Guidance of treatment decisions in risk-adapted primary radiotherapy for prostate cancer using multiparametric magnetic resonance imaging: a single center experience

MRI of the prostate has already been established as a valuable tool for radiation oncologists in improving target volume delineation [35-37] and has also been investigated for treatment planning [38]. It has been shown that morphologic MRI enables a more reproducible definition of the prostate gland [39-41], particularly when it comes to the definition of the prostate apex [42,43]; it leads to a smaller CTV volume and the sparing of more normal tissue relative to CT-based target volume definition [44]. Consequently, in prostate cancer it has been suggested that MRI-based treatment may reduce treatment-associated toxicity [30].

Apart from precise organ delineation, the accurate radiological tumor staging of prostate cancer is playing an increasingly important role in the era of IMRT [8]. Using highly-conformal RT techniques and small planning target volume margins, the detection and inclusion of extracapsular disease and seminal vesicle infiltration into the treatment volume is of paramount importance in avoiding a geographic miss and consequent tumor under-dosage [45].

MRI of the prostate can offer additional benefit in RT planning because it is regarded as the most reliable imaging modality regarding the determination of the local extension of prostate cancer, and it exhibits the highest congruence with the “gold standard” of post-operative pathological staging [10,11].

Surgical series have demonstrated that both the digital rectal examination, as well as other radiological techniques such as CT scans of the abdomen and transrectal ultrasound, only exhibit low correlations with definitive pathological staging after radical prostatectomy; indeed, they only accurately predict the extent of the local tumor 61–70% of the time [19,20]. In contrast, morphological T1- and T2-weighted MR imaging alone has already shown an accuracy of 84% for the detection of extraprostatic disease in clinically staged T2 prostate cancer [46]. In recent years, mpMRI sequences have been introduced into prostate cancer imaging in order to further enhance tumor staging [14]. As a result of the early rapid enhancement and early washout of prostate cancer tumors, DCE-MRI provides a means of determining tumor extension more accurately than morphological imaging [10], and has demonstrated an improved overall staging accuracy of 95% in a single-institution series [17]. Interestingly, this improvement in prostate cancer tumor staging due to DCE-MRI has been found to be particularly important for less experienced readers [47]. Likewise, recent literature reviews have confirmed that additional DWI can detect prostate cancer tumor foci more accurately than conventional anatomic imaging alone [12,48]. For these reasons, mpMRI of the prostate has been established in our institution as a standard imaging procedure in all patients before primary RT for prostate cancer. Although this approach is not uncommon among RT units, it cannot at the moment be considered as a modality with interdisciplinary consensus. Current guidelines recommend the use of mpMRI for local staging of prostate cancer ahead of curative therapy mainly for high-risk patients [28,31], whereas its value is questioned in patients presenting with low-risk features [49]. The purpose of the present study was therefore to analyze retrospectively the extent of the shift in tumor stage as a result of the use of mpMRI, and the consequent impact on risk group distribution and treatment decisions in primary RT for prostate cancer.

In our patient series, a considerable impact on prostate tumor staging as a consequence of the use of mpMRI was observed, with tumor upstaging in 43.4% and downstaging in 12.3% of patients. These findings are consistent with the findings from a previously published series of 199 patients with localized prostate cancer; in this series morphological MRI scans of the prostate, without DCE imaging or DWI, led to tumor upstaging in 52% and to tumor downstaging in 3% of patients before RT [25].

Although our study showed a trend towards a significantly lower probability of MRI-based tumor upstaging in patients with an intermediate or low-risk PSA level, there was no subgroup identifiable where it appeared reasonable to omit mpMRI before RT. Even in the low-risk subgroups (Gleason score < 7; PSA level < 10 ng/ml), a substantial upstaging of 45–50% could be observed. However, taking all of the established risk factors, such as the Gleason score, PSA level and tumor stage together, mpMRI of the prostate only led to a change in risk group in 25.4% of the patients in our series. This might be explained by the fact that patients with a locally advanced tumor stage detected by MRI may often present with other adverse features which had already been assessed before MRI.

Tumor stage shift and change in risk group taken together resulted in a change in the RT dose and target volume, as well as in the recommendations for additional ADT in 30% of patients. These findings are in concordance with recent data from a study by Chang et al. [24] who reported a significant upstaging of 29% before primary RT for prostate cancer in a comparable patient series, using morphological MRIs without multiparametric dynamic sequences. Interestingly, Clure et al. reported a similar extent of treatment adaptation of 27% as a result of MRI staging in a series of 104 patients who underwent robotic-assisted laparoscopic prostatectomy [50]. Similarly, Hricak et al. reported a change in surgical plan based on MRI scans in 39% of patients who underwent radical prostatectomy for localized prostate cancer [51].

Whereas it seems to be legitimate to increase treatment intensity for patients with mpMRI based upstaging, caution should be exerted for treatment de-intensification in cases where mpMRI led to downstaging: in our series, a considerable number of downstaging was found in patients who received ADT before RT. It is, however, mainly not determinable whether MR-based downstaging occured due to previous diagnostic inaccuracy of the pre-MRI staging or if it was an ADT effect. If the latter can’t be definitely ruled out, downstaging should not subsequently lead to de-intensification of radiotherapy. In order to avoid this uncertainty of ADT interference, mpMRI should be conducted as part of the initial staging procedure before start of any neoadjuvant treatment.

Our study had several limitations. First, all data were reviewed retrospectively and staging information was obtained from patient histories. In addition, the reporting radiologists had not been blinded to previously obtained clinical staging. However, these circumstances may represent the daily practice in a major cancer center more accurately than would be the case in a dedicated prospective trial. Second, clinical and MRI staging were not correlated to oncological outcome as has been reported in other studies [25,52]; this was because the follow-up time was limited as we evaluated a very recent patient series. The main focus of our study was on the impact of MRI staging on treatment decisions, not on its predictive value. mpMRI sequences were not reported separately from morphological MRI, so their additional value for treatment decisions could not be independently assessed. Additionally, in contrast to studies regarding the surgical treatment of prostate cancer, MRI staging could not be validated using definitive pathological staging. Finally, the impact of MRI on target volume delineation and treatment volume relative to CT-based planning was not analyzed in the current study, because it has been investigated extensively by others [29,30,53].