3T MRI evaluation of large nerve perineural spread of head and neck cancers

Journal of Medical Imaging and Radiation Oncology 59 (2015) 578–585 MEDIC AL I MAG I N G —O R I G I N A L A RTICLE bs_bs_banner 3T MRI evaluation of large nerve perineural spread of head and neck cancers Justin Baulch,1,2 Mitesh Gandhi,1,3 Jennifer Sommerville1 and Ben Panizza4 1 2 3 4 Department of Radiology, Princess Alexandra Hospital, Brisbane, Queensland, Australia QScan Radiology Clinics, Brisbane, Queensland, Australia Queensland X-Ray, Brisbane, Queensland, Australia Queensland Skull Base Unit, Princess Alexandra Hospital and the School of Medicine, University of Queensland, Brisbane, Queensland, Australia J Baulch MBBS; M Gandhi FRANZCR; J Sommerville FRANZCR; B Panizza MBA, FRACS. Correspondence Dr Justin Baulch, Department of Radiology, Princess Alexandra Hospital, 199 Ipswich Rd, Brisbane, Woolloongabba, Ql 4102, Australia. Email: justinbaulch@hotmail.com Conflict of interest: The authors declare that there is no conflict of interest. Submitted 5 February 2015; accepted 16 May 2015. doi:10.1111/1754-9485.12338 Abstract Introduction: Accurate definition of the presence and extent of large nerve perineural spread (PNS) is a vital component in planning appropriate surgery and radiotherapy for head and neck cancers. Our research aimed to define the sensitivity and specificity of 3T MRI in detecting the presence and extent of large nerve PNS, compared with histologic evaluation. Methods: Retrospective review of surgically proven cases of large nerve PNS in patients with preoperative 3T MRI performed as high resolution neurogram. Results: 3T MRI had a sensitivity of 95% and a specificity of 84%, detecting PNS in 36 of 38 nerves and correctly identifying uninvolved nerves in 16 of 19 cases. It correctly identified the zonal extent of spread in 32 of 36 cases (89%), underestimating the extent in three cases and overestimating the extent in one case. Conclusion: Targeted 3T MRI is highly accurate in defining the presence and extent of large nerve PNS in head and neck cancers. However, there is still a tendency to undercall the zonal extent due to microscopic, radiologically occult involvement. Superficial large nerve involvement also remains a difficult area of detection for radiologists and should be included as a ‘check area’ for review. Further research is required to define the role radiation-induced neuritis plays in the presence of false-positive PNS on MRI. Key words: 3T MRI; diagnosis; head and neck cancer; perineural spread. Introduction In 2008, an estimated 434 000 Australians were treated for non-melanoma skin cancer (NMSK), with 543 deaths in 2011.1 Perineural spread (PNS) is an important complication of cutaneous malignancy and is associated with increased aggressiveness and increased propensity for tumour recurrence.2 All too often, a patient with PNS of a head and neck NMSK is asymptomatic, or the symptoms do not present until months/years following excision of the cutaneous lesion.3 Although the incidence of PNS in NMSK is relatively low (0.5% for basal cell carcinomas and 2–14% for squamous cell carcinoma), it is also a feature of adenoid cystic carcinoma, melanoma, upper aerodigestive tract cancer, lymphoma and sarcoma.4 The trigeminal nerve (V) and facial nerve (VII) are the most commonly involved cranial nerves.4 578 The management options for PNS are varied and include surgical resection, radiotherapy, a combination of both or palliative care, in untreatable disease. Detection of the presence of PNS, and the extent of its spread, helps stratify patients into these treatment groups and is therefore critical in assigning the appropriate management. With the increased utilisation of 3T MRI, and the increased spatial resolution this affords, evaluation is required to define if the newer modality significantly increases the definition of extent of large nerve PNS, particularly in defining spread beyond the Gasserian ganglion. Patients and methods Patients Utilising the Perineural Database at the Princess Alexandra Hospital, patients were identified who underwent a © 2015 The Royal Australian and New Zealand College of Radiologists 3T MRI Evaluation of PNS 3T MRI ‘neurogram’ during preoperative assessment for head and neck cancers and