Plasma carotenoids and risk of breast cancer over 20 y of follow-up
Plasma carotenoids and risk of breast cancer over 20 y of follow-up1-3
A Heather Eliassen
Rulla M Tamimi
Shelley S Tworoger
Susan E Hankinson
0 Supported by the National Cancer Institute at the NIH (R01 CA131218 , R01 CA49449, and UM1 CA186107). Medicine, 181 Longwood Avenue, Boston, MA 02115 , USA
1 From the Channing Division of Network Medicine, Department of Medicine, Brigham and Women's Hospital and Harvard Medical School , Boston, MA (AHE, XL, BR, RMT, SST , and SEH); the Departments of Epidemiology (AHE , XL, BR, RMT, SST, and SEH) and Biostatistics (XL and BR) , Harvard T.H. Chan School of Public Health, Boston, MA; and the Division of Biostatistics and Epidemiology, School of Public Health and Health Sciences, University of Massachusetts , Amherst, MA, SEH , USA
2 Abbreviations used: ER , estrogen receptor; HER2, human epidermal growth factor receptor 2; ICC, intraclass correlation coefficient; NHS, Nurses' Health Study; PMH, postmenopausal hormone; PR, progesterone receptor; WHEL , Women's Health Eating and Living. First published online
Background: Increasing evidence suggests that carotenoids, which are micronutrients in fruit and vegetables, reduce breast cancer risk. Whether carotenoids are important early or late in carcinogenesis is unclear, and limited analyses have been conducted by breast tumor subtypes. Objectives: We sought to examine issues of the timing of carotenoid exposure as well as associations by breast tumor subtypes. Design: We conducted a nested case-control study of plasma carotenoids measured by using reverse-phase high-performance liquid chromatography and breast cancer risk in the Nurses' Health Study. In 1989-1990, 32,826 women donated blood samples; in 2000-2002, 18,743 of these women contributed a second blood sample. Between the first blood collection and June 2010, 2188 breast cancer cases were diagnosed (579 cases were diagnosed after the second collection) and matched with control subjects. RRs and 95% CIs were calculated by using conditional logistic regression adjusted for several breast cancer risk factors. Results: Higher concentrations of a-carotene, b-carotene, lycopene, and total carotenoids were associated with 18-28% statistically significantly lower risks of breast cancer (e.g., b-carotene top compared with bottom quintile RR: 0.72; 95% CI: 0.59, 0.88; P-trend , 0.001). Associations were apparent for total carotenoids measured $10 y before diagnosis (top compared with bottom quintile RR: 0.69; 95% CI: 0.50, 0.95; P-trend = 0.01) as well as those ,10 y before diagnosis (RR: 0.79; 95% CI: 0.64, 0.98; P-trend = 0.04, P-interaction = 0.11). Carotenoid concentrations were strongly inversely associated with breast cancer recurrence and death (e.g., b-carotene top compared with bottom quintile RR: 0.32; 95% CI: 0.21, 0.51; P-trend , 0.001) compared with not recurrent and not lethal disease (P-heterogeneity , 0.001). Conclusion: In this large prospective analysis with 20 y of followup, women with high plasma carotenoids were at reduced breast cancer risk particularly for more aggressive and ultimately fatal disease. Am J Clin Nutr 2015;101:1197-205.
biomarkers; breast cancer; carotenoids; plasma; nested case-control study
Carotenoids, which are essential for plant photosynthesis,
provide yellow-red pigments in fruit and vegetables. a-Carotene,
b-carotene, b-cryptoxanthin, lutein, zeaxanthin, and lycopene are
the most prevalent in the US diet, comprising 90% of circulating
), and hypothesized to be anticarcinogenic through
metabolism to retinoids that contribute to cellular
differentiation, antioxidation, immuno-enhancement, or the inhibition of
tumorigenesis and malignant transformation (
). In experimental
studies, carotenoids reduce the proliferation in breast cancer cell
lines and inhibit tumor progression (
Studies of fruit and vegetable intake and, more specifically,
carotenoids have been mixed. Most recently, in a pooled analyses of
18 cohort studies, inverse associations were observed between
a-carotene, b-carotene, and lutein and zeaxanthin intakes and
estrogen receptor (ER)4–negative, but not ER-positive, tumors (
The measurement of circulating carotenoids avoids recalled diet (
and inaccuracies of nutrient databases (
) and integrates cooking
), geographic and seasonal variation of foods (11),
and individual variation in absorption. In our recent pooled analysis
) of circulating carotenoids and subsequent breast cancer risk
(n = 3055 cases), we observed significant 13–22% reduced risks of
total breast cancer for the top (compared with bottom) quintiles of
a-carotene, b-carotene, lutein and zeaxanthin, lycopene, and total
carotenoids and 48% reduced risk of ER-negative tumors for
Although the pooled analysis allowed a comprehensive
examination of carotenoids, our ability to examine the importance of
exposure timing as well as tumor subtypes and outcomes was limited.
