Squamous cell carcinoma neck lymph nodes survival rate

INTRODUCTION

Section:

Nonmelanoma skin cancer (NMSC) is the most common cancer in the United States, accounting for more than one million new cases annually.1 Squamous cell carcinoma (SCC) and basal cell carcinoma (BCC) constitute nearly all NMSC, and the incidence of these tumors continues to increase.2 At least 75% of NMSC arises in the head and neck. Most NMSC cases are controlled successfully by complete excision, although they are associated with a lifelong risk for developing second primary skin cancers that can require repeated surgical resections and reconstructive procedures.3 Although rarely fatal, NMSC has tremendous adverse public health effects of high medical costs and, in advanced or aggressive cases, diminished quality of life from devastating aesthetic and psychosocial sequelae, functional impairment, and other serious consequences.4,5

Fortunately, only a small percentage of primary NMSCs, mostly SCCs, are refractory to standard dermatologic therapy.6-8 Compared with skin BCCs, skin SCCs not only are more likely to metastasize but also to cause mortality. Although the case-fatality rate is only approximately 1%, the national NMSC mortality figures equal or exceed those for melanoma, which is far more lethal but less common.9,10 The present prospective study of skin SCC was designed to determine whether certain clinical-pathologic features of skin SCC are associated with an increased risk of disease-specific mortality, as suggested by previous data.6

PATIENTS AND METHODS

Section:

We enrolled skin SCC patients consecutively and prospectively at a single institution (The University of Texas M.D. Anderson Cancer Center). All patients treated at M.D. Anderson for SCC of the skin of the head and neck or other sites are entered (after giving their consent) into the institution's Non-Melanoma Skin Tumor Registry, from which our patients were identified. The eligibility criteria were a minimum of one pathologically confirmed skin SCC lesion at presentation, definitive treatment of the SCC lesion(s) resulting in no evidence of disease (NED), and a minimum of 2 months of follow-up after definitive treatment. Patients were enrolled onto the study only after providing informed written consent. All skin SCC patients at M.D. Anderson were potentially eligible for our analysis; in other words, none were excluded for any reason related to the specific nature of the SCC (eg, nodal involvement).

Study Design

This study was a prospective, longitudinal analysis of risk factors for mortality from skin SCC. The study protocol and all accompanying forms and surveys were reviewed and approved by the institutional review board. Each patient's skin SCC was evaluated comprehensively at the beginning of study, or baseline. This evaluation included bidimensional measurements, depth of invasion, perineural involvement, lymphatic or vascular involvement, extracapsular spread, nodal involvement, and clinical, imaging, and pathology assessment. Study patients completed interviews and epidemiologic surveys and had a comprehensive history and physical examination. All patient identifiers were coded to assure patient confidentiality. Data were entered with internal validity checks. After enrollment and definitive treatment, all eligible patients had 4-month follow-up evaluations. These included a complete physical and skin examination and clinical and pathologic evaluations only as indicated for new or recurrent cancers, symptoms, or signs. Hematoxylin and eosin–stained sections from all patients were reviewed in a blinded fashion for different histopathologic features and recorded by two experienced pathologists. Cases with discrepancies were reviewed by a third pathologist.

End Point Evaluation

We established a Clinical End Points Committee to determine whether SCCs that developed during the present study were recurrences or second primary tumors (SPTs). All newly developed lesions were considered recurrences unless they differed from the pathologic/histologic criteria of or were located more than 5 cm away from the incident SCC. Criteria for postoperative radiation therapy included the presence of perineural invasion, extracapsular lymph node extension, more than one positive lymph node metastasis, and direct invasion of the bone. Patients previously treated with external-beam radiation therapy were not re-treated. Every deceased patient was evaluated by the Clinical End Points Committee to determine whether the cause of death was NMSC.

Statistical Analysis

Summaries of eligible patient data used standard descriptive statistics and frequency tabulation. Associations between categoric variables were assessed via cross-tabulation, χ2 test, and Fisher's exact test. Differences in numerical variables between groups, including those for size and depth of recurrent or nonrecurrent lesions, were compared via t test or Wilcoxon rank sum test.

