FCD.Castle.Restage.092818.4x4.SUBMITTED Improvement of risk assessment in cutaneous melanoma (CM) by a prognostic 31-gene expression profile (31-GEP) test over AJCC-based staging alone Giselle Prado, MD1, Robert W. Cook, PhD2, Kyle R. Covington, PhD2, Federico A. Monzon, MD, FCAP2, Darrell Rigel, MD, FAAD3 1National Society for Cutaneous Medicine, New York, NY; 2Castle Biosciences, Inc., Friendswood, TX; 3New York University School of Medicine, New York, NY BACKGROUND RESULTS RESULTS REFERENCES • A substantial number of melanoma-related deaths occur in patients originally diagnosed with early American Joint Committee on Cancer (AJCC) stage disease, suggesting aggressive tumor biology despite having clinicopathologic features associated with low-risk disease. • A 31-gene expression profile (31-GEP) test has been developed and validated in retrospective and prospective studies1-8 to predict 5-year metastatic risk from primary cutaneous melanoma (CM) tumor tissue with a high degree of technical reliability.9 • The 31-GEP test classifies melanoma as Class 1A (lowest risk), Class 1B (low risk), Class 2A (increased risk), or Class 2B (highest risk). • This prognostic information is used to inform patient management decisions, including frequency of follow-up and surveillance imaging, referrals, sentinel lymph node biopsy guidance, and consideration of adjuvant therapy.10-15 31-GEP Test Class 1 Low Risk (1A lowest) Class 2 High risk (2B highest) Stage I-III melanoma •Archival formalin-fixed paraffin-embedded CM tumor samples from 18 U.S. centers (n=690, Stage I-III) along with clinical, pathological, and outcomes data for each case were collected under an IRB- approved protocol1-4. Stage I-II cases were restaged according to AJCC 8th edition criteria. •The 31-GEP test was performed in a CAP-accredited/CLIA-certified laboratory using high-throughput RT-PCR assays as previously described1-5. •The Kaplan-Meier method was used to estimate 5-year recurrence-free (RFS; time to either a regional or distant metastatic event), distant metastasis-free (DMFS; time to any metastatic event beyond the regional nodal basin), and melanoma-specific survival (MSS; time from diagnosis to death documented as from melanoma) rates with significance determined by log-rank test. All non-recurrent cases had at least 5 years of follow-up. •Class 1A- and 2B-predicted MSS outcomes for each stage were compared to rates associated with AJCC 8th edition stage16. •Based on National Comprehensive Cancer Network (NCCN) guidelines for surveillance and follow-up, AJCC binary low and high-risk groups are defined as Stage I-IIA and Stage IIB-IV, respectively. Cox multivariate regression analysis for MSS was performed comparing AJCC binary risk and 31-GEP test results. METHODS Figure 1. Stage-specific survival rates for the 31-GEP cohort align with the AJCC 8th edition database survival rates AJCC 8th Ed. cohort14 Stage (n) 5-year I (10974) 98% II (4717) 90% III (4622) 77% Years Since Diagnosis M el an om a- S pe ci fic S ur vi va l ( % ) 0 5 10 100 80 60 40 20 0 AJCC 8th ed. cohort1631-GEP cohort with 8th ed. staging 31-GEP cohort AJCC 8th edition16 Earliest diagnosis year 1998 1998 Number of collaborating centers 18 10 Percent of cases from U.S. centers 100% 34% 31-GEP cohort with 8th Ed. staging 0 5 10 Years Since Diagnosis M el an om a- S pe ci fic S ur vi va l ( % ) Stage (n) 5-year I (344) 98.5% II (138) 90.7% III (208) 75.8% 100 80 60 40 20 0 Time (Years) % R e c u rr e n c e F re e RFS p<0.0001 Time (Years) % D is ta n t M e ta s ta s is F re e DMFS p<0.0001 Time (Years) % S u rv iv a l MSS p<0.0001 Class (n) 5-year RFS (95% CI) 5-year DMFS (95% CI) 5-year MSS (95% CI) 1A (312) 90% (87-93%) 94% (91-97%) 99% (98-100%) 1B (80) 81% (73-90%) 85% (77-93%) 95% (90-100%) 2A (84) 68% (58-79%) 75% (66-85%) 91% (85-98%) 2B (214) 37% (31-44%) 50% (43-58%) 75% (69-83%) Class 1A Class 1B Class 2A Class 2B Figure 2. 