PolyPEPI1018

PolyPEPI1018 is a 12-peptide off-the-shelf cancer vaccine from Treos Bio, a small UK-Hungarian biotech, paired with Roche's atezolizumab (a PD-L1 blocker that takes the brakes off T cells) in microsatellite-stable metastatic colorectal cancer [1][2]. This is the hardest CRC subtype to treat with immunotherapy - checkpoint inhibitors alone barely move the needle in MSS disease, which is roughly 95% of metastatic CRC. The Phase 2 trial NCT05243862 (OBERTO-301) finished enrolling 18 patients, and updated results presented at ASCO 2024 showed an objective response rate of 0%, disease control rate of 61%, median PFS of 2.7 months, and no Grade 3-5 treatment-related events [5][6]. In March 2026, Treos signed a partnership with GenDx to co-develop an HLA-typing companion diagnostic combining GenDx's NGSgo kits with Treos's PEPI Test software, intended for clinical validation in a planned Phase 3 [7]. The bet was that priming the immune system with tumor-associated peptides would give atezolizumab something to release; the Phase 2 readout suggests a soft signal at best in an unselected population.

PolyPEPI1018 is a novel investigational compound, never approved. The lead Phase 2 trial NCT05243862 (OBERTO-301: PolyPEPI1018 + atezolizumab in MSS mCRC) is active but no longer recruiting, with 18 patients enrolled and Treos Bio as sponsor [2]. Initial Stage 1 results were presented at ESMO 2023 (abstract 640P, Annals of Oncology) [5] and an updated 18-patient readout was presented at ASCO 2024: ORR 0%, DCR 61%, mPFS 2.7 months (95% CI 1.4-4.0), mOS not reached, and no Grade 3-5 treatment-related adverse events [6]. No FDA breakthrough, fast-track, or orphan designations have been disclosed. Two earlier Phase 1 trials are complete: NCT03391232 (OBERTO Phase 1b, n=11, Treos-sponsored, published as Hubbard et al. Clin Cancer Res 2022, showing immunogenicity and disease control in a maintenance setting) [1][4] and NCT05130060 (OBERTO-201, n=15, investigator-initiated at Mayo Clinic combining the vaccine with TAS-102/trifluridine-tipiracil, a standard third-line chemo for refractory CRC) [3]. On 17 March 2026, Treos and GenDx announced a partnership to co-develop an HLA-based companion diagnostic for PolyPEPI1018, integrating GenDx's IVDR-registered NGSgo HLA-A/B/C typing kits and NGSengine analysis software with Treos's proprietary PEPI Test software, with clinical validation planned during Treos's anticipated Phase 3 trial [7]. Treos Bio is a small private company without an active SEC filing trail; expect any further updates at ASCO GI or ESMO GI rather than via earnings disclosures.

PolyPEPI1018 is a fixed cocktail of 12 short peptides - fragments of proteins - drawn from 7 tumor-associated antigens that show up disproportionately in colorectal tumors. The idea is simple: inject the peptides under the skin with an adjuvant, dendritic cells pick them up, and trained T cells go hunt down tumor cells displaying those same fragments. Unlike neoantigen vaccines that target mutation-derived peptides unique to the tumor (such as Moderna's mRNA-4157 or BioNTech's autogene cevumeran), PolyPEPI1018 targets tumor-associated antigens (TAAs) - proteins the body overexpresses but does not mutate. The immune system may have already learned to tolerate these self-proteins through central and peripheral tolerance, which limits the depth and avidity of any T cell response a vaccine can generate. Treos calls this an 'off-the-shelf' vaccine because it isn't personalized to each patient's tumor. Instead, the company's PEPI platform predicts which patients have HLA types (immune-presentation molecules) likely to display enough of the 12 peptides to generate a response. The atezolizumab combination is the standard rationale for any cancer vaccine in 2026 - vaccines generate T cells; checkpoint inhibitors keep those T cells from being shut off by tumor PD-L1. Mechanistically the logic is clean. The track record for therapeutic cancer vaccines is not. Sipuleucel-T (Provenge) and T-VEC (Imlygic) are the only two FDA-approved therapeutic cancer vaccines, and neither generates meaningful revenue. Dozens of TAA-based peptide vaccines have shown peripheral immunogenicity without translating to survival benefit - the tolerance problem is a large part of why.

