Nature Medicine September 24, 2026

Evorpacept plus trastuzumab, ramucirumab and paclitaxel in HER2-positive gastric cancer: a randomized phase 2 trial

Kohei Shitara, Zev Wainberg, Josep Tabernero, Eric Van Cutsem, Clélia Coutzac, Christelle De La Fouchardière, Jeeyun Lee, Sun Young Rha, Yoon-Koo Kang, Philip Fanning, Alison Forgie, Cherry Mao, Daniel Brickman, Jaume Pons, Athanasios C. Tsiatis, Sophia Randolph, Keun-Wook Lee

Bottom Line

In a phase 2 trial of patients with pretreated HER2-overexpressing advanced gastric or gastroesophageal junction cancer, the addition of the CD47 inhibitor evorpacept to trastuzumab, ramucirumab, and paclitaxel improved objective response rates, particularly in patients with fresh biopsy-confirmed HER2 positivity, though it did not meet the prespecified statistical significance threshold compared to historical benchmarks.

Key Findings

1. In the intent-to-treat (ITT) population, the investigator-assessed objective response rate (ORR) was 40.3% for evorpacept plus TRP (n=63) compared to 26.6% for TRP alone (n=64), representing an absolute difference of 13.7%.
2. In the subgroup of patients with fresh biopsy-confirmed HER2 positivity, the ORR was 54.8% for the evorpacept arm (n=22) versus 23.1% for the control arm (n=26), an absolute difference of 31.7%.
3. Despite absolute improvements, the ORRs did not meet the prespecified statistical criterion (one-sided α = 0.025) when compared to the 30% historical benchmark for the ITT population (P = 0.0949, one-sided) or the fresh biopsy HER2+ subgroup (P = 0.030, one-sided).
4. While hematologic toxicities were more common with the addition of evorpacept, overall safety was considered similar and manageable between the two treatment arms.

Study Design

Design
Phase 2 RCT
Open-Label
Sample
127
Patients
Duration
Not reported
Median
Setting
Multicenter
Population Patients with pretreated, HER2-overexpressing, advanced gastric or gastroesophageal junction cancer
Intervention Evorpacept plus trastuzumab, ramucirumab and paclitaxel (evo + TRP)
Comparator Trastuzumab, ramucirumab and paclitaxel (TRP) alone
Outcome Investigator-assessed objective response rate (ORR)

Study Limitations

• The trial failed to meet its prespecified primary statistical objective against the historical 30% response rate benchmark (one-sided α = 0.025).
• Subgroup analyses, specifically the fresh biopsy HER2+ and high CD47 expression populations, were based on very small sample sizes (e.g., n=22 and n=26), limiting statistical power and broad generalizability.
• The open-label design introduces the potential for investigator bias in the assessment of the primary endpoint, investigator-assessed objective response rate.
• Biomarker evaluations correlating treatment efficacy with retained HER2 status and CD47 expression were conducted post hoc and require prospective validation.

Clinical Significance

This study provides clinical proof-of-concept for targeting the CD47-SIRPα 'don't eat me' signaling axis in solid tumors to enhance antibody-dependent cellular phagocytosis. Although the trial did not cross its stringent statistical boundaries against historical controls, the clinically meaningful increase in ORR to 54.8% in patients with confirmed retained HER2 expression suggests that combining innate immune checkpoint inhibitors with standard HER2-directed therapy and chemotherapy could offer a promising, biomarker-selected approach for patients with advanced gastric cancer who have progressed on prior therapies.

Historical Context

Published on September 24, 2026, this research addresses the limited options available for patients with advanced HER2-positive gastric and gastroesophageal junction cancers after progression on frontline anti-HER2 treatments. Historically, standard second-line therapy involving ramucirumab and paclitaxel yields an objective response rate of roughly 30%. Evorpacept is a novel, next-generation CD47-blocking fusion protein designed with an inactive Fc domain to avoid the severe red blood cell depletion seen in earlier CD47 inhibitors. The phase 2 ASPEN-06 trial sought to determine if neutralizing this innate immune evasion mechanism could safely resensitize tumors to HER2-directed antibodies and chemotherapy in the refractory setting.

Guided Discussion

High-yield insights from every perspective

Med Student
Medical Student

How does the mechanism of action of evorpacept, a CD47 inhibitor, theoretically complement the action of trastuzumab in HER2-overexpressing gastric cancer?

Key Response

CD47 acts as a 'do not eat me' signal frequently overexpressed on tumor cells to evade innate immunity. Evorpacept binds to CD47, blocking this signal. Trastuzumab, an anti-HER2 antibody, not only blocks HER2 signaling but also tags the tumor cell for antibody-dependent cellular phagocytosis (ADCP) by macrophages via its Fc domain. Combining the 'eat me' signal from trastuzumab with the blockade of the 'do not eat me' signal by evorpacept synergistically enhances macrophage-mediated tumor clearance.

Resident
Resident

The standard second-line therapy for advanced gastric cancer is ramucirumab plus paclitaxel. What is the clinical rationale for continuing trastuzumab beyond progression in this trial, and what unique toxicities must be monitored when adding a CD47 inhibitor to this regimen?

Key Response

Unlike in breast cancer, continuing HER2 blockade beyond progression in gastric cancer is traditionally debated due to lack of strong phase 3 data. This trial explores its continuation to synergize with CD47 inhibition. Residents must monitor standard toxicities of ramucirumab/paclitaxel (neuropathy, myelosuppression, bleeding, hypertension) alongside unique CD47 inhibitor toxicities. Although evorpacept has an inactive Fc domain designed to minimize RBC binding, CD47 inhibitors generally carry risks of anemia and thrombocytopenia due to the clearance of senescent red blood cells and platelets.

