Identification of fibroblast growth factor receptor 3 mutations in urine sediment DNA samples complements cytology in bladder tumor detection

Abstract

BACKGROUND

Mutations in fibroblast growth factor 3 receptor (FGFR3) are frequent events in low-grade bladder tumors. To assess the potential utility of the detection of FGFR3 mutations in a screening modality, the authors analyzed urine sediment DNA samples from 192 patients in a retrospective study.

METHODS

Urine sediment DNA samples from 192 patients were prepared. Seventy-two patients had undergone transurethral resection (TURBT group) of mainly Ta lesions and 120 patients had undergone cystectomy (cystectomy group). The majority of patients in the cystectomy group had more advanced tumors compared with patients in the TURBT group. DNA preparations were screened for FGFR3 mutations in exons 7, 10, and 15 using single-strand conformation polymorphism (SSCP) and DNA sequencing.

RESULTS

Using SSCP, 67% of patients in the TURBT group and 28% in the cystectomy group displayed FGFR3 mutations. Comparative analysis of cytology results and FGFR3 mutational analysis were performed in 122 cases. Within the TURBT group, FGFR3 mutation analysis outperformed cytology. FGFR3 mutation analysis identified change in 68% of urine sediment DNA samples whereas cytology recorded the presence of tumor cells in 32% of the DNA samples. In the cystectomy group, cytology outperformed FGFR3 mutation analysis. Cytology recorded tumor detection in 90% of patients, while SSCP identified mutational change in 24%.

CONCLUSIONS

Combining FGFR3 mutation results with cytology in both groups correctly identified tumor presence in 105 of 122 (86%) of patients. The greater sensitivity of FGFR3 mutation detection over cytology in identifying the presence of low-grade, superficial bladder tumors represents a potential new tool to complement standard cytology in screening patients for bladder tumors and recurrent disease. Cancer 2003;98:737–44. © 2003 American Cancer Society.

DOI 10.1002/cncr.11536

Bladder carcinoma is one of the most common malignancies worldwide. In 2003, approximately 50,000 people will be diagnosed with this disease in the United States. Bladder tumors have been categorized into two groups defined by distinct behaviors and molecular profiles.1–6 These groups are characterized by low-grade tumors (which are always papillary and usually superficial) and high-grade tumors (which can be either papillary or nonpapillary and often invasive). Clinically, superficial bladder tumors account for 75–80% of neoplasms, whereas the remaining 20–25% are invasive or metastatic at the time of presentation. Greater than 70% of patients with superficial tumors will have 1 or more disease recurrences after initial treatment. Approximately one-third of these patients will experience disease progression and will present at a later date with more aggressive lesions.

Superficial bladder carcinoma requires intensive clinical management after initial treatment. Cystoscopy and cytology are the gold standards for detecting bladder carcinoma. After transurethral resection of bladder tumors (TURBT), the standard practice is to screen patients for possible disease recurrence by cystoscopic evaluation every 3–4 months for 2 years and every 6 months thereafter. Much effort has been made to increase detection rates of bladder carcinoma in the follow-up period and to develop a noninvasive assay that is as sensitive as cytology in cancer detection.7–16 Although cytology performs well in the detection of late-stage bladder carcinoma, it is less successful in detecting the presence of low-grade superficial lesions. The main reason for this is that exfoliated cells from superficial bladder tumors may be morphologically indistinguishable from normal urothelial cells. Consequently, many groups continue to evaluate various molecular markers as indicators of tumor presence and/or progression in bladder carcinoma.17, 18

Fibroblast growth factor receptor 3 (FGFR3) belongs to a family of structurally related tyrosine kinase receptors. Germline mutations in FGFR3 have been identified as causative factor in disorders such as achondroplasia and hypochondroplasia.19, 20 Somatic mutations of the FGFR3 gene were reported in approximately 40% of the bladder tumors analyzed. FGFR3 mutations were localized in exons 7, 10, and 15, which represent putative ‘hotspot’ regions within the gene.21, 22 It is noteworthy that these mutations were detected at a higher frequency in superficial disease than in invasive bladder carcinoma.23–25

In the current study, we determined whether FGFR3 mutations could be detected in urine sediment DNA samples from patients with bladder carcinoma. We analyzed 192 urine sediment DNA samples for the presence of FGFR3 mutations. Comparison of results from cytology and mutation analysis revealed a significantly higher detection rate of Ta lesions using FGFR3 mutational analysis compared with cytology. However, cytology better identified invasive lesions (T1–T4) compared with the presence of an FGFR3 mutation. Combined cytology/FGFR3 mutation analysis results in the detection of 86.1% of patients with bladder carcinoma (Ta-T4).

