The recognition of hantavirus pulmonary syndrome (HPS) after the investigation of a cluster of unexplained respiratory deaths in the southwestern United States during the spring of 1993 showcased our ability to recognize new and emerging diseases, given the correct juxtaposition of a new clinical entity with circumscribed epidemiologic features that are analyzed with novel diagnostic methods. In less than a decade, HPS has become established as a pan-American zoonosis due to numerous viruses maintained by sigmodontine rodents with rodent- and virus-specific epidemiologic profiles. The classical features of the syndrome—acute febrile illness associated with prominent cardiorespiratory compromise after direct contact or inhalation of aerosolized rodent excreta—has been extended to include clinical variants, including disease with frank hemorrhage, that have confirmed that this syndrome is a viral hemorrhagic fever. Efforts are under way to refine prevention strategies, to understand the pathogenesis of the shock, and to identify therapeutic modalities.
In 1993, a “mystery” disease intruded into rural communities of the southwestern United States. Major efforts by private doctors, academia, state and local health departments, and the Centers for Disease Control and Prevention (CDC) framed the problem: young, healthy adults were dying of an infectious disease that had a high case-fatality rate, and the cause did not yield to any of the methods brought to bear during a proficiently executed and intensive investigation. Among those who participated in the public health response were a small group of investigators with experience with appropriate reagents and expertise in hantaviruses. Their experience with hemorrhagic fever with renal syndrome (HFRS) in Eurasia and with existing immune reagents was instrumental in obtaining the initial serologic clues that the etiologic agent of the disease was a hantavirus. Furthermore, the research orientation of this group of investigators, which was based in a public health institution, made it possible to confirm these findings by means of molecular biologic methods and modern infectious-disease pathology. The existence of new clinical paradigms (and, indeed, hantavirus pulmonary syndrome, or HPS, is a “new” disease) and specific laboratory tests has allowed rapid progress in our understanding of hantavirus disease in the Americas. Recent findings regarding the clinical picture, epidemiology (human and rodent), pathogenesis, and virology of this entity support the conclusion that the syndrome is a new American viral hemorrhagic fever.
What are Hantaviruses, and Where are they Found?
Hantaviruses are members of the genus Hantavirus of the family Bunyaviridae . They share similar replication strategies and certain genomic features, including phylogenetic relationships. Like other Bunyaviridae, they have 3 RNA segments and a lipid envelope, but, in contrast to the other members of the family, who usually have arthropod vectors, hantaviruses generally chronically infect rodents. Hantaviruses are found in Asia, Europe, and the Americas wherever rodents of the family Muridae are present, and they have been identified in rodents of the subfamilies Murinae, Arvicolinae, and Sigmodontinae (table 1; figure 1). Murinae comprise the largest number of Old World rodents and are the reservoirs for Hantaan, Dobrava, and Seoul viruses, which cause HFRS, and other viruses that are not associated with human disease. Arvicolinae are voles and are the reservoirs for such viruses as Puumala virus, which has been associated with mild HFRS in Europe, and Prospect Hill virus and related viruses in the United States, which have not. Sigmodontinae are the largest group of New World rats and mice and are the hosts of Sin Nombre virus (SNV) and numerous other viruses that are associated with HPS or that lack disease associations. The presence of hantaviruses has not yet been convincingly demonstrated in African Muridae for reasons that are not clear, and their absence from Australia and Antarctica is presumably the result of the lack of native Muridae. Thus, HPS seems to be a disease of the Americas because of some unexplained property of viruses associated with the sigmodontine American rodents.
The evolutionary split between Murinae and Sigmodontinae presumably dates to the divergence that occurred between subfamilies 30 million years ago, when the precursors of the sigmodontine rodents crossed the Bering Strait into the Americas. Peromyscus species are the principal rodent hosts for the major North American viruses, particularly SNV and its 2 close relatives, Monongahela virus and New York virus; as of September 2001, µ300, 1, and 2 cases of HPS, respectively, had been attributed to these viruses in the United States and Canada. Bayou virus (responsible for 3 cases of HPS) and Black Creek Canal virus (1 case of HPS) are more distantly related to the SNV-Peromyscus viruses but are closely related to each other, as one would expect on the basis of the relations among the rodents. In South America, the viruses are quite numerous and often closely related at the amino acid level, and their patterns of diversity are more complex (table 1). This large number of virus-host pairs presumably reflects the rapid and relatively recent evolution of rodents in that region. Disease has been identified from numerous Central American and South American countries, but it is increasingly apparent that most, if not all, sigmodontine rodents in the Americas will be found to have a characteristic hantavirus, that many of these viruses will cause HPS, and than HPS will be identified in every country from Canada to Chile.
