Implications for clinical staging of metastatic cutaneous squamous carcinoma of the head and neck based on a multicenter study of treatment outcomes




Cutaneous squamous cell carcinoma (SCC) of the head and neck is a common cancer that has the potential to metastasize to lymph nodes in the parotid gland and neck. Previous studies have highlighted limitations with the current TNM staging system for metastatic skin carcinoma. The aim of this study was to test a new staging system that may provide better discrimination between patient groups.


A retrospective multicenter study was conducted on 322 patients from three Australian and three North American institutions. All had metastatic cutaneous SCC involving the parotid gland and/or neck and all were treated for cure with a minimum followup time of 2 years. These patients were restaged using a newly proposed system that separated parotid disease (P stage) from neck disease (N stage) and included subgroups of P and N stage. Metastases involved the parotid in 260 patients (149 P1; 78 P2; 33 P3) and 43 of these had clinical neck disease also (22 N1; 21 N2). Neck metastases alone occurred in 62 patients (26 N1; 36 N2). Ninety percent of patients were treated surgically and 267 of 322 received radiotherapy.


Neck nodes were pathologically involved in 32% of patients with parotid metastases. Disease recurred in 105 (33%) of the 322 patients, involving the parotid in 42, neck in 33, and distant sites in 30. Parotid recurrence did not vary significantly with P stage. Disease-specific survival was 74% at 5 years. Survival was significantly worse for patients with advanced P stage: 69% survival at 5 years compared with 82% for those with early P stage (P = 0.02) and for those with both parotid and neck node involvement pathologically: 61% survival compared with 79% for those with parotid disease alone (P = 0.027). Both univariate and multivariate analysis confirmed these findings. Clinical neck involvement among patients with parotid metastases did not significantly worsen survival (P = 0.1).


This study, which included a mixed cohort of patients from six different institutions, provides further information about the clinical behavior of metastatic cutaneous SCC of the head and neck. The hypothesis that separation of parotid and neck disease in a new staging system is supported by the results. The benefit of having subgroups of P and N stage is uncertain, but it is likely to identify patients with unfavorable characteristics that may benefit from further research. Cancer 2006. © 2006 American Cancer Society.

Squamous cell carcinoma (SCC) is a common cutaneous malignancy occurring most frequently among older adult males of Anglo-Celtic descent. The metastatic potential of cutaneous SCC is uncertain but, overall, it is thought to be relatively low, about 5%.1 With a high proportion of these lesions occurring on the skin of the head and neck, involvement of lymph nodes within the parotid gland and in the neck is well recognized as being part of the natural behavior of this type of skin carcinoma.2

The current TNM staging system used internationally for cutaneous malignancy has a simple classification for regional metastatic disease: N0 for clinically absent metastatic disease and N1 signifying the presence of palpable metastatic disease. There are no subgroupings and this classification applies to all parts of the body, including the head and neck.

There has been recent interest in the clinical behavior of metastatic cutaneous SCC when it involves the lymph nodes of the parotid gland and cervical region2–5 and, in particular, the adequacy of the current TNM staging classification when applied to the head and neck has been questioned.7 An initial study from the Sydney Head and Neck Cancer Institute (SHNCI) at Royal Prince Alfred Hospital (RPAH) and the University of Sydney, Australia,7 hypothesized that, in relation to cutaneous SCC involving the skin of the head and neck, better prognostic discrimination may be achieved by separating parotid (P) and neck (N) disease. A revised staging system was described and tested. It was found that, among patients with metastatic cutaneous SCC involving the parotid gland, those who also had disease in the neck had a statistically significantly worse outcome. Subgroups of P stage and N stage were described and it was found there was a trend to decreasing disease control in the parotid region with increasing P stage, whereas survival decreased significantly with increasing clinical and pathologic N stage.

Although these initial findings supported the recommended separation of parotid and neck metastatic disease, we were concerned that the findings may be sample-specific. The suggested new staging system was subsequently tested on a separate patient cohort, also from Sydney, Australia.8 In that followup study the findings were somewhat different. Both local control in the parotid bed and survival varied with the extent of disease in the parotid gland (P stage), and, in marked contrast to the initial study, pathologic involvement of cervical nodes did not worsen the survival of patients with disease in the parotid gland. This may have been due to a lower incidence of nodal involvement in the followup study. The rate was over 50% in the initial study from the SHNCI and 36% in the followup study.

