CSF herpes virus and autoantibody profiles in the evaluation of encephalitis.

OBJECTIVE: To report the frequency of coexisting herpes viruses (herpes simplex virus 1 [HSV-1] or HSV-2, varicella zoster virus, Epstein-Barr virus [EBV], cytomegalovirus, or human herpes virus 6 [HHV-6]) and autoantibodies in patients with encephalitis (herpes or autoimmune) in clinical laboratory service.
METHODS: Three groups were evaluated for herpes viruses and antibodies: group 1-patients whose CSF was positive for a herpes virus by real-time PCR over a period of 6 months; group 2-patients whose CSF was positive for an autoimmune encephalitis-associated antibody over 5 years (e.g., NMDA receptor [NMDA-R] antibody), and the same number of controls without autoimmune/infectious disease; and group 3-incidental autoimmune parainfectious encephalitis cases encountered over 1 year.
RESULTS: In group 1, antibodies were detected in 27 of 100 herpes PCR-positive CSF specimens (CSFs), either unclassified neural or nonneural in all but one patient with NMDA-R antibody detected after EBV infection. Antibodies were also detected in 3 of 7 CSFs submitted for repeat PCR testing (unclassified, 2; AMPA receptor, 1). In group 2, herpes viruses were detected in 1 of 77 controls (HHV-6) and 4 of 77 patients with autoimmune encephalitis (EBV, 2; HHV-6, 2); autoantibodies targeted NMDA-R in 3/4 and GABAB-R in 1/4. In group 3, NMDA-R antibody was detected in 7 patients post-HSV-1 encephalitis. Of the remaining 3 patients, 2 had unclassified neural antibodies detected, and one had GABAB-R autoimmunity. Concomitant neoplasms were discovered in 2 patients each from groups 2 and 3.
CONCLUSIONS: Autoantibodies and herpes virus DNA frequently coexist in encephalitic CSF. Some patients develop parainfectious autoimmunity following viral CNS infection (usually HSV-1 encephalitis). The significance of detecting herpes nucleic acids in others remains unclear.

Neurologic autoimmunity may be triggered by cancer (paraneoplastic), infection (parainfectious), or it may be cryptogenic.[1] Clinical disorders for which experimental models have been developed include anti-Yo autoimmunity, in women with cerebellar degeneration and breast or gynecologic adenocarcinoma, and ganglioside autoimmunity, in patients with Guillain-Barré syndrome arising after Campylobacter infection.[2,3]

Ovarian teratoma is a frequent accompaniment of paraneoplastic autoimmune encephalitis targeting the GluN1 subunit of the NMDA receptor (NMDA-R).[4,5] Recent studies revealed that NMDA-R antibodies frequently coexist in patients with herpes simplex virus 1 (HSV-1) encephalitis, and that some anti–NMDA-R encephalitis cases develop as a parainfectious autoimmune phenomenon, within weeks of HSV-1 encephalitis.[6-11] Those studies confirmed suspicions that early relapse in HSV encephalitis, most notably in children, could be autoimmune.[12] Typically, a striking biphasic illness occurs, whereby initial remission from viral encephalitis is followed by relapse with autoimmune encephalitis within 1 to 7 weeks of the initial presentation.[7-11]

Other viruses from the Herpesviridae family may trigger neurologic autoimmunity, and the spectrum of antibodies encountered may extend beyond the NMDA-R.[7,13] In this study, we undertook a broad evaluation of CSF specimens (CSFs) clinically referred for either herpes or autoimmune testing in order to assess the following: (1) the frequency of neural and nonneural antibodies in CSFs from patients with herpes virus PCR positivity (identified over 6 months), (2) the frequency of herpes viruses in stored CSFs from patients with autoimmune encephalitis (identified over 5 years) and controls without inflammatory or infectious neurologic disorders, and (3) incidental cases of autoimmune encephalitis in which a parainfectious cause was considered (encountered in the Mayo Clinic Neuroimmunology Laboratory over 1 year).

METHODS

Standard protocol approvals, registrations, and patient consents.

The Mayo Clinic institutional review board (14-008716) approved this study. The figure demonstrates an outline of the patients.

Figure.

Groups 1–3, patients tested, and test results (viral PCR and antibodies)

Ab = antibody; AMPA-R = amino-3-hydroxy-5-methyl-4-isoxazole propionic acid receptor; ANA = antinuclear antibody; CBA = cell-based assay; CSFs = CSF specimens; EBV = Epstein-Barr virus; GABAB-R = γ-aminobutyric acid type B receptor; GPC = gastric parietal cell; HHV-6 = human herpes virus 6; HSV = herpes simplex virus; IFA = immunofluorescence assay; IgG = immunoglobulin G; NMDA-R = NMDA receptor; PCR = polymerase chain reaction; SMA = smooth muscle antibody; UNA = uncharacterized neural antibody.

