Biomedical Engineering Reference
In-Depth Information
binding such as binding to cells without the target antigen (i.e., CHO alone) or binding
to cells expressing the target antigen without the drug.
11.2.5 Preparation of Calibration (Reference) Standard
and Quality Control
Well-characterized, sufficiently purified protein drugs spiked into appropriate bio-
logical matrix (such as serum or plasma) should be used for reference standard and
quality control (QC) samples. The reference standard should be evaluated for any
biological matrix effect (or interference) that could result in enhanced or reduced
signal. Using drug spiked in buffer may also result in over- or underdetection of the
drug in study samples and is not recommended.
Pooled normal human (or animal) serum (or plasma) is typically used as the
biological matrix. In general, the pool is composed of individual samples fromhealthy
subjects screened for lack of interference. A minimal sample dilution needs to be
defined using criteria similar to that for immunoassays. The percentage of serum used
in preparing reference standard and QC samples should be the same as study samples
being tested.
Generation of a reference standard at different concentrations across the calibra-
tion standard curve concentration range is advised. The number of standard points
needed for a good curve fit will depend on the curve-fit model and assay performance
in general. An extended curve range should be evaluated, which can be narrowed to
improve curve fit.
QC samples typically consist of high (H), medium (M), and low (L) levels of drug
covering the linear portion of the standard curve. A negative control, pooled normal
human or species-specific serum without the drug may also be used.
11.2.6 Curve Fit
Multiple curve-fit models should be evaluated. Typically, multiple standard curves
from data collected by two-three analysts over multiple days are used for this
analysis. Most ligand binding assays, including the drug-target Ag binding in the
flow cytometric PK assay, have a sigmoidal dose-response, and a four-parameter
nonlinear curve fit is a commonly used model.
An appropriate fitted curve is expected to show random deviation from the best-
fitted curve (residual analysis) and acceptable percentage recovery of back-fit values
compared to the theoretical concentrations. Interpolated values from unknown
samples are ideally expected to have recovery within 80-120% from theoretical
values. Assay sensitivity is demonstrated to support assay's intended use.
Figures 11.3 and 11.4 show the dose-response curve from drug X and drug Y,
respectively. Corresponding histograms for drug X and representative histograms for
drug Y are shown in Figures 11.1 and 11.2, respectively. Four-parameter curve fit is
found to be suitable for both standard curves and criteria are met.
Search WWH ::
Custom Search