SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. Materials and Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. Disclosure Statement
  8. References

The fibroblast growth factor receptor (FGFR)-3 gene encodes a receptor tyrosine kinase that is frequently mutated in non-muscle invasive bladder cancer (NMIBC). A sensitive and quantitative assay using peptide nucleic acid-mediated real-time PCR was developed for detecting FGFR3 mutations in the urine samples and evaluated as a molecular marker for detecting intravesical recurrence of NMIBC in patients undergoing transurethral resection of bladder tumor. FGFR3 mutation was examined in tumor tissues and serially taken pre- and postoperative urine sediments in 45 NMIBC patients with a median follow up of 32 months. FGFR3 mutations were detected in 53.3% (24/45) of primary tumor tissues, among which intravesical recurrence developed in 37.5% (9/24) of cases. FGFR3 mutation in the primary tumor was not a significant prognostic indicator for recurrence, while the proportion of FGFR3 mutation (i.e. tumor cellularity was ≥11%) in the preoperative urine sediments was a significant indicator for recurrence in patients with FGFR3 mutations in the primary tumors. FGFR3 mutations were detected in 78% (7/9) of postoperative urine samples from recurrent cases with FGFR3 mutations in the tumor, while no mutations were detected in the urine of 15 non-recurrent cases. Urine cytology was negative in all cases with FGFR3 mutations in the primary tumors, while the sensitivity of cytological examination was as high as 56% (5/9) in cases showing wild-type FGFR3 in the primary tumors. Urine FGFR3 mutation assay and cytological examination may be available in the future as complementary diagnostic modalities in postoperative management of NMIBC. (Cancer Sci 2009)

Urothelial carcinoma (UC) is a histological subtype accounting for more than 90% of all bladder cancers, and there are 357 000 new cases every year worldwide.(1) Bladder UCs are generally divided into two groups for clinical management, depending on the pathological stage. Most of the newly diagnosed UCs are non-muscle invasive bladder cancer (NMIBC; i.e., pTa or pT1), and the initial treatment is transurethral resection of bladder tumor (TURBT). After the initial TURBT, the patients undergo intensive surveillance by cystoscopic examination at regular intervals; usually every 3 months, because up to 70% of these patients will experience intravesical recurrence, and 10–30% of the lesions will progress to life-threatening muscle-invasive disease (≥pT2).(2) Cystoscopy is an inconvenient, invasive, and expensive diagnostic modality, but currently it is the gold standard for detecting intravesical recurrence in the postoperative follow up. Although urine bound diagnostic tests including urinary cytology, nuclear matrix protein (NMP)22, and bladder tumour antigen (BTA) tests are used in the management after TURBT or bladder cancer screening, their usefullness is limited due to their poor sensitivity or specificity.(3) In previous reports, various molecular markers detectable in urine have been considered as a useful and non-invasive clinical assay improving the sensitivity of conventional tests.(4–10) In urine-based detection assays, contamination with normal urothelium or leucocytes can mask the signals of targeted somatic mutations.(11)

Fibroblast growth factor receptor (FGFR)-3 belongs to a family of structurally related tyrosine kinase receptors (FGFR1–4), and plays important roles in many biological processes including embryogenesis, proliferation, differentiation, and angiogenesis.(12) Recent reports have demonstrated that constitutively activated FGFR3 mutations exist in more than 50% of primary bladder UC.(13)FGFR3 mutations are especially prevalent in the low-grade papillary tumors (pTa/G1), but they are infrequent in high-grade or high-stage UC.(13,14)FGFR3 mutation in urine sediments may be a suitable biomarker for detection of low-grade and low-stage UC. Previous studies revealed that mutation of FGFR3 in the voided urine can be detected at high sensitivity in patients with FGFR3-mutated bladder UC.(15–17) However, there is no report validating the feasibility and usefulness of detecting FGFR3 mutation in the voided urine samples by serial determinations during follow up after TURBT. Recently, we have reported an assay protocol for detecting FGFR3 mutations in bladder tumor tissues and urine sediments by peptide nucleic acid (PNA)-mediated real-time PCR clamping assay.(17) In PNA-mediated PCR clamping, PNA is designed to anneal to a wild-type DNA sequence and inhibits the annealing of PCR primer to the wild-type alleles, resulting in preferential amplification of the mutated alleles. With 50 ng of genomic DNA as a template, this method allows sensitive and quantitative detection of the FGFR3 mutations in mutational hotspots in exons 7, 10, and 15 in bladder cancer. In the present study, we modified the protocol of the PNA-mediated PCR clamping assay to achieve quantitative detection of the FGFR3 mutations present in the urine samples at a concentration of 1% in only 1 ng of genomic DNA available as a template for PCR. With this the revised protocol, we assessed the usefulness of FGFR3 mutations as a diagnostic modality in the voided urine samples for the postoperative management of NMIBC. This is considered the first report addressing the significance of FGFR3 mutations in preoperative urine sediments as a novel indicator predicting the risk of intravesical recurrence of NMIBC.

