Am J Physiol Renal Physiol 307: F939F948, 2014.
First published August 20, 2014; doi:10.1152/ajprenal.00025.2013.
Comparison of serum creatinine and serum cystatin C as biomarkers to detect
sepsis-induced acute kidney injury and to predict mortality in CD-1 mice
Asada Leelahavanichkul,1,2,4* Ana Carolina P. Souza,1,2* Jonathan M. Street,1,2 Victor Hsu,1,2
Takayuki Tsuji,1,2 Kent Doi,1,2 Lingli Li,2 Xuzhen Hu,1,2 Hua Zhou,1,2 Parag Kumar,3
Jrgen Schnermann,2 Robert A. Star,1,2 and Peter S. T. Yuen1,2
1
Renal Diagnostics and Therapeutics Unit, National Institute of Diabetes and Digestive and Kidney Diseases, National
Institutes of Health, Bethesda, Maryland; 2Kidney Disease Branch, National Institute of Diabetes and Digestive and Kidney
Diseases, National Institutes of Health, Bethesda, Maryland, and 3Pharmacy Department, Clinical Center, National Institutes
of Health, Bethesda, Maryland; and 4Department of Microbiology, Faculty of Medicine, Chulalongkorn University, Bangkok,
Thailand
Submitted 14 January 2013; accepted in final form 14 August 2014
Leelahavanichkul A, Souza AC, Street JM, Hsu V, Tsuji T,
Doi K, Li L, Hu X, Zhou H, Kumar P, Schnermann J, Star RA,
Yuen PS. Comparison of serum creatinine and serum cystatin
C as biomarkers to detect sepsis-induced acute kidney injury
and to predict mortality in CD-1 mice. Am J Physiol Renal Physiol
307: F939 F948, 2014. First published August 20, 2014;
doi:10.1152/ajprenal.00025.2013.Acute kidney injury (AKI) dra-
matically increases sepsis mortality, but AKI diagnosis is delayed
when based on serum creatinine (SCr) changes, due in part, to
decreased creatinine production. During experimental sepsis, we com-
pared serum cystatin C (sCysC), SCr, and blood urea nitrogen (BUN)
to inulin glomerular filtration rate (iGFR) before or 318 h after cecal
ligation and puncture (CLP)-induced sepsis in CD-1 mice. sCysC had
a faster increase and reached peak levels more rapidly than SCr in
both sepsis and bilateral nephrectomy (BiNx) models. sCysC was a
better surrogate of iGFR than SCr during sepsis. Combining sCysC
with SCr values into a composite biomarker improved correlation
with iGFR better than any biomarker alone or any other combination.
We determined the renal contribution to sCysC handling with BiNx.
sCysC and SCr were lower post-BiNx/CLP than post-BiNx alone,
despite increased inflammatory and nonrenal organ damage biomark-
ers. Sepsis decreased CysC production in nephrectomized mice with-
out changing body weight or CysC space. Sepsis decreased sCysC
production and increased nonrenal clearance, similar to effects of
sepsis on SCr. sCysC, SCr, and BUN were measured 6 h postsepsis to
link AKI with mortality. Mice with above-median sCysC, BUN, or
SCr values 6 h postsepsis died earlier than mice with below-median
values, corresponding to a substantial AKI association with sepsis
mortality in this model. sCysC performs similarly to SCr in classify-
ing mice at risk for early mortality. We conclude that sCysC detects
AKI early and better reflects iGFR in CLP-induced sepsis. This study
shows that renal biomarkers need to be evaluated in specific contexts.
bilateral nephrectomy; glomerular filtration rate; survival; Kaplan-
Meier; receiver-operating characteristic curve
SERUM CREATININE (SCr) is currently used to detect and stage
acute kidney injury (AKI) and chronic kidney injury despite
well-known limitations. Creatine is synthesized primarily in
the liver and then transported to skeletal muscle for use in
storing ATP (3). Creatinine (molecular mass: 113 Da), a
metabolic end product of creatine, is released into plasma,
* A. Leelahavanichkul and A. C. P. Souza contributed equally to this work.
Address for reprint requests and other correspondence: P. Yuen, National
Institute of Diabetes and Digestive and Kidney Diseases, National Institutes of
Health, Bldg. 10, Rm. 3N108, 10 Center Drive, MSC 1268, Bethesda, MD
20892-1268 (e-mail: py@nih.gov).
http://www.ajprenal.org
Downloaded from http://ajprenal.physiology.org/ by 10.220.33.3 on July 11, 2017
freely filtered through the glomerulus, also secreted but not
absorbed by renal tubules. Whereas SCr is predominantly
determined by glomerular filtration rate (GFR), SCr is also
influenced by nonrenal factors that alter creatinine generation
(muscle mass, dietary creatine intake, and liver function) and
elimination (gastrointestinal excretion) (3). We previously
demonstrated that creatinine production is reduced in sepsis,
which limits the ability of SCr to detect and stage sepsis-
induced AKI (sepsis-AKI) (5). An endogenous molecule with
different properties could overcome these limitations, such as
serum cystatin C (sCysC; molecular mass: 13.3 kDa) (1) a
proteinase inhibitor that 1) prevents connective tissue destruc-
tion (24), 2) is constantly produced by most nucleated cells in
the body, 3) is freely filtered by the glomerulus, and 4) is then
entirely reabsorbed and catabolized in the proximal tubule (28,
31). sCysC has been proposed by some as a more ideal
endogenous biomarker of chronic kidney function (4, 12, 21,
28), although it is also affected by age, sex, muscle mass,
smoking, thyroid function, and malignancies (2).
Whether sCysC is a better biomarker of kidney injury is a
matter of controversy, and it may depend on the context of use
(14, 15). Within the context of chronic kidney disease, a
combined creatinine-CysC estimated GFR (eGFR) equation
can perform somewhat better than equations based on either
biomarker alone (15).
In the setting of human AKI, urinary CysC predicts AKI
48 72 h before SCr, whereas sCysC may detect AKI 12 days
earlier than SCr in intensive care unit patients who developed
AKI and were classified according to RIFLE criteria (13).
sCysC also outperforms SCr as an early biomarker of AKI in
the emergency room setting (29) and after cardiopulmonary
bypass in children (17).
But while some studies have shown the superiority of sCysC
as an early biomarker of AKI, other studies have shown that
sCysC performs as well as, or even worse, than SCr. In a study
by Wald et al. (32), serial measurements of sCysC in adult
patients undergoing cardiopulmonary bypass correlated with
the development of AKI, but the discriminatory capacity of
sCysC as an early biomarker of AKI was very limited. In the
postoperative period after cardiac surgery in elderly patients,
sCysC and SCr detected AKI similarly (25), and in a multi-
center prospective observational cohort study involving a het-
erogeneous adult population admitted to an intensive care unit,
both urinary CysC and sCysC were poor early predictors of
AKI and the need for renal replacement therapy (26).
F939
F940 SERUM CysC AS A BIOMARKER OF SEPSIS-INDUCED ACUTE KIDNEY INJURY