MATERIALS AND METHODS

Patients

Urine samples were collected from 72 patients9 who underwent TURBT and from 120 patients who underwent cystectomy at the Lahey Clinic (Burlington, MA). All patients provided consent under an institutional review board protocol. Urine samples were collected before surgery.

Extraction of DNA

Control cell lines or cells pelleted from urine samples were incubated in lysis buffer (10 mM Tris-HCl, pH 8.3; 100 mM NaCl; 1 mM ethylenediaminetetraacetic acid [EDTA]; 1% sodium dodecyl sulfate; and 100 μg/mL proteinase K) overnight at 37 °C. High–molecular weight DNA was isolated after phenol-chloroform extraction and ethanol precipitation.

Polymerase Chain Reaction (PCR)

PCR amplification was carried out in a 50 μL volume containing 50 ng of genomic cell line or urine sediment DNA, 1 × PCR buffer containing 15 mM MgCl₂, 100 μM each of dATP, dGTP, dTTP, 60 μM dCTP, 1 unit of Taq DNA polymerase (Perkin Elmer Gene Amp, Foster City, CA), 1 μM of forward and reverse primers, 2.0 μCi of [γ-³²P] dCTP (Amersham, Arlington Heights, IL), and 5% dimthylsulfoxide. After an initial denaturation at 95 °C for 5 minutes, 35 cycles of 94 °C for 1 minute, a specific annealing temperature (exon 7, 65 °C; exon 10, 62 °C; exon 15, 60 °C) for 1 minute, and 72 °C for 80 seconds were performed. The following primer pairs were used26: exon 7, 5′–AGTGGCGGTGGTGGTGAGGGAG-3′ and 5′-TGTGCGTCACTGTACACCTTGCAG-3′; exon 10, 5′-CAACGCCCATGTCTTTGCAG-3′ and 5′-CGGGAAGCGGGAGATCTTG-3′; exon 15, 5′-GACCGAGGACAACGTGATG-3′ and 5′-GTGTGGGAAGGCGGTGTTG-3′.

Single-Strand Conformation Polymorphism (SSCP) Analysis

Four microliters of the PCR product was mixed with 16 μL of stop solution (95% formamide, 20 mM EDTA, 0.05% bromophenol blue, and 0.05% xylene cyanol FF), heat-denatured at 95 °C for 5 minutes and rapidly loaded onto nondenaturing polyacrylamide gels. All samples were run under 2 different sets of gel conditions: 6% acrylamide and 10% glycerol run at room temperature with a cooling fan at 30 watts (W); and 8% acrylamide with no glycerol run at 4 °C at 30 W. Gels were dried at 80 °C and autoradiographed using Kodak BioMax film (Eastman Kodak, Rochester, NY) for 24–96 hours. The urine sample was scored as having a mutation if an abnormal SSCP pattern was detected in two or more repeat experiments involving separate PCR reactions.27

DNA Sequencing

Aberrantly migrating DNA fragments were excised from the gel and extracted after boiling in water for 15 minutes. DNA samples were ethanol precipitated in glycogen and used for PCR analysis as described above. One microliter of the PCR product was subcloned and used in the transformation of competent cells provided in the TA cloning kit (Invitrogen, San Diego, CA) according to the manufacturer's instructions. Colonies containing PCR product were grown overnight in Luria-Bertani medium (tryptone 1% wt/vol, yeast extract 0.5% wt/vol, NaCl 0.5% wt/vol, and 10 mM NaOH) and the plasmids were extracted. Sequencing was performed according to the dideoxy-chain-termination method using a Sequenase version 2.0 kit (United States Biochemical, Cleveland, OH) according to the manufacturer's instructions. Samples were loaded onto a 6% denaturing polyacrylamide gel containing 7 M urea and run at 50 °C. Dried gels were autoradiographed for 12–24 hours.

