Magnetic Resonance Imaging xxx (2014) xxx–xxx
Contents lists available at ScienceDirect
Magnetic Resonance Imaging journal homepage: www.mrijournal.com
Original contribution
Feasibility of diffusional kurtosis tensor imaging in prostate MRI for the assessment of prostate cancer: Preliminary results Michael Quentin a, 1, Gael Pentang a, 1, Lars Schimmöller a,⁎, Olga Kott a, 1, Anja Müller-Lutz a, 1, Dirk Blondin a, 1, Christian Arsov b, 2, Andreas Hiester b, 2, Robert Rabenalt b, 2, Hans-Jörg Wittsack a, 1 a b
University Dusseldorf, Medical Faculty, Department of Diagnostic and Interventional Radiology, D-40225 Dusseldorf, Germany University Dusseldorf, Medical Faculty, Department of Urology, D-40225 Dusseldorf, Germany
a r t i c l e
i n f o
Article history: Received 12 January 2014 Revised 2 April 2014 Accepted 12 April 2014 Available online xxxx Keywords: Prostate Cancer MRI Diffusion-weighted imaging Kurtosis
a b s t r a c t Purpose: To assess the feasibility of full diffusional kurtosis tensor imaging (DKI) in prostate MRI in clinical routine. Histopathological correlation was achieved by targeted biopsy. Materials and Methods: Thirty-one men were prospectively included in the study. Twenty-one were referred to our hospital with increased prostate specific antigen (PSA) values (N4 ng/ml) and suspicion of prostate cancer. The other 10 men were volunteers without any history of prostate disease. DKI applying diffusion gradients in 20 different spatial directions with four b-values (0, 300, 600, 1000 s/mm 2) was performed additionally to standard functional prostate MRI. Region of interest (ROI)-based measurements were performed in all histopathologically verified lesions of every patient, as well as in the peripheral zone, and the central gland of each volunteer. Results: DKI showed a substantially better fit to the diffusion-weighted signal than the monoexponential apparent diffusion coefficient (ADC). Altogether, 29 lesions were biopsied in 14 different patients with the following results: Gleason score 3 + 3 = 6 (n = 1), 3 + 4 = 7 (n = 7), 4 + 3 = 7 (n = 6), 4 + 4 = 8 (n = 1), and 4 + 5 = 9 (n = 2), and prostatitis (n = 12). Values of axial (Kax) and mean kurtosis (Kmean) were significantly different in the tumor (Kax 1.78 ± 0.39, Kmean 1.84 ± 0.43) compared with the normal peripheral zone (Kax 1.09 ± 0.12, Kmean 1.16 ± 0.13; p b 0.001) or the central gland (Kax 1.40 ± 0.12, Kmean 1.44 ± 0.17; p = 0.01 respectively). There was a minor correlation between axial kurtosis (r = 0.19) and the Gleason score. Conclusion: Full DKI is feasible to utilize in a routine clinical setting. Although there is some overlap some DKI parameters can significantly distinguish prostate cancer from the central gland or the normal peripheral zone. Nevertheless, the additional value of DKI compared with conventional monoexponential ADC calculation remains questionable and requires further research. © 2014 Elsevier Inc. All rights reserved.
1. Introduction
⁎ Corresponding author at: Department of Diagnostic and Interventional Radiology, University Dusseldorf, Medical Faculty, Moorenstr. 5, D-40225 Dusseldorf, Germany. Tel.: +49 211 81 17752; fax: +49 211 81 16145. E-mail addresses: [email protected] (M. Quentin), [email protected] (G. Pentang), [email protected] (L. Schimmöller), [email protected] (O. Kott), [email protected] (A. Müller-Lutz), [email protected] (D. Blondin), [email protected] (C. Arsov), [email protected] (A. Hiester), [email protected] (R. Rabenalt), [email protected] (H-J. Wittsack). 1 Tel.: +49 211 81 17752; fax: +49 211 81 16145. 2 Tel.: +49 211 81 18110; fax: +49 211 81 18676.
Functional prostate MRI has gained importance in the diagnosis of prostate cancer (PCa) over the last years since the primary tools for tumor detection like digital rectal examination, prostate specific antigen (PSA), transrectal ultrasonography (TRUS), and systematic transrectal biopsy have clear limitations. Although the widespread use of PSA screening has led to an increased incidence of PCa diagnosis and a shift towards earlier stages of detected cancers, the general benefit of PSA screening is still controversial due to considerable overdiagnosis and overtreatment [1]. Systematic TRUS-guided biopsy, on the other hand, is prone to undergrading and thus undertreatment [2]. Additionally, systematic TRUS-guided biopsy has a considerably low negative predictive value [3]. Therefore, there is a clinical need for improved tests for PCa diagnosis.
http://dx.doi.org/10.1016/j.mri.2014.04.005 0730-725X/© 2014 Elsevier Inc. All rights reserved.
Please cite this article as: Quentin M, et al, Feasibility of diffusional kurtosis tensor imaging in prostate MRI for the assessment of prostate cancer: Preliminary results, Magn Reson Imaging (2014), http://dx.doi.org/10.1016/j.mri.2014.04.005
2
M. Quentin et al. / Magnetic Resonance Imaging xxx (2014) xxx–xxx
Functional MRI, including diffusion-weighted imaging (DWI), dynamic contrast-enhanced MRI (DCE-MRI), and spectroscopic imaging (MRSI) has especially improved the role of MRI in detection, localization, and staging of PCa [4]. In the recently released prostate MRI guidelines of the European Society of Urogenital Radiology (ESUR), DWI is an integral part of functional prostate MRI [5]. DWI depends on the microscopic mobility of water molecules, known as “Brownian motion”, which is restricted in tissue compared with pure water. Several studies showed a significant association between the apparent diffusion coefficient (ADC), which quantifies the signal intensity loss in DWI, and the tumor aggressiveness expressed by the Gleason score [6–8]. Different models to evaluate DWI have been developed, including a simple monoexponential model, a biexponential model with two independent fractions, a statistical model with a distribution fraction and, recently, kurtosis imaging [9]. Kurtosis imaging has shown a better fit to diffusion-weighted data compared with the standard monoexponential ADC [10]. Although DWI is routinely applied in prostate MRI, only four studies dealing with kurtosis imaging in prostate MRI are available in the literature [11–14]. All four studies measured diffusion with an extended range of b-values in only three independent gradient directions. However, according to Jensen et al. [9,15], measuring the full diffusional kurtosis tensor requires at least 15 independent gradient directions and three different b-values. Therefore, this is the first study to assess the feasibility of full diffusional kurtosis tensor imaging in prostate MRI. Histopathological correlation was achieved by targeted biopsy.
