THE NEED TO DEFINE RISK IN CHRONIC KIDNEY DISEASE , 669
RISK FACTORS AND MECHANISMS OF CHRONIC KIDNEY DISEASE PROGRESSION , 671
DEMOGRAPHIC VARIABLES , 673
HEREDITARY FACTORS , 674
HEMODYNAMIC FACTORS , 676
Decreased Nephron Number, 676
Blood Pressure, 679
Obesity and Metabolic Syndrome, 679
High Dietary Protein Intake, 680
Pregnancy and Preeclampsia, 680
MULTISYSTEM DISORDERS , 681
Diabetes Mellitus, 681
PRIMARY RENAL DISEASE , 681
CARDIOVASCULAR DISEASE , 681
BIOMARKERS , 682
Urinary Protein Excretion, 682
Serum Albumin, 683
Serum Uric Acid, 684
Serum Bicarbonate, 684
Plasma Asymmetric Dimethylarginine, 685
Serum Phosphate, 685
Other Biomarkers, 685
NEPHROTOXINS , 685
RENAL RISK SCORES , 687
General Population Renal Risk Scores, 687
Risk Scores for Patients with Diagnosed Chronic Kidney Disease, 689
Risk Scores for Predicting Chronic Kidney Disease After Acute Kidney Injury, 691
FUTURE CONSIDERATIONS , 691
The Need to Define Risk in Chronic Kidney Disease
The proposal in 2002 of a simple definition for chronic kidney disease (CKD) and its subsequent worldwide adoption, coupled with the development of the four-variable Modification of Diet in Renal Disease (MDRD) formula that facilitated automated estimation of glomerular filtration rate (GFR) from a measurement of serum creatinine, has brought about an increase in awareness of CKD. Population-based studies from around the world have reported a prevalence of 8% to 16%, values substantially higher than anticipated. The large number of people known to be affected by CKD has major implications for the provision of health care—in particular, nephrology services. In the past decade, nephrology has moved from a position where it provided highly specialized services to a relatively small number of patients, with specific and relatively rare kidney disease or advanced CKD, to one where it must concern itself with the care of less advanced CKD in a substantial proportion of the general population. Furthermore, early-stage CKD is largely asymptomatic, and detection therefore requires a screening process. Studies have indicated that screening whole populations is not cost effective, and a means of identifying high-risk subgroups for targeted screening is therefore required. Successful screening programs are likely to identify large numbers of patients with previously undiagnosed CKD but, in most countries, nephrology services are unable to provide long-term care to all CKD patients, and the associated costs would be prohibitive. A solution to this problem was suggested by studies showing that there is substantial heterogeneity among patients who meet the diagnostic criteria for CKD, with most being at relatively low risk of ever progressing to end-stage kidney disease (ESKD). The Kidney Disease Outcomes and Quality Initiative (K/DOQI) classification system for CKD was widely adopted and proved valuable, particularly for identifying the prevalence of different stages of CKD in epidemiologic studies. It was noted, however, that the classification provided little information on the future risk of decline in renal function. The Kidney Disease Improving Global Outcomes (KDIGO) classification system therefore modified the K/DOQI system so that categories defined by GFR and albuminuria do correlate with risk, but this does not provide accurate, individual risk prediction. Previous studies identified a wide range of rates of decline in GFR among patients with CKD, and up to 15% may even show an increase over time. There is thus a need to develop methods for risk stratification within CKD to identify the relatively small subgroup of patients who are at risk of progression to ESKD and who may benefit from specialist intervention to slow or halt disease progression. Such risk stratification would be equally important for identifying individuals who are at low risk for progression who could be reassured and spared unnecessary referral to a nephrologist.
Another important aspect of CKD is its association with a substantially increased risk of future cardiovascular events (CVEs) that in most patients with mild CKD substantially exceeds the risk of ESKD. Whereas CKD is associated with a high prevalence of many traditional risk factors for cardiovascular disease, such as hypertension and dyslipidemia, risk prediction tools such as the Framingham risk score substantially underestimate cardiovascular risk in patients with CKD. It has been proposed that this observation is due to the role of several nontraditional cardiovascular risk factors that are specific to CKD.
From the above discussion, it is clear that there is a need to identify and understand factors associated with an increased risk of developing CKD and, once diagnosed, factors associated with an increased risk of progression to ESKD and CVE. In this chapter, we will review current knowledge of these risk factors and the methods being applied to predict risk in CKD patients. Risk factors for cardiovascular disease in patients with CKD, many of which overlap with risk factors for CKD progression, are discussed in Chapter 56 .
Definition of a Risk Factor
A risk factor is a variable that has a causal association with a disease or disease process such that the presence of the variable in an individual or population is associated with an increased risk of the disease being present or developing in the future. Thus, risk factors may be useful for identifying subjects at increased risk for a disease or particular outcome due to a disease process. In the course of epidemiologic research, many variables may show associations with a disease of interest but these may be chance associations, noncausal associations, or causal associations (true risk factors). The Bradford Hill criteria provide minimum requirements to be fulfilled to identify a causal relationship between a putative risk factor (exposure) and a disease (outcome; Table 22.1 ). In complex diseases such as CKD that result from the combined effects of multiple factors, it is likely that many risk factors will not fulfill all the criteria. Nevertheless, they do provide a useful framework for assessing the strength of a proposed causal relationship between risk factor and disease.
|Strength of association||The stronger the association, the more likely the relationship is causal.|
|Consistency||A causal association is consistent when replicated in different populations and studies.|
|Specificity||A single putative cause produces a single effect.|
|Temporality||Exposure precedes outcome (i.e., risk factor predates disease).|
|Biological gradient||Increasing exposure to risk factor increases risk of disease, and reduction in exposure reduces risk.|
|Plausibility||The observed association is consistent with biologic mechanisms of disease processes.|
|Coherence||The observed association is compatible with existing theory and knowledge in a given field.|
|Experimental evidence||The factor under investigation is amenable to modification by an appropriate experimental approach.|
|Analogy||An established cause and effect relationship exists for a similar exposure or disease.|
Epidemiologic Methods for Identifying Risk Factors
Studies to investigate associations between putative risk factors and a disease may be classified as observational or experimental. Observational studies include cross-sectional, case-control, and cohort studies, whereas the randomized controlled trial is the main experimental study.
In this study type, associations between putative risk factors and a disease are investigated in a study population at a single time point. Cross-sectional studies therefore have the advantage of being relatively quick and simple to perform but, because they are limited to a single point in time, they are unable to fulfill the Bradford Hill criterion for temporality. Thus, associations may be identified but inference regarding causality cannot be made. Nevertheless, these studies are useful as an initial search for putative risk factors and hypothesis generation.
These studies also examine subjects at a single point in time but, in this design, cases with a particular disease are identified according to specific criteria and are compared to controls similar to the cases with respect to age, gender, and other variables but who do not have the disease. Cases and controls are then compared with respect to the prevalence of a particular exposure or putative risk factor. One weakness of case-control studies is that they often rely on recollection of past exposure to the putative risk factor. A further challenge is to achieve adequate matching of cases and controls with respect to variables other than the putative risk factor(s).
Cohort studies are prospective studies in which a population of subjects with and without exposure to a putative risk factor (or variable exposure to a putative risk factor) are followed into the future, and the rate of disease occurrence is compared between the two groups. Advantages are that the temporality criterion for causality may be fulfilled and a direct measure of the incidence of disease obtained. Nevertheless, one weakness of cohort studies is the potential for confounding. This may occur when a variable is associated with both the putative risk factor (exposure) and the disease (outcome). Thus, the presence of a confounder may alter (strengthen, weaken, or mask) the association between exposure and outcome. Multivariable regression analysis may be used to adjust or control for potential confounding but may not completely eliminate the effects of confounding, and incomplete adjustment may result in residual confounding.
Another technique to assist in differentiating causality from association is mendelian randomization. This approach tests whether genetically determined variation in a particular biomarker (which is not affected by nongenetic confounding) is associated with outcomes in a similar manner to that observed in other observational studies. If the biomarker is directly involved in the pathogenesis of a disease, then inherited variation that changes the plasma concentration of the biomarker should be associated with the outcome in the manner predicted by the plasma concentration.
Randomized Controlled Trials
In this study design, a randomized controlled trial (RCT), subjects in a population are randomly assigned to one of two or more treatments or interventions. After a fixed period of follow-up, the randomized groups are compared with respect to the rate of a predefined outcome. To reduce the potential for bias, subjects and/or investigators are often blinded to the treatment. In a single-blinded study, only subjects are unaware of what treatment they receive, whereas in a double-blinded study, subjects and investigators are blinded, usually by the use of a matching placebo. Randomization, if successful, will produce close matching of the groups with respect to a wide range of known and unknown variables at baseline to reduce the possibility of confounding. Furthermore, the RCT is the only study design capable of fulfilling the causality criterion for experimental evidence. Nevertheless, although the RCT constitutes the gold standard for investigating the effect of a therapeutic intervention, it is not as definitive for evaluating putative risk factors. This is because a particular intervention may modify more than one risk factor, and it is therefore not possible to attribute a change in the outcome to the change in a single risk factor. Perhaps the best example of this in CKD is treatment with an angiotensin-converting enzyme (ACE) inhibitor that modifies blood pressure and proteinuria. It is therefore not possible to attribute the subsequent slowing of GFR decline to lowering of blood pressure or to reduction of proteinuria alone.