Thus, we examined these issues in a nested case-control study within
the Nurses’ Health Study (NHS) by using blood samples collected
10 y apart with 20 y of follow-up. The pooled analysis included 962
NHS cases; the current analysis was expanded to include 2188 cases.
In 1976, 121,701 female registered nurses aged 30–55 y were
enrolled in the NHS. Biennially, participants completed mailed
questionnaires on lifestyle, diet, reproductive history, and
disease diagnoses. In 1989–1990, 32,826 women aged 43–69 y
donated blood samples (
). Briefly, each woman arranged to
have her blood drawn and shipped overnight with an ice pack to
our laboratory where it was processed and archived in
liquidnitrogen freezers; 97% of samples arrived #26 h of collection.
In 2000–2002, a second sample was collected by using a similar
protocol from 18,743 of these women aged 53–80 y (
follow-up rate in the 32,826 women was 97% in 2010. The study
was approved by the Committee on the Use of Human Subjects
in Research at the Brigham and Women’s Hospital; the
completion of the self-administered questionnaire and blood
collection was considered to imply informed consent.
Case and control selection
Cases had no reported cancer (other than nonmelanoma skin)
before blood collection and were diagnosed with breast cancer
between the first collection and June 2010. Overall, 2188 breast
cancer cases were reported (n = 1750 invasive), which were
confirmed by medical record reviews (n = 2,152) or verbally by
the nurse (n = 36). The time from blood collection to diagnosis
ranged from ,1 mo to 20 y (median: 9.3 y) after blood
collection. One control was matched per case by using the
following factors (at both collections for subjects with 2 samples):
age (62 y), menopausal status and postmenopausal hormone
(PMH) use at blood collection and diagnosis (premenopausal,
postmenopausal and not taking PMHs, postmenopausal and
taking PMHs, and unknown), and month (61 mo), time of day
(62 h), and fasting status at blood collection (,10 h after a meal
or unknown; $10h).
Plasma carotenoids were assayed by reverse-phase HPLC (
at the Harvard T.H. Chan School of Public Health. Assays were
conducted in 8 batches; CVs from blinded quality-control
replicates (10% of samples) were generally #15% except for the
following batches for which CVs were #20%: a-carotene (n = 2),
b-carotene (n = 2), b-cryptoxanthin (n = 1), and lycopene (n = 1).
Variation in blinded quality controls across batches was noted,
which suggested laboratory variation; all batches were
recalibrated to an average standard batch (
Questionnaire, tumor, and outcome data
Information on breast cancer risk factors, including
anthropometric measures, reproductive history, and diet, was collected from
biennial and blood collection questionnaires. Detailed information
on case characteristics, including invasiveness, histologic grade, ER,
progesterone receptor (PR), and human epidermal growth factor
receptor 2 (HER2) status, was extracted from pathology reports. In
addition, cases with available tumor tissue included in tumor
microarrays were immunostained for ER, PR, and HER2 and read
manually by a study pathologist (
Breast cancer recurrences were documented from reported
second cancer diagnoses. If the second cancer was reported in the
lung, liver, bone, or brain, it was assumed that breast cancer
recurred because these are the most-common sites of breast cancer
recurrence; medical records were reviewed to exclude primary
lung cancer (
). Deaths were reported by the family or post office.
Nonresponders were searched in the National Death Index. More
than 98% of deaths in the NHS have been identified by these
methods. Physician reviewers blinded to exposure information
ascertained the cause of death from death certificates, which were
supplemented with medical records if necessary.
Included in the analysis were 2188 distinct cases and 2188
controls. Of these, 2147 cases and 2150 controls had at least one
carotenoid measure by using the first collection; 579 cases and
580 controls had 2 carotenoids measures from the first and second
collections. With the use of all available data, 1828 cases had
carotenoids measured ,10 y before diagnosis, and 895 cases
had carotenoids measured $10 y before diagnosis.
We calculated intraclass correlation coefficients (ICCs) over
10 y in women with 2 blood collections. Quintile cutoffs for
carotenoids were established in controls. RRs and 95% CIs were
calculated from conditional logistic regression models adjusted
for breast cancer risk factors. Tests for trend were conducted by
using the Wald test on quintile medians modeled continuously.
We examined the possibly nonlinear relation between carotenoids
and risk of breast cancer nonparametrically with restricted cubic
). Tests for nonlinearity used the likelihood ratio test
for comparison of the model with only the linear term to the
model with the linear and cubic spline terms.
We conducted separate analyses for carotenoids measured
,10 and $10 y before diagnosis by using all available measures
(i.e., first and second blood samples). In women with 2 blood
samples, we simultaneously modeled carotenoid concentrations
in the first (1989–1990) and second (2000–2002) collections.
