Prostate and Bladder Cancer Screening

Special Article: Urology

Martin M. Goldstein, MD, and Edward M. Messing, MD, FACS

Screening is the process by which asymptomatic people are tested to determine whether they are likely to have a particular disease. Those who are deemed likely by the screening test are evaluated further to determine whether they actually have the disease. Those diagnosed are usually treated for the screened disease. The goal, of course, is to detect and treat the disease earlier than if it were detected only after symptoms had occurred. The screening process is considered successful if these individuals not only are diagnosed earlier (ie, at earlier stages), but also have decreased cause-specific morbidity and mortality rates.

For a screening program to be of value, candidate diseases must meet certain criteria (Table 1). The disease must be a serious health concern. It must be relatively common and occur in a population that can be defined by easily ascertainable demographic and risk factors. There must be some apparent benefit (primarily for outcome, but also for disease- and treatment-related morbidity) for early treatment compared with later treatment. Additionally, the disease must have a preclinical phase of "reasonable" duration (eg, many months to years) during which the disease is detectable but undiagnosed because it has not yet caused symptoms. The interval between when the disease is detected in screened individuals and when it would have been detected if the diagnosis were delayed until symptoms provoked an evaluation is the "lead time." Because screening detects the condition before it would normally be diagnosed, time-interval survival data will favor the outcome of a screened population compared with an historic control unscreened population simply because of the lead time in diagnosis that screening has created, even if screened and unscreened patients eventually die at the same age from the disease. This is called "lead time bias."

Table 1. Bladder and Prostate Cancer: Appropriateness of Screening*

	Prostate Cancer	Bladder Cancer
Common problem	+ + + +	+ +
Associated morbidity and mortality from disease	+ +	+ +
Defined populations at increased risk	– –	+ +
Sensitivity of available screening test	–	+ + +
Specificity of available screening test	–	+
Cost of subsequent workup	– –	–
Do all diseases found need treatment?	– –	+ + +
Effectiveness of available treatment for late-stage disease	–	+
Effectiveness of treatment for early-stage disease	+	+ + +
Ability of available screening to detect the disease when still curable	+ ?	+ + +
Lower morbidity and expense of treatment for early compared with advanced disease	– – –	+ + ++

*+, reason to consider the disease suitable for screening (the more +'s, the more suitable); –, reason to consider the disease inappropriate for screening.

Length bias sampling is the tendency for screening tests to detect more slowly growing cancers because they are in the asymptomatic population longer than more rapidly growing ones, which quickly become symptomatic and no longer need screening to become detected. Length time bias can also give an impression of increasing survival in people with screening-detected cancers. These biases are important arguments against accepting the shift to earlier stages as proof that screening improves disease outcome (Fig. 1).

Sensitivity	=	n with a negative test n with with a disease and either a positive or negative test	=	true positives true positives & false negatives
Specificity	=	n without a disease and a negative test n without a disease and either a negative or positive test	=	true negatives true negatives & false positives
Positive predictive value	=	n with a positive test who have the disease n with a positive test whether they have the disease or not	=	true positives true positives & false positives
Lead time	=	the interval from detection because of screening to the time at which diagnosis would have been made without that screening (ie, when symptoms or signs would have provoked evaluation)
Length bias sampling	=	the tendency of a screening test to preferentially identify indolent disease with a long preclinical phase

Figure 1. Screening definitions.

PROSTATE CANCER SCREENING

Prostate cancer is the most frequently diagnosed noncutaneous cancer in American men1 and is a leading cause of cancer death in western society. Both the incidence and mortality rates of prostate cancer have been increasing in recent years, with incidence increasing far more rapidly than mortality.2,3 It has not yet been determined whether screening improves patient outcomes (ie, reduces mortality due to prostate cancer). Although the advantages are still being studied, advocates against prostate cancer screening maintain that the expense, anxiety, and morbidity of the screening tests, and of the subsequent workup and treatments they engender, may outweigh the benefits.

When prostate cancer screening programs are used routinely, the pathologic stages of the treated disease are lower (ie, earlier) than those found in historic controls.4,5 It is well established that early-stage prostate cancer responds far better to treatment than do advanced cancers, which generally have unfavorable outcomes.6 Screening increases the proportion of organ-confined prostate cancer from 35% to 70%,7 which theoretically should save lives and avoid future cancer-related deaths. Although the desired stage shift has occurred, there has not yet been any proven decrease in morbidity and mortality. The reasons for this could be as follows: Occult metastases are already present but not detectable at diagnosis (ie, the lead time is not long enough), current treatments even of localized disease are not invariably effective, or the disease normally grows so slowly—and ad lib or organized screening is such a recent phenomenon—that not enough time has elapsed to document the reduction in mortality that widespread screening will effect over the next several years. Although all three factors are probably involved, the last is most likely the most important.

Physical examination

The first prostate cancer screening test, the digital rectal examination (DRE), by itself is a relatively poor screening modality. Many patients now being diagnosed with organ-confined disease (and even extraprostatic disease) on the basis of prostate-specific antigen (PSA) testing have had DREs that would not have prompted a biopsy. In fact, it has been reported that a DRE screening program alone failed to increase the proportion of organ-confined cancers detected.8 It is best to use PSA and DRE as complementary tests. Catalona and associates4 reported that DRE alone would have missed approximately 40% and PSA alone would have missed about 25% of cancers detected in the first wave of a prostate cancer screening trial, but the two tests combined increased the prostate cancer detection rate to 78%.

Transrectal ultrasonography

Transrectal ultrasonography (TRUS) with 5- and 7-mHz probes is not much more unpleasant than DRE and can evaluate areas of the prostate not easily reachable by an examiner's finger. Additionally, it is an accurate means of assessing prostate size. Despite earlier contentions supporting its use as a screening tool, its inability to distinguish normal from malignant tissue is disappointing. Currently, TRUS is used primarily to guide biopsy needles into specific regions of the prostate and adjacent structures.

