Milea Timbergen

197 square test. Taken together these findings indicated an important association between tumour size and CTNNB1 mutation and led to the development of a second multivariable model (II) including tumour size as a continuous, log transformed variable. This model led to non-significant results for the association between CTNNB1 mutation and recurrence (p = 0.082). The results of both multivariable analyses are displayed in Table 4. Table 4. Multivariable Cox proportional hazards regression model for recurrence Multivariable analysis I a Multivariable analysis II a with additional correction for tumour size Variable n HR (95% CI) p-value n HR (95% CI) p-value Sex 0.434 0.482 Female 247 1.00 (reference) - 242 1.00 (reference) Men 82 1.23 (0.73-2.06) 0.434 80 1.21 (0.71-2.04) 0.482 Age 288 0.99 (0.98-1.01) 0.272 283 0.99 (0.98-1.01) 0.405 Mutation type 0.011 0.082 S45F 66 1.00 (reference) - 61 1.00 (reference) - T41A 154 0.70 (0.42-1.17) 0.178 153 0.83 (0.48-1.42) 0.490 S45P 24 0.32 (0.11-0.97) 0.043 24 0.37 (0.12-1.14) 0.084 WT 85 0.34 (0.17-0.69) 0.003 84 0.44 (0.21-0.92) 0.029 Tumour site <0.001 <0.001 Trunk/back 194 1.00 (reference) - 192 1.00 (reference) - Head/neck 9 1.67 (0.33-8.45) 0.534 8 1.60 (0.31-8.34) 0.581 Extremity 77 4.09 (2.11-7.92) <0.001 77 4.15 (2.14-8.05) <0.001 EA NS 49 2.42 (1.18-4.95) 0.016 45 2.04 (0.97-4.31) 0.061 Tumour size in mm b - - - 283 1.53 (1.03-2.280) 0.034 HR, hazard ratio; CI, confidence interval; WT, wild-type; EA NS, extra-abdominal not specified a Cohort was used as a stratification variable due to a violation of the proportional hazards assumption(p = 0.019) b Used as a log transformed value To examine the impact of tumour size on recurrence, we tested whether there was an effect modification of tumour size and mutation type and found a significant effect modification (p interaction  = 0.09) between those variables, adjusting for mutation, tumour size, tumour site, sex and age. To account for potential effect modification, we used the median tumour size (55 mm) as a cut-off value to perform a subgroup analysis of multivariable analysis (Supplemental Table 6). Supplemental Figure 2 presents the combined estimated effect (in terms of HR) of tumour size and CTNNB1 mutation for the multivariable model for all patients with the interaction between tumour size (log- transformed) and CTNNB1 mutation. 7