|
|
| |||||||||||||||||||||||||||||||||||||||||||||||
Background In patients with cancer, symptoms of limbic and brain-stem dysfunction may result from a paraneoplastic disorder. Paraneoplastic limbic or brain-stem encephalitis occurs more frequently with testicular cancer than with most other cancers. We sought antineuronal antibodies that might be used in a diagnostic test for this syndrome.
Methods Immunohistochemical and immunoblotting techniques were used to detect serum and cerebrospinal fluid antibodies. Serologic screening of a complementary DNA library and Northern blotting were used to clone the target antigen and determine which tissues expressed it.
Results Of 13 patients with testicular cancer and paraneoplastic limbic or brain-stem encephalitis (or both), 10 had antibodies in serum and cerebrospinal fluid against a 40-kd neuronal protein. These antibodies were used to clone a gene that we call Ma2, which codes for a protein (Ma2) that was recognized by serum from the 10 patients, but not by serum from 344 control subjects. Ma2 was selectively expressed by normal brain tissue and by the testicular tumors of the patients. Ma2 shares homology with Ma1, a "brain–testis–cancer" gene related to other paraneoplastic syndromes and tumors.
Conclusions The serum of patients with subacute limbic and brain-stem dysfunction and testicular cancer contains antibodies against a protein found in normal brain and in testicular tumors. Detection of these antibodies supports the paraneoplastic origin of the neurologic disorder and could be of diagnostic importance.
Paraneoplastic limbic encephalitis is a syndrome consisting of irritability, depression, seizures, severe memory deficit, and dementia.4 These symptoms are due to dysfunction of the limbic system (hippocampus, amygdala, hypothalamus, and insular and cingulate cortex), which is the area of the nervous system where most of the pathological changes that characterize the syndrome occur. Brain-stem encephalitis and abnormalities in other areas of the nervous system are frequent, but they may be clinically silent.5,6
Paraneoplastic limbic encephalitis is probably underdiagnosed, because of the diversity of its symptoms and the lack of specific diagnostic markers. In patients known to have cancer, symptoms of this paraneoplastic syndrome may be attributed to metastases to the brain, toxic or metabolic encephalopathies, infections, or toxic effects of cancer therapy.2 In approximately 60 percent of patients with paraneoplastic limbic or brain-stem encephalitis, the syndrome precedes the detection of the tumor, further complicating the diagnosis of the neurologic disorder.6,7,8 Abnormalities involving the mesial temporal lobes on magnetic resonance imaging (MRI) studies and the finding of changes due to inflammation in the cerebrospinal fluid (pleocytosis, increased levels of proteins, and oligoclonal bands) may be suggestive of paraneoplastic limbic encephalitis but do not establish the diagnosis.9
In 80 percent of patients with paraneoplastic limbic encephalitis, the primary tumor is a small-cell lung cancer, and about half of these patients have antibodies against the Hu family of neuronal RNA-binding proteins (human homologues of the drosophila embryonic lethal abnormal visual, or elav, protein) expressed in the brain and the tumor.8,10,11 With the exception of these antibodies, there are no other serologic markers of paraneoplastic limbic encephalitis, and the diagnosis relies on brain biopsy or is made at autopsy. In the remaining 20 percent of patients with the syndrome, testicular cancer occurs more frequently than expected. This observation, together with the detection of an antibody against a novel neuronal antigen in a patient with testicular cancer and paraneoplastic limbic encephalitis,12 led us to investigate other cases of this syndrome in patients with testicular tumors.
Methods
Serum and Tissue Samples
We analyzed serum samples (or cerebrospinal fluid samples when available) from 986 men and women with histologically proved cancer that had been sent to us for antineuronal-antibody testing. In addition, we obtained 344 serum samples for use as controls from patients with other paraneoplastic syndromes, patients with a variety of cancers without such syndromes, patients with multiple sclerosis or systemic lupus erythematosus, and normal subjects.
