Keywords
Depth of invasion - magnetic resonance imaging - tumor node metastasis staging - tongue
carcinoma
Introduction
Most tumors of the tongue occur on the lateral and under surface.[[1]] Dorsal tumors are uncommon but when they do occur, they are usually located near
the midline and more posteriorly.[[2]] Oral tongue tumors tend to remain in the tongue. Tumors in the anterior third of
the oral tongue invade the floor of the mouth.[[3]] Middle-third lesions infiltrate the musculature of the tongue and later, the lateral
floor of the mouth.[[4]] Carcinomas involving the posterior third of the tongue grow into the musculature
of the tongue, the floor of the mouth, the anterior tonsillar pillar, the tongue base,
the glosso-tonsillar sulcus, and the mandible.[[5]]
MRI provides valuable information both within and without the tongue. The tongue carcinoma
may extend far beyond the gross tumor margin seen on surgery, which is often deceiving.
It is known that the most important factor governing local recurrence is the resection
margin.[[6]] Whereas 1 cm is generally considered adequate for most squamous cell carcinomas,
the margins for tongue cancer should be 1.5–2.0 cm.[[7]] Tumors with deep margins are often difficult to assess during surgery. In addition,
these tumors are technically more difficult to resect. Hence, deep margins are frequently
the site of positive or inadequate resection margins. Up to 35% of patients have nodal
metastasis on presentation.[[8]] Five percent of these patients have bilateral lymph node involvement.[[9]] The first echelon nodes are the submandibular and jugulodigastric nodes.[[10]] Submental node involvement is uncommon except in patients with tumor at the tip
of tongue.[[11]] It should be noted that in patients with clinically N0 neck, the overall occult
metastatic rate is approximately 30%.[[12]] Various clinical studies have been performed to correlate the depth of tumor invasion
with the likelihood of cervical nodal metastasis. These studies reveal that the single
most important factor in predicting lymph node metastasis is the depth of tumor invasion.[[13]]
Tongue base carcinoma is a clinically silent region and tumors tend to spread with
deep infiltration. As a general rule, the extent of these tumors is underestimated
during clinical examination. Tongue base tumors tend to remain in the tongue except
for laterally placed lesions or late cases. Under such circumstances, tongue base
tumors may extend into the tonsillar fossa. Tonsillar carcinomas, on the other hand,
have a tendency to invade the tongue base. For tongue base carcinoma, the first echelon
nodes are the jugulodigastric nodes, followed by mid and lower jugular nodes. Retropharyngeal
nodes are occasionally involved. Submandibular nodes may be involved if there is anterior
tumor extension. Submental nodes are rarely involved. Seventy-five percent of patients
have positive nodes on presentation, while 30% have bilateral nodal metastases.[[14]] Patients with clinically N0 neck have a 30%–50% rate of occult metastases.[[15]]
Imaging anatomy
The tongue comprises dorsum, apex, inferior surface, and root. The root (base) is
attached to the hyoid bone and mandible while the apex forms the tip of the tongue.
The sulcus terminalis is a shallow groove with the circumvallate papillae just anterior
to it and divides the tongue into the oral (anterior two-thirds) and pharyngeal (posterior
third) parts. As a general guide on axial imaging, a line joining the anterior aspect
of the mandibular rami may be used as the dividing line between these two parts, which
differ in their developmental origins and hence their nerve supplies.[[16]]
The tongue muscles are divided into intrinsic and extrinsic groups. The intrinsic
muscles are entirely within the tongue with no bony attachment and are organized into
superior and inferior longitudinal, vertical, and transverse bands. Their principle
function is altering the shape of the tongue. The extrinsic muscles consist of genioglossus,
hyoglossus, styloglossus, and palatoglossus. These extrinsic muscles stabilize the
tongue and alter its position, as well as its shape. All the muscles of the tongue,
intrinsic and extrinsic, are thus innervated by the hypoglossal nerve. The exception
being palatoglossus, which being essentially a palate muscle, is supplied by the pharyngeal
plexus.
The anatomy of the tongue is well demonstrated on magnetic resonance imaging (MRI).
On axial T1-weighted images, fat with high signal intensity can be seen interspersed
between the muscles of intermediate signal intensity. MRI is the preferred modality
in the evaluation of tongue carcinomas. The abnormal signals seen on MRI are well
correlated with pathological findings. Tumor invasion of the floor of the mouth is
particularly well seen on coronal images. Sagittal images provide information on tongue
base involvement and the extent of pharyngeal infiltration.[[17]]
Genioglossus is the largest of all the tongue muscles and forms the bulk of the tongue.
It arises from the genial tubercle and is easily seen on MRI. It fans out widely and
inserts inferiorly into the hyoid bone; posteriorly into the tongue base; and superiorly
into the entire ventral surface of the tongue. Hyoglossus is a thin quadrilateral
sheet of muscle arising from the hyoid bone. It ascends superiorly, interdigitating
with the fibers of the styloglossus, and attaches to the side of the tongue. The hyoglossus
muscles define the lateral margins of the tongue and are readily identified on MRI.
