The present invention relates to methods and devices for the diagnosis of serious bacterial infection.

Serious bacterial infection (SBI) is an important contributor to mortality and morbidity in children, in particular in resource-poor countries. Child deaths attributable to invasive Streptococcus pneumoniae and Haemophilus influenzae type b infections alone are estimated to equal those attributable to HIV, Tuberculosis and malaria combined. The prompt recognition of SBI in children is challenging clinically, especially so in the context of co-infection with malaria and HIV, and is of the utmost importance in ensuring both effective, and rational antibiotic treatment. Diagnostic uncertainty in bacterial sepsis is compensated by the overuse of antibiotics, as delays in diagnosis of severe sepsis could lead to death. The over-prescription of antibiotics has the potential to lead to antibiotic resistance, which is an increasing public health problem.

Biomarkers may be used alone or in combination to allow classification of an individual to a unique group with defined characteristics. Biomarkers such as C-reactive protein (CRP) and procalcitonin have been shown to improve the accuracy of clinical diagnosis in serious infection with a variable degree of success. Of a panel of five markers investigated for their ability to detect SBI in Malawian children, procalcitonin (PCT) has previously been demonstrated to be the most discriminating.

Biomarkers which reflect the host response to infection, are useful adjuncts in confirming the diagnosis of sepsis, in addition to good clinical acumen, and improved molecular diagnostic techniques. It is desirable to identify biomarker panels that will prove to be useful in differentiating sepsis-associated inflammation from non-infectious inflammation.

The performance of biomarkers is traditionally assessed using receiver operator characteristic (ROC) curves, which evaluate the discrimination of a test (or combination of tests). However, such techniques may be insensitive to the addition of a new biomarker, especially if a good biomarker is already included in the model. Net reclassification improvement quantifies improvement in the model as a result of adding one or more new biomarkers. This approach has been verified in cardiovascular disease and acute lung injury as improving the accuracy of risk prediction. l Procalcitonin (PCT) is a precursor of calcitonin, and is a 1 16 amino acid protein with a molecular mass of 13 kDa. PCT is produced rapidly in response to inflammatory stimuli and has been extensively studied in the context of sepsis.

Resistin is an adipokine first related to insulin resistance in mice. In humans, resistin expression occurs predominantly in macrophages, and monocytes, and more recently has been reported in neutrophils. Systemic inflammation induces a dramatic increase in circulating resistin with levels peaking 8-16 hours after the administration of LPS in healthy subjects, mediated by pro-inflammatory cytokines and NF-κΒ. Resistin is significantly elevated in intensive care patients with sepsis, and elevated resistin levels correlate with severity of disease.

Neutrophil gelatinase-associated Lipocalin (NGAL) - or lipocalin 2 - is a member of the lipocalin protein family, and has a role in innate immunity through its ability to bind siderophores, required by many bacterial pathogens to scavenge iron from the host. Like resistin, it is highly upregulated by inflammatory stimuli via NF-κΒ. NGAL is more significantly elevated in sepsis than in the systemic inflammatory response syndrome (SIRS). In a large study of adult patients with suspected sepsis presenting to an emergency department, which aimed to define a biomarker panel to predict organ dysfunction, shock, and in-hospital mortality, the optimal 3-marker panel was NGAL, protein C, and interleukin-1 receptor antagonist. The area under the curve for the accuracy of the sepsis score derived from these three biomarkers was 0.80 for severe sepsis, 0.77 for septic shock, and 0.79 for death.

In view of the above, it can be seen that the diagnosis of SBI remains an area of important clinical need. It is an aim of certain embodiments of the invention to provide alternative methods and devices for the diagnosis of SBI.

There is also a need for provision of methods and devices for the diagnosis of SBI that are improved as compared to the methods or devices described in the prior art. It is an aim of certain embodiments of the invention to provide methods and devices that have improved sensitivity of diagnosis as compared to those known from the prior art. It is an aim of certain embodiments of the invention to provide methods and devices that have improved specificity of diagnosis as compared to those known from the prior art. It is an aim of certain embodiments of the invention to provide methods and devices that allow faster diagnosis compared to those known from the prior art. It is an aim of certain embodiments of the invention to provide methods and devices for the diagnosis of SBI that utilise integrated samples. For example, one or more samples may be processed by the same method or device. It is also an aim of certain embodiments of the invention to provide methods or devices that produce quantifiable results.

SUMMARY OF THE INVENTION

In a first aspect, the invention provides a method of diagnosing severe bacterial infection (SBI) in a patient, the method comprising investigating a patient sample to detect the presence of target molecules indicative of the expression of the biomarkers: neutrophil gelatinase-associated lipocalin (NGAL); and resistin; and procalcitonin (PCT);

wherein detection of the presence of target molecules indicative of expression of each of the biomarkers indicates the presence of SBI.

In preferred embodiments the methods of the invention are carried out in vitro, but it will be appreciated that the methods of the invention are also capable of being carried out in vivo.

In a second aspect, the invention provides a device for use in the diagnosis of SBI, the device comprising:

a loading area for receipt of a patient sample;

binding partners specific for target molecules indicative of the expression of the biomarkers: NGAL; and resistin; and PCT.

