Pralsetinib (BLU-667) demonstrates robust activity in 2057 QR Code Disclaimer Copies of this poster obtained through QR (Quick Response) code are for personal use onlyand may not be reproduced without RET-fusion-driven intracranial tumor models written permission of the authors. Erica K. Evans, Wei Hu, Fong Cao, Klaus Hoeflich, and Marion Dorsch Blueprint Medicines, Cambridge, MA, USA

RET kinase is oncogenic in diverse cancer subtypes Pralsetinib demonstrated strong anti-tumor Pralsetinib demonstrated anti-tumor activity Pralsetinib was active against intracranial RET 10 Oncogenic alterations in RET, a tyrosine kinase receptor, cause ligand-independent activity across -fusion in vivo models and strong target engagement in an intracranial metastases in the clinical setting kinase activation across a wide range of cancers, driving tumor formation and growth.1 CCDC6-RET-driven tumor model RET fusions are implicated as the underlying cause of disease in ~1–2% of patients 2,3 4 Pralsetinib Extends Survival in Mice With with NSCLC. Patients with NSCLC have an estimated 5-year survival rate of 18%. Ba/F3 Model 1800 Ba/F3 Model CCDC6-RET PDX Tumors Implanted Intracranially ) ) 2400 KIF5B-RET KIF5B-RET V804L 3 3 1600 Brain metastases, which portend a particularly poor prognosis, occur in about 40% of Vehicle BID Vehicle BID 1400 2000 60 mg/kg 60 mg/kg 100 5 cabozantinib QD 1200 cabozantinib QD these patients. 1600 ** 3 mg/kg 1000 3 mg/kg pralsetinib BID pralsetinib BID 80 RET Fusions 1200 800 10 mg/kg 10 mg/kg 800 *** pralsetinib BID 600 pralsetinib BID 30 mg/kg 400 30 mg/kg 60 Kinase 400 Tumor Volume (mm RET pralsetinib BID Tumor Volume (mm 200 *** pralsetinib BID 0 *** 0 *** Vehicle + *** 40 0 2 4 6 8 10 12 14 0 2 4 6 8 10 12 14 10 mg/kg pralsetinib BID Baseline Cycle 3, Day 1 Baseline Cycle 3, Day 1 Dimerization Days After Start of Treatment Days After Start of Treatment Survival, % Domain Fusion Partner 20 30 mg/kg pralsetinib BID • 52-year-old woman, RET-fusion-positive NSCLC, prior platinum • 59-year-old man, RET-fusion-positive NSCLC, prior platinum and checkpoint inhibitor and checkpoint inhibitor 0 • Near-complete resolution of previously untreated target brain • Complete resolution of previously untreated nontarget brain Dimerization NSCLC PDX Model 1000 Colorectal Cancer PDX Model metastasis after 2 months of pralsetinib 400 mg QD metastasis after 2 months of pralsetinib 400 mg QD ) ) KIF5B-RET Vehicle BID CCDC6-RET Vehicle BID 0 10 20 30 40 50 60 70 80 90 3 Domain Kinase RET Fusion 3 2000 • Continues to receive treatment with ongoing confirmed PR 60 mg/kg 60 mg/kg • Continues to receive treatment with ongoing confirmed 800 Treatment Days cabozantinib QD cabozantinib QD (70% shrinkage) at 10+ months (data cut-off 16 Aug 19) 1600 CR at 10+ months (data cut-off 16 Aug 19) 3 mg/kg 3 mg/kg Images courtesy of Dr. Stephen Liu, Georgetown University, Washington, DC Images courtesy of Dr. P Cassier Centre, Leon Berard, Lyon, France pralsetinib BID 600 pralsetinib BID 1200 ** 10 mg/kg 10 mg/kg pralsetinib BID 400 pralsetinib BID RET Mutations 800 RET Pathway Vehicle 10 mg/kg BID 30 mg/kg BID 30 mg/kg 30 mg/kg 4h 4h pralsetinib BID pralsetinib BID pRET 400 200 Tumor Volume (mm *** Tumor Volume (mm 60 mg/kg 60 mg/kg Pralsetinib demonstrated anti-tumor pralsetinib QD pralsetinib QD pSHC *** Extracellular 0 0 pSHC 0 4 8 12 16 20 24 28 0 2 4 6 8 10 12 14 16 18 20 10 Domain activity in patients with CNS involvement Days After Start of Treatment Days After Start of Treatment Mutations tSHC

