Introduction: SARS-CoV-2 is a respiratory virus with ‘direct’ transmission

Several recent studies investigated the evolutionary development of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), suggested that we should not rule out the possibility that it will become more pathogenic. For example, Markov et al. emphasized that SARS-CoV-2 infection frequently led to severe illness or death during the third-week post-infection. Since the contagious period generally spanned from day 2 to day 15, most transmissions (90%) occurred before the average time of death. The authors pointed out that the death of the host occurring after transmission would not reduce the fitness of a viral lineage and concluded that safe infections, mild COVID-19, and a low population mortality and morbidity burden should not be assumed [Reference Markov1]. A separate study by Rochman et al. constructed a maximally-general compartment model to chart the global phase space of human respiratory viruses, predicting that SARS-CoV-2 will likely continue to present a substantial disease burden [Reference Rochman, Wolf and Koonin2]. Roemer et al. highlight the potential for a future virus variant to emerge that is more dangerous than the currently circulating Omicron variant. They voiced concern that new variants might emerge through recombination with Delta strains since Delta generally has a higher severity than other variants [Reference Roemer3]. While increased pathogenicity cannot be ruled out, comparisons with other respiratory viruses suggest that the likely trajectory of SARS-CoV-2 pathogenicity is downwards. This review investigates the biological and human events that could affect the evolution of pathogenicity in SARS-CoV-2, both positively and negatively.

SARS-CoV-2, like other respiratory viruses, typically spreads via aerosols or droplets emitted when an infected individual talks, coughs, or sneezes. However, these particles cannot remain airborne for extended periods [Reference Bushmaker4]. Therefore, for the virus to spread, infected individuals generally need to be near susceptible individuals. In a study from the 1980s, Ewald pointed out that, beyond preserving the lives of their hosts, pathogens such as respiratory viruses can benefit by ensuring that their hosts remain mobile, helping to spread the virus [Reference Ewald5]. Humans and animals often refrain from moving around when sick, and data shows that respiratory viruses are typically much milder than viruses spread by vectors such as biting and bloodsucking insects [Reference Ewald5]. Since over 200 serologically distinct RNA and DNA virus species cause common colds, and almost all are mild (except in children and vulnerable individuals), the consistent mildness and lack of complications from colds, in contrast to many other viral illnesses, is remarkable [Reference Olszewska, Zambon and Openshaw6]. Walther and Ewald reported the infection fatality rates (IFRs) for nine respiratory viral illnesses, calculated as the mean percentage of deaths per infection [Reference Walther and Ewald7]. Leaving aside smallpox for the time being, these IFRs ranged from 0% for rhinovirus to 0.010% for influenza, as shown in Table 1. The IFRs for viruses transmitted through contaminated water, sexual contact, animal bites, or insects are frequently much higher. For instance, the human IFRs of untreated yellow fever, hepatitis B, HIV infection, and rabies vary from roughly 1% to almost 100%. For comparison, the IFR of SARS-CoV-2 in England and Australia prior to September 2022 was around 1.5% [Reference Levin8].

Table 1. The infection fatality rates and survival of 9 respiratory viruses [Reference Walther and Ewald7]

Furthermore, SARS-CoV-2 falls into the category that I refer to here as ‘direct’ respiratory viruses (DRVs). By this I mean its usual replication cycle involves multiplying in the cells that line the upper airways and being transmitted without passing through the lymphatic system or the bloodstream. (This does not imply the virus never reaches the bloodstream, and never attacks other tissues, but that SARS-CoV-2 virions that originate in organs other than the airways are seldom transmitted.) This contrasts with other respiratory viruses, including measles, mumps, rubella, and smallpox, which replicate in (or cause swelling of) the lymphatics and then pass into the bloodstream. The presence of the virus in the bloodstream allows it to reach all the organs of the body, which increases the opportunities for complications, suggesting that such viruses are likely to be more severe than DRVs. This is difficult to confirm with empirical data. Only a few respiratory viruses have been studied in detail, and data on fatality rates is not widely available. All of the respiratory viruses listed in Table 1 apart from rhinovirus are relatively pathogenic and the majority are lymphatic-passaging. Common colds are caused by at least 200 viruses, many of which are not widely studied, because they are mild (and/or rare). For example, 88 human adenoviruses in seven species are known, many of which cause respiratory illness [Reference Dhingra9]. Since most respiratory viruses do not pass through the lymphatics, and those that do are among the most pathogenic, we might tentatively conclude that DRVs are usually less pathogenic than measles, smallpox, mumps, etc.

Passage through the lymphatics might benefit the viruses involved in two ways: firstly, the virus can multiply invisibly before the host experiences any symptoms. For example, the measles virus has a relatively long incubation period (8–14 days), and it usually replicates in the respiratory tract, local lymph nodes, spleen, lungs, and thymus gland before any symptoms occur [Reference Moss and Griffin10]. This may slow up the development of the illness and extend the prime contagious period that lasts until the infected individual becomes too sick to interact with others. Secondly, since the virus suddenly appears all over the lungs and upper airways in great quantities, infected hosts may shed virus particles at a higher rate. The downside for the host (but not usually for the virus) is that such widespread internal infections can be fatal or lead to complications affecting other organs, such as the eyes, ears, and brain in the case of measles, or affecting unborn children in the case of rubella. Since SARS-CoV-2 does not normally pass through the lymphatics or enter the bloodstream before transmission, we might expect its pathogenicity to diminish.

