- Email: [email protected]
Orbital benign and malignant lymphoproliferative disorders: Differentiation using semi-quantitative and quantitative analysis of dynamic contrast-enhanced magnetic resonance imaging
European Journal of Radiology 88 (2017) 88–94
Contents lists available at ScienceDirect
European Journal of Radiology journal homepage: www.elsevier.com/locate/ejrad
Orbital benign and malignant lymphoproliferative disorders: Differentiation using semi-quantitative and quantitative analysis of dynamic contrast-enhanced magnetic resonance imaging Hao Hu a,1 , Xiao-Quan Xu a,1 , Hu Liu b , Xun-Ning Hong a , Hai-Bin Shi a , Fei-Yun Wu a,∗ a b
Department of Radiology, First Affiliated Hospital of Nanjing Medical University, Nanjing, China Department of Ophthalmology, First Affiliated Hospital of Nanjing Medical University, Nanjing, China
a r t i c l e
i n f o
Article history: Received 13 May 2016 Received in revised form 17 December 2016 Accepted 28 December 2016 Keywords: Orbit Lymphoproliferative disorders Differential diagnosis Magnetic resonance imaging Dynamic contrast-enhanced
a b s t r a c t Objectives: To assess the value of dynamic contrast-enhanced MR imaging (DCE-MRI) in differentiating benign from malignant orbital lymphoproliferative disorders (OLPDs). Methods: Thirty-nine patients with orbital lymphoproliferative disorders (21 malignant and 18 benign) underwent DCE-MRI scan for pre-treatment evaluation from March 2013 to December 2015. Both semi-quantitative (TTP, AUC, Slopemax ) and quantitative (Ktrans , kep , ve ) parameters were calculated, and compared between two groups. Receiver operating characteristic (ROC) curve analyses were used to determine the diagnostic value of each significant parameter. Results: Malignant OLPDs showed significantly higher kep , lower ve , and lower AUC than benign OLPDs, while no significant differences were found on Ktrans , TTP and Slopemax . ROC analyses indicated that ve exhibited the best diagnostic performance in predicting malignant OLPDs (cutoff value, 0.211; area under the curve, 0.896; sensitivity, 76.2%; specificity, 94.9%), followed by kep (cutoff value, 0.853; area under the curve, 0.839; sensitivity, 85.7%; specificity, 89.9%). Conclusion: DCE-MRI and specially its derived quantitative parameters of kep and ve are promising metrics for differentiating malignant from benign OLPDs. © 2017 Elsevier Ireland Ltd. All rights reserved.
1. Introduction Orbital lymphoproliferative disorders (OLPDs) constitute ten to fifteen percent of orbital masses [1]. They represent a broad spectrum of benign and malignant diseases, including lymphoid hyperplasia, atypical lymphoid hyperplasia, ocular adnexal lymphoma and idiopathic inflammatory pseudotumor [1,2]. Along with these, IgG4-related ophthalmic disease is becoming increasingly recognized and classified into benign OLPD group based on recent surveillance [3,4]. Differentiation of benign and malignant OLPDs is very crucial, because of the different treatment strategy and prognosis [5,6]. Orbital lymphomas are amenable to low-dose radiation therapy, while the benign mimics often exhibit a good response to corticosteroid therapy. The value of using clinical criteria for differentiating benign and malignant OLPDs is limited, because they
∗ Corresponding author at: Department of Radiology, The First Affiliated Hospital of Nanjing Medical University, No. 300, Guangzhou Road, Nanjing, China. E-mail address: wfydd [email protected] (F.-Y. Wu). 1 Dr. Hao Hu and Xiao-Quan Xu contribute equally to this work. http://dx.doi.org/10.1016/j.ejrad.2016.12.035 0720-048X/© 2017 Elsevier Ireland Ltd. All rights reserved.
often share similar clinical presentation [1,7,8]. Therefore, to find an efficient method to differentiate these two entities is in urgent need. Recently, dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) which utilizes fast T1-weighted imaging following a bolus injection of gadolinium contrast agent, has been increasingly used to assess the hemodynamic information of various tumors [9–11]. It allows noninvasive assessment of vascular permeability and blood flow, and has the potential to detect and characterize tumors, as well as evaluate treatment response [12,13]. Previous studies have confirmed that DCE-MRI and its derived quantitative metrics were helpful for predicting orbital malignancy [14,15]. However, they did not focus on the OLPDs, and enrolled several other orbital disorders. Few studies that specially used the DCE-MRI to discriminate benign from malignant OLPDs have been reported till now. In addition, they processed the DCE-MRI data using the model-free method, however we know that the main drawback of the model-free analysis is that they do not necessarily correlate with the physical essence [12]. Besides the model-free method, another method is the model-based
H. Hu et al. / European Journal of Radiology 88 (2017) 88–94
89
Table 1 Patient characteristics and pathologic findings of OLPD cases. Variable
Malignant OLPDs (n = 21)
Benign OLPDs (n = 18)
P value
Age Gender (M/F) Histologic subtypes
63.10 ± 14.83 14/7
50.39 ± 14.37 13/5
0.010 0.742
MALT lymphoma (17) DLBCL (2) Follicular lymphoma (2)
IIP (8) RLH (6) IgG4-related disease (4)
Note: M indicates male; F, female; MALT, mucosa-associated lymphoid tissue; DLBCL, diffuse large B-cell lymphoma; IIP, idiopathic inflammatory pseudotumor; RLH, reactive lymphoid hyperplasia. Data in parentheses indicates the number of the corresponding patients in our study.
calculation which is preferable as it provides greater pathophysiological insight [16]. Therefore, the aim of this study was to assess the value of DCE-MRI derived perfusion parameters, including both semiquantitative and quantitative measurements, for differentiating benign from malignant OLPDs. 2. Materials and methods 2.1. Patient population Our institutional review board approved this study and waived the informed consent requirement due to the retrospective nature. From March 2013 to December 2015, fifty-eight consecutive OLPDs patients underwent orbital MRI examination for pre-treatment evaluation. Nineteen patients were excluded because of the following exclusion criteria: no available DCE-MR images (n = 12), no adequate imaging quality (n = 1), lesions with diameter less than 1 cm (n = 2), secondary lymphoma (n = 1), prior history of corticosteroid or radiation therapy before MRI scan (n = 3). Finally, 39 OLPDs patients (21 malignant and 18 benign, 27 men and 12 women, mean age, 57.23 ± 15.79 years old) were enrolled in our study. The spectrum of OPLDs included: 1) malignant lesions (n = 21; 14 men and 7 women; mean age, 63.10 ± 14.83 years old): recorded as MALT lymphoma (n = 17), DLBCL (n = 2), and follicular lymphoma (n = 2). 2) Benign lesions (n = 18; 13 men and 5 women; mean age, 50.39 ± 14.37 years old): recorded as idiopathic inflammatory pseudotumor (n = 8), reactive lymphoid hyperplasia (n = 6), and IgG4-related ophthalmic disease (n = 4). Detailed demographic and pathologic information of our study population are displayed on Table 1. The final diagnosis was made based on the surgically pathological results in 35 patients, on the follow-up after steroid treatment in 4 patients with inflammatory pseudotumor. 2.2. MRI scan MR images were obtained using a 3T MR scanner (Verio; Siemens, Germany) with a 12-channel head coil. All patients underwent conventional unenhanced axial T1-weighted imaging (repetition time [TR]/echo time [TE], 600/10 msec), axial T2weighted imaging (TR/TE, 4700/79 msec) with fat saturation, and coronal T2-weighted imaging (TR/TE, 3500/79 msec) with fat saturation. Then the dynamic images were obtained by using a twodimensional (2D) turbo fast low angle shot (FLASH) sequence with integrated parallel acquisition technique (iPAT). Gadoliniumdiethylene triamine pentaacetic acid (Magnevist; Bayer Schering Pharma AG, Berlin, Germany) was intravenously bolus injected via a power injector at the rate of 4 mL/s at the dose of 0.1 mmol/kg, followed by a 20-mL bolus of saline administered at the same injection rate. Before the dynamic acquisition, an unenhanced T1 map based on a dual flip angle of 5◦ and 12◦ was obtained by using
the same sequence, which allows conversion of the changes of MR signal intensity into those of the gadolinium concentration during passage of the contrast agent [17]. The DCE acquisition consisted of 5 baseline sets and 90 contrastenhanced sets of images (total: 95 dynamics) without delay between acquisitions. The temporal resolution was 3.3 s, and the total acquisition time was 5 min 15 s. The other detailed imaging parameters for the DCE imaging were as follows: TR/TE, 474.66/1.43 msec; flip angle (FA), 12◦ ; Average, 1; field of view (FOV), 230 mm; matrix, 128 × 128; section thickness, 4.5 mm; number of sections, 7. After DCE-MRI scan, post-contrast axial, coronal and sagittal T1weighted images were obtained.
