Skip to main content
Advertisement
Browse Subject Areas
?

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here.

  • Loading metrics

A Meta-Analysis of the Relationship between FGFR3 and TP53 Mutations in Bladder Cancer

  • Yann Neuzillet ,

    Contributed equally to this work with: Yann Neuzillet, Xavier Paoletti

    Affiliations Department of Urology, Hôpital Foch, Université de Versailles – Saint-Quentin-en-Yvelines, Suresnes, France, Centre de Recherche, Institut Curie, Paris, France, UMR144, CNRS, Paris, France

  • Xavier Paoletti ,

    Contributed equally to this work with: Yann Neuzillet, Xavier Paoletti

    Affiliation Department of Biostatistics, Institut Curie, INSERM U900, Paris, France

  • Slah Ouerhani,

    Affiliations Centre de Recherche, Institut Curie, Paris, France, UMR144, CNRS, Paris, France, Unité 855, INSERM, Créteil, France, Faculté des Sciences de Tunis, Tunis, Tunisia, Institut Pasteur, Tunis, Tunisia

  • Pierre Mongiat-Artus,

    Affiliations Department of Urology, Hôpital Saint-Louis, APHP, Université Paris 7, Paris, France, INSERM U944, Institut Universitaire d'Hématologie, Université Paris 7, Paris, France

  • Hany Soliman,

    Affiliations INSERM U944, Institut Universitaire d'Hématologie, Université Paris 7, Paris, France, Department of Biochemistry, Hôpital Saint-Louis, APHP, Paris, France

  • Hugues de The,

    Affiliations Department of Biochemistry, Hôpital Saint-Louis, APHP, Paris, France, UMR7151, CNRS, Université Paris 7, Paris, France

  • Mathilde Sibony,

    Affiliation Department of Pathology, Hôpital Tenon, APHP, Paris, France

  • Yves Denoux,

    Affiliation Department of Pathology, Hôpital Foch, Suresnes, France

  • Vincent Molinie,

    Affiliation Department of Pathology, Hôpital Saint-Joseph, Paris, France

  • Aurélie Herault,

    Affiliations Centre de Recherche, Institut Curie, Paris, France, UMR144, CNRS, Paris, France

  • May-Linda Lepage,

    Affiliations Centre de Recherche, Institut Curie, Paris, France, UMR144, CNRS, Paris, France

  • Pascale Maille,

    Affiliation Department of Pathology, Hôpital Henri Mondor, APHP, Créteil, France

  • Audrey Renou,

    Affiliation Department of Pathology, Hôpital Henri Mondor, APHP, Créteil, France

  • Dimitri Vordos,

    Affiliation Department of Urology, Hôpital Henri Mondor, APHP, Créteil, France

  • Claude-Clément Abbou,

    Affiliations Unité 855, INSERM, Créteil, France, Department of Urology, Hôpital Henri Mondor, APHP, Créteil, France

  • Ashraf Bakkar,

    Affiliation Sinai University, Cairo, Egypt

  • Bernard Asselain,

    Affiliation Department of Biostatistics, Institut Curie, INSERM U900, Paris, France

  • Nadia Kourda,

    Affiliation Hôpital Charles Nicole, Tunis, Tunisia

  • Amel El Gaaied,

    Affiliation Faculté des Sciences de Tunis, Tunis, Tunisia

  • Karen Leroy,

    Affiliation Department of Pathology, Hôpital Henri Mondor, APHP, Créteil, France

  • Agnès Laplanche,

    Affiliation INSERM U794/CNRS FRE2939, Institut Gustave Roussy, Villejuif, France

  • Simone Benhamou,

    Affiliation INSERM U794/CNRS FRE2939, Institut Gustave Roussy, Villejuif, France

  • Thierry Lebret,

    Affiliation Department of Urology, Hôpital Foch, Université de Versailles – Saint-Quentin-en-Yvelines, Suresnes, France

  • Yves Allory,

    Affiliations Unité 855, INSERM, Créteil, France, Department of Pathology, Hôpital Henri Mondor, APHP, Créteil, France

  •  [ ... ],
  • François Radvanyi

    francois.radvanyi@curie.fr

    Affiliations Centre de Recherche, Institut Curie, Paris, France, UMR144, CNRS, Paris, France

  • [ view all ]
  • [ view less ]

Abstract

TP53 and FGFR3 mutations are the most common mutations in bladder cancers. FGFR3 mutations are most frequent in low-grade low-stage tumours, whereas TP53 mutations are most frequent in high-grade high-stage tumours. Several studies have reported FGFR3 and TP53 mutations to be mutually exclusive events, whereas others have reported them to be independent. We carried out a meta-analysis of published findings for FGFR3 and TP53 mutations in bladder cancer (535 tumours, 6 publications) and additional unpublished data for 382 tumours. TP53 and FGFR3 mutations were not independent events for all tumours considered together (OR = 0.25 [0.18–0.37], p = 0.0001) or for pT1 tumours alone (OR = 0.47 [0.28–0.79], p = 0.0009). However, if the analysis was restricted to pTa tumours or to muscle-invasive tumours alone, FGFR3 and TP53 mutations were independent events (OR = 0.56 [0.23–1.36] (p = 0.12) and OR = 0.99 [0.37–2.7] (p = 0.35), respectively). After stratification of the tumours by stage and grade, no dependence was detected in the five tumour groups considered (pTaG1 and pTaG2 together, pTaG3, pT1G2, pT1G3, pT2-4). These differences in findings can be attributed to the putative existence of two different pathways of tumour progression in bladder cancer: the CIS pathway, in which FGFR3 mutations are rare, and the Ta pathway, in which FGFR3 mutations are frequent. TP53 mutations occur at the earliest stage of the CIS pathway, whereas they occur would much later in the Ta pathway, at the T1G3 or muscle-invasive stage.

