So far, gene-expression microarrays
have served as research tools. New technology, standards, and protocol
initiatives, however, could help microarrays make the leap into clinical
and pre-clinical settings, helping to chart biomarkers for toxicology
and efficacy.
| Aug
1, 2005 |
| By:
Thomas
Goralski |
| Pharmaceutical
Discovery |
|
Thomas Goralski
|
DNA microarrays are used to measure relative transcript abundance in RNA
samples, allowing researchers to identify differences in gene
expression. To date gene expression microarrays have predominantly
served discovery based research efforts, however, recent technological
improvements and standardization initiatives, as well as need driven
incentives, have opened the possibility of using this high-complexity
analytical test in both pre-clinical and clinical settings. Gene
expression profiles are now being used to establish biomarker patterns
that correlate with phenotypic states for both drug toxicological
responses and efficacy investigations (1). The FDA's recently released
guidance on the submission of pharmacogenomic data (2) acknowledges the
powerful role gene expression profiles can play in clinical trials and
potentially clinical diagnostics as companions to drugs.
Steve Casey
|
Clinical studies that have a direct effect on patient treatment require
a higher degree of laboratory rigor than retrospective or discovery
based investigational studies. Each step along the process and each
reagent incorporated must be thoroughly tested and validated. Several
agencies monitor laboratory performance (i.e. CFR Part 58 Good
Laboratory Practices) and CLIA accreditation is required for
reimbursement from insurance companies. However, many of these processes
are designed for single analyte tests that monitor only one gene per
assay, where positive and negative controls are easily included.
Microarray technology poses unique challenges because this assay
generates data from thousands of genes simultaneously with different
levels of confidence in each measure. This article discusses challenges
and requirements necessary to transition a genetic testing laboratory
from research to the clinic from an analytical perspective.
The Test Process
The complexity of the gene expression microarray has posed
serious challenges to laboratories needing to integrate this test into a
standard specimen test environment. As total ribonucleic acid (RNA) is
purified from the tissue, the amount and integrity of the RNA needs to
be determined, and a predetermined amount is used in the first enzymatic
reaction. This first reaction converts the minor component of the total
RNA that encodes the majority of the genetic complexity (the messenger
RNA, or mRNA) into copy DNA (cDNA). A second enzymatic reaction
transcribes the cDNA into copy RNA (cRNA), which both amplifies the mRNA
up to 5,000 fold and incorporates a labeled nucleotide. This labeled
target (or probe depending on the platform. For this article, we will
refer to the labeled cRNA as the target and the microarray polymers as
the probe) is then added to the microarray, which contains a very large
number (often more than one million) of substrate bound oligonucleotide
probes. Over time, based on nucleic sequence complementarity, the
labeled target in solution will hybridize to the substrate bound probes
in an amount that is proportional to their concentration in the
specimen. In the post-hybridization processing the labeled target that
failed to hybridize to the microarray is removed, and the microarray is
exposed to reagents that are designed to associate a fluorescent dye to
the probe associated labeled target. Quantification of the amount of
hybridization is achieved by confocal laser scanner, where the amount of
fluorescence that detected is proportional to the amount gene transcript
present in the tissue specimen.
This testing process requires specimens to undergo a dozen container
transfers, employs over 42 reagents (6 of them enzymes) and requires
almost one hundred pipetting steps, some of which may necessitate
accurate transfer of single micro-liter volumes. All of these activities
must be accurately tracked, monitored and recorded. A robust laboratory
process such as this requires the development and implementation of a
quality system capable of supporting the simultaneous measurement of
tens of thousands of analytes (gene transcripts) in a reliable manner.
To ensure reproducibility of results it is essential that test
procedures be developed that support reliable testing in a moderate to
high throughput specimen processing system. The analysts performing
these tests must be well trained and follow comprehensive standard
operating procedures (SOPs) to ensure that these multiple steps be
performed as reproducibly and accurately as possible.
Good Laboratory Practices
Data quality of this test assay is ultimately dependent on the
reproducibility of the results. The ability to detect differential
transcript abundance levels between specimens is directly related to the
ability to minimize variation between identical specimens. This can be a
considerable challenge given the number of independent analytes
(transcripts) assayed simultaneously. To optimize the performance value
of this complex test, a range of activities should be in place as
described below.
The performance of this test is highly dependent upon the reagents
used in each step of the test. Procedures must be in place that
guarantee that the test is ultimately reproducible with changing reagent
lots, ensuring comparability across time and larger sample sets. Because
of the large number of reagents involved (many of which are single
vendor supplied), dependency on any single reaction component is not
completely understood. Therefore validation of the reagents requires a
functional test that resembles or represents the actual test as much as
possible. To ensure proper validation, however, frequency and
comprehensibility must be determined with each lot change and
incorporate whole protocol validation.
- Equipment/Instrument Validation
The equipment requirements for this type of assay are broad, and include
as a minimum refrigerators, freezers, centrifuges, micropipettes, pH
meters, balances, heat blocks, thermal cyclers, hybridizations ovens,
fluidic instruments, and a confocal scanning laser-detection system.
Each of these components must be validated individually following
standardized test methods using traceable references where possible. The
measured performance of each component must be within predetermined
acceptable ranges, and monitored over time to ensure stability. The
expertise required to thoroughly validate equipment and instrumentation
can sometimes exceed the resources of the testing laboratory.
Fortunately independent vendors exist where these activities may be
outsourced.
