'Urinary Protein & Albumin Both Predict Mortality & Renal Outcomes' -
Comparison of Urinary Albumin and Urinary Total Protein as Predictors of Patient Outcomes in CKD
American Journal of Kidney Diseases|
Volume 57, Issue 1 , Pages 21-28, January 2011
"Predictors evaluated in study: Baseline measurements of albumin-creatinine ratio (ACR), total protein-creatinine ratio (PCR), 24-hour albuminuria, and total proteinuria. In conclusion, total proteinuria and albuminuria perform equally as predictors of renal outcomes and mortality in patients with CKD. ACR and PCR were as effective as 24-hour urine samples at predicting outcomes and are more convenient for patients, clinicians, and laboratories. Both ACR and PCR stratify risk in patients with CKD."
"There is extensive literature linking albuminuria with cardiovascular disease and mortality in both renal and general populations......Identification and quantification of proteinuria is essential to identify those at high risk of progression to end-stage renal disease.4 Albuminuria usually is measured in patients with diabetes, and most intervention studies have used albuminuria.24, 25, 26 Some guidelines also recommend using albuminuria in patients with nondiabetic kidney disease,13, 14 but others recommend using total proteinuria.15, 16 It is notable that the benefits of interventions in patients with nondiabetic kidney disease are based on studies that used total proteinuria rather than albuminuria.27 This report shows that both are equally predictive of patient outcomes at all levels of proteinuria in a mixed population of patients attending a general nephrology clinic, who predominantly have nondiabetic kidney disease."
"A subsidiary finding in our study was that PCR and ACR were as powerful as 24-hour urine protein and albumin excretion at predicting patient outcomes, respectively. Traditionally, 24-hour urine samples have been seen as the gold-standard method to measure total proteinuria or albuminuria; however, spot urine samples are more convenient for patients, clinicians, and laboratories."
What to Measure-Albuminuria or Total Proteinuria? Editorial
American Journal of Kidney Diseases
Volume 57, Issue 1 , Pages 1-2, January 2011
Paul E. de Jong, MD, PhD Division of Nephrology, University Medical Centre Groningen, Hanzeplein 1, 9713 GZ Groningen, The Netherlands
Stephan J.L. Bakker, MD, PhD Ron T. Gansevoort, MD, PhD
Related Article, p. 21
The publication of the National Kidney Foundation's Kidney Disease Outcomes Quality Initiative guidelines on the classification of chronic kidney disease (CKD) in 20021 has drawn much attention to optimization of risk prediction in individuals with CKD. Since staging of CKD is based on glomerular filtration rate (GFR) and albuminuria, these 2 diagnostic tests have been discussed intensively in the last decade. Most reports dealt with the optimization of estimated GFR in large population databases. Fewer studies have addressed optimization of methods for assessing urinary albumin or the question of whether albumin is the optimal urinary protein to be measured. In this issue of the American Journal of Kidney Diseases, Methven et al shed some light on the latter issue, asking whether it is urinary albumin or total urinary protein that is more tightly linked to outcomes.2 This question should be considered in historical perspective.
In the period from the 1960s until the 1980s, nephrologists were interested in the urinary loss of large amounts of protein (typically more than 2-3 grams of protein per day) in patients with glomerular disorders. Laboratory methods for detection of total protein were available and sensitive enough for this purpose. In addition, nephrologists studied whether the loss of proteins was related only to loss of albumin (about 60,000 daltons) or also to loss of larger-molecular-weight proteins, such as immunoglobulins (about 160-180,000 daltons). It was shown in various studies that the presence of large-molecular-weight proteins is associated with poor kidney survival.3, 4 In the 1980s and 1990s, diabetologists became interested in the predictive value of small amounts of proteins in the urine. As the quantity of urinary total protein could not reliably be measured in these lower ranges, most studies used the more sensitive measurement of albumin, and referred to small amounts of albumin in the urine as "microalbuminuria."5, 6 Microalbuminuria, however, is not only related to diabetes, but is also associated with most traditional cardiovascular risk factors, including hypertension, obesity, and smoking.7
It has been argued that the loss of large amounts of albumin is a manifestation of kidney disease, while low amounts of albumin in the urine (microalbuminuria) is considered a manifestation of generalized vascular damage.8 Irrespective of the amount, albuminuria results from damage to the glomerular vascular endothelium. However, since the tubules reabsorb 90%-95% of all albumin filtered by the glomerulus,9 it has been questioned whether the presence of microalbuminuria might rather reflect a tubular disorder or a combination of a glomerular and tubular disorder.10 In the case of overt proteinuria or macroalbuminuria, usually around 20% of proteins are of nonalbumin nature, frequently including proteins with a higher molecular weight than albumin.11 In contrast, in microalbuminuria, and particularly in the normoalbuminuria ranges, up to 80% of the proteins can be of nonalbumin nature. It is most likely that these 80% of nonalbumin proteins are low-molecular-weight proteins (tubular) rather than high-molecular-weight proteins.