BRIEF DESCRIPTION OF THE DRAWINGS

Embodiments of the invention are further described hereinafter with reference to the accompanying drawings, in which:

Figure 1 shows graphs illustrating the progress of children through the Emergency

Department (ED) using a) the current diagnostic model, and b) the proposed model using an enhanced point of care (POC) diagnostic (PROMPT- SEPSIS). Alder Hey data from the Hospital Episodes Statistics (HES) database 2009-2010, reporting duration of time in the ED. This is divided into estimated proportions of children either awaiting assessment, or requiring ongoing evaluation. Serious bacterial infection ('diagnosed SBI') accounts for approximately 8% of all medical presentations (Craig et al, BMJ 2010). Alder Hey data reveal that 30% of medical presentations receive antibiotics. ^* abx = antibiotics a. Proportion of children at various stages of the clinical assessment. Serious bacterial infection is diagnosed in stages: triage, clinical assessment, blood tests, review with results of blood tests, diagnosis (see figure 1 ). Median time to diagnosis and intervention is 90 minutes. The sequential approach to investigation results in 193 500 'patient-hours' spent in the department beyond this median duration (calculated as the shaded area under the curve)

b. Our proposed POC diagnostic algorithm is modelled. SBI is identified early owing to an improved sensitivity (90% illustrated) 'ruling-in' the diagnosis and optimising early treatment. The ability of the algorithm to 'rule-out' SBI results in a decrease in the number of unnecessary antibiotic prescriptions. A similar effect upon hospital admission is anticipated. Overall time spent in the department as measured in 'patient-hours' is substantially reduced (shaded area). In this illustration, 'patient-hours' beyond a median duration of 90 minutes are reduced by 37.6%, equating to 72 800 patient-hours.

Figure 2 is a diagram of a LFD for a combination of biomarkers.

Figure 3 is a diagram of a multiplex LFD and a diagram of a traffic light system for a multiplex LFD.

Figure 4 shows successful demonstration of target biomarkers in simplex LFDs. Increasing intensity of 'test' line visible with known (standard) concentrations of each biomarker in buffer solutions. Also demonstrated was a good agreement between measured concentrations of biomarkers on archived human plasma samples, and reflectance readings on FD reader (not shown).

Figure 4a shows PCT detection at 24.7 ng/ml but not lower. C=control. Figure 4b shows NGAL detection as low as 3.1 ng/ml which is significantly below cut-off for detection of sepsis. NEG=control.

Figure 4c shows Resistin detection at 4 ng/ml which is significantly below cut-off for detection for sepsis.

Figure 5 is a Flow diagram showing the number of patients undergoing index tests and the number of patients with CBI, according to STARD guidelines.

Figure 6 is a Bar chart showing relative NGAL, granulysin and resistin expression in survivors and non-survivors.

NS=not significant Figure 7 isa Bar chart showing NGAL, granulysin and resistin concentrations in a) controls, NBI and SBI and b) in survivors and non-survivors.

NS=not significant, ^*** =p<0.0005, ^* =p<0.05

Figure 8 is a graph showing Receiver operator characteristic curves (ROC) of procalcitonin (PCT), granulysin, resistin and NGAL as markers of SBI .

Figure 9 is a graph showing Receiver operator characteristic curves (ROC) of combination of biomarkers (procalcitonin (PCT), resistin and NGAL) as markers of SBI.

Figure 10 is a flow diagram showing Pipeline of development of diagnostic panel of three biomarkers of SBI.

DETAILED DESCRIPTION OF THE INVENTION

The biomarkers utilised in the methods and devices of the invention have all previously been identified as associated with sever bacterial infection. However, the specific panels of biomarkers (e.g. a panel consisting of NGAL; resistin; and PCT) have not been disclosed before. Furthermore, the improvement in diagnosis of SBI provided by the biomarker panels utilised in the methods and devices of the invention has not previously been disclosed, and nor could it have been predicted. In fact, diagnosis utilising the methods and devices of the invention is able to discriminate SBI better than a three-marker panel previously described in the prior art as optimal.

There are a number of advantages provided by the diagnostic methods and devices of the invention. These methods and devices:

• Allow more accurate diagnosis of SBI than can be achieved using methods or devices described in the prior art. Uncertainty in diagnosis can lead to overuse of antibiotics, which contributes to the development of antibiotic resistant bacterial strains.

• Allow earlier diagnosis of SBI. It is well recognised that delays in diagnosis markedly increase the risk of SBI causing patient death.

• Prove simpler to use than panels comprising larger numbers of biomarkers.

• Prove cheaper to use than panels comprising larger numbers of biomarkers. In suitable embodiments of the methods or devices of the invention the group of biomarkers consists of NGAL; resistin; and PCT.

The inventors have found that methods utilising investigation of only these biomarkers are particularly advantageous. Limiting the number of biomarkers in this manner helps to reduce complexity of the methods or diagnostic devices of the invention, and this serves to reduce their cost, and the time taken to determine whether the presence of severe bacterial infection is indicated. The selection of the three biomarkers NGAL, resistin, and PCT provides advantages that have not previously been made available, in that the inventors have found that the combination of these biomarkers has an unusually high diagnostic quality, as indicated by the AUC score of 90% described in the Experimental Results section below. This figure is very high, illustrating the specific benefits of methods or devices of the invention investigating only NGAL, resistin, and PCT. This unusually and unexpectedly high value could not have been predicted from the prior art.

In alternative embodiments, the methods of the invention involve investigating the patient sample for a target molecule indicative of the presence of a further biomarker. Further biomarker selected from the group comprising matrix metalloproteinase 8 (MMP-8) and lactotransferrin.

In certain embodiments the methods or devices of the invention may further involve investigating physiological measurements selected from heart rate, temperature, respiratory rate and blood pressure.

The methods or devices of the invention may make use of target molecules selected from the group consisting of: the biomarker protein; and nucleic acid encoding the biomarker protein. In a suitable embodiment the nucleic acid is mRNA.

In certain embodiments of the methods and devices of the invention, the presence of the target molecules may be investigated using binding partners specific for the target molecules.

