- Email: [email protected]
Newly Diagnosed Bladder Cancer: The Relationship of Initial Symptoms, Degree of Microhematuria and Tumor Marker Status
0022-5347/02/1685-1955/0 THE JOURNAL OF UROLOGY® Copyright © 2002 by AMERICAN UROLOGICAL ASSOCIATION, INC.®
Vol. 168, 1955–1959, November 2002 Printed in U.S.A.
DOI: 10.1097/01.ju.0000034403.18207.66
NEWLY DIAGNOSED BLADDER CANCER: THE RELATIONSHIP OF INITIAL SYMPTOMS, DEGREE OF MICROHEMATURIA AND TUMOR MARKER STATUS HANS BOMAN, HANS HEDELIN, STEFAN JACOBSSON
AND
¨ NG STEN HOLMA
From the Department of Surgery, Alingsås Lasarett, Department of Urology, Ka¨rnsjukhuset, Sko¨vde, and Departments of Clinical Chemistry and Transfusion Medicine and Urology, Sahlgrenska University Hospital, Go¨teborg, Sweden
ABSTRACT
Purpose: We recorded initial symptoms and evaluated the frequency and intensity of hematuria in patients with newly diagnosed bladder cancer. We also evaluated and compared the sensitivity of bladder wash cytology, NMP22 (Matritech, Newton, Massachusetts), BTA Stat (Bion Diagnostic Sciences, Redmond, Washington) and UBC antigen (IDL Biotech, Sollentona, Sweden) with hematuria dipsticks and flow cytometry for determining the size of erythrocytes in urine. Materials and Methods: Urine samples were collected from 92 patients with newly diagnosed bladder cancer, 64 with idiopathic microhematuria and 42 with nephritis. Urine was analyzed for NMP22, BTA Stat, UBC and erythrocytes size using flow cytometry. Bladder wash cytology was done at cystoscopy. Urine was analyzed for microhematuria with hematuria dipsticks at home for 7 consecutive days immediately before the operation and in the hospital on the day of surgery. Results: Sensitivity was 75% for NMP22, 78% for BTA Stat, 64% for UBC and 61% for flow cytometry at 73% specificity. Cytology had 42% sensitivity at 97% specificity. Tumor size, grade and stage had a statistically significant influence on NMP22, BTA Stat, UBC and cytology. Of the patients 75% had microhematuria on the day of the operation and 75% had hematuria at least 1 of 7 days when tested at home the last week before transurethral bladder resection. The 70% of all patients with macroscopic hematuria as the initial symptom did not seem to differ from those without the condition in tumor size, grade, stage or tumor marker levels. Conclusions: Flow cytometry was not well enough able to distinguish patients with bladder cancer from controls. The sensitivity of all tested markers, including hematuria dipsticks, was high for large and high grade, high stage tumors. Further studies are needed to evaluate whether a marker could be used to determine priority among patients referred due to microhematuria. KEY WORDS: bladder; bladder neoplasms; hematuria; tumor markers, biological
Bladder cancer is a common malignancy in males. Most patients with bladder cancer are diagnosed after 1 or more episodes of macroscopic hematuria. More than 50% of patients with bladder cancer have a noninvasive tumor at initial diagnosis with an excellent prognosis.1 Muscle invasive or metastatic disease is evident in approximately 30% of cases at initial diagnosis and the prognosis is poor with a 5-year survival rate of 30% to 40% despite extensive treatment.1 Home screening for bladder cancer performed by Messing et al resulted in earlier diagnosis of high grade cancer and decreased bladder cancer mortality.2 The screening study was based on self-testing for hematuria with a chemical reagent strip. In the last decade urine bound markers for bladder cancer other than cytology and hematuria were introduced.3, 4 There are now at least 3 commercially available tumor markers. BTA Stat is a simple qualitative test performed within a few minutes in voided urine that measures complement factor H related proteins.3– 8 NMP22 is a quantitative test for nuclear matrix proteins.3, 4, 8 –10 UBC detects urinary fragments of cytokeratin 8 and 18.4, 11, 12 Patients with microscopic hematuria may be further studied by determining the size of erythrocytes in urine. Erythrocytes of normal size are noted when hematuria is caused by bleeding from the urinary tract mucosa (nonglomerular), while small asymmetrical erythrocytes indicate
that the cause is nephritis or idiopathic microhematuria.13 Microscopic hematuria has low specificity for bladder cancer but it may have the potential to increase when combined with erythrocyte size determination. We included this method in the current study since there were few patients with newly diagnosed bladder cancer in earlier reports of erythrocyte size determination. In a previous study we investigated the sensitivity of NMP22, BTA Stat, UBC and bladder wash cytology in 66 patients with newly diagnosed bladder cancer and 93 with recurrence. The main finding was that all markers had low sensitivity for small and low grade recurrences.14 The current study included 92 patients with newly diagnosed bladder tumors including 66 from the previous study. We evaluated whether various bladder cancer markers were influenced by tumor size, grade, stage and initial bladder tumor associated symptoms. We studied the frequency and intensity of microhematuria in patients with newly diagnosed bladder cancer. Flow cytometry for comparing erythrocyte size in urine was performed in patients with bladder cancer and microhematuria with hematuria secondary to nephritis. These patients served as controls. Thus, we studied and compared various bladder cancer markers and evaluated further the relationship of hematuria and bladder cancer.
