PII S0360-3016(00)01407-3 CLINICAL INVESTIGATION Skin PERINEURAL SPREAD OF CUTANEOUS SQUAMOUS AND BASAL CELL CARCINOMA: CT AND MR DETECTION AND ITS IMPACT ON PATIENT MANAGEMENT AND PROGNOSIS LORNA SOHN WILLIAMS, M.D.,* ANTHONY A. MANCUSO, M.D.,* AND WILLIAM M. MENDENHALL, M.D.â Departments of *Radiology and â Radiation Oncology, University of Florida, Gainesville, FL Purpose: To analyze the pretreatment imaging findings and outcome of patients with perineural spread of squamous or basal cell carcinoma of the face and scalp treated with radiotherapy, to determine whether CT (computed tomography) or MR (magnetic resonance) imaging can be effectively used to identify patients who would benefit from aggressive treatment, and to characterize the imaging features associated with cure. Methods: Thirty-five patients had perineural spread of squamous and basal cell carcinoma along the divisions of the trigeminal and/or facial nerves based on clinical findings and/or histopathological proof. Perineural extension seen on imaging was divided into three zones of involvement. The volume of perineural disease was graded semiquanitatively. All patients received radiotherapy with curative intent. Results: Eighteen of the 35 patients had imaging evidence of perineural spread of tumor, and the remaining 17 were imaging negative for perineural spread. The absolute 5-year survival of the imaging positive group was 50% compared with 86% in the imaging-negative group (p 5 0.048). Conclusions: Imaging can be used to identify patients with advanced perineural spread who warrant aggressive radiotherapy. Imaging evidence of perineural invasion worsens prognosis; however, low-volume and peripheral perineural disease is radiocurable. Greater perineural tumor volume with more central disease was associated with an unfavorable outcome. © 2001 Elsevier Science Inc. Perineural, Skin cancer, Cranial nerves, MR, CT. INTRODUCTION Perineural spread of head and neck cancer along nerves has been described in the literature for decades (1â4). Ballan- tyne used the term âneurotropismâ to describe the initial spread of the tumor along the loose connective tissue of the perineurium, which can allow a single tumor cell to dissem- inate without inflammation in the surrounding stroma (1). A pathological study by Carter et al. found that cancer cells invade the perineural space and use it as a conduit for spread (5). They also found that the neoplastic cells caused mild axonal and myelin degeneration and segmental infarction of the neural bundle (5). It was theorized that the neoplastic cells compromised the neural blood supply, resulting in hypoxia to the nerve and producing degeneration of the nerve (5). However, the exact biologic mechanism of peri- neural spread is not fully understood (5, 6). The subsequent extension is often contiguous, but skip lesions are known to occur (7, 8). There is often delay in the diagnosis of perineural spread (9). The first symptom frequently is a sensation of ants or worms crawling underneath the skin which will progress to pain, numbness, and/or motor deficits if untreated (9). The early signs of perineural spread can be missed unless the clinician maintains a high index of suspicion when assess- ing patients with recurrent or locally-advanced skin cancer that is located near one or more cranial nerves (9). Detecting early perineural spread may be accomplished by asking leading questions regarding remote history of previous skin cancer, or by performing neurological testing (9). Perineural spread of carcinoma can also be insidious, because patients may be asymptomatic for years before the symptoms of perineural spread manifest clinically (3, 6, 8). Before CT (computed tomography) or MR (magnetic reso- nance) imaging, radiographs or polytomography could de- tect only the most gross disease, and the diagnosis of perineural spread was frequently established only at surgery or by the pathologist (8). Advanced perineural spread has been often associated with a poor prognosis in the vast majority of cases (6, 9). However, it is increasingly being recognized that even in advanced cases of perineural spread, cure is possible (7, 8, 10). Surgeons have observed that Reprint requests to: Dr. L. S. Williams, Assistant Professorâ Neuroradiology Division, Department of Radiology, 1600 Archer Road, PO Box 100374, University of Florida, Gainesville, FL 32610-0374. Tel: (352) 395-0291; E-mail:
