Impact of radiographic findings on prognosis for skin carcinoma with clinical perineural invasion




The objective of the current study was to correlate pretreatment computed tomography and magnetic resonance imaging studies with outcomes for patients with squamous or basal cell carcinoma of the skin and clinical perineural invasion.


Between 1986 and 2002, 45 patients were treated with radiotherapy alone (21 patients) or combined with surgery (24 patients), and 4 patients received concomitant chemotherapy. Follow-up ranged from 0.85 years to 17.4 years (median, 3.8 years). Patients were stratified as follows: imaging negative, 10 patients; minimal or moderate peripheral disease, 14 patients; and central and/or macroscopic disease, 21 patients.


The 5-year local control rates were as follows: imaging negative, 76%; minimal or moderate peripheral disease, 57%; and central and/or macroscopic disease, 25%. The 5-year absolute and cause-specific survival rates were as follows: imaging negative, 90% and 100%, respectively; minimal or moderate peripheral disease, 50% and 56%, respectively; and central and/or macroscopic disease, 58% and 61%, respectively.


Patients who had symptomatic but imaging-negative perineural invasion had a relatively good prognosis after receiving definitive radiotherapy alone or combined with surgery. Patients who had imaging-positive minimal or moderate peripheral disease had a better local control rate but a similar survival rate compared with patients who had central and/or macroscopic disease. Cancer 2005. © 2005 American Cancer Society.

Perineural invasion is a relatively rare phenomenon and is reportedly observed in < 5% of patients with skin carcinoma.1 It is associated with a poor prognosis. Patients can be stratified into two subsets: 1) incidental, which is detected pathologically on biopsy or excision in an asymptomatic patient; and 2) clinical, in which the nerve invasion has produced symptoms. Approximately 30–40% of patients have clinical evidence of nerve invasion, and the remainder have incidental perineural invasion.2 Symptoms include formication (the sensation of worms or ants crawling underneath the skin), pain, numbness, and/or motor deficits.2 The most common nerves affected are the second division of the fifth cranial nerve and the seventh cranial nerve. Perineural spread is more common with squamous cell carcinoma than basal cell carcinoma; the presenting symptoms often are subtle, and the physician who fails to ask a leading question or to perform a thorough neurologic examination potentially can miss the diagnosis. Detecting perineural invasion requires a high index of suspicion.3 Factors associated with perineural spread include male gender, tumor size > 2 cm, location on the midface, and previous treatment of the lesion.4–6 Tumor spread usually is antegrade (toward the central nervous system), but retrograde spread also may occur.7 Achieving a cure while preserving both function and cosmesis requires the cooperation of a multidisciplinary team. Treatment is made even more difficult because perineural invasion often is associated with “skip” lesions along the nerve so that, even if resection with negative margins is accomplished, there is a significant risk of recurrence.8

Prior to the advent of computed tomography (CT) and magnetic resonance imaging (MRI), radiographs could detect only relatively extensive disease. MRI is particularly useful and, when fused with treatment-planning CT, optimally defines the extent of macroscopic disease.9 The objective of the current analysis was to correlate CT and MRI findings with prognosis for patients with clinical perineural invasion of cutaneous head and neck carcinomas who have been treated with radiotherapy (RT).


Between June 1986 and May 2002, 45 patients with squamous or basal cell carcinoma of the skin of the head and neck with clinical perineural invasion were treated with RT alone or combined with surgery with curative intent at the University of Florida. Patients ranged in age from 36 years to > 89 years, (median, 65 years). All patients had pretreatment cross-sectional imaging studies that were available for review. Over the period of the study, MRI has been preferred to evaluate the extent of perineural invasion, and CT has been used to assess the regional lymph nodes. Follow-up ranged from 0.85 years to 17.4 years (median, 3.8 years). Minimum follow-up in living patients was 2 years. All pretreatment radiographic studies were reviewed by one radiologist (A.A.M.).