In certain cases of such embodiments the binding partners may be selected from the group comprising: complementary nucleic acids; aptamers; antibodies or antibody fragments. In the context of the present invention, a binding partner specific to a target molecule indicative of expression of a biomarkers should be taken as requiring that the binding partners should be capable of binding to at least one such target molecule in a manner that can be distinguished from non-specific binding to molecules that are not target molecules. A suitable distinction may, for example, be based on distinguishable differences in the magnitude of such binding.

An individual binding partner may be able to bind to a target molecule, or molecules, indicative of expression of a single biomarker. Alternatively an individual binding partner may be able to bind to target molecules indicative of expression of two, three or more biomarkers.

In suitable embodiments of the methods or devices of the invention, the target molecule is a nucleic acid, and the binding partner is selected from the group consisting of: complementary nucleic acids; and aptamers.

In other suitable embodiments of the methods or devices of the invention, the target molecule is a biomarker protein, and the binding partner is selected from the group consisting of: aptamers; and antibodies or antibody fragments.

In certain embodiments of the methods or devices of the invention, the presence of the target molecule may be detected by direct assessment of binding of the target molecules and binding partners. Suitable examples of such methods in accordance with this embodiment of the invention may utilise techniques such as electro-impedance spectroscopy (EIS) to directly assess binding of binding partners (such as antibodies) to target molecules (such as the biomarker proteins).

In other embodiments of the methods or devices of the invention, the presence of the target molecules is detected using reporter moieties. Suitable examples of reporter moieties that may be used in such embodiments of such methods or devices include: fluorophores; chromogenic substrates; and chromogenic enzymes.

In certain embodiments of the methods or devices of the invention, the presence of the target molecules is detected using biosensors selected from impedance spectroscopy, surface plasmon resonance, quantum dots, and nanoparticle technology. In the certain embodiments where it is desired that methods or devices of the invention will make use of reporter moieties the reporter moieties may be directly attached to the binding partners. Examples of such embodiments include those utilising labelled antibodies.

In other embodiments of the methods or devices of the invention the reporter moieties may be attached to reporter molecules that interact with the binding partners. Examples of such embodiments include those utilising antibodies indirectly attached to a reporter moiety by means of biotin/avidin complex.

The methods or devices of the invention may make use of a range of patient samples. In suitable embodiments the patient sample is selected from the group comprising a tissue sample, such as a biopsy sample; and a body fluid sample.

In embodiments wherein the patient sample is a body fluid sample, suitable examples of such a sample may be selected from the group comprising: a blood sample, such as a whole blood sample; a plasma sample; a serum sample; a cerebrospinal fluid sample; a urine sample; a saliva sample, such as a buccal swab; a lung fluid sample; and a stool sample.

Suitable methods by which a patient sample may be investigated for the presence of a target molecule may be selected with reference to the nature of the patient sample; and/or the target molecule; and/or the binding partner.

For example, in certain embodiments of the methods or devices of the invention the binding partner is an antibody, or antibody fragment, and the detection of the target molecules utilises an immunological method. In certain embodiments of the methods or devices, the immunological method may be an enzyme-linked immunosorbent assay (ELISA). In other embodiments an immunological method may utilise a lateral flow device (as described elsewhere in the specification).

A method of the invention may further comprise quantification of the amount of the target molecules indicative of expression of the biomarkers that is present in the patient sample. Suitable methods of the invention, in which the amount of the target molecule present has been quantified, and the volume of the patient sample is known, may further comprise determination of the concentration of the target molecules present in the patient sample.

Once the amounts or concentrations of the target molecules in the patient sample have been determined, this information may be used as the basis of a qualitative assessment of the patient's SBI, which may, in turn, be used to suggest a suitable course of treatment for the patient.

For example, the methods or devices of the invention may determine: whether NGAL is present at a concentration of 100ng/ml or greater; and/or whether resistin is present at a concentration of 80ng/ml or greater; and/or whether PCT is present at a concentration of 2ng/ml or greater. If any of these protein biomarkers are determined to be present at, or above, the stated concentrations then this may indicate that the patient should receive treatment with antibiotics.

Additionally, or alternatively, the methods or devices of the invention may determine: whether NGAL is present at a concentration of between 100ng/ml and 500ng/ml; and/or whether resistin is present at a concentration of between 80ng/ml and 120ng/ml; and/or whether PCT is present at a concentration of between 2ng/ml and 10ng/ml. If any of these protein biomarkers are determined to be present at concentrations within the stated ranges, then this may indicate that the patient should be clinically re-evaluated, and may benefit from intravenous treatment with antibiotics.

Additionally, or alternatively, the methods or devices of the invention may further determine: whether NGAL is present at a concentration of greater than 500ng/ml; and/or whether resistin is present at a concentration of greater than 120ng/ml; and/or whether PCT is present at a concentration of greater than 10ng/ml. If any of these protein biomarkers are determined to be present at concentrations greater than the stated values, then this may indicate that the patient would benefit from immediate intravenous treatment with antibiotics, and may require review by a senior clinician.

The methods or devices of the invention may produce a binary output result which indicates the presence of absence of NGAL, the presence of absence of resistin and/or the presence of absence of PCT. In certain embodiments, the methods or devices of the invention may indicate the presence or absence of all three of NGAL, resistin and PCT.

In certain embodiments the device may have a reader. In alternative embodiments the device may be readerless. In a readerless device, the presence or absence of one or more of NGAL, resistin and PCT may be represented by a test line. A positive result may be represented by a positive test line. A negative result may be represented by a negative test line. In a device having a reader, the presence or absence of one or more of NGAL, resistin and PCT may be visible on a reader, such as a digital reader. For example, the concentration of one or more of NGAL, resistin and PCT may be quantified on a reader, such as a digital reader. In one embodiment, the digital reader may be visible using reflectance technology or fluorescence technology.

