Burden of tau pathology

Among the 22 cases, we observed a variable burden of the tau pathology along the tegmentum of the brainstem, the spinal cord, the hypothalamus/basal forebrain and the hippocampus—regions described in the original neuropathological criteria as being typically affected by the disease. It ranged from nearly absent or minimal/mild in form of isolated threads only visible at high magnification (× 400) (7 cases; 32%), moderate with a higher density of neuropil threads already identified at lower magnification (× 100) and the presence of a low density of pretangles and/or neurofibrillary tangles (4 cases; 18%), to prominent with a high density of neuropil threads, pretangles and/or tangles (15 cases; 50%). Details of the burden and topographical distribution of the tau pathology in the individual anatomical regions are shown in Figs. 1, 2 and Table 2. The neurodegenerative changes involving the brainstem tegmentum were variable and increased with increasing tau burden, but were observed in all cases, including those with absent or mild tau pathology (Figs. 1, 3).

Fig. 2

Left: graphical neuroanatomical representation of the main focus of anti-IgLON5 disease neuropathology and possible related clinical symptoms. Right: grouped analysis and heat map; source of variations: tau burden (0 = absent or mild; 1 = moderate or prominent) and neurodegeneration (mild/moderate vs severe) (two-way ANOVA, source of variation: anatomical region: p = 0.0004; degree of neurodegeneration: p < 0.0001)

Table 2 Heatmap of the tau and TDP-43 pathology load in different neuroanatomical regions for each caseFig. 3

Pathology burden. Graphical representation of the severity of neurodegeneration and tau burden in the brainstem regions observed at the different stages of pathology: midbrain (leftmost), pons (left middle), medulla oblongata (right middle), and upper cervical cord (rightmost), and their correlation with age at onset and duration of disease on the left. Color codes: upper line: the focus of neurodegeneration in the brainstem is represented as a pale-red area with minimal or even absent tau pathology; middle line: the mild-to-moderate deposition of tau is represented in softer red; lower line: prominent tau pathology is represented in intense red and areas with milder or occasional pathology in form of pale-red dots (s. nigra midbrain, pontine base, inferior olives). Figure adapted from the original criteria [23]. Stage and pathology burden. Stage 1 (upper row): mild/moderate neuronal loss/gliosis and isolated tau-positive neuropil threads in the lateral segments of the midbrain and pontine tegmentum and/or reticular formation/respiratory nuclei of the medulla oblongata and/or hypothalamic nuclei/preoptic area/pituitary stalk. Stage 2 (middle row): moderate neuronal loss/gliosis and moderate tau-positive neuropil threads, pretangles and NFT in the previous regions extending to adjacent tegmental nuclei and hypothalamic /prehypothalamic nuclei. Stage 3 (lower row): Prominent neuronal loss/gliosis and extensive tau pathology in the aforementioned regions + mild involvement of s. nigra, inferior olives, dentate nucleus of the cerebellum and glomerula of the granule cell layer of the cerebellar cortex

Identification of different stages of pathology

Detailed topographical mapping and grouped analysis showed a progressive increase in the tau pathology load along the brainstem and hypothalamic regions (Fig. 2, Table 2). Based on these findings, and the experimental data supporting that the tauopathy is a secondary event of the autoimmune disorder [33, 34, 48], we suggest a classification into three pathological stages of the anti-IgLON5 disease-related tauopathy. These stages showed a positive correlation with disease duration and an inverse relationship with age at onset (see below).

(I) Stage 1: brainstem degeneration without overt or minimal tau pathology

Stage 1 was observed in 7/22 (32%) patients. This stage is characterized by a mild-to-moderate gliosis and microglial activation in the anterior hypothalamus and brainstem tegmentum, particularly in the periaqueductal grey matter, lateral tegmentum, reticular formation and respiratory nuclei of the medulla oblongata (solitary nucleus and n. ambiguus). Enlarged neurons may be observed in the reticular formation (Fig. 1e1). Tau pathology may be absent or is minimal and presents in form of isolated tau-positive cell processes/neuropil threads in these regions (Fig. 1d1). The infundibulum may show mild-to-moderate tau-positive threads (suppl. Fig. 1). In some patients, the hippocampus may be affected by tau pathology in a type and distribution similar to PART, but with more frequent involvement of the CA2 sector and the dentate gyrus. The cortical areas, striatum, lenticular nucleus, n. subthalamicus, thalamus and cerebellum are devoid of tau pathology.

