Multisite sampling improves consensus cytologic interpretation in bone marrow evaluation of dogs

Abstract

BACKGROUND: Bone marrow (BM) evaluation is essential for diagnosing hematologic abnormalities in dogs. Standard practice often involves single-site sampling, despite some evidence suggesting that focal BM lesions may occur. We assessed the probability of obtaining an unrepresentative cytologic interpretation when BM cytology interpretation depends on a single site vs multiple sites. HYPOTHESIS/OBJECTIVES: Single-site BM sampling increases the probability of unrepresentative cytologic interpretation in dogs. ANIMALS: Sixteen client-owned dogs with hematologic disease requiring BM sampling were prospectively enrolled. METHODS: Randomized, masked clinical trial. We collected 4 BM aspirates per dog: right and left proximal humeri and iliac crests. Interpretative concordance was assessed by comparing individual site interpretations with a composite multisite interpretation. The primary outcome was the probability of a truth-consistent interpretation based on the number of sites sampled (k = 1-4) under 3 evaluation rules, accounting for inter-clinical pathologist agreement. RESULTS: Single-site sampling yielded a truth-consistent interpretation in only 50.0%-81.7% of cases, depending on diagnostic stringency. Evaluating 2 sites significantly improved the probability of a truth-consistent interpretation by 17.2%-23.3% (P < .002). Additional sites (2 or 4) offered only marginal further benefit. Inter-clinical pathologist agreement was moderate (kappa = 0.574). No single anatomical site provided superior diagnostic yield for overall interpretation. CONCLUSIONS AND CLINICAL IMPORTANCE: A single BM sampling site carries a higher probability of unrepresentative cytologic interpretation in dogs. Two-site sampling significantly improves recovery of a truth-consistent interpretation, providing a more reliable assessment for hematologic disease in dogs. This approach accounts for potential focal disease distribution, improving diagnostic confidence.