Background
Chronic kidney disease (CKD) is a progressive condition characterized by a gradual decline in kidney function, which can result in end-stage renal disease (ESRD) [
1]. The global prevalence of CKD is estimated to be between 8% and 16% [
2], and is thought to exceed 50% in certain high-risk populations [
3]. As of 2013, in the United States alone, more than 30 million adults were projected to have CKD, with the incidence rising profoundly among those aged 65 years and older [
4].
The etiology of CKD is multifactorial, with glomerular hypertension and hyperfiltration reflecting the most prominent mechanistic contributors to disease progression [
5]. Both systemic hypertension and glomerular hypertension resulting from glomerular hemodynamic changes, are known to provoke injury to the glomeruli. As a consequence, elevated blood pressure can overwhelm normal protection afforded from systemic hypertension to the kidney by autoregulation [
5]. Hyperfiltration initiates the renin angiotensin aldosterone system (RAAS), which in turn, increases glomerular permeability and gives rise to albuminuria, proteinuria, and dyslipidemia, while also diminishing glomerular filtration rate (GFR), thus reciprocally leading to hypertension.
Concurrent with hypertension, which occurs in approximately 65% of the diabetic population [
6], hyperglycemia and genetic predisposition are key factors in the development of kidney disease and in its progression [
7]. Diabetic kidney disease (DKD) is typically characterized by persistent albuminuria, increasing serum creatinine and a progressive decline in estimated GFR (eGFR). Over time, worsening DKD is associated with increased risk of cardiovascular (CV) and cerebrovascular events, as well as renal morbidity and mortality [
7]. Emerging data also suggest uric acid may be a potential marker of CKD [
8]. Routine assessment of these key biomarkers is an important dimension of preventive medicine, as it helps to identify those who might benefit from earlier intervention to lower the risk of adverse outcomes.
Although increasing urinary albumin excretion and serum creatinine levels, along with diminished eGFR, are likely important markers of kidney decline, it is unclear whether these biomarkers reflect independent risk factors for kidney disease progression, subsequent outcomes in patients with type 2 diabetes mellitus (T2DM), or both. By extension, numerous pathophysiological mechanisms are implicated in DKD, and in this context, kidney disease progression in patients with T2DM could possibly be delayed or prevented by controlling other measures such as blood glucose or blood pressure levels [
9‐
12,
7,
13]. Indeed, several randomized controlled trials have demonstrated that medications indicated for treatment of T2DM such as glucagon like peptide-1 (GLP-1) [
14,
15] and sodium-glucose co-transporter 2 (SGLT-2) inhibitors [
16,
17] can slow renal disease progression as well as reduce adverse CV outcomes. Other complementary therapeutic approaches such as utilizing selective endothelin receptor type A (ET-A) antagonists may also help delay renal function decline and subsequently lower the risk of related adverse renal outcomes [
18]; albeit the precise beneficial mechanisms conferred from these medications are still undergoing investigation [
19].
In light of these findings, a better understanding of the prognostic utility of DKD biomarkers and whether they might be implicated in increasing the risk of adverse renal outcomes in patients with T2DM, would help clarify their clinical value, and guide treatment decision-making in everyday clinical practice. The objective of this study was to conduct a systematic literature review (SLR) of the published evidence documenting the role of albuminuria, serum creatinine, eGFR, and uric acid in predicting kidney disease progression and cardio-renal outcomes among patients with T2DM.
Methods
Databases and search strategy
Embase, MEDLINE, and Cochrane Central Trials Register and Database of Systematic Reviews and other Cochrane Library assets were searched for human studies published in English between January 2000 and May 2016. Additionally, we performed a PubMed search from 2013 until May 2016, in an effort to capture more recently published studies that may not yet have been indexed in one of the other databases. Implementation and reporting of the present SLR adhered to the guidelines and standards advocated by the National Institute for Health and Care Excellence (NICE) and Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines [
20]. Although review papers identified in the searches were omitted from this study, the reference lists for each identified review article were hand-screened to include any relevant referenced publications (
n = 2) that met study inclusion criteria and had not been identified during the database searches.
The data sources above were searched to identify specific observational or non-randomized studies relevant to renal biomarkers and the need for renal replacement therapy. For the current SLR, we adopted the Population, Intervention, Comparator, Outcomes, and Study type (PICOS) criteria to establish our search strategy and inclusion/exclusion criteria. Relevant studies were subsequently identified using text word or MeSH headings, including but not confined to: population terms such as T2DM and non-insulin dependent diabetes; condition/disease outcome terms including renal or kidney replacement or transplantation, renal or kidney failure, chronic renal or kidney disease, end-stage renal or kidney disease; and biomarker/clinical measure terms including GFR, albuminuria, proteinuria, and creatinine.
Only full-text publications were included; congress abstracts were excluded. Studies of behavioral, educational, or clinical practice interventions were included. All review studies (expert, narrative, and systematic reviews, as well as meta-analyses), preclinical studies, pharmacovigilance/safety studies, case reports, and studies that employed a drug or device intervention were excluded from the SLR.
