Human parvovirus 4 ‘PARV4’ remains elusive despite a decade of study

Human parvovirus 4 (‘PARV4’) is a small DNA tetraparvovirus, first reported in 2005. In some populations, PARV4 infection is uncommon, and evidence of exposure is found only in individuals with risk factors for parenteral infection who are infected with other blood-borne viruses. In other settings, seroprevalence studies suggest an endemic, age-associated transmission pattern, independent of any specific risk factors. The clinical impact of PARV4 infection remains uncertain, but reported disease associations include an influenza-like syndrome, encephalitis, acceleration of HIV disease, and foetal hydrops. In this review, we set out to report progress updates from the recent literature, focusing on the investigation of cohorts in different geographical settings, now including insights from Asia, the Middle East, and South America, and discussing whether attributes of viral or host populations underpin the striking differences in epidemiology. We review progress in understanding viral phylogeny and biology, approaches to diagnostics, and insights that might be gained from studies of closely related animal pathogens. Crucial questions about pathogenicity remain unanswered, but we highlight new evidence supporting a possible link between PARV4 and an encephalitis syndrome. The unequivocal evidence that PARV4 is endemic in certain populations should drive ongoing research efforts to understand risk factors and routes of transmission and to gain new insights into the impact of this virus on human health.


Introduction
Human parvovirus 4 ('PARV4') was first described in 2005 1 , but many important features of its epidemiology, transmission, and clinical significance remain elusive 2 . PARV4 is a small non-enveloped, single-stranded DNA virus with a genome of approximately 5 kb, translated as two major and two minor open reading frames. Classified as a member of the genus Tetraparvovirus, in the family Parvoviridae (sub-family Parvovirinae), it sits phylogenetically alongside porcine and bovine hokoviruses 3 , while its closest human-tropic relatives are well-characterised B19V (genus Erythroparvovirus) and human bocaviruses (genus Bocavirus) (Table 1).
Ongoing investigation of PARV4 is being driven mainly by several intriguing features of its distribution: it is endemic in certain geographic areas 4 , but elsewhere is found confined only to certain high-risk groups 5 . The first reported identification of PARV4 was from an injecting drug user with hepatitis B virus (HBV) infection, who was screened for viral infections following presentation with influenza-like symptoms 1 . Based on this index case, subsequent studies in North America and Western Europe focused mainly on groups with risk factors for parenteral infection and those infected with other blood-borne viruses (BBVs). In these settings, a clear picture has emerged in which PARV4 infections are consistently and strongly associated with HIV, HBV and HCV, mostly in the setting of persons who inject drugs (PWIDs) and those with a history of multiple transfusion [5][6][7][8][9][10] . Despite attempts to identify it outside this setting, infection has not been detected in household contacts of seropositive individuals 11 and is infrequent in the general population 6,12 .
However, groups working in Africa have reported a completely different population epidemiology: in this setting, the detection frequency of PARV4 IgG (indicative of current or past exposure) ranges from 30% to 50% in the general population, without any consistent relationship with BBVs 4,13-15 . Acute infections associated with viraemia were detected at a frequency of 8.5% among young children in Ghana 16 .
The clinical syndromes that have been described in association with PARV4 are summarised in Table 1 and Figure 1. However, there are too few data to make any of these associations secure, and attributing causality in a robust way is inevitably challenging.
This article seeks to summarise recent advances that have been made in understanding and characterising the epidemiology, immunology, phylogeny, and pathophysiology of PARV4. We have maintained a primary focus on updates made since our previous review of the topic in 2014 2 , although we have also drawn upon some older references where they are fundamental in supporting our current understanding. Crucially, there are still many gaps in the story; therefore, we have also set out to identify and highlight important questions for future study.

New advances in the PARV4 literature Population prevalence of infection in different geographic locations
The most substantial updates to the PARV4 literature are studies to determine the prevalence of infection in geographic areas that had not previously been represented. This includes new studies in Scandinavia 17,18 , the Middle East 19 , and South America 20,21 ; we have summarised these data in Table 2. Below, we discuss these findings with further reference to their applicability to understanding risk factors for infection.

