Aquaporin-4 Autoantibodies in a Paraneoplastic Context | Allergy and Clinical Immunology | JAMA Neurology | JAMA Network
[Skip to Navigation]
Sign In
Table 1. 
Clinical Characteristics of Patients in Group 1 With Proven Cancer (n = 8) or Monoclonal Gammopathy (n = 1)a
Clinical Characteristics of Patients in Group 1 With Proven Cancer (n = 8) or Monoclonal Gammopathy (n = 1)a
Table 2. 
Clinical Characteristics of Patients in Group 2 With Proven Cancer (n = 7) or Monoclonal Gammopathy (n = 1)a
Clinical Characteristics of Patients in Group 2 With Proven Cancer (n = 7) or Monoclonal Gammopathy (n = 1)a
Original Contribution
May 2008

Aquaporin-4 Autoantibodies in a Paraneoplastic Context

Author Affiliations

Author Affiliations: Departments of Neurology, Laboratory Medicine and Pathology (Drs Pittock and Lennon), and Immunology (Dr Lennon), Mayo Clinic College of Medicine, Rochester, Minnesota.

Arch Neurol. 2008;65(5):629-632. doi:10.1001/archneur.65.5.629

Background  The neuromyelitis optica IgG autoantibody (NMO-IgG) is a validated biomarker for NMO and an emerging spectrum of inflammatory central nervous system–demyelinating disorders. Its antigen is the astrocytic water channel aquaporin-4; NMO-IgG has not been described in a cancer context.

Objectives  To report (1) neurologic and oncologic correlates for patients incidentally identified as NMO-IgG seropositive in a blinded evaluation for paraneoplastic autoantibodies and (2) the frequency of cancer in NMO-IgG–seropositive patients.

Design  Observational, retrospective case series.

Setting  Neuroimmunology Laboratory and Neurology Clinical Practice, Mayo Clinic College of Medicine.

Patients and Methods  From 1998 to 2007, we detected NMO-IgG in 2 patient groups: (1) 31 patients (88% female) identified incidentally among 180 000 patients evaluated for paraneoplastic autoantibodies and (2) 141 patients identified through physician-requested serological evaluation for a suspected NMO-spectrum disorder.

Results  In the first group, clinical information was available for 28 patients (90%). An NMO-spectrum disorder was diagnosed in 26 patients (93%), of whom 6 had a neoplasm (5 carcinomas [2 breast, 1 lung, 1 thymic, and 1 uterine cervical] and 1 B-cell lymphoma) and 1 had monoclonal gammopathy. In 4 patients, NMO-related symptoms followed neoplasia detection (median, 14 [range 3-18] months), and in 2 patients, symptoms preceded neoplasia detection (by 5 and 3 months). Two patients had carcinoma (1 breast and 1 lung) without neurological evidence of an NMO-spectrum disorder. In the second group, neoplasms were recorded in 7 seropositive patients (5.0%) with a clinically diagnosed NMO-spectrum disorder: 3 carcinomas (all breast), 1 thyroid Hürthle cell, 1 carcinoid, 1 pituitary somatotropinoma, and 1 B-cell lymphoma. An eighth patient had monoclonal gammopathy.

Conclusions  Aquaporin-4–specific IgG in some cases of NMO may reflect a paraneoplastic immune response. The clinical utility of this autoantibody as a cancer marker warrants prospective investigation.

Neuromyelitis optica IgG (NMO-IgG) is a sensitive and specific biomarker for inflammatory central nervous system (CNS)–demyelinating disorders collectively recognized as NMO-spectrum disorders (NMO, relapsing optic neuritis, and relapsing longitudinally extensive transverse myelitis).1 These disorders are often misdiagnosed as multiple sclerosis.1 Unlike patients with multiple sclerosis, patients with NMO-spectrum disorders frequently have clinical and serological stigmata of autoimmunity.2 To date, NMO-IgG has been reported only in patients with clinical evidence of an NMO-spectrum disorder.

The antigen of NMO-IgG is the astrocytic water channel aquaporin-4. Central nervous system lesions typical of aquaporin-4 autoimmunity with or without myelin loss are inflammatory with focal deposits of immunoglobulin and activated complement products and vasculocentric loss of aquaporin-4 in regions of astrocytic endfeet.1 The immunopathologic findings and reported benefit of plasmapheresis3,4 and monoclonal anti-CD20 B-cell antibody (rituximab) therapy5 support recent data demonstrating a role for complement-activating aquaporin-4–specific IgG as an initiator of the CNS-restricted accompaniments of NMO-IgG.6 We recently reported detection of NMO-IgG in a patient who presented with NMO in the setting of thymic carcinoma.7 That case and additional cases of cancer-associated NMO identified fortuitously in the course of paraneoplastic autoantibody evaluation in Mayo Clinic's Neuroimmunology Laboratory suggest the existence of a paraneoplastic subgroup of patients with NMO.