To clarify these issues further we decided to enlarge the study group by the inclusion of patients from other institutions. The purpose of this article is to report the findings of an expanded retrospective study that, again, aimed to clarify aspects of the clinical behavior of metastatic cutaneous SCC and to determine whether or not the proposed changes to the clinical staging system could be validated.


Retrospective clinical and pathologic information was accrued from three Australian and three North American intuitions. A minimum requirement was that high-quality clinical data could be provided for at least 20 patients, with histologically proven metastatic SCC involving the parotid gland and/or the lymph nodes of the neck with at least 2 years followup. To be eligible, patients needed to be previously untreated and to have received definitive therapy with the aim of cure. In some cases data had been accessioned prospectively to computerized databases, whereas in some institutions chart review was necessary. The study group included some of the patients from the two initial studies from Sydney.7, 8

Participating institutions were asked to provide patient demographic data along with clinical and pathologic information, details of radiotherapy, recurrence data, and survival. Patient data were de-identified and clinically restaged according to the proposed new staging system (Table 1). Data were entered using a numerical coding system onto a standardized spreadsheet and results transferred electronically to the SHNCI data management unit for analysis. All data were quality assured and patients lacking adequate staging, treatment, and followup data were deleted from the study.

Table 1. Clinical Staging System for Metastatic Cutaneous Squamous Cell Carcinoma of the Parotid and/or neck
 P0No clinical disease in the parotid
 P1Metastatic node up to 3 cm in diameter
 P2Metastatic node > 3 cm and up to 6 cm in diameter or multiple nodes
 P3Metastatic node > 6 cm in diameter or disease involving the facial nerve or skull base
 N0No clinical disease
 N1Single ipsilateral neck node up to 3 cm in diameter
 N2Single node > 3 cm in diameter or multiple nodes or contralateral nodes

Disease recurrence was called ‘local’ if tumor recurred in the parotid bed or in the skin over the parotid region. Recurrence in dermal lymphatics causing skin nodules in the parotid region is a pattern of recurrence recognized in this disease. Local recurrence does not refer to recurrence at the primary cutaneous site. Recurrence was designated as regional if it developed in cervical nodes.

Cumulative local control and survival results were calculated using the Kaplan–Meier method and curves were compared using the log rank test. Multivariate analysis of factors influencing survival was carried out using the Cox method of proportional hazards.


From the six participating institutions a total of 398 patients were enrolled, but it was necessary to exclude 76 patients because of missing data. Therefore, there were 322 patients in the final cohort for analysis. The three Australian institutions contributed 245 of these patients, whereas 77 came from the three North American institutions. The individual contributions were: SHNCI (RPAH/University of Sydney), 95; Westmead Hospital (University of Sydney), 97; Southern Zone Radiation Oncology (SZRO) and Royal Brisbane Hospital, 53; University of Florida, 23; University of Toronto, 31; Memorial Sloan Kettering Cancer Center (MSKCC), 23.

There were 239 men and 83 women with a median age of 68 years (range, 28–98). The earliest treatment date was 1960 and the latest patient was treated in 2003. Only seven patients, however, were treated before 1980 and the majority were treated after 1990.

Clinical Staging

Patients were allocated a P and N stage according to the proposed staging system (Table 1) and the results are tabulated in Table 2. The majority of patients (67%) had disease only in the parotid gland, and the majority of these (130 of 217) had early stage disease (P1). Neck disease alone was found in 62 (19%) of patients and more than half of these (36 of 62) had advanced metastatic disease in the neck (N2). A total of 43 of 322 patients (13%) had clinical disease in both the parotid gland and the neck.

