Table 5.
Multivariate logistic regression models for prediction of PD‐L1 positivity
| OR | 95% CI | P‐value | OR | 95% CI | P‐value | ||
|---|---|---|---|---|---|---|---|
| Model 1 (clinical variables) | Model 2 (clinical variables + CT radiomic features) | ||||||
| Age ≤ 59 years | 0.7281 | 0.3362–1.5232 | 0.408 | 0.6106 | 0.2691–1.3259 | 0.2227 | |
| Female sex | 1.0820 | 0.3793–3.1603 | 0.883 | 0.9354 | 0.3121–2.8772 | 0.9053 | |
| Current or ex‐smoker | 1.0635 | 0.3823–3.1009 | 0.907 | 0.9115 | 0.3129–2.7532 | 0.8658 | |
| Presence of EGFR mutation | 1.1710 | 0.5272–2.5799 | 0.695 | 1.1098 | 0.4821–2.5304 | 0.8042 | |
| Rad‐score > −0.715 | N/A | N/A | N/A | 3.4706 | 1.6919–7.2840 | 0.0008 | |
| C‐statistic (95% CI) | 0.550 (0.454–0.646) | 0.667 (0.575–0.760) | P‐value for comparison of c‐statistics between two models | 0.0299 | |||
| Bootstrapped c‐statistic (95% CI) | 0.550 (0.461–0.6488) | 0.667 (0.577–0.764) | |||||
CI, confidence interval; EGFR, epidermal growth factor receptor; OR, odds ratio; PD‐L1, programmed death ligand 1.