The times to recurrence, death, and disease-specific death were defined as the times from NED to these various events. The survival probabilities for time-to-event end points were estimated via the Kaplan-Meier method and log-rank test. The Cox proportional hazards regression model was used to assess patients' prognostic factors for disease-specific survival (DSS). DSS was defined as the time to cancer-related death, including from relapse or new skin primary cancers. Patients who were still alive were censored at the time of the last follow-up regardless of the disease status. Predictive variables were then selected via forward stepwise selection with a P value cutoff of ≤ .05; a previously deleted variable was allowed to re-enter the final model if its P value was ≤ .05. The likelihood ratio test was used to compare the models. Recursive partitioning analysis based on martingale residuals was applied to classify patients into various risk groups based on DSS. Cutoff values were identified by the recursive partitioning method for continuous variables included in the model (eg, lesion size) or by medians for variables not included in the initial model (eg, depth of invasion).

Statistical analysis was performed with SAS 8.0 (SAS Institute, Cary, NC) and S-plus 2000 (Version 3.3, Statistical Sciences, Seattle, WA) software. All tests of statistical significance were performed using a two-sided 5% type-I error rate.

RESULTS

Section:

Patient Characteristics/Demographics

Between July 1996 and June 2001, 260 patients at The University of Texas M.D. Anderson Cancer Center were diagnosed with SCC of the skin and entered into the Non-Melanoma Skin Tumor Registry. Twenty-four of these patients were ineligible for our study because they had in situ disease or did not have definitive therapy, and 26 patients were ineligible because they did not have adequate follow-up data (a minimum of 2 months). This left 210 patients with at least one SCC lesion at presentation, definitive treatment resulting in NED, and a minimum of 2 months of follow-up who were eligible and consecutively enrolled onto the present prospective, longitudinal study (Table 1). The median enrollment age was 68 years old (range, 34 to 94 years). There were 187 men, 23 women, five Hispanic American patients, and 205 white patients. To date, 52 study patients have died. The median follow-up of the 158 surviving patients after enrollment has been 22 months (range, 2 to 72 months).

All 210 patients had pathologically confirmed NMSC at presentation. Among them, 80 patients (38%) came to M.D. Anderson for treatment after referral for local or regional skin SCC recurrences (SPTs were not recorded for these patients). Sixty-seven patients (32%) reported having had previous skin lesions, and 63 patients (30%) presented with primary skin SCC. Regarding patients who were referred for or who self-reported previous skin lesions, we obtained pathologic confirmation of previous-lesion SCC histology for 35 of the 80 referred patients and two of the 67 self-reporting patients. Age was correlated significantly with the diagnosis of any previous NMSC (P = .004).

Lesion Characteristics at Presentation

Two hundred seventy-seven SCC lesions were present at baseline, or study enrollment, in our 210 eligible patients. Twenty-three percent (65 of 277) of the lesions were local, and 6.5% (18 of 277) were regional recurrent NMSC. Forty (19%) of the 210 patients presented with more than one synchronous lesion in the skin. This factor was not related significantly to age (P = .75). However, more than one lesion at presentation was correlated with reported multiple previous NMSCs (P = .018) and with previous other malignancies with or without NMSC (P = .004). The proportions of previous NMSC were 80% (32 of 40) in patients with multiple skin SCCs at diagnosis for the present study and 52.4% (89 of 170) in patients with only one skin SCC at diagnosis.

Eighty-seven percent (240 of 277) of the reported lesions occurred in the head and neck. Twenty percent of the head and neck skin lesions (49 of 240) occurred in the forehead, 19% (46 of 240) occurred in the face, 18% (42 of 240) occurred in the auricle, and 17% (40 of 240) occurred in the cheek. Twenty-five referred head and neck patients had delayed regional lymphatic recurrences of previously treated SCC lesions. Seventeen of these 25 patients had a history of multiple skin head and neck SCC lesions, and no primary site could be specified. Eight other patients with SCC of the midface skin presented with regional lymphatic failure.

Clinical and/or pathologic estimates of size (greatest diameter) were available for 257 of the total 277 skin SCC lesions. Overall, the median size of a lesion at study diagnosis was 2.0 cm (range, 0.2 to 17.0 cm). Locally recurrent lesions were significantly larger (median, 2.4 cm; range, 0.3 to 10.0 cm) than were nonrecurrent (de novo) lesions (median, 1.5 cm; range, 0.2 to 17 cm; P < .0001, Wilcoxon test).

