Cystatin C-estimated Glomerular Filtration Rate in Pediatric Autologous Hematopoietic Stem Cell Transplantation
Formal evaluation of kidney function is essential to determine chemotherapy dosing based on established treatment protocols in children undergoing autologous stem cell transplantation. Cystatin C has been widely studied as a marker of the glomerular filtration rate (GFR), although data regarding its use in stem cell transplantation are limited. We evaluated the performance of cystatin C-based equations and determined their sensitivity to detect a nuclear GFR of <100 mL/min/1.73 m(2) in children undergoing autologous transplantation. We performed a retrospective cohort analysis in 16 children undergoing 26 transplantations using a modified Bland-Altman analysis to account for repeated measures. Cystatin C-based equations published by Hoek, Le Bricon, Rule, Filler, Zappitelli, Larsson, and Schwartz (the New Chronic Kidney Disease in Children formula, New CKiD formula) were evaluated and compared to the creatinine-based modified Schwartz equation. We found that cystatin C-based equations demonstrated improved sensitivity to detect a nuclear GFR of <100 mL/min/1.73 m(2) compared to the creatinine-based modified Schwartz equation, which significantly overestimated GFR. Most cystatin C-based equations, however, tended to underestimate the nuclear GFR. The New CKiD formula, combining cystatin C and creatinine, offered a sensitivity of 100% and a specificity of 95% for detecting a nuclear GFR <100 mL/min/1.73 m(2). Institutions using cystatin C-based GFR estimation should be aware of the specific prediction formula and GFR measurement techniques available at their center, as each method's performance can vary considerably. As more research becomes available, this easily measured marker should become a valuable component of GFR estimation, providing cost savings (a nuclear GFR is 5.5 times more costly than a cystatin C) and reducing radiation exposure.
Cystatin C-estimated Glomerular Filtration Rate
in Pediatric Autologous Hematopoietic
Stem Cell Transplantation
Benjamin L. Laskin,
Jane C. Khoury,
Stella M. Davies,
Formal evaluation of kidney function is essential to determine chemotherapy dosing based on established
treatment protocols in children undergoing autologous stem cell transplantation. Cystatin C has been widely
studied as a marker of the glomerular ﬁltration rate (GFR), although data regarding its use in stem cell trans-
plantation are limited. We evaluated the performance of cystatin C-based equations and determined their
sensitivity to detect a nuclear GFR of \100 mL/min/1.73 m
in children undergoing autologous transplanta-
tion. We performed a retrospective cohort analysis in 16 children undergoing 26 transplantations using
a modiﬁed Bland-Altman analysis to account for repeated measures. Cystatin C-based equations published
by Hoek, Le Bricon, Rule, Filler, Zappitelli, Larsson, and Schwartz (the New Chronic Kidney Disease in Chil-
dren formula, New CKiD formula) were evaluated and compared to the creatinine-based modiﬁed Schwartz
equation. We found that cystatin C-based equations demonstrated improved sensitivity to detect a nuclear
GFR of \100 mL/min/1.73 m
compared to the creatinine-based modiﬁed Schwartz equation, which signif-
icantly overestimated GFR. Most cystatin C-based equations, however, tended to underestimate the nuclear
GFR. The New CKiD formula, combining cystatin C and creatinine, offered a sensitivity of 100% and a spec-
iﬁcity of 95% for detecting a nuclear GFR \100 mL/min/1.73 m
. Institutions using cystatin C-based GFR
estimation should be aware of the speciﬁc prediction formula and GFR measurement techniques available
at their center, as each method’s performance can vary considerably. As more research becomes available,
this easily measured marker should become a valuable component of GFR estimation, providing cost savings
(a nuclear GFR is 5.5 times more costly than a cystatin C) and reducing radiation exposure.
Biol Blood Marrow Transplant 18: 1745-1752 (2012) Ó 2012 American Society for Blood and Marrow Transplantation
KEY WORDS: Cystatin C, Kidney function, Autologous transplantation, Pediatrics
Measuring the glomerular ﬁltration rate (GFR) is
crucial for determining appropriate drug dosing,
monitoring the effects of therapeutic interventions,
and following the progression of chronic kidney
disease [1-3]. In pediatric autologous hematopoietic
stem cell transplantation (HSCT) treatment
protocols, chemotherapy dosing is commonly based
on renal function, as patients with a GFR \100 mL/
receive reduced dosages, which can
affect toxicity proﬁles and therapeutic beneﬁt [4,5].
The current standard of care is to evaluate GFR
pretransplantation using formal assessments (ie,
nuclear isotope testing or inulin or iohexol clearance),
which are repeated before each autologous HSCT in
children undergoing more than one transplantation.
Each of these methods is considered a gold standard,
but they are time and labor intensive, invasive, more
costly than serum-based or urine-based creatinine
methods, and may require exposure to radiation [6,7].
Newer, le ss invasive biomarkers may provide an al-
ternative to formal GFR measurements [1,3,7].
Perhaps the most widely studied and most promising
is cystatin C, a small housekeeping protein produced
by all nucleated cells at a constant rate that can
be measured in a single peripheral blood sample [8,9].
In contrast to creatinine, serum cystatin C
concentrations are not signiﬁcantly affected by age
(after 1 year) , gender, and most relevant to the
HSCT population, muscle mass [10,11]. However,
cystatin C concentrations may be inﬂuenced by other
Division of Nephrology and Hypertension;
and Epidemiology; and
Bone Marrow Transplantation and Im-
mune Deﬁciency, Cincinnati Children’s Hospital Medical Cen-
ter, Cincinnati, Ohio.