Group 1: Autoantibody testing in CSFs with herpes PCR positivity.

Residual CSFs from 100 patients evaluated in the Mayo Clinic Virology Laboratory between January 1, 2015, and June 30, 2015, that tested positive by real-time PCR for a herpes virus (HSV-1 and -2, varicella zoster virus [VZV], Epstein-Barr virus [EBV], cytomegalovirus [CMV], and human herpes virus 6 [HHV-6]), were prospectively collected and tested for immunoglobulin G (IgG) antibodies by indirect immunofluorescence assays (IFAs).

Patients in whom at least 500 μL of residual CSF was available on completion of virology testing were included. CSFs were refrigerated at 4°C and tested in the Neuroimmunology Laboratory within 5 days. Ninety-two of the 100 CSFs were referred from practitioners outside of Mayo Clinic; 8 were internally referred. Serum was not available for testing. Of 10 CSFs subsequently submitted from the same patients for repeat virology testing within the 6-month timeframe, 7 had sufficient residual volume for neural antibody testing. Clinical histories were obtained for patients in whom well-characterized synaptic plasma membrane protein-directed antibodies (such as NMDA-R) were detected and for those with antibody detected upon repeat testing.

The IgG-class antibody assays were all interpreted by experienced observers. The assays were indirect IFAs, which consisted of the following: (1) a tissue-based assay utilizing a composite substrate of mouse brain, kidney, and intestinal tissues to identify neural antibodies (either well-characterized, such as NMDA-R, or unclassified) and nonneural antibodies (such as antinuclear antibody [ANA], smooth muscle antibody [SMA] and gastric parietal cell [GPC] antibody) by established diagnostic criteria[14,15]; and (2) cell-based assays (CBAs) for encephalitis-associated neural antibodies targeting cell surface synaptic antigens (α-amino-3-hydroxy-5-methyl-4-isoxazole propionic acid receptor [AMPA-R], aquaporin-4 [AQP4], contactin-associated protein-like 2 [CASPR2], dipeptidyl-peptidase-like protein-6 [DPPX], γ-aminobutyric acid type B receptor [GABAB-R], metabotropic glutamate receptor 1 or 5 [mGluR1, mGluR5], leucine-rich glioma-inactivated 1 protein [LGI-1], or GluN1 subunit of the NMDA-R). For each CBA, HEK293 cells were transfected with the relevant complementary DNA and control cells were transfected with empty vector (EUROIMMUN, Luebeck, Germany).[16]

Five tests were performed in total on CSF at a dilution of 1:2 –1 tissue IFA, 3 composite CBAs, and 1 CBA to test for AQP4 antibody only. The composite CBAs contained chips expressing the following antigens: (1) NMDA-R, AMPA-R, and GABAB-R; (2) LGI-1 and CASPR2; (3) mGluR1, mGluR5, and DPPX. Insufficient quantity existed to determine titers. Twenty of 100 CSFs had insufficient quantity available for complete antibody testing. One test was lacking in 13 specimens, 2 tests in 6, and 3 tests in 1; 5 total CSFs were lacking tissue IFA.

Group 2: Herpes virus testing in CSFs with synaptic antibody positivity and controls.

Archived and frozen CSFs (2011–2015) from 77 patients with autoimmune encephalitis and known positivity for a neural antibody in one or more of the CBAs described above and 77 controls (antibody-negative CSFs from Mayo patients [2015] without known inflammatory or infectious CNS diseases) were tested retrospectively for herpes viruses (HSV-1/2, VZV, EBV, CMV, and HHV-6) by real-time PCR. The median patient age was 25 years (range, 2–85); 55 were female. Median control age was 49 years (range, 16–86); 49 were female. Control diagnoses were cognitive disorders (n = 33), epilepsy (n = 15), movement disorders (n = 11), no neurologic diagnosis (n = 17), or motor neuronopathy (n = 1).

For herpes virus real-time PCR testing, DNA from 200 μL of raw CSF was extracted on the MagNA Pure LC 2.0 (Roche Diagnostics, Indianapolis, IN) using the Total Nucleic Acid extraction protocol. Subsequently, 5 μL of extracted nucleic acid was added to a LightCycler capillary (Roche) containing 15 μL of master mix for one of the following: CMV (laboratory developed test [LDT]), EBV (Roche analyte specific reagents), HHV-6 (LDT), HSV-1 and -2 (Roche analyte specific reagents), and VZV (LDT). Testing was then performed on a LightCycler 2.0 (Roche) as previously described.[17-20]

Group 3: Incidental cases.

