A Sharper Image

Medical miracles abound, yet cancer continues to be a complex and challenging problem. "Cancer" is actually a generic, catchall term for the malignant tumors that are found in well over a hundred different diseases, but the basic concept is simple enough--a gene goes wrong and a tumor grows. Unfortunately, the reality is more complicated, involving an intricate sequence of phenomena and interactions in just a handful of the body's tens-of-trillions of cells. And therein lies the problem for rese

Apr 30, 2001
Bob Sinclair
Medical miracles abound, yet cancer continues to be a complex and challenging problem. "Cancer" is actually a generic, catchall term for the malignant tumors that are found in well over a hundred different diseases, but the basic concept is simple enough--a gene goes wrong and a tumor grows. Unfortunately, the reality is more complicated, involving an intricate sequence of phenomena and interactions in just a handful of the body's tens-of-trillions of cells. And therein lies the problem for researchers and physicians: once that handful of cells has multiplied to form a detectable mass, or one that directly results in symptoms such as pain or bleeding, it is frequently too late for effective intervention. Cancer diagnostics research is therefore focused on early discovery and identification of cancerous cells.

Fundamentally, cancer is a genetic problem that is either inherited or developed during an individual's lifetime, and cancer biologists have identified many of the genes that are involved in this process. These genes fall into two basic classes: proto-oncogenes and tumor-suppressor genes. In their normal states, a careful balance between these two classes of genes regulates cell growth and division, with proto-oncogenes accelerating cell growth and division and tumor-suppressor genes putting on the brakes. When mutated, proto-oncogenes can become oncogenes, producing too much or too active growth-stimulating product and driving the cell into a frenzy of unregulated division; loss of growth-inhibiting protein arising from mutation of a tumor-suppressor gene can have a similar effect.

Once the cancerous process is underway, a variety of detection options are available. The presence of cancer cell-specific components in the blood or lymph can be used to provide indirect evidence of cancer. Researchers have also been using cytohistological methods to identify cancerous cells based purely on conventional staining and cell morphology for nearly half a century.1 The diagnostic use of cancer cell-specific surface features, usually involving ELISA, flow cytometry, or cytologic staining using labeled antibodies, is a more sophisticated approach. Finally, as molecular techniques have become increasingly powerful, direct analysis of the DNA has become a viable option.

Once a detectable tumor has formed, in theory it is an easy matter to remove the mass surgically. However, this treatment loses efficacy if the cancerous cells have already metastasized. During metastasis, small numbers of individual cells or cell clusters detach from the tumor and migrate to another location in the body, where they form a new colony. Metastasis is a complex, multi-step process, and evidence suggests that rather than a few random cells from a tumor spreading, only specialized subpopulations of cancer cells are able to perform the metastatic transformation (which suggests that other genetic mechanisms, of which we are currently unaware, are also involved). It is these freely-circulating cells that often provide an initial handle on cancer detection.

Generally, samples for diagnostic evaluation (e.g., peripheral blood, bone marrow, lymph, or thin sections of biopsy tissue) are fixed onto slides and stained in a variety of ways prior to microscopic analysis. Microscopy can be performed in dozens of ways, each of which may have benefits or shortcomings for a given application. Two-color immunofluorescence, for example, has been shown to be superior to flow cytometry and immunohistochemistry for the detection of micrometastatic breast cancer in bone marrow,2 whereas DNA image cytometry has been shown to have improved sensitivity over conventional cytology in detecting cancerous cells in urine.3 Fluorescence in situ hybridization (FISH) analysis is also both highly sensitive and specific.4

Monoclonal antibodies are frequently used to label cancerous cells. There is an ongoing hunt for antibodies that are directed against new or novel cancer components, that are more specific, that bind more tightly, or that give lower background signals. The literature is rich with reports of new target molecules.5,6 Often, these same antibodies can be used in an initial flow-cytometric or magnetic purification step to enrich the cell population for the cell type of interest.7-9

Though detection of aberrant cell surface markers is useful, it is best to confirm these results by direct examination of the DNA. For example, in one study, PCR-single strand conformation polymorphism (verified by genomic sequencing) was shown to have an 80 percent concordance with immunochemical flow-cytometric evaluation of p53 overexpression in breast cancer tissue.1 But while DNA-based tests may be far more sensitive than microscopy (in terms of detecting a small number of cells among a large background), the presence of a DNA sequence, per se, may not be sufficient to make a confident diagnosis. Often, a mutated or amplified DNA sequence needs to be correlated with an expression phenotype such as the presence of a marker on or in the cell; this can only be detected via microscopy of appropriately stained cells. However, this leads to the problem of finding small numbers of aberrant cells on a microscope slide among perhaps millions of normal ones. Several companies have attempted to solve this problem with the creation of computer-controlled scanning microscopes that feature sophisticated image analysis and archiving combined with hands-off processing of many slides.

