- Email: [email protected]
Apparent diffusion coefficient values of necrotic and solid portion of lymph nodes: Differential diagnostic value in cervical lymphadenopathy
Clinical Radiology 68 (2013) 224e231
Contents lists available at ScienceDirect
Clinical Radiology journal homepage: www.clinicalradiologyonline.net
Original Paper
Apparent diffusion coefficient values of necrotic and solid portion of lymph nodes: Differential diagnostic value in cervical lymphadenopathy Y. Zhang a, b, y, *, J. Chen c, y, J. Shen c, J. Zhong c, R. Ye c, B. Liang c a
Diagnostic Imaging and Intervening Center, Cancer Center, Guangzhou, Guangdong, PR China State Key Laboratory of Oncology in South China, Guangzhou, Guangdong, PR China c Department of Radiology, The Second Affiliated Hospital, Sun Yat-sen University, Guangzhou, Guangdong, PR China b
article in formation Article history: Received 20 July 2010 Received in revised form 29 March 2011 Accepted 1 April 2011
AIM: To evaluate whether the analysis of the apparent diffusion coefficient (ADC) values of the necrotic and solid portions of lymph nodes aids differentiation between the causes of cervical lymphadenopathy. MATERIALS AND METHODS: Thirty-six patients with cervical lymph node metastasis from head and neck squamous cell carcinomas (SCC), 19 patients with lymphoma, and 23 patients with tuberculous lymphadenitis underwent conventional magnetic resonance imaging (MRI) and diffusion-weighted imaging (DWI). The ADC values of necrotic and solid portions of lymph nodes were measured and compared. Receiver operating characteristic (ROC) analysis was employed to investigate whether ADC values could help to discriminate between the causes of cervical lymphadenopathy, and to obtain the optimal ADC threshold values. RESULTS: The mean ADC values of the solid portions of metastatic nodes, lymphomatous nodes, and tuberculous nodes were (0.93 0.16) 103 mm2/s, (0.64 0.13) 103 mm2/s and (1.01 0.11) 103 mm2/s, respectively (p < 0.01). The mean ADC values of necrosis of metastatic and tuberculous nodes were (2.02 0.36) 103 mm2/s and (1.25 0.15) 103 mm2/s (p < 0.01). By using the ADC value of the solid portion, the optimal ADC threshold values for distinguishing between metastasis and lymphoma, between metastasis and tuberculosis, and between lymphoma and tuberculosis were 0.77 103, 0.98 103 and 0.81 103 mm2/s, respectively, and the sensitivities and specificities were 83 and 89%, 70 and 68%, 93 and 100%, respectively. By using ADC values of necrosis, the optimal ADC threshold value for distinguishing between metastasis and tuberculosis was 1.59 103 mm2/s, and the sensitivity and specificity were 88 and 100%, respectively. CONCLUSION: The ADC values both of the necrotic and solid portions of the lymph nodes are useful in differentiation between the causes of cervical lymphadenopathy. The ADC value of necrosis is especially helpful in discriminating metastasis from tuberculosis. Ó 2011 Published by Elsevier Ltd on behalf of The Royal College of Radiologists.
Introduction * Guarantor and correspondent: Y. Zhang, 651 Dongfeng Road, East, Guangzhou, PR China. Tel.: þ86 20 87343217, þ86 15920556877 (Mob); fax: þ86 20 87343217. E-mail address: [email protected] (Y. Zhang). y Both authors contributed equally to this work.
Diffusion-weighted imaging (DWI) can make subtle abnormality more obvious, and can provide characterization of tissues and their pathological processes at the microscopic level.1 Previous studies1e7 have shown that DWI can be
0009-9260/$ e see front matter Ó 2011 Published by Elsevier Ltd on behalf of The Royal College of Radiologists. doi:10.1016/j.crad.2011.04.002
Y. Zhang et al. / Clinical Radiology 68 (2013) 224e231
successfully used in the head and neck region, and can be helpful in distinguishing between causes of cervical lymphadenopathy.1,3,5e7 To the authors’ knowledge, in those studies, the apparent diffusion coefficient (ADC) values of lymph nodes without necrosis or the solid portions in the lymph nodes with necrosis were principally analysed, and were used for differential diagnosis. Although necrosis is very common in cervical lymphadenopathy, the diagnostic accuracy of the ADC value of necrosis to differentiate cervical nodes was seldom reported. Therefore, the purpose of this study was to analyse the ADC values of both the necrosis and solid portions in the cervical lymph nodes; in particular, the ADC values of necrosis, to differentiate causes of cervical lymphadenopathy.
Materials and methods Patients Between October 2004 and September 2007, 91 patients with enlarged cervical lymph nodes underwent conventional MRI and DWI. MRI examinations were performed before biopsy or further treatment. Four patients were excluded from this study due to image artefacts obscuring anatomy. Final diagnosis for all lymph nodes was made by histopathological examination after dissection for 44 patients, and surgical biopsy or core biopsy for 43 patients. According to the histopathological diagnoses, 36 patients with head and neck squamous cell carcinoma (SCC), 19 patients with lymphoma, and 23 patients with tuberculous lymphadenitis were included in this study. In 36 patients (28 men and eight women; mean age 55 years, age range 29e83 years) with head and neck SCC, the primary cancers arose in the larynx (n ¼ 16), tongue (n ¼ 10), nasopharynx (n ¼ 4), mouth floor (n ¼ 3), oral pharynx (n ¼ 2), and gingiva (n ¼ 1). Twenty-three patients were diagnosed with tuberculous lymphadenitis (16 men and seven women; mean age 32.1 years, age range 22e45 years). In 19 patients with lymphoma (11 men and eight women; mean age 56.5 years, age range 34e78 years), the histopathological types included diffuse large B-cell lymphomas (n ¼ 13), follicular lymphoma (n ¼ 5), and peripheral unspecified lymphoma (n ¼ 1). Six patients who did not suffer from head and neck SCC were excluded from this study including five patients with papillary carcinoma of thyroid, one patient with mucoepidermoid carcinoma of the parotid. Two patients with reactive lymphoid hyperplasia and one patient with infectious mononucleosis were also excluded from this study. To ensure that the biopsied lymph nodes or the nodes removed during surgery were the same as seen at MRI; images were analysed in presence of the surgeon performing the procedure. In total, 84 metastatic nodes, 53 tuberculous nodes, and 40 lymphomatous nodes were analysed in this study. The local ethics committee granted ethical approval for the study, and informed consent was obtained.
