FIELD OF INVENTION

Described are protein biomarkers of diabetic nephropathy. The biomarkers are present at differing concentrations in biological samples according to the health status of the individual, affording a simple flexible diagnostic means of differentiating individuals with diabetic nephropathy, from individuals with Type 2 diabetes and healthy individuals.

BACKGROUND

Diabetes Mellitus (DM) and associated multisystem complications have severe implications on global health, due to the increasing prevalence and subsequent management costs. This chronic condition arises from the inability of the body to either produce or utilise insulin correctly to maintain glucose homeostasis.

The International Diabetes Federation states that the global diabetes prevalence in adults of ages 20-79 in 2019 was 463 million people (9.3% of the global population) and is projected to rise to 578 million by 2030 (Saeedi et al., 2019) causing diabetes to be labelled a “serious threat to global health”. Diabetes and related complications were associated with 4.2 million deaths globally in 2019. The cost of management of diabetes and subsequent complications is large with a global estimate of global healthcare expenditure in 2017 of 760 billion United Stated Dollars (USD). In addition, it is estimated that 50.1% of the global population with diabetes remain undiagnosed (Saeedi et al., 2019). Diabetes is classified into several types, the most common of which are Type 1 (T1 DM) and Type 2 (T2DM) diabetes mellitus. Other classifications include diabetes in pregnancy (DIP) and gestational diabetes (GDM). It can also develop from an existing condition such as endocrine disorder, viral infection, or genetic disorder such as Prader-Willi syndrome, Down’s syndrome and Friedreich’s ataxia (IDF, 2019). T2DM is the most common type of diabetes and accounts for -90% of total diabetes cases worldwide. Occurrences of T2DM have increased in recent years and are predicted to continue to do so. Diagnosis is generally made in adulthood but incidence in children and young adults is increasing due to increasing sedentary lifestyles and high energy dietary practices (Saeedi et al., 2019). The diabetic environment is one of oxidative stress and inflammation, causing damage to renal cells and loss of kidney function in the case of diabetic nephropathy (DN).

Diabetic nephropathy is currently the leading global cause of end stage renal disease (ESRD) and accounts for 27.5 % of patients requiring dialysis or renal replacement therapy in the UK (Kidney Research UK and Diabetes UK 2018). The economic impact of late-stage diabetic nephropathy is much higher than that of other microvascular complications. It has been estimated that medical costs for T2DM patients with clinical nephropathy ($9,720) can be a third higher than those with no nephropathy or microalbuminuria (~$3,000). If the disease progresses to ESRD and dialysis is required costs increase up to four-fold ($41 ,117) and approximately twofold if a patient with neuropathy requires amputation ($22,798). Screening for chronic kidney disease in high-risk populations such as diabetes was found to be cost- effective according to the World Health Organisation. A systematic review investigated the risk factors of DN in T2DM patients. Nine independent factors were identified and validated in a risk prediction model: age, body mass index (BMI), smoking status, glycated haemoglobin (HbA1c), systolic blood pressure, DR, high- density lipoprotein cholesterol, triglyceride, urine albumin-to-creatinine ration (UCAR) and estimated glomerular filtration rate (eGFR) (Jiang et al., 2020). There is evidence to suggest that the presence of retinopathy acts as a predictor (Li et al., 2021). Hypertension is a particularly important risk factor for microvascular complications as increased pressure puts strain on small vessels and enters a damaging cycle of activation of the renin-angiotensin-aldosterone system (RAAS). A 2-year prospective cohort study carried out in Japan (KAMOGAWA-HBP study) on 477 patients with no diagnosed DN revealed systolic hypertension to be disproportionately high compared with diastolic hypertension, and a prognostic factor in predicting the development of DN (Kitagawa et al., 2019).

Pathophysiology DN is characterised by a variety of structural and molecular changes to the nephron, leading to an inflammatory and pro-fibrotic milieu, glomerulosclerosis and loss of the nephron. Nephrons are multifunctional filtration systems, located in the cortex, which, together with capillaries and collecting ducts, regulate waste and electrolyte homeostasis. Structurally, they are composed of a convoluted capillary which feeds into the surrounding Bowman’s capsule, delivering filtrate to an epithelial tubule and eventually to collecting ducts. Glomerular capillaries are surrounded by specialised cells, podocytes, which wrap around the vessels attached by foot processes and mesangial cells which are located in the space between capillaries. Fenestrated endothelium, podocytes and the glomerular basement membrane (GBM) formed by cellular secretions constitutes the glomerular filtration barrier (GFB) through which the acellular filtrate passes. It is permeable to water, small solutes and low molecular weight proteins but does not allow larger plasma proteins to pass through. Mesangial cells contribute to the generation of extracellular matrix (ECM) and control the availability of the surface area available for filtration by expansion and contraction mechanisms, regulated by systemic neural input and local neural input by a structure called the macula densa. This structure is an area of closely packed cells which line the wall of the distal tubule where the thick ascending limb of the Loop of Henle meets the distal convoluted tubule, proximate to the glomerulus. This specialised region senses and responds to osmolality fluctuations in the filtrate and acts upon mesangial cells by producing vasoconstrictive factors and activation of the juxtaglomerular apparatus. These cells produce rennin to active the rennin-angiotensin-aldosterone system (RAAS), a hormone system which regulates systemic blood-pressure and which is a significant contributor to diabetic nephropathy pathogenesis. Changes in diabetes lead to structural glomerular damage and dysfunction of specialised cells. Hyperglycemic conditions cause mesangial cell proliferation and hypertrophy and increased abnormal deposits of ECM which form visible lesions commonly characteristic of renal disease. In addition, GBM thickening and tubular basement membrane thickening also occurs. Podocyte injury leads to hypertrophy, detachment from vessel walls through loss of their foot processes, increasing the filtration gap in the filtration barrier and allowing larger molecules to pass through. Eventually apoptosis of podocyte cells occurs leaving the filtration barrier vulnerable. Deposits of plasma proteins accumulate in vessel lumen, and subepithelial deposits are also observed in the Bowman’s capsule and proximal renal tubules. As DN progresses the accumulation of these changes leads to sclerosis and eventually loss of the nephron. Continuous nephron loss leads to a need for renal replacement therapy and in some cases, transplant, in later stages of DN.

