The Value of Morphometry and DNA Flow Cytometry in Addition to Classic Prognosticators in Superficial Urinary Bladder Carcinoma

The Value of Morphometry and DNA Flow Cytometry in Addition to Classic Prognosticators in Superficial Urinary Bladder Carcinoma EDWARD C. M. BLOMJOUS, M.D., NELLIE W. SCHIPPER, M.Sc, JAN P. A. BAAK, M.D., PH.D., WINAND VOS, M.D., HERMAN J DE VOOGT, M.D., PH.D., AND CHRIS J. L. M. MEIJER, M.D., PH.D. In 8 patients with primary superficial bladder carcinoma Tumor Nodes Metastasis (TNM classification: stages Ta and Tl) with adequate follow-up of at least four years, the value of selective nuclear morphometry and DNA flow cytometry on paraffinembedded material in addition to classic prognosticators was assessed. Only the quantitative techniques appeared to be valuable predictors of new tumor occurrence. The recurrence rate in patients with large nuclei (mean nuclear area > 95 iim 2 ; n - 29) and in aneuploid cases (n = 3) was significantly higher (Wilcoxon: P =.5 and P -.1) than in those with small nuclei (mean nuclear area < = 95 Mm 2 ; n = 51) and diploid cases (n = 5). The prevalence of large nuclei and aneuploidy also appeared useful to predict progressive recurrence, i.e., grade 3 or/ and muscle invasive carcinoma (TNM classification: stages T2- T4) (chi-square: P <.1). Clinical follow-up showed that only 62.1% of the cases with large nuclei remained free of progressive recurrence, compared with 92.2% of those with small nuclei (Mantel-Cox: P <.1). For the aneuploid and diploid cases, these figures came to 53.3% and 98% (Mantel-Cox: P <.1). By multivariate analysis DNA ploidy was selected as the best discriminator. None of the classic prognosticators, including histologic grade, had additional prognostic value. Also, morphometry did not add to the prognosis prediction, which can be explained by the considerable overlapping between the prevalence of large nuclei and aneuploidy (24 of 29 and 3 cases, respectively). These findings practically suggest that patients presenting with superficial carcinoma with large nuclei (mean nuclear area > 95 nm 2 ) or aneuploid DNA values should be treated more aggressively. (Key words: Superficial bladder carcinoma; Morphometry; Flow cytometry) Am J Clin Pathol 1989;91:243-248 PREDICTION of the clinical behavior of superficial bladder carcinoma (stages Ta and Tl) in a given patient is a major problem in urologic practice. Despite transurethral resection, in a considerable proportion of the patients with superficial carcinoma recurrence develops (actually new occurrence) and 15-25% of cases ultimately progress to high-grade carcinoma (World Health Organization [WHO] classification: grade 3) and muscle invasive disease Tumor Nodes Metastasis (TNM-classification: stages T2-T4). 2,18,2,21 Identification of these patients at an early stage would enable a more appropriate treatment. Received May 9, 1988; received revised manuscript and accepted for publication July 13, 1988. Address reprint requests to Dr. Blomjous: Department of Pathology, Free University Hospital, De Boelelaan 1117, 17 MB Amsterdam, The Netherlands. Departments of Pathology and Urology, Free University Hospital, The Netherlands Prognosis partly depends on mucosal field alterations adjacent to the tumor (severe dysplasia or carcinoma in situ). las - 26 In addition, it is believed that prognostic information about progressive recurrence to some extent can be obtained by conventional clinicopathologic features of the initial tumor, such as tumor size, multiplicity, necrosis, histologic grade, and invasion of the lamina propria. 1 ' 318,2 However, these features do not impact heavily on prognosis. More refined techniques therefore have been used for the investigation of primary bladder carcinoma, in order to identify the patients with predilection to progression. These include identification of the pattern of immunohistochemical markers such as cell surface ABO(H) blood group antigen, T-antigen and carcinoembryonic antigen, chromosome analysis, DNA cytophotometry, and ultrastructural determination of alterations of the cell membrane. 9 "- 1419-24 - 2728 Although helpful to some degree, so far none of them appeared useful for routine practice by poor reproducibility or requirement of too sophisticated techniques. 29 Over the last few years nuclear morphometry has supplied an objective and practicable method for the examination of bladder carcinoma, 17,23 and especially the measurement of the areas of selectively sampled large and atypical nuclei has shown prognostic relevance. 5,6 As another objective technique, flow cytometry provides a fast and reliable method to determine the DNA content in large numbers of nuclei. In urology, flow cytometry has shown good correlation of DNA ploidy with tumor grade and stage and patient survival rate. 4,3 However, the value of both techniques as prognostic indicators in superficial bladder carcinoma has yet to be established. In this retrospective study we demonstrate the ability to predict the likelihood of tumor recurrence and progression by selective nuclear morphometry and by DNA flow cytometry on paraffin-embedded tissue of the initial superficial lesion. In addition, the prognostic value of these 243