Risk factors for PARV4 infection
Based on the strong relationship between PARV4 and HIV, HBV, and HCV co-infection in North American and European populations, several studies have compared PARV4 exposure (screening for PARV4 IgG or DNA or both) in HIV-positive and HIV-negative groups in different settings. In Iran, HIV infection (with or without HCV co-infection) was reported to be associated with a high prevalence of PARV4 viraemia (35%) 19 . However, in this study, the background prevalence of viraemia in healthy blood donors was also high at 17% 19 . Higher seroprevalence was also reported in Scandinavian subjects in the setting of HIV infection; the PARV4 IgG detection frequency was 8% among HIV-positive children 17 , The evidence of clinical associations for B19V (left-most column) are robust and consistent, in contrast to those for animal hokoviruses and PARV4, in which the evidence is often at the level of case reports or small case series only and for which replication in other cohorts is lacking. Other parvoviruses (for example, dependoparvoviruses and protoparvoviruses) have also been detected from human samples, but evidence is lacking for pathogenicity 59 .  Table 1. The question marks highlight that robust and consistent evidence is still lacking for all of these associations. GI, gastrointestinal; HIV, human immunodeficiency virus.
while the background population seropositivity rate was less than 1% in adults and absent in children 18,22 . Stratification of risk factors in a French study reported the highest PARV4 prevalence in PWIDs, especially among those who were HIV-infected, but also among men who have sex with men (MSM), particularly those likely to have been exposed in the 1980s 12 .
Our own study of a cohort of mothers and children in Kimberley, South Africa, was comparable to previous reports from sub-Saharan Africa in finding an overall PARV4 seroprevalence of 37% with no evidence that HIV infection was a risk factor 23 . Although vertical transmission has previously been demonstrated in another setting 24 , there was no concordance between maternal and child serostatus, suggesting that maternal transmission is probably not a major contributor to the overall burden of infection in South Africa 23 . The clear trend towards increasing acquisition with age is consistent with a broader environmental source of PARV4 infection.
A potential association between PARV4 exposure and infection with another BBV, human T-cell lymphotrophic virus (HTLV), was described in a Brazilian study 21 . Among 113 subjects with HTLV-1 infection, seven (6%) were positive for PARV4 DNA, two of whom also had HIV or HCV infection or both. Based on a history of blood transfusion in one of these individuals, it was postulated that this represents the most likely route of acquisition of all three viruses. However, the remaining five cases had no other co-infecting BBVs and no history of injecting drug use or blood transfusion, leading to the conclusion that other transmission routes of PARV4 transmission may also be operating 21 . Another study published by the same group identified PARV4 viraemia in 2-5% of individuals either donating blood or receiving transfusions of blood products 20 . Neither of these Brazilian studies screened for PARV4 IgG, so the prevalence of PARV4 exposure in this setting remains unknown 20,21 .
Viral biology/immunology PARV4 variants characterised to date fall into three distinct genetic groups (genotypes 1-3) but show relatively restricted sequence variability; in both the non-structural (NS) and capsidencoding genome regions, there is less than 3% amino acid (8-9% nucleotide) sequence divergence between genotypes (representations of diversity can be found in Figure 2 and in Supplementary data 1). The even greater restriction of sequence diversity within genotypes supports the hypothesis for relatively recent transmission to human populations 5 . Relatively few new PARV4 sequences have been reported over the period that we have reviewed, and although incremental additions to the sequence database have been made over recent years (Figure 3), these have not substantially extended the known diversity of this virus.
The distribution of the three genotypes is summarised as follows: • Genotype-1 is the genotype currently predominant in Europe and North America 11,25-27 . Sequencing of PARV4-positive plasma pools and blood products concentrates from various collection years 28,29 as well as analysis of the age distributions of PARV4-infected study subjects 26 suggests that genotype-1 replaced genotype-2 as the predominantly acquired genotype in these regions from the mid-1990s onwards, analogous to the genotype replacement that has been described for another parvovirus, B19V 30 . PARV4 sequences derived from a recent study in Iran also all grouped as genotype-1 19 .
• Genotype-2 (formerly known as PARV5) has been identified in European and North American cohorts 25,27 and appears to be the more common genotype in individuals likely infected in the 1980s or early 1990s 11,26 . It was also the predominant strain reported from Asia 31-33 . Limited amplification of PARV4 genomes from Brazil 21 yielded sequences that were most similar overall to genotype-2 strains previously identified in India 32 .
• Genotype-3 has been consistently identified from African cohorts 13 . We have recently derived full-length PARV4 sequences from an adult and child in South Africa (GenBank accession numbers KU871314 and KU871315), which cluster together with genotype-3 sequences.
Various reports provide preliminary evidence that PARV4 may not be completely cleared after acute infection, suggesting that it has the potential for latency and reactivation. Two studies in Taiwan have sought to describe serological responses to PARV4 infection by undertaking longitudinal follow-up of their cohorts 31,33 . Among a small cohort of health-care workers, IgG seropositivity was strikingly high at 60%; half of these had a  positive IgM that was sustained over the study period, and viraemia was detected at one of several sampling time-points 33 . In the second study, this time focusing on a high-risk PWID cohort, IgM was reported as remaining positive for up to 21 months 31 . Meanwhile, in vitro experiments have demonstrated a sustained high-magnitude ex vivo CD8 + T-cell response to PARV4 NS peptide 34 . Prolonged IgM-positivity and sustained high-magnitude T-cell responses together suggest a sustained or relapsing/remitting exposure to viral antigen. Tissue homing and sites of latency are unknown, but existing data suggest that bone marrow, the respiratory tract, liver and gut might represent potential sites of viral replication 5,15,26,27,35,36 . Skin has also been highlighted as a potential reservoir site for B19V 37 , although-to the best of our knowledge-no study has investigated this as a site of PARV4 replication. The role of capsid proteins in parvovirus infections has been recently reviewed 38 , although there are no specific functional or structural data for PARV4.

