Exposure‐Response Analyses of Datopotamab Deruxtecan (Dato‐DXd) in Patients with Hormone Receptor (HR) Positive, Human Epidermal Growth Factor Receptor 2 (HER2)‐Negative Breast Cancer

Abstract

Dato‐DXd (datopotamab deruxtecan) is an anti‐TROP2 antibody‐drug conjugate developed for the treatment of unresectable or metastatic hormone receptor‐positive, human epidermal growth factor receptor 2‐negative breast cancer (HR+/HER2‐BC) who have received prior endocrine‐based therapy and chemotherapy for unresectable or metastatic disease. Associations between Dato‐DXd or its payload DXd pharmacokinetics (PK) and various efficacy and safety endpoints were investigated in the current analysis. Exposure‐efficacy analysis was conducted in a total of 352 patients with HR+/HER2‐BC who received Dato‐DXd at 6 mg/kg every 3 weeks (Q3W). Area under the curve in cycle 1 (AUC1) was identified as a significant covariate of overall survival (OS), with higher Dato‐DXd exposure associated with longer OS. Progression‐free survival (PFS) showed a positive trend with higher exposure, though baseline tumor size emerged as a more significant predictor in multivariable analyses. Exposure‐safety analysis revealed correlations between higher Dato‐DXd or DXd PK metrics and increased risk of selected safety events, including stomatitis and ocular surface events. PopPK analysis showed that Dato‐DXd exposure increased with body weight, and implementing a capped dose of 540 mg for patients ≥90 kg helped normalize exposure across weight groups. Exposure‐safety analysis further supported that this approach may reduce the potential associated safety risks. These results support the recommended Dato‐DXd dosing regimen (6 mg/kg Q3W with dose capping for patients ≥90 kg) and highlight its favorable benefit‐risk profile in patients with HR+/HER2‐breast cancer.

Introduction

Hormone receptor‐positive/human epidermal growth factor receptor 2‐negative (HR+/HER2‐) breast cancer is the most common subtype of breast cancer, accounting for approximately 70% of all cases. ¹ Despite the availability of endocrine therapies and targeted agents such as CDK4/6 inhibitors, resistance frequently develops, leading to disease progression. Chemotherapy remains a cornerstone of treatment after endocrine resistance, but its efficacy is often limited by modest response rates, short progression‐free survival (PFS) and significant toxicity. ²

Trophoblast cell surface antigen 2 (TROP2), a transmembrane glycoprotein, is highly expressed in multiple solid tumors, including breast cancer, and is associated with aggressive disease and poor prognosis. ³ Datopotamab deruxtecan (Dato‐DXd) is an antibody‐drug conjugate (ADC) composed of a humanized anti‐TROP2 monoclonal antibody linked to a potent topoisomerase I inhibitor payload (DXd) via a stable, tumor‐selective tetrapeptide‐based cleavable linker. ⁴ Dato‐DXd enables targeted delivery of cytotoxic payload DXd to TROP2‐expressing tumor cells with an optimized drug‐to‐antibody (DAR) ratio of approximately 4. After Dato‐DXd internalization, the linker is selectively cleaved, releasing the DXd to eliminate both target and neighboring tumor cells while minimizing systemic toxicity. ⁴ , ⁵ Cytochrome P450 (CYP) 3A4 was the primary CYP enzyme involved in the metabolism of active payload DXd, while DXd did not exhibit any potential to inhibit or induce CYP enzymes at clinically relevant concentrations. ⁶ DXd does not undergo metabolism by UGT enzymes (data on file).

Dato‐DXd has demonstrated promising efficacy in HR+/HER2‐ BC in multiple clinical studies. ⁷ , ⁸ In the pivotal Phase 3 TROPION‐Breast01 (NCT05104866), patients receiving Dato‐DXd had statistically significant and clinically meaningful improvement in PFS compared with investigator's choice of chemotherapy (ICC) (PFS by Blinded Independent Central Review [BICR] hazard ratio (HR), 0.63 (95% CI, 0.52 to 0.76); P < .0001). ⁷ In 2025, Dato‐DXd was approved by the US Food and Drug Administration (FDA) for adult patients with unresectable or metastatic HR+/HER2‐ BC (immunohistochemistry (IHC) 0, IHC1+ or IHC2+/in situ hybridization [ISH]‐) who have received prior endocrine‐based therapy and chemotherapy for unresectable or metastatic disease. ⁹