Prostate-specific antigen

Prostate-specific antigen is a serum protease produced by prostatic epithelial cells and secreted into prostatic fluid. Some PSA also enters the systemic circulation. Serum PSA has been used with increasing frequency as a screening tool for prostate cancer. The traditional upper limit of normal has been 4 ng/mL. More than 70% of patients with what is believed to be biologically important prostate cancer that is organ-confined have serum PSA levels > 4 ng/mL, as compared with only 25% of patients with pathologically proven benign prostatic hyperplasia (BPH).9 In studies of volunteers, PSA values > 4 ng/mL are seen in 8-15% of American men between 50 and 70 years of age.10 The proportion of men with elevated serum PSA concentrations increases with increasing age.5

The positive predictive value (PPV) of a screening test is the proportion of those individuals who have a positive test who are found by diagnostic evaluation to have the disease (Fig. 1). The PPV of a PSA value that is elevated in a man aged 50-70 years is dependent on the degree of elevation. For levels between 4 ng/mL and 10 ng/mL, the PPV is approximately 25%. The PPV increases to 64% if the PSA is > 10 ng/mL.7,10 If the PSA level remains elevated and men who were not found to have cancer are rebiopsied 6-12 months later, the PPV increases to 34% for PSA of 4-10 ng/mL and to 70% for PSA > 10 ng/mL.11

The widespread use of PSA screening has led to an increase in the number of prostatic ultrasounds and biopsies performed. The number of cancers detected has increased and the stages of the cancers detected are lower (ie, earlier) than were cancers detected in the "pre-PSA" era.12 Indeed, those in whom cancer is detected under circumstances of repeated PSA screening have organ-confined disease nearly 80% of the time.4 Whether PSA screening effects a sufficiently early lead time to detect cancer before micrometastases occur is not yet known.

Catalona and colleagues7 performed followup studies in men who at initial screening had PSA values < 4 ng/mL. After up to 5 years, 2.6% had prostate cancer found on subsequent screening. The proportion of men with clinically advanced cancer was significantly lower among those with cancers detected through serial PSA testing (2%) than in those with cancers detected during initial screening (6%).13 This shows the value of serial screening with PSA even after an initially "normal" value.

Several additional concerns about prostate cancer screening with serial PSA levels have been raised. First, the test is not ideally sensitive. In several large screening series, roughly 20-30% of men with seemingly biologically important prostate cancer (based on histologic grade and cancer volume) still confined to the prostate (the patients one would most like to detect by screening) had PSA levels < 4 ng/mL. Because of this relative lack of sensitivity (true-positives/true-positives and false-negatives), PSA testing should be combined with DRE if prostate cancer screening is to be performed. Indeed, 80% of men with PSA levels < 4 ng/mL who have biologically important prostate cancer have cancer confined to the prostate gland, once again emphasizing the importance of not relying exclusively on PSA.

Second, the disease occurs in a population (aging men) in whom another entity that can elevate the PSA modestly—BPH—is ubiquitous. This lack of specificity (true-negatives/true-negatives and false-positives) leads many men who do not have prostate cancer, but who do have elevated PSA levels, to go through the expense ($692 in direct costs currently in Rochester, NY), morbidity, unpleasantness, and obvious anxiety that an elevated PSA provokes.

A theoretical concern raised about PSA screening for prostate cancer is that the test may be "too sensitive." Screening may be detecting the tiny foci of well-differentiated cancer that many middle-aged and elderly men have in their prostates and that are found incidentally at autopsy.4,7 It has been estimated that currently, ad lib screening still detects only 20% of all men with some adenocarcinoma in their prostates. Yet, even with this seemingly small number, it is possible that we are detecting the disease and possibly treating some men who are likely never to experience any ill effects from prostate cancer. Advocates for PSA screening have argued that such "autopsy" cancers are so small that they would not be expected to elevate serum PSA or to be detected by random needle biopsies. It is possible, though, that if the PSA is elevated from BPH, a biopsy may incidentally detect an "autopsy" cancer that would be indistinguishable from the more important large-volume cancers. If even a small proportion of the enormous pool of men with "autopsy" cancers were being diagnosed (and treated) with PSA screening, this would be a strong argument against implementing wide-scale screening. Currently, the evidence indicates that this is not happening. The distribution of histologic grades appears very similar in contemporary PSA-screened populations and those in the pre-PSA era (15-20% well-differentiated, 50-60% moderately differentiated, and 20-35% poorly differentiated tumors). One would expect a significant shift to well-differentiated tumors if PSA screening were enabling the detection of a large percentage of the "autopsy" cancers.

Despite these considerations, and assuming the patient is medically stable and willing to undergo treatment for prostate cancer, current standards of care are to proceed with TRUS-guided biopsy for PSA levels > 4 ng/mL or for suspicious DRE results. Yet with serum PSA, the high false-positive rate has been another major criticism. To improve the specificity of PSA (ie, to reduce the number of false-positives), alternative ways to analyze the PSA level to assess its clinical significance more accurately have been suggested (Table 2).

Table 2. Tests to Improve Specificity of PSA*

	Advantages	Disadvantages
Age-adjusted PSA	Considers that BPH increases with age and accepts that detection of disease in older men is less "valuable" than in younger men	Significant increase in biopsies for younger men; assumes similar PSA range for different races
PSA velocity	Useful for individuals with numerous PSA values over several years; may also detect cancer in patients whose PSA is < 4.0 ng/mL	Requires multiple PSA values performed by the same assay technique; requires testing over prolonged intervals
PSA density	Directly limits the effect of BPH	Inaccurate volume determinations using standard TRUS technique; expense and inconvenience of TRUS
Free PSA	Earlier cancer detection; eliminates PSA elevations due to BPH	Limited data at present on influence of noncancerous conditions

*PSA, prostate-specific antigen, 4-10 ng/mL; BPH, benign prostatic hyperplasia; TRUS, transrectal ultrasonography.

PSA density. One way to diminish the high false-positive rate associated with BPH is to use the PSA density: the ratio of serum PSA to the prostate volume, as measured by TRUS.14 Addition of this second test could reduce by approximately 50% the number of negative biopsies currently done, but nearly half of the organ-confined prostate cancers—the ones most desirous to find—would be missed as well.15 Moreover, because TRUS is needed to determine PSA density, the expense and inconvenience of TRUS would still be necessary. For these reasons, PSA density is rarely used to determine who should undergo biopsy.