Materials and Methods

  1. Top of page
  2. Abstract
  3. Materials and Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. Disclosure Statement
  8. References

Subjects and collection of the tumor tissues and voided urine samples after the initial TURBT.  The patients undergoing TURBT from April 2002 through March 2005 at the Departments of Urology at Tochigi Cancer Center Hospital and Nara Medical University Hospital were enrolled in this study. All participants had received study information and signed a written informed consent form. The voided urine samples before the initial TURBT were taken from the patients. The resected tumors were examined histologically and staged and graded according to the 2002 TNM classification and the 1973 World Health Organization (WHO) classification systems, respectively.(18,19) A total of 45 subjects with NMIBC were eligible for the study and were followed up until the histological diagnosis of tumor recurrence or up to 3 years postoperatively by routine cystoscopy and urine cytological examination. The median follow-up period was 32 months (range 4–36 months). The patients were monitored by routine cystoscopy and urine cytology at 1, 3, 6, 9, 12, 15, 18, 21, 24, 30, and 36 months after the initial TURBT. Intravesical recurrence was confirmed by histological diagnosis of tumor tissues obtained during TURBT for recurrence. The voided urine samples were divided and subjected to urine cytology and DNA extraction for gene testing. The urine samples were stored at −20°C until DNA extraction.

DNA extraction and measurement of DNA concentration.  DNA extraction from the tumor tissues and peripheral blood lymphocytes (PBL) was carried out as described previously.(6) DNA extraction from the urine samples was carried out with the QIAamp DNA Blood Mini Kit (Qiagen, Valencia, CA, USA) according to the manufacturer’s instructions. Briefly, the urine sample in a 50-mL tube was centrifuged at 180g for 5 min. The cell pellet was digested by Qiagen protease and subjected to DNA extraction by column centrifugation. In the final step, DNA was eluted from the column in 150 μL of the elution buffer. The genomic DNA concentration was determined by ultraviolet measurement using an ND-1000 spectrophotometer (NanoDrop Technologies, Wilmington, DE, USA). For analysis of samples with DNA concentrations less than 50 ng/μL, DNA concentration was quantified by real-time PCR using LightCycler (Roche Diagnostics, Mannheim, Germany). Quantification was carried out with the same primer pairs used for PNA-mediated real-time PCR clamping for amplification of FGFR3 exon 7.(17) Serially diluted assay standards were prepared by adjusting the genomic DNA concentrations to 100, 10, 1, and 0.1 ng/μL. DNA samples and assay standards were subjected to real-time PCR in a 20-μL reaction volume containing genomic DNA, 10 picomole of each primer, and 10 μL of QuantiTect PCR master mix (Qiagen) containing SYBR Green I dye. The conditions of real-time PCR are described in Table 1. DNA concentration was calculated from the crossing points (CP) of the assay standards and samples according to the fit points method on LightCycler Data Analysis software version 3.5 (Roche Diagnostics corporation).

Table 1.   Sequences of PCR primers and peptide nucleic acid (PNA), and PCR conditions
Real-time PCRSequence of primers and PNAPNA concentration (μM)Cycle no.PNA binding step (°C)Annealing step (°C)
DNA quantification5′-TGA GCG TCA TCT GCC CCC ACA GAG-3′ (sense)4564
5′-GGG CCC ACC TTG CTG CCA TTC A-3′ (antisense)    
Main amplifier for exon 75′-TGA GCG TCA TCT GCC CCC ACA GAG-3′ (sense)0.4457264
5′-GGG CCC ACC TTG CTG CCA TTC A-3′ (antisense)    
H2N-AGC GCT CCC CGC ACC-N2H (PNA)    
Main amplifier for exon 105′-CCA GGC CTC AAC GCC CAT GTC TTT-3′ (sense)1456758
5′-ACC CCG TAG CTG AGG ATG CCT GCA-3′ (antisense)    
H2N-CAT ACA CAC TGC CCG C-N2H    
Main amplifier for exon 155′-GCA ATG TGC TGG TGA CCG AG-3′ (sense)2457060
5′-CGG GCT CAC GTT GGT CGT CT-3′ (antisense)    
H2N-GGT CGT CTT CTT GTA GT-N2H    

