Autoimmune Limbic Encephalitis (ALE) is a rare neurological syndrome (prevalence is 13.7/100,000 people, while the incidence is 0.8/100,000 per year) characterized by an immune mediated inflammation of the brain and frequently associated with seizures, behavioral changes, memory and cognitive deficits (Kelley et al., 2017; Hermetter et al., 2018). It was initially identified as a paraneoplastic phenomenon, even if in recent years, a non-paraneoplastic variant has been characterized and several antibodies (Abs) have been described. Depending on the localization of the target antigen Abs, ALE can be divided into limbic encephalitis (LE) associated with Abs against intracellular neural antigen (INA-Abs) and LE associated with Abs against cell surface (CSA-Abs). The neuronal CAS-Abs most frequently associated with LE are leucine-rich glioma-inactivated 1 (LGI1), contactin-associated protein-like 2 (CASPR2), gamma-aminobutyric acid type B receptor (GABA-B receptor) and alpha-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptor (AMPA receptor). By contrast, the INA-Abs most frequently occurring with LE are anti-Hu (ANNA-1), anti-Ri (ANNA-2), anti-Ma/Ta (Ma2, also known as PNMA2), anti-CV2/CRMP5 (CRMP5) and glutamic acid decarboxylase (GAD) (Tüzün & Dalmau, 2007; Graus et al., 2016). Magnetic Resonance Imaging (MRI) findings often reveal hyperintense signals of the medial temporal lobes in T2-weighted or Fluid-Attenuated Inversion Recovery (FLAIR) images and the Electroencephalogram (EEG) often shows focal or generalized slow wave or epileptiform activity in the temporal lobe and in the striatum, diencephalon or rhombencephalon (Dubey et al., 2018). According to the clinical approach to diagnose ALE proposed by Graus and colleagues, cognitive difficulties and behavioral changes are the hallmark of the disorder and represent important biomarkers of disease acuity, progress and therapy. Anterograde amnesia, characterized by a severe autobiographical and episodic memory impairment, is the most frequently reported deficit (Loanen et al., 2019). However, there is growing evidence that cognitive changes in ALE may not be restricted to memory functions. Deficits in attention and executive functions including set-shifting, working memory and fluency may occur (Dodich et al., 2016; Wang et al., 2017; Witt & Helmstaedter, 2021; Gibson et al., 2020). Despite the prominence of cognitive dysfunctions in individuals with ALE, very few studies systematically investigate a wide range of cognitive domains. Moreover, some authors suggest that ALE associated Abs types exhibit distinct cognitive impairments. However, the degree to which different Abs types result in distinct neuropsychological (NPS) manifestations remains largely unknown and in literature the data are discordant. In their study, Frisch et colleagues found a distinct relationship between inductive processes and cognitive outcome in LGI1 and CASPR 2-Abs positive (CAS-Abs) and GAD-Abs-positive (INA-Abs) subforms of LE (Frisch et al., 2013). In addition, Muller in a systematic review shows that ALE patients with Abs against both LGI1 and CASPR2 (CAS-Abs) show higher percentages of NPS deficits compared to ALE patients with Abs against GAD (IN-Abs) (Mueller et al., 2023). Finally, Vrillon in a systematic review of 21 cases with anti-GAD confirms that anti-GAD LE appears to be associated with distinctive clinical features compared with other types of ALE (Vrillon et al., 2020). Conversely, this distinction is challenged by equally compelling evidence reporting no differences across Abs groups (Mueller et al., 2021; Mueller et al., 2022). ALE is treated by autoimmune suppression and better prognosis is associated with early treatment and with ALE associated to CSA-Abs (Hermetter et al., 2018; Dutra et al., 2018). However, the knowledge of cognitive phenotype over the course of the disease is limited and it has not been addressed in large studies. The significant variability in the methodology makes it challenging to derive solid conclusions. Isolated case reports and small series indicate that the clinical outcome is variable, as some reported residual memory deficits and others an apparent recovery (Titulaer et al., 2013., 2013; Van Sonderen et al 2016; Abboud et al., 2021). Against this background, the present thesis had the overarching objective of clarifying the nature, severity and trajectory of cognitive deficits in ALE through three complementary studies (Study 2-4), built upon a single shared data infrastructure (Study 1). The experimental design relied on a unified database developed to transparently integrate three studies with partially overlapping cohorts, (1) a cross-sectional comparison among ALE, non autoimmune temporal lobe epilepsy (NAI-TLE) and healthy controls (HCs), (2) an analysis focused exclusively on ALE patients stratified by serostatus and Abs subtypes and (3) a longitudinal assessment with 12 month follow up in an ALE subgroup. Recruitment took place at Fondazione IRCCS Istituto Neurologico Carlo Besta (Milan, Italy). Given the rarity of the condition and the lack of established effect size estimates, convenience sampling was adopted. For ALE, enrollment was conducted both retrospectively and prospectively and two time points were planned (baseline, T0; follow up, T1, the latter acquired only prospectively). General eligibility criteria were: age ≥18 years, at least 5 years of education, average abstract–logical reasoning on Raven’s Colored Progressive Matrices (RCPM >17.5), no diagnosis or ongoing evaluation for neurodegenerative disorders, no current major psychiatric episodes or severe systemic diseases, no substance abuse, ability to provide informed consent and no prior neurosurgery. Included ALE cases met Graus et al. (2016) criteria (possible autoimmune encephalitis [AE], definite ALE, or probable Abs negative encephalitis) and were assessed in the post acute phase. The NAI- TLE group met International League Against Epilepsy (ILAE) criteria for temporal lobe epilepsy (Scheffer et al., 2017), supported by semiology, EEG and MRI findings. HCs were recruited among hospital staff and visitors, without compensation, and matched for age (±5 years), education (±3 years) and sex. To establish a robust empirical foundation, the first study focused on curating and consolidating the patient database. ALE cohort construction began with a retrospective phase using a pre existing institutional database of 122 patients (since 2010) with “encephalitis” listed as diagnosis or diagnostic hypothesis. Each chart underwent joint manual review by the doctoral candidate and a neurologist with expertise in epilepsy to verify diagnostic accuracy, data completeness and T0 availability. A total of 55/122 were excluded due to diagnoses other than ALE (e.g., viral encephalitis, neurodegenerative disorders, glioma, stroke) or major psychiatric disorders; applying Graus et al. (2016) criteria, a further 36/67 were excluded, yielding a final cohort of 31 ALE patients (anti GAD=6; anti LGI1=6; anti CASPR2=5; seronegative=14). Since October 2022, prospective recruitment continued at the Epilepsy Unit. The final cohort reached 57 patients in the ALE group (31 identified retrospectively and 26 prospectively), alongside 36 patients with NAI-TLE and 58 HCs. For the patients, neurologists flagged potential candidates during routine clinical activity, followed by pre screening for compatibility with ALE or NAI-TLE and a structured consent process in accordance with the Declaration of Helsinki and Ethics Committee approval. All participants underwent comprehensive neurological assessment and a standardized NPS battery assessing nonverbal reasoning, language, attention, executive functions including set-shifting, switching and processing speed under executive load, working memory, verbal and visual memory (short- and long-term) and visuoconstructive abilities. The battery included: Raven Colored Progressive Matrices (CPM; Basso et al., 1987); Boston Naming