Abstract 3485: Developing a non-invasive, diagnostic test for stage I non-small cell lung cancer using circulating tumor cells

Introduction: Reliable, non-invasive diagnostic blood tests to assess high-risk pulmonary nodules are needed to avoid unnecessary invasive procedures and the risk of missing lung cancer. Methods: We used a prospective, observational, case-control study design of consecutively enrolled patients undergoing imaging for a lung nodule or diagnosed malignancy to investigate whether a non-EpCAM based circulating tumor cell (CTC) platform could accurately diagnose stage I non-small cell lung cancer (NSCLC). Phlebotomy was performed according to standardized protocols at all three participating medical centers and the number of CTCs was analyzed in a single blinded fashion on a continuous scale per mL of blood. For determining test characteristics, cases consisted of patients with stage I NSCLC, and controls consisted of advanced NSCLC, malignant non-NSCLC lung nodules, and benign lung nodules. Differences by disease group were compared formally by ANOVA or Chi-square analysis, and test performance for identifying stage I NSCLC compared to benign disease was assessed by sensitivity, specificity and C statistics. Results: Seventy-three of 79 samples were adequate for analysis after processing. The median age was 69 years (interquartile range [IQR] 65-76), 73% were male, 70% were Caucasian, 79% had a smoking history, and 45% had a history of cancer. Diagnoses consisted of 56 NSCLC patients, of which two-thirds were adenocarcinoma and 45 were stage I, 8 malignant non-NSCLC cases, and 9 benign nodules. There were no differences in basic clinical characteristics across diagnosis type other than plasma white blood cell count. Average nodule diameter was 2.2 cm (IQR 1.6-2.8) and this did not vary significantly by diagnosis. Median CTC/ml was 3.3, 3.0, 0.60 and 0.70 for Stage I NSCLC, Stage II-IV NSCLC, other non-NSCLCs, and benign nodules respectively. At a threshold of 1.0 CTC/mL, 2.0 CTC/mL and 5.0 CTC/mL respectively 63%, 54% and 44% of stage I NSCLCs had a detectable CTC burden. For these three thresholds sensitivity (89%, 89%, 100%) and specificity (64%, 53%, 44%) was calculated for discriminating stage I NSCLC from benign disease. The accuracy (C-statistic) for identifying stage I disease only relative to benign disease only improved from 0.84 to 0.91 when adding CTC/mL data to age, gender, smoking and cancer history, nodule size and location, and SUVmax. Discussion: CTCs may be useful for identifying stage I NSCLC from benign lung nodules. Addition of these data to