Viral load and immune response dynamics in patients with haemorrhagic fever with renal syndrome

Authors


Corresponding author: T. Avšič–Županc, Institute of Microbiology and Immunology, Medical Faculty, Zaloška 4, 1000 Ljubljana, Slovenia

E-mail: tatjana.avsic@mf.uni-lj.si

Abstract

Haemorrhagic fever with renal syndrome (HFRS) in Slovenia can be caused by infection with either Dobrava (DOBV) or Puumala (PUUV) virus, but a clear difference in disease severity is observed. We hypothesized that the wide spectrum of disease observed among HFRS patients might be related to differing immune responses and viral load kinetics. To test this hypothesis we analysed sequential blood samples from 29 HFRS patients hospitalized in Slovenia. Measuring viral RNA in patient samples revealed that viraemia lasts for longer than previously believed, with DOBV or PUUV-infected patients having viraemias lasting on average 30 days or 16 days, respectively. DOBV-infected patients were found to have a higher viral load than the PUUV-infected patients (107 vs. 105 RNA copies/mL). Both DOBV and PUUV-infected patients had IgM at the time of hospital admission, but there was a difference in IgG antibody dynamics, with only a minority of DOBV-infected patients having IgG antibodies. In our study, elevated levels of IL-10, TNF-α and IFN-γ were detected in all of the samples regardless of the causative agent. In DOBV-infected patients the decrease in cytokine secretion level appeared around day 20 post-infection, while in PUUV-infected patients the change was earlier. In general, our findings point toward notable differences between PUUV and DOBV infections, in terms of viral load and antibody and cytokine response dynamics, all of which may be reflected in differing disease severities and clinical outcomes.

Introduction

Human hantavirus infections can result in two different clinical syndromes depending on the causative agent involved, namely hantavirus (cardio) pulmonary syndrome (HCPS) in the Americas and haemorrhagic fever with renal syndrome (HFRS) in Europe and Asia [1]. In general, HFRS is clinically characterized by a sudden onset of acute fever, headache, back and abdominal pain, myalgia, nausea, vomiting and transient myopia. In more severe cases, acute renal insufficiencies, requiring treatment with transient haemodialysis, and haemorrhagic manifestations, due to capillary leakage, are observed [2, 3]. Hantaviruses have been documented to infect endothelial cells, causing a viraemia that typically clears within the first 2 weeks after the first appearance of symptoms [4, 5]. The virus has no direct cytopathic effect on primary cell targets; therefore host immune mechanisms seem to play an important role in the pathogenesis [2, 6]. Monocytes and macrophages appear to play roles both in systemic immunity to hantavirus infections and in pathogenesis through the release of cytokines and chemokines [7]. Several studies have shown the presence of elevated levels of TNF-α, IL-6, IL-2, IL-1 and IL-10 in HFRS patients [7-10]. Although humoral immune responses to hantaviruses are generally associated with protective immunity, an efficient CD8+T-lymphocyte response has also been observed in HFRS patients early in the course of disease [11, 12]. In a longitudinal study of primary CD8+ T-cell responses of HFRS patients from the first onset of clinical symptoms until viral clearance, vigorous CD8+ T-cell responses were detected during the first 2 weeks of illness. A subsequent decline in effector CD8+ T-cells coincided with a decline in viral load and clearance of the virus from the infected patients. Regulatory CD4+ T-cells remained stable throughout the course of infection [12]. Thus, it was proposed that a mixed pattern of Th1 and Th2 immune responses, high levels of pro-inflammatory cytokines and the absence of immunoregulatory mechanisms contribute to pathogenesis and lead to the severe disease form [4, 7, 12]. In Slovenia, HFRS is caused by the Puumala virus (PUUV) and Dobrava virus (DOBV). The clinical severity of the disease caused by PUUV is significantly lower than that of HFRS due to DOBV [10, 13-15]. Likewise, higher viral load levels were demonstrated in patients infected with DOBV than in those infected with PUUV, but there was no correlation between the level of antibody responses and the clinical course of the disease in these patients [16]. Comparing cytokine profiles revealed higher levels of IL-10 and TNF-α in DOBV-patients with severe HFRS than in those infected with PUUV. In contrast, higher levels of IL-12 were detected in patients infected with PUUV [10].

