Naoki Kotani (1), Yachi Chen (1), Sravanthi Cheeti (1), Luna Musib (1), Jennifer Lauchle (1), Jennifer Schutzman (1), Pascal Chanu (2)
(1) Genentech, Inc., South San Francisco, CA, USA (2) Clinical Pharmacology, Genentech/Roche, France
Objectives: Tumor Growth Inhibition (TGI) modeling is a model-based approach developed for oncology to assess treatment-induced anti-tumor activity based on tumor size data collected in early clinical trials [1-3]. In this approach, the use of metrics capturing tumor dynamics has been proposed to enable comparison of treatment effects. This allows early detection of dose/exposure-response signals to support drug development decisions based on a small number of patients in addition to the traditional objective response rate.
In this work, the TGI modeling approach was applied to guide dose selection of two novel compounds currently in clinical development to treat metastatic breast cancer: GDC‑0077, a p110α-selective and mutant-degrading phosphoinositide 3-kinase (PI3K) inhibitor, and GDC‑9545, a potent oral selective estrogen receptor antagonist/degrader (SERD). The utility of this modeling framework using early clinical data to support clinical dose selection for a Phase 3 studies is presented herein using examples from GDC‑0077 and GDC‑9545.
Methods: Data used in the analyses were from early clinical trials for each molecule; the data from an ongoing Phase I/Ib study of GDC‑0077 (NCT03006172) included 97 patients with 423 tumor assessments, and the data from an ongoing Phase I study of GDC‑9545 (NCT03332797) included 80 patients with 267 tumor assessments. The Phase I/Ib study of GDC‑0077 consists of a single agent portion (testing daily 6, 9, and 12 mg of GDC-0077) and a combination portion (testing daily 3, 6, and 9 mg of GDC-0077 in combinations with standard of care therapies). The Phase I study of GDC‑9545 consists of a single agent portion (testing daily 10, 30, 90/100 and 250 mg of GDC‑9545) and a combination portion (testing daily 100 mg of GDC‑9545 in combinations with standard of care therapies); only single agent results were taken into consideration to support dose selection for GDC‑9545. The tumor size data (sum of longest diameters [SLD]) were collected during the screening period, at the end of Cycle 2, and every 8 weeks thereafter (GDC‑0077); or during the screening period and every 8 weeks after the initiation of study drug treatment (GDC‑9545). The empirical bi-exponential model was used to derive TGI metrics [3,4]. Data from each study were analyzed independently, and baseline tumor size (TS0), growth rate constant (KG), and tumor shrinkage rate constant (KS) were estimated. TGI metrics such as log(KG) and time to growth (TTG), reported as robust predictors of long-term survival [2], were derived from these parameter estimates and used to assess the dose-response relationship and inform clinical dose selection.
Results: The empirical TGI model well described the observed tumor size data collected in each clinical trial. All the parameter estimates were obtained with good precision for both GDC‑0077 and GDC‑9545 study data (RSE < 30%). For GDC‑0077, although the distribution of TGI metrics were overlapping, the 9 mg dose showed slower tumor growth represented by smaller log(KG) and greater duration of response represented by longer TTG compared with the 6 mg dose as a single agent (median KG 0.00171 vs 0.00220 days-1; median TTG 162 vs 94.8 days). Similar results were obtained in combination therapy with GDC‑0077. In GDC‑9545, there was no clear difference in the estimated TGI metrics across the tested dose groups, and this was helpful to make a safety-driven decision on the dose selection.
Conclusions: The totality of data including model-based approaches using TGI metrics supported the dose selection in both early development programs. The TGI modeling based on tumor kinetics is a quantitative and useful approach to assist in dose selection during the early phase drug development in oncology.
References:
[1] Claret L et al. J Clin Oncol (2009) 27:4103-8.
[2] Bruno R et al. Clin Pharmacol Ther (2014) 95:386-93.
[3] Claret L et al. Clin Cancer Res (2018) 24:3292-8.
[4] Stein WD et al. Clin Cancer Res (2011) 17:907-17.
Reference: PAGE () Abstr 9399 [www.page-meeting.org/?abstract=9399]
Poster: Oral: Drug/Disease Modelling