Both the styloglossus (which arises from the styloid process and stylohyoid ligament)
and the palatoglossus (which originates from the palatine aponeurosis) cannot be seen
with certainty on imaging studies. Lymph from the tip of the tongue drains to the
submental nodes. Marginal lymphatics from the outer third of the rest of the oral
tongue are directed to ipsilateral submandibular and jugulodigastric nodes. Central
lymphatics of the inner two-thirds of the oral tongue have pathways to nodes of both
sides of the neck.[[18]]
TNM staging
Tumor node metastasis (TNM) classification is the most commonly used system for describing
malignant tumors, their regional involvement, and distant metastases.[[19]] The TNM and stage grouping are presented below:
Aims and objectives
Aim of the study is to correlate MRI and histopathological findings, to evaluate the
role of MRI in loco-regional TNM staging, and to assess the depth of invasion of tongue
carcinoma.
Materials and Methods
This study was undertaken in the Department of Radiology at a tertiary care hospital
in India over the 2-year period between July 2017 and June 2019. Before subjects were
recruited, the study protocol was approved by the institutional ethics committee (IEC),
in accordance with the ethical principles for human investigation outlined by the
Second Declaration of Helsinki, and written informed consent was obtained from all
patients prior to their enrollment in this study (IEC, Holy Family Hospital; IEC Approval
Reference Number: HFH/12/2017; IEC Approval Date: June 12, 2017). MR examinations
wereperformed using a 1.5-T scanner (Signa, General Electric Medical Systems, Milwaukee,
WI, USA). Neurovascular (NV) arraycoil was used. The patient’s head was secured using
relaxing cushion; ensuring that the shoulders touch the lower part of the coil. The
protocol included axial, sagittal, and coronal T1-weighted turbo spin echo (TSE),
axial and coronal T2-weighted turbo spin echo (TSE), and gadolinium-enhanced axial
and coronal T1-weighted sequences with fat suppression (FS) as well as diffusion-weighted
(DW) sequences [[Table 1]]. The tumor depth was measured at post contrast T1 coronal FS. The tumor thickness
was defined by the distance from the deepest point of invasion to the tumor surface.
At first, a vertical line joining the maximum length between tumor-mucosa junctions
was drawn as a reference line. The tumor thickness was determined by the summation
of two lines drawn perpendicular from the reference line to the point of maximum tumor
extension.
Table 1
Protocol for MRI Tongue in the current study is as below
|
Sequence
|
Slice
|
Slice thickness
|
Gap
|
Matrix
|
|
T1 Axial
|
29
|
4 mm
|
0.4 mm
|
512
|
|
T1 Coronal
|
23
|
4 mm
|
0.4 mm
|
512
|
|
STIR Coronal
|
23
|
4 mm
|
0.4 mm
|
256
|
|
T2 Fatsat Axial
|
29
|
4 mm
|
0.4 mm
|
512
|
|
T1 Fatsat Axial+C
|
23
|
4 mm
|
0.4 mm
|
512
|
|
T1 Fatsat Coronal + C
|
19
|
4 mm
|
0.4 mm
|
512
|
|
T1 Fatsat Sagittal+C
|
19
|
4 mm
|
0.4 mm
|
512
|
Clinical and MRI staging of tongue carcinoma was done preoperatively and correlated
[[Figures 1], [2], [3], [4], [5], [6], [7], [8], [9], [10], [11], [12], [13], [14], [15], [16]]. Post-surgery, histopathological TNM staging was done and correlated with clinical
and MRI TNM staging [[Figures 17], [18], [19], [20]] and [[Tables 2], [3], [4], [5]]. T1 tumor measures ≤2 cm in greatest dimension with depth of invasion (DOI) ≤5
mm. T2 tumor measures ≤2 cm with DOI >5 mm. T3 tumor measures >2 cm and ≤4 cm with
DOI >10 mm. T4a is moderately advanced local disease tumor >4 cm with DOI >10 mm.
T4b is very advanced local disease with tumor invasion into the masticator space,
pterygoid plates, or skull base, and/or tumor encases the internal carotid artery.
Figure 1 (A and B): Axial schematic representation (A) and T1-weighted magnetic resonance (MR) image
(B) demonstrate the root of the tongue. The high-signal-intensity lingual septum (ls)
is clearly seen and is flanked by the genioglossus muscles (gg), which form an inverted
V anteriorly before blending into the intrinsic muscles of the mobile tongue. The
sublingual spaces (sls) are lateral to the genioglossus and geniohyoid muscles and
also show high T1 signal intensity. Axial T1-weighted image (B) shows the tongue muscles,
genioglossus (long arrow), and hyoglossus (short arrow)
Figure 2 (A and B): Coronal schematic representation (A) and T1-weighted MR image (B) demonstrate genioglossus
muscles (gg) which resemble paramidline vertical pillars. Below the genioglossus muscles,
the geniohyoid muscles (gh) appear subtly wider than they do on axial images [Figure
1]. The sublingual spaces (sls) show high T1 signal intensity. ls = lingual septum.
Coronal T1- weighted image (B) shows lingual septum (short arrow) and mylohyoid (long
arrow), which form the floor of the mouth
Figure 3 (A and B): Sagittal drawing (A) and T1-weighted MR image (B) demonstrate the geniohyoid muscles
(gh) and the fanlike shape of the genioglossus muscles (gg). The mylohyoid muscle
(mh) extends from the mandible to the hyoid bone and supports the floor of the mouth.
