Doctor: Jules Grégory
Title: Bias and methodological specificities of randomized controlled trials in interventional radiology: the example of treatments for hepatocellular carcinoma
Supervisor: Isabelle Boutron
Doctoral school: ED 393 Epidemiology and Biomedical Information Sciences, Université Paris Cité
Date of thesis defense: 19/12/2022
Jury: Corine Alberti, Lydia Guittet, Hicham Kobeiter, Christophe Aubé, Isabelle Boutron, Maxime Ronot, Valérie Vilgrain, Boris Guiu
The randomized controlled trial (RCT) methodology was first used in interventional radiology (IR) in 1985 in a trial evaluating transarterial chemoembolization (TACE) for the treatment of hepatocellular carcinoma (HCC). The specialty has progressively adopted this evaluation method for new treatments and technical innovations. However, the application of this methodology to IR raises, as for other non-pharmacological treatments (TNPs) (i.e., involving treatments other than drugs), specific methodological issues related to the complexity of the interventions, the influence of care providers, the expertise of the center and the difficulties of blinding, which have a major influence on the planning, implementation and interpretation of RCTs.
In a first project, we assessed the extent to which the results of registered RCTs assessing TACE for treating HCC were publicly available. Among 67 identified RCTs, including a total number of 11 599 participants, almost two-thirds did not yield any public results, either on the registry platform or in scientific journals. Underreporting of trial results is a major cause of wasted medical research since inaccessible research results fail to help both patients and clinicians.
In a second project, we determined the proportion of RCTs on TACE for HCC, published after 2007, that were registered. We compared registered primary outcomes with those reported in publications to determine whether selective outcome reporting favored significant outcomes. We evidenced that half of these published RCTs were not registered, and when registered, one-third had major discrepancies between the registered and published primary outcomes. Selective outcome reporting favoring the trial results was observed in 7% of published reports.
In a third project we assessed the completeness of reporting of abstracts in RCTs for liver IR and the impact of the publication of the CONSORT extension for NPT 2017 update. We assessed the evolution trend over time using a time-series analysis. We found that completeness of reporting was poor in abstracts of trials for liver IR and did not improve after the publication of the CONSORT-TNP statement.
In a fourth project we worked on SIRT and developed a textbook outcome (TO) in order to propose a comprehensive standardization tool, suitable for routine care, clinical round, and research. This study involved two steps: (1) the steering committee first developed an extensive list of possible relevant items reflecting an optimal SIRT intervention based on a literature review, (2) then conducted an international and multidisciplinary survey which resulted in the final TO. This final TO consisted in 11 parameters across six domains corresponding to the different steps of a SIRT. Of these, all but one were applied in the institutions of >80% of experts.
Our work provides areas of improvement for IR RCTs through 1) systematic publication of trial results, including negative results, to reduce research waste; 2) prospective registration of trials to ensure quality and transparency; 3) improved reporting of trials through abstracts providing specificities of non-pharmacological interventions; and 4) standardization of interventions through dedicated outcomes capturing its essential dimensions.