Keratocystoma: A Distinctive Salivary Gland Neoplasm Characterized by RUNX2 Rearrangements

Keratocystoma is a rare salivary gland lesion that has been reported primarily in children and young adults. Because of a scarcity of reported cases, very little is known about it, including its molecular underpinnings, biological potential, and histologic spectrum. Purported to be a benign neoplasm, keratocystoma bears a striking histologic resemblance to benign lesions like metaplastic Warthin tumor on one end of the spectrum and squamous cell carcinoma on the other end. This overlap can cause diagnostic confusion, and it raises questions about the boundaries and definition of keratocystoma as an entity. This study seeks to utilize molecular tools to evaluate the pathogenesis of keratocystoma as well as its relationship with its histologic mimics. On the basis of targeted RNA sequencing (RNA-seq) results on a sentinel case, RUNX2 break-apart fluorescence in situ hybridization (FISH) was successfully performed on 4 cases diagnosed as keratocystoma, as well as 13 cases originally diagnosed as tumors that morphologically resemble keratocystoma: 6 primary squamous cell carcinomas, 3 metaplastic/dysplastic Warthin tumors, 2 atypical squamous cysts, 1 proliferating trichilemmal tumor, and 1 cystadenoma. RNA-seq and/or reverse transcriptase-PCR were attempted on all FISH-positive cases. Seven cases were positive for RUNX2 rearrangement, including 3 of 4 tumors originally called keratocystoma, 2 of 2 called atypical squamous cyst, 1 of 1 called proliferating trichilemmal tumor, and 1 of 6 called squamous cell carcinoma. RNA-seq and/or reverse transcriptase-PCR identified IRF2BP2::RUNX2 in 6 of 7 cases; for the remaining case, the partner remains unknown. The cases positive for RUNX2 rearrangement arose in the parotid glands of 4 females and 3 males, ranging from 8 to 63 years old (mean, 25.4 years; median, 15 years). The RUNX2-rearranged cases had a consistent histologic appearance: variably sized cysts lined by keratinizing squamous epithelium, plus scattered irregular squamous nests, with essentially no cellular atypia or mitotic activity. The background was fibrotic, often with patchy chronic inflammation and/or giant cell reaction. One case originally called squamous cell carcinoma was virtually identical to the other cases, except for a single focus of small nerve invasion. The FISH-negative case that was originally called keratocystoma had focal cuboidal and mucinous epithelium, which was not found in any FISH-positive cases. The tumors with RUNX2 rearrangement were all treated with surgery only, and for the 5 patients with follow-up, there were no recurrences or metastases (1 to 120 months), even for the case with perineural invasion. Our findings solidify that keratocystoma is a cystic neoplastic entity, one which appears to consistently harbor RUNX2 rearrangements, particularly IRF2BP2::RUNX2. Having a diagnostic genetic marker now allows for a complete understanding of this rare tumor. They arise in the parotid gland and affect a wide age range. Keratocystoma has a consistent morphologic appearance, which includes large squamous-lined cysts that mimic benign processes like metaplastic Warthin tumor and also small, irregular nests that mimic squamous cell carcinoma. Indeed, RUNX2 analysis has considerable promise for resolving these differential diagnoses. Given that one RUNX2-rearranged tumor had focal perineural invasion, it is unclear whether that finding is within the spectrum of keratocystoma or whether it could represent malignant transformation. Most important, all RUNX2-rearranged cases behaved in a benign manner.