Medical Device Link.

Originally Published IVD Technology March 2004

Molecular Diagnostics

Figure 1. The workflow in developing a prototype gene expression–based diagnostic array (Phase 1) for testing in prospective clinical trials (Phase 2). Click to enlarge.

Molecular diagnostic tests typically analyze key protein, DNA, or RNA markers to characterize diseases. Immunohistochemistry, Western blotting, DNA sequencing, polymerase chain reaction (PCR), Southern blotting, Northern blotting, and quantitative reverse-transcription PCR (qRT-PCR) are molecular techniques that have been effectively used in clinical diagnostic laboratories, where broadly trained technologists perform the tests rather than researchers with specific methodological expertise. However, more-parallel molecular tests that involve simultaneous analysis of tens to hundreds of markers within a clinical sample will soon be required to allow maximum translation of postgenomic research to patient care. 

Many basic research projects have focused on methods to measure the behavior of the approximately 30,000 genes in the human genome. This research has been done at different levels within the gene, measuring both genotypic and phenotypic changes. For example, parallel analysis has been applied at the DNA level to analyze sequence polymorphisms that distinguish closely related human alleles or microbial genes; at the level of DNA methylation patterns; at the mRNA level to analyze gene expression patterns; and at the protein level. 

With the development of microarray-based tests for diagnosing a wide range of diseases, the ability to gain more information from a limited clinical sample by using highly parallel expression-analysis techniques has emerged. This technology’s greatest potential exists for those diseases in which histological differentiation between disease entities is difficult and the consequences of inappropriate treatment can be serious. In the research environment, analyzing sets of gene products with microarrays, rather than individual proteins or RNAs, has been shown to provide effective diagnoses. However, improving high-throughput diagnostic methods will require the continued development of new second-generation custom microarrays. This article describes the challenges in developing this technology and the custom microarray techniques that are currently available.

Microarrays as Transcription Analysis Tools

Figure 2. Subgroups of diffuse large B-cell lymphoma (DLBCL) that are identified using high-density microarrays. Researchers used cDNA miroarrays to analyze the expression of 100 selected genes (represented by rows) in 240 clinical DLBCL specimens (represented by columns).8  Click to enlarge.

High-density DNA (HD DNA) microarrays have been useful in the semi-quantitative, or comparative, analysis of thousands of individual messenger RNA (mRNA) species. Experiments using HD DNA microarrays suggest that new tests measuring the expression of tens to hundreds of genes will be better able to characterize disease states than is currently possible using histopathology and other existing molecular methods. As stated above, clinical application of this DNA microarray-based research will require further development of sophisticated new second-generation custom DNA microarray platforms that are more suitable for routine clinical laboratory testing (see Figure 1).

HD DNA microarrays are small devices that hold tens of thousands of gene capture probes at known positions on a solid surface. One of the most widely used platforms is the arrays by Affymetrix Inc. (Santa Clara, CA), in which DNA oligonucleotides are chemically synthesized in situ on the array surface using masked photolithography. Another widely used platform is the cDNA array, in which PCR products are generated separately and placed on the array surface using a robotic arrayer.^1,2 Thousands of laboratories have used these two platforms to catalog the expression of large numbers of genes in normal and diseased cells and tissues.

When transitioning from a normal to a diseased state, cells and tissues undergo a variety of histological and biochemical changes. These changes reflect in large part the dysregulation in the expression of genes. The quantities and types of proteins that are expressed from their respective genes determine the normal and abnormal functions of cells. When genes are expressed, their DNA sequences are first copied into temporary molecules, or mRNAs. These mRNAs are then translated into proteins that perform the functions and generate the structures that allow cells to conduct their biological activities. 

