Patent Application
20130331283
This application claims the benefit of UK application 1017520.6 (filed 15 Oct. 2010), the complete contents of which are hereby incorporated herein by reference for all purposes.
The invention relates to biomarkers useful in diagnosis, monitoring and/or treatment of lupus.
Systemic lupus erythematosus (SLE) or lupus is a chronic autoimmune disease that can affect the joints and almost every major organ in the body, including heart, kidneys, skin, lungs, blood vessels, liver, and the nervous system. As in other autoimmune diseases, the body's immune system attacks the body's own tissues and organs, leading to inflammation. A person's risk to develop lupus appears to be determined mainly by genetic factors, but environmental factors, such as infection or stress may trigger the onset of the disease. The course of lupus varies, and is often characterised by alternating periods of flares, i.e. increased disease activity, and periods of remission. Subjects with lupus may develop a variety of conditions such as lupus nephritis, musculoskeletal complications, haematological disorders and cardiac inflammation.
Lupus occurs approximately 10 times more frequently in women than in men. It is part of a family of closely related disorders known as the connective tissue diseases which also includes rheumatoid arthritis (RA), polymyositis-dermatomyositis (PM-DM), systemic sclerosis (SSc or scleroderma), Sjogren's syndrome (SS) and various forms of vasculitis. These diseases share a number of clinical symptoms and abnormalities. Subjects suffering from lupus can present with a variety of diverse symptoms, many of which occur in other connective tissue diseases, fibromalgia, dermatomyositis or haematological conditions such as idiopathic thrombocytopenic purpura. Diagnosis can therefore be challenging.
It takes on average 4 years to obtain a correct diagnosis for lupus, in part due to the range and complexity of symptoms and the necessity to discount other possible causes. The American College of Rheumatologists has established eleven criteria to assist in the diagnosis of lupus for the inclusion of patients in clinical trials and developed the SLE Disease Activity Index (SLEDAI) to assess lupus activity. In addition to considering medical history, the subject's age and gender and a physical examination, a number of laboratory tests are also available to assist in diagnosis. These include tests for the presence of antinuclear antibodies (ANA) and tests for other auto-antibodies such as anti-DNA, anti-Sm, anti-RNP, anti-Ro (SSA), anti-Lb (SSB) and anti-cardiolipin antibodies. Other diagnostic tools include tests for serum complement levels, urine analysis, and biopsies of an affected organ. Some of these criteria are very specific for lupus but have poor sensitivity, but none of these tests provides a definitive diagnosis and so the results of multiple differing tests must be integrated to enable a clinical judgement by an expert. For example, a positive ANA test can occur due to infections or rheumatic diseases, and even healthy people without lupus can test positive. The ANA test has high sensitivity (93%) but low specificity (57%) [1]. Antibodies to double-stranded DNA and/or nucleosomes were associated with lupus over 50 years ago and active lupus is generally associated with IgG. The sensitivity and specificity of the Farr test for anti-DNA is 78.8% and 90.9%, respectively [2]. Thus it is clear that the status of multiple autoantibody species can provide information on the lupus status of a patient but to date these clinical analyses are performed individually in a piecemeal fashion. The necessity for a unified test offering both high sensitivity and specificity for lupus is clear.
Many autoantibody species have been described in connection with lupus [3] and their cognate antigens include numerous classes of proteins, subcellular organs such as the nucleus and non-protein species such as phospholipid and DNA. Frequently the antigen is either poorly described or uncharacterised at the molecular level e.g. antimitochondrial antibodies. Given the challenges in obtaining a correct diagnosis, there is a need for new or improved in vitro tests with better specificity and sensitivity to enable non-invasive diagnosis of lupus. Such tests can be based on biomarkers that can be used in methods of diagnosing lupus, for the early detection of lupus, subclinical or presymptomatic lupus or a predisposition to lupus, or for monitoring the progression of lupus or the likelihood to transition from remission to flare or vice versa, or the efficacy of a therapeutic treatment thereof. Such improved diagnostic methods would provide significant clinical benefit by enabling earlier active management of lupus while reducing unnecessary intervention caused by mis-diagnosis. It is an object of the invention to meet these needs.
The invention is based on the identification of correlations between lupus and the level of auto-antibodies against certain auto-antigens. The inventors have identified antigens for which the level of auto-antibodies can be used to indicate that a subject has lupus. Auto-antibodies against these antigens are present at significantly different levels in subjects with lupus and without lupus and so the auto-antibodies and their antigens function as biomarkers of lupus. Detection of the biomarkers in a subject sample can thus be used to improve the diagnosis, prognosis and monitoring of lupus. Advantageously, the invention can be used to distinguish between lupus and other autoimmune diseases, particularly other connective tissue diseases such as rheumatoid arthritis (RA), polymyositis-dermatomyositis (PM-DM), systemic sclerosis (SSc or scleroderma), Sjogren's syndrome and vasculitis where inflammation and similar symptoms are common.
The inventors have identified 50 such biomarkers and the invention uses at least one of these to assist in the diagnosis of lupus by measuring level(s) of auto-antibodies against the antigen(s) and/or the level(s) of the antigen(s) themselves. The biomarker can be (i) auto-antibody which binds to an antigen in Table 1 and/or (ii) an antigen in Table 1, but is preferably the former.
The invention thus provides a method for analysing a subject sample, comprising a step of determining the level of a Table 1 biomarker in the sample, wherein the level of the biomarker provides a diagnostic indicator of whether the subject has lupus.
Analysis of a single Table 1 biomarker can be performed, and detection of the auto-antibody/antigen can provide a useful diagnostic indicator for lupus even without considering any of the other Table 1 biomarkers. The sensitivity and specificity of diagnosis can be improved, however, by combining data for multiple biomarkers. It is thus preferred to analyse more than one Table 1 biomarker. Analysis of two or more different biomarkers (a “panel”) can enhance the sensitivity and/or specificity of diagnosis compared to analysis of a single biomarker. Each different biomarker in a panel is shown in a different row in Table 1 i.e. measuring both auto-antibody which binds to an antigen listed in Table 1 and the antigen itself is measurement of a single biomarker rather than of a panel.
Thus the invention provides a method for analysing a subject sample, comprising a step of determining the levels of x different biomarkers of Table 1, wherein the levels of the biomarkers provide a diagnostic indicator of whether the subject has lupus. The value of x is 2 or more e.g. 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15 or more (e.g. up to 50). These panels may include (i) any specific one of the 50 biomarkers in Table 1 in combination with (ii) any of the other 49 biomarkers in Table 1. Suitable panels are described below and panels of particular interest include those listed in Tables 2 to 16. Preferred panels have from 2 to 15 biomarkers, as using >15 of them adds little to sensitivity and specificity.
The Table 1 biomarkers can be used in combination with one or more of: (a) known biomarkers for lupus, which may or may not be auto-antibodies or antigens; and/or (b) other information about the subject from whom a sample was taken e.g. age, genotype (genetic variations can affect auto-antibody profiles [4]), weight, other clinically-relevant data or phenotypic information; and/or (c) other diagnostic tests or clinical indicators for lupus. Such combinations can enhance the sensitivity and/or specificity of diagnosis. Thus the invention provides a method for analysing a subject sample, comprising a step of determining:
The samples used in (a) and (b) may be the same or different.
The value of y is 1 or more e.g. 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15 (e.g. up to 50). When y>1 the invention uses a panel of different Table 1 biomarkers.
The invention also provides, in a method for diagnosing if a subject has lupus, an improvement consisting of determining in a sample from the subject the level(s) of y biomarker(s) of Table 1, wherein the level(s) of the biomarker(s) provide a diagnostic indicator of whether the subject has lupus.
The invention also provides a method for diagnosing a subject as having lupus, comprising steps of: (i) determining the levels of y biomarkers of Table 1 in a sample from the subject; and (ii) comparing the determination from step (i) to data obtained from samples from subjects without lupus and/or from subjects with lupus, wherein the comparison provides a diagnostic indicator of whether the subject has lupus. The comparison in step (ii) can use a classifier algorithm as discussed in more detail below.
The invention also provides a method for monitoring development of lupus in a subject, comprising steps of: (i) determining the levels of z1 biomarker(s) of Table 1 in a first sample from the subject taken at a first time; and (ii) determining the levels of z2 biomarker(s) of Table 1 in a second sample from the subject taken at a second time, wherein: (a) the second time is later than the first time; (b) one or more of the z2 biomarker(s) were present in the first sample; and (c) a change in the level(s) of the biomarker(s) in the second sample compared with the first sample indicates that lupus is in remission or is progressing. Thus the method monitors the biomarker(s) over time, with changing levels indicating whether the disease is getting better or worse.
