Population-based validation study of Adjuvant! for primary breast cancer patients in the Netherlands

Background: Adjuvant! is an online risk prediction tool that can support breast cancer patients and physicians in clinical decision-making regarding adjuvant systemic treatment. The aim of this study was to validate Adjuvant! on a Dutch population-based cohort of women diagnosed in 2003 and assess its calibration and discriminatory accuracy.

Methods: All Dutch patients diagnosed with invasive primary breast cancer in 2003 in the Netherlands meeting the criteria of Adjuvant!; unilateral, unicentric invasive breast cancer, no evidence of metastatic or residual disease, pathologically staged I-III and no pT4 features, were included. Local treatment consisted of breast conserving therapy or ablative surgery and axillary staging. Adjuvant systemic treatment was given according to national guidelines. Patient, tumor and treatment characteristics were retrieved from the Netherlands Cancer Registry. Endpoints were ten-year overall survival, breast cancer specific survival and event free survival.

Findings: A total of 8,195 patients were included. Out of all patients, 4,389 (53.6%) received some form of adjuvant systemic therapy. Throughout a ten-year follow-up, 2,156 (26.3%) patients died and 1,953 (23.8%) patients developed a recurrent disease (locoregional, distant or contralateral invasive cancer or DCIS). Ten-year observed overall survival was underestimated by -1.4% (95% CI -2.3 to -0.4, p=0.006), breast cancer specific survival was underestimated by -3.3% (95% CI -4.1 to -2.6, p<0.0001) and event free survival was underestimated by -3.5% (95% CI -4.5 to -2.6, p<0.0001). Discrimination was moderate for overall survival and breast cancer specific survival but poor for event free survival.

Interpretation: Adjuvant! does not accurately predict breast cancer specific survival and event free survival in this Dutch population-based study. Dutch patients showed better ten-year survival rates than predicted by Adjuvant!. For specific patient groups, the model could be used to estimate overall survival. We conclude that Adjuvant! should be used with caution in clinical practice in the Netherlands but probably also in other countries and the US.