The logic behind the study is deceptively simple but clinically consequential. CAR-T therapy works by engineering a patient’s own T cells to recognize a specific antigen on the surface of tumor cells. The strategy has produced remarkable results in blood cancers, but solid tumors have proven far more resistant, in part because tumor cells can simply stop displaying the antigen the therapy targets, a phenomenon known as target escape. One established countermeasure is to engineer CAR-T cells that recognize two different membrane molecules simultaneously, so that tumor cells remain visible to the immune system as long as they express either one of the two antigens. To justify such a design, researchers first need to know how often the two candidate antigens appear together in the same patients’ tumors, data that had been largely missing for GPC-3 and AXL in liver cancer.

The research team, led by Chenli Qiu, Jieyi Shi and Su Yan, who contributed equally, with senior authors Xiaoyan Zhang, Xiaowu Huang and Jianqing Xu, analyzed surgical specimens from 140 patients with hepatocellular carcinoma who underwent hepatectomy at Zhongshan Hospital in 2019. All patients had complete clinical information, and the samples were obtained following Institutional Review Board approval and written informed consent. The investigators used immunohistochemistry on 5-micrometer formalin-fixed, paraffin-embedded tissue sections, a standard pathology technique that uses antibodies to reveal where specific proteins sit within tissue. Sections were subjected to heat-induced antigen retrieval in Tris/EDTA buffer at 99 degrees Celsius, blocked with bovine serum albumin, and incubated overnight at 4 degrees Celsius with primary antibodies against AXL and GPC-3 before signal detection with HRP-labeled secondary antibodies and DAB chromogen.

Quantification followed a rigorous semiquantitative scheme. Only cytoplasmic or membranous staining counted as positive, and two pathologists independently scored each slide on a four-point scale, from 0 for no expression through weak, moderate and strong expression. The slides were digitized on a TissueFAXS platform, and categorical comparisons were analyzed with Pearson’s chi-square test, ordered data with the Goodman–Kruskal Gamma test, and survival outcomes with Kaplan–Meier curves and the log-rank test. This level of methodological detail matters because immunohistochemical scoring is inherently interpretive, and the discrepancy between different studies often traces back to the antibodies used and the thresholds applied, a point the authors themselves acknowledge.

The headline numbers are striking. GPC-3 was positive in 108 of 140 tumors, a rate of 77.1 percent, with strong expression accounting for the largest share, 42.1 percent of all cases. AXL was even more prevalent, positive in 120 of 140 tumors, or 85.7 percent, with strong expression in 48.6 percent. Critically, only 9 of the 140 samples, 6.4 percent, were negative for both proteins, meaning that 93.6 percent of hepatocellular carcinomas expressed at least one of the two antigens and 69.3 percent expressed both simultaneously. In adjacent non-tumor liver tissue, by contrast, GPC-3 was weakly positive in just one of 140 samples, and AXL staining, while present at low levels in some surrounding tissue, was consistently weaker than in the tumors themselves.

That contrast between tumor and normal tissue is what makes these two proteins attractive drug targets. GPC-3 is an oncofetal protein, a glycosylphosphatidylinositol-anchored membrane molecule that drives embryonic morphogenesis through the canonical Wnt/beta-catenin and Hedgehog pathways. It is essentially silent in the healthy adult liver but reactivated in hepatocellular carcinoma, where it appears in the cytoplasm and on the cell membrane in 70 to 100 percent of cases in previous analyses. Mechanistically, GPC-3 acts as a co-receptor that recruits Wnt, insulin-like growth factors and growth factors such as HGF, FGF-2 and TGF to their respective receptors, fueling proliferation, migration and survival. Overexpression also disrupts the Bax/Bcl-2 axis to increase resistance to apoptosis, promotes epithelial–mesenchymal transition through the ERK pathway, and modulates the CXCL12/CXCR4 axis to drive angiogenesis and metastasis.