The disease development can be either an improvement or a worsening, and this method may be used in various ways e.g. to monitor the natural progress of a disease, or to monitor the efficacy of a therapy being administered to the subject. Thus a subject may receive a therapeutic agent before the first time, at the first time, or between the first time and the second time. Increased levels of antibodies against a particular antigen may be due to “epitope spreading”, in which additional antibodies or antibody classes are raised to antigens against which an antibody response has already been mounted [5].
The value of z1 is 1 or more e.g. 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15 (e.g. up to 50). The value of z2 is 1 or more e.g. 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15 (e.g. up to 50). The values of z1 and z2 may be the same or different. If they are different, it is usual that z1>z2 as the later analysis (z2) can focus on biomarkers which were already detected in the earlier analysis; in other embodiments, however, z2 can be larger than z1 e.g. if previous data have indicated that an expanded panel should be used; in other embodiments z2=z1 e.g. so that, for convenience, the same panel can be used for both analyses. When z1>1 or z2>1, the biomarkers are different biomarkers.
The invention also provides a method for monitoring development of lupus in a subject, comprising steps of: (i) determining the level of at least w1 Table 1 biomarkers in a first sample taken at a first time from the subject; and (ii) determining the level of at least w2 Table 1 biomarkers in a second sample taken at a second time from the subject, wherein: (a) the second time is later than the first time; (b) at least one biomarker is common to both the w1 and w2 biomarkers; (c) the level of at least one biomarker common to both the w1 and w2 biomarkers is different in the first and second samples, thereby indicating that the lupus is progressing or regressing. Thus the method monitors the range of biomarkers over time, with a broadening in the number of detected biomarkers indicating that the disease is getting worse. As mentioned above, this method may be used to monitor disease development in various ways.
The value of w1 is 1 or more e.g. 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15 (e.g. up to 50). The value of w2 is 2 or more e.g. 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15 (e.g. up to 50). The values of w1 and w2 may be the same or different. If they are different, it is usual that w2>w1, as the later analysis should focus on a biomarker panel that is at least as wide as the number already detected in the earlier analysis. There will usually be an overlap between the w1 and w2 biomarkers (including situations where they are the same, such that the same biomarkers are measured at two time points) but it is also possible for w1 and w2 to have no biomarkers in common.
Where the methods involve a first time and a second time, these times may differ by at least 1 day, 1 week, 1 month or 1 year. Samples may be taken regularly. The methods may involve measuring biomarkers in more than 2 samples taken at more than 2 time points i.e. there may be a 3rd sample, a 4th sample, a 5th sample, etc.
The invention also provides a diagnostic device for use in diagnosis of lupus, wherein the device permits determination of the level(s) of y Table 1 biomarkers. The value of y is defined above. The device may also permit determination of whether a sample contains one or more of the known lupus biomarkers mentioned above e.g. ANA and/or anti-DNA antibodies.
The invention also provides a kit comprising (i) a diagnostic device of the invention and (ii) instructions for using the device to detect y of the Table 1 biomarkers. The value of y is defined above. The kit is useful in the diagnosis of lupus.
The invention also provides a kit comprising reagents for measuring the levels of x different Table 1 biomarkers. The kit may also include reagents for determining whether a sample contains one or more of the known lupus biomarkers mentioned above e.g. ANA and/or anti-DNA antibodies. The value of x is defined above. The kit is useful in the diagnosis of lupus.
The invention also provides a kit comprising components for preparing a diagnostic device of the invention. For instance, the kit may comprise individual detection reagents for x different biomarkers, such that an array of those x biomarkers can be prepared.
The invention also provides a product comprising (i) one or more detection reagents which permit measurement of x different Table 1 biomarkers, and (ii) a sample from a subject.
The invention also provides a software product comprising (i) code that accesses data attributed to a sample, the data comprising measurement of y Table 1 biomarkers, and (ii) code that executes an algorithm for assessing the data to represent a level of y of the biomarkers in the sample. The software product may also comprise (iii) code that executes an algorithm for assessing the result of step (ii) to provide a diagnostic indicator of whether the subject has lupus. As discussed below, suitable algorithms for use in part (iii) include support vector machine algorithms, artificial neural networks, tree-based methods, genetic programming, etc. The algorithm can preferably classify the data of part (ii) to distinguish between subjects with lupus and subjects without based on measured biomarker levels in samples taken from such subjects. The invention also provides methods for training such algorithms.
The invention also provides a computer which is loaded with and/or is running a software product of the invention.
The invention also extends to methods for communicating the results of a method of the invention. This method may involve communicating assay results and/or diagnostic results. Such communication may be to, for example, technicians, physicians or patients. In some embodiments, detection methods of the invention will be performed in one country and the results will be communicated to a recipient in a different country.
The invention also provides an isolated antibody (preferably a human antibody) which recognises one of the antigens listed in Table 1. The invention also provides an isolated nucleic acid encoding the heavy and/or light chain of the antibody. The invention also provides a vector comprising this nucleic acid, and a host cell comprising this vector. The invention also provides a method for expressing the antibody comprising culturing the host cell under conditions which permit production of the antibody. The invention also provides derivatives of the human antibody e.g. F(ab′)2 and F(ab) fragments, Fv fragments, single-chain antibodies such as single chain Fv molecules (scFv), minibodies, dAbs, etc.
The invention also provides the use of a Table 1 biomarker as a biomarker for lupus.
The invention also provides the use of x different Table 1 biomarkers as biomarkers for lupus. The value of x is defined above. These may include (i) any specific one of the 50 biomarkers in Table 1 in combination with (ii) any of the other 49 biomarkers in Table 1.
The invention also provides the use as combined biomarkers for lupus of (a) at least y Table 1 biomarker(s) and (b) biomarkers including autoantibodies including ANA, anti-Smith, anti-dsDNA, anti-phospholipid, anti-ssDNA, anti-histone, false positive test for serological test for syphilis, indicators of serositis, oral ulcers, arthritis, photosensitivity haematological disorder, renal disorder, antinuclear antibody, immunologic disorder, neurologic disorder, malar rash, discoid rash (and optionally, any other known biomarkers e.g. see above). The value of y is defined above. When y>1 the invention uses a panel of biomarkers of the invention.
In all embodiments of the invention, the biomarker(s) from Table 1 is/are preferably those in Table 18. Table 18 is a preferred subset of 44 of the 50 biomarkers in Table 1. Even more preferably, the biomarker(s) from Table 1 is/are also in Table 20. Table 20 is a preferred subset of 17 of the 50 biomarkers in Table 1.
Auto-antibodies against 145 different human antigens have been identified and these can be used as lupus biomarkers. Details of the 145 antigens are given in Table 17. Within the 145 antigens, 50 human antigens are particularly useful for distinguishing between samples from subjects with lupus and from subjects without lupus. Details of these 50 antigens are given in Table 1. A preferred subset of antigens are the 44 antigens given in Table 18. An even more preferred subset of antigens is the 17 antigens given in Table 20. Further auto-antibody biomarkers can be used in addition to these 50 (e.g. any of the other biomarkers listed in Table 17). The sequence listing provides an example of a natural coding sequence for each of these antigens. These specific coding sequences are not limiting on the invention, however, and auto-antibody biomarkers may recognise variants of polypeptides encoded by these natural sequences (e.g. allelic variants, polymorphic forms, mutants, splice variants, or gene fusions), provided that the variant has an epitope recognised by the auto-antibody. Details on allelic variants of or mutations in human genes are available from various sources, such as the ALFRED database [6] or, in relation to disease associations, the OMIM [7] and HGMD [8] databases. Details of splice variants of human genes are available from various sources, such as ASD [9].
As mentioned above, detection of a single Table 1 biomarker can provide useful diagnostic information, but each biomarker might not individually provide information which is useful i.e. auto-antibodies against a Table 1 antigen may be present in some, but not all, subjects with lupus. An inability of a single biomarker to provide universal diagnostic results for all subjects does not mean that this biomarker has no diagnostic utility, however, or else ANA also would not be useful; rather, any such inability means that the test results (as in all diagnostic tests) have to be properly understood and interpreted.
