Prognostic value of the proliferative index determined by Ki-67 immunostaining in superficial bladder tumors

Prognostic Value of the Proliferative Index Determined by Ki-67 Immunostaining in Superficial Bladder Tumors CHRISTIAN PFISTER. MD, LOUIS LACOMBE, MD, MARIE-CLAUDE VEZINA, LYNNE MOORE, MSc, HI~LI~NELARUE, PHD, BERNARD TETU, MD, FRAN(~OIS MEYER, MD, DSc, and YVES FRADET, MD ^

T h e biological behavior o f urothellal carcinomas remains unpredictable. The objective o f this study was to d e t e r m i n e the prognostic value o f Ki-67 i n d e x in superficial papillary bladder t u m o r s and to correlate it with the S-phase fraction (SPF) m e a s u r e d by flow cytometry. T h r e e h u n d r e d n i n e t e e n patients with newly diagnosed superficial (pTa, pT1) bladder t u m o r s were included between S e p t e m b e r 1990 and April 1992. Patients with bladder carcinoma in situ alone were excluded. We observed 255 pTa t u m o r s a n d 64 pT1 t u m o r s , whereas 111 lesions were classified as grade G1 a n d 208 as grade G2-G3. Ki-67 i m m u n o s t a i n i n g was p e r f o r m e d on p a r a f f i n - e m b e d d e d material using a 3-step i m m u n o p e r o x i d a s e p r o c e d u r e with the m u r i n e monoclonal antibody MiB1. T h e relation between Ki-67 expression and prognostic variables (stage, grade, t u m o r size, multifocality, age, and sex~ was investigated by the chi-square test. Cox regression was u s e d to describe the association between Ki-67 and t u m o r recurrence in 308 patients with follow-up while adjusting for potentially c o n f o u n d i n g prognostic variables. T h e frequency o f high Ki-67 expression (-->10%) increased with stage ( P = .005) and grade (P = .001), but not with t u m o r size or multifocality. Two h u n d r e d one patients e x p e r i e n c e d tumor recurrence in a m e d i a n follow-up o f 68 m o n t h s . Stage, grade,

t u m o r size, and multifocality were all i n d e p e n d e n t predictors of recurrence. Ki-67 i n d e x greater t h a n 10% was f o u n d to be an i n d e p e n d e n t predictor o f t u m o r recurrence a m o n g patients with t u m o r s larger than 3 c m in diameter [HR = 2.05, CI = 1.18-3.55), but not those with smaller size tumors. With regards to the DNA index, a significant but weak correlation was observed between Ki-67 express i o n and the SPF ( S p e a r m a n ' s c o r r e l a t i o n c o e f f i c i e n t = 0.23, P = .004). In addition, anenploid t u m o r s h a d significantly higher expression o f Ki-67 (22.5%) than diploid t u m o r s (10.1%) (P = .0006). Moreover, patients with DNA aneuploid bladder t u m o r s were m o r e likely to have m o r e than 10% Ki-67-positive cells than those with diploid tumors. In patients with newly d i a g n o s e d pTa or pT1 bladder tumors, a Ki-67 index above 10% is an i n d e p e n d e n t predictor o f shorter time to recurrence only in those with t u m o r s larger t h a n 3 cm. HUM PATHOL 30:1350-1355. Copyright © 1999 by W.B. Saunders Company Key words: bladder t m n o r s recurrence, Ki-67 index, DNA ploidy. Abbreviations: SPF. S-phase fraction; MAb, m o n o c l o n a l antibody; T U R , transurethral resection; PBS, p h o s p h a t e - b u f f e r e d saline; CI, confidence interval; HR, hazards ratio.

T h e natural history of bladder cancer remains unpredictable. Most urothelial carcinomas follow a relatively benign course. However, after initial endoscopic resection of papillary superficial tumors, m o r e than half of patients develop recurrences within 2 years, and the overall cumulative risk of muscle-invasive cancer is estimated at 7% after 5 years. 1"~Clinical prognostic variables contributing to recurrence, invasion, and survival have b e e n investigated in several cohort studies. 4-6 Primary t u m o r multiplicity, size of the tumor, pathological stage, a n d histological grade were f o u n d to be i n d e p e n d e n t predictors of earlier r e c u r r e n c e in a multivariate analysis. 7 T h e r e is growing interest in establishing clinical m a n a g e m e n t strategies, including intravesical therapy and frequency of periodic control cystoscopies, based on prognosis at initial diagnosis. 7,s Molecular p h e n o t y p i n g provides a n o t h e r dimension to the characterization of the biological potential of tumors that may help better predict their clinical outcome. Although several t u m o r markers have been studied in bladder cancer, few have b e e n evaluated in well-defined cohorts of patients to d e t e r m i n e their

