AbelZeta Pharma has highlighted the peer-reviewed publication of first-in-human clinical findings for C-CAR031, an investigational autologous glypican-3-targeted chimeric antigen receptor T-cell therapy developed for advanced hepatocellular carcinoma. The Nature paper reported an objective response rate of 44.4% among 36 heavily pretreated patients, while also identifying limited response durability and biological mechanisms that may allow tumors to escape treatment.
C-CAR031 is now being developed by AstraZeneca as AZD7003 after the pharmaceutical company acquired AbelZeta’s remaining development and commercialization rights in China. The publication therefore represents more than retrospective recognition of an early AbelZeta programme. It provides AstraZeneca with a peer-reviewed clinical and translational evidence package as it decides how aggressively to advance an armored CAR-T platform into later-stage liver cancer development.
The study provides one of the more closely watched signals yet that CAR-T cells can generate meaningful tumor regression in a solid malignancy with an immunosuppressive microenvironment. It does not establish C-CAR031 as an effective treatment, however. The trial was an early-phase, single-arm study without a comparator, and the median duration of response was 4.4 months, leaving durability as the central question behind the headline response rate.
What did the C-CAR031 first-in-human study show in heavily pretreated liver cancer?
The Nature publication evaluated 36 patients with advanced, treatment-refractory hepatocellular carcinoma who received C-CAR031 at four dose levels ranging from 0.75 million to 4 million cells per kilogram. The patients had disease that had progressed despite previous systemic treatment, placing the therapy in a population with substantial unmet need and relatively few attractive later-line options.
Tumor regression was observed in 32 of the 36 treated patients. The median best reduction in target-lesion size was 41.6%, with reductions ranging from 3.4% to 94.4%. Sixteen patients met the criteria for an objective response, producing the reported objective response rate of 44.4%.
The study reported median progression-free survival of 4.2 months and median overall survival of 14.2 months. Median duration of response was 4.4 months, with a 95% confidence interval of 2.9 to 7.4 months. These measures provide a fuller interpretation than the response rate alone because they indicate that tumor shrinkage did not always translate into prolonged disease control.
The findings nevertheless represent a notable signal in a difficult setting. CAR-T therapies have delivered major benefits in certain blood cancers, but solid tumors present additional barriers, including uneven antigen expression, restricted cellular trafficking, physical tumor architecture and immunosuppressive signaling within the tumor microenvironment. Demonstrating measurable regression across most patients suggests that engineered cells reached and attacked both liver and metastatic disease in at least a substantial proportion of the cohort.
The result should still be viewed as hypothesis-strengthening rather than confirmatory. The study was not randomized, was not designed to compare C-CAR031 with an established systemic treatment and included patients treated at different dose levels. A future expansion or Phase II study will need to reproduce the response signal in a larger and more consistently treated population.
Why is the dominant-negative TGFβ receptor important to the C-CAR031 design?
C-CAR031 targets glypican-3, a cell-surface protein frequently expressed in hepatocellular carcinoma and more limited in most healthy adult tissues. That expression pattern has made glypican-3 one of the most actively investigated targets for cellular therapy, antibody programmes and other targeted approaches in liver cancer.
Target recognition is only one part of the challenge. Hepatocellular carcinoma tumors can produce high concentrations of transforming growth factor beta, or TGFβ, which suppresses T-cell activity and contributes to an environment that protects malignant cells from immune attack. A conventional CAR-T cell may recognise its target but become less functional after entering this suppressive environment.
C-CAR031 was therefore engineered with a dominant-negative TGFβ receptor II. The additional component is intended to reduce the inhibitory effect of TGFβ signaling on the infused T cells, allowing them to maintain activity after reaching the tumor. AstraZeneca designed the anti-GPC3 targeting component and the dominant-negative receptor armour, while Shanghai AbelZeta contributed product engineering and manufacturing.
This mechanism is commercially important because it attempts to solve one of the most persistent obstacles facing solid-tumor CAR-T development. Rather than increasing target-binding strength alone, the therapy was designed to protect cell function against a specific immunological barrier associated with hepatocellular carcinoma.
