The study by Byrne et al1 in this edition of Heart, analysing the distribution of angiographic measures of restenosis after drug-eluting stent (DES) implantation, provides an opportunity for reflection and discussion of the applicability of surrogate end points in the evaluation of coronary stents (see page 1572). A surrogate end point is defined as a biomarker that is intended to substitute for a clinical end point and is expected to predict clinical outcomes based on epidemiological, therapeutic, pathophysiological, or other scientific evidence.2 The concept of using physical signs, laboratory assays, or imaging measures as a substitute for clinical outcomes has a long history and surrogate end points have become frequently used in all branches of medical research. In cardiology, blood pressure, cholesterol and haemoglobin A₁C are all used as surrogates for clinical vascular events, including death, stroke and myocardial infarction.3 They are attractive end points in the design of clinical studies as they allow for smaller sample sizes.

However, not all surrogates perform equally well as substitutes for a clinical end point. The ideal surrogate end point has the following characteristics: (a) a strong statistical correlation with the clinical end point of interest (Prentice criterion No 1)4 5; (b) however, “a correlate does not a surrogate make” and replacement of the clinical outcome requires that “the effect of the intervention on the surrogate predicts the effect on the clinical outcome—a much stronger condition than correlation” (Fleming criterion)6; (c) it captures the net effect of all of the treatment’s mechanisms of action on the clinical end point; thus any treatment difference in clinical outcome should be entirely explained by treatment differences in the surrogate (Prentice criterion No 2)4 5; (d) the magnitude of treatment difference in clinical outcome is …