Warthin tumour (WT) is thought to be a benign neoplasm of the salivary gland epithelium with a proliferative epithelial component associated with a variably prominent lymphoid stroma.¹ The lymphoid stroma is reactive in most cases, often with prominent secondary lymphoid follicles.² Although uncommon, the lymphoid stroma in WT may be involved by disseminated non-Hodgkin lymphoma (NHL).^2–11 The diagnosis of lymphoma may already be known in these patients.^2–7

“Warthin tumour is thought to be a benign neoplasm of the salivary gland epithelium”

The first time diagnosis of NHL at the time of histopathological assessment of a WT excision specimen is extremely rare, with 14 cases in the English language literature.^8–17 Staging procedures usually identify disseminated disease in these patients,^8,10,11,15 although there are reports of the disease being localised to the parotid region.^9,11,17 Follicular lymphoma is the most common lymphoma presenting as a collision tumour with WT.^8–13,15,16

We are aware of only one other case of small lymphocytic lymphoma (SLL) in WT in the English language literature,¹⁴ and report a second case of SLL detected in the diagnostic WT specimen. The diagnosis was confirmed by immunohistochemistry and molecular studies, and staging procedures immediately after diagnosis demonstrated disseminated disease.

CASE REPORT

This 60 year old man was seen for a swelling in the left parotid region of three to four years’ duration. On examination, the mass was firm, mobile, and non-tender. There were no systemic symptoms. He underwent excision of the mass. He also had chronic obstructive pulmonary disease, type II diabetes mellitus, obstructive sleep apnoea, and a past history of myocardial infarction.

Pathological findings

The superficial parotidectomy specimen was 6.5 × 4.0 × 3.0 cm, with a 2.3 cm diameter tan grey lesion close to one of the resection margins. There were satellite nodules measuring 0.6 to 1.0 cm in diameter adjacent to the main lesion.

In the histological sections of the largest lesion, classic findings of Warthin tumour were present; these included oncocytic epithelium—tall columnar cells with eosinophilic granular cytoplasm—present as papillary proliferations and lining the cystic areas, which were filled with pale eosinophilic fluid admixed with little cell debris (fig 1A).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

 Warthin tumour and small lymphocytic lymphoma. (A) Oncocytic epithelium and pale proliferation centres of small lymphocytic lymphoma (haematoxylin and eosin (H&E); original magnification, ×50). (B) Large paraimmunoblasts in a proliferation centre (H&E; original magnification, ×200).

The intervening stroma had a diffuse infiltrate of small lymphoid cells with no necrotic areas. There were interspersed small pale poorly defined nodular areas, the proliferation centres of which were composed primarily of large lymphoid cells with prominent nucleoli, the “paraimmunoblasts” (fig 1B). Both large and small lymphoid cells expressed CD20, CD5, and CD23; these cells were immunonegative for CD3 and CD10 (fig 2). The proliferative fraction of the lymphoid neoplasm was less than 5%, mostly in the proliferation centres. There was a small background population of T cells.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

 Small lymphocytic lymphoma tumour cells positive for (A) CD20, (C) CD5, and (D) CD23, but negative for (B) CD10 (immunoperoxidase with haematoxylin counterstain; original magnification, ×75).

The immunoglobulin heavy chain gene rearrangement was clonal, as identified by single step polymerase chain reaction using consensus primers against the VDJ region. There was no molecular evidence of the t(14, 18) or t(11, 14) translocations.

Staging and clinical course

The total leucocyte and the absolute lymphocyte counts were within the reference ranges at the time of diagnosis. Small superficial cervical and axillary lymphadenopathy was detected after the diagnosis. Radiographic staging procedures identified mediastinal and para-aortic lymphadenopathy, although there was no hepatosplenomegaly. The bone marrow was involved: 75% lymphocytes with a nodular and interstitial pattern of involvement. There were no B symptoms. The disease was considered to be in stage IV A.

The patient was treated with intermittent chlorambucil and the disease was stable for six years. He then developed generalised lymphadenopathy and was treated with six cycles of fludarabine and is currently in complete remission.

DISCUSSION

The lymphoid stroma of WT is part of the systemic lymphoid tissue and therefore is likely to be involved in disseminated lymphoma. Indeed, most patients with lymphomatous involvement of the WT lymphoid stroma have disseminated disease, either in the form of an already established diagnosis or disease that is discovered at staging after initial diagnosis in a WT specimen.^{2–8,10,11,15} In this last situation, NHL is usually brought to clinical attention by the superficial nature of the parotid gland mass.¹⁵ Inadequate staging information in some reports^8,12,14 does not permit the distinction between initial parotid presentation of systemic disease and lymphoma primarily arising in and restricted to WT.

“Disease localised to the parotid region for a variable period of time before dissemination has been offered as an argument supporting the parotid origin of lymphoma”

We were able to find 20 cases of coexisting WT and lymphoma in the English language literature.^2–17 Of these, 14 were first diagnosed in the WT excision specimen,^8–17 and most (10) were follicular lymphomas.^8–13,15,16 Single cases of diffuse large B cell lymphoma,¹⁰ diffuse mixed small and large cell lymphoma,¹⁷ and SLL¹⁴ have also been reported. As in our case, staging soon after diagnosis identified disseminated disease in six of the 14 cases,^{8,10–12,15,16} whereas systemic disease developed in four additional cases within three months to seven years.^9,11,13,17 Disease localised to the parotid region for a variable period of time before dissemination has been offered as an argument supporting the parotid origin of lymphoma.¹¹

The coexistence of two separate primary tumours without substantial histological admixture at the interface is referred to as a collision tumour.¹⁸ This may be coincidental or the pathogenesis may be related. In this second instance, one single agent may interact with different tissues, inducing different tumours,¹⁹ or one tumour may induce the formation of the other. It has also been proposed that lymphoma localised to WT may result from persistent antigenic stimulation provided by WT epithelial elements.²⁰ Most lymphomas in WT have been reported to be follicular and the WT antigenic stimulation theory is indeed plausible. It is also possible that the WT microenvironment expresses APRIL/BlyS, a B cell survival factor shown to be of importance in SLL/chronic lymphocytic lymphoma.²¹

There is also some controversy about whether the lymphoid stroma is acquired in response to WT, is part of an epithelial–lymphoid hamartoma, or whether the epithelial element arises from heterotopic salivary gland epithelial remnants in “parotid” lymph nodes.²² Because no marginal zone lymphomas have been reported in WT, it is possible that secondary mucosa associated lymphoid tissue is not acquired in WT, and that the lymphoid stroma is indeed part of the pre-existing lymph node. This concept is further supported by the occurrence of approximately 8% of WT in cervical lymph nodes.²²

At scanning magnification, the pale nodular areas of secondary germinal centres and proliferation centres in SLL may appear similar to an untrained eye. In our case, the SLL nature of the lymphomatous infiltrate was missed by some of the first year pathology residents.

In summary, this case is the second example of collision of SLL and WT where SLL was initially diagnosed in the WT excision specimen. It also highlights the importance of carefully evaluating the lymphoid stroma because the pale proliferation centres may mimic “pale” secondary follicles in a reactive stroma.