FIELD OF THE INVENTION The invention relates to detection of TB, in particular childhood TB.

BACKGROUND TO THE INVENTION

Detection of TB is a problem, particularly in children. The current gold standard involves bacteriological assessment. However, sputum samples can be difficult to obtain. Even when successfully obtained, sputum sampleefrom children can exhibit a paucity of bacilli, making direct detection in the sample difficult or imposeibla In these circumstances, culture must be carried out in order for the low number of bacilli in the sample to expand to detectable levels. This is laborious and costly and requires specialised laboratory resources, which ve drawbacks. More importantly, culture still lacks sensitivity in children. Moreover, culture takes approximately six weeks and this introduces a clinically significant delay to theobtajning of the diagnosis which is a serious problem for patient outcomes. In addition, it is particularly difficult to obtain sputum sampleefrom children, both in practice and in voluma Placing patients, eepedaJlychildren, on apeculative treatment without adefinitivediagnoalsisaserious cost burden as well as the medical risks and complicationswhich such a step would entail.

Dhanasekaran et al 2013 (Qenesand Immunity vol 14 pages 356-364) descr i be identification of biomarkera for Mycobacterium tuberculosis (M.tb.) infection and disease in BCG- accinated young children in Southern India, A combination of 11 biomarkera is described with only moderate disai minatory power. Unstimulated whole blood super natants did not identify cytokine expression differences (pages 359- 360).

WO2014 020343 (Protein logic Ltd) discloses biomarkera for diagnosing and/ or monitoring tuberculosis. This document isfocussed on adulta Childhood TB is mentioned only once, and the age of the subjects is not specified. The only

exemplification la confined to adulta

Kumar et aJ 2013 (Clinical and Vaccine Immunology vol 20 pages 704-711) discloses circulating biomarkera of pulmonary and extrapulmonary tuberculosis in children. It is disclosed that paediatricTB was associated with elevated plasma TGF-bet a, IL-21 and IL-23 level* It is disclosed that no significant diffsrenose were found for cytokines, for most type 17 and type 1 i nterferons, or moat cytoki nee associated with immune modulation.

Hur et aJ 2015 (Journal of Infection vol 70 pages 346-355) disclose adjunctive biomarkersfor improving diagnosis of tuberculosis and monitoring therapeutic effecta VEQF is discussed. No disclosure of childhood TB. Serene et a) 2012 (Biomarkersvol 17 pages 1 - 8) disclose host biomarkers of clinical relevance in tuberculosis: review of gene and protein expression studies. IL-6, IL-22 and I P-10 were mentioned. No disclosure of childhood TB.

Sutherland et aJ 2012 (PLoSONE vol 7epub number: Θ30324) disclose highly accurate diagnosis of pleural tuberculosis by immunological analysis of the pleural effusion. IL- 6 and I P-10 are mentioned. Thework isfocusead on adults. Thework isfocusssd on pleural fluid.

WO2015/ 040377 (Medical Research Council) discloses biomarkersfor tuberculosia IL-1ra, FQFand VEGF are mentioned. Thework isfocusssd on adults. Thework is focussed on sputum.

The current gold standard for diagnosis is direct detection of the infectious pathogen Mycobacterium tuberculosis (M.tb.) in adinicaJ specimen such as sputum.

Differentiating active tuberculosis (TB) disease from other respiratory tract infections (CO) constitutes a major challenge in the management of children with suspected intrathoracic TB disease. The present invention seeks to overcome pr obi em (s) associated with the prior art.

SUMMARY OF THE INVENTION

Prior art methods have been based on analysis of sputum, or have been baaed on extraction and manipulation (such as stimulation) of white blood cells. By contrast, the present inventors have based their detection on indicators which can be found directly in samples taken from the patient. In particular, the inventors have based their detection on a blood sample, and direct detection of markers present in that blood sampla This has the advantage of lending itself to development of a point of care test. This hasthe advantage of avoiding manipulations and stimulations which are part of the prior art techniques. In particular, the invention is advantageously baaed on unstimulated blood supernatant.

Thuein one aspect the invention provides a method for the diagnosis of TB in a subject, the method comprising;

(a) providing a sample from said subject, said sample being selected from the group consisting of: blood, serum and plasma;

(b) determining the concentration in said sample of the following biomarkers: IL-1ra, IL6, 1 L-7, 1 L-8, IL-12p70, FQF-basic IP-10, and VEQF;

(c) converting each biomarker concentration determined in (b) into adedlevaJue; and

(d) converting each decile value into a binary presence or absence by comparing the decile values of (c) to thefollowing specif icquantilecut- off values:

off value is converted into the binary presence of the biomarker, and a decile value lower than thespedficquantilecut-off valueisconverted intothebinary absence of the biomarker;

wherein detecting the presence of each of said biomarkers indicates that the subject has

TB.

Suitably step (c) converting each concentration determined in (b) into adedlevaJue comprises the steps of :

(d) comparing theconcentration of each biomarker determined in (b) to a reference frequency distribution of concentrations of said biomarker; and (di) reading out thededlevaluefrom the frequency distribution for the

concentration of said biomarker.

Suitably step (c) converting each concentration determined in (b) into adedlevaJue comprises the steps of:

(d) comparing theconcentration of each biomarker determined in (b) to a kernel density est! mate of concentrations of said biomarker ; and

(di) reading out thededlevaluefrom the kernel density estimate for the

concentration of said biomarker.

Suitably the reference frequency distribution or kernel density estimate is generated by measuring theconcentration of the biomarker in a number of subjects, for exam pie a minimum of 100 subjects, and compiling those measurements into a frequency distribution/ kernel density estimate, Alternatively the frequency distributions (kernel density est! mates) presented in Rgures2 to 9 herein maybeused. In this embodiment step (c) converting each concentration determined in (b) into adedlevaJuecomprisas the steps of:

(d) comparing theconcentration of each biomarker determined in (b) to the corresponding reference frequency distribution / kernel density estimate of concentrations of said biomarker selected from Figures 2 to 9; and

(di) reading out thededlevaluefrom the frequency distribution / kernel density estimate for theconcentration of said biomarker.

Suitably determining theconcentration of each biomarker comprises:

(bi) detection by contacting the sample with an antibody or antigen binding fragment thereof capable of specifically binding the biomarker; and

(bii) quantification of said binding.

Suitably determining theconcentration of each biomarker comprises detection of the mRNA for thebiomarker, wherein detection of the mRNA comprises:

(bi) contacting the sample with specific nucleic arid probe(s) or primer(s) for the biomarker; and

(bii) quantification of said probe(s) or primer(s). Suitably said probe or primer is a non-naturally occurring nudeicadd sequence.

Suitably said probe or primer is an artificial or man-made molecula Suitably said probe or primer is isolated and/ or purified. Suitably said probe or primer comprises dntfe stranded nucleic add. Suitably said probe or primer comprises a label moiety attached thereto. Suitably said label is covaJently attached. Suitably said label may be a fluorescent or radioactive label or aQdot, nanocrystaJ or nanopartide, most suitably a fluorescent label.

Suitably said sample is a sample of serum or plasma

Suitably said serum or plasma is essentially cell free. Suitably the subject is 16 years old or younger, preferably 15 years old or younger. Suitably the subject is 2 years old or older, preferably 5 years old or older. Suitably the subject is 5 to 15 years old.

In one embodiment, suitably said method further comprises determining the concentration in said sample of biomarker EC-stimulated VEQF (specif icquantile cutoff value 2).

In one aspect, the invention relates to a kit comprising reagsnt(s) for the specific detection of each of thefoHowing biomarkers: I L-1ra, I L6, 1 L-7, 1 L-8, 1 L-12p70, FQF- basic lP-IO. and VEGF.

In one aspect, the invention rdatesto akit comprising reagents for the specific detection of mRNA encoding each of the following biomarkers: IL-1ra, IL6, 1 L-7, 1 L-8, IL-12p70, FGF-basic IP- 10, and VEQF.

Suitably add reagents each comprise an antibody or antigen binding fragment thereof selected from the group consisting of a Fab fragment, a Fab' fragment, a F(ab')2 fragment, ascFv, aFv, arlgQ, and adiabody. In one aspect, the invention rdatesto a device comprising an array of materiaJswhich together are capable of specifically binding each of thefollowing biomarkers: I L-1ra, IL6, 1 L-7, 1 L-8, IL-t2p70, FQF-badc, IP-10, and VEQF, each material within the array being capable of specifically binding one of sdd biomarkers. In one aspect, the invention rdatesto a device comprising an array of materiaJswhich together are capable of detecting mRNA specific for each of thefollowing biomarkers: IL-1r¾ IL6, IL-7, IL-8, IL-12p70, FGF-baeic, IP-10, and VEQF, each material within the array being capable of specifically detecting one of said mRNAs.

Suitably said device is a lateral flow device.

In one aspect, the invention relates to a method of treating a subject comprising carrying out the method as described above, wherein if it isdetermined that the subject hasTB, a regimen of 2HRZE/4HR (2 months HRZE followed by 4 months HR wherein H - isoniazid, R« rifampicin. Z- pyrazinamide, E - ethambutol) is administered to said subject.

