ASH 2018 Investor Relations Event December 2, 2018 Forward Looking Statement

CHANGING THE COURSE OF HUMAN HEALTH THROUGH BOLD PURSUITS IN SCIENCE

This presentation contains forward-looking statements, which are generally statements that are not historical facts. Forward-looking statements can be identified by the words “expects,” “anticipates,” “believes,” “intends,” “estimates,” “plans,” “will,” “outlook,” “targets” and similar expressions. Forward-looking statements are based on management’s current plans, estimates, assumptions and projections, and speak only as of the date they are made. We undertake no obligation to update any forward-looking statement in light of new information or future events, except as otherwise required by law. Forward-looking statements involve inherent risks and uncertainties, most of which are difficult to predict and are generally beyond our control. Actual results or outcomes may differ materially from those implied by the forward- looking statements as a result of the impact of a number of factors, many of which are discussed in more detail in our Annual Report on Form 10-K and our other reports filed with the Securities and Exchange Commission.

2 CHANGING THE COURSE OF HUMAN HEALTH THROUGH BOLD PURSUITS IN SCIENCE

Mark Alles Chairman and Chief Executive Officer REVLIMID® US IP: Timeline

Mar 2022 Jan 31, 2026 All Remaining Generic Filers Natco Generic Entry Full Generic Entry

Earliest Forecasted Earliest Forecasted LITIGATION & Trial Decision Appeal Decision SETTLEMENT Q4 2019 to Q2 2020 Q2 2021 to Q4 2021 TIMELINES Volume-Limited Natco Sales Permitted*

2019 2020 2021 2022 2023 2024 2025 2026 2027 2028

COM REMS MDS MOU Polymorph Polymorph MCL MOU Oct 2019 Oct 2020 Apr 11, 2023 Nov 2024 Apr 2027 Jul 2028 PATENT MCL MOU 1 EXPIRATIONS May 15, 2023 MM MOU Oct 7, 2023

*Natco settlement provides for a phased, volume-limited generic entry beginning March 2022 with COM = Composition Of Matter MCL = Mantle Cell Lymphoma mid-single digits percentage of total capsules dispensed during first full year of entry gradually MOU = Method Of Use MDS = Myelodysplastic Syndromes increasing to approximately 30% of total capsules in 2025 and full generic entry in January 31, 2026 MM = Multiple Myeloma 1 This is a representative listing of the patents in suit 4 Advancing Innovative Pipeline to Offset Revlimid Sales Erosion

KEY PIVOTAL ASSETS NEXT WAVE Expected Peak Sales1 bb21212 +$2B

Liso-cel $3B

Luspatercept3 +$2B

Fedratinib $1B

Ozanimod $4-6B

Total Peak Sales ~ $12-14B1

Anticipated U.S. 2022 and Beyond Launches 2019-2020

1 ~$12-14 billion in unadjusted peak sales represents sum of individual potential peak 2bb2121 is being developed in collaboration with bluebird bio sales for each illustrated product, which may not coincide. 3 Luspatercept is being developed in collaboration with Acceleron Pharma 5 ASH 2018: Key Oral Presentations Highlighting Hematology Pipeline

Luspatercept in Beta Thalassemia (BELIEVE™) Luspatercept in Myelodysplastic REVLIMID Combo (R2) in Syndromes (MEDALIST™) Relapsed/Refractory Selected as the #1 Indolent Non-Hodgkin Plenary Presentation Lymphoma (AUGMENT™)

Liso-cel in bb21217 in Relapsed/Refractory Relapsed/Refractory Chronic Lymphocytic Multiple Myeloma Leukemia (TRANSCEND-CLL) JCARH125 in Relapsed/Refractory Multiple Myeloma (EVOLVE)

6 Meeting Agenda

Key Data Review at ASH 2018 Jay Backstrom, MD Chief Medical Officer

Advancing a Multi-Modal BCMA Kristen Hege, MD Portfolio CVP, Translational Development

Emerging Research and Early Rupert Vessey, MA, FRCP, DPhil Development Pipeline President, Research & Early Development

Diverse Portfolio with Significant Nadim Ahmed Potential President, Hematology & Oncology

Q&A Moderator: David Elkins EVP & Chief Financial Officer 7 CHANGING THE COURSE OF HUMAN HEALTH THROUGH BOLD PURSUITS IN SCIENCE

Jay Backstrom, MD Chief Medical Officer Addressing Unmet Needs in Myeloid Disease

1 MEDALIST™ in MDS

2 BELIEVE™ in Beta Thalassemia

9 Addressing Unmet Needs in Myeloid Disease

1 MEDALIST™ in MDS

Author Abstract Title List #1 The Medalist Trial: Results of a Phase 3, Randomized, Double-Blind, Placebo-Controlled Study of Luspatercept to Treat Anemia in Patients with Very Low-, Low-, or Intermediate-Risk Myelodysplastic Syndromes (MDS) with Ring Sideroblasts (RS) Who Require Red Blood Cell (RBC) Transfusions Plenary, Sunday, December 2, 2018 2:00 – 4:00PM

10 Luspatercept is a Potential First-in-Class Erythroid Maturation Agent

▪ Luspatercept is an investigational first-in-class erythroid maturation agent that neutralizes select TGF-β superfamily ligands to inhibit aberrant Smad2/3 signaling and enhance late-stage erythropoiesis in MDS models1 ▪ In a phase 2 study in LR, non-del(5q) MDS, luspatercept yielded a high frequency of transfusion reduction or RBC-TI in patients with MDS-RS vs other subtypes2

Luspatercept ActRIIB / IgG1 Fc recombinant TGF-β

fusion protein superfamily ActRIIB Modified ligand extracellular Cytoplasm P domain of Smad2/3 ActRIIB Nucleus Complex Human IgG1 Fc domain Erythroid maturation 1. Suragani RN, et al. Nat Med. 2014;20:408-414; ActRIIB, human activin receptor type IIB; IgG1 Fc, immunoglobulin G1 fragment crystallizable; RBC-TI, red blood cell transfusion independence; 2. Platzbecker U, et. A. Lancet Oncol. 2017; 18:1338. RS, ring sideroblasts; TGF-β, transforming growth factor beta. 11 MEDALIST™: Study Design A Randomized, Double-Blind, Placebo-Controlled, Phase III Study

MEDALIST™ Study Endpoints Luspatercept 1.0 mg/kg (s.c.) Primary endpoint: PATIENT POPULATION every 21 days n = 153 ▪ Red blood cell transfusion independence ≥ 8 Randomize Dose titrated up to a maximum of 1.75 mg/kg weeks (weeks 1–24) ▪ MDS-RS (WHO): ≥ 15% RS or 2:1 ≥ 5% with SF3B1 mutation Placebo (s.c.) every 21 days ▪ < 5% blasts in bone marrow n = 76 Key secondary endpoints: ▪ No del(5q) MDS ▪ Red blood cell transfusion independence ≥ 12 weeks (weeks 1–24 and weeks 1–48) ▪ IPSS-R Very Low-, Low-, or Intermediate-risk Disease & Response Assessment week 24 & every 6 months ▪ Prior ESA response Treatment discontinued for lack of Additional secondary endpoints: ⎻ Refractory, intolerant clinical benefit or disease progression per IWG criteria; no crossover allowed ▪ HI-E (IWG 2006 criteria1) for any ⎻ ESA naive: EPO > 200 U/L consecutive 56-day period ▪ Average RBC transfusion burden ▪ Reduction in red blood cell ≥ 2 units/8 weeks transfusion burden ≥ 4 RBC units/8 a ▪ No prior treatment with disease- weeks or Subjects followed ≥ 3 years post final modifying agents (e.g. IMiDs, ▪ Mean Hb increase of ≥ 1.5 g/dL/8 dose for AML progression, subsequent weeksb HMAs) MDS treatment and overall survival ▪ Duration of response ▪ Hb change from baseline

a In patients with baseline RBC transfusion burden ≥ 4 units/8 weeks. b In Data cutoff: May 8, 2018 Includes last subject randomized + 48 weeks. patients with baseline RBC transfusion burden < 4 units/8 weeks. EPO, erythropoietin; HMA, hypomethylating agent; iMID, immunomodulatory drug; IWG, International Working Group; s.c., subcutaneously; Hb, hemoglobin; HI-E, hematological improvement–erythroid. SF3B1, splicing factor 3b subunit 1; WHO, World Health Organization. 1. Cheson BD, et al. Blood. 2006;108:419-425. 12 MEDALIST™: Demographics and Baseline Disease Characteristics

Luspatercept Placebo Characteristic (n = 153) (n = 76) Age, median (range), years 71 (40–95) 72 (26–91) Male, n (%) 94 (61.4) 50 (65.8) Time since original MDS diagnosis, median (range), months 44.0 (3–421) 36.1 (4–193) WHO classification RCMD-RS, n (%) 145 (94.8) 74 (97.4) RBC transfusion burden, median (range), units/8 weeksa 5 (1–15) 5 (2–20) ≥ 6 units/8 weeks, n (%) 66 (43.1) 33 (43.4) < 6 units/8 weeks, n (%) 87 (56.9) 43 (56.6) Pre-transfusion Hb, median (range), g/dL 7.6 (6–10) 7.6 (5–9) IPSS-R risk categoryb Very Low, Low, n (%) 127 (83.0) 63 (82.9) Intermediate, n (%) 25 (16.3) 13 (17.1) SF3B1 mutation, n (%) 141 (92.2) 65 (85.5)c Serum EPO < 200 U/L, n (%) 88 (57.5)c 50 (65.8) ≥ 200 U/L, n (%) 64 (41.8)c 26 (34.2) a In the 16 weeks prior to randomization. b 1 (0.7%) patient in the luspatercept arm was classified as IPSS-R High-risk. c Data were missing for 1 patient. RCMD-RS, refractory cytopenia with multilineage dysplasia with RS.

