Kaplan-Meier curves with accompanying life tables.
Customize your JAMA Network experience by selecting one or more topics from the list below.
Levine DE, Karia PS, Schmults CD. Outcomes of Patients With Multiple Cutaneous Squamous Cell CarcinomasA 10-Year Single-Institution Cohort Study. JAMA Dermatol. 2015;151(11):1220–1225. doi:10.1001/jamadermatol.2015.1702
Copyright 2015 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.
Patients with multiple cutaneous squamous cell carcinomas (CSCCs) pose a management challenge for physicians, but their prognosis is unknown because outcomes have not been compared between patients who form single vs multiple CSCCs.
To compare outcomes in patients with 1 vs multiple CSCCs.
Design, Setting, and Participants
A 10-year retrospective single-institution cohort study at an academic tertiary care center of patients with dermally invasive (non–in situ) primary CSCC diagnosed from January 1, 2000, through December 31, 2009.
Main Outcomes and Measures
Electronic medical records were reviewed to determine the tumor stage (Brigham and Women’s Hospital tumor stage) and outcomes (local recurrence [LR], nodal metastases [NM], and death due to CSCC). Outcomes were compared between patients with 1 vs more than 1 CSCC via multivariable competing-risk regression adjusted for other significant cofactors.
Of 985 patients with CSCC, 727 had 1 CSCC, 239 had 2 to 9 CSCCs, and 19 had 10 or more CSCCs. Most patients with 10 or more CSCCs were immunosuppressed (15 of 19 [78.9%]). The median follow-up time was 50 months (range, 2-142 months). Patients with more than 1 CSCC had a higher risk of LR (subhazard ratio for 2-9 CSCCs, 1.8; 95% CI, 1.1-4.3; and for ≥10 CSCCs, 3.8; 95% CI, 1.4-10.0) and NM (subhazard ratio for 2-9 CSCCs, 3.0; 95% CI, 1.4-6.5; and for ≥10 CSCCs, 4.2; 95% CI, 1.4-10.4) compared with patients with 1 CSCC, adjusted for Brigham and Women’s Hospital tumor stage. The 10-year cumulative incidence of LR and NM was higher in patients with 2 to 9 CSCCs and markedly higher in those with 10 or more CSCCs compared with patients who had 1 CSCC (10-year cumulative incidence for 1 CSCC: LR, 3.0%; 95% CI, 2.0%-4.5%; and NM, 2.3%; 95% CI, 1.5%-3.8%; for 2-9 CSCCs: LR, 6.7%; 95% CI, 4.2%-10.6%; and NM, 5.9%; 95% CI, 3.5%-9.6%; and for ≥10 CSCCs: LR, 36.8%; 95% CI, 19.2%-59.0%; and NM, 26.3%; 95% CI, 11.8%-48.8%).
Conclusions and Relevance
Patients with multiple CSCCs warrant frequent follow-up because they have an elevated risk of LR and NM. In particular, patients with 10 or more CSCCs have markedly elevated risks of recurrence and metastasis.
The annual incidence of cutaneous squamous cell carcinoma (CSCC) in the United States is estimated to be between 400 000 and 700 000.1 Although most cases of CSCC are curable, a small subset recur, metastasize, and cause death.2 It is estimated that between 4000 and 8800 deaths due to CSCC occur annually in the United States. In southern and central states, deaths due to CSCC are estimated to exceed the number of deaths due to several common cancers, including melanoma.3
Several risk factors have been associated with poor outcomes in patients with CSCC in large cohort studies4-8 that used multivariable analysis, including large tumor diameter, depth of invasion, poor differentiation, perineural invasion, lymphovascular invasion, desmoplasia, immunosuppression, and ear, temple, or lip location. In Brigham and Women’s Hospital (BWH) tumor staging, tumors with 2 or more risk factors (diameter of ≥2 cm, tumor invasion beyond the subcutaneous fat, poorly differentiated histologic features, or large-caliber nerve invasion of ≥0.1 mm) are considered high-stage CSCC and carry an elevated risk of nodal metastases (NM) and death.9,10
A few studies11-15 have documented an increased risk of subsequent CSCC formation in patients with prior CSCCs. However, the effect of multiple tumor formation on CSCC outcomes is poorly quantified. There is only 1 cohort study,16 to our knowledge, that compared CSCC outcomes between patients with 3 or more nonmelanoma skin cancers and patients with fewer than 3 nonmelanoma skin cancers. It reported an elevated risk of new CSCCs and melanoma formation and a higher overall mortality rate among patients who formed multiple nonmelanoma skin cancers. There are no studies, to our knowledge, that specifically evaluate CSCC outcomes in individuals who form multiple vs single CSCCs. This study compares the outcomes of patients with 1 CSCC with those who form multiple CSCCs.
