Low-Risk Prostate Cancer: The Accuracy of Multiparametric MR Imaging for Detection 1

Transcription

1 Note: This copy is for your personal non-commercial use only. To order presentation-ready copies for distribution to your colleagues or clients, contact us at Jin Young Kim, MD See Hyung Kim, PhD Young Hwan Kim, MD Hee Jung Lee, PhD Mi Jeong Kim, MD Mi Sun Choi, PhD Low-Risk Prostate Cancer: The Accuracy of Multiparametric MR Imaging for Detection 1 Purpose: Materials and Methods: To retrospectively determine diagnostic performance with multiparametric magnetic resonance (MR) imaging for detection of cancer of different tumor volumes and Gleason grades in patients with clinically low-risk prostate cancer. The local ethical committee and institutional review board approved this study. Consecutive patients with clinically determined low-risk cancer (n = 100) were examined with multiparametric MR imaging (T2 weighted, diffusion weighted, and dynamic contrast material enhanced) by using a 3.0- T imager before prostatectomy. Two radiologists independently assessed the likelihood of cancer per sextant. Cancers with a volume of 0.5 cm 3 or more identified at histopathologic examination were compared with multiparametric MR images. Cancer detection with multiparametric MR imaging was assessed for tumors of different volumes and Gleason grades by using a logistic generalized estimating equation model with 95% confidence intervals (CIs) with two optimal dichotomized cutoff scores. Original Research n Genitourinary Imaging 1 From the Departments of Radiology (J.Y.K., S.H.K., Y.H.K., H.J.L., M.J.K.) and Pathology (M.S.C.), Keimyung University, Dongsan Hospital, 216 Dalsung-ro, Jung-gu, Daegu , Korea. Received April 9, 2013; revision requested May 21; revision received July 2; accepted July 23; final version accepted October 29. Address correspondence to S.H.K. ( kseehdr@dsmc.or.kr). q RSNA, 2014 Results: Conclusion: For cancers greater than or equal to 0.5 cm 3, with respect to cancer volume and Gleason grade, multiparametric MR imaging showed high diagnostic performance for the detection of cancer. Diagnostic accuracy with multiparametric MR imaging was significantly higher for cancers with a volume greater than 1 cm 3 than for those with a volume of cm 3 (87.7%; 95% CI: 82.4%, 94.3% vs 82.6%; 95% CI: 79.0%, 88.7%; P =.02) and for cancers with Gleason grades of 7 or more than for those with grades of 6 or less (89.2%; 95% CI: 85.4%, 93.8% vs 80.6%; 95% CI: 71.2%, 89.8%; P =.01). Detection rates for cancers with a volume more than 1 cm 3 and a Gleason grade of 7 or more were significantly higher than for those with a volume of cm 3 and a Gleason grade of 6 or less(87.8%; 95% CI: 85.3%, 93.7% vs 82.0%; 95% CI: 75.6%, 86.1%; P =.01). Detection of prostate cancer in patients with clinically low-risk cancer with multiparametric MR imaging is highly accurate, and larger cancer volume and higher Gleason grade are associated with higher detection accuracy. q RSNA, 2014 Radiology: Volume 271: Number 2 May 2014 n radiology.rsna.org 435

2 Widespread screening for prostate cancer by determining serum prostate-specific antigen levels has resulted in the detection of small and early-stage prostate cancers (1 3). Newly diagnosed prostate cancers that fit the standard definition of clinically low-risk cancer (clinical stages of T1c T2a, Gleason grade, 7, prostate-specific antigen level of, 10 ng/ ml [10 mg/l]) comprise up to one-half of prostate cancers detected (4 7). The current trends in practice patterns suggest that approximately 90% of patients with these cancers undergo radical Advances in Knowledge nn In patients with clinically low-risk prostate cancer, accuracy for cancer detection was higher for cancers with a volume greater than 1 cm 3 than for those with a volume of cm 3 at multiparametric MR imaging and two other combinations of acquisition types (T2 weighted and diffusion weighted [DW], 79.7% vs 83.6%; T2 weighted and dynamic contrast material enhanced [DCE], 78.2% vs 81.6%; and multiparametric, 82.6% vs 87.7% for cancer volumes cm 3 and greater than 1 cm 3, respectively); it was also higher for cancers with a Gleason grade of 7 or more than for those with a grade of 6 or less (T2 weighted and DW, 75.8% vs 81.2%; T2 weighted and DCE, 74.1% vs 78.5%; multiparametric, 80.6% vs 89.2% for Gleason grades 6 and 7, respectively). nn In patients with clinically low-risk prostate cancer, the accuracy of detection at multiparametric MR imaging was higher for cancers with high pathologic volume and Gleason grade; for lesions with volumes greater than 1 cm 3 and Gleason grades of 7 or more, the accuracy of detection was significantly higher than that for lesions with volumes of cm 3 and Gleason grades of 6 or less (82.0% vs 87.8%, P =.01). prostatectomy because the natural history of low-risk cancer is poorly understood, and there is no consensus on the best practice (4,6). At surgery, many prostate cancers initially considered low-risk at biopsy prove to be of higher grade, stage, and volume. Therefore, transrectal ultrasonographically guided biopsies that are not directed at specific areas of an abnormality may allow clinically important disease to be missed and may result in an inaccurate cancer grade (4 8). Authors of previous studies (7,8) have suggested that pathologic cancer volume of 0.5 cm 3 or more and Gleason grade of 7 or more were the main determinants of clinical