[email protected] Accepted for publication 24 August 2000. Int. J. Radiation Oncology Biol. Phys., Vol. 49, No. 4, pp. 1061â1069, 2001 Copyright © 2001 Elsevier Science Inc. Printed in the USA. All rights reserved 0360-3016/01/$âsee front matter 1061 complete cure of perineural spread may be accomplished with complete excision if the lesion, or through aggressive surgery and adjuvant radiation therapy (7, 10). The most common head and neck malignancies to spread perineurally are squamous cell carcinomas of the skin or mucosa and adenoid cystic carcinoma of the salivary glands (11). The reported incidence of perineural spread of skin carcinoma ranges from 4.9% to 14% (6, 8). CT and MR imaging can be used to detect gross perineural spread of head and neck cancers and define its extent (9). The precise role of imaging in radiation treatment plan- ning of perineural disease in cutaneous squamous and basal cell carcinomas has not been well defined in the literature. Furthermore, the exact relationship between the imaging findings of perineural spread and clinical outcome in this patient population had not been well characterized. This study retrospectively evaluates perineural spread of skin carcinoma along the divisions of the trigeminal and/or facial nerves imaged with CT and MR prior to treatment. Its purpose is to determine whether imaging has a decisive role in radiation therapy planning and if particular imaging find- ings predict a favorable outcome. METHODS AND MATERIALS A total of 96 patients with clinical and/or pathologic evidence of perineural spread of basal or squamous cell skin cancer were treated at this institution with radiation therapy and/or surgery from 1986 to 1995. Of these 96 patients, patients who met the following criteria were admitted to the study: (a) a cutaneous carcinoma located on face or scalp; (b) radiation therapy at this institution with curative intent; (c) a minimum 2-years of follow-up to determine disease- free clinical status; (d) pretreatment cross-sectional imaging performed at the time of diagnosis available for review; and (e) histopathological or highly suggestive clinical evidence of perineural spread at the time of diagnosis and treatment. Thirty-five patients met these criteria. Highly suggestive clinical evidence of perineural involvement was established by the treating radiation oncologist and was defined as trigeminal (V) or facial (VII) nerve dysfunction associated with a cutaneous carcinoma located in a corresponding distal site of innervation. Symptoms included pain, pares- thesia, motor function loss, and formication. The imaging findings at the time of diagnosis, the clinical course of these patients, and their ultimate outcome were examined retrospectively. The images were reviewed by two radiologists at different levels of expertise; one had 20 years of experience in head and neck imaging and the other was completing subspecialty training in neuroradiology. The findings were agreed upon by consensus. Previous studies have demonstrated the normal MR imaging appear- ance of the facial nerve and the trigeminal nerve branches with its ganglion (12â16). Radiographic evidence of peri- neural spread was defined as enlargement or abnormal en- hancement of the nerve, obliteration of the normal fat plane surrounding the nerve, and/or erosion or enlargement of its related foramen (12, 13). Abnormal enhancement was de- fined as diffuse enhancement with loss of the distinction been the nerve and its perineural vascular plexus. Obliter- ation of several specific fat planes was particularly impor- tant in assessing distal segments of V1-3 and VII. Obliter- ation of the superior medial orbital fat plane was used as a sign of distal V1 perineural spread; loss of the periantral fat plane was used as an indicator of distal V2 perineural extension; and loss of the fat in the stylomastoid foramen (SMF) was used as sign of perineural involvement of the distal portion of VII (16, 17). The volume of perineural disease was semiquanitatively graded as: minimal, moderate, and gross. Minimal disease was defined as abnormal enhancement without obvious enlargement of the nerve. Moderate disease was designated as nerve enlargement, two to three times greater than the normal mean diameter reported in previous normative stud- ies, with or without abnormal enhancement (16). Gross disease was defined as nerve enlargement, greater than three times the normal mean diameter reported in previous stud- ies, or obvious mass with or without abnormal enhance- ment. Perineural extension seen on imaging was further divided into three zones: zone 1, peripheral; zone 2, central/skull base; and zone 3, cisternal. For all three trigeminal divi- sions, zone 3 was the region from the cistern of the trigem- inal ganglion to brainstem. Specifically, for V1, these zones were designated as: zone 1, superiomedial orbit at the level of the orbital ridge up to superior orbital fissure (SOF); zone 