Characterization of Dengue Virus 4 Cases in Paraguay, 2019–2020

In 2019–2020, dengue virus (DENV) type 4 emerged to cause the largest DENV outbreak in Paraguay’s history. This study sought to characterize dengue relative to other acute illness cases and use phylogenetic analysis to understand the outbreak’s origin. Individuals with an acute illness (≤7 days) were enrolled and tested for DENV nonstructural protein 1 (NS1) and viral RNA by real-time RT-PCR. Near-complete genome sequences were obtained from 62 DENV-4 positive samples. From January 2019 to March 2020, 799 participants were enrolled: 253 dengue (14 severe dengue, 5.5%) and 546 other acute illness cases. DENV-4 was detected in 238 dengue cases (94.1%). NS1 detection by rapid test was 52.5% sensitive (53/101) and 96.5% specific (387/401) for dengue compared to rRT-PCR. DENV-4 sequences were grouped into two clades within genotype II. No clustering was observed based on dengue severity, location, or date. Sequences obtained here were most closely related to 2018 DENV-4 sequences from Paraguay, followed by a 2013 sequence from southern Brazil. DENV-4 can result in large outbreaks, including severe cases, and is poorly detected with available rapid diagnostics. Outbreak strains seem to have been circulating in Paraguay and Brazil prior to 2018, highlighting the importance of sustained DENV genomic surveillance.


Introduction
Dengue virus (DENV) is the most common arbovirus worldwide, with 50-100 million symptomatic infections resulting annually from four related viruses, designated DENV types 1-4 [1].The reported epidemiology and relative severity of DENV-4 have differed between regions and patient populations, with predominantly secondary cases and less severe disease reported in Southeast Asia compared to a mixture of primary and secondary onset.Cases were classified according to the 2009 WHO criteria as dengue without warning signs (DWS-), dengue with warning signs (DWS+), and severe dengue (SD) [30].

Clinical Samples and DENV Testing
Serum was obtained at the enrollment visit, aliquoted, and stored at −80 • C. Participants were screened for DENV by testing for the non-structural protein 1 (NS1) antigen and/or DENV RNA in a multiplex rRT-PCR for Zika, chikungunya, and dengue (the ZCD assay) [33,39,40].NS1 testing was performed at IICS-UNA using the Standard Q Dengue Duo rapid immunochromatographic test (SD Biosensor, Suwon, South Korea) according to manufacturer recommendations.Screening test results, both positive and negative, were confirmed in a DENV type-specific, quantitative rRT-PCR (the DENV multiplex test, DMPT) [41].The Standard Q Dengue Duo rapid immunochromatographic test also detects anti-DENV IgM and IgG.Results of antibody detection were recorded but not incorporated into the diagnostic algorithm of acute dengue cases.
The ZCD assay and DMPT were performed at both IICS-UNA and Emory University, following shipment of sample aliquots on dry ice.At IICS-UNA, RNA was extracted from 140 µL of serum using the Viral RNA Mini Kit (Qiagen, Germantown, MD, USA) and eluted into 60 µL of buffer, according to manufacturer recommendations.At Emory, total nucleic acid extraction was performed using either (1) an EMAG instrument (bioMérieux, Durham, NC, USA) or (2) the MagMaxViral RNA Isolation Kit in a KingFisher Apex system (both from ThermoFisher Scientific, Waltham, MA, USA).For automated extractions, nucleic acids were extracted from 200 µL of serum and eluted in 60 µL of buffer.A total of 5 µL of eluate was then used in ZCD and DMPT reactions, and both assays were performed and interpreted as previously described [39][40][41].Serum viral load was quantified from 4-point standard curves prepared with synthesized DENV target sequences and included on dedicated DMPT runs.

Case Definitions
Dengue case confirmation required a positive result in the DMPT.Cases that (1) tested negative for DENV in the ZCD assay or (2) had a positive screening test (NS1 or ZCD) that could not be confirmed in the DMPT were considered other acute illness (OAI).This case definition was employed to ensure rigorous confirmation of dengue cases with at least two different tests.

