If you have started reading about your cancer, you have almost certainly run into a wall of initials: OS, PFS, ORR, DFS, HR. A news story says a new treatment “improved survival.” Your oncologist mentions a trial. And you are left holding the only question that really matters: will this help me live longer, and live better? I want to walk you through what these endpoints actually measure, in plain language, because once you understand them you can read the same headline your neighbour read and understand far more than they do.
I will be honest with you throughout. Some of these measures genuinely tell you about living longer. Some tell you only that a tumour got smaller or stopped growing for a while — which can sound like the same thing but is not. Knowing the difference is one of the most useful things you can carry into a conversation with your team, and it can spare you a great deal of unnecessary fear.
What an “endpoint” even is
An endpoint is simply the yardstick a clinical trial uses to decide whether a treatment worked. Before a trial begins, researchers must state exactly what they will measure — you cannot judge a race without agreeing on the finish line first. Different endpoints answer different questions, and this is where confusion creeps in, because the very same drug can look impressive on one yardstick and unremarkable on another. When you understand which yardstick a headline is using, half the confusion disappears.
The two you will hear most often are overall survival and progression-free survival. Let me take them one at a time, then build out from there.
Overall survival (OS): the gold standard
Overall survival measures one thing, cleanly and without argument: how long people live after starting a treatment (or after being diagnosed), counting from that point until death from any cause. If a trial shows that a new drug improves overall survival, it means the people who received it lived longer, on average, than those who did not.
This is why researchers call OS the gold standard endpoint. It cannot be gamed. It does not depend on how a scan is read or when it happens to be taken. Death is unambiguous. When you see that a treatment improved overall survival, you are looking at the most meaningful result a cancer trial can produce.
Overall survival is the only endpoint that directly answers “did people live longer?” Everything else is, at best, a clue pointing toward that answer.
But OS has real drawbacks, and understanding them tells you why the other endpoints exist at all:
- It takes years to measure. To know whether people lived longer, you have to wait for enough time to pass. For many cancers that means five, ten, or more years. A drug that helps could sit unused for a long time while we wait for the survival data to mature.
- Later treatments blur the picture. After a trial drug stops working, people go on to other therapies, and those later treatments also affect how long someone lives. This can muddy whether the original drug was responsible for a survival difference — a problem researchers call “crossover” or “post-progression therapy.”
- “From any cause” cuts both ways. OS counts every death, including a car accident or a heart attack unrelated to the cancer. In a large trial this usually evens out between the groups, but it is worth knowing that OS is not purely a measure of cancer control.
Progression-free survival (PFS): faster, but a step removed
Progression-free survival measures how long a person lives without their cancer getting worse. “Progression” means the cancer has grown, spread to new places, or otherwise advanced — usually judged by scans (CT, MRI, PET) measured against a standardised ruler radiologists use, most commonly a set of rules called RECIST. The clock starts at treatment and stops at the first sign of progression or death, whichever comes first.
So PFS bundles together two good outcomes: staying alive, and keeping the cancer stable. Its great advantage is time. You can measure progression far sooner than you can measure death, because a scan showing tumour growth arrives long before the end of someone’s life. This lets promising drugs reach patients years earlier than they otherwise could.
Here is the crucial part, and I want to say it plainly: a longer PFS does not automatically mean people live longer. It is entirely possible for a drug to delay the cancer’s growth on scans without adding time to a person’s life. Sometimes the two go hand in hand; sometimes they do not. This is not a technicality — it is one of the most important ideas in modern oncology, and it is why you should never read “improved PFS” as though it said “improved OS.”
Why PFS and OS can disagree
- The cancer may be delayed for a few months, then return and behave just as aggressively, so the final survival time is unchanged.
- The treatment’s side effects may offset the benefit of slower growth, harming quality or length of life in other ways.
- People in the comparison group may go on to receive the same drug (or a similar one) later, catching up on survival even though their cancer progressed sooner.
- “Progression” is defined by scan measurements, which carry some subjectivity — a genuinely meaningful delay and a marginal one can look similar on paper.
