1. HIV i-Base: SMART Study & CROI Feedback UK-CAB - Feb 2006 UK-CAB 24 February 2006 CROI Feedback: SMART Study Simon Collins

2. HIV i-Base: SMART Study & CROI Feedback UK-CAB - Feb 2006 Background: Do you need to be on continuous treatment? can you start and stop based on CD4 counts that are generally known to be safe or low risk? Advantages: less drug use, less resistance (adherence), cost etc Disadvantages: risk, more infectious, more resistance (stopping), inflammatory responses

3. HIV i-Base: SMART Study & CROI Feedback UK-CAB - Feb 2006 SMART: Study design [Treatment interruptions] Drug Conservation (DC) [Stop or defer ART until CD4+ 350] [Continuous Treatment] Virologic Suppression (VS) [Use of ART to maintain viral load as low as possible throughout follow-up] CD4+ cell count >350 cells/mm 3 n = 3000 Plan: 910 primary endpoints, 8 years average follow-up Findings (11 Jan 06): 164 primary endpoints, 14 months average follow- up, 2% lost to follow-up

4. HIV i-Base: SMART Study & CROI Feedback UK-CAB - Feb 2006 Primary Endpoint HIV clinical disease progression or death Other Key Endpoints Death Serious HIV progression events Severe complications: cardiovascular, renal and hepatic Sub studies: bpdy shape, QoL etc

5. HIV i-Base: SMART Study & CROI Feedback UK-CAB - Feb 2006 Baseline Large, diverse study: Enrollment: 5473/6000 target 33 Countries, 318 Sites 57% US, 26% Europe, 10% L. America, 1% Africa 27% women, 30% Black, Age 46 Wide treatment experience: Baseline CD4 ~ 600; CD4 nadir ~250 cells/uL 70% <400 copies/mL; <5% patients were treatment naive ~25% prior AIDS Median 6 year prior ARVs ~ 6000 pyfu ~2%LTFU

6. HIV i-Base: SMART Study & CROI Feedback UK-CAB - Feb 2006 Baseline Follow-up and enrollment Enrollment started in US in Jan02 ~150-250 pts enrolled every 3 months to Sep 2004 - all mainly US So some pts followed for over 4 years Last 3000 pts all enrolled Oct04-Dec05 - especially new countries Wide baseline and nadir CD4 count: Baseline CD4 30% >750 cells/mm3 CD4 nadir - wide distribution: ~ 25% < 150 (inc, 10% <50 c/uL) ~ 10% 150-200 c/uL ~ 25% 200-300 c/uL etc

7. HIV i-Base: SMART Study & CROI Feedback UK-CAB - Feb 2006 % Patients on ART at Each Month of Follow- up by Treatment Group Percent Months from randomization Number of patients VS Group DC Group % of Follow-Up Time on ART VS arm: 93% DC arm: 33% Ie - A concern if these two lines become too close. At this distance, so long as it is maintain, ther trial remained ok for it’s main question VS 2308 1167 628 335 DC 2328 1188 613 337

8. HIV i-Base: SMART Study & CROI Feedback UK-CAB - Feb 2006 HIV Disease Progression or Death Logrank = 31.1 p < 0.0001 DC 2720 1170 589 322 34 VS 27521167625 334 41 Months from randomization DC Group VS Group Percent with Event 048121620242832364044 0 5 10 15 20

9. HIV i-Base: SMART Study & CROI Feedback UK-CAB - Feb 2006 Primary Endpoint Clinical Disease Progression or Death DC GroupVS Group RR (DC/VS) P-value N Rate* (95% CI) 117 3.7 47 1.5 2.5 (1.8, 3.6) <0.0001 * Per 100 person-years El-Sadr W, et al. 13th CROI, Denver, CO, February 5-8, 2006. Abst. 106LB

10. HIV i-Base: SMART Study & CROI Feedback UK-CAB - Feb 2006 Relative Risk (95% CI) Relative Risk of Primary Endpoint by Follow-up Interval 0-11-22-33-4 Year of Follow-up DC5637159 VS1911116 No. Events El-Sadr W, et al. 13th CROI, Denver, CO, February 5-8, 2006. Abst. 106LB

11. HIV i-Base: SMART Study & CROI Feedback UK-CAB - Feb 2006 Primary Endpoint and Components Favors VS ► ► Favors DC No. of Patients with Events Endpoints Relative Risk (95% CI) Progression of Disease or Death164 Death84 Serious Progression 21 Non Serious Progression 72 Serious Progression of Disease or Death101 > El-Sadr W, et al. 13th CROI, Denver, CO, February 5-8, 2006. Abst. 106LB

12. HIV i-Base: SMART Study & CROI Feedback UK-CAB - Feb 2006 HIV Progression or Death by Sex and Race No. of Patients with Events Subgroups All Patients164 Sex Female46 Race Black71 Non Black93 Favors VS ► ► Favors DC Relative Risk (95% CI) Male118 El-Sadr W, et al. 13th CROI, Denver, CO, February 5-8, 2006. Abst. 106LB