Introduction

Bladder cancer is one of the most common cancers worldwide. It is the fourth most prevalent cancer in men and the 11th most prevalent cancer in women in the United States [1]. More than 90% of bladder cancers are carcinomas, which may present at different stages. Ta tumours are papillary, generally low-grade tumours, which do not invade beyond the basement membrane. Carcinoma in situ (CIS) is a flat tumour that does not invade the basement membrane but is always of high grade. T1 tumours invade the subepithelial connective tissue but do not infiltrate the underlying muscularis propria. T2, T3 and T4 tumours invade the muscularis propria, perivesical tissue and adjacent organs, respectively [2].

There is clinical and molecular evidence for the existence of two pathways of bladder tumour progression: the Ta and CIS pathways [3][6]. Ta tumours often recur after surgical resection, but they progress only rarely (5–10% of cases) and unpredictably to high-grade T1 tumours and then to muscle-invasive tumours. By contrast, CIS often progress (in about 50% of cases) to T1 and then to muscle-invasive tumours. About 80% of muscle-invasive tumours are thought to arise through the CIS pathway [5], [7]. Activating mutations of FGFR3, which encodes a growth factor receptor of the fibroblast growth factor receptor family, have been shown to be associated mostly with the Ta pathway of tumour progression, as such mutations have been reported in 65% of pTa tumours, less frequently in pT1 (33%) and pT2-4 tumours (22%) and not at all in CIS [4], [8], [9] [Table S1]. By contrast, TP53 mutations are infrequent in Ta tumours (19% of cases) and frequent both in carcinoma in situ (52% of cases) and in muscle-invasive tumours (44% of cases) [3] [Table S2]. Conflicting results have been published concerning the relationship between TP53 and FGFR3 mutations. TP53 and FGFR3 mutations were initially thought to be essentially mutually exclusive, with FGFR3 mutations specific to the Ta pathway and TP53 mutations specific to the CIS pathway [10], [11]. However, Hernandez et al., in a study of a large series of pT1G3 tumours (n = 119), which are particularly difficult to manage clinically, reported FGFR3 and TP53 mutations to be independently distributed [12]. This was interpreted as indicating that pT1 tumours constitute a particular group of bladder tumours, not all of which fit into the two known pathways of bladder tumour progression [6]. Several other studies have also investigated both FGFR3 and TP53 mutations and have reported the presence of both types of mutation in some tumours. The number of double mutants was small in each of these reports (5 in Zieger et al. [13]; 2 in Lindgren et al. [14], 5 in Lamy et al. [15]; 9 in Ouerhani et al. [16]). In all these studies, P53 mutations and FGFR3 mutations were found to be inversely associated with the grade and the stage of the tumour. Stage and grade can therefore act as potential confusion factors that may create spurious associations between the risks of each of mutations. Only large sample sizes with tumours of each grade and stage would allow for properly adjusting association analysis on these two factors.

We made use of all the previously published data (535 tumours) and unpublished data from the Henri Mondor, Foch, IGR, and Saint-Louis hospitals (382 tumours) for analyses of both FGFR3 and TP53 mutations, in a meta-analysis investigating the relationship between these two mutations. We investigated whether FGFR3 and TP53 mutations were dependent (TP53 occurring more rarely in FGFR3-mutated tumours) or independent events (TP53 occurring at similar frequencies in tumours with and without FGFR3 mutations) in this large series of tumours. The frequency of FGFR3 and TP53 mutations depends strongly on tumour stage and grade. We therefore also performed the analysis on subgroups of tumours defined on the basis of stage, grade or both these parameters.

Results

Available data

We retained only tumours for which stage was documented from the various studies (published and unpublished) reporting mutations of both FGFR3 and TP53 in bladder cancer (Table 1). We excluded pure CIS and papilloma, as there were only two cases of CIS and one case of papilloma in total, in all the studies considered. We thus selected 917 tumours in total for study, and grade was documented for 827 of these tumours. Grades were provided in the study by Lamy et al., but it was impossible to retrieve information about both stage and grade for a given tumour [15]. We therefore excluded the data from the study by Lamy et al. from the combined investigation of stage and grade. The stages and grades of tumours for each study are summarised in Tables S1 and S2 (published studies) and Table S3 (unpublished studies). In total, there were 350 pTa, 358 pT1, 209 pT2-4 and 88 G1, 249 G2 and 490 G3 tumours. For the combined analysis of stage and grade, we considered the following five categories of tumours: pTaG1 plus pTaG2 (as a single category), pTaG3, pT1G2, pT1G3 and pT2-4 tumours. We classified pTaG1 and pTaG2 tumours together, and pT2, pT3 and pT4 tumours together as, in each of these groups, the tumours concerned are considered to constitute the same clinical entity, regardless of grade.

