CT-Guided Core Needle Biopsy of Nonspinal Bone Lesions: Comparison of Occult and Visible Bone Lesions

BACKGROUND. CT guidance may be used for biopsy of indeterminate bone lesions detected by MRI or PET/CT that are not visible (i.e., occult) on CT owing to equipment-, patient-, and operator-related factors. OBJECTIVE. The purpose of this study was to assess diagnostic yield (DY) and diagnostic performance of CT-guided core needle biopsy (CNB) of occult nonspinal bone lesions and to identify the most common benign and malignant diagnoses for occult lesions undergoing CNB. METHODS. This retrospective study included 1033 adult patients who underwent CT-guided nonspinal bone CNB between January 2004 and December 2020. Lesions were classified as occult or visible on CT; biopsies of occult lesions were performed by targeting anatomic landmarks using prebiopsy MRI or PET/CT. Pathologic results of CNB were classified as diagnostic or nondiagnostic to calculate DY of CNB. For nondiagnostic CNBs, final diagnoses were established by subsequent pathologic, clinical, and imaging follow-up. RESULTS. The sample included 70 patients with occult lesions (mean age, 56.8 years; 38 women, 32 men) and 963 patients with visible lesions (mean age, 59.6 years; 475 women, 488 men). Malignancy rate was lower for occult than for visible lesions (42.9% vs 60.9%, p = .004). DY was lower for occult than for visible lesions (37.1% vs 76.9%, p < .001). Diagnostic performance for detecting malignancy on the basis of final diagnoses was lower for occult than for visible lesions in terms of sensitivity (76.7% vs 93.7%, p = .003), specificity (7.9% vs 56.5%, p < .001), and accuracy (38.2% vs 80.0%, p < .001). Final diagnoses among malignant occult and visible lesions included metastasis (frequencies of 63.3% vs 65.4%), leukemia/lymphoma (33.3% vs 11.6%), and myeloma (3.3% vs 10.4%); final diagnoses among benign occult and visible lesions included red marrow (34.2% vs 8.2%), reactive marrow (26.3% vs 11.8%), and fracture (18.4% vs 3.8%). Occult lesions detected by MRI versus PET/CT had lower malignancy rate (39.3% vs 68.0%, p = .03) and lower DY (30.4% vs 60.0%, p = .01). CONCLUSION. At CT-guided CNB, malignancy rate and DY are lower for occult than for visible lesions. Leukemia/lymphoma and red marrow are more common among occult than visible lesions. CLINICAL IMPACT. Understanding these characteristics can help guide radiologists', referring providers', and patients' expectations when CNB of occult bone lesions is requested and performed.