Keywords Prostate, cancer, magnetic resonance imaging (MRI), diffusion-weighted imaging (DWI), Gleason score

Original Article The value of ADC, T2 signal intensity, and a combination of both parameters to assess Gleason score and primary Gleason grades in patients with known prostate cancer Acta Radiologica 2016, Vol. 57(1) 107 114! The Foundation Acta Radiologica 2014 Reprints and permissions: sagepub.co.uk/journalspermissions.nav DOI: 10.1177/0284185114561915 acr.sagepub.com Johannes Nowak 1,2, Uwe Malzahn 3, Alexander DJ Baur 2, Uta Reichelt 4, Tobias Franiel 2,5, Bernd Hamm 2 and Tahir Durmus 2 Abstract Background: The ability to non-invasively analyze tumor aggressiveness is an important predictor for individual treatment stratification and patient outcome in prostate cancer (PCA). Purpose: To evaluate: (i) whether apparent diffusion coefficient (ADC), the T2 signal intensity (SI), and a combination of both parameters allow for an improved discrimination of Gleason Score (GS) 7 (intermediate and high risk) and GS <7 (low risk) in PCA; and (ii) whether ADC may distinguish between 3 þ 4 and 4 þ 3 PCA (primary Gleason grades [pgg]). Material and Methods: Prostatectomy specimens of 66 patients (mean age, 63 5.6 years; 104 PCA foci) with a preceding multiparametric 1.5 T endorectal coil magnetic resonance imaging (MRI) were included. ADC (b values ¼ 0, 100, 400, 800 s/mm 2 ), standardized T2 (T2s), and the ADC/T2s ratio were tested for correlation with GS applying multivariate analysis. ADC cutoff values were calculated for prediction of GS and pgg, and logarithm of the odds (LOGIT) was used to express the probability for GS and pgg. Diagnostic accuracy was assessed by ROC analysis. Results: We found an almost linear negative relationship of ADC for GS 7 (P ¼ 0.002). The effect of ADC for GS 7 (adjusted odds ratio ¼ 0.995) was almost identical for peripheral and transition zone PCA (P ¼ 0.013 and P < 0.001, respectively). ADC showed an AUC of 78.9% for discrimination between GS <7 and GS 7. An ADC cutoff of <1.005 10 3 mm 2 /s indicated a GS 7 (90.5% sensitivity, 62.5% specificity). Within the group of GS ¼ 7PCA,an ADC > 0.762 10 3 mm 2 /s indicated a pgg of 3 (AUC ¼ 69.6%). Conclusion: T2s and the ADC/T2s ratio do not provide additional information regarding prediction of GS. ADC values have a good discriminatory power to distinguish tumors with GS 7 fromgs<7 and to predict pgg in GS ¼ 7PCA. Keywords Prostate, cancer, magnetic resonance imaging (MRI), diffusion-weighted imaging (DWI), Gleason score Date received: 24 May 2014; accepted: 8 November 2014 Introduction Prostate cancer (PCA) is the most frequent cancer in men in the Western world (1). Biologic aggressiveness of PCA varies greatly and is a key predictor of outcome (2). Although intensive research efforts have been made, and suspicious lesions of the prostate can now be identified at early stages, we still lack the ability to predict tumor aggressiveness non-invasively (3,4). Thus, radical prostatectomy and radiotherapy is the only definite treatment option for the majority of 1 Department of Radiology, Universitätsklinikum Würzburg, Germany 2 Department of Radiology, Charité Universitätsmedizin Berlin, Germany 3 Institute for Clinical Epidemiology and Biometry, Universität Würzburg, Germany 4 Institute for Pathology, Vivantes Klinikum am Urban Dieffenbachstrasse, Berlin, Germany 5 Department of Radiology, Universitätsklinikum Jena, Germany Corresponding author: Tahir Durmus, Department of Radiology, Charité Universitätsmedizin Berlin, Campus Mitte, Charitéplatz 1, 10117 Berlin, Germany. Email: tahir.durmus@charite.de

