By answering these questions, the author evaluates a study comparing oral ondansetron to placebo for cancer patients receiving emetogenic chemotherapy.

Nausea and vomiting caused by toxic chemotherapeutic agents may compromise effective therapy. Uncontrolled nausea and vomiting not only result in discomfort and distress but may also lead to dehydration and poor compliance with subsequent chemotherapy [1]. The introduction of ondansetron (Zofran; GR38032F) into medical practice marked a major advance in the management of chemotherapy-induced emesis. Ondansetron preferentially antagonizes the 5-hydroxytryptamine₃ (serotonin) receptor at peripheral and central sites [2]. Data from clinical trials support ondansetron as the antiemetic agent of choice in preventing acute cisplatin-induced emesis. Four previous studies have compared ondansetron to high-dose metoclopramide (Reglan) for emesis [3-6]. An analysis of these studies indicates that ondansetron provides better control of nausea and vomiting than metoclopramide during the first 24 hours after cisplatin therapy. Additionally, ondansetron has fewer and milder side effects than does metoclopramide.

In this issue of Annals, Beck and colleagues [7] report a randomized, double-blind study comparing 1 mg, 4 mg, and 8 mg of oral ondansetron to placebo in patients given cyclophosphamide (Cytoxan)-based chemotherapy for the first time. Of 349 patients, 174 also received 35 mg/m² of doxorubicin (Adriamycin), including 52 on the placebo arm. Another 108 patients received 600 mg/m² of cyclophosphamide. Ondansetron or placebo was given before chemotherapy and was repeated 4 and 8 hours later. To evaluate delayed emesis, the study drug was administered daily at three fixed times for 2 more days. Prochlorperazine (Compazine) tablets (10 mg) were provided as "rescue" medication for patients complaining of persistent nausea and vomiting. The investigators used valid end points for measuring the antiemetic effect of ondansetron. This study [7] shows that ondansetron at all doses is superior to placebo, but the optimal dose has yet to be established. Ondansetron is also effective in combating delayed nausea and vomiting. Toxicity was minimal and consisted of mild headache and constipation.

This trial had one troublesome issue: The control group received only a placebo to cope with the possibly severe side effects of the emetogenic chemotherapy with doxorubicin and cyclophosphamide. Questions must be raised about the ethical legitimacy of deliberately imposing such a burden on some patients. Competing concerns must be carefully weighed, because a scientifically validated antiemetic agent would have major therapeutic benefits for large numbers of patients, including those patients in this study [7] who have additional chemotherapeutic treatment.

The American Medical Association's Council on Ethical and Judicial Affairs has declared:

It is fundamental social policy that the advancement of scientific knowledge must always be secondary to primary concern for the individual [8].

I am not implying that clinical investigators should never expose patients to any risk at all. Some degree of risk is inevitable in medical care. The realistic ethical duty of the physician as clinician-investigator is to answer the question: What degree of additional risk of harm may a clinical trial impose on a patient?

In seeking a balanced answer to this question, the clinician-investigator's first principle must be to do no known harm. It is not ethical for the physician to withhold what is reasonably believed to be more beneficial nor to administer what is reasonably believed to be more harmful. The investigator must be able, in sincere good faith, to entertain the null hypothesis. If an arm of a study is known to be less beneficial or more harmful than alternatives, clinician-investigators must protect patients from that additional known risk.

By definition, a placebo is a clinically inert substance intended to provide no therapeutic benefit. An inert substance may help differentiate between a placebo effect and true therapeutic activity. Nevertheless, the requirement to "do no known harm" imposes a burden of justification on clinician-investigators using placebo-controlled studies. A placebo-controlled study must answer four questions.

1. Was any therapeutic intervention available that could be reasonably assumed to be less harmful than the placebo control? Doxorubicin and cyclophosphamide are potent emetogenic agents associated, respectively, with a moderate and high potential to induce vomiting [9]. Standard clinical practice is to use combination antiemetic therapy to control emesis. In a recent Cancer and Leukemia Group B study testing high, intermediate, and low dosages of cyclophosphamide, doxorubicin, and fluorouracil as adjuvant therapy in breast cancer [10], virtually all 500 patients in the low-dose arm received antiemetics for their first chemotherapy course (Budman D. Personal communication). These patients were given cyclophosphamide and fluorouracil, 300 mg/m² each, and doxorubicin, 30 mg/m², doses that were lower than those used by Beck and colleagues [7]. Gralla [9] recommends a combination of metoclopramide, dexamethasone, and Benadryl specifically to manage emesis induced by cyclophosphamide and doxorubicin. Although the best antiemetic treatment has not yet been defined, one or a combination of a number of drugs could have been selected as a concurrent treatment control using available data and clinical experience [9, 11-13].

To justify a placebo control, the authors note that no antiemetic has been approved by the Food and Drug Administration (FDA) for doxorubicin-induced emesis. This important justification cannot be easily dismissed. The criteria for FDA approval are that a study is deemed adequate and well controlled. However, patients withdrawing from the study were given the option of taking prochlorperazine. This is implicit recognition by the authors that even a low dose of prochlorperazine can be an effective antiemetic treatment.

Among physicians, a difference of opinion exists as to whether a specific therapeutic regimen for doxorubicin-induced emesis has been established. This points, in part, to a perceived lack of rigorous studies supporting many antiemetic agents that clinicians are using off-label for cancer patients.

2. Did the placebo control pose more than a minimal risk to those patients? Because the study is designed to be nontherapeutic for the patients in the placebo arm, those patients cannot be subjected to more than minimal risk. As defined in the Federal Register:

Minimal risk means that the probability of harm or discomfort anticipated in the research are not greater in and of themselves than those ordinarily encountered in daily life or during the performance of routine physical or psychological examinations or tests [14].

Persons involved in institutional review should consider whether any treatment arm poses more than minimal risk to patients. In making this decision, the Institutional Review Board determines whether the rights and welfare of patients are protected, whether the scientific design is sound, and whether informed consent can be obtained by adequate means. The investigators' Institutional Review Board approved the study; therefore, they must have decided that the study did not pose more than a minimal risk to the patients.

In considering the issue of minimal risk, one concern is that the patients' high probability of nausea and vomiting could have been reduced by giving an established antiemetic regimen. If the control arm had included such antiemetic agents, this would have strengthened the study by proving superiority to standard treatment. Additionally, it would have lessened concern about minimal risk, because standard antiemetics were used.

3. Was the study designed in every way possible to minimize potential harm to the patients receiving placebo. Patients receiving a placebo as an antiemetic agent in an ambulatory setting cannot be monitored closely, thus limiting the timing and type of therapeutic intervention necessary to relieve their symptoms. However, in Beck's study [7], patients could withdraw at any time, and antiemetic rescue (prochlorperazine) was available if needed. Therefore, measures were taken to minimize potential harm to patients receiving placebo.

4. Were the patients fully and accurately informed of the additional risk of being in the placebo group? The informed consent, approved by the Institutional Review Boards, clearly explained to patients that they might receive a placebo. Additionally, therapeutic alternatives including other antiemetics were listed. Thus, the authors followed established guidelines regarding institutional review and informed consent.

I would have preferred a control arm using antiemetic agents whose efficacy against a potent emetogenic agent are used in current clinical practice. Unquestionably, however, important countervailing considerations support the study as it was designed. It should now become possible to apply the results of this study to future antiemetic agent trials. Given ondansetron's established efficacy for chemotherapy-induced emesis, it should now be the standard for testing new antiemetic agents. Eliminating a placebo arm will minimize the risk to patients in clinical trials.