OUP user menu

Paraneoplastic limbic encephalitis: neurological symptoms, immunological findings and tumour association in 50 patients

S. Humayun Gultekin, Myrna R. Rosenfeld, Raymond Voltz, Joseph Eichen, Jerome B. Posner, Josep Dalmau
DOI: http://dx.doi.org/10.1093/brain/123.7.1481 1481-1494 First published online: 1 July 2000


Paraneoplastic limbic encephalitis (PLE) is a rare disorder characterized by personality changes, irritability, depression, seizures, memory loss and sometimes dementia. The diagnosis is difficult because clinical markers are often lacking, and symptoms usually precede the diagnosis of cancer or mimic other complications. The frequency of antineuronal antibodies in patients with PLE has not been investigated. We examined the neurological symptoms and the causal tumours in 50 patients with PLE to determine the utility of paraneoplastic antibodies and other tests. The diagnosis of PLE required neuropathological examination or the presence of the four following criteria: (i) a compatible clinical picture; (ii) an interval of <4 years between the development of neurological symptoms and tumour diagnosis; (iii) exclusion of other neuro-oncological complications; and (iv) at least one of the following: CSF with inflammatory changes but negative cytology; MRI demonstrating temporal lobe abnormalities; EEG showing epileptic activity in the temporal lobes. Of 1047 patients with neurological symptoms, whose sera or CSF were examined for paraneoplastic antibodies, 79 had the presumptive diagnosis of limbic encephalitis, dementia, cognitive dysfunction, or confusion. Fifty of these patients fulfilled our criteria for PLE. Pathological confirmation was obtained in 12 patients. The commonly associated neoplasms were of the lung (50%), testis (20%) and breast (8%). Neurological symptoms preceded the cancer diagnosis in 60% of patients (by a median of 3.5 months). Twenty-five of 44 (57%) patients with MRI studies had signal abnormalities in the limbic system. Thirty (60%) patients had antineuronal antibodies (18 anti-Hu, 10 anti-Ta, 2 anti-Ma), and 20 were antibody-negative or had uncharacterized antibodies (n = 4). The combination of symptoms, MRI findings and paraneoplastic antibodies established the diagnosis of PLE in 78% of the patients. Patients with anti-Hu antibodies usually had small-cell lung cancer (94%), multifocal neurological symptoms (78%) and a poor neurological outcome. Patients with anti-Ta (also called anti-Ma2) antibodies were young men with testicular tumours (100%), frequent hypothalamic involvement (70%) and a poor neurological outcome. In the group of patients without anti-Hu or anti-Ta antibodies, the tumour distribution was diverse, with cancer of the lung the most common (36%); 57% had positive MRI. Fifteen of 34 (44%) patients with a median follow-up of 8 months showed neurological improvement. Treatment of the tumour appeared to have more effect on the neurological outcome than the use of immune modulation. Improvement was observed in 38% of anti-Hu patients, 30% of anti-Ta patients and 64% of patients without these antibodies.

  • paraneoplastic
  • limbic
  • encephalitis
  • SCLC
  • PLE = paraneoplastic limbic encephalitis
  • SCLC = small-cell lung cancer

Based upon prior case reports and their own series, Corsellis and colleagues first described paraneoplastic limbic encephalitis (PLE) as a clinicopathological entity 30 years ago (Corsellis et al., 1968). To date, 137 patients with PLE have been reported in the English literature, 32 of them supported by autopsy (reviewed in Table 1). Most reports have emphasized the clinical and pathological involvement of the limbic structures, but the majority of patients had variable involvement of other areas of the nervous system, mainly the brainstem (Henson et al., 1954; Bakheit et al., 1990). The diagnosis of PLE is often difficult because similar symptoms (seizures, memory problems, irritability, depression, confusion and dementia) can be caused by many other cancer-related complications, including brain metastases, toxic and metabolic encephalopathies, infections (especially with herpes simplex encephalitis) and the side-effects of cancer therapy (Posner, 1995). In addition, neurological symptoms frequently precede the detection of the tumour, further confounding the diagnosis of the neurological disorder as paraneoplastic in origin.

View this table:
Table 1

One hundred and thirty-seven patients with PLE published in the English literature

