Molecular detection of MERS-CoV RNA in real time

The first open reading frames (ORF 1a and 1b; Fig. 2) have become a key diagnostic and taxonomic target for CoV species identification. With less than 80 % identity between the amino acid sequence of MERS ORF 1ab and betacoronavirus relatives, Tylonycteris bat HKU4 and Pipistrellus bat HKU5, it can be concluded that it is a novel and distinct virus. MERS-CoV is predicted to encode ten open reading frames with 5’ and 3’ untranslated regions [51]. The structural proteins include the spike (S), envelope (E), membrane (M) and nucleocapsid (N) [52]. The products of ORF1a and ORF1b are predicted to encode nonstructural proteins.

The majority of specimen testing to date has employed validated RT-rtPCR assays shown to be sensitive and specific [47, 48, 53]. The RealStar® kit uses these WHO-recommended assays [54]. The target sequences of these screening assays have not changed among genomes examined until at least mid-2015 (IMM observation). Other RT-rtPCR assays have been developed and validated for use as laboratory-based diagnostic tools [55–57]. Additionally, loop-mediated [58, 59] or recombinase polymerase [60] isothermal assays have been designed for field deployment.

MERS-CoV antigen detection

The detection of MERS-CoV antigen has not been common to date but the combination of short turnaround time from test to result, high throughput and identification of viral proteins makes this an attractive option. Detection of viral proteins rather than viral RNA indicates the likely presence of infectious virus. The first rapid immunochromatographic tool described could detect recombinant MERS-CoV nucleocapsid protein from DC nasal swabs with 94 % sensitivity and 100 % specificity compared to RT-rtPCR [61]. A different approach used a monoclonal antibody-based capture ELISA targeting the MERS-CoV nucleocapsid protein with a sensitivity of 10³ TCID₅₀ and 100 % specificity [62].

Assays to identify a humoral response to prior MERS-CoV infection among humans

Demonstration of a seroconversion to a MERS-CoV infection meets the current WHO definition of a case so optimized and thoroughly validated sero-assays employed alongside good clinical histories are useful to both identify prior MERS-CoV infection and help support transmission studies. Because serology testing is, by its nature, retrospective, it is usual to detect a viral footprint, in the form of antibodies, in the absence of any signs or symptoms of disease and often in the absence of any viral RNA [63].

Strategic, widespread sero-surveys of humans using samples collected after 2012 are infrequent. Much of the Arabian Peninsula and all of the Horn of Africa lack baseline data describing the proportion of the community who may have been infected by a MERS-CoV. However, sero-surveys have had widespread use in elucidating the role of DCs as a transmission source for MERS-CoV. Because of the identity shared between DC and human MERS-CoV (see Molecular epidemiology: using genomes to understand outbreaks), serological assays for DC sero-surveys should be transferrable to human screening with minimal re-configuration. Also, no diagnostically relevant variation in neutralization activity have been found from among a range of circulating tested MERS-CoV isolates and sera, so whole virus or specific protein-based sero-assays should perform equivalently in detecting serological responses to the single MERS-CoV serotype [49]. The development of robust serological assays requires reliable panels of well-characterized animal or human sera, including those positive for antibodies specific to MERS-CoV, as well as to likely sources of cross-reaction [64]. Obtaining these materials was problematic and slowed the development and commercialization of antibody detection assays for human testing [64]. A number of commercial ELISA kits, immunofluorescent assays (IFA) kits, recombinant proteins and monoclonal antibodies have been released [31, 65–68]. Initially, conventional IFAs were used for human sero-surveys. These relied on MERS-CoV-infected cell culture as an antigen source, detecting the presence of human anti-MERS-CoV IgG, IgM or neutralizing antibodies in human samples [18, 48, 69]. No sign of MERS-CoV antibodies was found among 2,400 sera from patients visiting Hospital in Jeddah, from 2010 through 2012, prior to the description of MERS-CoV [18]. Nor did IFA methods detect any sign of prior MERS-CoV infection among a small sample of 130 healthy blood donors from another Hospital in Jeddah (collected between Jan and Dec 2012) [70]. Of 226 slaughterhouse workers, only eight (3.5 %) were positive by IFA, and those sera could not be confirmed by virus neutralization (NT) test. The study indicated that HCoV-HKU1 was a likely source of cross-reactive antigen in the whole virus IFA [70]. Whole virus MERS-CoV IFA also suffered from some cross-reactivity with convalescent SARS patient sera and this could not be resolved by an NT test which was also cross-reactive [71]. IFA using recombinant proteins instead of whole-virus IFA, has been shown to be a more specific tool [31]. Since asymptomatic zoonoses have been posited [72], an absence of antibodies to MERS-CoV among some humans who have regular and close contact with camels may reflect the rarity of actively infected animals at butcheries, a limited transmission risk associated with slaughtering DCs [70], a pre-existing cross-protective immune status or some other factor(s) resulting in a low risk of disease and concurrent seroconversion developing after exposure in this group. IFA using recombinant proteins instead.

