23 Exploring the Validation of Bone Marrow Biopsies using a Digital Pathology System Approved for FFPE Cut Sections

Introduction Digital pathology (DP) with whole slide imaging (WSI) is a rapidly advancing field with great potential in applications such as telepathology and artificial intelligence (AI). Our laboratory uses a modern and popular high-throughput DP system that is approved for interpretation of formalin-fixed paraffin embedded (FFPE) sections cut on a microtome. Bone marrow (BM) biopsies represent off-label usage of our DP system because a BM case includes cytological (i.e. not cut) components; aspirate and blood smear. We performed an abbreviated mock validation of our DP system with BM biopsies from multiple myeloma (MM) patients using the latest DP validation guidelines from CAP, ASCP, and API to understand the analytical performance of our DP system with BM biopsies. Method Ten MM cases diagnosed within the past year at a single institution were selected and scanned. Each case included a peripheral blood smear, BM aspirate smear, clot, and core biopsy. Cases with secondary hematologic malignancies, additional testing, i.e., immunohistochemistry, or inadequate smears were excluded. De-identified glass and DP slides were assigned unique case numbers and evaluated by three pathologists, who signed them out within one week, with a two-week washout period between glass and digital reviews. Pathologists were provided with the patient’s age, flow cytometry results, two most recent complete blood counts, and prior history. A non-assigned reviewer compared diagnoses between glass and DP slides, categorizing discrepancies as “none,” “minor,” or “major.” Pathologists were given a qualitative evaluation form to rate their experience at the end of the study. Results All 10 cases demonstrated high (100%) diagnostic concordance between glass and digital slides. No major discrepancies impacting clinical management were observed (0/10). Of these, 6/10 cases showed no discrepancies, while 4/10 had minor discrepancies related to reported plasma cell percentages, with an average difference of 4.38% (range: 1–7.5%). The qualitative evaluation survey indicated that pathologists found the cases easy to diagnose using the DP platform (mean score: 1.3) on a 1–5 scale, (1 = strongly disagree that the diagnosis was challenging). Additionally, pathologists expressed strong interest (mean score: 4.3) in incorporating DP into their workflow if readily accessible. Conclusion This limited study demonstrated high concordance between digital and glass slide interpretations of BM cases in MM and suggests that DP systems designed for FFPE cut sections might have utility in the interpretation of BM cases. Future studies should include cases within the original exclusion criteria and should be expanded to the CAP-recommended guidelines of at least 60 cases with a target of > 95% concordance.