NCT05243862 (OBERTO-301) is a multicenter, open-label, single-arm Phase 2 study at Mayo Clinic sites in MSS mCRC patients who have progressed on 2-3 prior lines of chemotherapy, combining PolyPEPI1018 (1.2 mg sc) with atezolizumab (1,200 mg iv) every 3 weeks [2][5]. Eligibility is based on MSS status and prior therapy lines - the trial enrolls unselected patients and does not apply HLA-based PEPI score eligibility, which is the single most important design variable for a vaccine that depends on antigen presentation. This is why the GenDx partnership matters: an HLA-typing companion diagnostic is being built specifically to enable biomarker-stratified enrollment in a planned Phase 3 [7]. Any efficacy signal in the unselected Phase 2 cohort would, in principle, imply a stronger signal in an HLA-enriched Phase 3 population. Primary endpoint is incidence and severity of treatment-related adverse events - a safety endpoint, not efficacy [2]. That alone signals this is a signal-finding study, not registration-enabling. Target enrollment is 18 patients, hit. There is no comparator arm, so any efficacy signal has to be benchmarked against historical controls: late-line MSS mCRC gives roughly 0-2% ORR on checkpoint monotherapy (KEYNOTE-016 reported 0/18 responders in MSS vs 4/10 in MSI-H) [8]. The small sample size is the obvious limitation: 18 patients can show a safety signal and generate hypotheses but cannot establish efficacy with statistical confidence. Translational endpoints - peptide-specific T cell responses, tumor microenvironment changes on biopsy - matter more than RECIST response rates (tumor shrinkage of ≥30% by standardized imaging criteria) at this scale. The Mayo-led NCT05130060 study pairing the vaccine with TAS-102 is the more clinically pragmatic test, because TAS-102 is a real third-line option and any added benefit would translate directly to a regulatory path [3].

Our model puts this drug's chance of eventual approval at 6%. That starts from a 13% historical baseline for Phase 2 drugs in this area, then adjusts based on ten facts about the trial and the company behind it. The estimate drops mainly because the sponsor has a thin approval record, enrollment is smaller than typical for this phase, and earlier-phase results were weak. A non-randomized trial design pulls the number up slightly, but not enough to offset those drags.

Efficacy risk is the dominant concern, and the Phase 2 ORR=0% signal in 18 unselected patients does not refute it [6]. MSS colorectal cancer is the canonical 'cold' tumor - low mutational burden, few neoantigens, an immunosuppressive microenvironment dominated by myeloid-derived suppressor cells (MDSCs - immune cells the tumor recruits to block T cell attack). Atezolizumab monotherapy has essentially no activity here; the IMblaze370 Phase 3 testing atezolizumab ± cobimetinib vs regorafenib in chemotherapy-refractory mCRC failed on overall survival (median OS 8.87 vs 7.10 vs 8.51 months across arms) [7-Eng-2019]. Asking a TAA-based peptide vaccine to convert cold tumors hot is biologically ambitious; central tolerance to self-antigens is a structural headwind that limits T cell repertoire depth, which is why most TAA vaccines have failed to translate peripheral immunogenicity into tumor regression. Patient-selection risk compounds this: NCT05243862 does not apply HLA-based enrichment, so any peptide-specific response is diluted across patients whose immune systems can't even present the antigens. Trial-design risk is real - a single-arm n=18 study cannot distinguish a true signal from noise in a disease where spontaneous stable disease is common, and DCR 61% on its own is consistent with indolent biology rather than treatment effect. Safety risk is probably the smallest worry; the Phase 2 reported no Grade 3-5 treatment-related events and atezolizumab's immune-related adverse event (irAE) profile is well-characterized [6]. Execution and commercial risk hit hardest at the sponsor level. Treos Bio is private, small, and dependent on partnership or financing to take this past Phase 2. The GenDx CDx partnership is a positive signal for execution intent but is not the same as a Phase 3 partner. Big pharma has cooled on therapeutic cancer vaccines after a decade of disappointment; pricing power for any off-the-shelf vaccine in a heavily generic indication would be limited.

Worth a periodic check, not active monitoring - but the monitoring posture has been updated because Phase 2 data already exists. The OBERTO-301 readout at ASCO 2024 (n=18, ORR 0%, DCR 61%, mPFS 2.7 months, no Grade 3-5 events) [6] is a soft positive on safety, a null on RECIST partial responses (tumor shrinkage ≥30%), and ambiguous on disease control given the absence of a comparator arm. What to watch for next: (1) a presented or published Phase 2 update with HLA-stratified subgroup analysis showing whether PEPI-matched patients had better DCR or longer PFS than PEPI-mismatched patients - this is the single highest-information readout the program can produce; (2) any confirmed RECIST partial response in MSS patients in an expansion or follow-up cohort - even one would be unusual; (3) immune correlates showing peptide-specific T cells infiltrating the tumor, not just circulating in peripheral blood. Peripheral immunogenicity is cheap; intratumoral T cell infiltration is the real test. The GenDx partnership announced 17 March 2026 [7] is the concrete catalyst the program needed: it provides a credentialed HLA-typing workflow (NGSgo + NGSengine) integrated with Treos's PEPI Test, slated for clinical validation in a planned Phase 3. This converts 'HLA-based patient selection' from a hypothetical into an active development track. The Mayo-sponsored NCT05130060 readout is arguably more clinically informative because TAS-102 is a real comparator backbone and the trial is run by an academic group with no incentive to spin the data [3]. Treos Bio's PEPI platform is more interesting than this specific vaccine; if a Phase 3 demonstrates prospective HLA-based patient selection improving response rates, that would be a real story for the platform - and potentially for the field of off-the-shelf TAA vaccines more broadly. Check back at ASCO GI January 2027, ASCO 2027, or after any peer-reviewed Phase 2 follow-up publication. Until then, low priority.