Josep Dalmau, M.D., Ph.D. 1,3 Humayun S. Gultekin,M.D.2,3, Jerome B. Posner M.D 1,3

Departments of 1Neurology, 2Pathology, and the 3CotziasLaboratory of Neuro-Oncology. Memorial Sloan-KetteringCancer Center.

Since 1965 when the first paraneoplastic antineu-ronal antibody was reported by Wilkinson andZeromski (55), the number of immunologicalresponses detected in association with paraneoplas-tic syndromes of the nervous system has steadilyincreased. These responses are characterized by thepresence of antineuronal antibodies in serum andCSF and/or infiltrates of T-cells in the tumor and ner-vous system. A few syndromes are mediated by anti-bodies; they include those resulting from dysfunc-tion of the neuromuscular junction at the pre- orpost-synaptic level (Lambert-Eaton myasthenic syn-drome, myasthenia gravis) or ion channel dysfunc-tion in the peripheral nervous system (i.e, Voltage-gated potassium channel and neuromyotonia). Inmost other paraneoplastic syndromes, includingthose involving the central nervous system, the path-ogenic role of highly specific antineuronal antibodies(anti-Hu, anti-Yo, etc) has not been established; nev-ertheless these antibodies should be regarded asuseful markers of specific paraneoplastic syndromesand tumors. Moreover, there is increasing evidencethat in some of these syndromes T-cell mediatedmechanisms can cause the neurologic dysfunctionand contribute to tumor rejection. Some paraneo-plastic syndromes are caused by the tumor secretionof antibodies (macroglobulinemia and MAG antibod-ies), hormones, and cytokines. In other instances, thetumor may compete with the nervous system for anessential substrate (glucose, tryptophan) and resultin neurologic dysfunction.

IntroductionIn patients with cancer the most common neurologi-

cal complications are related to metastases and sideeffects of therapy (42). Paraneoplastic syndromes,defined as nervous system disorders related to cancerbut not caused by any of the complications shown inTable 1, are rare complications. This definition makesthe diagnosis of paraneoplastic syndromes difficult asthey come to include multiple syndromes that can resultfrom known or unknown pathogenic mechanisms. Forexample, the development of opsoclonus or acutemyelopathy in a cancer patient may either result from aviral process, or be considered paraneoplastic if a spe-cific cause is not identified (42). In recent years, the dis-covery of antibodies against onconeuronal proteins inthe serum and cerebrospinal fluid (CSF) of patients withcancer and some paraneoplastic syndromes has revealedthe existence of immune responses targeted both to thetumor and the nervous system (14). A major contribu-tion of the detection of these antibodies is that theyestablish the diagnosis of paraneoplastic syndromes andfocus the search for the underlying tumor to a limitednumber of organs. An increasing number of neurologicdisorders now can be considered paraneoplastic notbecause of the absence of known mechanisms, butbecause of the presence of specific paraneoplastic mark-ers (Table 2). The current work focuses on the paraneo-plastic disorders defined by the presence of specificimmunologic responses.

Pathogenic Mechanisms of the NeurologicDysfunction

Most paraneoplastic neurologic syndromes appear tobe immune-mediated (Table 2) either by antibodies orcytotoxic T-cell related mechanisms.

Antibody related mechanisms.Immune mediatedmechanisms have been demonstrated for the Lambert-Eaton myasthenic syndrome (LEMS), myastheniagravis, and neuromyotonia (16, 27, 31, 32). Patients

Brain Pathology 9: 275-284(1999)

Paraneoplastic Neurologic Syndromes:Pathogenesis and Physiopathology

MINI-SYMPOSIUM: Paraneoplastic Syndromes

Corresponding author:Josep Dalmau, MD, PhD. Department of Neurology, MemorialSloan-Kettering Cancer Center; 1275 York Ave, New York, NY,10021; Tel.: (212) 639-7045; Fax: (212) 717-3519;E-mail: dalmauj@mskcc.org

with LEMS develop antibodies against the P/Q-typevoltage gated calcium channels (VGCC) located at thepresynaptic level of the neuromuscular junction. Theseantibodies block the entry of calcium necessary for therelease of quanta of acetylcholine and result in neuro-muscular weakness. A similar mechanism has beendemonstrated in myasthenia gravis, where a tumor (thy-moma) triggers an immune response against the acetyl-choline receptor at the postsynaptic level of the neuro-muscular junction, resulting in weakness and fatigabili-ty. Thymoma is also the tumor most frequently associat-ed with paraneoplastic neuromyotonia. In this disorderthe development of antibodies against voltage gatedpotassium channels results in a syndrome characterizedby muscle cramps, myokymia, and difficulty in musclerelaxation due to continuous muscle fiber activity. Inaddition, patients frequently complain of peripheral sen-sorimotor neuropathy and hyperhydrosis (38).

Thus, LEMS, myasthenia gravis and neuromyotoniaare directly mediated by antibodies against cell surfaceantigens of neuronal structures located outside the bloodbrain barrier. Removal of these antibodies by plasmaexchange results in clinical and electrophysiologicalimprovement; their transfer to animals reproduces theclinical, electrophysiological and pathological abnor-malities (32, 36, 38-40).

