Purpose:

The STAKT study examined short-term exposure (4.5 days) to the oral selective pan-AKT inhibitor capivasertib (AZD5363) to determine if this drug can reach its therapeutic target in sufficient concentration to significantly modulate key biomarkers of the AKT pathway and tumor proliferation.

Patients and Methods:

STAKT was a two-stage, double-blind, randomized, placebo-controlled, “window-of-opportunity” study in patients with newly diagnosed ER+ invasive breast cancer. Stage 1 assessed capivasertib 480 mg b.i.d. (recommended monotherapy dose) and placebo, and stage 2 assessed capivasertib 360 and 240 mg b.i.d. Primary endpoints were changes from baseline in AKT pathway markers pPRAS40, pGSK3β, and proliferation protein Ki67. Pharmacologic and pharmacodynamic properties were analyzed from blood sampling, and tolerability by adverse-event monitoring.

Results:

After 4.5 days' exposure, capivasertib 480 mg b.i.d. (n = 17) produced significant decreases from baseline versus placebo (n = 11) in pGSK3β (H-score absolute change: −55.3, P = 0.006) and pPRAS40 (−83.8, P < 0.0001), and a decrease in Ki67 (absolute change in percentage positive nuclei: −9.6%, P = 0.031). Significant changes also occurred in secondary signaling biomarker pS6 (−42.3, P = 0.004), while pAKT (and nuclear FOXO3a) also increased in accordance with capivasertib's mechanism (pAKT: 81.3, P = 0.005). At doses of 360 mg b.i.d. (n = 5) and 240 mg b.i.d. (n = 6), changes in primary and secondary biomarkers were also observed, albeit of smaller magnitude. Biomarker modulation was dose and concentration dependent, and no new safety signals were evident.

Conclusions:

Capivasertib 480 mg b.i.d. rapidly modulates key biomarkers of the AKT pathway and decreases proliferation marker Ki67, suggesting future potential as an effective therapy in AKT-dependent breast cancers.

This article is featured in Highlights of This Issue, p. 1533

Translational Relevance

Capivasertib (AZD5363) monotherapy previously demonstrated anticancer activity in phase I trials in patients with advanced solid tumors, particularly those whose tumors harbored an AKT mutation. In the short time frame that window-of-opportunity studies permit, the STAKT study reported here determined that in primary breast cancers, capivasertib at the recommended monotherapy dose [480 mg twice daily (b.i.d.)] rapidly modulated the AKT pathway (after 4.5 days of treatment), as evidenced by significant decreases from baseline versus placebo in biomarkers of the AKT pathway (including pGSK3β and pPRAS40), and reduced cell proliferation (Ki67). Biomarker modulation was also observed at lower capivasertib doses of 240 and 360 mg b.i.d., and the inhibitory effects were dose and concentration dependent. Overall, the STAKT study provides further evidence that capivasertib has the potential to be an effective oral anticancer therapy via its impact on proliferative AKT signaling in patients with estrogen receptor–positive breast cancer.

Components of the AKT pathway (also known as the PI3K/AKT/mTOR signaling pathway) play a fundamental role in tumor cell survival, proliferation, and death (Supplementary Fig. S1; ref. 1). Mutations in signaling components can cause aberrant activation of the pathway, leading to the development of numerous solid and hematologic malignancies (1–3) and resistance to endocrine therapies (4, 5). Mutations in PIK3CA, AKT1, and PTEN are prevalent in estrogen receptor–positive (ER+) breast cancer (6), indicating that this pathway is important in ER+ breast cancer, which is further augmented by the observation of a reciprocal feedback between the ER and PI3K/AKT signaling pathways.

Capivasertib (AZD5363) is a new oral selective AKT1–3 inhibitor that demonstrated promising clinical activity and a tolerable safety profile as monotherapy treatment in a phase I study in heavily pretreated (median of five prior regimens) patients with AKT1 E17K-mutant metastatic solid cancers, with the strongest signal of activity observed in ER+ breast cancers (7). In cancer model systems, the inhibitor blocks the AKT pathway, depleting phosphorylation of pathway proteins GSK3β, PRAS40, and S6 and tumor cell growth (8). In preclinical experiments, reductions of 50% to 80% in phosphorylated PRAS40 (pPRAS40) and 40% to 70% in pGSK3β during the capivasertib dosing period were sufficient to cause significant antitumor activity in several xenograft models [e.g., 100 mg/kg twice daily (b.i.d.) in the BT474c xenograft model; ref. 8]. For Ki67, information was drawn from two previous presurgical studies of endocrine therapy in ER+ breast cancer (9, 10). The change in Ki67 in one of these, an endocrine study (NCT00259090), was a >80% decrease in all treated groups with clear statistical significance (10).

Successful drug development is critically dependent on an understanding of drug pharmacodynamics, including knowledge of the drug's rapidity of action and effect on molecular treatment targets. In this regard, the STAKT trial (NCT02077569) is the first two-stage, presurgical, window-of-opportunity biomarker study to focus on assessing the effects of a range of capivasertib monotherapy doses on key markers of the AKT pathway and expression of a protein strongly associated with tumor cell proliferation and growth (Ki67) in patients with newly diagnosed ER+ breast cancer. These markers were used to characterize the degree of biological activity and, thus, treatment potential in primary breast cancer arising from the inhibition of AKT signaling across a range of capivasertib doses.

The STAKT study aimed to help define the optimal dose of capivasertib for future studies, as well as aid further understanding of its impact on the AKT pathway and biomarker identification. Stage 1 of the study assessed the pharmacodynamic (PD) and pharmacokinetic (PK) effect and tolerability of capivasertib 480 mg b.i.d. versus placebo. Stage 2 examined the same, but at lower doses (320 and 240 mg b.i.d.) of capivasertib. It was hoped that exploring a range of doses in the study would also enable predictive model-based techniques to be developed to help identify an efficacious dosing range and dose reduction strategies for potential use in future stages of clinical development in breast cancer.

Here, data are presented from STAKT stages 1 and 2 that reveal the targeted effect of capivasertib on selected tumor and cell proliferation biomarkers, the PD properties of the dose– and exposure–response relationship, and tolerability from short-term exposure in patients with breast cancer.

Key eligibility criteria

Eligible patients were females aged ≥18 years with histologic confirmation of ER+ invasive breast carcinoma (stage 1–3 or stage 4 breast cancer with primary tumor in the breast amenable to biopsies) and tumors large enough to provide tissue for the biomarker assays. Additional key eligibility criteria included: subsequent standard of care determined to include chemotherapy, with or without surgery (as such treatment was considered to carry a greater risk of genotoxicity compared with exposure to capivasertib); WHO performance status 0–1 with no deterioration over the previous 2 weeks; and hepatic, renal, cardiac, lung, and gastrointestinal functions within normal limits, except if liver metastases were present, in which case liver enzyme levels could be up to three times the upper limit of normal. To reduce the potential risk of the drug exacerbating abnormal glucose profiles (11), patients with type 1 or 2 diabetes mellitus (irrespective of management), fasting glucose ≥7 mmol/L, or glycated hemoglobin (HbA1c) ≥64 mmol/mol or ≥8% at screening were excluded from participating in this study. Patients had no known history of hypersensitivity to the active ingredient or inactive excipients of capivasertib.

Study conduct

All patients provided written informed consent. The National Research Ethics Service Committee East Midlands – Northampton Research Ethics Committee (REC) approved the protocol. The STAKT study (trial registration ID: NCT02077569; REC ID: 13/EM/0112; Clinical Trial Authorization reference: 03057/0057/001) had local National Health Service (NHS) R&D approval and was conducted in accordance with the Declaration of Helsinki, Good Clinical Practice and the Department of Health Research Governance Framework for Health and Social Care 2005. The study was also performed in accordance with the Cancer Research UK Guidelines for Scientific Conduct (12). An independent data monitoring committee (IDMC) reviewed emerging data and, in particular, the results of stage 1 to approve the continuation of the study into stage 2. Trial oversight was provided by an independent trial steering committee on behalf of the funders and sponsors. The trial management group was responsible for the running of the trial. Tayside Clinical Trials Unit was responsible for the day-to-day management of the trial and undertook the data analyses in accordance with the statistical analysis plan for the study.

