25 Arenaviruses

There are few groups of viral zoonoses that have attracted such widespread publicity as the arenaviruses, particularly during the 1960s and 1970s when Lassa emerged as a major cause of haemorrhagic disease in West Africa. More than any other zoonoses, members of the family are used extensively for the study of virus-host relationships. Thus the study of this unique group of enveloped, single-stranded RNA viruses has been pursued for two quite separate reasons. First, lymphocytic choriomeningitis virus (LCM) has been used as a model of persistent virus infections for over half a century; its study has contributed, and continues to contribute, a number of cardinal concepts to our present understanding of immunology. LCM virus remains the prototype of the Arenaviridae and is a common infection of laboratory mice, rats and hamsters. Once thought rare in humans there is now increasing evidence of LCM virus being implicated in renal disease and as a complication in organ transplantation. Second, certain arenaviruses cause severe haemorrhagic diseases in man, notably Lassa fever in Africa, Argentine and Bolivian haemorrhagic fevers in South America, Guaranito infection in Venezuela and Chaparé virus in Bolivia. The latter is a prime example for the need of ever-continuing vigilance for the emergence of new viral diseases; over the past few years several new arenaviruses have been reported as implicated with severe human disease and indeed the number of new arenaviruses discovered since the last edition of this book have increased the size of this virus family significantly.

In common with LCM, the natural reservoir of these infections is a limited number of rodent species (Howard 1986). Although the initial isolates from South America were at first erroneously designated as newly defined arboviruses, there is no evidence to implicate arthropod transmission for any arenavirus. However, similar methods of isolation and the necessity of trapping small animals have meant that the majority of arenaviruses have been isolated by workers in the arbovirus field. A good example of this is Guaranito virus that emerged during investigation of a dengue virus outbreak in Venezuela (Salas et al. 1991).

There is an interesting spectrum of pathological processes among these viruses. All the evidence so far available suggests that the morbidity of Lassa fever and South American haemorrhagic fevers due to arenavirus infection results from the direct cytopathic action of these agents. This is in sharp contrast to the immunopathological basis of ‘classic’ lymphocytic choriomeningitis disease seen in adult mice infected with LCM virus and the use of this system for elucidating the phenomenon of H2-restriction of the host cytotoxic T cell response (Zinkernagel and Doherty 1979). Despite the utility of this experimental model for dissecting the nature of the immune response to virus infection and the growing interest in arenaviruses of rodents, there remains much to be done to elucidate the pathogenesis of these infections in humans.

The grouping was first recognized on the basis of a serological cross reaction observed between LCM and Machupo virus the latter being found to cause Bolivian haemorrhagic fever in the 1960s. The Arenaviridae take their name from their sand-sprinkled appearance when viewed in the electron microscope (Latin: arena = sand). The members of the family are listed in Table 25.1. The various strains and isolates of LCM are now considered to be a genus within the Arenaviridae. A close serological relationship exists between LCM, Lassa virus and other arenaviruses from Africa. For this reason, they are loosely referred to as the ‘Old World’ arenaviruses, in contrast to those from the Americas, although LCM can be found world-wide (Howard and Simpson 1980). The ‘New World’ arenaviruses show varying degrees of serological relationships with Tacaribe virus, first isolated in Trinidad. For this reason, viruses from the Americas are frequently regarded as members of the Tacaribe complex (Fig. 25.1).

With the exception of LCM, all are referred to by names that reflect the geographical locality in which they were isolated. Various strain designations are also commonly used, in particular for LCM and arenaviruses isolated from man. Multiple isolations of non-pathogenic viruses that infect New World rodents are made less frequently, with the exception of Pichinde virus where a large number of field isolates from Columbia have been characterized.

All but one of the 24 members of the Arenaviridae so far described have rodents as their natural reservoir hosts. The exception is the Tacaribe virus, which was originally isolated from the fruit-bat, Artibeus literatus. Although rodents are divided globally into over 30 families, arenaviruses are predominantly found within two major families, the Muridae (e.g. mice and rats) and Cricetidae (e.g. voles, lemmings, gerbils). The nature of the original reservoir for LCM virus remains obscure, but it appears to be a species of the Muridae which evolved in the Old World and subsequently spread to most parts of the globe. There is a wide range of tropism and virulence among laboratory strains of LCM virus originally isolated from laboratory mouse colonies.

The natural reservoir of Lassa virus, Mastomys natalensis, is also a member of Muridae and, in common with the host of LCM, frequents human dwellings and food stores. In contrast, nearly all arenaviruses isolated from South America are associated with cricetid rodents whose members frequent open grasslands and woodland.

In recent years, several new arenaviruses have come to our attention. In the instance of Guanarito virus, this new form of haemorrhagic fever from Venezuela was originally mistaken as dengue (Gonzalez et al. 1995). A virus from Brazil dubbed Sabiá virus caused a laboratory-associated infection during investigations of a new outbreak of hitherto unrecorded febrile illness in the southern provinces. Whitewater Arroyo virus was found in the USA in 1999 and a virus distinct from Machupo caused death in Bolivia in 2004. Each instance brings into sharp relief the need for continuing vigilance of these zoonoses.

Negative-staining electron microscopy of extracellular virus shows pleomorphic particles ranging in diameter from 80 to 200 nm (Fig. 25.2).

The virus envelope is formed from the plasma membrane of infected cells. A significant thickening of both bilayers of the membrane together with an increase in the width of the electron-translucent intermediate layer is characteristic of arenavirus development. Little is known about the internal structure of the arenavirus particle, although thin sections of mature and budding viruses clearly show the ordered, and often circular, arrangements of host ribosomes that are typical of this virus group, conferring the ‘sandy’ appearance from which its name is derived. Distinct well dispersed filaments 5–10 nm in diameter are released from detergent-treated virus. Two predominant size classes are present, with average lengths of 649 nm and 1300 nm respectively; these lengths do not show a close relationship with the two virus-specific L and S RNA species. Each is circular and beaded in appearance. Convoluted filamentous strands up to 15 nm in diameter can be seen in preparations of spontaneously disrupted Pichinde virus. These appear to represent globular condensations which arise from an association between neighbouring turns of the underlying helix. The basic configuration of the filaments shows a linear array of globular units up to 5 nm in diameter, probably representing single molecules of the viral polypeptide. These filaments progressively fold through a number of intermediate helical structures to produce the stable 15 nm diameter forms (Young 1987). Cryomicroscopy suggests that the nucleocapsid forms into a lattice just beneath the inner leaflet of the viral lipid layer (Neuman et al. 2005).