- Processing Standard Operating Procedures and Analyst Training
The complexity of the testing procedure necessitates that each and every
step is performed as precisely as possible to minimize variation. Robust
SOPs must be developed that minimize processing variability, such as
reducing the frequency of low volume transfers of viscous enzyme
solutions. These components tend to be assay critical, and variations in
the amounts delivered can have pronounced effects on the outcome of the
assay. These procedures must be documented and described in sufficient
detail to allow material activities to be precisely performed. Analysts
must be thoroughly trained in the SOPs to ensure that all steps are
performed as reproducibly as humanly possible. The incorporation of
robotics can minimize variation, though this does not eliminate the need
for standardized processing documentation and activities.
- Sample Tracking and Chain of Custody
To accurately track the thirteen container transfers that occur during
specimen processing, a specimen tracking system that is extensive,
accurate and secure must be put in place to ensure that transfers occur
correctly. The use of sample tube barcodes can reduce the probability of
incorrect transfers, and internal sample spikes ("molecular
barcodes") can detect and monitor the occurrence of these container
transfers, maintaining sample identification integrity throughout the
process. Additionally the use of robotics can also help to ensure the
fidelity of sample transfers. Documented knowledge of the specifics of
the process (including reagent lots, testing analyst, specific SOPs, and
equipment used) is an invaluable resource to understanding test
performance and variation, including failure analysis.
The goal of all of these efforts is to ensure that the test system is
capable of consistently producing results that are fit for intended use.
Process Validation can be described as "...establishing by
objective evidence that a process consistently produces a result or a
product meeting its predetermined specifications" [21 CFR
820.3(z)(1)]. Therefore the establishment of performance specifications
must proceed hand-in-hand with the specifics of the test assay.
Several levels of controls can be
employed to monitor certain portions of the process. Points exist where
intermediary test products can be evaluated to determine the progression
of the testing process. Acceptance criteria can be established at these
nodes that can help to minimize processing variation.
Additionally, the inclusion of
internal labeling and hybridization controls, whose hybridization
results are at least in part independent of test specimens, can be
useful in understanding assay performance. Microarrays frequently
contain a limited number of probes that are homologous to transcripts
that are not expressed in the specimen under test (e.g. bacterial
gene controls in a mammalian RNA sample). These probes typically derive
from bacterial or viral sequences, but may also come from plants or
correspond to "nonsense" or "synthetic" transcripts.
These controls can be used to determine the lower limit of detection and
quantification, precision of signal values, accuracy of differential
expression values, and robustness. Because these probes represent a
minor fraction of all probes on the array, or of the probes in a
biomarker pattern, their performance needs to be extrapolated to the
remainder of signal values of interest. Fortunately, in any given test,
a substantial portion of the transcripts can be identified that do not
vary in their abundance levels. These test-specific invariant
transcripts can, in theory, be used to compare the hybridization results
from one specimen to another.
Batch controls that consist of
standard specimens of positive and negative examples of the biomarker
pattern under investigation can provide invaluable information
concerning the performance of a test run. The results from these batch
controls can be compared to previous runs to monitor reproducibility
over time.
The entire system should be repeatedly tested to determine if the
process is routinely capable of performing to specifications. It is
essential to simulate test conditions used in actual specimen testing,
and repeat this analysis often enough to assure that results are
meaningful and consistent. Continual monitoring of hybridization data to
determine the normal performance ranges of variation can help to
identify and eliminate controllable causes. Replicate runs, especially
of batch controls, can be used to demonstrate that the output
consistently meets predetermined specifications. Monitoring of the
process during routine operation, using internal controls and
test-specific invariant transcripts, can also help to identify specimen
test failures.
- Performance Evaluations/Proficiency Testing
Several grass root standardization initiatives for microarrays are
currently in progress with the immediate goals of defining assay
capabilities, and understanding reproducibility both within laboratories
across time, or between different testing laboratories (3) This type of
analysis will evolve into standardized Proficiency Tests that will be
used to qualify testing facilities. (4)
Summary
As expression profiling begins
to play a more central role in drug development and diagnostics efforts,
it will be necessary to establish a laboratory infrastructure that is
capable of supporting this complex and sensitive test procedure. Some of
the features of the system have been described in this article and have
been implemented at microarray testing facilities. Establishing
processes that are compliant with the regulations of 21 CFR Part 58
(Good Laboratory Practices) in a microarray laboratory is a lengthy and
difficult process. For example, Expression Analysis' GLP initiative
required several years of effort at considerable expense, writing and
implementing SOPs along with a well-developed document control system
and extensive internal and external audits along with external
validation services. However, this level of laboratory commitment is
necessary in order to recognize the high complexity benefits of this
powerful clinical tool.
Thomas Goralski, PhD, is
laboratory director and Steve Casey is the founder and COO for
Expression Analysis Inc., a provider of regulatory compliant genomic
processing services. Steve Casey can be reached at scasey@expressionanalysis.com
References
1. S. McPhail, "Microarrays on
the Spot: Harnessing the Potential of Genomic Data." Pharmaceutical
Discovery: 5(1): 22-24 (2005).
2. http://www.fda.gov/cder/guidance/6400fnl.pdf
3.
External RNA Control Consortium Workshop: http://www.cstl.nist.gov/biotech/workshops/ERCC2003,
and The MAQC (Microarray Quality Control) Project: http://www.accessdata.fda.gov/scripts/oc/scienceforum/
sf2005/Search/preview.cfm?abstract_id=506&backto=author.
4. L.H. Reid, "Microarrays on
the Spot: The Value Of A Proficiency Testing Program To Monitor
Performance In Microarray Laboratories." Pharmaceutical
Discovery 5(3): 20-25 (2005).
|