We also conducted analyses of all available cases and controls
by using one carotenoids measure or, when available, the
average of 2 measures. Unconditional logistic regression models,
which were additionally adjusted for matching factors, were
used for stratified analyses.
We investigated interactions between carotenoids and
followup time and lifestyle factors, including BMI (in kg/m2), smoking,
alcohol intake, PMH use, and menopausal status, by using
likelihood ratio tests. We assessed associations with carotenoids
by ER status and luminal A (ER-positive or PR-positive,
HER2negative, and grade 1 or 2), luminal B (ER positive or PR
positive and either HER2-positive or HER2-negative and grade
3), and triple-negative tumors (ER-negative, PR-negative, and
). We also examined associations by tumor
invasiveness, histologic grade, tumor size, and lymph node
status. To test whether associations differed by tumor subtype,
we used polychotomous logistic regression (
) with a
likelihood ratio test for the comparison of a model with separate
slopes for carotenoids in each case group to one with a common
slope. We conducted a survival analyses in breast cancer cases
by using Cox proportional hazards models, calculating the
person-time from diagnosis to the first of recurrence or breast
cancer death, other death, or end of follow-up (June 2010).
Primary survival analyses were adjusted for breast cancer risk
factors; secondary analyses were further adjusted for tumor stage,
ER and PR status, hormone therapy, chemotherapy, and radiation
therapy. All P values were based on 2-sided tests and considered
statistically significant if #0.05; interactions with P . 0.05 but
#0.10 were considered suggestive. Analyses were conducted
with SAS version 9 software (SAS Institute) or STATA version
12.1 software (StataCorp).
At the first collection, cases were more likely than controls to
be nulliparous, have a history of benign breast disease, and have
a family history of breast cancer (Table 1). By the second
collection, women were, on average, 10 y older, slightly heavier,
and less likely to smoke. Approximately two-thirds of women
were postmenopausal at the first blood collection; nearly all
women were postmenopausal at the second collection.
Tenyear ICCs ranged from 0.30 (b-carotene) to 0.54 (lutein and
Significant inverse associations were observed between
a-carotene, b-carotene, lycopene, and total carotenoids and
breast cancer risk with overall 18–28% lower risk in the top
compared with bottom quintiles (Table 2). Results were similar
between simple and multivariate conditional logistic regression
models (data not shown). Additional adjustment for physical
activity, plasma cholesterol concentrations, or vitamin D intake
did not alter the results (data not shown). The association with
lycopene was suggestively stronger with measures 10–20
compared with ,10 y before diagnosis [top compared with bottom
quintile RRs (95% CIs): $10 y, 0.69 (0.50, 0.94; P-trend =
0.01); ,10 y: 0.87 (0.70, 1.07; P-trend = 0.14, P-heterogeneity =
0.09]. Associations with b-carotene and total carotenoids were
apparent for measures both 10–20 and ,10 y before diagnosis
(P-interaction = 0.55 and 0.11, respectively). Analyses restricted
to women with 2 blood samples yielded similar results (data not
shown). Tests for nonlinearity were NS for any individual
carotenoids or total carotenoids (data not shown).
The correlation between carotenoids was highest in the pro–
vitamin A carotenoids (a-carotene, b-carotene, and b-cryptoxanthin;
Spearman r = 0.48–0.76). Associations for a-carotene (except
when adjusted for b-carotene) and b-carotene remained when
adjusted for the other carotenoids (data not shown). The lycopene
association was attenuated with adjustment for a-carotene or
Associations between individual carotenoids and breast cancer
risk were similar by alcohol intake, PMH use, and menopausal
status at blood collection and in women who received screening
mammograms within 2 y of blood collection (data not shown).
The associations of b-carotene and total carotenoids with breast
cancer risk differed significantly by BMI (Table 3). Significant
inverse associations were observed in lean women [BMI ,25
top compared with bottom quintile RRs (95% CIs): b-carotene,
0.62 (0.47, 0.83; P-trend , 0.001); total carotenoids, 0.64 (0.48,
0.84; P-trend , 0.001)], whereas no association was observed
for overweight or obese women [e.g., BMI $30 RRs: b-carotene,
0.96 (P-trend = 0.86, P-interaction = 0.04); total carotenoids, 0.98
(P-trend = 0.75, P-interaction = 0.02)]. Although b-cryptoxanthin
was not associated with breast cancer overall, it was significantly
inversely associated with risk in lean women (top compared with
bottom quintile RR: 0.70; 95% CI: 0.53, 0.92; P-trend = 0.05,
P-heterogeneity = 0.08). The association of a-carotene with
breast cancer was significantly stronger in nonsmokers
(comparable RR: 0.74; 95% CI: 0.60, 0.92; P-trend = 0.01) than in
current smokers (RR: 1.23; 95% CI: 0.54, 2.80; P-trend = 0.22,
P-interaction = 0.03).