RESULTS

Two groups of patients were identified for urine collection in the current study. The first group of patients was scheduled for TURBT as identified by cystoscopy. Patients in the second group presented with invasive disease and were scheduled to undergo cystectomy. The characteristics of both groups of patients are shown in Table 1.

Table 1. Clinicopathologic Patient Profiles

Characteristic	TURBT group	Cystectomy group
TURBT: transurethral resection for bladder tumor; N/A: not applicable. a Two patients in the TURBT group had a positive cytology result and a negative cystoscopy. b One patient was lost to follow-up and two patients had positive cytology results with no evidence of bladder lesions and were not included. c Thirty-five patients were never free of disease due to lymph node involvement or underwent salvage cystectomy, eight patients were lost to follow-up, and two patients died in the immediate postoperative period. d Three patients underwent a cystectomy due to progression of disease.
No. of patients	72	120
Gender
Male	52	98
Female	20	22
Mean age (yrs)	69.9 (range, 39–83)	65.4 (range, 37–86)
Grade/stagea
G1/Ta	8	—
G1/T1	1	—
G2/Ta	33	1
G2/T1	7	1
G2/T3	—	2
G3/Ta	5	—
G3/T1	3	8
G3/T2	1	29
G3/T3	1	51
G3/T4	—	22
Cis/Tis	11	6
Disease recurrence (%)	56 (81)b	17 (23)c
Disease progression	12d	N/A

Urine samples were collected from both groups immediately before surgery and DNA was extracted from exfoliated cells after standard procedures were performed. Molecular analysis of exons 7, 10, and 15 of the FGFR3 gene was performed on urine sediment DNA using PCR-SSCP. Aberrantly migrating bands were excised from the gels, cloned, and sequenced.

Fibroblast Growth Factor Receptor 3 Mutations in the Transurethral Resection of Bladder Tumor Group

Of 72 urine samples collected from patients undergoing TURBT, 48 (66.7%) harbored mutations in FGFR3. The distribution of FGFR3 mutations according to tumor stage is shown in Table 2. Of the 48 samples displaying mutations in SSCP analysis, 30 (62.5%) were patients with Ta lesions. Mutations in FGFR3 were also identified in urine sediment DNA samples from 10 of the 13 (76.9%) patients with invasive (T1–T4) lesions and from 8 of the 11 (72.7%) with Tis lesions. In two patients with hematuria, no tumor was detected on cystoscopic evaluation and no FGFR3 mutation was detected in SSCP analysis.

Table 2. Tumor Stage and Fibroblast Growth Factor Receptor 3 Mutation Status in the Transurethral Resection of Bladder Tumor and Cystectomy Groups

Stage	TURBT		Cystectomy
	Positive (%)	Negative (%)	Positive (%)	Negative (%)
TURBT: transurethal resection of bladder tumor.
Ta	30 (65.2)	16 (34.8)	1 (100)	0 (0)
Tis	8 (72.7)	3 (27.3)	1 (16.7)	5 (83.3)
T1–T4	10 (76.9)	3 (23.1)	32 (28.3)	81 (71.7)
T0	0 (0)	2 (100)	0 (0)	0 (0)
Total	48 (66.7)	24 (33.3)	34 (28.3)	86 (71.7)

Comparison of Cytology and Fibroblast Growth Factor Receptor 3 Mutations in the Transurethral Resection of Bladder Tumor Group

Cytology was either not performed or nondiagnostic for 10 patients in the TURBT group. Of the remaining 62 patients, 20 (32.3%) were diagnosed as being positive for tumor cells by cytology and 42 (67.7%) harbored a mutation in FGFR3 as demonstrated using SSCP analysis (Table 3). When the data were analyzed according to stage, the largest disparity between cytology and FGFR3 mutation detection was recorded in patients with Ta lesions (positive results were recorded in 17.9% and 64.1% of patients, respectively). Although the sample numbers were smaller in the TURBT/T1–T4 group, FGFR3 mutations were identified in 90.9% of patients using SSCP analysis, whereas cytology identified the presence of tumor cells in 45.5% of the patients. In the TURBT group, cytology and SSCP performed comparably well at detecting patients with Tis lesions. In the two patients in whom no lesion was detected by cystoscopy, cytology identified both urine samples as positive whereas SSCP analysis of FGFR3 revealed no detectable mutation. To date, these patients have not presented with a bladder tumor.

