(T-096) Integrating Motor Function Scores of Spinal Muscular Atrophy in a Quantitative Systems Pharmacology Model of Neurofilament

Pranami Bora, NA – Senior researcher, Fondazione The Microsoft Research - University of Trento Centre for Computational and Systems Biology (COSBI), Rovereto, Italy; Silvia Parolo, NA – Head of Systems Biology, Fondazione The Microsoft Research - University of Trento Centre for Computational and Systems Biology (COSBI), Rovereto, Italy; Michael Monine, NA – Scientific Director, Clinical Pharmacology and Pharmacometrics, Biogen, Inc., Cambridge, Massachusetts, USA; Nick van der Munnik, NA – Associate Director, Clinical Pharmacology Modeling & Simulation, GlaxoSmithKline, Waltham, Massachusetts, USA; Stephanie Fradette, NA – Vice President, Neuromuscular Development Unit, Biogen, Inc., Cambridge, Massachusetts, USA; Enrico Domenici, NA – President, Fondazione The Microsoft Research - University of Trento Centre for Computational and Systems Biology (COSBI), Rovereto, Italy; Department of Cellular, Computational and Integrative Biology (CIBIO), University of Trento, Trento, Italy; Ivan Nestorov, NA – Executive Director, Pharmacometrics Unit, Biogen, Inc., Cambridge, Massachusetts, USA; Luca Marchetti, NA – Head of Computational Biology, Fondazione The Microsoft Research - University of Trento Centre for Computational and Systems Biology (COSBI), Rovereto, Italy; Department of Cellular, Computational and Integrative Biology (CIBIO), University of Trento, Trento, Italy

Objectives: Neurofilaments (Nf) are a non-specific marker of axonal injury and neurodegeneration. In motor neuron diseases, including Spinal Muscular Atrophy (SMA), Nf levels are prognostic for disease progression and survival and reduced in response to disease modifying therapies. To deepen the understanding of the relationship between Nf and motor function in SMA, we developed computational models of motor function scores in nusinersen-treated and untreated SMA patients, integrated with simulations from a previously developed Quantitative Systems Pharmacology (QSP) model of Nf trafficking [1].

Methods: For participants with infantile-onset SMA (most likely to develop Type I) and presymptomatic SMA, natural history data [2] and data from nusinersen trials (NURTURE [3] and ENDEAR [4]) were used. This included measurements of phosphorylated Nf heavy (pNfH) levels and the Children's Hospital of Philadelphia Infant Test of Neuromuscular Disorders (CHOP-INTEND) score measurements. For patients with later-onset SMA (most likely to develop Type II), Revised Upper Limb Module (RULM) score from CHERISH [5] and SHINE [6] were incorporated. We considered both the experimental time-series and the time-series simulated by a previously developed QSP model [1] to compare patient-specific Nf levels in different treatment scenarios and the untreated case.

Results: We developed a mixed-effect model for participants with infantile-onset SMA (age < 2 years). CHOP-INTEND natural history almost always shows a decreasing trend as a function of age. Inclusion of pNfH levels into the model improves agreement with the disease progression data (without treatment). In infantile-onset SMA, nusinersen treatment often results in a reversal of the trend, leading to an improvement of motor function. To link this effect to pNfH, we identify a correlation between the reduction of pNfH levels predicted by our QSP Nf model and the improvement of the score with respect to baseline.
In the case of later-onset SMA, natural history was expressed with an age-dependent model valid for participants between 2 and 5 years of age [5-7], when the improvement of motor functions typically stops. The response of RULM to treatment, resulting in a more rapid and prolonged improvement of the score, could be directly expressed in terms of pNfH time-series, correlating the reduction of pNfH with a more prolonged period of functional improvement.

Conclusions: We present a preliminary integration of SMA motor function scores with a previously developed QSP framework of Nf trafficking. Our results indicate that it is possible to link the pNfH time-series to the evolution of patient scores in different SMA subtypes, strengthening the use of Nf as a biomarker for SMA and supporting the usefulness of the QSP Nf platform.

Citations: [1] Paris A et al. CPT Pharmacometrics Syst Pharmacol. 2023; 12:196-206.
[2] Mercuri E et al. Orphanet J Rare Dis. 2020;15(1):84. Epub 20200405
[3] Darryl C et al. Neuromuscul Disord.2019;29(11):842-856.
[4] Finkel R et al. Eur J Paediatr Neurol. 2017;21:e14-e15.
[5] Mercuri E et al. New England Journal of Medicine. 2018;378(7):625-635.
[6] Castro D et al. Neurology. 2020; 94(15):1640.
[7] Coratti et al. Muscle & Nerve. 2021; 64:552–559.