Hypothalamic Immunopathology in Anti-Ma–Associated Diencephalitis With Narcolepsy-Cataplexy | Allergy and Clinical Immunology | JAMA Neurology | JAMA Network
[Skip to Navigation]
Sign In
Case Report/Case Series
October 2013

Hypothalamic Immunopathology in Anti-Ma–Associated Diencephalitis With Narcolepsy-Cataplexy

Author Affiliations
  • 1National Reference Network for Orphan Diseases (Narcolepsy, Hypersomnia, Kleine-Levin Syndrome), Sleep Unit, Department of Neurology, Gui de Chauliac Hospital, CHU Montpellier, France
  • 2INSERM U1061, Montpellier, France
  • 3Department of Neurology, Gui de Chauliac Hospital, CHU Montpellier, France
  • 4Center for Brain Research, Medical University of Vienna, Wien, Austria
  • 5Department of Pathology, CHU, Gui de Chauliac Hospital Montpellier, France
  • 6Immunology Department, CHU Nîmes, France
  • 7INSERM U1043, CNRS, UMR 5282, Toulouse, France
  • 8INSERM U1028, CNRS, UMR 5292, Center for Research in Neuroscience of Lyon, Team SLEEP, Lyon, France
JAMA Neurol. 2013;70(10):1305-1310. doi:10.1001/jamaneurol.2013.2831

Importance  Idiopathic narcolepsy with cataplexy is thought to be an autoimmune disorder targeting hypothalamic hypocretin neurons. Symptomatic narcolepsy with low hypocretin level has been described in Ma antibody–associated encephalitis; however, the mechanisms underlying such an association remain unknown.

Observations  We described a 63-year-old man with clinical criteria for diencephalic encephalitis with sleepiness, cataplexy, hypocretin deficiency, and central hypothyroidism, together with brainstem encephalitis reflected by supranuclear ophtalmoparesis and rapid eye movement sleep behavior disorder with underlying abnormalities on brain magnetic resonance imaging. An autoimmune process was demonstrated by the detection of antibodies against Ma protein. Death occurred 4 months after disease onset without any tumor detected. Neuropathology, immunohistochemistry, and immunoreactivity results were compared with those obtained in idiopathic narcolepsy-cataplexy and with normal control brains. The principal findings revealed almost exclusive inflammation and tissue injury in the hypothalamus. The type of inflammatory reaction suggests cytotoxic CD8+ T lymphocytes being responsible for the induction of tissue injury. Inflammation was associated with complete loss of hypocretinergic neurons. Autoantibodies of the patient predominantly stained neurons in the hypothalamus and could be absorbed with Ma2.

Conclusions and Relevance  The encephalitic process, responsible for narcolepsy-cataplexy and hypocretin deficiency, reflects a CD8+ inflammatory-mediated response against hypocretin neurons.

Idiopathic narcolepsy with cataplexy (NC) is thought to be an autoimmune disorder targeting hypothalamic neurons that produce hypocretin/orexin, as reflected by a low cerebrospinal fluid (CSF) hypocretin level.1 Symptomatic NC and low hypocretin level have been described in Ma antibody–associated encephalitis, an infrequent paraneoplastic autoimmune disorder.2-7 Ma1 and Ma2 are intracellular proteins expressed in the testes and throughout the brain, in particular in the brainstem, hypothalamus, limbic structures, and cerebellar nuclei.8,9 The mechanisms underlying such an association remain unclear and, to our knowledge, no neuropathological and immunohistochemical analyses have been performed in this rare condition.

Here, we report the clinical and neuropathological observations of a patient with NC and Ma antibody–associated encephalitis with rapidly fatal outcome. Results were compared with idiopathic NC and with control brains.

Report of a Case

A 63-year-old man without significant medical history presented with a 2-month duration of severe daytime sleepiness characterized by 10 short episodes of daytime naps per day. Typical and severe cataplexy occurred abruptly 1 month later together with hypnagogic hallucinations, rapid eye movement (REM) sleep behavior disorder with nocturnal agitation and frequent sleeptalking, dream-enacting behavior, and diplopia. He had a 15-pack-year history of smoking with rare alcohol intake. Clinical examination found a vertical supranuclear gaze palsy and bilateral ptosis, without any sign of dementia or parkinsonism. Polysomnography results demonstrated fragmented and reduced sleep efficiency with sustained muscle activity in REM sleep, with 2 sleep-onset REM periods and mean sleep latency at 7.2 minutes. A brain magnetic resonance imaging study revealed bilateral paramedian fluid-attenuated inversion recovery hyperintensities in the thalamus and hypothalamus without enhancement after contrast administration (Figure 1).

