Antiamphiphysin Antibodies Are Associated With Various Paraneoplastic Neurological Syndromes and Tumors FREE

SEVERAL TYPES of autoantibodies have been identified in patients with paraneoplastic neurological syndromes; these include antiamphiphysin antibodies that recognize a 128-kd protein in Western blots of rat and human brain subsequently shown to be the nerve terminal protein amphiphysin.^{1 - 2} Antiamphiphysin antibodies were first reported in women with breast cancer and stiff-man syndrome,¹ but studies indicate that they also occur in patients with other tumors and neurological disorders.^{3 - 4} Using a series of 5 patients, we have now extended the spectrum of neurological syndromes and tumors that are associated with antiamphiphysin antibodies and suggest that these antibodies are not specific for one type of neurological syndrome and one type of cancer. Also, we found that other neural and nonneural antibodies were present in several patients with antiamphiphysin antibodies, thus making it difficult to detect such antibodies by immunohistochemical analysis alone.

The serum samples from the 5 patients and all control patients were snap frozen and stored at −80°C until tested. The control group consisted of 40 patients with various nonparaneoplastic neurological diseases; 101 patients with cancer but without paraneoplastic neurological syndrome (37 with SCLC, 25 with gynecological cancer, 25 with colon adenocarcinoma, and 14 with non-SCLC lung tumors); 8 patients with SCLC, anti-Hu antibodies, and paraneoplastic neurological syndrome; 3 patients with M₂-type antimitochondrial antibodies but no neurological disorder; and 30 normal subjects. All serum samples were immunohistochemically tested as described below. For Western blot experiments, the serum samples from the 5 patients and from the controls (2 normal subjects, 1 patient with antimitochondrial antibodies, 16 patients with nonparaneoplastic neurological disorders, 7 patients with SCLC and 7 with breast carcinoma and no neurological disorder, and 8 patients with SCLC, paraneoplastic neurological syndrome, and anti-Hu antibodies) were tested using recombinant human amphiphysin and rat cerebellum extracts.

The method used to detect antiamphiphysin antibodies was slightly modified from that described elsewhere for the detection of paraneoplastic antibodies⁵ and conforms to the guidelines for the detection of such antibodies.⁶ Immunohistochemical testing was performed on frozen cerebellar and brainstem sections from brains removed from anesthetized rats perfused in vivo with 4% paraformaldehyde and 0.2% picric acid and postfixed for 24 hours in the same solution. The sections were incubated for 2 hours at room temperature with the serum samples and diluted at least 100-fold in phosphate-buffered saline and blocking goat serum; then, bound IgG was detected using fluorescein isothiocyanate–conjugated goat antibodies specific for the Fab‘2 fragment of human IgG (Sanofi-Pasteur, Paris, France). The antibody titer was determined by end point dilution. In some experiments, bound autoantibodies were detected by an indirect immunoperoxidase method, as previously described.⁵

Western blot analysis was performed using 12.5% acrylamide minigels in a semiautomatic apparatus (PhastSystem; Pharmacia Biothech, Orsay, France). The test proteins used were either a sample of fresh rat cerebellum, homogenized in potassium phosphate buffer containing Triton X-100 and methyl-phenyl sulfonyl fluoride, or a sample of recombinant human amphiphysin prepared as described previously.⁷ A serum sample from a rabbit immunized with recombinant amphiphysin protein⁷ was used as a positive control. Recombinant HuD fusion protein was used for the detection of anti-Hu antibodies. For Western blotting, 1 mg of protein was applied to each lane. After transfer of the proteins, the membranes were incubated with bovine serum albumin to saturate nonspecific protein binding sites and then for 12 hours with the test serum diluted at least 100-fold. Bound antibodies were detected by incubating the blots with either biotinylated rabbit anti–human IgG antibodies (Jackson, Baltimore, Md) or with pig anti–rabbit IgG antiserum (Dakopatts, Glostrup, Denmark), followed by streptavidin-peroxidase (Jackson). The color reaction was developed using diaminobenzidine.

Antimitochondrial antibodies were detected by screening serum samples from the patients and controls immunohistochemically on frozen sections of rat kidney, liver, and stomach for nonneural antibodies using the procedures previously described. M₂-type antimitochondrial antibodies demonstrate a typical granular cytoplasmic staining of the distal renal tubules, mainly of the medullary, hepatocytes, and gastric parietal cells.^{8 - 9} In our laboratory, a sample is considered positive at a titer of greater than 1:10.

The 5 patients described below were selected from 2800 patients whose serum samples were screened in our laboratories (Saint-Etienne and Lyon, France) for paraneoplastic antibodies.

