Integration between Novel Imaging Technologies and Modern Radiotherapy Techniques: How the Eye Drove the Chisel

Simple Summary This paper aims at showing the impact of novel imaging technologies and modern radiotherapy techniques on the management of cancer, with a particular focus on prostate adenocarcinoma. The manuscript explores the value of diagnostic imaging before treatment, the role of radiomics in predicting outcomes, the benefit of novel imaging in radiotherapy planning and the influence of advanced technologies in systemic treatment and in the management of other non-oncological conditions in order to tailor the best therapeutical strategies. Abstract Introduction: Targeted dose-escalation and reduction of dose to adjacent organs at risk have been the main goal of radiotherapy in the last decade. Prostate cancer benefited the most from this process. In recent years, the development of Intensity Modulated Radiation Therapy (IMRT) and Stereotactic Body Radiotherapy (SBRT) radically changed clinical practice, also thanks to the availability of modern imaging techniques. The aim of this paper is to explore the relationship between diagnostic imaging and prostate cancer radiotherapy techniques. Materials and Methods: Aiming to provide an overview of the integration between modern imaging and radiotherapy techniques, we performed a non-systematic search of papers exploring the predictive value of imaging before treatment, the role of radiomics in predicting treatment outcomes, implementation of novel imaging in RT planning and influence of imaging integration on use of RT in current clinical practice. Three independent authors (GF, IM and ID) performed an independent review focusing on these issues. Key references were derived from a PubMed query. Hand searching and clinicaltrials.gov were also used, and grey literature was searched for further papers of interest. The final choice of papers included was discussed between all co-authors. Results: This paper contains a narrative report and a critical discussion of the role of new modern techniques in predicting outcomes before treatment, in radiotherapy planning and in the integration with systemic therapy in the management of prostate cancer. Also, the role of radiomics in a tailored treatment approach is explored. Conclusions: Integration between diagnostic imaging and radiotherapy is of great importance for the modern treatment of prostate cancer. Future clinical trials should be aimed at exploring the real clinical benefit of complex workflows in clinical practice.


Background
Improvement of clinical outcomes through targeted dose-escalation and reduction of dose to adjacent organs at risk has been the main goal of radiotherapy in the last decade. Due to its characteristic radiobiology, prostate cancer benefited the most from this process, with a significant improvement in the benefit-to-risk ratio. Implementation of three-dimensional conformal planning  was the first step of this evolution, followed by the introduction of Cone-Beam Computed Tomography (CBCT) for set-up evaluation. In recent years, the development of Intensity Modulated Radiation Therapy (IMRT) and Stereotactic Body Radiotherapy (SBRT) techniques radically changed clinical practice and allowed us to re-consider the role of radiotherapy in different clinical scenarios. However, all these upgrades were made possible by the availability of modern imaging techniques, both for their diagnostic value and for the possibility of integrating their information with standard radiotherapy planning. In this paper, our purpose was to overview the close relationship between diagnostic imaging and prostate cancer radiotherapy techniques development.

Diagnostic Imaging and Its Predictive Value before Treatment: How Modern Imaging May Avoid Unnecessary Invasive Diagnostic Procedures (Tables 1 and 2 Below)
Similarly to other tumours [1][2][3][4], multiparametric magnetic resonance imaging (mpMRI) has been proposed as a widely adopted standard for diagnosis and baseline assessment of prostate cancer [5]. For example, Prostate Imaging Reporting and Data System (PI-RADS) represents one of the most important efforts performed in recent years to standardize mpMRI reporting [6]. However, despite the improvements in terms of implementation of this imaging method in clinical practice, it is well known that some issues still exist, like the management of Prostate Imaging Reporting and Data System (PI-RADS) score 3 lesions or the inter-reader variability among clinicians with different levels of expertise [7]. Moreover, various essays have been tested to improve the accuracy of mpMRI in this setting [8]. Indeed, many clinicians feel that the PI-RADS v2 scoring system can be inadequate in distinguishing clinically significant and insignificant groups in central gland tumours [9]. Moreover, it is important that radiologists are familiar with the common incidental findings associated with MRI to minimize the anxiety of the patient and to reduce costs associated with unnecessary further testing [10] since it is currently acknowledged that the widespread request for mpMRI significantly increased the identification of findings unrelated to the primary aim of the investigation [11]. Nonetheless, mpMRI confirmed its clinical usefulness by predicting prostate cancer Gleason Grade before biopsy [12,13] and by detecting local recurrence after radical prostatectomy [14,15]. All these features are of utmost clinical importance in planning either definitive or postoperative prostate radiotherapy.