Table 3. Tumor Stage, Cytology, and Fibroblast Growth Factor Receptor 3 Mutation Status in the Transurethal Resection of Bladder Tumor Group

Stage	Cytology		SSCP
	Positive (%)	Negative (%)	Positive (%)	Negative (%)
SSCP: single-strand conformation polymorphism.
Ta	7 (17.9)	32 (82.1)	25 (64.1)	14 (35.9)
Tis	6 (60)	4 (40)	7 (70)	3 (30)
T1–T4	5 (45.5)	6 (54.5)	10 (90.9)	1 (9.1)
T0	2 (100)	0 (0)	0 (0)	2 (100)
Total	20 (32.2)	42 (67.7)	42 (67.7)	20 (32.2)

Combining the data from cytology and SSCP analysis of FGFR3 revealed a high rate of detection of bladder tumors. Together, cytology and SSCP identified tumor presence in 90.9% of the T1–T4 lesions, 80% of the Tis lesions, and 74.4% of the Ta lesions. Across all grades and stages, 78.3% of the bladder tumors were detected in the TURBT group using cytology and/or FGFR3 mutation analysis.

Spectrum of Fibroblast Growth Factor Receptor 3 Mutations in the Transurethral Resection of Bladder Tumor Group

Table 4 shows the spectrum and location of FGFR3 mutations recorded in the TURBT group. In this group, 67 mutations in FGFR3 were identified in 48 urine samples. Fifty-two (77.6%) of these mutations were located in exon 7, either at codon 248 or codon 249. The majority of these mutations were identified in urine sediment DNA samples from patients with Ta lesions, which is consistent with previous reports.23–25 In addition, we identified 13 mutations in exon 10, 8 of which were located in codon 375 and 5 of which were located in codon 386. Mutations in codon 386 resulted in a phenylalanine to a leucine change. To our knowledge, this is the first report of this mutational change. Two mutations were detected in exon 15, both of which represented a lysine to glutamic acid change in codon 652.

Table 4. Spectrum of Fibroblast Growth Factor Receptor 3 Mutations in the Transurethral Resection of the Bladder Tumor and Cystectomy Groups

Exon	Codona	Mutation	Predicted effect	Frequency in TURBT group	Frequency in cystectomy group
TURBT: transurethral resection of bladder tumor. a Codons are numbered according to the FGFR3b cDNA open reading frame.
7	248	CGC → TGC	Arg → Cys	5	2
7	249	TCC → TGC	Ser → Cys	27	9
7	248 and 249	As above	As above	10	1
10	373	AGT → TGT	Ser → Cys	0	1
10	375	TAT → TGT	Tyr → Cys	8	14
10	386	TTC → CTC	Phe → Leu	5	3
15	652	AAG → GAG	Lys → Glu	2	7

In the majority of samples studied, only one mutation was identified. However, 19 patients had multiple mutations. In 10 of these patients, 2 separate mutations were identified in exon 7—1 mutation at codon 248 and 1 mutation at codon 249. In the remaining samples, mutations were detected in two separate exons.

Fibroblast Growth Factor Receptor 3 Mutations in the Cystectomy Group

Of 120 urine samples analyzed, 34 (28.3%) exhibited FGFR3 mutations in SSCP analysis (Table 2). Of the 34 urine sediment samples displaying mutations, 32 were collected from patients with invasive tumors (T1–T4) and with 1 Ta lesion and 1 Tis lesion. FGFR3 mutations were distributed throughout all pathologic stages.

