Table 3. Identification of the gene expression signature by univariate and multivariate Cox regression analyses.
| No. | Gene | Univariate analysis* | Multivariate analysis** | |||||
|---|---|---|---|---|---|---|---|---|
| HR | 95% CI | P | HR | 95% CI | P | |||
| 1 | GPX3 | 0.7399 | 0.6117–0.895 | 0.0019 | 0.6238 | 0.4789–0.8125 | 0.0005 | |
| 2 | PTPRM | 1.3738 | 1.1002–1.7154 | 0.0051 | ||||
| 3 | ASPM | 1.2755 | 1.0745–1.514 | 0.0054 | ||||
| 4 | CENPF | 1.2543 | 1.0607–1.4833 | 0.0081 | ||||
| 5 | TK1 | 1.3054 | 1.0733–1.5878 | 0.0076 | ||||
| 6 | PRR36 | 0.7927 | 0.6663–0.9431 | 0.0088 | ||||
| 7 | PLEK2 | 1.3833 | 1.1464–1.6691 | 0.0007 | 1.3876 | 1.0767–1.7883 | 0.0114 | |
| 8 | SLC2A1 | 1.2589 | 1.0766–1.4721 | 0.0039 | ||||
| 9 | FAM83A | 1.2135 | 1.0796–1.3639 | 0.0012 | ||||
| 10 | MIF | 1.4026 | 1.1054–1.7797 | 0.0054 | ||||
| 11 | PRC1 | 1.3344 | 1.0801–1.6486 | 0.0075 | ||||
| 12 | GJB2 | 1.1823 | 1.0597–1.319 | 0.0027 | ||||
| 13 | HMMR | 1.3209 | 1.105–1.579 | 0.0022 | ||||
| 14 | ANLN | 1.3082 | 1.1129–1.5377 | 0.0011 | ||||
| 15 | KCNK5 | 0.7836 | 0.6821–0.9003 | 0.0006 | ||||
| 16 | COL1A1 | 1.1987 | 1.0522–1.3655 | 0.0064 | 1.2173 | 1.02–1.4528 | 0.0293 | |
*, the 16 DEGs were significantly associated with OS (P<0.01) according to univariate Cox regression analysis; **, we then performed multivariate Cox regression analysis on these 16 DEGs to identify the most informative gene set for survival prediction. Finally, the three genes marked in grey in the table were selected for multivariate Cox regression analysis and generation of a prognostic risk model according to their respective regression coefficients. HR, hazard ratio; CI, confidence interval; DEGs, differentially-expressed genes; OS, overall survival.