Artificial intelligence reveals some surprises

Analysis of FSHD registry data overturns prior ideas about genetic and gender factors

by Natalie Katz, MD, University of Rochester

Facioscapulohumeral muscular dystrophy is a highly variable disease. While we have a good understanding of the underlying genetic mechanism that causes it, we have yet to understand why there is so much variability in the symptoms that patients experience. Our new study used a unique approach called machine learning, or artificial intelligence, to analyze patient-reported data collected by the FSHD US national registry. Our aim is to better understand what factors might influence the rate at which patients reach disease-related milestones, such as the need to use a wheelchair for mobility.

This study analyzed data collected over a span of almost two decades from 578 individuals with genetically confirmed FSHD type 1. Machine learning analysis showed that FSHD duration (the time from diagnosis to the most recent survey) was the factor most predictive of the need for wheelchair use, with longer disease duration associated with a higher probability of wheelchair use.

Surprisingly, overall medication use and other existing medical conditions were also found to influence the need to use a wheelchair. These traits were predicted to have a stronger influence on progression to wheelchair use than the more commonly analyzed factors of gender and D4Z4 repeat number. FSHD is caused by a loss of repeated DNA units called D4Z4 near the tip of chromosome 4. Normally, there are 11 or more units, but in FSHD, people have fewer units. Additionally, we found that women were more likely than men to progress to wheelchair use, independent of repeat number.

The results of this study raised several important considerations when thinking about FSHD progression and variability, and how these may impact patient care and clinical trial design. Traditionally, studies have analyzed individuals based on their D4Z4 repeat number, but it may be that disease duration has a stronger influence on functional performance during a clinical trial.

Overall general health of the individual and medication use may also influence disease progression and severity, which may have implications for how doctors clinically manage individuals with FSHD. Our finding that women are more likely to progress to wheelchair use compared to men challenges the notion that men with FSHD are more severely affected than women.

Ultimately, more studies will be needed to better discern how these risk factors influence the progression of weakness in FSHD. These findings suggest that we may need to shift our thinking about what factors influence FSHD progression and how meaningful outcomes are measured in clinical trials to show improvement.

As for what these findings mean for patients, it is hard to make any claims at this early stage of research. It is not known whether there are any causeand- effect relationships among the various factors. What’s exciting is that this type of analysis opens new doors to help us better understand FSHD.

Reference
Predictors of functional outcomes in patients with facioscapulohumeral muscular dystrophy. Katz NK, Hogan J, Delbango R, Cernik C, Tawil R, Statland JM. Brain. 2021 Sep 20:awab326. View abstract in PubMed.