Yoshimasa Kobayashi (1), Seiko Endo (1), Masato Fukae (1), Takatoshi Yonekura (1), Helen Kastrissios (2), Emi Kamiyama (1), Tushar Garimella (3), Takako Shimizu (1), Malaz A AbuTarif (3), Kazutaka Yoshihara (1)
(1) Daiichi Sankyo Co., Ltd. Tokyo, Japan, (2) Certara, L.P., NJ, USA, (3) Daiichi Sankyo Inc., NJ, USA
Introduction: Trastuzumab deruxtecan (T-DXd) is a newer antibody-drug conjugate consisting of an anti-HER2 antibody linked to a topoisomerase I inhibitor. T-DXd has been approved to treat HER2-positive metastatic breast cancer (BC) and advanced gastric cancer (GC) [1-3]. For the pivotal study of T-DXd for GC, a higher dose (6.4 mg/kg/Q3W) than the dose approved for BC (5.4 mg/kg/Q3W) was selected[4]. This dose selection was based on: (1) 6.4 mg/kg group showed a higher overall response rate (ORR) than the 5.4 mg/kg group in a Phase I study, while the adverse events (AEs) were tolerable and generally manageable, (2) the exposure of T-DXd at 6.4 mg/kg observed in the Phase I study in GC was similar to 5.4 mg/kg in BC, (3) several promising monoclonal antibodies (e.g. pertuzumab, cetuximab, trastuzumab emtansine (T-DM1)) failed to demonstrate efficacy for GC[5], some of which were reported to have lower exposure in GC than in other solid tumors[6-8].
Objective: Pharmacometric analysis conducted to justify 6.4 mg/kg in GC regarding risk/benefit balance.
Methods: Population pharmacokinetics (PPK) and exposure-safety (ES) analyses of T-DXd were performed using merged data from six Phase 1-2 studies including BC, GC and other tumor types at 0.8-8.0 mg/kg dose range. Exposure-efficacy (EE) analysis was performed using data from HER2-positive GC patients (N=160) primarily from the pivotal study for GC, DESTINY-Gastric01[4].
To construct the PPK model, a merged dataset including 14,044 intact T-DXd and 14,122 DXd PK samples collected from 808 subjects was used, after excluding observations below the lower limit of quantification (<1%) and 5 outlying observations. The base GC model was developed from the PPK model constructed for the initial BC application[9]. The final model was constructed by entering covariates selected from additional covariate candidates (tumor type (GC, BC, or other), total gastrectomy, race, alanine aminotransferase, aspartate aminotransferase, total bilirubin, creatinine clearance, and National Cancer Institute hepatic impairment score). Covariate selection was based on exploratory analyses and the likelihood ratio test.
In the EE analyses, ORR and confirmed ORR were analyzed by linear logistic regression. For ES analyses, drug-related interstitial lung disease (ILD) was analyzed by Cox regression. The other nine safety endpoints such as hematologic AEs were analyzed by linear logistic regression. Exposure metric was selected from Cmin and AUC at steady-state (ss) or at cycle1 of T-DXd and DXd, or from average serum concentration of them to the time of the event (Cave), based on statistical significance (the one with the minimum P value under 0.05). Further covariate analyses were conducted to assure that the relationships were not confounded by covariate imbalance.
Results: In the updated PPK model, tumor type was selected as a covariate affecting clearance and (central) volume of distribution of T-DXd and of DXd. The covariates for the base model and updated model were identical except for the removal of sex and the addition of tumor type. It was shown that the exposure of T-DXd in subjects with GC given a 6.4 mg/kg dose was similar to that with BC given the approved 5.4 mg/kg dose (e.g. post-hoc T-DXd Cmin,ss (μg/mL): 11.2 in BC with 5.4 mg/kg, 10.4 in GC with 6.4 mg/kg).
The EE analysis showed most statistically significant (P=0.023) relationship between Cmin,ss of T-DXd and confirmed ORR.
ES relationships were consistent with those observed in BC. The ES analysis showed most statistically significant (P<0.001) relationships between Cave of DXd and a hematologic AEs which were managed appropriately through routine clinical practice. The ES analysis showed most statistically significant (P<0.001) relationship between AUC,ss of T-DXd and the rate of ILD (any grade). The model-predicted incidence of ILD in Japanese subjects with GC given a 6.4 mg/kg dose was similar to that with BC given a 5.4 mg/kg dose (e.g. 10.2% vs 9.7% at 180 days), consistent with similar exposures of T-DXd at 6.4 mg/kg in GC and 5.4 mg/kg in BC.
Conclusions:The PPK and ES analyses showed that exposure of T-DXd and ILD risk at 6.4 mg/kg dose in GC are similar to those at the approved 5.4 mg/kg dose in BC. The EE analysis indicated additional benefit from using the higher 6.4 mg/kg dose compared to the 5.4 mg/kg. Based on these benefit-risk analyses, the dose of T-DXd 6.4 mg/kg was considered to be the optimal dose for GC population.
References:
[1] ENHERTU (fam-trastuzumab deruxtecan-nxki), Biologic License Application: 761139, Printed Labeling (2021), U.S. Food and Drug Administration, USA; https://www.accessdata.fda.gov/drugsatfda_docs/label/2021/761139s011lbl.pdf
[2] ENHERTU, Review Reports (2020), Pharmaceuticals and Medical Devices Agency, Japan; https://www.pmda.go.jp/files/000238706.pdf, https://www.pmda.go.jp/files/000238707.pdf
[3] ENHERTU, EPAR – Product information (2021), European Medicines Agency, EU; https://www.ema.europa.eu/en/medicines/human/EPAR/enhertu
[4] Shitara K, et al. N. Engl. J. Med. 2020;382:2419.
[5] Patel TH, et al. Curr. Treat. Options in Oncol. 2020;21:70.
[6] Han K, et al. AAPS J 2014;16:1056.
[7] Kirschbrown WP, et al. Cancer Chemother. Pharmacol. 2019;84:539.
[8] Chen SC, et al. Cancer Chemother. Pharmacol. 2017;80:1147.
[9] Yin O, et al. Clin. Pharmacol. Ther. 2020; doi.org/10.1002/cpt.2096
Reference: PAGE 29 (2021) Abstr 9868 [www.page-meeting.org/?abstract=9868]
Poster: Drug/Disease Modelling - Oncology