Sagittal T1-weighted image (B) shows the fan-shaped genioglossus (short arrow), the
longitudinal intrinsic muscle (long arrow), and darkly hypointense geniohyoid (star)
from genial tubercle to hyoid
Figure 4 (A-D): Mass lesion (star) in the right lateral aspect of the anterior 2/3rd of the tongue with inferior extension into the posterior aspect of right sublingual
space. The lesion appears isointense on T1 (A), hyperintense on T2 (B), and hyperintense
on STIR (C and D) and extends medially up to the lingual septum with no obvious extension
across the midline. Axial STIR image (C) demonstrates an ill-defined nodular mass
lesion involving intrinsic muscles of the anterior tongue including genioglossus,
myelohyoid, and geniohyoid. Inferiorly, the lesion invades the right lateral floor
of mouth and sublingual space (short arrow). Posteriorly, there is an invasion of
right pterygomandibular raphe (long arrow)
Figure 5 (A-F): Mass lesion (star) involving the alveolar margin of the mandible in the midline extending
along the lingual septum and genioglossus muscle with infiltration of the sublingual
spaces anteriorly. The lesion appears isointense on T1 (A), heterogeneously hyperintense
on T2 (B), and hyperintense on STIR (C and D) showing contrast enhancement on T1 +
c images (E and F) and a central focus of nonenhancement— suggestive of necrosis.
Axial T2 weighted fat-suppression MR image (B) demonstrates a bulky enhancing tumor
in genioglossus of anterior tongue extending into anterior alveolar margin of mandible
causing erosion of occlusal cortices of mandible and enhancement in the marrow (short
arrow). Sagittal T1 + c image (F) shows tumor invading the floor of the mouth (long
arrow). There is sparing of the mylohyoid muscle inferiorly
Figure 6 (A-F): Irregular shaped mass lesion (star) in the lateral aspect of the anterior 2/3rd of the left tongue with no extension across the midline and no obvious involvement
of the floor of the mouth/sublingual space. The lesion appears isointense on T1 (A),
heterogeneously hyperintense on T2 (B), and hyperintense on STIR (C and D) showing
contrast enhancement on T1 + c images (E and F) with a central focus of nonenhancement—s/o
necrosis. Axial T2 weighted fat-suppression MR image (B) demonstrates a nodular mass
lesion in the anterior 2/3rd of the oral tongue with infiltration of the genioglossus (short arrow) and reaching
up to sublingual space invading mylohyoid (long arrow). Coronal T1 + c image (E) shows
no obvious invasion of the floor of mouth
Figure 7 (A-F): Ill-defined mass lesion (star) extending along the anterior and left alveolar margins
of the mandible with infiltration of bilateral genioglossus muscles, the lingual septum,
bilateral sublingual spaces, left masticator space, left submandibular gland, and
left geniohyoid muscle inferiorly. The lesion appears isointense on T1 (A), hyperintense
on T2 (B), and hyperintense on STIR (C and D) showing contrast enhancement on T1 +
c images (E and F). There is sparing of the mylohyoid muscle. Axial T2-weighted fat-suppression
image (B) reveals an ill-defined heterogeneous signal intensity nodular mass lesion
involving the anterior tongue on left side. It crosses midline anteriorly and involves
genioglossus and geniohyoid. Coronal STIR image (D) shows tumor infiltration of the
floor of the mouth. Note the involvement of ipsilateral mylohyoid muscle (short arrow)
and normal contralateral mylohyoid. Note the tumor infiltration into bilateral sublingual
glands (long arrows) on T2FS axial image (B)
Figure 8 (A-F): Irregular shaped mass lesion (star) in the left side of the tongue appearing isointense
on T1 (A), heterogeneously hyperintense on T2 (B), and hyperintense on STIR (C and
D) showing contrast enhancement on T1 + c images (E and F). The lesion extends across
the midline into the right side. Inferiorly the lesion extends into the left sublingual
space causing loss of fat plane with the mylohyoid muscle and to the anterior aspect
of the right sublingual space. Axial STIR image (C) demonstrates an ill-defined nodular
hyperintense mass lesion of tongue invading genioglossus, myelohyoid, and geniohyoid
in the left lateral and anterolateral aspects of the tongue extending up to lingual
septum (short arrow) and crossing the midline (white line). Posteriorly, the lesion
invades the base of tongue and vallecula on left side and abuts the anterior tonsillar
pillar (long arrow)
Figure 9: Anteroposterior dimension: Axial three-dimensional image. Two perpendicular lines
“a” and “b” drawn are the anterior and posterior tumor-mucosal junction, respectively.
The length of the horizontal line “c” connecting these two perpendicular lines is
considered as anteroposterior dimension which corresponds to T3 stage
Figure 10: Craniocaudal dimension: Postcontrast T1-weighted image (Coronal reformat). Two horizontal
lines, “a,” “b,” were drawn on the superior and inferior tumor-mucosal junctions.
A line “c” is drawn perpendicular to lines “a” and “b” through the middle of the tumor.