Comparison of Cytology and Fibroblast Growth Factor Receptor 3 Mutations in the Cystectomy Group

All patients in this group underwent a cystectomy. Cytology was performed on only 62 of the patients, the results of which are presented in Table 5. Of these, 56 were positive by cytologic evaluation, including 52 T1–T4 lesions and 4 Tis lesions. Within this group of patients, only 15 urine sediment samples displayed FGFR3 mutations. Cytology identified tumor cells in all patients with Tis lesions, whereas FGFR3 mutations were detected in one of four Tis specimens. Combining the results from cytology and SSCP analysis resulted in a detection rate of 93.5% (58 of 62 patients) for bladder tumors in this patient group.

Table 5. Tumor Stage, Cytology, and Fibroblast Growth Factor Receptor 3 Mutation Status in the Cystectomy Group

Stage	Cytology		SSCP
	Positive (%)	Negative (%)	Positive (%)	Negative (%)
SSCP: single-strand conformation polymorphism.
Ta	0 (0)	0 (0)	0 (0)	0 (0)
Tis	4 (100)	0 (0)	1 (25)	3 (75)
T1–T4	52 (89.7)	6 (10.3)	14 (24.1)	44 (75.9)
Total	56 (90.3)	6 (9.7)	15 (24.2)	47 (75.8)

Spectrum of Fibroblast Growth Factor Receptor 3 Mutations in the Cystectomy Group

In the cystectomy group, we identified 38 mutations from 34 patients (Table 4). All but two of these mutations were detected in urine sediment DNA samples from patients with invasive disease (T1–T4). The exceptions include one Tis lesion with a mutation in codon 249 of exon 7 and a Ta lesion that harbored a mutation in exon 10. One novel point mutation was identified at codon 373, which resulted in a serine to cysteine alteration. In the cystectomy group, the majority of FGFR3 mutations were detected in exon 10 (Table 4).

Cytology and Fibroblast Growth Factor Receptor 3 Mutations in Combined Groups

When the TURBT and cystectomy groups were combined, 122 samples were analyzed by both cytology and SSCP. Among these samples, 29 of 39 Ta, (74%), 12 of 14 Tis (86%), and 64 of 69 T1-T4 (93%) lesions had changes, as detected using at least 1 of the techniques, that were consistent with the presence of a tumor. Although cytology identified tumor cells in 61% of specimens across all grades and stages, FGFR3 mutations were detected in 47% of the specimens. Combining results from cytology and SSCP analysis correctly identified tumor presence in 105 of the 122 (86%) urine sediment specimens analyzed in the current study.

Fibroblast Growth Factor Receptor 3 Mutation and Time to First Disease Recurrence

Within the TURBT group, recurrent tumors were detected in 56 patients (81%; Table 1) during follow-up periods of 16–103 months. Of these, 38 displayed the presence of an FGFR3 mutation. For these patients, the mean time to first disease recurrence was 13.7 months. In the 18 patients in whom no FGFR3 mutation was detected, the mean time to first recurrence was 18.1 months (data not shown). No statistically significant difference was found between these two groups. Thirteen patients had no reported tumor recurrence more than 5 years after TURBT.For these patients, analysis of urine sediment DNA samples revealed FGFR3 mutations in 10 patients, with the remaining 3 patients displaying a normal FGFR3 SSCP profile. Analysis of tumor recurrence with respect to mutation location revealed no significant association between tumor recurrence and specific mutations. Recurrent disease in cystectomy patients is defined by the detection of a secondary, bladder-derived lesion after cystectomy. Of 17 patients, 4 had FGFR3 mutations involving exon 10 or exon 15. The times to tumor recurrence in the presence or absence of a detectable FGFR3 mutation were 9 and 12.4 months, respectively.