PNA-mediated pre-main amplifier method for the low-copy number DNA template.  Previously, we reported a PNA-mediated real-time PCR clamping assay for detection of FGFR3 mutations.(17) This method enabled sensitive and reproducible detection of FGFR3 mutations in cases where 50 ng of genomic DNA were available as the template for PCR. In the PNA-mediated PCR-clamping, the chance of nucleotide misincorporation to the PNA binding sequence increases in reverse correlation with the amount of template DNA. When the amount of template DNA was 1 ng in genomic DNA (equivalent to 300 copies), mutations were hardly distinguishable from those caused by misincorporation of dNTPs. To overcome this pitfall, we modified the assay protocol to detect FGFR3 mutations at a concentration of 1% (three copies) in 1 ng (300 copies) of the template genomic DNA. We called the newly established method as PNA-mediated pre-main amplifier (PPA), which consisted of two steps of amplification (Fig. 1). Low-copy number DNA template was amplified by the pre-amplifier step and then set on the main amplifier to perform the PNA-mediated real-time PCR clamping. Pre-amplification was carried out in a PCR tube using DNA Engine Dyad Thermal Cycler (MJ Research, Watertown, MA, USA) in 20-μL aliquots consisting of 1 ng of genomic DNA, 10 μL of QuantiTect PCR master mix, and 10 picomole of each primer. The sequences of primer pairs were as reported previously.(17) Conditions of the thermal cycling in the pre-amplifier step were as follows: denaturing at 95°C for 15 min, amplification of seven cycles consisting of heat denaturation at 94°C for 15 s, annealing at 64°C (exon 7), 58°C (exon 10), and 60°C (exon 15) for 20 s, and extension at 72°C for 20 s. After final cooling to 4°C, 5 μL of the solution containing 2.5 μL of QuantiTect PCR master mix and 2.5 μL of PNA solution were added and mixed by gentle pipetting. The sequences of PNA and the final concentrations are listed in Table 1. Of 25 μL of the mixed solution, 20 μL was transferred to a capillary tube for the LightCycler and set on the main amplifier performing the real-time PCR (Table 1). CP of PPA were determined by the fit points method.

image

Figure 1.  Schematic diagram of the one-step and two-step peptide nucleic acid (PNA)-mediated PCR clamping. The feasibility of PNA-mediated PCR clamping was highly influenced by the amount of the template DNA. (a) Fifty nanograms of genomic DNA containing a low proportion of mutated DNA was used as a template. PNA-mediated PCR enabled preferential amplification of the mutated DNA, leading to enrichment of the mutated DNA fragments. The red circle indicates a mutated DNA sequence. (b) One nanogram of genomic DNA containing only wild-type DNA was used as the template. Misincorporation of dNTPs occurred in the sequences of the PNA-binding site, due to a failure in DNA synthesis brought about by DNA polymerase. When the nucleotide misincorporation (blue triangle) occurred in the early cycles of PNA-mediated PCR, a mutagenized sequence was subjected to subsequent amplification. (c) One nanogram of the template DNA containing only wild-type DNA was used for the PNA-free pre-amplifier step prior to the PNA-mediated main-amplifier. Seven PCR cycles of the pre-amplifier step generated sufficient copies of fibroblast growth factor receptor (FGFR)-3 DNA fragments, which were used as the template for the main-amplifier. The PNA-mediated reaction produced a randomly mutagenized DNA sequence that could slip from PNA clamping. However, all of these mutagenized fragments resulted in dispersion of mutagenesis signals and were scarcely detectable in the direct sequencing analysis. (d) One nanogram of genomic DNA with a low proportion of mutant DNA was used as the template. The PNA-free pre-amplifier increased the copy numbers of the FGFR3 molecules as a whole, leading to a successful preferential amplification of the mutated DNA fragments in the main-amplifier. The black, red, and blue fragments indicate the wild-type, mutated, and mutagenized DNA fragments, respectively.

Download figure to PowerPoint

Detection of FGFR3 mutations in the tumor tissues and urine samples.  The assay standards for mutation analysis of each exon were prepared as described previously.(17) In the clinical samples with DNA concentrations of ≥50 ng/μL, mutation analyses were carried out according to the one-step assay using 50 ng of genomic DNA as the template.(17) In the samples with DNA concentrations ranging from 0.125 to 50 ng/μL, a modified protocol was adapted using 1 ng of genomic DNA as the template. In each run, we defined CP of the assay standard corresponding to 1% tumor cellularity as the minimal detectable dose for FGFR3 mutations. Accordingly, a sample showing CP less than that of the 1% assay standard was considered mutation positive and subjected to direct sequencing to identify the mutational types.(17) The tumor cellularities of the mutation-positive samples were determined by a regression analysis using a standard curve obtained from 100, 10, and 1% assay standards. The samples with DNA concentrations less than 0.125 ng/μL were regarded as unavailable samples unless they could be enriched in DNA concentration.

Statistical analysis.  Statistical analyses and drawing figures were done using PRISM software version 4.00 (GraphPad Software, Inc., San Diego, CA, USA). Student’s t-test, Chi-square test, and Fisher’s exact test were used to analyze the correlations between the clinicopathological variables and FGFR3 mutational status in the primary tumors. Recurrence-free survival curves were plotted according to the Kaplan–Meier method, and the log-rank test was applied for statistical significance. A receiver operating characteristic (ROC) curve was used to define the optimal cut-off value of tumor cellularity in the urine sediments. The non-parametric variables were analyzed by the Mann–Whitney U-test. A P-value of <0.05 was considered significant.

