WO2024264081A1

WO2024264081A1 - Method for diagnosing tuberculosis disease

Info

Publication number: WO2024264081A1
Application number: PCT/ZA2024/050030
Authority: WO
Inventors: Novel Njweipi Chegou; Gerhard Walzl; Gian VAN DER SPUY
Original assignee: Stellenbosch University
Current assignee: Stellenbosch University
Priority date: 2023-06-22
Filing date: 2024-06-24
Publication date: 2024-12-26
Anticipated expiration: 2025-12-22

Abstract

A method for diagnosing TB or for ruling out TB is described. Devices and kits for use in performing the method are also disclosed. The method comprises testing a biological sample from a subject for CRP and at least one other cytokine biomarker selected from the group consisting of SAA, CXCL10/IP-10, SAP, CCL1/I-309, CXCL9/MIG and ferritin. Typically, the biological sample is contacted with capture agents which can bind to the biomarker(s) of interest, and binding of the capture agents to the biomarker(s) is detected. The sample is typically a blood sample, and more particularly a fingerstick (capillary) sample. A lateral flow assay can be used to perform the method, e.g. using up-converting phosphorous technology.

Description

METHOD FOR DIAGNOSING TUBERCULOSIS DISEASE

CROSS-REFERENCE TO RELATED APPLICATION

This application claims priority from South African provisional patent application number 2023/06459 filed on 22 June 2023, which is incorporated by reference herein.

FIELD

This disclosure relates to a method for diagnosing tuberculosis disease (TB).

BACKGROUND

Pulmonary tuberculosis (PTB) is still a leading cause of death and places severe pressure on health care systems of low- and middle-income countries. Each year, 3.6 million people with TB still go undiagnosed, and approximately 30% of patients diagnosed with TB are not treated. The high burden of undiagnosed and untreated TB further fuels ongoing transmission.

TB testing remains costly, imperfect, and of limited accessibility even in high TB prevalence areas. The highly sensitive and specific GeneXpert® MTB/RIF or Ultra (GeneXpert®) has shown great promise for rapid detection of active TB, but the cost is high. Furthermore, GeneXpert® is a central laboratory-based platform rather than point-of-care (POC), and is subject to logistical challenges and loss to follow-up. More recently, the Cepheid 3 gene host-response cartridge has been developed based on detection of a 3 gene signature (GBP5, DUSP3, KLF2) (Sweeney et al, Genome-wide expression for diagnosis of pulmonary tuberculosis: a multicohort analysis. Lancet Respir Med. 2016; 4: 213-24). This has been translated to a test (GeneXpert-MTB-HR-Prototype) which automatically calculates a TB score. However, this test will also have the same accessibility issues in high burden and resource-constrained settings as the GeneXpert® MTB/RIF and Ultra tests. Liquid culture is even less accessible than GeneXpert®, prone to contamination, and takes 42 days for a negative result. Traditional chest X-rays (CXR) are also unsuitable as rapid diagnostic or screening tools. They are relatively expensive, non-specific and at present, depend on skilled personnel for interpretation.

Moreover, active pulmonary TB will only be confirmed in approximately one-third of those that undergo costly and time-consuming testing for TB based on suggestive respiratory symptoms. This represents an inefficient use of sparse and expensive resources. Focusing TB investigation could streamline resource use, thereby improving TB care in overburdened health care systems. The World Health Organisation (WHO) has re-emphasised the need for efficient TB screening and seeks a non-sputum based screening tool to narrow TB testing to those patients with a high likelihood of PTB. Such a triage test should be appropriate for global use and have comparable validity to the WHO minimum target product profile (TPP) sensitivity of 90% and specificity of 70%.

There is therefore a need for a TB test which addresses the problems described above.

The preceding discussion of the background is intended only to facilitate an understanding of the present disclosure. It should be appreciated that the discussion is not an acknowledgment or admission that any of the material referred to was part of the common general knowledge in the art as at the priority date of the application.

SUMMARY

According to a first aspect of the disclosure, there is provided a method of diagnosing tuberculosis disease (TB) or ruling-out TB in a subject, the method comprising the step of testing a biological sample from a subject for the presence of CRP (C-reactive protein) and at least one other biomarker selected from the group consisting of SAP (Serum amyloid P), SAA (Serum amyloid A), CCL1/I-309 (Chemokine (C-C motif) ligand 1 , also abbreviated as I-309), CXCL9/MIG (C-X-C motif chemokine ligand 9, also known as monokine induced by interferon Gamma), ferritin and CXCL10/IP-10 (C-X-C motif chemokine ligand 10, also known as interferon gamma inducible protein 10).

The method may comprise the step of testing the biological sample for the presence of CRP and at least two other biomarkers selected from the group consisting of SAP, SAA, CCL1/I-309, CXCL9/MIG, ferritin and CXCL10/IP-10.

The method may comprise the step of testing the biological sample for the presence of CRP, SAP and CCL1/I-309.

The method may comprise the step of testing the biological sample for the presence of CRP, SAP and CXCL9/MIG.

The method may comprise the step of testing the biological sample for the presence of CRP, CXCL10/IP-10 and CXCL9/MIG. The method may comprise the steps of testing the biological sample for the presence of CRP, SAA and CXCL10/IP-10, and optionally also ferritin.

The method may include the steps of: contacting the biological sample with capture agents which bind to CRP and at least one other biomarker selected from the group consisting of SAP, SAA, CCL1/I-309, CXCL9/MIG, ferritin and CXCL10/IP-10; and detecting binding of the capture agents to the biomarkers.

The biological sample may be a blood sample, such as whole blood, serum or fingerprick blood.

The tuberculosis disease may be TB meningitis, pleural TB, TB pericarditis, pulmonary TB, TB lymphadenitis, skeletal TB, spinal TB, military TB, genitourinary TB, liver TB, gastrointestinal TB, TB peritonitis or cutaneous TB. Preferably, the TB is pulmonary TB.

One or more indicators may be provided to indicate when binding of each of the capture agents and biomarkers occurs.

Detection of two or more of the biomarkers in the sample or a measured signal which equates to a level of biomarker in the sample which is higher than a threshhold level of the same biomarker may be an indicator of TB.

A lateral flow assay may be used to detect eh biomarkers.

