Table 3. Cox regression analysis of 165 FIGO II/III/IV serous ovarian cancer patients.
| Factor n = 165, 78 death | Univariate Cox regression | Multiple Cox regression | ||
|---|---|---|---|---|
| HR (CI95) | p | HR (CI95) | p | |
| Age (decades) | 1.43 (1.16–1.75) | < 0.001 | 1.49 (1.20–1.86) | < 0.001 |
| FIGO (IV vs III vs II) | 2.51 (1.56–4.04) | < 0.001 | 2.69 (1.56–4.65) | < 0.001 |
| Grade (3 vs 1&2) | 2.11 (1.19–3.79) | 0.011 | 1.59 (0.87–2.94) | 0.133 |
| Residual tumor (yes vs no) | 1.76 (1.11–2.79) | 0.017 | 1.48 (0.91–2.40) | 0.117 |
| Mol. Subclass (II vs I) | 2.06 (1.29–3.28) | 0.002 | 1.82 (1.11–2.98) | 0.017 |
| Peritoneal carc. (yes vs no) | 3.72 (1.85–7.48) | < 0.001 | 2.73 (1.33–5.61) | 0.006 |
| EMF | 0.58 (0.31–1.08) | 0.086 | -1 | - |
| TSF | 1.86 (0.65–5.34) | 0.247 | 3.77 (1.14–12.39) | 0.029 |
1
Removed from the final Cox regression model by Akaike's information criterion selection (AIC, a variable-penalized criterion). If EMF is not excluded from the final model the TSF is even more predictive (HR = 4.24, CI95 (1.28–14.07), p = 0.018).