Journal ofNeurology, Neurosurgery, and Psychiatry 1993;56:361-364 361

A characteristic ganglioside antibody pattern in J Neurol Neurosurg Psychiatry: first published as 10.1136/jnnp.56.4.361 on 1 April 1993. Downloaded from the CSF of patients with amyotrophic lateral sclerosis

Andreas Stevens, Michael Weller, Horst Wietholter

Abstract Materials and methods Paired and serum Thirty five patients with amyotrophic lateral samples of patients with amyotrophic sclerosis, 16 men and 19 women, mean age lateral sclerosis (n = 35) revealed no con- 53-5 (2 4) years, range 25-79 years at diag- sistent abnormalities of CSF cell count, nosis, without a family history of the illness, CSF albumin, CSF IgG, CSF IgM, IgG were included in this study. The diagnosis or IgM index, or oligoclonal immuno- was clinical and based on the presence of globulin band formation in the CSF. both upper and symp- Determination of IgG and IgM CSF and toms and signs. Eight patients had prominent serum antibodies to gangliosides GM1, bulbar signs, 31 had definite spasticity. For GM2, GM3, AGMI, GDla, GDlb, and inclusion in this study, the onset of symptoms GTlb showed a characteristic pattern had to be insidious and not started before the which allowed the differentiation of age of 20. The clinical course had to be amyotrophic lateral sclerosis from con- progressive. Specific attention was given to trols and from patients with other neuro- alternative diagnoses of cervical spondylosis, logical disorders including multiple neoplastic lesions of cervical sclerosis. Specifically, patients with the and brain stem, polymyositis, peripheral disease had elevated CSF IgM antibodies neuropathy, progressive muscle atrophy, to all gangliosides except AGMI. The Charcot-Marie-Tooth disease, spinocerebel- lack of correlation between the CSF lar degeneration and familial spastic paraple- findings and corresponding serum anti- gia. CSF and serum samples were obtained bodies suggests a chronic, compart- after informed consent, centrifuged and mental, intrathecal immune response of stored at - 70°C. low activity in amyotrophic lateral scle- CSF parameters and ganglioside antibodies rosis. Whether this immune response is were estimated as previously described."'"3 primary and of pathogenetic signifi- Briefly, micro-enzyme linked immunosorbent cance, or an epiphenomenon of neuronal assay (ELISA) plates (M129A, Dynatech, degeneration, remains to be investigated. Denkendorf, Germany) were coated with purified ganglioside antigen, GM1, GM2,

GM3, AGM1, GDla, GDlb, or GT1b http://jnnp.bmj.com/ (J Neurol Neurosurg Psychiatry 1993;56:361-364) (150 ng per well, Sigma, St Louis, MO, United States) diluted in methanol. Antigen purity was confirmed by thin layer chro- Amyotrophic lateral sclerosis is characterised matography. Protein binding sites on the well by a progressive loss of upper and lower that might cause non-specific binding of motor neurons of unknown aetiology.' immunoglobulins were blocked by incubation Cerebrospinal fluid studies are scarce and with bovine serum albumin (0-1%) in phos- have produced no consistent abnormalities in phate-buffered saline for 24 hours before the on September 29, 2021 by guest. Protected copyright. or serum samples. The Department of patients with this disease.2' Several authors incubation of CSF , University have postulated that a disturbance of CSF samples were matched to an IgG con- of Tubingen, immunoregulation may play a role in the centration of 5 mg/l. Serum was diluted 1:50 Germany motor neurons in these in saline. No detergents A Stevens degeneration of phosphate-buffered H Wietholter cases.4-7 The detection of polyclonal serum were used throughout the assay. IgG or IgM National Institute of antibodies to GM1 ganglioside in such ganglioside reactive antibodies were labelled Mental Health, patients89 has reinforced the search for by anti-human IgG (Sigma A-5403) or anti- Clinical Neuroscience immunological events in the pathogenesis of human IgM (Sigma A-3914) antisera conju- Branch, Section on Molecular the disease. As the pathological changes are gated to alkaline phosphatase diluted 1:1000 Pharmacology, largely confined to the central in phosphate-buffered saline. Monoclonality Bethesda, Maryland, and because there is no significant blood- of positive samples was excluded by differen- United States on for K or chains as previously M Weller brain barrier disturbance, studies the CSF tial assays i. light are more likely to establish immune dysfunc- described. 12 Correspondence to: Dr Andreas Stevens, tion than investigation of the systemic Seven internal reference sera obtained Department of Psychiatry, response. This reports an from healthy volunteers or seven CSF sam- University of Tubingen, immune study Osianderstrasse, 7400 analysis of CSF and serum IgG and IgM anti- ples from patients with suspected, but even- Tubingen, Germany bodies to seven different gangliosides in 35 tually disproven, neurological disease served Received 24 October 1991 sclerosis as a control for interassay variability. The and in final revised form patients with amyotrophic lateral 24 July 1992 and examines their correlation with routine interassay correlation calculated for nine runs Accepted 31 July 1992 CSF parameters. was 0 91. The ratio of the optical density of a 362 Stevens, Weller, Wietholter