ReferenceNo. of patientsTumour typeNeurological improvement with treatment (type)Tissue
*Patients 1, 2, 3, 9; patient 2; patient 2; §patients 3, 4; pathological findings did not involve the limbic structures; #patients included in the present series. **4 non-seminomatous GCT, 2 seminomas, 2 mixed GCT; ††3 patients had been published previously (1 by Ahern et al. (1994), 1 by Burton et al. (1988) and 1 by Bennett et al. (1998). GCT = germ-cell tumour.
Henson et al., 1954 4*SCLCNo0/4
Brierley et al., 1960 1SCLCNoAutopsy
Verhaart et al., 1961 1SCLCNoAutopsy
Yahr et al., 1965 1SCLCNoAutopsy
Henson et al., 1965 1SCLCNo1
Croft et al., 1965 2§SCLCNo1, autopsy
Corsellis et al., 1968 32 SCLC, 1 non-SCLCNo3, autopsy
Kaplan et al., 1974 1SCLCNoAutopsy
Markham and Abeloff, 1982 1SCLCYes, tumour (chemotherapy)None
Carr et al., 1982 1Hodgkin's diseaseYes, tumour (chemotherapy and radiation therapy)None
Gritzman et al., 1983 1Oesophageal cancerNoAutopsy
Brennan and Craddock, 1983 1SCLCYes, tumour (radiation therapy)None
Duyckaerts et al., 1985 1Hodgkin's diseaseNoAutopsy
Case records, 1985 1Bladder cancerNoAutopsy
Yamada et al., 1985 1Mixed GCT of the testisNo1
Cornelius et al., 1986 1SCLCNoAutopsy
Van Sweden and Van 1Oesophageal cancerNoAutopsy
Peteghem, 1986
Camara and Chelune, 1987 1SCLCNoAutopsy
Janati et al., 1987 1Colon cancerNoNone
Kawaguchi et al., 1988 1Mediastinal GCT withNo informationAutopsy
teratoma elements
Tandon et al., 1988 1SCLCSymptomatic (haloperidol)Autopsy
Burton et al., 1988 1#Non-seminomatous GCT of the testis (predominant embryonic carcinoma)Yes, tumour (surgery and chemotherapy)None
McArdle and Millingen, 1988 1ThymomaNoAutopsy
Den Hollander et al., 1989 2SCLC1, yes, tumourNone
(chemotherapy) 1, no improvement
Newman et al., 1990 1SCLC, renal cancerNoAutopsy
Bakheit et al., 1990 32 SCLC, 1 non-SCLCNo3 Autopsy
Dirr et al., 1990 1SCLCNoBiopsy
Ingenito et al., 1990 1ThymomaNoAutopsy
Veilleux et al., 1990 1Atypical carcinoidNoAutopsy
Lacomis et al., 1990 1Breast cancerNoBiopsy
Pfliegler et al., 1990 1Hodgkin's diseaseYes, tumour (chemotherapy)None
Brashear et al., 1991 1SCLCNoAutopsy
Kodama et al., 1991 2Breast cancer, thymomaNo2, biopsy
Amir and Galbraith, 1992 1SCLCYes, tumour (chemotherapyNone
and radiation therapy)
Dalmau et al.,199214SCLC2, yes, tumour (chemotherapy)None
Baldwin and Henderson, 1992 1SCLCNoAutopsy
Kalkman et al., 1993 1SCLCYes, tumour (chemotherapy)None
Panegyres et al., 1993 1SCLCNoAutopsy
Tsukamoto et al., 1993 1Colon cancerYes, tumour (surgery)None
Kaniecki and Morris, 1993 1SCLCYes, tumour (chemotherapy)None
Sutton et al., 1993 1SCLCNoAutopsy
Cunningham and Burt,1994 1#ThymomaYes, tumour (surgery andNone
radiation therapy)
Sakai et al., 1994 1SCLCNo informationNone
Ahern et al., 1994 1#Non-seminomatous GCTNoNone
of the testis (predominant embryonic carcinoma)
Cher et al., 1995 1#Breast cancerYes, tumour (surgery andNone
protein A immune absorption)
Antoine et al., 1995 2ThymomaYes, tumour (surgery)None
Meyer et al., 1995 1NeuroblastomaNoNone
Heidenreich et al., 1995 1SCLCNoAutopsy
Honnorat et al., 1996 2SCLCNo informationNone
Martin et al., 1996 1Non-SCLCYes, tumour (surgery)None
Zacharias et al., 1996 1SCLCNoAutopsy
Deodhare et al., 1996 1Hodgkin's diseaseYes, tumour (chemotherapy)None
Alamowitch et al.,199716SCLCYes, 5/15 treated; tumour1, autopsy
Byrne et al., 1997 1#SCLCYes, tumour (chemotherapy)None
Nokura et al., 1997 1Teratoma of the ovaryYes, tumour (surgery)None
Okamura et al., 1997 1Teratoma of the ovaryYes, tumour (surgery)None
Kaluza et al., 1997 1Testicular GCTNoAutopsy
Riva et al., 1997 1Colon cancerNo informationNone
Provenzale et al., 1998 1SCLCNo informationNone
Wingerchuk et al., 1998 2Testicular GCTNoNone
Lucchinetti et al., 199823SCLCNo informationNone
Bell et al., 1998 1Renal cell cancerYes, tumour (surgery)None
Rosenbaum et al., 1998 2Ovarian cancer,1, yes, ovarian tumour
Hodgkin's disease(chemotherapy)
1, no, Hodgkin's disease1, biopsy
Hart et al., 1998 1SCLCNoNone
Aydiner et al., 1998 1#Teratoma of the ovaryNoNone
Antoine et al., 1999 1SCLCNo informationNone
Bennett et al.,1999 1#Embryonal carcinomaNoNone
of the testis
Voltz et al., 1999 8#Testis**Yes, 3/5 treated; tumour4, biopsy
(3 published twice)††(all 3 had orchiectomy and immune modulation)
Stern and Hulette, 1999 1Small-cell carcinoma of the prostateNoAutopsy

Antineuronal antibodies, when present in the serum and CSF, facilitate the diagnosis of PLE and often allow the early detection of the associated tumour (Alamowitch et al., 1997). However, the frequency of antineuronal antibodies in patients with PLE is largely unknown (Dalmau et al., 1999a). We report the spectrum of neurological symptoms and tumours in 50 patients with PLE and analyse the utility of the detection of paraneoplastic antibodies and other diagnostic tests for this disorder.

Design and methods

The clinical and laboratory data of 79 patients with presumptive paraneoplastic disorders, including limbic encephalitis, memory problems, dementia, cognitive dysfunction or confusion, were reviewed. These patients were identified from a database of 1047 patients with suspected paraneoplastic neurological syndromes, whose sera or CSF were examined for antineuronal antibodies at Memorial Sloan-Kettering Cancer Center. Twelve patients were examined by the authors (J.B.P., J.D.). Information on the other patients was obtained from their primary physicians. The diagnosis of PLE required neuropathological examination (biopsy or autopsy), or all four of the following criteria: (i) a clinical picture of short-term memory loss, seizures, or psychiatric symptoms suggesting involvement of the limbic system; (ii) an interval of <4 years between the onset of neurological symptoms and the cancer diagnosis; (iii) exclusion of other cancer-related complications (metastasis, infection, metabolic and nutritional deficits, cerebrovascular disorder or side-effects of therapy) that may cause symptoms of limbic dysfunction; and (iv) at least one the following: CSF with inflammatory changes (pleocytosis, oligoclonal bands, increased immunoglobulin content or increased protein content in the absence of measured immunoglobulin); MRI showing unilateral or bilateral temporal lobe abnormalities on T2-weighted images or atrophy on T1-weighted images; and EEG showing slow- or sharp-wave activity in one or both temporal lobes.