importance. Magnetic resonance (MR) imaging is the most detailed and accurate tool for imaging the prostate. Multiparametric MR imaging, including T2-weighted, diffusion-weighted (DW), and dynamic contrast material enhanced (DCE) MR imaging and MR spectroscopy, is currently the best imaging modality for the diagnosis and staging of prostate cancer (9 11). The combination of both anatomic and physiologic information is what makes multiparametric MR imaging such an appealing tool for the detection and grading of cancer. Authors of an increasing number of studies (12 14) are exploring multiparametric MR imaging, but relatively few authors have specifically evaluated the diagnostic performance of multiparametric MR imaging of cancers with different volumes and Gleason grades. Therefore, the purpose of our study was to retrospectively determine the diagnostic performance of multiparametric MR imaging for detection of cancers Implication for Patient Care nn In patients with clinically low-risk prostate cancer, cancer detection at multiparametric MR imaging is significantly dependent on cancer volume and Gleason grade; the detection rates for cancers with a volume of 1 cm 3 or more and a Gleason grade of 7 or more are significantly higher than those with a lower volume or Gleason grade. of different volumes and Gleason grades in patients with clinically low-risk prostate cancer. Materials and Methods Our study was approved by the local ethical committee and institutional review board, and the requirement for informed consent was waived. Patient Selection and Clinical Data Our hospital medical data were searched, and patients with clinically low-risk prostate cancer were identified by using the following inclusion criteria: (a) diagnosis of prostate cancer by means of 12-core transrectal prostate biopsy; (b) localized cancer with no more than two contiguous sextants involved, no core with cancer length greater than 7 mm, and a total cancer length of less than or equal to 10 mm at biopsy; (c) low D Amico risk (prostate-specific antigen level 10 ng/ ml [10 mg/l], Gleason grade 6, and positive biopsies 33%); (d) pretreatment multiparametric MR imaging; and (e) bone scan results negative for cancer. From January 2008 to September 2012, 142 patients met these inclusion criteria, 25 patients were excluded because of treatment with androgen deprivation or radiation therapy, and 17 patients were excluded because MR imaging was performed with a 1.5-T imager. Thus, 100 Published online before print /radiol Content codes: Radiology 2014; 271: Abbreviations: CI = confidence interval DCE = dynamic contrast material enhanced DW = diffusion weighted Author contributions: Guarantors of integrity of entire study, J.Y.K., S.H.K., H.J.L., M.J.K., M.S.C.; study concepts/study design or data acquisition or data analysis/interpretation, all authors; manuscript drafting or manuscript revision for important intellectual content, all authors; approval of final version of submitted manuscript, all authors; literature research, S.H.K., Y.H.K., H.J.L., M.J.K.; clinical studies, S.H.K., Y.H.K., H.J.L., M.J.K., M.S.C.; experimental studies, S.H.K.; statistical analysis, S.H.K., H.J.L., M.J.K.; and manuscript editing, J.Y.K., Y.H.K., H.J.L., M.J.K. Conflicts of interest are listed at the end of this article. 436 radiology.rsna.org n Radiology: Volume 271: Number 2 May 2014

3 Table 1 MR Imaging Parameters Methods and Orientation Repetition Time (msec) Echo Time (msec) Field of View (mm) Matrix Section Thickness (mm) No. of Sections Acquisition Time (sec) T2-weighted Axial Coronal Sagittal T1-weighted, axial DW, axial DCE, axial Figure 1 Figure 1: Flowchart shows selection of patients. Mp-MRI = multiparametric MR imaging. patients who underwent prostatectomy were included in our study (median age, 63 years; range, years) (Fig 1). To minimize hemorrhagic effects, all MR imaging examinations were performed at least 6 weeks after biopsy. The median time between MR imaging and prostatectomy was 14 days (range, days). The median level of serum prostate-specific antigen was 6.5 ng/ml (6.5 mg/l) (range, ng/ml [ mg/l]). MR Imaging MR imaging was performed with a 3.0- T MR imager (Magnetom Trio; Siemens Medical Solutions, Erlangen, Germany) with a pelvic phased-array coil (sixchannel Body Matrix coil; Siemens). An intramuscular injection of 1 mg of glucagon (Samil Pharmacy, Seoul, Korea) was given to suppress bowel peristalsis. The MR acquisition parameters are summarized in Table 1. After initial localizer images were obtained to determine the anatomic orientation of the prostate, T2-weighted fast spin-echo imaging was performed in three orthogonal planes. Spatial resolution was reduced in the sagittal plane to shorten acquisition time. T1- and T2-weighted fast spin-echo series of the entire pelvis were performed to detect lymph nodes and hemorrhage in the prostate. DW images were acquired by using two-dimensional echo-planar imaging with three b values (0, 100, and 1000 sec/mm 2 ) in three orthogonal directions. A three-dimensional fast low-angle shot sequence was used to perform DCE MR imaging. After four baseline acquisitions, gadobutrol (gadolinium chelate; Gadovist, Bayer, Germany) was administered as a bolus injection of 0.1 mmol/kg of body weight at a rate of 2 ml/sec, followed by a 20-mL flush of normal saline during a 10-second break before 21 contrastenhanced images were acquired with a temporal resolution of 