In late August, BioNTech told investors and investigators that it was ending a clinical trial that had been underway since 2020. The Phase 2 BNT122-01 study evaluated whether the fully personalized mRNA cancer vaccine autogene cevumeran could prevent recurrence of colorectal cancer after surgical resection. The vaccine was administered as adjuvant monotherapy, without chemotherapy or a checkpoint inhibitor. An independent Data Safety Monitoring Board (DSMB) reviewed the most recent data and determined that the study was unlikely to demonstrate a survival benefit for the vaccine compared with observation. The DSMB recommended that the trial be stopped. BioNTech’s U.S.-listed shares fell nearly 9% that morning.
The timing was notable. Nine days earlier, on August 19, Merck & Co. and Moderna announced that their personalized neoantigen vaccine, intismeran autogene, had met its primary endpoint in a Phase 3 melanoma trial. Moderna’s stock rose 177% in a single session, and the result renewed interest in the cancer vaccine field, which has experienced repeated disappointments in late-stage development.
BioNTech’s shares also rose after the melanoma announcement, reflecting expectations that a positive Phase 3 result for one individualized mRNA cancer vaccine could support the broader approach. Nine days later, however, one of BioNTech’s own trials of the same general therapeutic concept produced a very different result. The two studies are worth examining together, but they were not testing identical treatment strategies.
What BNT122-01 Was Actually Testing
BNT122-01 (ClinicalTrials.gov identifier NCT04486378) was a multicenter, open-label, randomized Phase 2 trial designed to enroll approximately 327 patients with surgically resected, high-risk stage II or stage III colorectal cancer who remained positive for circulating tumor DNA (ctDNA) after surgery. Persistent ctDNA indicates molecular evidence of residual disease despite the absence of detectable disease on conventional imaging. Patients were randomized to receive autogene cevumeran as adjuvant monotherapy or to observation. The primary endpoint was disease-free survival, with overall survival and safety included as secondary objectives.
The difficulty became apparent before the final decision to stop the trial. In October 2025, BNT122-01 crossed a prespecified futility boundary, indicating that the study was unlikely to demonstrate a benefit if it were to continue to its planned conclusion.
At that point, however, the DSMB considered the data immature and follow-up insufficient to support a definitive conclusion and allowed the trial to continue. Ten months later, with more mature data available, the DSMB identified a numerical imbalance in overall survival between the treatment groups and concluded that additional follow-up was unlikely to change the result.
BioNTech emphasized that the decision was based on efficacy rather than tolerability. No new safety signal was identified. BioNTech and Genentech, a member of the Roche Group and its development partner for autogene cevumeran, jointly made the decision. The companies have said they will conduct a detailed analysis of the trial data, including patient characteristics and treatment response, to inform subsequent development of mRNA-based cancer immunotherapies.
Built on the Same Platform, Tested in Different Settings
Autogene cevumeran, also known as BNT122 or RO7198457, is mechanistically related to the melanoma vaccine that generated headlines in August. Both are individualized neoantigen vaccines.
The process begins with sequencing a patient’s tumor after surgical resection. Bioinformatic analysis identifies mutations specific to that tumor, and a personalized messenger RNA construct is produced to encode selected neoantigens. Autogene cevumeran can encode up to 20 neoantigens. The resulting vaccine is manufactured specifically for the individual patient and delivered using a lipid-based nanoparticle formulation. Neither vaccine is an off-the-shelf product. Each dose is manufactured for a single patient. The major difference between the programs is how the vaccines have been incorporated into treatment.
Clinical studies of intismeran autogene, including the Phase 3 INTerpath-001 trial in melanoma, have evaluated the vaccine in combination with the PD-1 inhibitor pembrolizumab (Keytruda®; Merck & Co/MSD) BNT122-01, in contrast, evaluated autogene cevumeran alone in colorectal cancer. The study was designed to determine whether vaccine-induced immunity was sufficient to prevent recurrence in patients with molecular evidence of residual disease. This distinction is important because therapeutic cancer vaccines and immune checkpoint inhibitors have different functions. A vaccine is intended to generate or expand T cells capable of recognizing tumor-specific antigens. A checkpoint inhibitor acts on inhibitory pathways that can suppress T-cell activity after those cells encounter the tumor.
Generating a tumor-specific immune response, therefore, does not necessarily mean the response will remain effective within the tumor microenvironment. The BNT122-01 study directly tested whether the vaccine alone could provide sufficient antitumor activity in a high-risk population with colorectal cancer.
