A Comparison of Arterial Spin Labeling Perfusion MRI and DCEMRI In

Research article

Received: 25 December 2012, Revised: 21 March 2014, Accepted: 21 March 2014, Published online in Wiley Online Library: 9 May 2014

(wileyonlinelibrary.com) DOI: 10.1002/nbm.3124 A comparison of arterial spin labeling perfusion MRI and DCE-MRI in human prostate cancer Wenchao Caia, Feiyu Lia, Jing Wangb,c, Huarui Dub, Xiaoying Wanga,c*, Jue Zhangb,c, Jing Fangb,c and Xuexiang Jianga

Perfusion MRI has the potential to provide pathophysiological biomarkers for the evaluating, staging and therapy monitoring of prostate cancer. The objective of this study was to explore the feasibility of noninvasive arterial spin labeling (ASL) to detect prostate cancer in the peripheral zone and to investigate the correlation between the blood trans flow (BF) measured by ASL and the pharmacokinetic parameters K (forward volume transfer constant), kep (reverse reflux rate constant between extracellular space and plasma) and ve (the fractional volume of extracellular space per unit volume of tissue) measured by dynamic contrast-enhanced (DCE) MRI in patients with prostate cancer. Forty-three consecutive patients (ages ranging from 49 to 86 years, with a median age of 74 years) with pathologi- cally confirmed prostate cancer were recruited. An ASL scan with four different inversion times (TI = 1000, 1200, trans 1400 and 1600 ms) and a DCE-MRI scan were performed on a clinical 3.0 T GE scanner. BF, K , kep and ve maps were calculated. In order to determine whether the BF values in the cancerous area were statistically different from those in the noncancerous area, an independent t-test was performed. Spearman’s bivariate correlation was used to trans assess the relationship between BF and the pharmacokinetic parameters K , kep and ve. The mean BF values in the cancerous areas (97.1 ± 30.7, 114.7 ± 28.7, 102.3 ± 22.5, 91.2 ± 24.2 ml/100 g/min, respectively, for TI = 1000, 1200, 1400, 1600 ms) were significantly higher (p < 0.01 for all cases) than those in the noncancerous regions (35.8 ± 12.5, 42.2 ± 13.7, 53.5 ± 19.1, 48.5 ± 13.5 ml/100 g/min, respectively). Significant positive correlations (p < 0.01 trans for all cases) between BF and the pharmacokinetic parameters K , kep and ve were also observed for all four TI values (r = 0.671, 0.407, 0.666 for TI = 1000 ms; 0.713, 0.424, 0.698 for TI = 1200 ms; 0.604, 0.402, 0.595 for TI = 1400 ms; 0.605, 0.422, 0.548 for TI = 1600 ms). It can be seen that the quantitative ASL measurements show significant differences between cancerous and benign tissues, and exhibit strong to moderate correlations with the parameters obtained using DCE-MRI. These results show the promise of ASL as a noninvasive alternative to DCE-MRI. Copyright © 2014 John Wiley & Sons, Ltd.

Keywords: arterial spin labeling (ASL); prostate cancer; DCE-MRI; perfusion imaging; blood ﬂow (BF)

INTRODUCTION

Prostate cancer (PCa) is the second most frequently diagnosed * Correspondence to: X. Wang, Department of Radiology, Peking University First malignancy and the sixth leading cause of cancer-related mor- Hospital, Peking University, Beijing, China. tality in males worldwide (1). Angiogenesis plays a vital role in E-mail: [email protected] the growth, progression and metastasis processes of malignant tumors (2–6). Through histopathologic (7,8) and dynamic con- a W. Cai, F. Li, X. Wang, X. Jiang trast-enhanced (DCE) MRI findings (9–19), PCa has been shown Department of Radiology, Peking University First Hospital, Peking University, Beijing, China to be a hypervascular tumor. Unlike highly controlled normal physiologic vessels, the tumor vasculatures are chaotic and b J. Wang, H. Du, J. Zhang, J. Fang permeable, making them excellent targets for early detection, Department of Biomedical Engineering, Peking University, Beijing, China accurate localization and curative therapy (4,6,20). Noninvasive c J. Wang, X. Wang, J. Zhang, J. Fang MR biomarkers that are sensitive to angiogenic changes can Center for Functional Imaging, Advanced Academy of Interdisciplinary be good candidates for evaluating therapeutic effects (2,20). Sciences, Peking University, Beijing, China DCE-MRI has been used widely to acquire pharmacokinetic vascular characteristics of prostate cancerous tissues (such as time Abbreviations used: AIF, arterial input function; ASL, arterial spin labeling; fl to peak, speed of contrast uptake, Ktrans, k and v )(9–19,21), but it BF, blood ow; CNR, contrast-to-noise ratio; DCE, dynamic contrast enhanced; ep e DESPOT1, driven equilibrium single pulse observation of T1; EES, extravascular requires venous cannulation and administration of a gadolinium extracellular space; EPI, echo planar imaging; FAIR, flow-sensitive alternating fi chelate contrast agent. Nephrogenic systemic brosis (NSF) has inversion recovery; FOV, field of view; FSE, fast spin echo; kep, reverse reflux rate recently been linked to gadolinium-based contrast agents constant between extracellular space and plasma; Ktrans, forward volume fi (22–25). In this respect, DCE-MRI is not suitable for patients transfer constant; NSF, nephrogenic systemic brosis; PASL, pulsed arterial spin fi labeling; PCa, prostate cancer; PZ, peripheral zone; ROI, region of interest; SE, with renal insuf ciency and is especially inconvenient for spin echo; SNR, signal-to-noise ratio; SSFSE, single-shot fast spin echo; TI, inver- PCa patients who are required to have repeated follow-ups sion time; TRUS, transrectal ultrasound; ve, the fractional volume of extracellu- for therapeutic effect monitoring. lar extravascular space. 817

