Prostate Cancer: Comparison of Tumor Visibility on Trace Diffusion- Weighted Images and the Apparent Diffusion Coefficient Map

Transcription

1 Genitourinary Imaging Original Research Rosenkrantz et al. Tumor Visibility in Prostate Cancer Genitourinary Imaging Original Research Andrew B. Rosenkrantz 1 Xiangtian Kong 2 Benjamin E. Niver 1 Douglas S. Berkman 3 Jonathan Melamed 2 James S. Babb 1 Samir S. Taneja 3 Rosenkrantz AB, Kong X, Niver BE, et al. Keywords: apparent diffusion coefficient, diffusionweighted imaging, MRI, prostate cancer, T2-weighted imaging DOI: /AJR Received April 6, 2010; accepted after revision May 24, Department of Radiology, New York University Langone Medical Center, 560 First Ave., TCH-HW202, New York, NY Address correspondence to A. B. Rosenkrantz (andrew.rosenkrantz@nyumc.org). 2 Department of Pathology, New York University Langone Medical Center, New York, NY. 3 Department of Urology, New York University Langone Medical Center, New York, NY. AJR 2011; 196: X/11/ American Roentgen Ray Society Prostate Cancer: Comparison of Tumor Visibility on Trace Diffusion- Weighted Images and the Apparent Diffusion Coefficient Map OBJECTIVE. The purpose of our study was to compare the visibility of prostate cancer on trace diffusion-weighted (DW) images and the apparent diffusion coefficient (ADC) map. MATERIALS AND METHODS. In this retrospective study, 45 patients with prostate cancer underwent preoperative MRI, including DW imaging (DWI) (b values 0, 500, and 1,000 s/mm 2 ). A single observer reviewed the images in conjunction with tumor maps constructed from prostatectomy. For 132 peripheral zone (PZ) tumor foci, the visibility and contrast relative to benign PZ were recorded for T2-weighted imaging, trace DWI b500 images, trace DWI b1,000 images, and ADC maps. Trace DWI b1,000 images and ADC maps were compared in terms of Gleason score, size, normalized T2 signal intensity, ADC, and normalized ADC of visible tumors. RESULTS. For each image set, the percentage of visible tumor foci and contrast relative to benign PZ were as follows: T2-weighted imaging, 80.3% and 0.411; trace DWI b500, 26.5% and 0.131; trace DWI b1,000, 46.2% and 0.119; and ADC maps, 62.1% and Forty-seven tumor foci were visible on both trace DWI b1,000 images and ADC maps, 14 only on trace DWI b1,000 images, 35 only on ADC maps, and 36 on neither image set. There was no significant difference in Gleason score, size, normalized T2 signal intensity, ADC, or normalized ADC between tumors visible only on trace DWI b1,000 images and those visible only on ADC maps. CONCLUSION. Given a greater proportion of tumors visible on the ADC map than trace DWI and greater contrast relative to benign PZ on the ADC map, we suggest that, when performing DWI of the prostate, careful attention be given to the ADC map for tumor identification. T he accurate detection and localization of tumor within the prostate using MRI may be useful in the planning of surgery and radiation therapy, guidance of focal ablative therapies, and monitoring of active surveillance [1 4]. Standard anatomic T2-weighted imaging has been limited in this regard, with large variation in reported sensitivity and specificity for tumor detection [2]. Diffusion-weighted imaging (DWI) offers a functional assessment of the prostate that has been shown to provide added value compared with T2-weighted imaging alone for accurate tumor detection [5 7]. Validation of this technique is provided by studies showing a lower apparent diffusion coefficient (ADC) value in areas of histologically defined tumor compared with ADC values obtained within benign peripheral zone (PZ) [8, 9]. DWI remains an emerging technique for oncologic imaging, and its acquisition, processing, and interpretation are not yet stan- dardized. One source of variability in its application relates to the generation of two separate image sets when performing DWI, namely the trace DW images and the ADC map. Specifically, studies supporting the utility of DWI for prostate cancer detection and localization have variably reported that the interpreting radiologists reviewed the trace DW images alone [10], reviewed the ADC map alone [6, 11, 12], or performed a joint review of both the trace DW images and the ADC map [5, 13, 14]. It is important to recognize any difference between these two image sets in terms of ability to detect tumor because such a difference impacts the manner in which DWI of the prostate is optimally assessed in the clinical setting. Indeed, differences in utility for tumor detection between trace DW images and the ADC map have been observed for other organs. In the assessment of nodules in both the lung and the cirrhotic liver, greater utility for distinguishing benign and AJR:196, January