2.3. Imaging processing DCE-MR images were processed using a dedicated postprocessing software program (Omni-Kinetics; GE Healthcare) which supplies pharmacokinetic calculation on a pixel-by-pixel basis. The current tracer-kinetic modeling for quantitation of DCE images was based on a two-compartment modified Tofts model [18]. In terms of the arterial input function (AIF), it was extracted by manually drawing a small circle region of interest (ROI) on one side of carotid artery located proximal to the tumor [19]. Whole-tumor ROIs were manually drawn over DCE-MR images, and then voxel-wise perfusion maps, containing both model-free (semi-quantitative) and model-based (quantitative) parameters were automatically generated. Semi-quantitative parameters included AUC (Area under the gadolinium dynamic curve in mmol*min), TTP (Time from contrast arrival to peak in min) and Slopemax (Maximum concentration-time ratio in min−1 ). Quantitative parameters included Ktrans (the volume transfer constant between the plasma and the extracellular extravascular space [EES] in min−1 ), ve (the volume fraction of the EES in ml/ml), and kep (the rate constant from EES to blood plasma in min−1 , which equals the ratio Ktrans /ve ) [12,13]. In terms of the ROIs placement, they were outlined on all slices by encompassing as much as tumor area, while the visual necrotic, hemorrhagic areas and surrounding blood vessels were excluded with reference to the conventional MR images. To minimize the effect of partial volume, the edges of lesions were avoided. If bilateral lesions occurred, the lesion with the larger diameter was included for analysis. All the quantitative measurements were performed independently by two neuro-radiologists (reader 1: with 6 years of experience; reader 2: with 4 years of experience) who were blinded to the clinical information, pathological results and study design. The measurement results of these two readers were used to evaluate the inter-observer reproducibility. Meanwhile, to evaluate the intra-observer reproducibility, all the DCE-MR images were assessed again by the reader 1, spaced at least one month. The average of the two measurement results of reader 1 was adopted into statistical analysis.
90
H. Hu et al. / European Journal of Radiology 88 (2017) 88–94
2.4. Statistical analysis Numeric data were averaged over all patients and reported as the mean ± standard deviation. The Kolmogorov-Smirnov test was used for normally distributed analysis. If normally distributed, the difference would be tested using t-test. Otherwise, it would be tested using Mann-Whitney U test. The significance threshold for difference on DCE-MRI parameters was set at a P value less
than 0.010 (0.050/5) for Bonferroni multiple comparison correction. Receiver operating characteristic (ROC) curve analyses were performed to determine the value of each significant parameter in predicting malignant OPLDs. Sensitivity and specificity were calculated with a threshold criterion determined as the value would maximize the Youden index. The inter- and intra-observer reproducibility of DCE-MRI parameters measurement were evaluated using intraclass correlation coefficient (ICC) with 95% confidence
Fig. 1. Box-and-whisker plots show the comparisons of DCE-MRI derived parameters between benign and malignant OLPDs groups.
H. Hu et al. / European Journal of Radiology 88 (2017) 88–94 Table 2 DCE-MRI derived parameters of benign and malignant OLPDs. Parameters
Malignant OLPDs (n = 21)
Benign OLPDs (n = 18)
P value
Ktrans (min−1 ) kep (min−1 ) ve (ml/ml) AUC (mmol*min) TTP (min) Slopemax (min−1 )
0.212 ± 0.787 1.098 ± 0.445 0.210 ± 0.050 6.737 ± 1.781 2.325 ± 0.628 7.456 ± 3.157
0.228 ± 0.086 0.751 ± 0.266 0.328 ± 0.093 8.367 ± 1.569 3.022 ± 1.086 7.955 ± 3.262
0.587 <0.001 <0.001 0.005 0.020 0.549
Note: The numeric data were reported as the mean ± standard deviation. Data in parentheses indicates the unit of the corresponding parameters. For definitions of the indicated parameters please refer to the Image Processing section of the Materials and Methods.
Fig. 2. ROC curves of using DCE-MRI derived parameters to predict malignant OLPDs.
intervals and applying a two-way ICC with random rater assumption. The ICC value ranges between 0 and 1.00, and values closer to 1.00 represented better reproducibility. The ICC was interpreted as follows: (< 0.40, poor; 0.40–0.60, moderate; 0.61–0.80, good; ≥ 0.81, excellent). All statistical analyses were performed using the SPSS software package (version 19.0, SPSS, Chicago, IL). 3. Results There was significant difference on the age (P = 0.010), while no difference on the gender distribution (P = 0.742) of patients between benign and malignant OLPDs groups. Table 2 summarizes the detailed comparisons of DCE-MRI derived parameters between benign and malignant OLPDs groups. Malignant OPLDs had significantly higher kep (P < 0.001), lower ve (P < 0.001), and lower AUC (P = 0.005) than benign mimics. However, there were no significant differences on Ktrans (P = 0.587), TTP (P = 0.020) and Slopemax (P = 0.549) between two groups. The comparisons of DCEMRI parameters between two groups are plotted in Fig. 1. ROC curve analyses indicated that, ve exhibited the best diagnostic performance in predicting malignant OPLDs (cutoff value, 0.211; area under the curve, 0.896; sensitivity, 76.2%; specificity, 94.9%), followed by kep (cutoff value, 0.853; area under the curve, 0.839; sensitivity, 85.7%; specificity, 89.9%) and AUC (cutoff value, 7.457; area under the curve, 0.783; sensitivity, 76.2%; specificity, 72.2%). The ROC curves regarding using DCE-MRI derived parameters to discriminate malignant from benign OLPDs are shown in Fig. 2. kep showed the highest diagnostic sensitivity (85.7%) and ve demonstrated the highest diagnostic specificity (94.9%). Repre-
91
sentative cases of orbital lymphoma and idiopathic inflammatory pseudotumor are shown in Figs. 3 and 4. Excellent intra- and inter-observer reproducibility were obtained for both the semi-quantitative and quantitative measurements of DCE-MRI derived parameters (intra-observer ICCs: Ktrans , 0.983; kep , 0.989; ve , 0.985; AUC, 0.976; TTP, 0.984; Slopemax , 0.980, and inter-observer ICCs: Ktrans , 0.969; kep , 0.978; ve , 0.973; AUC, 0.940; TTP, 0.971; Slopemax , 0.954).