13 MEDALIST™: Treatment Exposure

Luspatercept Placebo Parameter (n = 153) (n = 76) Treatment duration, median (range), weeks 49 (6–114) 24 (7–89) Completed ≥ 24 weeks of treatment (primary phase), n (%) 128 (83.7) 68 (89.5) Completed ≥ 48 weeks of treatment, n (%) 78 (51.0) 12 (15.8) Number of doses received, median (range) 16 (2–37) 8 (3–30) Maximum dose escalation, n (%)a 1.0 mg/kg 35 (22.9) 5 (6.6) 1.33 mg/kg 28 (18.3) 8 (10.5) 1.75 mg/kg 90 (58.8) 63 (82.9) Patients remaining on treatment, n (%) 70 (45.8) 6 (7.9) Patients discontinued from treatment, n (%) 83 (54.2) 70 (92.1) Lack of benefit 51 (33.3) 50 (65.8) Patient withdrawal 14 (9.2) 10 (13.2) AE 10 (6.5) 4 (5.3) Disease progression 3 (2.0) 2 (2.6) Other 5 (3.3) 4 (5.3) a Dose may be titrated up to a maximum of 1.75 mg/kg. AE, adverse event. 14 MEDALIST™: Efficacy Results

Primary Endpoint: Key Secondary Endpoints: Red Blood Cell Transfusion Independence ≥ 8 Weeks Red Blood Cell Transfusion Independence ≥ 12 Weeks Luspatercept Placebo Luspatercept Placebo RBC-TI ≥ 8 weeks RBC-TI ≥ 12 weeks (n = 153) (n = 76) (n = 153) (n = 76) Weeks 1–24, n (%) 58 (37.9) 10 (13.2) Weeks 1–24, n (%) 43 (28.1) 6 (7.9) 95% CI 30.2–46.1 6.5–22.9 95% CI 21.14–35.93 2.95–16.40 P valuea < 0.0001 P valuea 0.0002 Weeks 1–48, n (%) 51 (33.3) 9 (11.8) a Cochran–Mantel–Haenszel test stratified for average baseline RBC transfusion requirement (≥ 6 units vs < 6 units of RBCs/8 weeks) and baseline IPSS-R score (Very Low or Low vs Intermediate). 95% CI 25.93–41.40 5.56–21.29 CI, confidence interval. P valuea 0.0003

a Cochran–Mantel–Haenszel test stratified for average baseline RBC transfusion requirement (≥ 6 units vs < 6 units of RBCs/8 weeks) and baseline IPSS-R score (Very Low or Low vs Intermediate).

15 MEDALIST™: Primary Endpoint: Subgroup Analysis

Luspatercept, n (%) Placebo, n (%) OR (95% CI) P Value Overall 58/153 (37.9) 10/76 (13.2) 5.06 (2.28–11.3) < 0.0001 Average baseline RBC transfusion requirement ≥ 6 units/8 weeks 6/66 (9.1) 1/33 (3.0) 3.20 (0.37–27.7) 0.2699 < 6 units/8 weeks 52/87 (59.8) 9/43 (20.9) 5.61 (2.40–13.1) < 0.0001 4 to < 6 units/8 weeks 15/41 (36.6) 1/23 (4.3) 12.7 (1.55–104) 0.0046 < 4 units/8 weeks 37/46 (80.4) 8/20 (40.0) 6.17 (1.95–19.5) 0.0013 Baseline serum EPO (U/L) < 100 23/51 (45.1) 7/31 (22.6) 2.82 (1.03–7.71) 0.0413 100 to < 200 14/37 (37.8) 2/19 (10.5) 5.17 (1.04–25.9) 0.0338 200–500 17/43 (39.5) 1/15 (6.7) 9.15 (1.10–76.2) 0.0188 Age group ≤ 64 years 17/29 (58.6) 3/16 (18.8) 6.14 (1.43–26.3) 0.0108 65–74 years 23/72 (31.9) 4/29 (13.8) 2.93 (0.91–9.41) 0.0635 ≥ 75 years 18/52 (34.6) 3/31 (9.7) 4.94 (1.32–18.5) 0.0120 Gender Male 32/94 (34.0) 4/50 (8.0) 5.94 (1.96–18.0) 0.0006 Female 26/59 (44.1) 6/26 (23.1) 2.63 (0.92–7.48) 0.0673 Time since initial diagnosis at baseline ≤ 2 years 14/40 (35.0) 3/19 (15.8) 2.87 (0.71–11.6) 0.1312 > 2–5 years 30/62 (48.4) 4/34 (11.8) 7.03 (2.21–22.3) 0.0004 > 5 years 14/51 (27.5) 3/23 (13.0) 2.52 (0.65–9.83) 0.1756 Baseline IPSS-R risk Very Low or Low 48/127 (37.8) 9/63 (14.3) 3.65 (1.65–8.05) 0.0009 Intermediate 10/25 (40.0) 1/13 (7.7) 8.00 (0.89–71.6) 0.0398 Baseline platelet count < 100  109/L 2/8 (25.0) 1/6 (16.7) 1.67 (0.11–24.3) 0.7171 100–400  109/L 42/128 (32.8) 8/61 (13.1) 3.24 (1.41–7.42) 0.0042 > 400  109/L 14/17 (82.4) 1/9 (11.1) 37.3 (3.31–422) 0.0006 OR, odds ratio. 0 0.1 0.25 0.5 2 10 20 100 1,500 Favors placebo Favors luspatercept

16 MEDALIST™: Duration of RBC-TI Response in Primary Endpoint Responders

1.0 Median duration (weeks) (95% CI): 30.6 (20.6–40.6) vs 13.6 (9.1–54.9) 0.9 Luspatercept 0.8 Placebo 0.7 Censored

TI TI 0.6 - 0.5

RBC 0.4 0.3 0.2

0.1 Probability Probability of Maintaining 0 0 10 20 30 40 50 60 70 80 90 100 110 120 Duration of RBC-TIa (week) Number of patients Luspatercept 58 49 37 29 22 18 10 6 3 2 1 1 0 Placebo 10 9 3 2 2 2 0 a During indicated treatment period. Patients who maintained RBC-TI at the time of analysis are censored.

17 MEDALIST™: Secondary Endpoint: Erythroid Response (HI-E)

Luspatercept Placebo (n = 153) (n = 76) Achieved HI-Ea (weeks 1–24), n (%) 81 (52.9) 9 (11.8) Reduction of ≥ 4 RBC units/8 weeks (baseline transfusion burden ≥ 4 units/8 weeks) 52/107 (48.6) 8/56 (14.3) Hb increase of ≥ 1.5 g/dL (baseline transfusion burden < 4 units/8 weeks) 29/46 (63.0) 1/20 (5.0) 95% CI 44.72–61.05 5.56–21.29 P valueb < 0.0001 Achieved HI-Ea (weeks 1–48), n (%) 90 (58.8) 13 (17.1) Reduction of ≥ 4 RBC units/8 weeks (baseline RBC transfusion burden ≥ 4 units/8 weeks) 58/107 (54.2) 12/56 (21.4) Hb increase of ≥ 1.5 g/dL (baseline RBC transfusion burden < 4 units/8 weeks) 32/46 (69.6) 1/20 (5.0) 95% CI 50.59–66.71 9.43–27.47 P valueb < 0.0001 a Defined as the proportion of patients meeting the HI-E criteria per IWG 2006 criteria (Cheson et al. 2006) sustained over a consecutive 56-day period during the indicated treatment period. b Luspatercept compared with placebo, Cochran–Mantel–Haenszel test.

18 MEDALIST™: Safety Summary

Luspatercept Placebo (n = 153) (n = 76) Patients with ≥ 1 TEAE, n (%) 150 (98.0) 70 (92.1) Patients with ≥ 1 serious TEAE 48 (31.4) 23 (30.3) Patients with ≥ 1 Grade 3 or 4 TEAE 65 (42.5) 34 (44.7) Patients with TEAEs leading to deatha 5 (3.3) 4 (5.3) Patients with ≥ 1 TEAE causing discontinuation, n (%) 13 (8.5) 6 (7.9)

▪ TEAEs were balanced between the armsb ▪ Progression to AML occurred in 4 patients (3/153 [2.0%] in the luspatercept arm; 1/76 [1.3%] in the placebo arm) a In luspatercept arm: sepsis (n = 2), multiple organ dysfunction syndrome, renal failure, and hemorrhagic shock; in placebo arm: sepsis, urosepsis, general physical health deterioration, and respiratory failure. b The most common grade 3 or 4 TEAEs reported in luspatercept-treated patients were anemia (6.5% of patients), fall (4.6%), and fatigue (4.6%). TEAE, treatment-emergent adverse event.