MRI All MRI examinations were performed using a 1.5 T MRI unit (Gyroscan Intera Master, Philips Medical Systems, the
225
Netherlands) with a synergy head and neck coil. All patients underwent conventional MRI and DWI to include nodes from the base of the skull to the suprasternal notch. Before imaging, all patients were told to try to avoid swallowing during MRI examination. Conventional MRI images were acquired as follows: transverse, coronal, and sagittal T1-weighted spin-echo (SE) [repetition time (TR)/echo time (TE), 400 ms/15 ms; 5 mm section thickness, 0.5 mm intersection gap] images, and transverse T2-weighted turbo SE (TSE) images (TR/TE 4000 ms/100 ms; 5 mm section thickness, 0.5 mm intersection gap). Subsequently, transverse diffusion-weighted images of the neck were obtained using a single-shot SE-type echoplanar imaging (EPI) sequence with sensitivity encoding (SENSE) technique. Imaging parameters for DWI were as follows: TR/TE/inversion time (TI) 6800 ms/70 ms/180 ms; b factors, 0 and 800 s/mm2; 160 256 matrix; 325 mm field of view (FOV); 4 mm section thickness with no gap; SENSE factor 2; 10 signals acquired; 7 min acquisition time. A short-tau inversion recovery (STIR) sequence was used for fat suppression in the diffusion-weighted sequence. Motion-probing gradient pulses were placed in the three orthogonal planes. Isotropic diffusion-weighted images were generated using three orthogonal-axis images. SE T1-weighted images (TR/TE 400 ms/15 ms; 5 mm section thickness, 0.5 mm intersection gap) in the transverse, coronal, and sagittal planes after an intravenous bolus injection of 0.2 ml/kg of body weight of gadopentetate demeglumine (Magnevist; Guangzhou Schering, Guangzhou, China) were obtained of all the patients.
Data analysis The ADC values of target lymph nodes were measured by an MRI radiologist without knowledge of the primary lesion or its histological type. A region of interest (ROI) was drawn manually on the ADC map around the margin of the lymph node and the ADC value was measured. For each node, the following measurements were made: (a) for nodes without necrosis, the ADC value of the whole node was measured, and (b) for nodes with necrosis, ADC values of the solid portion and the area of necrosis were measured. Necrosis was defined as areas that demonstrated hyperintensity on T2-weighted images, hypointensity on T1-weighted images, and had no contrast enhancement on T1-weighted contrast-enhanced images. ADC maps were generated on the scanner console using the b ¼ 0 and b ¼ 800 images.
Statistical analysis Statistical analyses were performed by using the software SPSS 13.0. p < 0.05 was considered statistically significant. Comparison of the ADC values of the solid portions of metastatic, lymphomatous, and tuberculous nodes was performed by using the one-way analysis of variance (ANOVA) test. Student’s t test was used to compare the ADC values of necrosis between metastatic and
226
Y. Zhang et al. / Clinical Radiology 68 (2013) 224e231
tuberculous nodes. Receiver operating characteristic (ROC) curve analysis was employed to investigate the discriminatory capability of (a) the ADC value of the solid portion in lymph nodes to distinguish between metastatic nodes, lymphomatous nodes, and tuberculous nodes; and (b) the ADC value of necrosis in lymph nodes for distinguishing between metastatic and tuberculous nodes. The area under
the ROC curve (AUC) was used to provide a measure of the performance of using the ADC values as effective indicators for discrimination. The ADC value that corresponded to the highest Yoden index (Yoden index ¼ sensitivity þ specificitye1) was chosen as the optimal ADC threshold value because it optimizes both the sensitivity and specificity.
Figure 1 MRI images of metastatic nodes with central necrosis in a 59-year-old woman with nasopharyngeal carcinoma. (a) Transverse T2-weighted image shows bilateral enlarged lymph nodes with and without central necrosis. (b) Transverse contrast-enhanced T1-weighted image shows the necrosis without contrast enhancement. (c) Transverse diffusion-weighted image acquired by using a b value of 800 s/mm2 shows the nodes with a hypointense central necrosis surrounded by hyperintense peripheral solid portions. (d) ADC map shows the central necrosis as hyperintensity, and solid portions as hypointensity. The ADC value of necrosis in the right-sided node was 2.35 103 mm2/s, and the ADC value of the solid portion in this node was 0.98 103 mm2/s.
Y. Zhang et al. / Clinical Radiology 68 (2013) 224e231
Results DWI and ADC values of lymph nodes Solid portions of all lymph nodes were clearly shown as a hyperintensity on DWI images (Figs 1, 2). The mean ADC value of the solid portion of tuberculous nodes [(1.01 0.11) 103 mm2/s] was significantly higher (p < 0.01) than that of metastatic nodes [(0.93 0.16) 103 mm2/s] and lymphomatous nodes [(0.64 0.13) 103 mm2/s]. The mean ADC value of the metastatic nodes
227
was higher (p < 0.01) than that of the lymphomatous nodes (Fig 3; Table 1). None of the lymphomatous nodes demonstrated necrosis. Twenty-six of 84 (31%) metastatic nodes and 38 of 53 (72%) tuberculous nodes demonstrated necrosis (Table 2). Areas of necrosis in both tuberculous nodes and metastatic nodes were hypointense relative to the solid portions of nodes on DWI images. (Figs 1 and 2); the areas of ROI were from 26 to 296 mm2. The mean ADC value of necrosis in metastatic nodes [(2.02 0.36) 103 mm2/s] was significantly higher (p < 0.01) than that
Figure 2 MRI images of a left-sided tuberculous node with central necrosis in a 35-year-old woman. (a) Transverse T2-weighted image and (b) Transverse contrast-enhanced T1-weitgted image show central necrosis in the lymph node. (c) Transverse diffusion-weighted image acquired by using a b value of 800 s/mm2 shows the nodes with a hypointense necrosis surrounded by a hyperintense peripheral rim. (d) ADC map shows the central necrosis as hyperintensity. The ADC value of the solid portion was 1.09 103 mm2/s, the ADC value of necrosis was 1.20 103 mm2/s.