Some sero-assays have bypassed the risks of working with infectious virus by creating transfected cells expressing recombinant portions of the MERS-CoV nucleocapsid and spike proteins [48, 73], or using a recombinant lentivirus expressing MERS-CoV spike protein and luciferase [74, 75]. A pseudo particle neutralization (ppNT) assay has seen widespread used in animal studies and was at least as sensitive as the traditional microneutralization (MNT) test. [10, 74, 76–78] Studies using small sample numbers and ppNT found no evidence of MERS-CoV neutralizing antibody in sera from 158 children with LRT infections between May 2010 and May 2011, 110 sera from 19 to 52 year old male blood donors and 300 self-identified animal workers from the Jazan Region of the KSA during 2012 [79, 80]. Similarly, a study of four herdsmen in contact with an infected DC herd in Al-Ahsa, eight people who had intermittent contact with the herd, 30 veterinary surgeons and support staff who were not exposed to the herd, three unprotected abattoir workers in Al-Ahsa and 146 controls who were not exposed to DCs in any professional role, found none with serological evidence of past MERS-CoV infection using the ppNT assay [10]. A delay in the neutralizing antibody response to MERS-CoV infection was associated with increased disease severity in South Korea cases with most responses detectable by week three of illness while others, even though disease was severe, did not respond for four or more weeks [81]. The implications for our ability to detect any response in mild or asymptomatic cases was not explored but may be a signifcant factor in understanding exposure in the wider community.

A Jordanian outbreak of acute LRT disease in a hospital in 2012 was retrospectively found to be associated with MERS-CoV infection, initially using RT-rtPCR, but subsequently, and on a larger scale, through positivity by ELISA and IFA or MNT test. [46, 82, 83] This outbreak predated the first case of MERS in the KSA. The ELISA used a recombinant nucleocapsid protein from the group 2 betacoronavirus bat-CoV HKU5 to identify antibodies against the equivalent cross-reactive MERS-CoV protein [71]. It was validated using 545 sera collected from people with prior HCoV-OC43, HCoV-229E, SARS-CoV, HCoV-NL63, HRV, HMPV or influenza A(H1N1) infections but was reportedly less specific than the recombinant IFA discussed above. It was still considered an applicable tool for screening large sample numbers [82]. A protein microarray expressing the S1 protein subunit has also been validated and widely used for DC testing [5, 84]. Detection of MERS-CoV infection using ELISA or S1 subunit protein microarray [84] is usually followed by confirmatory IFA and/or a plaque-reduction neutralization (PRNT) [69, 70, 85] or MNT test. [74, 85, 86] This confirmatory process aims toensure the antibodies detected are able to specifically neutralize the intended virus and are not more broadly reactive to other coronaviruses found in DCs (bovine CoV, BCoV) or humans (HCoV-OC43, HCoV-229E, HCoV-NL63, HCoV-HKU1, SARS-CoV). In the largest study of human sera, a tiered diagnostic process assigned both recombinant IFA and recombinant ELISA positive sera to ‘stage 1’ seropositivity. A stage 2 seropositive result additionally required a suitably titred PRNT result [87]. The study found 15 sera collected in 2012 to 2013 from 10,009 (0.2 %) people in 13 KSA provinces contained MERS-CoV antibodies, but significantly higher proportions in occurred in camel shepherds (two of 87; 2.3 %) and slaughterhouse workers (five of 140; 3.6 %) [87]. Contemporary surveys are needed.