To further define factors that contribute to severe HFRS, we have investigated differences in the kinetics of viral load, antibody responses and cytokine responses of DOBV and PUUV-infected HFRS patients during the disease progression. We hypothesized that differences in immune responses and viral load dynamics might lead to diverse disease progression. Here we present the results of our analysis of daily clinical samples that were prospectively collected from patients at their first presentation at the regional hospital with acute symptoms and during the entire time of hospitalization until the viral infection was resolved.

Materials and Methods

Collection of human blood and serum samples

The study was performed on 29 hospitalized HFRS patients, 18 infected with PUUV and 11 infected with DOBV. Patients included in the study met the following inclusion criteria: (a) verified diagnosis of acute hantavirus infection with immunofluorescence assay [13]; (b) hospitalization due to manifest HFRS; and (c) access to daily sequential samples of peripheral blood during the hospitalization time. The study was approved by the National Medical Ethics Committee and also written or oral informed consent was obtained from all included patients. In all, we have collected 183 blood samples from 18 HFRS patients infected with PUUV and 204 blood samples from 11 HFRS patients infected with DOBV. After each patient was discharged from the hospital, a detailed medical chart was collected and significant laboratory parameters were analysed.

Viral load dynamics

Total RNA was extracted from blood samples with TRIZOL LS Reagent (Invitrogen Life Technologies™, Foster City, CA, USA) in accordance with the manufacturer's instructions. The viral load was measured in daily samples of whole blood using Quantitative One-Step real-time reverse transcriptase polymerase chain reaction to detect and quantify DOBV and PUUV genomes [[17],[18]].

Antibody dynamics

The IgM and IgG antibodies against DOBV and PUUV were detected using commercially available enzyme immunoassays: Reagena DOBRAVA-HANTAN IgM EIA, Reagena DOBRAVA-HANTAN IgG EIA, Reagena PUUMALA IgM EIA and Reagena PUUMALA IgG EIA (Reagena International Oy Ltd, Toivala, Finland). The test was performed according to the manufacturer's instructions. A cut-off value for IgM is R > 1 and for IgG is R > 1.2. Based on our experiences, the average value measured, at admission to the hospital, is 2.8 (max 8.8) for IgM and 2.3 (max 5.0) for IgG.

Cytokine detection

Levels of IL-10, IL-12, IFN-γ and TNF-α were measured in plasma samples with commercial ELISA tests. Namely, for the determination of IL-10, IL-12 and INF-γ, the Human IL-10 ELISA kit, Total Human IL-12 ELISA kit and Human IFN-γ ELISA kit (all Thermo Fisher Scientific Inc., Waltham, MA, USA) were used. The TNF-α ELISA kit (BioSource Europe S.A., Nivelles, Belgium) was used to measure TNF-α concentrations. The limits of detection, as declared by the manufacturer, for IL-10, IL-12, IFN-γ and TNF-α were <3 pg/mL, <5 pg/mL, <2 pg/mL and <3 pg/mL, respectively. Normal values, measured in plasma samples of Slovenian healthy adult blood donors, for IL-10, IL-12, IFN-γ and TNF-α were determined to be <10.8 pg/mL, <203 pg/mL, <1.1 pg/mL and 4.76–12.4 pg/mL, respectively.

Statistical methods

All statistical analyses were performed using the R environment [19]. A statistical confidence level α of 0.05 was considered. Smooth lines in graphs were calculated by the loess smoothing procedure (local polynomial regression fitting [20]) in order to get an impression of general time trends in the data. Exploratory mixed-effects regression models (R package nlme) were used to identify statistically significant associations between viral load in time and a response variable (one of IgM, IgG antibody kinetic or cytokine response kinetics). In these models each patient is represented by a random effect with correlation modelled by the AR(1) process. Predictor variables (time and viral load) are considered to be either linear or non-linear (via restricted cubic splines formulation (R package rms)) [21]. As most patients were dismissed from the hospital before day 41 the measurements after that were omitted from the statistical analyses.

