Normal Central Zone of the Prostate and Central Zone Involvement by Prostate Cancer: Clinical and MR Imaging Implications

The central zone of the prostate can be distinguished from adjacent zones in most patients; detailed knowledge of the zonal anatomy and imaging features of the central zone could aid avoidance of pitfalls and accurate interpretation of prostate MR images.

Abstract

Purpose:

To describe the anatomic features of the central zone of the prostate on T2-weighted and diffusion-weighted (DW) magnetic resonance (MR) images and evaluate the diagnostic performance of MR imaging in detection of central zone involvement by prostate cancer.

Materials and Methods:

The institutional review board waived informed consent and approved this retrospective, HIPAA-compliant study of 211 patients who underwent T2-weighted and DW MR imaging of the prostate before radical prostatectomy. Whole-mount step-section pathologic findings were the reference standard. Two radiologists independently recorded the visibility, MR signal intensity, size, and symmetry of the central zone and scored the likelihood of central zone involvement by cancer on T2-weighted MR images and on T2-weighted MR images plus apparent diffusion coefficient (ADC) maps generated from the DW MR images. Descriptive summary statistics were calculated for central zone imaging features. Sensitivity, specificity, and area under the curve were used to evaluate reader performance in detecting central zone involvement.

Results:

For readers 1 and 2, the central zone was visible, at least partially, in 177 (84%) and 170 (81%) of 211 patients, respectively. The most common imaging appearance of the central zone was symmetric, homogeneous low signal intensity. Cancers involving the central zone had higher prostate-specific antigen values, Gleason scores, and rates of extracapsular extension and seminal vesicle invasion compared with cancers not involving the central zone (P < .05). Area under the curve, sensitivity, and specificity in detecting central zone involvement were 0.70, 0.30, and 0.96 for reader 1 and 0.65, 0.35, and 0.93 for reader 2, and these values did not differ significantly between T2-weighted imaging and T2-weighted imaging plus ADC maps.

Conclusion:

The central zone was visualized in most patients. Cancers involving the central zone were associated with more aggressive disease than those without central zone involvement.

© RSNA, 2012

Introduction

The contemporary description of the prostatic anatomy is based on the classic work by McNeal (1), who proposed the concept of prostatic “zones,” based on their anatomic location and histologic and embryologic features. The seminal vesicles, located posterosuperiorly to the prostate, drain into the mid prostatic urethra via the ejaculatory ducts, in the region of the verumontanum. The transition zone forms two bulges on either side of the urethra that extend superiorly, anteriorly, and laterally from the verumontanum. The central zone surrounds the ejaculatory ducts and is located posterior to the transition zone and the urethra proximal to the verumontanum. The peripheral zone surrounds both the central zone and the distal prostatic urethra (Fig 1). Anatomic studies from the early days of prostate magnetic resonance (MR) imaging showed that distinct imaging features allowed the three prostate zones to be identified on T2-weighted MR images (2,3). While the distinction between the peripheral zone and the remaining prostate was immediately accepted, there was a lack of consensus about whether the transition and central zones could be identified separately on MR images, and for a number of years, these two zones sometimes have been grouped under the term “central gland.” With the recognition that up to 30% of prostate cancers occur outside the peripheral zone (4–6), there has been increasing interest in defining the imaging features of the transition zone (7,8) and the “central gland” (9–18). However, to our knowledge, no imaging studies have addressed the importance of the central zone in prostate cancer.

Figure 1a:

(a) Oblique sagittal and (b) axial schematic representations of the prostate zones and their relationship to the prostatic urethra (white arrow) and ejaculatory ducts (black arrow). Central zone (red), transition zone (blue), peripheral zone (yellow) and anterior fibromuscular stroma (green) are shown.

Figure 1b:

(a) Oblique sagittal and (b) axial schematic representations of the prostate zones and their relationship to the prostatic urethra (white arrow) and ejaculatory ducts (black arrow). Central zone (red), transition zone (blue), peripheral zone (yellow) and anterior fibromuscular stroma (green) are shown.

The central zone has histologic features more akin to those of the ejaculatory ducts and seminal vesicles, suggesting it is a wolffian duct derivate, while the remaining prostate (including the transition and peripheral zones) derives from the urogenital sinus (19). This issue becomes particularly relevant when attempting to gain insight into prostate biology and metabolism at a molecular level through functional imaging techniques such as diffusion-weighted (DW) MR imaging, MR spectroscopy, or dynamic contrast material–enhanced MR imaging, as differences in the quantitative parameters obtained may solely represent sampling errors caused by grouping measurements from two embryologically and histologically distinct zones into one “central gland.” Perhaps even more important, pathologic studies have shown that malignant tumors involving the central zone have extremely different biological behavior and outcomes than do tumors arising in other prostatic zones (20,21).