Arenaviruses replicate in experimental animals in the absence of any gross pathological effect. However, cellular necrosis may accompany virus production, not unlike that seen in virus-infected cell cultures. The variable pathological changes associated with arenavirus infections are further complicated by the occasional appearance of particles in tissue sections that react strongly with fluorescein-conjugated antisera. Granular fluorescence with convalescent serum in the perinuclear region of acutely infected Vero cells is often seen. In addition, intracytoplasmic inclusion bodies are prominent in virus-infected cells both in vitro and in vivo. These usually appear early in the replication cycle and consist largely of single ribosomes which later become condensed in an electron-dense matrix, sometimes together with fine filaments (Murphy and Whitfield 1975).

The genome of arenaviruses consists of two single-stranded RNA segments of different sizes, designated L and S, with the S strand being more abundant. Analysis of RNA is complicated by the presence of ribosomal 18S and 28S RNA although these cellular RNA species are not essential for virus replication. The total ribosomal RNA content may in turn be influenced by the varying proportions of infectious to non-infectious particles present in virus stocks. Whether or not there is a role for these host RNA molecules in the establishment and maintenance of persistent infections is unclear. These are small amounts of both cell and viral low molecular weight RNA contained within virus particles, one of which codes for the Z protein required for replication (see below).

Genetic studies have shown that the S strand codes for the nucleoprotein (N) and the envelope glycoprotein precursor (GPC) in two main open reading frames located on RNA molecules of opposite polarity. The 3’ half of the S RNA codes for the N protein by production of a mRNA with a nucleotide sequence complementary to the viral genome. In contrast, the GPC is expressed by the 5’ half replication of the S RNA strand which is required to undergo replication before the production of a mRNA with a viral-sense sequence specific for the GPC protein. Thus expression of the genome is by synthesis of subgenomic RNA from full length templates of opposite polarities. This strategy of ‘ambisense’ coding for viral protein has so far been described only for the arenaviruses and some bunyaviruses. The reading frames for the two major gene products are separated by a hairpin structure of approximately 20 paired nucleotides. This intergenic region may act as a control mechanism for genome expression but there is as yet no experimental evidence to support this possibility.

The L RNA strand represents about 70% of the viral genome coding for both the viral polymerase (L) and the Z protein. Re-assortment studies with virulent and avirulent strains of LCM virus have shown that lethal disease in guinea-pigs is associated with the L RNA strand. The L protein is encoded by a large open reading frame covering 70% of the L RNA strand: it is expressed via a mRNA complementary in sense to the viral genome.

Genome sequencing can produce useful qualitative and quantitative comparisons between newly discovered arenavirus isolates and those already characterized. The first detailed study, reported by Bowen et al. (1997) used data from at least one strain of all arenaviruses recognized at that time. The resulting phylogenetic tree confirmed the distant relationship between Old World and New World arenaviruses, broadly in line with detailed antigenic analysis using monoclonal and polyclonal antibodies. Subsequent work on newer isolates coupled with the use of full length sequences shows that the New World arenaviruses are divisible into 3 lineages:

There seems to be much less variability among the Old World viruses with Mopeia and Mobala viruses being closely related to Lassa virus.

Nucleotide sequencing also shows extensive genetic diversity between different isolates that is not always correlated to location and/or the animal host reservoir. For example, Guanarito virus isolated from clinical cases of Venezuelan haemorrhagic fever revealed considerable heterogeneity of sequence, greater than that observed among rodent isolates (Weaver et al. 2000). Whether or not this is indicative of a mechanism whereby the virus evades the host immune response and is the forerunner of persistent infection in the rodent host, is unclear.

The propensity to cause serious human illness appears to have evolved quite separately among the Old World and New World arenaviruses. That the South American haemorrhagic fevers all appear confined to clade B suggests a single series of mutational events leading to a capacity to cause human disease. In contrast, Lassa virus has likely acquired its capacity to inflict serious human illness by a quite separate series of evolutionary events that are not clear from phylogenetic analysis alone.

The arenavirus genome codes for at least five proteins; an RNA-dependent RNA polymerase and a zinc-binding (Z) protein from the L strand and three structural proteins from the smaller, S strand. Extracellular particles contain a major nucleocapsid-associated protein of molecular weight 54–68 kD with two glycoproteins in the outer viral envelope. These envelope glycoproteins are not primary gene products but arise by proteolytic cleavage of a larger, 75 kD glycoprotein precursor polypeptide (GPC) at a unique cleavage site conserved among the majority of arenaviruses. The first 59 amino acids at the N terminus of GPC act as a signal sequence for membrane insertion. Maturation and release of virus does not seem to be markedly inhibited in the presence of tunicamycin, an inhibitor of glycosylation, but glycoprotein cleavage is essential for infectivity.

The major glycoprotein species (GP2) in the molecular weight range of 34–42 kD represents the C-terminal cleavage product of the GPC envelope glycoprotein precursor. A major antigenic site has been located between amino acids 390–405, and cross-reactive monoclonal antibodies bind to epitopes in this region. The corresponding N-terminal product of GPC cleavage (GP1) is probably highly glycosylated with at least four antigenic domains. Neutralizing monoclonal antibodies to LCM virus map to two of these regions and there is less sequence homology between the GP1 than between the GP2 molecules of different arenaviruses. Once cleaved, the glycoproteins GP1 and GP2 form tetrameric structures embedded in the viral envelope. The proximal GP2 molecules forming electrovalent bonds with the underlying viral nucleocapside, possibly mediated by the Z protein.

The internal nucleocapsid-associated (N) protein accounts for much of the virus-specific protein present in purified virus and infected cells, and remains bound to the virus genome after solubilization of the virus with non-ionic detergents. Molecular cloning studies have shown a surprisingly high degree of homology between the N proteins of Old and New World arenaviruses, and this would account for the serological cross-reactions seen using certain monoclonal antibodies to the N protein. A high degree of conservation between such epidemiologically distinct viruses may indicate precise functional roles for certain areas of the N polypeptide in virus replication. Cleavage products of the N protein are a consistent feature of both virus and virus-infected cells. Cleavage is not noticeable in Vero cells; yields of arenaviruses are lower in these cells, perhaps due to reduced availability of N for packaging. A fragment of the N protein is often seen in the nuclei although the exact function of this is not clear.