Results

Patients and clinical data

Twenty-nine HFRS patients hospitalized in Slovenia during the years 2007 and 2010 participated in the study. During the same period 83 cases of HFRS were confirmed in Slovenia. Based on clinical data and laboratory parameters [15, 17] the disease progression was evaluated. Among 18 patients infected with PUUV, 10 patients were categorized as mild-to-moderate and eight patients as severe disease manifestations. In the DOBV-infected HFRS patients, one patient had a fatal outcome, three patients experienced the severe form of the disease and seven patients had mild-to-moderate disease. This study group exemplified the representative number of patients with different clinical manifestations existing in Slovenia according to historical data.

In the PUUV-infected patients, the mean time from onset of symptoms until hospital admission was 4.7 days, and the mean time of hospitalization was 10.4 days. The longest hospitalization time was 30 days for the patient experiencing the severe form of HFRS with many complications, including: transient blindness, severe lumbar pain and meningitis. In the DOBV-infected patients, the mean time from the onset of the disease until hospitalization was 6.6 days and the mean hospitalization time was 28.5 days. The longest hospitalization time was 75 days, for the patient experiencing severe disease manifestations that led to chronic renal impairment. The most common symptoms and significant laboratory parameters for both groups of patients are shown in Table 1.

Table 1. Comparison of different clinical, laboratory and microbiological findings in Dobrava virus (DOBV) and Puumala virus (PUUV)-infected groups of patients
VirusPUUDOB
Time before hospital admission (days)4.76.6
Duration of hospitalization (days)10.428.5
Most common symptoms at admissionHigh fever, vomiting, severe lumbalgia and abdominal pain, blurred vision,impaired renal functionHigh fever, chills, severe lumbalgia and abdominal pain, severe impairment of renal function, conjunctival injections
Number of patients requiring haemodialysis1 (5.6%)5 (45.5%)
Laboratory findings
Min platelets (mean value)68 × 109/L44 × 109/L
Max creatinine (mean value)349.5 µM593.1 µM
Max urea (mean value)17.3 µM28.5 µM
Max AST/ALT (mean value)1.1/1.6 µkat/L1.8/1.7 µkat/L
Microbiological findings
Duration of viraemia (days)1630
Viral load (mean value)105 RNA copies/mL107RNA copies/mL
IgM (mean value)3.902.97
IgG (mean value)2.121.44
IL-10 (mean value) min/max value

400.4 pg/mL

1.1/5444.80 pg/mL

132.3 pg/mL

0.62/6000.0 pg/mL

IL-12 (mean value) min/max value

70.9 pg/mL

0.0/242.1 pg/mL

84.0 pg/mL

0.0/291.1 pg/mL

TNF-α (mean value) min/max value

44.4 pg/mL

1.0/451.5 pg/mL

64.9 pg/mL

14.2/237.2 pg/mL

INF-γ (mean value) min/max value

4.2 pg/mL

0.0/38.0 pg/mL

6.4 pg/mL

0.0/90.4 pg/mL

Kinetic of viral load

In all patients the viral load decreased with disease progression. In DOBV-infected patients the longest duration of viraemia observed was 40 days post-infection. All DOBV-infected patients were viraemic on the first 12 days post-infection, and most patients had detectable viral RNA in their blood up to 30 days post-infection (Fig. 1, above left). Most of the patients were dismissed from hospital care before the virus was cleared from the blood system. In addition, at admission all DOBV-infected patients had more than 107 RNA copies/mL (Table 1). The highest viraemia, 2.5 × 1010 RNA c/mL, was measured on the fifth day of illness in the DOBV-infected fatal patient (Fig. 1, below left, green line). As expected, in PUUV-infected patients the mean interval of viraemia was shorter, 16 days post-infection. PUUV-infected patients were generally hospitalized for shorter times than DOBV-patients; however, for one PUUV-patient, viraemia was detected at 30 days post-infection. The mean viral load in PUUV-infected patients was 105 RNA c/mL (Table 1). The highest viral load was 1.03 × 1010 RNA c/mL, measured on the seventh day of illness in a patient with severe disease progression (Fig. 1, below right, green line).

Figure 1.

Mean kinetic of viral load in patients infected with Dobrava virus DOBV (left) and Puumala virus PUUV (right). In the lower two figures the kinetics of viral load in a single patient infected with DOBV (left) and PUUV (right) is presented.

