Logo of nihpaAbout Author manuscriptsSubmit a manuscriptNIH Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
Kidney Int. Author manuscript; available in PMC 2009 Sep 16.
Published in final edited form as:
PMCID: PMC2745082

Urinary cystatin C as an early biomarker of acute kidney injury following adult cardiothoracic surgery


There is a need to develop early biomarkers of acute kidney injury following cardiac surgery, where morbidity and mortality are increased by its presence. Plasma cystatin C (CyC) and plasma and urine Neutrophil Gelatinase Associated Lipocalin (NGAL) have been shown to detect kidney injury earlier than changes in plasma creatinine in critically ill patients. In order to determine the utility of urinary CyC levels as a measure of kidney injury, we prospectively collected plasma and urine from 72 adults undergoing elective cardiac surgery for analysis. Acute kidney injury was defined as a 25% or greater increase in plasma creatinine or renal replacement therapy within the first 72 hours following surgery. Plasma CyC and NGAL were not useful predictors of acute kidney injury within the first 6 hours following surgery. In contrast, both urinary CyC and NGAL were elevated in the 34 patients who later developed acute kidney injury, compared to those with no injury. The urinary NGAL at the time of ICU arrival and the urinary CyC level 6 hours after ICU admission were most useful for predicting acute kidney injury. A composite time point consisting of the maximum urinary CyC achieved in the first 6 hours following surgery outperformed all individual time points. Our study suggests that urinary CyC and NGAL are superior to conventional and novel plasma markers in the early diagnosis of acute kidney injury following adult cardiac surgery.

Keywords: cystatin C, acute kidney injury, biomarker, cardiac surgery, Neutrophil Gelatinase Associated Lipocalin

Acute kidney injury (AKI) is a common and serious complication of cardiothoracic surgery;1 depending on the definition of AKI used, it may occur in over 40% of adults, with 1–5% requiring renal replacement therapy (RRT).29 In cardiac surgery patients, postoperative increases of serum creatinine of 20–25% from preoperative baselines are associated with increased morbidity and mortality.6,1012 The mortality associated with severe, RRT-requiring AKI after cardiac surgery can be has high as 60%, depending on the definition of the postoperative period and type of cardiac surgery performed.2,3,57,9,13,14 It is clear that the risk of both AKI and severe, RRT-requiring AKI is greatest in those with preoperative renal dysfunction, particularly among those undergoing more complex procedures.1,1517 Despite a variety of clinical tools to predict AKI and stratify its severity following cardiothoracic surgery, there have been no dramatic changes in AKI incidence, severity, or patient outcomes in this high-risk setting, or in others.1,15,1821 Unlike experimental models, in which primary or early secondary prophylaxis for AKI has been successful with a variety of agents, unsuccessful clinical trials of such therapies have generally randomized a heterogeneous population of patients with markedly elevated plasma creatinine values and established AKI. Validation of one or more biomarkers that are diagnostic in the early perioperative period would enrich trials of early therapy of evolving AKI in this setting. Similarly, if such a biomarker panel could also predict AKI severity, and in particular whether RRT is likely required, then accurate stratification of patients to receive pharmacotherapy or early RRT in randomized trials would be possible.

Serum creatinine, the currently accepted ‘gold standard’ to diagnose AKI, is a delayed and inadequate marker of acute changes in renal function. In AKI, serum creatinine elevation that reflects the development and severity of kidney damage does not occur until days after renal tubular injury has begun.14,22 In the setting of cardiac surgery, it is common for the serum creatinine to decrease in the immediate postoperative period, even in those who go on to develop severe AKI requiring RRT, further diminishing the diagnostic utility of serum creatinine elevation to detect AKI in the early postoperative period.23 Similarly, detection of postoperative oliguria is not a sensitive or specific tool to diagnose postoperative AKI or predict its potential reversibility or severity. Numerous past attempts to treat AKI have failed, conceivably because they were started too late and in the presence of established acute tubular necrosis (ATN).24 The failure to identify one or more reliable markers of early renal injury in serum or urine to serve as a ‘renal troponin’ is one of the major impediments to progress in AKI therapy, in comparison to the improvement in the outcome of acute coronary syndromes or chronic kidney disease in recent years.2528

International societies have made several aspects of AKI research top priorities, including discovery and validation of AKI biomarkers for diagnosis prior to the rise in serum creatinine.29 Recent years have seen several studies investigating novel serum and urine AKI biomarkers in a variety of settings. Coca et al.30 recently performed a systematic literature review evaluating 31 studies that assessed 21 unique serum and urine biomarkers of AKI. Although they note that several of these biomarkers performed well at a variety of tasks, including diagnosing early AKI (serum cystatin C (CyC), urinary neutrophil gelatinase-associated lipocalin (NGAL) and interleukin-18) as well as predicting mortality (urine kidney injury molecule-1 and interleukin-18), they concluded that these markers need validation in larger studies.

Recently, serum CyC was shown to detect AKI earlier than serum creatinine in critically ill patients.27,31 CyC is a nonglycosylated 13 kDa basic protein that is a member of the cystatin superfamily of cysteine protease inhibitors. It is produced by all nucleated cells, unaffected by muscle mass (unlike creatinine). It was long thought that its production rate was similarly unaltered by inflammatory conditions,3234 although recent data demonstrate that in the absence of chronic kidney disease, serum CyC may be associated with inflammatory biomarkers in an elderly, ambulatory population.35 CyC is excreted by glomerular filtration, then undergoes essentially complete tubular reabsorption and catabolism (without secretion), so that it is not normally found in urine in significant amounts.27,36,37 This is in agreement with recent findings that demonstrate that elevated levels of urinary CyC may reflect tubular dysfunction and tubulointerstitial disease.38