In one aspect, the invention relates to use of HRZE wherein H - isoniazid, R- rifampidn, Z - pyrazinamide, E - ethambutol for treatment of TB in a subject, wherein the method as described above is carried out for said subject, wherein if it is

determined that the subject has TB then HRZE is administered to said subject for two months and then H R is administered to said subject for four months. Suitably a treatment regimen of dosing said subject at least three times per week during the first two months is selected, preferably a treatment regimen of dosing said subject daily during the first two months is selected.

I n one aspect, the I nvention relates to tablets of H 75mg+ R 150 mg + Z 400 mg + E 275 mgfor treatment of TB in a subject, wherein the method as described above is carried out for said subject, wherein if it is determined that the subject has TB then HRZE is administered to said subject for two months, followed by 3 tablets of H 75 mg + 1.5 tablets of R 150 mg for four months.

In one aspect, the invention relates to a process for selecting a treatment regimen, said process comprising

carrying out the method as described above, wherein if it isdetermined that the subject has TB then a treatment regimen of 2HRZE/4HR (2 months HRZE followed by 4 months HR wherein H - isoniazid, R - rifampidn, Z - pyrazinamide, E - ethambutol) is selected.

I n one aspect, the i nvention relates to use of a combi nation of materials each of which recognises, specifically binds to or has affinity for one of the following biomarkers: IL- 1ra, IL6. IL-7, IL-8, IL-12p70, FQF-basic IP-10, and VEQF, wherein said combination includes at least one such material for each of said biomarkers, for aiding diagnosis of TB in a subject. Suitably said material comprises an antibody or antigen binding fragment thereof.

In one aspect, the invention relates to use for aiding diagnosis of TB in a subject, of a combination of materials each of which recognises, specifically binds to or has affinity for mRNAof one or moreof the following biomarkers: IL-1ra, IL6, IL-7, IL-8, IL-12p70, FGF-baaic, I P-10, and VEGF. Suitably said material comprises a nucleic add primer or probe. In one aspect, the invention relates to a apparatus comprising logic configured to carry out the method as described above.

In one aspect, the invention relates to a computer program product operable, when executed on a computer, to perform the method steps as described above.

DETAILED DESCRIPTION OF THE INVENTION

The inventors teach identification of novel host biomarkers for childhood TB. The inventors hypothesised that a unique bioslgnature for TB disease in children could be identified.

TB' means Tuberculosis; thieisadisease caused by the bacterium Mycobacterium tuberculosis (sometimes referred to as MTB).

It has been a challenge in the field of TBto provide a simple diagnostic test. The test nesdsto bereliabla Thetest needs to be accurate The test should ideally involve the minimum of tooling/ equipment, especially since TB is often a problem in developing oountrieswhere laboratory facilities can be few and/ or can be geographically distant from the patients being tested. SAMPLE

The sample may be from a subject. The subject is suitably a mammal, most suitably a human. Suitably the methods do not involve actual collection of the sample Suitably the sample is an in vitro sample Methods of the invention are suitably performed on an isolated sample from the subject being investigated. Thus, suitably the methods are methods which may be conducted in a laboratory setting without the need for the subject to be present. Suitably the methods are carried out in vitro i.a suitably the methods are in vitro methoda Suitably the methods are extracorporeal methoda

Suitably the invention may be applied to analysis of nucleic acids. Suitably, nucleic add is prepared from the sample collected from the subject of interest, 04. by extraction of nucleic add from white blood cells in the sampla Suitably, the sample comprises nudeic add. Suitably, the sample consists of nucleic add. Suitably, the nuddc add comprises, or is, mRNA or cONA, suitably mRNA.

Most suitably the invention may be applied to analysis of protein biomarkera Most suitably proteins in the sample are analysed.

Suitably the sample is an in vitro sampla

Suitably the sample is an extracorporeal sampla

Suitably the sample is blood.

Suitably the sample is blood supernatant.

Suitably the sample is serum. Serum may be obtained as the fluid collected from a blood sample which has been dotted.

Suitably the sample is plasma Plaemamay be obtained as the fluid collected from a blood sample which has been centrif uged to pellet the blood cells present. Alternatively plasma may be obtained by filtration to remove the blood cells present. Suitably the blood or blood supernatant/serum/plasmaisunstimulated.

It is an advantage of the invention that the analysis converts each measured biomarker into a binary output. For example, each cytokine biomarker examined isturned into a binary YES/ NO data point. By contrast, existing prior art approaches need quantitative readouts. It is an advantage of the invention that diagnostic accuracy is achieved at a level comparable to bacteriological culture It is an advantage of the invention that there are no timedelaysoomparableto those experienced with bacteriological cultura Prior art techniques use cells, either bacterial cells or isolated blood cella It is an advantage of the invention that no cella are required to be cultured, and no cells are required to be isolated.

It is an advantage of the invention that the host response/ host signature is analysed. Without wiahi ng to be bound by theory, the exist! ng attempts to create a blood test for TB havefocussed on challenging or stimulating immunecellsfrom the subject and studying the response. However, in principle this approach isaJways examining a "recall response". This is a response derived from "memory" of the immune system which haspreviously encountered aTB bacterium. Rrstly. thisrequiresastimulation of the sample either with bacteria or with antigens, which is labour intensive and costly. However, more importantly, in principle, this type of approach is only able to assay a secondary response. By contrast, the present invention is concerned with assessment of the primary response. This is especially important and useful when applied to children, since clearly each individual patient is at some point in their lives exposed to TB for the first ti ma If a method such as a prior art method isonlyfocussed on assessing a secondary response, then it is unlikely ever to be able to detect a response the first time such a subject encounters TB. It is an advantage of the present invention that thedirect primary response is being assessed. SuitaWythesampJeisacell freesampla SUitablythesampledoesnot comprisecells.

Suitably thecells are removed from the Wood sample by centrifugation and retrieval of the supernatant. OBIIS may be removed from the sample by any method known in the art. For example filtration.

A key reason for excluding cells from the analysis is because blood is deeply red in colour dueto the presence of the erythrocyte red blood cella Typically the detection step used to assess the presence or absence of the biomarkers is light ssnsitiva

Therefore, advantageously the sample is free of cells in order to avoid being

confounded by the red colour of whole blood. I n principle, any approach which d rcumvents the deep red colour of whole blood could be employed. Suitably this may be by cell lyaia MoresuitaHytheceMsere removed from the Mood before analysis. Suitably this is by csntrifugsiion. Suitably this may be by filtration. Suitably this may be by lateral flow.

In some settings, it may be possible to use a whole blood sample in the method of the invention, for example by using lateral flow. In this scenario, the whole Wood sample is placed in a lateral flow device. Thefluid component such as plasma/ serum then migrates, so at the point of assessment it is cell free.

LATERAL FLOW ASSAY AND OTHER DEVICES

Lateral flows teat assays (LFA) - also referred to as lateral flow

i mm unochrom at ographic assays- require minimal infrastructure and have been used to develop cheap and simpledevicesfor rapid medical diagnosis and screening, point of care tests or laboratory use. The assay is based on the detection of the presence of a specified target analytein a sample for mostly qualitative and occasional quantitative analysia Common applications of the LFA include its use in home pregnancy test, monitoring of diabetes and rapid diagnosis of H I V or parasitic and bacterial infections. As discussed extensively in review articles (such asSajidM, KawdeA, DaudM.

Designs, formats and applications of lateral flow assay: A literature review. Journal of Saudi Chem ical Society. 2014), a typical LFA strip is made up of four parts:

• Sample application pad: This is an absorbent pad made of cellulose or glass fibre on which the sample is applied and its main function is to transport the sample (ag. blood) containing the an alyte 'downstream' to the other part of the LFA strip. It may also pro-treat the sample, including separation into components such as plasma, before its transportation.

• Conjugate pad: this contains immobilized and labelled antibody that is

specific for the target anaJyta The antibody is conjugated to coloured particles such as latex or nanometre sized particles and the labelled antibody conjugate is released upon contact with moving liquid sampla

• Nitrocellulose or Reaction membrane: This membrane allows movement of complexes generated from the conjugated pad under capillary action. The membrane is further divided into test and control lines.

• Adsorbent pad: This pad works as a 'sink' at the end of the strip and it is designed to further draw the sample across the reaction by capillary action. RgurelO shows the standard layout of a lateral flow device (Saied Assadollahi, ChristianeReininger, Roland PaJkovits, Peter Pointl and Thomas SchaJkhammer. 'From Lateral Row Devices to a Novel Nano-Oolor MicrofluidicAseay.' Sensors 2009{9);6084-6100; doi:10.3390/S9080e08 ).

Lateral flow assay test can work i n two mai n formats, which are the sandwich and competitive assaya Sandwich LFA are designed for detection of large molecular weight molecules including proteins with multiple antigenic sites (ag. HIV, hCG), while competitive assays are designed to test small molecules with single epitopes. In sandwich LFA, apositiveteet will show coloured bands on the test line, while in competitive LFA the test line will show coloured band in negative samples.

M ultlplex detection format

In clinical diagnosis, the higher specificity or predictive accuracy of multiple inter- dependent analytesthat are detected simultaneously under the same condition has led to the development of multiplex LFA detection format used for the detection of more than one target anaJyte(PanhotraBR, Hassan ZU, Joshi CS, Bahrani A. Visual detection of multipieviraJ ampl icons by dipstick assay: its application in screening of blood donors a welcome tool for the limited resource settings. Journal of clinical microbiology.2005 Dec;43(12):6218; author reply -9. Pub ed P ID: 16333138.

