New research finds the protocol used to screen and detect lung cancer in the NELSON Trial is more sensitive than the protocol used in the National Lung Cancer Screening Trial (NLST), especially for early-stage cancers. Presented at the 2022 IASLC World Conference on Lung Cancer in Vienna, a study found that the NELSON protocol-Nederlands-Leuvens Longkanker Screenings Onderzoek-demonstrated a reduction in lung cancer mortality of 24 percent for screening with low-dose computed tomography (CT), compared to the 20 percent found in the NLST, according to an IASLC statement summarizing the presentation.
The NELSON and NLST trials differed in study population and trial design. Specifically, in contrast to the NLST, the NELSON trial employed a nodule management protocol that incorporated nodule volume and quantified volume growth as opposed to nodule diameter. Koen de Nijs, a PhD student at the Erasmus University Medical Center in Rotterdam, the Netherlands, and colleagues, employed the MISCAN-Lung model, which was previously used to evaluate the results of the NLST, to evaluate the outcomes of the NELSON trial. The model was used to reproduce lung cancer incidence and mortality by method of detection (clinical or screen-detected), sex, histology, and stage.
The researchers evaluated the potential differences in CT sensitivity by stage and histology, after accounting for the study population's characteristics, trial design, and lung cancer epidemiology in each trial. For both trials, the investigators considered a screening result to be a true positive if lung cancer was detected through the screening CT and related follow-up procedures.
The team found that the sensitivity in NELSON was estimated to be higher across all stages compared with the NLST. CT sensitivity was significantly higher for early-stage adenocarcinoma. For example, sensitivity in detecting Stage IA cancer was 73 percent in NELSON versus 52 percent in the NLST. For Stage IB, NELSON demonstrated a 90 percent sensitivity rate, compared to 64 percent for NLST.
"The differences in screening outcomes between the NELSON trial and the NLST were already known," de Nijs noted. "The NELSON trial showed a greater mortality reduction and larger proportion of early-stage cancers compared to the NLST."
There were, however, key differences between the two trials that may have contributed to these outcomes, he added, noting the number of screens in each trial (four in NELSON, compared to three in the NLST), the screening interval (2.5 years in NELSON and 1 year in the NLST), and the characteristics of the study population.
"Previous research had already shown that the NELSON trial was associated with less unnecessary follow-up procedure. However, reduced rates of follow-up may also lead us to speculate whether this is not associated with a reduced rate of early detection," stated de Nijs, adding that this study employed a lung cancer natural history model, "which allows us to control for the differences in study characteristics, by exactly replicating the circumstances of each trial in the model simulations.
"The estimates of the increased sensitivity indicate that, controlling for each of these elements, volume-based nodule management-the key remaining difference between the two trials-offers a higher probability of detection in spite of lower rates of follow-up," he continued.
Previous research has already shown that volume doubling time-the key metric of malignancy in the NELSON protocol-is highly predictive of lung cancer, according to de Nijs. "We interpret the result as indicating that consistent application of a volume-based nodule management protocol is associated with increased rates of early detection. This would suggest that diameter growth, while still predictive of lung cancer, is less discriminative between malignant and benign nodules."
de Nijs is hopeful these findings can help inform policies with regard to the nodule management protocol. "Volume-based measurement may require investment and training, such as for the use of computer-aided detection systems. However, in a comparison of the volume-based NELSON trial and the diameter-based NLST, we find considerable differences in the probability of the detection of disease at an early stage."
The results may also inform recommendations of future screening programs, he noted, pointing out that many published estimates of cost-effectiveness are still based on the results of the NLST.
"Any program employing volume-based nodule management may see greater benefits from the required investment, improving the cost-effectiveness of lung cancer screening on a population scale," said de Nijs. "Similarly, higher estimates of CT sensitivity may inform whether resource-limited lung cancer screening programs could achieve significant mortality reductions with more sparse screening."
Looking ahead, de Nijs' research group, together with a consortium of European partners, is leading the 4-IN-THE-LUNG-RUN trial, which will investigate the benefits of referring individuals with a negative baseline screen to a biannual screening interval rather than an annual screening interval.
"A higher CT sensitivity, such as demonstrated in this study, for detecting early-stage lung cancer is key to such a personalized deferment to biannual screening. Increased confidence that an individual is free of lung cancer at their baseline scan means that they are more likely to be equally well served with a 2-year repeat scan."
Mark McGraw is a contributing writer.