Volume 82, Issue 8 p. 1556-1562
Original Article
Free Access

Nonnasopharyngeal lymphoepithelioma of the head and neck

Praveen Dubey M.D.

Praveen Dubey M.D.

Department of Radiation Oncology, the University of Texas M. D. Anderson Cancer Center, Houston, Texas

Search for more papers by this author
Chul S. Ha M.D.

Corresponding Author

Chul S. Ha M.D.

Department of Radiation Oncology, the University of Texas M. D. Anderson Cancer Center, Houston, Texas

University of Texas M. D. Anderson Cancer Center, 1515 Holcombe Blvd., Box 97, Houston, TX 77030===Search for more papers by this author
K. Kian Ang M.D., Ph.D.

K. Kian Ang M.D., Ph.D.

Department of Radiation Oncology, the University of Texas M. D. Anderson Cancer Center, Houston, Texas

Search for more papers by this author
Adel K. El-Naggar M.D.

Adel K. El-Naggar M.D.

Department of Pathology, the University of Texas M. D. Anderson Cancer Center, Houston, Texas

Search for more papers by this author
Calvin Knapp M.D.

Calvin Knapp M.D.

Department of Head and Neck Surgery, the University of Texas M. D. Anderson Cancer Center, Houston, Texas

Search for more papers by this author
Robert M. Byers M.D.

Robert M. Byers M.D.

Department of Head and Neck Surgery, the University of Texas M. D. Anderson Cancer Center, Houston, Texas

Search for more papers by this author
William H. Morrison M.D.

William H. Morrison M.D.

Department of Radiation Oncology, the University of Texas M. D. Anderson Cancer Center, Houston, Texas

Search for more papers by this author

Presented at the 79th Annual Meeting of the American Radium Society, New York, New York, April 30-May 4, 1997.

Abstract

BACKGROUND

Lymphoepithelioma (squamous cell carcinoma with associated lymphoid stroma) commonly occurs in the nasopharynx, rarely at other sites. As a result, the clinical course and optimal treatment of nonnasopharyngeal lymphoepithelioma of the head and neck have not been well described. This retrospective study was undertaken to analyze the clinical course of the disease in patients treated at a single institution and to formulate recommendations for treatment based on that experience as well as results reported in the literature.

METHODS

Between 1950 and 1994, 34 patients with nonnasopharyngeal lymphoepithelioma of the head and neck were treated at the University of Texas M. D. Anderson Cancer Center. The patients' medical records were reviewed and their pathologic specimens evaluated. The primary tumor sites were: oropharynx (24 patients), salivary gland (4), laryngohypopharynx (4), and the maxillary sinus/nasal cavity (2). Assessed in accordance with the 1992 American Joint Committee Against Cancer TNM staging system, T classifications were TX-2, T1-7, T2-8, T3-10, and T4-7, and N classifications were N0-8, N1-5, N2-15, and N3-6. Treatment consisted of radiotherapy for 24 patients, excisional biopsy of the primary tumor followed by radiotherapy for 7 patients, and surgery for 3 patients. Of the patients treated with radiotherapy, neck dissections were performed on only two, both of whom had persistent lymph node masses after completing radiotherapy. The median dose delivered to the primary tumor was 65 gray (Gy) (range, 46-78 Gy). The median fraction size was 2.1 Gy (range, 1.6-3.2 Gy).

RESULTS

The 5-year actuarial disease specific survival and overall survival rates were 59% and 39%, respectively. The 5-year actuarial local control rate for all patients was 94%. For the irradiated patients, the 5-year regional control rates were 77% overall and 83% within the radiation field. The 5-year actuarial rate of distant metastasis for all patients was 30%. For patients who presented with and without regional adenopathy, the 5-year rates of distant metastasis were 36% and 12%, respectively (P = 0.27).

CONCLUSIONS

Nonnasopharygeal lymphoepithelioma is a radiosensitive disease. High rates of locoregional tumor control were achieved with radiotherapy at all head and neck sites. The main cause of treatment failure was distant metastasis, which occurred more frequently in patients with lymph node involvement. Radiotherapy is appropriate initial locoregional therapy for patients with this disease. Surgery should be reserved for patients who have persistent disease after completing radiotherapy. Systemic therapy is a reasonable approach for patients who present with regional adenopathy because they have a relatively high rate of distant metastasis. Cancer 1998;82:1556-62. © 1998 American Cancer Society.

