BBI Contributing Editor

SAN DIEGO, California The market for molecular diagnostic products for clinical use has expanded rapidly over the past decade, as new technologies have enabled tests to be developed that provide fundamentally new diagnostic capabilities. As shown in Table 1, the global market for clinical molecular diagnostic tests reached almost $885 million in 2001 and will exceed $1 billion in 2002, and there is an additional market of several hundred million attributable to home-brew products used by clinical labs to perform tests using methods developed in-house.

The total market at the provider level for molecular testing services offered by hospital and reference labs is estimated at $2.5 billion to $3.1 billion in the U.S., according to Thomas White, PhD, of Celera Diagnostics (Foster City, California). As discussed by White in a presentation at the annual San Diego Conference sponsored here by the American Association for Clinical Chemistry (AACC; Washington), tests for various infectious agents performed for patient diagnosis and screening as well as for screening of donated blood account for at least 95% of test volume at present. Genetic testing and cancer-related tests comprise the remainder of the market. Tests for only nine infectious agents account for essentially all of the testing performed with FDA-cleared kits, and seven companies supply most of the molecular in vitro diagnostics products used in U.S. clinical labs. Continued advances in genomics as well as in molecular test automation are expected to change the market's make-up in the future, and a number of additional companies are expected to become important players in the market. Low-cost automation of sample preparation techniques, along with the advent of homogeneous detection methods, will allow molecular testing to be adopted by a wider range of labs.

Another new aspect of molecular diagnostics, although not one that is likely to emerge as part of routine clinical testing for a few years, is haplotype analysis. Haplotype analysis is not yet feasible for clinical applications because a map of human haplotypes has not yet been generated by researchers. A consortium that includes the National Institutes of Health (NIH; Gaithersburg, Maryland) and a number of companies has recently initiated a large-scale effort to create a human haplotype map, but it will require at least three years at the existing rate of analysis to complete. Haplotype analysis and functional genomics will allow clinicians to redefine diseases in molecular terms and, as demonstrated by some applications that have already been developed, will enable the development of targeted diagnostics and associated therapies that are more effective and create fewer side effects.

Cancer screening and diagnosis is one of the applications of molecular diagnostics attracting considerable investment. The number of potential patients for cancer testing, including both diagnostic and screening tests, is estimated at 300 million annually worldwide, as shown in Table 2. More powerful tests for cancer detection and diagnosis will be enabled by the development of advanced analytical techniques described at the conference. Those technologies include new types of assays for single nucleotide polymorphism (SNP) analysis, proteomic assays and some emerging technologies that may allow whole genome sequencing for individual patients.

Focusing on chronic disease

Most existing clinical molecular diagnostic tests use technologies that are suited for detection of one or two target gene sequences, a capability that is desirable for infectious disease assays and certain cancer-related tests. However, newer technologies allow multiple genes to be detected in a single test, making the diagnosis of diseases with multi-gene defects possible. That capability will allow many chronic diseases such as diabetes, cardiovascular diseases and most cancers to be addressed by molecular diagnostic techniques. Recent developments such as the widespread adoption of viral load testing, which now represents a worldwide market at the supplier level of more than $200 million, demonstrate that monitoring of disease progression can be as important as initial diagnosis. Some multiple target molecular tests already have been launched, such as a microarray device now available from Nanogen (San Diego, California) for performing cystic fibrosis screening using a 25-mutation panel that is supplied as an analyte-specific reagent.

Although the analytical technology that will allow such multi-gene testing to be performed is advancing rapidly, the genomic data needed to make it useful for clinical diagnostics is only beginning to be elucidated. So far, very few functional SNPs have been identified, for example. The NIH haplotype mapping project is the primary program aimed at creating the required database. The project involves re-sequencing of 39 people who were originally sequenced as part of the Human Genome Project, along with one chimpanzee. The goal is to identify SNPs that impact the stability of proteins. So far, 23,400 genes have been sequenced, containing 200,000 SNPs. Some 70% of the SNPs are novel compared to all public databases. And 15% cause amino acid substitutions that impact protein function. By extrapolation, the researchers involved with the project estimate that 40,000 novel functional SNPs exist. The goal of the project is to perform the studies on a relatively large scale, using 500 cases and 500 controls from a given population group, and to also analyze two to three other populations in order to assess replication of the markers across various populations. In addition to Celera, companies involved in the project include Abbott Laboratories (Abbott Park, Illinois) and Applied Biosystems (Foster City, California), a subsidiary of Applera (Norwalk, Connecticut). The study is using kinetic RT-PCR in combination with microarrays to perform the analysis.

