April 1, 2008 (Vol. 28, No. 7)
K. John John Morrow Jr. Ph.D. President Newport Biotech
But Validation Must Be Improved to Obtain Truly Accurate Diagnostic Tools
It is widely acknowledged that early detection is the key to successful cancer treatment, yet such aggressive intervention requires accurate screening tools. Traditional diagnostic tests for cancer have a number of drawbacks including poor sensitivity, lack of specificity, and costly and invasive protocols. Now, finally, a variety of new biomarkers offers a wealth of possibilities. At CHI’s recent “PepTalk 2008” meeting the results of studies on markers for lung, pancreatic, and various malignancies were scrutinized.
The need for early diagnosis is a powerful driver in the search for new cancer biomarkers. Edouard Nice, Ph.D., and his colleagues at the Ludwig Institute for Cancer Research, recounted their efforts to develop early detection tools for malignancies. “Our strategy was to employ multidimensional, high-performance liquid chromatography to trace enrich low-level components such as growth factors in tumor material prior to the mass spectrometry analysis,” he said.
Dr. Nice’s studies have focused on a number of putative cancer biomarkers, but much of his discussion concerned telomerase, an enzyme long known to maintain the telomeres, or chromosomal ends, during cycles of cell replication. The decline of telomerase activity with age is thought to be responsible for chromosomal instability, resulting in aging and senescence.
There is a strong relationship between telomerase and tumor development; telomerase is expressed in more than 85% of cancers. In the Ludwig Institute’s work, the enzyme is measured by using magnetic beads conjugated to the telomere recognition site, a group of repeated TTAGGG nucleotides. The cells are lysed and telomerase activity is detected by the incorporation of multiple biotinylated nucleotides into the growing DNA strands, which are then reacted with streptavidin-conjugated horseradish peroxidase. Dr. Nice and his colleagues validated this assay procedure using a luminescent tag and then examined cells voided in the urine of bladder cancer patients and in the feces of colon cancer patients.
The gold standard for diagnosis and monitoring of bladder cancer has been cystoscopy combined with urinary cytology. This procedure has many drawbacks, including invasiveness, attendant morbidity, and high cost. Since the vast majority of bladder cancers arise from the urothelium, analysis of the voided cells in the urine provides a potential source of cancer-specific markers.
The magnetic bead telomerase procedure is exquisitely sensitive, detecting as few as 10 cancer cells, and its accuracy was confirmed by an independent pathological review panel, according to Dr. Nice. Samples from normal healthy volunteers were all telomerase negative. In the future, the assay could be configured for high throughput and automation and is a prime candidate for a first-line assay for detection and monitoring of colon and bladder cancer, he added.
To this end, Dr. Nice and his colleagues are working with Sienna Cancer Diagnostics to move the technology forward. Sienna recently announced a licensing agreement with Geron for worldwide exclusive rights to develop telomerase for the in vitro diagnostic market.
Liquid Chromatography & Mass Spec
The combination of the physical separation capabilities of liquid chromatography with the mass analysis capabilities of mass spectrometry is alive and well as a powerful diagnostic tool, reported Nigel Clarke, Ph.D., director of endocrinology at Quest Diagnostics Nichols Institute.
Presently LC-MS is expanding beyond its traditional role at the front line of small molecule diagnostic measurements to large molecule (mainly protein) diagnostics. Previously, radioimmune assay (RIA) for metabolites and hormone levels was the approach of choice, but it has too much baggage. Shortcomings to definitive RIA performance included the presence of autoantibodies in patients’ serum, difficulty in automating cumbersome protocols, and cross reactivity of antibodies, all causing a surfeit of false positives.
With the need to assess specific proteins, a different set of issues present themselves, including the requirement for high sensitivity to analytes present in vanishingly small amounts, coupled with their miscreant behavior. Dr. Clarke consigns to the bad-behavior category, molecules that are unstable, complex with other molecules, and clean up only with difficulty. Therefore, building new diagnostic tests requires robustness, small volumes, accuracy, precision, and reproducibility.