Diagnosis and Limitations Diagnosis of diabetic nephropathy through regular screening of diabetic patients involves assessment of a combination of urinary protein (proteinuria/albuminuria), estimated glomerular filtration rate (eGFR) and relevant clinical features such as diabetes duration, and the presence of co- morbidities. These methods have limitations and recent research has explored the use of alternative diagnostic markers. Urinary albumin and creatinine are measured to calculate a urine-albumin-creatinine ratio (UACR). Creatinine is also measured in serum and is used in various formulae to calculate eGFR. The most common of these formulae are the Modification of Diet in Renal Disease (MDRD) equation and the Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) which apply factors relating to age and ethnicity (Camargo et al., 2011). Studies have demonstrated the problematic use of these equations in T2DM. One study revealed both equations significantly underestimated eGFR when compared to measurements based on clearance of a dosed tracer, chromium-51 ethylenediaminetetraacetic acid ( ⁵¹ Cr-EDTA) (Silveiro et al., 2011). In addition, both were demonstrated to be less accurate in T2DM patients with GFRs > 60 ml/min compared with healthy volunteers (Camargo et al., 2011). However, comparison across a much larger cohort > 24,000 composed of patients with both T 1 DM and T2DM, revealed the MDRD formula to have a higher accuracy and least bias compared with several others (Schwandt et al., 2017). Proteinuria is also used as a diagnostic tool to detect renal damage through presence of urinary albumin in a spot sample. Low level UACR (30-300 mg/g) signifies microalbuminuria and is considered a risk factor for disease progression and a biochemical marker of DN. As kidney damage progresses to concentrations > 300 mg/g, macroalbuminuria is indicative of overt nephropathy and presents with eGFR decline (Marshall and Flyvbjerg, 2017). Limitations of this method of diagnosis is that the progression of micro-to-macro albuminuria is not definitive - some patients have demonstrated a return to normal albumin levels after being diagnosed with microalbuminuria. There are cases of non-proteinuric DN in which disease progresses to advanced stages with no significant increases in urinary albumin (Robles, Villa and Gallego, 2015). A cross-sectional study revealed that among diabetic patients with a low eGFR (< 60 ml/min/1 ,73m ² ) 26.7 % have a normal level of albumin in their urine. Therefore, the specificity and selectivity of this method has been challenged in recent years. The limitations in the sensitivity and specificity of conventional DN diagnostic methods are apparent and what is needed is more effective biomarkers of the early signs of DN, targeted to reflect biochemical and structural alterations in the nephron prior to albuminuria onset. References

Camargo, E. G., Soares, A. A., Detanico, A. B., Weinert, L. S., Veronese, F. V., Gomes, E. C. and Silveiro, S. P. (2011 ). The Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) equation is less accurate in patients with Type 2 diabetes when compared with healthy individuals, Diabetic Medicine, 28: 90-95.

Jiang, W., Wang, J., Shen, X., Lu, W., Wang, Y., Li, W., Gao, Z., Xu, J., Li, X., Liu, R., Zheng, M., Chang, Bai, Li, J., Yang, J. and Chang, B. (2020). Establishment and validation of a risk prediction model for early diabetic kidney disease based on a systematic review and meta-analysis of 20 cohorts, Diabetes Care, 43: 925-933. Kitagawa, Nobuko, Ushigome, E., Tanaka, T., Hasegawa, G., Nakamura, N., Ohnishi, M., Tsunoda, S., Ushigome, H., Yokota, I., Kitagawa, Noriyuki, Hamaguchi, M., Asano, M., Yamazaki, M. and Fukui, M. (2019). Isolated high home systolic blood pressure in patients with type 2 diabetes is a prognostic factor for the development of diabetic nephropathy: KAMOGAWA-HBP study, Diabetes Research and Clinical Practice, 158, p. 107920.

Marshall, S. M. and Flyvbjerg, A. (2017). Diabetic Nephropathy, in Holt, R. and Cockram, C. (eds) Textbook of Diabetes. 5th edn., John Wiley & Sons, Ltd, pp. 566- 579.

Robles, N., Villa, J. and Gallego, R. (2015). Non-Proteinuric Diabetic Nephropathy, Journal of Clinical Medicine, 4: 1761 -1773.