The length of this line “c” represents the craniocaudal dimensions which correspond
to T2 stage
Figure 11: Coronal T2-weighted image demonstrating muscle invasion. Shows muscle invasion by
the tumor. G = Genioglossus, H = Hyoglossus, S = Styloglossus, and M = Mylohyoid.
Tumor on the right side shows the invasion of all muscles except myelohyoid
Figure 12 (A and B): (A) Axial T2 weighted fat-suppression image shows a right-sided tongue cancer (T1N1M0)
extending more than 5 mm from the lateral margin of the tongue. (B) Coronal T2 weighted
fat-suppression image shows bilateral submandibular lymphadenopathy (arrows), a result
of the lymphatic drainage pathways of the inner two-thirds of the oral tongue. Radiological
depth of invasion (DOI) measured 3.5 mm and histopathological DOI measured 3.2 mm.
MRI and histopathology assessments of tumor spread were equivalent to within 0.5 mm
DOI
Figure 13 (A and B): (A) Axial T2 weighted fat-suppression image shows a right-sided tongue base cancer
(T4aN2aM0-long arrows). An enlarged right jugulodigastric node is also seen (short
arrow), the first echelon node of tongue base carcinoma. (B) Sagittal T2 weighted
fat-suppression image of the same patient shows the extent of pharyngeal invasion
of the tongue base tumor (arrow).Radiological DOI measured 13.2 mm and histopathological
DOI measured 12.8 mm. MRI and histopathology assessments of tumor spread were equivalent
to within 0.5 mm DOI
Figure 14: Axial T1 weighted image shows a tongue cancer (T4aN2aM0) with mandible invasion.
However, the early involvement of cortical bones is better seen on CT images. Radiological
DOI measured 14.4 mm and histopathological DOI measured 14.1 mm. MRI and histopathology
assessments of tumor spread were equivalent to within 0.5 mm DOI
Figure 15 (A and B): (A) Sagittal T2 weighted fat-suppression image shows carcinoma in the anterior third
of the oral tongue (T4aN1M0-arrow). (B) Sagittal T2 weighted fat-suppression image
(same patient) shows tumor invading the floor of the mouth (arrow). Radiological DOI
measured 13.2 mm and histopathological DOI measured 12.9 mm. MRI and histopathology
assessments of tumor spread were equivalent to within 0.5 mm DOI
Figure 16 (A and B): (A) Coronal T2 weighted fat-suppression image shows a carcinoma in the middle third
of the oral tongue (T3N1M0) with early infiltration (long arrow) of the tongue musculature
(genioglossus). Note the ipsilateral submandibular lymphadenopathy (short arrow).
(B) Coronal post-contrast T1 weighted fat-suppression image of a more advanced case
shows the tumor invading the lateral floor of the mouth (T4aN1M0-arrow). Radiological
DOI measured 9.2 mm and histopathological DOI measured 8.8 mm. MRI and histopathology
assessments of tumor spread were equivalent to within 0.5 mm DOI
Figure 17 (A-C): (A) An example of minimal residual squamous cell carcinoma of the oral tongue (hematoxylin
and eosin (H and E), original magnification ×40). (B) A minute focus of residual squamous
cell carcinoma, about 1mm wide and 1mm deep, with submucosal scar in the left lower
corner (H and E, original magnification ×100). (C) The diagnostic biopsy was represented
by five tissue fragments, all of which were smaller than 5mm in greatest dimension
and had invasive squamous cell carcinoma. The exact measurement of the depth of invasion
in this case is difficult given the fragmented nature of diagnostic biopsy. Only one
biopsy fragment had normal squamous mucosa allowing measurement of the depth of invasion
(H and E, original magnification ×40)
Figure 18 (A and B): An example of a T2 squamous cell carcinoma of the oral tongue with positive deep
margin, indicating that the depth of invasion may be underestimated. (A) The apparent
depth of invasion is 7mm; however, the deep margin is involved by carcinoma (hematoxylin
and eosin (H and E), original magnification ×20). (B) Carcinoma at deep margin. Hypothetically,
if there is additional 4mm (along the “plumb line”) of residual carcinoma in the tumor
bed, this carcinoma is more appropriately staged as T3 (H and E, original magnification
×100)
Figure 19 (A and B): Cross-section through left partial glossectomy with dorsal, lateral, and ventral
(toward floor of mouth) mucosa (clockwise, starting from the top). (A) The focus of
residual squamous cell carcinoma (between the white and black asterisks) shows no
connection to mucosa. (B) Residual carcinoma is represented by foci of extensive perineural
invasion. The potential reference points to measure the depth of invasion are along
the dorsal, lateral, and ventral mucosa. On this section, depth of invasion was measured
from the lateral mucosa (next to exclamation mark), because this area showed a focus
of moderate-to-severe dysplasia. Hematoxylin and eosin, frozen section, images taken