Financial disclosure: See Acknowledgment on page 1751.
Correspondence and reprint requests: Benjamin L. Laskin, MD,
Division of Nephrology and Hypertension, The Children’s
Hospital of Philadelphia, 34th Street and Civic Center Boule-
vard, Philadelphia, PA 19104 (e-mail: email@example.com).
Received March 12, 2012; accepted June 9, 2012
Ó 2012 American Society for Blood and Marrow Transplantation
factors, including steroid therapy, the WBC count, and
thyroid function [3,7].
Although cystatin C estimation of GFR has been
studied in those with chronic kidney disease, recipients
of liver and kidney transplantations, and pediatric and
adult oncology patients, limited data exist on its use in
patients undergoing HSCT [7,12-15].Multiple
cystatin C-based prediction equations have been
published, each with varied agreement with gold
standard methods depending on the patient group
studied and measurement techniques used [7,12].
Our primary objective was to evaluate the performance
of these estimating equations and determine their
sensitivity to detect a nuclear GFR of \100 mL/
in children undergoing autologous
transplantations, including multiple HSCTs. In
addition, we evaluated the association of demographic
and laboratory parameters with cystatin C, after
adjustment of GFR, in order to explore potential
covariates that may improve accuracy of cystatin
C-based equations. Demographic covariates included
age, gender, height, weight, and the height-to-
creatinine ratio, whereas laboratory covariates included
creatinine, blood urea nitrogen, and the WBC count.
MATERIALS AND METHODS
We conducted a retrospective cohort analysis by
reviewing the records of all consecutive autologous
HSCTs performed at Cincinnati Children’s Hospital
Medical Center (CCH MC) over the last 5 years (end-
ing June 2011) to identify patients having both a cysta-
tin C and formal nuclear medicine GFR measurement
Tc-labeled diethylene triamine pentaacetic
acid [DTPA]). If these measurements were performed
on different days, we only included patients with stable
kidney func tion (deﬁned as an intrapatient creatinine
SD of \0.1 mg/dL) and without intervening therapy
or events that would predispose to kidney injury.
The study was approved by the CCHMC Institution al
Laboratory Assessments and GFR Estimation
All patients at our institution receive nuclear GFR
measurements (DTPA) before autologous HSCT, in-
cluding each HSCT if multiple transplantations are
planned. Nuclear GFR was performed and calculated
using the methods described by Balachandran et al.
. Speciﬁcally, GFR is determined using plasma
DTPA disappearance curves obtained from 4 sampling
time points (120, 150, 180, and 210 minutes) after a sin-
gle injection of a known amount of radioisotope for
each patient. Scintillation co unting allowed for the
creation of time activity curves to calculate GFR.
Cystatin C concentrations were measured using
particle-enhanced immunonephelometry (Siemens
Healthcare Diagnostics Products, 2008) in the
CCHMC Division of Nephrology clinical labora tory.
To determine the optimal prediction formula, we
estimated GFR using several equations published
over the past decade including Larsson , Filler
, Zappitelli ,Hoek, Rule , and Le
Bricon . We also assessed the accuracy of formulas
published by Zappitelli  and Schwartz  (the
New Chronic Kidney Disease in Children formula,
New CKiD), as these equations, including both
serum creatinine and cystatin C, have shown
improved precision and accuracy [12,15,22]. Finally,
as a comparison, GFR was estimated with serum
creatinine using the modiﬁed Sc hwartz formula, with
a constant of 0.413 . All assessed GFR estimating
equations are shown in Table 1.
All patients had pediatric solid tumors and received
high-dose chemotherapy followed by autologous pe-
ripheral blood stem cell rescue according to national
or institutional protocols. Patients received uniform
prophylaxis and clinical monitoring, including nuclear
GFR to determine renal function for chemotherapy
dosing before each transplantation. Serum creatinine
was monitored daily duringinpatientstays and regularly
during outpatient follow-up between transplantations.
Descriptive statistics for continuous variables were
reported as means and SDs and as medians and inter-
quartile rang es when there was signiﬁcant departure
from normality. For those patients having more than
one serum creatinine or blood urea nitrogen value be-
tween cystatin C and nuclear GFR measurements, the
mean value was used in the analysis. Because
we included only patients with stable serum creatinine
values between GFR measurements, we assumed
any intrapatient differences between creatinine
Table 1. GFR Estimating Equations
Cystatin C-based formula
Filler 91.62*(cys C)
Zappitelli 75.94*(cys C)
Larsson* 77.24*(cys C)
Hoek 24.32+(80.35/cys C)
Rule 66.8*(cys C)
Le Bricon (78/cys C) + 4
Formulas including both cystatin C and creatinine
Zappitelli 43.82*(cys C)
New CKiD 39.1(height (m)/SCr)
SCr indicates serum creatinine (mg/dL); BUN, blood urea nitrogen
(mg/dL); m, meters.
GFR in mL/min/1.73 m
unless indicated by *mL/min Cys C, Cystatin C
1746 Biol Blood Marrow Transplant 18:1745-1752, 2012B. L. Laskin et al.
measurements would be random and normally distrib-
uted, favoring use of the mean. We included the WBC
count from the same day the cystatin C was perform ed.
