Eyelid and periocular sebaceous gland carcinoma: risk factors for recurrence, exenteration, metastasis, and death in 355 patients.

Objective: To analyze factors predictive of tumor recurrence, orbital exenteration, lymph node metastasis, systemic metastasis, and metastasis-related death in patients with eyelid and periocular sebaceous gland carcinoma (eSGC) undergoing protocol-based management.

Methods: Retrospective interventional study RESULTS: Of the 355 patients with eSGC, 248 (70%) were referred to us without any prior intervention, and 107 (30%) had a history of prior intervention. The treatment modalities after presentation to our center included excisional biopsy (n = 303, 85%), orbital exenteration (n = 36, 85%), neoadjuvant chemotherapy (n = 12, 3%), and topical mitomycin C (n = 4, 1%). The 15-year Kaplan Meier estimates of tumor recurrence, need for orbital exenteration, lymph node metastasis, systemic metastasis, and metastasis-related death were 23%, 26%, 12%, 8%, and 9%, respectively. Multivariate Cox regression model identified positive histopathological margins (p < 0.001) and perivascular invasion (p < 0.001) as predictors of tumor recurrence; diffuse tumor morphology (p = 0.009), tumor diameter >20 mm (p = 0.027), orbital extension of the tumor (p < 0.001), perivascular invasion (p = 0.013), and pagetoid tumor spread (p < 0.001) on histopathology as predictors of orbital exenteration; diffuse tumor morphology (p = 0.005), tumor diameter >10 mm (p = 0.015), and perivascular invasion (p = 0.008) as predictors of lymph node metastasis; diffuse tumor morphology (p = 0.024) as a predictor of systemic metastasis; and multicentric tumor origin (p = 0.035) as a predictor of metastasis-related death.

Conclusion: Clinical features, including diffuse tumor morphology, multicentric tumor origin, larger tumor diameter, and orbital extension of the tumor, and histopathological features, including pagetoid tumor spread, positive histopathological margins, and perivascular invasion, are predictors of poor outcomes in eSGC.