In certain embodiments the results provided by a method of the invention could be incorporated into an algorithm, for example, a mathematic equation on a computing device, such as a mobile computing device. The result of the algorithm would indicate identification of appropriate treatment regimes for patients.

A device of the invention may further comprise reporter moieties capable of indicating binding of one or more of the binding partners to the target molecules. Suitable reporter moieties may include those referred to above in connection with the methods of the invention. As set out above the reporter moieties may be directly attached to the binding partners, or may be attached to reporter moieties that interact with the binding partners.

Alternatively, a device according to the invention may further comprise means for directly assess binding of binding partners to target molecules. For example, a device of the invention may comprise means for conducting EIS. As referred to above, EIS is suitable for assessing the binding of antibody binding partners to biomarker protein target molecules.

In a suitable embodiment, a device according to the invention may be a biosensor diagnostic device.

In certain embodiments a device of the invention may comprise a point of care (POC) device. Such devices are preferably able to be used at the site of a patient without the need for further analysis at a centralised laboratory, or the like.

The devices of the invention can be manufactured inexpensively. This allows advantageous embodiments of the invention in which device of the invention comprises a disposable device. The disposability of such devices is advantageous in that it reduces the risk of cross-contamination or infection between patients, and obviates the need for costly sterilisation of such devices.

In suitable embodiments a device of the invention may comprise a lateral flow device (LFD). In certain embodiments a device of the invention may comprise a simplex LFD.

Alternatively, in other embodiments a device of the invention comprises a multiplex LFD.

A device according to the invention will generally comprise a reaction area in which binding partners provided in the device are able to bind specifically to target molecules present in the patient sample.

A device in accordance with the present invention will generally comprise a detection area in which binding of the binding partners to target molecules is detected. Such detection may be carried out by means of a reporter moiety.

In suitable embodiments a device according to the invention comprises means for moving a solution comprising target molecules from the loading area to the reaction area. Additionally, or alternatively, a device according to the invention may comprise means for moving a solution comprising target molecules from the reaction area to the detection area.

In suitable embodiments the means for moving the solution comprises a pump. In other suitable embodiments, the means for moving the solution comprises capillary flow means.

The diagnostic information provided by methods or devices of the invention may be of benefit in identification of appropriate treatment regimes for patients. Those patients indicated to have SBI will generally benefit from treatment with antibiotics. Those patients indicated not to have SBI may avoid unnecessary antibiotic treatment. This allows both groups of patients to receive more appropriate treatment, and also has the advantage of avoiding incurring costs that would otherwise be associated with ineffective treatment regimes, while reducing the risk of further promoting antibiotic resistance.

These benefits are illustrated in Figure 1. This compares progress of children through the Emergency Department using the current diagnostic model (panel "a"), with predicted progress based on diagnosis using a POC device in accordance with the present invention (panel "b"). It can be seen that the speed of diagnosis of SBI is increased, unnecessary antibiotic administration is reduced, and the average time spent by a patient in the Emergency Department is considerably curtailed.

As suggested above, POC devices are suitable for use in hospitals - where they avoid the need for centralised processing of samples in hospital laboratories. POC devices are also suitable for use in GP surgeries, walk-in health centres, pharmacies, and counselling centres. It will be appreciated that the ability to do away with centralised sample processing provides advantages in terms of the speed with which diagnoses can be reached, and this is helpful in reducing mortality rates.

Simplex LFDs are those able to detect expression of target molecules indicative of a single biomarker. In a certain embodiment a device according to the invention may comprise a number of simplex LFDs corresponding to the number of biomarkers expression of which is to be investigated. Thus a device in which only the expression of NGAL, resistin and PCT is to be investigated may comprise three simplex LFDs, each investigating expression of a different biomarker. An example of such a device comprising three parallel simplex LFDs is shown in Figure 2.

Alternatively, a device of the present invention may comprise one or more multiplex LFDs. Multiplex LFDs are those able to detect target molecules indicative of expression of more than one biomarker.

Figure 3 illustrates two alternative forms of such a device. The left hand panel of Figure 3 illustrates a multiplex device able to detect the presence of the three biomarkers NGAL, resistin and PCT. The right hand panel illustrates a "traffic light" system utilising a similar device. Here example 1 illustrates a positive control in which all three biomarkers are shown to be expressed. Example 2 illustrates results achieve in a patient sample exhibiting the same expression pattern. Here the expression of all three markers is diagnostic of SBI, and the "red" shown at the top of the LFD indicates the need for action to be taken in respect of treatment. Examples 3, 4, and 5 each illustrate LFDs in which only two of the panel of biomarkers are shown to be expressed. Here "amber" at the top of the LFD indicates that, though not currently suffering from SBI, the patient may require admission to hospital and future treatment with antibiotics. Finally, example 6 illustrates a result in which none of the biomarkers are being expressed. Here the "green" at the top of the LFD indicates that the patient should be free to return home.

It will be appreciated that the same considerations regarding the use of simplex or multiplex LFDs may also be applicable in respect of the methods of the invention.

Figure 4 illustrates development work undertaken in the development of a device of the invention utilising simplex LFDs for biomarker detection. Simplex LFDs for each of the biomarkers NGAL, resistin, and PCT (shown respectively in panels "b", "c", and "a") were provided with test solutions comprising the biomarker proteins at known concentrations in buffer. The concentrations used are marked on the various LFDs.

As can be seen, NGAL was detected at concentrations as low as 3.1 ng/ml - which is significantly below the concentrations that would generally have to be detected in order to allow SBI to be diagnosed. Similarly, resistin was detected at concentrations as low as 4ng/ml, also much lower than necessary in order to allow SBI diagnosis to take place. Finally PCT was detected at concentrations down to 24.7ng/ml, with further refinement of these particular test devices required in order to allow the desired diagnostic concentrations to be detected.