Saeedi, P., Petersohn, I., Salpea, P., Malanda, B., Karuranga, S., Unwin, N., Colagiuri, S., Guariguata, L., Motala, A. A., Ogurtsova, K., Shaw, J. E., Bright, D. and Williams, R. (2019). Global and regional diabetes prevalence estimates for 2019 and projections for 2030 and 2045: Results from the International Diabetes Federation Diabetes Atlas, 9th edition, Diabetes Research and Clinical Practice, 157.

Schwandt, A., Denkinger, M., Fasching, P., Pfeifer, M., Wagner, C., Weiland, J., Zeyfang, A. and Holl, R. W. (2017). Comparison of MDRD, CKD-EPI, and Cockcroft- Gault equation in relation to measured glomerular filtration rate among a large cohort with diabetes, Journal of Diabetes and its Complications, 31 : 1376-1383.

Silveiro, S. P., Araujo, G. N., Ferreira, M. N., Souza, F. D. S., Yamaguchi, H. M. and Camargo, E. G. (2011). Chronic Kidney Disease Epidemiology Collaboration (CKD- EPI) equation pronouncedly underestimates glomerular filtration rate in type 2 diabetes, Diabetes Care, 34: 2353-2355. Figures

Figure 1 Box plots representing participant serum H-FABP, serum L-FABP and urinary eGFR. Kruskal-Wallis significance between multiple groups is denoted by asterisks in the graph title. Mann Whitney U significance between two groups is denoted by asterisks above adjoining lines. *=p<0.05, **= p<0.01 , ***= p<0.001 , ****= p<0.0001.

Figure 2 Box plots representing participant serum H-FABP, serum L-FABP and urinary eGFR. Kruskal-Wallis significance between multiple groups is denoted by asterisks in the graph title. Mann Whitney U significance between two groups is denoted by asterisks above adjoining lines. *=p<0.05, **= p<0.01 , ***= p<0.001 , ****= p<0.0001.

Figure 3 Box plots representing participant serum H-FABP, serum L-FABP and urinary eGFR. Kruskal-Wallis significance between multiple groups is denoted by asterisks in the graph title. Mann Whitney U significance between two groups is denoted by asterisks above adjoining lines. *=p<0.05, **= p<0.01 , ***= p<0.001 , ****= p<0.0001.

Figure 4 Box plots representing participant serum H-FABP, serum L-FABP and urinary eGFR. Kruskal-Wallis significance between multiple groups is denoted by asterisks in the graph title. Mann Whitney U significance between two groups is denoted by asterisks above adjoining lines. *=p<0.05, **= p<0.01 , ***= p<0.001 , ****= p<0.0001.

Figure 5 Box plots representing participant serum H-FABP, serum L-FABP and urinary eGFR. Kruskal-Wallis significance between multiple groups is denoted by asterisks in the graph title. Mann Whitney U significance between two groups is denoted by asterisks above adjoining lines. *=p<0.05, **= p<0.01 , ***= p<0.001 , ****= p<0.0001.

Figure 6 Decision tree showing the use of sTNFRI , H-FABP and midkine for the stratification of diabetic nephropathy individuals (DiaNep), type 2 diabetic individuals (T2D) and healthy individuals (Con). MK=midkine, HFABP= heart fatty acid binding protein, sTNFRI = soluble tumour necrosis factor receptor I. Numbers following protein names and preceded by '<’ represent threshold values. SUMMARY OF THE INVENTION

The present disclosure describes methods and products for determining whether a patient has diabetic nephropathy or is at risk of developing diabetic nephropathy. A method is described of supporting the diagnosis of diabetic nephropathy in a patient who has type II diabetes or is suspected of having type II diabetes comprising determining the individual concentrations of two or more biomarkers chosen from sTNFRI , midkine, H-FABP, cystatin C, sTNFR2 and L-FABP in an in vitro sample of the patient and establishing the significance of the determined biomarker concentrations by comparing the measurements to a control. In a further aspect there is described a solid-state device, such as a biochip, which supports binding ligands to the biomarkers whose concentrations are to be determined. Also described is a kit comprising a solid-state device and detector binding ligands to the biomarkers, the detector binding ligands incorporating a detectable label. The analysis of multiple proteins in biological samples of three different age-matched cohorts, namely healthy patients, type 2 diabetes mellitus patients and diabetic nephropathy patients highlighted that sTNFRI , sTNFR2, midkine, cystatin C, H-FABP and L-FABP were at significantly greater concentrations in diabetic nephropathy patients. The concentration difference was evident in urine samples and blood samples.

DETAILED DESCRIPTION OF THE INVENTION

Although potential protein biomarkers of DN have previously been identified none have yet to be used in clinical practice and there is always a requirement for new diagnostic biomarkers; the current disclosure describes protein biomarker combinations that are able to identify type II diabetic patients who have diabetic nephropathy. Given the progressive nature of DN onset in T2D patients, the biomarkers can also be useful to support the identification of T2D patients likely to develop DN. It has been found that amongst multiple tested biomarkers a subset which are present in greater amounts in samples taken from individuals with type 2 diabetes mellitus can support the diagnosis of DN and thus facilitate the introduction of treatment measures before serious illness manifests.