According to a second embodiment of the disclosure, there is provided a device for diagnosing TB, the device comprising: a means for receiving a sample from a subject suspected of having TB; capture agents for binding CRP and at least one other biomarker selected from the group consisting of SAP, SAA, CCL1/I-309, CXCL9/MIG, ferritin and CXCL10/IP-10; and at least one indicator to indicate to a user of the device when the capture agents bind to the biomarkers.

The capture agents may be selected from the group consisting of antibodies, affibodies, ankyrin repeat proteins, armadillo repeat proteins, nucleic acid aptamers, peptides, carbohydrate ligands, synthetic ligands and synthetic polymers. Preferably, the capture agents are antibodies. The indicator may indicate binding of the capture agent to the biomarker by electrical, electronic, acoustic, optical or mechanical methods. The indicator may be up-converting phosphorous particles.

The device may further include measuring means for measuring the levels of the detected biomarkers.

The device may further include amplifying means for increasing the sensitivity of the detection of the biomarkers.

The device may be a hand-held point-of-care device or part of such a device. The device may be a lateral flow assay test strip.

According to a third embodiment of the disclosure, there is provided a cartridge comprising a plurality of lateral flow assay test strips.

According to a further embodiment of the disclosure, there is provided a system comprising at least one device or cartridge as disclosed and a reader for analysing the device(s) or cartridge.

According to a further embodiment of the disclosure, there is provided a kit for diagnosing TB, the kit comprising one or more of the following: capture agents for binding CRP and at least one other biomarker selected from the group consisting of SAP, SAA, CCL1/I-309, CXCL9/MIG and CXCL10/IP-10; means for obtaining or receiving a biological sample from a subject; a device for diagnosing TB; and/or instructions, in electronic or paper form, for performing the method as described above.

According to a further embodiment of the disclosure, there is provided a method of diagnosing a human subject as having TB and treating the subject, the method comprising the steps of: testing a biological sample from a subject suspected of having TB for the presence of CRP and at least one other biomarker selected from the group consisting of SAP, SAA, CCL1/I-309, CXCL9/MIG, ferritin and CXCL10/IP-10; determining whether the subject has TB based on the detection of the biomarkers in the sample; and administering an effective amount of TB treatment to the subject if the subject is in need thereof. According to a further embodiment of the disclosure, there is provided the use of capture agents for binding CRP and at least one other biomarker selected from the group consisting of SAP, SAA, CCL1/I-309, CXCL9/MIG, ferritin and CXCL10/IP-10 in the manufacture of a kit for diagnosing TB.

FIGURES

Figure 1 : Configuration of the up-converting phosphor technology-based lateral flow strip

(adapted from Yang et al.. PLoS One. 2017: 12(2): e0171376).

Figure 2: Configurations of lateral flow test strips that could be used to detect 3 biomarker proteins.

Figure 3: A) Receiver operator characteristics (ROC) curve showing the accuracies of the biomarker combination of CRP, SAP, CCL1/I-309 in the diagnosis of TB and B) scatterplot. AUC: Area under curve.

Figure 4: A) ROC curve showing the accuracies of the biomarker combination of CRP, SAP and CXCL9/MIG in the diagnosis of TB and B) scatterplot.

Figure 5: A) ROC curve showing the accuracies of the biomarker combination of CRP,

CXCL10/IP-10 and CXCL9/MIG in the diagnosis of TB and B) scatterplot.

Figure 6: CRP, CXCL10/IP-10 and CXCL9/MIG - performance in the training sample set

(n=677). A) = NoTB; B) = TB.

Figure 7: CRP, CXCL10/IP-10 and CXCL9/MIG - performance in the test sample set

(n=284). A) = NoTB; B) = TB.

Figure 8: ROC curve showing the accuracies of the biomarker combination of SAA, CRP,

CXCL10/IP-10 in the diagnosis of TB.

Figure 9: A) ROC curve showing the accuracies of the biomarker combination of CRP and

CCL1 9 in the diagnosis of TB and B) scatterplot. Figure 10: A) ROC curve showing the accuracies of the biomarker combination of SAA, CRP, Complement Factor H, ApoA-1 , NCAM, IFN-g and CXCL10/IP-10 in the diagnosis of TB and B) scatterplot.

Figure 1 1 : A) ROC curve showing the accuracies of the biomarker combination of SAA, CRP, Complement Factor H, ApoA-1 , ferritin, IFN-g, CXCL10/IP-10 in the diagnosis of TB and B) scatterplot.

Figure 12: SAA, CRP, CXCL10/IP-10 - MBT analysis evaluated against GeneXpert Ultra. ROC curve for MBT excluding GeneXpert Ultra Trace+ participants.

Figure 13: SAA, CRP, CXCL10/IP-10 - training set results using clinical diagnostic algorithm.

Figure 14: SAA, CRP, CXCL10/IP-10 - MBT analysis evaluated against a final diagnostic algorithm. A) ROC curve when definite/probable TB was compared with no TB. B) ROC curve for HIV- participants only.

Figure 15: SAA, CRP, CXCL10/IP-10 - evaluation of MBT test performance at individual sites. A) All sites, B) Pan-African.

DETAILED DESCRIPTION

A method for diagnosing TB, or alternatively for ruling out TB, is described herein. Devices and kits for use in performing the method are also disclosed.

Abbreviations of biomarkers referred to herein: CRP = C-reactive protein, SAP = Serum amyloid P, SAA =Serum amyloid A, CCL1 (l-309)= Chemokine (C-C motif) ligand 1 , also abbreviated as I-309, CXCL9 (MIG)= C-X-C motif chemokine ligand 9, also known as monokine induced by interferon Gamma (MIG), CXCL10 (IP-10) = C-X-C motif chemokine ligand 10, also known as interferon gamma inducible protein 10 (IP-10).

The method comprises testing a biological sample from a subject for CRP and at least one other cytokine biomarker selected from the group consisting of SAA, CXCL10/IP-10, SAP, CCL1/I-309, CXCL9/MIG and ferritin.

The at least one other biomarker can be SAA. Alternatively, the at least one other biomarker can be CXCL10/IP-10. Alternatively, the at least one other biomarker can be CXCL9/MIG. Alternatively, the at least one other biomarker can be CCL1/I-309. Alternatively, the at least one other biomarker can be SAP.

