Download Diffusion-weighted MRI in cervical lymph nodes: Differentiation

European Journal of Radiology 77 (2011) 281–286
Contents lists available at ScienceDirect
European Journal of Radiology
journal homepage: www.elsevier.com/locate/ejrad
Diffusion-weighted MRI in cervical lymph nodes: Differentiation
between benign and malignant lesions
Anna Perrone a , Pietro Guerrisi a , Luciano Izzo b , Ilaria D’Angeli c,∗ , Simona Sassi a ,
Luigi Lo Mele a , Marina Marini a , Dario Mazza a,1 , Mario Marini a
a
b
c
Department of Radiological Sciences, “Sapienza” University, Policlinico Umberto I, Rome, Italy
Department of General Surgery “Pietro Valdoni”, “Sapienza” University, policlinico Umberto I, Rome, Italy
Department of Heart and Great Vessels “Attilio Reale”, “Sapienza” University, Policlinico Umberto I, Rome, Italy
a r t i c l e
i n f o
Article history:
Received 24 March 2009
Received in revised form 28 July 2009
Accepted 31 July 2009
Keywords:
MRI
Lymph nodes
Malignant lesions
a b s t r a c t
Objective: Purpose of our study was to assess the potential role of diffusion-weighted imaging (DWI) in
the differential diagnosis between benign and malignant nodes.
Subject and methods: We enrolled 32 subjects: 14 with benign lymphadenopathy, 17 patients with histologically proved malignant disease before beginning treatment and 1 patient with lymphoma after
chemotherapeutic treatment.
In all patients we used fast spin echo T2-weighted images in axial and coronal planes, fast spin echo
T1-weighted images before and after contrast medium of administration in axial and coronal planes.
Before contrast administration diffusion sequences were acquired on the axial and coronal plane (b
factor of 0.500 and 1000 s/mm2 ) and then apparent diffusion coefficient (ADC) maps were reconstructed.
Results: On diffusion images, 13/14 patients with benign nodes showed low signal intensity and had high
signal on ADC maps, whereas all patients with malignant diseases appeared hyperintense on diffusion
images and with low signal intensity on ADC maps. Only a patient with tuberculosis showed a low ADC
value. The mean ADC value of malignant nodes was about 0.85 × 10−3 mm2 /s, the mean value of benign
nodes was 1.448 × 10−3 mm2 /s; this difference was statistically significant (p < 0.01). The mean ADC value
of treated nodes was 1.75 × 10−3 mm2 /s. The best threshold value was 1.03 × 10−3 mm2 /s, obtaining a
sensitivity of 100% and a specificity of 92.9%.
Conclusions: Diffusion imaging could be considered an important supportive tool for the diagnosis of
enlarged cervical lymphadenopathies.
© 2009 Elsevier Ireland Ltd. All rights reserved.
1. Introduction
The detection of cervical nodes metastasis is very important for
the prognosis and the treatment of head and neck tumours.
Up to today parameters used by conventional imaging techniques are shape, size, extracapsular spread and an abnormal inner
architecture. The size is certainly the most used criterion for the
diagnosis, whereas the presence of central necrosis is the most
reliable sign of malignity [1].
Nevertheless several reports showed that these parameters are
not enough to discriminate benign from malignant lesions [2,3].
∗ Corresponding author at: Department of Heart and Great Vessels “Attilio Reale”,
La Sapienza University of Rome, Viale del Policlinico 155, 00161 Rome, Italy.
Tel.: +39 339 4761533.
E-mail address: [email protected] (I. D’Angeli).
1
Free professional dentistry.
0720-048X/$ – see front matter © 2009 Elsevier Ireland Ltd. All rights reserved.
doi:10.1016/j.ejrad.2009.07.039
Diffusion-weighted MR imaging (DWI) analyzes intercellular
water motion: every change in the water protons movements
produces a variation of signal intensity in diffusion-weighted
sequences and as a consequence on ADC maps [4].
Purpose of our study was to assess the potential role of DWI in
the differential diagnosis between benign and malignant nodes.