Dato‐DXd PK have been evaluated in the dose range of 0.27 to 10 mg/kg Q3W in TROPION‐PanTumor01 (TP01). The PK profiles of total datopotamab antibody and Dato‐DXd were similar, suggesting high linker stability in plasma. The linear clearance was the major elimination pathway for Dato‐DXd doses of 4 mg/kg and above. The estimated linear clearance and volume of distribution of Dato‐DXd was 0.386 L/day and 5.94 L (3.06 + 2.88), respectively. ¹⁰ The mean half‐life of Dato‐DXd at cycle 1 was 4.8 days at 6 mg/kg Dato‐DXd. ¹¹ The estimated linear clearance and volume of distribution of DXd was 2.66 L/h and 25.1 L, respectively. ¹⁰ PK simulation and exposure safety analyses demonstrated that higher baseline body weight was associated with increased PK exposure of Dato‐DXd and DXd, as well as with a higher incidence of selected toxicities, including grade ≥ 3 treatment emergent adverse events (TEAE) and grade ≥ 2 stomatitis and ocular surface events. Based on the totality of these data, a weight‐based dose cap for Dato‐DXd is recommended, capping the dose of Dato‐DXd at 540 mg for patients weighing ≥90 kg, to ensure consistent PK exposure across weight groups and mitigate the risk of toxicities in high body weight patients. The dose capping is predicted to enable patients weighing ≥90 kg to achieve systemic exposures comparable to patients weighing <90 kg. ¹²

Dato‐DXd at 6 mg/kg Q3W was selected as the optimal dose balancing between efficacy and safety, based on data from Phase I TROPION‐PanTumor01 (NCT03401385). In TROPION‐PanTumor01, Dato‐DXd was evaluated at dose levels of 0.27, 0.5, 1.0, 2.0, 4.0, 6.0, 8.0, and 10.0 mg/kg Q3W in patients with non‐small cell lung cancer (NSCLC). ¹¹ The maximum tolerated dose (MTD) was determined to be 8 mg/kg Q3W based on the occurrence of dose‐limiting toxicities (DLTs). Based on a totality of data, 6 mg/kg Q3W was selected to be further evaluated in clinical studies for breast cancer, ¹³ , ¹⁴ including TROPION‐Breast01.

A population pharmacokinetics (PopPK) model was previously developed to characterize the PK of Dato‐DXd and free payload DXd. ¹⁰ , ¹⁵ Briefly, Dato‐DXd PK was characterized by a two‐compartment model with parallel Dato‐DXd linear and nonlinear clearance. DXd PK was described by a one‐compartment model with first order elimination. The release of DXd from the intact Dato‐DXd was driven by the total elimination rate of Dato‐DXd. ¹⁰ , ¹⁵ Exposure‐response (E‐R) analyses based on the PopPK model have informed the optimal dose justification of Dato‐DXd in patients with HR+/HER2‐ BC. Here, we evaluate the relationship between Dato‐DXd and/or DXd exposures and key efficacy and safety endpoints, using data from TROPION‐PanTumor01, TROPION‐Lung01 (NCT04656652), TROPION‐Lung05 (NCT04484142) and TROPION‐Breast01 (NCT05104866), to further justify optimal dose selection of Dato‐DXd in patients with HR+/HER2‐ BC.

Materials and Methods

Study Populations and End Points

Data for exposure‐response analyses were derived from four studies TROPION‐PanTumor01, TROPION‐Lung01, TROPION‐Lung05, and TROPION‐Breast01, which are summarized in Table 1. All four studies were conducted in accordance with the Declaration of Helsinki and were consistent with International Conference on Harmonization and Good Clinical Practice guidelines, and applicable regulatory requirements. Written informed consent from participants was obtained before performing any protocol‐related procedures. Dato‐DXd was administered Q3W, with varying doses used across studies (TROPION‐PanTumor01: 0.27–10 mg/kg; TROPION‐Lung01, TROPION‐Lung05, and TROPION‐Breast01: 6 mg/kg).

Table 1.

Clinical Studies Included in the Exposure‐Safety Analysis

Study ID	Study title	Number of patient a	Treatment groups
TROPION‐PanTumor01 1 , 2	Phase 1, two‐part, multicenter, open‐label, multiple dose, first‐in‐human study of DS‐1062a in subjects with advanced solid tumors	210 (lung cancer)	Dato‐DXd 0.27, 0.5, 1.0, 2.0, 4.0, 6.0, 8.0, and 10.0 mg/kg Q3W administered as an intravenous (IV) infusion.
		41 (HR+/HER2‐ BC)	Dato‐DXd 6.0 mg/kg Q3W IV
		44 (TNBC)	Dato‐DXd 6.0 and 8.0 mg/kg Q3W IV
TROPION‐Breast01 3	Phase III, open‐label, randomized study of Dato‐DXd versus Investigator's choice of single‐agent chemotherapy in participants with inoperable or metastatic HR+/HER2‐ breast cancer who have been treated with one or two prior lines of systemic chemotherapy	365 (HR+/HER2‐ BC)	Dato‐DXd 6.0 mg/kg Q3W IV
TROPION‐Lung05 4	Phase 2, single‐arm, open‐label study of DS‐1062a in advanced or metastatic non‐small cell lung cancer with actionable genomic alterations and progressed on or after applicable targeted therapy and platinum‐based chemotherapy (TROPION‐Lung05)	137 (NSCLC)	Dato‐DXd 6.0 mg/kg Q3W IV
TROPION‐Lung01 5	Phase 3 randomized study of DS‐1062a versus docetaxel in previously treated advanced or metastatic non‐small cell lung cancer with or without actionable genomic alterations	299 (NSCLC)	Dato‐DXd 6.0 mg/kg Q3W IV

^a Number of patients enrolled in the study may not be eligible to be included in the PopPK analysis or exposure‐response analysis. Data cut‐off: April 29, 2024.