Automated Systems for Tumor Cell Detection

Courtesy of ChromaVision

Detection of a single cell in the bone marrow using ChromaVision's ACIS
ChromaVision Medical Systems Inc. of San Juan Capistrano, Calif. (www.chromavision.com), produces an Automated Cellular Imaging System (ACIS) that allows researchers and pathologists to examine far larger numbers of cells of potential interest than can be processed manually. The ACIS microscopy system is an automated conventional light microscope and light source mounted in a shock-resistant housing. A computer- controlled X-Y stage holds carriers that contain four slides, and an input hopper holds up to 25 carriers, allowing walk-away scanning for up to 100 slides. Images are collected by a 640 x 480-pixel digital camera and processed by a computer running proprietary software for sensitive color detection as well as detection and analysis of a variety of size- and shape-based morphometric features.

Cells from clinical samples such as bone marrow or peripheral blood are first deposited on microscope slides via cytocentrifugation, a relatively gentle process that leads to minimal cell damage and distortion. These cells are then stained using appropriate antibodies, and processed by the ACIS. During the initial pass of each slide at low (10x) magnification, the ACIS counts the cells; those exhibiting positive staining are noted and their positions on the slide recorded. A second pass at higher magnification (40x or 60x) revisits these areas of interest, characterizing stained cells based on size and shape and collecting digital images that are stored for later examination by the pathologist.

As with many lab-automation systems, much of the power of the ACIS lies in its tireless and consistent performance of otherwise tedious manual tasks. While a human technician may get fatigued toward the end of the day, the ACIS continues to examine slides with the same rigor around the clock. This reliability, combined with scan speeds that exceed human capability, means that many more cells from a sample can be examined, leading to improved statistical parameters and a greater likelihood of detecting extremely rare cell types. After the run, the pathologist can instantly call up the digital images and when necessary can even view every positively stained cell from a set of slides. (Should the pathologist want to actually view the slide itself, a revisit feature allows the user to select an image from the ACIS display and to be precisely taken to that location on the slide.) The digital images allow for efficient archiving as well as sharing of images with colleagues and consultants.

One of most important-and difficult-tasks in breast cancer diagnostics is detecting migrating metastatic cells and other rare cancer cells in the bone marrow; these cells may be present at densities of only one in 10 million cells. Direct comparison of a pathologist's diagnoses with those of an ACIS-assisted pathologist confirm the power of the ACIS: the pathologist failed to find cancerous cells in 17 of 39 samples. However, direct examination of the unusual cells identified by ACIS from those same samples, by the same pathologist, confirmed the presence of the cancerous cells.10

The ACIS platform can be set up with several different software applications that are optimized for working with different analyses. For breast cancer and other malignancies, these applications include characterization of nuclei positively stained for estrogen receptor, progesterone receptor, Ki-67 stained with MIB-1 antibody, HER-2, and p53. In addition, the ACIS is capable of analyzing cell populations for microvessel density, DNA ploidy, and for rare cells in bone marrow and lymph node tissue.

Courtesy of MetaSystems

Metasystems' Metafer/Metacyte software

MetaSystems GmbH (www.metasystems.de) of Altlussheim, Germany, also has a number of automated options for scanning and imaging pathology slides. Based on Zeiss optics, MetaSystems' automated imaging instruments include a microscopy system, computer-controlled stage, digital imager, and light source. However, in these devices the light systems require an automatic filter changer because MetaSystems' instruments are fluorescence-capable.

MetaSystems' metafer/RCDetect system searches for and identifies rare cell types, primarily from blood, bone marrow, or tissue section samples. Cells are tagged with fluorescently labeled antibodies and automatically scanned at a rate of 20 x 106 cells per hour (the standard system holds eight slides, optionally expandable to 100 slides). As with ChromaVision's ACIS, a first pass examines cells at low magnification. With the increased options available using fluorescence, this first pass uses just the filter channel necessary to detect the labeled antibody. Once the position of positively stained objects has been recorded, objects of interest are re-examined using different excitation or filtration appropriate for visualizing counterstains and nuclear features. Images of objects that match certain criteria (such as appropriate stain, presence of a nucleus, size, or shape) are digitally stored for human examination. As nonspecific antibody reactions may occur, the verification of the cell type by means of an additional assay is crucially important in the assessment of minimal residual cancer. Thus, in a second pass, the objects of interest can be automatically relocated and reexamined at a higher optical magnification via FISH. Alternatively, sequential or simultaneous detection of multiple antibodies labeled with distinguishable fluorochromes can be employed for maximum specificity. MetaSystems' isis software package can be used to capture and correlate these immuno-label and FISH results.

