Researchers at the Institute of Pathology at Uniklinik RWTH Aachen, led by Professor Peter Boor, MD, and Assistant Professor Roman Bülow, MD, have devised a model for individualised treatment recommendations for IgA nephropathy in a retrospective study. The paper, entitled “Individualised treatment effects of corticosteroids in IgA nephropathy”, was published in July 2026 in the renowned journal eBioMedicine and provides a methodological framework for the further development of individualised treatment recommendations for various clinical conditions.
IgA nephropathy, formerly known as Berger’s disease, is the most common chronic disease of the glomeruli and can, in the long term, lead to a progressive decline in kidney function, ultimately resulting in kidney failure. The use of systemic corticosteroids to treat patients with this condition remains controversial, as the evidence from clinical trials regarding their therapeutic benefit is contradictory. This suggests that previous population-level studies may not have adequately reflected the individual circumstances of patients.
AI-assisted analysis identifies differences in response to treatment
The exploratory study, conducted largely by Dr David Laurin Hölscher in the Nephropathology Section of the Institute of Pathology, focused on the question of whether certain groups of patients with IgA nephropathy (IgAN) benefit from treatment with systemic corticosteroids and which characteristics can be used to identify these patients. For the analysis, the researchers combined clinical data with AI-assisted high-throughput analyses of digital tissue sections, known as pathomics data. Using causal machine learning methods, they investigated which patient groups might benefit from corticosteroid therapy. Across the entire study population, there was no measurable effect of corticosteroids on disease progression. However, the analyses revealed marked heterogeneity in individual response to treatment.
A basis for more precise treatment decisions
Based on these findings, the team developed a model for individualised treatment recommendations. In a retrospective analysis, its application could have reduced the use of systemic corticosteroids by around 60 per cent, whilst targeting treatment specifically at those patients likely to benefit from it. Furthermore, the researchers identified a histological pattern resembling interstitial nephritis as an important indicator of a potential response to corticosteroids. This association could only be identified through the systematic analysis of the Pathomics data.
Assistant Professor Dr Roman Bülow emphasises that the model developed cannot currently be used for clinical decision-making: “We see our study rather as a blueprint for the development of personalised treatment recommendations – a methodological framework that can also be applied to other biomarkers and clinical conditions in the future.”
The study provides compelling evidence of the potential offered by the combination of clinical data, AI and digital pathology.
Further information can be found in the complete publication.
