric means of AMPA 64 58 (47) 55 (44)
65e74 46 (37) 54 (43)
excretion values were compared between cases and controls, by
75 20 (16) 17 (14) 0.62
race/ethnicity, age groups (64 y, 65e74 y, 75 y and older), and Ethnicity
body mass index (BMI; normal: 18.5e24.9, overweight: 25e29.9, Japanese American 65 (52) 68 (54)
obese: 30) at urine collection, using general linear models and a White 40 (32) 39 (31)
Native Hawaiian 19 (15) 19 (15) 0.97
pairwise t-test.
Education
The association of urinary AMPA excretion with the risk of High school graduate 47 (38) 57 (45)
breast cancer was examined using conditional logistic regression Some college 30 (24) 24 (19)
with matched sets as strata. AMPA excretion was categorized into College graduate 21 (17) 23 (18)
quintiles based on the controls. Odds ratios (OR) and 95% confi- Graduate school 26 (21) 22 (18) 0.73
Family history of breast cancer
dence intervals (CI) were computed for each quintile, using the
No 107 (86) 105 (83)
lowest quintile as a reference. Models were adjusted for potential Yes 16 (13) 15 (12)
confounders of postmenopausal breast cancer risk, including BMI Unknown 1 (1) 6 (5) 0.16
at urine collection (continuous), family history of breast cancer BMI at urine collectiond
(yes, no), age at menarche (<13, 13e14, 15 or missing), age at first 18.5e24.9 74 (59) 68 (55)
25e29.9 31 (25) 36 (29)
live birth (20 or missing, 21e25, 26e30, 31 or nulliparous), 30 or higher 21 (17) 20 (16) 0.82
menopausal status (pre-, postmenopausal), hormone use (ever, Age at menarche, y
never), tobacco smoking (never, former, current smoker), daily <13 59 (48) 70 (56)
alcohol intake (0, <12 g/d, 12 g/d, missing), education (less than 13e14 49 (40) 43 (34)
>14 16 (13) 12 (10)
college degree, college degree or higher), mammography screening
Unknown 0 (0) 1 (1) 0.41
(ever, never), moderate to vigorous physical activity (hours/day, Menopausal status
continuous). Dietary factors such as total daily energy intake, Pre-menopausal 24 (19) 24 (19)
healthy eating index-2010 diet score (Guenther et al., 2013), and Post-menopausal 100 (81) 102 (81) 0.95
daily caffeine intake were also considered, but not included in the Age at first live birth, y
missing 0 (0) 3 (2)
final model as their inclusion did not change the risk estimates by Nulliparous 20 (16) 13 (10)
more than 10% (Mickey and Greenland, 1989). Linear trend was 20 21 (17) 25 (20)
evaluated by fitting a model with the nominal trend variable. 21e25 44 (36) 45 (36)
Heterogeneity in associations by race/ethnicity was tested using 26e30 29 (24) 35 (28)
31 10 (8) 5 (4) 0.22
the Wald test of an AMPA-race/ethnicity cross-product term in the
Number of children
logistic model. 0 20 (16) 13 (10)
Analyses were conducted using SAS version 9.4 (SAS Institute, 1 16 (13) 15 (12)
Inc., Cary, North Carolina). All tests were two sided, and P < 0.05 2e3 70 (57) 65 (52)
was considered statistically significant. 4 17 (14) 32 (25)
missing 1 (1) 1 (1) 0.18
Hormone replacement therapy use
3. Results Never used 45 (36) 53 (42)
Ever used 79 (64) 73 (58) 0.35
Study characteristics of breast cancer cases and controls are Smoking status
Never smoker 73 (59) 78 (62)
shown in Table 1. The largest racial/ethnic group among both cases
Former smoker 35 (28) 38 (30)
and controls was Japanese Americans (n ¼ 65 cases, n ¼ 68 con- Current smoker 16 (13) 10 (8) 0.44
trols), followed by Whites (n ¼ 40 cases, n ¼ 39 controls), and Alcohol intake, g/day
Native Hawaiians (n ¼ 19 cases, n ¼ 19 controls). Cases were less None 67 (54) 76 (60)
likely than controls to have started menarche before 13 years old or <12 37 (30) 35 (28)
12 19 (15) 15 (12)
to be parous; however, these differences were not statistically missing 1 (1) 0 (0) 0.56
significant. The majority of participants were postmenopausal at Mammography screening
baseline (81%) and reported no history of breast cancer among first Never had a 4 (3) 5 (4)
degree female family members (86% cases, 83% controls). Almost all mammogram
Ever had a mammogram 120 (97) 121 (96) 0.75
participants (97% cases, 96% controls) reported ever having a
a
mammogram. Characteristics at baseline questionnaire, unless specified.
b
The percentage of values above LLOQ was 90% and 84% among P-values are based on the chi-square test for categorical variables and on the t-
test for continuous variables.
breast cancer cases and controls, respectively (Table 2). Geometric c
Percentages may exceed 100 due to rounding.
mean urinary AMPA excretion after creatinine-adjustment was d
Body Mass Index, kg/m2.
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 65e74 y age 0.057, respectively). There were no differences in AMPA excretion
group (0.086) compared to 64 y and 75 y age groups, (0.068 and by race/ethnicity after adjusting for age, BMI and case/control
0.070, respectively, but these differences were not statistically status.
significant. Overweight women (BMI 25.0e29.9) had non- Non-creatinine normalized urinary AMPA concentrations of our
significantly higher AMPA excretion (0.092), followed by normal- samples (<LLOQ to 3698 ng/L) were within ranges found by others
weight (BMI 18.5e24.9) and obese women (BMI30), (0.075 and (Table 3).

