The utility of diagnostic tests for enteric fever in endemic locations

Abstract

Enteric fever, an infection caused by Salmonella enterica serovar Typhi and serovar Paratyphi A, is common and endemic in many areas of the Asian and African continents. In endemic areas, diagnostic tests are needed to diagnose acute cases for clinical management, to detect convalescent and chronic fecal carriage and for contact tracing. A suitable test may also allow an assessment of disease burden in a community to determine the need for vaccination programs. Each specific role may warrant a dedicated test, utilizing different samples, targets and methods to serve their respective purpose. Current diagnostic methods are poor. Blood culture is insufficiently sensitive and technically demanding, and bone marrow culture, although more sensitive, is infrequently performed. Antibody- and antigen-detection tests lend themselves to point-of-care format but remain insufficiently sensitive and specific for this role. There are concerns about the sensitivity of nucleic acid amplification tests and they have not become widely adopted. However, new approaches using genomics, proteomics, transcriptomics, in vivo-induced antigen and immunoaffinity proteomics-based technologies are being employed to identify new antigens, gene targets and metabolic products that could be used as a basis for more effective diagnostic tests. If novel tests are to be credible and widely used they require rigorous evaluation in endemic areas in studies with appropriate selection of patients, adequate sample sizes and proper attention to a gold standard reference. Here, we discuss the range of methods currently used for diagnosing enteric fever in endemic locations and we suggest new technologies which may improve enteric fever diagnostics over the coming years.

Keywords::

• biomarkers
• culture
• diagnostics
• enteric fever
• metabolomics
• paratyphoid
• PCR
• proteomics
• serology
• typhoid

Typhoid and paratyphoid fever (collectively called enteric fever) are serious systemic diseases caused by the bacteria Salmonella enterica serovar Typhi and other S. enterica serovars, including Salmonella Paratyphi A, B and C [1]. Enteric fever is endemic in low- and middle-income countries, where clean water is lacking and sanitation and hygiene standards are inadequate. In industrialized countries the incidence of enteric fever is low and invariably associated with travel to endemic locations [2]. The most recent estimate from population-based data suggests the annual global incidence of enteric fever is approximately 27 million cases, with 216,000 deaths [3]. The data on which this estimate is based is limited, and comes from isolated studies from countries with a healthcare infrastructure capable of assessing the burden of enteric fever. Accurate figures are compounded by the deficiencies of the currently available diagnostic tests [4,5]. South and Southeast Asia are amongst the most common regions affected by enteric fever, although the current burden in Africa is particularly unclear. Disease burden uncertainty in parts of the African continent reflects a lack of systematic studies and, in some areas with a high HIV prevalence, an enteric fever-like illness caused by invasive infections by other nontyphoidal Salmonella serovars (particularly Salmonella Typhimurium and Salmonella Enteritidis) in infants and in adults is highly prevalent [6,7].

The Gram-negative bacteria that cause enteric fever are transmitted via the fecal–oral route, usually through the consumption of contaminated food or water. These organisms are facultative intracellular pathogens and all are thought to follow a similar infection cycle. Disease is initiated following ingestion of the organisms, colonization of the small intestine, invasion of the gastrointestinal mucosal surface and dissemination throughout the body in the reticuloendothelial system including the liver, spleen and bone marrow [8]. Multiplication and subsequent bacteremia correlate with the onset of the clinical disease. The disease may be complicated by relapse and carriage in the gall bladder, resulting in long-term chronic fecal shedding [9,10]. The complete pathogenesis of enteric fever is still incompletely characterized. These intracellular pathogens use a variety of measures to avoid detection by the host, and these impinge on our ability to detect the organism or detect signs of contact with the organism [11,12]. Antibody- and cell-mediated immunity responses occur after infection and after immunization with live oral typhoid vaccines, yet these may not be specific to the infecting organism [13,14]. There are no satisfactory animal models of enteric fever and much of our understanding of pathogenesis has been derived from the S. Typhimurium mouse infection model and cell culture experiments [12]. Our inability to directly model the mechanism of infection has impaired our ability to develop specific diagnostic tests.

Enteric fever is generally considered a disease of school children and young adults, although there is evidence of a substantial disease burden in preschool children in some countries where disease is endemic [15,16]. Surveillance in India, Indonesia, Pakistan, Vietnam and China from the Diseases of the Most Impoverished (DOMI) study carried out by the International Vaccine Institute (IVI) found enteric fever to be predominantly a childhood illness, with similar incidence rates in preschool and school-aged children [17]. By contrast, in a recent study from Kathmandu, Nepal, the burden of disease was greatest in school-aged children and young adults, possibly because of a large transient workforce traveling from locations outside the city where exposure to the infecting organisms may be less common [18]. S. Typhi is the dominant cause of enteric fever in most areas, although the proportion of infections attributed to S. Paratyphi A has been increasing in the north of the Indian subcontinent and China [19,20].

Food handlers and contaminated water sources are considered to be important for maintaining enteric fever in endemic areas [21,22]. Acute enteric fever cases often coincide with the peak of the wet season in tropical and subtropical locations, consistent with an association with contaminated water, and several outbreaks have been directly linked to contaminated water sources [23]. Disease is presumably maintained in human populations through carriage; S. Typhi and S. Paratyphi A are able to survive for protracted periods in the gallbladder of apparently healthy people [9,10]. Estimates vary, but potentially between 1 and 5% of acutely infected patients may go on to chronically carry and shed the organisms into the local environment [24]. The relative epidemiological importance of direct human contact with those shedding the organisms, either transiently after acute infection or long-term as chronic carriers, and the role these people play in sustaining the disease is unclear [25].

Strategies to control and eventually eliminate enteric fever from an area include: improving the provision of clean water and adequate sanitation, the identification of carriers, and sustained vaccination programs [26]. For governments to make rational decisions about committing resources for this purpose, information concerning the burden of disease in the human population in defined areas is essential. The required information must include detecting the majority of cases. Case detection should incorporate those individuals that have been exposed but experienced an asymptomatic or mild infection that did not require medical attention and those that are transiently of chronically carrying the organism. There is a lack of current diagnostic tests that can serve any of these purposes [5].