As mentioned above, smallpox, with an IFR of 10%, was uniquely pathogenic among respiratory viruses. Walther and Ewald propose that one contributing factor was the exceptional durability of the virus outside the body; in one instance, it survived in a dried exudate in indirect light for over two years (Table 1) [Reference Walther and Ewald7]. This durability allowed the virus to adopt a sit-and-wait strategy, making its hosts’ mobility less critical. It also implies that smallpox was often spread by other means, not just through droplets and aerosols, differentiating it from typical respiratory viruses. Influenza, which can survive in the external environment for over 10 days, is also unusually enduring, possibly accounting for its higher-than-normal pathogenicity (Table 1). SARS-CoV-2 is principally transmitted by air, and the airborne virus has a half-life of approximately 3 h [Reference Bushmaker4]. It can also be transmitted via solid surfaces, having a half-life of around 5 h for the ancestral lineage WA1. The early variants of concern Alpha and Beta had slightly longer half-lives of around 5–6 h both in air and on surfaces, but the BA.1 Omicron variant was less stable than WA1. Since these periods are lower than those of the respiratory viruses shown in Table 1, we might anticipate selection for high host mobility and low pathogenicity in the future. However, we cannot rule out the emergence of increased transmission on solid surfaces and selection for greater durability if the mode of transmission changes.

Diarrhoea, sometimes bloody, occurs in approximately 10% to 20% of hospitalized patients with COVID-19 infection, and it is thought that the virus can replicate in the intestinal epithelium [Reference Friedel and Cappell11], opening up the possibility of transmission by the faecal-oral route. We should, therefore, consider the pathogenicity of viruses that are spread by this route. Norovirus typically causes non-bloody diarrhoea, vomiting, and stomach pain, sometimes with fever or headaches. It is spread predominantly through the faecal-oral route, either through contaminated food or surfaces, or directly from person to person [Reference Heijne12]. However, it rarely causes complications, and recovery typically takes one to three days. Other viruses spread by the faecal-oral route include several strains of coxsackievirus and enterovirus that cause hand, foot, and mouth disease, mainly in children. Symptoms include skin rashes, showing they reach the bloodstream. However, the disease is normally mild, and most patients recover within 10 days. This mildness may reflect the need for children to be physically close to spread the disease efficiently. Interestingly, the same viruses often cause common colds in adults. Occasionally, hand, foot, and mouth disease from enterovirus 71 causes complications, such as cardiac symptoms, viral meningitis, or encephalitis [Reference Huang13]. Another disease that more often causes complications is polio. Seventy-five per cent of polio infections are asymptomatic, but in a few cases, the virus migrates from the gut to the central nervous system, causing paralysis in 0.1%–1% of cases [14]. Increased transmission of SARS-CoV-2 by the faecal-oral route might be a very unwelcome development.

While observations show that virtually all DRVs are mild in most people, it is not obvious why this should be the case. It is interesting to speculate on the selective trends involved. It might be expected that common cold viruses would often be internalized and cause dangerous or unpleasant complications after the contagious period. This is similar to the argument of Markov et al. above when they point out that death from COVID-19 generally occurs after the contagious period, so they argue there is little natural selection to reduce host mortality. Ewald points out that we should focus on mobility rather than mortality, but the same difficulty remains: why should selection that preserves host mobility (during the contagious period) prevent complications later on? The answer seems to be that viral selection is influenced by two important factors: (1) preschool children are especially susceptible to cold viruses, having five to seven colds each year. About 10% to 15% of children will have more than 11 infections. Adults have two to three colds per year, but the incidence is increased by contact with children [Reference Turner15]. In a study where adults were challenged by administering nasal drops containing four common cold viruses, parents were less likely to develop cold symptoms than nonparents, suggesting that parents have had exposure to a wider range of respiratory viruses and therefore have greater immunity [Reference Sneed16]. This suggests that children can provide a breeding ground for novel virus variants that subsequently spread to adults. (2) Almost all DRVs are more prevalent during the winter, probably because they are thermally sensitive (as discussed below). For example, Du Prel et al. found that the number of hospitalizations of children for respiratory illnesses in Mainz, Germany, was about six times greater in midwinter than midsummer, caused by a variety of DRVs including RSV, rhinovirus, influenza, and adenovirus (Figure 1) [Reference du Prel17]. Since children are normally kept at home if they develop serious symptoms such a fevers, the need for respiratory viruses to spread and survive the winter in children might provide a filter that removes the more pathogenic strains that would otherwise reach the bloodstream and cause systemic infections in adults. Note that even relatively mild cold viruses such as rhinovirus frequently hospitalize young children [Reference du Prel17] (Figure 1). Selection in young children in winter may thus promote the emergence of viruses that rarely cause serious illness in healthy older children and adults throughout the year.

