Novel biomarkers for progression of chronic kidney disease

CHARACTERISTICS OF THE PROGRESSION OF CHRONIC KIDNEY DISEASE (CKD) Although there are different initiators of CKD, it is generally recognized that the secondary pathological pathway is quite common to all CKD. CKD may inevitably progress to end stage renal disease (ESRD) due to a vicious cycle of nephron destruction by progressive glomerulosclerosis and tubulointerstitial fibrosis. The chronic processes include kidney resident cell activation, secretion of proinflammatory cytokines and infiltration of inflammatory monocytes/macrophages and T cells. Subsequently, reactive oxygen species, fibrogenic and inflammatory cytokines in turn, stimulate the residential renal cells to undergo phenotypic activation or transition capable of producing excessive extracellular matrix (ECM). Then continuous deposition of ECM leads to the collapse of normal renal structure and the loss of kidney function.1,2 It is, therefore, concluded that the progression of CKD involves a complex interaction between resident renal cells and infiltrated cells mediated by a complex molecular network. CHALLENGES FOR CURRENT MONITORING PROGRESSION OF CKD The stages of CKD are presently defined on the level of estimated glomerular filtration rate (eGFR) and the presence of an elevated urinary albumin excretion according to the National Kidney Foundation Practice Guidelines.3 However, neither of markers can allow an accurate assessment of kidney function in healthy individuals and in CKD patients and are probably not ready for monitoring of disease progression.4 As for GFR, it is cumbersome to perform and because of the adaptation of remnant nephrons, significant loss of GFR may not indicate a clinical manifestation. Furthermore, current creatinine-based estimation equations have been reported to be less accurate in patients with inflammation or in older people with frailty or comorbid illnesses.5 And also one should realize that having a declined GFR does not directly indicate the subject suffering from progressive renal disease. The urinary albumin excretion is another classic biomarker for the progression risk in CKD, but has clear limitations. Many factors like diet, fever, hydration status, and physical activity may affect the level of albuminuria. And it is also limited because of the variability of protein levels in the individual's urine over time. The rate of progression shows considerable inter-individual variability and is influenced by multiple factors. An effective surveillance biomarker for CKD in addition to the classic ones was therefore in urgent need. It would enable us to early find CKD and more importantly, monitoring the progression of CKD, and finally reduce the prevalence of ESRD.6 NOVEL BIOLOGIC MARKERS FOR PROGRESSION OF CKD In recent years, two principle approaches have been applied in biomarker research: (1) the candidate biomarker approach that test the molecules involved in the pathogenesis process as stated above; (2) unbiased screening approach. Clinical samples are screened with unbiased proteomic approach to identify the differential proteins as potential markers. Tremendous clinic and basic studies have given birth to a steadily increasing panel of molecules that may serve as biomarkers for CKD. Compared with traditional markers, the novel additional biomarkers might lead to a better understanding and predicting of the progression of CKD.7 The US Food and Drug Adminstration and European Medicine Agency (EMEA) have recently quantified seven new nephrotoxicity biomarkers, kidney injury molecule-1 (KIM-1), albumin, total protein, beta2-microglobulin, cystatin C, clusterin, and trefoil factor-3. Scientists believe that these seven biomarkers may provide important advantages over the blood urea nitrogen (BUN) and creatinine tests. The novel biomarkers which might hold a great promise in the near future are addressed below. KIM-1 There is a growing body of evidence that support KIM-1 as specific marker for kidney tubular injury. KIM-1 expression is increased in various renal diseases and is primarily detected in areas of fibrosis and inflammation.8 A more recent study demonstrated that lowering urinary KIM-1 reflects amelioration of proteinuria-associated tubulointerstitial damage. And KIM-1 concentrations decreased in parallel with decreases in proteinuria as a result of treatment.9 But at the present time, more investigations are still required before its translation to clinical practice. Moreover, a reproducible measurement of the biomarker across different laboratories was needed for accelerating the transition. Fortunately, the commercially available assay kit for KIM-1 has recently been demonstrated a robust assay for detecting KIM-1 in urine.10 Neutrophil gelatinase-associated lipocalin (NGAL) Recent data indicate that NGAL is not only predictive for accurate kidney injury (AKI), but also for the progression of CKD. A study in 96 patients with nonterminal stage 2–4 CKD found that both serum and urinary NGAL levels were directly associated with the progression of CKD over a median follow-up of 18.5 months, independently of age and GFR at baseline.11 A pilot study in 78 patients with CKD showed that baseline urine NGAL levels correlated strongly with changes in serum creatinine and GFR.12 NGAL might also hold a promise in monitoring the status and treatment response for various renal diseases. Kuwabara et al13 showed that patients with nephrotic syndrome or interstitial nephritis had markedly elevated urinary NGAL levels that decreased in response to successful treatment. Urinary liver-type fatty acid binding protein (urinary L-FABP) Urinary L-FABP is another promising biomarker that has been demonstrated to be able to accurately reflect the tubulointerstitial damage and was useful as a real time indicator.14 A study investigating L-FABP levels in patients with diabetic nephropathy showed that urinary L-FABP was associated with the severity of disease.15 Cytokines Among those complex cytokine network involved in CKD, connective tissue growth factor (CTGF) and transforming growth factor β (TGFβ) has emerged as potential markers for progression of nephropathy. Suthanthiran and colleagues examined whether there was an association between serum levels of TGF-β1 and the risk factors for progression of CKD in 186 black and 147 white adults. Serum