What Are Surrogate Markers?

A few weeks ago, I was chatting with a longtime patient about why it is so hard to get good data regarding the possible effect of hormone therapy on dementia risk. We discussed the logistical challenges of treating women with hormone therapy in their 50s and following them for decades until they develop dementia in large enough numbers to see if the hormone therapy has an effect.

She laughed at that idea and remarked that it was too bad there was nothing you could measure that would tell you someone was going to get dementia in the future — essentially, to assess someone’s risk. This type of situation comes up in the literature about menopause and hormone therapy often. We would love to have a marker that would tell us, if we give hormone therapy now, what effect it will have decades into the future. In some cases, we do have this kind of test, called a surrogate marker.

Take me to the bottom line

Why is good data so hard to come by?

Sometimes a clinical trial is trying to answer a question with a very immediate answer. For example, we want to know if hormone therapy improves sleep disruption in perimenopausal women. We can randomize study participants into two groups. One group gets hormone therapy, and the other gets a placebo. We ask the participants about their sleep before and after they start the treatment. If we are really fancy, maybe we also do a sleep study on the participants before and after the medication. Then we look to see if the group that received the hormone therapy slept better after treatment compared with the placebo group.

That entire study can be performed over a few months. And sleep disruption is common in perimenopause, so we don’t need to enroll very many women to see an effect.

Other times we want to understand how giving a treatment now affects health well into the future. Let’s consider the example of heart disease in menopausal women. We want to know if treating women with hormone therapy decreases their risk of heart disease. This is more complicated, because heart disease is not very common in women in their 50s. We really want to know if giving hormone therapy prevents heart disease when women are at higher risk for it in their 60s, 70s, and beyond.

So we have three choices. First, we can give hormone therapy to older women who have a higher risk of heart disease — this was the route the Women’s Health Initiative study took in the 1990s. What we learned from that trial is that giving women hormone therapy in their 60s and 70s increases their risk of heart disease. It turns out we were probably asking the wrong question.

Alternatively, we can give women in their 50s hormone therapy and follow them for many decades. This is challenging and expensive — it is really hard to keep track of study participants over 20 or 30 years. Or we can give women in their 50s hormone therapy and look for signs that suggest they will develop heart disease later.

What is a surrogate marker?

Those signs that suggest that someone will get heart disease later are surrogate markers. Lots of areas of medicine use surrogate markers, but some use them more than others. Cardiology is an area of medicine that uses them a lot.

Rather than wait around for people to have heart attacks to know if a medication is working, researchers will use ultrasound to look at the carotid arteries that carry blood to the brain. When the walls of those arteries are thickened, it is a sign that the body has started to change in ways that lead to heart disease later. If a treatment prevents or reverses that thickening, it is likely also to keep you from developing heart disease. And we can see changes in the walls of the arteries change much more quickly than waiting for people to have cardiac events.

When are surrogate markers useful?

Surrogate markers are especially useful when we want to know how a treatment will affect a disease far into the future. In our example about hormone therapy and heart disease, using a surrogate marker might be a great way to get a better understanding of the role of estrogen and developing heart disease.

Surrogate markers can also be a great way to start to build a case for a treatment. It is faster and cheaper to do a clinical trial using surrogate markers to explore a new idea. Then, if the study using the surrogate marker is positive, researchers can move on to a larger, longer trial to confirm the results.

When should we be wary of surrogate markers?

Not all surrogate markers are created equal. When researchers design a study, they need to choose a marker that has been shown to be strongly connected to the real outcome they want to study. Measuring the thickness of the carotid artery walls has been shown in a whole body of medical literature to accurately predict future risk of heart disease. Other surrogate markers are less reliable or less specific.

Another commonly used surrogate marker is creatine phosphokinase (CPK). CPK is easy to measure in the blood and it changes quickly, but it is much less specific. It is found in the heart, brain, and muscle tissues and signals that there is damage or injury in those tissues. That damage could be caused by reduced blood flow to the heart or brain, suggesting heart disease, or it could be caused by something else. As a result, CPK is a less reliable surrogate marker.

There are also some diseases we would like to study using surrogate markers but for which no good markers exist. Researchers would love to have a simple, reliable marker for dementia. But thus far, reliable surrogate markers have been elusive. So we have to wait until people have cognitive changes that can be measured with tests. To see if giving hormone therapy in your 50s prevents dementia, we still need to wait decades to get an answer.

The bottom line

Studying how treatments affect disease many years after they are given — such as how taking hormone therapy affects a woman’s risk of heart disease later in life — is very challenging.
Surrogate markers are validated tests that reliably indicate that someone has or is at risk for a disease.
Using surrogate markers can allow researchers to study how a treatment affects a disease like heart disease more cheaply and more quickly than waiting for participants to have an event.
Not all surrogate markers are reliable, and not all diseases can be studied using surrogate markers.