AMPAR autoimmunity: Neurological and oncological accompaniments and co-existing neural autoantibodies

Abstract

α -amino-3-hydroxy-5-methyl-4-isoxazole propionic acid receptor (AMPAR) encephalitis is rare but treatable. We reviewed the clinical and autoantibody profiles of 52 AMPAR-IgG-positive patients (median age 48 years [range 12–81]; 38 female) identified at the Mayo Clinic neuroimmunology laboratory. Main presentation was encephalitis; symptoms other than encephalitis associated with co-existing antibodies (p = 0.004). A tumor was found in 33/44; mostly thymoma. Most patients had partial (14/29) or complete (11/29) immunotherapy response. Thirty-one patients had at least one co-existing antibody that predicted thymoma in paraneoplastic patients (p = 0.008). In conclusion, in AMPAR encephalitis co-existing antibodies predict clinical presentation other than encephalitis and thymoma.

1. Introduction

Autoimmune α-amino-3-hydroxy-5-methyl-4-isoxazole propionic acid receptor (AMPAR) encephalitis is rare and was first described in 2009 (Lai et al., 2009). Tumors are found in approximately two-thirds of patients, most commonly thymoma and small-cell lung cancer (SCLC) (Laurido-Soto et al., 2019; Höftberger et al., 2015). These numbers are based on review of published cases to date (Laurido-Soto et al., 2019) and several case series, the largest including 22 patients (Höftberger et al., 2015).

Co-existing neural antibodies are reported in 32% of patients (Höftberger et al., 2015), and the presence of “high-risk” antibodies for cancer were suggested to associate with more severe presentations, or poorer outcome (Jia et al., 2020; Martinez-Hernandez et al., 2020; Jia et al., 2021; Guasp et al., 2021). In addition, AMPAR-IgG can be found in patients with malignancy, as a tumor biomarker, without evidence of encephalitis (Zekeridou et al., 2016).

The analysis of larger cohorts of patients can clarify the clinical and laboratory features, tumor associations, and treatment response of patients with AMPAR-IgG associated autoimmunity, as well as the influence of co-existing neural antibodies on presentation and prognosis.

2. Material and methods

The study was approved by the Mayo Foundation Institutional Review Board. Patients who had paraneoplastic and/or encephalopathy neural antibody evaluation in the Mayo Clinic Neuroimmunology Laboratory and tested positive in serum and/or CSF for AMPAR-IgG by cell-binding assay through June 1, 2020 were included. Supplementary table 1 details the serum and CSF testing results of the cohort. Patients were tested on a clinical service basis at the time of sample collection, for neural antibodies, with previously published and validated techniques (Zekeridou et al., 2019), including antibodies specific for muscle acetylcholine receptor binding and modulating (grouped together, mAChR), ganglionic acetylcholine receptor (gAChR), striational (STR), P/Q and N-type voltage gated calcium channel (grouped VGCC), glutamic acid decarboxylase 65-kilodalton isoform (GAD65, included if values >20 nmol/L in serum), anti-neuronal nuclear antibodies 1, 2, and 3 (ANNA-1 or anti-Hu, ANNA-2 or anti-Ri and ANNA-3 or anti-DACH1), anti-glial nuclear antibody 1 (AGNA1 or anti-SOX1), purkinje cell cytoplasmic autoantibody type 1, 2 and Tr (PCA-1 or anti-Yo, PCA-2 or anti-MAP1B, and PCA-Tr or anti-DNER), amphiphysin, glial fibrillary acidic protein (GFAP), collapsin response-mediator protein-5 (CRMP5 or anti-CV2), N-methyl-d-aspartate receptor (NMDAR), gamma-aminobutyric acid B receptor (GABA_BR), leucine-rich glioma-inactivated 1 (LGI-1), and contactin-associated protein-like 2 (CASPR2). Clinical information was obtained from the electronic medical record (11; patients seen at Mayo Clinic) or provided by ordering physicians who either contacted our laboratory, or were contacted by phone to discuss the results of antibody testing (41 patients). Clinical response after immunotherapy was defined as partial (when there was improvement documented by the treating provider but not resolution of symptoms), or complete (when patients returned to baseline). Patients without clinical information or limited autoantibody testing were excluded.