who then went on to have a surgical resection for suspected PNS. Patients were excluded from the study if the MRI was not performed as a small field of view, thin slice, high-resolution matrix. They were also excluded if there were insufficient histological details to enable identification of the specific nerve involved and its zonal extent of spread. This resulted in 33 patients being identified (27 men and six women) with an age range of 34–86, with their surgery occurring between the years of 2009 and 2014. The distribution of malignancies includes: 27 squamous cell carcinomas (SCCs), three melanomas, two adenoid cystic carcinomas and one salivary gland tumour. Following surgical resection, 57 individual nerves were evaluated histologically. The average time between MRI and surgery was 25 days, ranging from 1 to 65. Imaging techniques Two different 3T MRI systems where utilised. At the Princess Alexandra Hospital, a Siemens Magneton Skyra system (Siemens, Erlangen, Germany) was used. The patients were scanned using a 64 Channel Head/Neck coil. T2 axial fat sat suppressed images (repetition time (TR), 4400 ms: time to echo (TE), 89 ms; Turbo Factor, 17; bandwidth (BW), 260 Hz/Px; 384 × 288 matrix, number of excitations (NEX) 1; 180-mm field of view (FOV); 40 slices; 2-mm thick, 0.4-mm interslice gap), T1 axial images precontrast (TR, 798 ms; TE, 9.1 ms; Turbo Factor, 3; BW, 260 Hz/Px; 320 × 240 matrix, 2 NEX; 180-mm FOV; 45 slices; 2-mm thick, 0.5-mm interslice gap), T1 coronal images precontrast (TR, 750 ms, TE, 9.1ms; Turbo Factor, 3; BW, 260 Hz/Px; 320 × 240 matrix, 2 NEX; 180-mm FOV; 45 slices; 2-mm thick, 0.5-mm interslice gap), T1 coronal fat suppressed images post gadolinium (TR, 986 ms; TE, 9.1ms; Turbo Factor, 3; BW, 260 Hz/Px; 320 × 240 matrix, 2 NEX; 180-mm FOV; 40 slices; 2-mm thick, 0.4-mm interslice gap), T2 SPACE (3D) coronal images (TR, 1000 ms; TE, 136 ms; Turbo Factor, 96; BW, 289 Hz/Px; 384 × 384 matrix, 1.4 NEX; 200-mm FOV; 0.5-mm isotropic voxels) and T1 Mprage (3D) sagittal images post-contrast (TR, 2300 ms: TE, 2.29 ms; Turbo Factor, 208; BW, 200 Hz/Px; 256 × 256, 1 NEX; 240-mm FOV; 0.94mm isotropic voxels). At the Mater Private Hospital, patients were scanned on a GE 3T Discovery MR750 (General Electric Medical Systems, Milwaukee, WI, USA). T1 axial fast spin echo (FSE) (frequency (freq) FOV 16, phase FOV 1, slice thickness 2, spacing 1, TR 808, slices 40, TE min full, echo train length (ETL) 3, NEX 2, BW 22.73); T1 coronal FSE (freq FOV 18, phase FOV 1, slice thickness 2, spacing 1, TR 711, slices 40, TE min full, ETL 3, matrix 288 × 256, NEX 2, BW 31.25, acceleration factor 2), T2 coronal fat sat (freq FOV 18, phase FOV 1, slice thickness 2, spacing 1, TR 7015, slices 40, TE 85, ETL 18, matrix 352 × 256, NEX 1, BW 83.33, acceleration factor 1.5), © 2015 The Royal Australian and New Zealand College of Radiologists 3D T1 fat sat contrast spoiled gradient echo (SPGR) (freq FOV 18, phase FOV 1, slice thickness 1, TR 6.6, TE minimum, matrix 256 × 256, NEX 1, BW 100, acceleration factor 2), T1 coronal fat sat (freq FOV 18, phase FOV 1, slice thickness 2, spacing 1, TR 683, slices 40, TE min full, ETL 4, matrix 352 × 224, NEX 0.5, BW 62.5, acceleration factor 2). Imaging analysis The images were reviewed by a radiologist with 17 years’ experience and subspecialty interest in head and neck radiology. They were initially reviewed blinded to the clinical details of the patient. They were also separately reviewed in conjunction with the clinical findings of the skull base surgeon. The radiological diagnosis of PNS was made when the nerve displayed asymmetrical thickening or enhancement5 (Figs 1–3), obliteration of perineural fat pads6 (Fig. 4) ± secondary denervation changes in the muscles of facial expression or mastication.6 The zonal classification system described by Williams et al.7 was used to define the extent of spread both radiologically and histologically (see Table 1). This system is used extensively by the head and neck multidisciplinary team, as it not only provides a standardised approach, but also helps to define the severity of disease and guide the management options. Histologic analysis Following either en bloc resection encompassing the involved nerve, or intraoperative sampling of large Fig. 1. Coronal T1 fat sat post-contrast MRI shows histologically proven perineural spread with thickening and enhancement of V1 in zone 1. 