There are three main uses for an enteric fever diagnostic test: the diagnosis of acute cases required for clinical management; the detection of convalescence and chronic fecal carriage for contact tracing, and the measurement of acute and convalescent cases for the assessment of disease burden. The patients and bacteria in these three situations have different physiological characteristics. Therefore, a diagnostic method suitable for acute infections may not be applicable to detect convalescent cases or those shedding the organisms postinfection as a result of asymptomatic carriage. Each function may warrant an individual test, utilizing different samples, targets and methods to serve their respective purpose. A rapid, simple, point-of-care test suitable for use in a healthcare center or outpatient clinic may fit the profile required for acute diagnosis. By contrast, burden of disease studies could be based on cross-sectional serological surveys without the need for a rapid result and the same stringency of sensitivity and specificity. Acknowledging these different requirements should be considered in a review of current diagnostics, and the development of future enteric fever tests. Here, we assess the current alternatives for diagnosing enteric fever in endemic settings and discuss potential improvements which may develop over the coming years. We have identified typhoid, paratyphoid and enteric fever diagnostic studies through a systematic search of the literature using PubMed and a hand search of the bibliography of identified papers.

Diagnosing the clinical syndrome

The clinical spectrum of enteric fever ranges from a mild illness with a low-grade fever, malaise and slight dry cough to a severe infection with serious complications, which can include toxic encephalopathy and intestinal perforation leading to peritonitis [24,27]. Apathy, blanching ‘rose spots’ on the trunk, hepatosplenomegaly and diarrhea are additional features that may occur. A recent studies suggests that S. Typhi and S. Paratyphi A infections cannot be distinguished clinically, as S. Paratyphi A causes a disease of equal severity to that of S. Typhi[28]. Most patients in endemic settings probably self-treat by visiting a local pharmacy or shop selling antimicrobials. Those that do not respond to self-treatment are managed in a health center or outpatient clinic and the majority do not require admission to hospital [15,16,29]. The common usage of antimicrobials in the community by patients prior to a visit to a healthcare facility means that the classic ‘textbook’ presentation of enteric fever with a slow ‘step-ladder’ rise in fever with increasing toxicity is less commonly observed and the overall clinical infection is less severe [27,30].

Distinguishing enteric fever from other causes of undifferentiated febrile illness such as dengue, leptospirosis, rickettsial infections and malaria in endemic countries is difficult without a differential diagnostic test. Healthcare workers are reliant on their clinical judgment to make an educated assessment of the cause of illness, and/or prescribe a broad-spectrum antimicrobial which targets several microorganisms that may be causing the observed infection. Clinical algorithms have been developed for febrile disease diagnosis and patient triage in resource-limited endemic areas [31–34]. Such algorithms, some of which use the Widal test result (discussed later), may be useful where there is simply no alternative, yet are hampered by both a lack of sensitivity and may be geographically specific/unspecific depending on the epidemiology of other infections.

When prescribing an antimicrobial, the treating clinician would ideally like to have information about the local circulation of pathogens causing febrile disease and their antimicrobial susceptibility patterns [35]. Unfortunately, this information is usually unavailable and, as a result, unnecessary or inappropriate antimicrobial therapy of unconfirmed enteric fever may be administered. Conversely, suitable therapy for patients with enteric fever (or indeed other infections) may be delayed while other febrile illnesses are being considered. Appropriate enteric fever therapy curtails the disease duration, prevents progression to severe disease, decreases the probability of relapse, and may also prevent shedding of the organism back into the community after recovery from an acute infection [24,30]. Conversely, overtreatment probably exerts a selective pressure on populations of S. Typhi and S. Paratyphi A and potentially contributes to the increasing antimicrobial-resistance levels that have been observed in S. Typhi and S. Paratyphi A in parts of Asia and Africa [35–37].

Microbiological culture

The isolation of S. Typhi or S. Paratyphi A from blood, bone marrow, rose spots or other sterile sites provides the most conclusive confirmation of enteric fever. Therefore, culture should be considered as the gold standard and used for evaluating all diagnostic tests, irrespective of their level of sophistication [38]. Bacterial isolation confirms the clinical diagnosis and allows antimicrobial-susceptibility testing which can direct appropriate therapy. Longitudinal culture information from sentinel locations provides estimates of disease burden, the relative proportion of the various human-specific Salmonella serovars and changing antimicrobial-resistance patterns. Such longitudinal datasets have been particularly important in documenting the increase in multidrug- and quinolone-resistant strains. Culturing has also demonstrated the emergence of S. Paratyphi A in certain geographical regions in Asia and the importance of nontyphoidal serovars in Africa [24,35,36,39]. The bacterial isolate can be characterized, formerly by phage typing, but now by molecular methods to identify the source of local outbreaks, permitting control and containment strategies [23]. Genotyping has highlighted strain circulation strains, particularly those with susceptibility to fluoroquinolones [37,40,41].