DENV Sequencing
Sixty-two 2019-2020 samples were selected for sequencing from individuals with confirmed DENV-4 infections and DMPT Ct values < 35.Samples were further selected to represent the distribution of all cases based on month of collection, city of residence, and severity of clinical illness.All samples from SD cases that met the Ct criterion were selected.A single DENV-4 case collected in 2018 as part of this ongoing study was also sequenced and included in phylogenetic analyses [33].
Extracted total nucleic acid underwent heat-labile dsDNase treatment (ArcticZymes, Tromso, Norway).cDNA was synthesized using random hexamer primers and SuperScript III RT (both from ThermoFisher Scientific) for first strand synthesis and New England Biolabs (New England Biolabs, Inc., Ipswich, MA, USA) reagents for second strand synthesis, without amplification.Sequencing libraries were fragmented and indexed using the Nextera XT DNA Library Prep kit (Illumina, San Diego, CA, USA) with dual indexes and 16 cycles of PCR.Libraries were quantified using the KAPA universal complete kit (Roche, Basel, Switzerland), pooled to equimolar concentration, and sequenced on a MiSeq with paired-end 150-bp reads (Illumina, San Diego, CA, USA).As a negative control, water was included with each batch of samples starting from DNase.As a positive control, in vitro transcribed ERCC spike-ins (NIST) were added to each sample prior to cDNA synthesis.
Maximum-likelihood (ML) phylogenies were estimated with IQ-TREE (version 1.6.12)with ultrafast bootstrap approximation to evaluate clade probabilities.ModelFinder was used to select the GTR+F+gamma4 nucleotide substitution model [43].Temporal signal was assessed using TempEst v1.5.1 [44], and 12 reference sequences with >0.01 distance from the best-fitting linear regression were excluded as outliers for possible low sequencing quality or misclassified dates.Downsampling was performed from this alignment to yield a set of unique sequences with high genome coverage of predominantly the same genotype identified in this study (genotype II; see Supplemental Material for complete details).Our final dataset included 61 DENV-4 sequences generated by our group from 2019 to 2020, 1 DENV-4 sequence generated by our group from 2018, 9 reference sequences from Paraguay in 2018, and 129 globally representative DENV-4 genotype II reference sequences.The final ML phylogenetic tree was rooted on the oldest DENV-4 sequence.

Statistical Analysis
Basic statistical analyses were performed using Excel software version 2312 (Microsoft, Redmond, WA, USA).Comparisons between group means and medians were made by ANOVA, Welch's test, both pooled and non-pooled two-sample t-tests, and Kruskal-Wallis tests.Comparisons of proportions were made using chi-squared tests or Fisher exact tests.Graphs were prepared with GraphPad Prism version 9 (GraphPad, San Diego, CA, USA).Crude associations and statistical analysis were performed using SAS version 9.4.Significance was set at two-sided p-values ≤ 0.05 for all analyses.

Geographical Distribution of Studied Cases
Participants included in the current study were enrolled between February 2019 and March 2020, and the distribution approximately mirrored country-wide numbers of suspected dengue cases, both confirmed and unconfirmed, reported to the Ministerio de Salud Pública y Bienestar Social, Paraguay (Figure 1).Patients from 14 of 17 departments and the capital district of Paraguay were included (Figure 2A).Most dengue (229/253, 90.5%) and OAI cases (501/546, 91.8%) came from the Central Department or capital district (Table S1).Dengue cases were confirmed among individuals who resided in 9 departments and the capital district (Figure 2).pected dengue cases, both confirmed and unconfirmed, reported to the Ministerio de Salud Pública y Bienestar Social, Paraguay (Figure 1).Patients from 14 of 17 departments and the capital district of Paraguay were included (Figure 2A).Most dengue (229/253, 90.5%) and OAI cases (501/546, 91.8%) came from the Central Department or capital district (Table S1).Dengue cases were confirmed among individuals who resided in 9 departments and the capital district (Figure 2).

