Post-treatment MRI aspects of photodynamic therapy for prostate cancer

Objectives Photodynamic therapy is a new focal therapy for prostate cancer. Methods In this technique, a photosensitising agent is introduced intravenously, then activated by local laser illumination to induce tumour necrosis. Treatment efficacy is assessed by magnetic resonance imaging (MRI). Results and Conclusions We illustrate specific post-treatment MRI aspects at early and late follow-up with pathological correlations. Teaching points • Dynamic phototherapy is a new and promising focal therapy for prostate cancer. • One-week MRI shows increased volume of the treated lobe and large, homogeneous necrosis area. • Six-month MRI shows significant changes of the prostate shape and signal. • Six-month MRI becomes “base line” appearance for further follow-up or monitoring.


Introduction
In 2012, with 417,000 new cases and 92,200 deaths, prostate cancer was the third most common cancer in Europe overall after female breast cancer and colorectal cancer [1]. In men only it was the most common cancer, followed by lung and colorectal cancers. The incidence of prostate cancer, especially for localised disease, has been growing since the 1990s, a dynamic that may be attributable to the widespread use of prostate-specific antigen (PSA) tests [2] and increasing life expectancies.
The management of prostate cancer is still under debate in the urological community. Radical prostatectomy, external beam radiotherapy or brachytherapy are still considered gold-standard treatments, but carry with them well-known side effects (urinary incontinence, erectile dysfunction, intestinal toxicity). Active surveillance and watchful waiting are attractive options for localised tumours [3]. Various focal treatments, such as phototherapy, cryotherapy, interstitial thermotherapy and dynamic high-frequency ultrasound (HIFU), are being evaluated [4,5].
Photodynamic therapy is a recent technology wherein a photosensitising agent (WST11) is administered intravenously, then activated by light to induce tumour necrosis. The light is applied using laser optical fibres positioned within the prostate as determined by magnetic resonance imaging (MRI) planning. The procedure is performed under general anaesthesia, in the dark. Hollow, transparent, plastic brachytherapy catheters were positioned into the prostate, using TRUS-image guidance in accordance with a previously devised MRI-based treatment plan (Fig. 1). Cylindrically diffusing optical fibres were inserted into the catheters. Laser light was delivered to the prostate, using a multichannel diode laser. The single TOOKAD Soluble intravenous (i.v.) dose was a 10-min infusion, followed by continuous illumination of the prostate gland for between 20 and 25 min. The total duration of the whole procedure was~2 h. The light activates the photosensitiser in the prostate, generating reactive oxygen species that activate thrombosis within the vessels. This results in obliteration of the microvessel anatomy with resultant deprivation of oxygen and nutrients to the tumour cells and surrounding prostate tissue in the treated area [6][7][8]. An optimal radius of 6.5 mm is expected around a 753-nm laser fibre [9]. The treatment appears promising for localised prostate cancer and benefits from a reasonably low rate of complications [10]. Currently, MRI is the reference imaging technique Fig. 1 Procedure performed under general anaesthesia, in the darkness (a). Catheters were positioned into the prostate, using TRUS-image guidance in accordance with a previously devised MRI-based treatment plan. Cylindrically diffusing optical fibres were inserted into the catheters (b) Fig. 2 Pre-treatment MRI of a 65-year-old patient with a PSA level of 9.7 ng/ml and a positive biopsy in the right prostate apex (adenocarcinoma Gleason 6). Axial and sagittal T2-weighted images (a, b) show a hypointense nodule in the right peripheral prostate with restriction on the apparent diffusion coefficient map of the diffusion-weighted image (c, low signal intensity) and early enhancement on the axial dynamic contrast-enhanced MR image (d, arrowhead) for the diagnosis of disease and follow-up after local treatment [11][12][13].
However, the changes induced by photodynamic therapy are poorly reported in the literature. Thus in the present pictorial review, we illustrate various MRI features in the follow-up of patients who underwent photodynamic therapy for localised prostate cancer. We studied early and late normal aspects, local complications and aspects of recurrence, with histopathological correlation.