The methods and devices of the invention may make use of microfluidic techniques to provide solutions comprising patient samples to multiple simplex LFDs. As will be appreciated, simplex LFDs are generally easier and cheaper to produce than multiplex LFDs, and so this constitutes an advantage provided by such embodiments of the invention.

Different types of samples may be processed simultaneously, sequentially or separately by a method or device of the invention. For example the sample may be selected from a blood sample, such as a whole blood sample; a plasma sample; a serum sample; a cerebrospinal fluid sample; a urine sample; a saliva sample, such as a buccal swab; a lung fluid sample; and/or a stool sample. In certain embodiments only one type of sample may be used at once. In alternative embodiments, two or more samples may be integrated into the same method or device of the invention. Preferably a blood sample, such as a sample of blood derived from a finger prick, may be used in the method or device. In certain embodiments a saliva sample, such as a Buccal swab of saliva, may be used in the method or device of the invention.

In certain embodiments the device may be primed to run a protocol, such as a fixed protocol, to perform a diagnosis. The device may comprise a microfluidic section such as a cartridge. The cartridge may comprise one or more channels that are configured to contain reagents and/or samples for use in a method of the invention. The channel volumes may be configured to contain a required amount of sample and/or reagent. The device may also comprise a pumping mechanism to regulate the flow rate for preparing the sample and/or reagents. The flow-rate may be regulated to optimise the preparation of the sample and/or reagents. Additionally or alternatively the device may comprise a pumping mechanism to regulate the flow rate for combining the sample and reagents. The flow-rate may be regulated to optimise the combination of the sample and/or reagents.

A device of the invention may be implemented to provide a method of separation of the sample. The method of separation may be selected from electrophoresis, chromatography or eluted conductivity detection. Embodiments which utilise antibody-antigen binding events may be processed in a lateral flow device, such as a hybrid system.

A device of the invention, such as an LFD, may be used for sample separation and sample detection. Alternatively or additionally a device of the invention, such as a microfluidic device may be used to prepare a sample. In certain embodiments a microfluidic device may be used for sample separation and sample detection.

In an embodiment of the invention, a sample may be prepared and simultaneously presented to one or more simplex LFD devices.

More than one sample may be used in a device or method of the invention, in particular where a microfluidic device is utilised. This constitutes an advantage provided by such embodiments of the invention.

The invention will now be further described with reference to the following Experimental Results that illustrate the effectiveness of a number of different methods and devices of the invention, using a number of different forms of target molecule and binding partner.

Experimental Results

The following studies illustrate the utility of the methods and devices of the present invention in the diagnosis of SBI. The results of these investigations also illustrate the benefits that can be achieved by such methods and devices, particularly in comparison with those of the prior art.

The studies described utilise the biomarkers NGAL, resistin, and PCT, as discussed in more detail above, as well as granulysin. Granulysin is an antimicrobial protein co-localised with perforin in the granules of cytotoxic T lymphocytes (CTLs) and natural killer (NK) cells. It has a direct cytotoxic effect on numerous extracellular pathogens, and is postulated to kill intracellular organisms such as Mycobacterium tuberculosis in conjunction with perforin. Granulysin levels correlate inversely with disease activity in M. tuberculosis and normalise with treatment. Granulysin is transiently elevated in acute viral infections, and has been suggested as a marker of host cellular immunity.

In these studies granulysin served as a comparator protein, which is differentially expressed in patients with SBI, but does not offer the diagnostic advantages conferred by the biomarkers NGAL, resistin and PCT. Thus the results achieved serve to highlight the surprising advantages conferred by this panel of biomarkers identified by the inventors.

Methods

We performed microarray analysis on twelve Malawian children with SBI, and three controls. We confirmed that the biomarkers NGAL and granulysin were differentially regulated in these patients, as was the comparator protein and resistin. In an independent cohort, gene expression was validated using real-time PCR, and protein products related to the NGAL, resistin and granulysin transcripts measured using commercial immunoassays.

Performance of the biomarkers NGAL, resistin and PCT, to predict SBI, was also assessed.

Results

Relative gene expression of NGAL and resistin were significantly increased, and granulysin was significantly decreased in patients with SBI (also referred to as "cases") compared to controls. Plasma concentrations of NGAL and resistin were significantly increased in children with confirmed SBI compared to children with signs of sepsis but no detectable bacterial infection, compared to controls (287 versus 128 versus 62 ng/ml and 195 versus 90 versus 18 ng/ml, respectively, p<0.0005). The areas under the curve (AUC) for NGAL, resistin and procalcitonin in predicting SBI were 0.79, 0.80 and 0.86 respectively, and the combined AUC (procalcitonin, resistin and NGAL) was 0.90.

Conclusions

The results demonstrated the utility of the diagnostic biomarkers using transcriptomics (PCR assays using nucleic acid target molecules), and translational concordance with the corresponding protein (ELISA assays using protein target molecules). The panel of NGAL, resistin and PCT significantly improves diagnosis of SBI. Our novel combination of diagnostic biomarkers may facilitate early and accurate diagnosis of SBI.

The initial part of this study aimed to confirm differential regulation of the biomarker genes and granulysin in children with serious bacterial infection using microarray technology. The changes of expression of the biomarkers and granulysin were validated by real-time PCR in an independent prospective cohort. In order to confirm the diagnostic utility of the biomarkers, we measured the proteins of the differentially regulated biomarker genes in plasma samples of children in the independent cohort with serious bacterial infection, to determine if they could reliably predict SBI, or outcome from SBI.

