Trends in targeted prostate brachytherapy: from multiparametric MRI to nanomolecular radiosensitizers

Diagnosis using MRI

Traditionally, definitive diagnosis of localized prostate cancer relied on the TRUS-guided sextant biopsy for identification of cancerous tissue; this approach alone was found to leave up to 15 % of cancers undetected and was prone to sampling error (Norberg et al. 1997). The information from TRUS biopsy is now routinely correlated with imaging information provided by multiparametric MRI (mpMRI); this offers a clearer representation of disease foci and extraprostatic disease extension as well as improved biopsy sampling and characterization as part of staging (Lawrentscuk and Fleshner 2009; Kaplan et al. 2002; Cirillo et al. 2008; Pullini et al. 2016). In 2012, the prostate imaging–reporting and data system (PI-RADS v.1, updated in 2015/2016) was developed to standardize the approach to diagnosis and reporting of mpMRI for prostate cancer (Weinreb et al. 2016); several recommendations were made. Diagnosis of prostate cancer should be performed using T2-weighted fast spin echo (T2W-FSE) sequences on a 1.5–3T MRI scanner, the latter being preferred due to its higher signal-to-noise ratio, along with inclusion of functional DCEI, DWI, and potentially MRSI sequences. A high degree of visibility of the prostate peripheral zones, along with urethral and rectal structures, seminal vesicles, and prostatic capsule (extra- and intracapsular disease) is typically required. Each imaging sequence of the mpMRI acquisition contributes supplementary information that facilitates diagnosis and/or staging of localized disease; the PI-RADS reporting system further standardizes this approach according to the respective imaging sequence.

T2W-FSE images (see Table 1 for recommended PI-RADS parameters) are typically used to discriminate between zonal anatomies of the prostate and discern normal tissue from various abnormalities; however, a number of benign conditions may mimic the T2W appearance of prostate cancer including benign prostatic hyperplasia (BPH) or prostatitis (Kitajima et al. 2010). Intraprostatic cancerous tissue conversely may also be difficult to detect on this imaging alone; Fig. 1 shows a typical example of a multiplanar T2W-FSE image of the prostate and surrounding anatomy. The limitations of purely anatomic imaging necessitate the supplementary use of functional imaging sequences.

Imaging sequence	Slice (mm)	FOV (cm)	In-plane dimensions (mm)	TR/TE (msec)	Additional	Contrast
T2W	3	12–20	≤0.7 × ≤0.4	Not specified	–	–
DWI	≤4	16–22	≤2.4 × 2.5	≤90/ ≥3000	b min = 50–100 s/mm2 b max = 800–1000 s/mm2	–
DCEI	3	Entire prostate + SV	≤2 × 2	<100/ <5	Observation time ≥2 min	0.1 mmol/kg GBCA; injection rate = 2–3 cc/sec

Since MRSI is still investigational, PI-RADS does not specify MRSI as standard for mpMRI sequences in prostate cancer diagnosis

DWI, a functional method of measuring random water molecule diffusion rates within tissue, is typically used to supplement T2W-FSE acquisitions. Prostate cancers present with restricted diffusion rates relative to the surrounding normal prostatic tissue, a phenomenon that is represented through apparent diffusion coefficient (ADC) maps computed at each image voxel (Weinreb et al. 2016; Kim et al. 2010). B values—an indicator of the rate of diffusion that is captured by the image dataset—have recommended ranges between 0 and 1000 s/mm²; larger b values are used to detect slower diffusion rates of water molecules (Kim et al. 2010). The inclusion of DWI sequences with T2W imaging improves detection of prostate cancer, particularly in the peripheral zone tumors (Haider et al. 2007). Figure 2 shows a typical DWI axial sequence of the prostate.

DCEI obtained by acquiring T1W image sequences during administration of an intravenous gadolinium-based contrast agent (GBCA), is used to further supplement the information obtained from both T2W and DWI sequences. DCEI takes advantage of cancer angiogenesis, a process that increases vascular density and permeability within tumors, to visualize prostatic regions of increased uptake of the GBCA (Verma et al. 2012; Singanamalli et al. 2016). Figure 3 shows the differential uptake of GBCA within areas of active tumor (the DIL), over a set of sequential axial frames. The addition of DCEI to the mpMRI sequence has demonstrated improvements in sensitivity and lesion detection accuracy (Kitajima et al. 2010; Alonzo et al. 2016).