BMO Capital Markets raised a similar issue in an investor note following the announcement, suggesting that the trial result could reduce confidence in autogene cevumeran and raise questions about whether vaccine-induced immune responses can prevent recurrence without concurrent checkpoint inhibition or chemotherapy.
A Difficult Setting
Colorectal cancer encompasses biologically distinct disease subsets. Approximately 4% to 5% of colorectal tumors have deficient mismatch repair (dMMR) or high microsatellite instability (MSI-H). These tumors carry a high mutational burden and can be substantially more immunogenic. Most colorectal cancers, approximately 85% to 90%, are mismatch-repair proficient and microsatellite stable (MSS). MSS colorectal cancer has historically shown limited sensitivity to immune checkpoint blockade as monotherapy, including in metastatic disease.
The dMMR/MSI-H population provides an important example of the difference between these biological subgroups. In the Phase 3 ATOMIC trial (NCT02912559), presented at the 2025 ASCO Annual Meeting, adding atezolizumab to standard adjuvant FOLFOX chemotherapy in patients with resected stage III dMMR colon cancer reduced the risk of recurrence or death by approximately 50% compared with chemotherapy alone. Three-year disease-free survival was 86.4% with atezolizumab plus chemotherapy versus 76.6% with chemotherapy alone (hazard ratio [HR], 0.50; 95% CI, 0.35-0.72).
ATOMIC, however, involved a molecularly selected population with dMMR disease and used a checkpoint inhibitor together with chemotherapy. BNT122-01 enrolled patients across high-risk stage II and stage III colorectal cancer and did not restrict enrollment to the dMMR/MSI-H subgroup. Most participants therefore would have been expected to have MSS disease. The difference in treatment strategy is equally important. BNT122-01 tested a personalized vaccine as monotherapy, whereas ATOMIC combined immunotherapy with a chemotherapy backbone.
ctDNA Identifies Residual Disease, but Does Not Define the Treatment
Circulating tumor DNA has become an important tool for identifying patients with molecular evidence of residual colorectal cancer after surgery. The clinical question, however, is what treatment should follow a positive ctDNA result. Two Phase 3 studies reported in 2026 illustrate approaches other than personalized vaccination. The CIRCULATE-US trial, presented at ASCO 2026, evaluated ctDNA-guided adjuvant chemotherapy in mismatch-repair-proficient stage II colon cancer. Three-year recurrence was 19% with ctDNA-guided treatment compared with 62% with observation (HR, 0.23).
The Phase 3 ALTAIR trial, conducted within the CIRCULATE-Japan platform and published in Nature Medicine in 2026, evaluated trifluridine/tipiracil in patients who became ctDNA-positive after completing standard adjuvant therapy. The study reported a modest, statistically nonsignificant improvement in disease-free survival. Neither study used a personalized cancer vaccine. Both used ctDNA positivity to identify patients for additional or alternative systemic therapy. BNT122-01 asked a different question: whether the same molecular evidence of residual disease could be used to select patients for treatment with a personalized vaccine alone. Based on the available results, the answer was no.
The Program Continues
BioNTech has stated that its other active Phase 2 program involving autogene cevumeran, IMcode003 (NCT05968326), will continue. Sponsored by Genentech, IMcode003 is evaluating autogene cevumeran in combination with atezolizumab and modified FOLFIRINOX chemotherapy versus modified FOLFIRINOX alone as adjuvant treatment for approximately 260 patients with resected pancreatic ductal adenocarcinoma (PDAC). The continued development of this program is supported by a different body of clinical evidence from that available in BNT122-01.
In an investigator-initiated Phase 1 study (NCT04161755), 16 patients with resected PDAC received sequential atezolizumab, autogene cevumeran, and modified FOLFIRINOX. Eight patients developed a detectable vaccine-induced T-cell response. Those patients had longer recurrence-free survival than patients without a detectable vaccine-induced response.(Rojas LA, et al. Nature. 2023;618:144-150. PMID: 37165196).
An extended follow-up published in Nature in 2025, with a median follow-up of 3.2 years, showed that the difference persisted. Median recurrence-free survival had not been reached among vaccine responders compared with 13.4 months among nonresponders (HR, 0.14; 95% CI, 0.03-0.6; P=0.007). The vaccine-induced CD8+ T-cell clones were estimated to have an average lifespan of 7.7 years. (Balachandran VP, et al. Nature. 2025;639:1042-1051. PMID: 39972124).