(II) Stage 2: brainstem degeneration with mild/moderate tau pathology

This grade of pathology was observed in 4/22 cases (18%). In this stage, gliosis and microglial activation in the anterior hypothalamus and prehypothalamic region, as well in the brainstem tegmentum are more prominent than in the previous stage, as well as the burden of the tau pathology. It involves additional regions of the basal forebrain, particularly the substantia innominata, and of the tegmentum of the brainstem, including the laterodorsal tegmental region, pedunculopontine nuclei, and locus coeruleus of the pons, and the pars reticulata of the substantia nigra in the midbrain. The medulla oblongata shows more prominent tau pathology in the aforementioned regions, in addition to a moderate involvement of the dorsal raphe and of the dorsal motor nucleus of the vagal nerve and the motor nucleus of the hypoglossus. In addition to clusters of neuropil threads, a low amount of pretangles and neurofibrillary tangles may be already observed (Fig. 1d2, f2). These may be also identified in the dorsal and anterior horns of the spinal cord (suppl. Fig. 2). In the cerebellar cortex a fine granular synaptic immunoreactivity pattern may be detected in the synaptic glomerula of the granule cell layer (suppl. Fig. 1). The dentate nucleus is usually not or only minimally involved. The cortical areas and the striatum are practically devoid of tau pathology, while the thalamus, n. subthalamicus and zona incerta may show mild tau-positive threads and isolated pretangles and/or tangles. The pars reticulata more than the pars compacta of the substantia nigra show at the most a moderate tau pathology.

(III) Stage 3: brainstem degeneration with prominent tau pathology—“classical” pattern

Stage 3 of prominent tau burden was observed in 11/22 cases (50%), and included the six original cases on which initial neuropathological criteria were based on and five additional cases. The pattern was very homogeneous among these cases and was readily identifiable.

At this stage of pathology, all areas reported in the original description of the anti-IgLON5 disease-related tauopathy are severely affected and show high burden of tau pathology (Figs. 1b3, d3, f3, 3, Table 2). The key regions include the prehypothalamic region, anterior hypothalamic nuclei, substantia innominata, n. basalis Meynert (moderate involvement), zona incerta, periaqueductal grey matter, laterodorsal tegmental area, pedunculopontine nuclei, locus coeruleus, raphe, dorsal motor nucleus of the vagal nerve, n. hypoglossus, and reticular formation. In some cases, the pontine base may show a variable amount of pretangle and/or tangles. The inferior olives may show single neurofibrillary tangles and perineuronal neuritic clusters. The anterior and posterior horns of the spinal cord show variable amounts of tau pathology (suppl. Fig. 2). In the cerebellum, tau positivity in the synaptic glomerula of the granule cell layer is usually focal, single Cajal cells may show a cytoplasmic immunoreactivity, while Purkinje cells and their apical dendrites remain mostly negative (suppl. Fig. 1). The dentate nucleus shows single tangles/pretangles/threads. The cortical areas remain devoid of pathology, while the striatum and thalamus may show at most single pretangles and/or tangles and/or neuropil threads. At this stage, the tau pathology in the hippocampus is variable but appears more extensive than in stage 1 and stage 2 in some cases, and involves more neurons of the CA1 sector and the ento- and transentorhinal regions. Although the involvement of the dentate gyrus and the CA2 sector is frequently observed in the anti-IgLON5 disease-related tauopathy, a clear delineation from PART pathology may be not possible and the presence of associated pathologies, like argyrophilic grain disease, pose a difficulty in the interpretation of hippocampal pathology.