Serum (and CSF when available) was examined for antineuronal antibodies using immunoblots of human neuronal protein extracts, recombinant paraneoplastic proteins (HuD, CDR62, Nova-1, Ma1 and Ma2) and immunohistochemistry on sections of human and rat cerebral cortex, as reported previously (Dalmau and Rosenfeld, 1995; Voltz, 1999). The presence of anti-CV2 antibodies was examined using immunohistochemistry with rat brain tissue fixed in paraformaldehyde (Honnorat et al., 1996).


Among 79 patients with the presumptive diagnoses of limbic encephalitis, memory problems, dementia, cognitive dysfunction or confusion, 50 fulfilled our criteria for PLE. Twelve of these 50 patients have been reported previously (Ahern et al., 1994; Cunningham and Burt, 1994; Cher et al., 1995; Alamowitch et al., 1997; Aydiner et al., 1998; Burton et al., 1997; Bennett et al., 1999; Voltz et al., 1999).

Patients with PLE

Twenty-three women and 27 men, ranging in age from 11 to 75 years (median 55 years), were included in the study. Neurological dysfunction developed before the tumour diagnosis in 29 patients (median 3.5 months, range 0.5–33 months) and after the tumour diagnosis in 19 (median 5 months, range 1–48 months). Two patients had no tumour identified during the course of the disease; one had anti-Hu antibodies and the other anti-Ma antibodies (see Discussion). The associated tumours are shown in Table 2. Cancer of the lung (50%), particularly small-cell lung cancer (SCLC) (40%), and testicular germ-cell tumours (20%) were the most frequently found neoplasms. Thirty-eight patients had localized disease and 5 had distant metastasis; no staging information was available for five patients.

View this table:
Table 2

Tumours associated with paraneoplastic limbic encephalitis

Type of tumourCurrent series (50 patients)English literature (72 patients)
*Includes 1 adenocarcinoma of the colon, 1 adenocarcinoma of the ovary, 1 chronic myeloid leukaemia and 1 plasma cell dyscrasia; includes 3 adenocarcinomas of the colon, 2 carcinomas of the oesophagus, 1 small-cell cancer of the ovary, 1 small-cell cancer of the prostate, 1 renal cancer, 1 cancer of the bladder, 1 neuroblastoma and 1 mediastinal germ cell tumour (1 patient included in the group of SCLC also had a renal cell cancer); 1 patient had anti-Hu antibodies and the other had anti-Ma antibodies.
Lung cancer25 (50%)43 (59%)
SCLC20 (40%)39 (54%)
Non-SCLC 5 (10%) 3 (4%)
Atypical carcinoid tumour 1 (1%)
Testicular germ-cell tumours10 (20%) 4 (6%)
Breast cancer 4 (8%) 2 (3%)
Hodgkin's disease 2 (4%) 5 (7%)
Immature teratoma (ovary) 2 (4%) 2 (3%)
Thymoma 1 (2%) 5 (7%)
Other 4* (8%)11 (15%)
Positive paraneoplastic antibodies without identifiable tumour 2 (4%)

Neurological symptoms developed over days or weeks in 41 (82%) patients. Slower development of symptoms, over months, occurred in 8 patients; for 1 patient the evolution of symptoms was not known.

Presenting symptoms included memory loss and confusion (24 patients), seizures (6 patients; 3 psychomotor, 1 focal facial twitching, 1 focal motor with secondary generalization, 1 generalized tonic-clonic), hypothalamic dysfunction (6 patients; 3 hypersomnia, 2 hyperthermia, 1 panhypopituitarism), psychiatric abnormalities (6 patients; 5 depression, 1 hallucinations), and symptoms of involvement of other areas of the nervous system (7 patients; 3 sensory neuronopathy, 2 cerebellar ataxia, 1 dizziness, 1 diplopia). For 1 patient, information about presenting symptoms was not available.

Table 3 shows the main symptoms during the course of the disease and the total number of patients having each symptom. Overall, 47 patients developed short-term memory loss (22 patients), seizures (25 patients) or both (20 patients). Among the 25 patients with seizures, 10 had temporal lobe or psychomotor seizures, 6 generalized seizures, and 9 a combination of seizure types. Only 3 patients did not have seizures or short-term memory loss: two developed a confusional state without epileptic activity on EEG and the other developed hypothalamic dysfunction characterized by diabetes insipidus, loss of libido and hypothyroidism. In these three patients the MRI showed mesial temporal changes compatible with PLE.