None of this makes PFS worthless. A meaningful delay in progression can mean more months of feeling well, fewer symptoms, and more time before switching to a harder treatment. Those are real goods, and for many people they matter enormously. Just do not mistake them for proof of a longer life.
Response rate (ORR): did the tumour shrink?
Overall response rate (ORR, or simply “response rate”) is the percentage of people whose tumours shrank by a defined amount after treatment. A complete response means the visible cancer disappeared on scans; a partial response means it shrank significantly (conventionally by at least about 30 percent). Add those together and you get the response rate.
Response rate is the fastest signal of all — shrinkage can often be seen within weeks. That makes it valuable early in a drug’s development, and it is frequently how the very first news about an exciting new therapy reaches the public.
But response rate is also the most limited of these measures, and I want you to hold it lightly:
- It says nothing about time. A tumour can shrink impressively and then start growing again a month later. Response rate alone does not tell you how long the benefit lasts — for that you need duration of response, which is often reported alongside it and is just as important.
- Shrinkage is not the same as living longer. As with PFS, a tumour getting smaller does not guarantee more months of life. It is a hopeful sign, not a promise.
- A single-arm number lacks context. “Seventy percent of patients responded” sounds dramatic, but without a comparison group you cannot know how those same people would have fared on standard treatment.
The other initials you may meet
A few more endpoints turn up often enough to be worth a plain definition:
- Disease-free survival (DFS) — used mainly after surgery or treatment intended to cure. It measures how long someone stays free of any detectable cancer. You will see this in trials of early-stage disease.
- Recurrence-free or relapse-free survival — very similar to DFS: how long before the cancer comes back.
- Time to progression (TTP) — like PFS, but it counts only cancer growth, not deaths.
- Quality of life and patient-reported outcomes — direct measures of how people actually feel and function: pain, fatigue, ability to do daily things. These matter enormously and, in my view, are still under-reported. When you see them included in a trial, take them seriously; they describe the part of “living better” that no scan can capture.
Hazard ratios and “a 30% reduction”: reading the number behind the claim
Sooner or later you will meet the hazard ratio (HR), and it trips up almost everyone, so let me demystify it. A hazard ratio compares the rate of an event — death, say, or progression — between the treatment group and the comparison group over the course of the trial. An HR of 1.0 means no difference between them. An HR below 1.0 favours the new treatment; an HR above 1.0 means it did worse.
You will often see an HR translated into a percentage. An HR of 0.70 is frequently reported as “a 30 percent reduction in the risk of death.” That is technically correct, but it is easy to misread. It does not mean 30 percent of people are cured, or that everyone lives 30 percent longer. It is a statement about the relative rate of the event across the group at any given moment. The real-world payoff — how many extra months or years, and for how many people — depends on the underlying numbers, and a striking-sounding percentage can sit on top of a difference of only a few weeks. Whenever you see a percentage reduction, look for the absolute figures underneath it: the actual median survival in each group, and the size of the gap between them.
Statistical significance and confidence intervals
Two more phrases are worth a moment. When a result is called statistically significant, that means only one thing: the difference is unlikely to be due to pure chance. It says nothing about whether the difference is large or meaningful to your life. A two-week improvement can be statistically significant in a big trial and still change very little for a person.
You may also see a confidence interval — a range, such as “HR 0.70 (95% CI 0.55–0.89).” Think of it as the trial’s honesty about its own uncertainty: the true effect probably lies somewhere in that range. A narrow range suggests a more precise estimate; a wide one, more uncertainty. If the range for a hazard ratio crosses 1.0, the trial has not clearly shown a benefit. You do not need to calculate anything — you just need to know that the single headline number is an estimate with a margin around it, not a fixed fact.
Surrogate endpoints: the shortcut, and its risk
PFS, response rate, and disease-free survival are often called surrogate endpoints. A surrogate is a stand-in — a measure we hope predicts the outcome we truly care about (living longer or better) but which is faster and easier to observe. The logic is sound on its face: if we waited for overall survival on every drug, patients would wait years for treatments that might have helped them today.