3
A.A. Franke, X. Li, Y.B. Shvetsov et al. Environmental Pollution 277 (2021) 116848

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 Adjusted geometric meanc 95% confidence interval p-valued
(%) (ng AMPA/mg CREAb) (ng AMPA/mg CREAb) (ng AMPA/mg CREAb)

All study participants 250 87 0.074

Breast cancer cases 124 90 0.089 0.087 0.055e0.119 0.21

Controls 126 84 0.065 0.063 0.032e0.095 reference

Age group:
64 y or younger 113 85 0.068 0.068 0.039e0.098 reference
65-74 y 100 87 0.087 0.086 0.054e0.120 0.38
75 y or older 37 95 0.074 0.070 0.019e0.123 0.95

Race/ethnicity:
Japanese American 133 84 0.075 0.072 0.041e0.104 reference
Native Hawaiian 38 92 0.079 0.076 0.027e0.127 0.91
White 79 90 0.078 0.076 0.041e0.111 0.88

BMIe at urine collection

Under 25 142 87 0.075 0.075 0.045e0.105 reference
25e29.9 67 93 0.094 0.092 0.054e0.131 0.45
30 or higher 41 78 0.057 0.057 0.010e0.107 0.53
a
Lower limit of quantitation (40 ng/L).
b
Creatinine.
c
Covariate-adjusted means, adjusted for all other factors in the table.
d
Based on pairwise t-test comparing adjusted means for each category to that for the reference category.
e
Body mass index (kg/m2).

Table 3
Aminomethylphosphonic acid found in human urine.

n, Population Country, sampling years LLODa, Urine Concentration (ng/L) % samples > LLOD, Ref
LLOQb LLOQ

399 healthy adults Germany, 2001e2005 100b <LLOQ-1880 15e60%d,e Conrad et al. (2017)
182, unknown population 18 European countries, 150b <LLOQ-2630 36% Hoppe (2013)
2013
40 lactating women USA, 2014e2015 30a, 100b <LLOD-1330 73% McGuire et al.
(2016)
a c e
100 elderly healthy adults USA, 1993e1996, 2014 40 years 1993-1996 114e222 years 1999-2000 205 5e71% Mills et al. (2017)
e2016 e384
years 2001-2002 185e339 years 2004-2005 164
e290
years 2014-2016 319-482
250 mostly postmenopausal Hawaii, USA, 2001e2006 15a, 40b <LLOQ-3698 87% current study
women
a
Lower limit of detection (ng/L).
b
Lower limit of quantitation (ng/L).
c
Taking into account only the amount of participants with levels above lower limit of detection.
d
At or above LLOQ.
e
Varies by year.