Pubmed Central PMCI D: 1317223; Corstjens PL, de Dood CJ, van der Ploeg-van Schip JJ, Wiesmeijer KC, Riuttamaki T, van Meijgaarden KE, et aJ. Lateral flow assay for simultaneous detection of cellular- and humoral immune responses. Clinical biochemistry.2011 Oct;44<14-15):1241-6. PubMed PMID: 21763300. Pubmed Central PMQ D: 3177995). I n this format, the assay is performed over a strip with number of test lines equal to the number of target analytes as recently described for thedetection of four common human papillomavirus (HPV) types (Xu Y, Liu Y, Wu Y, XiaX, Liao Y, Li Q. Fluorescent probe- based lateral flow assay for multiplex nucleic add detection. Analytical chemistry.2014 Jun 17;86(12):5611-4. PubMed PMID: 24892496).

Figure 11 shows multiplex detection format of LFA (YeXu, Yinghua Liu, Yan Wu, Xiaohu Xia, Yiqun Liao and Qingge Li. 'Rourescent Probe-Based Lateral Row Assay for multiplex Nucleic Add Detection.' AnaJ Chem., 2014, 86(12), 5611-5614). Clinical applications

Lateral Row Assays have found important use in clinical diagnosis and as point of care tests through thedetection of clinical analytes in plasma, serum, urine and other clinical samples. A ready example is the home pregnancy testing kits. In tuberculosis (TB), the urinary lipoarabinomannan (LAM) test is a LFA test with low sensitivity (52 - 59% but a high specificity (>94%) (Lawn SD, Dheda K, Kerkhoff AD, Peter JO, Dorman S, BoehmeCC, et aJ. DetermineTB-LAM lateral flow urine antigen assay for HIV- associated tuberculosis: recommendations on the design and reporting of dinicaJ studies. BMC infectious diseases.2013;13:407. PubMed PMID: 24004840. Pubmed Central PMCID: 3846798). The inventors also recently reported that the use of fluorescent up-converting phosphor (UCP) reporter technology combined with LFA to detect IP- 10 and CCL4 simultaneously on the same strip has potential to be developed asapoint of care test for pleural TB (Sutherland JS, Mendy JF, Gindeh A, WaJzl G, Togun T, Owolabi O, et aJ. Use of lateral flow assays to determine I P-10 and CCL4 levels in pleural effusions and whole blood for TB diagnosis. Tuberculosis (Edinb).2015). A multi-centre study in Africa similarly reported the applicability of the low-tech and robust UCP- LFA platform asaconvenient quantitative assay for detection of multiple chemokines in whole blood (Corstjens PL, Tjon Kon Fat ΕΞΜ , de Dood CJ, van der Rosg-van Schip JJ, Fran ken KL, Chegou NN, et aJ. Multi-center evaluation of a user- friendly lateral flow assay to determine I P-10 and CCL4 levels in blood of TB and non- TB cases in Africa ainicaJ biochemistry.2015 Aug 15. PubMed PMID: 26285074). Therefore, the invention relates to a device comprising an array of materials which together are capable of binding each of thefollowing biomarkers: I L-1ra, I L6, 1 L-7, 1 L- 8, IL-12p70, FQF-basic IP- 10, and VEEGF, each material within the array being capable of binding one of said biomarkers, wherein said device is a lateral flow device. I n this embodiment, the array of materials suitably comprises a number of test lines equal to the number of biomarkers. Suitatty ee* test MnecompriseB material capable of binding one such biomarker. Suitably each test line comprises material capable of binding a different such biomarker.

Alternatively, if the device comprises a'chip' or 'biochip', the array may comprise a spatial arrangement of the materials such as a grid or other defined arrangement such as a geometric pattern.

Suitably each material capable of binding one of said biomarkers is immobilised within the device

Suitably each material capable of binding one of said biomarkers is modified.

Suitably each material capable of binding one of said biomarkers is labelled. Suitably the label iscovaJently attached. Suitably the label is a dya Suitably each material capable of binding one of said biomarkers is a different antibody or antigen binding fragment thereof, wherein the antigen binding fragment thereof is selected from the group consisting of a Fab fragment, a Fab' fragment, a F(ab')2 fragment, ascFv, aFv, arlgO, and adiabody.

Suitably the antibody or antigen binding fragment thereof is a non-human antibody or antigen binding fragment thereof. Suitably the antibody or antigen binding fragment thereof is recombinant, ag. made by in vitro expression of a recombinant nucleic add sequence. Suitably the antibody or antigen binding fragment thereof is purified and/ or isolated.

Non-antibody reagents for detection ag. of protein biomarkers as described above may also be used, (ag. for detection in the methods, as material in the devices, as reagents in the kits, or other applications of the invention) for example phage display particles offering specific binding peptides, affimers, apt am era, nucleic adds binding specific protein or peptide sequences, small molecules with specific binding properties or other such spedficbinding partner(s) of the biomarkers described.

SIGNATURE

Prior art approaches have identified extremely large signatures such as requiring the analysis of 50 or more individual biomarkers. It is an advantage of the invention that the signature requires analysis of only 8 biomarkers.

Prior art attempts have involved flow cytometry. However, flow cytometry is extremely labour intensive, expensive, and is unsuitable for the provision of a bedside/ point of care test.

Prior art approaches have involved analysis of the transcriptome(i. a of mRNAs).

However, these approaches have typically also involved assessment of at least 50 genes, which is a problem.

Data is presented i n the exam pies section i n support of the AUG/ spedf idty/ sensitivity of the methods of the invention.

Most importantly, the positive and nepjative predictive values areprovided. In the field of TB, the predictive values may be regarded as more important even than the sensitivity/ specificity of the method. It is an advantage that the invention delivers extremely robust positive and negative predictive values. This is discussed in more detail in thessction below entitled "Predictive Values' Applications Of The Method"

The inventors undertook a large and complex analysis involving numerous intellectual choices in arriving at the 8 biomarker signatura In particular, it is important to note that this particular signature has special properties. For example, it is significantly different from a 7 biomarkers signature, which isunsuitabla Analysis of the signature attempting to drop any of the 8 biomarkers disclosed leads to adrop in specificity of approximately 50% and/ or adrop in predictive value of approximately 25%. These figures clearly illustratethat the teaching of an 8 biomarker signature according to the invention is not merely an iterative or arbitrary choice but presents clinically useful information which is a step change from that obtained with even one fewer marker. It is surprising that such a sharp and dramatic effect can be observed in this manner. Suitably the signature comprises 8 biomarkera

Suitably the biomarkers are as set out in thetablebelow.

REFERENCE SEQUENCES

Suitably the reference sequences of the biomarkers of interest are as defined in the following table:

Theakilled worker only has to identify the correct gene/ protein being used in the anaJysia The guidance provided is not intended to restrict the invention rigidly to the specific single exemplary sequences provided. Qene sequences (and therefore protein sequences) are known to vary between individuals ag. due to allelic variance or mutations between individuals. The information provided isto assist the operator in working the invention by assaying the correct gena Ultimately the gene product (such as mRNA or more suitably protein) is actually assayed. Therefore mi nor or minimal allelic or mutational differences between individuals are not important, what is important is that the correct gene (gene product) is assayed using the guidance provided. Suitably where the biomarker name is used, this means the corresponding amino add or nucleic acid sequence from the above tabla Suitably for amino add sequences, the canonical sequence is preferred. Suitably for nucleic add sequences, the most recent (ag. highest numbered) nucleic add sequence is preferred.

It will be understood that the invention may equally make use of detection of fragment(s), variant(s) or mutant(s) of these biomarkers. Suitably any such

fragment(s), variant(s) or mutant(s) have at least 80% sequence identity to the reference sequences along the whole length of said fragment(s), variant(s) or mutant(s), suitably 90%, suitably 95%, suitably 98% sequence identity along the whole length of said fragment(s), variant(s) or mutant(s).

DATABASE RELEASE

Sequences deposited in databases can changeover tima Suitably the current version of sequence database(s) are relied upon. Alternatively, the release in force at the date of filing is relied upon. As the skilled person knows, the accession numbers may be version/ dated accession numbers. The dteable aocess!on numbers for the current database entry are the same as above, but omitting the decimal point and any subsequent digits ag. for VEQF a version/dated accession number is P15692-18; the current entry is obtained using P15682 and so on.

GenBank Is the NIH genetic sequence database, an annotated collection of all pubJidy available DNA sequences (National Center for Biotechnology Information, U.S. National Library of Medidne 8600 Rockville Pike, Bethesda MD, 20894 USA; Nuddc Adds Research, 2013 Jan;41(D1):D36-42) and accession numbers provided relate to this unless otherwise apparent. Suitably the GenBank database release referred to is 15 October 2015, NCBI -GenBank Releaee210.0.