Radiomics: Role of Imaging in Predicting Treatment Outcomes: How Modern Imaging May Help to Predict Outcomes of a Determined Treatment (Tables 3 and 4 Below)
Integration between modern imaging and radiomics may also be helpful for diagnosis [16][17][18][19][20] and for predicting radiation toxicities before treatment [21]. Of course, treatment toxicity is one of the most important factors to consider to evaluate the benefit-to-risk ratio and to decide treatment strategy in modern radiotherapy [22], but radiological features may also be helpful in predicting clinical outcomes after definitive approaches for different neoplastic diseases [23][24][25][26][27][28][29][30][31]. Moreover, radiomics models were used to predict tumour histopathological features in different series [32][33][34][35][36][37][38]. Abdollahi et al. described a machine learning approach to explore rectal, bladder and femoral head toxicity, aimed at identifying predictive factors on baseline MRI images [39][40][41]. Radiomics features were also evaluated in a secondary analysis of the HYPRO trial, with a significant improvement in rectal bleeding predictability if compared to Dose Volume Histogram alone [42]. Early variations of the rectal wall were observed in patients undergoing RT on an MRI linear accelerator (MRI Linac) in the second week of treatment [43]. Radiomics features could become a valuable tool in this scenario, and machine learning may further refine treatment tailoring [44].

Implementation of Novel Imaging in Radiotherapy Planning: How Modern Imaging May Improve Planning Radiotherapy Techniques (Tables 5 and 6 Below)
Novel metabolic imaging improved staging sensibility in prostate cancer and has been shown to have a significant impact on radiotherapy management [45,46]. However, in the modern era, imaging influences radiotherapy planning in a deeper meaning. Since the introduction of three-dimensional imaging and planning, technical innovation has allowed us to increase the radiotherapy dose delivered to the target and to spare at the same time adjacent organs at risk. This trend was practice-changing, especially for prostate cancer, where a low alpha/beta ratio favours hypofractionated regimens [47] characterized by higher doses for a single fraction. Implementation of hypofractionated regimens in clinical practice is based on Image Guided Radiotherapy (IGRT), which has been used for the treatment of many different neoplastic diseases [48] and currently establishes a modern treatment standard, with favourable comparison to conventional treatment [49,50]. Lately, many reports about MRI-guided boosts to dominant intraprostatic lesions have been reported [51]. Moreover, modern treatment planning allowed us to push forward the ability to assess and report low radiation doses to organs at risk, thus helping to refine awareness about incidental irradiation of healthy tissue during treatment [52] and to explore the impact of dose to structures (e.g., bladder neck) on toxicity and quality of life [53]. Diagnostic imaging can be co-registered with planning imaging to drive modern high-dose intensity-modulated radiotherapy [54][55][56][57]. Kuisma et al. reported data about 30 men treated with radiotherapy and undergoing focal boost to Carbon-acetate PET/CT metabolically active areas. The authors concluded that this metabolic guidance for doseescalated radiotherapy was feasible and deserved further study [58]. Moreover, Prostate Specific Membrane Antigen (PSMA) PET/CT-guided local ablative radiotherapy was tested in a prospective phase 2 trial enrolling oligometastatic prostate cancer patients treated on all PSMA-positive metastases without any systemic treatment. Briefly, this approach showed to be safe and effective in a selected population [59]. The final evolution of integration between imaging and radiotherapy is expressed in the use of online MRI to perform modern IGRT and adaptive RT with MRI Linac [60].   PET, positron emission tomography; CT, computed tomography; PSMA, prostate-specific membrane antigen; IGRT, image-guided radiation therapy; HT, helicoidal tomotherapy; EBRT, external beam radiation therapy; bDFS, biochemical disease-free survival; VMAT, volumetric modulated arc therapy; HIFU, high-intensity focused ultrasound; IMRT, intensity-modulated radiation therapy, WPRT, whole-pelvic radiation therapy; SRT, stereotactic radiation therapy; SSRT, stereotactic synchroton radiotherapy; SBRT, stereotactic body radiation therapy.

