To our knowledge, this is the first study to report on MRI parameters in a collective incorporating only patients harboring PC of ISUP GG 4 and 5. MRI was highly sensitive in detecting these high-grade PC with specific MRI characteristics like almost entirely PI-RADS 4 and 5 classification, primarily high lesion diameter and LCC, low ADC value, and focal enhancement on DCE. A combination of clinical, biopsy, and MRI parameters had the best discrimination between men suffering from BCR in follow-up as compared to clinical and/or biopsy parameters only.

When focusing on the PSA values of our study population, it is notable that 8 patient revealed a PSA below 4 ng/ml. Previously published results showed the reduced sensitivity of PSA as a reliable predictor for both the diagnosis and the risk stratification for disease recurrence [10, 15, 18]. Furthermore, we observed a relevant number of patients with PSAD below 0.15 ng/ml/ml. Although, PSAD has been described as an additional decision guidance in indeterminate cases and equivocal lesions, it is not always trustworthy due to prostate volume enlargement (especially in patients with benign prostate hyperplasia) [19]. To our experience, it may especially help in patients with smaller prostate volumes as a supportive tool for referring patients to biopsy. On the other side, we included 80 of 145 patients with a PSA value above 10 ng/ml. None of them were younger than 55 years (mean age 71 years). MRI showed T3 stage with extracapsular extension in 48 of 80 cases, so PC was already advanced at diagnosis in most of the patients. If screening and PSA control would have been conducted earlier in these patients, PC might be detected in a less advanced stage. As there is an ongoing discussion about a commonly accepted PC screening strategy, our results reinforce the role of mpMRI.

Moreover, ISUP GG (Gleason Grade, respectively) in biopsy seems to be an important parameter to predict recurrence, a fact that we could demonstrate in our model as well [20]. Interestingly, we observed a relevant number of PC only detected in targeted biopsy and furthermore 36 patients with higher ISUP GG at targeted biopsy than in the systematic cores, which underlines the important role of pre-biopsy MRI in the high-risk cohort as well. Although PC tends to be larger at this stage, there is still a risk of missing the tumorous areas in case of performing systematic biopsy alone. However, the correct ISUP grade of MRI-targeted biopsy is a fact that has been discussed controversy, as a previous study reported downgrading of Gleason Grade at subsequent histopathology after RPE [21, 22]. We observed similar values for upgrading and downgrading after surgery. What is more important are the cases, where histopathology of biopsy cores underestimated the tumors ISUP GG. These are cases where MRI can contribute and may play a key role for clinical decision-making. As these tumors can clearly be depicted and appear aggressive on MRI, clinicians should be warned if biopsy reveals only low- or intermediate risk cancers. As MRI can reliably grade cancer aggressive, re-biopsy should be conducted to minimize the risk of missing high-grade PC [5].

Comparing patients with BCR at follow-up and cancer-free survival after RPE, we observed remarkable differences not only in clinical parameters and RPE histopathology, but also in defined MRI parameters. Our data support recent evidence that data derived from MRI can contribute to identify patients with higher and lower risk for BCR and may play a role to predict recurrence [14, 15, 21, 23]. Although, all our patients would be categorized as high risk (following EAU guidelines) based on their ISUP GG, the data suggest discrepancies at baseline which might have an influence on prognostic outcome. The correlation between ADC values and PC aggressiveness has been described earlier [7, 24].

Moreover, MRI has the ability to predict extracapsular extension and cT stage and is a valid predictor in PC staging [25, 26]. For non-metastatic disease in the high-risk group, surgical treatment remains the first option and adjuvant treatment (e.g., radiation therapy or androgen deprivation) might be necessary in some cases, based on the results of RPE histopathology.

In our multivariate logistic regression analysis incorporating information from PSA and biopsy results, PSAD and infiltration in biopsy core were independent parameters to estimate the risk of disease recurrence. These findings are not surprising and in accordance with previous studies showing the important role of Gleason grade [26]. When including MRI parameters into the multivariate regression analysis, the MRI T3 stage and PSAD correlated significantly with the risk for BCR. This is consistent with published data from pathology studies, showing an association between BCR and pathological T stage, pathological Gleason grade, and positive surgical margins [27]. As previously mentioned, extracapsular extension can be determined on MRI, and we observed a good correlation of MRI T stage with the histopathology of RPE [26]. A possible explanation for MRI T3 stage being the only independent MRI parameter could be a very high collinearity with maximum lesion diameter and LCC, which could mask their effect in the proposed analysis. There is already evidence in the literature about the benefit of MRI parameters in several prediction models [15, 28, 29]. However, these studies did not investigate on patients with a high-risk constellation only. As these groups of patients face higher risks for metastasis and prostate cancer-related death, a detailed knowledge about individual risk is helpful in clinical settings.

Some limitations of this study, besides the retrospective, single-center design, need to be discussed. First, mpMRI images were rated by individual radiologists. The rating might be influenced by experience, local in-house standards, and personal preference. This could probably bias the diagnostic performance of mpMRI. Second, we did not incorporate the information from RPE histopathology, as this study focused on the additional use of MRI in a high-risk collective at diagnosis. This means that we included patients for the analysis regardless of their R1-status. R1-status is known to have an influence on the risk for BCR and may act as a confounder. RPE histopathology contains important knowledge about cancer aggressiveness and can predict BCR at follow-up. Finally, at follow-up and subgroup analysis, we focused on patients treated with RPE only.

In conclusion, all high-grade PC could be clearly identified on mpMRI. PI-RADS classification was 5 in over 50%, lesion diameter and LCC tend to be higher than 12 mm, ADC was below 800 × 10−6 mm2/s, and DCE showed focal enhancement in over 90%. Higher MRI staging parameters next to higher PSAD seems to be associated to higher risk of BCR at follow-up and so may help clinicians to earlier identify these patients. However, the detection of advanced PC stages in older men might be reduced by standardized PSA screening incorporating MRI.