Recent years have seen a surge in the precision oncology market in two key dimensions. The first are efforts to guide precision oncology by sequencing tumors’ DNA and identifying actionable mutations in 300-400 cancer driver genes. The second complementary dimension is the development of new targeted therapies specifically inhibiting cancer drivers in tumors harboring actionable mutations in driver genes, a track followed now by many Pharma companies and a major source of new cancer drugs.
While these recent precision oncology efforts have led to exciting developments, their focus on just a few hundred driver genes has obvious limitations and regrettably, the fraction of patients that can currently benefit from the current precision-based therapies remains fairly modest (<10% of those sequenced in some estimations). Furthermore, the oncogene-centric approach does not provide an answer to an equivalently-significant genetic foundations of tumorigenesis – i.e. the loss of Tumor Suppressor Genes, the targeting of oncogenic events that are currently mostly unactionable and the targeting of non-oncogene addiction of cancer.
Despite the above-mentioned limitations, the ever-increasing volume of genomic data that is being collected (e.g., The Cancer Genome Atlas – TCGA) now offers unique opportunities for big data mining. A fundamental way to circumvent the limited number of actionable oncogenic-driving events is to identify molecular targets that become essential in a specific tumor genetic landscape. As mentioned above, Synthetic Lethality (SL) is a co-dependency on two non-essential genes, so when one of them is lost, the other becomes essential and hence a potential therapeutic target. Due to the genomic instability nature of cancer, many genes are deleted or under-expressed, and so, targeting their SL partner is a very attractive therapeutic approach. However, there are about 500 million such potential SL gene combinations, and reliably identifying SL partners via lab screens is an immense challenge, which is further complicated by the observation that the majority of SL identified in lab screens do not have translational significance.
Addressing these challenges, we developed a data-driven computational approach that mines the TCGA cohort to identify SL interactions that are more likely to be clinically relevant. Based on TCGA data analysis, our computational evaluation indicate that our approach can significantly increase the number of patients that could benefit from precision oncology treatments, meriting their further careful study.
Pangea’s state-of-the-art computational genomics enables to (1) recommend the most effective drugs to combat tumors of individual patients (2) to stratify the most relevant patients for a given treatment, and finally, (3) to facilitate the development of new SL-based drugs by providing a list of reliable new targets.
Proof of concepts using Pangea’s approach have been validated published in high Impact Factor journals.