[CANCER RESEARCH (SUPPL.) 52, 2766s, May 1, 1992]
Diagnostic Research: Breakout Group Report1
James L. Mulshine and Melvyn S. Tockman
Biomarkers and Prevention Research Branch, DCPC, National Cancer Institute, Rockville, Maryland 20850 [J. L. M.J, and Environmental Health Sciences, Johns
Hopkins School of Hygiene and Public Health, Baltimore, Maryland 21205 ¡M.S. TJ
In considering applications of biomarkers for preventive ap
proaches for the control of lung cancer, a critical step is to
specifically define the nature of the biomarker application. The
potential uses of markers include risk assessment, exposure,
early detection, intermediate end point, prognosis, or diagnosis.
Each of these applications imposes different requirements with
regard to formal validation. Since one of the principal objectives
of the Workshop was early detection, considerable discussion
focused on the unique resource requirements for this type of
research. All conventional diagnostic tools detect lung cancer
when métastaseshave already occurred. If one is to validate a
detection tool that detects lung cancer at an earlier stage, then
an archive of serially obtained specimens from high risk popu
lations (including a reasonable number of specimens from sub
jects who developed cancer during the period of the study) is
an absolute requirement. An example of such an archive is the
stored sputum serially acquired during the Johns Hopkins'
participation over the last decade in the National Cancer Insti
tute-sponsored lung cancer early detection trial. In addition to
serially obtained sputum specimens acquired during the 8-year
trial, the clinical follow-up of the study subjects allowed for
unequivocal designation of who did or did not go on to develop
lung cancer.
The Johns Hopkins' archive was acquired as part of a trial
to ask a prospective question. The workshop participants agreed
that having a motivated study group setting up such an archive
as part of a primary investigation was to be preferred over
having a vendor develop an archive as an end in itself.
A number of methodological issues were discussed including
the appropriate end point for an early detection trial. The most
definitive end point was the frequency of developing cancer.
Cancer-free survival was a parameter that was potentially sub
ject to lead-time bias.
The efficiency and economy of this type of research would be
improved by concentrating on very high risk populations and it
was observed that Stage I resected non-small cell lung cancer
patients comprised such a population. This population is also
attractive for intervention trials. Studies of early detection tools
generally should be evaluated separately from intervention trials
to minimize complicating design issues. As more investigators
recognize the unique opportunity provided by this study popu
lation then special provisions will have to be made to ensure
optimal conservation of this clinical trial resource.
An important goal for early detection research to help define
the range of potentially useful markers for early detection
research is to establish which markers are expressed in the
1Report from the NCI Workshop "Investigational Strategies for Detection
and Intervention in Early Lung Cancer," April 21-24, 1991, Annapolis, MD.
injured bronchial epithelium that comprises the field of carci-
nogenesis. A partial list would include markers of proliferation,
cell surface structures, differentiation, carbohydrates, onco-
genes, ploidy status, growth factors or their receptors, and cell
adhesion structures. The markers which define the biology of
the field of carcinogenesis would be likely to have utility both
as early detection markers and as intermediate end point mark
ers. As mentioned before, however, even if a particular marker
is used for multiple prevention-oriented applications, each spe
cific application needs to be validated independently.
As the lung cancer early detection research matures, it is
going to stimulate rational intervention research because con
ventional lung cancer treatments including surgery may not be
the ultimate modality to appropriately manage the very early
lung cancer lesions. An effective early detection tool for lung
cancer mandates the parallel effort to develop more appropriate
intervention tools. The better the intervention tool performs,
especially if the side effects of the intervention are mild, the
less is the demand for precision associated with the application
of the early detection tool.
Summary of Priorities
Resource support to permit the systematic acquisition of
appropriate clinical material to develop tissue archives from
individuals with high risk of developing cancer needs to be
identified. Long term clinical follow-up should be conducted to
permit classification of true cancer status through time. Quality
control standards for the clinical data and the analyses need to
be very rigorous. Investigator participation with the acquisition
of the clinical specimen archive should allow an archive resource
with sufficient quality to expedite the evaluation of new candi
date early detection markers, which were not necessarily envi
sioned as part of the original study design. Quality control
standards need to be defined for a range of parameters including
the optimal format for obtaining early detection specimens
including sputum, urine, blood, and tissue to maximize the
yield of information from those specimen analyses.
A wide range of specimens could be obtained to enable marker
research including any specimen acquired as a part of a clinically
indicated procedure as long as the high risk subject has provided
concurrence. Invasive procedures to obtain tissue for early
detection research should be generally restricted to defined high
risk populations and generally avoid general anesthesia-requir
ing procedures.
The ability to quantitate an end point for early detection
research is highly desirable. Incentives to accelerate the tempo
of such targeted research need to identified.
2766s
on July 17, 2021. © 1992 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from