Introduction

Prostate cancer (Pca) is the most frequently diagnosed cancer in men, and it is the second most common cause of cancer death in men in the United States [1, 2]. It is thus important to assess the aggressiveness of Pca in order to determine an appropriate initial treatment strategy [3]. The Gleason grading system is the pathologic reference standard for measuring Pca aggressiveness [4–7]. Because an invasive needle biopsy or prostatectomy is necessary for obtaining the Gleason score (GS), a noninvasive measure of Pca aggressiveness is desirable.

Magnetic resonance imaging (MRI) can yield information on tumor aggressiveness using diffusion-weighted imaging (DWI), dynamic contrast-enhanced MRI (DCE-MRI) and MR spectroscopy [8–14]. Amide proton transfer (APT) has been introduced as a novel endogenous contrast mechanism for MRI through the detection of low-concentration solutes such as mobile proteins and peptides in tissues or tumors that contain abundant amide (–NH) chemical constituents [15, 16].

According to previous reports of APT imaging of brain tumors, high-grade gliomas show high APT signal intensity (APT SI), and low-grade gliomas show low APT SI. The authors of these reports thus concluded that APT imaging is useful for assessing tumor aggressiveness [15, 17]. Although most clinical studies of APT imaging have involved neuroimaging [15, 17], a few studies have reported on APT imaging of breast cancer and Pca [18, 19]. In an APT imaging study of Pca, the APT SI was used to differentiate Pca from noncancerous tissue in the peripheral zone (PZ) [19]. However, the utility of this imaging method for assessing the GS of Pca remains unclear. The purpose of the present study was to evaluate the utility of APT imaging in assessing the GS of Pca through radiological-pathological comparison.

Materials and methods

Patients

From January 2012 to December 2013, 170 consecutive male patients were scheduled for MR examinations for suspected Pca based on elevated prostate-specific antigen (PSA) levels or needle biopsy results. A summary flowchart of the patient selection is presented in Fig. 1. A total of 125 patients (mean age 68.4 years, range 50–80 years) were included in the analysis. The mean (range) serum PSA level among the 125 patients was 14.9 (0.01–291.3) ng/mL. Fifteen patients underwent the MR examination before needle biopsies, and the other 110 underwent the MR examination at least 8 weeks after a needle biopsy to minimize the possible interference of post-biopsy hemorrhage in the accurate detection of Pca [20].

Fig. 1
figure 1

Patient selection flowchart. Imaging criteria of Pca lesion, noncancerous PZ and BPH lesion are described in the data analysis of the APT imaging and the DWI/ADC map in the “Materials and methods” section

Full size image

MRI

MR examinations were performed on a clinical whole-body 3.0 Tesla MRI system equipped with a dual-source parallel radiofrequency (RF) subsystem for transmission via a two-channel RF body coil (Achieva 3.0T TX; Philips Healthcare, Best, The Netherlands). A 32-channel torso-cardiac coil was used for parallel imaging. In addition to other MRI including T1-weighted imaging (T1WI), T2-weighted imaging (T2WI), DWI and DCE-MRI, APT imaging was performed prior to DCE-MRI. APT imaging was obtained in the transverse plane. T2WI, DWI and DCE-MRI were used for the localization of Pca and/or noncancerous regions.

The acquisition software was modified to alternate the operation of the two transmission RF sources during the RF saturation pulse, which enables long quasi-continuous RF saturation at a 100 % RF duty cycle. Special RF shimming for the saturation homogeneity of the alternated pulse was applied, adjusting the RF amplitudes of the two transmission RF sources to equalize the B 1-field [17, 21]. On a single slice corresponding to a maximum cross-section area of a tumor, two-dimensional (2D) transverse APT imaging was performed using a saturation pulse with a duration (T sat) of 0.5 s (10 × 50 ms, sinc-Gauss-shaped elements) and a saturation power level corresponding to B 1,rms = 2.0 μT. During each MR examination, a single slice on which a tumor appeared to be present was selected by an experienced radiologist, as referring to other MRI. For the acquisition of an APT Z-spectrum, imaging was repeated at 25 saturation frequency offsets from ω = −6.0 to +6.0 ppm in 0.5-ppm increments, as well as one far off the resonant frequency (ω = −1560.0 ppm) for signal normalization. In addition to B 0 shimming before the scan, a ΔB 0 map for off-resonance correction was acquired separately using a 2D dual gradient-echo sequence (ΔTE = 1 ms) with identical spatial resolution, which was used for a point-by-point ΔB 0 correction. The other MR parameters of the APT imaging and MR parameters of T1WI, T2WI, DWI and DCE-MRI are summarized in Table 1. An apparent diffusion coefficient (ADC) map was generated with reference to the signal intensities of DWI with b values of 0 and 1000 s/mm2.