108 Acta Radiologica 57(1) patients to date, although a benefit for all patients is currently under debate (5). Histopathologically, the Gleason grading correlates with patient outcome, with higher Gleason scores (GS) indicating more aggressive PCA (6,7). However, 28 47% of transrectal ultrasound (TRUS)-guided prostate biopsies do not detect a present PCA (false negatives), and up to one-third of patients were upgraded to higher GS from biopsy to prostatectomy specimen (4,8,9). Multiparametric magnetic resonance imaging (MRI) of the prostate has emerged as a promising diagnostic tool for both detecting and grading PCA (10,11). However, the majority of published studies analyze the diagnostic value of MRI in detecting PCA only (10,12,13). It has been shown that ADC values were negatively correlated with the postsurgical GS in PCA patients. The ADC was furthermore sufficient to predict the presence of low-grade and high-grade PCA in the peripheral zone (PZ) (14 16). Regarding T2- weighted (T2W) images, it has been demonstrated that PCA have lower T2 signal intensity (SI) than non-neoplastic tissue (11). In a pilot study, Wang et al. found lower T2 SI values at 1.5 T to be associated with higher GS, suggesting that T2W imaging might as well enable non-invasive assessment of PCA aggressiveness (17). Given that both ADC and T2W SI thus correlate independently with the GS in PCA, their combination might improve PCA grading with MRI. The first aim of this study was to test whether ADC, standardized T2W imaging (T2s) and the combination of ADC and T2W imaging (ADC/T2s ratio) allow a better discrimination between low-risk (GS < 7) and intermediate or high-risk (GS 7) tumors, using histopathology after prostatectomy as reference standard. Furthermore, GS ¼ 7 tumors comprise a very heterogenous group of PCA. While a primary Gleason grade (pgg) of 3 might be interpreted as intermediate risk, a pgg of 4 is associated with a higher clinical risk (18). We therefore intended to calculate ADC cutoff values in order to non-invasively differentiate 3 þ 4 from 4 þ 3 PCA. Material and Methods Patient characteristics This study was approved by the institutional ethical review board, and written consent was obtained from every participant. A collective of 69 consecutive patients with biopsy proven PCA underwent endorectal multiparametric MRI. MRI was performed at least 4 weeks after TRUS-guided biopsy. Patients had a mean age of 63 5.6 years (range, 48 73 years) and a mean PSA of 17.5 ng/ml (range, 1.7 272.4 ng/ml). Only patients with whole-mount, step-sectioned pathological evaluation of prostatectomy specimens were included in this retrospective analysis. Thus, three patients (one patient with metastatic disease underwent only lymphadenectomy and systemic therapy, two patients decided for primary radiation therapy) were excluded. The final study population consisted of 66 patients. MRI protocol MRI was performed on a 1.5 T MR scanner (Siemens Avanto, Erlangen, Germany) using a combined endorectal-pelvic phased array coil (Medrad, Pittsburgh, PA, USA). The MRI standard protocol included T1- weighted (T1W) images of the pelvis and prostate and multiplanar T2W images of the prostate. All MR sequences were performed with a 256 256 image matrix (section thickness, 3.0 mm; intersection gap, 0.6 mm; 100% phase oversampling). After axial, coronal, and sagittal localizer sequences were obtained, an angulated axial T2W turbo spin echo (TSE) sequence (TR/TE, 4850/85 ms; echo train length, 15; averages, three; field of view [FOV], 160 160 mm), an angulated coronal T2W