What's known on the subject? and What does the study add?
Over-treatment of indolent prostate cancer lesions is a problem which can result in increased human and medical costs.
Lesions with a low suspician level at mpMRI of the prostate have low risk of including high risk prostate cancer.
To determine whether multiparametric magnetic resonance imaging (mpMRI) has the potential to identify patients at low risk for cancer, thus obviating the need for biopsy. Prostate cancer is currently diagnosed by random biopsies, resulting in the discovery of multiple low-risk cancers that often lead to overtreatment.
Patients and Methods
We reviewed 800 consecutive patients who underwent a 3 Tesla mpMRI of the prostate with an endorectal coil from March 2007 to November 2011.
All suspicious lesions were independently reviewed by two radiologists using T2-weighted, diffusion-weighted, spectroscopic and dynamic contrast-enhanced MRI sequences.
Patients with only low suspicion lesions (maximum of two positive parameters on mpMRI) who subsequently underwent transrectal ultrasonography (TRUS)/MRI fusion targeted biopsy were selected for analysis.
In total, 125 patients with only low suspicion prostatic lesions on mpMRI were identified.
On TRUS/MRI fusion biopsy, 77 (62%) of these patients had no cancer detected, 38 patients had Gleason 6 disease and 10 patients had Gleason 7 (3+4) disease.
There were 30 patients with cancer detected on biopsy who qualified for active surveillance using 2011 National Comprehensive Cancer Network guidelines.
No cases of high-risk (≥ Gleason 4+3) cancer were identified on biopsy and, of the fifteen patients who underwent radical prostatectomy at our institution, none were pathologically upgraded to high-risk cancer.
Thus, for patients with only low suspicion lesions, 107 (88%) patients either had no cancer or clinically insignificant disease.
The results obtained in the present study show that low suspicion lesions on mpMRI are associated with either negative biopsies or low-grade tumours suitable for active surveillance.
Such patients have a low risk of harbouring high-risk prostate cancers.
Prostate cancer (CaP) is a common malignancy in the USA . Screening for CaP with a DRE and monitoring of serum PSA levels often leads to the detection of cancer via ultrasonography (i.e. TRUS-guided biopsy). However, TRUS-guided biopsies are not directed at specific areas of an abnormality and may miss clinically significant disease, as well as inaccurately stage the cancer. Therefore, there is a current need to improve the screening and staging algorithm for CaP.
The use of multiparametric MRI (mpMRI) has gained attention as a result of its role in the localization of CaP. It has been shown that mpMRI correlates with the histopathological tumour location on final pathology . By fusing mpMRI with TRUS, it is possible to superimpose pre-biopsy MRI images onto images obtained by real-time transrectal ultrasonography to allow for targeted biopsy. This method yielded a superior cancer detection rate of 54.4% compared to 27–40% for a standard random TRUS biopsy . In a similar study, the level of suspicion for cancer on mpMRI correlated with the D'Amico classification . Moreover, MRI was shown to better stage patients before treatment compared to TRUS biopsy . This suggests that lesions identified by mpMRI vary in significance according to the number of pulse sequences that are positive. Those lesions seen on only one or two sequences are often associated with the absence of cancer or the presence of low-grade cancers. Therefore, mpMRI could potentially serve a greater role with respect to the identification and management of low-risk CaP in these patients.
At our institution, prostate lesions are graded into three suspicion levels (i.e. low, moderate and high) depending on the number of positive MRI parameters . These parameters include T2-weighted (T2W), diffusion-weighted (DW)-MRI, spectroscopy (i.e. magnetic resonance spectroscopic imaging) and dynamic contrast-enhanced (DCE)-MRI sequences. Low suspicion lesions are defined as being positive on two or fewer of the four parameters (typically T2W and DW-MRI), whereas moderate suspicion lesions are defined as being positive on three of four parameters (typically, T2W, DW-MRI and DCE-MRI) and high suspicion lesions are defined as being positive on all four parameters. We hypothesize that patients with only low suspicion lesions represent a special low-risk group of men with either no disease or clinically insignificant disease, allowing them to be managed conservatively. In the present study, we analyzed a cohort of patients with low suspicion lesions on mpMRI who subsequently underwent image-guided biopsy using a TRUS/MRI fusion-guided platform to determine the prevalence of high-risk CaP in this group.
Patients and Methods
The present prospective study was approved by the Institutional Review Board of the National Cancer Institute of the National Institutes of Health. All patient information was protected in accordance with the Health Insurance Portability and Accountability Act. Patients eligible for the present study provided their appropriate consent and were informed about the potential harms and benefits.
Between May 2007 and November 2011, all patients with a high clinical suspicion for CaP (based on a higher PSA level, strong family history, referral for evaluation from community doctor) underwent mpMRI. Mean (median; range) age was 60 (59; 36–81) years and mean (median; range) PSA level was 7.11 (5.7; 0.3–64.7) ng/mL. Further patient demographics are presented in Table 1.