Arterial spin labeling (ASL) perfusion MRI is a noninvasive, Table 1. The clinical characteristics of 43 patients with PCa nonradioactive and non-contrast-enhanced method capable of quantitatively measuring microvascular perfusion characteristics Characteristics Value of tissues by tagging arterial water (26–29), significantly ben- – efitting PCa patients with renal dysfunction or other contraindi- Median patient age (years)* 74 (49 86) cations to MR contrast agents. Recently, a feasibility study Median PSA level (ng/ml)*↑ 22.9 (5.67–-5000) showed that normal prostate perfusion can be measured with Pathologic tumor stage♂ the ASL technique (30). To the best of our knowledge, ASL has T2a 5 not yet been used for human PCa detection and staging. T2b 7 The application of traditional ASL for PCa is limited by the intrin- T2c 1 sically low signal-to-noise ratio (SNR), and this is further exacer- T3a 13 fl bated by the relatively slow ow of blood into the prostate from T3b 11 multiple feeding vessels (31). On the other hand, the pulsed arterial T4 6 spin labeling (PASL) method provides a better balance between Gleason score♂ tagging efficiency and SNR, and has been proven to be effective 66 and repeatable in many studies of cerebral diseases (28,32,33) 712 and renal function (29). However, arrival times differ between path- 813 ological and physiological neovascularization settings, and this 912 necessitates the optimization of inversion times (TI) (26,30,33,34). In this study, we sought to explore the feasibility of the PASL *Numbers in parentheses are range. ↑PSA, prostate specific antigen. technique with an imaging readout that is different from the ♂ one that was employed in the previous feasibility study of Data are numbers of patients. prostate ASL perfusion imaging (30). The results were compared The staging was according to the AJCC 7th ed. (2010) (54). with those of DCE-MRI. Furthermore, in order to investigate possible transit time differences between benign and malignant tissues, we used four distinct inversion times. eight-channel pelvic phased-array coil. RF transmission was provided by the body coil. METHODS Transverse and coronal T2-weighted fast spin-echo (FSE) images (TR/TE = 3000/130 ms, NEX = 4 and matrix = 320 × 224), Patient population transverse T1-weighted FSE images (TR/TE = 620/7 ms, NEX = 1 The prospective study was approved by the institutional review and matrix = 320 × 224), diffusion-weighted spin-echo/echo board of the local ethics committee and informed consent was planar imaging (SE/EPI) images (TR/TE = 1000/57.3 ms, NEX = 4 obtained from all patients prior to MRI. Fifty-three consecutive and matrix = 128 × 128) were acquired. All sequences mentioned patients with PCa confirmed by transrectal ultrasound (TRUS)- above were performed using a 4 mm section thickness with an guided systemic biopsy (12 or more cores) were recruited between intersection gap of 1 mm. The field of view (FOV) was 26 cm July 2011 and October 2012. All of the patients underwent MR and the number of slices was 16. The phase encoding gradient examination before biopsy. The time between the MRI examina- was from right to left in order to reduce the motion artifacts tion and the biopsy was less than 1 month. caused by the bladder and the rectum.

Inclusion and exclusion criteria T1 mapping Inclusion criteria were as follows: (a) patients with histologically The baseline T1 mapping was performed via the driven equilibrium proven PCa; (b) visible PCa lesion on the MR images; (c) no prior single pulse observation of T1 (DESPOT1) method (35). A pair of history of PCa treatment (such as antiandrogen treatment, radia- spoiled gradient recalled echo images were acquired at optimized tion therapy or chemotherapy). flip angles before DCE-MRI scan. Other parameters were as follows: Exclusion criteria were as follows: (a) presence of a cardiac 3D spoiled gradient recalled echo sequence, TR/TE = 6.0/2.8 ms, pacemaker or other electronic implant; (b) known allergy to flip angle = 3°/12°, matrix = 256 × 256, FOV = 26 cm, section gadolinium-based contrast agents; (c) known renal insuffi- thickness = 5.0 mm, 26 slices, NEX = 2 and bandwidth = 31.5 kHz. ciency; (d) reported claustrophobia; (e) failure to give written informed consent. ASL perfusion imaging Four patients were excluded due to impaired renal elimination fl function (glomerular filtration rate less than 30 ml/min), which The PASL protocol used a ow-sensitive alternating inversion prevented their DCE-MRI examination. Six patients were ex- recovery (FAIR) sequence combined with a single-shot fast cluded from the data analysis owing to substantial motion spin-echo (SSFSE) imaging block (FAIR-SSFSE), with four different during image acquisition in either DCE-MRI (n = 4) or ASL inversion times (TI = 1000, 1200, 1400 and 1600 ms), and other fl imaging (n = 2). As a result, 43 patients were included in the final parameters remained unchanged (TR = 3500 ms, TE = 60 ms, ip analysis. Details of the patient population are given in Table 1. angle = 90°, FOV = 26 cm, matrix = 96 × 96, bandwidth = 20.83 kHz and slice thickness = 5 mm). The inversion times were determined based on previously published results (30) and our pre- MRI protocol liminary experiment. A single oblique axial plane was placed