Figure 1.  Brain Magnetic Resonance Imaging/Fluid-Attenuated Inversion Recovery Sequences
Brain Magnetic Resonance Imaging/Fluid-Attenuated Inversion Recovery Sequences

Images revealed bilateral paramedian hyperintensities in the thalamus, hypothalamus, and mammillary bodies, with lesions surrounding the third ventricle without enhancement after contrast administration.

Cerebrospinal fluid protein (0.34 g/L), IgG, and leukocyte levels were normal, and no oligoclonal bands were detected. Cerebrospinal fluid hypocretin-1 was undetectable. A central hypothyroidism was diagnosed together with low levels of testosterone but normal cortisol, prolactine, and luteinizing and follicle-stimulating hormone levels. A commercial immunoblot with recombinant antigens (Ravo Diagnostika) used to detect paraneoplastic autoantibodies revealed strong reactivity for Ma2 and weak reactivity against Ma1. Serum level results of tumor markers were normal. Testis echography and thoracoabdominal-pelvic computed tomography scan findings were normal, except for few mediastinal adenopathies. A body fluorodeoxyglucose positron emission tomography scan showed slight hypermetabolism in the cecum, mediastinal lymph nodes, and hypothalamus. A coloscanner showed normal results. Treatment for narcolepsy was started with venlafaxine and modafinil, with efficacy on sleepiness and cataplexy, along with levothyroxine.

A repeat evaluation for an underlying malignancy was negative. His neurological state deteriorated progressively, with severe cataplexy, frequent falls, and confusion; death occurred 4 months after disease onset owing to an unexpected infectious pneumonia potentially related to severe impairment of vigilance. Permission for brain and general autopsy was granted.

Two control patients were included: a woman aged 92 years affected with a long history (56 years) of typical idiopathic NC (undetectable CSF hypocretin-1 level) with CSF and serum samples stored at −80°C until use and a woman aged 39 years without neurological disease with cervix carcinoma as the cause of death. This study was approved by the local institutional review board.

Neuropathology and Immunohistochemistry

Multiple blocks from different brain regions were routinely embedded in paraffin. Paraffin sections (5-mm thickness) were stained with hematoxylin and eosin, Luxol fast blue for myelin, and Bielschowsky silver impregnation for axons. Immunocytochemistry was performed with a biotin avidin peroxidase technique to reveal the following primary antibodies: CD3, CD8, granzyme B, CD20, CD68, major histocompatibility complex (MHC) class I, HLA-D, human immunoglobulins, complement C9neo antigen, activated caspase 3, glial fibrillary acidic protein, aquaporin 4, and orexin/hypocretin. Biotinylated secondary antimouse and antirabbit antibodies (Jackson ImmunoResearch) and avidine peroxidase complex (Sigma) were used as secondary reagents. Peroxidase activity was visualized with diaminobenzidine.

Confocal Laser Fluorescence Microscopy

Fluorescence immunohistochemistry was performed on paraffin sections as described for light microscopy with few modifications. For confocal fluorescent double labeling, sections were incubated with patient serum and rat antihypocretin antibodies overnight at 4°C. After washing with phosphate-buffered saline, secondary antibodies consisting of Cy2-conjugated goat antihuman (Jackson ImmunoResearch; 1:200) and biotinylated antirabbit (Jackson ImmunoResearch) were applied simultaneously for 1 hour at room temperature. The staining was finished by application of streptavidin-Cy3 (Jackson ImmunoResearch; 1:75) for 1 hour at room temperature. Sections were embedded and examined using a confocal laser scan microscope (Leica SP5, Leica). Scanning for Cy2 (488 nm) and Cy3 (543 nm) was performed sequentially to rule out fluorescence bleed-through.

Quantification of Inflammation in Different Brain Regions

Perivascular inflammatory infiltrates were counted in 10 microscopic fields (magnification × 10; field size, 1.06 mm2) in each brain region in hematoxylin– and eosin–stained sections. Within the hypothalamic lesions, we also quantified different types of inflammatory cells. This was done in 10 high-power fields of 60 mm2 each.