19 MEDALIST™: Author’s Conclusions

▪ In lower-risk, RS-positive MDS, treatment with luspatercept resulted in a significantly higher percentage of patients who achieved RBC-TI, major RBC transfusion reduction, or hemoglobin increase, compared with placebo ▪ Erythroid responses are durable, with approximately 40% of patients achieving RBC-TI sustained at 12 months of treatment ▪ Luspatercept was generally well tolerated in this patient population ▪ Luspatercept is a potential new therapy for the treatment of patients with lower-risk, RS-positive MDS with RBC transfusion-dependent anemia

20 Addressing Unmet Needs in Myeloid Disease

2 BELIEVE™ in Beta-Thalassemia

Author Abstract Title Cappellini #163 The Believe Trial: Results of a Phase 3, Randomized, Double-Blind, Placebo-Controlled Study of Luspatercept in Adult Beta- Thalassemia Patients Who Require Regular Red Blood Cell (RBC) Transfusions Oral, Saturday, December 1, 2018 2:00PM

21 BELIEVE™: Study Design A Randomized, Double-Blind, Placebo-Controlled, Phase 3 Study

Study Endpoints β-thalassemiaa Primary endpoint: ▪ ≥ 33% reduction from baseline in RBC patients transfusion burden (with a reduction of ≥ Current 2 RBC units) during weeks 13–24 ▪ ≥ 18 years, requiring study regular RBC statusc Key secondary endpoints: transfusions Luspatercept 1 mg/kgb s.c. every 21 days + BSC ▪ ≥ 33% reduction from baseline in RBC ▪ (defined as: 6–20 (n = 224) transfusion burden during weeks 37–48 RBC units in the 24 Open- Post- ▪ ≥ 50% reduction from baseline in RBC weeks prior to May be titrated up to 1.25 mg/kg label treatment transfusion burden during weeks 13–24

randomization with no 2:1 (up to 5 follow-up ▪ ≥ 50% reduction from baseline in RBC ≥ 35-day transfusion-

Randomize transfusion burden during weeks 37–48 free during that years) (3 years) Placebob ▪ Mean change from baseline in RBC period) Study

s.c. every 21 days + BSC unblinding transfusion burden during weeks 13–24 ▪ (N = 336) (n = 112) Additional secondary endpoints: ▪ ≥ 33% or ≥ 50% reduction from 12-week period 12-week period Double-blind period Crossover permitted baseline in RBC transfusion burden historical screening / run- (48 weeks) during any 12 weeks or 24 weeks transfusions in transfusions a β-thalassemia or hemoglobin E/β-thalassemia (β-thalassemia with mutation and/or multiplication of α globin was allowed; b RBC transfusions and iron on study chelation therapy to maintain each patient’s baseline hemoglobin level; c The trial is fully enrolled and patients continue to receive treatment or follow-up. BSC, best supportive care; LPFD, last patient first dose; RCT, randomized clinical trial; s.c., subcutaneously. .

22 BELIEVE™: Demographics and Baseline Disease Characteristics

Luspatercept Placebo Characteristic (n = 224) (n = 112) Age, median (range), years 30 (18–66) 30 (18–59) Female, n (%) 132 (58.9) 63 (56.3) β0/β0 n (%) 68 (30.4) 35 (31.3) Hemoglobin (24 week),b median (range), g/dL 9.31 (4.5–11.4) 9.15 (5.8–11.7) RBC transfusion burden, median (range), units/12 weeks 6.12 (3–14) 6.27 (3–12) RBC transfusion burden, median (range), units/24 weeks 14 (6–24) 15 (6–26) Splenectomy, n (%) 129 (57.6) 65 (58.0) Serum ferritin, median (range), µg/L 1,447 (88–6,400) 1,304 (136–6,400) LIC, median (range), mg/g dry weight 6.14 (0.8–125.0) 5.05 (0.2–53.2) > 7 mg/g dry weight, n (%) 103 (46.0) 45 (40.2) Myocardial iron by T2*, median (Q1–Q3), ms 34.7 (27.4–40.3) 36.3 (29.0–42.0)

a According to HbVar database. b Defined as the mean of all documented pre-transfusion hemoglobin values during the 24 weeks prior to first dose for each patient. LIC, liver iron concentration.

23 BELIEVE™: Efficacy Results

A significantly greater proportion of luspatercept-treated patients achieved a A significantly greater proportion of luspatercept-treated patients achieved ≥ 33% reduction from baseline in transfusion burden during weeks 13 to 24 clinically meaningful reductions in transfusion burden of ≥ 33% and ≥ 50% Primary Endpoint Met: All Key Secondary Endpoints Met: Rate of Erythroid Response Rates of Erythroid Response 30 30 P < 0.0001 (OR 5.79, 95% CI 2.24-14.97 Luspatercept Luspatercept 25 25 P < 0.0001a Placebo Placebo 20 20

15 15 P = 0.0017c 21.4% b (n = 48) P = 0.0303

10 10 19.6 Patients Achieving 33% ≥ Achieving Patients

5 5 10.3 BurdenReduction (%)

Transfusion Burden Reduction (%) Reduction Burden Transfusion 7.6 (n = 5) 3.6 0.9 0 Transfusion Achieving Patients 0 1.8 Luspatercept Placebo ≥ 33% ≥ 50% ≥ 50% (n = 224) (n = 112) (from week (from week (from week 37–48) 13–24) 37–48) CI, confidence interval; OR, odds ratio. ▪ The least squares mean change in transfusion burden from baseline to weeks . 13–24 (luspatercept versus placebo) was −1.35 RBC units/12 weeks (95% CI −1.77 to −0.93; P < 0.0001)

a OR 6.44, 95% CI 2.27–18.26. b OR 4.55, 95% CI 1.03–20.11. c OR 11.92, 95% CI 1.65–86.29. 24 BELIEVE™: Primary Endpoint Subgroup Analysis Favors Luspatercept

Luspatercept Placebo Sub-groups OR (95% CI) P value n/N (%) n/N (%) Overall 48/224 (21.4) 5/112 (4.5) 5.79 (2.24, 14.97) < 0.0001 Region: North America & Europe 23/100 (23.0) 1/51(2.0) 14.94 (1.95, 114.12) 0.0009 Region: Middle East & North Africa 11/52 (21.2) 2/26 (7.7) 3.22 (0.66, 15.77) 0.1351 Region: Asia–Pacific 14/72 (19.4) 2/35 (5.7) 3.98 (0.85, 18.62) 0.0629 Age: ≤ 32 years 22/129 (17.1) 4/63 (6.3) 3.00 (0.98, 9.20) 0.0476 Age: > 32 years 26/95 (27.4) 1/49 (2.0) 17.50 (2.27, 134.98) 0.0004 Splenectomy: Yes 31/129 (24.0) 2/65 (3.1) 9.72 (2.22, 42.53) 0.0003 Splenectomy: No 17/95 (17.9) 3/47 (6.4) 2.94 (0.81, 10.69) 0.0918 Sex: Female 35/132 (26.5) 4/63 (6.3) 5.33 (1.80, 15.80) 0.0011 Sex: Male 13/92 (14.1) 1/49 (2.0) 8.05 (1.01, 64.16) 0.0218 β-thalassemia Gene: β0/β0 9/68 (13.2) 2/35 (5.7) 2.54 (0.48, 13.51) 0.2708 β-thalassemia Gene: Non-β0/β0 39/155 (25.2) 3/77 (3.9) 8.35 (2.47, 28.23) < 0.0001 Baseline Transfusion Burden: ≤ 6 units/12 weeks 27/112 (24.1) 3/56 (5.4) 5.61 (1.60, 19.65) 0.0033 Baseline Transfusion Burden: > 6 units/12 weeks 21/112 (18.8) 2/56 (3.6) 6.16 (1.38, 27.44) 0.0082 Baseline Hemoglobin: < 9 g/dL 22/87 (25.3) 4/51 (7.8) 3.78 (1.25, 11.42) 0.0128 Baseline Hemoglobin: ≥ 9 g/dL 26/137 (19.0) 1/61 (1.6) 14.17 (1.85, 108.79) 0.0012 Baseline Liver Iron: ≤ 3 mg/g dry weight 12/70 (17.1) 1/37 (2.7) 7.18 (0.88, 58.63) 0.0335 Baseline Liver Iron: > 3 to ≤ 7 mg/g dry weight 13/51 (25.5) 0/30 (0) Infinity 0.0053 Baseline Liver Iron: > 7 to ≤ 15 mg/g dry weight 10/38 (26.3) 1/19 (5.3) 5.41 (0.67, 43.34) 0.0741 Baseline Liver Iron: > 15 mg/g dry weight 13/65 (20.0) 3/26 (11.5) 1.79 (0.47, 6.78) 0.3831 0.1 1 10 100 Favors placebo Favors luspatercept

25 BELIEVE™: Reduction in Transfusion Burden During Any 12- and 24- Week Interval

During any 12- or 24-week interval, a significantly greater proportion of luspatercept-treated patients achieved clinically meaningful reductions in transfusion burden of ≥ 33% and ≥ 50%

Any 12-week interval Any 24-week interval

100 100 Luspatercept

P < 0.0001a Placebo 80 80

60 60 P < 0.0001b P < 0.0001c

40 40 P < 0.0001 Reduction (%) Reduction 70.5 (%) Reduction

20 40.2 20 41.1 16.5

29.5 Patients Achieving Transfusion Burden Transfusion Achieving Patients 6.3 Burden Transfusion Achieving Patients 2.7 0.9 0 0 ≥ 33% ≥ 50% ≥ 33% ≥ 50%

a OR 5.69, 95% CI 3.46–9.35. b OR 9.95, 95% CI 4.44–22.33. c OR 25.02, 95% CI 7.76–80.71. d OR 20.37, 95% CI 2.86–144.94.