That is a genuinely reasonable trade-off, and surrogates have brought good drugs to people sooner. But a surrogate is a bet, not a certainty. For some cancers and some treatments, a surrogate like PFS reliably tracks with overall survival. For others, the link is weak — a drug improves the surrogate but never delivers the longer life it seemed to promise. Regulators sometimes approve a drug on a surrogate endpoint (through pathways such as “accelerated approval” in the United States, or conditional approval elsewhere) on the understanding that survival data will be confirmed later. Usually it is; occasionally the later data does not confirm the benefit, and the approval is withdrawn or narrowed. This is the system working as intended, not a scandal — but it is why the exact wording of a headline matters.
When you read that a treatment was approved or celebrated on the basis of progression-free survival or response rate, the honest translation is: “This looks promising and may well help, but we do not yet have proof that it lets people live longer.”
Putting it together: one drug, three headlines
Imagine a single new drug for advanced cancer, and three different news stories about it. The first says, “New drug shrinks tumours in 65 percent of patients” — that is response rate, the earliest and thinnest signal. The second, a year later, says, “New drug delays cancer growth by four months” — that is progression-free survival, more substantial but still a surrogate. The third, years on, says, “New drug helps patients live three months longer” — that is overall survival, the result that actually settles the question.
All three headlines can describe the same drug at different stages of its evidence. None of them is dishonest. But they are not equal, and a reader who cannot tell them apart may feel a surge of hope at the first that the third does not fully justify — or may dismiss a genuinely useful drug because its survival benefit sounds modest. Knowing which endpoint you are looking at lets you feel the right amount of hope, which is a kind of protection.
How to read a headline or a statistic without fear
Here is the practical toolkit I would want a member of my own family to have. When you see a claim about a treatment, ask:
- Which endpoint is this? Overall survival is the strongest. PFS, response rate, and DFS are surrogates — meaningful, but a step removed from “living longer.”
- How big was the difference, in real terms? “Improved survival” can mean anything from a few extra weeks to several years. Look for the actual numbers — months of median improvement, the absolute figures — not just the word “significant” or a percentage on its own.
- Statistically significant is not the same as meaningful. Significance only means the result is probably not chance. Ask what the difference means for a person, not just for the statistics.
- Compared to what? A number in isolation (“60 percent responded”) tells you little without a comparison group and the standard-of-care context.
- Who was in the trial? The people studied may differ from you in age, cancer stage, biology, or overall health. Results from a trial of one group may not transfer neatly to another.
A word on the word “median”
Almost every survival number you meet is a median — the middle point, where half of people did better and half did worse. A “median overall survival of 18 months” does not mean you have 18 months. It means that in the group studied, half were still alive after 18 months and half were not — and some lived far longer, sometimes years beyond the median. A median is a description of a crowd, not a prediction about you. This is worth repeating to yourself whenever a number lands hard: it is a fact about a group of strangers from the past, not a sentence passed on your future.
Where population survival statistics fit in
You will also encounter figures like “five-year relative survival.” These deserve their own careful handling, because they frighten people more than almost anything else — and often unnecessarily.
- They are population averages, drawn from large groups of people diagnosed years ago. They describe groups, not individuals.
- Because they look back in time, they often lag behind today’s treatments. Someone counted in a five-year survival figure may have been diagnosed before the newest immunotherapy or targeted drugs existed. For many cancers, current outcomes are genuinely better than the historical numbers suggest.
- “Relative survival” compares people with the cancer to similar people without it, to separate the effect of the cancer itself from ordinary causes of death. It is a way of isolating the cancer’s impact, not a raw headcount.
- They vary enormously by cancer type, stage, tumour biology, and the individual person. A single headline number can hide vast differences between one situation and another.
So when you see a survival percentage, please read it as a rough map of the territory, not a verdict on your journey. It cannot know your specific cancer, your treatment, or how your particular disease will respond. Your oncologist, who does know those things, will always be a better guide than a national average.