To compare cystatin C-based equations (Table 1)
and the modiﬁed Schwartz equation  to nuclear
GFR estimation, mean bias and 95% limits of agree-
ment were calculated according to the methods of
Bland-Altman. Accounting for the repeated nature of
the measurements in patients having more than one au-
tologous HSCT, we used a modiﬁed Bland-Altman
analysis as previously described . Brieﬂy, mean
bias was calculated assuming that the true value of
GFR changed between measurements. The 95% limits
of agreement were calculated by partitioning the
variance of the difference into within-patient and
between-patientvarianceof differences,which were cal-
culated using one-way analysis of variance .We
computed each equation’s accuracy (the proportion of
the estimated GFR within 10% and 30% of the gold
standard value) and also reported the proportion of var-
) for eachequation. In addition, we calculated
the sensitivity and speciﬁcity of each GFR estimating
equation for detecting a nuclear GFR of \100 mL/
(the cutoff most relevant to chemotherapy
dosing in the pediatric autologous HSCT population).
To evaluate the signiﬁcance of covariates, we used
generalized linear modeling to correct for repeated
measures. Cystatin C was the main independent vari-
able, and nuclear GFR was the dependent variable; co-
variates were then added to the model. Because certain
covariates have been associated with cystatin C con-
centrations , we included age, gender, height,
weight, creatinine, blood urea nitrogen, the height-
to-creatinine ratio (as used in the Schwartz formula),
and the WBC count in the model. Cystatin C, creati-
nine, and nuclear GFR were log transformed to satisfy
the assumptions of linear regression modeling. There-
fore, the general model used to assess the signiﬁcance
of covariates was as follows:
log ðcystatin CÞ1b
where X represents covariates added to the model and
ε is the residual values under which the assumptions of
linear regression modeling have a normal distribution.
Partial F tests were used to assess the signiﬁcance of
added covariates (assessed at a P value \.15). All anal-
yses were conducted using SAS statistical software
(version 9.2; SAS Institute, Cary, NC).
Of 100 pediatric autologous HSCT performed in
the last 5 years, we identiﬁed 17 patients who had a cys-
tatin C measurement performed. One patient was ex-
cluded because the serum creatinine decreased by 0.2
mg/dL (SD of 0.14 mg/dL) between the cystatin C
and nuclear GFR measurements. The 16 included pa-
tients underwent a total of 33 autologous HSCTs (me-
dian of 2 transplantations per patient; range, 1-4
transplantations). Of these 33 transplantations, 26
had both a nuclear GFR and cystatin C perform ed
and were, therefore, included in the ﬁnal analysis.
Clinical and laboratory inform ation for these 16 pa-
tients and 26 transplantations are shown in Table 2.
The mean cohort nuclear GFR was 128.1 6 45.7
In 13 of the 26 transplantations (50%), patients
had a cystatin C and nuclear GFR performed on the
same day, whereas in the remaining 13 transplanta-
tions, patients had these measurements performed
a median of 6 days apart (range, 1-13 days). In these
13 transplantations, the serum creatinine was mea-
sured a median of 2 times (range, 2-8 times) between
the cystatin C and nuclear GFR measurements. In 7
of these 13 transplantations (54%), the creatinine
was unchanged between cystatin C and nuclear GFR
measurements, and in the remaining 6, the creatinine
Table 2. Clinical and Demographic Characteristics of the
Study Cohort (16 Patients and 26 Autologous HSCT)
Neuroblastoma 6/16 (37%)
Medulloblastoma 6/16 (37%)
Atypical teratoid rhabdoid tumor 2/16 (12%)
Relapsed clear cell sarcoma 1/16 (6%)
Relapsed hepatoblastoma 1/16 (6%)
Age at time of HSCT (years) 3.7 [3.1-6.3] (1.8-30.5)
Male gender 9/16 (56%)
White 15/16 (94%)
African American 1/16 (6%)
Height (cm) 99.9 [97.3-116.6] (79.5-170)
Weight (kg) 17.0 [14.0-22.2] (11.2-70.6)
No. of HSCT with cystatin C and nuclear
GFR on same day
Days between cystatin C and nuclear
GFR (n 5 13 HSCTs)
6 [3-9] (1-13)
Serum creatinine (mg/dL)* 0.3 [0.2-0.4] (0.2-0.7)
Blood urea nitrogen (mg/dL)* 12 [8-14.5] (3.5-25)
WBC count (K/L) 4.1 [2.3-5.7] (1.3-17.8)
Cystatin C (mg/L) 0.7 [0.6-0.9] (0.5-1.2)
Steroid therapy at time of cystatin C 0/16 (0%)
Thyroid replacement therapy at time
of cystatin C
Abnormal TSH 3/15 (20%)
Abnormal free thyroxine (T4) 1/13 (7%)†
HSCT indicates hematopoietic stem cell transplantation; GFR, glomeru-
lar ﬁltration rate; TSH, thyroid-stimulating hormone.
Data shown as median [interquartile range] (range) or number (%).
*For patients not having a cystatin C and nuclear GFR performed on the
same day (n 5 13 HSCT), the available serum creatinine and blood urea
nitrogen values between the cystatin C and nuclear GFR measurements
were averaged (median of 2 creatinine/blood urea nitrogen values per
HSCT; range, 2-8).
†One patient had a low TSH and elevated T4 and free T4 (hyperthyroid)
at the time of cystatin C measurement but 2 months later developed hy-
pothyroidism requiring replacement therapy.
Biol Blood Marrow Transplant 18:1745-1752, 2012
1747Cystatin C in Autologous Transplant
was stable, with an intrapatient creatinine SD of 0.05
to 0.07 mg/dL. Furthermore, betwee n GFR measure-
ments, all patients were outpatients and did not receive
any chemotherapy, antibiotics (other than stable acy-
clovir prophylaxis), or other nephrotoxic medications
that would predispose them to kidney injury.