579 J Baulch et al. Table 1. Zonal classification for perineural invasion in head and neck malignancies V1 V2 V3 VII Zone 1 Zone 2 Zone 3 To the superior orbital fissure To the external aperture of the foramen rotundum To the external aperture of foramen ovale To the external aperture of stylomastoid foramen From zone 1 to the Gasserian ganglion cistern From zone 1 to the Gasserian ganglion cistern From zone 1 to the Gasserian ganglion cistern From zone 1 to the lateral end of the internal auditory canal Into the prepontine cistern and brainstem Into the prepontine cistern and brainstem Into the prepontine cistern and brainstem Into the prepontine cistern and brainstem nerves (see Table 2), the specimens were evaluated with the use of keratin stains to identify intraneural and perineural spread. At a postoperative multidisciplinary meeting, the histology was reviewed in conjunction with the skull base surgeon to define the involved nerves and the extent of spread and to plan future therapy. The chart entry related to this meeting was reviewed to define the histological zonal extent (zones 1, 2 or 3). Results During surgery on the 33 patients, 57 individual nerves were identified and examined histologically, including: nine V1, 21 V2, 14 V3, 13 VII, one ATN (auriculotemporal nerve). Thirty-eight of these nerves displayed PNS: eight V1, 12 V2, nine V3, nine VII, 0 ATN (see Table 3). Four nerves were identified as involved radiologically but with no corresponding histological evaluation, and therefore were not included in the data. The radiologist, while blinded to clinical details, identified 33 out of 38 involved nerves and correctly identified uninvolved nerves in 15 of 19 cases (sensitivity 87%, specificity 79%). With the addition of relevant clinical details the radiologist identified 36 out of 38 involved nerves and correctly identified 16 of 19 uninvolved nerves (sensitivity 95%, specificity 84%). The degree of Fig. 2. Coronal T1 fat sat post-contrast MRI shows histologically proven perineural spread with thickening and enhancement of V3 in zone 2. Fig. 3. Coronal T1 fat sat post-contrast MRI shows histologically proven perineural spread with enhancement of CN VII in zone 2. 580 Fig. 4. Coronal T1 shows histologically proven perineural spread with effacement of perineural fat surrounding V2 within the pterygopalatine fossa. © 2015 The Royal Australian and New Zealand College of Radiologists 3T MRI Evaluation of PNS Table 2. Surgical approach Surgery 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 Pterional craniotomy R/O GG and supraorbital tissue and orbit Radical parotidectomy, neck dissection, mandible & ITF Orbital exenteration, parotidectomy and neck dissection Transfacial & pterional craniotomy for R/O V2/V3/GG/PPF WLE skin & transfacial V2 resection Transfacial R/O ION/PPF/V2 Radical parotidectomy and partial temporal bone resection Transfacial Microscope Resection V2 LTBR/radical parotidectomy/hemimandibular resection/ITF/pterional craniotomy/cavernours sinus resection/GG / Free flap Radical parotidectomy, ascending mandible, V3/GG/V2 to PPF Radical parotid/ascending mandible/ITF/LTBR/V3 to GG/VII to geniculate I-III neck Superficial parotidectomy & R/O midface VII branches WLE forehead skin, frontal bone, orbital exenteration, R/O front of GG Orbital exenteration R/O V1/V2 and ganglion and craniotomy WLE cheek + R/O ION & PPF Transfacial resection of infraorbital nerve Transfacial R/O V2 + PPF Lower lip WLE + segmental mandibulectomy + neck dissection Orbital exenteration, pterional craniotomy and R/O supraorbital nerve & ALT flap WLE + neck dissection + ascending mandible + ITF + LTBR + flap. Neck dissection for access/flap Hemi-mandible + infratemporal fossa + craniotomy WLE & resection of infraorbital nerve & orbital exenteration Orbital exenteration + WLE cheek/anterior maxilla + trigeminal nerve to PPF + retrograde parotidectomy + free flap recon (ALT) + glabella ± cheek flap ± facial sling suspension Excision cheek lesion; V2 chase to foramen rotundum; vidian nerve excision to geniculate ganglion; lesser and greater palatine nerve resection Partial pinnectomy, LTBR and excision