Lymphoepithelioma (LE), also known as squamous cell carcinoma with associated lymphoid stroma, occurs primarily in the nasopharynx. However, it has also been reported to present in other sites of the upper aerodigestive tract, including the major salivary glands, oral cavity, oropharynx, larynx, and hypopharynx.1-7 Because of the rarity of this tumor, the clinical course and optimal treatment for nonnasopharyngeal LE of the head and neck has not been well described. This retrospective study was undertaken to analyze the University of Texas M. D. Anderson Cancer Center experience with this disease and to formulate treatment recommendations based on that experience as well as the results reported in the literature.

METHODS

Patients

Between 1950 and 1994, 34 patients with LE of the head and neck whose primary tumor site was located outside of the nasopharynx underwent surgery and/or radiotherapy with curative intent at our institution. Pathologic specimens from all patients were rereviewed and confirmed as LE. Cases of LE metastatic to the neck from an unknown primary were excluded, as the nasopharynx was felt to represent the most likely site of the primary tumor.

Pretreatment evaluations included complete clinical examination, blood count, serum chemistry, and chest radiograph. Imaging of the head and neck was performed routinely after 1979 with computed tomography scans, magnetic resonance imaging, or both. Additional staging studies were conducted if indicated by clinical findings. Patients were retrospectively restaged using the criteria established by the American Joint Committee Against Cancer in 1992.8 Table 1 shows the T and N classifications of the patients. The two TX lesions represented patients who had tonsillar masses excised prior to presentation at the M. D. Anderson Cancer Center.

Table 1. Distribution of T and N Classifications
T N0 N1 N2 N3 Total
T1 2 0 5 0 7
T2 1 1 5 1 8
T3 2 2 3 3 10
T4 2 2 2 1 7
TX 1 0 0 1 2
Total 8 5 15 6 34

The median age at presentation was 60 years (range, 30-80 years). There were 22 men and 12 women. With regard to race, 24 patients were white, 5 were Hispanic, 4 were African American, and 1 was of Asian descent. The sites of the primary tumors were as follows: the tonsil (in 16 patients), the base of the tongue (in 8), the parotid gland (in 2), the submandibular gland (in 2), the supraglottic larynx (in 2), the hypopharynx (in 2), the nasal cavity (in 1), and the maxillary sinus (in 1). A neck mass was the only presenting sign in 10 patients. In another 10 patients, pain caused by the primary tumor was the only symptom at diagnosis. Other signs and symptoms at presentation included a visible mass at the primary tumor site, dysphagia, hoarseness, and otalgia. The incidence of lymph node involvement at diagnosis was 76%. Tumor morphology was documented in 24 of 29 clinically assessable primary tumors; the primary lesions were described as ulcerative, exophytic, and infiltrative in 10, 8, and 6 patients, respectively. Morphology was not assessable in patients with salivary gland or paranasal sinus tumors.

The median follow-up for living patients was 7.5 years. Complete follow-up information was available for all but 1 patient, who died 1 year after completing treatment. He was considered to have died of local recurrence. Actuarial curves were generated using the Kaplan-Meier method,9 starting from the date of initiation of therapy. Univariate analysis was performed using the log rank method to compare subgroups for significant differences.

Treatment

Three different strategies were used to treat the patients in this study. Twenty-four patients were treated with radiation therapy only; two of these patients had neck dissections after radiation therapy because they had residual lymph node masses. Seven patients, two of whom presented with clinical neck disease, underwent excision of the primary tumor followed by radiation therapy. The other three patients, two of whom had primary tumors in the supraglottic larynx and one of whom had the primary in the hypopharynx, were treated with surgery only (laryngectomy, partial pharyngectomy, and neck dissection); one of these patients was also given adjuvant chemotherapy postoperatively.

Twenty-eight patients were irradiated with either cobalt-60 or 6-MV photons. One of these 28 patients was treated with a mixture of photons and neutrons. The other three patients were treated in the early 1950s with orthovoltage X-rays (250 kVp). One patient, who was treated with orthovoltage X-rays for a carcinoma in the base of the tongue classified as T4, received a 40-gray (Gy) primary tumor boost with an interstitial implant.

