Urinary cytokeratin 20 mRNA expression has the potential to predict recurrence in superficial transitional cell carcinoma of the bladder

Cancer Letters 245 (2007) 121–126 www.elsevier.com/locate/canlet

Urinary cytokeratin 20 mRNA expression has the potential to predict recurrence in superficial transitional cell carcinoma of the bladder Frank Christoph a,*, Steffen Weikert a, Ingmar Wolff a, Martin Schostak a, Karim Tabiti b, Markus Mu¨ller a, Kurt Miller a, Mark Schrader a a

Department of Urology, Universita¨tsmedizin Berlin, Charite´-Campus Benjamin Franklin, Hindenburgdamm 30, 12200 Berlin, Germany b Roche Diagnostics, Penzberg, Germany

Received 31 October 2005; received in revised form 21 December 2005; accepted 22 December 2005

Abstract Higher levels of cytokeratin 20 (CK 20) mRNA are expressed in malignant urothelial tissue compared to normal tissue. We determined the CK 20 mRNA expression in urine from patients with transitional cell carcinoma (TCC) of the bladder and assessed the biological behavior of such tumors in a 5-year follow-up. Second voided urine was preoperatively collected from 56 patients with bladder carcinoma, from 20 patients with nonmalignant urological diseases and from 40 healthy volunteers. RNA extraction from exfoliated urothelial cells was followed by quantitative real-time RT-PCR with the Light Cyclerw. Patients in the superficial TCC group had a median expression of 8226 AU (arbitrary units) with and 1523 AU without tumor recurrence (PZ0.023). No such correlation was detected in the group with muscle-invasive tumors. Kaplan–Meier analysis revealed a significant difference between recurrent and nonrecurrent disease (PZ0.019) in superficial but not in muscle-invasive TCC (PZ0.84). CK 20 mRNA expression in urine has the potential to identify patients at risk for recurrence of noninvasive papillary urothelial tumors. It helps to categorize patients prior to TUR-B, so that the cystoscopy interval during follow-up may be extended in those with low-risk superficial TCC. q 2006 Elsevier Ireland Ltd. All rights reserved. Keywords: Tumor marker; Cytokeratin 20; Superficial bladder cancer; Real-time RT-PCR

1. Introduction Transitional cell carcinoma of the bladder is the second most common urological malignancy. It affects approximately 46,700 citizens in Europe each year [1,2]. Since superficial bladder cancer is usually diagnosed by cystoscopy and treated by endoscopic tumor resection, patients have to endure the repeated

* Corresponding author. Fax: C49 30 8445 4448. E-mail address: [email protected] (F. Christoph).

discomfort of regular cystoscopy in the routine followup. Many efforts have been made to investigate the molecular pathology of these superficial tumors with special attention to their occurrence and recurrence. Numerous commercial urinary bladder cancer tumor markers are now available, including BTA-STAT and BTA-TRAK (Bion Diagnostics, Redmond, Washington), NMP-22 (Matritech, Cambridge, Massachusetts) and the ImmunoCyt test (Diagnotime, Inc., Saint-Foy, Quebec, Canada), but they still lack sensitivity and specificity [3–6]. Urinary cytology is still also part of the clinical routine with 30–60% sensitivity rates and

0304-3835/$ - see front matter q 2006 Elsevier Ireland Ltd. All rights reserved. doi:10.1016/j.canlet.2005.12.038

