Skip to main content
. 2023 Dec 21;13:1212454. doi: 10.3389/fonc.2023.1212454

Table 3.

Independent predictors (Pi) building the multivariate Cox proportional hazard regression model (21, 22) for progression-free survival data-driven extracted via the stepwise-forward likelihood ratio selection method of covariates (2931) among HPV-driven head and neck squamous cell carcinoma.

Covariate Ref. Characteristic PFS events n (%) p-value univariate HR (95% CI) p-value multi-variate Loss in χ 2 for deleted term p-value for deleted term p-value for bootstrap Natural logarithm ln HR
Alcohol (g/day) <60 > 60 14 (73.7) 0.0002 3.839 (1.858 –7.931) graphic file with name fonc-13-1212454-i001.jpg 0.0003 11.955 0.0005 0.0040 1.345
Localization OPSCC HNSCC 20 (54.1) 0.0743 1.747 (0.917 –3.327) 0.0897 2.896 0.0888 0.1508 0.558
T category Other T T4 16 (61.5) 0.0457 1.884 (0.969 –3.664) 0.0619 3.323 0.0683 0.1209 0.633
ENE N0 or ENE- ENE+ 18 (52.9) 0.2642 2.374 (0.894 –6.304) 0.0826 2.888 0.0892 0.1928 0.865
Therapy Op+PORCT Other 26 (41.9) 0.6672 2.488 (0.910 –6.803) 0.0759 3.064 0.0800 0.1818 0.911
A1C Non-carrier Carrier 29 (44.6) 0.3678 0.415 (0.190 –0.903) 0.0266 4.896 0.0269 0.0579 -0.880
B12C Non-carrier Carrier 28 (62.2) 0.0075 4.403 (2.049 –9.463) 0.0001 15.488 0.0001 0.0040 1.482
A11 Non-carrier Carrier 6 (60.0) 0.1416 5.295 (1.870 –14.99) 0.0017 8.076 0.0045 0.0050 1.667
B51 Non-carrier Carrier 9 (64.3) 0.0220 4.522 (1.939 –10.54) 0.0005 10.175 0.0014 0.0050 1.509
DR15 Non-carrier Carrier 13 (52.0) 0.2515 2.719 (1.310 –5.642) 0.0072 6.867 0.0088 0.0340 1.000
DQB1 hom Non-carrier Carrier 9 (50.0) 0.5926 3.621 (1.189 –11.03) 0.0235 4.350 0.0037 0.0230 1.287
A3-B18 Non-carrier Carrier 4 (57.1) 0.7324 2.387 (0.697 –8.177) 0.1662 1.693 0.1931 0.1648 0.870
A1-B8 Non-carrier Carrier 7 (53.8) 0.8396 3.548 (1.347 –9.345) 0.0104 5.689 0.0171 0.0400 1.266

Shown are the number of events out of 44 events associated with each Pi in total and the percentage of patients sharing the characteristic experiencing an event, the p-value from univariate analysis, the hazard ratio (HR), and two-sided 95% confidence interval (95% CI) accompanied by the respective p-value as well as the loss in χ2 and the associated p-value for omitting the covariate (for the deleted term), the p-value from bootstrapping applying 1,000 iterations (29), and the natural logarithm (ln) of the HR used for summarizing the individual patient’s risk according to his individual characteristics. Significant p-values are shown in bold.