thumbnail
Table 1. Summary of the materials and methods and patients sections of the various published and unpublished studies.

https://doi.org/10.1371/journal.pone.0048993.t001

Distribution of FGFR3 and TP53 mutations by stage and by grade

FGFR3 mutation status was available for 916 of the 917 tumours with a documented stage and TP53 mutation status was available for 898 of the 917 tumours. This meta-analysis, like many previous studies, showed an inverse relationship between FGFR3 and TP53 mutations for both stage and grade (Figure 1). The frequency of FGFR3 mutations decreased with increasing stage and grade: 65% in pTa, 30.2% in pT1, 11.5% in pT2-4 and 69.8% in G1, a very similar rate in G2 (68%) and 18.6% in G3. These trends, for both stage and grade, were highly significant (p<0.0001 and p<0.0001, respectively). By contrast, the frequency of TP53 mutations increased with increasing stage and grade: 6.55% in pTa, 40.6% in pT1, 50.7% in pT2-4, and 3.8% in G1, 12.05% in G2 and 46.3% in G3. These trends, for both stage and grade, were highly significant (p<0.0001 and p<0.0001 respectively), suggesting that stage and grade may be confounding factors.

thumbnail
Figure 1. FGFR3 and TP53 mutation frequencies by stage (pT) or grade (G).

Proportion of wild-type tumours and of tumours with FGFR3 (upper row) and TP53 (lower row) mutations as a function of pathological stage (left column) and grade (right column). The number of cases in each subgroup is indicated on the bars of the graph.

https://doi.org/10.1371/journal.pone.0048993.g001

Association between FGFR3 and TP53 mutations, adjusting for stage or for grade

We then studied mutation status for both FGFR3 and TP53, as a function of stage (Figure 2). For pTa tumours, the most common of the four possible groups (wild-type FGFR3 plus wild-type TP53, wild-type FGFR3 plus mutated TP53, mutated FGFR3 plus wild-type TP53, mutated FGFR3 plus mutated TP53) was tumours with mutated FGFR3 and wild-type TP53 (208/336; 61.9% of cases), followed by tumours wild-type for both FGFR3 and TP53 (106/336; 31.5% of cases). A small number of tumours had TP53 mutations and were either wild-type for FGFR3 (11/336; 3.3%) or mutated for FGFR3 (11/336; 3.3%). For pT1 tumours, the two most common groups were tumours wild-type for both FGFR3 and TP53 (134/355; 37.7% of cases) or wild-type for FGFR3 and mutated for TP53 (115/355; 32.4% of cases). For invasive tumours (pT2-4), the two most common groups were also tumours wild-type for both FGFR3 and TP53 (88/207; 42.5% of cases) or wild-type for FGFR3 and mutated for TP53 (95/207; 45.9% of cases).

thumbnail
Figure 2. Combined FGFR3 and TP53 mutation frequencies by stage (pT).

Proportion of tumours with both FGFR3 and TP53 mutations (orange), with mutated FGFR3 and wild-type TP53 (grey), with wild-type FGFR3 and mutated TP53 (purple), or with wild-type FGFR3 and wild-type TP53 (blue), as a function of pathological stage. The number of cases in each subgroup is indicated in the corresponding rectangle.

https://doi.org/10.1371/journal.pone.0048993.g002

We then investigated whether FGFR3 and TP53 mutations were independent events. We defined four groups (wild-type FGFR3 plus wild-type TP53, wild-type FGFR3 plus mutated TP53, mutated FGFR3 plus wild-type TP53, mutated FGFR3 plus mutated TP53). We carried out a Mantel-Haenszel test, stratified for stage, to determine whether the proportion of tumours with TP53 mutations differed significantly between tumours with wild-type FGFR3 and tumours with FGFR3 mutations, assuming a common association for each stage. For all tumours considered together, we found a strong association between FGFR3 and TP53 mutations (OR = 0.49 [0.33, 0.72], p = 0.001), such that the odds of a given FGFR3-mutated tumour having a TP53 mutation were half those for FGFR3 wild-type tumours. The Briant and Day interaction test suggested that the association did not differ significantly across strata (p = 0.72). In subgroup analysis, FGFR3 and TP53 mutations were not significantly associated in pTa tumours (p = 0.20) or in pT2-4 tumours (p = 0.345), but these two types of mutation were strongly associated, even after Bonferroni correction (significance level for the three tests  =  0.017) in pT1 tumours (ORa = 0.52 [0.30, 0.88], p = 0.0009) (Table 2).

thumbnail
Table 2. Association between FGFR3 and TP53 mutations according to stage and grade.

https://doi.org/10.1371/journal.pone.0048993.t002

It was possible to carry out a similar analysis for grade, as opposed to stage, for seven studies (Mongiat-Artus UP, BladderCIT UP, Bakkar et al., 2003, Hernandez et al., 2005, Lamy et al., 2006, Lindgren et al., 2006, Ouerhani et al., 2009) including 638 patients for whom mutation status data were available for both TP53 and FGFR3. Some heterogeneity in the association was detected between grades (p = 0.05 in the Briant and Day test). We found an association between FGFR3 and TP53 mutations only in G3 tumours (OR = 0.57 [0.35–0.93], p = 0.0245), but this finding was of borderline difference after adjusting for multiple testing.

