Molecular profiling in bladder cancer
The Cancer Genome Atlas Project (TCGA) began as a three-year pilot project co-funded by the National Cancer Institute and National Human Genome Research Institute to characterize glioblastoma multiforme, lung and ovarian cancer.
The early success led to expanded funding to comprehensively characterize the genome of 25 common cancers and eight additional rare tumors. Genitourinary malignancies are prominent with completed or ongoing projects in clear cell, papillary and chromophobe renal cell carcinoma, muscle invasive bladder cancer, prostate adenocarcinoma and germ cell testicular cancer. Highly annotated fresh frozen tumor specimens that meet strict clinical and pathologic entry criteria are submitted to a central biospecimen-processing center where additional histology and molecular quality control are performed.
DNA, RNA and protein are then extracted and distributed to genome characterization centers. The molecular findings are then integrated to produce a comprehensive characterization of the genome unique to each tumor type resulting in a ‘human genome project’ for each cancer (Figure 1). After publication of the first “marker paper” for a given cancer, the data are stored in an open access repository creating a rich resource for the research community (https://tcga-data.nci.nih.gov/tcga and https://cghub.ucsc.edu/).
The unique aspects of the TCGA projects are the comprehensive description and integrated analysis of genomic data for each individual tumor. The TCGA project in bladder cancer reported the integrated genomic analysis of the first 131 patients in 20141.
Significantly mutated genes
We analyzed chemotherapy-naive, muscle-invasive urothelial cancers for somatic mutations, DNA copy number variants (CNVs), mRNA and microRNA expression, protein and phosphorylated protein expression (RPPA), DNA methylation, integrated with comprehensive clinical and pathologic data. In addition viral integration and translocations were characterized. This muscle invasive cohort has one of the highest somatic mutation rates with a mean and median of 7.7 and 5.5 per megabase (Mb), similar to lung adenocarcinoma and squamous cell carcinoma and melanoma.
We reported 32 significantly mutated genes (SMGs) involved in multiple pathways include cell cycle regulation (93% of tumors), chromatin remodeling (76%), DNA damage response, transcription factors, and RAS/RTK/PI3K (72%) signaling pathways. Four SMGs are involved in epigenetic regulation: ARID1A, MLL2, KDM6A, and EP300 and were mutated in up to one fourth of the tumors. One-third of the tumors were characterized by cancer–specific DNA hypermethylation.
An additional 281 tumors are in the pipeline and the complete data set will be analyzed beginning early 2015. We have preliminary mutation data on 238 samples (including the first 131) and six additional SMGs have been identified (Table 1). Several of the 38 SMGs have not previously been described in bladder cancer.
Three mutation clusters were identified combining copy number variation and somatic mutations characterized as: 1) “Focally amplified” - enriched in focal copy number alterations (eg. 3p loss/PPARG) and MLL2 mutations; 2) Enriched for TP53 and RB1 mutations, E2F3 amplifications; and 3) Papillary histology, FGFR3 mutant CDKN2A-deficient. This integrated analysis also indicated that the majority of tumors (69%) harbor one or more potentially actionable targets.
There are approved drugs targeting these genes in other tumor types that can be applied in clinical trials today as well as drugs in development. The future may be that therapies are target- specific rather than organ/tissue-specific. One example might be drugs targeting the PI3K/AKT/mTOR pathway which was altered in 42% of the TCGA cohort. These included activating point mutations in PIK3CA (17%), mutation or deletion of TSC1 and TSC2 (9%), and overexpression of AKT3 (10%).
These were largely mutually exclusive with one another suggesting that targeting these specific alterations may be effective. One example is TSC1mutations which may convey unique sensitivity to the mTOR inhibitor Everolimus (approved for as suggested by Iyer, et al2. While loss of function mutations in TSC1 occurred in only 6% of bladder cancers, they are also present in several other TCGA tumor types including lung adeno and squamous cancers, ovarian, colorectal and breast cancer suggesting that clinical trials targeting this pathway across these tumor types may be a rational approach.
The NCI has launched an innovative clinical trial (MATCH - Molecular Analysis for Therapy Choice) using this strategy that will treat patients with advanced solid and hematologic cancers targeting up to 30 specific gene alterations in small phase II trials. The Alliance cooperative group is planning to test this paradigm in locally advanced and metastatic bladder cancer where frontline therapy has failed. It is proposed that actionable targets for this trial will include alterations in FGFR3, the PI3K/AKT/mTOR pathway, and RB inactivation.
Integrated analysis of the 131 tumors in the initial TCGA muscle invasive bladder cancer cohort using unsupervised clustering of mRNA, miRNA and RPPA suggested four expression clusters, three of which were fully characterized as follows. Cluster I comprised 25% of tumors and shows papillary morphology and FGFR3 dysregulation including mutation, amplification and three tumors with translocations involving TACC3, which results in constituitive activation of the kinase domain of FGFR3 and reported previously by Williams, et al3.