Distribution of tau isoforms

In eight cases both, 3R and 4R tau isoforms contributed to the tau inclusions in the brainstem, which were mainly neuronal but could also include some oligodendroglial coiled-body type inclusions in the brainstem [23] and some ramified astrocytes in the limbic system. In three cases, however, there was a predominance of 4R tau isoforms. One case was retrospectively identified by reviewing the clinical information of a patient who had died as a result of a PSP-like syndrome with atypical features that were suggestive of anti-IgLON5 disease. Archived CSF confirmed the presence of anti-IgLON5 antibodies and the neuropathological examination showed an extensive brainstem tauopathy, as expected for the anti-IgLON5 disease-related tauopathy, which, however, formally fulfilled the neuropathological criteria of a PSP phenotype [8]. This included the presence of > 1 tufted astrocytes in the perirolandic region or the putamen, in this case in the putamen, and > 1 4R tau-positive neurofibrillary tangles in the globus pallidus, n. subthalamicus and s. nigra (two of the three regions), in addition to frequent coiled bodies, as reported [8]. Nevertheless, there was a clear brainstem predominance of the pathology which involved also the hypothalamic and brainstem nuclei affected in the other cases of anti-IgLON5 disease-related tauopathy at stage 3 with a mixture of 3R + 4R tau isoforms. This case also showed focal tau positivity in the synaptic glomerula of the cerebellar granule cell layer. Another case with a predominance of 4R tau isoforms was included in the very first description of the disease [49]. Finally, the third case (#8, clinical features consistent with a PSP-like phenotype and positive anti-IgLON5 antibodies), presented typical distribution of the anti-IgLON5 disease-related tau pathology with mainly neuronal tau pathology, which was, however, dominated by 4R tau isoforms, and showed no astrocytic tau pathology definitory of any other known 4R tauopathy. Interestingly, this case had a severe nigral degeneration with disproportionally low (mild-moderate) tau pathology in this region, and showed also some involvement of the pontine base and the dentate nucleus of the cerebellum, as can be also observed in PSP. However, there was no involvement of the basal ganglia or cortical regions. In the pallidum, an area of necrosis/infarct in the context of extensive small vessel disease was also observed.

A graphical summary of the main findings in the 22 patients is represented in Fig. 4.

Fig. 4

Summary of the neuropathological findings in 22 patients with anti-IgLON5 disease. Of these, 68% showed a brainstem tauopathy, while 32% did not. The latter group was represented by older patients with shorter disease duration, compared to the patients with brainstem tauopathy, who were younger at disease onset and had longer disease duration

Concomitant pathologies

(I) TDP-43 pathology in the brainstem and spinal cord

TDP-43 pathology in the brainstem and spinal cord was observed in 7/22 cases (32%): 5 had associated brainstem tau pathology (2 at stage 2 and 3 at stage 3) and 2 had minimal or no tau pathology (at stage 1) (suppl. Fig. 2). TDP-43 inclusions were most frequently identified in motor and non-motor neurons of the brainstem and spinal cord and/or in areas affected by tau pathology, i.e., the tegmentum medullaris, including reticular formation and the anterior horns of the spinal cord (suppl material in Ref. [49]). The morphology of the aggregates in motor neurons included skein-like inclusions and a fine granular diffuse cytoplasmic staining (suppl. Fig. 2j, k). Single Bunina bodies were also identified in motor neurons on HE-stained sections in single cases (e.g., case #4, case #20) (suppl. Fig. 2f), but no pale or Lewy-like inclusions were identified. Moreover, axonal spheroids in the anterior horn of the spinal cord were present in single cases with prominent motor neuron involvement (case #18) (suppl. Fig. 2d). In addition, few coiled-body- like TDP-43 inclusions were observed in oligodendrocytes (supp. Fig. 2l). Compact neuronal inclusions were also detected in the reticular formation. Minimal signs of corticospinal degeneration were observed in single cases (case #20), but without obvious TDP-43 inclusions in upper motor neurons. Neurons of the 3rd and 6th cranial nerves were also involved.