For sCysC to be considered a better biomarker of AKI than
SCr in some settings but not in others, rational criteria are
needed for each setting (context of use). Early detection bio-
markers are especially difficult to establish in patients, as the
timing of the initial renal injury is often difficult to discern.
Therefore, we compared the ability of SCr and sCysC to detect
kidney injury caused by sepsis (sepsis-AKI) under more con-
trolled circumstances using an experimental model of sepsis in
mice.
METHODS
Animals and animal models. We followed National Institutes of
Health (NIH) criteria for the use and treatment of laboratory animals.
All experiments were conducted on 6- to 8-wk-old male CD-1 mice
Survival experiments. Six hours after the induction of CLP-induced
sepsis, 50 l of blood were collected by the retroorbital sinus
approach under avertin anesthesia. Blood collection from younger
(6 8 wk old) mice after sepsis resulted in rapid mortality (data not
shown); hence, the use of older (1216 wk old) mice for survival
experiments. Animals were monitored for survival every 4 8 h after
surgery, and the time to death was recorded for each animal. SCr,
sCysC, and BUN were measured as described above. Mice were given
fluids and buprenorphine immediately after CLP; fluids, antibiotic,
and buprenorphine were given starting at 6 h after sepsis and then
given every 12 h until the time to death. Morbidly ill mice were
euthanized per protocol.
Production and kinetics of sCysC. Production and kinetics of
sCysC were performed in BiNx and BiNx CLP groups only. We
took advantage of the otherwise stable sCysC level 12 h after BiNx or
BiNx CLP to measure the pharmacokinetics of injected recombi-
nant CysC. At 12 h after BiNx or BiNx CLP, 25 l of capillary

Downloaded from http://ajprenal.physiology.org/ by 10.220.33.3 on July 11, 2017

(Charles River Laboratories) except that 12- to 16-wk-old male CD-1
mice were used for blood collections at 6 h after the induction of blood were collected via the retroorbital sinus, and recombinant CysC
sepsis and subsequent survival analysis. All animals had free access to
water and chow and were monitored postoperatively for symptoms
including piloerection, spontaneous activity, response to stimuli, la- A
bored breathing, and eye grooming. All procedures were performed
under isoflurane anesthesia (including the euthanasia of morbidly ill
mice, per protocol).
Cecal ligation and puncture (CLP) was performed to induce sepsis
as previously described (18, 20, 33). In brief, the cecum was ligated
12 mm from its tip and then punctured twice with a 21-gauge needle.
Bilateral nephrectomy (BiNx) was performed as previously described
(5); briefly, both kidneys were decapsulated to avoid adrenal damage
and then removed via flank incisions. BiNx was performed at the same
time as CLP in the BiNx CLP group. For sham surgery, the cecum
and/or kidneys were identified via simultaneous incisions. Normal
saline (15 ml/kg) was given intraperitoneally immediately after all
surgeries and then either antibiotic (imipenem-cilastatin, 14 mg/kg in
1 ml normal saline) for CLP or normal saline for BiNx was given
subcutaneously at 6 h. All mice subjected to any surgical procedure
also received buprenorphine solution (0.05 mg/kg ip) immediately
after and at 6 h after surgery. When we measured GFR 6 h after CLP,
we delayed fluid administration so that it would not dilute inulin
before blood sampling.
GFR measurements in conscious mice. GFR was measured at 0, 3,
6, 12, or 18 h after CLP by FITC-labeled inulin clearance (7). Each B
mouse was studied only once at a single time point (n 5 6
mice/time point). One microcapillary tube of blood (50 l) was
collected via a tail vein for the measurement of SCr, blood urea
nitrogen (BUN), and sCysC at the specified time point post-CLP. A
single dose of FITC-inulin (3.74 l/g body wt) was injected into the
retroorbital plexus, and 5-l blood samples were collected from the
tail vein 10, 15, 35, 55, and 75 min afterward. Plasma fluorescence
was measured by a Nanodrop-ND-3300 fluorescence spectrometer
(Nanodrop Technologies, Wilmington, DE). GFR was then calculated
using a two-compartment model, as previously described (7). The
correlation between SCr, sCysC, and BUN at the specified time
post-CLP and inulin GFR (iGFR) in the individual mice was calcu-
lated (SigmaStat 3.1, Systat Software, Point Richmond, CA).
Blood chemistries. Fifty microliters of blood were collected by the
retroorbital approach before surgery (0 h) and at 3, 6, 12, and 18 h
after surgery (n 5 6 mice/time point), and body weight was
recorded. SCr was measured by HPLC (34), BUN by colorimetric
assay (QuantiChrom Urea assay kit DIUR-500, Hayward, CA), and Fig. 1. Fold changes of biomarkers over baseline values (0 h) at different time
mouse sCysC by ELISA (BioVendor, Candler, NC). Under anesthesia points after cecal ligation and puncture (CLP). A: at 3 h after CLP, serum
at 18 h after surgery, blood was collected after cardiac puncture, and cystatin C (sCysC) was 3-fold higher than baseline, whereas serum creatinine
(SCr) was unchanged (1.16-fold) and blood urea nitrogen (BUN) was slightly
mice were euthanized. Serum aspartate transaminase, alanine changed (1.74-fold). SCr was not different from BUN at all time points. There
transaminase, and lactate dehydrogenase were measured by an auto- were no differences among biomarkers at 6 h. sCysC represents the earliest
analyzer (Hitachi 917, Boehringer Mannheim, Indianapolis, IN), and and highest relative change among all biomarkers. *P 0.05. **P 0.01.
serum TNF-, IL-6, and IL-10 were measured by ELISA (R&D ***P 0.001. B: changes in inunlin glomerular filtration rate (iGFR) at the
Systems, Minneapolis, MN). various time points after CLP. #P 0.001 relative to baseline.