At the time of assessment, no patient was pre-
scribed steroid therapy and no patient was receiving
thyroid hormone replacement. Fifteen of 16 patients
had thyroid function studies available, and 14 of 15
had this assessment performed before the autologous
HSCT that was included in the analysis. In the remain-
ing patient, thyroid studies were performed for the ﬁrst
time 5 months after transplantation and on the same
day as the cystatin C measurement. Three of 15 pa-
tients (20%) had a low thyroid-stimulating hormone
(TSH) level, and the remaining 12 of 15 patients
(80%) had a normal TSH. Of these 3 patients with
a low TSH, 2 (67%) had a nor mal free thyroxine
(T4), whereas the remaining patient had a mildly ele-
vated free T4 level when thyroid studies were obtained
5 months after transplantation. However, 2 months
later, this same patient became hypothyroid and re-
quired levothyroxine therapy.
We assessed the association between changes in
nuclear GFR and cystatin C in the 8 patients undergo-
ing .1 HSCT during the study period (6 patients with
2 HSCTs, 2 patients with 3 HSCTs each). Overall, the
nuclear GFR decreased by a mean 11% (median 16%)
between transplantations in these patients, whereas the
cystatin C estimated GFR (Larsson formula currently
used in clinical care at CCHMC) decreased by a mean
5% (median 7%). In 6 of 8 of these patients, the nu-
clear GFR and cystatin C estimated GFR both de-
creased between the ﬁrst 2 transplantations analyzed
(in the remaining 2 patients, the cystatin C and nuclear
GFR measurements moved in opposite directions). In
these 6 patients, the nuclear GFR decreased by a mean
20% (median 23%) between transplantations, whereas
the cystatin C GFR decreased by a mean 12% (median
8%). In 1 patient, 2 cystatin C measurements were
inadvertently obtained 11 days apart, before their 4th
autologous HSCT. In this patient, the L arsson esti-
mated cystatin C GFR decreased by 11% over the
11-day period, and the serum creatinine level rem ained
unchanged at 0.3 mg/dL.
Evaluation of Covariates
Speciﬁcally, the association between the nuclear
GFR and cystatin C concentration was independent
of patients’ age, gender, hei ght, weight, creatinine,
blood urea nitrogen, and the WBC count. However,
the height-to-creatinine ratio (as used in the Schwartz
formula ) was signiﬁcantly associated (P 5 .06)
with the nuclear GFR when included in the model
with the cystatin C concentration.
Sensitivity and Speciﬁcity to Predict a GFR \100
As shown in Tabl e 3, the sensitivity of each esti-
mating equation to detect a nuclear GFR \100 mL/
varied from 50% to 100%. The modiﬁed
Schwartz formula and the Zappitelli equation, includ-
ing both cystatin C and creatinine, performed the
worst with sensitivities of 50% and 66.7%, respec-
tively. Cystatin C formulas demonstrated improved
sensitivity with values ranging from 83% to 100%.
Overall, the New CKiD formula provided the best
combination of sensitivity (100%) and speciﬁcity
(95%). The Larsson formula, currently used for
GFR estimation at our institution, had a sensitivity
of 83.3% and a speciﬁcity of 85% for detecting a nu-
clear GFR \100 mL/min/1.73 m
. In our cohor t,
this meant that of 6 HSCTs with a nuclear GFR
\100 mL/min/1.73 m
, the Larsson formula correctly
detected this level of kidney function in 5 of 6 HSCTs
(83.3% sensitivity). In the remaining 20 HSCTs, the
Table 3. Overall Performance of GFR Estimating Equations
Formula Mean GFR (SD) Mean Bias (95% CI) 95% LOA Sensitivity (%) Speciﬁcity (%) R
30% Accuracy† (%) 10% Accuracy† (%)
Nuclear GFR 128.1 ± 45.7 – – – – – – –
New Schwartz 141.2 ± 38.2* 13.1 (1.5, 24.7) (243.9, 70.1) 50.0 100.0 0.61 73.1 34.6
Cystatin C formulas
Filler 139.0 ± 39.4 10.9 (20.3, 21.9) (243.4, 65.2) 83.3 95.0 0.65 73.1 30.8
Zappitelli 117.5 ± 34.6 210.6 (221.6, 0.3) (264.9, 43.7) 83.3 80.0 0.65 88.5 38.5
Larsson 124.1 ± 39.3 24.0 (215.0, 7.0) (258.6, 50.6) 83.3 85.0 0.65 76.9 34.6
Hoek 111.7 ± 29.4* 216.4 (227.7, 25.1) (272.3, 39.5) 83.3 80.0 0.65 84.6 26.9
Rule 109.0 ± 35.6* 219.1 (230.1, 28.1) (273.5, 35.3) 100.0 80.0 0.65 84.6 19.2
Le Bricon 116.7 ± 28.5* 211.5 (222.9, 20.1) (267.9, 44.9) 83.3 85.0 0.65 76.9 30.8
Combined cystatin C and creatinine formulas
Zappitelli 145.8 ± 44.1* 17.7 (7.2, 28.2) (234.4, 69.8) 66.7 95.0 0.69 73.1 30.8
New CKiD 118.4 ± 28.6 29.7 (221.2, 1.8) (267.0, 47.6) 100.0 95.0 0.64 80.8 38.5
GFR indicates glomerular ﬁltration rate; CI, conﬁdence interval; LOA, limits of agreement.