DISCUSSION

FGFR3 mutations have been identified in bladder tumors, with the highest frequency of mutation associated with low-grade, noninvasive lesions.21–25 Subsequent studies have confirmed this observation and suggested that the presence of an FGFR3 mutation is a strong indicator of superficial bladder cancer with a low tumor recurrence rate.25 From such observations, it was hypothesized that tumors harboring FGFR3 mutations shed tumor cells at a lower frequency or that tumor cells are less efficient at reattaching and seeding than tumor cells with a wild-type FGFR3 genotype. The observation that recurrences of tumors with mutated FGFR3 often present as solitary tumors in the same region of the bladder supports this hypothesis. In contrast, tumor cells with wild-type FGFR3 often appear in multiple lesions in different regions of the bladder. Although the identification of FGFR3 mutations is promising for the detection of tumor cells in urine samples, it was unclear whether tumor cell representation would be sufficient to enable detection with current techniques. The current study demonstrates the ability to detect FGFR3 mutations in exfoliated bladder cells in urine samples and presents a mutation profile that closely mirrors the profile reported in a range of bladder tumors.21–25

Holway et al.27 demonstrated that SSCP detects aberrantly migrating bands when there is a 5% representation of a mutated gene. In most cases in this study, the representation of tumor cells harboring FGFR3 mutations was considerably greater than 5%, as judged by the intensity of the signal generated by aberrantly migrating bands on SSCP analysis. This suggests that the release of cells from tumors with mutated FGFR3 is comparable to that observed in other bladder lesions.8, 9 The overall frequency of FGFR3 mutations detected in urine sediment DNA samples was 43% (82 of 192), a value comparable with the one reported in several series of bladder tumors.22–24 In addition, analysis of the FGFR3 gene and tumor stage revealed that the highest incidence of mutation was associated with noninvasive Ta bladder lesions.

Historically, low-grade, superficial lesions are difficult to detect using urine cytology because these tumor cells closely resemble normal cells sloughed from the bladder mucosa. In the TURBT group, the detection of FGFR3 mutations in urine sediment DNA samples from patients with Ta lesions correlated with tumor presence in 64% (25 of 39). Cytology performs very poorly in identifying these lesions, detecting only 18% of low-grade, noninvasive tumors. For two patients, cytology was recorded as positive, tumor absence was reported, and FGFR3 was recorded as normal in SSCP analysis. Follow-up of both patients for periods of 6 and 8 years, respectively, revealed that both remained bladder disease free.

In contrast with the findings in the TURBT group, cytology outperforms FGFR3 mutation analysis in the cystectomy group. Cytology identified the presence of 90% of late-stage tumors, whereas SSCP analysis identified only 24%. The combination of cytology and FGFR3/SSCP analysis across all tumor subgroups detected tumor presence in 86% of the samples analyzed.

The mutation profile recorded in the TURBT group confirmed the high incidence of point mutations associated with codons 248 and 249 of exon 7. The majority of mutations identified in this group (78%) were located in these codons, consistent with published data incorporating all FGFR3 mutations reported in bladder cancer to date.28 In addition, 13 point mutations were located in exon 10, and 2 were located in exon 15. With the exception of codon 386, the localization of mutations recorded in the TURBT group corresponded to mutation hotspots that had been reported by others.21–25, 28 However, a phenylalanine to leucine alteration at codon 386, which was identified in five patients, represents a novel mutation not previously reported. We also identified 10 cases in which mutations were found in codons 248 and 249 and 9 cases where multiple mutations were recorded in separate exons. The finding of double mutations in the FGFR3 gene in bladder carcinoma was reported recently.28 In the cystectomy group, a similar array of mutations were recorded involving the aforementioned mutation hotspots including representation of the novel mutation at codon 386. A second mutation found in codon 373, which resulted in a serine to cysteine change, has not been found in bladder carcinoma but has been reported with thanatophoric dysplasia.23 The frequency of mutations in the cystectomy group was lower than that recorded in the TURBT group, which is consistent with the finding that FGFR3 mutations are associated more frequently with superficial bladder tumors. Of 38 FGFR3 mutations detected in the cystectomy group, the majority of mutations were found in exon 10. This is in contrast with the findings in the TURBT group, in which most mutations were localized in exon 7. In both patient groups, only point mutation events were associated with alterations in the FGFR3 gene. In addition, analysis of mutations with respect to tumor grade, rather than stage, also revealed a strong association, showing that the two are linked closely in bladder carcinoma. Analysis of urine sediment DNA samples represents a sampling of cells from the bladder mucosa. Therefore, we do not know whether the FGFR3 mutation detected with SSCP analysis originates from the late-stage lesions that characterize this group of patients. Pathology reports most often record the most aggressive lesion present, but this does not exclude the possibility of a coexisting superficial lesion in the bladder mucosa field. To resolve this issue, we initiated a search for the mutation identified in exfoliated cells in the archived tumor tissue samples of patients. However, the lack of availability of tissue samples from the time of urine sampling prevented this approach. We have initiated a prospective trial that involves the analysis of FGFR3 in excised tumor tissue specimens and urine sediment DNA samples obtained on the same clinic visit.

