Vol. 85 No. 1 January 1998
ORAL AND MAXILLOFACIAL PATHOLOGY
Editor: Carl M. Allen
Warthin tumor A new hypothesis concerning its development J. M. Aguirre, MD, a M. A. Echebarrfa, MD, a R. Martfnez-Conde, MD, a C. Rodriguez, MD, b J. J. Burgos, MD, c and J. M. Rivera, PhD, c Vizcaya, Spain HOSPITAL DE CRUCES AND UNIVERSITY OF THE BASQUE COUNTRY
Objective. Warthin tumor is the second most common benign parenchyma[ salivary neoplasm. The purpose of this study was to evaluate the histopathologic features of Warthin tumor to determine whether a developmental pattern could be identified. Study design. Seventy-nine Warthin tumors in 63 patients (62 male and 1 female; average age, 58.62 years) were examined. All cases were histopathologically classified and morphometrica[ly analyzed with an Optomax system to measure the proportions of lymphoid tissue and cystic cavities. Results. The predominantly epithelial type (21% of all cases) of Warthin tumor was associated with a mean patient age at diagnosis of 53 years, a mean evolution time of 2.8 months, an average size of 1 cm, and a mean lymphoid component of 10%. In the case of the classicaltype (61% of all cases), the mean patient age at diagnosis was 58 years, the mean evolution time 9 months, the average size 2.4 cm, and the mean lymphoid component 40%. For the predominantly lymphoid type (18% of all cases), these values were, respectively, 62 years, 11 months, 2.8 cm, and 74%. The mean size of the classical and lymphoid tumors (79% of the Warthin tumors) was significantly greater (p < 0.0001 ) than the mean size of the epithelial type (21% of the Warthin tumors). A positive correlation between tumor size and evolution time was found. Conclusion. Our results suggest that the Warthin tumor initially develops as an adenomatous epithelial proliferation followed by lymphocytic infiltration. (Oral Surg Oral Med Oral Pathol Oral Radiol Endod 1998;85:60-3)
Warthin tumor (WT) is a benign neoplasm that represents 10% of all epithelial tumors of the major salivary glands. 1 W T is mainly observed in male patients and is almost exclusively limited to the parotid gland, where it may manifest itself bilaterally and as multiple lesions. The tumor is positively labeled by Tc-99 gammagraphy. Extraparotid locations 2 and malignant transformation have rarely been reported. 3,4 Histologically, W T consists of oncocytic epithelium that lines cystic structures and forms papillary projections in conjunction with a benign lymphoid component. Based on the relative amounts of these components, a number of histologic subtypes have been described, with possible implications related to the pathogenic development or evolution of the lesion. 5 aDepartment of Stomatology. bDepartment of Preventive Medicine and Public Health. CDepartment of Pathology. Received for publication Nov. 4, 1995; returned for revision Jan. 11, 1996; accepted for publication Sept. 17, 1997. Copyright © 1998 by Mosby, Inc. 1079-2104/98/$5.00 + 0 7114186384
The histogenesis of WT remains controversial. 1,6-9 At present, the epithelial component is thought to represent a neoplastic proliferation of salivary ducts trapped in the course of development within lymph nodes associated with the major salivary glands. 6,7,1° The lymphoid component is thought to be normal activated lymphoid tissue with a polyclonal B-cell proliferation. 1,11,12 In this clinicopathologic and morphometric study we examined the structural features of the different variants of W T and hypothesized a possible evolutionary pattern, with the aim of clarifying the histogenesis and development of the neoplasm.
MATERIALAND METHODS Seventy-nine WTs representing 63 patients (62 male and 1 female) were studied. The patients were treated in the following hospitals in Spain: Cruces (Vizcaya), La Paz (Madrid), Rfo Hortega (Valladolid), and Galdakao (Vizcaya). Patient age was 58.6 + 10.3 years (mean +_ SD). The surgical specimens were analyzed macroscopi-
ORAL SURGERY ORAL MEDICINE ORAL PATHOLOGY Volume 85, Number 1
Table I. Overall proportion of lymphoid stroma, cystic space, and epithelial component in Warthin's tumors (mean _+SD) Type E
L y m p h o i d stroma Cystic space Epithelium
Aguirre et al. 61
Table II. Mean patient age, tumor size, and evolution time of different types of Warthin's tumor Type E
n = 17
n = 48
n = 14
M e a n t u m o r size (cm)
53 -+ 10.8 A 1.05 + 0.7
57.7 + 10.3 A 2.42 _+ 0.7
61.57 + 7.7 A 2.80 +_ 1.3
M e a n evolution time (rot)
A 2.78 + 6.4 A
B 8.94 __. 10.3 A
B 10.78 +_ 8.2 A
n = 17
n = 48
n = 14
9.5 + 8.1 A 31.7 _ 12.7 A 58.7 _+ 15.6 A
39.6 + 13.4 B 43.6 + 15.2 B 29.0 _+ 19.0 B
74.2 _+ 8.8 C 31.4 + 10.5 A 8.2 + 9.2 C
Between lesion types, differences with the same letter are not significantly different (p > 0.05). Type E, Predominantly epithelial; Type C, classical; Type L, predominantly lymphoid.