Compared to the lowest quintile of AMPA excretion, women in the-art LCMS-assay (Franke et al., 2020). The preliminary results
the second highest and the highest quintile had a statistically presented here suggest that increased AMPA exposure, as reflected
significantly higher risk of breast cancer (OR: 3.03; 95% CI: by urinary AMPA excretion in overnight urine, may be associated
1.02e9.03 and OR: 4.49; 95% CI: 1.46e13.77, respectively; p for with increased breast cancer risk.
trend ¼ 0.029; Table 4). There was no evidence of heterogeneity in In humans, GLYP is not known to be metabolized to AMPA or
association by race/ethnicity (pheterogeneity ¼ 0.55; data not shown). other products (monographs on, 2017). Thus, AMPA measured in
urine samples of our population is likely due to direct AMPA
4. Discussion exposure. We speculate exposure to have occurred through the diet
and/or by drinking water contaminated with AMPA. This is a
To our knowledge, this is the first study to prospectively plausible theory owing to several studies that have detected GLYP
examine urinary excretion of AMPA, (a metabolite of the herbicide and AMPA residues in numerous food and water sources (Duke
GLYP as well as of phosphonate-based detergents (Jaworska et al., et al., 2003; Savini et al., 2019; Cessna et al., 1994; Cessna et al.,
2002; Battaglin et al., 2014)), and breast cancer risk in a popula- 2000; Chen et al., 2013; Battaglin et al., 2014; Cessna et al., 2002;
tion of predominantly postmenopausal women. While previous Aparicio et al., 2013).
reports investigated breast cancer risk and pesticides exposure Following application, GLYP is absorbed by foliage then
based entirely on self-reported information of pesticide exposure metabolized to AMPA within the plant or degraded mainly (70%) to
(Duell et al., 2000; Engel et al., 2017; Engel et al., 2005), our study AMPA by soil microbes (Duke, 2011; Bai and Ogbourne, 2016). Our
objectively measured urinary AMPA concentrations by a state-of- study participants were not explicitly asked about their current or

4
A.A. Franke, X. Li, Y.B. Shvetsov et al. Environmental Pollution 277 (2021) 116848

Table 4 Environmental Management Division Wastewater Branch). Thus,

Association between urinary aminomethylphosphonic acid (AMPA) excretion and AMPA exposure from this source for our population is most likely
breast cancer risk among Hawaii breast cancer cases and their controls nested in the
Multiethnic Cohort.a
negligible.
AMPA is a small compound (111 Da) (Grunewald et al., 2001),
AMPA quintileb Cases Controls OR 95% CI P-value does not contain a phenolic structure, and is highly water soluble
AMPA (continuous) c
123 123 3.49 0.35 34.51 0.286 (5.8 g/L near 25  C) (Grunewald et al., 2001), which would suggest
their non-persistence in humans. Thus, the fact that AMPA excre-
Quintile 1 [0.0005e0.0045] 16 25 1 (ref) tion was measurable (above the LLOQ), albeit at low levels, suggests
Quintile 2 (0.0045e0.0276] 27 24 2.59 0.93 7.17 0.068
Quintile 3 (0.0276e0.0427] 18 26 1.69 0.55 5.14 0.358
that other compounds in the body may have additively or syner-
Quintile 4 (0.0427e0.0758] 25 23 3.03 1.02 9.03 0.047 gistically enhanced the bioaccumulation or persistence of AMPA, a
Quintile 5 (0.0758e3.698] 37 25 4.49 1.46 13.77 0.009 common occurrence with EDCs and other environmental pollutants
P for trendd 0.029d (Diamanti-Kandarakis et al., 2009). To date, studies investigating
Note: OR, odds ratio; CI, confidence interval. the endocrine-disrupting properties of GLYP are diverging and
a
Modeled using conditional logistic regression with adjustment for body mass ongoing between independent investigators (Gasnier et al., 2009;
index at urine collection and for the following baseline factors: family history of Thongprakaisang et al., 2013; Manservisi et al., 2019) and the Eu-
breast cancer, age at menarche, age at first live birth, menopausal status, hormone
replacement therapy use, smoking status, alcohol use, education, moderate to
ropean Food Safety Authority (2017), while the endocrine-
vigorous physical activity, ever having had a mammogram. disrupting abilities of AMPA are unknown.
b
Quintiles based on the controls. Unfortunately, the cross-sectional nature of and limited infor-
c
Log-transformed AMPA levels. The number of cases and controls correspond to mation available for this pilot study makes it difficult to ascertain
the complete matched sets (1 case, 1 control).
d the exact source of AMPA exposure in our participants. Moreover,
P-value for trend is based on the quintile medians.
the lag time between exposure and disease occurrence (in our
breast cancer cases) creates further challenges in establishing
previous residential GLYP application. Thus, we can only surmise causal relationships. Nonetheless, regardless of the source of AMPA,
that general GLYP use may have been a source for AMPA exposure which remains inconsequential for our study purposes, the role of
in our population and that runoff from GLYP-treated areas reached AMPA as a possible EDC and/or human carcinogen merits further
surface or groundwater. study.
AMPA is also a key metabolite of phosphonate-based household The urinary AMPA concentrations in our study, when expressed
laundry and cleaning detergents (Jaworska et al., 2002; Battaglin as ng/L, are within ranges of those found in the literature among
et al., 2014). Thus, contamination of the environment by these supposedly non-occupationally exposed adults (Table 3). However,
detergents could also be a source of AMPA exposure (Fig. 1) if do- it is important to note that urinary analyte concentrations can vary
mestic wastewater entering wastewater treatment plants is re- widely depending on urine volume when urine specimens other
used/recycled into drinking water. However, domestic wastewater than timed samples are utilized (only Conrad et al. (2017) collected
in the State of Hawaii was not reused or recycled for drinking 24-h urines). To adjust for dilution, urine samples are typically
purposes between December 2001and May 2006, the period our normalized by a urinalysis parameter. Creatinine is a non-protein
samples were collected (personal communication with Sina Pruder, bound breakdown product of muscle and protein metabolism,
Engineering Program Manager, Hawaii Department of Health, and the most commonly used normalization parameter because it

