Table 4.
Cox proportional hazards regression model of overall survival in patients with non-small cell lung cancer compared high and low densities, close with long distances and mixed with unmixed pattern relative to malignant cells, adjusted for clinicopathologic variables
| Variable | B | SE | Wald | HR | 95% CI for Exp(B) | P* |
|---|---|---|---|---|---|---|
| Histologic type (ADC vs SCC) | −1.103 | 1.081 | 1.042 | 0.332 | 0.040–2.759 | 0.307 |
| Smoker (no vs yes) | 1.702 | 1.938 | .772 | 5.485 | 0.123–244.607 | 0.380 |
| Tumor size (≤1.5 cm vs >3.8 cm) | −2.328 | 0.854 | 7.432 | 0.098 | 0.018–0.520 | 0.006 |
| KRAS (wild-type vs mutant) | 1.452 | 0.981 | 2.191 | 4.273 | 0.625–29.231 | 0.139 |
| EGFR (wild-type vs mutant) | −1.411 | 1.207 | 1.367 | .244 | 0.023–2.597 | 0.242 |
| Low vs high densities | ||||||
| CD3 + | 3.377 | 0.996 | 11.506 | 29.289 | 4.161–206.159 | 0.001 |
| CD3 + CD8 + | 2.882 | 1.126 | 6.545 | 17.847 | 1.962–162.338 | 0.011 |
| CD3 + CD8 + GZB + | 0.524 | 1.296 | 0.164 | 1.689 | 0.133–21.429 | 0.686 |
| CD3 + CD45RO + | −2.725 | 1.669 | 2.665 | 0.066 | 0.002–1.727 | 0.103 |
| CD3 + CD8 + CD45RO + | 0.644 | 1.288 | 0.250 | 1.905 | 0.153–23.766 | 0.617 |
| CD3 + CD8negFOXP3 + | 0.913 | 0.670 | 1.854 | 2.491 | 0.670–9.268 | 0.173 |
| CD3 + CD45RO + FOXP3 + | −4.203 | 1.485 | 8.008 | 0.015 | 0.001–0.275 | 0.005 |
| CD3 + PD-1+ | −0.142 | 0.798 | 0.032 | 0.868 | 0.182–4.143 | 0.859 |
| CD3 + CD8 + PD-1+ | 4.311 | 1.405 | 9.414 | 74.550 | 4.746–1170.936 | 0.002 |
| CD3 + PD-L1 + | 8.036 | 1.729 | 21.612 | 3090.038 | 104.373–91482.972 | 0.000 |
| CD3 + CD8 + PD-L1 + | 0.486 | 1.224 | 0.157 | 1.625 | 0.148–17.880 | 0.692 |
| CD3 + PD-1+PD-L1 + | −3.646 | 1.074 | 11.519 | 0.026 | 0.003–0.214 | 0.001 |
| Mixed vs unmixed pattern | ||||||
| CD3 + | −3.073 | 1.124 | 7.471 | 0.046 | 0.005–0.419 | 0.006 |
| CD3 + CD8 + | 0.995 | 1.241 | .643 | 2.704 | 0.238–30.771 | 0.423 |
| CD3 + CD8 + GZB + | −0.593 | .800 | .551 | .552 | 0.115–2.648 | 0.458 |
| CD3 + CD45RO + | 3.454 | 1.152 | 8.990 | 31.623 | 3.307–302.378 | 0.003 |
| CD3 + CD8 + CD45RO + | 0.526 | 1.196 | 0.193 | 1.692 | 0.162–17.628 | 0.660 |
| CD3 + CD8negFOXP3 + | 7.917 | 61.229 | 0.017 | 2743.024 | 0.000–3.601E + 55 | 0.897 |
| CD3 + CD45RO + FOXP3 + | −1.087 | 0.844 | 1.661 | 0.337 | 0.065–1.761 | 0.197 |
| CD3 + PD-1+ | −1.861 | 1.136 | 2.683 | 0.155 | 0.017–1.442 | 0.101 |
| CD3 + CD8 + PD-1+ | 1.004 | 0.775 | 1.679 | 2.729 | 0.598–12.464 | 0.195 |
| CD3 + PD-L1 + | −.508 | 1.065 | 0.228 | 0.602 | 0.075–4.855 | 0.633 |
| CD3 + CD8 + PD-L1 + | 1.288 | 1.139 | 1.279 | 3.624 | 0.389–33.763 | 0.258 |
| CD3 + PD-1+PD-L1 + | 3.291 | 1.346 | 5.976 | 26.874 | 1.920–376.094 | 0.014 |
| Close vs long distance from malignant cells | ||||||
| CD3 + | −0.174 | 1.009 | 0.030 | 0.840 | 0.116–6.071 | 0.863 |
| CD3 + CD8 + | −1.376 | 1.253 | 1.206 | 0.253 | 0.022–2.945 | 0.272 |
| CD3 + CD8 + GZB + | −2.967 | 1.383 | 4.604 | .051 | 0.003–0.773 | 0.032 |
| CD3 + CD45RO + | −1.248 | 1.281 | 0.949 | 0.287 | 0.023–3.537 | 0.330 |
| CD3 + CD8 + CD45RO + | 1.043 | 1.273 | 0.672 | 2.839 | 0.234–34.388 | 0.412 |
| CD3 + CD8negFOXP3 + | 2.446 | 1.018 | 5.773 | 11.538 | 1.569–84.822 | 0.016 |
| CD3 + CD45RO + FOXP3 + | 3.032 | 1.150 | 6.956 | 20.735 | 2.179–197.331 | 0.008 |
| CD3 + PD-1+ | 1.461 | 1.100 | 1.763 | 4.311 | 0.499–37.260 | 0.184 |
| CD3 + CD8 + PD-1+ | 1.754 | 1.012 | 3.007 | 5.778 | 0.796–41.956 | 0.083 |
| CD3 + PD-L1 + | 5.445 | 1.706 | 10.188 | 231.550 | 8.178–6556.148 | 0.001 |
| CD3 + CD8 + PD-L1 + | 2.316 | 1.920 | 1.455 | 10.134 | 0.235–436.512 | 0.228 |
| CD3 + PD-1+PD-L1 + | −2.021 | 1.354 | 2.229 | 0.132 | 0.009–1.882 | 0.135 |
Note: B unstandardized regression weight, SE multiple linear regression, Wald Wald test, HR hazard ratio, CI confidence interval, ADC adenocarcinoma, SCC squamous cell carcinoma, GZB granzyme B.
*Boldface indicates statistically significant difference using Cox proportional-hazards model with un-adjusted P-values for clinicopathologic variables.
The table shows the analysis of densities, and pattern of cellular distribution and distances relative to malignant cells to the 12 relevant cell phenotypes of the multiplex immunofluorescence panels adjusted by clinicopathologic features.