Introduction

Prostate cancer is one of the most common cancer types in Europe [1] and primary radiotherapy is an established curative therapy option [2, 3]. Intensity-modulated radiation therapy and volumetric modulated arc therapy (VMAT) based techniques allow precise and high conformal radiation of the prostate with a favorable toxicity profile [4, 5]. While whole-gland radiotherapy and resection of the prostate are standard treatment modalities for prostate cancer, focal therapy trials are justified by recent findings that a dominant intraprostatic lesion, also called index lesion, determines prognosis and disease progression [6,7,8]. Local recurrence occurs at the site of the primary tumor [9, 10] which provides the rationale for focal therapy trials in a high proportion of patients [11]. Focal therapy of the prostate has been addressed recently by a position paper of the European Association of Urology wherein prospective recording of outcomes for this form of less radical treatment is encouraged [12]. In a recent radiotherapy-specific review the authors concluded that focal brachytherapy can be applied reasonably with decreased toxicity compared to whole-gland radiotherapy especially for low-risk prostate cancer patients [13]. Furthermore, contemporary efforts are underway to escalate the radiotherapy treatment to improve tumor control and limit toxicity to neighboring organs at risk (OAR) by applying a focal radiotherapy boost to the dominant intraprostatic lesion by MRI-guided brachytherapy or MRI-guided external beam radiation [14,15,16,17]. As MRI is increasingly becoming an integral part of diagnosis and treatment of prostate cancer, the aim of this retrospective single-institution trial was to study the possibility to detect and localize the index lesion with radiotherapy treatment planning MRI, to compare the results with pretreatment prostate punch biopsies and to analyze the dose distribution based on the anatomic location of the index lesion.

Materials and methods

The planning MRI, treatment plans and charts of 54 patients treated from March 2015 until March 2017 for low to high risk prostate cancer at the Radiation Oncology Department of the Universitätsklinikum Würzburg were reviewed. Radiotherapy was delivered according to institutional standard with volumetric modulated arc therapy in 33 fractions with simultaneous integrated boost and two dose levels of 1.82 and 2.31 Gy per fraction, resulting in a prescribed PTVBoost mean dose of 76.23 Gy and in a PTV dose of 60.06 Gy (prescribed to D95). In case of pelvic lymphatic radiation the prescribed dose was 45.5 Gy (D95) with 1.82 Gy per fraction. CTVP-SV consisted of the prostate without the seminal vesicles. PTVBoost was defined by a 5 mm margin around the CTVP-SV with avoidance of the rectum. The PTV was created by a 10 mm margin around the prostate including the seminal vesicles (CTVP+SV) in all but the dorsal direction, where a 7 mm margin was used.

All patients received pretherapeutic treatment planning MRI using a Siemens MAGNETOM Prisma with 3 T without an endorectal coil. MRI consisted of a native T2-weighted sequence, diffusion-weighted imaging (DWI; b‑value 800) with apparent diffusion coefficient (ADC) mapping and a contrast-enhanced T1-weighted sequence. Both T2-weighted sequence and ADC map were fused onto the radiotherapy treatment computed tomography (CT) within the Pinnacle3 (Philips Radiation Oncology Systems, Fitchburg, WI, USA) planning system which allowed the contouring of the DIL by one radiation oncologist trained in multiparametric MRI (mpMRI) of the prostate and the analysis of the dose distribution (Fig. 1).

Fig. 1
figure 1

Prostate cancer treatment plan showing an index lesion in the peripheral zone: a dose distribution, b native T2-weighted image, c apparent diffusion coefficient map

Full size image

The index lesion was defined as the lesion with the highest risk features based on the Prostate Imaging Reporting and Data System (PI-RADS) version 2 criteria [18]. If two similar lesions were present, the largest lesion was chosen. All patients had pathologic confirmed prostate cancer with diagnosis by TRUS-guided biopsy and in some patients additionally by transurethral prostate resection. The location of the DIL was documented by means of a sector map of the prostate (Fig. 2) and correlated with the pathology reports of the punch biopsies which provided the corresponding side of the cores and in 33 cases detailed location data. Statistical analysis was conducted with IBM SPSS v. 24.0 (IBM Corp., Armonk, NY, USA). An unpaired t‑test was used for the dose difference analysis between ventral and dorsal index lesions. Changes in accuracy were analyzed by applying a McNemar test. Differences were considered statistically significant in case of p < 0.05. For the mean dose analysis a 95% confidence interval (CI) of the difference in mean dose was calculated. Mean dose values are reported with standard deviation.

