Treated ER+ cohort
|
METABRIC
|
Loi
|
Zhang
|
Combined Fisher test
|
---|
Univariate |
n=926 |
n=250 |
n=136 | Â |
HR (95%) | 1.18 (1.07–1.31) | 1.6 (1.24–2.05) | 1.56 (0.97–2.51) |  |
P
| 0.001 | 0.0003 | 0.06 | <0.00001 |
Multivariate |
n=685 |
n=207 |
n=136a
| Â |
HR (95%) | 1.13 (1.00–1.27) | 1.49 (1.11–2.00) | 1.56 (0.97–2.51) |  |
P
| 0.041 | 0.008 | 0.06 | 0.001 |
Untreated ER+ cohort
|
METABRIC
|
Wang
|
Miller
|
Combined Fisher test
|
Univariate |
n=360 |
n=209 |
n=127 | Â |
HR (95%) | 1.12 (0.95–1.32) | 1.13 (0.91–1.39) | 1.32 (0.83–2.10) |  |
P
| 0.17 | 0.26 | 0.23 | 0.16 |
Multivariate |
n=272 |
n=209a
|
n=126 | Â |
HR (95%) | 1.09 (0.89–1.34) | 1.13 (0.91–1.39) | 0.96 (0.61–1.54) |  |
P
| 0.41 | 0.26 | 0.88 | 0.58 |
- These are for endocrine-treated [20,30] and untreated [29,31] ER+ cohorts, as indicated, as well as for a meta-analysis under a combined Fisher test. Untreated means cases did not receive either endocrine or chemotherapy, but may have received radiotherapy. The number of samples, hazard ratios (HRs), 95 % confidence intervals and log-rank test P values are given for both univariate and multivariate (adjusted for stage/nodal status, grade and tumor size) analyses. Perturbation activity scores were scaled to unit variance before running the Cox regressions to allow meaningful comparison of HR values.
-
aNote that for the Wang dataset [29], all samples were lymph node negative and no information on tumor size and grade was available. Similarly, for the Zhang dataset [30], no additional clinical information was available. For the Miller dataset [31], the multivariate analysis was adjusted for size and grade only since all untreated samples were lymph node negative. For the METABRIC set, overall survival was used as the endpoint; for all others we used distant-metastasis-free survival. In all cases, outcome data were censored at 10 years.