A minor component with a molecular weight in excess of 150 kD is often observed in infected cells. This L protein represents the virus-specific RNA polymerase (Fuller-Pace and Southern 1989). Amino acid sequences common to the viral polymerase are present along the open reading frame coding for the L protein, which suggests the conservation of certain functional domains. A small, 12 kD viral polypeptide, the so-called Z, or zinc-binding protein is considered to play a role in controlling the replication and expression of the genome: the Z protein may also modulate interferon responses in vivo (Djavani et al. 2001).

Arenaviruses replicate in a wide variety of mammalian cells although either BHK-21 cells or monkey kidney cell lines are used for molecular studies. Maximal virus adsorption to cell surfaces is at 2 hours at 37^˚C. At a low multiplicity of infection (i.e. below 0.1) the latent period is approximately 6–8 hours, after which cell-associated virus increases exponentially. The titre of extracellular virus reaches a maximum 36–48 hours after infection. The passage history of any particular virus stock is probably one of the most critical factors which determines the kinetics of arenavirus replication.

Infected cells undergo only limited cytopathic changes with little or no change in the total level of host cell protein synthesis; virus yields vary in different susceptible cell types. Cell metabolism is only minimally affected and in some cells only a reduction in differentiated, or ‘luxury’ cell functions can be observed. Cultures of persistently infected cells are readily established, the morphology and growth kinetics of which are similar to those of uninfected cells.

Only limited information is available concerning the replication and expression of viral RNA within infected cells, although possible replication events can be predicted from the nucleotide sequences of L and S genome segments. The major feature of an ambisense coding strategy is that it allows for independent expression and regulation of the N and GPC genes from the S RNA segment. The N protein is independently expressed late in acute infection and in persistently infected cells in the absence of low levels of glycoprotein production. This is explained by the production of subgenomic mRNA from a negative polarity, virus-sense template.

A control mechanism must therefore exist which determines the fate of nascent RNA of negative polarity, destined either for encapsidation or as a template for N protein-specific mRNA. In contrast, the template for glycoprotein-specific mRNA is of complementary sense to viral RNA and as such would not be required for nascent virus production. The lack of glycoprotein late in the replicative cycle or in persistently infected cells would therefore imply selective transcriptional or translational control of this gene product.

Both viral RNA and its complementary strand contain hairpin sequences which may provide recognition points for termination of transcription by viral RNA polymerase. The nucleotide sequence in the hairpin region is of coding sense and may be transcribed, either as a discrete mRNA species or as a result of extended transcription of N or GPC messengers through this region. The postulated reading frames for viral gene products transcribed from LCM and Pichinde viral genomes would fit this hypothesis. In addition, a sequence for ribosomal 18S subunit binding is present on both mRNA molecules although the significance of this is not clear.

Early diagnosis is essential but made difficult for the clinician as patients present with relatively non-specific symptoms. A history of travel to a region where arenavirus infections are common can help direct a laboratory investigation. Owing to the influenza-like nature of the early signs, other infections need to be excluded, for example yellow fever and dengue. If neurological signs are also present, then other causes of encephalitis or meningitis need to be excluded.

The diagnosis of arenavirus infections may be made by demonstration of a four-fold rise in specific antibody titre, the presence of IgM viral antibodies, or isolation of the virus. For routine isolation, the E6 clone of Vero cells is the cell line of choice, although all arenaviruses grow well in primate and rodent-derived fibroblast cell lines. However, a CPE is often difficult to see, and inoculated cultures often require examination by immunofluorescence (IF) or ELISA in order to detect the presence of viral antigens.

Both ELISA and IF tests have been used successfully for the diagnosis of human arenavirus infections. In the case of Lassa fever, infected cell substrates are used that have been treated by UV light, acetone, and cobalt irradiation to ensure safety. Drops of cell cultures dried onto glass slides can be prepared in a central laboratory and these preparations remain stable for many months. Most of the antigen detected within acetone-fixed infected cells represents cytoplasmic nucleocapsid protein. In the case of the New World arenaviruses, serological cross-reactions in the IF test (e.g. with sera from patients with Bolivian (Machupo) and Argentine (Junín) haemorrhagic fevers) are found with fixed cultures. Substrates prepared from other members of the Tacaribe complex, which includes Junín and Machupo viruses also react with sera taken from these patients during the acute phase and early convalescence. Greatest cross-reactivity can be seen between the closely-related Junín and Machupo antigens, closely followed by Tacaribe virus-infected cells. Recent experience in Bolivia however, has shown that new arenaviruses may be missed by placing sole reliance on serology (Delgado et al. 2008).

The use of PCR for arenavirus diagnosis requires the rigorous testing of primer sets and the optimization of both RNA extraction and temperature cycling. Even so, an arenavirus not previously found may be missed. The use of many primer sets may lead to a loss of sensitivity and non-specific amplification, although such pitfalls can be avoided with care (Drosten et al. 2002). But in the final analysis virus isolation should be attempted as soon as possible by referring samples to a specialist containment laboratory.

Monoclonal antibodies can distinguish between virus strains because they can be prepared against epitopes which go unrecognized when polyclonal antisera are used. Buchmeier et al. (1981) summarized the patterns of reactivity with a panel of monoclonal antibodies directed against laboratory strains of the homologous virus, and Lassa and Mopeia viruses. Reagents directed against the smaller, GP2 envelope glycoprotein cross-reacted by immunofluorescence with all substrates examined, whereas antibodies directed against the larger GP1 glycoprotein were either strain-specific or reacted with a subset only of the strains examined, presumably by binding to previously unrecognized epitopes. The observations that certain of these broadly cross-reactive antibodies also reacted with Pichinde virus suggests that epitopes on surface envelope structures among Old World and New World arenaviruses are conserved. A similar comparison has also been undertaken with monoclonal antibodies to Lassa tested against the Mopeia and Mobala viruses from Africa. Again, various degrees of cross-reactivity were observed with reagents specific for the GP2 external glycoprotein. Mobala virus from the Central African Republic, however, appears to be distinct, as several cross-reactive monoclonal antibodies originally prepared against LCM virus failed to recognize Mobala-infected substrates.