244 BLOMJOUS ET AL. A.J.C.P. March 1989 techniques is shown to be superior to that of classic parameters of tumor behavior, including tumor size, multiplicity, necrosis, histologic grade, and invasiveness. Patients and Materials Materials and Methods Of 158 patients who consecutively presented with primary and untreated superficial transitional cell carcinoma of the urinary bladder (stages Ta and Tl) in our hospital between January 1974 and December 1982, 8 were selected for this study. Patients were included only if the follow-up had lasted at least 48 months, unless they had died of tumor previously. Additional criteria were that formalin-fixed, paraffin-embedded tissue of the initial tumor was still available from the archives and that morphometric andflowcytometric analysis was not impeded by excessive cauterization damage. Eighty cases fulfilled these criteria. Five other cases were excluded because DNA histograms were uninterpretable. Histologic grade was assessed according to the WHO classification 22 by two independent pathologists, in addition to the grading by routine diagnostic procedures. Complete agreement between the three observations occurred for 72% of the cases, and discrepancies did not exceed more than one grade. In case of disagreement, the prevailing grade was assigned. Clinicopathologic stage was assessed according to the TNM classification. 15 Clinical Features Sixty-six patients were male and 14 were female. Age at clinical presentation ranged from 3 to 85 years, and median and mean age were 68 and 67 years, respectively. Of the 8 patients, 5 had initially been treated by transurethral resection only, whereas 27 patients had additional radiotherapy (4, rads) and 3 cytostatic bladder instillation (Adriamycin ) after transurethral resection of the primary tumor. Additional therapy had roughly been based on advanced grade, multiplicity, large size, or ulceration of the tumor, but the regimen was not standardized. Clinical follow-up ranged from 4 to 1 years, mean 7.1 years. Subsequent therapy varied, depending on the number, frequency, and stage of recurrences, general health condition, and likelihood of cure. Morphometry Morphometry was applied on 4-/im-thick hematoxylin and eosin-stained sections. One representative section was selected for each specimen. Analysis was carried out with a projection microscope, equipped with a 1X oil immersion objective, projecting at a graphic tablet (MOP- Videoplan, Kontron, Munich, Federal Republic of Germany) with an optical magnification of 2,X. The nuclear areas of selectively sampled nuclei were measured, according to a previously described morphometric method. 5 This method essentially entails selection of the most atypical area of the histologic slide (size: approximately 5 low-power fields, 1X) and subsequent measurement of ten nuclei that are selected on account of their large size. This method has shown satisfactory intraobserver and interobserver reproducibility (correlation coefficients of mean nuclear area and standard deviation >.9; regression coefficients between.75 and 1.25). Moreover, from a variety of other morphometric procedures, this selective method has emerged as most suitable for discrimination between different tumor grades, resulting in a considerable percentage of 93.2% correctly classified cases from a learning set of bladder tumors. 5 Flow Cytometry Paraffin sections were cut at 5 nm, adjacent to the sections used for morphometry. DNA ploidy was determined with the method described by Hedley and colleagues, 16 with slight modifications. The sections were dewaxed in xylene, rehydrated, and enzymatically dispersed by incubation with protease (Sigma, St. Louis, MO; P- 5255,.5% in.9% buffered saline, ph = 7) for 3 minutes at 37 C, with vigorously intermittent vortex mixing. After mechanical detachment by repeated syringing (21- gauge needles) andfiltrationthrough a 5-/um nylon gauze, the cell suspension was centrifugated, washed in TRIShydrochloride acid buffer, and recentrifugated. The cells werefinallystained with the DNAfluorochrome4',6'-diaminido-2-phenyl-indole dihydrochloride (DAPI, Sigma D-1388), final concentration 2 Mg DAPI/1 ml TRIShydrochloride acid buffer. The cell suspensions were analyzed with the PAS II Flow Cytometer (Partec, Arlesheim, Switzerland), using excitation light at 35 nm. Formalin-fixed mouse thymocytes served as an external standard for instrument setting. The DNA index was determined from the histograms. In agreement with general practice, the first modal cell peak was considered to represent the G/G1 peak of the diploid cell population. Samples were regarded as aneuploid when, in addition to the G/G1 and G2/M peaks, one or more other peaks were detected. Furthermore, samples of which the proportion of peritetraploid cells (DNA indices between 1.9 and 2.1) exceeded 1% of the whole cell population were considered as peritetraploid. Statistical Analysis Significance and clinical relevance of the single classic and quantitative features were analyzed by Wilcoxon's

