Lung cancer is the number one cause of cancer mortality in both men and women. Screening patients at risk for lung cancer might reduce mortality if it helps find cancers at an early stage while they are still resectable. Randomized studies done in the 1970s showed that screening for lung cancer with chest x-ray did not support this theory. Chest x-rays identified more small tumors, but resecting them did not improve mortality. The question of whether screening with chest CT can improve outcomes remains unanswered.
In October 2006, an observational study in the NEJM looked at screening of asymptomatic high-risk patients with CT. The International Early Lung Cancer Action Program (I-ELCAP) screened 31,000 patients and found 484 cancers. 85% of the cancers detected were stage I, and they estimated an 88% survival amongst these patients. This is in contrast to the 70% ten-year survival currently seen in patients with stage I lung cancer. They concluded that CT screening could detect lung cancer that is curable.
In stark contrast to these results, Bach et al recently published an article in JAMA that showed CT screening increases the number of cancers detected, increases the number of cancer surgeries performed, but does not decrease mortality. They screened 3200 asymptomatic patients with a smoking history with annual CT for four years and compared their outcomes with those expected from well-validated prediction models.
They expected to find 44 cases of cancer based on the prediction model but actually found 144 cancers. Screening made a patient three times as likely to have a cancer diagnosed. 109 cancer surgeries were performed as compared with 11 expected; screening increased the number of surgeries tenfold. 38 patients were predicted to die, and 38 patients did die; there was no evidence that CT screening reduced the risk of death.
Why the disparity between these two large observational studies? One study suggests screening will reduce mortality by 80%; the other that screening will have no effect. The editorialists have several theories. The disparity might be explained by chance, by difference in the populations screened, by problems with the prediction model used, by methods used to identify patient deaths, or by differences in their primary outcomes. It is important to remember that both of these studies were observational, and neither randomized patients to screening vs. no screening.
As it stands, there is little evidence to support lung cancer screening,and potential harms of excess diagnosis, surgeries and treatment are real. Two randomized controlled trials are underway, the National Lung Screening Trial and the NELSON trial. We will have to wait for results of these studies to clarify whether CT screening for lung cancer is beneficial. Despite this, legislation has been introduced that would require Medicare to cover the cost of screening. We need to take a step back before adopting screening interventions that may cause more harm than good. CT screening for the early detection of lung cancer cannot yet be recommended.
Image courtesy of Wikimedia Commons