Title: Newly Identified Viruses in Human Gastroenteritis
Abstract: Infectious diarrheal diseases remain a common cause of morbidity and mortality especially in children younger than 5 years in low-income and middle-income countries.1 The most common viral pathogens known to cause acute gastroenteritis in humans are rotavirus A, norovirus, sapovirus, enteric adenoviruses and astrovirus. In addition to these well-established etiologies, a large number of moderate to severe diarrhea cases go without etiological diagnosis and many unanswered questions surrounding the burden and etiology of diarrhea, and the factors determining clinical expression of infection remain. Sequence-independent amplification of nucleic acids combined with next-generation sequencing technology and bioinformatics analyses or viral metagenomics is a relatively new strategy for rapid identification of viruses in clinical and public health settings. In contrast to classical molecular detection techniques, even those detecting multiple viruses, viral metagenomics theoretically allows the characterization of the entire virome in a clinical sample, consisting of known pathogens, novel pathogens that elude conventional testing, bacteriophages, plant viruses and unannotated sequences that may be of viral origin. Since the first applications of metagenomics to the field of gastroenteritis,2,3 it has been increasingly applied for virus discovery purposes resulting in the identification of a plethora of previously unknown viruses in human and animal enteric specimens.4 Many previously unknown viruses have been characterized in human stool alone in recent years, such as salivirus, cosavirus, bufavirus, picobirnavirus, recovirus, anellovirus, astrovirus, circovirus and polyomavirus (see Table, Supplemental Digital Content 1, https://links.lww.com/INF/C297).5 The increasing application of such “unbiased” or catch all detection methods raises questions for the clinician, trying to understand which information is relevant for patient management, and for the public health professional focusing on population health decisions. The expanding universe of microbes, discovered through virome and microbiome studies, has opened an entirely new field of research, seeking to understand the interactions between hosts and the “healthy” and disease causing microorganisms that inhabit the human body. This is particularly the case for the gut microbiome and virome, which not only consists of human-associated microbes but also reflect the host environment. Viruses identified in human stool samples may originate from a dietary source or may replicate in the gut microbiota, such as bacteria, parasites, protozoan or nematodes. Even in the case of viruses with human tropism, not only many factors, such as immune status, age and nutritional status, but also geographic location and even seasonal differences, play an important role in the exposure to and clinical outcome of virus infection.6 Viruses may only become pathogenic in the context of certain host backgrounds (ie, immunodeficiency, coinfections, genetic factors), exemplified by simian immunodeficiency virus infections in different primate host species.7 This underlines the importance of studying both host and pathogen parameters in an integrated way in pathogenicity studies. NEWLY IDENTIFIED VIRUSES IN HUMAN GASTROENTERITIS: PATHOGENS OR NOT? Establishing disease causation in the molecular era has been debated since the start of the use of molecular detection techniques in biomedical research laboratories.8 Fulfilling Koch postulates for all newly identified viruses remains unlikely. Comparing viral incidence or prevalence in diseased versus matched controls, seroconversion and/or detection of antigen/nucleic acid in affected tissue may aid in studying viral pathogenicity. As nicely reviewed previously,9 major hurdles exist in performing these types of studies. The comparison of virus prevalence in disease cases with healthy controls requires careful subject matching to preclude misleading results explained by host differences instead of viral pathogenicity. Healthy control samples are often a limiting factor, as these are not ordinarily collected in medical settings. Measurement of antibody responses requires (often difficult) cell culture systems and/or proper antigen synthesis and positive control sera, and biopsy results of patients for antigen detection are often not available.9 In addition, the extent of viral genetic and phenotypic diversity is complicating development of specific and sensitive assays. Thus, finding a new virus does not automatically mean unveiling its clinical significance as is highlighted below by conveying what is currently known of a subset of the most studied viruses, focusing on virus families known to infect vertebrates and for which case-control studies were conducted (see Table, Supplemental Digital Content 1, https://links.lww.com/INF/C297). Picornaviridae Many new viruses belonging to the family Picornaviridae have been identified in recent years in stool samples. The Picornaviridae family contains human and animal viruses of considerable clinical and socioeconomic importance, such as enterovirus, parechovirus and hepatitis A virus. In addition, 4 new genera are recognized: cardiovirus (saffold virus), cosavirus, kobuvirus (Aichivirus) and salivirus (klassevirus), viruses which were identified in human stool but for which the clinical significance in gastroenteritis or other disease syndromes remains unclear.10,11 Cardiovirus Phylogenetic analysis suggests the existence of 8 distinct genetic saffold virus lineages. Screening for cardiovirus infections has occurred in many countries, including