cally so that the more representative fragments could be selected. The study involved only those specimens that were intact and without necrosis, secondary inflammation, or intense degenerative signs capable of significantly altering the tumor's morphologic features. Thirteen of 92 tumors that were initially evaluated (14.1%) were excluded on these grounds. The specimens were fixed in neutral 10% formalin, embedded in paraffin, prepared as 4-~tm sections, and stained with hematoxylin-eosin and periodic acid-Schiff. In all cases sectioning perpendicular to the lesional capsule was verified. The tumors were initially classified by two different pathologists according to histologic type; the classification was based on the criteria of Seifert et al., 5 according to which a tumor is (a) predominantly epithelial, (b) classical, (c) predominantly lymphoid, or (d) metaplastic. Sample kappa was 0.803. A morphometric study of the most representative stained sections of each tumor was carried out using an Optomax U (Synoptic) system linked to an Olivetti M24 microcomputer with Optomax software. A Nikon Optiphot light microscope equipped with a Kestrel 25 video camera, a Leitz-Grau 0.2% NG3 filter, and a Nikon Plan A p t 4x eyepiece were also used. After equipment calibration, direct and inverted contrast measurements were made of the lymphoid component and cystic cavities present in each case, with a minimum of 10 fields per tumor, except that in the smaller neoplasms the entire lesion was measured. The resulting areas and parameters were in turn used to calculate the percentages of lymphoid tissue and cystic spaces in each tumor. The data were statistically analyzed for mean age, time from first symptom until surgical intervention, size of tumor, and composition according to the percentage of the different tissues. After the equality of variances was contrasted (with Snedecor's F test), Student's t-dis-
M e a n age (yr)
For explanations, see Table I.
tribution test was applied for two-means comparisons. Comparison of more than two means was achieved with Tukey's pairwise comparisons method, based on the Studentized range for unequal cell sizes.13 The relationshi p between the size of the tumor and the time of evolution was studied with the corresponding scatter plot, correlation coefficient, and linear and polynomial (until third order) regression models.
RESULTS All of the WTs were located in the parotid gland; 82.5% were found in the lower portion of the superficial lobe of the gland. The lesions were bilateral in three cases (4.8%), with multiple lesions present in nine cases (14.3%). Based on the classification of Seifert et al., 5 there were 17 predominantly epithelial (E) tumors (21%), 48 classical (C) histologic presentations (61%), and 14 predominantly lymphoid (L) tumors (18%). No metaplastic tumors were observed. The morphometric analysis showed a lymphoid stroma of 39.29 + 23.36% (mean + SD) for the entire set of tumors (range: 1.46-90.63), with a cystic space of 38.89 _+15.04% (range: 7.94-97.06). The observed proportion of lymphoid tissue, cystic spaces, and epithelial component is represented in Table I for each histologic type. The classification into different histologic types (E, C, and L) is based on the quantities of the different tissues; the differences in the proportions of these tissues and the histologic type were statistically significant at the p = 0.05 level. Mean evolution time from clinical presentation to surgery was 7.47 months, and the average macroscopic lesion size was 2.2 cm. Table II shows mean patient age, tumor size, and evolution time for each of the three tumor types. With respect to mean size, the Type E tumors were statistically different from the Type C and Type L tumors: the Type E tumors were smaller than the others, but this size difference was not statistically significant. The relationship between size and evolution time is
ORAL SURGERY ORAL MEDICINE ORAL PATHOLOGY
Aguirre et al.
Table Ill. Estimation and contrast of the lineal regression model parameters to explain size of tumor as a function of time
Standard error of estimate
T for Ho: parameter =0
÷~----Y2_............. ~;...... --'"+ I:]
Intercept Time (slope)
> IT[ 0.0001 0.0143
p = Pr
. . . . . . . . . .