T2-weighted axial images were acquired using a turbo spin echo sequence (TR 10630 ms, TE 117 ms, field of view [FOV] 12.8 cm, voxel size 0.5 × 0.5 × 3.0 mm, image matrix 256 × 256, turbo factor 23, number of slices = 30). DKI was performed with a single-shot spin-echo echo-planar sequence (TR 1700 ms, TE 101 ms, FOV 20.4 cm, voxel size 1.5 × 1.5 × 6.0 mm, image matrix 136 × 136, GRAPPA parallel imaging scheme, acceleration factor 2, and acquisition time 7:00 min, number of slices = 10). To acquire DKI, four b-values (0, 300, 600, 1000 s/mm 2) with four averages, applying independent diffusion gradients in 20 directions for each b-value, were used. DCEMRI images, T2-weighted sagittal images, T2-weighted coronal images, and T1-weighted axial images were also acquired, but not assessed as part of this study. 2.3. Targeted biopsy Targeted biopsy of the described lesions was performed either with MR-guided in-bore biopsy or with stereotactic ultrasoundguided MRI fusion biopsy. MR-guided biopsies were performed on a 3-T MR scanner (Magnetom Trio; Siemens Medical Systems, Erlangen, Germany) using the dynatrim biopsy device (Invivo, Orlando, FL, USA) and the corresponding software DynaCAD (Invivo, Orlando, FL, USA). Stereotactic ultrasound-guided biopsies used realtime transrectal ultrasound guidance with MR/TRUS image fusion (Urostation®, Koelis, La Tronche, France). 2.4. Post-processing and measurements
2. Materials and methods 2.1. Subjects Subjects of this feasibility study were previously enrolled in ongoing, prospective trials assessing MRI-guided in-bore prostate biopsy and stereotactic ultrasound-guided MRI fusion biopsy at our institution. The detailed design of this prospective trial has been reported elsewhere [16]. The study was approved by the local ethics committee. Informed consent was obtained from all subjects. Thirty-one men were prospectively included in the study, who underwent functional prostate MRI between January 2012 and May 2013. Twenty-one of these men (mean age 71 years, range 55 to 77 years) were referred to our hospital with increased PSA values (N4 ng/ml). After performing T2-weighted axial images, a radiologist with 4 years’ experience in prostate MRI decided if patients had suspicious lesions. Lesions should be large enough to distinctively place region-of-interest (ROI)-based measurements without partial volume effects in the lesions. In case of suspicious lesions in T2weighted imaging, DKI measurement was performed in addition to the standard prostate MRI protocol. Following the local procedure, a maximum of three lesions was defined per patient in the written report, including suspicious and benign findings. Patients with suspicious lesions underwent targeted biopsy of all described lesions. The other 10 included men were volunteers (mean age 26 years, range 24 to 32 years) without any known prostate disease. 2.2. MR techniques All prostate MRI examinations were performed using a 3-T MRI scanner (Magnetom Trio, A TIM System; Siemens Medical Systems, Erlangen, Germany) with a six-channel phased-array body-coil and a 24-channel spine array coil. Before the examination, patients received 20 mg of butylscopolamine (Buscopan®, Boehringer Ingelheim Pharma, Ingelheim, Germany) intravenously and intramuscularly. The 10 volunteers did not receive any bowel relaxant.
DKI datasets were post-processed using the software tool Diffusional Kurtosis Estimator (DKE, Version 2.5.1, built on 16 Dec 17 2012) on an external workstation [17]. Kurtosis (axial, mean, radial) parametric maps and fractional anisotropy (FA) maps were calculated with the DKE. ROI-based measurements were performed in these parameter maps using the software MRIcro (www.mricro.com, Version 1.40, Chris Rorden, University of South Carolina, USA) on an external workstation. ROIs were defined as large as possible by a radiologist with 4 years’ experience in prostate MRI, in the areas histopathologically confirmed by targeted biopsy, using axial T2weighted images as reference. ROIs were copied in the different parametric maps, derived from diffusion-weighted imaging, to ensure that always the same areas were evaluated. Measurements in the peripheral zone and central gland were repeated three times in different areas to calculate mean values. ADC values were obtained from the MRI scanner software using the standard monoexponential model using ROI-based measurements in the picture archiving and communication system (Sectra Workstation IDS7, Sectra AB, Linköping, Sweden). 2.5. Statistics All data are expressed as mean ± standard deviation. Parameters were compared using the independent sample t-test. Statistical significance was defined as a p b 0.05. R 2 was calculated to assess the goodness of fit for the DKI and the monoexponential model to the diffusional data in all volunteers. Correlation coefficients were calculated for the different model parameters and the Gleason score. 3. Results Of the 21 included patients, 10 patients were biopsied using MRguided in-bore biopsy and another four patients received a stereotactic ultrasound-guided MRI fusion biopsy. Histopathological results were positive for PCa in nine patients and 17 different lesions. Gleason score distribution was as follows: Gleason score 3 + 3 = 6
Please cite this article as: Quentin M, et al, Feasibility of diffusional kurtosis tensor imaging in prostate MRI for the assessment of prostate cancer: Preliminary results, Magn Reson Imaging (2014), http://dx.doi.org/10.1016/j.mri.2014.04.005
M. Quentin et al. / Magnetic Resonance Imaging xxx (2014) xxx–xxx
3
Fig. 1. Normalized signal intensity in a representative tumor and the peripheral zone (PZ) of the same patient derived from diffusional data for the different b-values. The kurtosis formula shows considerably better goodness of fit compared with the monoexponential model, both visually and mathematically (expressed by higher R2).
(n = 1), 3 + 4 = 7 (n = 7), 4 + 3 = 7 (n = 6), 4 + 4 = 8 (n = 1), and 4 + 4 = 9 (n = 2), respectively. Biopsy revealed prostatitis in five patients in 12 different lesions. Seven patients did not receive a targeted biopsy and had to be excluded. DKI provided a substantially better fit to the diffusional data compared with the monoexponential model (Fig. 1). The goodness of fit of diffusional data in all volunteers was R 2 = 0.9967 in the peripheral zone and R 2 = 0.9947 in the central gland for DKI compared with R 2 = 0.9486 in the peripheral zone and R 2 = 0.9417 in the central gland for the monoexponential model. Image quality was sufficient in all examined patients and volunteers (Fig. 2). The different mean values and standard deviations for DKI and the monoexponential model are shown in Table 1. Values for axial
kurtosis, mean kurtosis, and radial kurtosis were not significantly different, each compared in the peripheral zone, the central gland, prostatitis, and PCa. For the different diffusion parameters, only the monoexponential ADC could significantly (p = 0.03, Table 2) distinguish between the peripheral zone, the central gland, prostatitis, and PCa (Fig. 3). Significant differentiation (p = 0.03) between PCa and prostatitis was possible with axial kurtosis, mean kurtosis, and ADC. Axial and mean kurtosis only failed to significantly differentiate the central gland from prostatitis (Fig. 4). The parameters with the worst performance were FA (Fig. 5) and radial kurtosis, which could only differentiate PCa or prostatitis from the normal peripheral zone. There was a weak correlation between the Gleason score and all kurtosis parameters (axial kurtosis r = 0.19, mean kurtosis r =
Fig. 2. Image of a dorsal right tumor in the peripheral zone in T2-weighted imaging (T2), apparent diffusion coefficient map (ADC), fractional anisotropy (FA), mean kurtosis (MK), axial kurtosis (AK), and radial kurtosis (RK). Kurtosis parameter maps are color-coded. The tumor is sufficiently visible in all parameters used.
Please cite this article as: Quentin M, et al, Feasibility of diffusional kurtosis tensor imaging in prostate MRI for the assessment of prostate cancer: Preliminary results, Magn Reson Imaging (2014), http://dx.doi.org/10.1016/j.mri.2014.04.005
4
M. Quentin et al. / Magnetic Resonance Imaging xxx (2014) xxx–xxx
4.2. Parameters derived from full diffusional kurtosis tensor imaging
Table 1 Mean values and standard deviation. FA PZ CZ PCa Pr
0.13 0.15 0.18 0.18
Kmean
Kax ± ± ± ±
0.03 0.02 0.05 0.05
1.09 1.40 1.78 1.41
± ± ± ±
0.12 0.12 0.39 0.19
1.16 1.44 1.84 1.50
± ± ± ±
Krad 0.13 0.17 0.43 0.17
1.18 1.48 1.81 1.54
ADC (10−3 mm2/s) ± ± ± ±
0.13 0.22 0.48 0.18
1.08 0.77 0.51 0.64
± ± ± ±
0.12 0.07 0.16 0.14
Values for the peripheral zone (PZ), central gland (CZ), prostate cancer (PCa), and prostatitis (Pr) using diffusional kurtosis tensor imaging for acquiring fractional anisotropy (FA), axial kurtosis (Kax), mean kurtosis (Kmean), radial kurtosis (Krad), and the monoexponential model for calculating the apparent diffusion coefficient (ADC, 10−3 mm2/s).