In April 1994, a 75-year-old woman developed sensory loss and pain in her lower limbs. On examination in December 1994, the absence of pallesthesia was noted in both legs. Pain and thermal sensation were almost normal. There was no motor deficit, and tendon reflexes were absent. Electrophysiological tests showed absent sensory responses in both legs. Motor conduction velocities were normal. The cerebrospinal fluid specimen contained 0.69-g/L protein, with a normal level of IgG and 1 lymphocyte per cubic millimeter. The patient's neurological status progressively deteriorated, and she developed cerebellar dysarthria, truncal ataxia, slight memory loss, and epilepsy. She had no stiffness. A computed tomographic scan of the brain showed no tumoral lesion or cerebellar atrophy. A breast adenocarcinoma, discovered in May 1996, was treated with surgery and radiotherapy. During the following months, the ataxia progressed and then stabilized. In September 1997, the patient was able to walk using a Zimmer frame.

In August 1996, a 55-year-old male smoker complained of mood changes and became abnormally anxious and depressed. His neurological status deteriorated progressively, and he had olfactory and auditory hallucinations. He developed severe anterograde memory loss, but his retrograde memory was normal. In February 1997, electroencephalographic tests showed epileptic discharges in the temporal lobes. The results of cerebrospinal fluid analysis were normal. T₂-weighted magnetic resonance images of the brain showed high signals in both hippocampi, especially the left. There was no stiffness. After 24 months of follow-up, a lymph node metastasis of small cell lung cancer (SCLC) was discovered in the mediastinum.

In December 1994, a 50-year-old woman complained of gait disturbance and paresthesia in her hands. Severe proximal and distal motor deficit with hand amyotrophy, diffuse areflexia, and a bilateral Babinski sign progressively developed. Severe sensory loss, affecting all modalities of sensation, appeared in all 4 limbs. There was no stiffness. The cerebrospinal fluid specimen contained 1.10-g/L protein with a normal level of IgG and 1 lymphocyte per cubic millimeter. Electromyographic tests showed denervation potentials in all 4 limbs. Motor conduction velocities were normal, and sensory conduction velocities were not measurable. An adenocarcinoma of the ovary, discovered in May 1995, was treated with surgery, radiotherapy, and chemotherapy. During the following months, the patient's neurological status progressively deteriorated, and she became virtually bedridden. Respiratory insufficiency necessitated intermittent mechanical respiration.

In January 1993, a 67-year-old man complained of progressive weakness in his legs. In February 1994, a moderate proximal motor deficit and depressed tendon reflexes were noted in both legs. Electrophysiological tests showed reduced motor action potentials in the lower limbs, with a 210% improvement after brief maximal exercise, a 40% decrease after stimulation at 3 Hz, and a 158% improvement after stimulation at 20 Hz, indicating Lambert-Eaton myasthenic syndrome (LEMS). The results of screening for antibodies to the P/Q-type voltage-gated calcium channel (VGCC) were positive. An SCLC was discovered in March 1994, and the patient received chemotherapy and radiotherapy, without improvement. He was discharged in May 1994 and was unavailable for follow-up until May 1995, at which time he was bedridden. There was severe truncal and appendicular cerebellar ataxia with dysarthria, overshadowing the symptoms of LEMS. The tendon reflexes were depressed. A computed tomographic scan of the brain showed severe cerebellar atrophia. There was no stiffness. The tumor was still localized to the mediastinum.

In 1972, A 73-year-old man was treated for carcinoma of the tongue with surgery and radiotherapy. In June 1996, he complained of increasing gait difficulties. Examination in July 1997 showed a proximal and distal motor deficit, with absence of both Achilles tendon reflexes. The patient's mouth was dry. There was no stiffness. Electrophysiological test results were indicative of LEMS, showing reduced motor action potentials in the lower limbs, with a 500% improvement after brief maximal exercise, a 20% decrease after stimulation at 3 Hz, and a 250% improvement after stimulation at 20 Hz. Screening for antibodies to the P/Q-type VGCC proved positive. There was no recurrence of the tongue carcinoma, and a computed tomographic scan of the chest showed a mediastinal tumor, identified as SCLC on surgical biopsy.

Of the 2800 serum samples routinely tested for paraneoplastic autoantibodies, 5 demonstrated labeling of the rat cerebellum and brainstem suggestive of antiamphiphysin antibodies. On Western blots using recombinant human amphiphysin, all 5 labeled a doublet of bands with an apparent molecular weight of 125 to 128 kd, identical to those recognized by the rabbit antiamphiphysin antiserum (Figure 1). When tested on Western blots of brain cerebellum, all 5 serum samples labeled a 128-kd band probably corresponding to amphiphysin. As with the control serum samples, additional bands were sometimes seen. The immunological data for these patients are summarized in Table 1.