Tumors harboring these translocations may be particularly sensitive to FGFR3 targeted therapy. Clusters I and II express high HER2 (ERBB2) and estrogen receptor beta signaling signature, sharing features with Luminal A breast cancer. Cluster III shows similarities to Basal-like breast and squamous cell head and neck carcinomas. Cluster III has a high cancer stem cell content, similar to the basal phenotype reported earlier by Volkmer and Ho and colleagues4,5. Cluster III is also enriched with tumors with mixed squamous histology similar to the subtype described by Sjodal, et al, which had the worst prognosis compared to three other subtypes which correlate well with the four subtypes described in the TCGA dataset6.
Cluster III also correlates with other TCGA subtypes in including basal-like breast carcinoma, head and neck squamous and lung squamous cell carcinomas1. This has been confirmed with a recent “cluster of cluster” TGCA pan cancer analysis, which stratified the bladder cancer cohort into three subtypes. Two were similar to lung adeno and lung squamous carcinoma as well as head and neck carcinoma while the third, which contained the majority of tumors, was unique to bladder cancer and had the best prognosis compared to the two other subtypes7.
The fourth cluster (10%) showed similarities to cluster I with FGFR3 alterations and high expression of miR-99a and 100 as well as similarities to cluster III with low expression of e-cadherin and mi-200a and b, both associated with epithelial-mesenchymal transition. Gordon Roberston and Andy Mungall have also performed unsupervised clustering of miRNA in 310 tumors and provide an integrated analysis of anti-correlations of clusters of miRNA regulating many pathways including epithelial-mesenchymal transition, DNA methylation, and FGFR3 expression among many others.
We have now integrated the mutation subtypes and the expression clusters for 234 tumors. Expression cluster I is enriched with FGFR3 mutations, amplification and expression and urothelial differentiation markers (UPK1A, 2, II, KRT 20). Cluster III is enriched with squamous morphology, basal markers (KRT 14, 5, TP63) and expression of several immune response genes. These include programmed cell death-ligand 1 (PD-L1) whose normal function to suppress the immune system during events such as pregnancy, but may be up-regulated in cancers thereby allowing them to evade the host immune system8.
Blockade of the interaction between PD-L1 and its receptor by anti–PD-L1 antibodies have been shown to potentiate immune responses and mediate antitumor activity9. The PD-L1 receptor PD-1 may be blocked by monoclonal antibodies including pembrolizumab, nivolumab and AMP-514, and several clinical trials in bladder cancer are planned or underway. PD-L1 may be targeted by human monoclonal antibodies such as MPDL3280A and MEDI4736.
A recent Phase I trial of MPDL3280A in advanced bladder cancer demonstrated rapid and sometimes sustained responses in tumors with high expression of PD-L110. A Phase II trial in patients who have progressed following cis-platinum-based chemotherapy (NCT02108652) and a Phase III trial comparing MPDL3280A to chemotherapy in a similar group of patients are underway (NCT02302807).
The total muscle invasive TCGA cohort now includes 412 tumors that have met the pre-specified quality controls for pathology and RNA quality and have been distributed to the genome characterization centers (See Figure). This complete data set will increase the power to detect additional low frequency events11, validate the cluster analyses, test hypotheses regarding chemotherapy resistance and provide a host of translational opportunities for functional validation and targeted therapy trials. Outcome analyses were deliberately not included in the analysis of the first 131 tumors as the follow-up data were not mature and will be included in the final analysis.
We expect to be able to include this in the final analysis of the full cohort. The analysis working group will reconvene in early 2015 to begin the final analysis with the expectation of publishing an update comprehensive integrated analysis.
There is also a large unmet need for comprehensive genomic characterization of non-muscle invasive bladder cancer (NMIBC). FGFR3 mutations characterize low-grade Ta tumors and high-grade tumors share similar genomic alterations as muscle invasive carcinomas.
The NCI is sponsoring a clinical trials planning meeting to be held in March 2015 with the goals of designing a targeted therapy trial and an immunotherapy-based trial and a more comprehensive understanding of the genomic landscape is a critical step in this process.
Prof. Seth Lerner. Scott Department of Urology Beth and Dave Swalm Chair in Urologic Oncology Baylor College of Medicine Medical Center Houston, Texas (USA)
Editorial Note: Article contributed on behalf of project co-chair John Weinstein (MD Anderson Cancer Centre), David Kwaitkowski (Broad Institute) and the Analysis Working Group.
The reference list has be retrieved at EUT@uroweb.org.
Saturday, 21 March8.30-11.00: Plenary Session 1Controversies in surgical oncology in bladder and kidneyHot topic lecture