Abstract P4-11-34: Validation of the CancerMath prognostic tool for breast cancer in Southeast Asia

Abstract

Abstract Background: CancerMath is a set of web-based prognostic tools for non-metastatic breast cancer patients. It predicts probability of nodal and nipple involvement as well as overall and cancer-specific survival with and without adjuvant therapy at any of the 15 years after diagnosis. But it has not been externally validated outside United States. This study assessed its performance in a Southeast Asian setting. Methods: Using Singapore Malaysia Hospital Based Breast Cancer Registry, clinical information were retrieved from 7064 Stage I to stage III breast cancer patients who were diagnosed between 1990 and 2011 and underwent surgery. Probability of positive nodes and overall survival at any of the 15 years after diagnosis were computed using the CancerMath algorithm. Predicted and observed outcomes were compared. Discriminative ability was evaluated by area under a receiver operating characteristic curve (AUC). And calibration was assessed by plotting observed survival against predicted survival for each decile of the predicted survival and a 45 degree line was added to illustrated perfect agreement. Results: Nodal status calculator predicted 40.6% of patients to be node positive which was similar to the observed 43.6%. The AUC was 0.71 (95% CI, 0.70-0.72). For outcome calculator, the observed and predicted overall survivals was 83.4% vs 87.3% at 5 year after diagnosis and 70.4% vs 75.3% at 10 year after diagnosis. The difference appeared smaller for patients from Singapore (5-year and 10-year predicted-observed= 2.5% and 0% respectively) comparing to patients from Malaysia (5-year and 10-year predicted-observed= 5.8% and 8.0% respectively), as well as for patients diagnosed in more recent years. The AUC for 5-year and 10-year overall survival was 0.77 (95% CI: 0.75-0.79) and 0.74 (95% CI: 0.71-0.76). For cases with relatively good and moderate prognosis, CancerMath predictions were similar to observed outcome as the plotted dots in the calibration plot were close to the 45 degree reference line from the fourth decile onwards. For therapy calculator, overestimation of survival persisted for most demographic, pathologic and treatment subgroups, especially for women younger than 40 years, with larger and high-grade tumor and with more lymph nodes involved. The AUC for 5-year overall survival is 0.73 (95% CI: 0.70-0.77). And with the exception of two groups, all plotted dots were below the 45 degree reference line in the calibration plot . Conclusion: The discriminative performance and calibration of CancerMath was fair in this validation study. The greater discordance observed in Malaysian patients might be explained by different life expectancy, adverse tumor characteristics, compliance to treatment and lifestyle after diagnosis between Singapore, Malaysia and United States. The results suggested that direct application of CancerMath should only be limited to certain subgroups in Southeast Asia. Citation Format: Nur Aishah Taib, Hui Miao, Helena M Verkooijen, Mikael Hartman, Hoong-Seam Wong, Cheng-Har Yip, Ern-Yu Tan, Soo-Chin Lee, Nirmala Bhoo-Pathy. Validation of the CancerMath prognostic tool for breast cancer in Southeast Asia [abstract]. In: Proceedings of the Thirty-Seventh Annual CTRC-AACR San Antonio Breast Cancer Symposium: 2014 Dec 9-13; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2015;75(9 Suppl):Abstract nr P4-11-34.