Clinicopathological characteristics and survival of lymphoepithelial carcinoma of the oral cavity and pharynx: a population-based study

Lymphoepithelial carcinoma (LEC) of the oral cavity and pharynx is uncommon, and the characteristics and survival remains unclear. The present study aims to describe the clinicopathological characteristics and determine the factors associated with survival of this uncommon cancer. A population-based study was carried out to investigate clinical characteristics and prognosis of LEC of the oral cavity and pharynx using the data from Surveillance, Epidemiology and End Results (SEER) database between 1988 and 2013. The propensity-matched analysis was conducted for prognostic analysis, and a prognostic nomogram was also constructed. Totally, 1025 patients with LEC of the oral cavity and pharynx were identified, including 769 nasopharyngeal LEC patients and 256 non-nasopharyngeal LEC patients. The median OS of all LEC patients was 232.0m (95% CI 169.0-258.0). The 1-, 5-, 10- and 20-year survival rates were 92.9%, 72.9%, 59.3%, and 46.8%, respectively. Surgery could significantly prolong the survival time of LEC patients (P<0.01, mOS: 190m vs. 255m). Radiotherapy, as well as radiotherapy after surgery, could prolong the mOS (P<0.01 for both). The survival analysis demonstrated that old age (>60 years), lymph node (N3) and distant metastases were independent factors for poor survival, whereas radiotherapy and surgery were independent factors for favorable survival. No significant differences in survival time between nasopharyngeal LEC and non-nasopharyngeal LEC patients were observed. The prognostic nomogram was established base on five independent prognostic factors (C-index=0.70; 95% CI 0.66-0.74). In conclusion, LEC of the oral cavity and pharynx is a rare disease, and old age, lymph node and distant metastases, surgery and radiotherapy were significantly associated with prognosis. The prognostic nomogram could be used to make individual predictions of OS.


Introduction
Lymphoepithelial carcinoma (LEC) is a rare malignant tumor which is defined as a carcinoma composed of undifferentiated malignant epithelial cells surrounded or infiltrated by a prominent component of characteristic lymphocytes and plasma cells; LEC accounts for approximately 5% of head and neck cancers [1] [2]. LEC was first described in the nasopharynx in 1921, and is also a tumor mostly located in the nasopharynx, where it represents 40% of all neoplasms [3]. Aside from the nasopharynx, this disease can also occur in other locations, including the oral cavity, salivary glands, parotid glands, etc. A previous study reported that the incidence of salivary LEC is second to nasopharynx LEC, but salivary LEC is exceedingly rare and comprises only 0.4% of salivary cancers [4].
The etiopathogenesis of LEC is not fully clear, tobacco smoking and alcohol consumption are identified to be contributing factors for its development. Nasopharynx LEC is almost invariably associated with Epstein-Barr virus (EBV) infection, and human papillomavirus (HPV) has also been identified to link to LEC of the larynx, hypopharynx and oropharynx [5,6]. Previous studies demonstrated LEC has significant lymphocytic infiltration that could cause a strong immune response, thus, LEC patients were always accompanied with good prognosis [7,8]. LEC is also believed to be more radiosensitive, and radiotherapy as the single modality of treatment for locoregional LEC has been described. The combination of surgery and postoperative radiotherapy has also been recommended for some non-nasopharyngeal LEC patients [7]. However, knowledge of LEC is currently limited to case reports and small case series, especially non-nasopharyngeal LEC. The clinicopathological characteristics and survival of nasopharyngeal LEC and non-nasopharyngeal LEC of the oral cavity and pharynx have not been well defined. Therefore, we performed a retrospective analysis of patients with LEC of the oral cavity and pharynx registered in SEER database to present the clinicopathological characteristics and prognosis. The characteristics and prognoses of nasopharyngeal LEC and non-nasopharyngeal LEC were also compared. Moreover, we constructed a prognostic nomogram to help physicians make an individualized survival prediction.

Methods and Materials Participants
All patients with a diagnosis of LEC (ICD-O-3:8310/3, ICD-0-3/WHO 2008) located in the oral cavity and pharynx between 1988 and 2013 were identified from the SEER database. The inclusion criteria: patients with primary LEC as their only cancer.
The demographic information and clinicopathological characteristics of these patients were extracted using SEER*Stat 8.2.2 software including age at diagnosis, sex, race, primary site, pathological grade, SEER historic stage classification, TNM stage, and the use of surgery, radiation and chemotherapy. The overall survival information was also identified and extracted from SEER database [9,10]. The Institutional Review Board of Jinan Stomatological Hospital approved this study.