Results

  1. Top of page
  2. Abstract
  3. Materials and Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. Disclosure Statement
  8. References

Optimization of PPA for detection of FGFR3 mutations in low-copy number DNA template.  The number of PCR cycles in the pre-amplifier step was critical for sensitive detection of FGFR3 mutations in the low-copy number DNA template. In a preliminary experiment to optimize the number of PCR cycles in the pre-amplifier step, the concentration of FGFR3 mutation in the sample was adjusted to either 1 or 0% and the difference in CP was maximal when seven cycles of PCR were used in the pre-amplification step (= 0.001). In this condition, we compared the coefficients of variation (CV) between the one-step and two-step assays using 1 ng of wild-type genomic DNA as the template (Fig. 2a,b left). The assay CV of CP in the PPA method was much smaller than that of the one-step assay (2.61 vs 7.38%, respectively). The sequencing analysis of the amplified DNA fragments in the one-step assay revealed point mutations caused by nucleotide misincorporation virtually in all samples (Fig. 2a right), whereas those amplified by the PPA assay showed no recognizable mutations except for a slight increase in the background signals. These results indicated that the PPA method circumvented the chance of a nucleotide misincorporation and minimized the CV of the CP for wild-type DNA or 0% standard (Fig. 2b right). In this condition, the assay standards with 100, 10, and 1% mutations in exon 7 of FGFR3 and 0% (wild type) were amplified by the PPA method using 1 ng of DNA template, and the results were compared with those of the one-step assays. The CPs of the assay standards were statistically significant between each other (Fig. 3a) and direct sequencing analysis of the 1% standard revealed that all of eight samples showed S249C mutation (TCC[RIGHTWARDS ARROW]TGC) in exon 7. These results demonstrated that the mutations were reliably detected in the samples containing ≥1% mutated DNA using only 1 ng of DNA template, and that the PPA method overcame the limitation of our prior study.

image

Figure 2.  Two-step peptide nucleic acid (PNA)-mediated real-time PCR clamping decreased the variance of crossing points (CP) and avoided the enhanced misincorporation of dNTPs. In the experiment for fibroblast growth factor receptor (FGFR)-3 exon 7, 1 ng of genomic DNA containing only wild-type FGFR3 was amplified in octaplicate (= 8) by the (a) one-step and (b) two-step PNA-mediated real-time PCR. The PNA-mediated pre-main amplifier method consisted of seven cycles of pre-amplification followed by 45 cycles of the main amplification step. The amplification curves and representative results of direct sequencing analysis of the PCR products are shown in the left and right panels, respectively. The mean CP ± SD and coefficient of variance (CV) are shown above the amplification curves. (a) Arrowheads indicate disincorporated nucleotides caused by PNA clamping. The uppermost sequence indicates the wild-type FGFR3 in exon 7. The PNA binding site is surrounded by a rectangle. Codons 248 and 249 are underlined.

Download figure to PowerPoint

image

Figure 3.  Validation for sensitive detection of fibroblast growth factor receptor (FGFR)-3 mutation in trace amounts of the template DNA. (a) One nanogram of genomic DNA of four assay standards with 0, 1, 10, and 100% of tumor cellularity was amplified in octaplicate samples (= 8) under the conditions of the two-step assays comprising seven amplification cycles of the peptide nucleic acid (PNA)-free pre-amplifier, followed by PNA-mediated real-time PCR clamping for exon 7 of the FGFR3 gene (solid lines). The data from the one-step assay are shown for comparison (dashed lines). The mean crossing points (CP) are plotted and the error bars represent 2 SD. Similar experiments were carried out to validate the assay protocols for (b) exon 10 and (c) exon 15.

Download figure to PowerPoint

In analysis of exons 10 and 15, seven amplification cycles in the pre-amplifier step were used to detect mutations in the samples containing ≥1% mutated DNA using 1 ng of genomic DNA as the template (Fig. 3b,c).

Correlation of FGFR3 mutations with the clinicopathological characteristics in NMIBC.  In analysis of FGFR3 mutations in 45 NMIBC samples, 24 (53.3%) tumors harbored activating mutations of FGFR3, and their correlations with the clinicopathological variables are summarized in Table 2. No variables showed significant correlation with FGFR3 mutations. Mutations were detected in six different codons. Mutations affecting the extracellular domain (exon 7) or transmembrane domain (exon 10) accounted for 95.8% (23/24) (Table 3). Intravesical recurrence was detected in 18 of 45 subjects (40%). The clinicopathological variables of the primary tumors, such as tumor stage, histological grade, tumor size, multiplicity, presence of carcinoma in situ lesion, and FGFR3 mutational status, did not correlate with the intravesical recurrence (Table 4).

Table 2.   Clinicopathological characteristics and fibroblast growth factor receptor (FGFR)-3 mutation status
VariablesTotalFGFR3 statusP-value
Wild-typeMutation% of mutation
  1. BCG, Bacille Calmette Guerin; CIS, carcinoma in situ.

