P1.01-69 Blood Serum Amyloid A as Potential Predictor of Response to First-Line Pembrolizumab in Patients with Advanced Non-Small-Cell Lung Cancer

Abstract

The selection of patients (pts) deriving clinical benefit from immune checkpoint agents represents the modern challenge for clinicians. Tumor-derived Serum Amyloid A (SAA) inhibits the immune response through the expansion of IL-10-secreting neutrophils in pts with melanoma. We investigated the predictive value of blood SAA monitoring in a cohort of Advanced Non-Small-Cell Lung Cancer (ANSCLC) pts receiving up-front Pembrolizumab (P). Pts with ANSCLC (PD-L1≥50%) receiving upfront P, were prospectively evaluated for blood SAA and radiological response at baseline and every 9 weeks during the treatment. The primary endpoint was response rate (RR) according Immune-related Response Evaluation Criteria in Solid Tumors (IrRECIST); the secondary endpoints were progression-free survival (PFS) and overall survival (OS). The most accurate SAA cut-off to predict response was established with a ROC-analysis. Forty-two patients were enrolled. Pts characteristics: male/female (71/29%), number of sites 1/2/3/≥ 4 (5/38/40/17%), ECOG PS 0/≥ 1 (38/62%); never or former/current smokers (12/78%); median age 70.5 (range 35-86) years. The overall RR was 38% (95%CI 25-53%). After a median follow-up of 13.5 months (m), baseline SAA ≤ the ROC-derived cut-off (29.9 mg/L, AUC 0.74, 95% CI 0.59-0.87, p=0.002; 14 [33%] pts) was significantly associated with a higher RR (53.6vs 7.1%, OR 15, 95% CI 1.72-130.7, p<0.01), longer PFS (17.4 vs 2.1 m, HR 0.18, 95%CI 0.06-0.51, p<0.0001) and OS (not reached [NR]vs7.2 m, HR 0.08, 95%CI 0.02-0.39, p<0.0001) compared with SAA >29.9 mg/L. Multivariate analysis confirmed pre-treatment low SAA as independent predictor of longer PFS (p=0.029) and OS (p=0.018). Considering SAA at baseline and the dynamic monitoring (pts=40), the median PFS was 17.4 m (95% CI 10.7-14.7) when SAA remained low (n=14) compared with 2.1 m (95%CI 1.3-5.6) when SAA remained high (n=12) (p<0.0001).The SAA monitoring was also significantly associated with OS (p=0.0002), with a median OS of 7.2 m (95% CI 5.6-13.4) in pts maintaining high SAA versus NR median at 18-m for pts with SAA remained low or changed. No deaths occurred in the permanently low-SAA group at the current data-lock. Baseline low SAA predicts good outcomes of 1stline P and the SAA monitoring throught simple blood test could help to easily identify patients who derived the greatest benefit from immunotherapy. The strong relationship with RR and the known immunosuppressive activity support a potential predictive value of this serum marker. A multi-institutional validation set is currently ongoing to confirm the role of SAA in this setting.