Serious Invasive Saffold Virus Infections in Children, 2009

This virus might have caused previously unexplained cerebral infections and deaths in children.

M olecular biology has revolutionized the diagnostics of infectious diseases through the introduction of more sensitive and specifi c diagnostic tests. Despite these advances, the etiologic agents of many apparent infections are still unidentifi ed. For example, the etiologic agent is unknown for many cases of apparent pneumonia (1); in a study conducted in California, USA, despite extensive testing and evaluation, an underlying cause of encephalitis was unidentifi ed for 207 (62%) of 334 patients (2).
During the past few years, intensive searches for new viruses, using conventional virologic methods and metagenomics, have resulted in the discovery of several new viruses. During the past decade, the family Picornaviridae has grown as the number of recognized genera has increased from 6 to 12 (3,4); the numbers of species, types, and subtypes have increased even more. However, only viruses from 3 genera (Enterovirus, Hepatovirus, and Parechovirus) have been fi rmly established as being capable of causing clinically signifi cant disease in humans. Viruses from other genera (Cardiovirus, Cosavirus, and Kobuvirus) have so far been detected only in noninvasively collected human sample material such as fecal and respiratory samples, and their clinical signifi cance remains to be fully elucidated. (Invasively collected sample material is that from tissues considered sterile, i.e., devoid of microorganisms.) The phylogenetic relationships of human picornaviruses are shown in Figure 1. Most picornaviruses that are pathogenic to humans are ubiquitous viruses capable of causing a variety of diseases, from monosymptomatic febrile infection to severe infection in the central nervous system and myocardium. However, most infections with these viruses are asymptomatic (5).
Saffold virus (SAFV) was discovered by Jones et al. in 2007 by sequence-independent genomic amplifi cation of virus isolated from a fecal sample (6). The sample had been obtained in 1981 from an 8-month-old child with fever of unknown origin. The genetic sequence of the virus indicated that the virus belonged to the species Theilovirus of the genus Cardiovirus, which contains 3 other members: Theiler's murine encephalomyelitis virus (TMEV), Vilyuisk human encephalomyelitis virus (VHEV), and Thera virus. In mice, TMEV is capable of causing infection in the central nervous system, and some variants of this virus cause a persistent infection and even multiple sclerosis -like disease (7). VHEV was isolated in the 1950s from cerebrospinal fl uid (CSF) from a patient with Vilyuisk encephalomyelitis, a progressive neurologic disorder that occurs in indigenous populations of an isolated part of eastern Siberia (8). However, the correlation between VHEV and Vilyuisk encephalomyelitis is still uncertain because VHEV has been isolated by multiple passages in mice and thus may represent a highly divergent strain of TMEV. Thera virus (previously named Theilerlike rat virus) has been isolated from rats, but the clinical signifi cance of infections with this virus is unknown (9,10). The genus Cardiovirus also contains a second species called encephalomyocarditis virus. Only 1 serotype is known, and it is capable of causing encephalitis and myocarditis in various animals (11,12).
Since the discovery of SAFV, several articles have provided insight into its epidemiologic and, to a minor degree, clinical signifi cance. Saffold viruses are distributed worldwide (6,(13)(14)(15)(16)(17)(18)(19), and 2 serologic studies have demonstrated that infection occurs early in life (14,20). However, fi nding an association with human disease has thus far been elusive. Most studies (13,15,17,(20)(21)(22) have tried to associate SAFV with gastroenteritis, but no convincing results have been produced. A few studies (16,18,21,23) analyzed the clinical signifi cance of SAFV virus in the respiratory system, but no substantial association between the virus and respiratory symptoms or disease has been made. Only 1 study (21) reports having tested invasively collected sample material (CSF samples), but no fi ndings were positive.
To investigate the possible invasive potential of SAFV in humans, we developed a diagnostic PCR and tested CSF samples from a group of children. SAFV was detected in 2 of these children.

Materials and Methods
We tested previously submitted CSF samples for SAFV, reviewed the patients' medical records, and sequenced the viruses isolated.

Fecal Samples
We selected fecal samples from 479 children <5 years of age with gastroenteritis that had been submitted for viral diagnostic testing from September 2009 through February 2010 and tested them for SAFV. Nucleic acid extracted from these samples was combined into 48 pools, with 9 or 10 samples per pool. Samples from pools with positive results were identifi ed, and new extractions from these pools were tested individually. However, enough sample material for new extractions was available for only about half of the samples.