Methods of data collection and outcome ascertainment for this cohort have been reported previously.5 In brief, the BWH Department of Pathology electronic database was searched using the keywords squamous and carcinoma to locate all pathology reports with a diagnosis of CSCC from January 1, 2000, through December 31, 2009. The diagnoses of noncutaneous squamous cell carcinoma, in situ CSCC, and recurrent CSCC were excluded. Electronic medical records of included patients were reviewed. Outcomes of interest, including local recurrence (LR), NM, and death due to CSCC, were recorded.
Outcome data were obtained from the notes of primary care physicians, dermatologists, oncologists, head and neck surgeons, plastic surgeons, and/or radiation oncologists. Local recurrence was considered to have occurred if the pathology report documented invasive CSCC in the same anatomic location as a prior CSCC and the second lesion was considered a recurrence of the index primary tumor. Nodal metastasis was defined as pathologically confirmed CSCC in a draining nodal basin of the primary CSCC with no other potential source. Death due to CSCC was considered to have occurred if the treatment team documented that the patient died because of a specific CSCC or complications that directly arose from it. Additional information recorded from the medical records included demographic factors, such as age at the diagnosis of the primary tumor, sex, race, ethnicity, treatment of the primary tumor, and the immune status of the patient at the time of data collection. The Partners Human Research Committee approved this study.
Tumor stages were assigned to all tumors using the BWH tumor staging system, which was found to have better prognostic discrimination than the 2010 American Joint Committee on Cancer and the 2010 International Union Against Cancer tumor staging systems.9,10 Perinanal, vulvar, and eyelid CSCCs were not assigned a BWH stage because these sites have their own American Joint Committee on Cancer tumor staging systems. In BWH staging, tumors are classified based on the presence of the following 4 risk factors: poorly differentiated histologic features, a diameter of 2 cm or greater, perineural invasion of 0.1 mm or greater in diameter, and invasion beyond the subcutaneous fat. A tumor is considered BWH T1 if it has 0 risk factors, T2a if it has 1 risk factor, T2b if it has 2 or 3 risk factors, and T3 if it has all 4 risk factors or bone invasion.5,9 Tumors were considered high stage if they were classified as BWH T2b or T3 and low stage if they were classified as BWH T1 or T2a because low-stage tumors have a reportedly small risk of LR (1.4%), NM (0.6%), and death due to CSCC (0.2%) compared with high-stage tumors (26.0%, 23.9%, and 16.3%, respectively).
Baseline patient demographic and CSCC clinical characteristics were analyzed using descriptive statistics and frequency tabulation. Univariate and multivariable model building was conducted to determine if patients with multiple vs single CSCCs had different outcomes when adjusting for other factors that may affect outcomes, such as tumor stage, age, and immune status. By necessity, 1 tumor per patient was entered into the multivariable analysis. If a patient had only 1 CSCC, that was the tumor analyzed. If a patient had multiple CSCCs and developed a poor outcome, the skin cancer that resulted in the poor outcome was selected for analysis (regardless of stage). For 3 patients with multiple poor outcomes arising from different tumors, the skin cancer that resulted in the worst outcome was selected (eg, the tumor resulting in NM was selected instead of the tumor that only caused LR). For patients with multiple CSCCs (n = 235) without poor outcomes, 1 tumor was randomly selected for analysis. If such a patient had both high- and low-stage tumors, the randomly selected tumor was chosen from the high-stage tumors because they were rare among the cohort; this process increased the number of high-stage cases with good outcomes in the analysis. This selection was considered appropriate because such patients did not develop a poor outcome due to their high-stage CSCC while their low-risk tumors likely contributed little to their overall risk of developing a poor outcome. A sensitivity analysis was also conducted. For this analysis, 1 tumor was selected at random for all patients with multiple CSCCs without prioritizing the selection of tumors that resulted in poor outcomes or were classified as high stage.
Competing risk analysis using the method of Fine and Gray17 was conducted to examine univariate and multivariable associations with each outcome of interest, comparing outcomes of patients with single vs multiple CSCCs. The proportional hazards assumption was checked via Schoenfeld residual plots.
For LR, NM, and death due to CSCC, cases were censored on the date of the last follow-up or on the date of death per the Social Security Death Index. If death occurred more than 3 months after the date of the last follow-up, the patient was considered unavailable for follow-up and was censored on the date of last follow-up rather than the date of death.