Blood culture and microbiological characterization is the mainstay of enteric fever diagnosis, yet it is only positive in approximately 40–60% of presumptive cases [42–49]. Various factors contribute to this lack of sensitivity. The window for detecting organisms circulating in the bloodstream is usually early in the course of the disease and particularly in the first week of the illness [50,51]. The volume of blood taken is critical and relates directly to the number of bacteria in the blood. Studies in Vietnam have documented a median of less than 1 cfu/ml of blood in adults and children with mild-to-moderate typhoid fever [52]. The quantity of bacteria in the bloodstream is higher in children than adults and higher in the first week of illness, compared with later weeks. The type of culture medium is also important. Studies have indicated that bile containing Oxgall media (BD Difco, UK) is a suitable media for recovering S. Typhi and S. Paratyphi from blood and other sites [50,53–55]. The advantage of Oxgall is attributed to the inhibition of the antibacterial activity of fresh blood caused by lysis of blood cells rather than direct enhancement of growth by the bile salts [56]. The limitation of Oxgall is that it does not permit the isolation of other infecting bacteria, and is not suitable as a general purpose blood culture media. Other media types used currently include those containing tryptone soya broth or brain–heart infusion broth, with additional sodium polyethanol sulphonate, or bespoke media developed for automated blood culture systems, such as BACTEC (Becton Dickinson, UK) and BactAlert (bioMérieux, France) [57]. Adequate dilution of blood in broth and the length of the incubation period in the microbiology laboratory are additional important factors. A very common reason for impaired sensitivity is pretreatment with antimicrobials prior to attending a hospital or healthcare facility [58].

Attempts have been made to further optimize blood culture methods, including the use of centrifugation to concentrate the bacteria and then culture the deposit or the Buffy coat onto an agar medium, which requires the incorporation of a lysis step to release intracellular bacteria within phagocytes [46,53–55,57,59–63]. These manipulations may improve sensitivity with a lower amount of starting material, yet are still limited by the absolute quantity of organisms per milliliter of blood. Alternative approaches that accelerate the time to produce a positive result from a blood culture have been used, by a short period of culturing followed by antigen detection or nucleic acid amplification (discussed later) [64–68]. To tackle the problem of antimicrobials, the bottles available for automated systems contain resins, which absorb and inactivate antimicrobials, thus potentially increasing sensitivity. However, such resins are unlikely to improve sensitivity in the face of prolonged antimicrobial usage prior to a blood sample being taken. In pediatric enteric fever study conducted in Pakistan, the authors demonstrated no difference in S. Typhi or S. Paratyphi A yields between anaerobic lytic and pediatric aerobic resin-containing blood culture media [69].

Many studies have demonstrated a high sensitivity (>80%) of culturing bone marrow aspirates even with prior antimicrobial therapy, regardless of the duration of disease prior to sampling [42–49,70]. The increased sensitivity of bone marrow culture, compared with blood culture, is principally due to approximately ten times the concentration of viable organisms in bone marrow than in the blood [71]. In fact, blood culture sensitivity can be comparable to that of bone marrow culture provided a sufficient volume of blood is cultured [55,72]. Despite the increased sensitivity of bone marrow culture, obtaining bone marrow by standard methods is technically challenging, invasive and not generally performed [73]. A modified technique to obtain bone marrow samples that is more acceptable to patients is highly desirable, and the use of low-diameter (fine) needles for this purpose is a possibility [74,75]. A less invasive method for obtaining a bone marrow aspirate may lead to a changed perception of the investigation as being normal for the hospital investigation of suspected enteric fever in the same way that a lumbar puncture is now considered routine for the diagnosis of meningoencephalitis.

Cultures from other sites such as rose spots, feces or rectal swab and urine, are less invasive and can also give positive results [42,44,45,73]. A duodenal string culture is a further option, although young children and those with severe disease may be unable to tolerate the string device [44,45]. A study from Indonesia in 1981 concerning 118 patients compared culturing blood, rectal swabs, bone marrow and duodenal strings, and gave positivity rates of 54, 36, 86 and 58%, respectively [44]. Yet, when all four were combined, the overall diagnosis rate was 98%, demonstrating that diagnostic sensitivity can be improved by culture of multiple sites. However, a positive culture from feces, duodenal contents or urine may require a cautious interpretation, as it may be indicative of acute enteric fever infection, but could also represent chronic carriage, with the acute infection syndrome caused by a different etiological agent.

An additional major limitation in the use of culture for diagnosis is that many endemic countries lack adequate microbiology laboratories [76]. Thus, blood culture is not routinely used to assist in the diagnosis of febrile patients and, even when laboratories are present, they may not comply with quality assurance standards. Encouraging the expansion of bacteriology facilities in endemic areas could have benefits for the diagnosis of enteric fever and other important endemic pathogens [77]. The development of alternative culture systems, at a lower cost and less dependent on expensive consumables, is needed to increase diagnostic capacity [4]. The time taken to generate a culture result is always greater than 24 h and often a result is not available until several days after the blood is taken for culture. Thus, the final culture result will not help healthcare workers in an outpatient clinic to make a decision about management and antimicrobial selection at the time of consultation.

Antibody detection

The Widal test, which measures agglutinating antibodies against lipopolysaccharide (LPS; O) and flagella (H) antigens of the S. Typhi in the serum of individuals with suspected enteric fever, was introduced more than a century ago and is still widely used [78,79]. It is simple and cheap to perform, and with the slide format, rather than tube, it takes only a few minutes. Use of the method has been hampered by a lack of standardization of reagents and inappropriate result interpretation [80–82]. The Widal test ideally requires both an acute and a convalescent-phase serum sample taken approximately 10 days apart, and a positive result is determined by a fourfold increase in antibody titer. However, antibody titers in infected patients often rise before the clinical onset, making it difficult to demonstrate the required fourfold rise between initial and subsequent samples for a confirmatory diagnosis [83–85]. Furthermore, in practice, the result of a single, acute-phase serum sample is often used to help patient management and, although useful in some cases, a single serum result may be confusing in others [74,80,81,85–87]. A single sample test is generally plagued by false-negative and false-positive results. A large number of cross-reacting antigenic determinants of typhoidal and nontyphoidal Salmonella organisms and other Enterobacteriacae are now recognized, as are several other diseases caused by non-Salmonella organisms such as malaria, dengue and TB [79,88]. A knowledge of the background levels of antibodies in the local population may aid interpretation of the Widal test, and it should only be used in patients where there is a reasonably high suspicion (prior probability) of enteric fever [74,86].

