Fax: (011) 61-2-98915814
Metastatic cutaneous squamous cell carcinoma of the head and neck
The Immunosuppression, Treatment, Extranodal spread, and Margin status (ITEM) prognostic score to predict outcome and the need to improve survival
Article first published online: 17 FEB 2009
Copyright © 2009 American Cancer Society
Volume 115, Issue 9, pages 1883–1891, 1 May 2009
How to Cite
Oddone, N., Morgan, G. J., Palme, C. E., Perera, L., Shannon, J., Wong, E., Gebski, V. and Veness, M. J. (2009), Metastatic cutaneous squamous cell carcinoma of the head and neck. Cancer, 115: 1883–1891. doi: 10.1002/cncr.24208
- Issue published online: 20 APR 2009
- Article first published online: 17 FEB 2009
- Manuscript Accepted: 20 OCT 2008
- Manuscript Revised: 2 OCT 2008
- Manuscript Received: 30 JUL 2008
- head and neck;
- lymph nodes;
- squamous cell carcinoma
The authors propose a prognostic score model using a prospective study of patients with regional metastatic cutaneous squamous cell carcinoma of the head and neck.
Two-hundred fifty patients were analyzed using a competing risks model to identify risk factors for survival. A risk score was obtained using the significant coefficients from the regression model, and cutoff points were determined that separated the score into 3 risk groups (low risk, moderate risk, and high risk).
At a median follow-up of 54 months (range, 1.3-212 months) 70 of 250 patients (28%) developed recurrent disease: Most were regional recurrences (51 of 70 patients; 73%) in the treated lymph node basin. After regional recurrence, a majority (73%) died of disease. The following 4 variables were associated significantly with survival: immunosuppression (hazard ratio [HR], 3.13; 95% confidence interval [CI], 1.39-7.05), treatment (HR, 0.32; 95% CI, 0.16-0.66), extranodal spread (HR, 9.92; 95% CI, 1.28-77.09), and margin status (HR, 1.85; 95% CI, 1.85-3.369); and those 4 variables (immuosuppression, treatment, extranodal spread, and margin status) were used to calculate the ITEM score. The 5-year risk of dying from disease for patients with high-risk (>3.0), moderate-risk (>2.6-3.0), and low-risk (≤2.6) ITEM scores were 56%, 24%, and 6%, respectively. Fifty-six of 250 patients (22%) died from another cause.
Patients who underwent surgery and received adjuvant radiotherapy had a better outcome compared with patients who underwent surgery alone. Patients who had moderate- or high-risk ITEM scores, usually because of extranodal spread and involved excision margins, had a poor outcome. The authors recommend considering these patients for inclusion in adjuvant chemoradiotherapy trials. Cancer 2009. © 2009 American Cancer Society.
