Risk factors for local recurrence and metastasis of eyelid sebaceous carcinoma after complete excision: A large cohort study in Japan

Abstract

Purpose:

To assess risk factors for local recurrence and metastasis in eyelid sebaceous carcinoma after complete excision in a Japanese population.

Methods:

In this retrospective study, we reviewed the medical records of 171 Japanese patients with histologically complete eyelid sebaceous carcinoma excision between December 2006 and August 2024. Patients who received adjuvant therapy other than surgery, did not have metastasis screening, or had metastasis detected prior to surgery were excluded. Sebaceous carcinomas were staged according to the American Joint Committee on Cancer staging system, 8^th edition. Kaplan–Meier analysis and the log-rank test were used to determine the risk factors for local recurrence and metastasis. The risk factors and the survival rate for local recurrence and metastasis were then assessed.

Results:

A total of 147 cases were included. Female sex (P = 0.024) and pagetoid spread (P = 0.006) were significantly associated with local recurrence. T stage ≥T1c (P = 0.041), lower eyelid tumor site (P = 0.021), and pagetoid spread (P = 0.011) were significantly associated with metastasis. At 5 years postoperatively, the local recurrence-free survival rate was 93.5% (95% CI: 86.9–96.9) and the metastasis-free survival rate was 89.6% (95% CI: 82.0–94.2).

Conclusion:

Excision margins, and follow-up and surveillance protocols for sebaceous carcinoma may need to reflect the significantly greater risk of local recurrence and metastasis with pagetoid spread, female sex, T stage (worse than T1b), and tumor location on the lower eyelid.

Sebaceous carcinoma is an aggressive neoplasm with a predilection for the periocular area.[1] It is generally rare but with wide geographical variation, and in some settings, such as Japan, it is the second most common eyelid malignancy after basal cell carcinoma.[2,3]

The sebaceous carcinoma mortality risk is 2–10%.[4,5] The reported risk factors for mortality are longer duration of symptoms (≥6M), larger tumor size, orbital invasion,[6] pagetoid spread,[7] and positive surgical margin.[8] Local recurrence is reported to occur in 6.0–32.7% of sebaceous carcinomas.[6,9] The risk factors for local recurrence are longer duration of symptoms,[10] perineural invasion, the involvement of more than one eyelid, orbital involvement,[11] pagetoid spread,[10] and tumor size.[12] Metastasis is reported to occur in 3.1–28.0% of sebaceous carcinomas.[9,13] The risk factors for metastasis are tumor location (medial canthal or lateral canthal involvement),[14] tumor size (≥10mm),[9,14,15,16] longer duration of symptoms,[10,15] pagetoid spread,[16] and orbital involvement.[11,15] However, these data derive mainly from small studies with a heterogeneous mix of patients including those with both clear and involved margins and some receiving adjunctive treatments such as radiotherapy. There is a paucity of data on the risk of local recurrence and metastasis after excision of sebaceous carcinoma with histologically clear margins.

The purpose of this retrospective study was to assess the clinical features of eyelid sebaceous carcinoma and elucidate the risk factors for local recurrence and metastasis in a large cohort of Japanese patients who underwent complete surgical excision of the tumor.

Methods

The protocols of this retrospective study were approved by the Institutional Review Board of Kyoto Prefectural University of Medicine, Kyoto, Japan (Approval No.: ERB-C-194-8). This study was performed in accordance with the tenets set forth in the Declaration of Helsinki.

In this study, we retrospectively analyzed the medical records of a consecutive series of 171 Japanese patients with a histological diagnosis of eyelid sebaceous carcinoma and clinical margins of >3mm and histologically negative margins of >0.5mm at the Kyoto Prefectural University of Medicine Hospital, Kyoto, Japan, and at the Seirei Hamamatsu General Hospital, Hamamatsu, Japan from December 2006 to August 2024. Patients who received adjuvant therapy other than surgery (n = 5), had metastasis prior to undergoing surgery (n = 7), or did not have metastasis screening (n = 12) were excluded from the study in order to evaluate the natural outcomes after complete surgical excision without the potential confounding effects of additional treatments or incomplete metastatic assessment.

All patients underwent a head and neck computed tomography and/or a positron emission tomography examination every 6 to 12 months to look for orbital invasive, regional spread, and metastasis. Local recurrence was identified clinically with biopsy of any suspicious changes.

The medical records of all patients were reviewed to obtain their clinical data, including the patient demographics (age, sex, and ethnicity), extent of tumor involvement, clinical tumor size (largest dimension of the lesion), growth pattern (nodular or diffuse), and histopathological features. Each patient was classified in accordance with the American Joint Committee on Cancer (AJCC) staging system, 8^th edition.[9]

Statistical methods

In statistical analysis, the probabilities of local recurrence-free survival (RFS) and metastasis-free survival (MFS) from eyelid sebaceous carcinoma excision were calculated with the Kaplan–Meier method. Survival was compared using the log-rank test. Statistical analysis of the data was performed using JMP version 18 (SAS Institute Inc., Cary, North Carolina, USA). A P value of < 0.05 was considered statistically significant.

