Urothelial cell carcinoma (UCC) of the urinary bladder is a serious health problem. The disease accounts for 3.2% of all new cancer cases worldwide. UCC of the bladder is the 5th most common cancer and the 9th leading cause of cancer death. In most patients (± 75%), UCC is superficial (i.e. pTa-pT1) at first presentation. There are two problems in the management of these patients. First, superficial UCC of the bladder is characterized by a high risk of recurrence (55­85%), with most recurrences in the first year after trans-urethral resection (TUR). These recurrences are detected by urethro-cystoscopy (UCS), the gold standard, which is performed every 3-4 months, often in combination with urine cytology (UC) as an adjunct. After a two-year recurrence free period, the intensity of follow up may be lowered. UCS and UC are routinely used in the daily clinical practice. However, UCS is invasive, bothersome for the patients and costly. Unfortunately, UC is characterized by a low sensitivity, especially for low grade UCC. Hence, there is room for improvement, i.e. a need for development of techniques to diagnose superficial bladder cancer recurrences in urine and/or to identify a subset of patients who require a less frequent follow up. The second concern in managing patients with UCC is to prevent progression (10-15%) to muscle invasive, potentially lethal UCC. The ability to distinguish superficial tumors with invasive capabilities from those unlikely to become invasive would be of great clinical benefit. Carcinoma in situ (CIS), a flat and high grade (G3) manifestation of superficial UCC is especially associated with an unfavorable disease course. However, individual prediction of prognosis by clinical and pathological features is still very difficult. Superficial UCC that maintains a more malignant phenotype might be better treated with radical cystectomy in an early stage of the disease. This explains the need for more accurate predictors of UCC-progression. With the recent advances in molecular biology, research is more and more focusing on understanding the molecular mechanisms associated with bladder carcinogenesis. Therefore, molecular markers may aid in solving the two UCC management problems outlined above. If molecular markers prove to have additional value over the established clinico-pathological variables, they could become part of the clinical practice. This thesis deals with these two problems concerning the management of patients with UCC. Molecular markers for diagnosis in urine and prognosis of UCC were analyzed in various clinical settings. Part one gives a short introduction on UCC in general perspective and contains the scope of this thesis (chapter 1). In the second part of the thesis, we evaluated Microsatellite analysis (MA); a test based on PCR reactions with DNA of the urine sediment, in different clinical settings. Chapter 2 provides an overview of the accuracy of the current urine markers to diagnose recurrent UCC. The sensitivity (patients with a recurrent lesion and positive urine test divided by all patients with recurrent UCC) and specificity (patients without recurrence and a negative test divided by all patients without recurrent UCC) were determined for 18 urine markers. All the urine markers reviewed had a higher sensitivity, but a lower specificity than conventional cytology for recurrent UCC. The sensitivities increased with tumor grade. In this meta-analysis confined to studies on recurrent disease, MA had a sensitivity of 82% and a specificity of 89% for detection of UCC recurrences. We found a considerably lower sensitivity to detect recurrent UCC for 13/18 urine markers in comparison with earlier review articles where no distinction between primary and recurrent UCC was made. The reason for this reduced sensitivity is most likely a higher percentage of small, pTa and low grade UCC among recurrent lesions compared to primary tumors. At this time, sufficient clinical evidence to change the cystoscopic follow-up scheme with the use of a urine marker is not available. Chapter 3 describes a phase II study on urinary MA, which we performed in a cross-section of 109 patients who visited the outpatient department for a follow-up UCS. Earlier investigations in other clinical settings and with a limited number of patients showed that MA had a high potential to diagnose UCC in urine. Our study was conducted to investigate whether MA was also capable of diagnosing UCC in patients who were under surveillance after TUR. MA detected 18/24 recurrent tumors. The undetected UCCs were small pTaG1 lesions. In 5 of 9 cases where no tumor was seen at UCS but MA was positive, an UCC was found within 6 months after MA. In the same group of patients, MA was more sensitive (74%) than the BTAstat test (56%) or conventional urine