Study design and treatment

STAKT was a multicenter, two-stage, double-blind, randomized, placebo-controlled, paired-biopsy, biomarker study conducted in NHS hospitals throughout the UK. The two-stage study format was designed to obtain early evidence that the drug is pharmacokinetically and pharmacodynamically active at the 480 mg b.i.d. dose, which, based on initial tolerability, PK, and preliminary efficacy data from a previous phase I study (NCT01226316; ref. 13), was the most likely dose and schedule to be used in clinical studies.

In stage 1, eligible patients were randomized (1:1) to either capivasertib 480 mg b.i.d. or placebo for 4.5 days. In stage 2, patients were randomized (1:1) to capivasertib 360 or 240 mg b.i.d. for 4.5 days (Supplementary Fig. S2). Enrollment in stage 2 was planned to commence seamlessly upon completion of stage 1, dependent on, in the opinion of the IDMC, the 480-mg b.i.d. dose producing a depletion to a prespecified margin of ≥50% over placebo for the primary biomarkers pPRAS40, pGSK3β (both key indicators of AKT signaling), and Ki67. In fact, there was a short break in patient recruitment between stages 1 and 2, supported by the study IDMC, as the study team made the decision to remove an originally planned placebo arm for stage 2 (protocol amendment approved by the REC and the Medicines and Healthcare Products Regulatory Agency in December 2015) because of unforeseen recruitment challenges. Changes in the lower doses of capivasertib were then assessed against baseline rather than placebo.

Patients received capivasertib for 4.5 days prior to their scheduled surgery or 6.5 days prior to scheduled chemotherapy (the final morning dose was taken on the day of surgery, which accounts for the final 0.5-day period). Ultrasound-guided tumor core biopsies were taken prior to the first dose and after 4.5 days of dosing (within 12 hours after the last dose). Additional “triple blinding” was performed by blinding laboratory staff to treatment arm and time of biopsy (baseline or after 4.5 days). Biopsy samples were formalin-fixed and paraffin-embedded and immunohistochemically stained using the prevalidated assays detailed in Supplementary Table S1, and tumor epithelial staining was assessed by H-scoring the primary biomarkers pGSK3B and pPRAS40 by consensus of two expert assessors. H-score was calculated as the sum (% weak [1+]) + (% moderate [2+] × 2) + (% strong [3+] × 3), of staining localized in the cytoplasm and/or nucleus. Ki67 was evaluated by percentage nuclear positivity only. Biopsy pairs were sectioned together and analyzed in the same immunostaining assay, with a positive internal control included in every assay for quality control purposes. Blood samples for PK were scheduled at predose; 2, 4, and optional 6 and 8 hours after the first dose on day 1; and after the last dose on day 5, as close to the time of the on-treatment biopsy as possible.

Study objectives

The primary objective of STAKT was to compare the AKT pathway biomarker and antiproliferative effect of 4.5 days' treatment with three dose levels of capivasertib. Primary endpoints were changes from baseline for the active phosphorylated forms of the selected primary biomarkers, pPRAS40 and pGSK3β, and Ki67. Secondary endpoints included changes in additional biological markers associated with the AKT pathway, comprising tumor epithelial staining for pAKT (cytoplasmic/nuclear/membrane), phosphorylated ribosomal protein S6 (cytoplasmic), and FOXO3a expression (nuclear and/or cytoplasmic), as well as an assessment of tolerability through monitoring of adverse events (AE) with Common Terminology Criteria for Adverse Events (CTCAE) version 4 (graded as mild, moderate, or severe). Additionally, relationships between primary biomarkers and capivasertib exposure (dose and plasma concentration) were explored post hoc to help identify a dosing range for potential use in later clinical development.

Statistical design

Up to 60 patients were planned to be recruited in each of stages 1 and 2. Twelve patients per arm with evaluable biomarker data would give 98% power to detect a difference of 30% (change from baseline) for pPRAS40 and pGSK3β levels between the treatment arms, at the 5% significance level. Assumptions for Ki67 in this study were for 80% power and a two-sided significance level of 5%. With 12 evaluable patients, the study would be powered to show a 50% reduction in Ki67.

Statistical analysis methods

The biomarker population consisted of all patients with evaluable biomarker data at baseline and evaluable pre- and on-treatment biopsies. Evaluable patients were defined as having both pre- and on-treatment biopsy samples with a minimum of 100 tumor epithelial cells for H-score assessment for both pGSK3β and pPRAS40, as well as a minimum of 500 tumor epithelial cells to count for percentage positivity for Ki67. In addition, the tumor had to express all three primary endpoint biomarkers in the pretreatment biopsy [defined as measureable percentage positivity for Ki67 nuclear staining (any score greater than zero is acceptable) and a total H-score, i.e., cytoplasm + nuclear, of ≥10 for each of pPRAS40 and pGSK3b], and patients also should have received the full planned 4.5-day dose of capivasertib treatment.

An analysis-of-covariance model was fitted to the biomarker data, including terms for treatment and adjusting for baseline biomarker level. For each biomarker, changes in expression between the matched baseline and on-treatment biopsies were evaluated and compared with changes seen in the placebo group (comparison with placebo in stage 1 only). Both absolute and percentage changes from baseline were recorded. Stage 2 comparisons were against baseline biomarker level. The strength and direction of the association between biomarkers were assessed by the Spearman rank correlation coefficient.

Dose– and exposure–response relationships were compared for three exposure variables: dose level given (i.e., 240 mg b.i.d., as opposed to the total daily dose of 480 mg per day); observed maximum capivasertib plasma concentration (Cmax) on day 1; and capivasertib plasma concentration on day 5. Linear and Emax models were compared for each exposure variable, and the Akaike information criterion was used as a goodness-of-fit diagnostic (14).

The safety population consisted of all patients receiving ≥1 dose of study drug and was assessed by the incidence and severity of treatment-emergent adverse events (TEAEs) reported within 7 days following the first dosing of capivasertib or up to the time of start of chemotherapy if earlier.

Patient disposition

In total, 36 patients from 12 participating hospitals were randomized in stage 1 (19 patients to capivasertib 480 mg b.i.d., 17 patients to placebo), and 12 patients in stage 2 (7 patients to capivasertib 240 mg b.i.d., 5 patients to capivasertib 360 mg b.i.d.). All 48 randomized patients (36 patients from stage 1, 12 patients from stage 2) received at least one dose of study treatment and were evaluable for safety analysis. Eight patients were deemed not evaluable for biomarker analysis in stage 1. All patients were deemed evaluable in stage 2, giving a total evaluable biomarker population of 40 patients.

Of the seven patients randomized to capivasertib 240 mg b.i.d. in stage 2, one was subsequently found not to meet the prespecified eligibility criterion of a total H-score of ≥10 for pGSK3β (total score was 6) and pPRAS40 (total score was 5) staining in the pretreatment sample. This was identified after the stage 2 data lock; hence, this patient was excluded from the biomarker analysis. Therefore, a final total of 39 patients were evaluable in the biomarker population (28 patients from stage 1, 11 patients from stage 2) according to the predefined evaluability criteria. No notable differences between the baseline characteristics in the patient populations enrolled across stages 1 and 2 were discerned (Table 1).

Table 1.

Characteristics of patients at study entry (biomarker analysis set).

