乌鸦传媒

Screening for lung cancer with volume-based low-dose CT led to a significant reduction in lung cancer mortality at ten years, according to long-term results from a large randomized trial.

Patients assigned to the screening arm had lung cancer mortality of 2.50 cases per 1,000 person-years versus 3.3 per 1,000 in the unscreened group. The difference translated into a 24% reduction in the risk of death from lung cancer at 10 years, reported Harry J. de Koning, MD, PhD, of Erasmus Medical Center in Rotterdam, The Netherlands, and coauthors.

Women appeared to derive even greater benefit from screening, as they had a 33% lower lung cancer mortality versus unscreened women, although the difference did not achieve statistical significance. Adherence to the screening protocol exceeded 90%.

The results confirmed earlier reports from the NELSON trial showing about a 25% reduction in the relative risk of lung cancer death in screened patients, as well as the suggestion that women benefited more, as .

"Volume CT screening enabled a significant reduction of harms (e.g., false-positive tests and unnecessary workup procedures) without jeopardizing favorable outcomes," de Koning and colleagues wrote. "Trial data suggest greater benefits in women than in men, but in a subgroup with a relatively low number of women. More research is required in women, as well as in other subgroups."

The results should close the door on any uncertainty regarding the benefits of lung cancer screening with low-dose CT, said the authors of .

"With the NELSON results, the efficacy of low-dose CT screening for lung cancer is confirmed," wrote Stephen W. Duffy, MSc, of Queen Mary University of London, and John K. Field, PhD, of the University of Liverpool in England. "Our job is no longer to assess whether low-dose CT screening for lung cancer works: It does. Our job is to identify the target population in which it will be acceptable and cost-effective."

Almost a decade ago, the showed that screening with low-dose chest CT reduced the risk of fatal lung cancer in high-risk patients. The 20% reduction in relative risk persuaded policymakers in the U.S. to back the approach to lung cancer screening, but doubts about the benefits persisted, particularly in Europe. The doubts were likely fostered, in part, by early publication of inconclusive results from several small European trials, Duffy and Field noted.

The NELSON trial included 13,195 men and 2,594 women 50 to 74, all current or former smokers. The participants were randomized to undergo CT screening or no screening. Screening occurred at baseline, 1 year, 3 years, and 5.5 years at four centers in The Netherlands and Belgium. Investigators used a volume-based approach to evaluate nodules observed on CT.

Computer software was used to perform nodule segmentation and determine nodule volume. When segmentation was not feasible with the software, nodule volume was corrected manually by a radiologist. Nodule volume and volume doubling time formed the basis for rating a screening image as negative, indeterminate, or positive.

The primary endpoint was lung cancer mortality at ten years. The screened group had a higher incidence of lung cancer, 5.58 versus 4.91 cases per 1,000 person years. Analysis of the between-group difference in the rate ratio for lung cancer mortality yielded a statistically significant reduction in favor of screening (HR 0.76, 95% CI 0.61-0.94, P=0.01). Comparison of the rate ratio for screened versus unscreened women produced hazard of 0.67 (95% CI 0.38-1.14).

Subgroup analyses showed a consistently lower lung cancer mortality hazard of about 25%.

The data suggested that screening led to an overdiagnosis rate of about 10% "at worst," well below the number of lives saved, Duffy and Field noted. They also pointed to a low rate of further investigation to clarify scan results. Previous studies suggested that about 20% of participants undergo at least one additional scan. In the NELSON trial, the frequency of additional testing reached that level only at the first screening, declining to 1.9%-6.7% at subsequent screenings, resulting in an overall rate of 10%.

On the other hand, positive predictive values averaged only 43.5% across four rounds of screening.