Table I Mean normalised optical density readings (SD) for ganglioside antibodies in sclerosis of chronic progressive (n = 13) and J Neurol Neurosurg Psychiatry: first published as 10.1136/jnnp.56.4.361 on 1 April 1993. Downloaded from CSF (adjusted to S mg/l IgG) and serum (diluted 1:50) ofpatients with amyotrophic lateral sclerosis and controls with other neurological diseases. relapsing remitting type (n = 35). The statis- tical analysis included testing for normal Controls ALS CSF OND CSF ALS serum OND serum serum distribution, Student's t test, linear correla- tion, discriminant analysis, and analysis of GMl-IgG 1-09 (0-38) 1-00 (0-15) 1-08 (0 28) 1-00 (0 14) variance. GMI-IgM 1-17 (0-17)* 1-06 (0-15) 1 09 (0 34) 1-00 (0 28) GM2-IgG 0-94 (0-14) 1-04 (0-11) 1 16 (0 21)* 1-00 (0 16) GM2-IgM 1 22 (0 25) 1 00 (0 14) 1 12 (0 36) 1 00 (0-28) GM3-IgG 0.91 (0-18) 1 00 (0 11) 1 11 (0-27)* 1 00 (0 19) GM3-IgM 1-30 (0 37)* 1-00 (0 11) 1 13 (0 33) 1 00 (0 25) Results AGMI-IgG 0 94 (0 17) 0.99 (0-14) 1 21 (0Q34)* 1-00 (0 20) The routine laboratory investigations of CSF AGMl-IgM 1 08 (0 16) 0 99 (0 06) 1 10 (0 34) 1-00 (0 21) GD1a-IgG 1-05 (0-17) 1-01 (0-12) 1-06 (0-18) 1-00 (0-15) and serum samples from amyotrophic lateral GDla-IgM 1-22 (0 28)* 0 98 (0 08) 1-00 (0-31) 1-00 (0 27) sclerosis did not consistent abnormali- GDlb-IgG 095 (0-16) 1-00 (012) 106 (018)* 100 (013) yield GDlb-IgM 1-57 (0 49)* 1-00 (0-06) 1 13 (0 39) 1 00 (0 26) ties. None of the patients had a paraprotein GTIb-IgG 1-18 (0 25) 1 00 (0-13) 1.20 (0 20) 1-00 (0 13) on serum immune GTIb-IgM 1-47 (0 40)* 1 00 1 electrophoresis. Patients (0-10) 1 05 (0 32) 00 (0 26) with the disease did not have higher total ALS = amyotrophic lateral sclerosis; OND = other neurological diseases. serum IgG or IgM levels than controls. There *Significantly (p < 0.03) elevated readings in patients with amyotrophic lateral sclerosis compared with those with other neurological diseases. were also no significant differences between patients and controls in CSF cell counts, CSF albumin, CSF IgG, CSF IgM, or IgG or Table 2 Incidence ofpathological CSF and senum ganglioside antibodies (normalised IgM index. Seven patients had elevated CSF optical density readings beyond the third percentile) in patients with amyotrophic lateral sclerosis. IgG, one had elevated CSF albumin, six had mild pleocytosis. Two had CSF IgG oligo- CSF IgG IgM Serum IgG IgM clonal immunoglobulin n % n % n % n % bands detected. CSF IgG and IgG index in these patients were GM1 3 9 5 14 5 14 4 11 GM2 1 3 10 29 5 14 3 9 normal. GM3 2 6 21 60 3 9 5 14 The results of CSF and serum ganglioside AGM1 0 0 9 26 5 14 6 17 GDla 3 9 13 37 2 6 1 3 antibody determination in amyotrophic lat- GDlb 1 3 27 77 5 14 5 14 eral sclerosis and the incidence of patho- GTIb 10 29 24 69 13 37 2 6 logical normalised optical density readings in these patients are provided in tables 1 and 2. The controls with other neurological diseases patient sample over the mean optical density were not different from the internal controls. of the seven internal standards was calcu- CSF IgG, but not age or gender, had an lated, thus arbitrarily assigning the value of influence on ganglioside antibodies in those one to this mean. These normalised optical with other neurological diseases, before density results obtained in amyotrophic later- adjustment. The Pearson coefficients were al sclerosis were then compared with a con- 014 (GM1), 0-56 (GM2), 037 (GM3), 0-23 trol group of patients with other neurological (AGMI), 0-48 (GDla), 0 47 (GDlb) and diseases (n = 24) such as tension or 0 35 (GTIb). Among the seven ganglioside normal pressure . Patients with antibodies examined, there was considerable neuroinflammatory and neuroimmunological intercorrelation in the amyotrophic lateral disorders were excluded. The means of the sclerosis samples (Pearson coefficient http://jnnp.bmj.com/ actual optical density readings in the group 0 3-0-6), suggesting that most of the sera with other neurological diseases at 30 min- contain either antibodies with multiple gan- utes were between 0-072 and 0-079 for the glioside reactivity or multiple antibodies seven CSF IgM ganglioside antibodies, against different gangliosides, a suggestion between 0138 and 0-191 for CSF IgG, that was confirmed by twin layer chromatog- between 0-294 and 0-462 for serum IgM, and raphy. The degree of normalised optical den- between 0265 and 0 394 for serum IgG.