Test (BNT; Goodglass et al., 1983); Attentive Matrices (AT; Spinnler & Tognoni, 1987); Trail Making Test A (TMT-A) and Trail Making Test B (TMT-B; Giovagnoli et al., 2016), with the derived index Trail Making Test B-A (TMT B-A); Word Fluency on Phonemic Cues (WF-P) and Word Fluency on Semantic Cues (WF-S; Novelli et al., 1986); Rey–Osterrieth Complex Figure – Copy (ROCF-Copy) and Rey–Osterrieth Complex Figure – Delayed Recall (ROCF-DR; Caffarra et al., 2002); Street’s Completation Test (SCT); Digit Span Forward (DS-F) and Digit Span Backward (DS-B); Corsi Block-Tapping Test (CBT; forward/backward); Short Story Recall (SSR; Novelli et al., 1986); Rey Auditory Verbal Learning – Immediate Recall (RAVL-IR) and Rey Auditory Verbal Learning – Delayed Recall (RAVL-DR; Carlesimo et al., 1996); and Corsi Block-Supra-Span (CBSST; Spinnler & Tognoni, 1987). ALE and NAI-TLE groups also underwent at least one EEG, an 18F Fluorodeoxyglucose Positron Emission Tomography (18F FDG PET) and MRI on clinical scanners (1.5 T Siemens Avanto Fit; 1.5 T Philips Achieva; 3 T Philips Achieva dStream). Serological testing for neuronal Abs was performed according to institutional procedures. HCs were enrolled with a similar information process and matched for age, education and sex. Leveraging the unified data infrastructure, the second study aimed to comprehensively characterize the cognitive profile in the post-acute phase of ALE using an observational cohort design, comparing patients with ALE with individuals NAI-TLE and with HCs. HCs provided a normative reference, whereas NAI-TLE was included as a seizure-bearing clinical control group and as a temporo-limbic benchmark, thereby enabling interpretation of cognitive impairment in ALE against a condition predominantly involving the temporo-limbic network. This comparative framework was intended to disentangle, insofar as feasible, cognitive sequelae plausibly attributable to seizures/temporal lobe epilepsy from deficits more consistent with an autoimmune etiology, with the specific objective of identifying clinically meaningful differentiating features that may facilitate recognition of ALE in seizure-dominant presentations, without implying equivalence between the two conditions. This single-center observational cohort study included ALE (n=57), NAI-TLE (n=36) and HCs (n=58). Between-group differences in cognitive performance were assessed with Kruskal–Wallis tests followed by Bonferroni-adjusted comparisons. Frequencies of cognitive deficits were summarized and multiple regression models were fitted with adjustment for disease duration and MRI lesion lateralization. Compared with HCs, ALE and NAI-TLE exhibited greater impairments in semantic fluency and in verbal and visuospatial memory, with smaller effects in attention and executive function. No significant mean differences emerged between ALE and NAI-TLE. However, deficit-frequency analyses suggested a dissociation: a higher proportion of ALE were impaired in long-term visuospatial memory, whereas more NAI- TLE showed difficulties in verbal learning. Notably, in regression models, the group factor (ALE versus NAI-TLE) did not predict performance after adjustment for disease duration and MRI lesion lateralization. Overall, the observed memory deficit pattern aligns with established features of ALE, characterized by predominant mesial temporal involvement and preferential autoimmune targeting of the hippocampus and neighboring limbic structures (Kelly et al., 2017; Budharam et al., 2019; Sanvito et al., 2024), which form the core neuroanatomical substrate of memory (Scoville & Milner, 1957; Squire & Zola-Morgan, 1991). Taken together with the relative preservation of executive functions in our cohort, these findings support a staged model of ALE whereby an initial phase of widespread inflammation affecting large-scale networks, including fronto-temporal circuits is followed by a chronic/residual phase characterized by persistent mesial temporal damage. Moreover, the lack of significant differences between clinical groups, even after controlling for confounders, suggests that in the post acute phase cognitive profiles are primarily determined by lesion topography rather than autoimmune etiology per se. The third study systematically delineated cognitive differences in ALE as a function of serostatus and Abs subtypes. The cohort was split into seropositive and seronegative groups and, within seropositives, stratified into INA Abs and CSA Abs subtypes. Analyses employed non-parametric tests and Quade’s non-parametric ANCOVA for subgroup comparisons, adjusting for disease duration. In a preliminary analysis, seronegative patients outperformed seropositives on semantic verbal fluency and executive measures (including set-shifting, switching and processing speed under executive load), with a trend toward better perceptual organization. However, subsequent analyses revealed that the pattern was driven by poorer performance in the INA-Abs group versus seronegatives, with no differences between INA-Abs and CSA-Abs. Taken together, these results suggest that cognitive impairment in ALE is more closely linked to the Abs profile than to serostatus alone. Specifically, INA-Abs forms are associated with more pronounced and persistent executive dysfunction, plausibly reflecting broader disruption of fronto-temporal networks. The persistence of these differences after adjustment for disease duration indicates that duration does not fully account for executive dysfunction in INA-Abs, prompting a reconsideration of the staged hypothesis of executive recovery proposed in Study 2 and supporting a role for Abs in shaping disease trajectories. In this context, the presence of INA-Abs appears to play a pivotal role, marking a qualitatively distinct and more aggressive pathological process that produces a more enduring disruption of neural circuits. The fourth study aimed to quantify within subject change between T0 and T1, identify clinical and radiological predictors of cognition at T1 and test whether trajectories differed by serostatus (seropositive versus seronegative) and Abs subtypes (INA Abs versus CSA Abs). A single-center longitudinal follow-up was conducted approximately one year after T0, excluding patients who developed new neurological disorders. The sample comprised 28 prospectively recruited patients, most of whom received pharmacological treatment, including IT as clinically indicated. The NPS battery was re-administered using parallel forms to mitigate practice effects. T0–T1 changes were evaluated with two-tailed Wilcoxon signed-rank tests and significant outcomes entered LASSO regression models with covariates including baseline performance, IT, T0–T1 interval and MRI findings, epileptic and psychiatric outcomes. Trajectories across groups were estimated with linear mixed-effects models (LMMs) including fixed effects of Time, Group and Time×Group interaction, a random intercept for participant and Satterthwaite degrees of freedom. Secondary analyses used a three level Group factor (INA Abs, CSA Abs and Seronegative). Selective improvement was observed in long term visuospatial memory and visuospatial working memory. Baseline performance and improvements on MRI emerged as robust predictors of visuospatial memory gains at T1, while working-memory improvement was associated earlier treatment initiation, improved seizure control and continuation of IT, in addition to baseline performance and serological status. No significant Time×Group interactions emerged in the LMMs for any outcome. Main effects of Time confirmed overall gains in working memory and visuospatial memory, while main effects of Group revealed level differences whereby seronegatives outperformed seropositives in visuospatial learning and verbal episodic memory and CSA Abs outperformed INA Abs in processing speed/visual search. Taken together, the results indicate selective yet consistent recovery in specific domains, plausibly supported by the widespread