228
Y. Zhang et al. / Clinical Radiology 68 (2013) 224e231 Table 2 Apparent diffusion coefficient (ADC) values ( 103 mm2/s) of necrosis for metastasis and tuberculosis. Histological findings
Number
Area of necrosis (mm2)
Mean ADC value (x s)
95% CI for mean
Range
Metastatic 26 (31%) 0.11e2.96 2.02 0.36 1.88e2.17 1.27e2.65 nodes Tuberculous 38 (72%) 0.05e1.01 1.25 0.15 1.20e1.29 1.01e1.59 nodes t-value 12.045 p-value <0.001
metastasis and tuberculosis was 1.59 103 mm2/s, and the sensitivity and specificity were 88 and 100% (Table 3).
Discussion Figure 3 ADC values of the solid portion of nodes due to metastases, lymphoma, and tuberculosis.
of tuberculous nodes ((1.25 1.02) 103 mm2/s; Fig 4; Table 2).
Diagnostic ability of DWI quantitative criteria in discriminating nodes The diagnostic ability of ADC value criteria in differentiating between causes of cervical lymphadenopathy was assessed by performing ROC analysis. Fig 5aec demonstrates the ROC curves of the ADC value of the solid portion used to differentiate lymphoma from metastasis, lymphoma from tuberculosis, and metastasis from tuberculosis; the AUC values were 0.94, 0.99, and 0.69, respectively. Fig 5d shows the ROC curve of the ADC value of necrosis used to differentiate metastasis from tuberculosis, and the AUC was 0.98. By using the ADC value of the solid portion, the optimal ADC threshold values to distinguish between metastasis and lymphoma, between metastasis and tuberculosis, and between lymphoma and tuberculosis were 0.77 103, 0.98 103 and 0.81 103 mm2/s, respectively. The sensitivity and specificity values were 83 and 89%, 70 and 68%, and 93 and 100%, respectively. By using the ADC value of necrosis, the optimal ADC threshold value to distinguish between
DWI is sensitive to the diffusion of water molecules in tissue, which can make subtle abnormalities more obvious.1 The contrast of DWI depends principally on the degree to which water molecules are free to move within tissues and pass through cell membranes.8 Diffusion is expressed as an ADC, which reflects the diffusion properties unique to each type of tissue. Previously, DWI has been used to evaluate various diseases of the central nervous system. The most established clinical application of DWI for the central nervous system is in the evaluation of acute stroke. The use of DWI in the head and neck region is challenging because of susceptibility artefacts.6 However, with the development of MRI hardware and sequences, DWI has been successfully used in the head and neck region,1e7 and previous studies have reported that the ADC value was helpful in distinguishing causes of cervical lymphadenopathy.1,3,5e7 Precious studies have analysed the ADC values of lymph nodes without necrosis or the solid portions in lymph nodes
Table 1 Apparent diffusion coefficient (ADC) values ( 103 mm2/s) of the solid portion in nodes for each nodal lesion. Histological findings
Mean short Mean ADC axis (mm) value (x s) (x s)
Lymphoma (n ¼ 40) 15 4.9 Metastasis (n ¼ 84) 14 6.1 Tuberculosis (n ¼ 53) 10 3.2 F-value p-value CI, confidence interval.
95% CI for mean
Range
0.64 0.13 0.60e0.67 0.37e0.90 0.93 0.16 0.90e0.97 0.67e1.41 1.01 0.11 0.98e1.04 0.81e1.26 82.928 <0.001
Figure 4 ADC values of necrosis in lymph nodes from metastases and tuberculosis.
Y. Zhang et al. / Clinical Radiology 68 (2013) 224e231
229
Figure 5 ROC curves for the ADC value of the solid portion (aec) and necrotic portion (d) in lymph nodes to discriminate between (a) lymphoma and metastasis, (b) lymphoma and tuberculosis, (c) metastasis and tuberculosis, (d) metastasis and tuberculosis.
Table 3 Apparent diffusion coefficient (ADC) threshold value (103 mm2/s) for discrimination between metastasis, lymphoma, and tuberculosis. Nodal lesions Solid portion M and L M and TB L and TB Necrosis M and TB
ADC Threshold value
AUC
Sensitivity
Specificity
Yoden index
0.77 0.98 0.81
0.94 0.69 0.99
83% 70% 93%
89% 68% 100%
0.72 0.38 0.93
1.59
0.98
88%
100%
0.88
AUC, area under the receiver operating characteristic (ROC) curve; M, metastasis; L, lymphoma; TB, tuberculosis.
with necrosis.1,3,5e7 The consensus in those studies was that the mean ADC value of lymphomatous nodes is lower than other causes of cervical lymphadenopathy including malignancy. In this study, a similar result was obtained, i.e., the mean ADC value of lymphoma was lower than those of metastasis and tuberculosis. This result indicated that DWI is very helpful in distinguishing lymphoma from nonlymphomatous lesions. The difference in restriction of diffusion in the tissue is mainly attributed to differences in cellularity; greater cellularity and less extracellular space result in more restricted diffusion.9 Lymphoma is expected to have greater tumour cellularity than other lymphadenopathies, which may explain the lower ADC values obtained for lymphoma than for SCC and tuberculosis.