Iconography and imaging protocol
The iconography presented in this paper comes from a group of 77 patients enrolled by our institution in a prospective international multicentric study (phase II/III), and who received photodynamic therapy with WST11 (Tookad soluble) for localised prostate cancer.
Inclusion criteria are: men over 18 years old; diagnosed prostate cancer histologically proved on biopsies (systematic biopsies most of the time, rarely targeted biopsies); eligible for active surveillance; no prior treatment for prostate cancer; TNM stage up to cT2b-N0/Nx-M0/Mx; Gleason score≤3+ 3, 3+4 accepted in certain conditions; PSA<10 ng/ml. The prostate cancer diagnosis was histological. All patients had a pre-treatment MRI then post-treatment MRIs at 1 week (early post-treatment) and 6 months (late post-treatment). Some of the patients also had a post-treatment MRI at 3 months. Systematic biopsies with TRUS (trans-rectal ultrasound)  guidance were performed at 6 months to explore correlations to MRI features. In case of positive biopsy at 6 months, function of the Gleason score, different options were proposed to the patients: treatment of the other lobe, retreatment on the same lobe, active surveillance or surgery.

MRI protocol
All MRIs were performed with a 1.5-T MRI system (GE Excite, General Electrics, Milwaukee, IL). An external body surface coil (eight channels) was used, with a field of view (FOV) equal to 24 cm, except for the dynamic contrastenhanced sequences for which FOV was 42 cm.
The imaging protocol included the following sequences: axial and coronal FRFSE T2 with fat saturation (4-mm slice thickness), axial FSE T1 (4-mm slice thickness), axial diffusion-weighted (b =600 s/mm 2 ); a three-dimensional (3D) dynamic contrast-enhanced T1 sequence (LAVA) with fat saturation after intravenous injection of 20 ml gadoterate meglumine (Dotarem, Guerbet, France), followed by axial and sagittal fat saturation contrast-enhanced FSE T1-     Pre-and post-treatment MRI exams were analysed on an imaging work station (Fujifilm Synapse 3D). We analysed MRI modifications for the areas treated with photodynamic therapy (optical fibre disposition map) and the histopathological results of systematic 6-month biopsies. Some patients underwent prostatectomy after the phototherapy treatment, permitting a more accurate radiological-pathological correlation.
PIRADS classification was not used for treatment planning. Pre-treatment MRI served only to plan the anatomic distribution of the optic fibres into the lobe to be treated, targeting the zones of positive biopsy. The PIRADS criteria were used for the follow-up MRI analyses.