Stage 1Stage 2
CapivasertibCapivasertibCapivasertib
Placebo480 mg b.i.d.360 mg b.i.d.240 mg b.i.d.
Clinical characteristics(n = 11)(n = 17)a(n = 5)(n = 6)b
Mean age, years (SD) 48.5 (7.2) 51.3 (9.6) 52.6 (12.9) 58.5 (11.1) 
Mean BMI, kg/m2 (SD) 27.3 (6.6) 28.8 (4.9) 31.5 (4.1) 27.4 (4.5) 
Mean weight, kg (SD) 75.6 (17.9) 78 (12.9) 86.2 (11.4) 72.7 (9.4) 
Mean height, m (SD) 1.7 (0.07) 1.7 (0.05) 1.7 (0.03) 1.6 (0.1) 
Ethnicity, n (%) 
 White (Caucasian) 10 (90.9) 17 (100) 5 (100) 6 (100) 
 Black (including Afro Caribbean) 1 (9.1) 
Stage 1Stage 2
CapivasertibCapivasertibCapivasertib
Placebo480 mg b.i.d.360 mg b.i.d.240 mg b.i.d.
Clinical characteristics(n = 11)(n = 17)a(n = 5)(n = 6)b
Mean age, years (SD) 48.5 (7.2) 51.3 (9.6) 52.6 (12.9) 58.5 (11.1) 
Mean BMI, kg/m2 (SD) 27.3 (6.6) 28.8 (4.9) 31.5 (4.1) 27.4 (4.5) 
Mean weight, kg (SD) 75.6 (17.9) 78 (12.9) 86.2 (11.4) 72.7 (9.4) 
Mean height, m (SD) 1.7 (0.07) 1.7 (0.05) 1.7 (0.03) 1.6 (0.1) 
Ethnicity, n (%) 
 White (Caucasian) 10 (90.9) 17 (100) 5 (100) 6 (100) 
 Black (including Afro Caribbean) 1 (9.1) 

Abbreviations: BMI, body mass index; SD, standard deviation.

aNineteen patients were randomized to capivasertib in stage 1, of whom 17 were evaluable for biological endpoints (two patients did not take all study medication and were therefore considered unevaluable for biological endpoints but were included in the safety analyses).

bSeven patients were randomized to capivasertib in stage 2 (240-mg b.i.d. group) and included in the primary analysis. One patient was subsequently found not to meet the prespecified eligibility criterion of a total H-score of ≥10 for pPRAS40 (total score was 5) and pGSK3β (total score was 6).

Changes in biomarkers

Significant reductions in the absolute change measurements in H-score for the primary biomarkers were observed for the 480 mg b.i.d. dose versus placebo: −55.3 (P = 0.006) for total pGSK3β and −83.8 (P < 0.0001) for total pPRAS40 (Table 2). The absolute percentage reduction in nuclei staining positive for Ki67 was also significant, at −9.6 (P = 0.031). The percentage change from baseline at the 480 mg b.i.d. dose versus placebo was also significant for total pGSK3β (−39%; P = 0.006) and total pPRAS40 (−50%; P < 0.0001). Reductions from baseline (absolute and percentage) were also observed for pGSK3β and pPRAS40 at the 360-mg b.i.d. and 240-mg b.i.d. doses. The secondary biomarkers pAKT and pS6 also showed significant change from baseline in H-score for the 480-mg b.i.d. dose versus placebo, but FOXO3a did not (Table 2).

Table 2.

Analysis of change from baseline for the primary and secondary biomarkers.

