Objective To identify factors associated with poor outcomes in perineurally invasive squamous cell carcinoma.
Design Retrospective cohort study.
Setting Two academic hospitals in Boston, Massachusetts.
Patients Adults with perineural SCC diagnosed from 1998 to 2008.
Main Outcome Measures Hazard ratios (HRs) for local recurrence, nodal metastasis, death from disease, and overall death, adjusted for known prognostic factors.
Results A total of 114 cases were included, all but 2 involving unnamed nerves. Only a single local recurrence occurred in cases with no risk factors other than nerve invasion. Tumors with large nerve (≥0.1 mm in caliber) invasion were significantly more likely to have other risk factors, including diameters of 2 cm or greater (P < .001), invasion beyond the subcutaneous fat (P < .003), multiple nerve involvement (P < .001), infiltrative growth (P = .01), or lymphovascular invasion (P = .01). On univariate analysis, large nerve invasion was associated with increased risk of nodal metastasis (HR, 5.6 [95% CI, 1.1-27.9]) and death from disease (HR, 4.5 [95% CI, 1.2-17.0]). On multivariate analysis, tumor diameter of 2 cm or greater predicted local recurrence (HR, 4.8 [95% CI, 1.8-12.7]), >1 risk factor predicted nodal metastasis (2 factors: HR, 4.1 [95% CI, 1.0-16.6]), lymphovascular invasion predicted death from disease (HR, 15.3 [95% CI, 3.7-62.8]), and overall death (HR, 1.1 [95% CI, 1.0-1.1]). Invasion beyond subcutaneous fat also predicted overall death (HR, 2.1 [95% CI, 1.0-4.3]).
Conclusions Squamous cell carcinoma involving unnamed small nerves (<0.1 mm in caliber) may have a low risk of poor outcomes in the absence of other risk factors. Large-caliber nerve invasion is associated with an elevated risk of nodal metastasis and death, but this is due in part to multiple other risk factors associated with large-caliber nerve invasion. A larger study is needed to estimate the specific prognostic impact of nerve caliber.
Squamous cell carcinoma (SCC) is the second most common type of skin cancer, with approximately 200 000 new cases of SCC recorded in 2000 in the United States alone.1 While most SCCs can be cured with excision or Mohs surgery, approximately 5% of all SCCs will metastasize, and the rate of metastasis increases up to 20% in SCCs with high-risk features.2 Various guidelines delineating these high-risk features have been published based on available data. The American Joint Committee on Cancer (AJCC) published a modified staging system for cutaneous SCC in 2011, which included the following high-risk characteristics that have an impact on tumor stage: depth greater than 2 mm, Clark level IV or greater, perineural invasion, location on ear or vermillion lip, and poorly differentiated or undifferentiated growth pattern on histologic analysis.2
Perineural invasion (PNI) is one of the new additions to the AJCC staging system and occurs in approximately 2.5% to 5% of primary SCCs.3,4 Perineural invasion portends poor outcomes with reported risks of lymph node metastasis and distant metastasis of 35% and 15%, respectively, compared with SCC without PNI (in which risks were 15% and 3.3%, respectively in the same study).5 Squamous cell carcinoma with PNI is also associated with increased mortality. One study reported a 64% 3-year disease-specific survival for patients with SCC with PNI compared with 91% for SCC without PNI.6Previous studies have shown that the extent of nerve invasion has an impact on prognosis. Ross et al7 found significantly higher risks of recurrence, metastasis, and disease-specific mortality in 24 SCCs with PNI of nerves 0.1 mm or greater in diameter compared with 24 SCCs with PNI less than 0.1 mm in diameter. Another study found no difference in 5-year recurrence for 34 SCCs with PNI that are 0.1 mm or smaller vs 28 SCCs with PNI larger than 0.1 mm.8 However, this study was limited to patients treated with both surgery and radiation, most of whom had positive surgical margins prior to salvage radiation, portending a worse baseline prognosis. The study did find that focal nerve invasion carried a better prognosis than more extensive invasion. Similarly, patients with only microscopic PNI have been shown to have improved local disease control and 10-year survival compared with patients with clinically symptomatic or radiographic perineural invasion.9-11 Thus, the degree of nerve invasion is likely prognostically important, although optimal classification of varying degrees of invasion remains uncertain.