12.9 seconds. In DCE MR imaging, there is a tradeoff between spatial and temporal resolution, and the chosen time resolution gave a good definition of the enhancement curves while maintaining sufficient spatial image resolution. Contrast agent distribution was observed for 4.5 minutes after injection. All axial images were identically angled along the prostate s longest axis, perpendicular to the urethra. Histopathologic Examination After open radical prostatectomy was performed, the prostate was fixed in 4% buffered formaldehyde for approximately 48 hours and was handled according to local clinical histopathologic routines for diagnostic purposes. The apical 3 7 mm of the prostate was amputated and longitudinally sliced. The remaining gland was serially sectioned from the apex to the base in 4-mm axial slices perpendicular to the urethra and was submitted for paraffin-embedded whole-mounts, from which 3.5-µm slides were stained with hematoxylin, eosin, and saffron. A pathologist (C.M.S, with more than 13 years of experience) who was blinded to the imaging results examined the hematoxylin-eosin and saffron stained slides, outlined cancer foci and described cancer location, and determined the cancer stage according to the TNM classification and cancer grade according to the Gleason grading system (15). The reported Gleason score was the total score for each patient. Correlation of MR Imaging and Histopathologic Results Multiparametric MR images were retrospectively and independently interpreted by two radiologists (K.S.H. and K.Y.H., with more than 7 years and 10 years of experience reading prostate MR images, respectively). All observers knew the patients had clinically determined low-risk disease but were blinded to patients specific clinical data and biopsy findings. By using the scoring system suggested by the European Society of Urogenital Radiology (Table 2), lesions identified at T2-weighted (with different descriptions for peripheral and transition zone), DW, and DCE MR imaging and MR spectroscopy were assigned Radiology: Volume 271: Number 2 May 2014 n radiology.rsna.org 437

4 Table 2 Scoring System of Multiparametric MR imaging for Prostate Cancer Detection Score T2-weighted (Peripheral Zone) T2-weighted (Transition Zone) DW DCE 1 Uniform high SI Heterogeneous adenoma with well-defined margins No reduction in ADC, no increase in SI on any high-b-value image (b 800) Diffuse, high SI at b 800 with low ADC Type 1 enhancement curve 2 Linear, wedge shaped, or geographic areas of lower SI Areas of more homogeneous low SI; however, margin is well defined Type 2 enhancement curve 3 Intermediate appearance Intermediate appearance Intermediate appearances Type 3 enhancement curve 4 Discrete, homogeneous areas of low Areas of more homogeneous low SI, Focal area of reduced ADC but iso Focal enhancing lesion with curve SI confined to prostate ill defined SI at high b value (b 800) type Discrete, homogeneous areas of low Involving anterior fibromuscular stroma Focal area of SI at high b value Asymmetric lesion or lesion at SI with extracapsular extension and/or or anterior horn of peripheral zone, (b 800) with reduced ADC unusual place with curve invasive behavior or mass effect usually lenticular or water-drop type 2 3 on capsule bulging or broad shaped (.1.5 cm) contact with surface Note. MR image review scales were as follows: score 1= clinically significant disease is highly unlikely to be present, score 2 = clinically significant cancer is unlikely to be present, score 3 = clinically significant cancer is equivocal, score 4 = clinically significant cancer is likely to be present, and score 5= clinically significant cancer is highly likely to be present. SI = signal intensity, ADC = apparent diffusion coefficient. scores on a five-point scale (15). Each lesion was given an overall score to predict the chance of it being a clinically significant cancer. The score for each parametric acquisition was added to generate a combined score, and optimal or cutoff values were defined for the assessment of clinically low-risk cancer by using multiparametric MR imaging. In addition, observers drew the locations of cancer on a schematic representation of the right and left peripheral and transition zone at the level of the base, midgland, and apex. For the purpose of our study, the prostate was divided into an upper third, which included the region at the base of the prostate; a middle third, which included the region at the level of the verumontanum; and a lower third, which included the remaining apical portion of the prostate. Whole-mount step-section histopathologic cancer maps served as the reference standard for MR imaging findings in each patient. A pathologist mapped prostate cancer foci in each section and determined the Gleason grade patterns present in each lesion. The volume for each cancer focus was calculated with absolute measurement planimetry by using the following equation (16): (TA ST SF)/PW 100, where TA is the sum of all tumor areas on the slides in square centimeters, ST is the section thickness (0.3 cm), SF is the shrinkage factor (1.33), and PW is the prostate weight in grams. All acquisitions were displayed on the screen of a workstation with automatic synchronization of magnification and section position. T2-weighted MR images was reviewed first, then DW and DCE MR images immediately after. The correlation of MR imaging to histopathologic results was