NGAL, a 25 kDa member of the lipocalin family, is markedly upregulated in the early postischemic mouse and rat kidney.39 Serum and urine NGAL levels are elevated earlier than serum creatinine in the setting of delayed graft function following kidney transplantation40 and percutaneous coronary intervention.41 Mishra et al.42 demonstrated in children undergoing cardiac surgery that NGAL concentrations increase in serum and urine within 2 h post-cardiopulmonary bypass, preceding the serum creatinine elevation, in those who go on to develop AKI. There was no increase in children with stable perioperative renal function.42,43 Wagener et al.23 similarly found that urinary NGAL is useful in the early diagnosis of AKI in adults after cardiac surgery, although the test performance was significantly inferior in this population to the seminal pediatric study (area under the curve (AUC) under the receiver operating characteristic (ROC) curve for urinary NGAL was 0.8 at 18 h, but only 0.68 at 1 h and 0.74 at 3 h postoperation, whereas it was 0.998 at 2 h and 1.0 at 4 h in the Mishra study). Taken together with the fact that only limited numbers of cardiac surgery patients (adult or pediatric) have been studied, it is clear that the role of NGAL in the diagnosis of perioperative AKI requires further investigation.


Patient characteristics

A total of 262 patients were screened for study enrollment. Of them 162 were excluded because they met one or more exclusion criteria and 49 declined to participate. In total 81 subjects consented for the study, of whom 73 subjects completed the protocol; the most common reason for failure to complete the protocol despite informed consent was an unanticipated preoperative cardiac catheterization within 24 h of surgery. Of these 73 individuals, 1 died in the operating room, had no postoperative data, and was excluded from the analysis. The clinical characteristics of the remaining 72 subjects are listed in Table 1.

Table 1
Clinical characteristics of subjects (n=72)

Study end points and patient outcomes

Of the 72 study subjects, 34 (47.2%, 95% confidence interval (CI, 35.3, 59.3%)) of patients developed postoperative AKI (the primary study end point) as defined by a peak increase of plasma creatinine of greater than or equal to 25% (from preoperative baseline) (Figure 1a) or need for RRT (Table 2) within 3 days of the surgery. The mean preoperative plasma creatinine for all 72 subjects was 1.25 mg/100 ml (median = 1.08 mg/100 ml), consistent with their mean preoperative Modification of Diet in Renal Disease-estimated glomerular filtration rate (eGFR) of 70.6 ml/min per 1.73 m2 body surface area. Baseline plasma creatinine (mg/100 ml; median (interquartile range)) was 1.05 (0.90–1.50) in the group that did not develop AKI (n = 38), not significantly different than the AKI group (1.14 (0.92–1.24), n = 34, P = 0.86). There was also no significant difference in baseline eGFR values between the two groups (P = 0.87; Table 1).

Figure 1
Plasma creatinine, cystatin C, and NGAL over time
Table 2
RRT cases, indications, and outcomes

The peak plasma creatinine (mg/100 ml) within 72 h postoperatively (‘72 h Max’) was significantly greater, by definition, in the AKI group than in the group that did not develop AKI (1.65 (1.28–2.22) vs 1.15 (1.06–1.48), respectively; P<0.001). Furthermore, peak plasma creatinine increased with increasing AKI severity (P<0.001): no AKI (1.15, 1.06–1.48), AKI without RRT (1.56, 1.26–1.84), and RRT (2.84, 1.42–5.41). Of the 34 AKI subjects, 13 (38.2%) had a ≥50% increase in their plasma creatinine without receiving RRT. Of these 34 patients, 7 (20.6%) went on to require RRT, all initiated with continuous venovenous hemodialysis. Clinical features, indications, and timing of initiation of the RRT cases are in listed in Table 2. Four (11.8%) of these AKI subjects died during their postoperative hospital course, with three of the four deaths in those requiring RRT; none of the cohort without AKI died during their hospitalization (P = 0.045). Similarly, there were significantly longer intensive care unit (ICU) stays and duration of hospitalization in the AKI group (Table 1). Finally, the AKI group had a trend towards longer mean cardiopulmonary bypass pump (CPB) times than those without AKI (220.4 vs 180.2 min, respectively; P = 0.065; Table 1).

Plasma cystatin C

Similar to plasma creatinine (Figure 1a), plasma CyC decreased from preoperative values at the initial postoperative time points (Figure 1b). There was no difference between the preoperative baseline plasma CyC values of those with and without AKI (P = 0.92; Table 3), or in a three-group comparison (no AKI, AKI without RRT, RRT; P = 0.55). There was no significant difference between the maximum plasma CyC values of those with and without AKI in the early postoperative period (post-CPB, ICU arrival, and 6 h ICU time points: the ‘early composite’ period), although there was a trend toward higher values in the AKI group (P = 0.071; Table 3). This composite time point represents a period approximately 24 h prior to the diagnosis of AKI. The trend toward higher plasma CyC values in the AKI group continued subsequently, so that there was a significantly higher 72 h peak value for plasma CyC in those who developed AKI compared to those without AKI (Table 3), and evidence for an overall difference in plasma CyC change over time (interaction P<0.01 from repeated-measures analysis of variance, ANOVA).

Table 3
Plasma and urinary cystatin and NGAL values in the perioperative period

We generated ROC curves for plasma CyC to detect AKI, a plasma creatinine elevation of at least 25% or RRT, at varied time points and combinations of time points throughout the protocol. The results are mentioned in Table 4a. Plasma CyC was not a useful early predictor of the development of AKI. None of the time points examined demonstrated significantly better diagnostic ability than what would be expected by chance, as indicated by the 95% CIs all including 0.5. The ROC curve for the maximum value of plasma CyC during the early composite period can be found in Figure 2.

