PowerPoint Presentation › Diagnostic discordance in cutaneous melanocytic lesions is well documented, and it is particularly prevalent among difficult-to-diagnose cases, for which histopathology may be insufficient for a definitive diagnosis.1-4 › The 23-gene expression profile (GEP) and 35-GEP tests are clinically available, objective ancillary tools that facilitate diagnosis of melanocytic lesions with ambiguous histopathology. The tests use proprietary algorithms to produce results of: suggestive of benign neoplasm; intermediate (cannot rule out malignancy); or suggestive of malignant neoplasm.5-7 › The GEP tests have demonstrated accuracy metrics of 90.4 – 94.9% sensitivity and 92.5 – 96.2% specificity for the 23-GEP, and 94.7 – 99.1% sensitivity and 89.5 – 94.3% specificity for the 35-GEP.5-7 › Today, both the 23- and 35-GEP tests are offered from a single laboratory. Under the current laboratory workflow, unless preferred otherwise by the ordering clinician, clinical samples are processed first through the 23-GEP test, and if a technical failure or intermediate result is received, processed through the 35-GEP (Figure 1). However, both are run independently of one another and can be ordered as stand-alone tests.8 Subtype performance of the ancillary diagnostic 23- and 35-gene expression profile (GEP) tests for difficult-to- diagnose melanocytic lesions Background Acknowledgments & Disclosures References › JAP has served as a consultant for Castle Biosciences, Inc. SIE is a consultant and shareholder of Castle Biosciences, Inc. KLA, JJS, BHR, JHR, JKW, SJK, and MSG are employees and shareholders of Castle Biosciences, Inc. KM has served as a consultant and investigator for studies supported by Castle Biosciences, Inc. This study was supported by Castle Biosciences, Inc. Results Jose A Plaza, MD1, Sarah I Estrada, MD2, Brooke H Russell, PhD3, Jennifer J Siegel, PhD3, Jason H Rogers, MSc3, Jeffrey K Wilkinson, PhD3, Sarah J Kurley, PhD3, Matthew S Goldberg, MD3,4, and Kiran Motaparthi, MD5 1 Depts of Dermatology & Pathology, The Ohio State Univ Wexner Medical Center, Columbus, OH, USA 2 Affiliated Dermatology, Scottsdale, AZ, USA 3 Castle Biosciences, Inc., Friendswood, TX, USA 4 Icahn School of Medicine at Mount Sinai, NY, USA 5 Dept of Dermatology, Univ of Florida College of Medicine, Gainesville, FL, USA For more information: JoseA.Plaza@osumc.edu Methods ›The 23- and 35-GEP diagnostic test workflow results in high accuracy across a large spectrum of subtypes of melanocytic neoplasms Conclusions Table 1. GEP workflow overall performance accuracy metrics Figure 1. Current clinical GEP workflow Presented at the Inaugeral Winter Clinical – Miami™ 2023 Meeting , February 17-20, 2023. Miami, Florida Scan or click here for more info Table 2. GEP workflow test result by lesion subtype (as indicated by submitting dermatopathologist) The performance of the 23- and 35-GEP tests using the clinical workflow was tested on unequivocal cases from a variety of subtypes ›. › Melanocytic lesions and associated de-identified clinical data from patients ≥ 18 years of age were included in this study. Samples were acquired under an IRB-approved protocol, including those previously submitted for clinical testing for the 31-GEP melanoma prognostic test. Samples were independently reviewed (blinded to the original diagnosis) by at least 3 total dermatopathologists for adjudication and included if they received at least 2 out of 3 diagnostic concordance with choices of benign, malignant, or uncertain malignant potential (UMP) (Table 1). Subtype in this analysis was determined by the submitting dermatopathologist. All cases not receiving a benign or malignant result from the 23-GEP were run on the 35-GEP. Final GEP workflow result Subtype* Benign, n Intermediate, n Malignant, n Melanomas (n=245) Acral lentiginous 15 Common 15 Desmoplastic 20 Lentigo maligna 1 30 Melanoma in situ 16 Nodular 4 77 Not specified 1 4 Spitzoid 3 17 Superficial spreading 1 41 Benign nevi (n=100) Blue 28 1 1 Compound 9 3 Compound dysplastic 26A 1 3B Deep penetrating 1 Intradermal 1 1 Junctional dysplastic 13C 1D 4E Spitz 7 Performance Cohort, n=350 95% Confidence interval Sensitivity 96.0% 92.0% – 99.0% Specificity 87.8% 80.8% – 93.8% Positive predictive value 89.0% 83.8% – 94.1% Negative predictive value 95.6% 91.1% – 98.9% Intermediate result 1.5% *5 samples did not have adequate subtype information. Dysplastic nevi had different degrees of atypia: A: 13 mild, 2 moderate; B: 2 mild, 1 moderate; C: 6 mild, 4 moderate; D: 1 moderate; E: 3 mild, 1 moderate. Table 3. GEP workflow performance accuracy metrics by lesion subtype Subtype* n Sensitivity Specificity Melanomas Acral lentiginous 15 100% Common 15 100% Desmoplastic 20 100% Lentigo maligna 31 96.8% Melanoma in situ 16 100% Nodular 81 95.1% Spitzoid 20 85% Superficial spreading 42 97.6% Benign nevi Blue 30 93.3% Compound dysplastic 30 86.7% Junctional dysplastic 18 72.2% *Only subtypes with n ≥ 15 are shown. Results 1. Shoo, B. A. et al. J Am Acad Dermatol 2010. 62 (5) 751–756. 2. Gerami, P. et al. Am J Surg Pathol 2010. 34 (6) 816–821. 3. Haws, B. et al. J Cutan Pathol 2012. 39 (9) 844–849. 4. Elmore, J. G. et al. BMJ 2017. 357 (1) j2813. 5. Clarke, L. E. et al. J Cutan Pathol 2015. 42 (4) 244–252. 6. Clarke, L. E. et al. Cancer 2017. 123 (4) 617–628. 7. Estrada, S. et al. SKIN 2020. 4 (6) 506–522. 8. Goldberg, M. et al. SKIN 2021.5 s79. *Technical fail includes samples with insufficient quantity of RNA and/or control or discriminant gene amplification failure based on the requirements for each test. Clinically, the technical failure rate for the GEP workflow is ~0.2%.8 https://castletestinfo.com/mypath-melanoma-diffdx-melanoma/scientific-references/ https://castletestinfo.com/mypath-melanoma-diffdx-melanoma/scientific-references/ DD_01_036 WC Miami 2023 Eposter dGEP workflow subtypes WIDE v1 Slide 1