ELISAs have been used to more precisely define the normal antibody response and its relevance for diagnosis [87,89–95]. The antigens used in ELISA-based studies have generally been LPS, flagella, Vi or outer membrane protein antigens. ELISAs measuring anti-LPS antibodies and anti-flagella antibodies have been found to be more sensitive than the Widal ‘O’ and ‘H’ test but all are hampered by similar limitations of specificity that occur with the use of the Widal Test.

Chart and coworkers at the Health Protection Agency’s National Salmonella Reference Centre in the UK have developed and evaluated an immunoassay (SDS-PAGE/immunoblotting) for the detection of serum antibodies against LPS and flagellar antigens of S. Typhi as well as S. Paratyphi A, B and C [96]. They evaluated 15 patients culture-positive for S. Typhi, 15 healthy controls and 300 serum samples submitted for Salmonella serodiagnosis. The 15 culture-positive S. Typhi patients demonstrated detectable serum antibodies to the O9 and O12 LPS antigens and ten of these had antibodies to the flagellar antigens. From the 300 reference sera, 22 had a positive antibody response to the O9 and O12 LPS antigens and 12 to the O6 and O7 LPS antigens. This method requires additional validation on a greater number of samples from endemic areas as the target O and H antigens may lead to an obvious lack of specificity.

Antibodies against the Vi antigen generally appear too late in the course of the illness to be helpful for diagnosis [96,97]. However, the detection of IgG to the Vi antigen has been proposed as a method to detect chronic carriers [98–101]. Sensitivities of >75% and specificities of >95% have been reported and the test has proved valuable in the context of outbreak investigations [102]. Its role for detecting carriers in the general population is less clear. In a study in Chile, anti-Vi antibody titers ≥160 had a 75% sensitivity and between 92 and 97% specificity for detecting chronic typhoid carriers, and a positive predictive value of between 8 and 17% in the general adult population [103]. When used in Vietnam for screening a population of more than 3000 adults for potential carriers, nearly 3% of those screened had elevated anti-Vi antibody titers but in none of those could S. Typhi be isolated from fecal samples [104]. The low positive predictive value of the test may limit its use as a general population screening tool for carriage.

Rapid diagnostic tests

Simple, reliable, point-of-care rapid diagnostics for enteric fever have been a long-felt need of clinicians working in endemic countries. Such tests need be robust and suitable for use in remote areas with limited laboratory facilities and the medical staff should not require any specific technical training. The overall utility and uptake of these tests depends on their simplicity. Such tests should have limited steps and be designed to yield a simple ‘positive/negative’ result at preset thresholds, similar to those detecting early pregnancy factor in rapid pregnancy tests. Ideally, the results should be available within 1 h of the initiation of the assay, so that they can be interpreted while the patient is in attendance. Rapid point-of-care diagnostic tests could, potentially, be combined with clinical algorithms for patients to differentiate febrile patients from endemic areas and help guide management.

Several attempts have been made to package serological tests into point-of-care rapid enteric fever tests and those that are currently available have differing methods and formats. Most have been developed to be used with blood (using venous whole blood or serum and/or capillary samples) and detect antibody directed against Salmonella antigens. The antibody class they detect is usually IgM, which is suggestive of a current or recent infection. Some rapid tests detect IgG, which may indicate a current infection or previous exposure.

Typhidot (Malaysian Biodiagnostic Research SDN BHD, Kuala Lumpur, Malaysia), was developed for the detection of specific IgM and IgG against a 50-kD S. Typhi outer membrane protein [105,106]. Typhidot is an ELISA-based method, miniaturized into an immunodot test format. Typhidot-M is a modified version of Typhidot, but detects IgM to the same outer membrane protein and is designed as a more specific marker of current acute infection [107]. TUBEX (IDL Biotech, Sollentuna, Sweden) detects antibody against S. Typhi LPS [108,109]. The specificity has been improved by means of an inhibition assay format and by detecting antibodies to a single antigen in S. Typhi only. TUBEX is similar to slide latex agglutination and it requires mixing of the analyte with reagent latex particles, producing a visual results readout. The S. Typhi-specific antibodies from the suspected patients are detected by their ability to inhibit the binding between colored indicator particles that are coated with a monoclonal antibody (mAb) specific for the S. Typhi O9 LPS antigen, and magnetic particles that are coated with S. Typhi LPS. The results of TUBEX can be read after 5 min following sedimentation. Improvements have been made to enhance the sensitivity of TUBEX, including the use of V-shaped tubes, and the use of two types of reagent particles and a modification for paratyphoid diagnosis [110]. The Royal Tropical Institute of The Netherlands (RTI) has developed a test detecting IgM against LPS using a latex agglutination, dipstick and, most recently, lateral flow [111–114]. Other rapid test kits include an ELISA and dipstick (Multi-Test Dip-S-Ticks) from PanBio (PanBio Indx, Inc., Baltimore, MD, USA) detecting anti-LPS IgG and IgM; Mega-Salmonella (Mega Diagnostics, Los Angeles, CA, USA), which is an ELISA detecting IgG and IgM antibodies against an undefined S. Typhi antigen; SD Bioline Typhoid rapid test (Standard Diagnostics, Kyonggi-do, Korea), which uses an immunochromatographic method to detect IgG and IgM antibodies against another undefined S. Typhi antigen; and a dipstick test named Enterocheck-WB (Zephyr Biomedicals, Goa, India), which detects anti-LPS IgM antibodies [115–118].

Table 1[87,105–108,110–116,118–126] summarizes selected studies investigating the performance of rapid diagnostic tests for typhoid. When compared with blood culture-positive typhoid cases, sensitivities for TUBEX vary between 56 and 100%, between 67 and 98% for Typhidot, between 47 and 98% for Typhidot-M and between 43 and 88% for the RTI test. The sensitivity of these rapid tests improves when paired samples are tested and with increasing duration of illness. Specificities measured against febrile controls with another defined infection and/or healthy nonfebrile controls range from 58 to 100% for TUBEX, 73 to 100% for Typhidot, 65 to 93% for Typhidot-M and 95 to 100% for the RTI test. The sensitivity of rapid tests in febrile patients with possible typhoid but with a negative blood culture vary from 8 to 100% depending on the selected population. In view of the low sensitivity of blood culture, it is impossible to know how many of these ‘clinical cases’ are true cases of enteric fever and such studies are open to investigator bias.

