Cognitive and Patient‐Reported Outcome Measures in LGI‐1‐IgG Autoimmune Encephalitis

ABSTRACT

Objective

Clinical outcome measures for autoimmune encephalitis (AE) are not yet well defined. Cognitive outcome measures (CogOs) and patient‐reported outcomes (PROs) may capture the symptoms of AE, beyond clinician‐reported outcomes (ClinROs) (the Modified Rankin Scale [mRS] and Clinical Assessment Scale in Autoimmune Encephalitis [CASE]).

Methods

This retrospective cohort study examined LGI‐1‐IgG AE patients at Cleveland Clinic. CogOs included the Rey Auditory Verbal Learning Test (RAVLT) and Brief Visuospatial Memory Test (BVMT). PROs included Neuro‐QoL, PROMIS Global Physical (GPH) and Mental Health (GMH). Abnormal RAVLT, BVMT (≤ −1.5 SD), Neuro‐QoL, and PROMIS (≥ 10 T‐scores) proportions were analyzed. CogOs and PROs were correlated with ClinROs using Spearman's rank correlation.

Results

Forty‐four patients (64% male, mean age 68 [SD = 11], median follow‐up 19 months) were included. ClinROs improved, with median mRS decreasing from 2.5 (IQR = 2–3) to 2 (IQR = 1–2), and CASE from 5 (IQR = 3–6) to 3 (IQR = 2–3) (all p < 0.05). RAVLT (n = 15), BVMT (n = 16), Neuro‐QoL (n = 21), and PROMIS (n = 34) were available. At > 12 months from diagnosis, cognition was impaired on BVMT (56%) and RAVLT (20%). PROs showed impairments in Neuro‐QoL lower (38%) and upper (27%) extremity function, cognition (31%), and PROMIS‐GPH (25%). RAVLT correlated with Neuro‐QoL cognition, depression, and stigma, while BVMT correlated with PROMIS‐GMH (all unadjusted p < 0.05). Neither RAVLT nor BVMT correlated with ClinROs. ClinROs correlated with Neuro‐QoL lower extremity function, cognition, and social roles (all p < 0.05).

Interpretation

LGI‐1‐IgG AE patients have long‐term impairments in cognition, physical health, and social roles, which can be objectively measured using CogOs and PROs and are not captured by ClinROs.

1. Introduction

There are no validated outcome measures for autoimmune encephalitis (AE). Clinician‐reported outcome measures (ClinRO), such as the modified Rankin Scale (mRS), are often broad and too motor‐focused to capture the full spectrum of AE symptoms [1, 2, 3, 4, 5]. The Clinical Assessment Scale in Autoimmune Encephalitis (CASE) score is a disease severity score and includes more encephalitis symptoms, but has several limitations including lacking granularity for psychiatric and mood symptoms, lack of consideration for sleep symptoms, emphasis on specific NMDAR‐IgG AE symptoms (e.g., dysautonomia) limiting its generalizability, and the lack of validation [2, 6, 7]. Despite this initial improvement in gross acute encephalitis symptoms measured by ClinROs, patients often report long‐term symptoms and impairments compared with their pre‐morbid abilities [8, 9]. In this study, we focused on one of the most common causes of AE in adults, leucine‐rich glioma‐inactivated 1 IgG (LGI‐1‐IgG) AE [10, 11, 12].

Many patients with LGI‐1‐IgG AE have cognitive impairment on longitudinal neuropsychological evaluation [8, 13, 14, 15]. Our previous study found that all patients in our LGI‐1‐IgG AE cohort had cognitive impairment on at least one neuropsychological test and 80% had impairment on at least 2 neuropsychological tests [16]. Specific neuropsychological testing, including the Rey Auditory Verbal Learning Test (RALVT) measuring verbal learning and memory, the Brief Visuospatial Memory Test (BVMT) measuring visual learning and memory, and the Rey‐Osterrieth Complex Figure Test (ROCF) measuring visuospatial memory, has shown to be impaired in patients with LGI‐1‐IgG AE [16, 17]. These findings highlight the high frequency of residual cognitive impairment in LGI‐1‐IgG AE patients—an issue often overlooked in clinical practice and may be underreported in research. For instance, some studies tended to prioritize outcomes like improvements in mRS or seizure freedom over cognitive measures [18, 19]. This underscores the need for further investigation into cognitive outcome measures (CogOs) for this disease.

