• sebaceous carcinoma;
  • SEER;
  • Asian/Pacific Islanders;
  • skin cancer


  1. Top of page
  2. Abstract
  6. Conflict of Interest Disclosures


Sebaceous carcinoma is a rare and aggressive cutaneous carcinoma. It is believed that this malignancy predominates in the periocular region and occurs more frequently in Asian populations and in women. The objective of the current study was to analyze demographic characteristics and outcomes for patients with this malignancy from a large United States-based population registry.


An analysis of the National Cancer Institute's Surveillance, Epidemiology, and End Results database from 1973 through 2004 was performed.


Of 1349 patients who were identified, 54% were men, 86.2% were white, and 5.5% were of Asian/Pacific Islander ancestry. The median age at diagnosis was 73 years. The most frequent site of disease was the eyelid (38.7%). The population-matched 5- and 10-year age-matched relative survival rate was 91.9% (standard error [SE], 1.9%) and 79.2% (SE, 3.7%), respectively. Cause of death was attributable to cancer in 31% of patients. Orbital involvement did not predict for worsened survival compared with nonorbital involvement (5-year overall survival, 75.2% vs 68%, respectively; P = .66). The overall population-matched rate of sebaceous carcinoma was highest in whites (2.03 cases per 1000,000; SE, 0.08) versus Asian/Pacific Islanders (1.07 per 1000,000; SE, 0.18; P = .0001) versus blacks (0.48 per 1000,000; SE, 0.11; P < .0001).


The current results support the finding of a predominance of men among patients with sebaceous carcinoma, and no difference was observed in the prognosis for orbital and periorbital involvement. This retrospective analysis also corroborated previous case reports of a higher incidence among patients with advanced age and the highest incidence for sites in the eyelid and skin of the face. The results also established that Asian/Pacific Islander ancestry is not a risk factor for developing sebaceous carcinoma. Cancer 2009. © 2008 American Cancer Society.

Sebaceous carcinoma is a rare and potentially aggressive cutaneous tumor.1 What is known about the incidence, risk factors, and prognosis of this disease is based on isolated case reports of <200 patients.1 On the basis of these previous clinical case reports, >70% of sebaceous carcinomas are located in the head and neck region, where sebaceous glands are the most common.1 Sebaceous carcinomas are primarily periorbital (especially in the meibomian and Zeis glands) and comprise the fourth most common neoplasms of the eyelid.2 It is believed that metastases spread typically to regional lymph nodes.3 Reported risk factors for sebaceous carcinoma include advanced age,4 Asian or South Asian race,1, 5–7 women,7–10 previous irradiation to the head and neck,11, 12 and a genetic predisposition for Muir-Torre syndrome13 or possibly familial retinoblastoma.12

The most confounding aspect of sebaceous carcinoma is its myriad of clinical presentations, which delay diagnosis by 1 to 2.9 years and may increase mortality.5, 8, 14 Because sebaceous carcinoma presents most commonly as a gradually enlarging, firm nodule, it can mimic more common ophthalmologic or dermatologic conditions.2 Definitive diagnosis is only by pathology.15 Involvement of preauricular or cervical lymph nodes has been associated previously with a 50% to 67% 5-year mortality rate.16 Extraocular sebaceous carcinoma and smaller tumor sizes are associated with lower metastatic potential and, consequently, lower mortality.8, 17, 18 Poor prognostic factors include involvement of both upper and lower eyelids concomitantly and involvement of the upper lid, compared with the lower lid,8 as well as lymphovascular invasion, multicentric disease, poorly differentiated tumors, pagetoid spread, or a high level of infiltrative growth.2

Standard surgical resection with wide margins and Mohs migrographic surgery are common treatments for sebaceous carcinoma.19, 20 Radiation and topical and systemic therapies also have been suggested as useful in the treatment of sebaceous carcinoma, although experience is limited.7, 20–25

Given the rarity of sebaceous carcinoma and the relatively limited information available in the literature regarding US populations, we initiated an investigation of the National Cancer Institute's Surveillance, Epidemiology, and End Results (SEER) database26 to study the demographics and clinical outcomes of patients with sebaceous carcinoma. The SEER database is a collection of large, population-based cancer registries that records cases and general demographic and pathologic information.26 The most current SEER database records provide incidence and survival statistics for >25% of the United States population.26 Given the large size of the SEER database, it is an ideal tool for investigating the natural history and clinical outcomes of rare diseases like sebaceous carcinoma. Here, we report the results from the largest study to date of this rare cutaneous malignancy, the results of which may aid in the development of new clinical approaches toward sebaceous carcinoma in US populations.