Clinical associations of PARV4
To date, determining the clinical impact (if any) of PARV4 infection remains perhaps the most uncertain area of research into this virus, despite the medical importance of establishing such disease associations. A wide range of potential infection outcomes has been proposed (summarised in Table 1), but the evidence for a pathogenic role of PARV4 in any of these has not been significantly expanded in recent years.
The most significant addition to the literature builds on an earlier report linking PARV4 with an encephalitis syndrome in Bellary, South India 32 . Further efforts have been made to substantiate this important potential disease association by screening ten patients in Northern India with an acute encephalitis syndrome in whom other viral causes had been ruled out. The investigators identified PARV4 DNA in the cerebrospinal fluid (CSF) and blood of two 39 . However, it remains uncertain whether PARV was actually the agent of the observed disease, or a bystander that was coincidentally detected in a highly exposed population. Neither this nor the previous study 32 included a control group in which the background incidence of PARV4 infections could have been evaluated. It is additionally possible that detection of PARV4 DNA in CSF in these two studies simply represents reactivation of the virus in the central nervous system in the context of other pathology.
Developing and refining methods for the study of PARV4 One approach to the determination of PARV4 serostatus is to use an enzyme-linked immunosorbent assay (ELISA)-based approach with an optical density read-out that is calibrated as positive or negative. However, the choice of a fixed cut-off above which threshold samples are termed 'positive' can be problematic and evidently has bearing on the final estimation of prevalence; the choice of this threshold evidently affects sensitivity and specificity of the test, and may need to be altered in light of the consideration of the primary reason for screening (diagnosis versus seroepidemiology). This effect is exemplified by recently published simulations, alongside discussion of alternative approaches that set out to quantify population prevalence 40 . One of these is to use 'mixture models', which estimate population seroprevalence from the complete screened dataset rather than classifying each individual sample as positive or negative 40 . Application of these new analytical methods to existing PARV4 serosurvey datasets has yet to be performed but may refine prevalence estimates and allow for better comparison between studies.
A novel polymerase chain reaction (PCR) assay for genotyping and quantifying PARV4 was published in 2014 41 . The authors report that their method reliably discriminates between PARV4 genotypes, and use quantitative PCR to measure viraemia 41 .