Figure 2
ROC curves for the maximum early composite
Table 4
AUC for ROC curves of CyC and NGAL for predicting AKI

Urinary cystatin C

Urinary CyC values are reported as the urinary concentration indexed to urinary creatinine excretion (subsequently referred to simply as urinary CyC and expressed in mg cystatin per gram of creatinine (mg/g)). At preoperative baseline, there was minimal urinary CyC excretion in the entire cohort.

There was no difference in the preoperative baseline urinary CyC concentrations between groups, whether looking at the cohort dichotomously (no AKI vs AKI, P = 0.42; Table 3) or in three groups (no AKI, AKI without RRT, and RRT; P = 0.49; Figure 3). The urinary concentration of CyC increased in all individuals following cardiac surgery, but the magnitude of these changes over time differed significantly by group (interaction P<0.01 from repeated-measures ANOVA). Subsequent post hoc comparisons indicated that as early as the first postoperative time point (while the patient is still in the OR), those who did not develop AKI increased their urinary CyC concentration by almost eightfold, but those who developed AKI increased on average by over 147-fold (P = 0.007). As further evidence of these differences, patients who developed AKI had both higher early composite postoperative maximum and 3-day maximum urinary CyC concentrations when compared to those without AKI (P<0.001 for both; Table 3). Figure 3 shows the early perioperative urinary CyC concentrations for the three groups (No AKI, AKI without RRT, and RRT). As with the two group comparison, changes over time differed significantly across groups (interaction P<0.01 from repeated-measures ANOVA). Although the most dramatic increase was seen in the RRT group, urinary CyC concentrations increased significantly according to the severity of AKI (no AKI < AKI without RRT < RRT) at each of the early postoperative time points (P<0.01).

Figure 3
Urinary cystatin C excretion over time

Exploratory subgroup analysis found that although there was very little difference in their baseline values (P = 0.32), women (n = 21) had significantly higher early postoperative maximum urinary CyC concentrations compared to men (P = 0.037). In addition, although there was no difference in their preoperative values (P = 0.52), those who had off-pump surgery (n = 8) had lower early composite postoperative maximum urinary CyC concentrations than those who were on CPB (0.107 (0.0681–0.392) vs 0.415 (0.12–1.41), respectively; P = 0.049). Among those who had CPB, there was a significant correlation between duration of CPB and maximum urinary CyC excretion during the early postoperative period (r = 0.30, P = 0.018).

ROC curves were constructed to test the ability of urinary CyC to predict AKI. The results are given in Table 4a. Urinary CyC at the 6 h ICU time point (AUC = 0.724, 95% CI (0.601–0.846); P = 0.002) and the maximum early composite value (AUC = 0.734, 95% CI (0.617–0.850); P<0.001; Figure 2) were excellent predictors of AKI after cardiac surgery. A sensitivity and specificity analysis for selected values of urinary CyC as a predictor of AKI development was performed. Positive and negative likelihood ratios were calculated (Table 5a).

Table 5
Performance of urinary CyC and NGAL to predict AKI

To examine whether urinary CyC had unique predictive ability beyond that provided by more standard risk factors (that is, age, race, gender, diabetes, CPB time, baseline eGFR, and baseline creatinine), logistic regression models were fit with and without urinary cystatin and the area under the ROC curves from these models was compared. The early composite maximum was an independent predictor of AKI development with an AUC of 0.725 compared to an AUC of 0.604 (P = 0.036). These data suggest that urinary CyC can aid in the early detection of AKI following cardiac surgery.

Fractional excretion of sodium

When comparing the fractional excretion of sodium (FENA) over time between those with and without AKI, there was no significant difference between these groups at baseline or any subsequent time point (interaction P = 0.51 and group main effect P = 0.71 from repeated-measures ANOVA). Nor was there an effect present when comparing the maximum or minimum value over the first 3 postoperative days or over the early composite postoperative period (data not shown).

Fractional excretion of urea

Similar to the FENA, there was no significant difference in the baseline fractional excretion of urea (FE Urea) values when comparing those with and without AKI (P = 0.26). There was evidence for a difference in FE Urea levels over time (interaction P = 0.067 and group main effect P<0.01 from repeated-measures ANOVA, data not shown). As a result, a statistically significant difference in the minimum value over the 3-day postoperative period (P<0.001) and the minimum over the early composite period (P = 0.004) was found, with those developing AKI having lower FE Urea values. From post hoc comparisons at individual time points, a significant difference between groups did not appear until the 6 h ICU time point and persisted through day 3 (P = 0.041 at 6 h ICU to P = 0.002 at day 2). Although the magnitude of the difference varied, those with AKI maintained consistently lower FE Urea values. A similar effect was seen when comparing the three groups; there was a negative correlation between the minimum FE Urea values and severity of AKI (no AKI>AKI without RRT>RRT; P<0.01).

Plasma NGAL

There was no significant difference in the baseline plasma NGAL values of those with and without AKI (P = 0.83) or in the three-group comparison (P = 0.26). Unlike plasma creatinine and CyC, plasma NGAL increased from preoperative values in the early postoperative period regardless of AKI outcome (interaction P = 0.93 and group main effect P = 0.79 from repeated-measures ANOVA; Figure 1c). As such, there was no difference between the maximum plasma NGAL value in those with and without AKI over the early composite (P = 0.60) or 3-day postoperative period (P = 0.54). A similar lack of effect was found when comparing across three groups (no AKI, AKI without RRT, and RRT). ROC curves were generated for plasma NGAL to detect AKI at varied time points and combination of time points throughout the protocol (Table 4b). Plasma NGAL was not a useful early marker of the development of AKI; at best the AUC (95% CI) was 0.536 (0.399–0.672); P = NS at the early composite time point (Figure 2).