sity elevation among the patients with the on September 29, 2021 by guest. Protected copyright. The results in amyotrophic lateral sclerosis disease was fairly heterogeneous (figs 1 and were also compared with two groups of previ- 2). The two patients with CSF oligoclonal ously characterised' patients with multiple bands did not differ in their ganglioside Figure 1 Individual 2.5 2.5 normalised optical density 0 0 readings for CSF IgM 00 0~~~~ ganglioside antibodies in 2.0 00 0 0 2.0 F patients with (left) 0 0 0 0 amyotrophic lateral sclerosis and (right) 1.5 00~~~ 1.5 k .

1.0 1 .0

0.5 0.5 0 0.0 i 0.0 GM1 GM2 GM3 AGM1 GD1la GD1lb GT1lb GMl GM2 GM3 AGMl GD1a GDlb GTlb A characteristic ganglioside antibody pattern in the CSF ofpatients with amyotrophic lateral sclerosis 363

Figure 2 Individual 2.5 2.5 normalised optical density J Neurol Neurosurg Psychiatry: first published as 10.1136/jnnp.56.4.361 on 1 April 1993. Downloaded from readings for CSF IgG ganglioside antibodies in 2.0 k 2.0 F patients with (left) amyotrophic lateral sclerosis and (right) 1.5 multiple sclerosis. 1.5 K