122

F. Christoph et al. / Cancer Letters 245 (2007) 121–126

almost 100% specificity compared to the 46% sensitivity and 93% specificity rates achieved by cystoscopy alone [7]. One of the most recently investigated markers is cytokeratin 20 (CK 20), which was first characterized by Moll et al. [8]. CK 20 is not expressed in normal urothelium, except by occasional umbrella cells and intermediate cells. First immunohistochemical investigations by Harnden et al. showed that CK 20 expression in urothelial carcinoma tissue correlated with the differentiation grade and also predicted early recurrence when strong CK 20 immunoreactivity was detected [9]. Subsequent studies have investigated the use of CK 20 expression in transitional cell carcinoma applying nonquantitative and quantitative RT-PCR [10–13]. Moreover, a number of studies have confirmed the observation that CK 20 immunoexpression alone or in an immunohistochemical profile together with CD44 or 34bE12 is useful for predicting the behavior of papillary tumors [14–17]. Therefore, if a tumor marker like CK 20 helps to differentiate benign from malignant disease, it might also help to predict tumor recurrence. To answer these questions, we evaluated the levels of CK 20 mRNA expression in voided urine by highly sensitive and quantitative real-time RT-PCR (QRT-PCR) in patients with superficial and muscle-invasive disease. We also compared the initial CK 20 expression levels with data from the 5-year follow-up after transurethral resection or radical cystectomy. Finally, we compared the CK 20 expression levels in a large control group of healthy individuals and patients with benign urological diseases.

2. Materials and methods This study included 56 patients (19 women and 37 men) with superficial and muscle-invasive disease treated at the Department of Urology between October 1997 and January 2000. All patients with superficial disease underwent transurethral resection of the bladder (TUR-B), whereas those with muscle-invasive disease were submitted to radical cystectomy. Histopathological staging and grading were done according to the TNM classification of the International Union Against Cancer [18]. All patients signed a consent form approved by the Committee on Human Rights in Research of our institution. Urine control samples were taken from 20 patients with nonmalignant urological diseases (benign prostatic hyperplasia, urolithiasis, cystitis) as well as from 40 healthy volunteers.

2.1. Sample preparation Second voided urine samples of 100 ml were obtained from each patient prior to the initial TUR-B. The urine was centrifuged, and the cells were washed with phosphatebuffered saline (PBS–Dulbecco, without calcium/magnesium). 2.2. RNA extraction Total RNA was extracted from urine samples using a commercially available RNA extraction kit (Trizol, Gibco, Carlsbad, California) according to the manufacturer’s instructions. To avoid contamination by DNA during RNA preparation, RNA was digested with RNAse-free DNAse I (Gibco), as recommended by the supplier. 2.3. RT-PCR Detection of CK 20 mRNA was conducted by a two-step procedure using the Light Cycler CK 20 mRNA Quantification Kit (Roche, Germany). First, cDNA was synthesized with a reverse transcription kit (Reverse Transcriptase Master Mix, Roche Diagnostics, Germany) using 5 ml of RNA and random hexamers serving as primers in a total volume of 20 ml. The second step was to amplify 2 ml of the cDNA by PCR using 2 ml of the LightCycler PCR Master Mix (CK 20 Detection Mix, Roche Diagnostics, Germany). During the PCR reaction, a 124 bp fragment of cytokeratin 20 encoding mRNA was amplified from the cDNA (CK 20 Detection Mix, Roche Diagnostics, Germany). The amplicon was detected by fluorescence using a specific pair of hybridization probes that were fluorescence-labeled at the 5 0 -end with LightCycler-Red 640 and at the 3 0 -end with fluorescein. The fluorescence emitted after hybridization to the template DNA was measured by the LightCycler Instrument. Using the same cDNA preparation but in a separate PCR reaction, mRNA encoding for porphobilinogen deaminase (PBGD) was processed as when used as a housekeeping gene. Its product served as a control for RNA and relative quantification. All PCR reactions were performed for 40 cycles of 60 s at 95 8C, 10 s at 60 8C and 5 s at 72 8C. Each sample was quantified by measuring its fluorescence resonance energy (FRET) with the LightCycler quantification software (LCQuant, Roche, Germany) as previously described [19]. Briefly, the amount of target is calculated as a ratio of target (T) gene copies to housekeeping (H) gene copies. The external standard for cytokeratin 20 mRNA consists of a stabilized fraction of a total RNA purified from an immortalized cell line constitutively expressing CK 20 (HT 29). Arbitrary units (AU) are the relative values derived by comparing the T:H ratio of a given sample to the T:H ratio of the external standard multiplied by 10,000. Statistical analysis was done using SPSS computer software. Various statistical tests were performed to evaluate CK 20 mRNA expression by RT-PCR (Pearson, Spearman,

F. Christoph et al. / Cancer Letters 245 (2007) 121–126

Mann–Whitney, Kruskal–Wallis). ROC and Kaplan–Meier analysis were used to determine the cutoff level of CK20 expression and to evaluate the follow-up data of the 56 tumor patients.