In a first aspect is described A method of supporting the diagnosis of diabetic nephropathy in patient who has type II diabetes or is suspected of having type II diabetes comprising determining the individual concentrations of two or more biomarkers chosen from sTNFR I, H-FABP, midkine, cystatin C, STNFR2 and L- FABP in an in vitro sample of the patient and establishing the significance of the biomarker concentrations by comparison to a control.

Soluble tumour necrosis factor receptor I (sTNFRI) is an anti-inflammatory protein and is represented by Uniprot number P19438. Soluble tumour necrosis factor receptor II (sTNFR2) is an anti-inflammatory protein and is represented by Uniprot number P20333. Heart fatty acid binding protein (H-FABP) or FABP-3 is prevalent in heart muscle tissue, connected to intracellular long-chain fatty acid transport and is represented by Uniprot number P05413. Liver fatty acid binding protein (L-FABP) or FABP-1 is prevalent in liver tissue, connected to intracellular long-chain fatty acid transport and is represented by Uniprot number P07148. Midkine is a cytokine and growth factor and is represented by Uniprot number P21741 . Cystatin C (cystatin-3) is an inhibitor of lyosomal proteinases and is represented by Uniprot number P01034. By ‘determining’ or ‘determined’ is meant measuring or measured. By ‘individual concentrations of two or more biomarkers’ is meant that each of the two or more biomarker concentration measurements represent the amount of a single different protein i.e. if the amounts of the biomarkers sTNFRI and H-FABP are determined, then two measurement values will be obtained, one for sTNFRI and one for H-FABP. Protein is also referred to as analyte herein unless the context suggests otherwise. By ‘establishing the significance of’ is meant that the biomarker concentrations measured in the in vitro sample are used to infer whether the patient has diabetic nephropathy or is at risk of developing diabetic nephropathy. To support the described method the clinical history of the patient and other patient health data may also be used. The term ‘by comparison to a control’ implies by comparison to one or more control values. The control, also known as the reference value, can be a measurement of sTNF1 , H-FABP, midkine, cystatin C, STNFR2 and L-FABP derived from biological samples of a suitable individual or population. For example, a control population can be a healthy cohort of patients i.e. patients who are or who are considered to be disease free, a cohort of patients with T2D or alternatively, a mixed cohort incorporating healthy and T2D patients. The control values can be further stratified using various parameters such as gender, age and ethnicity. In one instance the control can be an individual or population who/that has no underlying condition or has a condition that is known not to affect the level of the biomarker indicative of DN; in a further instance the control can be an individual or population who/that is known to have T2DM. Although it is likely that the individual who’s biological sample is being tested will have been previously diagnosed with T2DM, the method can also be used to screen individuals who have T2DM but have not yet been diagnosed. To qualify as a biomarker of DN the amount of the biomarker in a DN patient differs from the control amount of the biomarker; in the current invention it has been established that the concentration or amount of sTNFRI , H-FABP, midkine, cystatin C, sTNFR2 and L-FABP in DN patients have been found to be greater than control values. A population is represented by a plurality of individuals, and the control value that is used for comparison can be any suitable statistical measure of central tendency such as a median value, mean value, percentile e.g. quartile value, 95 ^th or 99 ^th percentile. The control value can also be a threshold or cut-off value that can be incorporated in an algorithm such as a decision tree. A control value for a specific patient may also be a personal reference value; for example, if the patient has had two or more of the biomarkers sTNFRI , H-FABP, midkine, cystatin C, sTNFR2 and L-FABP measured on at least two different occasions, for example at timepoint Ti and timepoint T ₂ , timepoint Ti preceding timepoint T ₂ chronologically, then timepoint Ti may be used as the control value. Each of the patient’s biomarker measurement values can be directly compared to the corresponding individual biomarker values from the population, the population measurement values for each individual biomarker being a statistical measure of central tendency e.g. the mean or median or the comparison can utilise both the mean and median. As a hypothetical example, the measurement value of the patient’s sTNFRI obtained from a blood sample is compared to the mean and/or median measurement value(s) of sTNFRI calculated from 50 healthy individual sTNFRI measurements (the control or reference). If the patient’s sTNFRI value is more than the mean and/or median sTNFRI value(s) calculated for the 50 healthy individuals, then this would support a diagnosis of DN in the patient. If in addition to the patient’s sTNFRI concentration being lower than the mean and/or mean sTNFRI value(s) calculated for the 50 healthy individuals, the patient’s measured concentration of midkine is greater than the mean and/or median midkine concentration value(s) of the 50 healthy individuals, then this adds further support to a diagnosis of DN in the patient. This approach can be applied to the other analytes; a patient H-FABP concentration which is greater than mean and/or median reference H-FABP concentration value(s) supports the diagnosis of DN; a patient L-FABP value which is more than the mean and/or median reference L-FABP value(s) supports the diagnosis of DN; a patient sTNFR2 value which is more than the mean and/or median reference sTNFR2 value(s) supports the diagnosis of DN; a patient cystatin C value which is more than the mean and/or median reference cystatin C value(s) supports the diagnosis of DN. Generally, the more biomarkers supportive of a diagnosis of DN increases the likelihood that the patient or individual has DN. The method of supporting the diagnosis of DN in a patient can comprise measuring in an in vitro biological sample of the individual the concentration of the biomarker sTNFRI and at least one further biomarker chosen from midkine, H-FABP, L-FABP, sTNFR2 and cystatin C, wherein each biomarker concentration in the in vitro biological sample of the patient is compared to a reference value of the corresponding biomarker to establish whether there is an increase or decrease, and in which an increase in the concentration of each of the biomarkers in the patient sample compared to the corresponding reference value supports the diagnosis of DN. In a preferred aspect of the invention the method comprises measuring sTNFRI and midkine. In a further preferred aspect of the invention the method comprises measuring sTNFRI , midkine and H-FABP; it has been found that by measuring these three analytes in the method, a powerful diagnostic is achieved and individuals with T2DM and those individuals with T2DM who have DN can be discriminated. Other biomarker combinations chosen from sTNFRI , midkine, H-FABP, cystatin C, sTNFR2 and L-FABP can also discriminate T2DM individuals with DN from T2DM individuals, but the combinations require more complex statistical methodologies and/or algorithms to implement the diagnosis. Further biomarker combinations that can diagnostically discriminate T2DM patients and T2DM patients with DN and that can be used in the methods of the invention include i. sTNFRI , midkine and cystatin C ii. sTNFRI , midkine and STNFR2 iii. sTNFRI , midkine and L-FABP iv. sTNFRI , H-FABP and cystatin C v. sTNFRI , H-FABP and sTNFR2 vi. sTNFRI , H-FABP and L-FABP vii. sTNFRI , cystatin-C and sTNFR2 viii. sTNFRI , cystatin C and L-FABP ix. midkine, cystatin C and STNFR2 x. midkine, cystatin C and H-FABP xi. midkine, cystatin C and L-FABP xii. H-FABP, cystatin C and sTNFR2 xiii. H-FABP cystatin C and STNFR2 xiv. H-FABP, cystatin C and sTNFR2 xv. H-FABP, cystatin C and L-FABP xvi. cystatin C, L-FABP and sTNFR2 xvii. sTNFRI , midkine, H-FABP and sTNFR2 xviii. sTNFRI , midkine, H-FABP and cystatin C xix. sTNFRI , midkine, H-FABP and L-FABP xx. sTNFRI , midkine, H-FABP, sTNFR2 and L-FABP xxi. sTNFRI , midkine, H-FABP, sTNFR2 and cystatin C xxii. sTNFRI , midkine, H-FABP, L-FABP and cystatin C.