The clinical findings suggest that the approach was biologically active, but they do not demonstrate that the armour completely neutralised the TGFβ problem. The translational analysis found that higher TGFβ levels could still be associated with resistance. The engineering may therefore improve resistance to suppression without making the cells immune to all concentrations, sources or downstream effects of TGFβ.
Why does the 4.4-month response duration temper enthusiasm around the 44.4% response rate?
The objective response rate is the most immediately impressive result because tumor shrinkage in almost half of a heavily pretreated advanced hepatocellular carcinoma cohort is clinically interesting. The median duration of response, however, indicates that the benefit was frequently temporary.
A median response duration of 4.4 months does not make the response rate irrelevant. In refractory disease, even a temporary response can demonstrate that a target and therapeutic mechanism are active. For drug development, though, the value of the platform will depend on whether later studies can extend those responses and translate them into more consistent progression-free and overall survival outcomes.
Durability is especially important for autologous CAR-T therapies because their manufacturing, conditioning and administration requirements are substantially more complex than those of an off-the-shelf infusion. Patients generally undergo cell collection, individualized manufacturing, lymphodepleting chemotherapy and clinical monitoring around the infusion. A therapy carrying that operational burden will need to deliver sufficient depth and persistence of benefit to justify its use against increasingly competitive systemic regimens.
Cross-trial comparisons with tyrosine kinase inhibitors, immune checkpoint combinations or other cell therapies would be unreliable at this stage. Those studies may differ in treatment line, liver function, disease burden, biomarker requirements, geographic population and assessment methods. The appropriate test is not whether an isolated percentage appears numerically higher than a percentage from another trial, but whether C-CAR031 can reproduce clinically meaningful and durable responses in a prospectively defined population.
The 14.2-month median overall survival is encouraging in context, but it cannot be attributed entirely to C-CAR031 without a control group. Subsequent anticancer treatment, baseline patient selection and differences in disease biology may also influence overall survival in a small early-phase cohort.
What does the safety profile reveal about the feasibility of armored CAR-T therapy in liver cancer?
Cytokine release syndrome was reported in 34 of the 36 treated patients, meaning the inflammatory toxicity occurred in nearly the entire cohort. Most cases were lower grade, while two patients experienced grade 3 cytokine release syndrome.
Nine patients had grade 3 or higher non-haematological adverse events. The published abstract characterized the overall safety profile as manageable, but that description should not be interpreted as an absence of substantial toxicity or monitoring requirements.
Cytokine release syndrome is a recognised complication of CAR-T treatment and can involve fever, low blood pressure, reduced oxygen levels and organ dysfunction. Its high incidence in this study reinforces the expectation that C-CAR031 would require delivery at specialized centres with cellular therapy expertise, rapid access to supportive treatment and the capacity to monitor patients after infusion.
The safety interpretation is also complicated by the underlying condition. Patients with advanced hepatocellular carcinoma may have impaired liver reserve, portal hypertension, viral hepatitis, bleeding risk and other comorbidities that can narrow the margin for treatment-related inflammation. The trial selected patients with relatively preserved performance status and liver function, including a Child-Pugh score of six or lower, which may limit generalizability to a broader advanced liver cancer population.
Later studies will need to determine whether a recommended dose can preserve antitumor activity while controlling severe inflammatory and organ-related events. They will also need longer follow-up for delayed toxicities and a more complete understanding of whether treatment risk varies with tumor burden, liver function, viral status or baseline inflammatory markers.
How could GPC3 antigen loss and rising TGFβ levels drive resistance to C-CAR031?
One of the paper’s most valuable contributions is its effort to explain why some tumors progressed despite initial sensitivity. High-throughput analysis of tumor samples and functional validation suggested that reduced GPC3 expression and elevated TGFβ levels may contribute to C-CAR031 resistance.
Antigen loss is a fundamental vulnerability for single-target cellular therapies. C-CAR031 requires sufficient glypican-3 expression to identify and attack malignant cells. If treatment eliminates cells with high GPC3 expression while leaving low-expressing or antigen-negative cells behind, the remaining population can expand and produce relapse.