In both groups the initial decrease in viral load was observed in the first few days after admission to the hospital, but in PUUV-infected patients the virus clearance was more rapid. Also, clear differences between patients in each group were observed, especially among the PUUV-infected patients (Fig. 1, below). We also observed interesting viral load kinetics in three patients infected with DOBV and five patients infected with PUUV, where an initial decrease in viraemia was followed by another increase (Fig. 1, below steep drops followed by steep increases). We were unable to correlate this observation with any specific treatment or therapy, or with clinical severity or antibody dynamics (data not shown). Three out of eight patients had underlying disease, such as rheumatoid arthritis, diabetes or arteriosclerotic vascular disease with chronic hypertension, which could influence the immune response.

IgM and IgG antibody kinetics

Puumala-infected patients were admitted to hospital care between 3 and 11 days after the onset of disease. On admission all patients had anti-PUUV IgM and 13 patients had IgG antibodies. DOBV-infected patients presented for hospital care between 4 and 11 days after the onset of disease. On admission all patients had anti-DOBV IgM antibodies, but only three had detectable IgG antibodies. The IgM and IgG mean values are shown in Table 1. DOBV-infected patients had lower humoral responses than PUUV-infected patients, but the kinetics of antibody development was more apparent (Fig. 2). The DOBV patients displayed clear time dependence for the appearance of IgM and IgG antibodies, where the IgM response was decreasing and the IgG response was increasing as the infection progressed. Although the kinetics of the IgG antibody response between PUUV and DOBV-infected patients was different, in the DOBV-infected patients the IgG response was constantly increasing in relation to the day of illness, but in the PUUV-infected patients a plateau appeared to be reached after 25 days post-infection. A non-linear negative association using the mixed-effects regression model (p <0.01) was observed between IgG responses and viral loads in DOBV-infected patients, in which a high viral load was associated with low IgG, regardless of the day of disease. The shape of the mean IgG response for the regression model where the day of illness and viral load were treated as non-linear can be seen in Fig. 3. No statistically significant association with viral load was observed for IgM responses in DOBV-infected patients or in IgM or IgG response in PUUV-infected patients.

Figure 2.

IgM and IgG antibody response against Dobrava virus (DOBV) and Puumala virus(PUUV) infection in plasma samples of haemorrhagic fever with renal syndrome(HFRS) patients.

Figure 3.

Shape of mean IgG response for the regression model where the day of illness and viral load were treated as non-linear. The scale on the y-axis corresponds to the fifteenth day of illness.

Cytokine response kinetics

Elevated levels of IL-10, TNF-α and IFN-γ were detected in almost all of the samples tested, regardless of the causative hantavirus (Table 1, Fig. 4). In DOBV-infected patients elevated cytokines slowly declined to a level of healthy blood donors at around 20 days post-infection. However, in PUUV-infected patients a change in cytokine dynamic was observed earlier, around day 10 post-infection. In DOBV-infected patients concentrations of IL-10, except for the fatal case, were lower than in PUUV-infected patients (22.4 vs. 400.4 pg/mL). Also in DOBV-infected patients a negative time dependence of IL-10 was observed, in contrast to PUUV-infected patients, where elevated levels of IL-10 seem to persist during the hospitalization time (Fig. 4, row 1). A significant association was observed between highly non-linear viral load dynamics and secretion of IL-10, where increase in viral load led to higher production of IL-10 regardless of the causative agent (p 0.04) (Fig. 5). Overall, all DOBV and PUUV-infected patients had levels of TNF-α high above the values of healthy blood donors during the whole hospitalization time, despite there being a trend toward negative time dependence for first 25 days (Fig. 4, row 2). An exception was the DOBV-infected fatal patient, where TNF-α concentrations increased as the disease progressed. The mean concentrations of TNF-α were higher in DOBV than PUUV-infected patients (64.9 vs. 44.4 pg/mL). The highest dynamic was observed for IFN-γ concentrations, which in most patients increased and decreased vigorously almost daily. On average, higher concentrations of IFN-γ were detected in patients infected with DOBV than PUUV (6.4 vs. 4.2 pg/mL). Namely, in PUUV-infected patients, concentrations of IFN-γ showed negative time dependence, but in some DOBV-infected patients a second increase after day 20 was observed (Fig. 4, row 3). The increase in IFN-γ concentrations follows the increase in viral load, but the models show no statistically significant association. Furthermore, in four patients (3 DOBV and 1 PUUV) IFN-γ could not be detected at all during the whole hospitalization.