of part of the mandible. V3 resection up to the foramen ovale. Total parotidectomy. Level 2–3 neck dissection. LTBR; Radical parotidectomy; V3 close to GG; Free flap reconstruction Orbital exenteration, WLE frontal mass Radical parotid, mastoid drill, VII resection, ITF clearance and partial mandibulectomy. Trigeminal taken to skull base Orbital exenteration w/ V1 & V2 resection Hemi-mandibulectomy/rad parotid/LTBR/ITF/V2-V3/PPF/latissimus dosi + scap flap. WLE/LTBR/radical parotidectomy/ascending mandible WLE Palatal Mass, R/O PPF, V2, Pterional Craniotomy + reconstruction Orbital exenteration; superficial parotidectomy; neck dissection 1–4 R/O, removal of; GG, Gasserian ganglion; ITF, infratemporal fossa; V1, ophthalmic nerve; V2, maxillary nerve; V3, mandibular nerve; VII, facial nerve; PPF, pterygopalatine fossa; WLE, wide local excision; ION, infraorbital nerve; LTBR, lateral temporal bone resection; ALT, anterolateral thigh. zonal spread was correctly identified in 27 of 33 nerves (82%) when no clinical details were available, and 32 of 36 nerves (89%) when clinical details were provided. As the addition of relevant clinical details more accurately © 2015 The Royal Australian and New Zealand College of Radiologists Fig. 5. Axial T1 fat sat post-contrast shows histologically proven perineural spread with thickening and enhancement of the buccal branch of CN VII. represents daily practice, these results form the basis of our discussion (see Table 3). In our case review, there were two false negatives, both involving CN VII. The first is in patient 23, where zone 1 CN VII involvement was identified histologically but no involvement reported on imaging. On retrospective review, zone 1 facial nerve involvement is evident on MRI as superficial enhancement and thickening of the buccal branch of the facial nerve (Fig. 5), not appreciated at the time of reporting. In patient 25, CN VII involvement was again present histologically but not reported on MRI. On retrospective review, no convincing CN VII trunk enhancement or thickening was evident; however, zone 1 disease of V3 was correctly identified, along with involvement of the ATN, although this nerve was not assessed individually on histology. There were three false positives, occurring in a single case, patient 20, who had a pre-auricular SCC initially diagnosed in 2011, and subsequent chemoradiotherapy, completed in 2012. There was radiological recurrence in late 2012, at which time the patient had a superficial parotidectomy, in which no tumour was seen histologically, followed-up with further chemoradiotherapy. PET examination in July 2013 showed avidity in the infratemporal fossa and MRI in August (the MRI included in our study) showed extensive enhancement and thickening of CN VII, ATN and V3 (see Fig. 6). Following wide local excision, mandiblectomy and infratemporal fossa clearance, multiple foci of SCC were identified but no evidence of perineural invasion. It was noted, however, that where SCC was evident, it was surrounded by dense sclerosis and active chronic inflammatory infiltrate. 581 J Baulch et al. Table 3. Results Age 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 58 56 34 81 77 71 61 70 73 46 63 56 56 61 79 69 63 50 58 55 60 86 68 52 66 61 61 60 55 67 59 63 43 Tumour SCC SCC SCC SCC SCC SCC Adenoid C SCC SCC SCC SCC SCC SCC SCC SCC SCC Melanoma Melanoma SCC SCC Melanoma SCC SCC SCC SCC SCC SCC SCC SCC SCC Salivary Adenoid C SCC Nerves excised V1,V2 V3,VII V1,V2 V2,V3 V2 V2 VII V2 V2,V3,VII V2,V3 V3,VII VII V1 V1,V2 V2 V2 V2 V3,VII V1 VII,V2,V3,ATN V2,V3 V2 V1,V2,VII V2 VII,V3 V3,VII V1 VII,V2,V3 V1,V2 V2,V3,VII V3,VII V2,V3 V1 MRI assessment V1 V2 Z2 0 Z1 0 Z2 Z2 Z2 V3 VII 0 Z1 Pathological assessment ATN V1 V2 Z2 0 Z1 0 0 Z2 Z2 Z2 V3 VII 0 Z1 Z1 Z2 Z1 Z2 0 Z2* Z2 Z2 Z1 0 0 Z2 Z2 Z1 Z2 Z1 0 0 Z2 0 Z2 Z1 Z1 0 0 Z1 Z1 Z2 Z2 Z2 Z2 ATN 0 Z1 Z2 0 0 Z2 Z2 0 0 0 Z1 Z2 Z2 ATN* Z1 Z2 0 0 Z1 Z1 Z1 0 Z1 0 Z1 Z2 0 ATN* 0 Z1 Z1 Z1 Z1 ATN 0 Z1 Z2 Z1 0 Correlation Z1 Z1 Z2 Z2 0 Z2 Z2 Z3 Z2 0 Z2 Z1 Z1 Z1* 0 0 Z2 Z2 Z2 Z2 0 Z2 Z1 0 0 Z2 Z1 Z1 0 Y Y Y Y Y Y N Y Y Y Y N Y Y Y Y Y Y Y N Y Y N Y N Y Y N Y N Y Y Y *Nerves identified on imaging but not individually assessed on histology. SCC, squamous cell carcinoma; Adenoid C, adenoid cystic carcinoma; V1, ophthalmic