Bioinformatics and clinic: predictive models in cancer patients stratification and novel therapies identification

Despite intense research efforts cancer remains difficult to treat due to chemotherapy resistance. The p53 tumor suppressor is crucial for cancer development, the DNA damage and chemotherapy response. We have created Boolean p53 interactome models and investigated their predictive power for cancer patients’ stratification. The comparison of model simulations obtained by knockout tests mimicking mutations with omics type of experimental data demonstrated a significant rate of successful predictions in osteosarcoma, colon cancer and mesothelioma cell lines, as well as in 71 mesothelioma patients. Model analysis allowed identification of deregulated pathways, prediction of therapeutic schemes and linking of the affected pathway with the patients’ clinical state. Model validations demonstrated successful predictions ranging from 52% to 85% depending on the drug, algorithm or sample used for validation. Patients were stratified depending on their p53 status and therapy received, then their clinical outcomes and simulation comparisons were used to identify 30 genes that correlated with survival. In patients with the wild-type p53, FEN1 and MMP2 exhibited the highest inverse correlation, whereas in untreated patients with p53 mutated, SIAH1 negatively correlated with survival. Using DRUGSURV, repositioned and experimental drugs targeting FEN1 and MMP2 were identified. Testing showed that drugs that target FEN1 (epinephrine and myricetin) have cytotoxic effect, whereas marimastat and batimastat, which target MMP2 have inhibitory effect on mesothelioma cell migration. In summary, p53 model has predictive properties with versatile potential for use in cancer treatment by identifying pathways crucial for tumor growth, by facilitating patients stratification and by identification of shifts in pathways required for chemoresistance. We believe that upon further testing in animal models and wider data base analysis, clinical decisions and personalized therapy can be devised based on individual patients’ genetic profile and previous chemotherapeutic treatment.