select search filters
briefings
roundups & rapid reactions
factsheets & briefing notes
before the headlines
Fiona fox's blog

expert reaction to conference abstract looking at blood autoantibody test and CT imaging, and lung cancer

A conference abstract, presented at the IASLC (International Association for the Study of Lung Cancer) 2019 World Conference on Lung Cancer, reports on a new test for the early detection of lung cancer.

 

Dr Jeanette Dickson, President of The Royal College of Radiologists and clinical oncologist specialising in lung cancer, said:

“Lung cancer is a disease with very poor survival rates.  Any way to improve early detection is to be welcomed, as those patients with smaller cancers do better.

“The preliminary results of this study are extremely interesting and give the promise of developing stratified lung cancer screening to improve patient outcomes, as well as being more cost effective for healthcare in the UK.

Does the press release accurately reflect the science, as far as you can tell from the available information?

“Yes – given the fact that it is an abstract not a peer reviewed paper.

Is this good quality research, as far as you can tell?  Is there enough info to be able to judge whether the conclusions backed up by solid data?

“Yes, it is a good quality study in an area of unmet need, however, it is an interim analysis.  The abstract is not peer reviewed.

How does this work fit with the existing evidence?

“It shows good evidence that LDCT lung screening impacts on mortality if used in the correct population.  This test gives a promise of better patient selection, which will reduce number of unnecessary tests and reduce unnecessary radiation exposure.

Have the authors accounted for confounders?  Are there important limitations to be aware of?

“This is difficult to assess in an abstract and the study was not powered to show reduction in mortality.

What are the implications in the real world?  Is there any overspeculation?

“This study projects the best case scenario but it is not unreasonable and fits with the future vision for stratified lung screening.”

 

Dr Heather Williams, Principal Medical Physicist, Nuclear Medicine, The Christie, said:

“This research involved over 12000 patients at high risk of lung cancer (because they are heavy smokers) having a blood test, which is designed to pick up signals given off by lung cancer.  If the test was positive, the patients were sent for X-rays and non-contrast CT scans to look for lung cancers.  The blood test shows promise, as the research indicates that it could lead to lung cancers being detected earlier, making it easier for more disease to be removed, so that patients have a greater chance of living longer after treatment.  However, more research is needed to prove that having this blood test would mean that lung cancer patients live longer, and to prove that the blood test really does pick out those patients who have lung cancer more often than just sending all heavy smokers to have X-rays and CT scans.

“It is also surprising that the authors have chosen to use non-contrast CT to determine whether there is abnormality within the lungs.  Lung cancer is normally diagnosed using a range of tests, including CT imaging with contrast, which is much more sensitive as it shows up the higher blood flow associated with even small tumours.

“Ultimately, the risks and benefits of this new technique need to be evaluated in relation to the combination of tests typically used to diagnose and assess lung cancer, not just non-contrast CT, to prove that having the blood test really does pick out those patients who actually have lung cancer from those who are most likely to have it because of their smoking habits.”

 

Prof Paul Pharoah, Professor of Cancer Epidemiology, University of Cambridge, said:

“In summary, this study has shown that this blood test might be effective as a test for early detection of lung cancer that might reduce lung cancer mortality.  Or it might not.

“The data are promising but much larger studies with longer follow up are needed before we can know whether or not it is an effective test for early detection.  Some key points are:

“1. This is just an abstract presented at a conference and it has not been peer reviewed and not all the relevant information needed to evaluate the study is presented.

“2. Oncimmune are the company who developed and own the test.  There are some statements in the Oncimmune press release are not supported by any data: “Further validates the use of Oncimmune’s platform technology as a screening modality, which can detect cancer four years or more before standard clinical diagnosis” – no data on the timing of diagnosis are presented.  And “The study also showed a lower rate of deaths among people in the intervention arm of the study after two years compared with people in the control group. Lung cancer-specific deaths were also lower in the intervention group.” – data on deaths are not presented in the abstract, so we cannot judge this statement.

“3. The IASLC press release states that there was a non-significant trend suggesting fewer deaths in the intervention group.  But a small, non-significant difference could either mean that there is a true difference or that there is no real difference.  The data are insufficient to judge.

“4. It is worth noting that there is actually a greater number of cancers detected in the control group (71 vs 56).  Probably not significant either, but an effective early detection test would be expected to reduce the rate of late stage disease AND increase the rate of early disease.  In this study the rate of early stage disease was the same in both groups.”