In the case of antibody-associated paraneoplasticsyndromes of the central nervous system (CNS), passivetransfer of the patients’ serum or IgG to animals, or theirimmunization with onconeuronal antigens have notreproduced the disease (table 3) (45, 49, 50). In thesedisorders the location of the antigens is intracellular,either predominantly cytoplasmic (CDR, Tr, Ulip,amphiphysin) or nuclear (Hu proteins, Nova) (14).These antibodies are specific markers of characteristicparaneoplastic syndromes and tumors, and indicatorsthat immunological mechanisms operate in these disor-ders. This concept is supported by the frequent detectionof intrathecal synthesis of antineuronal antibodies (17,24), and by their absence in other degenerative orinflammatory disorders of the nervous system (3, 46).To date there is no evidence that antineuronal antibodiesassociated with paraneoplastic syndromes of the CNSare pathogenic, but it is important to remember thatnone of the animal studies have reproduced the continu-ous intrathecal synthesis of antibodies characteristic ofthe human syndromes. Furthermore, autopsy studies ofpatients with the anti-Hu syndrome show deposits ofanti-Hu antibodies in the nervous system and tumor (6,10).

J. Dalmau et al: Paraneoplastic Neurologic Syndromes: Pathogenesis and Physiopathology276

•Metastatic•No Metastatic

•Side effects of therapy•Coagulopathy•Infections•Metabolic and nutritional deficits•Paraneoplastic

Table 1. Neurological Complications of Cancer.

Brain and cranial nerves•Limbic encephalitis•Brainstem encephalitis•Cerebellar degeneration•Opsoclonus-myoclonus•Paraneoplastic visual syndromes

•Cancer associated retinopathy (CAR)•Optic neuritis

Spinal cord•Necrotizing myelopathy•Myelitis•Motor neuron syndrome•Subacute motor neuronopathy

Dorsal root gangliaParaneoplastic sensory neuronopathy

Peripheral nerves•Autonomic neuropathy•Acute sensorimotor neuropathy

•polyradiculoneuropathy (Guillain-Barré)*•Brachial neuritis*

•Chronic sensorimotor neuropathy•Sensorimotor neuropathies, with plasma cell dyscrasias*

•Vasculitic neuropathy*•Neuromyotonia

Neuromuscular junction and muscle•Lambert-Eaton myasthenic syndrome•Myasthenia gravis•Polymyositis/Dermatomyositis*•Acute necrotizing myopathy•Cachectic myopathy•Carcinoid myopathy•Myotonia

Multiple level of involvement or uncertain site•Encephalomyelitis**•Stiff-man syndrome•Carcinomatous neuromyopathy

Italics indicate disorders associated with characteristic anti-neuronal antibodies.(*) Indicate syndromes associated with immunological dys-function, but without specific paraneoplastic markers.(**) Can include cerebellar symptoms, autonomic dysfunc-tion, and sensory neuronopathy.

Table 2. Paraneoplastic Syndromes of the Nervous System.

The concept that antibodies directed against intracel-lular antigens may be pathogenic is supported by thefollowing findings: 1) In vitro internalization of anti-Huantibodies has been demonstrated in neuronal and tumorcell lines (25, 29), 2) Hu antigen-like molecules areexpressed in the surface of neuroblastoma cell lines,suggesting a mechanism of cell surface antibody bind-ing and internalization (51), and 3) internalization ofantinuclear antibodies are pathogenic in other diseases(i.e, systemic lupus erythematosus and anti-dsDNA) (2,20).

T-cell related mechanisms.In some paraneoplasticsyndromes of the CNS, circumstantial evidence sug-gests that T-cell mediated mechanisms play a majorpathogenic role. Four findings support this hypothesis.

1) Autopsies of patients with paraneoplastic syn-dromes of the CNS show intense inflammatory infil-trates of mononuclear cells, including CD4 and CD8,that predominate in areas of the nervous system that aresymptomatic (figure 1) (10, 23, 30). Some T-cell infil-trates appear in close contact with neurons. The mecha-nism whereby CD4 or CD8 cytotoxic T-cells recognizeantigens expressed in neurons (which in normal circum-stances lack expression of the antigen presenting mole-cules MHC class I and II) is unknown. It is possible thatthese inflammatory infiltrates are not the primary causeof the neuronal damage, but a consequence of a primaryunknown mechanism (i.e. cytokines, antibodies) thatcauses neuronal dysfunction. Since neuronal damagecan result in expression of MHC molecules (37), the T-

cell infiltrates would be a “second hit” resulting in neu-ronal loss.

Extensive infiltrates of T-cells have been demonstrat-ed in the nervous system of patients with anti-Hu (10,23, 30), anti-Ri (28), anti-Ma (13) and anti-Ta antibod-ies (54). Milder mononuclear infiltrates have also beenfound in the brainstem and deep cerebellar nuclei ofpatients with anti-Yo associated cerebellar degeneration(19, 47, 52). Furthermore, the early presence of cere-brospinal fluid pleocytosis in most patients with theanti-Yo syndrome suggests that cell-mediated mecha-nisms play a pathogenic role (41).

2) The involvement of cytotoxic T-cell mediatedmechanisms in some antibody associated paraneoplasticsyndromes of the CNS is supported by a study of the T-cell receptor usage in the inflammatory infiltrates of theCNS of patients with the anti-Hu syndrome (53). In thatstudy 5 out of 5 patients showed a limited Vb T-cellreceptor repertoire. An over-representation (>10% oftotal CD3+) of certain Vb families (up to 45% of totalCD3+), which consisted mainly of CD8+ cells was iden-tified in 3 patients. Furthermore, the CDR3 sequencesobtained from one patient revealed an in situ expansionof two clones in the amygdala (one at a frequency of57%) and four clones in the tumor. This study demon-strates that the CNS and tumor infiltrating T-lympho-cytes are not indirectly recruited by a microinflammato-ry environment, but are specifically targeted to neuronaland tumor antigens.

