Surrogate Endpoints: Treating the Number, Not the Patient
Surrogate Endpoints
A surrogate endpoint is a substitute. It is a laboratory measurement (like blood pressure, cholesterol, or tumor size) used to predict a real clinical outcome (like heart attack, death, or pain).
They are faster and cheaper to measure. But they are often a trap.
The Symptom: The "Successful" Failure
"Drug X lowers cholesterol by 20%!" This sounds great. High cholesterol causes heart attacks. So Drug X must prevent heart attacks.
Not necessarily. A drug can fix the number but kill the patient.
The Mechanism: The Broken Link
We assume that A causes B causes C.
- A: Drug
- B: Surrogate (Cholesterol)
- C: Clinical Outcome (Death)
If the drug lowers B, we assume C will get better. But biology is complex. The drug might lower cholesterol (Good) but also cause liver toxicity (Bad).
The CAST Trial Tragedy (1989)
This is the most famous warning in medical history.
- The Problem: People who had irregular heartbeats (PVCs) after a heart attack were more likely to die.
- The Surrogate: Suppressing the irregular heartbeats.
- The Drugs: Encainide and Flecainide. They were perfect at the surrogate. They completely eliminated the irregular beats.
- The Result: The drugs killed more people than the placebo.
- Why? They fixed the rhythm but made the heart weaker in other ways.
If we had approved these drugs based only on the surrogate, we would have killed thousands of people.
Common Surrogates vs. Real Outcomes
| Disease | Surrogate (The Number) | Clinical Outcome (The Patient) | | :---------------- | :--------------------- | :------------------------------------ | | Cancer | Tumor Shrinkage | Survival (Living longer) | | Diabetes | HbA1c (Blood Sugar) | Blindness, Amputation, Kidney Failure | | Alzheimer's | Amyloid Plaque | Memory, Ability to dress oneself | | Heart Disease | LDL Cholesterol | Heart Attack, Stroke |
The Prescription: Ask "So What?"
When you read a medical news story, ask: "Did the patients actually feel better or live longer?"
- Be Skeptical of "Breakthroughs". Most "breakthroughs" are just improvements in a surrogate marker (e.g., "Shrinks tumors in mice!").
- Demand Hard Outcomes. If a drug has been on the market for 5 years, there is no excuse for not having data on survival or quality of life.
- The "Validated" Surrogate. Some surrogates are trustworthy (like Blood Pressure for Stroke). But most are not.
At PaperScores, we downgrade studies that rely solely on unvalidated surrogate endpoints. We care about you, not your blood test results.