A major drawback of a rapid diagnostic and indeed of any non-culture-based method is the lack of an isolated organism and antimicrobial susceptibility result. Additionally, from the currently available studies, it seems that rapid diagnostic tests still lack sensitivity when compared with culture and the background antibody levels in the general population and the cross-reactive nature of the selected antigens means that such tests continue to lack specificity. Overall, rapid tests may represent some improvement over the Widal test, but are still insufficiently sensitive, specific and consistent to be able to confidently recommend in endemic setting.

Antigen detection

Polyclonal and monoclonal antibodies have been used to detect S. Typhi antigens in bodily fluids. The antibodies used have been directed against the same targets used in serological testing, namely Vi, O9 and Hd. Studies originating from Indonesia, Papua New Guinea and Chile have employed specific antisera against the Vi antigen in a co-agglutination or ELISA format. Sensitivities of 62–97% and specificities of 83–95% have been reported using bacterial antigen detection [74,127]. In the study conducted in Chile, an ELISA gave a false-positive result in 64.7% of 34 culture-proven paratyphoid A or B patients and 47.1% of 21 patients with other nontyphoidal febrile illnesses [128]. A dot-blot ELISA format using a mAB against O9 has also been developed [129,130]. Studies using three serially collected urine samples from suspected enteric fever patients gave sensitivities of >92%, but with specificities of only approximately 71% [131,132].

An ELISA was used to detect Vi, O9 and Hd antigens in urine that was serially collected from enteric fever patients in Egypt [133]. The greatest level of sensitivity was found with Vi, which was positive at baseline (first urine sample taken during a culture-confirmed infection) in 73% of culture-confirmed cases. The O9 and Hd antigens fared less well than Vi, as they were positive in less than 50% of cases. Furthermore, the sensitivity of detecting Vi increased from 73 to 97% when all sequential urine samples were examined. These data indicate the intermittent nature of antigen excretion in the urine during an infection and sustained excretion after the point where the bacteria may be detected in the blood. Unfortunately, the method demonstrated limited specificity, particularly in patients with brucellosis. Studies detecting Vi and other antigens in serum by counter-immunoelectrophoresis and agglutination have also been performed [134–139]. Vi antigen has been detected in the serum of >95% of blood culture-positive cases and between 25 and 90% of patients with clinically suspected, blood culture-negative typhoid.

Nucleic acid amplification

Nucleic acid amplification tests, namely conventional PCR or real-time PCR, have been explored, yet not exhaustively, for the detection of both S. Typhi and S. Paratyphi A from several sterile sites, typically blood. Nucleic acid amplification is generally considered as a fundamental improvement to blood culture, as it is rapid and the small numbers of bacilli present in clinical samples may be amplified, and it may remove the issue of dead or nonculturable bacteria resulting from of pretreatment with antimicrobials. Conventional PCR detects amplification using an agarose gel, whilst real-time PCR amplification is detected by release of a fluorescent signal. Real-time PCR is by far the most robust method of the two, with several advantages over convention PCR, which induce less variability in the assay and the interpretation.

Targets for S. Typhi PCR-based assays have included the Hd flagella gene fliC-d[140], the Ha flagella gene (fliC-a), the Vi capsular gene viaB[141], the tyvelose epimerase gene (tyv; previously rfbE), fliC-d, the paratose synthase gene (prt; previously rfbS), groEL[142] and the 16sRNA gene [143]. Additional studies have used nested primers in an attempt to improve sensitivity, yet nested PCR adds inherent problems with unspecific amplification and contamination. Table 2[140–142,144–158] summarizes selected evaluations of nucleic acid amplification tests. Reported sensitivity in blood culture cases have been invariably >90% with specificities of 100%. Some studies have reported much lower sensitivities ranging from 38 to 42%, consistent with the number of bacteria in the blood [141,158], yet one of these did report a 100% with bone marrow aspirates [158].

Further analysis and understanding of the genomes of S. Typhi and S. Paratyphi A may lead to new and better targets for nucleic acid amplification tests [160]. We know that both S. Typhi and S. Paratyphi A have extremely limited genetic diversity within their populations, a fact which may aid DNA test specificity over other Gram-negative organisms. Between 1 and 3% of the gene content of the S. Typhi and S. Paratyphi A genomes are unique, and even though the genome sequence of S. Typhi CT18 was released in 2001, a significant proportion of the genes remain without known function. It seems that there is scope for further genomic exploration through functional genomics, yet whether gene characterization will lead to nucleic acid based tests is arguable. Developments in DNA-based tests for detecting bacterial pathogens have been rapid over the last 5 years, yet there remains a question of the utility for enteric fever diagnostics. The window and the type of sample are paramount, as the limited quantity and the transient nature of the bacteria in blood will hinder results [4]. An effective ‘capture’ system, with magnetic or nanoparticle technology may significantly enhance sensitivity particularly in stool samples, where the organism may be present in greater numbers and for a longer period than in the blood.

An alternative strategy to reduce the time to diagnosis and increase the sensitivity of PCR is a combination of bacteriology and molecular biology and involves a serotype specific PCR amplification on the blood culture after a short period of incubation [67,159]. The advantage of such a method is an increase in the speed of a positive confirmatory diagnosis, but is unlikely to produce a greater level of sensitivity than that of culturing alone. Although the use of PCR in developing countries will most likely be limited in the medium-term for reasons of cost and overall utility, it is worth noting that molecular biology is becoming more accessible in settings outside research laboratories and many molecular tests can be performed routinely in locations with suitable facilities. Efforts are being made to simplify the process for typhoid PCR using pre-prepared and aliquoted reagents freeze-dried in a single tube [153]. Furthermore, with the development of loop-mediated isothermal amplification (LAMP) there is less of a requirement for thermal cyclers, agarose gels or real-time machines. LAMP takes place in a single tube, at a constant temperature, and amplification detection can be detected by a change in turbidity or color in the reaction vessel [161].