Sequenom (San Diego, California) is pursuing an alternative tactic for creating a clinically useful genomic map using its MassExtend technology, which is based on mass spectrometry analysis and biochip arrays (SpectroCHIPs). As discussed by Michael Shi, PhD, senior director for genomics at the company, a brute force approach to developing a genome-wide map of clinically important SNPs would require analysis of 10 million SNPs for each of 1,000 patients and controls, or 20 billion genotypes, at a cost of at least $2 billion, making such an approach impractical using existing technologies. However, an alternative strategy is to pool DNAs from multiple individuals and analyze the pool rather than analyzing each subject's DNA. Sequenom already has created an SNP map using that strategy that is composed of 25,000 SNP markers. A prototype assay has been developed for cholesterol ester transfer protein (CETP) genotype using the Sequenom assay technology. CETP genotype exhibits a direct relationship to HDL cholesterol levels, and a correlation also has been demonstrated between CETP genotype and the degree of coronary artery narrowing over two years for patients undergoing treatment with Pravastatin, a compound manufactured by Bristol-Myers Squibb (New York). Sequenom has found a marked difference in response to Pravastatin treatment between patients with different CETP genotypes.

Another prototype Sequenom assay for ras/BRAF mutations in melanoma employs testing for three somatic mutations that are correlated with development of the disease. Other mutations in the BRAF and k-ras genes have been found to be associated with the development of colon cancer. The Sequenom method shows promise for use in developing clinically useful assays for gene abnormalities, including applications in cancer testing and pharmacogenomics.

A human SNP-based linkage map has been developed by the SNP Consortium, a non-profit group sponsored by 11 major pharmaceutical, medical and technology companies working with the Wellcome Trust (London). The chairman of the SNP Consortium is Arthur Holden, also chairman and CEO of First Genetic Trust (Deerfield, Illinois). As described by Tara Matisse, PhD, of Rutgers University (New Brunswick, New Jersey), at the San Diego Conference, one of the key researchers involved in the SNP Consortium, the consortium's SNP map provides an alternative to currently available STR (short tandem repeat) maps that offers faster throughput (six weeks vs. six months) when performing whole genome linkage studies, at least two-fold higher resolution, and lower genotyping cost (10 cents to 15 cents vs. 65 cents to $1). A total of 6,297 SNPs were genotyped in 90 individuals. Most of the SNPs were genotyped by Celera Genomics, but 851 were genotyped by Motorola Life Sciences, now Amersham Life Sciences (Phoenix, Arizona), using the CodeLink SNP platform. The final SNP Consortium map shows 99% correlation in terms of the order of the identified markers with the map created as part of the Human Genome Project, and has the potential for use in genetic studies.

Maps such as those developed by Sequenom and the SNP Consortium are expected to have a number of applications in genomics research, and eventually in developing new types of clinical tests for a wide range of diseases, including many cancers, cardiovascular diseases, neurological diseases, diabetes, and a wide range of other multigenic disorders. Because of the complexity of such diseases, however, labs will also need to adopt new types of informatics technology in order to analyze test results as they apply to individual patients. As discussed by David Wang, executive vice president of strategy, technology and operations of First Genetic Trust, the fields of informatics and genomics are now converging to create the analytical capabilities that will be needed, while also addressing the issues of patient privacy and medical ethics that must be successfully resolved if genomics research is to be translated into personalized medicine. First Genetic Trust has developed a web-based informatics platform over the past three years that provides electronic informed consent using electronic signatures, on-line counseling and disease education, an electronic patient record, and result reporting, as well as providing clinical follow-up including electronic adverse event tracking and reporting. About 250 centers in the U.S. and Europe act as data controllers, and the network now covers between 2 million and 3 million patients, including thousands in most of the main categories of major chronic diseases. First Genetic Trust is developing a cancer genetics service that will be available on its network, and has also developed data mining tools that can be used by its pharmaceutical and diagnostic testing partners to help develop new drugs and associated tests.

