151751-Najiba-Chargi

66 CHAPTER 4 Table 6. Spearman correlation coefficients of the different skeletal mass measurements Relation Correlation coefficient p-value LSMI - MSMI 0.291 0.002 LSMI – MCSA 0.335 <0.001 LSMI – MV 0.418 <0.001 LSMI - MT 0.342 <0.001 L3 CSA - MSMI 0.193 <0.001 L3 CSA - MCSA 0.469 <0.001 L3 CSA - MV 0.542 <0.001 L3 CSA - MT 0.439 <0.001 C3 CSA - MSMI 0.349 <0.001 C3 CSA – MCSA 0.573 <0.001 C3 CSA – MV 0.671 <0.001 C3 CSA - MT 0.527 <0.001 L3 CSA – C3 CSA 0.715 <0.001 Correlation between different masseter parameters, lumbar skeletal muscle index, cross-sectional area at level L3 and cross-sectional area at level C3 are shown. UNIVARIATE AND MULTIVARIATE COX REGRESSION ANALYSIS All clinically relevant characteristics or characteristics relevant to masseter muscle parame- ters were tested using Cox univariate regression analysis. In univariate analysis, MCSA, MSMI, L3 CSA, C3 CSA, BSA, CCI were all significantly associated with overall survival (Table 7). For variables that were strongly correlated or dependent on each other (e.g., MSMI, lowMSMI and MCSA) the variable with the lowest p-value was included in the multivariate analysis. As to not exceed the >10 events per variable rule BSA was excluded frommultivariate analysis based on expert opinion. This left LowMSMI-classification, C3 CSA, L3 CSA and CCI as included variables. Low MSMI and CCI score remained as the only independent predictors of overall survival (HR 3.032, p=0.002 and HR 1.338, p<0.001, respectively; Table 8). Table 7. Univariable Cox regression analysis Risk factor HR 95% CI p-value MCSA 1.0 0.99 - 1.0 0 .02 MV 0.95 0.91 – 1.0 0.06 MT 0.5 0.86 – 1.0 0.27 HU tot 1.0 0.99 – 1.0 0.23 Masseter ROI 1.0 0.98 – 1.1 0.55 Low MSMI 2.2 1.2 – 4.3 0.02 CSMI 0.92 0.80 – 1.1 0.28 LSMI 0.97 0.93 – 1.0 0.13