The causative viruses are widely distributed and cause a syndrome that is readily recognizable as HPS, which is characterized by a distinctive prodrome and pulmonary infection, cardiac depression, and hematologic manifestations. SNV infection presents the clearest syndrome, with minimal clinical manifestations outside the thoracic cavity. In some of the other American hantavirus infections, extrathoracic findings are seen more often than they are in SNV infection. Renal involvement is often present but rarely is a major consideration, just as increased pulmonary vascular permeability is often found in HFRS but does not commonly challenge the clinician in properly managed patients . The predominance of infection of pulmonary endothelial cells and the major role of pulmonary involvement in the clinical presentation initially led a consortium of scientists to name the condition “hantavirus pulmonary syndrome,” and no new findings have suggested that this appellation be changed.
HPS due to SNV. The original description of disease due to SNV infection  has been extended to include documented asymptomatic and mild infections that do not result in radiographic pulmonary compromise . The prodrome typically lasts 3–5 days, during which the patient has myalgia, malaise, and fever of abrupt onset. Onset of symptoms occurs a median of 14–17 days after a well-defined exposure . During the prodrome, cough and coryza are absent, but anorexia, nausea, vomiting, and abdominal pain may begin. The intensity of the syndrome will often cause patients to seek medical aid, but the absence of any particular finding indicating the need for hospital admission will often lead clinicians to give palliative therapy and send the patients home. According to the HPS registry (an unpublished record maintained by the CDC, Atlanta), approximately one-half of patients seek help from a doctor at least once before admission to the hospital. However, if there is sufficient clinical suspicion and a simple blood smear is performed, the presence of the syndrome may be suggested by the finding of thrombocytopenia.
At the time of hospitalization, patients are usually entering the next stage of disease, which is characterized by cardiopulmonary involvement. At this stage, cough is generally present, and gastrointestinal manifestations may dominate the clinical picture. Patients will often have tachypnea, tachycardia, and postural hypotension. Findings of a chest examination usually are not impressive, with a few rales. Results of clinical laboratory tests alter as the cardiopulmonary phase begins and become markedly abnormal. The first laboratory test to show abnormal results, and the most useful test, is the platelet count [6, 7]. A decreasing or low platelet count is reason for hospitalization for further observation; in other illnesses that have prodromes resembling that of HPS, thrombocytopenia is rare, and some of these diseases, such as rickettsial infections, plague, tularemia, and relapsing fever, may in themselves require inpatient treatment. Other critical laboratory findings include the presence of circulating immunoblasts, which are often interpreted to be atypical lymphocytes, and an elevated hematocrit. At the time of hospitalization, patients with HPS more commonly have a peripheral blood smear that reveals myelocytes, metamyelocytes, or promyelocytes and have significantly more severe thrombocytopenia, hemoconcentration, and hypocapnia than do patients with suspected HPS that is subsequently excluded .
Other abnormal test results may include a decreased serum sodium level, a usually mildly elevated serum glutamic oxaloacetic transaminase level, a decreased protein level, a usually mildly elevated low-density lipoprotein level, and a normal or slightly prolonged activated partial thromboplastin time. The mean serum creatinine level is µ2.0 mg/dL in 25% and µ3.0 mg/dL in 10% of patients with HPS. One-fourth of patients with HPS have a maximum urinary protein level of ≥3+ (as measured by use of a urine dipstick), and microscopic examination often reveals microscopic hematuria with ≥10 cells per high-power field.
Findings of pulmonary evaluation will be distinctly abnormal at this time, with low PO2 level or low pulse oximetry findings and, often, hypocapnia. In severe cases, there will be metabolic acidosis with an elevated serum lactate level. Almost all patients have chest radiographs with abnormal findings indicative of interstitial edema: specifically, Kerley B lines, hilar indistinctness, or peribronchial cuffing with normal cardiothoracic ratios. Approximately one-third of patients have evidence of air space disease on the initial radiograph. By 48 h after admission to the hospital, virtually all patients will have evidence of interstitial edema, and two-thirds will have developed extensive airspace disease that is initially bibasilar or perihilar, with some degree of pleural effusions. The initial lack of peripheral air space disease, the prominence of interstitial edema, and the presence of pleural effusions in the early stages of disease are in contrast to the typical radiographic findings in acute respiratory distress syndrome of other etiologies .