AJP-Renal Physiol doi:10.1152/ajprenal.00025.2013 www.ajprenal.org

 SERUM CysC AS A BIOMARKER OF SEPSIS-INDUCED ACUTE KIDNEY INJURY F941

A 500 E 500

400 400
iGFR (l/min)

iGFR (l/min)
300 300

200 200

100 100
R 2 = 0 .924
p < 0.001
0 0
0 2 4 6 8 10 0 100 200 300 400 500 600
sCysC (g/ml) 220.7564
eGFR ( 3.3411 + ) (l/min)
sCysC
B 500 F 500

Downloaded from http://ajprenal.physiology.org/ by 10.220.33.3 on July 11, 2017

400 400
iGFR (l/min)

iGFR (l/min)
300 300

200 200

100 100
R 2 = 0 .743
p < 0.001
0 0
0.0 0.2 0.4 0.6 0.8 1.0 1.2 0 50 100 150 200 250 300 350 400 450
Scr (mg/dl) 36.7066
eGFR ( 28.3925 + ) (l/min)
Scr
C 600 G 600

500 500
iGFR (l/min)

iGFR (l/min)

400 400

300 300

200 200

100 100 R 2 = 0 .813

p < 0.001
0 0
20 40 60 80 100 0 100 200 300 400 500
BUN (mg/dl) 8080.9724
eGFR ( 62.7609 + ) (l/min)
BU N
D 500 H 500

400 400
iGFR (l/min)

iGFR (l/min)

300 300

200 200

100 100
R 2 = 0 .944
p < 0.001
0 0
0.000 0.005 0.010 0.015 0.020 0.025 0.030 0.035 0.040 0 100 200 300 400 500
10.5599 174.2714 1 10.5599 174.2714
Composite ( + ) eGFR ( 22.7981 + + ) (l/min)
Scr sCysC Scr sCysC

0h 3h 6h 12h 18h

Fig. 2. Time course of kidney dysfunction during sepsis-induced acute kidney injury (AKI) and cross correlation with renal function biomarkers. After
sepsis-induced AKI, sCysC (A; R2 0.924), SCr (B; R2 0.743), and BUN (C; R2 0.813) all increased, and their reciprocals were correlated with iGFR,
which decreased. A linear regression analysis of the relationship between each biomarkers reciprocal and iGFR was performed. iGFR and estimated GFR (eGFR)
values based on sCysC (E), SCr (F), and BUN (G) values were plotted against each other. A multiple regression model was created combining SCr and sCysC
(R2 0.944). The relationship between a weighted composite value of SCr and sCysC generated based on the coefficients of the multiple regression model is
shown (D), together with iGFR plotted against composite eGFR (H).

AJP-Renal Physiol doi:10.1152/ajprenal.00025.2013 www.ajprenal.org

F942 SERUM CysC AS A BIOMARKER OF SEPSIS-INDUCED ACUTE KIDNEY INJURY

distribution (Vd) was calculated as follows: Vd injected dose/

A [sCysCpeak postinjection(12 h 5 min) sCysCpreinjection(12 h)], and the
sCysC net production was calculated as follows: net production
(sCysC12 h sCysC0 h) Vd, as nonrenal clearance cannot be distin-
guished from production. An alternative noncompartmental method of
calculating Vd (Phoenix WinNonlin software, version 6.02, Pharsight,
Mountain View, CA) yielded similar results for Vd estimates.
Statistical analysis. All data are expressed as means SE.
ANOVA with Bonferronis multiple comparison correction was per-
formed using Prism 4.0 (Graphpad Software). We compared survival
curves using a log-rank test (Prism 4.0, Graphpad Software). The area
under the receiver-operating characteristic curve was calculated for

Downloaded from http://ajprenal.physiology.org/ by 10.220.33.3 on July 11, 2017

B

Fig. 3. Time course of sCysC, SCr, and BUN after CLP-induced sepsis alone
or after bilateral nephrectomy (BiNx). AC: time courses of sCysC (A), SCr
(B), and BUN (C) in CLP and BiNx (n 5 6 mice/group). Two-way ANOVA
was performed with Bonferroni post hoc analysis. *P 0.05, **P 0.01, and
#P 0.001, BiNx vs. CLP at each time point.

