Intended for Healthcare Professionals
You are here
Home > Blog > Clinical Calculators > Eyelid Carcinoma (SGC): A Tool To Predict Recurrence

Eyelid Carcinoma (SGC): A Tool To Predict Recurrence

Eyelid Carcinoma (SGC): A Tool To Predict Recurrence


Eyelid sebaceous gland carcinoma (SGC) is a highly malignant condition with a significant recurrence risk. This study aimed to create a predictive nomogram for SGC recurrence based on risk factors.


A retrospective study involving 391 patients from one hospital (304) and grassroots hospitals (87) was conducted. Cox regression identified predictors for the nomogram, and its discriminative ability was evaluated using sensitivity, specificity, and the C-index.


Over a 4.12-year median follow-up, 52 patients (17.11%) experienced SGC recurrence, with 1-, 2-, and 5-year recurrence-free survival rates of 88.3%, 85.4%, and 81.6%. Key risk factors for recurrence included lymph node metastasis, Ki67 expression, histological differentiation, conjunctival pagetoid infiltration, and orbital involvement. The nomogram demonstrated strong discrimination in internal and external tests, with sensitivity and specificity values ranging from 0.722 to 0.893.


This study developed a predictive nomogram for eyelid SGC recurrence, complementing the TNM system. It offers clinical significance by aiding in the prompt identification of high-risk patients and tailoring individualized interventions, potentially improving treatment outcomes for this challenging malignancy.



Eyelid sebaceous gland carcinoma (SGC) is a malignant skin tumor that commonly originates from meibomian and Zeis gland epithelium, primarily affecting the periocular region. Globally, the annual incidence of eyelid cancer ranges from 5.1 to 15.7 cases per 100,000 individuals. In China, SGC ranks as the second most prevalent malignancy of ocular adnexa and eyelids, following basal cell carcinoma, comprising approximately 34.24% to 38.6% of cases.


Detecting SGC can be challenging as it often presents initially as diffuse thickening of the eyelid, later manifesting as isolated masses. This resemblance to conditions like blepharoconjunctivitis or chalazion can lead to misdiagnosis or delayed treatment.


One significant issue with SGC is its propensity for postoperative recurrence, attributable to its multicentric origin and pagetoid spread. Local recurrence rates can range from 5% to 18%. Poor prognostic factors include larger tumor size (>20 mm), more advanced T category, pagetoid spread, and orbital extension.


To better predict SGC recurrence and enhance patient management, it is essential to examine clinical data for recurrence risk factors and develop a personalized prediction model, such as a nomogram.


While the American Joint Committee on Cancer (AJCC) guidelines provide a basis for classifying eyelid carcinoma, other factors like pathological differentiation degree and Ki67 expression can impact prognosis. Therefore, incorporating additional potential risk factors into a nomogram scoring system can improve the accuracy of predicting patients’ recurrence-free survival rates, aiding in more tailored and effective patient care.





– Diagnosis of Eyelid Sebaceous Gland Carcinoma (SGC):

Patients who were diagnosed with eyelid SGC by the outpatient and pathology departments of Beijing Tongren Hospital, Capital Medical University.


Surgical Treatment and Pathological Examination:

Patients who underwent surgical treatment for eyelid SGC and had their surgical specimens subjected to pathological examination.


– Availability of Complete Medical Records and Pathology Data:

Patients with comprehensive medical records and pathology data pertaining to their eyelid SGC.


Exclusion Criteria

– Inaccessible or Uncooperative Patients:

Patients who could not be contacted or demonstrated uncooperative behavior.


– Surgery Conducted at Other Hospitals:

Patients who underwent surgery for eyelid SGC at hospitals other than Beijing Tongren Hospital.


– Follow-up Period Less than 6 Months:

Patients with a follow-up period of fewer than six months.


– Recurrence at Presentation:

Patients who already presented with recurrent eyelid SGC at the time of evaluation.


– History of Periocular Radiotherapy:

Patients who had previously received periocular radiotherapy.


– Incomplete Data:

Patients with incomplete or insufficient data related to their eyelid SGC.


This extensive retrospective study, conducted in accordance with the Declaration of Helsinki and approved by the Ethics Committee of Beijing Tongren Hospital, aimed to identify factors predictive of recurrence in eyelid Sebaceous Gland Carcinoma (SGC) patients. The study collected and analyzed medical records and pathological data from individuals who had undergone SGC surgeries at Beijing Tongren Hospital between January 2007 and December 2021.