performed by a radiologist who was not involved in the MR interpretation (L.H.J., with 15 years of experience in prostate MR imaging), taking into account the location of cancer determined in relation to benign anatomic landmarks such as the urethra and ejaculatory ducts, benign prostatic hyperplasia nodules, the size and shape of the peripheral zone, and distance from the apex or base. Cancer locations were assigned to the prostatic sextant regions (right, left; base, midgland, apex) on the basis of wholemount histopathologic cancer maps. In addition, the location of the lesion with the largest volume was determined in each patient. Statistical Analyses Quadratic k coefficients with Fleiss-Cohen (quadratic) weights were calculated to assess the interobserver agreement between the two radiologists for detection of cancer for the three different MR imaging modalities. A k value of was considered poor agreement; , fair agreement; , moderate agreement; , substantial agreement; and , excellent agreement. Statistical analyses were performed at the sextant level and at the lesion level for individual cancers with a volume of 0.5 cm 3 or more. For lesionlevel analyses, if a patient had multiple cancers greater than or equal to 0.5 cm 3, all were included. Detection of the largest lesion with a cancer volume of greater than or equal to 0.5 cm 3 in each patient was also assessed. Lesionlevel analysis was first performed at T2-weighted MR imaging alone, then at T2-weighted and DW MR imaging, and T2-weighted and DCE MR imaging. Radiologists interpretation scores were dichotomized as follows: A score of 3 or higher indicated intermediate or probable cancer and a score of 4 or higher indicated probable or definite cancer. The detection rates for cancers with volumes greater than or equal to 0.5 cm 3 of differing Gleason grades were assessed and compared by using a logistic generalized estimating 438 radiology.rsna.org n Radiology: Volume 271: Number 2 May 2014

5 equation model. The 95% confidence intervals (CIs) were calculated on the basis of the robust variance estimation. The empirical receiver operating characteristic curve and the corresponding area under the curve were estimated nonparametrically for the detection of cancer in combination with other MR imaging modalities (T2- weighted and DW MR imaging; T2- weighted and DCE MR imaging; T2- weighted, DW, and DCE MR imaging), by taking into account multiple sextants or multiple lesions per patient. The sensitivity, specificity, positive predictive value, negative predictive value, and accuracy were calculated along with the corresponding 95% CIs at the sextant and lesion levels by taking into account two more optimal dichotomized cutoff scores added on the basis of cancer volumes and Gleason grades. The values of the scores were added to define T2-weighted and DW MR imaging (n = 0 10); T2-weighted and DCE MR imaging (n = 0 10); and T2-weighted, DW, and DCE MR imaging (n = 0 15) values and the optimized cutoffs were calculated by using an empirical receiver operating characteristic analysis. The diagnostic accuracy of multiparametric MR imaging for the detection of cancer was assessed in univariate associations between both cancer volumes and Gleason grades by using the generalized estimating equation model with a Bonferroni-Holm correction and a robust covariance matrix. All statistical analyses were performed by using a software package (SAS 9.2; SAS Institute, Cary, NC). P values less than or equal to.05 were considered to indicate a significant difference. Results Patient Characteristics Patient characteristics are summarized in Table 3. Among 100 patients with clinically low-risk cancer, there were 44 (44%) patients with Gleason grades greater than or equal to 7 and 72 (72%) patients with stage T2b disease or higher. Among the 118 cancer foci that were greater than or equal to 0.5 cm 3, there were 56 cancers with Gleason grades greater than or equal to 7 and 49 cancers with volumes of 1 cm 3 or larger. As determined with wholemount histopathologic cancer maps, the mean cancer volume was 1.32 cm 3 (range, cm 3 ). There were 88 cancers in the peripheral zone (mean volume 6 standard deviation, 1.2 cm ) and 30 cancers in the transition zone (mean volume, 1.8 cm ). Cancers with Gleason grades of less than or equal to 6 and grades greater than or equal to 7 had mean volumes of 0.8 cm and 1.4 cm , respectively. Multiparametric MR Imaging Characteristics in Cancer Detection There was excellent agreement for lesion detection with the three different MR imaging modalities at the sextant level. (weighted k, k = 0.83 for T2- weighted, k = 0.86 for DW, and k = 0.87 for DCE MR imaging). Table 4 summarizes sensitivities, specificities, positive and negative predictive values, and accuracy for detection of cancer at the sextant level for each MR imaging modality on the basis of the scoring system we used. T2- weighted MR imaging alone demonstrated a detection rate of 69.4% (82 of 118, 95% CI: 64.5, 73.4) for lesions with at least a probable possibility of becoming clinically significant cancer, T2-weighted and DW MR imaging had a detection rate of 78.8% (93 of 118, 95% CI: 73.2%, 84.5%); and T2-weighted and DCE MR imaging had a detection rate of 68.6% (81 of 118, 95% CI: 63.7%, 75.6%). The detection rates for cancers greater than or equal to 0.5 cm 3 without regard to cancer volume or Gleason grade were significantly better with T2-weighted