#192      24 min 22 sec
Diagnostics overkill: Are we screening for cancer too frequently?

Epidemiologist Professor James Hanley scrutinizes the now standard practice of frequent screening for many types of cancer, and suggests optimal testing intervals may mean less screening overall. Presented by science host Dr Shane Huntington.

"We're taking out extra appendices - I don't mind taking out six to help five, but taking out six prostates to help one is something that you've got to think about." -- Prof Jim Hanley




Prof. James Hanley
Professor James Hanley

Prof James Hanley is a biostatistician who began his career with the Eastern Cooperative and Radiation Therapy Oncology Groups while at SUNY/Buffalo and the Dana Farber Cancer Institute/Harvard School of Public Health. He joined McGill in 1980. He has developed statistical methods to evaluate diagnostic tests, and collaborated extensively. His recent work has focused on prostate cancer and cancer screening. He has written several didactic articles on statistical methods, and on the history of epidemiology and statistics.

Publications and reprints

Department of Epidemiology, Biostatistics and Occupational Health at McGill University

Credits

Host: Dr Shane Huntington
Producers: Eric van Bemmel, Kelvin Param
Audio Engineer: Gavin Nebauer
Episode Research: Dr Dyani Lewis
Voiceover: Nerissa Hannink
Series Creators: Kelvin Param & Eric van Bemmel

View Tags  click a tag to find other episodes associated with it.

 Download mp3 (22.3 MB)

VOICEOVER
Welcome to Up Close, the research talk show from the University of Melbourne, Australia.

SHANE HUNTINGTON
I'm Shane Huntington. Thanks for joining us. Cancer screening is often heralded as a major advance in preventative medicine, and screening programs have been running for a number of decades with varying degrees of success.Today there are a myriad of tests that can be performed with the goal of early detection of potentially deadly forms of cancer, but are cancer screening tests always appropriate or beneficial? In recent years especially they have come under increasing scrutiny to work out whether they may be doing more harm than good.Today on Up Close we speak to a biostatistician about the design and interpretation of cancer screening studies. Professor James Hanley is from the Department of Epidemiology, Biostatistics and Occupational Health at McGill University in Montreal, Canada.Welcome to Up Close, Jim.

JIM HANLEY
Thank you.

SHANE HUNTINGTON
Now, Jim, when we talk about all these different types of screening, immediately we think that this is a good thing, that we're screening. 

JIM HANLEY
Absolutely, yes.

SHANE HUNTINGTON
How do we know that that's the case? Is early detection always a good thing? Or are there some sort of adverse consequences of some of these programs?

JIM HANLEY
Well, there are lots of adverse consequences. For some reason I think people think that if you get it early it's got to be better and it ought to work. You know of course, logically, it should. It's the down sides that I think are the problem. If you go to a doctor with a sore throat or a sore chest you're coming in with symptoms and you're asking the doctor for help. It's quite different when a doctor suggests to you - perfectly healthy - that you go and have something that could have a very long cascade of, possibly, negative things. So there's a bigger imperative, I think, to get it right when you're asking people to do something, when they're already healthy, and you might actually make them sick without good reason to.But to go back to what you were saying about it should always help. There is a classic case - a condition in children - called neuroblastoma. There was a screening program for a while in Quebec, where they were trying to find these early, and then there was very serious intervention if they were found. They stopped the trial early because they found, in fact, they were doing more harm than good.What was happening is that these conditions - tumours, whatever you want to call them - were regressing naturally on their own. So that's a danger, that if something can regress on its own, then you'd probably done more harm than good by intervening in the first place. So that's seen as one of the classics now of - you can't automatically assume that it's going to do good.

SHANE HUNTINGTON
When we look at screening versus diagnosis - and these are, obviously, two very different things - how do the ethical implications of these two things compare? Obviously, there's a lot around the way in which information is given to a patient when they're diagnosed with cancer, but how does that function in the screening processes?

JIM HANLEY
To emphasise here, I'm not a physician, but I do know there's one imperative that doctors have. That's, first, do no harm. It's a little different when you're suggesting to somebody maybe you should have a PSA test.

