Lung cancer screening works - but how well depends on who you ask. The two largest randomized trials of CT lung cancer screening have confirmed that screening with low-dose CT can reduce lung cancer deaths by 20-24%. However, these are population averages. The real question is whether screening works equally well for all types of eligible individuals, or whether some benefit far more than others.
Individual risk-factors may modify screening effectiveness. Factors like age, smoking history, and sex all influence a person's overall lung cancer risk. But the critical question this study asks is whether these factors also modify how much benefit someone gets from screening - a concept called 'heterogeneity in treatment effect.' A person at high risk of lung cancer may or may not be the same person who gains the most from screening.
Previous analyses used one variable at a time. Earlier attempts to identify who benefits most from screening compared subgroups based on a single characteristic - for example, comparing current versus former smokers. But this approach misses interactions between risk-factors and is prone to both false-positive and false-negative conclusions due to limited statistical power when splitting already modest trial populations into subgroups.
The methodological challenge: comparing multiple approaches. This study used individual-level data from the two largest lung cancer screening trials - NELSON (14,808 participants) and NLST (53,405 participants) - and applied three different analytical approaches simultaneously: traditional subgroup analyses, statistical risk-prediction models, and machine learning (causal forests). Comparing results across methods helps identify which findings are robust rather than artifacts of any single approach.
Former smokers benefit more than current smokers. Across both trials, screening was more effective for former smokers than current smokers. In NELSON, median screening effectiveness was 37.8-39.0% for former smokers versus 22.7% for current smokers. In NLST, the difference was most pronounced for long-term quitters (those who had stopped smoking 10 or more years ago), who showed 39.1% effectiveness versus 16.1% for current smokers.
Fewer pack-years means greater screening benefit. Counterintuitively, people with fewer lifetime cigarettes smoked - those with fewer pack-years - benefited more from screening. Screening effectiveness in the lowest pack-year groups reached 50.9% (NELSON) and 26.8% (NLST), compared to just 5.5-9.5% in the highest pack-year groups. This means the heaviest smokers, who carry the highest absolute lung cancer risk, appear to derive the least relative benefit from CT screening.
Women benefit at least as much as men. In NELSON, screening effectiveness was similar for women (25.3%) and men (24.9%). In NLST, women showed greater benefit (24.6%) than men (8.3%). This finding has implications for screening programs historically dominated by male participants in trials, and suggests that women should not be considered lower-priority candidates for lung cancer screening.
Age does not substantially modify screening effectiveness. Across both trials, screening effectiveness varied little by age group within the eligible screening age range (50-74 years). This suggests that within the currently recommended screening window, age alone should not be a primary factor in deciding whether an eligible individual will benefit from screening.
Histology is the primary driver of screening effectiveness. The central finding of this study is that tumor type - the specific cancer subtype - is the main factor determining whether screening reduces mortality. When the researchers analyzed effectiveness separately by cancer type rather than by patient characteristics, a clear pattern emerged that explained the differences seen across smoking, sex, and pack-year groups.
Adenocarcinoma: consistently responsive to screening. Adenocarcinoma, the most common lung cancer subtype, showed consistent and meaningful screening effectiveness across both trials: median mortality reductions of 17.8-23.0%. Adenocarcinomas typically grow in the outer lung regions, making them more visible and detectable at earlier, more curable stages on CT scans.
Small cell lung cancer: poorly served by screening. Small cell carcinoma showed the weakest screening response, with only 9.7-11.3% mortality reduction across both trials. Small cell cancers grow rapidly, are centrally located, and tend to have already spread beyond the chest by the time they are detectable on imaging - making the window for early detection very narrow regardless of screening frequency.
Squamous cell carcinoma: discordant results between trials. A striking and unexplained discordance emerged for squamous cell carcinoma. NELSON showed a robust 52.2% mortality reduction, while NLST showed no benefit - in fact, a 27.9% mortality increase (though confidence intervals were wide and overlapping zero). The authors suggest this may partly reflect differences in nodule measurement protocols: NELSON used automated volume measurements, which are more sensitive to growth than the manual diameter measurements used in NLST.
The explanation for the smoking paradox. Why do lighter smokers and former smokers benefit more from screening than heavy current smokers? The answer lies in cancer biology. People with fewer pack-years and those who quit smoking have a higher proportion of adenocarcinomas in their cancer distribution, and a lower proportion of small cell cancers. Because adenocarcinoma responds better to screening, groups with more adenocarcinoma appear to benefit more from screening overall.
Women benefit more because they get more adenocarcinoma. Similarly, the greater screening benefit seen in women compared to men in NLST is explained by the higher prevalence of adenocarcinoma among women who develop lung cancer. Women who smoke tend to develop adenocarcinoma more often than men, who more frequently develop squamous cell and small cell cancers. The screening benefit flows through histology, not directly through sex.
Where cancers grow determines whether screening can find them early. Adenocarcinomas typically develop in the periphery of the lungs, where they can be detected as small isolated nodules on CT before causing symptoms. Small cell cancers tend to arise centrally around the major airways, and their rapid growth means they are often advanced before they appear as distinct detectable masses. These anatomical and growth characteristics largely determine screen-detectability independent of how carefully or frequently screening is performed.
Absolute versus relative benefit: higher-risk patients gain more in absolute terms. While the relative (percentage) reduction in mortality was similar across cancer risk groups, the absolute number of deaths prevented was greater in higher-risk individuals. This means that in populations enriched for high lung cancer risk, screening prevents more deaths per person screened - even if the proportional effectiveness is similar. This supports targeting screening programs toward higher-risk populations as a matter of efficiency.
Expanding screening to lighter smokers is supported by histology data. Recent US guideline changes (USPSTF 2021) lowered the pack-year threshold for screening eligibility, and the American Cancer Society has recommended removing restrictions on years since smoking cessation. This study provides biological support for these expansions: people who smoked less and those who have quit longer have a higher prevalence of adenocarcinoma, the cancer type that benefits most from screening.
Racial and ethnic equity in screening. The findings have implications for health equity. Individuals of African-American ancestry are more likely to develop squamous cell carcinoma than those of European ancestry. Given the discordant squamous cell screening results between trials, further research specifically evaluating lung cancer screening effectiveness in African-American populations is warranted before assuming equal benefit.
Never-smokers with family history may be a candidate population. The study found a potential association between family history of lung cancer and adenocarcinoma risk. In regions where lung cancer in never-smokers is common - particularly in Asian populations - and given that adenocarcinoma is the predominant histotype in never-smokers, investigating screening for this population deserves further study.
Smoking cessation integration enhances screening effectiveness through histology. The study identifies a previously underappreciated mechanism by which smoking cessation support boosts the effectiveness of screening programs. Quitting smoking not only reduces overall lung cancer risk but also prevents further accumulation of pack-years, which this study shows is associated with lower screening effectiveness. Programs integrating cessation support with screening therefore maximize benefit through multiple pathways simultaneously.