4. Discussion In consideration of the entirely different therapeutic strategies for benign and malignant OLPDs, accurate pre-treatment differentiation of these two entities is very crucial for the individualized treatment. Our present study results indicated that DCE-MRI derived parameters, especially kep and ve , could help to distinguish benign from malignant OLPDs. kep demonstrated the highest diagnostic sensitivity, and ve showed the highest diagnostic specificity. To the best of our acknowledgement, our study is the first one that applying model-based analysis of DCE-MRI to differentiate benign from malignant OLPDs till now. Present study founded that malignant OLPDs demonstrated significantly lower AUC value than the benign mimics. The AUC value was viewed as a metric about the integration of the concentration of contrast agent observed in the tissue of interest over time [20]. Previous study by Haradome K. et al. has reported that the benign OLPDs showed higher contrast-enhancement ratio [6]. Meanwhile, Yuan Y et al. reported that the TIC pattern of benign OLPDs mostly showed as Type I or II (persistent or plateau pattern), while malignant ones mostly showed as Type III (washout pattern) [15]. In comprehensive consideration of the two reasons above, naturally the benign OLPDs would show the higher AUC than the malignant ones. However, AUC did not exhibit the attractive diagnostic ability (area under the curve, 0.783; sensitivity, 76.2%; specificity, 72.2%) for predicting orbital lymphoma in our study. It might be associated with the fact that the AUC value represents a conglomerate of physiologic processes, including blood flow, vascular permeability and fractional interstitial space [14]. Model-free analysis did not provide accurate histological insight of the tumor tissue, and therefore model-based metrics providing more histological information would be needed. Similar with previous studies those focused on lung cancer [9] and glioblastoma [21], malignant OLPDs in our study also exhibited significantly lower ve than the benign mimics. Previously numerous studies have noted that lymphoma was a hyper-cellularity tumor that constituted by numerous uniformly small-sized atypical lymphocytes [6,8,22,23]. The higher cellularity of lymphoma would result in more limited extracellular space. By contrast, the most important histological characteristic of benign OLPDs was interstitial edema, which would lead to the increased extracellular space [6]. Then, it was not surprising that lymphoma would exhibit lower ve than benign OLPDs, after considering that ve was a quantitative metric associated with the volume fraction of the extracellular extravascular space [24]. In our study, malignant OLPDs showed higher kep than benign mimics, which was consistent with one previous study of Ro SR et al. [25]. They compared the DCE-MRI derived parameters between overall orbital benign and malignant tumors, no just focusing on the lymphoproliferative disorders, and they explained that the higher kep might reflect the higher vascular permeability within the malignant tumors [25]. Meanwhile, they found orbital malignant tumors exhibited higher Ktrans than benign ones, however Ktrans did not differ significantly between malignant and benign OLPDs in our study. As to the possible interpretation, the mutability of Ktrans might be potential reason. Paldino MJ et al. [26]
92
H. Hu et al. / European Journal of Radiology 88 (2017) 88–94
Fig. 3. Representative images of a 79-year-old man with left orbital lymphoma. Pre-treatment DCE-MR imaging was conducted and ROIs were outlined on all slices (a). b–d show kep , ve , and AUC map, respectively. Obtained TIC shows as a washout pattern (e).
has ever clarified that, Ktrans could be influenced by several factors including blood flow to tissue, microvascular attenuation, vascular permeability, and fractional volume of EES (ve ). Knowing that lymphoma was a relatively poorer vascularized tumor
compared with the benign mimics [8], the local microvascular attenuation and blood flow might be decreased within lymphoma tissue, resulting in attenuation of Ktrans value. What’s more, given the proportional relationship between ve and Ktrans under a settled
H. Hu et al. / European Journal of Radiology 88 (2017) 88–94
93
Fig. 4. Representative images of a 45-year-old woman with right orbital idiopathic inflammatory pseudotumor. Pre-treatment DCE-MR imaging was exhibited and ROIs were outlined on all slices (a). b–d demonstrate kep , ve , and AUC map, respectively. Obtained TIC shows as a plateau pattern (e).
kep [12,13], the significantly decreased ve of lymphoma should also be responsible for the variation of Ktrans . Thus, we speculated that, although the vascular permeability increased within lymphoma tissue, which could assist the elevation of Ktrans value, the met-
ric might be more efficiently affected by other factors, such as the decreased microvascular attenuation, the lack of local blood flow and the decreased fractional volume of EES. Hence, kep might be
94
H. Hu et al. / European Journal of Radiology 88 (2017) 88–94
determined to a larger extent by vascular permeability than Ktrans , as in a previous study Zwick S et al. has clarified [27]. In terms of the measurement reproducibility, our study showed excellent inter- and intra-observer agreements for all quantitative measurements, with all ICCs more than 0.900, which seemed to be better than previous DCE-MRI-related studies [9,14,15]. The higher ICCs obtained in our study might be associated with the following reasons. First, our study subjects were OLPDs which were characterized by the homogeneous imaging appearance [1,6]. Tumor necrosis or hemorrhage were seldom found in OLPDs, and this characteristic would do beneficial to the placement of ROI. Second, the whole-tumor ROI approach was used for quantitative measurements, which was different from the selected ROI placement in previous studies [9,14,15,25]. As previous study indicated, wholetumor ROI approach could effectively reduce the inter-observer variability in ROI placement [28]. In addition to the limitations intrinsic to any retrospective study, several other limitations should be mentioned. First, the number of study population was relatively small. Further study with larger sample size would be needed to verify our results. Second, the imaging features of conventional MR images and DWI has also been proven to be useful in differentiating benign from malignant OLPDs. Conventional MR imaging focused on the tumor morphology, DWI focused on the tumorous cellular density, and DCE-MRI focused on the tumor angiogenesis. Further study combined using multi-modal MRI techniques would be more valuable in the differentiation of benign and malignant OLPDs. 5. Conclusion In conclusion, we present out initial clinical experience that using both semi-quantitative and quantitative analysis of DCE-MRI in the field of OLPDs. Our preliminary results suggest that quantitative DCE-MRI, especially kep and ve value, can help to differentiate benign from malignant OLPDs. Conflict of interest There was no any conflict of interest in this article. Funding information None. References [1] G. Akansel, L. Hendrix, B.A. Erickson, A. Demirci, A. Papke, A. Arslan, et al., MRI patterns in orbital malignant lymphoma and atypical lymphocytic infiltrates, Eur. J. Radiol. 53 (2) (2005) 175–181. [2] G.M. Espinoza, Orbital inflammatory pseudotumors: etiology, differential diagnosis, and management, Curr. Rheumatol. Rep. 12 (6) (2010) 443–447. [3] M. Takahira, Y. Ozawa, M. Kawano, Y. Zen, S. Hamaoka, K. Yamada, et al., Clinical aspects of IgG4-related orbital inflammation in a case series of ocular adnexal lymphoproliferative disorders, Int. J. Rheumatol. 2012 (2012) 635473. [4] A. Fujita, O. Sakai, M.N. Chapman, H. Sugimoto, IgG4-related disease of the head and neck: CT and MR imaging manifestations, Radiographics 32 (7) (2012) 1945–1958. [5] G.E. Valvassori, S.S. Sabnis, R.F. Mafee, M.S. Brown, A. Putterman, Imaging of orbital lymphoproliferative disorders, Radiol. Clin. North Am. 37 (1) (1999) 135–150. [6] K. Haradome, H. Haradome, Y. Usui, S. Ueda, T.C. Kwee, K. Saito, et al., Orbital lymphoproliferative disorders (OLPDs): value of MR imaging for differentiating orbital lymphoma from benign OPLDs, AJNR Am. J. Neuroradiol. 35 (10) (2014) 1976–1982.
