Characterization of cervical lymph nodes using DCE-MRI: Differentiation between metastases from SCC of head and neck and benign lymph nodes

Abstract

PURPOSES: To compare the characteristics of DCE-MRI between metastatic and benign lymph nodes (LNs). To explore the effect of sizes of LNs on the value of K^trans, Ve and iAUC.

MATERIALS AND METHODS: Metastatic and benign LNs were retrospectively analyzed. Chi-square test and two sample t-test were applied to TICs and parameters. One factor covariance analysis was used to explore the effect of diameters on the value of K^trans, Ve and iAUC.

RESULTS: The value of K^trans, Ve and iAUC of metastatic LNs were statistically higher (all p < 0.001) than those of benign LNs. The short-axis diameters had no influence on parameters.

CONCLUSIONS: DCE-MRI is useful to differentiate the metastatic LNs from benign LNs. Sizes of LNs has no effect on parameters.

Keywords

Dynamic contrast-enhanced magnetic resonance imaging lymph node head and neck squamous cell carcinoma

1 Introduction

Head and neck cancers, with a less than 59% 5-year survival rate [13, 18], account for more than 3.2% of all malignancies, according to a report from the American Cancer Society in 2010 [24]. For the patients with head and neck cancers, lymph node (LN) metastases are independent factors that can affect their survival rates, their distant metastases, distant recurrences and nodal recurrences, and LN metastases can reduce the patients’ survival rates by 11–35% [19 , 28]. LN status is also important in the planning of neck dissection and treatment [11, 22].

Clinical palpation is of very low sensitivity (<72%) [9, 11]. Ultrasound and cross section imaging have been severed as principal methods in the detection of lymphoid metastases. They provide morphological information that is helpful for diagnoses, such as LNs’ sizes, shapes, extra capsular invasions, necrosis and peripheral vascularization [4 , 32]. The specificity of the combination of these morphological assessments is 91–100%. However, the false negative rates ranges from 15% to 26% [5 , 31], which means that the cancer could probably recur if the “non-metastatic” LNs are left untreated. There are some efforts using sentinel LN biopsy to confirm the status of LNs before operations, but the complexities of the lymphatic drainages limit the application of this method [1]. During recent decades, functional and perfusion imaging has developed into a supplement of routine imaging and provides more helpful information [10 , 32]. Diffusion weighted MRI is one of the modalities that offers complementary functional information to anatomical imaging on diagnoses of LNs [2]. However, the lack of the sequence standardization limits the application of the technique. Positron emission tomography imaging is another functional imaging protocol with the highest accuracy (89.8%) and the lowest false negative rate (4.7%) [10]. But the limited availability and the high-cost of PET instruments make its widespread application difficult.

Dynamic contrast-enhanced MR imaging (DCE-MRI) provides information on the physiology of the microcirculation [17 , 29]. Quantitative parameters, such as volume transfer constant (K^trans), extravascular extracellular volume fraction (Ve) and initial area under the curve (iAUC), are known to be accurate in the analysis of the tissue microcirculation [14 , 23]. Studies using DCE-MRI have been reported on the diagnostic and prognostic assessments of head and neck cancers, and on the evaluations of the grading of glioma [3, 12]. We hypothesize that DCE-MRI is reliable in the diagnoses of metastatic LNs of head and neck cancers. The purposes of our study are to compare 1) the patterns of time-intensity curves (TICs) of metastatic and benign LNs; 2) the K^trans, Ve and iAUC of metastatic and benign LNs and their diagnostic ability; 3) the effects of LNs’ sizes on the K^trans, Ve and iAUC.

2 Materials and methods

2.1 Patients selection

The study was approved by the ethics committee of the hospital. Written informed consents had been signed up by all patients before the MR examination. No contraindications for MRI and the application of contrast medium were present.

