Medical Device Link.

Originally Published IVD Technology November/December 2001

Toward an automated paradigm for cancer screening

An integrated approach to product R&D may be the wave of the future for IVD manufacturers seeking to market molecular diagnostics.

Eric Gombrich and Susan Keesee

While there is great deal of interest in the discovery of novel markers and probes stemming from the human genome project, identifying and defining such biochemical entities will not by itself result in fulfilling the promise of the biomolecular testing paradigm. There is a great deal of additional work that must be done before companies with an interest in molecular diagnostics will be able to develop functional, cost-effective diagnostic tests from this information.

The development and introduction of a rapid, economic, and automated biomolecular testing system—one that employs multiple analytes and detection systems for the accurate analysis of complex biological mixtures—could result in new markets that dwarf current clinical testing markets. Furthermore, the use of molecular markers in clinical testing has the potential to fundamentally change the paradigm for IVD testing. Despite such potential rewards, however, industry-leading companies have seldom chosen to focus on the development of tests using biomolecular markers and probes.

This article examines the complexities that IVD manufacturers face in moving from the discovery of molecular markers to the development of molecular diagnostic tests, and the potential market for doing so. The application of cytology and pathology techniques in routine cancer screening is used as the basis for examining some of the obstacles inherent in the development and adoption of molecular technologies. Finally, the article describes a system that is now entering clinical trials with the intent of turning subjective cytology testing into a more-objective biomolecular test.

The Promise of Genomics

In the long term, the completed definition of the human genome could make it possible to control an individual's genetically preprogrammed susceptibility to disease. However, the near-term future is more likely to be focused on the use of proteins as a means to monitor and treat pathologic clinical symptoms. The next frontier of discovery for researchers and commercial entities is likely to be the definition of the human proteome—the seemingly endless array of proteins activated and inactivated under the control of the genome. These proteins combine and align themselves to form and govern all aspects of the human body.

As daunting as it may be to define the human proteome, however, such information may be only the starting point for developing and implementing that knowledge in the form of functional tests. Protein markers—or any other form of probe—are like ingredients in a complex recipe. While any one ingredient may be crucial, the other ingredients and the instructions for use of all ingredients are equally important. Unfortunately, in today's world of diagnostic assay development, there appears to be a singular focus on the discovery of novel markers and probes, and less of an effort dedicated to the development of such components into comprehensive tests.

Critical issues that must be addressed in the development of a comprehensive clinical test include the collection, preservation, and preparation of samples; biomarker detection; control of the mechanical, chemical, and programmatic aspects of the assay; and the way in which results are reported. Additionally, the questions of where, when, how quickly, and at what cost the test is to be performed are also critical design issues.

Buried in the midst of such design specifications is the question of whether a single marker is sufficient, or whether multiple markers may be necessary, for the detection of a specific biochemical indicator of disease. But there appears to be reluctance by a majority of the scientific community, and certainly the investment community, to embrace these concerns.¹

Current Uses of Novel Markers and Probes

Historically, it has been accepted practice to market novel markers, assays, or stains as analyte-specific reagents designated for research use only (RUO) status. Even in such a restricted status, RUO reagents have played a valuable role. In the field of pathology, for instance, where the judgment of the pathologist can often be subjective, such products have added to the arsenal of tools available for evaluating patient samples and verifying diagnoses. Companies such as BD (Franklin Lakes, NJ), Dako (Copenhagen, Denmark), Abbott Laboratories (Abbott Park, IL), Ventana Medical Systems Inc. (Tucson, AZ), and others have developed significant businesses centered around the provision of markers and stains, even as RUOs.

However, using such markers or stains can be a labor-intensive process. In pathology laboratories, for instance, a trained pathologist is still required to conduct poststaining microscopic analysis and interpretation to overcome factors of variability. Such labor-intensive processes run contrary to the general trend in biomolecular testing, including pathology, which is to eliminate subjectivity by developing systems that bring objective measures to the testing environment.

One example of such subjectivity is the adoption of Her-2/Neu testing as a method of breast cancer screening.² Most companies' Her-2/Neu kits rely on a pathologist to visualize the sample and decide whether the staining is positive or negative. And because each company's kit is slightly different, the pathologist must be trained to interpret results differently from one stain to the next. Another complexity is presented with different types of biological specimens. Apply the stain in histology to a tissue section, and assay results look one way. Apply the stain in cytology to fixed cells in suspension, and the markers present themselves entirely differently. Lyse the cells in the process of preparing the specimen, and a third model of marker presentation is seen. All of these nuances can be overcome simply by training the pathologist.