Data from RCTs may also be used to perform subgroup or post hoc analyses. Subgroup analyses may be prespecified in the trial design (preferable) or be performed post hoc. Although subgroup and post hoc analyses may be useful for exploratory analyses and hypothesis generation, they are prone to several weaknesses. First, they may be underpowered and therefore prone to type 2 errors (incorrect failure to reject the null hypothesis). Second, if too many hypotheses are tested, they may be prone to type 1 errors (incorrect rejection of a true null hypothesis).
Risk Factors and Mechanisms of Chronic Kidney Disease Progression
It has been appreciated for several decades that once GFR has decreased to below a critical level, CKD tends to progress relentlessly toward ESKD. This observation suggests that loss of a critical number of nephrons provokes a vicious cycle of further nephron loss. Detailed studies have elucidated a number of interrelated mechanisms that together contribute to CKD progression, including glomerular hemodynamic responses to nephron loss (raised glomerular capillary hydraulic pressure and single-nephron GFR [SNGFR]), proteinuria, and proinflammatory responses. A generally good prognosis after unilateral nephrectomy attests to the fact that a single pathogenic factor may be insufficient to initiate progressive CKD, but the multihit hypothesis proposes that multiple factors interact to overcome renal reserve and provoke progressive nephron loss. To meet the Bradford Hill criteria of plausibility and coherence, a putative risk factor should therefore somehow affect known mechanisms of CKD progression (see Chapter 52 for further details). Figure 22.1 shows how risk factors may interact with pathophysiologic mechanisms to initiate or accelerate CKD progression. Based on our understanding of the mechanisms underlying the pathogenesis of CKD and its progression, risk factors may be divided into susceptibility factors, initiation factors, and progression factors ( Table 22.2 ) . Nevertheless, distinguishing among these categories may in some cases be difficult because some factors (e.g., diabetes mellitus) may act in all three ways and, in some studies, it may be impossible to separate susceptibility factors from progression factors due to inadequate characterization of participants at study entry.
|Family history of CKD||+|
|Low nephron number||+||+|
|High protein intake||+||+|
|Primary renal disease||+|
|Genetic renal disease||+|
|Acute kidney injury||+||+|
|Low serum bicarbonate||+|
These are risk factors associated with an increased risk of an individual developing CKD after exposure to a factor that has potential to cause renal damage. An example is a reduced nephron number after uninephrectomy, which is associated with an increased risk of developing diabetic nephropathy if the individual develops diabetes. Studies to identify susceptibility factors should recruit subjects free of CKD at baseline who have been exposed to an initiating factor and followed over a prolonged period to allow ascertainment of outcomes. This could be achieved through a cohort study or subgroup analysis of an RCT.
Initiation factors directly cause or initiate kidney damage in a susceptible individual. Examples include exposure to nephrotoxic drugs, urinary tract obstruction, or primary glomerulopathies that may provoke CKD in some (but not all) exposed individuals. Studies investigating initiation factors should aim to recruit subjects without CKD at entry or known susceptibility factors, with variable exposure to a putative initiating factor. A cohort study design is best suited to investigate outcomes in subjects exposed versus not exposed to the factor of interest, or an RCT design could be used to assess the potential nephrotoxicity of a new drug.
These are factors that contribute to the progression of kidney damage once CKD has developed. An example is hypertension, which exacerbates raised intraglomerular hydraulic pressure and therefore accelerates glomerular damage. Studies investigating progression factors should recruit subjects with relatively early-stage CKD in a cohort study design. RCTs may also be used to study progression factors if the intervention being investigated modifies a putative progression factor. Outcomes may therefore be compared between the group in whom the risk factor was modified versus a control group. Unfortunately, however, many interventions alter several risk factors, and it may therefore not be possible to attribute an improved outcome to changes in a single risk factor.
The prevalence of CKD increases with age and is reported to be as high as 56% in those 75 years of age or older. Longitudinal studies of subjects without kidney disease have observed a decline in GFR with increasing age in some subjects, implying that nephron loss may be regarded as part of normal aging. On the other hand, aging is associated with an increase in several other risk factors for CKD, including hypertension, obesity, and cardiovascular disease, which may contribute to the rise in CKD prevalence. Several population-based studies have found a higher incidence of proteinuria and CKD as well as ESKD with increasing age. Similarly, the incidence of a decline in renal function over 5 years was greater among older patients with hypertension. One study reported that advanced age is a negative predictor of ESKD among patients with CKD, although older age was associated with a greater rate of decline in GFR. This apparent contradiction is most likely explained by the competing risks of death and ESKD in older patients, illustrated by the observation from one longitudinal study that for patients 65 years and older, the risk of ESKD exceeded the risk of death only when the GFR was 15 mL/min/1.73 m 2 or less ( Figure 22.2 ). On the other hand, another study found that in patients with CKD stage 4 or 5, renal function in those 65 years or older was associated with a slower decline than in those younger than 45 years.
A large individual-level meta-analysis that included data from 2,051,244 participants in 46 cohort studies has provided robust data regarding the effect of age on the risks associated with CKD. In general population and high cardiovascular risk cohorts, the increase in relative risk of mortality associated with lower GFR declined with increasing age but the increase in absolute risk of death was higher in older age groups. Similar trends were observed for the mortality risks associated with albuminuria. On the other hand, the relative increase in risk of mortality did not decrease with increasing age in data from CKD cohorts. Furthermore, there was no attenuation of the risk of ESKD with increasing age in any of the cohort categories. Thus, older age is a susceptibility factor for CKD, and the associated increase in risk of death and ESKD is observed at all ages. These observations suggest that targeted screening for CKD in older subjects would be a cost-effective strategy, but further studies are required to investigate the extent to which the risks associated with CKD in older adults may be attenuated by intervention. For further discussion of CKD in older age groups, see Chapters 24 and 85 .
In experimental studies, male rodents were more susceptible to age-related glomerulosclerosis than females, an observation that was independent of glomerular hemodynamics or hypertrophy and was attributable to a specific androgen effect. Data regarding the effect of gender on the risk of CKD and progression in humans are, however, somewhat contradictory. Many reports have indicated that male gender is associated with worse renal outcomes. Studies have reported a higher incidence of proteinuria and CKD among men in the general population and an increased risk of ESKD or death associated with CKD, a higher risk of decline in renal function in male hypertensive patients, a lower risk of ESKD in female patients with CKD stage 3, and a shorter time to renal replacement therapy (RRT) in male patients with CKD stage 4 or 5. In addition, most national registries, including the U.S. Renal Data Service (USRDS), have reported a substantially higher incidence of ESKD in males (413 per million population [pmp] in 2003) versus females (280 pmp). Previous meta-analyses have, however, yielded conflicting results, with one reporting a higher rate of decline in GFR in men and another reporting a higher risk of doubling of the serum creatinine level or ESKD in women after adjustment for baseline variables, including blood pressure and urinary protein excretion.
The largest meta-analysis to date included data from studies of 2,051,158 participants investigating the impact of gender on CKD-related outcomes. Whereas the risk of all-cause and cardiovascular mortality was higher in men than women, the relative risk of mortality increased with lower GFR and higher albuminuria in both, and the slope of increase in risk was steeper in women than men. Importantly, the relative risk of ESKD increased with lower GFR and higher albuminuria in both sexes and there was no evidence of a difference in the increase in risk between men and women ( Figure 22.3 ). One limitation of many of the studies quoted is that menopausal status of the women was not documented. Nevertheless, it is clear from the most robust data published that CKD is associated with at least the same relative increase in risk of death and ESKD in women as in men. The reasons for the higher absolute incidence of renal replacement therapy in men versus women require further investigation. For further discussion of the impact of gender on CKD, see Chapter 20 , Chapter 21 , Chapter 52 .
African Americans are overrepresented in the U.S. dialysis population, suggesting that ethnicity is a strong risk factor for the progression of CKD to ESKD. Population-based studies have found a higher incidence of ESKD among African Americans that was attributable only in part to socioeconomic and other known risk factors. Similarly, the risk of early renal function decline (increase in serum creatinine ≥ 0.4 mg/dL) was approximately threefold higher (odds ratio [OR], 3.15; 95% confidence interval [CI], 1.86 to 5.33) among black versus white diabetic adults, but 82% of this excess risk was attributable to socioeconomic and other known risk factors. The risk of renal function decline over 5 years among hypertensive patients was greater in African Americans, and African ancestry was independently associated with a greater rate of GFR decline in the MDRD study. Interestingly, data from the Reasons for Geographic and Racial Differences in Stroke (REGARDS) Cohort Study have shown a lower prevalence of estimated GFR (50 to 59 mL/min/1.73 m 2 ) among African American versus white subjects but a higher prevalence of estimated GFR (10 to 19 mL/min/1.73 m 2 ), suggesting that African American ethnicity acts as a progression factor but not as a susceptibility factor. A 2012 report from the USRDS showed a substantially higher incidence of ESKD in African Americans (3.3 times higher than whites), Hispanics (1.5 times higher than non-Hispanics), and Native Americans (1.5 times higher than whites). Similarly, the prevalence of ESKD in 2012 was higher among minority groups: African Americans, 5671 pmp; Native Americans, 2600 pmp; Hispanics, 2932 pmp; Asians, 2272 pmp; and whites, 1432 pmp. CKD and ESKD have also been reported to be more prevalent in other ethnic groups, including Asians, Hispanics Native Americans, Mexican Americans, and Aboriginal Australians.
