BRIEF REPORT
Dual Infections of the Central Nervous System with Epstein-Barr Virus Adriana Weinberg,1,2 Karen C. Bloch,7,8 Shaobing Li,1 Yi-Wei Tang,7,9 Megan Palmer,1 and Kenneth L. Tyler2,3,4,5,6 Departments of 1Pediatrics, 2Medicine, 3Neurology, 4Immunology, and 5Microbiology, University of Colorado Medical School, and 6Denver Veterans Affairs Medical Center, Denver; Departments of 7Medicine, 8Preventive Medicine, and 9Pathology, Vanderbilt University, Nashville, Tennessee
We describe clinical and laboratory characteristics of 16 patients with central nervous system (CNS) infection caused by Epstein-Barr virus (EBV) and another pathogen. Seven of 10 immunocompromised patients had coinfection with viruses (3 with cytomegalovirus, 2 with JC virus, and 2 with varicella zoster virus) and 3 with nonviral pathogens (2 with pneumococcus and 1 with Cryptococcus species). Three of 6 immunocompetent patients had coinfections with viruses (1 each with herpes simplex virus, varicella zoster virus, and West Nile virus), and 3 had coinfections with nonviral pathogens (2 with Ehrlichia chaffeensis and 1 with Mycoplasma pneumoniae). The EBV load was similar in immunocompromised and immunocompetent patients and in patients with viral and nonviral coinfections. EBV lytic-cycle mRNA was detected in the cerebrospinal fluid of 5 of 6 tested samples, indicating EBV replication in the CNS during coinfection. The use of polymerase chain reaction (PCR) on samples from cerebrospinal fluid (CSF) has greatly facilitated the diagnosis of central nervous system (CNS) infections. In a subset of PCRpositive CSF samples, there is evidence of simultaneous infection with 2 pathogens (dual infection). Herpesviruses, particularly Epstein-Barr virus (EBV), are the most frequent agents found in the CNS in association with other microorganisms [1–3]. The clinical significance of amplifying EBV DNA in the CSF from patients with suspected CNS infection remains uncertain. After primary infection, EBV becomes latent in B lymphocytes from which EBV DNA can be amplified, even in the absence of symptoms [4]. EBV DNA detected in the CSF by Received 19 May 2004; accepted 12 July 2004; electronically published 10 December 2004. Financial support: Emerging Infections Program (cooperative agreement U50/CCU41612304 with the Centers for Disease Control and Prevention). Reprints or correspondence: Dr. Adriana Weinberg, University of Colorado Health Sciences Center, Campus Box C227, Denver, CO 80262 (
[email protected]). The Journal of Infectious Diseases 2005; 191:234–7 2004 by the Infectious Diseases Society of America. All rights reserved. 0022-1899/2005/19102-0012$15.00
234 • JID 2005:191 (15 January) • BRIEF REPORT
PCR might represent latent DNA in the inflammatory infiltrate, as has been suggested by the failure in some subsets of patients to detect an EBV-specific intrathecal antibody response [3]. Conversely, our group has recently shown that the majority of patients with PCR-positive EBV in the CSF also have evidence of detectable lytic-cycle BZLF1 mRNA, which provides evidence of active EBV replication in the CNS [5]. To gain a better understanding of the clinical and pathological significance of EBV as a coinfecting agent in patients with CNS disorders, we identified cases of EBV-associated dual CNS infections and reviewed their clinical and laboratory characteristics and the nature of the associated EBV infection. Patients, materials, and methods. Patients with dual CNS infections were identified at 2 sites: the University of Colorado Health Sciences Center Virology Laboratory (UCHSC Virology) or through the Tennessee Unexplained Encephalitis Surveillance (TUES) study at Vanderbilt University. CSF specimens containing EBV DNA were identified between October 1995 and March 2001 at UCHSC Virology. Clinical samples were tested for EBV and for other pathogens as ordered by the referring physician. Clinical and laboratory information was obtained through a questionnaire submitted to the attending physician and/or neurology consultant and from a review of medical records. The study was approved by the institutional review boards of participating institutions. Immunocompetent patients who met criteria for encephalitis published elsewhere [6] were enrolled into the TUES study between January 2000 and October 2003. Clinical and laboratory information was collected by use of a standardized questionnaire at enrollment and at 3, 6, and 12 months. A series of tests, including PCR for EBV, was performed on CSF samples from all patients enrolled in the study. CSF samples with a positive qualitative PCR result for EBV at the TUES laboratory were sent to UCHSC Virology for quantitative PCR testing. CNS lymphoma (CNSL) was defined as typical histopathologic or neuroimaging findings without alternative diagnoses. Encephalitis was defined as fever, pleocytosis in the CSF, altered mental status, seizures, focal neurological deficits, and abnormal neuroimaging or electroencephalogram results. Meningitis was defined as headache, fever, nuchal rigidity, and pleocytosis in the CSF without the involvement of the brain parenchyma. Polyradiculomyelitis (PRM) was defined as fever, pleocytosis in the CSF, and signs and symptoms of combined involvement of the spinal cord and nerve roots. Qualitative EBV PCR was performed as described elsewhere [5, 7], by use of duplicate CSF aliquots. Each assay included 2