Accepted for publication June 28, 2002. Supported by Odd Fellow Logen Gustav II Adolf. 1955
1956
NEWLY DIAGNOSED BLADDER CANCER MATERIALS AND METHODS
The study was performed at the Department of Urology, Sahlgrenska University Hospital from January 1998 to May 2000. The intention was to include all 302 patients newly diagnosed with bladder cancer during the study period. Patients were excluded from analysis if they had prostate or renal cancer, urinary tract infection, stones or an indwelling catheter. Only 92 patients were included in the study partly due to the various difficulties of including patients at a busy clinic. Those who underwent transurethral bladder resection on an emergency basis or during the weekend were not included. Many were in poor general condition and could not participate, while others were not willing to participate. Symptoms and signs causing the patient to contact the physician were documented by 66 on a questionnaire and obtained from the clinical records of 26. Urine was sampled from 92 patients with a mean age of 71 years with newly diagnosed bladder cancer, including 25 women, on the morning of the day of transurethral bladder resection. All analyses were not performed in all cases, mainly because not enough urine was available. Tumor size was determined by the surgeons using the width of the resectoscope sling as a reference. Mean age and sex distribution were the same in the study group as in the whole bladder cancer population but the percent of patients with high grade, high stage tumors was lower. In the study group 30% of the patients had grade 3 cancer compared with 42% in the overall population of 302. Similarly 59% of the patients had stage Ta cancer compared with 52% in the population. The mean interval from diagnostic cystoscopy in the outpatient department to transurethral bladder resection in the hospital was 25 days. All tumors except a stage T4 tumor in a patient in poor general condition were verified by histopathological testing. To serve as controls urine was collected from 64 patients with a mean age of 52 years evaluated due to microhematuria, including 25 women, in 1998 and 1999. Urography and cystoscopy were normal and hematuria was considered idiopathic. Urine samples were obtained from 42 patients with a mean age of 57 years with various types of chronic nephritis, including 14 women, in whom microhematuria was noted at a routine followup visit to the nephrologist. These patients served as controls only for the investigation of erythrocyte size by flow cytometry. BTA Stat in 149 patients, including 88 of the 92 with bladder cancer, was assayed within 48 hours at the research laboratory of the urology department according to manufacturer instructions. In this qualitative test a line appears for visual identification if an antigen conjugate complex is formed. A single laboratory technician performed all analyses. NMP22 was done in 149 patients, including 89 of the 92 with bladder cancer. Urine was sampled and stabilized with the NMP22 test kit stabilizer within 30 minutes. The samples were centrifuged and frozen at ⫺70C, and analyzed within 3 months according to manufacturer instructions. Results are shown in units per ml. The cutoff value for a positive test was 4. The reason for this rather low value was that it provided the same 73% specificity as BTA Stat. Thus, the different markers could be compared more easily. UBC in 107 patients, including 59 of the 92 with bladder cancer, was analyzed according to manufacturer instructions in urine frozen (to ⫺20C) with enzyme-linked immunosorbent assay. It was not stabilized with a pH stabilizing liquid. The tests were blinded concerning clinical status. Results are shown in g./l. The cutoff value for a positive test was 1. The reason for this rather low value was that it provided the same 73% specificity as BTA Stat. Bladder wash cytology was performed in 146 patients, including 87 of the 92 with bladder cancer. The bladder
was washed through the cystoscope sheath with 100 to 200 ml. saline. The sample was immediately fixed in ethanol. Examination was performed by experienced cytopathologists blinded to clinical information. Sensitivity was determined at 2 levels, including 1 for malignant cells only, and 1 for malignant cells and unspecific cell changes. The dipstick test for hematuria was done in 54 of the 92 patients at home and in 85 of the 92 immediately before transurethral bladder resection. Those with newly diagnosed bladder cancer were asked to participate in the study and assess urine for hematuria in a midstream voided urine sample using Ecur-Test dipsticks (Boehringer Mannheim, Mannheim, Germany) daily at home the week before transurethral bladder resection. Many patients were in poor general or mental condition and others did not understand the information, which explains why only 54 patients returned the completed forms. Urine was also collected in the hospital a few hours before the operation and the dipstick was automatically analyzed for hematuria using a Miditron junior optical scanner (Boehringer Mannheim). The test was interpreted as positive when the hemoglobin content of the urine was at least grade 1 on a scale of 1 to 3. This value is equivalent to the trace level of reagent strips used by Messing et al.2 Flow cytometry in 126 patients, including 44 with tumor, 42 with nephritis and 40 with idiopathic microhematuria, was performed to measure erythrocyte size. Newly voided urine was immediately spun, re-suspended in isotonic phosphate buffer, pH 7.4, and maintained at room temperature. The sample was spun, the supernatant was discarded and the sample was re-suspended in 0.5 ml. buffer. A saturating concentration of a glycophorin A specific fluorescein isothiocyanate conjugated monoclonal antibody, GPA clone JC159 (Dakopatts, Glostrup, Denmark), for 15 minutes was used to label the erythrocytes. No nonreactive antibody was used as a control. A FACScan flow cytometer (Becton Dickinson, San Jose, California) was used. Forward scatter, side scatter and green fluorescence were collected and stored in list mode using commercially available software. Instrument settings were checked and adjusted on a daily basis using CaliBrite Beads (Becton Dickinson) and commercially available software. Erythrocytes present in the sample were identified as GPA positive events with appropriate forward and side scatter signals. The forward scatter median channel signal was used as an arbitrary parameter considered to be related to erythrocyte volume. Because the quantity of erythrocytes measured by flow cytometry and reported as a forward scatter signal is unknown and not calibrated to measure erythrocyte volume, no attempt was made to translate forward scatter into volume. The cutoff value for a positive test (nonglomerular bleeding) was greater than 931 since it provided the same specificity (73%) as the BTA Stat test. For comparing the cancer and control groups Fisher’s nonparametric permutation test15 was used for continuous and ordered variables, and Fisher’s exact test was used for dichotomous variables. For comparing the groups with and without macrohematuria at the initial diagnosis Fisher’s exact test was used for dichotomous variables and the MantelHaenszel chi-square test was used for ordered categorical variables. We used Pearson’s correlation coefficient for descriptive purpose and Pitman’s nonparametric permutation test,15 which is a powerful nonparametric test for correlations, based on measured values and not ranks. Adjustment for all confounders on correlation analysis was made with Spearman’s partial correlation analysis. Stepwise multivariate regression was done to evaluate whether a combination of tests was better than each univariate test. Cytology and BTA Stat were analyzed as dichotomous variables, and NMP22, UBC and flow cytometry were analyzed as continuous variables. All significance tests were 2-sided and with significance at 5%.
1957
NEWLY DIAGNOSED BLADDER CANCER RESULTS
Tumor characteristics. Table 1 lists the distribution of tumor size, grade and stage. Median size was 20 mm. (range 2 to 50), 60% were between 11 and 30 mm., and only 5 were less than 5 mm. There was a solitary tumor in 64 cases (70%). Tumor markers. There was a significant difference for NMP22, BTA Stat, UBC, flow cytometry and bladder wash cytology in patients with and without bladder cancer (p ⬍0.001 UBC p ⫽ 0.006). Figure 1 shows sensitivity and specificity curves. At 73% specificity for all tests the sensitivity for new tumors was 75% for NMP22, 78% for BTA Stat, 64% for UBC and 61% for flow cytometry. At 97% specificity with malignant cells and unspecific cell changes considered positive the sensitivity of cytology for newly diagnosed bladder tumors was 42%. At 100% specificity with only malignant cells considered positive the sensitivity of cytology for newly diagnosed bladder tumors was 35%. The positive predictive value for NMP22, BTA Stat, UBC and cytology was 80%, 81%, 74% and 94%, and the negative predictive value was 67%, 70%, 63% and 55%, respectively. None of the tests significantly correlated with the number of tumors (table 2). There was a statistically significant influence of tumor grade, stage and size on NMP22, BTA Stat, UBC and cytology (table 2). Grade did not significantly correlate with NMP22 (p ⫽ 0.094) or UBC (p ⫽ 0.18) after adjusting for patient age and sex. No other changes appeared after those adjustments. Sensitivity of the tests at different tumor sizes was 32% (cytology) for small tumors to 100% for large tumors (table 1). There was a low but significant influence of hematuria on BTA Stat, NMP22 and cytology but not on UBC with increasing hematuria levels even after compensating for confounding factors. The influence on cytology was weak and probably more of a coincidence. For BTA Stat and NMP22 the influence may have been due to cross reactions. Logistic regression was performed to evaluate whether a combination of tests could better discriminate patients with and without tumor. This analysis showed that many combinations of the tests provided a slight but clinically insignificant improvement. The model with the largest AUC included cytology (p ⫽ 0.015), NMP22 (p ⫽ 0.006) and BTA Stat (p ⫽ 0.009). The AUC in this model was 0.873. When adding age and sex to stepwise analysis age (p ⬍0.001), BTA Stat (p ⫽ 0.029) and cytology (p ⫽ 0.043) showed a significant effect. The AUC in this model was 0.886. Hematuria could not be included on regression analysis due to the lack of a control group. Hematuria home screening. A total of 54 patients self-
FIG. 1. Sensitivity and specificity curves
tested urine with dipsticks daily during the last week before transurethral bladder resection. According to the submitted protocols 14 patients (26%) had no hematuria and 14 (26%) had at least 1 day without any trace of microhematuria. On the other hand, 16 patients (30%) reported microhematuria daily and 10 (19%) reported macroscopic hematuria on at least 1 day. Hospital hematuria screening on the morning of the day of transurethral bladder resection. Table 3 shows the number of patients with and without microhematuria. Overall sensitivity for newly diagnosed bladder cancer was 74% for hematuria dipsticks with home testing and 75% for automatic analysis at the hospital. Hematuria significantly correlated with tumor size, stage and grade but not with the number of tumors (table 2). Grade (p ⫽ 0.062) and stage (p ⫽ 0.068) did not significantly correlate with hematuria after adjusting for age and sex. No other changes appeared after these adjustments. Flow cytometry. Urine from 44 of the 64 patients who had hematuria the morning of the day of transurethral bladder resection was examined with flow cytometry and the mean forward scatter value was 1,286 (range 141 to 2,597, fig. 2). In the 40 patients with idiopathic hematuria the mean forward scatter value was 574 (range 38 to 2,839) and in the 42 with nephritis it was 806 (range 78 to 1,911). Sensitivity was 50% at 95% specificity. There was no correlation of erythrocyte size with urine pH, urine density, proteinuria, serum creatinine, tumor size, grade or stage. Of the 14 patients 6 (43%) with grade 1 microhematuria had positive flow cytom-
TABLE 1. Test sensitivity No. Pts./Total No. (%) No. Pts.