Age (years)
 Mean ± SD63.0 ± 11.263.1 ± 12.762.8 ± 9.60.98
 Range36–8036–8045–78 
Sex
 Male351520570.27
 Female10 6 440 
Smoking history
 Present291118620.11
 Absent1610 638 
Tumor size (diameter, cm)
 <110 4 6600.73
 1–325131248 
 3<10 4 660 
Multiplicity
 Solitary271215560.71
 2–312 5 758 
 ≥4 6 4 233 
Pathological stage
 pTa19 712630.26
 pT126141246 
Tumor grade
 G1 6 3 3500.97
 G235161954 
 G3 4 2 250 
Concomitant CIS
 Present 3 2 1330.45
 Absent42192355 
BCG therapy
 No401921530.87
 Treated 5 2 360 
 Total45212453 
Table 3.   Fibroblast growth factor receptor (FGFR)-3 mutational types detected in this study
FGFR3Mutational typen%
CodonNucleotide
Exon 7R248CCGC[RIGHTWARDS ARROW]TGC416.7
S249CTCC[RIGHTWARDS ARROW]TGC833.3
Exon 10G372CGGC[RIGHTWARDS ARROW]TGC14.2
S373CAGC[RIGHTWARDS ARROW]TGC14.2
Y375CTAT[RIGHTWARDS ARROW]TGT937.5
Exon 15K652EAAG[RIGHTWARDS ARROW]GAG14.2
Total  24100
Table 4.   Correlation of the stage, histological grade, tumor size, multiplicity, concomitant CIS and fibroblast growth factor receptor (FGFR)-3 mutations in the tissue with intravesical tumor recurrence
VariablesTotalRecurrentNon-recurrentP-value*
  1. CIS, carcinoma in situ; *logrank test.

No. subjects451827 
 Stage
  pTa198111.00
  pT1261016
 Grade
  G16150.43
  G2351520
  G3422
 Tumor size (cm)
  <33512230.14
  ≥31064
 Multiplicity
  Solitary2710170.45
  Multiple18810
 Concomitant CIS
  Absent4216260.25
  Present321
 FGFR3 mutations in the tumor tissues
  Wild-type219120.89
  Mutation24915

Clinical usefulness of detecting FGFR3 mutation in the urine sediments.  A total of 429 voiding urine samples were taken from 45 cases, among which 61 samples were preoperative urine samples consisting of 35 from recurrent and 26 from non-recurrent cases (Table 5). The remaining 368 urine samples were obtained serially during follow up, among which 93 samples were from recurrent cases and 275 samples were from non-recurrent cases. The concentrations of genomic DNA extracted from the urine samples were quantified in all samples prior to the assay. Of 429 urine samples, 114 (26.6%) were not available for the assay because their DNA concentrations were <0.125 ng/μL. A total of 315 samples (73.4%) were subjected to the FGFR3 mutation detection assay. They were subjected to either the conventional PNA-mediated real-time PCR clamping assay or PPA method depending on their DNA concentrations.

Table 5.   Correlation of DNA concentration and fibroblast growth factor receptor (FGFR)-3 mutations in the urine samples with tumor recurrence
VariablesTotalRecurrentNon-recurrentP-value
  1. †Preoperative urine samples consisted of samples obtained in the initial transurethral resection of bladder tumor (= 18) and transurethral resection of bladder tumor for recurrence (= 17).

Collected urine samples429128301 
 Preoperative6135†26 
 Follow-up36893275 
DNA concentration (ng/μL)
 Total
  50≦83 (19.3%)23 (18.0%)60 (19.9%)0.885
  0.125–50232 (54.1%)71 (55.5%)161 (53.5%) 
  0.125<114 (26.6%)34 (26.6%)80 (26.6%) 
 Mean ± SD (ng/μL)  53.8 ± 187.6 36.2 ± 98.1  61.6 ± 214.3 
Detection of FGFR3 mutation in urine sediments
No. samples available for the assay31594221 
 Mutated26 (8.3%)20 (21.3%)6 (2.7%)<0.0001
 WT289 (91.7%)74 (78.7%)215 (97.3%) 

Risk of intravesical recurrence in patients showing FGFR3 mutation in tumor tissues and urine sediments.  In 21 of 24 cases with FGFR3 mutation in primary tumors, genomic DNA samples extracted from preoperative urine sediments before the initial TURBT were available for mutation detection assay. The sensitivity of FGFR3 mutation in the urine samples was 62% (13/21), and their mutational types coincided with those of the primary tumors in all cases. In 21 cases showing wild-type FGFR3 in primary tumor tissues, the preoperative urine samples were available for the assay in 13 cases and no mutations were found in these samples, showing the specificity of 100% (0/13). Tumor cellularities in the preoperative urine sediments were significantly higher in the recurrent cases than in the non-recurrent cases (Fig. 4a; = 0.008). An ROC curve analysis was performed to define the optimal cut-off value of tumor cellularity in the preoperative urine sediments. The area under the curve (AUC) was 0.847 (95% confidence interval, 0.669–1.026), and the cut-off value with optimal sensitivity and specificity was defined as 11% (Fig. 4b).

image

Figure 4.  The tumor cellularity in the preoperative urine predicted the risk of intravesical recurrence in patients with fibroblast growth factor receptor (FGFR)-3-mutated bladder urothelial carcinoma (UC). (a) The tumor cellularities in the preoperative urine sediments were plotted in the recurrent (= 9) and non-recurrent (= 12) groups of 21 cases with FGFR3-mutated bladder UC. The dashed line indicates the optimal cut-off point (≥11% and <11%) as determined with receiver operating characteristic (ROC) curve analysis shown in (b); *Mann–Whitney U-test. (b) A receiver operating characteristic curve was generated to define the optimal cut-off value of tumor cellularity in the preoperative urine sediments. The optimal cut-off value for predicting the intravesical recurrence was defined as the point closest to the upper-left corner of the graph (black arrow). (c) Recurrence-free survival curves according to the cut-off value (≥11% and <11%) of tumor cellularity in the preoperative urine sediments are shown. AUC, area under the curve; CI, confidence interval.