Vol.91 -No. 3 PROGNOSIS IN SUPERFICIAL BLADDER CARCINOMA 245 test and the chi-square test. To study, moreover, their prognostic value in respect to progression, univariate analysis according to Kaplan-Meier was performed. The analysis takes into account the time at risk from transurethral resection of the primary tumor to either progressive recurrence or last follow-up. Differences between the curves were analyzed with the Mantel-Cox test. Multivariate analysis using Cox' regression model was carried out to evaluate the prognostic value of a combination of features. Except from ranking the features to their predictive power, the purpose thereof is to reduce the number of variables because altogether they may contain redundant information. Values ofp<.5 were regarded as significant. Results Distribution of Classic and Quantitative Features and Outcome Fifty-three patients presented with a single primary tumor, 2 patients with two or three tumors, and 7 with more than three primary lesions. The diameter of the tumor or, in case of multiplicity, of the largest lesion, ranged from less than 1 cm to 7 cm: 56 patients had tumors up to 3 cm diameter and 24 carcinomas measured 4 cm or more. Ulceration of the tumor was observed at cystoscopic examination in only four cases. Table 1 shows the distribution of tumor grade and stage resulting from the pathologic assessments. Thirty-one cases were classed as grade 1,41 as grade 2, and 8 as grade 3 carcinoma. Invasive growth was absent in 36 cases (stage Ta); 44 cases showed invasion of the lamina propria (stage Tl). Nuclear morphometry showed mean nuclear areas ranging from 49.7 to 243.36 fim 2 (mean and standard deviation: 93.71 and 33.26 nm 2 ) and standard deviations of 4. to 7.29 nm 2 (mean and standard deviation: 9.92 and 8.2 nm 2 ). Flow cytometry showed 5 tumors to be diploid and 3 aneuploid, of which 14 were peritetraploid and 1 had multiple aneuploid tracings. The coefficient of variation of the diploid peak ranged from 3. to 11.7%, mean 6.5%. Clinical follow-up revealed 59 (74%) patients with one or more recurrences, defined as the number of follow-up cystoscopic studies at which recurrent carcinoma (single or multiple) was noted. Nineteen (24%) patients eventually had progressive recurrence develop at intervals of 3 to 12 months, with median and mean intervals of 24 and 35 months, respectively; progressive recurrence was defined as progression to high-grade (grade 3) carcinoma or to advanced stage (stages T2-T4) of disease. Figure 1 shows the distribution of the mean nuclear area and the DNA index in relation to progressive disease. Table 1. Distribution of Tumor Grade and Stage Grade 1 Stage Ta 23 Stage Tl 8 Total 31 Grade 2 13 28 41 Grade 3 Total 36 8 44 8 8 3 2 DNA l.b Index 1.6 1.4. 1.2 Features Related to Recurrence None of the classic prognosticators proved to be statistically significant in respect to the prediction of new occurrences. The quantitative parameters, on the other hand, appeared to be valuable predictors. To investigate the value of morphometry, the patients' tumors were divided into a subgroup of tumors with small nuclei (mean nuclear area < = 95 nm 2 ) (n = 51) that showed a relatively low mean recurrence rate (.31 recurrences/month), and a group with large nuclei (mean nuclear area > 95 nm 2 ) (n = 29) with significantly more recurrences (.69 recurrences/month) (Wilcoxon: 3.8; P =.5). Adjustment for therapy effect by omitting from consideration the 3 patients who received additional treatment after resection of the primary tumor resulted in similarfindings,although just not significant anymore for this subgroup (Wilcoxon: 2.92; P =.8). The practical relevance of the feature was limited, however, because analysis by the chi-square statistic did not yield a significant correlation between the presence of large nuclei (mean nuclear area > 95 ^m 2 ) and the occurrence of new carcinomas (chi-square:.21; P =.6 and chi-square:.2; P =.9). Also, DNA ploidy showed prognostic importance, because the recurrence rate was significantly higher for the patient group with aneuploid primary tumors (n = 3) (.87 recurrences/month) than for those with diploid carcinomas (n = 5) (.2 recurrences/month) (Wil- 2.8 2.6 2.4-2.2 O.B.6-.4' o no progression n = 61 progression n = 19 8 i B gblblb e 1 1 * * O H O» 1 o o g o 1 ( 1 1 // 1 Mean Nuclear Area (um21 FIG. 1. Distribution of the mean nuclear area and the DNA index in relation to progressive disease. Horizontally dotted line divides the diploid (n = 5) from the aneuploid (n = 3) tumors. Vertically dotted line divides the tumors with small nuclei (mean nuclear area & 95 fim 2 ) from those with large nuclei (mean nuclear area > 95 fim 2 ). u