3) To assess the issue of cell-mediated immunity inthe anti-Hu syndrome, Benyahia et al (4), examined the

J. Dalmau et al: Paraneoplastic Neurologic Syndromes: Pathogenesis and Physiopathology 277

Immunological findings Disorder Antigens(existence of animal model)

Antibodies to onconeuronal antigens. LEMS P/Q-type VGCC(Proved with animal model) Myasthenia gravis Acetylcholine receptor

Neuromyotonia VGKC

Antibodies to onconeuronal antigens. PEM/PSN Hu antigens (HuD, HuC/pL21, Hel-N1)(No animal model) PCD (breast, ovary cancers) Yo antigens (CDR34, CDR62)

PCD (Hodgkin’s lymphoma) Unknown (protein in the cytoplasm and dendrites of Purkinje cells and other neurons)

PCD/brainstem encephalitis Ma proteins (Ma1 and Ma2)PLE Ma2Ataxia/Opsoclonus Nova (Ri antigens)Stiff-man syndrome AmphiphysinEncephalomyelitis CV antigen (Ulip)

Direct synthesis of IgM or IgG by the tumor Sensorimotor neuropathy(No animal model) (Waldenström´s macrogl. MAG

Motor neuropathy (lymphoma) GD1

Table 3. Immune Pathogenesis of Paraneoplastic Neurologic Disorders. LEMS: Lambert-Eaton myasthenic syndrome, VGCC:Voltage-gated calcium channel, VGKC: Voltage-gated potassium channel, PEM: Paraneoplastic encephalomyelitis, PSN:Paraneoplastic sensory neuronopathy, PCD: paraneoplastic cerebellar degeneration, PLE: paraneoplastic limbic encephalitis, MAG:myelin-associated glycoprotein

peripheral blood lymphocyte (PBL) surface phenotypeof 15 patients suffering from the anti-Hu syndrome and2 suitable control groups. In addition, human recombi-nant HuD was used to stimulate in vitro peripheral bloodmononuclear cells of patients with anti-Hu syndromeand controls. These studies showed a significantincrease of the memory helper T-cells in the anti-Huseropositive group in comparison with the two controlgroups. Antigen specific proliferation of peripheralblood mononuclear cells was much higher in theseropositive group than in the controls. Furthermore, asignificant increase of the g-interferon/interleukin 4ratio was identified in the supernatants of the seroposi-tive group compared with the controls, suggesting thatthe HuD protein is an antigenic target for autoreactiveCD4+ T-cells, presumably of the Th1 subtype which

could be involved in direct cell-mediated injury.4) The tumors of patients with paraneoplastic disor-

ders frequently express MHC class I and II antigens,which make them more ‘visible’ to the immunologicalsystem (figure 2). For example, in a study of 15 small-cell lung cancers from patients with the anti-Hu syn-drome (11), 14 tumors were found to have high expres-sion of MHC class I antigens, while only 4 of 11 tumorsfrom patients without paraneoplastic symptoms hadmild expression of these antigens. In the same study,expression of MHC class II antigens was identified in6/15 tumors from patients with paraneoplastic symp-toms and 0/11 tumors from patients without paraneo-plastic symptoms.

Tumor Production of Antibodies. A third group ofparaneoplastic neurologic syndromes associated withimmunological disturbances are disorders associated

J. Dalmau et al: Paraneoplastic Neurologic Syndromes: Pathogenesis and Physiopathology278

Figure 1. Infiltrates of T-cells in the nervous system of patientswith paraneoplastic syndromes. Interstitial inflammatory infil-trates of T-cells in the mesial temporal lobe of a patient withparaneoplastic limbic encephalitis associated with testicularcancer and anti-Ta antibodies (A), and in the brainstem of apatient with anti-Hu associated encephalomyelitis and small-cell lung cancer (B).

Figure 2. Expression of MHC class I antigens by the small-celllung cancer of a patient with anti-Hu associatedencephalomyelits. Small-cell lung cancer of a patient with anti-Hu associated encephalomyelitis labeled with anti-Hu IgG (A)and the anti-MHC class I antibody (W6/32) (B). Note that bothpanels are consecutive sections of the same tumor. Neoplasticcells labeled with anti-Hu IgG in A, demonstrate high level ofexpression of MHC class I in B. Sections not counterstained.

with lymphoid neoplasms (table 3). Most of these para-neoplastic disorders affect the peripheral nervous sys-tem resulting in sensorimotor or motor neuropathy (34,44). In some instances (i.e., Waldenström’s macroglob-ulinemia) the monoclonal gammopathy has antibodyactivity against known antigens of the peripheral ner-vous system, such as myelin associated glycoprotein(MAG). In other instances (i.e., neuropathy associatedwith osteosclerotic myeloma) the specific target antigenin the peripheral nerve is unknown.

In addition to peripheral neuropathies, lower andupper motor neuron syndromes (sometimes identical toamyotrophic lateral sclerosis) have been identified morefrequently than expected in patients with lymphoma (15,43, 56). Although some of these patients have circulat-ing anti-ganglioside antibodies, the presence of theseantibodies is not consistent and their pathogenic role isuncertain.