SHANE HUNTINGTON
So what does PSA stand for and how does that test work?

JIM HANLEY
PSA stands for prostate-specific antigen. It's produced by normal prostate. A tumour produces about 10 times as much PSA per cc as the normal prostate. It was seen as something very specific that we should be able to track, but because of the noise from the normal prostate - and the normal prostate grows with age - it's hard to figure it out.The analogy I give is a little different, but it shows what's going on. Apparently, still, appendicitis is a difficult thing to diagnose. Doctors, they don't always get it right. Even in my own family they missed one once in one of our kids. But there the wisdom used to be that you were a good surgeon if, of the six appendices (sic) you took out, five were inflamed and one was white, one was normal. So you had to take out six to help five people.Now, taking out an extra appendix isn't a big deal. You didn't need it in the first place and it's a fairly minor thing. If you're a young person you'll get over it in a few days. Taking out six prostates to help one man not die of cancer is much more serious than taking out one extra appendix in six.I've heard men who said they were sitting around on a Sunday evening with friends and the friends said to them haven’t you had this new PSA test? They said no. Oh, you should have it. It could find cancer early. I heard that man, later, very much regretting that he'd ever taken that advice from his friends because every three months he was going to have another PSA test to be sure the cancer wasn't back. He had no prostate. He was impotent. He was incontinent for a while. So his life was miserable.The biggest problem with it is that you'll never know whether you were the one that needed to be cured or not. Some people do need to be cured, but then you're going to have to intervene on a lot more people in order for the few to benefit.

SHANE HUNTINGTON
So this sort of scenario - how do we determine what will be an effective screening process? I mean this sounds like one that would give back a lot of false positives.

JIM HANLEY
Yes. That was found fairly early on, after the first three or four rounds of using this in the United States in the early '90s, they realised that they were picking up a lot. So right away, just the numbers being picked up was way too many for how many were dying. We were asked by the Quebec Government to say whether we thought the Quebec Government should be paying for PSA tests. I remember one endocrinologist saying to us please pay for the PSA test; we'll save a lot of people. And we said how many? He said look, we will catch 4000 people with this PSA test every year, and the success rates, if you catch it early, are 90 per cent. So 90 per cent of 4000 people, that's 3600 people we're going to save. We had to point out to him that only 1000 men were dying each year of cancer. So you're saving more than were dying in the first place. That means you're not saving them, you're hurting them. So you know how many are dying at the moment without the benefit of screening, or in the absence of screening? So if the numbers being picked up are too high, then you have a problem and you have to somehow hone in on the ones that you think need the treatment  You can still go ahead and detect them, but you now have a problem of sorting them into bad actors and the ones that wouldn't hurt you. Unfortunately, we don't have very good tools for sorting them into bad actors and good actors. That's where the dilemma comes.

SHANE HUNTINGTON
This is Up Close, coming to you from the University of Melbourne, Australia. I'm Shane Huntington. In this episode we're talking about cancer screening studies with biostatistician, Professor James Hanley.Jim, what would be an optimum cancer screening program? How would you construct that so that you don't have this problem of too many false positives?

JIM HANLEY
You've got to have an exquisitely precise instrument for detecting those cancers, and only those cancers, that would prove fatal otherwise - these, and only these, cancers. But I mean that's an impossible prescription.Now, when you think of colon cancer and cervical cancer, the nice thing about colon cancer and cervical cancer is you can get in and actually see the cancer face to face. So that's helping you somewhat. While you're there you can take it out - in a small colon cancer or a polyp. So it also depends on where the cancer is and how invasive the test is.I suspect that that's where cervical cancer is a bit of a success, because it's right there on the surface. With a blood test you haven’t. With mammograms there's a little bit of it in between because you see something suspicious. You've then got a biopsy. One in four, one in five, one in six of the biopsies will be cancer. The others won't be. So a lot depends on the anatomy, how invasive the test is, and then how easy the treatment is. So the treatment itself has to be efficacious and the treatment has to be - it's a bit like the appendix. It's not got to be something that's worse than the cure. 