2, SOF up to trigeminal ganglion cistern. The boundary zones for V2 were: zone 1, fat plane of the periantral soft tissues, infraorbital canal, and the pterygopalatine fossa up to foramen rotundum; zone 2, foramen rotundum up to trigeminal ganglion cistern. For V3, they were: zone 1, inferior alveolar/lingual nerve up to foramen ovale; zone 2, foramen ovale up to trigeminal ganglion cistern. For CN VII these zones were: zone 1, facial region, which encompasses the branches within the parotid gland up to stylomastoid foramen (SMF); zone 2, from the SMF through and includ- ing the labyrinthine segment up to the internal auditory canal (IAC); and zone 3, IAC to brainstem. The specific imaging findings of perineural spread for trigeminal branches and facial nerve are summarized in Table 1. Thirty-five patients met the inclusion criteria. Eighteen patients (52%) had clinical and pathologic evidence of perineural spread of squamous or basal cell carcinoma. Of these 18 patients, 13 had imaging evidence of perineural spread and 5 were imaging negative for perineural spread. Eleven (31%) were asymptomatic, but demonstrated peri- neural extension on histopathology. Of these 11 patients, only 2 had imaging evidence of perineural spread of carci- noma. Six patients (17%) had clinical symptoms of peri- neural spread with negative histopathological findings of perineural spread. Of these 6 patients, 3 patients had imag- ing evidence of perineural spread of carcinoma. There were 26 males and 9 females, with a median age of 1062 I. J. Radiation Oncology c Biology c Physics Volume 49, Number 4, 2001 66 years. The median length of clinical follow-up was 4.3 years and the average length was 4.6 years. The average duration of symptoms was 5.9 months. Thirty (86%) had squamous cell carcinoma and 5 (14%) had basal cell carci- noma. Twenty-seven of the 35 patients (78%) were clinical stage T4N0 or N1; 4 patients (11%) were staged T2N0 or N2; and the remaining 4 patients (11%) were T1N0. When imaging or pathologic findings are considered, all patients would be classified as T4 because of perineural invasion. Twenty-six patients (74%) had recurrent skin cancers. Twenty-five patients (72%) were treated with wide local excision and 5 patients had (14%) Mohs surgery. The re- maining 5 patients (14%) had no surgery. One patient re- ceived adjuvant chemotherapy. Imaging negative asymp- tomatic patients (n 5 9) with only histopathological evidence of perineural invasion were treated with surgery and radiotherapy or radiotherapy alone, according to previ- ously described guidelines in McCord et al. (18). In these patients, the radiation fields encompassed the primary site with a 2â3-cm margin without treatment of the skull base. Asymptomatic patients with evidence of extensive patho- logical evidence of perineural invasion, involvement of a named nerve, and/or positive margins received treatment along the nerve to the skull base. Symptomatic as well as asymptomatic patients with ra- diographically-evident perineural disease were treated prox- imally to the central/skull base and cisternal region (zone 2 or 3), 2 cm beyond the most proximal extent of the tumor seen on imaging. These radiation fields often included at least part of the brainstem. Symptomatic patients who were imaging negative were treated with radiation fields that included the entire course of the nerve up to the skull base. Patients were treated twice daily and received an approxi- mate total dose of 75 Gy, at 1.2 Gy per fraction, according to previously described methods in Mendenhall et al. (19). The average radiation dose administered was 68.3 Gy. The typical dose to the brain stem was 70â75 Gy. All patients received radiation therapy with curative intent (Table 2). CT imaging was generally performed with 3-mm contig- uous sections on a GE 9800 with 14â16-cm FOV. One to 1.5-mm contiguous sections were also used for targeted CT imaging of the facial nerve with an FOV as small as 9.6 cm. MR imaging consisted of conventional spin echo and fast spin echo imaging in the axial and coronal planes with 3â4-mm contiguous sections and 16â18-cm fields of view. T1-weighted images were done without and with gadolin- ium administration. These images were acquired on a Sie- mens 1.0 T Magnetom, Siemens 1.5 T Visions, or GE 1.5 T Signa unit. In total, 10 MR and 25 CT studies, obtained at the time of the patientâs initial diagnosis and treatment, were reviewed. The outcomes were analyzed by determining the rates of absolute survival, cause-specific survival, and local control that were calculated actuarially using the Kaplan- Meier product-limit method (20, 21). The significance of the levels between curves