Magnetic resonance spectroscopic imaging (MRSI), while not as widely adopted as DWI or DCEI, is gaining interest as a valuable technique for non-invasively determining the biochemical concentrations of biomarkers associated with prostate cancer (Kurhanewicz et al. 2002; Picket et al. 2004; Pouliot et al. 2004). MRSI may be more specific in differentiating benign conditions from actively metabolic prostate cancer and may provide metabolic information associated with tumor progression (Kobus et al. 2014). In vivo prostate MRSI utilizes the relative or absolute concentrations of the metabolites such as choline, polyamines, creatine and citrate, as cancer specific biomarkers, with 98 % of cancers demonstrating an elevated (choline + creatine)/citrate ratio greater than the ratio for normal tissue (Fig. 4) (Westphalen et al. 2008; Scheenen et al. 2015). Addition of the functional information provided by MRSI could potentially improve diagnosis, staging, and disease localization for BT. The diagnostic value of combined magnetic resonance imaging and spectroscopic techniques has encouraged radiologists and oncologists to include it increasingly for clinical use.

Pre- and intra-operative image guidance and treatment planning

Once diagnosis and staging of prostate cancer have been completed (and BT is selected as a treatment option), the identification of critical target and normal tissue structures is made. During this phase, the target, typically the entire prostate gland and/or focal lesions, and organs at risk (OARs) are identified for treatment planning (Thomadsen et al. 2012; Yu et al. 1999). Historically, CT, TRUS, and more recently mpMRI, have been used to identify critical structures (Nath et al. 2009; Metcalfe et al. 2013; Rischke et al. 2013). Clinical use of CT or TRUS has traditionally been the mainstay of pre-treatment imaging for target identification, but recent improvements in the availability of diagnostic MRI scanners have improved access significantly to mpMRI for this purpose (Davis et al. 2012; Yamada et al. 2012; Nag et al. 2000; Mayer et al. 2016). Several studies have pointed to the subjectivity of contouring on conventional imaging modalities. Better imaging methods are being heralded as a means to provide objectivity to this aspect of the planning process (Rischke et al. 2013; Steenbergen et al. 2015; Fiorino et al. 1998).

Pre-operative guidance prior to MRI

CT-guided BT planning represents one of the earliest attempts to use volumetric imaging to accurately identify targets and OARs, as well as plan radioactive source positions, with good outcomes; Koutrouvelis et al. (2000) reported prostate-specific antigen (PSA) <2 ng/mL in 90 % of patients (n = 301) at median 26 month follow-up after being treated with permanent implant CT-guided BT (Koutrouvelis et al. 2000). Intra-operative TRUS-guided BT is rapidly growing in adoption largely due to its lower cost, widespread availability, and real-time guidance, and has allowed implant guidance during the BT procedure in addition to target localization. Stone et al. (2007), pioneers of the TRUS-guided permanent implant approach, reported excellent long-term toxicity outcomes of 325 patients (Stone and Stock 2007; Crook et al. 2011). TRUS-guided BT is now the recommended standard of care for prostate BT by both ABS and GEC-ESTRO guidelines for both LDR and HDR implants (Davis et al. 2012; Yamada et al. 2012; Hoskin et al. 2013). Despite these many advantages, significant TRUS artifact still make identification of the target and OARs highly subjective (Fig. 5) (Xue et al. 2006; Solhjem 2004). Further identification of DILs for dose escalation is simply not feasible using TRUS guidance alone, and a workflow incorporating mpMRI fusion with TRUS imaging is commonly required.

Intra-operative guidance prior to MRI

Current GEC-ESTRO and ABS guidelines for prostate brachytherapy recommend intra-operative TRUS imaging for visualization of the prostatic capsule, nearby anatomy, and implant guidance. Interpretation of TRUS images is highly subjective, leading to difficulties in target, OAR, and source identification during implantation. Figure 5 outlines some of the difficulties in identification of catheter trajectories for an HDR prostate BT procedure. For HDR BT, once catheter identification has been completed, an optimization algorithm is used to determine the optimal dwell times of the radioactive source within each catheter with the goal of maximizing the radiation dose to the target(s) while minimizing the dose to normal tissues (Fig. 6). A similar process in LDR BT is used to determine the configuration of implanted sources within the prostate gland.