246 BLOMJOUS ET AL. A.J.C.P. March 1989 coxon: 15.78; P =.1). The clinical relevance of the feature was moreover affirmed by the chi-square statistic because almost all patients with aneuploid tumors (26/ 3) had recurrent disease, whereas one-third of the diploid cases (17/5) remained free of recurrence (chi-square: 4.14; P =.4). Similar results were observed in the group of 5 patients without additional therapy of their primary tumor (Wilcoxon: 7.17; P =.7), although the clinical relevance was not any more obvious for this reduced number of patients (chi-square:.95; P =.3). Features Related to Progressive Recurrence Of the classic features, only advanced tumor grade showed a significant increase of the risk for progression (P =.6). This tendency was not significant for the subset of 5 patients without additional treatment of the primary tumor, which might result from the considerable reduction of especially the grade 3 patient group from eight to four patients. The quantitative techniques appeared valuable in the prediction of progressive recurrent carcinoma (Table 2). The mean and standard deviation of the nuclear area of the progressive cases (n = 19) were both significantly larger than those of the nonprogressive lesions (n = 61) (P =.1 and P -.1), and similar results were obtained for the reduced group of 5 patients without additional therapy (P =.1 and P =.6). The diagnostic value of the mean nuclear area was furthermore indicated by the chi-square statistic, because only 4 of the 51 cases (8%) with small nuclei (mean nuclear area < = 95 jtm 2 ) eventually had progressive recurrence, compared with 15 of the 29 cases (52%) with large nuclei (chi-square: 23.2; P <.1). Also, in the subset of 5 patients without additional treatment, this tendency appeared strongly significant (chi-square: 13.77; P =.2). The importance offlowcytometry was demonstrated by the low incidence of only one patient with progressive recurrence among the 5 diploid cases (2%), against 18 among the 3 aneuploid cases (6%) (chi-square: 34.83; P <.1). The findings were similar in the subset of 5 patients without additional therapy (chi-square: 17.56; P <.1). Univariate and Multivariate Analysis of Progression Figure 1A shows the Kaplan-Meier curves of the cases with relatively small (mean nuclear area < = 95 /*m 2 ; n = 51) and large (mean nuclear area > 95 /um 2 ; n = 29) nuclei: 92.2% and 62.1% respectively, remained free of progressive recurrence during four years follow-up (Mantel-Cox: 19.6; P <.1). Figure IB shows the patients with aneuploid (n = 3) and diploid primary tumors (n = 5): 98% and 53.3%, respectively, were free of progressive recurrence after four years of follow-up (Mantel-Cox: 37.24; P <.1). Adjustment for therapy effect by elimination of 3 patients who had additional therapy did not essentially change these curves (Mantel-Cox: 13.14; P =.3 and Mantel-Cox: 17.91; P <.1). In a stepwise regression analysis, DNA ploidy, being the most important prognosticator, was the first variable entered in the multivariate model. None of the remaining variables contributed significantly to the model. The mean nuclear area is apparently completely explained by the DNA ploidy, and the classic features also have no additional prognostic value. Also, after adjustment for treatment effect, DNA ploidy was the only variable selected in the subset of 5 patients without additional therapy. To evaluate the effect of treatment in addition to transurethral resection, the Kaplan-Meier curves of the patients with and without additional therapy were analyzed: no prognostic difference was observed (Mantel-Cox:.22; P =.9). Moreover, multivariate analysis also showed that additional treatment, when entered as covariate in the model, did not contribute to the prediction of prognosis. Discussion The most significant finding in the current study is that morphometry andflowcytometry both afford a sensitive means to predict whether a patient is at risk to develop new tumor occurrence or, more important, progressive recurrence after resection of primary superficial carcinoma. Quantitative features are shown to be by far superior to the classic prognosticators. In recent years morphometry has proved to be an objective and reproducible method in grading of urinary bladder carcinoma. 1217 ' 23 In a foregoing comparative study, which was undertaken in view of the different methodologic approaches of morphometry in the literature, the selective method as practiced in the present study emerged as the most suitable method for tumor grading. 5 The procedure of selective sampling explains our two to threefold larger values of the mean nuclear area in comparison with those from the literature, which in general are based on random sampling. 