Data for exposure‐efficacy analysis were derived from TROPION‐Breast01. The efficacy data set included efficacy‐ and PK‐evaluable patients (those who were treated with Dato‐DXd, had both baseline and ≥1 postbaseline tumor assessments, and had ≥1 post‐baseline PK record) with HR+/HER2‐ BC in TROPION‐Breast01. Efficacy end points included OS and PFS per independent central review (data cut‐off [DCO]: April 29, 2024).

The safety data set included evaluable patients (those who had received ≥1 dose of Dato‐DXd and had ≥1 post‐baseline PK record) from the TROPION‐PanTumor01, TROPION‐Lung01, TROPION‐Lung05, and TROPION‐Breast01 studies, including both patients with breast cancer and lung cancer who received Dato‐DXd from 0.27–10 mg/kg. Analyzed safety endpoints included (1) Grade 3+ treatment‐emergent adverse events (TEAE), (2) serious TEAEs (SAE), (3) TEAE leading to dose interruption (TROPION‐Breast01 and TROPION‐PanTumor01 only), (4) TEAE leading to dose reduction, (5) TEAE leading to treatment discontinuation, (6) any grade or (7) grade 2+ oral mucositis or stomatitis, (8) any grade or (9) grade 2+ ocular surface events (OSEs), and (10) any grade adjudicated drug‐related interstitial lung disease (ILD). Oral mucositis or stomatitis was defined by a single preferred term (Stomatitis) based on MedDRA v26.0. OSE was defined by preferred term list based on MedDRA v26.0.

A comprehensive glossary of abbreviations used throughout the manuscript is provided in the Appendix for reference (Table S1, Supporting Information).

PK Exposure Measures

The bioanalytical methods for measuring Dato‐DXd and DXd PK and the PopPK model describing Dato‐DXd and DXd PK were described previously. ⁹ , ¹⁰ The PK exposure metrics of Dato‐DXd and DXd were estimated for individual patients based on empirical Bayes estimates (EBE) of PopPK analysis. ¹⁵ , ¹⁶ Exposure metrics included maximum concentration at cycle 1 (Cmax1), AUC at cycle 1 (AUC1), and average concentration (C_avg) for both Dato‐DXd and DXd. C_avg was derived based on the PopPK model, calculated as cumulative AUC divided by time duration. The time duration was from the first dose to the end of the cycle of the event for patients with events. For patients who were censored, duration was from the first dose to the cycle of the last treatment or data cut‐off, whichever occurred first. C_avg is unique for each efficacy and safety endpoint and considering dose interruption and dose reduction.

Covariate Selection

A stepwise covariate selection procedure was applied for both Cox proportional hazards (CPH) and logistic regression analyses. ¹⁷ As the first step, the most statistically significant PK exposure metrics were prioritized. For PFS and OS, Dato‐DXd AUC1 and C_avg were evaluated in the exposure‐efficacy analysis. For safety endpoints, Cmax1, AUC1, and C_avg of both Dato‐DXd and DXd were assessed.

When statistically significant PK exposure metrics were identified, predefined covariate‐parameter relationships were tested using forward inclusion. Model selection was guided by the log‐likelihood ratio test. In PK exposure metric selection and forward inclusion, covariates with p‐values <.01 were considered significant and added sequentially. Each covariate was first evaluated univariately, and the one producing the greatest improvement over the base model was retained for the next round of testing. In backward elimination, covariates with p‐values >.001 were removed iteratively until no further insignificant covariate‐parameter relationships remained.