Expert commentary & five-year view

It is clear that all the currently available methods for the diagnosis of enteric fever are limited and there is a desperate need for simple and reliable enteric fever diagnostics, which can inform a clinician of the appropriate treatment and add vital insights into the disease burden.

Humans are the only host for S. Typhi and S. Paratyphi A and many countries where the infection was once endemic for enteric fever have eliminated the disease. Hence global elimination or, indeed, eradication of enteric fever should be possible.

Enteric fever diagnostics represent a paradigm of how technology must be driven by the human and microbiological realities of the natural infection, as the infection has a unique molecular pathogenesis with a specific host response. There are a wide range of potential clinical specimens and possible technological approaches, but methods must be conceived in physiological reality. Therefore, any new development of diagnostics for acute infection needs to address either the low count of bacteria in sterile sites or the cross-reactive nature of any potential antigens.

Diagnosis of acute infection must take into account the relevant differential diagnosis for the setting, be informative concerning drug resistance and locally available treatment options. In considering the product profile of a new rapid test, the specifications developed by the WHO Sexually Transmitted Diagnostics Initiative for the characteristics of an ideal diagnostic test in the developing country context could equally apply to typhoid rapid diagnostic tests. ‘ASSURED’ tests should be affordable by those at risk of infection, sensitive and specific, user-friendly, robust, equipment-free and able to be delivered to those who need it [162].

The results of current research into the identification of signature host–pathogen interactions with S. Typhi will form the basis of new diagnostic tests in enteric fever. S. Paratyphi A cannot be forgotten and should be considered as of equivalent importance to S. Typhi. If a simple test is to be developed, there is a clear need to identify novel markers of infection. Perhaps the most attainable method is the use of serology, requiring the detection of antigens that have a higher level of specificity than those used currently. Serological testing can also lend itself to be modified into a simple rapid diagnostic test and we suggest that these will be at the forefront of enteric fever diagnostics over the next 5 years. The selection, purification and stability of ‘novel’ antigens needs careful consideration, and extensive testing and validation is required to ensure preparations are free from contaminating material and do not exhibit cross-reactivity.

There are several methods currently being applied to investigate novel antigen–antibody interactions and potential new infection markers and genomics, proteomics and immunoscreening are leading the way [160,163,164]. There has been extensive sequencing of S. Typhi genomes, using conventional methods and new-generation technologies, thus allowing a greater understanding of the organism. Genomics has led to follow-on technologies, such as proteomics, and the proteome of S. Typhi has been studied under different culture conditions [164]. However, attention has moved away from antigens expressed in the in vitro-cultured organism to antigens potentially expressed in vivo with the appreciation that in vitro-expressed antigens may not be relevant in the infected patient. Additionally, gene expression and antigen production has been studied in cultured human-derived macrophages [165] and in human samples [166]. Immunoarrays containing a recombinant S. Typhi proteome are being developed and evaluated for investigating novel antigens, which may offer more specificity with serum from patients with acute infection, a successful approach in other infections. Furthermore, using in vivo-induced antigen technology, a group has screened a library of S. Typhi proteins, identifying those that were immunoreactive with convalescent-phase serum after adsorption against in vitro-cultured organisms [167].

Serum and salivary anti-S. Typhi LPS IgA antibody responses have also been examined in enteric fever [168] and antibody secreting cell and antibody in lymphocyte supernantant (ALS) responses that measure the transient presence of mucosal lymphocytes in the peripheral circulation after intestinal activation have been reported in individuals receiving oral attenuated strains of S. Typhi[169]. In a study in Bangladesh, typhoid patients were found to have significant IgA antibody responses against a membrane protein preparation [170]. The membrane preparation contained a cocktail of surface-exposed or cell surface S. Typhi antigens and attempts are currently being made to identify and refine these antigens. The authors also demonstrated anti-S. Typhi membrane protein IgA responses in ALS supernatant in 100% of individuals with blood culture-confirmed enteric fever and 70% of those with a compatible clinical syndrome. ALS requires ex vivo culturing of recovered lymphocytes and contains cells derived from mucosal priming in the gut wall that are transiently present in the peripheral circulation; therefore, this method may not be straightforward. However, using IgA to demonstrate acute infection is a rational and novel approach and obviously warrants further investigation.

Immunoaffinity proteomic-based technology has also used to identify immunogenic antigens and characterize the anti-S. Typhi antibody response in infected patients [171]. Columns charged with IgG, IgM and IgA antibody fractions from patients with S. Typhi bacteremia were used to capture S. Typhi antigens. Antigens were subsequently identified by mass spectroscopy. A total of 57 S. Typhi proteins, including proteins known to be immunogenic (e.g., PagC, HlyE, OmpA, and GroEL) and a number of proteins with high specificity to S. Typhi and S. Paratyphi A (HlyE, CdtB, PltA, and STY1364) were identified and will undoubtedly undergo further screening to confirm specificity.

