Outcome after treatment for sebaceous carcinoma: A multicenter study

Abstract Background Sebaceous carcinoma (SC) is a rare malignant tumour whereby, comprehensive long‐term data are scarce. This study aimed to assess the outcome of patients treated with resection for SC. Methods Patients treated at four tertiary centres were included. Cumulative incidence curves were calculated for recurrences. Results A total of 100 patients (57 males, 57%) were included with 103 SCs. The median age was 72 (range, 15–95) years with a median follow‐up of 52 (interquartile range [IQR], 24‐93) months. Most SCs were located (peri)ocular (49.5%). Of all SCs, 17 locally recurred (16.5%) with a median time to recurrence of 19 (IQR, 8–29) months. The cumulative incidence probability for recurrence was statistically higher for (peri)ocular tumours (p = 0.005), and for positive resection margins (p = 0.001). Two patients presented with lymph node metastases and additional seven patients (8.7%) developed lymph node metastases during follow‐up with a median time to metastases of 8 (IQR, 0.5–28) months. Three patients had concurrent in‐transit metastases and one patient also developed liver and bone metastases during follow‐up. Conclusion SC is a rare, yet locally aggressive tumour. Positive resection margins and (peri)ocular SCs are more frequently associated with local recurrence. SC infrequently presents with locoregional or distant metastases.


| INTRODUCTION
Sebaceous carcinoma (SC) is a rare malignant tumour of the sebaceous glands and only accounts for 0.7% of all cutaneous malignancies. SC has an incidence of 2:1.000.000 compared to an incidence of 164:1.000.000 for basal cell carcinoma in 2009 in the Netherlands. 1,2 It can occur at any site of the body where the glands are present, but are mostly found in the (peri)ocular area. The golden standard for treatment is wide local excision with a reported local recurrence rate of 4%-28%. [2][3][4][5] No standardised resection margins are described.
Radiotherapy as primary treatment has a higher recurrence rate and, therefore, this is only used in patients refusing excision. 2 Since SC is mostly found in the periocular region, these lesions are often divided into (peri)ocular and extraocular SCs. To date, there are only (small) cases series and literature reviews analysing the outcome of this disease at all anatomical locations, all emphasising the scarcity of data, and the need for more studies. 2,6 The majority of these cases refer to (peri)ocular SC. Extraocular SC is associated with lower metastatic potential and consequently lower mortality in comparison to (peri)ocular SC. However, these conclusions are based on small case series and the results are contradicted by other case series. 3,5,7 With an increased incidence of 3.31% annually in the US and only small cases series, or studies on (peri)ocular SC location, there is a need to better understand the prognosis and course of this disease. 3 Therefore, the aim of this study is to assess the rates of recurrence and metastases as well as survival and define prognostic factors for the outcome, for SC in all locations. The primary treatment was noted as the first treatment after pathological confirmation of SC. For the ocular SCs, some patients had a history of treatments going back multiple years, for chalazion, basal cell carcinoma, or squamous cell carcinoma, however without pathological confirmation of SC. Therefore, these treatments are not taken into account in the analyses. Seven patients were excluded due to the lack of treatment records.

| Statistical analysis
IBM SPSS statistics 25 and R (R Core Team, 2019) 8 were used for the statistical analyses. Recurrence rates were calculated using the cumulative incidence curves (CICs) accounting for competing risks.
Differences between CICs were calculated using Gray's test, 9 due to the small sample size multivariate analyses were not conducted. In addition, due to the small number of metastases, no statistical analysis on risk factors for metastases was conducted. Median survival was crudely derived using the Kaplan-Meier curve for descriptive purposes. . The extraocular tumours were significantly larger than the ocular tumours (p < 0.001) and showed a trend towards more male prevalence (p = 0.07) ( Table 1).