cytology (22%). A drawback of MA was the exclusion of 7 patients due to an insufficient amount of DNA in the urine sediment and another 9 patients with leukocyte abundance. Nevertheless, MA appeared a reliable molecular test in urine for detection of recurrent UCC, sometimes even before cystoscopical evidence of the disease. Chapter 4 describes MA in conjunction with the Fibroblast Growth Factor Receptor 3 (FGFR3) mutation, a genetic marker for favorable UCC. This study was initiated to explore the possibility to improve the sensitivity of MA by adding FGFR3 mutation analysis to the molecular marker panel. The tests were carried out on tumor- and urine samples of 59 UCC patients allowing a direct comparison of molecular changes in tumor and urine. Conventional cytology was used as a reference and a full clinical record of each patient was obtained. UCCs with a higher stage or grade had more microsatellite alterations per tumor. Conversely, FGFR3 mutations were only found in pTaG1-2 lesions. The sensitivity of MA in FGFR3 positive and negative UCC was 71% and 91%, respectively. When a Microsatellite alteration was detected in tumor tissue, the chance to find the same alteration in the urine was 35% regardless of FGFR3 status. The chance to find the same FGFR3 mutation in the urine was 52%. Additional alterations in the urine that were not present in the UCC were only seldom reported (1.8%). By combining the FGFR3 mutation with MA, the sensitivity of molecular cytology could be enhanced to 89% without compromising specificity. The similarity of the molecular profiles in urine and tumor corroborate the clonal relation of tumor cells in the urine and the original tumor(s). In chapter 5, the value of MA for patients under surveillance after radiotherapy was investigated. The follow up of these patients is more or less the same as for the patients with superficial UCC. However, UCS and UC are often difficult to interpret in case of previous radiotherapy. Therefore, MA was tested in this particular patient group. Although only 6 recurrences were found in the 49 patients, MA detected 5 of them. The specificity values for routine urine cytology, MA and expert urine cytology were 85%, 93% and 97%, respectively. In addition, the negative predictive value of MA (true negative MA / all negative MA) was 99% (70/71 urine samples) for this group. The latter is especially interesting because the positive predictive value of UCS (40% in this series) is much lower for irradiated patients. The results of this study indicate that MA may also be a useful tool for surveillance of patients who were previously treated by radiotherapy. Part three of this thesis focuses on the prognostic value of the Fibroblast Growth Factor Receptor 3 (FGFR3) mutation in UCC. Its relation to favorable disease will be discussed as well as its relation to clinical variables and other molecular markers. Chapter 6 gives an overview of the clinico-pathological and molecular prognostic features in (superficial) UCC. The last part of this chapter gives a short introduction of the FGFR3 gene. The chapters 7 & 8 contain the first two papers published on the favorable nature of the FGFR3 mutation in UCC. The FGFR3 mutation was detected in 70/98 (71%) of pTaG1-2 UCC. Conversely, invasive G3 lesions had a mutation in only 7% of tumors. Moreover, the mutation was absent in 20 cases of CIS, again pointing to the favorable nature of the FGFR3 mutation in UCC. White blood cells and normal-looking urothelial biopsies of patients with a FGFR3 mutation in their bladder tumor lacked FGFR3 mutations. This emphasized the somatic nature and the tumor specificity of these mutations in UCC. The FGFR3 mutation was also associated with a low recurrence rate of superficial UCC in the fist year after a TUR. In 57 patients with superficial UCC, only 6% of UCS was positive for patients with a FGFR3 mutation. This was 28% for patients who had no FGFR3 mutation in their tumor. These two preliminary studies showed that FGFR3 is the first gene to be mutated at a high frequency in favorable (pTa G1-2) UCC. This mutation is the first genetic marker to selectively identify a large group of patients with favorable disease characteristics. Furthermore, the results presented in these two chapters suggest that the frequency of UCS can be reduced considerably for patients with FGFR3 positive tumors. In chapter 9 the results of a large, retrospective multi­center study are displayed. The purpose of this study was to compare the FGFR3 mutations to other molecular markers and to obtain long term clinical follow up. In addition, the reproducibility of pathological grade and molecular variables was determined. In this study of 286 patients with primary UCC, the FGFR3 mutation was related to the expression of three immunohistochemical