Influence on Systemic Treatment and Different Medical Conditions: How Modern Imaging May Change Use of Radiotherapy Influencing Overall Treatment Management of Patients (Table 7 Below)
Local ablative treatment has been shown to be a practice-changing approach in oligometastatic prostate cancer [61], and the integration of imaging techniques and modern radiotherapy allowed us to increase the potential areas of exploitation of radiotherapy as a non-invasive, short and profitable technique. For example, stereotactic body radiotherapy (SBRT) is currently used in routine clinical practice as an ablative approach, able to control loco-regional sites of oligo-progression, thus deferring change of systemic treatment [62,63] or even in non-neoplastic diseases, thus opening emerging clinical scenarios [64]. Local ablative treatment is important in prostate cancer management due to the observation that metastasis-to-metastasis spread is a common phenomenon for this disease [65]. This innovative approach prompted radical changes in terms of clinical management of prostate cancer for clinical oncologists, especially considering the rising necessity for developing integrated treatment strategies with Androgen Deprivation Therapy and new hormonal agents (Abiraterone, Enzalutamide and Apalutamide) [66]. Moreover, the COVID-19 pan-demic prompted us to adapt patients' management [67][68][69][70][71], and the availability of shorter techniques to enable us to deliver safe and effective radiotherapy treatment was of utmost importance during the COVID-19 pandemic [72][73][74].