Pre-treatment MRI
Pre-treatment MRI was performed as a planning procedure to determine the position and the number of optical fibres and to exclude any loco-regional extension that would have contraindicated the treatment.
Only small tumours with low aggressiveness (Gleason 6) were considered for photodynamic therapy and thus not all of them were visible. When tumours were visible, the classical aspect was a hypointense nodule in T2-weighted MR images, with restriction on the apparent diffusion coefficient map and early enhancement on dynamic contrast-enhanced MR images (Fig. 2).
Early post-treatment MRI features One-week MRI examinations show T2 heterogeneous signals in the treated lobes (Fig. 3) that are related to the oedema and ischaemic modifications induced by phototherapy. These modifications are secondary to local intravascular coagulation triggered by the photosensitising agent when exposed to laser light.
MRI sequences show an increased volume of the treated lobe. When the procedure is unilateral, the volume variation is obvious and particularly well visualised in T2-weighted Among all heterogeneities within the treated lobe, besides the loss of physiological distinction between transitional and peripheral prostate tissue in T2 images, the paths of the optical fibres can be observed as small hyperintense spots in the axial plane and linear bands in the sagittal and coronal planes (Fig. 5).
On T1-weighted images before gadolinium injection, hyperintense, poorly defined areas can be identified. These areas are related to haemorrhagic suffusions (Fig. 6) and make T1weighted contrast-enhanced images difficult to analyse; in this situation the use of image subtraction may be helpful.
Necrosis is best defined on T1-weighted FAT-SAT contrast-enhanced images. It appears as non-enhancing areas [14] with irregular but well-defined and highly enhanced edges (Fig. 7). In most cases, a treated lobe presents a single, large, homogeneous necrosis area on T1 contrast-enhanced sequences. However, in some cases, islands of spared tissue may be observed with high early enhancement (Fig. 8). This aspect should not be wrongly analysed as residual tumour [14].  Another pitfall to be avoided is focal inflammatory enhancements in the treated lobe presenting a nodular feature (Fig. 9). These lesions are usually no longer present on 3month post-treatment MRIs.
In cases with concomitant bilateral treatment, the 1-week MRI shows the same types of modifications but they are symmetrical. A particularity of bilateral treatment is the higher risk of urethral wall necrosis. Normally, the intraprostatic urethra exhibits a continuous ring of peripheral enhancement [14]. A discontinuous ring (Fig. 10) on the 1-week postphotodynamic therapy MRI is thus suggestive of urethral wall necrosis. In this situation, despite earlier animal studies reporting structural and functional resistance of the urethra to photodynamic therapy [15], close clinical monitoring is essential.
Finally, it is important to note that on the 1-week posttherapy images the surroundings of the gland may also show enhancement due to local inflammation.  At 6 months after photodynamic therapy, important changes of the prostate shape and signal are found. Small areas of residual necrosis may still be present in the treated lobe, corresponding to coagulation necrosis (Fig. 11).
Classically, the volume of the treated lobe decreases, sometimes to less than the volume of the lobe before treatment. This is due to the apparition of retractile fibrous scar tissue (Fig. 12).
Residual, irregularly shaped fluid cavities related to atrophy in the treated lobe are frequently seen (Fig. 13).   Unilateral treatment induces an asymmetric prostate, whereas bilateral treatment results in a small, heterogeneous, but symmetrical prostate (Fig. 14). Fig. 21 Levator ani muscle. Right photodynamic therapy in a 65-year-old patient for Gleason 6 adenocarcinoma (two positive biopsies in right prostate). Day-7 post-treatment axial (a) and coronal (b) T2-weighted MR images visualise a heterogeneous high signal of the right puborectalis muscle (arrowheads); axial and sagittal T1 FS contrast-enhanced images (c, d) show a low-signal welldefined area surrounded by a hyperintense enhanced border after gadolinium injection (arrows) corresponding to extraprostatic necrosis affecting the right levator ani muscle Fig. 22 Levator ani muscle. Day-7 post-treatment axial T1 FS contrastenhanced (a) and axial T2-weighted (b) MR images after right photodynamic therapy show extension of necrosis to puborectalis muscle (star); at the 6-month post-treatment MRI follow-up, the axial T2 image (c) illustrated the complete disappearance of the area of necrosis (arrowhead). No anal functional abnormalities were seen in this patient Furthermore, the aspect observed at 6 months is definitive (Figs. 15 and 16). We thus underline that the 6-month MRI is very important as it will become the new "base line" appearance for any further follow-up or monitoring.
Three-month MRIs are not pertinent for treatment followup as they tend to show aspects that are between early necrosis and late fibrosis and thus difficult to analyse (Fig. 17).