13. HIV i-Base: SMART Study & CROI Feedback UK-CAB - Feb 2006 Progression or Death By Baseline CD4 No. of Patients with Events Subgroups All Patients164 Baseline CD4 (cells/mm 3 ) 350 - 44944 450 - 54936 550 - 64924 ≥ 65060 Favors VS ► ► Favors DC Relative Risk (95% CI) 2.5 1.5 4.3 3.1 2.9 El-Sadr W, et al. 13th CROI, Denver, CO, February 5-8, 2006. Abst. 106LB

14. HIV i-Base: SMART Study & CROI Feedback UK-CAB - Feb 2006 Progression of Disease or Death by Nadir CD4 No. of Patients with Events Subgroups All Patients164 Nadir CD4 (cells/mm 3 ) 100 – 19935 200 – 29939 300 – 39940 ≥ 40022 50 – 9910 Favors VS ► ► Favors DC Relative Risk (95% CI) < 5018 El-Sadr W, et al. 13th CROI, Denver, CO, February 5-8, 2006. Abst. 106LB

15. HIV i-Base: SMART Study & CROI Feedback UK-CAB - Feb 2006 Progression or death by baseline viral load No. of Patients with Events Subgroups HIV RNA (copies/ mL) - (pts on ART at baseline) ≤ 40087 > 40040 Favors VS ► ► Favors DC Relative Risk (95% CI) 2.5 3.8 1.1 All Patients This showed that the relative risk was almost 4x higher for people with undetectable viral load at baseline

16. HIV i-Base: SMART Study & CROI Feedback UK-CAB - Feb 2006 Causes of Death There were 29 deaths in the continuous treatment arm vs 47 in DC arm. The majority were related to non-AIDS events (ie approx 11 in each arm to heart disease, and similar number to violent/accident.) Only ~5% were defined as ‘AIDS-related’ Cancer was related to ~15% of deaths in each arm A larger number of ‘unknown’ reasons for death occurred in the DC arm (approx 20% vs 7%) Investigators said that primary outcome of the analysis was not changed, even when ‘violent and unknown deaths were taken out of the analysis

17. HIV i-Base: SMART Study & CROI Feedback UK-CAB - Feb 2006 Percent of Follow-up Time Below CD4+ Cell Count Thresholds by Treatment Group % of follow-up time VS GroupDC Group 31.7% 7.2% 8.2% 1.7% 3.1% 0.8% El-Sadr W, et al. 13th CROI, Denver, CO, February 5-8, 2006. Abst. 106LB

18. HIV i-Base: SMART Study & CROI Feedback UK-CAB - Feb 2006 Summary The DC strategy, compared to the VS strategy, is associated with increased risk of: –HIV disease progression or death –Death –Serious HIV disease progression –Severe complications (cardiac, renal or hepatic) Risk of disease progression or death in DC versus VS group –Did not differ by nadir CD4+ cell count –Was three-fold higher for patients on ART with baseline HIV RNA 400 copies/ml 3.2 [1.9,5.6] vs 1.1 [0.6,2.0] For other subgroups examined, risk was always greater in the DC group than the VS group

19. HIV i-Base: SMART Study & CROI Feedback UK-CAB - Feb 2006 Conclusion Episodic use of ART based on CD4+ cell count levels as per the SMART study design is inferior to continuous ART for the management of treatment- experienced patients. Thus, this strategy should not be recommended.

20. HIV i-Base: SMART Study & CROI Feedback UK-CAB - Feb 2006 Discussion.1 Other research: SMART was one of six studies to get an oral presentation - all slides are online One study with similar CD4 criteria to restart at ~250 CD4 also stopped early Other studies, restarting at <350 CD4 didn’t find the same risk (ie Stacatto) - but this was a much smaller study In SMART was that ‘proximal’ CD4 count was not predictive of risk - death and serious events occurred at all CD4 counts including > 600 CD4 etc - though the lower the CD4 count the higher the relative risk.

21. HIV i-Base: SMART Study & CROI Feedback UK-CAB - Feb 2006 Discussion.2 Two arms appear to show differences after 4-6 months or treatment interruption. Median time off treatment was 18 months in SMART - question of whether shorter fixed periods off-treatment would narrow the difference compared to continuous treatment SMART designed when treatment was more difficult - dosing, pill count, tolerability now are improved for more patients now Some individuals still have great difficulty with HAART Always an individual choice - and data needed to inform this risk when balancing risk vs benefit

22. HIV i-Base: SMART Study & CROI Feedback UK-CAB - Feb 2006 Discussion.3 SMART provided clear answer to its study question Follow-up and further analysis are ongoing Quality of Life sub-study should be presented in Toronto in July 2006 If SMART showed treatment to be ‘safer’ than anticipated when the study was first designed, does this reopen the ethics of a ‘when to start’ trial that randomises people to start at higher CD4 counts.

23. HIV i-Base: SMART Study & CROI Feedback UK-CAB - Feb 2006 Additional slides

24. HIV i-Base: SMART Study & CROI Feedback UK-CAB - Feb 2006 SMART 1 Staccato 2 Trivacan 3 CD4 Count Start Stop 250 350350 Disease progression per 100 patient years STI CT 3.7 1.5 0.2 0.4 17.6 6.7 Therapy Interruption: CD4 count guided 1SMART: 13 th CROI 2006 Abstract 106LB 2Staccato: 13 th CROI 2006 Abstract 102 3Trivacan: 13 th CROI 2006 Abstract 105LB