Immunoreactivity of Patient Serum and CSF and Absorption With Ma2 Antigen

Binding of the patient’s serum and CSF immunoglobulins to brain tissue was analyzed in paraffin sections from different brain regions of the control case. Sections were deparaffinized and nonspecific binding of immunoglobulins to brain tissue was blocked by preincubation of the sections with 10% fetal calf serum. Then the sections were incubated with patient serum (1:10, 1:100, 1:1000, and 1:10000) or patient CSF (1:2, 1:10, and 1:50). For control, we used the CSF of the patient with idiopathic NC at the same dilutions. Binding of the patient’s immunoglobulin was visualized with a biotin avidin peroxidase technique, as just described, using biotinylated antihuman immunoglobulin (Amersham).

Absorption of the patient serum was performed by preincubation of the serum at a dilution of 1:1000 with recombinant Ma2 (Norvus Biologicals LLC) at 10 µg/mL for 1 hour at 37°C. After this, sections were incubated with the preincubated or nonincubated patient serum for 1 hour at 37°C. Sections were washed with phosphate-buffered saline and bound primary antibodies were detected by biotinylated antihuman Ig (Jackson) followed by avidin-peroxidase and detection with diaminobenzidine as substrate.


Our patient fulfilled the clinical criteria for diencephalic encephalitis with sleepiness, cataplexy, hypocretin deficiency, and central hypothyroidism, together with brainstem encephalitis reflected by supranuclear ophtalmoparesis and REM sleep behavior disorder with underlying abnormalities on brain magnetic resonance imaging (Figure 1). An autoimmune process was demonstrated by the detection of antibodies against Ma protein. An autopsy was performed, showing infectious mediastinal adenopathies but no visceral cancer.

The hallmark of neuropathology in this case was a profound inflammation in the gray matter, especially in the hypothalamus, reflected by numerous perivascular inflammatory cuffs (Figure 2A), consisting of CD3+ and CD8+ T cells (Figure 2B and C). CD8+ T cells were diffusely infiltrating the neural parenchyma (Figure 2C), associated with intense upregulation of MHC class I molecules (Figure 2D), predominantly on CD68+ macrophages and activated microglia (Figure 2F). Major histocompatibility complex class I immunoreactivity was also seen on the surface of some neurons (Figure 2D, insert). Major histocompatibility complex class II expression was low and mainly seen within perivascular inflammatory cuffs (Figure 2E). In addition to MHC class I–restricted CD8+ T cells, B cells were also present but restricted to perivascular inflammatory infiltrates (Figure 2G). This inflammatory process was associated with loss of neurons, but no direct attachment of CD8+ or granzyme B+ T cells to neurons was observed and caspase 3 expression or apoptotic changes of neurons were absent (data not shown). A profound astrocytic gliosis was found in the hypothalamus (Figure 2H), with increased aquaporin 4 expression around the inflamed vessels (Figure 2I). No deposition of immunoglobulins or activated complement (C9neo) was seen. The composition of inflammatory infiltrates in the hypothalamus revealed 130 CD3+, 115 CD8+, 22 granzyme B+, 28 CD20+, 580 CD68+, and 160 MHC class II+ cells per square millimeter. Brain inflammation was particularly intense in the hypothalamus, with few other inflammatory infiltrates in the substantia nigra and the thalamus, but was almost absent in other brain regions (cortex, hippocampus, basal ganglia, cerebellar cortex, dentate nucleus, and white matter). The clinical presentation of our case suggested that hypocretin neurons had preferentially been destroyed. Thus, we analyzed hypocretin expression in the hypothalamus by immunohistochemistry in contrast to controls (Figure 2K and L) and no hypocretin-positive neurons were present in the inflamed hypothalamus of our case (Figure 2J).

Figure 2.  Pathological Alterations in the Hypothalamus of the Ma-Positive Patient
Pathological Alterations in the Hypothalamus of the Ma-Positive Patient

A, Image shows the basic pathology in the hypothalamus, consisting of inflammation and loss of neurons in a section stained with hematoxylin and eosin (H&E). Immunocytochemistry was performed on serial sections stained with the markers indicated on the upper right corner. There is a profound inflammation with perivascular inflammatory infiltrates, consisting of T cells (B and C), macrophages (F), and B cells (G). CD8+ T cells are not only present in the perivascular space but also diffusely dispersed in the tissue (C; the inserts show parenchymal T cells at high magnification). Inflammation is associated with the expression of major histocompatibility antigens (MHC) class I (D) and class II (E). Some neurons express MHC class I antigen on their surface (D, insert). Within the lesions, there is profound astrocytic gliosis (H) and increased expression of aquaporin 4 (AQP4) around inflamed vessels (I). No single hypocretin-positive neuron or axon was found in sections from our patient (J, original magnification ×400). K and l, Images show expression of hypocretin in neurons and axons in the hypothalamus of a control patient at different magnifications. GFAP indicates glial fibrillary acidic protein. For panels A-K, original magnification ×50; for panel L, original magnification ×400.