Tumor tissues were provided by referring physicians and the Tumor Procurement Service at Memorial Sloan-Kettering Cancer Center. They included testicular tumors from 4 patients with paraneoplastic limbic encephalitis, brain-stem encephalitis, or both; tumors from 85 patients without paraneoplastic syndromes (65 with testicular germ-cell tumors, 5 with colon cancer, 4 with breast cancer, 3 with lung cancer, 2 with parotid-gland cancer, and 6 with small-cell lung cancer); and tumors from 8 patients with other paraneoplastic syndromes (4 with small-cell lung cancer, 3 with ovarian cancer, and 1 with bladder cancer). Tissue from normal subjects and Wistar rats was processed and stored as reported elsewhere.13,14 For Western blot analysis, tissues were homogenized in 0.1 percent Nonidet P-40 and protease inhibitors.15
Immunohistochemical Analysis
Frozen sections of rat and human tissues that were 7 µm thick were fixed in 10 percent formalin or a combination of 30 percent methanol and 70 percent acetone at 4°C and incubated with a sample of serum, IgG, or cerebrospinal fluid from a patient with use of an avidin–biotin–peroxidase immunoassay.14,15 A monoclonal antibody against human CD8+ T cells (Dako, Carpinteria, Calif.) was used to examine the immunophenotype of the inflammatory infiltrates in brain-biopsy specimens.10
To avoid interference with endogenous IgG in the immunohistochemical studies with human tissue, we used IgG that had been purified from patients' serum samples and labeled with biotin.16 The same IgG was used for immune-competition assays: two serum samples were considered to be competing for the same epitopes when preincubation of the tissue with one sample abrogated the reactivity of the IgG isolated from the other sample.
Intrathecal Synthesis of Antibodies against Ma2 Onconeuronal Antigen
The presence of intrathecal synthesis of antibodies against an onconeuronal antigen that we have called Ma2 (also called anti-Ta) was assessed according to Schüller's formula.17 A ratio of Ma2 antibody reactivity in cerebrospinal fluid to Ma2 antibody reactivity in serum of more than 2 indicates that there is intrathecal synthesis of Ma2 antibody.
Cloning, Isolation, and Sequence Analysis of Ma2 Complementary DNA
Serum from a patient with paraneoplastic brain-stem dysfunction was plated at a density of 5x104 pfu per 150-mm plate and screened with a 
ZAP human cerebellar phage library (Stratagene, La Jolla, Calif.).13 Positive phage colonies were purified by several rounds of antibody screening, followed by subcloning into a pBluescript vector according to the phage-rescue protocol (Stratagene).
Double-stranded Ma2 complementary DNA (cDNA) was purified with the Qiagen plasmid midi-prep system (Qiagen, Santa Clarita, Calif.) and sequenced on an automated DNA sequencer (model ABI377, Applied Biosystems, Foster City, Calif.) with use of the dye-labeled terminator fluorescence method.18
Western Blot Analysis
Recombinant fusion proteins, Escherichia coli proteins, and proteins from human and rat tissues were obtained as described previously,10,14 subjected to 10 percent sodium dodecyl sulfate–polyacrylamide-gel electrophoresis, and transferred to nitrocellulose strips. The nitrocellulose strips were incubated with the patients' serum (dilution, 1:1000) and assessed for reactivity by an enhanced chemiluminescence assay (Amersham, Arlington Heights, Ill.).13
Northern Blot Analysis
The following sequence-specific oligonucleotide probes were used: Ma2, 5'GGGAATGGCCGAGACATC3' (cDNA base pairs, 234 to 217); Ma1, 5'GAAACCCAAGGACACGGG3' (cDNA base pairs, 647 to 630); and ß-actin, 5'GTCTTTGCGGATGTCCACG3'. Probes were end-labeled, purified, and hybridized to human I and II multiple-tissue Northern blots (Clontech, Palo Alto, Calif.) as described previously.13
Results
Patients
Among the 986 patients with cancer whose serum samples we examined for antibodies against onconeuronal antigens, 19 had testicular cancer and a paraneoplastic syndrome. Of these 19 patients, 13 had symptoms of limbic or brain-stem dysfunction (or both), and 10 had antibodies against an onconeuronal antigen we have named Ma2.