0.14, radial kurtosis r = 0.15, and FA r = − 0.01) and the ADC derived from the monoexponential model (r = − 0.23). 4. Discussion Our results show that mean kurtosis and axial kurtosis derived from full diffusional kurtosis imaging (DKI) can significantly differentiate PCa from prostatitis, the peripheral zone, or the central gland. These parameters only failed to significantly discriminate prostatitis from normal central gland. Beyond that, fractional anisotropy and radial kurtosis could not significantly differentiate PCa from the central gland or from prostatitis. 4.1. Diffusion models Using the out-of-the-box monoexponential model is a common and the easiest way to characterize the decay of signal intensity in DWI using the ADC. In recent years, more complex models were introduced. Since these models often use more free parameters they provide a better fit to the diffusion signal decay [13]. Some of these models make assumptions of the existence of different diffusion compartments, like the intra-voxel incoherent motion (IVIM) — model described by Le Bihan et al. [18]. DKI, on the other hand, does not make any assumption on the number or even existence of multiple compartments [19]. DKI is an extension of diffusion tensor imaging (DTI) [15]. With DTI, the degree of directionality can be obtained using FA and the overall movement of molecules using mean diffusivity (MD), and the ADC. DTI has been applied in prostate MRI in a few previous studies [20–24]. In DTI, a Gaussian distribution of diffusion is assumed, which can be hindered or restricted in complex biologic systems by the existence of various complex tissue structures [9]. DKI therefore describes the extent of deviation of the diffusion from the Gaussian form.
Table 2 P-values for the comparison of the different regions. P-values
PZ b −− N PCa
CZ b −− N PCa
PZ b −− N Pr
CZ b −− N Pr
PCa b −− N Pr
FA Kax Kmean Krad ADC
b0.01 b0.01 b0.01 b0.01 b0.01
0.10 0.01 0.01 0.05 b0.01
0.01 b0.01 b0.01 b0.01 b0.01
0.17 0.83 0.39 0.51 0.02
0.75 0.01 0.02 0.07 0.03
P-values derived from student’s t-test comparing values for the peripheral zone (PZ), central gland (CZ), prostate cancer (PCa), and prostatitis (Pr) calculated with diffusional kurtosis tensor imaging: fractional anisotropy (FA), axial kurtosis (Kax), mean kurtosis (Kmean), radial kurtosis (Krad), and the monoexponential model for calculating the apparent diffusion coefficient (ADC). Statistically significant differences (p b 0.05) are shaded in gray.
Although DWI as an integral part of functional prostate MRI has been investigated in numerous studies, there are only four studies available in the literature that deal with DKI in prostate MRI [11–14]. Measuring diffusion gradients in only three orthogonal directions, these studies did not assess the full diffusional kurtosis tensor. According to Jensen at al., who were the first to describe DKI, measurements in at least 15 independent gradient directions are necessary for calculation of the full diffusional kurtosis tensor [9,15]. Mean kurtosis, which is comparable to kurtosis in the three other studies, showed the same tendency, with higher values in PCa than in healthy tissue. In this study the different kurtosis parameters axial kurtosis, mean kurtosis, radial kurtosis showed comparable – not significantly different – values. A possible explanation might be the relatively isotropic diffusion in the prostate gland. In the literature, FA values gained only from DTI measurements are controversial. While some studies report higher values in PCa [20,21], others report lower [23] or equal [22,24] values compared with the peripheral zone. In this study, FA values were higher in PCa than in the peripheral zone. FA was able to discriminate PCa or prostatitis from the normal peripheral zone but was not able to distinguish between PCa and prostatitis, and failed in the central gland. Similar to several previous studies, this study demonstrated lower ADC values in PCa compared with the peripheral zone, and a negative correlation to the tumor grade [6,7]. The normal peripheral zone could be distinguished from PCa with every DKI parameter. Nevertheless, only the ADC was able to differentiate prostatitis from the central gland. This is in line with recent DWI studies reporting significant differences for the different tissues in the central gland (healthy tissue, PCa, different types of prostatitis) but also a considerably overlap using only DWI as functional imaging method [25,26].
4.3. Choice of b-values and signal-to-noise ratio The values of measured diffusional kurtosis tensor imaging parameters are different in all studies, most likely due to the different b-values used [13]. This study applied b-values up to 1000 s/mm 2, while the others used b-values up to 2000 [11,12], 2300 [14], and 800 [13] s/mm 2. Today, there is still limited consensus about the optimal b-value selection in DWI in functional prostate MRI [27]. The choice of b-value should not be too high because the quadratic term of the kurtosis formula will increase with b-values after a certain minimum [28]. According to Jensen et al., b-values should be somewhat larger than those usually applied in DWI, so that the departure from linearity is clearly apparent [9]. Fig. 1 shows that this departure from linearity could clearly be observed in our data when using b-values up to 1000 s/mm 2. Additionally there is limited consensus to include a minimum b-value of 0 s/mm 2. Maas et al. suggested to use a minimum b-value of 100 s/mm 2 to measure pure diffusion and to avoid perfusion effects in the lower b-value range [29]. On the other hand a minimum b-value of 0 s/mm 2 has been recommended for the monoexponential model in a European consensus meeting dealing with standardization of prostate MRI [27]. Closely connected to the choice of b-values is the signal-to-noise ratio with decreasing diffusion signal in higher b-values and constant background noise. According to Tabesh et al. [17] noise can artificially increase signal intensity and therefore yield falsely large kurtosis estimations. Finally an optimal choice of b-values for DKI is also tissue dependent, and therefore this may differ for prostate and other organs such as brain.
Please cite this article as: Quentin M, et al, Feasibility of diffusional kurtosis tensor imaging in prostate MRI for the assessment of prostate cancer: Preliminary results, Magn Reson Imaging (2014), http://dx.doi.org/10.1016/j.mri.2014.04.005
M. Quentin et al. / Magnetic Resonance Imaging xxx (2014) xxx–xxx
5
Fig. 3. Box plot of apparent diffusion coefficient (ADC; 10−3 mm2/s) values derived from the monoexponential model for the peripheral zone (PZ), the central gland (CZ), prostatitis, and prostate cancer.
4.4. Advantages of full diffusional kurtosis tensor imaging Altogether it is questionable whether the additional effort to measure and calculate DKI is justified. Rosenkrantz et al. suggested that DKI might be useful for predicting the final pathologic outcome for patients under active surveillance [12]. On the other hand, this study show that DKI measurements are feasible in clinical routine with little additional measuring time, since normal DWI can be skipped, and ADC calculation can be done with DKI measurements. 4.5. Limitations This study has some limitations. First of all, this study is limited by the rather small number of patients included. This is largely because patients were included before the biopsy and patients without a biopsy had to be excluded. We believe that a targeted biopsy performed either as in-bore MRI-guided biopsy or targeted ultrasound MRI fusion biopsy serves as a better reference than a random TRUS-guided biopsy. This limitation especially applies to the correlation to the Gleason score. Studies including a larger number
of patients should be performed to clarify the stability of the correlation between DKI and Gleason score. The second limitation is the restriction of the study population to patients with visible suspicious lesions in T2-weighted imaging. This had to be done to ensure the ROI-based measurements could be placed in the lesions without partial volume effects. Therefore, no sensitivity or detection rates were calculated in this study. The main aim was to test the feasibility of DKI in clinical routine for the first time. Another limitation is the – compared to the other imaging modalities – higher slice thickness of 6 mm for the DKI measurements, which was necessary to achieve adequate signal-to-noise ratio. 5. Conclusion In conclusion, although full DKI requires extended measuring and calculation time, it is feasible to utilize it in a routine clinical setting. DKI can significantly distinguish PCa from the normal peripheral zone or the central gland. The additional value of DKI compared with conventional monoexponential ADC calculation remains questionable and requires further research.
Fig. 4. Box plot of mean kurtosis (MK; dimensionless) values derived from diffusional kurtosis tensor imaging for the peripheral zone (PZ), the central gland (CZ), prostatitis, and prostate cancer.
