Here, we would like to highlight the need to develop the following key areas of imaging to help achieve value-based, precision cancer care: molecular (primarily nuclear) imaging and therapy, interventional radiology, and radiology informatics.
The ability to image not just anatomy but actual molecular and cellular processes as they occur has long tantalised the medical community. Molecular imaging has existed for decades, but until recently, its growth has remained frustratingly slow. For many years after the advent of clinical positron emission tomography (PET), 18F-FDG remained the only PET tracer approved by the United States Food and Drug Administration (FDA) for clinical use.
For a number of reasons, however, progress in molecular imaging has recently begun to speed up, giving much cause for optimism that the field will fulfill its potential. A number of new tracers have recently received FDA approval, and dozens more are in clinical or pre-clinical trials. Unlike 18F-FDG—which is a marker for the elevated glycolysis that is a hallmark of cancer but not specific to it—many of the newer PET tracers are highly specific, targeting molecular entities such as prostate-specific membrane antigen (PSMA, which is overexpressed in prostate cancer) or estrogen receptors.
In addition, new probes and approaches for optical imaging, as well as the advent of combined PET/MRI and clinical hyperpolarized MRI (a technique that allows in-vivo assessment of chains of metabolic events), are adding still more dimensions to molecular imaging for pre-clinical research and clinical use. As a result, we are seeing more and more potential applications of molecular imaging in clinical decision-making for oncology, including selection of conventional as well as molecularly targeted treatments; dose-finding; and early assessment of treatment response.
With regard to the selection of treatments for precision cancer care, interventional radiology has a key role to play, particularly given its capacity to allow targeted biopsies. Acquisition of tissue samples adequate for complex molecular analyses is crucial for the appropriate selection of targeted therapies but is far from a given in many cases. For example, in the United States, analysis of the interim results of the National Cancer Institute Molecular Analysis for Therapy Choice (MATCH) trial of targeted therapies, in which patients were matched to treatments based on in-depth molecular analyses, found that around 13% of patients’ biopsies were inadequate for this purpose. This realisation has catalysed research efforts to improve the quality of biopsies, bringing together tools and expertise from interventional radiology and other disciplines, including computer science. By enabling consistent, high-quality biopsies, interventional radiologists will have an opportunity to make their work increasingly central to value-driven, precision cancer care, particularly as the number of molecularly targeted treatments available grows.
Advanced interventional radiology suites now feature not only fluoroscopy but also hybrid imaging equipment, including combinations of cross-sectional imaging modalities such as CT, MRI or in some cases even PET/CT with single- or bi-planar angiography. Expanding the ability to characterise tissue biology via these modalities should enhance the capacity of interventional radiologists not only to perform optimal, biologically targeted biopsies but also to provide effective treatments. The use of molecular imaging in the IR suite has already been shown to aid target localisation and facilitate immediate assessment of treatment, before the patient leaves the operating table. Because of its focus on minimal invasiveness, IR is ideally suited to address the goals of achieving precision and value/efficiency in cancer care. To ensure that IR contributes as much as possible to precision oncology, it is essential that we aim to integrate cross-sectional and molecular imaging modalities into IR suites whenever possible and that we develop IR physicians with the capacity to interpret these modalities.
Alongside the development of molecular imaging probes, the development of novel targeted radionuclide therapies and of theranostic agents that allow both targeted imaging and treatment has also been picking up speed. Examples include the development of lutetium-177 (177Lu)-DOTATATE, which has been found to lengthen progression-free survival in patients with advanced midgut neuroendocrine tumours; and the development of radiolabeled ligands of PSMA, which, in early clinical studies, have yielded highly promising results for the treatment of metastatic prostate cancer.
To fulfill the potential of radionuclide therapies and theranostics and maintain progress in these areas, it will be critical to dramatically increase the recruitment of physicians for training in molecular imaging and nuclear medicine and the availability of training programmes in these fields. Furthermore, we must work to increase both the supply of radiochemists and other personnel and the maintenance and expansion of the complex infrastructure necessary for advancing these fields.
Last, but not least, we would like to highlight the growing importance of radiology informatics to the success and advancement of oncologic imaging as well as biomedical imaging as a whole. Rather than being a source of support in scattered areas, it is destined to become a pillar of radiology practice. We are now in the era of the “fourth industrial revolution,” in which the integration of disciplines and technologies, including machine learning and artificial intelligence, is leading to increasingly rapid innovation and the weaving of computer tools more and more deeply into the fabric of daily life.
Informatics tools for gathering and analysing data have the potential to monitor various aspects of daily radiological practice, facilitating efforts to improve both quality and efficiency. Furthermore, machine learning and artificial intelligence can be used to extract additional, clinically relevant data from images and enable faster identification and characterisation of abnormalities. Therefore, informatics tools will clearly be indispensible for helping radiology practices make the transition from volume-based healthcare to value-based healthcare that maximises both quality and efficiency.
Turning away from machine learning and AI out of fear that they will replace us is not an option. Rather, radiologists must embrace these new tools, remembering that our ultimate purpose is not to “decode” image patterns or analyse texture in images but to integrate all imaging findings with clinical and other findings and help solve clinical problems. We need to participate in developing machine learning and AI tools that will help us, and we need to become masters in their application to clinical care.
At present, the value that radiologists provide is, unfortunately, often overlooked. In the predominant value-based healthcare models, a correct diagnosis is taken for granted: Measurement of value begins only with the start of therapy, and the impact of radiology on value is calculated solely in negative terms (i.e., when radiology is a source of diagnostic error).
However, anyone who has been through residency training in radiology knows that being a good radiologist requires a great deal of learning and practice. Numerous studies have shown that sub-specialisation further improves radiologists’ interpretive abilities and that direct consultations between radiologists and referring physicians affect clinical decision-making. Moreover, radiologists carry out many other demanding responsibilities, from assessing the appropriateness of imaging requests, to adjusting imaging protocols, attending to radiation protection needs, communicating with patients, managing radiology personnel and performing research to move the field forward.
Thus, we already know that radiologists add value, but we need to make this clear to policymakers and the broader public. In addition to enabling us to contribute even more value to clinical care, informatics tools will be essential to develop metrics that demonstrate our work’s value. In short, if we embrace change, the future of radiology—and especially oncologic imaging—will shine more brightly than ever.