adoption of IT and improved control of epileptogenic/inflammatory activity. However, level differences between seronegatives and seropositives did not translate, within the observed window, into divergent recovery slopes, despite the hypothesis of greater damage in INA Abs phenotypes, likely due to the relatively short follow up and limited sample size. In summary, across these studies, ALE is associated with substantial cognitive impairment, characterized by core deficits in memory systems dependent on temporal circuitry and a more pronounced executive vulnerability in INA-Abs phenotypes. In the post-acute phase, the absence of differences relative to NAI-TLE suggests that cognitive deficits are driven predominantly by lesion topography rather than autoimmune etiology, supporting a model in which the location and extent of damage are the primary determinants of NPS outcomes. Longitudinally, selective and clinically meaningful improvements are observed in long-term visuospatial memory and visuospatial working memory. These gains were predicted by stronger baseline performance and earlier treatment initiation, as well as improvements in MRI findings and seizure control. Finally, recovery trajectories are broadly parallel across groups, albeit with a level offset favoring seronegative and CSA-Abs subgroups over INA-Abs, suggesting that Ab etiology contributes to initial severity and vulnerability profiles. This multidomain characterization of the NPS profile provides clinicians with a concrete tool for diagnostic framing, therapeutic planning and longitudinal monitoring, facilitating the development of more effective guidelines for cognitive rehabilitation. Stratification by Abs profile, rather than serostatus alone, supports more targeted therapeutic choices and early implementation of individualized non pharmacological interventions. Because cognitive sequelae broadly affect daily functioning (return to work, activities of daily living and interpersonal relationships), timely, targeted rehabilitation can meaningfully improve quality of life and social reintegration. The use of a unified, quality controlled data infrastructure integrating retrospective and prospective data, multimodal biomarkers and standardized NPS batteries also provides a foundation for future multicenter collaborations in rare diseases, accelerating evidence generation and clinical translatability. Despite this, some limitations remain: convenience sampling and the single center design may limit generalizability; the rarity of ALE constrains subgroup sizes, especially for Abs subtype analyses; between group differences in certain clinical variables (e.g., disease duration, MRI lesion lateralization), although analytically managed, may leave residual confounding; an approximately 12 month follow up may be insufficient to capture later divergence in recovery trajectories, particularly in INA Abs phenotypes. Additionally, despite the use of parallel forms, practice effects cannot be entirely excluded. Finally, some additional domains, particularly social cognition, were not included in the assessment because expanding the test battery would have entailed an excessive assessment burden (i.e., longer administration time and increased participant fatigue), potentially undermining data reliability and overall quality. In this study, we confined the assessment to core cognitive domains (attention, memory, language) and key executive components, as the primary objective was to obtain a standardized and comparable profile of basic cognitive functioning. More specialized and complex domains, such as social cognition, were therefore not evaluated, as they were considered ancillary to the study aims. Moreover, due to logistical and coordination constraints within the collaboration, social cognition data collected within a parallel research line could not be integrated into the analyses of the present study and may be explored in future work using more fine-grained, dedicated measures. In conclusion, this thesis proposes a methodologically coherent framework for characterizing the cognitive phenotype of ALE both cross sectionally and longitudinally, clarifying the relative roles of Abs profile. The findings provide immediately translatable elements for clinical practice diagnosis, therapeutic planning and cognitive rehabilitation finally lay the groundwork for multicenter studies and longer follow ups aimed at refining Abs specific recovery models and optimizing truly personalized interventions.
L’encefalite limbica autoimmune (Autoimmune Limbic Encephalitis, ALE) è una rara sindrome neurologica (la prevalenza è di 13,7/100.000 persone, mentre l’incidenza è di 0,8/100.000 all’anno) caratterizzata da un’infiammazione immuno-mediata del cervello e frequentemente associata a crisi epilettiche, alterazioni del comportamento e deficit di memoria e cognitivi (Kelley et al., 2017; Hermetter et al., 2018). Inizialmente è stata identificata come fenomeno paraneoplastico, sebbene negli ultimi anni sia stata descritta anche una variante non paraneoplastica e siano stati individuati diversi anticorpi (Abs). In base alla localizzazione dell’antigene bersaglio degli Abs, l’ALE può essere suddivisa in encefalite limbica (LE) associata ad Abs contro antigeni neuronali intracellulari (INA-Abs) e LE associata ad Abs contro antigeni di superficie cellulare (CSA-Abs). Gli anticorpi neuronali CSA-Abs più frequentemente associati alla LE sono quelli contro leucine-rich glioma-inactivated 1 (LGI1), contactin-associated protein-like 2 (CASPR2), recettore GABA-B (gamma-aminobutyric acid type B receptor) e recettore AMPA (alpha-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptor). Al contrario, gli INA-Abs più frequentemente riscontrati nella LE sono anti-Hu (ANNA-1), anti-Ri (ANNA-2), anti-Ma/Ta (Ma2, noto anche come PNMA2), anti-CV2/CRMP5 (CRMP5) e glutammato decarbossilasi (GAD) (Tüzün & Dalmau, 2007; Graus et al., 2016). I reperti alla risonanza magnetica (MRI) spesso rivelano segnali iperintensi dei lobi temporali mediali nelle immagini T2 o FLAIR (Fluid-Attenuated Inversion Recovery) e l’elettroencefalogramma (EEG) mostra frequentemente attività a onde lente focali o generalizzate oppure attività epilettiforme nel lobo temporale e nello striato, diencefalo o rombencefalo (Dubey et al., 2018). Secondo l’approccio clinico per la diagnosi di ALE proposto da Graus e colleghi, le difficoltà cognitive e le alterazioni comportamentali costituiscono il segno distintivo del disturbo e rappresentano importanti biomarcatori di acuzie di malattia, progressione e terapia. L’amnesia anterograda, caratterizzata da una grave compromissione della memoria autobiografica ed episodica, è il deficit più frequentemente riportato (Loanen et al., 2019). Tuttavia, vi sono crescenti evidenze che i cambiamenti cognitivi nell’ALE non siano limitati alle funzioni mnestiche. Possono infatti verificarsi deficit di attenzione e di funzioni esecutive, inclusi flessibilità cognitiva (set-shifting), memoria di lavoro e fluenza (Dodich et al., 2016; Wang et al., 2017; Witt & Helmstaedter, 2021; Gibson et al., 2020). Nonostante la rilevanza delle disfunzioni cognitive nelle persone con ALE, pochissimi studi hanno indagato in modo sistematico un ampio spettro di domini cognitivi. Inoltre, alcuni autori suggeriscono che i tipi di Abs associati all’ALE presentino compromissioni cognitive distinte. Tuttavia, il grado in cui diversi tipi di Abs determinino manifestazioni neuropsicologiche (NPS) differenti rimane in gran parte sconosciuto e i dati in letteratura sono discordanti. Nel loro studio, Frisch e colleghi hanno individuato una relazione specifica tra processi induttivi ed esito cognitivo nelle forme di LE positive per LGI1 e CASPR2 (CAS-Abs) e in quelle positive per GAD (INA-Abs) (Frisch et al., 2013). Inoltre, Muller, in una revisione sistematica, mostra che i pazienti con ALE e Abs contro LGI1 e CASPR2 (CAS-Abs) presentano percentuali più elevate di deficit NPS rispetto ai pazienti