(A) 480 mg b.i.d. versus placebo (stage 1)
Capivasertib 480 mg b.i.d. (n = 17)Mean difference in change between groups (mixed model)a
Type of change vs. baselineAdjusted (95% CI)bP value vs. placebo arm
Primary 
 Ki67 (% cells positive) 
  Nuclear Absolute −9.6 (−18.3, −0.98) 0.031 
 −23.4 (−59.1, 0.29) 0.052 
 pGSK3β (H-score) 
  Total Absolute −55.3 (−93.4, −17.7) 0.006 
 −39.0 (−65.5, −12.5) 0.006 
  Cytoplasmic Absolute −53.6 (−90.5, −16.8) 0.006 
 −39.2 (−65.8, −12.7) 0.006 
  Nuclear Absolute −2.8 (−5.9, 0.19) 0.065 
 % −36.5 (−74.4, 1.35) 0.058 
 pPRAS40 (H-score) 
  Total Absolute −83.8 (−111.6, −56.0) <0.0001 
 −50.2 (−68.7, −31.7) <0.0001 
  Cytoplasmic Absolute −90.1 (−120.9, −9.3) <0.0001 
 −55.8 (−75.4, −36.2) <0.0001 
  Nuclear Absolute 6.9 (−10.7, 24.6) 0.42 
 8.9 (−258.2, 276.2) 0.94 
Secondary 
 pAKT (H-score) 
  Total Absolute 81.3 (27.1, 135.5) 0.005 
 116.9 (29.3, 204.6) 0.011 
  Cytoplasmic Absolute 24.8 (−7.51, 57.1) 0.127 
 76 (−3.3, 155.2) 0.0595 
  Nuclear Absolute 53 (13.2, 92.8) 0.011 
 516.5 (−29.7, 1,062.6) 0.063 
 pS6 (H-score) 
  Cytoplasmic Absolute −42.3 (69.7, −14.8) 0.004 
 −30 (−48.8, −11.1) 0.003 
 FOXO3a (H-score) 
  Total Absolute 29.6 (−19.9, 79.2) 0.229 
 19.4 (−21.4, 60.2) 0.338 
  Cytoplasmic Absolute −46.2 (−77.0, −15.4) <0.0001 
 −42.4 (−97.2, 12.5) 0.124 
  Nuclear Absolute 79.3 (35.8, 122.8) 0.0009 
 843.5 (155.5, 1,531.5) 0.018 
(B) Mean change from baseline for all doses of capivasertib (stages 1 and 2) 
  Mean change from baseline (95% CI) 
  Capivasertib 480 mg b.i.d. (n = 17) Capivasertib 360 mg b.i.d. (n = 5) Capivasertib 240 mg b.i.d. (n = 7) 
Primary 
 Ki67 (% cells positive) 
  Nuclear Absolute −12.4 (−18.9, −5.92) 0.92 (−7.54, 9.38) 0.2 (−9.0, 9.3) 
 −38.3 (−57.6, −19.1) −0.33 (−30.3, 29.6) 21.6 (−15.7, 58.8) 
 pGSK3β (H-score) 
  Total Absolute −66.9 (−96.3, −37.6) −18.6 (−47.8, 10.6) −39.2 (−98.9, 20.6) 
 −41.6 (−57.5, −25.8) −27.1 (−74.2, 20.0) −9.3 (−94.9, 76.4) 
  Cytoplasmic Absolute −62.4 (−89.4, −35.3) −18.6 (−47.8, 10.6) −21.6 (−86.1, 43.0) 
 −40.7 (−56.5, −25.0) −27.1 (−74.2, 19.9) 192.0 (−305.4, 689.5) 
  Nuclear Absolute −4.6 (−10.9, −1.73) 0.14 (−0.21, 0.49) 
 % −37.1 (−61.3, −12.8) 
 pPRAS40 (H-score) 
  Total Absolute −71.3 (−93.9, −48.7) −69.6 (−111.2, −28.0) −43.8 (−94.0, 6.3) 
 −43.7 (−56.7, −30.7) −44.9 (−81.7, −8.12) −28.3 (−65.6, 8.9) 
  Cytoplasmic Absolute −76.5 (−103.3, −49.6) −71.0 (−114.6, 27.5) −35.3 (−94.1, 23.5) 
 −51.2 (−65.4, −37.0) −46.9 (−86.6, −7.29) 350.7 (−587.6, 1,288.9) 
  Nuclear Absolute 5.2 (−8.2, 18.5) 1.4 (−4.0, 6.8) 11.3 (−6.02, 28.6) 
 105.6 (−42.2, 253.4) 200 (−546.4, 946.4) 340 (−405.2, 185.2) 
Secondary 
 pAKT (H-score) 
  Total Absolute 70.4 (23.2, 117.5) 79 (−5.7, 163.7) 10 (39.8, 172.2) 
 107.0 (25.2, 188.9) 71.3 (−19.9, 162.4) 133.5 (21.8, 245.2) 
  Cytoplasmic Absolute 24.8 (−0.7, 50.4) 26 (−18.0, 70.0) 54.3 (18.5, 90.1) 
 69.7 (0.43, 139.0) 43.6 (−43.7, 130.9) 119 (−10.2, 227.8) 
  Nuclear Absolute 45.5 (12.0, 78.8) 28.4 (1.9, 54.9) 55.3 (6.8, 103.8) 
 508.2 (44.9, 971.4) 918.8 (−747.4, 2,584.9) 1,173.9 (−81.7, 2,429.5) 
 pS6 (H-score) 
  Cytoplasmic Absolute −63.1 (−81.5, −44.6) −51.8 (−92.0, −11.6) −66.4 (−107.3, −25.5) 
 −38.8 (−48.8, 28.9) −40.5 (−83.3, 2.3) −21.2 (−92.3, 49.8) 
 FOXO3a (H-score) 
  Total Absolute 4.18 (−30.0, 38.3) 6.0 (−24.2, 36.2) −21.4 (−74.1, 31.2) 
 12.1 (−13.5, 37.6) −12.8 (−73.6, 47.9) −23.1 (−118.9, 72.6) 
  Cytoplasmic Absolute −65.6 (−96.9, −34.3) 34.4 (−20.5, 89.3) 35.0 (−7.5, 77.5) 
 −40.2 (−76.4, −3.91) 144.9 (−218.0, 507.8) 145.3 (−54.7, 345.2) 
  Nuclear Absolute 69.8 (27.9, 111.6) 34.4 (−20.5, 89.3) 35.0 (−7.47, 77.5) 
 924.8 (307.9, 1,541.7) 144.9 (−218.0, 507.8) 145.3 (−54.7, 345.2) 
(A) 480 mg b.i.d. versus placebo (stage 1)
Capivasertib 480 mg b.i.d. (n = 17)Mean difference in change between groups (mixed model)a
Type of change vs. baselineAdjusted (95% CI)bP value vs. placebo arm
Primary 
 Ki67 (% cells positive) 
  Nuclear Absolute −9.6 (−18.3, −0.98) 0.031 
 −23.4 (−59.1, 0.29) 0.052 
 pGSK3β (H-score) 
  Total Absolute −55.3 (−93.4, −17.7) 0.006 
 −39.0 (−65.5, −12.5) 0.006 
  Cytoplasmic Absolute −53.6 (−90.5, −16.8) 0.006 
 −39.2 (−65.8, −12.7) 0.006 
  Nuclear Absolute −2.8 (−5.9, 0.19) 0.065 
 % −36.5 (−74.4, 1.35) 0.058 
 pPRAS40 (H-score) 
  Total Absolute −83.8 (−111.6, −56.0) <0.0001 
 −50.2 (−68.7, −31.7) <0.0001 
  Cytoplasmic Absolute −90.1 (−120.9, −9.3) <0.0001 
 −55.8 (−75.4, −36.2) <0.0001 
  Nuclear Absolute 6.9 (−10.7, 24.6) 0.42 
 8.9 (−258.2, 276.2) 0.94 
Secondary 
 pAKT (H-score) 
  Total Absolute 81.3 (27.1, 135.5) 0.005 
 116.9 (29.3, 204.6) 0.011 
  Cytoplasmic Absolute 24.8 (−7.51, 57.1) 0.127 
 76 (−3.3, 155.2) 0.0595 
  Nuclear Absolute 53 (13.2, 92.8) 0.011 
 516.5 (−29.7, 1,062.6) 0.063 
 pS6 (H-score) 
  Cytoplasmic Absolute −42.3 (69.7, −14.8) 0.004 
 −30 (−48.8, −11.1) 0.003 
 FOXO3a (H-score) 
  Total Absolute 29.6 (−19.9, 79.2) 0.229 
 19.4 (−21.4, 60.2) 0.338 
  Cytoplasmic Absolute −46.2 (−77.0, −15.4) <0.0001 
 −42.4 (−97.2, 12.5) 0.124 
  Nuclear Absolute 79.3 (35.8, 122.8) 0.0009 
 843.5 (155.5, 1,531.5) 0.018 
(B) Mean change from baseline for all doses of capivasertib (stages 1 and 2) 
  Mean change from baseline (95% CI) 
  Capivasertib 480 mg b.i.d. (n = 17) Capivasertib 360 mg b.i.d. (n = 5) Capivasertib 240 mg b.i.d. (n = 7) 
Primary 
 Ki67 (% cells positive) 
  Nuclear Absolute −12.4 (−18.9, −5.92) 0.92 (−7.54, 9.38) 0.2 (−9.0, 9.3) 
 −38.3 (−57.6, −19.1) −0.33 (−30.3, 29.6) 21.6 (−15.7, 58.8) 
 pGSK3β (H-score) 
  Total Absolute −66.9 (−96.3, −37.6) −18.6 (−47.8, 10.6) −39.2 (−98.9, 20.6) 
 −41.6 (−57.5, −25.8) −27.1 (−74.2, 20.0) −9.3 (−94.9, 76.4) 
  Cytoplasmic Absolute −62.4 (−89.4, −35.3) −18.6 (−47.8, 10.6) −21.6 (−86.1, 43.0) 
 −40.7 (−56.5, −25.0) −27.1 (−74.2, 19.9) 192.0 (−305.4, 689.5) 
  Nuclear Absolute −4.6 (−10.9, −1.73) 0.14 (−0.21, 0.49) 
 % −37.1 (−61.3, −12.8) 
 pPRAS40 (H-score) 
  Total Absolute −71.3 (−93.9, −48.7) −69.6 (−111.2, −28.0) −43.8 (−94.0, 6.3) 
 −43.7 (−56.7, −30.7) −44.9 (−81.7, −8.12) −28.3 (−65.6, 8.9) 
  Cytoplasmic Absolute −76.5 (−103.3, −49.6) −71.0 (−114.6, 27.5) −35.3 (−94.1, 23.5) 
 −51.2 (−65.4, −37.0) −46.9 (−86.6, −7.29) 350.7 (−587.6, 1,288.9) 
  Nuclear Absolute 5.2 (−8.2, 18.5) 1.4 (−4.0, 6.8) 11.3 (−6.02, 28.6) 
 105.6 (−42.2, 253.4) 200 (−546.4, 946.4) 340 (−405.2, 185.2) 
Secondary 
 pAKT (H-score) 
  Total Absolute 70.4 (23.2, 117.5) 79 (−5.7, 163.7) 10 (39.8, 172.2) 
 107.0 (25.2, 188.9) 71.3 (−19.9, 162.4) 133.5 (21.8, 245.2) 
  Cytoplasmic Absolute 24.8 (−0.7, 50.4) 26 (−18.0, 70.0) 54.3 (18.5, 90.1) 
 69.7 (0.43, 139.0) 43.6 (−43.7, 130.9) 119 (−10.2, 227.8) 
  Nuclear Absolute 45.5 (12.0, 78.8) 28.4 (1.9, 54.9) 55.3 (6.8, 103.8) 
 508.2 (44.9, 971.4) 918.8 (−747.4, 2,584.9) 1,173.9 (−81.7, 2,429.5) 
 pS6 (H-score) 
  Cytoplasmic Absolute −63.1 (−81.5, −44.6) −51.8 (−92.0, −11.6) −66.4 (−107.3, −25.5) 
 −38.8 (−48.8, 28.9) −40.5 (−83.3, 2.3) −21.2 (−92.3, 49.8) 
 FOXO3a (H-score) 
  Total Absolute 4.18 (−30.0, 38.3) 6.0 (−24.2, 36.2) −21.4 (−74.1, 31.2) 
 12.1 (−13.5, 37.6) −12.8 (−73.6, 47.9) −23.1 (−118.9, 72.6) 
  Cytoplasmic Absolute −65.6 (−96.9, −34.3) 34.4 (−20.5, 89.3) 35.0 (−7.5, 77.5) 
 −40.2 (−76.4, −3.91) 144.9 (−218.0, 507.8) 145.3 (−54.7, 345.2) 
  Nuclear Absolute 69.8 (27.9, 111.6) 34.4 (−20.5, 89.3) 35.0 (−7.47, 77.5) 
 924.8 (307.9, 1,541.7) 144.9 (−218.0, 507.8) 145.3 (−54.7, 345.2) 

Abbreviation: CI, confidence interval.

aMixed model: Difference between placebo (n = 11) and capivasertib (n = 17) corrected for baseline values.

bAdjusted for baseline value.

However, when the contribution to total H-score from cytoplasmic staining was separated from nuclear staining for the biomarkers whose total H-scores were derived from nuclear plus cytoplasmic staining (pGSK3β, pPRAS40, pAKT, and FOXO3a), it was noted that FOXO3a showed a significant decrease in cytoplasmic H-score (absolute change −46.2, P < 0.0001 vs. placebo), while there was a corresponding substantial increase in nuclear H-score for this marker (absolute change 79.3, P = 0.0009 vs. placebo) at the 480-mg dose level. Such a difference in staining profile for different cellular compartments was not observed for the other biomarkers (Table 2).

