key: cord-0806852-1s6dlcer
authors: Woolhouse, Mark E.J.; Brierley, Liam; McCaffery, Chris; Lycett, Sam
title: Assessing the Epidemic Potential of RNA and DNA Viruses
date: 2016-12-03
journal: Emerg Infect Dis
DOI: 10.3201/eid2212.160123
sha: 869d3d4d00795e9ca42618b9207f6dd315b19cd3
doc_id: 806852
cord_uid: 1s6dlcer

Many new and emerging RNA and DNA viruses are zoonotic or have zoonotic origins in an animal reservoir that is usually mammalian and sometimes avian. Not all zoonotic viruses are transmissible (directly or by an arthropod vector) between human hosts. Virus genome sequence data provide the best evidence of transmission. Of human transmissible virus, 37 species have so far been restricted to self-limiting outbreaks. These viruses are priorities for surveillance because relatively minor changes in their epidemiologies can potentially lead to major changes in the threat they pose to public health. On the basis of comparisons across all recognized human viruses, we consider the characteristics of these priority viruses and assess the likelihood that they will further emerge in human populations. We also assess the likelihood that a virus that can infect humans but is not capable of transmission (directly or by a vector) between human hosts can acquire that capability.

A series of recent emerging infectious disease outbreaks, including the 2014 Ebola virus disease (EVD) epidemic in West Africa and the continuing Zika virus disease epidemic in the Americas, have underlined the need for better understanding of which kinds of pathogens are most likely to emerge and cause disease in human populations. Many, although not all, emerging infectious diseases are caused by viruses, and these frequently emerge from nonhuman host reservoirs (1) (2) (3) . The enormous diversity (4) and high rates of evolution (5) of viral pathogens discourage attempts to predict with any precision which ones are most likely to emerge in humans. However, there is some consensus, at least in general terms, regarding the kinds of traits that are most essential in determining the capacity of a virus to infect, cause disease, and spread within human populations (Table 1) . We focus on one of these traits, the capacity of a virus to spread from one human to another (by any transmission route other than deliberate laboratory exposure), a key determinant of the epidemic potential of a virus.

A theoretical framework for studying the dynamics of infectious disease outbreaks is well established (6) . The capacity of an infectious disease to spread in a host population can be quantified in terms of its basic reproduction number, R 0 . R 0 is defined as the average number of secondary cases generated by a single primary case in a large, previously unexposed host population, and its value tells us a great deal about the epidemiology of a pathogen. R 0 = 0 indicates no spread in that population; this value would apply to zoonotic infections that do not spread between humans. R 0 in the range 0<R 0 <1 indicates that chains of transmission are possible but that outbreaks will ultimately be self-limiting. R 0 >1 indicates that major epidemics can occur or that the disease may become endemic in that host population. A higher value of R 0 also indicates that a greater reduction in transmission rates must be achieved to control an epidemic (6) . R 0 values have been estimated for >60 common human pathogens (7) , including human influenza A virus (R 0 <2), measles virus (R 0 <18), and dengue virus (R 0 <22). R 0 is determined by a combination of pathogen traits, such as its transmission biology, which is itself a complex interplay between the within-host dynamics of the pathogen and the host response to infection, and host traits, such as demography, behavior, genetics, and adaptive immunity. Consequently, for any given infectious disease, R 0 can vary between host species and between host populations. Infectious diseases with R 0 close to 1 are a particular concern because small changes in their epidemiologies can lead to major changes in the threat they pose to public health (8) .

R 0 is closely related to another conceptual approach to disease emergence, the pathogen pyramid. There are different versions of this scheme (3, 9) . We consider a pyramid of 4 levels ( Figure 1 ). Level 1 represents the background chatter of pathogens to which humans are continually or sporadically exposed but most of which are not capable of causing infection. Other levels can be considered in terms of the R 0 of the pathogen in humans: level 2 corresponds to R 0 = 0, level 3 to 0<R 0 <1, and level 4 to R 0 >1.

We updated our previous systematic literature review (10) of the capacity of virus species to transmit between humans (i.e., level 3 and level 4 viruses; online Technical Appendix, http://wwwnc.cdc.gov/EID/article/22/12/16-0123-Techapp1.pdf). Such viruses are found in 25 of 29 families containing viruses that infect mammals or birds (discounting 2 reports of family Nodaviridae species in mammals/birds). The 4 exceptions comprise 2 families that have no known human-infective viruses (Arteriviridae and Birnaviridae) and 2 with species that have been reported in humans but only at level 2 (Asfarviridae and Bornaviridae).