[7] A.L. Weber, L. Vitale Romo, N.R. Sabates, Pseudotumor of the orbit: clinical, pathologic, and radiologic evaluation, Radiol. Clin. North Am. 37 (1) (1999) 151–168. [8] R. Kapur, A.R. Sepahdari, M.F. Mafee, A.M. Putterman, V. Aakalu, L.J. Wendel, et al., MR imaging of orbital inflammatory syndrome, orbital cellulitis, and orbital lymphoid lesions: the role of diffusion-weighted imaging, AJNR Am. J. Neuroradiol. 30 (1) (2009) 64–70. [9] M. Yuan, Y.D. Zhang, C. Zhu, T.F. Yu, H.B. Shi, Z.F. Shi, et al., Comparison of intravoxel incoherent motion diffusion-weighted MR imaging with dynamic contrast-enhanced MRI for differentiating lung cancer from benign solitary pulmonary lesions, J. Magn. Reson. Imaging 43 (3) (2016) 669–679. [10] M.A. Bali, T. Metens, V. Denolin, M. Delhaye, P. Demetter, J. Closset, Tumoral and nontumoral pancreas: correlation between quantitative dynamic contrast-enhanced MR imaging and histopathologic parameters, Radiology 261 (2) (2011) 456–466. [11] C. Chakiba, F. Cornelis, E. Descat, M. Gross-Goupil, P. Sargos, G. Roubaud, et al., Dynamic contrast enhanced MRI-derived parameters are potential biomarkers of therapeutic response in bladder carcinoma, Eur. J. Radiol. 84 (6) (2015) 1023–1028. [12] S. Gaddikeri, R.S. Gaddikeri, T. Tailor, Y. Anzai, Dynamic contrast-enhanced MR imaging in head and neck cancer: techniques and clinical applications, AJNR Am. J. Neuroradiol. 37 (4) (2015) 588–595. [13] T. Chikui, M. Obara, A.W. Simonetti, M. Ohga, S. Koga, S. Kawano, 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. [14] J. Xian, Z. Zhang, Z. Wang, J. Li, B. Yang, F. Man, et al., Value of MR imaging in the differentiation of benign and malignant orbital tumors in adults, Eur. Radiol. 20 (7) (2010) 1692–1702. [15] Y. Yuan, X.P. Kuai, X.S. Chen, X.F. Tao, Assessment of dynamic contrast-enhanced magnetic resonance imaging in the differentiation of malignant from benign orbital masses, Eur. J. Radiol. 82 (9) (2013) 1506–1511. [16] M.J. Koh, H.S. Kim, C.G. Choi, S.J. Kim, Which is the best advanced MR imaging protocol for predicting recurrent metastatic brain tumor following gamma-knife radiosurgery: focused on perfusion method, Neuroradiology 57 (4) (2015) 367–376. [17] I. Ocak, M. Bernardo, G. Metzger, T. Barrett, P. Pinto, P.S. Albert, et al., Dynamic contrast-enhanced MRI of prostate cancer at 3 T: a study of pharmacokinetic parameters, Am. J. Roentgenol. 189 (4) (2007) 192–201. [18] P.S. Tofts, G. Brix, D.L. Buckley, J.L. Evelhoch, E. Henderson, M.V. Knopp, et al., Estimating kinetic parameters from dynamic contrast-enhanced T1-weighted MRI of a diffusable tracer: standardized quantities and symbols, J. Magn. Reson. Imaging 10 (3) (1999) 223–232. [19] Y.R. Kim, K.J. Rebro, K.M. Schmainda, Water exchange and inflow affect the accuracy of T1-GRE blood volume measurements: implications for the evaluation of tumor angiogenesis, Magn. Reson. Med. 47 (6) (2002) 1110–1120. [20] T.E. Yankeelov, J.C. Gore, Dynamic contrast enhanced magnetic resonance imaging in oncology: theory, data acquisition, analysis, and examples, Curr. Med. Imaging Rev. 3 (2) (2009) 91–107. [21] K. Özduman, E. Yıldız, A. Dinc¸er, A. Sav, M.N. Pamir, Using intraoperative dynamic contrast-enhanced T1-weighted MRI to identify residual tumor in glioblastoma surgery, J. Neurosurg. 120 (1) (2014) 60–66. [22] X. Wu, H. Pertovaara, P. Dastidar, M. Vornanen, L. Paavolainen, V. Marjomäki, et al., ADC measurements in diffuse large B-cell lymphoma and follicular lymphoma: a DWI and cellularity study, Eur. J. Radiol. 82 (4) (2013) e158–64. [23] X.Q. Xu, H. Hu, G.Y. Su, H. Liu, X.N. Hong, H.B. Shi, et al., Utility of histogram analysis of ADC maps for differentiating orbital tumors, Diagn. Interv. Radiol. 22 (2) (2016) 161–167. [24] M. Park, J. Kim, Y.S. Choi, S.K. Lee, Y.W. Koh, S.H. Kim, et al., Application of dynamic contrast-enhanced MRI parameters for differentiating squamous cell carcinoma and malignant lymphoma of the oropharynx, AJR Am. J. Roentgenol. 206 (2) (2016) 401–407. [25] S.R. Ro, P. Asbach, E. Siebert, E. Bertelmann, B. Hamm, K. Erb-Eigner, Characterization of orbital masses by multiparametric MRI, Eur. J. Radiol. 85 (2) (2016) 324–336. [26] M.J. Paldino, D.P. Barboriak, Fundamentals of quantitative dynamic contrast-enhanced MR imaging. Fundamentals of quantitative dynamic contrast-enhanced MR imaging, Magn. Reson. Imaging Clin. N. Am. 17 (2) (2009) 277–289. [27] S. Zwick, G. Brix, P.S. Tofts, R. Strecker, A. Kopp-Schneider, H. Laue, et al., Simulation-based comparison of two approaches frequently used for dynamic contrast-enhanced MRI, Eur. Radiol. 20 (2) (2010) 432–442. [28] C.J. Wu, Q. Wang, H. Li, X.N. Wang, X.S. Liu, H.B. Shi, et al., DWI-associated entire-tumor histogram analysis for the differentiation of low-grade prostate cancer from intermediate-high-grade prostate cancer, Abdom. Imaging 40 (8) (2015) 3214–3221.