We expect that such novel approaches will identify proteins that may be used as marker of an acute infection or even as an indicator of exposure. Such antigens need to be screened and evaluated in an appropriate manner. Field evaluations should inform the development process of new diagnostic tests, even at early conceptual stages. Independent field evaluations using the harshest deployment scenario should be imposed to evaluate any diagnostic claiming to be robust enough to meet clinical requirements. This evaluation with an understanding of the sensitivity and specificity of a diagnostic test is central to appropriate use in clinical practice and requires studies with an adequate sample size, inclusion of relevant patient groups and an appropriate gold standard. Unfortunately, many studies evaluating diagnostic tests for enteric fever have been insufficiently rigorous. The patients tested are an inadequate sample of the population at risk of the disease that would actually use the test in daily practice. Compliance with initiatives such as Standards for Reporting of Diagnostic Accuracy and Quality Assessment of Diagnostic Accuracy Studies should help to address inadequate study design [172]. Second, it is crucial that culture of bone marrow is added to blood and fecal culture in diagnostic test evaluation. Furthermore, creating a biobank of relevant samples from a study of patients with enteric fever that includes blood, bone marrow, feces, urine and saliva could also be a way of allowing well-characterized samples to be available for researchers in this field. Biobank curating is an approach that has been successfully used in TB and other infections. Finally, novel Bayesian approaches to analyzing diagnostic studies should be used, providing a scientific evidence base to the development and future use of diagnostic tests for typhoid, particularly in view of the lack of a good gold standard [173].

Table 1. Results of selected studies investigating the sensitivity and specificity of serology assays on blood from patients with suspected enteric fever.

Age group	Suspected enteric fever cases studied (n)	Nonenteric fever controls studied (n)	Test	Test positive in suspected enteric fever cases (sensitivity, %)	Test negative in controls (specificity, %)	Ref.
Child	42 BC ST pos18 BC neg	49 febrile	Typhidot^®	41/42 (98) BC ST pos17/18 (94) BC neg	49/49 (100)	[105]
All	21 BC ST/Widal pos	68 febrile	Typhidot	19/21 (90) BC ST/Widal pos	62/68 (91)	[106]
NS	14 BC ST pos11 BC neg	65 nonfebrile13 febrile	Tubex™	14/14 (100) BC ST pos11/11 (100) BC neg	78/78 (100)	[108]
Child	46 BC ST pos25 BC neg	26 febrile	TyphidotTyphidot-M^®	43/46 (93) BC ST pos7/25 (28) BC neg39/46 (84) BC ST pos13/25 (52) BC neg	20/26 (77)23/26 (88)	[119]
Child	28 BC/stool pos34 BC neg	72 febrile	TyphidotTyphidot-M	25/28 (89) BC ST pos31/34 (91 BC neg25/28 (89) BC ST pos31/34 (91) BC neg	67/72 (93)67/72 (93)	[107]
Child	64 BC ST pos	63 febrile	RTI LPS IgM DipTubex	35/45 (77) BC ST pos56/64 (87) BC ST pos	36/38 (95)48/63 (76)	[87]
NS	91 BC/BM ST pos36 BC/BM neg	80 febrile	RTI LPS IgM Dip	70/91 (77) BC ST pos4/36 (11) BC neg	77/80 (96)	[111]
ND	50 BC ST pos50 BC neg	44 febrile (21 due to nontyphoidal Salmonellae)	TyphidotTyphidot-MPanBio ELISA	41/50 (82) BC ST pos18/50 (36) BC neg49/50 (98) BC ST pos39/50 (78) BC ST pos12/50 (24) BC neg	32/44 (73)38/44 (86)	[115]
All	112 BC ST pos6 BC SPA pos61 BC neg	64 febrile633 nonfebrile	RTI LPS IgM Dip	73/112 (65) BC ST pos4/6 (67) BC pos SPA8/61 (13) BC neg	64/64 (100)631/633 (99)	[112]
NS	25 BC ST pos	55 febrile (25 BC pos Brucella spp.)	RTI LPS IgM Dip	22/25 (88) BC ST pos	53/55 (96)	[120]
All	59 BC ST pos	20 febrile	Dip-S-Ticks IgGTyphidotTubex	51/57 (89) BC ST pos46/58 (79) BC ST pos43/55 (78) BC ST pos	9/18 (50)17/19 (89)17/18 (94)	[116]
Community	103 BC ST pos	113 febrile	TubexTyphidot-M	58/103 (56) BC ST pos41/87 (47) BC ST pos	99/113 (88)33/40 (83)	[121]
ND	39 BC ST pos506 BC neg	ND	Typhidot	36/39 (92) BC ST pos42/545 (8) BC neg	ND	[122]
All	73 BC ST pos3 BC pos SPA	254 febrile	RTI LPS IgM LA	31/73 (43) BC ST pos1/3 (33) BC SPA pos42/95 (44) BC neg	246/254 (97)	[113]
AllCommunity	13 BC ST pos21 BC SPA pos1698 BC neg	ND	TubexTyphidot-M	9/13 (69) BC ST pos1/21 (5) BC SPA pos88/1698 (5) BC neg7/13 (54) BC ST pos2/21 (10) BC SPA pos146/1698 (9) BC neg	ND	[123]
All	75 BC ST pos	102 febrile	TubexTyphidot-MSD Bioline-IgMMega-Salm IgM	71/75 (95) BC ST pos41/55 (55) BC ST pos4//58 (69) BC ST pos68/75 (91) BC ST pos	82/102 (80)66/102 (65)73/92 (79)50/102 (49)	[118]
Child	34 BC ST pos209 BC neg	57 nonfebrile	Tubex™	31/34 (91) BC ST pos37/209 (18) BC neg	51/57 (90)	[124]
AllCommunity	43 BC ST pos800 BC neg	24 febrile	TubexTyphidot	26/43 (60) BC ST pos163/800 (20) BC neg29/43 (67) BC ST pos192/800 (24) BC neg	14/24 (58)13/24 (54)	[125]
ND	54 BC pos62 BC neg	93 febrile	RTI LPS IgM ICT	32/54 (59) BC ST pos40/62 (65) BC neg	91/93 (98)	[114]
All	58 BC SPA pos14 BC ST pos	53 nonfebrile	Tubex-PATM	47/58 (81) BC pos SPA7/14 (50) BC ST pos	52/53 (98)	[110]
Child	41 BC pos64 BC neg	ND	Typhidot-M	38/41 (93) BC ST pos24/64 (38) BC neg	ND	[126]

BC: Blood culture; BM: Bone marrow; Dip: Dipstick; ICT: Immunochromatographic test; LA: Latex agglutination; LPS: Lipopolysaccharide antigen; ND: Not done; Neg: Negative; NS: Not stated; Pos: Positive; RTI: Royal Tropical Institute of The Netherlands; SPA: Salmonella enterica serovar Paratyphi A; ST: Salmonella enterica serovar Typhi.