i n d e p e n d e n t prognostic value above known clinicopathologic criteria. With regard to biological parameters of prognostic significance, the study of D N A content and S-phase fraction (SPF) by flow cytometry has b e e n recognized as i m p o r t a n t in several t u m o r s 9 n Gerdes et al first described the nuclear antigen Ki-67 related to cell proliferation and expressed in the S, G2, and M phases of the cell cycle but absent in Go p h a s e ? 2 T h e m o n o c l o n a l antibody (MAb) MiB1 reacts with an epitope of the antigen that is stable on formalin-fixed tissues, t3 T h e ability to measure the growth fraction of tumors by i m m u n o h i s t o c h e m i s t r y is a simple, rapid and reproducible m e t h o d m o r e suitable for routine laboratory practice. In urinary bladder cancer, several groups have r e p o r t e d a correlation between the Ki-67 index and biological aggressiveness of the tumors. 14-21 Most studies, however, were p e r f o r m e d on a limited n u m b e r of h e t e r o g e n e o u s tumors, frequently including muscle invasive cancers. Moreover, there are few data on the relation of Ki-67 index a n d DNA S-phase fraction in solid tumors. We tested the hypothesis that the proliferative index of papillary bladder tumors m e a s u r e d by Ki-67 (MiB1) i m m u n o s t a i n i n g would provide additional p r o g nosuc information above pathologic criteria. To this end, we evaluated the relation of MiBl-labeled cell fraction to pathologic parameters and t u m o r recurrence using a b a n k of newly diagnosed papillary superficial (pTa, pT1) bladder tumors collected in a previous study in which the patients were submitted to a system-

From the Centre de Recherche e n Canc&rologie, Universitfi Laval, Qu6bec. Canada. Accepted for publication July 12. 1999. Address c o r r e s p o n d e n c e a n d reprint requests to Yves Fradet. MD. FRCSC, Centre de R e c h e r c h e en Canc6rologie, Pavillon H6tel Dieu de Qu6bec, 10, Rue Mac-Mahon. Q u e b e c G1R 2J6, Canada. Copyright © 1999 by W.B. Saunders Company 0046-8177/99/3011-0013510.00/0

1350

PROLIFERATIVEINDEXIN BLADDERTUMORS(Pfister et al) atic follow-up a n d received n o a d d i t i o n a l t r e a t m e n t after resection. 7 I n a d d i t i o n , the Ki-67 i n d e x was correl a t e d with D N A p l o i d y a n d SPF as m e a s u r e d by flow c y t o m e t r y o n a s u b s e t o f these t u m o r s .

MATERIAL AND MI=THODS Tissue a n d Patient Characteristics The 381 patients enrolled in this study had newly diagnosed superficial bladder tumors and were admitted for the first transurethral resection between September 1990 and April 1992 in 15 hospitals of the province of Quebec. Briefly, the patients' mean age was 65.1 years and the male-to-female ratio was 3:1. Criteria for patient eligibility included presence of histologically confirmed pTa or pT1 urothelial carcinomas. Patients with bladder carcinoma in situ alone were excluded in this series. Random biopsies of the bladder mucosa were performed in 236 patients at the time of transurethral resection (TUR). Associated carcinoma in situ was observed in only 5 cases and thus was not considered as a variable in the statistical analysis. After the initial TUR, cystoscopies were performed at 3-month intervals for 2 years, at 6-rnonth intervals for 2 years, and yearly thereafter. This schedule was repeated for any patient with tumor recurrence during the follow-up period. Small recurrent tumors were fulgurated immediately, and patients with larger tumors were admitted for new endoscopic resection u n d e r general anesthesia. In our series, tumor progression was defined as occurrence Of muscle invasion (> = pT2), metastasis, or death by bladder cancer. Patients received no additional intravesical treatment after initial resection until the first recurrence. 7 The immunostaining for Ki-67 was performed in a central laboratory on unstained slides of formalin-fixed, paraffin-embedded material that was available for 319 patients, all obtained by transuretral resection. M1 pathology specimens were reviewed and classified by a central pathologist (B.T.). According to the International Union Against Cancer (UICC) Tumor Node Metastases (TNM) classification, 255 lesions were determined as superficial papillary tumors (pTa), whereas 64 tumors invaded the lamina propria (pT1). According to the World Health Organization last recommendations, 22 we observed 111 tumors classified as low malignant potential papillary neoplasm (grade G1) and 208 tumors as louT- and high-grade papillary carcinoma (grades G2-G3). In cases with multiple tumors, the size of the largest tumor was always the one reported. At initial resection, most primary tumors were single (66.5%, 3 undetermined) and had a diameter of less than 3 cm (63.2%, 20 undetermined).

