PREDICT1: An observational study for identifying blood biomarkers associated with clinical benefit from carboplatin and pemetrexed (CbP) treatment in patients with non-squamous (NS) non-small cell lung cancer (NSCLC) (CJLSG1201)

9524 Background: At present, platinum-doublet chemotherapy or in combination with an immune check point inhibitor are standard treatment for patients with metastatic or recurrent NSCLC which lacks somatic gene alterations. Although CbP is one of the commonly used treatment options for NS-NSCLC, its clinical utility is limited due to lack of optimal biomarkers. Methods: Chemotherapy-naïve patients with pathologically proven advanced or recurrent NS-NSCLC received carboplatin (area under the curve = 5-6, at investigator’s discretion) plus pemetrexed (500 mg/m2) every 3 weeks followed by maintenance pemetrexed until disease progression. Blood samples were collected before treatment for proteomic analysis using mass spectrometry (MS). A classifier was constructed based on both an objective response assessed by radiologist independent of attending physicians in accordance with RECIST v1.1 and expression profiles of protein in a training cohort. The constructed classifier was then assessed with a validation cohort evaluating prediction accuracy of good responder, progression free survival (PFS) and overall survival (OS). Results: Of 244 patients with NS-NSCLC in a training cohort, proteomic profiles in blood from 96 patients who responded or progressed after treatment with CbP were analyzed to develop a classifier based on weighted voting. Details of the classifier will be presented at the 2020 ASCO Annual Meeting. The classifier was then applied to validation cohort (n = 94), and we successfully identified patients who benefit from the treatment (55 in good MS group) or not (39 in poor MS group). The objective response rate of the good MS group was significantly higher than that of the poor MS group (30.9% vs. 5.1%; p = 0.0018). The good MS group showed a significantly improved survival compared to the poor MS group (median PFS, 6.0 m vs. 2.3 m; hazard ratio [HR], 0.15; 95% confidence interval [CI], 0.09-0.27; p < 0.001; median OS, 25.7 m vs. 5.1 m; HR, 0.18; 95% CI, 0.1-0.34; p < 0.001). Conclusions: In the present study, we successfully developed and validated a predictive classifier using proteomic analyses with blood samples collected from patients before treatment with CbP, suggesting the clinical utility of the classifier in selecting NS-NSCLC patients for treatment with CbP. Clinical trial information: UMIN000008476 .