Contents lists available at ScienceDirect
European Journal of Radiology journal homepage: www.elsevier.com/locate/ejrad
Orbital benign and malignant lymphoproliferative disorders: Differentiation using semi-quantitative and quantitative analysis of dynamic contrast-enhanced magnetic resonance imaging Hao Hu a,1 , Xiao-Quan Xu a,1 , Hu Liu b , Xun-Ning Hong a , Hai-Bin Shi a , Fei-Yun Wu a,∗ a b
Department of Radiology, First Affiliated Hospital of Nanjing Medical University, Nanjing, China Department of Ophthalmology, First Affiliated Hospital of Nanjing Medical University, Nanjing, China
a r t i c l e
i n f o
Article history: Received 13 May 2016 Received in revised form 17 December 2016 Accepted 28 December 2016 Keywords: Orbit Lymphoproliferative disorders Differential diagnosis Magnetic resonance imaging Dynamic contrast-enhanced
a b s t r a c t Objectives: To assess the value of dynamic contrast-enhanced MR imaging (DCE-MRI) in differentiating benign from malignant orbital lymphoproliferative disorders (OLPDs). Methods: Thirty-nine patients with orbital lymphoproliferative disorders (21 malignant and 18 benign) underwent DCE-MRI scan for pre-treatment evaluation from March 2013 to December 2015. Both semi-quantitative (TTP, AUC, Slopemax ) and quantitative (Ktrans , kep , ve ) parameters were calculated, and compared between two groups. Receiver operating characteristic (ROC) curve analyses were used to determine the diagnostic value of each significant parameter. Results: Malignant OLPDs showed significantly higher kep , lower ve , and lower AUC than benign OLPDs, while no significant differences were found on Ktrans , TTP and Slopemax . ROC analyses indicated that ve exhibited the best diagnostic performance in predicting malignant OLPDs (cutoff value, 0.211; area under the curve, 0.896; sensitivity, 76.2%; specificity, 94.9%), followed by kep (cutoff value, 0.853; area under the curve, 0.839; sensitivity, 85.7%; specificity, 89.9%). Conclusion: DCE-MRI and specially its derived quantitative parameters of kep and ve are promising metrics for differentiating malignant from benign OLPDs. © 2017 Elsevier Ireland Ltd. All rights reserved.
1. Introduction Orbital lymphoproliferative disorders (OLPDs) constitute ten to fifteen percent of orbital masses [1]. They represent a broad spectrum of benign and malignant diseases, including lymphoid hyperplasia, atypical lymphoid hyperplasia, ocular adnexal lymphoma and idiopathic inflammatory pseudotumor [1,2]. Along with these, IgG4-related ophthalmic disease is becoming increasingly recognized and classified into benign OLPD group based on recent surveillance [3,4]. Differentiation of benign and malignant OLPDs is very crucial, because of the different treatment strategy and prognosis [5,6]. Orbital lymphomas are amenable to low-dose radiation therapy, while the benign mimics often exhibit a good response to corticosteroid therapy. The value of using clinical criteria for differentiating benign and malignant OLPDs is limited, because they
∗ Corresponding author at: Department of Radiology, The First Affiliated Hospital of Nanjing Medical University, No. 300, Guangzhou Road, Nanjing, China. E-mail address: wfydd [email protected] (F.-Y. Wu). 1 Dr. Hao Hu and Xiao-Quan Xu contribute equally to this work. http://dx.doi.org/10.1016/j.ejrad.2016.12.035 0720-048X/© 2017 Elsevier Ireland Ltd. All rights reserved.
often share similar clinical presentation [1,7,8]. Therefore, to find an efficient method to differentiate these two entities is in urgent need. Recently, dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) which utilizes fast T1-weighted imaging following a bolus injection of gadolinium contrast agent, has been increasingly used to assess the hemodynamic information of various tumors [9–11]. It allows noninvasive assessment of vascular permeability and blood flow, and has the potential to detect and characterize tumors, as well as evaluate treatment response [12,13]. Previous studies have confirmed that DCE-MRI and its derived quantitative metrics were helpful for predicting orbital malignancy [14,15]. However, they did not focus on the OLPDs, and enrolled several other orbital disorders. Few studies that specially used the DCE-MRI to discriminate benign from malignant OLPDs have been reported till now. In addition, they processed the DCE-MRI data using the model-free method, however we know that the main drawback of the model-free analysis is that they do not necessarily correlate with the physical essence [12]. Besides the model-free method, another method is the model-based
H. Hu et al. / European Journal of Radiology 88 (2017) 88–94
89
Table 1 Patient characteristics and pathologic findings of OLPD cases. Variable
Malignant OLPDs (n = 21)
Benign OLPDs (n = 18)
P value
Age Gender (M/F) Histologic subtypes
63.10 ± 14.83 14/7
50.39 ± 14.37 13/5
0.010 0.742
MALT lymphoma (17) DLBCL (2) Follicular lymphoma (2)
IIP (8) RLH (6) IgG4-related disease (4)
Note: M indicates male; F, female; MALT, mucosa-associated lymphoid tissue; DLBCL, diffuse large B-cell lymphoma; IIP, idiopathic inflammatory pseudotumor; RLH, reactive lymphoid hyperplasia. Data in parentheses indicates the number of the corresponding patients in our study.
calculation which is preferable as it provides greater pathophysiological insight [16]. Therefore, the aim of this study was to assess the value of DCE-MRI derived perfusion parameters, including both semiquantitative and quantitative measurements, for differentiating benign from malignant OLPDs. 2. Materials and methods 2.1. Patient population Our institutional review board approved this study and waived the informed consent requirement due to the retrospective nature. From March 2013 to December 2015, fifty-eight consecutive OLPDs patients underwent orbital MRI examination for pre-treatment evaluation. Nineteen patients were excluded because of the following exclusion criteria: no available DCE-MR images (n = 12), no adequate imaging quality (n = 1), lesions with diameter less than 1 cm (n = 2), secondary lymphoma (n = 1), prior history of corticosteroid or radiation therapy before MRI scan (n = 3). Finally, 39 OLPDs patients (21 malignant and 18 benign, 27 men and 12 women, mean age, 57.23 ± 15.79 years old) were enrolled in our study. The spectrum of OPLDs included: 1) malignant lesions (n = 21; 14 men and 7 women; mean age, 63.10 ± 14.83 years old): recorded as MALT lymphoma (n = 17), DLBCL (n = 2), and follicular lymphoma (n = 2). 2) Benign lesions (n = 18; 13 men and 5 women; mean age, 50.39 ± 14.37 years old): recorded as idiopathic inflammatory pseudotumor (n = 8), reactive lymphoid hyperplasia (n = 6), and IgG4-related ophthalmic disease (n = 4). Detailed demographic and pathologic information of our study population are displayed on Table 1. The final diagnosis was made based on the surgically pathological results in 35 patients, on the follow-up after steroid treatment in 4 patients with inflammatory pseudotumor. 2.2. MRI scan MR images were obtained using a 3T MR scanner (Verio; Siemens, Germany) with a 12-channel head coil. All patients underwent conventional unenhanced axial T1-weighted imaging (repetition time [TR]/echo time [TE], 600/10 msec), axial T2weighted imaging (TR/TE, 4700/79 msec) with fat saturation, and coronal T2-weighted imaging (TR/TE, 3500/79 msec) with fat saturation. Then the dynamic images were obtained by using a twodimensional (2D) turbo fast low angle shot (FLASH) sequence with integrated parallel acquisition technique (iPAT). Gadoliniumdiethylene triamine pentaacetic acid (Magnevist; Bayer Schering Pharma AG, Berlin, Germany) was intravenously bolus injected via a power injector at the rate of 4 mL/s at the dose of 0.1 mmol/kg, followed by a 20-mL bolus of saline administered at the same injection rate. Before the dynamic acquisition, an unenhanced T1 map based on a dual flip angle of 5◦ and 12◦ was obtained by using
the same sequence, which allows conversion of the changes of MR signal intensity into those of the gadolinium concentration during passage of the contrast agent [17]. The DCE acquisition consisted of 5 baseline sets and 90 contrastenhanced sets of images (total: 95 dynamics) without delay between acquisitions. The temporal resolution was 3.3 s, and the total acquisition time was 5 min 15 s. The other detailed imaging parameters for the DCE imaging were as follows: TR/TE, 474.66/1.43 msec; flip angle (FA), 12◦ ; Average, 1; field of view (FOV), 230 mm; matrix, 128 × 128; section thickness, 4.5 mm; number of sections, 7. After DCE-MRI scan, post-contrast axial, coronal and sagittal T1weighted images were obtained.
