Current Global Status of Dengue Diagnostics

Aims: Dengue Virus is a re-emerging infectious disease that is transmitted through mosquitos. Dengue is a significant health concern because of the number of people it affects globally. Clinical diagnosis of dengue is not possible because the symptoms are similar to other febrile-diseases. Therefore, the only way


INTRODUCTION
Dengue virus (DENV) is present in many countries around the world and is continuing to spread [1]. Dengue fever (DF) also known as "break bone fever" is caused by a singlestranded RNA virus that is transmitted through Aedes species mosquitos, primarily Aedes aegypti. There are four dengue virus serotypes (DENV-1, DENV-2, DENV-3 and DENV-4) and many different genotypes that occur worldwide. Different genotypes can occur in similar places at different times indicating multiple DENV introductions [2].

1997 World Health Organization (WHO) Dengue Case Classification System
Dengue virus can lead to the more serious forms of disease severity; dengue hemorrhagic fever (DHF) and dengue shock syndrome (DSS). These classifications were developed by the WHO [3]. The WHO 1997 case definition for DF include a probable, confirmed, and reportable definition. A probable DF case is an acute febrile illness with at least two of the following signs and symptoms: headache, retro-orbital pain, myalgia, arthralgia, rash, hemorrhagic manifestation or leukopenia and supportive serology or in close proximity to a confirmed case's location and time [3]. A confirmed DF case is based on laboratory criteria which would include virus isolation from patient serum or autopsy sample, fourfold or more change in reciprocal immunoglobulin M (IgM), immunoglobulin G (IgG) titers to a dengue antigen, dengue antigen or genome in live or deceased serum, tissue, or cerebrospinal fluid (CSF) via immunochemistry, immunofluorescence, ELISA or RT-PCR [3]. A reportable DF case is one that is either probable or confirmed [3]. The 1997 case definition of DHF requires that a patient present with the following conditions and symptoms; fever or history of one for 2-7 days which can be biphasic, thrombocytopenia (100,000 cells per mm 3 or less), evidence of plasma leakage from vascular permeability because of a rise in haematocrit greater than or equal to 20% above their normal, and one of the following: a positive tourniquet test, petechiae, ecchymoses, purpura, haematemesis, melaena, or a bleeding site [3]. The 1997 WHO case classification system defined DSS as the four criteria for DHF and a rapid and weak pulse with narrow pulse pressure due to hypotension, cold, clammy skin, and restlessness [3]. These definitions are difficult to use because they are not universally applicable, nor are they used consistently thus the development of a new classification system was needed [4,5].

2009 WHO Dengue Case Classification System
The new 2009 WHO classification system was developed to make the definitions more applicable to children by using signs over symptoms and to make it easier for clinicians to predict patients who may develop severe dengue [6,7]. The 2009 system includes definitions for dengue with or without warning signs and severe dengue. Probable dengue cases include: someone living in or has traveled to a dengue endemic area with fever and experiencing at least two of the following signs and symptoms: nausea, vomiting, rash, aches, pains, positive tourniquet test, any warning sign, and supportive serology or a confirmed case near the location or time [8]. Warning signs include abdominal pain and/or tenderness, persistent vomiting, clinical fluid accumulation, mucosal bleeding, lethargy, restlessness, liver enlargement of greater than or equal to 2 centimeters, increased haematocrit with rapidly decreased platelet count [8]. A severe dengue case is defined as any of the following: severe plasma leakage, severe bleeding that was evaluated by a clinician, or severe organ involvement [8]. However, these clinical case definitions are sometimes indistinguishable from other febrile illnesses therefore requiring laboratory confirmation [4]. Both of these systems have been reviewed, to further understand their applicability and usefulness, by [9][10][11][12][13].