was calculated using the log-rank test (21, 22). RESULTS Eighteen of the 35 patients admitted to the study had radiographical evidence of perineural spread of cancer at the time of diagnosis. Seventeen of the 35 had no radiographical evidence of perineural spread of carcinoma in spite of clinical symptoms and/or histopathological evidence of per- ineural invasion at the time of diagnosis. In the imaging- positive group, the distribution of nerves involved was as follows: V1, 5 patients; V2, 14 patients; V3, 3 patients; CN VII, 3 patients. Two patients had more than one nerve involved. The 2-year absolute survival rate between the imaging- Table 1. Imaging findings of perineural spread involving V1âV3 and VII Nerve Zone Distal Central Cisternal V1 Obliteration of the fat plane in the superior medial orbit. N/A Enhancement and/or enlargement in the trigeminal cistern. Brainstem edema at the root entry zone. V2 Obliteration of the fat plane beneath the muscles of facial expression. Enhancement and enlargement in foramen rotundum. Enhancement and/or enlargement in the trigeminal cistern. Enhancement and/or enlargement in the infraorbital canal. Brainstem edema at the root entry zone. Obliteration of the fat plane of the pterygopalatine fossa. V3 Obliteration of the fat plane of the masticator space. Enhancement and/or enlargement in foramen ovale. Enhancement and/or enlargement in the trigeminal cistern. Brainstem edema at the root entry zone. VII Enhancement and enlargement. Enhancement and/or enlargement including the labyrinthine segment of the facial nerve Enhancement and/or enlargement in the IAC. Obliteration of the fat plane of the stylomastoid foramen. Brainstem edema at the root entry zone. 1063Perineural spread of cutaneous carcinomas: CT and MR c L. S. WILLIAMS et al. positive and imaging-negative groups was similar; however, a significant decrease in local control rate and the cause- specific and absolute survival rates was observed at 5 years for the imaging positive subset compared to the imaging negative group (Table 3). Eight of the 18 (44%) imaging-positive patients had clinical progression of their disease. Two of the 17 (12%) imaging-negative patients had clinical progression of their disease, and a significant difference between the two groups was suggested ( p 5 0.060; Fisherâs exact test, two-tail). Table 2. Summary of imaging findings, treatment, and outcome Patient Age Sex Histology Evidence of PNS Imaging Symptoms Nerve(s) involved Volume Zone Surgical treatment RT dose (Gy) Clinical status Years of follow-up 1 73 M S CP Positive Sx V2 1 1 WLE 74.4 NED 9.2 2 42 F B CP Negative Sx V1 0 0 MCE 67.5 NED 1.9 3 33 M S P Negative Asx V3 0 0 MCE 65.0 NED 5.8 4 66 M S CP Positive Sx V2 1 1 WLE 76.8 NED 2.9 5 68 F S CP Positive Sx V2 1 1 MCE 76.8 NED 8.0 6 78 M S P Negative Asx V3 0 0 WLE 60.0 NED 5.1 7 51 M B CP Positive Sx V2 1 1 WLE 70.4 NED 2.1 8 69 F S C Negative Sx V1 0 0 WLE 70.4 NED 4.3 9 41 M S C Positive Sx V1, V2 3 1,2,3 None 79.2 NED 3.4 10 60 M S P Negative Asx V3 0 0 WLE 60.0 NED 6.5 11 46 F B P Negative Asx V2 0 0 WLE 58.8 NED 6.3 12 60 M S CP Negative Sx VII 0 0 WLE 70.0 NED 11.3 13 67 F S CP Positive Sx V2 1 1 MCE 74.4 NED 6.6 14 76 F B C Negative Sx V1 0 0 WLE 58.5 NED 6.3 15 73 M S CP Positive Sx V2 1 2 WLE 60.8 NED 4.3 16 80 M S CP Positive Sx VII 3 1 WLE 70.2 NED 2.4 17 56 M S P Negative Asx V3 0 0 WLE 70.2 NED 5.5 18 65 M S P Negative Asx V3 0 0 None 44.8 NED 6.0 19 74 M S P Negative Asx V1 0 0 WLE 58.8 NED 3.5 20 37 M S CP Negative Sx V3 0 0 WLE 70.0 NED 8.8 21 79 M S CP Positive Sx V2 1 1 MCE 69.6 NED 5.5 22 45 M S C Negative Sx V2 0 0 None 76.5 NED 9.0 23 79 F B P Negative Asx V3 0 0 WLE 64.7 NED 2.0 24 75 M S CP Positive Sx V2 2 1,2,3 WLE 70.0 DP 2.5 25 27 M S CP Negative Sx V2 0 0 WLE 59.3 DP 3.3 26 72 M S P Negative Asx V2 0 0 WLE 76.8 DP 1.1 27 41 M S CP Positive Sx V2 1 1 WLE 76.8 DP 1.0 28 66 M S P Positive Asx VII 3 1 MCE 70.0 DP 4.3 29 55 M S CP Positive Sx V3 3 3 None 70.4 DP 0.05 30 69 M B C Positive Sx V1â3,VII 1 1,2,3 WLE 71.4 DP 4.1 31 65 F S CP Positive Sx V1 1 1 WLE 70.0 DP 3.6 32 57 M S CP Positive Sx V2 2 2 WLE 70.0 DP 8.7 33 76 M S P Positive Asx V3 3 1 WLE 70.0 DP 1.3 34 67 M S CP Negative Sx V3 0 0 WLE 60.0 DP 2.9 35 91 F S C Positive Sx V2 1 1 None 76.5 DP 1.9 Abbreviations: S 5 squamous; B 5 basal carcinoma; C 5 only clinical evidence of PNS (perineural spread); P 5 only pathological evidence of PNS; CP 5 both clinical and pathological evidence of PNS; Asx 5 asymptomatic; Sx 5 symptomatic for PNS; WLE 5 wide local excision; MCE 5 microscopically-controlled excision; NED 5 no evidence of disease; DP 5 disease progression. Table 3. Two- and 5-year survival rates based on whether the patient was imaging positive or negative at the time of diagnosis and treatment Survival method of analysis Imaging positive n 5 18 (%) Imaging negative n 5 17 (%) Local control 2-year survival 78 88 *5-year survival 35 78 Cause-specific 2-year survival 83 94 â R5-year survival 54 94 Absolute 2-year survival 83 94 â¡5-year survival 50 86 * p-value 5 0.047. â p-value 5 0.031. â¡ p-value 5 0.049. 