Annual Meeting of the NCI Cohort Consortium (Abstract Submission): Submission #13

Lightning Talks Abstract
------------------------
Presenter's First Name: : Reuben
Presenter's Last Name:: Frost
Title (eg: professor, assistant professor, chair, etc):: {Empty}
Degree(s): MMath
Contact Email:: reuben.frost@icr.ac.uk
Organization:: The Institute of Cancer Research
Project Title:: Adding serum hormone measurement to improve the Tyrer-Cuzick breast cancer risk model in postmenopausal women
Additional Authors:
1. First Name: Tahania
   Last Name: Ahmad
   Degree(s): Ph.D.
   Organization: Queen Mary University of London
2. First Name: Anne
   Last Name: MacGregor
   Degree(s): M.D.
   Organization: Queen Mary University of London
3. First Name: Heather
   Last Name: Eliassen
   Degree(s): Sc.D.
4. First Name: Susan
   Middle Initial: E
   Last Name: Hankinson
   Degree(s): Sc.D.
5. First Name: Roger
   Middle Initial: L
   Last Name: Milne
   Degree(s): Ph.D.
   Organization: Cancer Council Victoria
6. First Name: Celine
   Middle Initial: M
   Last Name: Vachon
   Degree(s): Ph.D.
7. First Name: Amy
   Last Name: Berrington
   Degree(s): Ph.D.
   Organization: The Institute of Cancer Research
8. First Name: Adam
   Last Name: Brentnall
   Degree(s): Ph.D.
   Organization: Queen Mary University of London
9. First Name: Jack
   Last Name: Cuzick
   Degree(s): Ph.D.
   Organization: Queen Mary University of London
10. First Name: Judith
    Last Name: Offman
    Degree(s): Ph.D.
    Organization: Queen Mary University of London
11. First Name: Montserrat
    Last Name: García-Closas
    Degree(s): M.D.
    Organization: The Institute of Cancer Research

Abstract::
Introduction: The primary objective is to evaluate the potential of incorporating the ratio of serum estradiol (E2) and sex hormone binding globulin (SHBG) into breast cancer risk assessments using the Tyrer-Cuzick (T-C) breast cancer risk prediction model. Currently there is limited research on their integration in individual risk assessments.

Methods: E2/SHBG levels will be added to the current T-C model, using a two-stage process applied to data from 2,666 cases (invasive breast cancers) and 4,512 non-cases from three nested case-control studies and one case-cohort study, all part of the B2RISK consortium to: 
(1) develop a statistical model to explain variation in E2 and SHBG levels based on factors in the T-C model; and
 (2) estimate the odds ratio with the "hormone residual" (i.e., observed minus expected E2/SHBG from (1)), adjusted for T-C 10-year risk.  
The model will be validated in the Breast Cancer Now Generations (BGS) prospective cohort in the UK including 42,075 eligible participants, 847 with incident breast cancer diagnosed within a 5-years follow-up.

Results: Calibration of the current T-C model has been assessed in a BGS nested case-control sample with data on classical risk factors, 313-SNP polygenic risk score and mammographic density (1,777 controls and 610 breast cancer cases), using 5-year risk scores generated by T-C model v8 before adding hormones. The model was well calibrated (calibration slope 1.0 (95%CI 0.67-1.30) with an age-adjusted AUC=0.65 (95%CI 0.62-0.68).

Conclusion:  T-C scores are well calibrated with modest risk discrimination. Further analyses will evaluate whether adding E2/SHBG improves the model.