(5 g, R&D Systems) was injected by the tail vein. Additional

samples were taken at 5, 15, 30, 60, 120, and 180 min to determine
peak levels, and these limited time points enabled us to qualitatively
confirm a single compartment with no substantial distributive phase.
sCysC levels after CysC administration at 12 h were adjusted for
an extrapolated production rate from 0 to 12 h. The volume of
Fig. 4. AC: time courses of sCysC (A), SCr (B), and BUN (C) in BiNx with
and without CLP (n 5 6 mice/group). Two-way ANOVA was performed
with Bonferroni post hoc analysis. *P 0.05 and #P 0.001, CLP vs.
BiNx CLP at each time point. ns, Not significant.

AJP-Renal Physiol doi:10.1152/ajprenal.00025.2013 www.ajprenal.org

 SERUM CysC AS A BIOMARKER OF SEPSIS-INDUCED ACUTE KIDNEY INJURY
each biomarker for each time point during the survival study (Prism
4.0, Graphpad Software). We used the Statsmodels software package
(python programming language, http://statsmodels.sourceforge.net/)
to perform a linear regression analysis on the reciprocal value of the
biomarkers. A weighted regression analysis was used to validate
results, with weights based on inverse variances, when visual inspec-
tion of residual plots indicated a possible deviation from homosce-
dasticity. Models were compared using the Akaike information crite-
rion. P values of 0.05 were accepted as statistically significant.
RESULTS
sCysC outperforms SCr and BUN as a renal function bio-
marker early in the course of sepsis-AKI. We simultaneously
measured SCr, BUN, sCysC, and FITC-iGFR in conscious
mice at 0, 3, 6, 12, and 18 h after CLP using 5 6 mice/time
F943
in SCr and BUN (Fig. 1A). At 12 and 18 h, sCysC also had
significantly higher increases over baseline compared with SCr
and BUN, whereas at 6 h, the increased biomarker levels were
not statistically different (Fig. 1A). At 3 h after CLP, GFR
dramatically decreased 50% (Fig. 1B); this was associated
with a 3.2-fold increase in sCysC and a 1.7-fold increase in
BUN but no change in SCr (1.16-fold; Fig. 1A). Accordingly,
both sCysC and BUN outperformed SCr as early biomarkers of
sepsis-AKI, with sCysC being the best early biomarker. At
18 h after sepsis, GFR was reduced by 90% (from 428 to 5
l/min; Fig. 1B), which was associated with 12.7-, 4.8-, and
8.2-fold increases in sCysC, BUN, and SCr, respectively (Fig.
1A). In addition to detecting changes in GFR early in the
progression of disease, accurate assessment of GFR is also

Downloaded from http://ajprenal.physiology.org/ by 10.220.33.3 on July 11, 2017

point. At 3 h after CLP, sCysC was increased threefold over important. The level of each biomarker was correlated to iGFR
baseline (0 h) sCysC. At this time point, both SCr and BUN by including all time points. Because there is a reciprocal
less than doubled their values over baseline, and the sCysC relationship between GFR and net accumulation of each cir-
increase over baseline was significantly higher than increases culating biomarker, we used 1/biomarker values to avoid an

Fig. 5. Effect of CLP-induced sepsis after BiNx on inflammation and nonrenal injury biomarkers. sCysC (A), SCr (B), BUN (C), aspartate transaminase (AST;
D), alanine transaminase (ALT; E), lactate dehydrogenase (LDH; F), TNF- (G), IL-6 (H), and IL-10 (I) were measured at 18 h after surgery (n 5 6
mice/group). One-way ANOVA was performed with Bonferroni post hoc analysis. *P 0.05, **P 0.01, and #P 0.001 between the indicated groups.

AJP-Renal Physiol doi:10.1152/ajprenal.00025.2013 www.ajprenal.org

F944 SERUM CysC AS A BIOMARKER OF SEPSIS-INDUCED ACUTE KIDNEY INJURY

A B C
Fig. 6. Effect of sepsis on CysC volume of
distribution (Vd), production, and elimina-
tion. A: Vd of CysC was measured at 5 min
(as baseline) and at 12 h after surgery. B:
estimated CysC production was measured at
0 12 h after CLP surgery (n 5 6 mice/
group). C: clearance was measured at 12 h
after CLP surgery (n 5 6 mice/group).
*P 0.05 vs. baseline.