As an alternative to the use of a comparison of individual patient analyte measurements to a population analyte measurement of statistical central tendency, a suitable mathematical or machine learning classification model based on multiple variable analysis, such as a multiple logistic regression equation, can be derived from existing biomarker measurements of individuals and the derived model used to categorise an individual as having or not having DN. Further examples of multivariate techniques are decision trees, artificial neural networks, random forests, principal component analysis and support vector machine learning. These various multivariate statistical techniques are also referred to herein as ‘statistical methodologies’. When using a statistical methodology as part of the methods of the invention an output value is produced. The output value of the statistical methodology either supports a diagnosis of DN or does not support a diagnosis of DN. The word ‘amount’ is synonymous herein to ‘concentration’ or level’ and the words can be used interchangeably unless the context suggests otherwise. The terms participant, patient and individual are synonymous and can be used interchangeably unless the context suggests otherwise. All biological samples of the methods described have been obtained from the individual/patient prior to the analysis and measurement of the proteins i.e. the methods described use ‘in vitro’ or ‘ex vivo’ biological samples. The in vitro sample of the person that is analysed can be any suitable biological sample type including blood, plasma, serum, urine, saliva, cerebrospinal fluid, cell lysate and kidney tissue; the preferred sample types are blood, serum, plasma and kidney tissue. The most preferred sample type is serum. The aforementioned in vitro samples can also be obtained from any suitable mammal or can be a cell model system such as organoids or cultured cells. Any suitable technique can be used to detect and identify the protein target analytes including liquid chromatography, liquid chromatography-mass spectroscopy, and immunoanalytical techniques; for kidney tissue analysis mass spectrometry imaging (MSI) based on matrix-assisted laser desorption ionization (MALDI) coupled with Fourier-transform ion cyclotron resonance mass spectrometry or liquid extraction surface analysis-MSI (LESA-MSI) and liquid chromatography-mass spectrometry can be used. It is standard for the interpretation of biomarker results by an individual to be accompanied by the oversight of a medical professional or suitably trained individual as the concentration of a specific biomarker may vary in several different pathologies. Thus, the diagnosis of T2DM and DN using the described proteins can be supported by input from a medical professional or a suitably trained individual and can incorporate the use of further diagnostic measures including the measurement of further discriminant small molecule biomarkers ^approximately 1000 daltons) and macromolecular biomarkers e.g. proteins, polymeric carbohydrates.