Age-specific PSA. It has been suggested that because prostate volume and quantity of BPH increase with age, the 4-ng/mL cutoff may be too high for younger men (40-50 years) and too low for older men. Oesterling and coworkers16 proposed that the PSA threshold for prompting biopsy should change based on age, with 2.5 ng/mL for men aged 40-49 years, 3.5 ng/mL for 50-59-year-olds, 4.5 ng/mL for 60-69-year-olds, and 6.5 ng/mL for 70-79-year-olds. Catalona and associates17 reported instead that decreasing the PSA cutoff in younger men resulted in a significant increase in the number of biopsies performed, with only a minimal increase in cancer detection. Additionally, higher cutoffs in older men would result in fewer biopsies being done, but many organ-confined cancers would go undiagnosed.17 Age-specific PSA ranges also have varied with different races; the original age-specific PSA ranges were determined from the almost exclusively Caucasian northern European population that inhabits Olmstead County, MN. These ranges cannot be readily extrapolated to non-Caucasians or to Caucasians of other ethnic backgrounds. Finally, it is relatively difficult to assign men at either end of each age grouping a proper PSA cutoff. Despite these problems with using age-specific PSA values for individual patients, they may have value in large population-based screening efforts by decreasing the number of false-positive tests.

PSA rate of change. Another frequently used criterion for selecting men to biopsy is the rate of change of an individual's PSA level ("PSA velocity"). Sudden increases in PSA concentrations from an established baseline or slope suggest neoplastic transformation, and biopsies should be taken at that time. Carter and colleagues,18 using a population survey with retrospective analysis of at least three PSA determinations over several years, found that a rise of > 0.75 ng/mL annually could distinguish prostate cancer from BPH in most instances. Changes found at intervals < 12 months usually were too unreliable to extrapolate a rate of change over 12-24 months; this limits the utility of PSA velocity to the relatively few individuals having numerous PSA determinations over a span of several years, each performed by the same assay method.

Free and total PSA. One of the most promising developments for improving PSA's specificity has been the recognition that PSA in the circulation of men with prostate cancer tends to be bound by alpha1-antichromotrypsin and other proteins. Assays have been developed that can distinguish the amount of PSA in the circulation that is unbound from the total amount present. The proportion of free PSA (unbound PSA/total PSA) is greater in men with BPH than in men with cancer. Several groups have claimed that these calculations improve specificity without greatly reducing sensitivity and have advocated performing biopsies not on men whose total PSA is between 4 and 10 ng/mL, but on those whose free PSA is below a specific percentage. For example, using a free-PSA cutoff of 23% in men with a palpably "benign" gland would detect 90% of the cancers and eliminate 31% of negative biopsies.19 Others, using slightly different assays, have created receiver operating characteristic curves to determine free/total PSA proportions, which capture 90-95% of cancers detected (with the total PSA in the 4-10-ng/mL range). These investigators have estimated that a similar proportion (20-40%) of negative biopsies could be avoided.

Besides trying to improve the specificity of PSA in detecting prostate cancer, Catalona and associates20 have attempted to use percent-free PSA (<27%) to detect cancer even earlier without increasing the number of negative biopsies. If the threshold for biopsy were simply a total PSA of 4 ng/mL, then >20% of men with large volumes of localized prostate cancer would go undiagnosed, at least until their total PSA level rose above 4 ng/mL. If all men with a percent-free PSA of < 27% were biopsied (regardless of total PSA), most of the cancers associated with PSA levels < 4 ng/mL would be detected, as would all of those with total PSA > 4 ng/mL. Additionally, with this new index, 80% of the cancers would be organ-confined as compared with 70% using the 4-ng/mL threshold to trigger biopsy. In essence, use of this assay would eliminate the need for DRE as part of the first-line prostate cancer screening test (which is now total PSA and DRE). Currently, only a small number of subjects without cancer have been tested to calculate reliably the appropriate cutoff percentages, and almost no men with prostatitis or other nonmalignant prostate diseases have been studied to know the effects of these pathologic conditions on this ratio. The exact value of the free-PSA calculation in improving total PSA specificity (for values > 4 ng/mL) or sensitivity (< 4 ng/mL) remains unconfirmed in general clinical use.

Increasing the efficiency of screening

Another way to increase the predictive value of a screening test is to increase the prevalence of the disease among those being screened. Screening is not recommended until age 50 because of the low prevalence of prostate cancer in younger men. In the African-American community, there is a 31% higher incidence rate of prostate cancer and
a 117% higher mortality rate from this disease.21 This higher mortality may occur because African Americans have metastatic cancer twice as often as whites at the time of diagnosis.22 A stronger argument can be made for the selective screening of African Americans than for the general public because of the greater prevalence of disease in the former group. In fact, many are advocating PSA and DRE screening starting at age 40 for African Americans. Asians have a lower incidence of prostate cancer and may not be appropriate subjects for screening.

An additional population with substantial risk is first-degree relatives of men with prostate cancer. Men whose father or brother has prostate cancer are twice as likely to develop prostate cancer themselves as men with no relatives affected. Likewise, there is increased risk with a greater number of affected family members.23 Many are suggesting that this population be screened at age 40 as well, particularly if the affected relative was young when diagnosed.

Management of cancers detected by screening

For screening to be of value, there must also be effective treatment for early-stage disease at diagnosis. A final reservation about screening is that not all therapies are always effective, and all are associated with some morbidity and considerable expense. Although it is beyond the scope of this article to discuss in depth the therapies for localized prostate cancer and their associated morbidities, it is relevant to remember that most men with prostate cancer die with the disease, not of it, and that many patients who receive nothing but palliative treatment have very indolent disease courses. A variety of observational studies have demonstrated low (eg, 10-20%) 10-year cause-specific mortality rates in men with localized disease. These studies have been criticized, however, for lacking patients with aggressive (high-grade) tumors and for overrepresenting those with low-grade ones. They primarily have included elderly patients with considerable comorbidities (ie, 50-60% of the subjects died of non-prostate cancer causes within 10 years).24 Perhaps the best evidence of how men fare with untreated prostate cancer is a retrospective analysis by Albertsen and colleagues.25 In that study, 65-year-old men with well-differentiated cancers had the same life expectancy as the average 65-year-old man in Connecticut (> 15 years). Those with moderately and poorly differentiated cancers had their lives significantly shortened. Because these higher tumor grades encompassed > 80% of all men with prostate cancer in that study (and who are currently diagnosed with the disease), one could conclude that definitive treatment should be offered to men with life expectancies of > 8-10 years because their outlook beyond that point, if they are not treated definitively, is quite poor.