NMP22 (89 pts.)
BTA Stat (88 pts.)
Size (mm.): 92 10 or Less 16/21 (76) 13/22 (59) 11–20 20/30 (67) 23/29 (79) 21–30 18/24 (75) 19/23 (83) 31 or Greater 13/14 (93) 14/14 (100) Grade: 89 1 19/26 (73) 16/26 (62) 2 24/33 (72) 27/33 (82) 3 22/27 (81) 25/26 (96) Stage: 92 Ta 36/52 (69) 34/51 (67) T1 21/26 (81) 24/26 (92) T2 or greater 10/11 (91) 11/11 (100) Hospital hematuria screening: 85 None 14/21 (66) 12/21 (57) Grade 1 10/16 (63) 11/16 (69) Grade 2 8/10 (80) 9/10 (90) Grade 3 29/35 (82) 32/35 (91) Macroscopic 3/3 (100) 3/3 (100) Not all tests were done in each patient and all tumor variables were not known. Specificity was 97% for bladder wash cytology and it was set at 73% for NMP22, BTA Stat and UBC.
UBC (59 pts.)
Bladder Wash Cytology (87 pts.)
8/15 (53) 13/23 (56) 7/11 (63) 10/10 (100)
7/22 (32) 11/30 (37) 8/22 (36) 11/13 (85)
11/18 (61) 15/23 (65) 12/18 (67)
2/25 (8) 15/35 (43) 20/26 (77)
20/35 (57) 12/18 (67) 6/6 (100)
13/51 (25) 16/25 (64) 8/11 (73)
11/16 6/12 6/8 14/20 0/1
5/20 (25) 4/16 (25) 5/10 (50) 21/33 (63) 1/3 (30)
(69) (50) (75) (70)
1958
NEWLY DIAGNOSED BLADDER CANCER TABLE 2. Tests correlated with variables Tumor Size
NMP22 BTA Stat UBC Bladder wash cytology Flow cytometry Hospital hematuria screening
Grade
Stage
Hospital Hematuria Screening
No. Tumors
r
p Value
r
p Value
r
p Value
r
p Value
r
p Value
0.52 0.31 0.33 0.28 0.27 0.34
⬍0.001 0.005 0.013 0.01 0.09 0.02
0.37 0.33 0.32 0.54 0.19 0.24
⬍0.001 0.003 0.009 ⬍0.001 0.22 0.035
0.52 0.33 0.41 0.4 0.27 0.29
⬍0.001 0.001 0.005 ⬍0.001 0.08 0.008
0.15 0.09 ⫺0.1 0.21 0.01 0.01
0.18 0.51 0.5 0.07 0.97 0.9
0.37 0.36 0.22 0.36
⬍0.001 0.002 0.9 0.002
TABLE 3. Relationship of urine dipstick results immediately before transurethral bladder resection to tumor grade and stage No. Microhematuria (grade) 1
2
3
No. Macrohematuria
Grade: 19 16 1 10 2 2 6 10 3 3 4 Stage: 19 16 Ta 16 9 T1 3 6 T2 or greater 0 1 * Excluding 2 patients with nontransitional cell carcinoma.
10 2 5 3 10 6 4 0
35 9 11 15 35 15 12 8
3 2 1 0 3 3 0 0
No Hematuria*
Total No.
25 33 25 49 25 9
TABLE 5. Tumor grade and stage, and marker sensitivity in relation to initial symptoms Macrohematuria
FIG. 2. Plot of erythrocyte size shows median forward scatter channel signal, which was related to erythrocyte volume.
etry results compared with 4 of the 7 (57%) with grade 2 and 16 of the 19 (84%) with grade 3. No changes appeared after adjusting for age and sex. Initial symptoms (table 4). There was no significant difference in tumor size, grade and stage in the 64 patients with macrohematuria as the initial symptom compared with 25 without the condition. The sensitivity of the different markers was almost the same in the 2 groups (table 5). DISCUSSION
This study of 92 patients with newly diagnosed bladder cancer is one of the largest evaluating the applicability of NMP22, BTA Stat, UBC and cytology. To our knowledge it is the only report of the relationship of these markers, initial symptoms and the degree of microhematuria. Ramakumar et TABLE 4. Initial symptoms in patients with newly diagnosed bladder cancer Initial Symptoms
Men
Women
No. pts. % Macrohematuria % Voiding problems % Abdominal pain % Urinary tract infection % Hematuria screening % Prostate Ca screening % Gynecological health examination
67 71 18 3 3 2 3
25 64 4 20 12
No Macrohematuria*
No. pts. 64 25 % Grade: 1 30 28 2 36 48 3 34 24 % Stage: Ta 59 60 T1 28 32 T2 or Greater 13 8 % Pos. marker (No./total No.): NMP22 77 (48/62) 76 (19/25) BTA Stat 80 (49/61) 76 (19/25) UBC 68 (26/38) 60 (12/20) Cytology 41 (25/61) 48 (12/25) * Not available in 2 patients with nontransitional cell cancer and stage T4 cancer. † Not significant.