26 BELIEVE™: Safety Summary

Luspatercept Placebo Treatment-Emergent Adverse Events, n (%) (n = 223a) (n = 109a)

Patients with at least 1 TEAE (any grade) 214 (96.0) 101 (92.7)

Patients with at least 1 grade TEAE (grade ≥ 3)b 65 (29.1) 17 (15.6)

Patients with at least 1 serious TEAEc 34 (15.2) 6 (5.5)

Patients with at least 1 TEAE resulting in the following:

Deathd 0 1 (0.9)

Study drug discontinuation 12 (5.4) 1 (0.9)

a Safety population. b No one organ class or system was predominant.c Anemia was the only serious TEAE occurring in > 1% of patients in either arm (luspatercept, n = 3 [1.4%]; placebo, n = 0 [0%]). d TEAE of acute cholecystitis resulted in death in 1 of 109 (0.9%) placebo patients; no luspatercept-treated patients died due to TEAEs. TEAE, treatment-emergent adverse event.

27 BELIEVE™: TEAEs by Frequency ≥ 10% in Either Arm (all grades)

Luspatercept Placebo n (%) (n = 223a) (n = 109a) Back pain 61 (27.4) 32 (29.4) Upper respiratory tract infection 59 (26.5) 36 (33.0) Headache 58 (26.0) 26 (23.9) Bone pain 44 (19.7) 9 (8.3) Arthralgia 43 (19.3) 13 (11.9) Pyrexia 36 (16.1) 23 (21.1) Cough 32 (14.3) 12 (11.0) Fatigue 30 (13.5) 14 (12.8) Oropharyngeal pain 28 (12.6) 12 (11.0) Diarrhea 27 (12.1) 11 (10.1) Dizziness 25 (11.2) 5 (4.6) Asthenia 22 (9.9) 11 (10.1) Myalgia 22 (9.9) 11 (10.1) Pharyngitis 20 (9.0) 13 (11.9) a Safety population.

28 BELIEVE™: Grade 3–4 TEAEs by Frequency ≥ 1% in Either Arm

Luspatercept Placebo n (%) (n = 223a) (n = 109a) Anemia 7 (3.1) 0 Increased LIC 6 (2.7) 1 (0.9) Hyperuricemia 6 (2.7) 0 Hypertension 4 (1.8) 0 Syncope 4 (1.8) 0 Back pain 3 (1.3) 1 (0.9) Bone pain 3 (1.3) 0 Blood uric acid increased 3 (1.3) 0 Increased AST 3 (1.3) 0 Increased ALT 2 (0.9) 3 (2.8) Thromboembolic eventsb 2 (0.9) 0 ▪ In total, thromboembolic events (all grades) were reported in 8/223 (3.6%) luspatercept-treated patients (deep venous thrombosis, pulmonary embolism, portal vein thrombosis, ischemic stroke, thrombophlebitis, superficial phlebitis) and 1/109 (0.9%) placebo- treated patients (phlebitis). In all cases, patients had multiple risk factors for thromboembolic events

a Safety population. b Thromboembolic events included as a TEAE of interest; other events occurring in < 1% of patients are not shown. ALT, alanine aminotransferase; AST, aspartate aminotransferase.

29 BELIEVE™: Author’s Conclusions

▪ Luspatercept showed a statistically significant improvement in the primary endpoint of ≥ 33% reduction in transfusion burden compared with placebo ▪ Statistical significance was also demonstrated with luspatercept versus placebo for all key secondary endpoints, including ≥ 33% and ≥ 50% reductions in transfusion burden ▪ Luspatercept showed a statistically significant and clinically meaningful reduction in transfusion burden compared with placebo during any 12 or 24 weeks in the study period ▪ Luspatercept was generally well tolerated in this patient population ▪ Luspatercept is a potential new treatment for adult patients with β-thalassemia who require regular RBC transfusions

30 Luspatercept Clinical Development Plan

Phase I Phase II Phase III Program Status

MEDALIST™ (ESA refractory) ▪ Regulatory filings for MEDALIST™ & BELIEVE™ expected H1:19 MDS COMMANDS™ (front-line MDS) ▪ COMMANDS™ trial initiated Q3:18

- BELIEVE™ (Transfusion-dependent Beta-Thal)

▪ BEYOND trial currently enrolling Thal Beta BEYOND (NTD Beta-Thal) ▪ Phase II MF trial currently enrolling; Data expected H2:19

MF MF-001

Additional indications involving ineffective erythropoiesis being actively considered

31 Expanding Celgene’s Footprint in B-cell Malignancies

1 AUGMENT™ (R2) in R/R FL or MZL

2 TRANSCEND CLL – Liso-cel in R/R CLL

32 Expanding Celgene’s Footprint in B-cell Malignancies

1 AUGMENT (R2) in R/R iNHL

Author Abstract Title Leonard J #445 A Phase III Randomized Study of Lenalidomide Plus Rituximab (R2) Vs Rituximab/Placebo in Patients with Relapsed/Refractory Indolent Non-Hodgkin Lymphoma Oral, December 2, 2018 4:30 PM

33 AUGMENT™: Study Design

≤ 12 cycles or until PD, relapse, or intolerability

R-lenalidomide (R2) Relapsed/refractory Rituximab: 375 mg/m2 d1, 8, 15, 22 of cycle 1; d1 of cycles 2-5 Lenalidomide: 20 mg/d*, d1-21/28 (12 cycles) 5-year follow- FL and MZL up for PD, OS, *10 mg if CrCl between 30 to 59 mL/min. (N = 358) 1:1 response, and SPMs R-placebo Stratification Rituximab: 375 mg/m2 d1, 8, 15, 22 of cycle 1; d1 of cycles 2-5 • Prior rituximab treatment (yes vs no) Placebo: matched capsules (12 cycles) • Time since last antilymphoma therapy (≤ 2 vs > 2 years) • In patients with thromboembolism risk, prophylactic anticoagulation or antiplatelet • Histology (FL vs MZL) therapy at investigator discretion was recommended • Growth factor use was allowed per ASCO/ESMO guidelines1,2

▪ Primary endpoint: PFS by IRC (2007 IWG criteria w/o PET) ▪ Secondary endpoints include OS

NCT01938001 1. Crawford et al. Ann Oncol. 2010;21 Suppl 5:248-251. 2. Smith et al. J Clin Oncol. 2015;33:3199-3212. Lenalidomide is not approved as monotherapy or in combination in R/R FL or MZL in any country 34 AUGMENT™: Baseline Characteristics (ITT population)

R2 R-placebo Characteristic, n (%) (n = 178) (n = 180) Median age, y (range) 64 (26-86) 62 (35-88) Age ≥ 60 y 108 (61) 106 (59) Age ≥ 65 y 82 (46) 73 (41) ECOG PS 1-2* 62 (35) 52 (29) Positive BM involvement, n involved/performed (%) 33/106 (31) 31/111 (28) Ann Arbor stage III or IV at study entry 137 (77) 124 (69) Bulky disease† (≥ 7 cm or ≥ 3 cm x 3) 45 (25) 49 (27) High tumor burden per GELF criteria1,2 97 (54) 86 (48) Histology FL 147 (83) 148 (82) MZL 31 (17) 32 (18) LDH > ULN 43 (24) 39 (22) B-symptoms 16 (9) 12 (7) FLIPI score‡ 0 or 1 52 (29) 67 (37) 2 55 (31) 58 (32) 3 to 5 69 (39) 54 (30)

Data cutoff June 22, 2018. *Two patients in each group had ECOG PS of 2. †Bulky disease is defined as at least 1 lesion that is ≥ 7 cm or at least 3 lesions with ≥ 3 cm in the longest diameter by investigator review. ‡Two patients in R2 group and 1 patient in R-placebo group did not have FLIPI score recorded. 1. Salles et al. Lancet. 2011;377:42-51. 2. Brice et al. J Clin Oncol. 1997;15:1110-1117.

35 AUGMENT™: Prior Treatment History (ITT population)

R2 R-placebo Characteristic, n (%) (n = 178) (n = 180) Number of prior systemic antilymphoma regimens 1 102 (57) 97 (54) 2 31 (17) 42 (23) ≥ 3 45 (25) 41 (23) Prior rituximab treatment 152 (85) 150 (83) Prior rituximab-containing chemotherapy 130 (73) 129 (72) regimen ≤ 2 y since last antilymphoma therapy 89 (50) 92 (51) Relapse ≤ 2 y of initial diagnosis 56 (31) 61 (34) ≤ 2 y of initial therapy 66 (37) 75 (42) Refractory to last regimen* 30 (17) 26 (14)

*Refractory defined as no response or progressive disease < 6 months after last dose. Most common prior last regimens were R-CHOP (R2 37%, R-placebo 38%), combination chemotherapy (R2 33%, R-placebo 31%,) and rituximab monotherapy (R2 21%, R-placebo 23%) Data cutoff June 22, 2018. 36 AUGMENT™: Treatment Administration and Discontinuation (Safety Population)

R2 R-placebo (n = 176) (n = 180) Completed study treatment, n (%) 125 (71) 110 (61) Reasons for any drug discontinuation, n (%) Discontinued study treatment early 52 (30) 70 (39) Progression 21 (12) 54 (30) Adverse event 14 (8) 8 (4) Neutropenia 5 (3) 1 (< 1) Withdrawal by patient/other 15 (9) 8 (4) Death 2 (1) 0

Ending Dose Lenalidomide, n (%) R2 (n = 176) ▪ 66% of patients had at least 1 dose 20 mg 96 (55) interruption of lenalidomide due to AEs 15 mg 27 (15) ▪ Growth factors were administered to 36% 10 mg 32 (18) of R2 and 12% of R-placebo patients 5 mg 16 (9) 2.5 mg 5 (3)

Data cutoff June 22, 2018. 37 AUGMENT™: Primary Endpoint- Progression-Free Survival (ITT, IRC)

PFS by IRC*

Median follow up: 28.3 months

R2 R-placebo Median PFS (n = 178) (n = 180) HR (95% CI) P Value By IRC, mo (95% CI) 39.4 (22.9-NE) 14.1 (11.4-16.7) 0.46 (0.34-0.62) < 0.0001 By investigator, mo (95% CI) 25.3 (21.2-NE) 14.3 (12.4-17.7) 0.51 (0.38-0.69) < 0.0001

*Censoring rules based on FDA guidance. Data cutoff June 22, 2018. 38 AUGMENT™ : Prespecified Subgroup PFS Analysis (IRC, ITT)

Data cutoff June 22, 2018.

