Slide 1: Empowering people, informing care decisions
Empowering people, informing care decisions March 23, 2026 ©2026 Castle Biosciences 1
Slide 2: Disclaimers
Disclaimers Forward Looking Statements This presentation contains forward-looking statements within the meaning of Section 27A of the Securities Act of 1933, as amended, and Section 21E of the Securities Exchange Act of 1934, as amended, which are subject to the “safe harbor” created by those sectio
Slide 3: Speakers/Agenda
Speakers/Agenda J. Michael Guenther, M.D., Agenda Surgical Oncologist St. Elizabeth Physicians, Dr. Goldberg: Edgewood, Kentucky • Early- stage melanoma decision making • DecisionDx-Melanoma: Study Co-author • Answers two important clinical questions • Prospective validation data • DECIDE study intr
Slide 4: Dr. Matthew Goldberg
Dr. Matthew Goldberg ©2026 Castle Biosciences 4
Slide 5: Traditionally, staging and clinicopathology factors answer two key treatment questions
Traditionally, staging and clinicopathology factors answer two key treatment questions following diagnosis of cutaneous melanoma 1 Guideline-based treatment plans include: What is my • Wide local excision (WLE) with patient's risk of Assumed appropriate margin • Low frequency clinical follow-up, sen
Slide 6: DecisionDx-Melanoma uses tumor biology to assess a patient's risk
DecisionDx-Melanoma uses tumor biology to assess a patient’s risk GEP Score Class Result 0.00 Class 1A Individual risk of SLNB Individual risk Lowest risk positivity of recurrence Patient with of recurrence and/or metastasis within 5 years Stage I-III Melanoma Quantifies the expression Class 1B/2A Ul
Slide 7: DecisionDx-Melanoma demonstrates validated risk stratification
DecisionDx-Melanoma demonstrates validated risk stratification Gastman, 2019 - MSS Greenhaw, 2018 - MSS Hsueh, 2021 - OS Survival probability (%) Survival probability (%) Survival probability (%) 1.0 1.0 1.0 0.8 Class 1A 0.8 0.8 Class 1B Class 2A Class 1 Class 1 Class 2B Class 2 Class 2 0.6 N = 690;
Slide 8: Patients tested show improved survival compared to those not tested
Patients who receive DecisionDx-Melanoma testing show improved survival compared to those not tested 32% Melanoma-specific Hazard ratio Lower 3-year melanoma-specific P-value survival (95% CI) mortality rate in patients clinically tested vs untested Untested, 2016-2019 Reference -- 31-GEP tested, 2016
Slide 9: Current guidelines for SLNB patient selection
Current guidelines for SLNB patient selection Stage SLN+Risk SLNB Eligible Guidelines† recommend that the SLNB T1a <5% No procedure can be avoided for patients with an expected risk of <5% T1a-HR* Discuss and 5-10% T1b Consider T2a To safely forego the SLNB procedure, any test must demonstrate an ab
Slide 10: DecisionDx-Melanoma i31-SLNB result can accurately stratify risk
DecisionDx-Melanoma i31-SLNB result can accurately stratify risk Key clinical objectives evaluated in the DECIDE study: 1. Prospectively confirm the performance of the i31-SLNB result in predicting sentinel lymph node (SLN) positivity 2. Evaluate real-world use of the i31-SLNB result to guide SLNB de
Slide 11: Dr. J Michael Guenther
Dr. J Michael Guenther ©2026 Castle Biosciences 11
Slide 12: Review of publication data from DECIDE
Review of publication data from DECIDE: a prospective, multicenter clinical utility study Beard T, Guenther JM, Leong SP, et al. The integrated 31-gene expression profile test identifies low-risk patients with cutaneous melanoma who can forego the SLNB procedure: results from a prospective, multicente
Slide 13: DecisionDx-Melanoma i31-SLNB result accurately stratifies risk
DecisionDx-Melanoma i31-SLNB result accurately stratifies risk In a prospective, multicenter study, the test accurately predicts both SLN positivity and identifies low-risk patients who can forgo an SLNB Key clinical objectives evaluated in the DECIDE study: 1. Prospectively confirm the performance of
Slide 14: Patient demographics: n=912
Patient demographics: n=912 All Patients (n=912) All Patients (n=912) Age, years, median (range) 65 (20-90) Ulceration present Breslow thickness, mm, median (range) 0.8 (0.1-12) Absent 814 (89.3%) Transected base, n (%) Absenta 526 (57.7%) Present 82 (9.0%) Yes 386 (42.3%) SLN status assessed, n (%)
Slide 15: DECIDE Study Results: Predicting SLNB Positivity
DECIDE Study Results: Predicting SLNB Positivity A subset analysis of Stage IB (T1b-T2a) DecisionDx-Melanoma (i31-SLNB) A subset analysis of patients with Beard et al. Future Oncology 2026 Stage IB disease demonstrates Patient Population SLN Positivity %* that DecisionDx-Melanoma can identify…. Low
Slide 16: Comparing two prospective SLNB trials (DECIDE vs MERLIN_001)
Comparing two prospective SLNB trials (DECIDE vs MERLIN_001) A subset analysis of Stage IB (T1b-T2a) DecisionDx-Melanoma CP-GEP (i31-SLNB) Heiken et al., JAMA Beard et al. SSO Annual Conference 2026. Surg. 2025 Patient Population SLN Positivity %* SLN Positivity % A subset analysis of patients with
Slide 17: DECIDE study showed up to 13-fold difference in predicting SLNB positivity rates
DECIDE study showed up to 13-fold di erence in predicting SLNB positivity rates between lowest and highest categories SLN positivity rates by i31-SLNB result Fold Di erence <5% 5-10% >10% <5% vs. >10% Patients with >10% predicted T1-T4 risk had 8-13x 2.6% 7.0% 21.4% 8x higher % SLN+ higher SLN T1b-T
Slide 18: DecisionDx-Melanoma can identify patients with T1b-T2a tumors who can avoid a SLNB
DecisionDx-Melanoma can identify patients with T1b-T2a tumors who can avoid a SLNB The i31-SLNB result provided superior risk stratification because… • It can more precisely identify truly high- risk patients • Patients with an i31-SLNB score >10% have a markedly higher likelihood of SLN positivity.