| Recurrence
Of all SCs, 17 locally recurred (16.5%): Half of these patients had a previous R1 resection. The median time to recurrence was 19 months (IQR, . The cumulative incidence probability for recurrence was significantly higher for (peri)ocular tumours compared to extraocular tumours (p = 0.005), and for positive resection margins compared to clear resection margins (p = 0.001) (Figure 1). The cumulative

| Metastases
Nine patients (8.7%) developed regional and/or distant metastasis during follow-up with a median age of 69 years (range, 53-93) and a median time to metastasis of 8 months (IQR, 0.5-28) ( Table 2) Five patients also had local recurrent disease (56%), which is higher than the recurrence rate of 17% for all SCs.
Most of the patients with lymph node metastases were treated with lymph node dissections (n = 8), followed by postoperative radiotherapy in seven patients ( theoretically, wide excision with 5-to 6-mm margins in all cases might provide a higher cure rate. 13 However, wide surgical margins conflict with the aim to preserve a functional eye. At present, no guidelines or recommendations for the width of the surgical margin for extraocular SC are available. [10][11][12] Our results support higher local recurrence rates after positive resection margins and for (peri)ocular SCs. All published literature on risk factors for recurrences is based on ocular SCs. Haber et al. 14 found a recurrence rate of 16.6% for extraocular SC without analysing risk factors for recurrence. In the ocular region, a higher tumour stage, intraepithelial neoplasia, and an initial (benign) misdiagnosis are described as risk factors for recurrences. 15,16 This study observed a total metastasis rate of 8.7%, all involving the regional lymph nodes. The metastasis rate in most recent literature varies from 2.4% to 12%. 3,5,6 Tryggvason et al. 7 found a higher metastatic rate (regional or distant) in ocular SCs (4.4%) compared to extraocular sites (1.4%) only focusing on head and neck locations. In contrast, two publications including all SCs, based on the SEER database, suggest a higher prevalence of metastases in ocular SC, but a better overall survival for ocular SC. 3,5 Other studies showed that metastasis rates are associated with poor differentiation, larger tumour size, and increased tumour depth. 7,17,18 In this study, none of the well-differentiated tumours metastasised, although the differentiation state of many patients was unknown. Furthermore, not a single SC <10 mm in size metastasised, which is in line with the study by Lam et al. 11 who also did not find any metastases in tumours <10 mm in the ocular region.
In literature to date, only 30 cases of metastatic extraocular SC have been described. The most common site of metastasis including all locations were lymph nodes only (40%), lymph nodes and visceral organ (20%), visceral organs only (16%), and local spread (12%). 10,19,20 In this series, the one patient with bone and liver metastases was treated with the best supportive care, all other lymph node metastases were treated with lymph node dissection. Due to the lack of literature on metastatic SC, optimal treatment has not been firmly established.
Literature on lymph node involvement for SC recommends adjuvant radiotherapy after lymph node dissection. 6 For the follow-up of SC, no standard guidelines are available.
One recent guideline recommends follow-up every 6 months for the first 3 years and thereafter yearly consultations. 25  illustrates that SC can arise from different mutational mechanisms, whereby the UV damaged group has more poorly differentiated SCs in comparison to the Muir-Torre group.
Due to the retrospective nature of this study, caution should be taken in interpreting these results. As with any retrospective study, the investigator depends on the availability and accuracy of the medical record. This study includes patients referred to tertiary hospitals including an eye hospital which can cause a bias in the prevalence per tumour location. However, when compared to the literature, (peri) ocular tumour location is also described as the most prevalent anatomic region for SC. Only including tertiary centres in the analysis could potentially bias your results. Given that most rare cancer will be referred to tertiary centres, it is plausible that this bias is less significant for SC. Although this is a large series in the field, caution should also be exercised regarding the conclusions drawn, due to the relatively small patient population in this study. On the other hand, this is one of the first and largest studies describing the natural history and metastatic pattern of SC and risk factors for recurrence.

| CONCLUSION
Altogether, SC is a very rare, yet locally aggressive tumour in the elderly patient population. Patients with positive resection margins and (peri)ocular tumour location are more frequently associated with a local recurrence. Patients with SC infrequently present with locoregional or distant metastases, resulting in a good overall survival.