Disclosures

1. Using short-term MRI measures to predict long-term outcomes in the CombiRx randomized clinical trialclinicaltrials.gov - NCT00211887Study Support: NIH/NINDS 5-U01-NS045719Biogen Idec and Teva Neuroscience generously contributed study medications and matched placebowithout input into the design, conduct or analysis of the trial. Jerry S. Wolinsky, Ponnada A. Narayana, Flavia Nelson, Sushmita Datta, Brandi Gates, Gary R. Cutter, Robin Conwit, Stacey S. Cofield, and Fred D. Lublin, for the CombiRx Investigators and MRI Analysis Center

2. Disclosures Dr. Wolinsky received compensation from the following commercial entities: Astellas, consultant services, Bayer HealthCare, member/consultant Bayer Multiple Sclerosis Council, Celgene, consultant services, Eli Lilly, member Data Safety Monitoring Board, Hoffman LaRoche, member Steering Committee, Novartis Pharmaceuticals Corp., member Steering Committee and consultant, Sanofi, member Steering Committee, Teva Neuroscience, consultant, Teva Pharmaceuticals, member Scientific Advisory Board, member Data Safety Monitoring Board, consultant; royalties received for out licensed monoclonal antibodies through the UTHSCH to Millipore (Chemicon International) Corporation since 1993; contractual support from Sanofi. Dr. Narayana received compensation from the following commercial entities: Teva Neuroscience Advisory Board Meeting Dr. Nelson received compensation from the following commercial entities: Bayer Pharmaceuticals, Scientific Advisory Board/Speaker Teva Neurosciences, Scientific Advisory, Board/Speaker, Biogen Idec, Scientific Advisory Board, EMD Serono, Scientific Advisory Board/Speaker. Dr. Datta reports no conflicts. Ms. Gates reports no conflicts. Dr. Cutter received compensation for participating in Data and Safety Monitoring Committees: Sanofi-Aventis, Cleveland Clinic, Daichi-Sankyo, Glaxo Smith Klein Pharmaceuticals, Genmab Biopharmaceuticals, Eli Lilly, Medivation, Modigenetech, Ono Pharmaceuticals, Teva, NHLBI, NINDS, NMSS and Consulting services: Abbott, Bayer, Novartis, Consortium of MS Centers (grant), Genzyme, Klein-Buendel Incorporated, Nuron Biotech, Biogen, Somnus Pharmaceuticals and Teva; Dr. Cutter is president of Pythagoras, Inc.; Dr. Cutter reports received CMSC task orders which involve research from various Pharma organizations. Dr. Conwitreports no conflicts. Dr. Cofield received personal compensation from GlaxoSmithKline, TEVA Pharmaceuticals, Orthotech Biotech, the Department of Defense and the American Academy for Orthopedic Surgery for consulting services, research funds and/or DSMB service. Dr. Lublin received compensation from the following commercial entities: for consulting: Bayer HealthCare Pharmaceuticals; Biogen Idec, EMD Serono, Novartis, Pfizer, Teva Neuroscience, Genmab, Medicinova, Actelion, Allozyne, Sanofi, Acorda, Questcor, Avanir, Roche, Celgene, Abbott, MorphoSys, Johnson & Johnson; Revalesio, Coronado Bioscience, Genzyme, MedImmune; compensation from Elsevier for serving as Co-Chief Editor of Multiple Sclerosis and Related Diseases. The CombiRx Investigators have varied relationships . 2

3. Background: The Core Trial • CombiRx was a unique collaboration between NIH, academic institutions, practicing neurologists, pharma and biotech; the largest test thus far of the potential for combining immunotherapies in early RRMS. It provides the longest controlled clinical, MRI and biomarker dataset of any MS trial. • Drugs that are effective in disease management in isolation, particularly those with differing presumed mechanisms of action, have the potential for additive or synergistic effects when used in combination. We sought to determine if weekly interferon β-1a (IFN) and daily glatiramer acetate (GA) used together were more effective than either alone in relapsing remitting multiple sclerosis (RRMS). • CombiRx was a 3-year, double-blind, 86 center randomized clinical trial. Screening was initiated January 2005, enrollment closed April 2009 (1,008 subjects randomized), core study and extension closed January 2012, database lock 11 April 2012. • Primary endpoint annualized was the relapse rate over 3 years comparing combined IFN + GA versus the best single agent arm. • Multiple predefined secondary clinical and MRI defined secondary outcomes were analyzed over the 36 month core study on an intent to treat basis. 3

4. Background: Major Outcomes • The combination of weekly interferon β-1a and daily glatiramer acetate (IFN + GA) was not superior to either monotherapy for risk of relapse, using either protocol defined (PDE) or PDE + non-protocol defined (NPDE) relapse definitions. • GA was superior to IFN in reducing relapse risk using either definition. • IFN + GA was no better than either agent alone in lessening confirmed EDSS progression or change in MSFC. • IFN + GA was superior to either monotherapy in reducing new lesion activity and the accumulation of net lesion volumes. • All three treatment arms were effective in reducing MRI-defined disease within 6 months of therapy initiation. • Effects on other MRI measures of brain pathology statistically did not differ. • IFN + GA resulted in a higher proportion of participants with disease activity free state compared to either single arm, driven by the MRI results. 4