Thus, the purpose of this study was to describe the anatomic features of the central zone of the prostate on T2-weighted and DW MR images and evaluate the diagnostic performance of MR imaging in the detection of central zone involvement by prostate cancer. In the process, we also aimed to analyze potential pitfalls in MR imaging interpretation related to the central zone.

Materials and Methods

The institutional review board approved our retrospective study and waived the informed consent requirement. Our study was compliant with the Health Insurance Portability and Accountability Act.

Eligibility Criteria and Patient Characteristics

The inclusion criteria for our study were as follows: (a) patients who underwent endorectal prostate MR imaging, including DW MR imaging, performed at our institution for the assessment of prostate cancer between April 2008 and October 2009; (b) patients who underwent radical prostatectomy performed at our institution within 6 months after MR imaging; (c) patients in whom whole-mount step-section pathologic tumor maps were available. Two hundred twenty-seven patients satisfied these inclusion criteria. We excluded patients with (a) prior prostate cancer treatment, including surgery, focal therapy, hormones, or radiation therapy (n = 5); and (b) artifacts on DW MR images, which made the study results nondiagnostic (n = 11). Thus, our study included a total of 211 patients. A patient selection flowchart is shown in Figure 2. Patients’ clinical and histopathologic characteristics are summarized in Table 1.

Figure 2:

Patient selection flowchart.

Table 1.

Patient Characteristics

^*Numbers in parentheses are ranges.

^†To convert to Système International units in micrograms per liter, multiply by 1.0. Numbers in parentheses are ranges.

^‡Numbers in parentheses are percentages.

MR Image Acquisition, Analysis, and Interpretation

MR imaging studies were performed with a 1.5-T (168 patients) or 3-T (43 patients) whole-body unit (GE Medical Systems, Milwaukee, Wis). A body coil was used for excitation; a pelvic four-channel phased-array coil and a balloon-covered expandable endorectal coil (Medrad, Warrendale, Pa) were used for signal reception. The anatomic images were obtained by using the following sequences: transverse T1-weighted imaging (repetition time msec/echo time msec, 500–750/10–14 at 1.5 T and 400–650/10–14 at 3 T; section thickness, 5 mm; intersection gap, 1 mm; field of view, 28–36 cm; matrix, 256 3 192); transverse, coronal, and sagittal T2-weighted fast spin-echo imaging (4500–6000/120 [effective] at 1.5 T and 3500–5000/100 at 3 T; section thickness, 3 mm; no intersection gap; field of view, 12–14 cm; and matrix, 256 3 192). DW MR imaging was performed in the transverse plane with orientation and location identical to those prescribed for the transverse T2-weighted anatomic images by using a spin-echo echo-planar imaging sequence with a pair of rectangular gradient pulses along three orthogonal axes and the following: 3300–3600/60–110 [effective]; field of view, 14 cm; section thickness, 3 mm; no intersection gap; in-plane resolution, 1.9 3 1.9 mm; and b values of 0, 400, 700, and 1000 sec/mm². Parametric image maps that were based on the apparent diffusion coefficient (ADC) were generated by using a workstation (Advanced Workstation; GE Medical Systems).