The neutralization test is highly specific for all members of the Arenaviridae; it is notable that the few examples of cross-reactivity were obtained with high titre animal antisera raised against Junín, Tacaribe and Machupo viruses. However, the ease with which neutralizing antibodies can be quantified varies greatly. No cross-reactions have been observed between Junín and Machupo viruses in plaque-reduction tests with human convalescent sera despite a close antigenic relationship. A similar marked specificity of neutralization has been demonstrated with LCM and Lassa sera, and both viruses are readily distinguishable from one another by this technique (however, neutralizing antibodies to Lassa virus can be detected only with great difficulty). The sensitivity of the neutralization test for LCM virus can be increased by incorporating either complement or anti-gammaglobulin into the test system.

The mechanisms by which arenaviruses cause disease in man are not fully understood. There is no evidence that either immunopathological or allergenic processes play any part in causing disease; it appears to be more likely that disease is caused by direct damage of cells by the virus. Postmortem studies on patients who died from Junín virus infection have shown generalized lymphadenopathy, endothelial swelling in the capillaries and arterioles of almost every organ, and depletion of lymphocytes in the spleen. Virus first replicates in lymphoid tissue from whence it invades the reticuloendothelial system and those cells concerned in the immune and cellular immune responses; the host’s defence mechanisms are thus impaired. Fatal illness is invariably associated with capillary damage leading to capillary fragility, haemorrhages and irreversible shock (Johnson et al. 1973).

Disseminated intravascular coagulation is not a typical feature. Although Lassa fever is often regarded as being hepatotropic, the extent of hepatic damage is insufficient to account for the severity of the clinical disease. Studies of Lassa virus-infected rhesus monkeys have shown that changes in vascular function may play a much greater role in pathogenesis, as a result either of viral replication in the vascular epithelium or of secondary effects of virus activity in different organs. Platelet and epithelial cell functions fail immediately before death and are accompanied by a drop in the level of prostacyclin; these functions rapidly return to normal in animals surviving infection (Fisher-Hoch et al. 1987). Impairment of the functions of vascular epithelium in the absence of histological changes appears to be a common feature of the final stages of viral haemorrhagic diseases in general and suggests that hypovolaemic shock may be amenable to treatment with prostacyclin.

The pathogenesis of Argentine haemorrhagic fever caused by Junín virus has been studied in experimentally infected guinea-pigs, this being a suitable model of human disease. There is a pronounced thrombocytopenia and leucopenia characteristic of human infections, and animals die of severe haemorrhagic lesions. Bone marrow cells are destroyed with release of proteases and acid and alkaline phosphatases into the blood; this leads to consumption of the C4 component of complement. These effects may lead in turn to progressive alterations in vascular permeability and platelet function (Rimoldi and de Bracco 1980).

The most extensive histopathological studies have been made on tissues from patients with Lassa fever (Walker and Murphy 1987). However, there are many similarities in the pathological lesions found in man following Junín and Machupo virus infections. Focal non-zonal necrosis in the liver has been described in all three conditions with hyperplasia of Kupffer cells, erythrophagocytosis and acidophilic necrosis of hepatocytes. Councilman-like bodies can be observed together with cytoplasmic vacuolations and nuclear pyknosis or lysis. As with other organs, there is little evidence of cellular inflammation. Lesions in other organs have been described, including interstitial pneumonitis, tubular necrosis in the kidney, lymphocytic infiltration of the spleen and minimal inflammation of the central nervous system and myocardium (Walker and Murphy 1987). The hepatic changes may be grouped into three categories:

These changes are consistent with a direct cytolytic action of the virus; nevertheless, the simultaneous presence of Lassa virus and specific antibodies during the later stages of the acute disease suggest that antibody-dependent cellular immune reactions may also occur. Microscopic changes in the kidneys are minimal; it is not clear whether the impairment of renal function is due to the deposition of antigen-antibody complexes.

Until recently human infection was regarded as a rare, inapparent infection that may occasionally present as an influenza-like febrile illness, as aseptic meningitis or as severe meningoencephalomyelitis.

LCM virus may be a more important human pathogen than considered hitherto. Palacios et al. (2008) reported the presence of virus in patients receiving kidney and liver transplants who subsequently died of an acute febrile illness. The report follows others implicating LCM virus in fatal haemorrhagic-like illnesses following solid organ transplantation (Fischer et al. 2006).

The incubation period is 6–13 days. In the influenza-like illness there is fever, malaise, coryza, muscular pains and bronchitis. The meningeal form is more common; the same symptoms may remain mild and be of short duration and patients recover within a few days, but there can be more pronounced illness with severe prostration lasting 2 weeks or more. Chronic sequelae have been reported on occasion. They include headache, paralysis and personality changes. The few deaths have followed severe meningoencepalomyelitis. In one case there was mild pharyngitis and a diffuse erythematous rash followed by haemorrhages and death.

An early leucopenia followed by lymphocytosis is a constant finding. In central nervous system disease, the cerebrospinal fluid (CSF) is at increased pressure with a slight rise in protein concentration, normal or slightly reduced sugar concentration, and a moderate number of cells, mainly lymphocytes (150–400/mm³). These changes are not, of course, restricted to LCM infections. Virus can be isolated from blood, CSF and, in fatal cases, from brain tissue.

Man is usually infected through contact with rodents. Many infections have been acquired in laboratories, where LCM may be a contaminant in laboratory colonies of mice and hamsters. Hamsters kept as pet animals have also played a role in human infection. The mechanism of transmission of the virus to man is not fully understood but is likely to involve dust contaminated by urine, the contamination of food and drink, or via skin abrasions. Recent clusters of cases among transplant recipients would suggest that LCM virus may have become chronic in the otherwise asymptomatic donors. In at least one instance, donor infection could be traced back to ownership of a pet hamster infected with LCM virus (Fischer et al. 2006)

Lassa virus made a dramatic appearance in Nigeria in 1969 as a lethal, highly transmissible disease. The first victim was an American nurse who was infected in a small mission station in the Lassa township in north-eastern Nigeria, whence the virus and the disease derive their names. The origin of the infection was never determined, although it is thought to have been acquired through direct contact with an infected patient in Lassa. When the nurse’s condition steadily deteriorated she was flown to the Evangel Hospital in Jos, where she died the following day.