2 Rosenkrantz et al. malignant nodules has been shown for signal intensity ratios obtained from trace DW images than for ADC values obtained from the ADC map [15, 16]. Given the different approaches that are possible for the interpretation of the two image sets that are obtained when performing DWI of the prostate, the aim of this study was to compare the visibility of prostate cancer on trace DW images and the ADC map as well as to assess the features of prostate cancer that influence the visibility of tumor foci on these respective image sets. Materials and Methods Patients This retrospective study was compliant with HIPAA and approved by our institutional review board, with waiver of the requirement for written informed consent. Forty-seven consecutive men with a biopsy-proven diagnosis of prostate cancer who had undergone prostate MRI before radical prostatectomy were assessed; one patient was excluded because of poor signal-to-noise ratio on DW images, which can lead to spurious ADC values [17], and one patient was excluded for having received hormonal therapy before MRI, leaving a final cohort of 45 patients (mean age, 63 ± 8 years; age range, years). Patients were included regardless of the delay between biopsy and MRI. Among the final cohort of 45 men, the mean interval between biopsy and MRI was 69 ± 42 days (range, days), and the mean interval between MRI and prostatectomy was 19 ± 19 days (range, 2 76 days). The mean preoperative prostate specific antigen (PSA) was 6.0 ± 5.4 ng/ml (range, ng/ml). MRI All patients underwent prostate MRI on one of three 1.5-T clinical systems (Magnetom Avanto (n = 16), Sonata (n = 15), or Symphony (n = 14), Siemens Healthcare) using a torso phased-array coil (six-element anterior and posterior coil arrays for the Avanto scanner and four-element anterior and posterior coil arrays for the Sonata and Symphony scanners). Sequences included axial turbospin-echo (TSE) T2-weighted imaging through the prostate and seminal vesicles (TR range/te range, 4,000 4,500/ ; slice thickness, 3 mm; no interslice gap; field of view (FOV), mm to mm; matrix, ; no parallel imaging, and 3 signal averages), axial TSE T1- weighted imaging through the prostate and seminal vesicles using identical slice location as the T2- weighted imaging ( /9 12; slice thickness, 3 mm; no interslice gap; FOV, mm to mm; matrix , generalized autocalibrating partially parallel acquisition (GRAP- PA) with parallel imaging factor of 2; and 2 signal averages), and axial fat-suppressed single-shot echo-planar DW images through the pelvis ( /72 78; b values of 0, 500, and 1000 s/mm 2, slice thickness, 3 mm; FOV, mm; matrix, ; GRAPPA with parallel imaging factor of 2; 6 8 signal averages). DW images were obtained with tridirectional motion-probing gradients and inline reconstruction of the ADC map. Other sequences performed as part of the routine prostate MRI protocol at our institution at the time of these examinations but not assessed in this study included sagittal and coronal TSE T2- weighted imaging sequences through the prostate and seminal vesicles, axial in- and-out-of-phase T1-weighted sequence through the entire pelvis, and dynamic contrast-enhanced sequences after IV administration of gadolinium chelate. Histologic Analysis All prostatectomy slides for each case were reviewed by two experienced pathologists in consensus with 20 and 12 years of experience in prostate pathology. The two pathologists used these slides to generate diagrams depicting the location of tumor foci within the PZ; these diagrams were used as tumor maps during the retrospective review of MRI findings in the patient cohort as described later. The largest focus of PZ tumor within each sextant (right and left base, right and left midgland, and right and left apex) was depicted, along with the tumor size and Gleason score. The pathologists were unaware of MRI findings. Image Interpretation A single observer with 2 years of experience in prostate MRI reviewed the MR images for all cases during four