應用合併使用兩個已上市藥物的複合藥治療癌症病患已經是目前癌症醫療的重要發展，其中個別藥物可能具有重疊的劑量相關毒性而提高複合藥的毒性。本文根據模型建立複合藥的第一期臨床試驗設計，找尋該複合藥產生劑量限制毒性之機率在目標毒性機率之下的最大耐受劑量組合。本文設計啟動階段，藉以快速地蒐集病患的毒性反應資料以利模型的應用，其中描述劑量毒性關係的模型為納入藥物交互作用的芬內或邏輯斯模型。本文將該模型引入鍵盤複合藥設計中的劑量組合之升降試驗過程，最後估計最大耐受劑量組合，所以此設計稱為模型化鍵盤複合藥設計。本文也針對毒性晚發的情形修正上述的模型化設計。本文進一步在多種毒性機率情境下進行模擬研究，比較本文所提出的設計與文獻中既有的設計在正確選擇最大耐受劑量組合及病患中毒的相對表現。模擬結果顯示，本文提出的模型化鍵盤複合藥設計相較於文獻中的既有模型化設計或模型輔助設計具有較佳的表現。;It has become recently an important issue to develop a combination of two drugs for cancer treatment. However, the toxicity of the combined drugs may be enhanced when individual drugs possibly have overlapping dose-limiting toxicities (DLTs). In this thesis, model-based designs are constructed to find the maximum tolerated dose combination (MTC) at which the probability of DLT is closest to the target toxicity probability (TTP). To quickly collect information to facilitate the application of the model-based procedures, the start-up phase is designed. The dose combination-toxicity relationship is then described by the Finney or logistic model incorporating interaction between two individual drugs. The models are then employed into the Keyboard combined drug design for dose combination escalation/de-escalation and MTC estimation. Therefore, the proposed design is called the model-based Keyboard combined drug designs. We also modify the proposed designs for the setting with late-onset toxicity. Finally, a simulation study is conducted to investigate the performance of the proposed designs relative to some designs in the literatures under a variety of toxicity probability scenarios on the correct MTC selection and toxicity to patients. The simulation results show that the proposed model-based Keyboard combined drug designs outperform the model-based design or the original Keyboard combined drug design.