1989–1990 blood draw
2000–2002 blood draw
1NHS, Nurses’ Health Study.
2Mismatched case and control numbers were due to missing carotenoid measures for first or second blood collection.
3Mean 6 SD (all such values).
4Unchanged between first and second blood collection (all such values).
5All values are medians; 10th, 90th percentiles in parentheses.
1Multivariate conditional logistic regression models were adjusted for BMI at age 18 y; weight gain since age 18 y; ages at menarche, first birth, and
menopause; parity; alcohol intake; history of benign breast disease; and family history of breast cancer. Overall values use the 1990 blood collection or the
average of 1990 and 2000 blood collections if available. n cases: ,10 y, 1828; $10 y, 895; overall, 2147. P-heterogeneity ,10 compared with $10 y:
a-carotene, 0.85; b-carotene, 0.55; b-cryptoxanthin, 0.47; lutein and zeaxanthin, 0.20; lycopene, 0.09; total carotenoids, 0.11. NHS, Nurses’ Health Study.
BMI (in kg/m2) ,25
BMI from 25 to ,30
BMI from 25 to ,30
BMI from 25 to ,30
Lutein and zeaxanthin (mg/dL)
BMI from 25 to ,30
BMI from 25 to ,30
Total carotenoids (mg/dL)
BMI from 25 to ,30
1Multivariate unconditional logistic regression models were adjusted for matching factors; BMI at age 18 y; weight gain since age 18 y; ages at menarche,
first birth, and menopause; parity; alcohol intake; history of benign breast disease; and family history of breast cancer. NHS, Nurses’ Health Study.
Associations did not differ by tumor size, invasiveness, or nodal
involvement (data not shown). RRs were similar for ER-positive and
ER-negative cases (e.g., b-carotene top compared with bottom
quintile RRs: 0.70 and 0.72, respectively) although the number of
ER-negative cases was limited (n = 291–292) (Table 4).
Associations with b-carotene were suggestively stronger in poorly
differentiated tumors (P-heterogeneity = 0.08). Associations with
a-carotene and b-carotene were significantly inverse for luminal B,
but not luminal A, tumors, although tests for heterogeneity were
NS [e.g., b-carotene top compared with bottom quintile RRs (95%
CIs): luminal B, 0.47 (0.28, 0.77); P-trend = 0.003; luminal A: 0.80
(0.59, 1.09); P-trend = 0.08; P-heterogeneity = 0.32]. No significant
associations were observed in triple-negative tumors (n = 107–108).
a-Carotene, b-carotene, b-cryptoxanthin, and total carotenoids
were strongly inversely associated with risk of breast tumors that
recurred or were ultimately lethal (Table 4). Risks in the top
quintile (compared with bottom quintile) were 46–68% lower for
a-carotene (RR: 0.54; 95% CI: 0.35, 0.83; P-trend = 0.01),
b-carotene (RR: 0.32; 95% CI: 0.21, 0.51; P-trend , 0.001), and
total carotenoids (RR: 0.48; 95% CI: 0.31, 0.73; P-trend = 0.001)
(P-heterogeneity compared with nonrecurrent and nonlethal cases =
0.08, ,0.001, and 0.02, respectively). b-Cryptoxanthin, although
not associated with overall breast cancer risk, was associated with
significantly lower risk of recurrent or lethal disease (RR: 0.68;
95% CI: 0.45, 1.04; P-trend = 0.008); however, the test for
heterogeneity was NS (P = 0.30). b-Carotene and total carotenoid
concentrations were significantly inversely associated with
recurrence and breast cancer death in survival analyses in cases
[RRs (95% CIs): 0.47 (0.31, 0.71; P-trend = 0.002) and 0.65
(0.43, 0.96; P-trend = 0.04), respectively] (data not shown).
Associations were slightly attenuated with additional adjustment for
tumor and treatment characteristics although the b-carotene
association remained significant (RR: 0.52; 95% CI: 0.34, 0.79;
Ptrend = 0.003) (data not shown). When restricted to breast cancer
deaths only (n = 176), point estimates were similar, but CIs were
wider because of to the smaller sample size (data not shown).
1Multivariate unconditional logistic regression models were adjusted for matching factors; BMI at age 18 y; weight gain since age 18 y; ages at menarche,
first birth, and menopause; parity; alcohol intake; history of benign breast disease; and family history of breast cancer. ER, estrogen receptor; NHS, Nurses’
2P-heterogeneity across subtypes from polychotomous logistic regression.
In this large, nested, case-control study of plasma carotenoid
concentrations and breast cancer risk, we observed significant
modest inverse associations with a-carotene, b-carotene, and
total carotenoids and a suggestive inverse association with
lycopene. Associations were apparent for measures both ,10 and
$10 y before the diagnosis of breast cancer although
associations with some carotenoids were suggestively stronger with
distant measures. Associations were stronger in leaner women.