Please cite this article as: Quentin M, et al, Feasibility of diffusional kurtosis tensor imaging in prostate MRI for the assessment of prostate cancer: Preliminary results, Magn Reson Imaging (2014), http://dx.doi.org/10.1016/j.mri.2014.04.005
6
M. Quentin et al. / Magnetic Resonance Imaging xxx (2014) xxx–xxx
Fig. 5. Box plot of fractional anisotropy (FA; dimensionless) values derived from diffusional kurtosis tensor imaging for the peripheral zone (PZ), the central gland (CZ), prostatitis, and prostate cancer.
References [1] Pashayan N, Duffy SW, Pharoah P, Greenberg D, Donovan J, Martin RM, et al. Mean sojourn time, overdiagnosis, and reduction in advanced stage prostate cancer due to screening with PSA: implications of sojourn time on screening. Br J Cancer 2009;100:1198–204. [2] Rajinikanth A, Manoharan M, Soloway CT, Civantos FJ, Soloway MS. Trends in gleason score: concordance between biopsy and prostatectomy over 15 years. Urology 2008;72:177–82. [3] Heijmink SW, van Moerkerk H, Kiemeney LA, Witjes JA, Frauscher F, Barentsz JO. A comparison of the diagnostic performance of systematic versus ultrasoundguided biopsies of prostate cancer. Eur Radiol 2006;16:927–38. [4] Mazaheri Y, Shukla-Dave A, Muellner A, Hricak H. MR imaging of the prostate in clinical practice. MAGMA 2008;21:379–92. [5] Barentsz JO, Richenberg J, Clements R, Choyke P, Verma S, Villeirs G, et al. ESUR prostate MR guidelines 2012. Eur Radiol 2012;22:746–57. [6] Vargas HA, Akin O, Franiel T, Mazaheri Y, Zheng J, Moskowitz C, et al. Diffusionweighted endorectal MR imaging at 3 T for prostate cancer: tumor detection and assessment of aggressiveness. Radiology 2011;259:775–84. [7] Tamada T, Sone T, Jo Y, Toshimitsu S, Yamashita T, Yamamoto A, et al. Apparent diffusion coefficient values in peripheral and transition zones of the prostate: comparison between normal and malignant prostatic tissues and correlation with histologic grade. J Magn Reson Imaging 2008;28:720–6. [8] Hambrock T, Somford DM, Huisman HJ, van Oort IM, Witjes JA, Hulsbergen-van de Kaa CA, et al. Relationship between apparent diffusion coefficients at 3.0-T MR imaging and Gleason grade in peripheral zone prostate cancer. Radiology 2011;259:453–61. [9] Jensen JH, Helpern JA, Ramani A, Lu H, Kaczynski K. Diffusional kurtosis imaging: the quantification of non-gaussian water diffusion by means of magnetic resonance imaging. Magn Reson Med 2005;53:1432–40. [10] Jansen JF, Stambuk HE, Koutcher JA, Shukla-Dave A. Non-gaussian analysis of diffusion-weighted MR imaging in head and neck squamous cell carcinoma: a feasibility study. AJNR Am J Neuroradiol 2010;31:741–8. [11] Rosenkrantz AB, Sigmund EE, Johnson G, Babb JS, Mussi TC, Melamed J, et al. Prostate cancer: feasibility and preliminary experience of a diffusional kurtosis model for detection and assessment of aggressiveness of peripheral zone cancer. Radiology 2012;264:126–35. [12] Rosenkrantz AB, Prabhu V, Sigmund EE, Babb JS, Deng FM, Taneja SS. Utility of diffusional kurtosis imaging as a marker of adverse pathologic outcomes among prostate cancer active surveillance candidates undergoing radical prostatectomy. AJR Am J Roentgenol 2013;201:840–6. [13] Quentin M, Blondin D, Klasen J, Lanzman RS, Miese FR, Arsov C, et al. Comparison of different mathematical models of diffusion-weighted prostate MR imaging. Magn Reson Imaging 2012;30:1468–74. [14] Mazzoni LN, Lucarini S, Chiti S, Busoni S, Gori C, Menchi I. Diffusion-weighted signal models in healthy and cancerous peripheral prostate tissues: comparison
[15] [16]
[17]
[18]
[19]
[20] [21] [22]
[23]
[24]
[25]
[26]
[27]
[28]
[29]
of outcomes obtained at different b-values. J Magn Reson Imaging 2013;39:512–8. Jensen JH, Helpern JA. MRI quantification of non-gaussian water diffusion by kurtosis analysis. NMR Biomed 2010;23:698–710. Arsov C, Blondin D, Rabenalt R, Antoch G, Albers P, Quentin M. Standardised scoring of a multi-parametric 3-T MRI for a targeted MRI-guided prostate biopsy. Urologe A 2012;51:848–56. Tabesh A, Jensen JH, Ardekani BA, Helpern JA. Estimation of tensors and tensorderived measures in diffusional kurtosis imaging. Magn Reson Med 2011;65:823–36. Le Bihan D, Breton E, Lallemand D, Aubin ML, Vignaud J, Laval-Jeantet M. Separation of diffusion and perfusion in intravoxel incoherent motion MR imaging. Radiology 1988;168:497–505. Lu H, Jensen JH, Ramani A, Helpern JA. Three-dimensional characterization of non-gaussian water diffusion in humans using diffusion kurtosis imaging. NMR Biomed 2006;19:236–47. Gibbs P, Pickles MD, Turnbull LW. Diffusion imaging of the prostate at 3.0 tesla. Invest Radiol 2006;41:185–8. Gürses B, Kabakci N, Kovanlikaya A, Firat Z, Bayram A, Uluğ AM, et al. Diffusion tensor imaging of the normal prostate at 3 tesla. Eur Radiol 2008;18:716–21. Kozlowski P, Chang SD, Meng R, Mädler B, Bell R, Jones EC, et al. Combined prostate diffusion tensor imaging and dynamic contrast enhanced MRI at 3 T— quantitative correlation with biopsy. Magn Reson Imaging 2010;28:621–8. Manenti G, Carlani M, Mancino S, Colangelo V, Di Roma M, Squillaci E, et al. Diffusion tensor magnetic resonance imaging of prostate cancer. Invest Radiol 2007;42:412–9. Xu J, Humphrey PA, Kibel AS, Snyder AZ, Narra VR, Ackerman JJ, et al. Magnetic resonance diffusion characteristics of histologically defined prostate cancer in humans. Magn Reson Med 2009;61:842–50. Hoeks CM, Vos EK, Bomers JG, Barentsz JO, Hulsbergen-van de Kaa CA, Scheenen TW. Diffusion-weighted magnetic resonance imaging in the prostate transition zone: histopathological validation using magnetic resonance-guided biopsy specimens. Invest Radiol 2013;48:693–701. Nagel KN, Schouten MG, Hambrock T, Litjens GJ, Hoeks CM, ten Haken B, et al. Differentiation of prostatitis and prostate cancer by using diffusion-weighted MR imaging and MR-guided biopsy at 3 T. Radiology 2013;267:164–72. Dickinson L, Ahmed HU, Allen C, Barentsz JO, Carey B, Futterer JJ, et al. Magnetic resonance imaging for the detection, localisation, and characterisation of prostate cancer: recommendations from a European consensus meeting. Eur Urol 2011;59:477–94. Hui ES, Cheung MM, Qi L, Wu EX. Towards better MR characterization of neural tissues using directional diffusion kurtosis analysis. Neuroimage 2008;42:122–34. Maas MC, Fütterer JJ, Scheenen TW. Quantitative evaluation of computed high b value diffusion-weighted magnetic resonance imaging of the prostate. Invest Radiol 2013;48:779–86.