V804L/M M918T **P<0.01; ***P<0.001. pERK 40 DUSP6, GAPDH Pralsetinib Starting Dose 400 mg QDa Kinase Oral administration of pralsetinib demonstrated dose-dependent anti-tumor activity in RET-fusion-driven models, SPRY4 20 including a Ba/F3 allograft model expressing either a KIF5B-RET fusion or a KIF5B-RET(V804L) gatekeeper mutant. expression Cabozantinib was administered at the MTD in mice (60 mg/kg QD) and showed anti-tumor activity similar to pralsetinib in models with a WT RET-kinase domain, but not in models driven by gatekeeper mutant RET. Pralsetinib demonstrated 0 anti-tumor activity in an NSCLC KIF5B-RET fusion-driven model, as well as a CCDC6-RET fusion-positive colorectal Gene Expression Levels in Tumor From Brain Frequency of Oncogenic RET Alterations in Solid Tumors 2.0 cancer PDX. Pralsetinib was well tolerated in all models at all doses tested. -20 Indication RET Alteration Frequency Vehicle 1.5 10 mg/kg pralsetinib 4h -40 2,3 NSCLC Fusions ~1–2% 30 mg/kg pralsetinib 4h Lesions, % 1.0 10 mg/kg pralsetinib 12h -60 Advanced MTC6 Activating mutations ~90% Pralsetinib showed significant anti-tumor activity 30 mg/kg pralsetinib 12h 58% ORR (confirmed responses) in RET-fusion-positive NSCLC patients, 60% ORR (confirmed responses) in patients previously 0.5 -80 PTC7 Fusions ~20% Relative Gene treated with platinum chemotherapy

KIF5b-RET in Sum of Diameters Target No CNS involvement in -driven intracranial tumors Expression Level

Maximum Change From Baseline 71% (5/7) treatment-naïve patients had confirmed PR 8,9 CNS involvement Colon, breast, other tumor types Fusions <1% 0.0 -100 BA/F3-KIF5b-RET DUSP6 SPRY4 GSK3B aPatients enrolled by 14 Nov 18, with a data cut-off of 28 Apr 19. The response-evaluable population included patients with measurable disease at baseline and ≥1 evaluable Brain Implantation Model post-treatment disease assessment and excluded four patients who previously received >1 cycle of a selective RET inhibitor. All responses are confirmed on 2 consecutive Anti-tumor activity: Each mouse was intracranially injected with CCDC6-RET PDX cells. After tumor inoculation, body assessments as per RECIST 1.1. 100 weight was measured twice per week. The median survival for vehicle-treated mice was 31.5 days. Six of 10 mice Adverse Events Treatment-Emergent AE Treatment-Related AE Pralsetinib selectively inhibits Vehicle treated with 10 mg/kg pralsetinib BID survived through 95 days; all 10 mice treated with 30 mg/kg pralsetinib BID (Pralsetinib starting dose 400 mg) All (≥15% overall) Grade ≥3 All Grade ≥3 80 3 mg/kg pralsetinib PO BID survived through 95 days (both P<0.001). Constipation 30% 2% 17% 2% 10 mg/kg pralsetinib PO BID Neutropeniaa 26% 13% 26% 13% oncogenic RET fusions and gatekeeper mutants 60 30 mg/kg pralsetinib PO BID Intracranial tumor target engagement: Each mouse was intracranially injected with CCDC6-RET PDX cells. Enrollment was performed on a rolling basis, with body weight used as an early marker for tumor progression. Therapy was initiated AST increased 24% 5% 20% 2% Kinome Selectivity for RET Fatigue 21% 3% 13% 3% 40 immediately upon enrollment. Enrollment continued until there were ≥4 animals per group. Immunoblot demonstrated Survival, % high levels of pSHC inhibition after dosing with 10 or 30 mg/kg pralsetinib. Human DUSP6/SPRY4 transcripts decreased Hypertension 20% 13% 13% 10% Biochemical Anemia 18% 7% 11% 4% Variant 20 >90% with 10 or 30 mg/kg pralsetinib at 4h and indicated full pathway inhibition at doses demonstrating anti-tumor Diarrhea 18% 2% 9% - IC50 (nM) activity. Negative control, human GSK3B expression was not inhibited with pralsetinib treatment. Pyrexia 18% - 2% - RET wild-type 0.4 0 0 4 8 12 16 20 24 28 ALT increased 17% 3% 13% 2% RET V804L 0.3 Days Cough 17% - 3% - Dry mouth 17% - 12% - RET V804M 0.4 Ba/F3-KIF5b-RET Pralsetinib is being evaluated in a clinical trial Additional grade ≥3 treatment-related AEs (≥2%): increased CPK (3%), leukopeniab (3%). Luciferase Brain Implantation Model Across the entire study (n=276), the rate of discontinuation due to treatment-related toxicity was 4%. RET M918T 0.4 Day 0 Day 14 for patients with RET-altered tumors aCombined term including decreased neutrophils and neutropenia. bCombined term including leukopenia and white blood cell count decreased. CCDC6-RET 0.4 Vehicle Part 2 Ongoing CONCLUSIONS