UniProt (Universal Protein Resource) isacomprehensivecatalogueof information on proteins ('UniProt: a hub for protein information' Nuddc Adds Res.43: D204-D212 (2015).). For the avoidance of doubt, UniProt Release 2015_11 is relied upon. In moredetail. theUniProt consortium European E3ioinformaliC8 lnatitute(EE3l), 9B Swiss Instituted Bioinformaticsand Protein Information Resource (PI R)'sUniProt Kno^edgebaee(UniProtKB) Release 2015_11 (11-Nov-2015) is relied upon. TREATMENTS

It should Denoted that treatment for TB is a minimum six month program of drugs. This is expensive and can be very demanding on the patient. Dosage has to be very regular, such as multiple doeas per weak and ideally daily, which is a heavy burden on the healthcare provider as well asthepatient. Therefore, it is a problem in the art to avoid the mistreatment of patients i. a the mis-prescription of TB drugs to a patient who does not in fact haveTB. Thepreeent invention alleviates this problem by providing a robust tool for diagnosis (or for aiding diagnosis). If it is decided in view of the method of the invention that the subject hasTB, then the physician should prescribethetreatment for TB. In oneembodiment the invention provides a method of treating a patient comprising determining if they haveTB according to themethod(s) disclosed herein, wherein if the patient is determined to haveTB then the treatment for TB is prescribed, or more suitably administered. In another embodiment the methods of the invention are used to aid diagnosis of a patient who need not be present during the diagnostic step in the strict sense (i.a in this embodiment the invention is not directed at diagnosis for curative purposes stricto sensu); the physician may then take the information provided by the invention into account when diagnosing and/ or planning the treatment for said subject.

Thus it is an advantage of the invention that unnecessary drugs can be avoided. I n one embodiment, the test of the invention finds application as a rule-in teat rather than a rule-out test. I n other words, if the method(s) of the invention are used to determine that a subject hasTB, they should definitely be treated for TB. In the alternative, if the methods of the invention are used and it is not determined that a subject hasTB, then further investigation may be useful - a subject is suitably not 'ruled out ^* from possibly having TB if the methods of the invention do not determine that they haveTB.

I n terms of a 'rule-in' test, the methods of the invention have a comparable

performance to the gold standard in the art (bacterial culture). Thus it is an advantage of the invention that a positive finding using the methods of the invention provides a very high level of confidence that the subject has TB and should be treated for TB. Suitably TB treatment isae recommended by the World Health Organisation (WHO). The WHO may amend its guidelines from time to time- suitably treatment is as per the guidelines at the date of working the invention to treat a subject. More suitably this treatment is as per the guidelines at thefilingdateof thisdocument. If any further guidance is needed, most suitably treatment is as per the guidelines in the WHO 2010 publication "Guidelines for national programmes, fourth edition" ISBN:

97B92415A7B33 This

document is hereby incorporated herein by reference, apedficaJly for the teaching of the specific treatment regimefor TB.

[Exemplary treatments are set out below.

Suitably TB treatment compriaasthestandard 6-month course of 4 antimicrobial drugs as recommended by the WHO.

Suitably TB treatment comprises 6 months of rifampidn.

Suitably TB treatment comprises six (6) months treatment with specific anti-TB drugs (Isoniazid, Rifampidn, Pyrazinamideamd Ethambutol for first 2 months, followed by 4 months of Isoniazid and Rifampidn only).

Suitably patients with TB may receiveadaJly intensive phase followed by athreetimes weekly conti nuation phase [2H RZE/ 4(H R)3] ; suitably each dose is di rectly observed. Three times weekly dosing through out therapy [2(HRZE)3/4(HR)3] may be used as an alternative, provided that every dose is directly observed and the patient is NOT living with H I V or living in an HI V-prevaJent setting.

Suitably the dosing is not less than 3 times per weak. Suitably the dosing Is3 times per week or mora Most suitably the dosing is daily.

Most suitably the dosing frequency for patients with TB is daily throughout the course of therapy [2H RZE 4HRJ.

Suitably if the patient la living with HIV or living in an HIV-prevatent setting thedoaing is daily throughout the course of therapy. Summary of WHO treatment guidance (WHO 2010 publication "Guidelines for national programmes, fourth edition" page 5):

The standard treatment above is suitably not applied to muti drug resistant TB (MDR TB). Suitably drug susceptibility testing is undertaken before treatment, in accordance with WHO guidelines. If treating a TB patient whose treatment has failed or other patient with high likelihood of multidrug-realetant TB (MDR-TB), suitably treatment should be started on an empirical MDR regimen as recommended by WHO. Moat suitably the invention is applied to 'new ¹ TB, i .a new patients tested according to the invention.

Further or additional treatment may depend on whichever diagnosis is made; usually a course of antibioticsif it is a bacterial respiratory tract infection.

Administration may be by tablet or by injection such as intramuscular injection. For HRZE/HR regimes, suitably administration is by tablet.

Typical tablets comprisethe following doses:

H 75mg + R 150 mg + Z400 mg+ E 275 mg tablets.

Subjects are administered or prescribed a number of tablets according to their body weight (and/ or any other relevant factors if necessary) to achievethecorrect dose. This number may include fractions of a tablet. A typical dose for a subject of 30-39Kg body weight is 2 tablets of H 75mg+ R 150 mg + Z400 mg+ E 275 mg daily during the first 2 months of treatment, followed by 1.5 tablets of H 150 mg + R 150 mg during months 3 to 6 of treatment. Thedetermination of exact dose 04. baaed on body weight is a matter for the physician.

STATISTICAL ANALYSIS

The inventors considered standard approaches to statistical analysis. However, the inventors had insight that standard regression models tend to lead to over optimism. This is particularly true considering them

the large number of cytokine covariates some of which are highly correlated with each other. For various reasons, the inventors used a different approach, and

simultaneously tried to shrink the marker set (i.a to reducethedimensionaJity) whilst applying penal ties for shrinkage in the statistical analysis. Without wishing to be bound by theory, their reasoning was that if a signature assesses two markers which move in thesamedirection. then qualitatively the same information is being obtained from two comparable sources. This provides an opportunity to eliminate one of those markers, thereby simplifying the signature without compromising thequaJity of information obtained. Thecorollary of this is that if two analyses are providing information in two different directions, then this can be seen to provide extra information to improve the signatura In this manner, the inventors sought to remove any markers which might be considered as statistically related to one another, thereby arriving at an improved empirical (reduced) signature but which still delivered excellent diagnostic characteristics. MARKER SELECTION

The inventors undertook unbiased marker selection. Prior art based approaches have tended to use a "case control" approach. In brief, this might becharacterised by settling on TB as a subject to be addressed, selecting healthy patients, selecting patients havi ng TB, and com pari ng the healthy patients with the TB diseased patients. By contrast, thei nventors designed a case-control study nested within a prospective cohort (i.a nested case-control) approach. They selected a cohort of children using the same selection process to identify children in different geographical locations. Only then did they identify within those cohorts patients having TB and patients not having TB. More importantly, theinventorschosetocompareTBtoOD (i.a "other respiratory disease but not TB"). This is because it is a key clinical decision to identify those patients with TB compared to those presenting with other diseases which are not TB. Thus, it can be considered a problem in the art to differentiate TB patients from "other disease" patients.

Another drawback with prior art studies is that they have tended to settle on

biomarkerssuch as cytokines which have been published or associated with TB. By contrast, the inventors took an entirely unbiased approach and did not pick biomarkers by association with TB. They undertook a blind analysis and arrived at the signature of 8 biomarkers without knowing what those individual biomarkers wera Only then did the i nventors analyse the identities of the biomarkers in the! r signatura One example of the surprise of this approach is by consider! ng I FN-γ. The view i n the art is that I FN- Y is involved with TB. I ndeed, prior art approaches have tried to use I FN-γ and or IP- 10 as biomarkers for TB. It is extremely surprising that thebiomarker signature taught herein does not comprise I FN-γ.

Quantiferon ('OFT - available commercially from OJagen Inc.) is an interferon -gamma (IFN-y ) release assay, commonly known as an IQRA, and is a modern alternative to the tuberculin skin test (TST or Mantoux). In overview, OFT measures thecell-mediated immune response (cytokines) to very specific TBantigena The teat is performed by collecting whole Wood (1 mL) into each of three blood collection tubea When the blood of an infected patient is stimulated with the A/, tubercuhsis-spedftc antigens in OFT, their T-Cells respond by secreting acytokine called IFN-y. The IFN-y concentration in the plasma is determined using a sensitive ELISA.

Thus, Quantiferon (OFT) isacommerdaJ assay that still measures I FN-gamma following stimulation of blood with antigens specific for M.tb. Some IP- 10 assays have also bean investigated in the prior art. However, while I P-10 is reportedly more robust than IFN-gamma(i.a releaeed at a much higher level following stimulation), on its own it cannot distinguish between TB disease and TB sensitization similar to IFN-gamma, which is a problem in the art. Advantageously, in thepreeent invention, what istaught is acombination of markers, which when used together.- not individually - can distinguish TB from OD. This is based on the inventors' insight that, given the complexity of TB disease, a combination of markers rather than a single marker has a higher power and specificity to distinguish TB from TB sensitization and/ or other disease.

In more detail, the inventors took an unusual approach by analysing various cytokines and chamokines and turning the vaJues of each into deciles. Starting with 27 cytokines' chamokines as candidates, each was transformed into 10 deciles giving 270 for each patient stimulated and 270 for each patient unstimulated.

This represents a categorical (rather than continuous) approach which is itself a key part of the innovative approach taken. This approach has never before been applied to cytokine analysis. This has advantages which include removing the confounding effect of bias or selection in the biomarkers which are used.

The panel of 27 initial candidatasdid not have any previous association with TB. For example, they were not TB specific. At most, they may be regarded as an "immune panel". They are simply markers involved in lymphocytes as part of a commercial kit which is not in any way marketed or directed towards TB. To illustrate how surprising thefindingswere, even the inventors did not predict what they would find by using this approach.

