Predicting progression from SeLECTS with SWAS to EE-SWAS: risk factor identification and model development

TL;DR

This study identifies risk factors and develops a predictive model to determine which patients with SeLECTS are likely to progress to EE-SWAS, helping clinicians intervene early.

Contribution

The study introduces a validated predictive model for early risk stratification in SeLECTS patients progressing to EE-SWAS.

Findings

Prolonged spike-and-wave clusters and high-amplitude spikes predict EE-SWAS progression.

Younger age at first seizure is an independent predictor of progression.

The nomogram model achieved high accuracy with a C-index of 0.932 in derivation and 0.934 in validation.

Abstract

This study sought to identify early risk factors and develop a predictive model for progression from self-limited epilepsy with centrotemporal spikes (SeLECTS) accompanied by spike-and-wave activation in sleep (SWAS) to epileptic encephalopathy with SWAS (EE-SWAS), aiming to facilitate early clinical intervention.

From a pediatric cohort with spike-and-wave index >50%, we analyzed 77 SeLECTS patients (33 progressed to EE-SWAS, 36 remained stable over ≥2 years of follow-up). Baseline clinical and EEG features were comprehensively evaluated. Multivariate logistic regression identified independent predictors of cognitive regression, which were incorporated into a nomogram-based predictive model. Model performance was assessed using the C-index in both derivation and external validation cohorts.

Prolonged spike-and-wave clusters, high-amplitude spikes with secondary generalization, and younger age at first seizure emerged as independent predictors of EE-SWAS progression. The nomogram model demonstrated high discriminative ability, with a C-index of 0.932 in the derivation cohort and 0.934 in external validation.

This study provides the first validated tool for early risk stratification in SWAS-associated SeLECTS, enabling clinicians to anticipate EE-SWAS progression and optimize therapeutic strategies. The model’s robustness supports its potential utility in clinical decision-making to mitigate cognitive decline.

Figures