818 MR images were obtained with a clinical 3.0 T MR system (Signa through the maximum diameter of the PCa lesion. In order to Excite HD; GE Healthcare, Milwaukee, WI, USA) by using the avoid artifacts due to partially inverted spins at the margins,

wileyonlinelibrary.com/journal/nbm Copyright © 2014 John Wiley & Sons, Ltd. NMR Biomed. 2014; 27: 817–825 THE COMPARISON OF ASL PERFUSION MRI AND DCE-MRI IN PROSTATE CANCER the slab thickness for the section selective inversion in the FAIR The contrast-to-noise ratio (CNR) values of the FAIR signal on preparation was set to 30 mm. Twelve ASL images were obtained the selective and nonselective inversion images between at the same slice position, six with section-selective inversion cancerous and noncancerous areas were calculated respectively. and six with global inversion, and the images in each group were The signal intensities of prostate cancerous (SICa) and noncan- 2 averaged to ultimately yield pairs of averaged control and label cerous areas (SInonca) were measured with ROIs of 50–855 mm images for each of the four inversion times. M0 images were in size on the selective and nonselective inversion images. Image acquired by using SSFSE without FAIR preparation, the imaging noise (σN) was measured as the standard deviation of air signal parameters of which were the same as those of the FAIR-SSFSE intensity with an ROI of 50 mm2. sequence described above, while TR was set to 6000 ms to ensure complete longitudinal relaxation between measure- CNR ¼ ðÞSICa SInonca =σN: [2] ments. The total measurement time for one slice was within 1 min per TI value. Processing of DCE MR image Pharmacokinetic analysis based on the Tofts model and a two- DCE-MRI compartment model (39) yields the contrast agent transfer rate trans 1 The T1-weighted DCE-MRI was performed by running a modified between blood and the tissue K (min ), the extravascular ex- – 3D spoiled gradient-echo sequence (liver acquisition with volume tracellular space (EES) volume per unit volume ve (range 0.0 1.0) acceleration) (TR/TE = 4/1.9 ms, flip angle = 15°, matrix = 256 × 256, and the flux rate contrast between the EES and the plasma 1 FOV = 36 cm, in-plane reconstruction resolution = 0.7 × 0.7 mm, sec- kep (min ). Pixels were excluded if the pharmacokinetic fi – tion thickness/intersection gap = 3.0/1.5 mm, 36 slices, dynamic model tting did not converge or ve was outside the 0.0 1.0 range. time points = 15 and temporal resolution = 12 s). When the second The standardized contrast agent arterial input function (AIF) dynamic loop started, 0.5 mmol/ml Omniscan (GE Healthcare, Co. estimated in a previous clinical DCE-MRI study was used (40,41). trans Cork, Ireland) was administered at a dosage of 0.1 mmol/kg body The K , kep and ve maps were then fused with T2-weighted weight using a Spectris power injector (Medrad, Indianola, PA) with images using the automated image-registration software devel- a rate of 3 ml/s followed by an equal volume of saline flush, also at oped at our institution. The T1 map of the prostate was calculated 3ml/s.ThetotalscantimefortheDCEscanwasabout3min. by the DESPOT1 method (35).

Data postprocessing ROI analysis All images were transferred to an independent workstation for The cancerous and noncancerous areas were determined by quantitative analysis using an in-house program coded in considering the systemic biopsy results and the integrated Matlab2009 (MathWorks, Natick, MA, USA). interpretation of anatomic and functional MR findings, including T2-weighted, diffusion-weighted and DCE imaging, using the Prostate blood flow quantification scoring system recommended by the European Society of Urogenital Radiology (ESUR) prostate MR guidelines (42). Seven A quantitative model for the analysis of prostate blood flow patients were excluded from the benign area analysis due to (BF) with FAIR that made use of the extended Bloch equa- the diffuse lesion invading bilateral peripheral zones (PZs). A tions has been previously described (36). This model makes total of 43 cancerous and 36 noncancerous areas were used in it possible to simulate the difference ΔM(TI) between the lon- the analyses. Two radiologists (WC C, PhD and FY L, MD, PhD), gitudinal magnetizations M and M resulting from FAIR nonsel sel with 3 and 10 years of prostate MRI interpretation experience, experiments with, respectively, global preparation and section- drew ROIs by visual inspection in consensus on the BF map selective preparation (37): trans and K map. The T2-weighted image, apparent diffusion coef- fi ΔMTðÞ¼I MselðÞTI –MnonselðÞ¼TI 2M0TIðÞf=λ expðÞTI=T 1 [1] cient map and DCE image corresponding to the BF map and Ktrans map were observed to assist in the identification of the de- where M0 represents the tissue equilibrium magnetization per tailed anatomy of the prostate. While drawing the ROIs, the im- unit mass of the tissue, T1 is the longitudinal relaxation time of ages were magnified, and ROIs (50–855 mm2) covering as the tissue, f is the perfusion rate (usually expressed in milliliters much of the cancerous region as possible were drawn manually λ – per 100 g per minute) and is the blood tissue water partition on the slices with the largest PCa foci. The noncancerous ROI was fi coef cient, which is assumed to be nearly constant at 0.80. placed on the contralateral side of the tumor for each patient, Perfusion maps can be calculated pixel by pixel by analyzing and it was an oval ROI the same size as the cancerous ROI for Δ M for given TI, M0 and T1. each patient if the area of PZ without cancer was large enough. ASL images were postprocessed using an in-house MATLAB 2 trans If not, the ROI would be set to 50 mm . BF, K , kep and ve were program, which allowed for nonlinear image registration to then measured in the cancerous and noncancerous areas fi avoid BF quanti cation errors due to motion artifacts (38). located at the prostate PZ. All parameters were measured three Motion-corrected MR images with section-selective and global times at the same site, and the average was calculated. inversion were averaged separately, and the two average images were then subtracted to obtain the ΔM images. The T1 values for Statistical analysis the cancerous and noncancerous regions in the prostate were extracted from the T1 map acquired before the DCE-MRI. Pixels All statistical analyses were performed using SPSS 11.5 for with more than 400 ml/100 g/min perfusion were excluded from Windows. A two-sided value of p < 0.05 was considered to be fi the BF evaluation. The M0 image can provide the prostate statistically signi cant. anatomical information to facilitate placement of the regions of A paired t-test was used to compare the CNRs between the 819 interest (ROIs). selective and nonselective images, and we calculated the mean

