Statistical analysisThe SPSS versio

Statistical analysisThe SPSS version 26.0 (SPSS Inc., Chicago, IL, USA) software were used for statistical analysis. Spearman’s rank correlation (rho) [19] coefficient was calculated and used to assess the correlation intra SUV parameters ( SUVmean, SUVmax, SUVtotal, ΔSUVs ) and between SUV parameters with xerostomia. The P-value was calculated by Student's t-distribution to test the significance. In this study, acute and chronic xerostomia were analyzed separately.The SUV parameters with significant correlation (P < 0.05) to patient xerostomia were selected for the receiver operating characteristic (ROC) analysis [20]. The ‘True Positive’ was defined as ‘the patient who has a selected SUV parameter ≤ a given value will develop the grade 2 or higher xerostomia’, and the ‘True Negative’ was defined as ‘patient with a selected SUV parameter > a given value will not develop the grade 2 or higher xerostomia’. Meanwhile, the ‘False Positive’ and ‘False Negative’ were then defined accordingly. The sensitivity and specificity of each of the SUV parameters was determined by searching the optimal cut-off to maximize the Yoden’s index (i.e. sensitivity + specificity - 1). The area under the curve (AUC) was calculated and the corresponding P-value was obtained using Mann-Whitney U test (Null Hypothesis: AUC = 0.5). All statistical tests were calculated using IBM SPSS Statistic version 23 (IBM, Armonk, NY) and a two-sided P-value < 0.05 was considered statistically significant.