The two patients with paranasal sinus and nasal cavity tumors received no neck treatment. Seven patients who had well-lateralized primary tumors (in the salivary gland in four patients and in the tonsil in three) were given only ipsilateral neck treatment. All other patients underwent bilateral neck irradiation. Macroscopic lymph node disease was treated with similar doses as the primary tumor in all but two cases. In both cases, the regional treatment strategy was to irradiate the neck to 50 Gy and follow this with a planned neck dissection.

Figure 1 (6K) contains two scattergrams showing fractionation data of the irradiated patients. Two data points overlap on Figure 1a (6K). Relatively large fraction sizes were used to treat some patients in the earlier years of the experience. For patients who received radiotherapy, the median dose given to the primary tumor was 65 Gy (range, 46-78 Gy). The median fraction size was 2.1 Gy (range, 1.6-3.2 Gy). Radiation was delivered with fractionation once daily and with concomitant boost fractionation10 in 28 and 3 patients, respectively. Due to severe acute mucositis, 3 patients had treatment interruptions for 7, 7, and 21 days, respectively.

Details are in the caption following the image Details are in the caption following the image

(a) The primary dose by the number of fractions is shown. Two patients are excluded; one of them had an interstitial boost and the other received neutron therapy. (b) The primary dose by treatment days is shown, with the same two patients excluded.

RESULTS

Local and Regional Control

Local control for the various treatment strategies is shown in Table 2. Actuarial 5-year local control rates for all patients and for the 31 irradiated patients were 94% and 93%, respectively. Only 2 of the 34 patients were considered to have had local recurrences. One patient who was disease free at 6 months and died 1 year after treatment was arbitrarily scored as having had a local failure. The other patient, who was treated in 1959, had stable disease at the primary site after receiving 29 Gy; he was deemed to have experienced therapeutic failure and began to receive concurrent 5-fluorouracil and radiation. He died of neutropenic sepsis during his treatment.

Table 2. Local Control
Classification
Treatment n TX T1 T2 T3 T4 Total
XRT only 24 - 4/4 (0)a 4/4 (3) 5/6 (2) 1b/2 (3) 14/16 (8)
Primary tumor excision and XRT 7 2/2 (0) 3/3 (0) 1/1 (0) - 1/1 (0) 7/7 (0)
Surgery only 3 - - - 2/2 (0) 1/1 (0) 3/3 (0)
  • XRT: radiotherapy.
  • a Nos. in parentheses refers to the no. of patients who could not be evaluated for local control. Patients whose disease remained in local control but who had less than 18 months of follow-up due to death from distant metastases or intercurrent disease could not be evaluated for local control.
  • b The patient's treatment included an interstitial implant.

Actuarial 5-year regional control rates for all patients and for irradiated patients were 75% and 77%, respectively. Within the irradiated field, the 5-year actuarial regional control rate was 83%. The median time to regional failure was 7 months (range, 0-25 months). Table 3 shows the regional control rates stratified by treatment strategy for the neck.

Table 3. Regional Control
Neck treatment strategy n Could not be evaluateda Control after planned therapy Control in XRT field Neck recurrence outside of XRT field
Bilateral whole-neck XRT 22 17 13/17b 13 0
Partial-neck XRT for oropharyngeal primaries 3 2 0/2 2 2
Major salivary gland tumors: ipsilateral neck XRT 4 4 4/4 4 0
Observation 2 2 2/2 - -
Neck dissection only 3 3 2/3 - -
  • XRT: radiotherapy.
  • a Patients could not be evaluated for regional control if their disease remained in continuous regional control but they had less than 18 months of follow-up due to death from distant metastases or intercurrent disease.
  • b Two patients who had neck dissections for residual lymph node masses after the completion of radiotherapy are included. The pathologic specimens from both neck dissections contained no tumor.

Twenty-two patients had bilateral neck and supraclavicular irradiation. The 5-year actuarial regional control rate for these patients was 76%. Two patients who had residual neck masses after completing radiotherapy had neck dissections; neither of the pathologic specimens contained any disease. These two patients were considered to have had their disease regionally controlled by radiotherapy. Three of the four patients who had regional recurrences had distant metastases detected simultaneously.

Two patients whose primary tumors in the nasal cavity and maxillary antrum were treated by radiotherapy did not have any neck treatment; in both patients, the disease remained in regional control. Seven other patients had less than whole-neck radiotherapy. The regional treatment policy for the four patients with primary tumors of the salivary glands was ipsilateral neck irradiation; in all of them, the disease remained in regional control. The initial three patients with oropharyngeal primary tumors, all treated before 1955, were treated with various partial neck fields. Two of these patients developed recurrent neck disease outside of the radiation portals.