The lack of precise prognostic estimates for SCCs with different degrees of nerve invasion adjusted for other concomitant risk factors prevents clear guidance on the clinical approach to SCC with PNI. This is reflected in heterogeneity of management, as evidenced in a study of Mohs surgeons in which there was little consensus regarding when adjuvant radiation should be used.12 Clarification of the prognostic significance of different degrees of PNI in combination with other risk factors would aid clinicians in selecting appropriate workup and treatment while minimizing unnecessary treatment and resulting morbidity. This study was undertaken to address the current knowledge gap by comparing outcomes of patients with SCC with large- (≥0.1mm) vs small- (<0.1 mm) caliber PNI, adjusting for presence of other concomitant prognostic factors via multivariate analysis.
After approval from the Partners Healthcare institutional review board, all patients treated at Massachusetts General Hospital or Brigham and Women's Hospital from January 1, 1998, and December 31, 2008, with pathologically confirmed cutaneous SCC with PNI or neurotropism (NT) were identified via the dermatopathology and surgical pathology databases. The following words contained in the pathology reports were used to search the databases: squamous, nerve, nerves, neural, perineural, neurotropic, and neurotropism. Only patients with PNI in the primary tumor were included. Patients were excluded if the tumor did not have NT or PNI on review by the investigators, if the slides or medical records were not available, if the PNI was identified only in recurrent tumor, or if cutaneous origin of the primary tumor could not be confirmed.
Medical records were obtained, and the following information was extracted for each tumor: body location, date of procedure, tumor diameter, preceding scar, preceding radiation therapy, associated symptoms, duration of lesion prior to treatment, dates and methods of all treatments of the primary tumor, as well as any recurrence and/or metastasis, date and type of recurrence (local, in-transit, lymph node, distant organ), and whether a named nerve branch was involved. Patient-specific information was extracted including sex, race/ethnicity, skin cancer history, immunosuppression status, presence of noncutaneous malignant diseases, and, for deceased patients, cause of death as noted in medical records or death certificate.
Each case was assigned a study number, and slides were labeled with this unique study number to deidentify them. In a blinded fashion, slides of the tumors were reviewed (by M.M.J., J.B.C., and C.D.S.) for perineural invasion (defined as invasion of the nerve fiber or the presence of neoplastic cells within the perineural space, located between the perineurium and the nerve fiber) or neurotropism (defined as neoplastic cells adjacent to, but not within, the perineural space). Other histologic characteristics of the tumor were recorded, including maximum tumor depth in millimeters as well as by tissue level, involvement of margins on final excision of primary tumor, differentiation of tumor, and presence of lymphatic or vascular invasion. Recorded characteristics of the nerve(s) involved included diameter of the largest involved nerve, number of nerves involved, location of nerve invasion either within the tumor or discontiguous from the tumor, as well as tissue level of the perineural invasion. In cases in which any of this information was ambiguous, a consensus opinion was reached (by J.B.C. and C.D.S.). Both biopsy and excision specimens were evaluated for each primary tumor. Tumors were classified for study analysis according to the greatest degree of nerve involvement seen.
Cases were assigned to 1 of 2 groups: small-caliber nerve group (PNI of nerves <0.1mm) and large-caliber nerve group (PNI of nerves ≥0.1 mm). Tests for differences in patient characteristics by nerve caliber group were performed using χ2 test for nominal variables and t tests for continuous variables. In analyses with low case numbers of nominal variables, Fisher exact test was used. Large and small nerve caliber groups were compared regarding 4 outcomes of interest (local recurrence [LR], nodal metastasis [NM], death from SCC/disease-specific death [DSD], and all-cause death [ACD]) adjusted for other potential risk factors via multivariate Cox proportional hazard models developed through forward stepwise selection using a stay criterion of .05 or less. Other risk factors included age, sex, immunosuppression, tumor diameter, tumor depth, differentiation, number of nerves involved, lymphovascular invasion, and number of independent prognostic factors present (Jambusaria-Pahlajani et al13). Each tumor was analyzed as an independent case. Data analysis was performed using SAS statistical software (version 9.2; SAS Institute Inc). All hypothesis tests used a 2-sided significance level of .05.