Table 2. Results of selected studies investigating the sensitivity and specificity of nucleic acid amplification on blood from patients with suspected enteric fever.

Age group	Suspected enteric fever cases studied (n)	Nontyphoid controls studied (n)	Test	Target	PCR positive in suspected enteric fever cases (sensitivity)	PCR negative in controls (specificity)	Ref.
NS	12 BC ST pos4 BC neg	10 febrile	Nested PCR	fliC-d	11/12 (92%) BC ST pos4/4 (100%) BC neg	10/10 (100%)	[140]
NS	21 BC ST pos63 BC neg	20 nonfebrile	PCR	fliC-d	21/21 (100%) BC ST pos4/63 (6%) BC neg	20/20 (100%)	[144]
ND	82 (28 BC ST pos)	20 nonfebrile	PCR	ND	59/82 (72%)	20/20 (100%)	[145]
NS	20 BC ST pos20 BC neg	ND	Nested PCR	fliC-d	20/20 (100%) BC ST pos12/20 (60%) BC neg	ND	[146]
All	10 BC ST pos63 BC neg	ND	PCR	fliC-d	10/10 (100%) BC ST pos36/63 (57%) BC neg	ND	[147]
All	34 BC ST pos3 BC neg	ND	PCR	hilA	34/34 (100%) BC ST pos3/3 (100%) BC neg	ND	[148]
Child	17 BC ST pos46 BC neg	25 nonfebrile	Nested PCR	fliC-d	17/17 (100%) BC ST pos36/46 (78%) BC neg	25/25 (100%)	[149]
Child	26 BC ST pos177 BC neg	ND	PCR	viaB	10/26 (38%) BC ST pos12/177 (5%) BC neg	ND	[141]
All	68 BC ST pos51 BC neg	12 febrile	Nested PCR	fliC-d	67/68 (99%) BC ST pos26/51 (51%) BC neg	12/12 (100%)	[150]
Child	14 BC ST pos38 BC neg	11 febrile8 nonfebrile	Nested PCR	fliC-d	14/14 (100%) BC ST pos29/38 (76%) BC neg	19/19 (100%)	[151]
NS	8 BC ST pos47 BC neg	26 nonfebrile	TaqMan^®RT-PCR	fliC-d	8/8 (100%) BC ST pos40/47 (85%) BC neg	26/26 (100%)	[152]
All ages	33 BC pos broths	40 BC neg broths	PCR	ST-50	29/33 (88%)	40/40 (100%)	[153]
All ages	16 BC ST pos1 BC SPA pos43 BC neg	ND	Nested multiplexPCR	tyvprtfliC-dfliC-aviaB	16/16 (100%) BC ST pos1/1 (100%) BC SPA pos25/43 (58%) BC neg	ND	[142]
NS	51 BC ST pos7 BC neg	ND	PCR	STY-0312/SPA-2476STY-0313-0316	48/58 (83%)	ND	[154]
All ages	6 BC ST pos285 BC neg	10 febrile	Nested	fliC-d	6/6 (100%) BC ST pos8/285 (3%) BC neg	10/10 (100%)	[155]
NS	13 BC ST pos	2 febrile3 nonfebrile	PCR	ratAviaBtyv	13/13 (100%) BC ST pos	5/5 (100%)	[156]
All ages	78 BC ST pos742 BC neg	ND	PCR	fliC-d	73/78 (94%) BC ST pos95/742 (13%) BC neg	ND	[157]
All ages	54 BC ST pos46 BC SPA pos50 BC neg28 BM ST pos	25 febrile	Multiplex RT-PCR	STY0201SSPA2308	23/54 (42%) BC ST pos18/46 (39%) BC SPA pos0/50 (0%) BC neg28/28 (100%) BM ST pos	25/25 (100%)	[158]

BC: Blood culture; BM: Bone marrow; ND: Not done; Neg: Negative; NS: Not stated; Pos: Positive; RT-PCR: Real-time PCR.

Key issues

• • A sensitive and specific diagnostic test for enteric fever is a requirement for clinicians working in endemic areas. Bacterial culture is rarely available where enteric fever is common and differentiating the causes of fever in children and adults in endemic areas is a daily challenge.

• • A simple test for enteric fever, and tests for other common causes of fever, would allow appropriate clinical management and also permit the rational use of antimicrobials.

• • The agents of enteric fever have no known animal reservoir and are limited to areas with poor sanitation, thus the disease should be a target for elimination. We are currently unable to define the disease burden owing to the limitations of current diagnostics.

• • The low numbers of enteric fever bacteria in the blood and the antigenic cross-reactivity of Gram-negative microorganisms means that developing new methods is a considerable challenge and must be conceived in physiological reality.

• • New technological approaches utilizing genomics, proteomics, transcriptomics, in vivo-induced antigens and immunoaffinity proteomics are being employed to identify new antigen and metabolic products that could be developed as more effective diagnostic tests.

• • Independent field evaluations using the harshest deployment scenarios should be imposed on any diagnostic claiming to meet the needs of clinicians caring for enteric fever patients.

• • The lack of an adequate gold standard is a major handicap for developing new enteric fever diagnostic tests. It is crucial that culture of bone marrow is added to blood and fecal culture in diagnostic test evaluations.

• • The creation of a biobank of relevant samples from a study of patients with confirmed enteric fever is a way of allowing well-characterized samples to be available for researchers.

Financial & competing interests disclosure

The authors have no relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript. This includes employment, consultancies, honoraria, stock ownership or options, expert testimony, grants or patents received or pending, or royalties.

No writing assistance was utilized in the production of this manuscript.