Nonmelanoma skin cancer (NMSC) is the most common malignancy worldwide, and is at epidemic levels in many countries. With an aging population, increasing immunosuppression in the community, and changing social trends, the incidence is increasing.1 In Northern Australia, the annual incidence of cutaneous squamous cell carcinoma (cSCC) in men exceeds 1300 in 100,000 population.2, 3 Most lesions (80%-90%) arise on the sun-exposed head and neck (HN) after years of chronic ultraviolet (UV) radiation exposure, which often is associated with an outdoor occupation.4
Although only a minority of cSCC lesions (<5%) metastasize to regional lymph nodes, the absolute number of patients involved worldwide who may experience associated morbidity and mortality is not inconsequential.5, 6 Predicting the risk of developing lymph node metastases from cSCC is difficult, although several unfavorable primary features reportedly increase a patient's risk.7 Recent evidence from an Australian study suggested that horizontal tumor dimension was not associated strongly with the development of lymph node metastases.8 In that study, the median tumor size was only 15 mm, but the median tumor thickness was 6 mm. The current American Joint Committee on Cancer9 tumor-lymph nodes-metastasis (TNM) staging system for NMSC inadequately prognosticates for patients with a primary cSCC.10 In patients with cSCC, excluding T4 disease (extradermal invasion, such as muscle, bone, or cartilage), the only variable that is considered important in assigning T classification is horizontal dimension. Similarly, in the current TNM staging system, all patients with lymph node metastases are classified as N1 without any discrimination between the number and size of lymph nodes or their location (parotid vs cervical). An alternative lymph node staging system has been proposed that incorporates other important information.11
Patients who develop metastatic HN cSCC may die from uncontrolled lymph node recurrence despite treatment. Studies suggest that most patients who develop metastatic HN cSCC to lymph nodes should undergo surgery and adjuvant radiotherapy.12 Patients who receive that combined treatment achieve a better outcome (increased survival) compared with patients who receive either surgery alone or radiotherapy alone.13 Currently, there is no model with which to prognosticate the outcome of patients who develop metastatic HN cSCC. By using a large database of patients who were treated with curative intent, we developed a simple model using competing risks analysis that allows clinicians to assign a 3-tiered risk score. This current study builds on our extensive previous research14-17 and re-emphasizes the improved outcome with the addition of adjuvant radiotherapy.
MATERIALS AND METHODS
All patients between 1980 to 2005 who had metastatic cSCC to lymph nodes of the HN (parotid and/or cervical) and who were treated with curative intent were eligible for inclusion in this study. Patients must have undergone surgery and had to have biopsy-proven cSCC to HN lymph nodes. Most patients proceeded on to receive adjuvant radiotherapy within 6 weeks of surgery. A minimum follow-up of 2 years was required (unless recurrence and/or death occurred before the minimum 2 years). Patients who received radiotherapy alone were excluded.
Most patients underwent contrast-enhanced computed tomography scans of the HN and chest imaging (usually chest x-ray). Patients without an identifiable primary tumor underwent assessment of the upper aerodigestive tract to exclude another source, such as a mucosal SCC. In total, 250 eligible patients were identified. Management decisions were made within the context of the multidisciplinary HN cancer service at Westmead Hospital (Sydney, Australia). Relevant demographic, primary site, lymph node details, treatment, and outcome data were entered onto a comprehensive data form by the senior author (M.J.V.).
Prospectively (retrospectively before 1999) collected data were entered into a computer database. Because the population under investigation is aging, intercurrent deaths are not unusual. To properly interpret time-to-event data, these competing causes of death need to be incorporated into statistical analyses.18 Time-to-event outcomes (overall and progression-free survival) were analyzed using the cumulative incidence approach,19 whereas regression modeling in the presence of competing risks was performed using the method of Fine and Gray.20, 21A risk score was obtained using the significant coefficients from the regression model, and cutoff points were determined that separated the score into 3 groups. Although every attempt is made to select these cutoff points to produce groups of equal size, this may not always be possible if the score is not uniform over the patient risk profiles (which may be a function of both the number of significant factors and their distribution in the sample). Then, this score can be scaled to produce a usable algorithm for the classification of patients.
The median age at diagnosis was 67 years (range, 34-95 years) in 207 men and 43 women. All patients had a minimum follow-up of 24 months and a median follow-up of 54 months (range, 1.3-212 months). A minority of patients (15 of 250; 6%) was immunosuppressed, including 10 patients who had a diagnosis of leukemia/lymphoma and 5 patients who were renal transplantation recipients.
Primary Lesion Characteristics
One hundred seventy-eight of 250 patients (71%) had a documented index lesion either preceding or concomitant with the development of metastatic lymph node disease. It was not possible to identify an obvious primary tumor in 72 of 250 patients. In 26 of 153 patients (17%), the primary was recurrent. The median tumor size was 15 mm (range, 3-70 mm) and the median thickness was 6 mm (range, 0.5-30 mm). The most frequently documented primary sites were the ear (38 of 250 patients; 15%) and the frontotemporal area (31 of 250 patients; 12%). Only a minority of patients (30 of 250 patients; 12%) were assigned a well differentiated grade. With regard to treatment, most lesions (68%) either were excised (128 of 250 patients; 51%) or were treated with surgery and adjuvant radiotherapy (41 of 250 patients; 16%).