We further analyzed the binding of serum and CSF immunoglobulins from our case to normal control brain tissue. An intense intracytoplasmatic labeling (leaving nuclei unstained) was seen in hypothalamic neurons even at very high serum dilutions (Figure 3A-C). Double staining with confocal laser microscopy showed that some patient serum–reactive hypothalamic neurons also expressed hypocretin (Figure 3F and G). The control CSF from the patient with idiopathic NC did not label neurons in the hypothalamus (Figure 3D). Interestingly, no serum or CSF immunoreactivity was detected on sections of the patient’s inflamed hypothalamus (Figure 3E). In the control brain, neurons reactive with the patient’s serum or CSF were most numerous in the hypothalamus (Figure 3H). A lower number of immunoreactive neurons was seen in the substantia nigra and the subependymal nuclei of the medulla oblongata, without any reactivity present in the cortex, hippocampus, caudate nucleus, putamen, pallidum, thalamus, mammillary bodies, or white matter. Preabsorption of the patient’s serum with recombinant Ma2 protein completely abolished the neuronal immunoreactivity in the hypothalamus (Figure 3I).

Figure 3.  Immunoreactivity of the Ma-Positive Patient’s Serum and Cerebrospinal Fluid (CSF) With Neurons in Normal Hypothalamus
Immunoreactivity of the Ma-Positive Patient’s Serum and Cerebrospinal Fluid (CSF) With Neurons in Normal Hypothalamus

Multiple neurons in the normal hypothalamus are labeled with the Ma-positive patient’s serum (A and B) and CSF (C), while the CSF from a patient with idiopathic hypocretin-deficient narcolepsy-cataplexy showed no reactivity (D). No neurons with reactivity for the Ma-positive patient’s serum were detected in the inflamed hypothalamus of the Ma-positive patient (E). F and G, Double immunohistochemical staining with confocal laser microscopy shows that a fraction of the Ma-positive patient’s serum reactive neurons (red) also express hypocretin (green). Absorption of the patient serum with recombinant Ma2 completely abolishes neuronal immunoreactivity in the hypothalamus of the control patient (I) in comparison with nonabsorbed serum (H). For panel A, original magnification ×100; panels B-E, original magnification ×400; F and G, original magnification ×800; and H and I, original magnification ×200.


We report the first neuropathological findings in a patient with well-defined NC and anti-Ma antibody–associated encephalitis leading to death 4 months after clinical onset. Even if no tumor was detected at pre and post mortem levels, a paraneoplastic neurological syndrome remained a probable hypothesis because symptoms may precede tumor discovery by several years.

Severe neuronal loss and gliosis with extensive CD8+ T-cell perivascular and parenchymal inflammatory infiltrates were found in our case, particularly within the hypothalamus. This result contrasts with most reports on anti-Ma antibody–associated encephalitis describing widely distributed lesions.2,3,8-10 Few reports have identified an association between anti-Ma2 encephalitis and NC.4-7 Ma2 autoimmunity may be responsible for the pathogenesis of the lesions, as shown by predominant intracytoplasmatic serum immunoreactivity in hypothalamic neurons including hypocretin neurons. Immunoreactivity in our patient was fully abolished by preabsorption of the serum with Ma2. However, unexpectedly, the binding pattern of the patient’s serum in the central nervous system differed from that typical for the distribution of Ma2 protein, being nearly exclusively seen in the hypothalamus. This discrepancy is currently unresolved and requires future in-depth characterization of antigen and epitope binding of the patient’s serum autoantibodies.

In paraneoplastic diseases with immune reactions against intracellular antigens, the autoantibodies are good disease markers but are unlikely to be directly involved in neuronal killing. In our case, the target antigen could not be hypocretin per se, but hypocretin neurons apparently contain an antigen, specifically recognized by a T-cell–mediated autoimmune response. Despite only 4 months between disease onset and death, the lesions were apparently in an advanced stage with loss of all hypothalamic Ma+ neurons, being too late to detect the acute neuron destruction but early enough to identify the nature and composition of the inflammatory infiltrate.