A large meta-analysis that included data from 940,366 participants in 25 general population cohort studies investigated the impact of ethnicity on the risks associated with CKD in blacks, whites, and Asians. The absolute risk of all-cause or cardiovascular mortality (after adjustment for age) and ESKD was higher in black versus white versus Asian participants. However, the relative risk of all-cause or cardiovascular mortality and ESKD increased to a similar degree with lower GFR or greater albuminuria in all the ethnic groups. Thus, the risk between lower GFR or greater albuminuria and mortality or ESKD was not modified by ethnicity. The mechanisms underlying the associations between ethnicity and CKD remain to be elucidated, but possible explanations include genetic factors (see below, “ Hereditary Factors ”), increased prevalence of diabetes mellitus, lower nephron endowment, increased susceptibility to salt-sensitive hypertension, and environmental, lifestyle, and socioeconomic differences. Ethnicity and CKD are discussed further in Chapter 20 , Chapter 21 , Chapter 52 .
Hereditary renal diseases resulting from a single gene defect, such as autosomal dominant polycystic kidney disease, Alport’s disease, Fabry’s disease, and congenital nephrotic syndrome, account for a relatively small yet clinically important proportion of all patients with CKD. Nevertheless, evidence is rapidly accumulating that genetic factors account for familial clustering of many other forms of CKD with multifactorial causes. Among 25,883 incident ESKD patients, 22.8% reported a family history of ESKD, and screening of the relatives of patients with ESKD revealed evidence of CKD in 49.3%. In another case-control study, including 689 patients with ESKD and 361 controls, having one first-degree relative with CKD increased the risk of ESKD by 1.3 (95% CI, 0.7 to 2.6) and having two such relatives increased it by 10.4 (95% CI, 2.7 to 40.2) after controlling for multiple known risk factors, including diabetes and hypertension. Similarly, a case-control study of 103 U.S. white patients with ESKD reported a 3.5-fold increase in risk of ESKD (95% CI, 1.5 to 8.4) with the presence of a first-, second-, or third-degree relative with ESKD.
A genetic explanation for the high incidence of ESKD observed in African Americans was provided by groundbreaking research that identified strong associations between ESKD and two coding variants in the APOL1 gene. These gene variants confer resistance to infection with Trypanosoma brucei rhodesiense, which causes sleeping sickness. This observation explains how selection likely resulted in a high prevalence of these variants in the population. Subsequent studies have identified associations between APOL1 risk variants and several renal pathologies, including focal segmental glomerulosclerosis (FSGS), HIV-associated nephropathy (HIVAN), sickle cell kidney disease, and severe lupus nephritis. Moreover, cohort studies have reported associations between APOL1 risk variants and risk of progression to ESKD. Risk of progression was the lowest in European Americans (with no risk variants), intermediate in African Americans, with no or one risk variant, and highest in African Americans, with two risk variants. It is estimated that APOL1 variants account for 40% of disease burden due to CKD in African Americans.
Despite the strong association between inheritance of two APOL1 risk variants and ESKD, only a minority of people with this genotype actually develop kidney disease, suggesting that the action of a second factor is required to cause disease in genetically susceptible individuals. HIV is one example of such a second hit, but it has been proposed that other viruses and other gene variants may also be important.
Other studies have suggested that genetic factors also increase susceptibility to early manifestations of CKD. In a study of 169 families with one type 2 diabetic proband, the diabetic siblings of probands with microalbuminuria had a significantly increased risk of also having microalbuminuria, after adjustment for confounding risk factors (OR, 3.94; 95% CI, 1.93 to 9.01) than the diabetic siblings of probands without microalbuminuria. Furthermore, the nondiabetic siblings of diabetic probands with microalbuminuria had a significantly higher urinary albumin excretion rate (within the normal range) than the nondiabetic siblings of normoalbuminuric diabetic probands.
Genomewide association studies (GWAS) have identified multiple novel loci that are significantly associated with serum creatinine levels or CKD. Furthermore, a recent GWAS meta-analysis conducted in 63,558 participants of European descent identified significant associations between GFR decline over time and three gene loci— UMOD (previously associated with CKD and ESKD), GALNTL5/GALNT11, and CDH23 . It was estimated that the heritability of GFR decline in this population was 38%. Further studies have investigated the role of epigenetic factors (heritable changes in the pattern of gene expression not attributable to changes in the primary nucleotide sequence) that may affect the risk of CKD progression. One study compared the genomewide DNA methylation profile in 20 people from the Chronic Renal Insufficiency Cohort (CRIC) study with the most rapid decline in GFR and 20 with the most stable GFR. Results identified differences in the methylation of several genes associated with epithelial to mesenchymal transition and inflammation that may be involved in the mechanisms of CKD progression.
From this discussion, it is clear that genetic factors may act as susceptibility factors in some subjects, initiating factors in those with CKD due to a single gene defect, or progression factors in others. The rapid growth in knowledge of genetic aspects of CKD will likely result in genetic risk factors becoming increasingly important in risk prediction for patients with CKD. For a more detailed discussion of genetic aspects of kidney disease, see Chapter 43 , Chapter 44 , Chapter 45 , Chapter 46 .
Experimental studies have shown that glomerular hemodynamic responses (e.g., glomerular capillary hypertension and hyperfiltration) to nephron loss and chronic hyperglycemia are critical factors in establishing the vicious cycle of nephron loss characteristic of CKD. In addition, any factor that further increases glomerular hypertension and/or hyperfiltration may be expected to exacerbate glomerular damage and accelerate the progression of CKD (see Figure 22.1 ).
Decreased Nephron Number
Autopsy studies have revealed that the number of nephrons per kidney varies widely in humans, from 210,332 to 2,702,079 in one series. Multiple factors have been shown to influence nephron endowment, including those that affect the fetomaternal environment as well as genetic factors. A substantial body of evidence supports the hypothesis that low nephron endowment predisposes individuals to CKD by provoking an increase in SNGFR and, therefore, a reduction in renal reserve. The ascertainment of nephron number in living human subjects is currently not possible, but autopsy studies have shown an association between reduced nephron number and hypertension, as well as glomerulosclerosis. In human autopsy studies, low birth weight is directly associated with reduced nephron number, and birth weight may therefore serve as a marker of nephron endowment. Low birth weight is also a risk factor for later life hypertension and diabetes mellitus, both of which further increase the risk of CKD. One meta-analysis of 32 studies, which included data from over 2 million subjects, reported a significantly increased risk of albuminuria (OR, 1.81; 95% CI, 1.19 to 2.77) and ESKD (OR, 1.58; 95% CI, 1.33 to 1.88) associated with low birth weight. Thus low birth weight, acting as a marker of reduced nephron endowment, may be regarded as a susceptibility and progression risk factor for CKD. Factors affecting nephron endowment and the consequences of reduced nephron endowment are discussed in more detail in Chapter 23 .
Acquired Nephron Deficit
In experimental models of acquired nephron deficit, severe nephron loss (5/6 nephrectomy) alone initiates a cycle of progressive injury in the remaining glomeruli, mediated primarily through glomerular hypertension and hyperfiltration. In 14 patients subjected to similarly large reductions in nephron number following partial resection of a single kidney, 2 developed ESKD and 9 developed proteinuria, the extent of which was inversely correlated with the amount of renal tissue remaining. Lesser degrees of acquired nephron loss, such as removal of one of two previously normal kidneys (uninephrectomy), may not be sufficient to cause CKD in most subjects. However, nephrectomy for renal cell carcinoma is associated with an increased risk of developing CKD that is greater after radical nephrectomy than partial nephrectomy, suggesting that in the presence of subclinical kidney damage, acquired nephron loss may provoke CKD, and that the risk is proportional to the number of nephrons removed.
Nephron loss may also predispose individuals to CKD if they are also exposed to other risk factors. This is perhaps best illustrated by the observation that uninephrectomy exacerbates renal injury in experimental diabetic nephropathy and, in diabetics, uninephrectomy increases the risk of developing diabetic nephropathy.
The interaction between nephron loss and other risk factors is further illustrated by the observation that in a study of 488 people who had surgery for renal cell carcinoma, radical nephrectomy (compared to partial nephrectomy), diabetes, and increased age were each independently associated with an increased risk of developing CKD at least 6 months after surgery. In those who had a partial nephrectomy but no additional risk factors, only 7% developed CKD, but this increased to 24%, 30%, and 42% in those 60 years of age or older, those with hypertension, and diabetics, respectively.