Location of Lymph Nodes
Most patients (135 of 250; 54%) developed lymph node metastases after treatment of the primary tumor with a median time to lymph node recurrence of 11 months. The majority of metastatic lymph nodes (121 of 250; 48%) were located in the parotid. The remaining 85 of 250 patients (34%) had metastatic cervical lymph nodes (levels I-V), and 44 of 250 patients (18%) had multiple lymph node levels involved.
Pathology of Lymph Nodes
The median number of metastatic lymph nodes was 2 (range, 1-29 lymph nodes), and the median size was 26 mm (range, 10-92 mm). Extranodal spread was present in 82% of patients, and 57% had a close (<2 mm) or involved surgical margin.
All patients underwent either surgery alone (28 of 250 patients; 11%) or surgery and adjuvant radiotherapy (222 of 250 patients; 89%).
One hundred fifty-two patients (61%) underwent a parotidectomy (in 95%, this was facial nerve-sparing surgery) either as the only operation (27 of 250 patients; 11%) or in conjunction with a neck dissection (125 of 250 patients; 50%). Ninety-eight patients (39%) underwent a neck dissection, including 50 of 250 patients (20%) who underwent selective neck dissection and 48 of 250 patients (19%) who underwent comprehensive neck dissection. After surgery, 35% of patients who had a clinically negative neck (cN negative) at presentation were diagnosed with pathologically involved lymph nodes (pN positive).
Two hundred twenty-two patients (89%) received external beam adjuvant radiotherapy. Most patients were treated with a 3-dimensional, computed tomography-planned approach using a photon wedge pair/anterior lower neck technique (100 of 222 patients; 45%), a lateral opposed/anterior lower neck technique (30 of 222 patients; 14%), a single electron field (35 of 222 patients; 16%), or a wedge pair only technique (27 of 222 patients; 12%). The remaining patients received a combination of radiotherapy modalities. Most patients with parotid metastases (76%) who received parotid bed radiotherapy also received radiotherapy to the lower neck. The median radiotherapy dose to the dissected parotid or neck was 60 gray (Gy) (range, 50-74 Gy) in 2-Gy fractions and 50 Gy to sites of subclincal disease.
Seventy of 250 patients (28%) developed a recurrence (Table 1). Most recurrences were regional (51 of 70 patients; 73%) and occurred most often in treated lymph node basins. Only 9 of 70 patients (13%) had systemic failure as a first site of recurrence. Patients who underwent surgery alone had a higher rate of regional recurrence compared with patients who received combined treatment (13 of 27 patients [48%] vs 38 of 164 patients [23%]). No patient who underwent surgery alone developed distant metastases as a first site of recurrence. The median time to recurrence was 8 months (range, 2-34 months); and, after regional recurrence, most patients (73%) died of disease.
|Treatment||Disease Site: No. of Patients (%)|
|Sx alone, n=28||0||13||0||0||13 (48)|
|Sx+RTx, n=222||1||38||9||8||57 (35)|
|Total||1 (1)||51 (73)||9 (13)||8 (11)||70 (28)|
On univariate analysis, the following clinical and pathologic variables were considered (Table 2.): immunosuppression (yes vs no), clinical location of lymph nodes (parotid only vs other), treatment (surgery vs surgery and adjuvant radiotherapy), pathologic lymph node size (≤3 cm vs >3 cm), pathologic number of lymph nodes (single vs multiple), extranodal spread (yes vs no), margin status (clear vs involved), pathologic stage of parotid (P0, P1, P2, or P3) and cervical lymph nodes (N0, N1, or N2), and pathologic extent of disease (parotid only, parotid and cervical, or cervical only).
