SLIDES & TRANSCRIPTS
Tuesday, June 19

INTRODUCTION TO LUNG CANCER SCREENING, "WHAT IS THE SCOPE OF THE PROBLEM, WHAT ARE PLANS FOR STUDIES WITH SPIRAL CT SCREENING, AND WHY ARE DIAGNOSTIC AND THERAPEUTIC CONSIDERATIONS IMPORTANT."

John Ruckdeschel, MD

Of course the other problem is us. Those of you who are old enough may remember "Pogo" going through the swamp. "We have met the enemy, and they is us." Clearly, therapeutic nihilism is endemic. We are now into our third generation of cardiothoracic surgeons who are being trained to do cardiac surgery, and who are also being trained to treat lung cancer surgery or general thoracic surgery as if it was punishment for not filling in their billing forms in an orderly fashion.
It's funny. We laugh about it, but for those of us who live in the hinterlands, finding a city with a general thoracic surgeon is a challenge. And you would be shocked -- well, maybe you wouldn't be, as all of you sit in centers -- to realize the incredible mess that usually arrives with understaged or improperly staged patients when we see them.

The pulmonary community has for many years, had an extraordinarily antagonistic view of the problem of lung cancer. I would say over the last five or six years, however, that has begun to change and we are beginning to see a good dialogue go on at the pulmonary meetings. But there are several generations of pulmonologists out there who think that lung cancer is a diagnosis you make, not a disease you treat, and that one of their greater goals in life is to save lung cancer patients from those medical oncologists who are going to poison them.

The patients are socially ostracized. And I think most important is the balkanization of both therapy and research programs. What is going to come out of the Lung PRG meeting after several days of back and forth about issues of basic science and cancer control studies and tobacco control studies, the leading recommendation that came out of each of the groups, the thread that wove each of them together was that having groups like the lung SPOREs that are focused around the lung cancer problem, and bring together basic cancer control, imaging scientists, clinicians, et cetera was far more important than continuing to have a separately organized research group for each area of scientific interest -- a detection group, a prevention group, a biology group, a radiotherapy group, et cetera down the line.
And so with this balkanization, there are a limited number of investigators in lung cancer. That I think is one of the great problems that we have brought to the table.

TOP

I saw this at a meeting the other day, and I immediately wrote it down, because it summarizes how I feel about it, and also because I had a great fondness for H.L. Mencken in the years I spent in Baltimore. And some of you who were there and had to put up with the reception at Mack Harvey's house when you were a house officer there, will remember fondly that H.L. Mencken was related to his wife. And the entire Mencken library was kept by Mack Harvey. So I was quite enthralled to see that.
He was a wonderful iconoclast. And this is certainly a very true statement. And part of the reason of the meeting in March was that all sides of the issue of early detection of lung cancer had simple solutions to the problem.

TOP

I'll go quickly past prevention. This is probably one area in which our outcomes have been less successful, even than treating advanced disease. There is no clearly useful agent. In fact, some are actually harmful in current smokers. And even worse, we have no agreement on an intermediate marker, so that we can speed these studies up.

TOP

Early detection on the other hand, has two fundamental paradigms. One is radiology-based, and the other is sputum-based. And although I think most of us would agree that sputum-based approaches are conceptually more interesting, and much more likely to lead to a truly early diagnosis, there is in fact no agreed upon marker at the present time that we can turn to. So most of what we will look at today are issues related to radiographic screening.

TOP

Radiographic detection, certainly we don't need to go back through the NCI trial at Hopkins, Mayo, and Sloan. But clearly, from that trial there was a regular pick up of early, usually squamous cancers. There was a clear stage shift to detection of earlier disease in the screened subjects, but no improvement in overall survival.

TOP

Setting the stage for this meeting was Claudia and David's work in New York in the ELCAP [Early Lung Cancer Action Project] project, noting the obvious - that spiral CTs are better than chest x-rays for picking up small lung nodules. Screening 1,000 individuals, 27 cancers, 85% stage I in that group. A couple of Japanese studies that also had similar results, although their rates were lower. Again, reiterating what will be a common theme through all of this, that the population you choose to study, how you select them will very much determine what the outcome of your screening study will be. The more stringent the criteria, the more advanced stage you are going to find. The more pulmonary damage you have up front, the more advanced cases you are going to find. The looser your criteria, smoked a few packs anywhere, or general population, you are going to find far fewer cases.

TOP

Going into the meeting in March, there were clearly several unresolved issues from the spiral CT studies. One was this concept of overdiagnosis of lesions with limited malignant potential. Second was the sensitivity and specificity, especially this issue of the false positive rates. This is not a needle aspiration of a breast lump that we are doing, it's not a colonoscopy that we are we doing, it's a thoracotomy that we are doing on a number of these patients in order to make this diagnosis. And this is a particularly costly and potentially more escalated work-up. And of course we still have the issue of whether it impacts on survival.