Sensitivity and speciﬁcity to detect a GFR <100 mL/min/1.73 m
*Signiﬁcantly different from nuclear GFR (P < .05).
†Thirty percent accuracy is the percentage of estimated GFR values falling within 30% of the nuclear GFR, and 10% accuracy is the percentage of es-
timated GFR values falling within 10% of the nuclear GFR.
1748 Biol Blood Marrow Transplant 18:1745-1752, 2012B. L. Laskin et al.
Larsson formula correctly identiﬁed a GFR .100 mL/
in 17 of 20 HSCTs (85% speciﬁcity).
Conversely, the New CKiD formula correctly de-
tected a nuclear GFR \100 mL/min/1.73 m
in 6 of
6 HSCTs (100% sensitivity) and a GFR .100 mL/
in 19 of 20 HSCTs (95% speciﬁcity).
Performance of the GFR Estimating Equations
Table 3 summarizes the Bland-Altman analyses in
our cohort. In general, most cystatin C-only based for-
mulas (except the Filler equation) tended to underesti-
mate nuclear GFR, as evidenced by their negative
mean bias. However, based on the 95% conﬁdence in-
tervals, this negative mean bias was statistically signif-
icant only for the Hoek (unde restimated true GFR by
16.4 mL/min/1.73 m
), the Rule (underestimated true
GFR by 19.1 mL/min/1.73 m
), and Le Bricon formu-
las (underestimated true GFR by 11.5 mL/min/1.73
). Conversely, the modiﬁed Schwartz (creatinine-
based) and Zappitelli (including both cystatin C and
creatinine) signiﬁcantly overestimated GFR by 13.1
to 17.7 mL/min/1.73 m
. The remaining equations
shown in Table 3 (Filler, Larsson, cystatin C only Zap-
pitelli, and New CKiD) had a mean bias that was not
signiﬁcantly different from the nuclear GFR.
All estimating equations predicted .73% of GFR
values within 30% of the nuclear GFR. This 30% accu-
racy was best for the cystatin C-only Zappitelli formula
(88.5% of values within 30% of the true GFR). All es-
timating equations predicted .19% of GFR values
within 10% of the gold standard, nuclear GFR. The
Zappitelli formula with only cystatin C and the New
CKiD formulas had the best accuracy at this level, pre-
dicting 38.5% of values within 10% of the true GFR.
The 95% limits of agreement are also shown in
Table 3, with narrower limits representing greater pre-
cision. All estimating equations demonstrated similar
precision, with the combined Zappitelli formula having
the narrowest limits (10 4.2) and the New CKiD for-
mula having the widest (114.6). Finally, all formulas
demonstrated similar R
values (proportion of the var-
iability in true GFR explained by each estimating equa-
tion) ranging from 0.61 (modiﬁed Schwartz formula) to
0.69 (combined Zapp itelli formula).
To the best of our knowledge, this is the largest
study estimating renal function with cystatin C in pe-
diatric autologous HSCT recipients. We evaluated
the sensitivity, speciﬁcity, and overall performance of
GFR prediction formulas. Most of our patients did
not have signiﬁcant kidney disease, as shown by the co-
hort mean nuclear GFR of 128 mL/min/1.73 m
New CKiD formula , including both cystatin C
and creatinine, performed well, with a sensitivity of
100% and a speciﬁcity of 95% for detecting a nuclear
GFR of \100 mL/min/1.73 m
. It offered the best ac-
curacy in predicting GFR within 10% of the true value
and a mean bias that was not signiﬁcantly different
from zero. Overall, cystatin C-based equations dem-
onstrated improved sensitivity to detect a GFR of
\100 mL/min/1.73 m
compared to the creatinine-
based modiﬁed Schwartz formula. We found that
most cystatin C-only–based equations (except the
Filler equation) tended to underestimate GFR,
whereas the modiﬁed Schwartz formula overestimated
Serum creatinine is an established marker of kidney
function and is less than ideal, especially in patients with
chronic illnesses and with reduced muscle mass [25,26].
Current standard of care at our institution is to formally
measure GFR using a radioisotope study in children
undergoing autologous HSCT. The procedure is
invasive, time consuming, more costly than serum-
based methods, and requires repeated exposure to radi-
ation in multiple transplantations [3,6,7]. The costs of
running these tests vary by geographic location
(Table 4) [27-29]. The United States Medicare data
from 2012 show that a nuclear GFR is 5.5 times more
costly than a cystatin C, and a cystatin C is 2.7 times
more costly than a serum creatinine .
Accordingly, there is a need to develop novel
methods of GFR estimation that are reliable, less inva-
sive, and applicable to this patient population and
a wide range of others . Cystatin C is perhaps the
most promising marker of kidney function in this con-
text, as several estimating equations have already been
developed, the assay is easy to perform, and it has been
studied in several disease states [12,15,17-22].
Other investigators have assessed the ability of cys-
tatin C to estimate kidney function in patients with
malignancy. Bakoush et al.  measured cystatin C
using an immunoturbidimetry assay in 147 adults
with newly diagnosed cancer. Using the cystatin C es-
timating equation publ ished by Grubb et al. , the
authors reported a reasonable correlation (R
and mean bias of -5 mL/min/1.73 m
compared to io-
hexol GFR measurement. Conve rsely, the creatinine-
based equation overestimated GFR by 15 mL/min/
. Blufpand et al.  studied cystatin C
Table 4. Cost Estimates for Serum and Nuclear GFR Pre-
Bogota, Colombia (2006 US$)  $2.00 $27.00 –
Madrid, Spain (2006 US$)  $0.08 $3.00 –
United Kingdom (2003)  $0.15 $3.00 –
Belgium (2003)  $0.02 $2.00 –
United States Medicare (2012)*  $7.26 $19.25 $106.20
GFR indicates glomerular ﬁltration rate.