Results

Population characteristics

In this study, 147 consecutive patients who had undergone complete surgical excision (clinical margins >3 mm and histologically negative margins >0.5 mm) of histologically confirmed sebaceous carcinoma were included. The mean age at presentation was 72.2 ± 13.5 years, and 88 of the 147 (59.9%) cases were female. The site of the tumor was the upper eyelid in 99 cases (67.3%) and the lower eyelid in 48 cases (32.7%). The tumor was macroscopically nodular in 140 cases and diffuse in 7 cases. The mean tumor size was 10.0 ± 6.7 mm (2.0–26.0 mm), Wide local excision was undertaken in 141 cases and orbital exenteration in 6 cases [Table 1]. The surgical reconstructive approach is presented in Table 2. Histopathological examination identified pagetoid spread in 27 cases (18.4%). The mean follow-up period was 52.3 ± 42.7 months and was longer in men than in women (males: 59.7 ± 48.5 months; females: 47.4 ± 37.5 months).

Table 1.

Clinicopathologic findings of the patients who underwent complete surgical excision

Variable		Total		Local Recurrence				Metastasis
				Positive		Negative		Positive		Negative
		(n=147)		(n=7)		(n=140)		(n=11)		(n=136)
Mean Age, years		72.2±13.5		77.4±14.7		71.9±13.4		68.7±14.6		72.5±13.4
Sex, n (Female/male)		88/59		7/0		81/59		5/6		83/53
Tumor location, n		99/48		4/3		95/45		4/7		95/41
(Upper/Lower)
Mean tumor size, mm		10.0±6.7		13.9±6.9		9.8±6.6		15.2±11.7		9.6±5.9
Pagetoid Spread, n (%)		27 (18.4%)		4 (57.1%)		23 (16.4%)		5 (45.5%)		22 (16.2%)
Tumor presentation, n		140/7		6/1		134/6		10/1		130/6
(nodular/diffuse)
T stage
T1b		67		2		65		1		66
T1c		31		0		31		4		27
T2b		10		0		10		1		9
T2c		26		4		22		3		23
T3b		1		0		1		0		1
T3c		8		1		7		1		7
T4a		4		0		4		1		3
Mean follow-up, months		52.3±42.7		71.4±44.1		51.4±42.4		66.2±47.2		51.2±42.1

Continuous variables are described as mean±standard deviation; categorical variables are described as n (%)

Table 2.

Reconstructive methods of surgery

Reconstructive methods		Number
Simple closure		42
Hard palate graft		30
Hughes flap		24
Free tarsal flap		23
Switch flap		21
Auricular cartilage graft		1

Local recurrence and metastasis analysis

Post surgery, local recurrence occurred in 7 (4.8%) of the 147 patients [Table 1]. The mean elapsed time post excision until local recurrence was 20.9 ± 15.5 months; the RFS at 1, 3, and 5 years postoperatively was 97.6% (95% CI; 92.9–99.2), 95.8% (95% CI; 90.3–98.3), and 93.5% (95% CI; 86.9–96.9), respectively [Fig. 1a]. Female sex (P = 0.024) and the presence of pagetoid spread (P = 0.006) were significantly associated with recurrence [Fig. 2]. Metastasis occurred in 11 (7.5%) of the 147 patients [Table 1]. The mean elapsed time post excision until metastasis was 22.9 ± 13.4 months. All 11 patients had nearby lymph-node metastasis, and 5 also had distant metastasis (lung: n = 4 cases; brain: n = 1 case). One patient had local recurrence before metastatic spread. T stage of ≥T1c (P = 0.041), tumor location on the lower eyelid (P = 0.021), and the presence of pagetoid spread (P = 0.011) were significantly inversely associated with the MFS rate. At 1, 3, and 5 years postoperative, MFS was 97.5% (95% CI: 92.7–99.2), 92.0% (95% CI: 85.2–95.8), and 89.6% (95% CI: 82.0–94.2), respectively [Fig. 1b]. Tumor grade of ≥T1c, tumor on the lower eyelid, and the presence of pagetoid spread were significantly associated with metastasis [Fig. 3]. The RFS and MFS for each factor at 5 years postoperatively are presented in Table 3.

Figure 1.