, . .
. . . . . . . . .
Type," ~[ ~[ ]~
24 + + +
. . . . . . . . . . .
36 T~me (monU1s).
Correlation coe~clen! - 0.27
Fig. 1. Size and evolution time relationship distinguishing type of tumor~
presented in Fig. 1. A positive correlation is seen in the scatter plot. Size and evolution time of Type E tumors were less than those of Type C and Type L tumors. The continuous line represents a statistical model for mean tumor size as a function of evolution time. The discontinuous lines point out the limits at the 95% confidence interval for the mean size predicted by the regression model. Table III shows the intercept and slope coefficients of this model. A statistically significant association between size and evolution time was found. However, when the analysis simultaneously controlled for tumor type, the previous relation seen between tumor size and evolution time was no longer significant beyond chance (p = 0.24). DISCUSSION
WT has been the subject of controversy since its earliest description, One of the most consistent histologic characteristics of this tumor is its abundant lymphoid stroma, the origin of which has been given different interpretations. Some authors consider the-tumor lymphocytes to represent previously normal residual lymph node tissue within which the epithelial component has come to proliferate; this is the so-called heterotopic theory, Others consider the infiltrate to constitute an inflammatory lymphoid cell response to an epithelial neoplasm; this is the immune theory. (The idea that the lesion represents a lymphoepithelial hamartoma has generally been discarded. 1,6-s) Some authors 9,14,15 have suggested that Epstein-Barr virus (EBV) may play an important role in the pathogenesis of WT, especially in multiple and bilateral cases. This virus might infect ductal epithelial cells, and the release of EBV gene products or cytokine by infected cells might activate lymphoid tissue and result in a polyclonal B-cell response.
According to the heterotopic theory, the origin of most WTs is traceable to the neoplastic proliferation of nests of salivary gland tissue trapped in the course of embryonic development within intraparotid and paraparotid lymph nodes. In accordance with this theory, Seifert et al. 5 classified WTs into different subtypes mainly on the basis of their lymphoid content. These authors proposed a staged pathogenic development in which the subtype-III (lymphoid predominance) WT constitutes an initial stage from which the subtype-I (typical) and subtype-II (epithelial predominance) WTs subsequently evolve. Thus a subtype-II WT would develop as a result of adenomatous epithelial proliferation from a subtype-I lesion. However, the amount of lymphoid tissue present in a well-developed classical WT often exceeds that of the small lymph node from which it supposedly originates. In addition, the distribution and arrangement of the germinal centers of the lymphoid stroma suggest that its development and proliferation may in fact be secondary to the epithelial growth. 16 Because of the clinicopathologic and morphometric results of the present study, we suggest a new model of the structural pattern and evolution of WT in several phases. According to this model, most cases of WT would initially arise in response to an as-yet-unidentified stimulus (for example, tobacco or EBV) within the parotid lymph nodes, with a first-phase intranodal arrangement involving a predominantly adenomatous epithelial component and a sparse lymphoid stroma. This initial "predominantly epithelial" phase would correspond to the subtype-II lesion of the Seifert WT classification.5 Such a tumor would have a shorter duration and a smaller size. In the second phase, tumor growth would occur because of proliferation and infiltration of the lymphoid Component, secondary to immune phenomena produced during development of the epithelial component; the latter would in turn grow to form the characteristic papillary projections and cystic cavities. Most of the cases reported to date would correspond to this type of W T - - t h e equivalent of Seifert subtype-I (typical) WTsS--and would exhibit larger proportions of lym-
ORAL SURGERY ORAL MEDICINE ORAL PATHOLOGY Volume 85, Number 1
Aguirre et al. 63
phoid tissue, a longer clinical course, and greater lesion size. Finally, in the third phase, growth would be characterized by a more prominent lymphoid stroma as the result of continued infiltration and proliferation of the lymphoid component. These lesions would correspond to Seifert subtype-III (predominantly lymphoid) WTs. 5 The clinical course would be prolonged, and tumor size would increase. This growth pattern would account for the discrepancies observed in immunologic studies of the lymphoid component. Thus in the early stages the stroma would consist mainly of T-lymphocytes, include few B-cells, and exhibit an arrangement similar to that seen in normal or reactive lymph nodes, in agreement with the findings of Diamond and Braylan. 1° In contrast, in the later stages the lymphoid component would consist mainly of B-lymphocytes in the form of a polyclonal Bcell response.ll,12,17,18 There are no behavioral characteristics associated with the different subtypes of WT defined by Seifert et al.5; the histologic features are a continuous spectrum that evolves over time. We feel that the metaplastic type does not belong in that spectrum, because most of those cases represent tumors previously subjected to radiotherapy--a treatment now totally discarded for this conclifton. As none of the patients in our series was treated in this way, it is not surprising that no metaplastic-type tumors were observed. We consider WT to be a benign salivary adenoma in which the epithelial component is the fundamental neoplastic element. A number of etiopathogenic aspects remain to be clarified, however; these include the reasons for the marked male predominance of the tumor, the growing female incidence in certain countries, the factors that trigger epithelial proliferation, and the possible conditioning role of tobacco, viruses, and other stimuli. These aspects should be the subject of future research, and we recommend the testing of our hypothesis with a larger number of patients.