separate sessions, with a single session for each of the T2-weighted imaging, trace DWI b500 images, trace DWI b1,000 images, and the ADC map, incorporating 3-week intervals between sessions to minimize recall bias. Images were reviewed using a commercial workstation (Leonardo, Siemens Healthcare). The central gland was not assessed during these sessions. Assessment of Tumor Visibility During each session, the observer evaluated the single image set being reviewed in conjunction with the histologically defined tumor maps and the T1-weighted imaging. For each PZ tumor delineated on the tumor maps, the observer recorded whether a focal area of abnormal signal (low signal intensity (SI) on T2-weighted imaging, high SI on trace DWI b500 images, high SI on trace DWI b1,000 images, and low ADC value on the ADC map) was present within a similar region of the PZ on the given sequence. Areas of increased signal on T1-weighted imaging were considered to represent postbiopsy hemorrhage, and tumor foci were not regarded as visible if the signal abnormality was identified within an area of hemorrhage. Assessment of Tumor-To-PZ Contrast At the time of the assessments for tumor visibility, for each tumor that was recorded as visible on a given sequence, a region of interest (ROI) was placed over the tumor, with the mean value of the ROI recorded. The ROI was drawn manually around the periphery of the lesion to be as large as possible while excluding lesion margins. In addition to placing an ROI on visible tumor foci, an ROI was placed on benign PZ for each image set for each patient. For this purpose, nonhemorrhagic region measuring at least 1 cm 2 that showed homogeneous normal signal on the sequence under evaluation and that did not correspond with tumor on the tumor maps was selected. Using these values, the relative contrast between tumor and normal PZ (tumor-to-pz contrast ratio) was calculated for all visible tumor foci for all sequences with the formula [(SI PZ SI tumor ) / (SI PZ + SI tumor )]. This formula yields a larger value for increasing relative contrast between tissues, with a larger ratio representing superior tissue contrast. Normalization of T2 SI and ADC Values Given the use of different MRI scanners in this study with slightly varying acquisition parameters, normalization of T2 SI and ADC values was performed. For each case, the T2 SI of the right obturator internus muscle was obtained using a 2-cm 2 circular ROI and the ADC of the urine within the bladder was obtained using a 5-cm 2 ROI. Normalized T2 SI-to-muscle ratio and normalized ADCto-bladder ratio were then calculated for all tumor T2 SI and ADC measurements. Statistical Analysis Generalized estimating equations (GEEs) based on a binary logistic regression model were used to derive a 95% CI for the true percentage of times that tumor foci will be visible on each image set and to compare image sets with respect to these percentages. Mixed-model analysis of variance was used to compare image sets in terms of the relative contrast of visible tumor foci. Given greater visibility of tumor foci on trace DWI b1,000 images than on trace DWI b500 images, data pertaining to the trace DWI b1,000 images were then selected for a more detailed comparison with data from the ADC maps. Specifically, tumor foci were partitioned into four groups on the basis of whether they were visible on trace DWI b1,000 images but not on the ADC map, on the ADC map but not 124 AJR:196, January 2011

3 Tumor Visibility in Prostate Cancer on trace DWI b1,000 images, on both image sets, or on neither image set. Mixed-model analysis of variance was used to compare these four groups in terms of tumor size, normalized T2 signal intensity, ADC, and normalized ADC; GEE analysis was used to compare these four groups in terms of Gleason score (categorized as 6 or > 6 given limited variability in Gleason score). All reported p values are two-sided and considered statistically significant when < SAS version 9.0 (SAS Institute) was used for all computations. Results Pathologic Findings In the 45 patients, 132 foci of prostate cancer were marked on the tumor maps. These 132 tumor foci had a mean Gleason score of 6.6 (range, 6 9) and mean size of 10 mm (range, 3 28 mm). The T stages for the 45 patients based on prostatectomy were as follows: T2a (n = 6), T2c (n = 22), T3a (n = 16), and T3b (n = 1). Tumor Visibility Of the 