Discussion

There is an increasing incidence of HPV-positive oropharyngeal SCC, which represents a unique demographic, molecular and clinical entity with typically younger age, no or limited history of tobacco or alcohol use, and a more favourable prognosis.^{Reference Gillison, Castellsagué, Chaturvedi, Goodman, Snijders and Tommasino17–Reference Genden, Sambur, de Almeida, Posner, Rinaldo and Rodrigo19} These patients can present with cystic neck masses without a clinically apparent primary tumour, which could be confused with benign lateral cervical cysts.

Many authors emphasise a strong association between cystic cervical node metastases and HPV-related oropharyngeal SCC.^{Reference Goldenberg, Begum, Westra, Khan, Sciubba and Pai20–Reference Yasui, Morii, Yamamoto, Yoshii, Takenaka and Nakahara24} Yasui et al.^{Reference Yasui, Morii, Yamamoto, Yoshii, Takenaka and Nakahara24} have shown that HPV infection is consistent between the primary tumour and its corresponding node metastasis and that HPV-positive node metastasis is specific to oropharyngeal SCC, especially for a palatine tonsil or a tongue base primary tumour.

The incidence of cystic lymph node metastases arising from SCC of Waldeyer's ring has been reported to be from 35 to 64 per cent of those with nodal metastasis.^{Reference Regauer, Mannweiler, Anderhuber, Gotschuli, Berghold and Schachenreiter15,Reference Zengel, Assmann, Mollenhauer, Jung, Sotlar and Kirchner21,Reference Thompson and Heffner25} Cyst formation within metastases from Waldeyer's ring was found histologically in 50 per cent of all metastatically affected lymph nodes in 61 per cent of patients. However, clinically detectable cystic metastases (with a luminal diameter more than 5 mm) were identified in 26 per cent of all metastatically affected lymph nodes in one third of patients (34.6 per cent).^{Reference Regauer, Mannweiler, Anderhuber, Gotschuli, Berghold and Schachenreiter15}

The mechanism of cystic formation in the majority of cystic metastases is pseudocystic change and cystic degeneration. This results from spontaneous degradation of keratin and cellular debris within the carcinomatous lymph node deposit. However, some cystic metastases are true cysts. The origin of a true cyst is presumed to arise from submucosal minor salivary gland acini and ducts or from transitional squamous epithelium in the tonsillar crypts. Moreover, Regauer et al.^{Reference Regauer, Beham and Mannweiler26} found that cytokeratin CK7-positive carcinomas produce CK7-positive cystic nodal metastases, whereas no solid CK7-positive nodal metastases have been identified. Therefore, a subset of carcinomas occurring in the Waldeyer's ring area appear to arise from the large excretory ducts of submucosal minor salivary glands with only limited surface involvement, express CK7 and produce CK7-positive cystic metastases.^{Reference Regauer, Mannweiler, Anderhuber, Gotschuli, Berghold and Schachenreiter15,Reference Thompson and Heffner25,Reference Regauer, Beham and Mannweiler26}

The primary SCC arising in the Waldeyer's ring often demonstrates cystic spaces, especially deep inside crypts, and the cystic metastases simulate the growth behaviour and growth pattern of the parent cells. Therefore, cyst formation has been proposed as an intrinsic property of the tumour.^{Reference Regauer, Mannweiler, Anderhuber, Gotschuli, Berghold and Schachenreiter15}

In a series study of 121 patients with isolated lateral cervical cysts, Gourin and Johnson^{Reference Gourin and Johnson2} found that the incidence of unsuspected metastases was 9.9 per cent (12 of 121 patients). However, a significantly greater incidence of malignancy (23.5 per cent) was found in patients over 40 years of age. In a similar sample size in our study, we determined a 9.9 per cent total incidence of metastatic SCC in adults. Moreover, the incidence in patients older than 40 years of age was 18.5 per cent (10 of 54 patients), which approaches statistical significance (p = 0.003). The mean age of patients with cystic nodal metastases (52.7 years) was significantly higher (p = 0.003) than the mean age of patients with benign lateral cervical cysts (38.8 years). We confirmed that the incidence of cystic lymph node metastases significantly rises with increasing age (p = 0.013), so age would seem to be an important factor in the incidence of cystic metastases. In recent studies the average age of patients with cystic nodal metastasis ranges from 53 to 58 years, which corresponds with the younger population of HPV-positive oropharyngeal SCC patients.^{Reference Gourin and Johnson2,Reference Goyal, Zacharia and Goldenberg5,Reference Goldenberg, Begum, Westra, Khan, Sciubba and Pai20,Reference Thompson and Heffner25}

The incidence of cystic metastases in our series was higher in males than females (14 per cent and 5.6 per cent, respectively), but was not statistically significant (p = 0.204). Of note is the incidence of cystic nodal metastases in males over 40 years, which was 24.1 per cent, in contrast to females over 40 years, which was 12 per cent. This difference did not approach statistical significance (p = 0.167), probably because of the small sample size.