Patient‐reported outcome measures (PROs) have been proposed as outcome measures emphasized in other neurological disorders [20, 21, 22], and are often used as secondary endpoints in clinical trials [23, 24]. There is an emphasis on the need for multimodal outcome measures that reflect patient‐reported outcomes in addition to clinician‐reported [25, 26]. For clinicians, PROs facilitate symptom monitoring and personalized care [27]. Furthermore, PROs are simple outcome measures that can be incorporated into clinical visits using electronic surveys, enabling longitudinal sampling. Few studies on LGI‐1‐IgG AE have incorporated PROs to date, highlighting the need to further explore their value in both clinical practice and research [8, 28, 29].

This study aimed to explore the utility of CogOs and PROs to measure burden in patients with LGI‐1‐IgG AE. Specifically, we sought to identify the symptoms most frequently identified on PROs and determine whether PROs and CogOs correlate with ClinROs.

2. Methods

2.1. Study Design and Population

This retrospective cohort study included adult patients with LGI‐1‐IgG in serum or CSF and a clinical phenotype consistent with the diagnostic criteria for autoimmune encephalitis evaluated at the Cleveland Clinic between 2013 and 2024 [9]. CogOs were evaluated by neuropsychologists at the request of clinicians. PROs are completed prior to ambulatory clinic visits as standard care either through MyChart patient portal or on tablets in the waiting room. This study was approved by the Cleveland Clinic Institutional Review Board with a waiver of consent.

2.2. Cognitive Outcome Measures (CogOs)

2.2.1. Rey Auditory Verbal Learning Test (RAVLT)

The RALVT is an assessment of verbal memory [30]. RAVLT consists of immediate (sum of 5 learning trials) and delayed recall scores. Our previous study demonstrated that both the immediate and delayed recall scores were similarly impaired in LGI‐1‐IgG AE [16]. For this study, RAVLT immediate recall scores were used as a representative of RAVLT and is henceforth called RAVLT for simplicity. RAVLT scores were reported as age‐adjusted standard scores (mean = 100, SD = 15) [31]. In this study, an abnormal RAVLT score was defined as a score ≤ −1.5 standard deviations from the mean, consistent with clinical convention.

2.2.2. Brief Visuospatial Memory Test (BVMT)

The BVMT is an assessment of visuospatial memory [32]. BVMT includes both immediate (sum of 3 learning trials) and delayed recall scores. Our previous study demonstrated that both the immediate recall scores were more impaired than the delayed recall scores in LGI‐1‐IgG AE [16]. For this study, BVMT immediate scores were used as a representative of BVMT and are henceforth referred to as BVMT for simplicity. BVMT scores were reported as age‐adjusted T‐scores (mean = 50, standard deviation [SD] = 10) [33]. In this study, an abnormal BVMT score was defined as scores ≤ −1.5 standard deviations from the mean, consistent with clinical convention.

2.3. Patient‐Reported Outcome Measures (PROs)

2.3.1. Quality of Life in Neurological Disorders (Neuro‐QoL)

Neuro‐QoL is a self‐report instrument designed to evaluate the physical, mental, and social effects experienced by adults and children living with neurological conditions [34]. Neuro‐QoL scales are standardized to reference populations on a T‐scale with a mean of 50 and a SD of 10, where higher scores indicate more of the domain being measured [35]. For instance, higher Neuro‐QoL anxiety signifies worse anxiety, whereas higher upper extremity function indicates better function. These measures have been demonstrated as valid and reliable, and scores at least 1 SD worse than the population average (i.e., either > 10 or < 10 T‐score points, depending on the domain being measured) can be considered meaningfully worse (www.healthmeasures.net). In this study, impairment was defined as Neuro‐QoL score at least 1 SD below or above the reference population average.

2.3.2. Patient‐Reported Outcomes Measurement Information System (PROMIS)

PROMIS Global Health is a 10‐item measure of global health and includes a summary score for physical and mental global health [36]. PROMIS scales are also standardized to reference populations on a T‐scale with a mean 50 and a SD of 10. PROMIS Global Mental Health T‐scores less than 40 and Global Physical Health T‐scores less than 42 are considered fair/poor [37]. In this study, impairment was defined as PROMIS Global Physical Health T‐scores less than 42 and Mental Health T‐scores less than 40.