and the standard deviation of the kinetic parameters (BF, Ktrans, Figs 2 and 3). The BF values for different Gleason score subtypes of kep, ve) for the cancerous and noncancerous areas. An indepen- PCa are shown in Table 4. Moreover, the permeability and dent t-test was used to compare the BF values in the cancerous perfusion pharmacokinetic analyses showed signiﬁcantly higher trans and noncancerous areas for each of the four TI values. Scatter K , kep and ve in the cancerous areas (p < 0.01 for all parame- trans plots of BF, K , kep and ve were generated and Spearman’s ters, Table 3). When all 79 areas are taken into account, signiﬁcant bivariate correlation analysis was conducted to investigate the positive correlations are observed between the BF values obtained trans trans correlation between BF and the DCE parameters K , kep and at different TI values and the pharmacokinetic parameters K , ve in both cancerous and noncancerous areas. kep and ve derived from DCE-MRI (p < 0.01, Figs 4–6).

RESULTS DISCUSSION

The CNRs between cancerous and noncancerous areas (N = 36) on Previous studies (30,43,44) indicate that, compared with the the selective images were greater than those on the nonselective brain or other organs, BF in normal prostate PZ is intrinsically inversion images (p < 0.01 for all TI, Table 2). The BF maps of one low (Fig. 1), while BF in the prostate cancerous area obviously patient without and one patient with PCa are displayed in Figs 1 increases. Despite many technical challenges, our study demon- and 2 respectively. In addition, the mean T1 relaxation times for strates that the PASL technique can be used to visualize and the cancerous and noncancerous regions in the PZ were estimate the BF differences between noncancerous and cancerous 1481 ± 39 ms and 1067 ± 36 ms respectively. The BF values in the areas in the prostate PZ (Fig. 2, Table 3), which are comparable with cancerous regions were significantly higher than those in the those in the literatures when using 15O positron emission tomo- noncancerous regions of the prostate PZ (p < 0.01, Table 3 and graphy (15.7 ± 7.5 versus 29.4 ± 7.8 ml/100 g/min) (43) or contrast- enhanced MRI (23 ± 21 versus 138 ± 127 ml/100 cm3/min) (44) or ASL MRI (25.8 ± 7.1 ml/100 cm3/min in whole healthy prostate) (30). The higher BF values in the malignant areas (Table 3) indicate Table 2. The CNRs between cancerous and noncancerous the hypervascular property of the prostate carcinoma, which is areas (N = 36) on the selective and nonselective inversion consistent with the previous findings from the pathologic (7,8) – TI = 1000 ms TI = 1200 ms TI = 1400 ms TI = 1600 ms and DCE-MRI studies (9 19). In a mean vessel density study, Wilson et al. (7) found that CNRnonsel 3.0 ± 5.4 11.5 ± 5.2 10.8 ± 6.6 8.4 ± 5.8 prostate adenocarcinoma had microvessels that were signifi- CNRsel 1.4 ± 5.9 13.7 ± 5.5 12.3 ± 7.0 9.8 ± 5.8 cantly higher in number and larger in size. Brawer et al. (8) further reported that the vessel density correlated with the pathologic Data are mean ± standard deviation. stage and supported this concept of prostate neovascularity as a All p < 0.01. prerequisite to tumor progression.

820 Figure 1. (A), (B) A 72-year-old man with negative biopsy result. The PZ shows hyperintensity on T2-weighted image and isointensity on diffusion- 2 trans weighted image; (C)–(F) ROIs of 50 mm are placed in the PZ areas in the corresponding BF maps with different TI values; (G)–(I) K , kep, ve maps.

Figure 2. (A), (B) A 77-year-old man with biopsy proven cancer in right PZ. The cancer shows hypointensity on T2-weighted image and hyperintensity on diffusion-weighted image; (C)–(F) the corresponding BF maps with different T1 values show higher BF value in the cancerous zone than the trans noncancerous zone in left PZ; (G)–(I) K , kep, ve maps also show higher values in the cancerous zone.

Table 3. Results of comparison of perfusion parameter BF trans and kinetic parameters (K , kep, ve) between cancerous and noncancerous areas of prostate

Parameter PCa N* = 43 Non-PCa N*=36 BF TI = 1000 ms 97.1 ± 30.7 35.8 ± 12.5 TI = 1200 ms 114.7 ± 28.7 42.2 ± 13.7 TI = 1400 ms 102.3 ± 22.5 53.5 ± 19.1 TI = 1600 ms 91.2 ± 24.2 48.5 ± 13.5 Ktrans 0.093 ± 0.031 0.046 ± 0.021 kep 0.181 ± 0.037 0.139 ± 0.046 ve 0.506 ± 0.162 0.330 ± 0.101 *N = Number of cancerous or noncancerous areas. trans 1 BF unit, ml/100 g/min; K , kep unit, min ; ve, no unit. Data are mean ± standard deviation. All p < 0.01.