126 LNs collected from Jun. 2011 to Sep. 2014 were retrospectively analyzed. 64 of these LNs were metastatic lesions collected from 19 cases of SCC, and 62 were benign LNs from 32 controls. Diagnoses and locations of primary lesions of controls and patients were presented in Table 1. Tumor Lymph Node Metastasis (TNM) Classification of patients with SCC were listed in Table 2. All of the metastatic LNs were pathologically confirmed by lymph node excisions or fine needle biopsies. The matching between pathology and MR images were based on locations, prominent anatomical structures, sizes and shapes of the LNs, and histologic findings (necrosis, extranodal invasions and peripheral vascularization). Among the benign LNs, 56 LNs of 30 controls were clinically diagnosed. None of the controls suffered from known malignant lesions in other organs by the time of the examination and after 4–13 months’ follow-up. The clinically diagnosed benign LNs were defined by a set of criteria based on morphological images [8 , 31]: 1) a short-axis diameter smaller than 10 mm along anterior cervical chain and spinal accessory chain, and smaller than 15 mm in submandibular region; 2) the absence of necrosis, which presents as patchy increased signal intensity on T2 weighted images and decreased signal intensity on T1 weighted images; 3) the absence of extranodal invasions and obviously enhanced rings around LNs and 4) solitary LNs. Among the benign LNs, 6 of them collected from 2 patients suffering from squamous cell carcinoma were pathologically proven to be benign. Spatial resolution of MR limited the match between imaging and pathology for benign LNs. Based on experience and previous study [6], we excluded the LNs smaller than 4 mm in short-axis diameters. The morphological images were analyzed independently by two radiologists (have been working in this field for 2 years and 5 years, respectively). In cases of discrepancy, discussions and consultations with more experienced doctors (have been working as radiologists for 6 years to 15 years) would be performed.

All of the cases underwent routine MR and DCE-MR scans in the week before operations, biopsies and treatment.

2.2 Imaging acquisition

MRI and DCE-MRI were performed on a 3.0 Tesla unit (TRIO TIM, Siemens AG, Erlangen, Germany) with an 8 channel head coil and an 8 channel neck coil. Standard axial T1-weighted turbo spin echo (TSE) (TR = 500 ms, TE = 8.2 ms, slice thickness 2.5 mm), axial T2-weighted TSE (TR = 6000 ms, TE = 94.0 ms, slice thickness 2.5 mm), coronal T1-weighted TSE (TR = 600 ms, TE = 8.2 ms, slice thickness 3.0 mm), coronal T2-weighted TSE (TR = 4000 ms, TE = 84.0 ms, slice thickness 3 mm) and axial or sagittal T2-weighted fat-saturation sequences (TR = 4000 ms, TE = 84.0 ms, slice thickness = 2.5 mm for axial sequence and 3.0 mm for sagittal sequence, slice gap = 0.5 or 0.6 mm) without enhancement were acquired before the contrast medium injection. Coronal T1 maps were acquired with volumetric interpolated breath hold (VIBE) sequence (TR = 5.29 ms, TE = 1.81 ms, flip angle 2 and 15 degrees, slice thickness 3 mm, slice gap 0.6 mm). Coronal dynamic contrast enhanced images were acquired by time-resolved angiography with interleaved stochastic trajectories (TWIST) sequence (TR = 4.9 ms, TE = 2.0 ms, flip angle 12 degrees, slice thickness 3 mm, slice gap 0.6 mm, FOV = 240×200, voxel size = 1.1×0.8×3.0 mm). 50 consecutive scans were acquired with one scan taking about 6.4 seconds. The total scan time was about 5 minutes and 23 seconds. Three baseline images were acquired before the injection of a dose of 0.1 mmol/kg Magnevist (Bayer, Germany) and a 20 ml saline flush. The injection rate was 4 ml/s using a power injector.

2.3 Data analyses

All of the DCE-MR imaging of LNs was evaluated independently by two radiologists (have been working in the filed for 2 years and 5 years, respectively). The dynamic-enhanced data were analyzed with Tissue 4D of Syngo, provided by Siemens. Regions of interest (ROIs) were drawn by the guidance of T2 weighted images. A set of four ROIs were manually drawn on the dynamic enhanced images at a time. The whole volume of the LN was included by drawing ROIs slice by slice, with an exclusion of areas of necrosis, hemorrhages and vessels [4] (Fig. 1).