Discussion
Different types of diagnostic imaging are currently essential for prostate cancer diagnosis and treatment. Before radiotherapy delivery, mpMRI is widely used for diagnosis, and future development will allow the improvement of its diagnostic accuracy and to predict the biological aggressiveness of diseases even before a biopsy. This would allow an early and non-invasive assessment of disease in clinical practice. Moreover, early reports show that treatment toxicity after radiotherapy may be predicted by diagnostic imaging through radiomics features. During radiotherapy planning and delivery, three-dimensional imaging may allow the implementation of modern IGRT and refine the evaluation of dose distribution to healthy tissues. Moreover, information from diagnostic imaging may be integrated into radiotherapy planning to better tailor treatment (e.g., Simultaneous Integrated Boost to positive nodal disease detected with choline PET/CT) through co-registration with planning imaging. Finally, online MRI imaging allowed the development of MRI Linacs, able to safely deliver MRI-based IGRT with online adaptive protocols in order to shape radiotherapy planning on a daily basis, taking into account the inter-and intrafraction variability of patients' anatomy [75]. Limitations of these approaches are mainly related to three aspects: first, the sensitivity and specificity of novel diagnostic imaging (e.g., mp-MRI for prostate cancer) have to be further validated in clinical trials. Diagnostic imaging, with a particular focus on MRI, can play a crucial role in the distinction between benign and malignant conditions, thus avoiding the need for much more invasive approaches such as biopsies. In this regard, we cited Pietragalla et al. experience in 92 salivary gland patients, where diffusion-weighted imaging (DwI) and dynamic contrast-enhanced perfusion-weighted imaging (DCE-PwI) have proven useful in detecting benign neoplasms, epithelial malignancies, Warthin tumours and lymphomas [1]; distinguishing between benign chronic pancreatitis and pancreatic cancer by means of CT and MRI imaging was instead the main purpose of Srisajjakul and colleagues work [2]. Granata et al., always in the setting of gastrointestinal neoplasms, analyzed 88 patients and tried to outline peculiar MRI features which could allow the differentiation between intrahepatic cholangiocarcinoma and other forming-mass benign mimickers [3]. El-Shater Bosaily, Turkbey, Scialpi and Morote exploited the potential role of MRI diagnostic assessment in prostatic disease: they showed that the PI-RADS v2 scoring system could limit the detection of equivocal lesions, thus leading to the avoidance of unnecessary biopsies and to the accurate selection of patients who really deserve such an invasive diagnostic approach [5][6][7][8].
Among histologically confirmed tumours, new innovative technologies can provide better diagnostic refinement. As an example, Mungai and coll. showed how DCE MRI data can be helpful in recognizing the most frequent histopathological types of salivary neoplasms [4]; for what concerns prostatic disease, radiomic features can help in the process of risk-stratification: Gong et al., in their 489-prostate cancer patients experience, showed a correlation between radiomic features on mp-MRI and Gleason Score [12], as well as Santone et al. who demonstrated in a smaller sample size population the effectiveness of radiomics in grade group detection from MRI [13]. Functional MRI can also predict the risk of local recurrence after prostate surgery both at low PSA values, as demonstrated by Renard-Penna [14], and along with an increase in PSA with much more sensitivity, as in Coppola's work [15].
As a consequence, in the pre-treatment setting, this is of particular interest when diagnostic imaging could be integrated with clinical data to differentiate between benign or neoplastic findings, avoiding unnecessary histopathologic assessment through biopsy. Clinical trial design to address this issue is complicated and will probably require a large shared dataset of patients with a longitudinal follow-up. Moreover, as diagnostic imaging rapidly evolves, such databases could become obsolete if compared to current clinical standards. Currently, evidence from literature often does not allow to safely rely only on imaging for oncologic diagnosis. Larger series have been published for prostate cancer, but the number of patients included in these experiences is still too low (e.g., <1000 patients even in the largest series) to draw firm conclusions. Second, radiomics may offer the possibility to predict adverse events before treatment, but the assessment of this opportunity needs analysis of large populations and datasets. In this setting, radiomic features may play a role in assessing survival and toxicity outcomes and in predicting treatment responses in different cancer populations. Among prostate cancer patients, many studies investigated the potential of radiomic signatures in predicting radiotherapy-induced gastrointestinal (GI) and genitourinary (GU) toxicities. Mostafei's prospective work on 64 patients, whose pre-RT CT scans of the rectum and of the bladder were acquired and whose GI and GU toxicities were assessed, demonstrated that CT imaging features could predict radiationinduced toxicity and that integrating imaging and clinical/dosimetric features may enhance the predictive performance of radiotoxicity modelling [21]. Abdollahi explored RT-related bladder, rectal and femoral head injuries separately in patients who underwent IMRT for prostate cancer: in this study, 274 radiomic features were extracted from T2-weighted sequences, and changes from pre-RT and post-RT imaging showed a good correlation with radiation dose and radiation-induced urinary toxicity [40]; a similar work on prostate cancer patients treated with RT, and whose radiomic features on MRI T2-weighted scans were detected before and after treatment, resulting in the pre-IMRT predictive value of rectal toxicity [39]. Radiomic features can also predict femoral head injuries and the risk of post-treatment fractures in PCa patients. Abdollahi analyzed 30 patients and 60 femoral heads and extracted 34 features from T1, T2 and DWI MRI scans whose changes revealed useful in predicting the risk of fracture and with a potential role as biomarkers [41]. Lorenz conducted an exploratory study in which the analysis of delta-radiomic (variations in quantitative image metrics) profiles in organs at risk (OARs) revealed significant changes in the bladder and rectal wall just after 1 week of RT in prostate cancer patients whose MRI scans were acquired on an MRI linac [43]. A useful summary of available literature was provided by Desideri et al., who evaluated the current state of the art on radiomics prediction of radiation-induced toxicity in Head and Neck, Breast, Lung and Prostate cancer, with the conclusion that at present, radiomics is not readily applicable to patient management in clinical practice, but holds great potential for improved clinical decision making in precision radiation oncology [44].
For what concerns the prediction of survival as well, many series underlined the potential of radiomic features in NSCLC surgically treated patients [23], in hepatocellular carcinoma [19] and in neuroendocrine neoplasms [20].
Interestingly, in the setting of locally advanced rectal carcinomas treated with neoadjuvant chemo-radiation, texture analysis on T2-weighted MRI sequences was also investigated for the prediction of pathological response. If Crimì experience discouraged the use of radiomic features in the prediction of cPR in rectal cancer [25], on the other hand, Russo and colleagues proved the usefulness of MRI in the assessment of response in a setting of cervical cancer patients staged IB2-IIA1 [28].
The predictive function of radiomic features analysis for toxicity and treatment/survival outcomes is probably complicated by the fact that many radiomics features have been explored with different imaging methods, with significant heterogeneity in terms of aims and outcomes chosen for features of interest. Artificial intelligence will be needed to address this issue [76]. For these reasons, it is difficult to predict when radiomics features could be steadily implemented in clinical practice because standardization in this field is yet to come. The third issue regarding radiomic features addresses both radiotherapy planning and delivery. With a peculiar focus on prostate cancer, superior accuracy in both phases of the radiotherapy process has been related to the implementation of novel imaging technologies. Among these, we find new PET tracers, such as 18F-fluciclovine, PSMA, coline or Carbon-11-ACE as well, which play a part in decision-making and planning [45,46], promoting dose-escalation [54,58], and guiding local ablative radiotherapy, thus avoiding the need of immediate systemic therapy [59,60]. In the delivery phase, Image-Guided Radiation Therapy [IGRT] is crucial in improving the precision and the accuracy of treatment, as it increases the probability of tumour control with shorter treatment schedules. IGRT is nowadays the standard of care for all cancer types. In prostate cancer patients, for example, it allows for a safe, moderate hypofractionated regimen [49]; in addition, the guidance of molecular imaging was the central issue in Rigo and colleagues' work on 24 prostate cancer patients undergoing salvage radiotherapy after HIFU with efficacy and low toxicity [55].
The use of Cone-Beam CT (CBCT) in locally advanced rectal carcinoma is effective for mesorectum organ motion assessment, and it is an appropriate method for treatment implementation [48]. The role of MRI-guidance was also explored in this context: we cite Sanmamed's work, aimed at assessing the impact of doses to the bladder neck on the development of acute GU toxicity and on QoL in patients treated with MRI-guided brachytherapy and EBRT [53].
Paradoxically, the implementation of novel imaging in RT treatment planning is currently the most advanced field of research and, in our opinion, the most exploitable one. Indeed, RT technologies are often routinely available and are often already well implemented in clinical practice. Moreover, direct clinical outcomes and benefits are evaluable after treatment (e.g., survival outcomes, response to treatment, etc.), and trial design is more traditional if compared to other research fields discussed above. However, much literature evidence is limited to retrospective case series, and prospective evidence should be advocated because the real clinical advantages of all techniques should be thoroughly evaluated to select patients who may benefit from a certain radiotherapy approach. Interestingly, the relationship between modern imaging methods and radiotherapy may constitute a paradigmatic example of a research field in which it would be complicated to wait for pre-treatment evidence, and assessing clinical oncological benefit from a certain approach could be more effective and less demanding in terms of the number of patients included in trials. Thus, a tight relationship between radiologists and radiotherapists should be advocated both in routine clinical practice and trial design in order to improve knowledge and maximize clinical benefit from technical innovation. Lastly, we must mention the fact that local treatment implemented by the integration of modern technologies may have an impact on systemic therapy in both oncological and non-oncological conditions. In prostate cancer, complete metastatic ablation of oligometastatic disease with SABR and integrated PSMA-PET may provide an alternative to early initiation of androgen deprivation therapy [61]; SBRT can also be useful in early prostate cancer in particular conditions (such as covid19 pandemics) requiring shorter treatment time [72]. SBRT was successfully implemented in oligometastatic colorectal cancer patients [62] and in patients with lung disease, providing good local control with good tolerability [63]. For what concerns non-oncological settings, we cited Fiorentino and colleagues' work on cardiac arrhythmias: clinical data have proven the feasibility, efficacy and safety of SABR (25 Gy in one session) for refractory ventricular arrhythmias [64].