If Her-2/Neu testing truly has potential as a global screening tool for breast cancer, however, a process that relies on the availability of a trained pathologist may not offer the optimal model. Such testing is bound to be expensive, as the pathologist's time is valuable. And even trained professionals can process only a limited number of tests in a given period of time. Together, these factors effectively decrease the availability of the Her-2/Neu test—just the opposite of what biomolecular marker–based methods are intended to accomplish.

For all of these reasons, clinical application of such markers or stains has thus far been limited to relatively low-volume testing applications. To make the leap to a comprehensive test with validated results (and thus larger-volume testing) requires stringent protocol design and validation. It also requires a refined development effort to ensure that all of the test elements fit together in a way that facilitates a correct result.

The aforementioned list of issues that must be addressed is but a small subset of the functional and performance parameters that IVD manufacturers must define and control if they are to develop biomarker-based assays into validated clinical tests. Failure to address all such issues will relegate novel markers to applications that are limited in volume, and thus will certainly limit their clinical utility.

Realizing the Promise of Biomolecular Testing

Developing biomarkers for clinical use, and combining them with the platforms and analytical tools necessary to apply them, can result in fully automated, comprehensive tests for both screening and diagnosis. But developing such systems is not a simple task that can be pursued after the fact of research and development. It needs to be conceived and managed as a comprehensive concept from the outset.

Undertaking such a wide-ranging approach to the development of biomolecular tests is a novel business model. Not many companies are pursuing it at this time, and those that are tend to be small and entrepreneurial in nature.³ While there have been several attempts in the past to automate Pap-smear analysis through automated, morphologic analysis, it is only recently that companies such as ChromaVision (San Juan Capistrano, CA), and Ventana Medical Systems have begun to release products to automate laboratory analysis of histology and cytology specimens based on detection of marker-based stains. These current efforts center around the shift from interpretation based on structural presentation of cells and tissue, to detection of specific markers. The use of markers effectively turns what was a subjective interpretation into an objective chemistry test and simplifies the automation effort.

There are sound reasons for pursuing the goal of automation. Not only does automation increase laboratory throughput, it also decreases laboratories' costs for performing tests. And, as tests become more objective, the liability of laboratories also decreases. The challenge for companies that are focused on marker discovery is how to deliver viable tests—not just biomarkers—in order to tap into the broader range and volume of testing performed as a result of routine clinical use.

The market for comprehensive automated tests is vast in terms of both size (whether measured in test volume or dollars) and potential impact on people's lives. The example of cervical cancer screening helps to put this business challenge into perspective.

There are roughly 170 million Pap tests conducted each year around the world.⁴ If we presume an average of 7% of these will be abnormal and result in the need for a biopsy, there will be approximately 8 million cervical biopsies obtained and analyzed annually worldwide. This is not a small market if one has a marker that clearly identifies cervical dysplasia or cancer, so the majority of biomarker researchers are pursuing novel biomarkers to assist the pathologist in diagnosing cervical cancer in histological samples.³ By contrast, the potential market for cervical cancer screening is not merely the 8 million that is currently being pursued, it is the 170 million tests that are conducted for essentially the same purpose. And this does not include markets that are today unable to conduct Pap testing due to deficiencies in infrastructure and training. A rapid, low-cost, fully automated biomolecular screening test for cervical cancer could have application for more than 2 billion women globally.⁵

Automating Cervical Cancer Screening

One such system, the InPath system from Molecular Diagnostics Inc. (Chicago), is just entering FDA clinical trials. The InPath system is a fully automated biomolecular screening test for cervical cancer and its preinvasive precursor lesions.The system includes two novel stains: Cocktail-CVX, for the detection of cervical lesions through the use of multiple protein antibodies, and In-Cell HPV, for the detection of integrated human papillomavirus infection. The company is also developing its platform for use with other biomarkers to detect other cancers and cancer-related diseases.⁶

Staining Strategy. To stain patient samples in preparation for analysis with the Cocktail-CVX, the InPath system simultaneously applies multiple protein-based markers conjugated with separate fluorescent reporters. This approach is especially useful in screening for cervical and other types of cancers, which is conducted using cytology specimens that may include a myriad of cell types at different phases of their life cycle.