Non-Immune Mediated Mechanisms.Paraneoplas-tic syndromes can also result from non-immune mediat-ed mechanisms. They include: 1) synthesis of hormone-like substances (i.e, ADH, PTH) by the tumor that maycause encephalopathy (42), 2) competition for substratebetween the tumor and the nervous system (i.e, trypto-phan and carcinoid syndrome, hypoglycemia in patientswith large retroperitoneal tumors) (5, 18), and 3) secre-tion of cytokines (IL-1, IL-6, TNF-a) by the tumor orinflammatory infiltrates, resulting in cachexia (42, 48).

Effects of the Paraneoplastic Disorder on the TumorFor many years, cancer investigators had observed

that the tumors of patients with paraneoplastic syn-dromes behave more indolently than similar tumors inpatients without paraneoplastic syndromes. Two impor-tant limitations to testing this hypothesis are: 1) somesyndromes are so aggressive that the neurologic disorderresults in early death of the patient, thus interfering withany study regarding the natural course of the associatedtumor (12), and 2) the development of the paraneoplas-tic syndrome prior to cancer often leads to the earlierdetection of the tumor in these patients compared withpatients without paraneoplastic syndromes.

To overcome these limitations we took advantage ofthe observation that about 20% of patients with SCLC,but no neurological symptoms, harbor low titer of anti-Hu antibodies (9). Indeed, in patients with low titer ofanti-Hu antibodies the tumor is more likely to be limit-ed to the chest, to respond better to the chemotherapy,and to allow a longer survival than in patients withoutparaneoplastic immune responses (21). For other para-

neoplastic syndromes no similar studies are available,and the idea that the tumor of these patients is moreindolent than the tumor of patients without paraneoplas-tic syndromes remains a clinical impression (41).

A recent study using an animal model of anti-Huimmune response supports the concept that the specificimmunological reaction against the Hu antigens isresponsible for a more favourable outcome of the tumorof these patients (7). This study showed that animalsimmunized with HuD protein developed high titers ofantibodies without effect on the nervous system or on atumor (neuroblastoma) expressing Hu proteins.However, most animals immunized with HuD cDNAshowed smaller tumor volumes or rejection of the tumor.The major difference between these two models residesin the processing and presentation of the antigen to theimmunological system. Immunization with proteinsrequires protein uptake and processing via MHC classII, which primes CD4 cells and production of antibod-ies. Immunization with cDNA results in production ofprotein that is processed and presented to the immuno-logical system via MHC class I and II. Therefore, thistechnique not only primes the production of antibodiesbut also stimulates cytotoxic CD8 related mechanisms.

J. Dalmau et al: Paraneoplastic Neurologic Syndromes: Pathogenesis and Physiopathology 279

Figure 3. Effects of HuD cDNA immunization on a model ofmice with neuroblastoma (Neuro-2a cells). Neuro2a tumor vol-umes (mean 6 SEM) in A/J mice immunized with pcDNA3(n=14), pHuDsec (n=14), pHuM (n=10), pHuF (n=10), orrecombinant HuD protein (n=14). Days equals the number ofdays after tumor implantation. On day 22, pHuDsec immunizedanimals showed tumor volumes that were 51% smaller thancontrols (p = 0.0012). pHuM and pHuF immunized animalsshowed 16% and 14% reduction in tumor volumes, respective-ly (p = not significant). Animals immunized with recombinantHuD protein (HuD prot) showed no inhibition of tumor growth.Control animals were immunized with plasmid without insert.(courtesy of Drs. Carpentier and Rosenfeld).

In the model developed by Carpentier et al (7), animalsthat showed smaller tumor volumes had also larger infil-trates of T cells in the tumor (figure 3). The animals,however, did not develop neurological symptoms orneuropathological abnormalities. Thus, rather than amodel of paraneoplastic neurologic disease, this modelis closest to the phenomenon observed in patients withsmall-cell lung cancer and low titers of anti-Hu antibod-ies, who usually have smaller tumor volumes (orabsence of metastases) but no overt neurologicaldeficits.

Findings of an anti-tumor cytotoxic T-cell responsehas also been demonstrated in patients with anti-Yoassociated paraneoplastic cerebellar degeneration (RDarnell, personal communication). This cytotoxicresponse may account for the apparent indolent behav-ior of the tumors of these patients.

Implications of the Detection of Paraneoplastic

Antineuronal AntibodiesWhether or not antineuronal antibodies cause parane-

oplastic syndromes, the detection of these antibodies hasimportant clinical implications:

1) Antineuronal antibodies represent an importantshortcut to the diagnosis of paraneoplastic neurologicsyndromes. Although some paraneoplastic syndromes,particularly those involving the neuromuscular junction,neuromyotonia and stiff-person syndrome, can be diag-nosed using electrophysiological studies, most paraneo-plastic syndromes are difficult to diagnose. Before anti-neuronal antibodies were discovered, the diagnosis ofmost of these syndromes could only be inferred byexclusion of metastatic and non-metastatic complica-tions of cancer on the nervous system. In many instancesthe final diagnosis could only be established by biopsyor at autopsy, as it is the case for syndromes in which nomarkers have yet been identified (8).

2) The detection of antineuronal antibodies directsthe search of the underlying tumor to a few organs andreveals its presence at an early stage. About 60% ofpatients with paraneoplastic syndromes develop the neu-rologic symptoms before the tumor diagnosis (42). Thedetection of specific antibodies in these patients not onlyconfirms the paraneoplastic nature of the disorder butalso indicates the most likely primary location of theneoplasm. The association between specific antibodiesand associated tumors is shown in table 4.

