Why This Study Matters
CD47 is a checkpoint protein that lets cancer cells evade destruction by macrophages, and its overexpression is a negative prognostic factor across multiple tumor types, including HER2-positive gastric and gastroesophageal junction (GEJ) cancer. For patients whose disease progresses after first-line trastuzumab-based therapy, later-line options have historically been limited — typically ramucirumab plus paclitaxel or trastuzumab deruxtecan — leaving a persistent unmet need for regimens that can restore or enhance anti-HER2 antibody activity later in the treatment course.
Evorpacept is an engineered fusion protein that blocks the CD47–signal regulatory protein α (SIRPα) interaction, designed to enhance antibody-dependent cellular phagocytosis when paired with a tumor-targeting antibody such as trastuzumab. ASPEN-06 tested whether adding evorpacept to a standard later-line regimen could meaningfully improve outcomes.
Study Design
ASPEN-06 is a multicenter, open-label, randomized phase 2/3 study; this report covers the phase 2 portion, conducted at 49 academic and community healthcare institutions across 10 countries. Screening ran from February 28, 2022 through January 22, 2024, with a protocol-defined primary analysis data cutoff of May 24, 2024, and an updated, post hoc biomarker analysis data cutoff of May 15, 2025.
| Arm | Regimen (n) |
|---|---|
| Evorpacept + TRP | Evorpacept 30 mg/kg + trastuzumab (6 mg/kg initial, 4 mg/kg thereafter) + ramucirumab 8 mg/kg, each every 2 weeks, plus paclitaxel 80 mg/m² weekly for 3 of every 4 weeks (n = 63) |
| TRP alone | Trastuzumab + ramucirumab + paclitaxel at the same doses and schedule, without evorpacept (n = 64) |
The primary endpoint was investigator-assessed objective response rate (ORR), evaluated against two prespecified criteria: (1) superiority versus an assumed historical control ORR of 30% (from ramucirumab plus paclitaxel), tested at a one-sided α of 0.025, and (2) an ORR difference of at least 8.0 percentage points (ITT population) or 9.7 percentage points (fresh-biopsy HER2-positive subgroup) versus the internal TRP control arm.
Patient Population
Of 236 patients screened, 127 were randomized. All had pretreated, HER2-overexpressing (IHC3+ or IHC2+/ISH+) advanced or metastatic gastric/GEJ adenocarcinoma that had progressed on or after a previous HER2-directed agent and/or fluoropyrimidine/platinum-containing chemotherapy. At the primary analysis data cutoff, median follow-up was 7.8 months. Baseline characteristics were balanced between arms: median age was 64 years (evorpacept + TRP) versus 63 years (TRP alone), 81.1% of patients overall were male, and most were Asian (48.8%) or white (29.9%). A subset of 47 patients (37.0%) had a fresh HER2-positive biopsy at baseline, and in the later biomarker analysis, 95 patients (74.8%) were identified as having retained HER2-positive disease by fresh biopsy or circulating tumor DNA (ctDNA).
Primary Endpoint Results
Because the trial had two co-primary objectives — besting the internal TRP control by a set margin, and clearing the historical-benchmark bar with statistical significance — the result is genuinely mixed: the combination showed a clear, clinically meaningful contribution over TRP alone, but that was not sufficient on its own to declare the study a formal statistical success against history.
| Evo + TRP, ITT (n=63) | TRP alone, ITT (n=64) | Evo + TRP, fresh biopsy (n=22) | TRP alone, fresh biopsy (n=26) | |
|---|---|---|---|---|
| Confirmed ORR (95% CI) | 40.3% (28.1–49.3) | 26.6% (16.3–39.1) | 54.8% (32.3–66.3) | 23.1% (9.0–43.6) |
| Median DOR, months (95% CI) | 15.7 (11.0–NR) | 7.6 (6.3–NR) | 15.7 (3.6–NR) | 7.6 (7.4–NR) |
| Median PFS, months | 7.5 | 7.4 | 9.0 | 7.4 |
Overall survival was not mature at the primary analysis. In the updated analysis, OS was similar between arms: hazard ratio (HR) 1.07 (95% CI, 0.69–1.66) in the ITT population and HR 1.05 (95% CI, 0.46–2.38) in the fresh-biopsy HER2-positive subgroup.
