A Simple and Fast Method for the Simultaneous Detection of Nine Fibroblast Growth Factor Receptor 3 Mutations in Bladder Cancer and Voided Urine

Materials and Methods

Sample collection and DNA extraction. Paraffin-embedded tumor tissue samples were collected from patients who underwent surgery at the Erasmus MC or other hospitals in the Rotterdam area. All patients signed written informed consent. Voided urine samples were collected before surgery. DNA was obtained as previously described (14, 15).

Polymerase chain reaction. Initially, three regions of the FGFR3 gene, located in exons 7, 10, and 15, were amplified by PCR as described previously (8). These regions comprise the following potential codon mutations: R248C and S249C (exon 7), G372C, Y375C, and A393E (exon 10), and K652E, K652Q, K652M, and K652T (exon 15).

After analysis of the secondary structure and folding free energy of the exon 7 and exon 10 PCR products using DNAMAN software (Lynnon Corporation, Vaudreuil, Canada), new primer sequences were constructed. In the end, the following primer pairs were used: for exon 7, 5′-AGTGGCGGTGGTGGTGAGGGAG-3′ and 5′-GCACCGCCGTCTGGTTGG-3′; for exon 10, 5′-CAACGCCCATGTCTTTGCAG-3′ and 5′-AGGCGGCAGAGCGTCACAG-3′; and for exon 15, 5′-GACCGAGGACAACGTGATG-3′ and 5′-GTGTGGGAAGGCGGTGTTG-3′. A multiplex reaction for all three regions was successfully set up. The multiplex PCR was done in a volume of 15 μL, containing 1× PCR buffer, 1.5 mmol/L MgCl₂, 0.5 units Taq polymerase (Promega, Madison, WI), 0.17 mmol/L deoxynucleotide triphosphates (Roche, Basel, Switzerland), 10 pmol of exon 7 and exon 15 primers, 7.5 pmol of exon 10 primers (Invitrogen, Carlsbad, CA), 5% glycerol (Fluka, Buchs SG, Switzerland), and 1 to 250 ng of genomic DNA. Cycling conditions were as follows: 5 minutes at 95°C, 35 cycles at 95°C for 45 seconds, 60°C for 45 seconds, and 72°C for 45 seconds, followed by 10 minutes at 72°C. Products were treated with Exonuclease I (ExoI, Amersham Biosciences, Uppsala, Sweden) and shrimp alkaline phosphatase (Amersham Biosciences) to remove excess primers and deoxynucleotide triphosphates.

Fibroblast growth factor receptor 3 mutation analysis. PCR products were analyzed for mutations using the ABI PRISM SNaPshot Multiplex Kit (Applied Biosystems, Foster City, CA), according to the protocol supplied by the manufacturer. The SNaPshot method is based on the dideoxy single-base extension of unlabeled oligonucleotide primers. For each of the nine mutations described above, a primer annealing adjacent to the potentially mutant nucleotide was developed. In mutations K652E and K652Q, the same nucleotide is changed, therefore, one primer will detect both mutations. The same holds for mutations K652M and K652T. In total, seven SNaPshot primers were developed, with a different length of poly(dT) tail attached to the 5′-end to enable their simultaneous detection. All primers were designed with a similar melting temperature and were checked for the absence of base pairing with other SNaPshot primers.

PCR conditions were optimized in order to develop a multiplex SNaPshot reaction. Two new primers had to be developed to improve detection of the S249C and Y375C mutations (see Results). In order to enhance the detection of the most frequent mutation (S249C), two different lengths of poly(dT) tails were added. The final panel of SNaPshot primers is shown in Table 1. The multiplex SNaPshot reaction was done in a volume of 10 μL, containing 2 μL of PCR product (see above), 2.5 μL Ready Reaction Mix, 1× sequencing buffer, and SNaPshot primers with concentrations as indicated in Table 1. Cycling conditions were, 25 cycles of rapid thermal ramp to 96°C, 96°C for 10 seconds, rapid thermal ramp to 58.5°C, and 58.5°C for 40 seconds. After treatment with shrimp alkaline phosphatase to remove excess dideoxynucleotide triphosphates, labeled products were separated in a 25-minute run on 36-cm-long capillaries in an automatic sequencer (ABI PRISM 3100 Genetic Analyzer, Applied Biosystems). GeneScan Analysis Software version 3.7 (Applied Biosystems) was used for data analysis.