230
Y. Zhang et al. / Clinical Radiology 68 (2013) 224e231
Metastatic SCC lymph nodes are heterogeneous and contain variable degrees of necrotic parts. Based on the features, it is hypothesized that these lymph nodes are associated with alterations in water diffusivity, and DWI may play a role in their characterization. Several studies have reported the ability of DWI to discriminate malignant from benign lymph nodes in the neck.10,11 Abdel-Razek et al.1 and Holzapfel et al.7 reported that the ADC value of metastatic nodes is significantly lower than that of benign nodes. However, Sumi et al.3,5 reported significantly higher ADC values in metastatic lymph nodes than in benign lymphadenopathy. In the present study, the mean ADC value of the solid portion in metastasis was significantly lower than that of tuberculosis, but there was overlap in many cases. The AUC in ROC curve analysis was only 0.69, which was much lower than that in differentiating lymphoma from metastasis (AUC ¼ 0.94) and tuberculosis (AUC ¼ 0.99). With a threshold ADC value of 0.98 103 mm2/s, a low sensitivity of 70% and specificity of 68% were achieved in the differentiation of metastatic nodes from tuberculous nodes. This result indicated that using the ADC value of the solid portion to differentiate lymphoma from non-lymphomatous lymphadenopathy is much more effective than differentiating metastasis from tuberculosis. In this study the ADC values of necrotic portions of lymph nodes was useful in differentiating metastasis from tuberculosis. In fact, visible necrosis on MRI images is common in lymphadenopathy. Indeed, in this study, 31% of the metastatic nodes and 72% of the tuberculous nodes demonstrated necrosis. There are few differences between these two kinds of necrotic nodes on conventional MRI images. Necrosis is a factor that can influence diffusion of water molecules in tissue;2,10,12 as the amount of necrosis increases, the ADC value increases.13 Histopathological analysis has shown that micronecrosis is common in both lymphoma and SCC.4 Maeda et al.4 reported that the ADC values of tumours did not significantly differ with the degree of small foci of necrosis. Lyng et al.14 reported that the fraction of massive necrosis could be correlated with the ADC value but not the fraction of small foci of necrosis because the size of small foci of necrosis may be smaller than the voxel of the MRI images. In previous studies, only the ADC values of solid portions in lymph nodes were used to differentiate causes of cervical lymphadenopathy. Although the ADC values of necrosis in metastatic nodes were reported in the studies of both AbdelRazek et al. and Holzapfel et al., unfortunately, further analyses were not performed. In the present study, the ADC values of necrosis visible in lymph nodes was measured, and the mean ADC value of necrosis in metastasis [(2.02 0.36) 103 mm2/s] was found to be greater than that in tuberculosis [(1.25 1.02) 103 mm2/s]. It is speculated that this difference was due to different pathological mechanisms of necrosis. The degree of restriction to water diffusion in biological tissue is inversely correlated with tissue cellularity and the integrity of cell membranes. In necrotic tumours, a less cellular environment provides a larger extracellular space for diffusion of water molecules, and these molecules may also freely transgress defective or
disruptive cell membranes to move from the extracellular into the intracellular compartment,15 and this results in an increase in the ADC value. It was reported that a higher protein component in infective lesions increases the viscosity and reduces water proton mobility leading to lower ADC values.16 The ROC curve analysis indicated that the diagnostic potential of the ADC value in necrosis was much better than that in the solid portion of the node to differentiate metastasis from tuberculosis. If the ADC value of necrosis was used to distinguish between metastasis and tuberculosis, a much higher specificity and sensitivity (88% and 100%) could be obtained than by using the ADC value of the solid portion (specificity of 68%, sensitivity of 70%). This preliminary study indicates that the ADC value of necrosis in lymph nodes was very useful in characterizing cervical lymph nodes. As necrosis is common in metastasis and tuberculosis, when necrosis is detected in cervical lymph nodes, measurement of the ADC value of necrosis should be considered as a non-invasive method for differentiating metastatic nodes from tuberculous node, especially in nodes without confirmed primary tumours. A limitation of the present study was the absence of necrosis in the group of lymphoma. Although reports show that necrosis is also seen in lymphoma, lymphomatous nodes with necrosis were not found in this study. Further analysis of necrosis between lymphoma and other nodes was not undertaken. Another limitation of the present study was that the results were based on a relatively small sample of patients, so there is likely to be some uncertainty in these estimates. Therefore, it is important to perform a further study in a larger patient population in order to substantiate the role of DWI as a reliable method in differentiating among causes of cervical lymphadenopathy. The ADC values of both the necrotic and the solid portions in lymph nodes are useful in the discrimination of cervical lymphadenopathy. The ADC value of necrosis is especially helpful in differentiating metastasis from tuberculosis.
References 1. Abdel-Razek AA, Soliman NY, Elkhamary S, et al. Role of diffusion-weighted MR imaging in cervical lymphadenopathy. Eur Radiol 2006;16:1468e77. 2. Wang J, Takashima S, Takayama F, et al. Head and neck lesions: characterization with diffusion-weighted echo-planar MR imaging. Radiology 2001;220:621e30. 3. 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. AJNR Am J Neuroradiol 2003;24:1627e34. 4. Maeda M, Kato H, Sakuma H, et al. Usefulness of the apparent diffusion coefficient in line scan diffusion-weighted imaging for distinguishing between squamous cell carcinomas and malignant lymphomas of the head and neck. AJNR Am J Neuroradiol 2005;26:1186e92. 5. Sumi M, Cauteren MV, Nakamura T. MR microimaging of benign and malignant nodes in the neck. AJR Am J Roentgenol 2006;186:749e57. 6. King AD, Ahuja AT, Yeung DKW, et al. Malignant cervical lymphadenopathy: diagnostic accuracy of diffusion-weighted MR imaging. Radiology 2007;245:806e13. 7. Holzapfel K, Duetsch S, Fauser C, et al. Value of diffusion-weighted MR imaging in the differentiation between benign and malignant cervical lymph nodes. Eur J Radiol 2009;72:381e7. 8. Chan JH, Tsui EY, Luk SH, et al. Diffusion-weighted MR imaging of the liver: distinguishing hepatic abscess from cystic or necrotic tumor. Abdom Imaging 2001;26:161e5.
Y. Zhang et al. / Clinical Radiology 68 (2013) 224e231 9. Guo AC, Cummings TJ, Dash RC, et al. Lymphomas and high-grade astrocytomas: comparison of water diffusibility and histologic characteristics. Radiology 2002;224:177e83. 10. White ML, Zhang Y, Robinson RA. Evaluating tumors and tumors and tumor like lesions of the nasal cavity, the paranasal sinuses, and the adjacent skull base with diffusion-weighted MRI. J Comput Assist Tomogr 2006;20:490e5. 11. Chong VF, Fan YF, Khoo JB. MRI features of cervical nodal necrosis in metastatic disease. Clin Radiol 1996;51:103e9. 12. Kim YJ, Chang KH, Song IC, et al. Brain abscess and necrotic or cystic brain tumor: discrimination with signal intensity on diffusion weighted imaging. AJR Am J Roentgenol 1998;171:1487e90.
231
13. Lang P, Wendland MF, Saeed M, et al. Osteogenic sarcoma: noninvasive in vivo assessment of tumor necrosis with diffusion-weighted MR imaging. Radiology 1998;206:227e35. 14. Lyng H, Haraldseth O, Rofstad EK. Measurement of cell density and necrotic fraction in human melanoma xenografts by diffusion weighted magnetic resonance imaging. Magn Reson Med 2000;43: 828e36. 15. Koh DW, Collins DJ. Diffusion-weighted MRI in the body: applications and challenges in oncology. AJR Am J Roentgenol 2007;188: 1622e35. 16. Mishra AM, Gupta RK, Saksena S, et al. Biological correlates of diffusivity in brain abscess. Magn Reson Med 2005;54:878e85.