Researchers can take advantage of this flow of genetic information from DNA to mRNAs to proteins by purifying populations of all mRNAs expressed in specimens, synthesizing fluorescently labeled cDNA copies of the mRNAs in vitro, and hybridizing the labeled cDNAs to a microarray. The intensity of the fluorescent signal at each gene-specific probe on the microarray reflects the level of mRNA expressed from that gene in the specimen. By measuring the types and quantities of tens of thousands of individual mRNAs expressed in samples, researchers can infer information regarding the state of gene expression, the approximate quantities and types of proteins present, and the biological behavior of the sample. 

Figure 3. The synthesis of oligonucleotides for microarray experiments. Classical phosphoramidite synthesis relies on the acid catalyzed (red circle) removal of the DMT protecting group (blue circle) from an activated nucleotide, which is used to regulate the addition of the next nucleotide.  Click to enlarge.

For several decades, alternative methods have been available for identifying genes whose expression patterns are altered in disease. However, HD DNA microarrays have increased research scientists’ productivity in their gene- hunting efforts. As little as five years ago, identifying even a few differentially expressed genes may have taken several years and cost tens of thousands of dollars. Today, HD DNA microarrays can identify 10 times that number of genes in a few months and at a tenth of the cost.³

Advances in Lymphoma Diagnostics

Recent studies on the biology of non-Hodgkin’s lymphomas (NHL) illustrate some of the diagnostic advances made possible by HD DNA microarrays. In 2000, it is estimated that 54,900 patients were newly diagnosed with NHLs and 22,553 died of these diseases.⁴ Although there are many NHL subtypes, the diffuse large B-cell lymphoma (DLBCL) subtype is the most common, comprising 30% of all cases.⁵
 
Despite the complexity of this classification system, it is increasingly recognized that the NHL categories themselves encompass multiple disease subtypes with distinct molecular defects and clinical outcomes. For example, while 35–40% of patients diagnosed with DLBCL can be cured with chemotherapy, the remaining 60–65% die of this disease.⁶ Numerous attempts to distinguish patients who are likely to respond to chemotherapy from those unlikely to respond have been unsuccessful.⁷ In addition, these DLBCL tumors are indistinguishable using classical histopathology and currently available molecular markers.⁷

Researchers have approached this problem by using HD cDNA microarrays to profile gene expressions in DLBCL.⁸ These researchers identified genes that classified DLBCL specimens into distinct subclasses (see Figure 2). One subclass, germinal-center B-cell-like DLBCL, showed an expression pattern similar to mature germinal-center B-cells and portended a more favorable prognosis with a 60% five-year survival rate. On the contrary, two other DLBCL subclasses, activated B-cell-like DLBCL and type-3 DLBCL, showed much less favorable prognoses with a 35% five-year survival rate.⁸

Using Microarrays for Personalized Medicine

Research studies have predicted that in the near future, it may be possible to determine which patients might benefit from more-aggressive therapies, such as bone marrow transplants or alternative chemotherapy regimens, when DLBCL is diagnosed. HD DNA microarray studies have also highlighted molecular pathways important in subclasses of cancer. For example, researchers have observed that several genes in the activated B-cell subtype of DLBCL were downstream targets of the NFkB transcription factor.⁹ These investigators demonstrated that NFkB activity is indeed higher in this DLBCL subtype, suggesting that drugs specifically targeting the NFkB pathway may be effective in treating these tumors.⁹

Several recent papers have suggested that gene expression profiling of tumor specimens may be useful for preselecting patients who may benefit from drug treatment.^10-13 It may also be useful to examine gene expression profiles of cancers following chemotherapy in order to determine whether the tumors are responding to treatment.¹⁴ The goal of this approach is personalized medicine, in which detailed patient-specific molecular information would be assayed to predict an effective therapy.
 

Microarrays as Genotyping Tools for Infectious Diseases

Despite the abundance of protein diagnostics, diagnosing diseases by examining the qualitative presence of specific nucleic acid sequences has been adopted for infectious diseases. While the amplification of gene sequences through PCR or other similar methods can indicate the presence or absence of a pathogen genomic sequence, nucleic acid hybridization is almost always necessary to confirm that the amplification is sequence-specific. Suitably designed microarrays are versatile tools for this function since the hybridization data from an array can identify the pathogen genomic sequence in a more rapid and cost-effective manner than DNA sequencing.

