Approximately 25% of bladder cancers (BC) patients are diagnosed as muscle-invasive bladder cancer (MIBC). Radical cystectomy (RC) with lymphadenectomy, preceded by neoadjuvant chemotherapy in eligible patients, is the standard treatment for localized MIBC. However, despite undergoing radical surgery, up to 50% of MIBC patients will develop local relapse or distant metastasis within the first two years of RC (1). Metastatic BC patients have a poor prognosis with a five-year relative survival of 6% (2). These patients are generally offered platinum-based chemotherapy; however, the response rate achieved is around 50% (3). An improved understanding of the underlying biology of BC would lead to advances in the treatment and clinical outcomes of this tumor.

The rapid evolution of sequencing technologies has revealed the great complexity of the genomic abnormalities of individual cancer cells and substantial molecular intratumor heterogeneity (ITH), which may evolve over the course of disease and treatment exposure (4). These findings may explain the failure of some approaches toward predictive biomarker development that are exacerbated by tumor sampling bias driven by ITH while also indicating the way forward (5). Examining tumor clonal architecture and its evolution throughout treatment and disease progression may enable the identification of common tumor driver events and individual heterogeneous somatic events present in each tumor that contribute to drug resistance and treatment failure.

Liquid biopsy has emerged to eliminate tumor molecular profile limitations: it is a non-invasive technique, detects tumor clonal evolution in real-time, and reveals ITH (6). Notably, cell-free DNA (cfDNA)-based liquid biopsy has multiple clinical applications and has been studied in different types of tumors, including BC, as a biomarker for detecting molecular relapse or minimal residual disease, monitoring treatment response, or identifying therapy resistance mechanisms (7–11). However, cfDNA-based liquid biopsy has not been implemented in clinical urological practice yet. This is most likely because of the high BC genetic intra- and inter-lesion heterogeneity that makes a challenge to identify those biomarkers predicting which patients are at high risk of metastatic relapse and consequently guiding treatment decisions. In this study, we aimed to determine DNA mutations providing aggressive behavior to bladder tumors and analyze them in patients’ cfDNA during their follow up after RC. The identification of these mutations in bloodstream could be helpful to monitor tumor evolution in response to treatments, thus improving clinical management of bladder cancer patients.

Bladder cancer is one of the tumors with the highest genetic heterogeneity (4), presenting a challenge to identify a set of recurrent cfDNA biomarkers for use in the clinical setting to monitor tumor evolution in response to treatment (12).

Here, we have addressed the challenge of BC heterogeneity by analyzing genetic features of several different morphologically tumor regions from primary bladder tumor and locoregional/distant metastases from six BC patients. Furthermore, we have identified genetic alterations shared between primary bladder tumors and their metastases in different BC patients and we have been able to determine them in plasma cfDNA from these patients during their follow-up. The assessment of these BC genetic alterations providing aggressive behavior in plasma cfDNA would allow real-time monitoring of tumor evolution in a real clinical scenario.

Since some of the MIBC molecular subtypes have been associated with bladder tumor aggressivity, we first classified bladder tumors included in this study according to basal or luminal differentiation. As expected, the four MIBC tumors from our cohort that presented basal differentiation harbored mutations in the TP53 and/or RB1 genes. This subtype has been associated with aggressive tumors and a poor prognosis (13, 14) and indeed, these patients did have a poor prognosis, presenting a two-year mortality rate of 75%. The tumor from Pt#5 with luminal differentiation harbored an FGFR3 mutation, as expected for this molecular subtype. This is the only non-muscle-invasive BC patient in our series and the one with the longest progression-free survival (seven years). In fact, BC patients with luminal tumors have been found to have the best prognosis compared with all other molecular subtypes (13, 14). Finally, BC tumors with both basal and luminal subtypes have been suggested to present aggressive behavior since they are considered to have a similar prognosis to basal subtype (13, 14) and actually Pt#1 showed a short cancer-specific survival. This tumor did not harbor mutations in the TP53, RB1, or FGFR3 genes. However, we found genomic alterations in KRAS, promoting the activation of the RAS pathway in this tumor, which has been suggested to be associated with luminal differentiation (14, 15).

Germane to the mutations present in the distinct tumor regions of primary bladder tumors as well as locoregional/distant metastases, we were able to confirm previously reported ITH of bladder tumors, as well as lesion-to-lesion genetic heterogeneity (12, 16, 17). In fact, phylogenetic trees showing branched tumor evolution in each patient demonstrated the existence of more aggressive clones in primary tumors that would originate the metastasis. Additionally, we found that distant metastases were genetically more altered than locoregional ones, and both metastatic regions were genetically more altered than primary tumors, most probably due to the emergence of resistance mutations during disease progression (18). Interestingly, although both high genetic intra- and inter-lesion heterogeneity were identified, we were able to detect several mutations in common between primary tumors and metastatic lesions. All six patients harbored at least one of these four mutations in primary tumors and metastatic lesions: TERT c.1-124C>T, ATM c.1236-2A>T, RB1 c.13delA, and FGFR3 c.742C>T. These inter-lesion-shared mutations were analyzed in cfDNA samples during patient follow-up for real-time monitoring of disease relapse. Remarkably, a higher DNA mutation concordance was found between metastatic lesions and cfDNA samples during follow-up than between the primary tumor and cfDNA at the time of cystectomy. Multiple factors can account for this DNA mutation discordance between tumor and cfDNA at surgery, such as preanalytical methodological factors (amount of plasma analyzed, assay sensitivity, etc.) (10) or low cfDNA levels that could prevent variant detection (false negative). In fact, other authors previously reported better concordance between cfDNA and tissue samples when high cfDNA levels were present (19).