Moreover, several studies have recently shown that semi- quantitative and quantitative perfusion parameters derived from DCE-MRI measured in the prostate are associated with the proportion of speciﬁc histological components that differ between benign and malignant PZ tissues (9–11,19). Many other DCE investigations (12–17) showed that the DCE kinetic parame- Figure 3. The two lines of error bars demonstrate the mean perfusion ters (such as Ktrans and k ) indicated that, in the cancerous ep parameter (BF) changes with four different single TI values (1000, 1200, region, the contrast agent had a high penetration rate during 1400, 1600 ms) in cancerous and noncancerous areas of the prostate. the exchange between the blood plasma and the extracellular extravascular space. An experiment (18) on prostate carcinomas in rats showed that the mean tumor Akep values derived from maximum value of the enhancement in the absence of contrast

DCE-MRI decreased with increasing volume, which indicates that elimination and kep is the rate of the efﬂux of contrast from the 821 the tumor grows with the development of angiogenesis (A is the extracellular extravascular space to plasma).

Table 4. Perfusion BF values for different Gleason score subtypes of PCa

Gleason BF (ml/100 g/min) score TI = 1000 ms TI = 1200 ms TI = 1400 ms TI = 1600 ms 6(N = 6) 97.9 ± 13.2 109.3 ± 13.4 99.8 ± 17.9 93.4 ± 22.6 7(N = 12) 92.3 ± 19.8 117.5 ± 22.8 104.4 ± 19.8 98.0 ± 25.2 8(N = 13) 108.9 ± 30.3 137.3 ± 35.2 127.5 ± 43.7 107.9 ± 34.8 9(N = 12) 113.8 ± 38.9 143.0 ± 47.5 117.7 ± 40.9 106.5 ± 51.6

Figure 4. Scatter plots showing correlations between BFs in different TI (1000, 1200, 1400, 1600 ms) and the pharmacokinetic parameter Ktrans.

In our study, prostate cancer shows higher ve values, which is DCE-MRI could equally well elaborate the pathological or physi- not consistent with several previous studies (13,17). One possible ological angiogenesis in the brain and kidney (33,34,45–47). explanation is that the increased number of leaky tumor The noninvasive ASL perfusion technique and the conven- microcapillary might have contributed to ve. The number of tional dynamic susceptibility sequences requiring exogenous advanced PCa patients recruited in this study is larger, and their contrast agent (46,47) may offer similar information; however, lesions possess larger extravascular and extracellular volumes ASL has several advantages over the other techniques. ASL compared with those reported in the relatively early stages. scanning using endogenous rather than exogenous contrast Our results show that there is a moderate to high correlation agent can avoid the risk of NSF (22–25). Moreover, venous between the BF values derived from ASL and the pharmacoki- cannulation and the power injection process for the administra- trans netic parameters K , kep and ve derived from bolus-tracking tion of the contrast agent are not needed in ASL, which con- DCE-MRI, in the cancerous prostate regions and the benign siderably simplifies and accelerates the clinical workflow, and PZs. In most cases, the DCE pharmacokinetic parameters yield consequently reduces cost. composite information about the perfusion and capillary permeability characteristics (39). The uncontrolled angiogenic process Technology requires that new capillaries be recruited from existing blood vessels, in order to ensure a constant supply of nutrients and The FSE sequence was employed in our study, rather than the oxygen, and to allow for the elimination of metabolic waste typical EPI method, because EPI is particularly susceptible to (4,5,20). The increased cancerous immature vasculatures contrib- air–tissue interfaces (26). In addition, the FSE sequence results ute to higher perfusion and surface permeability, which result in in higher SNR. trans 822 higher mean BF, K , kep and ve values in the cancerous region. The factors that influence the accuracy of the BF measure- Some other researchers have also demonstrated that ASL and ments performed by the ASL technique include arterial transit

Figure 5. Scatter plots showing correlations between BFs in different TI (1000, 1200, 1400, 1600 ms) and the pharmacokinetic parameter kep.

Figure 6. Scatter plots showing correlations between BFs in different TI (1000, 1200, 1400, 1600 ms) and the pharmacokinetic parameter ve. time to the targeted voxels, local relaxation time of the tissue in our study and it was found that the BF values differed (T1), equilibrium magnetization of the blood, inversion pulse significantly between the benign and the malignant tissues for shape and efficiency, fast versus intermediate water exchange all TI values (Fig. 3). When TI is 1200 ms, the difference seems and bolus dispersion (26–28,32,33,45). Transit time is considered to be the largest. to play the most prominent role on the accuracy of the Another important influential factor is the T1 relaxation time 823 quantification (27,28,30,33,34). Four different TI values were used of the local tissue, which may lead to an inaccurate