from the scanned whole slide image with original magnification of ×1.2
Figure 20 (A-C): (A) Two clusters of invasive squamous cell carcinoma (each with about 15 cells, by
the black asterisk and in B and C) were 6.5 mm from the bulk of the tumor, suggestive
of lymphatic invasion and representing the deepest point of invasion. The black line
illustrates the way the distance between the invasive tumor front and remote foci
of carcinoma was measured. The T1 stage was assigned based on the depth of invasion
by the bulk of the tumor which was 4.5 mm. (B) One of the small clusters of carcinoma
is in the left upper corner and the second focus of carcinoma is in the right lower
corner. (C) Hematoxylin and eosin, images taken from the scanned whole slide image
with original magnification of ×1.2
Table 2
T-Primary tumor
|
Stage
|
Status of primary tumor
|
|
TX
|
Primary tumor cannot be assessed
|
|
TO
|
No evidence of primary tumor
|
|
Tis
|
Carcinoma in situ
|
|
Tl
|
Tumor 2 cm or less in greatest dimension
|
|
T2
|
Tumor more than 2 cm but not more than 4 cm in greatest dimension
|
|
T3
|
Tumor more than 4 cm in greatest dimension
|
|
T4a (lip)
|
Tumor invades through cortical bone, inferior alveolar nerve, floor of mouth, or skin
(chin or nose)
|
|
T4a (oral cavity)
|
Tumor invades through cortical bone, into deep/extrinsic muscle of tongue (genioglossus,
hyoglossus, palatoglossus, and styloglossus), maxillary sinus, or skin of face
|
|
T4b (lip and oral cavity)
|
Tumor invades masticator space, pterygoid plates, or skull base; or encases internal
carotid artery
|
Table 3
N — regional lymph nodes
|
Stage
|
Status of regional lymph nodes
|
|
NX
|
Regional lymph nodes cannot be assessed
|
|
N0
|
No regional lymph node metastasis
|
|
N1
|
Metastasis in a single ipsilateral lymph node, 3 cm or less in greatest dimension
|
|
N2
|
Metastasis as specified in N2a, 2b, 2c below
|
|
N2a
|
Metastasis in a single ipsilateral lymph node, more than 3 cm but not more than 6
cm in greatest dimension
|
|
N2b
|
Metastasis in multiple ipsilateral lymph nodes, none more than 6 cm in greatest dimension
|
|
N2c
|
Metastasis in bilateral or contralateral lymph nodes, none more than 6 cm in greatest
dimension
|
|
N3
|
Metastasis in a lymph node more than 6 cm in greatest dimension
|
Table 4
M — Distant metastasis
|
Stage
|
Status of distant metastasis
|
|
MX
|
Distant metastasis cannot be assessed
|
|
M0
|
No distant metastasis
|
|
M1
|
Distant metastasis
|
Table 5
Stage grouping
|
Group
|
Primary tumor
|
Regional lymph nodes
|
Distant metastasis
|
|
Stage 0
|
Tis (in-situ)
|
N0
|
M0
|
|
Stage I
|
T1
|
N0
|
M0
|
|
Stage II
|
T2
|
N0
|
M0
|
|
Stage III
|
T3
|
N0
|
M0
|
|
T1, T2, T3
|
N1
|
M0
|
|
Stage IVa
|
T4a
|
N0, N1
|
M0
|
|
T1, T2, T3, T4a
|
N2
|
M0
|
|
Stage IVb
|
T4b
|
Any N
|
M0
|
|
Any T
|
N3
|
M0
|
|
Stage IVc
|
Any T
|
Any N
|
M1
|
Statistical analysis
Descriptive statistics were reported using numbers and percentages for categorical
variables. Analysis was done using Microsoft Excel 2013, Microsoft Corp., Redmond,
WA, USA and SPSS Statistical Package Version 20.0, IBM Corp., Armonk, New York, USA.
P value (<0.05) was considered statistically significant. The inter-observer agreement
was assessed using Kappa statistics.
Results
This study was undertaken on 30 patients with clinical diagnosis of tongue carcinoma
referred for MR imaging at a tertiary care hospital over the 2-year period. 68% of
the patients belonged to age group of 51–60 years, which was followed by the age group
of 41–50 years comprising of 18% of the patients and 61–70 years comprising 13% of
the patients. The incidence of oral cancers is higher in males constituting 92% of
total patients. There was moderate agreement (k = 0.612) for the T stage between the
clinical and MRI staging assessments [[Table 6]] and fair agreement (k = 0.218) for N stage between MRI and clinical staging assessments
[[Table 7]]. Good (k = 0.822) agreement for the T stage was seen between MRI and histopathology
staging assessments [[Table 8]] and for N stage (k = 0.931) between MRI and histopathology staging assessments
[[Table 9]]. There was good agreement (k = 0.871) for M stage between the clinical and MRI
staging assessments. The agreement for the T stage was poor (k = 0.012) between the
clinical and histopathology staging assessments [[Table 10]]. Agreement for the N stage was poor (k = 0.091) between the clinical and histopathology
staging assessments [[Table 11]]. Mean depth of invasion by histology and MRI was14.22 mm and 16.12 mm, respectively.
Moderate agreement (k = 0.541) was noted between clinical and pathological tumor depth
[[Table 12]] and good agreement (k = 0.844) was noted between radiological and pathological
tumor depth [[Table 13]]. The correlation between depth of invasion reported on MRI and pathologic depth
of invasion (r = 0.93; P < 0.001).