1064 I. J. Radiation Oncology c Biology c Physics Volume 49, Number 4, 2001 In the imaging-positive group of 18 patients, a difference in clinical outcome based on perineural disease volume and extent was suggested ( p 5 0.054, Fisherâs exact test, two-tail), (Fig. 1). Ten patients had minimal or moderate perineural disease confined to zone 1 or 2 (Figs. 2â4). Two of these 10 (20%) with minimal volume and peripheral disease had clinical progression of their perineural spread. The remaining 8 imaging-positive patients had a greater perineural tumor volume and/or more central perineural spread (Figure 5â8). Seven patients in this group had mod- erate or gross disease with perineural extension to zone 1, 2, or 3, and 1 patient had minimal disease to zone 3. In these 8 patients with radiographically evident advanced disease, 6 of 8 (75%) had clinical progression of disease. Patients who had perineural spread symptoms at the time of diagnosis of were more likely to be imaging positive for perineural disease ( p 5 0.008, Chi-square test). However, no corre- Fig. 1. Disease progression is compared to both the volume of disease and the zone extent of disease. The volume of disease is graded as gross (GRO), moderate (MOD), and minimal (MIN). The eight patients above the dotted line had a greater degree and extent of perineural spread. Six of these eight had progression of their disease. The ten patients below the dotted line had a lesser degree and extent of perineural spread. Two of these ten had progression of their disease. The difference between the two was marginally significant, p 5 0.054 (Fisherâs exact test, two-tail). Fig. 2. Minimal, peripheral-zone 1 perineural disease of V2 at the level of the infraorbital canal. MR T1WI (weighted image) post contrast (TR 5 600 TE 5 15). The infraorbital nerve on the left (white arrow) is normal size but abnormally enhancing. Infiltration of the fat plane anterior to the maxillary antrum and beneath the muscles of facial expression on the left is also seen as a sign of perineural invasion (black arrow). The infraorbital nerve on the right (white arrowhead) has a normal MR appearance. Fig. 3. Minimal, central-zone 2 perineural disease of V2 at the level of the foramen rotundum. MR T1WI post contrast (TR 5 600, TE 5 15). V2 on the left is normal size and equal in diameter to V2 on the right, but is abnormally and homogeneously enhanc- ing (black arrow), depicting a minimal volume of perineural dis- ease. V2 on the right has a normal MR appearance consisting of a central hypointense structure (black arrowhead), representing V2 within foramen rotundum surrounded by its normally enhancing circumferential perineural vascular plexus (white arrow). 1065Perineural spread of cutaneous carcinomas: CT and MR c L. S. WILLIAMS et al. lation was seen between symptoms of perineural spread and outcome ( p 5 1.0, Fisherâs exact test, two-tail). A sum- mary of the patient data with regard to symptoms, imaging findings, and their outcomes is shown in Table 4. CONCLUSION Several treatment modalities are available for the treat- ment of squamous and basal cell carcinoma in addition to radiation therapy, including various surgical methods such as curettage with electrodessication and microscopically- controlled excision (Mohs technique) (9). The selection of a treatment modality depends on the potential for cure, pres- ervation of function, and cosmesis (9). Radiotherapy plays a significant role in the treatment of perineural disease of the head and neck, particularly in those patients whose perineu- ral involvement is either extensive, and thus not amenable to surgery, or involves the proximal portions of the cranial nerves (23). Untreated or treatment-resistant perineural disease of the head and neck region results in steady deterioration and death. Radiotherapy has been used to successfully cure some cases of perineural disease in the head and neck region. Outcome analysis in previous studies examining head and neck perineural disease were hampered by small numbers of less than 20 patients or heterogeneous patient populations comprised of various primary carcinomas (2, 8, 10). The larger studies of more than 30 patients that exam- ined perineural invasion of cutaneous squamous and basal cell carcinoma predated