2.3. Imaging processing DCE-MR images were processed using a dedicated postprocessing software program (Omni-Kinetics; GE Healthcare) which supplies pharmacokinetic calculation on a pixel-by-pixel basis. The current tracer-kinetic modeling for quantitation of DCE images was based on a two-compartment modified Tofts model [18]. In terms of the arterial input function (AIF), it was extracted by manually drawing a small circle region of interest (ROI) on one side of carotid artery located proximal to the tumor [19]. Whole-tumor ROIs were manually drawn over DCE-MR images, and then voxel-wise perfusion maps, containing both model-free (semi-quantitative) and model-based (quantitative) parameters were automatically generated. Semi-quantitative parameters included AUC (Area under the gadolinium dynamic curve in mmol*min), TTP (Time from contrast arrival to peak in min) and Slopemax (Maximum concentration-time ratio in min−1 ). Quantitative parameters included Ktrans (the volume transfer constant between the plasma and the extracellular extravascular space [EES] in min−1 ), ve (the volume fraction of the EES in ml/ml), and kep (the rate constant from EES to blood plasma in min−1 , which equals the ratio Ktrans /ve ) [12,13]. In terms of the ROIs placement, they were outlined on all slices by encompassing as much as tumor area, while the visual necrotic, hemorrhagic areas and surrounding blood vessels were excluded with reference to the conventional MR images. To minimize the effect of partial volume, the edges of lesions were avoided. If bilateral lesions occurred, the lesion with the larger diameter was included for analysis. All the quantitative measurements were performed independently by two neuro-radiologists (reader 1: with 6 years of experience; reader 2: with 4 years of experience) who were blinded to the clinical information, pathological results and study design. The measurement results of these two readers were used to evaluate the inter-observer reproducibility. Meanwhile, to evaluate the intra-observer reproducibility, all the DCE-MR images were assessed again by the reader 1, spaced at least one month. The average of the two measurement results of reader 1 was adopted into statistical analysis.
90
H. Hu et al. / European Journal of Radiology 88 (2017) 88–94
2.4. Statistical analysis Numeric data were averaged over all patients and reported as the mean ± standard deviation. The Kolmogorov-Smirnov test was used for normally distributed analysis. If normally distributed, the difference would be tested using t-test. Otherwise, it would be tested using Mann-Whitney U test. The significance threshold for difference on DCE-MRI parameters was set at a P value less
than 0.010 (0.050/5) for Bonferroni multiple comparison correction. Receiver operating characteristic (ROC) curve analyses were performed to determine the value of each significant parameter in predicting malignant OPLDs. Sensitivity and specificity were calculated with a threshold criterion determined as the value would maximize the Youden index. The inter- and intra-observer reproducibility of DCE-MRI parameters measurement were evaluated using intraclass correlation coefficient (ICC) with 95% confidence
Fig. 1. Box-and-whisker plots show the comparisons of DCE-MRI derived parameters between benign and malignant OLPDs groups.
H. Hu et al. / European Journal of Radiology 88 (2017) 88–94 Table 2 DCE-MRI derived parameters of benign and malignant OLPDs. Parameters
Malignant OLPDs (n = 21)
Benign OLPDs (n = 18)
P value
Ktrans (min−1 ) kep (min−1 ) ve (ml/ml) AUC (mmol*min) TTP (min) Slopemax (min−1 )
0.212 ± 0.787 1.098 ± 0.445 0.210 ± 0.050 6.737 ± 1.781 2.325 ± 0.628 7.456 ± 3.157
0.228 ± 0.086 0.751 ± 0.266 0.328 ± 0.093 8.367 ± 1.569 3.022 ± 1.086 7.955 ± 3.262
0.587 <0.001 <0.001 0.005 0.020 0.549
Note: The numeric data were reported as the mean ± standard deviation. Data in parentheses indicates the unit of the corresponding parameters. For definitions of the indicated parameters please refer to the Image Processing section of the Materials and Methods.
Fig. 2. ROC curves of using DCE-MRI derived parameters to predict malignant OLPDs.
intervals and applying a two-way ICC with random rater assumption. The ICC value ranges between 0 and 1.00, and values closer to 1.00 represented better reproducibility. The ICC was interpreted as follows: (< 0.40, poor; 0.40–0.60, moderate; 0.61–0.80, good; ≥ 0.81, excellent). All statistical analyses were performed using the SPSS software package (version 19.0, SPSS, Chicago, IL). 3. Results There was significant difference on the age (P = 0.010), while no difference on the gender distribution (P = 0.742) of patients between benign and malignant OLPDs groups. Table 2 summarizes the detailed comparisons of DCE-MRI derived parameters between benign and malignant OLPDs groups. Malignant OPLDs had significantly higher kep (P < 0.001), lower ve (P < 0.001), and lower AUC (P = 0.005) than benign mimics. However, there were no significant differences on Ktrans (P = 0.587), TTP (P = 0.020) and Slopemax (P = 0.549) between two groups. The comparisons of DCEMRI parameters between two groups are plotted in Fig. 1. ROC curve analyses indicated that, ve exhibited the best diagnostic performance in predicting malignant OPLDs (cutoff value, 0.211; area under the curve, 0.896; sensitivity, 76.2%; specificity, 94.9%), followed by kep (cutoff value, 0.853; area under the curve, 0.839; sensitivity, 85.7%; specificity, 89.9%) and AUC (cutoff value, 7.457; area under the curve, 0.783; sensitivity, 76.2%; specificity, 72.2%). The ROC curves regarding using DCE-MRI derived parameters to discriminate malignant from benign OLPDs are shown in Fig. 2. kep showed the highest diagnostic sensitivity (85.7%) and ve demonstrated the highest diagnostic specificity (94.9%). Repre-
91
sentative cases of orbital lymphoma and idiopathic inflammatory pseudotumor are shown in Figs. 3 and 4. Excellent intra- and inter-observer reproducibility were obtained for both the semi-quantitative and quantitative measurements of DCE-MRI derived parameters (intra-observer ICCs: Ktrans , 0.983; kep , 0.989; ve , 0.985; AUC, 0.976; TTP, 0.984; Slopemax , 0.980, and inter-observer ICCs: Ktrans , 0.969; kep , 0.978; ve , 0.973; AUC, 0.940; TTP, 0.971; Slopemax , 0.954).