Histopathologic Features

All study tumors were reviewed by two highly experienced pathologists, and skin SCC was confirmed in all study patients. Histopathologic features at presentation were as follows: 136 patients had one SCC; 38 patients had 59 SCC in situ lesions; 33 patients had SCCs and one or more skin BCCs (involving a total of 69 lesions); two patients each had one SCC and one spindle-cell variant of SCC histology; and one patient had two basosquamous plus seven SCC lesions.

Of the 277 total skin SCC lesions, 39 (14%) involved perineural invasion, 29 (10.5%) involved lymphatic or vascular invasion, and 43 (15.5%) had deep invasion beyond subcutaneous tissues. The rates of perineural, lymphatic or vascular, and deep invasion beyond subcutaneous tissues in recurrent versus nonrecurrent lesions were 24% v 10% (P = .002; perineural), 17% v 8% (P = .022; lymphatic or vascular), and 30% v 10% (P < .0001; invasion beyond subcutaneous tissues). Increment of size was correlated directly with the incidence of perineural invasion (P = .002) and lymphatic or vascular invasion (P < .001) in nonrecurrent SCC lesions. There was no detectable size-invasion association in locally or regionally recurrent lesions.

Deep Invasion Beyond Subcutaneous Tissues

Sixteen percent (43 of 277) of the baseline lesions (definitively resected at the start of study) invaded deeply into one or more of the adjacent tissues, and these 43 lesions occurred in 40 patients, of whom 37 had a single and three had two deeply invasive SCC lesions. The proportion of deep invasion was significantly higher for recurrent than for de novo skin SCC, 29.6% (24 of 81; recurrent) versus 9.8% (19 of 194; de novo; P < .0001). The proportion of deep invasion increased significantly with nonrecurrent SCC lesion size (P = .0001) but only marginally significantly with recurrent lesion size (P = .054). Lesions managed by standard dermatologic approaches were excluded from the deep-invasion analyses because invasiveness could not be assessed histopathologically.

Treatment

Surgery, including wide local excision, comprehensive dissection of regional lymphatics, or both, was part of treatment for 93.8% of all patients (197 of 210 patients; Table 2). The majority of surgery patients received wide local excision alone (n = 113), 29 patients received comprehensive dissection of regional lymphatics alone, and 16 patients received both wide excision and lymph node dissection. Thirty-nine patients received surgery plus adjuvant radiation therapy, and six patients received radiotherapy alone. No early-stage, low-risk patients received radiation therapy. Seven patients received standard dermatologic management (shave excisions, cryoablation therapy, and topical cytotoxic therapy) alone. Major soft tissue reconstruction alone (eg, pedicled flap and free flap) or craniofacial reconstruction was administered for 29% of patients (61 of 210 patients). Disease requiring major soft tissue reconstruction was associated with a larger size (mean of 4.47 cm ± 2.78 cm [standard deviation]) versus disease without major reconstruction (2.14 cm ± 2.05 cm; P ≤ .0001).

Study Outcomes

Survival was computed as the time between enrollment and either death or last follow-up. Three-year overall survival was 70% (95% CI, 62% to 79%; Fig 1A). Fifty-two patients were known to have died, 21 patients from skin SCC and 31 patients from other causes. The DSS analysis censored patients at the time of death if the cause of death was not skin cancer or at the time of last follow-up for patients still alive at that time. Three-year DSS was 85% (95% CI, 78% to 92%; Fig 1B). The 3-year estimates are derived from the complete follow-up on the 52 deceased patients and a median follow-up of 22 months in the 158 patients still alive at the time of our analysis. The use of 3-year estimates was justified by the reasonably tight range of the 95% CI estimations.

Recurrence

After skin SCC treatment, 44 recurrent lesions (22 local, eight regional, and 14 distant recurrences) occurred in 37 patients during the study period. Twenty-two of these patients were among the 80 patients who had prior recurrences, and 15 patients were among the 130 patients who had no prior recurrences (on-study recurrence rates were 27.5% v 11.5% in patients with and without prior recurrences, respectively; P = .003). Nine of the 22 patients with prior recurrences had received prior radiotherapy. Thirteen patients had distant/dermal metastases (11 patients had metastases only, one patient had metastasis and local recurrence, and one patient had metastasis and regional recurrence); 10 patients developed regional recurrences (eight patients had regional recurrences only, and two patients had regional and local recurrences); and 14 patients had local recurrences only. Fourteen of the 37 patients who experienced recurrence also developed second primary tumors. The Kaplan-Meier estimate showed that 80% (95% CI, 74% to 87%) of the study patients were free from recurrence at year 3 (Fig 1C).