contemporary high-resolution cross-sectional imaging capability (1, 6, 7). The large pop- ulation of cutaneous squamous and basal cell carcinoma patients and the availability of modern cross-sectional im- aging techniques permitted imaging finding and clinical outcome comparisons among patients in this study. Perineural invasion is observed in a minority (4.9â14%) Fig. 4. Moderate, central-zone 2 perineural spread involving V2 to the level of the foramen rotundum. MR T1WI post contrast (TR 5 600, TE 5 20). V2 on the left is enlarged approximately two times normal and abnormally enhancing (black arrow), demonstrating a moderate volume of perineural disease within the foramen rotun- dum. The cortical margins of the enlarged foramen rotundum on the left are seen as a markedly hypointense structure along its lateral border (white arrowheads), partially encircling the enlarged and enhancing nerve. V2 on the right is normal size. The normal nerve on the right (black arrowhead), is hypointense centrally and is incompletely surrounded by an enhancing normal perineural vascular plexus (white arrow), seen along the superior margin of its foramen rotundum. Fig. 5. Minimal, cisternal-zone 3, antegrade (toward the central nervous system) perineural spread from V2. MR T1WI post con- trast (TR 5 600, TE 5 15). The normal trigeminal ganglion on the right (black arrowhead) is seen as a non-enhancing crescentic structure located posterior and adjacent to border of the foramen ovale. Note the normal appearance of the trigeminal cistern on the right (white arrow), which is hypointense on T1WI. This is in contrast to the markedly abnormal trigeminal cistern in Fig. 6. The half-moon-shaped trigeminal ganglion on the left (black arrow) is normal sized but abnormally enhancing when compared with the normal appearing ganglion on the right. Fig. 6. Gross, cisternal-zone 3 perineural spread antegrade from V2. MR T1WI post contrast (TR 5 600, TE 5 20). A mass is seen filling the cistern of the trigeminal ganglion on the left (black arrow). The trigeminal cistern on the left is mildly expanded, and the normally non-enhancing, fluid-filled space is filled in with tumor and abnormally enhancing. The trigeminal cistern on the right is not seen in the same plane of section due to head rotation. On the right, a normal appearing V3 is seen exiting through foramen ovale (black arrowhead). V3 is seen on the uninvolved right side as a central, linear hypointense structure surrounded by an enhancing vascular plexus that lies along the medial and lateral borders of the nerve (white arrows). 1066 I. J. Radiation Oncology c Biology c Physics Volume 49, Number 4, 2001 of patients with cutaneous squamous and basal cell carci- noma and advanced perineural spread has been associated with a uniformly poor prognosis (6, 8, 9). This study dem- onstrates that certain patients with perineural spread of squamous and basal cell skin cancer are radiocurable if the perineural disease is recognized and appropriately treated. This work also demonstrates the decisive use of imaging in the treatment plan. Furthermore, to our knowledge, no other study has systematically examined the CT and MR imaging of perineural spread of cutaneous squamous and basal cell carcinoma to the cranial nerves with regard to patient out- come. This study attempts to determine if imaging findings of perineural spread have prognostic significance. It also intends to characterize the specific imaging features of perineural spread that may be used to identify those patients at high risk of treatment failure and to describe which specific imaging features may be used as indicators of clinical outcome. The clinical course of the imaging-negative patients com- pared to the imaging-positive patients differed significantly. The less favorable clinical course of the imaging-positive patients compared to the imaging-negative patients was demonstrated by a decreased absolute 5-year survival rate in those imaging-positive patients. The cause-specific and ab- solute survival rates were similar because two patients died of intercurrent disease. These data suggest that imaging findings of perineural spread of cutaneous squamous and basal cell carcinoma may be a risk factor of treatment failure. The imaging-positive group likely has more ad- vanced disease which simply implies a greater risk of treat- ment failure. MR and CT imaging is capable of detecting gross pathologic abnormalities and is expected to be insen- sitive to microscopic perineural disease. These observations may be anticipated; however, these data have important clinical value in the informed consent process allowing patients to be given reasonable odds for