nerve; V2, maxillary nerve; V3, mandibular nerve; VII, facial nerve; ATN, auriculotemporal nerve; Z1, zone 1; Z2, zone 2; Z3, zone 3; Y, complete correlation; N, incomplete correlation. Large nerve bundles traversed this region of sclerosis and some of the foci of SCC had foci of nerve fibres at their periphery, therefore, it was postulated that the foci of tumour may have arisen within the nerves, and then largely obliterated them. This may account for the multiple false positives in this case. The other consideration is that radiation-induced neuropathy may account for the apparent nerve thickening and enhancement. Although this is not a widely recognised entity in head and neck imaging, it is extensively described in other body regions, particularly the brachial plexus following lung and breast radiotherapy.8 In patient 28, as well as V3 and CN VII involvement, the MRI showed apparent thickening of V2 in zone 1, which was interpreted as further PNS. Due to the lack of clinical features of V2 involvement, the periphery of this 582 nerve was not resected, although it was sampled at the Gasserian ganglion intraoperatively and was negative. This indicates correct MR assessment of no V2 zone 2 disease; however, it does not allow us to assess for histological evidence of zone 1 disease, therefore this nerve has been excluded from the data. The extent of zonal spread was correctly identified in 32 of 36 patients, with three patients underestimated by a single zone and one patient overestimated by a single zone. Discussion PNS of head and neck malignancy is a now a wellrecognised complication in NMSK, melanoma and adenoid cystic carcinoma. As Australia has the highest © 2015 The Royal Australian and New Zealand College of Radiologists 3T MRI Evaluation of PNS Fig. 6. Axial T1 fat sat post-contrast in patient 20 shows apparent thickening and enhancement of the right V3, but no large nerve perineural spread was identified histologically. incidence of skin cancer in the world, with NMSK being the most commonly diagnosed cancer,9 the morbidity and mortality related to PNS of head and neck malignancy is a significant contributor to health-care costs. This is especially true in Queensland, with our institution, the Princess Alexandra Hospital, having a database of 146 cases of PNS, diagnosed since 1997. The incidence of PNS is quoted as 2–14% for squamous cell carcinomas and 0.5% for basal cell carcinomas.10 The mechanism of PNS has not been completely defined; however, the most likely theory is that large nerves represent the path of least resistance.11 It should be noted that large nerve PNS is a separate entity to ‘perineural invasion’ of small nerves, which is an incidental finding extensively reported on histopathological specimens. Involved nerves are identified on MRI as asymmetrically thickened/enhancing nerves5 with obliteration of perineural fat planes.6 There may also be associated denervation changes in the muscles of mastication or facial expression.6 The most commonly involved nerves are the branches of the trigeminal nerve and the facial nerve. The auriculotemporal nerve is also being increasingly recognised as an important pathway of PNS. The ATN anatomy, and its role in PNS, was extensively described initially by Schmalfuss et al.12 in 2002. It is formed by two roots that arise from V3 below the skull base to form a short trunk, with rami that join the facial nerve within the parotid gland, thus allowing communication between V3 and the facial nerve. © 2015 The Royal Australian and New Zealand College of Radiologists Clinicians at our institutions are encouraged to request specific ‘neurogram’ studies if there is concern regarding possible PNS of a head and neck cancer. These are performed exclusively on our 3T MRIs, utilising small FOV, thin slice, high-resolution matrix images with the shortest acquisition time in order to reduce patient motion. The clinicians are also encouraged to indicate if there is any clinical suspicion of PNS, in order to better target likely involved nerves, both at the time of imaging and at the time of reporting. Accurately defining the presence and extent of large nerve PNS greatly assists clinicians, both surgeons and radiation oncologists, in tailoring their treatment for specific cases. Jardim et