Statistical analysis
Continuous data were compared using Student's t-tests, while categorical data were examined using chi-square tests. To compare the differences in survival time between nasopharyngeal LEC and non-nasopharyngeal LEC patients, we conducted the propensity-matching (PSM) analysis with a 1:1 ratio based on age, race, pathological grade, TNM stage, and the use of surgery, radiotherapy and chemotherapy. The Kaplan-Meier method and log-rank test was used to evaluate the influence of each variable on survival time. Univariate and multivariate Cox regression survival analysis were utilized to assess the association of each variable with prognosis. The independent prognostic factors in the multivariate Cox analysis were also included to construct the prognostic nomogram. All statistical analysis was performed using MedCalc software (version 15.2.2, Mariakerke, Belgium) and R 3.1.3 software (http://www.r-project.org).
P<0.05 was considered as statistically significant.

Patients' characteristics
Totally, 1025 patients with LEC of the oral cavity and pharynx were identified.

Patient survival
The median OS (mOS) of LEC patients was 232.0 months (95% CI 169.0-258.0, Fig   1A). The (Fig 1B). Similarly, LEC patients with distant stages had significantly worse prognoses compared to individuals with localized or regional stages according to the SEER historic stage classification (P<0.01 for both); patients diagnosed with distant stage had a 5-year OS rate of 58.7%, 74.0% for localized stage and 80.0% for regional stage patients, respectively (Fig 1C).

Effect of different treatments on prognosis
As seen in Fig  prognoses compared with those with surgery alone (P<0.01) (Fig 3C).   Table   2). Subsequently, the multivariate Cox survival analysis demonstrated that only old age (>60 years), lymph node (N3) and distant metastases were independent factors for poor prognosis, whereas radiotherapy and surgery were independent factors for favorable survival in LEC patients ( Table 2).

Prognostic nomogram for LEC patients
To make an individualized survival prediction of LEC patients, we established a prognostic nomogram using all independent prognostic factors from the multivariate Cox survival analysis (Fig 4). The nomogram illustrated that the M category had the largest effect on survival, followed by age and N category. Tumor stage, race and the use of surgery and radiotherapy showed a moderate effect on prognosis. The calibration plots for the probability of overall survival at 3, 5 or 10 years in the LEC patient cohort yielded an optimal consistency between the prediction survival and the actual observation. The C-index for survival prediction in this prognostic nomogram was 0.70 (95% CI 0.66-0.74).   17, Fig 5A). Then, we carried out a PSM analysis to balance the characteristic and treatment regimens between these two groups. A total of 96 non-nasopharynx LEC patients were matched with 96 nasopharynx LEC patients (1:1) after PSM analysis (S2 Table). The survival analysis demonstrated that the survival of non-nasopharynx LEC patients was slightly better than that of nasopharynx LEC patients, but without statistically significance (166.0m vs 90.0m, P=0.20, Fig 5B). Both the univariate Cox analysis and multivariate Cox analysis revealed that location was not significantly associated with the prognosis of LEC patients.