2. Materials and methods
2.1. Patients
Our study was performed on 32 patients with enlarged neck
nodes who underwent Magnetic Resonance exam from November
2007 to March 2009.
The study cohort included 14 patients (8 males and 6 females;
mean age: 57 ± 10.8 years; range: 24–80 years) with benign lymphadenopathy (1 patient with tuberculosis, 3 with abscesses, 10
with a non-specific lymphadenitis), and 17 patients (10 males
and 7 females; mean age: 63 ± 12.8 years; range: 15–85 years)
282
A. Perrone et al. / European Journal of Radiology 77 (2011) 281–286
with histologically proved malignant disease before beginning
treatment (7 patients with lymphoma, 9 with squamous cell carcinoma and one patient with rabdomyosarcoma). Moreover we
studied a patient with Hodgkin lymphoma after chemotherapeutic
treatment.
Written informed consent was obtained from all patients.
2.2. MR technique and image analysis
All the exam were performed with a 1.5-T superconductive
scanner (Avanto, Siemens Medical Systems, Enlargen, Germany)
using a standard head–neck coil.
In all patients the following protocol was used:
- Fast spin echo (FSE) T2-weighted images (TR 5722 ms, TE 95 ms,
slice-thickness: 3 mm) in axial plane;
- Fast spin echo (FSE) T2-weighted images (TR 3850 ms, TE 75 ms,
slice-thickness: 5 mm), in coronal plane;
- Fast spin echo (FSE) T1-weighted images, with and without fat
suppression (TR 708 ms, TE 75 ms, slice-thickness: 3 mm) in axial
plane;
- Fast spin echo (FSE) T1-weighted images, with fat saturation after
contrast medium of administration (Omniscan, GE Healthcare);
in axial and coronal planes.
in all three orthogonal planes (X, Y, Z) with a b factor of 0.500 and
1000 s/mm2 per axis in each patient. The ADC value was automatically reconstructed by a standard software imager in the main
console. The whole-node ADC value was obtained drawing a region
of interest (ROI) covering all the pathologic node in all sections in
which it was present and averaging the results. In the study we
chose only the largest abnormal adenopathies and excluded from
analysis the necrotic areas.
2.3. Statistical analysis
Two experienced radiologists analyzed the results obtained
independently; disagreements regarding image findings were
resolved with a mutual accord. For each patient, both in the benign
group and in the malignant group, the average of ADC values of
examined nodes was calculated. Statistical analysis was done by
using the SPSS (“Statistical Package for Social Science”) program
software on the 21 untreated patients. We used Student’s t-test to
compare between two groups and a value of p < 0.05 was considered
significant. Subsequently we used a receiver operating characteristic (ROC) curve to evaluate diagnostic capability of ADC value and
to determine the cut-off value for differentiating malignant from
benign nodes.
3. Results
Before contrast administration Single-Shot Echo-planar diffusion sensitized sequences (DWI) (TR 8500, TE 99, Matrix 192 × 96,
slice-thickness 5 mm, bandwidth1582) were acquired on the axial
and coronal plane. The diffusion-sensitizing gradients were applied
On DWI images in 13/14 cases with inflammatory diseases
lymph nodes showed low signal intensity (b = 1000), whereas on
the ADC maps it presented high signal (Fig. 1).
Fig. 1. Patient with lymphadenitis. (A). Axial FSE-T2-weighted image showing an enlarged cervical node on the left side. (B) On axial diffusion-weighted MR image at
b = 0 s/mm2 the node shows high signal intensity while on diffusion-weighted MR images (C) at b = 1000 s/mm2 the same node is hypointense. (D) On ADC map the lymph
nodes is hyperintense with an ADC value = 1.52 × 10−3 mm2 /s.
A. Perrone et al. / European Journal of Radiology 77 (2011) 281–286
283
Fig. 2. Patient with tuberculosis. (A) Coronal FSE-T2-weighted image showing multiple enlarged neck nodes on the left side. (B) Diffusion-weighted image at b = 0 s/mm2
shows that nodes have high SI. (C) In the diffusion image at b = 1000 s/mm2 the same nodes exhibit high signal intensity. (D) On ADC map the lymph nodes show low SI; the
mean ADC value of the lymph node is 0.91 × 10−3 mm2 /s.