con ALE e Abs contro GAD (IN-Abs) (Mueller et al., 2023). Infine, Vrillon, in una revisione sistematica di 21 casi con anti-GAD, conferma che la LE anti-GAD sembra associarsi a caratteristiche cliniche distintive rispetto ad altri tipi di ALE (Vrillon et al., 2020). Al contrario, tale distinzione è messa in discussione da evidenze altrettanto convincenti che riportano assenza di differenze tra i gruppi di Abs (Mueller et al., 2021; Mueller et al., 2022). L’ALE viene trattata con immunosoppressione e una prognosi migliore è associata a un trattamento precoce e alle forme di ALE legate a CSA-Abs (Hermetter et al., 2018; Dutra et al., 2018). Tuttavia, le conoscenze sul fenotipo cognitivo nel corso della malattia sono limitate e non sono state affrontate in studi di grandi dimensioni. La notevole variabilità metodologica rende difficile trarre conclusioni solide. Report di casi isolati e piccole serie indicano un esito clinico variabile: alcuni riportano deficit di memoria residui, altri un’apparente ripresa (Titulaer et al., 2013., 2013; Van Sonderen et al 2016; Abboud et al., 2021). Su queste premesse, la presente tesi ha avuto come obiettivo generale quello di chiarire natura, gravità e traiettoria dei deficit cognitivi nell’ALE attraverso tre studi complementari (Studi 2–4), costruiti su una singola infrastruttura dati condivisa (Studio 1). Il disegno sperimentale si è basato su un database unificato, sviluppato per integrare in modo trasparente tre studi con coorti parzialmente sovrapposte: (1) un confronto trasversale tra ALE, epilessia del lobo temporale non autoimmune (NAI-TLE) e controlli sani (HCs), (2) un’analisi focalizzata esclusivamente sui pazienti ALE stratificati per sierostato e sottotipi di Abs e (3) una valutazione longitudinale con follow-up a 12 mesi in un sottogruppo di pazienti ALE. La struttura in quattro studi del programma di ricerca è riassunta in Figura 1. Il reclutamento è avvenuto presso la Fondazione IRCCS Istituto Neurologico Carlo Besta (Milano, Italia). Data la rarità della condizione e la mancanza di stime consolidate di dimensione dell’effetto, è stato adottato un campionamento di convenienza. Per l’ALE, l’arruolamento è stato condotto sia retrospettivamente sia prospetticamente e sono stati pianificati due time-point (baseline, T0; follow-up, T1, quest’ultimo acquisito solo prospetticamente). I criteri generali di eleggibilità includevano: età ≥18 anni, almeno 5 anni di scolarità, ragionamento astratto-logico nella norma alle Matrici Progressive Colorate di Raven (RCPM >17,5), assenza di diagnosi o valutazione in corso per disturbi neurodegenerativi, assenza di episodi psichiatrici maggiori in atto o gravi patologie sistemiche, assenza di abuso di sostanze, capacità di fornire consenso informato e nessun precedente intervento neurochirurgico. I casi ALE inclusi soddisfacevano i criteri di Graus et al. (2016) (encefalite autoimmune [AE] possibile, ALE definita o encefalite probabilmente Abs-negativa) ed erano valutati nella fase post-acuta. Il gruppo NAI-TLE soddisfaceva i criteri della International League Against Epilepsy (ILAE) per l’epilessia del lobo temporale (Scheffer et al., 2017), supportati da semiologia, EEG e MRI. I controlli sani sono stati reclutati tra personale ospedaliero e visitatori, senza compenso, e abbinati per età (±5 anni), istruzione (±3 anni) e sesso. Per stabilire una solida base empirica, il primo studio si è concentrato sulla cura e consolidamento del database dei pazienti. La costruzione della coorte ALE è iniziata con una fase retrospettiva utilizzando un database istituzionale preesistente di 122 pazienti (dal 2010) con “encefalite” indicata come diagnosi o ipotesi diagnostica. Ogni cartella clinica è stata sottoposta a revisione manuale congiunta da parte della dottoranda e di un neurologo esperto in epilessia per verificare accuratezza diagnostica, completezza dei dati e disponibilità di T0. Un totale di 55/122 è stato escluso per diagnosi diverse da ALE (ad es. encefalite virale, disturbi neurodegenerativi, glioma, ictus) o per disturbi psichiatrici maggiori; applicando i criteri di Graus et al. (2016), ulteriori 36/67 sono stati esclusi, ottenendo una coorte finale di 31 pazienti ALE (anti-GAD=6; anti-LGI1=6; anti-CASPR2=5; sieronegativi=14). Da ottobre 2022, il reclutamento prospettico è proseguito presso l’Unità di Epilessia. La coorte finale ha raggiunto 57 pazienti nel gruppo ALE (31 identificati retrospettivamente e 26 prospetticamente), insieme a 36 pazienti con NAI-TLE e 58 controlli sani. Per i pazienti, i neurologi segnalavano i potenziali candidati durante l’attività clinica di routine, seguita da un pre-screening di compatibilità con ALE o NAI-TLE e da un processo di consenso strutturato in accordo con la Dichiarazione di Helsinki e l’approvazione del Comitato Etico. Tutti i partecipanti hanno effettuato una valutazione neurologica completa e una batteria NPS standardizzata che valutava ragionamento non verbale, linguaggio, attenzione, funzioni esecutive (inclusi set-shifting, switching e velocità di elaborazione sotto carico esecutivo), memoria di lavoro, memoria verbale e visiva (a breve e lungo termine) e abilità visuocostruttive. La batteria includeva: Raven Colored Progressive Matrices (CPM; Basso et al., 1987); Boston Naming Test (BNT; Goodglass et al., 1983); Attentive Matrices (AT; Spinnler & Tognoni, 1987); Trail Making Test A (TMT-A) e Trail Making Test B (TMT-B; Giovagnoli et al., 2016), con l’indice derivato Trail Making Test B-A (TMT B-A); Word Fluency su indizi fonemici (WF-P) e su indizi semantici (WF-S; Novelli et al., 1986); Rey–Osterrieth Complex Figure – Copia (ROCF-Copy) e Richiamo differito (ROCF-DR; Caffarra et al., 2002); Street’s Completation Test (SCT); Digit Span in avanti (DS-F) e all’indietro (DS-B); Corsi Block-Tapping Test (CBT; avanti/indietro); Short Story Recall (SSR; Novelli et al., 1986); Rey Auditory Verbal Learning – Richiamo immediato (RAVL-IR) e differito (RAVL-DR; Carlesimo et al., 1996); e Corsi Block-Supra-Span (CBSST; Spinnler & Tognoni, 1987). I gruppi ALE e NAI-TLE hanno inoltre eseguito almeno un EEG, una PET con 18F-fluorodesossiglucosio (18F FDG PET) e una MRI su scanner clinici (1,5 T Siemens Avanto Fit; 1,5 T Philips Achieva; 3 T Philips Achieva dStream). I test sierologici per gli Abs neuronali sono stati eseguiti secondo le procedure istituzionali. I controlli sani sono stati arruolati con un processo informativo analogo e abbinati per età, istruzione e sesso. Sfruttando l’infrastruttura dati unificata, il secondo studio mirava a caratterizzare in modo comprensivo il profilo cognitivo nella fase post-acuta dell’ALE con un disegno di coorte osservazionale, confrontando i pazienti con ALE con soggetti NAI-TLE e controlli sani. I controlli sani fornivano un riferimento normativo, mentre la NAI-TLE era inclusa come gruppo di controllo clinico con crisi e come benchmark temporo-limbico, consentendo di interpretare la compromissione cognitiva nell’ALE rispetto a una condizione che coinvolge prevalentemente la rete temporo-limbica. Questo framework comparativo intendeva distinguere, per quanto possibile, le sequele cognitive plausibilmente attribuibili a crisi/epilessia del lobo temporale da deficit più coerenti con un’eziologia autoimmune, con l’obiettivo specifico di identificare caratteristiche differenzianti clinicamente significative che possano facilitare il riconoscimento dell’ALE nelle presentazioni a predominanza di crisi, senza implicare equivalenza tra le due condizioni. Questo studio osservazionale monocentrico includeva ALE (n=57), NAI-TLE (n=36) e controlli sani (n=58). Le differenze tra gruppi nelle prestazioni cognitive sono state valutate con test di Kruskal–Wallis seguiti da confronti con correzione di Bonferroni. Le frequenze dei deficit cognitivi sono state sintetizzate e sono stati stimati modelli di regressione multipla con aggiustamento per durata di malattia e lateralizzazione della lesione alla MRI. Rispetto ai controlli sani, ALE e NAI-TLE