PREDICT: A NEW UK PROGNOSTIC MODE THAT PREDICTS SURVIVAL FOLLOWING SURGERY FOR INVASIVE BREAST CANCER (GC Wishart, EM Azzato, PDP Pharoah, DC Greenberg, J Rashbass, O Kearins, G Lawrence, C Caldas C, PM Ravdin)

Add to your list(s) Download to your calendar using vCal

• Prof. Gordon Wishart, Addenbrooke's Hospital
• Tuesday 12 January 2010, 12:00-13:00
• CRI Lecture Theatre.

If you have a question about this talk, please contact Mala Jayasundera.

AIM : To develop and validate a prognostication model to predict overall survival for women treated for early breast cancer in the UK based on cancer registry data. Unlike SEER , this dataset includes accurate information on mode of detection as well as systemic therapy.

METHOD : Using the Eastern Cancer Registration & Information Centre (ECRIC) dataset, information was collated for 5818 women diagnosed with invasive breast cancer in East Anglia from 1999-2003. All patients underwent surgery, had records of pathological staging (tumour size, grade, lymph node status, ER status), systemic treatment (chemotherapy, hormone therapy, both), mode of detection (screen-detected, symptomatic) and follow up, censored on 31 December 2007. A model was derived from these data using Cox proportional hazards, with ER positive and ER negative tumours modelled separately, and this was subsequently validated in an external dataset of 5468 patients from the West Midlands Cancer Intelligence Unit (WMCIU). Validation was performed by comparing (a) observed and expected mortality (overall & breast cancer specific) at 5 & 8 years and (b) receiver operating characteristic (ROC) curves in both ECRIC & WMCIU datasets.

RESULTS ECRIC dataset Difference in overall observed/expected mortality ROC curve (AUC) was 0.81. Difference in breast cancer specific observed/expected mortality AUC was 0.84. WMCIU dataset: Difference in overall observed/expected mortality < 1% at 5 years (15.8% vs 16.5%) and 8 years (17.5% vs 17.8%). AUC was 0.79. Difference in breast cancer specific observed/expected mortality AUC was 0.82. Overall model fit was good across all subgroups although the ER positive model provided better discrimination (AUC 0.82) than ER negative (AUC 0.75). There was no significant difference between the ROC curves generated with ECRIC and WMCIU data (ER positive X2 = 0.17, p=0.68; ER negative X2 =0.00, P=0.95).

CONCLUSION We have developed a prognostication model for early breast cancer based on data from a UK cancer registry that has included mode of detection for the first time. The model is well calibrated, provides a high degree of discrimination and has been validated in a second patient cohort. This model will underpin a new web-based prognostication and treatment benefit tool for early breast cancer based on UK data (PREDICT).

This talk is part of the Cambridge Oncology Seminar Series series.

This talk is included in these lists:

• All Talks (aka the CURE list)
• Biology
• CCC talks for website
• CRI Lecture Theatre
• CRUK CI Seminars
• Cambridge Oncology Seminar Series
• Life Sciences
• Life Sciences
• ME Seminar
• MRC Cancer Unit Seminars
• PMRFPS's
• cancer
• cri
• my_list
• other talks
• se393's list
• sfm36

Note that ex-directory lists are not shown.