Beyond pain: Using Unsupervised Machine Learning to Identify Phenotypic Clusters of Small Fiber Neuropathy

Area of Science:

• Neurology
• Clinical Neuroscience
• Biomedical Research

Background:

• Small fiber neuropathy (SFN) involves loss of small nerve fibers, causing disabling sensory and autonomic symptoms.
• Current treatments primarily focus on pain and are often ineffective due to SFN's complex nature and patient variability.
• Understanding SFN's symptomatic heterogeneity is crucial for effective management.

Purpose of the Study:

• To investigate the symptomatic variability within small fiber neuropathy (SFN) patient cohorts.
• To determine if SFN patients can be sub-grouped based on distinct clinical characteristics and symptom profiles.
• To differentiate SFN phenotypes from those with mixed fiber neuropathy (MFN).

Main Methods:

• Recruited 105 SFN patients and 45 MFN patients with skin-biopsy-proven diagnoses.
• Employed unsupervised machine learning to cluster SFN patients based on symptom concurrence and severity.
• Compared demographics, clinical data, symptoms, and diagnostic findings between SFN and MFN groups, and among SFN clusters.

Main Results:

• MFN patients were more likely male, older, and had lower intraepidermal nerve fiber density compared to SFN patients.
• SFN patients were categorized into three distinct phenotypic clusters with significant differences in symptom severity, co-occurrence, and localization.
• One cluster (20%) had intense neuropathic pain, another had few, mild symptoms, and the largest cluster experienced fatigue, myalgias, and weakness.

Conclusions:

• This study introduces a data-driven method for subgrouping SFN patients based on clinical presentation.
• Identified three clusters potentially linked to different pathophysiological mechanisms, considering both pain and non-pain symptoms.
• Findings support a move towards stratified and personalized treatment strategies for small fiber neuropathy.