p Value†
0.64
0.75
1 0.86 0.72 0.72 1 with
al compared BTA Stat, NMP22, fibrin degradation products, telomerase, chemiluminescent hemoglobin and hemoglobin dipsticks for detecting bladder cancer.8 Telomerase in urine had the highest combination of 70% sensitivity and 99% specificity for bladder cancer but only 19 of 50 histologically confirmed cases were newly diagnosed cancer. Hemoglobin dipstick had 47% sensitivity and 84% specificity, values comparable to those of NMP22. The sensitivity of the different tests in the current study would probably have been higher if all tumors identified during the study period had been included since those that were not included had a somewhat higher stage and grade. A possible reason was that patients who underwent emergency surgery were not included. These patients had larger and frequently high grade, high stage tumors. The control group included 64 patients with microhematuria who had negative urography and cystoscopy results but overlooked bladder carcinoma or carcinoma in situ cannot completely be ruled out. Tests specificity may have been higher if healthy individuals without hematuria had been included. Still the overall sensitivity and specificity of NMP22, BTA Stat, UBC and cytology were in accordance with what others have reported.4, 8 On the other hand, we noted that the sensitivity of these markers was lower for small and low grade, low stage
1959
NEWLY DIAGNOSED BLADDER CANCER
tumors, a finding that only a few others have reported earlier.6, 11, 14 The sensitivity for small tumors was also unacceptably low using hematuria dipsticks. It is evident that none of the studied bladder cancer markers, including hematuria dipsticks, can be used in screening studies for detecting small, low grade tumors. An earlier finding of small, low grade and low stage tumors would probably not change the overall prognosis of bladder cancer more than marginally. Sensitivity is much higher for high grade bladder tumors and in the hematuria screening study of Messing et al they were diagnosed at an earlier stage, resulting in an improved prognosis.2 Hematuria dipsticks had 75% sensitivity for bladder tumors when grade 1 hematuria was included. The prevalence of microhematuria in the general population varies markedly from 2% to 38% depending on whether single or multiple tests are done and on the population studied.16, 17 The incidence of bladder cancer in patients with microhematuria is reported to be 1% to 5%.3 From these values it can be calculated that the specificity of hematuria for bladder cancer is 62% to 98%. Thus, hematuria dipstick sensitivity and specificity are comparable to those of the other markers tested. An advantage of hematuria dipsticks is the low cost when used in a standard manner to detect new tumors in risk groups and bladder tumor recurrence. Another advantage is that results are available immediately, as for BTA Stat. A disadvantage is that hematuria dipsticks are used in a nonstandard manner in a number of different situations. Still it seems reasonable that future reports of new markers should be compared with cytology and hematuria dipstick results. In this report we evaluated whether a combination of dipstick test and flow cytometry to compare erythrocyte size would increase the performance of hematuria screening by distinguishing glomerular from nonglomerular bleeding. The result was higher specificity but at the cost of unacceptably low sensitivity since 25% of all patients had no hematuria. Thus, erythrocyte size could not be determined. Nevertheless, it was interesting that sensitivity was 84% in patients with grade 3 microhematuria. Another interesting finding in our study was that patients with macroscopic hematuria as the initial symptom did not seem to differ from those without that condition in tumor size, grade, stage or tumor marker levels. The result must be considered with caution since there were only 25 patients. An explanation may be that other factors, for example tumor location and patient physical activity, are more important than tumor characteristics. We also noted that tumors missed by tumor markers were smaller with lower grade and stage than tumors above cutoff levels. This finding was logical since it can be hypothesized that small tumors do not leak as many red blood cells, cytokeratin or nuclear matrix proteins as large tumors. Another finding was that sensitivity and specificity were slightly higher with a combination of tests but it is questionable whether the increased costs for 2 tests would be justified. A group at risk for bladder cancer is men with lower urinary tract symptoms who today are examined with transrectal ultrasound and prostate specific antigen but not always with cystoscopy. A new marker may be helpful when selecting patients for cystoscopy. Of great concern is the fact that for all except BTA Stat 19% to 33% of grade 3 tumors would be overlooked. Further studies are needed to determine the costs and benefits of the new bladder tumor markers as well as of cytology. CONCLUSIONS
Tumor size, grade, stage and marker sensitivity were the same irrespective of whether macrohematuria was the initial
symptom. The sensitivity of all tested markers, including dipstick hematuria, was low for small and low grade, low stage tumors, and high for large and high grade, high stage tumors. Since low grade tumors already have a good prognosis, the potential value of the markers seems to be the earlier diagnosis of high grade tumors. Anita Fae provided technical assistance.
REFERENCES
1. Malmstrom, P. U., Busch, C. and Norlen, B. J.: Recurrence, progression and survival in bladder cancer. A retrospective analysis of 232 patients with greater than or equal to 5-year followup. Scand J Urol Nephrol, 21: 185, 1987 2. Messing, E. M., Young, T. B., Hunt, V. B., Gilchrist, K. W., Newton, M. A., Bram, L. L. et al: Comparison of bladder cancer outcome in men undergoing hematuria home screening versus those with standard clinical presentations. Urology, 45: 387, 1995 3. Lokeshwar, V. B. and Soloway, M. S.: Current bladder tumor tests: does their projected utility fulfill clinical necessity? J Urol, 165: 1067, 2001 4. Konety, B. R. and Getzenberg, R. H.: Urine based markers of urological malignancy. J Urol, 165: 600, 2001 5. Sarosdy, M. F., deVere White, R. W., Soloway, M. S., Sheinfeld, J., Hudson, M. A., Schellhammer, P. F. et al: Results of a multicenter trial using the BTA test to monitor for and diagnose recurrent bladder cancer. J Urol, 154: 379, 1995 6. Pode, D., Shapiro, A., Wald, M., Nativ, O., Laufer, M. and Kaver, I.: Noninvasive detection of bladder cancer with the BTA stat test. J Urol, 161: 443, 1999 7. Raitanen, M.-P., Marttila, T., Kaasinen, E., Rintala, E., Aine, R., Tammela, T. L. J. et al: Sensitivity of human complement factor H related protein (BTA stat) test and voided urine cytology in the diagnosis of bladder cancer. J Urol, 163: 1689, 2000 8. Ramakumar, S., Bhuiyan, J., Besse, J. A., Roberts, S. G., Wollan, P. C., Blute, M. L. et al: Comparison of screening methods in the detection of bladder cancer. J Urol, 161: 388, 1999 9. Wiener, H. G., Mian, Ch., Haitel, A., Pycha, A., Schatzl, G. and Marberger, M.: Can urine bound diagnostic tests replace cystoscopy in the management of bladder cancer? J Urol, 159: 1876, 1998 10. Casella, R., Huber, P., Blo¨ chlinger, A., Stoffel, F., Dalquen, P., Gasser, T. C. et al: Urinary level of nuclear matrix protein 22 in the diagnosis of bladder cancer: experience with 130 patients with biopsy confirmed tumor. J Urol, 164: 1926, 2000 11. Sa´ nchez-Carbayo, M., Herrero, E., Megias, J., Mira, A., Espasa, A., Chinchilla, V. et al: Initial evaluation of the diagnostic performance of the new urinary bladder cancer antigen test as a tumor marker for transitional cell carcinoma of the bladder. J Urol, 161: 1110, 1999 12. Sumi, S., Arai, K., Kitahara, S. and Yoshida, K. I.: Preliminary report of the clinical performance of a new urinary bladder cancer antigen test: comparison to voided urinary cytology in the detection of transitional cell carcinoma of the bladder. Clin Chim Acta, 296: 111, 2000 13. Tanaka, M., Kitamoto, Y., Sato, T. and Ishii, T.: Flow cytometric analysis of hematuria using fluorescent antihemoglobin antibody. Nephron, 65: 354, 1993 14. Boman, H., Hedelin, H. and Holma¨ ng, S.: Four bladder tumor markers have a disappointingly low sensitivity for small size and low grade recurrence. J Urol, 167: 80, 2002 15. Good, P.: Permutation Tests. A Practical Guide to Resampling Methods for Testing Hypotheses. New York: Springer, pp. 36 –37, 45– 47, 2000 16. Sultana, S. R., Goodman, C. M., Byrne, D. J. and Baxby, K.: Microscopic haematuria: urological investigation using a standard protocol. Br J Urol, 78: 691, 1996 17. Froom, P., Ribak, J. and Benbassat, J.: Significance of microhaematuria in young adults. Br Med J, 288: 20, 1984
Vol. 168, 1955–1959, November 2002 Printed in U.S.A.