Laboratory Tests
Laboratory confirmation can come from a number of different assays or tests. Dengue virus is comprised of 10 proteins; 3 are structural and 7 non-structural. One of the non-structural proteins, NS1, has been targeted for use in diagnosing dengue in humans because this protein is secreted in the bloodstream and can be found on day 1after the onset of fever [14,15]. This viral antigen can be detected via enzymelinked immunosorbent assays (ELISA) and immunochromatographic tests (ICT). These tests are available in rapid form and can detect NS1 up to 9 days after onset [16]. NS1 can be detected before viral RNA because it is produced so quickly [17,18]. Furthermore, some studies have shown that NS1 is proportionally associated with disease severity, i.e. the more NS1 is detected, the more severe is the dengue disease [19]. However, other studies have not shown a relationship between NS1 and disease severity and also that the NS1 test could not discern primary and secondary infections from each other [19].
Haemagglutination assays (HA) are also used to detect the presence of dengue virus antigen. ELISAs and ICTs can also use IgM, IgG, and/or immunoglobulin A (IgA) as markers to diagnose dengue. There are two types of ELISA for diagnosing dengue. One detects antibodies (IgM, IgG, or IgA) and the other detects antigen (NS1). However, ELISA, like other serological tests, has issues with specificity and cross-reactivity, but it can be more useful than the haemagglutination inhibition assay (HI) in distinguishing between primary and secondary infections [19,20]. HI can detect anti-dengue antibodies as well. This test has been regarded as a gold standard with great sensitivity while having the same issues as ELISAs regarding cross-reactivity and not being able to identify virus serotype specific antibodies [18,21]. HI also has the ability to screen many samples easily. The principles behind HI and antibody-capture ELISAs are different. Antibodycapture ELISAs sandwich antibodies and antigen on a plate, while HI uses red blood cells to agglutinate antibodies and antigen. Both still however detect the presence of dengue antibodies.
Antibody titers vary throughout the illness and between primary and secondary infections as shown in Fig. 1 [22]. Fig. 1 illustrates the level of various immunological markers that diagnostic tests can use to determine dengue status. Knowing this can help one understand which test to use and why certain tests can only be used during certain points of the disease progression [22]. For instance, if the specimen was collected early in the infection before an antibody response has occurred, then serological tests like ELISA and plaque reduction neutralization test (PRNT) may result in false negatives as shown in Fig. 1 [18,22]. When an antibody response has occurred, PRNT can identify serotype specific antibodies unlike ELISA and HI [18]. PRNT can distinguish between serotypes because each serotype of DENV is tested with the sample individually on a semi-solid media that allows the virus to propagate.
If a sample was collected early enough during viremia, when viral load would still be high, other diagnostic tests can be used to detect the presence of viral RNA to diagnose dengue, such as the reverse transcriptase-polymerase chain reaction (RT-PCR). RT-PCRs can show the results in real time (rRT-PCR), they can be nested (two pairs of primers for one locus) or semi-nested, and they can be simplex (testing for one serotype) or multiplex (testing for multiple) [23]. RT-PCRs can identify the virus's serotype. This test is sensitive, however, it is very technical requiring sophisticated instrumentation in a laboratory setting, highly trained personnel, and is very costly [18]. Virus isolation can also be used for diagnostic and typing purposes, but this approach can take a few days to grow the virus, and dengue is heat-labile which requires care in specimen handling [24]. One major issue with RT-PCR and virus isolation is the short viremic period which is required for diagnosis through these tests [7]. These tests can be used in different situations. RT-PCR, HA, HI, and PRNT require lab settings while ELISAs and ICTs can be used in the field and at point of care settings [18]. A summary of these tests' advantages and disadvantages has already been complied by the WHO Special Programme for Research and Training in Tropical Diseases (TDR) as shown in Fig. 2 [16]. Most of these tests are complex and require some degree of technical expertise and equipment to conduct. Therefore, companies have been creating rapid tests to try and address these issues. These rapid tests are not without their shortcomings however. Therefore, the purpose of this review is to update a metaanalysis, written by Stuart D. Blacksell, by comparing and contrasting how well commercial diagnostic tests perform [22].
Most of these commercially available rapid diagnostic tests can use various bodily fluids and tissues as samples such as whole blood, serum, and plasma, and some can use urine, CSF, and saliva. Serum is most commonly used because can differentiate primary and secondary infections via seroconversion of antibodies [16]. It is also the most well established, which saliva and urine are not. Urine is also rarely used because it cannot be used for IgM detection, a commonly used analyte [19]. CSF is only used when neurological symptoms have occurred, if not, the antibodies would be too low to detect. The rapid tests generally use serum, plasma, or whole blood for these