Table 1 shows the clinical features of the 10 patients with antibodies against Ma2. Of these 10 patients, 8 had paraneoplastic limbic encephalitis. Four patients had symptoms of brain-stem encephalitis, two of whom also had limbic encephalitis.
|
All 10 patients underwent orchiectomy, 5 received chemotherapy, and 1 received radiation therapy. After treatment of the testicular tumor, four patients had neurologic improvement (two of whom had a clinical remission), the neurologic status remained stable in three and deteriorated in one, and two died (one from complications of chemotherapy and the other as a result of the neurologic disease). In some patients the neurologic symptoms were treated with corticosteroids, plasma exchange, or intravenous immune globulin alone or in combination. In only one of these patients was treatment (corticosteroids and intravenous immune globulin) followed by improvement.
Three patients who had testicular tumors but no anti-Ma2 antibodies had symptoms suggestive of a paraneoplastic syndrome: two had brain-stem and cerebellar dysfunction, and the other had transient memory loss and confusion. Neurologic symptoms developed in these patients 3 months before and 12 and 24 months after the diagnosis of testicular cancer. In contrast to patients with anti-Ma2 antibodies, in whom the findings on MRI of the brain and studies of cerebrospinal fluid were usually abnormal, these three patients had normal results on MRI of the brain, and one (with brain-stem and cerebellar symptoms) had changes indicative of inflammation in the cerebrospinal fluid.
Characterization of Antibodies against Ma2
Serum samples from the 10 patients with anti-Ma2 antibodies reacted on Western blotting of an extract of purified human neurons with a 40-kd protein (Figure 1A). Cerebrospinal fluid samples were available from six of the patients, and all samples also reacted with the protein. No patient had antibodies exclusively in cerebrospinal fluid. The pattern of reactivity of antibodies in the 10 serum samples was examined immunohistochemically with the use of frozen human and rat tissues fixed in methanol–acetone and biotin-labeled purified IgG from each patient's serum. Under these conditions, all neurons of the human and rat brain, spinal cord, dorsal-root ganglia, intestinal autonomic neurons, and adrenal medullary ganglion cells showed discrete subnuclear and cytoplasmic immunoreactive structures (Figure 1B). Purkinje cells and other neurons of the cerebellar cortex had the weakest reactions. No reactivity was identified in liver, lung, and other non-neuronal tissues.
|
Cloning of the Gene Encoding the Ma2 Antigen
Screening of serum containing anti-Ma2 antibodies with a ZAP human cerebellar phage library allowed the isolation of a positive clone, which was recovered by subcloning into a pBluescript vector. After purification, a plasmid (p561A) was isolated that contained an insert of 614 bp. The sequence of this insert included an incomplete open reading frame corresponding to 195 amino acids, with a predicted molecular mass of 21.9 kd. We called this gene Ma2 (GenBank accession number AF037365) because of its partial homology with Ma1 (GenBank accession number AF037364), a gene that codes for an antigen associated with another paraneoplastic neurologic syndrome.13 A search of Genbank data bases revealed a gene that had 60 percent homology with Ma2 and that had been cloned from adult mouse testis (GenBank accession number AA498105).