Patient and tumor characteristics
The initial pathology database query identified 164 cases as primary cutaneous SCCs with PNI or NT. Twenty-three cases were excluded after histologic review because they had only NT (n = 13) or did not have any identifiable PNI or NT (n = 10). Of the 141 remaining cases, 15 cases were excluded owing to incomplete slide sets, and 12 cases were excluded owing to insufficient follow-up information in the medical record. The final cohort contained 114 cases of primary SCC, with PNI occurring in 96 patients, 46 cases with large-caliber nerve invasion, and 68 with small-caliber nerve invasion. The large-caliber and small-caliber groups were similar in many regards: there was a male predominance (71% male) and mean (SD) age at time of diagnosis was 71 (12) years. All cases occurred in patients identified as white with the exception of a single case in an African American and 5 cases in persons of unknown race. Two cases were in Latinos. Eight patients had more than 1 SCC with PNI. Six of these patients had 2 SCCs with PNI, 1 patient had 3, and a single heart transplant patient had 11 SCCs with PNI. Since the event rate was low in this subgroup (2 LRs and 1 NM occurring in 2 persons, both of whom had 2 SCCs with PNI), each case of SCC with PNI was analyzed independently. Immunostaining (cytokeratin) had been performed in only 4 cases. Two cases involved named nerves, all others involved unnamed nerves. Of 21 cases in which radiologic imaging of the primary tumor was obtained, no case had radiologic evidence of nerve invasion. One patient noted a “sharp pain” at the site of the tumor. There was no other clinical evidence of nerve invasion. Additional characteristics, including all variables that significantly differed between large- and small-caliber nerve groups, are summarized in Table 1.
Tumors with large-caliber nerve invasion were statistically more likely to have other risk factors than small-caliber nerve invasion cases, specifically tumor diameter greater than 2 cm (37% vs 7%; P < .001), maximum depth of tumor greater than 4 mm (76% vs 62%; P = .003), invasion beyond subcutaneous fat (41% vs 10%; P < .001), vascular invasion (17% vs 3%; P = .01) and lymphatic invasion (15% vs 2%; P = .02). They were also more likely to have multiple nerves involved (P < .001) and to have nerve involvement both within as well as discontiguous from the tumor (P = .002) (Table 1). Tumors with larger-caliber nerve invasion had significantly shorter follow-up time than those with small-caliber invasion (2.7 vs 4.0 years; P = .02), which would tend to underestimate occurrence of poor outcomes in the large-caliber group. However, this difference was adjusted for in Cox models.
Thirty-seven SCCs (24% of cases) occurred in immunocompromised patients (27 with organ transplants, 6 with chronic lymphocytic leukemia, 3 with lymphoma, and 1 with human immunodeficiency virus). Immunosuppression was equally common in large- and small-caliber nerve groups and was not significantly associated with any outcomes of interest on univariate or multivariate modeling.
Treatment of the primary tumor included standard excision (n = 72 cases), standard excision plus radiation (n = 15 cases), standard excision plus lymph node dissection (n = 3 cases), Mohs surgery (n = 15 cases), Mohs surgery plus radiation (n = 2 cases), radiation monotherapy (n = 1 case), electrodessication and curettage (n = 1 case), or other combination therapy (n = 5 cases). Adjuvant radiation was performed in a total of 20 cases (18%) (n = 5 in the small-caliber nerve invasion group, n = 15 in the large-caliber nerve invasion group). Treatment type was not associatedwith differences in outcomes of interest. Of the 17 cases treated with Mohs surgery, 2 had PNI identified during Mohs surgery that had not been identified on initial biopsy. In 1 of these cases, clear margins could not be obtained, and the patient underwent adjuvant radiation. Clear margins were obtained in the other 16 Mohs cases.
Risk factors for outcomes of interest
The overall risks of outcomes of interest were as follows: LR, 16%; NM, 10%; DSD, 7%; and ACD, 46%.Table 2 and Table 3 tabulate the number of events for each outcome by nerve caliber group and by number of non-PNI risk factors present (moderate or poor differentiation, diameter ≥ 2 cm, or invasion beyond subcutaneous fat). Outcomes were excellent in cases in which PNI was the only risk factor (regardless of nerve caliber) with only a single local recurrence in these 29 cases.
On univariate analysis (Table 4), large-caliber nerve invasion was associated with significantly increased risks of NM and DSD (hazard ratio [HR], 5.6 [95% CI, 1.1-27.9, and HR, 4.5 [95% CI, 1.2-17.0], respectively).On multivariate analysis, nerve caliber did not predict any outcome of interest. However, tumor diameter of 2 cm or greater predicted LR (HR, 4.8 [95% CI, 1.8-12.7), presence of multiple risk factors predicted NM (2 factors: HR, 4.1 [95% CI, 1.0-16.6], and 3 factors: HR, 14.1 [95% CI, 3.0-66.7]), and lymphovascular invasion predicted DSD (HR, 15.3 [95% CI, 3.7-62.8]) and ACD (HR, 1.1 [95% CI, 1.0-1.1]). Invasion beyond subcutaneous fat and older age also predicted ACD (HR, 2.1 [95% CI, 1.0-4.3], and HR, 1.1 [95% CI, 1.0-1.1], respectively). The results of the multivariate models are summarized in Table 5.