*Non facility price National Payment Amount.
Biol Blood Marrow Transplant 18:1745-1752, 2012
1749Cystatin C in Autologous Transplant
(measured by nephelometry) in 68 children with cancer
up to 3 months postchemotherapy. Cystatin C GFR
calculated with the Filler equation was compared
with the modiﬁed Schwartz formula and inulin. Com-
pared to inulin, GFR estimation using the Filler equa-
tion and the modiﬁed Schwartz formula had a mean
bias of 27.3 mL/min/1.73 m
and 214.3 mL/min/
, respectively. Regarding accuracy, 82.4%
(Filler) and 72.1% (modiﬁed Schwartz) of values were
within 30% of inulin. Finally, Chew et al.  com-
pared cystatin C (measurement assay not reported) to
DTPA-measured GFR in adult oncology patients.
They reported that the Le Bricon, Hoek, and Rule for-
mulas showed no signiﬁcant mean bias, and the Hoek
formula had the best accuracy at the 30% level (89%).
In contrast, fewer studies have measured cystatin C
in the HSCT population. Recently, Bacchetta et al.
 published a well-designed cross-sectional analysis
validating cystatin C-based equ ations in 252 children
at high risk of renal injury. However, only 7 of the chil-
dren in their cohort had an HSCT. Compared to inu-
lin, the authors assessed cystatin C (nephelome try)
estimating equations (Hoek, Le Bricon, Larsson,
Rule, Filler, Zappitelli, and Bouvet) and both the orig-
inal and modiﬁed Schwartz formulas. They were not
able to assess the New CKiD formula because they
did not have blood urea nitrogen values available. All
formulas performed well except for the Filler and orig-
inal Schwartz equations (which both overesti mated
GFR) and the Hoek and Bouvet formulas (which
underestimated GFR). The authors recommended us-
ing the Larsson formula (without correcting for body
surface area) or the Le Bricon equation in their study
population. Two other studies have assessed the Filler
equation to estimate GFR after HSCT. Hazar et al.
 prospectively analyzed 34 children after allogeneic
HSCT and compared nuclear (DTPA), creatinine, and
cystatin C-based GFR (nephelometry). Similar to our
ﬁndings, creatinine-based GFR formulas overesti-
mated DTPA GFR, whereas cystatin C did not. Inter-
estingly, creatinine-based GFR equations, which have
been shown to underestimate GFR in healthy adoles-
cents , overestimated GFR after HSCT. There-
fore, creatinine-based GFR estimation may not be
useful to detect decreased GFR after HSCT, possibly
due to the inﬂuence of decreased muscle mass in this
population. Finally, Aydin et al.  measured cystatin
C (nephelometry) and a 2-sample DTPA in 31 chil-
dren 1 month post-HSCT. Although the Filler equa-
tion had a low correlation with GFR (r 5 0.43), it
demonstrated reasonable mean bias (23 mL/min/
) and precision (95% limits of agreement
from 257.8 to 51.8).
Debate continues as to whether cystatin C GFR es-
timation is dependent on nonrenal factors such as ste-
roid use, malignancy, thyroid disease, or the WBC
count [3,13,24]. Some have reported that cystatin C is
disproportionally increased in patients with active
malignancy [37,38], whereas others have failed to
demonstrate this association . Similar controver-
sies exist regarding the independent effect of steroid
therapy and the WBC count on cystatin C levels
[24,39,40]. We were unable to ﬁnd an association
between the WBC count and cystatin C in our
cohort. As none of our patients were receiving steroid
therapy and most had normal thyroid function as
measured before autologous HSCT, we were unable
to generate conclusions ba sed on these covariates.
The strengths of our analysis include use of an es-
tablished methodology to compar e agreement using
repeated measures data, the inclusion of multiple
GFR estimating equations, a gold standard nuclear
medicine technique with 4 sampling time poin ts, and
cystatin C measurements with a nephelometric assay.
Nephelometry, as opposed to immunoturbidimetry,
has been shown to potentially result in more reliable
GFR estimation [3,8]. In contrast to most prior
studies in the HSCT population, we evaluated the
effect of several clinical and laboratory covariates on
cystatin C, including the WBC count.
Our analysis has several limitations. First, our co-
hort was relatively small. However, we did ﬁnd bias,
precision, and accuracy results that were similar to
the published literature. Furthermore, a large propor-
tion (50%) of our cystatin C and nuc lear GFR measure-
ments were performed on different days. We believe
the effect of this was likely small because the patients
had stable creatinine values and no intervening nephro-
toxic therapy between measurement s. Finally, we were
unable to assess the inﬂuence of all potential covariates
previously reported to affect cystatin C GFR estima-
tion . However, co mpared to a heterogeneous allo-
geneic HSCT population often receiving steroids,
our autologous HSCT cohort was relatively homoge-
nous, providing an opportunity to study GFR estimat-
ing equations in a uniform group of patients.
In conclusion, we demonstrate that cystatin C-
based equations performed better than creatinine in
identifying decreased GFR in pediatric patients who
undergo HSCT receiving multiple transplantations.
