Risk-based lung cancer screening tools can prevent significantly more lung cancer deaths than the current United States Preventive Services Task Force recommendations, but life-year gains were negligible or reduced and patients would experience greater overdiagnosis, according to new research.
“Current guidelines propose screening eligibility using age and smoking-related criteria, through combinations of accumulated pack-years and years since smoking cessation,” Kevin ten Haaf, PhD, of the department of public health at Erasmus Medical Center in Rotterdam, the Netherlands, and associates wrote in the Journal of the National Cancer Institute. “The USPSTF recommends annual screening between the ages of 55 and 80 years for current and former smokers (quit less than 15 years) who smoked 30 or more pack-years.”
Individual risk assessment utilizing established lung cancer risk–prediction models may have some superiority over pack-years in identifying those most likely to benefit from screening, they wrote, because the models incorporate smoking history in greater detail and consider risk factors such as chronic obstructive pulmonary disease.
Three risk-assessment models were used for the study, in addition to the USPSTF guidelines: the Bach, PLCOm2012, and Lung Cancer Death Risk Assessment Tool (LCDRAT). The study population was a simulated 1950 U.S. cohort from the Smoking History Generator aged between 55 years and 80 years; each simulated smoking history consists of whether and when the person initiates and ceases smoking, average number of cigarettes smoked per day by age, and the age of death from non–lung cancer causes.
The number of lung cancer deaths averted was significantly higher in the risk-based models, compared with the USPSTF recommendations (Bach, 693 per 100,000 population; PLCOm2012, 698 per 100,000 population; LCDRAT, 696 per 100,000 population; USPSTF, 613 per 100,000 population).
However, life-years gained was only modestly higher in the models, compared with the guideline (Bach, 8,660 per 100,000 life-years; PLCOm2012, 8,862 per 100,000 life-years; LCDRAT, 8,631 per 100,000 life-years; USPSTF, 8,590 per 100,000 life-years). In addition, life-years gained for every lung cancer death prevented was greater in the guideline (14.0 years) than in the risk-based models (12.1-12.4 years).
Overdiagnosis was also more common using risk-based tools (Bach, 149 per 100,000; PLCOm2012, 147 per 100,000; LCDRAT, 150 per 100,000; USPSTF, 115 per 100,000). This was mainly because of eligibility for risk-based screening tools increasing with age, the investigators noted.
According to a sensitivity analysis, risk-based models would retain the life-years gained by the USPSTF model if individuals with limited life expectancies (less than 5 years) were excluded. This would also reduce overdiagnosis by 65.3%.
“Future studies should investigate the cost-effectiveness of risk-based screening and the potential for reducing overdiagnosis in high-risk individuals,” the investigators concluded.
One coauthor developed the PLCOm2012 model, but the model is available free to noncommercial users, and the investigator has received no money from its usage. No other conflicts of interest were reported.
SOURCE: ten Haaf K et al. J Natl Cancer Inst. 2019 Nov 29. doi: 10.1093/jnci/djz164.