View this table:
Table 3

Clinical features

SymptomNo. of patients
*Some patients had more than one symptom.
Loss of short term memory42
Temporal lobe, `psychomotor'10
Generalized only 6
Mixture 9
Acute confusional state23
Psychiatric abnormalities21
Affective changes 7
Hallucinations 5
Disinhibition/personality changes 3
Mixed 6
Hyperthermia 4
Weight gain 2
Endocrine dysfunction 1
Hypersomnia 7
Cognitive dysfunction 7
Apraxia 3
Aphasia 1
Calculation, abstract thought 1
Visual recognition 1
Abulia 1
Cerebellar symptoms 9
Brainstem abnormalities13
Sensory loss 8
Lethargy, stupor 6
Autonomic dysfunction 4
Basal ganglion signs 2
Paresis 3
Olfactory and taste change 2
Myoclonus 1
Frontal release signs 1

Forty-nine patients had neuroimaging studies (44 MRI and 5 CT) (Table 4). The most frequent MRI abnormalities were identified on T2-weighted images and involved the mesial aspect of one (10 patients) or both (15 patients) temporal lobes (Fig. 1). In 5 patients, the mesial-temporal MRI abnormalities enhanced after contrast administration. A biopsy from the enhancing area obtained in 4 patients showed inflammatory infiltrates and neuronal loss (Fig. 2). In addition to the temporal lobe findings, 8 patients had MRI abnormalities in other areas: 4 in the brainstem, 4 in the hypothalamus, 1 in the thalamus, 1 in the cingulate gyrus and 1 in the basal frontal lobe; none of these abnormalities, except 1 in the hypothalamus, enhanced with contrast material. Three patients without temporal–limbic MRI findings had abnormalities involving other areas of the nervous system (Table 4).

View this table:
Table 4

Diagnostic tests

MRICSFParaneoplastic antibodiesEEG
*`Abnormal' indicates unilateral (10) or bilateral (15) temporal lobe abnormalities on T2-weighted sequences; in 5 of these 25 patients the lesions showed contrast enhancement. Three patients without temporal–limbic MRI findings had white matter abnormalities in other regions: 1 multifocal, 1 diffuse, and 1 perithalamic and deep white matter changes. `Abnormal' indicates inflammatory changes, including increased proteins (24/47), pleocytosis (24/47), increased IgG synthesis (15/15) and oligoclonal bands (10/13). `Abnormal' indicates temporal epileptic focus (unilateral 10; bilateral 3), periodic lateralized epileptiform discharges (2), general non-specific slowing (9) and abnormal but not specified (3). §Abnormal findings include temporal abnormalities (5) detailed in the text, multifocal white matter lesions (1), cerebral or cerebellar atrophy (3) and frontal lobe metastasis (1). Detailed in the text. #These include diffuse, non-specific slowing (3) and posterior triphasic waves (1).
50 patientsAbnormal*: 28Abnormal: 40Positive: 30Abnormal: 27
with PLENormal: 16Normal: 9(18 Hu, 10 Ta, 2 Ma)Normal: 6
Not available: 6Not available: 1Atypical: 4Not obtained: 17
(5 had CT study)Negative: 16
29 patientsAbnormal§: 10Abnormal: 10Atypical: 3Abnormal#: 4
without PLENormal: 12Normal: 5Negative: 26Normal: 5
Not available: 7Not available: 14Not available: 20
Fig. 1

MRI findings in PLE. Note the bright signal (arrows in A and B) in the medial aspect of the temporal lobes. In A, the more intense signal in the right temporal lobe corresponds to oedema after a brain biopsy (shown in Fig. 2).

Fig. 2

Pathological findings in a patient with anti-Ta-associated limbic encephalitis. Biopsy of the temporal lobe showing perivascular and parenchymal inflammatory infiltrates (A; haematoxylin–eosin, ×200) and multiple perineuronal infiltrates of CD8+ T-cells

(B; diaminobenzidine–haematoxylin, ×400).

The CSF was examined in 49 patients, and 40 showed one or more of the following abnormalities: 24 of 47 had pleocytosis (>5 cells/mm3; range 9–100 cells/mm3, median 29); 24 of 27 had elevated proteins (>45 mg/dl; range 47–170, median 74); 10 of 13 had oligoclonal bands; and 15 of 15 had intrathecal synthesis of IgG. Only 7 patients had increased proteins as an isolated finding. None of 49 patients had malignant cells in the CSF (Table 4).

Paraneoplastic antibodies were identified in the serum, CSF, or both, of 30 patients (60%): in 18 patients they were anti-Hu antibodies, in 10 they were anti-Ta antibodies (also known as anti-Ma2 antibodies) and in 2 they were anti-Ma antibodies (Figs 3 and 4). Four patients had antineuronal antibodies that have not been characterized: the serum from a patient with breast cancer reacted with several proteins in the 30–40 kDa range expressed predominantly in the cytoplasm of hippocampal pyramidal cells; the serum from a patient with non-SCLC reacted with several proteins in the range of 40–60 kDa expressed in the cytoplasm of neurons and Purkinje cells; the serum from a patient with ovarian cancer showed selective diffuse reactivity with the neuropil, but was negative in immunoblot studies; and the serum of a patient with non-SCLC showed faint labelling of the neuronal cytoplasm with negative immunoblot findings. Sixteen patients had no antineuronal antibodies. None of the sera contained anti-CV2 antibodies.

Fig. 3

Immunohistochemical analysis of antineuronal antibodies in PLE. Sections of normal human brain immunolabelled with normal human serum (A) and serum from patients with PLE and anti-Hu antibodies (B), anti-Ta (also called anti-Ma2) antibodies (C) and atypical antineuronal antibodies (D).

Fig. 4

Immunoblot analysis of human neuronal proteins with serum from patients with PLE. Lanes correspond to the same sera examined by immunohistochemistry in Fig. 3. Lane NHS corresponds to serum from a normal individual.

Overall, 39 (78%) patients had temporal lobe abnormalities on MRI studies and/or positive paraneoplastic antibodies in their serum. When the abnormal findings in one or both of these tests were combined with the detection of abnormal CSF, the number of patients with at least two abnormal tests was 49. The remaining patient, with Hodgkin's disease, had brain biopsy findings supporting the diagnosis of PLE.

On the basis of the presence or absence of anti-Hu and anti-Ta antibodies, we established three immunological groups of PLE patients. For practical purposes the two patients with anti-Ma antibodies and the 4 patients with non-characterized antineuronal antibodies were included in the group of patients without antibodies (Table 5). In 6 patients (3 anti-Hu, 3 anti-Ta) the detection of these antibodies led to the diagnosis of the paraneoplastic syndrome and assisted in identifying the underlying tumour.