The New CKiD formula, combining cystatin C, creat-
inine, and blood urea nitrogen offered a sensitivity of
100% and a speciﬁcity of 95% for detecting a GFR
\100 mL/min/1.73 m
. Cystatin C-only based formu-
las demonstrated improved sensitivity to detect a GFR
\100 mL/min/1.73 m
compared to the modiﬁed
Schwartz equation, although most tended to underes-
timate GFR. Institutions using cystatin C-based GFR
estimation for clinical care should be aware of the
speciﬁc prediction formula available at their center,
as each equation’s performance can vary considerably.
Similarly, clinicians should be mindful of the various
creatinine, cystatin C, and gold standard (ie, nuclear
GFR) measurement techniques used at their
1750 Biol Blood Marrow Transplant 18:1745-1752, 2012B. L. Laskin et al.
institution, as differenc es between technique and
across hospitals can lead to variable estimation of kid-
ney function. As more research becomes available on
standardized cystatin C assays , and the impact of
additional covariates (such as steroids or thyroid dis-
ease), this easily measured marker should become a use-
ful alternative to formal GFR measurements, offering
cost savings and decreasing radiation exposure. A pre-
diction formula that combines cystatin C, creatinine,
and additional covariates will likely provide the greatest
agreement compared with those formulas only includ-
ing a single biological marker [3,15]. Importantly, the
ease of measurement of cystatin C in a single
peripheral blood sample means that this assay can
easily be used for serial assessment of GFR during the
course of transplantation in persons with high risk of
kidney injury. Future rese arch should aim to validate
these ﬁndings in larger, multicenter cohorts against
standardized measurements of kidney function.
Financial disclosure: No ﬁnancial support was re-
ceived. The authors declare no conﬂicts of interest.
1. Stevens LA, Coresh J, Greene T, Levey AS. Assessing kidney
function–measured and estimated glomerular ﬁltration rate. N
Engl J Med. 2006;354:2473-2483.
2. Filler G, Sharma AP. How to monitor renal function in pediatric
solid organ transplant recipients. Pediatr Transplant. 2008;12:
3. Schwartz GJ, Work DF. Measurement and estimation of GFR in
children and adolescents. Clin J Am Soc Nephrol. 2009;4:
4. Matthay KK, Tan JC, Villablanca JG, et al. Phase I dose escala-
tion of iodine-131-metaiodobenzylguanidine with myeloabla-
tive chemotherapy and autologous stem-cell transplantation in
refractory neuroblastoma: a new approaches to Neuroblastoma
Therapy Consortium Study. J Clin Oncol. 2006;24:500-506.
5. Matthay KK, Villablanca JG, Seeger RC, et al. Treatment of
high-risk neuroblastoma with intensive chemotherapy, radio-
therapy, autologous bone marrow transplantation, and 13-cis-
retinoic acid. Children’s Cancer Group. N Engl J Med. 1999;
6. Kletzel M, Pirich L, Haut P, Cohn RA. Comparison of Tc-99
measurement of glomerular ﬁltration rate vs. calculated creati-
nine clearance to assess renal function pretransplant in pediatric
patients undergoing hematopoietic stem cell transplantation.
Pediatr Transplant. 2005;9:584-588.
7. Andersen TB, Eskild-Jensen A, Frøkiaer J, Brøchner-
Mortensen J. Measuring glomerular ﬁltration rate in children;
can cystatin C replace established methods? A review. Pediatr
8. Dharnidharka VR, Kwon C, Stevens G. Serum cystatin C is su-
perior to serum creatinine as a marker of kidney function:
a meta-analysis. Am J Kidney Dis. 2002;40:221-226.
9. Filler G, B
okenkamp A, Hofmann W, Le Bricon T, Mart
Bru C, Grubb A. Cystatin C as a marker of GFR–history, indi-
cations, and future research. Clin Biochem. 2005;38:1-8.
okenkamp A, Domanetzki M, Zinck R, Schumann G, Byrd D,
Brodehl J. Cystatin C–a new marker of glomerular ﬁltration rate
in children independent of age and height. Pediatrics. 1998;101:
11. Filler G, Witt I, Priem F, Ehrich JH, Jung K. Are cystatin C and
beta 2-microglobulin better markers than serum creatinine for
prediction of a normal glomerular ﬁltration rate in pediatric sub-
jects? Clin Chem. 1997;43(6 Pt 1):1077-1078.
12. Zappitelli M, Parvex P, Joseph L, et al. Derivation and validation
of cystatin C-based prediction equations for GFR in children.
Am J Kidney Dis. 2006;48:221-230.
oge U, Gerhardt T, Stoffel-Wagner B, et al. Cystatin C-based
calculation of glomerular ﬁltration rate in kidney transplant re-
cipients. Kidney Int. 2006;70:204-210.
14. Bacchetta J, Cochat P, Rognant N, Ranchin B, Hadj-Aissa A,
Dubourg L. Which creatinine and cystatin C equations can be
reliably used in children? Clin J Am Soc Nephrol. 2011;6:552-560.
15. Schwartz GJ, Mu
noz A, Schneider MF, et al. New equations to
estimate GFR in children with CKD. J Am Soc Nephrol. 2009;20:
16. Balachandran S, Toguri AG, Petrusick TW, Abbott LC. Com-
parative evaluation of quantitative glomerular ﬁltration rate
measured by isotopic and nonisotopic methods. Clin Nucl Med.