Motion correction was performed before further analysis. The TICs of ROIs were displayed according to the signal intensity of T1 map. Compared with those in the previous research [6], the TICs of studied LNs exhibited four patterns: (I) the slow inflow and slow washout pattern (type I), with a peak enhancement reached beyond 111 seconds after bolus injection and a final reduction ≤20% of peak enhancement; (II) the slow inflow and rapid washout pattern (type II), with a slow increase in signal intensity (>111 s) and a final reduction >20%; (III) the rapid inflow and slow washout pattern (type III), with a rapid increase in signal intensity (≤111 s) and a slow reduction (≤20%); or (IV) the rapid inflow and rapid washout pattern (type IV), with a rapid increase (≤111 s) and a rapid reduction (>20%) in signal intensity. The values of the K^trans, Ve and iAUC of the selected ROIs were calculated using Tofts Model, based on population-averaged AIF [20]. A histogram was displayed to observe the distribution of the K^trans in the selected ROI. If the K^trans was of Gaussian distribution, the mean values would be used for analysis; if not, the median values would be chosen.

2.4 Statistical analyses

All statistical analyses were performed using the software SPSS (version 16.0, SPSS Inc. Chicago, Ill., USA). Kolmogorov-Smirnov test was applied to check the normality. Kappa test was used to calculate the consistency of quantitative values measured by the two radiologists. Chi-square test was conducted to compare gender and TIC patterns between the two groups. Two sample t-test was performed to calculate the differences of ages, short-axis diameters and quantitative parameters between the two groups. Receiver operator characteristics analysis was conducted to assess the diagnostic ability and the cut-off value. One factor covariance analysis was used to analyze the effect of short-axis diameter on changes of quantitative parameters of metastatic LNs and benign LNs. A p value less than 0.05 was considered as statistical significant. All statistical tests were two-sided.

3 Results

3.1 Basic information and characteristics of LNs on routine MR images

The ages of patients with metastatic LNs and controls were (50.1±14.0) years old and (57.4±12.0) years old, respectively. The ratios of male to female were 13:6 and 16:16 for patients and controls. No statistical differences of ages and sexes were found between the two groups (p = 0.064 and 0.321, respectively). The short-axis diameters of metastatic LNs and benign LNs were (10.8±4.2) mm and (5.8±1.5) mm. One factor covariance analysis suggested that short-axis diameter had no effect on the K^trans, Ve and iAUC of metastatic and benign LNs (F = 0.484, p = 0.488). Necrosis was present in 18 metastatic LNs. There were nine metastatic LNs merged together in 2 patients (four for one and five for the other). Extranodal invasions were seen in 4 patients with metastases. Ring enhancement was observed in 15 patients with metastases.

3.2 Comparisons of TIC patterns between metastatic and benign LNs

The TICs of metastatic and benign LNs are listed in Table 3. TICs of 50% of metastatic LNs were type III (Fig. 2), and 58% of normal LNs were type IV (Fig. 3). TICs of rapid wash-in were relatively common in both metastatic and benign LNs, accounting for 93.7% and 90.3%, respectively. Wash-in patterns were not statistically different between groups (p = 0.527). The ratios of rapid to slow wash-out types of metastatic LNs and benign LNs were 28:36 and 36:26, respectively. Wash-out patterns were not statistically different between groups (p = 0.117).

3.3 Comparisons of quantitative parameters between metastatic and benign LNs

No statistical difference was found between evaluations by two independent radiologists (P Ϡ 0.05). The Kappa value was 0.809 (p < 0.001). Mean values of two evaluations were used. Values of the K^trans, Ve and iAUC of metastatic (n = 64) and benign (n = 62) LNs are listed in Table 4. Significant differences of the K^trans, Ve and iAUC were found between groups (all p < 0.001).

Results of ROC analysis was shown in Fig. 4. The cut-off values and diagnostic ability of parameters were also shown in Table 4. The Ve had the greatest diagnostic ability. The accuracy rate of Ve was 89.4%. The diagnostic sensitivity and specificity were 87.7% and 91.2%, respectively, by a cut-off value of 0.427. The positive predicted value was 91.4%. The negative predicted value was 89.4%.