TSE sequence (TR/TE, 4000/95 ms; echo train length, 13; averages, three; FOV, 200 200 mm), and an angulated axial T1W TSE sequence (TR/TE 691/12 ms; echo train length, three; averages, two; FOV, 160 160 mm) were acquired. Diffusion-weighted imaging (DWI) was acquired with the same slice thickness, gap, and axial angulation as the T2W TSE using gradient pulses applied along three orthogonal directions (x-, y-, and z-axes; TR/TE, 3000/69; number of signals acquired, eight; FOV, 220 220 mm; image matrix, 128 128; section thickness, 3.6 mm) with four b values of 0, 100, 400, and 800 s/mm 2. ADC maps were then automatically generated on the basis of a voxelwise calculation. Histopathological evaluation and MR correlation After surgical resection, whole prostate specimens were marked with ink (anterior, posterior, left, right) and then step-sectioned from apex to base in five to eight axial slices. After preparation for pathologic analysis (embedding of microslices in paraffin and staining with hematoxylin and eosin, segmentation into quadrants), cancer foci were identified and assigned a GS according to the modified Gleason grading system by a fellow pathologist experienced in genitourinary histopathology (UR) (6). Cancer foci were then graphically transferred to a standardized map. Only tumor foci larger than 5 mm were included into the analysis. If tumor foci touched each other in the drawings, they were regarded as one tumor focus. The pathologist and a radiologist (TD) jointly matched cancer foci in

Nowak et al. 109 pathologic maps with the corresponding T2W images and ADC maps. When necessary, anatomic landmarks, such as prostate zones, prostatic urethra, ejaculatory ducts, urinary bladder, and seminal vesicle tissue in superior imaging sections for guidance in order to find the most closely corresponding MRI for PCA foci were used. Image analysis and data acquisition Based on the pathological maps, elliptical region of interests (ROI) were placed on cancerous regions for SI measurements in T2W imaging and in the corresponding ADC maps by two radiologists in consensus (JN and TD, with 2 and 5 years of experience, respectively). ROIs were placed into the visually darkest tumor area, in order to avoid tumor edges and areas possibly containing hemorrhage. In patients with multifocal disease, each PCA lesion was measured separately and included in the study. Cancer size was measured in T2W imaging with the largest apparent tumor diameter. Furthermore, ROIs were placed in histologically confirmed PCA-free PZ in T2W imaging and ADC maps as controls. For normalization to an internal control as suggested by Wang et al., an additional ROI was placed in the internal obturator muscle. T2W imaging SI values of tumors in PZ and TZ were then divided by obturator muscle SI values, resulting in standardized T2 SI values (T2s) (17). Statistical analysis Ninety-five of the 104 cancers (91.3%) were GS of 6 or 7. Therefore the score variable was dichotomized and a binary target variable (GS < 7 versus GS 7) was obtained. Accounting for multiple cancer foci, measurements on different foci for the same patient were treated as correlated observations within the overall dataset. Thus a binary logistic modelling with GEE (generalized estimation equations) and LOGIT as link function (with LOGIT referring to natural logarithm of odds; LOGIT ¼ ln (p/1 p) with p ¼ probability for GS 7) under the mixed linear model framework was performed. We adjusted for PSA values and tumor diameter as covariates to correct for confounding factors. P values (P < 0.050 were regarded as statistically