Study Population
Seven hundred ninety-nine participants were enrolled and met inclusion criteria.This included 253 (31.7%) confirmed dengue and 546 (68.3%)OAI cases (Table 1).Dengue cases were older (mean 36.1 years, standard deviation (SD) 20.1) than OAIs (27.9, SD 19.3; p < 0.001) but were similar in gender makeup, comorbid illnesses, and days of symptoms trict (Table S1).Dengue cases were confirmed among individuals who resided in 9 departments and the capital district (Figure 2).

Phylogenetic Analysis
DENV sequences in the final alignment included 138 reference and 62 newly generated sequences from Paraguay: 61 from 2019 to 2020 (Tables S2 and S3, Figure S3) and a single sequence from 2018 (NCBI-GenBank accession numbers: OP811915-OP811976).All newly generated DENV-4 sequences belonged to genotype II.Reference sequences represented samples collected from 1956 to 2018, including sequences from Asia and the Americas (Table S4).
In the ML phylogenetic analysis, all Paraguay sequences clustered together (Figure 4A, demarcated with a dashed line box), and outbreak sequences were most closely related to 2018 sequences from Paraguay, which clustered just basal to the sequences from this study (Figure S4).Clades A and B differed by three synonymous single nucleotide polymorphisms, one each in the envelope, NS3, and NS5 genes.In ML analysis, Clade A appeared to be more closely related to the 2018 DENV-4 sequence generated for the current study, but in Bayesian analysis, that sequence was confirmed as ancestral to both.There was no phylogenetic clustering of cases by severity (Figure 4B), geographic location, or epidemic wave.
In time-scaled phylogenetic analysis, outbreak sequences again clustered together with high support and shared a most recent common ancestor in August 2017 (95% highest posterior density (HPD) February 2017-February 2018; Figure 4).Our inferred mean clock rate of 8.79 × 10 −4 (95% HPD 7.85-9.80× 10 −4 ) is slightly higher than the median reported rate in prior studies on DENV-4, 7.91 × 10 −4 , but well within the range of reported rates, 6.89 × 10 −4 to 20 × 10 −4 [46].All Paraguay sequences shared a common ancestor in January 2017 (95% HPD February 2016-September 2017) and diverged from their most recent ancestor, KP188564_Brazil_2013, in May 2011 (95% HPD June 2010-June 2012).These results suggest that there was unappreciated circulation of the outbreak lineage between 2011 and 2017.To assess whether the lineage was captured in prior studies of partial genome sequencing, we analyzed 743 DENV-4 reference sequences from the BV-BRC database collected between 2012 and 2018 with at least full envelope sequences (1485 bp), and we found no additional closely related reference sequences.