cure when accept- ing treatment-related morbidity. The imaging findings of perineural invasion were evalu- ated semi-quantitatively and subclassified according to sub- categories of tumor volume and of tumor extent in order to analyze the clinical outcome as a function of the imaging findings (Fig. 1). The improved outcome in those patients with minimal and peripheral perineural disease is likely related to the relative low volume of tumor and the periph- eral location. The greater treatment failures in advanced disease suggest that better treatment strategies are needed to manage high volume and/or proximal perineural spread of Fig. 7. Gross, cisternal-zone 3 perineural invasion to the level of the root entry zone of V and brainstem. (a) MR T1WI post contrast (TR 5 600, TE 5 20). A mass with abnormal enhancement is seen at the root entry zone of the trigeminal nerve on the left (white arrow). (b) MR T2 WI (TR 5 4000, TE 5 90). In this same patient, brain stem edema is seen on the left (black arrow) at the level of the trigeminal nerve nucleus. Fig. 8. Gross, peripheral-zone 1 perineural disease of VII. Con- trast-enhanced CT. The zygomatic branches of the facial nerve (white arrowheads) are grossly involved with tumor that spreads along the main trunk of the facial nerve on the left (white arrow). 1067Perineural spread of cutaneous carcinomas: CT and MR c L. S. WILLIAMS et al. squamous and basal cell carcinoma. As the number of patients in this database increases, future studies may be able to determine with statistical relevance the probability of survival based on the perineural tumor volume and extent of perineural disease seen on CT and MR imaging. Three patients had gross disease that was confined pe- ripherally to zone 1. Perineural disease progression oc- curred in two of these patients despite the peripheral loca- tion of the disease. This observation would suggest that tumor volume might be a more important prognostic factor than the extent of tumor spread. However, one patient with gross disease extending to the root entry zone of the trigem- inal nerve (zone 3) was cured (disease-free for 4 years and 6 months). Conversely, two patients with minimal periph- eral disease had progression of their perineural disease. These unanticipated results indicate that clinical outcomes could not be predicted in all cases solely based on tumor volume and perineural disease extent present on imaging. The degree of biological aggressiveness of the tumor may account for both the unanticipated disease progression in those two patients with minimal peripheral disease, and for the cure in that patient with gross and extensive disease. The imaging-negative patients who were symptomatic for perineural spread received radiotherapy to the nerve path- ways up to the skull base, in a fashion similar to those patients who were imaging positive. The presence of clini- cal symptoms of perineural involvement in the absence of imaging findings of perineural spread suggests microscopic perineural invasion. The decision to select these imaging negative, but symptomatic patients for aggressive radiother- apy, was based on previous data showing clinical symptoms of perineural spread to be associated with a significantly worse prognosis (1, 6, 19). This group was comprised of eight patients, seven treated with surgery and radiotherapy and one with radiotherapy alone. Only one treatment failure occurred in this group of eight, suggesting this aggressive radiotherapy strategy was justified in these patients who are at high risk for recurrence because of their symptoms of perineural invasion (1, 19). Although previous studies have reported patients with symptoms of perineural spread to have a significantly worse prognosis when compared with asymptomatic patients (1, 6, 19), no correlation between symptoms and a worse progno- sis was seen in this patient group of 35. This may be due to the small sample size, because these 35 patients represented a subset of a larger group of 96 patients. In the original 96 patients, symptoms of perineural spread were an indicator of worse prognosis (23). MR is the preferred imaging method for the evaluation of head and neck perineural tumor spread because of its supe- rior tissue contrast and multiplanar capabilities. In this study, a greater number of CT scans than MR exams were used for evaluation, because 22/35 (65%) of these patients were diagnosed and treated during or before 1992. MR imaging became the exam of choice for evaluating perineu- ral extension at this institution in approximately 1993. Given the outcome results in this patient population, an MR exam is recommended to assess the full extent of perineural disease for all cutaneous skin and basal cell carcinoma patients having symptoms suggesting perineural spread with a cutaneous lesion located in the corresponding cranial nerve distribution, and for those patients who had histopathological evidence of perineural invasion in prox- imity to a cranial nerve. Imaging would also be recom- mended in cases of locally-advanced or deeply infiltrating recurrent squamous or basal cell carcinoma, as the incidence of clinically occult perineural spread is increased (9). REFERENCES 1. Ballantyne AJ, McCarten AB, Ibanez ML. The extension of cancer of the head and neck through peripheral nerves. Am J Surg 1963;106:651â667. 