Despite unprecedented successes with immune checkpoint blockade in advanced cancers, these agents provide durable clinical benefit in only a subset of patients. In addition, the clinical management of patients receiving immune checkpoint blockade remains quite challenging due to the unpredictable and kinetically heterogeneous responses which can manifest as late responses, pseudoprogression, or hyperprogression in subsets of patients. The dynamics and heterogeneity of programmed cell death protein ligand 1 (PD-L1, aka B7-H1) expression confound its use as a predictive biomarker in cancer immunotherapy, but blood-based biomarkers have the potential to predict responders and detect mechanisms of resistance to immunotherapy. We previously reported that Bim (BH3-only protein) is a downstream signaling molecule of the PD-1 pathway, and that Bim levels in circulating tumor-reactive T-cells reflected patient responses to anti-PD-1 therapy in melanoma. We also discovered the existence of soluble PD-L1 (sPD-L1) in the plasma of cancer patients and showed a correlation of sPD-L1 with cancer stages in renal cell carcinoma and with prognosis in metastatic melanoma. Here we evaluated the frequency of tumor-reactive T cells with Bim expression along with effector memory profile and sPD-L1 as biomarkers of response in a new cohort of patients with metastatic melanoma and lung cancer undergoing anti-PD-1 therapy. We recruited 100 cancer patients treated with anti-PD-1 therapy who had peripheral blood collected at baseline and at each subsequent radiographic tumor evaluation. Frequencies of Bim+ and effector memory CD8+ T cells were measured by flow cytometry in gated tumor-reactive CD11a high PD1+ CD8+ T cells. We also measured levels of sPD-L1 in plasma at baseline and serially during treatment with ELISA. Baseline T cell markers and sPD-L1 levels and their percent changes in patients who had a radiographic response were compared to those who had progressive disease (PD). We found that patients with objective response after 4 cycles of anti-PD-1 therapy had higher frequency of Bim+ CD8+ T cells at baseline compared to patients with PD (mean 43% vs. 30%, P=0.0484). The frequencies of Bim+ T cells decreased significantly after the first 3 months of treatment in responders compared with progressors (mean -16% vs. + 40% P=0.0111). The frequency of effector memory CD8+ T cells also dramatically increased in responders in 3-6 months after anti-PD-1 therapy (mean 89% vs. 9.56%, p=0.002). High baseline sPD-L1 was associated with progression on anti-PD-1 therapy (mean 2.8 ng/mL vs. 0.7 ng/mL, p=0.07) or development of immune-related toxicities and the levels increased by 12 weeks in patients progressing on therapy.

Conclusion: Measurements of T cell biomarkers (Bim and effector memory) and sPD-L1 levels provide a new noninvasive way to predict and monitor patient responses to anti-PD-1 monotherapy in melanoma and lung cancers.

Note: This abstract was not presented at the conference.

Citation Format: Roxana S. Dronca, Aaron Mansfield, Xin Liu, Susan Harrington, Elizabeth Enninga, Svetomir Markovic, Lisa Kottschade, Rob Mcwilliams, Matthew Block, Wendy Nevala, Michael Thompson, Haidong Dong. Blood-based T cell biomarkers and soluble PD-L1 predict responses and immune-related toxicity to PD-1 blockade in melanoma and lung cancer [abstract]. In: Proceedings of the AACR Special Conference on Tumor Immunology and Immunotherapy; 2017 Oct 1-4; Boston, MA. Philadelphia (PA): AACR; Cancer Immunol Res 2018;6(9 Suppl):Abstract nr A38.