Studies on the biologic significance of FGFR3 mutations suggest that it plays an oncogenic role in tumorigenesis. For instance, Webster and Donoghue29 demonstrated that mutated FGFR3, when targeted to the membrane, transforms NIH3T3 cells. In addition, specific mutations in FGFR3, including those at codons 248 and 652, caused constitutive activation of the receptor, albeit through different mechanisms.19, 20 The mutation at codon 248 results in an amino acid substitution of an arginine for a cysteine, thereby forming an intermolecular disulfide bond, that leads to ligand-independent dimerization of the receptor. In the current study, 77 of the 94 identified missense mutations were predicted to create a cysteine residue. The mutation at codon 652, which was found in nine patients, constitutively activated the receptor by changing the conformation of the activation loop in the tyrosine kinase domain. The biologic effect of the novel mutation identified in eight patients at codon 386 still is not known.

In a previous study, in which 57 patients with superficial disease were followed prospectively over a 12-month period, FGFR3 mutation status was the strongest predictor of tumor recurrence when compared with grade and stage.25 In this study, 56 patients with recurrent disease were followed for 16–103 months. The time to first tumor recurrence ranged from 3 to 54 months. Within this group, the FGFR3 mutation status did not have predictive value for tumor recurrence or time to first tumor recurrence. A 12-month cutoff25 revealed that 68% and 39% of patients presented with recurrent tumors in this time period in the presence or absence of FGFR3 mutations, respectively. In the 13 patients in whom no tumor recurrence was observed, FGFR3 mutations were recorded in 10 patients, with 9 mutations occurring in either codon 248 or codon 249 of exon 7. Although the frequency of involvement of these two codons was high in these patients, a similar mutation profile was recorded in urine sediment DNA samples from patients displaying recurrent disease. Recurrent disease in cystectomy patients was characterized by the appearance of a transitional cell carcinoma at a secondary site after cystectomy. This was recorded in 17 patients, 4 of whom harbored FGFR3 mutations. No significant difference was recorded between the presence or absence of an FGFR3 mutation and the time to tumor recurrence. It is noteworthy that the four mutations identified in tumor recurrence in the cystectomy group harbored FGFR3 mutations in either codon 10 or codon 15.

During the preparation of this article, Van Rhijn et al.30 reported a highly sensitive and specific method for detection of tumor presence using a combination of mutation analysis and microsatellite analysis. The data of Van Rhijn et al. and the current study demonstrate the utility of using molecular markers for screening in bladder carcinoma. However, the evaluation of any single locus has shown limited clinical utility over established practices involving cystoscopy and cytology. The discovery of the high-frequency of mutations in FGFR3 associated with superficial bladder tumors identifies the most frequently altered gene linked to bladder carcinoma, albeit in tumors that generally are benign. FGFR3 mutation analysis and standard cytology present a noninvasive assay for monitoring tumor recurrence in the follow-up of bladder carcinoma. We have not determined the prognostic significance associated with the molecular status of the FGFR3 gene. Such links may be established in prospective studies involving a greater number of patients. In the current study, we demonstrated the feasibility of such an assay, although the labor-intensive nature of SSCP analysis indicates the need for a simpler assay for detecting mutations. As we learn more about the molecular events underlying different pathways to malignancy in the bladder, the potential of a noninvasive molecular screen to aid in the clinical decision-making process becomes more realistic.