Two radiologists retrospectively and independently interpreted the results of MR imaging studies, which were archived in a picture archiving and communication system (Centricity; GE Medical Systems). Reader 1 (H.A.V.) was a body imaging fellow with a focus of interest in prostate MR imaging and 2 years of experience in prostate MR imaging. Reader 2 (T.F.) was a radiologist with 4 years of experience in interpreting prostate MR images. Although the readers were aware that the patients had prostate cancer, they were blinded to the cancer location, clinical and laboratory findings (including prostate-specific antigen values), and histologic and imaging findings. On MR images, the central zone was defined anatomically as the region of the prostate surrounding the ejaculatory ducts from the prostatic base to the verumontanum (Fig 3) (4). The verumontanum was defined as the point where the two ejaculatory ducts reach the prostatic urethra as visualized in the axial plane (4). The readers first recorded whether they could identify the location of the verumontanum in the axial plane. Next, they determined whether they could distinguish the central zone from the remaining prostate on the basis of its signal intensity in any plane of T2-weighted imaging. They recorded the results on a three-point scale, as follows: zero, central zone not separately visualized; one, central zone partially visualized in the base of the prostate; and two, central zone visualized throughout its length from the base of the prostate to the verumontanum. When they were able to see the central zone, the radiologists then recorded the following imaging features on the basis of subjective visual assessment: maximum dimension of each side of the central zone (measured in the coronal plane), symmetry between the right and left sides of the central zone (assessed in the coronal plane), signal intensity of the central zone on a T2-weighted image and ADC map compared with the adjacent peripheral zone (homogeneously low, homogeneously intermediate, homogeneously high, and heterogeneous), and clear demarcation from the adjacent prostate zones. In addition, the central zone was divided into right and left sides (separated by the midline through the urethra and verumontanum in the coronal plane), and readers independently assigned scores for the likelihood of cancer on an index scale of one to five, as follows: score 1, definitely absent; score 2, probably absent; score 3, indeterminate; score 4, probably present; and score 5, definitely present. First, they assigned scores on the basis of the interpretation of T2-weighted images alone. In the same session, they then evaluated each region by using a combination of T2-weighted and ADC maps and rescored the images by using the same scale (Fig 4). Coregistration with the T2-weighted and anatomic landmarks (eg, ejaculatory ducts) were used to identify the central zone on the ADC maps.

Figure 3a:

(a) Axial and (b) coronal T2-weighted MR images (4750/119) in a 64-year-old patient with prostate cancer showing the typical homogeneous low signal intensity and symmetrical appearance on either side of ejaculatory ducts (dashed arrow on a) of the central zone (arrows). B = bladder, ERC = endorectal coil, PZ = peripheral zone.

Figure 4a:

(a) Axial and (b) coronal T2-weighted MR images (3500/100.7) and (c) ADC map (4000/100.1; b values were 0 and 700 sec/mm²) obtained at 1.5 T in a 58-year-old patient with prostate cancer demonstrate areas suspicious for cancer involving the right central zone (arrows), adjacent to the ejaculatory ducts (dashed arrow). (d) Representative image from step-section pathologic map demonstrates multifocal prostate cancer (green and black lines), with the dominant tumor involving the right central zone and corresponding to the abnormality on MR images. The location of a benign hyperplastic nodule (star) in the right transition zone is shown for anatomic correlation.

Figure 3b:

(a) Axial and (b) coronal T2-weighted MR images (4750/119) in a 64-year-old patient with prostate cancer showing the typical homogeneous low signal intensity and symmetrical appearance on either side of ejaculatory ducts (dashed arrow on a) of the central zone (arrows). B = bladder, ERC = endorectal coil, PZ = peripheral zone.

Figure 4b:

(a) Axial and (b) coronal T2-weighted MR images (3500/100.7) and (c) ADC map (4000/100.1; b values were 0 and 700 sec/mm²) obtained at 1.5 T in a 58-year-old patient with prostate cancer demonstrate areas suspicious for cancer involving the right central zone (arrows), adjacent to the ejaculatory ducts (dashed arrow). (d) Representative image from step-section pathologic map demonstrates multifocal prostate cancer (green and black lines), with the dominant tumor involving the right central zone and corresponding to the abnormality on MR images. The location of a benign hyperplastic nodule (star) in the right transition zone is shown for anatomic correlation.

Figure 4c:

(a) Axial and (b) coronal T2-weighted MR images (3500/100.7) and (c) ADC map (4000/100.1; b values were 0 and 700 sec/mm²) obtained at 1.5 T in a 58-year-old patient with prostate cancer demonstrate areas suspicious for cancer involving the right central zone (arrows), adjacent to the ejaculatory ducts (dashed arrow). (d) Representative image from step-section pathologic map demonstrates multifocal prostate cancer (green and black lines), with the dominant tumor involving the right central zone and corresponding to the abnormality on MR images. The location of a benign hyperplastic nodule (star) in the right transition zone is shown for anatomic correlation.

