Validity and reproducibility of histologic diagnosis and grading for adult soft-tissue sarcomas

Soft-tissue sarcomas in adults show great variations in histologic type and grade. A valid and reproducible prognostication system is needed to select patients with soft-tissue sarcomas who could benefit from adjuvant chemotherapy. This study was conducted to assess the validity and reproducibility of diagnosis of the histologic type, MIB-1 grade, and mitosis grade, as well as of the 3 components of these grading systems. MIB-1 grade is a recently proposed grading system for predicting the prognosis of patients with adult soft-tissue sarcomas on the basis of 3 criteria (tumor differentiation, necrosis, and MIB-1 score) and replaces the mitotic count in the French system with MIB-1 immunohistochemical staining. Four surgical pathologists from 4 institutions who had experience in diagnostic soft-tissue tumor pathology reviewed 130 cases of soft-tissue sarcoma and independently determined histologic type and grade. The validity of histologic diagnosis was measured by sensitivity and specificity, and that of grading was measured by kappa statistics and percentage agreement with the diagnosis of the expert panel at the National Cancer Center, which was defined as a gold standard. Interobserver reproducibility was measured by kappa and by percentage agreement between the diagnoses of the 4 pathologists. The validity of the diagnosis of histologic type was high for synovial sarcoma, small round-cell sarcoma, and liposarcoma (sensitivity 89% to 100%; specificity, 98% to 100%) but low for malignant fibrous histiocytoma (MFH) and spindle-cell sarcoma (73% to 75%; 93% to 95%). For the grading, the validity of the MIB-1 grade was substantial (kappa = 0.68; agreement, 79%) and higher than that of the mitosis grade (0.54; 69%). The most valid component was tumor differentiation (kappa = 0.79), followed by tumor necrosis (0.66), MIB-1 score (0.59), and mitotic score (0.37). Interobserver reproducibility of histologic diagnosis was high for small round-cell sarcoma, synovial sarcoma, and liposarcoma (kappa = 0.92, 0.90, 0.87; percentage agreement = 99%, 97%, 96%, respectively); for grading, reproducibility was highest for tumor differentiation (0.78; 87%) and second highest for MIB-1 grade (0.68; 79%). We conclude that diagnosis of the type of soft-tissue sarcoma for synovial sarcoma, small round-cell sarcoma, and liposarcoma and the MIB-1 grading system based on tumor differentiation are highly valid and reproducible among Japanese pathologists who are familiar with the grading system, whereas re-evaluation of histologic criteria is essential for other histologic types such as MFH and spindle-cell sarcoma.