Express Insight into Bladder Cancer

Bladder cancer: the ninth most common cancer worldwide, with nearly half a million new cases diagnosed each year. Unfortunately, less than three-quarters of those receiving a diagnosis will survive five years, and only about half can expect a decade. Why have those survival rates not significantly improved over time? The disease’s heterogeneity may be at least partly to blame, which is why Jun Zhu, Professor of Genetics and Genomic Sciences at the Icahn School of Medicine, opted to investigate a particular subtype of bladder cancer known as p53-like bladder cancer.

“p53-like muscle-invasive bladder cancers are generally resistant to cisplatin-based chemotherapy, but exhibit heterogeneous clinical outcomes with a prognosis intermediate to that of the luminal and basal subtypes,” explains Zhu, who is also Head of Data Science at patient-centered predictive health company Sema4. “There are recommended approaches for the treatment of luminal and basal bladder cancer subtypes. However, the optimal approach to p53-like bladder tumors remains poorly defined, so we urgently need novel therapeutic targets and better means to risk-stratify such tumors.”

MicroRNAs, or miRNAs, post-transcriptionally regulate gene expression by binding to mRNAs. In humans, base-pairing of a miRNA and its target mRNA is not perfect; rather, it is initiated by six to eight matched nucleotides, meaning that a single miRNA can target many genes and impact multiple biological pathways. The flip side? Such promiscuous binding means that the level of miRNA available for its target genes may be reduced. “Many studies show that miRNA expression levels don’t reflect miRNA functional activity well,” says Zhu. “Thus, we developed a computational approach – ActMiR – to infer miRNA activity based on expression levels of miRNAs and their predicted target genes.”

The procedure requires three pieces of information: (i) miRNA expression levels of each sample; (ii) mRNA expression levels of each sample; and (iii) the predicted target genes of each miRNA. The ActMiR method consists of three steps. First, the researchers estimated the average “baseline” expression of each target gene (while the miRNA was not impacting expression). Next, they defined the “degradation” levels as the difference between the observed expression levels of targeted genes for each sample that were affected by the miRNA and the baseline expression level. Finally, they used a linear model to represent the relationship between degradation levels and baseline expression of target genes for each sample; the coefficient represents the activity of miRNA.

When searching for miRNA biomarkers, says Zhu, most studies associate miRNA expression and clinical phenotypes. “Our approach is unique in multiple ways. First, our computational approach leverages miRNA activity instead of miRNA expression level to associate with clinical phenotypes. Second, we know that miRNA activities depend on genomic background, so we subtype bladder cancers and then quantify miRNA activity in each subtype.” The outcome? Two new prognostic miRNAs identified in p53-like bladder cancer.

“We examined the direct functional target genes of these two prognostic miRNAs and identified biological pathways significantly enriched for functional target gene set of each miRNA,” Zhu explains. “The functional target genes of miR-106b-5p were enriched in the bone morphogenetic protein (BMP) pathways, which are associated with bladder cancer invasiveness and tumor recurrence.” In in vitro experiments, the researchers showed that knocking down miR-106b-5p expression increased bladder cancer cells’ invasiveness, whereas overexpression of miR-106b-5p expression decreased invasiveness.

Bladder cancers are not routinely subject to molecular subtyping due to the heterogeneity within subtypes. “Our results suggest that miR-106b-5p activity can further categorize p53-like bladder tumors into more and less favorable prognostic groups, which provides critical information for personalizing treatment option for p53-like bladder cancers,” says Zhu. “We predicted potential therapeutic candidates that might specifically benefit miR-106b-5p underactive p53-like bladder cancers.” He and his colleagues are now refining their computational method by taking into account the combinatorial targeting of multiple miRNAs; they are also applying it to additional cancer types. For the p53-like bladder cancer project, Zhu is working with collaborators to test miRNAs and miRNA-drug combinations as therapeutics within in vitro and in vivo models. 

“We showed in our analysis that the percentage of p53-like bladder cancer patients responding to PD-L1 blockade immunotherapies is around 20 percent,” Zhu adds. “Even though tremendous progress has been made in immunotherapy development, personalized treatments for the p53-like bladder cancers are still urgently needed. We need to accelerate our in vitro and in vivo validation experiments to demonstrate the value of personalized treatments based on molecular subtypes of bladder cancer.”