The sudden emergence of new strains of pathogens has brought the need for rapid nucleic acid diagnostic tools to the forefront. Although customizable nucleic acid testing systems for the rapid development of diagnostics are still new and emerging, the urgent medical need for high-speed diagnostics is steadily increasing. For example, the recent outbreak of severe acute respiratory syndrome (SARS) and its rapid spread left no time for generating an antibody-based diagnostic test. Initial identification of the virus was based on symptom presentation and high-resolution electron microscopy of sputum samples. However, gene sequencing of the emergent SARS strains permitted rapid identification of the virus by using a nucleic acid amplification and hybridization/sequencing approach. 

Monitoring how the genotype of viruses evolves is critical for understanding the pathology of a disease as it spreads throughout a population. Viruses such as influenza, poliovirus, rhinoviruses, and HIV are known to have high rates of mutation, in some cases even causing attenuated strains that are developed for vaccines to revert to viruses with a wild-type phenotype. Although it remains possible to generate antibodies for all of these pathogen strains, identifying strains rapidly is more easily achieved using microarray tools developed specifically for this purpose.

Second-Generation Custom DNA Microarrays 

Figure 4. The CustomArray by CombiMatrix Corp. (Mukilteo, WA) is a super-parallel DNA synthesizer system controlled by semiconductor electrochemistry.  Click to enlarge.

Many researchers agree that in some form, DNA microarrays will be used as diagnostic tools in the near future.^{15, 16} The first prospective trials are under way to determine whether microarrays can predict the proper clinical course for patients.¹⁷ At the same time, FDA is anticipating working on the regulatory aspects of microarray-based diagnostic tests.¹⁸ However, extensive work will be necessary to develop microarray platforms that give reproducible results in different laboratories, which will be required for clinical applications.¹⁹ In addition, the basic parameters of microarray technology will also need to be established, including content selection (i.e., which oligonucleotides will be used), data analysis, data interpretation, and uniform sample preparation and assay conditions.

The use of a variety of generic whole-genome HD DNA microarrays in the research setting has led to important new basic research findings. However, these platforms are not suitable for use in clinical diagnostic laboratories for a number of reasons. 

First, ownership claims for probes to particular gene sequences can burden the expensive diagnostic development process before it has even begun. Second, the lack of sequence flexibility of the nucleic acid capture probes on the microarrays makes it difficult to be sure that the best capture probe has been identified. In other words, there is no chance to improve rapidly and iteratively the performance of the capture probe to its target. Third, many genes have significant polymorphisms or splice variants, which may or may not have been correctly represented on the array. The content arrays are based on the human genomic information from a comparatively small set of human genomic sequences. Finally, nearly all of these products are built around gene expression type applications, which are not necessarily the only approach for developing diagnostic tools.

In contrast, customizable microarrays give diagnostics developers complete freedom to identify rapid nucleic acid probes for specific applications, such as key changes in gene-expression approach, disease-related polymorphisms, patterns in DNA methylation analysis, or resequencing applications for infectious diseases. Although a number of technologies exist for making custom arrays, there are significant differences in the techniques used to fabricate the arrays, which have an impact on the length of time, cost, and complexity in developing nucleic acid diagnostics (see Figure 3).

For example, the Affymetrix system uses specially made chromium photomasks from the semiconductor industry to pattern the synthesis of nucleic acids. Although this method has been widely adopted in research scenarios, an important limiting factor concerns poor stepwise yields of light-based deprotection. Typically, ^20-25 nucleotides are used for light-based deprotection systems. However, while short oligonucleotides may work for some applications, they may not be ideal for developing many nucleic acid diagnostic tools.