significant) refer to the Wald test for GEE regression parameters. Receiver operating characteristic (ROC) analysis was performed to assess diagnostic accuracy and calculate ADC cutoff values for the target variables (GS < 7, GS 7). For this analysis only data corresponding to the tumor lesion with the highest GS were used. Cutoff values for ADC were calculated in order to distinguish 3 þ 4 from 4 þ 3 cancers in GS ¼ 7 PCA. Sensitivities and specificities were calculated. Statistical analysis was performed using IBM SPSS for Windows (Version 20.0; SPSS Inc., Chicago, IL, USA). Results A total of 81 cancers were assigned to the peripheral zone (PZ) of the prostate, whereas 23 cancers were localized in the transition zone (TZ). In total 104 cancers in 66 patients were analyzed (Table 1). With respect to histopathological grading, we found one tumor with a GS of 3 þ 2 ¼ 5, 38 tumors with 3 þ 3 ¼ 6, 40 tumors of 3 þ 4 ¼ 7 and 17 tumors with 4 þ 3 ¼ 7, five tumors with 4 þ 4 ¼ 8, and three tumors with 4 þ 5/5 þ 4 ¼ 9. Table 2 presents basic descriptive statistics for the MR signal intensities (ADC, T2s, ADC/T2s ratio; cancer-free T2s and ADC as controls) of all cancers stratified by GS. Representative T2 TSE images of PCA with their corresponding ADC maps are shown in Fig. 1. We found an almost linear relationship and highly significant (P ¼ 0.002) influence of ADC values on LOGIT (GS 7). This corresponds to a reduction of the odds by a factor of 0.995 per one unit increase in ADC SI (10 3 mm 2 /s). Lower ADC values were thus strongly associated with higher tumor aggressiveness in this study (Tables 2 and 3). ROC analysis of 66 tumors for ADC as marker for GS 7 showed an AUC of 78.9%, indicating that ADC values are appropriate for discrimination between GS < 7 and GS 7in our patient population (Fig. 2a). The effect (adjusted odds ratio [OR] ¼ 0.995) of ADC on the LOGIT of GS 7 was almost identical for PZ and TZ cancers, with P values of 0.013 and <0.001, respectively (Table 4). Thus, lower ADC values significantly correlate with higher GS in both PZ and TZ cancers. Analyzing our raw data, there was no linear relationship between T2s and LOGIT (GS 7). We Table 1. Distribution of tumors. Tumors/patient Total tumors (n) Frequency % PZ TZ 1 39 39 37,5 34 5 2 20 40 38,5 27 13 3 5 15 14,4 10 5 4 1 4 3,8 4 0 6 1 6 5,8 6 0 Total 66 104 100 81 23 Distribution of tumors demonstrating a single tumor focus in 39 patients and multifocal disease in 27 patients. PZ, peripheral zone; TZ, transition zone.

110 Acta Radiologica 57(1) Table 2. ADC and T2s values in PCA. Gleason 6 Gleason 3 þ 4 Gleason 4 þ 3 Gleason 7 Gleason 8 ADC TZ þ PZ 1.07 0.31 0.88 0.17 0.77 0.16 0.80 0.18 0.61 0.13 ADC TZ 0.78 0.13 0.74 0.19 0.63 0.03 0.68 0.15 0.58* ADC PZ 1.15 0.30 0.92 0.14 0.81 0.15 0.83 0.18 0.61 0.15 ADC PZ Ctr 1.70 0.32 1.75 0.17 1.80 0.13 1.74 0.26 1.60 0.56 T2s TZ þ PZ 4.03 1.01 3.42 0.60 3.89 0.60 3.60 0.66 3.61 0.87 T2s TZ 3.20 1.08 3.09 0.62 3.73 0.51 3.35 0.60 3.48* T2s PZ 4.22 0.92 3.50 0.57 3.94 0.64 3.66 0.67 3.63 0.95 T2s PZ Ctr 7.51 2.55 7.14 1.15 7.42 1.69 7.32 1.43 7.73 1.87 ADC/T2s TZ þ PZ 0.28 0.09 0.26 0.04 0.20 0.04 0.23 0.06 0.17 0.04 ADC/T2s TZ 0.26 0.10 0.26 0.04 0.17 0.03 0.21 0.05 0.17* ADC/T2s PZ 0.29 0.09 0.25 0.05 0.21 0.05 0.23 0.06 0.18 0.05 ADC/T2s PZ Ctr 0.25 0.08 0.25 0.05 0.25 0.06 0.24 0.06 0.20 0.07 ADC values [10 3 mm 2 /s] and standardized T2 SI (T2s) given as means standard error. Prostate tumors are listed in GS category, controls are measured in cancer-free PZ. *n ¼ 1. Ctr, control; PZ, peripheral zone; SI, signal intensity; TZ, transition