Overdiagnosis rate with lung cancer screening CT is 18%

Low-dose CT screening reduced death from lung cancer by about 20% in the National Lung Screening Trial, and the U.S. Preventive Services Task Force signaled it would recommend CT screening for most people with a heavy smoking history. The Affordable Care Act stipulates that USPSTF-recommended screening tests be completely free to consumers, so lung cancer screening will soon become a standard part of primary care for eligible patients.

Like any screening test, lung cancer screening also results in overdiagnosis: detecting relatively harmless tumors that would be best left undiscovered. Overdiagnosis is inherent to any screening program, and the additional tests, anxiety, and invasive procedures that result are considered (essentially) the cost of doing business in public health. Exactly how much overdiagnosis was in the NLST was unknown.

Researchers feel they now can say with confidence: as many as 18% of lung cancers detected in the NLST were actually slow-growing tumors that would never have harmed patients during their lifetimes, according to Edward Patz Jr. et al's report in the December 9, 2013 JAMA Internal Medicine. Their analysis suggests that for every 1,000 people screened with low-dose CT, about 3 deaths from lung cancer would be prevented, but 4 people would be "overdiagnosed," undergoing unnecessary surgical resections, chemotherapy and/or radiation for indolent cancers that did not require any treatment. Authors felt 18% was the upper limit of overdiagnosis, meaning the actual number could be lower.

The National Lung Screening Trial randomized 53,454 adults aged 55 to 74 who had a 30+ pack-year smoking history to be screened with either low-dose CT or chest films. During ~6.5 years of follow-up, CT screening diagnosed 1,089 lung cancers; chest X-rays diagnosed 969 among the subjects. Since both groups should have had equal rates of cancer, NLST researchers knew at study completion some of the extra cancers detected in the CT-screening arm might have been overdiagnoses. But they needed another 4-5 years of follow-up to be sure: this allowed people in the control chest X-ray arm to "catch up" to the CT arm and develop any lung cancers that might have been lurking during the trial (but which were detected early in their CT-screened counterparts, creating the imbalance). The behavior of these undetected tumors (lethal or non-lethal) would determine the overdiagnosis rate.

Bronchoalveolar lung cancers, well known to be indolent, were 80% likely to represent overdiagnosis. But even among all non-small cell lung cancers, as many as 22% may have been overdiagnosed and would never have harmed the person if left alone.

"These data raise the question as to the necessity and type of therapy required if a diagnosis of minimally invasive adenocarcinoma is established and challenge the diagnostic community to develop a classification scheme that could accurately phenotype all lung tumors," Patz and colleagues cautioned. The American College of Radiology did damage control, issuing a statement calling the overdiagnosis rate "modest" and on par with the expected overdiagnosis rates with other types of cancer screening: "Overdiagnosis is an expected part of any screening program and does not alter [the benefits of lung cancer screening]."

But treatment for lung cancer -- thoracotomy, lobectomy, chemotherapy, radiation -- is no walk in the park. More than one person treated unnecessarily for every death prevented is a sobering statistic. And if lung cancer screening creeps into lower-risk populations, as seems likely, overdiagnosis would be expected to rise.

"These overdiagnosis cases represent an important potential harm of screening because they incur additional cost, anxiety, and morbidity associated with cancer treatment," said Patz et al. They recommended doctors mention the risk of overdiagnosis to patients when explaining the risks of low-dose CT screening for lung cancer.

We certainly should. But doing so will make an already complex conversation mindblowing for the average person. Owing to decades of manic advocacy of disease screening in the U.S., most people (and their doctors for that matter) are not well-prepared to consider the idea that a lung cancer could be harmless, or that they could actually be harmed more by their surgery than by their cancer. And since most of these conversations will happen in harried, 15-minute primary care visits, it's safe to say that nuances essential to informed decision making will often be lost.

If you were counseled, The government recommends you get a lung cancer screening CT. There's a 1 in 300 chance it will save your life, but about the same chance you will get cancer treatment you don't need -- would you sign up?

For now, remedies to the overdiagnosis problem are scarce: if a biopsy shows non-small cell lung cancer, there's no good way today to predict its future behavior accurately enough to advise forgoing treatment. This same problem has plagued breast cancer screening programs for decades. Development of better biomarkers and lung cancer risk stratification models should someday help refine management and reduce unnecessary treatment for overdiagnosed lung cancers.

But in the present day, while we all get used to these new statistics, what they mean, and how to explain them clearly and responsibly, lung cancer screening programs are in full swing across the U.S. Professional societies have recommended lung cancer screening for all patients meeting NLST enrollment criteria. Most academic centers now offer lung cancer CT screening programs, and some advertise discounted lung cancer screening CTs in an attempt to lure potentially lucrative patients.

Relatively few people will undergo unnecessary surgery to resect harmless overdiagnosed lung cancers. Because they trust their doctors' good-faith intent to cure, most of them will not perceive it as harm. But we should recognize this is happening, even if we can't yet tell exactly to whom.

Overdiagnosis in Low-Dose Computed Tomography Screening for Lung Cancer. JAMA Internal Medicine 2013; December 09, 2013. doi:10.1001/jamainternmed.2013.12738

See also:

Lung cancer screening pro/con (Annals Internal Medicine)

Chest X-rays are useless for lung cancer screening

How to ration lung cancer screening CTs, rationally (NEJM)

Breath tests can accurately diagnose lung cancer

Risk of complications after transthoracic needle biopsy: population analysis

New lung cancer prediction tool promises better use of screening CT (NEJM)