Distant Metastases

The 5-year actuarial incidence of distant metastasis was 30%. Distant metastases occurred in nine patients and usually involved multiple sites, including liver, lung, and bone; four of these nine patients also had regional recurrences. The median interval between the initiation of therapy and the development of distant metastases was 7 months (range, 2-32 months). For patients whose disease remained in locoregional control, the 5-year rate of distant metastasis was 23%. The 5-year actuarial rates of distant metastasis for patients who presented with and without regional adenopathy were 36% and 12%, respectively (P = 0.27).

Disease Specific and Overall Survival

The actuarial disease specific survival (DSS) and overall survival rates at 5 years for all patients were 59% and 39%, respectively (Fig. 2 (2K)). The 5-year DSS rates for patients who presented with and without regional adenopathy were 53% and 80%, respectively (P = 0.13).

Details are in the caption following the image

Actuarial disease specific survival and overall survival are shown.

Treatment-Related Complications

There were two treatment-related deaths. One patient, who was treated in 1960, died 2 months after the completion of treatment from persistent dysphagia and dehydration. The other death was due to neutropenic sepsis (see the "Local and Regional Control" section). Grade 3 and 4 late complications included osteoradionecrosis (in 2 patients) (both were successfully managed with conservative measures), neck fibrosis (in 1 patient), tinnitis (in 1), and esophageal stricture (in 1). One of the three patients who underwent laryngectomy developed a tracheoesophageal fistula.

DISCUSSION

Lymphoepithelioma is a clinicopathologic entity that was described simultaneously but independently by Schminke and Regaud in 1921. In 1929, Ewing11 concurred that LE should be considered a separate category of nasopharyngeal carcinoma; he noted that the distinguishing feature of these tumors was their radiosensitivity.

It has been firmly established that the lymphocytes in the lymphoid stroma are neither neoplastic nor integral to the carcinomatous process.12 In 1978, the World Health Organization published a histologic typing system for nasopharyngeal carcinoma based on the appearance of the malignant epithelial cells.13 In this system, all three categories of nasopharyngeal tumors can be associated with a lymphoid stroma.12

Oropharyngeal LE is a rare disease. Between 1960 and 1983, only 13 cases of oropharyngeal LE were seen at the Mayo Clinic.1 At the Institut Gustave-Roussy, 18 patients were treated who had undifferentiated carcinoma of the nasopharyngeal type (UCNT) located in the tonsillar fossa;14 these 18 cases were identified in a retrospective pathologic review of 2262 patients with carcinoma of the tonsillar region. The report does not state whether a lymphoid stroma was present in these cases. Twenty-four of the 34 patients in our report had oropharyngeal LE.

Outside of Waldeyer's tonsillar ring, mucosal nonnasopharyngeal LE is even rarer. At the Mayo Clinic, 4 cases of laryngohypopharyngeal LE were seen between 1907 and 1984.2 Micheau et al.4 identified 3 cases of laryngeal LE in a review of 2430 laryngectomy and pharyngectomy specimens. Ferlito5 reviewed the pathologic specimens of 2052 laryngohypopharyngeal neoplasms and found only 1 case of LE of the hypopharynx.

Epstein-Barr virus (EBV) has been implicated in the pathogenesis of LE of the nasopharynx and some other sites.15 EBV has been closely linked to nasopharyngeal LE and undifferentiated carcinomas by several serology, immunofluorescence, and nucleic acid hybridization studies. Homogeneous EBV genome terminal fragment has been demonstrated in the neoplastic cells of undifferentiated carcinoma of the nasopharynx. EBV has also been definitely associated with LE-like carcinomas located in the stomach, salivary gland, lung, and thymus. None of our patients had special studies performed in which an EBV association was sought.

Salivary gland LE has been reported to occur almost exclusively in Greenland Eskimo, North American Eskimo, and Chinese patients.7, 16 In Greenland, over 90% of the salivary gland malignancies are LE. Canadian Eskimos have an incidence of salivary gland malignancies that is 30 times higher than the expected rate, with the majority of these tumors classified as LE. A very high percentage of the Eskimo and Chinese salivary gland LEs are associated with EBV.15, 17 None of our patients who had salivary gland LE were Chinese or Eskimos.