Subgroup Analyses
A post hoc biomarker analysis (data cutoff May 15, 2025) examined outcomes in patients with "retained" HER2-positive disease, defined by a fresh tumor biopsy or by ERBB2 gene amplification detected in ctDNA:
| Subgroup | n | Evo + TRP ORR | TRP alone ORR |
|---|---|---|---|
| Fresh-biopsy HER2-positive subgroup | 47 | 59.1% | 24.0% |
| ctDNA HER2-positive subgroup | 86 | 48.8% | 25.6% |
| Fresh biopsy or ctDNA HER2-positive subgroup | 95 | 48.9% | 25.0% |
In all three of these HER2-retained subgroups, the 95% confidence interval for the evorpacept + TRP ORR excluded the 30% historical benchmark. In the combined fresh biopsy or ctDNA HER2-positive subgroup, median duration of response was 15.7 versus 9.1 months (HR 0.59, 95% CI 0.24–1.45), the PFS hazard ratio numerically favored evorpacept + TRP (HR 0.72, 95% CI 0.44–1.18), and the OS hazard ratio was 0.95 (95% CI 0.58–1.56). A gender-based analysis of the primary endpoint found ORRs of 36.4% versus 27.1% in male patients (evo + TRP, n=55, vs. TRP, n=48) and 50.0% versus 25.0% in female patients (evo + TRP, n=8, vs. TRP, n=16).
Safety Profile
Treatment-emergent adverse events (TEAEs) of any grade occurred in 100% of patients in both arms. Hematologic toxicities were more common with the addition of evorpacept, though the study authors characterized overall safety as similar between arms.
| Adverse Event | Evo + TRP (any grade) | TRP alone (any grade) | Evo + TRP (Grade ≥3) | TRP alone (Grade ≥3) |
|---|---|---|---|---|
| Neutropenia | 69.8% | 55.6% | 55.6% | 39.7% |
| Anemia | 58.7% | 38.1% | 23.8% | 17.5% |
| Diarrhea | 41.3% | 36.5% | Not specified in source | Not specified in source |
| Leukopenia | Not specified in source | Not specified in source | 17.5% | 9.5% |
| Febrile neutropenia (any grade) | 3.2% | 7.9% | — | — |
Grade 5 (fatal) TEAEs occurred in 12 patients overall (5 in the evorpacept + TRP arm, 7 in the TRP-alone arm). Three grade 5 events (2.4%) were assessed as treatment-related: two in the evorpacept + TRP arm (an esophageal perforation possibly related to ramucirumab, and a case of renal insufficiency, acute respiratory insufficiency and anasarca possibly related to ramucirumab and trastuzumab) and one in the TRP-alone arm (grade 5 pneumonia related to TRP).
Interpretation and Broader Context
In summary, evo + TRP showed encouraging efficacy with manageable safety in advanced gastric/gastroesophageal junction cancer.
— Verbatim from the study abstract (Shitara et al., Nature Medicine, 2026)
Evorpacept added to trastuzumab, ramucirumab and paclitaxel produced numerically higher response rates and more than double the median duration of response compared with TRP alone, meeting the prespecified comparison against the internal control arm. However, because the trial did not meet its co-primary statistical objective versus the historical 30% benchmark, these phase 2 results alone do not establish evorpacept + TRP as a new standard of care. Post hoc biomarker analyses suggest that patients with retained HER2-positive disease — identified by fresh biopsy or ctDNA — may derive more pronounced benefit, a signal that could help refine patient selection going forward.
Future Directions
The study authors report that ASPEN-06 did not proceed to its phase 3 portion, for strategic reasons. They state that further investigation of this signal is warranted in larger controlled studies with prospective patient selection based on CD47-high expression and retained HER2-positive disease, in advanced gastric/GEJ cancer and other indications including HER2-positive breast cancer. Given the emergence of trastuzumab deruxtecan as second-line therapy in advanced gastric/GEJ cancer, the authors note that future development of evorpacept may be more relevant in earlier-line settings or in combination with next-generation Fc-competent HER2-directed antibodies such as zanidatamab.
Limitations
This is the phase 2 portion of an open-label study, and the sponsor did not advance ASPEN-06 to its phase 3 portion, so no confirmatory randomized trial data exist yet. Neither the ITT population nor the fresh-biopsy HER2-positive subgroup met the prespecified statistical significance threshold (one-sided α = 0.025) against the historical 30% ORR benchmark, so the comparison to that benchmark — drawn from a different, earlier trial rather than a randomized arm within this study — does not establish a statistically confirmed benefit. Overall survival was immature at the primary analysis, and in the updated analysis OS hazard ratios were close to or slightly above 1.0 in both the ITT population and the fresh-biopsy subgroup, showing no survival benefit signal to date. The biomarker subgroup analyses (fresh biopsy, ctDNA and gender-based ORR) were post hoc and exploratory, run on small sample sizes — as few as 8 to 26 patients per arm in some comparisons — and were not powered for statistical significance.
In the phase 2 portion of ASPEN-06, adding evorpacept to trastuzumab, ramucirumab and paclitaxel nearly doubled the objective response rate in previously treated, HER2-positive gastric and gastroesophageal junction cancer, exceeding the prespecified margin against the internal TRP control arm. But because the study did not clear the statistical bar against its other co-primary objective — a historical 30% ORR benchmark — and overall survival showed no benefit signal in an immature, updated analysis, these results support further investigation rather than an immediate change in standard of care; ASPEN-06 did not proceed to a phase 3 portion.