DOI: 10.1097/01.ju.0000034403.18207.66
NEWLY DIAGNOSED BLADDER CANCER: THE RELATIONSHIP OF INITIAL SYMPTOMS, DEGREE OF MICROHEMATURIA AND TUMOR MARKER STATUS HANS BOMAN, HANS HEDELIN, STEFAN JACOBSSON
AND
¨ NG STEN HOLMA
From the Department of Surgery, Alingsås Lasarett, Department of Urology, Ka¨rnsjukhuset, Sko¨vde, and Departments of Clinical Chemistry and Transfusion Medicine and Urology, Sahlgrenska University Hospital, Go¨teborg, Sweden
ABSTRACT
Purpose: We recorded initial symptoms and evaluated the frequency and intensity of hematuria in patients with newly diagnosed bladder cancer. We also evaluated and compared the sensitivity of bladder wash cytology, NMP22 (Matritech, Newton, Massachusetts), BTA Stat (Bion Diagnostic Sciences, Redmond, Washington) and UBC antigen (IDL Biotech, Sollentona, Sweden) with hematuria dipsticks and flow cytometry for determining the size of erythrocytes in urine. Materials and Methods: Urine samples were collected from 92 patients with newly diagnosed bladder cancer, 64 with idiopathic microhematuria and 42 with nephritis. Urine was analyzed for NMP22, BTA Stat, UBC and erythrocytes size using flow cytometry. Bladder wash cytology was done at cystoscopy. Urine was analyzed for microhematuria with hematuria dipsticks at home for 7 consecutive days immediately before the operation and in the hospital on the day of surgery. Results: Sensitivity was 75% for NMP22, 78% for BTA Stat, 64% for UBC and 61% for flow cytometry at 73% specificity. Cytology had 42% sensitivity at 97% specificity. Tumor size, grade and stage had a statistically significant influence on NMP22, BTA Stat, UBC and cytology. Of the patients 75% had microhematuria on the day of the operation and 75% had hematuria at least 1 of 7 days when tested at home the last week before transurethral bladder resection. The 70% of all patients with macroscopic hematuria as the initial symptom did not seem to differ from those without the condition in tumor size, grade, stage or tumor marker levels. Conclusions: Flow cytometry was not well enough able to distinguish patients with bladder cancer from controls. The sensitivity of all tested markers, including hematuria dipsticks, was high for large and high grade, high stage tumors. Further studies are needed to evaluate whether a marker could be used to determine priority among patients referred due to microhematuria. KEY WORDS: bladder; bladder neoplasms; hematuria; tumor markers, biological
Bladder cancer is a common malignancy in males. Most patients with bladder cancer are diagnosed after 1 or more episodes of macroscopic hematuria. More than 50% of patients with bladder cancer have a noninvasive tumor at initial diagnosis with an excellent prognosis.1 Muscle invasive or metastatic disease is evident in approximately 30% of cases at initial diagnosis and the prognosis is poor with a 5-year survival rate of 30% to 40% despite extensive treatment.1 Home screening for bladder cancer performed by Messing et al resulted in earlier diagnosis of high grade cancer and decreased bladder cancer mortality.2 The screening study was based on self-testing for hematuria with a chemical reagent strip. In the last decade urine bound markers for bladder cancer other than cytology and hematuria were introduced.3, 4 There are now at least 3 commercially available tumor markers. BTA Stat is a simple qualitative test performed within a few minutes in voided urine that measures complement factor H related proteins.3– 8 NMP22 is a quantitative test for nuclear matrix proteins.3, 4, 8 –10 UBC detects urinary fragments of cytokeratin 8 and 18.4, 11, 12 Patients with microscopic hematuria may be further studied by determining the size of erythrocytes in urine. Erythrocytes of normal size are noted when hematuria is caused by bleeding from the urinary tract mucosa (nonglomerular), while small asymmetrical erythrocytes indicate
that the cause is nephritis or idiopathic microhematuria.13 Microscopic hematuria has low specificity for bladder cancer but it may have the potential to increase when combined with erythrocyte size determination. We included this method in the current study since there were few patients with newly diagnosed bladder cancer in earlier reports of erythrocyte size determination. In a previous study we investigated the sensitivity of NMP22, BTA Stat, UBC and bladder wash cytology in 66 patients with newly diagnosed bladder cancer and 93 with recurrence. The main finding was that all markers had low sensitivity for small and low grade recurrences.14 The current study included 92 patients with newly diagnosed bladder tumors including 66 from the previous study. We evaluated whether various bladder cancer markers were influenced by tumor size, grade, stage and initial bladder tumor associated symptoms. We studied the frequency and intensity of microhematuria in patients with newly diagnosed bladder cancer. Flow cytometry for comparing erythrocyte size in urine was performed in patients with bladder cancer and microhematuria with hematuria secondary to nephritis. These patients served as controls. Thus, we studied and compared various bladder cancer markers and evaluated further the relationship of hematuria and bladder cancer.
Accepted for publication June 28, 2002. Supported by Odd Fellow Logen Gustav II Adolf. 1955
1956
NEWLY DIAGNOSED BLADDER CANCER MATERIALS AND METHODS
The study was performed at the Department of Urology, Sahlgrenska University Hospital from January 1998 to May 2000. The intention was to include all 302 patients newly diagnosed with bladder cancer during the study period. Patients were excluded from analysis if they had prostate or renal cancer, urinary tract infection, stones or an indwelling catheter. Only 92 patients were included in the study partly due to the various difficulties of including patients at a busy clinic. Those who underwent transurethral bladder resection on an emergency basis or during the weekend were not included. Many were in poor general condition and could not participate, while others were not willing to participate. Symptoms and signs causing the patient to contact the physician were documented by 66 on a questionnaire and obtained from the clinical records of 26. Urine was sampled from 92 patients with a mean age of 71 years with newly diagnosed bladder cancer, including 25 women, on the morning of the day of transurethral bladder resection. All analyses were not performed in all cases, mainly because not enough urine was available. Tumor size was determined by the surgeons using the width of the resectoscope sling as a reference. Mean age and sex distribution were the same in the study group as in the whole bladder cancer population but the percent of patients with high grade, high stage tumors was lower. In the study group 30% of the patients had grade 3 cancer compared with 42% in the overall population of 302. Similarly 59% of the patients had stage Ta cancer compared with 52% in the population. The mean interval from diagnostic cystoscopy in the outpatient department to transurethral bladder resection in the hospital was 25 days. All tumors except a stage T4 tumor in a patient in poor general condition were verified by histopathological testing. To serve as controls urine was collected from 64 patients with a mean age of 52 years evaluated due to microhematuria, including 25 women, in 1998 and 1999. Urography and cystoscopy were normal and hematuria was considered idiopathic. Urine samples were obtained from 42 patients with a mean age of 57 years with various types of chronic nephritis, including 14 women, in whom microhematuria was noted at a routine followup visit to the nephrologist. These patients served as controls only for the investigation of erythrocyte size by flow cytometry. BTA Stat in 149 patients, including 88 of the 92 with bladder cancer, was assayed within 48 hours at the research laboratory of the urology department according to manufacturer instructions. In this qualitative test a line appears for visual identification if an antigen conjugate complex is formed. A single laboratory technician performed all analyses. NMP22 was done in 149 patients, including 89 of the 92 with bladder cancer. Urine was sampled and stabilized with the NMP22 test kit stabilizer within 30 minutes. The samples were centrifuged and frozen at ⫺70C, and analyzed within 3 months according to manufacturer instructions. Results are shown in units per ml. The cutoff value for a positive test was 4. The reason for this rather low value was that it provided the same 73% specificity as BTA Stat. Thus, the different markers could be compared more easily. UBC in 107 patients, including 59 of the 92 with bladder cancer, was analyzed according to manufacturer instructions in urine frozen (to ⫺20C) with enzyme-linked immunosorbent assay. It was not stabilized with a pH stabilizing liquid. The tests were blinded concerning clinical status. Results are shown in g./l. The cutoff value for a positive test was 1. The reason for this rather low value was that it provided the same 73% specificity as BTA Stat. Bladder wash cytology was performed in 146 patients, including 87 of the 92 with bladder cancer. The bladder