Cellular RET-fusion NSCLC, treatment naïve RET Cell Lines • Pralsetinib has broad anti-tumor activity in intracranial tumor models IC (nM) 3 mg/kg 50 BID regardless of RET-fusion partner Ba/F3-KIF5b-RET 10.1 RET-fusion NSCLC, prior platinum Ba/F3-KIF5b-RET V804L 8.1 • Pralsetinib showed broad, durable anti-tumor activity in patients with RET-fusion NSCLC, both systemically and in the brain Ba/F3-KIF5b-RET V804M 14.1 Kinome illustration reproduced 10 mg/kg Ba/F3-KIF5b-RET V804E 8.1 courtesy of Cell Signaling Technology, BID MTC, prior cabozantinib and/or vandetanib Inc. (www.cellsignal.com).a • ARROW clinical trial enrollment continues in treatment naïve LC2/ad (CCDC6-RET) 3.7 400 mg RET-fusion-positive NSCLC (NCT03037385) TPC-1 (CCDC6-RET) 10.9 QD identified as MTC, no prior cabozantinib or vandetanib MZ-CRC (RET M918T) 4.2 30 mg/kg RP2D in dose TT (RET C634W) 15.4 BID escalation Other RET-fusion solid tumors aThe foregoing website is maintained by Cell Signaling Technology Inc., and Blueprint Medicines is not References responsible for its content. Acknowledgments Third-party editorial assistance was provided by Meredith Kalish, PhD, of 1. Subbiah V, et al. Cancer Discov. 2018;8(7):836–849. Pralsetinib was crafted to selectively target oncogenic RET fusions and activating mutations. Pralsetinib displays Ba/F3-KIF5B-RET cells engineered to express luciferase were injected intracranially into mice. Image analysis confirmed 2. Lipson D, et al. Nat Med. 2012;18(3):382-384. RET-altered solid tumors, prior selective RET TKI TM subnanomolar biochemical inhibitory activity across activated RET kinase fusions and mutations and low nanomolar intracranial tumor growth and luciferase detection. Imaging occurred periodically throughout the experiment to measure Ashfield Healthcare Communications and was funded by Blueprint Medicines 3. Takeuchi K, et al. Nat Med. 2012;18(3):378-381. 4. Reck M, et al. N Engl J Med. 2017;377(9):849–861. anti-proliferative activity against RET-fusion or mutant-driven cell lines. When screened against a panel of human disease. The health of vehicle-treated mice declined starting at day 14, with all animals requiring euthanasia by day 18. 5. Yawn BP, et al. Minn Med. 2003;86(12):32–37. Treatment with 3 mg/kg pralsetinib BID demonstrated a moderate survival advantage, while mice treated with 10 or 30 mg/kg 6. Romei C, et al. Oncotarget. 2018;9(11):9875-9884. kinases, pralsetinib inhibited RET (large dot) most potently. Those kinases inhibited by pralsetinib within 50x RET IC50 Abbreviations 7. Santoro M, et al. J Clin Invest. 1992;89(5):1517-1522. BID pralsetinib survived through 28 days (3 mg/kg BID, P<0.05; 10 or 30 mg/kg BID, both P<0.001). AE, adverse event; ALT, alanine aminotransferase; AST, aspartate aminotransferase, BID, twice daily; CI, confidence interval; CPK, creatine phosphokinase; CR, complete response; are shown with medium dot and within 100X RET IC50 shown with small dot. Other RET-mutated solid tumors 8. Kato S, et al. Clin Cancer Res. 2017;23(8):1988-1997. CRC, colorectal cancer; DCR, disease control rate (best response of SD or better); h, hour; IC50, half maximal inhibitory concentration; MTC, medullary thyroid cancer; MTD, maximum tolerated dose; NSCLC, non-small cell lung cancer; ORR, overall response rate; PD, progressive disease; PDX, patient-derived xenograft; PO, orally; PR, partial response; 9. Ballerini P, et al. Leukemia. 2012;26(11):2384-2389. PTC, papillary thyroid cancer; QD, once daily; RP2D, recommended phase 2 dose; SD, stable disease; TKI, tyrosine kinase inhibitor; WT, wild-type. 10. Gainor JF, et al. J Clin Oncol. 2019;37(suppl; abstr 9008).

Presented at the International Association for the Study of Lung Cancer 2019 World Conference on Lung Cancer; 7-10 Sept 2019; Barcelona, Spain.