|Immunosuppression (yes vs no)||3.36||1.51||7.5||.003|
|Location of nodes (parotid vs other)||1.16||0.69||1.94||.57|
|Treatment (Sx vs Sx + RTx)||0.47||0.25||0.99||.05|
|Lymph node size (<30 mm vs ≥30 mm)||1.02||0.1||1.03||.08|
|Lymph node count (single vs multiple)||1.07||0.99||1.15||.08|
|P stage (P0-P3)||0.94||0.67||1.32||.72|
|N stage (N0-N2)||1.3||0.94||1.79||.12|
|Extracapsular spread (yes vs no)||11.11||1.52||81.18||.02|
|Margins (clear vs involved)||1.83||0.99||3.39||.05|
|P−/N+ (N only metastases)||0.59||0.26||1.35||.21|
|P+/N− (P only metastases)||0.44||0.19||1.00||.05|
|P+/N+ (P and N metastases)||2.00||0.93||4.32||.08|
|Immunosuppression (yes vs no)||3.13||1.37||7.05||.006|
|Treatment (Sx vs Sx+RTx)||0.32||0.16||0.66||.002|
|Extracapsular spread (yes vs no)||9.92||1.28||77.09||.03|
|Margin status (clear vs involved)||1.85||1.02||3.37||.044|
|Treatment: Sx+RTx/Sx only||−1.8/0|
|Extracapsular spread: Yes/no||4.8/0|
|Margin status: Involved/clear||1.0/0|
Cox Regression Analysis
On multivariate analysis, the following variables were associated significantly with survival (Table 1): immunosuppression (yes vs no; hazard ratio [HR], 3.13; 95% confidence interval [CI], 1.39-7.05), treatment (surgery vs surgery and adjuvant radiotherapy; HR, 0.32; 95% CI, 0.16-0.66), extranodal spread (yes vs no; HR, 9.92; 95% CI, 1.28-77.09), and margin status (clear vs involved; HR, 1.85; 95% CI, 1.85-3.37). By using the coefficients of these significant variables as weights, a risk score was calculated for each patient. Patients were then classified into 3 groups according to cutoff scores. Patients with scores ≤2.6 were classified as low risk, those with scores >3 were classified as high risk, and those with scores within these limits were classified as moderate risk. Compared with the low-risk group, the HR was 4.53 (95% CI, 1.9-11.07) for the moderate-risk group and 9.46 (95% CI, 3.72-24.06) for the high-risk group (Table 2).
Survival curves (based on the cumulative incidence function) for each variable in the model are presented in Figures 1 through 4. The 5-year risk of dying from disease for the high-risk, moderate-risk, and low-risk patients was 56%, 24%, and 6%, respectively (Fig. 5). The overall survival rate (adjusted for competing risks) was 85% (95% CI, 81%-90%) at 2 years and 78% (95% CI, 73%-84%) at 5 years. Fifty-six of 250 patients (22%) died from another cause.
To our knowledge, this is the largest study to date of patients with metastatic HN cSCC from a single institution and reflects the management decisions of an established multidisciplinary team. Although metastatic HN cSCC is relatively uncommon, in absolute numbers, it afflicts many patients worldwide. Patients who develop metastatic lymph node metastases usually do so after they receive treatment for an index cSCC rather than presenting with a concomitant primary and metastatic cSCC. A few patients will not have an index cSCC and may be diagnosed with an unknown primary cSCC after the exclusion of other putative sources (eg, mucosal SCC). In most of these patients, the lymph node sites (eg, parotid gland, external jugular lymph nodes) are rarely if ever sites of metastases from mucosal HN SCC. In the current study, most patients were older (aged >65 years) white men, and many had been treated previously for an HN NMSC. Lesions located in proximity to the parotid gland, the so-called “metastatic basin”22 of the HN, have an increased potential to metastasize. However, accurately predicting which patients are at risk of developing lymph node metastases from an HN cSCC is an area of ongoing research as is the role of elective lymph node treatment and sentinel lymph node biopsy.