  1. Top of page
  2. Abstract
  6. Conflict of Interest Disclosures

The entire SEER database from 1973 to 2004 was investigated, and patients with sebaceous carcinoma were selected. The SEER population is more urban than the total United States population (88.2% vs 79%) and also is more likely to be foreign born (17.3% vs 11.1%). Nonetheless, because of its large size, (SEER, in its most recent years, encompasses 26.2% of the US population), SEER is considered an excellent representation of the US cancer population as a whole. The SEER registry comprises information from populations in Connecticut, New Jersey, Atlanta and Rural Georgia, Detroit, Iowa, Hawaii, New Mexico, Seattle-Puget Sound, Utah, Arizona, Alaska, and all of California (including San Francisco-Oakland, San Jose-Monterey, Los Angeles, and Greater California).27 Cases in the SEER database have histology and involved anatomic site recorded by the morphology and topography codes of the International Classification of Disease for Oncology-Third Edition.28

Population-matched relative survival was calculated by comparing the Kaplan-Meier overall survival (OS) and the predicted age-matched survival based on the 2000 US standard population.29 This age-matched relative survival signifies the relative chance of surviving for the cohort of patients with sebaceous carcinoma compared with a standardized US population of similar age.

Survival data, sex, age, race, skin site, stage and extent of disease, lymph node involvement, radiation use, histology grade, surgery type and extent, and the number and type of other malignancies were investigated. SEER does not record chemotherapy information or information about other medical interventions. For patients who had more than 1 incidence of sebaceous carcinoma recorded (16 patients), the first case of sebaceous carcinoma was used for analysis, and subsequent cases were ignored.

Survival relative to population-matched US Census data was analyzed with SEER*Stat (National Cancer Institute, Bethesda, Md). All other statistical analyses were performed with STATA/SE 9.2 (Stata Corporation, College Station, Tex).

The OS of patients who had orbital sites of disease (conjunctiva, skin of eyelid, orbit) was compared with the OS of patients who had nonorbital sites of disease (all other specified disease sites). Orbital sites were skin of eyelid, orbit, and conjunctiva. Nonorbital sites of disease included skin of lip, skin of external ear, skin of scalp or neck, skin of trunk, skin of upper limb and shoulder, skin of lower limb and hip, parotid gland, descending colon, nasal cavity, labia majora, labia minora, vulva, vagina, ovary, penis, prostate, and scrotum. Cases that were not included in this analysis were those for which the primary site of disease was ambiguous or unknown. Because this included 26.8% of cases, we also performed the analysis of orbital versus nonorbital survival with nonspecific disease sites included first as orbital cases and then as nonorbital cases. Neither analysis reported a significant P value (P = .91 and P = .41, respectively), indicating that the inclusion of these nonspecific cases did not change our conclusion that there was no difference in survival between patients with orbital disease and patients with nonorbital disease.

The relative racial incidence of sebaceous carcinoma versus other histologies for all tumors of the skin of the eyelid was determined by using the collapsed SEER code for racial classification. Age-matched rates of sebaceous carcinoma were compared using rate ratios. P values were generated by the SEER*Stat program using the method described by Fay.30 Because numerous small case studies of sebaceous carcinoma suggested an Asian predominance in incidence,1, 5–7 we sought to confirm this trend in our study. For each site of disease with >10 cases of sebaceous carcinoma, logistic regression was performed to analyze whether Asian race or non-Asian race was predictive of having sebaceous carcinoma versus nonsebaceous carcinoma histology. Patients with unrecorded or unknown race were excluded from this analysis.


  1. Top of page
  2. Abstract
  6. Conflict of Interest Disclosures

One thousand three hundred forty-nine cases of sebaceous carcinoma were identified. Patient information is reported in Table 1. The median age at diagnosis was 72 years (range, 13-102 years). There was a slight predominance of men (54%). Primary skin sites of disease, from the most frequent to the least frequent, were the eyelid, 38.7%; skin of other and unspecified parts of the face (excluding lip, eyelid, and external ear; 26.8%); trunk, 13.3%; scalp and neck, 8.7%; upper limb and shoulder, 3.9%; external ear, 3.2%; lower limb and hip, 1.6%; nonskin site of the head and neck, 1.3%; lip, 0.8%; and site not specified, <1%. Other sites of disease included the genitals and reproductive organs (1.1% of cases). It is noteworthy that there was 1 recorded case of sebaceous carcinoma of the descending colon. Race distribution was as follows: white, 86.2%; Asian/Pacific Islander, 5.5%; black, 2.7%; and unknown/other, 5.5%. From the data available, we knew that lymph nodes were sampled pathologically in only 37 patients. There was clinical or pathologic evidence of regional lymph node involvement in 1.3% of patients and evidence of distant or unspecified lymph nodes in 0.4% of patients. In 63.9% of patients, there was no clinical or pathologic evidence of lymph node involvement, and lymph node status was not recorded for 34.4% of patients.