It is essential to recognize and hospitalize patients with such findings for observation and therapy. There is usually a rapid decompensation, and 30%–40% of patients die 24–48 h after admission, even in well-run intensive care units (ICUs). Approximately 40% of patients will not require intubation and may be managed with judicious administration of fluids and careful monitoring. The remainder will pose a difficult problem of management. In addition to the marked increase in pulmonary capillary permeability, there will usually be primary myocardial dysfunction and high systemic vascular resistance. The balance between, on the one hand, maintaining high normal filling pressure (to maintain cardiac output) and, on the other hand, minimizing pulmonary edema in the face of positive end-expiratory pressure ventilation and other necessary procedures is best achieved by a skilled and experienced ICU staff. Early use of inotropic drugs, such as dobutamine, is encouraged . Other therapeutic interventions, such as intravenous administration of ribavirin, have had disappointing results, and the initial enthusiasm for extracorporeal membrane oxygenation for salvage therapy has been limited by the lack of a controlled trial or follow-up data from anecdotal reports of iatrogenic deaths and limb amputations . However, there are promising anecdotal reports of the use of supraphysiologic doses of corticosteroids in South America, and there are lessons to be learned from the treatment of the idiopathic systemic capillary leak syndrome (Clarkson's disease) .
Patients who die often experience disseminated intravascular coagulation, including frank hemorrhage and exceptionally high WBC counts [6, 7]. However, the key determinant, according to logistic-regression analysis, is the degree of thrombocytopenia. Survivors usually improve within a few days and undergo extubation in 7 days; they can be released from the hospital in 17 days even if they undergo intubation. Patients who survive usually show no major effects. Follow-up studies have detected defects in decreased small airways flow, increased residual volume, and decreased oxygen diffusion capacity, as well as more subjective complaints, including fatigue, myalgia, and shortness of breath. These symptoms may well be a consequence of hantavirus infection or the cytokine response, but the clinician must also consider the possible role of ICU therapy and administration of large amounts of oxygen to treat hypoxia . In addition, cognitive defects, perhaps also attributable to hypoxemia and intensive care, have been noted in survivors . Some patients who died after a relatively long stay in the ICU had histologic evidence of pulmonary fibrosis . Perhaps this is not unexpected, given that the cytokines responsible for the permeability defect are the same molecules involved in inflammation and fibrosis in other tissues.
HPS due to other American hantaviruses. Renal disease and myositis appear to be more common in cases of Bayou virus and Black Creek Canal virus infection in the southeastern United States than in SNV infection. Disease in South America appears to follow a similar pattern, with spot estimates of 40% seroprevalence in select populations and a greater prevalence among children . A systematic review of Andes virus infection in Argentina documented marked conjunctival injection, facial flushing (rubicundez, in Spanish; this symptom is also found in HFRS), pharyngeal congestion, and petechiae. All infected patients had proteinuria, 10% required hemodialysis, and one-half had aminotransferase levels 5–10 times the normal value . There are also accumulating reports of severe gastrointestinal bleeding and frank hemorrhagic shock confused with dengue hemorrhagic shock syndrome (unpublished data, CDC).
Although HPS represents a new syndrome with characteristic features, the final diagnosis can be made only on the basis of results from a number of different laboratory tests. By far the most valuable and widely used test is the IgM capture ELISA, which detects IgM in all acute cases . The format of the test, particularly if used with a negative antigen control for each serum sample, is highly reliable. An irradiated or recombinant SNV antigen can be used (one is provided free of charge to domestic and international pubic laboratories by the CDC); all cases of HPS in the Americas have been detected because of the broad cross-reactivity these reagents in different test formats, with well-conserved specificity for hantaviruses. In addition, the immunopathologic nature of the disease results in almost invariable positive results, often including results positive for IgG antibodies, even during the prodromal phase.
Levels of IgG antibodies are usually detectable during acute disease, peak during the first week of illness, and remain detectable for a long time . The low prevalence of hantavirus antibodies in the US general population makes measurement of antibody levels of value in acute disease diagnosis and population studies. However, if the prevalence of antibodies in the normal population has not been assessed, results of antibody testing can be misleading. For example, in Paraguay, a considerable seroprevalence of hantavirus antibodies, as high as 40%, is noted in people without any history of HPS . In this circumstance, the broad reactivity of the SNV antigens is a disadvantage, because there are 2 interpretations of the results: (1) there are multiple hantaviruses that are causing infection, and at least 1 is of low pathogenicity; or (2) there is a single hantavirus that is much less pathogenic than is SNV.
