Early Release – Rio Mamore Hantavirus Endemicity, Peruvian Amazon, 2020 – Volume 30, Number 12—December 2024 – Emerging Infectious Diseases journal

Molecular Analyses

We extracted nucleic acids from samples collected during routine AFI surveillance in Peru by using the MagNA Pure 96 DNA and Viral NA Small Volume Kit (Roche, https://www.roche.com). To elucidate potential co-infections, we tested all samples by nested reverse transcription PCR (RT-PCR) for hantavirus RNA and samples from Loreto by quantitative RT-PCR (qRT-PCR) for DENV (9,10). We conducted library preparation by using the KAPA RNA HyperPrep Kit (Roche), followed by enrichment via in-solution hybridization capture (Arbor Biosciences, https://arborbiosci.com) as previously described for hantavirus genomic sequencing (11). We sequenced the captured library by using an Illumina Miniseq High-Output Reagent Kit (paired-end 150-bp cycles) (https://www.illumina.com) and mapped reads against the RIOMV strain HTN-007 by using Geneious 2023.2.1 (https://www.geneious.com) and deposited sequence data in the European Nucleotide Archive (https://www.ebi.ac.uk; accession no. ERR11860590). We attempted to close sequence gaps by using PCR with specific bridging primers (Appendix 1 Table 1) and quantified the viral RNA load by using a set of specific oligonucleotides (Appendix 1 Table 2).

Phylogenetic Analyses

We conducted nucleotide alignments by using MAFFT with an L-INS-I algorithm (https://mafft.cbrc.jp) using Geneious 2023.2.1. We conducted Bayesian phylogenies by using MrBayes 3.2.6 (12) with a general time-reversible substitution model with gamma distribution and a complete deletion of all positions containing gaps in the alignment. We retrieved hantavirus reference sequences from GenBank for phylogenetic analysis (Appendix 1 Table 3). We conducted sequence identity plots of partial genomic sequences by using SSE 1.4 (13) with a fragment length of 200 and an increment between fragments of 50 nt. We constructed neighbor-joining trees of partial hantaviral genomic sequences in GenBank by 1,000 bootstrap replicates and the pairwise deletion option in MEGA X (14) (Appendix 1 Table 4). We determined translated amino acid sequence distances in MEGA X by using a pairwise deletion option (14).

Serologic Analyses

We tested serum samples by IgM/IgG ELISA by using a pool of recombinant nucleoproteins from ANDV and Sin Nombre virus (SNV) licensed for diagnostic use in Peru (EUROIMMUN, https://www.euroimmun.com) (Appendix 1). To provide additional validation for the IgM/IgG ELISA-positive samples, we conducted an end-point antibody titration by using an indirect immunofluorescence assay (IFA), as previously reported for hantavirus serologic investigations (15,16). For IFA, we tested serum samples at 1:10–1:10,000 dilutions in 10-fold dilution steps. A few samples still yielded weak reactivity at 1:10,000, so we also tested those at 1:12,500. IFA was based on cells infected with ANDV, SNV, Seoul virus, Hantaan virus, Puumala (PUUV), Dobrava (DOBV), and Saaremaa hantaviruses (EUROIMMUN). For IgM-based IFA and ELISA, we pretreated serum samples with the immunoadsorbent Eurosorb (EUROIMMUN) to deplete class IgM rheumatoid factors potentially present in the sample that might react with specifically bound IgG, causing false-positive results and in parallel depleting specific IgG, displacing IgM from the antigen causing false-negative results. IgM and IgG detection relies on specific secondary fluorescein-coupled antibodies for IFA or horseradish peroxidase–coupled antibodies for ELISA.

To test for potential causes of unspecific reactivity in ELISA, we used PCR for pathogens commonly eliciting polyclonal B cell stimulation, including Plasmodium spp. (17), Epstein-Barr virus (EBV) (18), and cytomegalovirus (CMV) (19) (TIB Molbiol, https://www.tib-molbiol.de). To assess potentially unspecific reactivity, we used IgM/IgG ELISAs for endemic arboviruses in serum from hantavirus-seropositive patients and controls, including Oropouche virus (OROV) (20), Mayaro virus, chikungunya virus (21), and nonendemic arbovirus Crimean-Congo hemorrhagic fever virus (CCHFV), and we used IgG ELISAs only for Plasmodium spp. (22,23), SARS-CoV-2 (nucleoprotein antigen-based) (24), and the nonendemic pathogen Middle East respiratory syndrome coronavirus (S1-based; IgM is not commonly tested for respiratory human coronaviruses) (all ELISAs from EUROIMMUN).