Urinary NGAL

Urinary NGAL values are reported, as previously published,42 as the urinary concentration indexed to creatinine excretion; subsequently referred to as urinary NGAL. There was minimal difference in the preoperative baseline NGAL concentrations between those with and without AKI (P = 0.71; Table 3) and between the three groups (no AKI, AKI without RRT, and RRT; P = 0.41; Figure 4). Urinary NGAL increased in all patients postoperatively, and there was evidence for a difference between the no AKI and AKI groups in urinary NGAL levels over time (interaction P = 0.068 and group main effect P<0.01 from repeated-measures ANOVA). On the basis of post hoc comparisons, significant differences were found at the ICU arrival (P = 0.003), 6 h ICU (P = 0.004), and day 1 (P<0.001) time points. Patients with AKI had higher early composite (P = 0.006; Table 3) as well as 3-day maximum urinary NGAL values (P = 0.002). Figure 4 shows the early perioperative urinary NGAL concentrations across the three groups (no AKI, AKI without RRT, and RRT). As with the two-group comparison, there were differences between groups over time (interaction P = 0.085 and group main effect P<0.01 from repeated-measures ANOVA). Although the most dramatic increase was seen in the RRT group, there was a significant positive correlation between urinary NGAL concentrations and the severity of AKI (no AKI < AKI without RRT < RRT) at each of the early postoperative time points (P<0.01) except for post-CPB (P = 0.068).

Figure 4
Urinary NGAL excretion over time

ROC curves were constructed to test the ability of urinary NGAL to predict AKI development (Table 4b). Urinary NGAL at the ICU arrival had the best performance (AUC = 0.705, 95% CI (0.581–0.829); P = 0.003) although not significantly better than the other individual time points or the maximum early composite (AUC = 0.691, 95% CI (0.567–0.815); P = 0.006; Figure 2). A sensitivity and specificity analysis for urinary NGAL as a test for predicting AKI development was performed. Similar to CyC, several cutoff values were chosen and positive and negative likelihood ratios were determined (Table 5b).

Comparison of cystatin C and NGAL

The area under the ROC curves for the plasma and urine of the two markers were compared. The early composite maximum was used in all cases. No statistically significant differences between the two markers were found (serum P = 0.36, urine P = 0.51), although in each case CyC had better predictive ability than NGAL. In addition, whether the combination of the urine levels of the two markers was better than either on its own was tested but did not yield a significant improvement (P = 0.69). However, the limited power available for these comparisons is acknowledged.


We hypothesized that concentrations of CyC in plasma (a GFR marker) and urine (a proximal tubular injury marker) would be early biomarkers of AKI following cardiac surgery, preceding significant plasma creatinine increases. Plasma CyC levels increase before creatinine levels in patients with progressive chronic kidney disease.32,34 Similarly, emerging data suggest that plasma CyC levels increase 1–2 days before plasma creatinine in patients developing AKI in a variety of settings,32 including radiocontrast nephropathy,44 and AKI in critically ill patients.31,45 This phenomenon is explained by the shorter plasma elimination half-life of CyC.46 To date, thyroid dysfunction, corticosteroids, and inflammatory biomarkers have been shown to alter serum CyC levels independent of GFR;47 although there is some dispute in the literature as to whether serum CyC concentration is altered by body/muscle mass and inflammation.35,48,49 Limited data suggest that appearance of increased concentrations of CyC in urine may be a sensitive marker of renal tubular injury,27,36,37 and may correlate with severity of AKI.27

In this study, we have shown that renal dysfunction following cardiac surgery is associated with increased urinary and plasma CyC concentrations. Similar to creatinine, plasma CyC initially decreases in the early postoperative period. Across the entire cohort, the plasma creatinine decreased by an average of 8% at the post-CPB time point, when the plasma CyC decreased by an average of 22%. This dramatic decrease in both markers makes their clinical utility to define AKI difficult to interpret in the hours immediately following cardiac surgery. Plasma NGAL, a marker of inflammation, appears to increase in all individuals following surgery and does not appear to suffer from this dilutional effect. Of course, as both CyC and creatinine can serve as markers for GFR, it is not surprising that both values subsequently rebound and increase more significantly in those who go on to develop AKI. However, contrary to the current literature in other forms of AKI, plasma CyC did not appear to be a useful AKI biomarker in the early postoperative period following cardiac surgery. Although the AUC values for plasma CyC were all above 0.60, none of the promising early time points had a value greater than 0.70, and all of the 95% CIs included 0.50, indicating that it was not significantly better than what would be expected by chance (Table 4a). This is in part because the mean plasma CyC for those who go on to develop AKI, depicted in Figure 1b, does not increase 25% (from preoperative baseline) until day 2, and does not increase 50% until day 3. Although several studies have reported that plasma CyC increases earlier than plasma creatinine in the setting of AKI,31 none of these were conducted in the setting of cardiac surgery. In summary, based on these findings, plasma CyC does not serve as a useful early AKI biomarker in this setting, perhaps confounded by the effects of perioperative hemodilution on plasma biomarker concentrations.