Exposure‐Efficacy Analysis

Initial exploratory plots were used to assess exposure‐response relationships. Kaplan–Meier curves were plotted by PK exposure quartiles, together with a log rank test. Following graphical exploration, the OS and PFS was analyzed with CPH model with Dato‐DXd PK exposures as one covariate and other potential covariates, according to the standard stepwise forward‐addition and backward‐elimination processes, as demonstrated in the covariate selection section. ¹⁸

$$\mathrm{h}\left(\mathrm{t}\right)={\mathrm{h}}_0\left(\mathrm{t}\right)\times {\mathrm{e}}^{{\theta }_{\mathrm{i}}\left[{\mathbf{x}}_{\mathbf{i}}\right]+{\theta }_{\mathrm{j}}\left[{\mathbf{x}}_{\mathbf{j}}/{\mathbf{x}}_{\mathbf{median}}\right]+{\theta }_{\mathrm{k}}\left[\mathbf{PK}\right]}$$

where h(t) is the hazard function for subject i at time t, h0(t) is the baseline hazard function, ${\theta }_{\mathrm{i}}$ is the coefficient for a binary covariate ${\mathbf{x}}_{\mathbf{i}}$, ${\theta }_{\mathrm{j}}$ is the coefficient for a continuous covariate ${\mathbf{x}}_{\mathbf{j}}$ normalized by the median (${\mathbf{x}}_{\mathbf{meidan}})$, and ${\theta }_{\mathrm{k}}$ is the coefficient for the exposure value PK.

The impact of covariates on PFS and overall survival OS was illustrated using forest plots of hazard ratios derived from the final CPH models. For OS, Dato‑DXd AUC1 was evaluated, and for PFS, baseline tumor size was assessed. Each covariate was stratified by the 5th, 25th, 75th, and 95th percentiles, with the median value serving as the reference. Hazard ratios and 95% confidence intervals were displayed to demonstrate the relative effect of covariates across the distribution.

Exposure‐Safety Analysis

Safety endpoints were modeled using linear logistic regression to characterize the relationship between Dato‐DXd or DXd exposure and event rates. For safety endpoints, Cmax1, AUC1, and C_avg of Dato‐DXd and DXd were evaluated. Covariates were tested on both the intercept and the slope of the exposure‐safety relationship. All categorical variables were converted to binary responses, and analyzed with linear logistic regression models:

$${\mathrm{P}}_{\mathrm{event}}=\frac{1}{1+{\mathrm{e}}^{-\left(\theta +{\theta }_{\mathrm{i}}\left[{\mathbf{x}}_{\mathbf{i}}\right]+{\theta }_{\mathrm{j}}\left[{\mathbf{x}}_{\mathbf{j}}/{\mathbf{x}}_{\mathbf{median}}\right]+{\theta }_{\mathrm{k}}\left[\mathbf{PK}\right]\right)}}$$

where logit is the logit transform and ${\mathrm{P}}_{\mathrm{event}}$ is the probability of an event (response). $\theta$ is the intercept that accounts for the baseline probability. ${\theta }_{\mathrm{i}}$ is the coefficient for a binary covariate ${\mathbf{x}}_{\mathbf{i}}$, ${\theta }_{\mathrm{j}}$ is the coefficient for a continuous covariate ${\mathbf{x}}_{\mathbf{j}}$ normalized by the median (${\mathbf{x}}_{\mathbf{median}})$, and ${\theta }_{\mathrm{k}}$ is the coefficient for the exposure value PK.

The impact of covariates on safety event probabilities was illustrated using forest plots of odds ratio derived from the final logistic regression model. Continuous covariate was stratified by the 5th, 25th, 75th, and 95th percentiles, with the median value serving as the reference. Odds ratios and 95% confidence intervals were displayed to demonstrate the relative effect of covariates across the distribution.

The final exposure‐safety logistic regression model for each safety endpoint was used to predict event rates at 4, 6, and 8 mg/kg, based on the median PopPK‐predicted PK metric values at each dose level.

Exposure‐response model evaluation and downstream analysis were performed using R (version 4.3).

Results

Baseline Characteristics

Patient baseline characteristics are summarized in Table S2, Supporting Information.

Data from 352 Dato‐DXd‐treated patients with HR+/HER2‐ BC in TROPION‐Breast01 were used in the exposure‐efficacy analyses. Eight patients without any PK records were excluded from the exposure‐efficacy analysis in TROPION‐Breast01. All 352 patients received Dato‐DXd at 6 mg/kg Q3W. The median age of patients included in the exposure‐efficacy analysis dataset was 55 years (range, 29–86), and the median body weight was 62 kg (range, 35.6–141). Of these patients, 98.6% were female, 50% were white, and 38.1% were from Europe. A total of 83% patients (292 of 352) received prior lines of CDK4/6 therapy, 18.8% (66 of 352) patients have received more than two last lines of CDK4/6 therapy.

Data from 1081 patients with breast or lung cancer were included in the exposure‐safety analysis. The majority of patients received Dato‐DXd at 6 mg/kg (85%, 919 of 1081). The median age of patients included in the exposure‐safety analysis dataset was 60 years (range, 26–86), and the median body weight was 64.2 kg (range, 35.6–156). A total of 67.2% of patients were female, 59.6% had lung cancer, 47.2% were white, and 29.7% were from the USA.

ExposureEfficacy Analyses

An apparent positive relationship was shown between Dato‐DXd AUC1 and OS (Figure 1A), and between baseline tumor size and OS (Figure 1B). Positive relationships with PFS were observed for baseline tumor size (Figure 1C) and Dato‐DXd AUC1 (Figure 1D).