Download figure to PowerPoint

FGFR3 mutational status in the primary tumors was not a significant predictor of intravesical recurrence after TURBT (Table 4). In cases harboring FGFR3 mutations in the primary tumor, the levels of FGFR3 mutations in the preoperative urine sediments significantly correlated with the 3-year recurrence-free survival rates (83.3 vs 22.2%) (Fig. 4c), whereas the results of preoperative urine cytology did not correlate with the recurrence-free survival.

Serial determination of FGFR3 mutations in the voided urine samples during the follow-up period after the initial TURBT. FGFR3 mutations in serially obtained urine samples were assayed quantitatively in the postoperative follow-up period. Low-copy number DNA samples were amplified by the PPA method using 1 ng of genomic DNA as the template. In 21 cases harboring FGFR3 mutations in primary tumors, assay results were plotted on a 3-D line chart (Fig. 5). The preoperative sensitivities of urine FGFR3 mutations in the recurrent and non-recurrent cases were 88.9% (8/9) and 41.7% (5/12), respectively. FGFR3 mutations were not detected 1 month after the initial TURBT in all cases. FGFR3 mutations in the postoperative urine samples were detected in 7 of 9 (78%) recurrent cases. In two cases, mutations preceded cystoscopical detection of tumor relapse by 6 and 9 months, respectively. In analysis of the non-recurrent cases with FGFR3 mutation in primary tumors, the pre- and postoperative urine samples were available for the assay in 12 cases and no cases were positive for FGFR3 mutations except for five urine samples obtained before the initial TURBT. The frequencies of FGFR3 mutations in the postoperative urine samples were compared with the results of cytological examinations between recurrent and non-recurrent cases (Table 6). The sensitivity and specificity of FGFR3 mutations in the postoperative urine samples were defined as the positive or negative rates of FGFR3 mutations in recurrent or non-recurrent cases, respectively. In cases with FGFR3 mutation in the primary tumors, the sensitivity and specificity of urine FGFR3 mutations in the recurrent and non-recurrent cases were 78% (7/9) and 100% (15/15), respectively. Cytological examination was negative in all cases with FGFR3 mutations in the tumors. In analysis of cases with wild-type FGFR3 in the primary tumors, FGFR3 mutations were not detected in the postoperative urine samples in the recurrent cases (0/9) except for one non-recurrent case (1/12), indicating sensitivity of 0% and specificity of 92% (11/12). In analysis of non-recurrent cases showing wild-type FGFR3 in the tumor, S249C mutation was detected in one case at the 36th postoperative month with tumor cellularity of 2.1% (data not shown). This was the last session in postoperative follow up and no further information was available as to the clinical outcome in this case, although no abnormality was detected by cystoscopical and cytological examinations carried out simultaneously. The sensitivity and specificity of the cytological examination were 56% (5/9) and 100% (12/12) in those with wild-type FGFR3 in their primary tumors. When they were used in combination, the assay sensitivity, specificity, and diagnostic accuracy were 67% (12/18), 96% (26/27), and 84% (38/45) (Table 6). It was elucidated that these two diagnostic modalities may be used as a complementary approach to the postoperative management of NMIBC.

image

Figure 5.  The time-course analysis of tumor cellularity in the voided urine sediments in cases with mutated-fibroblast growth factor receptor (FGFR)-3. Tumor cellularities serially determined after initial transurethral resection of bladder tumor (TURBT) are plotted on 3-D line chart. All cases had FGFR3 mutations in their tumor tissues. Red and blue lines indicate the data of recurrent (= 9) and non-recurrent (= 12) cases, respectively. Two asterisks indicate the events of positive results prior to the cystoscopical detection of recurrent tumors. In recurrent cases, the last point of serial determination indicates tumor cellularity in the preoperative urine sediments obtained before second TURBT.

Download figure to PowerPoint

Table 6.   Association of fibroblast growth factor receptor (FGFR)-3 mutational status in primary tumors with cytological examination and/or FGFR3 mutational status in voided urine samples
AssayFGFR3 status in primary tumorRecurrent casesNon-recurrent casesAssay performance
PositiveNegativePositiveNegativeSensitivity (%)Specificity (%)Accuracy (%)P-value
  1. †Positivity is defined by the positive result in either urine cytology or FGFR3 mutation detection assay. NA, not analyzed.