The population is described in Table 1. The primary tumor size was ≤ 2.0 cm in 6 patients, 2.1–5.0 cm in 12 patients, and > 5.0 cm in 6 patients; the size was not specified in 21 patients. Patients in the latter group had enlargement of one or more cranial nerves without a well defined primary tumor. The regional lymph nodes were staged according to the American Joint Committee on Cancer (AJCC) staging system for head and neck carcinoma.10 No patient had distant metastases. In addition to perineural invasion, 7 patients (16%) had radiographic evidence of bone invasion, and 5 patients (11%) had cartilage invasion.

Table 1. Patient Population (n = 45 patients)
VariableNo. of patients (%)
  1. NOS: not otherwise specified.

 Male35 (78)
 Female10 (22)
 Squamous cell carcinoma39 (87)
 Basal cell carcinoma4 (9)
 Metatypical basal cell carcinoma2 (4)
 Well differentiated4 (8)
 Moderately differentiated8 (18)
 Poorly differentiated16 (36)
 NOS17 (38)
Prior treatment 
 None12 (27)
 One surgery19 (42)
 Multiple surgeries14 (31)
Cranial nerves involved: Pretreatment symptoms 
 V18 (18)
 V225 (56)
 V31 (2)
 VII11 (24)
Cervical lymph node status 
 N041 (91)
 N13 (7)
 N2b1 (2)

Forty-four patients had physical signs and symptoms of perineural invasion. The presenting signs and symptoms were as follows: formication in 5 patients (11%), pain in 21 patients (48%), paresthesia/anesthesia in 31 patients (70%), and/or motor deficits in 15 patients (34%). One asymptomatic patient had perineural tumor that was apparent on radiographic evaluation.

Radiographic evidence of perineural spread was defined as an enlargement or abnormal enhancement of the nerve, obliteration of the normal fat plane surrounding the nerve, and/or erosion or enlargement of its related foramen. Abnormal enhancement was defined as diffuse enhancement with loss of the distinction between the nerve and the perineural vascular plexus. Several specific fat planes were assessed to evaluate the distal segments of V1–V3 and VII. Obliteration of the superior medial orbital fat plane was indicative of distal V1 perineural spread, loss of the periantral fat plane was indicative of distal V2 perineural extension, and loss of the fat in the stylomastoid foramen (SMF) was indicative of perineural involvement of the distal portion of VII.9

The volume of perineural disease was graded semiquantitatively as minimal, moderate, and macroscopic. Minimal disease was defined as abnormal enhancement without obvious enlargement of the nerve. Moderate disease was designated as nerve enlargement two to three times higher than the normal mean greatest dimension reported in previous normative studies, with or without abnormal enhancement.11 Macroscopic disease was defined as nerve enlargement greater than three times the normal mean greatest dimension or an obvious mass with or without abnormal enhancement.

Radiographic perineural extension was stratified into three zones: Zone 1, peripheral; Zone 2, central/skull base; and Zone 3, cisternal. For all three trigeminal divisions, Zone 3 was the region from the cistern of the trigeminal ganglion to the brainstem. For V1, the zones were: Zone 1, superiomedial orbit at the level of the orbital ridge up to the superior orbital fissure (SOF); and Zone 2, SOF up to trigeminal ganglion cistern. For the V2 zones, the zones were: Zone 1, fat plane of the periantral soft tissues, infraorbital canal, and the pterygopalatine fossa up to the foramen rotundum; and Zone 2, foramen rotundum to the trigeminal ganglion cistern. For V3, the zones were: Zone 1, inferior alveolar/lingual nerve to the foramen ovale, and Zone 2, foramen ovale to the trigeminal ganglion cistern. For cranial nerve VII, the zones were: Zone 1, facial region, which encompasses the branches within the parotid gland to the stylomastoid foramen (SMF); Zone 2, from the SMF through and including the labyrinthine segment to the internal auditory canal (IAC); and Zone 3, IAC to brainstem.9

On initial presentation, 43 patients (96%) had a single cranial nerve that was involved by tumor, 1 patient (2%) had 2 nerves involved, and 1 patient (2%) was asymptomatic (Table 1). Subsequent to the initial presentation, 11 patients (24%) experienced symptoms caused by invasion of an additional nerve, and 10 patients (22%) had symptoms of invasion of multiple other nerves.

