Endometrial cancer (EC) is the most common gynecologic malignancy worldwide and the leading cause of gynecologic cancer-related death in the United States. Unlike many other cancers, EC survival rates have declined over the past four decades, highlighting an urgent need for better treatment strategies (1). The Cancer Genome Atlas (TCGA) has classified EC into four molecular subtypes: POLE-ultramutated, microsatellite instability-high (MSI-H)/mismatch repair deficiency (dMMR), copy-number low (p53 wild-type), and copy-number high (p53 mutant) (2). This classification has refined risk stratification facilitated molecularly tailored therapies, and guided clinical management (3). While platinum-based chemotherapy has long been the standard for advanced EC, immune checkpoint inhibitors (ICIs) have transformed treatment paradigms, particularly for MSI-H tumors. ICIs have shown high efficacy in MSI-H but limited activity in MSS/pMMR tumors (4–6). To address this limitation, combined chemo-immunotherapy strategies have been studied. The RUBY and NRG-GY018 trials demonstrated significantly improved PFS, leading to the adoption of chemo-immunotherapy as the preferred first-line approach for advanced EC (7, 8). Nevertheless, disease progression remains a significant challenge in pMMR tumors, highlighting the need for more effective second-line therapies (9, 10). Preclinical studies have shown that combining lenvatinib, a multikinase inhibitor, and pembrolizumab, a PD-1 inhibitor, yields synergistic antitumor activity beyond either agent alone (11). The safety and effectiveness of the combination were first assessed in Study 111/KEYNOTE-146 and later validated in the phase III KEYNOTE-775 trial, which demonstrated meaningful improvements in PFS and OS over chemotherapy. This led to FDA approval of pembrolizumab (200 mg every 3 weeks) in combination with lenvatinib (20 mg daily) for previously treated advanced EC (12, 13). However, toxicity remains a major barrier to sustained treatment. Frequently reported grade ≥3 adverse events (AEs) included diarrhea, hypertension (HTN), musculoskeletal disorders, fatigue, nausea, decreased appetite, vomiting, stomatitis, weight loss, proteinuria, dyspnea, cough, and rash (13, 14). In Study 111/KEYNOTE-146, approximately 66% of patients experienced grade ≥3 AEs; this increased to 88.9% in KEYNOTE-775, compared to 72.7% seen with chemotherapy. These toxicities led to dose reductions in two-thirds of patients, treatment interruptions in over 70%, and discontinuation in one-third of patients—primarily due to lenvatinib (12, 13). Although the recommended starting dose is 20 mg daily, the actual average administered doses were lower: 14.4 mg/day in Study 111/KEYNOTE-146 and 13.8 mg/day in KEYNOTE-775. Only 8.9% of patients were able to maintain the full dose for ≥ 6 months (12, 13), emphasizing the need for close monitoring and individualized dose adjustments (14, 15). Recognizing the high incidence of treatment-related toxicity in clinical trials, and the broader range of patients encountered in real-world settings, many clinicians have adopted lower starting doses of lenvatinib. To better understand the impact of this approach, we conducted a retrospective cohort study to assess the efficacy and tolerability of initiating lenvatinib at ≤10 mg versus >10 mg.

The KEYNOTE-775 trial established lenvatinib and pembrolizumab as the standard of care for patients with pMMR advanced recurrent EC (12, 13). However, the recommended 20 mg daily dose of lenvatinib presents significant toxicity challenges in clinical practice (12, 14, 17), often requiring empirical dose reductions to improve patient tolerability. This concern is further illustrated by a Korean multicenter retrospective study, in which 83% of patients initiated lenvatinib at 20 mg but 56.2% required subsequent dose reductions due to toxicity, highlighting widespread tolerability issues. Notably, ORR and PFS in that cohort were lower than those reported in KEYNOTE-775, potentially reflecting genetic and demographic differences between Korean and Western populations, and underscoring the influence of ethnicity on treatment outcomes (18). Given these challenges, our real-world retrospective study evaluated whether initiating lenvatinib at a reduced dose (≤10 mg) could improve tolerability without compromising efficacy. Compared to our cohort, which reflected broader clinical diversity and individualized, physician-guided dosing strategies, the KEYNOTE-775 trial enrolled a more homogeneous population—characterized by better baseline performance status, fewer aggressive histologic subtypes, and universal initiation at 20 mg lenvatinib. In our real-world cohort, 62% of patients started treatment at ≤10 mg and only 14.1% at 20 mg. Our cohort also included a higher proportion of patients with ECOG ≥2 and aggressive histologic subtypes, including carcinosarcoma. Although unadjusted comparisons showed no statistically significant differences in PFS or OS between dose groups, age-adjusted analysis revealed a significantly higher risk of progression or death with initial dosing at ≤10 mg (HR: 2.92; p = 0.008). Notably, only 34% of patients in the higher-dose group received the full 20 mg, with most starting at intermediate doses, suggesting that intermediate starting doses may better balance efficacy and tolerability. Our findings align with those of How et al. (19), who observed no significant differences in baseline characteristics or survival outcomes (PFS or OS) between patients who received a reduced starting dose—most commonly 14 mg—and those who received the recommended 20 mg dose. In contrast, Zammarrelli et al. (20), reported shorter PFS in patients initiating treatment at 20 mg. These contrasting findings may reflect variability in clinical practice patterns and patient populations. Notably, our cohort had a higher proportion of ECOG 0-1 patients (43.5% vs. 28% in Zammarrelli et al) which may have contributed to more favorable outcomes. Additionally, initiating lenvatinib at 20 mg with subsequent dose reductions could lead to higher rates of treatment interruptions, potentially limiting the sustained therapeutic synergy between pembrolizumab and lenvatinib. Dose optimization has been observed in other malignancies, such as hepatocellular carcinoma, where the FDA-approved lenvatinib dose ranges from 8 to 12 mg daily (21). Our study has several limitations, including its retrospective design, which carries the risk of incomplete data capture and selection bias, and the lack of MSI-H subgroup analysis due to the small sample size. To enhance statistical reliability and align with real-world prescribing patterns, we used a dichotomous dose classification (≤10 mg vs. >10 mg), which may have masked finer dose-response patterns.

In conclusion, this is the largest real-world evaluation of the lenvatinib–pembrolizumab combination in advanced EC to date, offering important insights into the impact of initial dosing strategies on patient outcomes. We observed a clear trend to initiate lenvatinib at a reduced dose, with only 14.1% of patients receiving the full 20 mg, regardless of ECOG performance status. Clinicians should carefully weigh the risks and benefits of various lenvatinib starting doses. Our findings suggest that intermediate starting doses may provide a more favorable balance between efficacy and tolerability. Prospective, controlled studies are needed to determine the optimal lenvatinib starting dose that balances efficacy and tolerability in patients with pMMR advanced EC.