View this table:
Table 5

Immunological subsets of paraneoplastic limbic encephalitis

Clinical featuresAnti-Hu (18 patients)Anti-Ta (10 patients)No antibodies or other antibodies* (22 patients)
*Sixteen patients had no antineuronal antibodies, 2 had anti-Ma antibodies and 4 had antineuronal antibodies that are uncharacterized. The remaining 13 patients had adenocarcinoma of the breast (4), teratoma (2), Hodgkin's disease (2), plasma cell dyscrasia (1), malignant thymoma (1), adenocarcinoma of the colon (1), chronic myeloid leukaemia (1) and ovarian cancer (1). §Includes limbic dysfunction and symptoms involving brainstem and/or one or more of the following: basal ganglia; cerebellum; dorsal root ganglia; cerebral cortex; spinal cord; autonomic nervous system. Irrespective of the treatment used.
Sex 9 M, 9 F10 M 8 M, 14 F
Age range (median) (years)28–70 (61.5)22–45 (34)11–75 (57)
Predominant tumour16 SCLC10 testicular 8 lung (n = 4 SCLC)
(no tumour found)(1)(1)
Limited stage tumour14/15 9/1015/18
Neurological symptoms14 7 8
before tumour diagnosis
Limbic alone 4 313
Limbic–brainstem alone 2 5 0
Outside limbic–brainstem§12 2 9
Hypothalamic deficits 2 7 2
Psychiatric symptoms 5 610
Abnormal MRI 7/14 9/912/21
Abnormal CT 1/2 0/1 0/2
Abnormal CSF16/18 8/1016/21
Final neurological outcome: improvement 5/13 3/10 7/11

Of 18 patients with anti-Hu antibodies, 16 had SCLC, 1 was reported to have adenocarcinoma of the lung (the tumour was not examined by us), and no tumour was identified in another patient. All patients with anti-Ta antibodies had testicular germ-cell tumours. In the group of patients without antibodies or with atypical antibodies, 8 (36%) had cancer of the lung (4 SCLC), 4 adenocarcinoma of the breast, 2 immature teratoma, 2 Hodgkin's disease, 1 plasma cell dyscrasia, 1 malignant thymoma, 1 adenocarcinoma of the colon, 1 chronic myeloid leukaemia and 1 ovarian cancer (Table 5).

Seventy-eight per cent of patients with PLE and anti-Hu antibodies had symptoms suggesting dysfunction of areas of the nervous system distant from the limbic system (brainstem, cerebellum, dorsal root ganglia, cerebral cortex, spinal cord and autonomic nervous system) (Table 5). These areas were less likely to be involved in patients without antibodies or with atypical antibodies (41%).

Compared with the anti-Hu patients (median age 61.5 years), the anti-Ta patients were younger (median age 34 years) and had more restricted involvement of the limbic system and brainstem (80%); only 2 patients had other symptoms (1 with involvement of the basal ganglia and 1 with cerebellar dysfunction). As part of the limbic symptoms, 70% of these patients had severe hypothalamic dysfunction (clinical details of these patients have been reported elsewhere (Voltz et al., 1999). Unlike patients with anti-Hu antibodies, the anti-Ta patients did not develop signs of dorsal root ganglion involvement or myelitis.

Pathological examination of the brain was obtained in 12 patients: 8 had a brain biopsy and 4 an autopsy. In all instances, the main findings were perivascular cuffing and interstitial infiltrates of lymphocytes, microglial proliferation, gliosis and neuronal degeneration (Fig. 2). Three autopsy studies had results of detailed neuropathological examination showing chronic encephalitis without viral inclusion bodies that was not restricted to the limbic regions.

Clinical follow-up was available for 34 patients (median 8 months, range 1–84 months) (Table 6). Treatment of the tumour appeared to be associated with improvement of the neurological syndrome in 11 of the 15 patients (73%) in whom this assessment was possible. Eighteen patients received one or more immune modulatory treatments for the paraneoplastic disorder: 12 corticosteroids (60–100 mg prednisone), 4 cyclophosphamide, 4 intravenous immunoglobulin, 5 plasma exchange, and 1 removal of plasma IgG by immunoadsorption with a protein-A column. Four of these 18 patients (22%) had partial neurological improvement closely related to the dates when they were treated with immune modulation, although all 4 had simultaneous treatment of the tumour.

View this table:
Table 6

Neurological outcome in 34 patients with PLE

Treatment of the tumour only1121
Immune modulation only 004
Treatment of the tumour and immune modulation 464
No treatment 002

Irrespective of the treatment used, neurological symptoms progressed in 11 patients (resulting in death in 6), remained stable in 8 and improved in 15 (14 partial recovery, 1 full recovery). Improvement was observed in 5 of 13 (38%) patients with anti-Hu antibodies, 3 of 10 (30%) with anti-Ta antibodies and 7 of 11 (64%) without detectable antibodies (Tables 5 and 6).

Patients without PLE

The sera of 29 patients affected with memory loss, cognitive dysfunction, or seizures, but who did not fulfil our strict criteria for PLE, were examined for antineuronal antibodies. There were 16 men and 13 women with a median age of 64.5 years (range 29–92 years). Thirteen patients had cancer: neurological symptoms developed before the tumour diagnosis in 5 patients (median 12 months, range 1 month to 6 years) and after the tumour diagnosis in 8 (median 3.5 years, range 1–19 years). Tumours included SCLC in 2 patients, non-SCLC tumours in 2 patients, renal cell cancer in 2 patients, breast cancer in 1 patient, adenocarcinoma of the parotid gland in 1 patient, testicular germ-cell tumour in 1 patient, lymphoma in 1 patient, multiple myeloma in 1 patients and lung nodules (without biopsy) in 2 patients. No systemic tumour was identified in the other 16 patients after a median follow-up of 12 months (range 1 month to 8 years).