negative water controls and 2 positive controls of EBV-producing B95-8 cell-culture supernatant. The results were considered to be valid if the control samples yielded expected results and if the replicate results were concordant. Quantitative EBV PCR was performed by use of a competitive method described elsewhere [5]. Positive and negative controls were run with each clinical specimen. Transcription of the lytic-cycle gene BZLF1 was detected by reverse transcription (RT)–PCR, as described elsewhere [5]. In this target area, the RT-PCR generates 2 bands of 252 and 450 kb molecular weight, corresponding to cDNA and genomic DNA, respectively. The presence of both 450- and 252-kb bands was interpreted as evidence of transcription. The presence of the 450-kb band only was interpreted as a lack of transcription. The absence of both bands was interpreted as primer-site mismatch or an insufficient nucleic-acid template in the sample. Statistical analysis was performed by use of Instat3 software (version 3; GraphPad). Results. At UCHSC Virology, 528 CSF samples were analyzed for EBV DNA, with 39 (7%) positive results. Of these, 28 (72%) were cases in which EBV was the sole pathogen, and 11 (28%) contained EBV and another pathogen that was considered to have played a role in the patient’s illness (dual infection). Through the TUES study, 342 CSF samples were analyzed, and 21 (6%) tested positive for EBV DNA. Of these, EBV was the sole pathogen in 16 (76%) and was part of a dual infection in 5 (24%). The 16 coinfections were diagnosed in children and adults of both sexes and of diverse ethnicity (table 1). There were 10
Table 1.
Selected clinical and laboratory characteristics of patients with Epstein-Barr virus (EBV)–associated neurologic disorders.
Patient (sex) 1 (M) 2 (M) 3 (F) 4 (F) 5 (M) 6 (M) 7 (M) 8 (M) 9 (M) 10 (M) 11 (F) 12 (M) 13 (M) 14 (F) 15 (M) 16 (M)
immunocompromised hosts, including 9 patients with HIV infection and 1 patient who was receiving methotrexate for the treatment of rheumatoid arthritis. Three HIV-infected patients were diagnosed with CNSL and had cytomegalovirus as a copathogen. Ten patients presented with encephalitis or meningitis. Of these, 4 were immunocompromised; in addition to EBV, 1 had PCR-identified varicella zoster virus (VZV) CNS infection, 2 had culture-positive pneumococcal meningitis and sepsis, and 1 had culture-proven cryptococcal meningitis. The 6 immunocompetent patients with encephalitis had EBV associated with serologically diagnosed Ehrlichia chaffeensis (n p 2), PCR-diagnosed herpes simplex virus (HSV)–1 (n p 1), serologically diagnosed Mycoplasma pneumoniae (n p 1 ), PCRdiagnosed VZV (n p 1), and serologically diagnosed West Nile virus (n p 1). Two HIV-infected patients with progressive multifocal leukoencephalopathy (PML) had both JC virus and EBV DNA in the CNS. The last patient was HIV positive and presented with PRM. He had both VZV and EBV DNA in his CSF. The clinical presentations of the patients with meningoencephalitis were sometimes, but not always, indicative of the infectious agent. Of the 3 patients with VZV and EBV, patients 12 and 7 developed cutaneous lesions before or a few days after the onset of encephalitis. The patient with HSV-1 and EBV coinfection had temporal lobe involvement. Both patients with ehrlichia infection had leukopenia and thrombocytopenia consistent with a tickborne infection. The patients with JCV and EBV had radiographic studies consistent with PML. All patients survived, except for 2 with CNSL and AIDS. Six patients with encephalitis or PRM received acyclovir as part of
Age, years (race)
Underlying condition
27 (W) NA (W) 36 (H) 34 (B) 44 (W) 51 (W) 46 (W) 42 (B) 50 (B) 37 (W) 81 (W) 73 (W) 50 (W) 69 (W) 37 (W) 3 (W)
HIV HIV HIV HIV HIV HIV MTX HIV HIV HIV None None None None None None
Copathogen
EBV DNA, copies/mL
EBV RNA
CMV CMV CMV JCV JCV VZV VZV Streptococcus pneumoniae S. pneumoniae Cryptococcus neoformans HSV-1 VZV WNV Ehrlichia chaffeensis E. chafeensis Mycoplasma pneumoniae
34,000 615,000 5000 180,000 6200 438,000 5000 288,000 33,400 5900 33,900 900 63,490 11,668 25,310 56,027
+ + NA + NA NA NA NA NA NA + NA NA Indeterminate + NA
EBV serological result
Past Past Past Past
NA NA NA NA NA NA NA NA NA NA NA infection infection NA infection infection
Diagnosis CNSL CNSL CNSL PML PML PRM Enceph Enceph Mening Enceph Enceph Enceph Enceph Enceph Enceph Enceph
NOTE. +, positive; B, black; CMV, cytomegalovirus; CNSL, central nervous system lymphoma; Enceph, encephalitis; F, female; HSV-1, herpes simplex virus type 1; JCV, JC virus; Mening, meningitis; M, male; MTX, methotrexate; NA, not available; PML, progressive multifocal leukoencephalopathy; PRM, polyradiculomyelitis; VZV, varicella zoster virus; W, white; WNV, West Nile virus.