3) The discovery of antineuronal antibodies hasrevealed that different antibodies can be associated withthe same neurologic syndrome and, conversely, that thesame antibody may be associated with different syn-dromes. This occurs particularly in the initial stages ofthe illness, and has led to the concept of immunologic-phenotypic diversity. Three possibilities may arise (table5): 1) some immunological responses are consistentlyassociated with one syndrome (i.e., antibodies againstP/Q-type voltage gated calcium channels and theLambert-Eaton myasthenic syndrome) (35); 2) otherimmunological responses are highly specific markers ofmore than one syndrome, or syndromes with multifocalneurological deficits; an example is the anti-Hu immuneresponse in association with a number of symptomsdepending upon the areas of the nervous systeminvolved by the paraneoplastic encephalomyelitis; 3)several paraneoplastic immunological responses can beassociated with similar neurological symptoms. Thebest example is paraneoplastic cerebellar degeneration(table 6).

4) The detection of some antibodies has prognosticand therapeutic implications. It is known that some syn-

J. Dalmau et al: Paraneoplastic Neurologic Syndromes: Pathogenesis and Physiopathology280

Antibody Syndrome Tumor

Anti-Hu PEM/PSN SCLC

Anti-Yo PCD Ovary, breast

Anti-Ri Opsoclonus-ataxia Breast, Gyn

Anti-CV2 PEM/PCD Several: SCLC, thymoma

Anti-amphiphysin Stiff-person Breast, SCLC

Anti-Tr PCD Hodgkin’s, Non-Hodgkin’s

lymphoma

Anti-Ta Limbic/brainstem enc. Testis

Anti-Ma Brainstem enc. /PCD Several: colon, breast,

lung, parotid gland

Anti-VGCC LEMS SCLC, Lymphoma

Anti-VGKC Neuromyotonia SCLC, Thymoma

Anti-AChR Myasthenia gravis Thymoma

Table 4. Antineuronal Antibodies and the Most CommonAssociated Paraneoplastic Syndromes and Tumors.

Immune response Syndrome(s)

One Immune Response One SyndromeAnti-P/Q type VGCC: LEMSAnti-VGKC Neuromyotonia

One Immune Response Several SyndromesAnti-Hu: PEM, SN Anti-amphiphysin: Stiff-person, PEM

Several Immune Responses One SyndromeAnti-Yo, Anti-Tr, Anti-Hu: PCDAnti-Hu, anti-Ta: Limbic encephalitis

Table 5. Immunological and Clinical Diversity.

dromes, such as those directly mediated by antibodies(myasthenia gravis, LEMS, neuromyotonia) respond toimmunomodulation. In this case one specific antibodyassociates with one specific syndrome (first possibility,see above), and removal of the antibody (plasmaexchange) or immunomodulation (IVIg) may improvethe neurological deficit. In the situation where severalantibodies may be associated with similar neurologicalsymptoms (third possibility), the detection of a specificantibody not only indicates a specific type of neo-plasm(s), but also predicts whether the neurological dys-function will spread to other areas of the nervous sys-tem. For example, the detection of anti-Hu antibodies ina patient presenting with cerebellar dysfunction indi-cates that the underlying disorder is anencephalomyelitis and that symptoms will eventuallyinvolve other areas of the central nervous system, dorsalroot ganglia or autonomic nerves (33). However, inpatients with similar cerebellar symptoms but with anti-Yo or anti-Tr antibodies, symptoms will probablyremain restricted to the cerebellum, whereas patientswith anti-Tr antibodies symptoms may improve withtreatment of the tumor or spontaneously (41, 22).

5) The characterization of clinical-immunologicalsubsets of paraneoplastic disorders has made identifica-tion of new syndromes easier. An example, is the identi-fication of anti-Ta antibodies and the corresponding tar-get antigen, a neuronal protein called Ma2 (54). Twoobservations led to the discovery of this antibody: 1)Case report studies suggest a more frequent than expect-ed association between testicular cancer and paraneo-plastic limbic encephalitis, and 2) In a patient with thissyndrome an antineuronal antibody had been previouslyidentified (1). Taking these observations we retrospec-tively examined the tumors associated with symptomssuggesting paraneoplastic limbic encephalitis. SCLCwas the related tumor in 28 of 52 (53%) patients, andsome of these patients had anti-Hu antibodies. After

SCLC, the second leading neoplasm was testicular can-cer (9/52; 17% ). Eight of these 9 patients had a novelantineuronal antibody that was used to clone Ma2, thegene that codes for a protein (Ma2) recognized by thesera of all 8 patients, but not 304 controls. Therefore,Ma2 is the target antigen of anti-Ta antibodies, and theseantibodies appear to be highly specific for paraneoplas-tic limbic and brainstem encephalitis associated withtesticular tumors. Ma2 is highly homologous to Ma1,another gene identified by virtue of the immunogenicproperties of the protein (Ma1) that encodes. Ma1 is thetarget antigen of antibodies specifically associated with

J. Dalmau et al: Paraneoplastic Neurologic Syndromes: Pathogenesis and Physiopathology 281

Name Usual Tumor Clinical Findings Sex Onset Course/Cause of Death

Anti-Yo Gyn, Breast PCD F>M Neuro Subacute-severe/Tumor

Anti-Ri Breast, Gyn,SCLC Ataxia Opsoclonus F>M Neuro +/- Remitting

Anti-Tr Hodgkin’s,Non-Hodgkin’s PCD M>F Oncol +/- Remitting

Anti-Hu (+)SCLC SCLC PSN /PEM PCD F>M Neuro Subacute-severe/Neurological

Anti-Hu (-)SCLC SCLC PCD (>>LEMS) M>F Neuro Subacute-severe/Tumor

Anti-CV2 SCLC, thymoma,uterus PCD PEM/PSN M>F Neuro Severe/ Neurological

Anti-Ma Several Brainstem/ PCD F>M Neuro Severe/ Neurological

Table 6. Clinical-Immunological Subsets of Paraneoplastic Cerebellar Degeneration.