Fig. 2
figure 2

Sector map of the prostate, modified after Weinreb JC et al. [18]. AFS/AS anterior fibromuscular stroma, CZ central zone, PZ peripheral zone, TZ transition zone, US urethral sphincter. a: anterior, p: posterior, m: medial, l: lateral

Full size image

Results

Clinical characteristics

The reviewed 54 men had a median age of 74 years (range 54–81 years). The PSA value at first diagnosis (iPSA) was in 39 patients below 10 ng/ml, in 8 patients between 10–20 ng/ml and in 7 patients >20 ng/ml. Regarding the Gleason score, 11 patients had a score of 6, 23 patients a score of 7 and 20 patients had a score of 8 to 10. According to the D’Amico risk classification 10 patients had low-risk prostate cancer, 20 patients an intermediate-risk prostate cancer and 24 patients high-risk prostate cancer. All patients were diagnosed by pretreatment TRUS-guided biopsy. A total of 11 patients had prior transurethral prostate resection. Clinical characteristics are summarized in Table 1.

Table 1 Patient characteristics
Full size table

Location data and pathology analysis

In 10 out of 54 patients the location of the DIL was on the left side, in 34 patients on the right side and in 10 patients on both sides of the prostate. In 32 out of 54 patients the location of the DIL involved the peripheral zone, in 18 patients the transition zone, in 5 patients the anterior fibromuscular stroma, in 6 patients the central zone and in 3 patients the seminal vesicles. In 44 out of 54 patients the middle layer, in 32 patients the basis and in 23 patients the apex of the prostate was involved. Location data are summarized in Table 2.

Table 2 Location of the dominant index lesions
Full size table

Overall 82 intraprostatic lesions could be identified: 54 index lesions and 28 secondary lesions. The median volume of the index lesion was 1.66 cm3 (range 0.5–26.1 cm3). MRI-guided localization of the index lesion reached an accuracy of 68.3%, sensitivity of 66.7% and specificity of 75.0% in a side-based comparison with the TRUS-guided biopsy reports (n = 52; two patients not eligible because of lacking data). By inclusion of secondary lesions the accuracy could be significantly improved to 78.8% (McNemar test, p = 0.007, two-sided) with a sensitivity of 82.1% and a specificity of 65.0%. In a subgroup of 33 patients the pathology reports of the punch biopsy offered sextant location data which were compared to the prostate sector map (Fig. 2). In this subgroup accuracy was 66.2%, sensitivity 51.4% and specificity 82.8% for the index lesion. Including secondary lesions sextant comparison reached an accuracy of 67.2%, a sensitivity of 59.0% and a specificity of 76.3%.

Dose distribution analysis

In all, 41 out of 54 patients received VMAT-based radiotherapy of the prostate only and 13 patients received additional treatment of the pelvic lymph nodes. In case of low-risk prostate cancer reduction of delivered fractions from 33 to 32 was optional, resulting in 2 patients receiving a PTVBoost dose of 73.92 Gy (Dmean). One patient with lymph node metastasis received 30 fractions, resulting in a PTVBoost dose of 69.3 Gy (Dmean) in addition to antiandrogen treatment.

These three above described patients with an alternative fractionation were excluded from the subsequent analysis of the index lesion dose values because our aim was to detect deviances from the planned PTVBoost Dmean of 76.23 Gy in institutional standard fractionation. The mean Dmean of the DIL was 77.51 ± 0.77 Gy. Mean minimum dose was 74.55 ± 2.67 Gy and mean maximum dose was 79.12 ± 0.81 Gy. In 50 out of 51 patients Dmean DIL was within tolerance range of ±1% or above the prescribed dose. In most patients (39 out of 51) Dmean DIL was in the range of +1 to +3%. In 2 patients Dmean DIL was in the range of +3 to +5% and in 1 patient Dmean DIL was in the range of −1 to −2% (−1.4%).

There was a significant difference in mean Dmean between index lesions located in the ventral part of the prostate (peripheral zone anterior, transition zone anterior/posterior, anterior fibromuscular stroma; n = 21) and in the dorsal part (peripheral zone posterior lateral/medial, central zone, seminal vesicles; n = 30) with 77.87 ± 0.67 Gy versus 77.26 ± 0.77 Gy respectively (unpaired t‑test; df = 49, p = 0.005, two-sided; 95% confidence interval [CI] of difference = 0.2–1.03 Gy; Fig. 3).

Fig. 3
figure 3

Box plot showing the mean dose in Gy for ventral (n = 21) and dorsal (n = 30) index lesions with a significant mean dose difference. Unpaired t‑test, df = 49, p = 0.005, two-sided

Full size image

Discussion

MRI offers superior tissue contrast compared to CT which has led to an increasing use of mpMRI in the management of prostate cancer and in radiation oncology [19,20,21,22]. The localization and detection of intraprostatic lesions by MRI allows advanced treatment techniques with focal dose escalation and simultaneous OAR protection which was shown recently by the FLAME trial [14, 23]. As whole-gland radiotherapy is still the current standard of care, we studied the possibility to localize the index lesion with treatment planning MRI and to evaluate the dose coverage of the index lesion as clinically important target for avoidance of cancer progression [6].