3,1217,23 In another recent study, a close relationship was demonstrated between selective nuclear morphometry and the clinical course in various stages of urinary bladder carcinoma. 6 The present findings additionally show the prognostic usefulness within the category of superficial bladder carcinoma (stage Ta and Tl). To our knowledge, this has not been reported before for any other morphometric approach. Flow cytometric studies have indicated a strong correlation of nuclear DNA content with grade and stage of urinary bladder carcinoma, and many investigators have advocated use of the technique to monitor conservatively treated patients by scanning urine specimens. 78 Moreover, the current findings show that flow cytometry of the primary tumor provides a useful prognostic parameter, as has also been indicated by some previous reports. 4,3 Combinedflowcytometric and cytogenetic analysis has

Vol. 91-No. 3 PROGNOSIS IN SUPERFICIAL BLADDER CARCINOMA 247 Table 2. The Mean and Standard Deviation of the Nuclear Area (MNA and SD) of Primary Superficial Bladder Carcinoma in Patients with and Without Progressive Recurrence for the Total Set of Patients and the Subset of Patients Without Additional Therapy MNA (mean ± SD) Statistic SD (mean ± SD) Statistic Total set of patients (n = 8) No progression (n = 61) Progression (n = 19) Subset of patients without additional therapy (n = 5) No progression (n = 38) Progression (n = 12) 85. ± 24.61 121.7 ±41.87 85.66 ± 24.2 119.42 ±31.42 Wilcoxon: 15.35 P =.1 Wilcoxon: 1.4 P =.1 8.49 ± 3.87 14.51 ± 14.7 8.62 ± 3.39 11.87 rt 6.65 Wilcoxon: 6.33 P=.1 Wilcoxon: 3.47 P =.6 shown that DNA aneuploidy is associated with increased numbers of chromosomes. 31 The simultaneous occurrence of aneuploid DNA values in most of the present cases with large nuclei (mean nuclear area > 95 ^m 2 ) suggests that an increase of the nuclear area also relies at least in part on quantitative chromosomal abnormalities. On the other hand, diploid tumors can still contain qualitative chromosomal abnormalities that remain unrecognized byflowcytometry, 25 ' 31 and the same limitation may apply to morphometry. However, advanced tumor grade and stage are especially associated with gross aneuploidy and higher chromosome counts, 25,31 and a similar tendency is demonstrated for progressive recurrence in the current series of superficial carcinomas. Therefore, the inability of the quantitative techniques to recognize near-diploid and pseudodiploid chromosomal abnormalities does not substantially affect their clinical significance. Despite the considerable overlapping, aneuploid DNA values and large nuclei (mean nuclear area > 95 /im 2 ) do not coincide in each of our cases (Fig. 1). It may be questioned if, in thefivecases with enlarged nuclei and without aneuploidy (Fig. 1; lower, right), the number of cells with abnormal DNA content was large enough to exceed the detection limits of flow cytometry. Another explanation could be that in these cases nuclear enlargement relied on other causes, such as increased nuclear protein content or edema. Nevertheless, nuclear enlargement in these diploid cases was of minor importance because progressive recurrence was not revealed in any of the five patients. The superiority offlowcytometry is shown moreover by the large proportion of progressive recurrences (5%) among the six cases in which morphometry could not detect large nuclei, despite the abnormal DNA values found by flow cytometry (Fig. 1; upper, left). The considerable overlapping between the occurrence of large nuclei (mean nuclear area > 95 nm 2 ) and aneuploid DNA values (24 of 29 and 3 cases, respectively) explains why the combination of both methods does not substantially improve the prognostication as compared with the single techniques. It can be concluded from these findings that application of only one of the techniques would supply sufficient prognostic information, with some preference forflowcytometry. However, morphometry is more thanflowcytometry, a fairly simple and inexpensive technique that can be performed on routinely stained paraffin sections of the primary tumor. The advantage of additional morphometric investigation seems, for the present, that it may serve as an easy check-up for flow cytometric findings or as alternative if the specimen is unsuited for flow cytometry, as, for example, can be by admixture of a large quantity of tissue damaged by cauterization. The clinical implication of the current findings is that patients presenting with an aneuploid superficial primary 9 BO. 7 w. Mantel-Cox: 19.6 p<.1 Mantel-Cox: 37.24 p<.1 months HNA<-95 um2 (n=51) -- MNA> 95 um2 (n-29) - DNA Index - 1 (n-5) - DNA Index > 1 ln-3) FlG. 2. A (upper). Kaplan-Meier curves of progressive recurrence for patients with small (MNA S 95 jira 2 ) and large (MNA > 95 jtm 2 ) mean nuclear areas of the primary tumor. B (lower). Kaplan-Meier curves of progressive recurrence for patients with diploid and aneuploid primary tumors.