Case histories of patients with findings suspicious for parainfectious autoimmune encephalitis incidentally encountered by J.L. in the course of clinical laboratory service between July 2014 and June 2015 were collated. Virology findings (from CSF only) and antibodies detected (from serum and CSF) are reported.

RESULTS

Antibody testing results, virus PCR results, other CSF data, and clinical information are outlined in the figure and in tables 1–4.

Group 1: Autoantibody testing of CSFs with herpes PCR positivity.

One hundred CSFs that were positive by real-time PCR for a herpes virus (one virus per specimen) were encountered over the 6-month study period. Viruses detected included HSV-2 (n = 38), VZV (n = 30), HSV-1 (n = 18), EBV (n = 6), HHV-6 (n = 6), and CMV (n = 2).

Antibodies detected in 27 of 100 CSFs by tissue-based IFA included uncharacterized neural antibodies (n = 15), ANA (n = 12), SMA (n = 10), and GPC antibody (n = 1) (table 1). Only one CSF, which was EBV-positive and drawn 1 month into a meningoencephalitic illness, was antibody positive by CBA (for NMDA-R antibody). This patient was also positive for ANA and SMA by tissue-based IFA (tables 1 and 2) and had signs of systemic EBV infection as well as meningoencephalitis. CSF drawn 3 months after illness onset had neither antibodies nor EBV detected. The authors subsequently obtained an aliquot of CSF drawn 2 weeks into the illness from the referring provider. Viral testing was negative for EBV. Antibody testing revealed ANA alone.

Table 1.

Group 1 patient demographics, viral species detected, and tissue IFA antibody findings in CSF (n = 100)

Autoantibody testing of follow-up CSFs.

CSFs from 10 of the 100 patients with prior herpes virus PCR positivity (including the patient with EBV infection described above, Table 2, row 1) were subsequently tested by real-time PCR within a median of 7 weeks after the initial evaluation (range, 1.3–21). Two were still positive for herpes viruses, one for CMV (1.3 weeks after the first CSF test), and one for HHV-6 (3.5 weeks later). Sufficient specimen volumes were available for antibody testing in 7 of the 10 CSFs (including the 2 with repeat herpes positivity). Antibodies were detected in 3 of 7 repeat CSFs (43%, table 2, rows 2–4). Of these patients, the first 2 described were considered to have a parainfectious autoimmune disorder by the treating physicians. The first patient with a biphasic encephalitic course had GPC antibody detected in the initial HSV-1 PCR-positive CSF, and AMPA-R antibody (but not GPC antibody) detected in CSF submitted 9 weeks later. The second patient with zoster myelitis who developed encephalopathy during recovery from myelopathy had SMA detected in the initial VZV PCR-positive CSF, and SMA and an unclassified neural antibody (but not VZV) detected in CSF submitted 3 weeks later. A third patient, with encephalopathy, recurrent lymphoma with spread to the CNS, and HHV-6 viremia, tested positive for both HHV-6 and an uncharacterized neural antibody in both CSFs submitted. The significance of the viral and antibody findings was uncertain in that patient.

Table 2.

Group 1 patients with serial CSF specimens in whom neural autoantibodies were detected (n = 4)

Group 2: Herpes virus testing of CSFs with synaptic antibody positivity and controls.

Among stored CSFs from 77 patients diagnosed with autoimmune encephalitis, the following antibodies were detected: NMDA-R (n = 56), GABAB-R (n = 13), LGI-1 (n = 2), DPPX (n = 2), CASPR2 (n = 1), AMPA-R (n = 1), NMDA-R and AMPA-R (n = 1), and NMDA-R and GABAB-R (n = 1). Four of 77 CSFs were positive by herpes virus real-time PCR: 2 for EBV (both patients had anti–NMDA-R encephalitis) and 2 for HHV-6 (one patient had anti–NMDA-R encephalitis, the other had GABAB-R encephalitis [table 3]). None had a history of infectious encephalitis recorded. Two of the 4 patients also had neoplasia (bilateral ovarian teratomas and small cell lung carcinoma) detected and were diagnosed with a paraneoplastic (rather than parainfectious) encephalitis. Among 77 control CSFs (antibody negative, from patients without inflammatory or infectious CNS disease), one tested positive for HHV-6. The patient was a 21-year-old woman with chronic intractable cryptogenic focal epilepsy, without clinical, radiologic, or CSF evidence of CNS infection or autoimmunity.

