By Durga Chandrupatla
In recent years, several developments have dictated the role of radiotherapy in cancer care. There have been landmark gains in particle therapy, particularly in proton therapy; electron brachytherapy; treatment planning systems and applications; and real-time image-guided radiotherapy.
First, external beam radiation therapy (EBRT) systems that use photons or electrons are widely used for their proven clinical benefits, accessibility, and reimbursement policies. Meanwhile, proton and, to some extent, carbon-ion beam therapy are increasingly being recognized for their lower radiation doses to healthy tissue and better patient outcomes. However, proton and carbon-ion therapies warrant more multicenter clinical trials. Moreover, a significant return on investment factor represents a bottleneck for proton and carbon-ion therapy.
Second, despite its efficiency, brachytherapy is underutilized and is, to some extent, declining. The decline is due to advancements in image-guided radiotherapy, accurate dose calculation models, a low number of trained personnel, and inadequate reimbursement for brachytherapy. Despite these challenges, electronic brachytherapy has found new opportunities in IR therapy for breast and prostate cancer.
Last, treatment planning systems/softwares are evolving at a rapid rate due to advancements in AI. Machine learning applications are burgeoning in image simulation, treatment planning, dose delivery, and quality assurance.
MRI and CT-guided radiotherapy provide real-time tracking and tumor motion correction. These are especially important for lung and gastrointestinal tumors, which are subject to motion during radiotherapy. Elekta’s Unity and Viewray’s MRIdian are currently the only systems that combine the soft tissue visualization capabilities of an MRI with simultaneous irradiation of the tumor. Accuray recently introduced its Radixact system with helical tomotherapy, with the goal of improving dose accuracy and customizing the treatment planning based on changing tumor characteristics.
Globally, there are 75 operational proton therapy facilities, with 30 facilities in the United States. Proton therapy is substantially different from photon or electron therapy and provides efficient and accurate dose conformity. Individual clinical studies of proton vs photon therapy have demonstrated higher survival rates, lower risk to normal tissue, and organs at risk with proton therapy. For example, a study by Liu and colleagues demonstrates the superior dose conformity of proton over photon therapy. However, the biggest challenges for proton therapy are that it is capital intensive and there aren’t many long-term multicenter randomized clinical trials demonstrating a clear advantage over photon therapy. Reimbursement represents another bottleneck affecting adoption of proton therapy—proton therapy is not covered by insurance agencies.
Another particle therapy nearing international adoption is carbon-ion beam therapy. Japan, Italy, China, Austria, and Germany are the only countries currently offering carbon-ion therapy, with Japan operating five facilities that have treated more than 10,000 cancer patients. Clinical studies from Japan have shown efficacy in prostate, lung, and pancreatic cancer, with an overall five-year survival of >47%. At this time, Hitachi’s particle therapy system is the only source that offers commercial carbon-ion radiotherapy.
The application of brachytherapy has been steadily declining in recent years. In a survey among European Union countries, radiologists pointed out that the decline is due to a low number of qualified personnel, inadequate reimbursement, and underuse of brachytherapy equipment in image-guided interventions. Meanwhile brachytherapy has found a new window of opportunity, mainly in intraoperative radiation therapy (IORT). Breast and prostate cancer have been the major areas where brachytherapy is finding increasing acceptance. Moreover, with recent advancements in intraoperative CT and MRI, image-guided brachytherapy has seen new treatment approaches. IntraOp, Zeiss, and Xoft offer electronic brachytherapy, with IntraOp being the only vendor offering integrated, CT-guided IORT.
A long-term clinical study (>10 years) demonstrated that, when treated within eight weeks, brachytherapy is superior to chemoradiation for overall survival rate of cervical cancer. In localized prostate cancer, Tsubokura and colleagues showed that brachytherapy was superior to EBRT, with EBRT indicating more gastrointestinal toxicity and higher five-year biochemical failure-free survival rate. Additionally, 10-year results of the PORTEC-2 trial comparing vaginal brachytherapy with EBRT in patients with high- to intermediate-risk endometrial carcinoma demonstrated a decrease in gastrointestinal toxicity and better social functioning.
Another important application area is surgical guided brachytherapy for intracranial neoplasms. The results of a clinical study from the Barrow Neurological Institute were recently shared at the 2019 Annual Scientific Meeting of the American Association of Neurological Surgeons. Brachytherapy showed superior local tumor control and overall survival when compared with conventional treatment options. Furthermore, GammaTile, developed by GT Medical Technologies, has made advances in treating recurring brain tumors with a cesium seed embedded in a collagen-based implant. It has shown an excellent safety profile (8% radiation-related changes in the brain) compared with EBRT (5% to 24%).
There is a growing need for integrating AI in cancer treatment, especially in treatment planning systems/software. At present, it takes a significantly long time to manually mark cancerous and healthy tissue on CT and MRI scans ahead of radiotherapy. However, with machine learning algorithms, this process could be automated while achieving faster, more accurate, and more sensitive segmentation of medical images. Some of the key areas where AI can create a meaningful impact are in multimodality intraoperative image reconstruction, eg, CT to MRI and vice versa, accurate segmentation of organs at risk and tumors, remote treatment planning, adaptive radiotherapy treatment planning, and predicting long-term outcomes in cancer patients.
Nguyen and colleagues mapped the desired radiation dose distribution using deep learning from a patient’s planning target volume and organs at risk contours, which showed the feasibility for predicting radiation dose using AI. Radiology AI vendors provide tools such as triage, clinical support, and retrospective analysis using deep learning technologies. These tools can be leveraged by radiologists in detecting tumors, planning treatment, and monitoring treatment response. Machine learning models can also be built for predicting adverse events in cancer patients exposed to radiotherapy.
For example, Yu and colleagues developed a machine learning model that can predict radiation pneumonitis, which provides clinicians with an opportunity to make radiotherapy treatment changes. Moreover, Hosny and colleagues provided indication that deep learning models based on CT images can be used for mortality risk stratification from nonsmall cell lung cancer patients. In addition, many vendors have developed machine learning algorithms that can segment normal vs tumor tissue and automate treatment planning, thus improving the efficiency and consistency of radiotherapy in the clinic.