Discussion
From the end of 2019 to early 2020, DENV-4 caused the largest DENV outbreak in Paraguay's history [31,35].In our study population, dengue cases were poorly detected with available rapid diagnostics, associated with certain clinical manifestations, and some progressed to SD.The relative clinical severity of DENV-4 has varied in prior studies, which may reflect differences in virus strains and/or patient populations.In studies from Southeast Asia with documented transmission of all four DENV types, DENV-4 is often the least common, predominantly detected among secondary cases and associated with lower severity than other types [9,10,12].SD risk with DENV-4 may also be lower in the Americas, particularly compared to DENV-2 [3,4,[13][14][15].However, consistent with our findings, DENV-4 still causes SD, with an overall risk similar to DENV-1 [11,13,15] and increased risk among older patients and those with secondary infections [4][5][6][7][8][9]11,13,47].
Phylogenetic analysis indicated that the DENV-4 lineage responsible for the 2019-2020 outbreak was nearly identical to viruses detected in Paraguay in 2018, consistent with a recent phylogenetic study of DENV in the country [36].All Paraguay DENV-4 sequences shared a most recent common ancestor in 2017, and this aligns with a molecular clock analysis on 2018 DENV-4 sequences that estimated viral introduction into Paraguay in September 2017 [19].Thus, this large outbreak was not due to introduction of a new lineage into the country in 2019 but instead resulted from local DENV-4 evolution and emergence in a susceptible population [22,23,25].While we did not observe fixation of nonsynonymous mutations among DENV-4 sequences from Paraguay, these sequences all differed from their closest ancestor (KP188564_Brazil_2013) by seven amino acids, and the evolutionary history of this lineage over the decade preceding the outbreak is unclear due to limited DENV-4 sequences from the country and region.
DENV-4 has undergone multiple introductions into South America over the past 40 years, and genotype II, as identified in our study, has been predominant [5][6][7]16,18,20,21,24,25,47].After being absent for three decades, DENV-4 was detected in Brazil in 2010 and resulted in explosive epidemics in the following years, probably because of the population's sus-ceptibility [5,48].Similar to findings in Puerto Rico, previous research hypothesized that DENV-4 re-emergence or re-introduction in the state of Roraima in 2010 was preceded by cryptic or imperceptible circulation of the virus [20,49].Several studies have demonstrated that densely populated states like Sao Paulo and Rio de Janeiro play a key role in the spread of DENV-4 to other Brazilian locations [48,50], and notably, the MRCA for DENV-4 strains in Paraguay was detected in Sao Paulo state.Autochthonous DENV-4 evolution that precedes viral re-emergence and the ongoing risk for introduction of new strains, such as genotype I introduction into Brazil from Asia [20], highlights the importance of sustained genomic monitoring to trace the origin of new outbreaks.
It is notable that DENV-4 emerged in Paraguay in a population where DENV-1 had been predominant for the previous four years [33], as both the change in the predominant DENV type and waning cross-protective immunity could have contributed to the high numbers of symptomatic infections seen in 2019-2020 [51,52].The DENV-1/DENV-4 order of infections has been observed among SD cases [53], and prior DENV-1 infection has disproportionately contributed to SD elsewhere [54].The wave dynamics observed in 2019-2020 fit with the arrival of DENV-4 outbreak strains in a susceptible population relatively late in the DENV transmission season that ended in early 2019 in Paraguay [55], and consistent with this, genetic differences were not observed between the two waves.All sequenced SD cases were detected in the second wave.However, these did not cluster in phylogenetic analyses, and this finding may be attributed to higher case numbers in the second wave increasing observed SD by chance.
Rapid NS1 testing demonstrated poor sensitivity (52.5%) for dengue cases caused by DENV-4.This was lower than the sensitivity of the same assay observed during the 2018 DENV-1 outbreak in Asunción (71.4%), though specificity was high (>96%) in both studies [33].These data are similar to findings from Brazil, where rapid immunochromatographic tests for NS1 resulted in under-detection of DENV-4 [26][27][28]56].Poor NS1 performance may result from lower levels of NS1 in DENV-4 cases, though data to this effect are sparse [57], or high seroprevalence of anti-DENV IgG.Sensitivity of NS1 detection may improve with heat dissociation of IgG-NS1 complexes [5,56], but this requires instrumentation and detracts from the benefits of point-of-care testing.Due to NS1 test performance, this was implemented only as a screening test to determine further work-up by rRT-PCR.NS1 rapid tests continue to be a widely used tool in clinical practice, particularly in sites with limited resources due to simplicity and relatively low cost [58,59].However, it is important to consider the potential clinical and epidemiologic impact of their reduced sensitivity in comparison to DENV RNA detection shown in this work, particularly for DENV-4.This emphasizes the necessity of developing point-of-care diagnostic tests with improved performance features [60,61].
The clinical presentation of dengue cases differed from that of OAIs in this population, with arthralgia, myalgia, and nausea reported significantly more often among cases and cough and sore throat reported less often.DENV-4 has previously been associated with cutaneous manifestations when compared to other DENV types [34].Although rash was also more common among dengue cases in our population, this did not remain significant after adjustment for multiple comparisons.Sore throat was also less common among DENV-1 cases from Asunción in 2018, when dengue cases more commonly experienced headache and conjunctivitis [33].Although these remain relatively general complaints, identification of such symptom constellations will aid clinicians in the judicious use and interpretation of available diagnostics.

