Subscribe in iTunes
Today we are discussing lung cancer screening with Dr. Harold Sox and Dr. Peter Bach. Dr. Sox is an active emeritus professor of medicine at the Dartmouth Institute at the Geisel School of Medicine and the former editor of the Annals of Internal Medicine. Dr. Peter Bach is a physician at the Memorial Sloan-Kettering Cancer Center in New York City. Dr. Bach's research covers healthcare policy, including cancer quality and cost as well as lung cancer epidemiology.
Lung cancer is still the leading cause of cancer deaths and most patients are diagnosed with advanced-stage disease and have a low probability of survival. While there have been many advances in treatment for late-stage disease, screening and prevention are likely the most cost effective and best approaches to prevent lung cancer deaths.
CancerNetwork: Dr. Bach, you are the chair of the expert panel that has conducted a comprehensive benefit–risk analysis of computed tomography, or CT screening for lung cancer. The recommendations have been published in the Journal of the American Medical Association as of a few days ago. Could you briefly describe the recommendations and the research used to create them?
Dr. Bach: We conducted an evidence review comprising all English language literature that examined the potential benefits and risks of CT screening for lung cancer. The review was sponsored by 4 major medical organizations—the American Cancer Society, the American College of Chest Physicians (ACCP), the American Society of Clinical Oncology (ASCO), and the National Comprehensive Cancer Network. The expert panel that guided the review came from these four organizations and covered the major fields and had no conflicts of interest. The goal of the review was to form an evidence base on which practice guidelines could be based. In order to do that, we conducted a systematic review driven by specific questions related to risks and benefits. Once that was completed, two of the organizations involved, the ACCP and ASCO, as well as another organization, the American Thoracic Society, developed and endorsed clinical practice guidelines based on that.
What the review found is that there are only three studies of the correct design to assess the impact of CT screening on benefit—primarily the avoidance of death from lung cancer or the avoidance of death from other causes that could be either increased or decreased by screening. On that front, we found that in the largest trial—a US-based trial called the National Lung Screening Trial (NLST)—there was a relative decrease in lung cancer deaths among individuals who were screened on average three times, once each year for 3 years, and followed for about 6 years. A relative decrease of 20% or an absolute decrease of 0.33%, translating into for every 1,000 people like that, 3 lung cancer deaths would be avoided. The other two studies were smaller, European-based and neither found that benefit. We also found conclusive evidence that there is no effect of CT screening on changing the risk that someone will die of another cause. CT screening does not prevent death from causes like heart disease or emphysema, but it doesn't cause them through complications. When we look for harms, one that people are most concerned about is the detection of false positives, abnormalities on the CT scan in the lung that are not cancer but still cause anxiety and prompt follow-up—we found that across all of the studies, about 1 in 5 people, each time they are scanned, have one of these false positive which leads to follow-ups. What we found that was reassuring is that although this triggered many more diagnostic tests like CT scans and PET scans, it did not trigger many tests that were invasive or risky such as biopsies or small surgical procedures—most of those, although not all, were done on patients who were ultimately found to have lung cancer.
Other concerns are about risk-included radiation risk, and we found limited data on that, although using the best available models we estimated that the harm from the radiation scanning in this group of patients was relatively low—about a tenth as important as the potential benefit in terms of reduction in lung cancer death. We looked hard to see if we could find any evidence that CT screening affected smoking behavior. Some people are concerned that if you have a normal CT scan you will go back to smoking. Other people have argued that if you have a scary CT scan it will cause you to stop smoking. People have tried to look at that and have found no effect in either direction. We also addressed the questions regarding where CT screening has been shown to be effective. We found tilts in the literature of studies being done in large academic centers and where there was a lot of multispecialty expertise and concluded that those are the only places you could argue that CT screening could be done safely and effectively. That led to recommendations that patients who are like those who were studied—high-risk individuals who smoked 30-pack years or more, are 55 to 74, and if they have quit smoking, have quit within the last 15 years. For that group it is reasonable for physicians to suggest screening and discuss the risks and benefits, but that screening should be conducted in a place that is like those where the NLST was conducted, large academic centers—many of which were National Cancer Institute (NCI) designated where multispecialty care was available and there was diagnostic expertise.
We recommended against screening, screening should not be offered to individuals who are not like those in the NLST study. It is not scientific orthodoxy that causes us to do that. It is the data regarding the underlying risk of lung cancer. So we recommend against screening these individuals because if they smoked less than those in the NLST, their underlying risk of lung cancer is low enough that we don't believe the potential benefit outweighs the inherent risks and harms. So that led to that recommendation.
CancerNetwork: OK, so who in the United States is actually getting screened for lung cancer, besides smokers, who are the other potentially high-risk populations that should be getting screened, maybe that are not?