Contents lists available at ScienceDirect
Clinical Radiology journal homepage: www.clinicalradiologyonline.net
Original Paper
Apparent diffusion coefficient values of necrotic and solid portion of lymph nodes: Differential diagnostic value in cervical lymphadenopathy Y. Zhang a, b, y, *, J. Chen c, y, J. Shen c, J. Zhong c, R. Ye c, B. Liang c a
Diagnostic Imaging and Intervening Center, Cancer Center, Guangzhou, Guangdong, PR China State Key Laboratory of Oncology in South China, Guangzhou, Guangdong, PR China c Department of Radiology, The Second Affiliated Hospital, Sun Yat-sen University, Guangzhou, Guangdong, PR China b
article in formation Article history: Received 20 July 2010 Received in revised form 29 March 2011 Accepted 1 April 2011
AIM: To evaluate whether the analysis of the apparent diffusion coefficient (ADC) values of the necrotic and solid portions of lymph nodes aids differentiation between the causes of cervical lymphadenopathy. MATERIALS AND METHODS: Thirty-six patients with cervical lymph node metastasis from head and neck squamous cell carcinomas (SCC), 19 patients with lymphoma, and 23 patients with tuberculous lymphadenitis underwent conventional magnetic resonance imaging (MRI) and diffusion-weighted imaging (DWI). The ADC values of necrotic and solid portions of lymph nodes were measured and compared. Receiver operating characteristic (ROC) analysis was employed to investigate whether ADC values could help to discriminate between the causes of cervical lymphadenopathy, and to obtain the optimal ADC threshold values. RESULTS: The mean ADC values of the solid portions of metastatic nodes, lymphomatous nodes, and tuberculous nodes were (0.93 0.16) 103 mm2/s, (0.64 0.13) 103 mm2/s and (1.01 0.11) 103 mm2/s, respectively (p < 0.01). The mean ADC values of necrosis of metastatic and tuberculous nodes were (2.02 0.36) 103 mm2/s and (1.25 0.15) 103 mm2/s (p < 0.01). By using the ADC value of the solid portion, the optimal ADC threshold values for distinguishing between metastasis and lymphoma, between metastasis and tuberculosis, and between lymphoma and tuberculosis were 0.77 103, 0.98 103 and 0.81 103 mm2/s, respectively, and the sensitivities and specificities were 83 and 89%, 70 and 68%, 93 and 100%, respectively. By using ADC values of necrosis, the optimal ADC threshold value for distinguishing between metastasis and tuberculosis was 1.59 103 mm2/s, and the sensitivity and specificity were 88 and 100%, respectively. CONCLUSION: The ADC values both of the necrotic and solid portions of the lymph nodes are useful in differentiation between the causes of cervical lymphadenopathy. The ADC value of necrosis is especially helpful in discriminating metastasis from tuberculosis. Ó 2011 Published by Elsevier Ltd on behalf of The Royal College of Radiologists.
Introduction * Guarantor and correspondent: Y. Zhang, 651 Dongfeng Road, East, Guangzhou, PR China. Tel.: þ86 20 87343217, þ86 15920556877 (Mob); fax: þ86 20 87343217. E-mail address: [email protected] (Y. Zhang). y Both authors contributed equally to this work.
Diffusion-weighted imaging (DWI) can make subtle abnormality more obvious, and can provide characterization of tissues and their pathological processes at the microscopic level.1 Previous studies1e7 have shown that DWI can be
0009-9260/$ e see front matter Ó 2011 Published by Elsevier Ltd on behalf of The Royal College of Radiologists. doi:10.1016/j.crad.2011.04.002
Y. Zhang et al. / Clinical Radiology 68 (2013) 224e231
successfully used in the head and neck region, and can be helpful in distinguishing between causes of cervical lymphadenopathy.1,3,5e7 To the authors’ knowledge, in those studies, the apparent diffusion coefficient (ADC) values of lymph nodes without necrosis or the solid portions in the lymph nodes with necrosis were principally analysed, and were used for differential diagnosis. Although necrosis is very common in cervical lymphadenopathy, the diagnostic accuracy of the ADC value of necrosis to differentiate cervical nodes was seldom reported. Therefore, the purpose of this study was to analyse the ADC values of both the necrosis and solid portions in the cervical lymph nodes; in particular, the ADC values of necrosis, to differentiate causes of cervical lymphadenopathy.
Materials and methods Patients Between October 2004 and September 2007, 91 patients with enlarged cervical lymph nodes underwent conventional MRI and DWI. MRI examinations were performed before biopsy or further treatment. Four patients were excluded from this study due to image artefacts obscuring anatomy. Final diagnosis for all lymph nodes was made by histopathological examination after dissection for 44 patients, and surgical biopsy or core biopsy for 43 patients. According to the histopathological diagnoses, 36 patients with head and neck squamous cell carcinoma (SCC), 19 patients with lymphoma, and 23 patients with tuberculous lymphadenitis were included in this study. In 36 patients (28 men and eight women; mean age 55 years, age range 29e83 years) with head and neck SCC, the primary cancers arose in the larynx (n ¼ 16), tongue (n ¼ 10), nasopharynx (n ¼ 4), mouth floor (n ¼ 3), oral pharynx (n ¼ 2), and gingiva (n ¼ 1). Twenty-three patients were diagnosed with tuberculous lymphadenitis (16 men and seven women; mean age 32.1 years, age range 22e45 years). In 19 patients with lymphoma (11 men and eight women; mean age 56.5 years, age range 34e78 years), the histopathological types included diffuse large B-cell lymphomas (n ¼ 13), follicular lymphoma (n ¼ 5), and peripheral unspecified lymphoma (n ¼ 1). Six patients who did not suffer from head and neck SCC were excluded from this study including five patients with papillary carcinoma of thyroid, one patient with mucoepidermoid carcinoma of the parotid. Two patients with reactive lymphoid hyperplasia and one patient with infectious mononucleosis were also excluded from this study. To ensure that the biopsied lymph nodes or the nodes removed during surgery were the same as seen at MRI; images were analysed in presence of the surgeon performing the procedure. In total, 84 metastatic nodes, 53 tuberculous nodes, and 40 lymphomatous nodes were analysed in this study. The local ethics committee granted ethical approval for the study, and informed consent was obtained.