39 AUGMENT™: Prespecified Subgroup PFS Analysis (IRC, ITT) (cont.)

*Refractory defined as no response or progressive disease < 6 months after last dose. Data cutoff June 22, 2018. 40 AUGMENT™: Overall Survival (OS)

ITT Patients with FL (prespecified subgroup analysis)

(nominal)

▪ 42 total deaths (16 R2, 26 R-placebo) ▪ 35 total deaths (11 R2, 24 R-placebo) ▪ 2-year OS was 93% (95% CI, 87%-96%) for R2 and 87% (95% ▪ 2-year OS was 95% (95% CI, 90%-98%) for R2 and 86% CI, 81%-92%) for R-placebo (95% CI, 79%-91%) for R-placebo

Data cutoff June 22, 2018. 41 AUGMENT™: Summary of Author’s Conclusions

▪ AUGMENT met its primary endpoint as R2 demonstrated statistically significant and clinically relevant superiority over R-placebo for the primary endpoint of PFS – Interim analysis for overall survival had strong trend in the follicular lymphoma subgroup ▪ PFS advantage in prespecified subgroups, except MZL, was consistent with overall population ▪ Greater efficacy with manageable toxicity allowed more patients to complete R2 compared to R-placebo – Disease progression (more frequent with R-placebo) was the most common cause of treatment discontinuation ▪ AEs differed between arms – Neutropenia, infections, tumor flare, and cutaneous reactions were more frequent with R2 – Fewer cases of SPMs and histological transformations occurred with R2

. 42 Expanding Celgene’s Footprint in B-cell Malignancies

2 TRANSCEND CLL – Liso-cel in R/R CLL

Author Abstract Title Siddiqi #300 Rapid MRD-Negative Responses in Patients with Relapsed/Refractory CLL Treated with Liso-Cel, a CD19- Directed CAR T-Cell Product: Preliminary Results from Transcend CLL 004, a Phase 1/2 Study Including Patients with High-Risk Disease Previously Treated with Ibrutinib Oral, Sunday, December 2, 2018 8:45 AM

43 TRANSCEND CLL 004 Phase I Study Design (NCT03331198)

Key Eligibility Dose-escalation: mTPI-2 Designc 28-day DLT period • Relapsed/refractory CLL/SLL • Failed or ineligible for BTKia Primary Objective • Determine recommended dose • High-risk diseaseb: failed ≥ 2 prior therapies • Safety • Standard-risk disease: failed ≥ 3 prior therapies Exploratory Objectives • ECOG PS 0–1 • Antitumor activity • Pharmacokinetic profile

Dose Level Dose Treated (N=16) –1 2.5 × 107 CAR+ T cells -- 1 5 × 107 CAR+ T cells 6 2 1 × 108 CAR+ T cells 10

aFailure defined as SD or PD as best response, or PD after previous response, or discontinuation due to intolerance (unmanageable toxicity). Ineligibility defined as requirement for full-dose anticoagulation or history of arrhythmia. bComplex cytogenetics abnormalities, del(17p), TP53 mutation, or unmutated IGHV. cGuo W, Wang SJ, Yang S, et al. Contemp Clin Trials. 2017;58:23-33.

BTKi, Bruton's tyrosine kinase inhibitor; CAR, chimeric antigen receptor; CLL, chronic lymphocytic leukemia; DLT, dose-limiting toxicity; ECOG PS, Eastern Cooperative Oncology Group Performance Status; IGHV, immunoglobulin heavy chain variable region; mTPI, modified toxicity probability interval; PD, progressive disease; SD, stable disease; SLL, small lymphocytic lymphoma. Data on file 21 September 2018. 44 TRANSCEND CLL: Key Study Assessments

Enrollment & Measurable disease apheresis reconfirmed Bridging therapy FOLLOW-UP allowed Lymphodepletion Liso-cel On-study: 24 months Screen FLU 30 mg/m2 and 2–7 days Long-term: up to 15 years CY 300 mg/m2 × 3 days after FLU/CY after last Liso-cel treatment Liso-cel manufacturing

Months Post–Liso-cel

Baselinea 1 3 6 9 12 18 24

CT/PETb

BM Examc

MRDc,d

aBaseline refers to screening or if patient received bridging therapy, baseline disease assessments were repeated pre-lymphodepletion. bResponse assessment by iwCLL criteria (Hallek M et al. Blood. 2008;111(12):5446-5456). cCompleted when CR or PR with residual lymphadenopathy are suspected. dIn PB by 6-color flow cytometry (10-4) and, for those who are undetectable by flow, in BM by Immunoseq NGS (10-6).

BM, bone marrow; CR, complete response; CT, computed tomography; CY, cyclophosphamide; FLU, fludarabine; iwCLL, International Workshop on Chronic Lymphocytic Leukemia; MRD, minimal residual disease; NGS, next-generation sequencing; PB, peripheral blood; PR, partial response. 45 TRANSCEND CLL: Baseline Characteristics

Characteristic All Patients (N=16) Age, median (range) y 64.5 (51–76) Male, n (%) 8 (50.0) Stage, n (%) Rai Stage III/IV 10 (62.5) Binet Stage C 10 (62.5) High-risk features (any), n (%) 12 (75.0) TP53 mutation 10 (62.5) Complex karyotypea 8 (50.0) Del (17p) 7 (43.8) Prior lines of therapy, median (range) 4.5 (2–11) Prior ibrutinib, n (%) 16 (100.0) Ibrutinib relapse/refractory, n (%) 13 (81.3) Ibrutinib progression and prior venetoclax,b n (%) 8 (50.0)

aAt least 3 chromosomal aberrations. b7 patients progressed on venetoclax; 1 patient had best response of SD after 3 months of treatment.

46 TRANSCEND CLL: AEs of Special Interest

No Grade 4/5 AEs of special interest occurred Total DL1 DL2 (N=16) (n=6) (n=10) CRS – any grade, n (%) 12 (75.0) 6 (100.0) 6 (60.0) Median time to first onset, d (range) 6.5 (1–10) 6.5 (1–9) 5.0 (2–10) Median duration, d (range) 5.5 (2–30) 5.5 (3–30) 5.5 (2–13) Grade 3, n (%) 1 (6.3) 0 1 (10.0) Neurologic events (NE)a – any grade, n (%) 6 (37.5) 2 (33.3) 4 (40.0) Median time to first onset, d (range) 10.0 (4–21) 16.0 (11–21) 8.0 (4–11) Median duration, d (range) 6.5 (2–20) 4.0 (2–6) 8.0 (3–20) Grade 3b, n (%) 3 (18.8) 2 (33.3) 1 (10.0) Any, n (%) CRS or NEa 13 (81.3) 6 (100.0) 7 (70.0) CRS and NEa 5 (31.3) 2 (33.3) 3 (30.0) Tocilizumab and/or dexamethasone use 11 (68.8) 4 (66.7) 7 (70.0) Tumor lysis syndrome – any grade, n (%) 2 (12.5) 1 (16.7) 1 (10.0) Grade 3, n (%) 2 (12.5) 1 (16.7) 1 (10.0)

aNeurologic events are treatment-related events defined by the Investigator. bEncephalopathy n=1; aphasia n=1; confusional state and encephalopathy n=1. AE, adverse event; CAR, chimeric antigen receptor; DL, dose level; CRS, cytokine release syndrome.