A third biomarker can be SAA. Alternatively, the third biomarker can be CXCL10/IP-10. Alternatively, the third biomarker can be SAP. Alternatively, the third biomarker can be CXCL9/MIG. Alternatively, the third biomarker can be CCL1/I-309,

In one embodiment, the method comprises testing the biological sample for the presence of CRP, SAA and CXCL10/IP-10. Optionally, the sample can also be tested for ferritin.

In an alternative embodiment, the method comprises testing the biological sample for the presence of CRP, SAP and CCL1/I-309.

In an alternative embodiment, the method comprises testing the biological sample for the presence of CRP, SAP and CXCL9/MIG.

In an alternative embodiment, the method comprises testing the biological sample for the presence of CRP, CXCL10/IP-10 and CXCL9/MIG.

It will, of course, be possible to test the sample for additional biomarkers, i.e. by using a 4- biomarker signature, a 5-biomarker signature, a 6-biomarker signature, a 7-biomarker signature, and so forth.

Typically, (a) the biological sample is contacted with capture agents which can bind to the biomarker(s) of interest, and (b) binding of the capture agents to the biomarker(s) is detected.

The subject may be suspected as having TB or may have been exposed to a patient with a Mycobacterium tuberculosis infection.

A positive diagnosis for TB can be made when binding of the capture agents to one, two or three of the tested biomarkers is detected, or when the levels of the detected biomarkers are higher than a typical level of the same biomarker in subjects without TB. In another embodiment, a positive TB diagnosis can be made when the levels of the detected biomarkers are lower than a typical level of the same biomarker in subjects without TB.

Optionally, cut-off or threshold values can be determined based on levels of biomarkers which are typically found in patients without TB, and the levels of the biomarkers detected in the sample can be compared to the cut-off levels when making the determination of whether or not the subject has TB. In other words, the method can detect whether the biomarkers in the panels are under- or over-expressed relative to a subject who does not have TB.

The method can also be used as a rule-out (or triage test), so as to streamline the TB diagnostic process. In such a method, only subjects who test positive are further tested to confirm a diagnosis of active TB (in the interim, while waiting for the results of the second test, the subject can optionally be started on treatment). Subjects who do not test positive in the method described herein will not be sent for further TB testing, avoiding the high costs associated with these tests and allowing for redirection of resources to those with active TB.

In one embodiment, the method is for diagnosing or ruling-out active TB, rather than for latent M. tuberculosis infection (LTBI) or incipient TB.

In one embodiment, the TB is pulmonary TB. However, the TM may also be extra pulmonary TB, such as pleural TB, TB pericarditis, TB meningitis, TB lymphadenitis, skeletal TB, spinal TB, military TB, genitourinary TB, liver TB, gastrointestinal TB, TB peritonitis or cutaneous TB. The method is independent of HIV co-infection status.

The sample is typically a blood sample, and more particularly a fingerstick (capillary) sample. However, the sample can also be a venous sample (either whole blood or serum). No stimulation of the sample, e.g. with an antigen, is required.

The method is typically a multi-biomarker test (MBT) assay which measures individual concentrations of multiple biomarkers (in this case, at least two biomarkers and preferably at least three biomarkers).

The method described herein can be used for all human subjects, including adults and children (e.g. children 13 years and younger).

TB treatment can be administered to subjects who are diagnosed as having TB.

The biomarkers in the sample can be detected using commercially available techniques, such as ELISA techniques or multiplex bead array technology, although it is intended that a specific point- of-care (or bedside) diagnostic device will be used for rapid diagnosis, particularly in resource poor settings. Such a device will lead to a significant reduction in the costs and delays that are currently incurred in the diagnosis of TB, with a consequent reduction in death or long-term disability. In one embodiment, the device has a means for receiving the sample from the subject, such as a loading or receiving area onto or into which the sample is placed. Capture agents and indicators are present in the device, and once the sample has been loaded onto or received into the device, the sample is brought into contact with the capture agents, which are allowed to bind to the biomarkers if present. The indicator will indicate to the user of the device that binding of capture agents to one or more of the biomarkers has occurred. The device may further include amplifying means for increasing the sensitivity of the detection of the biomarkers.

The capture agents can be antibodies, affibodies, ankyrin repeat proteins, armadillo repeat proteins, nucleic acid aptamers, peptides, carbohydrate ligands, synthetic ligands or synthetic polymers. Typically, however, the capture agents are antibodies. The indicator can be a calorimetric, electrical, electrochemical, electronic, chromogenic, optical, fluorescent or a radio- labeled indicator.

In one embodiment, the device can be a lateral flow assay (LFA) similar to those known in the art.

LFA’s comprise at least four components: a sample pad (often made of glass fibre or woven mesh), a conjugate pad (similarly made of glass fibre or woven mesh), a nitrocellulose (NC) membrane with test and control lines, and a wicking pad (usually made of glass fibre or cellulose fibers).

The sequential steps of an LFA are as follows:

Sample application-. The sample, often dissolved in a running buffer, is applied to the sample pad. From here, it travels through the various components of the test strip. For blood samples, the buffer can include a blood lysing agent.

Conjugate pad interaction-. The sample flows to the conjugate pad, which contains dried labelled bioreceptors (typically primary antibodies). If the sample contains the target antigen, it will bind to these labelled bioreceptors.

Nitrocellulose migration-. The bound or unbound bioreceptors, along with the sample, continue to flow through the nitrocellulose membrane.

Test line formation-. The test line on the nitrocellulose membrane consists of capture antibodies that are permanently bound via electrostatic interactions. When the antigen in the sample is present, it binds to these capture antibodies, along with the labelled bioreceptors, resulting in the visual appearance of a line. If there is no antigen, no binding will occur at the test line, and thus, no line will be visible. The amount of label captured on the test line can be correlated to the amount of antigen present in the sample. Control line confirmation-. Simultaneously, the labelled bioreceptors will bind to secondary antibodies on the control line, also resulting in the visual appearance of a line. This control line serves to confirm that fluid has passed successfully from the sample application pad, past the test line.