CSF was examined in 15 patients: 4 showed inflammatory abnormalities, 6 showed increased proteins only and 5 were normal (Table 4). Three patients harboured atypical antibodies that reacted with neurons; none of these antibodies was similar to any of the 4 atypical antibodies found in the PLE patients. The other 26 patients were negative for antibodies.

MRI was obtained in 22 patients. Temporal lobe abnormalities were identified in 5 patients, whose final diagnoses were: herpes simplex encephalitis in 2 patients (1 associated with glioma); metastasis of renal cell cancer in 1 patient; bilateral temporal lobe abnormalities with frontoparietal and periventricular white matter changes in 1 patient with non-caseating bone marrow granulomatosis of uncertain significance; and bilateral temporal lobe T2-weighted abnormalities with contrast enhancement in 1 patient with primary CNS lymphoma, who eventually developed involvement of the left basal ganglia and occipital lobe.

The final diagnoses of the patients without MRI abnormalities in the temporal lobes included 1 of each of the following: viral encephalitis; primary CNS lymphoma; Creutzfeldt–Jakob disease; ischaemic cerebrovascular disease; psychiatric disorder with pseudodementia; frontal lobe metastasis; and intracranial hypertension due to remote subarachnoidal haemorrhage; and 17 patients had primary degenerative dementia (either of the Alzheimer type or of unknown cause).


In 1960, Brierley and colleagues described three patients suffering from `subacute encephalitis of later adult life mainly affecting the limbic areas' (Brierley et al., 1960). At autopsy, the first patient had a small leiomyoma of the left kidney but no other evidence of malignancy. The second patient had several mediastinal and hilar lymph nodes `extensively replaced by oat-shaped cells, appearing identical to those seen in anaplastic bronchial carcinoma'; no tumour was found in the lungs or elsewhere in the body. The third patient had `an encapsulated mass at the root of the right lung' which consisted of fibrotic lymph nodes. Although one of the authors had previously reported patients with a `mental disorder associated with primary lung carcinoma' (Charatan and Brierley, 1956), they believed that it was `most unlikely that this finding [lung cancer in patient 2] is in any way related to the encephalitis but its occurrence should be noted'. They also noticed that a lung cancer had been clinically suspected in patient 3 but they failed to consider that the fibrotic lymph nodes may have represented a remitting tumour. In 1961, Verhaart described two patients with inflammatory lesions of the medial temporal lobe and nerve cell loss in the brainstem and cerebellum, both of whom had tumour in the mediastinal lymph nodes but no identifiable pulmonary lesion (Verhaart, 1961). In 1962, Störring and colleagues described a patient with `undifferentiated bronchial carcinoma and an encephalitic-like picture in the limbic system' (Störring et al., 1962). In 1965, Yahr and colleagues reported a patient with `encephalopathy associated with carcinoma' who had SCLC and limbic system inflammatory changes (Yahr et al., 1965). In 1967, Ulrich and colleagues reported a patient with SCLC and neuronal degeneration in Ammon's horn and elsewhere with some perivascular lymphocytic cuffing (Ulrich et al., 1967).

The name limbic encephalitis was coined by Corsellis and colleagues (Corsellis et al., 1968) when they reported three additional patients and extensively reviewed the prior literature, including those patients first reported by Brierley and colleagues (Brierley et al., 1960). This description of limbic encephalitis generated more reports, and by 1990 Bakheit and colleagues were able to report three new patients with PLE, and reviewed the extant literature, which consisted of 16 cases that had been verified clinically and pathologically (Bakheit et al., 1990). Once the relationship between cancer and the nervous system lesion had been established, three pathogenic hypothesis were advanced: (i) a (not further defined) degeneration of the nervous system in which the inflammatory infiltrates were a secondary `reaction to the tissue breakdown' (Verhaart, 1961; Ulrich et al., 1967); (ii) a viral infection (Corsellis et al., 1968); and (iii) an immune-mediated response against the nervous system (Russell, 1961). The third hypothesis is the currently accepted one.

Despite the now fairly large number of patients reported, the diagnosis of PLE remains difficult. The presenting symptoms may be different from those considered typical of the disorder (i.e. short-term memory loss, seizures, and mood or behavioural changes). In 27% of our patients these symptoms were not part of the presentation of the paraneoplastic disorder. Usually, patients are not known to have cancer, as occurred in 60% of our patients. Symptoms resembling PLE are frequent in cancer patients and may result from multiple different metastatic and non-metastatic complications (Clouston et al., 1992; Hosaka and Aoki, 1996; Spiegel, 1996; Porta-Etessam, 1999). In fact, a 1956 report of a `mental disorder associated with primary lung carcinoma' described three patients who probably suffered from hepatic encephalopathy (Charatan and Brierley, 1956). With the exception of the anti-Hu antibody in some patients with SCLC and the anti-Ta antibody in patients with testicular tumours, there are no known serological markers of PLE. In the current study, 60% of the patients had paraneoplastic antibodies in their serum (see below). Identification of these antibodies, when combined with characteristic MRI findings, helped establish the diagnosis of PLE in 78% of the patients.

Symptoms of involvement of areas of the nervous system distant from the limbic system (particularly the brainstem and cerebellum) are frequent in patients with PLE. This occurred in 60% of our patients and similar findings were recognized by Bakheit and colleagues, who found that only 6 of 19 (32%) patients had isolated limbic encephalitis (Bakheit et al., 1990).