Table 1 Details of MR parameters
Full size table

Histopathological analysis

The needle biopsy was performed at 12 different sites of the prostate for each patient. Pathological examinations were reviewed by one pathologist (M.S., 8 years of experience in the examination of Pca). First, the presence/absence of cancer tissue at each biopsy site was recorded. Pca was then categorized into one of four GS groups as follows: GS-6 (3 + 3), GS-7 (3 + 4 or 4 + 3), GS-8 (4 + 4) or GS-9 (4 + 5 or 5 + 4).

Data analysis of the APT imaging and the DWI/ADC map

Two experienced radiologists (A.N. and Y.T., 20 and 14 years of experience in interpreting Pca, respectively) reviewed all of the MR images in consensus. In this study, because the APT imaging was of a single slice, one Pca lesion, one noncancerous PZ and one benign prostate hyperplasia (BPH) lesion were analyzed for each patient (Table 1). For patients with multiple Pca lesions, the largest Pca lesion was selected for analysis. First, the two radiologists identified nodules showing no high intensity on T1WI, low intensity on T2WI, high intensity on DWI with b values of 1000 and/or 2000 s/mm2, low ADC and early enhancement on DCE-MRI results for the same slice location as that used for the APT imaging. Second, they confirmed whether these nodules were located at the site of Pca based on the histopathological data obtained by the needle biopsy.

In addition to Pca lesions, noncancerous PZ and BPH lesions were included in the analysis to enable a comparison between benign and malignant lesions. Even if there were multiple areas of noncancerous PZ and BPH lesions were existed in each patient, one noncancerous PZ and one BPH lesion, which showed the largest area on the same image slice of Pca, were selected for the analysis. Areas of noncancerous PZ were defined as high SI on T2WI, low SI on T1WI and DWI with b values of 1000 and 2000 s/mm2 compared with noncancerous PZ, and they confirmed the absence of Pca by using the needle biopsy results. The area of BPH was defined as high or mixed high and low SI on T2WI, low SI on T1WI and DWI with b values of 1000 and 2000 s/mm2, and the absence of Pca was confirmed using the needle biopsy results.

We analyzed the APT imaging data with ImageJ software (National Institutes of Health, Baltimore, MD, USA), using a custom-made plug-in for APT imaging. The analysis method was reported in a previous study [17]. The plug-in was created to assess the Z-spectra and magnetization transfer ratio asymmetry (MTRasym), and was equipped with a correction function for B 0 inhomogeneity, using interpolation among the Z-spectral images. First, rigid-body motion correction was performed using the TurboReg algorithm. The local B 0 field shift in Hz was obtained from the B 0 map, which was created from dual gradient-echo images (TE1 and 2) according to the following equation:

$$\Delta B_{0} ( \times ) \, = \, ({\text{Phase}}[{\text{TE}}2 ]( \times ) \, - {\text{ Phase}}[{\text{TE}}1 ]( \times ))/({\text{TE}}2 \, - {\text{ TE}}1)*2*{\text{Pi}}$$

where phase [TEi](×) indicates phases of the images with echo times TE1 or TE2 at position × in radians, and TE1 and TE2 are given in seconds. The ΔB 0(×) is the resulting B 0 map measured in Hz. The image intensity of each voxel was corrected for the nominal saturation frequency offset by Lagrange interpolation among the neighboring Z-spectral images. This procedure corresponds to a frequency shift along the saturation frequency offset axis according to the measured B 0 shift.

The MTR was defined as 1  −  S sat/S 0, where S sat and S 0 are the SIs obtained with and without selective saturation, respectively. The analysis method was reported in a previous study [16]. To reduce the unwanted contributions from the conventional magnetization transfer (MT) effect and the direct saturation of bulk water, we performed an asymmetry analysis of MTR values with respect to the water frequency. For the APT imaging, the asymmetry analysis at 3.5 ppm downfield from the water signal was calculated as MTRasym (3.5 ppm):

$${\text{MTR}}_{\text{asym}}(3.5 \text{ppm})\,=\,[S_{\text{sat}}(3.5{\text{ppm}})-S_{\text{sat}}(+3.5{\text{ppm}})]/S_{0}\,=\, {\text{MTR}}^{\prime}_{\text{asym}}(3.5 \text{ppm})$$

where MTR′asym is the inherent asymmetry of the conventional MT effect, and APTR is the amide proton transfer ratio [16]. Thus, it should be noted that the measured MTRasym (3.5 ppm) values are the apparent APT SIs, and thus it is appropriate to define the calculated MTRasym (3.5 ppm) images as APT-weighted imaging (APTWI). In the present study, APT SI was defined as MTRasym (3.5 ppm) × 100 (%).