Fig. 1. Sources leading to human exposure of aminomethylphosphonic acid.

5
A.A. Franke, X. Li, Y.B. Shvetsov et al. Environmental Pollution 277 (2021) 116848

is produced in the body at a fairly constant rate (Alessio et al., 1985). Availability of data and material
Aside from the fact that three of the studies listed in Table 3 were
spot urine samples, a considerable percentage of samples were The data that support the reported findings of this study are
reported to contain AMPA (Table 3). It remains to be determined available from AAF and XL.
whether the extremely high percentage of detectable AMPA in our
samples (87% overall; Table 3) is a result of the higher analytical Declaration of competing interest
sensitivity of our analysis method, higher exposure to AMPA in our
multiethnic study population, other unknown factors, or a combi- The authors declare that they have no known competing
nation of these factors. financial interests or personal relationships that could have
The relatively small sample size of our current study prohibits us appeared to influence the work reported in this paper.
from further speculation. We intend to analyze approximately 1800
more MEC samples to increase study power. With these additional Acknowledgement
900 breast cancer cases and 900 controls, we will have 80% power
to detect as significant the OR ¼ 1.4 or higher in the entire sample We thank Drs. Anna Wu from the University of Southern Cali-
and OR ¼ 1.8 or higher within races/ethnicities. Thus, we will be fornia, and Iona Cheng from the University California, San Francisco
able to examine racial/ethnic differences, along with other envi- for their helpful review of this manuscript. Dr. Wu is also
ronmental exposures including BPA, phthalates, TCS, and paraben, acknowledged for initial data analysis that yielded similar results.
and other risk factors.
Another limitation is the imputation of a fixed value for AMPA
References
excretion below LLOQ; however, this did not affect our quintile
model association results. Nonetheless, these limitations should Alessio, L., Berlin, A., Dell’Orto, A., Toffoletto, F., Ghezzi, I., 1985. Reliability of urinary
not detract from the main findings of this study, which need to be creatinine as a parameter used to adjust values of urinary biological indicators.
further explored in a larger cohort study in the future. Int. Arch. Occup. Environ. Health 55 (2), 99e106. https://doi.org/10.1007/
bf00378371.
Aparicio, V.C., De Geronimo, E., Marino, D., Primost, J., Carriquiriborde, P., Costa, J.L.,
5. Conclusion 2013. Environmental fate of glyphosate and aminomethylphosphonic acid in
surface waters and soil of agricultural basins. Chemosphere 93 (9), 1866e1873.
https://doi.org/10.1016/j.chemosphere.2013.06.041.
To our knowledge, this is the first study to prospectively Authority, E.F.S., 2017. Peer review of the pesticide risk assessment of the potential
examine urinary AMPA excretion and breast cancer risk in a pop- endocrine disrupting properties of glyphosate. EFSA J. 15 (9), e04979.
Bai, S.H., Ogbourne, S.M., 2016. Glyphosate: environmental contamination, toxicity
ulation of predominantly postmenopausal women. Our pre- and potential risks to human health via food contamination. Environ. Sci. Pollut.
liminary results suggest a potential association between AMPA Res. Int. 23 (19), 18988e19001. https://doi.org/10.1007/s11356-016-7425-3.
exposure, as reflected by urinary AMPA excretion in overnight Battaglin, W.A., Meyer, M., Kuivila, K., Dietze, J., 2014. Glyphosate and its degrada-
tion product AMPA occur frequently and widely in US soils, surface water,