Table 2. Clinical P (Parotid) and N (Neck) Stage
P0 263662


Surgery was the definitive mode of treatment in 291 of the 322 patients (90%), whereas 10% of patients received definitive radiotherapy alone. Combined parotidectomy and neck dissection was carried out in 171 (59%) of the surgically treated patients, whereas 79 (27%) patients had parotidectomy only and 41 (14%) patients had neck dissection only. Among the 250 patients who had parotid surgery, pathologically positive metastatic disease was found in 227. Because the study is retrospective we cannot be sure of the indication for paratodectomy; however, 260 patients were known to have clinical parotid involvement (Table 2). Therefore, this could represent a 13% false-positive rate for clinical examination of the parotid gland. A total of 88 patients had therapeutic neck dissection and pathologically positive nodes were found in 76 (86%) of these. Among patients with disease in the parotid gland, 43 (17%) also had clinical neck disease, whereas 124 patients had elective neck dissections. Among those patients having elective neck dissections, 28 were pathologically positive, giving an overall incidence of 27% pathologic node involvement among patients with proven metastatic SCC in the parotid gland.


Adequate data concerning radiotherapy treatment were available for 300 of 322 patients. Thirty-three patients were known to have received no radiotherapy, whereas a total of 267 patients were irradiated. This represented definitive therapy in 31 (10%) patients of the study group. In the remaining 236 patients, radiotherapy was given as an adjuvant to surgery. Radiation treatment was directed at the parotid gland alone in 49 patients and to the neck in 32, and both parotid and neck in 186. Radiotherapy doses ranged from 46–80 Gy (median dose, 60 Gy) to the parotid and from 46–76 Gy (median dose, 50 Gy) to the neck. Reasons for patients not having radiotherapy include: patient refusal, early recurrence, ‘favorable’ pathology, and other reasons that are difficult to identify because of the retrospective nature of the study.


Disease recurred in one-third of patients (105 of 322). This involved the parotid region in 42, the neck in 33, and distant sites in 30. Figure 1 shows that disease control in the parotid bed did not vary significantly with P stage. Recurrence at the primary cutaneous site was not assessed in this study.

Figure 1.

There was little difference in disease control in the parotid region based on P stage among the 260 patients with clinical parotid disease.


At the time of analysis, just over half the patients were alive and disease-free, whereas 84 had died of disease, 7 were known to be alive with disease, and 62 had died of other causes with disease controlled. Cumulative disease-specific survival was 74% at 5 years (Fig. 2). There was a nonsignificant trend towards a worse survival as P stage increased (Fig. 3), but this became statistically significant when patients staged P1 (that is, metastatic disease in the parotid gland up to 3 cm in diameter) were separated from those clinically staged P2 and P3 (Fig. 4).

Figure 2.

Cumulative disease-specific survival at 5 years was 74% for the entire cohort of 322 patients.

Figure 3.

Survival did not vary significantly when patients were separated into stages P1, P2, or P3.

Figure 4.

There was a statistically significant difference in survival when patients with early P stage (P1) were compared with those with more advanced disease (combined P2 and P3).

Clinical N stage did not appear to have a significant influence on survival (Fig. 5); however, when patients clinically staged N0 were separated from those clinically staged N1 and N2 (that is, N-negative necks vs. N-positive necks), there was a significant survival difference (P = 0.026) (Fig. 6).

Figure 5.

Patients clinically staged N0 appeared to have a better survival than those staged N1 and N2, but this was not significant.

Figure 6.

When clinical N1 and N2 stages were combined (cN+), those patients did significantly worse than those staged N0 (cN–).

Among patients with metastatic involvement of the parotid gland, pathologic neck node involvement significantly worsened survival (P = 0.027) (Fig. 7). It can be seen that the survival curves of patients pathologically staged N1 and N2 were very similar. When multivariate analysis was carried out, the following variables were found to have an independent influence on survival: clinical (P3) parotid stage (P = 0.033) and pathologic (N1, N2) neck stage (P = 0.005). Clinical neck involvement lost its effect on survival when multivariate analysis was carried out (P = 0.11).

Figure 7.

Patients with pathologically positive necks had a significantly worse survival than those without neck disease (pN0 vs. pN1 and pN2).


Cutaneous SCC is a common disease among Caucasian populations, particularly those of Anglo-Celtic origin living in regions where high levels of sun exposure occur. The disease occurs most frequently in Australia, where the estimated incidence is 250–300/100,000 per year. In fact, the disease is so common in Australia that mandatory reporting is not carried out. It is recognized, however, that a high proportion of these cancers occur on the skin of the head and neck and that they have a low but definite metastatic potential.6 Indeed, metastatic cutaneous SCC is the most common parotid malignancy encountered in Australia and New Zealand.9 This is known to be a biologically aggressive disease,2, 10 afflicting elderly males most frequently. There is a high incidence of extranodal spread2 and combined surgery and radiotherapy are recognized as being the most effective form of treatment.3, 11 There is also evidence that metastatic cutaneous SCC can behave more aggressively in immunosuppressed patients,8 such as those with organ transplantation and chronic hematologic malignancy such as chronic lymphocytic leukemia.

