An article published recently in the Journal of Clinical Oncology presented a case study to highlight the value of using risk-prediction models as tools for patient-clinician communication on lung cancer screening. The study was authored by Stanford Cancer Institute members Summer Han, PhD, and Heather Wakelee, MD, PhD, and Stanford Cancer Institute trainee Julie Tsu-Yu Wu, MD, PhD.
“Lung cancer is one of the most deadly diseases because it’s often identified at more advanced stages. Implementing more sensitive screening would help catch the disease when it is more treatable and decrease mortality,” said Han.
Current U.S. Preventive Services Task Force (USPSTF) lung cancer screening criteria only consider age and smoking history for screening eligibility, while the National Comprehensive Cancer Network recommends an individualized risk assessment that looks at family history, environmental exposures, and presence of chronic lung disease.
Risk prediction models that take these additional risk factors into account have been created to identify patients at risk for lung cancer. Previous studies have shown these models have improved sensitivity, larger mortality benefits, and higher cost-effectiveness compared to screenings based on USPSTF criteria alone, however risk prediction models can also increase the risk of false positives, unnecessary procedures, and overdiagnosis. Identifying lung cancer biomarkers can be included as an adjunct to increase specificity, but lung cancer biomarker risk assessments currently lack validation. Further, there are outstanding questions and challenges for the wide-spread implementation of risk prediction models, such as defining key risk factors and choosing models that integrate risk factors.
Despite these challenges, the article posits that risk prediction models have utility in the clinical setting as tools for shared decision-making between the clinician and patient by helping the patient better understand their likelihood to benefit from screening. The article presents a case study of a patient who did not meet the USPSTF lung cancer screening criteria only to be diagnosed with lung cancer within a year of seeing his primary care provider. Risk prediction models would have been able to identify this patient as being at risk for lung cancer based on other criteria.
“At Stanford, we’re fortunate to have a dedicated, robust lung cancer screening program where we use these models as shared-decision-making tools as part of our practice. The hope is that these models are used in a clinical setting to help clinicians facilitate a conversation with the patient about the benefits and risks of screening so that they can reach a decision that makes sense for the patient’s unique circumstances,” said Han.
August 2023
By Katie Shumake