Practical implications: human tissue for transfusion and transplantation
Consistently high detection frequencies of anti-PARV4 antibodies in sera of haemophiliac individuals provide strong evidence for some transmission via the use of clotting factor concentrates 11,42 . Although modern procedures for viral inactivation may effectively remove parvovirus infectivity, PARV4 is more robust than B19V to heat and other virus inactivation methods used in blood product manufacture, and there is ongoing discussion about the need to implement donation screening for PARV4 DNA and/or to enhance measures to ensure that the virus is eradicated from blood products for transfusion 10,43-45 .
Blood donors have been screened as a component of several studies (see representative data in Table 1

Insights from new animal data
There is increasing evidence that infections with porcine and bovine hokoviruses are endemic in both wild and domestic animals; most of these recent animal data are simply detection studies and do not provide any new data about pathogenicity. These include the frequent detection of porcine hokovirus in hunted wild boar in Portugal 48 and a bovine hokovirus in domestic yaks 49 . In South America, another hokovirus was found to be endemic in a survey of Brazilian pig herds, in which post-weaning multisystemic wasting syndrome was common 50 , and a novel porcine hokovirus (referred to as porcine parvovirus 6, or PPV6) has been associated with foetal loss in pigs in China 51,52 .
As livestock pathogens, these organisms are potentially implicated as a cause of significant economic losses in agriculture 52,53 . However, since most of the suggested pathogenicity has been attributed to protoparvoviruses or copiparvoviruses, we cannot necessarily infer any significance to tetraparvoviruses. Nevertheless, the widespread isolation of parvoviruses, particularly from domestic animals, should raise concerns about the potential risk of inter-species transmission events leading to the possibility of more pathogenic viruses crossing into human populations.
A recent study of microRNA expression in porcine parvovirus infection describes upregulation of microRNAs that are known to be immune system regulators 52 . In the longer term, this kind of study may provide further insights into the nature and mechanism of host-virus interactions.

Discussion and conclusions
As this review illustrates, our understanding of PARV4 continues to advance gradually, but there are still many important unresolved questions, and the significance of this infection to human health remains uncertain. The striking difference in population epidemiology in developed and developing countries currently remains unexplained, and it is uncertain to what extent differences in virus properties or host characteristics account for the marked differences in population seroprevalence. For the former, it is plausible that genotype-specific amino acid substitutions underpin differences in the route, mechanism, or efficiency of transmission. From the host side, it is possible that differences in human genetics, prevalent co-infections, or environmental, behavioural, or iatrogenic exposure account for differences in seroprevalence and risk factors between continents.
Ongoing diligent scrutiny of carefully characterised clinical cohorts will be required to cement any associations with human disease; this may require larger collaborative efforts than have been undertaken to date. An in vitro system is still needed to help us understand the biology of PARV4 in more detail, including characterisation of tissue tropism, productivity of cellular infection, the influence of co-infecting viruses, and the mechanisms underlying viral suppression by the immune system.
This review highlights a number of compelling arguments in favour of advancing our understanding of PARV4. Firstly, this agent is endemic in some human populations; even if pathology emerges in only a small subset of infected individuals, this may still be important in terms of overall case numbers. Putative associations with significant pathology may be particularly relevant for certain vulnerable groups, including those with other chronic viral infection (HIV, HBV, or HCV), those with solid organ or bone marrow transplantation, and recipients of human blood or tissue transplant 11,35 .
Secondly, we know that viral infections are relevant to the host in more diverse ways than just the effects of primary infection; they may shape immune ontogeny, influence the outcomes of infection with other pathogens 54 , and drive or control oncogenesis. Understanding viral biology also allows us to draw on these agents as a resource for vaccine design, as exemplified by the adenovirus constructs used in vaccine trials for HIV 55,56 and HCV 57 .
Therefore, whether PARV4 turns out to be an innocent bystander or an elusive human pathogen, there are robust reasons for advocating ongoing clinical and research efforts to further our understanding of epidemiology and the current and future potential significance to human health.