Association between FGFR3 and TP53 mutations, adjusting for combined categories of both stage and grade

We then assigned the tumours to groups on the basis of both stage and grade (Figure 3). We defined five categories: pTaG1 and pTaG2 (a single category), pTaG3, pT1G2, pT1G3, pT2-4 (all stages, a single category). This division is based on that used in clinical practice. There was a strong association between TP53 mutations and category, as the frequency of TP53 mutations was 4.55% in pTaG1-2, 14.3% in pTaG3, 18.5% in pT1G2, 46.15% in pT1G3 and 50.25% in pT2-4 tumours, the largest difference thus being that between pT1G2 and pT1G3 tumours. An association between FGFR3 mutation status and category was also observed, although the trend was less clear-cut, because of the strong influence of both grade and stage on FGFR3 mutation rate: the frequency of FGFR3 mutation was very similar in pTaG1-2 tumours (70.2%) and pT1G2 tumours (64.6%), but significantly lower in pTaG3 tumours (40.5%) (χ2 test; p = 0.0001). The frequency of FGFR3 mutation was even lower in pT1G3 tumours (20.8%) and pT2-4 tumours (11.8%).

thumbnail
Figure 3. Combined FGFR3 and TP53 mutation frequencies according to the stage/grade group.

Proportion of tumours with both FGFR3 and TP53 mutations (orange), with FGFR3 mutations and wild-type TP53 wild-type (grey), with wild-type FGFR3 and mutated TP53 (purple), or with wild-type FGFR3 and wild-type TP53 (blue), as a function of stage/grade. The number of cases in each subgroup is indicated in the corresponding rectangle.

https://doi.org/10.1371/journal.pone.0048993.g003

We then studied the association between FGFR3 and TP53 mutations in the five tumour stage/grade categories (table 3). As before, we defined four groups within each category (wild-type FGFR3 plus wild-type TP53, wild-type FGFR3 plus mutated TP53, mutated FGFR3 plus wild-type TP53, mutated FGFR3 plus mutated TP53). Overall, no significant association was detected after adjusting for stage and group (OR = 0.69; 95% CI = 0.44 to 1.08, Mantel-Haenszel test = 0.075) and we detected no statistically significant heterogeneity across strata (p = 0.41 for heterogeneity). FGFR3 and TP53 mutations were considered to be independent events in all categories (Table 3), as the proportion of tumours with mutated TP53 did not differ significantly between the tumours with wild-type and mutated FGFR3 (Table 4).

thumbnail
Table 3. Association between FGFR3 and TP53 mutations according to the stage/grade group.

https://doi.org/10.1371/journal.pone.0048993.t003

thumbnail
Table 4. TP53 mutation rates in FGFR3-wild-type and FGFR3-mutated tumours, according to a combination of stage and grade.

https://doi.org/10.1371/journal.pone.0048993.t004

We assessed the robustness of the results, by carrying out the same analysis while accounting for the potential study effect. Similar results were obtained despite the continuity correction required to correct for too small sample sizes in some defined by the combination of stage/grade.

Discussion

Due to their inverse distributions as a function of stage and grade and the small number of double-mutated tumours (FGFR3 mutated, TP53 mutated) observed in small series, FGFR3 and TP53 mutations had been reported to be mutually exclusive events, with FGFR3 mutation strongly associated with the Ta pathway and TP53 mutation strongly associated with the CIS pathway [10], [11]. The group of Real showed for the first time, in a study of a large series of tumours, that FGFR3 and TP53 mutations were independent events in pT1G3 tumours (n = 119) [12]. Our meta-analysis of 917 tumours (including all published data (535 tumours) plus an additional series of 382 tumours of all stages and grades) confirms and extends the findings of Hernandez et al., by showing that FGFR3 mutations and TP53 mutations are independent events not only in pT1G3 tumours (confirmed here for 260 pT1G3 tumours) but also in pTa, pT1G2 and muscle-invasive tumours (pT≥2). By contrast, FGFR3 mutations and TP53 mutations were not independent events if we considered all tumours together, without accounting for stage and grade (p = 0.0001), or all pT1 tumours together (pT1G2 and pT1G3 tumours) (p = 0.0009).

The two known pathways of tumour progression and the different frequencies of FGFR3 and TP53 mutations in the different groups of tumours defined on the basis of grade and stage may account for these observations. At an early stage (pTaG1 and pTaG2), the frequency of FGFR3 mutation was high in the Ta low-grade pathway, whereas that of TP53 mutation was very low. If these tumours progress to muscle-invasive tumours, they will carry TP53 mutations at a frequency similar to that in tumours of the CIS pathway. Indeed, in our meta-analysis, muscle-invasive TP53 mutations were found in 42% of tumours with FGFR3 mutation (10/24). This proportion is lower than the 52% (95/183) of muscle-invasive tumours with wild-type FGFR3 observed, but not significantly so. For tumours of the Ta pathway, the frequency of TP53 mutations seems to increase gradually with stage, from pTa (11/219; 5% of tumours with FGFR3 mutations also have TP53 mutations) to pT1 (29/106; TP53 mutations in 27% of tumours) and pT2-4 tumours (10/24; TP53 mutations in 42% of tumours). Grade also seems to be important, as, for FGFR3-mutated tumours, the frequency of TP53 mutation was 17% in pT1G2 tumours and 37% in pT1G3 tumours. The frequency of FGFR3 mutation was found to be very similar in pT1G2 and pTa G1G2 tumours, suggesting that most pT1G2 tumours are derived from the Ta pathway rather than the CIS pathway. Consistent with this hypothesis, the frequency of TP53 mutation in pT1G2 tumours, although higher than that in pTa tumours (18% versus 7%), was found to be much lower than that reported in cases of dysplasia or CIS tumours (65–72%) [3], [17]. Thus, the lack of independence of FGFR3 and TP53 mutations when all pT1 tumours are considered together may reflect the two different pathways giving rise to tumours of this stage, with different frequencies of TP53 mutation.