Downloaded from http://ajprenal.physiology.org/ by 10.220.33.3 on July 11, 2017

asymptotic curve, which would visually and computationally
compress the differences at the low GFR/high biomarker end
of the curve. With this transformation, the relationship was
analyzed by linear regression. Although BUN changed quickly
(at 3 h after sepsis), it underperformed sCysC in reflecting
iGFR (R2 0.813 vs. R2 0.924, P 0.001; Fig. 2). Both
sCysC- and BUN-based eGFR equations were superior to SCr
(R2 0.743) in correlating to iGFR (P 0.001 and P 0.01,
respectively; Fig. 2). Next, we considered whether a combina-
tion of biomarkers would more accurately reflect iGFR than
any individual biomarker. The combination of sCysC with
SCr, but not BUN, improved on the correlation of sCysC
alone to iGFR (P 0.02 and P 0.6, respectively). sCysC
alone was superior to the combination of SCr and BUN (P
0.001). The combination of all three biomarkers was not
better than the combination of sCysC and SCr (P 0.83).
Using the coefficients from the linear regression analysis, a
weighted composite mouse eGFR equation was created from
SCr and sCysC, and the relationship with GFR was visual-
ized (Fig. 2, D and H).
Time courses of sCysC, SCr, and BUN accumulation after
CLP differ from those after BiNx. To determine the kinetic
response of these biomarkers to a sudden change in GFR, we
measured the time course of sCysC, SCr, and BUN after BiNx.
We found that sCysC stabilized 12 h after BiNx and CLP,
whereas SCr and BUN were still increasing at 18 h (Fig. 3,
AC). The increases in all biomarkers were diminished after
CLP compared with BiNx. This difference was statistically
significant starting at 3 h for SCr (Fig. 3B), 6 h for sCysC (Fig.
3A), and 12 h for BUN (Fig. 3C). Therefore, CLP caused an
accumulation of each circulating biomarker over time that was
distinct from that caused by complete (BiNx) loss of renal
function.
Sepsis blunts BiNx-induced increases in sCysC and SCr. We
(5) have previously demonstrated that sepsis decreases creati-
nine production. To determine if sepsis also affects sCysC
kinetics, CLP was performed simultaneously with BiNx. Loss
of kidney function by BiNx increased sCysC, SCr, and BUN,
as expected (Fig. 4, AC); BiNx combined with sepsis blunted
these increases in sCysC at 18 h and SCr at 12 and 18 h but did
not blunt BUN increases (Fig. 4, AC). In contrast, BiNx
combined with sepsis significantly increased other inflamma-
tory and organ damage parameters (inflammatory cytokines,
aspartate transaminase, alanine transaminase, and lactate de-
hydrogenase) compared with BiNx alone (Fig. 5). Hence,
consistent with our previous study with SCr (5), we now
demonstrate that within the context of BiNx, sCysC also
decreases after sepsis, in contrast to other circulating biomark-
ers, which do not decrease.
Sepsis decreases sCysC production and may increase non-
renal sCysC elimination after BiNx without changing Vd. We
next determined whether changes in sCysC kinetics (i.e., in-
creased Vd, production, metabolism, or elimination) could
account for the reduction of sCysC after sepsis. Exogenous
recombinant CysC was injected, and blood was sampled fre-
quently (see METHODS). Vd of sCysC was not different between
BiNx sham and BiNx CLP groups (Fig. 6A), similar to
our previous data on the FITC-inulin space (5). Additionally,
there were no differences in body weights between BiNx
sham and BiNx CLP groups (data not shown), as previously
reported (5). Next, we found that sepsis reduced estimated net
sCysC production (Fig. 6B); from this, we can infer that it may
have increased nonrenal elimination (see METHODS) (Fig. 6C).
Serum CysC does not outperform SCr on predicting mortal-
ity in the setting of sepsis. SCr, sCysC, and BUN were
measured at 6 h after CLP surgery, and the time to death was
measured. Post hoc assignment of mice to low and high
biomarker groups, using the median as a threshold, demon-
strated statistically significant differences in the Kaplan-Meier
survival curves for SCr, sCysC, and BUN (Fig. 7A). All mice
died within 52 h. SCr, sCysC, and BUN were each able to
predict the time to death (Fig. 7B), and higher biomarker
median values were associated with decreased survival (Fig.
7A). Both SCr (P 0.001) and sCysC (P 0.001) were
superior to BUN, and SCr was stronger than sCysC (P
0.015) in predicting the time to death. Combining multiple
biomarkers in the analysis did not improve on SCr alone. In
contrast, there was no difference in the ability of SCr and
sCysC to classify animals to earlier and later mortality groups.
Receiver-operating characteristic curves constructed compar-

Fig. 7. AKI biomarkers and mortality. Mice were grouped into above and below median (high vs. low, respectively) groups. Median/cutoff values were 1.186
mg/ml for sCysC, 0.241 mg/dl for SCr, and 54.57 mg/dl for BUN. A: mice with higher levels of sCysC, SCr, or BUN had a higher mortality rate. B: sCysC and
SCr correlated equivalently with the time to death, whereas BUN did not correlate with the time to death. C: there were no differences between areas under the
curves for both sCysC and SCr at different time points.

