Publication-only abstracts (abstract number preceded by an "e"), published in conjunction with the 2018 ASCO Annual Meeting but not presented at the Meeting, can be found online only.
A new algorithm predicting imminent disease progression in advanced NSCLC patients by machine-learning integration of five serum biomarkers.
Metastatic Non-Small Cell Lung Cancer
Lung Cancer—Non-Small Cell Metastatic
2018 ASCO Annual Meeting
J Clin Oncol 36, 2018 (suppl; abstr e21190)
Author(s): Yuri Kogan, Marina Kleiman, Shmuel Shannon, Moran Elishmereni, Eldad Taub, Larisa Aptekar, Ronen Mordechay Brenner, Raanan Berger, Hovav Nechushtan, Zvia Agur; Optimata Ltd., Bene Ataroth, Israel; Hadassah Medical Center, Jerusalem, Israel; Wolfson Medical Center, Tel Aviv, Israel; Sheba Medical Center, Ramat Gan, Israel
Background: Non-small cell lung cancer (NSCLC) is the leading cause of cancer death in America. Predicting disease progression just prior to its clinical manifestation would allow an earlier switch to the next treatment line, thus preventing major decline in the patient's state and potentially improving response to therapy. Yet, present serum tumor biomarkers, e.g. carcinoembryonic antigen (CEA), lack the power to foresee progression. We developed PrediCare, a new diagnostic algorithm for continuous monitoring and alerting to approaching progression in late-stage NSCLC, based on input from several tumor markers. Methods: The PrediCare algorithm was designed to process patient data in treatment by machine-learning modeling. Data of late-stage NSCLC patients under first-line standard-of-care therapies, collected in a retrospective observational trial (NCT02577627), served for algorithm training/testing. The algorithm’s predictive ability was evaluated using diverse features of 1-5 longitudinally measured serum tumor markers (CEA, CA125, CA15.3, CA19.9, NSE), as pre-selected by receiver-operating-characteristic analysis. Performance was evaluated by cross-validation. Results: NSCLC patients (n = 167) were assessed; median follow-up time was 101 days. The 5-marker combination showed strong prediction ability, with 66% sensitivity (accurate prediction of 109/165 progression events) and 91% specificity (15/165 false positives). Positive and negative predictive values, accuracy and Cohen’s kappa were 88%, 70%, 84% and 0.58, respectively. The algorithm’s performance with all five markers was superior to that of our previously shown 3-marker algorithm, and significantly surpassed the limited capacity (sensitivity/specificity) of each of these markers when used separately. Conclusions: Our algorithm is new in integrating five standard tumor biomarkers in a unique way for predicting imminent disease progression in advanced NSCLC. Use of our algorithm offers superiority over current interpretation of these markers in the clinic, and may aid in improving treatment navigation as well as potentially increasing survival.
1. Pembrolizumab (pembro) versus platinum-based chemotherapy (chemo) as first-line therapy for advanced/metastatic NSCLC with a PD-L1 tumor proportion score (TPS) ≥ 1%: Open-label, phase 3 KEYNOTE-042 study.