Conclusions
Integration between diagnostic imaging and radiotherapy is of cornerstone importance for the modern treatment of prostate cancer (Figure 1), and many enticing innovations are awaited to further improve the therapeutic ratio of radiotherapy in this field. However, awareness of the close interdependence between these two specialities is useful to underline the strengths and limitations of current radiotherapy approaches. Future clinical trial design should be aimed at exploring the real clinical benefit of complex workflows (e.g., diagnostic imaging, co-registration with planning imaging, dosimetric evaluation and online set-up correction) in clinical practice. Some recent advances in metabolic imaging may significantly improve staging in selected settings (e.g., PSMA-PET for prostate cancer disease), especially considering the detection of oligometastatic disease in patients candidates for SBRT [77]. However, many gaps still remain to be filled, radiomic and genomic biomarkers may further improve the predictive capacities of imaging methods, and the rise of deep learning techniques could increase synergism between modern imaging and radiotherapy [78]. innovations are awaited to further improve the therapeutic ratio of radiotherapy in this field. However, awareness of the close interdependence between these two specialities is useful to underline the strengths and limitations of current radiotherapy approaches. Future clinical trial design should be aimed at exploring the real clinical benefit of complex workflows (e.g., diagnostic imaging, co-registration with planning imaging, dosimetric evaluation and online set-up correction) in clinical practice. Some recent advances in metabolic imaging may significantly improve staging in selected settings (e.g., PSMA-PET for prostate cancer disease), especially considering the detection of oligometastatic disease in patients candidates for SBRT [77]. However, many gaps still remain to be filled, radiomic and genomic biomarkers may further improve the predictive capacities of imaging methods, and the rise of deep learning techniques could increase synergism between modern imaging and radiotherapy [78]   Funding: This research received no external funding.

Conflicts of Interest:
The authors declare no conflict of interest.