![]()
21
| 1 | 2006
Klinisk Biokemi i Norden
We want in this paper to challenge these statements,
especially the claim that S-Cystatin C as a measure
for GFR is clinically better than a calculation based
on creatinine, and worth the higher price.
Dr. A Larsson has kindly supplied us with raw
data on measurements of S-Creatinine, Cystatin C
and GFR measured as clearance of iohexol, used in
his paper (12).
Calculations and graphs presented in the next
sections are based on these data. (12)
Material and Methods
Difference plots were constructed for measured GFR
and GFR calculated from S-Creatinine or S-Cystatin
C results. Deviation from GFR and standard deviation
are shown.
Correctness in prediction of GFR was evaluated by
using the correlation equation stated by Larsson. Use
of the recommended MDRD formula for obtaining
better accuracy and precision of GFR estimate was
not possible as data on age were lacking.
Ability to detect impaired GFR was determined by
means of True Negative (TN), True Positive (TP), False
Negative FN), False Positive (FP) and Predictive value
of positive test (PVpos) and Predictive value of nega-
tive test (PV neg), when GFR of less than respectively
60, 70 and 90 mL/min was chosen as cut-off.
Results
As shown by Larsson Cystatin C correlates better with
GFR than serum creatinine. The results for a signifi-
cant part of the patients with creatinine concentration
around 150 µmol/L do not fit the correlation curve.
Overall comparison GFR and S-Creatinine /
S-Cystatin C
To examine the potential effect of this insufficient
correlation on clinical evaluation we presented
Larsson’s data as
differences
between GFR calculated
from S-Creatinine (respectively Cystatin C) and GFR
measured (Fig.1).
Mean difference was –2 ± 15 mL/min for S-
Creatinine and –1 ± 10 mL/min for Cystatin C.
Comparison GFR and S-Creatinine /
S-Cystatin C for different patients groups
We investigated if the correlation between GFR mea-
sured and GFR calculated depends on GFR value. For
this purpose the results were divided in four groups
as recommend by National Kidney Foundation (7)
Severely reduced
GFR < 30 mL/min*1.73 m
2
Moderately reduced GFR 31 – 60 mL/min*1.73 m
2
Grey zone/
slightly reduced
GFR 61 – 90 mL/min*1.73 m
2
Normal values
GFR > 90 mL/min*1.73 m
2
Fig. 2 shows GFR calculated from S-Creatinine (or
Cystatin C) as mean differences and standard devia-
tions from true GFR, in those defined ranges of GFR.
It can be seen that Cystatin C and creatinine
perform equally well for patients with GFR < 90
mLmin*1.73m
2
.
Deviations and variations are small in the important
range from 10 to the lower limit of normal range 90
mL/min. Above this creatinine underestimates glo-
merular filtration rate, and Cystatin C is markedly
better than S-Creatinine.
GFR cut-off
Marker
TP TN FP FN Sens.
Spec.
PV pos PV neg
60 mL/min
Creatinine
47 40
6
6
0.887
0.870
0.887
0.870
Cystatin C 50 40
6
3
0.943
0.870
0.893
0.930
70 ml/min
Creatinine
54 26 15
4
0.931
0.634
0.783
0.867
Cystatin C 56 33
8
2
0.966
0.805
0.875
0.943
90 ml/min
Creatinine
80
3
13
3
0.964
0.188
0.860
0.500
Cystatin C 78 12
4
5
0.940
0.750
0.951
0.706
Performance characteristic of S-Creatinine and Cystatin C to detect reduced GFR at different cut-off points.
(Fortsætter side 24)