Currently, no unequivocal recommendation can be made for prostate cancer screening. If it is carried out, it should probably be restricted to men with a >8-10-year life expectancy who would be willing to accept the morbidities of curative treatment were prostate cancer diagnosed. Randomized, prospective studies are underway to evaluate the efficacy of PSA and DRE screening and should address sampling issues of lead time bias and length bias. It will be another decade before results are in from such trials in terms of the impact of screening on prostate cancer mortality. Until that time, national mortality trends may be helpful in predicting screening's benefits. Because PSA screening became popular in the early 1990s, reduction in mortality from this disease should be apparent by 2000-2002 if screening is effective. If this is found, then the expenses and morbidities that screening creates will have to be analyzed carefully to determine whether it has sufficient value for the general population.

BLADDER CANCER SCREENING

Bladder cancer is the fourth most common malignancy in males and the eighth most common in females in the United States. An estimated 54,500 new cases of bladder cancer were expected to be diagnosed in 1997 in the United States.26 Nearly 95% of these are transitional cell carcinomas (TCCs). Prognosis is dependent on the grade and stage at the time of presentation. Superficial tumors are treated relatively easily with local therapies and have an excellent prognosis. Once the tumor invades the muscle, the prognosis is quite poor despite aggressive treatment.

Almost all bladder malignancies originate on the urothelial surface. The overwhelming majority of patients who die of the disease have metastases, and nearly all patients with metastatic disease have concomitant or previous muscle-invading tumors. Of all patients who have TCCs that have invaded the detrusor muscle, nearly 90% have
muscle invasion at the time of their initial diagnosis.27 If screening methods could detect bladder cancers destined to become muscle invading while they are still superficial (and therefore amenable to successful therapy), it is likely that a significant reduction in morbidity and mortality would result.

The prognosis of treated bladder cancer

Although TCCs can exhibit a wide spectrum of biologic aggressiveness ranging from low-grade superficial to high-grade invasive cancers, they generally have a distinctively dichotomous behavior. Low- to moderate-grade (grade 1, 2) superficial (stage Ta, T1) lesions frequently recur after "complete" endoscopic resection, but rarely invade; high-grade (grade 3) tumors usually are already muscle invading (stage T2+) at the time of diagnosis. It is almost exclusively the high-grade TCCs that are associated with substantial morbidity and mortality.

Low- and moderate-grade superficial TCCs account for approximately 55% of newly diagnosed cancers. Easily treated with cystoscopic resection, they recur in approximately 50% of patients,28 but progression to muscle invasion in subsequent recurrences is rare. High-grade TCC accounts for 45% of newly diagnosed TCCs. Roughly 40% of newly diagnosed high-grade cancers are confined to the epithelium (Ta) or lamina propria (T1); nearly 60% are muscle invasive (T<inf>2<reset>) or beyond when diagnosed. The high-grade superficial tumors progress to muscle invasion far more often than grade 1 and 2 TCCs do on subsequent recurrence. Use of intravesical immunotherapy Bacille Calmette-Gu|ferin (BCG) significantly decreases the rates of recurrence and progression of superficial TCC once it is resected transurethrally, even for high-grade tumors. Before the use of BCG, one representative study showed that stage Ta, T1 tumors progressed to muscle-invasive disease 2% of the time for grade 1, 11% for grade 2, and 45% for grade 3 lesions.29 The use of intravesical BCG has reduced the subsequent progression ratios, particularly for the high-grade superficial TCCs, which progress to muscle invasion roughly 25% of the time when treated with endoscopic resection and BCG.

Approximately 50% of patients with muscle-invasive or more extensive disease have occult metastases at diagnosis. Nearly all patients with metastatic disease die of bladder cancer within 2 years despite the administration of multidrug, highly toxic chemotherapy. Those who have muscle invasion without evidence of metastatic spread require either partial or radical cystectomy and systemic chemotherapy with or without radiation therapy. All of these treatments are associated with considerable morbidity and risks. Thus, screening that would permit earlier detection should not only afford patients a more favorable prognosis, but also potentially avoid much of the morbidity associated with treating advanced disease.

Rationale for early detection

Excellent cure rates are achieved for stage Ta and T1 TCC, even for high-grade disease. Once the disease is muscle invasive, prognosis is poor. Unfortunately, approximately 90% of patients with muscle-invasive disease will be found on the initial bladder cancer presentation27 and do not come from the pool of patients with recurrent superficial TCCs. Because all of these tumors originated in the urothelium, the opportunity exists to detect those destined to invade muscle before they actually do so. With current detection strategies, primarily evaluating hematuria detected
by patients or discovered on a urinalysis done for unrelated reasons, this is not often accomplished. Because more effective therapies for metastatic TCC are not on the immediate horizon, early detection strategies will need to be improved and implemented if we are to reduce significantly morbidity and mortality from this disease.

A common argument against screening for any chronic disease is that one may detect and treat a disease that, had it not been detected, would not have led to any adverse sequelae. This has been an important concern about screening for prostate cancer. This is not the case with bladder cancer, however, which is almost never found incidentally on autopsy.30 This almost always means that as opposed to prostate cancer, bladder cancer has a fairly brief preclinical duration or potential lead time (the interval between when it could be diagnosed if you looked for it and when it is diagnosed because of symptoms). Before a patient dies, symptoms (primarily hematuria) eventually lead to a diagnosis. Patients in whom bladder cancer is detected through screening would not undergo any unnecessary tests or treatments, only earlier ones.

