Progressive brain tissue loss precedes the onset of clinical MS by years

The onset of MS-related progressive brain tissue loss (PBTL) precedes clinical disease onset by 5–6 years, as demonstrated in a study using an artificial intelligence (AI)-based virtual model (termed ‘Digital Twin’) that can simulate patient conditions that would otherwise be difficult to capture.

Dr Christina Azevedo (University of Southern California, CA, USA) and colleagues aimed to estimate the onset of disease-related PBTL, defined as the age at which an MS patient’s thalamic atrophy curve starts to deviate from expected normal ageing [1]. To this end, they created an AI-based ‘Digital Twin’ for every participant. A total of 4,968 3T brain MRIs were analysed: 2,483 MRIs of 520 MS patients and 2,485 MRIs of 2,053 healthy controls, with an age range of 16–89 years.

To construct the ‘Digital Twin’ model, the researchers used normal ageing data to augment longitudinal data from a well-fitted spline model; they then compared different mixed spline models and identified the model with the best fit. The final covariates in this model included age, sex, baseline thalamic volume, age at clinical onset, lifetime disease-modifying treatment exposure, and intracranial volume. Based on 10-fold cross-validation, the final model reached a repeated measure correlation coefficient of 0.88 (P<0.01).

The results showed that the onset of disease-related PBTL preceded clinical MS onset by a mean of 5.1±3.8 years, and a median of 6 (IQR 3.1–8.1) years. Just over half (55.4%) of the investigated patients could be classified as having an earlier PBTL onset, while 44.6% were classified as having simultaneous PBTL and clinical onset. The figure shows an individual example of both phenotypes.

Figure: PBTL estimation in 2 MS patients, one with earlier and one with simultaneous PBTL onset [1]



PBTL, progressive brain tissue loss.

The results showed remarkable consistency with the reported timing of pre-morbid rise in serum neurofilament light chain. Patients with earlier PBTL onset were statistically significantly more likely to:

• be women (78.5% vs 59.7%; P<0.001);
• be older at clinical symptom onset (36.9 vs 29.3 years; P<0.001);
• be older at study entry (44.5 vs 40.4 years; P<0.001);
• have lower T2-lesion volumes at study entry (3,776 vs 5,719 mm³; P<0.001);
• have larger thalamic volumes at study entry (9.4 vs 9.1; P=0.012).

“Although PBTL onset is not synonymous with the true biological disease onset, it should be closer to the biological onset than the clinical onset we currently observe,” noted Dr Azevedo. She added that a more precise estimate of disease onset may lead to a better understanding of the natural history of MS and the early events in disease pathogenesis.

Future studies will focus on validating this work and exploring its clinical meaning in relation to disease severity.

1. Azevedo C, et al. An artificial intelligence approach to estimate the onset of MS-related brain tissue loss before clinical onset. O061, MSMilan 2023, 11–13 October, Milan, Italy.

 

Copyright ©2023 Medicom Medical Publishers



Posted on 4 December, 20234 December, 2023