mostravano maggiori compromissioni nella fluenza semantica e nella memoria verbale e visuospaziale, con effetti minori su attenzione e funzione esecutiva. Non emergevano differenze significative delle medie tra ALE e NAI-TLE. Tuttavia, le analisi di frequenza del deficit suggerivano una dissociazione: una proporzione più alta di ALE era compromessa nella memoria visuospaziale a lungo termine, mentre più NAI-TLE mostravano difficoltà nell’apprendimento verbale. In particolare, nei modelli di regressione, il fattore gruppo (ALE vs NAI-TLE) non prediceva la performance dopo aggiustamento per durata di malattia e lateralizzazione della lesione alla MRI. Nel complesso, il pattern di deficit di memoria osservato è coerente con le caratteristiche note dell’ALE, caratterizzata da un coinvolgimento mesiale temporale predominante e da un targeting autoimmune preferenziale dell’ippocampo e delle strutture limbiche adiacenti (Kelly et al., 2017; Budharam et al., 2019; Sanvito et al., 2024), che costituiscono il principale substrato neuroanatomico della memoria (Scoville & Milner, 1957; Squire & Zola-Morgan, 1991). Insieme alla relativa preservazione delle funzioni esecutive nella nostra coorte, questi risultati supportano un modello “a stadi” dell’ALE, in cui una fase iniziale di infiammazione diffusa che interessa reti su larga scala, incluse le connessioni fronto-temporali, è seguita da una fase cronica/residua caratterizzata da danno mesiale temporale persistente. Inoltre, l’assenza di differenze significative tra gruppi clinici, anche dopo il controllo dei confondenti, suggerisce che nella fase post-acuta i profili cognitivi siano determinati principalmente dalla topografia della lesione più che dall’eziologia autoimmune di per sé. Il terzo studio ha delineato in modo sistematico le differenze cognitive nell’ALE in funzione del sierostato e dei sottotipi di Abs. La coorte è stata suddivisa in gruppi sieropositivi e sieronegativi e, all’interno dei sieropositivi, stratificata in sottotipi INA-Abs e CSA-Abs. Le analisi hanno utilizzato test non parametrici e l’ANCOVA non parametrica di Quade per i confronti tra sottogruppi, aggiustando per la durata di malattia. In un’analisi preliminare, i pazienti sieronegativi mostravano prestazioni superiori ai sieropositivi nella fluenza verbale semantica e in misure esecutive (inclusi set-shifting, switching e velocità di elaborazione sotto carico esecutivo), con una tendenza verso una migliore organizzazione percettiva. Tuttavia, analisi successive hanno rivelato che tale pattern era guidato dalle peggiori prestazioni del gruppo INA-Abs rispetto ai sieronegativi, senza differenze tra INA-Abs e CSA-Abs. Complessivamente, questi risultati suggeriscono che la compromissione cognitiva nell’ALE sia più strettamente legata al profilo anticorpale che al solo sierostato. In particolare, le forme INA-Abs sono associate a una disfunzione esecutiva più marcata e persistente, plausibilmente riflettendo una perturbazione più ampia delle reti fronto-temporali. La persistenza di tali differenze dopo aggiustamento per la durata di malattia indica che la durata non spiega completamente la disfunzione esecutiva nelle INA-Abs, sollecitando una riconsiderazione dell’ipotesi “a stadi” di recupero esecutivo proposta nello Studio 2 e supportando un ruolo degli Abs nel modellare le traiettorie di malattia. In questo contesto, la presenza di INA-Abs sembra svolgere un ruolo cruciale, indicando un processo patologico qualitativamente distinto e più aggressivo che produce una disruzione più duratura dei circuiti neurali. Il quarto studio mirava a quantificare i cambiamenti intra-soggetto tra T0 e T1, identificare predittori clinici e radiologici della cognizione a T1 e testare se le traiettorie differissero per sierostato (sieropositivi vs sieronegativi) e per sottotipi di Abs (INA-Abs vs CSA-Abs). È stato condotto un follow-up longitudinale monocentrico circa un anno dopo T0, escludendo i pazienti che sviluppavano nuovi disturbi neurologici. Il campione comprendeva 28 pazienti reclutati prospetticamente, la maggior parte dei quali ha ricevuto trattamento farmacologico, inclusa terapia immunologica (IT) quando clinicamente indicata. La batteria NPS è stata ri-somministrata usando forme parallele per mitigare gli effetti di pratica. I cambiamenti T0–T1 sono stati valutati con test di Wilcoxon per ranghi con segno a due code e gli esiti significativi sono stati inseriti in modelli di regressione LASSO con covariate incluse: prestazione basale, IT, intervallo T0–T1 e reperti MRI, esiti epilettici e psichiatrici. Le traiettorie tra gruppi sono state stimate con modelli lineari a effetti misti (LMM) includendo effetti fissi di Tempo, Gruppo e interazione Tempo×Gruppo, un’intercetta casuale per partecipante e gradi di libertà di Satterthwaite. Analisi secondarie hanno utilizzato un fattore Gruppo a tre livelli (INA-Abs, CSA-Abs e sieronegativi). Complessivamente è stato osservato un miglioramento selettivo nella memoria visuospaziale a lungo termine e nella memoria di lavoro visuospaziale. La prestazione basale e i miglioramenti alla MRI sono emersi come predittori robusti dei guadagni in memoria visuospaziale a T1, mentre il miglioramento della memoria di lavoro era associato a inizio più precoce del trattamento, migliore controllo delle crisi e prosecuzione della IT, oltre alla prestazione basale e al sierostato. Non sono emerse interazioni Tempo×Gruppo significative negli LMM per nessun esito. Gli effetti principali del Tempo hanno confermato guadagni complessivi in memoria di lavoro e memoria visuospaziale, mentre gli effetti principali del Gruppo hanno rivelato differenze di livello: i sieronegativi superavano i sieropositivi nell’apprendimento visuospaziale e nella memoria episodica verbale e i CSA-Abs superavano gli INA-Abs nella velocità di elaborazione/ricerca visiva. Nel complesso, i risultati indicano un recupero selettivo ma consistente in domini specifici, plausibilmente supportato dall’adozione diffusa della IT e dal miglior controllo dell’attività epilettogena/infiammatoria. Tuttavia, le differenze di livello tra sieronegativi e sieropositivi non si traducevano, nell’arco temporale osservato, in pendenze di recupero divergenti, nonostante l’ipotesi di un danno maggiore nei fenotipi INA-Abs, verosimilmente per il follow-up relativamente breve e la dimensione campionaria limitata. In sintesi, attraverso questi studi, l’ALE risulta associata a una compromissione cognitiva sostanziale, caratterizzata da deficit centrali nei sistemi di memoria dipendenti dal circuito temporale e da una vulnerabilità esecutiva più pronunciata nei fenotipi INA-Abs. Nella fase post-acuta, l’assenza di differenze rispetto alla NAI-TLE suggerisce che i deficit cognitivi siano guidati prevalentemente dalla topografia della lesione piuttosto che dall’eziologia autoimmune, supportando un modello in cui la sede e l’estensione del danno sono i principali determinanti degli esiti NPS. Longitudinalmente, si osservano miglioramenti selettivi e clinicamente significativi nella memoria visuospaziale a lungo termine e nella memoria di lavoro visuospaziale. Tali guadagni sono stati predetti da una migliore performance basale e da un inizio più precoce del trattamento, nonché da miglioramenti alla MRI e dal controllo delle crisi. Infine, le traiettorie di recupero risultano ampiamente parallele tra i gruppi, sebbene con un “offset” di livello a favore dei sottogruppi sieronegativi e CSA-Abs rispetto agli INA-Abs, suggerendo che l’eziologia anticorpale contribuisca alla severità iniziale e ai profili di vulnerabilità. Questa caratterizzazione multidominio del profilo NPS fornisce ai clinici uno strumento concreto per l’inquadramento diagnostico, la pianificazione terapeutica e il monitoraggio longitudinale, facilitando lo sviluppo di linee guida più efficaci per la riabilitazione cognitiva. La stratificazione per profilo anticorpale, anziché per il solo sierostato, supporta scelte terapeutiche più mirate e