There are also more complex explanations for the findings of this meta-analysis. Mutation data for a series of tumours provide only a snapshot of the situation, from which the progression of individual tumours is inferred. Definitive validation of the model proposed will require extensive studies of TP53 and FGFR3 mutations in multiple tumours from the individual patients, including both non-muscle invasive and muscle-invasive tumours.

It should be noted that a high proportion of bladder tumors present a wild-type phenotype (no FGFR3 or TP53 mutations): 32% of Ta tumors, 40% of T1 tumors and 43% of T2-4 tumors). These tumours probably have mutations in genes other than FGFR3 or TP53, but with a similar effect (activation of the FGFR3 signaling pathway in Ta tumors of low grade and mutations of genes causing genetic instability in high-grade or high-stage tumors). RAS mutations, which are observed in about 10% of bladder cancers and are never found with FGFR3 mutations, are thought to affect the FGFR3 signaling pathway [5], [18]. MDM2 amplifications, which occur in about 6% of T1 and muscle-invasive bladder cancers [19], should lead to TP53 inactivation. Other mutations recently identified [20], or yet to be identified may also account for the absence of FGFR3 and TP53 mutations.

We recently showed that most (80%) muscle-invasive tumours with FGFR3 mutations harbour homozygous CDKN2A deletions, resulting in the loss of both P16INk4A and P14ARF [21]. The loss of P14ARF should lead to MDM2 activation, thereby decreasing TP53 levels. However, in tumours with FGFR3 mutations, homozygous CDKN2A deletion and TP53 mutation may occur together (data not shown), suggesting that these two events are not alternative mechanisms of inactivation for the same pathway. Indeed, the loss of CDKN2A also leads to the deletion of p16INk4A, which encodes a cyclin kinase inhibitor controlling RB protein activity through CDK4 in G1/S. Furthermore, p14ARF may have several TP53-independent activities [22] and TP53 mutations may have gain-of-function effects [23]. Thus, the various events (CDKN2A loss and TP53 mutation) may contribute in an additional manner to the transformed phenotype.

Meta-analyses are often performed on clinical data, but more rarely on biological data. Meta-analysis is much more powerful than individual analysis but has several drawbacks: 1) Different studies use different methodologies. Indeed, in the studies analysed here, different methods were used to assess the frequency of both FGFR3 (snapshot or sequencing) and TP53 (direct sequencing or FASAY) mutations. Furthermore, different exons were explored in different studies: exons 7, 10, 15 or 7, 10, 13, 15 for FGFR3 and exons 5 to 8, 4 to 9, 4 to 11 or 2 to 11 for TP53. Differences due to the choice of exons studied should be negligible; mutations of exon 15 of FGFR3 account for only 3% of all identified mutations when exons 7, 10, and 15 are considered, and mutations in exons 2, 3, 4, 9, 10 and 11 of TP53 mutations account for only 6.3% of all identified mutations when exons 2 to 11 of this gene are considered. FASAY and direct sequencing are very different in nature (FASAY being a functional assay and more sensitive), but they give very similar results [24]. Most of the tumours explored by the FASAY method in our analysis were pTaG1-G3 or pT1G2 tumours. A comparison of the FASAY data with direct sequencing data for the same tumours showed the frequencies of TP53 mutation obtained with these two techniques to be very similar. 2) This meta-analysis may also have been biased by the lack of review of stage and grade determination. The stages and grades assigned to tumours may therefore differ between the studies included in the meta-analysis. This could be problematic for pT1 tumours, because some pTa tumours may have been overstaged and classified as pT1 tumours [25].

This meta-analysis provides a good example of the effect of confounding factors that are not taken into account. Such factors may not only modify the magnitude of any association, but also create spurious associations. Larger sample sizes and more detailed data are therefore required for a valid statistical analysis. In such conditions, meta-analysis is an effective approach, making it possible to draw reliable conclusions, because it builds on the strengths of several studies that may not be sufficiently powerful individually to generate conclusive results. However, meta-analyses can only take into account the confounding factors that were actually measured. Other unknown variables may affect the relationship between the pathway and disease severity and remain undetected. All the publications retained for this analysis had a high quality of reporting, making it possible to extract the relevant information.

Despite the known weaknesses of meta-analysis, the large sample and the homogeneity of the results after adjustment add weight to our conclusions: we found that FGFR3 and TP53 mutations occurred independently when stage and grade were taken into account, and that the frequency of TP53 mutation was high in pT1G3 and pT2-4 tumours, regardless of the presence or absence of FGFR3 mutations in these tumours. Thus, TP53 mutations can occur in Ta pathway tumours, when these tumours progress. However, the time frame of TP53 mutation differs considerably between the two pathways: TP53 mutations occur before basement membrane invasion in the carcinoma in situ pathway, whereas they probably occur after or during basement membrane invasion in the Ta pathway. Our findings also indicate that care is required in the analysis of bladder tumours, as both pathways of tumour progression for bladder cancers must be taken into account when interpreting data.