Piet de Groen, associate professor of medicine at the Mayo Clinic (Rochester, Minnesota), described a computational database that has been constructed at that institution in collaboration with IBM (White Plains, New York), which also will have applications in clinical genomics. As discussed by de Groen, the increasing amount of data per patient that will be generated by molecular genetic tests, coupled with the increasing specialization characterizing medical practice, requires the use of computational medicine to allow clinical decisions to be made accurately and efficiently. While the existing database has focused on collecting phenotype data, de Groen's team is now building a genomic data section, plus microarray-based workflow automation. The database also provides data mining features, and allows identification of patients with appropriate characteristics to serve as participants in disease association studies. The key to constructing a useful data mining system is to acquire data for a large number of patients. The Mayo Clinic has more than 6 million patient records, 4 million of which are in electronic form, with lab data dating to 1993. About 400,000 patients are treated by the Mayo Clinic system each year, so the database is continuing to grow. While the database is only available to researchers who are physically on site at the Mayo Clinic at present, the institution is assessing ways in which the database can be made available to outside researchers.

Advances in sequencing, SNP detection

A number of new analytical technologies were described at the San Diego Conference that could have a major impact on the development of clinically useful molecular diagnostic tests. Table 3 describes new technologies now under development or available in the research market for rapid nucleic acid sequencing. Sequencing technology is used at present for genomic studies aimed at discovering new disease associations and drug targets, but commercially available sequencing technologies do not have sufficient throughput and are too costly for routine clinical use. A new sequencing technology under development by Dr. Daniel Branton at Harvard University (Cambridge, Massachusetts) uses nanopores formed by incorporating alpha-hemolysin into a membrane. When a voltage is applied across the membrane, the nucleic acid molecules migrate single-file through the nanopore with a transit time of 1.3 milliseconds. Molecules of up to 35 kilobases in length can be analyzed successfully and the direction of travel can be detected. In addition, a technique called zinc finger protein labeling is being evaluated as a possible method for haplotype and SNP detection in nanopores. At present, single-base detection, which is required for direct sequence analysis, has not been demonstrated. The existing device integrates the signal from 12 to 15 bases at a time. However, a new version of the device, employing nanofabricated pores in silicon nitride, shows promise for reducing noise and capacitance levels, and thereby improving resolution. In addition, use of a slower transit time will improve resolution. Another detection approach now under evaluation is injection of electrons by a tunneling probe. Use of highly localized electric and magnetic fields may also allow individual bases to be interrogated.

Solexa Ltd. (Cambridge, UK), a venture capital-funded start-up that has received $17 million in funding to date, is developing a whole genome sequencing method that promises to allow sequence data to be obtained at a fraction of the cost and time required for conventional sequencing techniques. Solexa has developed a Single Molecular Array chip that contains a random distribution of up to one billion different genomic DNA fragments of 20-30 bases in length on a 10 cm2 surface. Four nucleotides labeled with different removable fluorophores are added, and the complementary base is incorporated via the action of a polymerase. The nucleotides also act as reversible terminators of the polymerization reaction, allowing determination of the incorporated base via a fluorescence imaging measurement. Four fluorescence images are collected at four different wavelengths from each target cell on the chip in each polymerization cycle. Continuation of the cycling polymerization reaction allows determination of the sequence of the fragment bound at each location on the chip, and analysis of all chip locations in parallel allows the sequence of the entire genome to be determined, using the human genome map as a reference to allow the individual sequences of the fragments to be aligned in the proper order. Development is at an early stage, although the company has demonstrated that single molecules of DNA can be detected and analyzed, and that detection of SNPs at levels well above 95% will be possible. The cost to perform a complete genome sequence using the Solexa technology is estimated at between $1,000 and $10,000.

A number of other nucleic acid diagnostic assay techniques, including new techniques for genotyping and SNP analysis, also were described at the San Diego Conference. EraGen Biosciences (Madison, Wisconsin) is developing Aegis, a new system for quantitative real-time nucleic acid diagnostics. Aegis employs novel synthetic bases that only pair with each other, and not with natural bases. The technology has been commercialized for clinical applications by Bayer Diagnostics (Tarrytown, New York) in the latest version of its Versant HIV-1 RNA viral load assay, and has allowed development of an assay with improved sensitivity that can detect three-fold or greater changes in viral load throughout its entire quantification range. Other applications exist in genotyping and infectious disease testing.

Another new approach was described by Lloyd Smith, PhD, of Third Wave Technologies (Madison, Wisconsin), that uses the company's Invader technology in a microarray format. The technique, called Surface Invader Technology, has the potential to sequence an entire human genome in about two hours by only analyzing SNPs, which by definition represent the differences from the standard human genome. It is based on a modified version of Third Wave's existing Invader platform, a solution-phase methodology that is already in widespread use in research labs.