AXL tells a complementary story. It is a receptor tyrosine kinase of the TAM family, spanning TYRO3, AXL and MERTK, and it is implicated in cancer cell proliferation, stemness, invasion, angiogenesis and therapeutic resistance. When its ligand GAS6 binds, AXL dimerizes and activates downstream PI3K/Akt and MAPK/ERK signaling, pathways central to tumor growth and spread. AXL is overexpressed across many malignancies, at roughly 47 percent in renal cancer, 69 percent in lung cancer, 57.6 percent in breast cancer and 76.7 percent in colon cancer, and prior studies had placed its expression in hepatocellular carcinoma at 22 to 43 percent. The higher rate observed in the new study may reflect differences in patient populations, detection antibodies and scoring criteria. Notably, AXL positivity was associated with microvascular invasion in this cohort, though neither protein correlated with overall survival over the 50-month follow-up, possibly because the negative sample groups were small and follow-up relatively short.

The therapeutic implications extend beyond CAR-T design. A phase I dose-escalation study of autologous CAR-T cells targeting GPC-3 in advanced hepatocellular carcinoma already showed a manageable safety profile, with an overall objective response rate of 50 percent that reached 57.1 percent in the highest dose cohort. On the AXL side, the selective inhibitor bemcentinib has received US FDA fast-track designation in combination with a PD-1/PD-L1 agent, and AXL-directed CAR-T cells have been reported in non-small cell lung cancer and triple-negative breast cancer. Intriguingly, the authors highlight reports that anti-PD-L1 antibodies can promote liver cancer cell proliferation by activating a PD-L1–AXL signal relay, which suggests that the high AXL expression documented in this study may partly explain the suboptimal performance of immune checkpoint inhibitors in some hepatocellular carcinoma patients, and that AXL-targeted therapy could sensitize tumors to checkpoint blockade.

Dual-targeting is not a panacea, however, and the authors are candid about the challenges. AXL is also present on macrophages and dendritic cells, raising the specter of on-target, off-tumor toxicity if engineered T cells attack healthy immune populations. Proposed engineering countermeasures include optimizing CAR structural design, incorporating logic gates that require combinations of antigens to trigger killing, and building suicide gene systems that can eliminate the infused cells if toxicity emerges. Antigen escape, limited persistence of CAR-T cells in the hostile solid tumor microenvironment, manufacturing cost and toxicity management remain the central obstacles that any GPC-3/AXL dual-target program would need to address in preclinical and clinical testing.

The study’s limitations are equally clear: it was a single-center, retrospective analysis with a relatively modest sample size and short follow-up, and it found no correlation between either protein and most clinicopathological features such as cirrhosis, macrovascular invasion, tumor size, BCLC stage, recurrence or mortality. Yet the core result stands on firm ground. Nearly 94 percent of hepatocellular carcinomas in this cohort displayed at least one of the two antigens, and more than two-thirds displayed both, making GPC-3 and AXL together a coverage profile that few single or paired targets in liver cancer can match. As immunotherapy increasingly becomes a pillar of hepatocellular carcinoma treatment alongside surgery, locoregional and systemic options, this co-expression map offers exactly the kind of evidence base needed to move dual-target CAR-T cells for liver cancer from concept toward the clinic.

Subject of Research: Co-expression of Glypican-3 and AXL in hepatocellular carcinoma and their potential as dual targets for CAR-T cell immunotherapy

Article Title: GPC-3 and AXL are essential factors and therapeutic targets for hepatocellular carcinoma

Article References: Qiu, C., Shi, J., Yan, S., Zhu, C., Zhang, S., Feng, Y., Zhang, X., Huang, X., & Xu, J. (2025). GPC-3 and AXL are essential factors and therapeutic targets for hepatocellular carcinoma. Clinical Cancer Bulletin, 4(1), Article 22. https://doi.org/10.1007/s44272-025-00051-1