Results of reverse-transcriptase PCR performed on acute-phase serum specimens during the first 10 days of illness are also usually positive , provided sensitive primers and a nested reaction are used. This assay is useful because it identifies the infecting virus genotype, and sequencing of the PCR product (which is recommended to exclude contamination, in any case) may be helpful epidemiologically. It is worth emphasizing that one commonly used diagnostic modality, virus isolation, is not very useful for the study of possibly infected patients because of the low yield, even though a high level of viremia, correlated with hemoconcentration and the degree of thrombocytopenia, which promptly decreases with resolution of fever, is easily demonstrable by quantitative PCR . Genetic comparisons between patient tissues, rodent tissues, and virus isolates leave no doubt that the hantaviruses linked with human disease are indeed the causative agents, but the difficulty of propagating them continues to be a biological riddle.
Epidemiology in Humans
HPS due to SNV. The studies of the 1993 epidemic provided the basic definition of the epidemiology of HPS caused by SNV . It is a rural disease, with a few cases associated with suburban habitats where people were exposed to Peromyscus species. A preliminary analysis of data accumulated from patients with HPS around the country suggests that 10% of infections are acquired around the workplace, 5% during recreation, 50% in or around the home, and the remainder from mixed or unknown exposures . Often the exposures are obvious and egregious, involving imprudent behavior or massive rodent exposure. For some patients, the actual exposure is subtle and not readily ascertained without a close study of rodent sign and rodent traffic at the sites potentially involved. One interesting and important observation has been that a frequent antecedent of HPS is opening and inhabiting a long-unused cabin. This may be related to a combination of factors: entry disturbs deer mice, which often urinate as they flee; the closed cabin prevents dilution of aerosols by breezes; and the roof prevents inactivation of virus in aerosols by the strong ultraviolet component of sunlight.
HPS due to other American hantaviruses. The 2 close relatives of SNV carried by Peromyscus maniculatus or Permyscus leucopus have epidemiologic features similar to those of SNV (table 1). In addition, infection with New York virus has appeared in suburban neighborhoods, particularly in areas with a high incidence of Lyme disease. The few cases of Monongahela virus infection have been noted among hikers and persons who engage in other outdoor activities. Sporadic HPS cases have been associated with 2 other hantaviruses. Black Creek Canal virus is found among Sigmodon hispidus in southern Florida, where this rodent often frequents grassland with scattered brush. Bayou virus infects Oryzomys palustris, which is a rodent often found in areas with standing water, including drainage ditches and swampy habitats.
The overall pattern for the South American viruses has been similar to that of viruses in the United States: most persons who get infected work or live in rural areas or small towns in rural areas. Recreational visitors to natural habitats or rural areas have also been infected; one such patient exported a case to Europe. Increased transmission of the virus is linked to unusual weather conditions; it often occurs after heavy rainfall. The most perplexing difference between the pattern in South America and elsewhere is the unequivocal person-to-person transmission of Andes virus in Argentina and Chile ; there is no evidence of a similar propensity among the other hantaviruses, despite a thorough review of documented clusters in the United States, nor, for that matter, is there evidence of such a pattern of transmission among the ∼200,000 HFRS infections that occur each year in Eurasia. Elucidation of the mechanism of transmission may prove to be the Rosetta stone of HPS pathogenesis.
Virology and Pathogenesis
A continuing mystery is why hantaviruses are so difficult to isolate in cell culture. They generally require serial passage in a cell substrate before they infect an appreciable percentage of the cells, grow to a significant titer, or produce plaques under agar.
There is a better outline of the pathogenesis in humans, although many questions remain. Both HFRS and HPS are thought to be immunopathologic [1, 15] and share several important similarities in their pathogenesis. The process of human disease begins with inhalation of a small particle that bears infectious virus and deposition of the particle in a terminal respiratory bronchiole or alveolus. Then, possibly through infection of alveolar macrophages or other primary targets, a viremia is generated that results in widespread infection of pulmonary capillary endothelium and lesser degrees of infection of other cells in the body . Cellular entry of some hantaviruses is mediated by β3 integrins, which are critical adhesive receptors on endothelial cells and platelets that regulate vascular permeability and platelet activation and adhesion . Hantavirus infection usually induces little, if any, disturbance in cellular function or overt histologic damage; in vitro, cell monolayers must be tested for the presence of viral antigen or RNA to detect infection. Indeed, in vitro, infected pulmonary microvascular endothelial cells, if additional mediators are absent, maintain a normal permeability .
