Hepatology may have problems with putative surrogate outcome measures

Abstract

A surrogate outcome measure is a laboratory measurement, a physical sign, or another intermediate substitute that is able to predict an intervention's effect on a clinically meaningful outcome. A clinical outcome detects how a patient feels, functions, or survives. Surrogate outcome measures occur faster or more often, are cheaper, and/or are less invasively achieved than the clinical outcome. In practice, validation is surprisingly often overlooked, especially if a biologic plausible rationale is proposed. Surrogate outcomes must be validated before use. The first step in validation is to demonstrate a correlation between the putative surrogate and the clinical outcome, e.g., the higher the surrogate the shorter time to death. However, a correlation is not sufficient to validate the surrogate. The second step is to establish if the intervention's effect on the surrogate outcome accurately predicts the intervention's effect on the clinical outcome. In hepatology a number of putative surrogate outcomes are used both in clinical research and in clinical practice without having been properly validated. Sustained virological response to interferons and ribavirin in patients with chronic hepatitis C, serum bilirubin concentration following ursodeoxycholic acid or immunosuppressants for patients with primary biliary cirrhosis, and nutritional outcomes following artificial nutrition for liver patients may not be valid surrogates for morbidity or mortality. The challenge is to develop reliable surrogates, both to facilitate the development of new interventions and to ensure our patients and us that these interventions are effective clinically.