The ADC values were calculated on a commercially available PACS workstation (SYNAPSE, Fujifilm Medical Co., Ltd., Tokyo, Japan). The means and standard deviations (SDs) of the APT SIs or ADC values of Pca, noncancerous PZ and BPH were measured by drawing polygonal regions of interest (ROIs) on source images of APTWI and placing these ROIs on the same slice of the ADC map by consensus of the two readers. When the ROIs of the tumor in the noncancerous PZ or BPH were drawn on the APTWI and ADC maps, as much tumor as possible was included, referring to information from the T2WI, DWI and DCE-MRI. To minimize the effect of noise on the measurement of APT SIs and ADC values, only tumors with a transverse area ≥ 300 mm2 on APTWI and ADC maps (≥93 pixels on APTWI) were included for the analysis. The mean (range) area among analyzed Pca was 651.7 (306.7–3080.2) mm2. Similarly, only noncancerous PZ and BPH with an area ≥ 300 mm2 were included. The mean (range) areas of analyzed noncancerous PZ and BPH were 577.2 (306.7–2531.3) mm2 and 990.3 (326.1–3512.9) mm2, respectively.

Statistical analysis

The mean APT SIs and ADC values of the Pca, noncancerous PZ and BPH were compared by one-way analysis of variance (ANOVA) with Tukey’s honest significant difference (HSD) post hoc test. Spearman’s rank correlation coefficients (Spearman’s ρ) were calculated between the APT SI and GS groups and between ADC values and GS groups. All statistical analyses were performed with PASW Statistics for Windows, Version 18.0 (SPSS Inc., Chicago, IL, USA). Values of p < 0.05 were considered significant for each analysis.

Results

Numbers of analyzed tumor and noncancerous regions

Fifty-nine Pca lesions, 19 noncancerous PZ areas and nine BPH lesions were excluded due to small size (< 300 mm2; <93 pixels) on APTWI. Thirty-one noncancerous PZ and 46 BPH lesions were excluded due to discordance with the imaging criteria described in "Materials and methods." Ultimately, 66 Pca lesions, 75 noncancerous PZ areas and 70 BPH lesions were analyzed (Fig. 1). The 66 Pca lesions were categorized into four GS groups as follows: GS-6 (n = 23 lesions), GS-7 (n = 18), GS-8 (n = 11) and GS-9 (n = 14). No Pca with a GS less than 6 or greater than 9 was present in our population. Among the total 66 Pca lesions, 33, 10 and 23 were located in the PZ, transitional zone (TZ) and both PZ and TZ, respectively.

APT SIs and ADC values among the four GS groups

The results of APT SI comparisons among the four GS groups are shown in Fig. 2. Of these, GS-7 showed the highest APT SIs (5.17 ± 2.89 %) and GS-9 showed the lowest (1.96 ± 2.42 %). There were significant differences between the GS-6 and GS-7 groups and between the GS-7 and GS-9 groups (p < 0.05), but there were no significant differences between the GS-6 and GS-8 groups, the GS-6 and GS-9 groups, or the GS-7 and GS-8 groups.

Fig. 2
figure 2

Means and SDs of the APT SIs of each GS group of prostate cancer lesions, noncancerous peripheral zone (PZ) areas and benign prostate hyperplasia (BPH) lesions. The unit of APT SI is percentage. The GS-7 group showed the highest APT SIs among the four groups. There were significant differences between the GS-6 and GS-7 groups and the GS-7 and GS-9 groups. There was a significant difference between the BPH and GS-7 groups, but no significant differences between the noncancerous PZ and each of the GS groups. *p < 0.05 by one-way ANOVA and Tukey’s HSD test

Full size image

The results of the comparison of ADC values among the four GS groups are shown in Fig. 3. There were significant differences between the GS-6 and GS-7 groups, the GS-6 and GS-8 groups, and the GS-7 and GS-9 groups (p < 0.05), but there were no significant differences between any two of the GS-7, GS-8, and GS-9 groups.