4. Discussion In consideration of the entirely different therapeutic strategies for benign and malignant OLPDs, accurate pre-treatment differentiation of these two entities is very crucial for the individualized treatment. Our present study results indicated that DCE-MRI derived parameters, especially kep and ve , could help to distinguish benign from malignant OLPDs. kep demonstrated the highest diagnostic sensitivity, and ve showed the highest diagnostic specificity. To the best of our acknowledgement, our study is the first one that applying model-based analysis of DCE-MRI to differentiate benign from malignant OLPDs till now. Present study founded that malignant OLPDs demonstrated significantly lower AUC value than the benign mimics. The AUC value was viewed as a metric about the integration of the concentration of contrast agent observed in the tissue of interest over time [20]. Previous study by Haradome K. et al. has reported that the benign OLPDs showed higher contrast-enhancement ratio [6]. Meanwhile, Yuan Y et al. reported that the TIC pattern of benign OLPDs mostly showed as Type I or II (persistent or plateau pattern), while malignant ones mostly showed as Type III (washout pattern) [15]. In comprehensive consideration of the two reasons above, naturally the benign OLPDs would show the higher AUC than the malignant ones. However, AUC did not exhibit the attractive diagnostic ability (area under the curve, 0.783; sensitivity, 76.2%; specificity, 72.2%) for predicting orbital lymphoma in our study. It might be associated with the fact that the AUC value represents a conglomerate of physiologic processes, including blood flow, vascular permeability and fractional interstitial space [14]. Model-free analysis did not provide accurate histological insight of the tumor tissue, and therefore model-based metrics providing more histological information would be needed. Similar with previous studies those focused on lung cancer [9] and glioblastoma [21], malignant OLPDs in our study also exhibited significantly lower ve than the benign mimics. Previously numerous studies have noted that lymphoma was a hyper-cellularity tumor that constituted by numerous uniformly small-sized atypical lymphocytes [6,8,22,23]. The higher cellularity of lymphoma would result in more limited extracellular space. By contrast, the most important histological characteristic of benign OLPDs was interstitial edema, which would lead to the increased extracellular space [6]. Then, it was not surprising that lymphoma would exhibit lower ve than benign OLPDs, after considering that ve was a quantitative metric associated with the volume fraction of the extracellular extravascular space [24]. In our study, malignant OLPDs showed higher kep than benign mimics, which was consistent with one previous study of Ro SR et al. [25]. They compared the DCE-MRI derived parameters between overall orbital benign and malignant tumors, no just focusing on the lymphoproliferative disorders, and they explained that the higher kep might reflect the higher vascular permeability within the malignant tumors [25]. Meanwhile, they found orbital malignant tumors exhibited higher Ktrans than benign ones, however Ktrans did not differ significantly between malignant and benign OLPDs in our study. As to the possible interpretation, the mutability of Ktrans might be potential reason. Paldino MJ et al. [26]
92
H. Hu et al. / European Journal of Radiology 88 (2017) 88–94
Fig. 3. Representative images of a 79-year-old man with left orbital lymphoma. Pre-treatment DCE-MR imaging was conducted and ROIs were outlined on all slices (a). b–d show kep , ve , and AUC map, respectively. Obtained TIC shows as a washout pattern (e).
has ever clarified that, Ktrans could be influenced by several factors including blood flow to tissue, microvascular attenuation, vascular permeability, and fractional volume of EES (ve ). Knowing that lymphoma was a relatively poorer vascularized tumor
compared with the benign mimics [8], the local microvascular attenuation and blood flow might be decreased within lymphoma tissue, resulting in attenuation of Ktrans value. What’s more, given the proportional relationship between ve and Ktrans under a settled
H. Hu et al. / European Journal of Radiology 88 (2017) 88–94
93
Fig. 4. Representative images of a 45-year-old woman with right orbital idiopathic inflammatory pseudotumor. Pre-treatment DCE-MR imaging was exhibited and ROIs were outlined on all slices (a). b–d demonstrate kep , ve , and AUC map, respectively. Obtained TIC shows as a plateau pattern (e).
kep [12,13], the significantly decreased ve of lymphoma should also be responsible for the variation of Ktrans . Thus, we speculated that, although the vascular permeability increased within lymphoma tissue, which could assist the elevation of Ktrans value, the met-
ric might be more efficiently affected by other factors, such as the decreased microvascular attenuation, the lack of local blood flow and the decreased fractional volume of EES. Hence, kep might be
94
H. Hu et al. / European Journal of Radiology 88 (2017) 88–94
determined to a larger extent by vascular permeability than Ktrans , as in a previous study Zwick S et al. has clarified [27]. In terms of the measurement reproducibility, our study showed excellent inter- and intra-observer agreements for all quantitative measurements, with all ICCs more than 0.900, which seemed to be better than previous DCE-MRI-related studies [9,14,15]. The higher ICCs obtained in our study might be associated with the following reasons. First, our study subjects were OLPDs which were characterized by the homogeneous imaging appearance [1,6]. Tumor necrosis or hemorrhage were seldom found in OLPDs, and this characteristic would do beneficial to the placement of ROI. Second, the whole-tumor ROI approach was used for quantitative measurements, which was different from the selected ROI placement in previous studies [9,14,15,25]. As previous study indicated, wholetumor ROI approach could effectively reduce the inter-observer variability in ROI placement [28]. In addition to the limitations intrinsic to any retrospective study, several other limitations should be mentioned. First, the number of study population was relatively small. Further study with larger sample size would be needed to verify our results. Second, the imaging features of conventional MR images and DWI has also been proven to be useful in differentiating benign from malignant OLPDs. Conventional MR imaging focused on the tumor morphology, DWI focused on the tumorous cellular density, and DCE-MRI focused on the tumor angiogenesis. Further study combined using multi-modal MRI techniques would be more valuable in the differentiation of benign and malignant OLPDs. 5. Conclusion In conclusion, we present out initial clinical experience that using both semi-quantitative and quantitative analysis of DCE-MRI in the field of OLPDs. Our preliminary results suggest that quantitative DCE-MRI, especially kep and ve value, can help to differentiate benign from malignant OLPDs. Conflict of interest There was no any conflict of interest in this article. Funding information None. References [1] G. Akansel, L. Hendrix, B.A. Erickson, A. Demirci, A. Papke, A. Arslan, et al., MRI patterns in orbital malignant lymphoma and atypical lymphocytic infiltrates, Eur. J. Radiol. 53 (2) (2005) 175–181. [2] G.M. Espinoza, Orbital inflammatory pseudotumors: etiology, differential diagnosis, and management, Curr. Rheumatol. Rep. 12 (6) (2010) 443–447. [3] M. Takahira, Y. Ozawa, M. Kawano, Y. Zen, S. Hamaoka, K. Yamada, et al., Clinical aspects of IgG4-related orbital inflammation in a case series of ocular adnexal lymphoproliferative disorders, Int. J. Rheumatol. 2012 (2012) 635473. [4] A. Fujita, O. Sakai, M.N. Chapman, H. Sugimoto, IgG4-related disease of the head and neck: CT and MR imaging manifestations, Radiographics 32 (7) (2012) 1945–1958. [5] G.E. Valvassori, S.S. Sabnis, R.F. Mafee, M.S. Brown, A. Putterman, Imaging of orbital lymphoproliferative disorders, Radiol. Clin. North Am. 37 (1) (1999) 135–150. [6] K. Haradome, H. Haradome, Y. Usui, S. Ueda, T.C. Kwee, K. Saito, et al., Orbital lymphoproliferative disorders (OLPDs): value of MR imaging for differentiating orbital lymphoma from benign OPLDs, AJNR Am. J. Neuroradiol. 35 (10) (2014) 1976–1982.
