Download Surrogate End Points

Issue 07/09
July 2009
Surrogate End Points
When evaluating a piece of clinical evidence on a new treatment it is
important to look at the end point, the patient group included, the
comparator treatment and the duration of the study etc. etc.
Many studies just evaluate surrogate end points, as opposed to hard or
clinically meaningful outcomes, which allow our patients to live longer or
better lives.
In studies of prophylactic medical interventions, outcomes may be the disease itself (e.g.
stroke) or some substitute measure considered to be associated with the disease (e.g.
high cholesterol). Such intermediate measures are known as “surrogate end points” or
disease-oriented outcome. A definition of a surrogate endpoint is as follows
“A laboratory measurement or a physical sign used as a substitute for a clinically
meaningful end point that measure directly how a patient feels, functions or survives”.
What are the benefits of surrogate end points?
Clinically meaningful endpoints, also known as “hard outcomes”, or “patient-oriented
outcomes”, are often very time consuming and expensive (but not impossible) to study. In
a stroke prevention randomised controlled trail (RCT), for example, researchers would
have to recruit 10,000 participants and wait for five years for the results. In comparison,
surrogate endpoints are much faster and easier to study: for an RCT on a cholesterol
lowering drug it may suffice to study a few hundred patients over six months. Surrogate
end points are therefore much cheaper to evaluate than hard outcomes.
What are the problems with surrogate end points?
Surrogate end points may be misleading when they do not translate into clinically
important outcomes, or when the clinical outcome is the opposite of what was expected.
The latter was the case with type-I antiarrhythmics in patients who had heart rhythm
disturbances after myocardial infarction. Among these patients it had been shown that
the antiarrhythmic drugs ecainide and flecainide decreased ECG instances of arrhythmiac
(the surrogate end point), which was one of the reasons for their use. However, when
tested in an RCT patients who took ecainide and flecainide were twice as likely to die
from cardiac arrest or other causes than those on placebo1. The drugs improved ECGs but
killed patients !
How should they be used in clinical practice?
Because of this and other examples of failed surrogate end points, it is not advisable to
base treatment decisions on such results alone. This is nevertheless often done in clinical
practice particularly with new drugs that are heavily promoted. Part of the problem is
that the surrogate end point is often confused with the disease itself, rather than being
1
Issue 07/09
July 2009
more properly identified as a risk factor for it. This is further complicated by the fact
that such mixing of the concepts is at times warranted –some surrogate end points may
indeed be part of the disease and a risk factor for it. Blood glucose, for example, is a
surrogate end point as regards late complications of diabetes (e.g. myocardial infarction)
but at the same time high blood glucose in itself may cause a number of symptoms such as
tiredness and increased thirst.
Are they used in sales advertisements?
If hard outcome data are not available, pharmaceutical companies often use surrogate end
points in their sales materials. For example sales adverts for the antidiabetic drug,
sitagliptin refer to the fact that it, amongst other things “enhances incretin” and
“enhances physiological glucose control” 7. Such enhancements may well be of benefit, or
they many not be. No one yet knows the as there is as yet no information on the effect, if
any, of sitagliptin on mortality or diabetic complications. In other words we do not know
whether or not this drug will help our patients live longer or better lives.
Are there other examples of failed surrogate endpoints?
Choosing treatment is a complex decision so it is important not to over simplify it. The
examples listed below include medicines which improved surrogate endpoints but which
were later shown to cause harm.
There are many other examples of treatments that improve surrogate end points but
have not (yet) been shown to improve hard outcomes. These include angiotensin-II
receptor antagonists and the thiazolinediones or glitazones, all widely used locally.
Drug
Torcetrapib
Milronine
Epoprostenol
Metoprolol
Surrogate end point
Increased cholesterol
Increased cardiac
contractility
Increased cardiac
contractility
Reduced Perioperative
Ischaemia
Hard end point
Increased mortality
Increased mortality
Reference
2
3
Increased mortality
4
Increased mortality
5
In summary
Studies based on surrogate end points are common.
Large improvements in surrogate outcomes do not always result in improved hard
outcomes, which allow our patients to live longer or better lives.
Therefore results of such studies should be treated with caution.
Written by: Geraldine O’Riordan, Prescribing Advisor Tel: 01481-732460
References: 1. Edct , DS et al Mortality and morbidity in patients receiving encacide, flecainide or placebo. NEJM 1991,
Mar:324(12) 781-788.2. Barter et al Effects of torcetrapib NEJM 2007, Nov 357 (21) 2109-2122, 3.Packer et al The PROMISE
study , NEJM 1991,Nov 325(21) 1468-1475, 4. The FIRST trial Am Heart J. 1997 Jul 134(1) 44-54 5. Devereaux et al, Effect
of extended-release metoprolol Lancet 2008 May , 371 (9627) 1839-1847 6. Svensson et al A collection of misleading surrogate
end points , University of Gothenburg, Sweden. 7. Richter et al DPP-4 inhibitors for type 2 diabetes. Cochrane Database Syst
Rev 2008 (2) CD006739
2