Table 6
Correlation between MRI and clinical tumor (T) staging
|
Clinical "T"staging
|
MRI "T" staging
|
Total
|
|
T1
|
T2
|
T3
|
T4
|
|
By applying the Chi-square test and kappa statistics, P and k come out to be 0.01 and 0.512, respectively, showing moderate agreement between
the clinical and MRI staging assessments
|
|
T1
|
1
|
0
|
0
|
0
|
1
|
|
T2
|
0
|
6
|
2
|
0
|
8
|
|
T3
|
0
|
3
|
6
|
2
|
11
|
|
T4
|
0
|
1
|
4
|
5
|
10
|
|
Total
|
1
|
10
|
12
|
7
|
30
|
|
Sensitivity
|
|
|
90.1%
|
|
|
|
Specificity
|
|
|
93.8%
|
|
|
|
PPV
|
|
|
95.0%
|
|
|
|
Kappa coefficient
|
|
0.612, 95% CI (0.521-1.00)
|
|
Table 7
Correlation between MRI and clinical nodal (N) staging
|
Clinical "N"staging
|
MRI "N"staging
|
Total
|
|
By applying the Chi-square test and kappa statistics, P and k come out to be 0.03 and 0.218, respectively, which shows fair agreement between
the clinical and MRI staging assessments
|
|
NO
|
N1
|
N2
|
|
N0
|
6
|
3
|
4
|
13
|
|
N1
|
0
|
6
|
6
|
12
|
|
N2
|
0
|
0
|
5
|
5
|
|
Total
|
6
|
9
|
15
|
30
|
|
Sensitivity
|
|
93.7%
|
|
|
Specificity
|
|
95.2%
|
|
|
PPV
|
|
93.8%
|
|
|
Kappa coefficient
|
|
0.218, 95% CI (0.347-1.00)
|
|
Table 8
Correlation between MRI and histopathological tumor (T) staging
|
MRI "T"staging
|
HPE "T"staging
|
Total
|
|
By applying the Chi-square test and kappa statistics, P and k come out to be 0.01and 0.822, respectively, which shows good/substantial agreement
between the clinical and MRI staging assessments
|
|
T1
|
T2
|
T3
|
T4
|
|
T1
|
1
|
0
|
0
|
0
|
1
|
|
T2
|
0
|
10
|
0
|
0
|
10
|
|
T3
|
0
|
4
|
3
|
0
|
7
|
|
T4
|
0
|
4
|
4
|
4
|
12
|
|
Total
|
1
|
18
|
7
|
4
|
30
|
|
Sensitivity
|
|
|
94.2%
|
|
|
|
Specificity
|
|
|
96.1%
|
|
|
|
PPV
|
|
|
92%
|
|
|
|
Kappa coefficient
|
|
0.822, 95% CI (0.631-1.00)
|
|
Table 9
Correlation between MRI and histopathological (N) staging
|
MRI "N"staging
|
HPE"N"staging
|
Total
|
|
N0
|
N1
|
N2
|
|
By applying the Chi-square test and kappa statistics, P and k come out to be 0.01 and 0.931, respectively, which shows good agreement between
the clinical and MRI staging assessments
|
|
N0
|
5
|
0
|
0
|
5
|
|
N1
|
5
|
3
|
2
|
10
|
|
N2
|
4
|
3
|
8
|
15
|
|
Total
|
14
|
6
|
10
|
30
|
|
Sensitivity
|
|
94.3%
|
|
|
Specificity
|
|
95.1%
|
|
|
PPV
|
|
93%
|
|
|
Kappa coefficient
|
|
0.931, 95% CI (0.751-1.00)
|
|
Table 10
Correlation between clinical and histopathological tumor (T) staging
|
Clinical "T"staging
|
HPE"T"staging
|
Total
|
|
T1
|
T2
|
T3
|
T4
|
|
By applying the Chi-square test and kappa statistics, P and k come out to be 0.01and 0.012, respectively, which shows poor agreement between
the clinical and MRI staging assessments
|
|
T1
|
4
|
0
|
0
|
0
|
4
|
|
T2
|
0
|
9
|
2
|
0
|
11
|
|
T3
|
0
|
5
|
1
|
2
|
8
|
|
T4
|
0
|
3
|
1
|
3
|
7
|
|
Total
|
4
|
17
|
4
|
5
|
30
|
Table 11
Correlation between clinical and histopathological (N) staging
|
Clinical "N"staging
|
HPE"N"staging
|
Total
|
|
N0
|
N1
|
N2
|
|
By applying the Chi-square test and kappa statistics, P and k comeout to be 0.01 and 0.091, respectively, which shows poor agreement between
the clinical and MRI staging assessments
|
|
N0
|
6
|
0
|
2
|
8
|
|
N1
|
6
|
5
|
4
|
15
|
|
N2
|
3
|
0
|
4
|
7
|
|
Total
|
15
|
5
|
10
|
30
|
Table 12
Sensitivity and specificity of clinical depth in comparison to pathological depth
|
Clinical depth (mm)
|
Pathological depth (mm)
|
|
<5
|
>5
|
Total
|
|
Kappa coefficients were used to determine the agreement between measures once categorized
according to the cutoff point. Closer values to 1 mean higher agreement between categories.
