1. Switch to RAL-containing regimen  Canadian Study  CHEER  Montreal Study  EASIER  SWITCHMRK  SPIRAL

2.  Design  Endpoints –Primary: non inferiority in the proportion of patients with treatment failure at W48* (non completer = failure, intent-to-treat analysis), lower limit of the 95% CI for the difference = - 12.5%, 80% power ; * events occurring in the 2 weeks after W48 were included in the analysis –Secondary: virologic failure (confirmed HIV-1 RNA > 50 c/mL), CD4, fasting lipids, adverse events Switch to RAL 400 mg bid + continue other ARVs Continue PI/r + other ARVs * Randomisation was stratified by current use of lipid-lowering therapy ** Median time with virologic suppression was > 6 years Martinez E, AIDS 2010;24:1697-1707 SPIRAL SPIRAL Study: Switch PI/r to RAL Randomisation* 1 : 1 Open-label HIV+ ≥ 18 years On 2 ARV + PI/r HIV RNA 6 months** Raltegravir-naïve N = 140 N = 142 W48

3.  Treatment failure (intention-to-treat) –Progression to AIDS –Death –Virologic failure –Discontinuation of study medication –Consent withdrawn, lost to follow-up  Virologic failure (on-treatment) –Progression to AIDS –Death –Virologic failure during treatment –Patients who withdrew consent, were lost to follow-up, switched or stopped study medication were censored  Changes in plasma lipids –Analysis by intention-to-treat Martinez E, AIDS 2010;24:1697-1707 SPIRAL SPIRAL Study: Switch PI/r to RAL

4. RALPI/r Patients included in the efficacy analysis139134 Female19%28% ARV backbone: TDF + FTC/3TC ; ABC + 3TC/FTC58% ; 19%55% ; 20% PI/r at entry: LPV/r ; ATV/r ; FPV/r43% ; 37% ; 11%45% ; 33% ; 13% Patients on their first ARV regimen12%10% Patients with previous suboptimal ARV therapy*41%35% Patients with previous virologic failure40%36% Patients with previous suboptimal ARV therapy or virologic failure 57%49% CD4 cell count (/mm 3 ), median (IQR)529 (377-780)509 (369-726) Lipid-lowering therapy at entry19%21% Discontinuation before W48, n (%)13 (9%)14 (10%) For adverse event33 For virologic failure22 Baseline characteristics and patient disposition * 1 or 2 NRTI exclusively SPIRAL Study: Switch PI/r to RAL Martinez E, AIDS 2010;24:1697-1707 SPIRAL

5. Results: efficacy analyses Martinez E, AIDS 2010;24:1697-1707 SPIRAL Study: Switch PI/r to RAL SPIRAL Absence of treatment failure 95% CI for the ≠ = -5.2 ; 10.6 Primary efficacy endpoint Absence of virologic failure - 5.9 ; 17.6- 11.2 ; 10.9- 3.5 ; 7.5- 3.9 ; 13.9- 9.3 ; 7.6 Prior virologic failure or suboptimal therapy % 89.2 88.6 90 96.997.2 96.4 86.6 83.1 89.9 95.1 93.1 96.9 0 20 40 60 80 100 139 N= 1347965606972585664 All patients YesNo Prior virologic failure or suboptimal therapy All patients YesNo 128122 RALPI/r

6. SPIRAL Study: Switch PI/r to RAL 134131124121116 139138132130124 Log rank p = 0.4775 Weeks 0.6 4812162024283236404448 0.7 0.8 0.9 1 0 134122119 116 139128126125124 Log rank p = 0.4602 0.6 4812162024283236404448 0.7 0.8 0.9 1 0 Weeks Martinez E, AIDS 2010;24:1697-1707 SPIRAL Time to treatment failure by treatment group Time to virologic failure by treatment group RALPI/r

7.  Virologic failure –First of 2 consecutive measurements of HIV RNA ≥ 50 c/mL separated by a minimum of 2 weeks –VF at W48 : 4 (2.9%) in the RAL arm vs 6 (4.4%) in the PI/r arm No difference in patients with and without VF regarding –Demographics, HIV parameters, N(t)RTI backbone, PI, duration of viral suppression at entry Median time with virologic suppression prior to inclusion : 62.85 months in patients without previous VF vs 65 months in patients with previous VF –74/250 patients (50%) had previous VF with prior genotypic resistance tests GSS for backbone N(t)RTI was < 1 in 15/38 (39%) in the RAL group and in 9/36 (25%) in the PI/r group : VF developed in 0/15 vs 2/9 (22%), respectively (p=0.13) Moreover 0/11 subjects with GSS ≤ 0.5 backbone activity developed VF in the RAL arm SPIRAL Study: Switch PI/r to RAL Blanco JL. Antiviral Therapy 2015, epub ahead of print SPIRAL

