Daria Zhuravleva 1, Alzahra Hamdan 1, Matthis Synofzik 2,3, Andrew C. Hooker 1, Mats O. Karlsson 1
1 Uppsala University (Uppsala, Sweden), 2 Department of Neurodegenerative Diseases, Center for Neurology and Hertie Institute for Clinical Brain Research, University of Tübingen (Tübingen, Germany), 3 German Center for Neurodegenerative Diseases (DZNE) (Tübingen, Germany)
Background
The rapid development of personalized disease-modifying therapies, such as antisense oligonucleotides, has outpaced the evolution of the clinical trials used to evaluate them. In the European Union, where a disease is rare if it affects fewer than 5 in 10,000 people [1], the challenge of small sample sizes is acute. For neurodegenerative conditions like Autosomal Recessive Cerebellar Ataxias (ARCA), traditional multi-period crossover designs are frequently precluded by the disease-modifying, i.e., long-lasting effects of genetic treatments and the slow, irreversible nature of disease progression [2]. Consequently, there is an urgent need for robust statistical frameworks capable of capturing drug effects of disease-modifying treatments at the single-subject level. This study investigates the continuous-categorical item response theory (ccIRT) approach [3] as a framework to enhance the sensitivity of n-of-1 trials for disease-modifying treatments by integrating clinician-reported outcomes (ClinROs) with biomarkers (BMs).
Methods
The study was conducted through a series of stochastic simulation estimation (SSE) experiments using the Scale for the Assessment and Rating of Ataxia (SARA) IRT model as the example [4]. The experiments were structured into three primary investigative stages:
Exploratory study of biomarkers. A validation of the decrease in uncertainty metric, introduced in [3], was performed to quantify the precision gains in latent variable estimates upon integrating BMs with ClinROs. Artificial biomarkers (BM1–BM6) were constructed with varying residual error variances, ranging from 0 (BM1) to 1 (BM6), to represent a range of informativeness. These evaluations assessed the robustness of biomarker-specific IRT parameters estimation across various study durations, sample sizes, and longitudinal disease progression models.
Natural history of a single subject. The impact of observation duration (ranging from 1 to 10 years) and biomarker sampling density on the precision of individual disease progression parameters was evaluated. This stage focused on empirical Bayes estimates (EBEs) and their respective standard errors of baseline severity and the progression slope to establish a robust control for the subsequent interventional phase in a delayed-onset trial.
Evaluation of full n-of-1 clinical trials. A full n-of-1 [A|B] study design was evaluated, utilizing Phase A as the natural history control and Phase B as the active intervention period. The primary focus was on the statistical power to detect a drug-induced change in the progression slope across varying drug effect magnitudes, 0% (no change), 100% (complete inhibition of progression), and 150% (reverse of the progression), using a two-sided log-likelihood ratio test.
Model simulation and estimation were implemented using NONMEM. R was utilized for subsequent statistical analysis and the generation of graphical outputs.
Results
The decrease in the uncertainty metric proved to be a highly sensitive tool for biomarker informativeness ranking, demonstrating that the inclusion of biomarkers can significantly reduce the uncertainty of latent variable estimates, which ranges from 17% to 90% depending on the variance of the residual error of the biomarker model. At the same time, non-informative biomarkers had a negligible effect on the uncertainty of latent variable estimates (median decrease in uncertainty of 0.009%). The analysis of individual natural history studies showed that the precision of EBEs of baseline severity and progression slope aligns with the anticipated trend – extending the observational period with an enriched biomarker sampling schedule had a noticeable impact on root mean squared error. The ccIRT modelling approach was demonstrated as a framework for analyzing n-of-1 studies. Specifically, achieving an 80% power threshold is feasible within a five-year study (2 years for Phase A, and 3 years for Phase B) with frequent biomarker sampling (every 3 months or every 1 month).
Conclusion
This work demonstrates that the ccIRT approach effectively mitigates the limitations of small sample sizes by maximizing the informational yield of each subject. By combining clinical outcomes with biomarkers, this framework provides support for the transition towards personalized, disease-modifying interventions in rare disease research.
References:
[1] EURORDIS rare disease Europe: What is a rare disease? https://www.eurordis.org/information-support/what-is-a-rare-disease/ Accessed February 8, 2026.
[2] Kim-McManus O., Gleeson J.G., Mignon L. et al A framework for N-of-1 trials of individualized gene-targeted therapies for genetic diseases. Nature Communications. 2024 https://doi.org/10.1038/s41467-024-54077-5
[3] Hamdan A., Traschütz A., Beichert L., et al Integrated Modeling of Digital-Motor and Clinician-Reported Outcomes Using Item Response Theory: Towards Powerful Trials for Rare Neurological Diseases. CPT Pharmacometrics Systems Pharmacology. 2025; 14: 1857–1868 https://doi.org/10.1002/psp4.70081
[4] Hamdan A., Hendrickx N., Hooker A.C. et al Longitudinal analysis of natural history progression of rare and ultra-rare cerebellar ataxias using item response theory. Clinical Pharmacology & Therapeutics. 2024; 116(6): 1593–1605. https://doi.org/10.1002/cpt.3466
This work was funded through the project MMM (Medicine Made to Measure), an Innovative Training Network funded by the European Union under the Horizon Europe Marie Skłodowska-Curie Actions scheme, grant agreement no. 101120256; as well as by the European Union, project European Rare Disease Research Alliance (ERDERA), GA no. 101156595, funded under call HORIZON-HLTH-2023-DISEASE-07 (to M.S. and M.K.)
Reference: PAGE 34 (2026) Abstr 11961 [www.page-meeting.org/?abstract=11961]
Poster: Methodology - Study Design