The data we present suggest that aquaporin-4 autoimmunity may, in some patients, reflect a tumor-initiated immune response. We report clinical and oncologic associations for 28 of 31 patients in whom NMO-IgG was detected incidentally in the course of paraneoplastic serological evaluation. The neurological presentations (93% consistent with an NMO-spectrum disorder) confirmed the high specificity of NMO-IgG for inflammatory CNS-demyelinating disorders of this type. The frequency of cancer in patients whose neurological presentations were consistent with an NMO-spectrum disorder was 31% (compared with 5.0% among NMO-IgG–seropositive patients presenting to the Mayo Clinic's Neurology Department with an NMO-spectrum disorder). Most remarkably, all 7% of the incidentally identified seropositive patients without evidence of an NMO-spectrum disorder had recent diagnoses of cancer.


The study was approved by our institutional review board. We describe 2 groups of patients identified serologically as NMO-IgG positive in the Mayo Clinic's Neuroimmunology Laboratory during a 10-year period.

Group 1

Among approximately 180 000 patients whose serum samples were tested prospectively for paraneoplastic autoantibodies on a service basis, the indirect immunofluorescence screening component (on a composite substrate of mouse tissues) identified 31 serum samples with IgG yielding a staining pattern consistent with NMO-IgG. In no case had the ordering physician initially entertained a diagnosis of NMO. Clinical information was available for 16 of the 18 patients evaluated outside the Mayo Clinic by review of case records provided by outside physicians, physician telephone interviews, and physician-provided responses to form questionnaires.

Group 2

At the physician's request, 141 seropositive patients were evaluated for NMO-IgG, because the clinical presentation was consistent with an NMO-spectrum disorder. Information for these Mayo Clinic patients was obtained by review of their medical records.

Serologic testing

All serum samples were titrated in doubling dilutions to ascertain the endpoint dilution yielding positive immunofluorescence staining.1 Where serum was available (Table 1 and Table 2), IgG specific for aquaporin-4 was confirmed by quantitative green fluorescent protein–linked aquaporin-4 immunoprecipitation assay in all patients identified with cancer.8,9


In group 1, clinical information was available for 28 of 31 patients identified incidentally as NMO-IgG seropositive: 13 were evaluated at the Mayo Clinic; 26 (93%) were women and 8 (28%) were black. Neurological symptoms and signs in 26 patients (93%) fit the recognized spectrum of NMO1: 8 fulfilled 2006 diagnostic criteria for NMO, 6 had recurrent/relapsing longitudinally extensive transverse myelitis, 9 had a single episode of longitudinally extensive transverse myelitis, and 3 had relapsing optic neuritis. Seven of these 26 patients (27%) had an associated neoplasm (Table 1); NMO-related symptoms followed the diagnosis of cancer in 5 of those patients (median, 14 [range 3-18] months) and preceded the diagnosis of cancer in 2 (by 5 and 3 months).

The remaining 2 patients in group 1 (7%; patients 1 and 4) (Table 1) lacked symptoms and signs consistent with the currently recognized spectrum of NMO. Both had carcinoma (1 lung and 1 breast), with neurological symptoms attributable to brain metastases (magnetic resonance imaging results were compatible in both and spinal fluid cytology was positive in patient 1).

In group 2, 7 of 141 patients (5.0%) had a history of neoplasm (Table 2): 3 carcinomas (all breast), 1 thyroid Hürthle cell, 1 carcinoid, 1 pituitary somatotropinoma, and 1 B-cell lymphoma. An eighth patient had monoclonal gammopathy (Table 2).


This article highlights the specificity of NMO-IgG for a recently recognized spectrum of inflammatory CNS-demyelinating disorders.1 Among 180 000 patients evaluated in 10 years on a service basis for paraneoplastic autoantibodies, 93% of the 0.02% found incidentally to have NMO-IgG had a CNS disorder consistent with the NMO spectrum. Of those whose clinical picture was consistent with CNS aquaporin-4 autoimmunity, 27% had a recent cancer diagnosis. Most remarkable was the finding of carcinoma (breast in 1 and lung in 1) in 2 seropositive patients who lacked clinical evidence of an inflammatory CNS-demyelinating disorder. The findings in these patients suggest that, in some cases, aquaporin-4–specific IgG may be produced in the course of a tumor immune response.