Slide 19: DecisionDx-Melanoma can accurately predict SLN positivity in T1b-T2a patients
DecisionDx-Melanoma can accurately predict SLN positivity in T1b-T2a patients SLN positivity in T1b-T2a melanoma by i31-SLNB risk group 20 18 16 18.5% 9x higher SLN positivity 14 12 13.2x Di erence ~13x higher SLN positivity In T1–T2a melanoma, patients with <5% 10 in T1b-T2a i31-SLNB patients predi
Slide 20: i31-SLNB can accurately identify low-risk patients who can forgo SLNB
DecisionDx-Melanoma i31-SLNB can accurately identify low-risk patients who can forgo SLNB than AJCC staging alone False-negative Risk stratification method TN:FN ratio Rate NCCN SLNB Guidance 5% 19:1 Beard et al. is consistent with prior Previously reported combined studies (Prieto et al.) studies de
Slide 21: Patients with a low risk DecisionDx-Melanoma result can forego an SLNB
Patients with a low risk DecisionDx-Melanoma result can forego an SLNB • Patients with a low-risk i31-SLNB (<5%) have a low risk of recurrence Recurrence-Free Survival % • n=139* 3-year Recurrence Free Survival (RFS) = 97.8% • Patients with a high-risk i31-SLNB (≥5%) have a higher risk of recurrence
Slide 22: Identifying High-Risk Patients in Traditionally Low-Risk Cohorts
Identifying High-Risk Patients in Traditionally Low-Risk Cohorts CASTLE BIOSCIENCES
Slide 23: DecisionDx-Melanoma vs CP-GEP: Performance in T1a tumors
DecisionDx-Melanoma vs CP-GEP: Performance in T1a tumors Comparing the two largest studies including T1a tumors, DecisionDx-Melanoma identified high-risk patients among T1a tumors and CP-GEP did not. CP-GEP DecisionDx-Melanoma DecisionDx- Melanoma Amaral et al. EJC 2025 Joshi et al. JAAD 2025 identifi
Slide 24: Case Study: Class 2B
CASE STUDY 45-year-old male presented with invasive melanoma FINDINGS IMPACT TO PATIENT CARE • Breslow depth: 0.4 mm After DecisionDx-Melanoma results • Ulceration: Present • DecisionDx-Melanoma returned Class 2B (GEP • Regression: Present score 0.91), indicating the highest 5-year biologic • Mitoti
Slide 25: DecisionDx-Melanoma provides significant risk-stratification to inform management decisions
DecisionDx-Melanoma provides significant risk-stratification to inform management decisions Prognostic significance of Distribution of results result (5-year RFS) 0.00 Class 1A Lowest risk 66% 96% of metastasis within 5 years Class 1B/2A Increased risk 22% of recurrence and/or Stage I-IIA metastasis
Slide 26: Prospective data confirm i31-SLNB result improves SLNB decision-making
Prospective data confirm that DecisionDx- Melanoma i31-SLNB result improves SLNB decision- making and supports favorable patient outcomes 1 2 3 Accurate risk Confidence in clinical Favorable patient stratification decision-making outcomes Low-risk test results The i31-SLNB gives Patients with a <5% are
Slide 27: Q&A
Q&A Dr. Guenther Dr. Goldberg Derek Maetzold, Castle Biosciences, CEO Frank Stokes, Castle Biosciences, CFO ©2026 Castle Biosciences 27
Slide 28: Thank You
Thank You ©2026 Castle Biosciences 28
Slide 29: DecisionDx-Melanoma ~$540M
DER MA TO LO GY DecisionDx-Melanoma Provides comprehensive, personalized, genomic tumor information to guide management for patients with cutaneous melanoma Clinical Validity, Utility and SLNB Guidance and Demonstrated Patient Patient Outcomes1,2 Outcomes DecisionDx-Melanoma Demonstrated clinical va