Methven et al studied a cohort of 5,586 individuals that had been referred to a renal unit to evaluate the predictive value of albuminuria and proteinuria on 3 outcomes: (1) all-cause mortality, (2) start of renal replacement therapy, and (3) doubling of serum creatinine. The predictive utility of urinary total protein was comparable to that of urinary albumin for all 3 outcomes, regardless of whether it was expressed on a continuous scale or a categorical scale.
Individuals with a wide range of albuminuria were included in the study; about one-third had normoalbuminuria (albumin-creatinine ratio [ACR] <30 mg/g), one-third had microalbuminuria (ACR 30-300 mg/g), and one-third had macroalbuminuria (ACR >300 mg/g). Urinary total protein and urinary albumin predicted outcomes to a similar extent not only in those with macroalbuminuria but also in those with microalbuminuria.
At first glance, this is surprising. First, since the measurement of albumin is more specific, more sensitive, and better standardized than that of total protein,12, 13 one would expect this to result in better risk assessment.14 Second, when total protein is assessed, one is not informed of the nature of the proteins present in the sample. The finding that urinary total protein in the microalbuminuria range (most likely consisting of tubular proteins) and urinary albumin are equally associated with renal outcomes supports the hypothesis that albuminuria reflects a generalized tubular disorder. The relatively large contribution of tubular proteins to total urinary protein compared to albumin also makes one wonder whether the level of total protein would independently add to the level of albumin in the outcome prediction. It would have been interesting if the authors had performed such an evaluation in addition to the evaluations presented in their article. Although it is understandable that total protein may predict renal outcomes as well as albumin, it is more difficult to understand why both are comparable in their relationship to total mortality, which most likely relates to associations with cardiovascular outcomes. How could generalized vascular damage to the kidney result in tubular dysfunction? It has been argued that generalized atherosclerosis leads to medullary hypoperfusion and hypoxia, thus inducing renal tubular dysfunction, which causes impaired reabsorption of albumin and other tubular proteins.15
Importantly, Methven et al assessed both urinary albumin and urinary protein in fresh samples. This makes their data fairly unique, because other studies investigating the predictive value of proteinuria or albuminuria used stored, frozen samples. In a number of studies, we showed previously that frozen storage leads to increased variability in and decreased predictive value of albuminuria, in particular when urine samples are stored at -20°C without appropriate alkalinization prior to assessment.14, 16, 17
Should these findings influence our practice regarding risk prediction in CKD? The data emphasize that in patients referred for CKD who appear to have <300 mg of protein per day, we should perform more sensitive protein or albumin measurements to better predict future risk. Methven et al argue that the costs for a total protein assay are lower than for an albumin assay, but this argument may not withstand economic developments around these assays in the near future. The better argument would be one that is based on scientific grounds. First, these data should of course be confirmed before changing clinical practice. Second, as long as we do not know what kind of specific proteins we measure and thus do not know what exactly underlies the predictive power of the total protein assay, the present findings clearly warrant further research studies. Such investigations could lead to better prediction of cardiovascular and renal events, but will also increase our understanding of the pathophysiology of CKD and its link with atherosclerotic cardiovascular disease.
Comparison of Urinary Albumin and Urinary Total
Protein as Predictors of Patient Outcomes in CKD
Proteinuria is common, with a prevalence of 1.3% (frank proteinuria) to 8.2% (microalbuminuria) in the US general population,1 and is associated with adverse patient outcomes.
Total proteinuria is the single strongest predictor of renal risk, predicting progressive kidney disease and end-stage kidney disease.2, 3, 4 In patients with diabetes mellitus, albuminuria predicts progressive kidney disease,5 but the importance of albuminuria (as opposed to total proteinuria) in those with nondiabetic kidney disease is less well established. It predicts progression to end-stage renal disease in those with decreased estimated glomerular filtration rate (eGFR)6 and de novo decreases in kidney function in the general population.7
Albuminuria also is associated with increased risk of cardiovascular disease and death in both diabetic and nondiabetic populations,8, 9 even at levels less than microalbuminuria.10, 11 Proteinuria assessed using dipstick urinalysis is associated with increased cardiovascular risk12; however, dipsticks mainly measure albumin. The impact of total proteinuria on mortality has been less well characterized.