2.4. Clinician‐Reported Outcome Measures (ClinROs)

2.4.1. Modified Rankin Scale (mRS)

The mRS is a disability status scale from 0 (no symptoms) to 6 (death), developed to measure the degree of disability and dependence after stroke [38, 39]. The mRS is scored as follows: 0 for no symptoms, 1 for no significant disability, 2 for slight disability, 3 for moderate disability, 4 for moderately severe disability, 5 for severe disability, and 6 for death [1].

2.4.2. Clinical Assessment Scale in Autoimmune Encephalitis Score (CASE)

CASE is a 27‐item scoring system ranging from 0 to 27 designed specifically for grading severity of AE, comprised of symptom domains including seizure, memory dysfunction, psychiatric symptoms, consciousness, language, dyskinesia, gait instability, brainstem dysfunction, and weakness. Each subscale is graded on a severity scale from 0 to 3. For the seizure subscale, the grading is: 0 (none), 1 (controlled seizure), 2 (intractable seizures), and 3 (status epilepticus). For memory dysfunction, psychiatric symptoms, language problems, dyskinesia, gait instability, and weakness, the grading is: 0 (none), 1 (mild), 2 (moderate), and 3 (severe). The consciousness subscale is graded as: 0 (alert), 1 (drowsy), 2 (stupor), and 3 (comatose). Finally, the brainstem dysfunction subscale is graded as: 0 (none), 1 (gaze paresis), 2 (tube feeding), and 3 (ventilator care due to central hypoventilation) [6].

2.5. Statistical Analysis

Patient demographics, clinical characteristics, ClinROs, CogOs, and PROs were summarized by mean and SD or median and interquartile range (IQR) for continuous variables and count with percentage for categorical variables. PROs were summarized at two time points: 0–12 months and > 12 months after diagnosis for each patient. The time between diagnosis and PRO assessment varied by instrument and patient at each time point, as not all instruments were administered at every appointment in our routine practice (Figure S1 and Table S1). A Wilcoxon signed‐rank test was used only to determine whether CASE or mRS changed from diagnosis to the last follow up.

The relationships between time‐matched ClinROs, CogOs and PROs at the last clinical visit were examined using Spearmen's correlation. The levels of association were categorized using Spearman's correlation coefficients as follows: very weak (0–0.19), weak (0.20–0.39), moderate (0.40–0.59), strong (0.60–0.79), and very strong (0.80–1.0). Data analyses were conducted in R, version 4.3.1. All tests were two‐sided, and p‐values less than 0.05 were considered statistically significant. Given the limited sample size and exploratory nature of our study, we did not correct for multiple comparisons.

3. Results

3.1. Study Population

A total of 53 patients with LGI‐1‐IgG AE were evaluated, and 44 patients with available ClinROs were included (Figure S2), of whom 64% were male, 64% were of white race, and the mean age at disease onset was 68 (SD = 10.7) years (Table 1). The median time from symptom onset to diagnosis was 3.2 (IQR = 1.1–7.5) months. Tumors were present in 13 patients (30%), with thymoma identified in 1 patient.

TABLE 1.

Patient characteristics.

Characteristics	N	Statistics
Age at onset	44	68.0 (10.7)
Sex (male)	44	28 (63.6%)
Ethnicity
White	44	37 (84.1%)
Black		3 (6.8%)
Others		4 (9.1%)
Tumors a	44	13 (29.5%)
Time between symptom onset and diagnosis, months	44	3.2 [1.1, 7.5]
Seizures at onset	44	42 (95.5%)
Status epilepticus		8 (18.2%)
FBDS		27 (61.4%)
Cognitive symptoms at onset	44	39 (88.6%)
T2/FLAIR hyperintensity in the medial temporal lobe on brain MRI	44	24 (54.5%)
Time between symptom onset and acute treatment, months	39	2.0 [1.2, 7.0]
Acute treatment	44	37 (84.1%)
IVMP/high‐dose oral steroids		35 (79.5%)
IVIG		22 (50.0%)
PLEX		13 (29.5%)
Long‐term treatment	44	32 (72.7%)
Rituximab		22 (50.0)
IVIG		6 (13.6)
Mycophenolate mofetil		4 (9.1)
Azathioprine		1 (2.3)
Baseline mRS	44	2.5 [2, 3]
≤ 2		25 (56.8%)
> 2		19 (43.2%)
Baseline CASE	44	5 [3, 6]
Total follow‐up period, months	44	19.1 [8.6, 50.5]

Note: Statistics presented as mean (SD), median [P25, P75], N (column %).