In one embodiment the analysis may be carried out as follows:

frequency distribution >10 ded leomake each a binary quantila

In one embodiment the analysis may be carried out as follows:

frequency distribution > dedles>10 - equal sized quantiles> use each quantile as cut-off to generate a binary variabla 11 should be noted that each cytoki ne may be present across a different range of concentrations so that each of the deciles may not be the same between individual biomarkers. However, the distribution of each individual biomarker is suitably divided into 10 deciles (i. a 10 equal-sized quantiles) according to its own range of

concentrations when present; each biomarker value determined for a patient is then converted into a decile from that frequency distribution. It should Denoted that the invention isdirected at obtaining a clinical decision whether a subject hasTBor OD.

It isposBibleto use the invention as a screening tool such asapopulation screening test.

It is an advantage of the invention that it is helpful in assisting the dinical decision whether a patient hasTB or OD.

DECILE DETERMINATION

Deciles are generated according to standard statistical techniques. Generation of deciles is a mathematical frequentist procedure that can be derived or generated by any statistical softwara A specific variable (ag. a measured cytokine) with quantitative values when measured from several subjects will have a frequency distribution that is then usBdtogeneratethededlesmadeupof 10 equal-sized quantiles.

I n case any further guidance is needed, moat suitably deciles are generated as described in the examples section below. aiBJECJS

The present invention may be applied to any subject from newborn on war da

The present invention may be applied to adults or children.

Suitably the subject is 18 years or younger, suitably 16 years or younger, suitably 15 years or younger, suitably 7 years or younger, suitably 6 years or younger, suitably 5 years or younger, suitably 2 years or younger.

It should be noted that most immunesystemsfunction as "adult" from 2 to 5 years onwards. Thus, suitably the subject is at least 2 to 5 years old. Suitably the subject is at least 2 years old. Suitably the subject is at least 3 years, suitably at least 4 years, most suitably at least 5 years old.

Suitably the subject is a child. Suitably the subject is 16 years or younger.

Suitably the subject is 15 years or younger.

Suitably the subject is 2 to 16 years old, suitably 3 to 16, suitably 4 to 16, suitably 5 to 16 years old.

Suitably the subject is 2 to 15 years old, suitably 3 to 15, suitably 4 to 15, suitably 5 to 15 years old.

Suitably the subject to be tested has presented with at least one of the following symptoms: coughing, weight loss, sweating, swollen glands, and optionally sepsia

The invention may be applied to intrathoracic TB.

The invention may be applied to pulmonary TB.

The invention may be applied to extra-pulmonary TB.

The invention may be applied to patients which are uninfected with HIV.

The invention may be applied to patients which are infected with H I V.

It should be noted that si nee the method is based on the immune response, that any subject with aCD4 count of 50 or fewer is unlikely to show a response. Thus, suitably the subject has a CD4 count of 51 or greater. For reference, a normal CD4 count in a healthy human isapproximately 1000.

DETECTION

Suitably the biomarkers described herein are detected by the suitable means in the art. For example, the biomarkers may be detected by one or more antibodies which specifically recognise said biomarkers.

For example, the biomarkers may be detected by an antibody or antigen bi ndi ng fragment thereof as described above, wherein the antigen binding fragment thereof is selected from the group consisting of a Fab fragment, a Fab' fragment, a F(ab')2 fragment, aecFv, aFv, arlgO, and adiabody. The mode of aonooolng binding of an antibody or antigen binding fragment thereof to detect the markers is a matter of operator choice. I n case any guidance is needed, ELI SA's could be used for each of the cytokines identified, for example one ELI SA for each biomarker. Mow are examplee of suitable ELI SA reagenta

ELI SA's for the cytokines included in the signature of the invention may be obtained commercially from thefollowing companies, with exemplary product names' details where appropriate:

Quantikine sandwich Elisafrom R&D Systems UK, 19 [Barton Lane, Abingdon Science Park, Abingdon, 0X143NB, United Kingdom (Tel +44 (0)800 3734 15).

Human Platinum Elisaor high sensitivity Elisafrom eBosdence, Ltd. (Ireland, United Kingdom), 2nd ROOT, Titan Court, 3 Bishop Square, Hatfield, AL10 9NA, United Kingdom. BD OptEIA kits from BD Biosciences, Edmund H alley Road, Oxford Science Park, Oxford, OX44DQ (Tel.: +44 1865781666; Fax: +44 1865781627).

Astheskilled worker will be aware, individual ELISAsfor each cytokine might be laborious, and/ or require larger sample sizes. Therefore it is an advantage to carry out thedetection in a multiplex or singlesamplewhereposeibla This provides advantages such as low volume (particularly important for a paediatric test where lower volumes of blood/ serum aredeeirable), and combination of the cytokines (leas labour to complete thetest). There are numerous commercial suppliers of suitable multiplexing kit(s) useful to detect thebiomarkersof the signature, for example:

MILLIPLEX MAP Human Cytokine/ Chemokine Magnetic Bead Panel - Immunology Multiplex Assay, Catalogue number: HCYTOMAG-60K avaJlablefrom Merck-Millipore, Suite 21, Building 6, Croxley Green Business Park, Watford, Hertfordshire WD188YH, United Kingdom.

Human Luminex Performance Assay Base Kit, Panel A [catalogue number LUH000] from R&D Systems UK, 19 Barton Lane, Abingdon Science Park, Abingdon, 0X143NB, United Kingdom (Tel: +44 (0) 800 3734 15). Human Cytokine/ Chemokine/ Growth Factor Panel 1 (45 pi ex), (Catalogue number: EPX450- 12171-901) from eBiosdence, Ltd., (Ireland, United Kingdom), 2nd Floor, Titan Court, 3 Bishop Square, Hatfield, ALIO 9NA, United Kingdom. Moat suitably the antibodies used may be as In the commercially available E3io- Rad Human cytokine Th-VTh-227-plex kit (catalogue number *M500KCAF0Y from Bio- Rad Laboratories Ltd., Bio- Rad House, Maxted Road, Hem el Hempstead,

Hertfordshire, HP27DX, United Kingdom). Most suitably detection of the cytokines of the signature of the invention are detected/ assayed using this kit.

It is important to note that the prevailing view in the art is that stimulation of cells is required for use for analysia The stimulation may be by presentation with bacteria, or may be by presentation with antigen or any other appropriate form of stimulation. However, it should be noted that these types of stimulations are all directed at analysing recall responses. It is an advantage of the invention that unstimulated samples are analysed. In particular, it is an advantage that the sample is from unstimulated blood.

In particular when studying the key panel of 8 biomarkers, suitably the invention omits astimulation step; suitably the invention does not compriseastimulation step; suitably the invention excludes a stimulation step.

In some embodiments a ninth or further marker may be employed - astimulation step may be employed for such further marker(s) if appropriata

Most suitably the invention omits a stimulation step. Most suitably the invention does not compriseastimulation step. Most suitably the invention exdudesastimulation step.

Nucleic acid detection

For example, the biomarkers may be detected in nucleic add form, for example by detection of one or moremRNAsenoodingthebiomarkers.

If theskilled worker deal res to read-out/ detect nucleic adds via a microarray approach, reference is made to Anderson et aJ 2014 (N Engl J Med.2014 May 1;370(18):1712-23 'Diagnosis of childhood tuberculosis and host RNA expression in Africa.' I LULU Consortium; Kl DSTB Study Group.) mRNA technologies are suitably deployed in a laboratory setting. In outline, mRNA detection may comprise the following steps:

• stabilisation of RNA- can bedonewith a specific reagent,

• extracted of RNA,

• transcription to cDNA,

· amplification,

• array,

• data read out.

Of course the person skilled in the art will realise that some steps are optional or may be combined, for example stabilisation/ extraction may not be required if transcription can be performed di rectly on the sampla For example, array may not be requi red if the amplified material is assayed directly.

Thus in essence the required steps are:

• extraction of nucleic add

• assay of nucleic add to determine mRNA expression level of markers of interest

• data read out.

For multiplex detection, suitably afluorogenicoligonudeotideprabethat is specific to the target gene/ amplified target is used. Taqman probes are commonly used for multiplexing, but can also be used if multiplexing not required. Protocols are as stated by the manufacturer ag. Applied Biosystems (5791 Van Allen Way, Carlsbad, California 92008, US).

Different dyes can be used for thefluorogenic probes; examples that may be useful depending on the buffer conditions and type of thermal cyder are shown in the table below (Table from Qjagen Inc):

Protocols are well known in the art, for example using the StepOne and/ or StepOne Plus Real Time PCR systems according to manufacturer's instructions (Applied Biosystems(5791 Van Allen Way, Carlsbad, California 92008, US).

Other assays may be used if multiplexing was not desired, for example after reverse transcription, using dyes that bind double-stranded DNA and become florescent. For example SYBR green 1 (Qiagen IncJQiagen Ltd. Skelton House, Lloyd Street North, Manchester M156SH, UK) may boused.

Primer probe assays useful in the invention can be designed, or purchased predesigned, to the gene target of interest i.a the biomarkers of the invention. For example, Applied ESioeysterns Thermo sher Stientif 10 ^* 8(5791 Van Allen Way, Carlsbad, California 92008, US) protocol for SYBR green 1 custom design ("Design and optimization of SYBR Green assays") may be used, including the publicly available primer design tods discussed therein. This document is hereby incorporated herein by reference specifically for the primer/ probe design protocols and nucleic add detection teachings. I n case any further guidance is required, reference is made to the examples section below.

