Considerable unmet biomarker potential exists in the treatment of bladder cancer
By Dr. Patrick Hensley (Houston, US) & Dr. Ashish M. Kamat (Houston, US), University of Texas, MD Anderson Cancer Center
This article reflects the highlights of the lecture Dr. Ashish M. Kamat will be giving during “Thematic Session 09: Urinary biomarkers: Are we there yet?” at EAU21 Virtual. This session takes place in Virtual room 3 on Saturday 10 July, 13.15 – 14.15 CEST.
Of the various sources of biomarker material available for prognostic and predictive investigation in bladder cancer (BCa), urine has the strongest rationale for study. It is readily available, easy to collect in large quantities, and provides a non-invasive source with direct tumour contact. This has led to a deluge of urine-based biomarkers, but most have been developed for the diagnosis and surveillance of BCa, with relatively few having sufficient accuracy to predict therapeutic response.
Intravesical immunotherapy with Bacille Calmette- Guérin (BCG) is in most need of a ‘marker of treatment response.’ BCG is the mainstay treatment for intermediate and high-risk non-muscle-invasive BCa (NMIBC). While highly effective when used correctly, there are various nuances that clinicians need to be aware of when using BCG to obtain the best result for our patients, including appropriate dosing and scheduling, and how to tailor the therapy to avoid unnecessary toxicity. A marker of response to BCG would thus be invaluable and many have worked towards this goal. Unfortunately, as recognised in a recent review, an international panel concluded that the best markers of response remain clinicopathologic factors such as tumour stage, grade, size, presence or absence of carcinoma in situ, focality and recurrence history. 
A recent advance in NMIBC has been the adoption by the US Food and Drug Administration (FDA) of unified definitions to aid in novel, bladder-sparing therapeutic developments and single-arm trials in high-risk NMIBC [3,4]. The occurrence of high-grade disease within 6-12 months after receiving adequate BCG denotes a BCG ‘unresponsive’ state. [3,4] As drugs are studied and approved during this advance, markers of treatment response become ever more important to allow us to appropriately select personalised treatments for patients.
Several candidate markers studied in the preclinical setting primarily exploited the mechanism of BCG therapeutic response.  Interleukin (IL)-8 is one of the first cytokines expressed in the urine after BCG therapy. In a pilot study of 20 patients with NMIBC or upper tract urothelial carcinoma, Thalmann et al. profiled IL-8 expression in voided urine at 6 hours post-BCG instillation. Patients with high IL-8 expression had lower rates of recurrence and progression.  These findings were confirmed in a subsequent study, which identified IL-18 as another candidate cytokine with predictive capacity for BCG outcomes.  Additionally, when investigators profiled urinary Th1 response after BCG , failure to detect IL-2 during induction was associated with a shorter time to recurrence and progression.
Because BCG immunogenicity is complex and nonspecific, single candidate markers alone may be unreliable prognostic tools. We thus measured levels of 12 mechanistically relevant urinary cytokines in 130 patients with intermediate and high-risk NMIBC at our institution using an enzyme-linked immunosorbent assay (ELISA) at baseline and at specified time points throughout BCG therapy.  The final Cytokine Panel for Response to Intravesical Therapy (CyPRIT) nomogram was generated, including 9 inducible cytokines after BCG instillation (IL-2, IL-6, IL-8, IL-18, IL-1ra, TRAIL, IFN-γ, IL-12[p70], and TNF-α), which predicted the likelihood of recurrence with 85.5% accuracy (95% CI 77.9-93.1%).
In addition to induced cytokine expression in the acute phase after BCG therapy, it has also been proposed that high-baseline, pre-treatment levels of certain cytokines may promote tumorigenesis and progression. We profiled a broad panel of cytokine expression in urine samples and peripheral blood leukocytes at baseline prior to BCG.  Indeed, expression of IL-8 in urine was associated with recurrence in BCG-treated patients, with patients who had higher baseline urinary IL-8 levels experiencing a 4-fold increased risk of tumour recurrence (HR 3.72, 95% CI 1.49-9.28, P=0.005). High-baseline IL-8 expression in peripheral blood leukocytes similarly correlated with disease recurrence. This predictive capacity of urinary IL-8 was subsequently verified independently in a separate pilot trial of NMIBC patients treated with BCG ± intradermal HS-410.  Taken together, these studies confirm IL-8 as a putative negative pre-treatment prognostic marker for BCG response.