While she was in hospital she was cared for by two other American nurses, one of whom also became infected by direct contact, probably through a skin abrasion. This nurse became unwell after an 8-day incubation period and died following an illness lasting 11 days. The head nurse of the hospital, who had assisted at the post-mortem of the first patient, fell ill 7 days after the death of the second patient from who she had cared, and from whom she probably acquired the infection.

This third case was evacuated to the USA by air in the first-class cabin of a commercial airliner with two attendants and screened from economy-class passengers only by a curtain. After a severe illness under intensive care she slowly recovered. A virus, subsequently named Lassa, was isolated from her blood by workers at the Yale Arbovirus Unit. One of these virologists became ill but improved after an immune plasma transfusion donated by the third case. Five months after this infection, a laboratory technician in the Yale laboratories, who had not been working with Lassa virus, fell ill and died. The manner in which this infection was acquired has never been determined.

This trail of events not unnaturally earned for Lassa virus a formidable notoriety, which was sharply enhanced by two more devastating hospital outbreaks—one in Nigeria, the other in Liberia. A further epidemic was seen in Sierra Leone in October 1972. In sharp contrast to the previous outbreaks, this one was not confined to hospitals, although hospital staff were at considerable risk and several became infected. Most of the patients acquired their illness in the community and there was several intra-familial transmissions. This led to a revision of the initial view—formed from experience of nosocomial infections—that Lassa fever has a high mortality.

Lassa fever has since continued to occur in West Africa, usually as sporadic cases (Monath 1987). Between 1969 and 1978 there were 17 reported outbreaks affecting 386 patients in whom the mortality was 27%. Eleven of the episodes were in hospitals, where the case fatality rate reached 44%; two were laboratory infections, two were community-acquired outbreaks, and two were prolonged community outbreaks. Eight patients were flown to Europe or North America. One of them was evacuated with full isolation precautions and the remainder, of whom five were infectious, travelled on scheduled commercial flights as fare-paying passengers. Fortunately, no contact cases resulted.

Lassa virus causes a spectrum of disease ranging from subclinical to fulminating fatal infection. The incubation period ranges from 3 to 16 days and the illness usually begins insidiously. The disease is difficult to distinguish in the early stages from other systemic febrile illnesses, the most reliable clinical signs being a sore throat and vomiting. Between the third and sixth day of illness the symptoms suddenly worsen and there is high fever, severe prostration, chest and abdominal pains, conjunctival injection, diarrhoea, dysphagia and vomiting. Chest pain located substernally and along the costal margins, is often associated with tenderness on pressure and is exacerbated by coughing and deep inspiration. One important physical finding is a distinct pharyngitis; yellow-white exudative spots may be seen on the tonsillar pillars together with small vesicles and ulcers. The patient appears toxic, lethargic and dehydrated; the blood pressure is low and there is sometimes a bradycardia relative to the body temperature. There may be cervical lymphadenopathy, coated tongue, puffiness of the face and neck, and blurred vision. Occasionally a faint maculopapular rash may be seen during the second week of illness on the face, neck, trunk and arms. In severe cases, haemorrhages also occur. Cough is a common symptom, and light-headedness, vertigo and tinnitus appear in a few patients. Deafness has also been noted in about 20% of patients and, although it may be reversible, is more often permanent.

The fever generally lasts for 7–17 days and is variable. Convalescence begins in the second to fourth weeks, when the temperature returns to normal and the symptoms improve. Most patients complain of extreme fatigue for several weeks. Loss of hair happens occasionally and there may be brief bouts of fever.

Patients in whom the disease is fatal not uncommonly have a high sustained fever. Acutely ill patients suddenly deteriorate between days 7 and 14 with a sudden drop in blood pressure, peripheral vasoconstriction, hypovolaemia and anuria; there may be pleural effusions and ascites. In addition, coma, stupor, tremors and myoclonic twitching may occur. Death is due to shock, anoxia, respiratory insufficiency and cardiac arrest.

Lassa virus has been repeatedly isolated from the multimammate rat Mastomys natalensis in Sierra Leone and Nigeria. This rodent is a common domestic and peridomestic species, and large populations are widely distributed in Africa south of the Sahara. During the rainy season it may desert the open fields and seek shelter indoors. Some genetic variation has been shown in Mastomys populations inhabiting different ecological niches; however, there appears to be no difference in the prevalence of antibody and virus in at least two of the karyotypes found in West Africa. The animals are infected at birth or during the perinatal period. Like other arenaviruses, Lassa virus produces a persistent, tolerated infection in its rodent reservoir with no ill effects and without any detectable host immune response. The animals remain infected throughout their lifetime, freely excreting Lassa virus in urine and other body fluids. The correlation between the prevalence of antibody in a community and the degree of infestation by infected rodents, however, is poor.

Studies of the ratio of clinical illness to infections have recently confirmed that Lassa fever is endemic in several regions of West Africa. It has been estimated that only 1–2% of infections are fatal, substantially less than the figures of 30–50% originally associated with the early nosocomial outbreaks. However, there may still be up to 300, 000 infections per year with as many as 5,000 deaths (McCormick et al. 1986). The seroconversion rates among villagers in Sierra Leone vary from 4 to 22 per 100 susceptible individuals per year; up to 14% of febrile illness in such population groups is due to Lassa virus infection. These data confirm the relatively high rate of asymptomatic and mild infections in endemic areas. One reason for this may be the frequency of reinfections; although about 6% of the population lose antibody annually, rises in antibody titre are also often observed. It is not clear if reinfection results in clinical disease. A frequent finding of incomplete immunity after infection would have profound implications for the use of a vaccine.

There may be secondary spread from person to person in conditions of overcrowded housing and this is particularly important in rural hospitals. Medical attendants or relatives who provide direct personal care are most likely to contract the infection; as noted above, accidental inoculation with a sharp instrument and contact with blood have caused infection in few cases. Airborne spread may take place, as well as mechanical transmission. Although in Sierra Leone there has been no evidence of airborne spread in hospital outbreaks, one of the 1970 outbreaks in Nigeria is believed to have been caused by airborne transmission from a woman with severe pulmonary infection.

