Figure 3. Prognostic Analyses for Distant Recurrence.

Weighted univariate and multivariable Cox proportional hazards regression analyses are shown for distant recurrence-free interval. The hazard ratio (HR) compares the mutated or amplified status vs the wild-type or nonamplified status of the gene (or amplicon), respectively. Number of alterations was evaluated as a continuous variable per additional driver alteration. Number of events indicates the observed number of distant recurrences in the analysis cohort. The univariate Cox proportional hazards regression model was adjusted for treatment arm, and the multivariable model was adjusted for treatment arm, age, tumor size, nodal status, and grade as described in the Methods section. Focal gene amplifications on chromosomes 20q13, 11q13, and 8p11 are described in the Results section.