Fig. 3
figure 3

Means and SDs of the ADC values of each GS group of prostate cancer, noncancerous PZ areas and BPH lesions. The unit of ADC value is ×10−3 mm2/s. There was a significant difference between the noncancerous PZ and BPH values and between the noncancerous PZ areas and each GS group. There were also significant differences between the BPH and each of the GS groups. *p < 0.05 by one-way ANOVA and Tukey’s HSD test

Full size image

APT SIs and ADC values between noncancerous regions and prostate cancer

Comparisons of APT SIs among the noncancerous PZ regions, BPH and each GS group are shown in Fig. 2. There was a significant difference between BPH (2.83 ± 2.98 %) and the GS-7 group (p < 0.05), but there were no significant differences between the noncancerous PZ areas (3.33 ± 2.72 %) and any of the GS groups.

Comparisons of ADC values among the noncancerous PZ areas, BPH and each GS group are shown in Fig. 3. There were significant differences between the noncancerous PZ group and each GS group and between the BPH group and each GS group (p < 0.05). There were also significant differences between the noncancerous PZ areas and the BPH group (p < 0.05). Example cases of Pca lesions with GS-7 and GS-9 are shown in Figs. 4 and 5, respectively.

Fig. 4
figure 4

A 54-year-old man with GS-7 prostate cancer. a APT-weighted image of the prostate fused with a fat-suppressed proton density-weighted image; b T2-weighted image; c ADC map. The tumor was located in the right lobe of the PZ. The tumor showed APT SI of 5.4 % and ADC value of 0.68 × 10−3 mm2/s

Full size image
Fig. 5
figure 5

A 65-year-old man with GS-9 prostate cancer. a APT-weighted image of the prostate fused with a fat-suppressed proton density-weighted image; b T2-weighted image; c ADC map. The tumor was located in the left lobe of the PZ. The tumor showed APT SI of 2.2 % and ADC value of 0.62 × 10−3 mm2/s

Full size image

Rank correlations between the APT SIs and GS groups and between the ADC values and GS groups

There was no significant correlation between APT SIs and GS groups (Spearman’s ρ = −0.043, p = 0.53), but there was a significant correlation between the ADC values and GS groups (Spearman’s ρ = −0.795, p < 0.01).

Discussion

Pca lesions from the GS-7 group, which are categorized as intermediate-risk Pca, showed the highest APT SIs among the four GS groups. However, the APT SIs did not show a significant rank correlation with the GS groups. The results of the present study differ from those involving APT imaging of brain tumors [15, 17]. The relevant literature reveals that aggressive malignant tumors (e.g., high-grade gliomas) show high APT SIs, and less aggressive malignant tumors (e.g., low-grade gliomas) show low APT SIs. The APT signals of brain tumors are positively correlated with cellular density and/or proliferation, i.e., intracellular mobile proteins and peptides [15, 17]. A study using APT imaging of Pca showed that APT SIs for GS-8 Pca (1.44 %), which was the APT SI calculated from one tumor, were smaller than for GS-7 lesions (6.4 ± 2.6 % in 3 + 4, 6.0 ± 3.7 % in 4 + 3). The results of our study were consistent with those of a previous study using APT imaging of Pca, even though the saturation schemes differed between the present study (T sat = 0.5 s and B 1,rms = 2.0 μT) and the previous study (T sat = 0.496 s and B 1,rms = 3.8 μT) [19]. Another study suggested that T sat = 0.6 s and B 1,rms = 2.0 μT were optimal for obtaining APT MRI contrast between Pca lesions and noncancerous PZ areas [22]. Although there was a difference in T sat between the present and previous studies, the duration and power of the saturation RF powers in the present study might be appropriate for the assessment of APT SIs in Pca [22].

With regard to the relationship between the ADC values of the present Pca and GS groups, our results showed a significant negative correlation, which is consistent with previous findings [8, 12, 23–27]. The reduction in ADC values in malignant tumors is related to histopathological characteristics, such as cellular density and the percentage area of nuclei and cytoplasm, resulting in a restriction of the movement of water molecules [8, 28–31]. In addition, it is reported that there is a positive correlation between ADC and the percentage area of luminal space for Pca [8, 29, 30, 32]. That is, the ADC values of Pca decreased and the tumor cellularity of the prostate increased in accordance with Pca aggressiveness. The APT SIs and ADC values in the present study showed different changes even though those two indices are considered to be related to cellular density [8, 17, 28–30, 32]. We hypothesized that other factors except for cellular density may also affect the APT SI in Pca.

