Differences between treatment groups with 95% CIs at each time-point were summarized using Student’s t-test. Overall differences between the two treatment groups were evaluated using a linear mixed model (PROC MIXED, SAS 9.1, SAS Institute Inc., Cary, NC). As a sensitivity analysis we also conducted a regular on-treatment analysis censoring patients when they stopped any of the assigned drugs, and sensitivity analyses censoring patients from the date they received systemic steroids or bisphosphonates. Baseline age, body mass index (BMI), CD4 cell count, current smoking, gender, baseline BMD and treatment arm were evaluated as predictors of BMD loss during the first

24 weeks using univariate and multivariate linear regression. We conducted separate analyses for hip and spine BMD. Baseline age, gender, BMI, CD4 cell count and current smoking were also evaluated for associations with KU-60019 mw baseline spine and hip BMD. We used spss software (Norusis; SPSS Inc., Chicago, IL) and sas 9.1 (SAS Institute Inc., Cary, NC) for statistical analyses. Of the 104 randomized patients in the SPAR study, 63 participated in the BMD substudy. Fifty-nine patients (29 in the NRTI-sparing group and 30 in the PI-sparing group)

had at least one follow-up BMD measurement and were included in the present study. The majority were men (90%) and Caucasian (93%). Baseline median CD4 count was 190 cells/μL, median viral load was 197 000 HIV-1 RNA copies/ml, median age was 42 years and median BMI was 21.9. At baseline, median spine BMD was 1.09 g/cm2 and median hip BMD was 0.91 g/cm2. Thirty-three patients (55.9%) had a low BMD (T-score buy BEZ235 <−1.0) and of these eight had DEXA-defined osteoporosis. Baseline characteristics with IQRs according to treatment group are displayed in Table 1. A large proportion of patients switched part of the randomized treatment during the study period, but 44 (74.6%), 37 (67.3%), 28 (54.9%) and 22 patients (48.9%) were still on randomized triclocarban treatment at weeks 24, 48, 96 and 144, respectively. There were no differences between time to discontinuation of randomized

treatment between the two groups (P=0.76). A number of patients switched to a new drug within the same drug class and 51 (86.4%), 46 (83.6%), 38 (74.5) and 31 (68.9%) were still in the assigned drug class-sparing arm at weeks 24, 48, 96 and 144, respectively. Significantly more patients in the NRTI-sparing arm changed drug class (P=0.005). Four patients received systemic steroids during the study period and one patient in the NRTI-sparing arm received bisphosphonate treatment from week 72. The changes compared with baseline in spine and hip BMD in ITT analyses are displayed in Figures 1 and 2. Spine BMD declined in both arms until week 24, and thereafter BMD stabilized. Femoral neck BMD declined in both arms until week 48 and stabilized thereafter. There was no significant difference between the two treatment arms (P=0.7 for spine and P=0.3 for femoral neck).