In patients who experienced a recurrence event, lymph node failure was the dominant site of first recurrence (approximately 75%-80%) with distant metastases (lung, skeletal, etc) relative uncommon (<20%) as the first site of recurrence.12-17 Recurrence reflects clinical progression of residual microscopic cSCC either after surgery alone or after surgery and adjuvant radiotherapy. The objective of adjuvant radiotherapy is to decrease the risk of regional recurrence by eradicating microscopic cSCC. In our study, almost half of all patients (48%) who underwent surgery alone experienced lymph node recurrence compared with 38 of 222 patients (17%) who underwent surgery and received radiotherapy. The importance of achieving lymph node control is reflected by the poor outcome after recurrence, which, in most patients, was incurable. In our study, only a minority of patients who developed an isolated lymph node recurrence ultimately were cured despite treatment.
Patients who receive radiotherapy alone are unlikely to be cured. Although the data are limited, anecdotally these patients have a very poor outcome. Therefore, we suggest that there is no evidence to support a nonoperative curative approach as an alternative to surgery with or without adjuvant radiotherapy. Many patients who receive radiotherapy alone are inoperable medically (severe cardiopulmonary disease) or have advanced inoperable disease (skull base or carotid artery invasion). Accepting that most patients will undergo surgery, in the current study, we excluded patients who received radiotherapy only. Despite a lack of randomized data comparing surgery alone with surgery plus adjuvant radiotherapy the available evidence, albeit observational, provides compelling evidence of an improved outcome with combined treatment. Some authors still regard the addition of adjuvant radiotherapy as controversial23; however, given the emerging body of evidence supporting its benefits, we believe that this belief is incorrect. Again, as documented previously and as demonstrated in the current study, patients who received adjuvant radiotherapy experienced a significant survival benefit (5-year survival rate, 82% vs 60%; P = .05) as a result of improved regional control.
The current TNM staging system for lymph node metastatic HN cSCC is inadequate and assigns all patients as stage N1 without any consideration of lymph node size, count, or location. To improve on this flawed system and using a modified parotid (P1-P3) and lymph node (N0-N2) staging system, O'Brien et al24 have proposed a new staging system that incorporates these features. A recent multi-institutional study added support to this new system with findings suggesting that the separation of parotid and neck disease by PN stage is valid and prognostic.11 Despite those findings, important variables, such as extranodal spread, margin status, immune status, and treatment, were not considered in that new staging system.
We identified 4 significant variables on multivariate analysis that were important predictors of prognosis: immunosuppression, treatment, extracapsular spread. and margin status (ITEM). It is noteworthy that we were unable to confirm a significant difference in outcome when assigning a PN stage to our current cohort of patients; thus, consequently, lymph node size, lymph node count, and lymph node location are not considered in our weighted prognostic model. The importance of these 4 variables in our prognostic model is not surprising, because all of these are accepted poor prognostic variables in patients with HN cancer. Extranodal spread is a significant predictor of recurrence in patients with mucosal HN SCC25 and cutaneous HN SCC.14 The finding of extranodal spread reflects a high likelihood of residual microscopic cSCC in the dissected parotid or neck. In our study, most patients had a finding of extranodal spread. Immunosuppression is associated with an increased risk of developing cancer, especially skin cancer, and also of experiencing a poor outcome despite treatment. Studies of immunosuppressed patients who developed HN cSCC support this, and 1 study reported that no immunosuppressed patients remained alive at 2 years post-treatment compared with 87% of patients who remained in the immunocompetent group.26 In a large study (n = 170), patients who were immunosuppressed were significantly more likely to die compared with immunocompetent patients (HR, 3.8; P<.0001).27 Although only a minority of our patients (<10%) were immunosuppressed in the current study, this variable proved to be a powerful prognosticator. The inability to achieve clear surgical margins, particularly after parotidectomy, is accepted as an unfavorable variable in HN cancer because of the high likelihood of residual microscopic cancer and recurrence. We observed that almost 60% of patients had close or positive margins. Khurana et al also reported that positive surgical margins were associated with poor local disease control (P = .02).28 Similarly, Ch'ng et al reported that margin status had a significantly unfavorable impact on survival (P < .01).27, 29