Abbreviations: CASE, Clinical Assessment Scale in Autoimmune Encephalitis; FBDS, faciobrachial dystonic seizure; IVIG, intravenous immunoglobulin; IVMP, intravenous methylprednisolone; MoCA, Montreal Cognitive Assessment; mRS, modified Rankin Scale; PLEX, plasmapheresis.

^a Among the 13 patients with documented tumors, 4 had prostate cancer, 2 had pancreatic cancer, 2 had breast cancer, 2 had lung adenocarcinoma, 1 had bladder cancer, 1 had B‐cell lymphoma, and 1 had thymoma.

The median follow‐up time was 19.1 (IQR = 8.6–50.5) months. Most patients experienced seizures (42 [96%]), including 27 patients (61%) with FBDS, and cognitive symptoms (39 [89%]). Most patients (37 [84%]) received acute immunosuppressive therapies, including intravenous methylprednisolone/high‐dose oral steroids, intravenous immunoglobulin (IVIG), and/or plasmapheresis. Long‐term immunosuppressive/immunomodulatory therapies were prescribed for 32 patients (73%), most frequently with rituximab (22 [50%]) and IVIG (6 [14%]) (Table 1). Most patients (39 [89%]) were seizure‐free at the last follow‐up, although the majority (33 [75%]) continued treatment with anti‐seizure medications.

3.2. Longitudinal Disease Outcomes: CogOs

3.2.1. RAVLT and BVMT

There were 15 patients with RAVLT scores and 16 patients with BVMT scores. The mean RAVLT score, assessed at a median time of 13 (IQR = 9–61) months from disease onset, was 95.5 (SD = 14.1), with 3/15 patients (20%) showing impairment (Table 2). The mean BVMT T‐score, assessed at a median time of 13 (IQR = 9–39) months from disease onset, was 39.6 (SD = 12.1), with 9/16 patients (56.3%) showing impairment (Table 2).

TABLE 2.

Last available cognitive outcome measures.

Cognitive Assessment	N	Statistics
RAVLT a
Standard score	15	95.5 (14.1)
Impairment of standard score	15	3 (20.0%)
BVMT b
T‐score	16	39.6 (12.1)
Impairment of T‐score	16	9 (56.3%)

Note: Statistics presented as mean (SD), median [P25, P75], N (column %).

Abbreviations: BVMT, Brief Visuospatial Memory Test; RAVLT, Rey Auditory Verbal Learning Test.

^a RAVLT scores were evaluated at a median time of 13 (IQR = 9–61) months from disease onset, and impairment in standard scores was defined as 1.5 standard deviations below the normative mean, consistent with clinical convention.

^b BVMT scores were evaluated at a median time of 13 (IQR = 9–39) months from disease onset, and impairment in standard scores was defined as 1.5 standard deviations below the normative mean, consistent with clinical convention.

3.3. Longitudinal Disease Outcomes: PROs

Of the 44 patients in the study, 21 (47.7%) completed Neuro‐QoL and 34 (77.3%) completed PROMIS Global Health during the study period (Figure S3). The median time from diagnosis to the last visit with available PRO data > 12 months after diagnosis was 39.5 (IQR = 23.2–66.9) months for Neuro‐QoL and 43.8 (IQR = 19.4–68.1) months for PROMIS. Figures S4 and S5 illustrate the changes in PROs over time.

3.3.1. Neuro‐QoL

Within 0–12 months of diagnosis, four out of 15 patients (26.7%) had impaired Neuro‐QoL lower extremity and cognitive function. Impairments in Neuro‐QoL upper extremity function and social role participation were observed in three of 15 patients (20% each). After 12 months after diagnosis, Neuro‐QoL lower extremity function was impaired in 6 out of 16 (37.5%). This was followed by Neuro‐QoL cognitive and upper extremity function, which were impaired in 5 out of 16 patients (31.2%) and 4 out of 15 patients (26.7%), and Neuro‐QoL social role satisfaction, impaired in 3 out of 16 patients (18.8%) (Figure 1 and Table S2).

FIGURE 1.

Proportions of patients with impairment on Neuro‐QoL. A different subset of patients contributed to Neuro‐QoL assessment at each of the two time points. Neuro‐QoL, the Neuro‐Quality of Life scale.