8. Percentage changes in fasting lipid concentrations from baseline to W48  At entry, median total cholesterol (TC) was 198 mg/dL, 15% of the patients had TC > 240 mg/dL, 12% LDL-cholesterol > 160 mg/dL, 40% triglycerides > 200 mg/dL SPIRAL Study: Switch PI/r to RAL Martinez E, AIDS 2010;24:1697-1707 SPIRAL p < 0.0001 p < 0.001 p < 0.0001 p < 0.05 -22.09 -11.18 -6.49 -3.17 -4.85 4.72 1.82 2.96 5.84 -1.28 -20 -15 -10 -5 0 5 10 TriglyceridesTotal cholesterol LDL cholesterol HDL cholesterol Total to HDL cholesterol ratio % -25 RALPI/r

9. Martinez E, AIDS 2010;24:1697-1707 SPIRAL Study: Switch PI/r to RAL  At W48, significantly less patients had triglycerides > 200 mg/dL or total cholesterol > 240 mg/dL in the RAL group compared to the PI/r group: 14.6% vs 28.9% and 3.7% vs 17.2%, respectively  Differences in total cholesterol and triglycerides changes in patients assigned to RAL were significant when switching from LPV/r but not from ATV/r  There were no difference in the overall incidence of adverse events in the 2 groups  The incidences of serious adverse events and events leading to drug discontinuation were similarly low in both groups SPIRAL

10. Martinez E, AIDS 2010;24:1697-1707 SPIRAL Study: Switch PI/r to RAL vs continuation of PI/r  Conclusions –In HIV-infected adults with sustained plasma HIV-1 RNA < 50 c/mL on PI/r-containing ARV therapy, switching from the PI/r component to raltegravir results In non inferior efficacy And a better lipid profile SPIRAL

11. Cardiovascular biomarkers: median (95% CI) difference of percent change from baseline to W48, RAL (N = 119) minus PI/r (N = 114) SPIRAL Study: Switch PI/r to RAL SPIRAL Martinez E, AIDS 2012;26:2315-26 Markers of inflammation Endothelial dysfunction Insulin resistance Insulin resistance Hyper- coagulability Hyper- coagulability MCP-1 hsCRP OPG IL-6 IL-10 TNF-  ICAM-1 VCAM-1 E-selectin P-selectin Adiponectin Insulin D-dimer -70 -60 -50 -40 -30 -20 -10 0 20 %

12. Correlations between ∆ biomarkers and ∆ lipids No correlations between ∆ OPG, ∆ IL-6, ∆ IL-10, ∆ TNF-alpha, ∆ ICAM-1, ∆ VCAM-1, ∆ E-selectin, ∆ P-selectin, ∆ Adiponectin, ∆ D-dimer and any changes in lipids  Conclusion –Switching from PI/r to RAL led not only to significant changes in plasma lipids but also to significant changes in several cardiovascular biomarkers associated with inflammation, insulin resistance and hypercoagulability –There were few and weak significant correlations between changes in lipids and changes in biomarkers suggesting that decreases in biomarkers were rather independent of lipid changes SPIRAL Study: Switch PI/r to RAL SPIRAL ∆ Triglycerides∆ Total cholesterol∆ LDL cholesterol∆ HDL cholesterol ∆ hsCRP --0.2415 (p=0.0016)- ∆ MCP-1 -0.1608 (p=0.032) 0.1807 (p=0.0202) ∆ Insulin 0.2842 (p=0.0001)- 0.2125 (p=0.004)-- Martinez E, AIDS 2012;26:2315-26 Data expressed Spearman’s rho (p)

13. SPIRAL Study: Switch PI/r to RAL SPIRAL-LIP substudy (body composition)  Proceduresat baseline and W48 –Whole body, lumbar and hip DEXA scans –CT scan of abdomen (single cut 5 mm thick, at L4) –Standardized protocol performed by a single radiologist unaware of patient’s treatment  Endpoints –Primary: change in visceral adipose tissue (VAT) area (cm 2 ) –Secondary: changes in limb fat, trunk fat, total fat, total adipose tissue area, subcutaneous adipose tissue (SAT) area, SAT/VAT ratios, changes in bone mineral density and T scores in total body, spine (L1-L4) and hip (femoral neck and total hip) Curran A, AIDS 2012;26:475-81 SPIRAL