In one embodiment, the MBT is an up-converting phosphor lateral flow (UCP-LF) test. This comprises a lateral flow (LF)-based test device which measures individual concentrations of multiple biomarkers, using up-converting particles (UCP) as luminescent labels for quantitation. Unlike other conventional fluorescence techniques, UCPs are excited by infrared light and emit visible light. UCPs are mainly related compounds of rare earth elements that can be classified into vanadates, phosphates, sulfides, sulfur oxides, oxides, oxyhalides, molybdates, etc. The UCP label is not hampered by the red colour produced by haemolysis of erythrocytes, allowing convenient dilution of collected fingerstick blood using a lysis buffer. In comparison to other fluorescent labels, the UCP reporter is not light sensitive, does not fade and has an infinite lifetime. The test format is highly flexible in the number and identity of biomarkers that can be analysed.

A detailed description of UCP technology, including how to synthesise up-converting particles, can be found in Yang, R. (eds) Principles and Applications of Up-converting Phosphor Technology. Springer, Singapore, https://doi.org/10.1007/978-981 -32-9279-6_2, the contents of which are incorporated herein in their entirety. The UCP-LF test is described in more detail in van Hooij et al (Prototype multi-biomarker test for point-of-care leprosy diagnostics. iScience. 2020; 24(1 ): 102006) and Pierneef et al (Host biomarker-based quantitative rapid tests for detection and treatment monitoring of tuberculosis and COVID-19. iScience. 2023 Jan 20;26(1 ):105873), the contents of which are also incorporated herein in their entirety.

MBT results can be assessed on a stand-alone portable reader to yield immediate results.

Figure 1 shows the configuration of one embodiment of an up-converting phosphor technologybased lateral flow strip for use in detecting and quantifying CRP. (a) Infrared light is used to excite the UCP, which emits light in the visible spectrum, (b) The membrane of the UPC-LF strip has a test line (T line) and control line (C line), which are coated with mouse-anti-human CRP monoclonal antibody and goat anti-mouse antibody, respectively. The UCP particles are conjugated covalently with other mouse-anti-CRP antibody as a reporter, (c) For sample containing CRP, UCP reporters will be captured on both the T line and C line, while they will only be captured on the C line for sample without CRP. The signal from the UCP reporters indicates the concentration of CRP. The same principles apply to the detection of the other biomarkers in the signature, e.g. SAA and IP-10. Various test strip configurations can be used to detect the multiple biomarkers. Figure 2 shows four examples:

(a) singleplex strip - a separate test strip is used for each biomarker in the biosignature;

(b) extended MBT strip - a single test strip is used for all the biomarkers in the biosignature, the test strip having a test line and control line for each biomarker;

(c) basic MBT strip - a single test strip used for all the biomarkers in the biosignature, the test strip having a test line for each biomarker but only a single control line for each biomarker; and

(d) parallel MBT strip - separate test strips are used for each biomarker, each strip having standard dimensions. These can be contained together, e.g. on a single support or in the form a cartridge, so that they can be read together (in parallel) rather than separately (or sequentially/serially). The use of parallel strips, each specific for a single biomarker, minimizes potential cross reactivity. A portable reader capable of 2D scan (e.g. ESE Quant LR3 version) allows for application of a parallel MBT format.

In one embodiment, the assay can make use of biosensors comprising a transducer element, for the conversion of the biological signal to an electronic signal, to which antibodies against the biomarkers can be immobilised. The transducer element can use different conversion mechanisms, such as piezoelectricity or impedance changes, and can be implemented on different substrates, such as electrospun nanofiber meshes or paper. Depending on the chosen transducer element, the binding of the target molecules in the samples to the immobilised capture antibodies results in the generation of piezoelectric energy or a change in impedance, proportional to the amount of target molecule detected in the sample. The measured data are stored in the handheld device containing the biosensing elements, but can also be downloaded to a database or cloud for further analysis.

A kit can also be provided to enable the method of the invention to be performed. The kit could include one or more of the following: capture agents, such as antibodies, for binding the intended biomarkers; a means for obtaining or receiving the sample from a subject; fingerstick buffer; one or more devices as described above (e.g. lateral flow test strips or cartridges); a reader for analysing the device; and/or instructions, in electronic or paper form, for performing the method.

A system comprising a device (e.g. lateral flow test strips or cartridges) and a reader is also provided. The invention further provides the use of capture agents for CRP and at least one other biomarker selected from the group consisting of SAP, SAA, CXCL9/MIG and CXCL10/IP-10 in the manufacture of a kit or device for diagnosing TB.

In an initial part of this study, concentrations of 20 different host protein biomarkers were evaluated in serum samples from individuals that were presumed to have active TB across 7 field sites, situated in 6 different African countries (South Africa, Ethiopia, Uganda, Malawi, Namibia and The Gambia). The biomarkers that were evaluated were: GDF-15, SAA, CRP, fibrinogen, SAP, complement factor H, ApoA-1 , ApoCIII, NCAM, BDNF, Serpin C1 , ferritin, CCL14(HCC-1 ), CCL1 (I-309), CXCL9(MIG), interferon gamma (IFN-g), TNF-alpha, CXCL10(IP-10), TGF-alpha and VEGF-A (Table 1 ). Measurements were done using the Luminex® biomarker discovery and validation platform. When analysis was restricted to not more than 3 biomarkers in a biosignature to allow for possible easier conversion of the biosignatures into point of care tests, four different 3-marker combinations which showed potential in the diagnosis of active TB were identified, namely:

- CRP+SAP+ CCL1 (I-309);

- CRP+SAP+CXCL9(MIG);

- CRP+ CXCL10(l P- 10)+CXCL9(MIG); and

- CRP+IP-10+SAA.

Other potential biosignatures that were identified were:

- CRP+CXCL10(l P- 10)+SAA+ ferritin;

- CRP+CCL1 (I-309);

SAA + CRP + Complement Factor H + ApoA-1 + NCAM + IFN-g + CXCL10/IP-10; and SAA + CRP + Complement Factor H + ApoA-1 + ferritin + IFN-g + CXCL10/IP-10.

In a second part of the study, a multi-biomarker test (MBT) was designed to further evaluate the 3-marker biosignature of CRP+IP-10+SAA for diagnosing TB disease from fingerprick blood samples of participants from Gambia, Uganda, South Africa and Vietnam. The MBT was a UCP- LFA as described above, which is a low cost, quantitative, lateral flow platform.