l’implementazione precoce di interventi non farmacologici individualizzati. Poiché le sequele cognitive influenzano ampiamente il funzionamento quotidiano (rientro al lavoro, attività della vita quotidiana e relazioni interpersonali), una riabilitazione tempestiva e mirata può migliorare in modo significativo la qualità della vita e la reintegrazione sociale. L’uso di un’infrastruttura dati unificata e controllata per qualità, che integra dati retrospettivi e prospettici, biomarcatori multimodali e batterie NPS standardizzate, fornisce inoltre una base per future collaborazioni multicentriche nelle malattie rare, accelerando la produzione di evidenze e la traslazionalità clinica. Nonostante ciò, permangono alcune limitazioni: il campionamento di convenienza e il disegno monocentrico possono limitare la generalizzabilità; la rarità dell’ALE limita le dimensioni dei sottogruppi, soprattutto per le analisi dei sottotipi di Abs; le differenze tra gruppi in alcune variabili cliniche (ad es. durata di malattia, lateralizzazione della lesione alla MRI), pur gestite analiticamente, possono lasciare confondimento residuo; un follow-up di circa 12 mesi può essere insufficiente per cogliere una divergenza tardiva delle traiettorie di recupero, in particolare nei fenotipi INA-Abs. Inoltre, nonostante l’uso di forme parallele, gli effetti di pratica non possono essere del tutto esclusi. Infine, alcuni domini aggiuntivi, in particolare la cognizione sociale, non sono stati inclusi nella valutazione perché l’ampliamento della batteria avrebbe comportato un carico eccessivo (ossia tempi di somministrazione più lunghi e maggiore affaticamento dei partecipanti), con potenziale compromissione dell’affidabilità dei dati e della qualità complessiva. In questo studio, abbiamo limitato la valutazione ai domini cognitivi fondamentali (attenzione, memoria, linguaggio) e a componenti esecutive chiave, poiché l’obiettivo primario era ottenere un profilo standardizzato e comparabile del funzionamento cognitivo di base. Domini più specialistici e complessi, come la cognizione sociale, non sono quindi stati valutati, in quanto considerati accessori rispetto agli obiettivi dello studio. Inoltre, per vincoli logistici e di coordinamento nella collaborazione, i dati di cognizione sociale raccolti in una linea di ricerca parallela non hanno potuto essere integrati nelle analisi del presente studio e potranno essere esplorati in futuro con misure più fini e dedicate. In conclusione, questa tesi propone un framework metodologicamente coerente per caratterizzare il fenotipo cognitivo dell’ALE sia trasversalmente sia longitudinalmente, chiarendo i ruoli relativi del profilo anticorpale. I risultati forniscono elementi immediatamente traducibili per la pratica clinica (diagnosi, pianificazione terapeutica e riabilitazione cognitiva) e pongono le basi per studi multicentrici e follow-up più lunghi, finalizzati a raffinare modelli di recupero specifici per anticorpi e a ottimizzare interventi realmente personalizzati.
EXPANDING THE CLINICAL AND NEUROPSYCHOLOGICAL SPECTRUM OF AUTOIMMUNE LIMBIC ENCEPHALITIS: A LONGITUDINAL COHORT STUDY
Tallarita, Giulia
2026
Abstract
Autoimmune Limbic Encephalitis (ALE) is a rare neurological syndrome (prevalence is 13.7/100,000 people, while the incidence is 0.8/100,000 per year) characterized by an immune mediated inflammation of the brain and frequently associated with seizures, behavioral changes, memory and cognitive deficits (Kelley et al., 2017; Hermetter et al., 2018). It was initially identified as a paraneoplastic phenomenon, even if in recent years, a non-paraneoplastic variant has been characterized and several antibodies (Abs) have been described. Depending on the localization of the target antigen Abs, ALE can be divided into limbic encephalitis (LE) associated with Abs against intracellular neural antigen (INA-Abs) and LE associated with Abs against cell surface (CSA-Abs). The neuronal CAS-Abs most frequently associated with LE are leucine-rich glioma-inactivated 1 (LGI1), contactin-associated protein-like 2 (CASPR2), gamma-aminobutyric acid type B receptor (GABA-B receptor) and alpha-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptor (AMPA receptor). By contrast, the INA-Abs most frequently occurring with LE are anti-Hu (ANNA-1), anti-Ri (ANNA-2), anti-Ma/Ta (Ma2, also known as PNMA2), anti-CV2/CRMP5 (CRMP5) and glutamic acid decarboxylase (GAD) (Tüzün & Dalmau, 2007; Graus et al., 2016). Magnetic Resonance Imaging (MRI) findings often reveal hyperintense signals of the medial temporal lobes in T2-weighted or Fluid-Attenuated Inversion Recovery (FLAIR) images and the Electroencephalogram (EEG) often shows focal or generalized slow wave or epileptiform activity in the temporal lobe and in the striatum, diencephalon or rhombencephalon (Dubey et al., 2018). According to the clinical approach to diagnose ALE proposed by Graus and colleagues, cognitive difficulties and behavioral changes are the hallmark of the disorder and represent important biomarkers of disease acuity, progress and therapy. Anterograde amnesia, characterized by a severe autobiographical and episodic memory impairment, is the most frequently reported deficit (Loanen et al., 2019). However, there is growing evidence that cognitive changes in ALE may not be restricted to memory functions. Deficits in attention and executive functions including set-shifting, working memory and fluency may occur (Dodich et al., 2016; Wang et al., 2017; Witt & Helmstaedter, 2021; Gibson et al., 2020). Despite the prominence of cognitive dysfunctions in individuals with ALE, very few studies systematically investigate a wide range of cognitive domains. Moreover, some authors suggest that ALE associated Abs types exhibit distinct cognitive impairments. However, the degree to which different Abs types result in distinct neuropsychological (NPS) manifestations remains largely unknown and in literature the data are discordant. In their study, Frisch et colleagues found a distinct relationship between inductive processes and cognitive outcome in LGI1 and CASPR 2-Abs positive (CAS-Abs) and GAD-Abs-positive (INA-Abs) subforms of LE (Frisch et al., 2013). In addition, Muller in a systematic review shows that ALE patients with Abs against both LGI1 and CASPR2 (CAS-Abs) show higher percentages of NPS deficits compared to ALE patients with Abs against GAD (IN-Abs) (Mueller et al., 2023). Finally, Vrillon in a systematic review of 21 cases with anti-GAD confirms that anti-GAD LE appears to be associated with distinctive clinical features compared with other types of ALE (Vrillon et al., 2020). Conversely, this distinction is challenged by equally compelling evidence reporting no differences across Abs groups (Mueller et al., 2021; Mueller et al., 2022). ALE is treated by autoimmune suppression and better prognosis is associated with early treatment and with ALE associated to CSA-Abs (Hermetter et al., 2018; Dutra et al., 2018). However, the knowledge of cognitive phenotype over the course of the disease is limited and it has not been addressed in large studies. The significant variability in the methodology makes it challenging to derive solid conclusions. Isolated case reports and small series indicate that the clinical outcome is variable, as some reported residual memory deficits and others an apparent recovery (Titulaer et al., 2013., 2013; Van Sonderen et al 2016; Abboud et al., 2021). Against this background, the present thesis had the overarching objective of clarifying the nature, severity and trajectory of cognitive deficits in ALE through three complementary studies (Study 2-4), built upon a single shared data infrastructure (Study 1). The experimental design relied on a unified database developed to transparently integrate three studies with partially overlapping cohorts, (1) a cross-sectional comparison among ALE, non autoimmune temporal lobe epilepsy (NAI-TLE) and healthy controls (HCs), (2) an analysis focused exclusively on ALE patients