Highlight
- Lung cancer screening reduces mortality differently across subgroups defined by pack-years, smoking status, sex, and histology.
- Screening effectiveness ranges markedly, with better outcomes observed in former smokers, women, and cancers with favorable histology such as adenocarcinoma and other non-small cell types.
- Squamous-cell carcinoma showed inconsistent screening benefits between NLST (no benefit) and NELSON (substantial mortality reduction).
- Histological differences account for much of the observed heterogeneity in screening effectiveness, suggesting that relaxing smoking-related eligibility may improve screening impact.
Study Background and Disease Burden
Lung cancer remains the leading cause of cancer mortality globally with poor overall survival due to late diagnosis. Early-stage detection through low-dose computed tomography (LDCT) screening has emerged as an effective mortality reduction strategy. Landmark randomized controlled trials (RCTs), including the National Lung Screening Trial (NLST) in the United States and the Dutch-Belgian Lung Cancer Screening Trial (NELSON), have demonstrated over 20% reduction in lung cancer-specific mortality through screening high-risk populations.
Despite this, lung cancer screening effectiveness is not uniform across all patients and may vary according to personal risk factors such as smoking intensity, smoking status, sex, and tumor histology. This heterogeneity complicates population-based screening guidelines and may limit the overall benefit when strict eligibility criteria are applied. Understanding the sources of variability in screening benefit is essential to optimizing screening programs, improving personalized screening strategies, and expanding eligibility criteria to balance benefits and harms more effectively.
Study Design
This comparative analysis leverages individual-level data from two pivotal RCTs: the NELSON trial and the NLST. The NELSON trial enrolled 14,808 participants (predominantly men) from the Netherlands and Belgium, while the NLST included 53,405 participants from the United States, with a more balanced sex distribution.
Both trials compared lung cancer screening via low-dose CT against no screening or chest radiography, respectively, with the primary endpoint of lung cancer-specific mortality. The study employed traditional subgroup analyses and advanced predictive modeling using machine learning to assess heterogeneity of screening effectiveness based on pack-years of smoking, smoking status (current vs former smoker), sex, and lung cancer histological subtypes.
Key Findings
### Screening Effectiveness by Smoking Intensity (Pack-Years)
Screening effectiveness, defined as the relative reduction in lung cancer mortality, varied substantially across smoking intensity groups in both trials. In lower pack-year cohorts, effectiveness ranged from 26.8% to 50.9%, suggesting robust benefit. In contrast, the highest pack-year groups exhibited more modest benefits, from 5.5% to 9.5%. This finding challenges assumptions that highest cumulative smokers derive the greatest absolute benefit and suggests diminishing returns in very heavy smokers, potentially due to more aggressive cancer types less amenable to early detection.
### Influence of Smoking Status
Former smokers consistently experienced greater screening effectiveness (37.8%-39.1% reduction) compared to current smokers (16.1%-22.7%) across both trials. This difference aligns with the biological plausibility of tumor characteristics and lung parenchymal changes evolving after smoking cessation that may affect screening sensitivity and disease prognosis.
### Sex Differences
Women benefited more from screening than men, with mortality reductions of approximately 24.6%-25.3% versus 8.3%-24.9%, respectively. Possible explanations include variations in histological tumor types and differential biological susceptibility to tobacco carcinogens, underlining the need for sex-specific screening considerations.
### Heterogeneity by Histological Subtype
Heterogeneity analysis by lung cancer histology revealed that screening was more effective against adenocarcinoma (17.8%-23.0% mortality reduction) and other non-small cell lung cancers (24.5%-35.5%), with less pronounced benefits for small-cell carcinoma (9.7%-11.3%). Critically, squamous-cell carcinoma demonstrated trial-dependent outcomes: ineffective screening in NLST—with a mortality increase that was not statistically significant—but a significant 52.2% mortality reduction in NELSON. This disparity may reflect differences in screening protocols, population characteristics, or tumor biology between trials.
### Explaining Screening Benefit Variability
Statistical modeling indicated that the predominance of histologies with favorable screening responsiveness within subgroups (e.g., former smokers, women) accounted for the majority of observed variability in screening effectiveness by pack-years, smoking status, and sex. This insight underscores histology as the primary driver of screening benefit heterogeneity.
### Implications for Screening Eligibility Criteria
The findings support considering relaxed smoking-related eligibility thresholds for lung cancer screening to capture individuals with histologies more responsive to early detection, potentially improving screening yield and mortality benefit.
Expert Commentary
This comparative study elucidates the complex interplay between individual risk factors and lung cancer histology in determining screening effectiveness. It addresses a critical knowledge gap by integrating data from two of the largest lung cancer screening trials, thus enhancing generalizability.
However, there are limitations including demographic differences between trial populations and heterogeneity in screening protocols. The discrepancy in squamous-cell carcinoma outcomes invites further mechanistic investigation. Moreover, translating histology-driven insights into clinical practice will require refining risk models and developing biomarkers predictive of tumor biology.
Emerging machine learning approaches integrated in this analysis highlight the future direction for precision screening, tailoring interventions to individuals’ risk profiles and tumor characteristics. Incorporating these findings into clinical guidelines could optimize screening benefits and reduce harms related to overdiagnosis and unnecessary interventions.
Conclusion
Lung cancer screening significantly reduces mortality but shows marked heterogeneity across individual risk factors and tumor histology. Histological subtype emerges as the principal determinant of screening effectiveness, influencing observed differences by smoking intensity, smoking status, and sex.
Relaxing rigid smoking-related screening criteria could enhance population benefit by including individuals more likely to develop cancers with favorable screening profiles. These findings advocate for an evolution toward more personalized lung cancer screening strategies that integrate tumor biology and patient factors.
Future research should validate these results in diverse populations and explore biomarkers for predicting histological subtypes to guide individualized screening and early therapeutic strategies.
References
Welz M, van der Aalst CM, Alfons A, Naghi AA, Heuvelmans MA, Groen HJM, de Jong PA, Aerts J, Oudkerk M, de Koning HJ, Ten Haaf K; NELSON trial consortium. A comparative analysis of heterogeneity in lung cancer screening effectiveness in two randomised controlled trials. Nat Commun. 2025 Aug 28;16(1):8060. doi: 10.1038/s41467-025-63471-6. PMID: 40877276; PMCID: PMC12394595.