Phase I Drug Combination Trial Design: Walking the Tightrope

Abstract

Drug combination strategies are used in the palliative and curative settings for patients with cancer. Before a specific drug is incorporated into such a combination treatment strategy, it has to be explored using the clinical drug development route. The first critical step is the design of a combination phase I trial. As for all phase I trials, the design of a phase I drug combination trial is of utmost importance because it will dictate the course of the trial and, by its nature, it already incorporates, implicitly or explicitly, a short list of the potential answers to the primary end point question. While many consider phase I trials of drug combinations uninteresting and while there has not been much debate on their design, in reality, the complexity of the design of a phase I trial increases exponentially with the number of different drugs and treatment modalities included in the treatment strategy. If the treatment strategy to which the new drug is added already has curative intent, this adds a complicating dimension to the design of the trial. It means that the researchers will have to balance determining the feasible dose of the added experimental agent without compromising cure. In this issue of Journal of Clinical Oncology, Haddad et al report a phase I study of cetuximab/docetaxel/cisplatin/fluorouracil given as induction treatment to patients with locally advanced squamous cell carcinoma of the head and neck. Taking all the aforementioned factors into account, they should be applauded for performing a dose-finding study in the curative setting. Yet, the study raises some phase I methodology questions. In a series of theoretical phase I trials of drug A and drug B, considering only one dosing regimen per drug and disregarding all available drug data except for the maximum tolerated dose (MTD) and recommended phase II dose (RP2D) for each drug given as a single agent, as a result of the dose escalation track used, different theoretical MTDs and RP2Ds can be defined. For instance, a fixed dose of drug A (at its MTD) can be used while escalating drug B until the MTD of this combination has been determined. One should bear in mind that the MTD of drug B in the combination is not necessarily determined by its single-agent MTD; it could be higher than the single-agent MTD, perhaps because of drug-drug interactions with drug A. A similar trial could be designed for a fixed dose of drug B and dose escalation of drug A, potentially rendering a different MTD of the combination. A different approach would be a dose escalation in which both drugs are started at a relatively low dose and increased alternately. Depending on the escalation steps, different MTDs can be determined. Because the escalation steps can go back and forth between the two drugs whenever an MTD is determined, at least two MTDs can be determined. For example, the MTD is exceeded at dose level 4 for both drugs. If the previous dose level (drug A at dose level 4 and drug B at dose level 3) was safe, that can be called the MTD. However, one could also argue that the combination of drug A at dose level 3 and drug B at dose level 4 should be evaluated for feasibility and, if it is considered safe, that can also be called the MTD. As illustrated in this theoretical example in a simple combination phase I trial with only two drugs, the number of dose-escalating options to approach the MTD easily reaches four, potentially rendering four different MTDs. By adding drug C, this theoretical number rapidly increases to at least 18, and one does not want to think about the potential number of MTDs that are definable with four drugs, as used in the study by Haddad et al. Thus, one of the basic dilemmas facing investigators who design combination phase I trials is choosing the dose escalation approach that will determine the most biologically optimal and useful MTD. From a purely phase I methodology point of view, there is no right or wrong dose escalation approach, because at the end of the study, there will be an answer: one of the many possible MTDs. However, from a treatment practice point of view (even more so when there is curative intent), one should try to identify the dose escalation approach that is best supported by conceptual and actual data and will lead to dose recommendations that, after subsequent phase II and III studies, stand the best chance of changing practice by improved outcome. Guidance can be found in data on toxicity patterns, trials on comparable agents, and (if more than two drugs are used) in previously defined MTDs of the specific combinations to which a new drug is added. Guidance should not be sought in maximizing potential sales of the investigated drug or in beliefs not supported by evidence. As noted, the choice of the dose escalation approach should be determined by the best possible scientific and clinical practice rationale. Haddad et al elected to determine the MTD with a fixed dose of cetuximab (C), cisplatin (P), and docetaxel (D) while escalating the dose of fluorouracil (F). This approach is somewhat remarkable, because the same group of investigators reported that the current standard of induction chemotherapy in patients with squamous cell carcinoma of the head and neck consists of P F with or without D. The authors’ rationale for choosing their dose escalation approach is the potential overlap in the GI toxicity of C and F, but it is unclear why they compromised on the dose of the standard drug (F) and not on the JOURNAL OF CLINICAL ONCOLOGY E D I T O R I A L VOLUME 27 NUMBER 27 SEPTEMBER 2