Skin SPTs

A total of 177 SPTs (120 SCCs, 55 BCCs, two others) occurred during follow-up in 56 patients (SPTs alone in 42 patients and SPTs plus recurrences in 14 patients). The number of SCC SPTs per patient varied substantially. An average of one SCC SPT occurred every 2.5 person-years of follow-up. One patient developed 15 SCC SPTs over 46 months of intensive follow-up.

Risk Factors Associated With DSS

Table 3 shows the histopathologic factors that were investigated for their associations with DSS. Univariate analysis determined that the following baseline factors were statistically significantly associated with DSS: recurrent lesions (P = .05), deep invasion beyond subcutaneous tissues (P = .009), perineural invasion (P = .002), lesion size (P = .0003), and depth of invasion (P = .05). There was a statistically nonsignificant trend of an association between lymph node involvement and lower DSS: 3-year DSS of 69% with lymph node involvement versus 87% without lymph node involvement (P = .09).

Multiple covariate Cox models were constructed to determine which factors could independently affect DSS. Various models were compared by the likelihood ratio test. The best subset Cox model with two factors contained lesion size (≥ 4 cm v < 4 cm) with a hazard ratio of 4.5 (95% CI, 1.8 to 11.1; P = .002) and perineural invasion with a hazard ratio of 2.8 (95% CI, 1.2 to 6.6; P = .021). The associations among multiple prognostic factors were assessed via the log-linear model. Several significant two-way interactions were found: invasion beyond subcutaneous tissues was associated with the depth of invasion; lesion size was associated with recurrence (at the time of presentation) and depth of invasion; lymphatic/vascular invasion was associated with perineural invasion and lymph node involvement. No significant association was found between lesion size greater than 4 cm and perineural invasion. These factors, however, remain independent risk factors for DSS in the Cox model analysis.

A similar model was also identified by the recursive partitioning method. Lesion size (with a cutoff of 4 cm) was chosen as the first factor separating the high- and low-risk groups. In patients with lesions less than 4 cm, perineural invasion and deep invasion beyond subcutaneous tissues were subsequently chosen to further classify patients into various risk groups. Patients without these criteria and with lesions less than 4 cm in size had the lowest risk with respect to DSS (Fig 2). In fact, none of these patients died of disease, with the follow-up time up to 72 months. On the other hand, the 3-year DSS was 70% (95% CI, 59% to 83%) for patients who presented with at least one of the three risk factors. A clear separation of the DSS curves can be found in Figure 2. Importantly, delay in diagnosis or treatment was not significantly different between either of the two groups.

DISCUSSION

Section:

The present study prospectively evaluated features of skin SCC that can predict aggressive disease behavior and poor clinical outcome. Lesion size ≥ 4 cm and histologic evidence of perineural and deep invasion into underlying tissues were the clinical-pathologic factors associated most significantly with diminished DSS. There was a trend toward lower DSS in association with regional metastases that failed to reach statistical significance, possibly because of the low percentage and intensive management (eg, comprehensive dissection) of patients with nodal involvement and because only one node was involved in a majority of this small group of patients (Table 3). With a longer median follow-up (48 months) and larger number of patients with nodal metastasis (n = 74), a recent retrospective study in Australia11 (which has the highest incidence of skin SCC in the world) of advanced skin SCC patients treated with curative intent found a high recurrence rate (34%) and poor 5-year DSS (61%) in association with nodal involvement. This study also found that more intensively treated patients had better recurrence and DSS rates. All study patients came from a tertiary cancer center, which accounts for the much higher percentage of advanced or aggressive skin SCCs than generally would be seen in a community practice. Nonetheless, advanced cases do occur in the community, albeit less frequently, and so community practices can benefit from the findings we report here. We also assessed clinical factors (eg, sex and site of skin lesion), which were not significant predictors of DSS (data not shown).

