Scan-based competing death risk model for re-evaluating lung cancer computed tomography screening eligibility

Area of Science:

• Radiology
• Cardiology
• Pulmonology
• Oncology

Background:

• Baseline computed tomography (CT) scans in lung cancer (LC) screening may identify individuals who could benefit from reduced screening frequency.
• Certain participants may require focused early cardiac and respiratory clinical interventions instead of intensive LC screening.

Purpose of the Study:

• To develop and validate competing death (CD) risk models utilizing CT scan data.
• To identify participants with low LC risk but high CD risk.

Main Methods:

• A logistic regression model was developed to predict 5-year CD risk using demographic and quantitative CT measures from the National Lung Screening Trial (n=15,000).
• The model incorporated data on LC, cardiovascular disease, and chronic obstructive pulmonary disease.
• External validation was performed using the multicentric Italian Lung Detection dataset (n=2,287).

Main Results:

• The developed CD model demonstrated superior performance compared to an external pre-scan model (CD Risk Assessment Tool) in both derivation and validation cohorts (AUCs ranging from 0.677 to 0.744).
• The model identified a subgroup (27% in derivation, 34% in validation) with a significantly higher ratio of competing deaths to LC cases, suggesting potential for personalized screening strategies.
• The number needed to screen to detect one LC was substantially higher in this identified subgroup.

Conclusions:

• Integrating LC and CD risk assessments post-CT scan can enhance the efficiency of lung cancer screening programs.
• This dual-risk evaluation facilitates the timely initiation of multidisciplinary care for identified participants.
• Personalized screening and early clinical intervention can optimize resource allocation and patient outcomes.