Lassa virus infection is a common cause of spontaneous abortion in endemic areas of West Africa. Virus can easily be detected in the blood and tissues of the aborted foetuses. Pediatric infection occurs, although more frequently in male children. The acute febrile illness is accompanied by widespread oedema, abdominal distension and haemorrhaging with a case fatality rate as high as 30%.

The diagnosis of Lassa fever is confirmed by isolation of the virus or demonstration of a specific serological response. Infection in the early stages can be confused clinically with a number of other infectious diseases, particularly malignant malaria (Table 25.2).

Lassa virus grows readily in Vero cell culture and virus can usually be isolated within four days. Virus can be cultured from serum, throat washings, pleural fluid and urine; it is excreted from the pharynx for up to 14 days after the onset of illness and in urine for up to 67 days after onset. Lassa infection can be diagnosed early by detection of virus-specific antigens in conjunctival cells using indirect immunofluorescence. It is important to note that virus isolation should be attempted only in laboratories equipped to provide maximum containment (Biosafety Level 4) to protect the investigator. Suspected cases should be reported immediately to local and national public health authorities.

The most sensitive serological test for the detection of Lassa antibodies is indirect immunofluorescence; antibodies can be detected by this method in the second week of illness. On occasion antibodies fail to develop in patients from whom Lassa virus has been isolated. Neutralizing antibodies are difficult to measure in vitro, in sharp contrast to infections by the South American arenaviruses, for reasons that are unclear.

The two most reliable prognostic markers of fatal infections are the titres of circulating virus and of aspartate aminotransferase (AST). Patients in whom the titre of virus exceeds 10⁴ TCID₅₀/ml and with AST levels above 150 IU have a poor prognosis, and fatality rates approach 80%. In contrast, patients with virus and enzyme levels below these values have a greater than 85% chance of survival (Johnson et al. 1987). This demonstration of an association between the degree of viraemia and mortality is unique for virus infections and contrasts with the difficulty in predicting the outcome in patients with Argentine and Bolivian haemorrhagic fevers. Although Lassa fever can be diagnosed accurately from the presence of IgM antibodies on admission, there is no correlation between the time of appearance and the titre of specific antibodies and clinical outcome. Lassa fever is particularly severe in pregnant women. A study of 75 women in Sierra Leone showed that 11 of 14 deaths were the result of infection during the third trimester; a further 23 patients suffered abortion in the first and second trimesters.

The human host is clearly restricted in its ability to clear the virus and prevent its replication in tissues, possibly because of impairment of cytotoxic T cell reactions. The poor neutralizing antibody response and the high degree of viraemia contrast sharply with those in patients with South American haemorrhagic fevers, in whom there is little viraemia and neutralizing antibodies develop rapidly during acute infection.

Although the passive administration of Lassa immune plasma may suppress viraemia and favourably alter the clinical outcome, it does not always do so, particularly if the patient has a high virus burden (McCormick et al. 1986). Failure may be due to the difficulty in assessing accurately the titre of viral neutralizing antibodies in the plasma, the late and non-uniform nature of this response in convalescence, and antigenic variation. The widespread occurrence of human immunodeficiency virus (HIV) infections in West Africa precludes at present the use of immune plasma from convalescent individuals in this region. Conversely, immune plasma may be of benefit in the treatment of Junín infections (Maiztegui et al. 1979; Enria et al. 1984). This may be due to the high titre of neutralizing antibodies that develops soon after the acute phase.

There has been little progress towards developing a vaccine against Lassa virus, in part due to the lack of commercial interest in developing such products and in part due to the lack of knowing what constitutes a protective immune response. Many workers believe a strong cell-mediated response is required but as this would most likely require the use of live attenuated virus, this raises questions as to possible reversion of any vaccine strain to virulence.

Greater success has been achieved with antivirals. In one study of patients with a poor prognosis treated for 10 days with intravenous ribavirin (60–70 mg/kg/per day), begun within six days after the onset of fever, showed a reduced case fatality of 5% (McCormick et al. 1986). In contrast, patients who began treatment seven or more days after the onset of fever had a case fatality rate of 26%. In the Sierra Leone study, viraemia of greater than 10^3.6 TCID₅₀/ml on admission was associated with a case fatality rate of 76%. Patients with this risk factor who were treated with intravenous ribavirin within six days of the onset of fever had a case fatality rate of 9%, compared with 47% in those treated seven days or more after the onset of illness.

As with many examples of severe haemorrhagic disease, effective control requires early and strict isolation of cases coupled with rigorous disinfection and rodent control. Disinfection with 0.5% sodium hypochlorite or 0.5% phenol in detergent is recommended for surfaces and instruments. Owing to the high burden of virus replication, infection of relatives and others who have been in close contact is a distinct possibility and WHO guidelines suggest that non-casual contacts should be followed for up to 3 months for signs of illness. The taking of body temperature thrice daily is recommended, followed by hospitalization if the body temperature exceeds 38.5°C.

The emergence of Lujo virus in 2008 demonstrates how novel arenaviruses continue to emerge. The index case was a female resident of Zambia living in a semi-rural location where she kept horses, dogs and cats. She fell ill with diarrhoea and vomiting before being evacuated to a private hospital in South Africa where she died 12 days after onset of illness. Three hospital personnel—two nurses and one cleaner—also became infected, succumbing to the disease within two weeks of exposure to infection. An additional nurse was also infected but eventually recovered (Paweska et al. 2009).

Detailed clinical descriptions of these cases showed patients with signs typical of a viral haemorrhagic fever. The index case showed in particular a whole body rash accompanied by facial oedema, a feature frequently seen in cases with South American haemorrhagic fevers. Cerebral oedema and respiratory distress quickly followed. A marked thrombocytopenia, granulosis and elevated liver transaminases are all features common to viral haemorrhagic fevers. The single surviving case is notable in that this nurse received ribavirin within 24 hours after onset of illness, but the recovery was prolonged with levels of virus in the blood declining slowly over two months.

Phylogenetic analyses has confirmed that Lujo virus is an Old World arenavirus, although it its lineage suggests a close relationship with an ancestral virus rather than to Lassa, LCM and other Old World arenaviruses (Briese et al. 2009).