14. SPIRAL Study: Switch PI/r to RAL SPIRAL-LIP substudy (body composition) RALPI/r Patients included in the SPIRAL-LIP substudy, n3935 Female20%31% ARV backbone: TDF ; ABC ; ZDV62% ; 33% ; 3%69% ; 23% ; 17% PI/r at entry: LPV/r ; ATV/r ; FPV/r43% ; 51% ; 3%46% ; 34% ; 0% Time on previous PI/r, months (median)35.730 HCV co-infection28%17% Weight, kg (median)7068.5 BMI, kg/m 2 (median)23.423.6 Baseline characteristics of the 74 participants SPIRAL Curran A, AIDS 2012;26:475-81

15. SPIRAL Study: Switch PI/r to RAL SPIRAL-LIP substudy (body composition) RALPI/r Difference* (IQR) PI/r vs RAL CT scan VAT (cm 2 ) 10.920,7 (p=0.002)11.4 (- 9.5 ; 38.2) VAT (%) 12.811.91.6 (- 1.4 ; 1.9) TAT (cm 2 ) 3.721.4 (p=0.013)10.9 (- 22.8 ; 44.8) SAT (cm 2 ) - 3.25.12.1 (-18.0 ; 18.5) SAT (%) - 1.93.6- 1.1 (- 4.1 ; 1.9) SAT/VAT - 0.25- 0.11- 0.15 (-0.66 ; 0.15) DEXA scan Limb fat (kg) 3217185 (- 383 ; 795) Trunk fat (kg) - 28382323 (- 988 ; 1768) Total fat (kg) - 389307293 (- 1304 ; 2816) Limb/trunk ratio 0- 0.02- 0.01 (- 0.05 ; 0.05) Body fat distribution (median change from baseline to week 48) * p not significant for all measures SPIRAL Curran A, AIDS 2012;26:475-81

16. SPIRAL Study: Switch PI/r to RAL SPIRAL-LIP substudy (body composition) Curran A, AIDS 2012;64:475-81 SPIRAL 382 RALPI/r 21.4 23.73.6 11.9 12.8 10.9 20.7 5.1 -3.2 -1.9 -0.11 -0.25 171 32 - 28 307 -359 -0.02 0 cm 2 % TATSATVAT SATSAT/VAT % Limb fat Trunk fat Total fat Limb/trunk ratio g Median values All differences PI/r vs RAL not significant 0 + - 0 + -

17. RALPI/r Difference (IQR) PI/r vs RAL p DEXA scan Total BMD (g/cm 2 ) 0.01 (p=0.002)00.01 (- 0.01 ; 0.02)0.079 Femoral neck BMD (g/cm 2 ) 0- 0.010.01 (- 0.02 ; 0.02)0.032 Femoral neck T score 0.04- 0.100.01 (- 0.18 ; 0.18)0.016 Total hip BMD (g/cm 2 ) 0.01 (p=0.015)00.01 (- 0.01 ; 0.02)ns Total hip T score 0.12 (p=0.004)0.010.11 (- 0.05 ; 0.20)ns L1-L4 BMD (g/cm 2 ) 00.020 (- 0.02 ; 0.04)ns L1-L4 T score 0.030.100.09 (- 0.11 ; 0.31)ns Bone composition (median change from baseline to week 48) BMD: bone mineral density  No significant differences in BMD or T scores in either group even when controlling for TDF use SPIRAL Study: Switch PI/r to RAL SPIRAL-LIP substudy (body composition) SPIRAL Curran A, AIDS 2012;26:475-81

18. SPIRAL Study: Switch PI/r to RAL SPIRAL-LIP substudy (body composition)  Conclusion –Although there were no significant changes in body fat between groups, maintaining a PI/r-based regimen was associated with a significant increase in VAT and TAT –Switching to RAL led to a significant increase in femoral neck BMD when comparing between groups SPIRAL Curran A, AIDS 2012;26:475-81

19. SPIRAL Study: Comparison of ABC/3TC vs TDF/FTC Martinez E, AIDS Res Hum Retroviruses. 2013 Feb;29(2):235-41  In the RAL group, decrease in triglycerides and increase in HDL cholesterol at W48 tended to be more pronounced with ABC/3TC than with TDF/FTC  Differences in total-to-HDL cholesterol ratio between both combinations of NRTIs tended to be higher in the RAL group although differences at 48 weeks were not significant Switch to RALContinuation of PI/r NRTI ABC/3TC (N = 27) TDF/FTC (N = 73) ABC/3TC (N = 27) TDF/FTC (N = 70) Treatment failure* 3 (11.1%)8 (11%) 4 (14.8%) 12 (17.1%) Virologic failure (confirmed HIV-1 RNA > 50 c/mL) 1 (3.7%) 3 (4.1%) 2 (7.4%) 4 (5.7%) Discontinuation due to AE 000 4 (GFR decrease, N = 3), BMD decrease, N = 1) * Treatment failure: virologic failure, discontinuation of NRTI due to adverse event, consent withdrawal, loss to follow-up SPIRAL

