Summary
In a randomized phase 2 trial, evorpacept plus trastuzumab, ramucirumab and paclitaxel produced higher response rates than the same regimen without evorpacept. The trial met its prespecified response-difference threshold between treatment arms, but not its statistical target against a historical benchmark.
A randomized phase 2 trial found higher tumour response rates when evorpacept was added to trastuzumab, ramucirumab and paclitaxel for pretreated HER2-positive advanced gastric or gastroesophageal junction cancer. The combination met the study’s prespecified target for the response-rate difference between treatment arms, but it did not meet the separate statistical criterion comparing its response rate with a historical benchmark. The results were published in Nature Medicine on 24 September 2026.
How the trial tested the combination
The open-label phase 2 portion of ASPEN-06 randomized 127 patients whose cancer had progressed during or after previous treatment, including HER2-directed therapy. Sixty-three received evorpacept plus trastuzumab, ramucirumab and paclitaxel (TRP); 64 received TRP alone. The main outcome was the proportion of patients whose tumours shrank, assessed by investigators.
Evorpacept is designed to block the interaction between CD47 on a cell and SIRPα on macrophages. That interaction can act as a “don’t engulf” signal, limiting the removal of cells by phagocytosis. Evorpacept’s engineered, inactive Fc domain is intended to reduce that signal; trastuzumab, which binds HER2, supplies an active Fc signal that can promote macrophage phagocytosis. The combination was tested on the premise that blocking the inhibitory signal could strengthen this antibody-driven response.
In the full trial population, the investigator-assessed response rate was 40.3% with evorpacept plus TRP and 26.6% with TRP alone, a difference of 13.7 percentage points. Among patients whose HER2-positive status was confirmed on a biopsy taken after earlier HER2-directed treatment, the rates were 54.8% and 23.1%, respectively, a difference of 31.7 points. Both differences exceeded the study’s prespecified thresholds for comparing the randomized arms: 8 points in the full population and 9.7 points in the biopsy subgroup.
The trial also tested whether the combination’s response rate exceeded a 30% historical benchmark for ramucirumab and paclitaxel. It did not meet the required statistical criterion in either population: the one-sided P values were 0.0949 for the full population and 0.030 for the fresh-biopsy subgroup, against a prespecified significance threshold of 0.025. Thus, one of the two primary objectives was met, while the historical-benchmark objective was not.
Biomarker results and safety
An exploratory, post hoc analysis used data collected through 15 May 2025 to examine patients with evidence of retained HER2-positive disease, defined by a positive fresh biopsy or ERBB2 gene amplification detected in circulating tumour DNA. Within this group, researchers identified a candidate “CD47-high” category: at least 5% of tumour cells showing the highest membrane-staining intensity in an immunohistochemistry test.
Among 48 patients with retained HER2-positive disease and CD47-high tumours, the response rate was 63.6% with evorpacept plus TRP and 23.1% with TRP alone. Median progression-free survival was 19.5 months versus 7.0 months; the hazard ratio was 0.38 (95% confidence interval 0.17–0.84). In the CD47-low subgroup, the response rates were 36.4% and 30.0%. Because this biomarker analysis was conducted after the main analysis, involved small subgroups and derived its candidate threshold from the same dataset, the findings need prospective confirmation before the marker can guide treatment selection.
Adverse events were common in both groups. Grade 3 or higher treatment-emergent events occurred in 90.5% of patients receiving evorpacept plus TRP and 79.4% receiving TRP alone; neutropenia and anaemia were among the more frequent severe events and were more common with the combination. Three grade 5 treatment-related adverse events were reported: two in the combination group and one in the control group. The researchers described overall safety as similar between arms while noting the greater frequency of haematologic toxicities with evorpacept.
ASPEN-06 did not proceed to phase 3 for strategic reasons, according to the paper. The researchers say further controlled studies with prospective selection for retained HER2 and high CD47 expression would be needed to test whether the subgroup signal is reproducible.