SHANE HUNTINGTON
How much discrimination is there at the moment between different demographics and the way these screenings are applied? I can imagine - depending on a range of things - whether it's your...

JIM HANLEY
Family history...

SHANE HUNTINGTON
Family history, your age; a range of different factors would also influence the success of a screening program.

JIM HANLEY
That's right, that's right. For example, in breast cancer we have a couple of things: we have age, first of all. Under 50, as I understand it - again, I'm no expert in this area, but as I understand it - under 50 the biggest challenge is, number one, there are fewer cancers to be found under 50 than over 50, so the yield will not be as high.Second, as I understand it, the breast is denser when you're younger, so that goes against you. So I think that's two reasons why the over 50s were more certain about the benefits there, is that there are more of them to be found and the breast is less in the way. In breast cancer your family history and few things that double or triple the risk as well for some people.For prostate cancer there isn't much to go on other than age. In prostate cancer you could make the claim that there's no point in screening past 75 because the cancer wouldn't kill you until you're 95. By that time something else will probably have killed you.But in breast cancer we've just been looking at - again, a lot of the deaths happen in breast cancer in your 60s. Many more women die of breast cancer in their 60s than men die of their prostate cancer in their 60s. So it's very different because you're robbing women of a lot of years of life they could have had otherwise.There are not a lot of demographic things to go on. In lung cancer, yes, there's been a trial completed in lung cancer now. There were a few others earlier on in chest x-ray. Now, we've had one recently, in the United States, on CT screening for lung cancer. There you have the very good discrimination because they only screen smokers. There's a 50 fold risk difference between smokers and non-smokers. There it's getting to be a bit more efficient in that you know where the cancers are and you know where the cancers aren't to begin with, before you start looking for them.
SHANE HUNTINGTON
When we look at some of these recent studies of the screening programs that are around - and we note that there is a degree of, I guess, disappointment in just the efficacy and so forth of those studies. Why has that occurred, given the knowledge that you're talking about is there from the start regarding, I guess, how good these tests are, what the numbers are? Why is there so much disappointment regarding the outcomes of these programs?

JIM HANLEY
I think I have to separate it into two parts. The part I've been mainly concerned with is the benefit side. The down sides are serious in some of these and others - the number of extra people you have to treat and so on - that can be quite large, as, for example, in the PSA. In breast cancer, for example, the disappointment, I think, is because some of the scientists analysing this data from trials have missed an important principle, in that if you think about a screening program that you would have for women - for, say, 50 to 70 - that's 20 years of screening, every year or every second year.So you're doing something regularly for 20 years, but if you look at the trials, the trials themselves have only, maybe, had two or three rounds of screening in them. So there's a very transient deficit or hole made in the cancer deaths - deaths averted or taken away - because of a few rounds of a screening. So when you follow up these women much later, then you're not going to see the benefit because it's already disappeared, or some of them didn't even wait long enough for the benefit to show up.The benefit shows up seven or eight years later maybe, so you've got to have enough data in those years to see and have the benefit expressed. The trials have averaged the benefits over all years, even over the early years when there would be no benefit.Cancer that you find and cure today, that would have killed you otherwise, wouldn't kill you for, maybe, six or seven years. So the benefit is delayed by six to seven years and the trials have ignored that. The example I usually give is to make the contrast. In a lot of medicine - preventative medicine even - when we give HPV vaccine to girls in their teens or early life...

SHANE HUNTINGTON
HPV is?