Figure 4.

Mean dynamic of IL-10, TNF-α, IFN-γ and IL-12 in plasma samples of haemorrhagic fever with renal syndrome(HFRS) patients during the entire hospitalization time. The red line represents cytokine concentrations measured in healthy blood donors in Slovenia.

Figure 5.

The shape of the mean response of IL-10 for the regression model where the day of illness was treated as a linear and log viral load as a non-linear effect. The scale on the y-axis corresponds to the tenth day of illness.

In contrast to other cytokines, concentrations of IL-12 were decreased in comparison to levels measured in healthy blood donors (84.0 and 70.9 vs. 203 pg/mL) (Fig. 4, row 4). Considering the dynamic of IL-12 secretion, we have observed association between viral load and IL-12, where higher viral load is associated with higher IL-12 secretion (p 0.01).

Discussion

The mechanism behind hantavirus pathogenesis is not completely understood. Hantaviruses primarily infect endothelial cells lining the vascular system and cause increased permeability followed by capillary leakage. As hantaviruses are not cytopathogenic per se it has been suggested that host immune mechanisms directed against the infection cause the vascular permeability [2, 6]. For both HFRS and HCPS patients, a correlation between higher viral load levels and the severity of the disease has been reported [17, 22]. For HFRS, a wide spectrum of clinical disease is observed, thus it is possible that differences in immune response dynamics between HFRS patients infected with DOBV or PUUV might lead to diverse disease progressions and that viral load is also a contributing factor to those differences. To investigate this hypothesis, we collected sequential blood samples from 29 HFRS patients hospitalized in Slovenia (18 PUUV and 11 DOBV). Our study results are consistent with previous findings [13, 14] that DOBV is responsible for the severe form of HFRS in Slovenia, because the patients have advanced renal impairment and therefore more frequently require haemodialysis (Table 1). Also, all clinical parameters relating to kidney and liver function are higher in DOBV-infected patients than in PUUV-infected patients and the DOBV patients are hospitalized for longer periods of time.

Although DOBV and PUUV-infected patients both had IgM at the time of hospital admission, there was a difference in the subsequent viral loads, with DOBV-infected patients having much higher levels than PUUV patients. Therefore, our results are consistent with earlier studies indicating that more virulent hantaviruses (like DOBV and SNV) tend to have higher viral loads [17, 22]. Contrary to the generally accepted fact that viraemia in HFRS and HCPS patients lasts only a few days [5, 12, 23], our study showed that viraemias in DOBV or PUUV-infected patients averaged 30 days (max. 40 days) and 16 days (max. 30 days) post-infection, respectively. In addition to the unexpectedly prolonged viraemias in both groups of patients, differing viral kinetics were also observed. In DOBV-infected patients, after an initial drop in viral load, which occurred after a few days of hospitalization, the viral load tended to stabilize. In contrast, in the PUUV-infected patients the viral load continued to drop throughout the hospitalization period. Interestingly, all patients entered the convalescent phase before they were discharged from the hospital, but most of them still had detectable viral loads, especially those infected with DOBV.

Highly variable viral loads were observed among patients in both groups, which suggest that a host–virus interaction could be a factor in the pathogenesis seen in individual patients. The correlation between host immune genetics, especially HLA type and TNF-α polymorphism, and the severity of or susceptibility to hantavirus infection has already been shown for several hantaviruses, including DOBV and PUUV [15, 24-27]. In addition, in some HFRS patients, the virus seemed to be temporarily cleared, but this was usually followed by a second period of viraemia. This is similar to what we commonly observe in experimentally infected hamsters (data not shown), and we surmise that the second viraemia is the result of viral replication in target tissues and release of additional infectious virions. In the study reported here, we could not correlate this observation with medical treatment, severity of the disease or any underlying disease.