and DW MR imaging than with T2- weighted MR imaging alone (P =.01) or T2-weighted and DCE MR imaging (P =.01). The diagnostic accuracy of multiparametric MR imaging for detection of cancers greater than or equal to 0.5 cm 3 was associated with pathologic Table 3 Clinical and Pathologic Characteristics of the 100 Patients Characteristic Value Clinical characteristic* Age (y) 63.3 (51 76) Prostate volume (cm 3 ) 37 (20 115) Preoperative PSA (ng/dl) 6.5 ( ) Digital rectal examination Normal 92 (92) Suspicious 8 (8) Biopsy characteristic* No. of positive biopsies 2 (1 2) Tumor length (mm) 6 (1 16) Radical prostatectomy characteristics pt2a 28 pt2b 23 pt2c 36 pt3a 10 pt3b 3 Gleason score (primary + secondary score) (56) (34) (8) (2) Focality Unifocal 41 (41) Multifocal 59 (59) Bilateral 32 (32) Tumor foci location Peripheral zone 88 (76) Transition zone 30 (24) Note. Unless otherwise indicated, data are numbers, with percentages in parentheses. PSA = prostatespecific antigen. *Data are medians, with range in parentheses. cancer volumes and Gleason grades (Tables 5, 6). The accuracy of detection was higher for cancers with volumes of greater than 1 cm 3 than for cancers with volumes of cm 3 in either of the two other combined acquisitions (T2-weighted and DW MR imaging, 79.7% [55 of 69, 95% CI: 79.4%, 85.2%] vs 83.6% [41 of 49, 95% CI: 81.4%, 89.2%]; T2-weighted and DCE MR imaging, 78.2% [54 of 69, 95% CI: 70.9%, 82.1%] vs 81.6% [40 of 49, 95% CI: 75.7%, 84.2%]; and multiparametric MR imaging, 82.6% [57 of 69, 95% CI: 79.0%, Radiology: Volume 271: Number 2 May 2014 n radiology.rsna.org 439

6 Table 4 Cancer Detection at Prostate Sextant Level with Scores Based on Modality and Cutoff Level Modality and Cutoff Level Sensitivity (%) Specificity (%) Positive Predictive Value (%) Negative Predictive Value (%) Accuracy (%) T2 weighted (31/45) [65.1, 75.2] 71.2 (52/73) [63.5, 78.5] 65.8 (27/41) [57.5, 79.1] 76.6 (59/77) [65.4, 85.7] 57.6 (68/118) [51.5, 69.1] (11/34) [25.5, 37.2] 94.0 (79/84) [92.0, 96.1] 75.0 (24/32) [71.6, 86.2] 66.2 (57/86) [61.2, 72.0] 69.4 (82/118) [64.5, 73.4] DW and T2 weighted (43/54) [78.7, 84.3] 81.2 (52/64) [77.6, 87.8] 63.8 (30/47) [56.1, 70.8] 83.0 (59/71) [73.5, 85.8] 72.0 (85/118) [69.2, 80.4] (17/39) [38.8, 50.8] 91.1 (72/79) [89.3, 94.0] 77.1 (27/35) [72.2, 85.0] 69.8 (58/83) [65.8, 74.5] 78.8 (93/118) [73.2, 84.5] DCE and T2 weighted (15/49) [25.2, 38.1] 79.7 (55/69) [75.6, 84.8] 59.0(26/44) [49.8, 66.2] 75.6 (56/74) [56.9, 80.8] 61.8 (73/118) [57.5, 68.1] (12/44) [22.7, 32.6] 93.2 (69/74) [91.6, 94.9] 71.7 (28/39) [67.3, 79.9] 63.2 (50/79) [58.6, 69.5] 68.6 (81/118) [63.7, 75.6] Note. Data in parentheses are numbers used to calculate the percentage and data in brackets are 95% CIs. Scores of equal or greater were cut-off values on European Society of Urogenital Radiology scoring system. Table 5 Cancer Detection at Prostate Sextant Level with Scores Based on Cancer Volume Cancer Volume, Modality, and Score Sensitivity (%) Specificity (%) Positive Predictive Value (%) Negative Predictive Value (%) Accuracy (%) cm 3 T2 weighted and DW, 7 T2 weighted and DCE, 6 T2 weighted, DW, and DCE, 10.1 cm 3 T2 weighted and DW, 6 T2-weighted and DCE, 6 T2-weighted, DW, and DCE, (28/36) [71.4, 84.9] 71.0 (27/38) [68.9, 78.8] 60.6 (20/33) [48.1, 68.9] 83.3 (30/36) [79.4, 89.2] 79.7 (55/69) [79.4, 85.2] 79.4 (31/39) [71.7, 84.3] 73.3 (22/30) [67.3, 80.7] 57.1 (20/35) [51.3, 68.1] 76.4 (26/34) [70.5, 84.4] 78.2 (54/69) [70.9, 82.1] 80.9 (34/42) [74.6, 86.1] 70.3 (19/27) [64.7, 77.3] 60.5 (23/28) [56.0, 68.2] 83.8(26/31) [81.3, 89.7] 82.6(57/69) [79.0, 88.7] 61.2 (19/30) [59.4, 70.3] 84.2 (16/19) [82.9, 90.8] 85.1 (23/27) [80.9, 88.9] 72.7 (16/22) [68.5, 77.4] 83.6 (41/49) [81.4, 89.2] 71.4 (20/28) [67.3, 78.8] 80.9 (17/21) [78.7, 84.6] 80.0 (20/25) [78.5, 87.3] 66.6 (16/24) [65.5, 72.6] 81.6 (40/49)[75.7, 84.2] 55.8 (19/34) [52.6, 64.1] 80.0 (12/15) [78.7, 84.3] 83.3 (25/30) [76.0, 85.4] 73.6 (14/19) [71.4, 89.8] 87.7 (43/49) [82.4, 94.3] Note. Data in parentheses are numbers used to calculate the percentage and data in brackets are 95% CIs. Scores of equal or greater were cutoff values of European Society of Urogenital Radiology scoring system. 88.7%] vs 87.7% [43 of 49, 95% CI: 82.4%, 94.3%], for volumes of cm 3 and.1 cm 3, respectively); it was also higher for cancers with Gleason grades greater than or equal to 7 than for cancers with Gleason grades of less than or equal to 6 (T2-weighted and DW MR imaging, 75.8% [47 of 62, 95% CI: 73.4%, 85.2%] vs 82.1% [46 of 56, 95% CI: 76.5%, 85.8%]; T2-weighted and DCE MR imaging, 74.1% [46 of 62, 95% CI: 67.5%, 80.9%] vs 78.5% [44 of 56, 95% CI: 69.7%, 82.5%]; and multiparametric MR imaging, 80.6% [50 of 62, 95% CI: 71.2%, 89.8%] vs 89.2% [50 of 56, 95% CI: 85.4%, 93.8%] for Gleason grades less than or equal to 6 and greater than or equal to 7, respectively). T2-weighted and DW MR imaging, T2-weighted and DCE MR imaging, and multiparametric MR imaging performed significantly better than T2-weighted MR imaging alone for detection of cancer foci (P =.02, P =.03, P =.02, respectively). The best diagnostic performance was obtained with multiparametric MR imaging, which performed significantly better than T2-weighted and DW MR imaging and T2-weighted and DCE MR imaging (P =.01, P =.02, respectively) (Fig 2). T2-weighted and DW MR imaging performed better than T2-weighted and DCE MR imaging, but the difference was not significant. 440 radiology.rsna.org n Radiology: Volume 271: Number 2 May 2014