In the assay, an Invader oligo and a probe are both attached to the same cell of an array on a gold film, and target DNA and other required reagents are then added. The assay uses a flap endonuclease that, in combination with sequence-specific hybridization, provides a high degree of binding specificity, and catalyzes the cleavage of a signal probe to provide signal amplification. At present, the sensitivity of the assay is not sufficient to allow testing to be performed with the amounts of DNA that typically are available in a clinical blood sample. However, Third Wave is investigating a number of approaches, including use of the Rolling Circle Amplification technology from Molecular Staging (New Haven, Connecticut), as well as switching from gold to diamond films in the surface array to improve sensitivity. Third Wave reported pro forma revenues of $24.7 million for the first nine months of calendar 2002, a decline from $28.1 million in the like period of 2001, as a result of a strategic decision to emphasize clinically focused products with improved profit margins. The company secured its 90th clinical reference laboratory customer during the third quarter of 2002, and revenues from clinical products continue to grow.

Genohm (San Diego, California), a company that was founded by Jacqueline Barton, PhD, of the California Institute of Technology (Pasadena, California), is commercializing a new electronic nucleic acid sensor technology. The technology makes use of the conductivity properties of nucleic acids, which depend on the proper pairing of bases in the double-stranded molecule. A single base mismatch is sufficient to stop electron transfer though the molecule, allowing such events to be detected with a high degree of sensitivity. Cyclic voltammetry is combined with electrocatalysis to measure conductivity. At present, about 100 million DNA molecules are needed in order to obtain a detectable signal, but Barton said she believes that can be improved.

Other new developments in nucleic acid diagnostics described at the San Diego Conference include a new genotype assay for the genetic determination of osteoporosis, the COL1A1 assay developed by Axis-Shield (Dundee, Scotland); a cystic fibrosis assay capable of detecting the 25 most prevalent mutations in that disease that uses the eSensor technology developed by Motorola Life Sciences (Pasadena, California); and a fully automated microarray platform for genomic testing (the Infiniti system) from AutoGenomics (Carlsbad, California), with prototype assays for cystic fibrosis (detecting 31 mutations) and the Connexin 26 mutation associated with deafness in children. Another technology, the RiboMaker Detection System, was described by Designer Genes (Phoenix, Arizona), which is an isothermal, PCR-free, gel-free process to produce multiple signal molecules per target. It is not a target amplification system, nor is it a true signal amplification system. Although the technology is at an early stage of development, it can detect a single 1,000 nucleotide DNA or RNA target in three-hour reaction without PCR. The technique also can be used to analyze proteins, and to determine the methylation status of DNA. Another new optical detection method applicable to nucleic acid diagnostics, the MagiProbe, was described by Akio Yamane of Wakunaga Pharmaceutical Co. Ltd. (Hiroshima, Japan). The MagiProbe is a novel fluorescence quenching probe that only emits light when the base to which it is attached hybridizes to a complementary base. The probe has the ability to allow direct detection of single base mismatches.

Important advances in specimen preparation for molecular diagnostics were described by Rajiv Raja, PhD, of Arcturus (Mountain View, California). Arcturus has developed the PixCell II Laser Capture Microdissection (LCM) System used in preparing tissue specimens for molecular analysis. The system allows an operator to visually select target cells for analysis and isolate them using laser dissection. Importantly, studies have shown that results from analysis of breast biopsy specimens collected by LCM can give completely opposite diagnostic information vs. specimens collected by conventional biopsy techniques. Those findings are interpreted as showing that the heterogeneous nature of most biopsy specimens can result in very different test results depending on which cells are collected and analyzed. LCM allows the pathologist to identify suspicious cells and isolate only those cells for subsequent molecular analysis, thus avoiding dilution with normal cells or degraded tissue that otherwise would occur in the absence of cell selection.

Arcturus also has introduced the RiboAmp process that can be used to amplify small amounts of RNA prepared from LCM samples, increasing the amount of RNA available for analysis by up to one million fold. A system has been developed that combines RiboAmp with LCM to create a microgenomic analysis system. A study using the system to analyze breast cancer biopsy specimens revealed that well-defined clusters of gene expression could be identified that are correlated with different types of breast cancer. Another application is profiling of oocytes vs. embryos to assess early stages of development. Arcturus has identified 758 genes that can be used to delineate stages of embryonic development.

