Blood oxygenation level-dependent MRI at 3T for differentiating prostate cancer from benign tissue: a preliminary experience

Abstract

Objective:

Blood oxygenation-level dependent (BOLD) MRI may identify or quantify the regional distribution of hypoxia within a tumor. We aimed to evaluate the feasibility of BOLD MRI at 3 T in differentiating prostate cancer from benign tissue.

Methods:

A total of 145 patients with biopsy-proven prostate cancer underwent BOLD MRI at 3 T. BOLD MRI was performed using a multiple fast field echo sequence to acquire 12 T ₂*-weighted images. The R2* value (rate of relaxation, s⁻¹) was measured in the index tumor, and benign peripheral (PZ) and transition zone (TZ), and the results were compared. The variability of R2* measurements was evaluated.

Results:

Tumor R2* values (25.95 s⁻¹) were significantly different from the benign PZ (27.83 s⁻¹) and benign TZ (21.66 s⁻¹) (p < 0.001). For identifying the tumor, the area under the receiver operating characteristic of R2* was 0.606, with an optimal cut-off value of 22.8 s⁻¹ resulting in 73.8% sensitivity and 52% specificity. In the Bland–Altman test, the mean differences in R2* values were 8.5% for tumors, 13.3% for benign PZ, and 6.8% for benign TZ. No associations between tumor R2* value and Gleason score, age, prostate volume, prostate-specific antigen, or tumor size.

Conclusion:

BOLD MRI at 3 T appears to be a feasible tool for differentiating between prostate cancer and benign tissue. However, further studies are required for a direct clinical application.

Advances in knowledge:

The R2* values are significantly different among prostate cancer, benign PZ, and benign TZ.

Introduction

Hypoxia is known to be prognostic marker for a poor treatment outcome with radiotherapy, chemotherapy, or surgery in solid tumors. ¹ Tumor hypoxia can be evaluated using either a polarographic needle electrode or tissue immunohistochemistry, but it is invasive, dependent on electrode placement, and offers a limited sample for immunohistochemistry. ² Thus, non-invasive imaging tools for identifying or quantifying the regional distribution of hypoxia in a tumor show potential for selecting an appropriate treatment strategy in oncology.

In prostate cancer, multiparametric MRI is the best imaging tool for tumor localization, staging, guiding targeted biopsy, monitoring therapeutic response, following up active surveillance, or detecting recurrence. ³ Of the functional MRI techniques, blood oxygenation-level dependent (BOLD) MRI might offer some information on hypoxia in a non-invasive manner. ² Many studies have reported the utility of BOLD MRI in cervical cancer, breast cancer, and renal cell carcinoma and in assessing the dysfunction of native or transplanted kidneys. ^1,2,4 To date, several studies have demonstrated that BOLD MRI at 1.5 T is a feasible tool for evaluating prostate cancer. ^5–9 However, few studies have reported the value of BOLD MRI at higher field strengths in evaluating prostate cancer, ^7,10 especially without vasomodulation such as carbogen breathing. Therefore, the purpose of this study was to evaluate the feasibility of BOLD MRI at 3 T without carbogen breathing in differentiating prostate cancer from benign tissue.

Methods and materials

Patients

This retrospective study was approved by our institutional review board, and the need for informed consent was waived. 154 patients with biopsy-proven prostate cancer patients underwent pretreatment MRI at Samsung Medical Center between September 2014 and March 2015. The inclusion criteria were as follows: (1) biopsy-proven prostate cancer, (2) BOLD MRI at our institution, and (3) no previous treatments, such as surgery, radiotherapy, or hormonal therapy. Of these, nine patients were excluded due to severe artifacts. Thus, 145 patients (mean age, 66.8 years) treated with surgery (n = 101) and hormonal therapy or radiotherapy (n = 44) were enrolled in this study.

MRI and biopsy

Pretreatment multiparametric prostate MRI was performed using a 3 T MR scanner (Intera Achieva 3.0 TX; Philips Medical System, Best, The Netherlands) with a phased-array body coil. The routine MRI protocols included T ₂ weighted, T ₁ weighted, diffusion-weighted, and dynamic contrast-enhanced MR images according to the PI-RADS v. 2 guidelines. ³ In addition to the routine protocols, transverse BOLD MRI was acquired before the administration of contrast materials using a multiple fast field echo sequence to acquire 12 T ₂* weighted images. The parameters were as follows: repetition time, 260 ms; range of echo time, 5–60 ms; flip angle, 27°; slice thickness, 5 mm; interslice gap, 1 mm. The 12 T ₂* weighted images corresponding to 12 different gradient echoes were obtained for each section. The acquisition time was about 59 s.