In most forms of human CKD, initial nephron loss due to primary renal disease, multisystem disorders that involve the kidney or exposure to nephrotoxins is focal, but hemodynamic adaptations in the remaining glomeruli are thought to contribute to nephron loss by provoking further glomerulosclerosis (see Chapter 52 ). Several epidemiologic studies have supported this hypothesis by showing that patients with a reduced GFR are at increased risk of a further decline in renal function. Two large meta-analyses of cohort studies identified baseline GFR as a strong predictor of ESKD. Among 845,125 participants from the general population, the estimated GFR (eGFR) was independently associated with an increased risk of developing ESKD when it fell below 75 mL/min/1.73 m 2 . For groups of patients with average eGFRs of 60, 45, and 15 mL/min/1.73 m 2 , the hazard ratios for developing ESKD were 4, 29, and 454, respectively, when compared to a reference group with an eGFR of 95 mL/min/1.73 m 2 . Similar findings were reported in a further 173,892 participants selected for being at increased risk of developing CKD ( Figure 22.4 ). Among 21,688 patients selected for having CKD, a lower eGFR was an independent risk factor for ESKD, such that a fall of 15 mL/min/1.73 m 2 below a threshold of 45 mL/min/1.73 m 2 was associated with a pooled hazard ratio of 6.24. Further analyses by the CKD Prognosis Consortium have confirmed that the association between reduced GFR and increased risk of ESKD persists independently of gender, age, ethnicity, diabetes, and hypertension. Additionally, analysis of data from 1,530,648 participants has shown that change in GFR over time is strongly predictive of future risk of ESKD (and mortality), suggesting that a GFR decline of 30% may be useful as a surrogate marker of CKD progression in clinical trials. Thus, in different contexts, acquired nephron deficit may be regarded as a susceptibility factor (e.g., after donor nephrectomy in a healthy kidney donor), initiation factor (when severe nephron loss provokes glomerulosclerosis in remaining previously normal glomeruli), or progression factor (when nephron loss accelerates pre-existing damage in remaining glomeruli).
The importance of GFR as a risk factor has emphasized the need for more accurate methods to estimate it. Adoption of the MDRD equation improved detection of CKD and made possible much of the epidemiologic research on CKD, but it was recognized from the outset that the MDRD equation was imperfect and, in particular, tended to underestimate true GFR at values above 60 mL/min/1.73 m 2 . This is important because this is the threshold below which CKD may be diagnosed without other evidence of kidney damage. Several other equations have been developed to estimate GFR from serum creatinine concentration, culminating in a recommendation by KDIGO that the MDRD equation should be replaced by the Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) equation, which is more accurate than the MDRD equation and results in less bias, particularly at GFR values above 60 mL/min/1.73 m 2 . Further analysis by the CKD Prognosis Consortium found that the eGFR determined by the CKD-EPI equation results in a lower prevalence of CKD stages 3 to 5 (8.7% vs. 6.3%) and affords better risk prediction than the MDRD equation. Among those classified as having an eGFR of 59 to 45 mL/min/1.73 m 2 by the MDRD equation, 34.7% were reclassified by the CKD-EPI as having an eGFR of 89 to 60 mL/min/1.73 m 2 , and those reclassified had a lower incidence of adverse outcomes versus those not reclassified (e.g., incidence of all-cause mortality 9.9 vs. 34.5/1000 person-years).
The limitations of creatinine as a marker of GFR due to nonrenal factors that affect serum creatinine concentration, including muscle mass and diet, have prompted a search for alternatives. Cystatin C, a peptide produced by all nucleated cells and therefore not affected by muscle mass, has emerged as the most promising alternative. The production of reference material to standardize cystatin C assays has greatly improved the potential for clinical application, and equations have been developed to estimate the GFR from the serum cystatin C concentration or from creatinine and cystatin C levels together. The combined equation and CKD-EPI (creatinine) equation have similar bias, but the combined equation yields better precision and accuracy. Further work by the CKD Prognosis Consortium has reported that when cystatin C is used to estimate GFR, reclassification to a higher GFR category (higher than that assigned by eGFR creatinine) is associated with a lower risk of all-cause mortality, cardiovascular mortality, and ESKD. It should be noted, however, that in this analysis, CKD was defined by a single eGFR value. It therefore remains to be established whether the use of cystatin C will improve risk prediction in those with CKD defined by two eGFR values determined at least 90 days apart, as required by the KDIGO definition.
Acute Kidney Injury
Despite previous perceptions that patients who recover from acute kidney injury (AKI) regain normal renal function and have a good prognosis, several cohort studies have reported that recovery from AKI is associated with a substantially increased risk of CKD and death. Among 3769 adults who required dialysis for AKI and survived dialysis free for at least 30 days, the incidence rate for chronic dialysis was 2.63/100 versus 0.91/100-person years in 13,598 matched controls (adjusted hazard ratio [HR], 3.23; 95% CI, 2.70 to 3.86). The relative risk was particularly high for those with no previous diagnosis of CKD (adjusted HR, 15.54; 95% CI, 9.65 to 25.03). There was no difference in survival between the groups. In another study of similar design, outcomes were investigated in 343 patients with a preadmission eGFR more than 45 mL/min/1.73 m 2 who required dialysis for AKI but survived for at least 30 days after discharge without dialysis. After controlling for potential confounders, AKI that required dialysis was associated with a 28-fold increase in the risk of developing CKD stage 4 or 5 (adjusted HR, 28.1; 95% CI, 21.1 to 37.6) and more than double the risk of death (adjusted HR, 2.3; 95% CI, 1.8 to 3.0) versus 555,660 adult patients hospitalized during the same period but without AKI.
Analysis of data from a cohort of 233,803 Medicare beneficiaries 67 years or older who were hospitalized in 2000 reported a substantially increased risk of developing ESKD in those who developed AKI on a background of CKD (HR, 41.2; 95% CI, 34.6 to 49.1) or without previous CKD (HR, 13.0; 95% CI, 10.6 to 16.0) versus those who did not develop AKI. The importance of AKI as a risk factor for CKD initiation was further illustrated by the observation that among patients who had AKI without preexisting CKD ( N = 4730), 72.1% developed CKD within 2 years of the AKI episode. Furthermore, 25.2% of those who developed ESKD had a history of AKI. In a similar study, a cohort of 113,272 patients hospitalized with a primary diagnosis of acute tubular necrosis (ATN), AKI, pneumonia, or myocardial infarction (control group) was studied. Overall, 11.4% progressed to CKD stage 4 during follow-up, including 20.0% of those with ATN, 13.2% of those with AKI, 24.7% of those with preexisting CKD, and 3.3% of the control patients. After controlling for other variables, having a diagnosis of AKI, ATN, or CKD increased the risk of developing CKD stage 4 by 303%, 564%, and 550%, respectively, versus controls. After controlling for covariates, AKI and CKD were associated with an increased risk of death of 12% and 20%, respectively, versus controls.
The multiplicative effect of AKI on CKD progression is further illustrated by a study of 39,805 patients with an eGFR less than 45 mL/min/1.73 m 2 prior to hospitalization. Those who survived an episode of dialysis-requiring AKI had a very high risk of developing ESKD within 30 days of hospital discharge (i.e., nonrecovery of AKI) that was related to the preadmission eGFR. For an eGFR of 30 to 44 mL/min/1.73 m 2 , the incidence of ESKD was 42% and, for an eGFR of 15 to 29 mL/min/1.73 m 2 , it was as high as 63%, whereas the incidence of ESKD was only 1.5% in those who did not have dialysis-requiring AKI. In patients who survived longer than 30 days after hospital discharge without ESKD, the incidence of ESKD and death at 6 months was 12.7% and 19.7%, respectively, versus 1.7% and 7.4% in the comparator group with CKD but no AKI. After adjustment for multiple risk factors, AKI was associated with a 30% increase in long-term risk for death or ESKD (adjusted HR, 1.30; 95% CI, 1.04 to 1.64).
Consistent with the findings of individual studies, a meta-analysis of 13 cohort studies reported a significantly increased risk of developing CKD and ESKD in patients who had survived an episode of AKI versus participants without AKI (pooled adjusted HR for CKD, 8.8; 95% CI, 3.1 to 25.5; pooled HR for ESKD, 3.1; 95% CI, 1.9 to 5.0). Taken together, these data show that AKI should be regarded as an important risk factor for CKD initiation and progression. The mechanisms responsible for these observations require further elucidation but have been proposed to include nephron loss, loss of peritubular capillaries, cell cycle arrest, cell senescence, pericyte and myofibroblast activation, fibrogenic cytokine production, and interstitial fibrosis. The incidence of AKI has increased, and it is likely to become an increasingly important risk factor for CKD among older patients.
Hypertension is an almost universal consequence of reduced renal function but is also an important factor in the progression of CKD. In the hypothesis of CKD progression presented in Figure 22.1 , it is clear that elevated systemic blood pressure transmitted to the glomerulus would contribute to glomerular hypertension and thus accelerate glomerular damage. Hypertension has been shown to be predictive of ESKD risk in several large population-based studies. Furthermore, a close association between the magnitude of increased risk and level of blood pressure has been reported in several studies, so that even elevations in blood pressure below the threshold for the diagnosis of hypertension were associated with increased risk of ESKD.