132 tumor foci identified on prostatectomy, 106 (80.3%) were visible on T2- weighted imaging, 61 (46.2%) were visible on trace DWI b1,000 images, 35 (26.5%) were visible on trace DWI b500 images, and 82 (62.1%) were visible on the ADC map (Figs. 1 3). The differences in percentage of tumor foci that were visible on each image set were statistically significant for all combinations of image sets (p values ranging from < to ) (Table 1). Figure 4 shows the number of tumor foci visible on all possible combinations of image sets when considering T2-weighted imaging, trace DWI b1,000 imaging, and the ADC map, indicating that 115 (87%) tumor foci were visible on at least one of these three image sets. Tumor-to-PZ contrast Tumor-to-PZ contrast was greatest for T2-weighted imaging (0.411) and the ADC map (0.309) and lowest for trace DWI b1,000 images (0.119) and trace DWI b500 images (0.131). The differences in tumor-to-pz contrast were statistically significant for all combinations of image sets (p values < ) aside from the difference in tumor-to-pz contrast between trace DWI b1,000 images and trace DWI b500 images (0.6742) (Table 2). Comparison of trace DWI b1,000 images and the ADC map in terms of features of visible tumors The 132 tumor foci were classified into four groups of tumor foci as follows: 47 (35.6%) that were visible on both the ADC map and trace DWI b1,000 images, 35 (26.5%) that were visible on the ADC map but not on trace DWI b1,000 images, 14 (10.6%) that were visible on the trace DWI b1,000 images but not on the ADC map, and 36 (27.3%) that were not visible on either the trace DWI b1,000 images or the ADC map. There were trends toward a higher percentage of tumor foci with a Gleason score greater than 6 as well as a larger mean tumor size among those tumor foci that were visible on both trace DWI b1,000 images and the ADC map in comparison with the remaining groups of tumor foci. However, these differences reached statistical significance only for the comparison of Gleason score between this group and those tumor foci visible only on the ADC map (p value = 0.027) as well as for the comparison of tumor size between this group and those tumor foci not visible on either image set (p value = 0.002) but not for the remaining comparisons (p values ranging from to 0.933). There was no significant difference in normalized T2 signal A C Fig year-old man with prostate cancer. A D, Axial turbo spin-echo T2-weighted image (A), trace diffusion-weighted imaging (DWI) b500 image (B), trace DWI b1,000 image (C), and ADC map reconstructed from b values of 0, 500, and 1000 s/mm 2 (D). Right apical tumor focus (Gleason 3 + 3) that was identified at radical prostatectomy was visible on T2-weighted image (arrow, A), trace DWI b1,000 image (arrow, C), and ADC map (arrow, D) but was not visible on trace DWI b500 image (B). intensity between any of the four groups of tumor foci (p values ranging from to 0.715). There was no significant difference in ADC or normalized ADC when comparing tumor foci visible on both image sets with those visible only on the ADC map (p value of and 0.677, respectively) (Table 3). Discussion The primary aim of this study was to compare the visibility of prostate cancer on trace DW images and the ADC map. The clinical utility of such a comparison is indicated by the varying methods of interpretation of DWI of the prostate described in the literature to date. We observed that significantly more foci of PZ tumor that were identified on review of prostatectomy specimens were visible on the ADC map than on trace DW images. The decreased visibility of prostate cancer foci on trace DW images is possibly due to the persistent high signal within the normal PZ of the prostate on high b value B D AJR:196, January

4 Rosenkrantz et al. A C A C B D B D Fig year-old man with prostate cancer. A D, Axial turbo spin-echo (TSE) T2-weighted image (A), trace diffusion-weighted imaging (DWI) b500 image (B), trace DWI b1,000 image (C), and ADC map reconstructed from b values of 0, 500, and 1000 s/mm 2 (D). Right midgland tumor focus (Gleason 3 + 4) that was identified at radical prostatectomy is visible on axial TSE T2-weighted image (arrow, A) and ADC map (arrow, D) but not visible on trace DWI b500 image (B) or trace DWI b1,000 image (C). Fig year-old man with prostate cancer. A D, Axial turbo spin-echo (TSE) T2-weighted image (A), trace diffusion-weighted imaging (DWI) b500 image (B), trace DWI b1,000 image (C), and ADC map reconstructed from b values of 0, 500, and 1000 s/ mm 2 (D). Right apical tumor focus (Gleason 3 + 4) that was identified at radical prostatectomy is visible on axial TSE T2-weighted image (arrow, A) and ADC map (arrow, D) but not visible on trace DWI b500 image (B) or trace DWI b1,000 image (C). 126 AJR:196, January 2011