Please cite this article as: Quentin M, et al, Feasibility of diffusional kurtosis tensor imaging in prostate MRI for the assessment of prostate cancer: Preliminary results, Magn Reson Imaging (2014), http://dx.doi.org/10.1016/j.mri.2014.04.005
Contents lists available at ScienceDirect
Magnetic Resonance Imaging journal homepage: www.mrijournal.com
Original contribution
Feasibility of diffusional kurtosis tensor imaging in prostate MRI for the assessment of prostate cancer: Preliminary results Michael Quentin a, 1, Gael Pentang a, 1, Lars Schimmöller a,⁎, Olga Kott a, 1, Anja Müller-Lutz a, 1, Dirk Blondin a, 1, Christian Arsov b, 2, Andreas Hiester b, 2, Robert Rabenalt b, 2, Hans-Jörg Wittsack a, 1 a b
University Dusseldorf, Medical Faculty, Department of Diagnostic and Interventional Radiology, D-40225 Dusseldorf, Germany University Dusseldorf, Medical Faculty, Department of Urology, D-40225 Dusseldorf, Germany
a r t i c l e
i n f o
Article history: Received 12 January 2014 Revised 2 April 2014 Accepted 12 April 2014 Available online xxxx Keywords: Prostate Cancer MRI Diffusion-weighted imaging Kurtosis
a b s t r a c t Purpose: To assess the feasibility of full diffusional kurtosis tensor imaging (DKI) in prostate MRI in clinical routine. Histopathological correlation was achieved by targeted biopsy. Materials and Methods: Thirty-one men were prospectively included in the study. Twenty-one were referred to our hospital with increased prostate specific antigen (PSA) values (N4 ng/ml) and suspicion of prostate cancer. The other 10 men were volunteers without any history of prostate disease. DKI applying diffusion gradients in 20 different spatial directions with four b-values (0, 300, 600, 1000 s/mm 2) was performed additionally to standard functional prostate MRI. Region of interest (ROI)-based measurements were performed in all histopathologically verified lesions of every patient, as well as in the peripheral zone, and the central gland of each volunteer. Results: DKI showed a substantially better fit to the diffusion-weighted signal than the monoexponential apparent diffusion coefficient (ADC). Altogether, 29 lesions were biopsied in 14 different patients with the following results: Gleason score 3 + 3 = 6 (n = 1), 3 + 4 = 7 (n = 7), 4 + 3 = 7 (n = 6), 4 + 4 = 8 (n = 1), and 4 + 5 = 9 (n = 2), and prostatitis (n = 12). Values of axial (Kax) and mean kurtosis (Kmean) were significantly different in the tumor (Kax 1.78 ± 0.39, Kmean 1.84 ± 0.43) compared with the normal peripheral zone (Kax 1.09 ± 0.12, Kmean 1.16 ± 0.13; p b 0.001) or the central gland (Kax 1.40 ± 0.12, Kmean 1.44 ± 0.17; p = 0.01 respectively). There was a minor correlation between axial kurtosis (r = 0.19) and the Gleason score. Conclusion: Full DKI is feasible to utilize in a routine clinical setting. Although there is some overlap some DKI parameters can significantly distinguish prostate cancer from the central gland or the normal peripheral zone. Nevertheless, the additional value of DKI compared with conventional monoexponential ADC calculation remains questionable and requires further research. © 2014 Elsevier Inc. All rights reserved.
1. Introduction
⁎ Corresponding author at: Department of Diagnostic and Interventional Radiology, University Dusseldorf, Medical Faculty, Moorenstr. 5, D-40225 Dusseldorf, Germany. Tel.: +49 211 81 17752; fax: +49 211 81 16145. E-mail addresses: [email protected] (M. Quentin), [email protected] (G. Pentang), [email protected] (L. Schimmöller), [email protected] (O. Kott), [email protected] (A. Müller-Lutz), [email protected] (D. Blondin), [email protected] (C. Arsov), [email protected] (A. Hiester), [email protected] (R. Rabenalt), [email protected] (H-J. Wittsack). 1 Tel.: +49 211 81 17752; fax: +49 211 81 16145. 2 Tel.: +49 211 81 18110; fax: +49 211 81 18676.
Functional prostate MRI has gained importance in the diagnosis of prostate cancer (PCa) over the last years since the primary tools for tumor detection like digital rectal examination, prostate specific antigen (PSA), transrectal ultrasonography (TRUS), and systematic transrectal biopsy have clear limitations. Although the widespread use of PSA screening has led to an increased incidence of PCa diagnosis and a shift towards earlier stages of detected cancers, the general benefit of PSA screening is still controversial due to considerable overdiagnosis and overtreatment [1]. Systematic TRUS-guided biopsy, on the other hand, is prone to undergrading and thus undertreatment [2]. Additionally, systematic TRUS-guided biopsy has a considerably low negative predictive value [3]. Therefore, there is a clinical need for improved tests for PCa diagnosis.
http://dx.doi.org/10.1016/j.mri.2014.04.005 0730-725X/© 2014 Elsevier Inc. All rights reserved.
Please cite this article as: Quentin M, et al, Feasibility of diffusional kurtosis tensor imaging in prostate MRI for the assessment of prostate cancer: Preliminary results, Magn Reson Imaging (2014), http://dx.doi.org/10.1016/j.mri.2014.04.005
2
M. Quentin et al. / Magnetic Resonance Imaging xxx (2014) xxx–xxx
Functional MRI, including diffusion-weighted imaging (DWI), dynamic contrast-enhanced MRI (DCE-MRI), and spectroscopic imaging (MRSI) has especially improved the role of MRI in detection, localization, and staging of PCa [4]. In the recently released prostate MRI guidelines of the European Society of Urogenital Radiology (ESUR), DWI is an integral part of functional prostate MRI [5]. DWI depends on the microscopic mobility of water molecules, known as “Brownian motion”, which is restricted in tissue compared with pure water. Several studies showed a significant association between the apparent diffusion coefficient (ADC), which quantifies the signal intensity loss in DWI, and the tumor aggressiveness expressed by the Gleason score [6–8]. Different models to evaluate DWI have been developed, including a simple monoexponential model, a biexponential model with two independent fractions, a statistical model with a distribution fraction and, recently, kurtosis imaging [9]. Kurtosis imaging has shown a better fit to diffusion-weighted data compared with the standard monoexponential ADC [10]. Although DWI is routinely applied in prostate MRI, only four studies dealing with kurtosis imaging in prostate MRI are available in the literature [11–14]. All four studies measured diffusion with an extended range of b-values in only three independent gradient directions. However, according to Jensen et al. [9,15], measuring the full diffusional kurtosis tensor requires at least 15 independent gradient directions and three different b-values. Therefore, this is the first study to assess the feasibility of full diffusional kurtosis tensor imaging in prostate MRI. Histopathological correlation was achieved by targeted biopsy.
