1. The Foundation Role of Immunomodulation Therapy for Long-Term Efficacy Safety, Outcome Measures, and Disability Mitigation Investigations Innovation Clinical Application Program Chairman Bruce A. Cree, MD, PhD, MCR Assistant Professor of Neurology Department of Neurology University of California San Francisco Multiple Sclerosis Center San Francisco, California

2. Overview ► Mechanisms of action of first line therapies ► Outcome measures in clinical trials ► Comparison of landmark trials ► Longitudinal studies: what do they tell us?

3. Goals of Treatment ► Reduce frequency of relapse ► Slow progression of disability ► Reduce MRI activity ► Prevent morbidity from symptoms and provide palliative care ► Maintain adherence ► Provide long-term efficacy and safety

4. Treg Th2/ Th3 MO IL-4 IL-5 IL-6 IL-13 TGF  B Histamine Proteases TNF  NAA, ATP NO O 2 5-HT Mast Cell IL-12 APC Thp CD4 CD40LCD40 IL-4 & IL-10 CD4 APC Thp CD28 B7 Th2/ Th3 B7 CD40 Microglia CD40L CD28 Th1 Th17 B Glutamate  T CD8 MMP- 2/9 VCAM-1ICAM-1 VCAM-1 IFN  TNF IL-17 IL-10 TGF  Ab+C 9neo CD8 Mast Cell  T Granutocyte Complement Monocyte Pl Figure courtesy of Dhib-Jalbut S, 2008 Immunopathogenesis of the MS Lesion IFN  TNF Th17 NO Oi TNFa MMP LFA-1 VLA-4 Th1 Th17 Oligo BBB MCP-1 MIP-1   P-10 RANTES Astrocyte IL-23 Treg CD4+CD25+ Myelin Ag Microbial Ag HLA Virus TCR

5. IFN-  : Activity T H 1+ Resting T cell MMP Activated (+) T cells T H 1+ MMP BBBBloodCNS TNF-α IFN-γ IL-2 TH1TH1 APC IFN- β Myelin protein Antigen T H 1+ Adapted from Yong VW. Neurology. 2002;59:802-808.

6. Glatiramer Acetate: Activity Adapted from Ziemssen T et al. J Neurol Sci. 2005;233:109-112. BBB GA- specific T cell APC GA therapy TH1TH1TH2TH2 APC Microglia MHC CNS Ag TCR Macrophage PeripheryCNS TH2TH2 MHC GA TCR Neuroregeneration Bystander suppression effect Anti-inflammatory cytokines Neurotrophins + + TCR IL-4 IL-10 BDNF

7. Low (0-1 attacks in 2 years) Intermediate (2-4 attacks in 2 years) High (> 5 in 2 years) Weinhenker B et al. Brain. 1989;112:1422 Long-Term Disability Effect of Early Relapses

8. Number of Attacks, 1 st 2 years p <0.001 Interval Between 1 st 2 Attacks p <0.001 DSS at 2 years p < 0.001 DSS at 5 years p < 0.001 Development of Disability Effect of Early Clinical Course Clinical Characteristic Significance* * Controlled for age at onset, remitting at onset, cerebellar, cerebral Weinshenker B et al. Brain. 1991;114 ( Pt 2):1045-56.

9. Relapses in MS ► Relapses are the most obvious evidence of inflammatory disease activity in RRMS ► Relapse frequency in typical untreated RRMS populations enables treatment effect to be rapidly assessable in a 12-month clinical study Total number of relapses during the study period Total in-study person-years

10. Effect on Annualized Relapse Rates: Summary of Phase III Trials – 2 years in-study 35 % Reduction in relapse rates 30 25 20 15 10 5 0 31% 8 MIU qod IFN beta-1b P=0.0001 29% 4.4 MIU tiw IFN beta-1a P<0.001 32% 8.8 MIU tiw IFN beta-1a P<0.0001 29% 20 mg qd glatiramer acetate P=0.055 6 MIU qw IFN beta-1a P=0.04 18% N.B.: Results are from separate clinical trials Jacobs et al. Ann Neurol. 1996;39:285; IFNB MS Study Group. Neurology. 1993;43:655; IFNB MS Study Group and University of British Columbia MS/MRI Analysis Group. Neurology. 1995;45:1277; Johnson et al. Neurology. 1995:45:1268; Johnson et al. Neurology. 1998;50:701; PRISMS Study Group. Lancet. 1998;352:1498; Rebif package insert.