was washed through the cystoscope sheath with 100 to 200 ml. saline. The sample was immediately fixed in ethanol. Examination was performed by experienced cytopathologists blinded to clinical information. Sensitivity was determined at 2 levels, including 1 for malignant cells only, and 1 for malignant cells and unspecific cell changes. The dipstick test for hematuria was done in 54 of the 92 patients at home and in 85 of the 92 immediately before transurethral bladder resection. Those with newly diagnosed bladder cancer were asked to participate in the study and assess urine for hematuria in a midstream voided urine sample using Ecur-Test dipsticks (Boehringer Mannheim, Mannheim, Germany) daily at home the week before transurethral bladder resection. Many patients were in poor general or mental condition and others did not understand the information, which explains why only 54 patients returned the completed forms. Urine was also collected in the hospital a few hours before the operation and the dipstick was automatically analyzed for hematuria using a Miditron junior optical scanner (Boehringer Mannheim). The test was interpreted as positive when the hemoglobin content of the urine was at least grade 1 on a scale of 1 to 3. This value is equivalent to the trace level of reagent strips used by Messing et al.2 Flow cytometry in 126 patients, including 44 with tumor, 42 with nephritis and 40 with idiopathic microhematuria, was performed to measure erythrocyte size. Newly voided urine was immediately spun, re-suspended in isotonic phosphate buffer, pH 7.4, and maintained at room temperature. The sample was spun, the supernatant was discarded and the sample was re-suspended in 0.5 ml. buffer. A saturating concentration of a glycophorin A specific fluorescein isothiocyanate conjugated monoclonal antibody, GPA clone JC159 (Dakopatts, Glostrup, Denmark), for 15 minutes was used to label the erythrocytes. No nonreactive antibody was used as a control. A FACScan flow cytometer (Becton Dickinson, San Jose, California) was used. Forward scatter, side scatter and green fluorescence were collected and stored in list mode using commercially available software. Instrument settings were checked and adjusted on a daily basis using CaliBrite Beads (Becton Dickinson) and commercially available software. Erythrocytes present in the sample were identified as GPA positive events with appropriate forward and side scatter signals. The forward scatter median channel signal was used as an arbitrary parameter considered to be related to erythrocyte volume. Because the quantity of erythrocytes measured by flow cytometry and reported as a forward scatter signal is unknown and not calibrated to measure erythrocyte volume, no attempt was made to translate forward scatter into volume. The cutoff value for a positive test (nonglomerular bleeding) was greater than 931 since it provided the same specificity (73%) as the BTA Stat test. For comparing the cancer and control groups Fisher’s nonparametric permutation test15 was used for continuous and ordered variables, and Fisher’s exact test was used for dichotomous variables. For comparing the groups with and without macrohematuria at the initial diagnosis Fisher’s exact test was used for dichotomous variables and the MantelHaenszel chi-square test was used for ordered categorical variables. We used Pearson’s correlation coefficient for descriptive purpose and Pitman’s nonparametric permutation test,15 which is a powerful nonparametric test for correlations, based on measured values and not ranks. Adjustment for all confounders on correlation analysis was made with Spearman’s partial correlation analysis. Stepwise multivariate regression was done to evaluate whether a combination of tests was better than each univariate test. Cytology and BTA Stat were analyzed as dichotomous variables, and NMP22, UBC and flow cytometry were analyzed as continuous variables. All significance tests were 2-sided and with significance at 5%.
1957
NEWLY DIAGNOSED BLADDER CANCER RESULTS
Tumor characteristics. Table 1 lists the distribution of tumor size, grade and stage. Median size was 20 mm. (range 2 to 50), 60% were between 11 and 30 mm., and only 5 were less than 5 mm. There was a solitary tumor in 64 cases (70%). Tumor markers. There was a significant difference for NMP22, BTA Stat, UBC, flow cytometry and bladder wash cytology in patients with and without bladder cancer (p ⬍0.001 UBC p ⫽ 0.006). Figure 1 shows sensitivity and specificity curves. At 73% specificity for all tests the sensitivity for new tumors was 75% for NMP22, 78% for BTA Stat, 64% for UBC and 61% for flow cytometry. At 97% specificity with malignant cells and unspecific cell changes considered positive the sensitivity of cytology for newly diagnosed bladder tumors was 42%. At 100% specificity with only malignant cells considered positive the sensitivity of cytology for newly diagnosed bladder tumors was 35%. The positive predictive value for NMP22, BTA Stat, UBC and cytology was 80%, 81%, 74% and 94%, and the negative predictive value was 67%, 70%, 63% and 55%, respectively. None of the tests significantly correlated with the number of tumors (table 2). There was a statistically significant influence of tumor grade, stage and size on NMP22, BTA Stat, UBC and cytology (table 2). Grade did not significantly correlate with NMP22 (p ⫽ 0.094) or UBC (p ⫽ 0.18) after adjusting for patient age and sex. No other changes appeared after those adjustments. Sensitivity of the tests at different tumor sizes was 32% (cytology) for small tumors to 100% for large tumors (table 1). There was a low but significant influence of hematuria on BTA Stat, NMP22 and cytology but not on UBC with increasing hematuria levels even after compensating for confounding factors. The influence on cytology was weak and probably more of a coincidence. For BTA Stat and NMP22 the influence may have been due to cross reactions. Logistic regression was performed to evaluate whether a combination of tests could better discriminate patients with and without tumor. This analysis showed that many combinations of the tests provided a slight but clinically insignificant improvement. The model with the largest AUC included cytology (p ⫽ 0.015), NMP22 (p ⫽ 0.006) and BTA Stat (p ⫽ 0.009). The AUC in this model was 0.873. When adding age and sex to stepwise analysis age (p ⬍0.001), BTA Stat (p ⫽ 0.029) and cytology (p ⫽ 0.043) showed a significant effect. The AUC in this model was 0.886. Hematuria could not be included on regression analysis due to the lack of a control group. Hematuria home screening. A total of 54 patients self-
FIG. 1. Sensitivity and specificity curves
tested urine with dipsticks daily during the last week before transurethral bladder resection. According to the submitted protocols 14 patients (26%) had no hematuria and 14 (26%) had at least 1 day without any trace of microhematuria. On the other hand, 16 patients (30%) reported microhematuria daily and 10 (19%) reported macroscopic hematuria on at least 1 day. Hospital hematuria screening on the morning of the day of transurethral bladder resection. Table 3 shows the number of patients with and without microhematuria. Overall sensitivity for newly diagnosed bladder cancer was 74% for hematuria dipsticks with home testing and 75% for automatic analysis at the hospital. Hematuria significantly correlated with tumor size, stage and grade but not with the number of tumors (table 2). Grade (p ⫽ 0.062) and stage (p ⫽ 0.068) did not significantly correlate with hematuria after adjusting for age and sex. No other changes appeared after these adjustments. Flow cytometry. Urine from 44 of the 64 patients who had hematuria the morning of the day of transurethral bladder resection was examined with flow cytometry and the mean forward scatter value was 1,286 (range 141 to 2,597, fig. 2). In the 40 patients with idiopathic hematuria the mean forward scatter value was 574 (range 38 to 2,839) and in the 42 with nephritis it was 806 (range 78 to 1,911). Sensitivity was 50% at 95% specificity. There was no correlation of erythrocyte size with urine pH, urine density, proteinuria, serum creatinine, tumor size, grade or stage. Of the 14 patients 6 (43%) with grade 1 microhematuria had positive flow cytom-
TABLE 1. Test sensitivity No. Pts./Total No. (%) No. Pts.
