Primary Malignant Lymphoma of the Parotid Gland

Objectives  To document the clinicopathologic features of primary malignant lymphoma of the parotid gland based on analysis of our cases and to compare the results with similar studies in the literature.

Design  Retrospective, nonrandomized case study.

Setting  Academic, tertiary medical center.

Patients  Forty-one consecutive cases of malignant lymphomas of the parotid gland were identified among 820 patients who had undergone parotid surgery during the course of 22 years. Thirty-three (80%) of these were primary lymphomas and were included in the study. Eight (20%) occurred in patients with a history of malignant lymphoma and were therefore excluded.

Intervention  Diagnosis was established by open parotid biopsy in 8 patients, superficial lobectomy in 23, and total parotidectomy in 2. After diagnosis, lymphomas were staged and treated with local irradiation and/or chemotherapy.

Results  Fifteen men and 18 women aged 26 to 100 years (mean, 66 years) had an enlarging painless mass on initial examination. Seven (21%) had an underlying autoimmune disease and 20 (61%) had Ann Arbor stage I disease at diagnosis. Of 25 patients available for a minimum 2-year follow-up, 16 (64%) were alive with or without disease. Histological grade was the only prognostic feature associated with outcome (P<.01).

Conclusions  Our study, when viewed collectively with those in the literature, indicates that malignant lymphomas of the parotid gland are uncommon and often not suspected clinically. The disease affects both sexes equally and is unusual before the age of 50 years. Most are B-cell, non-Hodgkin lymphomas, and about 80% of patients have Ann Arbor stage I or II disease at diagnosis.

PRIMARY malignant lymphomas of the salivary glands are uncommon, comprising only 1.7% to 3.1% of all salivary neoplasms^1-3 and 0.6% to 5% of all tumors and/or tumorlike lesions of the parotid gland.^1,4-8 Although the parotid gland is by far the most common site of origin, malignant lymphomas may also involve, in descending order of frequency, the submandibular gland, minor salivary glands, and sublingual gland (Table 1). ^2,3,9-13

In the parotid gland, malignant lymphomas are often clinically unsuspected, manifesting as nonspecific masses indistinguishable from other more common epithelial tumors. We therefore decided to review our experience with parotid lymphomas during a span of 22 years to see if they exhibited a characteristic clinicopathologic profile and, if so, whether there were any clinical features that might suggest the diagnosis before surgery.

A retrospective review of the clinical records and anatomical pathology files of the Eye and Ear Institute and the University of Pittsburgh Medical Center, Pittsburgh, Pa, from September 1, 1972, through October 31, 1994, disclosed 820 patients who had undergone surgery of the parotid gland for the presence of a tumor or tumorlike mass. Of these, 542 patients (66%) had benign lesions and 278 (34%) had malignant lesions. Forty-one (5%) of the 820 cases were malignant lymphomas and, of these, 8 were from patients who had a history of malignant lymphoma and were therefore excluded. The remaining 33 cases were considered to be primary lymphomas and are the subject of this study.

The medical records of these 33 patients were reviewed for the following features: age, sex, initial symptoms, physical findings, extent of surgical procedure, stage of disease, and histological type of lymphoma. The Ann Arbor system (Table 2) was used for staging and the Working Formulation was used for histological classification (Table 3). ^14,15 Cases that did not conform to the Working Formulation were reviewed by one of us (L.B.) and reclassified accordingly.

Follow-up was obtained from medical records, from a tumor registry, or by letter or telephone contact with the attending physician, patient, or family member. Data were analyzed statistically by χ² and 2-tailed Fisher exact tests when appropriate, with a level of significance of P<.05. Multiple logistic regression of variables and Spearman rank correlation were also performed.

There were 15 men and 18 women with a mean age of 66 years (range, 26-100 years). Twenty-four patients (73%) were between 50 and 80 years of age, while only 4 (12%) were younger than 50 years.

The most common clinical manifestation was a progressively enlarging, nonfixed mass that was painless in 30 patients (91%) and painful in 3 others (9%). Only 3 patients (9%) had coexistent enlarged cervical lymph nodes. Five patients (15%) exhibited facial nerve paresis, but of these, only 1 complained of pain.