Two recently completed Phase III trials performed on patients with nasopharyngeal carcinoma are of interest. The Intergroup trial randomized patients with nasopharyngeal carcinoma to receive either radiation only or radiation given concurrently with cisplatin followed by postradiation chemotherapy.18 Forty-two percent of the patients in the trial had LE. The 3-year survival rates for the radiotherapy arm and combined modality arms were 44% and 76%, respectively (P = 0.001). The rates of distant metastasis in the two study arms and the impact of pathologic category have not been reported yet. Another Phase III trial, conducted by the International Nasopharynx Cancer Study Group,19 randomized patients with Stage IV UNCT to receive either neoadjuvant chemotherapy (bleomycin, epirubicin, and cisplatin) followed by radiotherapy or radiotherapy only. Patients in the combined modality treatment arm had a significant improvement in disease free survival (P < 0.01) but not in overall survival. The crude incidence of distant metastasis as a type of first progression was reduced in the combined modality arm compared with the radiotherapy arm (18% vs. 32%, respectively). The statistical significance of this reduction in distant metastasis was not reported.

Patients who present with LE of the head and neck with regional disease have a high rate of distant metastasis. In our study, the 5-year actuarial rate of distant metastasis in patients who presented with regional adenopathy was 36%. Bansberg et al.1 reported that 4 of 10 nonnasopharyngeal LE patients who presented with regional adenopathy developed distant metastases. Other nonnasopharyngeal LE series have been smaller but have reported the occurrence of distant metastases in their patients. Mesic et al.20 reported that the crude incidence of distant metastasis in 113 patients with nasopharyngeal LE treated at the M. D. Anderson Cancer Center was 35%; distant metastases developed in 43% of patients who presented with N2 and N3 disease.

In our series, the incidence of regional adenopathy at the time of diagnosis was 76%. In the reports of oropharyngeal LE of Bansberg et al.,1 the incidence of regional lymph node involvement at presentation was 77%. In the earlier years of our series, patients with oropharyngeal LE were treated with partial neck irradiation. Two of these patients had recurrences in the unirradiated lower neck and contralateral neck. We recommend that patients with oropharyngeal LE be treated with bilateral whole-neck irradiation unless they have small, well-lateralized tumors of the tonsillar fossa. In our four patients with salivary gland LE, who were treated with ipsilateral neck irradiation only, the disease remained in regional control. Our strategy for postoperative irradiation of salivary gland malignancies is to treat the unilateral neck only; our limited data suggest that unilateral neck irradiation is also a reasonable approach to treating salivary gland LE.

Previous reports have documented that nasopharyngeal LE is a highly radiosensitive disease for which excellent local control rates can be achieved with radiotherapy.20, 21 In the report of oropharyngeal LE of Bansberg et al.,1 only 1 of 11 patients treated with definitive radiotherapy had a local recurrence; the median radiation dose was 55 Gy (range, 45-60 Gy). The fraction size used and the duration of the radiation course were not reported. In our series, the local control rate was very high with a variety of fractionation schedules and treatment durations. Thus, organ preservation should be possible for all patients with LE primary tumors in the head and neck. The two treatment-related deaths could probably have been averted with the aggressive supportive care now available.

In summary, our data suggest that the clinical features of nonnasopharyngeal LE parallel the features of its nasopharyngeal counterpart. Patients with nonnasopharyngeal LE have a high incidence of lymph node metastasis and a propensity for distant metastasis. Our locoregional treatment strategy for patients with nonnasopharyngeal LE will be to irradiate all the primary tumors. Given the excellent local control rates that were achieved using widely differing fractionation schemes, we cannot formulate clear dose recommendations from our data. Our future strategy will be to treat T1, T2, and moderate-size T3 tumors with a dose of 66 Gy in 33 fractions. For large T3 and T4 tumors, we recommend delivering 69 Gy over 6 weeks using a concomitant boost fractionation schedule. Because the disease is radiosensitive, we favor treating the regional lymph node disease with radiotherapy; neck dissections should be performed only on those patients who have persistent lymph node disease 6 weeks after completing treatment. Our data suggest that patients who present with regional adenopathy develop distant metastasis at a moderately high rate. Extrapolating from the data of Phase III trials of nasopharyngeal carcinoma, we recommend that patients who present with regional adenopathy be treated with neoadjuvant chemotherapy in an attempt to decrease the rate of distant metastasis.

Acknowledgements

The authors acknowledge the statistical analysis performed by Pamela Allen, M.P.H.