Figure 4d:

(a) Axial and (b) coronal T2-weighted MR images (3500/100.7) and (c) ADC map (4000/100.1; b values were 0 and 700 sec/mm²) obtained at 1.5 T in a 58-year-old patient with prostate cancer demonstrate areas suspicious for cancer involving the right central zone (arrows), adjacent to the ejaculatory ducts (dashed arrow). (d) Representative image from step-section pathologic map demonstrates multifocal prostate cancer (green and black lines), with the dominant tumor involving the right central zone and corresponding to the abnormality on MR images. The location of a benign hyperplastic nodule (star) in the right transition zone is shown for anatomic correlation.

Histopathologic Analysis and Image Correlation

Prostatectomy specimens were sliced from apex to base at 3–4-mm intervals. The distal 5-mm portion of the apex was amputated and coned. The seminal vesicles were amputated and submitted separately. After paraffin embedding, microslices were placed on glass slides and stained with hematoxylin-eosin. A pathology fellow (K.U.) with 3 years of experience in genitourinary pathology, supervised by the faculty genitourinary pathologist, retrospectively evaluated all the pathologic specimens and outlined the central zone on the whole-mount maps, taking into account the location surrounding the ejaculatory ducts and typical histologic features such as large-caliber glands with a “Roman bridge” architecture, prominent basal cell layer, and tall columnar cells, with eosinophilic cytoplasm lacking prominent nucleoli (22). Cancer involving the central zone was defined as cancer foci involving the expected location of the central zone around the ejaculatory ducts and at the base of the prostate. Other pathologic criteria reported in the literature for aiding the diagnosis of “primary” central zone cancer were not used in our study; for example, we did not use the presence of high-grade prostatic intraepithelial neoplasia as a precursor lesion and at least 80% of the tumor occupying the central zone, as we aimed to evaluate central zone “involvement” by cancer rather than primary central zone cancer. Correlation between pathologic findings and MR imaging findings was performed in consensus by three of the authors (K.U., H.A.V., T.F.). The central zone outlined on the whole-mount pathologic maps was evaluated in conjunction with the MR images to establish the location and visibility of the central zone on MR images by using anatomic landmarks such as the urethra, verumontanum, and ejaculatory ducts.

Statistical Analysis

The median and range were calculated for each continuous variable. The Wilcoxon rank-sum test (continuous variables) and the Fisher exact test (categorical variables) were used to examine associations between the two groups (patients with central zone involvement and patients without central zone involvement). The frequency and percentage of patients in whom the central zone was visualized separately from the remaining prostatic zones were calculated, and the corresponding MR imaging features were summarized according to each reader. Reader accuracy in identifying patients with central zone involvement by cancer was assessed by calculating sensitivity and specificity, along with 95% exact Clopper-Pearson binomial confidence intervals (CIs). For this purpose the scoring scale of one to five was dichotomized according to two different cutoff points. First, scores 1 and 2 were considered negative and scores 3–5 were considered positive for central zone involvement. Then, sensitivity and specificities were again calculated considering scores 1–3 as negative and scores 4 and 5 as positive for central zone involvement. To evaluate reader accuracy in determining the presence of central zone involvement by cancer, the area under the curve of the empirical receiver operating characteristic and the corresponding 95% CIs were estimated nonparametrically by using the method proposed by Obuchowski (23), accounting for multiple assessments per patient. To evaluate agreement between the two readers, the Cohen k coefficient, along with the standard error, was calculated and interpreted as follows: slight agreement, less than 0.20; fair agreement, 0.20–0.40; moderate agreement, 0.41–0.60; substantial agreement, 0.61–0.80; and almost perfect agreement, 0.81–1.00. P values < .05 were considered to indicate a significant difference. All statistical analyses were performed with software (Stata 11; StataCorp, College Station, Tex).

Results

Central Zone Visibility

The verumontanum could be identified on axial T2-weighted images in the mid urethra of 197 (93%) and 196 (93%) of 211 patients according to readers 1 and 2, respectively (k = 0.64; standard error, 0.06). According to reader 1, the central zone was visible in 177 of 211 (84%) patients; in 117 of 177 (66%) patients, it was visible in the region of the prostatic base only, and in 60 of 177 (34%), it was visible throughout its length (from the prostatic base to the verumontanum). According to reader 2, the central zone was visible in 170 of 211 (81%) patients; it was visible at the level of the prostatic base only in 78 of 170 (46%) patients and throughout its length (from the prostatic base to the verumontanum) in 92 of 170 (54%) patients (Fig 3). Interreader agreement for central zone visibility was substantial (k = 0.64; standard error, 0.06).