trans quantification of BF, K , kep and ve for different tissues. In this Luo and Bin Fan for providing invaluable assistance with the study, prostate BF quantification was performed using the local collection of conventional MR scanning information and contrast tissue T1 measured by the DESPOT1 method (35), which is able agent injection preparation. The authors also gratefully acknowl- to yield T1 relaxation time estimates that are comparable in qual- edge Cihat Eldeniz and Hongyang Yuan (Department of ity to those computed using the gold standard inversion recov- Radiology and Biomedical Research Imaging Center, University ery approach. of North Carolina at Chapel Hill, USA) for their linguistic assis- Furthermore, the FAIR related relative signal change between tance during the preparation of this manuscript and Jinnan selective and global inversion is typically of the order of a few Wang for his assistance in revising the manuscript. percent, and the CNR is of the order of 5–10. Blurring artefacts could easily mimic such a signal. In our study, there were no REFERENCES obvious blurring artefacts on the M0 images. The quality of M0 images was acceptable for analysis. 1. Jemal A, Bray F, Center MM, Ferlay J, Ward E, Forman D. Global cancer statistics. CA Cancer J. Clin. 2011; 61(2): 69–90. The lack of a standard protocol for DCE-MRI often leads to 2. Folkman J. Tumor angiogenesis: therapeutic implications. New Engl. inconsistent results. In this study, we performed a 3 min DCE-MRI J. Med. 1971; 285(21): 1182–1186. scan with a temporal resolution of 12 s. A higher temporal 3. Bergers G, Benjamin LE. Tumorigenesis and the angiogenic switch. resolution (preferred <5 s interval) was recommended by some Nat. Rev. Cancer 2003; 3(6): 401–410. 4. Hanahan D, Weinberg RA. The hallmarks of cancer. Cell 2000; 100(1): of the previous studies (17,19,48,49) to obtain more accurate AIFs 57–70. and more precise perfusion/permeability measurements. How- 5. Folkman J, Watson K, Ingber D, Hanahan D. Induction of angiogene- ever, some researchers (13,15,18) used longer intervals (about sis during the transition from hyperplasia to neoplasia. Nature 1989; trans 339(6219): 58–61. 10 s) and found that the K and kep values obtained this way can be used to distinguish the PCa from the noncancerous tissue. 6. Russo G, Mischi M, Scheepens W, De la Rosette JJ, Wijkstra H. Angiogenesis in prostate cancer: onset, progression and imaging. Some former studies recommended that having an imaging BJU Int. 2012; 110(11 Pt C): E794–E808. period that is at least 5 min in length would be better to capture 7. Wilson NM, Masoud AM, Barsoum HB, Refaat MM, Moustafa MI, the washout of contrast, but Chen et al. (19) found that a 2 min Kamal TA. Correlation of power Doppler with microvessel density in acquisition protocol could potentially avoid rectal motion (50) assessing prostate needle biopsy. Clin. Radiol. 2004; 59(10): 946–950. 8. Brawer MK, Deering RE, Brown M, Preston SD, Bigler SA. Predictors of and was still adequate to differentiate between cancerous and pathologic stage in prostatic carcinoma. The role of neovascularity. normal tissues. Our study demonstrated that signal intensity in Cancer 1994; 73(3): 678–687. most of the PCa nodules begins to drop before 3 min. Therefore, 9. Ren J, Huan Y, Wang H, Chang YJ, Zhao HT, Ge YL, Liu Y, Yang Y. the temporal resolution and the acquisition time of a DCE-MRI Dynamic contrast-enhanced MRI of benign prostatic hyperplasia scan need to be chosen judiciously. and prostatic carcinoma: correlation with angiogenesis. Clin. Radiol. 2008; 63(2): 153–159. 10. Isebaert S, De Keyzer F, Haustermans K, Lerut E, Roskams T, Roebben Study limitations I, Van Poppel H, Joniau S, Oyen R. Evaluation of semi-quantitative dynamic contrast-enhanced MRI parameters for prostate cancer in Several limitations of this study need to be mentioned. First, the correlation to whole-mount histopathology. Eur. J. Radiol. 2012; single-slice PASL acquisition would compromise the evaluation 81(3): e217–e222. of the overall characteristics of the PCa foci. An appropriate 11. Franiel T, Ludemann L, Rudolph B, Rehbein H, Staack A, Taupitz M, Prochnow D, Beyersdorff D. Evaluation of normal prostate tissue, multi-slice ASL sequence is currently under development. chronic prostatitis, and prostate cancer by quantitative perfusion Second, the prostate tissue properties were determined based analysis using a dynamic contrast-enhanced inversion-prepared dual- on the results of a TRUS-guided systemic extended needle contrast gradient echo sequence. Invest. Radiol. 2008; 43(7): 481–487. biopsy rather than the whole-mount prostatectomy specimens. 12. Franiel T, Ludemann L, Rudolph B, Rehbein H, Stephan C, Taupitz M, Beyersdorff D. Prostate MR imaging: tissue characterization with The systemic extended needle biopsy, which was used in this fl fi pharmacokinetic volume and blood ow parameters and correlation study as well, has been proven to signi cantly improve the with histologic parameters. Radiology 2009; 252(1): 101–108. ability of detection and localization of cancer in prostate (51), 13. Langer DL, van der Kwast TH, Evans AJ, Plotkin A, Trachtenberg J, and many previous MR studies based on the TRUS biopsy Wilson BC, Haider MA. Prostate tissue composition and MR measure- trans (9,14,52,53) were able to provide relatively reliable results. Last, ments: investigating the relationships between ADC, T2, K , ve, and – the sample size in this study is small. The ASL technique needs corresponding histologic features. Radiology 2010; 255(2): 485 494. 14. Kozlowski P, Chang SD, Jones EC, Berean KW, Chen H, Goldenberg to be optimized and evaluated on a larger population before it SL. Combined diffusion-weighted and dynamic contrast-enhanced can be used as a clinical sequence for prostate examination. MRI for prostate cancer diagnosis – correlation with biopsy and In summary, this study demonstrates that a PASL sequence histopathology. J. Magn. Reson. Imaging 2006; 24(1): 108–113. with appropriately chosen parameters allows the extraction of 15. Oto A, Kayhan A, Jiang Y, Tretiakova M, Yang C, Antic T, Dahi F, BF information in prostate, and may be an alternative for DCE-MRI. Shalhav AL, Karczmar G, Stadler WM. Prostate cancer: differentiation of central gland cancer from benign prostatic hyperplasia by using Noninvasive, noncontrast ASL MRI is therefore a promising method diffusion-weighted and dynamic contrast-enhanced MR imaging. Ra- in PCa evaluation, staging, treatment determination and therapeutic diology 2010; 257(3): 715–723. effect monitoring. 16. Buckley DL, Roberts C, Parker GJ, Logue JP, Hutchinson CE. Prostate cancer: evaluation of vascular characteristics with dynamic contrast-enhanced T1-weighted MR imaging – initial experience. Acknowledgements Radiology 2004; 233(3): 709–715. 17. Ocak I, Bernardo M, Metzger G, Barrett T, Pinto P, Albert PS, Choyke Presented in part at the 20th Annual Meeting ISMRM, PL. Dynamic contrast-enhanced MRI of prostate cancer at 3 T: a study Melbourne, Australia, 2012. of pharmacokinetic parameters. Am. J. Roentgenol. 2007; 189(4): 849. 18. Yaligar J, Thakur SB, Bokacheva L, Carlin S, Thaler HT, Rizwan A, Lupu The authors would like to thank Yudong Zhang for his ME, Wang Y, Matei CC, Zakian KL, Koutcher JA. Lactate MRSI and DCE 824 technical assistance in the T1 relaxation measurement and Sha MRI as surrogate markers of prostate tumor aggressiveness. NMR Li, Yanbin Dou, Ran Wan, Xiufeng Yang, Baoyu Zhu, Jingjing Biomed. 2012; 25(1): 113–122.