Although disease-specific mortality is an important concern for advanced skin SCC patients, our study demonstrates that morbidity is equally, if not more, important, at least in the short term (our median analysis was < 2 years). DSS was 85%, compared with a 70% overall survival. The morbidity rate from disease requiring major reconstruction, either soft tissue alone (such as pedicled flap and free flap) or craniofacial reconstruction, was 29%. A striking 38% (79 of 210) of patients developed 221 new skin lesions after enrollment, 80% of which were SPTs. We believe that the morbidity and economic and health care utilization costs of skin SCC are largely underestimated and that the opportunities for preventive intervention strategies are great.

A multidisciplinary team of otolaryngologists, head and neck surgeons, Mohs micrographic surgeons, and dermatologists, plastic surgeons, general surgeons, medical oncologists, and radiation therapists enrolled patients onto this study. Notwithstanding this multidisciplinary breadth, 87% of the baseline skin SCCs in our study occurred in the head and neck region, reflecting the high incidence of NMSC in sun-exposed areas.12 This percentage of skin SCCs that occurred in the head and neck is higher than that reported in retrospective reviews.3,13

Our present results from prospective analyses of several prognostic factors in a wide range of skin SCC patients are unique in the literature. Previous studies have involved either early-stage patients with virtually 100% cure rates14 or advanced-disease patients usually assessed retrospectively for only one prognostic factor (eg, perineural invasion15 or nodal involvement11). This report also presents the first prospective study of the incidence and outcome of skin SCC patients in the southern latitudes of the United States. Although the incidence of skin SCC doubles every 8 to 10 degrees of declining latitude,16 most large skin cancer studies in the United States have been in the northerly latitudes of Minnesota,17 New York,18 and the Pacific Northwest.3 Therefore, the data reported here may differ from previously reported data because of population- and geography-based differences in NMSC, particularly skin SCC.

Our ongoing study and future plans in this setting include long-term follow-up of the present study's population to extend the relatively short-term findings of this report and further define the factors that best predict poor clinical outcomes. We plan to analyze the molecular characteristics of the aggressive skin SCC defined in this report, and this analysis may result in novel biologic therapies that can reduce skin SCC treatment- and disease-related morbidity and mortality. Further study of and advances in risk assessment, surveillance, and prevention will be critical for controlling the substantial threat of skin SCC to the public health.

Authors' Disclosures of Potential Conflicts of Interest

Section:

The authors indicated no potential conflicts of interest.

Section:

Fig 1. Kaplan-Meier curves (for all 210 squamous cell carcinoma patients) of (A) overall survival in months (70% at 3 years; 95% CI, 62% to 79%); (B) disease-specific survival in months (85% at 3 years; 95% CI, 78% to 92%; censored at non–skin-cancer death or last follow-up of living patients); and (C) recurrence-free survival in months (80% at 3 years; 95% CI, 74% to 87%). E, number of events; N, total sample size.

Fig 2. Kaplan-Meier disease-specific survival (DSS) curves for high- (perineural invasion, deep invasion beyond subcutaneous tissues, lesion size ≥ 4 cm) and low-risk patients (all others). Three-year DSS was 70% (high risk) versus 100% (low risk; P < .0001, log-rank test). E, number of events; N, total sample size.

Table 1. Demographic Information of 210 SCC Patients Eligible for Analysis

Table 2. Treatment in 210 Skin SCC Patients

Table 3. 3-Year Disease-Specific Survival As a Function of Selected Histopathologic Features

© 2005 by American Society of Clinical Oncology

Supported in part by National Cancer Institute grant No. P01-5P01CA68233 (G.L.C.), grant No. CA97007, and Cancer Center Support Grant No. 5P30 CA16672.

G.L.C., J.J.L., and F.C.H. contributed equally to this work.

Authors' disclosures of potential conflicts of interest are found at the end of this article.

We sincerely thank our patients for providing the foundation of our analysis; our entire interdisciplinary skin cancer team, especially Shirley Taylor, Maride Uy, and Nicole Sieffert for their collaborative efforts; and Michael S. Worley for critically reviewing this manuscript.

REFERENCES

Section:

What happens if squamous cell carcinoma spreads to lymph nodes?

What is the survival rate when cancer spreads to lymph nodes?

Is neck lymph node cancer curable?

Is squamous neck cancer curable?