Screening for hematuria

Practically all bladder cancers will cause hematuria at some point,31 but hematuria is often intermittent, even when caused by serious disease. Repeated testing for hematuria is needed, and once one specimen is positive, further evaluation should be done to determine its cause. Screening for hematuria can be done by microscopic urinalysis or by using a chemical reagent strip to test for hemoglobin. False-positive results can be found with reagent strips in conditions such as hemoglobinuria or myoglobinuria, but in general, the strips are a very accurate reflection of microscopic urinalyses for detecting hematuria.32

The importance of complete urologic workup for asymptomatic hematuria, including intravenous urography and cystoscopy, has been known for some time. In 1956, Greene and coworkers,33 and subsequently many others, confirmed that important lesions are identified in 13-20% of patients who are referred for asymptomatic microscopic hematuria, and cancer was found in 6.5-13% of patients.34-36 Two screening studies of the general population using hematuria home testing have been published. Messing and colleagues37-39 and Britton and associates40,41 solicited from general-practice patient-care rosters all middle-aged and elderly men residing in geographically defined regions [south central Wisconsin37-39 and Leeds, England]40 who were not believed to have urologic malignancies or other known causes of hematuria, and requested them to test their urine 10-14 times at home with reagent strips for hemoglobin. If the result was positive even once, the subjects were asked to undergo formal urologic evaluation including intravenous urography, cytology, and cystoscopy. The two studies had similar findings. Fifteen to 20% of the screened populations had hematuria. Of those who completed the workup, 6-8% were found to have urothelial cancers. Overall, 1.2-1.3% of those screened had bladder cancer diagnosed. Neither study had a prospective, randomized control population.

The Wisconsin study looked at state tumor registry data to compare the screening participants' outcomes with those of an unscreened population. Additionally, pathology materials from all men screened in whom bladder cancer developed and from all men age > 50 years old in Wisconsin (not screened) in whom bladder cancer developed during 1 year of the study were compared by a referee pathologist. In both populations, roughly the same percentage had low-grade (grade 1, 2) superficial bladder cancers (56.8% unscreened, 52.4% screened). Approximately 45% of the cancers (43.2% unscreened, 47.6% screened) were high grade. The proportion of muscle-invasive tumors was significantly higher in the unscreened population: 23.9% of all cancer registry patients versus 4.9% of all screened cancers. By 24 months after diagnosis, 16.4% of the unscreened patients had died of bladder cancer. In contrast, none of the 21 men participating in the screening study in whom bladder cancer was detected has died of bladder cancer after 4-9 years of followup.42 It would seem that screening allowed the successful diagnosis of cancers destined to become muscle invasive at preinvasive stages. Compared with other diseases for which screening has been accepted as beneficial and worth the expense (eg, mammography for breast cancer in postmenopausal women, fecal occult blood testing or colonoscopy for colorectal cancer, and blood pressure checks for hypertension), bladder cancer screening with chemical reagent strips appears to be quite cost-effective.43

Because this was not a prospective, randomized, controlled study, certain biases must be considered.37-39 Lead time bias (discussed earlier) could have contributed to the decreased mortality seen in the screened group. The followup time for the screened population was 4-9 years, though, as compared with 2 years for the unscreened population. Likewise, it is unlikely that in the screened population, more indolent tumors with longer preclinical durations were detected (length bias sampling), as the distributions of the grades of cancers detected were similar in both groups. The worse outcomes in the unscreened population could be explained if they received less effective therapy than the screenees. The unscreened patient outcomes, however, paralleled those reported in contemporary
series of optimum treatment for similar tumor stages and grades, so inferior therapy in the control population does not appear to have been involved. The control group was not randomized, and its subjects may have differed from the screenees in terms of their health consciousness or other risk factors. This screening method has other problems as
well. Although the test's sensitivity approaches 100% for cystoscopically detectable tumors,37-39,44 the specificity is only 8% and the PPV is also only 8% (although the PPV is 11% for all malignancies and 33% for serious disease).

In summary, in the absence of a prospective randomized trial, the available information strongly suggests that screening shifts the diagnosis to an earlier noninvasive stage, with a resulting decrease in mortality. If other screening tests could be used to reduce the number of false-positive results (thereby reducing the number of negative workups), both cost effectiveness and public and physician acceptance would be enhanced even further.

Other bladder cancer screening methods

Cytology and DNA flow cytometry. Another possible method for early detection of bladder cancer is cytologic analysis of exfoliated cells found in urine. Although easy to perform, this test lacks the sensitivity to be of practical use in widespread screening. In particular, cytology fails to detect well- or moderately differentiated tumors. Although these low-grade lesions are rarely life threatening, if a screening modality cannot detect these cancers, it would undermine the confidence of patients and physicians in the test. Flow cytometry is more sensitive than classic cytology but also lacks sufficient sensitivity to detect low-grade tumors.

Genetic abnormalities. Chromosomal abnormalities in bladder cancers have been identified with fluorescence in situ hybridization using labeled DNA probes specific for deleted or amplified tumor marker chromosomal regions. Anomalies involving chromosomes 1, 3, 5, 7, 9, 10, 11, 13, and 17 have been reported from examining bladder cancer specimens or exfoliated cells,45 and identifying these anomalies by fluorescence in situ hybridization is being used as an adjunct to cystoscopic examination in the followup of patients with histories of bladder cancer. Recently, a cosmid-directed probe for detection of small genetic anomalies in the p21 region of chromosome 9 has been used on cells collected by bladder irrigation. This has led to an important improvement in sensitivity45 and may become a useful wide-scale screening modality.