The invention further describes a solid-state-device for use in the detection of diabetic nephropathy in a patient with type 2 diabetes mellitus, which supports three antibodies at discrete locations, one antibody specific to sTNFRI , the second antibody specific to midkine and the third antibody specific to H-FABP. The solid- state device can be a biochip, a microtitre plate, a nanoparticle, a slide or a bead system. A solid-state device may also be referred to as a substrate. Examples of bead materials are a single solid element such as carbon, silicon, silver, gold etc., an alloy, a mineral or a polymer or a combination of two or more materials. The antibodies engage with the substrate (i.e. supported by the substrate) by, for example, passive adsorption or, or more preferably, can be chemically bonded to the substrate attached by way of, for example, covalent bonds. Such covalent bonding generally requires the initial introduction of a chemically active compound covalently attached to the substrate surface prior to antibody addition, leading to the so-called chemically-activated surface. The antibody itself may also require the addition of a chemical activating group to achieve substrate bonding. The attachment of the antibodies to the chemically-activated surface requires that the binding characteristics of the antibodies are not affected, and that the specificity and affinity of the antibodies following the bonding process remain fit for purpose. The substrate is preferably of a planar conformation such as a glass slide, microtitre plate or a biochip. A biochip is the preferred substrate due to its stability and adaptability. A biochip is a thin, wafer-like substrate with a planar surface which can be made of any suitable material such as glass or plastic but is preferably made of ceramic. The biochip is able to be chemically-activated prior to antibody bonding or is amenable to the passive adsorption of antibodies. Various aspects of the biochip technology are described in EP0874242. A microlayer coating of material such as an ink composition can optionally be added to the planar surface of the substrate, preferably the biochip, prior to antibody placement, to increase the hydrophobic properties of the substrate and decrease non-specific binding of proteins in the biological sample. Either the upper surface or both surfaces of the substrate can be coated. The method of the invention preferably makes use of the sandwich immunoassay which proceeds by way of addition of the sample to be analysed followed by the binding of the biomarkers in the sample to antibodies (capture antibodies) adsorbed or bonded to the substrate surface, followed by addition of the detector antibodies. The further antibodies, the detector antibodies, promote or provide a detectable and measurable signal enabling the captured biomarkers to be detected and quantified. The detector antibodies are preferably bound to an enzyme, bioluminescent or radioactive compound which are commonly used detectable labels; in a preferred embodiment of the methods and products of the invention the detector antibodies comprise an antibody conjugated to a detectable label that is an enzyme. The signal can be any suitable electromagnetic radiation based on for example phosphorescence, fluorescence, bioluminescence, a radioactive compound, chemiluminescence (e.g. HRP/luminol system) etc. Preferably, chemiluminescence is used. An example of a detector antibody is an antibody conjugated to biotin (Ab-biotin) which subsequently binds to a streptavidin-biotin- enzyme complex to give Ab-biotin-streptavidin-biotin-enzyme; a further example is an antibody conjugated to biotin (Ab-biotin) which subsequently binds to a streptavidin-enzyme complex to give Ab-biotin-streptavidin-enzyme. Avidin can be used in place of streptavidin. A particularly preferred detector antibody is an antibody bound to the detectable label horseradish peroxidase (HRP) i.e. Ab-HRP. As with the capture antibodies, it is preferable that a detector antibody for an individual biomarker binds specifically to that individual biomarker without any measurable cross-reactivity to a different biomarker as this further ensures the specificity of detection of individual biomarkers. Alternatively, if the protein biomarkers to be detected have common epitopes, that is common amino acid sequences detected by an antibody, then a single detector antibody can be used. The capture and detector antibodies incorporated in products and methods of the invention can be any suitable antibody-based species including intact monoclonal antibodies or polyclonal mixtures comprising intact monoclonal antibodies, antibody fragments for example Fab, Fab’, and Fv fragments, linear antibodies single chain antibodies and multispecific antibodies comprising antibody fragments, single-chain variable fragments (scFvs), multi-specific antibodies, chimeric antibodies, humanised antibodies and fusion proteins comprising the domains necessary for the recognition of a given epitope on a target. Preferably, references to antibodies in the context of the present invention refer to polyclonal or monoclonal antibodies. As an alternative or in addition to an antibody species , other suitable binding ligands such as aptamers, molecular imprinted polymers, oligonucleotides and phages can be used in the methods and products of the invention.

In a further aspect of the present invention, there is provided a kit for use in the methods of the preceding claims, wherein the kit comprises a solid-state device supporting two or more binding ligands to detect two or more of the biomarkers sTNFRI , midkine, H-FABP, cystatin C, sTNFR2 and L-FABP; the solid-state device is preferably a biochip, and most preferably a ceramic biochip; the binding ligands are preferably (capture) antibodies each specific to one of the biomarkers and the binding ligands are bonded to the solid-state device surface. The kit preferably incorporates two or more binding ligands to detect two or more of the biomarkers sTNFRI , midkine, H-FABP, cystatin C, sTNFR2 and L-FABP; the binding ligands are preferably (detector) antibodies each specific to one of the biomarkers, the detector antibodies incorporating a detectable label.

Methods

Participant Cohort This study was ethically approved by the Ulster University School of Biomedical Sciences Filter Committee, the London - Chelsea Research Ethics Committee and the Northern Health and Social Care Trust (NHSCT). In accordance with Ulster University research and governance guidelines, all electronic data was stored in an access-restricted folder on a shared drive within the university and all hard data copies were stored in a clearly labelled folder in a locked office. Three groups of participants were recruited as follows: Group 1 : Non-DM (control group), Group 2: T2DM, Group 3: DN participants (T2DM) Participants were over 18 years of age with no autoimmune diseases (e.g. HIV/AIDS) or any other diagnosed condition or illness that may affect the kidney (e.g. long-lasting Hepatitis B/C, polycystic kidney disease, kidney stones).