Table 3.

Group 2 patients with diagnoses of autoimmune encephalitis in whom herpes viruses were detected (n = 4)

In the course of clinical consultative correspondence with outside physicians over 12 months, 10 patients were encountered with CSF positivity for both herpes PCR and neural antibodies concurrently or sequentially (table 4). Eight patients had HSV-1 alone detected (3 were children), one had both EBV and HHV-6 detected, and one had EBV alone.

Table 4.

Group 3 (incidental cases) in whom parainfectious autoimmunity was considered (n = 10)

Five patients had biphasic clinical courses and were recognized to have a postinfectious autoimmune disorder. All had HSV-1 detected during the first phase and NMDA-R antibody detected during the second phase. Three were older than 60 years, and 2 were younger than 2 years; all were female. Initial encephalitic illnesses were followed by improvement attributed to antiviral therapy. Relapse of encephalitic symptoms occurred a median of 4 weeks later (range, 1–6). One pediatric patient developed a hyperkinetic movement disorder during the second phase. Immunotherapy data were available for 3 patients, 2 of whom improved.

Five patients had monophasic encephalitic courses. In one child, HSV-1 infection was initially suspected and was confirmed by PCR. Further evaluation was prompted by lack of improvement and development of orofacial dyskinesias despite prompt treatment with antiviral therapy. Anti–NMDA-R encephalitis was suspected and confirmed by CSF testing. A second patient had a clinical course typical for severe anti–NMDA-R encephalitis from the outset, and had both NMDA-R antibody and EBV detected in the same CSF specimen. The significance of the virology finding was unknown. Two further patients had both HSV-1 and neoplasia detected during evaluations for encephalitis. One of those 2 patients had GABAB-R encephalitis and small cell lung carcinoma, and was concluded to have paraneoplastic encephalitis. The other had an unclassified neural antibody and renal cell carcinoma detected, but the cause of encephalitis (infectious, parainfectious, or paraneoplastic) was uncertain. The final patient had encephalopathy, which progressed to a minimally responsive state over the course of 1 year. Despite extensive evaluations, a cause was not determined. Both EBV and HHV-6 were detected in the initial CSF. Repeat CSF testing over the year persistently demonstrated an elevated white cell count, HHV-6 PCR positivity, and an unclassified neural antibody. There were no neurologic improvements after multiple antiviral treatments and immunotherapies.

DISCUSSION

In this study based in the Mayo Clinic Virology and Neuroimmunology clinical laboratories, autoantibodies and herpes nucleic acids were commonly codetected in CSFs submitted for herpes and autoimmune encephalitis evaluation. The species of herpes viruses (HSV-1/2, VZV, EBV, HHV-6, and CMV) and antibodies detected (most commonly unclassified neural or nonneural IgGs) were diverse. For some patients in whom longitudinal clinical data and specimens were available (mostly anecdotal), clinical and CSF testing results indicated sequential clinical events (remitting viral encephalitis, followed by autoimmune encephalitis). Some patients had the now well-described parainfectious phenomenon of anti–NMDA-R encephalitis after HSV-1 encephalitis. Consistent with previous reports, many (half) of those patients were children and neoplasms were not found.[7-11] For others, testing results were also suggestive of sequential viral and parainfectious autoimmune disorders, although distinct clinical phases could not be discerned. In other patients, the clinical significance of the detected herpes viruses and antibodies was uncertain. Latent neurotropic virus may be reactivated or shed as a consequence of CNS inflammation, and autoimmune responses to disparate neural antigens during viral infection may occur.

The herpes virus PCR-positive cohort (group 1), which consisted of patients with CSFs submitted on a clinical basis for herpes virus testing, and evaluated on a research basis for autoantibodies, provided a snapshot of other parainfectious autoimmune phenomena arising during CNS herpes infection. Consistent with a previous report, uncharacterized neural antibodies were commonly detected in CSF.[7] In addition, we also detected a high frequency of nonneural autoantibodies, either organ-specific (such as SMA and GPC antibody) or non–organ-specific (ANA). A similar profile of nonneural antibodies is frequently detected in patients with paraneoplastic autoimmunity.[21] We did not have clinical data or sera available to determine whether the detected non–organ-specific autoimmunity was attributable to coexisting systemic autoimmune disease or advanced age, rather than occurring in the context of CNS infection.[22] Nonneural autoantibodies detected in CSF may also indicate nonspecific immune activation during infection, rather than heralding impending neural-specific autoimmunity. Repeat CSF data were limited to those patients who had CSF resubmitted on a clinical basis.