TOP

So out of this came a group of people at the meeting who I will call the "spiralists." The "spiralists" -- Claudia and Dave and others being in the fore -- believe that screening is a form of early diagnosis. It is not a therapeutic intervention, and should lead to a quantitative assessment of all of the parameters relative to practice, not just overall survival.

TOP

The "spiralists," and I take great liberty at this statement, would say if a chest x-ray leads to a clear stage shift, and spiral CT is more sensitive, and true early detection would lead to improved early survival, let's start doing the procedure more widely, and assess the benefits and limitations without waiting for the results of a trial.
Was that okay to paraphrase, Dave, with that?

TOP

As Scott indicated, the controversy between the "survivalists" and "spiralists" has been heated, and in fact several of these meeting were originally set up because Rick Klausner, who was being pummeled from both sides of the issue at a meeting, turned around and happened to see the two largest people he could find, myself and Marty Abeloff and asked us to help get him out of the problem, which is where our involvement came from.

TOP

The "survivalists," on the other hand, feel that no screening test should be adopted in the population until the full impact on survival is known, and that diagnostic procedures, such as a thoracotomy, are not trivial from a clinical or a cost perspective.

TOP

But given the frequency and lethality of lung cancer, both potential gains in survival and the costs could be highly significant. This led to the joint meeting sponsored by the NCI and the American Cancer Society in March. It was hosted by Bob Smith from ACS, Dan Sullivan, and myself.

TOP

We were called upon to discuss this rapidly emerging technology, various methods to evaluate the technology, and the data systems and other needed infrastructure. The focus of the workshop was the recognition that observational trials of spiral CT were ongoing in a number of locales, and that the data from these trials needed to be optimized to inform the larger goal of reducing mortality from lung cancer. And this is really one of the key themes that came out of the meeting, and that Scott reiterated this morning. Like it or not, we are not operating in a vacuum. This is not a laboratory where we can control who does or does not perform or receive spiral CT scans. They are being done now by the hundreds and thousands, and will increasingly be done over the next several years.

TOP

The workshop divided up and looked at four major topics, the impact of technology training, changes on screening studies. And the issue here is very clear. On the side of the "spiralists" is the concern of launching a ten year or more survival-based study when the technology is changing so rapidly that what we are using today will be irrelevant in two or three years. You will end up with a study using a procedure that nobody wants to use anymore, and the results, when you finally get them, won't be very productive.

The second group was actually the open warfare group, the methodologies for screening studies. And this was a survival study, or anything but a survival study here. And it was very clear that at this particular point in time, I think Rick Klausner set the tone for the meeting right from the beginning, and then Peter Greenwald carried it through into this meeting, that in fact this would not be an either/or. That there was a necessity for both sets of studies, and therefore gave the blessing to go ahead with a whole series of different approaches to this, and this meeting really being an offshoot of that. Data comparability and informatics. And then new frontiers of screening science.

TOP

Barbara McNeil chaired the impact of changing technology. And the questions that they were posed were how do we account for changing technology in the course of a trial? And how do we integrate the computer-aided diagnosis into screening technologies?

TOP

The assumptions that they made were that the CT studies are going to have a relatively short accrual period, and then a long follow-up period, and that technology changes should be considered only for a five year time frame. That looking beyond five years was pure speculation, and not likely to be helpful. They stressed, as did every one of these groups, that cost data needed to be collected prospectively.

TOP

Their other assumption was that resolution will improve. The ability to get down to 1-3 millimeter lesions relatively reliably would change, but that that wasn't a major change over where we were now. That clearly computer-assisted diagnosis would enhance image processing, and was likely to move rather rapidly, since many of the algorithms developed for breast imaging could be modified for nodule imaging. And their feeling was that biomarkers may actually better define the at-risk population who should be getting screened. I think this is really the key thing that has not been addressed yet, is actually who to screen. Right now obviously, those in the commercial world would do it on anybody with a pulse and an insurance card or a checkbook, but that's probably not the correct population.

TOP

They felt that both the non-experimental and the randomized controlled trial designs were useful. That there is a whole series of manuscripts and techniques in the area of adaptive designs and sequential statistical methods and Bayesian methods that would allow the analysis of non-randomized data in order to give us useful, helpful, and informative data from the non-randomized studies. And their feeling was that the issue will not be technology changes. The issue will be observer variation. And they cited extensively the data on mammography. That the issue is not how well the machines can resolve issues, it's how poorly physicians actually can use the data that they see in front of them, how much variation there is within and between observers, and were very strong on their recommendations for instituting strong quality control and centralized quality control techniques right from the beginning.