Figure 4. Anti-Hu immunolabeling as an index of neuronal dif-ferentiation in human brain tumors. Medulloblastoma (row 1)and neurocytoma (row 2) reacted with anti-Hu Fab and anti-synaptophysin. Both tumors were synaptophysin negative, butshowed reactivity with the anti-Hu antibody, an early marker ofneuronal differentiation.

paraneoplastic cerebellar degeneration, brainstemencephalitis and several types of tumors (13).

6) Because the expression of some paraneoplasticantigens is highly restricted to neurons or neuronal pre-cursors, the corresponding antibodies have become use-ful markers in surgical pathology. The best example isthe anti-Hu antibody. This antibody reacts exclusivelywith neurons and tumors that express neuroendocrinemarkers, mainly SCLC and neuroblastoma. In a recentstudy a recombinant Fab fragment against Hu proteinswas used as an index of neuronal differentiation inhuman brain tumors (26). Among 112 tumors of centralneuroepithelial lineage, the anti-Hu reactivity identifiedneoplasms of mature neuronal phenotype and embryon-al tumors with immature neuronal phenotype better thanthe staining with antibodies against synaptophysin (fig-ure 4). No anti-Hu reactivity was detected in 58 of 59glial tumors, the only exception being one of twelveoligodendrogliomas which may in fact have been anextraventricular neurocytoma.

References

1. Ahern GL, O’Connor M, Dalmau J, Coleman A, PosnerJB, Schomer DL, Herzog AG, Kolb DA, Mesulam M(1994) Paraneoplastic temporal lobe epilepsy with testic-ular neoplasm and atypical amnesia. Neurology 44: 1270-1274

2. Alarcon-Segovia D, Ruiz-Arguelles A, Llorente L (1996)Broken dogma: penetration of autoantibodies into livingcells. Immunol Today 17: 163-164

3. Anderson NE, Rosenblum MK, Graus F, Wiley RG,Posner JB (1988) Autoantibodies in paraneoplastic syn-dromes associated with small-cell lung cancer. Neurology38: 1391-1398

4. Benyahia B, Liblau R, Merle-Béral H, Tourani J-M,Dalmau J, Delattre J-Y (1998) Cell-mediated auto-immu-nity in paraneoplastic neurologic syndromes with anti-Huantibodies. Ann Neurol, in press

5. Bourin M, Hascoet M, Deguiral P (1996) 5-HTP induceddiarrhea as a carcinoid syndrome model in mice?Fundam Clin Pharmacol 10: 450-457

6. Brashear HR, Caccamo DV, Heck A, Keeney PM (1991)Localization of antibody in the central nervous system ofa patient with paraneoplastic encephalomyeloneuritis.Neurology 41: 1583-1587

7. Carpentier AF, Rosenfeld MR, Delattre J-Y, Whalen RG,Posner JB, Dalmau J (1998) DNA vaccination with HuDinhibits tumor growth of a neuroblastoma in mice. ClinCancer Res 4: 2819-2824

8. Dalmau J (1997) Paraneoplastic syndromes of the ner-vous system: General pathogeneic mechanisms anddiagnostic approaches. In: Handbook of ClinicalNeurology, Vinken PJ, Bruyn GW (eds.), Volume 69,Chapter 17, pp. 319-328, Elsevier: Amsterdam

9. Dalmau J, Furneaux HM, Gralla RJ, Kris MG, Posner JB(1990) Detection of the anti-Hu antibody in the serum ofpatients with small cell lung cancer - a quantitative west-ern blot analysis. Ann Neurol 27: 544-552

10. Dalmau J, Furneaux HM, Rosenblum MK, Graus F,Posner JB (1991) Detection of the anti-Hu antibody inspecific regions of the nervous system and tumor frompatients with paraneoplastic encephalomyelitis/sensoryneuronopathy. Neurology 41: 1757-1764

11. Dalmau J, Graus F, Cheung N-KV, Rosenblum MK, Ho A,Cañete A, Delattre JY, Thompson SJ, Posner JB (1995)Major histocompatibility proteins, anti-Hu antibodies, andparaneoplastic encephalomyelitis in neuroblastoma andsmall cell lung cancer. Cancer 75: 99-109

12. Dalmau J, Graus F, Rosenblum MK, Posner JB (1992)Anti-Hu-associated paraneoplastic encephalomyelitis/sensory neuronopathy. A clinical study of 71 patients.Medicine 71: 59-72

13. Dalmau J, Gultekin SH, Voltz R, Hoard R, DesChamps T,Balmaceda C, Batchelor T, Gerstner E, Eichen J, FrennierJ, Posner JB, Rosenfeld MR (1999) Ma1, a novel neu-ronal and testis specific protein, is recognized by theserum of patients with paraneoplastic neurologic disor-ders. Brain 122: 27-39

