Statistical Models Used to Predict Friedreich’s Ataxia Progression
Models provided reasonably accurate predictions for scores on SARA, ADL
Researchers have developed statistical models to help predict the progression of Friedreich’s ataxia using data such as age at disease onset and genetic information.
Although the models are not yet accurate enough to be employed in clinical use, the researchers said this is a first step toward individualized medicine for the disease.
“The idea here is that a set of core variables allowing for prediction of the future disease course within reasonable margin of error can augment clinicians’ work for better planning of routine examinations as well as pointing out patients at risk for certain symptoms,” they wrote. “Of course, at the moment the predictive performance we report here is not good enough for clinical use and open questions remain, but we demonstrate the general feasibility of such approaches in a rare disease like [Friedreich’s ataxia].”
The study, “Prediction of the disease course in Friedreich ataxia,” was published in Scientific Reports.
Friedreich’s ataxia is a progressive disease, meaning symptoms tend to gradually worsen as time goes on. Its progression tends to occur slowly and there can be a lot of variation from person to person, which makes it difficult to accurately predict how a person’s disease is likely to change in the future.
Prediction models could help develop individualized treatments
Finding better methods for such prediction “could help in individualizing medicine for individuals with this disease and in the long-term such data could support planning the interval of examinations and also help patients and their caretakers making preparations for potential future disabilities and limitations affecting daily life,” the researchers wrote.
To develop predictive models, the researchers analyzed data from 602 people with Friedreich’s ataxia from the European Friedreich’s Ataxia Consortium for Translational Studies (EFACTS) database. The team used most of the data to construct the models then used the remaining data to test them, with statistical tests to judge their accuracy for predicting disease progression.
The researchers noted that, in the statistical models, the most important factor for predicting progression was disease duration (the length of time a person has been living with symptoms). Other data that were important for predicting progression included age, age at onset, and the number of GAA repeats in the FXN gene that cause Friedriech’s ataxia.
“To the best of our knowledge, this is the first work using a purely predictive approach in [Friedreich’s ataxia] with the greater aim of enabling individualized medicine and care in the long term,” the researchers wrote.
The models were able to provide reasonably accurate predictions for scores on the scale for rating and assessment of ataxia (SARA) and the activities of daily living (ADL), which are measures of disease severity. The models also showed fairly high accuracy for predicting whether patients would completely lose the ability to walk independently.
The models were not very accurate at predicting other outcomes, including scores on the spinocerebellar ataxiafunctional assessment (SCAFI) sub-scales — that include measures of physical function — and the presence of heart disease.
“For some measures of disease severity modeling worked reasonably well, such as for the SARA and ADL clinical scales and loss of ambulation. However, modeling performance was rather insufficient when it came to the SCAFI subtests as well as most targets on cardiac symptoms,” the researchers said, calling their work a “promising first step toward a more individualized medicine in [Friedreich’s ataxia].”
“Much more work, both on optimizing modeling for the most promising targets, as well as on the understanding of [Friedreich’s ataxia] as a whole is necessary,” they said.