5. Objectives • Evaluate the predictive validity of baseline, 6 month and 12 month MRI measures for selected clinical and MRI outcomes at 1, 2 and 3 years on study. • Determine if these predictors and outcomes are differentially affected by randomization arm assignments. 5

6. Methods: Imaging and Image Analysis

7. Methods: Clinical Definitions 7

8. Results 8

9. Baseline Demographics *IFN + GA v GA, p=0.03 9

10. Available Image Data 10

11. Does entry MRI Gd+ status influence on trial Gd+ activity? Gd negative Enhancement Free Survival Gd positive Month on Trial Baseline Gd status strongly influenced time remaining Gd+ activity free on study. In this Cox Proportional Hazards model, adjusted for baseline age and treatment group, those with any Gd+ lesions at baseline had a 3.1 (95% CI: 2.4, 3.9) increased hazard of new Gd+ activity on trial (p<0.0001). The KM curve presented assumed that interval missing scan data had no Gd+ activity; there were no substantive differences when the analysis assumed subjects censored at the first missing scan. 11

12. Does entry MRI Gd+ status influence on trial Gd+ activity? Gd negative Enhancement Free Survival Gd positive Month on Trial In this conservative model, missing scans were assumed to be Gd+. Those with any Gd+ lesions at baseline had a 1.9 (95% CI: 2.4, 3.9) increased hazard of additional Gd+ activity on trial (p<0.0001). The KM curve presented assumed that interval missing scan data had Gd+ activity and declared the subject as a failure at that time point regardless of subsequent scan information. 12

13. What are the transitional probabilities of Gd+ activity state conversion on study? Those with any Gd+ lesions at baseline were 8 times more likely to be Gd negative at month 6 than were Gd negative subjects to convert to Gd+. Those who lacked enhancement at baseline had only an 8% chance of conversion to Gd+ at month 6. 13

14. Do transitional probabilities of Gd+ activity state conversion differ by study group? Those on IFA+GA with Gd+ lesions at entry were 17.5 times more likely to be Gd neg at month 6 than those Gd neg were to convert to Gd+. Those on IFA only with Gd+ lesions at entry were 5.7 times more likely to be Gd neg at month 6 than those Gd neg were to convert to Gd+. Those on GA alone with Gd+ lesions at entry were 4.9 times more likely to be Gd neg at month 6 than those Gd neg were to convert to Gd+. 14

15. What is the predictive value of 6 month Gd+ activity for Gd+ activity further on study? 81% of those without Gd activity at month 6 had no further enhancements on study, compared to only 37.4% of those who had activity at 6 months who remained activity free for the remainder of the trial (p < 0.001). However, 93.3% of those without Gd activity at baseline had no enhancement activity on their 36 month scans, compared to 77.6% of those who had activity at entry who were activity free at month 36 (p < 0.001). 15

16. Does entry MRI Gd+ status influence on trial combined unique activity (CUA)? Gd negative MRI activity Free Survival Gd positive Month on Trial Baseline Gd status strongly influenced time remaining Gd+ activity free on study. In this Cox Proportional Hazards model, adjusted for baseline age and treatment group, those with any Gd+ lesions at baseline had a 1.84 (95% CI: 1.6, 2.2) increased hazard of new CUA activity on trial (p<0.0001). The KM curve assumes that missing scans had no new MRI activity; there were no substantive overall differences when an alternative analysis assumed that subjects censored at the first missing scan due as there were relatively small numbers of interval missing scans. 16

17. What are the transitional probabilities of MRI-defined lesion activity (CUA) state conversion on study? Transitional probabilities were quite similar from 12 to 24 months and again from 24 to 36 months on study for those subjects with available scans over those intervals 17

18. What are the transitional probabilities of MRI-defined lesion activity (CUA) state conversion on study? Gd+ ± New T2 Gd- New T2 CUA - Transitional probabilities were quite similar from 12 to 24 months and again from 24 to 36 months on study for those subjects with available scans over those intervals 18

19. What is the predictive value of 6 month combined unique activity (CUA) for further on study activity? 73.3% of those with no CUA at month 6 had no new CUA at months 24 and 36, compared to only 38.8% of those who had CUA at 6 months who remained CUA free at months 24 and 36 (p < 0.001). However, 93.3% of those without Gd activity at baseline had no enhancement activity on their 36 month scans, compared to 77.6% of those who had activity at entry who were activity free at month 36 (p < 0.001). 19