urine, and breast cancer risk. In our study, women with higher groundwater, and precipitation. JAWRA J. Am. Water Resour. Assoc. 50 (2),
urinary AMPA excretion had a significantly increased risk of breast 275e290.
cancer compared to women with lower AMPA excretion. Analysis in Benbrook, C.M., 2016. Trends in glyphosate herbicide use in the United States and
globally. Environ. Sci. Eur. 28 (1), 3. https://doi.org/10.1186/s12302-016-0070-0.
a larger cohort is warranted to confirm our preliminary findings.
Benbrook, C.M., 2019. How did the US EPA and IARC reach diametrically opposed
conclusions on the genotoxicity of glyphosate-based herbicides? Environ. Sci.
Eur. 31 (1), 2.
Ethics approval
Cessna, A., Darwent, A., Kirkland, K., Townley-Smith, L., Harker, K., Lefkovitch, L.,
1994. Residues of glyphosate and its metabolite AMPA in wheat seed and foliage
The study protocol was approved by the institutional review following preharvest applications. Can. J. Plant Sci. 74 (3), 653e661.
boards of the University of Hawaii and the University of Southern Cessna, A., Darwent, A., Townley-Smith, L., Harker, K., Kirkland, K., 2000. Residues of
glyphosate and its metabolite AMPA in canola seed following preharvest ap-
California. plications. Can. J. Plant Sci. 80 (2), 425e431.
Cessna, A., Darwent, A., Townley-Smith, L., Harker, K., Kirkland, K., 2002. Residues of
glyphosate and its metabolite AMPA in field pea, barley and flax seed following
Funding preharvest applications. Can. J. Plant Sci. 82 (2), 485e489.
Chen, M.X., Cao, Z.Y., Jiang, Y., Zhu, Z.W., 2013. Direct determination of glyphosate
This work was supported by the Hawaii Community Foundation and its major metabolite, aminomethylphosphonic acid, in fruits and vegetables
by mixed-mode hydrophilic interaction/weak anion-exchange liquid chroma-
(grant# 20ADVC-102165) and the National Cancer Institute (P30 tography coupled with electrospray tandem mass spectrometry. J. Chromatogr.
CA71789 and U01 CA164973). A 1272, 90e99. https://doi.org/10.1016/j.chroma.2012.11.069.
Conrad, A., Schroter-Kermani, C., Hoppe, H.W., Ruther, M., Pieper, S., Kolossa-
Gehring, M., 2017. Glyphosate in German adults - time trend (2001 to 2015) of
Author Contribution human exposure to a widely used herbicide. Int. J. Hyg Environ. Health 220 (1),
8e16. https://doi.org/10.1016/j.ijheh.2016.09.016.
Diamanti-Kandarakis, E., Bourguignon, J.P., Giudice, L.C., Hauser, R., Prins, G.S.,
Adrian A. Franke: Conceptualization, Methodology, Validation, Soto, A.M., et al., 2009. Endocrine-disrupting chemicals: an Endocrine Society
Project administration, Writing e original draft, Reviewing & scientific statement. Endocr. Rev. 30 (4), 293e342. https://doi.org/10.1210/
Editing Xingnan Li: Investigation, Resources, Writing- Reviewing & er.2009-0002.
Duell, E.J., Millikan, R.C., Savitz, D.A., Newman, B., Smith, J.C., Schell, M.J., et al., 2000.
Editing Yurii B. Shvetsov: Formal analysis, Data curation, Writing-
A population-based case-control study of farming and breast cancer in North
Reviewing & Editing Jennifer F. Lai: Writing e original draft, Carolina. Epidemiology 11 (5), 523e531. https://doi.org/10.1097/00001648-
Reviewing & Editing. 200009000-00007.
Duke, S.O., 2011. Glyphosate degradation in glyphosate-resistant and -susceptible
crops and weeds. J. Agric. Food Chem. 59 (11), 5835e5841. https://doi.org/
Consent to participate 10.1021/jf102704x.
Duke, S.O., Rimando, A.M., Pace, P.F., Reddy, K.N., Smeda, R.J., 2003. Isoflavone,
glyphosate, and aminomethylphosphonic acid levels in seeds of glyphosate-
Informed consent was obtained from all individual participants treated, glyphosate-resistant soybean. J. Agric. Food Chem. 51 (1), 340e344.
included in the study. https://doi.org/10.1021/jf025908i.

