- To evaluate predictors of more aggressive disease and the role of multiparametric 3.0-T magnetic resonance imaging (MRI) in selecting patients with prostate cancer for active surveillance (AS).
Active surveillance (AS) is gaining in popularity as a treatment option for selected men with low-risk localised prostate cancer, because it reduces the risk of overtreatment of patients with insignificant disease [1, 2]. The rationale for AS is that most low-risk prostate cancer has an indolent course and the slow growth rate allows sufficient time during follow-up to detect cancers destined to become more aggressive during a window of curability .
The success of AS as a management strategy for prostate cancer relies primarily on the accurate identification of patients with low-risk disease unlikely to progress . These patients are usually identified based on DRE results, PSA kinetics, and repeat biopsy pathology findings. Application of this diagnostic pathway can lead to reclassification of cancers as more aggressive cancer in 20–30% of patients [1, 5].
Prostate MRI is accepted as one of the best imaging methods for detecting and staging prostate cancer because it produces excellent anatomical images of the prostate gland . It has been used with varying levels of effectiveness in an attempt to more accurately stage patients with newly diagnosed prostate cancer . Multiparametric MRI can provide additional information about tissue microvascular properties (dynamic contrast-enhanced imaging, DCEI) and facilitates lesion characterisation (diffusion-weighted imaging, DWI). Multiparametric MRI has been shown to be an efficient method for localising and detecting tumour foci within the prostate . Currently, the focus of prostate cancer studies using MRI is shifting from the diagnosis of advanced prostate cancer to the identification and accurate location of low-volume prostate cancer.
We hypothesised that tumour visibility on MRI might allow identification of patients suitable for AS. We therefore investigated clinical factors associated with the risk of more aggressive disease and the role of multiparametric 3.0-T MRI in selecting patients with clinically low-risk prostate cancer eligible for AS.
We retrospectively reviewed our radical prostatectomy (RP) database from June 2005 to December 2011, with appropriate Institutional Review Board approval. Informed consent was waived. From this database, which contained 1703 consecutive RP cases at the time of the study, we identified patients who underwent preoperative 3.0-T MRI for staging and met the Prostate Cancer Research International: Active Surveillance (PRIAS) criteria of clinical stage T1c or T2 disease, a PSA level of ≤10 ng/mL, a Gleason score of ≤6, a PSA density (PSAD) of <0.2 ng/mL2, and fewer than three positive biopsy cores . We excluded patients who underwent a prostate biopsy or MRI at other institutions and the patients who had <12-core specimens after prostate biopsy. We identified 345 (20.3%) patients who fulfilled the PRIAS criteria for AS. Of these, 47 patients were excluded and 298 patients were enrolled in this study. All 298 patients had had preoperative 3.0-T MRI after prostate cancer was identified on prostate biopsies. The mean interval between MRI examination and RP was 16 days.
MRI examination was performed at least 3 weeks after TRUS-guided biopsy. Before MR scanning, all patients were injected i.m. with 20 mg butyl scopolamine (Buscopan; Boehringer, Ingelheim, Germany) to reduce bowel peristalsis. Patients underwent MRI using a 3.0-T scanner (Intera Achieva 3T; Philips Medical System, Best, the Netherlands) with a phased-array coil. An axial T1-weighted turbo spin-echo sequence, multi-planar turbo spin-echo T2-weighted imaging (T2WI), DWI in the axial plane, and DCEI were performed after i.v. injection of 0.1 mmol/kg gadopentetate dimeglumine (Magnevist; Schering, Berlin, Germany). All MR images were interpreted independently in a routine clinical setting by two radiologists who each had >7 years’ experience in prostate MRI; Radiologist 1 and 2 interpreted prostate MRI from 150 and 149 patients, respectively. Both readers were aware that the patients had biopsy confirmed prostate cancer, but were ‘blind’ to other clinical and pathological findings. On T2WI, prostate cancer was defined as a hypointense region in the peripheral zone relative to the adjacent parenchyma, excepting an area of high signal intensity on T1-weighted image suggestive of haemorrhage after prostate biopsy. Prostate cancer was defined as a focal hyperintense lesion at b = 1000 s/mm2 of DWI, with a low focal apparent diffusion coefficient (ADC) value relative to that of the adjacent normal parenchyma on the ADC map images. On DCEI, prostate cancer was defined as a hyperintense area in the peripheral zone relative to a normal peripheral zone, especially within 1 min after bolus contrast injection [10, 11]. The likelihood of the presence of cancer was assigned using a Likert scale between 1 and 5 (1, not suspect; 2, hardly suspect; 3, ambiguous; 4, suspect; 5, highly suspect). The assigned scores of 3–5 were considered positive, and scores of 2–1 were considered negative for cancer. The likelihood score was assigned to each patient in the present study. In cases of several lesions in a patient, we considered them as positive if at least one of the lesions scored >2.