Immunohistochemical Staining for Ki-67 Briefly, 5-pro-thick sections were deparaffnized with toluene and rehydrated with graded alcohols. The sections in a citrate buffer (0.01 mol/L, pH 6) were heated in a microwave oven for 17 minutes at 700 W, then cooled at room temperature for 20 minutes. A 3-step immunoperoxidase procedure (Ultra Steptavidine kit, Signet Ldt., Dedham, MA) was used after a phosphate-buffered saline (PBS) wash. After being blocked with 2% normal horse serum, the sections were incubated at room temperature for 20 minutes in a 1:50 dilution of the murine monoclonal antibody MiB1 (Dako Ldt., Mississauga, Ontario) against the Ki-67 antigen. After a further PBS (IX) wash, they were then incubated for 20 minutes with peroxidase conjugated rabbit-anti-mouse antibody. Sections were washed twice more, and peroxidase activity was revealed using hydrogen peroxide-diaminobenzidine solution. Finally, sections were counterstained with

Mayer's haematoxylin before mounting. A negative control was performed for each batch of slides by omission of the primary antibody and tonsilar tissue was used as the positive control. A square grid (10 × 10 mm) in the ocular was used at a 40× magnification to delineate the area in which both the stained and unstained cells were counted. Nuclei in morphologically malignant cells were considered positive for the Ki-67 antigen when they showed dark brown granular staining. Random fields in well-preserved areas of tumor were examined and at least 1,000 malignant cells were counted in each case by 2 i n d e p e n d e n t observers (C.P. and B.T.). Cell counting was performed Without prior knowledge of the tumor features, and the Ki-67 result was reported as an index of percent of tumor cells stained with the antibody.

Flow Cytometry Technique A paraffin block of tissue in formalin was available for DNA-FCM in 213 cases; however, SPF analysis was performed for 165 patients by the central study laboratory. As previously reported, ploidy and S-phase fraction were determined on single parameter histograms With the use of ModFit (Very Software House, Topsham, ME). 1° G0/G1 and G2/M were defined from a Gaussian curve, and SPF was evaluated with the trapezoid model. Because of the lack of standard DNA reference on paraffin materiel, the first peak of an aneuploid histogram was interpreted as an internal diploid control and, in the presence of a single pea k, the histogram was interpreted as diploid. The DNA index was expressed as the ratio of the peak channel n u m b e r of the tumor cell population on the control diploid population.

Statistical Analysis To explore different cut-off points for Ki-67 expression, the proportion of patients with histological grade G2-G3 tumors was compared for different categories of the Ki-67 index (<10%, 10% to 14%, 15% to 19%, 20% to 24%, ->25%). A P-value for trend was calculated using logistic regression whereby a linear variable representing the 5 categories of Ki-67 expression was entered in the model. This analysis did not lead us to change the cut-offvalue of 10% suggested in the literature. 16,17'23 The relation between Ki-67 expression and the prognostic variables (tumor stage, histological grade, tumor size, multifocality, age, and sex) was investigated by the chi-square test. To explore the relation of Ki-67 expression to the risk of tumor recurrence. Cox proportional hazards models were used to generate hazards ratios (HRs) and their 95% confidence intervals (CI) while adjusting for potentially confounding prognostic variables such as tumor stage, histological grade, tumor size, and multifocality, defined as multiple simultaneous papillary tumors. Age and sex of the patient were not added to the model, because they were not related to tumor recurrence. A similar analysis was performed among the subpopulafion of patients with a tumor diameter o v e r 3 cm at diagnosis. In this population, the relation of the Ki-67 index to tumor recurrence was modeled with adjustment for tumor grade and multifocality only because stage, sex, and age were not found to be confounding factors. Kaplan-Meier survival curves were generated to describe the survival difference between tumors With a high or low expression of Ki-67 for 2 populations: patients with an initial tumor diameter 3 cm or smaller and those with larger than 3 cm. Because only 18 patients had invasive progression, we

1351

HUMAN PATHOLOGY

Volume30, No. 11 (November 1999) TABLE 2. The Relation of Prognostic Variables to the Ki-67 Index: P-Values are Taken From the Chi-Square Distribution (*statisticatly significant at ~ = 0,05)

were unable to perform multivariate analysis for this parameter. To describe the relationship between Ki-67 expression and cancer progression, we used the chi-square distribution. We compared the median Of Ki-67 expression for patients with diploid and those with aneuploid minors using the Wilcoxon rank sum test for the 213 available patients. The correlation between Ki-67 expression and the SPF was described for 165 cases using SPearman's correlation coefficient.