zone. (a) (c) (e) (g) (b) (d) (f) (h) Fig. 1. Axial T2 TSE with corresponding ADC Maps. (a, b) Patient with an anterior prostate cancer lesion (arrows) in the PZ with GS of 3 þ 3 ¼ 6 and T2s of 2.36 and an ADC value of 0.95. (c, d) Patient with localized cancer focus in the left apex with a GS 3 þ 4 ¼ 7, T2s of 2.44 and an ADC of 0.89. (e, f) Patient with a laterally localized tumor in the PZ with a GS 4 þ 3 ¼ 7 and a T2s of 3.92 and an ADC of 0.74. (g, h) Large right-sided PCA focus with bulging and histologically confirmed extracapsular extension of GS 5 þ 4 ¼ 9; T2s of 4.64 and an ADC of 0.55. subsequently categorized T2s values into quartiles. A borderline effect was seen for overall T2s SI and prediction of GS 7 (P ¼ 0.050), with very high T2s values (fourth quartile) indicating a significantly lower probability of cancers with GS 7 (P ¼ 0.009), compared to low T2s values (first quartile; generalized linear modeling, adjusted for categorized PSA and cancer diameter variables; Table 3). T2s values in PZ cancers (n ¼ 81) show a similar effect as compared to the overall calculation, with an even stronger impact of high T2s values (fourth quartile) on the probability for lower tumor aggressiveness. There was no monotone

Nowak et al. 111 relationship between T2s with LOGIT (GS 7) in our subsample of TZ cancers, not allowing for further statistical analysis. In order to evaluate independency and potential additive effects of ADC and T2s SI with PCA grading, a ratio of these parameters was tested. However, the ADC/T2s ratio only confirmed the result that was seen for ADC values alone, but no additional information could be identified regarding correlation with GS (P ¼ 0.155). ADC cutoff values were calculated for different criteria: (i) for maximum values of both sensitivity (90.5%) and specificity (62.5%), ADC values Table 3. Results of multivariate analysis for ADC, T2s, and ADC/T2s. P value Adjusted OR 95% CI ADC 0.002 0.995 (0.992, 0.998) T2s 0.050 T2s fourth quartile 0.009 0.129 (0.028, 0.606) T2s third quartile 0.508 0.636 (0.166, 2.430) T2s second quartile 0.634 0.698 (0.159, 3.061) T2s first quartile 1 ADC/T2s 0.155 0.994 (0.985, 1.002) Multivariate analysis (separate models for ADC, T2s, and ADC/T2s) shows the relation of tumor ADC and T2s with LOGIT (GS 7) in all lesions. Adjusted for covariates (tumor diameter, PSA values). CI, confidence interval. lower than 1.005 10 3 mm 2 /s indicated a GS 7; (ii) for high sensitivity (95.2%) and specificity of 50%, the cutoff ADC value for GS 7 was 1.052 10 3 mm 2 /s (Table 5). In our analysis an ADC of >0.762 10 3 mm 2 /s indicated rather a 3 þ 4 type Gleason grade with an AUC of 69.6%, corresponding Table 4. Results of multivariate analysis for ADC and T2s in lesions of the PZ and TZ. P value Adjusted OR 95% CI PZ tumors ADC 0.013 0.995 (0.990, 0.999) T2s 0.026 T2s fourth quartile 0.009 0.103 (0.018, 0.573) T2s third quartile 0.935 0.933 (0.175, 4.966) T2s second quartile 0.384 0.447 (0.739, 2.744) T2s first quartile 1 TZ tumors ADC <0.001 0.995 (0.992, 0.998) T2s * Multivariate analysis (separate models for ADC, T2s) shows the relation of tumor ADC and T2s with LOGIT (GS 7) in tumors of the PZ and TZ. Adjusted for covariates (tumor diameter, PSA values). *No monotone relationship between T2s and LOGIT (GS 7) in our subsample of TZ cancers, not allowing for further statistical analysis. PZ, peripheral zone; TZ, transition zone; CI, confidence interval. Fig. 2. (a) ROC analysis for ADC and GS (AUC ¼ 0.79). (b) Subgroup analysis of tumors with GS ¼ 7. ADC values to distinguish 3 þ 4 and 4 þ 3 Gleason grades (AUC ¼ 0.70).