T2-weighted axial images were acquired using a turbo spin echo sequence (TR 10630 ms, TE 117 ms, field of view [FOV] 12.8 cm, voxel size 0.5 × 0.5 × 3.0 mm, image matrix 256 × 256, turbo factor 23, number of slices = 30). DKI was performed with a single-shot spin-echo echo-planar sequence (TR 1700 ms, TE 101 ms, FOV 20.4 cm, voxel size 1.5 × 1.5 × 6.0 mm, image matrix 136 × 136, GRAPPA parallel imaging scheme, acceleration factor 2, and acquisition time 7:00 min, number of slices = 10). To acquire DKI, four b-values (0, 300, 600, 1000 s/mm 2) with four averages, applying independent diffusion gradients in 20 directions for each b-value, were used. DCEMRI images, T2-weighted sagittal images, T2-weighted coronal images, and T1-weighted axial images were also acquired, but not assessed as part of this study. 2.3. Targeted biopsy Targeted biopsy of the described lesions was performed either with MR-guided in-bore biopsy or with stereotactic ultrasoundguided MRI fusion biopsy. MR-guided biopsies were performed on a 3-T MR scanner (Magnetom Trio; Siemens Medical Systems, Erlangen, Germany) using the dynatrim biopsy device (Invivo, Orlando, FL, USA) and the corresponding software DynaCAD (Invivo, Orlando, FL, USA). Stereotactic ultrasound-guided biopsies used realtime transrectal ultrasound guidance with MR/TRUS image fusion (Urostation®, Koelis, La Tronche, France). 2.4. Post-processing and measurements
2. Materials and methods 2.1. Subjects Subjects of this feasibility study were previously enrolled in ongoing, prospective trials assessing MRI-guided in-bore prostate biopsy and stereotactic ultrasound-guided MRI fusion biopsy at our institution. The detailed design of this prospective trial has been reported elsewhere [16]. The study was approved by the local ethics committee. Informed consent was obtained from all subjects. Thirty-one men were prospectively included in the study, who underwent functional prostate MRI between January 2012 and May 2013. Twenty-one of these men (mean age 71 years, range 55 to 77 years) were referred to our hospital with increased PSA values (N4 ng/ml). After performing T2-weighted axial images, a radiologist with 4 years’ experience in prostate MRI decided if patients had suspicious lesions. Lesions should be large enough to distinctively place region-of-interest (ROI)-based measurements without partial volume effects in the lesions. In case of suspicious lesions in T2weighted imaging, DKI measurement was performed in addition to the standard prostate MRI protocol. Following the local procedure, a maximum of three lesions was defined per patient in the written report, including suspicious and benign findings. Patients with suspicious lesions underwent targeted biopsy of all described lesions. The other 10 included men were volunteers (mean age 26 years, range 24 to 32 years) without any known prostate disease. 2.2. MR techniques All prostate MRI examinations were performed using a 3-T MRI scanner (Magnetom Trio, A TIM System; Siemens Medical Systems, Erlangen, Germany) with a six-channel phased-array body-coil and a 24-channel spine array coil. Before the examination, patients received 20 mg of butylscopolamine (Buscopan®, Boehringer Ingelheim Pharma, Ingelheim, Germany) intravenously and intramuscularly. The 10 volunteers did not receive any bowel relaxant.
DKI datasets were post-processed using the software tool Diffusional Kurtosis Estimator (DKE, Version 2.5.1, built on 16 Dec 17 2012) on an external workstation [17]. Kurtosis (axial, mean, radial) parametric maps and fractional anisotropy (FA) maps were calculated with the DKE. ROI-based measurements were performed in these parameter maps using the software MRIcro (www.mricro.com, Version 1.40, Chris Rorden, University of South Carolina, USA) on an external workstation. ROIs were defined as large as possible by a radiologist with 4 years’ experience in prostate MRI, in the areas histopathologically confirmed by targeted biopsy, using axial T2weighted images as reference. ROIs were copied in the different parametric maps, derived from diffusion-weighted imaging, to ensure that always the same areas were evaluated. Measurements in the peripheral zone and central gland were repeated three times in different areas to calculate mean values. ADC values were obtained from the MRI scanner software using the standard monoexponential model using ROI-based measurements in the picture archiving and communication system (Sectra Workstation IDS7, Sectra AB, Linköping, Sweden). 2.5. Statistics All data are expressed as mean ± standard deviation. Parameters were compared using the independent sample t-test. Statistical significance was defined as a p b 0.05. R 2 was calculated to assess the goodness of fit for the DKI and the monoexponential model to the diffusional data in all volunteers. Correlation coefficients were calculated for the different model parameters and the Gleason score. 3. Results Of the 21 included patients, 10 patients were biopsied using MRguided in-bore biopsy and another four patients received a stereotactic ultrasound-guided MRI fusion biopsy. Histopathological results were positive for PCa in nine patients and 17 different lesions. Gleason score distribution was as follows: Gleason score 3 + 3 = 6
Please cite this article as: Quentin M, et al, Feasibility of diffusional kurtosis tensor imaging in prostate MRI for the assessment of prostate cancer: Preliminary results, Magn Reson Imaging (2014), http://dx.doi.org/10.1016/j.mri.2014.04.005
M. Quentin et al. / Magnetic Resonance Imaging xxx (2014) xxx–xxx
3
Fig. 1. Normalized signal intensity in a representative tumor and the peripheral zone (PZ) of the same patient derived from diffusional data for the different b-values. The kurtosis formula shows considerably better goodness of fit compared with the monoexponential model, both visually and mathematically (expressed by higher R2).
(n = 1), 3 + 4 = 7 (n = 7), 4 + 3 = 7 (n = 6), 4 + 4 = 8 (n = 1), and 4 + 4 = 9 (n = 2), respectively. Biopsy revealed prostatitis in five patients in 12 different lesions. Seven patients did not receive a targeted biopsy and had to be excluded. DKI provided a substantially better fit to the diffusional data compared with the monoexponential model (Fig. 1). The goodness of fit of diffusional data in all volunteers was R 2 = 0.9967 in the peripheral zone and R 2 = 0.9947 in the central gland for DKI compared with R 2 = 0.9486 in the peripheral zone and R 2 = 0.9417 in the central gland for the monoexponential model. Image quality was sufficient in all examined patients and volunteers (Fig. 2). The different mean values and standard deviations for DKI and the monoexponential model are shown in Table 1. Values for axial
kurtosis, mean kurtosis, and radial kurtosis were not significantly different, each compared in the peripheral zone, the central gland, prostatitis, and PCa. For the different diffusion parameters, only the monoexponential ADC could significantly (p = 0.03, Table 2) distinguish between the peripheral zone, the central gland, prostatitis, and PCa (Fig. 3). Significant differentiation (p = 0.03) between PCa and prostatitis was possible with axial kurtosis, mean kurtosis, and ADC. Axial and mean kurtosis only failed to significantly differentiate the central gland from prostatitis (Fig. 4). The parameters with the worst performance were FA (Fig. 5) and radial kurtosis, which could only differentiate PCa or prostatitis from the normal peripheral zone. There was a weak correlation between the Gleason score and all kurtosis parameters (axial kurtosis r = 0.19, mean kurtosis r =
Fig. 2. Image of a dorsal right tumor in the peripheral zone in T2-weighted imaging (T2), apparent diffusion coefficient map (ADC), fractional anisotropy (FA), mean kurtosis (MK), axial kurtosis (AK), and radial kurtosis (RK). Kurtosis parameter maps are color-coded. The tumor is sufficiently visible in all parameters used.
Please cite this article as: Quentin M, et al, Feasibility of diffusional kurtosis tensor imaging in prostate MRI for the assessment of prostate cancer: Preliminary results, Magn Reson Imaging (2014), http://dx.doi.org/10.1016/j.mri.2014.04.005
4
M. Quentin et al. / Magnetic Resonance Imaging xxx (2014) xxx–xxx
4.2. Parameters derived from full diffusional kurtosis tensor imaging
Table 1 Mean values and standard deviation. FA PZ CZ PCa Pr
0.13 0.15 0.18 0.18
Kmean
Kax ± ± ± ±
0.03 0.02 0.05 0.05
1.09 1.40 1.78 1.41
± ± ± ±
0.12 0.12 0.39 0.19
1.16 1.44 1.84 1.50
± ± ± ±
Krad 0.13 0.17 0.43 0.17
1.18 1.48 1.81 1.54
ADC (10−3 mm2/s) ± ± ± ±
0.13 0.22 0.48 0.18
1.08 0.77 0.51 0.64
± ± ± ±
0.12 0.07 0.16 0.14
Values for the peripheral zone (PZ), central gland (CZ), prostate cancer (PCa), and prostatitis (Pr) using diffusional kurtosis tensor imaging for acquiring fractional anisotropy (FA), axial kurtosis (Kax), mean kurtosis (Kmean), radial kurtosis (Krad), and the monoexponential model for calculating the apparent diffusion coefficient (ADC, 10−3 mm2/s).
