The topic of lung cancer screening is a very charged one, with most people, patients and physicians alike, having a strong opinion, either for or against. This is also an area in which there can be suspicion that any argument against screening is due to a financial calculation in which saving people from lung cancer isn’t worth the cost of imaging. Any screening discussion also entails a consideration of cost, financial and other, vs. benefit, but here I’ll focus on the issues related to the possible shortcomings of lung cancer screening in terms other than cost.
First, while CT screening can detect lung cancers, it also detects lots of little nodules that turn out to not be cancer. In fact, only a small proportion of nodules on CT are actually lung cancers, and nearly half of the patients involved in some screening trials have nodules identified. Those people who are told there is a nodule to follow now are subject to the anxiety of knowing they have a visible abnormality that is unlikely to be but possibly could be cancer. I have seen a few patients in my clinic who don’t have a cancer diagnosis but have a nodule that is being followed that appears unlikely to be cancer but which leaves them sleeping with one eye open all the time, worried about the possibility that they have lung cancer. There is also the risk that they will need to go through invasive procedures, ranging from a bronchoscopy or CT-guided biopsy to a video-assisted thoracoscopic surgery or even a full thoracotomy (open lung surgery) to get tissue and obtain a definitive diagnosis. In one series (abstract here), twenty percent of the thoracotomies were done for what turned out to be a benign cause. Of course, there are complications that may occur with an invasive procedure, such as a collapsed lung (pneumothorax), pain, risk of infection, and many other problems, even including death. Fortunately, that’s unlikely, but the mortality risk (i.e., risk of dying) from a lobectomy and mediastinal staging is in the 1-3% range, not zero. And it’s especially tragic if you’re chasing down a diagnosis that wasn’t actually cancer.
The other concern is that even detecting cancers may not actually translate to improving survival. To have screening be successful, you need to have a disease that is asymptomatic, can reliably be detected while there are no symptoms, and that can be effectively treated after early detection to improve the survival from the disease. But there are several potential biases that can lead us to think a screening test is more effective than it really is (figures below are all stolen borrowed from an older but very good New England Journal of Medicine review with reference here). For instance, there is lead time bias, which means that you find out something earlier but don’t actually change the outcome, and therefore the apparent survival after diagnosis is longer without better survival (the circle with the dot inside represents the screened person with a cancer):
(click to enlarge)
This would be like being told that a meteor is going to hit earth and destroy the planet on January 5, 2009. Not much help to know if we can’t change the outcome, but instead just learn the bad news sooner.
Another potential problem is length-time bias, which is that screening is most likely to pick up less aggressive cancers, because they have a longer interval of being visible on scans while remaining asymptomatic. The more aggressive cancers grow quickly enough that a yearly CT scan would have less opportunity to detect them before they cause symptoms:
The extreme version of this can cause something called overdiagnosis bias, which is when a cancer can be so slowly moving that it doesn’t really impact survival, but screening allows you to find these non-threatening cancers and show that survival is remarkably good, but in the control group these people would never have had symptoms, never have been diagnosed with cancer, and would have died of unrelated causes without ever knowing they had cancer:
If that concept seems familiar, it may be because you read a similar sentiment I recently described in a post about the potential of overtreating indolent bronchioloalveolar carcinomas. I wasn’t arguing against screening in that post. I was just saying that as someone who sees and treats a lot of BAC, I am seeing a significant fraction of these patients with a form of the disease that I’m concerned have such a slow moving cancer that it would never be a threat to their survival, and the treatment could be worse than the disease. That was me expressing my concern about overdiagnosis bias with BAC (although it can happen with other lung tumors, just less commonly), without using that terminology. And some publications have shown that screening studies have detected a significant number of tumors that have a remarkably slow doubling time (abstracts here and here), potentially taking years for a tumor of just a few millimeters to grow to 1.5 or 2 cm, and likely much, much longer to be a threat to survival.
One other issue is that even the advocates of screening aren’t suggesting that never-smokers or patients under 40 be screened routinely, but I know that there are many never-smokers with lung cancer who feel that screening would have been remarkably beneficial for them. But once you start expanding a screening program to a broad population at much lower risk, it both dramatically increases the cost of a screening program and raises the likelihood that what you find will be something other than what you’re looking for. So you end up spending way more money to cause far more anxiety, remove many more benign nodules with a risk for real complications, to find fewer and fewer real cancers. So even a CT screening process that the major proponents envision wouldn’t detect the vast majority of the 20,000-25,000 never-smokers in the US who are diagnosed with lung cancer each year (or the 30-50% never-smokers in most recent series from Asia).
I’ll conclude next time with some general comments of where we stand with screening in the US these days. But I’ll just clarify one point here, which is that while I have mentioned some areas in this discussion that may be shortcomings of screening, the arguments in favor of screening that I described in my last post also have some clear validity. I am not attacking motherhood and apple pie here. I am not arguing for or against a formalized lung cancer screening program here, just trying to offer everyone an idea of the ways in which people are considering this complex and very polarizing topic.
posted by Dr. West @ 7:02 pm link to this post
Loading ...