A six-year follow-up presented at the AACR Annual Meeting in April 2026 extended the observation further, with investigators from Memorial Sloan Kettering Cancer Center reporting that nearly 90% of the original T-cell responders remained alive.
A separate Phase 1 study of autogene cevumeran, administered with or without atezolizumab in 213 patients with a range of advanced, previously treated solid tumors, was published in Nature Medicine in 2025. The vaccine elicited detectable neoantigen-specific T-cell responses in 71% of patients, with responses remaining measurable for up to 23 months after treatment began. Objective tumor responses were observed in a small monotherapy cohort and among patients treated with atezolizumab. (Lopez J, et al. Nat Med. 2025;31:1187-1197. PMID: 39762422.
These findings do not establish that IMcode003 will improve clinical outcomes. The pancreatic cancer Phase 1 data were generated in a small study, and an association between vaccine-induced immune responses and recurrence-free survival does not establish that the vaccine caused the difference in outcomes. The pancreatic program, however, differs from BNT122-01 in several important respects. It incorporates a checkpoint inhibitor and chemotherapy, and its development has included a repeated association between vaccine-induced immunity and clinical outcomes in published studies.
BioNTech is also conducting IMcode004, a Phase 2 trial evaluating autogene cevumeran plus the checkpoint inhibitor nivolumab (Opdivo®; Bristol Myers Squibb), versus nivolumab alone in patients with resected muscle-invasive urothelial carcinoma. Also, in this case, the vaccine is being evaluated as part of a combination regimen rather than as monotherapy.
Two Trials, Two Different Questions
INTerpath-001 and BNT122-01 should not be viewed as contradictory tests of a single treatment concept. They evaluated individualized neoantigen vaccines across different tumor types and, importantly, in combination with different treatment partners. INTerpath-001 evaluated an individualized neoantigen vaccine in combination with checkpoint inhibition in melanoma and demonstrated a statistically significant reduction in recurrence in a randomized Phase 3 study.
BNT122-01 evaluated autogene cevumeran alone in colorectal cancer and did not demonstrate a survival benefit in a randomized Phase 2 study. The difference is clinically relevant. The melanoma trial was conducted in a tumor type with high immunogenicity and a well-established history of benefit from checkpoint blockade. BNT122-01, by contrast, included a predominantly MSS colorectal cancer population, a disease setting in which single-agent checkpoint inhibition has historically shown limited activity.
The manufacturing platforms are similar: both approaches require sequencing the patient’s tumor, identifying tumor-specific mutations, and producing a personalized mRNA construct. The clinical treatment strategies are not. The available results therefore do not support treating “cancer vaccines” as a single therapeutic category with one expected outcome. Tumor biology, patient selection, and the therapies administered with the vaccine are central components of the clinical question.
What Comes Next?
BioNTech has said it will conduct a more detailed analysis of the BNT122-01 results. That analysis should provide additional information about the overall-survival finding, which has so far been described publicly only as a numerical imbalance. A hazard ratio and absolute event counts have not been disclosed. The currently available information does not establish the reason for that imbalance. The randomized results from IMcode003 in pancreatic cancer will be particularly important for evaluating whether the immune response and recurrence-free survival findings observed in early-stage studies translate into improved outcomes in a larger clinical trial.
For now, the BNT122-01 result provides a specific clinical lesson rather than a verdict on personalized cancer vaccines as a whole: in this colorectal cancer population, a personalized mRNA vaccine given as adjuvant monotherapy did not demonstrate the expected clinical benefit. Ongoing combination studies will determine whether the same platform yields different results when used with chemotherapy and/or immune checkpoint blockade.
Related Onco’Zine Coverage
- A Billion-Dollar Verdict: What the Economics Behind INTerpath-001 Tell Us About the Future of Cancer Vaccines [Article]
Clinical trials
- ClinicalTrials.gov. A Study of Efficacy and Safety of Autogene Cevumeran in Combination With Standard of Care Compared With Standard of Care Alone in Participants Who Have Undergone Surgical Resection of Colorectal Cancer (BNT122-01). Identifier NCT04486378.
- ClinicalTrials.gov. A Study of the Efficacy and Safety of Adjuvant Autogene Cevumeran Plus Atezolizumab and mFOLFIRINOX Versus mFOLFIRINOX Alone in Participants With Resected Pancreatic Ductal Adenocarcinoma (IMCODE003). Identifier NCT05968326.