An immunohistochemistry assay for cleaved caspase-3, as a measure of apoptosis, was additionally performed, and the percentage positivity was gauged in a total population of 500 and, where possible, 3,000 tumor cells. Absolute and percentage changes were assessed by analysis of variance (ANOVA) adjusted for treatment and baseline values. Only one of eight ANOVA analyses (considering percentage apoptosis in 3,000 cells adjusted for baseline) gave a P value of <0.05 (data not shown).

A heat map showing the changes in the primary and secondary biomarkers for each patient in the three capivasertib groups, as well as the placebo group, is shown in Fig. 1A. Compared with placebo, more substantial and frequent decreases were seen with the 480-mg dose for Ki67 and the AKT pathway markers pGSK3β, pPRAS40, and pS6, with 13 of 17 patients showing some change in all three primary biomarkers. Increases in turn were more common with such treatment for pAKT (12/17 patients) and nuclear FOXO3a expression (13/17 patients). At doses of 360 and 240 mg b.i.d., declines in the biomarkers pGSK3β, pPRAS40, and pS6 were also observed, albeit of smaller magnitude for the primary markers than seen with the 480-mg dose (particularly for Ki67; Table 2). Increases were again noted for pAKT and nuclear FOXO3a. Representative immunohistochemistry staining images for both primary and secondary biomarkers at baseline and on treatment can also be seen in Fig. 1B, showing declines in Ki67, pGSK3B, pPRAS40, and pS6, and increases in pAKT and nuclear FOXO3a, with the 480-mg dose that are not apparent with placebo.

Figure 1.

A, Heat map for stages 1 and 2 showing changes in primary and secondary biomarkers by individual patient. B, Examples of immunohistochemistry staining images. A, Green denotes a change in the direction expected with AKT pathway inhibition using capivasertib. Notably, capivasertib induces increases in pAKT because of its ATP-competitive mechanism of action. Patients are ranked in this figure by percentage change in Ki67 (nuclear percentage positivity); bars and numbers represent percentage change from baseline (except for pAKT, where bars are capped at −100% and +100% change for presentation purposes). B, Original magnification of staining images was 20×.

Figure 1.

A, Heat map for stages 1 and 2 showing changes in primary and secondary biomarkers by individual patient. B, Examples of immunohistochemistry staining images. A, Green denotes a change in the direction expected with AKT pathway inhibition using capivasertib. Notably, capivasertib induces increases in pAKT because of its ATP-competitive mechanism of action. Patients are ranked in this figure by percentage change in Ki67 (nuclear percentage positivity); bars and numbers represent percentage change from baseline (except for pAKT, where bars are capped at −100% and +100% change for presentation purposes). B, Original magnification of staining images was 20×.

Close modal

Correlation coefficient analyses

A correlation coefficient analysis (Spearman rank) of the percentage changes in biomarkers at the 480-mg b.i.d. dose (Fig. 2) indicated significant positive correlation between changes in pGSK3β (total) and Ki67 (nuclear; R = 0.52, P = 0.031), and between pGSK3β (total) and pS6 (cytoplasmic; R = 0.54, P = 0.025). Negative correlations were determined between FOXO3a (nuclear) and Ki67 (nuclear; R = 0.75, P < 0.0001), FOXO3a (nuclear) and pGSK3β (R = 0.71, P = 0.0014), and FOXO3a (nuclear) and pS6 (R = 0.61, P = 0.0092). These correlations observed at the 480-mg b.i.d. dose were not replicated in the placebo group (Fig. 2). Correlation coefficients for the lower doses were not robust, most likely because of the small patient numbers in these groups, although a reduced PD effect cannot be ruled out.

Figure 2.

Scatter plot matrices comparing the percentage changes in primary and secondary biomarkers for capivasertib 480 mg b.i.d. and placebo. Nuclear score: Ki67 and FOXO3a; total score: pGSK3β, pPRAS40, and pAKT; cytoplasmic score: pS6. Relationships between biomarkers are quantified by Spearman rank correlation coefficient (R) with P value. The red and blue solid lines represent the linear regression lines for capivasertib and placebo, respectively, and the red and blue zones represent 95% CI.

Figure 2.

Scatter plot matrices comparing the percentage changes in primary and secondary biomarkers for capivasertib 480 mg b.i.d. and placebo. Nuclear score: Ki67 and FOXO3a; total score: pGSK3β, pPRAS40, and pAKT; cytoplasmic score: pS6. Relationships between biomarkers are quantified by Spearman rank correlation coefficient (R) with P value. The red and blue solid lines represent the linear regression lines for capivasertib and placebo, respectively, and the red and blue zones represent 95% CI.

Close modal

Pharmacokinetics

The capivasertib Cmax was generally observed 2 hours after dose. There was a trend toward increasing Cmax and concentration on day 5 with increasing dose (Fig. 3). The variation in concentration on day 5 is influenced by the variability in sampling times, which is noted in the discussion.

Figure 3.

Relationship between (A) capivasertib dose and Cmax on day 1 and (B) capivasertib dose and capivasertib plasma concentration on day 5 at time of biopsy.

Figure 3.

Relationship between (A) capivasertib dose and Cmax on day 1 and (B) capivasertib dose and capivasertib plasma concentration on day 5 at time of biopsy.

Close modal

Dose– and exposure–response relationships

Biomarker modulation was dose and concentration dependent. The dose–response relationship for percentage change from baseline could be described by a nonlinear (Emax) model for all primary biomarkers (Fig. 4). Similar correlations were observed for the change in the biomarkers and PK exposure (Cmax on day 1 or concentration at the time of biopsy on day 5).

Figure 4.

Observed and model-predicted dose– and exposure–response relationships for (A) pGSK3β, (B) pPRAS40, and (C) Ki67 as percentage change from baseline. Day 5 concentrations are at the time of biopsy.

Figure 4.

Observed and model-predicted dose– and exposure–response relationships for (A) pGSK3β, (B) pPRAS40, and (C) Ki67 as percentage change from baseline. Day 5 concentrations are at the time of biopsy.

Close modal

Tolerability

In total, 78 TEAEs (i.e., within 7 days of first dosing or before scheduled chemotherapy) were reported in 27 of 31 (87%) patients receiving capivasertib (all doses), and 23 TEAEs in 10 of 17 (59%) patients receiving placebo (Table 3). Among the 27 patients receiving capivasertib (all doses) and reported to have a TEAE (all grades—mild, moderate, or severe), diarrhea was observed most frequently, with 13 (17%) events out of a total of 78, followed by nausea with eight events out of 78 (10%).

Table 3.

TEAEs (all grades) observed at a frequency of ≥2 events in any treatment arm within 7 days of first dosing (safety analysis set).