Perineural invasion of unnamed nerves without other risk factors may have a better prognosis than previously thought. There was only a single local recurrence in this 29-patient subset. Conversely, in cases with other risk factors, a 4- and 14-fold increased risk of nodal metastasis was found in patients with 2 and 3 other risk factors, respectively. Cases with large-caliber nerve invasion were significantly more likely to have other concomitant risk factors, including nerve diameter of 2 cm or greater, invasion beyond the subcutaneous fat or a 4-mm depth, multiple nerve involvement, infiltrative growth, or lymphovascular invasion. Large-caliber nerve invasion was associated with an increased risk of nodal metastasis and death from SCC on univariate analysis. However, on multivariate analysis, other risk factors predominated in predicting poor outcomes, including tumor diameter of 2 cm or greater, presence of more than 1 non-PNI risk factor, lymphovascular invasion, and invasion beyond subcutaneous fat.
Perineural invasion was recently added to the AJCC SCC staging criteria as a high-risk characteristic. However, there are data to suggest that limited nerve involvement may not have an impact on prognosis as adversely as advanced nerve invasion (Table 6).7-11 The findings of this study are consistent with those data, because large-caliber PNI was associated with a 4- to 5-fold increased risk of nodal metastasis and death due to SCC on univariate analysis, indicating that it may be associated with a worse prognosis than small-caliber PNI. Thus, SCC with large-caliber nerve invasion may be considered high-risk, and adjuvant therapy may be considered in such cases. The results further indicate that regardless of nerve caliber, patients without other risk factors (moderate or poor differentiation, diameter of 2 cm or greater, or tumor penetration beyond subcutaneous fat) have a good prognosis. This is an important finding because it suggests that there are factors other than PNI that may be more robust prognostic indicators. Because large-caliber PNI occurs together with the strong prognostic predictors described herein more often than small-caliber nerve invasion, it may be a marker for tumors with a particularly poor prognosis owing to the presence of numerous high-risk features.
This study was limited to patients diagnosed as having SCC with PNI at 2 academic medical centers. It is possible that there are differences between patients seen at academic centers and those seen elsewhere. However, the study population was confined to those with primary tumors. Thus, the tendency for overestimation of poor outcomes at academic hospitals owing to tertiary care of recurrent tumors was minimized. Although to our knowledge this is the largest cohort study of primary SCC with PNI to date, it was powered only to define the strongest prognostic predictors. A larger study might show that additional risk factors significant on univariate analyses, such as caliber of involved nerves and number of nerves involved, are also independent predictors of outcomes in SCC with PNI. Because only 2 cases in this cohort had invasion of named nerves, the prognosis of this type of extensive PNI could not be fully evaluated. Since prognosis has previously been reported to be poor in patients with involvement of named nerves, named nerve invasion should continue to be regarded as having a high risk of poor outcomes, regardless of presence or absence of other risk factors.
Because all patients in this study had PNI, the prognostic importance of PNI as a whole, relative to other risk factors, could not be evaluated. In a larger cohort study of SCC outcomes that contains SCC with and without PNI (Schmults et al15), PNI invasion is an independent prognostic factor in multivariate analysis (Jambusaria-Pahlajani et al13). However, a still-larger, population-based study of SCC outcomes will likely be necessary to establish the prognostic impact of different gradations of PNI (eg, nerve caliber) in multivariate analysis. Based on the univariate data herein showing good outcomes with small nerve invasion and worse outcomes with large nerve invasion, and other studies that have shown a poor prognosis for patients with SCC with large named nerve or otherwise extensive nerve invasion, it is likely that larger studies will bear out in multivariate modeling that extensive nerve invasion is independently associated with worse outcomes, while lesser nerve invasion is not. Since small-caliber nerve invasion without other risk factors had an excellent prognosis in this study, adjuvant therapy in such cases may not be necessary.
In conclusion, SCC invading unnamed nerves of caliber smalller than 0.1 mm without other risk factors may have a good prognosis. However, perineurally invasive SCCs with other risk factors (moderate or poor differentiation, diameter of ≥2 cm, or deep invasion beyond subcutaneous fat) have a high risk of poor outcomes and may be considered for adjuvant therapy. Large-caliber nerve invasion commonly occurs concomitantly with other risk factors and may therefore be a marker of poor prognosis. Larger studies are needed to estimate the specific prognostic impact of large-caliber nerve invasion on SCC outcomes.
Accepted for Publication: June 20, 2012.
