Optimal design of clinical outcome studies in nutrition and cancer: future directions.

Abstract

Many studies have assessed the benefits of nutrition support in cancer patients. Except for studies directed at the treatment of severe malnutrition, most clinical trials have failed. Although prospective randomized controlled clinical trials (Phase III) remain the most reliable means of evaluating the efficacy of therapy, the available literature reports only results from small trials (Phase II to III), most of which appear to be contradictory and none of which conclusively answer the question being considered. To address this gap in knowledge, tools such as meta-analysis have been adapted from the field of statistics. Meta-analysis involves pooling results across several studies and provides a more precise estimate of treatment effect than can each individual study. However, clinical trials selected for meta-analysis, although broadly similar, can differ significantly in terms of therapies used and clinical populations studied. Major cancer types with differing effects on food intake and malnutrition (eg, the mechanical obstruction in head and neck cancer vs the cytokine-induced metastases associated with lung, ovarian, colon, and breast cancer) cannot be subject to the same analytic criteria. In this paper, the current state of clinical outcome trials in nutrition and cancer is examined, and the desired design for future studies is proposed. Research priorities include the conduct of Phase II clinical trials that use as outcome measures quality of life, performance status, and survival to identify optimal cancer-specific and patient-specific nutrition support. The next round of Phase III efficacy studies should establish the appropriate use of nutrition support in cancer therapy.