11. Is MS All About Relapses? ► Hypothesis: if relapses cause long-term disability then patients with frequent relapses should be at higher risk for disability ► From the London Ontario natural history studies patients with frequent attacks are at highest risk for future ambulatory disability ► Assumption: modifying the relapse rate will influence long-term disability Weinshenker et al. 1989 Brain 112:1419

12. Proportion of Placebo Groups with Clinical Activity Jacobs et al. Ann Neurol. 1996;39:285; IFNB MS Study Group. Neurology. 1993;43:655; IFNB MS Study Group and University of British Columbia MS/MRI Analysis Group. Neurology. 1995;45:1277; Johnson et al. Neurology. 1995:45:1268; Johnson et al. Neurology. 1998;50:701; PRISMS Study Group. Lancet. 1998;352:1498. Relapses EDSS Progress IFNβ-1b (3 year) 86%39% IFNβ-1a (QW) (2 year) 77%35% IFNβ-1a (TIW) (2 year) 84%38% Glatiramer acetate (2 year) 73%25%

13. How is Sustained Progression Measured? ► Most clinical trials define progression by comparing the change in EDSS from baseline to study conclusion, and then confirm the change in EDSS at 3 or 6 months ► Does this measure of confirmed progression reflect permanent disability? ► If so, then confirmed changes in EDSS during the course of the trial should be sustained by the end of the study

14. Does Sustained Disability Measure Permanent Disability? ► 50% of patients with a 1 point change, confirmed at 3 months will improve to a lower EDSS ► 33% of patients with a 1 point change, confirmed at 6 months, will improve to a lower EDSS ► More stringent measures of change are harder to demonstrate in 2-year trials because relatively few MS patients will progress ► Conclusions: 6 months sustained EDSS change is more rigorous than a 3-month sustained change, but neither is a good predictor of long term disability Liu C & Blumhardt LD J Neurol Neurosurg Psychiatry. 2000;68:450-7.

15. *  1 EDSS point sustained for 6 months in 6 MIU qw phase III trial and for 3 months in all other phase III trials. Jacobs et al. Ann Neurol. 1996;39:285; IFNB MS Study Group. Neurology. 1993;43:655 IFNB MS Study Group and University of British Columbia MS/MRI Analysis Group. Neurology. 1995;45:1277 Johnson et al. Neurology. 1995:45:1268; Johnson et al. Neurology. 1998;50:701 PRISMS Study Group. Lancet. 1998;352:1498 Effect on Sustained Disability*: Summary of Phase III Trials 40 Reduction in sustained disability progression (%) 12% 22% 30% NS 30 20 10 0 8.8 MIU tiw IFN beta-1a 4.4 MIU tiw IFN beta-1a 20 mg qd glatiramer acetate p<0.05 p=NS 37% 6 MIU qw IFN beta-1a p=0.02 29% 8 MIU qod IFN beta-1b p=NS

16. Summary ► Relapses and disability progression represent different but complimentary aspects of MS natural history ► Relapse rate reduction and the mean change in EDSS are the most sensitive clinical outcome measures in MS trials ► The generally accepted sustained change in EDSS measure is not a reliable marker of long term disability ► Phase III trials results showed: The interferons and glatiramer acetate modestly reduce the relapse rate The interferons and glatiramer acetate modestly reduce the relapse rate IFN beta-1a has a statistically significant impact on sustained disability progression over two years IFN beta-1a has a statistically significant impact on sustained disability progression over two years IFN beta-1a and glatiramer acetate have a statistically significant impact on the mean EDSS over two year IFN beta-1a and glatiramer acetate have a statistically significant impact on the mean EDSS over two year

17. Are direct comparator studies needed in MS or can we make valid conclusions from cross trial comparisons?

18. Cross Trial Comparisons Relative Efficacy (RR) IFNβ-1a 30 µg qw qw IFNβ-1b, 250 µg qod IFN β-1a 44 µg tiw GA 20 mg qd Relapse rate (annualized) -18%-34%-32%-29% Relapse-Free (2 years) +42%+95%+100%+36% Progression free -37%-29%-30%-12% New T2 Lesions -36%-83%-78%-38% Gd+ Lesions -42%--88%-33% BOD - 4% -17%-15%-8%

19. 672 days (96 weeks) IFNβ-1a GA Time to first relapse (days) Hazard ratio (95% CI): 0.943 (0.74, 1.21) p = 0.643 0 100 200 300400 500 600700 0.00 0.25 0.50 0.75 1.00 Survival distribution function The REGARD Trial Time to First Relapse (1 o endpoint)

20. The BEYOND Trial Relapse Risk (1 o Endpoint) ► No significant difference in relapse risk between any group Interferon beta-1b 500 vs. Interferon beta-1b 250 Interferon beta-1b 500 vs. Glatiramer acetate Interferon beta-1b 250 vs. Glatiramer acetate Primary Analysis Sensitivity Analysis (no major protocol violations, 100% of doses, post-hoc) P-values(one-sided) P-values(one-sided) P=0.16P=0.73P=0.43 P=0.29P=0.30P=0.18 0.5 1.0 1.5

21. What can be learned from long-term follow up studies?

22. Long-Term Follow Up ► Do long-term follow up studies adequately address medication safety? ► Do long-term studies adequately address longitudinal efficacy? ► Have methods of analysis for longitudinal studies been optimized?