3. Results This study included 56 patients with transitional cell carcinoma of the bladder: 29 pTa tumors, 14 pT1 tumors, and 13RpT2 tumors. TCC was low- or intermediate-grade (G1/2) in 26 patients and highgrade (G3) in 30. Patients had a median age of 72 in the superficial group and 78 in the group with muscleinvasive tumor disease. In a control group, urinary CK 20 expression was measured in seven patients with benign prostatic hyperplasia (BPH), seven patients with urolithiasis and six patients with cystitis. Urine from 40 healthy volunteers was also included in this control group. Cytokeratin 20 mRNA expression was detectable in all urine samples examined. The median CK 20 expression level, measured in arbitrary units (AU), was 278 AU in the control group versus 5565 AU in the carcinoma group (P!0.0001). Optimal sensitivity and specificity (82 and 93%) were obtained by applying 932 AU as the cytokeratin 20 mRNA cutoff value (mean CK 20 mRNA value C1SD in the control group). Optimal performance with 71% sensitivity and 98% specificity was achieved with a cutoff value of 1893 AU. Considering the different tumor stages, the median expression of CK 20 was lowest in the pTa G1/2 group with 3134 AU and highest in the RpT2 G3 group with 59,897 AU. The low median expression level of CK 20 in the subgroup of pT1 G1/2 tumors with 133 AU may be explained by the small amount of samples with extremely heterogeneous expression levels ranging from 44 to 14,678 AU. A comparison of tumor grades in the individual stages disclosed no statistical significance. In contrast, comparing the individual stages revealed stage-related CK 20 mRNA expression (PZ0.014 for pTa versus pT1 and PZ0.025 for pT1 versus RpT2). Table 1 gives further details on different CK 20 expression levels in relation to the stage and grade. Comparing superficial with muscle-invasive tumor disease revealed a median CK 20 expression of 4588 AU in the superficial group (pTa and pT1) versus 59,897 AU in the muscle-invasive (RpT2) group, which was also statistically significant (PZ0.0002). Dividing the superficial tumor group into G1/2 and G3 tumors showed a median CK 20 expression of 2651 AU in low- and

123

intermediate-grade versus 10,865 AU in high-grade tumors, which was statistically significant (PZ0.005). Follow-up of patients with superficial tumor disease consisted of regular cystoscopy in our outpatient clinic or by the attending urologist. Patients with muscleinvasive disease had a follow-up with computed tomography of the abdomen every 3 or 6 months as well as regular chest X-rays during the first 5 years after surgery. The median follow-up was 51.5 months for both groups. A comparison of individual tumor groups with and without relapse according to the tumor stage and grade revealed higher CK 20 mRNA expression levels for all patients with tumor recurrence, but the differences were not statistically significant. In the pTa group, 11 of 29 patients (38%) had tumor recurrence with a median time-to-relapse of 11 months. The median CK 20 mRNA expression in this group was 4293 versus 1335 AU in the group of pTa tumors without recurrence (PZ0.09). In the group of pT1 tumors, 6 of 14 patients had a relapse after a median time of 16.5 months (43%) with a median CK 20 mRNA expression of 9763 AU in the recurrence-free group versus 19,288 AU in the recurrence group (PZ 0.28). Distant metastases became apparent after a median time of 23 months in seven (54%) of the muscle-invasive tumor patients initially treated by cystectomy, and six of them (86%) died after a median survival of 27 months. The median expression was 59, 897 AU with tumor relapse and 81,771 AU without it (PZ0.84). Table 1 Urinary CK 20 mRNA expression in tumor patients in arbitrary units (AU) G1/2 pTa Median 3134 Mean 4498 SD 4821 pT1 Median 133 Mean 4952 SD 8423 RpT2 Median n.e. Mean SD Superficial TCC Median 2651 Mean 4548 SD 5103

n

G3

n

P-value*

24

5100 6730 7189

5

0.686

3

11,369 32,813 36,964

11

0.126

0

59,897 72,952 55,049

13

27

10,865 24,662 32,872

16

0.005

n.e., not evaluated in this study. *PZ0.014 for pTa versus pT1 tumors; PZ0.025 for pT1 versus RpT2 tumors.