7 Table 6 Cancer Detection at Prostate Sextant Level with Scores Based on Gleason Grades Gleason Grade, Modality, and Score Sensitivity (%) Specificity (%) Positive Predictive Value (%) Negative Predictive Value (%) Accuracy (%) Gleason grade 6 T2 weighted and DW, 7 T2 weighted and DCE, 7 T2 weighted, DW, and DCE, 10 Gleason grade 7 T2 weighted and DW, 7 T2 weighted and DCE, 6 T2-weighted, DW, and DCE, (27/34) [74.4, 88.6] 67.8 (19/28) [63.9, 75.4] 56.6 (17/30) [45.1, 65.2] 81.2 (26/32) [78.3, 87.1] 75.8 (47/62) [73.4, 85.2] 80.6 (25/31) [75.3, 87.2] 70.9 (22/31) [62.8, 75.7] 53.5 (15/28) [45.1, 63.1] 73.5 (25/34) [67.2, 83.1] 74.1 (46/62) [67.5, 80.9] 84.2(32/38) [82.6, 89.1] 66.6 (16/24) [61.5, 78.5] 62.8 (23/28) [56.4, 70.6] 85.1 (23/27) [79.3, 91.4] 80.6 (50/62) [71.2, 89.8] 59.5 (25/42) [57.4, 69.3] 78.5 (11/14) [78.1, 89.6] 87.5 (35/40) [79.3, 94.9] 68.7 (11/16) [64.1, 76.4] 82.1 (46/56) [76.5, 85.8] 66.6 (26/39) [61.4, 73.3] 82.3 (14/17) [73.1, 87.9] 81.0 (30/37) [76.4, 84.3] 66.6 (12/18) [62.3, 75.8] 78.5 (44/56) [69.7, 82.5] 60.0 (27/45) [ ] 81.8 (9/11) [ ] 81.8 (36/44) [ ] 75.0 (9/12) [ ] 89.2 (50/56) [ ] Note. Data are percentages, with numbers used to calculate percentages in parentheses and 95% CIs in brackets. Scores of equal or greater were cutoff values on European Society of Urogenital Radiology scoring system. In the analysis of diagnostic performance for different cancer volumes with optimal cutoff score from the receiver operating characteristic curves, the sensitivity and specificity of the multiparametric MR imaging were 80.9% (34 of 42, 95% CI: 74.6%, 86.1%) and 70.3% (19 of 27, 95% CI: 64.7%, 77.3%), respectively, in cancer foci of cm 3 volume (score 10) and 55.8% (19 of 34, 95% CI: 52.6%, 64.1%) and 80.0% (12 of 15, 95% CI: 78.7%, 84.3%), respectively, in cancer foci greater than 1 cm 3 in volume (score 9). The diagnostic accuracy of multiparametric MR imaging for cancer volume was 82.6% (57 of 69, 95% CI: 79.0%, 88.7%) and 87.7% (43 of 49, 95% CI: 82.4%, 94.3%), respectively, and was significantly higher for cancers with a volume greater than 1 cm 3 than for cancers with volumes of cm 3 (P =.02). According to Gleason grades, the sensitivity and specificity of multiparametric MR imaging were 84.2% (32 of 38, 95% CI: 82.6%, 89.1%) and 66.6% (16 of 24, 95% CI: 61.5%, 78.5%), respectively, in cancer foci with Gleason grades less than or equal to 6 (score 10) and 60.0% (27 of 45, 95% CI: 56.6%, 65.2%) and 81.8% (nine of 11, 95% CI: 76.8%, 89.3%), respectively, in cancer foci with Gleason grades greater than or equal to 7 (score 9). The diagnostic accuracy of multiparametric MR imaging for Gleason grades was 80.6% (50 of 62, 95% CI: 71.2%, 89.8%) and 89.2% (50 of 56, 95% CI: 85.4, 93.8), respectively, and was significantly higher for cancers of Gleason grades greater than or equal to 7 than for cancers with grades less than or equal to 6 (P =.01). The diagnostic accuracy of multiparametric MR imaging increased with higher pathologic cancer volumes and Gleason grades compared in the subanalysis of the combined effects of these variables on diagnostic performance (Table 7). The accuracy for detection of cancers with volumes greater than 1 cm 3 and Gleason grades greater than or equal to 7 was significantly higher than that for cancer volumes of cm 3 and Gleason grades less than or equal to 6 (87.8% [29 of 33, 95% CI: 85.3%, 93.7%] vs 82.0% [32 of 39, 95% CI: 75.6%, 86.1%], respectively; P =.01). The accuracy of the other two combined acquisitions was also significantly increased with high cancer volumes and Gleason grades, similar to that of multiparametric MR imaging (Fig 3). Discussion Our results showed that the patients with clinically low-risk prostate cancer had a moderate likelihood of harboring high-risk disease. Fifty-six (56%) of 100 patients suspected of having only clinically low-risk disease had low-risk cancers. Previous studies have shown significant pathologic upgrading at radical prostatectomy, with authors of recent studies reporting upgrading percentages of 20.3% 54% (17). This potentially creates a problem with accurately identifying and monitoring patients with presumably low-risk disease before surgery. Our results confirmed findings from retrospective studies in smaller groups of patients, which have suggested relationships between cancer volume, Gleason grade, and cancer detection with MR imaging. Combined functional or metabolic MR imaging techniques were investigated in various studies (17 21) for the detection of cancer. Studies (22 25) have reported that cancers of Gleason grades less than or equal to 6 or volumes less than 1 cm 3 were difficult to detect, even with various combined MR acquisitions. In addition, approximately 20% of low-risk cancers were Radiology: Volume 271: Number 2 May 2014 n radiology.rsna.org 441

