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