4 Discussions

There are three important findings reported in this paper. Firstly, TICs of metastatic and benign LNs are not accurate to differentiate LNs of different status. Secondly, the K^trans, the Ve and the iAUC are reliable to differentiate metastatic LNs from benign LNs. The most accurate parameter is the Ve, with a diagnostic accuracy of 89.4% and a cut-off value of 0.427. Thirdly, sizes of LNs have no effect on the K^trans, Ve and iAUC of metastatic and benign LNs.

Several studies on the diagnostic ability of DCE-MRI on metastatic LNs have been performed in past few years [6 , 16]. All of these studies focused on TICs. The study by Fischbein et al. [6] concluded that LNs with TICs of slow wash-in patterns were inclined to be metastases. In their study, the mean time to peak of 25 tumor-involved LNs of head and neck squamous cell carcinoma was 111 s. In the present study, we defined 111s as the threshold for wash-in patterns. As shown in Table 3, the results were not satisfying. Another study by Kvistad et al. [16] found that the rapid wash-in patterns are more common in breast cancers involved LNs. Results of another small sample study by Wendle et al. [32] found that the TICs of metastatic LNs of oral squamous cell carcinoma were not regular [32]. The reason for the differences is not clear. It might be related to the locations and pathological types of both primary tumors and LNs. Individual circulation time of the patients also influences the occurrences of the signal enhancement after the injection of contrast medium [32]. Another possibility that can’t be ignored is that, as Tofts described in one of his original works on DCE-MR imaging [29], the TICs were easily affected by imaging protocols.

In our study, values of the K^trans, Ve and iAUC are higher in metastatic LNs than in benign LNs and the Ve is the most accurate parameter in the differentiation. The Ve is a symbol of the extravascular extracellular volume fraction [29, 30]. Several researches on DCE-MR imaging showed that the Ve was considered as an index of necrosis and an inverse index of cellularity [3, 17]. Metastatic LNs of squamous cell carcinoma are prone to necrosis. When necrosis is obvious and detectable on routine and dynamic enhanced images, it will be excluded from ROIs. But if the necrosis area is not big enough to be detected on images, it will be included in the analysis. We suspect that the undetectable necrosis resulted in the changes of the Ve. The K^trans stands for the volume transfer constant, which is closely related to the perfusion and integrity of the endothelial cells of vesicles [29, 30]. The iAUC integrates the area under the kinetic curve during the first 90 seconds, and is related to the increased permeability and angiogenesis. According to previous research, the K^trans has been found as a parameter related to a decrease of percentage area of stroma and an increase of expression of vascular endothelial growth factor [2, 17]. Tumor vessels’ endothelial cells are often pathologically immature, lacking in pericyte and smooth muscle coverage [7]. Once a LN is invaded by tumor cells, new vessels with high permeability are generated. The increase in the number of new vessels can be detected by contrast enhanced ultrasound [32]. The increased values of K^trans and iAUC of metastatic LNs correspond to the histopathological changes.

In this study, one factor covariance analysis proved that the increases of K^trans, Ve and iAUC of metastases are not related to LNs’ sizes. In our opinion, it is an important finding, because it demonstrates that the K^trans, Ve and iAUC are helpful for detections of normal- sized metastatic LNs. According to previous studies and experience, 15–26% metastatic LNs from head and neck cancers are of normal size and misdiagnosed by routine methods [5 , 31]. Quantitative DCE-MR imaging increases positive predictive rate to 89.4–91.2%. The reason for this is that quantitative parameters of DCE-MR imaging detect metastases by assessing the changes of physiology of microcirculation [3 , 30].

There are also some limitations for this study. First, not all the studied benign LNs were pathologically confirmed. Spatial resolution of MR imaging limited the matches between the images and with pathology of benign LNs. We defined a set of criteria based on mature studies [8 , 31] and ruled out patients with malignant lesions. Second, we didn’t analyze the data of different locations of LNs due to the small sample. Therefore, a large-cohort study is needed.

In conclusion, this study shows that the quantitative parameters of metastatic LNs are higher than those of benign LNs, and the Ve is the most accurate parameter for LNs’ diagnoses. Sizes of LNs have no effect on quantitative parameters of LNs. TICs are not reliable in the diagnose of metastatic LNs.

