Whole Blood Protocol
Below is a protocol for calibration of creatinine concentration reported for whole blood to creatinine concentration in simultaneously collected venous serum or plasma.
The NIDDK Laboratory Working Group has recommended that creatinine concentrations measured in whole blood should be adjusted and reported to providers as equivalent creatinine concentrations measured in simultaneously collected venous serum or plasma, where the serum or plasma measurements are traceable1 to a high-level isotope dilution mass spectrometry (IDMS) Reference Method.2 The purpose of this protocol is to provide manufacturers of whole blood creatinine measurement devices practical guidance regarding the calibration of their devices to an IDMS-traceable serum or plasma method. The protocol may also be used by laboratories reporting creatinine measured in whole blood to verify agreement of the measurement to an IDMS traceable serum or plasma method.
Sensors for measurement of creatinine in whole blood used in clinical practice are enzyme-based biosensors with electrochemical detection. Like the more commonly used glucose sensors based on the same technology, creatinine sensors respond to the molality of creatinine in the sample (amount of creatinine per unit mass of water in the sample). It has been known for some time that molality of creatinine in erythrocyte fluid is equal to molality in plasma3, 4 and that creatinine is transported by passive diffusion through the lipid bilayer of the erythrocyte membrane.5 When these two sample phases are in equilibrium, the measured quantity in whole blood and its separated plasma by the direct biosensor method should be identical. The presence of erythrocytes may affect the whole blood measurement in other ways, for example, by hindering diffusion of analyte from bulk sample to the surface of the sensor, or by affecting the electrical conductivity between different elements of the electrochemical cell. Therefore, in practice, results obtained with whole blood and simultaneously collected serum or plasma often differ significantly, are usually sensitive to hematocrit, and correction for the presence of erythrocytes may be needed to force agreement between the whole blood and serum or plasma measurements.
This protocol assumes that interfering substances for the whole blood device have been characterized separately, and that samples used in Part 1 of the protocol are free of interfering substances. This protocol also assumes that the whole blood device has been shown to be linear toward creatinine concentrations over the manufacturer's claimed measurement range.
1. Calibration of whole blood creatinine method to a heparinized plasma IDMS-traceable method at variable hematocrit levels
- A heparinized venous whole blood sample from a healthy volunteer should be collected in heparin anticoagulated tubes and determined to be within the adult reference ranges for the following analytes.
|US Units||SI Units|
|Creatinine||0.8 - 1.5 mg/dL||70 - 132 µmol/L|
|Hematocrit||35 - 47% (w) 40 - 52% (m)||0.35 - 0.47 (w)
0.40 - 0.52 (m)
|Total protein||63 - 79 g/L||63 - 79 g/L|
|Cholesterol||150 - 250 mg/dL||5.7 - 9.4 mmol/L|
|Triglycerides||50 - 150 mg/dL||0.56 - 2.83 mmol/L|
- A portion of the whole blood sample should be centrifuged and the plasma removed.
- Divide the remaining blood sample into three aliquots. Adjust an aliquot of whole blood to low (~20 to 30 percent) and another aliquot to high (~55 to 65 percent) hematocrit by addition or removal of plasma. The actual low and high hematocrit to be evaluated should be determined based on the hematocrit reportable range claimed for the device. Retain the third blood aliquot at the unadjusted hematocrit level.
- The actual hematocrit value should be determined for each sample by microcentrifugation.
- Divide each blood sample, representing each of the three hematocrits, into three more aliquots. Adjust two of these aliquots to mid- and high-creatinine concentrations (see note below). The high target concentration should be approximately –10 percent from the highest concentration claimed by the manufacturer for their device. Retain the third aliquot at the unadjusted creatinine concentration. Two additional intermediate concentrations should be prepared by admixture of the unadjusted, mid- and high-creatinine blood samples.
- Note: Samples should be supplemented by addition of a concentrated stock solution of creatinine based on estimated water content of the whole blood sample, not the entire volume of whole blood.
- After spiking, the plasma concentration of samples that are supplemented with creatinine may continue to change over a period of time as the molality of the plasma and erythrocyte phases approach equality. This process requires approximately five hours at room temperature.6 If desired, the plasma concentration of creatinine from samples that have been spiked with creatinine may be measured as a function of time to determine how much time is required to reach equilibrium. These measurements can be done at the highest hematocrit only, which should require the longest time. Allow sufficient time after sample spiking before continuing with the assays.
