Evorpacept

Evorpacept (ALX148) is ALX Oncology's CD47-blocking fusion protein, designed to flip off the 'don't eat me' signal that tumors use to hide from macrophages. The lead Phase 2 program tested here combines evorpacept with pembrolizumab and stereotactic body radiation therapy (SBRT) as neoadjuvant therapy in HPV+ oropharyngeal cancer (NCT05787639, n=29, UC San Diego); the primary endpoint is pathologic complete response (pCR) at resection [1]. The bigger commercial bet sits in HER2+ tumors - gastric, breast, salivary - where evorpacept added to trastuzumab-based regimens has shown the most consistent signal across the ASPEN franchise, anchored by positive ASPEN-06 randomized Phase 2 data in HER2+ gastric cancer (NCT05002127) [2][3].

Evorpacept is a novel, unapproved biologic. ALX Oncology has been running it across at least eight active studies spanning HPV+ head and neck (NCT05787639, n=29, active not recruiting) [4], HER2+ metastatic breast post-trastuzumab-deruxtecan (ASPEN-09-03, NCT07007559, single-arm Phase 2, n≈80, recruiting) [5], platinum-resistant ovarian (NCT05467670) [6], and B-cell non-Hodgkin lymphoma (ASPEN-01, completed and published) [7]. The key-readout-relevant program is ASPEN-06 (NCT05002127), a randomized Phase 2/3 trial in second/third-line HER2+ gastric/GEJ cancer. The Phase 2 portion enrolled 127 patients; ALX reported positive topline data in 2023 and updated results at ASCO GI 2025, showing objective response rate of 54.8% with evorpacept + trastuzumab + ramucirumab + paclitaxel vs. 23.1% with the same regimen minus evorpacept in the prespecified fresh HER2+ biopsy population - the first positive randomized data for any CD47 blocker [3][13]. No FDA breakthrough or fast track designation has been granted for evorpacept to date and no regulatory submission is filed. Financial picture is better than I initially read: as of Q1 2026 (10-Q reported May 2026), ALX held $169.1M cash, up from $48.3M at year-end 2025 following a $150M financing in Q1 2026, with reported runway into 1H 2028. R&D burn for Q1 2026 was $13.6M (down from $23.9M) reflecting prior workforce reductions and pipeline prioritization [14]. The HPV+ HNSCC neoadjuvant study is investigator-sponsored at UCSD; pCR readout will be reported via academic channels (likely AHNS or ESMO 2026) rather than driving an ALX label.

CD47 is a surface protein that nearly every cell in your body uses as an ID badge - it tells patrolling macrophages 'don't eat me, I'm one of you.' Tumors hijack this. They jack up CD47 expression to avoid being chewed up by innate immune cells. Evorpacept is a decoy: a high-affinity CD47-binding domain fused to an inert IgG1 Fc tail. It plugs into CD47 on tumor cells and blocks the handshake with SIRPα (the inhibitory receptor on macrophages), removing the 'don't eat me' brake. The inert Fc is deliberate - earlier CD47 antibodies like magrolimab carried active Fc regions that hit CD47 on red blood cells and caused anemia. Evorpacept's design minimizes that on-target hematologic toxicity, which is the cleanest part of its story [7]. Why HER2+ combinations specifically: trastuzumab kills tumor cells partly through antibody-dependent cellular phagocytosis (ADCP) - macrophages recognize trastuzumab-coated tumor cells and engulf them. CD47 blockade removes the inhibitory counterweight, so the two mechanisms are mechanistically synergistic: trastuzumab provides the 'eat me' signal via Fc engagement, evorpacept removes the 'don't eat me' brake. That's the biological rationale behind ASPEN-06 and ASPEN-09-03. The validation case is mixed. Genetically, CD47-SIRPα is well-characterized as a phagocytosis checkpoint. Clinically, Gilead paid $4.9B for Forty Seven to acquire magrolimab in 2020, then discontinued the entire program in 2024 after MDS and AML trials failed and FDA imposed partial holds for safety. That's the elephant in the room - the most advanced CD47 program in the industry blew up [11].

NCT05787639 is a single-arm Phase 2 investigator-sponsored study at UCSD enrolling 29 patients with HPV+ oropharyngeal squamous cell carcinoma (HNSCC). Design: neoadjuvant SBRT plus pembrolizumab plus evorpacept, then surgery. Primary endpoint is pCR at resection [4]. Status is active, not recruiting - enrollment is essentially complete. n=29 single-arm is a signal-finding study, not registrational. There's no comparator arm, so any pCR rate has to be benchmarked against historical controls and against the new perioperative pembrolizumab standard from KEYNOTE-689 (median event-free survival 51.8 vs. 30.4 months), which received FDA approval in June 2025 for resectable PD-L1+ HNSCC [15]. The commercially relevant evorpacept trial designs live in ASPEN-06 (HER2+ gastric, NCT05002127, randomized Phase 2 with 127 patients completed plus an ongoing Phase 3 portion, ORR by Blinded Independent Central Review (BICR)) [3][13] and ASPEN-09-03 (HER2+ metastatic breast post-T-DXd, NCT07007559, single-arm Phase 2 with ~80 patients, primary endpoint ORR by BICR in the HER2 copy-number-amplified ctDNA subpopulation) [5]. Note this corrects my earlier read of ASPEN-09-03 as a randomized n=120 study - it is single-arm. Those are the studies that drive ALX's valuation. The HPV+ HNSCC study matters as proof-of-concept that CD47 blockade adds something on top of PD-1 plus radiation in an immunologically 'hot' tumor - if pCR is meaningfully above historical pembrolizumab+RT controls, it opens the door for ALX or a partner to run a randomized Phase 2.