The period of invasion presumably results in induction of IFN-α, which may well be responsible for the prodromal manifestations. Later, when the immune response is well under way, immunoblasts are found in the peripheral blood; these are CD4+ and CD8+ T cells that are also DR positive . T cell clones derived from peripheral blood at this stage of infection have been shown to be hantavirus specific . These cells presumably enter the lung, where large numbers of T immunoblasts are found in the interstitium, in conjunction with activated macrophages . On the basis of in vitro studies of SNV-infected microvascular capillary endothelial cells, the lung capillaries are normally permeable at this time. However, the secretion of soluble mediators, such as TNF-α, nitric oxide, and IFN-γ, have profound effects on capillary endothelial permeability, which, in turn, leads to the pulmonary edema that precipitates the patient's admission to the hospital. The increase in permeability is so profound that, in severe cases, the lung edema fluid can approximate plasma in its protein concentration and electrophoretic pattern . Further evidence for this mechanism is found in the peripheral blood, where high concentrations of cytokines are found, including IFN-γ, IL-4, IL-6, and soluble receptors for TNF-α and IL-2 . Coincidentally, high-dose IL-2 therapy for resistant cancer can reproduce a toxicity syndrome that closely mimics HPS .
In addition to the severe cytokine response, hantaviruses are readily neutralized in vitro, and indeed, neutralizing antibodies are present in the serum of patients at admission to the hospital. The hantavirus-specific RNA level in serum is decreasing, presumably representing clearance of virus or viral nucleocapsid immune complexes. Interestingly, higher titers of neutralizing antibodies at admission have been correlated with increased likelihood of survival [17, 25].
In addition to the severe permeability defect in the lung, myocardial depression is common, despite normal findings of microscopic and immunohistochemical examination, which suggests that soluble mediators are the basis for the cardiogenic portion of the shock [6, 7, 10]. This pattern of mixed hypovolemic shock (i.e., internal fluid shift) and cardiac shock with high peripheral vascular resistance is found (unlike classic septic shock) in HFRS, dengue hemorrhagic fever, and perhaps other hemorrhagic fevers [26, 27]. The cause is unknown, but circulating TNF-α is a candidate. The remarkable finding that Andes virus causes typical HPS in hamsters  provides an opportunity to study the disease process of either HPS or HFRS for the first time.
Rodent infection and rodent behavior intersect with human behavior to cause human hantavirus infection. P. maniculatus, for example, is an ecologic generalist, with no inhibitions about entering human habitations. We have seen that it is possible to predict the occurrence of the ecologic conditions that are associated with a high risk of hantavirus infection by monitoring critical biomes, such as piñon-juniper forest, and prospectively predicting more precise areas of high risk on the basis of information obtained by satellites. At least in the southwestern United States, where rodent population numbers are limited by the pattern and availability of precipitation, we can predict times of extraordinary risk on the basis of climate predictions (e.g., if an El Niño event is expected) and follow these predictions with measurements of the number of infected deer mice that are placing the human population at hazard. But what can we do about human infection? One obvious approach is to minimize rodent-human contact. Unfortunately, this approach is not feasible, because up to one-half the infections occur outside the peridomestic setting. CDC recommendations call for the rodent-proofing of homes, the reduction of rodent cover around houses, the minimization of food available for rodents, the trapping of rodents in and around dwellings, and the careful disposal of dead rodents .
The lesson for clinicians is that HPS can be caused by a large number of closely related viruses found in American rodents from the far north to Tierra del Fuego. These syndromes are sufficiently similar to be recognizable as HPS, and they can all be diagnosed in a timely and sensitive fashion by an IgM ELISA that uses SNV antigens. The major clinical manifestations are a major pulmonary edema accompanied by cardiac depression. These febrile prodromes are also associated with early thrombocytopenia and a left shift in the differential count. Some of the South American viruses have been associated with more extrathoracic manifestations, including hemorrhage, which have further established HPS as a viral hemorrhagic fever. Because of the diversity of rodent hosts, various activities can result in exposure to hantaviruses, but, in general, a major risk factor is residence in or a visit to a rural area, perhaps in ecological and other conditions that have led to rodent proliferation or the movement of rodents into structures.
Human disease is induced by the host immune response, and no therapy specific for hantaviruses or their pathogenic mediators is available. Education of the public and of receiving doctors can help by encouraging early hospitalization and supportive care. Prevention can be partially achieved by rodent avoidance, but real protection will require a vaccine or a host-modification strategy.
To all those who toil day in and day out in obscurity on infectious diseases that do not currently occur in the United States, even though they often represent profound international public health problems, we thank you for remembering that science has no boundaries. We thank Amera Khan, Joni Young, and John O'Connor for editorial assistance.