14. Dalmau JO, Posner JB (1997) Paraneoplastic syndromesaffecting the nervous system. Semin Oncol 24: 318-328

15. Forsyth PA, Dalmau J, Graus F, Cwik V, Rosenblum MK,Posner JB (1997) Motor neuron syndromes in cancerpatients. Ann Neurol 41: 722-730

16. Fukunaga H, Engel AG, Lang B, Newsom-Davis J,Vincent A (1983) Passive transfer of Lambert-Eatonmyasthenic syndrome with IgG from man to mousedepletes the presynaptic membrane active zones. ProcNatl Acad Sci USA 80: 7636-7640

17. Furneaux HM, Reich L, Posner JB (1990) Autoantibodysynthesis in the central nervous system of patients withparaneoplastic syndromes. Neurology 40: 1085-1091

18. Gilbert JA, Bates LA, Ames MM (1995) Elevated aromat-ic-L-amino acid decarboxylase in human carcinoidtumors. Biochem Pharmacol 50: 845-850

19. Giometto B, Marchiori GC, Nicolao P, Scaravilli T, Lion A,Bardin PG, Tavolato B (1997) Sub-acute cerebellardegeneration with anti-Yo autoantibodies: immunohisto-chemical analysis of the immune reaction in the centralnervous system. Neuropathol Appl Neurobiol 23: 468-474

20. Golan TD (1997) The nuclear compartment of living cellsis not an absolute immunologically sequestered site: evi-dence for spontaneous intranuclear entry of lupus autoan-tibodies to living cells. Leukemia 11: 6-9

21. Graus F, Dalmau J, Reñé R, Torà M, Malats N,Verschuuren JJ, Cardenal F, Viñolas N, García del MuroJ, Vadell C, Mason WP, Rosell R, Posner JB, Real FX(1997) Anti-Hu antibodies in patients with small cell lungcancer: association with complete response to therapyand improved survival. J Clin Oncol 15: 2866-2872

J. Dalmau et al: Paraneoplastic Neurologic Syndromes: Pathogenesis and Physiopathology282

22. Graus F, Dalmau J, Valldeoriola F, Ferrer I, Reñé R, MarinC, Vecht ChJ, Arbizu T, Targa C, Moll JWB (1997)Immunological characterization of a neuronal antibody(anti-Tr) associated with paraneoplastic cerebellar degen-eration and Hodgkin’s disease. J Neuroimmunol 74: 55-61

23. Graus F, Ribalta T, Campo E, Monforte R, Urbano A,Rozman C (1990) Immunohistochemical analysis of theimmune reaction in the nervous system in paraneoplasticencephalomyelitis. Neurology 40: 219-222

24. Graus F, Segurado OG, Tolosa E (1988) Selective con-centration of anti-Purkinje cell antibody in the CSF of twopatients with paraneoplastic cerebellar degeneration.Acta Neurol Scand 78: 210-213

25. Greenlee JE, Parks TN, Jaeckle KA (1993) Type IIa (‘anti-Hu’) antineuronal antibodies produce destruction of ratcerebellar granule neurons in vitro. Neurology 43: 2049-2054

26. Gultekin SH, Dalmau J, Graus Y, Posner JB, RosenblumMK (1998) Anti-Hu immunolabeling as an index of neu-ronal differentiation in human brain tumors: a study of 112central neuroepithelial neoplasms. Am J Surg Pathol 22:195-200

27. Hart IK, Waters C, Vincent A, Newland C, Beeson D,Pongs O, Morris C, Newsom-Davis J (1997)Autoantibodies detected to expressed K+ channels areimplicated in neuromyotonia. Ann Neurol 41: 238-246

28. Hormigo A, Dalmau J, Rosenblum M, River ME, PosnerJB (1994) Immunological and pathological study of anti-Riassociated encephalopathy. Ann Neurol 36: 896-902

29. Hormigo A, Lieberman F (1994) Nuclear localization ofanti-Hu antibody is not associated with in vitro cytotoxici-ty. J Neuroimmunol 55: 205-212

30. Jean WC, Dalmau J, Ho A, Posner JB (1994) Analysis ofthe IgG subclass distribution and inflammatory infiltratesin patients with anti-Hu-associated paraneoplasticencephalomyelitis. Neurology 44: 140-147

31. Lang B, Newsom-Davis J, Wray D (1981) Autoimmuneaetiology for myasthenic (Eaton-Lambert) syndrome.Lancet 1: 224-226

32. Lindstrom J, Shelton D, Fujii Y (1988) Myasthenia gravis.Adv Immunol 42: 233-284

33. Mason WP, Graus F, Lang B, Honnorat J, Delattre J-Y,Valldeoriola F, Antoine JC, Rosenblum MK, RosenfeldMR, Newsom-Davis J, Posner JB, Dalmau J (1997)Small-cell lung cancer, paraneoplastic cerebellar degen-eration and the Lambert-Eaton myasthenic syndrome.Brain 120: 1279-1300

34. Miralles GD, O’Fallon JR, Talley NJ (1992) Plasma-celldyscrasia with polyneuropathy. N Engl J Med 327: 1919-1923

35. Motomura M, Johnston I, Lang B, Vincent A, Newsom-Davis J (1995) An improved diagnostic assay for Lambert-Eaton myasthenic syndrome. J Neurol NeurosurgPsychiat 58: 85-87