3.3.2. PROMIS

Within 0–12 months of diagnosis, impairment in PROMIS Global Physical Health was reported in 9 of 24 patients (37.5%), while 9 of 28 patients (32.1%) exhibited impairment in PROMIS Global Mental Health. After 12 months from diagnosis, 5 of 20 patients (25%) had impairment in PROMIS Global Physical Health, and 3 of 20 patients (15%) had impairment in PROMIS Global Mental Health (Figure 2 and Table S2).

FIGURE 2.

Proportions of patients with impairment on PROMIS‐GH. A different subset of patients contributed to PROMIS‐GH assessment at each of the two time points. PROMIS‐GH, the Patient Reported Outcomes Measurement Information System Global Health.

3.4. Longitudinal Disease Outcomes: ClinROs

3.4.1. mRS and CASE

The median mRS improved from 2.5 (IQR = 2–3) at diagnosis to 2 (IQR = 1–2) at the last follow‐up (p < 0.001), (Figure S6). Twenty of 44 patients (46%) had mRS < 2 at the last follow‐up visit. The median CASE improved from 5 (IQR = 3–6) at diagnosis to 3 (IQR = 2–3) at the last follow‐up (p < 0.001), (Figure S6). All CASE subscale scores decreased from disease onset to the last follow‐up visit (Figure S7).

3.5. Subgroup Analysis

3.5.1. Categorized by Seizure Status at the Last Follow‐Up Visit

Patients with controlled or intractable seizures showed meaningfully worse Neuro‐QoL social role satisfaction and stigma, with trends toward higher Neuro‐QoL depression, anxiety, fatigue, and sleep disturbance than those without seizures (Table S5).

3.5.2. Categorized by Sex

Females reported worse Neuro‐QoL cognitive function, social role participation and satisfaction, sleep disturbance, fatigue, anxiety, depression, and stigma, and also had lower RAVLT scores compared to males (Table S6).

3.6. Correlation Between CogOs and ClinROs

Time‐matched mRS was not correlated with RAVLT (r = 0.16, p = 0.61) or BVMT (r = 0.11, p = 0.71) (Tables 3 and 4). Time‐matched CASE score was not correlated with RAVLT (r = 0.11, p = 0.73) or BVMT (r = 0.10, p = 0.71) (Tables 3 and 4).

TABLE 3.

Rey Auditory Verbal Learning Test, clinician‐reported outcome measures, and Patient‐reported outcome measures at matching time points.

Outcome measures	N	RAVLT standard score
		Correlation coefficients (95% CI)	p
Clinician‐reported outcome measures
mRS	13	0.16 (−0.43, 0.65)	0.608
CASE	13	0.11 (−0.47, 0.62)	0.727
Patient‐reported outcome measures
Neuro‐QoL
Upper extremity function	13	−0.04 (−0.58, 0.52)	0.896
Lower extremity function	13	0.06 (−0.51, 0.59)	0.855
Cognitive function	14	0.55 (0.03, 0.84)	0.040
Social role participation	14	0.24 (−0.33, 0.68)	0.409
Social role satisfaction	14	0.34 (−0.23, 0.74)	0.236
Sleep disturbance	13	−0.45 (−0.80, 0.13)	0.121
Fatigue	14	−0.46 (−0.80, 0.10)	0.100
Anxiety	13	−0.28 (−0.72, 0.33)	0.363
Depression	14	−0.68 (−0.89, −0.23)	0.008
Stigma	13	−0.61 (−0.87, −0.09)	0.027
PROMIS
Global physical health	14	0.60 (0.10, 0.86)	0.0234
Global mental health	14	0.46 (−0.09, 0.80)	0.097

Note: Bold: Patient‐reported outcome measures significantly correlated with the Rey Auditory Verbal Learning Test.

Abbreviations: CASE, Clinical Assessment Scale in Autoimmune Encephalitis; mRS, modified Rankin Scale; Neuro‐QoL, the Neuro‐Quality of Life scale; PROMIS, the Patient Reported Outcomes Measurement Information System Global; RAVLT, Rey Auditory Verbal Learning Test.

TABLE 4.

Brief Visuospatial Memory Test, clinician‐reported outcome measures and patient‐reported outcome measures at matching time points.