Among patients with CKD in the MDRD study, higher baseline mean arterial pressure (MAP) independently predicted a greater rate of GFR decline. These observations have led to the suggestion that blood pressure be viewed as a continuous rather than dichotomous risk factor for CKD, with less emphasis on traditional definitions of hypertension and normotension. Despite these close associations, the causality criterion for a risk factor requires evidence from an RCT. Three large RCTs have sought to investigate the effect on CKD progression of intensive versus standard blood pressure lowering. Whereas the primary analysis of data from the MDRD study found no significant difference between the rate of decline in GFR between patients randomized to intensive blood pressure control (target MAP < 92 mm Hg, equivalent to <125/75 mm Hg) versus standard blood pressure control (target MAP < 107 mm Hg, equivalent to 140/90 mm Hg), a secondary analysis did show benefit associated with the low blood pressure target in patients with higher levels of baseline proteinuria. Further secondary analysis showed that the lower achieved blood pressure was also associated with a slower GFR decline, an effect that was more marked in patients with higher baseline proteinuria. Furthermore, long-term follow-up (mean, 6.6 years) of patients from the MDRD study reported a significant reduction in the risk of ESKD (adjusted HR, 0.68; 95% CI, 0.57 to 0.82) or a combined end point of ESKD or death (adjusted HR 0.77; 95% CI, 0.65 to 0.91) in patients randomized to low blood pressure targets, even though treatment and blood pressure data were not available beyond the 2.2 years of the original trial.
In the African American Study of Kidney Disease and Hypertension (AASK), no significant difference in the rate of GFR decline was observed between subjects randomized to MAP goals of ≤92 mm Hg or lower versus 102 to 107 mm Hg. It should be noted, however, that patients in AASK generally had low levels of baseline proteinuria (mean urine protein, 0.38 to 0.63 g/day) . Furthermore, prolonged follow-up of the AASK cohort after completion of the randomized trial found no significant differences in the primary outcome for the whole cohort; however, it did show a significantly reduced risk of creatinine doubling, ESKD, or death in subjects with a baseline urine protein/creatinine ratio more than 0.22 g/g who were initially randomized to intensive blood pressure control. Thus, the MDRD and AASK study results suggest a significant interaction between blood pressure and proteinuria as risk factors for CKD progression.
In a third study, additional blood pressure reduction with a calcium channel blocker in patients with nondiabetic CKD on ACE inhibitor (ACEI) treatment failed to produce additional renoprotection, but the degree of additional blood pressure reduction was modest (4.1/2.8 mm Hg) and may have been insufficient to improve outcomes in patients already receiving optimal ACEI therapy. A recent RCT has reported significant benefit associated with a lower blood pressure target in young people with autosomal dominant polycystic kidney disease (age < 50 years) and GFR higher than 60 mL/min/1.73 m 2 . Participants randomized to a low blood pressure target (110/75 to 95/60 mm Hg) evidenced a slower rate of increase in kidney volume and a greater decrease in albuminuria and left ventricular mass index than those randomized to usual blood pressure control (120/70 to 130/80 mm Hg). Interestingly blood pressure was not an independent predictor of ESKD in diabetic patients in the RENAAL study or in predominantly nondiabetic subjects in the Chronic Renal Insufficiency in Birmingham (CRIB) study. This is likely due to the fact that blood pressure was well controlled in all subjects (in the RENAAL study) and illustrates how risk factors may vary in importance, depending on the population studied.
Taken together, there is convincing evidence that elevated blood pressure is an important risk factor for progression of CKD, although unequivocal evidence from RCTs is lacking, and some uncertainty remains regarding optimal treatment targets. It is hoped that the ongoing Systolic Blood Pressure Intervention Trial (SPRINT; NCT01206062 at www.clinicaltrials.gov ), which randomized more than 9000 subjects, including about one third with nondiabetic CKD, to systolic blood pressure targets of less than 120 or 140 mm Hg, will yield new data to guide future recommendations.
Obesity and Metabolic Syndrome
In experimental models, obesity is associated with hypertension, proteinuria, and progressive renal disease. Micropuncture studies have confirmed that obesity is another cause of glomerular hyperfiltration and glomerular hypertension that can be predicted to exacerbate the progression of CKD. Furthermore, several other factors associated with obesity and the metabolic syndrome may contribute to renal damage, including hormones and proinflammatory molecules produced by adipocytes, increased mineralocorticoid levels and/or mineralocorticoid receptor activation by cortisol, and reduced adiponectin levels. In humans, severe obesity is associated with increased renal plasma flow, glomerular hyperfiltration, and albuminuria, abnormalities that are reversed by weight loss. Obesity, as defined by increased body mass index (BMI), has been associated with increased risk of developing CKD in several large population-based studies. Furthermore, one study has found a progressive increase in relative risk of developing ESKD associated with increasing BMI (RR, 3.57; 95% CI, 3.05 to 4.18 for BMI of 30.0 to 34.9 kg/m 2 vs. BMI of 18.5 to 24.9 kg/m 2 ) among 320,252 subjects confirmed to have no evidence of CKD at initial screening.
There is evidence that obesity may directly cause a specific form of glomerulopathy characterized by proteinuria and histologic features of focal and segmental glomerulosclerosis, but it is likely that it also acts as a risk factor in the development of several other forms of renal disease. One study has identified childhood obesity as a risk factor for CKD in adulthood. Among 4340 participants born in one week in 1946, pubertal-onset obesity and obesity throughout childhood were associated with an increased risk of CKD (defined by eGFR < 60 mL/min/1.73 m 2 or albuminuria) at age 60 to 64 years. Interest has also focused on the role of the metabolic syndrome (insulin resistance), as defined by the presence of abdominal obesity, dyslipidemia, hypertension, and fasting hyperglycemia, in the development of CKD. An analysis of data from the Third National Health and Nutrition Examination Survey (NHANES III) data found a significantly increased risk of CKD and microalbuminuria in subjects with the metabolic syndrome as well as a progressive increase in risk associated with the number of components of the metabolic syndrome present. Furthermore, a large longitudinal study of 10,096 patients without diabetes or CKD at baseline identified metabolic syndrome as an independent risk factor for the development of CKD over 9 years (adjusted OR, 1.43; 95% CI, 1.18 to 1.73). Again, there was a progressive increase in risk associated with the number of traits of the metabolic syndrome present (OR, 1.13; 95% CI, 0.89 to 1.45 for one trait vs. OR, 2.45; 95% CI, 1.32 to 4.54 for five traits). In another study, patient hip/waist ratio, a marker of insulin resistance, was independently associated with impaired renal function, even in lean individuals (BMI < 25 kg/m 2 ), among a population-based cohort of 7676 subjects.
The effect of obesity on the progression of established CKD is less well documented. Increased BMI has been identified as a risk factor for CKD progression among subjects with immunoglobulin A (IgA) nephropathy, renal mass reduction surgery or renal agenesis, and renal transplants. On the other hand, BMI was unrelated to the risk of ESKD among a cohort of patients with CKD stage 4 or 5. It is widely recognized that weight loss is difficult to achieve in obese individuals, but surgical intervention in the form of gastric banding or bypass appears to offer the most effective long-term outcomes. Two large cohort studies have shown significant survival benefit in subjects who underwent bariatric surgery, but unfortunately renal end points were not reported in these studies. Beneficial renoprotective effects of weight loss have been reported in a meta-analysis of observational studies that found an association between weight loss and reduction in proteinuria independent of blood pressure, as well as smaller studies that reported improvement or stabilization of renal function or reduction in proteinuria following bariatric surgery in subjects with CKD.
The best method for assessing obesity in CKD remains to be determined. A further systematic review analyzed the effects of weight loss achieved by bariatric surgery, medication, or diet in 31 studies and found that in most studies, weight loss was associated with reductions in proteinuria. In people with glomerular hyperfiltration, the GFR tended to decrease with weight loss, and in those with a reduced GFR, it tended to increase. BMI is the most widely applied method but does not take body composition into account. One study has reported a high sensitivity but relatively low specificity of BMI to detect obesity in subjects with CKD.
High Dietary Protein Intake
Protein feeding provokes an increase in GFR in rodents and humans. Consistent with the hypothesis that the glomerular hemodynamic changes associated with hyperfiltration accelerate glomerular injury, experimental studies have reported that a high-protein diet accelerates renal disease progression, whereas dietary protein restriction results in normalization of glomerular capillary hydraulic pressure as well as SNGFR and marked attenuation of glomerular damage. Observational studies in humans have reported an increased risk of microalbuminuria associated with higher dietary protein intake in subjects with diabetes and hypertension (OR, 3.3; 95% CI, 1.4 to 7.8) but not in healthy subjects or those with isolated diabetes or hypertension, again illustrating the interaction between risk factors for CKD. In another study, high intake of protein, particularly nondairy animal protein, was associated with a greater rate of GFR decline among women with an eGFR of 80 to 55 mL/min/1.73 m 2 but not in those with an eGFR of more than 80 mL/min/1.73 m 2 . Randomized trials investigating the effects of high protein are lacking, but several studies have sought to examine the potential renoprotective effects of dietary protein restriction. In the MDRD study, primary analysis revealed no significant difference in the mean rate of GFR decline in subjects randomized to low- or very low-protein diets, but secondary analysis of outcomes according to achieved dietary protein intake indicated that a reduction in protein intake of 0.2 g/kg/day correlated with a 1.15-mL/min/year reduction in the rate of GFR decline, equivalent to a 29% reduction in the mean rate of GFR decline. On the other hand, long-term follow-up of participants in study 2 of the MDRD trial found no renoprotective benefit among those randomized to very low-protein diet in the original study, but did report a higher risk of death in this group (HR, 1.92; 95% CI 1.15 to 3.20). Nevertheless, three meta-analyses of smaller studies have all reported a significant renoprotective benefit associated with dietary protein restriction. The role of dietary protein restriction in the management of CKD is discussed further in Chapters 52 and 61 .