stratified by serostatus and Abs subtypes and (3) a longitudinal assessment with 12 month follow up in an ALE subgroup. Recruitment took place at Fondazione IRCCS Istituto Neurologico Carlo Besta (Milan, Italy). Given the rarity of the condition and the lack of established effect size estimates, convenience sampling was adopted. For ALE, enrollment was conducted both retrospectively and prospectively and two time points were planned (baseline, T0; follow up, T1, the latter acquired only prospectively). General eligibility criteria were: age ≥18 years, at least 5 years of education, average abstract–logical reasoning on Raven’s Colored Progressive Matrices (RCPM >17.5), no diagnosis or ongoing evaluation for neurodegenerative disorders, no current major psychiatric episodes or severe systemic diseases, no substance abuse, ability to provide informed consent and no prior neurosurgery. Included ALE cases met Graus et al. (2016) criteria (possible autoimmune encephalitis [AE], definite ALE, or probable Abs negative encephalitis) and were assessed in the post acute phase. The NAI- TLE group met International League Against Epilepsy (ILAE) criteria for temporal lobe epilepsy (Scheffer et al., 2017), supported by semiology, EEG and MRI findings. HCs were recruited among hospital staff and visitors, without compensation, and matched for age (±5 years), education (±3 years) and sex. To establish a robust empirical foundation, the first study focused on curating and consolidating the patient database. ALE cohort construction began with a retrospective phase using a pre existing institutional database of 122 patients (since 2010) with “encephalitis” listed as diagnosis or diagnostic hypothesis. Each chart underwent joint manual review by the doctoral candidate and a neurologist with expertise in epilepsy to verify diagnostic accuracy, data completeness and T0 availability. A total of 55/122 were excluded due to diagnoses other than ALE (e.g., viral encephalitis, neurodegenerative disorders, glioma, stroke) or major psychiatric disorders; applying Graus et al. (2016) criteria, a further 36/67 were excluded, yielding a final cohort of 31 ALE patients (anti GAD=6; anti LGI1=6; anti CASPR2=5; seronegative=14). Since October 2022, prospective recruitment continued at the Epilepsy Unit. The final cohort reached 57 patients in the ALE group (31 identified retrospectively and 26 prospectively), alongside 36 patients with NAI-TLE and 58 HCs. For the patients, neurologists flagged potential candidates during routine clinical activity, followed by pre screening for compatibility with ALE or NAI-TLE and a structured consent process in accordance with the Declaration of Helsinki and Ethics Committee approval. All participants underwent comprehensive neurological assessment and a standardized NPS battery assessing nonverbal reasoning, language, attention, executive functions including set-shifting, switching and processing speed under executive load, working memory, verbal and visual memory (short- and long-term) and visuoconstructive abilities. The battery included: Raven Colored Progressive Matrices (CPM; Basso et al., 1987); Boston Naming Test (BNT; Goodglass et al., 1983); Attentive Matrices (AT; Spinnler & Tognoni, 1987); Trail Making Test A (TMT-A) and Trail Making Test B (TMT-B; Giovagnoli et al., 2016), with the derived index Trail Making Test B-A (TMT B-A); Word Fluency on Phonemic Cues (WF-P) and Word Fluency on Semantic Cues (WF-S; Novelli et al., 1986); Rey–Osterrieth Complex Figure – Copy (ROCF-Copy) and Rey–Osterrieth Complex Figure – Delayed Recall (ROCF-DR; Caffarra et al., 2002); Street’s Completation Test (SCT); Digit Span Forward (DS-F) and Digit Span Backward (DS-B); Corsi Block-Tapping Test (CBT; forward/backward); Short Story Recall (SSR; Novelli et al., 1986); Rey Auditory Verbal Learning – Immediate Recall (RAVL-IR) and Rey Auditory Verbal Learning – Delayed Recall (RAVL-DR; Carlesimo et al., 1996); and Corsi Block-Supra-Span (CBSST; Spinnler & Tognoni, 1987). ALE and NAI-TLE groups also underwent at least one EEG, an 18F Fluorodeoxyglucose Positron Emission Tomography (18F FDG PET) and MRI on clinical scanners (1.5 T Siemens Avanto Fit; 1.5 T Philips Achieva; 3 T Philips Achieva dStream). Serological testing for neuronal Abs was performed according to institutional procedures. HCs were enrolled with a similar information process and matched for age, education and sex. Leveraging the unified data infrastructure, the second study aimed to comprehensively characterize the cognitive profile in the post-acute phase of ALE using an observational cohort design, comparing patients with ALE with individuals NAI-TLE and with HCs. HCs provided a normative reference, whereas NAI-TLE was included as a seizure-bearing clinical control group and as a temporo-limbic benchmark, thereby enabling interpretation of cognitive impairment in ALE against a condition predominantly involving the temporo-limbic network. This comparative framework was intended to disentangle, insofar as feasible, cognitive sequelae plausibly attributable to seizures/temporal lobe epilepsy from deficits more consistent with an autoimmune etiology, with the specific objective of identifying clinically meaningful differentiating features that may facilitate recognition of ALE in seizure-dominant presentations, without implying equivalence between the two conditions. This single-center observational cohort study included ALE (n=57), NAI-TLE (n=36) and HCs (n=58). Between-group differences in cognitive performance were assessed with Kruskal–Wallis tests followed by Bonferroni-adjusted comparisons. Frequencies of cognitive deficits were summarized and multiple regression models were fitted with adjustment for disease duration and MRI lesion lateralization. Compared with HCs, ALE and NAI-TLE exhibited greater impairments in semantic fluency and in verbal and visuospatial memory, with smaller effects in attention and executive function. No significant mean differences emerged between ALE and NAI-TLE. However, deficit-frequency analyses suggested a dissociation: a higher proportion of ALE were impaired in long-term visuospatial memory, whereas more NAI- TLE showed difficulties in verbal learning. Notably, in regression models, the group factor (ALE versus NAI-TLE) did not predict performance after adjustment for disease duration and MRI lesion lateralization. Overall, the observed memory deficit pattern aligns with established features of ALE, characterized by predominant mesial temporal involvement and preferential autoimmune targeting of the hippocampus and neighboring limbic structures (Kelly et al., 2017; Budharam et al., 2019; Sanvito et al., 2024), which form the core neuroanatomical substrate of memory (Scoville & Milner, 1957; Squire & Zola-Morgan, 1991). Taken together with the relative preservation of executive functions in our cohort, these findings support a staged model of ALE whereby an initial phase of widespread inflammation affecting large-scale networks, including fronto-temporal circuits is followed by a chronic/residual phase characterized by persistent mesial temporal damage. Moreover, the lack of significant differences between clinical groups, even after controlling for confounders, suggests that in the post acute phase cognitive profiles are primarily determined by lesion topography rather than autoimmune etiology per se. The third study systematically delineated cognitive differences in ALE as a function of serostatus and Abs subtypes. The cohort was split into seropositive and seronegative groups and, within seropositives, stratified into INA Abs and CSA Abs subtypes. Analyses employed non-parametric tests and Quade’s non-parametric ANCOVA for subgroup comparisons, adjusting for disease duration. In a preliminary analysis, seronegative patients outperformed seropositives on semantic verbal fluency and executive measures (including set-shifting, switching and processing speed under executive load), with a trend toward