May 1, 2007 (Vol. 27, No. 9)
Kate Marusina Ph.D.
Technology Is Essential to Personalized Medicine
Molecular diagnostics uses molecular signatures for disease identification, monitoring treatment effectiveness and even predicting a patient’s response to a new treatment. Although adoption of this technology in the clinical laboratory setting has been slower than anticipated, novel tools continue to create vastly superior alternatives to the established methods. Molecular diagnostics has gained more significance as medicine shifts toward targeted or personalized treatment methods.
Select Biosciences’ “Molecular Diagnostics World Congress”, which was held in Philadelphia recently, provided opportunities for companies in the field to exchange information and design new strategies.
“The tipping point in the field of molecular diagnostics will occur once pharmaceutical companies fully realize the value of integrating drug development with companion diagnostics,” said Eddie Blair, M.D., Ph.D., managing director, Integrated Medicines (www.integratedmedicines.co.uk). “Companion diagnostics could not only expedite the development of drugs but could also decrease the attrition rate. Stratification of the population based on predictive biomarkers could increase efficacy and safety of drugs developed for target populations. Physicians could thus make more informed decisions. Companion diagnostics could also increase the commercial value of drugs, and possibly extend the life of some patents.”
A well-known case of such co-development is Herceptin (trastuzumab) and the HER-2 test. Overexpression of the HER-2 receptor is linked to decreased patient survival. Herceptin was approved in 1998 to treat metastatic breast cancer in patients who overexpress HER-2 protein. Its FDA approval was dependent on the concomitant approval of a diagnostic immunohistochemistry (IHC) assay (now known as Dako HercepTest®).
Another example of a companion program is LpPLA2 (lipoprotein-associated phospholipase A2) immunoassay, developed by diaDexus(www.diadexus.com) and approved for predicting the risk of heart disease and ischemic stroke. GlaxoSmithKline(www.gsk.com) is developing a small molecule designed to inhibit this enzyme, thus reducing the risk of adverse cardiovascular events.
“Our calculations demonstrate that the net present value of a drug with a companion test nearly tripled in comparison with a stand-alone drug,” continued Dr. Blair, “but the established mentality does not change overnight. However, additional pressure from the regulatory agencies is aiding the paradigm shift.”
In 2005, the FDA issued guidance encouraging drug and biologic developers to conduct pharmacogenomic tests during drug development. FDA’s Critical Path Initiative further emphasizes development of predictive diagnostics.
“The immediate goals of Integrated Medicines include facilitating more widespread use of approved diagnostic tests in the promotion and marketing of medicines,” concluded Dr. Blair. “We also promote exclusive co-development of diagnostic tests with new medicines. We may be a little ahead of the curve in this respect.”
Barcoding Single DNA Molecules
Rapid and unequivocal identification of bacterial pathogens has been an elusive goal for both biodefense and healthcare. U.S. Genomics (www.usgenomics.com) circumvents many limitations of existing methods by analyzing individual DNA molecules in clinical samples. The basis for DirectLinear™ Analysis is barcoding of DNA strands with fluorescent tags. These tags are derived from either bisPNAs (peptide-nucleic acids) or restriction enzymes. The specific probe locations generate a map unique to each organism. Bound tags can be visualized as a genetic fingerprint or barcode.
“The power of this approach lies in in silico bioinformatics,” said Carolyn Conant, Ph.D., staff scientist for U.S. Genomics. “We perform in silico digests of infectious organisms using multiple restriction enzymes. Then for each digest, we try several different barcode patterns. After that, we find a combination that will uniquely identify each organism in the sample. A single generic fluorescent tag may be sufficient to identify, say, all nosocomial infectious organisms.”
Barcoded fragments of genomic DNA are prepared using a fully automated process, including DNA digestion, labeling with an intercalating agent, and tagging with fluorescent barcodes. Proprietary microfluidics stretch the DNA to its full contour length. The stretched fragments pass through the laser excitation and detection region and the resulting data is compared with the database.