Table 1. Patient demographics
Race, n (%)
PSA level (ng/mL)
MpMRI was performed at 3.0 Telsa (Achieva; Philips Healthcare, Best, The Netherlands). MRI acquisition was performed using a combination of a six-channel cardiac surface coil (SENSE, Philips Healthcare) placed over the pelvis and an endorectal coil (BPX-30; Medrad, Pittsburgh, PA, USA) as described previously . MRI sequences included triplanar T2W MRI, axial DW-MRI, three-dimensional MR spectroscopy (i.e. magnetic resonance spectroscopic imaging) and axial DCE-MRI. MpMRI was subsequently subjected to blinded centralized radiological evaluation by two experienced genitourinary radiologists (BT and PLC). For mpMRI analysis of the peripheral zone lesions, on T2W MRI images and apparent diffusion coefficient maps of DW-MRI, the criterion for a ‘visible’ lesion was a well circumscribed, round-ellipsoid low-signal-intensity region within the prostate gland . The three-dimensional MR spectroscopy analysis evaluated choline/citrate ratios within the image voxels in the lesion core sites. Voxels were considered abnormal when the choline/citrate ratio was 3 or more sds above the mean healthy choline/citrate ratio value (≥0.373), which was defined as a mean (sd) of 0.13 (0.081) based on results previously recorded from 433 healthy voxels from peripheral zone regions . DCE-MRI images were evaluated by direct visual interpretation of raw dynamic enhanced T1-weighted images and the diagnostic criterion for CaP was defined as a focus of asymmetric, early and intense enhancement with rapid washout compared to the background . For central gland lesions, on T2W MRI images and apparent diffusion coefficient maps of DW-MRI, the criterion for a ‘visible’ lesion was a homogenous low-signal-intensity lesion with irregular margins and no capsule, often invading the pseudocapsule, with lenticular extension into the urethra or anterior fibromuscular zone . The criteria for lesion analysis in the peripheral zone were the same for MR spectroscopy and DCE-MRI sequences. Intraprostatic lesions were categorized as low suspicion if two or fewer parameters (typically T2W and DW-MRI were positive; moderate suspicion if three modalities (typically T2W and DW-MRI and DCE-MRI) were positive; and high suspicion if all four modalities were positive (Fig. 1). Additionally, the prostate gland was manually contoured planimetrically by an experienced radiologist and MRI-derived prostate volumes were calculated.
Patients with lesions suspicious for cancer on MRI were enrolled in our image-guided prostate biopsy protocol, which fuses the MRI to the TRUS. All patients undergoing outpatient, office-based prostate biopsy were given a 3-day course of antibiotic prophylaxis and a cleansing Fleet Enema (CB Fleet Company, Inc., Lynchburg, VA, USA) on the morning of the procedure. Lidocaine jelly and injectable lidocaine for analgesia was used to obtain a periprostatic block. Initially, all patients underwent a standard ‘extended sextant’ 12-core transrectal ultrasonographic biopsy. During this TRUS biopsy, the physician was blinded to the location of suspicious lesions previously detected on the mpMRI. In the same biopsy session, patients underwent a TRUS/MRI fusion-guided biopsy of suspicious lesions found on mpMRI. An electromagnetic field generator was placed above the pelvis to track the real-time transrectal ultrasonographic probe during the biopsy. After a two-dimensional sweep of the rectal probe, the real-time ultrasonographic image was fused to the previous MRI image, allowing the operator to guide the biopsy needle to previously identified suspicious lesions. At least two biopsy cores were taken from each lesion (one in the axial plane and one in the sagittal plane). Details of the biopsy platform and a description of the biopsy technique have been described previously [3, 7, 8]. All biopsies were subjected to blinded centralized pathological evaluation by a single genitourinary pathologist.
Descriptive statistics were used to describe patient characteristics, including patient age, PSA level, PSA density, MRI volume and number of lesions. Student's t-test was used to determine any differences between biopsy-positive and biopsy-negative patients with respect to lesions per patient, MRI prostate volume, PSA level and PSA density.
A total of 800 patients underwent mpMRI of the prostate between 2007 and 2011. Clinical suspicion of CaP (higher PSA level, family history or referral from a community centre) was suspected in all patients who presented for evaluation. We identified 125 (16%) patients with only low suspicion lesions on mpMRI for analysis in this series.