A total of 22 of these families contain level 4 viruses with epidemic potential in humans (sometimes described as human-adapted viruses) (11) . This finding indicates that this capability is widely distributed among virus taxa. The 3 families with level 3 viruses but no level 4 viruses are the Arenaviridae, Bunyaviridae, and Rhabdoviridae. A list of 37 presumptive level 3 virus species is provided in Table 2 . These species cover a wide taxonomic range and a variety of transmission routes, including vectorborne. Several level 3 viruses have historically been associated with sizeable outbreaks (>100 cases) in human populations: Bwamba, Oropouche, Lake Victoria Marburg, Sudan Ebola, and o'nyong-nyong viruses. For some other viruses, including Guanarito, Junin, lymphocytic choriomeningitis, and Sabia (all arenaviruses); simian virus 40; Titi monkey virus; and influenza A(H5N1) virus, humanto-human transmission is rare or merely suspected. In addition, several viruses are only known or believed to transmit between humans by iatrogenic routes or vertical transmission; this group (Table 2) might be regarded as unlikely epidemic threats.

When a virus is transmitted by a vector, it can be particularly difficult to confirm or exclude the infectiousness of human cases, as with Semliki forest, Barmah forest, and Rift Valley fever viruses. Similarly, even when humanvector-human transmission is believed to occur, it is often difficult to quantify its contribution to a given outbreak, as with Venezuelan equine encephalitis virus.

Level 2 viruses are those that can infect humans (>100 species) but have never been reported to be transmitted by humans (10) . In at least some instances, such as influenza A(H5N1) virus (12) , this finding is attributable to tissue tropisms during human infection that are incompatible with onward transmission.

Shifts in pyramid level equate to shifts in the public health threat posed by a virus. We consider possible shifts in the following sections.

Virus species of mammalian and, more rarely, avian origin are sometimes observed to be transmissible between humans when first found in humans, which constitutes a jump from level 1 straight to level 3 or 4 ( We still have incomplete knowledge of the diversity of viruses that infect mammals and birds; the few hundred recognized species (4) surely represent only a small fraction of the total (3). Moreover, we have few predictors of potential human-to-human transmissibility. One possible indicator is emergence from nonhuman primates, with suggestions that primate viruses are more likely to be able to, or to acquire the ability to, spread in human populations (13, 14) . However, emergence of human transmissible viruses from bat (e.g., severe acute respiratory syndrome coronavirus) or bird (e.g., influenza) reservoirs indicates that this trait is associated with a wide range of reservoirs.

The possibility that level 2 viruses might acquire the capacity to be transmitted between humans (i.e., move into level 3 or 4) is a major concern, especially in the context of influenza A(H5N1) virus and other avian influenza virus subtypes. However, there are few examples of this transition throughout the entire recorded history of human viruses going back to 1901. One possible example involves the simian immunodeficiency virus (SIV) and HIV. A SIV smmderived laboratory strain of SIV has been reported to infect humans, but without onward transmission (15) . SIV smm is related to HIV-2. SIV cpz , which is related to HIV-1, has not been directly observed in humans. However, different HIV-1 lineages, independently derived from SIV cpz , are variably transmissible in humans, and the pandemic HIV-1 M lineage was the only virus to overcome a key host restriction factor (human tetherin) (16) . The only other examples of viruses newly transmitted between humans relate to rare instances of iatrogenic transmission (e.g., Colorado tick fever or rabies viruses). Epidemiologic and phylogenetic considerations routinely inform our assessment of the likelihood of humanto-human transmission being observed in the future. For example, there is markedly less concern about rabies virus than about avian influenza virus, and we suggest 2 reasons for this observation. First, rabies virus has a much longer history of and a much higher incidence of human infection, but human-to-human transmission is extremely rare. Second, there is no evidence that other rhabdoviruses viruses (with the possible exception of Bas-Congo virus, which represents a novel genus) are transmissible in humans (or primates more generally).

Level 3 viruses can also become level 4 viruses. We note that virus evolution is not (necessarily) required for R 0 to become >1 in human populations. Differences in host (or vector) behavior, ecology, or demography might be sufficient (8) .