Thymoma and thymic carcinoma have previously been reported as sporadic accompaniments of NMO.7,10 Our documentation of 9 patients (7 with an NMO-spectrum disorder) in whom NMO-IgG was associated temporally with cancer justifies consideration of an underlying neoplasm in NMO-IgG–seropositive patients presenting with transverse myelitis or optic neuritis. For 3 of those 9 patients (33%), breast carcinoma was the most recently identified neoplasm. The detection of NMO-IgG in 2 patients whose neurological symptoms were attributable to CNS metastases of breast and lung carcinomas supports our suggestion that neoplastic cells may provide the antigen initiating an aquaporin-4 immune response.

Tumor cells express, as onconeural antigens, proteins that are normally expressed by mature neurons, glia, or muscle. Cancer-directed immune responses initiated by those antigens have the potential to target autoantigens in the nervous system.11-14 Aquaporin-4–IgG is the third glial-reactive IgG autoantibody recognized in a paraneoplastic context. The first glial antigen reactive with IgG autoantibody, specific for the collapsin-response mediator protein 5 (CRMP-5), is expressed in neurons and oligodendrocytes in the adult nervous system.14,15 The second, the nuclear transcription factor SOX1, is expressed in progenitors of neurons and astrocytes in adult brain.16 Collapsin-response mediator protein 5–IgG reflects an immune response initiated by small cell lung carcinoma or thymoma. The neurological presentation is multifocal in most CRMP-5–IgG–positive patients, but some patients present with a syndrome of myelopathy and optic neuropathy that mimics NMO.14 Its imaging and histopathologic findings are distinct from NMO17,18; SOX1-IgG19,20 (the anti-glial nuclear antibody/anti-neuronal nuclear antibody type 4 [AGNA/ANNA-4]21,22) is also a marker of immune responses initiated by small cell lung carcinoma antigens.

Aquaporin-4 is the most abundant water channel in the CNS. It is mercurial insensitive23 and highly expressed in the astrocytic end-feet at the glia limitans of the blood-brain barriers,1,8,23 synapses,23 and paranodes6 of myelinated axons.5 Outside the CNS, aquaporin-4 is expressed in membranes of skeletal muscle, glandular epithelia (breast and salivary glands), the lungs (tracheal and bronchial epithelium), kidneys (basolateral membranes of distal collecting tubules), stomach (parietal cells), and colon (epithelium).24,25

Tissue microarray analyses have revealed aquaporin-4 immunoreactivity in 11 of the 11 tested cancer types26 encountered in this study. This observation supports our hypothesis that aquaporin-4 may be pertinent clinically as a tumor antigen. Biological properties recently assigned to aquaporins are essential determinants of tumor cell extravasation and metastatic potential. These include enablement of cell migration (by facilitating transmembrane water influx into lamellipodia [dynamic cellular protrusions at the leading edge of migrating cells])27,28 and adhesion properties conferred by an alternatively spliced isoform of aquaporin-4 (lacking N-terminal residues 1-22).29 It remains to be determined whether aquaporin-4 autoimmunity protects against tumor spread.

The pathogenicity we propose for NMO-IgG is analogous to that demonstrated for the muscle acetylcholine-receptor antibody, which causes myasthenia gravis by impairing neuromuscular transmission.30 The acetylcholine-receptor antibody is a clinically useful marker for myasthenia gravis as well as for thymoma in patients without myasthenia gravis. Furthermore, we recognize acetylcholine-receptor antibodies in patients with cancer (small cell lung cancer or thymoma)31 and other paraneoplastic autoantibodies (eg, ANNA-1, 5%; CRMP-5–IgG, 8%)32 in patients who lack any clinical or electrophysiologic signs of myasthenia gravis. Prospective studies are needed to investigate the frequency of aquaporin-4 autoantibodies in patients with cancer and the protein and messenger RNA expression in neoplasms derived from seropositive patients.

Correspondence: Vanda A. Lennon, MD, PhD, Neuroimmunology Laboratory, Hilton 3-79, Mayo Clinic, 200 First St SW, Rochester, MN 55905 (

Accepted for Publication: November 15, 2007.