- After adjustment of hematocrit and sample spiking to elevated creatinine concentrations, blood samples will be available according to the matrix below.
|Percent of Hematocrit|
- Split each blood sample into two portions. One portion should be centrifuged and the plasma removed and the other portion kept as whole blood.
- The 15 blood samples should be analyzed on the whole blood measurement device in random order, with the number of replicates per sample determined using the known imprecision of the whole blood measurement device. See Table 1 in the Appendix.7 More than one whole blood measurement device may be included in this phase of the protocol, at the discretion of the manufacturer.
- The 15 plasma samples should be assayed using an IDMS-traceable method in random order. Data from the isotope dilution mass spectrometry (IDMS)-traceable method should be normalized to creatinine in NIST Standard Reference Material (SRM) 967. The number of replicates per sample should be equal to the number of replicates for the corresponding whole blood sample.8
- Calculate mean and standard deviation for each set of measurements. The means of the determinations by the whole blood and plasma methods for the unspiked sample should be within the bias limits given in Table 1 of the Appendix, at each hematocrit level, for a given whole blood imprecision.
- If the required agreement is not obtained, the results from the two methods should be analyzed by linear regression, at each level of hematocrit, using the means of the whole blood results as the "y" value and the means of the plasma results as the "x" value. The resulting slope and intercept should be adjusted to one and zero, respectively, to force agreement between the two methods. This correction may then be made to subsequent whole blood results, in software, based on the slope and intercept adjustments.
corrected "y" = [(1/slope) (uncorrected "y")] + [(0 – intercept)/slope]
The slope and intercept may vary as a function of hematocrit. A secondary correction to slope and intercept based on measured hematocrit should be included, if necessary.9
2. Verification of agreement with an IDMS-traceable plasma method using patient samples
Patient specimens should be used to verify the above standardization procedure. It is recommended that at least 5 different whole blood creatinine measurement devices be used as part of this verification protocol in order to show device-to-device consistency.
- At minimum, 40 heparinized blood samples from hospitalized patients, chosen to span the measured range for creatinine as thoroughly as possible, should be included.
- Each sample should be divided in two parts. One part should be centrifuged and the plasma extracted and the other part retained as whole blood.
- Each blood specimen should be assayed in duplicate (at a minimum) on the whole blood device. The duplication is necessary to improve the imprecision of the estimate and may aid in determining if any measurements should be discarded as outliers.
- Each plasma sample should be assayed in duplicate (at a minimum) using an IDMS-traceable method, normalized to NIST SRM 967.
- Perform regression analysis on the data using the means of the whole blood results as the "y" or dependent variable and the means of the plasma results as the "x" or independent variable.
- Using CLSI Document EP9-A2 as a guide, determine if any points should be excluded as outliers and compute the average bias between the two methods. Any points discarded as outliers should be further investigated.
- Determine if the whole blood measurement device meets the NIDDK guidelines for agreement to the IDMS traceable method, especially in the creatinine concentration range of 1.0 – 1.5 mg/dL.
Figure 1: Total error budget for creatinine measurement in the range 88.4-133 µmol/L (1.00-1.50 mg/dL). The line represents the limit of systematic biases and random imprecisions that produce a relative increase of less than 10 percent in the root mean squared error when estimating GFR using the MDRD Study equation. Adapted from: Myers et al. Recommendations for Improving Serum Creatinine Measurement: A Report from the Laboratory Working Group of the National Kidney Disease Education Program. Clinical Chemistry. 2006;52:5-18.
Table 1: Number of replicates required based on the targeted agreement between the whole blood and IDMS-traceable methods and the known imprecision of the whole blood method (derived from Figure 1).
|Known within run standard deviation of the whole blood method (mg/dL)||Targeted bias between whole blood and IDMS-traceable plasma method (mg/dL)||Number of replicates required to detect targeted bias|
|0.09||-0.045 > x <>||52|
|0.08||-0.06 > x <>||21|
|0.07||-0.075 > x <>||11|
|0.06||-0.085 > x <>||6|
|0.05||-0.095 > x <>||4|
|0.04||-0.10 > x <>||2|