To address the possibility that a single biomarker might not provide universal diagnostic results, and to increase the overall confidence that an assay is giving sensitive and specific results across a disease population, it is advantageous to analyse a plurality of the Table 1 biomarkers (i.e. a panel). For instance, a negative signal for a particular Table 1 antigen is not necessarily indicative of the absence of lupus (just as absence of antibodies to DNA is not), confidence that a subject does not have lupus increases as the number of negative results increases. For example, if all 50 biomarkers are tested and are negative then the result provides a higher degree of confidence than if only 1 biomarker is tested and is negative. Thus biomarker panels are most useful for enhancing the distinction seen between diseased and non-diseased samples. As mentioned above, though, preferred panels have from 2 to 15 biomarkers as the burden of measuring a higher number of markers is usually not rewarded by better sensitivity or specificity. Preferred panels are given below.
Where a biomarker or panel provides a strong distinction between lupus and non-lupus subjects then a method for analysing a subject sample can function as a method for diagnosing if a subject has lupus. As with many diagnostic tests, however, and as is already known for other diagnostics tests e.g. the PSA test used of prostate cancer, a method may not always provide a definitive diagnosis and so a method for analysing a subject sample can sometimes function only as a method for aiding in the diagnosis of lupus, or as a method for contributing to a diagnosis of lupus, where the method's result may imply that the subject has lupus (e.g. the disease is more likely than not) and/or may confirm other diagnostic indicators (e.g. passed on clinical symptoms). The test may therefore function as an adjunct to, or be integrated into, the SLEDAI analysis, or similar methodologies e.g. adjusted mean SLEDAI, European League Against Rheumatism (EULAR). Dealing with these considerations of certainty/uncertainty is well known in the diagnostic field.
The invention is used for diagnosing disease in a subject. The subject will usually be female and at least 10 years old (e.g. >15, >20, >25, >30, >35, >40, >45, >50, >55, >60, >65, >70). They will usually be at least of child-bearing age as the risk of lupus increases in this age group, and for these subjects it may be appropriate to offer a screening service for Table 1 biomarkers. The subject may be a post-menopausal female.
The subject may be pre-symptomatic for lupus or may already be displaying clinical symptoms. For pre-symptomatic subjects the invention is useful for predicting that symptoms may develop in the future if no preventative action is taken. For subjects already displaying clinical symptoms, the invention may be used to confirm or resolve another diagnosis. The subject may already have begun treatment for lupus.
In some embodiments the subject may already be known to be predisposed to development of lupus e.g. due to family or genetic links. In other embodiments, the subject may have no such predisposition, and may develop the disease as a result of environmental factors e.g. as a result of exposure to particular chemicals (such as toxins or pharmaceuticals), as a result of diet [10], of infection, of oral contraceptive use, of postmenopausal use of hormones, etc. [11].
Because the invention can be implemented relative easily and cheaply it is not restricted to being used in patients who are already suspected of having lupus. Rather, it can be used to screen the general population or a high risk population e.g. subjects at least 10 years old, as listed above.
The subject will typically be a human being. In some embodiments, however, the invention is useful in non-human organisms e.g. mouse, rat, rabbit, guinea pig, cat, dog, horse, pig, cow, or non-human primate (monkeys or apes, such as macaques or chimpanzees). In non-human embodiments, any detection antigens used with the invention will typically be based on the relevant non-human ortholog of the human antigens disclosed herein. In some embodiments animals can be used experimentally to monitor the impact of a therapeutic on a particular biomarker.
The invention analyses samples from subjects. Many types of sample can include auto-antibodies and/or antigens suitable for detection by the invention, but the sample will typically be a body fluid. Suitable body fluids include, but are not limited to, blood, serum, plasma, saliva, lymphatic fluid, a wound secretion, urine, faeces, mucus, sweat, tears and/or cerebrospinal fluid. The sample is typically serum or plasma.
In some embodiments, a method of the invention involves an initial step of obtaining the sample from the subject. In other embodiments, however, the sample is obtained separately from and prior to performing a method of the invention. After a sample has been obtained then methods of the invention are generally performed in vitro.
Detection of biomarkers may be performed directly on a sample taken from a subject, or the sample may be treated between being taken from a subject and being analysed. For example, a blood sample may be treated to remove cells, leaving antibody-containing plasma for analysis, or to remove cells and various clotting factors, leaving antibody-containing serum for analysis. Faeces samples usually require physical treatment prior to protein detection e.g. suspension, homogenisation and centrifugation. For some body fluids, though, such separation treatments are not usually required (e.g. tears or saliva) but other treatments may be used. For example, various types of sample may be subjected to treatments such as dilution, aliquoting, sub-sampling, heating, freezing, irradiation, etc. between being taken from the body and being analysed e.g. serum is usually diluted prior to analysis. Also, addition of processing reagents is typical for various sample types e.g. addition of anticoagulants to blood samples.
The invention involves determining the level of Table 1 biomarker(s) in a sample. Immunochemical techniques for detecting antibodies against specific antigens are well known in the art, as are techniques for detecting specific antigens themselves. Detection of an antibody will typically involve contacting a sample with a detection antigen, wherein a binding reaction between the sample and the detection antigen indicates the presence of the antibody of interest. Detection of an antigen will typically involve contacting a sample with a detection antibody, wherein a binding reaction between the sample and the detection antibody indicates the presence of the antigen of interest. Detection of an antigen can also be determined by non-immunological methods, depending on the nature of the antigen e.g. if the antigen is an enzyme then its enzymatic activity can be assayed, or if the antigen is a receptor then its binding activity can be assayed, etc. For example, the MAP2K5 kinase can be assayed using methods known in the art.
A detection antigen for a biomarker antibody can be a natural antigen recognised by the auto-antibody (e.g. a mature human protein disclosed in Table 1), or it may be an antigen comprising an epitope which is recognized by the auto-antibody. It may be a recombinant protein or synthetic peptide. Where a detection antigen is a polypeptide its amino acid sequence can vary from the natural sequences disclosed above, provided that it has the ability to specifically bind to an auto-antibody of the invention (i.e. the binding is not non-specific and so the detection antigen will not arbitrarily bind to antibodies in a sample). It may even have little in common with the natural sequence (e.g. a mimotope, an aptamer, etc.). Typically, though, a detection antigen will comprise an amino acid sequence (i) having at least 90% (e.g. ≧91%, ≧92%, ≧93%, ≧94%, ≧95%, ≧96%, ≧97%, ≧98%, ≧99%) sequence identity to the relevant SEQ ID NO disclosed herein across the length of the detection antigen, and/or (ii) comprising at least one epitope from the relevant SEQ ID NO disclosed herein. Thus the detection antigen may be one of the variants discussed above.
Epitopes are the parts of an antigen that are recognised by and bind to the antigen binding sites of antibodies and are also known as “antigenic determinants”. An epitope-containing fragment may contain a linear epitope from within a SEQ ID NO and so may comprise a fragment of at least n consecutive amino acids of the SEQ ID NO:, wherein n may be 7 or more (e.g. 8, 10, 12, 14, 16, 18, 20, 25, 30, 35, 40, 50, 60, 70, 80, 90, 100, 150, 200, 250 or more). B-cell epitopes can be identified empirically (e.g. using PEPSCAN [12,13] or similar methods), or they can be predicted e.g. using the Jameson-Wolf antigenic index [14], ADEPT [15], hydrophilicity [16], antigenic index [17], MAPITOPE [18], SEPPA [19], matrix-based approaches [20], the amino acid pair antigenicity scale [21], or any other suitable method e.g. see ref. 22. Predicted epitopes can readily be tested for actual immunochemical reactivity with samples.
Detection antigens can be purified from human sources but it is more typical to use recombinant antigens (particularly where the detection antigen uses sequences which are not present in the natural antigen e.g. for attachment). Various systems are available for recombinant expression, and the choice of system may depend on the auto-antibody to be detected. For example, prokaryotic expression (e.g. using E. coli) is useful for detecting many auto-antibodies, but if an auto-antibody recognises a glycoprotein then eukaryotic expression may be required. Similarly, if an auto-antibody recognises a specific discontinuous epitope then a recombinant expression system which provides correct protein folding may be required.
The detection antigen may be a fusion polypeptide with a first region and a second region, wherein the first region can react with an auto-antibody in a sample and the second region can react with a substrate to immobilise the fusion polypeptide thereon.
A detection antibody for a biomarker antigen can be a monoclonal antibody or a polyclonal antibody. Typically it will be a monoclonal antibody. The detection antibody should have the ability to specifically bind to a Table 1 antigen (i.e. the binding is not non-specific and so the detection antibody will not arbitrarily bind to other antigens in a sample).