AJP-Renal Physiol doi:10.1152/ajprenal.00025.2013 www.ajprenal.org

 SERUM CysC AS A BIOMARKER OF SEPSIS-INDUCED ACUTE KIDNEY INJURY F945

A B 2.5

2.0

sCysC (g/ml)
1.5

1.0

0.5
15 20 25 30 35 40 45 50 55

Downloaded from http://ajprenal.physiology.org/ by 10.220.33.3 on July 11, 2017

0.9

0.8

0.7

Scr (mg/dl)
0.6

0.5

0.4

0.3

0.2

0.1
15 20 25 30 35 40 45 50 55

120

110

100
BUN (mg/dl)

90

80

70

60

50

40

30
15 20 25 30 35 40 45 50 55

AJP-Renal Physiol doi:10.1152/ajprenal.00025.2013 www.ajprenal.org

F946 SERUM CysC AS A BIOMARKER OF SEPSIS-INDUCED ACUTE KIDNEY INJURY
ing mortality at earlier and later time points demonstrated no
difference between SCr and sCysC (Fig. 7C).
DISCUSSION
An ideal serum kidney filtration biomarker should be con-
stantly produced, freely filtered, neither secreted nor reab-
sorbed by the renal tubule, and lack nonrenal elimination
pathways (27). The biomarker might either have fast kinetics to
rapidly track minute changes in GFR or slower kinetics that
integrate changes in GFR over a long time interval (similar to
glucose vs. HbA1C). Unfortunately, the production rate of
creatinine is influenced by many extrarenal factors (muscle
mass, age, sex, and reduced production in sepsis-AKI) and
tubular secretion (3). Similarly, BUN is influenced by noncon-
stant production and significant tubular reabsorption, espe-
cially during volume depletion (6). In contrast, sCysC produc-
tion is more predictable, although it is modestly influenced by
smoking, obesity, hyperthyroidism, and low-grade chronic in-
flammation (10, 11, 16), but it is largely unaffected by acute
inflammation, including sepsis-induced inflammation (19, 23,
28). Thus, sCysC could meet most of the ideal filtration
biomarker criteria with less interindividual variation (9).
At least two classes of AKI biomarkers are needed, each
with a different context of use or role in clinical decision
making: an early detection biomarker that is tightly coupled to
GFR, enabling rapid detection after the insult with an oppor-
tunity to identify patients that may be at a higher risk, and a
prognosis biomarker that can predict severity and/or mortality.
We demonstrated in an animal model of sepsis that sCysC
changes rapidly after injury and that a sCysC-based eGFR
equation better reflects iGFR throughout the course of sepsis-
AKI. In comparison, BUN increased rapidly, but a BUN-based
eGFR equation only moderately correlated with iGFR; SCr
increased slowly, and a SCr-based eGFR equation only poorly
reflected GFR. The combination of sCysC with SCr, but not
BUN, further improved eGFR equations when correlated to
iGFR. However, sCysC is still not an ideal biomarker because
sepsis decreases the production and potentially increases the
nonrenal clearance of sCysC. Furthermore, sCysC did not
outperform the widely used SCr to predict mortality. The
combination of SCr with sCysC also did not improve the
performance of either individual biomarker with respect to
sepsis mortality.
sCysC outperforms creatinine early after AKI. GFR cannot
be calculated in patients from SCr during AKI until a steady
state is reached days after injury; this period is prolonged
because of the slow creatinine kinetics and decreased produc-
tion (8). Therefore, we measured iGFR by the best available
method of plasma disappearance of FITC-inulin, which uses a
two-compartment model of two-phase compartment decay, to
approximate how quickly GFR falls during sepsis. This iGFR
measurement has limitations during sepsis because 1) the
calculation is based on an assumption that GFR does not
change appreciably during the 75-min sampling window and 2)
the number of samples that can be collected is limited by
potential artifacts introduced by multiple blood draws. A more
sophisticated model could be developed, but there are limits to
how many time points can be collected before the collection
itself can cause its own artifacts. Nevertheless, sepsis rapidly
reduced iGFR by 50% as early as 3 h after CLP (Fig. 1B),
AJP-Renal Physiol doi:10.1152/ajprenal.00025.2013 www.ajprenal.org
even before systemic sepsis symptoms, such as lethargy, di-
minished response to stimulus, or piloerection, developed at
6 h. Despite these large and rapid changes in iGFR, SCr was
slow to react; sCysC and BUN had much faster kinetics: both
increased rapidly and achieved a steady state approximately
within 12 h (Fig. 1A).
Influence of sepsis on sCysC production and metabolism.
We have recently found that sepsis blunted the increase in SCr
after sepsis in BiNx mice and that creatinine production was
reduced by sepsis (5). In the present study, we found that sepsis
similarly blunted the increase in sCysC after sepsis in BiNx
mice.
The effect of sepsis on sCysC was complicated, as sepsis
both reduced sCysC production and may have also enhanced
nonrenal clearance. The latter could be due to increased clear-
Downloaded from http://ajprenal.physiology.org/ by 10.220.33.3 on July 11, 2017
ance by the reticuloendothelial system (31). Both of these
could conspire to reduce the ability of sCysC to function as an
early detection biomarker, even though sCysC did increase
earlier (at 3 h) than SCr and BUN in the same mice. Although
BUN did not correspond to iGFR as well as sCysC, it also
increased rapidly after BiNx or CLP, and CLP did not blunt the
increase observed after BiNx. As such, although BUN is not
the best biomarker for GFR, it may still be useful for detecting
early changes in GFR. The usefulness of this early rise may be
somewhat diminished by 6 h when systemic symptoms of
sepsis begin to manifest, as both sCr and sCysC outperformed
BUN as predictors of mortality.
AKI biomarkers and mortality. In this study, although sCysC
was increased more than SCr early after CLP, it did not
outperform SCr in predicting mortality. The areas under the
receiver-operating characteristic curves of sCysC and SCr with
the outcome of mortality were comparable. Our result is
consistent with previous work by Perianayagam et al. (22),
which demonstrated that a single measurement of sCysC at the
time of nephrology consultation did not surpass SCr on pre-
dicting in-hospital mortality among critically ill patients diag-
nosed with AKI. To date, very few clinical studies have
analyzed sCysC as a predictor of mortality or early AKI
biomarker among septic patients. Recently, in a large multi-
center study (30), it was found that sCysC was less sensitive
for AKI detection than SCr among postcardiac surgery pa-
tients.
For the mouse survival experiments, we chose to measure
sCysC, SCr, and BUN at 6 h after the induction of sepsis
because this is when the animals become clinically sick and,
hence, in this model, mimics the time needed to diagnose
sepsis in a clinical setting. Both sCysC and SCr did not show
a very strong correlation with the time to death (with R2 0.7
for both), as AKI is not expected to be the primary cause of
sepsis mortality, and dysfunction of other organs/systems also
contributes to mortality during sepsis.
Conclusions. We evaluated the ability of sCysC to monitor
the progression of sepsis-AKI in the clinically relevant mouse
model of polymicrobial sepsis in CD-1 mice. Among the three
studied filtration biomarkers (sCysC, SCr, and BUN), sCysC
showed a more robust increase early after sepsis. The blunted
increase in sCysC after sepsis in BiNx, despite the increase in
other nonrenal injury biomarkers, was caused by a sepsis-
induced reduction of sCysC production and possibly the in-
crease in sCysC nonrenal elimination. Thus, sCysC is a better
early detection biomarker than SCr in sepsis but is still influ-
 SERUM CysC AS A BIOMARKER OF SEPSIS-INDUCED ACUTE KIDNEY INJURY
enced by nonrenal factors that conspire to limit the accurate
prediction of GFR during the evolution of sepsis-AKI. BUN
levels after BiNx were not affected by sepsis but corresponded