Multivariable models were built through forward stepwise selection and backward elimination. Variables were added based on their relative effect estimates on univariate modeling and retained if the Wald test comparing the smaller model with the larger model was significant at P ≤ .05 or if the P value comparing the 2 models was borderline (>.05 to >.99) and addition of the variable changed the subhazard ratio by at least 10%. Life tables and Kaplan-Meier curves were generated to illustrate event-free survival for LR and NM and disease-specific survival for death due to CSCC. All statistical tests were conducted using a 2-sided 5% type I error rate. Analyses were conducted using Stata, version 12 (StataCorp LP).
The search verified 985 patients with 1832 cases of CSSC. The median follow-up time was 50 months (range, 2-142 months). Patients were divided into those with 1 CSCC (727 of 985 [73.8%]), those with 2 to 9 CSCCs (239 of 985 [24.3%]), and those with 10 or more CSCCs (19 of 985 [1.9%]). These groupings were chosen because they demonstrated the most significant risk differences in poor outcomes based on preliminary descriptive statistics. The median number of tumors in patients with 10 or more CSCCs was 15 (range, 10-80 tumors). Patients with multiple CSCCs were significantly more likely to be younger (P = .005), male (P = .03), and immunosuppressed (P < .001). Fourteen cases were not assigned a BWH tumor stage owing to their location on the vulva, perianal region, or eyelid. Although most patients (893 of 985 [90.7%]) formed low-stage tumors (BWH T1 or T2a), patients with multiple CSCCs were significantly more likely to form high-stage (BWH T2b or T3) tumors compared with patients who had 1 CSCC (P < .001). A total of 143 patients (14.5% of the cohort) were immunosuppressed for a variety of reasons, including organ transplantation (n = 58), rheumatoid arthritis (n = 36), chronic lymphocytic leukemia (n = 23), chronic chemotherapy (n = 8), or other causes (n = 18). A high fraction (15 of 19 [78.9%]) of patients with 10 or more CSCCs were immunosuppressed (Table 1).
Table 2 compares the 10-year cumulative incidence of poor outcomes by the number of CSCC tumors. The incidence of LR and NM was higher in patients with 2 to 9 CSCCs and markedly higher in those with 10 or more CSCCs compared with patients who had 1 CSCC (10-year cumulative incidence for 1 CSCC: LR, 3.0%; 95% CI, 2.0%-4.5%; and NM, 2.3%; 95% CI, 1.5%-3.8%; for 2-9 CSCCs: LR, 6.7%; 95% CI, 4.2%-10.6%; and NM, 5.9%; 95% CI, 3.5%-9.6%; and for ≥10 CSCCs: LR, 36.8%; 95% CI, 19.2%-59.0%; and NM, 26.3%; 95% CI, 11.8%-48.8%). No difference in death due to CSCC was observed between groups (1 CSCC, 2.2%; 95% CI, 1.4%-3.5%; 2-9 CSCCs, 2.1%; 95% CI, 1.0%-4.8%; and ≥10 CSCCs, 0%).
Poor outcome events were stratified by the number of CSCCs subdivided by the BWH tumor stage in Table 3. Poor outcomes were significantly higher in patients with high-stage tumors compared with those who had low-stage tumors regardless of the number of CSCCs that were formed in these patients. A higher proportion of patients with multiple high-stage CSCCs were immunosuppressed compared with those with a single high-stage CSCC (4 of 6 [66.7%] vs 18 of 75 [24.0%]; P = .04). This difference may be attributed to the increased median number of CSCCs in immunosuppressed patients with multiple CSSCs (median, 4 tumors; range, 2-80 tumors) compared with nonimmunosuppressed patients who had multiple CSCCs (median, 2 tumors; range, 2-38 tumors), giving immunosuppressed patients with multiple CSCCs a higher chance of developing high-stage CSCCs over time.
Formation of multiple CSCCs was associated with worse outcomes on multivariable analysis adjusted for BWH tumor stage. Patients with 2 to 9 CSCCs and 10 or more CSCCs had higher risks of LR and NM compared with patients who had 1 CSCC (subhazard ratios for LR with 2-9 CSCCs, 1.8; 95% CI, 1.1-4.3; for ≥10 CSCCs, 3.8; 95% CI, 1.4-10.0; and NM with 2-9 CSCCs, 3.0; 95% CI, 1.4-6.5; and ≥10 CSCCs, 4.2; 95% CI, 1.4-10.4). No difference in death due to CSCC was observed in patients with 2 to 9 CSCCs or 10 or more CSCCs compared with patients who had 1 CSCC (Table 4). Immunosuppression was not an independent predictor of outcomes on multivariable analysis. The results of the sensitivity analysis did not differ from the results of the primary analysis. The Figure shows Kaplan-Meier survival curves and life tables for LR, NM, and death due to CSCC for patients with 2 to 9 or 10 or more CSCCs vs 1 CSCC, adjusted for BWH tumor stage.