Only in the patient with tuberculosis, the ADC value was low as
for the malignant group (Fig. 2). In fact the cases with malignant
disease appeared hyperintense on diffusion images (b = 1000) and
with low signal intensity on ADC maps (Figs. 3 and 4). The mean
ADC value of metastatic and lymphomatous nodes was about
0.85 × 10−3 mm2 /s (range: 0.581 × 10−3 –1.03 × 10−3 mm2 /s),
lower than the mean value of benign nodes (1.448 × 10−3 mm2 /s,
range: 0.91 × 10−3 –2.246 × 10−3 mm2 /s); this difference was
statistically significative (p < 0.01) with t = 3.7497 (Table 1).
In the only patient treated with CHT lymph nodes were
hypointense on DWI (b = 1000) and hyperintense on the ADC maps
with a mean value of 1.75 × 10−3 mm2 /s (Fig. 5). The best threshold value for differentiating malignant from benign nodes was
1.03 × 10−3 mm2 /s, obtaining a sensitivity of 100% and a specificity
of 92.9% (Fig. 6). Fig. 7 shows the receiver operating characteris-
Table 1
Histogram relative to the distribution of ADC values among the examined
patients.
tic (ROC) curve of the ADC value used for differentiating benign
from malignant lymph nodes. The area under the curve was
0.983.
4. Discussion
The evaluation of cervical nodes is important not only for diagnosis and staging of malignant diseases, but also for planning
treatment and follow-up.
Even if ultrasound image, contrast-enhanced computed tomographic and contrast-enhanced MRI allow the detection of enlarged
cervical lymphodenopathies, none of these methods reaches the
ideal accuracy [5,6].
US guided fine needle aspiration biopsy (US-FNAB) of lymph
nodes has been shown to be an accurate method but it is an
invasive and operator-dependent exam with a high incidence of
false-negative cases [7,8].
Moreover these imaging methods use standard parameters
(shape, size, internal architecture, extranodal diffusion and vascular features) that showed to be scarsely reliable [2,3].
SPECT (single photon emission CT) and PET (photon emission
tomography) are new image techniques which supply functional
information (blood flow and glucidic metabolism) but they are
invasive (exposure to radiations), expensive, low available and with
a relatively low spatial resolution [9–11].
Recently diffusion-weighted imaging with Magnetic Resonance
was introduced which could improve the diagnostic accuracy in the
differential diagnosis between benign and malignant nodes [12,13].
Magnetic Resonance with diffusion-weighted imaging is a noninvasive technique that measures the motion of water in the
284
A. Perrone et al. / European Journal of Radiology 77 (2011) 281–286
Fig. 3. Patient with Hodgkin lymphoma. (A) Axial T2-weighted image showing multiple enlarged cervical nodes on the right side that present homogenous high SI, except
for a small necrotic area excluded from measurements. (B and C) Both on diffusion-weighted images at b = 0 and at 1000 s/mm2 one, the lymph nodes are hyperintense. (D)
ADC map shows hypointensity of lymphomatous nodes with a value of 0.62 × 10−3 mm2 /s. It can see that signal intensity of necrotic area is opposite to nodes.
Fig. 4. Patient with carcinoma of the tongue (arrow). (A) Axial T2-weighted image shows an enlarged node on the right side. (B) Diffusion-weighted image at b = 0 s/mm2
shows that node is hyperintense as well as in the (C) diffusion image at b = 1000 s/mm2 . (D) On ADC map the node appears hypointense with an ADC value of 0.71 × 10−3 mm2 /s.
A. Perrone et al. / European Journal of Radiology 77 (2011) 281–286
285
Fig. 5. Patient with nasopharingeal carcinoma. (A) Axial T2-weighted image shows multiple enlarged nodes on the both sided. (B) Diffusion-weighted image at b = 0 s/mm2
shows that nodes are hyperintense as well as in the (C) diffusion image at b = 1000 s/mm2 . (D) On ADC map the node shows low signal intensity with an ADC value of
0.76 × 10−3 mm2 /s.
extracellular space. As showed in several studies, metastatic nodes
present a reduction of diffusivity, which can be attributed to a
hypercellularity, to an increased nuclear-to-cytoplasmatic ratio
and to perfusion [13].