Consistent with earlier studies, we found that the humoral immune response to infection is already present when the first symptoms of HFRS appear [18, 28, 29], consequently the antibodies do not appear to completely control viral load kinetics. There was a difference in antibody kinetics, however, in that while most PUUV-infected patients had IgG antibodies at admission, only a minority of DOBV-infected patients had IgG antibodies. In addition, DOBV patients had lower IgG antibody responses in general than PUUV patients and displayed a rapid decrease in IgM. Although observed differences between DOBV and PUUV-infected patients could be also influenced by the quality of commercial ELISAs, we believe that at least this is not the only factor in such observations, mainly because a low IgG response was correlated with a high viral load for the DOBV-infected patients but not for the PUUV-infected patients.

It is generally accepted that neutralizing antibodies are a primary correlate of protective immunity for HFRS. However, once HFRS is established, the humoral immune response may not be sufficient to resolve the disease. The importance of a robust cell-mediated response is supported by earlier studies in which it was shown that a decline in effector CD8+ T-cells coincided with decline in viral load and clearance of the virus from HFRS patients [12]. Besides clearing the viral infection, CD8+ T-cells are also a rich source of cytokines and chemokines, such as TNF-α and IFN-γ [30]. In HFRS patients several studies have shown elevated levels of TNF-α, IL-6, IL-2, IL-1 and IL-10 patients [7-10]. In our study, elevated levels of IL-10, TNF-α and IFN-γ were detected in all of the samples regardless of the causative agent. Again, differences in both groups of patients were observed; in DOBV-infected patients the decrease in cytokine secretion level appeared around day 20 post-infection, while in PUUV-infected patients the change was earlier. In agreement with the work of others [9, 28, 29] we observed increased levels of TNF-α in both DOBV and PUUV-infected patients. Although the increased levels of TNF-α were observed in both groups, there was no association with viral load in our study. We did observe a significant correlation between the viral load dynamics and secretion of IL-10; an increase in viral load led to higher production of IL-10. IL-10 displays a potent ability to suppress the antigen-presentation capacity of antigen presenting cells. Although IL-10 is an anti-inflammatory cytokine and is capable of inhibiting synthesis of pro-inflammatory cytokines such as IFN-γ and TNFα, there was no association of IL-10 and IFN-γ or TNF-α observed in studied patients.

Another potent mediator of immunopathogenesis, NK cells, is activated by IFN-α, IFN-β, TNF-α, IL-12 and IL-15. In addition, NK cells can also influence the adaptive immune response directly by secreting IFN-γ [31]. IFN-γ has been shown to inhibit HTNV replication in vitro, and in PUUV-infected patients IFN-γ was found to be significantly decreased in the acute phase of HFRS [32]. In our study, IFN-γ dynamics changed almost daily in observed patients. On average, in most DOBV and PUUV-infected patients elevated concentrations of IFN-γ, in comparison to healthy blood donors, were measured, although in four patients with severe disease progression, IFN-γ could not be detected for almost the entire length of the patients' hospitalization. Our data suggest that concentrations of IFN-γ from different patients, measured in the early or late phase of acute HFRS, cannot be correlated with each other, as there are obvious differences between patients.

Consistent with earlier data [7], we observed a decrease in most measured cytokines in PUUV-infected patients during the late phase of acute disease. Interestingly, in DOBV-infected patients increases in measured cytokines during the late phase of disease were observed. In general, our findings point toward notable differences between PUUV and DOBV infections, in terms of viral load and antibody and cytokine response kinetics, all of which may be reflected in differing disease severities and clinical outcomes. The disparity between pro- and anti-inflammatory cytokines might be in part responsible for the pathology seen in HFRS patients, but the extreme variations observed among patients suggest that other factors, such as host genetics, are also important. This is also the reason why we believe that serial samples from single patients are the only way to obtain meaningful data comparing different biological factors in the early and late phases of the disease.

Acknowledgments

We thank the following infectologists for their help in collecting samples and retrieving patients' case records: Rafael Černuta, Anica Berginc Dolenšek, Professor Dr Stanka Lotrič-Furlan, Silvija Mörec Jakopič, Professor Dr Marko Noč, Dr Jernej Pajk, Emil Pal, MSc and Academic Professor Dr Franc Strle.

Transparency Declaration

This study was supported by the US Civilian Research and Development Foundation [SIB1-2964-LJ-09] and by the Ministry of Higher Education, Science and Sport of Slovenia (grant No. P3-0083).

Ancillary