Abstract
The critical issue in characterizing the usefulness of tumor markers in screening or monitoring for urothelial cancer and in determining prognosis revolves around the question, "What do we need to know?" During surveillance of patients with low-risk disease, highly sensitive markers may detect disease persistence or early recurrence (on the basis of field-effect changes or tumor cell implantation); however, they may not indicate the possibility of progression. If cystoscopy is negative in these instances, is it appropriate to treat? Does earlier treatment actually prevent recurrence or progression? Some investigators have suggested a positive answer to the former question, although lead-time bias may a role. The issue in the latter question may be moot because low-risk disease is by definition unlikely to progress. A corresponding question is whether one should treat the patient if the marker has a low specificity. In such instances, a marker with low specificity may incur more frequent instrumentation and a panoply of additional costs. A marker may have some value in providing earlier diagnosis in low-risk disease if treatments are available to prevent clinical expression and recurrence. A tumor marker would have additional value if it could detect biologic change from low-risk to high-risk disease and permit a corresponding change in treatment. Moreover, if a marker could provide earlier diagnosis in high-risk disease, it might permit identification of those patients likely to progress at earlier and curable stages of disease. To be successful, such markers must have high sensitivity and specificity in diagnosis. It has been suggested that "a simple noninvasive highly sensitive and specific method for detecting bladder cancer would decrease the morbidity associated with current surveillance methods, improve quality of life for patients, and reduce costs by substituting a less expensive test for the more expensive endoscopic procedures" (Dr. M. Soloway, personal communication 2000). In considering this statement, it is tempting to question some of its implications. The term "simple" suggests that point-of-care tests may be the easiest, but tests that are available may be inconsistent and not sufficiently sensitive or specific. The need for a "highly sensitive and specific" test is of critical importance for applicability in the individual, even when such tests are based on the results in large populations. When considering the "morbidity of current methods," it is tempting to question whether cystoscopy is that morbid a procedure. Given the information it provides, it seems improbable that it can be readily replaced. Currently, no markers provide information over and above that provided by cystoscopy and histology- or cytology-based understanding of the biology of the disease especially in considering distinctions between low-risk and high-risk disease. The ability to "improve quality of life" will depend on the use of the predictive value made on the basis of a test's sensitivity and specificity. A false-positive test may lead to anxiety and extensive unnecessary evaluation, whereas a false-negative test may lead to missed diagnosis, delayed treatment, and the possibility of disease progression until accurate diagnosis is obtained. In each case, quality of life is compromised. "Substituting a less expensive test" depends on the benefits of that substitution and the practicality of omitting or delaying standard assessment. Costs depend on the frequency of testing, those incurred by false-positive tests, and those incurred by false-negative tests. Several caveats emerge when taking into account these various considerations. In screening for low-risk disease, the urgency of diagnosis may be less important. High sensitivity is of less importance because the risk of a missed diagnosis has less ominous repercussions. Nevertheless, high specificity is important to avoid unnecessary frequent instrumentations. In considering surveillance for low-risk disease, cytology may be more useful in indicating a change from negative (persistent or recurrent low-risk disease) to positive (high-risk disease) with the consequent need for more aggressive treatment approach. No urinary "screening" markers reliably provide this information. In considering surveillance for high-risk disease, sensitive markers would detect persistence or early recurrence. If such markers are negative, intervals between repeated cystoscopies may be prolonged; however, a high specificity is needed to ensure the validity of a truly negative interpretation. Otherwise, diagnosis may be missed and the risk from progression and the development of incurability increased. Markers for urothelial cancer must be expressed exclusively as a consequence of the presence of cancer cells. The test must be sensitive in detecting disease and must be validated in nonselected populations. It must be sensitive in excluding nondisease, and noncancer conditions must have little influence on its accuracy. There must be little or no interassay or intra-assay variability. Interpretation of an assay for a tumor marker should be all or none, or based on a threshold level. The assays must have sufficient sensitivity and specificity in populations and in the individual to have practical clinical applicability.
Collapse