Variables were summarized using mean and range for continuous variables or count and percentage for categorical. Fisher’s exact test was used to test the association between several categorical outcome variables and multiple biological variables. Ordinal logistic regression was used to examine the relationship between treatment response and various factors. All analyses were performed using Rv4.0.3 (R Foundation for Statistical Computing, Vienna, Austria).

3. Results

3.1. Clinical presentation

Fifty-two patients were included; median age was 48 years (range 12–81); 38 (73.1%) were female. The main clinical presentations (Table 1) were one or more of encephalitis without (n = 35), or with seizures (n = 15), ataxia with (n = 4) or without (n = 1) encephalitis, and MG with (n = 6) or without (n = 1) encephalitis (Fig. 1). Five (83%) of 6 patients with MG and encephalitis had a diagnosis of MG predating onset of encephalitis by a maximum of 5 years. Five patients had other movement disorders and encephalitis: dyskinesias with co-existing CRMP5-IgG (n = 1), dyskinesias +/− tremor with co-existing NMDAR-IgG (n = 2), tremor with co-existing ANNA1 (n = 1) and parkinsonism with co-existing mAChR antibodies (n = 1). Additional symptoms were pain (n = 3) and weakness (n = 3 with no additional clinical details; two had multiple co-existing antibodies).

Table 1.

Demographics, clinical characteristics, ancillary testing, treatments and outcomes.

	Entire cohort	AMPAR-IgG alone	1 co-existing antibody						2 or more co-existing antibodies
Number of patients	52	21	13						18
Median age at time of first AMPAR antibody positivity, y	48 (12–81)	53 (18–81)	56 (12–79)						52 (15–68)
Female sex	38 (73%)	14 (67%)	11 (85%)						13 (72%)
			mAChR-IgG	CRMP5-IgG (anti-CV2)	NMDAR-IgG	LGI-1-IgG	PCA-1-IgG (anti-Yo)	gAChR-IgG
Number of patients			4	1	5	1	1	1
Neurologic manifestations
Encephalitis	50 (96%)	21 (100%)	4 (100%)	1 (100%)	5 (100%)	1 (100%)	0	1 (100%)	17a (94%)
Seizures	15 (29%)	4 (19%)	2 (50%)	0	3 (60%)	1 (100%)	0	1 (100%)	4 (22%)
Ataxia	5 (10%)	1 (5%)	1 (25%)	0	0	0	1b (100%)	0	2 (11%)
Myasthenia gravis	7 (13%)	0	2 (50%)	0	0	0	0	0	5 (28%)
Pain	3 (6%)	1 (5%)	0	0	0	0	0	0	2 (11%)
Weakness	3 (6%)	1 (5%)	0	0	0	0	0	0	2 (11%)
Other movement disorder	5 (10%)	0	1c (25%)	0	1 (20%)	0	0	0	3d (17%)
CSF
	Pleocytosis	Pleocytosis	Pleocytosis						Pleocytosis
	19/29 (66%)	5/10 (50%)	7/10 (70%)						7/9 (78%)
MRI
	Normal	Normal	Normal						Normal
	12/36 (33%)	2/13 (15%)	7/12 (58%)						3/11 (27%)
Malignancy Data e
	Entire cohort	AMPAR-IgG alone	Any co-existing antibody
Number of patients with available data	44	18	26
			1 co-existing antibody						2 or more co-existing antibodies
			10						16
Malignancy present	33 (75%)	14 (78%)	7 (70%)						12 (75%)
			mAChR-IgG	NMDA-IgG		PCA-1-IgG (anti-Yo)		gAChR-IgG
Number of patients			3	5		1		1
Malignancy present	33 (75%)	14 (78%)	2 (67%)	4 (80%)		1 (100%)		0 (0%)
Thymoma	13f	1	2f	1		0		0
Small cell lung cancer	7	3	0	1		0		0
Breast cancer	5f	4	1f	0		0		0
Ovarian cancer	3	2	0	0		1		0
Teratoma	3	1	0	2		0		0
Other	3g	3	0	0		0		0
Treatments
Steroids	32/37 (87%)	13/14 (93%)	9/10 (90%)						10/13 (77%)
IVIg	21/42 (50%)	10/19 (53%)	4/9 (44%)						7/14 (50%)
PLEX	22/39 (56%)	12/19 (63%)	5/7 (71%)						5/13 (39%)
Rituximab	14/30 (47%)	8/15 (53%)	2/5 (40%)						4/10 (40%)
Cyclophosphamide	1/28 (4%)	1/12 (8%)	0/7 (0%)						0/9 (0%)
Treatment response
None	4/29 (14%)	3/16 (19%)	0/5 (0%)						1/8 (13%)
Partial	14/29h (48%)	9/16 (56%)	2/5 (40%)						3/8 (38%)
Complete	11/29i (38%)	5/16 (31%)	2/5 (40%)						4/8 (50%)