1 .0 - 1.0

0.5 F Ilhi"il 0.5 1a11,nI 0.0 0.0 GM1 GM2 GM3 AGM1 GD1a GD1b GT1b GM1 GM2 GM3 AGM1 GD1a GD1b GT1b

antibody pattem from the other patients with well as IgM antibodies to AGMI were the illness. entered into the analysis, patients with the A comparison of serum ganglioside anti- disease were still correctly identified in 62%, bodies between patients with the disease and but 14 controls were now misclassified as controls or those with other neurological dis- having the illness. Analysis of variance detect- ease showed elevated IgG to GM2, GM3, ed no effect of age, gender or distribution of AGM1 and GTlb in the disease but no dif- clinical signs on CSF or serum antibodies in ference in IgM. Serum reactivity in amyo- the disease. trophic lateral sclerosis differed from either type of multiple sclerosis only in their higher IgG anti-AGM1 optical density readings. Discussion CSF IgM to all gangliosides except AGMI The growing number of experimental and was elevated in the patients compared with clinical studies on altered immune function in those with other neurological diseases. CSF amyotrophic lateral sclerosis has not lead to IgM to GM2, GM3, GD1a, GD1b, and the identification of a consistent abnormality,7 GTIb was also higher than in chronic pro- except for the few cases associated with gressive multiple sclerosis. On the other plasma cell dyscrasia. '4 The detection of hand, patients with chronic progressive multi- immune complexes in serum3- and nervous ple sclerosis had higher CSF IgG optical den- tissue'5 as well as of IgG deposits in motor sity readings to GM2, GM3, GD 1 a, and neurons'6 and lymphocytic infiltrates'7 and GDlb than patients with amyotrophic lateral major histocompatibility complex class II sclerosis. CSF IgG antibodies to GM2, GM3, antigen expression in nervous tissue'8 may GDla, and GDlb, as well as IgM antibodies well be epiphenomenal to neuronal degenera- to GDla, GDlb, and GTlb were lower than tion and does not prove an autoimmune in relapsing remitting multiple sclerosis. aetiology. There was no correlation between CSF or We could not confirm elevated serum IgG http://jnnp.bmj.com/ serum ganglioside antibodies in amyotrophic or IgM levels or quantitative intrathecal IgG lateral or multiple sclerosis. or IgM synthesis or blood-brain barrier Discriminant analysis performed on nor- breakdown as characteristic features of the malised CSF optical density readings to all disease in our patients.3 Previous efforts to seven gangliosides allowed correct reclassifi- identify anti-neuronal protein antibodies in cation of 75% of the patients with the disease. such patients were not successful.'9 Poly- clonal or monoclonal serum antibodies to When variables were selected by stepwise on September 29, 2021 by guest. Protected copyright. entry depending on Wilk's ). reduction, the GM1, which were once thought to be associ- same classification results were obtained ated with the disease,89 have now been recog- using IgG antibodies to GM2 and GM3, nised to be probably more tightly linked to a which are low in amyotrophic lateral sclerosis, lower motor neuron syndrome distinct from in association with IgM antibodies to GDla amyotrophic lateral sclerosis.'22024 Moreover, and GDlb, which are high. None of the con- the spectrum of neurological disease with trols and 3% of the patients with multiple similar levels of serum anti-GM1 antibodies sclerosis were falsely classified as having is broad.'223 amyotrophic lateral sclerosis. To our knowledge, the present work is the The same procedure was then applied to first comprehensive study of ganglioside anti- serum ganglioside antibodies: this analysis of bodies in the CSF of patients with amyo- 35 serum samples from patients with amyo- trophic lateral sclerosis. We have identified trophic lateral sclerosis, according to their different patterns of ganglioside antibody pattern of ganglioside reactivity, allowed reactivity in serum and CSF, both of which correct classification in 83%. Six per cent of are nevertheless characteristic of the disease. the controls with other neurological diseases This pattern may be the only currently avail- were mistakenly diagnosed as having the dis- able biochemical test for the diagnosis of ease (false positives). No single patient with amyotrophic lateral sclerosis. The normalised multiple sclerosis received a diagnosis of serum optical density readings obtained amyotrophic lateral sclerosis. When only IgG in this study are lower than those found antibodies to GM1, GM2, and AGMI, as in patients with multifocal motor neur- 364 Stevens, Weller, Wietholter