Similar to CyC, plasma NGAL did not show much promise as an AKI biomarker in the early postoperative period. All of the AUC values were less than 0.60, and as with plasma CyC they all had 95% CIs that included 0.50 (Table 4b). This was relatively surprising, as the Mishra study found that plasma NGAL could detect AKI before plasma creatinine elevation following pediatric cardiac surgery (AUC under the ROC curve was 0.998 at 2 h post-CPB).42 Several possibilities exist for the discrepancy between our data and the earlier published report. Our samples were immediately frozen at −80 °C and analyzed in the same lab as the Mishra paper. However, we chose to measure NGAL in plasma rather than serum (as in prior publications). Although unpublished data from their lab demonstrate no significant difference between plasma and serum levels of NGAL, the difference in our approach is acknowledged. We also used a different definition of AKI than the Mishra study, which used a 50% serum creatinine increment cutoff. More importantly, our adult patient population differed significantly from the children of the Mishra study. The similarity of our urine NGAL findings to those of Wagener et al., who also studied adult cardiac surgery patients, is supportive of this hypothesis (of note, Wagener et al. did not measure serum NGAL, only urine). It remains to be determined whether inflammation-induced increases in serum NGAL following cardiac bypass in adults with numerous comorbidities (our study subjects were chronically ill individuals, on average 61.3 years old, 23.6% were diabetics, 27.8% were having their second (or more) cardiac surgery) result in inferior performance of this marker in this population compared to children. Processes of care also differ between the adult and pediatric cardiac surgery populations; for example, although we excluded patients who received radiocontrast within 24 h of surgery, such procedures are not typically performed in the preoperative period before pediatric cardiac surgery, and may have had some effects in the adult studies. Operative factors may also have played a major role in this phenomenon. It should be noted that the mean (± s.e.) CPB time for those who had bypass was 199 ± 11 min. This is a 2- to 3-fold increase from the pump times reported in previous NGAL literature, which is another plausible explanation for the marked increased in plasma NGAL seen in all of our study subjects, irrespective of AKI development. Preexisting vascular and renal disease may also explain the poorer performance of urinary NGAL as an early AKI biomarker in the adults studied by Wagener and our cohort, compared to the pediatric population. Of course, the growing literature studying AKI biomarkers in both pediatric and adult cardiac surgery patients will continue to expand our knowledge regarding the impact of such clinical variables on test performance in validation studies.

Similar to urine NGAL, this study showed that AKI following cardiac surgery is associated with an early postoperative increase in urinary CyC excretion. Urinary CyC concentrations have not been widely studied; values are customarily indexed to the urine creatinine. We have clearly demonstrated that an early and persistent increase of urinary CyC after cardiac surgery correlates with the development and severity of AKI. The latter finding is similar to previous work by Herget-Rosenthal et al.27, who demonstrated that increased urinary CyC excretion in the setting of AKI correlated with the requirement for RRT. Our data suggest that urinary CyC concentrations have utility beyond the ability to provide an earlier diagnosis of AKI. The observed correlation with AKI severity offers the potential to stratify cases of ATN, clearly separating those with established ATN bound for RRT from those with lesser tubular injury or reversible, prerenal azotemia. This point is underscored by the comparatively poor performance in this regard of two standard urine chemistry indices in our study. Whereas the FENA was simply not useful to predict AKI or stratify severity, the FE Urea was actually misleading in our cohort. Specifically, the FE Urea values were significantly lower at several of the early postoperative time points in those who required RRT compared to the rest of the group (with a negative correlation between FE Urea values and AKI severity); yet this same RRT group had extremely elevated urinary CyC concentrations at these same early time points. This increase in urinary CyC suggested the presence of tubular injury, whereas the low FE Urea values suggested that these AKI cases had a ‘prerenal’ etiology. Further larger studies that include other AKI biomarkers will be of great interest as we seek further explanation of this phenomenon.

Our study has some limitations. It is a single-center study of limited size, and thus should be confirmed in a larger multicenter study. We did not measure GFR directly, but plasma concentrations of two GFR markers (creatinine and CyC) exhibited similar patterns over time (Figures 1a and b). We defined AKI as a 25% increase in plasma creatinine within 72 h of surgery. Although some previous studies have used a 50% increase as their definition, emerging data suggest that smaller changes have significant clinical impact including increased ICU and hospital length of stay as well as increased morbidity and mortality.6,10 Regardless, urinary CyC concentrations were significantly increased at the first postoperative time point in those who would develop AKI compared to those who did not, although plasma creatinine did not peak until day 2 (a full 48 h after ICU arrival; Figure 1a), and AKI could not be diagnosed by plasma creatinine elevation of any degree before 24–28 h postoperatively. We anticipate that a larger trial could more definitively establish a cutoff value for urinary CyC that is predictive of AKI. Although we could not precisely distinguish prerenal azotemia from ATN in all cases of AKI, our data suggest that marked elevation of urinary CyC (but not FE Urea or FENA) in the early postoperative period is likely diagnostic of AKI caused by severe, dialysis-requiring ATN.

In summary, our study found that increased urinary CyC excretion is a useful early biomarker of AKI in adults following cardiac surgery, and the extent of excretion of this marker correlates with the severity of AKI. Urinary NGAL is a similarly useful early biomarker of AKI in this setting, although it does not perform as well as in some previously published studies. Urinary CyC should be a component of urinary AKI marker panels assembled and tested to diagnose and evaluate AKI.

Materials and Methods

Patients and methods

We screened all patients admitted to the University of Chicago Cardiac Surgery service for elective surgery between August 2005 and August 2007. All patients eligible for enrollment were approached. Exclusion criteria included (1) age <18 years; (2) preexisting end-stage kidney disease receiving RRT or post-renal transplantation; (3) emergent cardiac surgery; (4) unstable renal function (change in serum creatinine of ≥0.2 mg per 100 ml within the past month, or oliguria <400 ml per day); (5) use of radiocontrast ≤24 h prior to the surgery; (6) preexisting hypothyroidism with a change in thyroid hormone dose in the past 2 weeks; and (7) patients receiving corticosteroid therapy with a change in their dose within the past 2 weeks. Written informed consent was obtained from all patients at the time of their enrollment. The study was approved by the institutional review board of the University of Chicago.

All patients were prospectively followed from the time of their enrollment. Blood and urine samples were collected simultaneously at predetermined time points; at the time of study enrollment, on the day of the surgery preoperatively (immediately post-anesthesia induction), postoperatively (after coming off CPB), upon arrival in the ICU, 6 h after arriving in the ICU, daily for the next 7 days (days 1–7), then weekly for the next 3 weeks (days 14, 21, and 28) and at the time of hospital discharge. Patients were considered to have completed the study at the time of their hospital discharge.

