Application of the model-informed drug development paradigm to datopotamab deruxtecan dose selection for late-stage development

Abstract

To replace the conventional maximum tolerated dose (MTD) approach, a paradigm for dose optimization and dose selection that relies on model-informed drug development (MIDD) approaches has been proposed in oncology. Here, we report our application of an MIDD approach during phase I to inform dose selection for the late-stage development of datopotamab deruxtecan (Dato-DXd). Dato-DXd is a TROP2-directed antibody-drug conjugate being developed for advanced/metastatic non-small cell lung cancer (NSCLC) and other tumors. Data on pharmacokinetics (PKs), efficacy, and safety in NSCLC were collected in the TROPION-PanTumor01 phase I dose-expansion and -escalation study over a wide dose range of 0.27-10 mg/kg administered every 3 weeks. Population PK and exposure-response analyses were performed iteratively at three data cutoffs to inform dose selection. The 6 mg/kg dose was identified as the optimal dose by the second data cutoff analysis and confirmed by the subsequent third data cutoff analysis. The 6 mg/kg dose was more tolerable (i.e., lower rates of interstitial lung disease, stomatitis, and mucosal inflammation) than the MTD (8 mg/kg) and was more efficacious than 4 mg/kg (simulated mean objective response rate: 23.8% vs. 18.6%; mean hazard ratio of progression-free survival: 0.74) - a candidate dose studied just below 6 mg/kg. Therefore, the 6 mg/kg dose was judged to afford the optimal benefit-risk balance. This case study demonstrated the utility of an MIDD approach for dose optimization and dose selection.

FIGURE 1

Model‐derived relationship between exposure (Dato‐DXd C _avg) and ORR in a reference patient (top) and observed Dato‐DXd C _avg at different doses (bottom). Top panel: Solid curve and shaded area represent the mean and 95% CI of the C _avg‐ORR relationship. Bottom panel: Horizontal box plots represent the C _avg distributions at the specified dose levels, the whiskers indicate 5th and 95th percentiles, the boxes cover 25th to 75th percentiles, and the vertical bars indicate the medians. C _avg, average concentration of Dato‐DXd to tumor response or, for those who did not respond, the end of the last dosing cycle up to the data cutoffs; CI, confidence interval; Dato‐DXd, datopotamab deruxtecan; ORR, objective response rate.

FIGURE 2

Model‐simulated results that informed benefit–risk assessment to select 6 mg/kg as an optimal dose for further development. (a) Simulated steady‐state AUC_tau for Dato‐DXd and unconjugated DXd; (b) simulated probability of response (complete or partial response); (c) simulated PFS descriptors; (d) simulated probabilities for indicated TEAEs; and (e) simulated hazard ratios for experiencing the indicated TEAEs when treated with Dato‐DXd 6 mg/kg relative to 4 mg/kg in a virtual population with non‐small cell lung cancer. Dashed lines represent 20% and 30% probabilities. AE, adverse event; AUC_tau, area under the curve over a dosing cycle; CI, confidence interval; Dato‐DXd, datopotamab deruxtecan; HR, hazard ratio; ILD, interstitial lung disease; PFS, progression‐free survival; TEAE, treatment‐emergent adverse event.