JIM HANLEY
HPV is the human papillomavirus, the one that causes the cervical cancer and warts as well. Once those girls have three doses of the vaccine, then they're fully immune. The polio vaccine was the same. Once you've had your three doses you're fully protected for many years against HPV infections that are coming at you from the outside. But there you'll see - in the trials we've seen 95 per cent protection already, as soon as the immunity is up to full speed. So right from the starting gate you've got a 95 per cent protection every year. In prostate case the 50 per cent benefit kicks in after about seven years. That's a major difference, but the statisticians and epidemiologists have analysed the data the same way, as though it didn't matter whether it was an immediate benefit or a delayed benefit. They've missed and underestimated the benefit.I like saying if you were going to go on statins to keep your cholesterol down and wanted to know very quickly - as soon as you went on statins - how much did my cholesterol fall, your doctor wouldn't say to you let's come back in a week and we'll test and that's how much you'll fall. You would want to be on them for three or four weeks, and then you stay on them and you keep taking them. Then you'll keep a 50 per cent reduction or 40 per cent reduction going. If you go off statins, or if you test too early, you won't see the full benefit. That's the analogy I make with the screening studies. They haven’t had deep enough studies that screen for long enough.I've been trying to think out why it is. I think, in most trials in medicine, what we test in the trial is, pretty well, what we're going to do later in the program. If I'm going to do the surgery this way, or going to do the radiation this way, or going to give this dose, that's - what's tested in the trial is what's then used in the community. In screening there is no 20 year screening program that's been tested.

SHANE HUNTINGTON
I'm Shane Huntington. My guest today is biostatistician, Professor James Hanley. We're talking about cancer screening here on Up Close, coming to you from the University of Melbourne, Australia.James, you've looked at a lot of this data and you've reanalysed it. What have you found in terms of, I suppose, outcomes where there has been this disappointment, where, in reality, that's not the case? So what screening programs have actually proved to be better than you'd have thought?

JIM HANLEY
I think - well, we knew it from 15 years ago in our own reports in Quebec. There had been cost effectiveness studies done on mammography for breast cancer. The man who was involved with that - and the technology assessments there - said to me it's 40 per cent, Jim, it's 40 per cent. Then, over the last 15 years, we've had report and report reanalysing and merging together all the trials, and they're coming up with 20 per cent.Our own Canadian taskforce, just last November of 2011, said the benefit's only a 19 per cent reduction. We looked at the five trials where we could see what was going on year by year. We could see 40 per cent holes in every one of the five trials, but the hole was only there for as long as they were screening - with a gap, with a lag. So we're saying come on, it's double what you're saying; and, if you continued it, it will stay at double what you're saying. There's no doubt that the down sides of screening are getting bigger because the screening techniques are getting more and more sensitive and picking up more and more cancers that probably wouldn't kill. We haven’t studied that. Other people have studied that. All we're saying is that the reduction is probably 40 per cent when people are saying it's 20. I suspect the reduction won't get any higher, but if we get more and more exquisitely sensitive tests, the numbers of cancers detected will go up. So, you see, there's the trade off. That's what most of the publicity has been about recently, is you've got to tell women the full story; the down side as well as the up side; there will be false positives; we may be treating you when we didn't need to, but we won't be able to know that; we'll never know that. Not every woman who is alive today because of having had breast cancer found and treated needed to have it found and treated. Again, we don't know that. The calculations are quite scary there that, indeed, many women who attribute their being around today to having had breast cancer detected early would still be around today even if they hadn’t.

SHANE HUNTINGTON
Now, tell us a bit about those calculations because when people hear this their immediate reaction, I think, will be somewhat negative, that you would automatically assume you detect cancer in a person, that's a good thing, it allows you to have treatment. What factors go into your calculations to indicate otherwise?

JIM HANLEY
Well, I think part of it is that when people think cancer and hear the word cancer today they're still thinking back to what cancer was like 40 years ago. If you got a diagnosis of cancer 40 years ago it meant it was fatal. I think people are still thinking that way. That number of fatal cancers are still there - unless they're intervened on - but the number of other cancers that are there are now being found when they used not to be found. If you do autopsies on men who died in a car accident at age 75 or age 80, and you look hard, you will find cancer in their prostates, in 50 per cent of them., and it didn't kill them. So they died with it, but not of it.If you now go out searching for those you're going to find what they call indolent cancers: cancer that wouldn't do any damage in the person's lifetime. So that's the problem. It's like the appendices (sic). We're taking out extra appendices - I don't mind taking out six to help five, but taking out six prostates to help one is something that you've got to think about. I'm not even sure, if we take out the six, we'll have the one. We might have half of one or something like that. It's the, I guess, the over-detection or over-diagnosis is the new word for this. It wasn't there 40 years ago. It's the down sides of the technology, is that tests are overly sensitive.