Inverse associations with carotenoids were not different by ER
status but were suggestively stronger for more-aggressive tumors,
including poorly differentiated and luminal B tumors. a-Carotene,
b-carotene, b-cryptoxanthin, and total carotenoids were
associated with strong, significant reduced risks of recurrent or lethal
This study built on our recent pooled analysis (
) of plasma
carotenoids and breast cancer risk with more than double the
number of NHS cases and the investigation of carotenoid exposure
timing and breast cancer subtypes. Our current observations of
inverse associations with plasma a-carotene, b-carotene, and total
carotenoid concentrations overall were in agreement with the
pooled results. However, we did not observe a significant
association with lutein and zeaxanthin concentrations in the current
study and only a suggestive association with lycopene
concentrations. Our results were also consistent with one (
) of 2 (
) recent studies that were not included in the pooled analysis in
which inverse associations were observed with a-carotene and
We investigated whether the carotenoid–breast cancer
associations varied across exposures that may have contributed to
oxidative stress because women with a greater likelihood of oxidative
stress, such as smokers, may be more likely to benefit from high
amounts of antioxidants such as carotenoids. Similar to the pooled
), we did not observe interactions with alcohol intake,
menopausal status, or PMH use. Our ability to detect interactions
with smoking was limited by the small number of current
smokers. Although adiposity contributes to oxidative stress (
in both pooled (
) and current analyses, we observed stronger
inverse associations with plasma carotenoid concentrations in lean
women than in overweight and obese women. However, in
contrast to the pooled (
) results, we did not observe positive
associations in overweight or obese women. Because fat-soluble
carotenoids are stored in adipose tissue (
), and BMI is inversely
associated with plasma carotenoid concentrations (
circulating carotenoid concentrations in overweight women likely
incorporate more exposure misclassification. Although it is not clear
if adipose carotenoids are bioavailable, there is evidence to
suggest that an exchange with other tissue sources contributes to
plasma concentrations (26).
To our knowledge, the Women’s Health Initiative (
) is the
only other study of multiple measures of carotenoids and breast
cancer risk, although the time between measures (#6 y) and
follow-up time (median: 8 y; maximum: 12 y) were shorter than
in our study. Although associations with more-recent measures of
a-carotene and b-carotene in the Women’s Health Initiative were
suggestively stronger, CIs for each time period (1–3, 2–4, and 3–5 y)
before diagnosis were wide given the small number of cases (n =
190), and baseline a-carotene concentrations also were associated
with significantly lower risk. In the current analysis, with the
unique advantage of 2 samples collected 10 y apart, we examined
proximate (,10 y before diagnosis) and distant (10–20 y before
diagnosis) carotenoid concentrations. Although we reported very
good reproducibility of carotenoids over a 2–3-y period (ICCs:
), the 10-y reproducibility was attenuated (ICCs:
0.39–0.54), which was similar to 15-y correlations published
), which suggests that distant measures are not
simply a proxy for recent exposure. Because of this, and because
we observed inverse associations with both proximate and distant
measures, carotenoids may play important roles both early and
late in carcinogenesis. However, associations with lycopene and
total carotenoids were suggestively stronger for distant than for
proximate measures, which is consistent with the stronger
doseresponse observed with total carotenoids measured 10–15 y before
diagnosis (compared with 1–9 y before diagnosis) in a previous
publication of 295 cases (
). Together, these results suggest
carotenoids may inhibit tumor initiation, which is compatible with
hypothesized mechanisms, including the conversion of pro–
vitamin A carotenoids to retinol, which regulates cell growth,
differentiation, and apoptosis (
8, 30, 31
), and the antioxidant
capacity to scavenge reactive oxygen species and prevent DNA
5–7, 32, 33
). Although experimental evidence also
suggested that carotenoids may inhibit progression after initiation
), the observed suggested association with distant
carotenoids measures warrants additional investigation.
Although we did not observe significant heterogeneity by ER
status, we had fewer ER-negative cases than in the pooled analysis
(n = 292 compared with 417) (
). RRs were comparably inverse
for both ER-positive and ER-negative tumors, and carotenoids
have inhibited the growth of both ER-positive and ER-negative
cell lines (
). These findings support carotenoids as a modifiable
risk factor for ER-negative as well as ER-positive breast cancer.
