Table 4.
Univariate and Multivariate analyses for OS, PFS, and ORR.
| Univariate analyses | Multivariate analyses* | |||||
|---|---|---|---|---|---|---|
| OS | HR | 95% CI | P | HR | 95% CI | P |
| T19007C PG | 2.08 | 0.80–5.37 | 0.131 | 4.90 | 1.69–14.22 | 0.004 |
| C8092A PG | 0.40 | 0.13–1.24 | 0.112 | 0.19 | 0.06–0.62 | 0.006 |
| P for interaction: NA | ||||||
| PFS | HR | 95% CI | P | HR | 95% CI | P |
| T19007C PG | 1.14 | 0.56–2.30 | 0.717 | 2.47 | 1.16–5.26 | 0.018 |
| C8092A PG | 0.39 | 0.17–0.89 | 0.025 | 0.17 | 0.07–0.44 | <0.001 |
| P for interaction: NS | ||||||
| ORR | Odds ratio | 95% CI | P | Odds ratio | 95% CI | P |
| T19007C PG | 1.20 | 0.32–4.51 | 0.787 | 0.08 | 0.004–1.59 | 0.099 |
| C8092A PG | 22.5 | 3.88–130.41 | <0.001 | 301.13 | 8.21–11045.48 | 0.002 |
| P for interaction: NS | ||||||
For OS and PS, a Cox Proportional Hazard Model was fitted to the data starting with the following covariates: T19007C PG (Positive vs. Negative), C8092A PG (Positive vs. Negative), Histology (Squamous vs. Non-Squamous), Age (≤ 65 vs. >65 years), ECOG PS (0 vs. 1–2), Gender (Male vs. Female), Number of prior systemic regimens (1 vs. 2 vs. >3). In both final models (OS and PFS) only T19007C PG, C8092A PG and gender were retained (P < 0.05).
For ORR, a logistic regression model was used with objective response as the dependent variable and same covariates used in the PFS and OS models. Only T19007C PG, C8092A PG, and age were retained in the final model (P < 0.05).
The interaction between T19007C PG and C8092A PG was assessed within the same 3 final models, by evaluating the change in the log-likelihood of the model associated with the introduction of the interaction term. However, for OS, the p for interaction could not be estimated because of failure of model convergence.