A network based approach to drug repositioning identifies plausible candidates for breast cancer and prostate cancer

Abstract

Background: The high cost and the long time required to bring drugs into commerce is driving efforts to repurpose FDA approved drugs-to find new uses for which they weren't intended, and to thereby reduce the overall cost of commercialization, and shorten the lag between drug discovery and availability. We report on the development, testing and application of a promising new approach to repositioning.

Methods: Our approach is based on mining a human functional linkage network for inversely correlated modules of drug and disease gene targets. The method takes account of multiple information sources, including gene mutation, gene expression, and functional connectivity and proximity of within module genes.

Results: The method was used to identify candidates for treating breast and prostate cancer. We found that (i) the recall rate for FDA approved drugs for breast (prostate) cancer is 20/20 (10/11), while the rates for drugs in clinical trials were 131/154 and 82/106; (ii) the ROC/AUC performance substantially exceeds that of comparable methods; (iii) preliminary in vitro studies indicate that 5/5 candidates have therapeutic indices superior to that of Doxorubicin in MCF7 and SUM149 cancer cell lines. We briefly discuss the biological plausibility of the candidates at a molecular level in the context of the biological processes that they mediate.

Conclusions: Our method appears to offer promise for the identification of multi-targeted drug candidates that can correct aberrant cellular functions. In particular the computational performance exceeded that of other CMap-based methods, and in vitro experiments indicate that 5/5 candidates have therapeutic indices superior to that of Doxorubicin in MCF7 and SUM149 cancer cell lines. The approach has the potential to provide a more efficient drug discovery pipeline.

Keywords: Cancer treatment; Computational drug repositioning; Drug screening.

Fig. 1

Analytic workflow. (1) After mapping mutated genes to the FLN, identify the functional neighbors that are up or down regulated (DEG: differentially expressed genes) and within significantly enriched disease pathways (FDR < 0.05). (2) Map the genes that are down or up regulated by drug candidates to the FLN (3) Compute the MP score; i.e. the significance of the functional overlap between the drug and disease perturbed genes (see text). (4) Rank the compounds according to the MP score. (5) Compute the sensitivity and specificity of the ranked list of compounds. (6) Repeat the process with different groups of MAG and DRG (Drug Response Gene) generated by looping over the parameters (m & k). (7) Choose the parameter set that has highest sensitivity and specificity. (8) The drug candidates are chosen form the ranked list generated by the best parameter set. (9) The top ranked drug candidates are chosen for in vitro experimental validation

Fig. 2

Comparison of performance for the MFM with other methods. We applied CMap datasets to compare performance of MFM with Shegemizu et al. and Lamb et al. The sensitivity and specificity were calculated as explained in the Methods section, and the area under the ROC curve was used as a measure of performance. UCDB: prediction of drug candidates that can down-regulate genes up-regulated in cancer. DCUB: prediction of drug candidates that can up-regulate genes down-regulated in cancer. It shows that MFM consistently outperforms the two methods in different datasets and diseases

Fig. 3

a FDA approved indications of predicted drug candidates; b Half maximal inhibitory concentration (IC50) (μM) of predicted drug candidates and Doxorubicin against MCF7, SUM149 and MCF10A; c and d Therapeutic index (TI) and maximal inhibitory concentrations (E_max) of predicted repositioned drug candidates on MCF7, SUM149 and MCF10A. (*Currently used FDA drug for breast cancer; Therapeutic index (TI) was calculated as a ratio of the IC50 of MCF10A, to the IC50 of MCF7 and SUM149)