Similarly, spotted cDNA arrays encounter problems with spot nonuniformity and misidentification of spots.²⁰ These problems limit the quality of data derived from high-density microarrays, making it difficult to distinguish experimental noise from biologically meaningful measurements of gene expression. For example, a significant percentage of genes that are identified as being differentially expressed from specimen types using high-density microarrays cannot be confirmed using the more-laborious gold standard technology of qRT-PCR. This is particularly true for genes that are expressed at low levels.²¹

In addition, several recent papers have described similar experiments that were performed on both the Affymetrix and spotted cDNA platforms and produced different lists of differentially expressed genes.²² If the gene expression patterns were genuine, researchers should be able to identify the same genes regardless of which platform is used. This noncorrelation of data across platforms suggests that care should be taken to understand the technical limitations of these platforms.

An alternative technology that addresses the key issues of turnaround time, interplatform reproducibility, and oligonucleotide quality is the CustomArray, a semiconductor-based technology by CombiMatrix Corp. (Mukilteo, WA).²³ This technology uses the electrochemical generation of acid at microelectrodes, rather than light-based photochemistry, to pattern nucleic acid synthesis. With a 16–36-hour turnaround time, this system can synthesize oligonucleotides of high quality and long lengths on an array format (see Figure 4).²⁴
 
This technology enables testing a wide range of diagnostic approaches, including screening genotypic information from both virus and host cells concurrently, using multiple assay types on one chip, and employing variable oligonucleotide lengths and non standard chemistries.²⁵ This technology also benefits from employing established manufacturing processes that have been developed in the semiconductor industry. This is an important consideration when determining how a diagnostic device can be manufactured in large volumes at a reasonable cost.

Conclusion

Andy McShea, PhD, is director of microarray technology; Ali Arjormand, PhD, is director of business development; and Amit Kumar, PhD, is president and chief executive officer at CombiMatrix Corp. (Mukilteo, WA). They can be reached here, here, and here, respectively. Stephen Schmechel, MD, PhD, is a professor of pathology at the University of Washington (Seattle); he can be reached here.

An ideal custom DNA microarray platform should allow a user complete freedom in defining which genes are assayed using the array. The microarray platform should also be compatible with computer-based probe design software and molecular methods so that measurements can be performed with high sensitivity and specificity, thereby making independent validation of the assay unnecessary. 

The use of manufacturing techniques that allow miniaturization and massive scale-up and result in low costs and minimal biological sample requirements should contribute to a well-engineered microarray platform.²⁶ Because of these factors, there should be standardization among various clinical laboratories using microarrays, such that results from different laboratories can be directly compared. This is particularly important under existing and proposed regulations that give the Centers for Disease Control and Prevention (Atlanta) and FDA shared responsibility for monitoring molecular diagnostic testing.²⁷

References

1. RJ Lipshutz et al., “High Density Synthetic Oligonucleotide Arrays,” Nature Genetics 21 (1999): 20–24.

2. M Schena et al., “Parallel Human Genome Analysis: Microarray-Based Expression Monitoring of 1000 Genes,” Proceedings of the National Academy of Sciences 93 (1996): 10614–10619. 

3. I Haviv and IG Campbell, “DNA Microarrays for Assessing Ovarian Cancer Gene Expression,” Molecular and Cellular Endocrinology 191 (2002): 121–126. 

4. LAG Ries et al., (eds). SEER Cancer Statistics Review, 1975–2000, (Bethesda, MD: National Cancer Institute, 2003; [accessed 15 December 2003]) available from Internet: http://seer.cancer.gov/csr/1975_2000.

5. NL Harris, “Mature B-Cell Neoplasms: Introduction,” in Pathology and Genetics of Tumours of Haematopoietic and Lymphoid Tissues, ES Jaffe, NL Harris, H Stein, and JW Vardiman (eds). (Lyon, France: IARC Press, 2001): 121–126. 

6. B Coiffier, “Diffuse Large Cell Lymphoma,” Current Opinions in Oncology 13 (2001): 325–334. 
7. AA Alizadeh et al., “Distinct Types of Diffuse Large B-Cell Lymphoma Identified by Gene Expression Profiling,” Nature 403 (2000): 503–511. 