Figure 1. Case numbers of children aged 16 years or below hospitalized by various DRVs in Mainz, Germany, 2000–2008. The four most frequent pathogens were, in order, rhinovirus (RV), respiratory syncytial virus (RSV), influenza virus A (IVA), and adenovirus (AV). Note that rhinovirus, usually considered mild, hospitalized substantial numbers of children. The timing of all illnesses is variable. For example, RSV cases mainly occurred before 1 January in the winter of 2002/2003, but after 1 January from 2005 onwards. In spite of this variation, the peak for all hospitalizations combined was normally around February, and the minimum was around August. Adapted from du Prel JB et al. Clin Infect Dis 2009;49(6):861–83.

Biological factors that may support the evolution of reduced pathogenicity of direct respiratory viruses

Table 2 shows some important biological factors that may support the evolution of reduced pathogenicity.

Table 2. Factors that may decrease the pathogenicity of respiratory viruses

Factors that may increase pathogenicity

Table 3 lists some of the biological factors that might increase the pathogenicity of SARS-CoV-2 and other respiratory viruses. As discussed above, SARS-CoV-2 that replicates in the warmer parts of the body (lungs, lymphatics, etc.) does not normally pass back to the airways and escape as droplets and aerosols. However, replication in many parts of the body might increase viral shedding, and circumstances that favour transmission by this (systemic) route could, in principle, arise in hospitals, refugee camps, or other settings that offer the potential for continual transmission from very sick individuals. Similarly, circumstances encouraging transmission via exudates on solid surfaces could favour more pathogenic strains that increase total virus shedding, even if the shedding period is brief. Additionally, Ewald pointed out that faecal transmission, including through contaminated drinking water, could increase virulence because large numbers of susceptible people could become infected by the pathogens released by a few immobilized hosts [Reference Ewald18]. Another danger is that the recombination of SARS-CoV-2 with animal coronaviruses might result in a strain with increased pathogenicity.

Table 3. Biological and behavioural factors that could increase the pathogenicity of respiratory viruses

We should also consider human activities and events that could increase the pathogenicity of SARS-CoV-2, some of which are listed in Table 4. It might be expected that the introduction of mass testing for viruses for the first time during the COVID-19 pandemic would have had an important impact on the evolution of pathogenicity. PCR testing followed by tracing of the contacts of infected individuals may have been harmful in this respect. Since there was a delay in getting an appointment for a test, and it took one or more days to receive the results, we might expect a strong selection of strains that could replicate and spread so quickly that testing and tracing could not stop them. Such aggressive strains might outrun the track and trace system, while milder strains, which might otherwise have had an advantage, might have been stopped. The effects of the instant lateral flow tests, which arrived later, are more difficult to predict. In previous pandemics, in the absence of testing, individuals infected by strains that were mild and caused ‘colds’ would have continued to socialize and spread the virus for several days. Free lateral flow tests would have prevented this in many cases. On the other hand, strains that sometimes gave completely asymptomatic infections might have been favoured, since lateral flow testing would not normally have detected them.

Table 4. Human activities that could increase the pathogenicity of SARS-CoV-2

Accidental release of pathogens from laboratories is not uncommon, and laboratory alterations to the virus, such as alterations to cleavage sites and receptor binding domains, might result in increased pathogenicity [Reference Segreto and Deigin44]. Repeated serial passage through cells is used to make viral strains for live attenuated vaccines. It tends to reduce the fitness of viruses to be transmitted naturally, reducing the danger of accidental release [Reference Badgett45]. On the other hand, selection during laboratory passaging can be very powerful because scientists often transfer orders of magnitude more viruses to cells than are transmitted during natural infections. Therefore, adaptation in cell cultures can be very rapid, and one or a few passages in cells might give rise to strains with novel, possibly more harmful, features.

Poor sanitation in developing countries might allow a more pathogenic strain specializing in faecal transmission to evolve. However, extensive adaptation to this route would probably prevent significant transmission by the airborne route.

Any social, political, or cultural pressures for individuals to continue working and socializing when sick might increase the virulence of respiratory viruses, including SARS-CoV-2. A crisis such as a war might have this result.

Note that some changes can create ‘tipping points’ where changes to the tropism of SARS-CoV-2 occur that are unlikely to be reversed. For example, strains released from laboratories or arising from poor sanitation might stably pass through the lymphatics or evolve increased durability. By contrast, other factors, such as increased testing in a few countries, or transmission in crowded environments, might increase pathogenicity temporarily and reversibly.

A note on vaccines

The effect of vaccination on the evolution of virus pathogenicity is unclear and may depend on the vaccine’s effectiveness. In the cultural climate that currently exists in many countries, it seems unlikely that high vaccine uptake levels will be achieved soon. Nevertheless, by reducing the number of susceptible individuals in a population, even vaccines with low uptake might support the emergence of mild strains with longer incubation periods that can better preserve host mobility. On the other hand, vaccines that are ineffective or confer antibody-based but not T-cell receptor-based immunity may promote the emergence of novel variants (possessing altered surface proteins) to which the population may be susceptible. Aggressive strains with shorter incubation periods may emerge if susceptible hosts become more abundant due to novel surface proteins.