Recently, researchers have used machine learning algorithms to investigate genomic alterations associated with metastatic prostate cancer (mCRPC), in the hope of discovering novel drug targets.
If mPC responds to medical castration, a surgical treatment option, it is referred to as metastatic castration-sensitive prostate cancer (mCSPC). Unfortunately, after around 16 months there is a high likelihood that the disease will progress into a lethal form called metastatic castration-resistant prostate cancer (mCRPC).
Current mCRPC therapeutic options
Chemotherapies have been approved for mCRPC, but each only prolongs life by about 4 months. Single-agent targeted therapies are a type of precision medicine that interfere with molecular targets to stop cancer growth and spread. Currently, Poly-ADP-Ribose-Polymerase (PARP) is the only approved single-agent targeted therapy for mCRPC.
The limited number of therapeutic options available for men with mCRPC is largely due to the extremely difficult nature of metastasis in mPC patients – 80% of men with prostate cancer have bone-only metastases and obtaining a biopsy from bone metastases is not usually feasible. Therefore, this is a sizeable obstacle for the development of targeted therapies for the disease.
But, as the progression of mCRPC leads to a significantly worse prognosis, it is essential that targeted drugs are identified that prevent or delay this process. Recently, it has been investigated as to whether genomic profiling of tumour somatic mutations in cell-free DNA (cfDNA) could provide insight into novel targets for mCRPC drug development.
Using machine learning to study prostate cancer
For the first time, researchers have used machine learning algorithms to explore cfDNA data to reveal targetable genomic alteration patterns linked to the progression of mCRPC. It was found that amplifications in several signalling pathways were enriched in mCRPC – specifically RTK, MAPK, PI3K and G1/S.
The identification of these pathways is hugely exciting as none of them are currently targeted by any approved therapies for metastatic prostate cancer. Therefore, this discovery may provide a starting point for the development of novel therapeutic combination strategies for men with prostate cancer. Further research is now required into treatments that could provoke mechanisms of resistance in mCRPC, including an in-depth focus on inhibitors of these newly identified signalling pathways.
This research has shown that significant advances are achievable using machine learning-based analysis of genomic data obtained from cfDNA. Moreover, using these methods overcame a major limitation that is currently faced in mCRPC research – obtaining biopsies from bone metastases. Therefore, similar approaches could not only determine therapies to target the genomic aberrations that cause metastatic prostate cancer, but also have the potential to be replicated to provide key biological insights for drug discovery in other diseases too.
Image credit: FreePik sewcream