Prognostic Stage Grade

RESULTS

Tumor size

Ki-67 Index a n d Pathological Features

Multifocality

T h e Ki-67 index of the tumors varied widely f r o m 0% to 70% (mean, 16.9% _+ 12.4%). In 222 patients, the Ki-67 index value was m o r e than 10%. T h e proportion of patients with grade G2-G3 tumors increased with the value of Ki-67 expression. A m o n g patients with less than 10% Ki-67 expression, the p r o p o r t i o n of grade G2-G3 tumors was 49.5%, and for those with 25% or greater Ki-67 expression, this p r o p o r t i o n was 80.5%. T h e Pvalue for trend was statistically significant at .0001 (Table 1). With regard to the relation of prognostic variables to the Ki-67 index in o u r series, Ki-67 was expressed at over 10% for 65.9% of pTa tumors and 84.4% of pT1 tumors ( P = .004). Moreover, 76.9% of patients with grade G2-G3 h a d Ki-67 expression of over 10% comp a r e d with only 55.9% of grade G1 patients (P = .001). A greater p r o p o r t i o n of patients with large tumors, multiple tumors, those aged over 70 years and female had Ki-67 index expressed over 10% when c o m p a r e d with their counterparts, but these differences did not reach statistical significance (Table 2). Ki-67 Index a n d Tumor Recurrence O f the 319 patient s for w h o m paraffin-embedded material was available to p e r f o r m Ki-67 index evaluation, 308 had follow-up data available. The median follow-up was 67.7 months (range, 2.1 to 92.2). We observed 201 cases of bladder t u m o r recurrence. Eighteen patients h a d t u m o r progression defined as muscle invasion (>-pT2), or metastasis, or death by bladder cancer. Because of missing data in few patients, the multivariable analysis included 287 patients with recurrence data and prognostic variables (Table 3). Several prognostic variables were i n d e p e n d e n t predictors of the risk of t u m o r recurrence. Patients with p T l tumors h a d 47% greater risk of r e c u r r e n c e when c o m p a r e d with those with pTa tumors (95% CI = 1.02 to 2.12). Superficial bladder tumors over 3 cm d i a m e t e r h a d 40% TABLE 1. Proportion of Patients With Histological Grade G2-G3 Tumors Were Compared for Different Categories of the Ki-67 Index Ki-67 Index Value

<10

10-14

15-19

20-24

>=25

Patients (n) % Grade G2-G3

97 49.5

36 58.3

53 62.3

46 78.3

87 80.5

NOTE. A P-value for trend ( P = .0001) was calculated using logistic regression whereby a linear variable representing the 5 categories of Ki-67 expression was entered in the model.

Age Sex

Variables

n

Ki-67 Index ->10% (%)

pTa pT1 G1 G2-G3 < = 3 cm > 3 cm Yes No <70 years > = 7 0 years Male Female

255 64 111 208 189 110 210 106 189 130 236 83

65.9 84.4 55,9 76,9 66.7 72.7 69.1 71.7 67.2 73.1 67.8 74.7

PValue .004* .001" .28 .63 .26 .24

greater risk of r e c u r r e n c e (95% CI = 1.02 to 1.86), and patients with multiple tumors h a d over twice the risk of recurrence when c o m p a r e d with those with only 1 t u m o r at diagnosis (95% CI = 1.53 to 2.76). In this population, high grade increased the risk of t u m o r r e c u r r e n c e with b o r d e r l i n e statistical significance (HR = 1.34, 95% CI = 0.96 to 1.88). T h e Ki-67 index was not an i n d e p e n d e n t predictor of t u m o r r e c u r r e n c e for the overall cohort of patients. Neither age n o r sex of the patient were related to the risk of recurrence. Survival curves show that a m o n g patients with b l a d d e r tumors smaller than 3 cm in diameter, survival is not associated with the Ki-67 index (Fig 1A). However, for the population of patients with large tumors (>-3 cm), those with Ki-67 expression over 10% have a signific a n @ lower survival than their counterparts (Fig 1B). W h e n limiting the analysis to the 105 patients with bladder tumors larger than 3 cm, a Ki-67 index > 1 0 % was an i n d e p e n d e n t predictor of t u m o r r e c u r r e n c e and was associated with over twice the risk of r e c u r r e n c e (HR = 2.05, 95% CI = 1.18 to 3155) (Table 4). In this population, histological grade G2-G3 (papillary carcinomas) was also associated with a greater risk of recurrence (HR = 2.04, 95% CI = 1.13-3.68), a n d patients with m o r e than 1 t u m o r at diagnosis had an increased risk of borderline statistical significance (HR = 1.53, 95% CI = 0.96 to 2.44). Neither t u m o r stage, age, n o r sex of the patient were associated with the risk of recurrence. A cumulative risk index for r e c u r r e n c e TABLE 3.