Instances of shifts from level 3 to level 4 in recent times have been infrequent. Three candidates are Ebolavirus, Zika virus, and chikungunya virus. However, although these viruses have caused epidemics of unprecedented size in humans populations in the past decade, the condition R 0 >1 in human populations might had been previously met for all 3 viruses (17) (18) (19) .

For Ebola virus, the epidemic in West Africa in 2014 constituted the first appearance of this virus in high-density, urban populations, which is expected to correspond to a higher value of R 0 . The chikungunya virus epidemic in the Indian Ocean region in 2005 was associated with a vector species jump (from Aedes aegypti to Ae. albopictus mosquitoes) that has been linked to a mutation in the virus envelope 1 protein gene (18) . The chikungunya virus epidemic in the Caribbean region in 2013 followed the first appearance of chikungunya in the Americas and infected populations that had no history of exposure to the virus. The current Zika virus epidemic in South America appears to be another example of a transition from a level 3 to a level 4 arbovirus associated with geographic spread into areas with high densities of vectors (19) . Occasional Zika virus transmission directly from infected humans to other humans are of considerable interest, but probably contribute little to R 0 . Chikungunya, Zika and the other level 4 arboviruses (yellow fever and dengue viruses) illustrate that, for arboviruses, a high potential for spread in human populations is linked to carriage by anthropophilic vector species, particularly mosquitoes of the genus Aedes. In contrast, no tick species are regarded as anthropophilic, and there are no level 4 and few level 3 tickborne arboviruses.

The preceding sections illustrate that identifying transitions of viruses between level 2 and level 3 or between level 3 and level 4 is not always straightforward. Standard epidemiologic theory can help clarify our expectations.

As we discussed, pyramid level is related to the basic reproduction number R 0 in human populations. In turn, the value of R 0 is indicative of expected outbreak dynamics. Some key results (Figure 2) are the probability that a single primary case will generate >1 secondary cases (for any value of R 0 ), the expected average size of an outbreak generated (over the range 0<R 0 <1), and the probability that an epidemic will spread in the human population (for R 0 >1). These results strictly apply to homogeneous infections in a homogeneous host population, although more general frameworks can accommodate host or pathogen heterogeneity (20) (21) (22) . Nonetheless, the key predictions that secondary cases do not always occur even if R 0 >0 and that major epidemics do not always occur even if R 0 >1 are robust.

From an epidemiologic perspective, our confidence that a putatively level 2 virus is truly incapable of humanto-human transmission is thus a function of the number of index cases observed. The transition between level 3 and level 4 can be studied in terms of the expected distribution of outbreak sizes (23) . In the range 0<R 0 <1, an overdispersed distribution of outbreak sizes is expected: most outbreaks are small (often just single cases) with a long tail of larger outbreaks. This pattern has been reported for a range of emerging viral diseases ( Figure 3 ). As the critical threshold R 0 = 1 is approached, this value is signaled in the outbreak size distribution (Figure 3 ). This framework has been used successfully to monitor the epidemiology of measles virus in the United Kingdom after a decrease in childhood vaccination rates in the late 1990s and indicated the approach to the critical threshold that corresponded to loss of herd immunity (23) .

Outbreak size distribution analysis has been applied to human case data for Andes virus (24), monkeypox virus (20) , and MERS-CoV (25) (Figure 3 ). For EVD up to 2013, data are clearly inconsistent with theoretical expectation for R 0 <1 (Figure 3 ), which suggests that large numbers of small outbreaks have remained undetected or that R 0 was already >1 in at least some settings. Either way, R 0 ≈1 for EVD in humans implies that small differences in the biology or epidemiology of the virus would lead to large changes in scale of outbreaks (8) , which could make events such as the EVD epidemic in 2014, if not predictable, then much less unexpected.

Changes in pyramid level might be mediated by virus evolution or changes in virus ecology (28) . A major issue is whether the capacity of a virus to spread in human populations arises as a result of adaptation (evolution of transmissibility that occurs during human infection) or preadaptation (genetic variation within nonhuman reservoirs that predisposes a virus not only to infect humans but also transmit between humans, noting that RNA viruses often show high levels of genetic variation such that they are sometimes described as quasi-species [29] ). These alternatives have been characterized as tailor-made and off-the-shelf, respectively (28) . The first alternative implies a progression from no or low transmissibility between humans to moderate or high transmissibility. The second alternative implies moderate or high transmissibility at first infection of humans.

