151751-Najiba-Chargi

222 CHAPTER 12 chi-square test. The risk parameters were calculated and presented with corresponding 95% confidence intervals (95% CI) and p-values. The predictive value of low SMM on cetuximab DLT was evaluated using univariate and multi- variate logistic regression. A Cox proportional hazard regressionmodel was used for univariate and multivariate analysis of OS and DFS. Covariates used in the multivariate analysis were selected based on clinical significance or statistical significance (p < 0.05) in univariate cox or logistic regression analysis. Statistical significance was evaluated at the 0.05 level using 2-sides tests. OS and DFSwere visualized using KaplanMeier survival curves and number at risk tables. RESULTS STUDY POPULATION Between 2007 and 2018, 110 HNSCC patients were treated with primary or adjuvant cetux - imab and radiation for oropharynx, hypopharynx, or larynx tumor. Of these patients, 100 had pre-treatment imaging of the C3 vertebra which is necessary for the determination of SMM. Additionally, patients receiving cetuximab with palliative intent were excluded. As can be seen in Figure 2, 91 patients were included in the analysis, 28 patients (30.8 %) experienced cetuximab DLT and 63 (69.2%) experienced no cetuximab DLT. Figure 2. Flowchart of patient inclusion Determining the optimal cut-off value for low SMM The cut-off value for low SMM was determined by calculating the log likelihood using a tech - nique previously described by Williams et al. 15 The cut-off value best associated with the pres - ence of cetuximab DLT (lowest Log-Likelihood value) was LSMI ≤ 45.2 cm 2 /m 2 . Using this cut-off value for the study population, 68 (74.7%) were identified with low SMM and 23 (25.3%) were identified without low SMM.