MERS-CoV does not appear to be easily transmitted from DCs to humans, or perhaps it is [72], but generally does not trigger a detectable immune response if only mild disease or asymptomatic infection results. Serology assays are in need of further validation in this area so care is required when moving newly developed diagnostic serology algorithms from a research setting to one that informs public health decisions. This was reinforced when a false positive US case, purported to have been infected after a handshake and two face-to-face meetings, did not withstand further confirmatory analysis using a more specific, NT assay and was subsequently retracted [88, 89].

Specimen types for RT-PCR and length of viral shedding

The WHO recommends sampling from the LRT for MERS-CoV RT-rtPCR testing, especially when sample collection is delayed by a week or more after onset of symptoms. [53] LRT samples are also best for attempting isolation of infectious virus, although the success of culture is reduced when disease persists [49]. Recommended sample types include bronchoalveolar lavage (BAL), tracheal/tracheobronchial aspirate, pleural fluid and sputum [53, 90]. Fresh samples yield better diagnostic results than refrigerated material [69] and if delays in testing of ≥72 h are likely, samples (except for blood) should be frozen at −70 °C [90]. If available, lung biopsy or autopsy tissues can also be tested [53]. The URT is a less invasive and more convenient sampling site however, and an oropharyngeal and throat swab or a nasopharyngeal aspirate/wash are recommended when URT sampling is to be conducted [90]. Paired sera, collected two to three weeks apart are preferable for serological testing while a single sample is suggested to be sufficient if collected two weeks after onset of disease or a single serum collected during the first 10–12 days if conducting RT-rtPCR [53, 90]. Human urine and stool have been found to contain MERS-CoV RNA 12 to 26 days after symptom onset [25, 69, 91] and are listed as samples that should be considered [53, 90]. In two cases that arrived in the Netherlands, urine was RT-rtPCR negative but faeces was weakly positive and sera were RT-rtPCR positive for five days or more [25]. The finding of MERS-CoV viral RNA in serum provides an avenue for retrospective PCR-based studies if respiratory samples are unavailable [83]. RNAaemia may also correlate with disease severity; signs of virus were cleared from the serum of a recovered patient, yet lingered until the death of another [92].

Clinically suspected MERS cases may return negative results by RT-rtPCR. Data have shown one or more negative URT samples may be contradicted by further URT sampling or the use of LRT samples, which is preferred [2, 43, 93]. Higher viral loads occur in the LRT compared to the URT. [22, 69, 88, 94] This fits with the observation that the majority of disease symptoms are reported to manifest as systemic and LRT disease [21]. However, on occasion, even LRT specimens from MERS cases may initially be negative, only to later become positive by RT-PCR [95]. This may be due to poor sampling when a cough is absent or non-productive or because the viral load is low [95]. Despite this both the largest human MERS-CoV studies [32, 96–98] and smaller ones [22, 25, 99], use samples from the URT. It is then noteworthy that one study reported an association between higher loads in the URT and worse clinical outcome including intensive care and death [94]. At writing, no human data exist to define whether the virus replicates solely or preferentially in the LRT or URT, or replicates in other human tissues in vivo although MERS-CoV RNA has been detected from both the URT and LRT in a macaque monkey model [100].The distribution of DPP4 in the human upper airways is also not well described.

Individual human case studies report long periods of viral shedding, sometimes intermittently and not necessarily linked to the presence of disease symptoms. [25, 69, 99, 101] In one instance, a HCW shed viral RNA for 42 days in the absence of disease [99]. It is an area of high priority to better understand whether such cases are able to infect others. Over three quarters of MERS cases shed viral RNA in their LRT specimens (tracheal aspirates and sputum) for at least 30 days, while only 30 % of contacts were still shedding RNA in their URT specimens [91, 102].