Dr. Clarke and his associates have studied the angiotensin system in the development of a superior LC/MS assay. Angiotensin II is the final active messenger of the renin-angiotensin pathway, it binds to a specific receptor, precipitating vasoconstriction and fluid retention, leading to an increase in blood pressure.
Since many drugs developed to control hypertension regulate blood pressure at various points in the renin-angiotensin system, it is critical to have accurate measures of the members of this cascade.
Dr. Clarke described his group’s efforts to improve the preparative steps, automating them, and lowering the volumes of material required. The group found that LC/MS/MS assays required much less sample volume and were more specific than traditional RIA. One of the major problems of RIA that the group has resolved is its limited dynamic range, which extends from 0.65 ng/mL/hr to 14 ng/mL/hr. Dr. Clarke reported that his group’s LC/MS/MS procedure improved sensitivity by an order of magnitude.
“We need to face new problems with protein analysis that do not exist in small-molecule assays,” he said, “and external testing agencies need to be educated with regard to these new methodologies.”
Ovarian Cancer Antigens
“There are several strategies for the early detection of cancer,” says Michael Tainsky, Ph.D., professor at Wayne State University School of Medicine. “One can screen patients’ serum for plasma DNA, antigenic proteins, marker proteins, or for antibodies.”
This range of options opens up the possibility of developing a number of different detection protocols. It is unlikely that any single protein can serve as an unambiguous marker for malignancy, therefore effective detection systems will require that a number of different proteins be used. Thus, a definitive diagnosis would represent the composite value of several cancer-related markers.
Dr. Tainsky’s program harnesses antibodies in patients’ serum for the detection of cancer-specific epitopes using peptides selected for IgG binding from phage-display cDNA libraries. This detection method is limited to amino acid epitopes and cannot detect epitopes produced through post-translational modifications such as those involving glycosylation or addition of lipid molecules to the original amino acid chain.
A prime target for a serum-based early detection test is ovarian cancer. Survival rates are quite high when the disease is confined to the ovary, yet early detection is challenging since symptoms are easily confused with many other illnesses. Often by the time the condition clearly manifests itself it is too late. Indeed, only 29% of cases are detected early enough to take remedial action.
Dr. Tainsky and his colleagues use microarrays to which phage displaying different peptides from tumor cDNA libraries are bound. The phage are put through a number of rounds of selection and enrichment in order to produce highly tumor-specific clones, and the end products are robotically spotted to build the microarrays. Using this approach, the Tainsky team was able to isolate 65 antigen-producing phage clones positive for ovarian cancer samples, yet with low binding to sera from women with benign gynecological conditions.
The question of why these biomarker proteins stimulate an immune response in the patients goes to the origins of the cancerous state. According to Dr. Tainsky, the proteins are either normal proteins that are overexpressed so much so that the autoimmune-protective mechanisms of the individual break down, or they are part of the normal repertoire, mutated to a state in which they appear foreign to the host.
The autoantibodies revealed through this technology can be used to detect ovarian cancer, and immunohistochemical studies have determined that some of the epitopes can serve as tissue biomarkers. They also offer the prospect of developing patient-specific cancer vaccines that could be individualized for the patient, Dr. Tainsky said.
High-Density Antibody Arrays
Designing a new generation of biomarker detection platforms requires reducing background to the lowest level possible. Bryce P. Nelson, Ph.D., research and development vp at Gentel Biosciences, discussed progress in building antibody arrays for research use.
In order to lower background, Gentel employs a nonporous, optically clear, ultrathin nitrocellulose film that is attached to a 1 x 3 inch glass slide. The completed unit can be subdivided into 16 wells for multiplex assays containing more than 45 analytes per well.
The company has developed a single capture assay in which antibodies are bound to the surface. The sample to be tested is labeled with fluorescein and serum proteins are specifically captured. The sample is then reacted with a detector antibody (antifluorescein) labeled with a fluorescent dye, which can then be measured using a standard fluorescence scanner.