The Relation of Prognostic Variables to the Risk of Tumor Recurrence (n : 287)

Prognostic Variables

Hazard Ratios

95% CI

Ki-67 Index > 10 % Stage pT1 Grade G2/G3 Tumor size > = 3 cm Multifocality Patient age > =70 years Patient sex female

1.11 1.47" 1.34 1.38" 2.06* 1.10 0.97

0.79-1.54 1.02-2.12 0.96-1.88 1.02-1.86 1.53-2.76 0.82-1.49 0.70-1.35

NOTE. Hazard ratios (HR) and their 95% confidence intervals (CI), taken from Cox proportional hazards models, are adjusted for stage, grade, tumor size, and multifocality. *Statistically significant at e~ = .05.

1352

PROLIFERATIVE INDEX IN BLADDER TUMORS (Pfister et al) A

100-

100 T

78.1

80

75. 60.6

60

©

==':001

5O'L

©

56,3

"-h ,

3°t

25-

20

,<

i:L:d : ; i+,~

14.3

, : : 2' j ~¢,2i

10

0

# patients

o

~

Grade Ki67

Follow-up (years)

B

!

75-

7 G1 <10%

--H-16 G1 >10%

22 G2/G3
62 G2/G3 >10%

Ki-67 expression below 10% (6.6%) and those with over II

10% (5.6%). Ki-67 Index a n d D N A Flow Cytometry

50nu

"~/=.,

r

FIGURE 2. Percent recurrence at 2 years of follow-up a m o n g patients with initial tumor diameter > 3 cm.

I00-

t -~

t::z

01

25 t

With regard to the DNA index, a significant positive correlation was observed between Ki-67 expression and the S-phase fraction (Spearman's correlation coefficient = 0.23 and P = .004). In addition, aneuploid tumors had significantly higher expression of Ki-67 (median = 22.5%) than diploid tumors (median = 10.1%) ( P = .0006). Furthermore, 42.4% of diploid tumors had Ki-67 expression over 10%, compared with 82.5% o f a n e u p l o i d tumors (P = .02).

-1 --~

L---I

ol

0

1 I

~

~

~

~

~

;

Follow-up (years)

FIGURE I. (A) Recurrence-free survival a m o n g patients with tumors under 3 cm in diameter (n = 183) for Ki-67 expression under 10% (plain line) a n d a b o v e 10% (broken line). P v a l u e of the log-rank test = .91. (t3) Recurrence-free survival a m o n g patients with tumors over 3 cm in diameter (n = 107) for Ki-67 expression under 10% (plain line) and a b o v e 10% (broken line). P value of the log-rank test = .02.

within 2 years can be calculated for tumors larger than 3 cm according to whether the tumors have greater than 10% Ki-67-positive cells or are grade G2- G3 (Fig 2). No statistical difference (P = .72) was observed in the proportion of progression among patients with TABLE 4. The Relation of Prognostic Variables to the Risk of Tumor Recurrence Among Patients With Tumor Diameter >3 Cm (n = 105) Prognostic Variables Ki-67 index >10% Stage pT1 Grade G2/G3 Multifocality Patient age Patient sex

Hazard Ratios

95 % CI

2.05* 1.04 2.04* 1.53 1.15 0.81

1.18-3.55 0.61-1.77 1.13-3.68 0.96-2.44 0.72-1.84 0.46-1.41

NOTE. Hazard ratios (HR) and their 95% confidence intervals (CI), taken from Cox proportional hazards models, are adjusted for grade and mulfifocality. *Statistically significant at c~ = .05.

DISCUSSION

The proliferative potential of tumors is one biological characteristic that has been associated with poor prognosis in many tumor types. Bladder tumor proliferation has been extensively studied by different methods, including the mitotic index. BrdUrd incorporation, and flow cytometry3 °,H,23,24More recently, antibodies against antigens such as Ki-67 and proliferating cell nuclear antigen expressed on cells in active proliferation provide a more practical m e t h o d to assess clinically the growth fraction of tumors. Few studies have correlated the Ki-67 index with other methods such as mitotic index and BrdU labeling index. 24-26 We observed a significant correlation between Ki-67 index a n d the S-phase fraction measured by DNA fl0w cytometry. However, the Ki-67 values were always higher than the S-phase fraction, a finding that is expected because Ki-67 nuclear antigen is expressed on cells in the S phase, G2-M, and most of G1 cells. ]2'27 Our study also confirmed, in initial papillary tumors, a significantly higher Ki-67 index in aneupl0id tumors. Aneuploidy is commonly associated with anomalies of the cell-cycle regulators p53 and Rb. Wright et al 2s reported that Ki-67 index for tumors lacking detectable Rb protein was significantly greater than for those showing Rb expression. Likewise, p53-positive tumors had greater Ki-67