47 TRANSCEND CLL: Response Rates

Total DL1 DL2 Best Overall Response, n (%) (N=16) (n=6) (n=10) ORR 13 (81.3) 6 (100.0) 7 (70.0) CR/CRi 7 (43.8) 5 (83.3) 2 (20.0) PR/nPR 6 (37.5) 1 (16.7) 5 (50.0) SD 2 (12.5) 0 2 (20.0) PD 1 (6.3) 0 1 (10.0)

Total DL1 DL2 Response at 30 Days Post–liso-cel, n (%) (N=16) (n=6) (n=10) ORR 12 (75.0) 6 (100.0) 6 (60.0) CR/CRi 5 (31.3) 3 (50.0) 2 (20.0) PR/nPR 7 (43.8) 3 (50.0) 4 (40.0) Response at 3 Months Post–liso-cel, n (%) (N=10) (n=6) (n=4) ORR 8 (80.0) 5 (83.3) 3 (75.0) CR/CRi 5 (50.0) 3 (50.0) 2 (50.0) PR/nPR 3 (30.0) 2 (33.3) 1 (25.0)

CR, complete response; CRi, complete remission with incomplete blood count recovery; DL, dose level; nPR, nodular partial remission; ORR, overall response rate; PD, partial disease; PR, partial remission; SD, stable disease. 48 TRANSCEND CLL: Minimal Residual Disease

uMRD4 at any time point, n (%) Total DL1 DL2 (N=15) (n=6) (n=9) Blood, flow 11 (73.3) 6 (100.0) 5 (55.6) (N=8) (n=5) (n=3) Bone marrow, NGS 7 (87.5) 4 (80.0) 3 (100.0)

▪ 11 of 15 (73.3%) patients had uMRD4 in blood by flow at Day 30 – All continue to remain undetectable at latest follow-up ▪ 5 patients have post-dose follow-up at Month 6 – All continue to maintain uMRD4 response (CR, n=4 and PR, n=1 by iwCLL criteria)

CR, complete response; DL, dose level; iwCLL, International Workshop on Chronic Lymphocytic Leukemia; NGS, next-generation sequencing; PR, partial remission; uMRD4, undetectable minimal residual disease sensitivity 10-4.

49 TRANSCEND CLL: Author’s Conclusions

Liso-cel demonstrated promising activity in a heavily pretreated patient population with high-risk CLL, all of whom had received prior ibrutinib ▪ Liso-cel toxicities were manageable at both dose levels tested – Low rates of grade 3 CRS (6.3%); neurologic events (18.8%) ▪ High best ORR (81.3%) and a CR/CRi rate (43.8%) – Responses have deepened over time at 3- and 6-month follow-up – CR continues in 5 of 6 patients with at least 3 months of follow-up ▪ Early uMRD4 responses were observed in a majority of patients (73.3%) and were maintained at 3 and 6 months ▪ Following analysis of dose escalation data and selection of a RP2D, the phase 2 portion of the trial will open for accrual (expected 1H 2019)

1H, first half; CLL, chronic lymphocytic leukemia; CR, complete response; CRi, complete remission with incomplete blood count recovery; CRS, cytokine release syndrome; uMRD4, undetectable minimal residual disease sensitivity 10-4; NE, neurologic events; ORR, overall response rate; RP2D, recommended phase 2 dose.

50 Liso-cel’s Broad Clinical Development Plan

Phase I Phase II Phase III

*TRANSCEND NHL-001 (3L+ R/R DLBCL)

* TRANSCEND WORLD (3L+ R/R DLBCL EU and Japan; 2L transplant ineligible)

*TRANSFORM (2nd line R/R transplant eligible)

DLBCL *PILOT (2nd line R/R transplant ineligible)

TRANSCEND OUTREACH (3L+ R/R community centers)

PLATFORM (3L+ R/R combinations)

*TRANSCEND CLL 004 (R/R CLL) CLL

Ped ALL (Pediatric R/R ALL and NHL) ALL

U.S. Approval for R/R DLBCL Expected in mid-2020

* Pivotal trials 51 CHANGING THE COURSE OF HUMAN HEALTH THROUGH BOLD PURSUITS IN SCIENCE

Kristen Hege, MD CVP, Translational Development

52 Industry Leading Myeloma Pipeline Focused on Two Major Campaigns: BCMA & CELMoD® Agents BCMA CELMoD® Agents

A potentially disruptive approach to myeloma Developing Novel CELMoD® Compounds to ▪ bb2121* CAR T – KarMMa™ pivotal trial enrollment complete Address Unmet Needs in Myeloma ▪ bb21217* CAR T – Phase I ongoing ▪ Iberdomide (CC-220) – Phase I/II trial ongoing ▪ JCARH125** CAR T – Phase I ongoing ▪ CC-92480 – Phase I ongoing ▪ CC-93269 TCE – Phase I ongoing

▪ BCMA ADC*** – IND filing anticipated in H1:2019 *In collaboration with bluebird bio **JCARH125 managed by Juno Therapeutics, a Celgene Company *** In collaboration with Sutro Biopharma 53 BCMA: Striving to Change the Treatment Paradigm

CAR T Cell Program Updates at ASH

Author Abstract Title Shah N #488 Initial results from a phase 1 clinical study of bb21217, a next-generation anti-BCMA CAR T therapy. Oral Abstract, Sunday, December 2, 4:45PM

Author Abstract Title Mailankody S #957 JCARH125, Anti-BCMA CAR T-cell Therapy for Relapsed/Refractory Multiple Myeloma: Initial Proof of Concept Results from a Phase 1/2 Multicenter Study (EVOLVE). Oral Abstract, Monday, December 3, 5:00PM

54 ASH 2018, Abstract 488

Initial Results From a Phase I Clinical Study of bb21217, a Next-Generation Anti-BCMA CAR T Cell Therapy

Nina Shah, MD1, Melissa Alsina, MD2, David Siegel, MD3, Sundar Jagannath, MD4, Deepu Madduri, MD5, Jonathan L. Kaufman, MD6, Ashley Turka7, Lyh Ping Lam, PharmD, RPh7, Monica Massaro, MPH7, Kristen Hege, MD8, Fabio Petrocca, MD7, Jesus G Berdeja, MD9 and Noopur Raje, MD10 1University of California San Francisco, San Francisco, CA; 2Department of Blood and Marrow Transplantation and Cellular Immunotherapy, H. Lee Moffitt Cancer Center and Research Institute, Tampa, FL; 3Myeloma Division, John Theurer Cancer Center, Hackensack University Medical Center, Hackensack, NJ; 4Mount Sinai Hospital, New York, NY; 5Department of Hematology and Medical Oncology, Tisch Cancer Institute, Icahn School of Medicine at Mount Sinai, New York, NY; 6Winship Cancer Institute / Department of Hematology and Medical Oncology, Emory University School of Medicine, Atlanta, GA; 7bluebird bio, Inc., Cambridge, MA; 8Celgene Corporation, San Francisco, CA; 9Sarah Cannon Research Institute, Nashville, TN; 10Cancer Center, Massachusetts General Hospital, Boston, MA bb21217: Next-Generation Anti-BCMA CAR T Cell Therapy for Multiple Myeloma

• bb21217 uses the same CAR construct design as bb21211

T cell Terminally • bb21217 is cultured with PI3 kinase inhibitor, plasticity differentiated bb007, to enrich for T cells displaying a Self renewal No self renewal Long-lived Short-lived memory-like phenotype CD62L+, CD62L-, CD27- • CAR T cells enriched for this phenotype may CD27+ persist and function longer than non-enriched CAR T cells2 • Persistence of functional CAR T cells after infusion may be one determinant of duration of response

BCMA, B-cell maturation antigen; PI3K, phosphoinositide 3 kinase. 1. Friedman et al. Hum Gene Ther 2018;29:585-601. 2. Fraietta JA, et al. Nat Med. 2018 May;24:563-571 CRB-402 Phase I Study Design and Status

3 + 3 dose escalationa

CAR+ T cell 150 × 106 300 × 106 450 × 106 800 × 106 dose Study Status as of Oct 18, 2018

N ≈ 50 Collected N=13 • R/R MM • ≥3 prior regimens – IMiDs and Dosed N=12 proteasome (150 × 106 dose) inhibitors b b – Or double- HTB n=6 LTB refractory n=6 • ≥50% BCMA expression (dose Median (min, escalation only) max) follow-up Primary endpoints: AEs, DLTs 26 wk (4, 51) c NCT03274219 Other endpoints: Response , PFS, OS, MRD, CAR+ T cell expansion and persistence

AE, adverse events; BCMA, B-cell maturation antigen; DLT, dose-limiting toxicity; HTB, high tumor burden; IMiD, immunomodulatory imide drugs; LTB, low tumor burden; MRD, minimal residual disease; OS, overall survival; PFS, progression-free survival; R/R MM, relapsed/refractory multiple myeloma. aAll patients to date received 150 × 106 CAR+ T cells; an intermediate dose of 300 × 106 CAR+ T cells will be the next dose level. bHTB defined as ≥50% bone marrow plasma cells pre-infusion; LTB <50%. cPer International Myeloma Working Group criteria.