DETERMINATION OF QUANTI LESV DEO LES

Suitably converting each biomarker concentration determined into a decile value comprises the steps of: (d) comparing the concentration of each biomarker determined to a reference frequency distribution of concentrations of said biomarker; and

(di) reading out the decile value from thefrequency distribution for the

concentration of said biomarker.

A "frequency distribution" shows a summarized grouping of data divided into mutually exclusive classes and the number of occurrences in a class. So it is possible to prepare a frequency distribution even with small numbers of data points such as 30 (ag. exam scores of 30 children in a class). The larger the number of subjects used (i.a data points), the more representative the distribution will be of the true population La the more normally distributed thefrequency distribution will be. Etiological variables ag. weight, height, blood pressure, Haemoglobin concentration, electrolytes in blood, cytokine measurements eta are usually skewed. Thus, to get a normally distributed curve for such biological variables using a normal histogram for exampla very large numbers of data points are needed which may be problematic. Therefore, non- parametric methods such as Kernel, which isadata-smoothing density estimator, are used. This is especially helpful when we want to present the representation of distribution of such data with a reasonable good sample size ag. of 100 or mora Therefore, when frequency distribution' is mentioned herein, suitably this may comprise a non-parametric equivalent such as a non-parametric density estimator, such as a data-smoothing density estimator, most suitably a Kernel density estimator.

Considering Figures 2 to 9, the Y-axes are appropriately labelled 'Density' because they indicate the Kernel Density Estimata

In moredetaJI, the Kernel frequency distribution isadata-smoothing statistical approach to displaying frequency distribution. Unlike the ordinary 'histogram', it is a non-parametric method which makes it very appropriate for our type of data, which like most biological data does not follow the normal Gaussian distribution. The basic histogram displays frequency in number or proportion and the problems with histograms include the fact that it is not smooth and depends on the width of the bins (i.a the bars) and end point of the bins which can be arbitrarily chosen. However, Kernel density esti mate removes the dependence on end poi nt of the bi ns by assumi ng no gap in the interval of cytokine measurements within a specified range, giving a smooth density estimata The Y-axis of Figures 2 to 9 can simply be summarised as:

"theprobability density function of each respective marker." Suitably the reference frequency distribution (or Kernel Density Estimate) is generated by measuring the concentration of thebiomarker in a number of subjects, for example a minimum of 100 subjects, and compiling those measurements into a frequency distribution (or Kernel Density [Estimate).

Alternatively the frequency distributions (Kernel Density Estimates) presented in Figures 2 to 9 herein may be used. Suitably step (c) converting each concentration determined in (b) into a decile value comprises the steps of:

(d) comparing the concentration of each biomarker determined in (b) to the corresponding reference frequency distribution or Kernel Density Estimate of concentrations of said biomarker selected from Figures 2 to 9; and

(cii) reading out thededlevaluefrom the frequency distribution or Kernel Density (Estimate for the concentration of said biomarker.

In one embodiment, decile/ quantilecutoffs may be augmented or replaced by absolute cut-offs expressed as absolute concentrations of the biomarker (s) in thesampla Exemplary absolutecut-off values are provided in the table below:

In one embodiment, decile/ quantilecutoffs may be augmented or replaced by the concentration range for the specific quantlle of interest expressed as the range of absolute concentrations of thebiomarker(s) in thesampla Exemplary concentration ranges are provided in the table above. POINT OF CARE TEST

In many embodiments, theskilled operator may choose to analyse the concentrations of the markers in a laboratory or teat facility.

The invention mayalso be applied asapoint of caretest. Suitably the invention may also be applied as a bedside teat.

When the invention is a point of care/ bedside test, suitably the sample is blood or plasma When the invention is a point of care/ bedside test, suitably the markers are analysed in protein form.

When the invention is a point of care/ bedside test, suitably thedetection is

immunological detection.

When the invention is a point of care/ bedside test, suitably the test is the a format of a lateral flow assay. PREDICTIVE VALUED APPLICATIONS OF THE ΜΕΞΤΗΟΡ

The table below provides additional results of predictive vaJueeof thebioaignatura We show and compare the PPV and NPV of thebiosignatureto prior art methods. We further show it has demonstrable comparable performance to the gold standard of cultura The high specificity and the positive predictive value of the invention lends itself to change treatment dedsiona This enables accurate prescription for TB detected according to the invention. This also avoids waste of resources in prescription of unnecessary drugs. This further demonstrates the utility of the invention.

FURTHER APPLICATIONS

The ke set of 8 biomarkers are advantageously assooaod from unstimulated samples. However, in some embodiments it may be helpful to further assess a ninth or further marker; suitably such a ninth or further marker comprisssastimulated marker such as EC-stimulated VEQF. Suitably the cutoff for EC-stimulated VEGFisdedle2. For the avoidance of doubt, details of the 'VEQF biomarker of EC-stimulated VEQF are as above in the key panel of 8 biomarkers of the invention ("VEQF).

A method for aiding thediagnosisof TB in a subject, the method comprising;

(a) providing a sample from said subject, said sample being selected from the group consisting of: blood, serum and plasma;

(b) determining the concentration in said sample of the following biomarkers: I L-1ra, I L6, 1 L-7, 1 L-8, 1 L-12p70, FGF-basic I P-10, and

VEQF;

(c) converting each biomarker concentration determined in (b) into adedlevaJue; and

(d) ccfiverting each decilevalueinto abinary presence or absence by comparing the dedlevaJuee of (c) to thefollowing specif icquantilecut- off values:

off value is converted into the binary presence of the biomarker, and a decile value lower than thespscificquantilecut-off vaJueisconverted intothebinary absence of the biomarker;

wherein detecting the presence of each of said biomarkers indicates an increased likelihood that the subject hasTB.

A method for differentiating TB from CO in a subject, the method comprising; carrying out steps (a) to (d) above

wherein detecting the presence of each of said biomarkers indicates that the subject has TB.

A method of collecting information useful in diagnosis of TB in a subject, the method comprising:

carrying out steps (a) to (d) above

wherein detecting the presence of each of said biomarkers identifies the subject as having TB.

A method for thediagnosisof TB in a subject, the method comprising;

carrying out steps (a) to (d) above

wherein detecting the presence of each of said biomarkers provides thediagnoais that the subject hasTB.

A method for selecting a subject to receive treatment for TB, the method comprising; carrying out steps (a) to (d) above

wherein detecting the presence of each of said biomarkers selects said subject to receive said treatment.

A method comprising the steps of selecting a subject to receive treatment for TB by; carrying out steps (a) to (d) above

wherein detecting the presence of each of said biomarkers selects said subject to receive said treatment; and administering said treatment to said subject.

A method of treating TB in a subject comprising administering a regimen of

2HRZE/4HR (2 months H RZE followed by 4 months H R wherein H - isoniazJd, R - rifampidn, Z- pyracdnamide, E - ethambutol) to a subject determined to have each of thefollowing biomarkers: IL-1ra. IL6, IL-7, IL-8, IL-t2p70, FGF-basic, IP- 10, and

VEGF. The invention also relates to said method further comprising testing the subject prior to the administering step to determinethat the subject has thefollowing biomarkers: IL-1ra, IL6, IL-7, IL-8, IL-12p70, FGF-basic IP-10, and VEGF. Suitably testing is carried out by carrying out steps (a) to (d) above.

In so far as the embodiments of the invention described above are implemented, at least in part, using software-controlled data processing apparatus, it will be appreciated that a computer program providing such software control, and a storage medium by which such a computer program is stored, are envisaged as aspects of the present invention. Clearly in several of the methods or processes of the invention, one step (typically step (a)) comprises providing a sample from the subject - clearly that step would not typically be performed using software-controlled data processing apparatus; suitably that step is manually executed, or omitted, in embodiments implemented using software-controlled data processing apparatus

Thus the invention relates to an apparatus such as a computer comprising logic, circuitry or code configured to carry out the method as described above.

Thus the i nvention relates to a computer program product operable, when executed on a computer, to perform the method as described above.

Further particular and preferred aspects are set out in the accompanying independent and dependent claJ ma Features of the dependent claims may be combi ned with features of the independent claims as appropriate, and in combinations other than those explicitly set out in thedaJma

Where an apparatus feature is described as being operable to provideafunction, it will be appreciated that this includes an apparatusfeature which provides that function or which is adapted or configured to provide that function.

BRIEF DESCRIPTION OF THE DRAWINGS

Embodiments of the present invention will now be described byway of example, with reference to the accompanyi ng drawi nge, i n which :

Figure 1 shows optimal LASSO modelswith cytokine covariates, adjusted for age and origin. Panels a, b, c: Optimal LASSO model, adjusted for age and origin, determined by

5-fold cross validation in training set. Panels d, e, f : box-and-whisker plot showing probability of TB disease in thebacteriologicaJly-confirmed TB, clinically diagnosed TB and OD subjects in the training set as predicted by the identified biosignatura Panels g, h, i: All C showing discriminating ability of identified biosignatureto classify confirmed

TBfrom OD in the independent test set

Figure 2 shows a frequency distribution for IL-1ra

Figure 3 shows a frequency distribution for IL-6

Figure 4 shows a frequency distribution for IL-7

Rgure5ahowsafrequencydistribution for IL-8

Figure 6 shows a frequency distribution for IL-12p70

Figure 7 shows a frequency distribution for FQF-baaic

Figure 8 shows a frequency distribution for IP- 10

Figure 9 shows a frequency distribution for VEGF

Figure 10 shows a lateral flow device.