2. Van der Waal JE, Davids JJ, Van der Waal I. Extraparotid Warthin's tumours: report of 10 cases. Br J Oral Maxillofac Surg 1993 ;31:43 -4. 3. Hall G, Tesluk H, Baron S. Lymphoma arising in an adenolymphoma. Hum Pathol 1985; 16:424-7. 4. Onder T, Tiwari RM, Van der Waal I, Snow GB. Malignant adenolymphoma of the parotid gland: report of carcinomatous transformation. J Laryngol Otol 1990;104:656-61. 5. Seifert G, Bull HG, Donath K. Histologic subclassification of the eystadenolymphoma of the parotid gland: analysis of 275 cases. Virchows Arch (Pathol Anat) 1980;388:13-38. 6. Thompson AS, Bryant HC Jr. Histogenesis of papillary cystadenoma lymphomatosum (Warthin's tumor) of the parotid salivary gland. Am J Pathol 1950;26:807-49. 7. Allegra SR. Warthin's tumor: a hypersensitivity disease?: Ultrastmctural, light and immunofluorescent study. Hum Pathol 1971;2:403-20. 8. Eveson JW, Cawson RA. Warthin's tumor (cystadenolymphoma) of salivary glands: a elinicopathologic investigation of 278 cases. Oral Surg Oral Med Oral Pathol 1986;61:256-62. 9. Gallo O. New insights into the pathogenesis of Warthin's tumour. Oral Oncol Eur J Cancer 1995;31B:211-5. 10. Diamond LW, Braylan RC. Cell surface makers on lymphoid cells from Warthin's tumors. Cancer 1979;44:580-3. 11. Caselitz J, Salfelder A, Seifert G. Adenolymphoma: an immunohistochemical study with monoclonal antibodies against lymphocyte antigens. J Oral Pathol 1984;13:438-47. 12. Rueo LR Rosai S, Remotti D. Immunohistology of adenolymphoma (Warthin's tumor): evidence for a role of vascularitation in the organization of the Iymphoepithelial structure. Histopathology 1987;11:557-65. 13. Kramer CY. Extension of multiple range tests to group means with unequal numbers of replications. Biometrics 1956;12:30710. 14. Scully C. Viruses and salivary gland disease: are there associations? Oral Surg Oral Med Oral Pathol 1988;66:179-83. 15. Santucci M, Gallo O, Calzolari A, Bondi R. Detection of Epstein-Barr viral genome in tumor cells of Warthin's tumor of parotid gland. Am J Clin Pathol 1993;100:662-5. 16. Thackray AE, Lucas RB. Tumors of the major salivary glands. In: Firminger HI. Atlas of tumor pathology. Washington: Armed Forces Institute of Pathology, 1974. 17. Cossman J, Deegan MJ, Batsakis JG. Warthin's tumor: B lymphocytes within the lymphoid infiltrate. Arch Pathol Lab Med 1977;101:354-6. 18. Tubbs RR, Sheibani K, Weiss RA, Lee V, Sebek BA, Valenzuela R. Immunohistoehemistry of Warthin's tumor. Am J Clin Pathol 1980;74:795-7.
The authors thank Dr. A. del Villar, Dr. I. Zabalza, Prof. E Contreras, and Dr. J. M. Viguer for their valuable collaboration in this study.
REFERENCES 1. Chapnik JS. The controversy of Warthin's tumor. Laryngoscope 1983;93:695-716.
Reprint requests: J. M. Aguirre, MD Medicina Bucal, Departamento de Estomatologfa Facultad de Medicina y Odontologfa Universidad-del Pals Vasco Aptdo. 699, 48080-Bilbao Spain