Updating the International IgA Nephropathy Prediction Tool for use in children

Sean J. Barbour,Rosanna Coppo,Lee Er,Maria Luisa Russo,Zhi-Hong Liu,Jie Ding,Ritsuko Katafuchi,Norishige Yoshikawa,Hong Xu,Shoji Kagami,Yukio Yuzawa,Francesco Emma,Alexandra Cambier,Licia Peruzzi,Robert J. Wyatt,Daniel C. Cattran,Caihong Zeng,Biage Su,Xuhui Zhong,Koichi Nakanishi,Yihui Zhai,Maki Urushihara,Motoshi Hattori,Francesca Diomedi Camassei,Antonella Barreca,Thomas Robert,Larisa Prikhodina,Ulla Berg,Rezan Topaloglu,Malgorzata Mizerska-Wasiak,Aikaterini Papagianni,Shubha S. Bellur,Ian Roberts
DOI: https://doi.org/10.1016/j.kint.2020.10.033
IF: 19.6
2021-06-01
Kidney International
Abstract:<p>Although IgA nephropathy (IgAN) is a common cause of glomerulonephritis in children, the absence of a method to predict disease progression limits personalized risk-based treatment decisions. The adult International IgAN Prediction Tool comprises two validated Cox survival models that predict a 50% decline in estimated glomerular filtration rate (eGFR) or end stage kidney disease (ESKD) using clinical risk factors and Oxford MEST histology scores. Here, we updated the Prediction Tool for use in children using a multiethnic international cohort of 1,060 children with IgAN followed into adulthood. The updated pediatric Prediction Tool had better model fit than the original adult tool with lower Akaike Information Criterion, higher R<a class="workspace-trigger" href="#bib2"><sup>2</sup></a><sub>D</sub> and similar C-statistics. However, calibration showed very poor agreement between predicted and observed risks likely due to the observed disease trajectory in children. Therefore, the Tool was updated using a secondary outcome of a 30% reduction in eGFR or ESKD, resulting in better R<a class="workspace-trigger" href="#bib2"><sup>2</sup></a><sub>D</sub> (30.3%/22.2%) and similar C-statistics (0.74/0.68) compared to the adult tool but with good calibration. The trajectory of eGFR over time in children differed from adults being highly non-linear with an increase until 18 years old followed by a linear decline similar to that of adults. A higher predicted risk was associated with a smaller increase in eGFR followed by a more rapid decline, suggesting that children at risk of a 30% decrease in eGFR will eventually experience a larger 50% decrease in eGFR when followed into adulthood. As such, these two outcomes are analogous between pediatric and adult Prediction Tools. Thus, our pediatric Prediction Tool can accurately predict the risk of a 30% decline in eGFR or ESKD in children with IgAN.</p>
urology & nephrology
What problem does this paper attempt to address?
The problem that this paper attempts to solve is to develop an IgA nephropathy (IgAN) prediction tool suitable for children in order to accurately predict the risk of children with IgAN in the disease process. Specifically, the researchers aim to update the existing international IgAN prediction tool so that it can be used for children. The original adult IgAN prediction tool predicts the time before a 50% estimated glomerular filtration rate (eGFR) decline or the occurrence of end - stage kidney disease (ESKD) through clinical risk factors and Oxford MEST histological scores. However, since there are significant differences in the disease progression trajectories between children and adults, especially the eGFR change trajectory in children is nonlinear, it is necessary to adjust the existing tool to more accurately predict the risk of disease progression in children. The main objectives of the study include: 1. **Update the prediction model**: Recalibrate and select the coefficients in the prediction model according to the characteristics of children with IgAN to improve the prediction performance of the model. 2. **Improve the prediction indicators**: Since the eGFR change trajectory in children is nonlinear and the events of 50% eGFR decline are fewer, the study introduced a new secondary outcome indicator - 30% eGFR decline or ESKD, to increase the applicability and accuracy of the prediction model. 3. **Evaluate the model performance**: Evaluate the performance of the updated prediction tool in children through indicators such as model fit (such as \(R^2_D\)), discrimination ability (such as C - statistic), and calibration (i.e., the consistency between predicted risk and actual risk). Through these efforts, the researchers hope to provide a reliable tool to help doctors more accurately assess the risk of disease progression in patients at the early diagnosis of IgAN in children, thereby formulating more personalized treatment plans.