1353

HUMAN PATHOLOGY

Volume 30, No. 11 (November 1999)

growth fractions in comparison with p53-negative tumors. 28 Comparative analysis of Ki-67 using MiB1 antibody and p53 expression showed that the relationship between these two markers is stage i n d e p e n d e n t , z9 High MiB1 and p53 overexpression were associated with an increased probability of t u m o r r e c u r r e n c e b u t were not better indicators for cancer progression than aneuploidy alone. 29-3° Several reports have analyzed the Ki-67 index of bladder carcinoma. Most studies have used the initially described Ki-67 MAb, which requires testing frozen sections of tumors. 17,18,2°,24 In our study, we used the MiB1 MAb, raised against r e c o m b i n a n t parts of the partially cloned Ki-67 molecule, and reactive with an epitope detectable in formalin-fixed tissuesJ 3,31 A comparative study of MAbs Ki-67 and MiB1 in breast carcinomas showed no difference between results with MiB1 in frozen or formalin-fixed samples, b u t a m u c h higher average index with MiB1 (21%) than with Ki-67 (13%) in frozen specimens. 32 This may explain the slightly higher Ki-67 index observed in o u r study for pTa and pT1 tumors of 16.9% c o m p a r e d with earlier reports. According to O k a m u r a et al, 16 the Ki-67 index was highly correlated with the grade of urothelial carcinomas, although the values for growth fractions in grade 2 tumors varied widely. T h r e e different groups were suggested: less than 5% indicated indolence and was equal to grade 1, Ki-67 index value of 5% to 15% represented the intermediate range and in contrast a value higher than 15% predicted for a highly malignant potential. 16 Mulder and coworkers r e p o r t e d m o r e than 10% of Ki-67 positive cells in all deeply infiltrating urothelial carcinomas (pT2 to pT4). However, a high p r o p o r t i o n of the noninfiltrating papillary bladder tumors (pTa) h a d very low indices, whereas the remaining noninfiltrating urothelial carcinomas showed a high proliferative activi.ty. 17 Stavropoulos et al observed a significant difference in Ki-67 growth fractions between pTa and pT1 tumors and. in contrast, the Ki-67 index did not significantly differ between p T ! a n d pT2.18 In our population of 319 newly diagnosed superficial papillary bladder tumors, using the same cut-off value of 10% for the Ki-67 index, we also observed a statisti, cally significant correlation between Ki-67 growth fraction and histological grade and pathological stage of the tumor. O n e of the m a i n concerns with papillary b l a d d e r tumors is to identify those with a potential to progress toward muscle invasive disease with recurrences. Re, centlv, Mostofi and colleagues p r o p o s e d a new classification of urothelial neoplasms of the urinary bladder. 22 Briefly, the histological features of urothelial papillary lesions were classified in 3 different groups: papillary neoplasm of low malignant potential, with the risk of developing recurrence usually of a similar histology (grade G1); low-grade papillary carcinomas, which may invade the lamina p r o p r i a and have a low risk ( < 5 % ) of further progression (grade G2); high-grade papillary carcinomas, where the risk of progression becomes higher than low-grade lesions, varying f r o m 15% to 40% (grade G3~. This is why in the c u r r e n t study, we

categorized tumors in grade G1 versus grade G2-G3 in correlation with the Ki-67 index. T h e m e d i a n Ki-67 index for G1 tumors was 10% (0 to 50%) and 20% (0 to 70%) for G2-G3 tumors. However, there was significant overlap o f Ki-67 indexes between G1 and G2-G3 tumors, particularly in cases with Ki-67 < 1 0 % , which were c o m p o s e d of 50°A G2-G3 tumors. T h e p r o p o r t i o n of patients with grade G2-G3 tumors increased with the value of Ki-67 expression (P = .0001), but 20% of cases with Ki-67 > 2 5 % were G1 tumors (Table 1). Thus, papillary neoplasms of low m a l i g n a n t potential cannot be distinguished f r o m papillary carcinomas on the basis of growth fraction. Nevertheless, the m e d i a n Ki-67 value observed in o u r G1 tumors provides further rationale to the c o m m o n use of a 10% cut-off value in the literature.16,17.23 Beyond their potential biological significance, the clinical usefulness of t u m o r markers is to provide prognostic information b e y o n d that obtained with standard pathological variables. Ki-67 index has b e e n strongly correlated with invasive bladder cancer. In a series of 59 patients, Bush et al r e p o r t e d a m e d i a n Ki-67 index value of 25.3% with a significant positive correlation between Ki-67 and b o t h t u m o r stage (P = .002) and histological grade (P = .001). Moreover, patients with invasive disease (pT2 to pT4) had higher Ki-67 indices (median, 26.3%) than those with disease confined to the bladder (CIS, pTa, pT1) (median, 12.5%).19 In a retrospective study of 31 formalin-fixed radical cystectomy specimens (11 pT1, 1 pT2, 10 pT3a, 4 pT3b, 5 pT4), Tsuji and colleagues observed a m e d i a n Ki-67 index value of 28.6%. No significant correlation was found between Ki-67 index and t u m o r stage. However, patients with tumors of a high Ki-67 index had a worse prognosis than those with tumors of a low Ki-67 index. 2° In a recent series, where a Ki-67 cut-off limit of 20% was chosen, Rey and coworkers observed a correlation with different TNM categories (P = .048 ). Survival was better for the patients with low proliferating tumors than for high proliferative ones, and the authors concluded that the Ki-67 index was able to detect high-risk patients that could not be cured by radical surgery alone, raising the n e e d for adjuvant chemotherapy. 21 These different data, using a higher Ki-67 index cut-off value, are usually in relation with radical cystectomy specimens and cannot be c o m p a r e d with o u r study o f p T a and pT1 tumors, where only 18 patients h a d invasive progression. Although this study could n o t address the value of Ki-67 index to predict cancer progression, it was perfectly designed to d e t e r m i n e its predictive value for t u m o r recurrence. Only p r i m a r y superficial papillary lesions were included in o u r prospective study, a n d no additional intravesical t r e a t m e n t was p e r f o r m e d after initial resection until the first recurrence. 7 Multivariate analysis showed that Ki-67 index was n o t an independ e n t predictor of recurrence for the entire cohort. However, for patients with an initial t u m o r larger than 3 cm, a Ki-67 index > 1 0 % was an i n d e p e n d e n t predictor of shorter time to recurrence. In the same category of tumors, having a G2-G3 versus a G1 t u m o r was also an i n d e p e n d e n t prognostic variable. Consequently, it is