The present study represents an attempt to accumulate data that may shed further light on the clinical behavior of this disease. Aggregated data of this kind provide a heterogeneous population that may give a fairer view of the real world or may, conversely, cloud the true picture. Nonetheless, an honest attempt has been made to gather data from a number of centers to increase our knowledge. This and previous studies have been predicated, to some extent, on the hypothesis that the current TNM staging classification is too simplistic and limited to provide adequate prognostic discrimination among patient groups. For example, patients with metastatic cutaneous SCC are simply designated N1, irrespective of the extent of that metastatic disease. In the head and neck region this has the potential to lead to marked heterogeneity among patient groups. The N1 staging classification, as currently used, would apply equally to a patient with a single 2-cm parotid metastasis, a patient with a 6-cm parotid metastasis and facial nerve invasion, or a patient with metastatic disease in both the parotid and the neck.

A modification to the staging system has previously been recommended,7 aiming particularly to separate parotid and neck disease by the addition of a P stage to the TNM system, as it relates to cutaneous SCC involving the head and neck. It is not our contention that the recommended new staging system is necessarily definitive. Ultimately, a larger number of patients and mathematical modeling to identify appropriate subgroupings may be necessary. Nonetheless, the present study does confirm our original hypothesis that separation of P and N stages provides more accurate prognostic information, with the presence of neck disease significantly worsening survival.

Local control in the parotid bed did not vary statistically significantly with parotid stage, but survival was influenced by P stage. A larger patient cohort may be required to yield a sufficient number of events to demonstrate any difference in disease control in the parotid region. Analysis of cumulative survival by the Kaplan–Meier method, however, demonstrated that patients with P1 stage had significantly better survival (82%) than those staged P2 and P3 (69%). Furthermore, on multivariate analysis, patients with parotid stage P3, that is, those with disease more than 6 cm in diameter or those patients with facial nerve or skull base invasion, had a significantly worse survival than the others. Patients with advanced parotid metastases should be candidates for further clinical research into a possible role for adjuvant systemic chemotherapy.

Whereas there was a trend toward worse survival among patients with clinical neck disease in addition to their parotid disease, the results by multivariate analysis did not reach statistical significance (P = 0.111). This may be due to the confounding effect of false-positive and false-negative neck disease. Pathologic neck node involvement, however, did independently and significantly worsen survival among patients with parotid SCC. Interestingly, those staged N1 and N2 pathologically had an equally poor outcome, indicating that subgroupings of N stage did not provide prognostic significance in this study.

The overall incidence of metastatic neck involvement among patients with SCC in the parotid was 27% in the current study, just over half the incidence of neck involvement reported from the patient cohort analyzed in the initial study from the SHNCI. That institution contributed nearly 30% of the patients in the present analysis and this difference in incidence in metastatic disease may reflect a number of factors, including referral patterns, extent of elective neck dissection, and techniques of pathologic examination of neck dissection specimens. Nonetheless, the high incidence of neck involvement has been confirmed and previous recommendations that the neck should always be treated among patients with metastatic cutaneous SCC involving the parotid gland5 is reiterated.

There are clearly deficiencies with a study of this type. Despite the fact that the participating institutions are all recognized as being major head and neck oncology centers, there was variability in the extent of surgery and radiotherapy scheduling and dosage that have not been corrected for in the current analysis. It is beyond the scope of this study to make specific clinical recommendations about the appropriate extent of parotidectomy and management of the facial nerve.

Nonetheless, this multicenter retrospective study of patients with histologically proven metastatic cutaneous SCC involving the parotid gland or cervical lymph nodes has provided further information about the clinical behavior of this disease and also further evidence that supports a review of the current staging system. The data confirm initial impressions that the addition of P and N stages adds valuable prognostic information about cancer-specific survival. The value of subgroupings within the P and N stages remains uncertain.