CT imaging improves histopathological grading of retroperitoneal leiomyosarcomas

Background: Initial grading of retroperitoneal leiomyosarcoma (LMS) is performed by core biopsy (CB) however, discrepancy between grade of tumour at initial CB and surgical excision is recognised, raising concerns about the accuracy of CB for directing neoadjuvant therapy. The histological grading system used for staging LMS consists of 3 components: tumour differentiation, mitotic index and proportion of necrosis. We postulate that assessment of necrosis by histopathology alone is inadequate, resulting in under-grading of LMS. We propose and assess a combined grading system that incorporates CT scan findings into pre-surgical grading. Methods: Retrospective, blinded review of CT, CB histology and final surgical histology of patients with retroperitoneal LMS was undertaken. A modified grading system, H-CT-Grade, was derived by replacing the CB necrosis score with a CT-derived necrosis score. The sensitivity and specificity of H-CT-Grade, the standard histopathology scoring, H-grade were compared. Inter-observer variability in assessment of CT necrosis was also assessed. Results: 53 patients fulfilled criteria for inclusion. CT was more sensitive at detection of necrosis than CB histology alone with sensitivity of 100% vs 53%. The use of (CT)HGrade resulted in increased detection of high-grade tumours with H-CT-grade having sensitivities of 80% and 35% for Grade 2 and 3 tumours respectively vs 53% and 15% with H-Grade. Assessment of reader agreement demonstrated Kappa scores of 0.8. Conclusion: Histology from CB under-grades LMS due to undersampling of tumour necrosis. CT is more sensitive in assessing necrosis and its incorporation into a modified CT-histopathology grading system (H-CT-Grade) improves accuracy of grading with significant implications for patient management. (C) 2019 Published by Elsevier Ltd.