3.0-Telsa mpMRI and MRI/Ultrasonography Fusion Biopsy Results
Of the total cohort of 125 patients, 77 (62%) did not have cancer on either on the TRUS or MRI targeted biopsy. There were 48 (38%) patients who had cancer detected on biopsy (four patients were positive on MRI targeted fusion biopsy alone, 24 were positive on TRUS alone and 20 were positive on both MRI targeted and TRUS biopsy). When evaluating for disparities between disease-positive and -negative patients, no differences were noted in the mean number of lesions per patient, PSA level, prostate volume or PSA density (Table 2).
Table 2. Patient characteristics
Patients, n (%)
Lesions, n (%)
Lesions per patient, n
MRI prostate volume (mL)
PSA level (ng/mL)
PSA density (ng/mL/mL)
Of the 48 patients who had positive biopsies, 38 (79%) patients had Gleason 6 (3+3) disease, whereas 10 (21%) patients had Gleason 7 (3+4) disease (Table 3). The mean (range) number of positive TRUS cores was 2.59 (1–9) per patient. The mean (range) percentage involvement of TRUS biopsy and MR-guided biopsy cores was 20.7% (1–80%) and 21.65% (3–70%), respectively. No patients had primary Gleason 4 disease or higher (Table 4).
In total, 21 patients (18 prostatectomy, three external beam radiation) chose to undergo whole gland therapy at the time of analysis. There were 15 patients who chose to undergo surgical intervention at our institution, whereas three patients underwent surgery at other institutions. Of the patients who underwent surgery at our institution, all were found to have either Gleason 6 (3+3) or Gleason 7 (3+4) disease on postoperative pathology upon review by an experienced genitourinary pathologist. No patients were found to have primary Gleason 4 disease or greater, extracapsular extension or seminal vesicle invasion. All patients had pT2 disease on postoperative pathology.
Active Surveillance Patient Qualification
Of the 48 patients found to have cancer on biopsy, nine patients with Gleason 7 disease were disqualified from active surveillance using 2011 National Comprehensive Cancer Network (NCCN) guidelines as a result of disease grade. Furthermore, four patients with Gleason 6 disease found on MRI-guided fusion biopsy alone were also disqualified because no existing guidelines incorporate this novel biopsy system. Of the remaining patients, 30 (29 patients with Gleason 6 disease and one patient with Gleason 3+4 disease) qualified for active surveillance using 2011 NCCN guidelines (either ‘very low risk’ or ‘low risk’). At the time of analysis, 27 patients were still on an active surveillance at the National Cancer Institute. There were three patients who opted to undergo surgical intervention subsequent to the diagnosis of cancer as a result of personal preference.
Screening for PSA has resulted in a dramatic increase in the number of men diagnosed with CaP. This has led to concerns that an overdiagnosis of indolent cancers is occurring at great human and medical costs. Many small low-grade tumours are inconsequential, yet lead to definitive treatment. mpMRI offers the ability to survey the entire gland for suspicious lesions . Many lesions depicted by MRI are low in suspicion. To determine the value of detecting low suspicion CaP with mpMRI, only patients with a clinical suspicion (higher PSA level, family history, outside referral) of CaP underwent imaging and received a biopsy in our protocol.
The results obtained in the present study show that patients with only low suspicion lesions on mpMRI had a very low likelihood of harbouring high-risk disease. Among the 125 patients found to have only low suspicion lesions on mpMRI, 77 did not have cancer by standard or MRI-guided biopsy, whereas 38 patients had Gleason 6 (3+3) cancer. Thus, 115/125 (92%) of patients with only low suspicion lesions either did not have cancer or had low-risk cancers. Of the 38 patients with Gleason 6 cancer, 29 patients were classified as either ‘very low’ risk or ‘low’ risk under the 2011 NCCN guidelines, qualifying them for active surveillance. There was one patient with Gleason 3+4 disease who also belonged in this group because he had low volume disease and was aged 75 years with a life expectancy of less than 10 years, qualifying him for active surveillance. Thus, using the 2011 NCCN guidelines, 107 (88%) patients with only low suspicion lesions on mpMRI had either no cancer or qualified for active surveillance. With further show that low-risk and low volume CaP should be monitored with active surveillance  and that these patients may avoid radical whole gland treatments along with their associated morbidities.
A total of 18 patients in our cohort did not qualify for active surveillance. There were three patients who were disqualified from active surveillance as a result of a PSA level ≥10 ng/mL but met all other parameters (all had a small focus of Gleason 6 cancer). Upon further analysis, these patients had a mean (range) prostate volume of 103 (69–146) mL on manual MRI planimetric volume estimation. Based on a previous study , PSA level correlates with prostate size at large volumes. There was one patient who had areas of chronic inflammation on biopsy, which is another potential explanation for the high PSA. Because these patients met all other criteria for active surveillance and the high PSA levels are probably explained by their large prostate volumes or inflammation, they may not necessarily have needed to be disqualified from active surveillance. In addition, four patients had Gleason 6 cancer detected on MRI-guided biopsy alone. Because no guidelines have been established for this novel method of biopsy, these patients were not considered for active surveillance. However, all four patients were classified as ‘low risk’ based on the D'Amico classification  (all had one core of Gleason 6 disease) and may represent appropriate patients for active surveillance.