Disclosure of funding

None.

References

Antonio

J.K.

, Santini

, Politi

, et al., Sentinel lymph node biopsy in squamous cell carcinoma of the head and neck: 10 years of experience, Acta Otorhinolaryngol Ital 32 (2012), 18–25.

Chikui

, Obara

, Simonetti

A.W.

, et al., The principal of dynamic contrast enhanced MRI, the method of pharmacokinetic analysis, and its application in the head and neck region, Int J Dent 2012 (2012), 480659.

Choi

H.S.

, Kim

A.H.

, Ahn

S.S.

, Shin

N.Y.

, Kim

and Lee

S.K.

, Glioma grading capability: Comparisons among parameters from dynamic contrast-enhanced MRI and ADC value on DWI, Korean J Radiol 14 (2013), 487–492.

Craciunescu

O.I.

, Yoo

D.S.

, Cleland

, et al., Dynamic contrast-enhanced MRI in head-and-neck cancer: The impact of region of interest selection on the intra- and interpatient variability of pharmacokinetic parameters, Int J Radiat Oncol Biol Phys 82 (2012), e345–e350. doi: 10.1016/j.ijrobp.2011.05.059.

Curtin

H.D.

, Ishwaran

, Mancuso

A.A.

, Dalley

R.W.

, Caudry

D.J.

and McNeil

B.J.

, Comparison of CT and MR imaging in staging of neck metastases, Radiology 207 (1998), 123–130.

Fischbein

N.J.

, Noworolski

S.M.

, Henry

R.G.

, Kaplan

M.J.

, Dillon

W.P.

and Nelson

S.J.

, Assessment of metastatic cervical adenopathy using dynamic contrast-enhanced MR imaging, Am J Neuroradiol 24 (2003), 301.

Gaya

, Daley

, Taylor

N.J.

, et al., Relationship between human tumour angiogenic profile and combretastatin-induced vascular shutdown: An exploratory study, Br J Cancer 99 (2008), 321–326.

Giacomini

C.P.

, Jeffrey

R.B.

and Shin

L.K.

, Ultrasonographic evaluation of malignant and normal cervical lymph nodes, Semin Ultrasound CT MR 34 (2013), 236–247.

Gor

D.M.

, Langer

J.E.

and Loevner

L.A.

, Imaging of cervical lymph nodes in head and neck cancer: The basics, Radiol Clin North Am 44 (2006), 101–110.

10.

Greco

, Crippa

, Agresti

, et al., Axillary lymph node staging in breast cancer by 2-fluoro-2-deoxy-d-glucose positron emission tomography, J Natl Cancer Inst 4 (2001), 630–635.

11.

Haberal

, Celik

, GöÇmen

, Akmansu

, Yörük

and Ozeri

, Which is important in the evaluation of metastatic lymph nodes in head and neck cancer: Palpation, ultrasonography, or computed tomography? Otolaryngol Head Neck Surg 130 (2004), 197–201.

12.

Hoskin

P.J.

, Saunders

M.I.

, Goodchild

, Powell

M.E.

, Taylor

N.J.

and Baddeley

, Dynamic contrast enhanced magnetic resonance scanning as a predictor of response to accelerated radiotherapy for advanced head and neck cancer, Br J Radiol 72 (1999), 1093–1098.

13.

Jafari

N.S.A.

, Moradi

, Kharazifard

and Khami

, Delay in the diagnosis and treatment of oral cancer, J Dent (Shiraz) 14 (2013), 146–150.

14.

Jansen

J.F.

, Carlson

D.L.

, Lu

, et al., Correlation of a priori DCE-MRI and ¹H-MRS data with molecular markers in neck nodal metastases: Initial analysis, Oral Oncology 48 (2012), 717–722.

15.

Jansen

J.F.

, Schöder

, Lee

N.Y.

, et al., Tumor metabolism and perfusion in head and neck squamous cell carcinoma: Pretreatment multimodality imaging with ¹H magnetic resonance spectroscopy, dynamic contrast-enhanced MRI, and [¹⁸F]FDG-PET, Int J Radiat Oncol Biol Phys 82 (2012), 299–307.