TOP

Again, cost data from the start. And I think their major recommendations were to try looking at intermediate endpoints, so that we are not locked into a 10-20-year study right from the beginning. And their very strong recommendation is that we not freeze the technology during any trial. That telling people that they could only do it with a certain type of machine, at a certain sensitivity, and a certain set of settings, and then holding to that because that issue of technology was important, was the wrong approach to this. The technology had to float, and had to change as it went forward.

TOP

Jack Mandell chaired the second group on methodologies for evaluation of screening. And this was an assessment of the various approaches.

TOP

Their assumptions were that a randomized trial would be desirable, but might take a prolonged time, and that feasibility might be a real issue. Many of the "spiralists" and many of us who are on neither side of that operation feel that -- and I think Val Rusch and others in New York have felt that-once this opens in a community in the commercial radiology field, once the advertisements start to hit in a big way, that the ability to randomize smokers to a trial of screening CT versus no screening CT, and have them stick to it as the data comes out (and as the newspaper articles go on about the cases that are truly found early, which is what happens in the communities), that randomization would be impossible. And so right off the bat the PLCO trial is looking at this feasibility question of whether or not people can be randomized. It was also their very strong conclusion that alternative designs were a reality, and could not be dismissed by those who were purists about the randomized trial.

TOP

This was the first group where it became clear that there was an urgent need for practice-relevant questions to be answered. In other words, what do we do with small lesions? How do we deal with these things? Is the algorithm that Claudia and her group developed for watching and waiting below a certain size, when to restudy, when to operate, when to do PET scans, et cetera correct? What is the right answer to this?
And actually one of the first randomized trials that had nothing to do with survival was proposed here, which was a randomized trial of the management of small lesions, sort of a question of operate now versus operate later.

TOP

They felt that the data that is being accumulated, both in the non-randomized studies and globally in radiology practices should be coordinated and standardized. And that developing models to assess changes in strategy would be helpful.

TOP

The community trials if you will, the ongoing, uncontrolled study of spiral CT are not going to be coordinated centrally, but quality control could be in the same way that mammographic screening has been done by setting certain minimal standards for performance of these. They also felt strongly that all of the trials needed a multidisciplinary team to deal with the outcome of this, and that lacking a team of pulmonologist, thoracic surgeon, et cetera to deal with the problem after the lesions were found was in fact not appropriate. And what we are seeing increasingly in practice, and I know many of you have commented on this to me at meetings over the last year, our patients walking in, our staff walking, our scientists walking in with their CT scans done somewhere, where they walk in, have a CT, walk out, and there is no follow-up of it, and they want to know what to do with this small lesion that has been found in various places. It has also been felt that all of these studies really do provide a unique time for smoking cessation interventions. That when you get someone sitting across the table from you, writing their check for their spiral CT scan in most instances, that that is a unique opportunity to introduce or to reinforce smoking cessation techniques.

TOP

The data standardization and informatics group was chaired by Rachel Boward-Barbash. They were looking at what types of comparisons are feasible across all types of studies, and what will be the influence of computerized systems on data management in the clinical practice.

TOP

They felt that there needed to be support from NCI for the development of a repository of standard data elements, with an explicit data dictionary that defines the key characteristics.

TOP

And that really anyone could support, and that there would be within that, a minimal data element that a private radiologist who was doing these on his own, could provide that data to these studies. Then there was obviously a larger data set that the research units would work from. And that these could actually be shared across a diverse set of study designs. And of course setting up the expert panel to define the dictionary and data elements was critical and urgent.

TOP

The final group, and the group whose report was about 12 pages long, instead of 2 pages long was the new frontiers of science for screening. John Field chaired this group. The purpose was given that these populations are undergoing screening. Are there other aspects of screening science that should be co-studied before we lose this opportunity?

TOP

They felt strongly that blood, sputum and all epithelial smears from samples of the screened and unscreened populations needed to be collected. And again, refining the work-up of early lesions, and making sure that we collect the tissue from all of these. And that we have a repository and a bank. Whatever we decide to do with these lesions, that we clearly keep tissue.

TOP

They felt obviously that support for biomarker development needed to be sped up, as this would complement screening. And they felt it was urgent to correlate radiographic markers and characteristics with these biomarkers.

