Editors' ChoiceCancer

Personalized cancer medicines

See allHide authors and affiliations

Science Translational Medicine  25 Mar 2015:
Vol. 7, Issue 280, pp. 280ec48
DOI: 10.1126/scitranslmed.aaa9872

Over 100 different types of cancers afflict the human population. This large family of diseases is united by a common theme—abnormal cell growth that has the potential to spread across the body. Cancer is treated with a combination of approaches including chemotherapy, radiation, surgery, hormonal therapy, targeted therapy, and palliative care. The selection of the treatment depends on the cancer type, stage, and tissue(s) affected. Some of these therapies are relatively nonspecific, such as chemotherapy, which acts by killing cells that divide rapidly such as cancer cells. Targeted therapy, on the other hand, works on specific molecular differences between cancer and normal cells for better efficacy and reduced toxicity. However, high intertumor heterogeneity is a major hurdle in applying therapeutic targeted agents to treat most cancer patients.

High-throughput sequencing technologies are enabling molecular understanding of disease processes like never before. Rubio-Perez et al. now use these techniques to study drivers of tumorigenesis in over 4000 tumor samples. They present an in silico prescription strategy which began by identifying drivers of tumorigenesis in pan-cancer cohort, mapping therapeutic agents targeting these drivers, and linking this knowledge for personalized treatment options for patients. The driver mutations in the 4068 samples (pan-cancer cohort) were identified on the basis of a selective advantage to tumor cells for growth and faster proliferation. These drivers broadly acted by loss of function (LoF), gain of function (GoF), or switch of function (SoF). In the current study, 45% of the driver mutations caused loss of function, whereas 36.8% were activating mutations. Overall, 459 mutated driver genes acted in one or more tumor types. Some of these had been previously established cancer-causing genes, but some have not been previously reported. Interestingly, all identified drivers showed abnormally high connectivity in interactome networks, significant enrichment for rare germline variants, and low baseline tolerance to germline variants with functional impact. The drivers belonged to predictable processes like cell cycle and DNA damage, angiogenesis, apoptosis, proliferation, and cell adhesion, and also included molecules in emerging processes like chromatin regulatory factors, splicing factors, and members of the ubiquitin proteolysis system. At least one driver alternation was observed in 90% of the tumors analyzed in the study. FDA-approved drugs, clinical trials, and preclinical candidates were then explored as therapeutic options. A total of 96 drivers were targeted by 57 FDA-approved drugs; 47 clinical trial candidates and 20470 ligands showing binding to 77 drivers. Of the 57 FDA-approved drugs, 32 targeted multiple drivers, stressing the fact that a large number of cancer drugs have poor selectivity.

This comprehensive assessment of over 4000 tumor samples indicates that only 5.9% tumors are tractable by approved agents following clinical guidelines. However, almost 40% of patients could benefit from different repurposing options based on the genetic profile of the patients. Almost 73% patients can also benefit from drug candidates currently under trial. The information generated as part of this approach is curated in the Cancer Driver Actionability Database. This information can be exploited for interpretation of newly sequenced tumor genomes or design of sequencing panels as also in identifying repurposed candidates for prospective drug development. Despite several limitations, the approach described in the study establishes a proof of principle to intelligently exploit cancer genome information towards precise cancer medicine.

C. Rubio-Perez et al., In silico prescription of anticancer drugs to cohorts of 28 tumor types reveals targeting opportunities. Cancer Cell 27, 382–396 (2015). [Full Text]

Navigate This Article