57 Overall Safety and Tolerability

Grade ≥3 AEs in >1 Patienta • CRSb occurred in (8/12) 67% patients – Mostly grade 1/2; 1 grade 3, no grade 4 Neutropenia 8% 75% – Median time to onset of CRS was 4.5 days Thrombocytopenia 25% 25% (2, 11) Anemia 42% Grade 3 – Managed with or without tocilizumab Grade 4 Leukopenia 17% 25% • Neurotoxicityc occurred in 3/12 (25%) patients

Lymphopenia 8% 17% – 1 patient with high tumor burden and rapidly accelerating disease at baseline experienced Hypophosphatemia 17% DLT (grade 4 encephalopathy and grade 3 0% 20% 40% 60% 80% 100% CRS) – Dosing continued at the 150 × 106 dose split AEs of Special Interesta by tumor burden Grade, n (%) – No other DLTs occurred 1 2 3 4 • 1 grade 3 catheter-related infection; no other CRSb 4 (33) 3 (25) 1 (8) – severe infections reported to date Neurotoxicityc 1 (8) 1 (8) – 1 (8) • 4 patients experienced 1 or more SAEs, no AE, adverse event; DLT, dose-limiting toxicity; SAE, serious adverse event. aAEs occurring between bb21217 infusion and disease progression. bCytokine release syndrome (CRS) uniformly graded according to Lee et al., Blood 2014;124:188-195. c deaths on study to date Events selected as CAR T neurotoxicity on the case report form occurring within 90 days after bb21217 infusion. Data Extract: 18OCT2018 58 Clinical Responses and Duration of Response at the 150 × 106 CAR+ T Cell Dose

Clinical Responses Over Time Clinical Response bb21217-Treated (N=12) ORR,c n (%) [95% CI] 10 (83.3) [51.6, 97.9] a sCR/CR 3 (25) b ≥VGPR 6 (50) b MRD status in bone marrow, n MRD-evaluable respondersd 4 MRD-neg 4e Median time to first response (min, max),c,f 1 (1, 2) mo Median time to best response (min, max),c,f 1 (1, 10) mo Median follow-up duration (min, max), mo 5.9 (1.0, 11.8)

• 10/12 patients (83%) achieved an objective response at the first tested dose (150 × 106 CAR+ T cells) • Responses deepening over time; CR achieved as late as month 10 • Responses ongoing in all but 1 responder; first patient dosed continues in response >1 year after treatment CR, complete response; MRD, minimal residual disease; ORR, objective response rate; PR, partial response; VGPR, very good partial response. *Patients with high tumor burden. • 100% MRD negativity in 4/4 responders evaluable for MRD status; 2/2 aProgression based exclusively on appearance of new bone lesions. bMRD status not available. cIncludes unconfirmed responses. dPatients with ≥PR and valid MRD assessments. eTwo MRD- non-responders were MRD positive neg. responses at 10-6 and 2 at 10-5 sensitivity level by Adaptive next-generation sequencing. f Among 10 responders with ≥PR. Data Extract: 18OCT2018 59 Robust Expansion of Infused CAR+ T Cells Enriched for Memory-like T Cells

CD27+ or CD62L+ CD45RA– CD8+ Cellsa Vector Copy Number Over Time by Baseline Tumor Burden

Wilcoxon P=0.2188; n=6 Wilcoxon P=0.0391; n=8 (n=5) (n=6)

1 0 0 1 0 0

- -

8 0 A 8 0

)

5 4

6 0 c 6 0

8 8

4 0 4 0

6 (

( 2 0 2 0

C C

0 0

C A R + C A R - C A R + C A R -

• Enrichment for memory-like cells within CAR+ cell population in blood post infusion Month 1 Month 3 Month 6 Month 9 • Robust and consistent CAR+ T cell expansion post-infusion independent of tumor burden At risk, n 9 7 3 1 With detectable vector, n (%) 9 (100) 6 (86)b 3 (100) 1 (100) • Detectable CAR + T cells up to 9 months post-infusion HTB, high tumor burden; LLOQ, lower limit of quantitation; LTB, low tumor burden. aImmunophenotyping occurred at time of peak CAR T expansion. bOne patient with undetectable vector received cyclophosphamide on day 15 for grade 4 encephalopathy.

Data Extract: 18OCT2018 60 bb21217: Author’s Conclusions

• bb21217 demonstrated promising early clinical activity in heavily pretreated patients with relapsed/refractory multiple myeloma at first dose level tested – 83% ORR with 90% of responses ongoing – Elimination of MRD in the bone marrow of all 4 evaluable responders • Safety profile appears consistent with known toxicities of CAR T cell therapies • Early clinical evidence of enrichment for memory-like T cells among circulating CAR+ T cells and sustained functional CAR+ T cell persistence • Dose escalation is ongoing • Longer follow-up in a larger patient population will clarify the depth and durability of bb21217 tumor responses and dose response

61 JCARH125: Data from the Phase I/II EVOLVE Trial at ASH

Abstract will be presented on Monday, Dec. 3, 5:00 p.m (Abstract #957)

62 BCMA Pipeline: T Cell Engager and Antibody Drug Conjugate

BCMA T Cell Engager (CC-93269) Novel TCE Format ▪ Unique 2+1 format: 2 high-affinity BCMA binders enhances avidity to BCMA+ cells ▪ Optimized structure supports extended half-life leading to potential dosing once every 1-4 weeks ▪ Phase I trial (MM-001) in RRMM ongoing

BCMA Antibody Drug Conjugate

▪ Sutro’s proprietary site-specific conjugation technology allows for Novel precise and consistent conjugation of the linker and warhead ADC Format ▪ Option to a global license from Sutro Biopharma ▪ IND planned in H1:2019

63 CHANGING THE COURSE OF HUMAN HEALTH THROUGH BOLD PURSUITS IN SCIENCE

Rupert Vessey, MA, BM, BCh,FRCP,DPhil President, Research & Early Development Developing Novel CELMoD® Compounds to Address Unmet Needs in Myeloma

Protein Homeostasis

▪ Novel drugs modulating cereblon to enhance the speed and efficiency of critical substrate degradation ▪ These compounds have the potential to overcome clinical resistance to currently approved agents in myeloma ▪ Two programs in the clinic with novel preclinical profiles – Iberdomide (CC-220) – CC-92480

65 Iberdomide (CC-220) Modulates Cereblon to Induce the Degradation of Aiolos and Ikaros More Efficiently Than Lenalidomide or Pomalidomide*

CRBN:CC-220 Cereblon Binding Affinity

100

80 Pom

60 Len

PoC PoC 40

20 CC-220

0 0.001 0.01 0.1 1 10 100 Compound Conc. (mM) Conc, nM Matyskiela et al J Med Chem. 2017

* In preclinical models 66 Iberdomide Induces Degradation of Aiolos and Ikaros in Pomalidomide Refractory Patients

CC-220-MM-001 Ph Ib/IIa Study Design PD effects in POM refractory BM samples Heavily Pre-Treated RRMM Population Nuclear Aiolos Nuclear Ikaros Prior LEN or POM, PI and anti-CD38 Nuclear Aiolos Nuclear Ikaros

All patients must be refractory to the last prior line 300 300

Part 1: Dose Escalation Part 2: Dose Expansion

200 200

S -

IBER IBER + IBER + -

Score Score

- -

100 H H mono DEX DEX H 100

21/28d 21/28d 21/28d 0 0 g 5 g 5 in 1 n 1 n D i D e 2 n 2 e C e C r re ▪ Preclinical hypothesis is being validated c c S S ▪ Initial data show a compelling clinical profile in Len/Pom refractory patients Unpublished data from CC-220-MM-001 ▪ Data expected at a major medical meeting in 2019

67 Iberdomide: ASH Data Support Updates to Phase I/II Trial

Synergistic Activity with Dara and Bortezomib in Myeloma Cell Lines Iberdomide MM-001: Updated Study Design H929 Effect on CDC of H929 Cells of H929 Cells Vehicle

100

e 0.1mg/mL Dara n

i Triplet Cohorts Added Based on Early Safety and Efficacy

i i

t 1mg/mL Dara

Incorporation o

h 10 mg/mL Dara

r -

) 50

o H

c Part 1: Dose Escalation Part 2: Dose Expansion

(

Thymidine

- H)

3 0 ( 0 1 .1 1 .0 0 0 CC-220Iberdomide Concentration (mM) (nM) IBER + IBER + IBER + IBER + DARA + BTZ + DARA + BTZ + DEX DEX MM1.S Proliferation DEX DEX

ClinicalTrials.gov Identifier: NCT02773030.

Thymidine Incorporation

3 (

Amantangelo, M. et al. ASH 2018. Abstract 1935. 68 CC-92480: Substrate-Selective CELMoD® for Multiple Myeloma

CC-92480-MM-001 First-in-Human Study Design RRMM Population Prior exposure to IMiDs and PIs Progression on or within 60 days of last MM therapy

“More Continuous” Schedule “More Intermittent” Schedule CC-92480 + dex CC-92480 + dex

MTD MTD Cohort Expansion (Part 2) DL2 DL2 DL1 DL1

CC-92480 Program Objectives

▪ Demonstrate activity in Len/Pom/PI/anti-CD38 relapsed/refractory patients ▪ Demonstrate differentiated PD and MM cytotoxicity as proof-of-concept ▪ Planning to evaluate additional patient populations

69 Broad Emerging Early Stage Portfolio Expected to Contribute to Diversification and Growth with Potential Approvals as Early as 2022

CC-220 CC-90002 CC-95251 CC-90010 CC-90011 CELMoD 8 programsBCMA ADC with keyanti -dataCD47 supportingSIRPa advancementBETi LSD1 to NHL RRMMmid-to-late stage development expected in 12-18 months Etigilimab* GEMoab* CC-90010 CELMoD Anti-TIGIT CD3xPSCA Multiple BET inhibitor Lymphoma Solid NHL bb21217 Myeloma CC-93269 NHL & Leukemia Tumors BCMA CAR T BCMA TCE RRMM 6 RRMM bb21217 7 Iberdomide 8 Tislelizumab+ BCMA CAR T CELMoD®AG-270 Motolimod Liso-cel AG-270 Mat2A PD-1 + TLR8 agonist CD19 CAR T Mat2A CC-92480 JCARH125** CLL NHL CELMoD® BCMA CAR T MSC-1* CC-95251 RRMM RRMM TRPH-222* anti-LIF SIRPa JCARH125 CD22 ADC CC-90009 BCMA CAR T CELMoD®

