Human Cardioviruses, Meningitis, and Sudden Infant Death Syndrome in Children

Cardioviruses cause myocarditis and encephalomyelitis in rodents; human cardioviruses have not been ascribed to any disease. We screened 6,854 cerebrospinal fluid and 10 myocardium specimens from children and adults. A genotype 2 cardiovirus was detected from a child who died of sudden infant death syndrome, and 2 untypeable cardioviruses were detected from 2 children with meningitis.

Cardioviruses cause myocarditis and encephalomyelitis in rodents; human cardioviruses have not been ascribed to any disease. We screened 6,854 cerebrospinal fl uid and 10 myocardium specimens from children and adults. A genotype 2 cardiovirus was detected from a child who died of sudden infant death syndrome, and 2 untypeable cardioviruses were detected from 2 children with meningitis.

T he cardioviruses (family Picornaviridae, genus
Cardiovirus) are pathogens of rodents and include a murine encephalomyocarditis virus and Theiler's virus and related strains (species Theilovirus), the latter serving as laboratory models of the pathogenesis of multiple sclerosis in mice (1). The existence of specifi c human cardioviruses was suspected in the 1960s in conjunction with a rare infectious neurodegenerative disease known as Vilyuisk encephalitis (2,3). Recently, human cardioviruses (hCVs) were identifi ed in archived diagnostic cell culture supernatants (4) and in clinical samples from children with diarrhea or respiratory infection (5,6). Up to 8 different putative hCV types have since been characterized in human feces (7).
Despite the remarkable pathogenicity of rodent cardioviruses, specifi c disease associations of hCV could not be made. An initial clinical study yielded no evidence of hCV in cerebrospinal fl uid (CSF) of 400 patients with aseptic meningitis, encephalitis, or multiple sclerosis (8).
To evaluate the pathogenetic potential of these emerging viruses, we investigated 6,854 CSF specimens from adults and children with neurologic disease and 10 myocardium specimens from infants who had died of sudden infant death syndrome (SIDS).