Antibodies against Recombinant Ma2 in Serum and Cerebrospinal Fluid
Recombinant Ma2 protein was expressed in E. coli from the p561A plasmid.10,13 With the use of Western blots containing the recombinant Ma2 fusion protein, all 10 serum samples (Figure 2A) and all 6 cerebrospinal fluid samples from patients with paraneoplastic limbic or brain-stem encephalitis (or both) reacted with a band of approximately 30 kd. There was no reactivity with the control protein (an extract of E. coli containing pBluescript without an insert). None of 344 control serum samples reacted with recombinant Ma2.
|
In five patients, the ratio of Ma2 antibody reactivity in cerebrospinal fluid to that in serum was measured and was 0.74, 4.4, 6.2, 16.9, and 23.5, indicating intrathecal synthesis of the antibodies in four patients.17
Expression of Ma2 by Normal Brain and by Testicular Tumors
Northern blot analysis of messenger RNA (mRNA) from normal human tissues showed that Ma2 mRNA occurs in brain but not in placenta, lung, liver, spleen, thymus, prostate, ovary, testis, small intestine, colon, or peripheral-blood leukocytes (Figure 3). In brain tissue the mRNA was present as a single transcript of approximately 6.5 kb. Immunohistochemical and Western blot analysis of the same tissues, with biotinylated IgG from patients with Ma2 antibodies used as a probe, showed reactivity (presumably with Ma2) only in brain (data not shown).
|
|
We have previously described four patients with paraneoplastic neurologic syndromes (brain-stem encephalitis and cerebellar degeneration) and serum antibodies (called anti-Ma) against neuronal proteins of 37 and 40 kd.13 These anti-Ma antibodies were used to clone Ma1, which codes for a 37-kd protein in brain and normal testis. Because of the similarities between the nucleotide sequences of Ma2 and Ma1, we examined whether serum containing anti-Ma or anti-Ma2 antibodies reacted with both of the Ma proteins. Serum containing anti-Ma2 antibodies reacted exclusively with Ma2, but serum containing anti-Ma antibodies reacted with both Ma1 and Ma2 proteins. Preincubation of serum containing anti-Ma antibodies with either of these proteins did not abrogate its reactivity with the other protein, indicating that Ma1 and Ma2 contain distinctive epitopes. In addition, preincubation of rat-brain sections or immunoblots of neuronal or recombinant Ma2 proteins with serum containing anti-Ma antibodies decreased but did not abolish the reactivity of these blots with anti-Ma2 IgG antibodies. These findings suggest that some Ma2 epitopes are recognized by both types of antibodies (data not shown). The clinical and immunologic findings associated with the presence of anti-Ma and anti-Ma2 antibodies are summarized in Figure 5.
|
We found that serum and cerebrospinal fluid from 10 of 13 patients with testicular cancer and paraneoplastic limbic or brain-stem encephalitis (or both) contained antibodies against a 40-kd neuronal protein. A recombinant preparation of this protein (called Ma2, also known as Ta)21 was bound by serum samples from all 10 patients, but not by serum samples from patients with testicular cancer who did not have a paraneoplastic syndrome. Moreover, these antibodies reacted not only with the neuronal protein but also with the patients' testicular-tumor tissue.
Antineuronal antibodies in the serum of patients with this paraneoplastic syndrome were used to clone the Ma2 gene. Ma2 was found to resemble Ma1, a gene that we previously identified using antibodies (called anti-Ma) in serum from patients with paraneoplastic cerebellar or brain-stem dysfunction (or both) associated with lung, breast, parotid-gland, or colon cancer.13 Anti-Ma antibodies react with a 37-kd neuronal protein (Ma1)13 and a 40-kd protein, which we have identified as Ma2. Unlike the anti-Ma antibodies, which recognize both proteins, anti-Ma2 antibodies react only with Ma2 (Figure 5).
Ma2 and Ma1 are most likely members of a novel gene family that includes KIAA0883, a gene cloned from the brain of an adult (GenBank accession number AB020690). The KIAA0883 gene is almost identical to Ma2, but the protein it encodes contains additional 3' sequences that have homology with the sequence of the corresponding region of Ma1. Whether Ma2 represents a truncated or alternatively spliced form of KIAA0883 is not known.