FGFR3 assayMutated7201578100920.0001
Wild-type09111092521.0000
Total7111263996730.0042
Urine cytologyMutated09015010063NA
Wild-type5401256100810.0062
Total51302728100710.007
Combination†Mutated7201578100920.0001
Wild-type541115692760.0464
Total126126679684<0.0001

Discussion

  1. Top of page
  2. Abstract
  3. Materials and Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. Disclosure Statement
  8. References

In the PNA-mediated real-time PCR clamping, low yields of DNA from the urine samples seem to be a major limiting factor to define the assay sensitivity.(17) To increase the copy numbers of the targeted gene in the template DNA, we added a pre-amplification step comprising seven cycles of PCR prior to the PNA-mediated real-time PCR clamping. This modified assay protocol achieved a sensitive method detecting FGFR3 mutations at concentrations of ≥1% in 1 ng of the template DNA. One nanogram of genomic DNA with 1% of FGFR3 mutations corresponds to as few as three copies of mutated FGFR3. Estimated from Poisson distribution, the probability of missing mutated FGFR3 molecules was 4.98% on sampling the aliquot containing FGFR3 mutations at the concentration of 1% in 1 ng of the genomic DNA. The probabilities of missing mutated DNA molecules in the template increased up to 13.5% for two copies and 35.8% for a single copy, resulting in frequent false negatives. These assumptions suggested that 1% of tumor cellularity in 1 ng of the template DNA is actually the minimal detectable dose in the assay.

We developed a quantitative method to determine the proportion of FGFR3 mutations in the sample as tumor cellularity and this approach disclosed the correlation between tumor cellularity and intravesical recurrence of NMIBC.

Assessing the risk for intravesical recurrence and progression after TURBT is another major concern in the clinical management of NMIBC. Many efforts have been reported to establish any molecular alterations in tumor tissues as prognostic markers.(2,20) Although some earlier studies have analyzed FGFR3 mutation as a potential prognostic marker, the true prognostic value of FGFR3 mutation is still controversial.(21–23) Quantitative analysis revealed that the tumor cellularity in the preoperative urine sediments strongly correlates with the tumor recurrence (= 0.008). Moreover, the ROC analysis of the preoperative urine sediments determined the optimal cut-off value of FGFR3 mutations (11%) in the urine sediments that differentiates between the recurrent and non-recurrent tumors (Fig. 4). Using this cut-off value, the sensitivity, specificity, and diagnostic accuracy for detecting tumor recurrence were 77.8, 83.3, and 80.9%, respectively. This is the first report elucidating that the FGFR3 mutational status in the urine sediments may serve as a prognostic indicator for tumor recurrence in NMIBC (Fig. 4c), indicating that cell detachment or exfoliation into the urine strongly correlates with the tumor recurrence.

Sequential analyses of the urine samples taken after the initial TURBT revealed intravesical recurrence in almost 78% (7/9) of the recurrent cases showing FGFR3 mutations in the primary tumors, while the cytological examination of the same sample showed no positive results (Table 6). In two cases, FGFR3 mutations were detected prior to the cystoscopical detection, indicating that the urine FGFR3 assay can diagnose intravesical recurrence reliably in the postoperative management of NMIBT cases harboring FGFR3 mutations in primary tumors. As shown in Table 6, the sensitivity of FGFR3 mutations and urine cytology in the follow-up samples were mutually exclusive, indicating that the combination of urine FGFR3 assay and cytological examinations improved diagnostic sensitivity in detecting tumor relapse. van Rhijn et al. reported that the median sensitivity of the cytology for patients under surveillance was 48% and varied between histological grades, ranging from 17% for Grade 1, 34% for Grade 2, and 58% for Grade 3.(24) In the present study, the proportion of tumor grades was 13% (6/45), 78% (35/45), and 9% (4/45) for Grades 1, 2, and 3. Grade 1 and 2 tumors occupied more than 90% of the cohort and the sensitivities of urine cytology in the recurrent cases were 0% (0/9) and 56% (5/9) depending on the presence or absence of the FGFR3 mutations.

In conclusion, our results indicate the potential usefulness of the urine FGFR3 mutation assay as both a significant predictor of intravesical recurrence and a complementary urinary marker for monitoring patients undergoing TURBT in combination with urine cytology. Once FGFR3 mutations are detected in the primary tumors, the patients should be indicated for monitoring FGFR3 mutations in the urine sediments, particularly when the tumor cellularity in the preoperative urine sediments is above the cut-off level of 11%.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Materials and Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. Disclosure Statement
  8. References

This work was supported in part by a Grant-in-Aid for Cancer Research and for the 3rd Term Comprehensive 10-Year Strategy for Cancer Control from the Ministry of Health, Labor, and Welfare, Japan. Makito Miyake, MD is the awardee of a Research Resident Fellowship from the Foundation for Promotion of Cancer Research (Japan) for the 3rd Term Comprehensive 10-Year Strategy for Cancer Control (Tokyo, Japan).