Post-retreatment features
The indication criteria of retreatment are positive biopsies in treated or contralateral lobe during the follow-up, with Gleason score ≤3+3 or 3+4 in certain conditions. When ipsilateral retreatment is needed, the modifications seen after 1 week are similar to those seen 1 week after the initial treatment (Fig. 18). After 6 months, the atrophy in the retreated lobe is more marked than after the first treatment.
When a renewed intervention is for the contralateral lobe (Fig. 19) the asymmetric aspect is more pronounced at 1 week because of the atrophy of the previously treated lobe, but at 6 months, this asymmetry disappears and the prostate takes on the same appearance as a concomitant bilateral treatment.
Complications Photodynamic therapy appears to have a reasonably low rate of complications [10]. In a study on different types of focal therapies performed in a cohort of men with low-risk prostate cancer, the overall complication rate was 13 %, with only two Clavien-Dindo grade 3 complications. (Clavien Dindo is a to the muscularis propria. Note the complete recovery of the rectal wall at 6 months. Also no fistulae were clinically identified grading system from 1 to 5, used for the classification of surgical complications; grade 3 or more are serious complications: grade 3-complication which requires surgical, endoscopic or radiological intervention-up to grade 5, which corresponds to the death of the patient). In that series, there were no complications described for the patients who received photodynamic therapy [16]. Azzouzi et al. [8] found that most treatment-emergent adverse events (TEAEs) post photodynamic therapy were mild or moderate and only 9 % of patients reported serious TEAEs ( Table 1).
The main post-photodynamic therapy complication is extraprostatic necrosis, defined as a lack of enhancement on T1-weighted contrast-enhanced MR images [14] involving the different structures surrounding the prostate.
When present, necrosis most frequently extends to periprostatic fatty tissue (Fig. 20). The muscles close to the prostate can also be affected, especially the levator ani (Fig. 21). However, in our experience with this situation, anal functional abnormalities are not observed and a normal aspect is recovered after 6 months (Fig. 22).
Extraprostatic necrosis may also affect the internal obturator muscle (Fig. 23) and rarely it may extend to the anterior rectal wall (Fig. 24). In our experience, fistulisation does not seem to occur and the rectal wall recovers its normal aspect on the 6-month follow-up MRI.
Sliding of the optical fibre may result in an extension to a seminal vesicle (Fig. 25).    As mentioned earlier, necrosis may affect the intraprostatic urethra; a break of the periurethral enhancement ring or its complete disappearance at 1 week is an interesting sign suggestive of urethral parietal necrosis (Fig. 26).
Recurrence in the treated area MRI detection of tumour recurrence in the treated area is made difficult by the loss of the normal hyperintensity of the peripheral prostate in T2-weighted images and the signal of the scar.
Thus, after photodynamic therapy, the diagnosis of recurrence is mainly based on PSA levels and systematic biopsy.
There is no reliable feature of recurrence and cases of small unaggressive recurrence are usually not visible on MRI (Fig. 27). Early dynamic contrast-enhanced images and apparent diffusion coefficient mapping may be useful for detecting recurrence (Fig. 28) but they lack reliability. Therefore, any suspicious nodule seen on follow-up MRI should be biopsied (Fig. 29). Targeted biopsies with TRUS are performed in nodules presenting PIRADS criteria on follow-up MRI.

Conclusions
Photodynamic therapy is a new and promising focal treatment for localised prostate cancer. Early and late post-treatment MRI aspects are specific. Necrosis is obtained at 1 week and the final post treatment aspect at 6 months. Knowledge of these features permits the appreciation of treatment efficacy and the accurate diagnosis of complications or recurrence.
Open Access This article is distributed under the terms of the Creative Commons Attribution License which permits any use, distribution, and reproduction in any medium, provided the original author(s) and the source are credited. Fig. 29 Mismatch. PSA increase in a patient treated by left photodynamic therapy 1 year earlier. MRI was performed: axial T2 (a), axial ADC (b) and axial T1 dynamic contrast-enhanced (c) MR images show a suspicious left median focus with T2 hyposignal, early enhancement and mild diffusion restriction on ADC map (arrowhead). The location of suspicious lesion seen on MRI is indicated on a prostate map. Targeted biopsy was performed with TRUS. Pathology slide at high magnification (d) shows a 1mm tumour focus in a scar tissue. Crushed appearance of several glands (circled), suspect, confirmed by immunohistochemical study, and surrounded by thick collagen fibres (fibrous scar). The nodule on MRI corresponds to the scar tissue; the tumour focus is too small to be seen on MRI