Results:

Of the 377 children included in the study, 282 (74.8%) presented with meningitis and 95 (25.2%) presented with pneumonia. The baseline characteristics are summarised in Table 1 . There were no significant differences in clinical characteristics between children with confirmed bacterial infection (CBI) and children with signs of severe sepsis but no detectable bacterial infection (NBI), but as expected, the proportion of children presenting with signs of meningitis was significantly higher in the SBI group. The numbers of patients that had measurement of NGAL, resistin, granulysin and PCT are shown in Figure 5 according to STARD guidelines.

Microarray bioinformatic analysis

In the biomarker discovery set using microarray, we studied 12 children with serious bacterial infection, and 3 controls. Among the several genes demonstrating large foldchanges (<-3 or >3), our study concentrated on three genes which had been previously reported to be biomarkers of bacterial infection: resistin (RETN), neutrophil gelatinase associated lipocalin (NGAL, also referred to as LCN2) and granulysin (GNLY). The fold changes in the controls compared to the cases were resistin: -5.88, p=0.06, neutrophil gelatinase associated lipocalin (NGAL) -4.19, p=0.002 and granulysin 3.14, p=0.002.

Relative gene expression

In the independent validation set we studied 163 children with serious bacterial infection, and the 15 children from the microarray discovery set were excluded. Relative gene expression of resistin and NGAL were significantly increased in cases compared to controls (p<0.001 ), and relative gene expression of granulysin was decreased in cases compared to controls. Relative gene expression of NGAL granulysin and resistin were increased in non-survivors compared to survivors (medians: 41 .2 versus 17.7, 0.07 versus 0.1 and 13.7 versus 1 1 .0, p= 0.012, 0.04 and 0.26 respectively (Figure 6).

Plasma concentrations of NGAL, granulysin and resistin in NBI, CBI and controls

Plasma concentrations of NGAL and resistin were significantly increased in children with CBI compared to NBI, compared to controls (287 versus 128 versus 62 ng/ml and 195 versus 90 versus 18 ng/ml, p<0.0005). Plasma concentrations of granulysin were not significantly different between the groups (Figure 7).

Plasma concentrations of NGAL, granulysin and resistin in survivors and non- survivors

Plasma concentrations of NGAL and resistin were significantly increased in non-survivors compared to survivors (306 versus 21 1 versus ng/ml and 214 versus 150 ng/ml, p=0.05 and 0.02 respectively). Plasma concentrations of granulysin were not significantly different between the groups (Figure 7).

Performance of NGAL, resistin, PCT, and granulysin in predicting SBI

The areas under the curve (AUCs) for NGAL, granulysin and resistin in predicting SBI were NGAL 0.79 (95% CI 0.73-0.85), granulysin 0.56 (95% CI 0.48-0.63) and resistin 0.80 (95% CI 0.72-0.88). PCT, a previously evaluated biomarker in this cohort (5), had an AUC of 0.86 (0.79-0.92). The ROC curves of the four biomarkers and the combined biomarker panel of NGAL, resistin and PCT (the panel used in methods and devices in accordance with the invention) are shown in Figures 8 and 9.

Additive value of the biomarkers using net reclassification index

The comparator protein granulysin was had poor performance as a single biomarker, even though it was differentially expressed between cases and controls. This illustrates that differential expression alone is not sufficient to yield a useful biomarker.

In contrast, addition of the biomarkers resistin and NGAL to PCT (to yield the biomarker panel used in the methods and devices of the invention) significantly improved prediction (diagnosis) of SBI as compared to the currently used biomarker, PCT alone.

The NRI for the new prediction models are shown in Table 4. Including resistin or NGAL or both resistin and NGAL in the current SBI prediction model (i.e. PCT alone) significantly improve the overall performance. Including resistin in the current model decreases the proportion of correct classification among events (SBI) by 4% (95% CI:-0.22, 0.12; p= 0.60), but significantly increases the proportion of correct classification among non-events (not SBI) by 54% (95% Cl:0.26,0.82; p= 0.0002). In other words, including resistin in the current prediction model results in 4% loss (non-significant) in terms of sensitivity, but gains 54% (significant) in terms of specificity. These changes, and the advantages that they provide, could not be predicted from the prior art.

Including NGAL in the current model results in 1 % (95% Cl:-0.15, 0.18; p= 0.87) gain in sensitivity and a significant 56% (95% Cl:0.33,0.79; p <0.0001 ) gain in specificity. Including both resistin and NGAL in the current model results in 14% (95% CI:-0.06,0.33; p= 0.17) gain in sensitivity and a significant 67% (95% Cl:0.38,0.97; p= < 0.0001 ) gain in specificity. As above, these changes, and the advantages that they provide, could not be predicted from the prior art.

Discussion:

The diagnosis of bacterial sepsis is complicated by the variable and non-specific nature of the signs and symptoms. The search for an accurate biomarker of sepsis to influence clinical practice has become the holy grail of sepsis, but to date the search has been disappointing. We have demonstrated concordance between transcript and protein expression for three biomarkers of SBI initially identified from microarray screening, validation using RT-PCR, to the measurement of protein concentrations in plasma samples. To our knowledge, this is the first study to identify a combination of biomarkers for the detection of SBI using a systems biology approach. Our data suggest possible interrelationships between pathways involving resistin, cytotoxic T cell activity and lipocalin-2, and a significant role for these proteins in the host defence against bacterial infection, but the exact mechanisms require further detailed study. Resistin has been shown to attenuate both cytokine production and T call activity in a dendritic cells stimulated with lipoteichoic acid, and adiponectin (another adipokine) is a negative T cell regulator.