Fellow
Fellow

The study noted improved objective response rates specifically in patients with fresh biopsy-confirmed HER2 positivity. How does the spatial and temporal heterogeneity of HER2 expression in gastric cancer complicate second-line trial designs, and how should this affect your clinical practice?

Key Response

HER2 expression in gastric cancer is notoriously heterogeneous and is lost in approximately 30-40% of patients following first-line trastuzumab therapy. The reliance on fresh biopsies in this trial highlights that patients who retain HER2 positivity are the ones who benefit from continued HER2-targeted therapy. For a fellow, this underscores the critical importance of re-biopsying metastatic lesions at progression to confirm biomarker status before committing to biomarker-driven subsequent lines of therapy.

Attending
Attending

This trial demonstrated a clinically promising objective response rate in the biomarker-selected subgroup but failed to meet its prespecified statistical significance threshold against historical benchmarks. How do you integrate these 'biologically positive but statistically negative' findings into your teaching and future clinical trial enrollment strategies?

Key Response

Attendings must teach trainees to critically differentiate between a statistically negative trial (often due to study design, powering, or flawed historical benchmarks) and a biologically inactive drug. While this combination cannot be adopted into routine clinical practice due to the statistical failure, the strong signal in the fresh HER2-positive cohort provides a compelling teaching point on biomarker drift and justifies steering appropriate patients toward phase 3 randomized trials evaluating this combination in carefully selected, biopsy-proven HER2-retained populations.

Scholarly Review

Critical appraisal through the lens of expert reviewers and guideline development

PhD
PhD

The study utilized a comparison against historical benchmarks rather than a fully powered randomized control arm to determine statistical significance. What are the major methodological pitfalls of using historical controls in heavily pretreated gastric cancer trials, and how does this affect the interpretation of a four-drug regimen?

Key Response

Historical controls in oncology are vulnerable to stage migration, improvements in supportive care, and shifting baseline prognostic factors. Comparing a novel four-drug regimen (evorpacept + trastuzumab + ramucirumab + paclitaxel) to historical two-drug (ramucirumab + paclitaxel) benchmarks makes it impossible to isolate the independent treatment effect of evorpacept versus the effect of simply continuing trastuzumab. A rigorous phase 2/3 design must use a concurrent, randomized, placebo-controlled arm to control for these confounding variables.

Journal Editor
Journal Editor

As a peer reviewer, how would you evaluate the authors' emphasis on an 'improved objective response rate' in the abstract and discussion, given that the trial did not meet its prespecified primary statistical endpoint?

Key Response

A seasoned reviewer would flag this as potential 'spin.' If a trial fails its primary prespecified statistical threshold, it is technically a negative trial. The editorial challenge is to ensure the authors transparently report the primary endpoint failure prominently, while ensuring that secondary, exploratory, or subgroup analyses (such as the fresh HER2-positive cohort) are strictly framed as hypothesis-generating rather than definitive evidence of efficacy.

Guideline Committee
Guideline Committee

Considering this phase 2 trial did not meet its prespecified statistical threshold, what level of evidence does it provide, and does it warrant any updates to the NCCN or ESMO guidelines for second-line HER2-positive gastric cancer?

Key Response

This study provides lower-level evidence (Level 3/Category 2B at best) due to its failure to meet the primary statistical endpoint, meaning evorpacept should not be incorporated into current treatment algorithms. NCCN and ESMO currently recommend ramucirumab plus paclitaxel or fam-trastuzumab deruxtecan-nxki (T-DXd) for second-line HER2-positive gastric cancer. However, the strong signal in fresh-biopsy patients reinforces existing, but sometimes loosely applied, guideline recommendations to consider repeat biomarker testing upon progression to ensure accurate selection for targeted agents like T-DXd.

Clinical Landscape

Noteworthy Related Trials

2010

ToGA Trial

n = 594 · Lancet

Tested

Trastuzumab plus chemotherapy

Population

Patients with HER2-positive advanced gastric or gastro-oesophageal junction cancer

Comparator

Chemotherapy alone

Endpoint

Overall survival

Key result: Median overall survival was significantly improved with trastuzumab plus chemotherapy compared to chemotherapy alone (13.8 vs 11.1 months).
2014

RAINBOW Trial

n = 665 · Lancet Oncol

Tested

Ramucirumab plus paclitaxel

Population

Patients with advanced gastric or gastroesophageal junction adenocarcinoma progressing on first-line chemotherapy

Comparator

Placebo plus paclitaxel

Endpoint

Overall survival

Key result: The combination of ramucirumab and paclitaxel significantly increased overall survival compared to placebo and paclitaxel (9.6 vs 7.4 months).
2020

DESTINY-Gastric01

n = 188 · NEJM

Tested

Trastuzumab deruxtecan

Population

Patients with HER2-positive advanced gastric cancer who progressed on prior regimens including trastuzumab

Comparator

Physician's choice of chemotherapy

Endpoint

Objective response rate

Key result: Trastuzumab deruxtecan demonstrated a significantly higher objective response rate (51% vs 14%) and longer overall survival than standard chemotherapy.

Tailored to your role

Want this tailored to you?

Add your specialty or training stage to get role-specific takeaways and more questions.

Personalize this analysis
↑