The challenges of measuring response in interventional oncology


A rapid overview of three of these treatments: ablation, chemoembolization and selective interval radiation therapy, should help clarify the challenges and difficulties of determining tumour response in interventional oncology, write Maxime Ronot and Valérie Vilgrain.

Since the introduction of anticancer treatments, a simple relationship between tumour burden—ie. the quantity of cancer—and survival—ie. the length of life—has been suggested. Even if this relationship has been shown to be non-linear, it is the common ground for analysis in oncology. More important, the prediction of outcome in oncology is based on the assumption that tumour burden is a valid surrogate for patient survival. Within this context, imaging is expected to provide an accurate evaluation of tumour burden, and more importantly its progression following treatment, or tumour response. Because the entire volume of a tumour cannot be quantified in each patient within a delay that would be compatible with daily practice, imaging criteria have been developed based on two simplifications that should be kept in mind: first, that a selected sample of a tumour is representative of the entire tumour volume, the notion of a target lesion, and second, that one or two-dimensional measurement of lesions can provide an acceptable approximation of their volume. World Health Organization (WHO) criteria and more recently Response Evaluation Criteria In Solid Tumours (RECIST) are both first generation imaging-based morphological response criteria. Since they were first introduced, these criteria have been extensively validated and endorsed by national and international medical and health organisations.

Interventional oncology as part of the broader oncological community is no exception to this. However, because much of interventional oncology deals with liver cancer, assessment of tumour response is facing new challenges because of newer procedures performed and the tumour being treated. Indeed, the goal of most procedures is to induce necrosis rather than tumour shrinkage. As a result, conventional dimension-based or morphological response criteria are not accurate, and have been shown to markedly over- or more frequently under-estimate tumour response. It is important to note that many tumours treated in interventional oncology procedures are hypervascular. Once again, the consequence of treatment is more a change in the inner structure of the lesions, especially the microvascular network, than in tumour size.

Based on these observations, in the past 15 years, a paradigm shift has occurred and several teams have suggested that response criteria should focus on the viable portion of lesions, defined as the portion showing contrast enhancement, rather than on their size. As a result, numerous new imaging criteria have been introduced. A second consequence is that response criteria have gradually been adapted to the different types of treatment available. A rapid overview of three of these treatments: ablation, chemoembolization and selective interval radiation therapy, should help clarify the challenges and difficulties of determining tumour response in interventional oncology. Ablation is a good example of consensual and robust imaging criteria defining tumour response. There is an accepted and reproducible definition of response, with valid radio-pathological correlations, consensual terminology and guidelines and recommendations endorsed by national and international societies. Therefore, major changes are unlikely in the years to come and collective efforts are being made to define the optimal follow-up schedule.

Assessment of tumour response after chemoembolization is more complex and there is less consensus. The European Society for the Study of the Liver (EASL) criteria and the modified RECIST criteria have been introduced in the past decade. It is interesting to note that they were not based on experimental or observational studies, but were proposed as revised versions of WHO and RECIST criteria. Initial reports showed that they were better than the latter for assessment of response, and both have been shown to be independent prognostic factors. Nevertheless, these criteria have been shown to have several limitations, mainly the lack of standardisation, and there are concerns about applicability and reproducibility that have been raised. Indeed, they may be difficult to use, especially in heterogeneous lesions, and their use is dependent on operator experience. Although recent guidelines have acknowledged the potential value of these new criteria, they are not considered robust enough to replace older morphological criteria in trials. As a result, since they were first introduced, numerous studies have been published to better define the type and optimal number of target lesions, the ideal imaging technique, and the follow-up schedule. At present most teams perform one-dimensional (mRECIST) or two -dimensional (EASL) measurement of the enhanced portion of a maximum of two target lesions. Nevertheless, very recent data have suggested that three-dimensional evaluation of the whole tumour burden using specific software may be of interest. This is an interesting approach that will probably offer a more reproducible evaluation of tumour response in the years to come.

The effects of Selective Internal Radiation Therapy (SIRT) are even more complex. This treatment induces a combination of necrosis, inflammation and fibrosis, and morphological changes are delayed, very often for up to three months after treatment. Thus, although dimension-based criteria such as EASL/mRECIST have been shown to be useful, they are even more difficult to apply. In these cases, functional and molecular imaging can play an important role. Indeed, diffusion-weighted or perfusion imaging, and positron emission tomography have been shown to confirm response assessed according to morphological criteria, more important to identify responders earlier and more accurately than the latter. Thus, functional parameters may be interesting biomarkers and be complementary to more conventional criteria. Recent studies have also reported that they can predict response before treatment is begun. Unfortunately, these studies have only included small populations. These criteria cannot yet be used in daily practice and further standardisation and validation are needed.

These three examples clearly show the recent advances in this field, and illustrate different steps to the evaluation of tumour response. They outline the areas that still need to be developed in order to improve best practice: the importance of the viable portion of tumour; adapting imaging criteria to the different types of treatment; using common terminology; and identifying the optimal time to evaluate response. They also show the main challenges to be met in years to come: defining the optimal follow-up schedule, standardising criteria using (semi) automatic approaches, evaluating the whole tumour burden, and finally defining the respective role of morphological and functional imaging in this process.

Maxime Ronot is with the Radiology department, Beaujon University Hospital, Clichy, France. Valérie Vilgrain is with the Radiology department. Beaujon University Hospital, Assistance-Publique, APHP, Clichy, France. Both Ronot and Vilgrain are affiliated to the Université Paris Diderot. Vilgrain is also affiliated to the Sorbonne Paris Cité, France, and INSERM U1149 CRB3, Paris, France. The authors have reported no disclosures pertaining to this article