BRIEF REPORT • JID 2005:191 (15 January) • 235
their clinical management, and they improved. However, the other 7 patients with encephalitis or meningitis, who did not receive acyclovir or other antiviral agents active against EBV, also improved or remained in stable condition. The EBV load in the CSF varied between 900 and 615,000 DNA copies/mL (mean, 112,612 EBV DNA copies/mL; median, 33,650 EBV DNA copies/mL). There were no significant differences in the EBV load associated with the patients’ immune status or clinical presentation. Six samples obtained from patients with either CNSL (n p 2 ), encephalitis (n p 3 ), or PML (n p 1) were tested for the presence of EBV mRNA. Five samples were positive, and 1 was indeterminate. In the indeterminate case, the primers used for RNA amplification were not able to amplify the DNA, which indicated mismatches at the primer annealing sites and precluded any interpretation of the negative result. These data suggest that EBV actively replicates in the CNS when it occurs in association with another pathogen, even in cases in which the dominant clinical findings could be ascribed to the other pathogen, as in the case of PML or HSV-1. EBV serological testing was performed in 4 patients with encephalitis, all of whom had evidence of past EBV infection, which suggests that the CNS infection was due to reactivation or reinfection. We analyzed the cytological, biochemical, and virological characteristics of the CSF from patients with dual infections and compared them with those seen in encephalitis caused by EBV as the sole pathogen [5]. For this analysis, we excluded the 3 cases of CNSL, because EBV has a well-established causative role in CNSL, and the pathogenesis of CNSL is different from inflammatory conditions, in which tissue destruction and inflammation are the predominant pathogenic mechanisms. The cases were stratified by viral versus nonviral coinfection. The white blood cell (WBC) counts in the CSF were 7–1400 cells/mL (mean, 424 cells/mL; median, 130 cells/mL). There were no significant differences between the mean SE of WBCs in the cases with a viral copathogen (558 212 cells/mL) and the cases with a bacterial or fungal copathogen (267 182 cells/ mL) (figure 1A). By comparison, in our previous study of 10 patients with EBV encephalitis, patients had a mean SE WBC count of 143 62 cells/mL [5], which tended to be lower than the counts seen in patients with dual infection but reached statistical significance only in comparison with the patients with viral coinfections (P p .02). There was a significant linear association of the EBV load with the total number of cells in the CSF (r 2 p 0.4; P p .02) but not with the number of lymphocytes (r 2 p 0.16; P p .2). The protein concentrations in the CSF were 42–940 mg/dL (mean, 291 mg/dL; median, 148 mg/dL). There were no significant differences between viral and nonviral coinfections (mean SE, 250 88 and 337 163 mg/dL, respectively). This compares with values of 164 31 mg/dL that we previ236 • JID 2005:191 (15 January) • BRIEF REPORT
Figure 1. Cytochemical and virological characteristics of the cerebrospinal fluid of Epstein-Barr virus (EBV)–associated dual infections of the central nervous system and of meningoencephalitis caused by EBV alone. Columns represent the means SE of observations from 7 patients with viral coinfections, 6 patients with nonviral coinfections, and previously published information [5] on 11 patients with EBV meningoencephalitis. WBCs, white blood cells.