8 Figure 2 Figure 2: Prostate cancer in a 67-year-old patient with serum prostate-specific antigen level of 7.5 ng/ml (7.5 mg/l). One left biopsy core shows 7-mm cancer with Gleason grade of 6 (3 + 3). (a) T2-weighted image shows discrete, homogeneous areas of low signal intensity in left peripheral zone (score, 4 of 5; arrow). (b) DW image shows apparent diffusion coefficient for intermediate appearance of cancer focus (score, 3 of 5; arrow). (c, d) DCE images and color maps overlaid on T2-weighted MR images show focal enhancement (arrow; red circle: cancer focus; green circle: normal prostate tissue), compared with reference prostate tissue. (e) Graph shows region with type 3 enhancement curve (red curve, wash-out). Green curve is type 1, normal prostate tissue. Multiparametric MR imaging score was 11 in left peripheral zone. Cancer foci were not found in right peripheral or transition zones on multiparametric MR images. (f) Photomicrograph of whole-mount stepsection pathologic specimen shows Gleason grade 7 (3 + 4) cancer focus with volume greater than 1 cm 3 in left peripheral zone (black outline), and four cancer foci with Gleason grade of 6 (3 + 3) and volumes of less than 0.5 cm 3 in right peripheral zone (black outlines) (hematoxylin-eosin stain; original magnification, 31.05). Roi = region of interest. not visible on MR images, even with knowledge of the cancer location from pathologic results (24). To our knowledge, there was only one study (26) in which investigators used T2-weighted MR imaging combined with MR spectroscopy to assess the detection of cancer in patients with low-risk prostate cancer, similar to those in our study. Detection of cancers with a volume of cm 3 is significantly dependent on Gleason grades. In addition, authors of another study (27) assessed the detection of low-risk cancer. By using imaging scores as a threshold to define cancer by means of qualitative assessment, the diagnostic accuracy was significantly increased with multiparametric MR imaging. However, in clinically low-risk cancer, no attempt was made to assess associations between the diagnostic performance of multiparametric MR imaging and cancer volumes or Gleason grades. In our study, T2-weighted, DW, and DCE MR imaging were synchronously examined, and cancer foci detected with each acquisition were compared among two or three combined acquisitions. Cancers with only an equivocal or probable level at T2-weighted, DW, or DCE MR imaging were upgraded to a probable or definite level with combined acquisitions. Therefore, the accuracy for the detection of cancer was increased with multiparametric MR imaging. Our results are in close accord with those of Tamada et al (28), who obtained a sensitivity and specificity of 69% and 85%, derived retrospectively, for multiparametric MR imaging. In our study, we used whole-mount stepsection histopathologic examination results and compared them with the results of multiparametric MR imaging 442 radiology.rsna.org n Radiology: Volume 271: Number 2 May 2014

9 Table 7 Cancer Detection Based on Cancer Volume and Gleason Grade for Multiparametric MR Imaging Volume and Grade T2-weighted and DW T2-weighted and DCE T2-weighted, DW, and DCE cm 3 and Gleason (29/39) [68.2, 88.7] 71.7 (28/39) [64.3, 76.8] 82.0 (32/39) [75.6, 86.1] cm 3 and Gleason (20/27) [72.5, 82.6] 74.0 (20/27) [70.8, 80.4] 85.1 (23/27) [80.1, 91.3].1 cm 3 and Gleason (14/19) [72.1, 85.1] 63.1 (12/19) [69.2, 81.6] 84.2 (16/19) [79.4, 90.9].1 cm 3 and Gleason (27/33) [77.2, 89.3] 78.7 (26/33) [72.6, 85.3] 87.8 (29/33) [85.3, 93.7] Note. Data are percentages, with numbers used to calculate percentages in parentheses and 95% CIs in brackets. Figure 3 Figure 3: Graphs show receiver operating characteristic curve representing the diagnostic accuracy of other combined acquisitions according to pathologic cancer volumes and Gleason grades. (a) T2-weighted and DW MR imaging, (b) T2-weighted and DCE MR imaging, and (c) T2-weighted, DW, and DCE MR imaging (black line = volume. 1 cm 3 and Gleason 7; gray line = volume of cm 3 and Gleason 6). for cancers with volumes greater than or equal to 0.5 cm 3. In cancers with volumes of cm 3 and greater than 1 cm 3, the detection rate was higher for those with Gleason grades greater than or equal to 7 than in those with Gleason grades less than or equal to 6. Cancer volume also had an important effect on detection. The detection of cancer with multiparametric MR imaging is affected by both cancer volume and Gleason grade. Our study had limitations. First, we did not perform a subanalysis for cancer detection in the peripheral and transition zones of the prostate. The sextant grid that was used to record detection was not designed to make a distinction between the peripheral and transition zone. However, we are aware that detection of transition zone cancers remains very challenging, and we Disclosures of Conflicts of Interest: J.Y.K. No relevant conflicts of interest to disclose. S.H.K. No relevant conflicts of interest to disclose. Y.H.K. No relevant conflicts of interest to dishave explored the role of functional MR imaging in discrimination of cancers in the central gland of the prostate. Second, the use of histopathologic examination as the gold standard can be difficult because of problems in matching. This was because of the deviating angle between the MR images and the histopathologic sectioning of the prostate. The histopathologic slides lacked orientation marks, precluding correlation with MR images. Furthermore, nonuniform histopathologic slice and MR imaging section thickness makes matching even more difficult. This emphasizes the importance of establishing solid and precise routines in the histopathologic work up. Third, we used cutoff scores derived on the basis of a retrospective analysis with European Society of Urogenital Radiology guidelines. However, this scoring system has not been validated independently, and the criteria for assigning scores to lesions identified in each sequence are not yet established. Fourth, both radiologists were aware of the fact that all patients had biopsy-proved low-risk prostate cancer. This could have biased the image interpretation toward higher sensitivity. In summary, multiparametric MR imaging has high accuracy in the detection of cancer in patients with clinically low-risk cancer, and the cancers with higher volumes and Gleason grades allowed for higher detection accuracy. Acknowledgment: We are grateful to Sun Young Lee, PhD, for the statistical analyses in this manuscript. Radiology: Volume 271: Number 2 May 2014 n radiology.rsna.org 443