Transrectal ultrasound-guided biopsy was performed using an ultrasound device (IU22, Philips Healthcare, Bothell, WA) with an end-firing transrectal transducer, and an 18 G, 22 cm biopsy gun (ACECUT, TSK Laboratory, Tochigi, Japan) was used. From the prostate base to the apex bilaterally, standard systemic biopsy obtained 10‒12 cores covering the peripheral zone (PZ), and targeted biopsy obtained 2‒3 cores. The mean time interval between MRI examination followed by biopsy and biopsy was 27.3 days (range, 11‒58 days).

R2* measurement

All images were evaluated by a radiologist (C.K.K., with 14 years of experience in prostate MRI). The radiologist was aware that patients had biopsy-proven prostate cancer, but the clinical and pathologic results of each patient were blinded. Clinical variables including age, prostate-specific antigen (PSA), prostate volume, PSA density, tumor size at MRI, and digital rectal examination, and pathological findings including Gleason score, tumor location, tumor volume, extracapsular extension, and seminal vesicle invasion were recorded.

All T ₂* weighted images were analyzed using an independent workstation (PRIDE, v. 4.2, Philips Healthcare, Best, The Netherlands). The color-coded R2* (s⁻¹) map was created from a voxel basis to an exponential function. On the R2* map, red indicates the highest R2* values, which indicates a high concentration of deoxyhemoglobin; blue indicates the lowest R2* values, indicating a low concentration of deoxyhemoglobin.

With manual placement of a region of interest (ROI) on the R2* map, the R2* values of tumor and benign prostate tissues, including the peripheral zone (PZ) and transition zone (TZ), were calculated by a radiologist (C.K.K.). The ROI was manually placed as multiparametric MRI findings based on the histopathological results. The tumor ROI was drawn for the index lesion in each patient. The index lesion was considered as the lesion with the largest diameter or the lesion demonstrating extraprostatic extension. The ROI measurement of the tumor encompassed as much as possible in an image with the greatest visibility, and the ROI measurements of the benign PZ and TZ were performed with pixel numbers of 12‒15 (Figure 1). The average of two measurements was obtained for each tumor and benign tissue. The mean number of pixels in the tumors was 21.8 (range, 10‒238). If ROI measurement was impossible in the benign PZ or TZ due to a diffuse tumor, ROI measurement was performed only in the tumor. Thus, R2* measurements were performed for the tumor in 145 patients, the benign PZ in 138 patients, and the benign TZ in 141 patients. To evaluate inter-reader variability, a radiologist (Y.K., with 2 years of experience in prostate MRI) independently calculated the R2* values for 40 patients using the same method as the experienced radiologist on different days.

Figure 1.

R2* calculation on the R2* map using a ROI. (a, b) Transverse T ₂ weighted image (a) and apparent diffusion coefficient map (b) show a 1.7 cm hypointense mass (arrow) in left TZ. (c) The ROIs of the tumor (red-colored), benign PZ (yellow-colored) and benign TZ (white-colored) are demonstrated on T ₂* weighted image (upper left) and R2* map (upper right). The R2* values of the tumor, benign PZ, and benign TZ are 31.9, 19.8, and 22.2 s⁻¹, respectively. The right lower image shows an example of the expected signal decay as a function of echo time in the tumor. PZ, peripheral zone; ROI, region of interest; TZ, transition zone.

Statistical analysis

All statistical analyses were performed using SPSS software (v. 26.0; IBM, Armonk, NY) and MedCalc (v. 13.0; MedCalc Software, Mariakerke, Belgium). Two-sided p-values < 0.05 were considered statistically significant. Differences in R2* values between tumor and benign tissue were compared using the paired Student’s t-test. Differences in R2* values among tumors, benign PZ, and benign TZ were tested using one-way analysis of variance. For multiple pairwise comparisons, Tukey’ post hoc analysis was performed. Differences in R2* values between patients with and without clinically significant cancer (i.e. defined as Gleason score ≥7) were compared using an independent t-test. To evaluate diagnostic performance and identify optimal cut-off values for differentiating between tumor and benign tissue, receiver operating characteristic (ROC) curve analysis was used, and the area under the curve (AUC), sensitivity, and specificity were also calculated. The Youden index was calculated to identify the optimal cut-off value. Inter-reader variability was evaluated using a Bland–Altman plot. The associations between tumor R2* value and variables including age, PSA, prostate volume, PSA density, Gleason score, and tumor size at MRI were evaluated using Spearman rank correlation.