The results showed excellent performance, with an AUC of 0.88 when evaluated against a composite clinical algorithm. Overall performance when evaluated against GeneXpert Ultra achieved a specificity of 79% when sensitivity was set to 90% with NPV of 97%. Some differences were observed between performance across sites, with Gambia and Uganda performing well and lower performance seen for both South Africa and Vietnam. The increase in specificity when only African participants were included in the analysis is likely due to the fact these markers were derived from pan-African meta-analyses. In this multi-country study, the test reached the WHO target product profile for a screening test for TB regardless of geographical location or HIV status.

The results show the utility of fingerstick blood point-of-care (or near point-of-care) for screening of TB patients. This is preferable to venous blood for (rapid) POC tests, particularly for use in children.

The test is suitable for use in remote settings and provides rapid sample-to-result time and does not require extensively trained personnel. Same day test results will allow immediate and appropriate referrals for confirmatory testing and faster time to treatment initiation.

The invention will now be described in more detail by way of the following non-limiting examples.

Example 1 - Identification of biosignatures

Concentrations of 20 different host protein biomarkers were evaluated in serum samples from individuals that were presumed to have active TB across 7 field sites, situated in 6 different African countries (South Africa, Ethiopia, Uganda, Malawi, Namibia and The Gambia).

Out of the 961 study participants whose data was analysed, 263 (27%) were finally diagnosed with active TB and 698 (73%) with other respiratory diseases (no TB). 184 (19%) of all study participants were infected with HIV, of whom 66 (25% of all the TB patients) had TB (i.e. HIV/TB co-infection). 118 (17%) of the no TB patients also had HIV. Using general discriminant analysis, the abilities of combinations were evaluated between different biomarkers to diagnose active TB in all study participants, regardless of HIV infection and regardless of the study site. Data was split into a training set and a test/validation set (70 to 30% split) and biosignatures identified on the training set (n=677 study participants) were validated on the test/validation set (n=284 study participants). The performance of each signature was benchmarked against the World Health Organization (WHO) target product profiles (TPP) for a triage TB test. That is, sensitivity was set at a minimum of 90% and the resultant specificity calculated. The WHO TPP targets for a triage TB test are sensitivity >90% and specificity >70%.

The biomarkers that were evaluated were: GDF-15, SAA, CRP, fibrinogen, SAP, complement factor H, ApoA-1 , ApoCIII, NOAM, BDNF, Serpin C1 , ferritin, CCL14(HCC-1 ), CCL1 (I-309), CXCL9(MIG), interferon gamma (IFN-g), TNF-alpha, CXCL10(IP-10), TGF-alpha and VEGF-A. Measurements were done using the Luminex® biomarker discovery and validation platform.

Correlations between different host biomarkers and identified biomarkers that are correlated or anticorrelated with the biomarkers that were included in the different 3-marker signatures, were also assessed. The cut-off values and diagnostic accuracies of all individual biomarkers evaluated are shown in Table 1. The accuracies of these biomarkers when used in combination (e.g. 3- marker signatures) in the diagnosis of TB are shown in Table 2. Table 1: Cut-off values and accuracies of the different biomarkers evaluated in the diagnosis of

TB individually. 95% confidence intervals for each parameter are shown. Cut-off values are in pg/ml.

Table 2: Accuracy of the different biosignatures in the diagnosis of active TB in the training dataset (n=677) and the validation (test) dataset (n=284). Biosignatures were derived from the training dataset and their accuracy confirmed by evaluating them on an independent test/validation set which comprised 30% of study participants. Analysis was done regardless of HIV infection status or country of sample origin.

AUC= Area under the receiver operator characteristics curve, Sens = Sensitivity, Spec = specificity, nd = not determined.

*The WHO TPP targets for a triage TB test are sensitivity >90% and specificity >70%.

(a) CRP+SAP+ CCL1 (1-309)

A combination of CRP (Pg/ml), SAP (pg/ml) and CCL1/I-309 (pg/ml) was identified as a promising 3-marker signature from the dataset. The data for this combination are shown in Tables 3-6 and Figure 3.

Table 3: Training data set for CRP, SAP, CCL 1/1-309

Table 4: Test data set for CRP, SAP, CCL 1/1-309

Table 5: Training data set for CRP, SAP, CCL 1/1-309, optimized for Sensitivity (at least >90%), Posterior prob cutoff=0. 12

Table 6: Test data set for CRP, SAP, CCL 1/1-309, optimized for Sensitivity (at least >90%>), Posterior prob cutoff=0. 12

(b) CRP+SAP+CXCL9 (MIG) A promising 3-marker signature of markers with medians >100pg/ml (excludes CCL1/I-309, IFN- g, TGF-alpha, TNF-alpha) was CRP, SAP and CXCL9/MIG (Tables 7-10 and Figure 4).

Table 7: Training data set for CRP, SAP, CXCL9/MIG

Table 8: Test data set for CRP, SAP, CXCL9/MIG

Table 9: Training data set for CRP, SAP, CXCL9/MIG, optimized for sensitivity

Table 10: Test data set for CRP, SAP, CXCL9/MIG, optimized for sensitivity

(c) CRP+CXCL10(IP-10)+CXCL9 (MIG)

The 3 markers with the best ROC AUC were determined to be CRP, CXCL10/IP-10 and CXCL9/MIG (Figures 5-7).