Various assay formats can be used for detecting biomarkers in samples. For example, the invention may use one or more of western blot, immunoprecipitation, silver staining, mass spectrometry (e.g. MALDI-MS), conductivity-based methods, dot blot, slot blot, colorimetric methods, fluorescence-based detection methods, or any form of immunoassay, etc. The binding of antibodies to antigens can be detected by any means, including enzyme-linked assays such as ELISA, radioimmunoassays (RIA), immunoradiometric assays (IRMA), immunoenzymatic assays (IEMA), DELFIA™ assays, surface plasmon resonance or other evanescent light techniques (e.g. using planar waveguide technology), label-free electrochemical sensors, etc. Sandwich assays are typical for immunological methods.
In embodiments where multiple biomarkers are to be detected an array-based assay format is preferable, in which a sample that potentially contains the biomarkers is simultaneously contacted with multiple detection reagents (antibodies and/or antigens) in a single reaction compartment. Antigen and antibody arrays are well known in the art e.g. see references 23-29, including arrays for detecting auto-antibodies. Such arrays may be prepared by various techniques, such as those disclosed in references 30-34, which are particularly useful for preparing microarrays of correctly-folded polypeptides to facilitate binding interactions with auto-antibodies. It has been estimated that most B-cell epitopes are discontinuous and such epitopes are known to be important in diseases with an autoimmune component. For example, in autoimmune thyroid diseases, auto-antibodies arise to discontinuous epitopes on the immunodominant region on the surface of thyroid peroxidase and in Goodpasture disease auto-antibodies arise to two major conformational epitopes. Protein arrays which have been developed to present correctly-folded polypeptides displaying native structures and discontinuous epitopes are therefore particularly well suited to studies of diseases where auto-antibody responses occur [27].
Methods and apparatuses for detecting binding reactions on protein arrays are now standard in the art. Preferred detection methods are fluorescence-based detection methods. To detect biomarkers which have bound to immobilised proteins a sandwich assay is typical e.g. in which the primary antibody is an auto-antibody from the sample and the secondary antibody is a labelled anti-sample antibody (e.g. an anti-human antibody).
Where a biomarker is an auto-antibody the invention will generally detect IgG antibodies, but detection of auto-antibodies with other subtypes is also possible e.g. by using a detection reagent which recognises the appropriate class of auto-antibody (IgA, IgM, IgE or IgD rather than Ig). The assay format may be able to distinguish between different antibody subtypes and/or isotypes. Different subtypes [35] and isotypes [36] can influence auto-antibody repertoires. For instance, a sandwich assay can distinguish between different subtypes by using differentially-labelled secondary antibodies e.g. different labels for anti-IgG and anti-IgM.
As mentioned above, the invention provides a diagnostic device which permits determination of whether a sample contains Table 1 biomarkers. Such devices will typically comprise one or more antigen(s) and/or antibodies immobilised on a solid substrate (e.g. on glass, plastic, nylon, etc.). Immobilisation may be by covalent or non-covalent bonding (e.g. non-covalent bonding of a fusion polypeptide, as discussed above, to an immobilised functional group such as an avidin [32] or a bleomycin-family antibiotic [34]). Antigen arrays are a preferred format, with detection antigens being individually addressable. The immobilised antigens will be able to react with auto-antibodies which recognise a Table 1 antigen.
In some embodiments, the solid substrate may comprise a strip, a slide, a bead, a well of a microtitre plate, a conductive surface suitable for performing mass spectrometry analysis [37], a semiconductive surface [38,39], a surface plasmon resonance support, a planar waveguide technology support, a microfluidic devices, or any other device or technology suitable for detection of antibody-antigen binding.
Where the invention provides or uses an antigen array for detecting a panel of auto-antibodies as disclosed herein, in some embodiments the array may include only antigens for detecting these auto-antibodies. In other embodiments, however, the array may include polypeptides in addition to those useful for detecting the auto-antibodies. For example, an array may include one or more control polypeptides. Suitable positive control polypeptides include an anti-human immunoglobulin antibody, such as an anti-IgM antibody, an anti-IgG antibody, an anti-IgA antibody, an anti-IgE antibody or combinations thereof. Other suitable positive control polypeptides which can bind to sample antibodies include protein A or protein G, typically in recombinant form. Suitable negative control polypeptides include, but are not limited to, β-galactosidase, serum albumins (e.g. BSA or HSA), protein tags, bacterial proteins, yeast proteins, citrullinated polypeptides, etc. Negative control features on an array can also be polypeptide-free e.g. buffer alone, DNA, etc. An array's control features are used during performance of a method of the invention to check that the method has performed as expected e.g. to ensure that expected proteins are present (e.g. a positive signal from serum proteins in a serum sample) and that unexpected substances are not present (e.g. a positive signal from an array spot of buffer alone would be unexpected).
In an antigen array of the invention, at least 10% (e.g. ≧20%, ≧30%, ≧40%, ≧50%, ≧60%, ≧70%, ≧80%, ≧90%, ≧95%, or more) of the total number of different proteins present on the array may be for detecting auto-antibodies as disclosed herein.
An antigen array of the invention may include one or more replicates of a detection antigen and/or control feature e.g. duplicates, triplicates or quadruplicates. Replicates provide redundancy, provide intra-array controls, and facilitate inter-array comparisons.
An antigen array of the invention may include detection antigens for more than just the 44 different auto-antibodies described here, but preferably it can detect antibodies against fewer than 10000 antigens (e.g. <5000, <4000, <3000, <2000, <1000, <500, <250, <100, etc.).
An array is advantageous because it allows simultaneous detection of multiple biomarkers in a sample. Such simultaneous detection is not mandatory, however, and a panel of biomarkers can also be evaluated in series. Thus, for instance, a sample could be split into sub-samples and the sub-samples could be assayed in series. In this embodiment it may not be necessary to complete analysis of the whole panel e.g. the diagnostic indicators obtained on a subset of the panel may indicate that a patient has lupus without requiring analysis of any further members of the panel. Such incomplete analysis of the panel is encompassed by the invention because of the intention or potential of the method to analyse the complete panel.
As mentioned above, some embodiments of the invention can include a contribution from known tests for lupus, such as ANA and/or anti-DNA tests. Any known tests can be used e.g. Farr test, Crithidia, etc.
Thus an array of the invention (or any other assay format) may also provide an assay for one or more of these additional markers e.g. an array may include a DNA spot.
The invention involves a step of determining the level of Table 1 biomarker(s). In some embodiments of the invention this determination for a particular marker can be a simple yes/no determination, whereas other embodiments may require a quantitative or semi-quantitative determination, still other embodiments may involve a relative determination (e.g. a ratio relative to another marker, or a measurement relative to the same marker in a control sample), and other embodiments may involve a threshold determination (e.g. a yes/no determination whether a level is above or below a threshold). Usually biomarkers will be measured to provide quantitative or semi-quantitative results (whether as relative concentration, absolute concentration, titre, etc.) as this gives more data for use with classifier algorithms.
Usually the raw data obtained from an assay for determining the presence, absence, or level (absolute or relative) require some sort of manipulation prior to their use. For instance, the nature of most detection techniques means that some signal will sometimes be seen even if no antigen/antibody is actually present and so this noise may be removed before the results are interpreted. Similarly, there may be a background level of the antigen/antibody in the general population which needs to be compensated for. Data may need scaling or standardising to facilitate inter-experiments comparisons. These and similar issues, and techniques for dealing with them, are well known in the immunodiagnostic area.
Various techniques are available to compensate for background signal in a particular experiment. For example, replicate measurements will usually be performed (e.g. using multiple features of the same detection antigen on a single array) to determine intra-assay variation, and average values from the replicates can be compared (e.g. the median value of binding to quadruplicate array features). Furthermore, standard markers can be used to determine inter-assay variation and to permit calibration and/or normalisation e.g. an array can include one or more standards for indicating whether measured signals should be proportionally increased or decreased. For example, an assay might include a step of analysing the level of one or more control marker(s) in a sample e.g. levels of an antigen or antibody unrelated to lupus. Signal may be adjusted according to distribution in a single experiment. For instance, signals in a single array experiment may be expressed as a percentage of interquartile differences e.g. as [observed signal−25th percentile]/[75th percentile−25th percentile]. This percentage may then be normalised e.g. using a standard quantile normalization matrix, such as disclosed in reference 40, in which all percentage values on a single array are ranked and replaced by the average of percentages for antigens with the same rank on all arrays. Overall, this process gives data distributions with identical median and quartile values. Data transformations of this type are standard in the art for permitting valid inter-array comparisons despite variation between different experiments.