Fig. 2. Advantages and disadvantages of dengue diagnostic tests [16]
viremia and antibodies can be easily detected paired (acute and convalescent samples) can differentiate primary and secondary infections via seroconversion of antibodies [16]. It is also the most wellestablished, which saliva and urine are not. Urine is also rarely used because it cannot be used for sed analyte [19]. CSF is only used when neurological symptoms have occurred, if not, the antibodies would be too low to detect. The rapid tests generally use serum, plasma, or whole blood for these reasons. These tests are someti accurate and therefore require a confirmatory test.

Accuracy Calculations
Sensitivity (SN), specificity (SP), positive predictive value (PPV), and negative predictive value (NPV) are calculations that can be used to determine the accuracy of a test. The formulas for these calculations are shown in Table 1. reasons. These tests are sometimes less accurate and therefore require a confirmatory Sensitivity (SN), specificity (SP), positive predictive value (PPV), and negative predictive value (NPV) are calculations that can be used to determine the accuracy of a test. The formulas for these calculations are shown in Table 1

Current Diagnostic Methods
Currently, Brazil uses serology to diagnose dengue. They generally do not use RT-PCR or NS1 ELISA often because they are costly and require highly trained personnel [

MATERIALS AND METHODS
The terms, "dengue diagnostics" and "evaluation" were searched on the database Science Direct between 2011-Present. This resulted in 976 relevant articles. The content filter was applied to only produce articles from journals. This left 760 articles which were then organized by relevance. Of the 760 articles the first 100 articles' abstracts were read to determine relevance to this review. Inclusion criteria: Articles that were included in this review mentioned a specific test being evaluated. Only articles written in English were considered for this review. Exclusion criteria: Articles that studied concurrent infections of dengue with another disease were not included. Of the articles reviewed, 10 were additionally included in this report with the meta-analysis [22]. These same procedures were used on the databases PubMed, Academic Search Premier, Google Scholar, and the Directory of Open Access Journals which resulted in an additional 5 articles. This resulted in 16 total articles that are included in this analysis along with the aforementioned meta-analysis. A brief schematic of this is shown in Fig. 3. Fig. 3 shows the process of including and excluding articles that was used for each database search. Among these articles was a meta-analysis. The studies in this meta-analysis were out of the original year range, but were still included in this review for thoroughness.