6
A.A. Franke, X. Li, Y.B. Shvetsov et al. Environmental Pollution 277 (2021) 116848

Engel, L.S., Hill, D.A., Hoppin, J.A., Lubin, J.H., Lynch, C.F., Pierce, J., et al., 2005. development and endocrine system. Environ. Health 18 (1), 15. https://doi.org/
Pesticide use and breast cancer risk among farmers’ wives in the agricultural 10.1186/s12940-019-0453-y.
health study. Am. J. Epidemiol. 161 (2), 121e135. https://doi.org/10.1093/aje/ McGuire, M.K., McGuire, M.A., Price, W.J., Shafii, B., Carrothers, J.M., Lackey, K.A.,
kwi022. et al., 2016. Glyphosate and aminomethylphosphonic acid are not detectable in
Engel, L.S., Werder, E., Satagopan, J., Blair, A., Hoppin, J.A., Koutros, S., et al., 2017. human milk. Am. J. Clin. Nutr. 103 (5), 1285e1290. https://doi.org/10.3945/
Insecticide use and breast cancer risk among farmers’ wives in the agricultural ajcn.115.126854.
health study. Environ. Health Perspect. 125 (9), 097002 https://doi.org/10.1289/ Medalie, L., Baker, N.T., Shoda, M.E., Stone, W.W., Meyer, M.T., Stets, E.G., et al., 2020.
ehp1295. Influence of land use and region on glyphosate and aminomethylphosphonic
Franke, A., Halm, B., Ashburn, L., 2008. Isoflavones in children and adults consuming acid in streams in the USA. Sci. Total Environ. 707, 136008. https://doi.org/
soy. Arch. Biochem. Biophys. 476, 161e170. 10.1016/j.scitotenv.2019.136008.
Franke, A.A., Li, X., Lai, J.F., 2020. Analysis of glyphosate, aminomethylphosphonic Mickey, R.M., Greenland, S., 1989. The impact of confounder selection criteria on
acid, and glufosinate from human urine by HRAM LC-MS. Anal. Bioanal. Chem. effect estimation. Am. J. Epidemiol. 129 (1), 125e137. https://doi.org/10.1093/
412 (30), 8313e8324. https://doi.org/10.1007/s00216-020-02966-1. oxfordjournals.aje.a115101.
Gasnier, C., Dumont, C., Benachour, N., Clair, E., Chagnon, M.C., Seralini, G.E., 2009. Mills, P.J., Kania-Korwel, I., Fagan, J., McEvoy, L.K., Laughlin, G.A., Barrett-Connor, E.,
Glyphosate-based herbicides are toxic and endocrine disruptors in human cell 2017. Excretion of the herbicide glyphosate in older adults between 1993 and
lines. Toxicology 262 (3), 184e191. https://doi.org/10.1016/j.tox.2009.06.006. 2016. Jama 318 (16), 1610e1611. https://doi.org/10.1001/jama.2017.11726.
Grunewald, K., Schmidt, W., Unger, C., Hanschmann, G., 2001. Behavior of glypho- IARC, 2017. IARC Monographs on the Evaluation of the Carcinogenic Risks to
sate and aminomethylphosphonic acid (AMPA) in soils and water of reservoir Humans-Volume 112: Some Organophosphate Insecticides and Herbicides.
Radeburg II catchment (Saxony/Germany). J. Plant Nutr. Soil Sci. 164 (1), 65e70. IARC, World Health Organization, Lyon, France.
Guenther, P.M., Casavale, K.O., Reedy, J., Kirkpatrick, S.I., Hiza, H.A., Kuczynski, K.J., Park, S.Y., Wilkens, L.R., Henning, S.M., Le Marchand, L., Gao, K., Goodman, M.T.,
et al., 2013. Update of the healthy eating index: HEI-2010. J. Acad. Nutr. Diet. 113 et al., 2009. Circulating fatty acids and prostate cancer risk in a nested case-