248 BLOMJOUS ET AL. A.J.C.P. March 1989 tumor or with relatively large nuclei (mean nuclear area > 95 nm 2 ) should be treated more aggressively. Actually, a minority of 3 patients included in this study received radiotherapy (n = 27) or cytostatic bladder instillation (n = 3) in addition to transurethral resection of their primary tumor. Despite this, their clinical outcome did not appear to be significantly more favorable, and, also, in multivariate analysis additional treatment did not contribute significantly to the prognosis. Although thesefindingsmight suggest that additional radiotherapy or intravesical chemotherapy is ineffective, this conclusion is not justified because therapy was administered to a selected patient group, which might have had a worse prognosis without additional therapy. It is concluded that selective nuclear morphometry and DNA flow cytometry in contrast to the classic prognosticators are rather sensitive predictors of recurrent carcinoma and progressive recurrence. Because the methods result in a considerable overlapping in the determination of the cases at risk, combined application of both techniques does not improve the prognostication, except in case of failure of one of them. Practically, the findings suggest that patients presenting with a tumor with large nuclei (mean nuclear area > 95 urn 2 ) or aneuploid DNA values should be treated more aggressively. Future trials will be required to evaluate the preventive role of treatment in addition to transurethral resection of the primary tumor. References 1. Althausen AF, Prout GR Jr, Daly JJ: Noninvasive papillary carcinoma of the bladder associated with carcinoma in situ. J Urol 1976;116:575-58. 2. Barnes R, Hadley H, Dick A, Johnston O, Dexter J: Changes in grade and stage of recurrent bladder tumors. J Urol 1985; 118: 177-178. 3. Bjelkenkrantz K, Herder A, Grontoft O, Stal O: Cytophotometric characterization of the WHO grades of transitional cell neoplasms. Pathol Res Pract 1982;174:68-77. 4. Blomjous CEM, Schipper NW, Baak JPA, Van Galen EM, De Voogt HJ, Meijer CJLM: Retrospective study of prognostic importance of DNA flow cytometry of urinary bladder carcinoma. J Clin Pathol 1988;41:21-25. 5. Blomjous CEM, Smeulders AWM, Vos W, Baak JPA, Van Galen EM, Meijer CJLM: A comparative study in morphometric grading of transitional cell carcinoma of the urinary bladder. Anal Quant Cytol Histol 1988 (in press). 6. Blomjous CEM, Vos W, Schipper NW, Baak JPA, De Voogt HJ, Meijer CJLM: The prognostic significance of selective nuclear morphometry in urinary bladder carcinoma (submitted for publication). 7. Chin JL, Huben RP, Nava E, et al: Flow cytometric analysis of DNA content in human tumors and irrigation fluids. Cancer 1985;56:1677-1681. 8. Collste LS, Devonec M, Darzynkiewiicz Z, et al: Bladder cancer diagnosis by flow cytometry. Correlation between cell samples from biopsy and bladder irrigation fluid. Cancer 198;45:2389-2394. 9. Coon JS, Weinstein RS, Summers JL: Blood group precursor T- antigen expression in human urinary bladder carcinoma. Am J Clin Pathol 1982;77:692-699. 1. Dalesio O, Schulman CC, Sylvester R, et al: Prognostic factors in superficial bladder tumors. A study of the European Organization for Research on Treatment of Cancer: Genitourinary Tract Cancer Cooperative Group. J Urol 1983;129:73-733. 11. Decenzo JM, Howard P, Irish CE: Antigenetic deletion and prognosis of patients with stage A transitional cell bladder carcinoma. J Urol 1975;114:874-878. 