The stage of disease was I in 20 patients (61%), II in 7 (21%), III in 2 (6%), and IV in 4 (12%) (Table 2). Seven patients (21%) had a history of an autoimmune disease (Sjögren syndrome in 5 and rheumatoid arthritis in 2), and at least 3 (11%) of 28 patients had other coexistent malignant neoplasms (squamous cell carcinoma of the lung in 1, adenocarcinoma of the breast in 1, and adenocarcinoma of the colon in 1).

Although some of our cases predated the immunohistochemical era, the diagnosis of malignant lymphoma was confirmed in 28 of the 33 cases in which the slides were available for review. The remaining 5 cases had been thoroughly evaluated by appropriate immunohistochemical stains, and the stated diagnosis was therefore accepted. The lymphomas were all of the non-Hodgkin type and included 8 variants, the most common of which was the follicular, small cleaved lymphoma, which accounted for 14 (42%) of the cases (Table 3). Sixteen tumors were nodular (follicular) and 17 were diffuse. Twenty (61%) of the lymphomas were low grade according to the Working Formulation, 15% were intermediate grade, and 24% were high grade (Table 3).

Of 28 cases in which the slides were available for examination, the lymphoma was confined to the parenchyma of the parotid gland in 17 (61%), and in 1 of these the cervical lymph nodes were also affected. Seven lymphomas (25%) involved only the intraparotid lymph nodes, and in 1 of these the cervical lymph nodes were also diseased. Four cases (14%) involved both the parenchyma and intraglandular lymph nodes, and in 1 of these the cervical lymph nodes were also involved.

Three (11%) of the 28 cases also showed histological evidence of "benign lymphoepithelial lesion"; all 3 of these were from patients with clinical evidence of Sjögren syndrome.

The diagnosis of malignant lymphoma of the parotid gland was established by open biopsy in 8 patients (24%), superficial lobectomy in 23 (70%), and total parotidectomy in 2 (6%). Five patients who underwent superficial lobectomy also had a cervical lymph node biopsy.

Three patients had a fine-needle aspiration of their salivary tumor before surgery; of these, 2 specimens were interpreted as benign and 1 as probably malignant. Seven (21%) of 33 patients had a frozen section performed; of these, 5 were interpreted as malignant and 2 as possibly malignant.

After diagnosis, the patients were referred to an oncologist for staging and treatment. Staging during the 22-year time interval varied according to existing medical dogma and technology, but it generally included a thorough history and physical examination, complete blood cell count with bone marrow aspirate and biopsy, liver function tests (usually total bilirubin, alkaline phosphatase, aspartate aminotransferase, and alanine aminotransferase), and computed tomography of the chest, abdomen, and pelvis. A few patients had lymphangiograms and random needle biopsies of the liver, but none underwent staging laparotomies.

Specific details of therapy were not always available and, moreover, are beyond the scope of this article. In general, patients with localized disease were treated with postoperative radiation, while those with multifocal or systemic disease received chemotherapy.

A minimum 2-year follow-up was available in 25 of the 33 patients (Table 4).

Statistical evaluation of selected variables showed that only histological grade (high vs intermediate and low) of the Working Formulation was clinically significant (P<.01); the higher the grade, the worse the prognosis. Stage (I vs II, III, and IV) almost achieved significance (P<.07). Sex (P<.14), age at onset (P<.86), presence of an autoimmune disorder (P<.59), and specific histological type of lymphoma (P<.39) were not important prognostic variables.

A mass in the parotid gland may be the first manifestation of malignant lymphoma or it may represent a secondary focus of disease in a patient with a known history of malignant lymphoma. In our series of 41 parotid lymphomas, 33 (80%) were primary and 8 (20%) were secondary. These statistics on the frequency of primary and secondary malignant lymphomas, however, may be misleading, since the occurrence of a parotid mass in a patient with a history of malignant lymphoma is often assumed to be lymphomatous and, therefore, not excised for pathological evaluation.