19. Chen YJ, Chu WC, Pu YS, Chueh SC, Shun CT, Tseng WY. Washout gradient using a B-spline grid [C]//SPIE Medical Imaging. International Society in dynamic contrast-enhanced MRI is associated with tumor aggressive- for Optics and Photonics. 2011; 796220-796220-8. ness of prostate cancer. J. Magn. Reson. Imaging 2012; 36(4): 912–919. 39. Tofts PS, Brix G, Buckley DL, Evelhoch JL, Henderson E, Knopp MV, 20. Turkbey B, Kobayashi H, Ogawa M, Bernardo M, Choyke PL. Imaging Larsson HB, Lee TY, Mayr NA, Parker GJ, Port RE, Taylor J, Weisskoff of tumor angiogenesis: functional or targeted? Am. J. Roentgenol. RM. Estimating kinetic parameters from dynamic contrast- 2009; 193(2): 304–313. enhanced T1-weighted MRI of a diffusable tracer: standardized quan- 21. Cyran CC, von Einem JC, Paprottka PM, Schwarz B, Ingrisch M, tities and symbols. J. Magn. Reson. Imaging 1999; 10(3): 223–232. Dietrich O, Hinkel R, Bruns CJ, Clevert DA, Eschbach R, Reiser MF, 40. Parker GJ, Roberts C, Macdonald A, Buonaccorsi GA, Cheung S, Wintersperger BJ, Nikolaou K. Dynamic contrast-enhanced com- Buckley DL, Jackson A, Watson Y, Davies K, Jayson GC. Experimen- puted tomography imaging biomarkers correlated with immunohis- tally-derived functional form for a population-averaged high- tochemistry for monitoring the effects of sorafenib on experimental temporal-resolution arterial input function for dynamic contrast- prostate carcinomas. Invest. Radiol. 2012; 47(1): 49–57. enhanced MRI. Magn. Reson. Med. 2006; 56(5): 993–1000. 22. Heinz-Peer G, Neruda A, Watschinger B, Vychytil A, Geusau A, 41. Fritz-Hansen T, Rostrup E, Larsson HB, Sondergaard L, Ring P, Haumer M, Weber M. Prevalence of NSF following intravenous Henriksen O. Measurement of the arterial concentration of Gd-DTPA gadolinium-contrast media administration in dialysis patients with using MRI: a step toward quantitative perfusion imaging. Magn. Reson. endstage renal disease. Eur. J. Radiol. 2010; 76(1): 129–134. Med. 1996; 36(2): 225–231. 23. Rydahl C, Thomsen HS, Marckmann P. High prevalence of 42. Barentsz JO, Richenberg J, Clements R, Choyke P, Verma S, Villeirs G, nephrogenic systemic fibrosis in chronic renal failure patients Rouviere O, Logager V, Futterer JJ. ESUR prostate MR guidelines exposed to gadodiamide, a gadolinium-containing magnetic reso- 2012. Eur. Radiol. 2012; 22(4): 746–757. nance contrast agent. Invest. Radiol. 2008; 43(2): 141–144. 43. Inaba T. Quantitative measurements of prostatic blood flow and 24. Hope TA, Herfkens RJ, Denianke KS, LeBoit PE, Hung YY, Weil E. blood volume by positron emission tomography. J. Urol. 1992; 148 Nephrogenic systemic fibrosis in patients with chronic kidney dis- (5): 1457–1460. ease who received gadopentetate dimeglumine. Invest. Radiol. 44. Ludemann L, Prochnow D, Rohlfing T, Franiel T, Warmuth C, Taupitz 2009; 44(3): 135–139. M, Rehbein H, Beyersdorff D. Simultaneous quantification of perfu- 25. Sadowski EA, Bennett LK, Chan MR, Wentland AL, Garrett AL, Garrett sion and permeability in the prostate using dynamic contrast- RW, Djamali A. Nephrogenic systemic fibrosis: risk factors and enhanced magnetic resonance imaging with an inversion-prepared incidence estimation. Radiology 2007; 243(1): 148–157. dual-contrast sequence. Ann. Biomed. Eng. 2009; 37(4): 749–762. 26. Petersen ET, Zimine I, Ho YC, Golay X. Non-invasive measurement of 45. Qiu M, Paul Maguire R, Arora J, Planeta-Wilson B, Weinzimmer D, perfusion: a critical review of arterial spin labelling techniques. Br. J. Wang J, Wang Y, Kim H, Rajeevan N, Huang Y, Carson RE, Constable Radiol. 2006; 79(944): 688–701. RT. Arterial transit time effects in pulsed arterial spin labeling CBF 27. Williams DS, Detre JA, Leigh JS, Koretsky AP. Magnetic resonance mapping: insight from a PET and MR study in normal human imaging of perfusion using spin inversion of arterial water. Proc. Natl. subjects. Magn. Reson. Med. 2010; 63(2): 374–384. Acad. Sci. U. S. A. 1992; 89(1): 212–216. 