(II) Other associated pathologies

Other associated pathologies were identified in 74% of cases, including common findings in apparently neurologically unimpaired aged persons, like low to moderate density of ßA4-amyloid deposits in form of diffuse and/or compact plaques, few alpha-synuclein aggregates, limbic TDP-43 pathology, argyrophilic grain pathology, mild-to-moderate ento/transentorhinal and hippocampal tau pathology, and vascular lesions. Co-pathologies were detected in both, patients with and without the brainstem tauopathy (Table 1). A single case had, in addition to the anti-IgLON5 disease-related brainstem tauopathy and TDP-43 aggregates in brainstem motor neurons as also seen in ALS, frequent intranuclear hyaline inclusions in an extent and distribution that were consistent with neuronal intranuclear hyaline inclusion body disease (NIHID) [25]; genetic testing to exclude fragile-X tremor ataxia syndrome or NOTCH2NLC gene alterations for NIHID was not possible. In two elderly patients, we found concomitant ßA4-amyloid pathology, and aging-related tau astrogliopathy (ARTAG), respectively; in two other cases, we identified incidental Lewy body pathology [8]; and in further two patients, we detected argyrophilic grain pathology. No FUS aggregates were detected in any of the cases. Small vessel disease and an old lacunar infarction was identified in the basal ganglia of the patient in which the brainstem tauopathy was dominated by 4R tau isoforms. This patient had in addition neuropathological changes consistent with argyrophilic grain disease (AGD) and limbic age-related TDP-43 encephalopathy (LATE) associated with incipient hippocampal sclerosis (patient #8, Table 1).

In those cases, in which other additional/co-existing pathologies are found independently of a “brainstem/motor neuron predominant TDP-43 pathology”, we suggest reporting them preliminarily as additional pathologies, as long as evidence for a strong association with anti-IgLON5 disease is lacking (and co-incidentality, age-related changes, vascular disease, or other mechanisms cannot be excluded).

Clinicopathological correlations (Table 3, Figs. 5, 6, suppl. Fig. 3–6)

Clinical features and their corresponding pathological stages (stages 1 to 3) for each patient are detailed in Tables 1 and 3. The patients with stage 3 pathology and severe neurodegeneration were younger at disease onset (48–77 years; median 61 years) and had a longer disease duration (median 9 years) compared to patients at stage 1 and stage 2, who were older (69–85 years; median 79 years) and had a shorter disease duration (median < 1 year) (Tables 1 and 3; Figs. 5, 6, suppl. Fig. 3). Clinical symptoms at onset were similar at stage 1, stage 2 and stage 3 with the exception of sleep disorder that was more common in patients with moderate/severe tauopathy (Fig. 5C). Correlations with other symptoms are shown in suppl. Fig. 5. Patients at stage 3 developed significantly more oculomotor abnormalities during the course of the disease (Table 1, suppl. Table 2). Concerning cognition, despite the significantly longer duration of disease, only 3/11 patients at stage 3 pathology showed at most mild cognitive impairment, whereas 4/11 patients at stage 1 or 2 developed pronounced cognitive impairment during the course of the disease. However, these patients had slightly more PART-like tau pathology in the hippocampus/limbic system (neurofibrillary stage III) than the younger patients in stage 3 (neurofibrillary stage II) and more widespread ßA4 amyloid deposits.

Table 3 Summary of demographic and clinical features in patients of different disease stages of anti-IgLON5 pathologyFig. 5

Correlation analysis between the different pathology stages and A age of onset and B disease duration; and C between the presence of sleep disorder and degree of neurodegeneration (A, B one-way ANOVA with Tukey’s multiple comparisons test, C Fisher’s exact test)

Fig. 6

Potential relationship between grades of severity of neurodegeneration of the brainstem (mild—severe), tau accumulation (few—prominent), age at onset (older—younger) and rate of disease progression (fast – slow)

Consistent with the anatomical distribution of the neuropathological findings with prominent involvement of the anterior horns of the spinal cord, the patient at “stage 1 and brainstem ± spinal TDP-43 pathology” (patient #6, Tables 1, 2) and the patient with “stage 2 without TDP-43 pathology” (patient #11) showed, in addition to bulbar dysfunction, the clinical picture of motor neuron disease with marked muscle atrophy, paresis, and fasciculations. These symptoms were not clinically evident in the other patients with brainstem ± spinal TDP-43 co-pathology associated with tau lesions at stage 3. A higher female to male ratio was observed in the TDP-43 positive cases (4:2) compared to the TDP-43 negative group (3:13). Beyond this, the patient groups did not differ in terms of their age at onset (median 69.5 years [TDP-43] vs. 67 years [non-TDP-43]) or the distribution of initial clinical symptoms.

