The immunologic responses that mediate viral clearance of and recovery from hantavirus cardiopulmonary syndrome (HCPS) due to Sin Nombre (SN) virus are unknown. Serial serum samples from 26 patients with acute SN virus infection were tested for IgG, IgA, and IgM reactivity to recombinant viral nucleocapsid (N) and glycoprotein G1 antigens by a novel strip immunoblot assay. The titers of antibodies capable of neutralizing SN virus in vitro also were determined for each sample. At admission, patients with severe disease had lower titers of IgG antibodies to SN virus N antigen (P < .033) and lower neutralizing antibody titers (P < 3.4 × 10−5), compared with patients with mild disease. These data suggest that a strong neutralizing antibody response may be a predictor of effective clearance of and recovery from SN virus infection and raise the possibility that passive immunotherapy may be useful in HCPS.
Hantaviruses are rodentborne, negative-strand RNA viruses of the family Bunyaviridae [1, 2]. At least 10 distinct members of the genus Hantavirus have been associated with hemorrhagic fever with renal syndrome or hantavirus cardiopulmonary syndrome (HCPS) in humans [1, 2]. Sin Nombre (SN) virus, the prototype etiologic agent of HCPS, is carried by the deer mouse, Peromyscus maniculatus [3–6]. Although there are 3 other viruses in North America and ⩾2 in South America, SN virus accounts for most of the >250 known North American cases [1, 2] (J. Young and H. Artsob, personal communication).
Transmission of SN virus is thought to occur primarily via inhalation of contaminated aerosols of rodent urine, feces, or saliva . The incubation period is ∼8–20 days. The prodromal phase of HCPS lasts 1–6 days and consists of fever, myalgia, headache, malaise, gastrointestinal disturbances, and thrombocytopenia . Most patients develop acute pulmonary edema, hypotension, and shock . In practice, most patients with HCPS are recognized as such and are admitted to a hospital on the first day of pulmonary edema. Death, when it occurs, usually comes 1–3 days after the onset of respiratory symptoms and is usually caused by cardiogenic shock. Since most deaths are caused by cardiac depression rather than by hypoxia, many investigators have begun to refer to the syndrome as HCPS, rather than the older term hantavirus pulmonary syndrome. Recently, it has been recognized that a small subset of patients with acute SN virus infection manifest the symptoms of only the prodromal phase of HCPS .
All persons with acute SN virus infection have detectable antibodies against the SN virus nucleocapsid (N) antigen of the IgM class by the onset of clinical symptoms, and almost all have IgG antibodies directed against the N and Gl antigens . Little is known about the kinetics of antibody responses, and almost nothing is known of the neutralizing antibody responses in patients with HCPS. We sought to determine the relationship, if any, of specific antibody responses to the outcome of infection in patients with mild-to-severe infection with SN virus.
Subjects and Methods
The case patients with SN virus infection were diagnosed at TriCore Laboratories (Albuquerque) and/or were treated at the University of New Mexico Hospital. Patients were considered to have acute SN virus infection on the basis of the presence of IgM and IgG class antibodies directed against the SN virus N antigen and the presence of IgG antibodies against the viral G1 antigen. The IgG response to the G1 antigen is highly specific for SN virus infection, as opposed to other hantavirus infections [11–13]. For this study, 82 serum samples obtained from 26 patients hospitalized with acute SN virus infection were subjected to moreextensive serologic examination. Of the 26 samples, 24 were collected from patients within 24 h of admission. These 24 samples included 12 from patients with mild disease (those who did not require endotracheal intubation) and 12 from patients with severe disease (those who required intubation). Since the antibody titers of these 24 admission samples were judged to be most likely to display a relationship with disease severity, these antibody titers were compared statistically by disease severity.
Strip immunoblot assay (SIA)
An SIA was developed for the detection of antibodies to SN virus . This technique is minimally different in principle from the Western blot method [12, 13]. The SIA is similar to one that once was under commercial development by Chiron (Emeryville, CA), except that the Chiron prototype was developed for use with IgG conjugates only, and IgM and IgA versions were not available . The SN virus N antigen was expressed from the pET23b vector (Novagen, Madison, WI) in Escherichia coli and was purified over a metal chelation column by using a C-terminal polyhistidine moiety, as described elsewhere . Residues 35–117 of the G1 antigen were expressed in yeast in fusion with human Superoxide dismutase and were purified by ammonium sulfate precipitation and gel filtration chromatography, as described elsewhere .
On the basis of formal proficiency tests with the Oregon Department of Health (Portland) and of informal proficiency testing with the California Department of Health (Berkeley), the sensitivity and specificity of the IgM and IgG SIA test appear to approach 100%. We are not aware of any sample that we diagnosed incorrectly with the SIA test. Similar data were obtained with the Chiron SIA when it was available. On the basis of the greater band intensities we saw with the current test in side-by-side comparisons, we believe that the current IgM/IgG SIA is less likely to miss a positive sample than was the Chiron SIA.
