AACR 102nd Annual Meeting, 2—6 April 2011, Orlando, Florida

Blood test for lung cancer in never-smokers

Dr Charlie Birse (Celera Corporation, California, USA)

We continue now, here in Orlando, with Dr Charlie Birse. Charlie, you’ve been addressing the difficult problem of detecting lung cancer early and specifically we have an increasing population of never-smokers who have lung cancer. What is it that you’ve been doing?

Basically we have been trying to develop a blood-based test that can detect lung cancer in the broad smoking population but also in never-smokers. And you’re correct, the proportion of never-smokers is expected to increase in coming years as smoking cessation and prevention programmes start to be successfully introduced.

What are the big issues here? Because mass radiography, simple X-rays, prove to be no help really, then CT is now looking interesting but you’re swamped by the numbers of people you need to look at, aren’t you?

Sure. The results coming from the national lung screening trial look very exciting. It looks for the first time as though we do have an imaging approach which does reduce lung cancer mortality. That’s a great step forward for the community, however, there are issues still with CT scanning and most of them revolve around the number of false positives that are resolved with CT scanning. Our test, we hope to be able to use as a compliment to CT scanning and to quickly and in a non-invasive fashion resolve which of those nodules that you find in your chest are malignant and which are benign.

So your test uses a biomarker, tell us what it is.

The biomarkers are natural proteins that are found in the bloodstream. We’ve taken an innovative approach to try and identify markers going back to the tumour tissues themselves and asking which markers are found in the tissues and then asking are they released into the bloodstream and could act as biomarkers.

Now this was a bit of a needle in a haystack job as well; you had to look at a lot of potential markers and you boiled it down to quite a small number.

That’s right. Initially we identified more than 500 candidate markers and we’ve selected only the kind of cream of the crop to move forward in our validation studies. So we’re encouraged by our initial results but we still have some other candidates that we could consider, depending on how things go.

So now you’ve got nine biomarkers and a model for relating it to lung cancer?

That’s right. So we actually took nine biomarkers through a series of validation studies and we’ve actually selected a six marker model which performs rather well. So those six markers seem to complement each other rather well in detecting most of the lung cancer cases.

And can you explain how you’ve validated this because you’ve had to look at patients and cases and controls and work out whether it’s specific and sensitive?

Exactly. So what we’ve done is we’ve actually undertaken four independent case control studies and in those case control studies we’re basically collecting cases, which are lung cancer samples, and controls, which are samples collected from individuals who don’t have cancer but are relevant controls. So they might be smokers or they might have a benign lesion in their chest.

And did you look at the difference between smokers who have lung cancer caused by smoking and never-smokers who, nevertheless, have lung cancer?

That’s right. In our most recent study we have extended the analysis to evaluate the performance of our test in a never-smoking population. We were encouraged to find that the test works rather well in the never-smoking population as well. Given the potential increase in relative numbers in that never-smoking population, it’s important to know how well the test will perform there.

Give us some ballpark figures of how well it performed in the never-smokers.

In the never-smokers we were seeing a sensitivity of around 68% whilst maintaining a very high specificity of 95%.

So typically, which never-smokers are you looking at and how valuable might this be in a practical situation?

In terms of the never-smokers we were looking at all types of never-smokers that spanned all stages of disease and we’re looking at the key histological type. Never-smokers often present as adenocarcinomas, one of the key histological types found in non-small cell lung cancer, so we biased our study to include many adenocarcinomas.

If you were using this, however, as an adjunct to screening of individuals thought to be at high risk, how would you then use it and how well would it perform?

If we were to use it post-CT, so after a nodule had been identified, we’re expecting to use it with that very high specificity. We want to be absolutely sure that we can tell individuals that the lesion in their chest is actually benign so they’re not going to have the concern leaving the clinic that they have a lump in their chest and it may be a malignant lesion.

Can you summarise how doctors would use this test in practice?

In today’s practice many of the nodules that are identified in the chest are identified incidentally. It’s a mistake that they’re identified but they say, ‘Lo and behold, you have a nodule in your chest. We need to quickly and accurately distinguish whether this nodule is malignant or benign.’ That would be where our test would come in and the clinician could quickly refer the patient to undertake our test and, within a matter of days, have a description of whether they have a malignant or benign lesion.

Is there a possibility of using such a biomarker test to look at the population in general - patients or people who are not ill?

That is a possibility and certainly something that we would consider for the future. The real issue with developing that kind of test is that we would have to have extremely high performance, in particular on the specificity side it would have to be very, very specific so that not too many false positives would be identified. And we’re some ways away from that right now.

Well could you then, finally, sum up in a few words where we are at for screening using biomarkers for lung cancer in smokers and never-smokers?

I think we’re at quite an exciting place, actually, today with breaking news coming from the CT trial where imaging is, for the first time, being able to help us in detecting early lung cancer. We, and others, are developing novel complimentary tests that are going to improve the overall performance in that screening process. I think the future is looking fairly encouraging, going forward.

Charlie Birse, thank you very much for coming here from California. It’s good to talk with you here at the American Association for Cancer Research on ecancer.tv.

It’s been a pleasure, thank you.