Background: Combination therapy is routine in oncology, although it may be associated with adverse drug-drug interactions (DDIs). Machine-learning (ML) and Artificial Intelligence (AI) based tools have been applied to this problem, however they often lack predictive rationales. We have leveraged published data, including FDA drug label information, to create a rule-based tool designed to identify and classify potential drug-drug interaction candidates based on shared gene mechanisms, mechanistic impact, and supporting clinical evidence. Design: A proprietary database was developed and curated from the FDA drug labels of 338 approved drugs (271 small molecules and 67 biologics), capturing detailed information on drug-gene and drug-drug interactions. A rule-based Drug-Drug Interaction (DDI) assessment tool was then developed based on shared gene mechanisms, mechanistic impact, and supported clinical evidence. The performance of the tool was evaluated in 2 independent data sets. Results: The tool was first tested using a curated set of known DDI-positive and DDI-negative drug pairs, (20 DDI-positive, 10 DDI-negative) yielding 100 percent accuracy. A second set of 544 drug pairs predicted to have efficacy based on a mechanistic model (Cellworks) was then assessed, with 6.8% pairs (37/544) predicted to be DDI-positive. Of these, 34 pairs shared a common gene and showed a clear victim-perpetrator relationship, indicating a strong mechanistic link for possible interactions, while the 3 additional DDI-positive pairs (Fluorouracil, Oxaliplatin; Carboplatin, Nab-Paclitaxel; Trastuzumab, Paclitaxel) showed clinical evidence of interaction in the absence of shared genes. Evaluation of ClinicalTrials.gov showed that 7 of these pairs (18.9%) had failed clinical development due toxicity-related findings. 460 (93%) of the DDI-negative drug pairs lacked shared gene pathways, mechanistic relationships, and clinical corroboration. Conclusions: Developed specifically for oncology applications, the tool couples FDA label data and clinical information relevant to cancer treatment with mechanistic insights, providing a method to predict and mitigate toxicity risks arising from drug combinations. Future studies will aim to prospectively validate the tool.