To our knowledge, this is the first study to investigate the
associations with prediagnostic plasma carotenoid concentrations by
molecular subtype or recurrent or lethal breast cancer. Although
heterogeneity was NS, associations with plasma carotenoids
tended to be stronger for luminal B than for luminal A tumors. For
the subset of tumors that recurred or caused death, inverse
associations with plasma carotenoids were considerably stronger
with a .46% reduction in risk in top quintiles of a-carotene,
b-carotene, b-cryptoxanthin, and total carotenoids. Although
increased fruit and vegetable intake after diagnosis was not
associated with breast cancer recurrence in the Women’s Health
Eating and Living (WHEL) Study, reduced risk of recurrence
was observed with higher baseline (
) and cumulative (
plasma carotenoids, which suggested that early concentrations
may be associated with longer-term response. Although the
WHEL Study assessed postdiagnostic carotenoid concentrations,
it is possible, given that one measure of carotenoids is
representative of longer-term exposure, that WHEL results also were
reflective of prediagnostic concentrations as were assessed in our
analysis. This issue of timing relative to diagnosis and breast cancer
survival deserves additional study. The combination of these results
and results of our prior pooled analysis (
) suggest that
carotenoids may be particularly important for the prevention of aggressive,
and deadly, breast tumors.
With many breast cancer cases, our study had several strengths.
To our knowledge, it is the first comprehensive study of the
importance of carotenoid exposure timing with blood samples
collected 10 y apart and with 20 y of follow-up. It is also the first
investigation of the associations by molecular subtype and
recurrent or lethal breast cancer. However, there were also limitations
in our study. Although we could not eliminate the possibility of
residual confounding, the comprehensive information on breast
cancer risk factors in the NHS allowed for thorough adjustment for
potential confounders. Although only one blood sample was
available for the majority of women, reproducibility over a 2–3-y
period in the NHS is very good (
). In addition, we reduced
measurement error by averaging the values of 2 blood samples
10 y apart when available. We had too few cases of HER2-type
and basal-like tumors to examine these subtypes separately; future
pooled analyses of these rarer subtypes would be beneficial.
Although there are biologically plausible mechanisms through
which carotenoids may reduce breast cancer risk (
is possible that the observed association may have been due to
other phytochemicals in fruit and vegetables correlated with
carotenoids or an interaction between various phytochemicals (36).
In conclusion, the results of this large prospective analysis
suggest that women with higher circulating carotenoid
concentrations are at reduced breast cancer risk, particularly for tumors
that are more aggressive and have worse prognosis. In addition,
carotenoid measures both close to and more than a decade before
diagnosis appear protective but may be particularly important for
preventing tumor initiation. Although additional work is necessary
to confirm the causal role of carotenoids, and the use of specific
carotenoid supplements is not advised (
carotenoid concentrations are responsive to changes in dietary intake
of carotenoid-rich fruit and vegetables, such as carrots, sweet
potatoes, leafy greens, and tomatoes (
). Because intake of
fruit and vegetables is beneficial for many reasons, the association
of higher plasma carotenoid concentrations with lower risk of
aggressive and lethal breast cancer may encourage women to
increase their consumption of carotenoid-rich fruit and vegetables.
We thank the participants and staff of the NHS for their valuable
contributions as well as the following state cancer registries for their help: Alabama,
Arizona, Arkansas, California, Colorado, Connecticut, Delaware, Florida,
Georgia, Idaho, Illinois, Indiana, Iowa, Kentucky, Louisiana, Maine,
Maryland, Massachusetts, Michigan, Nebraska, New Hampshire, New Jersey,
New York, North Carolina, North Dakota, Ohio, Oklahoma, Oregon,
Pennsylvania, Rhode Island, South Carolina, Tennessee, Texas, Virginia,
Washington, and Wyoming. In addition, this study was approved by the Connecticut
Department of Public Health Human Investigations Committee. Certain data
used in this publication were obtained from the Department of Public Health.
The authors assume full responsibility for analyses and interpretation of these
data. Study sponsors had no role in the design of the study, collection,
analysis, and interpretation of data, writing of the manuscript, or decision to
submit the manuscript for publication.
The authors’ responsibilities were as follows—AHE, BR, RMT, SST, and
SEH: designed the research; AHE, RMT, SST, and SEH: conducted the
research; XL performed statistical analyses; AHE: had primary
responsibility for the final content of the manuscript; and all authors: wrote the
manuscript and read and approved the final manuscript. No authors
declare a conflict of interest. None of the authors reported a conflict of interest
related to the study.
1. Motchnik PA , Frei B , Ames BN . Measurement of antioxidants in human blood plasma . Methods Enzymol 1994 ; 234 : 269 - 79 .
2. Willett W. Nutritional Epidemiology . New York: Oxford University Press; 1998 .
3. Krinsky NI . Carotenoids and cancer: basic research studies . In: Frei B, editor. Natural antioxidants in human health and disease . San Diego (CA): Academic Press; 1994 : p. 239 - 62 .
4. Bertram JS . Dietary carotenoids, connexins and cancer: what is the connection? Biochem Soc Trans 2004 ; 32 : 985 - 9 .