TGF-β1 protein levels was positively associated risk factors for the progression of CKD in the African-American population.16 In a prospective long-term (12.8 years) follow-up study (198 type 1 diabetic patients with established diabetic nephropathy and 188 type 1 diabetic patients with persistent normoalbuminuria), it was clearly demonstrated that plasma CTGF contributes significantly to prediction of ESRD and mortality in patients with type 1 diabetic nephropathy.17 Urine mRNAs More recent studies with urinary mRNA quantification have indicated a novel strategy in exploring biomarkers for kidney disease.18 Podocyte mRNAs (nephrin and podocin mRNA) excretion in urine was suggested to be a potential clinical tool for the diagnosis and monitoring of glomerular diseases.19 Although urine mRNA detection is simple and non-invasive, it requires further research to define its role in routine clinical practice. Multiplex signatures for CKD detection Given the complexity and multiple pathologic mechanisms of CKD as described above, finding a single biomarker with sufficiently high sensitivity and specificity is obviously a challenging issue. The proteomic technology, on the other hand, offers a promising opportunity to identify a panel of disease biomarkers for CKD. Technological platforms that are used in the proteome-wide research include mass spectrometer (MS), 2-D electrophoresis, high performance liquid chromatography (HPLC), capillary electrophoresis and protein microarray, and so on.20 Many investigators have defined various protein patterns as diagnosis biomarkers of different kidney diseases using those approaches. Urinary proteomics has recently gained a wide interest and has been extensively applied to various primary glomerular diseases, including IgA nephropathy, membranous nephropathy, focal segmental glomerulosclerosis, and so on.21 2-D difference in gel electrophoresis (DIGE) was applied to define IgA nephropathy-specific biomarkers in the urine; comparing between IgA nephropathy (n=17) and healthy controls (n=10), they observed increased levels of several urine proteins in IgA nephropathy.22 A recent study using capillary electrophoresis MS (CE-MS) on 3600 samples from 20 different centers established a database of more than 5000 urinary peptides. This database was used to define biomarkers for CKD and the validation in a blinded and independent cohort of 134 individuals yield 88% sensitivity and 100% specificity.23 Knickerbocker et al24 demonstrated that the combination of multiple cytokines detected by protein microarray with clinic variables could predict the early mortality in patients initiating kidney dialysis. In a study with 305 individuals, biomarkers were defined and validated in blinded data sets using high-resolution capillary electrophoresis coupled with electrospray-ionization mass spectrometry. A panel of biomarkers was identified being able to detect diabetic nepthropathy and may provide prognostic information.25 CONSIDERATION OF KEY ISSUES IN IDENTIFYING BIOMARKERS Despite the long list of promising biomarkers discovered for CKD, the translation to clinic application is quite limited. We suggest exploring those potential biomarkers through different phases of biomarker study that would consequently indicate different levels of evidence. Five phases were proposed for biomarker studies, i.e. preclinical exploratory study, clinical assay development for clinical disease, retrospective longitudinal repository studies, prospective screening study and disease control studies.26 However, different phase design may be applied according to different research purpose and the available information from previous studies about the candidate biomarkers. And results for each phase should be subsequently subjected to validation that is the process of confirming the conclusions, statistical as well as biological. The development of statistical algorithms for discovering promising biomarkers from the large amount of data becomes an active area of research, especially for proteomic study. Data preprocessing is the first step in analysis, which includes normalization of the data to remove systematic variation, transformation of data and data filtering. Normalization is an important preprocessing step, especially for microarray experiment.27 A number of normalization methods were proposed in previous studies, for example, adding a control reagent to the sample, printing identical control features on each array and then to calculate their median value and use this to normalize data between arrays. Different normalization method may be needed for different experiment platforms; an optimal one can be chosen with the criteria of reproducibility between replicate data sets. The advancement of proteomic techniques has generated many analytic challenges, among which is to discover "signature" protein profiles specific to each pathologic state (e.g. normal vs. cancer) from high-dimensional data. Analytical tools for disease classification have included threshold correlation coefficients, k-nearest-neighbors, support vector machines, principle component analysis and hierarchical clustering. In proteomics studies, the number of samples is usually low compared to the number of variables which consequently leads to overfitting: that is the created model is specific for the training data and does not generalize well to new set of samples. To avoid overfitting, samples are usually split into training and test sets and used the test set to provide independent estimates of classification errors for the algorithm derived from the training set. Alternatively, one may consider a cross-validation of the entire samples.28 The effects of sample collection, processing and storage and the inherent properties of the provider of the sample such as age and gender are all the critical preanalytical variables that are often overlooked. Implementing standard operating procedures for sample preparation for the different groups is a necessity to minimize bias. Fortunately, the Human Kidney and Urine Proteome Project (HKUPP) will soon release standard protocols and guidelines for proteome analysis of non-proteinuric urine.29 Finally, it will be important in future studies to validate the sensitivity and specificity of promising biomarkers in clinical samples from large cohorts and from multiple clinical situations.