112 Acta Radiologica 57(1) Table 5. ADC cutoff values for differentiation of Gleason grades. to a sensitivity and specificity of 77.5% and 64.7%, respectively (Fig. 2b). Discussion Sensitivity Specificity Cutoff for GS 7 ADC 1.005 0.905* 0.625 ADC 1.052 0.952 y 0.5 Cutoff for Gleason 3 þ 4 ADC > 0.762 0.775* 0.647 ADC cutoff values for differentiation of Gleason grades in prostate tumors. *Maximum sensitivity and specificity. ymaximum sensitivity. DWI and ADC are rapidly gaining importance as reliable non-invasive markers for assessing the treatment response in a large variety of cancers (19). DWI is sensitive to the Brownian motion (¼diffusion) of free water molecules. Studies have been published that propose a relationship between DWI (ADC) and PCA aggressiveness (GS) (11,14,20). Lower ADC values in our study were significantly associated with intermediate or high risk PCA (GS 7), which is basically consistent with earlier research work on this topic (14,15,21). T2W imaging can be seen as water imaging, with PCA showing a high cell density with comparatively low T2 SI due to a decreased amount of free water (22). Wang et al. showed a negative correlation of standardized T2 values with GS (17). In our study, we could not generally confirm this finding; only very high T2s values (fourth quartile of our collective) were indicative of PCA with lower GS, whereas no significant relationship was found for the remaining T2s values. This finding might in part explain the finding of Doo et al., that addition of ADC to T2W imaging improves the accuracy of detecting intermediate or high risk PCA (GS 7), but not of low risk PCA (12). We also tested whether the combination of ADC with T2W imaging provides additional information regarding discrimination of low-grade from combined intermediate and high-grade PCA, and our results show that a ratio of the two parameters has no additive effect regarding prediction of GS, compared to ADC values alone. More importantly, we show that the ADC value might be able to separate PCA with a GS of 7 into the subgroups of 3 þ 4 ¼ 7 and 4 þ 3 ¼ 7 cancers, with an ADC > 0.762 10 3 mm 2 /s indicating presence of 3 þ 4 Gleason grades (77.5% sensitivity, 64.7% specificity). This is consistent with a recent study of Verma et al., who found that mean values of Gleason grade 3 þ 4 cancers were significantly different from those of the reference category of 4 þ 3 cancers (15). The finding that PCA with GS of 3 þ 3, 3 þ 4or4þ 3 might be distinguished by ADC values indicates a major role for MRI in re-evaluation of patients during active surveillance (3,4,18). This is of special interest since studies suggest that Gleason 3 þ 4 and 4 þ 3 PCA are heterogeneous in terms of cancer-specific mortality, reflecting the different biological aggressiveness of GS ¼ 7 cancers and thus a need for different treatment regimens (2,18). Some other publications used prostatectomy specimens as reference standard for histopathology when correlating ADC values with GS (14,15,20,21,23 25); this is important since GS determined by TRUS tend to underestimate tumor grades (4,26). Additionally, a number of studies were restricted to cancer foci in the PZ of the prostate (14,20,23,24). Taking into consideration that PCA is a multifocal disease with approximately 30% of PCA arising in the TZ, the inclusion of cancers of the central gland is essential for clinical practice (27). In two publications the number of TZ cancers was too low for detailed statistical analysis (21,25). Verma et al. included 47 patients with at least one cancer focus in the TZ (15). They found a negative correlation of ADC values with GS in the PZ, as did the other studies. To the best of our knowledge, only a few studies with histological correlation on prostatectomy (investigating the value of high b-values at 3 T) described the predictive value of ADC for PCA in the TZ, which is consistent with our results, obtained with a maximal b-value of 800 s/mm 2 (28). In the present study, we were able to demonstrate negative correlation of ADC with GS in both the PZ and TZ. This is a clinically very important finding and further supports the qualification of DWI as a