Pretreatment CT and/or MRI findings are shown in Table 2. Ten patients had clinical symptoms of perineural invasion but no radiographic evidence of disease.

Table 2. Pretreatment Computed Tomography and/or Magnetic Resonance Findings
ParameterNo. of patients (%)
  • a

    Some patients had 2 or more nerves involved.

No. of cranial nerves involved 
 None10 (22)
 119 (42)
 ≥ 216 (36)
Pretreatment cranial nerves involveda 
 V15 (11)
 V222 (49)
 V312 (27)
 VII11 (24)
Extent of nerve invasion 
 None10 (22)
 Minimal or moderate peripheral14 (31)
 Central and/or macroscopic disease21 (47)

Twenty-three patients underwent surgery and received postoperative RT, and 1 patient received preoperative RT and underwent surgery. Twenty-one patients received definitive RT alone (17 patients) or combined with concomitant chemotherapy (4 patients). Patients treated with RT alone or combined with chemotherapy had more advanced, incompletely resectable tumors, usually because of tumor extension to the skull base. Patients who were treated with concomitant chemotherapy were treated with the most recent years of the study based on the efficacy of chemoradiation in mucosal head and neck malignancies.

Thirteen patients received once-daily RT, and 32 patients received twice-daily treatment. Doses ranged from 55.80 grays (Gy) to 79.20 Gy minimum tumor dose (MTD) (median, 72.60 Gy MTD). Treatment techniques have been described previously.2, 12 The philosophy at our institution has been to treat the entire tumor to the MTD, even if a radiosensitive structure, such as the brainstem or optic chiasm, is within the target volume.

All statistical computations were performed using SAS software.13 Estimates of local control, and absolute, and cause-specific survival were attained using the Kaplan–Meier product-limit method.14 The log-rank test statistic was generated to test for significant differences in these outcomes across levels of selected covariates.


Patients were stratified as follows: 1) clinically positive, imaging negative (n = 10 patients); 2) minimal or moderate peripheral disease (n = 14 patients); and 3) central and/or macroscopic disease (n = 21 patients). Table 3 details the 5-year rates of local control and survival stratified by increasing radiographic extent of perineural invasion. The radiographic extent of perineural invasion also was recategorized as none versus observed; absolute and cause-specific survival both were significant (P = 0.0373 and P = 0.0102, respectively), whereas local control was not significant (P = 0.1011).

Table 3. Treatment Outcomes
EventEstimate of freedom from event at 5 yrs (%)P value
Imaging negative (n = 10 patients)Minimal or moderate peripheral disease (n = 14 patients)Central and/or macroscopic disease (n = 21 patients)
Local control7657250.2027
Cause-specific survival10056610.0206
Absolute survival9050580.0817


Previous studies from this institution have demonstrated a survival rate of approximately 50% for patients with clinical perineural invasion.3, 6 Our data demonstrate that, with aggressive treatment, even patients with extensive disease can be cured.

The current data revealed no significant survival difference between patients who had minimal or moderate peripheral disease and patients who had central and/or macroscopic disease. This suggests that outcomes cannot be predicted solely on the basis of tumor volume and extent as determined by pretreatment imaging. Thus, even patients with perineural invasion extending to the central skull base and brainstem should be treated aggressively.

We currently use MRI to assess the extent of the perineural invasion and CT to evaluate the regional lymph nodes and potential sites of bone invasion. Patients who have tumors that appear to be resectable undergo surgery and postoperative RT. Patients with incompletely resectable disease are treated with definitive RT and concomitant weekly cisplatin at a dose of 30 mg/M2. Hyperfractionated RT (74.4 Gy MTD in 62 twice-daily fractions) usually is employed to reduce the risk of damage to the optic nerves and retina; the regional lymph nodes are irradiated electively.