Image Credits: AI Generated

DOI: 10.1007/s44272-025-00051-1

Keywords: hepatocellular carcinoma, Glypican-3, AXL, CAR-T cell therapy, immunotherapy, biomarkers, liver cancer, targeted therapy, immunohistochemistry, receptor tyrosine kinase, oncofetal protein, dual-target therapy

What makes natural killer cells so attractive as an anti-cancer weapon is their biology. Unlike cytotoxic T cells, which must be primed against a specific antigen before they can attack, NK cells are hardwired to detect and destroy transformed tissue on contact, without prior sensitization. They patrol the body constantly, integrating activating and inhibitory signals through a sophisticated array of germline-encoded receptors. When a cell loses the surface expression of major histocompatibility complex class I molecules, a common immune-evasion trick employed by tumors, the inhibitory brakes on NK cells are released and the killer cell delivers its lethal hit through the polarized release of lytic granules containing perforin and granzymes. NK cells also deploy activating receptors such as NKG2D and the natural cytotoxicity receptors NKp30, NKp44, and NKp46, which recognize stress-induced ligands upregulated on malignant cells. Beyond direct killing, they shape the broader immune response by secreting interferon-gamma and tumor necrosis factor alpha, which in turn promote the maturation of dendritic cells and support T cell priming, making them both executioners and orchestrators of antitumor immunity.

The liver, however, is a uniquely hostile arena for these cells. The review devotes considerable attention to the hepatocellular carcinoma tumor microenvironment, describing it as a formidable barrier that systematically disables NK cell function. Chronic exposure to transforming growth factor beta, an immunosuppressive cytokine abundantly produced in cirrhotic and tumor-bearing livers, downregulates activating receptors and blunts cytotoxicity. Persistent hypoxia, a hallmark of rapidly growing tumors with disorganized vasculature, further erodes NK cell metabolism and effector function. Perhaps most strikingly, the review highlights metabolic exhaustion driven by lipid accumulation: tumor cells and their supporting stroma flood the microenvironment with lactic acid and lipids, reprogramming NK cell metabolism and starving the cells of the glycolytic capacity they need to mount an effective attack. Studies cited in the review show that lactate dehydrogenase-associated lactic acid production directly blunts tumor immunosurveillance by both T and NK cells, and that fructose-1,6-bisphosphatase-induced inhibition of glycolysis cripples NK cell function during tumor progression.

The clinical evidence of NK cell dysfunction in liver cancer is equally sobering. Single-cell sequencing studies of human liver tumors have mapped the landscape of infiltrating immune cells and revealed that tumor-resident NK cells are phenotypically and functionally compromised. Tissue-resident CD56-bright NK cells enriched in the adult human liver, characterized by high Eomes and low T-bet expression, show reduced proinflammatory potential compared with their blood counterparts. In patients with hepatocellular carcinoma, myeloid-derived suppressor cells inhibit NK cell activity through the NKp30 receptor, and elevated levels of soluble NKG2D ligands and immunomodulatory molecules such as DKK3 further dampen antitumor responses. Exhaustion markers tell a similar story: NK cells expressing the inhibitory receptors TIGIT, TIM-3, CD96, and programmed death 1 correlate with disease progression in hepatitis B virus-related hepatocellular carcinoma, and high CD96 expression on NK cells predicts poorer prognosis. Downregulation of the PR/SET domain 10 protein has also been implicated as an underlying mechanism of NK cell dysfunction in these patients.

Against this backdrop, the review maps out the therapeutic strategies now being pursued to restore or supercharge NK cell activity. The most straightforward approach is cytokine-supported adoptive transfer: NK cells harvested from patients or healthy donors are expanded ex vivo using cytokines such as interleukin-15 and interleukin-21, then infused back in large numbers. Interleukin-15 has shown particular promise, with experimental work demonstrating that it can overcome hepatocellular carcinoma-induced NK cell dysfunction. The foundational proof of concept came from early clinical studies showing successful adoptive transfer and in vivo expansion of haploidentical NK cells in patients with cancer, and more recent trials have extended the approach to liver cancer specifically. A phase I trial of NK cell infusion in patients with recurrent hepatocellular carcinoma after liver transplantation reported that the treatment was well tolerated and showed preliminary efficacy, while an open-label phase I study protocol is testing adoptive immunotherapy with NK cells derived from peripheral blood CD34-positive stem cells to prevent recurrence after curative hepatectomy.

