For DXA, both Lunar (General Electric Company, Brussels, Belgium) and Hologic (Hologic Inc., Bedford, MA, USA) equipment was utilized. Calibration procedures Caspase activity assay and quality control checks were performed daily. A special phantom was made available to each of the sites before study initiation to calibrate the DXA equipment for the assessment of lean and fat mass, in order to allow accurate centralized analysis of the data. Given that CT and DXA scanning had not been part of the SSAR 2004/0002 protocol, participants in that protocol were excluded from all body composition analyses. Glomerular filtration rate (GFR) was estimated using various equations: Cockcroft and Gault (C&G) [26], Modification of Diet

in Renal Disease (MDRD) formulas [27], the Cystatin C (CysC) equation [28] and the Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) estimate [29]. Patients from the SSAR 2004/0002 study

were excluded from the analyses of the estimated GFR (eGFR) estimated using the CysC and MDRD-6 equations, as cystatin C and urea levels were not available. Plasma HIV-1 RNA (lower limit of quantification Cell Cycle inhibitor 50 HIV-1 RNA copies/mL), CD4 T-cell count, routine haematology and chemistry were monitored at all visits. The latest version (version 1, December 2004) Adult Clinical Trials Group (ACTG) table for grading the severity of adverse events was used for the reporting of adverse events. The primary objective of this trial was to demonstrate noninferiority of an SQV/r-based regimen compared with an ATV/r-based regimen with respect to mean changes in TC. The sample size was calculated using results from the AI424-008 trial in which mean TC increased from 169 to 177 mg/dL Florfenicol (with the standard deviation of the mean difference being 37.1) after 48 weeks of treatment [4]. Noninferiority is demonstrated when the upper limit of the 95% confidence interval of the difference between study arms is <10%. Setting alpha at 5%, a sample size of 60 subjects per study arm results in a power of more than 80% to

demonstrate noninferiority, assuming a true difference in TC between arms of 0%. The patient population used in the analyses included all randomized patients who received at least one dose of study medication. All analyses were performed on the intent-to-treat (ITT) population. For the ITT analysis, all missing values of the outcome measurements (other than the week 12 lipids in the SSAR 2004/0002 study) were imputed using a last observation carried forward (LOCF) approach. In addition, an on-treatment (OT) analysis was performed for blood lipids, glucose metabolism, body composition, virological and immunological responses. In the OT analysis, data from patients who prematurely discontinued study medication were censored at the time of study drug discontinuation, and no LOCF imputation was used. Changes in metabolic and renal parameters were assessed using linear mixed models incorporating repeated measurements.