Pregnancy and Preeclampsia
Physiologic adaptations during pregnancy provoke glomerular hyperfiltration that usually does not cause renal damage. In the context of preexisting CKD, however, the glomerular hyperfiltration of pregnancy can be predicted to exacerbate proteinuria and glomerular injury. Several studies have shown an increased risk of CKD progression during pregnancy, particularly when the pregestational serum creatinine is 1.4 mg/dL or higher (≥124 µmol/L). In one study of 82 pregnancies in 67 women with primary renal disease and serum creatinine level of 1.4 mg/dL or more, blood pressure, serum creatinine, and proteinuria increased during pregnancy. In 70 pregnancies with postpartum data available, persistent loss of maternal renal function at 6 months was reported in 31%, and by 12 months 8 women had progressed to ESKD. Adverse obstetric outcomes included preterm delivery in 59% and low birth weight in 37%, although fetal survival was 93%.
In a more recent series of 49 women with CKD stage 3 to 5 before pregnancy, the mean GFR declined during pregnancy (from 35 ± 12.2 to 30 ± 13.8 mL/min/1.73 m 2 ), but there was no change in the mean postpartum rate of GFR decline. Nevertheless, a pregestational GFR less than 40 mL/min/1.73 m 2 , combined with proteinuria of more than 1 g/day, was associated with a more rapid postpartum GFR decline and a shorter time to ESKD or halving of GFR and low birth weight. Although earlier reports suggested good outcomes, one recent study has reported adverse effects associated even with early-stage CKD. In 91 pregnancies with predominantly CKD stages 1 and 2, modest increases in hypertension, serum creatinine, and proteinuria were observed. An increase in adverse obstetric outcomes, including preterm delivery, lower birth weight, and admission to a neonatal intensive care unit versus low-risk pregnancy controls was also reported; this remained true, even when only those with CKD stage 1 were considered, although there were no perinatal deaths. On the other hand, pregnancy was not associated with a more rapid decline in the GFR over 5 years in a cohort of 245 women of childbearing age with IgA nephropathy and serum creatinine level of 1.2 mg/dL or lower (in the majority).
Complications of pregnancy and, in particular, hypertension and preeclampsia, may also cause renal damage. In one large population-based study, renal outcomes were assessed in 570,433 women who had had at least one singleton pregnancy. Only 477 women developed ESKD at a mean of 17 ± 9 years after the first pregnancy (overall rate, 3.7/100,000/women/year), but preeclampsia was associated with a significant increase in the risk of ESKD, ranging from a relative risk of 4.7 for preeclampsia in a single pregnancy (95% CI, 3.6 to 6.1) to a relative risk of 15.5 for preeclampsia in two or three pregnancies (95% CI, 7.8 to 30.8). The risk was further increased if the pregnancy resulted in a low-birth weight or preterm infant. Causes of ESKD were glomerulonephritis in 35%, hereditary or congenital disease in 21%, diabetic nephropathy in 14%, and interstitial nephritis in 12%. Similarly, in women with diabetes prior to pregnancy, preeclampsia and preterm birth were associated with significantly increased risks of ESKD and death, illustrating how different risk factors for CKD may interact to increase risk.
A large cohort study reported an increased risk of multiple adverse health outcomes after hypertension during pregnancy, including cardiovascular disease, diabetes mellitus, and CKD (HR, 1.91; 95% CI, 1.18 to 3.09). Similarly, a very large case-control study found that hypertension during pregnancy was associated with a substantially increased risk of subsequent CKD (HR, 9.38; 95% CI, 7.09 to 12.4) or ESKD (HR, 12.4; 95% CI, 8.53 to 18.0). In both these studies, the risks of CKD were substantially higher if preeclampsia developed during the pregnancy. Possible explanations for these observations include the presence of pathogenic factors common to CKD and preeclampsia, including obesity, hypertension, insulin resistance, and endothelial dysfunction, exacerbation by preeclampsia of preexisting subclinical CKD, and effects of preeclampsia on the kidney that increase the risk of CKD later in life. That preeclampsia may provoke renal damage has been suggested by several studies showing an increased incidence of microalbuminuria after preeclampsia. A meta-analysis of seven of these studies reported a 31% prevalence of microalbuminuria at a weighted mean of 7.1 years after preeclampsia versus 7% in a control group with uncomplicated pregnancies. Further research is required to identify which mechanisms are most relevant but, even without further information, preeclampsia should be regarded as a risk factor for the development and progression of CKD.
Diabetic nephropathy has rapidly become the single most common cause of ESKD worldwide. Diabetes was associated with a substantially increased risk of ESKD or death associated with CKD in one population-based study of 23,534 subjects (HR, 7.5; 95% CI, 4.8 to 11.7), as well as an increased risk of moderate CKD (estimated creatinine clearance < 50 mL/min) in another study of 1428 subjects with an estimated creatinine clearance of more than 70 mL/min at baseline. Evidence that glycemic control is a key risk factor for the development of diabetic nephropathy has been shown in randomized trials that found a reduced risk of developing nephropathy in subjects with type 1 and type 2 diabetes randomized to tight glycemic control. The pathogenesis of diabetic nephropathy is complex and involves multiple mechanisms, including glomerular hemodynamic factors, advanced glycation end product formation, generation of reactive oxygen species, and upregulation of profibrotic growth factors and cytokines. In at least one study, diabetic nephropathy was associated with more rapid progression to ESKD than other causes of CKD. Thus, diabetes may be regarded as a susceptibility, initiation, and progression risk factor for CKD. For further discussion of the pathogenesis of diabetic nephropathy, see Chapter 39 .
Primary Renal Disease
Whereas substantial variation in the rate of GFR decline has been observed among subjects with a common cause of CKD, there is also evidence that some forms of CKD may provoke more rapid progression than others. In the MDRD study and the Chronic Renal Insufficiency Standards Implementation Study (CRISIS), a diagnosis of adult polycystic kidney disease was an independent predictor of a greater rate of GFR decline. In several cohort studies, diabetic nephropathy was associated with shorter time to ESKD or a more rapid GFR decline than other diagnoses.
Multiple studies have reported that CKD is associated with a substantial increase in the risk of cardiovascular disease (CVD), and it is therefore not surprising that CVD is also associated with an increased risk of CKD. Among hospitalized Medicare beneficiaries, the prevalence of CKD stage 3 or worse was 60.4% for those with heart failure and 51.7% for those with myocardial infarction. The presence of CKD in addition to heart disease was associated with a significantly increased risk of death and progression to ESKD. These observations may in part be explained by the fact that CVD and CKD share many risk factors, including obesity, metabolic syndrome, hypertension, diabetes mellitus, dyslipidemia, and smoking. In addition, CVD may exert effects on the kidneys that promote the initiation and progression of CKD, including decreased renal perfusion in heart failure and atherosclerosis of the renal arteries. For example, renal atherosclerosis was detected in 39% of patients (≥70% stenosis in 7.3%) undergoing elective coronary angiography. Furthermore, arterial stiffness may result in greater transmission of an elevated systemic blood pressure to glomerular capillaries and exacerbate glomerular hypertension. In one study, pulse wave velocity (PWV) and augmentation index (AI), markers of arterial stiffness, were identified as independent risk factors for progression to ESKD among subjects with CKD stage 4 or 5 ; in another study, AI was an independent determinant of rate of creatinine clearance decline among subjects with CKD stage 3. On the other hand, neither PWV nor AI were predictors of the rate of GFR decline in a cohort of subjects with CKD stages 2 to 4. In two relatively small cohort studies of those with CKD, a diagnosis of CVD was associated with an increased risk of progression to ESKD but, in the CRIC study, a history of any CVD at baseline was not associated with an increased risk of the primary end point of ESKD or 50% GFR reduction among 3939 participants. Conversely, in the same study, a history of heart failure was independently associated with a 29% higher risk of the primary outcome. For further discussion of cardiovascular disease in patients with CKD, see Chapter 56 .