- ClinicalTrials.gov. A Study to Evaluate the Efficacy and Safety of Autogene Cevumeran With Nivolumab Versus Nivolumab Alone in Participants With High-risk Muscle-invasive Urothelial Carcinoma (IMCODE004). Identifier NCT06534983.
Highlights of Prescribing Information
Pembrolizumab (Keytruda®; Merck & Co/MSD)[Prescribing Information]
Atezolizumab (Tecentriq®; Genentech/Roche)[Prescribing Information]
Nivolumab (Opdivo®; Bristol Myers Squibb/BMS)[Prescribing Information]
References
[1] BioNTech SE. BioNTech Provides Update on Phase 2 Clinical Trial of Autogene Cevumeran in Resected Colorectal Cancer. Press release, August 28, 2026. Available at: Online. Last accessed on August 31, 2026.
[2] BioNTech SE. Form 6-K, Current Report of Foreign Private Issuer. U.S. Securities and Exchange Commission, August 28, 2026. Online. Last accessed on August 31, 2026.
[3] BioSpace. BioNTech halts mid-stage trial of personalized mRNA vaccine in colorectal cancer. August 2026. Online. Last accessed on August 31, 2026.
[4] RTTNews. BioNTech Terminates Phase 2 Trial Of Autogene Cevumeran In Colorectal Cancer. August 2026. Online. Last accessed on August 31, 2026.
[5] Benzinga. BioNTech Stock Dips After BNT122-01 Trial Termination. August 2026. Online. Last accessed on August 31, 2026.
[6] Rojas LA, Sethna Z, Soares KC, et al. Personalized RNA neoantigen vaccines stimulate T cells in pancreatic cancer. Nature. 2023;618(7963):144-150. doi:10.1038/s41586-023-06063-y. PMID: 37165196.
[7] Balachandran VP, et al. RNA neoantigen vaccines prime long-lived CD8+ T cells in pancreatic cancer. Nature. 2025;639(8057):1042-1051. doi:10.1038/s41586-024-08508-4. PMID: 39972124.
[8] Lopez J, Powles T, Braiteh F, et al. Autogene cevumeran with or without atezolizumab in advanced solid tumors: a phase 1 trial. Nat Med. 2025;31(4):1187-1197. doi:10.1038/s41591-024-03334-7. PMID: 39762422.
[9] CancerNetwork. Autogene Cevumeran & The Future of Personalized Cancer Vaccines. 2026. Online. Last accessed on August 31, 2026.
[10] OncLive. mRNA Vaccines May Help Increase Long-Term Disease Control in PDAC. 2026. Online
[11] Alliance for Clinical Trials in Oncology. Alliance Presents Results from Phase III ATOMIC Trial (A021502). Press release, June 1, 2025. Online.Last accessed on August 31, 2026.
[12] Sinicrope FA, Ou FS, et al. Randomized trial of standard chemotherapy alone or combined with atezolizumab as adjuvant therapy for patients with stage III deficient DNA mismatch repair (dMMR) colon cancer (Alliance A021502; ATOMIC). J Clin Oncol. 2025;43(17_suppl):LBA1.
[13] American Society of Clinical Oncology Educational Book. Immunotherapy in Localized Colorectal Cancer: Current Practice and Future Directions. 2026. doi:10.1200/EDBK-26-520600.
[14] OncologyNewsCentral. ctDNA Use in Colorectal Cancer: Highlights From ASCO 2026 (CIRCULATE-US). June 2026. Online. Last accessed on August 31, 2026.
[15] Nakamura Y, et al. Post-adjuvant chemotherapy in ctDNA-positive patients with resected colorectal cancer: a randomized phase 3 trial (ALTAIR). Nat Med. 2026. doi:10.1038/s41591-026-04428-0.
[16] Siegel RL, Kratzer TB, Giaquinto AN, Sung H, Jemal A. Colorectal cancer statistics, 2026. CA Cancer J Clin. 2026. PMID: 41769777.
[17] American Cancer Society. Colorectal Cancer Statistics | How Common Is Colorectal Cancer? 2026. Online. Last accessed on August 31, 2026.
This article is intended for informational purposes for healthcare professionals and does not constitute medical advice. Autogene cevumeran remains investigational and is not approved by the FDA, the European Medicines Agency, or any regulatory authority for any indication. Prescribing information referenced here for atezolizumab pertains to its approved indications only and is not a complete summary of safety data for the investigational combinations discussed in this article; consult the full FDA-approved label before making treatment decisions.
Featured image © 2026 CH/JCO Used with permission.
DOI