All surgically removed specimens were coated with India ink and fixed overnight in 10% buffered formalin. Transverse step sections were made at 3-mm intervals in a perpendicular plane to the prostatic urethra and were embedded in paraffin. After microsectioning, the tissue slices were stained with haematoxylin-eosin. One of two uropathologists unaware of the MRI results recorded the percentage tumour volume, prostate size, Gleason score, presence or absence of extracapsular extension (ECE), and seminal vesicle invasion (SVI).
Patients were classified into two groups: those with one or more visible cancer lesions on MRI (Fig. 1), and those without visible cancer lesions. Patients with no evidence of tumour on all MR images (T2WI, DWI, DCEI) were included in the no-visible-cancer group (Fig. 2). Preoperative clinical variables (age, body mass index [BMI], PSA level, prostate volume, PSAD) and pathological findings (ECE, SVI, Gleason score, positive surgical margin, percentage tumour volume) from RP specimens were compared between the two groups. Unfavourable disease was defined based on a Gleason score of >6 (upgrading) and/or a pathological stage >pT2 (upstaging) upon examination of the RP specimens.
Continuous variables were compared using the independent t-test or Mann–Whitney U-test, while categorical variables were compared using the chi-square test or Fisher's exact test. Univariate and multivariate logistic regression analyses were used to identify potential preoperative parameters predicting unfavourable disease at final pathology. A P < 0.05 was considered to indicate statistical significance.
The preoperative and histopathological characteristics of the 298 patients with prostate cancer who underwent multiparametric MRI and met the PRIAS criteria are shown in Table 1. In all, 21 patients (7%) were upstaged, 136 patients (45.6%) were upgraded, and 142 patients (47.7%) were diagnosed with unfavourable disease.
|No. of patients||298|
|Median (sd) age, years||65 (6.6)|
|BMI, kg/m2||24.3 (17.5–30)|
|Preoperative PSA level, ng/mL||4.1 (1.2–9.1)|
|Prostate volume, mL||36 (14–85)|
|PSAD, ng/mL2||0.11 (0.03–0.19)|
|Total number of biopsy cores,||12 (12–15)|
|1 positive core||209 (70.1)|
|2 positive cores||89 (29.9)|
|Gleason score 6||298 (100)|
|RP Gleason score:|
|Positive surgical margin||18 (6)|
|Upstaging (> T2)||21 (7)|
|Upgrading (Gleason score >6)||136 (45.6)|
|Unfavourable disease||142 (47.7)|
When the patients were classified according to the presence or absence of cancer on MRI, 263 (88.3%) of the 298 patients were included in the visible-cancer group, while 35 (11.7%) were included in the no-visible-cancer group. BMI, PSA level, prostate volume, PSAD, and positive surgical margins were not significantly different between the two groups. Patients in the no-visible-cancer group were younger on average than those in the visible cancer group (62.6 vs 65.3 years, P = 0.02). The incidence of upgrading and unfavourable disease was significantly lower in the no-visible-cancer group than the visible cancer group (14.3% vs 49.8% and 14.3% vs 52.1%, respectively, P < 0.001). Additionally, the no-visible-cancer group had a smaller percentage tumour volume than the visible cancer group (1% vs 5%, P = 0.007). The incidence of upstaging was not significantly different between the two groups (Table 2).