Blood samples were centrifuged at 3600 r.p.m. for 15 min and the plasma was stored at −80 °C. Urine samples were collected and some unspun urine was stored at −80 °C. In addition, as previously described,42 urine was centrifuged at 2000 r.p.m. for 5 min and the supernatants were aliquoted and stored at −80 °C. Sample analysis was batched and deferred until the time of study completion (hospital discharge).

Preoperative patient characteristics, significant intraoperative risk factors and peri- and postoperative complications were recorded on all patients. All therapeutic and clinical care decisions were made by the primary cardiac surgery service. In addition, the decision to initiate RRT or make other renal interventions was made by the primary service and the Nephrology Consult Service attending without involvement of the study investigators. A variety of clinical scoring systems1,21 were used to assess the preoperative risk for AKI.

The primary end point was the development of AKI, defined as a 25% or greater increase in the plasma creatinine from the preoperative baseline or the need for RRT within the first 3 postoperative days. Secondary end points included a 50% or greater increase in plasma creatinine from preoperative levels, in-hospital mortality, and length of ICU and hospital stays. Other variables recorded included age, sex, ethnicity, CPB time, cross-clamp time, past medical history (hypertension, congestive heart failure, diabetes, and so on), previous cardiothoracic surgery, preoperative creatinine clearances estimated by both Cockcroft and Gault and Modification of Diet in Renal Disease equations,50,51 serial plasma and urine creatinine, sodium and urea nitrogen.

Cystatin C enzyme-linked immunosorbent assay

The levels of CyC in plasma and urine were quantitated using an enzyme-linked immunosorbent assay (Human Cystatin C ELISA kits: BioVendor LLC, Candler, NC, USA). Briefly, kit-supplied plates were purchased precoated with polyclonal anti-human CyC-specific antibody. The plates were loaded with 100 μl of sample (urine or plasma at appropriate dilution) or standards (CyC concentration ranging from 200 to 10,000 ng/ml), and run according to the manufacturer's instructions. Assays were run in duplicate or triplicate, and read on a Bio-Tek Synergy HT plate-reader (BioTek Instruments, Winooski, VT, USA). The mean intraassay coefficient of variation was 3.9% for plasma CyC and 5.1% for urine CyC for batched samples prepared on the same day. The mean interassay coefficient of variation was 10.3%. Laboratory investigators were blinded to the sample source and clinical outcomes. Plasma and urine creatinine were measured by the alkaline picrate reaction, plasma and urine urea nitrogen were measured by the urease method, and plasma and urine sodium were measured with an ion-selective electrode (Beckman Unicell DxC 600, Beckman Coulter, Fullerton, CA, USA). FENA and FE urea were calculated as follows: FENA% = 100 × (UrineNa × PlasmaCr/PNa × UCr);FE urea% = 100 × (Urea × PlasmaCr/BUN × UCr).

Enzyme-linked immunosorbent assay for NGAL

Previously published protocols for the detection of NGAL from neutrophils52 were modified for this project. This protocol itself has been previously published.42 Briefly, microtiter plates were coated overnight at 4 °C with a mouse monoclonal antibody raised against human NGAL (HYB211-05; AntibodyShop, Gentofte, Denmark). All subsequent steps were undertaken at room temperature. Plates were blocked with buffer containing 1% bovine serum albumin, coated with 100 μl of samples (urine or plasma) or standards (NGAL concentrations ranging from 1 to 1000 μg/l), and incubated with a biotinylated monoclonal antibody against human NGAL (HYB211-01B; AntibodyShop) followed by avidin-conjugated horseradish peroxidase (Dako, Carpinteria, CA, USA). Tetramethylbenzidine substrate (BD Biosciences, San Jose, CA, USA) was added for color development, which was read after 30 min at 450 nm with a microplate reader (Benchmark Plus; Bio-Rad, Hercules, CA, USA). All measurements were made in triplicate in a blinded fashion.


Results are expressed as mean ± s.e. or median and interquartile range (25–75%). SigmaStat 3.5 (Systat Software Inc., San Jose, CA, USA) or Stata version 9 (StataCorp, College Station, TX, USA) were used for analyses. For continuous variables, a t-test or a Mann–Whitney U-test, as appropriate, was performed for comparison between two groups. When comparing more than two groups, ANOVA or the Kruskal–Wallis test was used. Spearman's rank correlation coefficients were utilized for further assessment of the correlation between marker levels and severity of AKI. For an overall comparison of change over time between groups, a repeated-measures ANOVA was performed with particular interest in the interaction between group and time. A χ2 or Fisher's exact test was used for analysis of categorical variables. A P-value of <0.05 was considered statistically significant. ROC curves were generated for the various markers at the postoperative, ICU admit and 6 h ICU time points, as well as the maximum value at these three time points (referred to as the early composite time point). The area under the ROC curve (AUC) was calculated as a measure of the utility of the CyC and NGAL concentrations as early markers of AKI. An area under the curve of 0.5 indicates a marker with accuracy no better than expected by chance, whereas a value of 1.0 signifies a biomarker with perfect diagnostic accuracy. The AUCs were compared using methods described by DeLong et al.53 In addition, logistic regression was employed to further explore the unique predictive ability of these markers.


Part of this work was presented at the American Society of Nephrology meeting in San Francisco, CA on 3 and 4 November 2007. This project was supported by the following grants to Dr Murray: NIH 1K23GM00713-01A1 (from NIGMS); National Kidney Foundation of Illinois Research Grant for the Young Investigator; University of Chicago General Clinical Research Center Award (PHS MO1RR000055); Dr Koyner was supported by NIH 2 T32 DK007510; and Dr Worcester was supported by NIH NIDDK PO1 56788.