124

F. Christoph et al. / Cancer Letters 245 (2007) 121–126

Table 2 Urinary CK 20 mRNA expression in recurrent and nonrecurrent patients

Median Mean SD n P-value a b

Superficial TCC

Muscle-invasive TCC

Recurrenta

Nonrecurrent

Recurrentb

Nonrecurrent

8226 18,462 29,038 17

1523 7828 15,869 26

59,897 74,147 55,908 7

81,771 71,558 59,310 6

0.023

0.84

Intravesical, local tumor recurrence. Systemic, metastatic tumor recurrence.

1.1

recurrence (log-rank test) than those with lower expression levels (PZ0.019). No such correlation could be detected in the group with muscle-invasive disease, since CK 20 expression was extremely heterogeneous in all samples investigated (PZ0.84). In the benign control group, there was the significantly higher expression of urinary CK 20 mRNA (median of 1381 AU) in patients with BPH than in those with urolithiasis (157 AU) or cystitis (163 AU) or in healthy individuals (279 AU) (PZ0.009).

4. Discussion p=0.019

Recurrence-free survival

1.0 0.9 A 0.8 0.7 0.6 0.5 B 0.4

0

10

20 30 40 50 Follow-up (months)

60

70

Fig. 1. Disease-free survival of patients with superficial TCC. Patients with median CK 20 mRNA expression lower (A) or higher (B) than 4600 AU.

The median CK 20 mRNA expression in patients with superficial tumors differed significantly between relapsed and nonrelapsed patients with values of 8226 and 1523 AU, respectively (PZ0.023) (see Table 2). Comparing patients at higher risk for recurrence with pTa G3 and pT1 G3 tumors, the median CK 20 expression was 13,987 AU (relapse) versus 347 AU (relapse-free) in the pTa G3 group and 23,897 AU (relapse) versus 10,957 AU (relapse-free) in the pT1 G3 group, but neither group evidenced statistical significance with PZ0.35 for the pTa G3 group and PZ0.79 for the pT1 G3 group. The 5-year recurrence-free interval was evaluated by Kaplan–Meier analysis with a cutoff of 4600 AU after ROC analysis for patients with superficial TCC (Fig. 1). Patients with an initial CK 20 mRNA expression above 4600 AU had a significantly higher risk of tumor

Highly sensitive real-time RT-PCR has already shown its ability to determine the stage and grade of tumor tissue and to differentiate between malignant and benign tissue, since CK 20 mRNA expression is significantly higher in invasive and poorly differentiated bladder TCC and lower in low-grade tumors or benign urothelial tissue [19]. In the present study, CK 20 mRNA expression by urothelial cells in urine was significantly higher in patients with tumor disease than in those with nonmalignant diseases or in healthy volunteers. Interestingly, CK 20 mRNA expression was significantly higher in patients with benign prostatic hyperplasia than in those with cystitis or stone disease or in the healthy control group. CK 20 mRNA expression in urine was able to clearly differentiate the individual tumor stages. CK 20 mRNA expression was lower in pTa tumors and highest in RpT2 tumors. The latter group displays a very heterogeneous CK 20 expression pattern that has also been described in tumor tissue. Moreover, CK 20 expression was strongly gradedependent. It was expressed at significantly higher levels in high-grade superficial TCC, suggesting a loss of expression control in progressive tumor disease. We were also able to identify the patient group with a higher risk of recurrence, as evidenced by higher CK 20 mRNA expression in the specific subgroup. Similar findings were obtained by Ramos et al. using CK 20 immunohistochemistry in tumor tissue and evaluating its predictive value in low-grade papillary tumors [16]. As the log-rank test could differentiate patients with and without tumor recurrence in the 5-year follow-up according to their initial CK 20 mRNA expression level in urine, our preliminary data indicate that CK 20 may not only serve as a marker for tumor detection, but also has the potential to predict tumor recurrence in superficial TCC.