In the only study to examine the effect of sample type on molecular analysis, 64 nasopharyngeal aspirates (NPA; an URT sample), 30 tracheal aspirates, 13 sputa and three BAL were examined. The tracheal aspirates and BAL returned the highest viral load values followed by NPA and sputum. Unsurprisingly, higher viral loads generally paralleled whole genome sequencing and culture success and, in NPA testing, were significantly correlated with severe disease and death [49, 94, 103]. This study demonstrated the importance of LRT sampling for whole genome sequencing.

Virus survival in the environment

Infectious MERS-CoV added to DC, goat or cow milk and stored at 4 °C could be recovered at least 72 h later and, if stored at 22 °C, recovery was possible for up to 48 h [150]. MERS-CoV titre decreased somewhat when recovered from milk at 22 °C but pasteurization completely ablated MERS-CoV infectivity [150]. In a subsequent study, MERS-CoV RNA was identified in the milk, nasal secretion and faeces of DCs from Qatar [151].

A single study has examined the ability of MERS-CoV to survive in the environment [150]. Plastic or steel surfaces were inoculated with 10⁶ TCID₅₀ of MERS-CoV at different temperature and relative humidity (RH) and virus recovery was attempted in cell culture. At high ambient temperature (30 °C) and low RH (30 %) MERS-CoV remained viable for 24 h [150]. By comparison, a well known and efficently transmitted respiratory virus, influenza A virus, could not be recovered in culture beyond four hours under any conditions [150]. Aerosol experiments found MERS-CoV viability only decreased 7 % at low RH at 20 °C. In comparison, influenza A virus decreased by 95 % [150]. MERS-CoV survival is inferior to that previously demonstrated for SARS-CoV [152]. For context, pathogenic bacteria can remain viable and airborne for 45 min in a coughed aerosol and can spread 4 m. MERS-CoV’s ability to remain viable over long time periods gives it the capacity to thoroughly contaminate a room’s surfaces when occupied by an infected and symptomatic patient [153]. Whether MERS-CoV can remain adrift and infectious for extended periods (truly airborne) remains unknown. Such findings expand our understanding of the possibilities for droplets to transmit respiratory viruses in many settings, including hospital waiting rooms, emergency departments, treatment rooms, open intensive care facilities and private patient rooms. The nature and quality of air exchange, circulation and filtration are important variables in risk measurement and reduction as is the use of negative pressure rooms to contain known cases. Droplet spread between humans is considered the mechanism of human-to-human transmission and the need for droplet precautions was emphasized after the Al-Ahsa hospital, the KSA and the South Korean outbreaks [21, 23, 154, 155]. By extrapolation, aerosol-generating events involving DCs (urination, defecation, and preparation and consumption of DC products) should be factored into risk measurement and reduction efforts and messaged using appropriate context. The provision of evidence supporting the best formulation of personal protective equipment to be worn by HCWs who receive, manage or conduct procedures on infectious cases remains a priority.

Transmission of MERS-CoV among humans

MERS-CoV was found and characterized because of its apparent association with severe, and therefore more obvious, illness in humans; we were the canaries in the coal mine. Sero-assays and prospective cohort studies have yet to determine the extent to which milder or asymptomatic cases contribute to MERS-CoV transmission chains. However, transmission of MERS-CoV is defined as sporadic (not sustained), intra-familial, often healthcare associated, inefficient and requiring close and prolonged contact [22, 31, 63, 93, 97, 102, 156] In a household study, 14 of 280 (5 %) contacts of 26 MERS-CoV positive index patients were RNA or antibody positive; the rate of general transmission, even in outbreaks is around 3 % [31]. It seems that the majority of human cases of MERS-CoV, even when numbers appear to increase suddenly, do not readily transmit to more than one other human so to date, the localized epidemic of MERS-CoV has not been self-sustaining [157–161]. That is to say, the basic reproduction number (R₀) - the average number of infections caused by one infected individual in a fully susceptible population – has been close to one throughout various clusters and outbreaks. If R₀ was greater than 1, a sustained increase in case numbers would be expected. Some R_o calculations may be affected by incomplete case contact tracing, limited community testing and how a case is defined. That MERS has had a constant presence in the Arabian Peninsula since 2012 is due to ongoing, sporadic spill-over events from DCs amplified by poorly controlled hospital outbreaks.