The Study
CSF specimens were collected from 3 cohorts. The fi rst cohort comprised 2,562 specimens sent during 1998-2008 to the Institute of Virology, University of Bonn Medical Center (UBMC), Bonn, Germany, for routine investigation of meningoencephalitis (333 from the Department of Pediatrics and 2,229 from other departments). The second cohort comprised 3,960 specimens collected during 1982-2008 at the UBMC children's hospital from children with cancer and neurologic complications during chemotherapy. The third cohort comprised 348 specimens from hospitalized children with clinical meningitis or encephalitis in which no etiologic agent had been found; the specimens were sent for virologic investigation to the Institute for Hygiene and the Environment in Hamburg, ≈400 km from UBMC, during 2006-2008. Myocardium specimens were collected during 2010 at the UBMC Institute for Forensic Medicine from 10 epidemiologically unlinked children who died of SIDS.
Viral RNA was purifi ed from clinical specimens by using the Viral RNA Mini and RNeasy Mini kits (QIAGEN, Hilden, Germany). Detection of hCV RNA was done in pools of 2-10 specimens by using quantitative realtime reverse transcription PCR (RT-PCR) and nested RT-PCR specifi c for the viral 5′ untranslated region (5′-UTR), as described (6). Amplifi cation of further hCV genomic regions from individual positive specimens was conducted by using ≈20 sets of different nested RT-PCRs (primers available on request from C.D.).
In 2 of 681 CSF specimens (n = 333 and n = 348 from cohorts 1 and 3, respectively) from children with meningitis (online Appendix Table, wwwnc.cdc.gov/EID/ article/17/12/11-1037-TA1.htm), hCV RNA was detected at low concentrations (1.14 × 10 4 and 9.63 × 10 2 copies/ mL). In 1 of these patients, hCV was also detectable in feces (9.50 × 10 2 copies/g). In 1 of 10 myocardium specimens, hCV was detected by nested RT-PCR, and results of quantitative real-time RT-PCR were negative. Underquantifi cation because of nucleotide mismatches below oligonucleotide binding sites and contamination of nested RT-PCR was excluded by sequence comparison (up to 5% nt divergence from other hCV strains, including the positive control). Serum and liver specimens from the patient who died of SIDS were negative according to realtime RT-PCR. No histopathologic alterations could be observed in myocardial tissue from this same patient.
To evaluate whether detected hCV strains differed from previously described genotypes, amplifi cation and nucleotide sequencing of additional genomic regions was attempted. In a case of meningoencephalitis (specimen 07/03981), we sequenced a 1,297-nt fragment comprising the near complete 5′-UTR and the fi rst 489 nt of the structural protein gene (leader, viral protein [VP] 4 domain, and upstream VP2 domain, GenBank accession no. JN209931). Despite repeated trials, further sequence fragments could be amplifi ed neither from the specimen from this patient nor from that from the second patient with meningoencephalitis that showed very low virus concentrations (specimen VI1607). From the specimen from the SIDS patient (specimen 347/10), amplifi cation of the complete structural genome and partial nonstructural genome was successful (5,333 nt, GenBank accession no. JN209932). This virus belonged to hCV genotype 2 in the VP1 genomic region (i.e., the region used for the designation of genotypes) (Figure, panel A). The CSF specimen 07/03981 was also phylogenetically related to genotype 2 viruses in the 5′-UTR and Leader-VP2 genomic regions ( Figure, panels B and C). On the basis of the 5′-UTR sequences, the closest known relative to both viruses was D/VI2229, obtained in Germany in 2004 (nucleotide percentage distance 4.7% for the SIDS specimen and 0.9% for the CSF specimen). In the structural protein gene fragment, the closest relative of both viruses was a strain obtained in the Netherlands in 2008 (Nijmegen2008, nucleotide distance 13.9% for the SIDS specimen and 3.5% for the CSF specimen). This suggested geographic rather than phylogenetic clustering of viruses detected within and beyond the respiratory and enteric tracts. However, formal and fi nal virus typing is pending because VP1 regions could not be sequenced from 2 viruses.
Absence of other detectable pathogens in 1 of the meningoencephalitis case-patients (07/03981) made causation by hCV plausible (online Appendix Table). For the second case (VI1607), an enterovirus was co-detected by real-time RT-PCR in CSF and feces. Serotyping from feces classifi ed this virus as echovirus type 30, known to cause aseptic meningitis. For the specimen from the child who died of SIDS, a rhinovirus was co-detected at low concentrations (real-time RT-PCR threshold cycle value >40), most compatible with shedding after previous respiratory infection (9).

Conclusions
The detection of hCVs in body compartments beyond the respiratory and enteric tracts is novel and suggests a role of these viruses in organ-related disease. A low detection rate in CSF does not contradict a general potential of these viruses to cause meningoencephalitis, as exemplifi ed by enteroviruses for which lack of detection in CSF despite clear association with disease is not uncommon (10). Considering links between the related Theilovirus and demyelinating disease in laboratory models (1) outcomes of patients with hCV infection of the central nervous system should be followed up. Such longitudinal studies should include suffi cient numbers of patients because natural infections with Theilovirus in rodents are common and will less frequently result in multiple sclerosis-like disease than in laboratory models (1). The rarity of hCV detection in our study suggests the assembly of such cohorts to be a diffi cult and lengthy task that could benefi t greatly from international coordination. Despite the absence of histopathologic alterations, the detection of hCV in a child who died of SIDS is remarkable because the related encephalomyocarditis virus constitutes a prototypic model for myocarditis in mammals (11). Again, the high human seroprevalence against hCV (12) will complicate epidemiologic studies, yet investigations of links between hCV and SIDS are highly justifi ed because diarrhea is an acknowledged risk factor for SIDS (13).
A limitation of our study is that the VP1 genomic region of the viruses detected in CSF could not be obtained. In analogy to enteroviruses and parechoviruses, certain genotypes may be associated with distinct disease profi les, like polioviruses with encephalitis or parechovirus 3 with meningitis (14). Although we were able to classify the virus detected in the child who died of SIDS as a common genotype 2, the partial hCV sequence from a patient with meningitis did not permit typing because hCVs, as all picornaviruses, recombine frequently (15). We thus cannot exclude that the viruses detected in the meningitis cases may have acquired distinct features in their capsid protein or elsewhere that might infl uence pathogenicity.