Afghanistan, Canada, China, Germany, Pakistan and the US, showing that this virus has spread worldwide with polymerase chain reaction (PCR) prevalence rates in enteric specimens ranging from 0.5% to 12%.10–15 Cardioviruses are mostly found in infants and children <6 years of age.10–15 Seroprevalence studies of saffold viruses 2 and 3 using virus neutralization assays in Africa, Asia and Europe revealed that generally 75% of children <24 months are seropositive versus >90% in older children and adults.13,15 This epidemiologic pattern suggests that cardiovirus infection is acquired early in life, similar to other viruses in the family Picornaviridae. So far, epidemiological studies and 2 case-control studies in young children have failed to provide a clear picture of the relationship between cardiovirus infection and actual disease in humans.10,12 Rather, available data suggest that saffold viruses rarely cause disease and most likely go unnoticed in a high proportion of infections.11,16 However, this is also the case for other picornaviruses, that nonetheless cause serious disease: for polio, an estimated 1:100 to 1:1000 infected individuals develop neurological illness. Such disease associations, however, cannot be detected through the studies that have been conducted so far. Cosavirus Cosaviruses are classified into 5 different species11 and were originally identified in 2008 in the feces of South Asian children with non-polio acute flaccid paralysis17 and shown to be present at high prevalence in feces of both healthy (44% and 25.4%) and paralyzed (49% and 42.8%) children from Pakistan and Tunisia, respectively.17,18 In China, cosavirus was detected in children with (3.2%) and without (1.6%) diarrhea.19 Cosavirus prevalence was also reported in the feces of healthy Brazilian children from a community child-care center (49% in 2008 and 6.5% in 2011) and with gastroenteritis in a pediatrics department (3.6%).20 Strong associations of cosavirus infection with gastroenteritis were not obtained. Kobuvirus The Kobuvirus genus contains 3 species, Aichivirus A–C, of which only Aichivirus 1 from species A infects humans.21 Seroprevalence studies performed in Asia, Europe and North Africa demonstrated high levels of Aichivirus 1 antibodies (80–99%) in adults, suggesting a worldwide distribution of the virus.21 In contrast, much lower reverse-transcription–PCR prevalence data were obtained, with Aichivirus 1 being detected in 0.5%–3% of human gastroenteritis cases in Asia, Europe, South-America and Africa, similar to observations with saffold viruses.10,11,20,21 The epidemiological data suggest that the virus might circulate without causing any symptoms or with mild symptoms that do not require medical attention. Salivirus Since the identification of saliviruses, they have been detected in stool samples of children with gastroenteritis worldwide, with prevalence rates ranging from 2.1% to 8.6%.10,11,22 A possible association was reported between salivirus detection and gastroenteritis in case-control studies.10,23,24 However, the virus has also been detected in healthy children and in association with other known enteric pathogenic viruses, such as norovirus and adenovirus, and thus, indicating that the potential association of salivirus with enteric disease needs to be studied in more detail, similar to other members of the family Picornaviridae.22,23 Polyomaviridae Virus discovery programs using clinical samples from a range of conditions have yielded multiple newly identified viruses in the family Polyomaviridae, including KI polyomavirus (KIPyV), WU polyomavirus (WUPyV), Merkel cell polyomavirus (MCPyV), human polyomavirus 6 (HPyV6), HPyV7, trichodysplasia spinulosa-associated polyomavirus, HPyV9, MW polyomavirus (MWPyV), HPyV10, MX polyomavirus, STL polyomavirus and HPyV12. A case-control study to explore the prevalence of 10 human polyomaviruses (BKPyV, JCPyV, KIPyV, WUPyV, MCPyV, HPyV6, HPyV7, trichodysplasia spinulosa-associated polyomavirus, HPyV9 and MWPyV) was conducted in China by testing fecal specimens from 211 hospitalized children with diarrhea and from 208 asymptomatic control subjects collected between April 2011 and January 2012.25 Only KIPyV (0.5 and 0%), WUPyV (4.3% and 1.9%), MCPyV (30.3% and 27.9%) and MWPyV (1.4% and 2.9%) were detected in healthy controls and children with diarrhea, respectively.25 STL polyomavirus was present in 2.2% of hospitalized Chinese children with gastroenteritis and 3.0% of healthy children.26 Despite the relatively frequent detection of (some of the) human polyomaviruses in fecal samples, a causative role of polyomaviruses in gastroenteritis was not supported.25 Parvoviridae New human bocaviruses (HBoVs) and bufaviruses were identified in the family Parvoviridae. HBoV subtypes 1–4 have been detected in respiratory and gastrointestinal infections worldwide.27 In contrast to HBoV1, HBoV2, HBoV3 and HBoV4 occur mainly in stool.28 HBoV2 is the most common with PCR prevalence up to 26%, followed by HBoV3 (5%) and HBoV4 (2%). The detection rates in adults may be lower than in children.28 Case-control studies to investigate a link between bocavirus 1 and gastroenteritis in China between 2006 and 2008 showed no statistically significant differences in the prevalence (3.5%–4.3%), clinical disease presentation and virus loads between children with and without gastroenteritis.29,30 Human bocavirus 2 was detected at much higher prevalences of 20.4% and 12.3% in children with diarrhea and healthy controls, respectively.30 An association of human bocavirus 2 infection and acute gastroenteritis was shown in a case-control study in Australian children, but not in adults or children