For the SIA, the viral antigens were suctioned onto a nitrocellulose membrane under vacuum (0.5 m Hg for 3 min) with a MiniSlot 10 apparatus (Immunotics R, Cambridge, MA). We diluted 2 μL of highly purified (0.9 mg/mL) recombinant SN virus N antigen to 1 mL in PBS, pH 7.4, and 1.5 μL of purified recombinant Gl antigen (1.06 mg/mL) was diluted to 1 mL of PBS and vacuumed onto a wetted 8 × 10-cm nitrocellulose membrane . Separate wells were filled with human serum at dilutions that provided control lanes of 3 + and 1 + intensity for anti-human IgM, IgG, or IgA conjugates. A Coomassie brilliant blue lane was also added to allow proper orientation of the SIA strips. After the antigens were transferred fully by vacuum, the nitrocellulose was removed from the apparatus and allowed to dry for 2 min. It was then cut lengthwise into 2-mm-wide strips with a hand-held paper shredder. The strips were then stored at 4°C in PBS buffer containing 5% nonfat dry milk [11, 12].
Serum samples from patients were preincubated for 1 h in a Western blot tray (BioRad Laboratories, Richmond, CA) with a blocking reagent that consisted of a detergent lysate of E. coli in PBS buffer containing 5% milk [11–14]. We used a series of 4-fold dilutions (1:125–1:32,000). Samples that showed reactivity at 1:32,000 were subjected to further examination at dilutions of 1:128,000 and 1:512,000. After the preincubation, the SIA strips were placed into the wells in the tray and rocked gently overnight at room temperature. After 3 washes in detergent-PBS buffer (10 mM sodium phosphate, pH 7.4; 0.5% deoxycholic acid; 0.5% Triton X-100; 0.1 M NaCl), a 1:1000 dilution of alkaline phosphataseconjugated goat anti-human IgA, IgG, or IgM antibodies was added and rocked gently for 1 h. After 3 more washes, bound alkaline phosphatase was detected with nitroblue tetrazolium and 5-bromo-4-chloro-3-indoyl-phosphate substrate, at 0.33 mg/mL and 0.165 mg/mL, respectively. The development solution also contained 100 mM Tris-HCl (pH 9.5), 100 mM NaCl, and 5 mM MgCl2. Development was stopped after 10 min by decanting the development solution while rinsing the trays repeatedly with water. End-point reactivities were scored as the final dilution in which a visually detectable signal was evident above background.
Focus reduction neutralization test (FRNT)
FRNT is used widely for the study of hantavirus neutralization responses, as an alternative to the plaque reduction neutralization test, because some hantaviruses do not produce consistent plaques on cell monolayers . The serum samples from patients with HCPS were examined by FRNT individually in 48-well tissue-culture plates. Samples were serially diluted (1:50, 1:200, 1:800, 1:3200) and mixed with equal volumes of ∼45 focus-forming units of SN virus strain CC107 for 1 h at 37°C before incubation on Vero E6 cells. The dilution buffer consisted of complete MEM (Gibco/BRL, Grand Island, NY) containing 2.5% fetal bovine serum (HyClone Laboratories, Logan, UT). After adsorption for 4 h at 37°C, the cells were washed in PBS and were overlaid with media containing 1.2% methylcellulose for 7 days. The methylcellulose layer was removed, and the cells were fixed with 100% methanol containing 0.5% hydrogen peroxide . Viral antigen was visualized by the addition of hyperimmune rabbit anti-SN virus N protein (1:5000), followed by peroxidase-conjugated goat anti-rabbit IgG (Jackson ImmunoResearch Lab, West Grove, PA) and l,4-dideoxy-l,4,imino-D-arabinital/metal concentrate as substrate (Pierce Chemicals, Rockford, IL). The neutralization activity of a patient's serum sample was expressed as the maximum serum dilution that would reduce the number of foci by ⩾80%. These analyses were conducted by a person who was blinded to the severity of each patient's illness.
Unpaired data were analyzed with the single-factor analysis of variance test.
In total, 26 patients were studied during the acute phase of disease. For most patients, we had ⩾2 consecutive serum samples, which allowed us to examine the antibody titers as a function of time. Table 1 shows the geographic distribution, age range, sex, and clinical characteristics of the 26 patients. Blood samples were collected within the first 24 h of admission in all but 2 patients.
We wished to determine the course of antibody response as a function of time in the mildly ill patients (classes 0 and I), in comparison with those patients with severe illness (classes II and III). The evolution of antibody responses in 1 patient is exemplified in figure 1. Serial 4-fold dilutions were used, from 1:125 to 1:32,000, for 3 different immunoglobulins (IgM, IgG, and IgA). For this patient, the specific IgM response decreased with time, whereas the IgG reactivity increased. After 15 months, the IgM and IgA antibodies were diminished.
