How long have I got, Doc?

I once had a rather odd conversation with a patient, which went something like this:

“Tell me, Doc, how long have I got left?”

“I can’t really give you a very good answer to that.”

“Well, put it this way: Do you think I will last the Summer?”

“Yes, I think there’s a reasonably good chance that you will”

“Good. You see, I’m a joiner. Not a carpenter, a joiner. You know the difference?”

“I think so.”

“Well, I reckon I can save my wife a thousand pounds if I make my own coffin.”

Once they have got over the initial shock of their diagnosis, a lot of cancer patients ask about their prognosis, although I’m not always sure what they really want to know, or what they understand by the answers that they are given. I think often they just want to an idea of whether what they have got is going to be immediately fatal, or whether they will be able to go ahead with important plans that they have, such as attending a daughter’s Wedding or going on a family holiday.

Three score years and ten

So what do we mean when we quote survival figures? I think it is helpful to start by looking at life expectancy. Psalm 90 tells us “The days of our years are three score years and ten”, but I prefer to use some more recent statistics. According to the World Bank, the life expectancy at birth of a British baby born in 2015 is 81 years but for a baby born in sub-Saharan africa it is only 60 years. Going back to 1960, the figures are 71 and 40. However, the Global Age Watch index gives the life-expectancy for a 60-year-old living in the UK in 2015 to be 25 years, and in Sudan, Congo, Togo and Ghana it is 17 years. What is going on here? The life expectancy in all of these countries has been going up steadily for decades, so why is a British 60-year-old who was born in 1955 expected to reach the age of 85, whereas a baby born in 1960 might only live to 70. Why is there a 31-year discrepancy between the life-expectancy of an African and a British baby born in 1960, but only an 8-year difference for 60-year-olds?

You may quite rightly object that I am comparing data compiled from different sources which are not looking at exactly the same populations. But the main differences stem from how life expectancy depends on age. Obviously life expectancy is an average figure and for various reasons we use the median rather than the mean (most people think that average and mean are the same thing, but there are several ways of calculating an average as an indicator of what is going on, and the mean is only valid in particular circumstances). The median life expectancy is by definition the age at which half the population we are interested in will be dead, i.e. half of all babies born in the UK in 1960 died before they reached the age of 71, and of course the converse is true, that half of them didn’t. In Africa half of them died before reaching the age of 40, so those surviving to 60 are the lucky minority, who have run the gamut of measles, diarrhoea (still a major killer of children worldwide), malaria and AIDS and come through unscathed. This may be due to a tougher constitution, richer parents, better lifestyle choices or pure luck. Whatever the reason, it is a mathematical inevitability that once you have excluded the people who have died young, the ones that are left can expect to live to a greater age. And the narrowing of the gap between the UK and Africa with increasing age tells us that the major part of the difference between the two countries is due to early deaths, probably in childhood.

So how long have I got, then?

Coming back to cancer patients, as a junior doctor I once worked for an oncologist who loved to use ward rounds as a teaching opportunity:

“Look at this gentleman! He has pancreatic cancer and the survival from that is only two months. I have been giving him chemotherapy now for two years and doesn’t he look well? Don’t ever let anyone tell you that chemotherapy doesn’t work!”

What this particular oncologist didn’t seem to realise is that the two-month survival figure included those who died before they ever had a chance to see them in the first place. It also included the ones who had one cycle of chemotherapy and then died. Those poor souls don’t stick in the memory nearly as well as the lucky fellows who keep coming back for more treatment and so make up the majority of the clinic. Undoubtedly chemotherapy made a huge difference to some of them, but an oncologist relying mainly on personal experience is going to have a rather skewed view of what is going on. On the other hand their patients might feel very encouraged by their optimism (they certainly thought the world of the colleague I am thinking of).

Cancer Registries

Another source of data are published survival figures for different forms and stages of cancer. They can be found through Cancer Research UK and also the National Office of Statistics. The data come from the National Cancer Registry, which relies on coding information collected at hospital attendences, but is nevertheless reasonably accurate and complete, and possibly also from death certificates (I’m not quite sure about this), where the cause of death is recorded by whatever doctor is on hand at the time who has seen the patient at least once while they were alive, and may not be accurate at all (for instance listing the chest infection that killed them rather than the cancer that caused it).

Anybody relying on these statistics to give a prognosis to an individual with cancer needs to be aware of a few pitfalls. For a start, that isn’t really what the data are collected for, and what they really tell us is how mortality has changed over time, and how it compares between different countries. For any such comparison to be possible, the raw figures have to be adjusted to take into account differences in age structure between populations, for instance, to give “standardised mortality rates”, and can end up looking a bit different. Mortality rates also hide the distinction between somebody who has died from their cancer, and somebody who has had a cancer diagnosis and then died of something else – this is particularly affected by age; you shouldn’t expect a high 10-year survival rate in a group of 80-year-olds whatever their state of health.