Table 1. Patient Characteristics (n = 1349)
CharacteristicNo. of Patients (%)
  • *

    It is noteworthy that this includes surgery of regional and/or distant site(s) only without surgery directed at the site of initial disease.

Age, y
 10-191 (<0.1)
 20-293 (0.2)
 30-3924 (1.8)
 40-4987 (6.4)
 50-59180 (13.3)
 60-69277 (20.5)
 70-79381 (28.2)
 80-89310 (23)
 ≥9086 (6.4)
 Men729 (54)
 Women620 (46)
 White1163 (86.2)
 Black37 (2.7)
 Asian/Pacific Islander75 (5.5)
 Unknown/other74 (5.5)
Site of disease
 Nonskin site of the head and neck17 (1.3)
 Skin of lip11 (0.8)
 Skin of eyelid522 (38.7)
 Skin of external ear43 (3.2)
 Skin of other and unspecified parts of face361 (26.8)
 Skin of scalp, neck117 (8.7)
 Skin of trunk179 (13.3)
 Skin of upper limb and shoulder52 (3.9)
 Skin of lower limb and hip22 (1.6)
 Skin of overlapping regions or not specified8 (0.6)
 Genitals/reproductive organs15 (1.1)
 Descending colon1 (<0.1)
 Unknown primary site1 (<0.1)
Surgery type
 No site cancer-directed surgery performed*143 (10.6)
 Local tumor destruction, local excision, or wide local excision995 (73.8)
 Major amputation40 (3)
 Unknown171 (12.7)
Lymph node dissection
 No lymph node dissection1300 (96.4)
 Regional lymph nodes sampled or dissected23 (1.7)
 Sentinel lymph node dissection with or without regional lymph node dissection14 (1)
 Unknown12 (0.9)
 No radiation1262 (93.6)
 Radiation (beam, brachytherapy)74 (5.4)
 Unknown type of radiation or unknown whether radiation given13 (1)
Combination therapy
 No surgery or radiation132 (9.8)
 Surgery alone977 (72.4)
 Radiation alone11 (0.8)
 Radiation and surgery53 (3.9)
 Unknown or not recorded176 (13)
First malignancy
 Yes1029 (76.3)
 No320 (23.7)
No. of primary malignancies
 1944 (70)
 ≥2404 (30)
 Unknown or not recorded1 (<0.1)
Year of diagnosis
 1973-197946 (3.4)
 1980-1989154 (11.4)
 1990-1999448 (33.2)
 2000-2004701 (52)
Lymph node involvement
 No lymph nodes involved862 (63.9)
 Regional or distant lymph nodes involved23 (1.7)
 Lymph node status unknown or not recorded464 (34.4)

The OS rate at 5 years and 10 years was 71.1% (standard error [SE], 1.5%) and 45.9% (SE, 2.1%), respectively. The 5- and 10-year survival rate relative to an age-matched 2000 US standard population was 91.9% (SE, 1.9%) and 79.2% (SE, 3.7%), respectively (Fig. 1, Table 2). Cause of death was attributable to cancer in 147 of 473 deaths (31% of cases). Survival curves for orbital and periorbital sites of disease did not differ significantly from the survival curves for nonorbital sites (P = .66) (Table 3).

thumbnail image

Figure 1. Overall survival and survival relative to a United States Census- matched population.

Download figure to PowerPoint

Table 2. Overall and Relative Survival—All Patients With Sebaceous Carcinoma
Variable% SurvivingSE, %
  1. SE indicates standard error of the mean.

Overall survival
 5 Y71.11.5
 10 Y45.92.1
Relative population-matched survival
 5 Y91.91.9
 10 Y79.23.7
Table 3. Estimated Survival for Orbital and Periorbital Sites Versus Nonorbital Sites*
Primary Disease Site (n = 978)5-Year OS (SE), %P
  • OS indicates overall survival; SE, standard error.

  • *

    Nonspecific sites of disease (371 patients) were excluded from analysis.

  • Comparison was made by using the log-rank test.

Orbital (n = 530)75.2 (2.2) 
Nonorbital (n = 448)68 (2.7).66

Radiation use was rare: Only 5.3% of all patients received external-beam radiation, and 2 of 1349 patients received brachytherapy. Over 76% of patients underwent some form of surgery. No chemotherapy data are recorded by SEER.