α-amino-3-hydroxy-5-methyl-4-isoxazole propionic acid receptor (AMPAR), muscle acetylcholine receptor binding antibody (mAChR), collapsin response-mediator protein-5 antibody (CRMP5), N-methyl-d-aspartate receptor antibody (NMDA), leucine-rich glioma-inactivated 1 antibody (LGI-1), purkinje cell cytoplasmic autoantibody type 1 (PCA-1), ganglionic acetylcholine receptor (gAChR) antibody.

Cerebral spinal fluid (CSF), magnetic resonance imaging (MRI), intravenous immunoglobulin (IVIg), plasma exchange (PLEX).

^aPatient had isolated ataxia syndrome without altered mental status or seizures with “encephalitis”.

^bOne patient with MG had neuropathic pain associated, but without altered mental staus with “encephalitis”.

^cAkathisia and tics.

^dPatients had tremor, dyskinesias and parkinsonism.

^eMalignancy data available for 44 patients.

^fOne patient had breast cancer and co-existing thymoma and so is counted twice.

^gOne patient had melanoma, a second had lymphoma and prostate cancer, and a third patient had non-small cell lung cancer.

^hOne patient have received treatment for underlying malignancy without immunotherapy.

ⁱOne patient developed symptoms while on checkpoint inhibitor cancer immunotherapy which was ceased in addition to traditional immunotherapy for immunosuppressive/immunomodulating therapy.

Fig. 1.

AMPAR-IgG positive patients presented primarily with encephalitis; 39% had additional neurological symptoms and 2 patients did not have encephalitis (one with ataxia and one with myasthenia gravis and pain).

The clinical presentations seen in patients with single co-existing antibodies are summarized in Table 1. For patients with multiple co-existing antibodies they presented with: isolated encephalopathy (3), encephalopathy and MG (2), or encephalopathy with seizures, weakness, and hyperkinetic movement disorder (1) for those with co-existing mAChR-IgG, VGCC-IgG and CRMP5-IgG. Other presentations included isolated encephalopathy (1) or encephalopathy with ataxia and hyperkinetic movement disorder (1) for those with co-existing mAChR-IgG and STR-IgG; isolated encephalopathy for those with co-existing NMDAR-IgG and CASPR2-IgG (1), high titer GAD65-IgG, VGCC-IgG and GABA_BR-IgG (1), STR-IgG, ANNA1-IgG, VGCC-IgG and mAChR-IgG (1), or mAChR-IgG, CRMP5-IgG and VGCC-IgG (1); neuropathic pain and MG with co-existing mAChR-IgG and VGCC-IgG (1); encephalopathy and MG with co-existing mAChR-IgG and CRMP5-IgG (1); encephalopathy, seizures and MG with co-existing mAChR-IgG, CRMP5-IgG, VGCC-IgG, and CASPR2-IgG (1); encephalopathy, seizures and ataxia with co-existing VGCC-IgG, and high-titer GAD65-IgG (1); encephalopathy and weakness with co-existing VGCC-IgG, CRMP5-IgG and CASPR2-IgG (1); and encephalopathy, seizures and hyperkinetic movement disorder with co-existing ANNA1-IgG and VGCC-IgG (1).Two patients did not have encephalitis: one had MG and neuropathic pain with mAChR-IgG and thymoma, the other had ataxia, PCA-1-IgG and ovarian adenocarcinoma. The presence of symptoms other than encephalitis was associated with at least one co-existing antibody (14% of AMPAR-IgG positive alone had additional symptoms vs. 55% of patients with co-existing antibodies, p = 0.004).