The lack of correlation between 10( Weller M, Stevens A, Sommer N, Wietholter H. opathy.'223 Cerebrospinal fluid interleukins, immunoglobulins, and J Neurol Neurosurg Psychiatry: first published as 10.1136/jnnp.56.4.361 on 1 April 1993. Downloaded from our serum and CSF data suggests that fibronectin in . Arch Neurol 1991 ;48: a intrathecal 837-41. there is specific compartmental 11 Marcus DM, Latov N, Hsi BP, Gillard BK, and partici- immune reaction of IgM type in the disease. pating laboratories. Measurement and significance of for CSF antibodies against GM1 ganglioside. J Neuroimmunol As conventional isoelectric focusing 1989;25:255-9. oligoclonal bands is useful for IgG analysis 12 Weller M, Stevens A, Sommer N, Dichgans J, Kappler B, does not readily detect IgM,25 it is not Wietholter H. Ganglioside antibodies. Lack of diagnos- but tic specificity and clinical utility? _7 Neurol 1992 (in surprising that oligoclonal bands were seen in press). 13 Stevens A, Weller M, Wietholter H. Differing CSF and only two of our patients. Even sophisticated serum ganglioside antibody patterns in chronic progres- techniques of IgM oligoclonal band analysis sive and relapsing remitting multiple sclerosis. Neurology in the disease.2627 1992;42 (Suppl. 3):247. fail to yield positive results 14 Shy ME, Rowland LP, Smith T et al. Motor neuron dis- The lack of elevation of total CSF IgM or the ease and plasma cell dyscrasia. Neurology 1986;36: appears to prove that the immune 1429-36. IgM index 15 Donnenfeld H, Kascsak RJ, Bartfeld H. Deposits of IgG activation detected in this study cannot be and C3 in the spinal cord and motor cortex of ALS J we have identified patients. Neuroimmunol 1984;6:51-7. very strong. Nevertheless, 16 Engelhardt JI, Appel SH. IgG reactivity in the spinal cord a phenomenon which distinguishes amy- and motor cortex in amyotrophic lateral sclerosis. Arch from other neurolog- Neurol 1990;47:1210-6. otrophic lateral sclerosis 17 Troost D, van den Oord JJ, de Jong JMBV, Swaab DF. ical diseases and normal controls which Lymphocytic infiltration in the spinal cord of patients than with amyotrophic lateral sclerosis. Clin Neuropathol warrants further investigation. Rather 1989;8:289-94. claiming a primary pathogenetic role of gan- 18 Lampson LA, Kushner PD, Sobel RA. Strong expression we suggest that our find- of class II major histocompatibility (MHC) antigens in glioside antibodies, the absence of detectable T cell infiltration in amyo- ings may be related to an alteration of the trophic lateral sclerosis (ALS) spinal cord. J Neuropathol f in the brain28 and Exp Neurol 1988;47:353. ganglioside composition 19 Brown RH, Johnson D, Ogonowski M, Weiner HL. spinal cord29 of patients with the disease. Antineural antibodies in the serum of patients with amyotrophic lateral sclerosis. Neurology 1987;37:152-5. 20 Shy ME, Evans VA, Lublin FD et al. Antibodies to GM1 and GDlb in patients with with- out plasma cell dyscrasia. Ann Neurol 1989;25:511-3. 