Nucleic Acid Purifi cation
Nucleic acids were extracted from 200 μL of CSF or blood (from SAFV-positive patients) by using the QIAamp DNA Blood Mini Kit (QIAGEN, Hilden, Germany) and semiautomatic extraction on the QIAcube instrument (QIAGEN). Nucleic acid was extracted from 200-μL fecal suspension (10% in phosphate-buffered saline) by using the Total Nucleic Acid Isolation Kit (Roche Diagnostics GmbH, Mannheim, Germany) on the MagnaPure LC instrument (Roche Diagnostics GmbH).

Nucleic Acid Amplifi cation and Detection
Five microliters of extracted material was used per reverse transcription PCR (total volume 25 μL) by using the OneStep RT-PCR Kit (QIAGEN). The reaction mixtures contained 1 μmol/L of each primer and 0.2 μmol/L of probe. Design of the primers and probe was based on an alignment of all available SAFV sequences in GenBank (www.ncbi.nlm.nih.gov/genbank) in July 2010 by using ClustalW (www.clustal.org) and Primer3 (http:// frodo.wi.mit.edu/primer3) software. The primers and probe are selective for a highly conserved part of the 5′ untranslated region (Table). The Strategene Mx3005P real-time thermocycler instrument (Agilent Technologies A/S, Horsholm, Denmark) was used for amplifi cation and detection with the following settings: 50°C for 20 min, 95°C for 15 min, followed by 45 cycles of 95°C for 15 s and 55°C for 1 min.

Genotyping by PCR and Sequencing
Genotyping was conducted by nested PCR and sequencing of parts of the viral protein (VP) 1 and VP2 regions of the capsid gene by using primers listed in the Table. The inner VP1 and VP2 primers amplifi ed DNA fragments of ≈599 and 577 bp, respectively. PCR products were purifi ed by using the High Pure PCR Purifi cation Kit (Roche Diagnostics GmbH) before sequencing, which was performed by using the inner PCR primers on an ABI automated sequencer and BigDye version 1.1 chemistry (both from Applied Biosystems, Darmstadt, Germany). Sequences were aligned, and phylogenetic analysis with known reference sequences was performed by using MEGA4 software (www.megasoftware.net). The sequences have been submitted to GenBank under accession nos. JF693612-23.

Results
SAFV was detected in CSF from 2 of the 319 children. Additional sample material from these 2 children was subsequently obtained and tested. From child 1, blood and CSF collected at the same time and a fecal sample collected 2 weeks later were tested; only test results for the fecal sample were positive for SAFV. From child 2, a postmortem blood sample and a myocardial biopsy sample were tested; test results for each sample were positive for SAFV.

Child 1
Child 1 was a 16-month-old, previously healthy boy who became ill in May 2009. The boy had a fever 6 days before hospital admission, followed 1 day later by sudden onset of monosymptomatic ataxia, with no fever. The ataxia fl uctuated from causing an insecure gait to walking into things and falling. The patient also had intermittent diffi culty controlling his arms when trying to eat. No history of recent travel was reported. The boy was in otherwise good health; he had no abnormal psychological symptoms and retained a normal degree of consciousness throughout the acute phase of the disease. Differential diagnoses at hospital admission were intracranial tumor or viral cerebellitis. The boy's 4-year-old sister remained healthy.
Laboratory test results are listed in the online Appendix Table  (wwwnc.cdc.gov/EID/article/18/1/11-0725-TA1. htm). At hospital admission, CSF values (leukocyte count, protein level, and glucose level) were within reference ranges, and no microorganisms were detected. A magnetic resonance imaging scan of the brain showed a small venous anomaly in the left frontal lobe but no tumor, hemorrhage, or infl ammation. A fecal sample collected 2 weeks later yielded positive test results for parechovirus type 3 and negative results for enterovirus and adenovirus. Parechovirus was not found in the CSF or blood.
During the next 2 months, the ataxia remitted completely without sequelae. The diagnosis at this time was viral encephalitis, possibly caused by parechovirus type 3.