Another unique approach to sample preparation for molecular diagnostics, the CloneSaver Card, has been developed by Whatman (Newton, Massachusetts). The card uses FTA Technology, a technology that uses Whatman's proprietary filter matrix impregnated with a chemical formulation that performs essentially all of the preparative steps needed for molecular analysis in less than 20 minutes by simply applying a blood specimen. The device performs cell lysis, protein denaturation, entrapment of nucleic acids, and inactivation of enzymes and viruses, including those from infected patients. Nucleic acids from the sample are preserved for up to three years at room temperature. PCR amplification can be performed directly from a punch taken from the card. While the technique is only applicable to genomic DNA in cells, it allows a standardized approach to specimen processing, and the card format provides savings in storage space. The FTA matrix also can be supplied in a 96 well format, allowing automation using a punch device configured for 96-well plates.

Improving early disease detection

Molecular diagnostics has proven particularly useful in allowing disease to be detected with greater sensitivity, a capability that translates to earlier detection of diseases such as cancer and infectious disease. Molecular testing already has proven valuable in improving the ability to detect cervical cancer more reliably than conventional PAP smear analysis by enabling certain strains of human papilloma virus (HPV) to be analyzed in PAP specimens. A test marketed by the Vysis (Downers Grove, Illinois) unit of Abbott Laboratories that uses in situ hybridization technology to detect amplification of the her-2/neu oncogene also has proven valuable for improving the ability to identify those patients with breast cancer who will benefit from treatment with Herceptin, a targeted drug manufactured by Genentech (South San Francisco, California). The test is typically used in conjunction with an immunohistochemical assay that detects expression of the her-2/neu protein product in biopsy specimens. A total of 205 clinical labs in the U.S. are now using the Vysis PathVysion FISH her-2/neu assay, and the number of labs using FISH for her-2/neu testing almost doubled (from 35 to 63, based on data from proficiency testing surveys conducted by the College of American Pathologists Chicago, Illinois) between 2000 and 2001.

New applications of molecular testing in cancer diagnosis continue to emerge. A molecular assay using pyrosequencing technology from Pyrosequencing AB for detection of k-ras oncogene mutations in pancreatic cancer patients was described at the San Diego Conference by Manjula Gupta of the Cleveland Clinic Foundation (Cleveland, Ohio). The test allows determination of malignant potential of a pancreatic tumor prior to surgery, aiding in planning of treatment. Exact Sciences (Maynard, Massachusetts) reported on advances in colorectal cancer detection using its Oligonucleotide Tiling Assay (OTA). The OTA assay measures cleavage of target DNA from either peripheral blood cells or colonic tissues that occurs when mismatches prevent binding to an array of oligonucleotide probes specific for the Mutation Cluster Region of the APC gene, one of the key genes known to be involved in the development of colorectal cancer. Exact Sciences also described the use of DNA integrity as a market for detection of colorectal cancer. The assay is based on the theory that tumor cells undergo less apoptosis (cell death) than normal cells, and therefore their DNA has higher molecular weight. A model assay indicated a sensitivity of 56% and a specificity of 97% for detection of colorectal cancer. Exact Sciences, in collaboration with the Mayo Clinic, is focused on the development of molecular tests to improve the early detection of colorectal cancer.

Another application of molecular diagnostics in cancer management is gene expression profiling of multiple myeloma. Multiple myeloma is a disease that is likely to increase in prevalence because of its association with aging. The disease arises from a benign dyscrasia that afflicts about 20% of all individuals over the age of 60. About 2% of individuals with benign dyscrasia convert to multiple myeloma each year. Survival varies from two to more than 80 months, and only about 20% of the variation in survival is explained by current lab parameters such as cytogenetics, LDH levels, beta-2-microglobulin levels and other established markers.

A program headed by John Shaughnessy Jr., PhD, of the University of Arkansas Medical Center (Little Rock, Arkansas), is evaluating the use of gene expression profiling as an alternative method for predicting outcome in multiple myeloma patients. A set of 14 genes has been identified that differentiates the benign condition from multiple myeloma. The genes can be analyzed to stratify patients with the disease to predict response to chemotherapy. Shaughnessy has found a subgroup of proliferation genes that are strongly associated with response, and has identified gene expression subgroups that are linked to distinct developmental stages of the disease. Future studies will assess the potential to replace the battery of biochemical tests that are now used to monitor disease progression with gene expression profiling.