Genetic engineering has pushed the field considerably further. Chimeric antigen receptor-engineered NK cells, or CAR-NK cells, are equipped with synthetic receptors that redirect their killing capacity toward tumor-associated antigens. In hepatocellular carcinoma, the leading targets are glypican-3, or GPC3, a cell surface proteoglycan abundantly expressed on malignant hepatocytes, and CD147, a transmembrane glycoprotein implicated in tumor invasion. GPC3-specific CAR-NK cells have been developed and validated in preclinical models, and anti-CD147 chimeric antigen receptors have demonstrated efficacy against hepatocellular carcinoma cells. A first-in-human phase 1 trial of induced pluripotent stem cell-derived CD19-directed CAR-NK cells in B-cell lymphoma has provided an important clinical template for the broader platform, demonstrating that off-the-shelf, genetically defined NK cell products can be manufactured and safely administered. The review notes, however, that head-to-head comparisons suggest CAR-T cells can outperform CAR-NK cells in some CAR-mediated effector functions, underscoring that the technology still has room to mature.

A parallel and rapidly evolving strategy involves bispecific and trispecific killer cell engagers, known as BiKEs and TriKEs. These antibody-like molecules physically bridge NK cells to tumor cells while simultaneously delivering activating signals. Trispecific killer engagers incorporating an interleukin-15 linker have been shown to direct NK cells toward CD33-positive targets while inducing persistence, in vivo expansion, and enhanced function. Tri-specific engagers targeting mesothelin have been used to steer NK cells toward lung cancer, and the platform is now being applied to liver cancer: a GPC3-targeted NK engager bispecific antibody has entered a phase 1/2a clinical trial in patients with hepatocellular carcinoma, and novel NKp46-based engagers targeting GPC3 have undergone preclinical characterization. Recent work has also produced therapeutic bispecific antibodies targeting CD16A on NK cells and GPC3 on tumor cells, and tetravalent bispecific engagers designed to activate NK cells against cancer without mediating fratricide, a key safety consideration when the targeting antigen is also expressed on the therapeutic cells themselves.

Combination strategies form another pillar of the review. The multikinase inhibitor sorafenib, a longstanding standard of care in advanced hepatocellular carcinoma, has been shown to perpetuate cellular anticancer effector functions by modulating crosstalk between macrophages and NK cells, and preclinical studies of sorafenib combined with memory-like NK cells have reported encouraging antitumor activity in models of the disease. Xenograft experiments mapping the signaling pathways engaged by combined sorafenib and NK cell therapy have reinforced the mechanistic rationale. Lenvatinib, a newer tyrosine kinase inhibitor, appears to activate antitumor immunity by suppressing immunoinhibitory infiltrates in the tumor microenvironment, creating a permissive setting for adoptive cell therapy. Immune checkpoint blockade is also being explored in combination with NK cell approaches, with single-cell RNA sequencing in humanized mice revealing antitumor potency of NK and CD8-positive T cells when PD-1 and TIGIT are co-targeted. Even oncolytic viruses have entered the picture, with an engineered measles virus shown to enhance the antitumor responses of adoptively transferred NKG2D-positive immune cells in hepatocellular carcinoma models.

The review is candid about the knowledge gaps that still separate laboratory promise from routine clinical benefit. The complex interplay between NK cells and the tumor microenvironment remains incompletely understood, particularly the roles of hepatic stellate cells and cancer-associated fibroblasts, which secrete chemokines such as CXCL5 that promote immunosuppressive signaling. The metabolic reprogramming of NK cells in the lipid-rich liver environment, the mechanosensory properties that govern NK cell function in dense tumor tissue, and the education and licensing processes that calibrate NK cell responsiveness all demand further study. Chronic alcohol consumption, a major risk factor for liver disease, has been shown to impair peripheral NK cell development by reducing interleukin-15 availability, adding another layer of complexity in a patient population where such exposures are common. The authors argue that resolving these questions is essential for optimizing NK cell immunotherapy and improving prognostic outcomes.