Post-implant quality assurance

Typically, post-implant dosimetry is performed 1 month after an LDR BT implant using CT guidance (Fig. 7); the position of the implanted sources are identified, along with normal tissues and target volumes, and the isodose distribution is recomputed to evaluate the quality of the delivered treatment (Potters et al. 2001). Post-operative dosimetry metrics have been poorly correlated with intra-operative dosimetry; a study by Acher et al. (2010) showed that this was largely due to the subjective nature of the contoured prostate volume on CT (Merrick et al. 1999; Acher et al. 2010). MRI-based post-implant dosimetry has been proposed to offer improved discrimination between the prostate and OARs; however, signal voids around the metallic casings of brachytherapy sources pose a challenge for accurate source localization. Thomas et al. (2009) demonstrated that signal voids surrounding brachytherapy sources exist for increasing field strengths (3–4.5 mm separation for 1.5 T, and 4.5–6 mm for 3 T), which affect the accuracy with which the sources are localized; a proton density-weighted FSE sequence was used to limit this effect with good results (Thomas et al. 2009). Various studies demonstrated techniques to minimize these artifacts; Kuo et al. (2010) examined an in-phantom method using inversion recovery with ON-resonant suppression (IRON) to generate positive contrast in areas of high magnetic susceptibility artifact (Kuo et al. 2010). Similarly, positive contrast agents such as cobalt dichloride-N-acetyl-cysteine (C4) have been used as encapsulated markers and show promise at a number of different parameters and field strengths, without altering the dosimetry of the delivered therapy, and showing minimal patient toxicities (Lim et al. 2014; Frank et al. 2008).

Focal and salvage therapy using MRI guidance

Traditionally, the use of CT and TRUS guidance for prostate BT planning was limited to treating the entire gland, largely due to the inability to resolve the multifocal nature of intraprostatic disease with confidence. The significant improvement in the ability to differentiate individual focal lesions using mpMRI enables targeted dose escalation, while decreasing the whole-gland dose, thereby significantly reducing the dose to the OARs (Muller et al. 2014; Crehange et al. 2014).

Studies evaluating focal boosting to both single and multifocal DILs using both LDR and HDR have emerged recently. Ahmed et al. (2012) showed excellent 12 month outcomes and urinary toxicity following LDR BT focal boost (Ahmed et al. 2012). Banerjee et al. (2015) and Gomez-Iturriaga et al. (2016) demonstrated similar results with HDR BT showing significant escalation of DIL doses without an expected increase in urinary toxicities (Gomez-Iturriaga et al. 2016; Banerjee et al. 2015). This approach has also shown promise in focal salvage therapy, using both LDR and HDR, in reduction of urinary toxicities while maintaining excellent outcomes comparable to whole-gland salvage treatment (Hsu et al. 2013; Duijzentkunst et al. 2016).

These promising hypothesis-generating studies require robust long-term control and toxicity data, but they do indicate that focal boosting (and focal salvage therapy) using mpMRI-guided HDR and LDR is a feasible method for dose escalation while minimizing OAR toxicities.

There exist ample opportunities for multimodality fusion of mpMRI as well as intra-operative MRI for focal dose escalation. Significant technological hurdles to the implementation of intra-operative MRI-guided BT make it more likely that multimodality fusion will be readily adopted as a means to deliver focal or whole-gland BT. For centers that have the ability to implement intra-operative guidance for BT, it is likely that closed-bore, high-strength MRIs will see increased uptake due to their wider availability and ability to deliver high-quality anatomical and functional imaging.

Radiosensitization using gold nanoparticles

Another promising new technique for improving local dose escalation, potentially acting synergistically with the improved image guidance afforded by mpMRI, is the use of gold nanoparticles (GNP) for selective radiosensitization of tumors (Jain et al. 2012; Babaei and Ganjalikhani 2014). The main mechanism of action of these nanoparticles is the production of photo- and Auger electrons by the photoelectric effect after bombardment with photons (Spiers 1949; Castillo et al. 1988). The short range of these electrons requires that cytosolic uptake of GNPs occurs to create DNA single- and double-strand breaks (Zheng et al. 2008). The selective dose enhancement factor (DEF)—the increased effect of local radiation dose deposition due to the GNPs—has been observed with high GNP concentrations in conjunction with keV photon energies as outlined by Zhang et al. 2008, Rahman et al. (2009) and Roeske et al. (2007). The introduction of GNP-mediated radiosensitization both for LDR and HDR brachytherapy, which emit gamma photon energies in the ideal range for maximizing DEF, holds significant promise.