Materials and Methods

Studies

Studies of adult patients with bladder carcinoma were eligible for inclusion in the analysis if they reported FGFR3 and TP53 mutations as a function of stage. Paediatric patients (13 urothelial carcinoma cases) were excluded because the process of carcinogenesis in these patients is probably different from that in adult cases [26].

We searched databases (MEDLINE, PubMed, Embase) with cross-referenced search terms customised for each of the search engines used, to identify relevant articles published in English and dealing with FGFR3 and TP53 mutations in bladder carcinomas. We searched the reference lists of review articles manually to identify additional publications. In accordance with PRISMA guidelines [27], the eight identified studies including a total of 963 patients are presented in Figure 4. All but one (Hernandez et al., 2005) of these studies were retrospective single- or multicenter studies investigating the association between gene mutations and pathways in urothelial cell carcinoma. One study focused on the pT1G3 stage (Hernandez et al., 2005), whereas the others included patients at early stages (pTa and pT1) (Lindgren et al. 2006), late stages (pT1-4) (Ouerhani et al., 2009) or at any stage. Three studies concerned newly diagnosed patients (Bakkar 2003, Hernandez et al., 2005; Lamy 2006). Two of these studies have not yet been published and are denoted below by the name of the main investigator or the program supporting the study plus “UP” (for unpublished; thus, Mongiat-Artus UP and BladderCIT UP).

thumbnail
Figure 4. PRISMA 2009 flow diagram of the study selection process.

https://doi.org/10.1371/journal.pone.0048993.g004

Information about treatment (irradiation or chemotherapy) before biopsy was unavailable for a large proportion of the patients. We therefore included all patients, regardless of their prior treatment. The main characteristics of these eight studies are summarised in Table 1. We refer to the publications for all details in the cases of published data. In the Bladder CIT UP study, the pTa and pT1 tumours were from incident cases and the pT2-4 tumours were from both newly diagnosed and progressing cases (patients with a history of previous non-muscle-invasive tumours). FGFR3 mutations were assessed by the SNaPshot technique in the BladderCIT UP study [28] and by allele-specific PCR [29] in the Mongiat-Artus UP study. TP53 mutations were assessed by direct sequencing on both strands, followed by confirmation of the identified mutations in an independent PCR in the BladderCIT UP study and by the FASAY method [30], [31] followed by confirmation of the identified mutations by an independent PCR in the Mongiat-Artus UP study.

We did not include the study by van Rhijn et al. (2004) [32] because TP53 alterations were assessed by immunohistochemistry in this study and it has been shown that there is only 57 to 71% similarity between the alterations detected by mutation assessment and those detected by immunohistochemistry [Table S3].

The following variables were collected: TP53 and FGFR3 mutations, stage and grade of the disease and the type of mutation. Data for individual patients were available for four studies (Mongiat-Artus UP, BladderCIT UP, Lindgren et al., 2006, Ouerhani et al., 2009) [Table S4 and S5]. For the other reports, data were extracted directly from the publications by cross-referencing tables. For three studies (Bakkar et al., 2003; Lindgren et al., 2008; Ouerhani et al., 2009), the authors were contacted and provided additional information.

Biological and pathological data

The frequencies of FGFR3 and TP53 mutations by grade and tumour stage were recorded in all studies. Data were extracted from the publications or from the raw data, when available [Table S4 and S5].

Grading was assessed with the WHO 1973 classification in all but one of the studies (Zieger et al., 2005), the remaining study using the Bergkvist classification. Grades 1 and 2 of this classification were classified as grades 1 and 2, respectively, of the WHO 1973 grading. Grades 3 and 4 (one tumour) of the Bergkvist classification were considered to correspond to grade 3 of the WHO 1973 grading. In five studies, a central review by a senior pathologist was reported in the publication or in the analysis report for unpublished data. No such information was available for the other three studies.

In cases in which multiple tumour samples from individual patients were studied (BladderCIT UP, Zieger et al., 2005), the first tumour diagnosed was retained for the analysis. In the Zieger series, only two tumours were reported to have different mutations in the first and second sample: one tumour had FGFR3 mutations in the second biopsy sample but not the first, whereas the other had TP53 mutations only in the last sample. A sensitivity analysis, based on the use of the last sample rather than the first one, was carried out and gave very similar results.

Statistical analysis

Tumours were classified into three different groups for stage (T): Ta, T1 and T2-4. CIS and papilloma were excluded because too few cases were reported for reliable analysis. We also established five groups for exploration of the combination of stage and grade (G). These “TG” groups — TaG1-2, TaG3, T1G2, T1G3 and T2-4 — are clinically relevant. Most TaG1 and TaG2 tumours are treated by simple resection. T2-4 tumours are treated by cystectomy. T1G3 tumours are the most aggressive of the non-muscle-invasive tumours and TaG3 tumours behave very differently from TaG1 and TaG2 tumours [33], [34].

We analyzed the relationships of FGFR3 or TP53 mutation frequency with stage and grade. We then focused on the association of mutations of FGFR3 and TP53.