We consider that our survey of documented changes of pyramid level is most consistent with the off-the-shelf model of virus emergence. In particular, we can find no convincing examples of level 2 viruses becoming level 3 or 4 viruses, which suggests that, if this happens at all, it typically happens sufficiently rapidly (i.e., requires a sufficiently small number of introductions) that we fail to observe the level 2 phase. In contrast, we regularly observe viruses at levels 3 or 4 the first time they are detected in human populations. Nonetheless, the possibility of virus evolution of transmissibility in a new host has been demonstrated experimentally for influenza A(H5N1) virus in ferrets (30) . A theoretical study (31) suggested that the fact that this virus subtype has been circulating widely in poultry populations, with frequent human exposure and sporadic human infection for almost 20 years, provides little or no reassurance about its future evolutionary trajectory.

HIV lineages show clear evidence of adaptation to humans (16) , but as discussed earlier, it is not clear whether the SIV lineages that gave rise to HIV-1 or HIV-2 were capable of transmission between human hosts. We speculate that extended infection times make tailor-made emergence more likely for retroviruses.

Demonstrating that an infected human has the potential to transmit the infection to another human is not always straightforward. High virus titers in body secretions and excretions, blood, or skin are considered indicative. Case clusters are suggestive, but if persons occupy the same environment (e.g., household), then it might be difficult to rule out common exposure. Case clusters must be epidemiologically plausible (i.e., delimited in space and time in a manner consistent with the known or assumed epidemiology of the virus). Genotyping techniques are useful tools for confirming a cluster but do not resolve the source of infection.

For several of the viruses we studied (e.g., Bas-Congo, Lujo, Nelson Bay, and severe fever with thrombocytopenia syndrome viruses) (Table 2) , the evidence for human-to-human transmission is best regarded as tentative, particularly where putative clusters were small. Such assessments can be even more difficult for vectorborne viruses. In many situations, the best evidence for the human-to-human transmission will come from analysis of virus genome sequences.

One approach to resolving the question of human-to-human transmission is analysis of nucleotide sequence data, sometimes referred to as forensic phylogenetics. Nucleotide substitution rates in fast-evolving RNA viruses, such as MERS-CoV and Ebola virus, are ≈1-5 × 10 -3 /site/year (32, 33) , making it possible to use sequences isolated from different hosts at different times to estimate time-resolved phylogenetic trees. Estimates of the transmission chain from temporal sequence data can be improved by incorporating additional information on the date of onset of individual cases, duration of latent and infectious periods, and overall prevalence (34).

We provide some example phylogenetic trees generated from simulated epidemics (Figure 4 ). In an epidemic in an animal reservoir with occasional transmission to humans ( Figure 4 , panel A), for each human sequence, the most closely related next sequence is of animal origin. Clusters of closely related human sequences are shown, and the distribution of the expected cluster sizes is a function of R 0 (Figure 4 , panels B, C) (35) .

In an outbreak, it might be difficult to find and sample the putative source animal cases. However, estimating the time to most recent common ancestor (TMRCA) of the human cases will indicate how long the infection has been spreading. For sporadic zoonoses (Figure 4, panel A) , most transmission has occurred unobserved in the animal reservoir, and the TMRCA of pairs of human cases will be long because these sequences are not closely related. For outbreaks involving human-to-human transmission (Figure 4, panels B, C) , the TMRCA of the cluster of human cases will be closer to the date of the first human infection (whether sampled or not) and provides the estimated date of the zoonotic event.

Use of sequence data to distinguish between multiple instances of human infection from a common animal source and human-to-human transmission in the early stages of an outbreak is extremely challenging because of short timescales, and involvement of few mutations. However, genetic differences and phylogenetic evidence show that at least 2 of the first 3 reported cases of influenza A (H7N9) virus infection in humans were believed to originate from distinct domestic avian sources (36) . Further sequencing of avian samples implied that a low-pathogenicity influenza A(H7N9) virus strain had been spreading in domestic birds for ≈1 year before sporadic cases were detected in humans (37) . Similarly, detection of genetically distant lineages of MERS-CoV, which persisted for only a few months each, suggest multiple introductions from an animal reservoir and only limited human-to-human transmission to date (32) . In contrast, the influenza A(H1N1) pandemic in 2009 and the EVD epidemic in West Africa in 2014 were believed to be the results of single zoonotic events, followed by sustained human-to-human transmission (33) , as shown by a single rapidly expanding lineage.