When tested immunohistochemically on rat cerebellum and brainstem, serum samples from patients 1, 3, and 5 demonstrated a pattern typical of antiamphiphysin antibody. In the cerebellum, there was diffuse staining of the neuropil in the molecular layer, the cytoplasm of the Purkinje cells being less intensively labeled, and intense granular staining was seen in the periphery of the perikarya of the Purkinje cells and the granule cell layer, suggesting staining of the synaptic terminals (Figure 2, A). In the brainstem, fiber tracts were labeled and large neurons showed the same pattern of staining as Purkinje cells, whereas glial cells were never stained. Serum samples from patients 2 and 5 demonstrated a somewhat different pattern. In addition to the labeling typical of antiamphiphysin antibody, granular cytoplasmic staining of neurons and glial cells was seen, with diffuse labeling of the neuropil. When tested immunohistochemically on rat kidney, liver, and stomach, these 2 serum samples demonstrated labeling characteristic of M₂-type antimitochondrial antibody (titer, 1:160 and 1:2500), while the serum samples from the 3 patients with M₂-type antimitochondrial antibody, but no neurological disease, demonstrated granular labeling in the neuropil and cytoplasm of neurons and glial cells ( Figure 2, B). Thus, some of the granular cytoplasmic staining seen in the serum samples from patients 2 and 5 may be due to antimitochondrial antibodies. Also, a second serum sample from patient 2, obtained 11 months after the first antibody determination, showed nuclear labeling of neurons, with sparing of the nucleoli, suggestive of anti-Hu antibodies; this was confirmed by Western blotting using recombinant HuD protein and whole cerebellar extracts (Figure 3).

On immunohistochemical testing, none of the 182 control serum samples demonstrated labeling suggestive of antiamphiphysin antibodies. On Western blot analysis using recombinant human amphiphysin, none of the 57 control serum samples proved positive, except samples from 3 of 8 patients with anti-Hu antibodies.

Although antiamphiphysin antibodies have been reported in several patients, to our knowledge detailed clinical data are available in only 6 cases.^{1 ,4} In 1993, Folli et al¹ described 3 women with breast carcinoma who had contracture and stiffness of the limbs and continuous motor unit activity at rest on electromyographic testing, consistent with stiff-man syndrome. The trunk was usually spared, and oligoclonal bands of IgG were present in the cerebrospinal fluid specimens from 2 of the patients. Their serum samples reacted with a protein doublet of 128 kd on Western blots of human cortex. In 1993² and 1994,⁷ 2 more patients were described by the same group as having stiff-man syndrome and serum samples that were reactive with the 128-kd protein, now identified as amphiphysin. In 1993, Grimaldi et al³ mentioned 2 patients with stiff-man syndrome and antibodies that reacted with a 125-to 130-kd protein on Western blots of total rat brain homogenate; one had carcinoma of the colon and the other Hodgkin disease. In 1996, Dropcho⁴ reported the detailed clinical description of 3 patients with SCLC and antibodies that reacted with human amphiphysin: 2 had subacute sensory neuronopathy (SSN) and paraneoplastic encephalomyelitis (PEM), and 1 had stiffness and myoclonus associated with symptoms of encephalomyelitis. Finally, in 1997, Lennon et al¹⁰ described, in abstract form, 11 patients with antibodies that reacted with recombinant human amphiphysin; these patients did not have stiff-man syndrome, but they did have PEM with various features, including limbic encephalitis, choreoathetosis, myoclonus, stiffness, and rigidity. The tumors involved were breast or lung carcinomas; 2 patients had no known cancer.

Our patients extend the spectrum of neurological syndromes associated with antiamphiphysin antibodies. Two patients (Nos. 4 and 5) had SCLC and LEMS with a typical electrophysiological pattern and anti-VGCC antibodies; in patient 4, paraneoplastic cerebellar degeneration appeared in the following months. The association of LEMS and paraneoplastic cerebellar degeneration has been reported in several other patients and is probably not fortuitous, suggesting a pathophysiological link between the 2 disorders.¹¹ Patient 1 had breast cancer, SSN with the electrophysiological abnormalities of SSN, and central nervous system manifestations of PEM. Patient 3 had ovarian carcinoma and developed severe sensory-motor neuropathy, with electrophysiological abnormalities indicative of a neuronopathy. The presence of pyramidal syndrome showed that she had encephalomyelitis. Although rare, the combination of SSN and PEM has already been reported in patients with gynecological cancer¹² and is a classic paraneoplastic disorder. In the last patient (No. 2), the clinical and radiological data indicated limbic encephalitis that was associated with SCLC. Taken as a whole, the 26 patients described with antiamphiphysin antibodies (including our 5) show that these antibodies occur in association with several types of tumors, with gynecological cancer occurring in only 9 of the 26 patients. The clinical features are also heterogeneous. However, stiffness was present in 12 of the 26 patients, but it should be clarified whether a true stiff-man syndrome or a PEM with rigidity and spasm,¹³ or both, can be associated with antiamphiphysin antibodies.