We compared the neutralizing and nonneutralizing antibody titers of the earliest available blood samples from the patients who had milder disease with those from patients who were severely ill (table 2, figure 2). Only the 24 samples that were collected before admission or within the first 24 h after admission were compared statistically. None of the patients with mild disease had neutralizing antibody titers <1:800; however, among those with severe disease, nearly all had titers ⩽1:800. The exception was a 44-year-old man with severe disease and a titer ⩾1:3200. He was treated successfully with extracorporeal membrane oxygenation and survived. The P value for the comparison of the severely ill patients with those with milder disease, by rank sum statistic, was 3.4 × 10−5 (table 2). In contrast, similar rank sum comparisons of the various antibody titers (IgA, IgG, and IgM against recombinant Gl and N antigens) generally showed no differences between the 2 patient groups. The IgG titers against the viral N antigen were somewhat lower in the patients with severe disease than they were in those with milder disease (P < .033, table 2).
The evolution of neutralizing and nonneutralizing antibody titers in the 2 groups is depicted in figure 2. Figure 2A shows the neutralization titers, whereas figure 2B shows the titers of IgG antibodies directed against the SN virus N antigen. Although the antibody titers of the severely ill patients appear to catch up with those of patients with milder disease during the course of hospitalization, 7 class III patients died during their hospitalization (table 1). Data from the far end of each curve are of limited significance because of the few patients examined.
Our efforts to characterize the antibody responses of patients with severe and mild forms of HCPS were motivated by our interest in identifying new therapeutic options. Our reasoning that poor antibody responses might be associated with a poor outcome was inspired by the experience with other viral hemorrhagic fevers . If admission antibody titers do differ significantly between the 2 groups, it may lead to the investigation of hyperimmune plasma or human monoclonal antibodies as a therapeutic option.
Early diagnosis of hantavirus infection is hampered by the limited availability of diagnostic tests for antibodies or viral RNA. It is now clear that most, if not all, patients in the prodromal phase of HCPS have anti-SN virus antibodies of at least the IgM class [11–13] (B.H., unpublished data). Early diagnosis could be effective at facilitating early triage to a tertiary care center, if a specific antibody test were available in the rural clinics where most patients with HCPS first present for treatment. It is unfortunate that such a test is not available, because early referral to a tertiary care center with experience with HCPS probably will reduce mortality. After the cardiopulmonary phase of illness begins, therapeutic options are limited. The cardiopulmonary phase advances rapidly in persons destined to have a fatal outcome, which makes otherwise reasonable interventions (e.g., administration of intravenous ribavirin) less likely to produce a significant effect.
Therapy for Argentine hemorrhagic fever, which is caused by the Junin arenavirus, often includes administration of hyperimmune plasma from patients who have recovered from the disease. As little as 2–3 U of plasma may be sufficient, depending on the titer of neutralizing antibodies. This intervention has reduced the case-fatality rate of acute Junin virus infection from 20%–30% to l%–2% and remains effective even if therapy is begun several days into the disease . There are important biologic differences between diseases caused by arenaviruses and those caused by hantaviruses. For example, the antibody responses induced by arenaviruses are often quite delayed, compared with those induced by hantaviruses. Nevertheless, we wished to determine whether there was any scientific basis for the consideration of immune plasma therapy for HCPS. For this reason, we looked for a correlation between the severity of infections with SN virus and the initial titer of neutralizing antibodies.
Neutralization tests are used widely for diagnosis, but their use in predicting outcomes of viral infection or in treatment is less well developed [15, 17–19]. We showed a strong correlation between neutralizing antibody titer and the severity of disease caused by SN virus (table 2). By contrast, the differences between the 2 groups in nonneutralizing antibody titer were much less obvious (table 2, figure 2). The most straightforward explanation is that patients who are able to develop strong neutralizing antibody response also can reduce plasma viremia and viral replication and that antibodies serve a useful role in clearing the infection during the acute phase of illness. The reasons for the variability of disease severity are not well understood but probably include genetic differences among the hosts, differing dosages of virus inoculum, and/or viral genetic differences.
Most patients who had a mild course of illness had a neutralizing antibody titer ⩾1:3200 at the time of the initial blood sampling. None had a titer <1:800, even when the sample was collected ⩽2 days before hospitalization. In contrast, most patients with severe infections had titers ⩽1:200, and some of those patients who survived a few days after hospitalization showed delayed or nonexistent increases in titer during their stays in the intensive care unit (figure 2). We considered the first day of hospitalization to be the most useful reference point, since it gave us some ability to establish that the severe cases and the mild cases were being studied at roughly the same stage of illness. We also noted that the date of onset of symptoms or the time of exposure often is difficult to ascertain accurately, especially for patients in whom disease progressed rapidly.
We are grateful to K. M. Johnson for consultations regarding this manuscript and to C. S. Schmaljohn for providing the Sin Nombre virus isolate CC107.