ously reported regarding patients with EBV as a sole pathogen [5] (figure 1B). There was no significant association between the intensity of pleocytosis and the protein content of the CSF. Glucose levels were abnormally low in only 2 patients, both of whom had a concurrent bacterial infection: 1 with pneumoccocal meningitis and 1 with ehrlichiosis. The EBV load was not significantly different between viral and nonviral CNS coinfections (mean SE, 103,900 60,530 and 70,500 44,180 EBV DNA copies/mL, respectively) (figure 1C), and it was comparable to the viral load previously reported by us in cases where EBV was the sole pathogen (mean SE, 46,340 20,610 EBV DNA copies/mL). Discussion. Our data indicate that ∼25% (16/60) of cases in which EBV CNS infection is diagnosed by CSF PCR are, in fact, dual infections in which a second pathogen is associated
with EBV. This is likely an underestimate, given that we only included cases in which definite evidence for a second pathogen was present. Dual infection might have gone unrecognized in some cases because of a lack of comprehensive diagnostic testing and/or a failure to detect a copathogen by use of currently available diagnostic tests. CNS coinfection with EBV affected both immunocompetent and immunocompromised hosts. Most copathogens differed between the 2 groups, except for VZV, which was found in both types of patients. Although the relative frequency of dual infections in these 2 groups of patients could not be determined in the present study, it is interesting to note that the incidence of CSF coinfections with EBV was similar at the 2 participating sites, despite the patient populations referred to these laboratories being quite different. UCHSC Virology draws specimens from tertiary-care facilities with many immunocompromised hosts. In contrast, organ-transplant recipients and patients with AIDS were excluded from the TUES study. Most EBV-associated CNS coinfections had no clinical, virological, or CSF characteristics that could distinguish them from cases in which EBV was the sole pathogen [5]. There was a trend for viral loads to be higher in patients with dual infections than in those with single infection, but this difference did not reach statistical significance. The number of CSF WBCs in patients with viral coinfections was higher than that in patients with EBV encephalitis, although this difference was statistically significant only for the group with a viral copathogen. We did not detect significant differences in CSF WBC counts between EBV infections associated with another viral versus with a nonviral pathogen, despite the fact that single-agent bacterial infections typically have higher CSF WBC counts and protein content and lower glucose levels than viral CNS infections. These data suggest that CNS coinfections with multiple viral agents are associated with increased local inflammatory responses. This finding, if confirmed, might be used in clinical practice by investigating the presence of multiple pathogens in cases of viral encephalitis with highly cellular CSF. The clinical significance of EBV coinfection of the CNS is unclear. Previous work has suggested that coinfection caused by 2 herpesviruses might worsen the prognosis of CNS disease [7]. In contrast, all of our patients, with the exception of 2 with CNSL, improved or recovered. Although our patients seemed to recover independently of EBV-specific therapy, it is
conceivable that only the most severe cases received acyclovir, which makes the outcome data difficult to interpret. With respect to the pathogenesis of the dual infections, our findings are consistent with the hypothesis that the intensity of the inflammatory response caused by a primary CNS infectious agent may trigger the reactivation of EBV and superinfection of the CNS. The latency site in these cases could be the CNS, as has been reported elsewhere [8, 9], or it could be infiltrating lymphocytes. In conclusion, given the increasing availability of molecular diagnostic testing for CSF, it is important to recognize that dual infections are common and that the identification of EBV in the CSF does not preclude further testing for a second, potentially treatable agent. Conversely, EBV actively replicates in the CNS even when it is found in association with another infectious agent.
References 1. Studahl M, Hagberg L, Rekabdar E, Bergstrom T. Herpesvirus DNA detection in cerebral spinal fluid: differences in clinical presentation between alpha-, beta-, and gamma-herpesviruses. Scand J Infect Dis 2000; 32:237–48. 2. Cinque P, Vago L, Dahl H, et al. Polymerase chain reaction on cerebrospinal fluid for diagnosis of virus-associated opportunistic diseases of the central nervous system in HIV-infected patients. AIDS 1996; 10: 951–8. 3. Portolani M, Cermelli C, Meacci M, et al. Epstein-Barr virus DNA in the cerebrospinal fluid of patients with human immunodeficiency virus infection and central nervous system disorders. New Microbiol 1999; 22:369–74. 4. Maurmann S, Fricke L, Wagner HJ, et al. Molecular parameters for precise diagnosis of asymptomatic Epstein-Barr virus reactivation in healthy carriers. J Clin Microbiol 2003; 41:5419–28. 5. Weinberg A, Li S, Palmer M, Tyler KL. Quantitative CSF PCR in EpsteinBarr virus infections of the central nervous system. Ann Neurol 2002; 52:543–8. 6. Glaser CA, Gilliam S, Schnurr D, et al. In search of encephalitis etiologies: diagnostic challenges in the California Encephalitis Project, 1998– 2000. Clin Infect Dis 2003; 36:731–42. 7. Tang YW, Espy MJ, Persing DH, Smith TF. Molecular evidence and clinical significance of herpesvirus coinfection in the central nervous system. J Clin Microbiol 1997; 35:2869–72. 8. Sanders VJ, Felisan S, Waddell A, d’Tourtellotte WW. Detection of herpesviridae in postmortem multiple sclerosis brain tissue and controls by polymerase chain reaction. J Neurovirol 1996; 2:249–58. 9. Vinters HV, Wang R, Wiley CA. Herpesviruses in chronic encephalitis associated with intractable childhood epilepsy. Hum Pathol 1993; 24: 871–9.
BRIEF REPORT • JID 2005:191 (15 January) • 237