Quantification of proteinuria therefore is essential to stratify risk, but should it be measured as total proteinuria or albuminuria? Some guidelines recommend using albumin-creatinine ratio (ACR) for all patients with chronic kidney disease (CKD),13, 14 whereas others recommend restricting ACR to patients with diabetes mellitus and using total protein-creatinine ratio (PCR) for all others.15, 16
The biochemistry laboratory in Glasgow Royal Infirmary routinely analyzes urine samples for both albumin and total protein. We assessed which was the superior predictor of renal outcomes and mortality in patients with CKD attending our clinic.
Proteinuria is common and is associated with adverse patient outcomes. The optimal test of proteinuria to identify those at risk is uncertain. This study assessed albuminuria and total proteinuria as predictors of 3 patient outcomes: all-cause mortality, start of renal replacement therapy (RRT), and doubling of serum creatinine level.
Retrospective longitudinal cohort study.
Setting & Participants
Nephrology clinic of a city hospital in Scotland; 5,586 patients with chronic kidney disease (CKD) and proteinuria measured in random urine samples (n = 3,378) or timed urine collections (n = 1,808).
Baseline measurements of albumin-creatinine ratio (ACR), total protein-creatinine ratio (PCR), 24-hour albuminuria, and total proteinuria.
All-cause mortality, start of RRT, and doubling of serum creatinine level were assessed using receiver operating characteristic curves and Cox proportional hazards models.
Blood pressure, serum creatinine level, ACR, PCR, date of death, date of starting RRT.
Patients were followed up for a median of 3.5 (25th-75th percentile, 2.1-6.0) years. For all outcomes, adjusted HRs were similar for PCR and ACR (derived from random urine samples and timed collections): death, 1.41 (95% CI, 1.31-1.53) vs 1.38 (95% CI, 1.28-1.50); RRT, 1.96 (95% CI, 1.76-2.18) vs 2.33 (95% CI, 2.06-3.01); and doubling of serum creatinine level, 2.03 (95% CI, 1.87-2.19) vs 1.92 (95% CI, 1.78-2.08). Receiver operating characteristic curves showed almost identical performance for ACR and PCR for the 3 outcome measures. Adjusted HRs for ACR and PCR were similar when derived from random urine samples or timed collections and compared with 24-hour total protein and albumin excretion for each outcome measure.
This is a retrospective study.
Total proteinuria and albuminuria perform equally as predictors of renal outcomes and mortality in patients with CKD. ACR and PCR were as effective as 24-hour urine samples at predicting outcomes and are more convenient for patients, clinicians, and laboratories. Both ACR and PCR stratify risk in patients with CKD.
Our study shows that total urinary protein and albumin excretion are equally powerful predictors of all-cause mortality and renal outcomes in patients with CKD attending a hospital kidney clinic. We showed this when assessing albuminuria with ACR or total proteinuria with PCR derived from a random urine sample and confirmed in a large subgroup with both ACR and PCR derived from a timed urine collection and 24-hour urine albumin and total protein excretion results. The analysis held true whether PCR and ACR were assessed as continuous or categorical variables. Unexpectedly, PCR also performed well at low levels (PCR of 150-500 mg/g, equivalent to protein excretion of 0.15-0.5 g/d of total proteinuria), for which albuminuria traditionally has been seen as the superior marker of risk.
We reviewed the literature for other studies directly comparing albuminuria and total proteinuria (MEDLINE 1950-2009 using the search terms albuminuria, proteinuria, mortality, renal replacement therapy). The Australian Diabetes, Obesity and Lifestyle (AusDiab) Study20 compared albuminuria with total proteinuria in a cross-sectional study of the general population, and 2 groups studied a renal population, Collier et al21 and Birmingham et al,22 who focused exclusively on lupus nephritis. None of these reports assessed patient outcomes. We identified one study comparing albuminuria and nonalbumin proteinuria as a predictor of outcomes in a kidney transplant population,23 which found that both were independent predictors of transplant loss and provided different information. We could identify no study examining outcomes in a CKD population.
Identification and quantification of proteinuria is essential to identify those at high risk of progression to end-stage renal disease.4 Albuminuria usually is measured in patients with diabetes, and most intervention studies have used albuminuria.24, 25, 26 Some guidelines also recommend using albuminuria in patients with nondiabetic kidney disease,13, 14 but others recommend using total proteinuria.15, 16 It is notable that the benefits of interventions in patients with nondiabetic kidney disease are based on studies that used total proteinuria rather than albuminuria.27 This report shows that both are equally predictive of patient outcomes at all levels of proteinuria in a mixed population of patients attending a general nephrology clinic, who predominantly have nondiabetic kidney disease.