The level of a biomarker relative to a single baseline level may be defined as a fold difference. Normally it is desirable to use techniques that can indicate a change of at least 1.5-fold e.g. ≧1.75-fold, ≧2-fold, ≧2.5-fold, ≧5-fold, etc.
As well as compensating for variation which is inherent between different experiments, it can also be important to compensate for background levels of a biomarker which are present in the general population. Again, suitable techniques are well known. For example, levels of a particular antigen or auto-antibody in a sample will usually be measured quantitatively or semi-quantitatively to permit comparison to the background level of that biomarker. Various controls can be used to provide a suitable baseline for comparison, and choosing suitable controls is routine in the diagnostic field. Further details of suitable controls are given below.
The measured level(s) of Table 1 biomarker(s), after any compensation/normalisation/etc., can be transformed into a diagnostic result in various ways. This transformation may involve an algorithm which provides a diagnostic result as a function of the measured level(s). Where a panel is used then each individual biomarker may make a different contribution to the overall diagnostic result and so two biomarkers may be weighted differently.
The creation of algorithms for converting measured levels or raw data into scores or results is well known in the art. For example, linear or non-linear classifier algorithms can be used. These algorithms can be trained using data from any particular technique for measuring the marker(s). Suitable training data will have been obtained by measuring the biomarkers in “case” and “control” samples i.e. samples from subjects known to suffer from lupus and from subjects known not to suffer from lupus. Most usefully the control samples will also include samples from subjects with a related disease which is to be distinguished from the disease of interest e.g. it is useful to train the algorithm with data from rheumatoid arthritis subjects and/or with data from subjects with connective tissue diseases other than lupus. The classifier algorithm is modified until it can distinguish between the case and control samples e.g. by adding or removing markers from the analysis, by changes in weighting, etc. Thus a method of the invention may include a step of analysing biomarker levels in a subject's sample by using a classifier algorithm which distinguishes between lupus subjects and non-lupus subjects based on measured biomarker levels in samples taken from such subjects.
Various suitable classifier algorithms are available e.g. linear discriminant analysis, naïve Bayes classifiers, perceptrons, support vector machines (SVM) [41] and genetic programming (GP) [42]. GP is particularly useful as it generally selects relatively small numbers of biomarkers and overcomes the problem of trapping in a local maximum which is inherent in many other classification methods. SVM-based approaches have previously been applied to lupus datasets [43]. The inventors have previously confirmed that both SVM and GP approaches can be trained on the same biomarker panels to distinguish the auto-antibody/antigen biomarker profiles of case and control cohorts with similar sensitivity and specificity i.e. autoantibody biomarkers are not dependent on a single method of analysis. Moreover, these approaches can potentially distinguish lupus subjects from subjects with (i) other forms of autoimmune disease and (ii) rheumatoid arthritis. The 50 biomarkers in Table 1 can be used to train such algorithms to reliably make such distinctions.
It will be appreciated that, although there may be some biomarkers in Table 1 which always give a negative absolute signal when contacted with negative control samples (and thus any positive signal is immediately indicative of lupus), it is more common that a biomarker will give at least a low absolute signal (and thus that a disease-indicating positive signal requires detection of auto-antibody levels above that background level). Thus references herein detecting a biomarker may not be references to absolute detection but rather (as is standard in the art) to a level above the levels seen in an appropriate negative control. Such controls may be assayed in parallel to a test sample but it can be more convenient to use an absolute control level based on empirical data, or to analyse data using an algorithm which can (e.g. by previous training) use biomarker levels to distinguish samples from disease patients vs. non-disease patients.
The level of a particular biomarker in a sample from a lupus-diseased subject may be above or below the level seen in a negative control sample. Antibodies that react with self-antigens occur naturally in healthy individuals and it is believed that these are necessary for survival of T- and B-cells in the peripheral immune system [44]. In a control population of healthy individuals there may thus be significant levels of circulating auto-antibodies against some of the antigens disclosed in Table 1 and these may occur at a significant frequency in the population. The level and frequency of these biomarkers may be altered in a disease cohort, compared with the control cohort. An analysis of the level and frequency of these biomarkers in the case and control populations may identify differences which provide diagnostic information. The level of auto-antibodies directed against a specific antigen may increase or decrease in a lupus sample, compared with a healthy sample.
In general, therefore, a method of the invention will involve determining whether a sample contains a biomarker level which is associated with lupus. Thus a method of the invention can include a step of comparing biomarker levels in a subject's sample to levels in (i) a sample from a patient with lupus and/or (ii) a sample from a patient without lupus. The comparison provides a diagnostic indicator of whether the subject has lupus. An aberrant level of one or more biomarker(s), as compared to known or standard expression levels of those biomarker(s) in a sample from a patient without lupus, indicates that the subject has lupus.
The level of a biomarker should be significantly different from that seen in a negative control. Advanced statistical tools can be used to determine whether two levels are the same or different. For example, an in vitro diagnosis will rarely be based on comparing a single determination. Rather, an appropriate number of determinations will be made with an appropriate level of accuracy to give a desired statistical certainty with an acceptable sensitivity and/or specificity. Antigen and/or antibody levels can be measured quantitatively to permit proper comparison, and enough determinations will be made to ensure that any difference in levels can be assigned a statistical significance to a level of p<0.05 or better. The number of determinations will vary according to various criteria (e.g. the degree of variation in the baseline, the degree of up-regulation in disease states, the degree of noise, etc.) but, again, this falls within the normal design capabilities of a person of ordinary skill in this field. For example, interquartile differences of normalised data can be assessed, and the threshold for a positive signal (i.e. indicating the presence of a particular auto-antibody) can be defined as requiring that antibodies in a sample react with a diagnostic antigen at least 2.5-fold more strongly that the interquartile difference above the 75th percentile. Other criteria are familiar to those skilled in the art and, depending on the assays being used, they may be more appropriate than quantile normalisation. Other methods to normalise data include data transformation strategies known in the art e.g. scaling, log normalisation, median normalisation, etc.
The underlying aim of these data interpretation techniques is to distinguish between the presence of a Table 1 biomarker and of an arbitrary control biomarker, and also to distinguish between the response of sample from a lupus subject from a control subject. Methods of the invention may have sensitivity of at least 70% (e.g. >70%, >75%, >80%, >85%, >90%, >95%, >96%, >97%, >98%, >99%). Methods of the invention may have specificity of at least 70% (e.g. >70%, >75%, >80%, >85%, >90%, >95%, >96%, >97%, >98%, >99%). Advantageously, methods of the invention may have both specificity and sensitivity of at least 70% (e.g. >70%, >75%, >80%, >85%, >90%, >95%, >96%, >97%, >98%, >99%). As shown in Tables 9-16, the invention can consistently provide specificities above 90% and sensitivities greater than 80%.
Data obtained from methods of the invention, and/or diagnostic information based on those data, may be stored in a computer medium (e.g. in RAM, in non-volatile computer memory, on CD-ROM) and/or may be transmitted between computers e.g. over the internet.
If a method of the invention indicates that a subject has lupus, further steps may then follow. For instance, the subject may undergo confirmatory diagnostic procedures, such as those involving physical inspection of the subject, and/or may be treated with therapeutic agent(s) suitable for treating lupus.
As mentioned above, some methods of the invention involve testing samples from the same subject at two or more different points in time. In general, where the above text refers to the presence or absence of biomarker(s), the invention also includes an increasing or decreasing level of the biomarker(s) over time. An increasing level of an auto-antibody biomarker includes a spread of antibodies in which additional antibodies or antibody classes are raised against a single antigen. Methods which determine changes in biomarker(s) over time can be used, for instance, to monitor the efficacy of a therapy being administered to the subject (e.g. in theranostics). The therapy may be administered before the first sample is taken, at the same time as the first sample is taken, or after the first sample is taken.
