Clinical examination revealed an extensive tumour in his right tonsillar region, which largely occluded his pharynx and caused airway narrowing. An enlarged level 2 lymph node, estimated diameter 3 cm, was palpated. On fibreoptic nasendoscopy, the larynx and supraglottis were free of tumour.

Histologically, the specimen included tissue fragments of basaloid appearance in which the architecture varied from solid to microcystic with a few ducts and possible micropapillary structures (figure 2A). Peripheral palisading was seen. The nuclei were generally ovoid and a few were angulated (figure 2B). There was no nuclear pleomorphism or mitotic activity. Because of the fragmented nature of the tumour, it was not possible to fully assess invasiveness although one fragment did demonstrate invasive features.

At the follow-up 1 month later the mass had disappeared completely and his tracheostomy was removed. No oral mass could be found at the next consultation 2 months later.

Further follow-up in February 2004 showed a small prominence at the previous tumour site. Biopsy was considered; however, the mild swelling at the right tongue base did not change in appearance for 2 years until February 2006 when the swelling had increased in size again. Considering a potential infection, a course of antibiotics was prescribed, without effect, and a second biopsy was obtained under local anaesthetic a month later. Deep samples from the right of the tongue base showed only normal tissue. In May 2006, the swelling had significantly reduced and no further changes were detected. By March 2007, his chronic obstructive pulmonary disease had deteriorated to the extent he could no longer attend clinic and regular follow-up was discontinued.

In this case, immunohistochemical stainings were unhelpful and did not differentiate between the differential diagnoses named above. The possible diagnosis of basal cell adenocarcinoma (BCAC) was not considered, although it is well documented that neither immunohistochemical methods nor ultrastructural characteristics do distinguish between BCAC and BCA.⁴

However, even though no tumour invasion could be demonstrated, the clinical presentation was that of a malignant tumour. The general consensus at the multidisciplinary team meeting was strongly in favour of malignant disease. The tumour had grown rapidly, was painful and completely occluding the pharynx requiring an emergency tracheostomy. The enlarged cervical lymph nodes and the nodular opacity in the left lung were suggestive of metastatic disease. The size of the tumour, measuring more than 4 cm in diameter, indicated a high clinical stage of tumour. Although tumour grading was not available, a poor outcome was suggested considering the relative importance of stage versus grade in minor salivary gland tumours.⁵ Treatment options were discussed and surgery would have been the preferred approach if the patient's health and fitness had allowed. Alternatively, radiotherapy was considered based on evidence from previous studies⁶ of patients receiving radiotherapy alone for minor salivary gland carcinomas. These demonstrated long-term local tumour control especially in patients unfit for surgery. In this case, as the clinical findings had much improved without starting any treatment, the decision was made to observe and follow the patient up at regular intervals.

When the first biopsies were taken, a surgical trauma occurred that might have led to a change in blood supply resulting in tumour necrosis. This represents one of many mechanisms thought to be responsible for spontaneous remission of cancer.⁷ In 1935 spontaneous regression of a carcinoma of the tongue after biopsy was reported by David Roxburgh.⁸ However, without a confirmed histological diagnosis of malignancy, our case did not fulfil the strict criteria of spontaneous regression of a cancer defined by Everson and Cole in 1956. They postulated spontaneous regression of cancer to be a partial or complete disappearance of a malignant tumour in the absence of all treatment, or in the presence of treatment that is considered inadequate to exert a significant influence on neoplastic disease.⁹ In our case, a second attempt to biopsy the lesion was unsuccessful and no further tissues samples were taken at the postmortem examination; hence, the exact nature of the tumour, as well as the mechanism leading to its regression, remain uncertain. Histological distinction between benign and malignant basaloid neoplastic proliferations of the salivary glands is difficult and may, ultimately, rely on the demonstration of an infiltrative growth pattern or perineural invasion.