Stage 1Stage 2
PlaceboCapivasertib 480 mg b.i.d.Capivasertib 360 mg b.i.d.Capivasertib 240 mg b.i.d.All capivasertib doses
(n = 17)(n = 19)(n = 5)(n = 7)(n = 31)
Number of patients experiencing a TEAE,an (%) 10 (59) 15 (79) 5 (100) 7 (100) 27 (87) 
Total number of TEAEs observed,an 23 58 11 78 
Number of TEAEs observed by preferred term, n (%)b 
 Diarrhea 3 (13.0) 11 (19.0) 1 (9.0) 1 (11.1) 13 (16.7) 
 Nausea 3 (13.0) 6 (10.3) 1 (9.0) 1 (11.1) 8 (10.3) 
 Headache 4 (6.9) 4 (5.1) 
 Dizziness 3 (5.2) 1 (9.0) 4 (5.1) 
 Proteinuria 1 (4.3) 3 (5.2) 3 (3.8) 
 Fatigue 1 (4.3) 2 (3.4) 1 (11.1) 3 (3.8) 
 Vomiting 1 (4.3) 2 (3.4) 2 (2.6) 
 Pain in extremity 2 (3.4) 2 (2.6) 
 Seroma 2 (18.2) 2 (2.6) 
 Constipation 2 (8.7) 1 (1.7) 1 (1.3) 
 Stomatitis 2 (8.7) 
Stage 1Stage 2
PlaceboCapivasertib 480 mg b.i.d.Capivasertib 360 mg b.i.d.Capivasertib 240 mg b.i.d.All capivasertib doses
(n = 17)(n = 19)(n = 5)(n = 7)(n = 31)
Number of patients experiencing a TEAE,an (%) 10 (59) 15 (79) 5 (100) 7 (100) 27 (87) 
Total number of TEAEs observed,an 23 58 11 78 
Number of TEAEs observed by preferred term, n (%)b 
 Diarrhea 3 (13.0) 11 (19.0) 1 (9.0) 1 (11.1) 13 (16.7) 
 Nausea 3 (13.0) 6 (10.3) 1 (9.0) 1 (11.1) 8 (10.3) 
 Headache 4 (6.9) 4 (5.1) 
 Dizziness 3 (5.2) 1 (9.0) 4 (5.1) 
 Proteinuria 1 (4.3) 3 (5.2) 3 (3.8) 
 Fatigue 1 (4.3) 2 (3.4) 1 (11.1) 3 (3.8) 
 Vomiting 1 (4.3) 2 (3.4) 2 (2.6) 
 Pain in extremity 2 (3.4) 2 (2.6) 
 Seroma 2 (18.2) 2 (2.6) 
 Constipation 2 (8.7) 1 (1.7) 1 (1.3) 
 Stomatitis 2 (8.7) 

aTreatment emergent is defined as occurring within 7 days of first dosing or before scheduled chemotherapy.

bPercentage per dose = (number of TEAEs by preferred term/total number of TEAEs observed) × 100.

One patient (1.7%) receiving capivasertib 480 mg was reported to have a decreased ejection fraction. No hyperglycemic TEAEs were reported in the capivasertib-treated patients. Three patients had an AE assessed by the investigator as “possibly” causally related to the study drug (one in each patient): 480 mg b.i.d., atypical migraine; 360 mg b.i.d., neutropenia; and 240 mg b.i.d., neutropenia. All three patients subsequently recovered from the AE.

Three treatment-emergent serious adverse events (SAE) were recorded at the 480-mg b.i.d. dose (delayed recovery from anesthesia, nausea, and vomiting) and two at the 340-mg b.i.d. dose (dizziness and nausea). No SAEs were recorded in the 240-mg b.i.d. or placebo groups. No patients discontinued capivasertib (any dose) because of an AE.

Window-of-opportunity studies performed in the time between diagnosis and standard-of-care treatment, in which patients are briefly exposed to an investigative compound, can facilitate early understanding of the PD and dose/exposure–response effects of a drug and may confirm identification of predictive response markers for retrospective analyses of ongoing or completed clinical trials (15). Equally, such studies provide a unique opportunity to monitor drug mechanism in vivo, including providing evidence of hit on the target pathway and its important cellular endpoints.

The principal research questions addressed in the STAKT study reported here were whether the new selective pan-AKT inhibitor capivasertib, particularly at the recommended monotherapy dose of 480 mg b.i.d., could reach its therapeutic target AKT pathway in primary breast cancer patients, and if so, whether this was to the extent necessary to produce anticancer efficacy; these were assessed by the effects on markers indicative of functional AKT signaling and tumor proliferation, respectively. Our results show that capivasertib 480-mg b.i.d., after 4.5 days of treatment, does indeed reach its therapeutic AKT pathway target in such patients, as evidenced by statistically significant decreases from baseline compared with placebo in activity (phosphorylation) of the pathway biomarkers GSK3β and PRAS40, as well as pS6, accompanied by significant decreases in tumor Ki67 (Table 2). Pathway biomarker modulation was also observed at the lower doses of 240 and 360 mg b.i.d., albeit generally to a lesser degree, indicating that the inhibitory effects of capivasertib were dose and concentration dependent.

A heat map of the degree of biomarker modulation (Fig. 1) further revealed that most patients receiving capivasertib 480 mg b.i.d. showed a reduction from baseline in the primary biomarkers pGSK3β and pPRAS40, as well as in Ki67, and a (high) degree of correlation in the modulation of biomarkers within the same patient. By comparison, in the placebo group, changes were smaller in magnitude and more randomly distributed between increases and decreases, indicating no direct, substantive effects in the placebo group, contrasting with capivasertib, which was having a targeted modulating effect across multiple biomarkers in the AKT pathway. At the 480-mg b.i.d. dose, the heat map showed that marked primary biomarker changes were often seen in the same individuals, as opposed to the placebo group, in which such changes tended to be spread across several patients. Correlations between changes in pGSK3β and pS6, and between pGSK3β and Ki67, were identified for the 480-mg b.i.d. dose that were further suggestive of pathway element interplay, its impact on proliferation, and its targeting by capivasertib. Also, in all the capivasertib-treated groups, changes in multiple primary and/or secondary endpoint signaling pathway markers were seen in the majority of individuals. In total, these observations are indicative of capivasertib reaching its intended target as a controlled “hit” down the whole AKT pathway in breast cancers. At doses of 360 and 240 mg b.i.d., similar observations can be made to those at 480 mg, albeit of smaller magnitude for the primary markers (particularly for Ki67, whereby decreases were also less frequent), again suggesting that after 4.5 days' dosing, capivasertib was reaching its intended signaling pathway target in a consistent manner.

The secondary biomarkers analyzed (pAKT, pS6, and FOXO3a) also showed significant changes in this study that are consistent with inhibition of the AKT pathway. The observed induction of pAKT is consistent with previous preclinical and clinical data; capivasertib increases phosphorylation of AKT itself, and it has been reported that this is due to the protein being held in a hyperphosphorylated but catalytically inactive form as a consequence of compound binding (8, 13, 16). Results were obtained from paired biopsies collected during two previous phase I studies (NCT01226316 and NCT01353781), albeit in metastatic solid malignancies (including those with PIK3CA mutation). These phase I studies also showed downregulation of PD biomarkers (e.g., pPRAS40 and pGSK3b) following capivasertib treatment, increased phosphorylation levels of AKT (consistent with ATP-competitive mechanism of action), and inhibition of FOXO nuclear translocation (13, 17).

In contrast to these phase I studies, the STAKT study was more controlled, i.e., paired samples were collected at similar time points and from patients with similar disease, and a placebo control group was included. The STAKT study results we report are consistent with the data from the phase I studies and demonstrate the PD effects more robustly and for the first time in primary breast cancers. For FOXO3a, the large decrease in cytoplasmic staining and corresponding large increase in nuclear staining after 4.5 days' exposure seems likely to be a result of redistribution of FOXO3a within the cancer cell as a consequence of upstream changes in AKT signaling pathway activity. This is consistent with the data in paired biopsies from solid tumors in the capivasertib phase I study (13) and validates the preclinical observation of induction of nuclear accumulation of FOXO3a in BT474c breast cancer cells [human epidermal growth factor receptor 2 (HER2)-amplified PIK3CA-mutant] exposed to capivasertib (18). It has been reported that AKT pathway–driven phosphorylation of FOXO3a is arrested by AKT inhibition (18), permitting translocation of FOXO3a to the nucleus; in keeping with this, inverse associations were seen in our study between changes in nuclear FOXO3a and the pathway signaling markers pGSK3β and pS6. Once inside the nucleus, as a transcription factor, FOXO3a is able to initiate expression of the tumor-suppressive genes BIM, FasL, and p27, which collectively can induce cell-cycle arrest and so may contribute to proliferation inhibition with the drug and/or apoptosis (18). The effect of capivasertib on such tumor proliferation, within the treatment window of 4.5 days, is clearly seen in the Ki67 results here, and changes in nuclear FOXO3a with the 480-mg dose also showed an indirect correlation with changes in Ki67. The impact of the drug on apoptosis, whereby cell survival can be a further consequence of AKT signal transduction, has not been clearly shown within the short time frame of the STAKT study by the cleaved caspase-3 staining results. Future translational studies should look to address this question.

