Introduction

Evaluation of new antiemetic agents follows principles similar to the development of new antineoplastic agents. Preclinical models, such as the ferret [1] and the least shrew [2], with emetic responses similar to those in patients, can be used to identify agents that may have antiemetic properties against standard challenge agents and can also be used to characterize the potency and time course of antiemetic activity, placing these agents in perspective to those currently in clinical use. Phase I trials in normal subjects may be employed to characterize toxicity and pharmacokinetic parameters as well as exploring potential drug interactions. However, Phase II and Phase III studies take on an added dimension of complexity since new antiemetic agents are designed to be used against emetic agonists which may be of different potency and have different time-course characteristics themselves. In view of the lack of highly effective salvage therapy for patients with refractory emesis, it would be reasonable to evaluate new agents in this setting, and success of a new single agent in attaining antiemetic protection would indeed be an impressive achievement [8]. However, using these agents against the most difficult emetogenic challenge will not gain a realistic estimate of efficacy in standard first-line use, and other factors, such as anticipatory emesis, will also come into play [14]. Promising antiemetic agents will therefore rapidly move into the Phase III clinical trials setting and are likely to be evaluated as components of combination regimens (Table 1).

Table 1 Recommendations on evaluation of new agents
Full size table

New antiemetic agents can be divided into those that are members of known antiemetic families and those that are unique in structure or mechanism of action. In view of the success of antiemetic drug development over the last 30 years, it is seldom advisable to compare a new antiemetic agent against highly or moderately emetogenic chemotherapy to a true placebo. If a new agent is a member of a known antiemetic family, it is likely to be included as a replacement for a known agent in a standard regimen, with the new regimen then compared for non-inferiority to the original standard regimen [3]. Although the non-inferiority design will require much larger patient populations, it is preferable to design such a study, with results that can later be evaluated for suggestions of superiority [13], than to design a superiority trial that may prove to be too small to establish the equivalent value of a new agent that could be added to the antiemetic armamentarium.

Identification of a new family of antiemetics allows a different form of Phase III comparison. In these cases, randomization of a standard regimen with or without the new agent (preferably in a randomized double-blind placebo-controlled design) can be used to assess the contribution of the new family to overall antiemetic protection and to characterize this protection as related to the emetogenicity of the challenge agents, the time course of protection, and the unique or additive effects compared to the standard regimen [4]. It should be noted that changes in the standard regimen may require a re-evaluation of these study designs and results. For example, the combination of an NK-1 antagonist with dexamethasone and ondansetron has been found to be superior to dexamethasone and ondansetron alone [7]. However, since recent data has suggested that the efficacy and mechanism of action of palonosetron may differ from that of ondansetron [10], then an NK-1 antagonist combined with dexamethasone and palonosetron cannot be immediately assumed to be superior to dexamethasone and palonosetron and may require further evaluation. In addition, as indicated by the overall Multinational Association of Supportive Care in Cancer (MASCC) guidelines, the relative value of different regimens must also be considered specifically in regard to the emetogenicity of the antiemetic challenge itself. Relative efficacy in the setting of moderately emetogenic chemotherapy, for example, may be similar but not identical to that against highly emetogenic chemotherapy [15].

Classification of emetogenicity of antineoplastic agents has established the framework against which antiemetic efficacy is defined. At the 2004 Perugia Antiemetic Consensus Conference, the previous five-level antiemetic classification was simplified to a four-level classification by combining the previous Level 3 (30–60% of patients with emesis) and Level 4 (60–90% of patients with emesis) into a single moderately emetogenic classification (emesis in 30–90% of patients), since clinical separation of these two classes of patients was found to be extremely difficult [6]. The categories of highly emetogenic (previously Level 5—>90% of patients with emesis), low emetogenic (previously Level 2—10–30% of patients with emesis), and minimal emetogenic (previously Level 1—<10% of patients with emesis [clinically, agents considered to be non-emetogenic]) were left intact. However, this resulted in a wide range of stimuli gathered under the single heading of moderately emetogenic chemotherapy. At the 2009 Consensus Conference, this classification was left intact as was the basic principle that the emetogenic classification scheme should be used to describe single agents, since the potential variety of combination doses and schedules of even a few chemotherapeutic agents might defy meaningful classification. However, it was recognized that the commonly used combination of the moderately emetogenic agents cyclophosphamide and doxorubicin that forms the basis of many breast cancer regimens did appear to create a particularly potent moderately emetogenic combination that commonly served as the basis for antiemetic clinical trials and that might require more aggressive antiemetic regimens.

As new antineoplastic agents have been developed, we have attempted to add them to the emetogenic classification schema. Such efforts continue to be hampered by the limited recording of “common” toxicities, such as emesis, during antineoplastic drug development, and the unregulated use of prophylactic antiemetics during antineoplastic drug development even before emetogenicity of the agents has been specifically established. Classification of new agents must therefore depend to a certain extent on expert opinion and on synthesis of various limited data sources, still allowing significant consensus but limiting confidence to that allowed by the quality of the underlying data. The accompanying tables represent the present emetogenic classification of commonly used antineoplastic agents. Numerous new agents have been added since 2004, and some agents have been reclassified based on additional data (Table 2).

Table 2 Emetogenic potential of single intravenous antineoplastic agents
Full size table

As recognized at the 2004 Consensus Conference, the increasing use of oral agents (both cytotoxic agents and biologic agents) has created an additional challenge, since such agents tend to be used in extended regimens of daily oral use rather than the single bolus administration commonly seen with intravenous agents. Whether emetogenicity of such agents should be defined based on the acute emetogenicity of a single dose or the cumulative emetogenicity of a full course of chronic administration remains an issue for discussion. This is particularly critical since some of the newer agents may only become consistently emetogenic after a week or more of continuous administration, so that evaluation of only a single day would greatly underestimate the clinical concern. In general, emetogenic classification has therefore been established based on that of a full course of therapy as clinically employed (Table 3). Chronic oral administration also erases the distinction between acute and delayed emesis so that definitions for oral agents must intrinsically differ from those of intravenous agents.

Table 3 Emetogenic potential of single oral antineoplastic agents
Full size table

Control of acute vomiting allowed the identification of delayed vomiting as a separate entity. Control of both acute and delayed vomiting has now led to the realization that nausea may be the greatest remaining emetogenic challenge [11]. Although vomiting and nausea seem to appear and respond in parallel, they are not the same phenomena. While vomiting can be objectively measured in terms of number of emetic episodes, nausea is a subjective phenomenon that requires different measurement tools and definitions. It must also be recognized that the standard primary endpoint for emetogenic trials, Complete Response, is defined as “no vomiting and no use of rescue medication” and does not specifically refer to nausea or protection from nausea at all. Certain populations, such as young women being treated for breast cancer, have been identified as being particularly susceptible to nausea out of proportion to the appearance of vomiting [5]. Preliminary clinical trials of several agents have also suggested that just as some agents may be more effective against acute vomiting and some against delayed vomiting, other agents may be more effective against nausea than against vomiting and vice versa [9, 12]. Identification and characterization of antinausea agents and rational inclusion of these agents into antiemetic regimens may be the primary challenge in coming years (Table 4). Characterization of the nausea phenomenon and its relationship not only to vomiting but also to anorexia and cachexia will lead to greater insights and improved control of these various gastrointestinal toxicities.

Table 4 Remaining challenges in antiemetic therapy
Full size table