 

Prof Kevin McConway, Emeritus Professor of Applied Statistics, The Open University, said:

“It really is a mug’s game trying to make a careful assessment of a study like this, where no full research paper is yet available, and the study has not yet gone through full peer review by other experts in the field.  One has only partial information about what was done and what was found.  There may be odd-looking features of the data that eventually turn out to have a perfectly reasonable explanation, but that explanation isn’t available yet.  But that said, there are some points here that I want to comment on.

“The top line on the press release from the International Association for the Study of Lung Cancer is, in my view, very misleading.  The results of the comparison of mortality between people who did and did not have the new test are not even mentioned in the abstract of the research.  They are mentioned as a quote from the lead researcher in the press release, but the wording there makes two things clear.  First, the study was not planned to include enough participants to make a statistically meaningful comparison of death rates after only two years of follow-up.  (That’s what “not powered to detect a difference” means.)  Second, and not surprisingly given the way the study was planned, the difference in death rates after two years was not statistically significant.  That means that, while there were indeed fewer deaths in the group that had the new testing procedure, it is still plausible that that happened simply because of statistical variability – that, entirely by chance, more of the people who were going to die within two years anyway happened to end up in the group that didn’t have the new test.  The fact that the researcher describes this as a ‘trend’ does not by any means remove the possibility that it’s all due to random chance and has nothing to do with the new test.  It’s still true that the new test may in the end turn out to reduce lung cancer mortality, but that is not shown by this particular study, despite what the top line says.

“Lung cancer certainly is a particularly important health problem in Scotland, and indeed it is important in much of the world, so I certainly hope that this test eventually develops to a useful clinical tool.  But there is still quite a way to go.  The abstract and the press releases point out positive aspects of the findings, but there are some less positive points that are not brought out.  For instance, the new test ‘identifies 41 per cent of lung cancers’, but that means that it does not identify 59% of them.  The high specificity of 90 per cent means that, of people who in the end are going to turn out not to have lung cancer, 90 per cent will not have a positive test.  Compared to the figures given for CT scanning, the new antibody test gives fewer positive results, both for those who do not have lung cancer (which is a good thing) and those who do have lung cancer (which is a bad thing).  Is the balance between these two percentages optimal, or does that need more consideration?  Is the cut-off between positive and negative test results on the new test in the right place?  Maybe more will be said about this in the full research paper, but we do not have that yet.  To put it another way, in all 598 people had a positive result on the new test (out of 6087 who had the test), but only 18 of these were diagnosed with cancer in the study period.  All 598 would have received the follow-ups of a chest X-ray and CT scans, but they would all have had a period of concern about whether they had a cancer, even though the great majority of them did not.  But I’m more concerned about the people who had the test and got a negative test result.  There were far more of them than of people who got a positive result, as is to be expected.  But it appears that 38 of them had a cancer diagnosis during the period – that’s quite a lot of people who would, initially, have been falsely reassured.  Can that number be reduced?

“Taking into account all diagnoses of lung cancer, there were actually more in the group that did not have the new test than in the group that did have the new test.  That difference was not statistically significant (by my calculation), so it might all be down to random variability.  But there’s no mention of this as a “non-significant trend” in the press release, as was said about the death rates, even though the apparent strength of the relationship is about as strong as for the difference in mortality, by my calculation.

“I’m not mentioning these points with the aim of rubbishing the research, which does appear to have been carefully and appropriately conducted.  I’ve brought them up only because the abstract and, particularly, the press releases do concentrate on the positive findings and don’t mention that not everything was so positive.  Probably the full research paper, when it appears after proper peer review, will give a clearer and more balanced view – but we haven’t got that yet.”

 

Abstract title: ‘Early Detection of Cancer of the Lung Scotland(ECLS): Trial results’.

This is a conference abstract from the IASLC 2019 World Conference on Lung Cancer hosted by the International Association for the Study of Lung Cancer (IASLC), and it was presented at 9.15am UK time on Monday 9 September 2019.

There is no paper as this is not published work.

 

Declared interests

Dr Jeanette Dickson: “None.”

Prof Paul Pharoah: “I have no conflicts of interest to declare.”

Prof Kevin McConway: “Prof McConway is a member of the SMC Advisory Committee, but his quote above is in his capacity as a professional statistician.”

None others received.

in this section

filter RoundUps by year

search by tag