20. What is the predictive value of 6 and 12 month combined unique activity (CUA) for further on study activity? • Models for CUA in the first year of the trial, adjusted for baseline Gd status and treatment group, are predictive of new CUA later in the trial. • Those with CUA in the first year of the trial are 4.6 times (CI 3.2, 6.5; p <0.0001)more likely to have additional new CUA by month 24 or 36. Those without CUA at 6 months, but with CUA at 12 months were 4.9 times more likely to develop additional new CUA on trial (2.8, 8.5; p < 0.0001). • Baseline Gd+ subjects were 1.8 times more likely to have new CUA in in the later portions of the trial (CI 1.3, 2.6; p = 0.0005). • Those randomized to IFN alone and GA alone were more likely have new CUA positive in the later 2/3 of the trial compared to those on IFN + GA (p = 0.0034, global for any treatment difference). 20

21. Does relapse in the first year predict subsequent relapse or combined unique activity (CUA) on study? Note: There were 24 combinations of CUA and relapse activity in the first year and subsequent two years on trial. However, most participants (82.3%) did not experience a relapse after the first year of the trial, regardless of MRI activity in the first year. Thus, the addition of clinical activity to these combinations adds little predictive value. These relationships were similar using only Gd activity in the first year with relapse activity. 21

22. Does MRI activity in the 6 months predict subsequent clinical progression? Neither Gd activity status nor CUA status at 6 months was predictive of progression during the core 3 year trial (p > 0.96 and p > 0.58, respectively). However, these are potentially ‘pseudo’ mutually exclusive measures, as the definitions of progression used required participants to stay on trial up to 9 months for progression to be confirmed. 22

23. Conclusions • In this early RRMS cohort, baseline Gd activity strongly influenced on trial MRI-defined activity. • On study Gd+ lesions were suppressed in all three treatments, with combined treatment (IFN + GA) more effective than either monotherapy. • In this low frequency MRI monitoring protocol, CUA provided 3–4 times more mean lesions per subject than did Gd+ lesion counts alone. However, as the scan reader noted, the use of Gd+ was especially helpful in subjects with high T2 lesion loads at entry. • Loss of disease free activity state was driven by on study CUA. • Lack of a placebo only arm and similar treatment effects across the three study arms may have damped the predictive power of early change on MRI for later clinical outcomes. • Nevertheless, the presence of MRI-activity over the first year was predictive of continued activity over years 2 and 3. • Analysis of the extension phase of CombiRx may show stronger predictive associations. 23

24. Acknowledgments (1) • Data and Safety Monitoring Board: • Current Members: Anne Lindblad (Chair), Jason Fine, Gary Franklin, David Miller, Emanuelle Waubant; Past Chair: Stan van den Noort • External Medical Safety Monitors: • Rene Elkin, Aaron Miller • Clinic Coordinating Center Staff: • Tarah Gustafson, Ruthie Perez, Susan Mallinson, Christine Hannigan, YadiraBencosme, Colleen Farrell, Shelly Phelps, CuongTroung, Michele Cohen, Tai Baker, Francine Diaz, Heidi Sadowsky • Statistical & Data Management Center Staff: • Steve Powell, Mary Jo Sewell, Theresa McVie, Amber Salter, Jie Zhang, Todd Jenkins, Yvonne Gulley, Jeffry Hebert, Micheal Tyler, Chistopher Parks, Michael Brascome, Richard Mailhot, Monika Baier • MRI-AC Staff - current: • John Lincoln, Irina Vainrub, Rimma Brokin, Kelly Ton, Lucille Lambert, Yanjie Zhang, Chelsa Pitt • NINDS Clinical Trials Advisors: • Robin Conwit, Ursula Utz, Peter Gilbert, Joanne Odenkirchen • Funding for Contrast Letter Acuity: • National Multiple Sclerosis Society 24

25. Acknowledgments (2) • Site Staff, MSFC Raters, Coordinators, Examiners and Investigators:: • Drs. Agius, Bashir, Baumhefner, Birnbaum, Blevins, Bomprezzi, Boster, Brown, Burkholder, Camac, Campagnolo, Carter, Cohen, Cooper, Corboy, Cross • Drs. Dewitt, Dunn, Edwards, Eggenberger, English, Felton, Fodor, Ford, Freedman, Galetta, Garmany, Goodman, Gottesman, Gottschalk, Gruenthal, Gudesblatt • Drs. Hagan, Hamill, Herbert, Hoffman, Holub, Honeycutt, Hughes, Hutton, Jacobs, Johnson, Kasper, Kattah, Kaufman, Keegan, Khan, Khatri, Kita, Koffma • Drs. Lallana, Lava, Lindsey, Loge, Lynch, McGee, Mejico, Metz, O'Connor, Pandey, Panitch, Pelletier, Preiningerova, Rammohan, Riley, Riskind, Rolak, Royal • Drs. Scarberry, Schulman, Scott, Sheppard, Sheremata, Stone, Stuart, Subramaniam, Thadani, Thomas, Thrower, Tullman, Turel, Vollmer, Waldman, Weinstock-Guttman, Wendt, Williams, Wynn, Yeung 25