Our combination of three biomarkers (NGAL, resistin, and PCT) demonstrates good discrimination of SBI. The addition of resistin and NGAL to PCT demonstrates a net reclassification improvement of 81 % as compared to PCT alone. These findings will be further validated in a prospective cohort of febrile children, including an assessment of the impact of the combination biomarkers on clinical decision making. Transcription profiling has been used in other infectious diseases to identify biosignatures of disease groups. The use of multiple biomarkers allows the capture of multi-dimensional patterns and pathways, such as those that can occur in complex biological processes like sepsis, and is therefore more robust. Gene expression studies are typically estimates of effect size, and further validation in larger studies is imperative, if accurate differences are to be identified. Our study has fulfilled these criteria and is now undergoing further validation in a new prospective cohort of febrile children presenting to a tertiary centre emergency department in England.

In this study, resistin and NGAL singly provided good discrimination of SBI in archived samples (AUC: 0.80 and 0.79 respectively), but individually did not perform as well as procalcitonin in our previous study. Our combination biomarkers performed better (AUC: 0.90) than an optimal 3-marker panel (NGAL, protein C, and interleukin-1 receptor antagonist) described in the prior art connection with in North American adults presenting to an emergency department with suspected sepsis (Shapiro, N.I., Trzeciak, S., Hollander, J.E., Birkhahn, R., Otero, R., Osborn, T.M., Moretti, E., Nguyen, H.B., Gunnerson, K.J., Milzman, D., et al. 2009. A prospective, multicenter derivation of a biomarker panel to assess risk of organ dysfunction, shock, and death in emergency department patients with suspected sepsis. Crit Care Med 37:96-104).

Our study exemplifies translational medicine, the use of transcriptomics and bioinformatics to develop a better functional understanding, and to identify potent biosignatures for accurate clinical decision-making (Figure 10). Obstacles to the pipeline for diagnostics have been reported to be at the "front end" i.e. earlier stages of discovery and development, and also later, with lack of samples for validating and testing biomarkers. Our study has overcome both these barriers, and is a major step in addressing the pipeline problem in the diagnosis of severe sepsis.

The present invention provides methods in which expression of the biomarkers NGAL, resistin, and PCT is used to diagnose SBI. It also provides devices able to detect expression of these biomarkers. These are suitable for use as a point-of-care (POC) device at low cost, and the POC devices of the invention may be used in primary and secondary care, and in resource-poor settings, with great potential to reduce delays in diagnosis, unnecessary hospital admission, unnecessary antibiotic prescribing, and morbidity and mortality. In conclusion, we have identified methods and devices using a unique combination of biomarkers of SBI, which could help guide antibiotic management and decisions on referral to hospital. Our data demonstrate exciting improvements in classification, which need to be validated prospectively in a larger cohort. We are currently conducting a prospective impact study to further confirm clinical validity and clinical utility, and to assess the influence of our novel combination biomarkers on clinical decision making, in particular the decision to prescribe antibiotics or admit to hospital, in febrile children presenting to the emergency department.

Materials and methods:

Study setting and population:

We prospectively enrolled children between April 2004 and October 2006 who presented to the Accident and Emergency Department and the Admissions Unit of Queen Elizabeth Central Hospital, Blantyre, Southern Malawi. The study setting, population and characteristics of study patients have been described previously (5).

The primary outcome measure was bacteriological confirmation of infection (SBI) and the secondary outcome measure was death/survival in hospital. Afebrile children, without malaria parasitaemia, from the same villages as the cases, were used as controls. All controls were HIV-uninfected. Ethical approval for this study was granted from The College of Medicine Research Committee (COMREC), Malawi and The Liverpool School of Tropical Medicine Research Ethics Committee. Parents or guardians gave written informed consent for children to enter the study.

Definitions

Cases: Children who presented with signs and symptoms of meningitis or pneumonia, as defined previously.

Confirmed serious bacterial infection (CBI): Children who presented with either bacterial meningitis or bacterial pneumonia, in whom a bacterial pathogen was identified by culture, polysaccharide antigen test or PCR (Streptococcus pneumoniae, Neisseria meningitidis, and Haemophilus influenzae b).

No detectable bacterial infection (NBI): Children who presented with bacterial meningitis or bacterial pneumonia, but who were negative for any bacteria on culture, polysaccharide antigen test or PCR (S. pneumoniae, N. meningitidis, and H. influenzae b). Microbiological methods and malaria diagnosis

Blood, CSF and lung aspirate culture and PCR were performed using standard microbiological techniques as previously described. Blood films were examined by microscopy according to best standard practice of district-hospital laboratories in Africa.

RNA Extraction and quantification

Total RNA was extracted from whole blood using an optimised method for the PAXgene™ blood RNA kit (Qiagen, West Sussex, UK), as previously described (36). The RNA then concentrated by precipitating overnight at -20°C with 2μΙ linear acrylamide (5mg/ml), (Ambion, Warrington, UK), 0.5 volumes 7.5M Ammonium acetate (Sigma, Dorset, UK) and 2.5 volumes ice cold 100% ethanol. The samples were centrifuged at 13,000g, 4°C for 30 minutes. The RNA pellet was washed twice with 0.5ml ice cold 80% ethanol, each time centrifuging at 4°C for 10 minutes at 13,000g. The pellet was air dried for approximately 5 minutes and re-suspended in 1 1 μΙ of nuclease free water.

The total RNA concentration (ng/μΙ) and ratios (260/280 and 260/230) were measured using a NanoDrop ND-100 UV-vis spectrophotometer (Labtech International, Ringmer, UK) and the RNA integrity was assessed using the Agilent 2100 bioanalyser (Agilent Technologies, Edinburgh, UK) pre- and post-concentration.