criterion for patient stratification. Our study may be limited by its retrospective design. Since this study was not intended to investigate the primary detection rate of MRI but its grading ability, matching of imaging with prostatectomy specimen seemed to be best qualified for this aim. We identified cancerous lesions in MRI according to pathologic maps of patients with known PCA before therapy. Slice thickness of prostatectomy specimens was not identical to MRI, which might have led to inaccuracies in lesion matching. Lesions were matched on a subjective basis in consensus of the pathologist and radiologist to minimize these inaccuracies. However, a digitalized prostatectomy specimen preparation would have been another objective method to match pathology and imaging. A potential limitation of our study is that we did not perform correction for endorectal coil artifacts using analytic coil correction software or a phantom study in order to identify SI isosurfaces on T2W imaging. However, Wang et al. showed that correction

Nowak et al. 113 did not significantly affect the results of their study compared to uncorrected images (17). Another limitation of our study might be related to the distribution of GS in our subgroup of patients, with the majority of cancers (91.3%) being of low and intermediate risk (with GS of 6 or 7). To address this issue, we dichotomized the score variable, leading to a binary target variable (GS < 7 representing low-risk and GS 7 representing intermediate and high-risk cancers). Furthermore, we included both PZ and TZ cancers and used a multivariate statistical analysis acknowledging multifocal tumors as dependent variables. Finally, ADC values may vary between institutions due to a number of reasons such as the used MRI scanner, field strength, and b-values. Further studies are needed to confirm our results and evaluate possible additive effects of absolute T2 relaxation times to ADC. In conclusion, our study indicates that ADC, but not T2W imaging as suggested by Wang et al, is an important and accurate parameter that correlates with GS in PCA of the PZ and TZ, and might furthermore help to differentiate 3 þ 4 and 4 þ 3 tumors in PCA with GS ¼ 7. Thus, the ADC value might be considered when classifying PCA patients into different treatment groups (according to low-, intermediate-, and high-risk cancers). Conflict of interest The study was performed at the Department of Radiology of the Charite Universita tsmedizin Berlin (CUB), Germany. The authors had full control of all primary data and all authors have participated sufficiently in the preparation of this work. The material in this paper represents exclusively our own work unless cited otherwise. Although not for this study explicitly, the Department of Radiology, CUB received grants from a variety of organizations as listed in the conflict disclosure form of the Department Director (Prof. Hamm). The other authors do not have a potential conflict of interest to disclose. The paper has not been published previously and is not currently under consideration for publication elsewhere. Funding This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors. References 1. Jemal A, Siegel R, Xu J, et al. Cancer statistics, 2010. CA Cancer J Clin 2010;60:277 300. 2. Wright JL, Salinas CA, Lin DW, et al. Prostate cancer specific mortality and Gleason 7 disease differences in prostate cancer outcomes between cases with Gleason 4 þ 3 and Gleason 3 þ 4 tumors in a population based cohort. J Urol 2009;182:2702 2707. 3. Bul M, Zhu X, Rannikko A, et al. Radical prostatectomy for low-risk prostate cancer following initial active surveillance: results from a prospective observational study. Eur Urol 2012;62:195 200. 4. Shapiro RH, Johnstone PA. Risk of Gleason grade inaccuracies in prostate cancer patients eligible for active surveillance. Urology 2012;80:661 666. 5. Wilt TJ, Brawer MK, Jones KM, et al. Radical prostatectomy versus observation for localized prostate cancer. N Engl J Med 2012;367:203 213. 6. Epstein JI. An update of the Gleason grading system. J Urol 2010;183:433 440. 7. Gleason DF, Mellinger GT. Prediction of prognosis for prostatic adenocarcinoma by combined histological grading and clinical staging. J Urol 1974;111:58 64. 