According to the company, DirectLinear Analysis could correctly identify microbial pathogens with greater than 99.5% certainty with an associated false positive rate as low as 1 in 1,000. The assay requires less than 100 copies of the organism in a sample, and only five fragments of any organism are required for its positive ID. The data can also be analyzed by unsupervised learning, with no prior knowledge of the infectious agents. The software then identifies and classifies recurring barcode patterns. This is particularly relevant because less than 1% of microbial organisms have been sequenced. An upload of new sequences to the database enables immediate identification of new pathogens.
“Our approach has a number of advantages, specifically in the hospital setting,” added David Hoey, vp of business development. “The entire process takes just six to eight hours to complete. This means that physicians will be able to refine the treatment much sooner in comparison with the two to three days required for culturing. We can detect and identify organisms that are notoriously hard or impossible to culture, like fungi or tuberculosis. Theoretically, a single test can simultaneously identify an unlimited number of pathogens in a sample.”
Radio Frequency Signals ID Individual Reactions
Analysis of genetic polymorphisms is now commonly carried out by genotyping, using beads conjugated to SNP-specific oligonucleotides. Each bead, such as Illumina’s(www.illumina.com) BeadArray, is identified by a unique nucleotide signature or by embedded digital holographic codes (Veracode). PharmaSeq (www.pharmaseq.com) is developing its own version of a DNA carrier—an electronic radio frequency microtransponder (MTP). Each MTP integrates a photocell, read-only memory (containing the chip ID), transmit logic circuitry, and an antenna, all in a 500-µm x 500-µm x 100-µm chip. Light, collected by photocells, is converted into electrical power, while memory-modulated current is transmitted through the antenna, which emits a radio-frequency signal corresponding to the chip ID.
“We can manufacture a virtually unlimited number of unique ID values,” said Wlodek Mandecki, Ph.D., president and CSO of PharmaSeq. “Unique identification of each carrier is essential for tracking its history from manufacturing through assay and storage. Moreover, each assay can contain a large number of MTPs. Such degree of multiplexing in a microfluidic environment cannot be achieved with currently existing bead technologies.”
For genotyping analysis, the entire surface of an MTP is coated with a polymer derivatized with hydroxyl or amino groups. Oligos could be synthesized directly on the surface of the chip, or attached by conjugation. After hybridization with the target DNA, the suspension of MTPs is passed though a narrow channel of the benchtop analyzer. The first product will be a genotyping test of the CFTR gene, implicated in onset of cystic fibrosis.
“We can detect all key CFTR mutations in the same test tube,” added Dr. Mandecki. “Our first version is able to deliver almost 100% correct determinations.” MTP technology is also amendable to immunoassays and cell-based assays. Cell lines could be grown directly on the surface of MTPs. Cell-coated MTPs can then be studied as a group, not only saving time and costs, but also eliminating variations in experimental conditions.
For many years a biopsy has been the gold standard for diagnosing and differentiating various types of cancer. In many cases, however, incomplete tissue sampling can result in false negative ID. Biopsies are also inadequate for monitoring post-operative cancer patients or for predicting the recurrence of cancer. Molecular diagnostic techniques aim to fill this gap in providing noninvasive, sensitive, and highly correlative measurements based on cancer-specific molecules.
Rubicon Genomics (www.rubicongenomics.com) is developing diagnostic kits based on methylation analysis of multiple genes implicated in tumor origination and progression. “DNA methylation is an excellent analyte, providing a stable and specific cancer signature,” says John P. Langmore, Ph.D., vp of commercial development. “Methylation analysis is able to detect one cancer cell on the background of a thousand normal cells, enabling non-invasive testing of tumor DNA and cells in serum or urine. Because methylation patterns change with tumor progression, these tests will identify the cancer stage and aggressiveness as well as predict disease recurrence and response to treatment.”
The established assays to measure DNA methylation rely on conversion of unmethylated cytosine to uracil by sodium bisulfite, followed by resequencing of the converted regions. The assay is labor intensive, involves organic reactions, and requires a large amount of patient sample for each test. Rubicon’s MethylPlex assays depend on enzymatic reactions to completely destroy the unmethylated sequences and create an in vitro amplifiable library of all the methylated sequences.