The first known MERS human-to-human transmission event was one characterized by acute LRT disease in a healthcare setting in Jordan. In stark contrast, a sero-survey of HCW who were sometimes in close and prolonged contact with the first, fatal MERS-CoV case in 2012 [162], found none of the HCW had seroconverted four months later, despite an absence of eye protection and variable compliance with required PPE standards [162].

Early on in the MERS story, samples for testing were mostly collected from patients with severe illness and not those with milder acute respiratory tract infections. Contacts of confirmed MERS cases were often observed for clinical illness, but not tested. These omissions may have confounded our understanding of MERS-CoV transmission and biased early data towards higher numbers of seriously ill and hospitalized patients, inflating the apparent proportion of fatal cases. Case–control studies were not a focus. As testing paradigms changed and contacts were increasingly tested, more asymptomatic and mild infections were recognized [163].

A rise in the cases termed asymptomatic (which enlarge the denominator for calculations of the proportion of fatal cases, defined in [164]) resulted in a drop in the proportion of fatal cases during the Jeddah-2014 outbreak. Historically, such rises are consistent with changing definitions and laboratory responses and clinical management of a newly discovered virus infection that was first noted only among the severely ill. Upon follow-up, over three-quarters of such MERS-CoV RNA positive people did recall having one or more symptoms at the time, despite being reported as asymptomatic [165] raising some question over the reliability of other reported data.

The proportion of fatal MERS cases within the KSA compared to outside the KSA, as well as the age, and sex distribution change in different ways when comparing MERS outbreaks. Approximately 43 % of MERS cases (549 of 1277) in the KSA were fatal betwen 2012 and December 2015 while 21 % (72 of 330) died among those occurring outside of the KSA. The total number of male cases always outnumber females and the proportion of male deaths is always greater than the proportion of females who die. However the proportion of male deaths from total males with MERS is a similar figure to that for females. In the KSA, there is a greater proportion of younger males among cases and deaths than were observed from the 2015 South Korean or the Jeddah-2014 outbreaks (Additional file 2: Figure S2). Why these aspects have differed may be due to differences in the time to presentation and diagnosis, the nature and quality of supportive care, the way a person became infected (habits, exposure to a human or zoonotic source, viral load, route of infection) or the extent to which different populations are burdened by underlying diseases [40].

As a group, HCWs comprised 16 % of MERS cases in the KSA and South Korea. It is apparent that the weekly proportion of infected HCWs increases alongside each steep rise in overall detections (Fig. 5). In May 2013, the WHO published guidelines for IPC during care of probable or confirmed cases of MERS-CoV infection in a healthcare setting [166]. This is explainable because to date, each case rise has been intimately associated with healthcare-facility related outbreaks [118]. These rises in MERS-CoV detections can decrease the average age during each event because HCWs are usually younger than inpatients with MERS. Healthcare facilities have been a regular target for suggested improvements aimed at improving infection prevention and control (IPC) procedures [115, 118].

Open in a separate window

Fig. 5

Data on MERS-CoV detections among HCWs based on publicly described laboratory confirmed cases collated into the author’s curated line list as at 4^th September 2015. Sources of these public data include the WHO, Ministries of Health and FluTrackers [207–209]. Earlier and subsequent versions of this chart are maintained on a personal blog [210]

Molecular epidemiology: using genomes to understand outbreaks

Most of the analysis of MERS-CoV genetics has been performed using high throughput or “deep” sequencing methods for complete genome deduction [167–169]. MERS-CoV was the first subject of such widespread use of deep sequencing to study an emerging viral outbreak with global reach. The technique can produce genomic length coverage in a single experiment with highly repetitious measurement of each nucleotide position [52, 140]. Despite assays having been published early on, subgenomic sequencing, once the mainstay of viral outbreak studies, has less often been published during MERS-CoV characterization [48]. As more genomes from both humans and DCs have been characterized, two clades have become apparent; A and B (Fig. 6). Clade A contains only human-derived MERS-CoV genomes from Jordan, while Clade B comprises the majority of human and camel genomes deduced thus far [168].