Urinary Protein Excretion
Abnormal excretion of protein in the urine indicates dysfunction of the glomerular filtration barrier and is therefore a marker of glomerulopathy and an index of disease severity. Experimental evidence has suggested that proteinuria may also contribute to progressive renal damage in CKD (see Chapter 53 ). A large body of evidence attests to a strong association between proteinuria and the risk of CKD progression, as well as cardiovascular and all-cause mortality. Mass screening of a general population of 107,192 participants by dipstick urinalysis identified proteinuria as the most powerful predictor of ESKD risk over 10 years (OR, 14.9; 95% CI, 10.9 to 20.2). Similarly, among 12,866 middle-aged men enrolled in the multiple risk factor intervention trial (MRFIT), proteinuria detected by dipstick test was associated with a significantly increased risk of developing ESKD over 25 years (HR for 1+ proteinuria, 3.1; 95% CI, 1.8 to 3.8; HR for ≥2+ proteinuria, 15.7; 95% CI, 10.3 to 23.9). Furthermore, detection of 2+ proteinuria or more increased the hazard ratio for ESKD associated with an eGFR less than 60 mL/min/1.73 m 2 from 2.4 without proteinuria (95% CI, 1.5 to 3.8) to 41 with proteinuria (95% CI 15.2 to 71.1).
Similar associations have been reported for measurements of urinary albumin in the general population. In the Nord-Trøndelag Health (HUNT 2) study, which included 65,589 adults, micro- and macroalbuminuria were independent predictors of ESKD after 10.3 years (HR, 13.0 and 47.2, respectively) and combining reduced eGFR with albuminuria substantially improved the prediction of ESKD. In the Prevention of Renal and Vascular End-stage Disease (PREVEND) study, albuminuria was an independent predictor of a decline in eGFR to less than 60 mL/min/1.73 m 2 .
Among patients selected for having CKD from a wide variety of causes, baseline proteinuria has consistently predicted renal outcomes. In three large prospective studies that included patients with nondiabetic CKD (MDRD study, Ramipril Efficacy In Nephropathy [REIN] study, and AASK), higher baseline proteinuria was strongly associated with a more rapid decline in GFR. Similarly, among patients with diabetic nephropathy, the baseline urinary albumin/creatinine ratio was a strong independent predictor of ESKD in the RENAAL study and Irbesartan in Diabetic Nephropathy Trial (IDNT). The findings of these individual studies have been confirmed by two large meta-analyses. In one analysis, which included nine general population cohorts ( N = 845,125) and eight cohorts with increased risk of developing CKD ( N = 173,892), urine albumin/creatinine ratios of more than 30, 300, and 1000 mg/g were independently associated with progressive increases in the risk of ESKD, progressive CKD, and AKI, respectively (see Figure 22.4 ). Among 21,688 patients known to have CKD from 13 studies, an eightfold higher urine albumin/creatinine or protein/creatinine ratio was associated with increased all-cause mortality (pooled HR, 1.40) and risk of ESKD (pooled HR, 3.04). Further meta-analyses by the CKD Prognosis Consortium have shown that the magnitude of proteinuria remains a risk factor for ESKD independent of gender, ethnicity, age, diabetes, or hypertension.
Secondary analyses of prospective RCTs have found that the extent of residual proteinuria that persists, despite optimal treatment with an ACEI or angiotensin receptor blocker (ARB), also predicts renal prognosis. In the REIN study, percentage reduction in proteinuria over the first 3 months and the absolute level of proteinuria at 3 months were strong independent predictors of the subsequent rate of decline in GFR. In the IDNT, a greater reduction in proteinuria at 12 months was associated with a greater reduction in the risk of ESKD (HR, 0.44; 95% CI, 0.40 to 0.49 for each halving of baseline proteinuria). In the AASK study, a change in proteinuria from baseline to 6 months predicted subsequent progression. Similarly, a meta-analysis of data from 1860 patients with nondiabetic CKD showed that during antihypertensive treatment, the current level of proteinuria was a powerful predictor of the combined end point of doubling of the baseline serum creatinine level or onset of ESKD (relative risk [RR], 5.56; 95% CI, 3.87 to 7.98 for each 1.0-g/day increase in proteinuria). Furthermore, a meta-analysis of 21 randomized trials of drug treatment in CKD, which included 78,342 participants, found that for each 30% initial reduction in albuminuria on treatment, the risk of ESKD decreased by 23.7% (95% CI, 11.4% to 34.2%) independent of the class of drug used for treatment. These data support the proposal that proteinuria, like blood pressure, should be regarded as a continuous risk factor for CKD progression. Proteinuria thus appears to be a powerful predictor of renal risk in the general population, in patients with CKD prior to treatment, and in CKD patients on treatment. Recognition of the importance of proteinuria as a risk factor in CKD prompted the addition of an albuminuria category (A1 to A3) to the CKD classification proposed by KDIGO.
These important observations raise the question of how best to measure proteinuria. As discussed, all measurements of proteinuria have been reported to predict renal outcomes, including dipstick urinalysis, urine albumin/creatinine ratio or protein/creatinine ratio (ACR and PCR) and 24-hour urinary albumin or protein excretion. A secondary analysis of data from the RENAAL trial found that urine ACR measured on the first morning void was better than 24-hour urinary protein or albumin concentration as a predictor of time to doubling of the serum creatinine level or ESKD among patients with diabetes and CKD. On the other hand, retrospective analysis of data from 5586 patients with CKD reported similar HRs associated with urinary ACR and PCR for the outcomes of all-cause mortality, start of renal replacement therapy, or doubling of serum creatinine. Further analysis of these data identified a cohort of patients with a normal urine ACR but elevated urine PCR in whom the risk of ESKD or death was intermediate between the groups with both urine ACR and PCR abnormal or normal. Analysis of data from the CRIC study also reported that urine ACR and PCR had similar associations with complications of CKD. Together, these data imply that any measure of proteinuria is better than no measurement. If the goal is to detect and monitor low levels of albuminuria (category A1 and A2), the urine ACR measured on the first morning void is best. For patients with CKD, urine ACR or PCR may be used, and there is some evidence that there may be added information gained by requesting both.
Serum albumin levels are widely regarded as a marker of nutritional status but may also be reduced due to proteinuria or inflammation. Several studies have identified lower serum albumin levels as a risk factor for CKD progression. In the MDRD study, higher baseline serum albumin was associated with slower subsequent rate of GFR decline but, in a multivariable analysis, this was displaced by a similar correlation with baseline serum transferrin levels, another marker of protein nutrition. Three studies have found associations between serum albumin and renal outcomes in patients with type 2 diabetes and CKD. Among 182 patients with a mean serum creatinine of 1.5 mg/dL at baseline, hypoalbuminemia was an independent risk factor for ESKD. In a long-term follow-up of 343 patients, lower baseline serum albumin was an independent predictor of CKD progression and, in the RENAAL study, lower serum albumin was an independent predictor of ESKD. Similar observations have been reported in other forms of CKD. In a large cohort of patients with IgA nephropathy ( N = 2269), lower serum total protein (composed largely of albumin) was an independent risk factor for ESKD. In a cohort of 3449 patients with CKD referred to a nephrology service, lower serum albumin was an independent risk factor for ESKD. In these studies, the predictive value of serum albumin was independent of and additional to that of proteinuria, indicating that it was not merely acting as a marker of albuminuria.
Chronic anemia due to inherited hemoglobinopathy is associated with increased renal plasma flow, glomerular hyperfiltration, and subsequent development of proteinuria, hypertension, and ESKD. Anemia is a common complication of CKD from any cause, and several studies have shown that it is also an independent predictor of CKD progression. In the RENAAL study, baseline hemoglobin was a significant independent predictor of ESKD among diabetic patients—each 1-g/dL decrease in hemoglobin was associated with an 11% increase in the risk of ESKD. Baseline hemoglobin was also one of four variables included in the renal risk score developed from the RENAAL data. Similarly, a higher hemoglobin level was independently associated with lower risk of progression to ESKD (halving of GFR or need for dialysis) or death among 131 patients with all forms of CKD (HR, 0.778; 95% CI, 0.639 to 0.948 for each 1-g/dL increase). Furthermore, time-averaged hemoglobin of less than 12 g/dL was associated with a significantly increased risk of ESKD among 853 male veterans with CKD stages 3 to 5 (HR, 0.74; 95% CI, 0.65 to 0.84 for each 1-g/dL increase in hemoglobin).
Two other cohort studies have identified lower hemoglobin as an independent risk factor for a more rapid decline in GFR in patients with CKD stage 4 and ESKD in patients with CKD stage 3 or 4. Consistent with the hypothesis that anemia contributes directly to CKD progression, two small randomized studies have reported a renoprotective benefit associated with erythropoietin therapy. Among patients with serum creatinine of 2 to 4 mg/dL and hematocrit less than 30%, erythropoietin treatment was associated with significantly improved renal survival. In nondiabetic patients with serum creatinine of 2 to 6 mg/dL, early treatment (started when hemoglobin < 11.6 g/dL) with erythropoietin alpha was associated with a 60% reduction in the risk of doubling the serum creatinine level, ESKD, or death versus delayed treatment (started when hemoglobin < 9.0 g/dL).