The invention can be used to monitor a subject who is receiving lupus therapy. There is presently no cure for lupus. Current therapies for lupus include therapeutic drugs, alternative medicines or life-style changes. Approved drugs include non-steroidal and steroidal anti-inflammatory drugs (e.g. prednisolone), anti-malarials (e.g. hydroxychloriquine) and immunosupressants (e.g. cyclosporin A). A series of new drugs are being developed, many of which target B-cells, such as Rituximab which targets CD20 and Belimumab which is directed against B-lymphocyte stimulator (BlyS). The appropriate treatment regime will depend on the severity of the disease, and the responsiveness of the patient. Disease-modifying antirheumatic drugs can be used preventively to reduce the incidence of flares. When flares occur, they are often treated with corticosteroids. Given the similarities between rheumatic diseases, discussed below, it is not surprising that many of the therapeutics developed for one disease may have efficacy in another. In particular, the success of cytokine inhibitors in treating RA has advanced our understanding of these diseases and has opened up the possibility that some of these new classes of therapeutics will be of use in multiple disease areas. For example, Belimumab failed to meet its target in RA but has demonstrated efficacy in a phase III trial for lupus. Another anti-CD20 antibody, Ocrelizumab, is being investigated for use in RA and lupus and Imatinib which targets kit, abl and PDGFR kinases is in Phase II for RA and scleroderma. Other representative molecules which are directed towards rheumatic diseases are (target in parentheses): Tocilizumab (IL-6 receptor), AMG714 mAb (IL-15), AlN457 mAb (IL-17), Ustekinumab (IL-23/IL-12), Belimumab (BLyS/BAFF), Atacicept (BLyS/BAFF and APRIL), Baminercept (LTα/LTβ/LIGHT), Ocrelizumab (CD20), Ofatumumab (CD20), TRU-015/SMIP (CD20), Epratuzumab (CD22), Abatacept (CD80/CD86), Denosumab (RANKL), INCB018424 (JAK1/JAK2/Tyk2), CP-690,550 (JAK3), Fostamatinib (Syk), multiple compounds (p38), Imatinib (PDGF-R, c-kit, c-abl), ARRY-162 (ERK/MEK), AS-605240 (PI3Kγ), Maraviroc (CCR5), IB-MECA/CF101 (Adenosine A3 receptor agonist) and CE-224,535 (P2X7 antagonist).
In related embodiments of the invention, the results of monitoring a therapy are used for future therapy prediction. For example, if treatment with a particular therapy is effective in reducing or eliminating disease symptoms in a subject, and is also shown to decrease levels of a particular biomarker in that subject, detection of that biomarker in another subject may indicate that this other subject will respond to the same therapy. Conversely, if a particular therapy was not effective in reducing or eliminating disease symptoms in a subject who had a particular biomarker or biomarker profile, detection of that biomarker or profile in another subject may indicate that this other subject will also fail to respond to the same therapy.
In other embodiments, the presence of a particular biomarker can be used as the basis of proposing or initiating a particular therapy (patient stratification). For instance, if it is known that levels of a particular auto-antibody can be reduced by administering a particular therapy then that auto-antibody's detection may suggest that the therapy should begin. Thus the invention is useful in a theranostic setting.
Normally at least one sample will be taken from a subject before a therapy begins.
Where the development of auto-antibodies to a newly-exposed auto-antigen is causative for a disease, early priming of the immune response can prepare the body to remove antigen-exposing cells when they arise, thereby removing the cause of disease before auto-antibodies develop dangerously. For example, one antigen known to be recognised by auto-antibodies is p53, and this protein is considered to be both a vaccine target and a therapeutic target for the modulation of cancer [45-47]. The antigens listed in Tables 1 and 17 are thus therapeutic targets for treating lupus.
Thus the invention provides a method for raising an antibody response in a subject, comprising eliciting to the subject an immunogen which elicits antibodies which recognise an antigen listed in Table 1. The method is suitable for immunoprophylaxis of lupus.
The invention also provides an immunogen for use in medicine, wherein the immunogen can elicit antibodies which recognise an antigen listed in Table 1. Similarly, the invention also provides the use of an immunogen in the manufacture of a medicament for immunoprophylaxis of lupus, wherein the immunogen can elicit antibodies which recognise an antigen listed in Table 1.
As discussed above for detection antigens, the immunogen may be the antigen itself or may comprise an amino acid sequence having identity and/or comprising an epitope from the antigen. Thus the immunogen may comprise an amino acid sequence (i) having at least 90% (e.g. ≧91%, ≧92%, ≧93%, ≧94%, ≧95%, ≧96%, ≧97%, ≧98%, ≧99%) sequence identity to the relevant SEQ ID NO disclosed herein, and/or (ii) comprising at least one epitope from the relevant SEQ ID NO disclosed herein. Other immunogens may also be used, provided that they can elicit antibodies which recognise the antigen of interest.
As an alternative to immunising a subject with a polypeptide immunogen, it is possible to administer a nucleic acid (e.g. DNA or RNA) immunogen encoding the polypeptide, for in situ expression in the subject, thereby leading to the development of an antibody response.
The immunogen may be delivered in conjunction (e.g. in admixture) with an immunological adjuvant. Such adjuvants include, but are not limited to, insoluble aluminium salts, water-in-oil emusions, oil-in-water emulsions such as MF59 and AS03, saponins, ISCOMs, 3-O-deacylated MPL, immunostimulatory oligonucleotides (e.g. including one or more CpG motifs), bacterial ADP-ribosylating toxins and detoxified derivatives thereof, cytokines, chitosan, biodegradable microparticles, liposomes, imidazoquinolones, phosphazenes (e.g. PCPP), aminoalkyl glucosaminide phosphates, gamma inulins, etc. Combinations of such adjuvants can also be used. The adjuvant(s) may be selected to elicit an immune response involving CD4 or CD8 T cells. The adjuvant(s) may be selected to bias an immune response towards a TH1 phenotype or a TH2 phenotype.
The immunogen may be delivered by any suitable route. For example, it may be delivered by parenteral injection (e.g. subcutaneously, intraperitoneally, intravenously, intramuscularly), or mucosally, such as by oral (e.g. tablet, spray), topical, transdermal, transcutaneous, intranasal, ocular, aural, pulmonary or other mucosal administration.
The immunogen may be administered in a liquid or solid form. For example, the immunogen may be formulated for topical administration (e.g. as an ointment, cream or powder), for oral administration (e.g. as a tablet or capsule, as a spray, or as a syrup), for pulmonary administration (e.g. as an inhaler, using a fine powder or a spray), as a suppository or pessary, as drops, or as an injectable solution or suspension.
The antigens listed in Tables 1 and 17 can be useful for imaging. A labelled antibody against the antigen can be injected in vivo and the distribution of the antigen can then be detected. This method may identify the source of the antigen (e.g. an area in the body where there is a high concentration of the antigen), potentially offering early identification of lupus. Imaging techniques can also be used to monitor the progress or remission of disease, or the impact of a therapy.
The antigens listed in Table 1 can be useful for analysing tissue samples by staining e.g. using standard immunocytochemistry. A labelled antibody against a Table 1 antigen can be contacted with a tissue sample to visualise the location of the antigen. A single sample could be stained with different antibodies against multiple different antigens, and these different antibodies may be differentially labelled to enable them to be distinguished. As an alternative, a plurality of different samples can each be stained with a single antibody.
Thus the invention provides a labelled antibody which recognises an antigen listed in Table 1. The antibody may be a human antibody, as discussed above. Any suitable label can be used e.g. quantum dots, spin labels, fluorescent labels, dyes, etc.
The invention has been described above by reference to auto-antibody and antigen biomarkers, with assays of auto-antibodies against an antigen being used in preference to assays of the antigen itself. In addition to these biomarkers, however, the invention can be used with other biological manifestations of the Table 1 antigens. For example, the level of mRNA transcripts encoding a Table 1 antigencan be measured, particularly in tissues where that gene is not normally transcribed (such as in the potential disease tissue). Similarly, the chromosomal copy number of a gene encoding a Table 1 antigen can be measured e.g. to check for a gene duplication event. The level of a regulator of a Table 1 antigen can be measured e.g. to look at a microRNA regulator of a gene encoding the antigen. Furthermore, things which are regulated by or respond to a Table 1 antigen can be assessed e.g. if an antigen is a regulator of a metabolic pathway then disturbances in that pathway can be measured. Further possibilities will be apparent to the skilled reader.
Preferred embodiments of the invention are based on a panel of biomarkers. Panels of particular interest consist of or comprise the combinations of biomarkers listed in Tables 3 to 16 (which show ten panels of 2, 3, 4, . . . , 14 and 15 biomarkers). Table 19 shows 13 further preferred panels.
