Highlight
– STELLAR-303, a global randomised phase 3 trial, reports a statistically significant overall survival (OS) benefit for zanzalintinib plus atezolizumab versus regorafenib in previously treated metastatic colorectal cancer that is not MSI-H/dMMR (ITT HR 0.80; p=0.0045).
– Median OS was 10.9 months with zanzalintinib–atezolizumab versus 9.4 months with regorafenib. The regimen is chemotherapy-free but associated with more grade ≥3 treatment-related adverse events (60% vs 37%).
Background and disease burden
Metastatic colorectal cancer (mCRC) remains a major cause of cancer mortality worldwide. Standard lines of systemic therapy for patients with microsatellite-stable (MSS) and mismatch repair–proficient (pMMR) tumours include cytotoxic chemotherapy, anti-VEGF and anti-EGFR agents where appropriate, followed by later-line options such as regorafenib and trifluridine–tipiracil. Immune checkpoint inhibitors (ICIs) have produced durable responses for the subset of tumours with high microsatellite instability (MSI-H) or deficient mismatch repair (dMMR), but responses in the much larger MSS/pMMR population have been limited. There is therefore an urgent unmet need for effective and tolerable therapies in heavily pretreated MSS mCRC.
Study design and methods
STELLAR-303 is a global, open-label, randomised phase 3 trial (121 centres, 16 countries) comparing the combination of zanzalintinib (a multitargeted tyrosine kinase inhibitor) plus atezolizumab (anti–PD-L1 antibody) against regorafenib monotherapy in patients aged ≥18 years with metastatic adenocarcinoma of the colon or rectum, previously treated with standard-of-care therapies and without MSI-H/dMMR. Patients were randomised 1:1 (blocks of four) to zanzalintinib 100 mg orally daily plus atezolizumab 1200 mg IV every 3 weeks, versus regorafenib 160 mg orally days 1–21 of each 28-day cycle. Stratification factors included geographic region, RAS mutation status, and presence of liver metastases.
Dual primary endpoints were overall survival in the intention-to-treat (ITT) population and overall survival in the prespecified subset without liver metastases. Safety was assessed in all patients who received at least one dose of study drug. This report corresponds to a planned overall survival analysis with data cutoff April 30, 2025. The trial remains active for the final OS analysis in the subset without liver metastases. ClinicalTrials.gov identifier: NCT05425940. The trial was funded by Exelixis.
Key findings
Patient population and follow-up: 901 patients were randomised between Sept 7, 2022, and July 15, 2024 (n=451 to zanzalintinib–atezolizumab; n=450 to regorafenib). Median follow-up was 18.0 months (IQR 14.6–21.5). Baseline characteristics were balanced across arms (59% male; race distribution: 54% White, 38% Asian; prior treatments consistent with standard care for refractory disease).
Primary efficacy — Overall survival (ITT)
At the planned OS analysis, the trial met its primary ITT endpoint. Zanzalintinib plus atezolizumab demonstrated a statistically significant OS improvement versus regorafenib with a stratified hazard ratio (HR) of 0.80 (95% CI 0.69–0.93; p=0.0045). Median OS was 10.9 months (95% CI 9.9–12.1) for the combination versus 9.4 months (95% CI 8.5–10.2) for regorafenib. The absolute median OS gain was 1.5 months, and the HR indicates a 20% reduction in the hazard of death in the combination arm during the observation period.
Subset analysis — patients without liver metastases
At the interim analysis for the prespecified subgroup without liver metastases, the stratified HR was 0.79 (95% CI 0.61–1.03; p=0.087). Median OS in this subset was 15.9 months (95% CI 13.5–17.6) with zanzalintinib–atezolizumab versus 12.7 months (95% CI 10.9–15.5) with regorafenib. The difference did not reach statistical significance at the interim time point; the trial will continue to final analysis in this subgroup.
Safety
Safety and tolerability were important differentiators. Grade 3 or worse treatment-related adverse events (TRAEs) occurred in 268 of 446 patients (60%) receiving zanzalintinib–atezolizumab versus 161 of 434 patients (37%) receiving regorafenib. There were five treatment-related deaths (≈1%) in the combination arm and one (<1%) in the regorafenib arm. The combination therefore carries a higher serious-toxicity burden despite being chemotherapy-free. Specific adverse-event profiles were consistent with expectations for TKI plus checkpoint inhibitor regimens (eg, hypertension, hepatotoxicity, immune-related events), although the detailed breakdown of AEs by type and grade is in the primary manuscript.
Context and interpretation
STELLAR-303 is, to our knowledge, the first phase 3 trial to demonstrate an OS benefit for an immunotherapy-based regimen in MSS/pMMR relapsed or refractory mCRC. Historically, ICIs have produced meaningful clinical benefit primarily in MSI-H/dMMR tumours; for MSS disease, response rates to PD-1/PD-L1 blockade have been low. Previous later-line systemic options for MSS mCRC that improved OS include regorafenib and trifluridine–tipiracil. Regorafenib’s activity in the refractory setting was established in the CORRECT phase 3 trial, and regorafenib has been a standard salvage option since then.
