#5309 MODELLING LONG-TERM OUTCOMES FOR PATIENTS WITH IMMUNOGLOBULIN A NEPHROPATHY (IGAN) FROM SHORT-TERM PROTEINURIA DATA

Abstract

Abstract Background and Aims Immunoglobulin A nephropathy (IgAN) is rare kidney disease that leads to glomerular injury, progressive loss of kidney function and progression to kidney failure. Pre-specified interim proteinuria data from ongoing clinical trials is being used as the basis of regulatory approval of new therapies but little is known about the translation of these results to long-term patient outcomes; a topic of particular interest to heath technology assessment (HTA) groups looking to inform decisions on reimbursement before final renal outcome data are available. The objective of this study was to model long-term outcomes for patients with IgAN based on short-term proteinuria data. Method We developed a de novo model including a short-term (36-week) decision tree and a long-term Markov model to capture expected lifetime outcomes associated with treatment for IgAN. We used achievement of proteinuria <1 g/day as a clinically relevant treatment target. The short-term model included data on proteinuria level and distribution of chronic kidney disease (CKD) health states (CKD 1/2, 3, 4, and 5). At the end of 36 weeks of treatment, patients were categorized into groups based on their proteinuria level (<1g/day and ≥1 g/day) and CKD stage before transitioning to the long-term Markov model. In the long-term model, proteinuria level-based, CKD health state transition matrices derived from analysis of UK National Registry of Rare Kidney Disease (RaDaR) data, and transition matrices to dialysis and transplant derived from the US Renal Data System (USRDS) were used. Other included clinical inputs of complications and mortality were derived from extension of US Optum data analyses and literature. Health-related quality of life (HRQoL) was assessed as health state utilities and disutility associated with complications. Hypothetic scenarios of improvement in short-term proteinuria and impact on long term outcomes were tested. Model outputs included life years (LYs) and quality-adjusted life years (QALYs). Effectiveness was discounted at 3% annually. Deterministic sensitivity analysis was conducted. Results Over a lifetime, patients receiving treatment that increased the probability of achieving proteinuria <1 g/day by 10% in the short-term, were modelled to have gained an additional 0.388 LYs from reduced CKD/transplant/dialysis-related mortality and 0.633 QALYs from delaying CKD progression to kidney failure. As expected, results improved when the probability of achieving proteinuria <1 g/day increased; with a 30% increase in the probability of achieving proteinuria <1 g/day, LYs gained were 0.562 (Figure 1) and QALYs 0.968 (Figure 2). Results were most sensitive to time horizon and long-term extrapolation of CKD health state transition matrices. Conclusion Despite challenges inherent in the translation of surrogate endpoints to long-term outcomes, combining short-term proteinuria from clinical trials and long-term CKD data from real-world registry and claims-based data provides a solution. Based on modelling, treatments that increase the probability of achieving <1 g/day in the short-term, are expected to provide benefits to patients with IgAN in the long-term including less time with advanced kidney disease, less time on dialysis and avoiding the need for kidney transplant resulting in improved survival, and better quality of life. Refinement and customization of country-specific model inputs will help ensure outputs are relevant for different jurisdictions and as representative of anticipated real-world outcomes as possible.