The ten different panels listed in each of Tables 3 to 16 can be expanded by adding further biomarker(s) to create a larger panel. The further biomarkers can usefully be selected from known biomarkers (such as ANA, anti-DNA antibodies, etc.; see above), from Table 17, or from Table 1. In general the addition does not decrease the sensitivity or specificity of the panel shown in the Tables. Such panels include, but are not limited to:
Preferred panels have between 2 and 15 biomarkers in total.
All definitions herein which refer to biomarkers of Table 1 are also disclosed by reference to Table 21 instead. Thus, for instance, the invention provides a method for analysing a subject sample, comprising a step of determining the level of a Table 21 biomarker in the sample, wherein the level of the biomarker provides a diagnostic indicator of whether the subject has lupus.
The term “comprising” encompasses “including” as well as “consisting” e.g. a composition “comprising” X may consist exclusively of X or may include something additional e.g. X+Y.
References to an antibody's ability to “bind” an antigen mean that the antibody and antigen interact strongly enough to withstand standard washing procedures in the assay in question. Thus non-specific binding will be minimised or eliminated.
References to a “level” of a biomarker mean the amount of an analyte measured in a sample and this encompasses relative and absolute concentrations of the analyte, analyte titres, relationships to a threshold, rankings, percentiles, etc.
An assay's “sensitivity” is the proportion of true positives which are correctly identified i.e. the proportion of lupus subjects who test positive by a method of the invention. This can apply to individual biomarkers, panels of biomarkers, single assays or assays which combine data integrated from multiple sources e.g. ANA, anti-DNA and/or other clinical test such as those included in the SLEDAI index. It can relate to the ability of a method to identify samples containing a specific analyte (e.g. antibodies) or to the ability of a method to correctly identify samples from subjects with lupus.
An assay's “specificity” is the proportion of true negatives which are correctly identified i.e. the proportion of subjects without lupus who test negative by a method of the invention. This can apply to individual biomarkers, panels of biomarkers, single assays or assays which combine data integrated from multiple sources e.g. ANA, anti-DNA and/or other clinical tests such as those included for consideration in the SLEDAI index. It can relate to the ability of a method to identify samples containing a specific analyte (e.g. antibodies) or to the ability of a method to correctly identify samples from subjects with lupus.
Unless specifically stated, a method comprising a step of mixing two or more components does not require any specific order of mixing. Thus components can be mixed in any order. Where there are three components then two components can be combined with each other, and then the combination may be combined with the third component, etc.
References to a percentage sequence identity between two amino acid sequences means that, when aligned, that percentage of amino acids are the same in comparing the two sequences. This alignment and the percent homology or sequence identity can be determined using software programs known in the art, for example those described in section 7.7.18 of ref. 48. A preferred alignment is determined by the Smith-Waterman homology search algorithm using an affine gap search with a gap open penalty of 12 and a gap extension penalty of 2, BLOSUM matrix of 62. The Smith-Waterman homology search algorithm is disclosed in ref. 49.
Table 17 lists 145 biomarkers. From within these 145, a preferred subset is SEQ ID NOs:1-139.
Table 1 lists 50 biomarkers. From within these 50, a preferred subset is the 44 listed in Table 18.
In all embodiments of the invention, where only one biomarker is used, the biomarker is preferably not PIAS2 or PABPC1. In all embodiments of the invention, where only two biomarkers are used, these two biomarkers are preferably not PIAS2 and PABPC1.
Three separate protein arrays were developed which were enriched for proteins associated with transcription (TRN array), kinases and kinase-associated proteins (KIN array) and cancer associated antigens (CAG array) described in sources such as the cancer immunome and SEREX databases. Full-length open reading frames for target genes encoding the 999 proteins present on the arrays were cloned in-frame with a sequence encoding a C-terminal E. coli BCCP-myc tag [23, 33] in a baculovirus transfer vector and sequence-verified. Several of the kinases which were integral membrane proteins were cloned as N- or C-terminal truncations representing the extracellular or cytoplasmic domains. Recombinant baculoviruses were generated, amplified and expressed in Sf9 cells using standard methods adapted for 24-well deep well plates. Recombinant protein expression was analyzed for protein integrity and biotinylation by Western blotting. Cells harbouring recombinant protein were lysed and lysates were spotted in quadruplicate using a QArray2 Microarrayer equipped with 300 μm solid pins on to streptavidin-coated glass slides. Spotted proteins project into an aqueous environment and orient away from the surface of the slide, exposing them for binding by auto-antibodies. In addition to the proteins on each array, four control proteins for the BCCP-myc tag (BCCP, BCCP-myc, β-galactosidase-BCCP-myc and β-galactosidase-BCCP) were arrayed, along with Cy3/Cy5-labeled biotin-BSA, dilution series of biotinylated-IgG and biotinylated IgM, a biotinylated-myc peptide dilution series and buffer-only spots.
Serum samples were obtained from two groups of subjects:
Serum samples from both groups were individually analysed using each of the three types of arrays. Serum samples were incubated with each of the three array types separately. Serum samples were clarified by centrifugation at 10-13K rpm for 2 minutes at 4° C. to remove particulates, including lipids. The samples were then diluted 200-fold in 0.1% v/v Triton/0.1% v/v BSA in 1×PBS (Triton-BSA buffer) and then applied to the arrays. Diluted serum (4 mL) sample was added to each array housed in a separate compartment of a plastic dish. All arrays were incubated for 2 hours at room temperature (RT, 20° C.) with gentle orbital shaking (˜50 rpm). Arrays were removed carefully from the dish and any excess probing solution was removed by blotting the sides of the array onto lint-free tissue. Probed arrays were washed three times in fresh Triton-BSA buffer at RT for 20 minutes with gentle orbital shaking. The washed slides were then blotted onto lint-free tissue to remove excess wash buffer and were incubated in a secondary staining solution (prepared just prior to use) at RT for 2 hours, with gentle orbital shaking and protected from light using aluminium foil. The secondary staining solution was a labelled anti-human IgG antibody. Slides were washed three times in Triton-BSA buffer for 5 minutes at RT with gentle orbital shaking, rinsed briefly (5-10 seconds) in distilled water, and centrifuged for 2 minutes at 240 g in a container suitable for centrifugation. To help wick away excess liquid on the arrays, a lint-free tissue was placed at the bottom of the arrays during centrifugation.
The probed and dried arrays were then scanned using a microarray scanner capable of using an excitation wavelength suitable for the detection of the secondary staining solution, to detect auto-antibodies bound by the array and to determine magnitude of auto-antibody binding. The microarray scans produced images for each array that were used to determine the intensity of fluorescence bound to each protein spot which were used to normalize and score array data.
Raw median signal intensity (also referred to as the relative fluorescent unit, RFU) of each protein feature (also referred to as a spot or antigen) on the array was subtracted from the local median background intensity. Alternative analyses use other measures of spot intensity such as the mean fluorescence, total fluorescence, as known in the art.
The resulting net fluorescent intensities of all protein features on each array were then normalized to reduce the influence of technical bias (e.g. laser power variation, surface variation, binding to BCCP, etc.) by a multiscaling procedure. Other methods for data normalization suitable for the data include, amongst others, quantile normalization [40], multiplication of net fluorescent intensities by a normalisation factor consisting of the product of the 1st quartile of all intensities of a sample and the mean of the 1st quartiles of all samples and the “VSN” method [50]. Such normalization methods are known in the art of microarray analysis. The normalized fluorescent intensities were then averaged for each protein feature.
The multiscaling method was applied to all 3996 quadruplicate signals from 326 protein arrays. Data were arbitrarily split in test and training sets and the data from the training set was then used with GP to identify classifiers which would successfully distinguish case from control samples. Classifiers were then assessed for performance by referring to the combined sensitivity and specificity (S+S score) using the test set. Data were repeatedly split into test and training sets and analysis cycles repeated until a stable set of classifiers (“panel”) was identified.
The number of biomarkers in each panel was limited to n where n=1-15. Multiple combinations of putative biomarkers were derived and the performance of the derived panels was then ranked by combined S+S score. The top 6000 panels for each n-mer panel were taken and the frequency of appearance of each protein in these panels was used to rank the predictive power of each protein for that specific n-mer. The top 10 biomarkers for each n-mer, as judged by frequency of appearance were also identified and then combined into a single list (Table 18). These represent biomarkers of particular interest as they represent the subset of biomarkers with the greatest predictive properties.