Because histology specimens typically contain a limited number of cell types, a single-marker assay can be useful for identifying a particular protein or gene in histologic tissue samples. The P16 marker by MTM (Heidelberg, Germany), for instance, is marketed as a novel stain to aid pathologists in their visual interpretation of such histology specimens.⁷

By contrast, it is difficult to employ a single-marker assay in the analysis of cytology specimens, because the heterogeneous cell populations in such samples may express the target protein at different levels or only during certain phases of their life cycle. Consequently, the marker may bind indiscriminately to either a diseased cell or, at a lower level, to a completely normal cell also residing in the sample. When used in cytology testing, such single-marker assays may possess relatively good sensitivity, but very poor specificity.

The InPath system addresses this issue by using the multiple markers contained in the Cocktail-CVX stain, which makes it possible to analyze three distinct parameters of the cervical epithelial cell. This strategy promotes a high degree of sensitivity for cervical dysplasia and cancer while also delivering very high specificity—a prerequisite for any screening test.

The first parameter is addressed by quantifying the occurrence of known cancer-related proteins such as CD71 via fluorescence intensity measurements. This is accomplished through the use of specifically cloned antibodies targeting these proteins.

The second parameter is evaluated through the use of yet another specifically cloned antibody targeting endocervical cells. These markers are also quantified, and then subtracted from the positive cell analysis to calculate the level of cellular terminal differentiation in the specimen.

The third parameter measured is a determinant of cellular integrity, which is used to eliminate artifacts and debris from the analysis. This measurement is accomplished through the use of a fluorescent dye that binds to the minor groove of double-stranded DNA and, in essence, identifies viable cell nuclei. This third characteristic of the assay focuses the system's image-analysis software on true cellular events, thereby increasing the specificity of the assay. Through this multiparameter design, the assay is able to identify specimens that contain cellular abnormalities.

Figure 1. The InPath Slide-Based Test platform includes an automated microscope with integrated analytical software designed to operate specifically with the InPath assays for fully automated slide-based analysis.

The automated stainers in use include the Zymed ST5050 by Zymed Laboratories Inc. (South San Francisco) and the NexES IHC staining system by Ventana Medical Systems Inc. To meet the requirements of these stainers, Cocktail-CVX stain comes prepacked in a series of vials, along with controls. Staining chambers in the immunostainer are loaded according to the directions for the InPath system, and staining is conducted under guidance of the integrated software. Batches of slides can be stained in unison.

Analysis is carried out by the AcCell instrument, which comprises a fully automated microscopy station equipped for the capture and processing of fluorescent images. A multistage three-color image-analysis algorithm is employed to identify cells of potential interest. The initial phase of the algorithm employs a calibration slide to perform a series of calibration and normalization processes to ensure that the system is performing properly; that the specimen is suitable for evaluation; and that the images are properly conditioned, scaled, and normalized for analysis.

The AcCell unit incorporates a high-precision automated stage that electronically controls the motion of the specimen slide along the x, y, and z (focus) axes. The electronic controller is coupled to an automated slide handler for loading and unloading slides to and from the stage. Controls for the unit's Olympus BX microscope are routed to an external PC, which manages both the camera and imaging software. All communications are integrated and accessed via a serial communications protocol. The entire system is contained in a housing that prevents ambient light from interfering with fluorescent analysis.

Principles of Operation

The first step performed by the InPath image-analysis software is to evaluate the image captured in the blue spectral channel, which detects and localizes each cell on the slide. A nuclear DNA dye incorporated into the staining assay selectively stains cell nuclei blue, thus distinguishing them from the background of nonnucleated cells, cell debris, and mucus. This step also makes it possible for subsequent image processing operations to be applied only to cellular material.

In the second analytical step, the system evaluates the image captured in the red spectral channel, which is generated by selective staining of the cells with fluorescently labeled antibodies against extracellular epitopes of transferrin receptor (CD71). The expression of this receptor—and thus the level of immunostaining that occurs—is substantially enhanced in cells that are metabolically active, as is characteristic of dysplastic or malignant cells with high levels of DNA synthesis. Since the cells in a normal exfoliative cervical cytology specimen are nominally not proliferating and do not exhibit a significant level of metabolic activity, any red fluorescence detected in the immediate vicinity of a blue cell nucleus can be taken as an indicator of a potentially abnormal cell. An intensity threshold determines whether the level of red fluorescence is significant. The red fluorescent label can also be simultaneously applied to other antibodies as deemed necessary for the application.