|
|
<5
|
11
|
3
|
14
|
|
>5
|
5
|
11
|
16
|
|
Total
|
16
|
14
|
30
|
|
Sensitivity
|
-
|
70%
|
|
Specificity
|
-
|
78.5%
|
|
PPV
|
-
|
70%
|
|
Kappa coefficient
|
-
|
0.541, 95% CI (0.327-0.807)
|
Table 13
Sensitivity and specificity of radiological depth in comparison to pathological depth
|
Radiological depth (mm)
|
Pathological depth (mm)
|
|
<5
|
>5
|
Total
|
|
Kappa coefficients were used to determine the agreement between measures once categorized
according to the cutoff point. Closer values to 1 mean higher agreement between categories
|
|
<5
|
14
|
1
|
15
|
|
>5
|
2
|
13
|
15
|
|
Total
|
16
|
14
|
30
|
|
Sensitivity
|
-
|
90%
|
|
Specificity
|
-
|
92.8%
|
|
PPV
|
-
|
90%
|
|
Kappa coefficient
|
-
|
0.844, 95% CI (0.563-1.00)
|
Cutoff values for histopathological (HP) depth and MRI depth
The cutoff value of HP depth that could determine the existence of nodal metastasis
was 8 mm. The cutoff value for T1WGd MRI depth was 5 mm. With the HP depth cutoff
value of 5 mm as a standard, groups were subdivided into those >5 mm and those <5
mm; the nodal metastasis rates for each group were 52% and 24%, respectively (P = 0.040).
Correlation between histopathological (HP) depth and MRI depth
Pearson’s correlation coefficient of HP depth and T1WGd MRI depth was 0.851 (P < 0.001) suggesting that HP depth shows a strong correlation with T1WGd MRI depth.
Discussion
In the current study, the extent of primary tumor (T) and metastasis to regional lymph
nodes (N) was initially evaluated by clinical examinations followed by MR imaging.
The final diagnosis was made by histopathological examination (HPE). Kappa Index was
used for data analysis which showed moderate agreement (kappa value 0.512) between
the clinical and MRI “T” staging. This is consistent with the studies performed by
Paiva et al.[[20]] and Hirunpat et al.[[21]] which also showed that mis-staging by clinical examination in the overall stage
grouping was high. Also, there was good agreement (kappa value 0.822) for the T staging
(tumor depth and width) between MRI and HPE assessments. The final staging assessed
by MR imaging in the current study remains the same in 30 patients who underwent surgery
and final staging by HPE. These results are consistent with the study conducted by
Tetsumura et al.[[22]] in which the tumor depth and width were measured on both MR images and HPE and
the authors observed a high correlation between the values measured by MRI and HPE.
In this study, clinical examination and MRI were both adequate at determining depth
of invasion compared with final pathology when tumors were≥5 mm in depth, but not
for those less than 5 mm. We used 5 mm as a cutoff as this is the depth at which the
risk of nodal metastases increases, based on the literature.[[13], [23]] Since the clinical importance is to be able to detect deeper tumors, the decreased
ability of either examination to be able to accurately predict the depth of superficial
lesions is less clinically significant.
There have been previous studies investigating the accuracy of MRI in predicting the
depth of invasion of oral tongue SCC; however, these studies primarily have a small
sample size and retrospective study design, and none have compared MRI with clinical
examination. Preda et al. investigated 33 oral tongue SCC in a retrospective series.[[24]] The authors demonstrated that MRI thicknesses correlated strongly with histological
tumor thicknesses (correlation coefficient = 0.68, P < 0.0001). Park et al.[[25]] evaluated 114 patients with oral cavity and oropharyngeal SCC of which 49 patients
had oral tongue SCC. Relationship between MRI and histologic depth of invasion in
oral tongue subsite was high with a correlation coefficient of 0.949. In the current
study, the mean depth of invasion by histology and MRI was 14.2 mm and 16.1 mm, respectively.
This group reported on deeper tumors, explaining the better correlation.
As pointed out by Lwin et al.,[[26]] there is tumor shrinkage after resection affecting all oral cavity subsites, including
the oral tongue. The tumor shrinkage factor for oral tongue cancer has been reported
to be 87%. Most of the studies assessing the relationship between tumor depth of invasion
and risk of nodal metastases are based on pathologic assessment and not clinical or
radiographic assessment. Therefore, clinical and MRI examination may under or over
estimate depth of invasion and may not have the same ability to predict nodal metastases.
Sentinel lymph node (SLN) biopsy has been evaluated in recent years in head and neck
cancer. A few studies evaluated SLN for oral and oropharyngeal cancer; however, most
of these studies included advanced T stage and did not study specific subsite.[[27], [28], [29]] Sagheb et al. did a pilot study to examine the role of SLN in early T stage tongue SCC with N0
neck. A SLN was followed by a neck dissection during the same operation.[[30]] It was concluded that the sensitivity of SLN is about 75% and further investigation
is needed.
While MRI was shown to correlate well with pathological depth and is more sensitive
and specific for depth measurements than clinical assessment, the latter test is complementary
and useful in situ ations where either MRI is unavailable or difficult to interpret
due to artefacts. In a prospective study, Yuen et al.[[31]] examined the correlation between ultrasound and pathologic tumor thickness in 45
oral tongue carcinoma patients during general anesthesia and before commencing surgery.
There was a statistically significant correlation coefficient of 0.940 (P <.005). While this technique may be difficult to perform in clinic due to pain or
trismus, its improved ability to measure tumor thickness does warrant further investigation.