TOP

Overall, there was a clear need for adequate support personnel for these trials. I don't know what the NCI budget for cheese went up that week, because we did a fair amount of whining about the lack of support, both in cooperative groups and prevention trials and several other areas for the infrastructure support needed to do these. And the tradition of having academic centers pick up the slack was probably not going to be able to continue when you were talking about doing thousands of patients, and the way that it might happen when you were doing hundreds of patients on a clinical trial.Support for the informatics and the tissue repository were clearly spoken, but I think also felt that NCI was well along the pathway of providing infrastructure support in these two areas.

TOP

So the conclusions of this group, which is being directly incorporated in the Lung Progress Review Group report which is being written now, said that although a randomized controlled trial may be preferable, there are numerous alternative design studies underway that will answer crucial questions on the biology of early lesions.

TOP

The question is not whether non-randomized trials should be done, but how we learn the maximum amount from these trials. And understanding what to do clinically with these lesions is a high priority. And of course it's now the major focus for this meeting.

TOP

Let me give you my biases from having sat through about four of these early detection meetings in the last three months in several cities, with many of you actually. I have been doing lung cancer now for about a quarter of century, which is a scary thought. I have never met a benign lung cancer. In the years when I took care of breast cancer and prostate cancer as part of my training, or as part of my general oncology practice back in New York, I knew that there were prostate cancers, and there were breast cancers that acted benignly. And I worked with some of the earliest oncologists in the field what had very little other than hormonal manipulations, and they had breast cancer patients who went on and on and on, and prostate cancer patients who died with prostate cancer, but not from prostate cancer. Now I don't know about you, but we could probably muster a half a dozen cases in all of our collective experience of lung cancers that just sat there and didn't do anything, and left the patient alive for years and years and years. And so this issue of possible overdiagnosis with CT scan of benign lesions, I'm having trouble getting my hands around that.

TOP

I think the major problem is not what to do with these lesions, nor is it what procedure to use, it's which population to study. We haven't published the data yet, but some of you are aware of our own screening study. We started out using a screening that is quite different from the one in New York, where the cutoff is more than a 20 pack year history of smoking. What's the cut off level? Ten pack year history? Others have used 20 or more pack year history, with either current or former smokers.
Our cut off was a 30% reduction in FEV-1, which Mel Tochmanhad previously reported was in itself was a strong predictor for the subsequent development of lung cancer. In our screening group, the majority of the patients with less than 1 centimeter lesions in fact have advanced disease, not stage I disease.
And so I think that that is really going to be the critical issue. Who goes in the pipeline at the front end? What group of patients do we choose? And I think that when you do larger population-based studies, as the Japanese have done, and as we will hear later today, their rates of finding disease are lower. The more you ratchet down to an exposure history, the more you are going to find lesions, and more often the lesions will already be metastatic at the time that we see them.

TOP

As we have already said, there is no question that in our society, where access to CT scan is open, it is a private capitalistic society, and the response of the consumers will press this issue.
How many men in this room are over 50? Could you raise your hand? How many of you had a PSA last year? About a third to a half in that group. Of the others, how many are planning to have one? How many are planning not to have one, because they don't think the data is secure enough yet on prostate-specific antigen? Well, this is what's going to happen in CT. We have literally tens of millions of former smokers who are sitting out there now, many of them reasonably well educated, saying to themselves, "you mean there is a test I can get for a couple of hundred dollars, that not only will tell me whether or not I have an early lung cancer, but just like it says here in this advertisement, tell me whether I have significant coronary disease as well? Where can I sign up for this?" And this is happening with increasing regularity, and the discussion is going on even in academic radiology departments, not about whether they should introduce spiral CT scanning, but when and how to introduce it.

I think spiral CT is going to work. I actually think it's going to lead to a major change in the stage presentation of lung cancer, and that we are totally and completely ill equipped to deal with that issue. We don't have enough good thoracic surgeons in the country to handle a sudden burst in the number of cases coming through. We don't have enough multidisciplinary teams to deal with these people sensibly.
I also think that spiral CT will reduce the mortality from lung cancer, but this will be greatly enhanced as we get developed sputum biomarkers, and other biomarkers to help elucidate who to screen.

Therefore, I think the purpose for this meeting is to recognize that there is currently an American College of Radiology randomized study underway, the PLCO trial. It may or may not be fully expanded to a randomized trial. And there may or may not be another randomized trial put underway, but that there are literally dozens now of trials underway that are not randomized, and are looking at various aspects of which population, and which procedure, and how to approach this.

And we are lacking an agreed upon consensus of what to do with these lesions. We are now facing an entire new terminology of both histopathology, cytology, and radiographic terms such as ground glass appearance. I don't think any of us grew up with that in Squire's textbook of radiology as one of the things that we were looking for. So that is where we are right now. We are early in the process. And I think that getting some agreement on what to do with these lesions will go a long way towards making the data sets that come out of these studies useful.
Thank you.

TOP