We initially analysed the association of mutation rates for these two genes stratified on stage and grade separately.

Overall associations were tested in adjusted Mantel-Haenszel tests, after checking for interactions of associations by clinical T stage or by grade. Interactions were explored using Cochran homogeneity tests. In cases of interaction, if association was estimated to be in opposite direction, subgroup analysis by stratum was performed. Fisher's exact tests were used when the sample size per stratum was too small.

The magnitude of the association is expressed as an adjusted odds ratio (OR), comparing the odds of FGFR3 mutation in the tumours with wild-type and mutated TP53. Adjusted ORs were estimated from the contingency table.

A significance threshold of 5% was used for all global tests. Subgroup analyses (defined by stage, grade or a combination of both) were adjusted for multiple testing, by the Bonferroni method, assuming the tests to be independent.

Supporting Information

Table S1.

Overview of FGFR3 mutations studies in bladder carcinoma.

https://doi.org/10.1371/journal.pone.0048993.s001

(DOC)

Table S2.

Overview of TP53 mutations studies in bladder carcinoma.

https://doi.org/10.1371/journal.pone.0048993.s002

(DOC)

Table S3.

Overview of FGFR3 and TP53 mutations in bladder carcinoma in the two unpublished studies.

https://doi.org/10.1371/journal.pone.0048993.s003

(DOC)

Table S4.

Available individual data from unpublished, Bakkar, Lindgren, Ouerhani, and Zieger studies.

https://doi.org/10.1371/journal.pone.0048993.s004

(DOC)

Table S5.

Joint distribution of FGFR3 and P53 mutations frequencies by stage (T) and grade (G) group.

https://doi.org/10.1371/journal.pone.0048993.s005

(DOC)

Acknowledgments

We thank Gaëlle Pierron for assistance with TP53 mutation analysis. The “bladderCIT” unpublished work is part of the Cartes d'Identité des Tumeurs ® (CIT) national program. We thank Pierre Hainaut for his advice.

Author Contributions

Conceived and designed the experiments: YN XP SO YA FR. Performed the experiments: HS MS YD VM AH MLL PM AR DV AB NK. Analyzed the data: PMA HdT CCA BA AEG KL AL SB TL. Contributed reagents/materials/analysis tools: XP FR. Wrote the paper: YN XP FR.

