Table 5

Cox proportional hazards regression analysis of significant variables

MarkerHazards ratioSEzP
Model Including All Factors
 Age increase per yr0.3940.289−3.120.0013
Compressed Model
 Age increase per yr0.3850.306−3.120.0018
 Cisplatin resistanceb1.540.2072.080.037
  • a The addition of age to the model reduced the significance of the individual markers. Each additional year of age decreased the risk of death. However, the total overall effect of the three biological markers was statistically significant (P = 0.03; three degrees of freedom).

  • b A variable created as the number of cisplatin markers expressed (range, 0 to 2 for GST-π and P-gp) improved the significance of the total model (P = 0.0001).