It is known that a normal exfoliative cervical cytology specimen can contain polymorphonuclear lymphocytes (PMNs), bacteria, and endocervical cells, some percentage of which can be proliferative or metabolically active and thus stained by the CD71 reagents in the cocktail.^8,9 To eliminate the small number of stained, activated PMNs that may be present in a sample, the InPath system uses a series of size gates. The gates are set large enough to exclude bacteria and PMNs from further consideration while retaining small dysplastic squamous cells and adenocarcinoma cells.

The third analytical step evaluates the image captured in the green spectral channel, which is generated by a fluorescently labeled endocervical antibody in the staining cocktail. This antibody specifically binds to a cytokeratin that is exclusively associated with endocervical cells. Any green fluorescence detected in the immediate vicinity of a blue cell nucleus indicates that the object is an endocervical cell. Using this information, the system can count the number of endocervical cells to assess the adequacy of the specimen, and can discriminate between endocervical and squamous cells to classify the specimen. Endocervical cells that also exhibit red fluorescence are proliferating, differentiating, or metabolically active, and are thus indicative of either a reparative process or endocervical abnormalities.

Figure 2. The image on the left is a standard liquid-based, cervical cytology sample that has been Pap stained. The image on the right is the same field of view as the Pap-stained image, but based on staining with the InPath Cocktail-CVX. Note the fluorescent red objects and compare them with the same cells via Pap stain in the prior image. InPath allows for Pap counter-staining so a single sample can be processed through the InPath system and routine Pap testing.

The InPath system was designed to permit adjunctive staining of a cytology sample with the Cocktail-CVX, followed by reflex testing using a conventional Pap stain, with no disruption in the morphologic presentation of the cells. This enables a single sample to be used as both experiment and control for purposes of validation and clinical trials. The same practice can be followed in actual laboratory use. A single slide can be first analyzed using the InPath method and, if positive, subsequently Pap stained for adjudication.

A large group of dysplastic cells stained using the Cocktail-CVX, as opposed to Pap staining, fluoresce in red, while other nondiagnostic cells are not detected (see Figure 2). Preclinical studies using the InPath system have resulted in 100% sensitivity to high-grade lesions (CIN 2, CIN3) and specificity of more than 80%.¹⁰

The Integrated Approach to R&D

For many IVD organizations and researchers, product R&D begins and ends with the search for a single "silver bullet" biomarker for a particular disease. Although highly sophisticated and time-consuming, such primary science research stops short of creating fully formed diagnostic platforms for use in real-world clinical applications.

By way of contrast, development of the InPath system was conducted as an integrated whole that included the biomolecular, hardware, and software elements. Each of these components was developed simultaneously with a single application in mind.

Different types of work were required to optimize the operation of each part of the system. Much of the R&D for the Cocktail-CVX stain, for instance, focused on evaluating different clones of CD71 and other antibodies and combining them with the appropriate tags for the intended application. And on the hardware side, a host of technical attributes came under scrutiny, including the types of microscope slides to be used, how cells would be preserved and deposited on the slides, how the staining would be conducted, and so on.

In the IVD marketplace, pursuit of such an integrated approach to R&D is still rare enough to be considered distinctive, but it may also represent an important trend for the future. Clinical researchers are increasingly concluding that very few diseases can be characterized by a single detectable element, meaning that the future of biomolecular diagnostics is likely to require the ability to detect combinations of multiple markers. Integrated R&D is especially suited to the task of creating comprehensive, automated systems for diagnosing and monitoring diseases where no single marker is sufficient.

Changing the IVD Paradigm

Figure 3. The InPath e2 Collector is a single-use balloon shaped like a mirror image of the cervix. It is operated by the reusable handle. The device allows for complete sampling of the entire cervix in a single step, with greater comfort for the patient, and less blood and other obscuring material in the sample.

The InPath system includes an example of such a novel technology, a unique device for the collection of cervical specimens (see Figure 3).¹¹ While this device may have advantages simply as a collection tool, it was actually designed as a means to an end. Eventually, the device is intended to entirely eliminate the need for a specimen slide.