Due to the long half-life of most LDR brachytherapy sources, GNPs introduced into cancerous cells must remain within the cell cytosol for extended periods of time to sufficiently provide a dose-enhancement effect. Shorter periods of GNP uptake by prostate cancer cells may be suitable for HDR delivery, due to the rapid radiation delivery (in the order of 10–30 min), and it therefore may serve as an easier pathway to GNP integration within the BT framework. Further development and customization of GNPs to specifically tailor them for use in prostate BT should focus on the route of administration, pharmacokinetics, and cellular uptake.

The geometric and functional parameters of GNPs play an important role in their uptake as outlined in Albanese et al. (2012), Perrault et al. (2009) and Favi et al. (2015). Naïve GNPs without additional ligands preferentially accumulate at sites of porous and leaky tumor vasculature (Jain et al. 2012). The addition of ligands such as polyethylene glycol (PEG) allow improved non-specific uptake by receptor-mediated endocytosis (RME) within tumors, as well as improved transit times in systemic circulation (Zhang et al. 2008; Kumar et al. 2013; Lechtman et al. 2013; Chithrani et al. 2006). Further functionalization by addition of tumor-specific ligands may enable GNPs to further target prostate cancer with increased specificity. Of importance in uptake and clearance is GNP size: smaller GNPs tend to have rapid circulations when administered intravenously and are quickly cleared by the renal system but more rapidly permeate tumor vasculature, while excessively large particles may have significantly increased uptake by the Reticuloendothelial System (RES), a process which decreases selective DEF (Arnida and Ghandehari 2010; Maeda et al. 2001).

In contrast to systemic administration, the intra-operative nature of BT procedures may enable interstitial injection of GNPs directly within focal lesions outlined on mpMRI. In this proposed workflow, the pre-treatment mpMRI-TRUS—or intra-operative MRI—could be used to localize focal lesions and the high spatial resolution of MRI could be used to guide deposition of high concentrations of GNPs directly within these cancerous foci during the BT procedure. This approach may significantly reduce the need for larger GNP sizes with their increased residence time, thereby taking advantage of the benefits of small GNP sizes. Evidence suggests that a medium range (6–50 nm) PEGylated, spherical GNP, administered interstitially, could allow sufficient tumor uptake and retention over the course of LDR BT to offer a significant dose-enhancement effect (Chithrani et al. 2006). For HDR BT, a smaller size (<6 nm) PEGylated, spherical GNP may be more suitable, allowing for a residence time and dispersion that correlated with the procedure duration, thereby producing the required dose enhancement with fewer potential side effects. GNPs could be administered during the procedure and rapidly cleared from circulation afterward. Commercially available GNPs specifically optimized for radiation therapy are also being developed to have longer circulation, small sizes (2–3 nm), higher tumor uptake, and improved clearance (Kumar et al. 2013). It remains to be seen if the pharmacokinetics of these commercially available GNPs is similar when administered locally within the prostate gland.

Conjugation of gadolinium with GNPs (Gd-GNP) could also allow for visualization on intra-operative MRI and offer a means of calculating the biological effective dose from the additive effects of GNPs during HDR BT (Harisinghani et al. 2003; Debouttiere et al. 2006; Le Duc et al. 2014). Vartholomeos et al. 2011 also examined the use of MRI-compatible nanoparticles to act as drug-delivery nanorobots under MRI-guided steering; similar methodologies may be applied to both track and steer injected GNPs to tumor vasculature within the prostate gland during BT (Vartholomeos et al. 2011).

There remain significant challenges with local deposition of high GNP concentrations within the prostate gland largely relating to their diffusion and uptake within the tumor vasculature. The assumption of homogenous GNP distribution with local administration may not be accurate, especially with local administration; problems with inconsistent tumor vasculature (particularly in hypoxic tumor regions) may cause poor uptake or heterogeneous GNP distribution even within small focal lesions. It is prudent to assume that focal lesions neighboring normal tissues, such as the urethra, may be at risk of excessive local hot spots with uneven GNP distribution and therefore further exploration is needed. Additionally, although systemic toxicities of GNP administration have been noted as minimal in some studies (Alkilany and Murphy 2010; Fratoddi et al. 2014), the examination of toxicities from localized injection of high GNP concentrations within the prostate requires further study if they are to be used in BT.