Another promising development for bladder cancer screening has been the use of microsatellite analysis of DNA in voided urine.46 Microsatellites are repeating, small DNA sequences that are unique to each individual and have very low inherent mutation rates.47 With mechanisms of DNA repair disrupted in malignant cells, DNA mutations can be detected in exfoliated tumor cells found in the urine of patients with TCC. Polymerase chain reaction is used to amplify the DNA recovered from urine on a preselected panel of microsatellite loci and chromosomal regions commonly deleted in bladder cancer. These are compared with the same loci in "normal" DNA in peripheral white blood cells. Mao and coworkers46 reported that using 13 DNA markers (which included the loss of heterozygosity of chromosome 9), tumors were detected in 19 of 20 patients (95%) with bladder cancer; cytology was positive in only 50%. Equally promising results were seen using 20 microsatellite markers in patients undergoing surveillance for recurrent bladder tumors.48 As with cytology and DNA image analyses, false-positive results in patients without TCC were quite rare. This technique has not been tested on populations without histories of bladder cancer, in which only slightly > 1% would be expected to have the disease.

Marker antigens. Antibodies can be used to detect tumor-associated antigens found on exfoliated urothelial cells. When combined with cytology or image analysis, the technique of immunologic staining potentially is an excellent way to detect premalignant or malignant cells. The Lewis X blood group-related antigen is normally absent from adult urothelial cells, except for occasional umbrella (superficial transitional epithelial) cells, but bladder tumor cells have enhanced expression of the Lewis X antigen. For screening purposes, its expression is independent of tumor grade or stage.49 Immunocytologic staining for Lewis X antigen on two voided urine specimens in patients with cystoscopically evident bladder tumors (where a positive test required one of two specimens to stain positively) had a sensitivity of 95%, specificity of 85%, PPV of 76%, and negative predictive value of 98%49 in a group of patients with a high rate of bladder cancer (32%). It remains to be seen whether these excellent numbers would be obtained in a screening population with a much lower expected incidence.

The antigen M344 is expressed on approximately 70% of superficial TCCs. Staining for this antigen has been used to help increase sensitivity for the low-grade superficial tumors often missed by standard cytology.50 Another antigen, DD23, is absent from normal urothelium and is found on 80% of bladder tumors regardless of grade or stage.51 A third antigen, T138, is detectable with advanced-stage lesions as well as with aggressive superficial
cancers.52 To date, no antigen described is either extremely sensitive or tumor specific. Were these antigens to be used for screening purposes, a combination would likely be required to increase sensitivity.53

Growth factors, their receptors, and other biomarkers involved in urothelial carcinogenesis. Soluble factors found in urine and their receptors on urothelial cells have been studied as possible screening tools. The expression of
epidermal growth factor receptors (EGFRs) is restricted to the basal layer of transitional epithelial cells in normal urothelium and nonmalignant disease. In both low- and high-grade bladder tumors, these receptors are located throughout all cell layers. In patients with TCC, this transepithelial expression of EGFRs occurs throughout the
urothelium, even in regions remote from bladder tumors. The density of EGFRs in tissue also correlates with grade and aggressiveness of bladder cancer54 and is a predictor of survival and stage progression.55 Concentrations of EGF are lower in voided urine of patients with TCC, possibly because of the increased density and widespread urothelial expression of EGFRs. Measurement of EGF in voided urine and detection of EGFR on urothelial cells, already important in predicting outcomes, may have additional value in screening.

Urinary autocrine motility factor is a soluble cytokine that is present in the urine of patients with bladder cancer in higher concentrations than in healthy controls.56 It is likely to be involved with cellular motility, a property required for tumor invasion and angiogenesis. Analysis of autocrine motility factor and its urothelial receptors may become an important future screening tool.

The Bard Bladder Tumor Antigen (CR Bard, Inc., Redmond, WA) test measures urinary complexes of fragmented basement membranes (presumably from enzymes originating from the tumor, which degrade the basal lamina). This commercially available test showed moderate sensitivity for grade 2 tumors (71%) and exquisite sensitivity for grade 3 tumors (100%) as well as carcinoma in situ. Its drawbacks include limited sensitivity for low-grade tumors and a high incidence of false-positives resulting from infection or trauma (iatrogenic).57

The Nuclear Matrix Protein 22 test (NMP-22; Matritech) (Matritech, Inc., Newton, MA) is an immunoassay performed on 24-hour urine collections that measures a nuclear protein that forms part of the chromatin lattices on which DNA lies. Unaffected by pyuria or trauma, this test has been shown to be effective in predicting recurrence in
patients shortly after local surgical resection (TURBT).58 It is likely to be used for surveillance primarily, and its role in screening for the general population remains unclear. As with the Bard Bladder Tumor Antigen test, sensitivity for well- and moderately differentiated TCCs is not sufficient to use it as an independent
screening modality.

The AuraTek FDP test, a rapid urine dipstick immunoassay test for the detection of fibrin/fibrinogen degradation products, has had similarly positive results for all grades and stages of TCC. Its low cost and ease of use may make this an important instrument for bladder cancer screening, but its sensitivity is only 68%, a low value for a sole screening test (Schmetter et al, unpublished data). It should be noted that none of these tests (Bard Bladder Tumor Antigen, NMP-22, AuraTek FDP) are yet used routinely in clinical practice, let alone for screening patients without histories of bladder cancer. It remains to be seen how well they will perform in large populations, in which bladder cancer is present in only a small number of subjects.

Future perspectives for bladder cancer screening

Although screening for bladder cancer seems rational and effective, this cannot be proved without a prospective study of screening with randomized unscreened controls, using outcomes data with death from bladder cancer as the end point. If screening is shown to save lives, then perhaps the most cost-effective method will be combining one or more tests with hematuria home reagent-strip testing (ie, evaluating with cystoscopy those subjects who are positive on hematuria testing as their first screen who then have one or a combination of other tests also positive) (Table 3).

Table 3. Bladder Cancer Detection*

	Sensitivity	Specificity	PPV	Low-Grade Detection	Cost/ Complexity
Home hematuria	+ + +	–	–	–	+ + + +
Cytology	– – –	+	+	–	+ + +
Flow cytometry	+	+	+	–	+ +
Immunocytologic staining with antigens	+	+	+	+	+
Growth factors and receptors	?	+	–	+	+
Bard Tumor Antigen Test	–	+	–	–	+
NMP-22	–	+	–	–	+
FDP test	–	+	–	+	+
Microsatellites	+ +	+	+	+	– –

*+, favors use of this test in screening (the more +'s, the more favorable); –, favors not using this test in screening. PPV, positive predictive value; NMP, Nuclear Matrix Protein.

