OPPORTUNITY 3
Detection Technologies
Detecting the Signatures of Cancer Cells
In general, the smaller the tumor at the time of diagnosis, the better
the prognosis. Similarly, the absence of tumor spread at the time of diagnosis
means a more favorable outcome than when patients present with metastatic
disease. Hence, the earlier a patient is diagnosed, the better. Accurate
early detection methods give us a chance to catch a tumor before it has
reached a stage at which effective care is compromised. It has been shown
over and over again that early detection saves lives. Even earlier detection
should therefore save even more lives.
Currently, three major approaches are used to detect cancer. The first
relies on the physical detection of a tumor mass, such as by manual palpation,
endoscopy, or x-ray imaging, as in mammography for breast cancer. The second
perceives an abnormal consequence of tumor development, such as the presence
of blood in the stool--one potential sign of colon 
cancer. The third class of methods, and the newest, relies on detecting
molecular markers of the existence of malignant cells. In the most common
use of this kind of technology, routine blood samples are examined for high
levels of proteins secreted from certain types of tumor cells. For example,
Prostate Specific Antigen (PSA) from prostate cancer cells is a test now
being performed in millions of men. This advance notwithstanding, the vast
majority of tumors go undetected until patients present with clinical symptoms,
and that is often too late to guarantee a good clinical outcome. One reason
for this is that we lack sensitive and specific methods for detecting most
common tumors.
The Goal
Develop new methodologies that will allow tumor detection at the earliest
stage, when the number of tumor cells is small.
The Opportunity
New opportunities exist to make major advances in early detection methodology.
They include detecting solid tumors by testing for proteins secreted by
them but not by their normal cell counterparts. Tumor cells also harbor
certain altered genes that can be detected in body fluids with which they
come into contact, signaling the presence of a nearby cancer. Cancer cells
regularly influence the behavior of both neighboring and distant tissues.
Blood vessels, the kidney, the brain, endocrine glands and other organs
are all susceptible to changes in structure and function as tumors grow.
The proteins secreted by tumors that account for these changes are being
discovered rapidly, making sensitive methods for their detection feasible.
Detecting such tumor products in a blood sample early in the course of disease
could signal the presence of small numbers of tumor cells. New methods of
sensing these tumor cell specific signatures should provide opportunities
to detect tumors at their earliest stages.
The Plan
Much is now known about how certain proteins are secreted by normal and
cancer cells. Secreted proteins can be recognized, in part, because they
carry certain molecular "flags" that denote them as secreted;
moreover, these molecules participate in the secretion process. We can now
use molecular, genetic, and protein assay methods of high sensitivity to
catalog proteins secreted specifically by a given tumor cell type. Individual
proteins in the catalog of a given cell type can then be measured in blood
samples by sensitive protein detection methods in search of those that accurately
mirror the presence of small numbers of tumor cells and their growth. Specific
tests can then be developed for these signals of the presence of a given
tumor type. This approach should be applied to all of the common solid tumors--e.g.,
breast, colon, lung, prostate, ovarian, brain, and bladder. We now have
gene probing systems for detecting extremely small numbers of tumor cells
by virtue of the altered cancer genes they contain. These methods must now
be adapted to accurately and rapidly detect small numbers of tumor cells
in readily obtained samples of tissue, blood, and/or other body fluids.
Numerous protein molecules made by tumor
cells alter the behavior of neighboring, and even distant, organs and tissues.
In some cases, tumors cause such tissues (e.g., blood vessels) to grow and
to serve their nutritional needs. In others, they create the opportunity
for a tumor cell to invade normal tissue and to metastasize. Detection of
these protein molecules would signal the presence of tumor cells in a patient
which could then be localized by clinical methodology. With increasing knowledge
of how these factors work and of how to detect them, it should be possible
to design detection methods that can be used efficiently on fresh tissue
and body fluid samples.
Consequences: Investing vs. Waiting
Experience with many cancers has already provided firm evidence that
it is better to treat cancer early than late. Even for patients with advanced,
widespread disease, treatment is more effective the earlier it is used.
We have known for years that certain tumors, such as testicular cancer in
men and choriocarcinoma in women, have distinct signature molecules circulating
in the blood, and we have been using these characteristic markers to diagnose
and treat disease much earlier than if we had to wait for visible or palpable
lumps. It is logical to infer that all cancers have distinctive signatures.
As we continue to improve therapy for cancers of all types, we must also
develop methods to diagnose cancers at the earliest possible stage to ensure
the best possible outcome for patients. These molecular methods of diagnosis
are now within our grasp. Failure to exploit this opportunity will ultimately
delay the treatment of early cancer and needlessly handicap the potential
for treatment success.
Detecting Cancer Cells Through Diagnostic Imaging
One hundred years ago Roentgen demonstrated that x-rays can be used to
visualize internal structures of the body. Progressive refinements in technique
since then have steadily improved the quality and versatility of the x-ray
picture, so that for many decades now, the diagnostic power of the x-ray
has pervaded the practice of medicine; the chest film, barium contrast studies
of the gastrointestinal tract, and detailed visualization of the coronary
arteries are familiar examples. These and other imaging techniques have
made it possible to diagnose localized abnormalities often before they have
produced irreversible damage. In no field of medicine has the diagnostic
usefulness of the x-ray been more dramatic than in oncology. In many anatomic
locations, cancers too small to be detected by physical examination can
be pinpointed by imaging and treated before spread to distant sites has
occurred. This is why x-ray mammography saves the lives of many women diagnosed
with early breast cancer.
