Molecular evaluation of early bladder cancer to improve clinical decision making

Abstract

Bladder cancer (BC) is the 12th most common cancer worldwide, with approximately 5,300 deaths every year in the UK. Currently, diagnosis is only possible through the assessment of the tissue by invasive techniques such as cystoscopy and transurethral resection of bladder tumour (TURBT). Depending on its invasiveness in muscularis propria, it is clinically divided into non-muscle invasive BC (NMIBC) and muscle-invasive BC (MIBC). Low-grade NMIBCs usually undergo surveillance for many years with recurrent cystoscopies and TURBTs, while high-grade tumours are treated with BCG (Bacillus Calmette–Guérin) instillations in addition to life-long surveillance. Although the efficiency and quality of the surgical techniques have improved over the years, BC, even at early stages, continues to have a distinctively high incidence of recurrence and progression to invasive stages.

This project addresses the two main unmet needs of bladder cancer clinical management. First, the need to identify patients who are likely going to recur: all patients must undergo years of surveillance, which is not only extremely unpleasant but also a burden for the health system. Identifying markers that predict recurrence would help clinicians to avoid unnecessary surveillance and promptly provide treatments to those who will recur. Second, there is the necessity to identify patients who will not respond to BCG treatment. In 40% of cases, patients fail to respond to BCG and can have serious side effects. Being able to predict BCG would help nonresponders to have access to alternative treatment straightaway and avoid side effects.

To address these aims, the tumour immune microenvironment was investigated at the RNA and protein levels. Clinical collaborators identified a discovery set of sixty-five archival Formalin-Fixed Paraffin-Embedded (FFPE) biopsies, composed mostly of NMIBC but containing also MIBC in order to represent the spectrum of stages and grades. Genes of interest were identified through literature or RNA analysis using the NanoString PanCancer IO 360 panel. To be able to quickly translate our results into the clinic, interesting markers were selected and stained on biopsy’s sections.

Antibodies were optimised in immunohistochemistry and multiplex immunofluorescence with Quantum Dots or Tyramide Signal Amplification fluorophores. Images were analysed with machine learning image analysis and data-driven analysis using a newly identified analysis pipeline.

The composition of the immune cell population in NMIBC showed higher CD4+ and CD8+ T-cells and lower CD68+ macrophages when compared with MIBC. Interestingly, the density of CD8+ cells was also statistically higher in patients that did not recur compared to patients who recurred.

RNA analysis of 27 NMIBCs identified four genes that were differentially expressed in patients who recurred, KIR3DL1, S100A8, CCNA1 and KIR2DL3, while CCND1, LAMB3, SERPINB5, COL17A1 were linked to patients who did no recurred. KIR2DL3, CCNA1, CCND1 and LAMB3 were further evaluated at the protein level in 53 NMIBCs, however, the results did not confirm the earlier discovery.

Interestingly, when comparing the response to BCG treatment in NMIBC, two genes were identified and confirmed to be differentially expressed at the protein level: MMP7 and IFI27. IFI27, also known as ISG12a, was linked to a positive response to BCG, possibly through the regulation of tumour necrosis factor-related apoptosis-inducing ligand (TRAIL) and was identified as the most promising marker. IFI27 could identify 78% of BCG responders with a specificity of 76% on a cohort of forty-three patients treated with BCG. Further evaluation of IFI27 immediately before the start of BCG therapy and afterwards could provide a more comprehensive view of its role in the response to BCG.