|No visible cancer||Visible cancer|
|No. of patients (%)||35 (11.7)||263 (88.3)|
|Mean (sd) age, years||62.6 (6.6)||65.3 (6.5)||0.02|
|BMI, kg/m2||23.9 (20.8–29.5)||24.4 (17.5–30)||0.46|
|PSA level, ng/mL||4.4 (2.9–8.2)||4.0 (1.2–9.1)||0.07|
|prostate volume, mL||35 (25–81.5)||36.8 (14–85)||0.68|
|PSAD, ng/mL2||0.10 (0.05–0.19)||0.09 (0.03–0.19)||0.11|
|T2||35 (100)||242 (92)|
|T3 (upstaging)||0 (0)||21 (8)|
|ECE||0 (0)||19 (7.2)||0.14|
|SVI||0 (0)||2 (0.8)||1.00|
|Prostatectomy Gleason score:||<0.001|
|6||30 (85.7)||132 (50.2)|
|7||5 (14.3)||128 (48.7)|
|8–10||0 (0)||3 (1.1)|
|Upgrading (>6)||5 (14.3)||131 (49.8)||<0.001|
|Unfavourable disease||5 (14.3)||137 (52.1)||<0.001|
|Positive surgical margin||1 (2.9)||17 (6.5)||0.71|
|Median (range) % tumour volume||1 (0.1–20)||5 (0.6–50)||0.007|
Among the 35 cases with no visible cancer on MRI, the median (range) size of the five cases with Gleason score 7 and 30 cases with Gleason score 6 on RP specimen was 1.03 (0.06–1.79) mL and 1.2 (0.06–6.2) mL, respectively. There was no statistically significant difference between the groups. Furthermore, there was no definite evidence of prostate cancer or prostatitis when we retrospectively reviewed prostate MRIs of the five cases with no visible cancer and with Gleason score 7 according to the final pathological report.
Univariate analysis of the postoperative histopathological results showed a strong correlation between unfavourable disease and a visible cancer lesion on MRI (P < 0.001), patient age of ≥65 years (P = 0.009), and PSAD of >0.08 ng/mL2 (P = 0.01). Multivariate logistic regression analysis revealed that a visible cancer lesion on MRI (P < 0.001, odds ratio [OR] 6.4), a PSAD of >0.08 ng/mL2 (P = 0.004, OR 2.41), and a patient age of ≥65 years (P = 0.008, OR 1.95) were significant predictors of pathologically unfavourable disease (Table 3).
|OR (95% CI)||P||OR (95% CI)||P|
|Age ≥65 years||1.87 (1.17–2.99)||0.009||1.95 (1.19–3.2)||0.008|
|MRI finding (visible vs not visible)||6.52 (2.46–17.33)||<0.001||6.4 (2.36–17.37)||<0.001|
|Positive cores (two vs one)||1.63 (0.99–2.69)||0.06||1.4 (0.83–2.37)||0.21|
|PSAD >0.08 ng/mL2||2.08 (1.17–3.69)||0.01||2.41 (1.33–4.39)||0.004|
|BMI, kg/m2||1.01 (0.93–1.1)||0.86|
|Prostate volume, mL||0.98 (0.96–1.01)||0.33|
|PSA level, ng/mL||0.96 (0.83–1.11)||0.58|
Of the 298 patients in the present study who met the criteria for AS, there were statistically significant differences in the incidence of upgrading and unfavourable disease between those who had cancer lesions visible on MRI and those who did not.
Several previous studies have evaluated the role of MRI in patients with prostate cancer. Fradet et al.  retrospectively reported on 114 men with biopsy confirmed prostate cancer who met AS criteria. They found that a lesion suggesting cancer on MRI conferred a three-fold increased risk of overall cancer progression. The present findings are also consistent with those of Margel et al. . They prospectively assessed 60 consecutive patients with clinically localised prostate cancer who fulfilled AS criteria by performing DCEI and DWI in addition to T2WI. When no cancerous lesion was identified on MRI, the chance of reclassification on repeat biopsy was only 3.5%, which is extremely low.
Cabrera et al.  retrospectively reported on 93 patients with prostate cancer selected for AS who underwent endorectal MRI and MR spectroscopy. They found no association among clinical stage, serum PSA level, Gleason score and apparent tumours on endorectal MRI. However, the only endpoint of that study was high PSA velocity . In another retrospective study, Guzzo et al.  evaluated endorectal MRI of 172 men with prostate cancer who would have qualified for AS. The authors found that discrete tumour identification on endorectal MRI was not predictive of adverse pathological features (upgrading, upstaging) in patients who otherwise qualified for AS. However, all patients in that study underwent only T1-weighted imaging and T2WI without functional imaging sequences for staging.