Our survey of the capacity of RNA and DNA virus infections to be transmitted, directly or indirectly, between humans leads to several conclusions and practical suggestions for improving surveillance of emerging infectious diseases and targeting efforts to identify future public health threats. In support of these conclusions, the World Health Organization recently published list of priority emerging infectious diseases and corresponding viruses (38) included 6 of the viruses in Table 2 .

A major observation is that the taxonomic diversity of viruses that are possible threats to public health is wide, but bounded. Most human infective viruses are closely related to viruses of other mammals and some to viruses of birds. There are no indications that humans acquire new viruses from any other source. However, diversification within human populations occurs and is a prominent feature of some DNA virus taxa (e.g., family Papillomaviridae) (4).

In general, however, our knowledge of origins of human viruses is still incomplete. Although the origins of HIV-1 have been extensively investigated (16) , for most other viruses, even level 4 viruses, little or no research has occurred. An origins initiative (9) most human transmissible viruses that emerge already have that capability at first human infection or acquire it relatively rapidly. If transmission from humans would require a change in a phylogenetically conserved trait, such as tissue tropism or transmission route (4), then such viral paradigm shifts will probably be extremely rare (39) . Even when a virus is capable of transmission between humans, the critical threshold R 0 >1 is not always achieved. However, because changes in virus traits or host population characteristics can influence R 0 , level 3 viruses (Table 2) are of special interest from a public health perspective, and of special concern when, like MERS-CoV, they also cause severe illness. Demonstrating human transmissibility is often difficult, but essential. The best evidence is likely to come from virus genome sequencing studies. These studies should be a public health priority (40) .

We currently have few clues to help us predict which mammalian or avian viruses might pose a threat to humans and, especially, which might be transmissible between humans. One argument in favor of experimental studies of these traits, including controversial gain of function experiments (30) , is that they could help guide molecular surveillance for high-risk virus lineages in nonhuman reservoirs.

The first line of defense against emerging viruses is effective surveillance (40) . A better understanding of which kinds of viruses in which circumstances pose the greatest risk to human health would enable evidence-based targeting of surveillance efforts, which would reduce costs and increase probable effectiveness of this endeavor.

Global trends in emerging infectious diseases

Prediction and prevention of the next pandemic zoonosis

Virus taxonomy: ninth report of the International Committee on Taxonomy of Viruses

Emerging pathogens: the epidemiology and evolution of species jumps

Infectious diseases of humans: dynamics and control

Global mapping of infectious disease

Population biology of emerging and re-emerging pathogens

Origins of major human infectious diseases

RNA viruses: a case study of the biology of emerging infectious diseases

The diversity of human RNA viruses

Host species barriers to influenza virus infections

Phylogeny and geography predict pathogen community similarity in wild primates and humans

Identifying future zoonotic disease threats

Brief report: infection of a laboratory worker with simian immunodeficiency virus

Origins of HIV and the AIDS pandemic

The basic reproductive number of Ebola and the effects of public health measures: the cases of Congo and Uganda

Biology and pathogenesis of chikungunya virus

Zika virus in the Americas: yet another arbovirus threat

Inference of R (0) and transmission heterogeneity from the size distribution of stuttering chains

Aggregation and distribution of strains in microparasites

How do pathogen evolution and host heterogeneity interact in disease emergence?

Measles outbreaks in a population with declining vaccine uptake

Ecological origins of novel human pathogens

Interhuman transmissibility of Middle East respiratory syndrome coronavirus: estimation of pandemic risk

The transmission potential of monkeypox virus in human populations

Outbreaks chronology: Ebola virus disease

Emergence of new infectious diseases

The quasispecies (extremely heterogeneous) nature of viral RNA genome populations: biological relevance: a review

Airborne transmission of influenza A/H5N1 virus between ferrets

Evolution and emergence of novel human infections

Spread, circulation, and evolution of the Middle East respiratory syndrome coronavirus

Genomic surveillance elucidates Ebola virus origin and transmission during the 2014 outbreak

Bayesian reconstruction of disease outbreaks by combining epidemiologic and genomic data

Phylodynamics of infectious disease epidemics

Human infection with a novel avian-origin influenza A (H7N9) virus

The genesis and source of the H7N9 influenza viruses causing human infections in China

World Health Organization. Blueprint for R&D preparedness and response to public health emergencies due to highly infectious pathogens

Pacing a small cage: mutation and RNA viruses

Lessons from Ebola: improving infectious disease surveillance to inform outbreak management