Open in a separate window

Fig. 6

The genetic relationship between MERS-CoV nucleotide sequences (downloaded from GenBank using the listed accession numbers and from virological.org [212]). This neighbour joining tree was created in MEGA v6 using an alignment of human and DC-derived MERS-CoV sequences (Geneious v8.1 [211]). Clades are indicated next to dark (Clade A) or pale (Clade B) blue vertical bars. Camel icons denote genomes from DCs. Healthcare or community outbreaks are boxed and labelled using previously described schemes [212, 213]

Two studies during 2015, one looking at Jeddah-2014 MERS-CoV variants and another looking at a variant exported from South Korea to China, have now identified signs of genetic recombination among MERS-CoV variants. While human and camel whole genome sequences have retained >99 % identity with each other, members of genetically distinct lineages can and do swap genetic material when suitable conditions and coinfections co-occur [170–172]. Shared identity implies that the major source for human acquisition is the DC, rather than another animal, although more testing of other animal species is needed to confirm that conclusion. Over a month, a DC virus sequenced on different occasions did not change at all indicating a degree of genomic stability in its host, supporting that DCs are the natural, rather than intermediate, host for the MERS-CoV we know today [77]. To date, recombination has been localised to breakpoints near the boundary between ORF1a and ORF1b regions, within the spike gene [170] and in the ORF1b region (Fig. 2) [172]. It is not unexpected that recombination should occur since it is well known among other CoVs [124] and because the majority of MERS-CoV whole genomes collected from samples spanning three years (2012–2015) and from humans, camels and different countries have shown close genetic identity to each other, with just enough subtle variation to support outbreak investigations so long as whole genome sequencing is applied [52, 77, 135, 138, 168, 173–175].

Changes in genome sequence may herald alterations to virus transmissibility, replication, persistence, lethality or response to future drugs. If we have prior knowledge of the impact of genetic changes because of thorough characterization studies, we can closely monitor the genomic regions and better understand any changes in transmission or disease patterns as they occur. Genetic mutations noted during the largest of human outbreaks, Jeddah-2014, did not impart any major replicative or immunomodulatory changes when compared to earlier viral variants in vitro [156, 176]. However, we understand very little of the phenotypic outcomes that result from subtle genetic change in MERS-CoV genomes. To date no clinical relevance or obvious in vivo changes to viral replication, shedding or transmission has been reported or attributed to mutations or to new recombinant viruses [156]. But vigilance and larger, more contemporary and in vivo studies are needed.

Genome sequence located to a distinct clade were identified from an Egyptian DC that was probably imported from Sudan. This does not fit into either of the current clades [125, 168, 177]. A virus sequenced from a Neoromicia capensis bat was more closely related to MERS-CoV than other large bat-derived sequences had been to that point, but the genome of a variant of a MERS-CoV has yet to be discovered and deduced from any bat [125].

Analyses of MERS-CoV genomes have shown that most single nucleotide differences among variants were located in the last third of the genome (Fig. 2), which encodes the spike protein and accessory proteins [168]. At least nine MERS-CoV genomes contained amino acid substitutions in the receptor binding domain (RBD) of the spike protein and codons 158 (N-terminal region), 460 (RBD), 1020 (in heptad repeat 1), 1202 and 1208 bear investigation as markers of adaptive change [140, 169]. The spike protein had not changed in the recombinant MERS-CoV genome identified in China in 2015 but was reported to have varied at a higher rate than that for complete MERS-CoV genomes, among South Korean variants [172, 178]. This highlights that subgenomic regions may not always contain enough genetic diversity to prove useful for differentiating viral variants. Despite this, one assay amplifying a 615 nucleotide fragment of the spike S2 domain gene for Sanger sequencing agreed with the results generated by the sequencing of a some full genomes and was useful to define additional sequence groupings [177].