1354

PROLIFERATIVE INDEX IN BLADDER TUMORS (Pfister et al)

possible to cumulate these 2 factors to distinguish patients with tumors larger than 3 cm at very low risk (14%) of r e c u r r e n c e if G1 and Ki-67 less than 10% from those at intermediate risk (60%) with one of these features, and those at highest risk (78%) if G2-G3 and Ki-67 greater than 10% (Fig 2). Such information has direct implications for a more appropriate use of intravesical therapy. In conclusion, the results of this cohort study suggest that the determination of Ki-67 index should be incorporated in the analysis of initial papillary bladder tumors larger than 3 cm, which represent approximately one third of all cases of superficial disease.

Acknowledgment. T h e a u t h o r s are grateful for t h e technical assistance given by N a n c y R o b e r g e a n d P i e r r e Simard. S u p p o r t e d bY grants f r o m t h e N a t i o n a l C a n c e r Institute (CA 47526) o f USA; a n d C h r i s t i a n Pfister received a fellowship f r o m ' T Association Frangaise d ' U r o l o g i e , le G r o u p e m e n t des E n t r e p r i s e s Frangaises d a n s la L u t t e c o n t r e le Cancer, la Ligue C o n t r e le C a n c e r de H a u t e - N o r m a n d i e , la Soci~t~ Frangaise du Cancer." REFERENCES 1. Parmar MKB, Freedman LS, Hargreave TB, et ah Prognostic factors for recurrence and follow-up policies in the treatment of superficial bladder cancer: Report from the British Medical Research Council Subgroup on Superficial Bladder Cancer (Urological Cancer Working Party).J Uro1142:284-288, 1989. 2. Koch M, Hill GB, McPhee MSi Factors affecting recurrence rates in superficial bladder cancer. J Natl Inst 76:1025-1029, 1986 3. Allard P, Bernard P, FradetY, et al: The early clinical course of primary Ta and T1 bladder cancer: A proposed prognostic index. BrJ Uro181:692-698, 1998 4. Kurth KEI, Denis L, Bouffioux C, et al: Factors affecting recurrence and progression in superficial bladder tumors. Eur J Cancer 31:1840-1846, 1995 5. Malmstrom PU, Busch C, Norlen BJ: Recurrence, progression and survival in bladder cancer.J Urol Nephro121:185-195, 1986 6. Kiemeney LAL, WitjesJA, Heijbroek RP, et al: Predictability of recurrent and progressive disease in individual patients with primary superficial bladder cancer.J Uro1150:60-64, 1993 7. Allard E Fradet Y, Tatu B, et al: Tumor-associated antigens as prognostic factors for recurrence in 382 patients with primary transitional cell carcinoma of the bladder. Clin Cancer Res 1:11951202, 1995 8. Reading J, Hall RR, Parmar MKB: The application of a prognostic factor analysis for Ta-T1 bladder cancer in routine urological practice. BrJ Urol 75:604-607, 1995 9. Koss LG, Czerniak B, Herz F, et al: Flow-cytometric measurements of DNA and other cell components in human tumours: A critical appraisal. HUM PATHOL20:528-548, 1989 10. T~m B, Allard P, Fradet Y, et al: Prognostic significance of nuclear DNA content and S-phase fractio n by flow-cytometry in primary superficial bladder cancer. HUM PATHOL27:922-926, 1996 11. De Vere White RW, Deitch AD: Evaluation o f DNA flow cytometry as a screening test for bladder cancer. J Cell Biochem 161:80-84, 1992 (suppl) 12. GerdesJ, Schwab U, Lemke H, et al: Production of a mouse monoclonal antibody reactive with a h u m a n nuclear antigen associated with cell proliferation, l n t J Cancer 31: ] 3-20, 1983