5. Bendich A . Carotenoids and the immune response . J Nutr 1989 ; 119 : 112 - 5 .
6. Krinsky NI . Actions of carotenoids in biological systems . Annu Rev Nutr 1993 ; 13 : 561 - 87 .
7. Shultz TD , Chew BP , Seaman WR , Luedecke LO . Inhibitory effect of conjugated dienoic derivatives of linoleic acid and beta-carotene on the in vitro growth of human cancer cells . Cancer Lett 1992 ; 63 : 125 - 33 .
8. Sporn MB , Roberts AB . Role of retinoids in differentiation and carcinogenesis . Cancer Res 1983 ; 43 : 3034 - 40 .
9. Prakash P , Russell RM , Krinsky NI . In vitro inhibition of proliferation of estrogen-dependent and estrogen-independent human breast cancer cells treated with carotenoids or retinoids . J Nutr 2001 ; 131 : 1574 - 80 .
10. Zhang X , Spiegelman D , Baglietto L , Bernstein L , Boggs DA , van den Brandt PA , Buring JE , Gapstur SM , Giles GG , Giovannucci E , et al. Carotenoid intakes and risk of breast cancer defined by estrogen receptor and progesterone receptor status: a pooled analysis of 18 prospective cohort studies . Am J Clin Nutr 2012 ; 95 : 713 - 25 .
11. Mangels AR , Holden JM , Beecher GR , Forman MR , Lanza E. Carotenoid content of fruits and vegetables: an evaluation of analytic data . J Am Diet Assoc 1993 ; 93 : 284 - 96 .
12. Micozzi MS , Beecher GR , Taylor PR , Khachik F. Carotenoid analyses of selected raw and cooked foods associated with a lower risk for cancer . J Natl Cancer Inst 1990 ; 82 : 282 - 5 .
13. Eliassen AH , Hendrickson SJ , Brinton LA , Buring JE , Campos H , Dai Q , Dorgan JF , Franke AA , Gao Y-T , Goodman MT , et al. Circulating carotenoids and risk of breast cancer: pooled analysis of eight prospective studies . J Natl Cancer Inst 2012 ; 104 : 1905 - 16 .
14. Hankinson SE , Willett WC , Manson JE , Hunter DJ , Colditz GA , Stampfer MJ , Longcope C , Speizer FE . Alcohol, height, and adiposity in relation to estrogen and prolactin levels in postmenopausal women . J Natl Cancer Inst 1995 ; 87 : 1297 - 302 .
15. Zhang X , Tworoger SS , Eliassen AH , Hankinson SE . Postmenopausal plasma sex hormone levels and breast cancer risk over 20 years of follow-up . Breast Cancer Res Treat 2013 ; 137 : 883 - 92 .
16. El-Sohemy A , Baylin A , Kabagambe E , Ascherio A , Spiegelman D , Campos H . Individual carotenoid concentrations in adipose tissue and plasma as biomarkers of dietary intake . Am J Clin Nutr 2002 ; 76 : 172 - 9 .
17. Rosner B , Cook N , Portman R , Daniels S , Falkner B . Determination of blood pressure percentiles in normal-weight children: some methodological issues . Am J Epidemiol 2008 ; 167 : 653 - 66 .
18. Tamimi RM , Colditz GA , Hazra A , Baer HJ , Hankinson SE , Rosner B , Marotti J , Connolly JL , Schnitt SJ , Collins LC . Traditional breast cancer risk factors in relation to molecular subtypes of breast cancer . Breast Cancer Res Treat 2012 ; 131 : 159 - 67 .
19. Holmes MD , Chen WY , Li L , Hertzmark E , Spiegelman D , Hankinson SE . Aspirin intake and survival after breast cancer . J Clin Oncol 2010 ; 28 : 1467 - 72 .
20. Durrleman S , Simon R . Flexible regression models with cubic splines . Stat Med 1989 ; 8 : 551 - 61 .
21. Marshall RJ , Chisholm EM . Hypothesis testing in the polychotomous logistic model with an application to detecting gastrointestinal cancer . Stat Med 1985 ; 4 : 337 - 44 .
22. Kabat GC , Kim M , Adams-Campbell LL , Caan BJ , Chlebowski RT , Neuhouser ML , Shikany JM , Rohan TE , Investigators WHI . Longitudinal study of serum carotenoid, retinol, and tocopherol concentrations in relation to breast cancer risk among postmenopausal women . Am J Clin Nutr 2009 ; 90 : 162 - 9 .
23. Maillard V , Kuriki K , Lefebvre B , Boutron-Ruault M-C , Lenoir GM , Joulin V , Clavel-Chapelon F , Chajes V . Serum carotenoid, tocopherol and retinol concentrations and breast cancer risk in the E3N-EPIC study . Int J Cancer 2010 ; 127 : 1188 - 96 .
24. Keaney JF , Larson MG , Vasan RS , Wilson PWF , Lipinska I , Corey D , Massaro JM , Sutherland P , Vita JA , Benjamin EJ. Framingham Study . Obesity and systemic oxidative stress: clinical correlates of oxidative stress in the Framingham Study . Arterioscler Thromb Vasc Biol 2003 ; 23 : 434 - 9 .