Above all, in the cases of lymphoma the increate cellularity and
the reduced extracellular space have an important role. However
this restriction in diffusion in metastatic nodes is represented as an
area of hyperintensity on diffusion images with a low value of ADC.
Only few authors have examined characterization of head and
neck lesions with diffusion-weighted MR imaging [13–17].
In our series we considered 14 subjects with benign diseases
who showed enlarged cervical nodes and 17 patients with known
malignant lymph nodes (metastasis or lymphoma) before treatment.
The evaluation with DWI showed that metastatic and lymphomatous nodes appeared hyperintense (b = 1000 mm2 /s) and
hypointense on ADC maps; adversely inflammatory nodes were
hypointense (b = 1000 mm/s) and hyperintense on ADC maps. So
mean ADC value for malignant lesions, that was 0.85 × 10−3 mm2 /s,
resulted lower than benign ones, that was 1.448 × 10−3 mm2 /s
(t = 3.7497 per p < 0.01). The best ADC threshold value for distinguishing benign from malignant nodes was 1.03 × 10−3 mm2 /s,
with a sensitivity of 100% and a specificity of 92.9%. This value
Fig. 6. The mean ADCs of benign and malignant nodes are compared. The horizontal
line is our threshold value (1.03 × 10−3 mm2 /s): the ADCs of benign nodes are significantly higher than those of malignant nodes, except for patient with tuberculosis.
Fig. 7. Receiver operating characteristic (ROC) curve of the ADC value used for
differentiating benign from malignant lymph nodes. The area under the curve is
0.983.
286
A. Perrone et al. / European Journal of Radiology 77 (2011) 281–286
of specificity is due to a case of benign lymphadenopathy
which showed a lower value than cut-off one: in tuberculosis
(0.934 × 10−3 mm2 /s) this restriction on DWI can be due to the presence of inflammatory cells in the pus that impede the motion of
water molecules.
Many author analyzed the capability of DWI to differentiate
between several causes of lymph nodes. Both Chang et al. [18] and
Wang et al. [13] reported a threshold value of 1.22 × 10−3 mm2 /s,
with a sensitivity of 91% and a specificity of 93%, accordingly with
our results. Razek et al. [14] in a recent study reported metastatic
mean ADC value lower than benign ones, with a cut-off value of
1.38 × 10−3 mm2 /s, obtaining a sensitivity of 98% and a specificity
of 88%.
Our data are not in agreement with Sumi et al. [12,15]
who found, for metastatic nodes, significantly higher mean
ADC value (1.167 ± 0.447 × 10−3 mm2 /s) than flogistic lymphadenopathies (0.652 ± 0.101 × 10−3 mm2 /s) and than lymphomatous ones (0.601 ± 0.427 × 10−3 mm2 /s).
Differences among these studies can be attributed to several
causes. Above all, the choice of the b values: a lower b values
increase signal-to-noise ratio but makes worsen the sensitivity to
diffusion. Other factors are the selection of the region of interest
on ADC maps and the use of sequences which reduce the artefacts
in order to make more precise the measurement of the interested
area.
4.1. Limits
Our study has some limits such as a small study cohort. In fact,
our statistical tests were performed on the number of the patients
rather than the number of involved nodes, in order to avoid bias or
confounding effects. Moreover the patients with lymphoma were
not treated surgically, but with radio and/or chemotherapy so the
malignancy of nodes was assessed by imaging criteria.
Another limitation was represented by the low spatial resolution of echoplanar imaging that worsens using high b values
(b = 1000 s/mm2 ), that on the other hand are necessary to improve
the sensitivity of diffusion imaging. For this reason the small nodes
with a diameter inferior to 9 mm are difficult to detect on ADC maps.