References

  1. Top of page
  2. Abstract
  3. Materials and Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. Disclosure Statement
  8. References
  • 1
    Parkin DM, Bray F, Ferlay J, Pisani P. Global cancer statistics, 2002. CA Cancer J Clin 2005; 55: 74108.
  • 2
    Sugano K, Kakizoe T. Genetic alterations in bladder cancer and their clinical applications in molecular tumor staging. Nat Clin Pract Urol 2006; 3: 64252.
  • 3
    Wiener HG, Mian C, Haitel A, Pycha A, Schatzl G, Marberger M. Can urine diagnostic tests replace cystoscopy in the management of bladder cancer? J Urol 1998; 159: 187680.
  • 4
    Sidransky D, Von Eschenbach A, Tsai YC et al. Identification of p53 gene mutations in bladder cancers and urine samples. Science 1991; 252: 7069.
  • 5
    Mao L, Schoenberg MP, Scicchitano M et al. Molecular detection of primary bladder cancer by microsatellite analysis. Science 1996; 271: 65962.
  • 6
    Sugano K, Tsutsumi M, Nakashima Y et al. Diagnosis of bladder cancer by analysis of the allelic loss of the p53 gene in urine samples using blunt-end single-strand conformation polymorphism. Int J Cancer 1997; 74: 4036.
  • 7
    Shigyo M, Sugano K, Fukayama N et al. Allelic loss on chromosome 9 in bladder cancer tissues and urine samples detected by blunt-end single-strand DNA conformation polymorphism. Int J Cancer 1998; 78: 4259.
  • 8
    Van Rhijn BW, Lurkin I, Kirkels WJ, Van Der Kwast TH, Zwarthoff EC. Microsatellite analysis-DNA test in urine competes with cystoscopy in follow-up of superficial bladder carcinoma: a phase II trial. Cancer 2001; 92: 76875.
  • 9
    Shigyo M, Sugano K, Tobisu K, Tsukamoto T, Sekiya T, Kakizoe T. Molecular followup of newly diagnosed bladder cancer using urine samples. J Urol 2001; 166: 12805.
  • 10
    Steiner G, Schoenberg MP, Linn JF, Mao L, Sidransky D. Detection of bladder cancer recurrence by microsatellite analysis of urine. Nat Med 1997; 3: 6214.
  • 11
    Van Rhijn BW, Lurkin I, Chopin DK et al. Combined microsatellite and FGFR3 mutation analysis enables a highly sensitive detection of urothelial cell carcinoma in voided urine. Clin Cancer Res 2003; 9: 25763.
  • 12
    Johnson DE, Williams LT. Structural and functional diversity in the FGF receptor multigene family. Adv Cancer Res 1993; 60: 141.
  • 13
    Billerey C, Chopin D, Aubriot-Lorton MH et al. Frequent FGFR3 mutations in papillary non-invasive bladder (pTa) tumors. Am J Pathol 2001; 158: 195559.
  • 14
    Kimura T, Suzuki H, Ohashi T, Asano K, Kiyota H, Eto Y. The incidence of thanatophoric dysplasia mutations in FGFR3 gene is higher in low-grade or superficial bladder carcinomas. Cancer 2001; 92: 255561.
  • 15
    Rieger-Christ KM, Mourtzinos A, Lee PJ et al. Identification of Fibroblast Growth Factor Receptor 3 mutations in urine sediment DNA samples complements cytology in bladder tumor detection. Cancer 2003; 98: 73744.
  • 16
    Van Oers JM, Lurkin I, Van Exsel AJ et al. A simple and fast method for the simultaneous detection of nine Fibroblast Growth Factor Receptor 3 mutations in bladder cancer and voided urine. Clin Cancer Res 2005; 11: 77438.
  • 17
    Miyake M, Sugano K, Kawashima K et al. Sensitive detection of FGFR3 mutations in bladder cancer and urine sediments by peptide nucleic acid-mediated real-time PCR clamping. Biochem Biophys Res Commun 2007; 362: 86571.
  • 18
    Sobin LH, Wittekind C, eds. TNM Classification of Malignant Tumors, 6th edn. New York: John Wiley & Sons, 2002.
  • 19
    Mostofi FK, Sobin LH, Torloni H. Histological Typing of Urinary Bladder Tumors: Intenational Classification of Tumors, No. 10. Geneva: World Health Organization, 1973: 2131.
  • 20
    Knowles MA. Role of FGFR3 in urothelial cell carcinoma: biomarker and potential therapeutic target. World J Urol 2007; 25: 58193.
  • 21
    Van Rhijn BW, Lurkin I, Radvanyi F, Kirkels WJ, Van Der Kwast TH, Zwarthoff EC. The fibroblast growth factor receptor 3 (FGFR3) mutation is a strong indicator of superficial bladder cancer with low recurrence rate. Cancer Res 2001; 61: 12658.
  • 22
    Hernández S, López-Knowles E, Lloreta J et al. Prospective study of FGFR3 mutations as a prognostic factor in nonmuscle invasive urothelial bladder carcinomas. J Clin Oncol 2006; 24: 366471.
  • 23
    Burger M, Van Der Aa MN, Van Oers JM et al. Prediction of progression of non-muscle-invasive bladder cancer by WHO 1973 and 2004 grading and by FGFR3 mutation status: a prospective study. Eur Urol 2007; 54: 83543.
  • 24
    Van Rhijn BW, Van Der Poel HG, Van Der Kwast TH. Urine markers for bladder cancer surveillance: a systematic review. Eur Urol 2005; 47: 73648.