RESULTS AND DISCUSSION
Of the 15 articles and one meta-analysis included in this review 5 were conducted in Thailand, 4 in each Brazil and Vietnam, 3 in Malaysia, 2 each in Martinique, Mexico, India, Singapore, Sri Lanka and Taiwan, and 1 in Bangladesh, China, Colombia, French Guiana, the United States of America (USA), and Venezuela. Most of these countries are endemic with dengue except the USA and China. However, some states in China are endemic [26]. Each of the studies tested the accuracy of at least one commercially available assay as listed in Table 2. Most of the serological/virological tests are either ICT or ELISA. Three tests are molecular real time RT-PCR kits. All of the tests accept serum as the sample material. Not all companies' websites listed the sensitivity and specificity of their tests. Panbio Dengue Early Rapid kit should be used alongside other dengue serological tests [31].  Panbio Dengue IgM Capture ELISA does not require serial dilutions, has break-apart wells, and color-coded reagents [35]. Panbio Pan-E ELISA has break-apart wells [36]. SD BIOLINE Dengue Duo can detect dengue specific antibodies and antigen simultaneously in acute or convalescent serum, plasma, or whole blood [37]. SD Dengue IgG Capture ELISA and SD Dengue NS1 Ag ELISA include all of the required reagents for the test [38,39]. SD Dengue IgM Capture ELISA has been included in the WHO procurement scheme [40].
Both Bio-Rad products detect NS1 in human serum or plasma [41,42]. Omega's PATHOZYME Dengue IgM kit is an ELISA suitable for screening that has color-coded reagents and coated plates [43]. Geno-Sen's DENGUE 1-4 Real time PCR kit is to be used with the Rotor Gene 2000/300/6000 [44]. Real Star Dengue RT-PCR kit 1.0 can be used with various real-time PCR platforms [45]. Simplexa Dengue Kit is to be used with the 3M Integrated Cycler [46]. DENV Detect IgM Capture ELISA by In BiOS is the first in vitro test to have Food & Drug Administration (FDA) approval and was developed with the new generation of CDC licensed recombinants expressed in mammalian cells [47]. Dengue Fever IgG and IgM combo Device by Merlin can be used to determine primary or secondary infection, but is not allowed for sale or use in the USA [48]. ASSURE Dengue IgA Rapid Test is a rapid point of care test that is the only IgA ICT included in this review [49]. ASSURE, SD BIOLINE Dengue Duo, IMMUNO Quick, and Dengue Eden Test Bioeasy show results in 10-20 minutes unlike Panbio Pan-E ELISA, Panbio Dengue IgM Capture ELISA, and Panbio Dengue Early ELISA which show results in 2 hours and 10 minutes [15, [35][36][37]49,50].
The results of the literature review are summarized on Tables 3-7. The Tables all include the country where the study was conducted, the sample size (n), SN, SP, PPV, and NPV. Tables 4 and 5 include the analyte because these combo tests can use more than one analyte to determine dengue status. Table 7 also includes analyte because many tests are listed and each test has a unique analyte it detects. The country was listed on the Table as a means to conduct comparisons and to understand the population being studied. Sample size was included on all tests as an indicator of validity. Sensitivity and specificity were included because these are the characteristics in which the accuracy of the test is being evaluated in this review. PPV and NPV are included because these calculations also describe the performance of a test, but are not being used for comparison in this review. Table 3 summarizes the results found through the literature review about the STRIP by Bio-Rad. Table 4 is a summary of the studies that evaluated the accuracy of the Platelia Dengue NS1 Ag made by Bio-Rad. Table  5 summarizes what many studies have found when assessing the accuracy of the SD BIOLINE Dengue Duo NS1 antigen and IgG and IgM combo device. Table 6 shows the results of different studies conducted on Panbio Dengue Duo Cassette. Table 7 summarizes many different tests across many studies. Almost every study has included the sensitivity of the diagnostic assays they tested and almost all have included the specificity. About half of the studies reported PPV and NPV. Some of the studies did not report specificity, PPV, and NPV; however they were calculated if the study provided the raw data. Each calculation not included in the original article is marked with an "*" on the Tables. The STRIP by Bio-Rad was the most tested assay according to this literature review (Table 3). Platelia Dengue NS1 Ag (Table  4), BIOLINE Dengue Duo NS1 antigen and IgG and IgM combo device (Table 5), and Panbio Dengue Duo Cassette (Table 6) were also studied very often. The sensitivities and specificities of the Panbio Dengue Duo Cassette and the BIOLINE Dengue Duo NS1 antigen and IgG and IgM combo device were reported on each different analyte and on all of them.