2. Dodd GD, Dolan PA, Ballantyne AJ, et al. The dissemination of tumors of the head and neck via the cranial nerves. Radiol Clin North Am 1970;8:445â461. 3. Curtin HD, Williams R, Johnson J. CT of perineural tumor extension: Pterygopalatine fossa. AJNR 1984;5:731â737. 4. Laine FJ, Braun IF, Jensen ME, et al. Perineural tumor ex- tension through the foramen ovale: Evaluation with MR im- aging. Radiology 1990;174:65â71. 5. Carter RL, Foster CS, Dinsdale EA, et al. Perineural spread by squamous carcinomas of the head and neck: A morphological study using antiaxonal and antimyelin monoclonal antibodies. J Clin Pathol 1983;36:269â275. 6. Goepfert H, Dichtel WJ, Medina JE, et al. Perineural invasion in squamous cell skin carcinoma of the head and neck. Am J Surg 1984;148:542â547. Table 4. Clinical symptoms of perineural spread compared to imaging findings and clinical outcome of all 35 patients Symptoms of perineural spread Imaging findings of perineural spread Clinical Status *Disease-free n 5 25 Disease progression n 5 10 Symptomatic (n 5 24) Positive n 5 16 10 6 Negative n 5 8 7 1 Asymptomatic (n 5 11) Positive n 5 2 0 2 Negative n 5 9 8 1 * Two patients died of intercurrent disease but had no evidence of disease greater than 2 years of follow-up. 1068 I. J. Radiation Oncology c Biology c Physics Volume 49, Number 4, 2001 7. Lawrence N, Cottel WI. Squamous cell carcinoma of skin with perineural invasion. J Am Acad Dermatol 1994;31:30â33. 8. Cottel WI. Perineural invasion by squamous-cell carcinoma. J Dermatol Surg Oncol 1982;8:589â600. 9. Mendenhall WM, Million RR, Mancuso AA, et al. Carcinoma of the skin. In: Million RR, Cassisi NJ, editors. Management of head and neck cancer: a multidisciplinary approach. Phil- adelphia: Lippincott Co; 1994. p. 643â691. 10. Ampil FL, Hardin JC, Peskind SP, et al. Perineural invasion in skin cancer of the head and neck: A review of nine cases. J Oral Maxillofac Surg 1995;53:34â38. 11. Ginsberg LE, DeMonte F. Imaging of perineural tumor spread from palatal carcinoma. AJNR 1998;19:1417â1422. 12. Gebarski SS, Telian SA, Niparko JK. Enhancement along the normal facial nerve in the facial canal: MR imaging and anatomic correlation. Radiology 1992;183:391â394. 13. Curtin HD, Wolfe P, Snyderman N. The facial nerve between the stylomastoid foramen and the parotid: Computed tomo- graphic imaging. Radiology 1983;149:165â169. 14. Daniels DL, Pech P, Pojunas KW, et al. Trigeminal nerve: Anatomic correlation with MR imaging. Radiology 1986;159: 577â583. 15. Runistein D, Stears RL, Stears JC. Trigeminal nerve and ganglion in the Meckel cave: Appearance at CT and MR imaging. Radiology 1994;193:155â159. 16. Sohn Williams L, Mancuso AA, Schmalfuss I. MR of the trigeminal nerve and its perineural vascular plexus: Normal appearance and variants. Presented at Radiological Society of North America Scientific Meeting, Chicago, IL, November 29-December 4, 1998. 17. Silverman CS, Mancuso AA. Periantral soft-tissue infiltra- tion and its relevance to the early detection of invasive fungal sinusitis: CT and MR findings. AJNR 1998;19:321â 325. 18. McCord MW, Mendenhall WM, Parsons J, et al. Skin cancer of the head and neck with incidental microscopic perineural invasion. Int J Radiat Oncol Biol Phys 1999;43:591â595. 19. Mendenhall WM, Parsons J, Price Mendenhall N, et al. Car- cinoma of the skin of the head and neck with perineural invasion. Head Neck 1989;11:301â308. 20. Kaplan EL, Meier P. Nonparametric estimation from incom- plete observations. J Am Stat Assoc 1958;53:457â481. 21. SAS Institute Inc. SAS Technical report P-179: Additional SAS/STAT procedures. Release 6.03 ed., Cary, NC: SAS Institute Inc; p. 49â89. 22. Lawless JF. Statistical models and methods for lifetime data. New York: Wiley; 1982. p. 420â422. 23. McCord MW. Perineural invasion by skin cancer of the head and neck. Presented at the 27th Annual Radiation Oncology Clinical Research Seminar, Gainesville, FL, February 2â8, 1997. 1069Perineural spread of cutaneous carcinomas: CT and MR c L. S. WILLIAMS et al.