Figure 11 shows the multiplex detection format of a lateral flow device. EXAMPLES

Methods

Brief cohort description

Children aged less than IS yean who were exposed to an adult infectious TB case in the household setting were actively traced and screened for symptoms suggestive of

TB disease in the respective households. Those with suspected intrathoracic TB disease thereafter had further detailed clinical evaluation and investigations to ascertain their TB disease status. A total of 173 child TB contacts with suspected intrathoracic TB disease, prospectively recruited both in The Gambia (n=150) and

United Kingdom (n=23), were included in the biosignature discovery experiments using an immuno-epidemiological approach.

Whole blood gtimulatioH assay (WBA)

For the Gambia cohort, a WBA was set up at the recruitment within four hours of venepuncture. 100μ 1 of undiluted heparinised whole blood was incubated in duplicates with M.tb antigens ESAT-6VCFP-10 fusion protein (EC; lOug/ml final concentration; kindly provided by Professor Tom OttenhofF, Leiden University Medical Center, The Netherlands) and positive (PHA-L, Sigma Chemicals, UK; 10μ g/ml final concentration) and negative (RPMI 1640 medium; BioWittaker, Verviers, Belgium) controls. After overnight incubation at 37°C with 5% C0 _¼ supernatants were harvested, duplicates pooled and stored at -20°C prior to analysis. Samples were added to mis cohort from an equivalent set up of a household contact study in the UK, conducted by BK. For the children from the UK cohort, we had also obtained the relevant demographic and clinical data and derived supernatants from IGRA (Quantiferon-TB Gold In-Tube (QFT-G) test (CeUestis, Australia). Similar to our in-house assay, mis commercially available in-vitro IFN-γ release assay uses stimulation of fresh whole blood in three separate tubes containing Mtb antigens (ESAT-6, CFP10 and TB 7.7), positive (Phytoheamagiiitinin-L) and negative (Nil) controls respectively. These samples were shipped frozen to The Gambia for joint analysis.

The WBA supernatants from the Gambian children and the QFT supematants from me children from the UK cohort were used for a multiplex cytokine detection assay (MCA). The MCA was carried cut on site in the MRC TB Immunology laboratory in The Gambia, with the Gambian and UK samples randomly distributed over the multiplex plates.

Multiplex cytokine detection assay (MCA)

We carried out a comprehensive MCA by Luminex using the unstimulated and EC- stimulated WBA supematants of the Gambian children and QFT supernatants (from antigen and nil QFT tubes) of the children from the UK cohort Culture supernatants were analysed using the Bio- Rad Human cytokine Th-l/Th-2 27-plex kit according to the manufacturer's instruction and as described previously [1]. Cytokines assessed were:

1(MCAFX MUMa, MIP-lb, PDGF-bb, RANTES, TNFa and VEGF. Following pre-wetting of the filter plate, 50ul of bead suspension was added to each well and washed twice. 50ul of samples and standards, tested singly and in duplicates respectively, were then added and the plate sealed and shaken for 30 seconds at llOOrpm and then incubated for one hour at 300rpm. The plate was washed three times, 25 μ 1 of pre-diluted detection antibody was added and the plate shaken and incubated for 30 minutes at 300rpm in the dark. After washing, 50μ 1 of 1 x streptavidin-PE was added to each well and incubated for 10 minutes with shaking at 300rpm. The plate was again washed and resuspended in 125μ 1 of the assay buffer, sealed, mixed and immediately read on the Bio-plex analyser using Bioplex manager software (version 4.0; Bio-Rad, USA) and a low photomultiplier tube (PMT) setting. Cytokine concentrations below the level of detection - repotted as 'OOR' in the Bioplex software - were calculated as zero in the analysis.

Statia tical analysis

We analysed the data obtained from the multiplex cytokine assay (MCA) of unstimulated and EC-stimulated whole blood culture supernatants for the identification of the host-specific multi cytokine bioeignature associated with TB in children. For this analyses, the unstimulated and antigen-specific cytokine responses from the 27-plex MCA were analysed as separate variables. We randomly assigned the study subjects into a training set (80% of subjects) and an independent test set (20%). We men used Generalized Linear Model (GLM) applying Least Absolute Shrinkage and Selection Operator (LASSO) penalty to fit logistic regression models in the training set adjusting for age in years and origin of sample, initially with bacteriologically confirmed TB (gold standard) compared to other respiratory diseases mimicking TB but not TB (OD group) as the binary outcome variable. The LASSO model applies a maximum penalised likelihood to the absolute size of the regression coefficients, shrinking mem towards zero i.e. an LI norm penalty is applied to the regression coefficients. This procedure results in both variable selection (some regression coefficients equal zero) and estimates of non-zero regression coefficients shrunk towards zero. This methodology is suitable to our cytokine data in which there were very many measures, many of which could potentially be highly correlated. We fitted the cytokiiie covariates as categorical and continuous variables and as a combination of both. Categorical cytokine covariates were constructed by a split of the cytokine values into deciles by dividing the frequency distribution of each cytokine value into 10 equal-sized quantiles, which were then fitted into the model as 10 binary variables for each cytokine using each of 10 quantiles as a cut off. The optimal LASSO model was determined using a 5-fold cross-validation in the training set, which was subsequently applied to classify bacteriologically confirmed TB from OD in the independent test set naive of origin. The optimal model was defined as the model with the highest penalty parameter ('lambda) resulting in the smallest prediction error and the best mean cross-validation AUC. The cross- validation process accounts for, and replaces the classical method of adjusting for multiple testing. In addition, it naturally protects against over-fitting and it is a way of assessing how a model will generalise to an independent dataset The prediction performance of the optimal LASSO model was evaluated by estimating prediction probabilities for TB and area under the receiver operating characteristics curves (AUC).

Results

Overall, S3 children from the combined cohorts were diagnosed with TB and started on standard TB treatment; 24 had bactcriologicaUy-anfinned TB and 29 had TB diagnosed based on clinical and radiological features with no positive microbiological tests. One hundred and twenty were diagnosed and treated for OD. In detail, thirty-five of the ISO Gambian children had TB, comprising 16 bacteriologicaUy-canfirmed and 19 clinical diagnosed TB cases, while 115 had OD.

Of the 23 children from the UK, 18 were diagnosed with TB (8 confirmed and 10 clinically diagnosed TB) while S had OD. None of the children recruited from The Gambia or UK was HIV infected. Table 1 shows mat the distributioa of the baseline profiles of the children from Gambia and UK were comparable.

Pattern of cytokine and chemokine production in confirmed TB vs OD

The concentrations of cytokines and chemokines obtained by a 27-plex multiplex cytokine analysis of unstimulated supematants and EC-stimulated supematants from children with bacteriologically confirmed TB were compared with the levels in children with OD by multi van able linear regression analyses, adjusting for age in years and origin. The unstimulated and EC-specific values (i.e. EC-stimulated minus unstimulated negative control values) for each cytokine or chemokine were analysed as separate variables.

As shown in Table 2, of the 27 cytokines and chemokines analysed, the unstimulated concentrations of ILlra, IL7, IL12p70, IL13, IL15, IL17, Eotaxin, basic FGF, GCSF,

IFN-Y, IP10, MiPlb and VECT were significantly higher in children with confinned

TB compared to children with OD. Of the EC-specific concentrations of all the analytes, only TL2 and IL7 were significantly different between the two groups. Furthermore, we found that the concentrations in unstimulated samples were significantly higher for all the analytes in children from the UK compared to Gambian children regardless of age or diagnosis (p-value <0.001 for all), with the exception of unstimulated-IL7 value in which there was no significant difference (data not shown). On the contrary, the EC- specific concentrations of the analytics were significantly lower in children from the UK compared to Gambian children regardless of age and diagnosis (p-value <0.001 in all), with the exception of EC-specific concentrations of IL2, ILS, JL7, IL13, IFN-γ and RANTES in which there were no significant differences.

Identification of a host specific biosignature for the diagnosis of childhood TB

Using GLM with LASSO penalty to fit a binary logistic regression model, adjusting for age in years and origin and with 5 -fold cross-validation in the training set, a combination of nine categorical cytokines optimally predicted TB or OD with a mean cross- validation AUC of 0.82. Each of the nine cytokines is a binary variable with a cut-off value.

The cytokines - with the specific quantile cut-off value for each in bracket - were: unstimulated IL-lra (3), IL6 (6), IL-7 (8), IL-8 (9X IL-12p70 (9), FGF-basic (3), HMO (4), VEGF (9) and EC-stimulated VEGF (2). When applied to the independent test set, this model reliably classified confirmed TB from OD with an AUC of 0.91(95% CI 0.80 -1.0) as depicted graphically in Figure 1.