Local recurrence/metastasis-free survival after complete excision. (a) Kaplan–Meier analysis findings showing that at 1, 3, and 5 years post complete excision, RFS in eyelid sebaceous carcinoma was 97.6% (95% CI: 92.9–99.2), 95.8% (95% CI: 90.3–98.3), and 93.5% (95% CI: 86.9–96.9), respectively. (b) Kaplan–Meier analysis findings showing that at 1, 3, and 5 years post complete excision, MFS in eyelid sebaceous carcinoma was 97.5% (95% CI: 92.7–99.2), 92.0% (95% CI: 85.2–95.8), and 89.6% (95% CI: 82.0–94.2), respectively

Figure 2.

Recurrence-free survival stratified by clinical factors. Kaplan–Meier curves for time to outcomes of local recurrence by (a) T category, (b) sex of the patient, (c) tumor location, and (d) pagetoid spread

Figure 3.

Metastasis-free survival stratified by clinical factors. Kaplan–Meier curves for time to outcomes of metastasis by (a) T category, (b) sex of the patient, (c) tumor location, and (d) pagetoid spread

Table 3.

5 years postoperative local recurrence-free survival and metastasis-free survival

				Local Recurrence-free Survival		Metastasis-free Survival
				(95% CI)		(95% CI)
T stage		T1b		95.9% (84.8-99.0)		96.5% (87.0-99.1)
		≥T1c		91.3% (80.6-96.4)		83.3% (70.5-91.2)
Sex		Female		89.0% (78.3-94.7)		92.6% (83.3-96.9)
		Male		100% (100.0-100.0)		85.8% (71.8-93.5)
Tumor location		Lower		90.8% (74.9-97.0)		79.8% (62.9-90.1)
		Upper		95.1% (87.6-98.2)		94.5% (86.2-98.0)
Pagetoid spread		Positive		77.9% (53.0-91.7)		74.5% (50.9-89.1)
		Negative		97.0% (91.0-99.0)		92.9% (84.9-96.8)

During the postoperative follow-up period, tumor-associated death occurred in 3 cases at 25, 50 and 59 months after surgery, all due to lung metastasis.

Discussion

The identification of risk factors for local recurrence and metastatic spread of sebaceous carcinoma can help refine surgical treatment algorithms and postoperative surveillance. This study is the first large (over 140 cases) long-term follow-up study after histologically complete excision of sebaceous carcinoma and identified female gender, pagetoid spread, T stage ≥T1c, and tumor of the lower eyelid to be risk factors for recurrence and/or metastasis.

Previous studies have generally reported a much higher risk (6.0–32.7%) of local recurrence than the 4.8% in the present study. However, most of these studies included patients with a wide range of complete and incomplete margins and they typically reported, as expected, that narrow or incomplete margins were associated with recurrence in contrast to the present study of a more homogeneous cohort of patients, all of whom have had surgical excision of the primary tumor with 3 mm or greater clinical margins and negative margins.[6,17,18,19,20,21] None of the patients in the present study underwent adjunctive radiotherapy or chemotherapy, whereas Takagawa et al.[12] reported an increased risk of local recurrence with radiotherapy, although this may be a confounder for incomplete excision or large tumor size. It also remains uncertain whether there are geographical or ethnic differences in the risk of recurrence, and it is extremely hard to unpick this from the speed of presentation and diagnosis, the surgery and excision margins, and the length and frequency of follow-up. The ongoing risk of recurrence and metastasis with apparently clear margins indicates microscopic spread outside the excision margins that is not identified in histopathological examination (which is probably more likely with pagetoid spread) and/or the presence of microscopic metastasis that is not identified on imaging prior to excision. It reaffirms the need for long-term follow-up of all patients with sebaceous carcinoma.

To our knowledge, this is the first study to show female sex is significantly associated with increased local recurrence. Interestingly, Goto et al.[17] also reported that all cases of local recurrence were female. Androgen, estrogen, and progesterone receptor are presented in the meibomian gland, which is the most common site of origin of sebaceous carcinoma in the eyelid.[22] The meibomian gland is an androgen target organ, and androgen signaling has been shown to regulate lipid synthesis and secretion in this gland.[23] Therefore, it is possible that there are hormonal influences on the biological behavior of sebaceous tumors arising from this tissue.

The risk of postsurgery metastasis was 7.5% in the present study. This is in keeping with other studies which reported risks of between 3.1% and 28.0%.[9,13,14,16,18,19,24] It raises the question of whether a histologically margin negative is adequate to minimize the risk of metastasis or whether micro-metastasis is already present at the time of tumor excision, or perhaps even seeded during excision. Tumor size is one of the major risk factors for metastasis.[11,14,24] In 2016, the 8^th edition AJCC staging system was released, and reported a stronger correlation between tumor size and metastasis and tumor-related death than the staging system in the 7^th edition.[9,11] The significantly increased risk of metastasis is highly likely to reflect the tumor being present for longer and/or having a greater propensity to invade surrounding tissues, blood vessels, and lymphatic vessels. However, the correlation is still not very strong and metastasis can occur at any T stage and even with relatively small tumors that have been excised with wide margins.