Despite the importance of depth of invasion, other histopathological parameters have
been found to correlate with nodal metastasis including size of the tumor in greatest
dimension, and other pathologic features such as pattern of invasion, density of cancer-associated
fibroblasts, and perineural and vascular invasion.[[32]] All these need to be taken in account to determine the risk of regional metastasis.
Multiple pulse sequences had been used in the previous works to detect small tongue
carcinoma and accurately identify tumor margins, including T2WI, STIR, and T1-weighted
fat-suppressed contrast-enhanced sequences. Lam et al.[[33]] reported that particularly contrast-enhanced T1-weighted MRI provides a satisfactory
accurate correlation between MRI tumor thickness and histologic tumor thickness in
oral tongue cancer. Background diffusion-weighted imaging obtained with magnetic resonance
(DW-MRI) is a noninvasive imaging tool potentially able to provide information about
micro-structure tumor characteristics. The inclusion of DWI/ADC values might be helpful
for differentiation between true tumor margin and edema, and also for distinction
between benign and malignant head and neck tumors. Multiple studies reported high
diagnostic accuracy of DWI for differentiation of malignant from benign status of
metastatic cervical lymph nodes.[[34]]
There are several studies[[35]] which tested the reliability of MRI in measuring tongue tumor thickness and correlated
it well with histologic tumor thickness. Spiro et al.[[36]] postulated that disease-related death is apparently unusual when oral tumors are
thin, regardless of tumor stage, and that tumor thickness rather than stage may have
the best correlation with treatment failure and survival. However, tongue carcinoma
may vary in shape and growth pattern. Therefore, depth of invasion (represented by
para-lingual distance), not merely tumor thickness, is another important prognostic
factor.
The current study evaluated the clinical assessment of tumor thickness in comparison
to radiographic interpretation. There are strong correlations between pathological,
radiological, and clinical measurements. Specifically, for oral tongue, cut-off of
5mm has been suggested. Finally, just where to measure DOI from can be difficult to
determine in oral tongue (with mucosa on dorsal, lateral, and ventral aspects) and
in undulating hyperplastic epithelium, which can create an uneven basement membrane.
One has to imagine an arcuate reference line and then drop a “plumb-line” which can
be equally as difficult due to variations in normal mucosa and DOI at different tumor
section. The study highlighted the potential impact on T staging of extratumoral foci
of SCC due to perineural invasion. It must be noted that a large proportion of extratumoral
NI or LI occurs in tumors that are already T3, thus diminishing their impact on staging.
Extratumoral perineural invasion represents a challenge to DOI measurement in isolated
cases only. These scenarios are not currently directly addressed in the AJCC 8th edition description of DOI. However, they are covered under a more general TNM principle:
when in doubt, the less advanced attribute should be selected (i.e., smaller DOI measurement,
not including the extratumoral perineural invasion).
The oral tongue is covered by mucosa on its dorsal, lateral, and ventral aspects and
a simple “plumb line” method may be difficult to apply in some cases. When residual
carcinoma is small and not connected to the mucosal surface, the reference point from
which to measure the DOI is perhaps best represented by mucosa with squamous dysplasia.
In oral tongue, the level of the basement membrane of the closest adjacent normal
mucosa is probably better represented by an arcuaterather than a straight line, especially
when the line is drawn through two points, i.e., normal mucosa on both sides of carcinoma.
The current study showed that in up to 12% of apparently T2 cases, DOI may be underestimated
due to the positive deep margin. Rarely, extratumoral perineural invasion may be the
deepest point of invasion, but it is unlikely to affect T stage. DOI measurement for
early SCC of the oral tongue may require re-examination of the diagnostic biopsy slides
in up to 20% of cases due to the absence or only minimal residual carcinoma in glossectomy
specimens. A proactive assessment and reporting of DOI on diagnostic biopsies or documentation
of factors limiting DOI measurement (e.g., fragmentation, lack of normal mucosa, absence
of intrinsic tongue musculature) may minimize the need to re-review the original diagnostic
biopsy when the glossectomy reveals no or minimal residual carcinoma.
Conclusion
MRI is the imaging modality of choice for evaluation of tongue carcinoma as MRI helps
in the accurate staging of the tumor using TNM classification which is crucial for
optimizing treatment options. The current study shows a high correlation between MRI
and HPE findings regarding thickness of tumor and depth of invasion. MRI and histopathology
assessments of tumor spread were equivalent to within 0.5 mm DOI. In conclusion, estimation
of invasion depth using MRI as a preoperative study in oral tongue carcinoma is essential
in planning surgical treatment strategies such as the extent of elective neck dissection.
Invasion depth, which greatly affects occult node metastases, must be included in
the TNM staging of oral tongue carcinoma.
Limitations of the study
The limitations of our study include a relatively small number of cases and errors
caused by manual measurement of tumor thickness during clinical examination.
Declaration of patient consent
The authors certify that they have obtained all appropriate patient consent forms.
In the form, the patients have given their consent for their images and other clinical
information to be reported in the journal. The patients understand that their names
and initials will not be published and due efforts will be made to conceal their identity,
but anonymity cannot be guaranteed.