SHANE HUNTINGTON
When we factor that into the way we look as a society, of the prevalence of cancer, do people take this into consideration? We sometimes hear that the number of people with cancer is getting higher...

JIM HANLEY
For sure it is.

SHANE HUNTINGTON
...and people attribute that to diet and environment and other factors. Is it just this testing scenario that, potentially, [causes it]?

JIM HANLEY
Some of it, indeed, is the testing, yes. I mean we saw an increase in prostate cancer diagnoses when men started having operations for an enlarged prostate. It used to be surgical, so they would do something to reduce the prostate. When they did they would take a sample for pathology, and then they'd find cancer there. So even then there was a blip, well before the PSA test came in. There was a blip and it was an artefact of the fact that they were looking more, so that was part of the problem. So a lot of it is the looking and the looking more carefully and the slicing more carefully. We don't have good enough tests then to say this you can leave alone. The boundaries of what's cancer and what's not have changed. I think that's the problem. 

SHANE HUNTINGTON
When you look at the coming years in your research where do you think this is heading in terms of the way this reanalysis of the data will affect the screening programs?

JIM HANLEY
Well, first of all, when we look back at the screening studies in breast cancer, for example, the studies are all now 20 something years old or 30 years old and the technology's changed. I'm not sure that we'll have the resources to go and mount another 20 year study again. More and more, I think, people are going to be watching what happens naturalistically. When a program, for example, in Norway was introduced in breast cancer screening a few years ago, the nice thing was that it was staggered in its entry. Some counties got it before others, so that meant that, later on, you could be able to figure out - tease out, if you like - how much benefit there was because you could compare the counties that had been screen longer for the ones that they're screening.So I think we're going to see more and more of that. Unfortunately, we're probably going to have to let the screening in and then see what it does to the community. It's dangerous because now people are somewhat regretting the huge hype on the PSA screening for prostate cancer. I think people are retreating on that now, and the guidelines have changed. The man who invented or discovered PSA for monitoring cancer is very sorry that he ever did let it go out into the public for screening.So there's a down side to it, but I don't see any other way at the moment than that. By the time a trial is finished - 20 years later - the technology has changed. So we may have to just let these in and watch them. We may have to let them in under controlled circumstances and then have documentation. But time is against us because we won't see the benefit for eight or 10 years; so it's a real Catch 22 in that sense. 

SHANE HUNTINGTON
Jim - just finally - a message to medical professionals and patients on how to deal with this information?

JIM HANLEY
I think doctors have to take their time, before they promote tests, to sit down and talk fully with the patient about what is going to happen. It's not good enough to have a doctor check off - we'll do your PSA this time - and not even have talked to you first about it and what the implications are; because once you get on that treadmill there's no getting off of it. Even doing the first version of it - especially at PSA - because it means the second. Then you'll wonder has the value gone up and whatever. So a full disclosure, I think, or a full information pamphlet on what it's likely to do, how many are likely to benefit and so on. 

SHANE HUNTINGTON
Professor James Hanley from the Department of Epidemiology, Biostatistics and Occupational Health, McGill University in Montreal, Canada, thank you for being our guest on Up Close today and helping us scrutinise the value of cancer screening programs.

JIM HANLEY
Thank you.

SHANE HUNTINGTON
Relevant links, a full transcript and more info on this episode can be found at our website, at upclose.unimelb.edu.au. Up close is a production of the University of Melbourne, Australia.This episode was recorded on 8 March 2012. Our producers for this episode were Kelvin Param and Eric van Bemmel. Audio engineering by Gavin Nebauer. Background research by Dyani Lewis. Up Close is created by Eric van Bemmel and Kelvin Param. I'm Shane Huntington. Until next time, goodbye. 

VOICEOVER
You've been listening to Up Close. We're also on Twitter and Facebook. For more info visit upclose.unimelb.edu.au. Copyright 2012, the University of Melbourne.


show transcript | print transcript | download pdf