5 Tumor Visibility in Prostate Cancer TABLE 1: Tumor Foci Visible on Each Image Set Image Set Visible (%) DW images [18], thereby lessening the conspicuity of tumor foci, which themselves manifest as areas of increased signal on trace DW images. Indeed, we observed a statistically significant greater relative contrast between tumor and benign PZ for the ADC map than for trace DWI b1,000 images. Radiologists are familiar with the role of the ADC map in confirming that a lesion that is primarily detected on trace DW images exhibits true restricted diffusion rather than T2 shine-through [19]. However, the role of the ADC map as the primary image set to depict a lesion is less intuitive, as evidenced by studies showing greater utility for trace DW images than for the ADC map in the assessment of pathology in the brain [20], liver [16], and lung [15]. Our findings indicate the importance of the method of interpretation for DWI of the prostate in the clinical setting in that some tumor foci may not be detectable if attention is given solely to the trace DW images without specific assessment of the ADC map for primary identification of areas of cancer. p Value for Comparison With Trace DWI b500 Trace DWI b1,000 ADC Map T2-weighted imaging 80.3 (106/132) < < Trace DWI b (35/132) < Trace DWI b1, (61/132) ADC map 62.1 (82/132) < Note Data in parentheses are fraction of total. Values for p listed in bold are statistically significant. DWI = diffusion-weighted imaging, ADC = apparent diffusion coefficient. 17 (13%) 3 (2%) 11 (8%) Trace DWI b1000 T2-Weighted Imaging 19 (14%) 45 (34%) 2 (2%) 31 (23%) 4 (3%) ADC Map Fig. 4 Venn diagram illustrates total number and percentage of peripheral zone tumor foci identified at prostatectomy that were visible on all possible combinations of T2-weighted imaging, trace diffusion-weighted b 1,000 images, and ADC maps. Shimizu et al. [21] also performed an assessment of the visibility of prostate cancer when evaluating various MR image sets aware of histologic findings from prostatectomy. These authors used a torso phased-array coil at 1.5 T, as was done in our study. Unlike our findings, these authors reported nearly identical visibility of 113 foci of PZ cancer between trace DW images and the ADC map (50.4% and 52.2%, respectively). However, it is not clear that the image sets were interpreted during separate imaging sessions, and it is possible that the findings on one image set influenced the interpretation of the other set, leading to the observed similar visibility. Woodfield et al. [14] also reviewed DW images of the prostate aware of histologic findings and reported that 81 of 185 PZ tumor foci (44%) were visible on both DWI b1,000 images and the ADC map, slightly greater than the 36% of tumor foci that were visible on both image sets in our study. Unlike in our study, these authors used an endorectal coil and correlated imaging findings with results from transrectal ultrasound-guided biopsy rather than from prostatectomy; in addition, the number of tumor foci visible on only one of the two image sets is not reported. In a separate study, Kim et al. [22] reported that ADC values were measurable on 59 of 61 PZ tumor foci (96.7%) detected at prostatectomy. The greater visibility of PZ tumor in their study may be due to the much larger size of tumor foci in their sample (mean tumor size of 2.1 cm vs 1.0 cm) as well as the use of a higher field-strength magnet (3 T vs 1.5 T). We do note that the 62.1% of foci of PZ tumor that were visible on the ADC map in our current study is similar to the 68.2% of foci of PZ tumor that were visible on the ADC map in an earlier study from our institution in a separate cohort of patients [23]. However, our earlier study did not also include an assessment of trace DW images. One interesting finding in this study was that some tumor foci were visible solely on either the ADC map or the trace DWI b1,000 images (35 and 14 tumor foci, respectively). We were unable to identify