16.

Kvistad

K.A.

, Rydland

, Smethurst

H.B.

, Lundgren

, Fjøsne

H.E.

and Haraldseth

, Axillary lymph node metastases in breast cancer: Preoperative detection with dynamic contrast-enhanced MRI, Eur Radiol 10 (2000), 1464–1471.

17.

Langer

D.L.

, van der Kwast

T.H.

, Evans

A.J

, et al., Prostate tissue composition and MR measurements: Investigating the relationships between ADC, T2, K^trans, Ve, and corresponding histologic features, Radiology 255 (2010), 485–494.

18.

Lefebvre

J.L.

, Chevalier

, Luboinski

, Kirkpatrick

, Collette

and Sahmoud

, Larynx preservation in pyriform sinus cancer: Preliminary results of a European organization for research and treatment of cancer phase III trial. EORTC Head and Neck Cancer Cooperative Group, J Natl Cancer Inst 88 (1996), 890–899.

19.

Mochiki

, Sugasawa

, Nibu

, Asai

, Nakao

, Asakage

, Prognostic factors for hypopharyngeal cancer: A univariate and multivariate study of 142 cases, Acta oto-laryngologica, Supplementum (2007), 136–144.

20.

Parker

G.J.

, Roberts

, Macdonald

, et al., Experimentally-derived functional form for a population-averaged high-temporal-resolution arterial input function for dynamic contrast-enhanced MRI, Magn Reson Med 56(200), 993–1000.

21.

Som

P.M.

, Lymph nodes of the neck, Radiology 165 (1987), 593–600.

22.

Prades

J.M.

, Timoshenko

A.P.

, Schmitt

T.H.

, et al., Planned neck dissection before combined chemoradiation for pyriform sinus carcinoma, Acta Oto-laryngologica 128 (2008), 324–328.

23.

Rodjan

, de Graaf

, vanderValk

, et al., Retinoblastoma: Value of dynamic contrast-enhanced MR imaging and correlation with tumor angiogenesis, Am J Neuroradiol 33 (2012), 2129–2135.

24.

Society

. Cancer facts and figures 2010. Atlanta: American cancer society. (2010).

25.

Steppan

, Reimer

, Müller-Holzner

, et al., Breast cancer in women: Evaluation of benign and malignant axillary lymph nodes with contrast-enhanced ultrasound, Ultraschall Med 31 (2010), 63–67.

26.

Sumi

, Sakihama

, Sumi

, et al., Discrimination of metastatic cervical lymph nodes with diffusion-weighted MR imaging in patients with head and neck cancer, Am J Neuroradiol 24 (2003), 1627–1634.

27.

Tang

, Li

, Mao

, et al., Retropharyngeal lymph node metastasis in nasopharyngeal carcinoma detected by magnetic resonance imaging: Prognostic value and staging categories, Cancer 113 (2008), 347–354.

28.

Tham

I.W.

, Hee

S.W.

, Yap

S.P.

, Tuan

J.K.

and Wee

, Retropharyngeal nodal metastasis related to higher rate of distant metastasis in patients with N0 and N1 nasopharyngeal cancer, Head Neck 31 (2009), 468–474.

29.

Tofts

P.S.

, Modeling tracer kinetics in dynamic GD-DTPA MR imaging, J Magn Reson Imaging 7 (1997), 91–101.

30.

Tofts

P.S.

, Brix

, Buckley

D.L.

, et al., Estimating kinetic parameters from dynamic contrast-enhanced T(1)-weighted MRI of a diffusible tracer: Standardized quantities and symbols, J Magn Reson Imaging 10 (1999), 223–232.

31.

van den Brekel

M.W.

, Stel

H.V

, et al., Cervical lymph node metastasis: Assessment of radiologic criteria, Radiology 177 (1990), 379–384.

32.

Wendl

C.M.

, Müller

, Meier

, et al., High resolution contrast-enhanced ultrasound and 3-tesla dynamic contrast-enhanced magnetic resonance imaging for the preoperative characterization of cervical lymph nodes: First results, Clin Hemorheol Microcirc 52 (2012), 153–166.