There is extensive literature linking albuminuria with cardiovascular disease and mortality in both renal and general populations.5, 11, 28, 29, 30 The relationship shown in this report between total proteinuria and all-cause mortality is less well documented. Several studies31, 32, 33, 34 have shown a link between proteinuria and mortality; however, these studies assessed proteinuria using urinary dipstick testing, which predominantly measures albuminuria and is semiquantitative at best.35 One Finnish study of 1,056 patients with type 2 diabetes mellitus and 1,375 nondiabetic participants showed an association between spot urine concentration of total protein (ie, not adjusted for creatinine concentration) and subsequent cardiovascular and all-cause mortality.12
In urine, total protein is composed predominantly of albumin, but also of physiologic proteins (such as Tamm-Horsfall protein) and other nonalbumin proteins of various molecular weights. Proportions of these proteins vary widely in pathologic states, and nonalbumin proteins are less well defined compared with albumin. There is less inter- and intralaboratory variation in albumin assays than in total protein assays,35 and efforts are underway to standardize the albumin assay across laboratories.36 However, ACR is 2-10 times more costly than PCR. Given the technical challenges, it perhaps is surprising that PCR performed as well as ACR in predicting risk in our study. High molecular weight proteinuria correlates more strongly with rate of progression of kidney disease than intermediate- or low-molecular-weight or even total proteinuria. This is believed to result from increased tubular toxicity, although an alternative hypothesis would be that this finding is a consequence of loss of glomerular size selectivity.37 Furthermore, there is substantial variation in the amount of nonalbumin proteinuria between individuals at clinically significant levels of albuminuria. We previously have found that using only ACR to identify patients with significant proteinuria (protein excretion >1 g/d) would lead to more than one-fifth of patients (22%) being undetected who would have been identified using PCR as a result of high proportions of nonalbumin proteinuria.38 Nonalbumin proteinuria may carry some additional prognostic significance that is not captured by measuring albumin alone.
A subsidiary finding in our study was that PCR and ACR were as powerful as 24-hour urine protein and albumin excretion at predicting patient outcomes, respectively. Traditionally, 24-hour urine samples have been seen as the gold-standard method to measure total proteinuria or albuminuria; however, spot urine samples are more convenient for patients, clinicians, and laboratories. The ability of spot urine PCR and ACR to predict 24-hour total proteinuria and albuminuria has been extensively investigated and shown to be accurate, reliable, and reproducible.39 We could find no other study examining the comparative performance of ACR and PCR and 24-hour urine samples at predicting mortality and renal outcomes. PCR and ACR potentially could be superior for 2 reasons. First, 24-hour urine collections are difficult for patients to collect accurately; thus, spot urine samples may represent a more accurate estimate of true 24-hour urine protein excretion. Second, urinary creatinine also may be contributing to the predictive power of the test. Creatinine excretion correlates with muscle mass40; therefore, malnutrition or muscle wasting would lead to a higher PCR or ACR for any given level of urinary protein excretion and may contribute to the risk. In a retrospective study, we cannot differentiate between these explanations. Furthermore, this analysis should be replicated in a study comparing random spot urine samples with 24-hour urine samples.
Our study has several limitations. It was retrospective and we therefore cannot confirm the causation of the observed relationships. Our lack of complete data for race and primary kidney disease has hampered our ability to produce subgroup analyses. Twenty-four-hour urine collections were available for approximately one-third of patients. There may have been drift in the assays during such a prolonged period. However, this will affect all assays and we have no reason to expect a systematic bias. These issues affect our assessment of ACR and PCR equally. The relationships shown may apply to only the assays used in our study. However, the strengths of this study are the large numbers and the representative nature of the population; an unselected adult population attending a general nephrology clinic. It is not representative of primary care-based CKD populations, which may have a lower prevalence of proteinuria and different age distribution.
Further research is required to clarify the roles of total proteinuria and albuminuria as predictors of patient outcomes. Results of our study should be replicated prospectively and examined in other populations and subgroups using other total proteinuria and albuminuria assays. The importance of specific nonalbumin proteins in urine should be examined both for prognostication and to shed light on the underlying pathophysiologic process.
In conclusion, total proteinuria and albuminuria perform equally as predictors of renal outcomes and mortality in patients with CKD. ACR and PCR were as effective as 24-hour urine samples at predicting outcomes and are more convenient for patients, clinicians, and laboratories. Both ACR and PCR stratify risk in patients with CKD.