To conclude, the encephalitis process, responsible for NC and hypocretin deficiency as one of the specific targets, reflected a CD8+ inflammatory-mediated response against hypocretin neurons.

Back to top
Article Information

Corresponding Author: Yves Dauvilliers, MD, PhD, Service de Neurologie, Hôpital Gui-de-Chauliac, 80 Ave Augustin Fliche, 34295 Montpellier Cedex 5, France (ydauvilliers@yahoo.fr).

Accepted for Publication: March 4, 2013.

Published Online: August 12, 2013. doi:10.1001/jamaneurol.2013.2831.

Author Contributions:Study concept and design: Dauvilliers, Liblau, Peyron, Lassmann.

Acquisition of data: Dauvilliers, Bauer, Rigau.

Analysis and interpretation of data: Dauvilliers, Bauer, Rigau, Lalloyer, Labauge, Carlander, Lassmann.

Drafting of the manuscript: Dauvilliers, Bauer.

Critical revision of the manuscript for important intellectual content: Dauvilliers, Rigau, Lalloyer, Labauge, Carlander, Liblau, Peyron, Lassmann.

Obtained funding: Dauvilliers.

Administrative, technical, and material support: Dauvilliers, Bauer, Rigau, Lalloyer, Peyron.

Study supervision: Dauvilliers, Labauge.

Conflict of Interest Disclosures: Dr Dauvilliers has received speaker’s honoraria and funding for travel to conferences from UCB Pharma, JAZZ, Novartis, and Bioprojet, and has participated in advisory boards of UCB and Bioprojet. Dr Carlander has received consultancy honoraria from Merck Serono and travel accommodation expenses from Biogen Idec, Merck Serono, Teva, and Novartis. Dr Lassmann serves on scientific advisory boards for Micromet, Biogen, and Baxter, and has given lectures for Novartis, Biogen, and Serono.

Dauvilliers  Y, Arnulf  I, Mignot  E.  Narcolepsy with cataplexy.  Lancet. 2007;369(9560):499-511.PubMedGoogle ScholarCrossref
Dalmau  J, Graus  F, Villarejo  A,  et al.  Clinical analysis of anti-Ma2-associated encephalitis.  Brain. 2004;127(pt 8):1831-1844.PubMedGoogle ScholarCrossref
Blumenthal  DT, Salzman  KL, Digre  KB, Jensen  RL, Dunson  WA, Dalmau  J.  Early pathologic findings and long-term improvement in anti-Ma2-associated encephalitis.  Neurology. 2006;67(1):146-149.PubMedGoogle ScholarCrossref
Overeem  S, Dalmau  J, Bataller  L,  et al.  Hypocretin-1 CSF levels in anti-Ma2 associated encephalitis.  Neurology. 2004;62(1):138-140.PubMedGoogle ScholarCrossref
Landolfi  JC, Nadkarni  M.  Paraneoplastic limbic encephalitis and possible narcolepsy in a patient with testicular cancer: case study.  Neuro Oncol. 2003;5(3):214-216.PubMedGoogle ScholarCrossref
Adams  C, McKeon  A, Silber  MH, Kumar  R.  Narcolepsy, REM sleep behavior disorder, and supranuclear gaze palsy associated with Ma1 and Ma2 antibodies and tonsillar carcinoma.  Arch Neurol. 2011;68(4):521-524.PubMedGoogle ScholarCrossref
Compta  Y, Iranzo  A, Santamaría  J, Casamitjana  R, Graus  F.  REM sleep behavior disorder and narcoleptic features in anti-Ma2-associated encephalitis.  Sleep. 2007;30(6):767-769.PubMedGoogle Scholar
Voltz  R, Gultekin  SH, Rosenfeld  MR,  et al.  A serologic marker of paraneoplastic limbic and brain-stem encephalitis in patients with testicular cancer.  N Engl J Med. 1999;340(23):1788-1795.PubMedGoogle ScholarCrossref
Rosenfeld  MR, Eichen  JG, Wade  DF, Posner  JB, Dalmau  J.  Molecular and clinical diversity in paraneoplastic immunity to Ma proteins.  Ann Neurol. 2001;50(3):339-348.PubMedGoogle ScholarCrossref
Dalmau  J, Gultekin  SH, Voltz  R,  et al.  Ma1, a novel neuron- and testis-specific protein, is recognized by the serum of patients with paraneoplastic neurological disorders.  Brain. 1999;122(pt 1):27-39.PubMedGoogle ScholarCrossref