8. A Rosenwald et al., “The Use of Molecular Profiling to Predict Survival After Chemotherapy for Diffuse Large B-Cell Lymphoma,” New England Journal of Medicine 346 (2002): 1937–1947. 

9. RE Davis et al., “Constitutive Nuclear Factor KappaB Activity Is Required for Survival of Activated B-Cell-Like Diffuse Large B-Cell Lymphoma Cells,” Journal of Experimental Medicine 194 (2001):1861–1874. 

10. M Sakamoto et al., “Analysis of Gene Expression Profiles Associated with Cisplatin Resistance in Human Ovarian Cancer Cell Lines and Tissues Using cDNA Microarray,” Human Cell 14 (2001): 305–315.

11. KE Luker et al., “Overexpression of IRF9 Confers Resistance to Antimicrotubule Agents in Breast Cancer Cells,” Cancer Research 61 (2001): 6540–6547. 

12. U Certa et al., “High Density Oligonucleotide Array Analysis of Interferon-Alpha2a Sensitivity and Transcriptional Response in Melanoma Cells,” British Journal of Cancer 85 (2001): 107–114. 

13. JC Chang et al., “Gene Expression Profiling for the Prediction of Therapeutic Response to Docetaxel in Patients with Breast Cancer,” Lancet 362 (2003): 362–369. 

14. TA Buchholz et al., “Global Gene Expression Changes During Neoadjuvant Chemotherapy for Human Breast Cancer,” Cancer Journal 8 (2002): 461–468. 

15. L Pusztai et al., “Clinical Application of cDNA Microarrays in Oncology,” Oncologist 8 (2003): 252–258. 

16. H Brown, “The Real Value of Microarray Technology,” The Lancet Oncology 4 (2003): 326. 

17. M Branca, “Genetics and Medicine: Putting Gene Arrays to the Test,” Science 300 (2003): 238.

18. EF Petricoin et al., “Medical Applications of Microarray Technologies: A Regulatory Science Perspective,” Nature Genetics 32 Supplement (2002): 474–479.

19. B Vastag, “Gene Chips Inch Toward the Clinic,” Journal of the American Medical Association 289 (2003): 155–156,159. 

20. E Taylor et al., “Sequence Verification as Quality-Control Step for Production of cDNA Microarrays,” Biotechniques 31 (2001): 62–65. 
21. MS Rajeevan et al., “Use of Real-Time Quantitative PCR to Validate the Results of cDNA Array and Differential Display PCR Technologies,” Methods 25 (2001): 443–451. 

22. WP Kuo et al., “Analysis of Matched mRNA Measurements from Two Different Microarray Technologies,” Bioinformatics 18 (2002): 405–412. 

23. K Dill et al., “Antigen Detection Using Microelectrode Array Microchips,” Analytica Chimica Acta 444 (2001): 69–78.

24. A McShea, “A Novel Semiconductor-Based Microarray Platform for Rapid Development and Validation of Gene Expression Diagnostic Assays,” (paper presented at Macroresults for Microarrays Conference, Boston, May 13–14, 2003).

25. A McShea et al., “Semiconductor-Based Technology for the Generation of Oligonucleotide Arrays, siRNA Pools for Knockdown Analysis and Gene Fragments for Protein Function Studies” (paper presented at the Annual Northwest Gene Expression Conference, Seattle, August 27–29, 2003).

26. MS Talary, JP Burt, and R Pethig, “Future Trends in Diagnosis Using Laboratory-on-a-Chip Technologies,” Parasitology 117 Supplement (1998): S191–S203.

27. Secretary’s Advisory Committee on Genetic Testing (SAGT) Web site (Bethesda, MD: SAGT, 2003 [accessed 15 December 2003]); available from Internet: 
www4.od.nih.gov/oba/sacgt/gtdocuments.html.    
  

Copyright ©2004 IVD Technology