Asian/Pacific Islander patients who had cancers of the skin of the eyelid were 6.21 times more likely (95% confidence interval, 3.8-10.1 times more likely; P < .001) to have sebaceous carcinoma compared with non-Asian/Pacific Islander patients who had cancers of the skin of the eyelid (Table 4). Forty-nine of 76 (65%) Asian/Pacific Islander patients who had eyelid malignancies recorded in SEER had a sebaceous carcinoma. Sebaceous carcinoma accounted for 22.6% (449 of 1982) of eyelid malignancies in white patients. For white patients, this was the third most common eyelid malignancy after lentigo maligna (520 of 1982 patients) and melanoma (458 of 1982 patients). Asian/Pacific Islander patients with cancers of the skin of scalp, neck, and trunk also were more likely to have a larger relative proportion of sebaceous carcinomas compared with other skin cancers in non-Asian/Pacific Islander patients (see Table 4). It is worth noting that basal cell carcinomas and squamous cell carcinomas are not recorded frequently in the SEER database.

Table 4. Odds Ratios for the Likelihood That a Skin Cancer Will Be Sebaceous Carcinoma in Asian/Pacific Islander Patients Versus Non-Asian/Pacific Islander Patient
Skin SiteOR95% CIP
  • OR indicates odds ratio; CI, confidence interval.

  • a

    *Patients with unrecorded race were excluded from the analysis.

  • ORs signify the odds that a patient of Asian/Pacific Islander ancestry who has any skin cancer at the site specified in the column ‘Skin Site’ will have a cancer with sebaceous carcinoma histology at that skin site versus the odds that a non-Asian Pacific Islander patient with any skin cancer at the same skin site will have sebaceous carcinoma. A significant P value indicates that it is unlikely that Asian/Pacific Islander patients will have the same proportion of sebaceous carcinomas as non-Asian/Pacific Islander patients as a proportion of their total number of all skin cancers of a specified skin site.

  • Could not be calculated, because there were no Asian/Pacific Islander patients with sebaceous carcinoma of the lower limb and shoulder.

Skin of eyelid6.213.8-10.1<.001
Skin of lip13.81.5-126.4.02
Skin of external ear3.90.52-29.0.18
Skin of scalp and neck7.13.1-16.6<.001
Skin of trunk5.42.7-11.1<.001
Skin of upper limb and shoulder2.70.37-9.5.33
Skin of lower limb and shoulder

However, the incidence of sebaceous carcinoma based on the crude relative incidence rates relative to the US Census population from 2000 to 2004 (Table 5) increased for whites (2.03 per 1000,000; SE, 0.08) versus Asian/Pacific Islanders (1.07 per 1000,000; SE, 0.18; P = .0001) versus blacks (0.48 per 1000,000; SE, 0.11; P < .0001).

Table 5. Overall Rates of Sebaceous Carcinoma (per 1,000,000) Matched to the 2000 U.S. Census Population
  • SE indicates standard error.

  • *

    Index value.

Asian/Pacific Islander1.070.18.0001
American Indian0.390.28.0044


  1. Top of page
  2. Abstract
  6. Conflict of Interest Disclosures

This study of 1349 cases that were identified over a period of 31 years was a retrospective review on a mixed population from data reported in the National Cancer Institute's SEER database. Given the rarity of sebaceous carcinoma, the SEER database is an ideal method for describing the demographic attributes of this cutaneous malignancy. Our data support previous clinical reports that sebaceous carcinomas occur primarily in the skin of the orbital region or in the skin of the nonorbital face, head, and neck. We also corroborate previously published data associating sebaceous carcinoma with advanced age1; in the current study, the median age at diagnosis was 73 years. Age-matched relative survival data from our analysis also indicate that sebaceous carcinoma remains a significant cause of mortality in affected US populations. Significantly, our data do not support several previously reported and accepted demographic features of sebaceous carcinoma. We did not detect a predominance of incidence among women. In addition, there was not a poorer prognosis for periorbital or orbital involvement. Finally, and perhaps most noteworthy, there was not an Asian predominance in the overall incidence of sebaceous carcinomas. Each of these points is detailed below.