If 2 or more of the following conditions hold, then the subject is classified as TB:

CRP >29877 ng/ml,

CXCL10/IP-10 >29 pg/ml, and

CXCL9/MIG >942pg/ml

(d) CRP+CXCL10(IP-10)+SAA

Another combination identified in this study was SAA, CRP, CXCL10/IP-10 (Tables 11-14 and Figure 8). Table 11: Training data set for SAA, CRP, CXCL10/IP-10

Table 12: Test data set for SAA, CRP, CXCL10/1 P-10

Table 13: Training data set for SAA, CRP, CXCL10/IP-10, optimized for sensitivity

Table 14: Test data set for SAA, CRP, CXCL10/IP-10, optimized for sensitivity

(e) CRP + CCL1 (I309) A potential 2-biomarker signature of CRP and CCL1 (I309) was identified (Tables 16-19 and Figure 9). Table 15: Training data set for CRP, CCL 1 (1309)

Table 16: Test data set for CRP, CCL1(I3O9)

Table 17: Training data set for CRP, CCL 1(1309), optimized for sensitivity

Table 18: Test data set for CRP, CCL1(I3O9), optimized for sensitivity

(f) SAA + CRP + Complement Factor H + ApoA-1 + NCAM + IFN-g + CXCL10/IP-10 A potential 7-biomarker signature of SAA, CRP, Complement Factor H, ApoA-1 , NCAM, IFN-g and CXCL10/IP-10 was identified (Tables 19-22 and Figure 10). Table 19: Training data set for SAA, CRP, Complement Factor H, ApoA-1, NCAM, IFN-g, CXCL10/1 P-10

Table 20: Test data set for SAA, CRP, Complement Factor H, ApoA-1, NCAM, IFN-g, CXCL10/1 P-10

Table 21: Training data set for SAA, CRP, Complement Factor H, ApoA-1, NCAM, IFN-g, CXCL 10/1 P- 10, optimized for sensitivity

Table 22: Test data set for SAA, CRP, Complement Factor H, ApoA-1, NCAM, IFN-g, CXCL 10/1 P- 10, optimized for sensitivity

(g) SAA + CRP + Complement Factor H + ApoA-1 + ferritin + IFN-g + CXCL10/IP-10

Another potential 7-biomarker signature of SAA, CRP, Complement Factor H, ApoA-1 , ferritin, IFN-g, CXCL10/IP-10 was identified (Tables 22-25 and Figure 11 ). Table 23: Training data set for SAA, CRP, Complement Factor H, ApoA-1, ferritin, IFN-g, CXCL10/1 P-10

Table 24: Test data set for SAA, CRP, Complement Factor H, ApoA-1, ferritin, IFN-g, CXCL10/1 P-10

Table 25: Training data set for SAA, CRP, Complement Factor H, ApoA-1, ferritin, IFN-g,

CXCL 10/1 P- 10, optimized for sensitivity

Table 26: Test data set for SAA, CRP, Complement Factor H, ApoA-1, ferritin, IFN-g, CXCL 10/1 P- 10, optimized for sensitivity

Example 2: Testing of CRP+CXCL10(IP-10)+SAA for identifying or ruling out TB

Methods

Ethics statement

Local ethics approval was obtained through the MRC/LSHTM/Gambian government joint ethics committee (MRCG at LSHTM); the Oxford Tropical Research Ethics Committee, the institutional review board at Pham Ngoc Thach Hospital, and the ethics committee of the Ministry of Health, Vietnam (OUCRU); the Stellenbosch University Health Research Ethics Committee (SU HREC) and the Uganda National Council for Science and Technology (MAK). Written informed consent was obtained from all participants prior to sample collection. For participants aged 12-18, parent/guardian consent was obtained alongside assent of the participant. All participants had blood sampling performed prior to knowledge of current TB status.

Patient recruitment and classification

Patients were consecutively recruited from local health clinics at each of the sites. Inclusion criteria included cough > 2 weeks and at least one other symptom suggestive of TB (i.e., weight loss, hemoptysis, night sweats, fever). Sputum was taken for microbiological analysis and fingerstick (capillary) blood was taken for MBT analysis. Patients were classified based on a) GeneXpert Ultra (Cepheid, USA) results above Trace and b) a composite algorithm of microbiological, chest X-ray and clinical parameters. In the composite diagnostic algorithm, the Definite TB group consisted of those with sputum liquid culture positive for Mtb, or those with Xpert Ultra positive (>trace). The Probable group consisted of those with Xpert Ultra trace positive and chest X-ray (CXR) appearance suggestive of TB. Participants for whom all tests were negative, and an alternative diagnosis were made or had resolution of symptoms without TB treatment were classified as ‘no TB’ or ‘other respiratory disease (ORD)’. Unknown or possible classifications were made in cases of contradictory or missing results. Production of a multi-biomarker test (MBT)

Lateral flow strips: 4 mm wide UCP-LF strips specific for a single host protein - CRP, IP-10 and SAA1/A2 - were produced as described in Chegou et al (“Diagnostic performance of a sevenmarker serum protein biosignature for the diagnosis of active TB disease in African primary healthcare clinic attendees with signs and symptoms suggestive of TB”. Thorax. 2016 Sep;71 (9):785-94), the contents of which are incorporated herein. Each Test (T) line comprised 200 ng of the following antibodies: mouse-anti-human CRP mAb (C5; Labned.com, Amstelveen, the Netherlands), mouse-anti-human IP-10 mAb (B-C55; Diaclone Research, Besancon, France) and mouse-anti-human SAA1/A2 mAb (865504; R&D systems, Minneapolis, MN, USA). To detect unbound UCP-conjugate, the downstream Flow-Control (FC) line comprised 100 ng goat-anti- mouse (M8642; Sigma-Aldrich, St. Louis, MO, USA).

UCP coniuoates

Antibodies were conjugated to luminescent up-converting reporter particles (UCP) allowing quantitative measurements. Polyacrylic acid functionalized UCPs (200 nm, NaYF₄:Yb³⁺, Er ³⁺; Intelligent Material Solutions Inc., Princeton, NJ, USA) were conjugated according to previously described protocols. Mouse-anti-CRP (CRP135; Labned.com, Amstelveen, Netherlands), mouse-anti-IP-10 (B-C50; Diaclone Research, Besancon, France) and mouse-anti-SAA1/A2 (924903; R&D systems, Minneapolis, MN, USA) were bound at a concentration of 50 pg antibody per mg UCP. UCPs were incorporated in the sample/conjugate pad at a density of 200 ng per 4 mm (CRP, SAA1/A2) or 400 ng per 4 mm (IP-10) and air dried.