Patient‐reported outcome measures	N	BVMT T‐score
		Correlation coefficients (95% CI)	p
Clinician‐reported outcome measures
mRS	15	0.11 (−0.43, 0.59)	0.705
CASE	15	0.10 (−0.43, 0.59)	0.710
Patient‐reported outcome measures
Neuro‐QoL
Upper extremity function	14	−0.30 (−0.72, 0.28)	0.301
Lower extremity function	14	−0.23 (−0.68, 0.34)	0.423
Cognitive function	15	0.26 (−0.29, 0.68)	0.351
Social role participation	15	0.21 (−0.34, 0.65)	0.588
Social role satisfaction	15	0.11 (−0.43, 0.59)	0.454
Sleep disturbance	14	−0.16 (−0.64, 0.41)	0.594
Fatigue	15	−0.21 (−0.65, 0.34)	0.562
Anxiety	14	0.16 (−0.41, 0.63)	0.458
Depression	15	−0.16 (−0.62, 0.38)	0.695
Stigma	13	−0.18 (−0.65, 0.39)	0.533
PROMIS
Global physical health	14	0.49 (−0.05, 0.81)	0.073
Global mental health	14	0.67 (0.21, 0.88)	0.009

Note: Bold: Patient‐reported outcome measures significantly correlated with the Rey Auditory Verbal Learning Test.

Abbreviations: BVMT, Brief Visuospatial Memory Test; CASE, Clinical Assessment Scale in Autoimmune Encephalitis; mRS, modified Rankin Scale; Neuro‐QoL, the Neuro‐Quality of Life scale; PROMIS, the Patient Reported Outcomes Measurement Information System Global.

3.7. Correlation Between CogOs and PROs

RAVLT score was associated with Neuro‐QoL cognitive function (r = 0.55, p = 0.040), depression (r = −0.68, p = 0.008), stigma (r = −0.61, p = 0.027), and PROMIS Global Physical Health (r = −0.57, p = 0.007), (Tables 3 and 4, Figures 3 and 4). BVMT T‐score was associated with PROMIS Global Mental Health (r = 0.67, p = 0.009), (Tables 3 and 4, Figures 3 and 4).

FIGURE 3.

Correlation between clinician‐reported outcome measures and cognitive outcome measures and Neuro‐QoL at matching time points (Last Available Paired Scores). *p < 0.05. BVMT, Brief Visuospatial Memory Test; CASE, Clinical Assessment Scale in Autoimmune Encephalitis; mRS, modified Rankin Scale; Neuro‐QoL, the Neuro‐Quality of life scale; RAVLT, Rey Auditory Verbal learning Test.

FIGURE 4.

Correlation between clinician‐reported outcome measures and cognitive outcome measures and PROMIS at matching time points (Last Available Paired Scores). *p < 0.05. BVMT, Brief Visuospatial Memory Test; CASE, Clinical Assessment Scale in Autoimmune Encephalitis; mRS, modified Rankin Scale; PROMIS‐GH, the patient reported outcomes measurement information system global health; RAVLT, Rey Auditory Verbal Learning Test.

3.8. Correlations Between PROs and ClinROs

Last time‐matched mRS was inversely associated with four Neuro‐QoL: social role participation (r = −0.59, p = 0.004), social role satisfaction (r = −0.51, p = 0.019), cognitive function (r = −0.57, p = 0.007), and lower extremity function (r = −0.45, p = 0.039) (Figures 3 and 4, Table S3).

Last time‐matched total CASE score was significantly correlated with several PROs in a negative direction (worse CASE associated with worse function) including: Neuro‐QoL upper extremity function (r = −0.57, p = 0.007), Neuro‐QoL lower extremity function (r = −0.66, p = 0.001), Neuro‐QoL social role participation (r = −0.72, p < 0.001), Neuro‐QoL social role satisfaction (r = −0.61, p = 0.004), Neuro‐QoL cognitive function (r = −0.64, p = 0.002), and PROMIS Global Physical Health (r = −0.44, p = 0.024).

Last time‐matched total CASE score was also correlated with several PROs in a positive direction (worse CASE associated with greater symptoms) including Neuro‐QoL anxiety (r = 0.47, p = 0.037) and stigma (r = 0.48, p = 0.030). However, last time‐matched CASE score did not significantly correlate with Neuro‐QoL depression, fatigue, sleep disturbance, and PROMIS Global Mental Health (Figures 3 and 4, Table S4).