A total of 304 patients met the inclusion criteria, and their surgical procedures included margin control with frozen sections. Resected tumor margins were carefully marked and subjected to immediate histopathological examination. Any presence of tumor cells led to a wider mass excision until the resection margin was devoid of tumor cells. The excision maintained a safe margin of clinically unaffected tissue, usually greater than 3 mm, and subsequent eyelid reconstruction was tailored to each patient’s specific condition. For cases involving periocular or periorbital tissue invasion, enucleation or orbital exenteration was performed. Lymphadenectomy was carried out if nodal metastasis was confirmed, with radiotherapy administered as adjuvant therapy.


Patients underwent regular postoperative follow-ups, each having at least two postoperative follow-up records. Key indicators, such as tumor recurrence and metastasis, were closely monitored during these follow-up visits, with tumor recurrence diagnosed by specialized physicians through pathological and imaging examinations.


Statistical Analysis

In this study, data analysis was conducted using R software (version 4.2.2) and SPSS software (version 25.0). The analysis involved the utilization of R packages “rms,” “foreign,” and “survival” for the development of a nomogram. Categorical variables were represented as frequency (percentage), while continuous variables with normal and skewed distributions were described as mean ± standard deviation and median (interquartile range), respectively. The Kaplan–Meier (K-M) method was employed to compute survival and recurrence rates, with the log-rank test used for comparisons. A significance threshold of p < 0.05 was set for statistical significance.


To determine the predictors to be included in the nomogram, the study initially employed univariate Cox proportional hazards regression to screen potential recurrence predictors for Sebaceous Gland Carcinoma (SGC). Subsequently, factors displaying potential statistical significance (p < 0.1 in univariate Cox regression) were included in the multivariate Cox regression analysis, using the forward stepwise method (likelihood ratio). Hazard ratios (HRs) along with their corresponding 95% confidence intervals (CIs) were recorded as part of the Cox regression analysis. The nomogram itself was established based on the results of the multivariate Cox proportional hazard regressions, allowing for the calculation of the prognostic index (PI) and the prediction of recurrence-free survival probabilities at 1, 2, and 5 years post-treatment.


The study’s dataset of 304 individuals was randomly divided into a training set (216, 70%) and an internal test set (88, 30%) utilizing the “caret” package in R. To assess the nomogram prediction model’s discriminatory ability, various metrics including the concordance index (C-index), sensitivity, specificity, and the area under the curve (AUC) of the time-dependent receiver operating characteristic (ROC) curve were calculated.


Risk factors were quantified, and their scores were summed to derive the total points equivalent to the PI. Patients were then categorized into four risk groups based on their PI scores: low, medium, high, and very high risk, defined by PI scores of <60, 60–80, 80–100, and >100, respectively. The study employed K-M analysis and the log-rank test to evaluate differences in recurrence rates among these risk subgroups. Additionally, the discrimination accuracy of the TNM staging system was compared by using the T category as the sole predictor, calculating C-index, sensitivity, specificity, and AUC. The advantages of the nomogram over the T category were further assessed through the net reclassification improvement (NRI) method. Lastly, calibration curves for 1-, 2-, and 5-year recurrence-free survival rates were generated to evaluate the alignment between predicted and actual outcomes. All analyses were two-sided, and significance was determined at p < 0.05. The bootstrap method, with 500 repetitions, was employed to yield unbiased performance estimates.



This study involved the inclusion of 304 patients, consisting of 129 (42.43%) men and 175 (57.57%) women, from Beijing Tongren Hospital. The average age of these patients at the time of their initial diagnosis was 63.00 ± 12.51 years, with a range spanning from 29 to 90 years. Most tumors were located in the upper eyelid (58.89%), followed by the lower eyelids (32.57%) and other locations, including both upper and lower eyelids, inner and outer canthal areas. Notably, some patients presented with specific characteristics such as lymph node metastasis on their initial visit, orbital involvement, or conjunctival pagetoid infiltration.


