Prediction of late distant recurrence in patients with oestrogen-receptor-positive breast cancer: a prospective comparison of the breast-cancer index (BCI) assay, 21-gene recurrence score, and IHC4 in the TransATAC study population

Background Biomarkers to improve the risk-benefi t of extended adjuvant endocrine therapy for late recurrence in patients with oestrogen-receptor-positive breast cancer would be clinically valuable. We compared the prognostic ability of the breast-cancer index (BCI) assay, 21-gene recurrence score (Oncotype DX), and an immuno histochemical prognostic model (IHC4) for both early and late recurrence in patients with oestrogen-receptor-positive, node-negative (NO) disease who took part in the Arimidex, Tamoxifen, Alone or in Combination (ATAC) clinical trial. Methods In this prospective comparison study, we obtained archival tumour blocks from the TransATAC tissue bank from all postmenopausal patients with oestrogen-receptor-positive breast cancer from whom the 21-gene recurrence score and IHC4 values had already been derived. We did BCI analysis in matched samples with sufficient residual RNA using two BCI models-cubic (BCI-C) and linear (BCI-L)-using previously validated cutoffs. We assessed prognostic ability of BCI for distant recurrence over 10 years (the primary endpoint) and compared it with that of the 21-gene recurrence score and IHC4. We also tested the ability of the assays to predict early (0-5 years) and late (5-10 years) distant recurrence. To assess the ability of the biomarkers to predict recurrence beyond standard clinicopathological variables, we calculated the change in the likelihood-ratio. chi(2) (LR-Delta chi(2)) from Cox proportional hazards models. Findings Suitable tissue was available from 665 patients with oestrogen-receptor-positive, NO breast cancer for BCI analysis. The primary analysis showed significant differences in risk of distant recurrence over 10 years in the categorical BCI-C risk groups (p< 0 . 0001) with 6 . 8% (95% CI 4 . 4-10 . 0) of patients in the low-risk group, 17 . 3% (12 . 0-24 . 7) in the intermediate group, and 22 . 2% (15 . 3-31 . 5) in the high-risk group having distant recurrence. The secondary analysis showed that BCI-L was a much stronger predictor for overall (0-10 year) distant recurrence compared with BCI-C (interquartile HR 2 . 30 [95% CI 1 . 62-3 . 27]; LR-Delta chi(2) = 22 . 69; p< 0 . 0001). When compared with BCI-L, the 21-gene recurrence score was less predictive (HR 1 . 48 [95% CI 1 . 22-1 . 78]; LR-Delta chi 2 = 13 . 68; p= 0 . 0002) and IHC4 was similar (HR 1 . 69 [95% CI 1 . 51-2 . 56]; LR-Delta chi(2) = 22 . 83; p< 0 . 0001). All further analyses were done with the BCI-L model. In a multivariable analysis, all assays had signifi cant prognostic ability for early distant recurrence (BCI-L HR 2 . 77 [95% CI 1 . 63-4 . 70], LR-Delta chi(2) = 15 . 42, p< 0 . 0001; 21-gene recurrence score HR 1 . 80 [1 . 42-2 . 29], LR Delta chi(2) = 18 . 48, p< 0 . 0001; IHC4 HR 2 . 90 [2 . 01-4 . 18], LR-Delta chi(2) = 29 . 14, p< 0 . 0001); however, only BCI-L was signifi cant for late distant recurrence (BCI-L HR 1 . 95 [95% CI 1 . 22-3 . 14], LR-Delta chi(2) = 7 . 97, p= 0 . 0048; 21-gene recurrence score HR 1 . 13 [0 . 82-1 . 56], LR-Delta chi(2) = 0 . 48, p= 0 . 47; IHC4 HR 1 . 30 [0 . 88-1 . 94], LR-Delta chi(2) = 1 . 59, p= 0 . 20). Interpretation BCI-L was the only signifi cant prognostic test for risk of both early and late distant recurrence and identified two risk populations for each timeframe. It could help to identify patients at high risk for late distant recurrence who might benefit from extended endocrine or other therapy.