20. SPIRAL-MET substudy: LDL subclasses and lipoprotein- phospholipase A2 activity  81 patients, PI/r group (N = 41), RAL group (N = 40)  Baseline and week 48 assessment : –LDL size and phenotype : Phenotype A : LDL size > 26.8 nm with predominance of large buoyant LDL subfractions Phenotype intermediate : LDL size 26.0-26.8 nm Phenotype B : LDL size < 26.0 nm with a predominance of small, dense LDL subfractions –Total lipoprotein-associated phosholipase A2 (Lp-PLA2) –Proproteinconvertasesubtilisin/kexin type 9 (PCSK9) –Standard lipid parameters –Insulin, C-peptide, HOMA index –Cardiovascular risk assessment (Framingham equation) Saumoy M, Atherosclerosis 2012;225:200-7 SPIRAL

21. RAL (n = 40)PI/r (n = 41) Age, years (median)4443 Female15%27% ARV backbone: TDF/FTC ; ABC/3TC43% ; 35%74% PI/r at entry: LPV/r ; ATV/r ; FPV/r43% ; 38% ; 10%37% ; 39% ; 12% Time on previous PI/r, months (median)35.730 Weight, kg (median)7371 BMI, kg/m 2 (median)24.524.4 Cardiovascular risk estimation (Framingham)7.428.27 Lipid lowering therapy17.5%22.8%  Baseline characteristics Lipid parameters were not significantly different between groups, except Apo B, which was lower in the RAL arm (p = 0.035) Saumoy M, Atherosclerosis 2012;225:200-7 SPIRAL-MET substudy SPIRAL

22.  Results – Insulin, waist Significant difference in insulin levels between arms favorable to RAL at W48 (p = 0.020) HOMA index decreased in RAL group (p = 0.032) at W48, remaining unchanged in the PI/r arm At W48, increase in waist circumference (3.95 cm ; p = 0.004) and waist-to-hip ratio (0.01 ; p = 0.022) in the PI/r arm, where as no change in RAL group – No change in number of patients on lipid-lowering therapy – Cardiovascular risk assessment at W48 Increase in the PI/r arm (0.8% ; p = 0.032) No change in the RAL arm No change between arms at W48 Significant increase of systolic (+ 5 mm Hg ; p = 0.016) and diastolic (+ 8,5 mm Hg ; p = 0.005) blood pressure in the RAL arm, no change in the PI/r group Saumoy M, Atherosclerosis 2012;225:200-7 SPIRAL-MET substudy SPIRAL

23. Saumoy M, Atherosclerosis 2012;225:200-7 SPIRAL-MET: median changes in lipid parameters between baseline and W48 according to therapy SPIRAL

24. LDLc phenotype A: large & low density of cholesterol esters (non-atherogenic) LDLc phenotype B: small & high density of cholesterol esters (atherogenic) Saumoy M, Atherosclerosis 2012;225:200-7 100 80 60 40 20 0 AIntermediateB BaselineW48BaselineW48BaselineW48 Group 1 (n = 21)**Group 2 (n = 19)RAL (n = 40) Group 1: LPV/r, FPV/rGroup 2: ATV/r, SQV/r 0.439* 0.088* < 0.001* % LDL-c particles phenotype * Homogeneity marginal test to compare proportions at baseline and W48 in each arm ** Statistically significant difference between LPV and RAL at W48 (Pearson Chi-square test) SPIRAL-MET: median changes in the percentage of LDL-c phenotype in RAL arm and PI arm stratified by PI/r used (group 1 vs group 2) at W48 SPIRAL LDLc phenotype

25. SPIRAL substudy: endothelial function  35 patients, PI/r group (n = 16), RAL group (n = 19)  Endothelial function was evaluated through flow-mediated dilatation (FMD) of the brachial artery at baseline, W24 and W48  Total cholesterol, LDL cholesterol and triglycerides decreased at W16 and W32 in the RAL arm, while no changes were observed in the PI/r arm  Triglyceride levels were significantly lower in the RAL arm than in the PI/r arm at W16, 32 and 48  No significant changes from baseline occurred in FMD at W24 and W48 within or between the RAL and PI/r arms. Adjustment for baseline artery diameter did not have a significant effect on the FMD differences –Median baseline FMD values within normal range ( > 5%) + limited sample size might have precluded detection of any RAL effect or clinically relevant differences SPIRAL Masia M, JAC 2013;68:409-413