Raising the lung cancer CT screening threshold may curb false positives

Increasing the lung nodule threshold for a positive finding on an initial lung cancer screening from 5 mm to 7 or 8 mm in the baseline CT screening round may provide an optimal balance between downstream work-ups and delays in diagnosis, according to a study published online Feb. 18 in Annals of Internal Medicine. Although the “provocative data could substantially” reduce screening costs, the approach may overlook other risks, according to an accompanying editorial.

As lung cancer screening programs expand, concerns about false positives, excessive diagnostic work-ups and costs have grownn. “As large national programs screen millions of high-risk participants, the definition of a positive result will have enormous cost implications,” wrote Claudia I. Henschke, PhD, MD, from Mount Sinai Medical Center in New York City, and colleagues.

The International Early Lung Cancer Action project (I-ELCAP) has applied a 5 mm threshold to define a positive finding in the initial screening round, resulting in a 16 percent positive-finding rate. The National Lung Screening Trial (NLST) used a 4 mm threshold. The positive result rate in the baseline screening round in the NLST was 28 percent.

Henschke and colleagues explored the implications of using progressively higher thresholds for nodule size in the baseline round of screening.

The researchers focused on 21,136 I-ELCAP participants who underwent baseline CT screening for lung cancer between 2006 and 2010. Although the positive result rate was highest among participants with noncalcified nodules between 5 and 9 mm, the frequency of lung cancer diagnosis within 12 months in this group was 0.3 percent.

At the 5-mm threshold, 3,396 participants had a positive result. The figure dropped to 2,159; 1,498; 1,077 and 838 for thresholds of 6 mm, 7 mm, 8 mm and 9 mm, respectively. The less restrictive thresholds would reduce the work-up in the baseline round by 55.9 percent at the 7-mm threshold and 75.3 percent at the 9-mm threshold.

However, the diagnosis of lung cancer within 12 months would have held steady at 119 cases at both the 5- and 6-mm threshold, and dropped to 113, 112, and 111 at the 7-mm, 8-mm and 9-mm thresholds, respectively.

“Our results suggest that changing the threshold for a parenchymal, solid or part-solid NCN [noncalcified nodule] to 8.0 mm may be reasonable in the baseline round of CT screening,” wrote Henschke et al. Although delays in diagnosis may lead to stage progression and reduced curability in some cases, the researchers used I-ELCAP data to infer that more than 90 percent of 7-mm nodules would remain stage I if diagnosis were delayed 12 months.

Henschke and colleagues emphasized the need for continual evaluation and revision of the positive finding threshold, which they said will reduce unnecessary surgery and maximize the diagnosis and treatment of curable lung cancers.

However, an accompanying editorial expressed some reservations about increasing the nodule cut point.

Stephen Lam, MD, from the British Columbia Cancer Agency in Vancouver, British Columbia, Canada, and colleagues noted the 4-mm threshold was used in the NLST. Other variables, such as the presence or absence of emphysema or nodule speculation, also may predict cancer, they added.

Lam et al referred to their work on a comprehensive computer-based risk prediction model that includes patient demographics and CT features such as nodule size, type and spiculation to help estimate risk and guide clinical decisions. “To be accurate, such models must incorporate more information and provide probabilities of lung cancer over a range of nodule sizes, not just for 1 cut point,” they wrote.

Henschke et al pointed out that imaging biomarkers may emerge as the lung cancer CT screening experience base grows.