The Gentel researchers identified candidate biomarkers through the literature and sourced high-performance antibodies, eliminating those that showed cross reactivity. The antibodies can then be quality controlled for reproducibility. Because of the small volumes required and the ability to subdivide the array into many individual sample wells, it is possible to generate a large body of data from a single slide. Many samples can be leveraged allowing the characterization of disease-related proteins and the response of regulatory proteins to therapeutic interventions.
Charles Birse, Ph.D., lung biomarker group leader at Celera, reported that the currently available selection of lung cancer biomarkers is inadequate, represented by carcinoembryonic antigen, cytokeratin fragment 19, and neuron specific enolase. Not one of these markers provides the sensitivity and specificity required of a truly accurate diagnostic tool.
This shortage has spurred the company’s search for better serum markers. The huge range of concentrations of different proteins in serum makes the search difficult and requires that the 97% most concentrated proteins such as albumin and immunoglobulin be removed prior to evaluation. But albumin is notorious for binding to other proteins, and its elimination may inadvertently remove possible biomarkers.
So Celera has an extensive program of discovery and validation, screening for shed proteins as well as cell-surface proteins in tissue specimens, cell lines, and serum samples. Using mass spec as a screening method, Celera researchers have identified over 500 possible disease-associated proteins. These investigations narrowed down the candidates that are expressed in the serum of patients, and finally yielded 27 targets as promising lung cancer biomarkers. The Celera researchers are also examining tissues, and Dr. Birse cautions that serum may not be appropriate for the validation of all biomarkers.
Pancreatic Cancer Biomarkers
Pancreatic cancer results in 30,000 deaths per year in the U.S., almost equal to the number of new cases. The five year survival rate is less than 5%, reported Ru Chen, Ph.D., a research scientist in the department of medicine at the University of Washington. Early detection could significantly improve the prognosis of this deadly disease. When pancreatic cancer is detected in the early stage (when the tumor is smaller than 2 cm), the five-year survival rate is about 46%.
Dr. Chen’s screening program is similar to many other cancer biomarker searches described in this article, including evaluation of pancreatic tissue, pancreatic secretions, and serum profiles from patients and matched controls. Proteins from normal and cancerous tissue are differentially labeled, broken into peptides, and affinity separated. This allows protein classification according to whether they are increased, decreased, or remain at the same level in the normal as compared to the malignant tissue.
Dr. Chen and her colleagues identified 500 pancreatic proteins that were unchanged, 95 that were upregulated, and 61 that were downregulated. They then designed a microarray platform in which paraffin specimens from a variety of normal and diseased tissue were mounted on an assembly that scanned them, using standard pathology staining, immunofluorescence staining, and immunohistochemistry.
Dr. Chen’s group is now faced with a large collection of biomarker candidates. In order to prioritize them she has developed a list of criteria including degree of overexpression, role of the protein, availability of ELISA technology, whether the protein is extracellular or membrane bound, correlation with other analyses such as RNA and DNA detection, and the specificity of the protein as an authentic pancreatic cancer marker.
“Using this approach, we have identified several potential biomarker candidates that really work well on the cancer tissues. To be clinically useful for screening the general population, these tissue biomarkers need to be developed into blood assays and further validated in serum/plasma,” Dr. Chen stated.
Discovery to Clinical Validation
There is an active field of cancer biomarker study that focuses on proteins that have been recognized as being associated with states of malignancy. Philip Hemkin, Ph.D., principle scientist, cancer diagnostics R&D, Abbott Laboratories, discussed his group’s investigations of the tissue inhibitor of metalloproteinases-1 (TIMP-1).
Over the last decade a number of cancer researchers have observed elevated levels of TIMP-1 in the plasma of cancer patients, while healthy individuals maintain lower levels within a narrow range. These studies suggested to Dr. Hemkin and his colleagues that TIMP-1 might serve as an early detection signal for colon cancer.