31-GEP results identify significantly different risk groups4 Figure 3. Addition of 31-GEP test results improves risk obtained by AJCC 8th edition staging alone Stage I Stage II Stage III Class (n) 5-year MSS (95% CI) Event Rate Class (n) 5-year MSS (95% CI) Event Rate Class (n) 5-year MSS (95% CI) Event Rate 1A (249) 99.6% (98.8-100%) 0.4% 1A (21) 100% (100-100%) 0% 1A (42) 94.8% (88.0-100%) 7% 2B (19) 89.5% (76.7-100%) 11% 2B (83) 84.7% (76.3-94.1%) 13% 2B (112) 61.2% (50.1-74.7%) 28% *For surveillance and follow-up purposes, NCCN guidelines propose dividing patients into binary risk groups based on AJCC Stage: low (Stage I-IIA) and high risk (Stage IIB-IV). Cox multivariate regression analysis n=690 58 events MSS HR (95% CI) p value AJCC high risk 4.84 (2.3-10.1) <0.0001 GEP Class 1B 3.38 (0.9-12.8) 0.073 GEP Class 2A 4.67 (1.3-16.4) 0.02 GEP Class 2B 10.1 (3.4-16.4) <0.0001 Stage I II III M el an om a- S pe ci fic S ur vi va l ( % ) Low Risk Stage I-IIA High Risk Stage IIB-III 99.6% ≈IA 89.5% ≈IIA/IIB 100% ≈IA 84.7% ≈IIB/IIC 94.8% ≈IIA 61.2% ≈IIIC+ 98% AJCC Binary Risk Category 100 90 80 70 60 Class 1A Class 2B AJCC MSS 90% 77% CONCLUSIONS • In the study cohort of Stage I-III melanoma cases1-4 with similar survival outcomes to the 8th edition AJCC cohort, the 31-GEP test result was able to add information to further stratify patients with lower and higher risks than predicted by clinicopathologic staging alone. Multivariate analysis demonstrated that a 31-GEP Class 2B result was an independent predictor of MSS with a greater hazard ratio than AJCC binary risk. • As accurate risk assessment is important for patient management decisions, use of the 31-GEP test can help guide these choices, including follow-up, sentinel lymph node biopsy guidance, surveillance and possible adjuvant therapy, as has been previously published10-15. To determine the impact on risk prediction when results from the 31-GEP test are used with AJCC 8th edition staging. OBJECTIVE The authors wish to acknowledge the following collaborating physicians and institutions for their contributions to this study: Drs. Sancy Leachman and John Vetto, Oregon Health and Science University, Drs. Pedram Gerami and Jeff Wayne, Northwestern University, Drs. Jane Messina and Jonathan Zager, Moffitt Cancer Center, Dr. Rene Gonzalez, University of Colorado Cancer Center, Drs. David Lawson, Keith Delman, and Maria Russell, Emory University, Dr. Stephen Lyle, University of Massachusetts Medical School, Dr. Gilchrist Jackson, Kelsey-Seybold Clinic, Dr. Anthony Greisinger, Kelsey Research Foundation, Dr. Lee Cranmer, University of Arizona Cancer Center, Dr. T. Christopher Windham, Florida Hospital Memorial Medical Center, Dr. Lewis Kaminester, Dermatology North Palm Beach, Dr. Martin Fleming, University of Tennessee Health Science Center, Drs. Laura Ferris and Jonhan Ho, University of Pittsburgh Medical Center, Dr. Alexander Miller, START Center for Cancer Care, Dr. Sarah Estrada, Affiliated Dermatology, Dr. Jason Robbins, Pathology Associates, Dr. David Pariser, Pariser Dermatology Specialists, Dr. Brian Gastman, Cleveland Clinic and Dr. Daniel Rosen, Baylor College of Medicine. 1. Gerami P, et al. Clin Cancer Res 2015;21:175-83. 2. Gerami P, et al. J Am Acad Dermatol 2015;72:780-5 e783. 3. Zager JS, et al. BMC Cancer 2018;18(1):130. 4. Gastman BR, et al. JAAD 2018; doi: 10.1016/j.jaad.2018.07.028. 5. Hsueh EC, et al. J Hematol Oncol 2017;10:152. 6. Greenhaw B, et al. Dermatol Surg 2018; 7. Renzetti M, et al. Society of Surgical Oncology Annual Meeting. 2017. 8. Hsueh EC, et al. J Clin Oncol 2016; 34(15_suppl):9565. 9. Cook RW et al. Diagn Pathol. 2018;13(1):13. 10. Svoboda RM et al. J Drugs Dermatol. 2018;17(5):544. 11. Berger AC et al. Curr Res Med Opin 2016;32(9):1599-604. 12. Dillon LD et al. SKIN: J Cut Med 2018;2(2):111-21. 13. Cook RW et al. Winter Clinical Dermatology Conference. 2018. 14. Schuitevoerder D et al. J Drugs Dermatol 2018;17(2):196-199. 15. Farberg AS et al. J Drugs Dermatol 2017;16(5):428-431. 16. Gershenwald et al. CA Cancer J Clin 2017;67(6):472-492. ACKNOWLEDGEMENTS FUNDING & DISCLOSURES This study was sponsored by Castle Biosciences, Inc., which provided funding to contributing centers for tissue and clinical data retrieval. RWC, KRC, & FAM are employees and options holders of Castle Biosciences, Inc. GP is a fellow with the National Society for Cutaneous Medicine which receives funding from Castle Biosciences, Inc. DR is a consultant for Castle Biosciences, Inc.