In the present study, we found that tumour visibility on MRI (P < 0.001, OR 6.4), higher PSAD (P = 0.004, OR 2.41), and old age (P = 0.008, OR 1.95) were significant independent predictors of unfavourable disease. Our present findings suggest that patients with prostate cancer without a discrete tumour on multiparametric MRI are more suitable for enrolment in AS than those with a visible tumour on multiparametric MRI. Many previous studies have reported that functional imaging sequences, e.g. DCEI and DWI, improve the accuracy of prostate cancer localisation. DCEI using a gadolinium-based contrast medium is considered the most sensitive sequence for identification and staging of organ-confined peripheral or transition zone prostate cancer [16, 17]. However, DCEI alone is not specific, as it can show false-positive results in prostatitis or BPH , and Noworolski et al.  showed that DCEI has low sensitivity for detecting small or low-grade prostate cancer due to partial volume averaging with normal tissue and reduced angiogenesis in small tumours. Several studies have documented reliable correlations between ADC values obtained with DWI and Gleason scores [20, 21]. Other groups have also shown that less differentiated and dense prostate cancers have lower ADC values, better contrast, and higher detection DWI rates than more differentiated and less dense prostate cancers [22-24].
A research group from Johns Hopkins analysed PSA kinetics and biopsy results of 376 patients from an AS cohort . They concluded that a PSAD of >0.08 ng/mL2 predicted progression on biopsy (P < 0.001). San Francisco et al.  validated a PSAD threshold value of >0.08 ng/mL2 as a predictor of subsequent progression. The present results support the importance of PSAD as a significant predictor of unfavourable disease. However, the ideal threshold value for PSAD has yet to be determined.
Suardi et al.  evaluated PRIAS criteria according to age, and found that in patients aged >70 years, the rate of unfavourable prostate cancer characteristics was 41% compared with 23.2% and 24.1% in patients in younger age tertiles (age 63 years and 63.1–69 years, respectively). In another analysis of the PRIAS criteria, El Hajj et al.  reported that a threshold of 65 years (age <65 years) was a significant predictor of favourable disease (P = 0.005, OR 1.61).
An interesting finding of the present study was the relationship between upstaging and tumour visibility on MRI. There was no upstaging in the no-visible-cancer group; nevertheless, the incidence of upstaging was not significantly different between the visible-cancer group and the no-visible-cancer group. This could be due to the fact that only 21 patients had pathological stage T3 disease. A larger, multicentre study is needed to explore this relationship in more detail.
The present study has some limitations. First, this was a retrospective study, and there was a selection bias in that only patients who underwent RP were enrolled. Furthermore, we could not evaluate the relationship between our MRI findings and ultimate disease progression in AS patients. Second, all multiparametric 3.0-T MRI interpretations were performed with the same diagnostic criteria in a clinical setting by two radiologists without central review. There may therefore have been reader variation; however, we argue that our approach more accurately reflects how MRI findings are analysed in clinical practice. Third, we did not analyse the correlation between visible cancer lesions on MRI and the final specimen histopathology in all cases. Because in the present study, we focused not on the cases with lesions that were visible on MRI but on those that were not visible on MRI. Also, all patients in the present study had already been diagnosed with prostate cancer from prostate biopsies. Fourth, we selected the present study population based on PRIAS criteria, thus the generalizability of our results is limited to patients with prostate cancer who meet these criteria for AS. Nevertheless, our observations strongly support a role for multiparametric MRI in selecting patients with prostate cancer eligible for AS.
In conclusion, we found that clear tumour identification on 3.0-T multiparametric MRI (combination of T2WI, DCEI, and DWI) was predictive of adverse histopathological features in patients with prostate cancer eligible for AS. Multiparametric MRI could potentially be used to identify patients with prostate cancer eligible for AS as an initial management strategy. Furthermore, PSAD and age could help clinicians select appropriate candidates for AS.
The authors declare that they have no relevant financial interests.
apparent diffusion coefficient
body mass index
dynamic contrast-enhanced imaging
the Prostate Cancer Research International: Active Surveillance (criteria)
seminal vesicle invasion