NMP22 (89 pts.)
BTA Stat (88 pts.)
Size (mm.): 92 10 or Less 16/21 (76) 13/22 (59) 11–20 20/30 (67) 23/29 (79) 21–30 18/24 (75) 19/23 (83) 31 or Greater 13/14 (93) 14/14 (100) Grade: 89 1 19/26 (73) 16/26 (62) 2 24/33 (72) 27/33 (82) 3 22/27 (81) 25/26 (96) Stage: 92 Ta 36/52 (69) 34/51 (67) T1 21/26 (81) 24/26 (92) T2 or greater 10/11 (91) 11/11 (100) Hospital hematuria screening: 85 None 14/21 (66) 12/21 (57) Grade 1 10/16 (63) 11/16 (69) Grade 2 8/10 (80) 9/10 (90) Grade 3 29/35 (82) 32/35 (91) Macroscopic 3/3 (100) 3/3 (100) Not all tests were done in each patient and all tumor variables were not known. Specificity was 97% for bladder wash cytology and it was set at 73% for NMP22, BTA Stat and UBC.
UBC (59 pts.)
Bladder Wash Cytology (87 pts.)
8/15 (53) 13/23 (56) 7/11 (63) 10/10 (100)
7/22 (32) 11/30 (37) 8/22 (36) 11/13 (85)
11/18 (61) 15/23 (65) 12/18 (67)
2/25 (8) 15/35 (43) 20/26 (77)
20/35 (57) 12/18 (67) 6/6 (100)
13/51 (25) 16/25 (64) 8/11 (73)
11/16 6/12 6/8 14/20 0/1
5/20 (25) 4/16 (25) 5/10 (50) 21/33 (63) 1/3 (30)
(69) (50) (75) (70)
1958
NEWLY DIAGNOSED BLADDER CANCER TABLE 2. Tests correlated with variables Tumor Size
NMP22 BTA Stat UBC Bladder wash cytology Flow cytometry Hospital hematuria screening
Grade
Stage
Hospital Hematuria Screening
No. Tumors
r
p Value
r
p Value
r
p Value
r
p Value
r
p Value
0.52 0.31 0.33 0.28 0.27 0.34
⬍0.001 0.005 0.013 0.01 0.09 0.02
0.37 0.33 0.32 0.54 0.19 0.24
⬍0.001 0.003 0.009 ⬍0.001 0.22 0.035
0.52 0.33 0.41 0.4 0.27 0.29
⬍0.001 0.001 0.005 ⬍0.001 0.08 0.008
0.15 0.09 ⫺0.1 0.21 0.01 0.01
0.18 0.51 0.5 0.07 0.97 0.9
0.37 0.36 0.22 0.36
⬍0.001 0.002 0.9 0.002
TABLE 3. Relationship of urine dipstick results immediately before transurethral bladder resection to tumor grade and stage No. Microhematuria (grade) 1
2
3
No. Macrohematuria
Grade: 19 16 1 10 2 2 6 10 3 3 4 Stage: 19 16 Ta 16 9 T1 3 6 T2 or greater 0 1 * Excluding 2 patients with nontransitional cell carcinoma.
10 2 5 3 10 6 4 0
35 9 11 15 35 15 12 8
3 2 1 0 3 3 0 0
No Hematuria*
Total No.
25 33 25 49 25 9
TABLE 5. Tumor grade and stage, and marker sensitivity in relation to initial symptoms Macrohematuria
FIG. 2. Plot of erythrocyte size shows median forward scatter channel signal, which was related to erythrocyte volume.
etry results compared with 4 of the 7 (57%) with grade 2 and 16 of the 19 (84%) with grade 3. No changes appeared after adjusting for age and sex. Initial symptoms (table 4). There was no significant difference in tumor size, grade and stage in the 64 patients with macrohematuria as the initial symptom compared with 25 without the condition. The sensitivity of the different markers was almost the same in the 2 groups (table 5). DISCUSSION
This study of 92 patients with newly diagnosed bladder cancer is one of the largest evaluating the applicability of NMP22, BTA Stat, UBC and cytology. To our knowledge it is the only report of the relationship of these markers, initial symptoms and the degree of microhematuria. Ramakumar et TABLE 4. Initial symptoms in patients with newly diagnosed bladder cancer Initial Symptoms
Men
Women
No. pts. % Macrohematuria % Voiding problems % Abdominal pain % Urinary tract infection % Hematuria screening % Prostate Ca screening % Gynecological health examination
67 71 18 3 3 2 3
25 64 4 20 12
No Macrohematuria*
No. pts. 64 25 % Grade: 1 30 28 2 36 48 3 34 24 % Stage: Ta 59 60 T1 28 32 T2 or Greater 13 8 % Pos. marker (No./total No.): NMP22 77 (48/62) 76 (19/25) BTA Stat 80 (49/61) 76 (19/25) UBC 68 (26/38) 60 (12/20) Cytology 41 (25/61) 48 (12/25) * Not available in 2 patients with nontransitional cell cancer and stage T4 cancer. † Not significant.