0.14, radial kurtosis r = 0.15, and FA r = − 0.01) and the ADC derived from the monoexponential model (r = − 0.23). 4. Discussion Our results show that mean kurtosis and axial kurtosis derived from full diffusional kurtosis imaging (DKI) can significantly differentiate PCa from prostatitis, the peripheral zone, or the central gland. These parameters only failed to significantly discriminate prostatitis from normal central gland. Beyond that, fractional anisotropy and radial kurtosis could not significantly differentiate PCa from the central gland or from prostatitis. 4.1. Diffusion models Using the out-of-the-box monoexponential model is a common and the easiest way to characterize the decay of signal intensity in DWI using the ADC. In recent years, more complex models were introduced. Since these models often use more free parameters they provide a better fit to the diffusion signal decay [13]. Some of these models make assumptions of the existence of different diffusion compartments, like the intra-voxel incoherent motion (IVIM) — model described by Le Bihan et al. [18]. DKI, on the other hand, does not make any assumption on the number or even existence of multiple compartments [19]. DKI is an extension of diffusion tensor imaging (DTI) [15]. With DTI, the degree of directionality can be obtained using FA and the overall movement of molecules using mean diffusivity (MD), and the ADC. DTI has been applied in prostate MRI in a few previous studies [20–24]. In DTI, a Gaussian distribution of diffusion is assumed, which can be hindered or restricted in complex biologic systems by the existence of various complex tissue structures [9]. DKI therefore describes the extent of deviation of the diffusion from the Gaussian form.
Table 2 P-values for the comparison of the different regions. P-values
PZ b −− N PCa
CZ b −− N PCa
PZ b −− N Pr
CZ b −− N Pr
PCa b −− N Pr
FA Kax Kmean Krad ADC
b0.01 b0.01 b0.01 b0.01 b0.01
0.10 0.01 0.01 0.05 b0.01
0.01 b0.01 b0.01 b0.01 b0.01
0.17 0.83 0.39 0.51 0.02
0.75 0.01 0.02 0.07 0.03
P-values derived from student’s t-test comparing values for the peripheral zone (PZ), central gland (CZ), prostate cancer (PCa), and prostatitis (Pr) calculated with diffusional kurtosis tensor imaging: fractional anisotropy (FA), axial kurtosis (Kax), mean kurtosis (Kmean), radial kurtosis (Krad), and the monoexponential model for calculating the apparent diffusion coefficient (ADC). Statistically significant differences (p b 0.05) are shaded in gray.
Although DWI as an integral part of functional prostate MRI has been investigated in numerous studies, there are only four studies available in the literature that deal with DKI in prostate MRI [11–14]. Measuring diffusion gradients in only three orthogonal directions, these studies did not assess the full diffusional kurtosis tensor. According to Jensen at al., who were the first to describe DKI, measurements in at least 15 independent gradient directions are necessary for calculation of the full diffusional kurtosis tensor [9,15]. Mean kurtosis, which is comparable to kurtosis in the three other studies, showed the same tendency, with higher values in PCa than in healthy tissue. In this study the different kurtosis parameters axial kurtosis, mean kurtosis, radial kurtosis showed comparable – not significantly different – values. A possible explanation might be the relatively isotropic diffusion in the prostate gland. In the literature, FA values gained only from DTI measurements are controversial. While some studies report higher values in PCa [20,21], others report lower [23] or equal [22,24] values compared with the peripheral zone. In this study, FA values were higher in PCa than in the peripheral zone. FA was able to discriminate PCa or prostatitis from the normal peripheral zone but was not able to distinguish between PCa and prostatitis, and failed in the central gland. Similar to several previous studies, this study demonstrated lower ADC values in PCa compared with the peripheral zone, and a negative correlation to the tumor grade [6,7]. The normal peripheral zone could be distinguished from PCa with every DKI parameter. Nevertheless, only the ADC was able to differentiate prostatitis from the central gland. This is in line with recent DWI studies reporting significant differences for the different tissues in the central gland (healthy tissue, PCa, different types of prostatitis) but also a considerably overlap using only DWI as functional imaging method [25,26].
4.3. Choice of b-values and signal-to-noise ratio The values of measured diffusional kurtosis tensor imaging parameters are different in all studies, most likely due to the different b-values used [13]. This study applied b-values up to 1000 s/mm 2, while the others used b-values up to 2000 [11,12], 2300 [14], and 800 [13] s/mm 2. Today, there is still limited consensus about the optimal b-value selection in DWI in functional prostate MRI [27]. The choice of b-value should not be too high because the quadratic term of the kurtosis formula will increase with b-values after a certain minimum [28]. According to Jensen et al., b-values should be somewhat larger than those usually applied in DWI, so that the departure from linearity is clearly apparent [9]. Fig. 1 shows that this departure from linearity could clearly be observed in our data when using b-values up to 1000 s/mm 2. Additionally there is limited consensus to include a minimum b-value of 0 s/mm 2. Maas et al. suggested to use a minimum b-value of 100 s/mm 2 to measure pure diffusion and to avoid perfusion effects in the lower b-value range [29]. On the other hand a minimum b-value of 0 s/mm 2 has been recommended for the monoexponential model in a European consensus meeting dealing with standardization of prostate MRI [27]. Closely connected to the choice of b-values is the signal-to-noise ratio with decreasing diffusion signal in higher b-values and constant background noise. According to Tabesh et al. [17] noise can artificially increase signal intensity and therefore yield falsely large kurtosis estimations. Finally an optimal choice of b-values for DKI is also tissue dependent, and therefore this may differ for prostate and other organs such as brain.
Please cite this article as: Quentin M, et al, Feasibility of diffusional kurtosis tensor imaging in prostate MRI for the assessment of prostate cancer: Preliminary results, Magn Reson Imaging (2014), http://dx.doi.org/10.1016/j.mri.2014.04.005
M. Quentin et al. / Magnetic Resonance Imaging xxx (2014) xxx–xxx
5
Fig. 3. Box plot of apparent diffusion coefficient (ADC; 10−3 mm2/s) values derived from the monoexponential model for the peripheral zone (PZ), the central gland (CZ), prostatitis, and prostate cancer.
4.4. Advantages of full diffusional kurtosis tensor imaging Altogether it is questionable whether the additional effort to measure and calculate DKI is justified. Rosenkrantz et al. suggested that DKI might be useful for predicting the final pathologic outcome for patients under active surveillance [12]. On the other hand, this study show that DKI measurements are feasible in clinical routine with little additional measuring time, since normal DWI can be skipped, and ADC calculation can be done with DKI measurements. 4.5. Limitations This study has some limitations. First of all, this study is limited by the rather small number of patients included. This is largely because patients were included before the biopsy and patients without a biopsy had to be excluded. We believe that a targeted biopsy performed either as in-bore MRI-guided biopsy or targeted ultrasound MRI fusion biopsy serves as a better reference than a random TRUS-guided biopsy. This limitation especially applies to the correlation to the Gleason score. Studies including a larger number
of patients should be performed to clarify the stability of the correlation between DKI and Gleason score. The second limitation is the restriction of the study population to patients with visible suspicious lesions in T2-weighted imaging. This had to be done to ensure the ROI-based measurements could be placed in the lesions without partial volume effects. Therefore, no sensitivity or detection rates were calculated in this study. The main aim was to test the feasibility of DKI in clinical routine for the first time. Another limitation is the – compared to the other imaging modalities – higher slice thickness of 6 mm for the DKI measurements, which was necessary to achieve adequate signal-to-noise ratio. 5. Conclusion In conclusion, although full DKI requires extended measuring and calculation time, it is feasible to utilize it in a routine clinical setting. DKI can significantly distinguish PCa from the normal peripheral zone or the central gland. The additional value of DKI compared with conventional monoexponential ADC calculation remains questionable and requires further research.
Fig. 4. Box plot of mean kurtosis (MK; dimensionless) values derived from diffusional kurtosis tensor imaging for the peripheral zone (PZ), the central gland (CZ), prostatitis, and prostate cancer.