The risk HLA haplotype HLA-DQB1*05:01 was equally represented in cases with mild/moderate and severe neurodegeneration, whereas the HLA-DRB1*10:01 allele was more frequently represented in cases with mild/moderate neurodegeneration (suppl. Fig. 5C). No statistical significance was reached when comparing its distribution between stages, age at onset or the presence of a PSP phenotype (suppl. Fig. 4). The limited availability of other laboratory parameters like CSF pleocytosis and anti-IgLON5 antibody titers in CSF and/or serum prevented us to draw firm conclusions (suppl. Fig. 5C).

Applicability of the original neuropathological criteria of the anti-IgLON5 disease-related tauopathy and proposed update

Of the 22 cases, 15 (68%) fulfilled the original neuropathological research criteria of a brainstem-predominant tauopathy (11 definite, 4 probable). In 12/15 (80%), the tau pathology consisted of a mixture of 3R and 4R tau isoforms, whereas in the other three cases, 4R tau isoforms predominated. In contrast, 7/22 cases (32%) did not meet the neuropathological criteria of the anti-IgLON5 disease-related tauopathy due to the lack of overt/prominent brainstem tau pathology (Fig. 4), although patients had the clinical features of anti-IgLON5 disease and had anti-IgLON5 antibodies.

In view of these findings and the spectrum of pathologies observed in anti-IgLON5 disease in an extended autopsy series, and keeping in mind that the disease as such is defined by (1) the presence of anti-IgLON5 antibodies in the CSF and/or serum, and (2) the presence of neurological symptoms reflecting a predominantly brainstem involvement, we propose adapting the original research criteria that defined the unique tauopathy associated with the disease [23] accordingly (Table 4), and to delete the diagnostic category “possible”.

Table 4 Updated neuropathological criteria to define the tauopathy associated with anti-IgLON5 disease

As it is still unclear whether the presence of a 4R tau predominant tauopathy is a separate subtype or an earlier/later stage of the disease, we recommend that the diagnosis should be only established in the presence of anti-IgLON5 antibodies in the CSF and the presence of brainstem symptoms including a sleep disorder and/or a supportive HLA genetic background. The same holds for cases where neuropathological findings are those of stage 1, with no overt or only mild tau pathology in the brainstem, which may also be identified in individuals without symptoms of a neurodegenerative process [9, 10]. In this situation, as well as in cases with TDP-43 proteinopathy “only”, we also recommend establishing the diagnosis in an integrated manner, i.e., in the presence of anti-IgLON5 antibodies in the CSF and the presence of brainstem symptoms including a sleep disorder and/or a supportive HLA genetic background. Conversely, as autoantibodies at certain titers may be rarely discovered as an incidental finding of a panel workup of individuals without clinical features consistent with the disease, as well as the possibility of false-positive autoantibody testing that occurs in any clinical testing scheme [2], the diagnosis of anti-IgLON5 disease should be only established in the presence of CSF antibodies and clinical symptoms.

Due to the consistent involvement of the brainstem and the hypothalamus as observed in the pathology burden map, we suggest that the neuropathological assessment could be performed in a minimal set of brain regions that represent the “high interest areas” of the disease. These areas partly overlap with other samplings protocols, e.g., as suggested for PSP [47] and should be easily applicable by other investigators. They should include at least the perirolandic region, striatum and/or lenticular nucleus, hypothalamus, hippocampus, midbrain, pons, medulla oblongata, and cerebellum. If a broad sampling is locally possible, it should include further brain regions (see “Materials and methods”), and the spinal cord.