46. Warmuth C, Gunther M, Zimmer C. Quantification of blood flow in 28. Wong EC, Buxton RB, Frank LR. Implementation of quantitative brain tumors: comparison of arterial spin labeling and dynamic perfusion imaging techniques for functional brain mapping using susceptibility-weighted contrast-enhanced MR imaging. Radiology pulsed arterial spin labeling. NMR Biomed. 1997; 10(4–5): 237–249. 2003; 228(2): 523––532. 29. Wang J, Zhang Y, Yang X, Wang X, Zhang J, Fang J, Jiang X. Hemo- 47. Wu WC, Su MY, Chang CC, Tseng WY, Liu KL. Renal perfusion 3-T MR dynamic effects of furosemide on renal perfusion as evaluated by imaging: a comparative study of arterial spin labeling and dynamic ASL-MRI. Acad. Radiol. 2012; 19(10): 1194–1200. contrast-enhanced techniques. Radiology 2011; 261(3): 845–853. 30. Li X, Metzger GJ. Feasibility of measuring prostate perfusion with 48. Jackson AS, Reinsberg SA, Sohaib SA, Charles-Edwards EM, Jhavar S, arterial spin labeling. NMR Biomed. 2013; 26(1): 51–57. Christmas TJ, Thompson AC, Bailey MJ, Corbishley CM, Fisher C, 31. Clegg EJ. The arterial supply of the human prostate and seminal Leach MO, Dearnaley DP. Dynamic contrast-enhanced MRI for pros- vesicles. J. Anat. 1955; 89(2): 209–216. tate cancer localization. Br. J. Radiol. 2009; 82(974): 148–156. 32. Luh WM, Wong EC, Bandettini PA, Hyde JS. QUIPSS II with thin-slice 49. Franiel T, Hamm B, Hricak H. Dynamic contrast-enhanced magnetic TI1 periodic saturation: a method for improving accuracy of quanti- resonance imaging and pharmacokinetic models in prostate cancer. tative perfusion imaging using pulsed arterial spin labeling. Magn. Eur. Radiol. 2011; 21(3): 616–626. Reson. Med. 1999; 41(6): 1246–1254. 50. Padhani AR, Khoo VS, Suckling J, Husband JE, Leach MO, Dearnaley 33. Deibler AR, Pollock JM, Kraft RA, Tan H, Burdette JH, Maldjian JA. DP. Evaluating the effect of rectal distension and rectal movement Arterial spin-labeling in routine clinical practice, part 1: technique on prostate gland position using cine MRI. Int. J. Radiat. Oncol. Biol. and artifacts. Am. J. Neuroradiol. 2008; 29(7): 1228–1234. Phys. 1999; 44(3): 525–533. 34. Hendrikse J, van Osch MJ, Rutgers DR, Bakker CJ, Kappelle LJ, Golay X, 51. San Francisco IF, DeWolf WC, Rosen S, Upton M, Olumi AF. Extended van der Grond J. Internal carotid artery occlusion assessed at pulsed prostate needle biopsy improves concordance of Gleason grading arterial spin-labeling perfusion MR imaging at multiple delay times. between prostate needle biopsy and radical prostatectomy. J. Urol. Radiology 2004; 233(3): 899–904. 2003; 169(1): 136–140. 35. Deoni SC, Rutt BK, Peters TM. Rapid combined T1 and T2 mapping 52. Hara N, Okuizumi M, Koike H, Kawaguchi M, Bilim V. Dynamic using gradient recalled acquisition in the steady state. Magn. Reson. contrast-enhanced magnetic resonance imaging (DCE-MRI) is a Med. 2003; 49(3): 515–526. useful modality for the precise detection and staging of early 36. Roberts DA, Detre JA, Bolinger L, Insko EK, Lenkinski RE, Pentecost prostate cancer. Prostate 2005; 62(2): 140–147. MJ, Leigh JS, Jr. Renal perfusion in humans: MR imaging with spin 53. Ito H, Kamoi K, Yokoyama K, Yamada K, Nishimura T. Visualization of tagging of arterial water. Radiology 1995; 196(1): 281–286. prostate cancer using dynamic contrast-enhanced MRI: comparison 37. Kim SG. Quantification of relative cerebral blood flow change by flow- with transrectal power Doppler ultrasound. Br. J. Radiol. 2003; 76(909): sensitive alternating inversion recovery (FAIR) technique: application to 617–624. functional mapping. Magn. Reson. Med. 1995; 34(3): 293–301. 54. Edge S, Byrd D, Compton C, Fritz A, Greene F, Trotti A. American Joint 38. Klein A, Kroon DJ, Hoogeveen Y, Kool L. J S, Renema WKJ, Slump CH. Committee on Cancer, American Cancer Society: AJCC Cancer Staging Multimodal image registration by edge attraction and regularization Manual. Springer-Verlag: New York, NY, 2010; 41: 1–4. 825