Then you have to consider the age of the data. 10-year survival figures come from people diagnosed with cancer at least ten years ago. While I was practising as an oncologist I found that if I was treating my patients the same way two years in a row then I wasn’t keeping up with the latest research, let alone a decade. I would hope that survival now has improved further compared with ten years ago but we will have to see…

Less obvious is what happens when you look at the survival for different stages of the same cancer. Here is becomes very misleading to compare recent with older data, the reason being that the methods we use for determining the stage of a cancer are getting steadily more and more sensitive (for instance with improvements in scanner technology). A better scanner means that cancer spread that would have been missed before is now picked up, and the result is that the patient is categorised as being at a more advanced stage at the time of diagnosis. Even with no change in treatment this will improve the survival figures for all stages. Another way of putting it is that if you are found to have a stage 3 cancer today, say, then your disease probably isn’t actually as advanced as in someone with stage 3 cancer ten years ago.

Trials and tribulations

In my experience, oncologists are very fond of quoting data from clinical trials. These tend to be results that they are very familiar with, and are what they are basing their treatment decisions on. A typical example is an important study, codenamed Keynote-189, investigating new forms of treatment for certain types of lung cancer, and was published in May 2018. The study compared chemotherapy alone (platinum followed by pemetrexed) with chemotherapy combined with immunotherapy (pembrolizumab), and it was quite clear that the patients having the combination did much better. The results were reported in terms of progression-free survival, overall 1-year survival and odds ratio for survival, none of which are likely to mean very much to the average cancer patient. The median survival wasn’t even given at all, since in the combination group more than half the patients were still alive at the end of the follow-up period.

While writing this I have just found that an update to the same trial was published online yesterday in the Journal of Clinical Oncology which does give the survival figures – for chemotherapy alone the overal survival was 10.7 months, and 22 months when it was combined with immunotherapy. As an oncologist I regard this as an impressive result, though in practice this sort of thing can cause quite a headache when you return from a conference with an ethical obligation to give your patients the new gold standard of treatment but no funding is in place yet to provide it.

So how does this help the newly-diagnosed patient wondering what will happen to them, remembering that the trial data were collected for the purpose of comparing different treatments, and as such are not entirely valid for other purposes? Well, for a start they are probably going to differ a bit from the average subject in the trial. They might not be the same age, have the same sub-type of cancer, live in the same country or share the same habits. Even if they are reasonably similar, remember that the overall survival is a median figure, which is the point at which half that group have died (with the other half still being alive). Some of those will have died very quickly, perhaps as many as one in ten of them within the first few months. On the other hand, some are going to survive much longer. This is always the case with cancer patients, but not something that everyone realises, and it is seized upon by practitioners of alternative medicine to justify what they do. You can easily find Web sites of private clinics offering dubious and unproven treatments which proudly display testimonials saying things like “My doctor only gave me a few months to live, but I came here instead and I am still alive two years later” (of course you will never find testimonials, or any figures at all, relating to the ones who did badly).

Of course to somebody who is destined to do better than average it doesn’t really matter what happens to the rest, but most of the time we can’t really predict who that will be. What we can do is try not to be too specific. If you tell a patient that their expected survival is a year, then they often won’t understand that this is a median survival and in some way they will subconsciously hold you to it. In nine months’ time they may well believe that they only have another three months left, though in fact their life expectancy is now longer because they haven’t fallen by the wayside already.

To treat or not to treat…

I should just add that you need to bear all this in mind if you are using the data to decide what treatment to have. A survival benefit of a year is quite a lot in oncology trials, and new drugs are regularly approved on the basis of prolonging life by as little as three months. That might not seem very long, particularly when you factor in the side-effects. Well, for one thing, if different treatments are used sequentially, three months here, six months there etc. can add up to a couple of years, which might seem much more worthwhile. More to the point, quoting a survival benefit of three months hides the truth that the treatment won’t help some people at all, whereas others will gain a great deal more from it. To my mind the important thing is to be able to stop it at once if it doesn’t seem to be working so that you can switch to something else.

It is much easier to give a prognosis when you have seen how an individual is over a period of time. If someone is obviously deteriorating, then how quickly this is happening can be very informative, and this forms the basis for an oncologists’ rule of thumb (note that it doesn’t apply to other diseases). If they are going downhill over a matter of months, then months is probably what they have left. If they are visibly worse over a few weeks, then weeks. Or days, as the case may be.

I think really what many people want is hope. If they know that it is possible to live a long time with their cancer, even if only a minority achieve it, then maybe, just maybe, it is better than the death sentence they thought they were being given at first.