PBAC and UrC are rare and aggressive malignancies with a median survival for locally advanced or metastatic disease ranging between 12–24 months (9). This poor prognosis is the result of many different factors, such as 1) delayed symptoms resulting in diagnosis at advanced tumour stages, 2) no standard-of-care therapeutic recommendations, and 3) poorly known molecular pathogenesis and genomic landscape of the tumour (23). As randomized trials for the evaluation of clinical benefit of various drugs in UrC and PBAC are not feasible, precision medicine is of prominent therapeutic interest in these rare cancers (24). Comprehensive genome profiling could therefore be fundamental to drive therapeutic decisions in these patients. In the present study, we performed mutational analyses of tumour tissues from UrC and PBAC patients and sought to identify targetable alterations and corresponding drugs that could potentially be effective for these patients.
Our genomic analysis, in line with previous reports, found TP53 (UrC: 79%, PBAC: 42%) to be the most commonly affected gene in both UrC and PBAC (10, 15, 16, 18, 19, 21). Since TP53 mutated tumours progress faster and have a poor response to anticancer therapy, targeting p53 for cancer therapy seems to be an attractive strategy (25). Although TP53 has previously been considered as undruggable due to its essential role in cell survival, recently many drugs targeting TP53 mutant tumours are being tested in early-phase (Phase I/II) clinical trials (26).
KRAS mutations in UrC have been extensively investigated, as it is a commonly affected oncogene in CRC, a tumour type sharing large histological and molecular similarity with UrC. After summarizing published literature on the prevalence of KRAS mutations in both UrC and PBAC, it proved the most frequently tested gene with alterations in ~ 30% of UrC and 25% of PBAC cases, which is in agreement with our present results showing 33% mutational frequency in UrC and 25% in PBAC (11, 15, 16, 18, 19). These data underscore the importance of the RAS pathway in both UrC and PBAC (16, 18, 21, 27). The vast majority (11/14) of KRAS alterations were missense mutations in codon 12 (G12V, G12D and G12A) which is a similar pattern to that found in CRC (28). In recent years, several breakthrough structural and mechanistic studies have led to the clinical development of selective KRAS inhibitors. Last year, the FDA granted accelerated approval to sotorasib, the first KRAS-blocking drug for non-small cell lung cancer (NSCLC) patients. Phase I/II studies (e.g. NCT03600883, NCT04699188) are currently investigating the efficacy of sotorasib and other G12C-inhibitors in other tumour types as well.
In our UrC cohort, four patients carried loss-of-function alterations in their BRCA2 gene. When considering the therapeutic significance of these alterations, only one patient proved to bear a potentially significant (Tier 2) SNV, the rest were categorized into Tier 3. Recently, PARP inhibitors (olaparib, rucaparib, niraparib) have becomeavailable for patients with alterations in their BRCA1/2 or other homologue recombinant repair genes. Accordingly, our drug prediction recommended PARP inhibitors for the UrC patient with Tier 2 mutation. Although little is known about the efficacy of PARP inhibitors in UrC. The only report on the use of a PARP inhibitor for the treatment of an UrC patient came from a Japanese phase I dose escalation study of niraparib and described progression during therapy. As BRCA positivity was not an inclusion criterion in the study, the BRCA status of the UrC patient is unknown(13).
When considering copy number alterations, MYC amplifications were detected at the highest frequency (5/33, 15%) in our UrC cohort, which is lower compared to a previously published study (6/17, 35%). In addition, we found EGFR amplification in 6% (2/33) of our UrC patients, which was also lower, compared to the frequency of 20% found by Lee et al. (19). Furthermore, we found recurrent copy number gains in members of the FGF/FGFR signalling pathway (in four UrC and six PBAC patients). Although, previous genomic analyses of PBAC did not reveal amplifications of the MYC gene, here we found MYC as the most frequently amplified gene (25%).
MYC is a global transcription factor and a driver of many human malignancies, that has proven to be difficult to inhibit directly (29). In this context, it is interesting that CDK12 was found to be a synthetic lethal gene with MYC. These findings were corroborated in an independent study demonstrating that CDK inhibition triggered massive downregulation of MYC expression and its related genes (30). The overlap between MYC and the known cellular functions of CDK12, as well as the requirement of CDK12 for optimal processing of MYC, collectively indicate CDK12 as a potential therapeutic target for MYC-dependent cancers (31).
EGFR is a widely used therapeutic target. To date, numerous anti-EGFR compounds, both tyrosine kinase inhibitors (TKIs) and monoclonal antibodies, have been developed and approved for different cancers (32). This is also reflected by our results identifying the second highest number of recommended drugs for patients with tumours harboring an EGFR amplification. In the literature, two UrC patients have been reported to receive an EGFR inhibitor. One patient with an EGFR amplification experienced a persistent partial response to cetuximab in the third line setting (10), while the other UrC patient with immunohistochemically proven EGFR overexpression experienced a transient 55% decrease in tumour size with gefitinib treatment (14). These results suggest EGFR as a potent therapeutic target in UrC.