The function of the Ma1 and Ma2 proteins is unknown, but they are both target antigens in diseases that are probably initiated by an immune response to neuronal proteins expressed by tumors. The intrathecal synthesis of anti-Ma2 antibodies in four of five patients whom we studied indicates that there is an immunologic response against Ma2 within the nervous system of these patients.17 The absence of intrathecal synthesis of anti-Ma2 antibodies in one patient may have resulted from treatment with corticosteroids and intravenous immune globulin 10 days before testing, which ameliorated the neurologic symptoms and abnormalities on MRI.
We do not know whether anti-Ma2 antibodies, an associated cytotoxic T-cell response, or both cause the neurologic disease. The inflammatory infiltrates of one patient showed neurons closely surrounded by CD8+ T cells, suggesting that cytotoxic T cells may be the effectors of the neuronal damage. Three of the six families of previously identified "cancer–testis" antigens (MAGE, BAGE, and GAGE) were originally identified through the use of cytotoxic T cells to define antigens expressed by the tumor cells of one patient. The other three families of antigens (SSX2, NY-ESO-1, and SCP1) were identified from recombinant cDNA expression libraries with the use of serum samples from patients with cancer.22
Our findings suggest that patients with symptoms of paraneoplastic limbic or brain-stem encephalitis, particularly if they are young men, should be examined for serum antibodies against Ma2. Detection of these antibodies supports the diagnosis of a paraneoplastic syndrome and guides the search for the tumor to the testis.
Supported in part by grants from the National Institutes of Health (NS-26064, to Drs. Posner and Dalmau) and the Deutsche Forschungsgemeinschaft Habilitationsstipendium (Vo 497/2-1). Dr. Posner holds the Evelyn Frew American Cancer Society Clinical Research Professorship.
Memorial Sloan-Kettering Cancer Center has an agreement with Athena Diagnostics (Worcester, Mass.) licensing it to use the Ma2 protein for diagnostic testing.
We are indebted to Drs. G.L. Ahern, S.L. Galetta, P.W. Benedetto, G. Stoll, M. Pause, C.J.M. Sindic, M. Wick, and M.K. Schwartz for providing serum samples and clinical information; to Drs. F. Graus and C. Cordón-Cardo for critical review of the manuscript and for providing serum samples; and to R. Hoard and T. DesChamps for technical assistance.
Source Information
From the Department of Neurology and the Cotzias Laboratory of Neuro-Oncology (R.V., S.H.G., M.R.R., E.G., J.E., J.B.P., J.D.) and the Department of Pathology (S.H.G.), Memorial Sloan-Kettering Cancer Center, New York.
Address reprint requests to Dr. Dalmau at the Department of Neurology, Memorial Sloan-Kettering Cancer Center, 1275 York Ave., New York, NY 10021, or at dalmauj{at}mskcc.org.
References
| |||||||||||||||||||||||||||||||||||||||||||||||
Related Letters:
A Serologic Marker of Paraneoplastic Limbic and Brain-Stem Encephalitis in Patients with Testicular Cancer
Debourdeau P., Gligorov J., Zammit C., Dalmau J., Gultekin H. S., Posner J. B.
Extract |
Full Text
N Engl J Med 1999;
341:1475-1476, Nov 4, 1999.
Correspondence
This article has been cited by other articles:
HOME | SUBSCRIBE | SEARCH | CURRENT ISSUE | PAST ISSUES | COLLECTIONS | PRIVACY | HELP | beta.nejm.org Comments and questions? Please contact us. The New England Journal of Medicine is owned, published, and copyrighted © 2008 Massachusetts Medical Society. All rights reserved. |
Previous
-32P–labeled oligonucleotide probes specific for Ma2, Ma1, or ß-actin revealed a single Ma2 transcript of 6.5 kb only in brain tissue, whereas a single Ma1 transcript of 2.6 kb was expressed by both brain and testis tissue.