Limitations
This study focused on acute symptomatic dengue cases.Therefore, results may not be generalizable to mild or subclinical DENV-4 infections, and primary/secondary infection status could not be fully characterized.Second, although participants resided in 14/17 departments in Paraguay, over 90% of individuals lived in Asunción or the Central Department, which impacts the power of phylogenetic studies to detect regional differences in DENV-4 sequences.Third, the genomic record for DENV-4 in Paraguay dates back only to 2018 and is limited in South America as a whole.This complicated analyses of DENV-4 introduction into Paraguay and the emergence of the two clades identified in the current outbreak.Nevertheless, this study provides important information on a large dengue epidemic that occurred in an endemic country like Paraguay and could serve to improve our understanding of dengue epidemiology in the region.

Conclusions
Findings from the 2019-2020 DENV outbreak in Paraguay highlight the capacity of DENV-4 to cause explosive outbreaks and the need for sustained genomic monitoring of circulating DENV strains in a population.DENV-4 is poorly detected with available rapid diagnostics and, without high rates of symptomatic disease and widespread molecular testing, may remain under-reported and insufficiently characterized.

Supplementary Materials:
The following supporting information can be downloaded at: https://www.mdpi.com/article/10.3390/v16020181/s1,Supplemental methods.Downsampling of DENV-4 sequences.S1.Department or district of residence of dengue and other acute illness cases.Table S2.Data for DENV-4 samples sequenced in this study.Table S3.Demographic data for participants from whom DENV-4 whole genome sequences were obtained.Table S4.Sequences retrieved from NCBI GenBank included for the phylogenetic analysis.

Institutional Review Board Statement:
The study was conducted according to the guidelines of the Declaration of Helsinki.The study protocol was reviewed and approved by the Scientific and Ethics Committee of the Instituto de Investigaciones en Ciencias de la Salud, Universidad Nacional de Asunción (IICS-UNA, IRB00011984) and the Emory University Institutional Review Board (IRB00000569).
Informed Consent Statement: Written informed consent was obtained from all subjects.Parents or legal guardians provided consent for children, and children older than six years of age provided assent.
Data Availability Statement: Upon publication, data supporting the presented results will be made freely available through the Emory Dataverse, which is an open data repository offered through a partnership between Emory and the Odum Institute at the University of North Carolina at Chapel Hill.

Figure 1 .
Figure 1.Suspected dengue cases reported in the country and enrolled acute cases during the study period.(A) All suspected dengue cases reported to the Ministerio de Salud Pública y Bienestar Social, Paraguay by month.The number of cases for February 2020 is shown above the broken bar.(B) Acute cases included in the current study by date of symptom onset.

Figure 2 .
Figure 2. (A) Map of Paraguay displaying departments from which dengue cases were enrolled (dark orange) or other acute illness cases were enrolled but dengue was not identified (light orange).Three departments from which no cases were enrolled are shown in grey (from west to east: Ñeembucú, Misiones, Caazapá).(B) Map of dengue cases detected in the study by city in the capital district (Asunción) and surrounding area.Maps were prepared using (A) Mapchart (www.mapchart.net;date accessed, 23 August 2022) and (B) Google Maps, 2022 (www.google.com/maps;date accessed, 23 August 2022).

Figure 1 .
Figure 1.Suspected dengue cases reported in the country and enrolled acute cases during the study period.(A) All suspected dengue cases reported to the Ministerio de Salud Pública y Bienestar Social, Paraguay by month.The number of cases for February 2020 is shown above the broken bar.(B) Acute cases included in the current study by date of symptom onset.