There were nine patients in our cohort who had Gleason 3+4 cancer detected on biopsy and were not eligible for active surveillance. Classical NCCN guidelines categorize most patients with Gleason 3+4 disease as ‘intermediate’ risk, thereby disqualifying them from active surveillance. However, there is increasing evidence to suggest that patients with intermediate-risk CaP are appropriate for active surveillance . Using the validated University of California, San Francisco Cancer of the Prostate Risk Assessment (CAPRA) score [14-16], all of the remaining nine patients with Gleason 3+4 disease scores were between 3 and 5, categorizing them as ‘intermediate’ risk. Cooperberg et al.  reported that patients with intermediate-risk (Gleason 7 or CAPRA score of 3–5) CaP are not necessarily prone to higher rates of progression and, thus, careful active surveillance is a legitimate option for appropriately selected patients in this group.
Studies have shown significant pathological upgrading on radical prostatectomy, with recent studies reporting numbers in the range 20.3–54% [17, 18]. This potentially creates a problem with respect to accurately monitoring a patient with supposedly low-risk disease. In the present study, we found that, in the 15 patients who underwent radical prostatectomy at our institution, disease in the prostate was limited to either Gleason 6 or 7 (3+4) tumours. There were five patients who were upgraded from Gleason 6 disease to Gleason 3+4 disease on postoperative histology, which is consistent with the pathological upgrading from biopsies noted in previous studies [17, 18]. Although some patients were pathologically upgraded on pathology to Gleason 7 (3+4), no patients were upgraded or discovered to have high-risk cancer (Gleason 4+3 or greater) on pathology. In an attempt to elucidate the role of MRI in selecting patients with low-risk disease using postoperative pathology, Guzzo et al.  reported that tumour identification on endorectal MRI was not predictive of adverse pathological features for men who otherwise qualified for active surveillance. However, the patients in that study  were imaged with a 1.5-Telsa scanner and only T2W sequences were used. By contrast, by correlating the MRI with patient-specific MRI-based prostate molds to generate whole-mount histological sections correlated with axial MRI images, Turkbey et al.  showed that the combination of additional MRI parameters significantly increased sensitivity with respect to detecting suspicion lesions. The patients in the present study were imaged using mpMRI and a 3.0-Telsa scanner, allowing greater accuracy when identifying and characterizing each lesion. Although a limited number of patients underwent prostatectomy at our institution, no patients were found to have high-risk cancer on pathology, again indicating that select patients with low suspicion lesions are suitable for active surveillance.
mpMRI is emerging as a tool for urologists in the localization and management of CaP. In the future, mpMRI may be able to localize CaP without the use of a biopsy. If a patient has only low suspicion lesions, the probability that they do not have cancer or will qualify for active surveillance is high (88%). Unlike Choi et al. , who reported that patients on active surveillance are imaged less compared to those who choose surgery, we consider that mpMRI has a significant role in the oncological management of patients with low-risk disease. In total, 27 of the 30 patients (90%) who qualified for active surveillance remain on a surveillance protocol at the National Cancer Institute, which includes MRI at the time of biopsy.
The present study has several limitations. First, the present study was conducted at a single institution with a high level of expertise in mpMRI. We consider that the validation of our hypothesis in a matched cohort from another institution would be of value. Additionally, the mpMRI scoring system used in the present study comprises a simple and non-weighted system in which all sequences had an equal contribution with respect to assessing the suspicion level of lesions. Although emerging data suggest that specific parameters can correlate with the aggressiveness of CaP , a clear, weighted MRI scoring system has not been validated. In the current scoring system used in the present study, each parameter was considered as either positive or negative, whereas a five-point scale can used to grade each parameter in the weighted scale. The reproducibility of this five-point scale, however, has not been well established. Currently, international efforts are being made to develop such a grading system for mpMRI .
In conclusion, mpMRI represents a non-invasive technique for the localization of CaP that allows the clinician to stratify the risk of CaP based on MR-determined suspicion of the lesions. The results obtained in the present study show that low suspicion lesions on mpMRI tend to predict for the absence of high-risk cancer and the presence of low-grade disease or no disease. In those individuals with only low suspicion lesions, comprising ≈ 16% of our population, the risk of significant disease is sufficiently low to consider deferring biopsy or, if cancer is found by biopsy, to choose active surveillance. These single-institution results should be verified in multiple institutions, which would help to determine the role of mpMRI in the management of low-risk CaP.
This research was supported (in part) by the Intramural Research Program of the National Cancer Institute, National Institutes of Health.