Figure 1.

Overall survival (OS) and progression‐free survival (PFS) of patients with HR‐positive/HER2‐negative breast cancer (HR+/HER2‐ BC) in TROPION‐Breast01. (a) Kaplan–Meier analysis stratified by Dato‐DXd AUC1 quartiles or (b) baseline tumor size quartiles. Legends represented median OS (95% CI) by quartiles. (c) Kaplan–Meier analysis stratified by baseline tumor size quartiles or (d) Dato‐DXd AUC1 quartiles. Legends represented median PFS (95% CI) by quartiles. (e) Forest plots of covariates on OS. (f) Forest plots of covariates on PFS. In (e) and (f), the dot represents the median hazard ratio, while the bar represents the 95% CI. The reference value is expressed in the variable name, median for continuous variables.

The CPH model for OS was developed based on patients with HR+/HER2‐ BC in TROPION‐Breast01 (N = 352). Dato‐DXd AUC1 was selected in the forward addition (p < .01) and kept in backward elimination (p < .001). No other covariates were identified as significant, as described by the CPH model below.

$$\mathrm{Overall}\mathrm{survival}:h\left(t\right)={\mathrm{h}}_0\left(\mathrm{t}\right)\times {\mathrm{e}}^{-0.0031\left[\mathrm{Dato}-\mathrm{DXd}\mathrm{AUC}1\right]}$$

Higher values of Dato‐DXd AUC1 were significantly associated with longer OS. Hazard ratios were derived from the final CPH model for OS and presented in Figure 1E. The 95th percentile and 5th percentile of Dato‐DXd (864 and 467 µg·day/mL, respectively) showed a hazard ratio of approximately 0.515 (95% CI, 0.39–0.68) and 1.76 (95% CI, 1.39–2.24), respectively, with respect to median Dato‐DXd (650 µg·day/mL; Figure 1E).

Dato‐DXd AUC1 was not considered as a significant covariate for PFS based on the CPH evaluation (p > .001). Baseline tumor size was selected as the only statistically significant covariate (p < .001). The relationship between baseline tumor size and PFS can be described with the model below:

$$\begin{array}{c}\hfill \mathrm{progression}-\mathrm{free}\mathrm{survival}:\mathrm{h}\left(\mathrm{t}\right)={\mathrm{h}}_0\\ \hfill \left(\mathrm{t}\right)\times {\mathrm{e}}^{0.28\left[\mathrm{Baseline}\mathrm{tumor}\mathrm{size}\right]}\end{array}$$

Greater baseline tumor size is associated with higher PFS risk. For baseline tumor size, the 95th percentile (138 mm) and the 5th percentile (17 mm) showed hazard ratios of approximately 1.52 (95% CI, 1.24–1.86) and 0.825 (95% CI, 0.752–0.906) when compared to the median tumor size of 55 mm (Figure 1F).

Exposure‐Safety Analysis

A total of 1081 patients (n = 437 with breast cancer) were evaluable for safety across the four studies. Among 1081 patients, four, five, seven, and six patients received the lower doses of 0.27, 0.5, 1, and 2 mg/kg, respectively, whereas 50, 919, 82, and 8 patients received the higher doses of 4, 6, 8, and 10 mg/kg, respectively. A summary of the observed incidences of events for the safety end points is provided in Table 2.

Table 2.

Observed Incidences of Events for the Safety End Points at 4, 6, and 8 mg/kg in the Safety Analysis Dataset

Event	4 mg/kg	6 mg/kg	8 mg/kg
Grade 3+ TEAE	30% (15/50)	41% (376/917)	60% (49/82)
SAE	20% (10/50)	23% (212/919)	50% (41/82)
TEAE leading to dose interruption a	8% (4/50)	24% (117/485)	35% (29/82)
TEAE leading to dose reduction	2% (1/50)	21% (193/918)	28% (23/82)
TEAE leading to dose discontinuation	16% (8/50)	8% (72/919)	23% (19/82)
Any grade stomatitis	42% (21/50)	55% (504/919)	55% (45/82)
Grade 2+ stomatitis	8% (4/50)	29% (268/919)	34% (28/82)
Adjudicated drug‐related ILD	10% (5/50)	5% (42/919)	12% (10/82)
Any grade OSE	26% (13/50)	34% (311/919)	42% (34/82)
Grade 2+ OSE	8% (4/50)	10% (88/919)	13% (11/82)

Abbreviations: ILD, interstitial lung disease; OSE, ocular surface events; SAE, serious TEAEs; TEAE, treatment emergent adverse events.

^a Analysis of TEAE leading to dose interruption included TROPION‐Breast01 and TROPION‐PanTumor01 only.