Dr. Bach: The studies that have shown a benefit focused on those whose risk factor for lung cancer is a combination of their age and smoking. Studies have not been done looking at important end points in these other groups. There are concerns, obviously, about other lung cancer risk factors, both genetic and environmental, but we don't know enough to recommend screening these individuals. Many of these other risk factors like emphysema or asbestos exposure can increase your risk of lung cancer but they also change the anatomy of the lung and we don't know how CT screening will perform in groups like that where it is just frankly much harder to read a CT and be confident that what you are looking at is or is not something malignant. We don't know who is being screened right now. We think the prevalence is quite low. There was recently a survey published that suggested that it was really just a de minimis number of people being screened. It is easy to find on the Internet ads for places that are offering screening and sometimes on the Internet, or if you call these places, it appears that many of them are not following any of these recommendations and it appears that the key qualifications for being screened is that you have a checkbook.
CancerNetwork: Dr. Sox, do you feel that CT screening is cost effective? Since the publication of the National Lung Screening Trial, are more patients being screened by this method?
Dr. Sox: Well, to my knowledge, nobody has published cost effectiveness data analysis based on the NLST results. There is a very reputable group that is working on one and I expect it will be published probably within the year. Right now, we don't know.
CancerNetwork: Do you feel that based on these results and others, that overdiagnosis is seen as a major issue? Is there evidence of this from trials and cohort studies?
Dr. Sox: Several trials have shown evidence for overdiagnosing of lung cancer. Overdiagnosis in cancer occurs when somebody discovers a cancer that has no important effect on the patient. Apparently, some cancers can grow to a certain size and then stop growing or maybe get smaller, or grow so slowly that they never get big enough to cause any problems. As an example, many men die with small cancers of the prostate that nobody knew about because they never caused any symptoms. So clearly overdiagnosis can occur with cancer and several trials have shown the overdiagnosis of lung cancer. It can be a problem because when you discover a cancer you don't have any way of knowing whether it is going to be one that is going to grow big enough to cause problems and so you end up treating all of the cancers. The patients whose cancers were not going to cause any harm get treated often with surgery that they really wouldn't need if we had a way of knowing what the future of that cancer was. So these patients end up suffering the same harms from treatment and incurring the same costs of treatment, but they don't get any benefit.
CancerNetwork: In terms of other screening methods such as chest radiograph, for example. What is the most frequently utilized screening method? Dr. Sox, let's start with you.
Dr. Sox: This is an interesting question because the bottom line is we don't know. The reason we don't know is that when you do a chest x-ray or a CT scan, sometimes you do it on a patient who has symptoms that are worrisome like persistent cough or blood in the sputum, so you are worried about lung cancer and the screen is to diagnose lung cancer. Other times you are doing a procedure to screen for the disease, which by definition means that the patient does not have any symptoms of the disease you are screening for. The problem with measuring lung cancer screening rates is that doctors don't routinely record the result or the reason for the test and as a result you can't make the distinction between doing the test for diagnosis and doing the test on someone who is perfectly asymptomatic and being screened because of their extra risk. The bottom line is we don't know which methods have been favored in the past. The only option prior to low dose CT have been chest x-ray and lung sputum cytological examination, but those were shown 20 to 25 years ago to have no effect on lung cancer mortality in randomized trials so they really never caught on.
CancerNetwork: Dr. Bach, do you have anything to add?
Dr. Bach: No, I agree with Dr. Sox.
CancerNetwork: OK, and as a last question, can either CT scans or chest radiography, can these screening methods distinguish between small-cell and non–small-cell lung cancers? Is this a goal of future screening methods?
Dr. Bach: It is a good question, and I think the answer is yes and no. There are certain features of different kinds of lung cancer that allow you to distinguish to a reasonable degree on CT scanning. But the gold standard is tissue diagnosis. It isn't particularly a goal of screening to make these distinctions. The objective of screening is to try to intercept those lung cancers that you can intercept in early stage where your treatment also changes the natural history of the disease. In laymen's terms where you can cure the disease rather than when it is in advanced stages when it is very difficult or nearly incurable. It is worth noting that the impact of screening in the NLST, which is now the most carefully done study—where as I noted before, the centers are uniformly excellent and a level of adherence to screening and follow-up screening among the enrolled subjects is higher than we have ever seen in a routine screening setting—that only about 1 in 5 individuals that were, if you will, destined to die of lung cancer had their deaths prevented (that is what that relative risk reduction of 20% means). Which means even carefully done expert centers with high follow-up, 4 out of 5 of the lung cancers in laymen's terms snuck through and were incurable despite routine screening. So the simple notion that screening intercepts cancers before they become advanced may be correct, but the extent to which you can do it is really quite limited at least as this study suggests.
CancerNetwork: Dr. Sox, do you have anything to add to that?
Dr. Sox: Only to say that the issue raised by Dr. Bach, on whether you can get these results in typical community practice is really a big issue. For a lot of reasons of which the most dramatic is the mortality of lung cancer surgery—the patients in this trial got the absolute best care imaginable. For example, a mortality rate of surgical treatment of lung cancer was 1% in the trial, and the best information we have about the death rate from lung cancer surgery in the broad community practice, which includes specialized centers as well as community hospitals, is about 4%. One of the interesting pieces of information we can get from the cost effectiveness analyses is how much the benefit of lung cancer screening is affected if you substitute a 4% mortality rate for a 1% mortality rate in the models that the analysts used that tried to predict the results of the lung cancer screening.
CancerNetwork: Thank you so much to both of you for joining us today.