During a session at the 2018 Genitourinary Cancers Symposium, Peter Black, MD, professor, department of urologic sciences, University of British Columbia, discussed using molecular subtypes, the Coxen model, and gene mutations to select patients and therapies for neoadjuvant chemotherapy.
During a session at the 2018 Genitourinary Cancers Symposium, Peter Black, MD, professor, department of urologic sciences, University of British Columbia, discussed 3 clinical biomarkers with potential to select patients and therapies for neoadjuvant chemotherapy (NAC).
Black began by explaining that there are multiple prospective randomized trials and meta-analyses that demonstrate survival advantage from NAC. The 2 main limitations with NAC are: only about 40% of patients seem to be benefitting from the treatment, and the other 60% are potentially suffering from side effects from chemotherapy and potentially also unnecessary delay in definitive radical cystectomy. The other main limitation is that NAC has not been widely adopted either here in North America or in Europe. Some countries do not give it routinely at all.
“I think one of the ways forward to overcome both of these limitations is with biomarkers,” said Black. “If we have a biomarker that would tell us which patients are likely to respond or not to respond, we can avoid using NAC in the likely non-responders, and if we had better patient selection, we’d also probably have better buy in for adoption of NAC. “
Black identified 3 clinical biomarkers that are in development and close to potential clinical implementation.
There are several research groups that identified a molecular taxonomy for bladder cancer based on RNA expression. The idea is that basal and luminal split, with basal tumors having a gene expression profile that resembles the basal layer of the urothelium; it’s more stem-like and less differentiated. The luminal tumors, on the other hand, have a gene expression profile that resembles more differentiated luminal cells, Black explained.
According to Black, these subtyping systems have a limitation in that you need a whole patient cohort to create a cluster to be able to label an individual patient as belonging to one of the subtypes, which is not practical in clinical practice.
So, Black’s team developed a single-patient classifier called a GSC. It includes 4 cohorts: luminal, luminal-infiltrated, basal, and claudin-low. They put together a multicentric cohort of patients who were receiving NAC, which included 269 patients who received cisplatin-based chemotherapy, who were used for all survival analyses. There were 305 total patients for developing the classifier. The control population of almost 500 patients who did not receive NAC was based on publically available data and literature.
“The bottom line with this classifier was, and if you look at the non-NAC patient population, you can see that the luminal patients clearly do better than everyone else,” said Black. “The other 3 cohorts are relatively closer together. In the NAC data set you can see they jump up the most, they get the most benefit from NAC with respect to survival.”
A research team at the University of Virginia developed the Coxen model, which is currently being evaluated in a clinical trial. The model was initially based on the NCI-60 panel of cancer cell lines, none of which were bladder cancer, but it's a comprehensive database of gene expression and IC50 values for over 100,000 compounds, said Black. The team was able to integrate bladder cancer data into the gene expression data as well as a panel of bladder cancer cell lines, and through bioinformatics techniques, were able to come up with a predictive model for any given drug based on gene expression.
Genomic alterations, mutations, and copy number changes
A research team led by Eliezer Van Allen, MD, from the Dana-Farber Cancer Instiitute, and Jonathan Rosenberg, MD, from Memorial Sloan Kettering Cancer Center, examined EERC2 mutations in a cohort of 50 patients. The team compared 25 patients who received cisplatin-based chemotherapy and had a complete response, with 25 who had residual muscle-invasive disease. According to Black, it was the ERCC2 gene that was a predominant in the responders, while non-responders did not carry mutations in ERCC2 mutations.
Combining subtypes and mutations
Black concluded by demonstrating what would happen if the subtyping and mutation biomarkers were put together.
“One-third of patients are basal, and if you add in the patients who have the DNA repair gene mutations, you end up with about 50% of patients who we would predict would respond well to cisplatin-based chemotherapy based on a combination of mutations and subtyping,” said Black.