Central Zone Characteristics

The median values for the maximum dimensions of the right and left sides of the central zone on coronal T2-weighted images were 6 × 5 mm for reader 1 and 6 × 6 mm for reader 2. The two sides of the central zone appeared symmetrical in 142 of 177 (80%) patients according to reader 1 and in 143 of 170 (84%) patients according to reader 2 (k = 0.56; standard error, 0.07) (Fig 3). The signal intensity of the central zone on T2-weighted images and ADC maps was homogeneously low in 121 of 177 (68%) and 146 of 170 (86%) patients according to readers 1 and 2, respectively (Fig 3). Neither reader classified the central zone as having homogeneous high signal intensity. Heterogeneous signal intensity was identified in the central zone of a small proportion of patients (nine of 177 [5%] for reader 1 and eight of 170 [5%] for reader 2). There were no patients in whom the signal intensity of the central zone differed between the T2-weighted images and the ADC maps. The central zone was clearly demarcated from the adjacent peripheral and transition zones in 141 of 177 (80%) and 120 of 170 (71%) patients according to readers 1 and 2, respectively (k = 0.60; standard error, 0.07). Imaging features of the central zone, along with interreader agreement and standard error, are shown in Table 2.

Table 2.

MR Imaging Features of the Central Zone

^*Numbers in parentheses are percentages.

^†Numbers in parentheses are standard errors.

Central Zone Involvement by Cancer

The central zone was involved by cancer in 14 (7%) of 211 patients (Table 1). There were significant differences in prostate-specific antigen levels (P = .04), prostatectomy Gleason score (P = .001), extracapsular extension (P < .001), and seminal vesicle invasion (P < .001) between patients with and without central zone involvement (Table 1). Area under the curve for detecting central zone involvement with T2-weighted imaging and T2-weighted imaging plus ADC maps (Table 3) were 0.70 (95% CI: 0.59, 0.81) and 0.73 (95% CI: 0.61, 0.84), respectively, for reader 1 and 0.65 (95% CI: 0.53, 0.77) and 0.65 (95% CI: 0.52, 0.77), respectively, for reader 2 (Fig 5). Depending on the cutoff from the suspicion index scale used, sensitivity varied between 0.20 and 0.45 and specificity varied between 0.93 and 0.99 (Table 3). No significant differences in area under the curve, sensitivity, specificity, positive predictive value, or negative predictive value were present between assessments made with T2-weighted imaging and T2-weighted imaging plus ADC maps for either reader (P = .25 to > .99). Figure 6 shows similarity in T2 signal intensity of the normal central zone and prostate cancer foci.

Table 3.

Accuracy in Detection of Central Zone Involvement by Prostate Cancer

Note.—Cutoff A refers to a dichotomized scoring scale, with scores 1 and 2 considered negative and scores 3–5 considered positive for central zone involvement. Cutoff B refers to a dichotomized scoring scale, with scores 1–3 considered negative and scores 4 and 5 considered positive for central zone involvement. Numbers in parentheses are 95% CIs. T2 = T2-weighted imaging

Figure 5a:

Receiver operating characteristic curves and areas under the curves for (a) reader 1 and (b) reader 2 in detecting central zone involvement by prostate cancer with use of T2-weighted images (T2WI) with and without qualitative assessment with DW MR imaging.

Figure 6a:

(a) Axial and (b) coronal T2-weighted MR mages (3150/105.4) in a 57-year-old patient with multifocal peripheral zone prostate cancer demonstrate similarity in T2 signal intensity of the normal central zone (arrows) and prostate cancer foci (dashed arrows).

Figure 5b:

Receiver operating characteristic curves and areas under the curves for (a) reader 1 and (b) reader 2 in detecting central zone involvement by prostate cancer with use of T2-weighted images (T2WI) with and without qualitative assessment with DW MR imaging.

Figure 6b:

(a) Axial and (b) coronal T2-weighted MR mages (3150/105.4) in a 57-year-old patient with multifocal peripheral zone prostate cancer demonstrate similarity in T2 signal intensity of the normal central zone (arrows) and prostate cancer foci (dashed arrows).

