Abstract Background: Capivasertib (AZD5363), an AKT1,2,3 inhibitor, significantly improved progression-free and overall survival when added to paclitaxel in triple negative breast cancer (BC) patients (Schmid et al. ASCO 2018). We have previously reported in STAKT, robust target inhibition at 480mg BD versus placebo, including significant decreases in the primary biomarkers (PBs) - Ki67, pPRAS40 & pGSK3β - in primary BCs (Robertson et al. SABCS 2017). We now report the dose- and exposure-response relationship of capivasertib and the correlation between primary and secondary (pAKT, pS6, nuclear FOXO3a) tumor biomarkers. Design: STAKT was a two-stage, double blind, randomized, placebo controlled 'window-of-opportunity' trial in newly diagnosed ER+ BC patients. Stage 1 assessed capivasertib at a dose of 480mg BD p.o. versus placebo. Stage 2 assessed capivasertib at two lower doses 360mg and 240mg BD. Tumor biopsies were taken prior to 1st dose and after 4.5 days of dosing. Evaluable patients (who required pre-defined minimum baseline PD values for PBs) included placebo (n=11), capivasertib at 480mg (n=17), 360mg (n=5) and 240mg (n=6). Blood samples for pharmacokinetic (PK) studies were scheduled at pre-dose; 2, 4, optional 6 & 8 hrs post first dose on Day 1; ˜2-4 h post last dose on Day 5 (before biopsy). The % change from baseline for PBs were evaluated against the following exposure variables (placebo=0): i) Dose, ii) Observed Cmax Day 1 (˜2h post-dose), iii) Observed plasma concentration on Day 5, iv) Model-predicted plasma concentration Day 5 at time of biopsy, and v) Model-predicted AUC on Day 5. Spearman correlation coefficient measured the strength and direction of association between biomarkers. Results: · Significant mean reductions in % change from baseline were observed for the PBs pGSK3β (-39%; p<0.006), pPRAS40 (-50%; p<0.0001) and Ki67 (-23%; p=0.052) at 480mg versus placebo. At 360mg and 240mg, mean % changes from baseline in pGSK3β were -27% and -9%, respectively; in pPRAS40 -45% and -28%, respectively; and in Ki67 0% and +22%, respectively. · Dose-response relationships for individual % change from baseline could be described by an Emax model for all PBs. Overall, the correlation to PK exposure (observed or predicted) was similar to the correlation to dose. · Correlation coefficient analyses between biomarkers at capivasertib 480mg BD identified- i) Positive correlations for pGSK3β with Ki67 (ρ = 0.52, p-value < 0.05) & with pS6 (ρ = 0.54, p-value<0.05); ii) Negative correlations between FOXO3a and Ki67 (ρ = -0.75, p-value<0.001) pGSK3β (ρ = -0.71, p-value<0.001) & also pS6 (ρ = -0.61, p-value<0.001).Correlation coefficients for lower doses are not robust due to small sample size in these groups. Conclusions · Capivasertib caused dose- and concentration- dependent effects on biomarkers after only 4.5 days. · Significant changes in the PBs were demonstrated at 480 mg BD. Biomarker changes was observed at 360mg and 240mg BD, but statistical analysis was limited by the small sample size at lower doses. · Correlation between a number of tumor biomarkers (relative changes) were identified for capivasertib 480mg BD. Citation Format: Gee J, Coleman RE, Cheung KL, Evans A, Holcombe C, Skene A, Rea D, Ahmed S, Jahan A, Horgan K, Rauchhaus P, Littleford R, Finlay P, Cheung A, Cullberg M, de Bruin E, Foxley A, Koulai L, Pass M, Schiavon G, Rugman P, Deb R, Robertson JFR. Dose- and exposure-response relationship and biomarker correlation analysis in breast tumors from patients treated with capivasertib, an AKT inhibitor, in the STAKT randomized, placebo controlled pre-surgical study [abstract]. In: Proceedings of the 2018 San Antonio Breast Cancer Symposium; 2018 Dec 4-8; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2019;79(4 Suppl):Abstract nr P2-12-01.