Following a median follow-up duration of 4.12 years (ranging from 1.20 to 6.50 years), 52 patients (17.11%) experienced recurrence of sebaceous gland carcinoma (SGC). The initial diagnosis, based on TNM staging, revealed 18 patients in stage T1, 22 in stage T2, 4 in stage T3, and 8 in stage T4. Furthermore, 18 patients (5.92%) developed postoperative metastasis, affecting various sites such as nasal lymph nodes, cervical lymph nodes, parotid glands, maxillofacial lymph nodes, and preauricular lymph nodes. Additionally, two patients (0.66%) had systemic metastasis, and two others presented with lymph node metastases at two different sites. The median time to recurrence was 35 months, with a range from 0 to 154 months.


The Kaplan–Meier (K-M) method was used to calculate the 1-, 2-, and 5-year recurrence-free survival rates, which were found to be 88.3%, 85.4%, and 81.6%, respectively.


To develop the nomogram for predicting SGC recurrence, initial screening of predictors involved both univariate and multivariate analyses. In the univariate analysis, factors such as lymph node metastasis at initial diagnosis, Ki67 expression, histology differentiation degree, conjunctival pagetoid infiltration, and orbital involvement were identified as potential risk factors for recurrence.


A prognostic nomogram was then constructed, incorporating these independent predictors. The predictive efficiency of this nomogram was evaluated in training and test sets, demonstrating high discriminatory ability with impressive metrics like the concordance index (C-index), sensitivity, specificity, and area under the curve (AUC) of the receiver operating characteristic (ROC) curve. Notably, the nomogram outperformed the TNM staging system in terms of discrimination accuracy, as indicated by the net reclassification improvement (NRI) results.


Furthermore, risk stratification based on patients’ prognostic index (PI) scores allowed for the differentiation of recurrence risk categories. K-M curves, plotting risk subgroups, showed substantial discrimination ability compared to those based solely on the T category of the TNM system.


In summary, this study effectively identified risk factors for SGC recurrence and developed a predictive nomogram with a strong discriminatory capacity. The nomogram’s performance surpassed that of the conventional TNM staging system, offering improved accuracy in predicting SGC recurrence risk and enhancing patient management.



This study aimed to provide a comprehensive analysis of the long-term prognosis of patients with eyelid Sebaceous Gland Carcinoma (SGC), shedding light on their clinical characteristics and recurrence patterns. The study reported recurrence, nodal metastasis, and mortality rates of SGC at 11.7%, 5.59%, and 0.99%, respectively. The recurrence rate observed in this study was found to be lower than that in a U.S. study by Shields et al. (11/60.18%) but higher than that in a Korean study by Choi et al. (2/40.5%) and fairly consistent with a U.S. study by Ford et al. (7/65.11%). These variations could be attributed to differences in sample size, surgical techniques, ethnic backgrounds, and environmental factors.


Nomograms were constructed based on five independent risk factors: lymph node metastasis at initial diagnosis, Ki67 expression, histological differentiation (well, moderately, or poorly differentiated), orbital involvement, and conjunctival pagetoid infiltration. Patients presenting with lymph node metastasis at the time of initial diagnosis had a notably high recurrence rate of 90.9%. Ki67, a marker for cellular proliferation, was identified as a prognostic marker in SGC, with higher Ki67 expression associated with poor outcomes in terms of recurrence and lymph node metastasis.


Histological differentiation played a significant role, with well-differentiated SGC cells showing a lower recurrence rate compared to poorly differentiated cells. Additionally, patients with orbital invasion had a recurrence rate of 42.6%, often resulting in adverse outcomes, including systemic metastasis or parotid lymph node metastasis. Conjunctival pagetoid infiltration was linked to increased recurrence risk due to its infiltrative nature.


The study highlighted the limitations of the traditional AJCC TNM staging system, emphasizing that factors such as pathological differentiation, Ki67 expression, and pagetoid conjunctival infiltration can significantly impact recurrence risk and should be considered in predictive models.


The developed nomogram demonstrated superior predictive accuracy compared to the T staging system, both in terms of sensitivity and specificity. It enhanced the discrimination ability in risk staging for SGC recurrence, providing valuable guidance for physicians in decision-making and patient management. For instance, patients at higher recurrence risk may benefit from more frequent follow-ups.


While this study made significant strides in predicting SGC recurrence, it acknowledged some limitations. These include a focus on 1, 2, and 5-year recurrence predictions due to limited data at year 3, a relatively small sample size, a predominantly Chinese participant group, and the absence of certain potential predictors such as radiotherapy and perineural invasion.


Oncology Related Tools


Latest Research


About Author

Similar Articles

Leave a Reply