CC-90009 CC-90011 CC-90006 CC-99677 CELMoD CC-93269LSD1 TAPA CC-90002 MK2 RRAML AML MyeloidBCMA TCE CD47Inflammation mAb Disease & Immunology 4 3

CC-92480GEM333 CC-90011 FT-1101 CELMoD ®CD3xCD33 LSD1 inhibitor BETi AML

TCE- T cell engager CC-92252 IL-2 mutein *Celgene has an exclusive option to license and/or option to acquire: TRPH-222 , CD3xPSCA., Etigilimab, and MSC-1; **JCARH125- Program managed by Juno Therapeutics, a Celgene Company; IND expected for CELMoD BCM and BCMA ADC in H1:2019 70 CHANGING THE COURSE OF HUMAN HEALTH THROUGH BOLD PURSUITS IN SCIENCE

Nadim Ahmed President, Hematology & Oncology

71 Strengthening Leadership in Multiple Myeloma & Myeloid Diseases While Expanding into Lymphoma/CLL and Myelofibrosis

Multiple Myeloid Myeloma Diseases ▪ bb2121 ▪ Luspatercept ▪ bb21217 ▪ CC-486 ▪ JCARH125* ▪ CC-90009 ▪ Iberdomide (CC-220) ▪ GEM333 ▪ CC-92480 ▪ CC-92369

Lymphoma/ Myelofibrosis CLL ▪ Fedratinib ▪ Luspatercept ▪ Liso-cel ▪ CC-90002 ▪ CC-90010

*JCARH125- managed by Juno Therapeutics, a Celgene Company 72 Comprehensive Myeloma Pipeline Focused on Two Major Campaigns: BCMA & CELMoD® Agents BCMA CELMoD® Agents

A disruptive approach to myeloma Developing Novel CELMoD® Compounds to

▪ Bb2121 CAR T – KarMMaTM pivotal trial fully accrued Address Unmet Needs in Myeloma: ▪ CC-220 (iberdomide): ▪ Bb21217 CAR T – Phase I data presented at ASH 2018 ▪ Pre-clinical data presented at ASH 2018 ▪ JCARH125 CAR T – Phase I data presented at ASH 2018 ▪ Phase I/II study ongoing in RRMM ▪ CC-93269 TCE – Phase I ongoing ▪ CC-92480 – Phase I study ongoing ▪ BCMA ADC – IND planned H1:19

73 Advancing bb2121 into Earlier Lines of Multiple Myeloma

Phase I Phase II Pivotal

Phase II studies planned in front-line setting In planning for 2019

Phase II study planned in ™ 2nd line setting KarMMa 2 (MM-002)

Phase III study in ™

3rd line+ initiated KarMMa 3 (MM-003) Multiple Multiple Myeloma

Pivotal trial in late line fully ™ accrued KarMMa (MM-001)

74 Aligning Novel Investigational Therapies with Current and Future Multiple Myeloma Patient Segments

Transplant Eligible Non-Transplant Eligible

BCMA CAR T BCMA BCMA TCE CAR T Standard Risk High Risk Standard Risk High Risk NDMM CELMoDs CELMoDs BCMA TCE Elderly Frail

CELMoDs High Risk Standard Risk (1 to 3 Prior Therapies) (1 to 3 prior therapies) BCMA BCMA

CAR T TCE CELMoDs Early Relapse Early

Triple Class (or Penta) BCMA ADC CELMoDs Refractory BCMA BCMA Refractory CAR T (IMiD agents, PIs & CELMoDs

anti-CD38) Late Relapse Late

Higher Unmet Need Lower Unmet Need

75 Luspatercept: A Potential First-in-Class Erythroid Maturation Agent

MEDALIST™:

▪ High unmet need – lower risk, Phase I Phase II Phase III Program Status transfusion dependent MDS patients have poorer outcomes MEDALIST (ESA refractory) ▪ Luspatercept resulted in clinically ▪ Regulatory filings for MEDALIST™ meaningful reductions in RBC MDS ™ transfusion burden vs. placebo in RS+ COMMANDS™ (front-line MDS) & BELIEVE expected H1:19 lower risk MDS patients requiring RBC transfusions ▪ COMMANDS™ trial initiated Q3:18 BELIEVE™ (Transfusion-dependent Beta-Thal) ▪ BEYOND trial currently enrolling BEYOND (NTD Beta-Thal) BELIEVE™: Thal Beta

▪ High unmet need - β-thalassemia ▪ Phase II MF trial currently patients receive chronic transfusions MF-001 MF enrolling; Data expected H2:19 leading to iron overload and associated morbidities ▪ Luspatercept resulted in clinically meaningful reductions in RBC Additional indications involving ineffective erythropoiesis being transfusion burden vs. placebo in actively considered adult β-thalassemia patients requiring RBC transfusions

U.S. approval in MDS and Beta Thalassemia expected in 2020

MDS= myelodysplastic syndrome; β thal = beta-thalassemia; MF= myelofibrosis 76 77 77 Building Leadership in a New Disease: Myelofibrosis (MF)

Fedratinib – A Potential Blockbuster in MF ▪ Selective JAK2 kinase inhibitor ▪ Myelofibrosis clinical program completed to date: − Ph III trial in treatment-naïve patients − Ph II trial in patients resistant or intolerant to ruxolitinib ▪ Myelofibrosis program to start recruitment Q4 2018/Q1 2019: − FREEDOM: Single-arm trial in patients resistant or intolerant to ruxolitinib - US and Canada − FREEDOM 2: Ph III trial in patients resistant or intolerant to ruxolitinib - ex.US ▪ NDA submission for myelofibrosis on track for year-end 2018 ▪ Ongoing myelofibrosis program with luspatercept, in addition to early pipeline compounds

Addressing Unmet Needs in MF

~30% First-Line ruxolitinib, well-controlled ~40% ~20% First-Line ruxolitinib, not well-controlled (low dose/low platelets) ~30% Ruxolitinib failures

78 Expanding NHL/CLL Opportunity with Multiple Mechanisms

Key Mechanisms Diffuse Large B-cell Chronic Lymphocytic Follicular Lymphoma IMiD® Lymphoma Leukemia ABC UC GCB REVLIMID® ▪ AUGMENT™ positive data presented ASH 2018 ▪ ROBUST® data (event-driven) REVLIMID®

ROBUST®

Line - CD19 CAR T

CELMoD® Agents Front Liso-cel ▪ Potential best-in-class profile ▪ Pivotal r/r DLBCL program under way ▪ Encouraging CLL activity SCT SCT Relapsed / Refractory Eligible Ineligible ® CELMoD Agents ® Relapse Liso-cel REVLIMID

st AUGMENT™ 1 ▪ R/R DLBCL and CLL CELMoD® Agents ▪ Additional patient segments planned ▪ Next-gen CELMoD® IND submission imminent Liso-cel Anti-CD47 CELMoD® Agents Liso-cel

Relapse+ CC-90002 nd CC-90002 2 (Anti-CD47) ▪ R/R DLBCL Note: REVLIMID® is not approved for follicular or diffuse Large B-cell non-Hodgkin lymphoma 79 Liso-cel Results in Relapsed/Refractory CLL in Perspective

Phase III MURANO Studya Phase I/II TRANSCEND CLL

8.2% 3.6% CR 43.8% PR 84% 68.7% 37.5%

Bendamustine-Rituximab Venetoclax-Rituximab Liso-cel

(N = 195) (N = 194) (N = 16)

▪ Median number of prior treatments = 1 ▪ Median number of prior treatments = 4.5 ▪ Prior BTK inhibitor exposure: ▪ Prior ibrutinib = 100% 1.5-2.5% of patients ▪ Prior ibrutinib and venetoclax = 50%

aSeymour et al. N Engl J Med 2018; 378:1107-1120

Data presented to show potential profile of JCAR-017, which is subject to ongoing investigation in R/R CLL, within context of RR/CLL treatment landscape. Because clinical trials are conducted under widely varying conditions, response rates in different clinical trials cannot be directly compared 80 Four New Hematology Products Expected to Launch Through 2020

– Selective JAK2 inhibitor for myelofibrosis Fedratinib – U.S. NDA submission expected by YE18 – EU MAA submission planned in 2019

– Encouraging data from Ph I CLL trial presented at ASH 2018 Liso-cel – Pivotal study in CLL planned – U.S. approval expected in relapsed/refractory DLBCL in mid-2020

– First-in-class erythroid maturation agent Luspatercept – Potential new standard of care for MDS and Beta-thalassemia patients – U.S. and EU regulatory submissions expected in H1:19

– Clinical program in earlier lines advancing, including front-line patient segments bb2121 – KarMMaTM enrollment complete – U.S. approval expected in relapsed/refractory MM in 2020

81 Panel Discussion and Q&A MODERATOR David Elkins Executive Vice President & Chief Financial Officer PANEL Mark Alles Chairman and Chief Executive Officer

Jay Backstrom, MD Chief Medical Officer Kristen Hege, MD CVP, Translational Development Rupert Vessey, MA, BM, BCh, FRCP, DPhil. President, Research and Early Development Nadim Ahmed President, Hematology & Oncology Mark Gilbert, MD Chief Medical Officer- Juno Therapeutics, a Celgene company 82 Advancing Innovative Pipeline to Offset Revlimid Sales Erosion