p Value†
0.64
0.75
1 0.86 0.72 0.72 1 with
al compared BTA Stat, NMP22, fibrin degradation products, telomerase, chemiluminescent hemoglobin and hemoglobin dipsticks for detecting bladder cancer.8 Telomerase in urine had the highest combination of 70% sensitivity and 99% specificity for bladder cancer but only 19 of 50 histologically confirmed cases were newly diagnosed cancer. Hemoglobin dipstick had 47% sensitivity and 84% specificity, values comparable to those of NMP22. The sensitivity of the different tests in the current study would probably have been higher if all tumors identified during the study period had been included since those that were not included had a somewhat higher stage and grade. A possible reason was that patients who underwent emergency surgery were not included. These patients had larger and frequently high grade, high stage tumors. The control group included 64 patients with microhematuria who had negative urography and cystoscopy results but overlooked bladder carcinoma or carcinoma in situ cannot completely be ruled out. Tests specificity may have been higher if healthy individuals without hematuria had been included. Still the overall sensitivity and specificity of NMP22, BTA Stat, UBC and cytology were in accordance with what others have reported.4, 8 On the other hand, we noted that the sensitivity of these markers was lower for small and low grade, low stage
1959
NEWLY DIAGNOSED BLADDER CANCER
tumors, a finding that only a few others have reported earlier.6, 11, 14 The sensitivity for small tumors was also unacceptably low using hematuria dipsticks. It is evident that none of the studied bladder cancer markers, including hematuria dipsticks, can be used in screening studies for detecting small, low grade tumors. An earlier finding of small, low grade and low stage tumors would probably not change the overall prognosis of bladder cancer more than marginally. Sensitivity is much higher for high grade bladder tumors and in the hematuria screening study of Messing et al they were diagnosed at an earlier stage, resulting in an improved prognosis.2 Hematuria dipsticks had 75% sensitivity for bladder tumors when grade 1 hematuria was included. The prevalence of microhematuria in the general population varies markedly from 2% to 38% depending on whether single or multiple tests are done and on the population studied.16, 17 The incidence of bladder cancer in patients with microhematuria is reported to be 1% to 5%.3 From these values it can be calculated that the specificity of hematuria for bladder cancer is 62% to 98%. Thus, hematuria dipstick sensitivity and specificity are comparable to those of the other markers tested. An advantage of hematuria dipsticks is the low cost when used in a standard manner to detect new tumors in risk groups and bladder tumor recurrence. Another advantage is that results are available immediately, as for BTA Stat. A disadvantage is that hematuria dipsticks are used in a nonstandard manner in a number of different situations. Still it seems reasonable that future reports of new markers should be compared with cytology and hematuria dipstick results. In this report we evaluated whether a combination of dipstick test and flow cytometry to compare erythrocyte size would increase the performance of hematuria screening by distinguishing glomerular from nonglomerular bleeding. The result was higher specificity but at the cost of unacceptably low sensitivity since 25% of all patients had no hematuria. Thus, erythrocyte size could not be determined. Nevertheless, it was interesting that sensitivity was 84% in patients with grade 3 microhematuria. Another interesting finding in our study was that patients with macroscopic hematuria as the initial symptom did not seem to differ from those without that condition in tumor size, grade, stage or tumor marker levels. The result must be considered with caution since there were only 25 patients. An explanation may be that other factors, for example tumor location and patient physical activity, are more important than tumor characteristics. We also noted that tumors missed by tumor markers were smaller with lower grade and stage than tumors above cutoff levels. This finding was logical since it can be hypothesized that small tumors do not leak as many red blood cells, cytokeratin or nuclear matrix proteins as large tumors. Another finding was that sensitivity and specificity were slightly higher with a combination of tests but it is questionable whether the increased costs for 2 tests would be justified. A group at risk for bladder cancer is men with lower urinary tract symptoms who today are examined with transrectal ultrasound and prostate specific antigen but not always with cystoscopy. A new marker may be helpful when selecting patients for cystoscopy. Of great concern is the fact that for all except BTA Stat 19% to 33% of grade 3 tumors would be overlooked. Further studies are needed to determine the costs and benefits of the new bladder tumor markers as well as of cytology. CONCLUSIONS
Tumor size, grade, stage and marker sensitivity were the same irrespective of whether macrohematuria was the initial
symptom. The sensitivity of all tested markers, including dipstick hematuria, was low for small and low grade, low stage tumors, and high for large and high grade, high stage tumors. Since low grade tumors already have a good prognosis, the potential value of the markers seems to be the earlier diagnosis of high grade tumors. Anita Fae provided technical assistance.
REFERENCES
1. Malmstrom, P. U., Busch, C. and Norlen, B. J.: Recurrence, progression and survival in bladder cancer. A retrospective analysis of 232 patients with greater than or equal to 5-year followup. Scand J Urol Nephrol, 21: 185, 1987 2. Messing, E. M., Young, T. B., Hunt, V. B., Gilchrist, K. W., Newton, M. A., Bram, L. L. et al: Comparison of bladder cancer outcome in men undergoing hematuria home screening versus those with standard clinical presentations. Urology, 45: 387, 1995 3. Lokeshwar, V. B. and Soloway, M. S.: Current bladder tumor tests: does their projected utility fulfill clinical necessity? J Urol, 165: 1067, 2001 4. Konety, B. R. and Getzenberg, R. H.: Urine based markers of urological malignancy. J Urol, 165: 600, 2001 5. Sarosdy, M. F., deVere White, R. W., Soloway, M. S., Sheinfeld, J., Hudson, M. A., Schellhammer, P. F. et al: Results of a multicenter trial using the BTA test to monitor for and diagnose recurrent bladder cancer. J Urol, 154: 379, 1995 6. Pode, D., Shapiro, A., Wald, M., Nativ, O., Laufer, M. and Kaver, I.: Noninvasive detection of bladder cancer with the BTA stat test. J Urol, 161: 443, 1999 7. Raitanen, M.-P., Marttila, T., Kaasinen, E., Rintala, E., Aine, R., Tammela, T. L. J. et al: Sensitivity of human complement factor H related protein (BTA stat) test and voided urine cytology in the diagnosis of bladder cancer. J Urol, 163: 1689, 2000 8. Ramakumar, S., Bhuiyan, J., Besse, J. A., Roberts, S. G., Wollan, P. C., Blute, M. L. et al: Comparison of screening methods in the detection of bladder cancer. J Urol, 161: 388, 1999 9. Wiener, H. G., Mian, Ch., Haitel, A., Pycha, A., Schatzl, G. and Marberger, M.: Can urine bound diagnostic tests replace cystoscopy in the management of bladder cancer? J Urol, 159: 1876, 1998 10. Casella, R., Huber, P., Blo¨ chlinger, A., Stoffel, F., Dalquen, P., Gasser, T. C. et al: Urinary level of nuclear matrix protein 22 in the diagnosis of bladder cancer: experience with 130 patients with biopsy confirmed tumor. J Urol, 164: 1926, 2000 11. Sa´ nchez-Carbayo, M., Herrero, E., Megias, J., Mira, A., Espasa, A., Chinchilla, V. et al: Initial evaluation of the diagnostic performance of the new urinary bladder cancer antigen test as a tumor marker for transitional cell carcinoma of the bladder. J Urol, 161: 1110, 1999 12. Sumi, S., Arai, K., Kitahara, S. and Yoshida, K. I.: Preliminary report of the clinical performance of a new urinary bladder cancer antigen test: comparison to voided urinary cytology in the detection of transitional cell carcinoma of the bladder. Clin Chim Acta, 296: 111, 2000 13. Tanaka, M., Kitamoto, Y., Sato, T. and Ishii, T.: Flow cytometric analysis of hematuria using fluorescent antihemoglobin antibody. Nephron, 65: 354, 1993 14. Boman, H., Hedelin, H. and Holma¨ ng, S.: Four bladder tumor markers have a disappointingly low sensitivity for small size and low grade recurrence. J Urol, 167: 80, 2002 15. Good, P.: Permutation Tests. A Practical Guide to Resampling Methods for Testing Hypotheses. New York: Springer, pp. 36 –37, 45– 47, 2000 16. Sultana, S. R., Goodman, C. M., Byrne, D. J. and Baxby, K.: Microscopic haematuria: urological investigation using a standard protocol. Br J Urol, 78: 691, 1996 17. Froom, P., Ribak, J. and Benbassat, J.: Significance of microhaematuria in young adults. Br Med J, 288: 20, 1984