AS316 - Validation of Promising Breast Cancer Early Detection Biomarkers


Investigator Names and Contact Information

Christopher Li []


An estimated 192,370 women were diagnosed with breast cancer in the United States in 2009,

accounting for 27% of all cancers diagnosed among U.S. women.1 With respect to mortality, 40,170 women

will die of breast cancer in 2009, accounting for 15% of all cancer related deaths among U.S. women. Breast

cancer mortality rates have declined over the past decade in the U.S. primarily as a result of the

implementation of population-based screening programs and advances in breast cancer treatment. Despite

these changes, breast cancer is still associated with significant morbidity (recurrence, anxiety, and treatment

side effects) and mortality. Continued improvements in our ability to detect breast cancer early offer the

promise of further reducing the burden of this disease, as breast cancer detected at an earlier stage is much

more curable than is metastatic disease. The current 5-year survival rate for localized breast cancer in the U.S.

is 98%, but is only 27% for metastatic disease.1


We recently completed a unique large scale study aimed at discovering breast cancer early detection

biomarkers described in detail in Section 3 using WHI OS samples. Our validation of EGFR as a potential

breast cancer early detection biomarker described in this section provides important evidence that there may

indeed be breast cancer early detection biomarkers detectable in plasma, that our discovery platforms were

able to identify viable candidates, and that more comprehensive validation of our discovered candidates,

especially those much higher ranked, is warranted. A limitation of the validation work we have conducted thus

far is that we have only been able to assess candidates that have a commercially available ELISA assay.

Unfortunately this has restricted our work to the follow-up of only relatively low ranked candidates. For

example, the log2 ratio for EGFR in our discovery experiments focused on ER+ breast cancer was 0.22

(p=0.072), placing it near the bottom of our list when sorted by log2 ratio or p-value. Our top three candidates

for ER+ disease from IPAS were catalase (log2 ratio=1.92, p-value=0.007), Notch3 (log2 ratio=1.35, p-value

=0.001), and stress induced phosphoprotein 1 (log2 ratio=1.02, p-value=0.002). Not only are their log2 ratios

and p-values much more compelling than EGFR, for all three there is literature linking them to breast cancer.

Our overall goal is to reduce suffering and death due to breast cancer by validating plasma biomarkers

that can be used to detect breast cancer early, when it is most treatable. We believe that the necessary next

step toward this goal is to cull our long list of candidate biomarkers to those we can have greater confidence in

given the inherent false discovery rate of our discovery work. In order to do this we propose to conduct a

second screen of promising candidates using an antibody array platform. This approach is feasible using

relatively small sample volumes and is promising because ~95% of our candidates have at least one

commercially available antibody, and ~90% have five or more available antibodies. Thus, through this second

screen we can confirm the potential performance of our candidates on individual samples, which is critical

given that one of our primary discovery platforms, IPAS, only evaluated pooled samples. Thus we propose to

complete the following aim:

Conduct a second screen of the most promising breast cancer early detection biomarkers to identify

the candidates most suitable for subsequent formal validation:

We have identified a lengthy list of promising candidates, however due to the inherent false discovery rate of this work this list needs to be culled.

In completing this aim we will consider breast cancer both as a whole and according to clinically relevant

subtypes based on tumor marker expression and histology. Analyses will also be conducted that consider

various patient characteristics that are known or suspected to influence the plasma proteome, such as use of

menopausal hormone therapy. Given the biological heterogeneity of breast cancer, we expect that there will be

markers unique to different clinically and molecularly defined breast cancer subtypes.


1. Jemal A, Siegel R, Ward E, Hao Y, Xu J, Thun MJ. Cancer statistics, 2009. CA Cancer J Clin