The average size of the largest cystic lymph node metastasis in the literature varies from 1.8 to 3.9 cm.^{Reference Goyal, Zacharia and Goldenberg5,Reference Zengel, Assmann, Mollenhauer, Jung, Sotlar and Kirchner21,Reference Thompson and Heffner25} In our study, the mean maximal axial diameter of cystic metastases was 4.0 cm and the mean minimal axial diameter of cystic metastases was 2.9 cm. We did not discover statistically significant differences in the measurements of either diameter (p = 0.520, p = 0.918). We presume that mass size does not correlate with the biological nature of the lesion. On the other hand, in a study where the branchial cleft cysts and malignant cystic adenopathy were radiographically compared, benign cystic masses were found to be larger on the long axis, short axis and in height. The average measurement for the short axis and long axis for a branchial cleft cyst was 2.2 cm and 2.8 cm, respectively. The average measurement for the short axis and long axis for a malignant cyst was 1.4 cm and 1.8 cm, respectively. In contrast to our study, where we only analysed patients with visible neck masses, branchial cleft cyst and head and neck cancer patients who underwent contrast-enhanced CT of the neck were evaluated in the previously mentioned study. Many of these head and neck cancer patients probably did not present with visible neck masses, so CT-identified cystic lesions were presumably smaller, which could explain the size differences.^{Reference Goyal, Zacharia and Goldenberg5}

In a majority of cases, both lateral cervical cysts and cystic metastases were located in level II and then level III. We found cervical cystic masses (benign and malignant) in levels II and III in 80 per cent of the cases. The lateral cervical cysts in our series were located in level II and III in 88 per cent of cases. Other authors found 95.2 per cent of the lateral cervical cysts in levels II and III.^{Reference Goyal, Zacharia and Goldenberg5} In our study, all cystic metastases were identified in levels II and III (100 per cent). Other authors report that malignant cystic lymph nodes were localised mostly in level IIA (52.6 per cent) and in levels II and III in 63 per cent.^{Reference Goyal, Zacharia and Goldenberg5}

A cystic mobile mass with a history of previous inflammation would suggest a benign lateral cervical cyst rather than a metastasis. The collapse of the cyst after aspiration was also classically described as a lateral cervical cyst characteristic. However, Mallet et al. emphasise that none of these signs are sufficient to eliminate cervical metastasis.^{Reference Mallet, Lallemant, Robin and Lefebvre27}

In our study, we did not find statistically significant differences between benign cysts and cystic metastases in the occurrence of inflammatory signs such as pain or local inflammatory changes, growth or pressure, or antibiotic use (p = 0.454, p = 0.341, p = 1.000). Similarly, CT imaging did not show a significant difference between benign cystic masses and cystic node metastases when used to look for the presence of streaking as an inflammation feature in adjacent fat surrounding a cystic lesion.^{Reference Goyal, Zacharia and Goldenberg5}

Cystic metastasis should be distinguished from necrosis in an otherwise solid node. The imaging criteria of cystic nodal metastases were defined by Goldenberg et al.^{Reference Goldenberg, Begum, Westra, Khan, Sciubba and Pai20} Cystic nodes were described as round or ovoid masses with a thin (less than 2 mm) enhancing capsule, homogeneous fluid content, and no internal complex, irregular, or solid area. Nodes with thicker solid walls and irregular, complex central low attenuation are classified as necrotic. Of 20 patients who met these criteria for cystic metastases, 10 had a tongue base primary tumour, 7 had a palatine tonsil primary tumour and 3 had a histologically unknown primary tumour. In situ hybridisation confirmed the presence of HPV DNA in 87 per cent of patients.

Yasui et al.^{Reference Yasui, Morii, Yamamoto, Yoshii, Takenaka and Nakahara24} found that radiographically identifiable cystic node metastases are specific to oropharyngeal SCC and cancer of unknown primary and are more likely to be HPV-positive when compared with solid or necrotic node metastasis. None of 146 non-oropharyngeal SCCs showed radiographically detected cystic node metastases. On contrast-enhanced CT imaging, 6 per cent of node-positive oropharyngeal SCC and 11 per cent of node-positive cancer of unknown primary showed cystic node metastasis. Yasui et al.^{Reference Yasui, Morii, Yamamoto, Yoshii, Takenaka and Nakahara24} stated that this prevalence of radiographically identifiable cystic node metastases in oropharyngeal SCC and cancer of unknown primary is relatively low, when compared with histopathologically identified cystic node metastases. Therefore, the capacity of contrast-enhanced CT imaging to visualise a cystic lesion is limited, particularly when the lesion is small.^{Reference Yasui, Morii, Yamamoto, Yoshii, Takenaka and Nakahara24}

Ferris et al.^{Reference Ferris, Branstetter and Nayak4} sought to determine whether radiographic imaging using combination PET-CT can distinguish benign from malignant cystic masses. They reviewed the PET-CT imaging of cystic neck masses in 5 patients over 40 years of age and concluded that PET-CT is not a reliable modality for identifying malignancy in adults with suspicious cystic neck masses.^{Reference Ferris, Branstetter and Nayak4} On the other hand, in a series study of 58 patients with single cystic neck lesions, Abadi et al.^{Reference Abadi, Johansen, Godballe, Gerke, Høilund-Carlsen and Thomassen28} showed that fluoro-deoxy-glucose PET-CT could reliably rule out malignancy, albeit with a high frequency of false-positive scans, requiring further diagnostic analysis.