Performance by Country
Multiple studies were conducted in Brazil and the Bio-Rad NS1 STRIP had the best overall sensitivity at 99.5%. When evaluated in a different study, this test was only sensitive 91% of the time [26, [51][52][53]. Bio-Rad's Platelia NS1 ELISA came in a very close second (99.3%) with regards to sensitivity [51]. Master Diagnostic's BioEasy Eden, the Bio-Rad NS1 STRIP, and the Panbio Early Rapid were the most specific tests in studies that occurred in Brazil [26, [51][52][53]. In Colombia, the SD BIOLINE Dengue Duo test performed the best with regards to sensitivity and specificity [54]. Platelia NS1 ELISA was the most sensitive test in French Guiana (82.4%) and Panbios's Pan-E Early ELISA was the most specific (97.9%) [55]. In India, the Panbio Dengue Duo Cassette had the best sensitivity when detecting IgG, but when detecting IgM it had the best specificity across both studies (75%) [56,57].The three studies conducted in Malaysia have found that the Bio-Rad NS1 STRIP is the most sensitive (90.4%) and specific test (99.5%) [51,58,59]. All of the tests evaluated in Martinique had 100% specificity except the RealStar RNA kit (98%) and the most sensitive test was the Simplexa RNA kit(93.2%) [60,61]. Since Simplexa is an RNA kit, it would not be efficient in any capacity (cost and ease) to use in Martinique, therefore the best test to use, that is not an RNA kit, would be Bio-Rad's Platelia NS1 ELISA (sensitivity: 61.2%, specificity: 100%) [60].

Performance by Test
It is of interest to note that the companies promote different accuracies for their tests than have been illustrated in this article. Sensitivity varied between a 1% difference to as much as 27%. Companies' statements about test specificty overall are very accurate. However, a few tests were off by 2%-26%.

CONCLUSION
A limitation of this review and its analyses is the accuracy of the papers it cites. Some papers were more detailed than others about what tests they used, therefore, including some data in the analysis was impossible. Some articles did not report specificity, PPV, NPV nor the raw numbers (TP, FP, TN, FN) which made it impossible to include these tests in our analysis. Articles also relied on the accuracy of the reference test. These tests may not be 100% sensitive and specific therefore making the accuracy of the tests they reported on higher or lower than its true accuracy and our analysis less meaningful. Another limitation to this review is the limited number of articles evaluating the accuracy of commercial kits in non-endemic areas. Only two studies were conducted in non-endemic countries and one was conducted in the endemic area of the country.  Overall when evaluating Performance by country, the Bio-Rad NS1 STRIP seems to have performed the best. There may be some bias in this statement because it was evaluated more often than any other test. Also, many factors could have contributed to this bias as well. For example, if the study only used disease-free samples, then they would be less rigorous than the studies that used dengue-negative /Flavivirus-positive samples because of the cross-reactivity that could occur. Flaviviruses can be concurrently endemic, which requires a very sensitive test, and therefore requires this type of rigorous evaluation. Overall when evaluating performance by test, the Standard Diagnostics' (SD) BIOLINE dengue duo was most specific in Vietnam (100%) and most sensitive in Sri Lanka (92.9%) [22,54,63,75]. Standard Diagnostics' (SD) Dengue NS1 Ag ELISA was most sensitive in Malaysia (76.8%) and most specific in Thailand (98.6%) [54,68,81]. That said, the sensitivity of this test was not very good at all (44.8%-76.8%) [54,68,81]. Overall, most tests performed the best in Brazil, but again, this statement is biased due to the fact that Brazil evaluated many tests, which means it has the opportunity to be the best more often.
Infection with dengue virus can lead to severe disease or it can be asymptomatic. Many diagnostic tests are available to help determine the dengue status of individuals. The purpose of this review was to compare and contrast the accuracy of commercial dengue diagnostic tests. The best performing test across all studies and countries seems to be Bio-Rad NS1 STRIP. Those who evaluate the accuracy of commercial tests should consider using the Bayesian Latent class model as described in Pan-ngum et al., 2013 and elsewhere because this model does not assume a perfect test. Taking away this assumption would present a better understanding of the true accuracy of a test. One thing is clear; these commercially available diagnostic tests should be evaluated more often and across more populations in a standardized way to create reliability.

ETHICAL APPROVAL
The authors declare that this work was not against public interest.