The immune impact of PI3K-AKT pathway inhibition in colorectal cancer.

154 Background: Our prior work has shown that PI3K-altered colorectal cancer (CRC), with PIK3CA mutation or PTEN loss, has increased expression of key immune checkpoints (including PD-L1) resulting in immune evasion, despite increased immune engagement. Here, we investigated the impact of PI3K-AKT inhibition on the immune repertoire of CRC. Methods: Multiplex immunofluorescence was performed using two Vectra panels [1: AE1/AE3, CD3, CD8, PD-1, PD-L1, CD68; and 2: AE1/AE3, CD3, CD8, Granzyme B (GzB), CD45RO, FoxP3] on paired biopsies (baseline and cycle 1 day 15) from 6 patients with PI3K-altered metastatic CRC (mCRC) treated with AKT inhibitor, MK2206 (200 mg oral weekly), on a phase 2 clinical trial. Separately, one million CT26 CRC cells were implanted in BALB/C-e mice. After 48 hours, 10 mice/group were randomized for treatment with pan-PI3K inhibitor copanlisib (C, 10 mg/Kg IV 2x/week), anti-PD-1 (P, 200 µg IP 2x/week), copanlisib + anti-PD-1 (C+P), or control (Ct), for 21 days. Mouse tumors were stained with 6-plex immunohistochemistry (CD3, CD8, PD-L1, Ki67, GzB, AE1/AE3). Data were analyzed using related-samples Wilcoxon Signed-Rank test, Mann-Whitney U test, Kruskal-Wallis test, and Student’s t-test, as appropriate. Results: In PI3K-altered mCRC patients, AKT inhibition resulted in a trend towards increased median densities of intratumoral CD8 + T cells (0.8 vs 4.8 density/mm 2 , P = 0.14) and memory T cells (0 vs 10.3, P = 0.07), and decreased density of macrophages (12.4 vs 0, P = 0.07). No antigen experienced T cells were seen and activated CD8 + T cells were present in 1 patient only. In CT26 mice, PI3K and PD-1 co-inhibition resulted in the smallest mean tumor volumes (C+P 12% of Ct vs C 40% and P 42% of Ct, P < 0.05 for both), and the highest median % of intratumoral CD8 + Ki67 + T cells as compared to all other treatment arms (C+P 1.6% vs C 0.5%, P 0.4%, Ct 0.6%, P < 0.05 for each pairwise comparison). C+P also increased the % of total CD3 + and CD8 + cells as compared to Ct and C (P < 0.05 for all). C alone did not increase immune infiltration in this non-PI3K activated model. Conclusions: PI3K-AKT pathway inhibition has the potential to improve effector T cell infiltration in PI3K-altered CRC. PI3K inhibitor synergizes with anti-PD-1 to improve treatment efficacy and CD8 + T cell proliferation. The mechanisms behind this immune repertoire shift are yet to be elucidated, such as via cytokine modulation. Therapeutic approaches to activate the proliferating CD8 + cells would be useful, and may require PI3Kα/β specific inhibitors to allow early T cell activation through PI3Kδ/γ isoforms.