21 Salazar-Grueso EF, Routbort MJ, Martin J, Dawson G, 1 Matsumoto H, Hanson MR, Chad DA. Amyotrophic Roos RP. Polyclonal IgM anti-GMI ganglioside anti- lateral sclerosis. Recent advances in pathogenesis and body in patients with motor neuron disease and vari- therapeutic trials. Arch Neurol 1988;45:189-202. ants. Ann Neurol 1990;27:558-63. 2 Younger DS, Rowland LP, Latov N et al. Motor neuron 22 Pestronk A, Chaudhry V, Feldman EL et al. Lower motor disease and amyotrophic lateral sclerosis: relation of neuron syndromes defined by patterns of weakness, high CSF protein content to paraproteinemia and clini- nerve conduction abnormalities, and high titres of cal syndromes. Neurology 1990;40:595-9. antiglycolipid antibodies. Ann Neurol 1990;27:316-26. 3 Apostolski S, Nikolic J, Bugarski-Prokoplievic C, Miletic 23 Sadiq SA, Thomas FP, Kilidereas K et al. The spectrum V, Pavlovic S, Filipovic S. Serum and CSF immunolog- of neurologic disease associated with anti-GM1 anti- ical findings in ALS. Acta Neurol Scand 1991;83:96-8. bodies. Neurology 1990;40: 1067-72. 4 Oldstone MBA, Wilson CB, Perrin LH, Norris Jr FH. 24 Lange DJ, Trojaborg W, Latov N et al. Multifocal motor Evidence for immune complex formation in patients neuropathy with conduction block: is it a clinical entity? with amyotrophic lateral sclerosis. Lancet 1976;2: Neurology 1992;42:497-505. 169-72. 25 Sharief MK, Keir G, Thompson EJ. Glutaraldehyde- 5 Bartfeld H, Dham C, Donnenfeld H et al. Immunological enhanced immunofixation: a sensitive new method for profile of amyotrophic lateral sclerosis patients and their detecting oligoclonal IgM. Jf Neuroimmunol 1989;23: cell-mediated immune response to viral and CNS anti- 149-56. gens. Clin Exp Immunol 1982;48:137-47. 26 Sharief MK, Thompson EJ. Immunoglobulin M in the 6 Appel SH, Stockton-Appel V, Stewart SS, Kerman RH. cerebrospinal fluid: an indicator of recent immun- Amyotrophic lateral sclerosis. Associated clinical disor- ological stimulation. _7 Neurol Neurosurg Psychiaty 1989; ders and immunological evaluations. Arch Neurol 1986; 52:949-53. 43:234-8. 27 Sharief MK, Keir G, Thompson EJ. Intrathecal synthesis http://jnnp.bmj.com/ 7 Drachman DB, Kuncl RW. Amyotrophic lateral sclerosis: of IgM in neurological diseases: a comparison between an unconventional autoimmune disease? Ann Neurol detection of oligoclonal bands and quantitative estima- 1989;26:269-74. tion. J Neurol Sci 1990;96:131-42. 8 Pestronk A, Adams RN, Clawson L et al. Serum antibod- 28 Rapport MM, Donnenfeld H, Brunner W, Hungund B, ies to GM1 ganglioside in amyotrophic lateral sclerosis. Bartfeld H. Ganglioside patterns in amyotrophic lateral Neurology 1988;38: 1457-61. sclerosis brain regions. Ann Neurol 1985;18:60-7. 9 Li F, Pestronk A. Autoantibodies to GM1 ganglioside: 29 Dawson G, Stefansson K. Gangliosides of human spinal different reactivity to GMl-liposomes in amyotrophic cord: aberrant composition of cords from patients with lateral sclerosis and lower motor neuron disorders. amyotrophic lateral sclerosis. _7 Neurosci Res 1984;12: Neurol Sci 199 l;104:209-14. 213-20. on September 29, 2021 by guest. Protected copyright.