8. Bak JB, Landas SK, Haas GP. Characterization of prostate cancer missed by sextant biopsy. Clin Prostate Cancer 2003;2:115 118. 9. Fink KG, Hutarew G, Esterbauer B, et al. Evaluation of transition zone and lateral sextant biopsies for prostate cancer detection after initial sextant biopsy. Urology 2003;61:748 753. 10. Turkbey B, Pinto PA, Mani H, et al. Prostate cancer: value of multiparametric MR imaging at 3 T for detection histopathologic correlation. Radiology 2010;255: 89 99. 11. Tamada T, Sone T, Jo Y, 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;28:720 726. 12. Doo KW, Sung DJ, Park BJ, et al. Detectability of low and intermediate or high risk prostate cancer with combined T2-weighted and diffusion-weighted MRI. Eur Radiol 2012;22:1812 1819. 13. Tan CH, Wei W, Johnson V, et al. Diffusion-weighted MRI in the detection of prostate cancer: meta-analysis. Am J Roentgenol 2012;199:822 829. 14. Nagarajan R, Margolis D, Raman S, et al. Correlation of Gleason scores with diffusion-weighted imaging findings of prostate cancer. Adv Urol 2012;2012:374805. 15. Verma S, Rajesh A, Morales H, et al. Assessment of aggressiveness of prostate cancer: correlation of apparent diffusion coefficient with histologic grade after radical prostatectomy. Am J Roentgenol 2011;196:374 381. 16. desouza NM, Riches SF, Vanas NJ, et al. Diffusionweighted magnetic resonance imaging: a potential noninvasive marker of tumour aggressiveness in localized prostate cancer. Clin Radiol 2008;63:774 782. 17. Wang L, Mazaheri Y, Zhang J, et al. Assessment of biologic aggressiveness of prostate cancer: correlation of MR signal intensity with Gleason grade after radical prostatectomy. Radiology 2008;246:168 176. 18. Rasiah KK, Stricker PD, Haynes AM, et al. Prognostic significance of Gleason pattern in patients with Gleason score 7 prostate carcinoma. Cancer 2003;98:2560 2565. 19. Harry VN, Semple SI, Parkin DE, et al. Use of new imaging techniques to predict tumour response to therapy. Lancet Oncol 2010;11:92 102. 20. Bittencourt LK, Barentsz JO, de Miranda LC, et al. Prostate MRI: diffusion-weighted imaging at 1.5T correlates better with prostatectomy Gleason Grades than

114 Acta Radiologica 57(1) TRUS-guided biopsies in peripheral zone tumours. Eur Radiol 2012;22:468 475. 21. Vargas HA, Akin O, Franiel T, et al. Diffusion-weighted endorectal MR imaging at 3 T for prostate cancer: tumor detection and assessment of aggressiveness. Radiology 2011;259:775 784. 22. Gibbs P, Liney GP, Pickles MD, et al. Correlation of ADC and T2 measurements with cell density in prostate cancer at 3.0 Tesla. Invest Radiol 2009;44:572 576. 23. Hambrock T, Somford DM, Huisman HJ, et al. Relationship between apparent diffusion coefficients at 3.0-T MR imaging and Gleason grade in peripheral zone prostate cancer. Radiology 2011;259:453 461. 24. Itou Y, Nakanishi K, Narumi Y, et al. Clinical utility of apparent diffusion coefficient (ADC) values in patients with prostate cancer: can ADC values contribute to assess the aggressiveness of prostate cancer? J Magn Reson Imaging 2011;33:167 172. 25. Oto A, Yang C, Kayhan A, et al. Diffusion-weighted and dynamic contrast-enhanced MRI of prostate cancer: correlation of quantitative MR parameters with Gleason score and tumor angiogenesis. Am J Roentgenol 2011;197:1382 1390. 26. Epstein JI, Feng Z, Trock BJ, et al. Upgrading and downgrading of prostate cancer from biopsy to radical prostatectomy: incidence and predictive factors using the modified Gleason grading system and factoring in tertiary grades. Eur Urol 2012;61:1019 1024. 27. Bouye S, Potiron E, Puech P, et al. Transition zone and anterior stromal prostate cancers: zone of origin and intraprostatic patterns of spread at histopathology. Prostate 2009;69:105 113. 28. Kitajima K, Takahashi S, Ueno Y, et al. Do apparent diffusion coefficient (ADC) values obtained using high b- values with a 3-T MRI correlate better than a transrectal ultrasound (TRUS)-guided biopsy with true Gleason scores obtained from radical prostatectomy specimens for patients with prostate cancer? Eur J Radiol 2013;82: 1219 1226.