To the best of our knowledge, there have been no previous reports on the difference in risk of metastasis between the upper and lower eyelids. In fact, anecdotally oculoplastic surgeons report that the risk is higher with upper eyelid tumors, but this is probably confounded by the more challenging excision surgery of the upper lid, resulting in a higher likelihood of incomplete excision or narrow margins. However, in the present study, all tumors had at least 3 mm of clinical clearance, allowing an accurate comparison of recurrence rates. Moreover, tumor size is not responsible for this difference as they were marginally larger (mean 10.4 mm, range 2 mm to 26 mm) in the upper eyelid than in the lower eyelid (mean 9.4 mm, range 2 mm to 25 mm). One possible explanation for this finding is that anatomically, the tarsal plate of the lower eyelid is shorter in height than that of the upper eyelid.[25] In addition, eyelid malignancies frequently metastasize via lymphatic pathways to regional lymph nodes, and differences in lymphatic drainage between the upper and lower eyelids may potentially influence metastatic patterns.[26,27] Thus, it is possible that lower eyelid tumors escape the tarsal plate and invade surrounding tissue where there are few natural barriers, earlier than upper eyelid tumors.

Pagetoid spread is a specific pattern of neoplastic invasion of the surface of the epithelium that is similar to that observed in Paget’s disease of the breast.[20] The present study found that it is significantly associated with both local recurrence and metastasis. In sebaceous carcinoma, Hsia et al.[11] reported that pagetoid spread is a risk factor for local recurrence, and Choi et al.[16] reported that pagetoid spread is a risk factor for lymph node or distant metastasis. Clinically, pagetoid spread tends to occur in the diffuse type of tumor, and cases of sebaceous carcinoma with pagetoid spread are more likely to masquerade as blepharitis,[21] which is associated with a great difficulty with diagnosis, and therefore a longer period prior to biopsy and surgical excision.[4,6,28,29] Early biopsy and then mapping biopsies of the palpebral and bulbar conjunctiva are vital to identify the extent of spread and may also identify the presence of pagetoid spread which may help guide surgical excision and margins.[18] It also raises the question as to whether wider excision margins should be used if pagetoid spread is identified on biopsy or suspected clinically.

In this study, all cases had complete excision (>3mm clinical margin) and negative margins by frozen section. Yadlapati reported Mohs micrographic surgery has superior outcomes for local, regional, and distant recurrence than wide local excision.[30] In their report, the local recurrence and metastasis risk were 23.4% and 11.0%, respectively, which is higher than the present study. This may reflect a different case mix, particularly as there is a high index of suspicion in a Japanese population. Tumors may be identified earlier than other settings. It still remains uncertain whether Mohs surgery reduces the risk of recurrence or metastasis than wide local excision with large margins and histological examination by an experienced pathologist.[31]

In addition to clinical and histopathological factors, genetic and molecular alterations may also influence the biological behavior of sebaceous carcinoma. Recent genomic studies have identified recurrent mutations in tumor suppressor genes such as TP53 and RB1 in sebaceous carcinoma, which have been associated with tumor progression and aggressive clinical behavior.[32] Tumors harboring these alterations may therefore have a greater propensity for recurrence or metastasis.

This study has several limitations. As patients who received adjuvant therapy or lacked metastatic screening were excluded, a degree of selection bias cannot be completely excluded. Additionally, the number of outcome events was relatively small (7 recurrences and 11 metastases), and multivariable Cox proportional hazards analysis was not performed due to the risk of model overfitting. Therefore, potential confounding among risk factors could not be fully adjusted for.

Conclusion

This study reveals that after complete surgical excision of an eyelid sebaceous carcinoma, the risk factors for local recurrence are female sex and pagetoid spread, and that the risk factors for metastasis are T stage (≥T1c), tumor location (lower eyelid), and pagetoid spread. These factors should be considered in both surgical planning, where wider surgical margins might be considered, and postoperatively, where more frequent or longer-term follow-up may be beneficial.

Authors' contributions

Oku, Watanabe, Rajak, Nakayama, Yoneda, Kumakiri and Sotozono had full access to all of the data in the study, and Yoshii had full access to all the study data for statistical analysis. All authors take responsibility for the integrity of the data and the accuracy of the data analysis; Concept: Oku, Watanabe; Design: Oku, Watanabe; Definition of intellectual content: Oku, Watanabe; Literature search: Oku; Data acquisition: Oku, Watanabe, Yoneda, Kumakiri, Sotozono; Data analysis: Oku; Statistical analysis: Oku, Yoshii; Manuscript preparation: Oku, Watanabe, Rajak, Sotozono; Manuscript editing and manuscript review: All authors; Guarantor: Oku.

Conflicts of interest

There are no conflicts of interest.

Funding Statement

Nil.