Results

Table 1 presents the clinical characteristics of the included patients. The mean PSA level was 21.59 ± 62.96 ng ml⁻¹. The mean prostate volume was 34.0 ± 15.8 cm³. The median Gleason score at biopsy was 7 (range, 6‒10). The tumors were distributed as follows: in the PZ for 106 patients, TZ for 20 patients, and both zones for 13 patients. On surgical findings, the mean tumor volume was 3.69 ± 5.83 cm³. Extracapsular extension and seminal vesicle invasion were found in 36 and 18 patients, respectively.

Table 1.

Patient characteristics (n = 145)

Variable	Value
Age (years)	66.8 ± 8.4
PSA (ng/mL)	21.59 ± 62.96
Prostate volume (cm3)	34.0 ± 15.8
PSA density	0.63 ± 1.62
Tumor size at MRI (cm)	2.03 ± 1.19
Biopsy Gleason score	7 (6‒10)
6	45 (31.0)
7	52 (35.9)
8‒10	48 (33.1)
Tumor location
PZ	106 (73.1)
TZ	26 (17.9)
Both	13 (9.0)
Digital rectal examination
Positive	34 (23.4)
Negative	111 (76.6)
Treatment
Surgery	101 (69.7)
Hormonal therapy or radiotherapy	44 (30.3)
Surgical findings
Tumor volume (cm3)	3.69 ± 5.83
Gleason score	7 (6‒9)
Extracapsular extension	36 (35.6)
Seminal vesicle invasion	18 (17.8)

PSA, prostate-specific antigen; PZ, peripheral zone; TZ, transition zone.

Data are presented as mean ± standard deviation, median (range), or n (%).

The R2* values in the tumors and benign prostatic tissues are shown in Table 2. The mean R2* values of tumors were 25.95 ± 5.35 s⁻¹, which were significantly higher than those of benign tissues (24.71 ± 4.93 35 s⁻¹) (p = 0.040). To differentiate between tumors and the PZ or TZ, the mean R2* values of tumors were significantly different from those of the PZ or TZ (p < 0.001; Figure 2). In pairwise comparisons, the mean R2* values were significantly lower in tumors than in the PZ (27.83 ± 8.88 s⁻¹) (p = 0.041). The mean R2* values of the TZ (21.66 ± 3.04 s⁻¹) were significantly lower than those of the tumors (p < 0.001). The mean R2* values of the PZ were significantly higher than those of the TZ (p < 0.001).

Table 2.

R2* values of tumor and benign tissue

Variable	R2*	p-value
Tumor	25.95 ± 5.35	0.040a
Benign tissue	24.71 ± 4.93	<0.001b
PZ	27.83 ± 8.88
TZ	21.66 ± 3.04

PZ, peripheral zone; TZ, transition zone.

Data are presented as mean ± standard deviation.

^aStatistical comparison between tumor and benign tissues.

^bStatistical comparison among tumor, PZ, and TZ.

Figure 2.

Box-Whisker plots show R2* values of the tumor, benign PZ, and benign TZ. Central horizontal line = median, and box = 25 and 75% confidence intervals. PZ, peripheral zone; TZ, transition zone.

For differentiating between tumors and benign tissues, the AUC for R2* values was 0.606 (95% confidence interval, 0.558–0.653). With an optimal cut-off value of 22.8 s⁻¹, the sensitivity and specificity were 73.8 and 52%, respectively.

For inter-reader variability, the Bland–Altman test demonstrated that the mean differences in R2* values were 8.5% for tumors, 13.3% for benign PZ, and 6.8% for benign TZ.

No significant associations were found between tumor R2* value and Gleason score, age, prostate volume, PSA density, or tumor size at MRI (all p > 0.05). There was no significant difference in tumor R2* value between patients with and without clinically significant cancer (p = 0.756).