Disclosure: P Devarajan—licensing agreements with Abbott Diagnostics and Biosite Inc. for developing NGAL as a biomarker of acute renal failure.


1. Wijeysundera DN, Karkouti K, Dupuis JY, et al. Derivation and validation of a simplified predictive index for renal replacement therapy after cardiac surgery. JAMA. 2007;297:1801–1809. [PubMed]
2. Abel R, Buckley M, Austen W, et al. Etiology, incidence and prognosis of renal failure following cardiac operations. Results of a prospective analysis of 500 consecutive patients. J Thorac Cardiovasc Surg. 1976;71:323–333. [PubMed]
3. Andersson L, Ekroth R, Bratteby L, et al. Acute renal failure after coronary surgery: a study of incidence and risk factors in 2009 consecutive patients. Thorac Cardiovasc Surg. 1993;41:237–241. [PubMed]
4. Chertow G, Levy E, Hammermeister K, et al. Independent association between acute renal failure and mortality following cardiac surgery. Am J Med. 1998;104:343–348. [PubMed]
5. Conlon P, Strafford-Smith M, WD W, et al. Acute renal failure following cardiac surgery. Nephrol Dial Transplant. 1999;14:1158–1162. [PubMed]
6. MoraMangano C, Diamondstone L, Ramsey J, et al. Renal dysfunction after myocardial revascularization: risk factors, adverse outcomes, and hospital resource utilization. Ann Intern Med. 1998;128:194–203. [PubMed]
7. Ostermann M, Taube D, Morgan C, et al. Acute renal failure following cardiopulmonary bypass: a changing picture. Intensive Care Med. 2000;26:565–571. [PubMed]
8. Tuttle K, Worrall N, Dahlstrom L, et al. Predictors of ARF after cardiac surgical procedures. Am J Kidney Dis. 2003;41:76–83. [PubMed]
9. Zanardo G, Michielon P, Paccagnella A, et al. Acute renal failure in the patient undergoing cardiac operation: prevalence, mortality rate, and main risk factors. J Thorac Cardiovasc Surg. 1994;107:1489–1495. [PubMed]
10. Lassnigg A, Schmidlin D, Mouhieddine M, et al. Minimal changes of serum creatinine predict prognosis in patients after cardiothoracic surgery: a prospective cohort study. J Am Soc Nephrol. 2004;15:1597–1605. [PubMed]
11. Loef B, Epema A, Smilde T, et al. Immediate postoperative renal function deterioration in cardiac surgical patients predicts in-hospital mortality and long-term survival. J Am Soc Nephrol. 2005;16:195–200. [PubMed]
12. Ryckwaert F, Boccara G, Frappier J, et al. Incidence, risk factors, and prognosis of a moderate increase in plasma creatinine after cardiac surgery. Crit Care Med. 2002;30:1495–1498. [PubMed]
13. Bove T, Calabro MG, Landoni G, et al. The incidence and risk of acute renal failure after cardiac surgery. J Cardiothorac Vasc Anesth. 2004;18:442–445. [PubMed]
14. Rosner MH, Okusa MD. Acute kidney injury associated with cardiac surgery. Clin J Am Soc Neprol. 2006;1:19–32. [PubMed]
15. Mehta RH, Grab JD, O'Brien SM, et al. Bedside tool for predicting the risk of postoperative dialysis in patients undergoing cardiac surgery. Circulation. 2006;114:2208–2216. quiz 2208. [PubMed]
16. Roques F, Nashef SA, Michel P, et al. Risk factors and outcome in European cardiac surgery: analysis of the EuroSCORE multinational database of 19,030 patients. Eur J Cardiothorac Surg. 1999;15:816–822. discussion 822–3. [PubMed]
17. Thakar C, Liangos O, Yared JP, et al. Predicting acute renal failure after cardiac surgery: validation and redefinition of a risk stratification algorithm. Hemodial Int. 2003;7:143–147. [PubMed]
18. Eriksen B, Hoff K, Solberg S. Prediction of acute renal failure after cardiac surgery: retrospective cross-validation of a clinical algorithm. Nephrol Dial Transplant. 2003;18:77–81. [PubMed]
19. Fortescue E, Bates D, Chertow G. Predicting acute renal failure after coronary bypass surgery: cross-validation of two risk-stratification algorithms. Kidney Int. 2000;57:2594–2602. [PubMed]
20. Parsonnet V, Dean D, Bernstein AD. A method of uniform stratification of risk for evaluating the results of surgery in acquired adult heart disease. Circulation. 1989;79(6 Part 2):I3–I12. [PubMed]
21. Thakar CV, Arrigain S, Worley S, et al. A clinical score to predict acute renal failure after cardiac surgery. J Am Soc Nephrol. 2005;16:162–168. [PubMed]
22. Murray P, Le Gall J, Dos Reis Miranda D, et al. Physiologic endpoints (efficacy) for acute renal failure studies. Curr Opin Crit Care. 2002;8:519–525. [PubMed]
23. Wagener G, Jan M, Kim M, et al. Association between increases in urinary neutrophil gelatinase-associated lipocalin and acute renal dysfunction after adult cardiac surgery. Anesthesiology. 2006;105:485–491. [PubMed]
24. Tang I, Murray P. Prevention of perioperative ARF: what works? Best Prac Res Clin Anesthesiol. 2004;18:91–111. [PubMed]
25. Bonventre JV, Weinberg JM. Recent advances in the pathophysiology of ischemic acute renal failure. J Am Soc Nephrol. 2003;14:2199–2210. [PubMed]
26. Dagher PC, Herget-Rosenthal S, Ruehm SG, et al. Newly developed techniques to study and diagnose acute renal failure. J Am Soc Nephrol. 2003;14:2188–2198. see comment. [PubMed]