[7] A.L. Weber, L. Vitale Romo, N.R. Sabates, Pseudotumor of the orbit: clinical, pathologic, and radiologic evaluation, Radiol. Clin. North Am. 37 (1) (1999) 151–168. [8] R. Kapur, A.R. Sepahdari, M.F. Mafee, A.M. Putterman, V. Aakalu, L.J. Wendel, et al., MR imaging of orbital inflammatory syndrome, orbital cellulitis, and orbital lymphoid lesions: the role of diffusion-weighted imaging, AJNR Am. J. Neuroradiol. 30 (1) (2009) 64–70. [9] M. Yuan, Y.D. Zhang, C. Zhu, T.F. Yu, H.B. Shi, Z.F. Shi, et al., Comparison of intravoxel incoherent motion diffusion-weighted MR imaging with dynamic contrast-enhanced MRI for differentiating lung cancer from benign solitary pulmonary lesions, J. Magn. Reson. Imaging 43 (3) (2016) 669–679. [10] M.A. Bali, T. Metens, V. Denolin, M. Delhaye, P. Demetter, J. Closset, Tumoral and nontumoral pancreas: correlation between quantitative dynamic contrast-enhanced MR imaging and histopathologic parameters, Radiology 261 (2) (2011) 456–466. [11] C. Chakiba, F. Cornelis, E. Descat, M. Gross-Goupil, P. Sargos, G. Roubaud, et al., Dynamic contrast enhanced MRI-derived parameters are potential biomarkers of therapeutic response in bladder carcinoma, Eur. J. Radiol. 84 (6) (2015) 1023–1028. [12] S. Gaddikeri, R.S. Gaddikeri, T. Tailor, Y. Anzai, Dynamic contrast-enhanced MR imaging in head and neck cancer: techniques and clinical applications, AJNR Am. J. Neuroradiol. 37 (4) (2015) 588–595. [13] T. Chikui, M. Obara, A.W. Simonetti, M. Ohga, S. Koga, S. Kawano, 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. [14] J. Xian, Z. Zhang, Z. Wang, J. Li, B. Yang, F. Man, et al., Value of MR imaging in the differentiation of benign and malignant orbital tumors in adults, Eur. Radiol. 20 (7) (2010) 1692–1702. [15] Y. Yuan, X.P. Kuai, X.S. Chen, X.F. Tao, Assessment of dynamic contrast-enhanced magnetic resonance imaging in the differentiation of malignant from benign orbital masses, Eur. J. Radiol. 82 (9) (2013) 1506–1511. [16] M.J. Koh, H.S. Kim, C.G. Choi, S.J. Kim, Which is the best advanced MR imaging protocol for predicting recurrent metastatic brain tumor following gamma-knife radiosurgery: focused on perfusion method, Neuroradiology 57 (4) (2015) 367–376. [17] I. Ocak, M. Bernardo, G. Metzger, T. Barrett, P. Pinto, P.S. Albert, et al., Dynamic contrast-enhanced MRI of prostate cancer at 3 T: a study of pharmacokinetic parameters, Am. J. Roentgenol. 189 (4) (2007) 192–201. [18] P.S. Tofts, G. Brix, D.L. Buckley, J.L. Evelhoch, E. Henderson, M.V. Knopp, et al., Estimating kinetic parameters from dynamic contrast-enhanced T1-weighted MRI of a diffusable tracer: standardized quantities and symbols, J. Magn. Reson. Imaging 10 (3) (1999) 223–232. [19] Y.R. Kim, K.J. Rebro, K.M. Schmainda, Water exchange and inflow affect the accuracy of T1-GRE blood volume measurements: implications for the evaluation of tumor angiogenesis, Magn. Reson. Med. 47 (6) (2002) 1110–1120. [20] T.E. Yankeelov, J.C. Gore, Dynamic contrast enhanced magnetic resonance imaging in oncology: theory, data acquisition, analysis, and examples, Curr. Med. Imaging Rev. 3 (2) (2009) 91–107. [21] K. Özduman, E. Yıldız, A. Dinc¸er, A. Sav, M.N. Pamir, Using intraoperative dynamic contrast-enhanced T1-weighted MRI to identify residual tumor in glioblastoma surgery, J. Neurosurg. 120 (1) (2014) 60–66. [22] X. Wu, H. Pertovaara, P. Dastidar, M. Vornanen, L. Paavolainen, V. Marjomäki, et al., ADC measurements in diffuse large B-cell lymphoma and follicular lymphoma: a DWI and cellularity study, Eur. J. Radiol. 82 (4) (2013) e158–64. [23] X.Q. Xu, H. Hu, G.Y. Su, H. Liu, X.N. Hong, H.B. Shi, et al., Utility of histogram analysis of ADC maps for differentiating orbital tumors, Diagn. Interv. Radiol. 22 (2) (2016) 161–167. [24] M. Park, J. Kim, Y.S. Choi, S.K. Lee, Y.W. Koh, S.H. Kim, et al., Application of dynamic contrast-enhanced MRI parameters for differentiating squamous cell carcinoma and malignant lymphoma of the oropharynx, AJR Am. J. Roentgenol. 206 (2) (2016) 401–407. [25] S.R. Ro, P. Asbach, E. Siebert, E. Bertelmann, B. Hamm, K. Erb-Eigner, Characterization of orbital masses by multiparametric MRI, Eur. J. Radiol. 85 (2) (2016) 324–336. [26] M.J. Paldino, D.P. Barboriak, Fundamentals of quantitative dynamic contrast-enhanced MR imaging. Fundamentals of quantitative dynamic contrast-enhanced MR imaging, Magn. Reson. Imaging Clin. N. Am. 17 (2) (2009) 277–289. [27] S. Zwick, G. Brix, P.S. Tofts, R. Strecker, A. Kopp-Schneider, H. Laue, et al., Simulation-based comparison of two approaches frequently used for dynamic contrast-enhanced MRI, Eur. Radiol. 20 (2) (2010) 432–442. [28] C.J. Wu, Q. Wang, H. Li, X.N. Wang, X.S. Liu, H.B. Shi, et al., DWI-associated entire-tumor histogram analysis for the differentiation of low-grade prostate cancer from intermediate-high-grade prostate cancer, Abdom. Imaging 40 (8) (2015) 3214–3221.