The Abbott group developed an automated, two-step, double-mouse monoclonal sandwich assay in which an antibody against TIMP-1 is attached to a magnetic microparticle. A second antibody labeled with acridium will generate a light signal.
The group optimized the assay procedure, which has a large dynamic range, extending from zero to 2,500 ng/mL. The kit shows great precision and stability, he reported.
With a robust assay available, Dr. Hemkin and his colleagues are well positioned to continue obtaining a detailed clinical evaluation of the efficacy of TIMP-1 as an early marker for colon and other cancers.
DNA immunization, first introduced in 1999, is a technique for rapid generation of antibodies that until recently had not reached it full potential. Strategic Diagnostics adopted this technology for the purpose of producing large numbers of antibodies to putative cancer biomarkers. The approach involves the introduction of genes for target proteins (usually a sequence coding for a representative peptide of about 100 amino acids) into the animal to be used for the immunization procedure.
This results in a savings of time and resources since purification of a protein to a state satisfactory for use as an immunogen can be extremely challenging. With DNA immunization, a precise sequence is inserted into the carrier plasmid and there is no possibility of the inadvertent introduction of contaminating proteins.
According to Damon Hostin, marketing development manager, Strategic Diagnostics has optimized the procedure, overcoming problems such as poor immune response by the host animal that plagued the procedure in the past. “We broke the method up into its component parts, and then optimized the conditions of each step,” said Hostin. “The algorithms for protein modeling have improved greatly in the last few years, so we are able to generate DNA sequences for peptides that are much better immunogens.”
With present-day modeling, peptides can be produced that assume their appropriate 3-D folding, thereby ensuring that the authentic epitopes are available for stimulating the immune system of the host.
This means that antibodies can be produced on an assembly-line basis, avoiding the difficulty of expressing and purifying a protein that could be unstable and technically demanding. Over the course of its existence, the company has produced around 1,500 antibodies at its GMP facility, Hostin reported.
According to Leigh Anderson, Ph.D., CEO of the Plasma Proteome Institute, the search for cancer biomarkers is a harsh, unforgiving affair. “It is a high-risk enterprise, with a 99% attrition rate,” he says. “This may be a reflection of the huge level of genetic variability within the human population.”
What Works and What Doesn’t
There is no guiding theory of biomarker science, Dr. Anderson noted, and it is completely empirically based. There is also a plethora of candidates that are difficult to resolve, and verification is slow, and appears to be a rate-limiting step. Despite the critical importance of cancer biomarkers in early detection, the protein-based in vitro diagnostics industry is only 1% as large as the pharmaceutical business. This may explain why, after years of investigation, the number of clinically successful biomarkers obtained through proteomics is zero.
“The major roadblock is the need for large (one to two thousand patients) sample sets when one is evaluating a candidate biomarker,” Dr. Anderson said. “There are lots of possible biomarkers out there, but the amount of effort required for a really decisive assessment is immense.”
Karin A. Hughes, Ph.D., worldwide product director for sepsis, stroke and protein C at Biosite, feels the problem may lie in the fact that discovery tools do not intersect with validation tools. “There are a large number of potential biomarkers but discovery technologies are not suitable for biomarker validation. Because the skills to validate them do not always reside in-house nor with discovery partners, biomarkers get stuck in the research phase and are not moving forward.”
In order to move cancer treatment forward, an emphasis on diagnosis as opposed to therapy will be required. With therapeutic protocols running to hundreds of thousands of dollars per patient, effective biomarkers that detect cancer while it is still treatable look like a pretty good investment.
Dr. Anderson is optimistic that this day is dawning. “We have a project in place with the NCI to study 300 possible cancer biomarkers,” he stated. “This will provide definitive information on the first large cohort of candidates.”
K. John Morrow Jr., Ph.D., is president of Newport Biotech Consultants and contributing editor for GEN. Web: www.newportbiotech.com. E-mail: [email protected].