Please cite this article as: Quentin M, et al, Feasibility of diffusional kurtosis tensor imaging in prostate MRI for the assessment of prostate cancer: Preliminary results, Magn Reson Imaging (2014), http://dx.doi.org/10.1016/j.mri.2014.04.005
6
M. Quentin et al. / Magnetic Resonance Imaging xxx (2014) xxx–xxx
Fig. 5. Box plot of fractional anisotropy (FA; dimensionless) values derived from diffusional kurtosis tensor imaging for the peripheral zone (PZ), the central gland (CZ), prostatitis, and prostate cancer.
References [1] Pashayan N, Duffy SW, Pharoah P, Greenberg D, Donovan J, Martin RM, et al. Mean sojourn time, overdiagnosis, and reduction in advanced stage prostate cancer due to screening with PSA: implications of sojourn time on screening. Br J Cancer 2009;100:1198–204. [2] Rajinikanth A, Manoharan M, Soloway CT, Civantos FJ, Soloway MS. Trends in gleason score: concordance between biopsy and prostatectomy over 15 years. Urology 2008;72:177–82. [3] Heijmink SW, van Moerkerk H, Kiemeney LA, Witjes JA, Frauscher F, Barentsz JO. A comparison of the diagnostic performance of systematic versus ultrasoundguided biopsies of prostate cancer. Eur Radiol 2006;16:927–38. [4] Mazaheri Y, Shukla-Dave A, Muellner A, Hricak H. MR imaging of the prostate in clinical practice. MAGMA 2008;21:379–92. [5] Barentsz JO, Richenberg J, Clements R, Choyke P, Verma S, Villeirs G, et al. ESUR prostate MR guidelines 2012. Eur Radiol 2012;22:746–57. [6] Vargas HA, Akin O, Franiel T, Mazaheri Y, Zheng J, Moskowitz C, et al. Diffusionweighted endorectal MR imaging at 3 T for prostate cancer: tumor detection and assessment of aggressiveness. Radiology 2011;259:775–84. [7] Tamada T, Sone T, Jo Y, Toshimitsu S, Yamashita T, Yamamoto A, et al. Apparent diffusion coefficient values in peripheral and transition zones of the prostate: comparison between normal and malignant prostatic tissues and correlation with histologic grade. J Magn Reson Imaging 2008;28:720–6. [8] Hambrock T, Somford DM, Huisman HJ, van Oort IM, Witjes JA, Hulsbergen-van de Kaa CA, et al. Relationship between apparent diffusion coefficients at 3.0-T MR imaging and Gleason grade in peripheral zone prostate cancer. Radiology 2011;259:453–61. [9] Jensen JH, Helpern JA, Ramani A, Lu H, Kaczynski K. Diffusional kurtosis imaging: the quantification of non-gaussian water diffusion by means of magnetic resonance imaging. Magn Reson Med 2005;53:1432–40. [10] Jansen JF, Stambuk HE, Koutcher JA, Shukla-Dave A. Non-gaussian analysis of diffusion-weighted MR imaging in head and neck squamous cell carcinoma: a feasibility study. AJNR Am J Neuroradiol 2010;31:741–8. [11] Rosenkrantz AB, Sigmund EE, Johnson G, Babb JS, Mussi TC, Melamed J, et al. Prostate cancer: feasibility and preliminary experience of a diffusional kurtosis model for detection and assessment of aggressiveness of peripheral zone cancer. Radiology 2012;264:126–35. [12] Rosenkrantz AB, Prabhu V, Sigmund EE, Babb JS, Deng FM, Taneja SS. Utility of diffusional kurtosis imaging as a marker of adverse pathologic outcomes among prostate cancer active surveillance candidates undergoing radical prostatectomy. AJR Am J Roentgenol 2013;201:840–6. [13] Quentin M, Blondin D, Klasen J, Lanzman RS, Miese FR, Arsov C, et al. Comparison of different mathematical models of diffusion-weighted prostate MR imaging. Magn Reson Imaging 2012;30:1468–74. [14] Mazzoni LN, Lucarini S, Chiti S, Busoni S, Gori C, Menchi I. Diffusion-weighted signal models in healthy and cancerous peripheral prostate tissues: comparison
[15] [16]
[17]
[18]
[19]
[20] [21] [22]
[23]
[24]
[25]
[26]
[27]
[28]
[29]
of outcomes obtained at different b-values. J Magn Reson Imaging 2013;39:512–8. Jensen JH, Helpern JA. MRI quantification of non-gaussian water diffusion by kurtosis analysis. NMR Biomed 2010;23:698–710. Arsov C, Blondin D, Rabenalt R, Antoch G, Albers P, Quentin M. Standardised scoring of a multi-parametric 3-T MRI for a targeted MRI-guided prostate biopsy. Urologe A 2012;51:848–56. Tabesh A, Jensen JH, Ardekani BA, Helpern JA. Estimation of tensors and tensorderived measures in diffusional kurtosis imaging. Magn Reson Med 2011;65:823–36. Le Bihan D, Breton E, Lallemand D, Aubin ML, Vignaud J, Laval-Jeantet M. Separation of diffusion and perfusion in intravoxel incoherent motion MR imaging. Radiology 1988;168:497–505. Lu H, Jensen JH, Ramani A, Helpern JA. Three-dimensional characterization of non-gaussian water diffusion in humans using diffusion kurtosis imaging. NMR Biomed 2006;19:236–47. Gibbs P, Pickles MD, Turnbull LW. Diffusion imaging of the prostate at 3.0 tesla. Invest Radiol 2006;41:185–8. Gürses B, Kabakci N, Kovanlikaya A, Firat Z, Bayram A, Uluğ AM, et al. Diffusion tensor imaging of the normal prostate at 3 tesla. Eur Radiol 2008;18:716–21. Kozlowski P, Chang SD, Meng R, Mädler B, Bell R, Jones EC, et al. Combined prostate diffusion tensor imaging and dynamic contrast enhanced MRI at 3 T— quantitative correlation with biopsy. Magn Reson Imaging 2010;28:621–8. Manenti G, Carlani M, Mancino S, Colangelo V, Di Roma M, Squillaci E, et al. Diffusion tensor magnetic resonance imaging of prostate cancer. Invest Radiol 2007;42:412–9. Xu J, Humphrey PA, Kibel AS, Snyder AZ, Narra VR, Ackerman JJ, et al. Magnetic resonance diffusion characteristics of histologically defined prostate cancer in humans. Magn Reson Med 2009;61:842–50. Hoeks CM, Vos EK, Bomers JG, Barentsz JO, Hulsbergen-van de Kaa CA, Scheenen TW. Diffusion-weighted magnetic resonance imaging in the prostate transition zone: histopathological validation using magnetic resonance-guided biopsy specimens. Invest Radiol 2013;48:693–701. Nagel KN, Schouten MG, Hambrock T, Litjens GJ, Hoeks CM, ten Haken B, et al. Differentiation of prostatitis and prostate cancer by using diffusion-weighted MR imaging and MR-guided biopsy at 3 T. Radiology 2013;267:164–72. Dickinson L, Ahmed HU, Allen C, Barentsz JO, Carey B, Futterer JJ, et al. Magnetic resonance imaging for the detection, localisation, and characterisation of prostate cancer: recommendations from a European consensus meeting. Eur Urol 2011;59:477–94. Hui ES, Cheung MM, Qi L, Wu EX. Towards better MR characterization of neural tissues using directional diffusion kurtosis analysis. Neuroimage 2008;42:122–34. Maas MC, Fütterer JJ, Scheenen TW. Quantitative evaluation of computed high b value diffusion-weighted magnetic resonance imaging of the prostate. Invest Radiol 2013;48:779–86.
Please cite this article as: Quentin M, et al, Feasibility of diffusional kurtosis tensor imaging in prostate MRI for the assessment of prostate cancer: Preliminary results, Magn Reson Imaging (2014), http://dx.doi.org/10.1016/j.mri.2014.04.005