That possibility strengthens the case for stricter biomarker selection and repeat tissue analysis. A larger trial may need to establish whether the intensity, percentage or distribution of GPC3 expression predicts response. A simple positive or negative classification may prove inadequate when expression is heterogeneous within the same tumor or differs between primary and metastatic lesions.
The persistence of high TGFβ as a resistance factor presents a different challenge. It suggests that the dominant-negative receptor may have a functional threshold, or that TGFβ contributes to tumor escape through pathways extending beyond direct suppression of the infused cells. Combination approaches could therefore become important, although combining C-CAR031 with other therapies could also increase toxicity and complicate manufacturing and treatment sequencing.
The original investigator-initiated study registry includes planned exploration of C-CAR031 with lenvatinib, regorafenib or AstraZeneca’s durvalumab. Those combinations may help alter tumor vasculature, immune signaling or disease control, but their clinical value must be demonstrated rather than inferred from biological rationale.
What does AstraZeneca’s control of AZD7003 mean for AbelZeta and the development strategy?
AbelZeta and AstraZeneca initially partnered to co-develop C-CAR031 in China. In January 2026, AstraZeneca agreed to acquire AbelZeta’s remaining 50% share of the Chinese development and commercialization rights, giving AstraZeneca sole global control of the programme, now identified as AZD7003.
AbelZeta said the China agreement could provide up to $630 million through a combination of an upfront payment and development, regulatory and sales milestones. That headline value should not be treated as cash already received because much of it depends on future clinical and commercial achievements. AbelZeta also remains eligible for additional milestones and royalties connected with development outside China under the companies’ earlier arrangement.
The Nature publication supports AbelZeta’s scientific and manufacturing credentials even though control of the asset has transferred. It demonstrates that a therapy engineered and manufactured with AbelZeta’s participation produced peer-reviewed clinical activity and translational findings in a solid tumor.
For AstraZeneca, the publication gives AZD7003 a more developed biological narrative. The company now has evidence of tumor regression, a defined safety profile and identified resistance mechanisms that can inform dose selection, biomarker strategy and combination development.
The commercial hurdle remains high. Autologous CAR-T manufacturing is costly and operationally demanding, while hepatocellular carcinoma is increasingly served by systemic combinations that can be administered across a wider hospital network. AZD7003 will need to show not only that it can induce responses, but that the responses are sufficiently durable and clinically valuable to justify the treatment infrastructure.
Which clinical milestones will determine whether AZD7003 can progress beyond an early efficacy signal?
A separate Phase I/II study listed on ClinicalTrials.gov is evaluating C-CAR031 in GPC3-positive advanced or recurrent hepatocellular carcinoma. The open-label, single-arm study is targeting enrolment of approximately 121 participants who have progressed after, or cannot tolerate, at least two previous lines of systemic therapy.
Its Phase I component is designed to establish a recommended expansion dose and a recommended Phase II dose. The Phase II portion will assess objective response rate through independent review under RECIST 1.1, alongside duration of response, progression-free survival, overall survival and safety measures.
That development programme should provide a more reliable estimate of efficacy than the first 36-patient dataset, particularly if patients are treated at a consistent dose and selected through an analytically validated GPC3 assay. Independent response review will also reduce some of the interpretive uncertainty associated with investigator-assessed results.
The most important future readouts will be the reproducibility of the response rate, the proportion of responses lasting six or 12 months, the safety profile at the selected dose and the relationship between GPC3 expression and clinical benefit. Investigators will also need to determine whether manufacturing can reliably deliver treatment before rapidly progressing patients deteriorate or require another therapy.
The Nature publication moves C-CAR031, now AZD7003, beyond a conference-stage signal and into a more credible peer-reviewed development phase. It does not resolve the programme’s central risk. AstraZeneca must now show that engineering CAR-T cells to resist the liver cancer microenvironment can produce benefits that last long enough to outweigh the toxicity, complexity and cost of individualized cellular treatment.