Correspondence: Chrysalyne D. Schmults, MD, MSCE, Mohs and Dermatologic Surgery Center, Department of Dermatology, Brigham and Women's Hospital, 1153 Centre St, Ste 4349, Jamaica Plain, MA 02130 (cschmults@partners.org).
Author Contributions: Drs Carter and Schmults had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the analysis. Study concept and design: Carter and Schmults. Acquisition of data: Carter, Johnson, Chua, and Karia. Analysis and interpretation of data: Carter, Karia, and Schmults. Drafting of the manuscript: Carter and Karia. Critical revision of the manuscript for important intellectual content: Carter, Johnson, Chua, Karia, and Schmults. Statistical analysis: Karia and Schmults. Administrative, technical, and material support: Carter and Karia. Study supervision: Carter and Schmults.
Conflict of Interest Disclosures: None reported.
Additional Contributions: We thank Travis Hollmann, MD, PhD, and Victor Neel, MD, PhD.
1.Christenson LJ, Borrowman TA, Vachon CM,
et al. Incidence of basal cell and squamous cell carcinomas in a population younger than 40 years.
JAMA. 2005;294(6):681-69016091570
PubMedGoogle ScholarCrossref 2.Farasat S, Yu SS, Neel VA,
et al. A new American Joint Committee on Cancer staging system for cutaneous squamous cell carcinoma: creation and rationale for inclusion of tumor (T) characteristics.
J Am Acad Dermatol. 2011;64(6):1051-105921255868
PubMedGoogle ScholarCrossref 3.Leibovitch I, Huilgol SC, Selva D, Hill D, Richards S, Paver R. Cutaneous squamous cell carcinoma treated with Mohs micrographic surgery in Australia, II: perineural invasion.
J Am Acad Dermatol. 2005;53(2):261-26616021121
PubMedGoogle ScholarCrossref 5.Goepfert H, Dichtel WJ, Medina JE, Lindberg RD, Luna MD. Perineural invasion in squamous cell skin carcinoma of the head and neck.
Am J Surg. 1984;148(4):542-5476486325
PubMedGoogle ScholarCrossref 6.Clayman GL, Lee JJ, Holsinger FC,
et al. Mortality risk from squamous cell skin cancer.
J Clin Oncol. 2005;23(4):759-76515681519
PubMedGoogle ScholarCrossref 7.Ross AS, Whalen FM, Elenitsas R, Xu X, Troxel AB, Schmults CD. Diameter of involved nerves predicts outcomes in cutaneous squamous cell carcinoma with perineural invasion: an investigator-blinded retrospective cohort study.
Dermatol Surg. 2009;35(12):1859-186619889009
PubMedGoogle ScholarCrossref 8.Lin C, Tripcony L, Keller J,
et al. Perineural infiltration of cutaneous squamous cell carcinoma and basal cell carcinoma without clinical features.
Int J Radiat Oncol Biol Phys. 2012;82(1):334-34021093171
PubMedGoogle ScholarCrossref 9. McCord MW, Mendenhall WM, Parsons JT,
et al. Skin cancer of the head and neck with clinical perineural invasion.
Int J Radiat Oncol Biol Phys. 2000;47(1):89-9310758309
PubMedGoogle ScholarCrossref 10. McCord MW, Mendenhall WM, Parsons JT, Flowers FP. Skin cancer of the head and neck with incidental microscopic perineural invasion.
Int J Radiat Oncol Biol Phys. 1999;43(3):591-59510078643
PubMedGoogle ScholarCrossref 11.Garcia-Serra A, Hinerman RW, Mendenhall WM,
et al. Carcinoma of the skin with perineural invasion.
Head Neck. 2003;25(12):1027-103314648861
PubMedGoogle ScholarCrossref 12.Jambusaria-Pahlajani A, Hess SD, Katz KA, Berg D, Schmults CD. Uncertainty in the perioperative management of high-risk cutaneous squamous cell carcinoma among Mohs surgeons.
Arch Dermatol. 2010;146(11):1225-123121079057
PubMedGoogle ScholarCrossref 13.Jambusaria-Pahlajani A, Kanetsky PA, Karia PS,
et al. Evaluation of AJCC Tumor (T) Staging for Cutaneous Squamous Cell Carcinoma and a Proposed Alternative Tumor Staging System.
JAMA DermIn PressGoogle Scholar 14.Ballantyne AJ. Perineural invasion by SCC.
J Dermatol Surg Oncol. 1984;10(7):502-504
Google Scholar 15.Schmults CD, Karia PS, Carter JB, Han J, Qureshi AA. Factors predictive of recurrence and death in cutaneous squamous cell carcinoma: a 10-year single institution study.
JAMA DermIn PressGoogle Scholar