FSB was collected using disposable 20pl Minivette® collection tubes (Heparin coated; Sarstedt) and directly mixed with 180pl high salt finger stick (HSFS) buffer: 100 mM Tris pH 8.0, 270 mM NaCI, 1% (v/v) Triton X-100, and 1% (w/v) BSA.⁶⁸ 1 OOpI of the lysed blood was added to a disposable microwell after which lateral flow strips were added corresponding to each of the 3 markers (SAA, CRP, IP10). The LF strips were dried and then analysed using the Lateral flow Quant reader (DIALUNOX, Germany). For the individual biomarkers, test results are displayed as the ratio (R=T/FC) between Test and Flow-Control UCP-signal measured at the respective lines. All scans were quality controlled to ensure the peaks were in the appropriate place and were rescanned if any errors had occurred. Tests that failed could not be repeated and participants were excluded from further analysis.

HIV testing

All participants without known HIV infection received voluntary counselling and testing using a rapid fingerstick blood HIV- 1/2 test (Alere, USA). If the rapid test was positive, the result was confirmed either with serology testing or a second rapid test. CD4 counts were determined for all HIV+ individuals at baseline.

Statistical analysis

A Fisher’s exact test was used to compare sex, age, prior TB, and HIV status between the groups. Data were exported from the REDCap database and imported into R (v. 4.3.1 ) running on Ubuntu 22.04.3 on a Dell PowerEdge R650. Except where otherwise stated, participants classified as definite TB or probable TB were regarded as TB Positive. Participants classified as Possible TB or Unknown were excluded from the analysis.

Modelling was done using the tidymodels framework (v. 1.1.1 ). The dataset was divided into training and test sets using a 3:1 ratio split stratified by outcome. All numerical variables were scaled to have a mean of zero and a standard deviation of 1. Nominal variables were dummy- coded. Four models were tuned using 10-fold nested cross-validation with 50 bootstrap repeats on the inner loop. The selected algorithms were k-Nearest Neighbors (kknn v. 1.3.1 ), Random Forest (ranger v. 0.15.1 ), Elastic-Net Logistic Regression (glmnet v. 4.1.8) and XGBoost (xgboost v. 1.7.5.1 ). Parameter optimisation learning was done using a Bayesian search process on the inner bootstrap process with each optimal solution then being evaluated against the corresponding left-out fold of the outer loop. The algorithm that performed best on the combined 10 outer fold evaluations was chosen to be trained on the full training dataset, re-balanced using the SMOTE algorithm (themis v. 1 .0.2). Cut-point thresholds for Youden’s J as well as a selection of defined specificities (0.8, 0.85, 0.9 & 0.95) were determined from the training ROC curve. The thresholds are the probabilities predicted by the regression model for any combination of marker values. The trained model was then used to predict outcomes on the test dataset and performance metrics were calculated at the thresholds previously established with bootstrap (n=5000) 95% bias-corrected and accelerated confidence intervals. Only results for the test set are presented.

Results

Patient information

1262 adults and adolescents (12-70 years) presenting at primary health care facilities with respiratory symptoms were screened prior to microbiological confirmation. There were 52 screen failures. 205 participants were evaluated using a different version of the MBT test and were thus excluded from final analysis. Thus, final analysis was performed on a total of 1003 participants. Both microbiological and clinical parameters were included to define a diagnostic algorithm. This included 'definite TB’ (n=224), ‘probable TB’ (n=7) 'possible TB’ (n=6), ‘Unknown’ (n=30) and ‘No TB’ (TB negative but other respiratory disease (ORD), n=735) (data not shown). Results were divided into a training set (n=721 ) and a test set (n=281 ).

Definite TB and probable TB were combined for final data analysis. 73% of the definite TB group were males compared to 53% of the ORD group (p=0.0001 ). Additionally, 5% of the definite TB group were HIV positive compared to 10% of the no TB group (ns). No significant difference in age was seen between the groups with median [IQR] of 32.1 [16.3-47.9] years for the definite TB group and 34.7 [15.2-54.2] years for the ORD group. 9% of the definite/probable group had prior TB compared to 24% for the unknown and 16% for the no TB group. No significant difference in BMI was observed between the groups although the definite/probable and unknown groups had lower BMI compared to the possible and no TB groups (18/18.2 compared to 20.5/20.9).

Two large production lots of UCP-LF strips were used with materials stored and shipped at ambient temperature. Major differences were not observed between the batches, hence no adjustments were made in the final analysis.

Evaluation of the MBT against GeneXpert Ultra

When the performance of the MBT test was evaluated against GeneXpert Ultra classification of participants with readings above Trace (n=281 for the test set), it was able to classify TB with an AUC of 0.869, sensitivity of 73% (95% Cl, 58-88%) and specificity of 86% (95% Cl, 80-92%) (Figure 12). When performance was evaluated at 90% training sensitivity threshold, specificity was 74% (95% Cl, 66-82%) (Table 27).

Table 27: MBT performance (set at 90% training sensitivity)

Evaluation of MBT against a composite algorithm

The test was evaluated against a composite algorithm including microbiological and clinical information. Definite and probable TB groups were combined and possible and unknown groups were excluded. Analysis of the training data set resulted in a classification of TB with an AUC of 0.88, sensitivity of 79% and specificity of 86% (Figure 13). For the test dataset, TB was classified with an AUC of 0.88, sensitivity of 71% (95% Cl, 55-86%) and specificity of 90% (95% Cl, 84- 96%) (Figures 14A and B). When sensitivity was set to 90% for a triage test, specificity was 80% (72-87%) for MBT (at 90% training sensitivity threshold) (Table 26).

When only HIV negative participants were analyzed, no change (AUC 0.88, sensitivity 70% (95% Cl, 53-86%) and specificity 91% (95% Cl, 86-97%)) was observed (Figure 14B). When only HIV positive participants were analyzed, test performance reduced but this is likely due to the smaller sample size in the cohort (n= 17 for definite TB and 83 for no TB) (data not shown).

Analysis of performance at individual sites

Due to different demographics, host and bacterial genetics, the performance of the test was evaluated at individual sites (Figure 15). The site with the best discriminatory performance was The Gambia with an AUC of 0.95 (Figure 15A). When sensitivity was set to 90% for a triage test, specificity was 56% (39-72%). Analysis of Ugandan participants classified TB with an AUC of 0.95 for the MBT test (Figure 15A). When sensitivity was set to 90%, specificity was 86% (95% Cl, 74- 95%). Analysis of South African participants classified TB with an AUC of 0.79. When sensitivity was set to 90% for a triage test, specificity was 70% (58-83%). However, sensitivity reduced to 75% for the test set. Analysis of Vietnam participants classified TB with an AUC of 0.86. When sensitivity was set at 90%, specificity was 76% (95% Cl 63-88%); however, sensitivity reduced to 77% for the test set.