Discussion
The majority of lymphoepithelial carcinoma studies comprises case reports or small series due to the rarity of this condition, especially non-nasopharynx LEC such as salivary gland LEC, tonsil ELC, laryngeal LEC, etc [11][12][13]. Therefore, the clinicopathological characteristics and survival for this disease are not fully clear, especially regarding the differences between nasopharynx LEC and non-nasopharynx LEC. In the present study, we present the clinicopathological characteristics and prognosis of those patients and determine the factors that affect survival based on the data of 1205 patients from the SEER database. We also conducted a comparison analysis of patients with nasopharynx LEC and non-nasopharynx LEC. Furthermore, we constructed a prognostic nomogram for patients with LEC of the oral cavity and pharynx to make an individualized survival prediction.
Nasopharyngeal LEC is the most commonly found in the oral cavity and pharynx.
Our data showed that the tumor lesions of 75.0% of LEC patients were located in nasopharyngeal area. The patients with nasopharyngeal LEC are young, with an average age of 45.8 years. Previous studies have reported several pediatric patients with nasopharyngeal carcinoma [14,15]. The nasopharyngeal LEC patients were predominantly male, with a male to female ratio of 2.73:1. Moreover, our data showed that the incidence of nasopharyngeal LEC is much higher within the nonwhite/black population than in the white population. This is attributed to the prevalence and distribution of cancer-related viruses, such as EBV and HPV [16,17]. Nasopharyngeal LEC patients can remain asymptomatic for a long time because nasopharynx is a clinically occult site. Nonspecific symptoms often caused delay a definitive diagnosis, one case series reported a mean delay period of 7.2 months [18]. Consequently, more than 90% of patients with a diagnosis of nasopharyngeal LEC present locally or regionally advanced disease [19]. Additionally, the previous studies provided those patients had a high incidence of distant metastases, ranging from 20% to 40% [20,21]. In our study, 91.5% of nasopharyngeal LEC patients had regional or distant metastases, and 81.0% of cases had lymph node metastases. Due to anatomical limitations on surgical interventions, radiotherapy is undoubtedly the preferred choice of treatment and chemotherapy is combined in advanced disease [21].  [23]. All of these studies suggested that non-nasopharyngeal LECs were often located in the oropharynx, salivary gland, tonsil, tongue, etc. Chan JY and Dubey reported that the primary occurrence was in an oropharyngeal site in the majority of cases, followed by the salivary gland, while the present study found that the salivary gland was the most common site of non-nasopharyngeal LEC [23]. In addition, the clinical characteristics of non-nasopharyngeal LEC patients from the present study are consistent with those of previous reports. Our study found that most non-nasopharyngeal patients were men, with a male to female ratio of 1.6:1. Most non-nasopharyngeal LEC patients were aged younger than 60 years (62.5%) and were white (72.7%). Lymph node metastases in non-nasopharyngeal LEC are common with the incidence ranged from 10% to 50%. In the present study, the incidence of lymph node metastases was 70.2%, which is much higher than that from previous studies. One possibility for this discrepancy is that the lymph node status of 125 out of the 256 (48.8%) non-nasopharyngeal LEC patients was unknown in the present study. In addition, most non-nasopharyngeal LEC patients had advanced-stage (III/IV) disease, ranging from 59.4% to 80.2% at diagnosis in these studies. Our data reported that 78.2% patients were diagnosed with an advanced stage. In our cohort, 186 patients were white; in contrast, 53 cases of advanced-stage disease occurred in patients who were nonwhite, including American Indian/AK Native, Asian/Pacific Islander. Our findings support those of Chan JY, who identified that 23.7% of their non-nasopharyngeal LEC patients were nonwhite/black.
A strong association of EBV with LEC has been reported in Southeast Asia, Greenland, and Alaska, but not in the white population, especially for salivary gland LEC [7,24,25]. Several previous studies from high incidence regions have observed EBV positivity in non-nasopharyngeal LEC cases; for example, Ma reported that all 38 Chinese salivary gland LEC with EBV encoded RNA positive [7]. One potential explanation is associated with the geographic distribution of the 2 major types of EBV. In accordance with other reports, we found that patients with LEC of the oral cavity and pharynx had a much better prognosis. Almost 50% of the LEC patients could survive for 20 years in our cohort. In addition, the survival analysis in both unmatched cohort and matched cohort did not yield any significant difference in survival time between nasopharyngeal LEC and non-nasopharyngeal LEC patients. Therefore, we constructed a prognostic nomogram for individualized survival prediction of LEC patients using the long-term follow-up data from the SEER database. With this easy-to-use scoring system, both physicians and patients could calculate the survival probability of individual LEC patients. When a prognostic nomogram is completed, the validation of the nomogram is essential to avoid overfitting the model and determine its generalizability [27]. [28,29].
For common cancers, validation of the prognostic nomogram should be performed in the primary cohort and an independent cohort. However, this prognostic nomogram for LEC patients can only be validated in the primary cohort due to the rarity of LEC. In addition, several important prognostic indices were not included in this prognostic nomogram such as serum tumor markers. Further studies with more comprehensive information are required to confirm the accuracy of this nomogram.
In the present study, we described the clinicopathological characteristics and survival of patients with LEC of the oral cavity and pharynx. The results showed that patients with LEC of the oral cavity and pharynx often had a favorable prognosis. Old age, lymph node and distant metastases, surgery and radiotherapy were significantly associated with prognosis. Non-nasopharynx LEC patients were more likely to be young and male and have an undifferentiated grade and late TNM stage, but no any significant difference in prognosis between nasopharyngeal LEC and non-nasopharyngeal LEC patients could be observed. Meanwhile, we also constructed the first prognostic nomogram to predict the individual survival. In conclusion, the present study is the largest series concerning on LEC of the oral cavity and pharynx, and these results are vital to disease management and future prospective studies for this rare cancer.