Participant Recruitment Recruitment of control (group 1) participants was carried out through email circulation within Ulster University and via advertisement in the newsletter of a local organisation, University of the Third Age (U3A). T2DM and DN participants (Group 2 and 3, respectively) were identified and approached by members of the NHSCT clinical care team as they attended routine clinic appointments. Prospective participants received detailed information on the study and if they wished to take part, verbal consent to an appointment was obtained, to be carried out at their next routine visit to the clinic.

Data Collection /Appointment Appointments for control (group 1) participants took place at Ulster University within the Human Intervention Studies Unit (HISU). Appointments for T2DM (group 2) took place at the Whiteabbey Hospital Diabetes Clinic (NHSCT clinic) and DN (group 3) appointments were spread between the Whiteabbey Hospital Diabetes Clinic, for patients in earlier stages of DN and the Renal Unit at Antrim Area Hospital (NHSCT clinic) for participants in late-stage DN. Informed, written consent was obtained for all data and sample collection. Participants were asked to complete a short questionnaire to obtain information on medical history and lifestyle. Anthropometric measurements of height (cm), weight (kg) and blood pressure (mmHg) were taken. For T2DM and DN participants. For the control group, these measurements were taken according to Ulster University HISU protocols (HISUEQUIP SOP 002 and CLN/WI004/001) and measured using digital scales, a stadiometer and a portable blood pressure monitor within the HISU facility. Blood pressure measurements were taken immediately following the questionnaire. A reading was taken from each arm and the arm with the highest reading used for further measurements. An average measure of mmHg was determined from two or more separate readings which did not vary more than 5 mmHg. An average of two readings was taken for weight and height measurements.

Sample Collection and Processing Blood samples were obtained and processed according to the Ulster University protocol for phlebotomy (HTA SOP 003) and the Sample Collection, Processing and Transport protocol (Protocol 002). For DN participants who were receiving dialysis, the blood sample was obtained from the dialysis line by nursing staff on the renal unit. For all other participants blood samples were collected by venipuncture using a 22-gauge (G) butterfly needle. 2 x 10 ml serum vacutainers (red tops) and 3 x 10 ml EDTA (purple tops) (plasma) were obtained (50 ml total). In the case of dialysis participants, this was reduced to 1 x 10ml serum and 2 x EDTA vacutainers (30 ml total). After collection, vacutainers were inverted ~ 10 times and placed at 4 °C for processing. Urine samples for T2DM and DN participants were obtained using kidney trays and transferred to a sterile container: universal tubes (Analab Ireland, Scientific Laboratory Supplies Ltd, Lisburn, UK) or a 10 ml urine aspirator (Urine Monovette, Sarstedt, Numbrecht, Germany). For control groups, the participant was supplied with a sealed sterile foil bowl (Medisave UK, Ltd, Weymouth, UK). The sample was then transferred to a sterile universal tube. All urine was stored at 4 °C before processing. Serum tubes were left at room temperature to clot for 30-60 min. Serum and EDTA vacutainers were then centrifuged for 10 min at 4 °C and 1 ,300 g. The supernatant was transferred to a pre-labelled polypropylene tube on ice using a Pasteur pipette. Aliquots (500 pl) were transferred to pre-labelled cryovials on ice within 1 h of centrifugation. The urine sample container was transferred to ice. Aliquots (500 pl) were transferred to pre-labelled cryovials also on ice. All serum, plasma and urine aliquots were placed upright into a pre-labelled cryobox and stored at -80°C until time of analysis.

Point of care assay Aution sticks (10AE) were used for dipstick urinalysis and were interpreted using a PocketChem analyser from Arkray Inc., Japan, according to manufacturer’s instructions.

Bradford Assay (Urinary Protein) Total urinary protein levels (mg/ml) were determined, in duplicate, by Bradford assay (Pierce, Rockford, IL, USA) using a stock solution of BSA (Sigma) as standard (2 mg/ml). Patient urine samples (10 pl/patient), after centrifugation (1200g, 10 minutes, 4°C), were mixed with Bradford reagent (1 ml) and allowed to stand for 5 minutes. The samples were read at room temperature on a Hitachi Spectrophotometer (Model No. U-2800) at a wavelength of A ₅ 9 ₅ nm. Total urinary protein (mg/ml) was determined from a BSA calibration chart (1 to 5 mg/ml).

Osmolality Urine osmolality (mOsm) was determined, in duplicate, for each study participant using a Loser Micro-osmometer according to manufacturer’s instructions (Loser Messtechnik, Berlin, Germany).