There are many studies which show the feasibility and sensitivity of diagnostic mpMRI-based detection of prostate cancer lesions with reference to whole-mount pathology [24,25,26]. In patients with primary radiotherapy treatment whole-mount pathology is not routinely available. Therefore, we correlated the treatment MRI with TRUS-guided biopsy reports as ground truth although TRUS-guided biopsy itself may miss significant proportions of high-grade prostate cancer foci [27] with a sensitivity level of around 50% [28]. On the other hand, mpMRI seems to outperform TRUS-guided biopsy for high-risk prostate cancer detection but may miss lower grade foci [28,29,30]: The PROMIS trial reported significant better sensitivity of mpMRI versus TRUS-guided biopsy (93% vs. 48%, p < 0.0001) for clinically significant cancer which was defined as Gleason score ≥ 4 + 3 or ≥6 mm of any cancer [28]. Furthermore, the recently published results of the PRECISION trial suggest the superiority of MRI-targeted biopsy over TRUS-guided biopsy [31]. Pal et al. confirmed the high sensitivity of mpMRI, but also reported that the sensitivity is lowered to 72% if International Society of Urological Pathology (ISUP) grade 2 cancer is included in the analysis [30]. Pal et al. concluded that reliance on mpMRI could deny patients with ISUP grade ≤ 3 potentially beneficial treatment [30].

In our study MRI and TRUS-biopsy derived location of the index lesion was matched in 68.3% (side-based assessment, n = 52) with an increase in accuracy by inclusion of secondary lesions to 78.8% which could be possibly explained by an overdetection of clinically insignificant cancer by TRUS-guided biopsy [28]. In the subgroup with sextant biopsy-based location data accuracy was only 67.2% with inclusion of secondary lesions, but the patient number was low (n = 33) and correlation and registration of MRI and pathology proves to be challenging, even in whole-mount pathology studies [32] and even by usage of the PI-RADS version 2 sector map for location description [33]. Sensitivity was decreased from 82.1 to 59.0% for the detection of the index lesion inclusive secondary lesions due to the fact that the sextant biopsy data are based on triple the amount of prostate segments.

There was a significant, but small difference in planned mean dose for dorsally located lesions compared to ventral locations with a mean Dmean of 77.26 ± 0.77 Gy versus 77.87 ± 0.67 Gy (p = 0.005). An explanation for this dose difference is the observance of OAR constraints especially for the rectum with the aim of dose reduction for the dorsal half respectively third of the rectum. In the definition of PTVBoost the rectum was routinely excluded which led to a dorsal dose gradient. Nevertheless, mean planned Dmean exceeded the prescribed dose of 76.23 Gy even for dorsal index lesions and each plan was critically reviewed before application for target volume coverage and simultaneous OAR protection. In our study only one out of 54 cases showed a slight underdosage of more than 1% in DIL mean dose, due to the location of the index lesion adjacent to the rectal wall. Thus, in more than 98% of all patients the planned dose coverage of the index lesion was adequate.

There are limitations to this study: We conducted a retrospective review of the histologic biopsy reports, as such no prospective confirming biopsy or whole-mount pathology were available. Implementation of automated segmentation [34,35,36] and deformable image registration [37] could improve accuracy of the MRI-histology comparison and minimize uncertainty arising from rigid image registration and from observer dependent variability in delineation [38, 39]. Our treatment planning MRI did not utilize dynamic contrast-enhanced imaging but a T1-weighted post contrast sequence for radiotherapy treatment planning with emphasis on the T2-weighted and DWI sequences for cancer delineation. A meta-analysis by Chen et al. concludes that dynamic contrast-enhanced imaging plays a supplementary, but still only confirmatory role in prostate cancer detection [40]. Although high diagnostic performance is desirable, we adopted a more practical, time and cost effective approach. Recently, DWI-sequences with higher b‑values have been proposed for lesion detection to increase sensitivity [41], but have not been implemented yet in this study collective. Additionally, we showed the planned radiotherapy dose, but for applied dose intra- [42,43,44] and interfraction [45,46,47] organ movement has to be considered. Adaptive radiotherapy techniques with compensation for movement errors as well as high-quality treatment planning as shown in this study form the basis of precision radiotherapy for prostate cancer [44].

Conclusion

Our data show that contemporary VMAT-based radiotherapy treatment planning allowed adequate dose coverage of the index lesion in more than 98% of all patients. Special attention must be given to index lesions located adjacent to the rectal wall where OAR constraints require careful treatment planning and contemporary image-guidance systems are of paramount importance. Integration of MRI in radiation oncology allows the detection of these lesions and may guide treatment planning concerning dose coverage. In the future it may form the basis for further trials of mpMRI-based focal radiotherapy. A diagnostic accuracy in the range of only 70% however challenges the concept of focal treatment only.