poorly to iGFR.
sCysC did not outperform SCr in identifying mice at risk of
early or late mortality in this model of sepsis. sCysC can be
used interchangeably with SCr for mortality risk stratification,
and it increases more, early after the insult, and correlates
better with iGFR than SCr.
sCysC may not be an ideal biomarker for sepsis-AKI, but
because an early biomarker may enhance risk stratification and
facilitate early diagnosis, sCysC may be, indeed, a better
biomarker of sepsis-AKI than SCr. Although BUN also in-
creased early after CLP, it did not correlate well with iGFR and
was inferior to sCysC and SCr for classification of mortality
risk.
ACKNOWLEDGMENTS
The authors thank our late colleague Richard Chen for his invaluable
contribution to statistical analysis and interpretation. The authors thank Chris-
toph Eisner for iGFR measurements.
GRANTS
This research was supported by the Intramural Research Program of the
National Institutes of Health (National Institute of Diabetes and Digestive and
Kidney Diseases).
DISCLOSURES
No conflicts of interest, financial or otherwise, are declared by the author(s).
AUTHOR CONTRIBUTIONS
Author contributions: A.L., A.C.P.S., K.D., J.B.S., R.A.S., and P.S.Y.
conception and design of research; A.L., A.C.P.S., V.H., L.L., X.H., and
P.S.Y. performed experiments; A.L., A.C.P.S., J.M.S., V.H., L.L., X.H., H.Z.,
P.K., J.B.S., and P.S.Y. analyzed data; A.L., A.C.P.S., J.M.S., V.H., T.T.,
K.D., H.Z., P.K., J.B.S., R.A.S., and P.S.Y. interpreted results of experiments;
A.L., A.C.P.S., J.M.S., and P.S.Y. prepared figures; A.L. and A.C.P.S drafted
manuscript; A.L., A.C.P.S., J.M.S., V.H., T.T., K.D., L.L., X.H., H.Z., P.K.,
J.B.S., R.A.S., and P.S.Y. approved final version of manuscript; A.C.P.S.,
J.M.S., V.H., T.T., K.D., L.L., X.H., P.K., J.B.S., R.A.S., and P.S.Y. edited and
revised manuscript.
REFERENCES
1. Baas MC, Bouman CS, Hoek FJ, Krediet RT, Schultz MJ. Cystatin C
in critically ill patients treated with continuous venovenous hemofiltration.
Hemodial Int 10, Suppl 2: S33S37, 2006.
2. Bagshaw SM, Bellomo R. Cystatin C in acute kidney injury. Curr Opin
Crit Care 6: 533539, 2010.
3. Bjornsson TD. Use of serum creatinine concentrations to determine renal
function. Clin Pharmacokinet 4: 200 222, 1979.
4. Christensson A, Ekberg J, Grubb A, Ekberg H, Lindstrom V, Lilja H.
Serum cystatin C is a more sensitive and more accurate marker of
glomerular filtration rate than enzymatic measurements of creatinine in
renal transplantation. Nephron Physiol 94: p19 p27, 2003.
5. Doi K, Yuen PS, Eisner C, Hu X, Leelahavanichkul A, Schnermann J,
Star RA. Reduced production of creatinine limits its use as marker of
kidney injury in sepsis. J Am Soc Nephrol 20: 12171221, 2009.
6. Dossetor JB. Creatininemia versus uremia. The relative significance of
blood urea nitrogen and serum creatinine concentrations in azotemia. Ann
Intern Med 65: 12871299, 1966.
7. Eisner C, Faulhaber-Walter R, Wang Y, Leelahavanichkul A, Yuen
PS, Mizel D, Star RA, Briggs JP, Levine M, Schnermann J. Major
contribution of tubular secretion to creatinine clearance in mice. Kidney
Int 77: 519 526, 2010.
8. Endre ZH, Westhuyzen J. Early detection of acute kidney injury:
emerging new biomarkers. Nephrology (Carlton) 13: 9198, 2008.
AJP-Renal Physiol doi:10.1152/ajprenal.00025.2013 www.ajprenal.org
F947
9. Filler G, Bokenkamp A, Hofmann W, Le Bricon T, Martinez-Bru C,
Grubb A. Cystatin C as a marker of GFR history, indications, and future
research. Clin Biochem 38: 18, 2005.
10. Fricker M, Wiesli P, Brandle M, Schwegler B, Schmid C. Impact of
thyroid dysfunction on serum cystatin C. Kidney Int 63: 1944 1947, 2003.
11. Galteau MM, Guyon M, Gueguen R, Siest G. Determination of serum
cystatin C: biological variation and reference values. Clin Chem Lab Med
39: 850 857, 2001.
12. Grubb AO. Cystatin Cproperties and use as diagnostic marker. Adv Clin
Chem 35: 6399, 2000.
13. Herget-Rosenthal S, Marggraf G, Husing J, Goring F, Pietruck F,
Janssen O, Philipp T, Kribben A. Early detection of acute renal failure
by serum cystatin C. Kidney Int 66: 11151122, 2004.
14. Inker LA, Okparavero A. Cystatin C as a marker of glomerular filtration
rate: prospects and limitations. Curr Opin Nephrol Hypertens 20: 631
639, 2011.
15. Inker LA, Schmid CH, Tighiouart H, Eckfeldt JH, Feldman HI,
Greene T, Kusek JW, Manzi J, Van Lente F, Zhang YL, Coresh J,
Downloaded from http://ajprenal.physiology.org/ by 10.220.33.3 on July 11, 2017
Levey AS. Estimating glomerular filtration rate from serum creatinine and
cystatin C. N Engl J Med 367: 20 29, 2012.
16. Knight EL, Verhave JC, Spiegelman D, Hillege HL, de Zeeuw D,
Curhan GC, de Jong PE. Factors influencing serum cystatin C levels
other than renal function and the impact on renal function measurement.
Kidney Int 65: 1416 1421, 2004.
17. Krawczeski CD, Vandevoorde RG, Kathman T, Bennett MR, Woo
JG, Wang Y, Griffiths RE, Devarajan P. Serum cystatin C is an early
predictive biomarker of acute kidney injury after pediatric cardiopulmo-
nary bypass. Clin J Am Soc Nephrol 5: 15521557, 2010.
18. Leelahavanichkul A, Yasuda H, Doi K, Hu X, Zhou H, Yuen PS, Star
RA. Methyl-2-acetamidoacrylate, an ethyl pyruvate analog, decreases
sepsis-induced acute kidney injury in mice. Am J Physiol Renal Physiol
295: F1825F1835, 2008.
19. Martensson J, Martling CR, Oldner A, Bell M. Impact of sepsis on
levels of plasma cystatin C in AKI and non-AKI patients. Nephrol Dial
Transplant 27: 576 581, 2012.
20. Nemeth K, Leelahavanichkul A, Yuen PS, Mayer B, Parmelee A, Doi
K, Robey PG, Leelahavanichkul K, Koller BH, Brown JM, Hu X,
Jelinek I, Star RA, Mezey E. Bone marrow stromal cells attenuate sepsis
via prostaglandin E(2)-dependent reprogramming of host macrophages to
increase their interleukin-10 production. Nat Med 15: 4249, 2009.
21. Newman DJ, Thakkar H, Edwards RG, Wilkie M, White T, Grubb
AO, Price CP. Serum cystatin C: a replacement for creatinine as a
biochemical marker of GFR. Kidney Int Suppl 47: S20 S21, 1994.
22. Perianayagam MC, Seabra VF, Tighiouart H, Liangos O, Jaber BL.
Serum cystatin C for prediction of dialysis requirement or death in acute
kidney injury: a comparative study. Am J Kidney Dis 54: 10251033,
2009.
23. Randers E, Kornerup K, Erlandsen EJ, Hasling C, Danielsen H.
Cystatin C levels in sera of patients with acute infectious diseases with
high C-reactive protein levels. Scand J Clin Lab Invest 61: 333335, 2001.
24. Reed CH. Diagnostic applications of cystatin C. Br J Biomed Sci 57:
323329, 2000.
25. Ristikankare A, Poyhia R, Kuitunen A, Skrifvars M, Hammainen P,
Salmenpera M, Suojaranta-Ylinen R. Serum cystatin C in elderly
cardiac surgery patients. Ann Thorac Surg 89: 689 694, 2010.
26. Royakkers AA, Korevaar JC, van Suijlen JD, Hofstra LS, Kuiper
MA, Spronk PE, Schultz MJ, Bouman CS. Serum and urine cystatin C
are poor biomarkers for acute kidney injury and renal replacement therapy.
Intensive Care Med 37: 493501, 2011.
27. Shemesh O, Golbetz H, Kriss JP, Myers BD. Limitations of creatinine
as a filtration marker in glomerulopathic patients. Kidney Int 28: 830 838,
1985.
28. Simonsen O, Grubb A, Thysell H. The blood serum concentration of
cystatin C (gamma-trace) as a measure of the glomerular filtration rate.
Scand J Clin Lab Invest 45: 97101, 1985.
29. Soto K, Coelho S, Rodrigues B, Martins H, Frade F, Lopes S, Cunha
L, Papoila AL, Devarajan P. Cystatin C as a marker of acute kidney
injury in the emergency department. Clin J Am Soc Nephrol 5: 17451754,
2010.
30. Spahillari A, Parikh CR, Sint K, Koyner JL, Patel UD, Edelstein CL,
Passik CS, Thiessen-Philbrook H, Swaminathan M, Shlipak MG.
Serum cystatin C- versus creatinine-based definitions of acute kidney
injury following cardiac surgery: a prospective cohort study. Am J Kidney
Dis 60: 922929, 2012.
F948 SERUM CysC AS A BIOMARKER OF SEPSIS-INDUCED ACUTE KIDNEY INJURY