Child 2
Child 2 was a 27-month-old, previously healthy girl who was found dead in her bed in August 2009; she had no Emerging Infectious Diseases • www.cdc.gov/eid • Vol. 18, No. 1, January 2012 9 known history of disease or symptoms. During necropsy, signs of cerebral herniation were detected. A small vascular malformation surrounded by edema was found in the brain. No signs of encephalitis or hemolytic uremic syndrome were found. The following sample materials were collected: CSF, blood, feces, myocardium, pericardial aspirate, lung tissue, and respiratory secretions (online Appendix  Table). In the CSF, mononuclear pleocytosis was noted. Results of routine bacteriologic culture found coagulasenegative staphylococci in the CSF, a few nonhemolytic streptococci in the lung tissue, and a few nonhemolytic streptococci and a few Staphylococcus aureus organisms in the pericardial aspirate. Verotoxin-producing Escherichia coli was cultured from the fecal sample. The conclusion of the autopsy and laboratory fi ndings was that cerebral herniation was the immediate cause of death.
Virus from each of the 2 children was characterized by sequencing part of the VP1 region of the capsid gene, and the sequences were compared with those of other SAFVs detected in fecal samples from the patients with gastroenteritis ( Figure 2). Of the 48 fecal sample pools, 10 were positive for SAFV by PCR. From these pools, 6 individual samples were available for further testing; the VP2 capsid region was successfully sequenced for 4 of these 6 samples, and they were all SAFV type 2. Later, an SAFV type 2-specifi c VP1 PCR was designed, which provided VP1 sequences from all 6 fecal samples and samples from the 2 children reported here: the fecal sample from child 1 and the blood sample from child 2. The phylogenetic analysis ( Figure 2) showed that all viruses were SAFV type 2 and that the sequences arranged in 2 clusters with 8% nt differences and 5% aa differences between the clusters.

Discussion
Several novel viruses have recently been discovered by using microarrays or mass-sequencing methods (6,21). Almost all of these viruses have been found in noninvasively collected patient materials, making the correlation to clinical disease diffi cult to establish. For the picornavirus group, this issue is further complicated by knowledge that this virus family can cause a wide variety of diseases with a high proportion of nonspecifi c symptoms or asymptomatic infections (5).
We report 2 cases of invasive infection with SAFV type 2 in children. For each child, SAFV was detected in at least 1 compartment other than the central nervous system. This fi nding strengthens the evidence of an acute infection as the cause of clinical disease. In child 1, SAFV virus was found in the CSF and a fecal sample. In child 2, fi ndings were even more convincing because SAFV was found in 3 invasively collected samples: CSF, myocardium, and blood. No other credible cause of infection was found for either of the 2 children. In child 1, the only other positive fi nding was parechovirus type 3 in a fecal sample taken 2 weeks after onset of disease. Parechovirus was not found in the CSF and therefore seems unlikely as the cause of the acute symptoms. From child 2, several types of bacteria were identifi ed, but these seem unlikely to be the cause of death because culture of postmortem samples often grows commensal organisms. The bacteria from this child were not found consistently in the tissue samples, thus making systemic bacterial infection less likely. Verotoxinproducing E. coli also seems an unlikely cause of death because no diarrhea or signs of hemolytic uremic syndrome were present before the child's death or at autopsy.
At autopsy of child 2, a small vascular malformation surrounded by edema was found. This edema could be the cause of the herniation. However, the edema might also have been caused by cerebral infection or septicemia. This possibility is supported by the fi nding of mononuclear 10 Emerging cells in the spinal fl uid and the direct virus identifi cation in the CSF and blood. The exact cause of death and whether there is a connection between the infection and the changes surrounding the malformation are unclear. However, our investigations show that before her death, the child's blood contained SAFV. Child 1 had monosymptomatic ataxia preceded by 1 day of fever. The symptoms receded over the next few months, and the patient recovered fully. Child 2 died without any preceding symptoms or any known predisposing factors; this clinical picture is sometimes found for patients with enteroviral infections (24).
The fi nding of SAFV in invasively collected samples from the 2 children described here fi ts well with the knowledge about the picornavirus group. SAFV probably behaves similarly to the viruses in the enterovirus group, producing mainly asymptomatic infections but also producing nonspecifi c symptoms in other patients and severe disease in a few patients. Enteroviruses are known to cause many more or less organ-specifi c diseases such central nervous system infection (meningitis, encephalitis, and myelitis), myocarditis, enanthema, exanthema, and septicemia. These 2 cases fi t well within the expected range of diseases attributed to picornaviruses, but further studies are needed to determine the exact correlation of SAFV to disease in humans.
In a previous study, Chiu et al. looked for SAFV in 400 CSF samples but found none (21). Lack of detection could be explained by different assay sensitivities or by the selection of samples tested. In their study, patient selection was based on neurologic disease and patient ages were not reported.
Our study has also shown that Saffold virus type 2 circulated in Denmark in 2009, the same year that the children reported here became ill. Other studies have shown type 2 to be a common type of SAFV and to be circulating worldwide (13,15,16,18,20,21,23). Because SAFVs are single-stranded RNA viruses, a nucleotide variation of 8% among the restricted number of samples in this study is expected.
In conclusion, we have established SAFV virus as a cause of invasive infection and a highly probable cause of severe disease in children. More studies are needed to further illuminate the role of SAFV as a human pathogen.