Since the initial signatures were developed based on pan-African samples, and due to differences in disease severity in Vietnam (significantly lower numbers of GeneXpert high readings compared to Gambia (p<0.0001 )), pan-African analysis was also performed. This increased classification of TB to an AUC of 0.90 (Figure 15B). When sensitivity was set to 90% for a triage test, specificity increased to 78% (95% Cl 68-87%) (Table 26).

Classification of unknowns

Using predictive models, the possible and unknown TB groups were classified using the MBT test. 24/26 (92%) unknown and 4/4 (100%) possible TB were classified as TB when the same threshold was applied as for the definite TB group. For individuals where results for both tests were available, MBT classified 16/30 (53%) and 28/30 (93%) unknown/possible as ‘TB’ (p=0.0009).

Claims

1 . A method of diagnosing tuberculosis disease (TB) or ruling-out TB in a subject, the method comprising the step of testing a biological sample from a subject for the presence of C- reactive protein (CRP) and at least one other biomarker selected from the group consisting of SAA (Serum amyloid A), C-X-C motif chemokine ligand 10 / interferon gamma inducible protein 10 (CXCL10/IP-10), Serum amyloid P (SAP), Chemokine (C-C motif) ligand 1 (CCL1/I-309), C-X-C motif chemokine ligand 9 / monokine induced by interferon Gamma (CXCL9/MIG) and ferritin.

2. A method according to claim 1 , which comprises testing the biological sample for the presence of CRP and either or both of SAA and CXCL10/IP-10.

3. A method according to claim 1 or 2, which comprises testing the biological sample for the presence of CRP, SAA and CXCL10/IP-10.

4. A method according to any one of claims 1 to 3, which comprises testing the biological sample for the presence of CRP, SAA, CXCL10/IP-10 and ferritin.

5. A method according to claim 1 , which comprises testing the biological sample for the presence of :

- CRP, SAP and CCL1/I-309;

- CRP, SAP and CXCL9/MIG;

- CRP, CXCL10/IP-10 and CXCL9/MIG; ferritin;

- SAA + CRP + Complement Factor H + ApoA-1 + NCAM + IFN-g + CXCL10/IP-10; or

- SAA + CRP + Complement Factor H + ApoA-1 + ferritin + IFN-g + CXCL10/IP-10.

6. A method according to any one of claims 1 to 5, which includes the steps of: contacting the biological sample with capture agents which bind to the biomarkers intended to be detected; and detecting binding of the capture agents to the biomarkers.

7. A method according to any one of claims 1 to 6, wherein the biological sample is a blood sample, such as capillary or fingerstick blood.

8. A method according to any one of claims 1 to 7, wherein the TB is pulmonary TB.

9. A method according to any one of claims 6 to 8, wherein a lateral flow assay is used to detect the biomarkers.

10. A method according to any one of claims 6 to 9, wherein the capture agents are conjugated to luminescent up-converting reporter particles.

11. A device for diagnosing TB according to the method of any one of claims 1 to 10, the device comprising: a means for receiving a sample from a subject suspected of having TB; capture agents for binding at least one of the biomarkers intended to be detected; and at least one indicator to indicate to a user of the device when the capture agents bind to the biomarkers.

12. A device according to claim 11 , which is a lateral flow assay strip.

13. A device according to claim 11 or 12, wherein the capture agents are conjugated to luminescent up-converting reporter particles.

14. A device according to any one of claims 11 to 13, wherein the capture agents are antibodies.

15. A device according to any one of claims 11 to 14, which includes capture agents for binding only one of the biomarkers intended to be detected.

16. A device according to any one of claims 11 to 14, which includes capture agents for binding all of the biomarkers intended to be detected.

17. A cartridge for diagnosing TB according to the method of any one of claims 1 to 10, the cartridge comprising a plurality of lateral flow assay strips, each lateral flow assay strip comprising capture agents for binding a single biomarker intended to be detected, wherein the cartridge comprises a lateral flow assay strip for each biomarker to be detected.

18. A system including a device according to any one of claims 11 to 16 or a cartridge according to claim 17 and a reader for analysing the device.

19. A kit for diagnosing TB according to the method of any one of claims 1 to 10, the kit comprising one or more of the following: capture agents for binding the biomarkers intended to be detected; means for obtaining, collecting or receiving a biological sample from a subject; fingerstick buffer; one or more devices according to any one of claims 11 to 16; a cartridge according to claim 17; and/or instructions, in electronic or paper form, for performing the method.

20. A method of diagnosing a human subject as having TB and treating the subject, the method comprising the steps of: testing a biological sample from a subject suspected of having TB for the presence of CRP and at least one other biomarker selected from the group consisting of SAP, SAA, CCL1/I-309, CXCL9/MIG and CXCL10/IP-10; determining that the subject has TB based on the detection of the biomarkers in the sample; and administering an effective amount of TB treatment to the subject if the subject is in need thereof.

21 . The use of capture agents for binding CRP and at least one other biomarker selected from the group consisting of SAA, CXCL10/IP-10, SAP, CCL1/I-309, CXCL9/MIG and ferritin in the manufacture of a kit or device for diagnosing TB.

22. The use according to claim 21 , wherein the capture agents specifically bind CRP and either or both of SAA and CXCL10/IP-10.

23. The use according to claim 21 or 22, wherein the capture agents specifically bind CRP, SAA and CXCL10/IP-10.

24. The use according to any one of claims 21 to 23, wherein the capture agents specifically bind CRP, SAA, CXCL10/IP-10 and ferritin.

25. The use according to claim 21 , wherein the capture agents specifically bind:

- CRP, SAP and CCL1/I-309;

- CRP, SAP and CXCL9/MIG;

- CRP, CXCL10/IP-10 and CXCL9/MIG; ferritin;

- SAA + CRP + Complement Factor H + ApoA-1 + NCAM + IFN-g + CXCL10/IP-10; or

- SAA + CRP + Complement Factor H + ApoA-1 + ferritin + IFN-g + CXCL10/IP-10.