Results

Development of an fibroblast growth factor receptor 3 mutation detection assay. Mutation analysis was based on the ABI PRISM SNaPshot Multiplex Kit. We first designed primers adjacent to the seven nucleotides that are most frequently mutated in bladder tumors. These seven primers enabled the detection of nine mutations in the FGFR3 gene. They were modified by adding different lengths of poly(dT) tails to the 5′-end in order to allow their separation based on differences in size. These primers were annealed to a combination of FGFR3 exon 7, 10, and 15 PCR products from control DNA samples. In a subsequent reaction (SNaPshot reaction), the primers were extended using a substrate mixture of all four fluorescently labeled dideoxynucleotide triphosphates, thus allowing the addition of only one nucleotide to each primer. The products were analyzed on an ABI automatic sequencer in a 25-minute run. With this procedure, all seven wild-type nucleotides were detected as shown in Fig. 1A.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1.

Detection of wild-type and mutant FGFR3 nucleotides using the FGFR3 SNaPshot mutation assay. A, analysis of control DNA. Peaks are labeled with the relevant FGFR3 codon. Please note that two different primers were used for codon 249; S, size standard. B to J, analysis of tumor DNA samples containing the following mutations: R248C, S249C, G372C, Y375C, A393E, K652E, K652M, K652Q, and K652T.

When we subsequently analyzed known mutations in bladder tumor DNA samples, it seemed that the mutations in codons 248, 372, 393, and 652 could be detected reliably. However, for codon 249, the mutant peak was often too small to be distinguished from background signals, thus leading to false-negative results. Conversely, false-positive results were frequently experienced for codon 375 because of the presence of a background peak with the same color and position as expected for the mutant nucleotide. A new SNaPshot primer for codon 375 was therefore designed that annealed to the antisense strand. When using this forward primer, the background signal disappeared. A new primer on the opposite strand was also developed for codon 249. Two different lengths of poly(dT) tails were added to the new S249C forward primer, in order to enhance the possible detection of this most frequent mutation. However, the detection of the S249C signal continued to be inefficient. We next analyzed the secondary structure of the exon 7 PCR product, and observed that the single strand was able to fold into a very stable secondary structure with a free energy of −36.11 kcal/mol. When the codon 249 mutation was present, the structure was even more stable (free energy −40.05 kcal/mol). Consequently, annealing of the S249C SNaPshot primer was inefficient. A new exon 7 reverse primer was developed to create a shorter PCR product with a lower GC content, resulting in a less stable secondary structure for both wild-type (free energy −20.31 kcal/mol) and codon 249 mutant (free energy −22.88 kcal/mol) DNA. As a result, we now observed that the S249C peaks were clearly visible next to the wild-type peaks (detection of the codon 249 mutation is depicted in Fig. 1C). Please note that the mutant peaks appear at a slightly different position than the wild-type products. This is due to the differences in size of the fluorescent labels of the dideoxynucleotide triphosphates. In addition, the assay is not quantitative because the labels also have different emission efficiencies. Based on the secondary structure, a new reverse primer was also developed for exon 10, and this improved the detection of the codon 375 mutation even further (Fig. 1E). In Fig. 1B-J, examples of the detection of all nine mutations are shown.

Sensitivity of the fibroblast growth factor receptor 3 SNaPshot assay. We next determined the minimal amount of DNA required for reliable mutation detection. To this end, tumor DNA samples harboring the four most common mutations (S249C, Y375C, R248C, and G372C) were diluted, and for each of these mutations, we observed that 1 ng of DNA was sufficient as a starting concentration for the multiplex PCR reaction (data not shown). We subsequently determined whether these mutations could also be detected in a background of normal DNA. To this end, we mixed tumor DNA heterozygous for the mutations (FGFR3 mutations are usually heterozygous; refs. 15, 17) with control DNA in different ratios (1:1, 1:4, 1:9, and 1:19). Figure 2 shows the result of the mutation detection using a 1:9 (G372C and Y375C) or 1:19 (R248C and S249C) ratio of mutant versus normal DNA. All four mutations were detected. This means that one mutant allele can still be detected against a background of up to at least 39 wild-type alleles.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 2.

DNA samples from tumors harboring the most common FGFR3 mutations were diluted at a ratio of 1:9 (G372C and Y375C) or 1:19 (R248C and S249C) with control (nonmutant) DNA. A, R248C; B, S249C; C, G372C; D, Y375C.

