A new analysis of National Lung Screening Trial (NLST) data has found that as many as one-fifth of all lung cancer diagnoses may represent overdiagnosis -- that is, detection of indolent or slow-growing lung cancers that may never have harmed the patient, according to a report in JAMA Internal Medicine.
In a study funded by the U.S. National Institutes of Health, researchers from several institutions analyzed data from more than 53,000 NLST subjects for signs of overdiagnosis. Following them for almost five years, they measured the difference in the number of lung cancers in the CT arm of NLST versus the chest x-ray arm of the study, aiming to determine how many CT-detected lung cancers didn't manifest clinically over the follow-up period.
The results showed that more than 18% of cancers appeared to be from overdiagnosis, they concluded (JAMA Internal Medicine, December 9, 2013).
Clinically insignificant tumors?
Low-dose CT (LDCT) screening has been shown in clinical trials to be an effective screening tool for some individuals, "but some of the tumors it finds may be indolent (slow growing) or clinically insignificant," wrote lead author Dr. Edward Patz Jr., from Duke University Medical Center, and colleagues.
Overdiagnosis is defined as the detection of cancer with a screening test that wouldn't otherwise have become clinically apparent. The difference between clinically manifest cancers and screen-detected malignancies is known as the lead time.
Of course, overdiagnosis is not easy to quantify. Doing so requires sufficient "catch-up" time to ensure that cancers that are initially indolent have time to manifest clinically and be counted as cancers, the authors explained. And not every nodule can be followed up with biopsy to determine if it was a slow-growing lethal tumor or truly indolent.
"In the NLST, there were four to five years of follow-up after screening, which may not be sufficient to detect all cancers in the [chest x-ray] arm; with longer follow-up, additional detection may occur," Patz and colleagues wrote. "Therefore, the present study calculated rates of excess cancers in the LDCT versus [chest x-ray] arm, for all lung cancer and for various histologic subtypes."
The excess cancer rates provide an "upper bound" on the true overdiagnosis rate that occurs with LDCT screening, they noted.
Among 1,089 lung cancers reported during a mean follow-up of 6.41 years in the CT group and 6.37 years in the chest x-ray group, the authors estimated that 18.5% represented overdiagnosis. They also estimated that 22.5% of non-small cell lung cancer (NSCLC) detected by LDCT was overdiagnosed, and 78.9% of bronchioalveolar lung cancer detected by CT was overdiagnosed.
"In other words, if these individuals had not entered the NLST, they would not have received a lung cancer diagnosis or treatment, at least for the next five years," Patz and his team wrote. "This is most striking in patients with [bronchioloalveolar-cell carcinoma (BAC)]," they wrote, noting that based on the most recent guidelines, many of the BACs would have been classified as minimally invasive carcinomas.
The researchers also applied a convolution model to the NLST data that explored the effects of factors such as other screening models and a longer follow-up period.
This model found that the excess cancer rates associated with three annual low-dose CT screening sessions decreased substantially when calculated for lifetime follow-up versus just the five years available to the investigators -- and these lower rates represented true overdiagnosis, the authors wrote.
Consistent with previous studies, the convolution model also demonstrated that overdiagnosis is more common with low-dose CT than with chest x-ray.
Use with caution
All models must be approached with caution, the group noted, particularly when results are extrapolated beyond the true follow-up period. In addition, models do not always reflect true practice because people and clinicians don't always behave like a model predicts. What the model does offer is a "framework" for exploring different screening scenarios and potential outcomes, the study team wrote.
Screening with low-dose CT "has the potential to detect indolent tumors, reflecting overdiagnosis," Patz and colleagues concluded. "Whereas the NLST demonstrated a relative mortality reduction with LDCT, the limitations of the screening process, including the magnitude of overdiagnosis, should be considered when guidelines for mass screening programs are considered."
In the future, better biomarkers to predict which individuals with screen-detected lung cancers will actually develop lethal tumors will render mass screening programs more valuable, they wrote.
Commenting on the results, the American College of Radiology (ACR) stated that the demonstrated mortality reduction of CT lung cancer screening "significantly outweighs the comparatively modest rate of overdiagnosis." It said it will continue to support current guidelines for lung cancer screening with CT.
"As the JAMA study authors observed, medical science largely cannot determine which cancers will never progress to harm patients and which will become lethal," ACR continued, adding that it supports the search for better biomarkers and imaging techniques to reduce potential overdiagnosis of lung and other cancers.
"Additionally, ACR advises that preparations for lung cancer screening programs proceed as the medical community continues to address this issue," the organization wrote.