In addition, we found for the first time D-type cyclin (CCND1/2/3) genes to be affected by activating mutations in both UrC and PBAC samples with frequencies of 15% and 25%, respectively. CCNDs promote cell cycle progression from G1 to S phase by binding to and activating the cyclin dependent kinases CDK4/CDK6, thereby imparting oncogenic properties. The cyclin D-CDK4/6 complex is often hyperactivated in various tumours (e.g. NSCLC, head and neck, renal cell, breast, pancreatic and colorectal cancer) partly by gene amplification, and is therefore an attractive therapeutic target (33, 34). Recently, multiple CDK4/6 inhibitors have been approved for the treatment of breast cancer (33). In addition, several clinical trials are ongoing to assess the efficacy of palbociclib, abemaciclib, or ribociclib in other cancers (e.g. NCT03446157, NCT02022982 and NCT03356223). Although, molecular alterations of cell cycle pathway genes are not mandatory for the prescription of these drugs, their presence suggest a favourable effect. Accordingly, this mutation-drug association is being tested in an ongoing phase II pan-cancer trial (NCT04439201) assessing the efficacy of palbociclib in patients with various malignancies harboring CCND1/2/3 amplifications. Due to our present results, UrC and PBAC patients carrying activating amplification in their CCND genes may be good candidates for future studies.
Our analyses identified one UrC and one PBAC patient with MET amplification. MET alterations besides their primary cancer driver role can mediate resistance to other targeted therapies such as EGFR inhibitors. They do this through activation of downstream signal transduction that leads to escape from therapy-induced cell death (35). This effect was reflected in our QCI® drug prediction as it suggested resistant association between MET amplification and anti-EGFR drugs.
According to recent data, not only the above mentioned amplification of MET but also its exon 14 skipping alteration (METex14) is associated with acquired resistance to EGFR-targeting compounds (36). METex14 has been identified in about 3% of lung NSCLCs and other solid tumours like breast cancer and glioblastoma. In 2020 and 2021, two MET-inhibitors; capmatinib and tepotinib, were approved for use as monotherapies in NSCLC patients carrying METex14 (37). Here we are the first to report the METex14 alteration in treatment-naïve UrC and PBAC with an overall incidence of 3/38 (7–8%). Considering the durable response observed in NSCLC patients with METex14 alteration, UrC and PBAC patients with this alteration may also benefit from tepotinib or capmatinib therapy. Accordingly, in the only published UrC patient treated with tepotinib a durable disease stabilization could be observed (12).
We identified copy number gain of ERBB2 (HER2) in one UrC and one PBAC sample. In breast cancer, ERBB2 amplification is a well-known prognostic biomarker for worse survival in the absence of anti-HER2 therapy (38). There is an abundance of approved HER2-targeted agents not just for breast cancer but also for other cancer entities such as metastatic gastric or gastroesophageal junction cancers (39). Currently, several ongoing clinical trials are evaluating the potential benefit of targeting HER2 in various tumour types (e.g. NCT02465060, NCT02675829). Little is known about the prevalence of ERBB2 amplification in UrC and PBAC. A study investigating the prevalence of ERBB2 amplification in different tumours identified ERBB2 amplification in 2 of 7 (28%) UrC samples (39). This study also showed clinical benefit of HER2-targeted therapy in tumours for which HER2-inhibitors are not yet approved (39). In the present study, the QCI® drug prediction algorithm recommended the highest number of drugs for ERBB2 amplified tumours, suggesting this alteration as an attractive therapeutic target.
FLT3 amplification might also be a potentially actionable molecular alteration, although the majority of the kinase inhibitors approved so far are relatively nonspecific for FLT3 (e.g. nintedanib, ponatinib, sorafenib and sunitinib). Off-target activities of these multi-kinase inhibitors can contribute to higher toxicity causing severe adverse events (40). Accordingly, in the case report of Loh et al. an UrC patient with FLT3 amplification received sorafenib therapy, but it had to be discontinued shortly after drug initiation due to a serious adverse event. Sunitinib was subsequently administered without toxicity, but an additional treatment change was required due to disease progression (11). Next-generation inhibitors, such as quizartinib or gilteritinib, are more specific and potent FLT3 inhibitors, with more favourable toxicity profiles, however, to date these drugs are approved only for acute myeloid leukaemia and thus no data on their activity in solid tumours is available (41).
This study has several key strengths. This is one of the largest studies of UrC and PBAC with respect to case number and number of the assessed genes. It is the first study in UrC and PBAC that systematically applies a clinical decision support tool to match driver aberrations with clinically approved drugs. Finally, we report here for the first-time recurrent alterations in CCND1-2, NOTCH3, RNF43, CDK12, FGFR4, CREBBP and SMARCA4 genes in UrC and mutations in MYC, CDKN2A and FLT3 genes in PBAC.
On the other hand, this study has several limitations. Due to the rarity of UrC and PBAC, a retrospective approach was needed to collect samples from multiple institutions over a long time period. Associated differences in sample handling and specimen age may result in heterogeneous quality of FFPE tissues. Consequently, a relatively high rate (~ 9%) of samples did not pass quality control. In addition, as the clinical interpretation could be carried out in a retrospective manner, we were not able to assess whether the recommended drugs would have been effective in the assessed UrC and PBAC patients. A further limitation is the heterogeneity of databases regarding both judgment of pathogenicity and druggability of certain variants. In addition, as the CNVs were predicted by bioinformatics tools, orthogonal validation for example by fluorescence in-situ hybridization (FISH) would be needed to validate the presence of the alteration. Some of these limitations could be addressed with the addition of in vitro (e.g., organoid) and in vivo (patient-derived tumour xenografts) models to improve our ability to predict drug response, which could then improve treatment selection for patients with rare cancers.
In conclusion, our results suggest significant overlaps in the genomic landscape of UrC and PBAC. The cell cycle pathway is the most affected pathway, followed by the DNA damage control, RAS and PI3K pathways. However, large individual heterogeneity was observed in the mutation patterns. In the majority of cases at least one potentially druggable alteration was identified, highlighting the promise of genetic profiling to guide treatment of these rare malignancies.