Abstracts

Rationale: Lennox–Gastaut syndrome (LGS) and Dravet syndrome (DS) are epileptic encephalopathies characterized by refractory epilepsy and varying levels of intellectual disability. Existing Clinical Outcome Assessment (COA) measures are not appropriate for conditions with heterogenous symptoms and intellectual disability, including LGS and DS, due to items that lack optimal content validity or sensitivity to change. To account for these limitations, we developed a novel clinical global impression (CGI) measure for COA of nonseizure symptoms.

Methods: A qualitative literature review identified COA measures used historically in studies of patients with LGS or DS, leading to a shortlist of nine existing measures and a conceptual model. Interviews were conducted with expert clinicians and patient advocates to assess the suitability of existing COA measures. An item-level analysis of COA data from phase 2 trials, collected with various measures, was also conducted to assess missing data and sensitivity to change. Based on the findings, items were drafted to assess nonseizure symptoms using a global impression approach (the CGI-I Nonseizure Symptoms instrument). To refine and validate this measure, an online Delphi panel was convened with 15 experts (9 epileptologists/neurologists, 6 patient advocates). The a priori consensus threshold for agreement was ≥ 70% comprising no revisions or minor revisions.

Results: The findings demonstrated that some measures may have applicability for LGS and DS but are not fit-for-purpose. Certain measures were emotionally burdensome for caregivers to complete. A number of clinicians recommended a global impression approach for assessment of within-patient change in observable domains. Item‑level analysis of caregiver responses to some existing measures found floor effects and a high level of missing data. The draft CGI-I Nonseizure Symptoms items were reviewed by experts in a three round Delphi panel to confirm item relevance and refine descriptions, reduce overlap, and limit respondent burden. Three items reached consensus (≥ 70% no revisions required): communication, alertness, and disruptive behaviors. These aligned with high priority areas identified in a previous evaluation of qualitative literature and interviews with clinicians and patient advocates.

Conclusions: Heterogeneity of symptoms and intellectual disability in LGS and DS contribute to challenges in nonseizure symptom evaluation in clinical trials. Floor effects and high levels of missing data are observed with existing measures, suggesting they may lack relevance for LGS and DS. A global impression approach was used to develop a novel measure that addresses these limitations by assessing within-patient change. Further work will evaluate psychometric evidence to support the use of the 3-item CGI-I Nonseizure Symptom measure in these patient populations.

Funding: Study funded by Takeda Pharmaceutical Company Limited.