For each n-mer, the 25 panels which provide the highest combined S+S score are presented in Tables 2-16. The biomarkers frequently appearing in the top 25 panels for all the presented n-mers were combined to produce the set of 44 markers in Table 18. The top panels in Tables 5-16 each have a S+S score higher than the value of 1.5 (i.e. above the typical value for ANA [1]).
Overall, Tables 2-16 produced the biomarkers of SEQ ID NOs:1-139 in Table 17, a subset of 44 of which are presented in Table 18. Many of these 44 biomarkers has significant predictive power across multiple n-mers. For example, IGHG1 has the greatest combined S+S score for a single marker but is not a significant contributor to panels above 2-mers in size. In contrast, KIT is important for all sizes of panels from n=1 to n=15 Thus the contribution that a particular biomarker provides to the discriminatory power of a panel can depend on the number of markers in that panel as well as on their identity.
Some markers have previously been identified in association with lupus in particular or more generally with diseases with an autoimmune component. In particular, STAT1 has been previously linked with active pathways in lupus [51] and SSX2 and SSX4 were originally identified as antigens to which autoantibodies were raised in cancer.
The presence of antibodies to the Table 18 antigens was confirmed to be significantly different between the two groups. A back propagation algorithm was used to confirm biomarkers that can distinguish between the two groups. The data analysis was validated by two permutation assays. These assays confirmed that the chosen biomarkers are related to the disease status of the sera. The core biomarker set was successfully validated by depleting the set of 999 proteins of the 44 identified biomarkers and repeating the analysis. With the data from these biomarkers removed, it was no longer possible to derive a panel which could distinguish between healthy and diseased serum samples with comparable performance.
In a second analysis, the identical raw data as described previously was used. The identification of biomarkers was performed essentially as described above with the following changes. The raw array data was normalized by consolidating the replicates (median consolidation), followed by normal transformation and then median normalisation. Outliers were identified and removed. There is no method of normalisation which is universally appropriate and factors such as study design and sample properties must be considered. For the current study median normalisation was used. Other normalisation methods include, amongst others, quantile normalisation, multiplication of net fluorescent intensities by a normalisation factor consisting of the product of the 1st quartile of all intensities of a sample and the mean of the 1st quartiles of all samples and the “VSN” method. Such normalisation methods are known in the art of microarray analysis.
This normalised data was then used for the identification of biomarker panels. It is not possible to predict a priori which classifier will perform best with a given dataset, therefore data analysis was performed with 5 different feature ranking methods (1-5) plus forward feature selection:
Other classification methods as known in the art could be used. Classifiers were then assessed for performance by referring to the combined sensitivity and specificity (S+S score) and area under the curve (AUC). Data were repeatedly split and analysis cycles repeated until a stable set of classifiers (“panels”) was identified. Nested cross validation was applied to the classification procedures in order to avoid overfitting of the study data. The performance of the classification was compared to a randomized set of case-control status samples (permutation assay) which should give no predictive performance and provides an indication of the background in the analysis. A FIGURE close to 1.0 is expected for the null assay (equivalent to a sensitivity+specificity (S+S) score of 0.5+0.5, respectively) whereas an S+S score of 2.0 would indicate 100% sensitivity and 100% specificity. The difference between the values for the permutation analysis and the classifier performance indicates the relative strength of the classifier. For each analysis, multiple combinations of putative biomarkers were derived and the performance of the derived panels was then ranked by combined S+S score. The top 13 panels for the best performing n-mer panel (containing 3 biomarkers; shown in Table 19) were taken and the frequency of appearance of each protein in these panels was used to rank the predictive power of each protein included in these panels. The biomarkers with the greatest diagnostic power, as judged by frequency of appearance in the panels derived were identified and combined into a single list (Table 20). These represent biomarkers of particular interest as they correspond to the subset of biomarkers with the greatest predictive properties.
The maximum S+S score was obtained with the forward feature selection method (S+S=1.41; sensitivity=0.54, specificity=0.87) which gave an AUC value of 0.75 and corresponding to panels consisting of 3 biomarkers. The sensitivity reached 0.54 and the specificity was 0.87. The biomarkers which showed greatest diagnostic power include KIT, PIAS2, RPL15, ACTL7B, EEF1G and TCEB3, many of which were also identified in the previous analysis.
The performance of biomarker panels containing 3 proteins, identified by forward selection is shown below:
It will be understood that the invention has been described by way of example only and modifications may be made whilst remaining within the scope and spirit of the invention.
(i)The “Symbol” column gives the gene symbol which has been approved by the HGNC. The symbol thus identifies a unique human gene. This symbol can be related via Table 17 to the gene's Official Full Name provided by NCBI.
(ii)This number is the SEQ ID NO: for the coding sequence for the auto-antigen biomarker, as shown in Table 17.
(iii)The HUGO Gene Nomenclature Committee aims to give unique and meaningful names to every human gene. The HGNC number thus identifies a unique human gene.
Table 1 lists biomarkers useful with the invention. The measured biomarker can be (i) presence of auto-antibody which binds to an antigen listed in Table 1 and/or (ii) the presence of an antigen listed in Table 1, but is preferably the former.
(i)This is the symbol for the relevant biomarker (or, for Tables 3-16, biomarkers in the panel).
(ii)S + S is the sum of the sensitivity and specificity columns. These final two columns show the sensitivity and specificity of a test based solely on the relevant biomarker (or, for Tables 3-16, panel) shown in the left-hand column when applied to the samples used in the examples.
(i)This number is the SEQ ID NO: for the coding sequence for the auto-antigen biomarker, as shown in the sequence listing.
(ii)The “Symbol” column is as described for Table 1.
(iii)This name is taken from the Official Full Name provided by NCBI. An antigen may have been referred to by one or more pseudonyms in the prior art. The invention relates to these antigens regardless of their nomenclature.
(iv)A “GI” number, “GenInfo Identifier”, is a series of digits assigned consecutively to each sequence record processed by NCBI when sequences are added to its databases. The GI number bears no resemblance to the accession number of the sequence record. When a sequence is updated (e.g. for correction, or to add more annotation or information) it receives a new GI number. Thus the sequence associated with a given GI number is never changed.
(v)The “ID” column shows the Entrez GeneID number for the antigen marker. An Entrez GeneID value is unique across all taxa.
(i)The “Symbol” column gives the gene symbol which has been approved by the HGNC. The symbol thus identifies a unique human gene. This symbol can be related via Table 17 to the gene's Official Full Name provided by NCBI.
(ii)This number is the SEQ ID NO: for the coding sequence for the auto-antigen biomarker, as shown in Table 17.
(iii)The HUGO Gene Nomenclature Committee aims to give unique and meaningful names to every human gene. The HGNC number thus identifies a unique human gene.
(i)The “Symbol” column is as described for Table 1.
(ii)This name is taken from the Official Full Name provided by NCBI. An antigen may have been referred to by one or more pseudonyms in the prior art. The invention relates to these antigens regardless of their nomenclature.
(iii)A “GI” number, “GenInfo Identifier”, is a series of digits assigned consecutively to each sequence record processed by NCBI when sequences are added to its databases. The GI number bears no resemblance to the accession number of the sequence record. When a sequence is updated (e.g. for correction, or to add more annotation or information) it receives a new GI number. Thus the sequence associated with a given GI number is never changed.
(iv)The “ID” column shows the Entrez GeneID number for the antigen marker. An Entrez GeneID value is unique across all taxa.
(i)This number is the SEQ ID NO: for the coding sequence for the auto-antigen biomarker, as shown in the sequence listing.
(ii)The “Symbol” column is as described for Table 1.
(iii)This name is taken from the Official Full Name provided by NCBI. An antigen may have been referred to by one or more pseudonyms in the prior art. The invention relates to these antigens regardless of their nomenclature.
(iv)A “GI” number, “GenInfo Identifier”, is a series of digits assigned consecutively to each sequence record processed by NCBI when sequences are added to its databases. The GI number bears no resemblance to the accession number of the sequence record. When a sequence is updated (e.g. for correction, or to add more annotation or information) it receives a new GI number. Thus the sequence associated with a given GI number is never changed.
(v)The “ID” column shows the Entrez GeneID number for the antigen marker. An Entrez GeneID value is unique across all taxa.
| Number | Date | Country | Kind |
|---|---|---|---|
| 1017520.6 | Oct 2010 | GB | national |
| Filing Document | Filing Date | Country | Kind | 371c Date |
|---|---|---|---|---|
| PCT/IB2011/054572 | 10/14/2011 | WO | 00 | 8/22/2013 |