F. Christoph et al. / Cancer Letters 245 (2007) 121–126

However, urinary CK 20 mRNA expression was not helpful in patients with muscle-invasive disease, since expression levels failed to predict tumor recurrence defined by the appearance of metastatic disease. Moreover, the median CK 20 mRNA expression was lower in the patient group with recurrent disease. There are no comparable data relating urinary CK 20 expression to tumor recurrence in muscle-invasive disease, but this interesting finding has already been observed when using nested RT-PCR to identify CK 20 mRNA expression in peripheral blood. In this connection, CK 20 expression was shown to be an independent prognostic factor for superficial but not invasive disease in a small group of 26 patients [20]. Twenty of these patients had received chemotherapy, which is supposed to lower the recurrence rate in muscle-invasive disease and might therefore alter the prognostic power in this special setting. However, only five patients in our study had received chemotherapy. The lack of a correlation between the expression level and the relapse rate may also be attributed to the small number of patients in our study. Nevertheless, caution is necessary when comparing exfoliated tumor cells in urine with those that circulate in peripheral blood. Moreover, expression level of CK 20 mRNA may vary within the same sample. Lassmann et al. detected heterogeneous CK 20 expression in single tumor samples by immunohistochemistry and RT-PCR in colorectal carcinoma [21]. In urothelial carcinoma, the tumor cells on the surface that are exfoliated into the urine may differ in their expression levels from those infiltrating the muscularis mucosa or adjacent tissue. Nevertheless, this remains highly speculative, since CK 20 immunohistochemistry shows a diffuse CK 20 expression pattern in advanced superficial bladder TCC, but no representative data are available for muscle-invasive disease [15]. The results of this study confirm that the level of CK 20 mRNA expression in urine can differentiate various benign conditions from malignant disease. Moreover, the data presented show that urinary CK 20 mRNA expression levels appear to predict time to recurrence in noninvasive tumors. In superficial tumor disease, preoperative CK 20 detection in urine is a reliable tool for obtaining prognostic information and estimating the risk of recurrence after transurethral resection. Thus, low-risk patients may benefit by an extension of the cystoscopy interval in the routine follow-up, whereas high-risk patients may profit from a more aggressive approach with intravesical instillation therapy. In muscle-invasive disease, urinary CK 20 detection is possible but has no diagnostic value in risk estimation.

