Predictors and Clinical Characteristics of Relapses in LGI1-Antibody Encephalitis

Abstract

Background and objectives: Relapses occur in 15%-25% of patients with leucine-rich glioma-inactivated 1 antibody (LGI1-Ab) autoimmune encephalitis and may cause additional disability. In this study, we clinically characterized the relapses and identified factors predicting their occurrence.

Methods: This is a retrospective chart review of patients with LGI1-Ab encephalitis diagnosed at our center between 2005 and 2022. Relapse was defined as worsening of previous or appearance of new symptoms after at least 3 months of clinical stabilization.

Results: Among 210 patients, 30 (14%) experienced a total of 33 relapses. The median time to first relapse was 23.9 months (range: 4.9-110.1, interquartile range [IQR]: 17.8). The CSF was inflammatory in 11/25 (44%) relapses, while LGI1-Abs were found in the serum in 16/24 (67%) and in the CSF in 12/26 (46%); brain MRI was abnormal in 16/26 (62%) relapses. Compared with the initial episode, relapses manifested less frequently with 3 or more symptoms (4/30 patients, 13% vs 28/30, 93%; p < 0.001) and had lower maximal modified Rankin scale (mRS) score (median 3, range: 2-5, IQR: 1 vs 3, range: 2-5, IQR: 0; p = 0.001). The median mRS at last follow-up after relapse (2, range: 0-4, IQR: 2) was significantly higher than after the initial episode (1, range: 0-4, IQR: 1; p = 0.005). Relapsing patients did not differ in their initial clinical and diagnostic features from 85 patients without relapse. Nevertheless, residual cognitive dysfunction after the initial episode (hazard ratio:13.8, 95% confidence interval [1.5; 129.5]; p = 0.022) and no administration of corticosteroids at the initial episode (hazard ratio: 4.8, 95% confidence interval [1.1; 21.1]; p = 0.036) were significantly associated with an increased risk of relapse.

Discussion: Relapses may occur years after the initial encephalitis episode and are usually milder but cause additional disability. Corticosteroid treatment reduces the risk of future relapses, while patients with residual cognitive dysfunction after the initial episode have an increased relapse risk.

Figure 1. Disease Course and Treatment of Patients With Relapsing LGI1-Ab Encephalitis

Swimmer plot illustrating the disease course of the 30 relapsing patients, with time on the abscissa axis extending from clinical onset (at time 0) until the last follow-up (yellow circle). Each patient is represented by a different number and line on the ordinate axis (1–30). The treatment received at the initial episode (green and orange empty rhombi) and the occurrence of a first (red triangle) and second relapse (blue triangle) are illustrated. Rhombi (green and orange) represent the treatment received for the relapse. The time point of clinical stabilization (light blue square) was established retrospectively to mark the beginning of an interval of at least 3 months during which the patient did not manifest any neurologic change. The modified Rankin scale (mRS) scores at clinical stabilization and last follow-up are depicted inside the respective indicators. LGI1-Ab = leucine-rich glioma-inactivated 1 antibody.

Figure 2. Evolution of Symptoms Before and After the Relapse

Heatmap and bar graphs illustrating the prevalence of different symptoms (memory impairment, FBDS, seizures, psychiatric/behavioral symptoms and sleep disturbances) during the initial encephalitis episode, stabilization, first and second relapses, and at last follow-up. The severity of cognitive symptoms is graded through different shades of blue, as specified in the legend. Symptoms that remained stable between stabilization and relapse are marked with a cross. ADL = activities of daily living; FBDS = faciobrachial dystonic seizures; IADL = instrumental activities of daily living; memory = memory impairment; psychiatric = psychiatric/behavioral symptoms; sleep = sleep disturbances.

Figure 3. Cox Proportional Hazards Regression Curve for Relapse Occurrence in Patients With and Without Residual Cognitive Dysfunction After the Initial Episode and in Patients With and Without Prior Administration of Corticosteroids at Initial Episode

The Cox regression model was built including variables associated with relapse risk. The abscissa axis represents follow-up time (in months) after disease onset. Relapse-free survival on the ordinate axis describes the probability of remaining relapse-free over time. Patients with residual cognitive dysfunction after the initial episode (in red, A) and without prior administration of corticosteroids at initial episode (in red, B) have a higher risk of relapse.