To exclude a nonspecific reaction to components of the ELISA other than viral antigens, we tested all hantavirus IgM/IgG ELISA–positive samples on non–antigen-coated ELISA plates acquired from the manufacturer, following the same ELISA protocols. We compared serologic reactivity in hantavirus IgM/IgG ELISA-positive serum with 38 hantavirus IgM-negative and 38 IgG-negative samples from the same areas and time for which sufficient volumes were available, while ensuring comparable age distribution (hantavirus IgM-negative serum, mean patient age 20 years, SD 21.13; hantavirus IgM-positive serum, mean patient age 22 years, SD 22.58; t-test, t = 0.29, p = 0.77; hantavirus IgG-negative serum, mean patient age 26 years, SD 17.04; hantavirus IgG-positive serum, mean patient age 28 years, SD 24.53; t-test, t = 0.26, p = 0.80). We considered p<0.05 to indicate statistical significance and conducted all tests by using a 2-tailed approach. We performed statistical analyses by using R software version 2024.04.2 (The R Project for Statistical Computing, https://www.r-project.org).

Virus Isolation

We used Vero E6 and BHK-21 cell lines for virus isolation attempts, as previously described for RIOMV and other hantaviruses (25). For both lines, we seeded a monolayer of 1.8 × 10⁵ cells per well in a 12-well dish with Dulbecco Eagle modified medium supplemented with 10% fetal bovine serum and 1% penicillin/streptomycin. We diluted the serum sample 1:25 and inoculated 250 μL onto the cell monolayer. After 1 hour, we removed the inoculum and replaced it with fresh medium. We checked the cells daily to assess the development of cytopathic effects and tested by qRT-PCR. We performed 2 blind passages over 1 week each.

Cohort Description

Figure 1

Figure 1. Locations and climate classifications related to study of Rio Mamore hantavirus endemicity in the Peruvian Amazon, 2020. A) Location of Peru (green) in South America. B) Regions in Peru where…

During a 1-year period (January 2020–2021), INS collected and stored 3,400 serum samples from febrile patients for AFI surveillance during a DENV outbreak that overlapped with COVID-19 (24) in 3 diverse ecoregions: Loreto (Amazon; n = 1,972 samples), Lambayeque (coastal desert /dry forest; n = 743 samples), and Lima (coastal desert; n = 685 samples) (Figure 1). The mean age of patients analyzed was 27 years (SD 18.8); 51.1% (n = 1,735) were female and 48.9% (n = 1,663) male. The overall age and sex distributions were comparable to those of the total population of Peru (mean age 31 years; 50.8% female, 49.2% male) (https://www.inei.gob.pe/media/MenuRecursivo/publicaciones_digitales/Est/Lib1743/Libro.pdf).

Molecular Testing for DENV and Hantaviruses

In January 2020, a serum sample from a 5-year-old boy from Iquitos, Loreto region, was positive for hantavirus RNA by nested RT-PCR (9) (Figure 1). Because sampling was conducted during a dengue outbreak and because other viruses may co-occur during dengue outbreaks, as was illustrated by detection of a Fort Sherman orthobunyavirus in a patient from Lambayeque (27), DENV infection in Loreto was assessed by qRT-PCR (10). The hantavirus-positive serum was negative for DENV, whereas the overall rate of DENV detection in Loreto was 56.8% (95% CI 54.7%–59.0%; n = 1,121/1,972) during 2020–2021. The patient experienced fever, muscle pain, headache, and vomiting for 3 days before sampling; no travel history preceding AFI was reported. No further information about the clinical signs/symptoms and disease outcome of the patient was available. A BLAST (http://blast.ncbi.nlm.nih.gov) comparison of the 347-bp region amplified by hantavirus RT-PCR after the exclusion of primer binding sites showed the highest nucleotide sequence identity (97.1%) to the rodent-associated RIOMV strain from 1996 in Loreto (8).

The hantavirus RNA load in the serum sample was low at 3.8 × 10³ copies/mL, which was consistent with the time since symptom onset, because the hantavirus RNA load is known to decline rapidly from ≈10⁵–10⁶ copies/mL 3–6 days after symptom onset (28). Virus isolation was unsuccessful, most likely because of low viral load and sample degradation under tropical conditions and repeated freeze–thaw cycles.