125

Acknowledgements We thank Ms Petra von Kwiatkowski for her expert technical assistance and Dr Joanne Weirowski for her linguistic advice. References [1] F.o.e.c. societies, The ten cancers with highest mortality in 1998, 2005. [2] A. Jemal, A. Thomas, T. Murray, M. Thun, Cancer statistics, 2002. CA Cancer J. Clin. 52 (2002) 23–47. [3] J. Landman, Y. Chang, E. Kavaler, M.J. Droller, B.C. Liu, Sensitivity and specificity of NMP-22, telomerase, and BTA in the detection of human bladder cancer, Urology 52 (1998) 398–402. [4] V. Serretta, D. Lo Presti, P. Vasile, E. Gange, E. Esposito, I. Menozzi, Urinary NMP22 for the detection of recurrence after transurethral resection of transitional cell carcinoma of the bladder: experience on 137 patients, Urology 52 (1998) 793–796. [5] L. Thomas, H. Leyh, M. Marberger, E. Bombardieri, P. Bassi, F. Pagano, et al., Multicenter trial of the quantitative BTA TRAK assay in the detection of bladder cancer, Clin. Chem. 45 (1999) 472–477. [6] Y. Fradet, C. Lockhard, Performance characteristics of a new monoclonal antibody test for bladder cancer: immunoCyt trade mark, Can. J. Urol. 4 (1997) 400–405. [7] P. Jichlinski, L. Guillou, S.J. Karlsen, P.U. Malmstrom, D. Jocham, B. Brennhovd, et al., Hexyl aminolevulinate fluorescence cystoscopy: new diagnostic tool for photodiagnosis of superficial bladder cancer—a multicenter study, J. Urol. 170 (2003) 226–229. [8] R. Moll, A. Lowe, J. Laufer, W.W. Franke, Cytokeratin 20 in human carcinomas. A new histodiagnostic marker detected by monoclonal antibodies, Am. J. Pathol. 140 (1992) 427–447. [9] P. Harnden, I. Eardley, A.D. Joyce, J. Southgate, Cytokeratin 20 as an objective marker of urothelial dysplasia, Br. J. Urol. 78 (1996) 870–875. [10] A. Klein, R. Zemer, V. Buchumensky, R. Klaper, I. Nissenkorn, Expression of cytokeratin 20 in urinary cytology of patients with bladder carcinoma, Cancer 82 (1998) 349–354. [11] V. Buchumensky, A. Klein, R. Zemer, O.J. Kessler, S. Zimlichman, I. Nissenkorn, Cytokeratin 20: a new marker for early detection of bladder cell carcinoma? J. Urol. 160 (1998) 1971–1974. [12] D. Rotem, A. Cassel, N. Lindenfeld, Y. Mecz, Y. Sova, M. Resnick, et al., Urinary cytokeratin 20 as a marker for transitional cell carcinoma, Eur. Urol. 37 (2000) 601–604. [13] T. Inoue, H. Nakanishi, K. Inada, T. Hioki, M. TatematsuY, Y. Sugimura, Real time reverse transcriptase polymerase chain reaction of urinary cytokeratin 20 detects transitional cell carcinoma cells, J. Urol. 166 (2001) 2134–2141. [14] A. Alsheikh, Z. Mohamedali, E. Jones, J. Masterson, C.B. Gilks, Comparison of the WHO/ISUP classification and cytokeratin 20 expression in predicting the behavior of low-grade papillary urothelial tumors. World/Health Organization/International Society of Urologic Pathology, Mod. Pathol. 14 (2001) 267–272. [15] S. Desai, S.D. Lim, R.E. Jimenez, T. Chun, T.E. Keane, J.K. McKenney, et al., Relationship of cytokeratin 20 and CD44 protein expression with WHO/ISUP grade in pTa and pT1 papillary urothelial neoplasia, Mod. Pathol. 13 (2000) 1315–1323.

126

F. Christoph et al. / Cancer Letters 245 (2007) 121–126

[16] D. Ramos, S. Navarro, R. Villamon, M. Gil-Salom, A. Llombart-Bosch, Cytokeratin expression patterns in lowgrade papillary urothelial neoplasms of the urinary bladder, Cancer 97 (2003) 1876–1883. [17] D. Golijanin, A. Shapiro, D. Pode, Immunostaining of cytokeratin 20 in cells from voided urine for detection of bladder cancer, J. Urol. 164 (2000) 1922–1925. [18] L.H. Sobin, C. Wittekind, TNM Classification of Malignant Tumours, sixth ed., Wiley, New York, 2002. [19] F. Christoph, M. Muller, M. Schostak, R. Soong, K. Tabiti, K. Miller, Quantitative detection of cytokeratin 20 mRNA

expression in bladder carcinoma by real-time reverse transcriptase-polymerase chain reaction, Urology 64 (2004) 157–161. [20] T. Okegawa, M. Kinjo, K. Nutahara, E. Higashihara, Value of reverse transcription polymerase chain assay in peripheral blood of patients with urothelial cancer, J. Urol. 171 (2004) 1461–1466. [21] S. Lassmann, M. Bauer, R. Soong, J. Schreglmann, K. Tabiti, J. Nahrig, et al., Quantification of CK20 gene and protein expression in colorectal cancer by RT-PCR and immunohistochemistry reveals inter- and intratumour heterogeneity, J. Pathol. 198 (2002) 198–206.