Screening, critical appraisal, and quality assessment
At the first-level screening, one reviewer screened the titles of publications identified from the search according to the eligibility criteria. A second reviewer then performed a quality check of 10% of all screened publications. Where discrepancies existed in a screening decision among the 10% of publications, the second reviewer examined all of the publications that had been excluded based on the same reason as the study marked with the discrepancy. At the second-level screening, one reviewer screened the abstracts, and if necessary, full texts of publications retained from the first screening stage against the study eligibility criteria. A second reviewer then performed a quality check of 20% of all screened studies. In the event of a discrepancy, the second reviewer was tasked with checking all studies that had been excluded for the same reason as the study marked with the discrepancy. Any remaining discrepancies regarding study inclusion were resolved by a third reviewer.
Publications of longitudinal studies meeting eligibility criteria that described the role of albuminuria/proteinuria or serum creatinine/uric acid/eGFR in renal disease progression were included in this review. Although uric acid is a less well established potential biomarker of CKD risk, we examined it in this SLR with creatinine/eGFR. All studies meeting inclusion criteria were assessed for quality using the Downs and Black (D&B) instrument [
21], which is suitable for non-randomized longitudinal clinical and observational studies. The D&B instrument evaluates (a) the methodological quality (10 items), (b) statistical power (one item), (c) bias (seven items), (d) confounding (six items), and (e) external validity of the study (three items).
Study outcomes
The primary outcomes examined in this study were renal disease progression, ESRD, non-fatal CV events, and all-cause mortality. Renal disease progression reflected any mention of GFR loss, a steep decline in eGFR, at least a 50% decline in eGFR, a doubling in serum creatinine, or nephropathy progression. Non-fatal CV events comprised non-fatal myocardial infarction or non-fatal stroke.
Statistical considerations
Key data were extracted from each of the selected publications, where available, and included: study type (e.g. retrospective or prospective); geographic region where the study was conducted; sample size; baseline patient demographics, clinical characteristics and reported medication use; clinical indicators and outcomes of kidney decline and disease progression; follow up duration; and time to renal mortality, CV mortality, all-cause mortality, or receipt of any renal replacement therapy (RRT) including chronic dialysis or transplantation. Descriptive reporting included tabulations of study design and patient characteristics. Risk estimates were extracted from the included publications. These primarily were hazard ratios, and in some cases relative risks (RR), with 95% confidence limits (where provided) for the association between renal biomarkers and outcomes of interest. Risk estimates reported in the publications were assessed qualitatively by visual inspection for statistical significance as well as directionality (e.g. an increase or reduction in the risk of an outcome of interest according to a particular biomarker). Due to the variation of study biomarker predictors, covariates, outcomes, statistical methods, and risk estimates used in the included publications, quantitative comparison of risk estimates across studies was not possible. Data extracted from the included publications were reported, tabulated, and descriptively assessed in Microsoft Excel 2016 worksheets (Microsoft Corporation, Redmond, WA, USA).
Discussion
This study aimed to systematically review the published literature regarding the role of the biomarkers albuminuria, serum creatinine, eGFR, and uric acid in predicting kidney disease progression and associated cardio-renal outcomes in persons with T2DM. Of the 42 longitudinal publications identified in this review, data reported for the baseline population, biomarkers, and outcomes investigated were for the large part heterogeneous. Biomarker measures tended to be dissimilar between studies, with multiple measures employed to assess eGFR and albuminuria. Further contributing to study heterogeneity, 12 different outcome measures were identified for kidney decline or disease progression alone, which made cross-study comparisons a challenge.
Irrespective of these discrepancies, the data reviewed in this study showed that albuminuria, serum creatinine/eGFR, or the combination of both, were robust predictors of adverse outcomes in persons with T2DM. Of the three biomarkers, albuminuria was the most frequently evaluated, with the majority of studies identified in this review displaying an association with kidney decline and related outcomes, which is fitting with prior investigations [
57,
31,
58]. Additional evidence in patients with T2DM has indicated that baseline micro- and macro-albuminuria as well as increasing albuminuria carry higher risks of declining kidney function and associated outcomes, beyond other existing renal biomarkers [
59]. In one meta-analysis of five studies encompassing patients with T2DM, patients with micro-albuminuria displayed an almost 4-fold (95% CI 1.6–8.4) increased RR for developing ESRD as compared with those who had normo-albuminuria [
60]. Absolute changes in albuminuria were also considered as precursors to kidney disease progression in individuals with T2DM [
61], with ample data further emphasizing the relationship between albuminuria and incipient nephropathy as well as CV outcomes [
62,
63]. Taken together, these findings support the utility of albuminuria as a robust predictor of renal decline. Hence, routine screening for albuminuria in persons with T2DM is considered effective clinical practice for monitoring the onset and progression of kidney disease.