4. Discussion

This study assessed CogOs (RAVLT and BVMT) and PROs (Neuro‐QoL and PROMIS) in a cohort of LGI‐1‐IgG AE and identified several important findings for clinical practice and outcomes research. The most frequent long‐term symptoms detected on CogOs and PROs were cognitive difficulties, poor overall physical and mental health, and motor function limitations, many of which are not captured in the currently used ClinROs. ClinROs (mRS and CASE) did not show a correlation with CogOs, suggesting cognitive is not captured well in common clinical outcome measures reported. ClinROs (mRS and CASE) correlated well with PROs that evaluate function, including Neuro‐QoL cognition, motor function, social roles, and PROMIS Global Physical Health, but did not capture the majority of AE symptoms, including Neuro‐QoL depression, fatigue, sleep disturbance, and PROMIS Global Mental Health. Overall, this study suggested that CogOs and PROs may be a useful adjunct to ClinROs in capturing AE symptomatology.

4.1. Main AE Symptoms Captured on CogOs and PROs

The most common symptoms identified in this study were memory impairment, reduced overall physical and mental well‐being, and difficulties with motor function. Residual verbal and visuospatial memory impairment was identified in many patients with LGI‐1‐IgG AE (20%–56%) using CogOs. Patients may exhibit difficulties with recalling word lists and people's names, recognizing faces, and locating objects. Abnormal BVMT results were found in over half of the patients, while abnormal RAVLT results were observed in one‐fifth. PROs assessing cognition (Neuro‐QoL cognitive function) also revealed abnormalities in more than one‐fourth of the patients in this cohort. Together, these findings underscore the importance of residual cognitive problems, particularly in verbal and visuospatial memory, in patients with LGI‐1‐IgG AE. This aligns with previous studies that have reported long‐term cognitive impairment in LGI‐1‐IgG AE, which may be linked to macrostructural and network changes within the hippocampal memory system [8, 40]. Notably, more than half of the patients in this cohort exhibited abnormal MRI findings in the medial temporal region, further supporting potential disruption of regions impacting cognition.

A high proportion (32%) of LGI‐1‐IgG AE patients experienced mental health problems and related issues, as assessed by PROMIS Global Mental Health, particularly within the first 12 months after diagnosis. Neuro‐QoL depression and anxiety picked up a small proportion of patients with long‐term abnormalities (about 7% each). The differences in proportions are likely due to the different survey questions, with PROMIS Global Mental Health measure more broadly capturing emotional, social, and quality of life issues. This finding suggests that patients continue to experience mental health problems despite treatment with immunosuppressive therapies, highlighting the need to investigate additional approaches, such as symptom‐based therapies (e.g., antidepressants).

PROMIS Global Physical Health consists of the questions related to overall physical health, abilities to perform everyday physical activities, fatigue, and pain, and Neuro‐QoL extremity function were abnormal in a significant proportion of LGI‐1‐IgG AE patients. These questions reflect the ability to perform routine activities, such as getting on and off the toilet, transitioning from bed to a chair, brushing teeth, and washing and drying the body [41]. These tasks require multiple domains of function and do not specifically measure motor weakness and may indicate functional impairment secondary to cognitive changes and also contributed to by comorbidities or medications (e.g., anti‐seizure medications). The PROMIS Global Physical Health is based on a reference sample of the U.S. general adult population (aged 18 and older). Given the higher mean age of this cohort (68 years, SD = 10.7), age‐related limitations in daily function may also have contributed to these findings.

4.2. Application of PROs in Clinical Practice to Measure AE Symptoms

Our data suggest that both Neuro‐QoL and PROMIS both have advantages in identifying symptoms at different time points. At 0–12 months after symptom onset, our observations suggest that PROMIS can serve as a useful screening tool to capturing AE symptoms (both mental and physical symptoms), activity limitation, and impaired quality of life. Specifically, PROMIS Global Mental Health captured a greater proportion of mental health symptoms compared to Neuro‐QoL.

After 12 months from diagnosis, however, Neuro‐QoL cognitive function captured abnormalities in approximately 30% of the patients, while PROMIS Global Mental Health reported abnormalities in only 15% of the patients. Therefore, using PROMIS Global Mental Health alone might not capture all cognitive function abnormalities at this time point.