Typical MRI findings of PLE include unilateral or bilateral mesial temporal lobe abnormalities that are best seen on T2-weighted images. On T1 sequences, the temporal–limbic regions may be hypointense and atrophic, and may sometimes enhance with contrast injection. These abnormalities, although well known (Dirr et al., 1990; Lacomis et al., 1990), had not been examined in a large series of patients. In our study, 64% of the patients with PLE had abnormal MRI studies, which in 89% of the cases showed the changes indicated above. Patients with herpes simplex encephalitis may have similar MRI findings in the early stages of the disease. However, these patients usually develop prominent signs of oedema and mass effect involving one or both inferior–medial temporal lobes, the inferior frontal lobes and the cingulate gyrus. In addition, gyral enhancement is present in more than half of the patients and signs of haemorrhage are not uncommon (Demaerel et al., 1992; Kapur et al., 1994).

CSF analysis assists in making the diagnosis of PLE in two ways. First, a negative cytological analysis for malignant cells in combination with the absence of meningeal enhancement on the MRI helps exclude leptomeningeal metastases. Secondly, the detection of inflammatory abnormalities (pleocytosis, intrathecal synthesis of IgG, oligoclonal bands) supports the diagnosis of an inflammatory or immune-mediated neurological disorder; 64% of our patients had one or more of these abnormalities. We did not find any patient with antibodies present only in the CSF; however, two patients had barely detectable serum anti-Ta titres, while the CSF titres were several orders of magnitude higher (data not shown).

The EEG has limited usefulness in making the diagnosis of PLE, but it is useful in assessing whether changes in the level of consciousness or behaviour are related to temporal lobe seizures. In a review of 19 patients, Bakheit and colleagues identified temporal lobe seizures in only one patient (Bakheit et al., 1990). In our study, the incidence of EEG abnormalities was limited by the fact that only patients in whom seizures were suspected received EEG evaluation. Nevertheless, among the 35 patients with EEG studies, 45% had epileptic activity.

Symptoms resembling PLE may result from several disorders, including systemic lupus erythematosus (Glanz et al., 1998; Stubgen et al., 1998), Wernicke–Korsakoff encephalopathy with or without cancer (De Reuck et al., 1981; Engel et al., 1991), toxic effects of doxifluridine (an antineoplastic agent) (Heier and Fossa, 1986) and herpes simplex encephalitis (Aimard et al., 1979; Perentes and Herbort, 1984). Two of our patients with PLE were initially considered to have herpes simplex encephalitis and received treatment with acyclovir without neurological improvement. However, patients with herpes simplex encephalitis usually develop acute or subacute mental confusion and seizures that often lead to stupor or coma. The MRI findings (see above) and the presence of red blood cells in the CSF point to the diagnosis of herpes simplex encephalitis, which diagnosis can be confirmed by PCR (polymerase chain reaction) analysis of the CSF or brain biopsy (Cinque et al., 1996).

Symptoms and pathological findings of limbic encephalitis have been reported in the absence of cancer (Brierley et al., 1960; Langston et al., 1975; Horoupian and Kim, 1982; Daniel et al., 1985; Kohler et al., 1988; Kepes et al., 1990). We believe that in most cases a cancer was present but not detected. In one of our patients with PLE and life-threatening neurological deterioration, the family consented to right orchiectomy because of the detection of serum anti-Ta antibodies and ultrasound findings of a mild abnormality in the right epididymis (consistent with epididymitis). The original pathological evaluation was negative for testicular cancer, but detailed review of the slides demonstrated microscopic intratubular germ-cell tumour (Dalmau et al., 1999b).

We compared the histological types of tumours of our patients with those reported in the English literature, bearing in mind two considerations. First, 53 of 137 reported patients come from three studies of patients selected on the basis of suffering from SCLC or harbouring anti-Hu antibodies (Dalmau et al., 1992; Alamowitch et al., 1997; Lucchinetti et al., 1998). Secondly, 12 of our patients have been reported previously. Thus, after excluding these 65 patients from Table 1, the remaining 72 patients were compared with the 50 patients of the current study (Table 2). This comparison revealed a preferential association of PLE with certain tumours: SCLC, germ-cell tumours of the testis, breast cancer, Hodgkin's lymphoma, immature teratoma and thymoma.

The reason for the preferential association of PLE with certain tumours is unknown. SCLC and testicular germ-cell tumours express a number of proteins (including the Hu and Ma proteins) that in normal adult tissues are restricted to neurons and germ cells of the testis (Carpentier et al., 1998; Dalmau et al., 1999c; Voltz et al., 1999). In addition, these tissues are immunoprivileged by the presence of endothelial barriers and atypical or absent expression of antigen-presenting molecules (MHC class I) (Haas et al., 1988; Neumann et al., 1997; Corriveau et al., 1998). These findings suggest a mechanism whereby the tumour expression of brain/testis proteins is the trigger of autoimmunity against the nervous system. This hypothesis is supported by our study, in which SCLC and testicular germ-cell tumours were consistently associated with antineuronal antibodies in patients with PLE. Immunological disturbances have also been identified in patients with PLE and thymoma (Antoine et al., 1995), a tumour frequently related to other autoimmune neurological disorders, including myasthenia gravis and neuromyotonia (Newsom-Davis and Mills, 1993; Drachman, 1994).

Thirty of our patients (60%) had positive paraneoplastic antibodies (anti-Hu, anti-Ta or anti-Ma). We took into consideration a possible bias of serum referral to our laboratory for anti-Ta antibody testing. However, only 2 of the 10 patients with positive anti-Ta antibodies were included in the study after the discovery of these antibodies (Voltz et al., 1998); the other 8 were retrospectively identified from the 1047 patients in the database.