Genomic Characterization of RIOMV

Figure 2

Figure 2. Phylogenetic relationships between partial concatenate sequences of RIOMV from Peru (RIOMV PER 2020, depicted in red) and reference sequences. The phylogenetic trees were constructed by using MrBayes 3.2.6 (http://mrbayes.csit.fsu.edu…

Figure 3

Figure 3. Sequence identity plot comparing RIOMV variants from Peru (RIOMV PER 2020 and RIOMV PER 1996) and Brazil (RIOMV BRA 2005). The identity plot was calculated by using SSE (http://www.virus-evolution.org…

Using a mixed approach of high-throughput and Sanger sequencing to close multiple gaps after high-throughput sequencing, probably because of low viral RNA load, we reached a genome coverage of 41.4% (4,956 nt; 543,993 reads), of which 1,681 nt were from Sanger sequencing, with coverage of 72.7% for small (S), 52.5% for medium (M), and 24.7% for large (L) segments (GenBank accession nos. OR902838–40) (Appendix 1 Figure 1). Sequence comparisons within a dataset of all hantavirus reference sequences (Appendix 1 Table 3) showed that the 3 segments of the RIOMV strain from our study were most closely related to the rodent-associated RIOMV from 1996 (S segment, 96.8%; M segment, 96.4%; L segment, 97.0%). Separate Bayesian phylogenetic analyses of each segment placed the RIOMV strain consistently in a clade sharing the most recent common ancestry with the rodent-associated RIOMV (Figure 2). The phylogenetic placement of all genomic segments within the same RIOMV clade and similar sequence distance across complete genomic segments (Figure 3) spoke against reassortment or recombination events in the available dataset (29). The clustering of our sequence within the RIOMV clade was also supported by phylogenies relying on all available partial genomic sequences of viruses belonging to the Orthohantavirus mamorense species (Appendix 1 Table 4; Figure 2).

Serologic Markers of Acute Hantavirus Infection

To investigate whether the molecular detection of RIOMV was epidemiologically linked to additional hantavirus cases, we selected all AFI samples from Loreto that were negative for DENV during January 2020–2021 (n = 832). We selected DENV-negative serum because acute DENV infection can affect serologic analysis for other pathogens (e.g., SARS-CoV-2) (30). Because a RIOMV-specific IgM ELISA was not available, we used a commercially available IgM ELISA licensed for diagnostic use in Peru that relies on pooled ANDV and SNV nucleoproteins, which are suitable for detecting hantavirus-specific immune responses shortly after infection (31). The IgM ELISA yielded a detection rate of 1.6% (95% CI 0.9%–2.7%; n = 13/832) averaged over 2020 (Appendix 2), including detectable IgM in the patient who tested positive by RT-PCR and exhibited high IgM IFA end-point titers of 1.0 × 10⁴ against SNV and ANDV individually (Table 1; Appendix 1 Figure 3, panel A). Among the 13 samples positive for IgM by ELISA, 30.8% (95% CI 12.4%–58.0%; n = 4/13) were positive by IgM IFA (Appendix 1 Figure 3, panel B). The incidence based solely on IFA results was 0.5% (95% CI 0.07%–1.11%; n = 4/832). All IgM-positive samples, except those positive by RT-PCR, were negative for IgG, which may be attributed to the facts that all patients were febrile and that IgG production is expected during the first weeks after symptom onset (Table 1) (32,33).

In January 2020, the 2 patients who were positive according to IgM IFA, including the patient who was positive according to RT-PCR, were 1 and 5 years of age. In September 2020, the patients who were IgM positive by IFA were a 27-year-old woman and a 66-year-old man. Beyond the patient positive by PCR, IgM end-point titers were generally low at 1.0 × 10¹ against SNV and SNV/ANDV/PUUV, suggesting circulation of hantaviruses antigenically related to SNV/ANDV, potentially including RIOMV (Table 1; Appendix 1 Figure 4). Peru is not in the geographic distribution of the primary host of SNV, the North American deer mouse (Peromyscus maniculatus) (34), and PUUV is not endemic to South America. The reactivity for those viruses is probably explained by cross-reactivity between hantaviral epitopes, as previously described in full virus-based IFA (35).

Figure 4

Figure 4. Investigation of unspecific reactivity in serum positive for hantavirus IgM by ELISA, Peru. A) OD₄₅₀in noncoated ELISA plate. B) Comparison of IgM ELISA reactivity for different arboviruses: Control…