References

  1. 1. Jemal A, Siegel R, Xu J, Ward E (2010) Cancer statistics, 2010. CA Cancer J Clin 60: 277–300.
  2. 2. Edge SB, Byrd DR, Compton CC, Fritz AG, Greene FL, et al.. (2010) AJCC Cancer Staging Manual. Springer-Verlag: New York, USA.
  3. 3. Spruck CH 3rd, Ohneseit PF, Gonzalez-Zulueta M, Esrig D, Miyao N, et al. (1994) Two molecular pathways to transitional cell carcinoma of the bladder. Cancer Res 54: 784–788.
  4. 4. Billerey C, Chopin D, Aubriot-Lorton MH, Ricol D, Gil Diez de Medina S, et al. (2001) Frequent FGFR3 mutations in papillary non-invasive bladder (pTa) tumors. Am J Pathol 158: 1955–1959.
  5. 5. Wu XR (2005) Urothelial tumorigenesis: a tale of divergent pathways. Nat Rev Cancer 5: 713–725.
  6. 6. Knowles MA (2007) Role of FGFR3 in urothelial cell carcinoma: biomarker and potential therapeutic target. World J Urol 25: 581–593.
  7. 7. Vallot C, Stransky N, Bernard-Pierrot I, Hérault A, Zucman-Rossi J, et al. (2011) A novel epigenetic phenotype associated with the most aggressive pathway of bladder tumor progression. J Natl Cancer Inst 103: 47–60.
  8. 8. Cappellen D, De Oliveira C, Ricol D, de Medina S, Bourdin J, et al. (1999) Frequent activating mutations of FGFR3 in human bladder and cervix carcinomas. Nat Genet 23: 18–20.
  9. 9. Kimura T, Suzuki H, Ohashi T, Asano K, Kiyota H, et al. (2001) The incidence of thanatophoric dysplasia mutations in FGFR3 gene is higher in low-grade or superficial bladder carcinomas. Cancer 92: 2555–2561.
  10. 10. Bakkar AA, Wallerand H, Radvanyi F, Lahaye JB, Pissard S, et al. (2003) FGFR3 and TP53 gene mutations define two distinct pathways in urothelial cell carcinoma of the bladder. Cancer Res 63: 8108–8112.
  11. 11. van Rhijn BW, van Tilborg AA, Lurkin I, Bonaventure J, de Vries A, et al. (2002) Novel fibroblast growth factor receptor 3 (FGFR3) mutations in bladder cancer previously identified in non-lethal skeletal disorders. Eur J Hum Genet 10: 819–824.
  12. 12. Hernández S, López-Knowles E, Lloreta J, Kogevinas M, Jaramillo R, et al. (2005) FGFR3 and Tp53 mutations in T1G3 transitional bladder carcinomas: independent distribution and lack of association with prognosis. Clin Cancer Res 11: 5444–5450.
  13. 13. Zieger K, Dyrskjøt L, Wiuf C, Jensen JL, Andersen CL, et al. (2005) Role of activating fibroblast growth factor receptor 3 mutations in the development of bladder tumors. Clin Cancer Res 11: 7709–7719.
  14. 14. Lindgren D, Liedberg F, Andersson A, Chebil G, Gudjonsson S, et al. (2006) Molecular characterization of early-stage bladder carcinomas by expression profiles, FGFR3 mutation status, and loss of 9q. Oncogene 25: 2685–2696.
  15. 15. Lamy A, Gobet F, Laurent M, Blanchard F, Varin C, et al. (2006) Molecular profiling of bladder tumors based on the detection of FGFR3 and TP53 mutations. J Urol 176: 2686–2689.
  16. 16. Ouerhani S, Rouissi K, Kourda N, Marrakchi R, Bougatef K, et al. (2009) Combined analysis of smoking, TP53 and FGFR3 mutations in Tunisian patients with invasive and superficial high grade bladder tumors. Cancer Invest 27: 998–1007.
  17. 17. Hartmann A, Schlake G, Zaak D, Hungerhuber E, Hofstetter A, et al. (2002) Occurrence of chromosome 9 and p53 alterations in multifocal dysplasia and carcinoma in situ of human urinary bladder. Cancer Res 62: 809–818.
  18. 18. Jebar AH, Hurst CD, Tomlinson DC, Johnston C, Taylor CF, et al. (2005) FGFR3 and Ras gene mutations are mutually exclusive genetic events in urothelial cell carcinoma. Oncogene 24: 5218–5225.
  19. 19. Simon R, Struckmann K, Schraml P, Wagner U, Forster T, et al. (2002) Amplification pattern of 12q13–q15 genes (MDM2, CDK4, GLI) in urinary bladder cancer. Oncogene 21: 2476–2483.
  20. 20. Gui Y, Guo G, Huang Y, Hu X, Tang A, et al. (2011) Frequent mutations of chromatin remodeling genes in transitional cell carcinoma of the bladder. Nat Genet 43: 875–878.
  21. 21. Rebouissou S, Hérault A, Letouzé E, Neuzillet Y, Laplanche A, et al. (2012) CDKN2A homozygous deletion is associated with muscle invasion in FGFR3-mutated urothelial bladder carcinoma. J Pathol 227: 315–324.
  22. 22. Eymin B, Leduc C, Coll JL, Brambilla E, Gazzeri S (2003) p14ARF induces G2 arrest and apoptosis independently of p53 leading to regression of tumours established in nude mice. Oncogene 22: 1822–1835.
  23. 23. Olive KP, Tuveson DA, Ruhe ZC, Yin B, Willis NA, et al. (2004) Mutant p53 gain of function in two mouse models of Li-Fraumeni syndrome. Cell 119: 847–60.
  24. 24. Camplejohn RS, Sodha N, Gilchrist R, Lomax ME, Duddy PM, et al. (2000) The value of rapid functional assays of germline p53 status in LFS and LFL families. Br J Cancer 82: 1145–1148.
  25. 25. Platz CE, Cohen MB, Jones MP, Olson DB, Lynch CF (1996) Is microstaging of early invasive cancer of the urinary bladder possible or useful? Mod Pathol 9: 1035–1039.
  26. 26. Giedl J, Wild PJ, Stoehr R, Junker K, Boehm S, et al. (2006) Urothelial neoplasms in individuals younger than 20 years show very few genetic alterations and have a favourable clinical outcome. Verh Dtsch Ges Pathol 90: 253–263.
  27. 27. Liberati A, Altman DG, Tetzlaff J, Mulrow C, Gøtzsche PC, et al. (2009) The PRISMA statement for reporting systematic reviews and meta-analyses of studies that evaluate healthcare interventions: explanation and elaboration. J Clin Epidemiol 62: e1–34.
  28. 28. van Oers JM, Lurkin I, van Exsel AJ, Nijsen Y, van Rhijn BW, et al. (2005) 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 11: 7743–7748.
  29. 29. Bakkar AA, Quach V, Le Borgne A, Toublanc M, Henin D, et al. (2005) Sensitive allele-specific PCR assay able to detect FGFR3 mutations in tumors and urine from patients with urothelial cell carcinoma of the bladder. Clin Chem 51: 1555–15557.
  30. 30. Ishioka C, Frebourg T, Yan YX, Vidal M, Friend SH, et al. (1993) Screening patients for heterozygous p53 mutations using a functional assay in yeast. Nat Gene 5: 124–129.
  31. 31. Flaman JM, Frebourg T, Moreau V, Charbonnier F, Martin C, et al. (1995) A simple p53 functional assay for screening cell lines, blood, and tumors. Proc Natl Acad Sci USA 92: 3963–3967.
  32. 32. van Rhijn BW, van der Kwast TH, Vis AN, Kirkels WJ, Boevé ER, et al. (2004) FGFR3 and P53 characterize alternative genetic pathways in the pathogenesis of urothelial cell carcinoma. Cancer Res 64: 1911–1914.
  33. 33. Millán-Rodríguez F, Chéchile-Toniolo G, Salvador-Bayarri J, Palou J, Vicente-Rodríguez J (2000) Multivariate analysis of the prognostic factors of primary superficial bladder cancer. J Urol 163: 73–78.
  34. 34. Reuter VE (1999) Bladder. Risk and prognostic factors–a pathologist's perspective. Urol Clin North Am 26: 481–492.