better perceptual organization. However, subsequent analyses revealed that the pattern was driven by poorer performance in the INA-Abs group versus seronegatives, with no differences between INA-Abs and CSA-Abs. Taken together, these results suggest that cognitive impairment in ALE is more closely linked to the Abs profile than to serostatus alone. Specifically, INA-Abs forms are associated with more pronounced and persistent executive dysfunction, plausibly reflecting broader disruption of fronto-temporal networks. The persistence of these differences after adjustment for disease duration indicates that duration does not fully account for executive dysfunction in INA-Abs, prompting a reconsideration of the staged hypothesis of executive recovery proposed in Study 2 and supporting a role for Abs in shaping disease trajectories. In this context, the presence of INA-Abs appears to play a pivotal role, marking a qualitatively distinct and more aggressive pathological process that produces a more enduring disruption of neural circuits. The fourth study aimed to quantify within subject change between T0 and T1, identify clinical and radiological predictors of cognition at T1 and test whether trajectories differed by serostatus (seropositive versus seronegative) and Abs subtypes (INA Abs versus CSA Abs). A single-center longitudinal follow-up was conducted approximately one year after T0, excluding patients who developed new neurological disorders. The sample comprised 28 prospectively recruited patients, most of whom received pharmacological treatment, including IT as clinically indicated. The NPS battery was re-administered using parallel forms to mitigate practice effects. T0–T1 changes were evaluated with two-tailed Wilcoxon signed-rank tests and significant outcomes entered LASSO regression models with covariates including baseline performance, IT, T0–T1 interval and MRI findings, epileptic and psychiatric outcomes. Trajectories across groups were estimated with linear mixed-effects models (LMMs) including fixed effects of Time, Group and Time×Group interaction, a random intercept for participant and Satterthwaite degrees of freedom. Secondary analyses used a three level Group factor (INA Abs, CSA Abs and Seronegative). Selective improvement was observed in long term visuospatial memory and visuospatial working memory. Baseline performance and improvements on MRI emerged as robust predictors of visuospatial memory gains at T1, while working-memory improvement was associated earlier treatment initiation, improved seizure control and continuation of IT, in addition to baseline performance and serological status. No significant Time×Group interactions emerged in the LMMs for any outcome. Main effects of Time confirmed overall gains in working memory and visuospatial memory, while main effects of Group revealed level differences whereby seronegatives outperformed seropositives in visuospatial learning and verbal episodic memory and CSA Abs outperformed INA Abs in processing speed/visual search. Taken together, the results indicate selective yet consistent recovery in specific domains, plausibly supported by the widespread adoption of IT and improved control of epileptogenic/inflammatory activity. However, level differences between seronegatives and seropositives did not translate, within the observed window, into divergent recovery slopes, despite the hypothesis of greater damage in INA Abs phenotypes, likely due to the relatively short follow up and limited sample size. In summary, across these studies, ALE is associated with substantial cognitive impairment, characterized by core deficits in memory systems dependent on temporal circuitry and a more pronounced executive vulnerability in INA-Abs phenotypes. In the post-acute phase, the absence of differences relative to NAI-TLE suggests that cognitive deficits are driven predominantly by lesion topography rather than autoimmune etiology, supporting a model in which the location and extent of damage are the primary determinants of NPS outcomes. Longitudinally, selective and clinically meaningful improvements are observed in long-term visuospatial memory and visuospatial working memory. These gains were predicted by stronger baseline performance and earlier treatment initiation, as well as improvements in MRI findings and seizure control. Finally, recovery trajectories are broadly parallel across groups, albeit with a level offset favoring seronegative and CSA-Abs subgroups over INA-Abs, suggesting that Ab etiology contributes to initial severity and vulnerability profiles. This multidomain characterization of the NPS profile provides clinicians with a concrete tool for diagnostic framing, therapeutic planning and longitudinal monitoring, facilitating the development of more effective guidelines for cognitive rehabilitation. Stratification by Abs profile, rather than serostatus alone, supports more targeted therapeutic choices and early implementation of individualized non pharmacological interventions. Because cognitive sequelae broadly affect daily functioning (return to work, activities of daily living and interpersonal relationships), timely, targeted rehabilitation can meaningfully improve quality of life and social reintegration. The use of a unified, quality controlled data infrastructure integrating retrospective and prospective data, multimodal biomarkers and standardized NPS batteries also provides a foundation for future multicenter collaborations in rare diseases, accelerating evidence generation and clinical translatability. Despite this, some limitations remain: convenience sampling and the single center design may limit generalizability; the rarity of ALE constrains subgroup sizes, especially for Abs subtype analyses; between group differences in certain clinical variables (e.g., disease duration, MRI lesion lateralization), although analytically managed, may leave residual confounding; an approximately 12 month follow up may be insufficient to capture later divergence in recovery trajectories, particularly in INA Abs phenotypes. Additionally, despite the use of parallel forms, practice effects cannot be entirely excluded. Finally, some additional domains, particularly social cognition, were not included in the assessment because expanding the test battery would have entailed an excessive assessment burden (i.e., longer administration time and increased participant fatigue), potentially undermining data reliability and overall quality. In this study, we confined the assessment to core cognitive domains (attention, memory, language) and key executive components, as the primary objective was to obtain a standardized and comparable profile of basic cognitive functioning. More specialized and complex domains, such as social cognition, were therefore not evaluated, as they were considered ancillary to the study aims. Moreover, due to logistical and coordination constraints within the collaboration, social cognition data collected within a parallel research line could not be integrated into the analyses of the present study and may be explored in future work using more fine-grained, dedicated measures. In conclusion, this thesis proposes a methodologically coherent framework for characterizing the cognitive phenotype of ALE both cross sectionally and longitudinally, clarifying the relative roles of Abs profile. The findings provide immediately translatable elements for clinical practice diagnosis, therapeutic planning and cognitive rehabilitation finally lay the groundwork for multicenter studies and longer follow ups aimed at refining Abs specific recovery models and optimizing truly personalized interventions.| File | Dimensione | Formato | |
|---|---|---|---|
|
TesiPhd_Completa_Tallarita.pdf
accesso solo da BNCF e BNCR
Licenza:
Tutti i diritti riservati
Dimensione
1.74 MB
Formato
Adobe PDF
|
1.74 MB | Adobe PDF |
I documenti in UNITESI sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/20.500.14242/362046
URN:NBN:IT:UNICATT-362046