3.2. Ancillary testing

Ancillary testing results are summarized in Table 1. CSF pleocytosis was present in 19/29 (66%) patients with available information. Most patients (24/36, 67%) had abnormal MRI with inflammatory changes (Fig. 2): typical changes of limbic encephalitis (n = 16, 2 with lesions extending into occipital regions), multifocal T2-signal abnormalities (n = 4), cerebellar T2-signal abnormalities (n = 3), and isolated basal ganglia T2-signal abnormalities (n = 1; presenting with isolated altered mental status and no co-existing antibodies). Twenty-five patients had both MRI and CSF data available, of which 4 (16%) had normal results on both. All 4 patients had an encephalitis presentation and 3 of them had accompanying cancer (2 SCLC, 1 breast cancer).

Fig. 2.

Representative MRIs of patients with AMPAR encephalitis (axial FLAIR sequences). A. Hyperintensities of the mesio-temporal lobes bilaterally with no gadolinium enhancement (not shown) B. Caudate and putamen hyperintensities as the only radiological finding in a patient with encephalitis presentation and no co-existing antibodies with no gadolinium enhancement (not shown). C–D. Multifocal abnormalities in a patient with multifocal encephalitis and seizures with no gadolinium enhancement (not shown).

3.3. Cancer association

Tumor data were available for 44 patients (Table 1). Thirty-three patients (75%) had associated tumor, most commonly thymoma (13/33, 39%) and SCLC (7/33, 21%). Patients with tumor were older (mean age 54 [range 16–81] vs. 35 [range 15–75], p = 0.004). There was only 1 malignancy (ovarian teratoma) found among the 5 patients aged 18 or less and this patient had co-existing NMDAR-IgG.

3.4. Co-existing neural autoantibodies

AMPAR-IgG was detected by CBA in serum and/or CSF in all 52 patients. Forty-three (83%) of the patients had AMPAR-IgG detected on tissue immunofluorescence (IFA) in addition to CBA; 14 in serum, 15 in CSF, and 14 in both. For details of the testing results see supplementary table 1. For the 6 patients with AMPAR-IgG detected by serum CBA alone, all but one had encephalitis (encephalitis with co-existing high titer GAD65-IgG without malignancy, 1; encephalitis with SCLC, 1; encephalitis and seizures with breast cancer, 1; MG and neuropathic pain without encephalitis with co-existing mAChR-IgG and VGCC-IgG and thymoma, 1; encephalopathy, seizures and movement disorder with co-existing ANNA1-IgG and VGCC-IgG presented and SCLC, 1; encephalitis with co-existing LGI-1-IgG and no tumor,1).

At least one co-existing antibody was found in 31/52 (60%) patients. Most common were mAChR antibodies (17/46 [40%]; binding median titer 4.36 nmol/L, range 0.7–28.70 nmol/L; modulating median titer 96% loss, range 11–100%). VGCC-IgG were detected in 11/46 (24%) patients tested (P/Q median 0.15 nmol/L [0.12–0.65 nmol/L]; N-type median 0.05 nmol/L [0.02–0.24 nmol/L]); 9 of these patients also had mAChR-IgG. CRMP5-IgG was found in serum and/or CSF in 11 (21%) patients. NMDAR-IgG was present in 6 patients: 5 in CSF and 1 in serum without CSF available (patient presenting with encephalitis). One patient was positive for LGI-1-IgG in serum (CSF not available), and 2 for CASPR2-IgG in serum, one of whom was also positive in CSF. One patient had PCA-1-IgG detected in serum in addition to AMPAR-IgG; both antibodies were detected on the tissue immunofluorescence and confirmed by western blot (PCA-1-IgG) or CBA (AMPAR-IgG). CSF was not available for this patient. Two or more co-existing antibodies were found in 18 patients (34.6%). All 7 patients with MG had co-existing mAChR-IgG and thymoma.