It should be remembered that for a test with a given sensitivity and specificity, the efficiency of screening increases with the disease's prevalence in the screened population. For example, the PPV would be improved by limiting the screening program to people at high risk. For bladder cancer, this may mean restricting participation to people with
occupational exposure to known bladder carcinogens. If this is done, only 10-15% of people who die of the disease would be involved. Alternatively, limiting screening to men aged > 50 years with smoking histories would significantly reduce the numbers to be screened and still would include the considerable majority (65%) of people who
eventually die from it.

CONCLUSION: PROSTATE AND BLADDER CANCER SCREENING

Morbidity and mortality from both prostate and bladder cancer depend directly on the stage of cancer at the time of diagnosis. For prostate cancer, survival is best while the disease is localized within the gland, and for bladder cancer, survival is marginal once there is muscle invasion deep to the lamina propria. The goal of screening is to detect the tumors in their presymptomatic phase to provide early treatment, presumably effecting a decrease in morbidity and mortality. Although its biologic behavior may make bladder cancer more suitable for mass screening, it is far less common than prostate cancer and is responsible for less than one third the number of cancer deaths.

Although the debate about screening for prostate cancer is intense in the United States, screening with DRE and PSA is currently performed widely on an ad lib basis. Screening for bladder cancer has not been widely implemented. Although not randomized, the evidence available from studies on home hematuria reagent-strip testing strongly suggests that screening led to an earlier stage of detection with improved survival. We await prospective randomized studies to clarify the value of mass screening for these diseases.

References

1. Wingo PA, Tong T, and Bolden S. Cancer statistics. Cancer 1995;45:8-30.

2. Potosky AL, Miller BA, Albertson PC, and Kramer BS. The role of increasing detection in the rising incidence of prostate cancer. JAMA 1995;273:548-552.

3. Miller BA, Ries LAG, Hankey BF, et al, eds. Cancer Statistics Review: 1973-1989. NIH publication no. 92-2789. Bethesda, MD: National Cancer Institute; 1992.

4. Catalona WJ, Richie JP, Ahmann FR, et al. Comparison of digital rectal examinations and serum prostate specific antigen in the early detection of prostate cancer: results of a multicentered clinical trial of 6,630
men. J Urol 1994;151:1283-1290.

5. Richie JP, Catalona WJ, Ahmann FR, et al. Effect of patient age on early detection of prostate cancer with serum prostate specific antigen and digital rectal examination. Urology 1993;42: 365-374.

6. Boring CC, Squires TS, Tong T, and Montgomery S. Cancer statistics, 1994. CA Cancer J Clin 1994;44:7-26.

7. Catalona WJ, Smith DS, Ratliff TL, and Basler JW. Detection of organ confined prostate cancer is increased through prostate specific antigen based screening. JAMA 1993;270:948-954.

8. Gerber GS, Thompson IM, Thisted R, and Chodak GW. Disease specific survival following routine prostate cancer screening by digital rectal examination. JAMA 1993;269:61-64.

9. Oesterling JE. Prostate specific antigen: a critical assessment of the most useful tumor marker for adenocarcinoma of the prostate. J Urol 1991;145:907-923.

10. Brawer MK, Chetner MP, Beatie J, et al. Screening for prostatic carcinoma with prostate specific antigen. J Urol 1992;147:841-845.

11. Catalona WJ. Screening for prostate cancer: enthusiasm. Urology 1993;42:113-115.

12. Jacobsen SJ, Katusic SK, Bergstralh EJ, et al. Incidence of prostate cancer diagnosis in the eras before and after serum prostate specific antigen testing. JAMA 1985;274:1445-1449.

13. Smith DS, Catalona WJ, and Hersaschman JD. Longitudinal screening for prostate cancer with prostate specific antigen. JAMA 1996;276:1309-1315.

14. Benson MC, Whang IS, Pantuck A, et al. Prostate specific antigen density: a means of distinguishing benign prostatic hypertrophy and prostatic cancer. J Urol 1992;147:815-816.

15. Catalona WJ, Richie JP, deKernion JB, et al. Comparison of prostate specific antigen (PSA) concentration versus prostate specific antigen density (PSAD) in the early detection of prostate cancer: receiver operation characteristics (ROC) curves. J Urol 1994;152:2031-2036.

16. Oesterling JE, Jacobsen SJ, Schute CG, et al. Serum specific antigen in the community based population of healthy men: establishment of age specific reference ranges. JAMA 1992;270: 860-864.

17. Catalona WJ, Hudson MA, Scardino PT, et al. Selection of optimal prostate specific antigen (PSA) cutoffs for early detection of prostate cancer: receiver operating characteristics (ROC) curves. J Urol 1994;152:2037-2042.

18. Carter HB, Pearson JD, Mettler EJ, et al. Longitudinal evaluation of PSA levels in men with and without prostate disease. JAMA 1992;267:2215-2220.

19. Catalona WJ, Smith DS, Wolfert RL, et al. Evaluation of percentage of free serum prostate specific antigen to improve specificity of prostate cancer screening. JAMA 1995;274:1214-1220.

20. Catalona WJ, Smith DS, and Ornstein DK. Prostate cancer detection in men with serum PSA concentrations of 2.6-4.0 ng/ml and benign prostate examination: enhancement of specificity with free PSA measurements. JAMA 1997;277:1452-1455.

21. Miller BA, et al, eds. Cancer Statistics Review: 1973-1989. NIH publication no. 92-2789. National Cancer Institute; 1992.

22. Steele GD Jr, Osteen RT, Winchester DP, et al. Clinical highlights from the national cancer database: 1994. Ca Cancer J Clin 1994;44:71-80.

23. Steinberg GD, Carter BS, Beaty TH, et al. Family history and the risk of prostate cancer. Prostate 1990;17:337-347.

24. Johansson JE, Holmberg L, Johansson S, et al. Fifteen year survival in prostate cancer. A prospective, population based study in Sweden. JAMA 1997;277:467-471.