Our model puts this drug's chance of eventual approval at 6%. That starts from the historical approval rate for Phase 2 drugs in this area, which is about 13%, then adjusts based on ten facts about the trial and sponsor. The estimate is nudged up by the trial's non-randomized design and open-label blinding, but pulled down by the sponsor's weak approval record and limited earlier-phase results. The remaining factors are close to average for this stage, so they don't shift the number much.

Mechanism risk first: the CD47 class has one major graveyard. Magrolimab - Gilead's $4.9B acquisition - was discontinued in 2024 after futility in MDS and safety holds in AML. Evorpacept's inert Fc was the design hypothesis that distinguished it, and the ASPEN-01 NHL data support cleaner hematologic safety, but the class-wide efficacy question is not resolved [7][11]. The MSS mCRC Phase 2 with evorpacept + cetuximab + pembrolizumab was terminated early for immune-related toxicity, showing combination safety is not a solved problem [8]. Trial-specific risk: n=29 single-arm with pCR endpoint is fine for hypothesis-generation but cannot support registration. If pCR comes in at 30-40% - overlapping plausible pembro+RT historical controls - the readout will be ambiguous and won't drive a Phase 3. Commercial risk for ALX itself is less acute than I first wrote: with $169.1M cash and runway into 1H 2028 [14], the immediate cash-runway concern has been addressed by the Q1 2026 financing. The longer-term question is whether ALX can run an ASPEN-06 Phase 3 and prep commercial infrastructure on its own, or whether it needs a partner. The earlier 2025 restructuring and prioritization happened from a position of weakness; the 2026 financing buys time but not enough for full commercialization without partnership. Competitive risk: PD-1 plus radiation is the de facto standard in HPV+ HNSCC neoadjuvant trials, and KEYNOTE-689 just put perioperative pembrolizumab on label [15]. The bar for adding a third agent is high - payers and KOLs will want a clear pCR or event-free survival delta, not just a numeric trend. CD47 class competition: Akeso's ligufalimab (AK117) is the most credible remaining CD47 antibody, with a Phase 3 trial (AK117-302) underway in 1L PD-L1+ R/M HNSCC pairing ligufalimab with ivonescimab (PD-1/VEGF bispecific) against pembrolizumab - the first Phase 3 of a CD47 antibody in solid tumors [16]. Lemzoparlimab was dropped by AbbVie in November 2023 with I-Mab continuing only a China Phase 3 in MDS [17]; letaplimab data is sparse. So the surviving CD47 field is essentially evorpacept and ligufalimab.

Worth watching, but not at the top of the queue. The interesting evorpacept readouts for commercial purposes are ASPEN-06 in HER2+ gastric (already positive Phase 2, Phase 3 portion pending) and ASPEN-09-03 in HER2+ metastatic breast post-T-DXd - those drive ALX's equity story and the HER2+ thesis where evorpacept has shown the cleanest signals so far [2][3]. This HPV+ HNSCC study is a small academic-led proof-of-concept; the signal that would matter is pCR ≥40% with manageable toxicity, which would justify a randomized follow-on. Anything in the 25-35% range is noise against pembro+RT historicals - and the bar is now higher with perioperative pembrolizumab FDA-approved [15]. Check back when UCSD reports pCR data - likely at a 2026 conference (ASCO June, AHNS July, or ESMO fall). On the company: cash position is no longer the headline risk after the Q1 2026 financing extended runway into 1H 2028 [14]. The real question is partnership versus go-it-alone for ASPEN-06 Phase 3 commercialization. Competitive watch item: Akeso's ligufalimab (AK117) Phase 3 in 1L PD-L1+ R/M HNSCC versus pembrolizumab is the first true CD47 Phase 3 readout in solid tumors and will set the bar for the whole class - if ligufalimab fails, evorpacept's mechanism story takes another hit; if it wins, it validates the class but introduces a direct competitor in HNSCC [16]. The cleanest tell on the mechanism itself remains ASPEN-06 Phase 3 confirmatory data - that's the trial that decides whether evorpacept becomes a real franchise or follows magrolimab into the CD47 graveyard.