Our findings confirm that parainfectious autoimmunity may occur in the context of EBV and VZV, as reported previously.[10,13,23] In 4 of 5 cases (from groups 1–3) where the clinical histories were known, patients with EBV-positive CSF also had NMDA-R antibodies detected. With this study, we additionally demonstrated AMPA-R autoimmunity post–HSV-1 encephalitis and neurally directed antibodies in the setting of HHV-6 infection. Together, these results demonstrate a spectrum of coexisting herpes viruses and antibodies extending beyond HSV-1 and NMDA-R antibodies.

The rate of resubmission of CSF (10%) from group 1 is consistent with a sizable number of patients with herpes encephalitis having prolonged or complicated courses, prompting clinicians to re-evaluate. A recent prospective study of herpes simplex encephalitis cases found that 25% of patients had immune-mediated relapsing neurologic sequelae.[12] It was also found in the same study that adults who did not have hyperkinetic movements and altered consciousness often had delayed diagnoses when compared with children who had these symptoms. Those findings and our data suggest that postviral autoimmune encephalitis remains underrecognized in clinical practice, and that opportunities for neurologic improvement with immunotherapy may be missed.[12]

Our autoimmune encephalitis cohort (group 2) demonstrated herpes virus PCR positivity (HHV-6 and EBV) in 4 of 77 patients (5%), although the timing of CSFs submission may have affected sensitivity. CSFs may have been submitted for antibody testing weeks to months after symptom onset, by which time herpes PCR testing might have proven negative.

The significance of herpes virus PCR positivity was uncertain in 3 of 4 patients with autoimmune encephalitis (group 2) and 4 of 10 incidental cases (group 3). Of those 7 patients, a paraneoplastic cause (rather than parainfectious) seemed most likely in 3. This uncertainty seemed particularly true for those patients in whom HHV-6 was detected, and was exemplified further by one control patient with cryptogenic intractable chronic focal epilepsy who was HHV-6–positive (and neural autoantibody–negative). Of 4 patients in groups 1, 2, or 3 who had HHV-6–positive CSF and histories known, one had HHV-6 viremia and likely CNS infection, one had encephalopathy of unknown cause, and 2 had malignancies (small cell carcinoma and T cell lymphoma) prompting consideration of a paraneoplastic rather than parainfectious encephalitis. Post hoc, we reviewed the available Mayo Clinic histories of 2 additional antibody-negative patients from group 1 with HHV-6 CSF positivity. In contrast to our antibody-positive patients, both were immunocompromised and had unambiguous clinical and radiologic features supportive of HHV-6 encephalitis. One had myelodysplastic syndrome and the other had diffuse systemic (excluding CNS) large B cell lymphoma.

It is possible that CSF herpes PCR positivity may occur in some patients because of shedding of latent neurotropic viral DNA in the course of paraneoplastic encephalitis. Reactivation of infection in that context cannot be excluded either (and we did not undertake confirmatory viral cultures from archived CSFs).[24] PCR positivity for HSV-1, HHV-6, and EBV in CSF, occurring without additional evidence supporting CNS infection, has also been reported among immunosuppressed patients and those with neurodegenerative disorders or stroke.[25,26]

Although serum was not available for groups 1 and 2, antibody data were available for both serum and CSF among our incidental cases. In 6 of those 10 cases, antibodies were detected in CSF alone, supporting the concept of a CNS-predominant immune response (antiviral and autoimmune). An animal model study of herpes simplex encephalitis, which did not include testing for neural autoantibodies, demonstrated a sustained inflammatory response after recovery from herpes infection.[27] Ours and previous data support the possibility of autoimmunity contributing to this persistent CNS inflammation.[7] Before the availability of antiviral therapy, corticosteroids (anti-inflammatory) had been reported to be beneficial for reducing edema in HSV encephalitis.[28,29] Use of corticosteroids combined with antiviral therapy was associated with improved outcome in HSV encephalitis in a retrospective study, but controlled clinical trials are lacking in this area.[30,31]

Detection of a herpes virus (usually HSV-1) and diverse autoantibodies in the CSF may signify a parainfectious autoimmune response in many patients. That response may initially be nonneural but, in some cases, matures into an attack directed at specific neural receptors (such as NMDA-R), which may manifest as a relapsing biphasic illness. The frequent coexistence of autoimmune markers in patients with CNS herpes infection raises several mechanistic and therapeutic questions, which may be answered by animal model studies and clinical trials.[32] Caution regarding the interpretation of herpes virus PCR, unclassified, and nonneural antibody results in CSF is advised where other clinical or supportive features of CNS infection or autoimmunity are absent.