Argentine haemorrhagic fever has been known since 1943 and Junín virus, the causative agent, was first isolated in 1958. The virus causes annual outbreaks of severe illness—with between 100 and 3,500 cases—in an area of intensive agriculture known as the wet pampas in Argentina. Mortality in some outbreaks has ranged from 10% to 20%, although the overall mortality is generally 3–15% unless supportive therapy is provided early in the course of the disease.

After an incubation period of 7–16 days, the onset of illness is insidious, with chills, headache, malaise, myalgia, retro-orbital pain and nausea; these are followed by fever, conjunctival injection and suffusion and an enanthem, exanthema and oedema of the face, neck and upper thorax. A few petechiae may be seen, mostly in the axilla. There is hypervascularity and occasional ulceration of the soft palate. Generalized lymphadenopathy is common. Tongue tremor is an early sign, and some patients present with pneumonitis. In the more severe cases the patient’s condition becomes appreciably worse after a few days, with the development of hypotension, oliguria, haemorrhages from the nose and gums, haematemesis, haematuria and melaena. Oliguria may progress to anuria and pronounced neurological manifestations may develop. Laboratory findings have included leucopenia with a decrease in the number of CD4-positive cells, thrombocytopenia and urinary casts containing viral antigen. Patients recover when the fever falls, followed by diuresis and rapid improvement. Death may result from hypovolaemic shock. Subclinical infections also occur. Man-to-man transmission has not been observed.

Elevated levels of interferon can be detected in the early stages of Argentine haemorrhagic fever, and these coincide with the onset of fever and backache. Although there is no correlation between the titres of interferon and circulating virus, Levis and colleagues (1984) have suggested that at least some of the clinical signs may be directly attributable to interferon, particularly the depression of platelet and lymphocytic numbers that result from Junín virus infection of leucocytes and macrophages.

Argentine haemorrhagic fever has a marked seasonal incidence, coinciding with the maize harvest between April and July, when rodent populations reach their peak. Agricultural workers, particularly those harvesting maize, are, not surprisingly, the most commonly affected.

The main reservoir hosts of Junín virus are Calomys field voles that live and breed in burrows under the maize fields and in the surrounding grass banks. Other rodent species may also be infected. Calomys spp. has a persistent viraemia and viruria, and virus is also present in considerable quantities in the saliva. The mode of transmission of Junín virus to man has not been conclusively established. The virus may be carried in the air from dust contaminated by rodent excreta or may enter by ingestion of contaminated foodstuffs.

In contrast to Lassa fever, antibodies play a major role in recovery from Junín infection. Controlled trial of immune plasma collected from patients at least 6 months into convalescence have shown a dramatic reduction in mortality if plasma is given within the first 8 days of illness (Maiztegui et al. 1979). The efficacy of this therapy is directly related to the titre of neutralizing antibody in the plasma; as a result, a dose of no less than 3,000 ‘therapeutic units’/kg body weight has been recommended (Enria et al. 1984).

The late development of a neurological syndrome is seen in up to 10% of patients treated with immune plasma; it is often benign and self-limiting but points to the possible persistence of viral antigens on cells of the CNS well into convalescence. Treatment with immune plasma also restores the response of peripheral blood lymphocytes to antigenic stimuli, suggesting that the administration of plasma also results in the modulation of cellular immunity.

Of interest is the finding of Andes virus, a member of the hantavirus genus of the family Bunyaviridae, among the cohorts presenting with presumptive Argentine haemorrhagic fever but found to be serologically negative for Junín virus. Were it not for the activities of the investigators primarily working with cases of haemorrhagic disease, this new cause of acute respiratory disease would most likely have remained unrecognized.

There have been several attempts to produce a vaccine against Argentine haemorrhagic fever. The XJ-Cl₃ strain of virus grown in the brains of suckling mice is relatively non-pathogenic and was administered to 636 volunteers between 1968 and 1970. However, the vaccine often induced a mild febrile reaction or a subclinical infection, and its use was discontinued despite the fact that over 90% of vaccinees maintained neutralizing antibody for up to nine years.

A second vaccine was developed using a pedigree strain of Junín virus extensively passaged in cells and plaque purified. This ‘candidate 1’ vaccine has been administered to over 280 000 individuals in the endemic regions of Argentina with greater than 95% protection. As a result, the number of clinical cases has declined sharply to less than 100 per year. Protective efficacy has been hard to judge, however, owing to annual variations in rodent numbers and, more importantly, the prevalence of infection among the local rodent populations.

Bolivian haemorrhagic fever was first recognized in 1959 in the Beni region in north-eastern Bolivia. The most notable outbreak affected 700 people in the San Joaquin township between late 1962 and the middle of 1964. The mortality was 18%. Originally referred to locally as ‘black typhus’ this disease, predominantly of males, erupted at a time of abnormally low rain fall combined with a decline in the cat population. The resulting explosion in number of Calomys callosus increased vastly the risk to humans: rodent trapping in half the township led to a precipitate drop in the number of new cases. The disease continued in that region more or less annually for a number of years in the form of sharply localized epidemics. Its incidence has decreased considerably since the late 1970s and human infections are now rarely reported. The mortality in individual outbreaks varied from 5% to 30%. It is worth noting that the discovery of a common morphology and serological cross-reaction between Machupo and LCM virus led to the concept of the arenavirus family.

The clinical disease is similar to Argentine haemorrhagic fever. The incubation period ranges from 7–14 days and the onset is insidious. About one-third of patients show a tendency to bleed, with petechiae on the trunk and palate, and bleeding from the gastrointestinal tract, nose, gums, and uterus. Almost half the patients develop a fine tremor of the tongue and hands, and some may have more pronounced neurological symptoms. The acute disease may last 2–3 weeks and convalescence may be protracted, generalized weakness being the most common complaint. Clinically inapparent infections are rare.

Machupo virus, the responsible agent, is readily isolated from lymph nodes and spleen taken at necropsy. Isolation of the virus from acutely ill patients has proved difficult, however, the best results being obtained from specimens taken 7–12 days after the onset of illness.