We acknowledge the many potential biases in comparing outcomes from different treatments delivered over many years and using nonrandomized methodology. In our study, only 28 patients underwent surgery alone compared with 222 patients who received combined treatment. However, it is unlikely that randomized data will become available to support our assertion that adjuvant radiotherapy is beneficial; and, to our knowledge, no such studies have been registered or proposed. Despite these misgivings, previous studies have highlighted the importance of adjuvant radiotherapy in conferring better regional control and improved survival compared with either surgery alone or radiotherapy alone.12-17 In a series of patients with metastatic cSCC to cervical lymph nodes, the 5-year disease-free survival rate was improved significantly in patients who underwent surgery and received adjuvant radiotherapy compared with patients who underwent surgery alone (73% vs 18%; P = .001), and locoregional control also was improved (77% vs 15%).14 Jol et al also reported a decreased locoregional failure rate in patients who underwent surgery and adjuvant radiotherapy compared with patients who underwent surgery alone (17% vs 44%).30 We observed a significant risk reduction of 68% (P = .006) for patients who underwent combined treatment, similar to what was reported in another study (n = 170 patients; 77% of those patients received adjuvant radiotherapy) that also documented a highly significant HR of 0.4 (P = .0003).27 Similarly, in a study of 184 patients who had unfavorable primary parotid tumors, adjuvant radiotherapy (vs surgery alone) reduced the risk of death by 50% (HR, 0.5; P = .05).31
High-risk pathologic and clinical variables can define risk groups of patients with metastatic lymph node HN cSCC. In our study, immunocompetent patients who had ITEM scores ≤2.6 had a low risk (6%) of dying from cancer. However, most patients in our study had either an involved surgical margin and/or the presence of extranodal spread. Even assuming that they received adjuvant radiotherapy, many patients still would be in the high-risk group because of the powerful negative effect of extranodal spread. It is these patients, who made up almost 60% of the patients in our study, who may benefit the most in absolute terms from means to improve outcome. There is compelling evidence in patients with mucosal HN SCC that adjuvant chemoradiotherapy improves locoregional control and survival, albeit with some increased toxicity.32 Data from a single-arm study at Peter MacCallum Cancer Institute33 (Melbourne, Australia) in which patients were given weekly, concomitant, platinum-based chemotherapy and adjuvant radiotherapy suggest a possible role for adjuvant chemoradiotherapy to improve outcome in patients with unfavorable cSCC (extracapsular spread, close/positive margins). An international, multi-institutional, randomized trial testing this hypothesis, under the auspices of the Australian and New Zealand Trans Tasman Radiation Oncology Group, currently is underway (Postoperative Skin Trial 05.01; available at: www.trog.com.au accessed January 2009). Pending the results of that important trial, the future management of patients with moderate- to high-risk, metastatic lymph node HN cSCC may include adjuvant chemoradiotherapy.
Conflict of Interest Disclosures
The authors made no disclosures.
- 9AJCC Cancer Staging Handbook: From the AJCC Cancer Staging Manual, 6th ed. New York, NY: Springer-Verlag, 2002..
- 19Depicting time-related events after cardiac surgery: Kaplan-Meier or competing risk? Asia Pacific Heart J. 1998; 7: 98-102., , .
- 33Post-operative chemo-therapy for high-risk head and neck squamous cell carcinoma (HNSCC) [abstract]. Presented at the 12th International Congress of Radiation Research, Brisbane, Australia, August 17-22, 2003. Abstract PP07/0267:97., , , et al.