MRI All MRI examinations were performed using a 1.5 T MRI unit (Gyroscan Intera Master, Philips Medical Systems, the
225
Netherlands) with a synergy head and neck coil. All patients underwent conventional MRI and DWI to include nodes from the base of the skull to the suprasternal notch. Before imaging, all patients were told to try to avoid swallowing during MRI examination. Conventional MRI images were acquired as follows: transverse, coronal, and sagittal T1-weighted spin-echo (SE) [repetition time (TR)/echo time (TE), 400 ms/15 ms; 5 mm section thickness, 0.5 mm intersection gap] images, and transverse T2-weighted turbo SE (TSE) images (TR/TE 4000 ms/100 ms; 5 mm section thickness, 0.5 mm intersection gap). Subsequently, transverse diffusion-weighted images of the neck were obtained using a single-shot SE-type echoplanar imaging (EPI) sequence with sensitivity encoding (SENSE) technique. Imaging parameters for DWI were as follows: TR/TE/inversion time (TI) 6800 ms/70 ms/180 ms; b factors, 0 and 800 s/mm2; 160 256 matrix; 325 mm field of view (FOV); 4 mm section thickness with no gap; SENSE factor 2; 10 signals acquired; 7 min acquisition time. A short-tau inversion recovery (STIR) sequence was used for fat suppression in the diffusion-weighted sequence. Motion-probing gradient pulses were placed in the three orthogonal planes. Isotropic diffusion-weighted images were generated using three orthogonal-axis images. SE T1-weighted images (TR/TE 400 ms/15 ms; 5 mm section thickness, 0.5 mm intersection gap) in the transverse, coronal, and sagittal planes after an intravenous bolus injection of 0.2 ml/kg of body weight of gadopentetate demeglumine (Magnevist; Guangzhou Schering, Guangzhou, China) were obtained of all the patients.
Data analysis The ADC values of target lymph nodes were measured by an MRI radiologist without knowledge of the primary lesion or its histological type. A region of interest (ROI) was drawn manually on the ADC map around the margin of the lymph node and the ADC value was measured. For each node, the following measurements were made: (a) for nodes without necrosis, the ADC value of the whole node was measured, and (b) for nodes with necrosis, ADC values of the solid portion and the area of necrosis were measured. Necrosis was defined as areas that demonstrated hyperintensity on T2-weighted images, hypointensity on T1-weighted images, and had no contrast enhancement on T1-weighted contrast-enhanced images. ADC maps were generated on the scanner console using the b ¼ 0 and b ¼ 800 images.
Statistical analysis Statistical analyses were performed by using the software SPSS 13.0. p < 0.05 was considered statistically significant. Comparison of the ADC values of the solid portions of metastatic, lymphomatous, and tuberculous nodes was performed by using the one-way analysis of variance (ANOVA) test. Student’s t test was used to compare the ADC values of necrosis between metastatic and
226
Y. Zhang et al. / Clinical Radiology 68 (2013) 224e231
tuberculous nodes. Receiver operating characteristic (ROC) curve analysis was employed to investigate the discriminatory capability of (a) the ADC value of the solid portion in lymph nodes to distinguish between metastatic nodes, lymphomatous nodes, and tuberculous nodes; and (b) the ADC value of necrosis in lymph nodes for distinguishing between metastatic and tuberculous nodes. The area under
the ROC curve (AUC) was used to provide a measure of the performance of using the ADC values as effective indicators for discrimination. The ADC value that corresponded to the highest Yoden index (Yoden index ¼ sensitivity þ specificitye1) was chosen as the optimal ADC threshold value because it optimizes both the sensitivity and specificity.
Figure 1 MRI images of metastatic nodes with central necrosis in a 59-year-old woman with nasopharyngeal carcinoma. (a) Transverse T2-weighted image shows bilateral enlarged lymph nodes with and without central necrosis. (b) Transverse contrast-enhanced T1-weighted image shows the necrosis without contrast enhancement. (c) Transverse diffusion-weighted image acquired by using a b value of 800 s/mm2 shows the nodes with a hypointense central necrosis surrounded by hyperintense peripheral solid portions. (d) ADC map shows the central necrosis as hyperintensity, and solid portions as hypointensity. The ADC value of necrosis in the right-sided node was 2.35 103 mm2/s, and the ADC value of the solid portion in this node was 0.98 103 mm2/s.
Y. Zhang et al. / Clinical Radiology 68 (2013) 224e231
Results DWI and ADC values of lymph nodes Solid portions of all lymph nodes were clearly shown as a hyperintensity on DWI images (Figs 1, 2). The mean ADC value of the solid portion of tuberculous nodes [(1.01 0.11) 103 mm2/s] was significantly higher (p < 0.01) than that of metastatic nodes [(0.93 0.16) 103 mm2/s] and lymphomatous nodes [(0.64 0.13) 103 mm2/s]. The mean ADC value of the metastatic nodes
227
was higher (p < 0.01) than that of the lymphomatous nodes (Fig 3; Table 1). None of the lymphomatous nodes demonstrated necrosis. Twenty-six of 84 (31%) metastatic nodes and 38 of 53 (72%) tuberculous nodes demonstrated necrosis (Table 2). Areas of necrosis in both tuberculous nodes and metastatic nodes were hypointense relative to the solid portions of nodes on DWI images. (Figs 1 and 2); the areas of ROI were from 26 to 296 mm2. The mean ADC value of necrosis in metastatic nodes [(2.02 0.36) 103 mm2/s] was significantly higher (p < 0.01) than that
Figure 2 MRI images of a left-sided tuberculous node with central necrosis in a 35-year-old woman. (a) Transverse T2-weighted image and (b) Transverse contrast-enhanced T1-weitgted image show central necrosis in the lymph node. (c) Transverse diffusion-weighted image acquired by using a b value of 800 s/mm2 shows the nodes with a hypointense necrosis surrounded by a hyperintense peripheral rim. (d) ADC map shows the central necrosis as hyperintensity. The ADC value of the solid portion was 1.09 103 mm2/s, the ADC value of necrosis was 1.20 103 mm2/s.
