< Back to previous page

Project

Balancing Innovation With Optimization in Oncology: Avenues for Bridging the Cancer Clinical Research Gap

Cancer is an umbrella term referring to a group of debilitating and potentially fatal diseases characterized by the uncontrolled growth and eventual spread of cells that have undergone genetic mutations. Together, these diseases pose an enormous burden on healthcare systems worldwide, killing millions of people each year, and costing the global economy close to a trillion dollars annually. There are various therapeutic modalities available to treat malignant neoplasms, with the three main ones being pharmacotherapy, surgery, and radiotherapy. In accordance with the principles of evidence-based medicine, any anticancer treatment should be thoroughly tested before being given to patients in the clinic, with the goal of empirically confirming its safety and efficacy. It is against this backdrop that clinical trials are undertaken, which are interventional studies evaluating the performance of an investigational therapy in a select number of human volunteers, who suffer from the condition of interest themselves if that therapy is intended to shrink or stabilize tumors. When these studies feature a control group and rely on randomization for deciding which intervention is administered to a particular participant, they are labelled randomized controlled trials (RCTs).

RCTs individually and collectively deliver some of the most robust evidence that can underpin medical and policy-related decisions. However, in oncology, such trials have several important shortcomings that undermine the applicability of their findings. Firstly, their comparator arms may not always reflect the standard of care, or more broadly speaking, constitute a clinically relevant product or procedure, rendering them inappropriate benchmarks. Secondly, their samples are frequently poorly representative of the general patient population due to the myriad of inclusion and exclusion criteria that are employed during their recruitment process. Thirdly, their primary endpoints are usually surrogate outcome markers which oftentimes lack validation as proxies for more patient-centric measures like overall survival or quality of life. Fourthly, since they may take years to conclude and are subject to heavy regulatory scrutiny, they are typically expensive to perform, and it is commonly only the pharmaceutical industry that can afford to sponsor them. As companies primarily seek to increase their profits, they will ordinarily neglect to pursue research ideas which are not commercially viable, even though ideas of this kind can lead to the development of highly effective strategies for treating cancers.

These shortcomings are not being adequately addressed within the current framework for developing antineoplastic treatments, resulting in the emergence of an evidentiary void that is situated at the interface between the pre- and post-licensing research activities carried out under that framework. This void, which has been called the cancer clinical research gap, severely complicates the work of decision-makers operating downstream from the marketing authorization milestone, such as health technology assessment bodies, payers, physicians, and patients. The uncertainties with which those stakeholders are faced are detrimental to their decision-making and contribute to the undue uptake of ineffective and possibly unsafe interventions into the existing armamentarium of oncologists. This is not just harmful to the individuals who are treated with those interventions and yet whose real needs remain unmet, but also to society as a whole, because it gives rise to the unwarranted coverage of drugs, devices and techniques which are extremely costly owing to their exorbitant prices.

To bridge the cancer clinical research gap, evidence is needed that can help answer the questions which are left unaddressed at the moment. In this PhD project, three distinct (though not mutually exclusive) strategies to produce such evidence were explored. The first strategy involves the generation of real-world evidence (RWE) through the analysis of observational real-world data (RWD), i.e. health data collected outside of RCTs, generally (but not necessarily) in a routine fashion and for other purposes originally than supporting any research efforts. The second strategy consists of conducting treatment optimization studies, which is to say, clinical trials that are designed to optimize the way health technologies are applied in real-life circumstances by reducing the toxicity and/or costs associated with the use of those technologies. These interventional studies focus for instance on investigating whether the dosing regimen of antitumor agents can be safely de-escalated. The third strategy entails executing trials that are set up to emulate how the experimental medicine or practice would be deployed in an off-study environment. Trials of this nature are denoted by the descriptor ‘pragmatic’ and characterized by their eligibility criteria being more relaxed than those of their explanatory counterparts, their primary endpoints being of direct relevance to patients, and their follow-up schedules not being very intense, among other things.

In total, five studies were performed to scrutinize these three strategies.

The observational approach to closing the gap was examined by means of two surveys, one of which was sent to cancer clinicians who are part of the network of the European Organisation for Research and Treatment of Cancer (EORTC), and the other to representatives of European, North American, South American, Asian and Oceanian cooperative groups known for undertaking academic clinical research in oncology. The former found that those clinicians had positive views of RWE, and that they were already actively assisting with its production. At the same time however, the respondents to this survey did not think such evidence could fill the gap by itself, perceiving the methodological hurdles of analyzing observational RWD as particularly challenging to overcome, and not believing these data were equally useful in all contexts. The questionnaire that was targeted towards the aforementioned groups revealed that they had run analyses of non-interventional RWD before, and that they had experienced the methodology-linked limitations of those analyses as being difficult to surmount. Moreover, it showed that they had tackled some uncertainties more often than others using RWD of that type.

Although they can be combined with each other, the two interventional methods by which the gap could be filled were assessed separately. With regard to treatment optimization studies, semi-structured interviews were organized with experts belonging to different stakeholder factions, and an analysis was conducted of an extensive cohort of these studies extracted from the database of the EORTC, which is the largest and oldest sponsor of investigator-initiated cancer RCTs in Europe. The persons who were interviewed felt that there was a strong need for optimizing the application of anticancer therapies, and while sometimes expressing diverging opinions, had certain preferences regarding the design, execution, funding, timing, and setting of trials launched in this respect. Furthermore, the interviewees identified three policy options through which these trials could be integrated into the present clinical research paradigm. The review of the earlier referenced cohort of EORTC-coordinated treatment optimization studies yielded an empirical model for carrying out such studies, highlighting that they can be multi-country RCTs displaying at least some elements of pragmatism, and that they have the potential to be impactful from a scientific point of view. Concerning pragmatic trials, studies which had been assigned the ‘pragmatic’ label and which investigated antineoplastic interventions were extracted from the literature and evaluated in terms of the degree of pragmatism they exhibited, based on the validated PRECIS-2 instrument. The extraction exercise underscored how rare it still is for oncology trials to carry the ‘pragmatic’ tag despite its rising popularity, and the subsequent PRECIS-2 evaluation demonstrated that this tag should not be taken at face value, considering how few cancer trialists seem to apply it correctly.

Overall, the results of the five studies which have been incorporated into the present PhD dissertation indicate that, notwithstanding the fact that the three proposed strategies for bridging the cancer clinical research gap are complementary to each other, they are not all capable of narrowing it to the same extent. More specifically, observational RWD have inherent weaknesses which undermine their ability to conclusively deal with the unanswered questions underlying said gap, and which are perceived as problematic by the actors who have to interpret the RWE derived from these data. Additionally, such RWD can exclusively offer insights into non-original uses of treatments. For those reasons, the conduct of RCTs in the form of treatment optimization studies with a pragmatic setup is necessary. Ideally, these trials are executed on an international scale to maximize their impact and to ensure the widespread uptake of their conclusions. In addition, it stands to reason that independent parties should undertake them, and that any research endeavors of this sort should be supported with public means, in light of how informative they are to stakeholders involved in setting or enacting healthcare policies. The cooperative groups may be best positioned to start performing them systematically, having the partnerships and the infrastructure in place to manage large RCTs running across multiple countries.

Date:10 Oct 2018 →  12 Sep 2023
Keywords:Clinical research, Artificial intelligence, Pragmatic clinical trials, Big data, Treatment optimization, Real-world data, Real-world evidence, Machine learning, Precision medicine, Oncology
Disciplines:Drug discovery and development, Medicinal products, Pharmaceutics, Pharmacology, Pharmacotherapy, Other pharmaceutical sciences
Project type:PhD project