31. Tenstad O, Roald AB, Grubb A, Aukland K. Renal handling of
radiolabelled human cystatin C in the rat. Scand J Clin Lab Invest 56:
409 414, 1996.
32. Wald R, Liangos O, Perianayagam MC, Kolyada A, Herget-
Rosenthal S, Mazer CD, Jaber BL. Plasma cystatin C and acute
kidney injury after cardiopulmonary bypass. Clin J Am Soc Nephrol 5:
13731379, 2010.
33. Yasuda H, Leelahavanichkul A, Tsunoda S, Dear JW, Takahashi Y, Ito S,
Hu X, Zhou H, Doi K, Childs R, Klinman DM, Yuen PS, Star RA.
Chloroquine and inhibition of Toll-like receptor 9 protect from sepsis-induced
acute kidney injury. Am J Physiol Renal Physiol 294: F1050 F1058, 2008.
34. Yuen PS, Dunn SR, Miyaji T, Yasuda H, Sharma K, Star RA. A
simplified method for HPLC determination of creatinine in mouse serum.
Am J Physiol Renal Physiol 286: F1116 F1119, 2004.

Downloaded from http://ajprenal.physiology.org/ by 10.220.33.3 on July 11, 2017

AJP-Renal Physiol doi:10.1152/ajprenal.00025.2013 www.ajprenal.org