5. Conclusions
We found a significant difference between benign and
malignant cervical nodes on diffusion-weighted imaging and
on ADC maps, identifying a threshold ADC value equal to
1.03 × 10−3 mm2 /s. This value, at our opinion, could be used not
only in a pre-treatment phase, but also after therapy to detect recurrent disease or as sign of improvement. That is demonstrated by
our most important data: the case of post-chemotherapy Hodgkin
lymphoma with hypointense nodes on diffusion sequences and
hyperintensity on ADC maps, indicative for improvement.
In summary, diffusion imaging could be considered an important supportive tool for the diagnosis of enlarged cervical
lymphadenopathies.
References
[1] King AD, Tse GM, Ahuja AT, et al. Necrosis in metastatic neck nodes: diagnostic
accuracy of CT, MR imaging, and US. Radiology 2004;230:720–6.
[2] Van den Brekel MW, Castelijns JA, Snow GB. The size of lymph nodes in the
neck on sonograms as a radiologic criteria for metastasis: how reliable is it?
AJNR Am J Neuroradiol 1998;19(4):695–700.
[3] Van den Brekel MW, Castelijns JA, Snow GB. Detection of lymph node metastases in the neck, radiologic criteria. Radiology 1994;192(3):617–8.
[4] Rowley H, Grant E, Roberts T. Diffusion MR imaging. Theory and application.
Neuroimaging Clin North Am 1999;9:343–61.
[5] Curtin HD, Ishwaran H, Mancouso AA, et al. Comparison of CT and MR imaging
in staging of neck metastsis. Radiology 1998;207:123–30.
[6] Esen G. Ultrasound of superficial lymph nodes. Eur J Radiol 2006;58:345–59.
[7] Van den Breckel MWM, Castelijns JA, Stel HV, et al. Occult metastatic neck
disease: detection with US and US-guided fineneedle aspiration cytology. Radiology 1991;180:457–61.
[8] Van den Breckel MWM, Castelijns JA, Stel HV, et al. Modern imaging techniques and ultrasound guided aspiration cytology for the assessment of neck
node metastases: a prospective comparative study. Eur Arch Otorhinolaryngol
1993;250:11–7.
[9] Yamazaki Y, Saitoh M, Notani K, et al. Assessment of cervical lymph node metastases using FDG-PET in patients with head and neck cancer. Am Nucl Med
2008;22:177–84.
[10] Jeong HS, Baek CH, Son YI, et al. Use of integrated (18)F-FDG PET/CT to improve
the accuracy of initial cervical nodal evaluation in patients with head and neck
squamous cell carcinoma. Head Neck 2007;29(3):203–10.
[11] Kotani J, Kawabe J, Higashiyama S, et al. Evaluation of diagnostic abilities of
Ga-SPECT for head and neck lesions. Ann Nucl Med 2008;22(4):297–300.
[12] Sumi M, Van Cauteren M, Nakamura T. MR microimaging of benign and malignant nodes in the neck. Am J Radiol 2006;186:749–57.
[13] Wang J, Takashima S, Takayama F, et al. Head and neck lesions: characterization with diffusion/weighted eco planar MR imaging. Radiology 2001;220:
621–30.
[14] Razek AAKA, Soliman NY, Elkhamary S, Alsharaway MK, Tawfik A. Role of
diffusion-weighted MR imaging in cervical lymphadenopathy. Eur Radiol
2006;16:1468–77.
[15] Sumi M, Sakihama N, Sumi T, et al. Discrimination of metastatic cervical lymph
nodes with diffusion-weighted MR imaging in patients with head and neck
cancer. Am J Neuroradiol 2003;24:1627–34.
[16] Herneth AM, Czerny C, Krestan C. Role of diffusion weighted MRI in the characterization of lymph node metastasis. In: XVI International Congress of Head
and Neck Radiology. 2003.
[17] King AD, Ahuja AT, Yeung DK, et al. Malignant cervical lymphadenopathy: diagnostic accuracy of diffusion-weighted MR imaging. Radiology
2007;245(3):806–13.
[18] Chang VHF, Fan YF, Khoo JBK. MRI features of cervical nodal necrosis in
metastatic diseases. Clin Radiol 1996;51:103–9.