The majority (84%-97%) of all parotid lymphomas are of the non-Hodgkin type and of B-cell origin, either nodular or diffuse.^{2,3,10,12,16,17} Only 3% to 16% are caused by Hodgkin disease and, while T-cell lymphomas have been described, they are exceptional.¹⁷

The lymphoma may arise in intraglandular lymph nodes normally found in the parotid gland or from the parenchyma (mucosa-associated lymphoid tissue [MALT]) or both. In our study, of the 28 cases in which slides were available for examination, 7 (25%) of the lymphomas arose in intraparotid lymph nodes, 17 (61%) from the parenchyma, and 4 (14%) from both sites.

The distinction of nodal vs parenchymal (MALT) origin of a salivary lymphoma has some clinical significance, since many of the parenchymal (MALT) lymphomas tend to be low grade, are localized at the time of diagnosis and often remain so for extended periods, frequently are associated with a "benign lymphoepithelial lesion," and are often curable.^18-25 A few, however, may disseminate to lymph nodes or other MALT sites or even transform to a higher-grade lymphoma.²⁶

One of the deficiencies of the Working Formulation classification of malignant lymphomas is that it does not contain a specific category for MALT lymphomas (Table 3). Since MALT lymphomas may have a range of cellular sizes, they are often spread among several categories of the Working Formulation, including small lymphocytic, small cleaved, and mixed small and large cell.²² The new Revised European-American Lymphoma (REAL) classification, however, does take MALT lymphomas into consideration.^27,28

There is considerable confusion in the literature on the use of the terms primary and secondary, and nodal and extranodal, in regard to malignant lymphomas of the parotid gland.^2,29,30 We use the term primary for any malignant lymphoma that first manifests in the parotid gland, regardless of the subsequent stage of the disease, whether it arises in the parenchyma or intraglandular lymph nodes, or whether it is associated with an autoimmune disorder. When a parotid lymphoma develops in a patient with a known history of malignant lymphoma, we arbitrarily designate it as secondary. Others use the term primary only when the lymphoma originates in the parenchyma (MALT) as opposed to intraparotid lymph nodes. Still others apply the term primary only when the lymphoma appears de novo and secondary when it is associated with an underlying autoimmune disorder, such as Sjögren syndrome. While some investigators distinguish nodal from parenchymal lymphomas, others consider all parotid lymphomas, regardless of site of origin, as extranodal and indicate that they constitute about 8% of all extranodal lymphomas of the head and neck.^31,32 These inconsistencies in terminology and the ever-changing histological classification of malignant lymphomas make comparison of various studies on salivary lymphomas difficult.

Some investigators have noticed a progressive increase in the incidence of salivary lymphomas during the last several decades.^3,16 This may result from a combination of factors: (1) referral patterns, (2) increased life span of individuals who are at risk for developing the disease, and (3) the realization that with the use of sophisticated molecular techniques, many of the "benign lymphoepithelial lesions" are actually subtle examples of lymphoid malignant neoplasms.

Malignant lymphomas of the parotid gland are uncommon in patients younger than 50 years. The peak age of occurrence is between 50 and 80 years, with a mean or median age in most series of 55 to 65 years (range, 20-100 years).^2,3,7-13,22 Although various studies have indicated a male-female ratio varying from 2:1 to 1:1 to 1:2, our collective review of 269 cases, including 33 in the current series, shows the sex distribution to be about equal: 125 men (46%) and 144 women (54%).^2,3,7-13,33

Most patients with malignant lymphomas of the parotid gland show a progressively enlarging, unilateral, painless mass of 0.5 to 8.0 cm in greatest dimension associated with enlarged cervical lymph nodes in anywhere from 9% to 69% of cases.^2,3,7,11,33 Although bilateral parotid lymphomas have been described in 4% to 21% of patients, it is not always clear whether these patients actually have bilateral disease or whether they have a unilateral lymphoma associated with bilateral "benign lymphoepithelial lesion." In rare instances, a patient may have a simultaneous malignant lymphoma involving both the parotid and submandibular glands.¹¹