Affymetrix Microarray Analysis

RNA ^g) was reverse transcribed into double stranded cDNA using the Superscript double stranded cDNA synthesis kit [Invitrogen] according to the manufacturers' guidelines. The double stranded cDNA was purified using a GeneChip® sample clean-up module (Invitrogen, Paisley, UK). The purified cDNA was then biotin labelled with the ENZO BioArray high yield RNA transcript labelling kit (Affymetrix, High Wycombe, UK) and cleaned with a cRNA clean-up module (Invitrogen, Paisley, UK). Aliquots of labelled cRNA (20μg) were fragmented at 94°C for 35 minutes and then hybridized to a Human Genome U133A GeneChip® array for 16 hours rotating at 60rpm at 45°C in a GeneChip® Hybridization Oven 640 (Affymetrix, High Wycombe, UK). Each chip was washed and stained on a GeneChip® Fluidics Station 450 (Affymetrix, High Wycombe, UK) and scanned using a GeneChip® Scanner 450 (Affymetrix, High Wycombe, UK) employing standard recommended protocols (Affymetrix, High Wycombe, UK). Microarray bioinformatics analysis

The experiment was made up of four different groups (HIV-infected non-survivor n=3, HIV- infected survivor n=3, H I V-un infected non-survivor n=3, HIV-uninfected survivor n=3), and the control group n=3. The experimental design was the comparison of the expression levels of probe sets between each group and between each group and the control. Microarray analysis was carried out using Bioconductor packages Simpleaffy (37) and Limma (38) with using the Mas5 algorithm. Mas5 generates a Present, Marginal or Absent call for each probeset, and these calls were used to remove any probe sets that entirely absent across all arrays. This way only real biological change is analysed, and probe sets in the low signal intensity, noisy area are not included. Once this filtering step was complete, a differential analysis using the Limma Bioconductor package was carried out. This analysis was set up as a two-group analysis using one-colour arrays, applying the Limma fit model and using an empirical Bayes method to rank probe sets in order of evidence for differential expression. In both the FT and FFPE groups only probe sets with a foldchange of less than -1 and greater than 1 averaged across the subtype groups were accepted. These probesets were then merged with annotation to give gene level information.

Reverse transcription for qPCR

RNA samples were DNAse treated (DNA Free, Ambion, Warrington, UK] to remove any contaminating genomic DNA. RNA (^g) was reverse transcribed with Superscript II™ RNase H ^" Reverse Transcriptase and oligo (dT) _12- i8 (50C^g/ml) following the manufacturer's guidelines ((Invitrogen, Paisley,UK). The cDNA was stored at -40°C until required.

Real-Time quantitative PCR measurement of target genes

The Human Universal Probe Library system (Roche, Burgess Hill ,UK) employing proprietary locked nucleic acids (LNA) analogues was used for qPCR to measure expression levels in genes of interest. Using the Roche Online Assay Design Centre, specific primers and an associated probe were selected for the reference and target transcripts. Gene expression was determined using real time qPCR on a Roche LightCycler 480. Assays were designed using the Roche Universal ProbeLibrary system. Table 1 shows the primer sequences and the probe number for each assay. Assays were run in triplicate 10μΙ reactions in 384 well plates. Each well contained 5μΙ LightCycler 480 Probes Master (2x concentration), 0.1 μΙ each primer (20μΜ), 0.1 μΙ probe (10μΜ), 0.7μΙ water and 4μΙ of cDNA (1 :40 dilution). The thermocycling conditions were: 95°C for 5 minutes, then 50 cycles of 95°C for 10 seconds and 60°C for 30 seconds, a final cooling step of 40°C for 10 seconds completes the program. To compensate for variations in cell number, RNA isolation, reverse transcription and PCR amplification efficiency, 12 endogenous transcripts were screened and analysed using the GeNORM algorithm (39), HPRT1 , GAPDH and YWHAZ were selected as suitable reference genes. A total of 188 randomly selected samples were analysed using RT PCR.

NGAL, Resistin and Granulysin measurement by ELISA

NGAL (R&D Systems, Abingdon, UK), resistin (Universal Biologicals, Cambridge, UK) and granulysin (ABO Swiss, Xiamen, China) were measured in archived plasma samples using commercially available sandwich enzyme immunoassays according to the manufacturers' instructions. Assays were performed on archived samples if there was enough sample left.

Statistical analysis:

Basic analyses was performed using SPSS for Windows, version 15.0, (Illinois, USA), and ROCR and Hmisc packages in R (http://www.r-project.org/) were used in summarising the performances of biomarkers and computing the net reclassification improvement (NRI). The data, when plotted, did not follow a normal distribution, and therefore the Mann Whitney Test was used to compare distributions, and Spearman's correlation coefficient for correlations of clinical and laboratory variables. Receiver operator characteristic (ROC) curves were used to determine the areas under the curve (AUCs) with 95% confidence intervals for the biomarkers to predict SBI and death. The laboratory assays for bacteriological confirmation and the markers of infection were performed by investigators blinded to the clinical data. The clinicians involved in managing the cases were not involved in performing any of the laboratory assays.

Net reclassification improvement (NRI) was used to measure the additive value of measuring multiple biomarkers and to quantify improvement in performance over procalcitonin alone, which has previously been shown to be the best biomarker of SBI in Malawian children with signs of severe sepsis. The magnitude of NRI is more important than the statistical significance, therefore we reported NRI with confidence intervals, as well as significance values.

The study was reported according to STARD guidelines which includes method of recruitment of patients, orders of test execution, and numbers of patients undergoing the tests under evaluation and the numbers of patients with the reference standard (Figure 5). Table 1. Characteristics of children with confirmed serious bacterial infection (CBI) and those with no bacterial infection (NBI) detected. Numeric values are median and interquartile range (IQR)

Table 2 Primer sequences and probe numbers for each assay

Table 3. Performance characteristics of NGAL and resistin as biomarkers of SBI, singly, and in combination with procalcitonin

Table 4. Reclassification improvement in new prediction models versus current PCT alone model