differences in the features of tumors in these two groups, which showed nearly identical findings in terms of Gleason score, tumor size, and normalized T2 signal intensity. It is possible that histologic, metabolic, or molecular features of tumors [24, 25], not assessed in this study, may impact the relative visibility of tumor foci on these sets of images. In addition, the T2 shine-through effect seems unable to account for the visibility of some tumor foci on only the trace DW images in view of the generally T2-hypointense appearance of prostate cancer. Finally, although we noted tendencies for tumor foci visible on both trace DWI b1,000 images and the ADC map to have a larger size and higher Gleason score than tumor foci visible on only one of the two image sets, these trends did not consistently reach statistical significance. Although trace DWI b500 images showed nearly identical tumor-to-pz contrast among visible tumor foci as the trace DWI b1,000 images, the fraction of PZ tumor foci that were visible on this image set was poor at only 26.5%, significantly lower than the fraction that were visible on trace DWI b1,000 images. This poor visibility on the trace DWI b500 images may be attributed to the weak- TABLE 2: Tumor-to-Peripheral Zone (PZ) Contrast of Tumor Foci That Were Visible on Each Image Set Image Set Tumor-to-PZ Contrast (mean ± SD) p Value for Comparison With Trace DWI b500 Trace DWI b1,000 ADC Map T2-weighted imaging ± < < < Trace DWI b ± < Trace DWI b1, ± < ADC map ± < Note Values for p listed in bold are statistically significant. DWI = diffusion-weighted imaging, ADC = apparent diffusion coefficient. AJR:196, January

6 Rosenkrantz et al. TABLE 3: Comparison of Four Groups of Tumor Foci Visible on No. of Foci (%) Percentage With Gleason Score > 6 Size (mm) Normalized T2 Signal Intensity ADC ( 10 3 mm 2 /s) Normalized ADC ( 10 3 mm 2 /s) Both image sets 47 (36) ± ± ± ± 0.09 ADC map only 35 (27) ± ± ± ± 0.08 Trace DWI b1,000 only 14 (11) ± ± 0.98 Neither image set 36 (26) ± ± 1.30 Note DWI = diffusion-weighted imaging, ADC = apparent diffusion coefficient. Dashes = not applicable. er diffusion weighting of the lower b value image as well as the obscuring of some tumor foci by the greater residual signal within the benign PZ than is encountered when using a higher b value. It is difficult to find a previous study reporting a similar comparison of tumor visibility on trace DW images with varying b values. Kim et al. [22] reported greater sensitivity and accuracy for tumor detection with a b value of 1,000 s/mm 2 than for a b value of 2,000 s/mm 2 when performing prostate MRI at 3 T with a phased-array coil, whereas Van As et al. [26] reported the greatest difference between benign and malignant prostate tissue for a maximum b value of 300 s/mm 2 compared with higher b values when performing prostate MRI at 1.5 T with a phased-array coil. However, both of these studies compared the ADC maps produced from different b values and not the trace DW images themselves. Kitajima et al. [27] observed greater contrast-to-noise ratio (CNR) between benign and malignant PZ on trace DW images obtained with a b value of 1,000 s/mm 2 than with a b value of 2,000 s/ mm 2 when performing prostate MRI using a phased-array coil at 1.5 T. However, the absolute CNR values obtained using the two different b values were similar, and the actual visibility of tumor foci on the respective image sets was not directly assessed. Finally, the use of very high b values, even greater than the maximum b value of 1,000 s/mm 2 in this study, may be limited by poor signal-tonoise ratio that can lead to systematic errors in ADC determination given non-gaussian distribution of image noise and subsequent artifactual signal elevations on trace DW images that may introduce bias toward lower ADC values [17, 28]. In this study, tumor foci had the greatest visibility on T2-weighted imaging, with the vast majority of PZ tumors (80.3%) being detectable on this image set. Note that despite the high visibility of tumor foci on T2-weighted imaging, this sequence is limited by low specificity, given the numerous benign processes leading to low T2 signal within the PZ [1, 12, 29]. Such nonspecificity of areas of T2-hypointensity within the PZ contributes to the added value of functional MR sequences for tumor detection [30, 31]. We observed a very small number of tumor