Discussion

The introduction of MR imaging into clinical practice in the 1980s for the first time allowed detailed in vivo visualization of the prostate zonal anatomy, originally described by McNeal (1) decades earlier. Researchers in one of the first studies ever published in the prostate MR imaging literature reported that, on T2-weighted images obtained with use of a 0.3-T system, the central zone was identified in 31 of 32 men aged 25 to 35 years but only in eight of 23 men aged 40 years or older (2). Despite constant refinements and improvements in MR imaging technology, no studies in which investigators reevaluated the anatomy of the central zone have followed. However, the notion that the central and transition zones of the prostate cannot be visualized separately on MR images, at least in patients who undergo imaging for the assessment of prostate cancer, remains fairly widespread. While some investigators have focused on the MR imaging features of cancer in the transition zone (7,8), many others have grouped the central and transition zones as the “central gland” when reporting their findings (9–18). In this study, we showed that, in a population undergoing MR imaging for the assessment of prostate cancer (mean age, 60 years), the central zone was visible, at least partially, in 81%–84% of patients. This suggests that when imaging findings in the central gland are evaluated, an attempt should be made to identify the central and transition zones separately. To aid identification of the central zone, we report its most common features on images, namely its location surrounding the ejaculatory ducts from the prostatic base to the verumontanum (the location of the verumontanum was identified on MR images by both readers independently in 93% of the patients in this study), its symmetrical appearance on either side of the verumontanum on coronal T2-weighted images, and its homogeneous low signal intensity on T2-weighted images and ADC maps. The central zone was better visualized in the base of the prostate compared with the midgland in the region of the verumontanum. There are several reasons for this finding: The central zone comprises the majority of the total prostatic tissue at the level of the base; hence, it is not surprising that this is the area where it is best visualized. Furthermore, because the transition zone originates on either side of the urethra and extends superiorly from the verumontanum, the central zone may become more compressed and less well visualized at this level, particularly in the presence of benign prostatic hyperplasia in the transition zone.

The normal central zone is characterized histologically by the presence of a prominent basal cell layer, tall columnar cells with eosinophilic cytoplasm that lack prominent nucleoli (22). The glands of the central zone are usually of larger caliber than those of the peripheral zone and show a Roman bridge architecture with the formation of intraglandular lacunae and compact stroma, features that may explain the typical homogeneous low signal intensity on T2-weighted images observed in this study. Although the normal central zone is morphologically distinct from normal peripheral zone tissue, involvement of the central zone by prostate cancer is not routinely reported by all pathologists who evaluate prostatectomy specimens (21). This finding may be partially due to the fact that tumors arising in the central zone are usually indistinguishable from peripheral zone tumors of similar histologic grade; thus, differentiation with regard to zonal origin relies on the identification of morphologic features in the normal prostate tissue adjacent to the tumor (21). However, in accordance with data in previous studies from the pathology literature (20,21), our data suggest that tumors involving the central zone—tumors demonstrating higher prostate-specific antigen values, Gleason scores, and rates of extracapsular extension and seminal vesicle invasion—tend to be more aggressive than tumors that do not involve the central zone. Therefore, it may be prudent to routinely alert referring clinicians to central zone involvement when it is identified on MR images, particularly given the high specificity achieved by the two readers in this study (0.93–0.99), albeit with a relatively low positive predictive value (0.19–0.45), which is expected given the low prevalence of central zone involvement of prostate cancer.

Given that both prostate cancer and normal central zone typically display homogeneous low signal intensity on T2-weighted images and ADC maps, it is important for the radiologist to recognize the normal central zone to avoid a false-positive diagnosis of cancer. This is a particular problem in regard to findings in the posterior aspect of the prostatic base, which is composed predominantly of central zone, and may explain why MR imaging interpretation of findings in the base of the prostate often poses a challenge even to experienced genitourinary radiologists. This pitfall is reflected in the relatively low sensitivity (0.20–0.35) and area under the receiver operating characteristic curve (0.65–0.73) we found for the detection of central zone involvement by prostate cancer. DW MR imaging was noncontributory, as we observed no significant difference in accuracy when the T2-weighted images were interpreted alone or in conjunction with ADC maps. The use of other MR techniques such as MR spectroscopy and dynamic contrast-enhanced MR imaging for the assessment of the central zone has not been reported, although MR spectroscopy is unlikely to be helpful in this setting, as metabolite contamination from the adjacent seminal vesicles is known to hinder interpretation of findings in the prostate base (24).