ANTIAMPHIPHYSIN ANTIBODIES contrast with the well-known anti-Hu and anti-Yo antibodies, which show a strong correlation with the type of neurological disorder and the type of cancer. Most patients with anti-Hu antibodies have SCLC and SSN/PEM, while anti-Yo antibodies are strongly associated with paraneoplastic cerebellar degeneration and gynecological tumors. This observation, together with the intracellular localization of amphiphysin, is evidence against a direct role of antiamphiphysin antibodies in neurological disorders. Such a conclusion is further supported by the fact that 2 of our patients had LEMS, a disorder of the neuromuscular junction thought to depend on anti-VGCC antibodies.¹⁴ It is possible that the humoral response directed against amphiphysin may simply be an epiphenomenon of the autoimmune process that directly mediates the neurological disease.

The current hypothesis concerning paraneoplastic neurological disorders is that the autoimmune response is directed primarily against tumor antigens that are also present in the nervous system.¹⁵ Dropcho⁴ reported that SCLC and breast carcinomas in patients without neurological disorders can express amphiphysin messenger RNA. Thus, patients with tumors may be prone to develop antiamphiphysin antibodies. However, Folli et al¹ found no antiamphiphysin antibodies in 30 control patients with cancer, and we found no labeling suggestive of antiamphiphysin antibodies in 101 patients with SCLC, gynecological cancer, or colon carcinoma. Thus, the incidence of antiamphiphysin antibodies in patients with cancer, but without paraneoplastic disorders, is probably low. So far, these antibodies have only been seen in patients with neurological disorders consistent with paraneoplastic syndromes; therefore, whatever their pathogenic role, they are useful tools for the diagnosis of tumors in patients with suspected paraneoplastic disorders.

Many laboratories use routine immunohistochemical testing as a screening method for the detection of paraneoplastic antibodies. In 2 of our patients, the typical labeling of antiamphiphysin antibodies was significantly modified by the presence of M₂-type antimichondrial antibodies or anti-Hu antibodies, emphasizing the need to test serum samples with an unusual staining pattern on noncerebral tissues and by Western blotting using recombinant protein or brain homogenates.^{6 ,16} Three of our 5 patients had several autoantibodies in their serum samples, and of these 3 patients, 1 had antiamphiphysin, anti-VGCC, and antimichondrial antibodies, and 1 had anti-Hu, anti-mitochondrial, and antiamphiphysin antibodies. Antiamphiphysin antibodies have also been reported in association with anti-Hu⁴ and anti–glutamic acid decarboxylase antibodies,¹⁷ and we found that the serum samples from 3 of 8 patients with PEM/SSN and anti-Hu antibodies reacted with recombinant amphiphysin. These data are in agreement with those of reports indicating that patients with paraneoplastic neurological disorders can sometimes produce several autoantibodies that react with the nervous system¹⁸ or that non–organ-specific antibodies are seen in patients with cancer¹⁹ or paraneoplastic neurological syndromes.²⁰ Patients with cancer, especially those with paraneoplastic neurological disorders, are probably prone to develop a complex state of autoimmunity leading to the simultaneous production of several antibodies, both specific and nonspecific for the nervous system.

Accepted for publication March 24, 1998.

This study was supported in part by La Ligue Contre le Cancer de la Loire et du Rhône, La Fondation de France and l'Association pour la Recherche Contre le Cancer.

The screening for antibodies to the P/Q-type VGCC was performed by Dr J. Newsom-Davis, Oxford University, Oxford, England, in case 4 and by Dr N. Martin-Moutot, Marseille, France, in case 5. The recombinant HuD fusion protein was kindly provided by Dr J. Dalmau, Sloan Kettering Cancer Center, New York, NY.

Reprints: Jean Christophe Antoine, MD, Service de Neurologie, Hôpital de Bellevue, Boulevard Pasteur, 42,055 Saint-Etienne Cedex, France.