Table. Pathologic Complete Response (pCR) Rates and Hazard Ratios of Event-Free Survival (EFS) and Distant Recurrence–Free Survival (DRFS) for pCR vs Not pCR by Molecular Subtypea.
| Subtype | No. | pCR rate, % (95% CIb) | Hazard ratio (95% CI) | |
|---|---|---|---|---|
| EFS | DRFS | |||
| HR+ ERBB2− | 361 | 17 (14-22) | 0.14 (0.03-0.55) | 0.16 (0.04-0.64) |
| HR+ ERBB2+ | 173 | 40 (33-48) | 0.15 (0.03-0.63) | 0.10 (0.01-0.77) |
| HR− ERBB2− | 326 | 42 (36-47) | 0.18 (0.09-0.34) | 0.20 (0.10-0.40) |
| HR− ERBB2+ | 90 | 68 (57-77) | 0.14 (0.05-0.41) | 0.18 (0.06-0.53) |
| All | 950 | 35 (32-38) | 0.19 (0.12-0.31) | 0.21 (0.13-0.34) |
Abbreviation: HR, hormone receptor.
a
The observation that every hazard ratio for molecular subtype is smaller than the overall hazard ratio is not a typographic error. It is an example of the Simpson paradox, which was observed as well in the I-SPY1 trial.23 This observation demonstrates the importance of considering molecular subtype when evaluating the association of pCR with EFS and DRFS.
b
Based on binomial exact (Clopper-Pearson) CI method.