foci that were visible on trace DWI b1,000 images or on the ADC map but that were not visible on T2-weighted imaging. Gibbs et al. [32] observed no significant correlation between T2 relaxation times and ADC values in both benign and malignant peripheral zone regions. Their results suggest that these two parameters are influenced to some extent by distinct properties of prostate tissue, possibly accounting for the visibility of some tumor foci on trace DWI b1,000 images or the ADC map but not on T2-weighted imaging [32]. When T2-weighted imaging, trace DWI b1,000 images, and the ADC map were considered jointly, 87% of tumor foci were visible on at least one image set. This high tumor visibility between the three image sets supports the utility of a multiparametric approach to achieve maximal tumor detection with MRI, an aim that is of growing importance not just for the individual tailoring of conventional therapies such as surgery or radiation but also for emerging focal ablative therapies for which precise tumor localization can be critical [2, 4]. A number of limitations of this study warrant mention. First, because the imaging was reviewed in conjunction with histologically determined tumor maps, it is not possible from our data to show the specificity of abnormal areas within the PZ that may be detected on trace DW images and the ADC map when DWI of the prostate is performed in a clinical context without knowledge of histologic findings. Our intent was to show inherent differences on these two image sets regarding tumor visibility and contrast relative to benign PZ. For this purpose, direct correlation with histologic findings was important to ensure that our results truly reflect the features of PZ tumor foci on these image sets. Second, because only a single observer reviewed the MR images, interobserver variability for the visibility of tumor foci was not assessed. Third, we included all patients despite the presence of postbiopsy hemorrhage, which may have obscured some tumor foci. However, two prior studies indicate that hemorrhage tends to be less prevalent within regions of tumor [10, 23]. Fourth, because we only evaluated PZ tumors, possible differences in visibility of central gland tumors on trace DW images and the ADC map were not assessed. Finally, patients were imaged using a pelvic surface phased-array coil. Although an endorectal coil would have provided higher SNR, this approach has a number of limitations, including increased cost and examination time, nonuniform signal across the prostate, and increased motion artifact from rectal peristalsis. In conclusion, we observed that some foci of PZ tumor will be visible on either the trace DW images or the ADC map but not on the other image set. Of these two image sets, a significantly greater proportion of tumor foci were visible on the ADC map, and the ADC map exhibited significantly greater tumor-to-pz contrast. Therefore, in contrast to the methodology used in some previous studies, review of both the trace DW images and the ADC map may be useful for maximal tumor detection when performing DWI of the prostate. Moreover, unlike the approach that is often applied for interpreting DWI in other organs, careful attention should be given to the ADC map as a primary means for tumor detection when interpreting DWI of the prostate. References 1. Scheidler J, Hricak H, Vigneron DB, et al. Prostate cancer: localization with three-dimensional proton MR spectroscopic imaging clinicopathologic study. Radiology 1999; 213: Turkbey B, Albert PS, Kurdziel K, Choyke PL. Imaging localized prostate cancer: current approaches and new developments. AJR 2009; 192: AJR:196, January 2011

7 Tumor Visibility in Prostate Cancer 3. Kelloff GJ, Choyke P, Coffey DS. Challenges in clinical prostate cancer: role of imaging. AJR 2009; 192: Hricak H, Choyke PL, Eberhardt SC, Leibel SA, Scardino PT. Imaging prostate cancer: a multidisciplinary perspective. Radiology 2007; 243: Miao H, Fukatsu H, Ishigaki T. Prostate cancer detection with 3-T MRI: comparison of diffusionweighted and T2-weighted imaging. Eur J Radiol 2007; 61: Haider MA, van der Kwast TH, Tanguay J, et al. Combined T2-weighted and diffusion-weighted MRI for localization of prostate cancer. AJR 2007; 189: Kim CK, Park BK, Lee HM, Kwon GY. Value of diffusion-weighted imaging for the prediction of prostate cancer location at 3T using a