10 close. H.J.L. No relevant conflicts of interest to disclose. M.J.K. No relevant conflicts of interest to disclose. M.S.C. No relevant conflicts of interest to disclose. References 1. Polascik TJ, Oesterling JE, Partin AW. Prostate specific antigen: a decade of discovery what we have learned and where we are going. J Urol 1999;162(2): Jemal A, Siegel R, Ward E, Hao Y, Xu J, Thun MJ. Cancer statistics, CA Cancer J Clin 2009;59(4): Cooperberg MR, Lubeck DP, Meng MV, Mehta SS, Carroll PR. The changing face of low-risk prostate cancer: trends in clinical presentation and primary management. J Clin Oncol 2004;22(11): Hayes JH, Ollendorf DA, Pearson SD, et al. Active surveillance compared with initial treatment for men with low-risk prostate cancer: a decision analysis. JAMA 2010;304(21): Cooperberg MR, Broering JM, Carroll PR. Time trends and local variation in primary treatment of localized prostate cancer. J Clin Oncol 2010;28(7): Thompson IM, Klotz L. Active surveillance for prostate cancer. JAMA 2010;304(21): Bill-Axelson A, Holmberg L, Ruutu M, et al. Radical prostatectomy versus watchful waiting in early prostate cancer. N Engl J Med 2011;364(18): Noldus J, Graefen M, Haese A, Henke RP, Hammerer P, Huland H. Stage migration in clinically localized prostate cancer. Eur Urol 2000;38(1): Delongchamps NB, Rouanne M, Flam T, et al. Multiparametric magnetic resonance imaging for the detection and localization of prostate cancer: combination of T2-weighted, dynamic contrast-enhanced and diffusion-weighted imaging. BJU Int 2011;107(9): Shukla-Dave A, Hricak H, Kattan MW, et al. The utility of magnetic resonance imaging and spectroscopy for predicting insignificant prostate cancer: an initial analysis. BJU Int 2007;99(4): Seitz M, Shukla-Dave A, Bjartell A, et al. Functional magnetic resonance imaging in prostate cancer. Eur Urol 2009;55(4): Tan CH, Wang J, Kundra V. Diffusion weighted imaging in prostate cancer. Eur Radiol 2011;21(3): Alonzi R, Padhani AR, Allen C. Dynamic contrast enhanced MRI in prostate cancer. Eur J Radiol 2007;63(3): Langer DL, van der Kwast TH, Evans AJ, Trachtenberg J, Wilson BC, Haider MA. Prostate cancer detection with multi-parametric MRI: logistic regression analysis of quantitative T2, diffusion-weighted imaging, and dynamic contrast-enhanced MRI. J Magn Reson Imaging 2009;30(2): Barentsz JO, Richenberg J, Clements R, et al. ESUR prostate MR guidelines Eur Radiol 2012;22(4): Montironi R, van der Kwast T, Boccon- Gibod L, Bono AV, Boccon-Gibod L. Handling and pathology reporting of radical prostatectomy specimens. Eur Urol 2003;44(6): Turkbey B, Mani H, Shah V, et al. Multiparametric 3T prostate magnetic resonance imaging to detect cancer: histopathological correlation using prostatectomy specimens processed in customized magnetic resonance imaging based molds. J Urol 2011;186(5): Rastinehad AR, Baccala AA Jr, Chung PH, et al. D Amico risk stratification correlates with degree of suspicion of prostate cancer on multiparametric magnetic resonance imaging. J Urol 2011;185(3): 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(3): Zakian KL, Sircar K, Hricak H, et al. Correlation of proton MR spectroscopic imaging with gleason score based on step-section pathologic analysis after radical prostatectomy. Radiology 2005;234(3): Tanimoto A, Nakashima J, Kohno H, Shinmoto H, Kuribayashi S. Prostate cancer screening: the clinical value of diffusion-weighted imaging and dynamic MR imaging in combination with T2- weighted imaging. J Magn Reson Imaging 2007;25(1): Turkbey B, Shah VP, Pang Y, et al. Is apparent diffusion coefficient associated with clinical risk scores for prostate cancers that are visible on 3-T MR images? Radiology 2011;258(2): D Amico AV, Renshaw AA, Cote K, et al. Impact of the percentage of positive prostate cores on prostate cancer-specific mortality for patients with low or favorable intermediate-risk disease. J Clin Oncol 2004;22(18): Delongchamps NB, Beuvon F, Eiss D, et al. Multiparametric MRI is helpful to predict tumor focality, stage, and size in patients diagnosed with unilateral low-risk prostate cancer. Prostate Cancer Prostatic Dis 2011;14(3): Mazaheri Y, Hricak H, Fine SW, et al. Prostate tumor volume measurement with combined T2-weighted imaging and diffusion-weighted MR: correlation with pathologic tumor volume. Radiology 2009;252(2): Vargas HA, Akin O, Shukla-Dave A, et al. Performance characteristics of MR imaging in the evaluation of clinically low-risk prostate cancer: a prospective study. Radiology 2012;265(2): 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(1): Tamada T, Sone T, Jo Y, et al. Prostate cancer: relationships between postbiopsy hemorrhage and tumor detectability at MR diagnosis. Radiology 2008;248(2): radiology.rsna.org n Radiology: Volume 271: Number 2 May 2014