Our results are suggestive of a slight predominance of incidence among men (54%). Most case reports of sebaceous carcinoma demonstrate that women are more likely to develop this cutaneous malignancy, and 1 major review stated that >70% of cases occur in women.1, 7–10 Our data are more supportive of other studies, which reported a relatively equal predisposition for the development of sebaceous carcinoma in men and women.3

Our results also do not support higher mortality associated with periorbital or orbital disease, as suggested previously in the literature.8, 17, 18 Survival analysis for orbital and periorbital versus nonorbital sites supports a similar prognosis (5-year survival rate, 75.2% vs 68%, respectively; P = .66) in these 2 groups. Limited staging data from 2004 (n = 141 cases) were difficult to interpret, because almost 70% of cases had unknown or inapplicable staging.

Finally, our results demonstrate that a higher proportion of whites are affected with sebaceous carcinoma than was reported previously.1 The work cited most to support an Asian predominance is the article by Ni et al.,7 in which the incidence of sebaceous carcinoma in Shanghai is compared with that in Boston. Ni et al. observed that sebaceous cell carcinomas accounted for 32.7% of malignant eyelid tumors in Shanghai; whereas, at the Massachusetts Eye and Ear Institute, sebaceous carcinomas accounted for only 1.5% of eyelid tumors. Our analysis demonstrated that sebaceous carcinomas did account for a higher proportion of malignant eyelid tumors in Asian/Pacific Islander patients but that this was because of the relative lack of other eyelid skin tumors rather than an increased incidence of sebaceous carcinomas overall in Asian/Pacific Islander patients.

Therefore, our results indicate that being of Asian/Pacific Islander ancestry is not a risk factor for sebaceous carcinoma. The proportion of Asian/Pacific Islander patients with skin cancers of the face who have sebaceous carcinoma is higher than the proportion in whites; therefore, the Shanghai experience appeared to indicate a higher relative number of sebaceous carcinomas than the Massachusetts Eye and Ear Institute. Because the populations of patients were from 2 institutions of different size, comparing the absolute number of cases also may have contributed to the results published by Ni et al.7 It is important to note that the ethnic makeup of the SEER database is different from the that of US Census population. Whereas 26.1% of the entire US population is included in the SEER database, 53.3% of the US Asian population is included, leading to a relative Asian predominance of patients.31 This likely would bias our study in favor of an increased incidence of sebaceous carcinoma among Asian patients, and it is significant that, in the face of this bias, our study still identified significantly less sebaceous carcinoma than expected.

Cause of death was attributable to cancer in 31% of cases, and the relative population-matched survival rate at 10 years was 79.2%. Although our study demonstrated that patients with sebaceous carcinoma were more likely to die from noncancer-related causes (rather than cancer-related causes) by a factor of 2.2:1, sebaceous carcinoma remained a significant cause of morbidity and mortality.

It is noteworthy that sebaceous carcinoma comprised 1% of all eyelid tumors and 4.7% of malignant epithelial eyelid tumors in 1 study.32 Although the high incidence of sebaceous carcinomas relative to other malignancies of the skin of the eyelid observed in our study is likely because of the under-reporting of basal cell carcinomas by cancer registries in the SEER database, the description of relative prevalence between racial groups remains valid. Also limiting the SEER database is a lack of information recorded on chemotherapy, tumor burden, recurrence data, margin data, performance data, information on comorbid conditions, and the dose, technique, or intent of radiation. Staging information is available only for 2004, and 69.5% of those cases did not have stage of disease recorded. We also were unable to corroborate previous data suggesting poorer outcomes in patients with preauricular and cervical lymph node involvement,14 because lymph node dissections were not performed or recorded in 97.3% of patients. In fact, the 1.7% of patients in our study who had clinical or pathologic evidence of lymph node involvement was an unexpectedly low percentage. However, despite these limitations of the SEER database, the results of this study allow elucidation of the demographics of this disease and clarification of expected survival, both overall and relative to others of similarly advanced age.

In conclusion, to our knowledge, the current study, with 1349 cases from a US population, is the largest report in the literature to date of sebaceous carcinoma, a very rare cutaneous malignancy. Our retrospective review corroborates previous case reports of higher incidence in patients with advanced age and highest prevalence in the eyelid and skin of the face. Significantly, our results do not support a predominance of incidence among women or a poorer prognosis for orbital or periorbital presentations. We have demonstrated that Asian/Pacific Islander ancestry is not a risk factor for sebaceous carcinoma. However, because of the relatively low number of other skin malignancies, Asian or Pacific Islander populations with cancers of the skin of the eyelid, lip, head and scalp, and trunk have a higher proportion of sebaceous carcinoma than other races.

Patients with sebaceous carcinoma have an age-matched OS rate of 91.9% at 5 years and 79.2% at 10 years. Thus, this report sheds new light on the incidence and natural history of this rare cutaneous malignancy.


  1. Top of page
  2. Abstract
  6. Conflict of Interest Disclosures