Genomic sequence can also be used to define the geographic boundaries of a cluster or outbreak and monitor its progress, based on the similarity of the variants found among infected humans and animals when occurring together, or between different sites and times (Fig. 6) [169]. This approach was employed when defining the geographically constrained MERS hospital outbreak in Al-Ahsa, which occurred between 1^st April and 23^rd May 2013, as well as clusters in Buraidah and a community outbreak in Hafr Al-Batin, the KSA. Genomic sequencing identified that approximately 12 MERS-CoV detections from a community outbreak in Hafr Al-Batin between June and August 2013 may have been triggered by an index case becoming infected through DC contact [175]. Sequencing MERS-CoV genomes from the 2013 Al-Ahsa hospital outbreak indicated that multiple viral variants contributed to the cases but that most were similar enough to each other to be consistent with human-to-human transmission. Molecular epidemiology has revealed otherwise hidden links in transmission chains encompassing a period of up to five months [179]. However, most outbreaks have not continued for longer than two to three months and so opportunities for the virus to adapt further to humans through co-infection and sustained serial passage have been rare [169]. In Riyadh-2014, genetic evidence supported the likelihood of multiple external introductions of virus, implicating a range of healthcare facilities in an event that otherwise looked contiguous [23, 168, 179]. Riyadh is a nexus for camel and human travel and has had more MERS cases than any other region of the KSA to date but also harbours a wide range of MERS-CoV variants [128, 167, 179]. However the South Korean outbreak originated from a single infected person, resulting in three to four generations of cases [180, 181]. Studies of this apparently recombinant viral variant did not find an increased evolutionary rate and no sign of virus adaptation thus the outbreak seems to have been driven by circumstance rather than circumstance together with mutation [181].

Contact tracing and the possible importance of asymptomatic cases

For many MERS cases detected outside the Arabian Peninsula, extensive contact tracing has been performed and the results described in detail. Contact tracing is essential to contain the emergence and transmission of a new virus and today it is supported by molecular epidemiology. Although it is an expensive and time consuming process, contact tracing can identify potential new infections and through active or passive monitoring, react more rapidly if disease does develop. Results of contact tracing to date have found that onward transmission among humans is an infrequent event. For example, there were 83 contacts, both symptomatic and asymptomatic, of a case treated in Germany who travelled from the UAE but no sign of virus or antibody were found in any of them [73]. The very first MERS case had made contact with 56 HCWs and 48 others, but none developed any indication of infection [162]. In a study of 123 contacts of a case treated in France, only seven matched the definition for a possible case and were tested; one who had shared a 20 m² hospital room while in a bed 1.5 m away from the index case for a prolonged period was positive [26]. None of the contacts of the first two MERS cases imported into the USA in 2014 contained any MERS-CoV footprint [182] and none of the 131 contacts of two travellers returning to the Netherlands developed MERS-CoV antibodies or tested RNA positive [25, 183]. Analyses of public data reveal many likely instances of nosocomial acquisition of infection in the Arabian Peninsula and these data may be accompanied by some details noting contact with a known case or facility. One example identified the likely role of a patient with a subclinical infection, present in a hospital during their admission for other reasons, as the likeliest index case triggering a family cluster [93]. Contact tracing was a significant factor in the termination of a 2015 outbreak involving multiple South Korean hospitals [184]. Such studies demonstrate the necessity of finding and understanding a role for mild and asymptomatic cases, together with restricting close contact or prolonged exposure of infected people to others, especially older family members and friends with underlying disease (Fig. 4c).

Any unreferenced opinions expressed herein are those of the authors and do not necessarily represent the views of any employer or institution. Our thanks go to Andrew Rambaut, Maia Majumder, Marion Koopmans and Hale Abdali for helpful discussions on social media and CMAM and RIAM for patience.