al.13 and Kosec et al.14 have both recently showed that the presence of PNS is a negative prognostic factor for patient outcomes. Broadly speaking, patients are treated with either combination surgery and radiotherapy, radiation alone or with palliation. As described by Panizza et al.,15 surgical techniques include subcranial resection or skull base resection. Subcranial resection includes transfacial resection of the trigeminal nerve and its branches, including the content of the pterygopalatine fossa or orbit up to the skull base, and facial nerve resection to its mesotympanic portion. Skull base resection is surgical resection extending intracranially. The surgery is designed to halt the central spread of disease, as retrograde or centrifugal spread is not common.16 PNS beyond the Gasserian ganglion of the trigeminal nerve, or the Geniculate ganglion of the facial nerve, usually deems the patient inoperable. Previous research has demonstrated a high level of accuracy of 1.5 T MRI in detecting the presence of PNS of head and neck cancers.17 Nemzek et al.18 reported the sensitivity for detection of PNS was 95% and the sensitivity for defining the zonal extent was only 63%. These images were obtained on a variety of imaging units and protocols. Hanna et al.19 evaluated 38 cranial nerves and defined the sensitivity of MRI as 100% and specificity of 85%. They did not assess the accuracy of zonal extent. Our study again demonstrates a high level of accuracy of ‘targeted’ 3T MRI neurogram studies in the detection of PNS, with a sensitivity of 95% and a specificity of 84%. This accuracy reduces when the radiologist is blinded to relevant clinical details regarding focal neurology, with a sensitivity of 87% and a specificity of 79%. This serves to remind both clinicians and radiologists of the importance of clinical details in requesting and reporting MRI neurograms. The main limitation of our study is that the gold standard for identification of PNS, which is histological evaluation, is available in a relatively small percentage of cases, therefore limiting the statistical analysis. While including patients with clinical evidence of nerve involvement or progressive disease on follow-up examinations may have increased the study numbers, it would not allow us to compare the extent of spread, which is an 583 J Baulch et al. important part of our study. It would also introduce the problem of those patients with focal neurology secondary to causes other than PNS, such as post-surgical paraesthesia’s following previous superficial surgical resections, as described by Lee et al.20 The other limitation is the lack of histological sampling of radiological zone 3 disease, which therefore excluded these cases from the study and limits our assessment of the accuracy of 3T MRI in the detection of zone 3 disease. We may infer an increased sensitivity from the fact that only one case of false negative, histologically proven zone 3 disease was seen in our study, out of a total of 59 nerves reviewed. Overall, our study has displayed a high level of accuracy of ‘targeted’ 3T MRI in the detection of the presence of PNS of head and neck cancers and its extent of spread. There was only a single case of false negative PNS where, even on review of the imaging following histological evaluation, the involved nerve appears to have a normal MRI appearance. The other case in which PNS was not reported preoperatively represents superficial involvement of the facial nerve, which was clearly evident on retrospective review of the MRI following histology. Although many of these superficial nerves are well within the resolution capabilities of 3T MRI, this remains somewhat of a ‘blind-spot’ for reporters and should become a check area for review. All three cases of false positive PNS involved a single patient who had extensive preoperative radiotherapy and evidence of possible nerve ‘obliteration’ by SCC, which in combination, may account for the radiological findings. Conclusions In the evaluation of head and neck cancers, 3T MRI targeted to the course of the trigeminal and facial nerves has a high sensitivity and specificity for the detection of large nerve PNS. It is also highly accurate in the assessment of the extent of spread along the nerve, although there is a