Using cytology to distinguish benign cervical cysts from metastatic adenopathy continues to be challenging. A multivariate analysis of 19 cytomorphological features for cytological separation of a branchial cleft cyst from metastatic cystic SCC shows that the distinction is cytomorphologically difficult, and diagnostic accuracy still remains imperfect.^{Reference Layfield, Esebua and Schmidt6} Image cytometry DNA analysis on cytological specimens can increase diagnostic sensitivity, but only 53 per cent of cases with cystic metastasis showed aneuploidity, indicating malignancy, whereas DNA analysis showed diploidity in all benign cysts.^{Reference Nordemar, Tani, Högmo, Jangard, Auer and Munck-Wikland9}

Polymerase chain reaction analysis of the fine needle aspiration fluid enables non-surgical detection of HPV DNA, which may facilitate the differential diagnosis of cystic node metastasis and lateral cervical cyst.^{Reference Yasui, Morii, Yamamoto, Yoshii, Takenaka and Nakahara24} False-negative FNAC still seems to be high, so surgical excision with thorough histological examination can prove the diagnosis. Despite ultrasonography guidance to assess solid components or the cyst wall (regularly performed in our study), our cytological results from cystic neck masses were repeatedly non-diagnostic. Consequently, only a small number of our patients underwent FNAC, and open excisional biopsy to confirm the diagnosis was recommended to these patients. However, we identified two patients (2 of 20) with false-negative cytological results.

Panendoscopy under general anaesthesia, with direct biopsies, primarily of the base of the tongue, and tonsillectomy are an essential part of the diagnosis of cystic malignancy. In our study, we found an occult primary tumour in 9 of 11 patients (6 in the palatine tonsils and 3 in the base of the tongue) in this way. All the primary tumours were small and clinically occult. Two patients remained cancer of unknown primary.

In a study by Zengel et al.,^{Reference Zengel, Assmann, Mollenhauer, Jung, Sotlar and Kirchner21} the average size of the primary tumour was 1.1 cm in the palatine tonsils and 0.8 cm in the base of the tongue. In 65 per cent of the cases, the size of the primary tumour was 1.0 cm or less. Because of the size and deep submucosal location with contact with basal parts of the tonsillar crypts, pre-operative clinical or radiological detection of these carcinomas is very difficult or impossible. Their study strongly supports a causal relationship between HPV-associated oropharyngeal SCC and cancer of unknown primary.^{Reference Zengel, Assmann, Mollenhauer, Jung, Sotlar and Kirchner21} We agree with this relationship, because all the patients in our study had a primary cancer in the oropharynx or had cancer of unknown primary. In addition, all patients with cystic nodal metastases were found to be p16-positive by immunohistochemistry analysis in samples from the primary tumour or from the metastasis.

Thompson and Heffner^{Reference Thompson and Heffner25} show that primary tumours arising from tonsillar tissue present with slower growth than is usually expected for SCC, with some tumours remaining undiscovered for many years. They explain this indolent or suppressed growth of the primary tumour by the nature of the tonsillar crypt lymphoepithelium from which these tumours arise. Primary tumours were found within a wide period of time ranging from the day of initial surgery up to 11 years (average of 12.4 months). Tumour infiltrating lymphocytes and inflammatory response are associated with host response resulting in local tumour growth control and even regression of the tumour.^{Reference Regauer, Mannweiler, Anderhuber, Gotschuli, Berghold and Schachenreiter15,Reference Thompson and Heffner25} Meticulous histological analysis of the surgical specimen is necessary in order not to miss small carcinoma. Immunohistological staining can help in detection, especially of small carcinomas with lymphoepithelial differentiation, which resemble physiological cryptal epithelium.^{Reference Zengel, Assmann, Mollenhauer, Jung, Sotlar and Kirchner21}

The management of lateral cystic neck masses in patients aged over 40 years is not yet well defined. Patients over 40 years with isolated lateral cervical cystic masses should be presumed to have cystic SCC metastasis until proven otherwise, and the general recommendation is to perform surgical interventions in one stage. If the FNAC reveals SCC, it is possible to plan elective treatment, panendoscopy, direct biopsies of Waldeyer's ring, tonsillectomy and neck dissection in one stage. If FNAC is non-diagnostic, excisional biopsy of the cystic mass with frozen section biopsy is indicated. If frozen section biopsy is positive for SCC, then panendoscopy, direct biopsies of Waldeyer's ring, tonsillectomy and neck dissection should be performed in one stage.^{Reference Gourin and Johnson2,Reference Goldenberg, Sciubba and Koch3,Reference Cinberg, Silver, Molnar and Vogl16,Reference Andrews, Giddings and Su29,Reference Devaney, Rinaldo, Ferlito, Silver, Fagan and Bradley30}