Nevertheless, the trajectory of the field is unmistakably upward. From the NK-92 cell line, whose three decades of use have shaped much of modern NK cell research, to iPSC-derived CAR-NK products and GPC3-directed engagers now in clinical trials, the toolkit for deploying natural killer cells against hepatocellular carcinoma has expanded dramatically in a short period. The review’s synthesis suggests that the coming years will be defined by rational combinations, pairing engineered NK cells with kinase inhibitors, checkpoint blockade, and metabolic interventions designed to neutralize the tumor microenvironment’s defenses. For a cancer that has long resisted immunotherapeutic advances and continues to claim hundreds of thousands of lives each year, the renaissance in NK cell biology may represent the most credible path yet toward durable, off-the-shelf immunotherapy for liver cancer, and the scientific foundation laid by this comprehensive analysis offers researchers a clear map of both the opportunities and the obstacles that lie ahead.

Subject of Research: Natural killer cell-based immunotherapy for hepatocellular carcinoma

Article Title: Natural killer cell-based therapy for hepatocellular carcinoma (HCC): A literature review

Article References: Widowati, W., Wargasetia, T. L., Sadri, B., Nainggolan, I. M., Azis, R., Rifana, S. D., Rismani, E., & Vosough, M. (2026). Natural killer cell-based therapy for hepatocellular carcinoma (HCC): A literature review. Medical Oncology, 43(11), Article 293. https://doi.org/10.1007/s12032-026-03393-5

Image Credits: AI Generated

DOI: 10.1007/s12032-026-03393-5

Keywords: natural killer cells, hepatocellular carcinoma, immunotherapy, CAR-NK cells, tumor microenvironment, glypican-3, bispecific killer cell engagers, interleukin-15, liver cancer, adoptive cell therapy, NK cell dysfunction, sorafenib

The product at the centre of the discussion, known as C-CAR031, consists of autologous T cells — the patient’s own immune cells, collected and modified outside the body — that express an affinity-tuned CAR directed against glypican 3. Affinity tuning is a critical design choice: because GPC3 is expressed at lower levels on some healthy tissues, the receptor’s binding strength is calibrated so the engineered cells preferentially recognise tumour cells bearing high densities of the antigen while sparing normal tissue. This balancing act between potency and safety has long been recognised as one of the central challenges of targeting solid-tumour antigens with CAR T cells.

What distinguishes C-CAR031 from earlier GPC3-targeted constructs is its armour. The engineered T cells are modified to secrete a dominant-negative transforming growth factor beta receptor II (dnTGFβRII), a truncated receptor that sits on the cell surface and soaks up TGFβ, an immunosuppressive cytokine that is abundant in the hepatocellular carcinoma tumour microenvironment. TGFβ normally suppresses T cell activity, proliferation and effector function, effectively disarming infiltrating immune cells. By expressing a dominant-negative receptor, the CAR T cells become resistant to this suppressive signal, allowing them to retain their killing capacity inside the tumour rather than being shut down on arrival.

The clinical evaluation included 36 patients with hepatocellular carcinoma, a cancer that remains one of the leading causes of cancer-related death worldwide and for which advanced-stage options are limited. According to Greten and Ho, the initial efficacy results were promising, providing some of the clearest signals yet that armoured CAR T cells can achieve meaningful anti-tumour activity in a solid organ malignancy. The commentary’s authors suggest that the strategy of overcoming TGFβ-mediated immunosuppression represents a rational and technically feasible way to improve CAR T cell function in liver tumours, where the fibrotic, inflamed and cytokine-rich microenvironment has historically blunted adoptive cell therapies.