27. Herget-Rosenthal S, Poppen D, Husing J, et al. Prognostic value of tubular proteinuria and enzymuria in nonoliguric acute tubular necrosis. Clin Chem. 2004;50:522–558. [PubMed]
28. Star RA. Treatment of acute renal failure. Kidney Int. 1998;54:1817–1831. [PubMed]
29. Nephrology TASo. The American Society of Nephrology Renal Research Report. J Am Soc Nephrol. 2005;16:1886–1903. [PubMed]
30. Coca SG, Yalavarthy R, Concato J, et al. Biomarkers for the diagnosis and risk stratification of acute kidney injury: a systematic review. Kidney Int. 2007;19:19. [PubMed]
31. Herget-Rosenthal S, Marggraf G, Husing J, et al. Early detection of acute renal failure by serum cystatin C. Kidney Int. 2004;66:1115–1122. [PubMed]
32. Laterza O, Price C, Scott M. Cystatin C: an improved estimator of glomerular filtration rate? Clin Chem. 2002;48:699–707. [PubMed]
33. Massey D. Commentary: clinical diagnostic use of cystain C. J Clin Lab Analysis. 2004;18:55–60. [PubMed]
34. Newman DJ, Thakkar H, Edwards RG, et al. Serum cystatin C measured by automated immunoassay: a more sensitive marker of changes in GFR than serum creatinine. Kidney Int. 1995;47:312–318. [PubMed]
35. Keller CR, Odden MC, Fried LF, et al. Kidney function and markers of inflammation in elderly persons without chronic kidney disease: the health, aging, and body composition study. Kidney Int. 2007;71:239–244. [PubMed]
36. Herget-Rosenthal S, Feldkamp T, Volbracht L, et al. Measurement of urinary cystain C by particle-enhanced nephelometric immunoassay: precision, interferences, stability, and reference range. Ann Clin Biochem. 2004;41(Part 2):111–118. [PubMed]
37. Uchida K, Cotoh A. Measurement of cystatin-C and creatinine in urine. Clin Chim Acta. 2002;323:121–128. [PubMed]
38. Herget-Rosenthal S, van Wijk JA, Brocker-Preuss M, et al. Increased urinary cystatin C reflects structural and functional renal tubular impairment independent of glomerular filtration rate. Clin Biochem. 2007;40:946–951. [PubMed]
39. Mishra J, Ma Q, Prada A, et al. Identification of neutrophil gelatinase-associated lipocalin as a novel early urinary biomarker for ischemic renal injury. J Am Soc Nephrol. 2003;14:2534–2543. [PubMed]
40. Parikh CR, Jani A, Mishra J, et al. Urine NGAL and IL-18 are predictive biomarkers for delayed graft function following kidney transplantation. Am J Transplant. 2006;6:1639–1645. [PubMed]
41. Bachorzewska-Gajewska H, Malyszko J, Sitniewska E, et al. Neutrophil-gelatinase-associated lipocalin and renal function after percutaneous coronary interventions. Am J Nephrol. 2006;26:287–292. [PubMed]
42. Mishra J, Dent C, Tarabishi R, et al. Neutrophil gelatinase-associated lipocalin (NGAL) as a biomarker for acute renal injury after cardiac surgery. Lancet. 2005;365:1231–1238. [PubMed]
43. Parikh CR, Mishra J, Thiessen-Philbrook H, et al. Urinary IL-18 is an early predictive biomarker of acute kidney injury after cardiac surgery. Kidney Int. 2006;70:199–203. [PubMed]
44. Rickli H, Benou K, Ammann P, et al. Time course of serial cystatin levels in comparison with serum creatinine after application of radiocontrast media. Clin Nephrol. 2004;61:98–102. [PubMed]
45. Delanaye P, Lambermont B, Chapelle J, et al. Plasmatic cystain C for the estimation of glomerular filtration rate in intensive care units. Intensive Care Med. 2004;30:980–983. [PubMed]
46. Filler G, Bokenkamp A, Hofmann W, et al. Cystatin C as a marker of GFR—history, indications, and future research. Clin Biochem. 2005;38:1–8. [PubMed]
47. Fricker M, Wiesli P, Brandle M, et al. Impact of thyroid dysfunction on serum cystatin C. Kidney Int. 2003;63:1944–1947. [PubMed]
48. Macdonald J, Marcora S, Jibani M, et al. GFR estimation using cystatin C is not independent of body composition. Am J Kidney Dis. 2006;48:712–719. [PubMed]
49. Shlipak MG. Cystatin C as a marker of glomerular filtration rate in chronic kidney disease: influence of body composition. Nat Clin Pract Nephrol. 2007;3:188–189. [PubMed]
50. Cockcroft DW, Gault MH. Prediction of creatinine clearance from serum creatinine. Nephron. 1976;16:31–41. [PubMed]
51. Levey AS, Bosch JP, Lewis JB, et al. A more accurate method to estimate glomerular filtration rate from serum creatinine: a new prediction equation. Modification of Diet in Renal Disease Study Group. Ann Intern Med. 1999;130:461–470. [PubMed]
52. Kjeldsen L, Koch C, Arnljots K, et al. Characterization of two ELISAs for NGAL, a newly described lipocalin in human neutrophils. J Immunol Methods. 1996;198:155–164. [PubMed]
53. DeLong ER, DeLong DM, Clarke-Pearson DL. Comparing the areas under two or more correlated receiver operating characteristic curves: a nonparametric approach. Biometrics. 1988;44:837–845. [PubMed]
PubReader format: click here to try


Related citations in PubMed

See reviews...See all...

Cited by other articles in PMC

See all...


  • MedGen
    Related information in MedGen
  • PubMed
    PubMed citations for these articles
  • Substance
    PubChem Substance links

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...