On the other hand, two other studies that had left ventricular mass as their primary end point and the Trial to Reduce Cardiovascular Events with Aranesp Therapy (TREAT) found no effect of a high versus low hemoglobin target on the rate of decline in the GFR. Several studies have, however, reported adverse outcomes associated with normalization of hemoglobin in patients with CKD. In the Cardiovascular Risk Reduction by Early Anemia Treatment with Epoetin Beta (CREATE) study, randomization to a higher hemoglobin target (13 to 15 mg/dL) was associated with a shorter time to initiation of dialysis than a lower target (10.5 to 11.5 mg/dL). In TREAT, randomization to a higher hemoglobin target was associated with an increased risk of stroke and, in the Correction of Hemoglobin and Outcomes in Renal Insufficiency (CHOIR) study, a higher hemoglobin target was associated with an increased incidence of the combined end point of all-cause mortality, myocardial infarction, or hospitalization for congestive cardiac failure.
Lipid abnormalities are common in patients with CKD, and several studies have identified dyslipidemia as a susceptibility and progression factor for CKD. In population-based studies, several lipid profile abnormalities have been associated with an increased risk of developing CKD, including an elevated low-density lipoprotein (LDL) to high-density lipoprotein (HDL) cholesterol ratio, higher triglyceride and lower HDL cholesterol levels, lower HDL cholesterol levels and elevated total cholesterol, low HDL cholesterol, and elevated total to HDL cholesterol. Several observational studies have reported dyslipidemia as a risk factor for CKD progression. In the MDRD study, lower HDL cholesterol levels independently predicted a more rapid decline in GFR ; in a smaller study of patients with CKD, total cholesterol, LDL cholesterol, and apolipoprotein B levels were all associated with a more rapid decline in the GFR. Among 223 patients with IgA nephropathy, hypertriglyceridemia was independently predictive of CKD progression. Hypercholesterolemia was reported to predict loss of renal function in patients with type 1 or 2 diabetes and, among nondiabetic patients, CKD advanced more rapidly in those with hypercholesterolemia and hypertriglyceridemia.
RCTs of lipid lowering have produced mixed results with respect to renal outcomes. Subgroup analysis of a prospective randomized trial of pravastatin treatment in patients with previous myocardial infarction found that pravastatin slowed the rate of decline in patients with an eGFR less than 40 mL/min/1.73 m 2 , an effect that was also more pronounced in those with proteinuria. Similarly, in the Heart Protection Study, patients with previous cardiovascular disease or diabetes randomized to simvastatin treatment had a smaller increase in serum creatinine than those who received placebo. In a placebo-controlled, open-label study, atorvastatin treatment in patients with CKD, proteinuria, and hypercholesterolemia was associated with preservation of creatinine clearance, whereas it declined in those receiving placebo. On the other hand, lipid lowering with fibrates was not associated with renoprotection in two studies, although one study did show a reduced incidence of microalbuminuria in patients with type 2 diabetes receiving fenofibrate.
One meta-analysis of 13 small controlled trials found that lipid-lowering therapy was associated with a significantly slower rate of GFR decline (0.156 mL/min/month; 95% CI, 0.026 to 0.285; P = 0.008) among patients with CKD. On the other hand, several other studies found no association between dyslipidemia and risk of CKD progression. Analysis of data from a relatively small subgroup of studies with renal end points recorded in a meta-analysis found that statin therapy was associated with a reduction in proteinuria but with no improvement in creatinine clearance in participants with CKD. Furthermore, analysis of data from 3939 participants in the CRIC study found no association between total or LDL cholesterol and the risk of ESKD or 50% reduction in eGFR. Indeed, among participants with proteinuria of less than 0.2 g/day, higher LDL and total cholesterol were associated with a lower risk of reaching this end point. The Study of Heart and Renal Protection (SHARP) is the largest RCT to investigate the cardiovascular and renoprotective effects of lipid lowering in CKD. Patients with CKD or undergoing dialysis were randomized to treatment with simvastatin and ezetimibe or placebo. In 6245 participants with CKD not requiring dialysis, treatment resulted in an average reduction in LDL cholesterol of 0.96 mmol/L but was not associated with a reduction in the primary outcome of ESKD or secondary outcome of ESKD or creatinine doubling. Similarly, a meta-analysis of 38 studies, which included 37,274 participants with CKD, found that statin therapy was associated with a reduction in mortality and cardiovascular events but no clear effect on CKD progression. Together, evidence that dyslipidemia is a risk factor for CKD progression remains mixed, with the most recent studies indicating no association. Mechanisms whereby dyslipidemia may contribute to CKD progression are discussed in Chapter 52 .
Serum Uric Acid
Hyperuricemia is a common consequence of chronic renal failure and may also contribute to CKD progression. Several cohort studies have investigated uric acid as a risk factor in CKD and were summarized in a recent review. Most but not all population-based studies have identified hyperuricemia as an independent risk factor for the development of incident CKD. Similarly, most cohort studies that included people with CKD have identified a higher serum uric acid level as a risk factor for CKD progression. Possible mechanisms whereby hyperuricemia may contribute to CKD progression are exacerbation of glomerular hypertension, endothelial dysfunction, and proinflammatory effects. On the other hand, it is possible that an elevated uric acid concentration is acting as a marker of reduced kidney function or oxidative stress—uric acid is produced by xanthine oxidase, which also generates reactive oxygen species.
To date, only small studies investigating the effect of uric acid–lowering therapy on CKD progression have been published. A meta-analysis of eight trials found no difference in the eGFR among participants treated with allopurinol versus those with no treatment or placebo in five trials, whereas three trials that reported only serum creatinine reported benefit in favor of allopurinol. In five trials that measured proteinuria, no benefit was observed. Together, published evidence suggests that an elevated serum uric acid level may act as a susceptibility and progression risk factor in CKD, but large randomized trials are still required to determine whether treatment of hyperuricemia is beneficial for slowing CKD progression.
Studies in animal models have shown that acidosis may promote the progression of CKD through several mechanisms, including activation of the alternative complement pathway by elevated cortical ammonia levels, increased production of endothelin and aldosterone, and calcium deposition. At least five studies have investigated serum bicarbonate as a risk factor in human CKD. In all except the MDRD study, lower serum bicarbonate levels, even within the normal range, were independently associated with an increased risk of CKD progression. Two small randomized trials have reported slowing of CKD progression with bicarbonate supplementation, and another trial found that correction of acidosis with oral sodium bicarbonate or a diet rich in fruits and vegetables was associated with a lower rate of GFR decline. Bicarbonate supplementation is already recommended for patients with levels below 22 mEq/L, but several studies are underway to investigate further whether it is beneficial in the setting of less severe acidosis.
Plasma Asymmetric Dimethylarginine
Asymmetric dimethylarginine (ADMA) is formed by the breakdown of arginine methylated proteins and acts as an endogenous inhibitor of nitric oxide synthase to reduce nitric oxide production. The increased ADMA levels observed with a reduced GFR have been proposed as one mechanism for the endothelial dysfunction associated with CKD. Elevated ADMA levels are associated with CVD and cardiovascular mortality in patients with CKD. In animal models, administration of ADMA was associated with the development of hypertension, increased deposition of collagen I and III and fibronectin in glomeruli and blood vessels, and rarefaction of peritubular capillaries. Conversely, the overexpression of dimethylarginine dimethylaminohydrolase (DDAH), the enzyme responsible for degradation of ADMA, was associated with reduced ADMA levels and amelioration of renal injury in rats after 5/6 nephrectomy, implying that ADMA may also promote CKD progression.
Several relatively small studies have identified increased ADMA levels as a risk factor for CKD progression. Among 131 patients with CKD, a higher plasma ADMA level was an independent risk factor for ESKD or death (HR, 1.20; 95% CI, 1.07 to 1.35 for each 0.1-µmol/L increase). In 227 relatively young patients with mild-to-moderate nondiabetic CKD, higher ADMA levels predicted progression to the combined end point of creatinine doubling or ESKD (HR, 1.47; 95% CI, 1.12 to 1.93 for each 0.1-µmol/L increase). Finally, retrospective analysis of data from 109 patients with IgA nephropathy showed associations between ADMA levels and glomerular and tubulointerstitial injury. Furthermore, the plasma ADMA level was an independent determinant of annual GFR reduction rate.
When rats were fed a high-phosphate diet after uninephrectomy, renal calcium and phosphate deposition, as well as tubulointerstitial injury, were observed within 5 weeks. Furthermore, in animals and humans with CKD, dietary phosphate restriction or treatment with oral phosphate binders was associated with reductions in proteinuria and glomerulosclerosis and attenuation of CKD progression. Together, these data suggest that phosphate loading and/or hyperphosphatemia exacerbate renal injury in CKD. In addition, higher levels of the phosphatonin fibroblast growth factor 23 (FGF23) have been identified as an independent predictor of CKD progression. Three cohort studies of patients with CKD have identified higher serum phosphate levels as an independent risk factor for progression. On the other hand, the largest study to date, which included 10,672 participants with CKD, found no independent association between higher serum phosphate and risk of progression. It should be noted, however, that the number of ESKD events was low, and the study therefore had limited power to detect a moderate association between serum phosphate levels and CKD progression.
A number of other biomarkers are currently being investigated as risk factors in CKD. Although many have been reported to be associated with adverse outcomes, the challenge is to identify biomarkers that add to the predictive power of established risk factors. For a comprehensive discussion of novel biomarkers, see Chapter 30 .