25. Virtanen SM , van't Veer P , Kok F , Kardinaal AF , Aro A . Predictors of adipose tissue carotenoid and retinol levels in nine countries . The EURAMIC Study. Am J Epidemiol 1996 ; 144 : 968 - 79 .
26. Krinsky NI , Mayne ST , Sies H. Carotenoids in health and disease . Boca Raton (FL): CRC Press; 2004 .
27. Kotsopoulos J , Tworoger SS , Campos H , Chung F-L , Clevenger CV , Franke AA , Mantzoros CS , Ricchiuti V , Willett WC , Hankinson SE , et al. Reproducibility of plasma and urine biomarkers among premenopausal and postmenopausal women from the Nurses' Health Studies . Cancer Epidemiol Biomarkers Prev 2010 ; 19 : 938 - 46 .
28. Comstock GW , Burke AE , Hoffman SC , Norkus EP , Gross M , Helzlsouer KJ . The repeatability of serum carotenoid, retinoid, and tocopherol concentrations in specimens of blood collected 15 years apart . Cancer Epidemiol Biomarkers Prev 2001 ; 10 : 65 - 8 .
29. Sato R , Helzlsouer KJ , Alberg AJ , Hoffman SC , Norkus EP , Comstock GW . Prospective study of carotenoids, tocopherols, and retinoid concentrations and the risk of breast cancer . Cancer Epidemiol Biomarkers Prev 2002 ; 11 : 451 - 7 .
30. Gudas L , Sporn M , Roberts AB . Cellular biology and biochemistry of the retinoids . In: Sporn MB , Roberts AB , Goodman DS , editors. The retinoids: biology , chemistry, and medicine. New York: Raven Press; 1994 . p. 443 - 520 .
31. Simeone A -M, Tari AM . How retinoids regulate breast cancer cell proliferation and apoptosis . Cell Mol Life Sci 2004 ; 61 : 1475 - 84 .
32. Sies H , Stahl W. Vitamins E and C, beta-carotene, and other carotenoids as antioxidants . Am J Clin Nutr 1995 ; 62 ( 6 Suppl) : 1315S - 21S .
33. Frei B . Reactive oxygen species and antioxidant vitamins: mechanisms of action . Am J Med . 199497 : 5S - 13S ; discussion: 22S - 28S .
34. Rock CL , Flatt SW , Natarajan L , Thomson CA , Bardwell WA , Newman VA , Hollenbach KA , Jones L , Caan BJ , Pierce JP . Plasma carotenoids and recurrence-free survival in women with a history of breast cancer . J Clin Oncol 2005 ; 23 : 6631 - 8 .
35. Rock CL , Natarajan L , Pu M , Thomson CA , Flatt SW , Caan BJ , Gold EB , Al-Delaimy AK , Newman VA , Hajek RA , et al. Longitudinal biological exposure to carotenoids is associated with breast cancer-free survival in the Women's Healthy Eating and Living Study . Cancer Epidemiol Biomarkers Prev 2009 ; 18 : 486 - 94 .
36. Liu RH . Health benefits of fruit and vegetables are from additive and synergistic combinations of phytochemicals . Am J Clin Nutr 2003 ; 78 ( 3 Suppl) : 517S - 20S .
37. Omenn GS , Goodman GE , Thornquist MD , Balmes J , Cullen MR , Glass A , Keogh JP , Meyskens FL , Valanis B , Williams JH , et al. Effects of a combination of beta carotene and vitamin A on lung cancer and cardiovascular disease . N Engl J Med 1996 ; 334 : 1150 - 5 .
38. The effect of vitamin E and beta carotene on the incidence of lung cancer and other cancers in male smokers . The Alpha-Tocopherol, Beta Carotene Cancer Prevention Study Group. N Engl J Med 1994 ; 330 : 1029 - 35 .
39. US Department of Agriculture, Agricultural Research Service. 2011 . USDA National Nutrient Database for Standard Reference , release 24 [Internet]. [cited 2014 Sep 23 ]. Available from: http://www.ars.usda. gov/nutrientdata
40. Pierce JP , Natarajan L , Sun S , Al-Delaimy W , Flatt SW , Kealey S , Rock CL , Thomson CA , Newman VA , Ritenbaugh C , et al. Increases in plasma carotenoid concentrations in response to a major dietary change in the women's healthy eating and living study . Cancer Epidemiol Biomarkers Prev 2006 ; 15 : 1886 - 92 .
41. Yeum KJ , Booth SL , Sadowski JA , Liu C , Tang G , Krinsky NI , Russell RM . Human plasma carotenoid response to the ingestion of controlled diets high in fruits and vegetables . Am J Clin Nutr 1996 ; 64 : 594 - 602 .