(4), 569e580. https://doi.org/10.1016/j.jand.2012.12.016. control study: the Multiethnic Cohort. Cancer Causes Control 20 (2), 211e223.
Hawaii at a Glance 2009-2013 Hawaii Tumor Registry d U, 2016. Hawaii Tumor https://doi.org/10.1007/s10552-008-9236-4.
Registry. University of Hawaii Hawaii Cancer. Savini, S., Bandini, M., Sannino, A., 2019. An improved, rapid, and sensitive ultra-
Hoppe, H.W., 2013. Determination of Glyphosate Residues in Human Urine Samples high-performance liquid chromatography-high-resolution orbitrap mass spec-
from 18 European Countries. Medical Laboratory Bremen, D-28357 Bremen/ trometry analysis for the determination of highly polar pesticides and con-
Germany. taminants in processed fruits and vegetables. J. Agric. Food Chem. 67 (9),
Jaworska, J., Van Genderen-Takken, H., Hanstveit, A., van de Plassche, E., Feijtel, T., 2716e2722. https://doi.org/10.1021/acs.jafc.8b06483.
2002. Environmental risk assessment of phosphonates, used in domestic Somasundaram, L., Coats, J.R., 1991. Pesticide transformation products in the envi-
laundry and cleaning agents in The Netherlands. Chemosphere 47 (6), 655e665. ronment. Pestic. Transform. Prod. 459 (1), 2e9. https://doi.org/10.1021/bk-
https://doi.org/10.1016/s0045-6535(01)00328-9. 1991-0459.ch001.
Kohler, B.A., Sherman, R.L., Howlader, N., Jemal, A., Ryerson, A.B., Henry, K.A., et al., Thomas, D.B., 1984. Do hormones cause breast cancer? Cancer 53 (3 Suppl. l),
2015. Annual report to the nation on the status of cancer, 1975-2011, featuring 595e604. https://doi.org/10.1002/1097-0142(19840201)53:3þ<595::aid-
incidence of breast cancer subtypes by race/ethnicity, poverty, and state. J. Natl. cncr2820531304>3.0.co;2-y.
Cancer Inst. 107 (6), djv048. https://doi.org/10.1093/jnci/djv048. Thongprakaisang, S., Thiantanawat, A., Rangkadilok, N., Suriyo, T., Satayavivad, J.,
Kolonel, L.N., Henderson, B.E., Hankin, J.H., Nomura, A.M., Wilkens, L.R., Pike, M.C., 2013. Glyphosate induces human breast cancer cells growth via estrogen re-
et al., 2000 Feb 15. A multiethnic cohort in Hawaii and Los Angeles: baseline ceptors. Food Chem. Toxicol. 59, 129e136. https://doi.org/10.1016/
characteristics. Am. J. Epidemiol. 151 (4), 346e357. j.fct.2013.05.057.
Kolpin, D.W., Thurman, E.M., Lee, E.A., Meyer, M.T., Furlong, E.T., Glassmeyer, S.T., Vihko, R., Apter, D., 1989. Endogenous steroids in the pathophysiology of breast
2006. Urban contributions of glyphosate and its degradate AMPA to streams in cancer. Crit. Rev. Oncol. Hematol. 9 (1), 1e16. https://doi.org/10.1016/s1040-
the United States. Sci. Total Environ. 354 (2e3), 191e197. https://doi.org/ 8428(89)80012-5.
10.1016/j.scitotenv.2005.01.028. Wu, A.H., Franke, A.A., Tseng, C., Wilkens, L.R., Conroy, S.M., Li, Y., et al., 2021.
Manservisi, F., Lesseur, C., Panzacchi, S., Mandrioli, D., Falcioni, L., Bua, L., et al., 2019. Exposure to Phthalates and Risk of Breast Cancer: the Multiethnic Cohort study.
The Ramazzini Institute 13-week pilot study glyphosate-based herbicides Breast Cancer Res. In Press.
administered at human-equivalent dose to Sprague Dawley rats: effects on