In an IgM ELISA devoid of viral antigen, the optical density (OD) of hantavirus IgM IFA-confirmed serum did not differ significantly from that of IFA IgM-negative serum that had previously tested positive in the hantavirus IgM ELISA (Mann-Whitney U test, U = 20; p = 0.28) (Figure 4, panel A). We therefore compared reactivity with that of a control group composed of serum hantavirus IgM negative by ELISA. The OD of serum hantavirus IgM reactive by ELISA was higher (mean 0.61, SD = 0.28) than that of serum hantavirus IgM nonreactive by ELISA (mean 0.07, SD = 0.09; Figure 4, panel A). Among the IgM-positive samples, 83.33% (95% CI 54.0%–96.5%; n = 10/12) also showed reactivity in real-time PCR for CMV, EBV, and Plasmodium spp. in IgM ELISAs for several endemic arboviruses (OROV, chikungunya, and Mayaro virus) and even for the nonendemic arbovirus CCHFV (Table 2). Arbovirus IgM ELISA reactivity in hantavirus IgM ELISA-reactive serum was consistently higher than that in hantavirus IgM ELISA-nonreactive serum (Figure 4, panel B). To avoid false-positive results, when calculating the IgM detection rate, we included only samples that tested positive for IFA.

Serologic Markers of Past Hantavirus Infection

The scarce hantavirus detections in humans and rodents in Peru all originated from the Amazon Basin. To compare whether persons living in other ecozones had had previous contact with hantaviruses, we tested 830 samples from Loreto, Lambayeque, and Lima that were DENV-negative by qRT-PCR and were of sufficient serum volume for hantavirus IgG testing by using the same ELISA that we used to detect IgM. The hantavirus IgG ELISA detection rate by region was 4.0% (95% CI 2.1%–7.0%; n = 11/278) in Loreto, 1.4% (95% CI 0.4%–3.8%; n = 4/279) in Lima, and 2.9% (95% CI 1.4%–5.8%; n = 8/273) in Lambayeque (Appendix 1 Figure 5, panel A; Appendix 2). We found no significant difference in hantavirus IgG detection rates by ELISA between sites (χ² = 3.32; p = 0.20).

To confirm past hantavirus infections, we again performed IFA-based end-point titration. IgG IFA confirmed 60.9% (95% CI 40.7%–77.9%; n = 14/23) of the samples IgG positive by ELISA (Appendix 1 Figure 5, panel B). The sample positive by RT-PCR had IgG titers of 1.0 × 10⁴ against SNV and ANDV, a finding consistent with early IgG responses in some hantavirus-infected patients (32). End-point titration suggested 7 past infections with viruses antigenically more related to SNV than to ANDV, according to the highest SNV titers in all 3 ecozones. One serum sample each showed monotypic reactivity with SNV or ANDV (Table 3). Another 6 samples positive by IFA had similar titers against >2 hantaviruses, including DOBV/Saaremaa virus, Hantaan virus/DOBV, and SNV/ANDV (Table 3; Appendix 1 Figure 5, panel B). Nine samples IgG positive by ELISA samples were negative by IFA (Table 3).

Figure 5

Figure 5. Investigation of unspecific reactivity in hantavirus in IgG-positive serum by ELISA, Peru. A) OD₄₅₀in noncoated ELISA plate. B) Comparison of IgG ELISA reactivity for different viruses excluding the…

Complementary testing as before revealed that 2 serum samples (Table 4, samples 3562 and 3937) from children <1 year of age that were positive by the hantavirus IgG ELISA but negative by hantavirus IgG IFA were CMV positive by PCR. In an IgG ELISA plate devoid of viral antigens, the ODs in the 2 CMV-positive serum samples were strikingly higher at 1.03 and 2.31 (mean 1.67, SD = 0.9) compared with the rest of the samples that were hantavirus IgG positive by ELISA (mean 0.05, SD = 0.09). The ODs of samples that were hantavirus IgG-positive by ELISA but IgG-negative by IFA, including those 2 samples, differed significantly from those of the control group (Table 4; Figure 5, panel A). After excluding the 2 CMV-positive samples, we detected no significant differences between the ODs of the 3 compared groups. The 2 CMV-positive samples also showed significantly higher reactivity in most complementary ELISAs, including pathogens not found in South America, such as CCHFV and Middle East respiratory syndrome coronavirus (Table 4). After excluding the 2 CMV-positive samples, we found significant differences in ELISA reactivity levels for OROV, SARS-CoV-2, and Plasmodium spp. (Figure 5, panel B), whereas the overall number of positive serum samples did not differ among groups (Appendix 1 Table 8). Thus, we calculated the ELISA-based IgG detection rate without the 2 CMV-positive serum samples. The adjusted hantavirus IgG detection rate for Loreto was 3.2% (95% CI 1.6%–6.1%; n = 9/278;); 66.7% (95% CI 45.2%–82.9%; n = 14/21) of the samples also tested positive by IFA, and the overall IgG detection rate was 1.7% (95% CI 1.0%–2.8%; n = 14/830) across the 3 ecozones.