March 27th, 2007 at 9:08 pm
Jim,
It’s true that there is often a “selection bias” in research, in which the patients who are included on trials are not necessarily representative of the general population. I’m in the process of reviewing the small amount of research on treating elderly patients with post-operative chemotherapy. The studies typically include patients with a median age of around 59 or 60, and yet the median age of a new lung cancer patient is now closer to 70. So we really don’t have good info on what to do with real life patients from the younger, healthier patients who are more typically enrolled on trials.
One of the problems with poor performance status is that there are many unique ways that patients with a poor performance status are ill, so the study population ends up as a mish-mash of very different patients, which makes it harder to determine what your treatment is doing/can do. The other problem with patients who have a worse performance status is that some of them have major medical problems that could be as serious or more serious than lung cancer. If we need to learn what a treatment is doing for lung cancer, it becomes hard to interpret when several patients die of heart disease or emphysema or a different cancer instead.
That said, there have been a growing number of trials done for elderly and/or poor performance status in lung cancer, and elsewhere in cancer. There previously was next to no research, and now there are many more offerings, but they tend to be different trials than the ones for a healthier population. However, we’ve found that some of these trials can’t get a reasonable number of patients on them. They have trouble enrolling because the people most likely to go to a research center are the younger, fitter, more aggressive patients. Sicker patients tend to get treated closer to home, rather than make it to a major cancer center. In fact, people in oncology are often skeptical about the results that come from the most renowned cancer centers like MD Anderson and Memorial Sloan Kettering because many of the people who go there have come from vast distances and have already separated themselves from the people who don’t have the health or other resources to get on a plane and go to a huge cancer center. There is even data showing that one of the strong predictors of survival in cancer is the distance that someone lives from their oncologist’s office. It’s not that it’s better to travel a distance away to get chemo, but rather that the people who are going past 3 oncology offices to get to a bigger research center are doing it because they are fitter and more aggressive.
So I guess I’d summarize by saying that you’re right, the patients on trials don’t perfectly represent the real world, but we’re getting better all the time.
-Dr. West
March 27th, 2007 at 12:15 am
Just thinking again today about the various biases mentioned in the previous discussions about against CT Scanning for at risk individuals.
One of the reasons against scanning was that there are likely biases built in to early screening and that early detection does not necessary prolong life (which I think is both incorrect and irrelevant.) But I was also reading some of the things that are favorable prognostic indicators including things like sex, race, age, performance status, etc. in some of these other studies/treatments for various cancers.
I got to thinking how can performance status be considered a favorable prognostic indicator in any type of cancer? Nobody with poor performance status is ever eligible for entry into clinical trials, and even the gold standard cancer treatments are not offered to people with poor performance status.
It seems to me that unless you include persons with poor performance status in these studies, you are building in a bias in your research, making your survival statistics even more puffy than the true effectiveness of the drug regimen.
Perhaps studies should be based on their effectiveness with the persons with the poorest performance status instead.
It would be a much more significant statistic to know that a treatment regimen produced 5 year survivals in the 40% range of patients with the poorest performance status, than if it produced 55-60% of the healthiest individuals that you can find and in the cancer’s earliest stages. Seems like researchers are puffing up (biasing) the effectiveness of their treatments and of their research statistics.
Jim
February 18th, 2007 at 9:19 pm
I’ll try to find out more about this. The concept as been out there but hasn’t been nearly as well described and tested as imaging studies for screening.
February 18th, 2007 at 8:35 pm
I’ve just heard that they have developed a breath test of sorts that is like a billion times more sensitive than that of police breath alyzer tests. They think that they will be able to use this as an early detection for lung cancer as it will be able to detect the minute chemical changes in the breathe that are associated with lung cancer.
I’m not sure why they are wasting their time on this stuff, there has never been any evidence that detecting lung cancer early will improve overall survival. Seems like someone forgot to send these guys the memo.
January 27th, 2007 at 11:00 pm
TK and Jim,
I do understand where you’re coming from. I get a bit defensive and frustrated because I, too, find it hard to see a situation outside my own perspective.
I’m glad the puppy comment can bring us to smile — there’s never enough humor on a cancer website.
And for the record, I had no comment about cats…
January 27th, 2007 at 10:57 pm
Yes Dr. West, these frustrations are not directed towards you personally. Presenting both sides serves to educate all of us more about the subject, and it certainly opens up the topic for discussion.
JIM
January 27th, 2007 at 6:27 pm
Dr. West,
I hope you do not become too gunshy after all the emotional responses you have gotten on this topic. You are to be commended for presenting both sides of what appears to be a very controversial issue. You are doing a great service to all your readers.
You do have to understand that a lung cancer diagnosis is devastating to the patient and their loved ones. In advanced disease there is so little hope to hold onto, and we are all grasping for some kind of direction that will give us all a chance of a longer, better quality life. It is sometimes hard to be objective when we are in that situation.
There is also the issue of who will pay for these screenings. I know of a few folks who are paying out of pocket to get screened for their own piece of mind. That is always an option for any of us, lacking support from the insurance community.
You have made many excellent points and please do not stop trying to inform and educate us, in spite of the sometimes heated backlash that you may get.
I was very relieved to hear that you have never tortured a puppy.
Tk