12. De Sanctis PN, Tannenbaum M, Tannenbaum S, Olssen C: Morphologic quantitation of nuclear size in various grades of transitional cell carcinoma of urinary bladder. Urology 1982;2:196-199. 13. Fitzpatrick JM, West AB, Butler MR, Lane V, O'Flynn JD: Superficial bladder tumors (stage pta, grade 1 and 2): the importance of recurrence pattern following initial resection. J Urol 1986; 135: 92-922. 14. Fossa SD, Kalhus O, Scott-Knudsen O: The clinical and histopathological significance of Feulgen DNA values in transitional cell carcinoma of the human urinary bladder. Eur J Cancer 1977; 13: 1155-1162. 15. Harmer MH: TNM classification of malignant tumours. 3rd ed. Geneva: International Union against Cancer, 1978, pp 113-117. 16. Hedley DW, Friedlander ML, Taylor IW, Rugg CA, Musgrove EA: Method for analysis of cellular DNA content of paraffin-embedded pathological material using flow cytometry. J Histochem Cytochem 1983;31:1333-1335. 17. Helander K, Hofer PA, Holmberg G: Karyometric investigations on urinary bladder carcinoma, correlated to histopathological grading. Virchows Arch [Pathol Anat] 1984;43:117-125. 18. Heney NM, Nocks BN, Daly JJ, et al: Ta and Tl bladder cancer: location, recurrence and progression. Br J Urol 1982;54:152-157. 19. Lamb D: Correlation of chromosome counts with histological appearances and prognosis in transitional cell carcinoma of bladder. Br Med J 1967;1:273-277. 2. Lutzeyer W, Ruebben H, Dahm H: Prognostic parameters in superficial bladder cancer: an analysis of 315 cases. J Urol 1982; 127: 25-252. 21. Melicow MM: Tumors of the urinary bladder: a clinicopathological analysis of 2,5 specimens and biopsies. J Urol 1955;74:498-53. 22. Mostofi FK, Sobin LH, Torloni H: Histological typing of urinary bladder tumours. In: WHO, eds. International histological classification of tumours, no. 1. Geneva: World Health Organization. 1973:17. 23. Ooms ECM, Kurver PHJ, Veldhuizen RW, Alons CL, Boon ME: Morphometric grading of bladder tumors in comparison with histologic grading by pathologists. Hum Pathol 1983;14:144-15. 24. Pauli BU, Cohen SM, Alroy J, Weinstein RS: Desmosome ultrastructure and the biological behavior of chemical carcinogen induced urinary bladder carcinomas. Cancer Res 1978;38:3276-3285. 25. Pauwels RPE, Smeets AWGB, Schapers RFM, Geraerdts JPM, Debruyne FMJ: Grading in superficial bladder cancer. Cytogenetic classification. Br J Urol 1988;61:135-139. 26. Schade ROK, Swinney J: Pre-cancerous changes in bladder epithelium. Lancet 1968;2:943-946. 27. Spooner ME, Cooper EH: Chromosome constitution of transitional cell carcinoma of the urinary bladder. Cancer 1972;29:141 1412. 28. Wiley EL, Mendelsohn G, Droller M, Eggleston JC: Immunoperoxidase detection of carcinoembryonic antigen and blood group substances in papillary transitional cell carcinoma of the bladder. J Urol 1982;128:276-28. 29. Srinivas V, Kiruluta HG: ABO(H) isoantigens in bladder tumours: a new technique of quantitative analysis. J Urol 1984;131:245 248. 3. Tribukait B, Gustafson H, Esposti PL: The significance of ploidy and proliferation in the clinical and biological evaluation of bladder tumours: a study of 1 untreated cases. Br J Urol 1982;54: 13-135. 31. Wijkstroem H, Granberg-Oehlman I, Tribukait B: Chromosomal and DNA patterns in transitional cell bladder carcinoma. A comparative cytogenetic and flow cytofluorometric DNA study. Cancer 1984;53:1718-1723.