Diagnostic accuracy and added value of die muMcytoUme biosignature

The quantile specific cut-off values for each cytokine in the biosignature enabled us to convert the biosignature into a binary - positive or negative - test We investigated the diagnostic accuracy of the biosignature as two separate variables i.e. "biosignature 1" (combination of all the 9 identified cytokines) and "biosignature 2" (combination of only the 8 cytokines from unstimulated supernatants). When we compared the results of the novel multi cytokine biosignatures to the disease certainty classification (i.e. bacteriologicaUy-confinned TB (= confirmed TB), clinically diagnosed TB (= probable TB) and other respiratory diseases but not-TB (OD) as defined by the WHO [2], 'biosignature! ' was positive in 8% of confirmed TB and 7% of clinically diagnosed TB cases. However 'biosignature2' had a relatively higher sensitivity with positive results in 21% and 24% of confirmed and clinically diagnosed TO cases respectively. Bom biosignature versions were negative in 119 of 120 OD cases giving a very high specificity of 99.2% (Table 3).

Using a composite reference standard of all children diagnosed with active TB disease, 'biosignaturel ' derived from only markers in unstimulated supematants was positive in 12 of all the S3 children diagnosed with active TB disease giving a sensitivity of 23% (95% CI 12 -36). As individual tests, the sensitivity of

'biosignaturel' was significantly higher than that of smear microscopy (5.7%; p- value = 0.020) but comparable to that of Μ.Λ culture (35.9%; p-value = 0.127). The combination of 'biosignaturel' and smear microscopy were positive in 15 of 53 children with active TO disease giving a sensitivity of 28.3%, which was significantly higher than the sensitivity of smear microscopy alone (5.7%; p <0.001). Similarly, 'biosignaturel' combined with MA culture had a sensitivity of 49.1%, which was significantly higher than the sensitivity of MA culture used alone (35.9%; pO.001). The sensitivity of 'biosignaturel' combined with MA culture was significantly higher than that of 'biosignaturel ' combined with microscopy (p < 0.001), but comparable to the sensitivity of the combination of 'biosignaturel' _t microscopy and MA culture (p =0.320). The use of 'biosignature2' in combination with these diagnostic tests did not result in any change in the specificity of the tests.

Summary

Since host immune factors such as IFN-γ have been shown to be important, but insufficient to confirm or exclude TB [3, 4], the aim of this study was to investigate cytokines other than IFN-γ that may help to differentiate TB from OD in Gambian and UK children, since the distinction of these two clinical presentations is important to initiate the right therapy. Previous studies have reported other cytokines such as TNF-α, IL-12(p40), JL-6, IL-10, IL-18 and IL-17, FGF and VEGF that have been found to be important in the immune response against M.tb and/or in distinguishing TB from OD [1, 5, 6].

We identified a unique, quantUe-specific, 9-cytokine biosignature that optimally distinguished bacteridogically-confirmed TB from OD irrespective of age and origin of the children. The biosignature also predicted the probability of active TB disease in children with clinically-diagnosed TB that was comparable to that of bacteriologically- confirmed TB cases. Specifically, we used an age and origin adjusted LASSO regression model to identify a quantile-specific combination of unstimulated IL-lra, IL- 6, IL-7, IL-8, IL-12p70, basic FGF, IP-10, VEGF and EC-stimulated VEGF that optimally (fistinguished between bactenologicaUy-conflrmed TB and OD with an AUC of 0.91 in an independent test set. The performance of mis biosignature was regardless of sensitization to M.tb in the clinical outcome groups, while eight of the 9-cytokines in the biosignature were from unstimulated supematants. A major strength of our study is the prospective approach used in an exclusively paediatric active case finding study setting while the identified biosignature in our cohort contains cytokines known to be closely associated with TB immunity. We investigated separately the diagnostic accuracy of the 8 -cytokines bioagnature comprising only the markers from unstimulated supematants, and our full 9-cytokine biosignature by comparing their results to disease certainty classifications according to WHO case definitions and a composite reference standard of all children diagnosed with active TB disease. We found that while the unstimulated 8-cytokine biosignature had a relatively higher sensitivity than the full 9-cytokine biosignature, both biosignature versions distinguished active TB disease from OD with a very high specificity of 99.2%. The unstimulated 8-cytokine biosignature detected a comparable number of TB cases among all children diagnosed with active TB disease in our study relative to M.tb culture, and demonstrated substantial added value when combined with routine TB diagnostic tests, It showed a comparably higher specificity but a lower sensitivity relative to a risk score based on a 51 -whole blood gene transcript that was identified in a multi-country childhood TB biomarker study in south and eastern Africa, as well as a three marker combination of TNF-α, IL-12(p40) and IL-17 in antigen stimulated whole blood supematants of adult Gambians [1, 7]. However, the specificity of this paediatric biosignature is comparable to mat of the combination of IL-13, FGF and IFN-γ in ex vivo sputum samples in another study in adults in The Gambia, which resulted in 96% correct classification of consecutively recruited cdture-confirmed TB cases from OD with a sensitivity of 85% and specificity of 96% [6].

A number of factors makes this unstimulated multicytokine biosignature a particularly promising approach for use in high TB burden countries. First, the quantile-specific cutoff values for each of the component cytokines mean die readouts can be easily converted into a binary test with either a positive or negative result, which makes it more easily interpretable. Secondly, this multicytokine biosignature, derived from only markers in unstimulated supematants, had similar epidemiological properties to M.tb culture but is potentially not subject to the same time delay or risk of contamination as culture. Thirdly, it has a demonstrable potential to reduce presumptive treatment of TB disease in children in primary care settings in developing countries where TB diagnosis is mostly based on the use of smear microscopy, as well as at referral centres with X-ray, Xpert and/or culture facilities. This is because of the substantial increases in the number of children who would be deemed to have active TB disease when used in combination with the routine diagnostic tests. Fourthly, this unstimulated multicytokine biosignature could potentially be measured directly in serum or plasma samples without the additional cost, training or infrastructure needed for antigen stimulation and incubation in the laboratory. Thus the 8-cytokine biosignature is the most preferred embodiment of the invention.

Example 2: Nudek add baaed detection In this example we describe processing of samples to illustrate the application of the invention/gene signature in aiding diagnosis of TB (such as childhood TB)

via nucleic acid detection, such as RNA (mRNA) detection.

1. Sample collection:

The blood sample is collected into a tube containing a stabilising agent for RNA, such as a PaxGene tube or tempus tube.

Alternatively trizol is added to sample.

Such sample may be stored in fridge or freezer until processing.

If stored in the fridge, the storage time is days, if stored in the freezer the storage time can be months. 2. Sample processing

Depending on the collection tube and stabilising agent, suitable commercially available RNA extraction kit(s) are selected and used according to the manufacturer's instructions.

In mis example, Qiagen kits containing spin∞lurnns and wash buffers for RNA extraction are used.

Total RNA including micro RNA is extracted using these established methods.

3. Reverse transcription

In order to obtain DNA which can be amplified, a transcription reaction needs to first convert mRNA to cDNA. This is done by addition of RT random primers, dNTPs, Reverse Transcriptase and RT buffers in the appropriate amounts, followed by thermal cycling incubation as is known in the art

In this way, the RNA is reverse transcribed into DNA.

4. Amplification

cDNA can then be amplified using primers and/or probes specific for the transcripts of interest of the biomarkers as described above, suitably together with primers and/or probes specific for reference genes for normalisation.

For convenience, in this example primers and/or probes for each cytokine in question, internal controls and endogenous control genes are added to a mastermix along with the cDNA template, and triplicate PCR reactions in either single-plex or multi-plex are carried out using a real-time PCR instrument

References:

1. Sutfaeriand JS, de Jong BC, Jeffiies DJ, Adetifa IM, Ota MO. Production of TNF- alpha, IL-12(p40) and IL-17 can discriminate between active TO disease and latent infection in a West African cohort. PloS one 2010: 5(8): el236S.

2. W JLO. Definitions and reporting framework for Tuberculosis - 2013 revision., Geneva, Switzerland.

www.who.intfris/bitstn^ [Accessed 03

December 2014], 2013.

3. Kairfmann SH. Fact and fiction in tuberculosis vaccine research: 10 years later. The Lancet infectious diseases 2011 : 11(8): 633-640.

4. Flynn JL, Chan J, Triebdd KJ, Dalton DK, Stewart TA, Bloom BR. An essential role for interferon gamma in resistance to Mycobacterium tuberculosis infection. The Journal of experimental medicine 1993: 178(6): 2249-2254.

5. Algood HM, Chan J, Flynn JL. Chemokines and tuberculosis. Cytokine Growth Factor Rev 2003: 14(6): 467-477.

6. Ota MO, Mendy JF, Donkor S, Togun T, Daramy M, Gomez MP, Chegou NN, Sillah AK, Owolabi O, Kampmann B, Walzl G, Sutherland JS. Rapid diagnosis of tuberculosis using ex vivo host biomarkers in sputum. The European respiratory journal : official journal of the European Society for Clinical Respiratory Physiology 2014: 44(1): 254-257.

7. Anderson ST, Kaforou M, Brent AJ, Wright VJ, Banwdl CM, Chagaluka G, Crampin AC, Dockrell HM, French N, Hamilton MS, Hibberd ML, Kern F, Langford PR, Ling L, MlothaR, Ottenhoff TH, Pienaar S, Pillay V, Scott J A, TwahirH,

Wilkinson RJ, Coin LI, Heyderman RS, Levin M, Hey B. Diagnosis of childhood tuberculosis and host RNA expression in Africa. The New England journal of medicine 2014: 370(18): 1712-1723.