Adjudicated drug‐related ILD and TEAE leading to treatment discontinuation had no significant exposure‐response relationships to all tested PK metrics (i.e., Cmax1, AUC1, and C_avg of Dato‐DXd and DXd; Figures S1 and S2, Supporting Information). Comparably low adjudicated drug‐related ILD rates were observed at dose levels of 4, 6, and 8 mg/kg, with rates of 10%, 5%, and 12%, respectively. Analysis through distribution plots of Dato‐DXd and DXd PK metrics (Figure S1, Supporting Information) and covariate selection of a logistic regression model, indicated no clear relationship between PK metrics and adjudicated drug‐related ILD. While the distribution plots suggested an apparent relationship between Dato‐DXd or DXd PK exposure and treatment discontinuation (Figure S2, Supporting Information), covariate analysis suggested no significant E‐R relationship between Dato‐DXd or DXd PK and the incidences of treatment discontinuation (p > .001).

All the other safety end points (i.e., Grade 3+ TEAE, SAE, TEAE leading to dose interruption, TEAE leading to dose reduction, any grade or Grade 2+ oral mucositis or stomatitis, and any grade or Grade 2+ OSEs, respectively) were correlated with PK exposure metrics. The final exposure‐safety models derived through stepwise covariate selection suggested increased event probabilities with increasing exposure of intact Dato‐DXd or DXd (Table 3). DXd C_avg is associated with increased risks of Grade 3+ TEAEs and SAEs. Dato‐DXd exposure metrics are correlated with the other six safety endpoints. Dato‐DXd AUC1 is correlated with TEAEs leading to dose interruption, TEAEs leading to dose reduction, as well as with Grade 2+ stomatitis. Dato‐DXd C_avg is associated with the incidence of any grade stomatitis, any grade OSE, and Grade 2+ OSE. Figure 2 shows logistic regression plots for Grade 3+ TEAE (Figure 2A), SAE (Figure 2B), Grade 2+ stomatitis (Figure 2C), and Grade 2+ OSE (Figure 2D). Caution is warranted when interpreting the analysis results, given the potential bias introduced by the derivation of C_avg. ¹⁹ , ²⁰ For the safety endpoints where C_avg was identified as the most statistically significant PK covariate, other PK exposure metrics (e.g., AUC1) also demonstrated a consistent exposure‐safety relationship (data on file). Exposure‐safety logistic regression plots for the other safety endpoints are shown in Figure S3, Supporting Information.

Table 3.

Overview of Significant Exposure Metrics and Covariates in the Final Logistic Regression Model

Safety endpoints	Significant exposure metrics	Significant covariates
Grade ≥3 TEAEs	DXd Cavg	Region, smoke status
Serious TEAEs	DXd Cavg	Baseline albumin, tumor type, race
TEAE leading to dose interruption	Dato‐DXd AUC1	None
TEAE leading to dose reduction	Dato‐DXd AUC1	Baseline tumor size
TEAE leading to treatment discontinuation	None	Not tested
Oral mucositis/stomatitis any grade a	Dato‐DXd Cavg	None
Oral mucositis/stomatitis ≥ Grade 2 a	Dato‐DXd AUC1	None
Adjudicated drug‐related ILD	None	Not tested
Ocular surface event any grade b	Dato‐DXd Cavg	Tumor type (NSCLC vs BC)
Ocular surface event ≥ Grade 2 b	Dato‐DXd Cavg	None

Note: Significant covariates obtained a p‐value < .01 when selected from forward addition, and a p‐value < .001 when removed from backward elimination.

^a Defined by single preferred term (Stomatitis) based on MedDRA v26.0.

^b Defined by preferred term list based on MedDRA v26.0.

Figure 2.

Logistic regression plots for (a) Grade 3+ TEAE, (b) SAE, (c) Grade 2+ stomatitis, and (d) Grade 2+ OSE. Solid back lines and shaded areas represent model predicted safety event incidences and corresponding 95% confidence interval. Points around the solid back lines represent the observed proportions of event per exposure quartile. Vertical error bars represent the 95% confidence interval of the observed data. Fractions indicate the total number of events over the total number of patients within the exposure quantile; All dose levels from 0.27–10 mg/kg were included in the analysis, and doses at 4, 6, and 8 mg/kg were highlighted by the horizontal bars. Horizontal bars show the exposure at 4 mg/kg (n = 50), 6 mg/kg (n = 919), and 8 mg/kg (n = 82) in the pooled population with lung or breast cancer. Lower ends, black dots, and upper ends of the horizontal bars represent the 25th percentile, median, and 75th percentile, respectively. Vertical dashed lines represent the median of PK exposure at 4, 6, and 8 mg/kg.

