Correlation between tumor growth rate and survival in patients with metastatic breast cancer treated with trastuzumab deruxtecan

Abstract

Background: Previous studies in multiple metastatic tumors treated with diverse anticancer agents including immunotherapy, chemotherapy, mAb, and TKIs have suggested the rate of tumor growth (g-score) is inversely associated with survival.

Methods: We performed a retrospective analysis of patients with metastatic breast cancer (mBC) treated with trastuzumab deruxtecan (T-DXd), ado-trastuzumab emtansine (T-DM1), or chemotherapy to investigate the impact of those therapies on g-score and explore the association of g-score with clinical outcomes. This is the first report assessing g-score in tumors treated with an ADC.

Results: We investigated the association of g-score with progression-free (PFS) and overall survival (OS) in 2 phase 3 studies in patients with HER2 + mBC (DESTINY-Breast03 (DB-03)) and HER2-low mBC (DESTINY-Breast04 (DB-04)). After grouping patients according to quartiles of g-scores, we explored the association between g-score and PFS/OS using Kaplan-Meier plots and Cox regression models. The median g-score was higher for T-DM1, suggesting a faster growth rate at 0.0009/day vs that for T-DXd at 0.0002/day (P < .0001). Additionally, with data collection stopped at the time of database lock, 23% and 48% of tumors demonstrated only regression without growth in the T-DM1 and T-DXd arms, respectively. In DB-04, median g was 0.0018/day and 0.0006/day (P < .0001); with 17% and 32% of tumors demonstrating only regression with treatment of physician's choice (TPC) and T-DXd, respectively.

Conclusions: Compared to T-DM1 and TPC therapies, T-DXd significantly reduced the rate of tumor growth in the overall population and across subgroups. In both studies, the tumor growth rate was inversely associated with PFS and OS. In addition, it showed improved concordance with survival compared to ORR. The use of tumor growth rate as an intermediate endpoint may potentially accelerate drug development and reduce a patient's exposure to agents with limited or no activity.

Keywords: biomarkers; intermediate endpoint; metastatic breast cancer; tumor growth rate.

Figure 1.

Illustration of g-score modeling from Stein et al (2008).

Figure 2.

T‐DXd showed significant superiority in reducing tumor growth (g) in both DB-03 (a) and DB-04 (b) studies. C: Capecitabine. E: Eribulin. G: Gemcitabine. NP: Nab-Paclitaxel. P: Paclitaxel.

Figure 3.

g is highly associated with OS in DB-03 (top row) and DB-04 (bottom row) studies. No growth: dx subjects; Q1‐Q4: quartiles of g are calculated for subjects pooled from 2 treatment groups. In each quartile group, the median g is calculated: quartile (median g).

Figure 4.

g has improved concordance with survival compared to tumor response in DB-03 (top row) and DB-04 (bottom row) studies among subjects with tumor growth ($g>0$). The concordance is calculated based on Cox regression models.

Figure 5.

T‐DXd’s improvement in g‐score can be demonstrated with as few as first 60 subjects, provided the 60th subject had been followed for 14 weeks in DB-03 (top) and DB-04 (bottom) studies. In DB-03 study (1:1 randomization ratio), the analyses were based on first randomized 40, 60, and 80 subjects when the 40th, 60th, and 80th subject followed for 14 weeks respectively; in DB-04 study (2:1 randomization ratio), the analyses were based on first randomized 60, 90, and 120 subjects.