Peak plasma concentration for capivasertib was generally observed 2 hours after dose, and there was a trend toward increasing Cmax and concentration on day 5 with increasing dose. Variability in time of sampling after dose caused variation in day 5 concentrations between patients that contributed to the large overlap noted between dose groups (Fig. 3). The PK data reported here are consistent with the emerging PK profile for capivasertib that indicates a median time to Cmax of 2 hours (range, 0.5–6 hours), with a terminal half-life of approximately 10 hours (range, 7–15 hours) after the first dose (11, 13, 17). Weak dose–response relationships (percentage change from baseline) were identified for the two primary biomarkers (pGSK3β and pPRAS40) and Ki67. Generally, the correlation between the primary biomarkers and PK exposure was similar to the correlation with dose.

The safety profile of capivasertib was previously reported, and the most frequently reported side effects are gastrointestinal events (diarrhea, vomiting, and nausea), fatigue, hyperglycemia, and maculopapular rash (7, 11, 13). The safety assessment of capivasertib in the STAKT study reported was, by nature of a window-of-opportunity study, short in comparison with prior and ongoing studies of the compound. The TEAEs and SAEs observed in STAKT were as previously reported for capivasertib, and no new safety signals became evident.

Since the STAKT trial was completed, the phase II study (PAKT: NCT02423603) has reported improved progression-free survival (PFS) and overall survival (OS) from the combination of capivasertib and paclitaxel as first-line therapy in patients with triple-negative breast cancer; an enhanced effect on PFS was observed in patients with mutations in the PIK3CA, AKT1, or PTEN genes (19). More recently, in a phase II, randomized, double-blind, placebo-controlled study (FAKTION: NCT01992952; ref. 20), capivasertib in combination with fulvestrant significantly improved PFS in patients with advanced ER+ HER2 breast cancer previously treated with aromatase inhibitors, with an observed OS improvement of approximately 6 months, although this was not statistically significant (37% OS data maturity). The successful development of a targeted AKT inhibitor such as capivasertib, either as monotherapy or in combination, may provide a new treatment option in breast cancer that helps circumvent endocrine therapy resistance from aberrant activation of the pathway (21). In a separate phase II trial (BEECH: NCT01625286), adding capivasertib to weekly paclitaxel did not prolong PFS in a population of patients with advanced ER+/HER2 breast cancer or in a subpopulation whose tumors harbored a PIK3CA mutation (22). Notably, no concomitant endocrine therapy was permitted during the BEECH study. The STAKT translational study, with the PK/PD data showing that capivasertib is able to reach its therapeutic target in sufficient concentration to significantly modulate key biomarkers of the AKT pathway and tumor proliferation, provides biological support for the improved clinical outcomes seen in a likewise hormone receptor–positive tumor population in the FAKTION trial.

Study limitations

The small sample size in stage 2, from unforeseen recruitment challenges, was mitigated by a scientific rationale endorsed by ethical approval to withdraw the placebo arm in stage 2: it was agreed that assessment of dose-related effects could be achieved by comparing on-treatment PD results with baseline in the three dose groups, and that the placebo arm in stage 2 could be dropped. Biomarker changes after 4.5 days were therefore assessed against baseline biomarker levels in stage 2. This was different from stage 1, in which changes in biomarkers were compared with changes in the placebo arm. The reduced patient numbers in stage 2 also meant that robust statistical analyses of biomarker expression between the three doses were not possible. A second limitation was that the mutational status of these tumors was not assessed; therefore, it is not clear at this time whether the extent of the PD effect of capivasertib monotherapy is associated with alterations in key genes of the pathway (i.e., PIK3CA/AKT1/PTEN). This warrants further investigation.

Conclusions

The STAKT study has shown that capivasertib causes dose- and concentration-dependent effects on primary endpoint markers of its target AKT pathway (pGSK3β and pPRAS40) and the proliferation marker, Ki67, after only 4.5 days' exposure. To our knowledge, this is the first such study with this drug in primary breast cancer and is also the shortest presurgical window-of-opportunity study in breast cancer to show a significant decrease in tumor proliferation with a targeted therapy, which also exemplifies the effectiveness of such studies for investigating novel potential treatment compounds. Changes in secondary markers in the pathway (e.g., pAKT, FOXO3a, and pS6) were also in accordance with the drug targeting this pathway. Correlations between a number of tumor biomarkers were identified for capivasertib 480 mg b.i.d. (13). These correlations were in keeping with the drug's expected mechanism of inhibitory impact on the AKT signaling pathway, as were changes in the primary and secondary endpoints. Biomarker modulation was also observed at the lower capivasertib doses of 240 and 360 mg b.i.d., although statistical analysis was limited by the small sample size at the lower doses.

The data presented confirm that capivasertib at the recommended monotherapy dose (480 mg b.i.d.) rapidly modulates the AKT pathway, and the significant resultant decrease in tumor Ki67 also raises the potential that it may be an effective anticancer therapy in AKT-dependent breast cancers. These findings, together with the positive results in the phase II randomized trial FAKTION, support the further development of capivasertib in patients with hormone receptor–positive metastatic breast cancer.

J.F.R. Robertson is a paid consultant for Carrick Therapeutics, Cullinan Oncology, and AstraZeneca, holds ownership interest (including patents) in Carrick Therapeutics, and reports receiving commercial research grants, speakers bureau honoraria, and other remuneration from AstraZeneca. R.E. Coleman is a paid consultant for Inbiomotion, Scancell, and Biocon, reports receiving speakers bureau honoraria from Amgen, Genomic Health, and Eisai, holds ownership interest (including patents) in Inbiomotion, and is an advisory board member/unpaid consultant for Astellas and Boehringer Ingelheim. K.-L. Cheung reports receiving commercial research grants from AstraZeneca. D. Rea reports receiving commercial research grants from Celgene, Roche, and Biotheranostics, speakers bureau honoraria from Novartis, Pfizer, Genomic Health, and Prosigna, and other remuneration from Pfizer and Novartis. S. Ahmed reports receiving other remuneration from AstraZeneca. S.Y.A. Cheung is an employee/paid consultant for Certara and AstraZeneca. M. Cullberg, L. Koulai, J.P.O. Lindemann, and P. Rugman are employees/paid consultants for AstraZeneca. E.C. de Bruin, M. Pass, G. Schiavon, and A. Foxley are employees/paid consultants for and hold ownership interest (including patents) in AstraZeneca. No potential conflicts of interest were disclosed by the other authors.

Conception and design: J.F.R. Robertson, R.E. Coleman, K.-L. Cheung, R. Littleford, S.Y.A. Cheung, J.P.O. Lindemann, M. Pass, A. Foxley

Development of methodology: J.F.R. Robertson, K.-L. Cheung, A. Skene, R. Littleford, S.Y.A. Cheung, J.P.O. Lindemann, P. Finlay, J.M.W. Gee

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): J.F.R. Robertson, R.E. Coleman, K.-L. Cheung, A. Evans, C. Holcombe, A. Skene, D. Rea, S. Ahmed, A. Jahan, K. Horgan, R. Littleford, S.Y.A. Cheung, J.P.O. Lindemann, P. Finlay, J.M.W. Gee

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): J.F.R. Robertson, K.-L. Cheung, A. Evans, D. Rea, S. Ahmed, P. Rauchhaus, S.Y.A. Cheung, M. Cullberg, E.C. de Bruin, L. Koulai, J.P.O. Lindemann, P. Rugman, G. Schiavon, A. Foxley, J.M.W. Gee

Writing, review, and/or revision of the manuscript: J.F.R. Robertson, R.E. Coleman, K.-L. Cheung, A. Evans, A. Skene, D. Rea, S. Ahmed, A. Jahan, K. Horgan, P. Rauchhaus, R. Littleford, S.Y.A. Cheung, M. Cullberg, E.C. de Bruin, L. Koulai, J.P.O. Lindemann, M. Pass, P. Rugman, G. Schiavon, R. Deb, J.M.W. Gee

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): J.F.R. Robertson, S. Ahmed, K. Horgan, R. Littleford, S.Y.A. Cheung, J.P.O. Lindemann, P. Rugman, P. Finlay, J.M.W. Gee

Study supervision: J.F.R. Robertson, R.E. Coleman, K.-L. Cheung, A. Jahan, K. Horgan, R. Littleford, J.P.O. Lindemann, J.M.W. Gee

