Clinical Trials in Kidney Transplantation: Design Considerations and Novel Approaches

Abstract

Background: Clinicians face considerable uncertainty in the management of common complications related to kidney transplantation including cardiovascular disease, cancer and rejection. A well-designed randomised controlled trial (RCT) is the ideal study type to help answer questions in areas of clinical equipoise. The deficiencies of nephrology RCTs are well known to the nephrology community. Together with the known issues of logistics, cost and lack of infrastructure, kidney transplantation trials face unique challenges that impact study design, trial feasibility and ascertainment of trial outcomes. Given current failure to improve long-term outcomes for kidney transplant recipients, there is a need to enhance trial design and feasibility by making trials more efficient and more focussed on what matters to patients and clinicians. Methods: This thesis utilises novel approaches to trial design in kidney transplantation and is presented as a thesis of published works. Various clinical research methods were used to test the hypothesis that improvements in trial design can lead to the generation of new data, at lower cost and greater efficiency to improve patient outcomes. The first section, presented in Chapters 2 and 3 of the thesis, reports on the trial design considerations and implementation of the Canadian-Australasian RCT of Screening Kidney Transplant Candidates for Coronary Artery Disease (CARSK). Trial procedures were incorporated into routine care, improving recruitment feasibility and complete outcome ascertainment. A pretrial Markov microsimulation model was also constructed using clinical, costs and utility estimates derived from published literature to estimate the benefits (e.g. survival) and trade-offs (e.g. costs and complications) of the two interventions being tested from a health systems perspective. The pre-trial model defined areas of evidence uncertainty in to help inform data collection for the trial. The second section of the thesis, presented in Chapters 4 and 5, describes a novel method for long-term outcome ascertainment of hard outcomes of kidney transplantation trials. Participant outcomes from five multi-centre RCTs of everolimus-based immunosuppression were linked to the Australian and New Zealand Dialysis and Transplant (ANZDATA) Registry. Meta-analyses using individual participant data were performed to ascertain long term cancer incidence, graft function and patient survival. The third section, chapter 6, presents a systematic review and meta-analysis of controlled trials in the treatment of antibody-mediated rejection (AMR). The risk of bias for all studies and the quality of evidence (for graft survival) were analysed for each treatment regimen, highlighting the deficiencies in existing data, and exemplifying the challenges facing kidney transplantation trials. Results: New data were generated from each of the studies. The CARSK trial was rolled out at 12 Australasian sites, achieving 525 (48%) of the Australasian recruitment target over 2.5 years. Recruitment remains on-going at these sites, with Canada commencing recruitment in early 2019. Further expansion of new sites into Asia and Europe are expected. The pretrial model for CARSK provided strong reassurances that the new intervention would be cost-effective and improve patient survival, based on evidence synthesised from current literature. The lack of robust evidence (in particular of the costs of transplantation in the modern era) provided strong justifications for doing the trial and also for including a strong health-economic component. By linking individual participant data to an established registry, novel evidence for the benefits and potential problems with everolimus-based immunosuppression were derived. Everolimus with reduced dose calcineurin inhibitor (CNI) combination was associated with a 56% reduction in the development of non-melanoma skin cancers compared with standard CNI-based triple therapy. However, the poor tolerability of everolimus in trials suggests widespread use of everolimus-based immunosuppression as first-line maintenance therapy may be problematic. In the systematic review and meta-analysis of the treatment of acute and chronic AMR, plasma exchange and IVIG have emerged as the standard-of-care, despite the lack of any highquality evidence supporting use of this combination. Furthermore, newer agents targeting B cells, plasma cells and complement blockade have failed to demonstrate efficacy in the treatment of AMR. The quality of evidence for the majority of treatment strategies for AMR is low and has not improved with time. This remains an area of great unmet clinical need. Conclusion: The thesis has demonstrated that careful planning and innovative use of available resources can improve future trial design and generate meaningful data. Multicentre collaborations are required to achieve acceptable power and generalisability. Pre-trial modelling should be considered as an adjunct to trials to examine evidence uncertainties and further improve the efficiency and quality of data collection. Trial procedures should be integrated into routine care where possible to maximise efficiency, engagement and potentially enhance translation. Linkage of outcome data to existing registries such as ANZDATA should be considered a priori so that long-term patient-centred outcomes, as opposed to short-term surrogate outcomes, can be ascertained with reliability and at acceptable cost.