A fluorescence in situ hybridisation (FISH) assay which detects aneuploidy in chromosomes 3, 7, and 17 and loss of the 9p21 locus in voided urine samples (UroVysion®) has been approved by the FDA as an adjunct to cystoscopy for screening patients with haematuria as well as for the surveillance of patients with a history of BCa. Investigators at the Mayo Clinic (US) studied 37 patients primarily receiving BCG for NMIBC: all 12 patients with positive post-treatment UroVysion FISH suffered a tumour recurrence, with over half being muscle-invasive (MIBC).  Positive post-treatment UroVysion was confirmed to be a predictor of recurrence in several other independent studies with variable adjuvant intravesical agents for NMIBC. [13-16]
Our group subsequently investigated the role of FISH as a dynamic marker at various time points to predict recurrence and progression in NMIBC patients treated with induction and maintenance BCG. In a cohort of 126 patients, those who had a positive FISH result during therapy were 3-5 times more likely to develop recurrence and 5-13 times more likely to be faced with progression in comparison with patients with negative mid-treatment FISH.  This was subsequently validated in an independent, multi-centre trial where FISH was predictive of recurrence and/or progression events at baseline (HR 2.59, 95% CI 1.42-4.73) prior to the 6th induction instillation (HR 1.94, 95% CI 1.04-3.59) and at 3-month follow-up (HR 3.22, 95% CI 1.65-6.27).  While not specifically FDA-approved for this indication, use of UroVysion FISH is supported by the AUA Guidelines for assessing the response to intravesical BCG. 
The identification of so-called ‘molecular BCG failure’ patients, defined as positive FISH at 6 weeks and 3 months with negative 3-month cystoscopy, has tremendous clinical relevance to the identification of those at the highest risk of BCG failure with continued therapy. Patients with such a molecular BCG failure have significantly higher rates of recurrence and progression than patients with a negative FISH and may be candidates for early enrolment into clinical trials that compare novel agents with the continued standard of care BCG therapy. 
Unmet needs and future directions
Despite the candidate predictive urine markers mentioned above, considerable unmet biomarker potential exists in the treatment of BCa. Clearly in times of BCG shortage, alternative intravesical therapies such as with chemotherapy are increasing in use. Additionally, with emerging intravesical and systemic therapeutic options for BCG unresponsive disease, as well as immunotherapeutics and antibody-drug conjugates with proven efficacy in earlier disease states, there is an obvious need for growth in our biomarker armamentarium.
A viable option is to translate available tissue-based predictive protein and molecular biomarkers into urine assays. Targeted exon sequencing of pretreatment NMIBC tumours identified ARID1A mutations as being predictive of BCG failure.  Additionally, recent thorough molecular classification of NMIBC has correlated candidate molecular subtypes to innate sensitivity and resistance to BCG therapy.  This is in addition to the well-characterised molecular subtypes of MIBC with the ability to predict response to systemic chemotherapy.  Sensitive biomarkers to predict complete clinical response to intravesical and systemic therapies would have tremendous implications on bladder preservation. The Southwest Oncology Group (SWOG) 1314 trial prospectively profiled the ability of the COXEN tissue-based genetic classifier to predict complete pathologic response to neoadjuvant cisplatin-based chemotherapy.  As sequencing technology becomes more refined and clinically applicable, urine-based genetic material (exfoliated tumour cells, cell-free DNA, exosomes, etc.) may prove a viable source for molecular subtyping and predictive biomarker development.
Using the clinical cohort from our CyPRIT study, collaborators at Cedars Sinai (Los Angeles, US) studied the Oncuria™ test, which measures 10 cancer-associated biomarkers.  They found that pre-treatment urinary concentrations of MMP9, VEGFA, CA9, SDC1, PAI1, APOE, A1AT, ANG and MMP10 were increased in subjects with disease recurrence. A combinatorial predictive model of treatment outcomes reached an area under the receiver operating curve of 0.89 (95% CI: 0.80 – 0.99), outperforming any single biomarker, with a test sensitivity of 81.8% and a specificity of 84.9%. Patients with higher urinary levels of ANG, CA9 and MMP10 had a significantly higher risk of disease recurrence.
Additionally, significant progress has been made in identifying immunohistochemistry (IHC)-based protein signatures in pre-neoadjuvant chemotherapy-treated MIBC specimens. These signatures that are predictive of pathologic response  could rationally be profiled in pre-treatment urine specimens using high-fidelity ELISA-based platforms.
Lastly, we can presume there exists a predictive biomarker rationale for characterising the expression and molecular constitution of therapeutic targets. For example, UroSEEK is a urine-based molecular assay designed for detection and surveillance of BCa. It detects alterations in 11 commonly mutated genes, including TERT, FGFR3, PIK3CA, TP53, HRAS, KRAS, ERBB2, CDKN2A, MET, MLL, and VHL , most of which are druggable targets with agents that have been actively studied in clinical trials or that have recently received FDA approval.
There are no current urine biomarkers FDA-approved for predicting therapeutic response along the spectrum of NMIBC or MIBC. Off-label use of the FISH assay (UroVysion®) in voided specimens of patients with NMIBC undergoing treatment with intravesical BCG has a predictive capacity and is supported by the AUA Guidelines. The most promising preclinical evidence of urine-based predictive biomarker potential in the NMIBC setting involves the profiling of baseline-and-elicited-cytokine response to BCG therapy. As novel agents transition from intravesical and systemic therapies with nonspecific targets and host responses (i.e. BCG and cytotoxic chemotherapies) to targeted therapeutics (monoclonal antibodies and antibody-drug conjugates), we expect biomarkers to become equally predictable and precise.
The reference list can be made available to interested readers upon request by sending an email to: communications@ uroweb.org.