Many advances in the treatment and care of cancer patients have been closely linked to the availability of more effective pharmaceutical agents. As research continues to develop new and improved chemotherapeutic agents, it is heartening that the FDA maintains a flexible approach to the drug approval process and will consider an array of indicators of drug efficacy, including patient self-reports of health-related quality of life. Dr. Beitz’s article provides an overview of some of the issues related to study design and data analysis and interpretation that emerge when health-related quality of life is used as an end point in such trials.
The article includes a number of helpful suggestions for the development of strategies in drug evaluation. It also raises some additional questions. I will focus on two issues: (1) the interpretation of multiple end points, and (2) the meaning of effectiveness.”
Multiple Indicators of Health-Related Quality of Life
Dr. Beitz cites studies that included multiple indicators of drug efficacy. For example, studies examining the effects of mitoxantrone(Drug information on mitoxantrone) (Novantrone) plus prednisone(Drug information on prednisone) in patients with hormone-refractory prostate cancer included patient-reported changes in pain scores and analgesia, in addition to standard assessments of tumor response. Studies comparing gemcitabine (Gemzar) and fluorouracil(Drug information on fluorouracil) used the outcome of “clinical benefit response,” a composite of both patient self-reports (ie, pain intensity, analgesic use), observer ratings (ie, Karnofsky performance status), and physical data (ie, weight change). Whether a patient was rated as having a clinical benefit response was specified by a complex algorithm.
When multiple indicators of health-related quality of life are used, the investigator is faced with the possibility that the measures will yield inconsistent results. For example, Malia Wilson and I recently reviewed the use of quality-of-life assessment in cancer clinical trials.[1] We identified three randomized, placebo-controlled studies comparing well-being in patients with advanced non–small-cell lung cancer (NSCLC) or colorectal cancer who received treatment with hydrazine sulfate or placebo.[2-4] All three studies included survival, toxicity, and quality-of-life end points.
The findings of our review indicated inconsistent interrelationships among outcomes: For example, toxicity and quality of life yielded a similar pattern of results in the two NSCLC studies[2,4] that was not upheld in the colorectal study.[3] Also, decrements in survival correlated with decrements in quality of life in the colorectal study by Loprinzi et al[3] but not in the same group’s NSCLC study.[4]
In these three studies, the interpretation was facilitated because hydrazine sulfate did not result in better outcomes, regardless of indicator. However, what if an agent led to better survival but worse toxicity and quality of life? Moreover, is it possible that patients might experience more toxicity but better quality of life?
In the case of the studies of gemcitabine(Drug information on gemcitabine) vs fluorouracil discussed by Dr. Beitz, an analysis of individuals who could complete two cycles of therapy indicated comparable reductions in pain intensity in the two groups but lower analgesic use and higher Karnofsky status in the gemcitabine group. It is somewhat difficult to reconcile these findings, which raise a number of questions: For example, is there something about gemcitabine that lowers analgesic use, or, alternatively, something about fluorouracil that increases its use? Is a physician’s perception of how well a patient is doing affected by how many prescriptions are being written? Thus, could Karnofsky ratings have been affected by whether a patient was taking analgesics? If patients had been asked about their satisfaction with treatment, would pain intensity have been a more powerful predictor than analgesic use or physician ratings, thus indicating equivalent effects of both agents on health-related quality of life from the patient’s perspective? Clearly, relationships among study outcomes are complex and warrant further study.
Meaning of Effectiveness
Throughout Dr. Beitz’s paper, health-related quality of life is referred to as providing evidence of “drug effectiveness” that will lead to drug approval. “Effectiveness” is also the theme of the FDA documents cited at the end of the article. However, Dr. Beitz and her FDA colleagues appear to be using this term somewhat differently than it is being applied in much of the health services literature. For example, Hays et al[5] define “efficacy” as “what is possible,” as compared to “effectiveness,” which is what happens under existing circumstances.
The studies described by Dr. Beitz seem more relevant to establishing what is possible—ie, whether a drug can confer patient benefit—than to determining how a drug performs outside of the setting of controlled research. This inference is supported by Dr. Beitz’s recommendation of phase III clinical trials as the preferred design to assess health-related quality of life outcomes.
Advantages and Disadvantages of Phase III Trials
There are distinct advantages to a phase III design. For example, phase III studies control for many biases that threaten the interpretation of study results: Patient selection biases are largely eliminated, and exposure to the experimental condition (eg, drug administration) can be specified precisely.
At the same time, phase III design is subject to other biases, particularly those related to external validity: Patients in clinical trials are often highly selected, and settings in which clinical trials are conducted may have standards of care that differ significantly from those in nonresearch settings.[6]
In fact, only a small percentage of cancer patients participate in clinical trials; this percentage has been estimated to be 6%.[7] Most clinical trial participants are unrepresentative of typical cancer patients in a number of ways. For example, trial participants are likely to be younger: An analysis of Southwest Oncology Group therapeutic registrations over a 4-year period indicated that individuals over 65 years of age constituted only 23% of trial participants, whereas 54% of cancer cases are diagnosed in patients who fall into this age group.[8]
Beyond Phase III Trials
Thus, to provide adequate examination of the full effectiveness of cancer drugs—ie, how these drugs work in the patients and setting typically found in oncology clinical practice—one may need to go beyond phase III clinical trials. In fact, phase IV, “postmarketing” trials provide an ideal opportunity to examine how approved drugs actually perform in the context of patient and environmental factors that may influence acceptance, tolerance, and adherence to therapy. These factors can include comorbidity and intercurrent disease, availability of supportive and symptomatic care, transportation to treatment centers, economic factors (including health insurance and systems of health care), and patient and family preferences. Factors such as these can have a major influence on the effectiveness of drug therapies in the “real world” and can be examined through a variety of study designs.[5]
Multiple kinds of assessments can yield useful information about health-related quality of life in cancer patients, and various experimental and analytic designs may be appropriate, depending on the objectives of a particular study. Dr. Beitz’s paper clearly communicates the FDA’s flexible approach to evaluating new agents so that they will be available to cancer patients. It is hoped that the FDA will encourage continued evaluation of how these pharmaceuticals affect health-related quality of life as they are used in everyday clinical practice.
A 52-year-old man presented with an erythematous lesion in the axilla of unknown duration. Surgical excision was performed. What is your diagnosis?