markers (MIB-1, P53 and P27kip1) known to be associated with prognosis. FGFR3 mutations were detected in 172/286 (60%) of UCCs. G1-tumors had a FGFR3 mutation in 88% of cases, G3-tumors in 16%. Conversely, aberrant expression patterns of MIB-1, P53 and P27kip1 were seen in 5%, 2% and 3% of G1-tumors and in 85%, 60% and 56% of G3-tumors, respectively. In multivariate analysis with recurrence rate, progression and disease specific survival as endpoints, the combination of FGFR3 and MIB-1 proved of independent significance for all 3 disease parameters. With regard to disease progression of superficial UCC, the difference in the clinical course between UCC combining FGFR3 mutation / MIB-1 normal (N=138) and those combining FGFR3 wild type / MIB-1 high (N=47) was evident. Interestingly, the 61 remaining patients with tumors characterized by absence of FGFR3 mutation and normal MIB-1 expression or by a FGFR3 mutation and high MIB-1 expression behaved more or less the same representing an intermediate prognosis. These results led to the proposal of a molecular grading model for UCC distinguishing three molecular grades (mG1-3): mG1 (favorable prognosis), mG2 (intermediate prognosis) and mG3 (poor prognosis). Moreover, the molecular Grade was more reproducible than pathological grade (91% vs. 47–61%). Based on the observations in the chapters 7, 8 and 9, it was concluded that the FGFR3 mutation represents the favorable molecular pathway of papillary UCC. Furthermore, molecular Grading provides a new, simple and highly reproducible tool to determine UCC prognosis. Chapter 10 shows that FGFR3 mutations are also common in urothelial papilloma, a rare urothelial neoplasm. Papillomas are generally considered as a separate diagnostic entity because of their benign natural behavior. Therefore, they are currently classified apart from papillary lesions of low malignant potential (PUN-LMP, G1) and low grade papillary urothelial carcinomas (LG-PUC, G1-2) by the 1998 World Health Organization / International Society of Urological Pathology consensus classification. In our study, FGFR3 mutations were found in 9/12 papillomas and in 68/79 pTaG1 lesions (reclassified as PUN-LMP and LG­PUC in 62 and 17 cases, respectively). Besides urothelial papillomas, the FGFR3 mutation analysis was also performed for 21 ovarian Brenner tumors. This tumor was chosen for its resemblance to low grade UCC from a cytological point of view. However, no FGFR3 mutations were found in Brenner tumors. This report was the first to describe a genetic defect in urothelial papilloma and we suggested from a molecular perspective, that urothelial papillomas should be classified together with all well differentiated urothelial neoplasms. Chapter 11 describes the distribution of FGFR3 mutations and P53 over-expression (assumed if expression level >10%) for 260 patients in UCC. As described in the chapters 7 and 8, the FGFR3 gene is mutated in a high percentage of low grade UCCs. Therefore, the purpose of this study was to evaluate the relationship between FGFR3 mutations and over-expression of the P53 tumor suppressor protein as a manifestation of a mutation in the P53 gene. The over-expression of P53 and/or the mutant P53 gene have been reported in earlier studies as an unfavorable marker in UCC. FGFR3 mutations and P53 over-expression were observed in 59% and 25% of tumors, respectively. Together, the FGFR3 and P53 over-expression described 79% of UCC. However, alterations in FGFR3 and P53 coincided in only 5.7% of UCC. Because each of these alterations was also associated with opposite clinico-pathological parameters, it was proposed that mutations in FGFR3 and P53 mark two alternative genetic pathways in papillary UCC pathogenesis. Chapter 12 gives an overview of the distribution of FGFR3 mutations detected in UCC and the corresponding germline mutations, which are responsible for a wide variety of skeletal anomalies. Almost all FGFR3 mutations in earlier studies corresponded to the mutations that lead to thanatophoric dysplasia, a congenital disorder that is generally lethal shortly after birth. This study reported a few new FGFR3 mutations in UCC that matched non-lethal skeletal syndromes. Therefore, patients with skeletal syndromes associated with FGFR3 mutations and a relative normal life span (for example: achondroplasia) might be at a higher risk for development of bladder tumors than the general population provided that the FGFR3 mutation on its own is capable of bladder tumor formation.

Erasmus University Rotterdam
hdl.handle.net/1765/6758
Erasmus MC: University Medical Center Rotterdam

van Rhijn, B. (2005, June). Molecular Diagnosis and Prognosis of Bladder Cancer : towards the implementation of molecular markers in clinical practice. Retrieved from http://hdl.handle.net/1765/6758