The rodent reservoir of Machupo virus is Calomys callosus; over 60% of this species of vole caught during the San Joaquin epidemic were found to be infected. The distribution of cases in the township was associated with certain houses and Calomys callosus was trapped in all households where cases occurred. Transmission to man is probably by contamination of food and water or by infection through skin abrasions. Transmission from man to man is unusual but a small episode took place in 1971, well outside the endemic zone. The index case, infected in Beni, carried the infection to Cochabamba and, by direct transmission, caused five secondary cases, of which four were fatal. Abnormally low rainfall, combined with an increase in the use of insecticide, led to a rapid decline in the numbers of cats, with the result that the population of Machupo-infected rodents increased dramatically thus increasing the opportunity for human contact with contaminated soil and foodstuffs. This balance has since been restored and largely accounts for the reduction in the number of reported cases over the past two decades.

Recently a second cause of haemorrhagic disease in Bolivia has been ascribed to an unrelated arenavirus, tentatively called Chaparé virus (Delgado et al. 2008). Worryingly this new agent shows no serological cross-reactivity with Machupo virus; phylogenetically Chaparé virus is more closely related to Sabiá virus.

There is little documented evidence that Bolivian haemorrhagic fever can be successfully treated, although laboratory studies would suggest this is possible. That animals can be partially protected against Machupo virus by immunization with Junín virus vaccine also suggests vaccination may be successful but has yet to be investigated by means of a clinical trial.

This agent was first described in 1989, with 26 deaths being recorded among 105 cases originally suspected as being dengue infections (Salas et al. 1991). Most of the cases have been adults and all from the state of Portuguesa in the mid-western part of Venezuela. Lasting from 3 to 12 days, the infection is typified by fever, sore throat, nausea with vomiting, and other symptoms associated with arenavirus infections in the New World. Up to 90% of the patients showed a marked thrombocytopenia and leucopenia. Post-mortem examination of the fatal cases revealed extensive haemorrhage in the lungs and liver accompanied by cardiomegaly, splenic enlargement, and congestion of the lungs. Oedema of the kidneys was also observed, together with blood in the intestines and bladder.

The virus has since been isolated repeatedly from the cotton rat Sigmodon alstoni, although there have been no significant recorded cases since the original outbreak. The route of transmission remains unclear, with human-to-human spread being rare. The infections are likely to have been acquired peridomestically, however, as in the study of Salas et al. virus was recovered from a rodent trapped in the house of one case. The relatively high mortality of the infection parallels that seen in the early reported cases of Machupo and Junín infections; in the event of further outbreaks this level should be reduced as diagnosis improves and appropriate treatment instigated earlier in the course of the disease.

Isolated in 1990 from human cases at autopsy, Sabiá virus is thought to infect rodents in the agricultural regions surrounding Sao Paulo. The potential seriousness of this infection is highlighted by a laboratory worker at Yale University becoming critically ill after an accidental exposure to an aerosol containing Sabiá Virus. The individual developed a febrile illness accompanied by a marked leucopemia and thrombocytopenia. As is often the case with emerging haemorrhagic diseases, involvement of the liver suggests initially a case of yellow fever but this can quickly be ruled out. The recent emergence of the phylogenetically-related Chaparé virus some 1,000 km away in Bolivia is a stark reminder that distribution of these pathogenic viruses may be more extensive than is currently appreciated.

This arenavirus has been isolated from the field rodent Bolomys obscures trapped within the endemic region for Argentine haemorrhagic fever (Bowen et al. 1996). There is as yet little indication that Oliveros virus causes human disease, although approximately 25% of captured Bolomys rodents have been found to contain antibodies to the virus (Mills et al. 1996).

The 1993 hantavirus outbreak in the ‘Four Corners’ region of the USA stimulated extensive studies of feral rodent populations in order to measure the extent of Sin Nombre virus distribution and the corresponding risk to rural inhabitants. During one such study, an unusually high level of arenavirus antibodies was detected among trapped pack rats (Neotoma spp) found in the Whitewater Arroyo of New Mexico (Kosey et al. 1996), similar findings were reported by Fulhorst and collaborators (1996) from trapped White-throated Woodrats (N.albigula). Fulhorst and colleagues showed that this new arenavirus can be passed through the urine and thus could pose a threat if excreted in recreational areas and around isolated households. Members of the Neotoma family are ubiquitous throughout the south west USA and some evidence of human infection has been obtained from three female patients presenting with acute respiratory symptoms and non-specific febrile symptoms. All three of the latter patients died within 1–8 weeks of onset: although evidence of Whitewater Arroyo virus was obtained by PCR and by virus isolation in one case, the link between Whitewater Arroyo virus and human disease has yet to be conclusively proven.

There have also been further examples of arenavirus infection among feral rodents of the USA. Infectious virus was recovered from 5 of 27 examples of the Californian mouse Peromyscus calfornicus caught in the Santa Ana mountains close to the Bear Canyon trailhead. Catarina virus is another arenavirus associated with the southern plains woodrat, Neotoma micropus, found in southern Texas (Cajimat et al. 2007) but there is no evidence as yet that either of these newly discovered arenaviruses causes human disease.

Unique among the viral zoonoses, the arenaviruses show a host-parasite relationship which has received intensive study. In the case of LCM virus, this has resulted in the discovery of fundamental concepts in viral immunopathogenesis and clearance. Yet much remains to be learnt, particularly from the standpoint of public health. The emergence of three new arenaviruses potentially capable of causing serious human disease since the first edition of this book illustrates vividly the need for public health microbiologists to be ever vigilant for hitherto unknown agents causing unexpected outbreaks. The epidemiology of almost all arenaviruses remains poorly understood; for example, Lassa is clearly widespread among the rural areas of West Africa, but in contrast to South America there is an inexact correlation between the distribution of infected rodents and human infections. There is also much to be learnt in terms of the susceptibility of the natural hosts to infection; a rodent of the Calomys family wild-caught in Venezuela, for example, may be refractory to infection whereas its cousin from elsewhere in South America can be readily infected. It is tempting to speculate that arenaviruses are instrumental in controlling rodent population numbers and that only when man radically alters the rodent habitat do zoonotic infections result. Thus there is ample scope for further studies of the natural history, epidemiology and pathology of this unique and fascinating group of viruses. By such work, we may better understand the host-parasite relationship of these agents and thus be better prepared for preventing further outbreaks of severe and debilitating human infections.