Author Contributions:Study concept and design: Pittock and Lennon. Acquisition of data: Pittock and Lennon. Analysis and interpretation of data: Pittock and Lennon. Drafting of the manuscript: Pittock. Critical revision of the manuscript for important intellectual content: Pittock and Lennon. Obtained funding: Lennon. Administrative, technical, and material support: Pittock and Lennon. Study supervision: Pittock and Lennon.

Financial Disclosure: The authors disclose that, in accordance with the Bayh-Dole Act of 1980 and Mayo Foundation policy, Dr Lennon stands to receive royalties for the discovery related to the aquaporin-4 autoantigen. To date, Dr Lennon has received a total of less than $1000 in royalties. Drs Pittock and Lennon are named inventors on a patent application filed by the Mayo Foundation for medical education and research that relates to the NMO (aquaporin-4) antibody and its application to cancer.

Funding/Support: This study was supported by the Mayo Foundation (Drs Pittock and Lennon) and the Ralph Wilson Medical Research Foundation (Dr Lennon).

Wingerchuk  DMLennon  VALucchinetti  CFPittock  SJWeinshenker  BG The spectrum of neuromyelitis optica.  Lancet Neurol 2007;6 (9) 805- 815PubMedGoogle Scholar
Pittock  SJLennon  VAde Seze  J  et al.  Neuromyelitis optica and non organ-specific autoimmunity.  Arch Neurol 2008;65 (1) 78- 83PubMedGoogle Scholar
Weinshenker  BGO'Brien  PCPetterson  TM  et al.  A randomized trial of plasma exchange in acute central nervous system inflammatory demyelinating disease.  Ann Neurol 1999;46 (6) 878- 886PubMedGoogle Scholar
Keegan  MPineda  AAMcClelland  RLDarby  CHRodriguez  MWeinshenker  BG Plasma exchange for severe attacks of CNS demyelination: predictors of response.  Neurology 2002;58 (1) 143- 146PubMedGoogle Scholar
Cree  BALamb  SMorgan  KChen  AWaubant  EGenain  C An open label study of the effects of rituximab in neuromyelitis optica.  Neurology 2005;64 (7) 1270- 1272PubMedGoogle Scholar
Hinson  SRPittock  SJLucchinetti  CF  et al.  Pathogenic potential of IgG binding to water channel extracellular domain in neuromyelitis optica [published online ahead of print October 10, 2007].  Neurology 2007;69 (24) 2221- 2231PubMed10.1212/01.WNL.0000289761.64862.ceGoogle Scholar
Pittock  SJWeinshenker  BGWingerchuk  D  et al.  Autoimmune neurological accompaniments of neuromyelitis optica (NMO).  Ann Neurol 2006;60S41Google Scholar
Lennon  VAKryzer  TJPittock  SJVerkman  ASHinson  SR IgG marker of optic-spinal multiple sclerosis binds to the aquaporin-4 water channel.  J Exp Med 2005;202 (4) 473- 477PubMedGoogle Scholar
Banwell  BTenembaum  SLennon  VA  et al.  Neuromyelitis optica-IgG in childhood inflammatory demyelinating CNS disorders [published online ahead of print December 19, 2007].  Neurology 2008;70 (5) 344- 359PubMed10.1212/01.wnl.0000284600.80782.d5Google Scholar
Antoine  JCCamdessanche  JPAbsi  LLassabliere  FFeasson  L Devic disease and thymoma with anti-central nervous system and antithymus antibodies.  Neurology 2004; (62(6)) 978- 980PubMedGoogle Scholar
Lennon  VAKryzer  TJGriesmann  GE  et al.  Calcium-channel antibodies in Lambert-Eaton myasthenic syndrome and other paraneoplastic syndromes.  N Engl J Med 1995;332 (22) 1467- 1474PubMedGoogle Scholar
Sciamanna  MAGriesmann  GEWilliams  CLLennon  VA Nicotinic acetylcholine receptors of muscle and neuronal α7 types coexpressed in a small cell lung carcinoma.  J Neurochem 1997;69 (6) 2302- 2311PubMedGoogle Scholar
Lennon  VAErmilov  LGSzurszewski  JHVernino  S Immunization with neuronal nicotinic acetylcholine receptor induces neurological autoimmune disease.  J Clin Invest 2003;111 (6) 907- 913PubMedGoogle Scholar