phased-array coil: preliminary results. Invest Radiol 2007; 42: Xu J, Humphrey PA, Kibel AS, et al. Magnetic resonance diffusion characteristics of histologically defined prostate cancer in humans. Magn Reson Med 2009; 61: Riches SF, Payne GS, Morgan VA, et al. MRI in the detection of prostate cancer: combined apparent diffusion coefficient, metabolite ratio, and vascular parameters. AJR 2009; 193: Tamada T, Sone T, Jo Y, et al. Prostate cancer: relationships between postbiopsy hemorrhage and tumor detectability at MR diagnosis. Radiology 2008; 248: Morgan VA, Kyriazi S, Ashley SE, DeSouza NM. Evaluation of the potential of diffusion-weighted imaging in prostate cancer detection. Acta Radiol 2007; 48: Lim HK, Kim JK, Kim KA, Cho KS. Prostate cancer: apparent diffusion coefficient map with T2-weighted images for detection a multireader study. Radiology 2009; 250: Kitajima K, Kaji Y, Fukabori Y, Yoshida K, Suganuma N, Sugimura K. Prostate cancer detection with 3 T MRI: Comparison of diffusion-weighted imaging and dynamic contrast-enhanced MRI in combination with T2-weighted imaging. J Magn Reson Imaging 2010; 31: Woodfield CA, Tung GA, Grand DJ, Pezzullo JA, Machan JT, Renzulli JF 2nd. Diffusion-weighted MRI of peripheral zone prostate cancer: comparison of tumor apparent diffusion coefficient with Gleason score and percentage of tumor on core biopsy. AJR 2010; 194:1013; [web]w316 W Uto T, Takehara Y, Nakamura Y, et al. Higher sensitivity and specificity for diffusion-weighted imaging of malignant lung lesions without apparent diffusion coefficient quantification. Radiology 2009; 252: Vandecaveye V, De Keyzer F, Verslype C, et al. Diffusion-weighted MRI provides additional value to conventional dynamic contrast-enhanced MRI for detection of hepatocellular carcinoma. Eur Radiol 2009; 19: Dietrich O, Heiland S, Sartor K. Noise correction for the exact determination of apparent diffusion coefficients at low SNR. Magn Reson Med 2001; 45: Saremi F, Knoll AN, Bendavid OJ, Schultze- Haakh H, Narula N, Sarlati F. Characterization of genitourinary lesions with diffusion-weighted imaging. RadioGraphics 2009; 29: Koh DM, Collins DJ. Diffusion-weighted MRI in the body: applications and challenges in oncology. AJR 2007; 188: Chong J, Lu D, Aragao F, et al. Diffusion-weighted MR of acute cerebral infarction: comparison of data processing methods. AJNR 1998; 19: Shimizu T, Nishie A, Ro T, et al. Prostate cancer detection: the value of performing an MRI before a biopsy. Acta Radiol 2009; 27: Kim CK, Park BK, Kim B. High-b-value diffusion-weighted imaging at 3 T to detect prostate cancer: comparisons between b values of 1,000 and 2,000 s/mm2. AJR 2010; 194:172; [web]w33 W Rosenkrantz AB, Kopec M, Kong X, et al. Prostate cancer vs. post-biopsy hemorrhage: diagnosis with T2- and diffusion-weighted imaging. J Magn Reson Imaging 2010; 6: Shukla-Dave A, Hricak H, Ishill NM, et al. Correlation of MR imaging and MR spectroscopic imaging findings with Ki-67, phospho-akt, and androgen receptor expression in prostate cancer. Radiology 2009; 250: Langer DL, van der Kwast TH, Evans AJ, et al. Intermixed normal tissue within prostate cancer: effect on MR imaging measurements of apparent diffusion coefficient and T2 sparse versus dense cancers. Radiology 2008; 249: Van As N, Charles-Edwards E, Jackson A, et al. Correlation of diffusion-weighted MRI with whole mount radical prostatectomy specimens. Br J Radiol 2008; 81: Kitajima K, Kaji Y, Kuroda K, Sugimura K. High b-value diffusion-weighted imaging in normal and malignant peripheral zone tissue of the prostate: effect of signal-to-noise ratio. Magn Reson Med Sci 2008; 7: Kristoffersen A. Optimal estimation of the diffusion coefficient from non-averaged and averaged noisy magnitude data. J Magn Reson 2007; 187: Casciani E, Polettini E, Bertini L, et al. Prostate cancer: evaluation with endorectal MR imaging and three-dimensional proton MR spectroscopic imaging. Radiol Med 2004; 108: Choi YJ, Kim JK, Kim N, Kim KW, Choi EK, Cho KS. Functional MR imaging of prostate cancer. RadioGraphics 2007; 27:63 75; discussion, Seitz M, Shukla-Dave A, Bjartell A, et al. Functional magnetic resonance imaging in prostate cancer. Eur Urol 2009; 55: Gibbs P, Tozer DJ, Liney GP, Turnbull LW. Comparison of quantitative T2 mapping and diffusionweighted imaging in the normal and pathologic prostate. Magn Reson Med 2001; 46: AJR:196, January