4.3. CogOs and Their Correlation With ClinROs and PROs

Neither the RAVLT nor the BVMT were correlated with ClinROs (mRS and CASE). This finding highlights the limitations of ClinROs in capturing memory impairment in LGI‐1‐IgG AE. RAVLT was correlated with Neuro‐QoL cognitive function, suggesting that incorporating Neuro‐QoL cognitive function could help identify patients needing further neuropsychological evaluation. RAVLT also correlated with depression, suggesting a relationship between mental health and memory impairment in LGI‐1‐IgG AE, similar to other neurological disorders (e.g., dementia or stroke) as well as in the general population [42, 43, 44].

4.4. Correlation Between PROs and ClinROs

The correlation between time‐matched ClinROs and PROs showed that higher CASE scores and mRS were associated with worse function, including Neuro‐QoL cognition, motor function, social roles, and PROMIS Global Physical Health. This suggests ClinROs and PROs that measure function are linked: the more severe and disabling the disease, the worse the patient's functional outcomes.

In contrast, among the PROs that measure symptoms, CASE correlated only with Neuro‐QoL anxiety. Many PROs assessing symptoms of fatigue, sleep, and mental health did not correlate with mRS or CASE, suggesting that these symptoms are not adequately captured by these ClinROs. The discrepancy in correlations between CASE scores and certain PROs may partly reflect the absence of fatigue and sleep domains in CASE, despite the inclusion of memory dysfunction, psychiatric symptoms, and weakness.

4.5. Future Directions

Future studies should focus on developing a composite outcome measure that integrates ClinROs, PROs, and objective cognitive outcome measures to be used longitudinally in both clinical practice and clinical trials. Moreover, as residual symptoms were identified across multiple domains among patients with LGI‐1‐IgG AE, further research is also needed to understand the immunobiological basis for these long‐term symptoms and whether they respond to immunosuppressive therapy or whether there are potential benefits for rehabilitation or other symptom‐based therapies (such as antidepressants) for these chronic symptoms.

4.6. Limitations

This study has several limitations inherent to its retrospective design. The timing of CogOs, PROs, and ClinROs assessments varied for each patient, as evaluations were conducted based on clinical needs; however, this approach reflects real‐world clinical practice. A comprehensive neuropsychological battery was not included. The number of patients who underwent CogOs or completed each PRO varied. The small size of our cohort, typical of studies on rare neurological diseases, may limit generalizability. Additionally, p‐values were not adjusted for multiple comparisons, so the possibility of false‐positive results due to chance cannot be excluded. Although most patients completed self‐reported outcomes on tablets, device use in older adults may still be a potential limitation. Finally, as some of the mRS and CASE scores were retrospectively assigned, assessment bias may have occurred.

5. Conclusion

Despite improvements in ClinROs, many patients with LGI‐1‐IgG AE had memory impairment detected using CogOs, and residual symptoms and functional impairments detected on PROs. This study highlights the need to shift clinical practice and research from relying solely on ClinROs (such as mRS and/or CASE) to adopting multimodal outcome measures that prioritize CogOs and PROs in the assessment of LGI‐1‐IgG AE.

This study identified key abnormal domains of PROs in LGI‐1‐IgG AE, including motor function, cognition, and social roles, which may help clinicians and researchers select appropriate PRO domains for longitudinal outcomes. The study also revealed that CASE and mRS scores were more strongly correlated with PROs that measure function and less with PROs that measure symptoms and CogOs. Together, this highlights the gap in outcome measures not currently captured by ClinROs.

Author Contributions

T.O., C.S., A.K. contributed to the conception and design of the study. T.O., C.S., N.T., R.G. contributed to the acquisition and analysis of data. T.O., C.S., A.A., V.P., B.L., J.A.C., R.G., A.K. contributed to drafting the text or preparing the figures.

Conflicts of Interest

C.S. has received personal compensation for scientific advisory boards with Genentech and Amgen. T.O., R.G., N.T., A.A., and B.L. report no disclosures. V.P. received personal compensation from Xenon Pharmaceuticals, Catalyst Pharmaceuticals, Ovid Therapeutics, Eisai Inc., and UNEEG. J.A.C. has received personal compensation for consulting for Astoria, Bristol‐Myers Squibb, Convelo, and Viatris, and chairing a DSMB for Celltrion. A.K. has received personal compensation for consulting with Alexion, and for serving as an Editor of Neurology Open Access.

Supporting information

Data S1.

Data Availability Statement

Anonymized data not published can be made available by reasonable request from any qualified investigator, subject to approval by the Cleveland Clinic Institutional Review Board.