Discussion

Tumor hypoxia impacts therapeutic response by increasing resistance to radiation and gene expression of metastasis-promoting proteins. ¹¹ Moreover, hypoxia can cause aggressive behavior in tumors resulting from restrained proliferation, differentiation, necrosis, or apoptosis. ¹² In prostate cancer, several studies have reported that hypoxia is associated with poorer outcomes after surgery or radiotherapy. ^13–15 Therefore, stratifying patients based on tumor hypoxia before or during treatment may allow oncologists to plan a treatment strategy based on tumor characteristics, which may offer opportunities for personalized treatment. Recently, BOLD MRI has been used to evaluate hypoxia in prostate cancer. ^5–10 However, investigations of BOLD MRI in human prostate cancer are limited, especially at higher MR field strengths or in situations without carbogen inhalation. The present study was performed to investigate the feasibility of BOLD MRI at 3 T for evaluating prostate cancer in cases of room-air inhalation without carbogen breathing. Our results demonstrated that the mean R2* values of tumors were significantly different from those of benign tissues, the mean difference in inter reader variability of R2* measurement was 8.5% for tumors and 6.8‒13.3% for benign tissue. Therefore, BOLD MRI at 3 T might be a feasible tool to evaluate prostate cancer without carbogen breathing that improves oxygenation of normal prostatic tissue and prostate cancer.

In BOLD MRI, tissue contrast is influenced by intrinsic tissue properties, which can be affected by blood flow and paramagnetic deoxyhemoglobin. BOLD MRI offers a sensitive index of the oxygenation status of tissue immediately adjacent to perfused microvessels. Accordingly, the BOLD effect can be quantified by plotting the natural log of signal intensity against the echo time: R2* (rate of relaxation)=1/T2*. Thus, high R2* value suggests a high concentration of deoxyhemoglobin and a relatively low concentration of oxygenation in tissues. ¹⁶

Few studies have reported potential feasibility of BOLD MRI at 1.5 T to evaluate prostate cancer. ^5,7,9 Alonzi et al ⁹ showed that the mean R2* values of the tumor, benign PZ, and benign TZ in 20 patients were 15.59, 14.09, and 10.35 s⁻¹, respectively. Chopra et al ⁵ reported that the mean R2* value of tumors in nine patients was 17.6 s⁻¹. The BOLD effect increases linearly with increasing field strength. ¹⁷ In our study at 3 T, the mean R2* value of the tumor, benign PZ, and benign TZ were 25.95, 27.83, and 21.66 s⁻¹, respectively, which were higher (an increase of less than a factor of 2 than that reported by previous studies at 1.5 T). A recent study at 3 T reported that the median R2* values of the tumor, benign PZ and benign central zone were 24.0, 24.9, and 23.6 s⁻¹ without the use of an endorectal coil, and 24.3, 27.9, and 25.6 s⁻¹ with the use of an endorectal coil, respectively. ¹⁰ These results are consistent with our results. However, a study by Zhou et al ¹⁸ was performed at 3 T with an endorectal coil in 10 patients with prostate cancer and the mean R2* values of the tumor (46.05 s⁻¹) and benign prostate tissue (41.91 s⁻¹) were higher than our results. These conflicted results might be caused by different numbers of patients, the use of different coils at MRI examinations, or different time interval of MR examinations after the biopsy.

Contrary to a previous study, ⁹ Luttje et al ¹⁰ and our study demonstrated that the mean R2* value of the benign PZ was significantly higher than that of the tumor, and that of the benign TZ was significantly lower than that of the tumor. Zhou et al ¹⁸ reported that the mean tumor R2* values were significantly higher than benign prostate tissue, although R2* measurements in the PZ were not obtained due to susceptibility artifacts by the use of endorectal coil. The R2* value does not necessarily reflect tissue pO2, while the R2* value reflects the oxygenation status of hemoglobin within perfused vessels. Accordingly, R2* depends on blood volume and flow. ⁸ Compared to the benign PZ, prostate cancer has increased blood flow or volume due to neoangiogenesis, ⁷ which increases oxygenation leading to a decreased R2* value in the tumor. Benign prostate hyperplasia (BPH) in the TZ can show comparable enhancement with the tumor because BPH is a benign proliferative process characterized by increased microvessel density, which results in increased oxygenation status and decreased R2* value. Interestingly, the R2* value of benign TZ was significantly lower than that of the tumor. This might be explained as follows: (1) cancers with a low Gleason score may show similar tumor perfusion to the benign PZ, resulting in similar R2* values as the benign PZ; or (2) post-biopsy hemorrhage in the prostate might affect magnetic field inhomogeneity, leading to an impact on the R2* value in the tumor or benign tissue: e.g. elevation of R2* values. In addition to oxygenation, the R2* value is affected by other factors, such as transverse relaxation parameters, iron content, fibrosis, blood volume fraction, blood flow, or macroscopic field inhomogeneity. ^19,20

