Using Metabolomics to Explore the Role of Postmenopausal Adiposity in Breast Cancer Risk
JNCI J Natl Cancer Inst (
Using Metabolomics to Explore the Role of Postmenopausal Adiposity in Breast Cancer Risk
Jessica A. Lasky-Su 0 1
Oana A. Zeleznik 0 1
A. Heather Eliassen heather.eliassen@channing 0 1
0 The Author(s) 2018. Published by Oxford University Press. All rights reserved. For Permissions , please
1 Affiliations of authors: Channing Division of Network Medicine, Department of Medicine, Brigham and Women's Hospital and Harvard Medical School , Boston, MA (JALS, OAZ , AHE); Department of Epidemiology, Harvard T.H. Chan School of Public Health , Boston, MA, AHE , USA
Excess adiposity after menopause is a well-established, and
potentially modifiable, breast cancer risk factor (
). Breast cancer
has a clear hormonal origin, and adiposity is likely to influence
breast cancer risk through estrogens given that aromatization
of androgens in adipose tissue is the primary estrogen source
after menopause. In fact, overweight and obese
postmenopausal women have 50% higher circulating estrogen levels
compared with lean women (
). Underpinning the prominent
hormonal hypothesis, excess adiposity is more strongly
associated with estrogen receptor–positive (ERþ) breast cancers (
addition to estrogen (
), other mechanisms that may contribute
to the body mass index (BMI)–breast cancer association include
inflammation and insulin resistance. Perturbations in metabolic
systems are hallmarks of obesity, and a logical place to
investigate additional mechanisms.
The study of metabolomics, relatively recently applied in
epidemiologic settings, offers great potential to explore
differences in metabolomic profiles and subsequent breast cancer
risk, such as the Moore et al. report in this issue of the Journal
). However, to date, the only other study to prospectively
assess the association between small molecule metabolites and
breast cancer risk is from the same group of investigators, who
reported several diet-related metabolites associated with
In this issue of the Journal, Moore et al. (
) report on the
correlation between BMI and circulating metabolites, and then on
the association between BMI-related metabolites and breast
cancer risk. The nested case–control study within the Prostate,
Lung, Colorectal, and Ovarian Cancer Screening Trial (PLCO)
included 621 invasive breast cancer case subjects and 621
matched controls. A total of 617 known metabolites, measured
in archived serum samples, were included in the analysis.
Correlation analyses resulted in 67 metabolites that were
modestly associated with BMI (partial Pearson correlation range
¼ –0.26 to 0.35). Of these, seven metabolites were statistically
significantly associated with overall breast cancer risk after
correcting for multiple comparisons (false discovery rate < 0.2),
including two androgen-related metabolites and three
branchedchain amino acid (BCAA)–related metabolites. Twenty-three
metabolites were statistically significantly associated with ERþ
breast cancer (n ¼ 418 case subjects), including the two
androgen-related metabolites and a total of six BCAA-related
metabolites. No metabolites were statistically significantly
associated with the smaller group of 144 ER- breast cancers. With
forward stepwise selection, one androgen-related metabolite
(16alpha-hydroxy-DHEA-3-sulfate) was selected for overall and
ERþ breast cancer risk, and one (overall) or three (ERþ)
BCAArelated metabolites were selected.
The methodological approach of Moore et al. (
), focusing on
the subset of metabolites correlated with BMI, has an implicit
advantage in its reductionist nature. While this approach is
effective for identifying metabolites specific to the BMI–breast
cancer relationship, there are consequences of this strategy
that deserve consideration. From the outset, such an approach
leaves the lingering question of which remaining metabolites
from the full set of 617 may be important in breast cancer
etiology. Within a reductionist approach, the choices of analytic
techniques also may result in missed metabolites. For instance,
a Bonferroni approach in the screening step does not account
for the high correlations among the metabolites and is likely
overly conservative. A less conservative approach that accounts
for correlations among metabolites (eg, “number of effective
tests” ) would decrease the possibility of missing metabolites
in the initial BMI screening stage. Second, the agnostic
datadriven approach of Pearson correlations does not accommodate
metabolites or pathways previously reported to be associated
with BMI. For example, while all three BCAAs (leucine,
isoleucine, and valine) have been associated with BMI in prior
), Moore at al. identified only valine as associated
with BMI. Although a systematic comparison with published
BMI-associated metabolites would enhance this approach, we
recognize that the task of comparing metabolomics results
across studies is daunting given the lack of consistent reporting
across studies. Without standard identifiers, such as those from
the human metabolome (
), reported along with the metabolite
names, it is unclear if there is overlap between metabolites with
different names. Complicating this issue is the rapidly growing
nature of the field of metabolomics, where many platforms
identify metabolites that are not yet catalogued in the Human
The final step in the analysis by Moore et al. (
) was to assess
the change in the BMI–breast cancer association when
accounting for BMI-related metabolites associated with breast cancer.
In the overall analysis, the association was attenuated with the
addition of the two selected metabolites. The authors estimate
that 57.6% of the BMI–breast cancer association is due to the
indirect effect through these metabolites. While impressive, it
seems likely that this is an overestimate given the
wellestablished associations between BMI and estradiol (
estradiol and breast cancer risk (
). Indeed, in a collaborative
analysis of eight prospective studies (
), adjusting for circulating
estradiol resulted in even more attenuation of the BMI–breast
cancer association than reported by Moore et al., although they
reported no correlation between the selected metabolite
16alpha-hydroxy-DHEA-3-sulfate and estradiol. Although the BMI–
breast cancer association is stronger among ERþ tumors (
was not the case in the analysis by Moore et al. This complicates
the mediation interpretation when metabolites are included in
a model of BMI and ERþ breast cancer risk.
