Biology Reference
In-Depth Information
7.4.3.8 Precision
The reliability and reproducibility of any assay must be clearly demonstrated, and is the
focus of assay validation. The first step in defining the precision of the device is to identify
the key sources of variability. For some assays this is an easy process, yet for others it is much
more complex. Achieving precision of the device should be performed well in advance of
the actual analytical validation. Precision studies will then be designed for each of the key
sources of variability. For an example, we will utilize a hypothetical assay whose primary
sources of variability are the operator, day of run, and instrument. Typically factorial designs
are utilized to test the contribution of each of the sources of variability. To assess the contri-
bution of the sources to the variation, the samples are kept consistent throughout. The fol-
lowing is an example of a factorial design for the theoretical assay being utilized (
Fig. 7.7
).
This same simple design can be expanded to address additional sources of variability as
necessary. It is important to note that as additional factors are added to this design, the num-
ber of tests increases on an exponential scale; therefore, only the key sources of variability
should be included. Statistical models such as variance components or mixed effects can then
be used to identify and quantify the contribution of each factor to the observed variability. As
these studies are the focus of analytical validation, properly powering them and evaluating
samples that are representative of the intended patient population are critical.
7.4.3.9 Analytical Specificity
Each assay comprising the companion diagnostic device should be specific to its particular
analyte. Cross reactivity of the analytes with other intended targets can significantly compro-
mise the reliability of the assay call. Consideration for the cross reactivity of each assay should
be investigated early in the development of the assay. For a nucleic acid-based assay, basic log-
ical alignment search tool (BLAST) searches and other sequence alignment methods to assess
the overlap of the target with similar targets are useful tools in this effort. Careful attention to
the design of assays in the development phase will save significant time and ease the assess-
ment of specificity. Specificity should be demonstrated through specific cross reactivity studies
utilizing synthetic targets. Each assay should be tested for its cross reactivity with other assays
in the device to ensure it is specifically capturing data for the target of interest and not giving
signal from other targets. A graphical representation of this type of cross reactivity is shown
below with a theoretical device consisting of five assay targets (
Fig. 7.8
).
Assay performed by 3
different appropriately
trained operators
Assay performed on 3
different days
Assay performed on 3
different instruments
X
X
Results in each of the individual operators running 9 separate runs
incorporating sources of variability
FIGURE 7.7
A graphical representation of a standard factorial design utilized in assay validation to test multi-
ple sources of variation.
Search WWH ::
Custom Search