Lung cancer is the leading cause of cancer death worldwide, with overall 5-year survival rates in the United States of 15% but approaching 50% when diagnosed at an early stage. The National Lung Screening Trial (NLST) reported that low-dose computed tomography (LDCT) screening reduced lung cancer mortality by 20% in adults who were at high risk of lung cancer. These dramatic results come with high human and societal cost because of the extremely low yield associated with the screening criteria and high false positive rate by LDCT. NLST entry criteria, based on smoking history and age, yielded 1 lung cancer for every 156 screened. One quarter of those screened required expensive, sometimes invasive diagnostic work-up, yet 96.4% of them turned out to be false positives. A better lung cancer risk prediction model can provide better selection criteria and make LDCT more effective in balancing benefit versus harm. This study team has developed 10 blood-based biomarkers (protein: pro-SFTPB, HE4, IGFBP2, LRG1; lipid: DAS; autoantibody: LTF, ADCK1, STK10, TRIM10, KM2) and validated them using pre-diagnostic sera from the Beta-Carotene and Retinol Efficacy Trial. Pro-SFTPB was further validated on the Pan-Canadian Early Detection of Lung Cancer Screening Study and Physician Health Study and shown to complement lung cancer risk prediction models based on epidemiologic data. Recently, 4 circulating inflammation biomarkers (CRP, IL-1RA, BCA- 1/CXCL, MDC/CCL22) were found to be independently associated with lung cancer risk. The proposed study will incorporate these 14 biomarkers into PLCOm2012, a 6-year lung cancer risk prediction model developed and validated by this team using PLCO epidemiological data, to improve lung cancer risk prediction.
In Aim 1, using a nested case-control study design these 14 biomarkers will be assayed using sera collected at baseline from 549 lung cancer patients diagnosed within 6 years after baseline and 1,098 matched controls, and then incorporated into the PLCOm2012 model to improve lung cancer risk prediction and the selection criteria for LDCT.
In Aim 2, sera collected annually up to five years since baseline for these subjects will be analyzed using two Bayesian models to incorporate biomarker trajectories to improve early detection of lung cancer.
In Aim 3, the models from Aim 1&2 will be evaluated for their potential clinical utilities. If successful, the proposed study will challenge the paradigm of epidemiological modeling (age, smoking history, etc.) for lung cancer risk prediction and single threshold for early detection, improve selection criteria for LDCT screening, increase yield of lung cancer by LDCT screening, reduce LDCT-associated harm, and improve early detection of lung cancer.
The proposed study will develop better selection criteria for lung cancer screening with low-dose computed tomography. This will increase the ratio of lung cancer identified among the people who are screened, and thereby reduce lung cancer mortality, while sparing people who have low risk for lung cancer from receiving unnecessary procedures, therefore reducing screening- related harm.
| Tayob, Nabihah; Stingo, Francesco; Do, Kim-Anh et al. (2018) A Bayesian screening approach for hepatocellular carcinoma using multiple longitudinal biomarkers. Biometrics 74:249-259 |
| Bantis, Leonidas E; Yan, Qingxiang; Tsimikas, John V et al. (2017) Estimation of smooth ROC curves for biomarkers with limits of detection. Stat Med 36:3830-3843 |