Other (study supervision refers to the immunohistochemical biomarker component of the study): J.M.W. Gee

Capivasertib was discovered by AstraZeneca following collaboration with Astex Therapeutics (and its collaboration with the Institute of Cancer Research and Cancer Research Technology Limited). We are grateful for the assistance of the Tayside Clinical Trials Unit at the University of Dundee, the valuable insight from Barry Davies and Paul Elvin generated during early stages of capivasertib development (PK/PD in preclinical models, biomarker hypotheses, and selection), and the contribution of Steve Kelley, Martine Roudier, Claire Smith, and all the investigators and site staff, with special thanks to the patients and their families. AstraZeneca and Cancer Research UK provided funding for the study (award number A16395), and Julia Gee expresses gratitude for funding from the Breast Cancer Now Fellowship. Medical writing assistance was provided by Martin Goulding, DPhil, and Kristin Almond, PhD, from Mudskipper Business Ltd, supervised by the University of Nottingham.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1.
Brown
JS
,
Banerji
U
. 
Maximising the potential of AKT inhibitors as anti-cancer treatments
.
Pharmacol Ther
2017
;
172
:
101
15
.
2.
Lindsley
CW
. 
The Akt/PKB family of protein kinases: a review of small molecule inhibitors and progress towards target validation: a 2009 update
.
Curr Top Med Chem
2010
;
10
:
458
77
.
3.
Hassan
B
,
Akcakanat
A
,
Holder
AM
,
Meric-Bernstam
F
. 
Targeting the PI3-kinase/Akt/mTOR signaling pathway
.
Surg Oncol Clin N Am
2013
;
22
:
641
64
.
4.
Janku
F
,
Wheler
JJ
,
Naing
A
,
Falchook
GS
,
Hong
DS
,
Stepanek
VM
, et al
PIK3CA mutation H1047R is associated with response to PI3K/AKT/mTOR signaling pathway inhibitors in early-phase clinical trials
.
Cancer Res
2013
;
73
:
276
84
.
5.
Li
J
,
Davies
BR
,
Han
S
,
Zhou
M
,
Bai
Y
,
Zhang
J
, et al
The AKT inhibitor AZD5363 is selectively active in PI3KCA mutant gastric cancer, and sensitizes a patient-derived gastric cancer xenograft model with PTEN loss to Taxotere
.
J Transl Med
2013
;
11
:
241
.
6.
Lee
JJ
,
Loh
K
,
Yap
YS
. 
PI3K/Akt/mTOR inhibitors in breast cancer
.
Cancer Biol Med
2015
;
12
:
342
54
.
7.
Hyman
DM
,
Smyth
LM
,
Donoghue
MTA
,
Westin
SN
,
Bedard
PL
,
Dean
EJ
, et al
AKT inhibition in solid tumors with AKT1 mutations
.
J Clin Oncol
2017
;
35
:
2251
9
.
8.
Davies
BR
,
Greenwood
H
,
Dudley
P
,
Crafter
C
,
Yu
DH
,
Zhang
J
, et al
Preclinical pharmacology of AZD5363, an inhibitor of AKT: pharmacodynamics, antitumor activity, and correlation of monotherapy activity with genetic background
.
Mol Cancer Ther
2012
;
11
:
873
87
.
9.
Robertson
JF
,
Nicholson
RI
,
Bundred
NJ
,
Anderson
E
,
Rayter
Z
,
Dowsett
M
, et al
Comparison of the short-term biological effects of 7alpha-[9-(4,4,5,5,5-pentafluoropentylsulfinyl)-nonyl]estra-1,3,5, (10)-triene-3,17beta-diol (Faslodex) versus tamoxifen in postmenopausal women with primary breast cancer
.
Cancer Res
2001
;
61
:
6739
46
.
10.
Robertson
JF
,
Dixon
JM
,
Sibbering
DM
,
Jahan
A
,
Ellis
IO
,
Channon
E
, et al
A randomized trial to assess the biological activity of short-term (pre-surgical) fulvestrant 500 mg plus anastrozole versus fulvestrant 500 mg alone or anastrozole alone on primary breast cancer
.
Breast Cancer Res
2013
;
15
:
R18
.
11.
Dean
E
,
Banerji
U
,
Schellens
JHM
,
Krebs
MG
,
Jimenez
B
,
van Brummelen
E
, et al
A phase 1, open-label, multicentre study to compare the capsule and tablet formulations of AZD5363 and explore the effect of food on the pharmacokinetic exposure, safety and tolerability of AZD5363 in patients with advanced solid malignancies: OAK
.
Cancer Chemother Pharmacol
2018
;
81
:
873
83
.
12.
Cancer Research UK
. 
Research integrity: guidelines for scientific conduct
; 
2016
.
Available at
: https://www.cancerresearchuk.org/sites/default/files/scientific-conduct-guidelines.pdf.
13.
Banerji
U
,
Dean
EJ
,
Pérez-Fidalgo
JA
,
Batist
G
,
Bedard
PL
,
You
B
, et al
A phase 1 open-label study to identify a dosing regimen of the pan-AKT inhibitor AZD5363 for evaluation in solid tumors and in PIK3CA-mutated breast and gynecologic cancers
.
Clin Cancer Res
2018
;
24
:
2050
9
.
14.
Akaike
H
. 
A new look at the statistical model identification
.
IEEE Trans Automat Contr
1974
;
19
:
716
23
.
15.
Marous
M
,
Bieche
I
,
Paoletti
X
,
Alt
M
,
Razak
AR
,
Stathis
A
, et al
Designs of preoperative biomarkers trials in oncology: a systematic review of the literature
.
Ann Oncol
2015
;
26
:
2419
28
.
16.
Okuzumi
T
,
Fiedler
D
,
Zhang
C
,
Gray
DC
,
Aizenstein
B
,
Hoffman
R
, et al
Inhibitor hijacking of Akt activation
.
Nat Chem Biol
2009
;
5
:
484
93
.
17.
Tamura
K
,
Hashimoto
J
,
Tanabe
Y
,
Kodaira
M
,
Yonemori
K
,
Seto
T
, et al
Safety and tolerability of AZD5363 in Japanese patients with advanced solid tumors
.
Cancer Chemother Pharmacol
2016
;
77
:
787
95
.
18.
Zhang
X
,
Zhuang
T
,
Liang
Z
,
Li
L
,
Xue
M
,
Liu
J
, et al
Breast cancer suppression by aplysin is associated with inhibition of PI3K/AKT/FOXO3a pathway
.
Oncotarget
2017
;
12
:
63923
34
.
19.
Schmid
P
,
Abraham
J
,
Chan
S
,
Wheatley
D
,
Brunt
M
,
Nemsadze
G
, et al
AZD5363 plus paclitaxel versus placebo plus paclitaxel as first-line therapy for metastatic triple-negative breast cancer (PAKT): a randomised, double-blind, placebo-controlled, phase II trial
.
J Clin Oncol
2018
;
36
(
Suppl 15
):
abst 1007
.
20.
Jones
RH
,
Carucci
M
,
Claire
A
,
Butler
R
,
Alchami
F
,
Bale
CJ
, et al
Capivasertib (AZD5363) plus fulvestrant versus placebo plus fulvestrant after relapse or progression on an aromatase inhibitor in metastatic ER-positive breast cancer (FAKTION): a randomized, double-blind, placebo-controlled, phase II trial
.
J Clin Oncol
2019
;
37
(
Suppl 15
):
abst 1005
.
21.
AlFakeeh
A
,
Brezden-Masley
C
. 
Overcoming endocrine resistance in hormone receptor-positive breast cancer
.
Curr Oncol
2018
;
25
(
Suppl 1
):
S18
S27
.
22.
Turner
NC
,
Alarcon
E
,
Armstrong
AC
,
Philco
M
,
Lopez Chuken
YA
,
Sablin
MP
, et al
BEECH: a dose-finding run-in followed by a randomised phase II study assessing the efficacy of AKT inhibitor capivasertib (AZD5363) combined with paclitaxel in patients with estrogen receptor-positive advanced or metastatic breast cancer, and in a PIK3CA mutant sub-population
.
Ann Oncol
2019
;
30
:
774
80
.