in the United Kingdom.31,32 Bufaviruses were first detected in West African children with diarrhea33 and have subsequently been shown in diarrhea specimens from all over the world with PCR prevalence rates between 0.8% and 4% in patients with gastroenteritis.34,35 Astroviridae The classic human astroviruses serotypes 1–8 have been associated with acute gastroenteritis in humans. Many more different human astroviruses have been identified in diarrhea specimens in recent years, such as MLB1-3 and VA1-4, which seem to have a worldwide distribution.36 A case-control study on MLB1 and classic astrovirus infections in India showed that classic astroviruses were significantly associated with diarrhea, whereas the MLB1 strain was not. In addition, MLB1 loads were not different between symptomatic and asymptomatic children.37 However, MLB1 was significantly associated with diarrhea in Kenya but not in The Gambia.38 A seroprevalence study in North America indicated that MLB1 seropositivity was high in children <6 months old, decreased to a nadir at 12–23 months old and increased to 100% by adulthood.39 Similar high prevalence rates were seen for astrovirus VA1 in healthy U.S. blood donors and evaluation of serum samples from different pediatric age groups revealed that the prevalence of antibodies in 6–12 months, 1–2 years, 2–5 years and 5–10 years old was 20%, 23%, 32% and 36%, respectively, indicating rising seroprevalence with age.40 The seroprevalence studies indicate that infection with these new astroviruses is common, but a causative role in gastroenteritis remains to be shown. Overall, the development of next-generation sequence platforms in combination with metagenomics approaches has yielded a growing list of viruses detected in human gastroenteritis cases, for which the (possible) clinical disease association remains to be elucidated, highlighting the fact that identifying new viruses is at present “easier” than showing association with disease. FUTURE ROLE OF VIRAL METAGENOMICS IN DISEASE ASSOCIATION STUDIES? It is clear that virome profiling at present does not add much to the needs of the clinician or public health specialist. Nevertheless, viral metagenomics can be used as a potentially cost-saving research tool to study viral incidence, level of viral coinfections and their correlation to clinical diarrheic disease in well-designed cohort studies in the years to come and to resolve some of the outstanding questions simultaneously for a wide range of viruses. Provided samples from large enough matched cohorts are analyzed from different human populations with and without disease, the obtained data may not only provide basic insight into the viral burden of diarrhea in general instead of focusing on one or a few viral pathogens as is classically done, but it may allow (i) elucidation of viral pathogens or combinations of pathogens involved in disease burden, (ii) surveillance for as yet unidentified enteric viruses and zoonotic events, (iii) the study of effects of vaccination on viral incidence levels (eg, rotavirus) and whether other viruses fill the niche that vaccination leaves behind and (iv) an appraisal of host differences in viral disease burden. Theoretically, it would be possible to decipher the entire spectrum of viruses present in the human enteric tract at a certain point in time. This gut virome is not a given: it will vary over time, with diet, with genetic profile and health status of the host, among others.6 Finding clinically relevant information on viruses and their evolutionary changes becomes theoretically possible but requires in depth bioinformatics analyses of systematically sampled stool with sufficient metadata for meaningful analysis. Currently, such analyses are time-consuming, expensive and complex precluding routine use of next-generation sequencing in clinical and public health settings. The fast developments in the metagenomics field, however, undoubtedly will lead to advances in this respect. CONCLUSIONS When talking about newly identified viral pathogens of human gastroenteritis, one can conclude that breakthroughs in the field of metagenomics have had far-reaching effects on the identification and characterization of new viruses. It also led to the recognition that a growing number of enteric gastroenteritis cases that were once attributed to unknown causes may be caused or triggered by infectious viral agents. Solid evidence, however, that these new viruses actually cause disease is difficult to obtain. Still, the technology is fast evolving, and viral metagenomics when applied judiciously could address some important outstanding questions, relevant both for virus diagnostics and surveillance. Future metagenomics will potentially enable generating the complete genomic information from clinical samples independent of the sector (public health, veterinary health and food safety), the type of pathogen (viruses, bacteria and parasites) and the sample type. The primary sequence output and derived data, such as assemblies and functional annotation data, in combination with associated clinical, microbiological, and epidemiological data would lead to a more in depth view of the human virome in correlation to disease. This does require engagement of clinicians, bioinformaticians, epidemiologists and laboratory scientists, to fully capture the potential of these novel technologies for infectious disease research. ACKNOWLEDGMENTS This work was partially funded by the European Commission COMPARE H2020 project under grant agreement No 643476, the Dutch government project number FES0908, by Netherlands Genomics Initiative (NGI) project number 050-060-452 and ZonMW TOP project 91213058.