A need to raise the bar — A systematic review of temporal trends in diagnostics for Japanese encephalitis virus infection, and perspectives for future research

Highlights • Japanese encephalitis virus (JEV) remains a leading cause of neurological infection in Asia.• A systematic review identified 20,212 published human cases of laboratory-confirmed JEV infections from 205 studies.• 15,167 (75%) of cases were confirmed with the lowest confidence diagnostic test, i.e., level 3 or 4, or level 4.• Only 109 (53%) of the studies reported contemporaneous testing for dengue-specific antibodies.• A fundamental pre-requisite for the control of JE is lacking — that of a simple and specific diagnostic procedure that can be adapted for point-of-care tests and readily used throughout JE endemic regions of the world.


Introduction
The mosquito-borne flavivirus Japanese encephalitis virus (JEV) accounts for an estimated 68,000 cases of Japanese encephalitis and 709,000 disability adjusted life years annually (1, 2). Japanese encephalitis virus (JEV) primarily affects children in rural areas when JEV-infected mosquitoes feed on humans rather than their primary amplifying hosts, pigs or reservoir hosts, i.e. aquatic birds (3).
Sustained efforts from international agencies have supported the introduction of immunisation programmes into routine health control schedules in countries with endemic JEV transmission, Table   1 (4). The evidence suggests that vaccination has had an impact on JE incidence (4)(5)(6)(7)(8)(9)(10). However, JEV remains a leading cause of neurological infection in endemic countries, and the Joint World Health Organisation (WHO)/United Nations Children's Fund (UNICEF) surveillance data do not substantiate the improvements cited in the past ten years, with sustained numbers of reported patients over this period, see Figure 1. Table 1: Country-specific data on the introduction of Japanese encephalitis virus vaccine in JE endemic countries. Data adapted from the CDC report by Heffelfinger et al. 2017 and updated with WHO surveillance data (4,11,12).   (13). Data include probable* and laboratory confirmed cases reported by JE endemic countries *WHO definition of a probable case (14) = A case that meets the clinical case definition for acute encephalitis syndrome (AES) that occurs in close geographical and temporal relationship to a laboratory-confirmed case of JE, in the context of an outbreak. Note that these data represent only reported cases, and are not considered to be a true representation of global JE incidence. Weaknesses of these data are discussed in the main text.
JE cases reported to WHO/UNICEF have important limitations. For example, increased awareness of the disease and access to laboratory capacity may contribute to increased case reporting. Conversely, surveillance data are likely to represent only a small proportion of patients (15). This is particularly relevant for JE, occurring predominantly in rural areas lacking diagnostic capacity (16). There are no rapid or point-of-care tests for JE in clinical use (17), and the WHO recommended standard diagnostic assay is an ELISA test that requires trained professionals, appropriate resources and several hours for the results of the tests to be obtained (18). In a survey performed by WHO/UNICEF in 2017, 21 countries responded, of which 11 met the minimum surveillance standards (19,20). Equally, there are problems of specificity of the most widely used diagnostic test, JE MAC-ELISA (21). This is an increasing issue, with increasing endemicity of other flaviviruses and vaccination coverage.
The reasons for persistence of JE as a public health problem are complex and multifactorial. A key principle that must be kept in mind is that JE is a zoonotic infection, human immunisation will never eradicate it in the natural environment, and therefore sustained vaccination coverage is necessary.    Angola. The studies incorporated a variety of methods for the diagnostic tests, including conventional and novel methods, as summarised in Table 3. The data do not provide evidence of change in the certainty of diagnosis through time, see Table 4.

J o u r n a l P r e -p r o o f
Overview of JE diagnostic testing The first isolation of JEV was in 1934, when Hayashi demonstrated that a filterable agent inoculated into monkeys produced encephalitis (31). The experiment was performed using homogenised brain, obtained at post-mortem, from a fatally-infected child who presented with encephalitis in Tokyo during the 1924 epidemic. Early studies relied on clinicopathological correlates in infected humans, when compared with those observed following animal inoculation of post-mortem samples.
Subsequently, hamster, porcine, and human cell culture systems were developed which revealed cytopathic effects when inoculated with JEV-infectious specimens (32,33). As these procedures improved, mosquitoes and mosquito-cell cultures were added to the resources for isolation and identification of JEV (34,35). Cerebrospinal fluid (CSF) and other body fluids were also included for analysis (36,37). Subsequently, JEV antigen detection procedures including complement fixation, immunofluorescence microscopy of cells in CSF, reverse passive haemagglutination and staphylococcal coagglutination (38)(39)(40)(41) were added to the list of diagnostic tests. Nonetheless, assays involving direct virus detection are minimally useful for the diagnosis of JE as the level of viraemia is usually low and the virus is detectable only briefly early in the infection (42).
In the mid-twentieth century, investigation of the antigenic properties of JEV soon led to the development of serological assays including complement fixation (43), inhibition of haemagglutination (44), and virus neutralisation tests (45,46). Early reports of human infection in 1947 used a seroneutralisation technique in which a patient sample was mixed with virus and inoculated into mice (45,47,48). In 1941, Casals and Palacios published a report on the application of the complement fixation technique (49). The method was used for many years, although it was insensitive, particularly during the acute illness (50). Accordingly, in 1958, Clarke and Casals published a report on the application of the haemagglutination inhibition test (HI) (44). The principle exploits the fact that JEV envelope protein agglutinates erythrocytes. Anti-JEV antibodies, developed following infection, bind to JEV protein and thus prevent erythrocyte agglutination, hence the term haemagglutination-inhibition. This remained the method of choice for JE diagnosis, by serological methods, for many years (51)(52)(53)(54) and was subsequently adapted as a more convenient microtiter J o u r n a l P r e -p r o o f method (34,46,55). However, limitations in the sensitivity and specificity of the assay were recognised by Clarke and Casals (44). Moreover, the test relies on the combined paired results obtained from acute and convalescent serum samples, thus taking weeks for confirmation (56). Other serological methods such as single-radial haemolysis (57)(58)(59) were also introduced. However, inadequacies were readily acknowledged and it was accepted practice to perform these tests in parallel with others, thus increasing the workload and extending the time for results to be obtained (60,61).
The plaque-reduction neutralisation test (PRNT) was subsequently developed as the gold-standard for