In this single-center cohort of patients with CSCC, CSCCs were more likely to recur (locally or via nodal metastasis) in patients with multiple CSCCs compared with patients who had 1 CSCC. Patients with multiple CSCCs were significantly more likely to develop high-stage tumors than patients with 1 CSCC on multivariable analysis of poor outcomes, so analyses were adjusted for tumor stage. However, even after adjusting for tumor stage, patients with multiple CSCCs had worse outcomes, especially those with 10 or more CSCCs. In fact, one-fourth of these patients developed nodal metastasis.
People with multiple CSCCs were more likely to be immunosuppressed, consistent with prior studies,14-16 yet immunosuppression alone had no significant effect on CSCC outcomes in multivariable analysis (LR, NM, or death due to CSCC). In addition, patients with 10 or more CSCCs were significantly younger than those with 1 or 2 to 9 CSCCs (mean age, 63 vs 69 and 71, respectively; P = .005). This difference is likely owing to many patients in this group having been immunosuppressed for many years. Of 19 patients with 10 or more CSCCs, 15 were immunosuppressed; the mean duration of immunosuppression among these patients was 20 years. With many years of continuous immunosuppression in these relatively young patients, the potential for multiple CSCC formation is likely increased.
This study may have been underpowered to fully evaluate the effect of immunosuppression or other cofactors that may have some interplay with multiple CSCC formation and outcomes of patients with CSCC. However, more than 75% of patients with multiple CSCC formations were immunocompetent, indicating that CSCC formation can be a significant problem for the immunocompetent and risks of poor outcomes are still elevated with apparently intact immunity when multiple tumors form in people. It is known that with advancing age, there is a relative decrease in immune surveillance in the skin, which may drive multiple CSCC formation in persons who are otherwise immunocompetent.18,19
Nearly 80% of the patients with 10 or more CSCCs (15 of 19 patients) were immunosuppressed. This group may be a particularly high-risk subgroup worthy of further outcome studies with larger cohorts because more than 25% of these patients developed NM. Somewhat surprisingly, none of these patients died of CSCC. It may be that high vigilance resulting in early detection of metastasis, aggressive treatment, and immunosuppression reduction contributed to high survival in this subgroup. However, it may also be that profound immune suppression resulted in death due to other causes (which affected 9 of these 19 patients [47.4%]) before death due to CSCC occurred.
There is a body of literature supporting oral retinoids for reduction of CSCC formation. In this cohort, only 6 of 239 (2.5%) patients with 2 to 9 CSCCs and 6 of 19 (31.6%) patients with 10 or more CSCCs were prescribed acitretin and were able to continue taking the drug for a minimum of 6 months (range, 6-83 months). It was effective in this group, with no additional CSCCs developing in 5 of 12 patients (41.7%).
The increased risk of poor outcomes with multiple CSCCs may be because each CSCC is its own event and carries its own inherent risk of recurrence and metastasis, yet these individual risks still accumulate in an individual. As the baseline risk of a tumor rises, so does the incremental increase in cumulative risk with additional occurrences of CSCC, particularly high-stage CSCC.
To our knowledge, this study is the first to compare the risk of LR, NM, and death due to CSCC between patients with multiple CSCCs and those with 1 CSCC. These findings substantiate the importance of close follow-up for dermatologic patients with multiple CSCCs, especially those with many tumors, and highlight the necessity for dermatologists to document prior CSCC sites, examine the scar sites of prior CSCCs, and perform lymph node examinations in those patients. Larger studies are required to determine which factors affect multiple tumor formation and subsequent outcomes.
Accepted for Publication: April 29, 2015.
Corresponding Author: Chrysalyne D. Schmults, MD, MSCE, Department of Dermatology, Brigham and Women’s Hospital, 1153 Centre St, Ste 4349, Jamaica Plain, MA 02130 (email@example.com).
Published Online: July 15, 2015. doi:10.1001/jamadermatol.2015.1702.
Author Contributions: Dr Levine and Mr Karia had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Schmults.
Acquisition, analysis, or interpretation of data: Levine, Karia.
Drafting of the manuscript: Levine, Karia.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Levine, Karia.
Administrative, technical, or material support: Levine, Karia.
Study supervision: Schmults.
Conflict of Interest Disclosures: None reported.