Nevertheless, in light of the extant epidemiologic evidence, the clinical importance of albuminuria as a prognosticator of kidney disease has recently been challenged [
64,
61,
65]. Although albuminuria is commonly used to assess kidney disease development and progression among patients with T2DM, high variability has been observed in this individual marker along with insufficient sensitivity or specificity to detect kidney disease end points on its own [
66,
61,
67,
68]. By extension, the predictive utility of micro-albuminuria has further come into question due to observed spontaneous remission in people with diabetes [
69,
61,
70]. On this basis, seeking out novel biomarkers that can more reliably identify individuals at risk of experiencing poor renal outcomes is warranted. Yet to date, most studies have only assessed the relationship of individual, rather than combined or composite biomarkers, with renal outcomes [
71], and no new single biomarker has been shown capable of outperforming albuminuria [
72]. Given the numerous pathophysiological processes that encompass DKD (e.g. hyperfiltration, pro-inflammatory, pro-fibrotic, and angiogenic processes) [
73], it is doubtful that albuminuria, or any other single biomarker, is individually capable of accurately forecasting the development and progression of renal damage in persons with T2DM [
68]. On the other hand, a more comprehensive selection of multiple biomarkers that capture distinct pathways involved in renal impairment would presumably augment risk prediction more effectively than the use of a single biomarker alone. Indeed, extending our efforts in this study beyond the conventional “albuminuric pathway” has been described elsewhere [
66,
61,
67,
71,
68], and provides impetus for others to establish a panel of biomarkers intended to improve early detection of kidney disease development, as well as its progression in diabetes. For effective clinical use, this panel would need to be highly sensitive and specific, minimally invasive to collect, and cost-effective.
Several publications identified in this review demonstrated that patients with T2DM may present with low eGFR yet be normo-albuminuric. In this context, normo-albuminuric low eGFR has been reported to be more prevalent among older women [
67], with these patients often exhibiting a more diverse set of clinical characteristics compared with those who present with low eGFR and higher concentrations of albuminuria [
22,
74,
75]. Additionally, renal biomarkers are known to be influenced by clinical features such as age, gender, and genetic disparities, as well as modifiable factors [
61,
76]. To this end, distinct sets of risk factors have been confirmed on the background of low eGFR, and separately for increased albuminuria [
13], making it challenging to disentangle whether changes in these biomarkers are complementary of other underlying pathophysiological conditions not linked to kidney disease, or whether they truly reflect important markers of kidney disease development and progression in T2DM. Moving forward, other avenues of research appear warranted that will focus on assessing the predictive value of albuminuria, eGFR and other renal markers for kidney disease end points according to relevant subsets of conventional risk factors (i.e. age, race, or sex-specific disparities) in people with T2DM.
The findings from this SLR illustrated that the risk of kidney function decline and related outcomes tended to vary depending on which biomarker was studied. The available findings demonstrated that the risk as well as the onset of kidney function decline appeared to occur sooner in patients with T2DM who demonstrated an increase in albuminuria, or in the presence of a diminished eGFR over time, and that use of the two biomarkers combined had better predictive ability over time than either alone; although data were sparse. For this reason, studies incorporating these biomarkers in tandem, and their relationship with cardio-renal outcomes, would likely enrich our understanding of the numerous pathways involved in provoking renal impairment, especially given the high individual variability that often accompanies these biomarkers. To further improve prediction based on these studied biomarkers, standardized measures are also needed, as well as more reliable reporting of baseline kidney function and outcomes, particularly for kidney disease progression. Such standardizations will facilitate understanding of the clinical utility of albuminuria and serum creatinine as predictive biomarkers for kidney disease progression as well as related renal and CV outcomes and mortality risk among patients with T2DM, with the risk perhaps differing among select subgroups.
Limitations
The observed heterogeneity in the published data clearly limits the generalizability of our findings regarding the importance of albuminuria, serum creatinine, and eGFR as biomarkers for adverse renal outcomes. Although a rigorous, objective, and transparent systematic methodology was implemented and quantitative data were extracted from the included publications, the present SLR was descriptive and qualitative in nature due to the widespread disparity observed among the study designs, baseline population characteristics, biomarkers measured, outcomes reported, and study methods and reporting employed. The majority of included publications did not clearly report baseline CKD stage for the study populations. We therefore referred to eGFR, where available, to estimate baseline kidney function; and it should be noted that these estimates were based on heterogeneous data, as numerous methods were used to define eGFR. The evaluated publications, despite meeting inclusion criteria for this SLR, did not report outcomes consistently, particularly for kidney disease progression, which made it difficult to compare predictors across the studies examined. The scientific quality of the data presented varied; for example, some studies had small sample sizes and not all publications reported risk estimates such as hazard ratios or measures of precision. Others failed to report a test statistic for significance. Of the studies that did report hazard ratios, the range of covariates included in survival models also varied, further making it challenging to directly compare risk estimates across each investigation.
Acknowledgments
Editorial assistance was provided by Bríain Ó Hartaigh, PhD, Market Access Solutions, and Brett Wells, PhD, Editorial Solutions, both of Envision Pharma Group. The authors wish to thank Harry Otun, PhD, Envision Pharma Group, for assistance with screening, data extraction, and validation in this study. Funding for these services was provided by AbbVie, Inc.