36. Nagel A, Engel AG, Lang B, Newsom-Davis J, Fukuoka T(1988) Lambert-Eaton myasthenic syndrome IgGdepletes presynaptic membrane active zone particles byantigenic modulation. Ann Neurol 24: 552-558

37. Neumann H, Cavalie A, Jenne DE, Wekerle H (1995)Induction of MHC class I genes in neurons. Science 269:549-552

38. Newsom-Davis J, Mills KR (1993) Immunological associ-ations of acquired neuromyotonia (Isaac’s syndrome).Report of five cases and literature review. Brain 116: 453-469

39. Newsom-Davis J, Murray NM (1984) Plasma exchangeand immunosuppressive drug treatment in the Lambert-Eaton myasthenic syndrome. Neurology 34: 480-485

40. O’Neill JH, Murray NM, Newsom-Davis J (1988) TheLambert-Eaton myasthenic syndrome. A review of 50cases. Brain 111: 577-596

41. Peterson K, Rosenblum MK, Kotanides H, Posner JB(1992) Paraneoplastic cerebellar degeneration. I. A clini-cal analysis of 55 anti-Yo antibody positive patients.Neurology 42: 1931-1937

42. Posner JB (1995) Neurologic Complications of Cancer.F.A.Davis Company: Philadelphia

43. Rowland LP, Sherman WH, Latov N, Lange DJ, McDonaldTD, Younger DS, Murphy PL, Hays AP, Knowles D (1992)Amyotrophic lateral sclerosis and lymphoma: bone mar-row examination and other diagnostic tests. Neurology 42:1101-1102

44. Schold SC, Cho ES, Somasundaram M, Posner JB (1979)Subacute motor neuronopathy: a remote effect of lym-phoma. Ann Neurol 5: 271-287

45. Sillevis Smitt PA, Manley GT, Posner JB (1995)Immunization with the paraneoplastic encephalomyelitisantigen HuD does not cause neurologic disease in mice.Neurology 45: 1873-1878

46. Sillevis-Smitt P, Manley G, Moll JWB, Dalmau J, PosnerJB (1996) Pitfalls in the diagnosis of autoantibodies asso-ciated with paraneoplstic neurologic disease. Neurology46: 1739-1741

47. Sindic CJ, Andersson M, Boucquey D, Chalon MP,Bisteau M, Brucher JM, Laterre EC (1993) Anti-Purkinjecells antibodies in two cases of paraneoplastic cerebellardegeneration. Acta Neurologica Belgica 93: 65-77

48. Strassmann G, Jacob CO, Fong M, Bertolini DR (1993)Mechanisms of paraneoplastic syndromes of colon-26:involvement of interleukin 6 in hypercalcemia. Cytokine 5:463-468

49. Tanaka K, Tanaka M, Igarashi S, Onodera O, Miyatake T,Tsuji S (1995) Trial to establish an animal model of para-neoplastic cerebellar degeneration with anti-Yo antibody.2. Passive transfer of murine mononuclear cells activatedwith recombinant Yo protein to paraneoplastic cerebellardegeneration lymphocytes in severe combined immunod-eficiency mice. Clin Neurol Neurosurg 97: 101-105

50. Tanaka M, Tanaka K, Onodera O, Tsuji S (1995) Trial toestablish an animal model of paraneoplastic cerebellardegeneration with anti-Yo antibody. 1. Mouse strains bear-ing different MHC molecules produce antibodies onimmunization with recombinant Yo protein, but do notcause Purkinje cell loss. Clin Neurol Neurosurg 97: 95-100

J. Dalmau et al: Paraneoplastic Neurologic Syndromes: Pathogenesis and Physiopathology 283

51. Tora M, Graus F, de Bolos C, Real FX (1997) Cell surfaceexpression of paraneoplastic encephalomyelitis/sensoryneuronopathy-associated Hu antigens in small-cell lungcancers and neuroblastomas. Neurology 48: 735-741

52. Verschuuren J, Chuang L, Rosenblum MK, Lieberman F,Pryor A, Posner JB, Dalmau J (1996) Inflammatory infil-trates and complete absence of Purkinje cells in anti-Yo-associated paraneoplastic cerebellar degeneration. ActaNeuropathol (Berl ) 91: 519-525

53. Voltz R, Dalmau J, Posner JB, Rosenfeld MR (1998) T-cellreceptor analysis in anti-Hu associated paraneoplasticencephalomyelitis. Neurology 51:1146-1150.

54. Voltz R, Gultekin H, Posner JB, Rosenfeld MR, Dalmau J(1998) Sera of patients with testicular cancer and parane-oplastic limbic/brainstem encephalitis harbor an antineu-ronal antibody (anti-Ta) that recognizes a novel neuronalprotein. J Neurol 245: 379

55. Wilkinson PC, Zeromski J (1965) Immunofluorescentdetection of antibodies against neurons in sensory carci-nomatous neuropathy. Brain 88: 529-538

56. Younger DS, Rowland LP, Latov N, Hays AP, Lange DJ,Sherman W, Inghirami G, Pesce MA, Knowles DM,Powers J (1991) Lymphoma, motor neuron diseases, andamyotrophic lateral sclerosis. Ann Neurol 29: 78-86

J. Dalmau et al: Paraneoplastic Neurologic Syndromes: Pathogenesis and Physiopathology284