The Gleason grading system, which is commonly used for evaluating Pca [5, 33], consists of two digits that signify the primary and secondary patterns in Pca, based on a scale of 1 (least aggressive) to 5 (most aggressive) [5, 33]. In normal prostate tissue, the glands are typically large, with an arbitrary shape, and contain abundant nuclei on the boundary [5]. The normal prostate gland comprises a network of water-rich ducts and acini supported by stroma [29]. The stroma is much looser in the PZ than in the central gland, resulting in a larger extracellular space [29]. PZ tissue contains numerous glands, and the glandular cavity is full of liquids and mobile proteins and peptides [29]. Gleason pattern 3 cancer consists of discrete glandular units, typically with smaller glands than those seen in Gleason pattern 1 or 2 cancers, in addition to infiltrates in and among non-neoplastic prostate acini, marked variations in size and shape, and smoothly circumscribed small cribriform tumor nodules [5]. Glands of Gleason pattern 4 cancer tend to be fused together rather than isolated as those in Gleason pattern 3 [34]. Gleason pattern 4 cancer consists of fused microacinar glands, ill-defined glands with poorly formed glandular lumina, large cribriform glands, cribriform glands with an irregular border, and a hypernephromatoid pattern [5]. Gleason pattern 5 cancer shows essentially no glandular differentiation, and it is composed of solid sheets, cords, or single cells, and comedocarcinoma with central necrosis surrounded by papillary, cribriform or solid masses [5, 34]. As the prostate gland contains liquid and soluble proteins, we hypothesize that the change in glandular structures, i.e., extracellular mobile proteins and peptides, might affect the APT SI of the GS groups [29].

The higher APT SI of the GS-7 group may be the result of two counteracting effects. With regard to the comparison among noncancerous PZ and Pca with GS-6 and GS-7, the cellular density of Pca increases in accordance with the progression of GS [28–30, 32]. On the other hand, glandular structures containing liquids and mobile proteins and peptides still exist in Pca with GS-6 and GS-7. Contrarily, the cellular density in Pca with GS-8 or GS-9 is increased compared to GS-7, but the number of glandular structures is decreased. We speculate that the decrease in the number of glandular structures reduces the amount of liquids and mobile proteins and peptides. If we consider that APT SI may be affected by both cellular density (“intracellular” mobile proteins and peptides) and glandular structures (“extracellular” mobile proteins and peptides), the higher APT SI in the GS-7 group may be explained by a particular balance between the contributions of these two phenomena.

One major limitation of this study is that the histopathological results for the Gleason scores were obtained from needle biopsies. Needle biopsies have been shown to lead to a 25–69 % underestimation of GS compared with GS established from prostatectomy specimens, as sampling errors are occasionally unavoidable due to the histological heterogeneity of Pca [11, 26, 35]. Therefore, there is a risk of underestimating true Pca aggressiveness using our methodology. However, it was difficult for us to collect a sufficient number of patients with a highly aggressive Pca who had undergone a prostatectomy without prior hormone therapy and/or radiotherapy.

There were other drawbacks in the study. First, because of time limitations for the imaging protocol, APT imaging is limited to one slice per patient. In this study, only the largest Pca lesion was selected in patients with multiple Pca lesions, as assessment of the entire prostate gland was deemed impossible. In addition, the spatial resolution of the APT imaging was 1.8 × 1.8 × 5.0 mm3, making it difficult to assess small Pca lesions. We also did not compare APT imaging with other multi-parametric MR techniques such as DCE-MRI and MR spectroscopy. These imaging techniques are known to be useful as noninvasive means of estimating Pca aggressiveness. Another study limitation was that we could not divide the GS-7 tumors into two groups of GS 3 + 4 and GS 4 + 3, and the GS-9 tumors into two groups of GS 4 + 5 and GS 5 + 4. According to the literature, the prognosis of patients with a GS of 4 + 3 is worse than that of patients with a GS of 3 + 4 [36], but there were insufficient numbers of Pca lesions in the present study for such comparisons.

Conclusions

The mean APT SI in Pca with a GS of 7 was higher than that for the other GS groups, but the APT SIs did not show significant rank correlation with GS groups.