Substantial evidence suggests that the detection of circulating tumor DNA (ctDNA; the variable fraction of plasma cfDNA derived from tumor) in the bloodstream following potential curative treatment is associated with a high risk of tumor relapse (7, 11, 20–22). However, the identification of specific mutations in cfDNA can also be an advantage for applying targeted therapy. For instance, those patients with advanced or metastatic BC harboring FGFR3 mutations in their primary tumor are already being treated with the Food and Drug Administration (FDA)-approved drug Erdafitinib^® (NCT05316155), a fibroblast growth factor receptor (FGFR) kinase inhibitor (23). Interestingly, an FGFR3 missense mutation (c.742C>T; p.Arg248Cys), a gain-of-function mutation that promotes cell proliferation and migration (24, 25), was found in Pt#5’s cfDNA samples, suggesting that liquid biopsy could be a useful tool to identify potential therapeutic targets during patient follow-up. In fact, this specific FGFR3 mutation was also identified in cfDNA samples from a series of BC patients who were candidates for Erdafitinib (12), emphasizing that the detection of the FGFR3 mutation in plasma cfDNA could be helpful to guide targeted treatment in a non-invasive way. FGFR3 alterations occur mainly in non-muscle-invasive BC (25), and it was mutated in the only patient of our series (Pt#5) with non-invasive disease at the time of RC. However, this patient did not present an RB1 mutation in their primary bladder tumor; however, they did in their metastatic tissue. This is most likely due to RB1 mutations being commonly found in higher BC stages and associated with poor patient outcomes (26).

As regards somatic mutations in DNA repair genes, such as ATM and RB1, these have been associated with better responses to cisplatin-based chemotherapy in MIBC (27–29). Interestingly, three patients from our cohort (Pt#3, Pt#4, and Pt#5) harbored an RB1 mutation both in primary tumors/metastatic sites and plasma cfDNA. This mutation was undetectable in cfDNA from Pt#3 and Pt#5 after treatment with gemcitabine-cisplatin chemotherapy. The remaining patient (Pt#4) did not receive cisplatin-based chemotherapy, and the RB1 mutation was present in their cfDNA during the entire follow-up. RB1 is a tumor suppressor gene that has been associated with DNA repair as well as cell proliferation and division (27). Several studies have demonstrated that mutations in RB1 are frequently found in bladder cancer (26, 27), although, as far as we know, the specific mutation (c.13delA; loss-of-function mutation) found in our cohort had not been described before. On the other hand, ATM is a core component of the DNA repair system, and its signaling pathway is known to play a role in the development of breast cancer and other tumors (30). Several mutations in ATM have been previously described in BC (26, 31), however, interestingly, the specific ATM mutation found in our cohort (c.1236-2A>T) was first described in our previous study analyzing bladder tumors (32). Finally, the TERT promoter mutation (c.1-124C>T, gain-of-function mutation) has been widely reported in several solid tumors, including bladder cancer (33–36), where it plays an important role in tumorigenesis. In a previous study (22), which included both progressive and non-progressive BC patients, we showed that none of the non-progressive patients harboring ATM and TERT mutations in their primary tumors, present these mutations in their cfDNA during their follow-up after RC. Furthermore, in another previous study, we described that the presence of the mutation TERT c.1-124C>T could be considered a biomarker of aggressivity in MIBC (32). This data suggests that the presence of mutation in ATM and TERT in plasma cfDNA after treatment is related to tumor dissemination and confers a poor prognosis to patients.

To the best of our knowledge, this is the first work to evaluate in plasma cfDNA a set of genetic alterations associated with aggressive behavior in bladder cancer, initially identified in primary and metastatic sites from BC patients. Nonetheless, we must acknowledge some study limitations. First, some tumors might not have shed ctDNA into the plasma at the time of cystectomy, even if the tumor had a targetable mutation, with the consequent risk of a false-negative result. In addition, this study includes only six patients. However, it must be considered the difficulty of obtaining metastatic biopsies from these patients. In fact, our cohort of metastatic patients is one of the largest in the literature considering the type of samples analyzed from each patient (different morphologically tumor regions from primary and metastatic sites). Nonetheless, more patients need be investigated to determine the final role of these four genes as biomarkers for monitoring tumor evolution and assessment of treatment response in advanced, metastatic BC patients.

A higher number of mutations was identified in distant metastases compared with locoregional metastases. Moreover, mutations in metastatic tissue had a high concordance with mutations in cfDNA samples. Despite BC genetic intra- and inter-lesion heterogeneity, specific mutations in TERT, ATM, RB1, and FGFR3 were found to be shared between primary and metastatic lesions in different BC patients. These shared mutations have been identified in cfDNA samples during patient follow-up, allowing real-time monitoring of these patients. Consequently, the analysis of BC mutations associated with poor prognosis in plasma cfDNA could be a valuable tool to monitor tumor evolution in response to treatments, thus improving the clinical management of BC patients.

The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found in the article/ Supplementary Material .

The studies involving humans were approved by Clinical Research Ethics Committee of the Hospital Clinic of Barcelona (HCB/2018/0026). The studies were conducted in accordance with the local legislation and institutional requirements. The participants provided their written informed consent to participate in this study.

RC: Investigation, Methodology, Writing – original draft. MI-T: Resources, Writing – review & editing. JO: Resources, Writing – review & editing. FR: Writing – review & editing. LR-C: Methodology, Writing – review & editing. SH: Writing – review & editing. BM: Data curation, Writing – review & editing. AA: Writing – review & editing. LI: Writing – review & editing. LM: Funding acquisition, Investigation, Supervision, Writing – review & editing.