Moore et al. (
) make important strides through their
exploration of the BMI–metabolomics–breast cancer association.
Further steps could include examination of the metabolic
pathways where the selected metabolites operate, to strengthen the
understanding of the underlying biology. Several analytic
approaches could be utilized for this. One approach would
evaluate the association between all metabolites in the metabolic
pathway and breast cancer risk with a more liberal statistical
significance threshold given the correlation among metabolites
within a pathway. A more sophisticated approach would assess
the pathway metabolites simultaneously through multivariate
techniques, such as principal components analysis, partial least
squares discriminant analysis, network approaches (eg,
weighted gene co-expression network analysis), or metabolite
set enrichment analysis. Evaluation of metabolic pathways may
yield important additional information on the functional nature
of the relationship between the selected metabolite and breast
The two potential mechanisms identified, steroid hormone
and BCAA, are of particular interest in breast cancer etiology.
While the steroid hormone mechanism has been investigated
in detail, the hypothesis that BCAA metabolism may contribute
to breast carcinogenesis is intriguing. BCAA levels are predictors
of insulin resistance and type 2 diabetes risk, independent of
), and have been associated with pancreatic cancer (
The reported association with breast cancer warrants additional
investigation and replication in future studies. In conclusion,
Moore et al. provide an exciting look into the possible metabolic
underpinnings of breast cancer etiology related to adiposity,
and we anticipate that future investigations of the role of
metabolomics in breast cancer will work to confirm these
findings and also expand to a broader snapshot of the metabolome
to better understand the biologic networks of importance.
Dr. Lasky-Su is a consultant to Metabolon. Drs. Zeleznik and
Eliassen have no disclosures.
1. Lauby-Secretan B , Scoccianti C , Loomis D , et al. Body fatness and cancerviewpoint of the IARC Working Group . N Engl J Med . 2016 ; 375 ( 8 ): 794 - 798 .
2. Endogenous Hormones Breast Cancer Collaborative Group . Circulating sex hormones and breast cancer risk factors in postmenopausal women: Reanalysis of 13 studies . Br J Cancer . 2011 ; 105 ( 5 ): 709 - 722 .
3. Key TJ , Appleby PN , Reeves GK , et al. Body mass index, serum sex hormones, and breast cancer risk in postmenopausal women . J Natl Cancer Inst . 2003 ; 95 ( 16 ): 1218 - 1226 .
4. Moore SC , Playdon MC , Sampson JN , et al. A metabolomics analysis of body mass index and postmenopausal breast cancer risk . J Natl Cancer Inst . 2018 ; 110 ( 6 ): 588 - 597 .
5. Playdon MC , Ziegler RG , Sampson JN , et al. Nutritional metabolomics and breast cancer risk in a prospective study . Am J Clin Nutr . 2017 ; 106 ( 2 ): 637 - 649 .
6. Gao X , Starmer J , Martin ER . A multiple testing correction method for genetic association studies using correlated single nucleotide polymorphisms . Genet Epidemiol . 2008 ; 32 ( 4 ): 361 - 369 .
7. Moore SC , Matthews CE , Sampson JN , et al. Human metabolic correlates of body mass index . Metabolomics . 2014 ; 10 ( 2 ): 259 - 269 .
8. Newgard CB , An J , Bain JR , et al. A branched-chain amino acid-related metabolic signature that differentiates obese and lean humans and contributes to insulin resistance . Cell Metab . 2009 ; 9 ( 4 ): 311 - 326 .
9. Ho JE , Larson MG , Ghorbani A , et al. Metabolomic profiles of body mass index in the Framingham Heart Study reveal distinct cardiometabolic phenotypes . PLoS One . 2016 ; 11 ( 2 ): e0148361 .
10. Wurtz P , Wang Q , Kangas AJ , et al. Metabolic signatures of adiposity in young adults: Mendelian randomization analysis and effects of weight change . PLoS Med . 2014 ; 11 ( 12 ): e1001765 .
11. Park S , Sadanala KC , Kim EK . A metabolomic approach to understanding the metabolic link between obesity and diabetes . Mol Cells . 2015 ; 38 ( 7 ): 587 - 596 .
12. Shin AC , Fasshauer M , Filatova N , et al. Brain insulin lowers circulating BCAA levels by inducing hepatic BCAA catabolism . Cell Metab . 2014 ; 20 ( 5 ): 898 - 909 .
13. Wishart DS , Feunang YD , Marcu A , et al. HMDB 4 . 0: The human metabolome database for 2018 [Epub ahead of print November 11, 2017 ]. Nucleic Acids Res . 2017 . doi: 10 .1093/nar/gkx1089.
14. Key T , Appleby P , Barnes I , Reeves G , Endogenous H , Breast Cancer Collaborative Group. Endogenous sex hormones and breast cancer in postmenopausal women: Reanalysis of nine prospective studies . J Natl Cancer Inst . 2002 ; 94 ( 8 ): 606 - 616 .
15. Wang TJ , Larson MG , Vasan RS , et al. Metabolite profiles and the risk of developing diabetes . Nat Med . 2011 ; 17 ( 4 ): 448 - 453 .
16. Mayers JR , Wu C , Clish CB , et al. Elevation of circulating branched-chain amino acids is an early event in human pancreatic adenocarcinoma development . Nat Med . 2014 ; 20 ( 10 ): 1193 - 1198 .