17. Larsson A, Malm J, Grubb A, Hansson LO. Calculation of glo-
merular ﬁltration rate expressed in mL/min from plasma cysta-
tin C values in mg/L. Scand J Clin Lab Invest. 2004;64:25-30.
18. Filler G, Lepage N. Should the Schwartz formula for estimation
of GFR be replaced by cystatin C formula? Pediatr Nephrol.
19. Hoek FJ, Kemperman FA, Krediet RT. A comparison between
cystatin C, plasma creatinine and the Cockcroft and Gault for-
mula for the estimation of glomerular ﬁltration rate. Nephrol
Dial Transplant. 2003;18:2024-2031.
20. Rule AD, Bergstralh EJ, Slezak JM, Bergert J, Larson TS. Glo-
merular ﬁltration rate estimated by cystatin C among different
clinical presentations. Kidney
21. Le Bricon T, Thervet E, Froissart M, et al. Plasma cystatin C is
superior to 24-h creatinine clearance and plasma creatinine for
estimation of glomerular ﬁltration rate 3 months after kidney
transplantation. Clin Chem. 2000;46(8 Pt 1):1206-1207.
22. Bouvet Y, Bouissou F, Coulais Y, et al. GFR is better estimated
by considering both serum cystatin C and creatinine levels.
Pediatr Nephrol. 2006;21:1299-1306.
23. Bland JM, Altman DG. Measuring agreement in method com-
parison studies. Stat Methods Med Res. 1999;8:135-160.
24. Stevens LA, Schmid CH, Greene T, et al. Factors other than
glomerular ﬁltration rate affect serum cystatin C levels. Kidney
25. Laskin BL, Goebel J, Davies SM, et al. Early clinical indicators
of transplant-associated thrombotic microangiopathy in pediat-
ric neuroblastoma patients undergoing auto-SCT. Bone Marrow
26. Laskin BL, Goebel J, Davies SM, Jodele S. Small vessels, big
trouble in the kidneys and beyond: hematopoietic stem cell
transplantation-associated thrombotic microangiopathy. Blood.
27. Centers for Medicare & Medicaid Services. Physician fee sched-
ule look-up and clinical laboratory fee schedule. Available at:
http://www.cms.gov/home/medicareasp.Accessed March6, 2012.
28. Narvaez-Sanchez R, Gonzalez L, Salamanca A, et al. Cystatin C
could be a replacement to serum creatinine for diagnosing and
monitoring kidney function in children. Clin Biochem. 2008;41:
29. Lamb EJ, O’Riordan SE, Webb MC, Newman DJ. Serum cys-
tatin C may be a better marker of renal impairment than creat-
inine. J Am Geriatr Soc. 2003;51:1674. Author reply 1674-1675.
30. Bakoush O, Grubb A, Rippe B. Inaccuracy of GFR predictions
by plasma cystatin C in patients without kidney dysfunction
and in advanced kidney disease. Clin Nephrol. 2008;69:331-338.
31. Grubb A, Nyman U, Bj
ork J, et al. Simple cystatin C-based pre-
diction equations for glomerular ﬁltration rate compared with
the modiﬁcation of diet in renal disease prediction equation
Biol Blood Marrow Transplant 18:1745-1752, 2012
1751Cystatin C in Autologous Transplant
for adults and the Schwartz and the Counahan-Barratt predic-
tion equations for children. Clin Chem. 2005;51:1420-1431.
32. Blufpand HN, Tromp J, Abbink FC, et al. Cystatin C more ac-
curately detects mildly impaired renal function than creatinine
in children receiving treatment for malignancy. Pediatr Blood
33. Chew JS, Saleem M, Florkowski CM, George PM. Estimating
renal function in oncology patients using cystatin C-based equa-
tions. Clin Oncol (R Coll Radiol). 2009;21:425-426.
34. Hazar V, Gungor O, Guven AG, et al. Renal function after he-
matopoietic stem cell transplantation in children. Pediatr Blood
35. Fadrowski JJ, Neu AM, Schwartz GJ, Furth SL. Pediatric GFR
estimating equations applied to adolescents in the general pop-
ulation. Clin J Am Soc Nephrol. 2011;6:1427-1435.
36. Aydin F, Tezcan G, G€ung
or O, et al. Can serum cystatin C re-
ﬂect the glomerular ﬁltration rate accurately in pediatric patients
under chemotherapeutic treatment? A comparative study with
Tc-99m DTPA two-plasma sample method. Nucl Med Commun.
S, Akan O, Can M, Elmali E, Akan H. Cystatin C can
be affected by nonrenal factors: a preliminary study on leukemia.
Clin Biochem. 2006;39:115-118.
38. Kos J, Stabuc B, Cimerman N, Br€unner N. Serum cystatin C,
a new marker of glomerular ﬁltration rate, is increased during
malignant progression. Clin Chem. 1998;44:2556-2557.
39. Bardi E, Dobos E, Kappelmayer J, Kiss C. Differential effect of
corticosteroids on serum cystatin C in thrombocytopenic pur-
pura and leukemia. Pathol Oncol Res. 2010;16:453-456.
onroos MH, Jahnukainen T, Irjala K, et al. Comparison of
glomerular function tests in children with cancer. Pediatr Neph-
41. Weinert LS, Camargo EG, Soares AA, Silveiro SP. Glomerular
ﬁltration rate estimation: performance of serum cystatin C-
based prediction equations. Clin Chem Lab Med. 2011;49:
1752 Biol Blood Marrow Transplant 18:1745-1752, 2012B. L. Laskin et al.