13. Key G, Becker MHG, Baron B, et al: New Ki-67 equivalent murine monoclonal antibodies (MiB1-3) generated against bacterially expressed parts of the Ki-67 cDNA containing three 62 base pair repetitive elements encoding for the Ki-67 epitope. Lab Invest 68:629-636, 1993 14. Fontana D, Bellina M, Gubetta L, et al: Monoclonal antibody Ki-67 in the study of the proliferative activity of bladder carcinoma. J Uro1148:1i49-1151, 1992 15. Krfiger S, Mfiller H: Correlation of morphometry, nucleolar organizer regions, proliferating cell nuclear antige n and Ki-67 antigen expression with grading and stagging in urinary bladder carcinomas. BrJ Uro175:480-484, 1995 16. Okamura K, Miyake K, Koshikawa T, et ai: Growth fractions of transitional cell carcinomas of the bladder defined by the monoclonal antibody Ki-67, J Uro1144:875-878, 1990 17. Mulder AH, Van HootegemJCSP, Sylvester R, et al: Prognos~ tic factors in bladder carcinoma: Histologic parameters and expression of a cell cycle related nuclear antigen (I~-67).J Pathol 166:37-43, 1992 18. Stravropoulos NE, Ioackim-Velogianni E, Hastazeris K, et ah Growth fractions in bladder cance r grade, category and recurrence rate of superficial lesions. BrJ Uro172:736-739, 1993 19. Bush C, Price P, Norton J, et ah Proliferation in human bladder carcinoma measured by Ki-67 antibody labelling: Its potential clinical importance. BrJ Cancer 64:357-360, 1991 20. Tsuji M, Kojima K, Murakami Y, et ah prognostic value of Ki-67 antigen and p53 protein in urinary bladder cancer: Immunohistochemical analysis o f radical cystect0mY specimens. Br J Urol 79:36%372, 1997 21. ReyA, Lara PC, Redondo E, et ah Ki-67 proliferation index in tumors of the urinary tract as related to established prognostic factors and long term survival. Arch Esp Ur01 51:204-210, 1998 22. Epstein JI, Amin MB, Reuter VR, et al: The World Health Organisation/International Society of Urological Pathology ConsenSus: Classification of urothelial (transitional cell) neoplasms of the urinary bladder. A m J Surg Pathol 22:2435-1448, 1998 23. Rudolph P, Peters J, Lorenz D, et al: Correlation between mitotic and Ki-67 labelling indices in paraffin embedded carcinoma specimens. Hu?x PATHOL29:1216-2222, 1998 24. Tsujihashi H, Nakanishi A, Matsuda H, et ah Cell proliferation of human bladder tumours determined by BrdUrd and Ki-67 immunostaining.J Uro1145:846-849, 1991 25. Lipponen PK, Eskelinen MJ: Cell proliferation of transitional cell bladder determined by PCNA/cyclin immunostaining: A histological descfiptionl Anficancer Res 12:57%583, 1990 26. Mellon K, Neal DE, Robinson MC, et al: Cell cycling in bladder carcinoma determined by monoclonal antibody Ki-67. Br J Uro166:281-285, 1990 27. Campanella R, Russo A, Plaja S, et ah Study of cellular DNA content by flow-cytometry in primary bladder carcinomas:Significance of monoclonal and multiclonal varieties of DNA aneuploidy. Eur Uro121:58-63, 1992 28. Wright C, Thomas D, Meilon K, et ah Expression of retinoblastoma gene product and p53 protein in bladder carcinoma: Correlation with Ki-67 index. BrJ Uro175:173-179, 1995 29. Pfister C, Buzelin F, Casse C, et ah Comparative analysis of MiBI and p 5 3 expression in human bladder tumours and their correlation with cancer progression. Eur Urol 33:278-284, 1998 30. Popov Z, Hoznek A, Colombel M, et ah The prognostic value of p53 nuclear overexpression and MiB1 as a proliferative marker in transitional cell carcinoma of the bladder. Cancer 80:1472-1481, 1998 31. Cohen MB, Waldman FM, Caroll PR, et ah Comparison of five histopathologic methods to assess cellular proliferation in transitional cell carcinoma of the urinary bladder. HuM PATNOL24:772-778, 1993 32. Barbaresghi M, Girlando S, Mauri FM, et al: Quantitative growth fraction evaluation with MiB1 and Ki-67 antibodies in breast carcinomas. AmJ Clin Pathol 102:272-175, 1994

1355