Of the multiparametric MR techniques, T2 signal, apparent diffusion coefficient, quantitative dynamic contrast-enhanced MRI parameters, and choline-to-citrate are associated with tumor aggressiveness in prostate cancer. ^21,22 Many oncologic studies have reported that the R2* value from BOLD MRI might be a useful quantitative marker for predicting clinical outcome in breast cancer, cervical cancer, colon cancer, or ovarian cancer. ^23–25 However, no studies have demonstrated the role of the R2* value in prostate cancer to evaluate tumor aggressiveness. In our study, there was no significant association between tumor R2* and the Gleason score. Further studies with a well-designed and large population are necessary.

In our study, ROC curve analysis demonstrated poor diagnostic performance (AUC = 0.606) of the R2* value in differentiating between tumors and benign prostate tissue, and with an optimal cut-off value of 22.8 s⁻¹, 74% sensitivity and 52% specificity were observed. These results were comparable with those of a previous study ⁶ which the sensitivity and specificity of R2* in depicting tumor hypoxia were 88 and 36%, respectively. The poor specificity of R2* in demonstrating tumor hypoxia may be explained by the lower blood volume in prostate cancer. ²⁶ Although R2* values differed among tumors, the benign PZ, and the benign TZ, the poor performance of the R2* value in ROC curve analysis suggests limited applications in clinical practice.

All quantitative measurements from functional MRI have inter- or intrareader variability. In our study of R2* measurements, inter reader variability was less than 13.3% in Bland–Altman plots, which was in line with a previous study. ^9,27 Accordingly, our results suggest that R2* measurement in the prostate might be a reliable quantitative marker.

In BOLD MRI, breathing carbogen is one possibility and may increase tissue oxygenation through decreased deoxyhemoglobin concentration and improved blood perfusion. ^7,28 In a recent study, R2* measurements were performed before and during a period of carbogen gas inhalation in 17 patients with prostate cancer. A 64% reduction in tumor R2* values during carbogen inhalation was identified, while a 22% reduction in tumor R2* values was observed in the native state. ²⁹ Diergarten et al ²⁸ performed BOLD MRI at 1.5 T in 32 prostate cancer patients. Their results demonstrated that significant difference between signal enhancement during carbogen breathing in the tumor and benign prostate tissue was found. Interestingly, a study at 3 T with oxygen breathing demonstrated that tumor R2* values decreased significantly upon oxygen challenge, suggesting improved oxygenation. ¹⁸ Although our study performed BOLD MRI without carbogen or oxygen inhalation, our results revealed the potential feasibility of R2* values in differentiating between tumor and benign tissue. However, further studies on carbogen or oxygen inhalation are needed.

There are several limitations to our study. First, this study had a retrospective design with a relatively small population. Second, 30.3% of patients did not undergo surgery because of advanced-stage cancer (≥T3a), nodal metastasis, or distant metastasis. There might be limited correlations between BOLD effects and histopathological findings, but any potential selection bias would only be present in the study population who underwent surgery because advanced-stage prostate cancer with a higher Gleason score could be excluded. Further studies are needed to identify the pathophysiological or hemodynamic aspects of prostate cancer. Third, the mean R2* values within the ROI of the tumor have limited utility, considering tumor heterogeneity. More dedicated software development or histogram analysis is required. Fourth, the R2* value may directly reflect blood oxygenation in the tumor. In particular, post-biopsy hemorrhage in prostate or rectal gas might affect the R2* value due to potential susceptibility artifacts. Finally, to date, no optimal protocols for BOLD MRI have been determined.

In conclusion, our results demonstrated that BOLD MRI at 3 T appears to be a feasible technique for differentiating between prostate cancer and benign tissue. However, further studies are required for a direct clinical application.