Metasystem's metafer/Msearch is an automatic metaphase finder that determines the optimal plane of focus for a metaphase-spread slide and then rapidly scans for cells. The positions of features for specific metaphases are recorded, and the spreads themselves are displayed as an image gallery. From this gallery, the user can select optimal spreads for further analysis. MetaSystems' ikaros karyotyping system, a fast and intuitive graphical program for organizing and analyzing karyotypes from a spread, can be integrated with Msearch and directly called from any selected metaphase image. In addition to karyotyping, well-separated metaphase spreads identified by Msearch can also be subjected to a variety of other probe-specific analyses such as FISH.

FISH involves hybridizing a fluorescently labeled DNA probe to chromosomal DNA without significantly damaging the cells.11 MetaSystems' isis FISH imaging system features automated filter selection, focus, slide scanning, and image capture. Isis is capable of imaging up to nine color channels from each sample, which, if used in combinations, can allow for several different types of experiments to be performed, including:

  • color karyotyping, in which differently colored probes are used to identify specific chromosomes of interest;
  • comparative genomic hybridization, where differently labeled normal and aberrant DNA is hybridized to normal DNA, leading to quantitative differences in fluorescence or color at the chromosomal locations of rearrangements and other changes;
  • mFISH, in which every chromosome is labeled with a different color or combination of colors to facilitate analysis of ploidy changes and translocations; and
  • mBAND, where the haploid complement is effectively bar-coded using different colors or combination of colors on metaphase spreads to allow a variety of analyses, especially scans for intra-chromosomal rearrangements like insertions, deletions, and inversions.

FISH analysis isn't limited to metaphase cells. The metafer/MetaCyte system can be thought of as a lower-resolution FISH system that fills a niche somewhere between metaphase FISH and flow cytometry. For example, this scanning image cytometer captures images from several focal planes and automatically measures a variety of morphometric and intensity features of interphase cells. Typical applications include automatic FISH signal counting, automatic measurement of 3-D distances between FISH signals to detect signal co-localizations representing chromosome translocations, and assessment of DNA content. MetaCyte can analyze up to 200 cells per minute and images are captured and stored in a gallery during scanning. Histogram display of the measured features and multi-parameter gating for the selection of sub-populations facilitate the analysis of even large data amounts.

Finally, Applied Imaging (www.aicorp.com) of Santa Clara, Calif., provides fully automated scanning and image analysis systems. Their MDS system (previously reviewed in The Scientist12) provides automated slide scanning using brightfield or fluorescent illumination. Classification of cells and structures and automatic movement back to the coordinates of features of interest is supported. Further analysis of candidate cells can be performed either via the microscope or using digitally stored images on the monitor. The MDS software allows for acquisition of FISH images and for quantitative analysis of FISH fluorescent signals and of staining intensity. Applied Imaging provides a number of options and a variety of platforms, different combinations of applications, and alternative configurations to allow standard karyotyping, FISH, and comparative genomic hybridization, as well as integrations with specific diagnostic probes and systems such as Vysis' PathVysion for HER-2/neu.13

Bob Sinclair (bobsinclair@tech-write-edit.com)is a freelance writer in Salt Lake City.
1. G. Chakravarty et al., "A comparative study of detection of p53 mutations in human breast cancer by flow cytometry, single-strand conformation polymorphism and genomic sequencing," British Journal of Cancer, 74:1181-7, 1996.

2. J.J. Vredenburgh et al., "A comparison of immunohistochemistry, two-color immunofluorescence, and flow cytometry with cell sorting for the detection of micrometastatic breast cancer in the bone marrow," Journal of Hematotherapy, 5:57-62, 1996.

3. B. Planz et al., "Diagnostic accuracy of DNA image cytometry and urinary cytology with cells from voided urine in the detection of bladder cancer," Urology, 56:782-6, Nov., 2000.

4. K. Truong et al., "Quantitative FISH by image cytometry for the detection of chromosome 1 imbalances in breast cancer: a novel approach analyzing chromosome rearrangements within interphase nuclei," Laboratory Investigation: A Journal of Technical Methods and Pathology, 78:1607-13, 1998.

5. K. Kobiki, et al., "Detection of endometrial cancer by flow cytometry using two monoclonal antibodies," Cytometry, 36:150-6, 1999.