Analyses of covariates were conducted in eight safety endpoints that have positive exposure‐safety relationships. Table S3, Supporting Information, included the pre‐defined covariates evaluated in the stepwise covariate selection. Significant covariates retained in the final model and parameter estimates for the logistic regression models are summarized in Table S4, Supporting Information, and Table 3. Forest plots were generated to illustrate the odds ratios by covariates (Figure 3 and Figure S4, Supporting Information).

Figure 3.

Forest plots for (a) Grade 2+ stomatitis and (b) Grade 2+ OSE. The point represents the median hazard ratio, while the bar represents the 95% CI. The reference value is expressed in the variable name, median for continuous variables. OSE, ocular surface events.

Tumor type was identified as a significant covariate in SAEs and any grade OSE. There is a lower probability of SAE in patients with breast cancer compared to patients with lung cancer (odds ratio [95% CI] = 0.425 [0.301 − 0.601], Figure S4B, Supporting Information), and a higher risk of any grade OSE (Odds ratio [95% CI] = 2.83 [2.14 − 3.74], respectively. Figure S4, Supporting Information). Tumor type was not a significant covariate in other safety analysis.

Regional effect or race effect are significant covariates for Grade 3+ TEAE or SAE. Patients from the United States had a higher likelihood of presenting with Grade 3+ TEAEs compared to other regions. The rest of the world, China, Europe, or Japan decreased the Grade 3+ TEAE odds ratio to 0.465, 0.671, 0.707, and 0.531, respectively (Figure S4A, Supporting Information). Compared to White patients, Asian patients had a lower probability of developing SAE (Odds ratio [95% CI] = 0.424 [0.303 − 0.593], respectively (Figure S4B, Supporting Information).

Patients who are former or current smokers had a higher probability of developing Grade 3+ TEAEs compared to the patients who never smoked (Odds ratio [95% CI] = 1.57 [1.21 − 2.04], respectively (Figure S4A, Supporting Information). Patients who had higher baseline albumin levels were less likely to have SAE (Figure S4B, Supporting Information). Baseline albumin levels at the 95th and 5th percentiles (46 and 29 g/L) showed a predicted odds ratio of 0.587 and 2.14, relative to the median albumin level (39 g/L, respectively). Patients with higher tumor baseline were less likely to have a dose reduction (Figure S4C, Supporting Information). Baseline tumor size at the 95th and 5th percentiles (166 and 18 mm) showed a predicted odds ratio of 0.489 and 1.33, relative to the median tumor size baseline (60 mm). No significant covariates besides PK metrics were identified for TEAE leading to dose interruption, any grade or Grade 2+ oral mucositis or stomatitis, or Grade 2+ OSEs.

The final model developed for each AE endpoint has been applied to predict the probability of each AE incidence at 4, 6, and 8 mg/kg doses of Dato‐DXd, based on the median PopPK‐predicted PK exposure metrics (Figure S5, Supporting Information). Predicted probabilities of different safety events at different dose levels well‐agreed with observed incidences (data not shown). Dose‐response projections suggested a numerical increase in safety risks from 4 to 8 mg/kg. Absolute increases in projected Grade 3+ TEAE were 11.8% from 6 to 8 mg/kg (39.5% vs 51.3%), and 6.2% from 4 to 6 mg/kg (33.3% vs 39.5%). Grade 2+ OSE incidence increased by 7% from 6 to 8 mg/kg of Dato‐DXd, but only increased by 3.3% from 4 to 6 mg/kg.

Dato‐DXd and DXd exposure increase with increasing body weight (Figure S6, Supporting Information). Compared to patients with body weight below 60 kg (N = 405), patients with progressively higher weight, including >90–100 kg (N = 56), >100–110 kg (N = 27), >110–120 kg (N = 11), and above 120 kg (N = 7), had 34%, 67%, 56%, and 75% increase in Dato‐DXd AUC1, respectively. Similar trends were observed for DXd PK exposure. Implementing a capped dose of 540 mg for patients with a body weight above 90 kg is predicted to align Dato‐DXd and DXd exposures with those observed in lower‐weight patients (<90 kg) receiving the standard 6 mg/kg dose (Figure S6, Supporting Information). Switching from a weight‐based dose to a flat dose of 540 mg for patients > 90 kg results in decreases in the odds ratio of multiple safety endpoints when compared to patients weighing 60–80 kg (data not shown).

Discussion and Conclusions

The pivotal TROPION‐Breast01 trial established Dato‐DXd as a new treatment option for patients with HR+/HER2‐ BC, addressing a critical unmet need in this population. The current analysis represents the first comprehensive evaluation of the relationships between Dato‐DXd PK exposure and PFS and OS in patients with HR+/HER2‐ BC, as well as the exposure‐safety relationship in 1081 patients with breast or lung cancer.