tendency to undercall the zonal extent due to the presence of radiologically occult microscopic involvement, which is still beyond the resolution of 3T MRI. Difficulty may arise in patients with history of previous radiotherapy due to radiation-induced neuropathy masquerading as PNS, although the significance of this entity in head and neck cancers requires further evaluation. Some difficulty also remains in the detection of superficial nerve involvement; however, this appears to represent more of a radiologist ‘blind-spot’ rather than a technical limitation of 3T MRI. Acknowledgements The authors gratefully acknowledge the technologists and radiologists of Queensland X-Ray and the Princess Alexandra Hospital for their help and encouragement. 584 References 1. Australian Institute of Health and Welfare (AIHW) 2015. Australian Cancer Incidence and Mortality (ACIM) books: Non-melanoma skin cancer, all types. Canberra: AIHW. 2. Feasel AM, Brown TJ, Bogle MA, Tschen JA, Nelson BR. Perineural invasion of cutaneous malignancies. Dermatol Surg 2001; 27: 531–42. 3. Catalano P, Sen C, Biller HF. Cranial neuropathy secondary to perineural spread of cutaneous malignancies. Am J Otol 1995; 16: 772–7. 4. Parker GD, Harnsberger HR. Clinical-radiologic issues in perineural tumour spread of malignant diseases of the extracranial head and neck. Radiographics 1991; 11: 383–99. 5. Caldemeyer KS, Mathews VP, Righi PD, Smith RS. Imaging features and clinical significance of perineural spread or extension of head and neck tumours. Radiographics 1998; 18: 97–110. 6. Ginsberg LE. MR imaging of perineural tumour spread. Magn Reson Imaging Clin N Am 2002; 10: 511–25. 7. Williams LS, Mancuso AA, Mendenhall WM. Perineural spread of cutaneous squamous and basal cell carcinoma: CT and MR detection and its impact on patient management and prognosis. Int J Radiat Oncol Biol Phys 2001; 49: 1061–9. 8. Bowen BC, Seidenwurm DJ. Plexopathy. AJNR Am J Neuroradiol 2008; 29: 400–2. 9. Australian Institute of Health and Welfare and Australiasian Association of Cancer Registries, Cancer in Australia: an overview. Canberra 2008. 10. Parker GD, Harnsberger HR. Clinico-radiologic issues in perineural tumour spread of malignant disease of the extracranial head and neck. Radiographics 1991; 11: 383–99. 11. Larson DL, Rodin AE, Roberts DK, O’Steen WK, Rapperport AS, Lewis SR. Perineural lymphatics: myth or fact. Am J Surg 1966; 112: 488–92. 12. Schmalfuss IM, Tart RP, Mukherji S, Mancuso AA. Perineural spread along the auricultemporal nerve. AJNR Am J Neuroradiol 2002; 23: 303–11. 13. Jardim JF, Francisco AL, Gondak R, Damascena A, Kowalski LP. Prognostic impact of perineural invasion and lymphovascular invasion in advanced stage oral squamous cell carcinoma. Int J Oral Maxillofac Surg 2015; 44: 23–8. 14. Kosec A, Svetina L, Luksic I. Significance of clinical stage, extent of surgery and outcome in cutaneous squamous cell carcinoma of head and neck. Int J Oral Maxillofac Surg 2013; 42: 82–8. 15. Panizza B, Solares CA, Redmond M, Parmar P, O’Rourke P. Surgical resection for clinical perineural invasion from cutaneous squamous cell carcinoma of the head and neck. Head Neck 2012; 34: 1622–7. 16. Panizza B, Warren T. Perineural invasion of head and neck skin cancer: diagnostic and therapeutic © 2015 The Royal Australian and New Zealand College of Radiologists 3T MRI Evaluation of PNS implications. Curr Oncol Rep 2013; 15: 128–33. 17. Gandhi MR, Panizza B, Kennedy D. Detecting and defining the anatomic extent of large nerve perineural spread of malignancy-comparing targeted MR imaging with the histological findings following surgery. Head Neck 2011; 33: 469–75. 18. Nemzek WR, Hecht S, Gandour-Edwards R, Donald P. Perineural spread of head and neck tumour: how accurate is MR imaging? AJNR Am J Neuroradiol 1998; 19: 701–6. © 2015 The Royal Australian and New Zealand College of Radiologists 19. Hanna E, Vural E, Prokopakis E, Carrau R, Snyderman C, Weissman J. The sensitivity and specificity of high-resolution imaging in evaluating perineural spread of adenoid cystic carcinoma to the skull base. Arch Otolaryngol Head Neck Surg 2007; 133: 541–5. 20. Lee KC, Panizza B. Existence of MRI-negative clinical (large nerve) perineural squamous cell carcinoma spread. Head Neck 2009; 31: 1531. 585