Figure 4. The InPath POS test is an instrument designed to operate with the InPath e2 Collector. The POS instrument automatically stains cervical samples collected with the e2 Collector, and then uses lasers and other methods to scan the Collector's surface of suspect material. The complete process takes less than 20 minutes and provides point-of-care screening capabilities.

The process developed for the InPath POS instrument may eliminate the need to perform Pap testing on 80% of samples, as they could be identified as disease free even before the patient leaves the physician's office. Such a process could significantly reduce the operational requirements of large-volume Pap-testing laboratories, thereby improving the ability of pathologists to manage only cases where disease is present.

The prototype of the InPath POS instrument is currently in development and is expected to be available for clinical testing before the end of 2001.

Conclusion

The organizations that dominate the IVD industry are often seen to be wrestling to manage single-digit growth. At the same time, they may be missing opportunities to double, triple, or quadruple their business.

Moving to the next frontier of the clinical laboratory is likely to require a new set of skills that many companies do not currently possess. Developing the next generation of comprehensive, integrated test devices for the detection of disease will require companies to have the ability to fully integrate chemistry, hardware engineering, and software development. In many cases, doing so will require companies to make dramatic changes in the definition of their business.

The integrated R&D that will be needed to produce next-generation IVDs will not be attained by merely aligning an assay developer with an automated microscope manufacturer through a comarketing agreement, and applying a third company's software. Accomplishing such a fundamental shift in business thinking requires breaking the business models employed by the largest and most successful companies in the IVD industry.

The promise of biomolecular testing lies in providing more-accurate, objective test results, and disseminating disease screening and diagnostics to a broader market. Continuing to partition the elements of biochemicals, hardware, and software among separate research, development, and regulatory entities is a recipe for denying that promise.

References

1. S Hensely, "Proteins—Not Genes—Could Be Clue to Human Complexity, Disease," The Wall Street Journal (February 13, 2001), B1.

2. S Wang et al., "Assessment of Her-2/Neu Status Breast Cancer; Automated Cellular Imaging System (ACIS)–Assisted Quantitation of Immunohistochemical Assay Achieves High Accuracy in Comparison with Fluorescence In Situ Hybridization Assay as Standard," American Journal of Clinical Pathology 116, no. 4, (2001): 495–503.

3. EG Gorman, "Building a Diagnostics Market: One Molecule at a Time," IVD Technology 7, no. 7 (2001): 59–67.

4. RJ Kurmanetalo et al., "Interim Guidelines for Management of Abnormal Cervical Cytology," Journal of the American Medical Association 271 (1994): 1866–1869.

5. Planning Appropriate Cervical Cancer Prevention Programs (Seattle: PATH Foundation, 2000).

6. "Molecular Diagnostics Inc. Announces Strategic Venture with Leading Provider of Tissue Analysis Systems," in Molecular Diagnostics Inc. home page [on-line] (Chicago: Molecular Diagnostics Inc., 10 October 2001 [cited 15 October 2001]); available from Internet: http://www. molecular-dx.com.

7. H zur Hausen, "Human Papillomaviruses," Cancer Research 49 (1989): 4677–4681.

8. RC McGlennen et al., "Expression of Cytokine Receptors and Markers of Differentiation in Human Papilloma Infected Cervical Tissue," American Journal of Obstetrics and Gynecology 165, no. 3 (1991): 696–705.

9. JM Lloyd et al., "Demonstration of an Epitope of the Transferrin Receptor in Human Epithelium as a Potentially Useful Cell Marker," Journal of Clinical Pathology 37, no. 2 (1984): 131–135.

10. B Patterson et al., "Molecular Biomarker–Based Screening for Early Detection of Cervical Cancer," Acta Cytologica 45, no. 1 (2001): 36–47.

11. "Molecular Diagnostics Submits 510(k) Application to the FDA for the InPath e2 Collector," in Molecular Diagnostics Inc. home page [on-line] (Chicago: Molecular Diagnostics Inc., 1 October 2001 [cited 15 October 2001]); available from Internet: http://www.molecular-dx.com.

Eric Gombrich is vice president for business development and Susan Keesee, PhD, is vice president for molecular science at Molecular Diagnostics Inc. (Chicago). The authors can be reached via egombrich@molecular-dx.com and skeesee@molecular-dx.com, respectively.

Photos Courtesy Molecular Diagnostics Inc.

Copyright ©2001 IVD Technology