228
Y. Zhang et al. / Clinical Radiology 68 (2013) 224e231 Table 2 Apparent diffusion coefficient (ADC) values ( 103 mm2/s) of necrosis for metastasis and tuberculosis. Histological findings
Number
Area of necrosis (mm2)
Mean ADC value (x s)
95% CI for mean
Range
Metastatic 26 (31%) 0.11e2.96 2.02 0.36 1.88e2.17 1.27e2.65 nodes Tuberculous 38 (72%) 0.05e1.01 1.25 0.15 1.20e1.29 1.01e1.59 nodes t-value 12.045 p-value <0.001
metastasis and tuberculosis was 1.59 103 mm2/s, and the sensitivity and specificity were 88 and 100% (Table 3).
Discussion Figure 3 ADC values of the solid portion of nodes due to metastases, lymphoma, and tuberculosis.
of tuberculous nodes ((1.25 1.02) 103 mm2/s; Fig 4; Table 2).
Diagnostic ability of DWI quantitative criteria in discriminating nodes The diagnostic ability of ADC value criteria in differentiating between causes of cervical lymphadenopathy was assessed by performing ROC analysis. Fig 5aec demonstrates the ROC curves of the ADC value of the solid portion used to differentiate lymphoma from metastasis, lymphoma from tuberculosis, and metastasis from tuberculosis; the AUC values were 0.94, 0.99, and 0.69, respectively. Fig 5d shows the ROC curve of the ADC value of necrosis used to differentiate metastasis from tuberculosis, and the AUC was 0.98. By using the ADC value of the solid portion, the optimal ADC threshold values to distinguish between metastasis and lymphoma, between metastasis and tuberculosis, and between lymphoma and tuberculosis were 0.77 103, 0.98 103 and 0.81 103 mm2/s, respectively. The sensitivity and specificity values were 83 and 89%, 70 and 68%, and 93 and 100%, respectively. By using the ADC value of necrosis, the optimal ADC threshold value to distinguish between
DWI is sensitive to the diffusion of water molecules in tissue, which can make subtle abnormalities more obvious.1 The contrast of DWI depends principally on the degree to which water molecules are free to move within tissues and pass through cell membranes.8 Diffusion is expressed as an ADC, which reflects the diffusion properties unique to each type of tissue. Previously, DWI has been used to evaluate various diseases of the central nervous system. The most established clinical application of DWI for the central nervous system is in the evaluation of acute stroke. The use of DWI in the head and neck region is challenging because of susceptibility artefacts.6 However, with the development of MRI hardware and sequences, DWI has been successfully used in the head and neck region,1e7 and previous studies have reported that the ADC value was helpful in distinguishing causes of cervical lymphadenopathy.1,3,5e7 Precious studies have analysed the ADC values of lymph nodes without necrosis or the solid portions in lymph nodes
Table 1 Apparent diffusion coefficient (ADC) values ( 103 mm2/s) of the solid portion in nodes for each nodal lesion. Histological findings
Mean short Mean ADC axis (mm) value (x s) (x s)
Lymphoma (n ¼ 40) 15 4.9 Metastasis (n ¼ 84) 14 6.1 Tuberculosis (n ¼ 53) 10 3.2 F-value p-value CI, confidence interval.
95% CI for mean
Range
0.64 0.13 0.60e0.67 0.37e0.90 0.93 0.16 0.90e0.97 0.67e1.41 1.01 0.11 0.98e1.04 0.81e1.26 82.928 <0.001
Figure 4 ADC values of necrosis in lymph nodes from metastases and tuberculosis.
Y. Zhang et al. / Clinical Radiology 68 (2013) 224e231
229
Figure 5 ROC curves for the ADC value of the solid portion (aec) and necrotic portion (d) in lymph nodes to discriminate between (a) lymphoma and metastasis, (b) lymphoma and tuberculosis, (c) metastasis and tuberculosis, (d) metastasis and tuberculosis.
Table 3 Apparent diffusion coefficient (ADC) threshold value (103 mm2/s) for discrimination between metastasis, lymphoma, and tuberculosis. Nodal lesions Solid portion M and L M and TB L and TB Necrosis M and TB
ADC Threshold value
AUC
Sensitivity
Specificity
Yoden index
0.77 0.98 0.81
0.94 0.69 0.99
83% 70% 93%
89% 68% 100%
0.72 0.38 0.93
1.59
0.98
88%
100%
0.88
AUC, area under the receiver operating characteristic (ROC) curve; M, metastasis; L, lymphoma; TB, tuberculosis.
with necrosis.1,3,5e7 The consensus in those studies was that the mean ADC value of lymphomatous nodes is lower than other causes of cervical lymphadenopathy including malignancy. In this study, a similar result was obtained, i.e., the mean ADC value of lymphoma was lower than those of metastasis and tuberculosis. This result indicated that DWI is very helpful in distinguishing lymphoma from nonlymphomatous lesions. The difference in restriction of diffusion in the tissue is mainly attributed to differences in cellularity; greater cellularity and less extracellular space result in more restricted diffusion.9 Lymphoma is expected to have greater tumour cellularity than other lymphadenopathies, which may explain the lower ADC values obtained for lymphoma than for SCC and tuberculosis.