The biological rationale for combining anti-angiogenic or multitarget TKIs with checkpoint inhibition is plausible: TKIs can modulate the tumour microenvironment, reduce immunosuppressive myeloid populations, normalize abnormal vasculature, and increase immune cell infiltration, potentially converting an immunologically “cold” MSS tumour into a “hotter” one more responsive to checkpoint blockade. STELLAR-303 provides large-scale clinical evidence that a TKI plus anti–PD-L1 antibody can translate to an OS benefit in this difficult-to-treat population.
Strengths
- Large, multicentre, randomised phase 3 design with global participation and prespecified stratification variables relevant to prognosis (RAS status, liver metastases).
- Dual primary endpoints focused on both the overall population and a clinically relevant subset without liver metastases.
- Clinically meaningful endpoint (OS) met in the ITT population, a high bar in refractory mCRC studies.
Limitations and considerations
- Open-label design: while OS is an objective endpoint, open-label trials can introduce bias in subsequent therapies and supportive care decisions.
- Toxicity: the combination increased the rate of grade ≥3 TRAEs and had more treatment-related deaths. Patient selection, toxicity management pathways, and dose-modification strategies will be critical if the regimen is adopted.
- Magnitude of benefit: median OS improved by 1.5 months in the ITT population. Clinicians and patients must weigh this incremental benefit against toxicity and quality-of-life implications, which require detailed reporting from the trial.
- Subset analyses: the non-liver-metastases subgroup showed a numerically larger benefit but did not meet statistical significance at interim analysis. Further data are needed to confirm whether certain subgroups derive greater benefit.
- Biomarkers: the trial stimulates the need for predictive biomarkers beyond MSI status (eg, tumour immune infiltration signatures, angiogenesis markers, circulating tumour DNA) to identify patients most likely to benefit and least likely to be harmed.
- Industry sponsorship: Exelixis funded the trial. While industry sponsorship is common and not inherently problematic, independent confirmation and real-world experience will be important for broad adoption.
Clinical implications and next steps
For clinicians treating heavily pretreated MSS mCRC, zanzalintinib plus atezolizumab may represent a new chemotherapy-free option supported by a statistically significant OS improvement vs regorafenib. However, integration into practice should be informed by individual patient factors (performance status, comorbidities, prior toxicities), close attention to toxicity management, and shared decision-making about the balance of modest median OS gain and increased serious adverse events.
Key next steps include:
- Detailed reporting of safety management algorithms, quality-of-life data, and subsequent therapies received after progression in each arm.
- Final OS analysis in the no-liver-metastases subgroup, and exploration of other clinically meaningful subgroups (eg, RAS/BRAF status, PD-L1 expression, tumour mutational burden, ctDNA dynamics).
- Biomarker-driven translational studies to identify predictors of benefit and mechanisms of resistance.
- Post-approval real-world evidence to evaluate tolerability, sequencing with other agents, and comparative effectiveness in diverse populations.
Expert commentary
Experts will likely view STELLAR-303 as a landmark study because it addresses a long-standing negative space for immunotherapy in MSS mCRC. The demonstration of an OS benefit is compelling; nevertheless, the toxicity signal warrants caution. Oncologists experienced with TKI–ICI combinations (eg, in renal cell carcinoma) know that careful monitoring and proactive management can mitigate many severe events, but the higher rate of treatment-related deaths underscores that this is not a low-risk option. When choosing between regorafenib and zanzalintinib–atezolizumab, multidisciplinary input and clear patient counseling will be crucial.
Conclusion
STELLAR-303 shows that zanzalintinib plus atezolizumab improves overall survival compared with regorafenib in patients with previously treated MSS/pMMR metastatic colorectal cancer, marking the first phase 3 demonstration of an immunotherapy-based OS benefit in this population. The combination is a promising chemotherapy-free option, but increased serious toxicity and treatment-related mortality require careful patient selection, robust toxicity-management strategies, and continued investigation into predictive biomarkers. Pending full subgroup analyses, quality-of-life data, and independent corroboration, this regimen may change the therapeutic landscape for a subset of heavily pretreated patients with mCRC.
Funding and trial registration
Funding: Exelixis. ClinicalTrials.gov: NCT05425940.
Key references
Hecht JR, Park YS, Tabernero J, et al.; STELLAR-303 study investigators. Zanzalintinib plus atezolizumab versus regorafenib in refractory colorectal cancer (STELLAR-303): a randomised, open-label, phase 3 trial. Lancet. 2025 Oct 20:S0140-6736(25)02025-2. doi:10.1016/S0140-6736(25)02025-2.
Grothey A, van Cutsem E, Sobrero A, et al. Regorafenib monotherapy for previously treated metastatic colorectal cancer (CORRECT): a randomised, placebo-controlled, phase 3 trial. Lancet. 2013;381(9863):303–312.
Le DT, Durham JN, Smith KN, et al. PD-1 blockade in tumours with mismatch-repair deficiency. N Engl J Med. 2015;372:2509–2520.
Thumbnail prompt (AI image generator)
An oncologist in a modern hospital office looking at a tablet displaying survival curves and molecular diagrams; on the desk, pills and an infusion bag subtly labeled; cool blue and neutral tones conveying clinical seriousness; background shows a diverse clinical trial team conferring; soft light, high realism.