Our study had several limitations. First, it was vulnerable to the inherent disadvantages of its retrospective design. Second, a degree of selection bias was introduced by including only patients with prostate cancer who underwent radical prostatectomy, thus influencing the generalizability of our findings to the general population. Although it is well known that prostate volume and the relative proportion occupied by each prostatic zone vary with age, prostate imaging is performed almost exclusively in patients with suspected or confirmed prostate cancer, and therefore it is probably in this patient population that the characteristics of the different prostatic zones are most clinically relevant. Third, although interreader agreement for central zone visibility was substantial (k = 0.64), interreader agreement on central zone features including symmetry, T2 signal intensity, and clear demarcation from adjacent peripheral and transition zones was only moderate (k = 0.41–0.60), suggesting that these observations may be quite challenging to make at times. Fourth, to maximize the number of eligible patients, we included MR imaging examinations performed at both 1.5 and 3 T. The effects of higher magnetic field strength on diagnostic accuracy of MR imaging in prostate cancer have not been clearly established (25). Fifth, although distortions in the prostate zonal anatomy, size, and shape caused in vivo by the presence of the endorectal coil and caused ex vivo by the preparation of the whole-mount pathologic specimen (eg, tissue shrinkage during fixation) were subjectively accounted for during correlation of pathologic and MR imaging findings, the influence of these factors on the visibility of the prostate zonal anatomy on MR images remains unknown. Finally, our aim was not to evaluate the capability of MR imaging to depict primary central zone cancer. Instead, we evaluated patients with central zone involvement by cancer, even if the exact location from which the tumor arose could not be confirmed. This likely accounts for the higher percentage of cancers involving the central zone in this study (7%) compared with that in the pathology literature (approximately 3%) (20,21). This low prevalence of central zone involvement by cancer (14 of 211 patients) in the study population also limits the power of our analysis. However, our results also support an association between central zone involvement by prostate cancer and increased cancer aggressiveness.

In summary, on MR images, the central zone of the prostate could be distinguished from the adjacent prostate zones in most patients, and prostate cancers involving the central zone were associated with more aggressive disease than those without central zone involvement. Detailed knowledge of the zonal anatomy and imaging features of the central zone could aid avoidance of pitfalls and accurate interpretation of prostate MR images.

Advances in Knowledge.

• The central zone of the prostate could be distinguished from the adjacent prostate zones by two independent readers in 170 and 177 (81% and 84%, respectively) of 211 patients undergoing MR imaging for the evaluation of prostate cancer.

• The most common MR imaging features of the central zone were homogeneous low signal intensity on T2-weighted images and apparent diffusion coefficient maps (121 of 177 [68%] and 146 of 170 [86%] of patients according to two independent readers) and symmetrical appearance on either side of the verumontanum (142 of 177 [80%] and 143 of 170 [84%] of patients according to two independent readers).

• Area under the curve, sensitivity, and specificity in detecting central zone involvement by cancer were 0.70, 0.30, and 0.96 for reader 1 and 0.65, 0.35, and 0.93 for reader 2, and these values did not differ significantly between T2-weighted imaging and T2-weighted imaging plus apparent diffusion coefficient maps.

• There were significant differences in prostate-specific antigen level (P = .04), prostatectomy Gleason score (P = .001), extracapsular extension (P < .001), and seminal vesicle invasion (P < .001) between patients with and without central zone involvement.

Implication for Patient Care.

• Given that the MR imaging signal intensity of prostate cancer and normal central zone are similar, it is important for the radiologist to recognize the normal central zone to avoid a false-positive diagnosis of cancer; furthermore, tumors involving the central zone tend to be more aggressive than tumors that do not involve the central zone, and therefore it may be prudent to routinely alert referring clinicians to central zone involvement when it is identified with MR imaging, particularly given the high specificity achieved by the two readers in this study (0.93–0.99).

Disclosures of Potential Conflicts of Interest: H.A.V. No potential conflicts of interest to disclose. O.A. No potential conflicts of interest to disclose. T.F. Financial activities related to the present article: none to disclose. Financial activities not related to the present article: employed by Charité-Universitätsmedizin Berlin, received temporary research grant DFG-FR 2891/1-1for January 2010 to June 2010 from Deutsche Forschungsgemeinschaft, and was paid travel and accommodation expenses for RSNA 2010 by Department of Radiology, Charité, Universitätsmedizin Berlin. Other relationships: none to disclose. D.G. No potential conflicts of interest to disclose. K.U. No potential conflicts of interest to disclose. K.A.T. No potential conflicts of interest to disclose. V.R. No potential conflicts of interest to disclose. H.H. Financial activities related to the present article: none to disclose. Financial activities not related to the present article: none to disclose. Other relationships: none to disclose.