25. Albertsen PC, Fryback DG, Storer BE, et al. Long term survival among men with conservatively treated localized prostate cancer. JAMA 1995;274:626-631.

26. Parker SL, Tong T, Bolden S, and Wingo PA. Cancer statistics 1997. Ca Cancer J Clin 1997;47:5-27.

27. Kaye KW, and Lange PH. Mode of presentation of invasive bladder cancer: reassessment of the problem. J Urol 1992;128:31-33.

28. Prout GR Jr, Barton PA, Griffin PP, et al. Treated history of noninvasive grade I transitional cell carcinoma. J Urol 1992;148:1413-1419.

29. Heney NM, Ahmed S, Flanagan MJ, et al. Superficial bladder cancer: progression and recurrence. J Urol 1983;130:1083-1086.

30. Marshall VF. Current clinical problems regarding bladder tumors. Cancer 1956;3:543-550.

31. Messing EM, and Vaillancourt A. Hematuria screening for bladder cancer. J Occup Med 1990;32:835-845.

32. Messing EM, Young TB, Hunt BB, et al. The significance of asymptomatic microhematuria in men 50 or more years old: findings of a home screening study using urinary dipsticks. J Urol 1987;137:919-922.

33. Greene LF, O'Shaughnessy EJ Jr, and Hendricks ED. Study of 500 patients with asymptomatic microhematuria. JAMA 1956;161:610-613.

34. Carson CC III, Segura JW, and Greene LF. Clinical importance of microhematuria. JAMA 1979;241:149-150.

35. Golin AL, and Howard RS. Asymptomatic microscopic hematuria. J Urol 1980;124:389-391.

36. Thompson IM. The evaluation of microscopic hematuria: a population based study. J Urol 1987;138:1189-1190.

37. Messing EM, Young TB, Hunt VB, et al. Urinary tract cancers found by home screening with hematuria dipsticks in healthy men over 50 years of age. Cancer 1989;64:2361-2367.

38. Messing EM, Young TB, Hunt VB, et al. Home screening for hematuria: results of a multiclinic study. J Urol 1992;148: 289-292.

39. Messing EM, Young TB, Hunt VB, et al. Hematuria home screening: repeat testing results. J Urol 1995;154:57-61.

40. Britton JP, Dowell AC, and Whelan P. Dipstick haematuria and bladder cancer in men over 60: results of a community study. BMJ 1989;299:1010-1012.

41. Britton JP, Dowell AC, Whelan P, et al. A community study of bladder cancer screening by the detection of occult urinary bleeding. J Urol 1992;148:788-790.

42. Messing EM, and Catalona WJ. Urethral tumors of the urinary tract. In: Walsh PC, Retik AB, Vaughn ED, and Wein AJ, eds. Campbell's Urology, 7th ed. Philadelphia: WB Saunders 1998:2354-2357.

43. Lawrence WF, Messing EM, and Bram LL. The cost effectiveness of screening men for bladder cancer using chemical reagent strips to detect microscopic hematuria. Abstract presented at the 1995 Annual Meeting of the American Urological Association, April 27, 1995, Las Vegas, NV.

44. Messing EM, and Vaillancourt A. Hematuria screening for bladder cancer. J Occup Med 1990;32:838-845.

45. Wheeless LL, Reeder JE, Han R, et al. Bladder irrigation specimens assay by fluorescence in situ hybridization to interphase nuclei. Cytometry 1994;17:319-326.

46. Mao L, Schoenberg MP, Scicchitano M, et al. Molecular detection of primary bladder cancer by microsatellite analysis. Science 1996;271:659-662.

47. Brentnall TA. Microsatellite instability: shifting concepts in tumorigenesis. Am J Pathol 1995;147:561-563.

48. Steiner G, Scoenberg MP, Linn JF, et al. Detection of bladder cancer recurrence by macrosatellite analysis of urine. Nature Med 1997;3:621-624.

49. Golijanin D, Sherman Y, Shapiro A, and Pode D. Detection of bladder tumors by immunostaining of the Lewis X antigen in cells from voided urine [Review]. Urology 1995;46:173-177.

50. Bonner RB, Hemstreet GP III, Fradet Y, et al. Bladder cancer risk assessment with quantitative fluorescence image analysis of tumor markers in exfoliated bladder cells. Cancer 1993;72:2461-2469.

51. Grossman HB, Washington RW Jr, Care TE, and Liebert HB. Alterations in antigen expression in superficial bladder cancer [Review]. J Cell Biochem Suppl 1992;161:63-68.

52. Fradet Y, Tardif M, Bourget L, et al. Clinical cancer progression in urinary bladder tumors evaluated by multiparameter flow cytometry with monoclonal antibodies. Cancer Res 1990;50:432-437.

53. Huland E, Huland H, Meier TH, et al. Comparison of 15 monoclonal antibodies against tumor associated antigens of transitional cell carcinoma of the human bladder. J Urol 1991;146:1631-1636.

54. Messing EM. Clinical implications of the expression of epidermal growth factor receptors in human transitional cell carcinoma. Cancer Res 1990;50:2530-2537.

55. Mellon K, Wright C, Kelly P, et al. Long term outcome related to epidermal growth factor receptor status in bladder cancer. J Urol 1995;153:1919-1925.

56. Guirguis R, Schiffmann E, Liu B, et al. Detection of autocrine motility factor in urine as a marker of bladder cancer. J Natl Cancer Inst 1988;80:1203-1211.

57. D'Hallewin M, and Baert L. Initial evaluation of the bladder tumor antigen test in superficial bladder cancer. J Urol 1996;155:475-476.

58. Soloway MS, Briggman JB, Carpinigo GA, et al. Use of the new tumor marker, urinary NMP 22, in detection of occult or rapidly recurring transitional cell carcinoma of urinary tract following surgical treatment. J Urol 1996;156:363-367.

Received October 26, 1997; Accepted October 28, 1997.

From the Department of Urology, University of Rochester Medical Center, Rochester, NY, USA
Correspondence address: Martin M. Goldstein, MD, University of Rochester Medical Center, Department of Urology, 601 Elmwood Avenue, Room 1-5336, Rochester, NY 14642.