151751-Najiba-Chargi

268 CHAPTER 14 , = , × (1 + ) = ( − + × 6 ) × ( ℎ 70 ) 0.75 = × (1 + × ( − )) (S3) Where CL bound represents the total clearance of protein-bound cisplatin, CL non-renal represents the non-renal clearance, CL renal represents the renal clearance, and GFR represents the glo - merular filtration rate as calculated by the Cockcroft-Gault formula or the Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) cystatin C formula, which is divided by 6 L/ hour as ‘normal’ glomerular filtration rate, the clearance is standardized for a 70 kg person 25 . In case body composition was evaluated in combination with GFR the part in equation S3 which standardizes for a 70 kg person, was replaced by the body composition description part of equation 5. Albumin was evaluated on CL of free cisplatin and V of protein-bound cisplatin, according to equation S4: (3) (4) (5) , = , × (1 + ) = ( − + × 6 ) × ( ℎ 70 ) 0.75 = × (1 + × ( − )) (S4) Where θ i represents the parameter estimate for individual i , θ pop represents the typical param- eter estimate for the population with a median albumin, and θ albumin represents the fractional increase per unit albumin. MODEL EVALUATION The fit of the baseline model was evaluated using goodness-of-fit (GOF) plots. Addition of the covariates GFR, albumin and body composition descriptors was evaluated by GOF plots, a drop in the objective function value (OFV; minus twice the log likelihood), successful minimi - zation, parameter precision ($COVARIANCE option of NONMEM) and a drop in inter-individual variability (IIV). SOFTWARE Nonlinear mixed-effects modeling was performed using NONMEM (version 7.3, ICON Develop - ment Solutions, Ellicott City, USA) and Perl-speaks-NONMEM (PsN, version 4.7.0) 23 . In order to obtain parameter estimates the first-order conditional estimation with interaction (FOCE-I) method was used. Model management was performed using Pirana (version 2.9.9) 26 . R (version 3.6.3) was used for data management and graphical diagnostics 27 .