230
Y. Zhang et al. / Clinical Radiology 68 (2013) 224e231
Metastatic SCC lymph nodes are heterogeneous and contain variable degrees of necrotic parts. Based on the features, it is hypothesized that these lymph nodes are associated with alterations in water diffusivity, and DWI may play a role in their characterization. Several studies have reported the ability of DWI to discriminate malignant from benign lymph nodes in the neck.10,11 Abdel-Razek et al.1 and Holzapfel et al.7 reported that the ADC value of metastatic nodes is significantly lower than that of benign nodes. However, Sumi et al.3,5 reported significantly higher ADC values in metastatic lymph nodes than in benign lymphadenopathy. In the present study, the mean ADC value of the solid portion in metastasis was significantly lower than that of tuberculosis, but there was overlap in many cases. The AUC in ROC curve analysis was only 0.69, which was much lower than that in differentiating lymphoma from metastasis (AUC ¼ 0.94) and tuberculosis (AUC ¼ 0.99). With a threshold ADC value of 0.98 103 mm2/s, a low sensitivity of 70% and specificity of 68% were achieved in the differentiation of metastatic nodes from tuberculous nodes. This result indicated that using the ADC value of the solid portion to differentiate lymphoma from non-lymphomatous lymphadenopathy is much more effective than differentiating metastasis from tuberculosis. In this study the ADC values of necrotic portions of lymph nodes was useful in differentiating metastasis from tuberculosis. In fact, visible necrosis on MRI images is common in lymphadenopathy. Indeed, in this study, 31% of the metastatic nodes and 72% of the tuberculous nodes demonstrated necrosis. There are few differences between these two kinds of necrotic nodes on conventional MRI images. Necrosis is a factor that can influence diffusion of water molecules in tissue;2,10,12 as the amount of necrosis increases, the ADC value increases.13 Histopathological analysis has shown that micronecrosis is common in both lymphoma and SCC.4 Maeda et al.4 reported that the ADC values of tumours did not significantly differ with the degree of small foci of necrosis. Lyng et al.14 reported that the fraction of massive necrosis could be correlated with the ADC value but not the fraction of small foci of necrosis because the size of small foci of necrosis may be smaller than the voxel of the MRI images. In previous studies, only the ADC values of solid portions in lymph nodes were used to differentiate causes of cervical lymphadenopathy. Although the ADC values of necrosis in metastatic nodes were reported in the studies of both AbdelRazek et al. and Holzapfel et al., unfortunately, further analyses were not performed. In the present study, the ADC values of necrosis visible in lymph nodes was measured, and the mean ADC value of necrosis in metastasis [(2.02 0.36) 103 mm2/s] was found to be greater than that in tuberculosis [(1.25 1.02) 103 mm2/s]. It is speculated that this difference was due to different pathological mechanisms of necrosis. The degree of restriction to water diffusion in biological tissue is inversely correlated with tissue cellularity and the integrity of cell membranes. In necrotic tumours, a less cellular environment provides a larger extracellular space for diffusion of water molecules, and these molecules may also freely transgress defective or
disruptive cell membranes to move from the extracellular into the intracellular compartment,15 and this results in an increase in the ADC value. It was reported that a higher protein component in infective lesions increases the viscosity and reduces water proton mobility leading to lower ADC values.16 The ROC curve analysis indicated that the diagnostic potential of the ADC value in necrosis was much better than that in the solid portion of the node to differentiate metastasis from tuberculosis. If the ADC value of necrosis was used to distinguish between metastasis and tuberculosis, a much higher specificity and sensitivity (88% and 100%) could be obtained than by using the ADC value of the solid portion (specificity of 68%, sensitivity of 70%). This preliminary study indicates that the ADC value of necrosis in lymph nodes was very useful in characterizing cervical lymph nodes. As necrosis is common in metastasis and tuberculosis, when necrosis is detected in cervical lymph nodes, measurement of the ADC value of necrosis should be considered as a non-invasive method for differentiating metastatic nodes from tuberculous node, especially in nodes without confirmed primary tumours. A limitation of the present study was the absence of necrosis in the group of lymphoma. Although reports show that necrosis is also seen in lymphoma, lymphomatous nodes with necrosis were not found in this study. Further analysis of necrosis between lymphoma and other nodes was not undertaken. Another limitation of the present study was that the results were based on a relatively small sample of patients, so there is likely to be some uncertainty in these estimates. Therefore, it is important to perform a further study in a larger patient population in order to substantiate the role of DWI as a reliable method in differentiating among causes of cervical lymphadenopathy. The ADC values of both the necrotic and the solid portions in lymph nodes are useful in the discrimination of cervical lymphadenopathy. The ADC value of necrosis is especially helpful in differentiating metastasis from tuberculosis.
References 1. Abdel-Razek AA, Soliman NY, Elkhamary S, et al. Role of diffusion-weighted MR imaging in cervical lymphadenopathy. Eur Radiol 2006;16:1468e77. 2. Wang J, Takashima S, Takayama F, et al. Head and neck lesions: characterization with diffusion-weighted echo-planar MR imaging. Radiology 2001;220:621e30. 3. 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. AJNR Am J Neuroradiol 2003;24:1627e34. 4. Maeda M, Kato H, Sakuma H, et al. Usefulness of the apparent diffusion coefficient in line scan diffusion-weighted imaging for distinguishing between squamous cell carcinomas and malignant lymphomas of the head and neck. AJNR Am J Neuroradiol 2005;26:1186e92. 5. Sumi M, Cauteren MV, Nakamura T. MR microimaging of benign and malignant nodes in the neck. AJR Am J Roentgenol 2006;186:749e57. 6. King AD, Ahuja AT, Yeung DKW, et al. Malignant cervical lymphadenopathy: diagnostic accuracy of diffusion-weighted MR imaging. Radiology 2007;245:806e13. 7. Holzapfel K, Duetsch S, Fauser C, et al. Value of diffusion-weighted MR imaging in the differentiation between benign and malignant cervical lymph nodes. Eur J Radiol 2009;72:381e7. 8. Chan JH, Tsui EY, Luk SH, et al. Diffusion-weighted MR imaging of the liver: distinguishing hepatic abscess from cystic or necrotic tumor. Abdom Imaging 2001;26:161e5.
Y. Zhang et al. / Clinical Radiology 68 (2013) 224e231 9. Guo AC, Cummings TJ, Dash RC, et al. Lymphomas and high-grade astrocytomas: comparison of water diffusibility and histologic characteristics. Radiology 2002;224:177e83. 10. White ML, Zhang Y, Robinson RA. Evaluating tumors and tumors and tumor like lesions of the nasal cavity, the paranasal sinuses, and the adjacent skull base with diffusion-weighted MRI. J Comput Assist Tomogr 2006;20:490e5. 11. Chong VF, Fan YF, Khoo JB. MRI features of cervical nodal necrosis in metastatic disease. Clin Radiol 1996;51:103e9. 12. Kim YJ, Chang KH, Song IC, et al. Brain abscess and necrotic or cystic brain tumor: discrimination with signal intensity on diffusion weighted imaging. AJR Am J Roentgenol 1998;171:1487e90.
231
13. Lang P, Wendland MF, Saeed M, et al. Osteogenic sarcoma: noninvasive in vivo assessment of tumor necrosis with diffusion-weighted MR imaging. Radiology 1998;206:227e35. 14. Lyng H, Haraldseth O, Rofstad EK. Measurement of cell density and necrotic fraction in human melanoma xenografts by diffusion weighted magnetic resonance imaging. Magn Reson Med 2000;43: 828e36. 15. Koh DW, Collins DJ. Diffusion-weighted MRI in the body: applications and challenges in oncology. AJR Am J Roentgenol 2007;188: 1622e35. 16. Mishra AM, Gupta RK, Saksena S, et al. Biological correlates of diffusivity in brain abscess. Magn Reson Med 2005;54:878e85.