Figure 1 .
Figure 1.Suspected dengue cases reported in the country and enrolled acute cases during the study period.(A) All suspected dengue cases reported to the Ministerio de Salud Pública y Bienestar Social, Paraguay by month.The number of cases for February 2020 is shown above the broken bar.(B) Acute cases included in the current study by date of symptom onset.

Figure 2 .
Figure 2. (A) Map of Paraguay displaying departments from which dengue cases were enrolled (dark orange) or other acute illness cases were enrolled but dengue was not identified (light orange).Three departments from which no cases were enrolled are shown in grey (from west to east: Ñeembucú, Misiones, Caazapá).(B) Map of dengue cases detected in the study by city in the capital district (Asunción) and surrounding area.Maps were prepared using (A) Mapchart (www.mapchart.net;date accessed, 23 August 2022) and (B) Google Maps, 2022 (www.google.com/maps;date accessed, 23 August 2022).

Figure 2 .
Figure 2. (A) Map of Paraguay displaying departments from which dengue cases were enrolled (dark orange) or other acute illness cases were enrolled but dengue was not identified (light orange).Three departments from which no cases were enrolled are shown in grey (from west to east: Ñeembucú, Misiones, Caazapá).(B) Map of dengue cases detected in the study by city in the capital district (Asunción) and surrounding area.Maps were prepared using (A) Mapchart (www.mapchart.net;date accessed, 23 August 2022) and (B) Google Maps, 2022 (www.google.com/maps;date accessed, 23 August 2022).

Figure 3 .
Figure 3. Quantifiable DENV-4 serum viral load (A) distribution for dengue cases in the study po ulation and (B) by day of symptoms at presentation.
All ten Paraguay sequences from 2018 differed from the outbreak sequences by only two synonymous mutations, one in the NS3 gene and the other in NS5.The closest reference sequence from outside Paraguay came from a sample collected in São José do Rio Preto, Brazil in 2013 (KP188564.1).Outbreak strains comprised two clades, designated clade A (n = 24) and clade B (n = 37) (Figure 4B, tree branches shown in different shades of blue).

Figure 4 .
Figure 4. MCC tree of DENV-4 genotype II.(A) Full tree with 200 sequences from the final 61 DENV-4 from Paraguay, 2019-2020 (this study); 1 DENV-4 sequence from Paraguay, 20 study); 9 reference sequences from Paraguay in 2018; and 129 globally representative DEN quences.(B) Magnified and cropped MCC tree to show Paraguay 2019-2020 outbreak seq and closest references.SD cases are indicated with a solid red box; all other dengue cases a cated with an empty box.Years are indicated by vertical lines.Nodes for the MRCA are label the month and year (95% confidence interval).

Figure 4 .
Figure 4. MCC tree of DENV-4 genotype II.(A) Full tree with 200 sequences from the final dataset: 61 DENV-4 from Paraguay, 2019-2020 (this study); 1 DENV-4 sequence from Paraguay, 2018 (this study); 9 reference sequences from Paraguay in 2018; and 129 globally representative DENV-4 sequences.(B) Magnified and cropped MCC tree to show Paraguay 2019-2020 outbreak sequences and closest references.SD cases are indicated with a solid red box; all other dengue cases are indicated with an empty box.Years are indicated by vertical lines.Nodes for the MRCA are labeled with the month and year (95% confidence interval).

Table 1 .
Demographic and clinical data for dengue cases versus other acute illness cases (N = 799).
Abbreviation: SD, standard deviation.a Data presented as n/N (%).

Table 1 .
Demographic and clinical data for dengue cases versus other acute illness cases (N = 799 Abbreviation: SD, standard deviation.a Data presented as n/N (%).

Table 3 .
Symptoms reported in the preceding 7 days among dengue and other acute illness cases.
a Results presented as n/N (%): the number of participants who reported a symptom (n) over the total number