Specific findings of JEV diagnostics review
Studies reporting on the use of seroneutralisation, IgM ELISA and RT-PCR are discussed below, since these assays are, at present, the ones most widely incorporated into clinical diagnostics.
Seroneutralisation assays: Thirty-two of these included articles identified evidence of JE using neutralisation assays, see   Since JEV is a human pathogen with high individual risk, seroneutralisation has to be performed in biosafety level 3 laboratory, placing additional burdens on time, cost and qualified personnel. Another potential complication may arise when sera from patients who have previously been exposed to JEVrelated flaviviruses may contain higher titres against the closely related flaviviruses than against JEV ("doctrine of original antigenic sin") (100). For example, the titres of anti-YFV neutralising antibodies were higher than anti-JEV neutralising antibodies in JE patients who had previously received the yellow fever vaccine (101). Similarly, in a study testing West Nile virus and JEV, 18 patients' data remained equivocal due to high levels of antigenic cross-reactivity between these viruses (91). The neutralisation test may only be strictly applicable as the gold-standard for vaccine efficacy studies, in which a baseline serum sample is compared with a convalescent sample taken at a fixed interval 1-3 months later. For the purpose of confirming acute JEV, neutralisation is an imperfect gold standard.

J o u r n a l P r e -p r o o f
Severe constraints on being able to arrange for sample testing by neutralisation, and the results being interpretable without cross-reactive positivity due to other flaviviruses (which is relatively rare in JEV endemic areas), impede 'neutralisation confirmation'. The neutralisation titres obtained may be affected by the particular strain of challenge virus utilised (102). A final issue with the neutralisation test is the inability to detect non-neutralising antibodies, thus potentially reducing the analytical sensitivity (103). Therefore, the practicalities of PRNT and diagnostic yield when testing field samples can be low, although the specificity is potentially high. However, there does appear to be higher analytical sensitivity in studies that used nested and heminested techniques as compared with single RT-PCR, but these techniques are notoriously prone to J o u r n a l P r e -p r o o f contamination causing false positive results. It is also recognised that the sensitivity of nucleic acid detection (and protein) detection will continue to increase as technology improves (106,107).
Evidence to suggest that this will be the case arises from the high cycle threshold (Ct) of patients that are confirmed, and the fact that blood donor transmission has been seen in WNV patients who were negative by RT-qPCR tests (108). Recent detection of JEV RNA in throat swabs of JE patients suggests that this non-invasive sample may marginally improve diagnostic yield (109). There have now been two cases of JE, confirmed by RT-PCR, that were first identified by metagenomic nextgeneration sequencing (mNGS) (110,111); the first detection of JEV RNA in human urine, and JEV detection in serum from an African patient with a co-YFV infection. The latter was not confirmed by an orthogonal method, and remains questionable. Nonetheless, unbiased mNGS technology (see below), application and reporting will continue to improve, and could potentially detect JEV in novel locations (112).

Requirements of a new test for the detection of JEV infection
CNS infections are challenging syndromes to diagnose and treat, even in the most highly resourced centres (113,114). It is estimated that they may be caused by >100 different pathogens, including novel and emerging pathogens (115). Current approaches to diagnosis in routine clinical practice Evidence suggests that the secreted viral JEV non-structural protein 1 (NS1) is present at very low concentrations in serum or CSF, unlike in dengue (164,165). A novel alternative approach would be to analyse the host response, using transcriptomics or proteomics. However, questions of specificity J o u r n a l P r e -p r o o f and also how these would be translated into point-of-care tests, would require detailed investigation and the development of innovative methodologies.
In summary, whilst the diagnosis of JE has been possible for many years, it still requires specialised high containment laboratories and appropriately trained scientists and therefore cannot be reliably carried out in many resource limited regions where JEV is endemic/epidemic. A fundamental prerequisite in the public health strategy for the control of JE is lacking, that of a reliable and simple diagnostic procedure that can be adapted for point-of-care tests, and readily available for use throughout JEV endemic regions of the world. Improved diagnostic capabilities throughout JEV affected areas will not only benefit individual patients (through accurate diagnosis) but lead to higher quality surveillance data and better understanding of the distribution of JE risk, enabling improved targeting and evaluation of interventions. The lack of diagnostic capabilities for JE is a barrier to understanding the true disease burden and impact of public health strategies.
J o u r n a l P r e -p r o o f   Virus isolation by innoculation of any specimen in cell culture or animal with characteristic cytopathic effect and confirmation by detection of JEV RNA or virus antigen.
JEV virus antigen detected from brain tissue or CSF by immunofluorescence or immunohistochemistry