Yet the enthusiasm is tempered by two persistent concerns. The first is antigen loss: cancers are genetically unstable populations, and tumour cells that downregulate or lose GPC3 expression can escape recognition entirely, leaving the infused T cells blind to the disease. The second is limited progression-free survival, indicating that even when responses are achieved, they may not be durable. These twin problems — immune escape at the level of the target antigen and insufficient persistence of the therapeutic cells — echo the experience of CAR T cell therapy across solid tumours more broadly, and they frame the central question posed by the commentary’s title: how much armouring is enough?

The field has been experimenting with a growing arsenal of engineering strategies to answer that question. One alternative approach, tested in a separate clinical study, involved CAR T cells armoured to secrete interleukin-15, a cytokine that supports T cell survival and proliferation. Other preclinical work has focused on generating persistent, polyfunctional GPC3-specific CAR T cells that eliminated orthotopic hepatocellular carcinomas in mouse models, demonstrating that sustained multi-cytokine function is achievable with the right design. Each armouring strategy addresses a different vulnerability — cytokine deprivation, suppressive signalling, exhaustion — and the emerging lesson is that no single modification is likely to be sufficient on its own.

Beyond the engineered cells themselves, Greten and Ho highlight the importance of the surrounding clinical and diagnostic infrastructure. Regulatory agencies, including the US Food and Drug Administration, have issued guidance on the development of CAR T cell products, reflecting the unique manufacturing and safety considerations these living drugs entail. Meanwhile, advances in molecular imaging — such as a first-in-human case series of a glypican-3-targeted diagnostic radiopharmaceutical — may eventually allow clinicians to non-invasively assess GPC3 expression in a patient’s tumour before committing to a GPC3-directed therapy, and to monitor antigen heterogeneity across lesions and over time. Such companion diagnostics could become essential for selecting patients most likely to benefit and for detecting the antigen loss that undermines responses.

Broader momentum in the solid-tumour CAR T field lends context to the hepatocellular carcinoma results. A randomised phase 2 trial of satri-cel, a claudin-18 isoform 2-specific CAR T cell therapy for previously treated advanced gastric or gastro-oesophageal junction cancer, demonstrated that solid-tumour CAR T products can compete with standard treatment options in controlled studies. Taken together with the GPC3 data, these findings suggest that the field is approaching an inflection point, in which rational target selection, affinity tuning and microenvironmental armouring converge to produce clinically meaningful outcomes in cancers once considered off-limits to cell therapy.

What comes next, the commentary argues, is a systematic effort to determine which combinations of engineering features deliver durable benefit without unacceptable toxicity. Potential directions include multi-antigen targeting to pre-empt antigen escape, logic-gated receptors that improve tumour specificity, and rational sequencing or combination with other immunotherapies and liver cancer treatments. The authors, who declare no competing interests, emphasise that bringing CAR T cells for solid tumours into routine clinical practice will require more than promising initial efficacy: it will demand durable responses, manufacturable and reproducible products, and a deeper understanding of how engineered cells behave in the uniquely immunosuppressive landscape of the liver. C-CAR031 offers a compelling proof of principle that armouring works; the task now is to determine how much is enough, and to keep building.

Subject of Research: Armoured glypican 3-targeted CAR T cell therapy for hepatocellular carcinoma

Article Title: Armouring GPC3 CAR T cells for hepatocellular carcinoma: how much is enough and what comes next?

Article References: Greten, T. F., & Ho, M. (2026). Armouring GPC3 CAR T cells for hepatocellular carcinoma: how much is enough and what comes next?. Nature Reviews Clinical Oncology. https://doi.org/10.1038/s41571-026-01206-2

Image Credits: AI Generated

DOI: 10.1038/s41571-026-01206-2

Keywords: CAR T cells, hepatocellular carcinoma, glypican 3, TGFβ, cancer immunotherapy, solid tumours, dominant-negative TGFβ receptor II, antigen loss, tumour microenvironment, C-CAR031, adoptive cell therapy, liver cancer