Diagnostic yield and concordance of image-guided biopsy in musculoskeletal lesions.

Abstract

Image-guided biopsy is well-established in the diagnosis of musculoskeletal lesions. While several studies have demonstrated a high diagnostic yield in image-guided biopsy, there are no current guidelines around procedural factors such as number of cores. Furthermore, there have been mixed results regarding which lesions are more favourable to a diagnostic biopsy. We wanted to evaluate diagnostic yield and concordance for image-guided biopsies for musculoskeletal lesions. The null hypothesis was that there are no controllable factors that contribute to positive yield. Retrospective review of consecutive patients who underwent image-guided biopsy at a large teaching hospital for musculoskeletal lesions discussed at the sarcoma multi-disciplinary meeting. The formal biopsy histology report was evaluated, and biopsies were considered diagnostic or non-diagnostic. In those that had subsequent surgery (by wide excision or open biopsy), final and initial histology was compared and biopsies were considered concordant or not. Overall diagnostic yield and concordance were calculated. Statistical analysis was performed with Stata 13.0 (StataCorp). Over the 14-year period, 429 biopsies were included. Diagnostic yield was 85% and concordance was 100%. No cases of malignant lesions were initially called benign on biopsy. One biopsy had a complication (0.2%). Factors associated with higher diagnostic yield included soft tissue versus bone lesions, three or more cores and longer total specimen length. Factors that were not associated included core size, use of FNA cytology, gender, age, benign versus malignant, anatomic location and lesion appearance. The null hypothesis is rejected. The main predictor of diagnostic biopsy was total specimen length, independent of number of cores taken. Three or more cores and longer cores are optimal, though these factors are influenced by lesion biology and not always able to be controlled.