Fibroblast growth factor receptor 3 SNaPshot mutation detection is superior to single-stranded conformation polymorphisms. To compare the new method with analysis by SSCP, we did FGFR3 mutation analysis on primary tumor DNA samples from 92 patients diagnosed with bladder cancer. Mutations that were found by SSCP could all be confirmed using the SNaPshot method, except for one codon 249 mutation (Table 2). Four additional codon 375 mutations were observed, thereby almost doubling the detection efficiency for this mutation. This mutation was difficult to detect by SSCP because the wild-type and mutant bands migrated closely together. In summary, the SNaPshot method proved to be more sensitive than SSCP.

Fibroblast growth factor receptor 3 mutation analysis in urine. We next evaluated the sensitivity of the assay with regard to the detection of recurrent bladder tumors. To this end, we selected urine samples from patients with a recurrent tumor. Urine samples from patients with primary tumors were excluded because primary tumors are often larger than recurrences and this may result in an overestimated sensitivity of the assay. FGFR3 mutations in the tumors were detected by SSCP (14) or SNaPshot analysis.² A total of 64 urine samples were analyzed. Of these, 29 were derived from patients with a mutant tumor, and in 35 cases, the corresponding tumor was wild-type for FGFR3. Table 3 illustrates the results obtained with the SNaPshot assay on DNA from urinary cells. The results show that the sensitivity of the assay was 62% (18 of 29) and the specificity 89% (31 of 35). The positive predictive value of the test was 82% (18 of 22) and the negative predictive value was 74% (31 of 42). We conclude that FGFR3 mutations are readily observed in a considerable proportion of urinary DNA samples.

Discussion

FGFR3 mutation detection is of great clinical importance in bladder cancer. Here, we describe a simple assay to detect FGFR3 mutations in bladder tumors and urine samples. Previous mutation detection was by SSCP and sequencing, both very laborious techniques as the mutation hotspots on exons 7, 10, and 15 have to be screened separately. With the SNaPshot technique, the analysis of the three exons can be combined into one assay, thus allowing a sample to be screened for all FGFR3 mutations simultaneously. Furthermore, there is no need for additional sequencing because both techniques are identical.

The frequency of FGFR3 mutations in primary bladder tumors is ∼50%. Because the presence of an FGFR3 mutation is associated with a favorable disease course (6), a first analysis of tumors for mutations can select this 50% of patients with a low chance of progressive disease, which could reduce the number of cystoscopies. FGFR3 also has potential as a molecular marker in urine. A previous study by Rieger-Christ et al. (16) showed an overall frequency of FGFR3 mutations in urine DNA samples of 43%. In this work, the mutation status of the corresponding tumors was not analyzed, therefore it is not possible to calculate the sensitivity of FGFR3 mutation detection. We have shown that FGFR3 mutation detection on urine by the SNaPshot assay is an efficient method to detect recurrent bladder tumors (i.e., the sensitivity of the assay was 62%). That the mutations in the urine DNA samples were not observed in all cases might in part be due to the fact that FGFR3 mutant tumors are often pTaG1, and it is possible that these tumors shed fewer cells than tumors containing a wild-type FGFR3 gene, as these are more often of higher grade and stage. In three cases, the mutation in the urinary DNA sample did not match the mutation in the tumor, and in four tumors that were wild-type for FGFR3, the corresponding urine contained an FGFR3 mutation. A possible explanation for these findings is that we have observed in another study that there is a small group of patients in whom wild-type and mutant bladder tumors seem to occur simultaneously, and that some tumors also appear heterogeneous for the FGFR3 mutation.² For follow-up of patients with superficial bladder cancer (pTa, pT1), we suggest the following scheme: cystoscopy is done at 3 months after transurethral resection and subsequently once per year. FGFR3 mutation analysis on urine DNA is done every 3 months. When positive, cystoscopy follows. When negative, further screening is done by loss of heterozygosity analysis. Thus, FGFR3 mutation analysis on voided urine could replace a large percentage of cystoscopies without the risk of negatively affecting the disease course in these patients. In addition, a decrease in the number of invasive cystoscopies would considerably improve the quality of life of the patients.

In conclusion, we developed a high-throughput method for FGFR3 mutation analysis that is more sensitive than SSCP, and that can be used on both tumor and urine DNA samples. In theory, it is possible to add additional primers to the SNaPshot reaction; for instance, for the detection of other less common mutations such as G382C and S373C (7, 16, 18).