Common myths worth clearing up
- “The drug improved survival, so it will help me live longer.” Only if “survival” here means overall survival — and even then, it is an average benefit across a group, not a guarantee for one person.
- “My tumour shrank, so I’m cured.” Response (shrinkage) is encouraging but is not the same as cure, or even as durable control. Ask about the duration of response.
- “PFS improved but OS didn’t, so the drug is useless.” Not necessarily. More time without progression can mean more months of feeling well and a delay before harder treatment — real benefits, even without a proven survival gain. It depends on the size of the benefit and its cost in side effects.
- “The five-year survival rate is my prognosis.” It is a group average from the past, not a personal forecast.
- “A statistically significant result is a big result.” It only means “probably not chance.” The size and the meaning are separate questions.
- “A 30 percent reduction in risk means I’m 30 percent more likely to be cured.” No — a hazard ratio describes the relative rate of an event across a group, not your personal odds of cure.
What to ask your own oncology team
These endpoints are not abstract when a treatment is being offered to you. Here are questions I would encourage you to bring to your team. Write them down before your appointment — anxiety makes us forget the things we most wanted to ask, and there is no shame in reading from a list.
- “What is the main benefit this treatment is expected to give me — longer life, longer time before the cancer grows, fewer symptoms, or something else?”
- “Is that benefit based on overall survival, or on a surrogate like progression-free survival or response rate?”
- “In plain numbers, how much benefit are we talking about, on average?”
- “What are the likely side effects, and how might they affect how I feel day to day?”
- “How will we know if it is working, and how soon?”
- “If this doesn’t work, what options come next?”
- “How do people like me — my age, my stage, my health — tend to do, as best we can tell?”
There are no foolish questions here. A good team will welcome them, and asking makes you a genuine partner in the decision rather than a passenger being carried along by it.
When to contact your team
Understanding endpoints is about the long view. But if you are on treatment now, some things need attention sooner rather than later. Contact your oncology team promptly if you notice:
- New or worsening symptoms — increasing pain, breathlessness, or unexplained weight loss.
- Signs your treatment may not be well tolerated — persistent vomiting, an inability to keep fluids down, or severe, unrelenting fatigue.
And treat these as urgent — call your team’s emergency line or seek emergency care — if you have:
- A fever (many teams use 38°C / 100.4°F as the threshold), chills, or shaking, especially during chemotherapy — this can signal a serious infection and is a medical emergency, even if you otherwise feel well.
- Sudden shortness of breath, chest pain, new confusion, a severe headache, or any bleeding that will not stop.
These are general signposts, not a substitute for the specific instructions your own team has given you. If they told you something different — a different number to call, a different temperature threshold — follow theirs.
What to take from this
- Overall survival (OS) is the gold standard — it directly measures whether people lived longer, and it is the strongest evidence you can have.
- Progression-free survival (PFS) measures time without the cancer worsening. It is faster to obtain but does not guarantee a longer life — though a meaningful delay can still mean more good months.
- Response rate measures whether tumours shrank. It is the quickest signal but the most limited — always ask how long the response lasts.
- Surrogate endpoints (PFS, response rate, DFS) are useful stand-ins that bring drugs to patients sooner, but they are a bet on benefit, not proof of it.
- Hazard ratios and percentage “risk reductions” describe a group’s rate of an event, not your personal odds — look for the absolute numbers underneath them.
- Almost every figure is a median — a middle point for a group, with many people doing better — and a prediction about a crowd, never about you.
- Population survival statistics are group averages from the past and often understate what today’s treatments can do.
- When reading any claim, ask: which endpoint, how big, compared to what, and in people like me?
You do not need to become a statistician. You only need enough understanding to ask better questions and to worry about the right things. The numbers describe groups; your team treats you. Take what you have read here back to your own oncologist — they know your cancer, your body, and your options in a way no article ever can, and they are the right people to help you turn these ideas into the decision that is right for you.