Abstract
In almost all current Phase I designs, toxicity response is treated coarsely as a binary indicator of dose limiting toxicity (DLT) and a lot of useful toxicity information is discarded. We are the first to establish a novel toxicity scoring system to treat toxicity response as a quasi-continuous variable and utilize all toxicities in Phase I trial. The generally accepted and objective parts, such as a logistic function, grade and type of toxicity, and whether the toxicity is DLT, are used so that the toxicity scoring system is relatively objective. Our toxicity scoring system has been successfully applied to an isotonic design (ID) to develop an extended isotonic design (EID). Simulation study and application of EID to the data of a real Phase I trial demonstrate that EID can always estimate a more accurate maximum tolerated dose (MTD) according to the exact toxicity profile under any toxicity profiles without additional cost or length of the trial. These cannot be accomplished in designs using a binary indicator of DLT, such as Standard 3+3 design, ID, and continual reassessment method (CRM). Moreover, our EID is relatively objective, model free, and simple to use. Our toxicity scoring system can also be applied to other designs, such as CRM and escalation with overdose control (EWOC), to improve their efficiency and accuracy in MTD estimation by utilizing all toxicities. Our novel toxicity scoring system and EID may help to begin a new era in which toxicity response is treated as a continuous variable.