Co-existing antibodies did not predict the presence of tumor. However, in patients with AMPAR-IgG and tumor, when co-existing antibodies were present, the tumor was most likely thymoma (46% vs. 6%, p = 0.006). In addition, in the paraneoplastic cases, the presence of mAChR-IgG or CRMP-5-IgG predicted thymoma (60% vs. 13%, p = 0.004 and 70% vs. 19%, p = 0.005, respectively) and NMDAR-IgG predicted teratoma (40% vs. 3%, p = 0.032).

3.5. Outcome

Thirty-four of 37 patients received immunotherapy (Table 1) with response data available for 29. Most patients had partial (n = 14, 48%) or complete (n = 11, 38%) immunotherapy response. Presence of co-existing antibodies (any neural antibody, or antibody against intracellular vs. synaptic antigen) did not affect immunotherapy response. In 22 patients with follow up >3 months (average 22.6 months, range 4–77), 15 (68%) had a modified Rankin Score (mRS) ≤ 2 and 4 patients (18%) died.

4. Discussion

In this large cohort of AMPAR-IgG-associated autoimmunity, most patients presented with immunotherapy-responsive autoimmune encephalitis supported by inflammatory CSF and MRI. A small percentage of patients (4/25; 16%) had both normal CSF and imaging, of which 3 presented with isolated subacute mental status change without any other features fulfilling criteria for a diagnosis of possible autoimmune encephalitis (Graus et al., 2016) highlighting the importance of neural antibody testing to confirm the diagnosis, focus the cancer search, and guide treatment in patients with a high index of suspicion (Graus et al., 2016). Seventy-five percent of all patients had associated tumor; this would place AMPAR-IgG as a “high-risk” antibody in the adult population, according to the updated diagnostic criteria for paraneoplastic neurologic syndromes (Graus et al., 2021).

Most patients presented with encephalitis with or without other symptoms, except for two. AMPAR-IgG in patients with thymoma without encephalitis was previously described (Zekeridou et al., 2016); in our cohort, we also encountered a patient with PCA-1 autoimmunity, ataxia and ovarian adenocarcinoma. We postulate that the presence of AMPAR-IgG is reflective of the antigenic composition of the underlying malignancy (Day et al., 2014; Pittock et al., 2004). In contrast to GAD65 autoimmunity, where GAD65-IgG associates with different phenotypes (cerebellar ataxia, stiff-person spectrum disorders, epilepsy, and limbic encephalitis (Budhram et al., 2021)), AMPAR autoimmunity is associated with encephalitis and co-existing antibodies contribute to clinical symptoms other than encephalitis. Similar to what is described in neuromyelitis optica spectrum disorder and MG (Jarius et al., 2012), when MG co-exists with AMPAR autoimmunity, the MG typically predates the onset of encephalitis, sometimes by several years.

Neural co-existing antibodies occurred in 60% of patients, higher than previously described possibly due to more extensive testing (Höftberger et al., 2015). Unlike previously reported (Jia et al., 2021), the presence of co-existing antibodies did not increase the likelihood of associated tumor. Rather, the presence of co-existing antibodies in patients with tumor predicted thymoma, which is associated with multiple neural antibodies and neurological syndromes (Zekeridou et al., 2016; Vernino and Lennon, 2004). As expected, the presence of NMDAR-IgG predicted teratoma whereas mAChR and CRMP-5-IgGs predicted thymoma.

Response toimmunotherapy was present in >85% of patients, regardless of co-existing antibodies, which argues for aggressive treatment regardless of co-existing antibodies and 68% had a favorable outcome (mRS ≤ 2). Unfortunately, this retrospective study precluded evaluating long-term outcomes, best treatment practices, and relapses. Additional limitations of the study include missing malignancy data on a proportion of the cohort (potentially affecting the overall frequency of malignancy in our cohort), inconsistent investigations used for malignancy detection, limited data available for some of the patients not seen at Mayo Clinic, as well as an absence of matched serum and CSF for a proportion of the cohort.

5. Conclusions

Our study enhances understanding of AMPAR encephalitis, which is a treatable condition, and elucidates the role of co-existing neural autoantibodies in defining clinical presentation and predicting the associated tumor found in paraneoplastic subtypes of the disease.

Data availability

Data will be made available on request.