On the basis of the presence of paraneoplastic antibodies, we established three immunological subsets of PLE (Table 5) that differ from each other by the age of the patients, the type of tumour association and some clinical features. The most distinctive of these include the presence of symptoms from areas outside the limbic system in most anti-Hu patients and the frequent hypothalamic and brainstem involvement in anti-Ta patients. In addition, PLE patients with anti-Ta antibodies are more likely to have abnormal MRI findings than other PLE patients.

It is important to note that the group of patients `with no antibodies or with other antibodies' is an ill-defined, heterogeneous group that includes patients without antineuronal antibodies, and 6 patients with antibodies other than anti-Hu and anti-Ta (4 non-characterized and 2 anti-Ma). One anti-Ma patient was a woman, without a cancer diagnosis, who developed confusion, lethargy and T2 MRI abnormalities involving the medial temporal lobes, hypothalamus and upper midbrain; the other was a woman with breast cancer, typical symptoms of PLE, and T2 MRI abnormalities in the medial temporal lobes. Unlike other anti-Ma sera that react similarly with both Ma1 and Ma2 proteins (Dalmau et al., 1999c), the sera and CSF of these two patients showed more reactivity with Ma2 than Ma1 (data not shown). This finding and the data from patients with anti-Ta antibodies (which recognize Ma2) (Voltz et al., 1999) suggest that Ma2 contains epitopes related to autoimmunity associated with limbic dysfunction.

In contrast to most paraneoplastic syndromes of the CNS, which do not improve with treatment, PLE may respond to therapy. In our study, 44% of patients with a median follow-up of 8 months showed neurological improvement regardless of the type of treatment. Treatment of the tumour appeared to have more effect on neurological improvement than the use of immunosuppressive therapies. We found similar results in the literature review of 137 patients (Table 1). Information about the response to treatment was available for 107 patients; 25 (23%) had neurological improvement, in 24 the improvement was attributed to tumour treatment, and in 1 it was attributed to symptomatic treatment (haloperidol). These findings, coupled with the high frequency of limited stage disease (88% of our patients, irrespective of antineuronal antibody status), underscore the need for prompt detection and treatment of the tumour, which may improve the neurological deficits and control the cancer.

We are aware that some of our findings may result from the design of the study. To explore this possible bias, we examined the clinical information of a group of 29 patients with memory loss, cognitive dysfunction or confusion (considered to be paraneoplastic) whose sera were sent to us for antineuronal antibody testing. When compared with our group of 50 patients with PLE, the two main discriminating diagnostic tests were the MRI studies and antineuronal antibody testing. None of the 29 patients had paraneoplastic antibodies, and only 2 had MRI abnormalities similar to those identified in PLE; one had herpes simplex encephalitis, and the other a primary CNS lymphoma that eventually showed multifocal areas of enhancement.

From this and previous studies (Table 1), it appears likely that PLE is an immune-mediated disorder that can be caused by several types of tumour-induced autoimmunity. When PLE is suspected, the following tests should be considered: (i) MRI of the brain without and with contrast; (ii) paraneoplastic antibody testing of serum and CSF; (iii) CSF studies to rule out the presence of metastatic cells and demonstrate the presence of inflammatory abnormalities (oligoclonal bands, intrathecal synthesis of IgG, pleocytosis); and (iv) EEG studies, particularly in patients with an acute confusional state of unknown cause. Patients older than 40 years, particularly if they are smokers, should be examined for serum and CSF anti-Hu antibodies; the detection of this antibody confirms that the limbic syndrome is paraneoplastic and indicates that the tumour is probably SCLC. In patients older than 40 years without paraneoplastic antibodies, the tumours most frequently encountered are non-SCLC, carcinoma of the breast and thymoma; however, the absence of anti-Hu antibodies does not rule out the possibility that the underlying tumour is SCLC. Patients younger than 40 years, particularly if they are male, should be studied for serum and CSF anti-Ta antibodies; the detection of these antibodies indicates that the disorder is paraneoplastic and that the tumour is likely to be located in testis. In patients younger than 40 years and without paraneoplastic antibodies, the tumours most frequently involved are Hodgkin's lymphoma and immature teratoma. Regardless of the antineuronal antibody status, MRI studies show T2 abnormalities in mesial temporal regions in 57% of the patients; these findings, although not pathognomonic, are highly suggestive of PLE, particularly if they are associated with inflammatory changes in the CSF. Although we did not find anti-CV2 antibodies, which are not consistently associated with PLE, there are 3 reported patients (2 with SCLC and 1 with thymoma) in whom detection of anti-CV2 antibodies helped to establish the paraneoplastic origin of a limbic dysfunction (Antoine et al., 1995; Honnorat et al., 1996).

From our study and review of the literature, no definitive treatment is available for PLE. However, early identification and treatment of the tumour is the approach that appears to offer the greatest chance for neurological improvement.


We thank Drs James Miller, Nicolas Miret, Ian Sutton, Francesc Graus, Antoine Carpentier, Jerome Honnorat, Jean-Yves Delattre, Charles F. Bolton, Fred Hochberg, Casilda Balmaceda, Hakan Gurvit, Joseph Landolfi, Michael Homer-Ward, Lisa Rogers, Louisa Thoron, Jeffrey Raizer, Bruce Cree, Daniel Park, Ronald Kanner, Lawrence Haber, Jon Wilson and Nuno Antunes, for providing clinical information and Dr Victor Reuter for the pathological evaluation of the tumour of one of the patients. This work was supported by NIH grant NS-26064 (J.B.P, J.D.) and the ACS Evelyn Frew Clinical Research Professorship (J.B.P.).


  • * Present address: Neurologische Klinik, Klinikum Grosshadern, D-81366 München, Germany

  • ** Present address: Department of Neurology, University of Arkansas for Medical Sciences, Little Rock, Arkansas, USA


View Abstract