6. C.W. Lin et al., "Detection of exfoliated bladder cancer cells by monoclonal antibodies to tumor-associated cell surface antigens," Journal of Occupational Medicine, 32:910-6, 1990.

7. W. Kruger et al., "Immunomagnetic tumor cell selection- implications for the detection of disseminated cancer cells," Transfusion, 40:1489-93, Dec., 2000.

8. H. Iinuma et al., "Detection of tumor cells in blood using CD45 magnetic cell separation followed by nested mutant allele-specific amplification of p53 and K-ras genes in patients with colorectal cancer," International Journal of Cancer, 89:337-44, July 20, 2000.

9. S. Sumi et al., "Preliminary report of the clinical performance of a new urinary bladder cancer antigen test: comparison to voided urine cytology in the detection of transitional cell carcinoma of the bladder," Clinica Chimica Acta, 296:111-20, June, 2000.

10. K.D. Bauer et al., "Reliable and sensitive analysis of occult bone marrow metastases using automated cellular imaging," Clinical Cancer Research, 6:3552-9, Sept., 2000.

11. B. Sinclair, "Deep into that darkness peering," The Scientist, 13[17]:17, Aug. 30, 1999.

12. L.S. Thurston, "Applied Imaging targets tumors," The Scientist, 14[24]:25, Dec. 11, 2000.

13. L. Defrancesco, "Fish or chips," The Scientist,13[5]:16, Mar. 1, 1999.

Bacterial Detectives

BCR Diagnostic's LEXSAS bacteriologic biosensor

Courtesy of BCR Diagnostics

A cancer cell (green) is detected using BCR Diagnostic's LEXSAS bacteriologic biosensor

Bacteria are renowned for their fecundity; given unlimited nutrients and a couple of weeks, the world could be neck-deep in the children of a single bacterial cell. The notion that this phenomenon might be exploited to allow early detection of cancer cells--nature's other prodigious replicators--led Boris Rotman and colleagues, at Brown University, and at BCR Diagnostics of Jamestown, R.I. (www.bcrbiotech.com), to develop a novel assay system, the bacterial chain reaction (BCR).1

In BCR, blood cells are fixed on a microscope slide, washed, and treated with an antibody that is specific to the cancer cell of interest. (The published work used an anti-cytokeratin mouse monoclonal antibody that was known to bind human adenocarcinoma cells). In an almost-standard sandwich-type configuration, penicillinase enzyme was attached to the target cells using anti-penicillinase and bridging antibodies. After a final wash, the slide was overlaid with nutrient agar, containing 3 x 106 Sarcinia lutea cells and 34.5 nM penicillin, and incubated overnight.

Penicillinase coupled to the surface of the cancer cells depleted penicillin in its immediate vicinity, and the S. lutea cells began to grow, producing more penicillinase and further lowering penicillin levels; overnight a visible S. lutea colony developed around the cancer cell (whose presence was later confirmed by microscopy). The system detected one cancer cell in a background of 2 x 107 blood cells.

But in this age of instant gratification, the requirement for overnight growth was unsatisfying. How could the cells be coerced into greater speed? "Bacterial spores germinate in a matter of minutes," notes Rotman. "Our LEXSAS system uses germinogenic substrates in an assay that is actually much simpler than the original BCR." It's faster too, producing a positive result from as little as a single cell in 20-30 minutes.

There are many ways this assay can be set up. In one variant, target particles (cells, viruses, or DNA-coated microspheres) are placed in a 5-picoliter well and coupled to alkaline phosphatase (AP) using an antibody, oligonucleotide, or any other specific ligand. After washing, the well is filled with nutrient media that lacks any germinogenic substrates but contains up to 5000 Bacillus cereus spores and cAMP. B. cereus spores germinate when adenosine is present, but fail to recognize cAMP as a germinogenic substrate; any AP captured in the well will cleave the cAMP, causing the spores to germinate. Thus, germination indicates that a target cell is present in the microwell.

"Spore germination leads to the de novo synthesis of many different enzymes which we can use as indicators of germination," continues Rotman. Germinating B. cereus spores secrete an esterase, which cleaves non-fluorescent fluorescein diacetate included in the germination medium, resulting in an easily detectable, highly fluorescent product.

A patent for LEXSAS should be issued within the next few weeks. BCR's current focus for LEXSAS is the rapid and sensitive detection of pathogenic bacteria in food, but any substance that can be targeted with an AP-conjugated ligand--cancer cells included--should be detectable with the technology.

1. B. Rotman et al., "Sensitivity and specificity of in situ bacterial chain reaction (BCR) in detecting sparse human tumor cells in peripheral blood," Biochemical and Biophysical Research Communications, 229:80-5, 1996.