**1. Introduction**

A biomarker is defined, by a collaborative working group involved with both the United States National Institutes of Health (NIH) and the Food and Drug Administration (FDA), as "a characteristic that is measured as an indicator of normal biological processes, pathogenic processes, or responses to an exposure or intervention, including therapeutic interventions" [1]. This working group definition was formed on the initiative of the NIH with the aim of accelerating the development and clinical application of reliable biomarkers based on shared definitions. Indeed, an "ideal" biomarker is defined with the presence of some analytic features: (1) it should be measured and readily available in biological samples, such as blood or urine; (2) it should be reproducible, non-invasive, and not expensive [2]. In addition, several clinical features should also be provided to complete the biomarker's definition; it needs to allow for an early detection of a disease status, while it also needs to have high sensitivity and specificity, i.e., the biomarker needs to differentiate the pathologic status from the normal one and from other clinical conditions, as accurately as possible [3]. The effort made by the NIH–FDA working group is considerable ever since it forecasted and tried to solve the problems of biomarker development, from discovery to clinical application. Indeed, once a biomarker is found to be involved in one or more pathophysiological mechanisms of a disease, it may be introduced into clinical practice to see whether it offers advantages in clinical management and after completion of the validation phase that is considered a crucial step [4]. An important example of the pitfalls of biomarker development is illustrated in the chronic kidney disease (CKD) scenario. CKD is a chronic disease characterized by a poor prognosis, due to the strong association with the development of cardiovascular events, all-cause mortality, and renal events such as renal replacement therapies (RRT, i.e., dialysis or kidney transplantation) [5,6]. The Kidney Disease Improving Global Outcomes Work Group (KDIGO), in 2012, defined CKD with the presence of either decreased kidney function (estimated glomerular filtration rate (eGFR) < 60 mL/min/1.73 m2) and/or albuminuria, namely, an abnormal amount of protein excretion with urine, for at least three months [7]. The global dimension of the disease is so important that CKD started to be considered a relevant public health problem. Indeed, according to the Global Burden of Kidney Disease, the CKD incidence and prevalence increased by 88.76% and 86.95%, respectively, from 1990 to 2016 [8]. Moreover, the mortality attributed to CKD increased by 41.5% between 1990 and 2017, a percentage that exceeded the mortality due to several neoplasms or cardiovascular (CV) disease [9]. Hence, great effort is advocated toward improving clinical decision-making and reinforcing treatment and prevention of CKD. The KDIGO working group proposed a classification called "CGA" that incorporates the cause (C) of kidney disease, as well as the eGFR (G) and albuminuria (A) levels, to stratify risk in patients with CKD and to better address the importance of the underlying disease. However, considering eGFR and albuminuria levels in cross-tabs, like those reported in KDIGO guidelines, does not take into consideration the complex mechanisms of CKD. In fact, all patients are stratified on the basis of eGFR and albuminuria and allocated in risk categories for prognostic estimations. This approach was also defined as "reductionist", since it does not consider the different etiologies of CKD and many other parameters, including serum or urine biomarkers, which could help clinical management of CKD [10]. Moreover, although the KDIGO also suggested considering the causes of kidney disease to predict poor outcomes in CKD patients, how to incorporate these parameters remains unclear [11]. Few studies included renal diagnoses in risk prediction models and, when diagnoses are present, they are classified in different ways (i.e., four, five, or even six categories) with large heterogeneity between studies, thus making an univocal interpretation difficult [12,13]. There is also a growing debate regarding the question whether an eGFR reduction below the threshold of 60 mL/min/1.73 m<sup>2</sup> represents the consequence of a physiologic senescence or a marker of renal pathology [14,15]. In fact, despite the evidence that eGFR reduction foresees both the onset of end-stage kidney disease (ESKD) and the all-cause death regardless of age, in large general and high-risk populations, other human studies showed that kidneys undergo structural and functional change with aging such as nephrosclerosis and a reduction in measured GFR [14,16]. At the same time, these changes are not associated with a reduction in single-nephron GFR; this is the likely reason why, in the absence of albuminuria, the GFR reduction alone determines only a small increased risk for age-standardized mortality and ESKD [16]. Owing to these important controversies and in an attempt to improve risk stratification and care of CKD patients, as well as to share clinical findings with the nephrology community, the International Society of Nephrology (ISN) started an international project called "closing the gaps", which encompasses a set of activities that have to be performed to improve the prognosis of CKD patients [17]. The core of this project is

about the development, implementation, and clinical application of novel biomarkers in CKD patients. It indeed emerged that, with the exception of cystatin C, which is a filtration marker useful to estimate the eGFR and to improve risk prediction in CKD, the vast majority of previously tested biomarkers in CKD did not reach any clinical application [18]. The aim of this review is to report the principal pieces of evidence regarding biomarker development in CKD, the contribution of these biomarkers to both observational and interventional (i.e., randomized controlled trials) studies, and the possible strategies that could be followed to ameliorate this important branch of clinical research.
