Is stromal microinvasion in primary mucinous ovarian tumors with "mucin granuloma" true invasion?

Stromal microinvasion has been recently described in ovarian mucinous borderline tumors (MBTs), leading to proposals for new classifications, including a category of MBTs with stromal microinvasion. This study was conducted to test the validity of this belief. To determine whether stromal microinvasion can be distinguishable from extruded neoplastic epithelium from an adjacent ruptured gland, particularly if accompanied by a mucin granuloma (MG) on hematoxylin and eosin (H&E)-stained sections, we compared the histopathologic features of 138 primary ovarian mucinous tumors, consisting of 81 MBTs, 37 MBTs with stromal microinvasion, 11 intraglandular carcinomas, 2 with microinvasive foci, and 7 mucinous adenocarcinomas with extensive stromal invasion. Immunohistochemical analysis for cytokeratin was performed in 72 cases. Of 77 cases containing MGs, 52% contained "microinvasive foci," consisting of individually scattered epithelial or irregular glandular components in the intervening stroma, as determined by H&E staining. Of the 37 negative cases on H&E, cytokeratin reactivity was detected in 15 additional cases with epithelial/glandular components. Therefore, "stromal microinvasion" can be missed on H&E. All patients with MBT with or without microinvasive foci and localized intraglandular carcinoma with or without microinvasive foci had stage I tumors; none died of tumor-related causes during the follow-up period. Tumor-related deaths were identified only in 5 patients with extensive stromal invasion. This study suggests that some cases of stromal microinvasion in stage I MBTs not accompanied by high grade nuclear atypia represent mucocele-like stromal reaction to ruptured mucinous glands rather than true stromal invasion because isolated tumor cells were exclusively confined to the boundary of MGs and devoid of obvious cellular atypia, and no patient with stromal microinvasion died of tumor-related causes.