Biomarker Measurements Biochip Array Technology (BAT) obtained from Randox Laboratories Ltd (Crumlin, Northern Ireland, UK) was used according to the provided kit instructions for the simultaneous detection of multiple biomarkers from a serum and/or urine patient sample for the following proteins (abbreviations and limit of assay detection are shown in parenthesis): liver-type fatty acid-binding protein-1 (L- FABP) (1.06 ng/ml), macrophage inflammatory protein-1 alpha (MIP-1 a) (0.68 pg/ml), soluble tumour necrosis factor receptor 1 (sTNFRI) (0.05 ng/ml) and sTNFR2 (0.01 ng/ml) were analysed on a Chronic Kidney Disease 1 (CKD1) biochip array. C-reactive protein (CRP) (500 ng/ml), C3a des Arg (C3DA) (30 ng/ml), neutrophil gelatine-associated lipocalin (NGAL) (3.52 ng/ml) and adiponectin (ADP) (39.06 ng/ml) were analysed using the Chronic Kidney Disease 2 (CKD2) biochip array. Interleukin (IL)-2 (2.97 pg/ml), IL-4 (2.12 pg/ml), IL-6 (0.12 pg/ml), IL-8 (0.36 pg/ml), IL-10 (0.37 pg/ml), IL-1 a (0.19 pg/ml), IL-1 (0.26 pg/ml), vascular endothelial growth factor (VEGF) (3.24 pgp/ml), interferon-y (IFNy) (0.44 pg/ml), monocyte chemoattractant protein-1 (MCP-1) (3.53 pg/ml), epithelial growth factor (EGF) (1.04 pg/ml) and tumour necrosis factor a (TNFa) (0.59 pg/ml) were measured using the high-sensitivity Cytokine (hs-CTK) biochip array. Clusterin (10.82 ng/ml), cystatin C (0.87 ng/ml), NGAL (0.4 ng/ml) and kidney injury molecule 1 (KIM-1) (54.6 pg/ml) were analysed in urine only using the Acute Kidney Injury (AKI) biochip array.

Other biomarker assays The following biomarkers were measured in serum; transferrin (0.08 g/l), microalbumin (5.11 mg/l), heart-type fatty acid-binding protein (H-FABP) (2.94 ng/ml), cystatin C (0.4 mg/l), albumin (3.2 g/l), total antioxidant status (TAS) (0.21 mmol/l), urea (0.51 mmol/l) and creatinine (11.4 pmol/l) using the Imola analyser (Randox Laboratories Ltd, Crumlin, UK), according to the manufacturer’s instructions. HDL cholesterol (HDL) (0.189 mmol/l), LDL cholesterol (LDL) (0.189 mmol/l), total cholesterol (0.865 mmol/l), triglycerides (22.9 mg/dl) and creatinine (311 pmol/l) were measured on a Daytona analyser (Randox Laboratories Ltd, Crumlin, UK). Insulin (2.78 pmol/l) was measured using a Cobas e801 analyser (Roche, Basel, Switzerland). Microalbumin and creatinine were also measured in urine.

Midkine ELISA Urine and serum midkine concentrations were measured using a commercial ELISA from LyraMid, Sydney, Australia, according to the manufacturer’s instructions.

Statistical Analysis Statistical analysis was undertaken using R [31], Data below the limit of detection (LOD) or mean detectable dose (MDD) was inputted as 90% of the LOD or MDD [28], Biomarker data were analysed using Kruskal-Wallis to identify which factors were differentially expressed between control, T2D and DN groups. Statistical significance was taken at the p<0.05 level and results are presented as mean ± SD where appropriate. Spearman’s rho correlations were also performed. Correlations > 0.7 were considered significant. sTNFRI and sTNFR2 exhibited correlation coefficients of >0.90 in serum and urine samples of each of the study cohorts and sTNFRI can be replaced by sTNFR2 in the diagnostic biomarker combinations exemplified herein.

Results

Assignment of kidney disease was based on eGFR function and creatinine levels. Dip Stick Urinalysis Results: urinary glucose, pH and protein levels were significantly different across groups.

Serum Biomarkers: in total, 34 serum biomarkers were investigated and 25/34 (73.5%) were significantly different across patient groups (Kruskal-Wallis) (Table 1). Urine Biomarkers: in total, 25 urine biomarkers were investigated and 13/25 (52.0%) were significantly different across patient groups (Kruskal-Wallis) (Table 2). Biomarkers that Differentiated Control from T2D, and T2D from DN Subjects The AUROC for each of the biomarkers that differentiated control from T2D, and T2D from DN subjects, are described in Table 3. The study investigated biomarkers which were differentially expressed in patients with T2D and DN. Biomarkers with an AUROC >0.84 were considered significant. Biomarkers which met this criterion included creatinine, cystatin C, H-FABP, midkine, NGAL, sTNFRI and sTNFR2. Serum H-FABP and L-FABP increased across groups with the most significant increase observed between T2D and DN patients (Figure 1A and 1 B). eGFR levels were not significantly different between control and T2D however, for T2D versus DN, the eGFR levels were decreased almost 70% (Figure 1 C). Urine and serum sTNFRI , sTNFR2 and midkine exhibited the same pattern observed for H-FABP (Figure 2). Urine midkine levels were significantly different between control and the other groups. Urine midkine levels in T2D were not significantly different from DN patients. Serum midkine levels were significantly higher in the DN patients when compared to T2D, which may suggest that systemic levels of midkine are increasing because of kidney dysfunction, as noted by the decline in eGFR.

Table 1 serum biomarkers (mean ± standard deviation) Table 2 urine biomarkers (mean ± standard deviation)

Table 3 AUROC for control vs. T2D and T2D vs. DN