CSF complements serum for evaluating paraneoplastic antibodies and NMO-IgG

The detection of neural-reactive immunoglobulin G (IgG) autoantibodies aids the diagnosis of organ-specific autoimmune neurologic disorders. Many paraneoplastic autoantibodies reliably predict a particular cancer type and are accompanied by varied neurologic presentations of subacute onset.¹ The detection of neuromyelitis optica (NMO)–IgG predicts a relapsing inflammatory demyelinating disorder predominated by optic neuritis and transverse myelitis.² When an autoimmune neurologic disorder is suspected, serologic testing of serum is frequently undertaken before more invasive CSF evaluation. However, CSF evaluation can complement testing of serum when suspicion for an autoimmune etiology persists despite a negative serum result. Here we report, for a 25-year period of testing by standardized indirect immunofluorescence protocols, the frequency of neural autoantibody detection in serum and CSF.

Methods.

The immunofluorescence protocols we used were validated in this laboratory for detection of paraneoplastic antibodies (anti-neuronal nuclear antibody [ANNA]-1; ANNA-2; ANNA-3; Purkinje cell cytoplasmic antibody [PCA]-1; PCA-2; PCA-Tr; collapsin response-mediator protein [CRMP]-5-IgG; amphiphysin antibody; antiglial/neuronal nuclear antibody [AGNA]-1; NMDA receptor antibody) and NMO-IgG. We searched the Mayo Clinic Neuroimmunology Laboratory database (January 1986 to March 2010) for all patient samples submitted for service evaluation of paraneoplastic or NMO-IgG. We included both Mayo Clinic and non-Mayo patients for whom both serum and CSF were submitted, and reviewed available oncologic data for patients with antibodies identified by CSF testing.

Results.

Testing was performed on a clinical service basis for a median of 12 years (range 2–25 years). The antibody detection rate in all specimens ranged from 0.01% for PCA-Tr to 7% for NMO-IgG (table).

Table.

Results of immunofluorescence testing for paraneoplastic autoantibodies and NMO-IgG, 1986–2010

Abbreviations: IgG = immunoglobulin G; NMO = neuromyelitis optica.

Coexisting antibodies, n of patients for each:

^aCRMP-5 IgG, 16; PCA-2, 1; amphiphysin antibody, 1.

^bAmphiphysin antibody, 1; CRMP-5 IgG, 1.

^cCRMP-5 IgG, 8; ANNA-1, 1.

^dCRMP-5 IgG, 1.

^eAGNA-1.

^fANNA-1, 16; PCA-2, 8; amphiphysin antibody, 3; ANNA-2, 1.

^gPCA-2, 1.

^hCRMP-5 IgG, 3; ANNA-1, 1; ANNA-2, 1.

In patients for whom paired serum and CSF samples were tested, the antibody detection rate ranged from 0.08% (PCA-Tr) to 9% (NMDA receptor antibody). One or more neural autoantibodies were detected in 462 patients (497 antibodies detected). In 405 of those 462 patients, both serum and CSF yielded a positive result (88%). In 57 patients, serum or CSF alone was positive (12%). Among those patients, serum alone yielded a positive result in 31 (54%) and CSF alone in 26 (46%). For classic paraneoplastic antibodies, CSF alone yielded a positive result in 20 patients, twice as commonly as serum alone (10 patients). For NMO-IgG, serum alone yielded a positive result in 21 patients 3.5 times more commonly than CSF alone (6 patients).

Discussion.

From our review of a 25-year experience with immunofluorescence testing on a service basis in the Mayo Clinic Neuroimmunology Laboratory, we found that the rate of clinically pertinent autoantibody detection was highest when both serum and CSF were tested. It is plausible that this finding may reflect a greater likelihood of physicians deciding to test both serum and CSF in patients with the highest index of clinical suspicion.

When both serum and CSF were tested, CSF was more commonly informative than serum for paraneoplastic antibody detection. This raises concern that clinically important neural antibodies may be missed when only serum is tested. This finding was most prominent for NMDA receptor-specific IgG. Consistent with this finding, Kumar et al.³ recently reported 3 patients, each of whom had NMDA receptor IgG detected in CSF but not in serum. Where there is a high suspicion for cancer, additional testing for paraneoplastic antibodies on CSF may be helpful.

For detection of NMO-IgG, the positivity rate was greater for serum alone than for CSF alone. However, testing of CSF is helpful when the serum is negative. Klawiter et al.⁴ reported 3 patients who had an NMO spectrum disorder but were seronegative for NMO-IgG (those patients are included in this report also). Detection of NMO-IgG in CSF allowed unambiguous distinction of NMO from MS. CSF testing offers the additional advantage of generally lacking the non-organ-specific IgG autoantibodies (e.g., antinuclear, antimitochondrial, and smooth muscle) that are common in serum of patients with NMO and also with classic paraneoplastic autoimmune disorders.

Testing of serum or CSF for paraneoplastic autoantibodies or NMO-IgG should be reserved for clinically suspicious cases. Risk factors for a paraneoplastic disorder which may prompt testing include a subacute onset of neurologic symptoms and signs, a personal or family history of autoimmunity or cancer, and smoking history. Seronegativity for paraneoplastic antibodies does not exclude a diagnosis of cancer, and testing for occult malignancy should proceed in highly suspicious cases.⁵ The low positivity rates for most paraneoplastic antibodies in this study are consistent with their rarity, and reflects the ordering of this testing in patients with diverse neurologic presentations, where no cause has been established.¹ The more frequent detection of NMO-IgG and NMDA receptor antibody is consistent with the more narrow clinical contexts for ordering these 2 tests.^2,6