Other unusual signs and symptoms that have been described are pain (9% in this series), facial nerve paresis (4% to 15%), and fixation of the mass to the overlying skin or deep tissues.^2,3,7 The incidence of finding an underlying coexistent autoimmune disorder, such as Sjögren syndrome and rheumatoid arthritis, has ranged from 0% to 44%.^2,8-10,13,33

At the time of diagnosis, most patients are found to have Ann Arbor stage I (51%) or II (30%) disease and low-grade (52%) to intermediate-grade (37%) malignant lymphomas according to the Working Formulation (Table 5 and Table 6).

Our review indicates that primary malignant lymphoma of the parotid gland is rarely suspected preoperatively. The following clinical features might suggest the diagnosis: development of a parotid mass in a patient with a known history of malignant lymphoma; occurrence of a parotid mass in a patient with an immune disorder, such as Sjögren syndrome, rheumatoid arthritis, or acquired immunodeficiency syndrome; occurrence of a parotid mass in a patient with a previous diagnosis of "benign lymphoepithelial lesion"; multiple masses in a unilateral parotid gland or bilateral parotid masses; or parotid mass associated with multiple, enlarged unilateral or bilateral cervical lymph nodes.

Our approach to any salivary mass, even those suspected of being a primary malignant lymphoma, is excision. In the case of the parotid gland, this usually entails a superficial lobectomy. The rationale of this approach is as follows: (1) A variety of epithelial salivary tumors either contain or are often associated with an exuberant lymphoid element, such as the Warthin tumor, sebaceous lymphadenoma, lymphoepithelial carcinoma, and acinic cell carcinoma.³⁴ On small biopsy specimens, these tumors may be confused with malignant lymphomas or result in an uncertain diagnosis. (2) Malignant lymphomas developing in a background of "benign lymphoepithelial lesion" may initially be focal and subtle and easily missed on a small biopsy specimen. (3) Excisional biopsy also provides enough tissue for thorough routine histological evaluation and for additional specialized studies, such as T- and B-cell markers, flow cytometry, and gene rearrangement. (4) Parotid masses, even in a patient with known malignant lymphoma, are not always lymphomatous. Some are epithelial in origin and may be even more aggressive than the lymphoma. A superficial parotid lobectomy in the hands of an experienced head and neck surgeon is a low-risk procedure. Concerns in the past about facial nerve damage, salivary fistula, and cosmetic deformities are no longer valid. Fine-needle aspiration and incisional biopsies, however, may have merit in the medically compromised patient who is a poor operative risk.

Once the diagnosis is established, the patient should be referred to an oncologist for staging and appropriate treatment. As noted above, the inconsistencies in terminology make comparison between studies difficult. With this in mind, the prognoses from several reviews are shown in Table 7. ^2,8,12,35

The impact of histological grade, stage of disease, and presence or absence of an underlying autoimmune disease on clinical outcome is controversial.^2,8,36,37 In our study, the only variable to achieve prognostic significance was the histological grade according to the Working Formulation. Grades I and II lymphomas, when grouped together, had a significant survival advantage over grade III lesions (P<.01). Although stage of disease was important, it did not quite achieve significance when Ann Arbor stage I was compared, collectively, with stages II, III, and IV. While controversy exists as to the exact risk of development of malignant lymphoma in Sjögren syndrome, most authorities agree that there is an increased risk.^38,39 However, its presence had no impact on outcome in our study (P<.59).

In summary, primary malignant lymphomas of the parotid gland are uncommon. They affect both sexes equally, are rare before 50 years of age, and manifest clinically as a unilateral, painless mass. Most are of the B-cell, non-Hodgkin type and, at the time of diagnosis, about 80% of patients have Ann Arbor stage I or II disease.

Accepted for publication November 10, 1997.

Corresponding author: Leon Barnes, MD, Department of Pathology, University of Pittsburgh Medical Center, PUH, 200 Lothrop St, Pittsburgh, PA 15213 (e-mail: barnes@a1.isd.upmc.edu).