“After the methylated DNA has been amplified about 1,000-fold, Rubicon determines methylation at multiple promoters using quantitative PCR or microarrays,” continued Dr. Langmore. “This amplification is essential for conducting the multi-analyte assays necessary to achieve high clinical sensitivity and specificity. For example, in a Rubicon study of a specific solid tumor, individual methylation markers were as little as 20% sensitive for detecting cancer and 6% effective in differentiating between tumor subtypes. However, the Rubicon 35 multianalyte test delivers the correct diagnosis in 100% of cases.”
Another advantage of MethylPlex is that the patient DNA can be stored and re-tested at a later time. “Our assays have 95–98% confidence level for detecting only 30 copies of a methylated sequence. A short-term, noninvasive, blood-based test, such as MethylPlex, presents a perfect adjuvant to tissue pathology. As our knowledge of methylation patterns grows, we hope to supplant biopsies by providing a multianalyte assay,” added Dr. Langmore.
Evolving Regulation of Assays
IVDMIAs (in vitro diagnostic multivariate index assays) are based on scoring data from multiple indicative tests to derive a patient-specific result. Interpretation of the assays depends on the algorithm provided by the test developer. A new FDA guideline would change regulation of IVDMIA components, also known as Analyte-Specific Reagents (ASRs), and place them under 510(k) or PMA regulations.
“In anticipation of changes in FDA regulation of tests previously considered ASRs, we are implementing an ISO13485 compliant production line and will be registering with the FDA for the production of components in our customers’ compliant Class II and III medical devices. The analytic and production capabilities are mostly in place, and we are working on strengthening our document controls and supporting systems,” said Trey Martin, COO, Integrated DNA Technologies (IDT; www.idtdna.com).
Originally, an ASR was defined as a single active ingredient of an in-house diagnostic test that is developed and used by the same diagnostic laboratory. In 1997 FDA asserted its control over the sale, distribution, and use of ASRs. Most ASRs were classified as Class I and were exempt from the agency’s 510(k) or PMA premarketing notification requirements.
This new guidance created a lot of confusion in the ASR manufacturing industry. ASRs were regulated differently from the medical devices that they were a part of. “The industry anticipates that the 2007 guidance will subject most existing ASRs to Class II or Class III regulations. When this comes into effect, IDT will be able to deliver medical device-level material in bulk gram quantities, or formulated and replicated in thousands of tubes or plates,” said Martin. “We expect the demand for diagnostic oligos to increase over the next several years as research methodologies become diagnostic tests.
Genomics is a rapidly progressing field of molecular diagnostics. Genomic data is already widely used for screening for inherited traits, predispositions, and even for prediction of recurrence of a disease. Next-generation sequencing technologies promise to inundate the research community with data that could change the drug development paradigm. Multigenome comparisons of hundreds and thousands of human genomes may become a reality in our lifetime.
The bioinformatic hurdle, however, is inhibiting the practical analysis of whole genomes. Novel sequencing technologies generate massive numbers of short sequencing reads. Mapping such reads to a reference genome using the conventional reading of DNA stored as a flat file takes an inordinate amount of time. Synamatix (www.synamatix.com) uses a different approach for storing and analyzing genomic data.
“We break the human genome sequence into patterns of nucleotides, storing every overlapping, forward, and reverse pattern from one to n, until it becomes unique,” said Arif Anwar, Ph.D, general manager, Synamatix. “Patterns, relationships between patterns, and their significance are maintained in a proprietary structured network database called SynaBASE. By using a search application built on top of SynaBASE, one million 120 mer reads can be mapped back to the human genome in just under one hour. Mapping of 30 mers can be achieved at an average rate of over 1,000 reads per second. We provide over 200-fold improvement in performance in comparison with conventional tools.”
The company is positioning itself as an agnostic solution provider that is able to process, map, and analyze sequence data from any second-generation sequencer faster than it is generated. The Synamatix approach offers a solution that is able to handle all types of reads on the same software platform. SynaBASE is able to correctly call variable quality reads by using position-specific scoring, according to the company.
“We also offer more than 20 different applications based on SynaBASE for high-throughput sequence comparisons, searching, mining, and mapping of bulk genomic data,” said Dr. Anwar.
Kate Marusina, Ph.D., is a life sciences consultant. E-mail: firstname.lastname@example.org.