Yu  ZKryzer  TJGriesmann  GEKim  KBenarroch  EELennon  VA CRMP-5 neuronal autoantibody: marker of lung cancer and thymoma-related autoimmunity.  Ann Neurol 2001;49 (2) 146- 154PubMedGoogle Scholar
Ricard  DRogemond  VCharrier  E  et al.  Isolation and expression pattern of human Unc-33-like phosphoprotein 6/collapsin response mediator protein 5 (Ulip6/CRMP5): coexistence with Ulip2/CRMP2 in Sema3a- sensitive oligodendrocytes.  J Neurosci 2001;21 (18) 7203- 7214PubMedGoogle Scholar
Wegner  MStolt  CC From stem cells to neurons and glia: a Soxist's view of neural development.  Trends Neurosci 2005;28 (11) 583- 588PubMedGoogle Scholar
Cross  SASalomao  DRParisi  JE  et al.  Paraneoplastic autoimmune optic neuritis with retinitis defined by CRMP-5-IgG.  Ann Neurol 2003;54 (1) 38- 50PubMedGoogle Scholar
Keegan  MPittock  SLennon  VA Autoimmune myelopathy associated with CRMP-5-IgG [published online ahead of print February 27, 2008].  Ann Neurol 2008;63 (4) 531- 534PubMedGoogle Scholar
Vural  BChen  L-CSaip  P  et al.  Frequency of SOX group B (SOX1, 2, 3) and ZIC2 antibodies in Turkish patients with small cell lung carcinoma and their correlation with clinical parameters.  Cancer 2005;103 (12) 2575- 2583PubMedGoogle Scholar
Sabater  LTitulaer  MSaiz  AVerschuuren  JGüre  AOGraus  F SOX1 antibodies are markers of paraneoplastic Lambert Eaton myasthenic syndrome [published online ahead of print November 21, 2007].  Neurology 2008;70 (12) 924- 928PubMed10.1212/01.wnl.0000281663.81079.24Google Scholar
Graus  FVincent  APozo-Rosich  P  et al.  Anti-glial nuclear antibody: marker of lung cancer-related paraneoplastic neurological syndromes.  J Neuroimmunol 2005;165 (1-2) 166- 171PubMedGoogle Scholar
Lachance  DPittock  SJKryzer  TJ  et al.  Anti-neuronal nuclear antibody type 4 (ANNA-4): a novel paraneoplastic marker of small cell lung carcinoma (SCLC)  Neurology 2006;66 (5) A340- A340Google Scholar
Amiry-Moghaddam  MOttersen  OP The molecular basis of water transport in the brain.  Nat Rev Neurosci 2003;4 (12) 991- 1001PubMedGoogle Scholar
Frigeri  AGropper  MAUmenishi  FKawashima  MBrown  DVerkman  AS Localization of MIWC and GLIP water channel homologs in neuromuscular, epithelial and glandular tissues.  J Cell Sci 1995;108 (pt 9) 2993- 3002PubMedGoogle Scholar
Frigeri  AGropper  MATurck  CWVerkman  AS Immunolocalization of the mercurial-insensitive water channel and glycerol intrinsic protein in epithelial cell plasma membranes.  Proc Natl Acad Sci U S A 1995;92 (10) 4328- 4331PubMedGoogle Scholar
 Human Protein Atlas Web site. Accessed July 20, 2007
Hu  JVerkman  AS Increased migration and metastatic potential of tumor cells expressing aquaporin water channels.  FASEB J 2006;20 (11) 1892- 1894PubMedGoogle Scholar
Auguste  KIJin  SUchida  K  et al.  Greatly impaired migration of implanted aquaporin-4-deficient astroglial cells in mouse brain toward a site of injury.  FASEB J 2007;21 (1) 108- 116PubMedGoogle Scholar
Hiroaki  YTani  KKamegawa  A  et al.  Implications of the aquaporin-4 structure on array formation and cell adhesion.  J Mol Biol 2006;355 (4) 628- 639PubMedGoogle Scholar
Lennon  VALambert  EH Myasthenia gravis induced by monoclonal antibodies to acetylcholine receptors.  Nature 1980;285 (5762) 238- 240PubMedGoogle Scholar
Vernino  SLennon  VA Autoantibody profiles and neurological correlations of thymoma.  Clin Cancer Res 2004;10 (21) 7270- 7275PubMedGoogle Scholar
Pittock  SJKryzer  TJLennon  VA Paraneoplastic antibodies coexist and predict cancer, not neurological syndrome.  Ann Neurol 2004;56 (5) 715- 719PubMedGoogle Scholar