Over the past quarter century, the entire
imaging field has taken a quantum leap forward. Indeed, the practice of
diagnostic radiology has been revolutionized--more dramatically so than
any other area of clinical medicine. A Rip Van Winkle radiologist, awakening
today after a 25-year nap, would be simply astounded by the sheer richness
and precision of the information available in a routine computed tomographic
(CT) scan of the body. He would find that internal organs deep within the
body can be biopsied by long, thin needles guided safely to their targets
by CT or ultrasound scanning; this capability has eliminated in many cases
the need for general anaesthesia and an open surgical procedure. He would
note with particular satisfaction that the crude and often painful techniques
of the past for visualizing the brain and spinal cord, myelography and pneumoencephalography,
have given way to non-invasive, painless, and vastly more informative CT
and magnetic resonance imaging (MRI). And he would see that adaptations
of MRI permit the refined visualization even of the arteries of major organs
without the need for painful and potentially hazardous needle-sticks of
these vessels for the injection of contrast material.
The Goal
Discover and develop techniques that will further increase the precision,
accuracy, and scope of imaging diagnosis and integrate imaging further into
the practice of clinical oncology.
The Opportunity
We already know that several different types of physical processes can
interact with living tissue and produce useful images. X-rays can be collected,
recorded, and analyzed to produce plain images on film or computed tomographic
(CT) scans. Radioactive tracers that seek out a particular organ or structure
(such as a tumor) can yield an image of that organ or structure when the
decay of the tracer is detected by appropriate sensing devices (nuclear
medicine). The responses of tissue exposed to a changing magnetic field
can be recorded as magnetic resonance images. Sound waves of high frequency
can pass through the body and produce images in real time of rapidly moving
or stationary anatomical structures (ultrasound). We still have far to go
to realize the full potential even of the techniques already available to
us.
Consider just two examples:
* Most routine imaging techniques tell us the anatomic extent of an organ
or of an abnormality within an organ. Sometimes the appearance of an abnormality
is so characteristic that it allows us to infer what the abnormality is
(i.e., strongly suggests a specific diagnosis), but most often it does not.
Certain currently available techniques, such as positron emission tomography
(PET) or single photon emission computerized tomography (SPECT) imaging
permit visualization of the physiological or metabolic characteristics of
a tissue, including tumor tissue. These characteristics might include, for
example, the glucose utilization rate or the kinds of receptors covering
the surface of the tumor. Such information may allow an assessment of how
to target particular kinds of therapy to a tumor, or to assess, without
the need for biopsy, how a tumor is responding to a treatment recently administered.
Gazing much further into the future, it is even possible to imagine that
metabolic imaging techniques may eventually be extended to give information
about disruption of cellular signaling pathways or specific patterns of
gene expression.
* The integration of electronic and computational capability to manipulate
images so that they can be better appreciated by people has been most dramatically
illustrated in the pictures beamed back to earth from orbiting satellites
or from interplanetary probes. Extensions of this technology have potentially
profound implications for analyzing and refining patterns detected in medical
images. The application of neural networks to patterns visualized in standard
x-ray mammography has already suggested that neural networks can be "taught"
to distinguish between malignant and non-malignant breast images with impressive
accuracy. Since we already know that mammography saves lives, the implications
of further developing image enhancement and pattern recognition for medical
applications are very compelling.
The Plan
To exploit the opportunities in this area, we propose three new initiatives:
* Metabolic and Physiological Imaging. A major expansion of effort
in this area will greatly improve our capability in the non-invasive physiological
characterization of tumors, in a manner that can be expected to have profound
implications for therapy selection, development, and assessment. Areas of
particular interest include functional MRI, magnetic resonance spectroscopy,
SPECT scanning, and image fusion, in which the data relating to both anatomical
and physiological characteristics of a tumor are combined into one image.
* Pattern Recognition and Image Enhancement. All areas of diagnostic
imaging can be expected to benefit from an intensified effort to improve
how visual or digital information is processed, so that it may be made maximally
informative. The successful development of digital detectors for mammography
is the first step in what will likely be a greatly enhanced ability to acquire,
process, store, and transmit digital information from a variety of imaging
settings. Digitization of imaging information will then facilitate image
manipulation and optimization.
* Integration of Imaging and Therapy.
The influx of technology from defense and electronics applications has raised
the potential of image-guided tumor diagnosis and therapy to a new level.
We will expand existing technology transfer programs between the NCI and
several other Federal agencies that are actively seeking ways to exploit
these technologies for medical uses. This effort will encompass such diverse
possibilities as three-dimensional imaging, intraoperative guidance, applications
of virtual reality for training and therapy, surgical workstation design,
telesurgery, and telerobotics. Liaisons have been established between the
medical and electronics communities so that promising areas of common interest
can be defined.
Consequences: Investing vs. Waiting
These areas are all of interest to imaging scientists in academia and
industry. If NCI is able to capitalize on the opportunities outlined here,
translation of imaging science into clinical reality for people with cancer
and those at risk will occur much sooner than is possible at our current
level of involvement. The formation of productive consortia between academia
and industry will occur much more rapidly if catalyzed by NCI interest and
resources. Our participation will ensure the application of emerging technologies
to the cancer problem.
Imaging advances will bring early and more accurate diagnosis of many
cancers, fewer invasive procedures for patients, and a heightened ability
to monitor tumor response to treatment. Significant advances in imaging
are now possible and will translate directly into larger numbers of lives
saved, but their development will be stunted without NCI leadership and
investment at this important time.