Adjuvant! online is overoptimistic in predicting survival of Asian breast cancer patients

• Main Authors: Bhoo-Pathy, N., Yip, C.H., Hartman, M., Saxena, N., Taib, N.A., Ho, G.F., Looi, L.M., Bulgiba, A.M., van der Graaf, Y., Verkooijen, H.M.
• Format: Article
• Language: English
• Published: 2012
• Subjects: R Medicine
• Online Access: http://eprints.um.edu.my/3731/1/Adjuvant%21_Online_is_overoptimistic_in_predicting_survival.pdf; http://eprints.um.edu.my/3731/
• Institution: Universiti Malaya

BACKGROUND: Adjuvant! Online is a free web-based tool which predicts 10-year breast cancer outcomes and the efficacy of adjuvant therapy in patients with breast cancer. As its prognostic performance has only been validated in high income Caucasian populations, we validated the model in a middle income Asian setting. METHODS: Within the University Malaya Hospital-Based Breast Cancer Registry, all 631 women who were surgically treated for invasive non-metastatic breast cancer between 1993 and 2000 were identified. The discriminative performance of Adjuvant! Online was tested using receiver operating characteristic (ROC) analysis. Calibration of the model was evaluated by comparing predicted 10-year overall survival with observed 10-year survival. FINDINGS: Adjuvant! Online was fairly capable in discriminating between good and poor survivors, as attested by the area under ROC curve of 0.73 (95 Confidence Interval: 0.69-0.77). Overall, Adjuvant! Online predicted 10 year survival (70.3) was significantly higher than the observed 10 year survival (63.6, difference of 6.7; 95 CI: 3.0-10.4). The model was especially overoptimistic in women under 40 years and in women of Malay ethnicity, where survival was overestimated by approximately 20 (95 CI: 9.8-29.8) and 15 (95 CI: 5.3-24.5) respectively. INTERPRETATION: Even though Adjuvant! Online is capable of discriminating between good and poor survivors, it systematically overestimates survival. These findings suggest that the model requires adaptation prior to use in Asian settings.