Exposure‐efficacy analysis suggested a clear trend of improved OS with higher PK exposure. CPH model analysis further suggested Dato‐DXd AUC1 as a significant covariate for OS hazard, indicating that higher Dato‐DXd exposure could be associated with longer survival (Figure 1). Please note the analysis was conducted at one dose level (6 mg/kg) in TB‐01 study, therefore the results should be interpreted with caution for potential confounding, even though none of other covariates was identified with significant association with OS. Positive exposure‐efficacy relationship with Dato‐DXd AUC1 was also observed for PFS. Patients in the fourth quartile (Q4) of Dato‐DXd AUC1 are estimated to have 98‐day longer PFS compared to those in the first quartile (Q1). CPH analysis suggested baseline tumor size was more significantly associated with PFS, and Dato‐DXd AUC1 was removed in the backward elimination (p = .0011). Further evaluation of TROPION‐Breast01 was conducted to explore the potential association between Dato‐DXd exposure and disease related covariates. Patients in the lowest exposure quartile tend to have larger baseline tumor size as well as lower baseline albumin values (Table S5, Supporting Information), suggesting exposure and these covariates might be confounded in the exposure‐response analysis. ²¹ , ²²

A limited number of adjudicated drug‐related ILD cases were observed, with 57 cases out of 1081 patients overall, and 12 out of 352 patients in the TROPION‐Breast01 analysis population. The incidence rate was not higher at 6 mg/kg compared to 4 mg/kg (5% vs 10%, respectively, Table 2). No exposure‐response relationship was observed between Dato‐DXd or DXd PK exposure and adjudicated drug‐related ILD. These findings indicate that the Dato‐DXd dose of 6 mg/kg is acceptable for managing adjudicated drug‐related ILD.

Exposure‐safety relationships were observed between Dato‐DXd or DXd exposure and eight endpoints: Grade 3+ TEAEs, SAEs, TEAEs leading to dose interruption, TEAEs leading to dose reduction, any grade or grade 2+ stomatitis, and any grade or grade 2+ OSE. The increase in incidence of safety events, including stomatitis and OSE, were higher between 6 to 8 mg/kg than that of 4 to 6 mg/kg (Figure S5, Supporting Information). Predicted probabilities of TEAE leading dose reduction increased 7% from 4 to 6 mg/kg, compared to 14% from 6 to 8 mg/kg. No exposure‐safety relationships were identified for TEAE leading to treatment discontinuation.

The TEAE leading to treatment discontinuation rate was not higher at 6 mg/kg than 4 mg/kg (8% vs 16%, respectively, Table 2). The rate of TEAE leading to treatment discontinuation in the TROPION‐Breast01 was low (2.5%, respectively). The pooled exposure‐safety analysis of NSCLC and BC showed that tumor type was not a significant covariate for most safety endpoints. Higher risk of any grade OSE in patients with breast cancer compared to lung cancer may be associated with frequent ophthalmologic assessments that were mandated throughout the TROPION‐Breast01 study per regulatory requirement. ⁷ Based on a totality of efficacy and safety data, a lower dose of 4 mg/kg may only provide a small amount of safety/tolerability benefit.

With a capped dose of 540 mg in patients with body weight ≥90 kg, PopPK predictions showed that Dato‐DXd and DXd AUC1 increased with increasing body weight up to 90 kg and then decreased as expected in higher body weight patients. E–R analysis suggested that a dose cap at ≥90 kg may mitigate the risk of safety profiles, including OSE, in higher body weight patients.

In conclusion, the results of these analyses demonstrate the benefit‐risk profile of Dato‐DXd in patients with HR+/HER2‐ BC, with a dosage cap of 540 mg for patients weighing ≥90 kg; this corresponds to the dose regimen recommended in the recently approved prescribing information.

Author Contributions

Zoey Tang and Diansong Zhou analyzed the data and wrote the manuscript. KyoungSoo Lim, Yu Jiang, Neelima Denduluri, Nana Rokutanda, Song Ren, Pavan Vajjah, Yuzhuo Pan, Ying Hong, and Diansong Zhou reviewed and edited the manuscript. Zoey Tang, KyoungSoo Lim, Yu Jiang, Neelima Denduluri, Nana Rokutanda, Song Ren, Pavan Vajjah, Yuzhuo Pan, Ying Hong, and Diansong Zhou performed the research.

Funding

This study was sponsored by AstraZeneca. In July 2020, AstraZeneca entered into a global development and commercialization collaboration with Daiichi Sankyo for datopotamab deruxtecan (Dato‐DXd).

Conflicts of Interest

Yu Jiang, Neelima Denduluri, Nana Rokutanda, Song Ren, Pavan Vajjah, and Diansong Zhou are current employees of and shareholders in AstraZeneca. Zoey Tang and KyoungSoo Lim were previously employed by AstraZeneca, and the work presented in this article was conducted during the tenure at the company. Yuzhuo Pan and Ying Hong are current employees of Daiichi Sankyo and own stock in Daiichi Sankyo.

Supporting information

Supporting information