23. BiasImpactStrategy Ascertainment Modified therapeutic effect dependent on characteristics of participating patients. F/U must be as complete as possible Directly compare baseline and on- RCT characteristics of those patients in LTF to those not in LTF Informed Therapeutic Decisions Inflated estimate of therapeutic benefit because patients doing well continue therapy whereas failing patients switch or stop therapy. MPR: Use percent of total possible time on therapy instead of absolute time to assess exposure. Treatment Selection Modified therapeutic effect dependent on patient selection characteristics. Propensity Scoring: Adjust for the propensity (i.e., likelihood) that a particular treatment will be selected based on available patient characteristics Multiple Testing Increased risk of Type 1 error from the use of multiple predictor variables and weighting schemes Create a single model and apply adjustments to p-values according to the number of predictors tested in the model. Sources of Bias in LTFU Studies

24. Glatiramer Acetate 15 year LTFU Ford C et al. Mult Scler. 2010;16:342-50.

25. Glatiramer Acetate 15 year LTFU Ford C et al. Mult Scler. 2010;16:342-50.

26. Glatiramer Acetate 15 year LTFU ► In a small cohort of patients followed for 15 years, glatiramer acetate was safe and well tolerated ► 65% of continuously treated patients did not progress to SPMS ► 41% of patients withdrawing from the study did so because of disease progression ● Propensity scores were used to try to adjust for differences between ongoing and withdrawing patients ► EDSS at baseline predicts EDSS at 15 years

27. IFN β -1a (QW) LTFU Disposition Bermel R et al. Mult Scler. 2010 Feb 18. [Epub ahead of print] Complete 2-year follow-up (n=172) Unascertained (n=36) Ascertained for ASSURANCE (n=136; 79%) Able to locate, Unable to contact (n=13) Unable to locate (n=23) Living (n=122; 90%) Deceased (n=14; 10%) ICF and question booklet completed LOCF

28. IFN β -1a QW LTFU Outcomes Bermel R et al. Mult Scler. 2010 Feb 18. [Epub ahead of print] Currently receiving IM IFN ß-1a (n=56) Not currently receiving IM IFN ß-1a (n=66) Patients, % P=0.062 EDSS Milestone P=0.326 P=0.114 P=0.006

29. IFN β -1a QW LTFU Conclusions ► 79% of eligible patients were located for the 15 year follow up ► At 15 years, patients currently on IFN β-1a had less progression in EDSS scores than patients not on IFN β- 1a ► However, patients not currently on IFN β-1a had higher baseline EDSS scores suggesting more severe baseline MS ● Propensity scores were used to adjust for these differences ► Inferences with regard to association with lower EDSS and ongoing treatment were not made Bermel R et al. Mult Scler. 2010 Feb 18. [Epub ahead of print]

30. 124 125 123 52 58 56 IFNβ-1b 250 µg IFNβ-1b 50 µg Placebo 19881993 Pivotal Study (n=372) LTF 2005 Cross-sectional investigation of: - clinical outcomes (disability, relapse rate) - imaging (brain and spinal MRI) - cognition and mood - QoL, resource use - lab parameter including NAb's and PgX Patients under regular medical care - no trial 1990 IFN β-1b LTFU Design Ebers G et al. presented at ECTRIMS, Madrid, Spain, September 2006: P666 Ebers G et al. presented at AAN, October 2006: M-3

31. Event Rates and Long-Term Efficacy Clinical and Radiological Endpoints 1.Need to demonstrate that the short-term event-rates are correlated with long-term outcome. 2.Need to demonstrate that the short-term event-rates contribute independently to predicting outcome. 3.Need to demonstrate that therapies which reduce the event-rates, are also associated with improved long-term outcome. Ebers G et al. presented at ECTRIMS, Madrid, Spain, September 2006: P666 Ebers G et al. presented at AAN, October 2006: M-3

32. IFN β -1b LTFU Adjusted OUtcome Any Variable + Any Exposure Weighting – Any Negative Outcome EDSS p<0.001 1 Exposure p<0.001 2 High Low Ebers G et al. presented at ECTRIMS, Madrid, Spain, September 2006: P666 Ebers G et al. presented at AAN, October 2006: M-3

33. Event Rates and Long-Term Efficacy Conclusions 1.The LTF study demonstrates that the short-term event- rate is correlated with long-term outcome. 2.The LTF study also demonstrates that the short-term event-rate contributes independently to predicting long- term outcome. 3.The LTF study provides convincing evidence that early initiation and sustained use of IFNβ-1b has a beneficial impact on long-term outcome in MS. 4.The analysis strategy employed provides a methodological framework for mitigating bias in assessing long-term efficacy in other clinical trials having similar non-randomized data.

34. ►Disease modifying therapy seems favorably effect the long-term course of MS ►Propensity score adjusted analysis and other statistical methods for controlling biases inherent in long term, unblinded studies are important statistical advances for interpreting these studies ►Once the MS community agrees on the relevant covariates, these methods can be used to sort out some of these issues without the cost (and ethical dilemmas) posed by long-term placebo-controlled trials. Conclusions