Quality-of-Life End Points in Oncology Drug Trials

Introduction

With a small number of exceptions, notably in the treatment of refractory tumors, oncologic drugs are marketed on the basis of a demonstrated clinical benefit, such as survival. Health-related quality of life (HRQL) is a multidimensional construct that represents the patient’s perspective on valued aspects of health and functioning.[1,2] For oncologic drugs that cannot be shown to have an impact on patient survival, demonstration of a favorable effect on HRQL outcomes could be considered evidence of clinical benefit.[3,4] Randomized, controlled clinical trials are the preferred venue for assessing HRQL outcomes in oncology.

Issues Related to Study Conduct and Design

Measurement of HRQL is accomplished through the use of scales or instruments that solicit patients’ responses to questions related to their physical, psychological, and social functioning. Other dimensions of HRQL, such as cognitive or sexual function, may also be evaluated, and there may be global assessments of health or life satisfaction. Measures directed at tumor-related symptoms are preferable to measures of “larger” psychosocial consequences, as the former are likely to be more sensitive.

Choosing an Appropriate HRQL Instrument

A spectrum of HRQL instruments has been developed for the evaluation of cancer patients. These range from general questionnaires to disease- or treatment-specific measurement scales to ad hoc instruments that are specific to a single study.[5] To balance the trade-offs inherent in the various types of instruments, many oncology experts have proposed using a battery of several HRQL instruments in a clinical trial. For example, a cancer-specific core instrument that measures aspects common to many cancers, such as the general Functional Assessment of Cancer Therapy Scale (FACT-G), may be combined with one or more modules that measure concerns specific to a certain kind of cancer or anticancer treatment, as appropriate.[6]

Although the use of batteries of instruments provides maximal flexibility, this practice can make cross-study comparisons problematic if instruments or modules are not used consistently across various trials. In addition, the multiplicity of outcomes may make it difficult to draw statistical inferences.

It is therefore important for investigators to identify, in a prospective manner, the specific HRQL outcomes that will be assessed as evidence of drug effectiveness, the measurement times that will be considered critical, and the minimal differences in HRQL scores that will be considered clinically meaningful. In some cases, an exploratory effort may be needed to construct specific plans for definitive studies. It is very important to address the issue of whether the instrument is measuring benefit, toxicity, or both.

Minimizing Bias

Every attempt should be undertaken to minimize bias in the conduct of trials that assess HRQL outcomes. Double-blinding is the preferred approach for minimizing bias, and it can be carried out successfully in studies involving oral agents. The Breast Cancer Prevention Trial (P-1) conducted by the National Surgical Adjuvant Breast and Bowel Project, for example, successfully randomized over 13,000 participants to tamoxifen(Drug information on tamoxifen) (Nolvadex) or placebo in double-blind fashion.[7] If double blinding is not feasible, it may nonetheless be possible for study personnel who perform HRQL assessments to be blinded to patients’ treatment assignments and their responses to treatment.

Procedures for the conduct of interviews or patient completion of questionnaires should be standardized across interviewers, patients, and clinical sites. Study personnel should be carefully trained to ensure data quality. Ideally, HRQL outcomes should be measured prior to patients’ discussions with their health care providers regarding treatment response, adverse events, or other concerns that could affect their evaluation of HRQL. Clearly, knowledge of one’s treatment assignment could introduce bias, resulting in expressions of disappointment about having received the standard therapy on the one hand, or heightened expectations about receiving the newer, experimental therapy on the other.

Case in Point: Importance of Pilot-Testing Novel Treatment Concepts

Uncontrolled studies offer researchers the opportunity to gain experience with the use of HRQL instruments prior to large-scale implementation in a randomized, controlled trial. The following example illustrates the importance of testing novel treatment concepts in a pilot trial.

Historically, it has been difficult to demonstrate the benefits of cytotoxic chemotherapy in patients with hormone-refractory prostate cancer using traditional end points, such as objective tumor response or survival. Tumor response evaluation has been particularly problematic for the osteoblastic bone disease that commonly occurs in these patients.[8]

A small, uncontrolled study examined the palliative response to treatment with mitoxantrone(Drug information on mitoxantrone) (Novantrone) plus prednisone(Drug information on prednisone), as measured by changes in pain scores and analgesic scores (the latter derived from data entered in patients’ medication diaries).[8] These symptomatic patients with hormone-refractory prostate cancer experienced a measurable palliative response despite a disappointing objective tumor response to therapy.

The performance of this study allowed critical design issues related to patient selection, choice of pain scales, and pain response definitions to be refined prior to the initiation of a larger randomized trial. Logistical concerns involving patient compliance with serial pain assessments and completion of analgesic diaries were also addressed. In addition, the safety of the treatment regimen in this elderly patient population with comorbid illnesses was assessed.

A confirmatory randomized trial then compared mitoxantrone plus prednisone vs prednisone alone and served as the basis for marketing approval of mitoxantrone for the treatment of pain related to hormone-refractory prostate cancer.[9] Like its predecessor, the randomized trial also evaluated the relief of pain in symptomatic patients, with palliative response defined as a 2-point reduction in pain intensity on a 6-point scale, associated with stable analgesic use and lasting a minimum of 6 weeks. The pain intensity scale was derived from the McGill-Melzack Pain Questionnaire.[10,11]

Combination treatment produced a superior palliative response rate and superior median duration of palliative response with an acceptable safety profile. Median time to progression (defined on the basis of progression of pain, analgesic use, or radiographic end points) was also prolonged in the patients treated with mitoxantrone plus prednisone. Median survival was similar in both arms—not an unexpected finding in this patient population.

Although this trial was not blinded, members of the FDA’s Oncologic Drugs Advisory Committee agreed that a 2-point improvement in pain intensity measured on a 6-point scale was clinically meaningful in this patient population.[12] Furthermore, the magnitude of the effects of the mitoxantrone-prednisone arm on response duration and time to progression could not be readily attributable to patients’ knowledge of their treatment assignment or investigator bias.

Thus, the advisory committee concluded that the trial was an adequate, well-controlled study that provided convincing evidence of net clinical benefit for the combination of mitoxantrone plus prednisone over prednisone alone in patients with hormone-refractory prostate cancer. In 1996, the FDA granted marketing approval for mitoxantrone in combination with corticosteroids as initial chemotherapy for patients with pain related to advanced hormone-refractory prostate cancer.[13]