Umbilical cord blood natural killer cells improve anti-GD2 antibody efficacy in neuroblastoma: from mouse to human Authors: Mengjia Song, Yingxia Lan, Xiya Tan, Linnan Wu, Jia Zhu, Suying Lu, Feifei Sun, Min Gao, Junting Huang, Yi Que, Yu Zhang, Xiaona Fang, Mengzhen Li, Yumiao Mai, Juan Wang, Tong Xiang, Yizhuo Zhang Published: 2026 ABSTRACT AbstractDespite advances in anti-disialoganglioside (GD2) immunotherapy, high-risk neuroblastoma (NB) remains a clinical challenge. Natural killer (NK) cell-mediated antibody-dependent cellular cytotoxicity is a potent mechanism of action of anti-GD2 antibody. However, dose-intensive chemotherapy causes NK cell depletion. Here, a positive association between intratumoral NK cell infiltration and anti-GD2 efficacy was first observed in clinical samples. We then established the unique advantages of ex vivo expanded and activated umbilical cord blood (UCB)-derived NK cells in terms of cytotoxicity, persistence, exhaustion resistance, and safety from preclinical models to clinical cases compared with patient-derived PB NK cells. Anti-GD2 antibody further enhanced the superiority of UCB NK cells in terms of their functional status, persistence, and capacity to remodel an immune-activated tumor microenvironment. Combined treatment with UCB NK cells and an anti-GD2 led to synergistic effects in vitro and in mice and achieved complete and partial responses in two patients with relapsed/refractory NB, with no additive toxicity. Mechanistically, the anti-GD2 antibody induced a “high activating-low inhibitory” phenotype in UCB NK cells, and donor UCB NK cells showed dynamically lower levels of immune checkpoints but higher levels of memory-like markers than those of recipient NK cells when combined with anti-GD2 therapy. Collectively, this study is the first translational investigation of UCB NK cells combined with anti-GD2 therapy in NB, bridging preclinical mechanistic insights into an early clinical proof-of-concept. A phase I study of UCB NK cell infusion combined with anti-GD2 therapy in children with high-risk, relapsed/refractory NB is ongoing at our center (NCT06631391).Keywords: Neuroblastoma, umbilical cord blood natural killer cells, anti-GD2 therapy, combined immunotherapy, tumor microenvironment Introduction Neuroblastoma (NB) is the most common extracranial solid tumor in childhood, accounting for 15% of childhood cancer-related deaths. 1 , 2 Approximately half of the patients have high-risk disease at diagnosis. 1 For patients in whom first-line therapy fails, their prognosis is poor, with a long-term survival rate of less than 10%. 3 Over the past few decades, immunotherapy with an anti-disialoganglioside (GD2) monoclonal antibody (Ab) has been implemented as the standard treatment regimen and has substantially improved patient outcomes. 4 , 5 However, approximately 60% survival in patients with high-risk NB remains poor, suggesting the urgent need to develop novel and effective immunotherapeutic approaches. GD2 belongs to a unique class of carbohydrate tumor-associated antigens expressed at high densities in primary and metastatic NB cells. 6 Binding of the terminal pentasaccharide of the GD2 molecule by anti-GD2 Ab can induce Fcγ receptor-mediated Ab-dependent cellular cytotoxicity (ADCC) and complement-mediated cytotoxicity (CDC) via natural killer (NK) cells, 7 , 8 neutrophils, 9 and macrophages. 10 NK cells played an important role in the antitumor immunity of NB. 11 , 12 However, pediatric patients display low NK cell frequencies in peripheral blood (PB) and tumor microenvironment (TME), 13 and dose-intensive chemotherapy severely compromise NK cell immunity in patients with NB. 11 Additional barriers such as reduced FCGR3A genotype 14 and suppressive TME further derail the ADCC mediated by NK cells. 11 Numerous efforts have been devoted to combining ex vivo expanded NK cell infusion with anti-GD2 therapy. 15-17 Nevertheless, the clinical application of haploidentical NK cells has been hindered by human leukocyte antigen (HLA) rejection, insufficient expansion, and transient persistence post-infusion in our group 18 and other colleagues. 16 , 17 , 19 Autologous NK cells derived from patients experience irreversible functional damage because of prior cytotoxic therapies. 11 , 17 Recently, chimeric antigen receptor T-cell and NK T-cell immunotherapies targeting GD2 have demonstrated potent antitumor properties in NB, but pose substantial safety risk. 20 , 21 Given these obstacles, we focused on NK cells derived from umbilical cord blood (UCB NK cells). 22 Expanded UCB NK cells not only have excellent cellular vitality and cytokine-secreting capacity 23 , 24 but also express high-level C-X-C motif chemokine receptor 4, which may facilitate their homing to tumors. 25 Therefore, it is logical to assume that combining the anti-GD2 Ab with expanded UCB NK cells may result in potent antitumor effects in patients with NB. In the present study, we expanded and activated NK cells from the UCB of donors and the PB of patients with NB. We further compared the antitumor activity, phenotypic features, and safety of UCB and PB NK cells with or without the anti-GD2 Ab. Moreover, this study presents clinical evidence of the adoptive transfer of UCB NK and anti-GD2 Ab in two cases of relapsed/refractory NB. Materials and methods Clinical trial registration and study design This phase I clinical study was conducted at Sun Yat-sen University Cancer Center, Guangzhou, China. This study was registered at ClinicalTrials.gov (Registration Number: NCT06631391) and started in December 2024, with primary completion expected in December 2026 and follow-up anticipated to be completed in December 2029. The inclusion criteria included: age ≤18 y; Karnofsky (≥16 y old) or Lansky (<16 y old) physical status score of at least 50; patients diagnosed with high-risk, recurrent/refractory NB according to clinical diagnostic criteria, who have undergone comprehensive treatment; expected survival period of at least 12 weeks; fully recovered from acute toxic effects of prior anticancer therapy. The exclusion criteria included: symptomatic brain metastasis (stable for >2 months post-treatment, cranial imaging confirming no cerebral hemorrhage); Grade II or higher myocardial ischemia, myocardial infarction, poorly controlled arrhythmias, New York Heart Association class III-IV heart failure, left ventricular ejection fraction below 50%; interstitial lung disease history or current condition; coagulation disorders with a bleeding tendency or thrombolytic/anticoagulant therapy; and arterial/venous thromboembolism within 12 months prior to enrollment. This study was approved by the Ethics Committee of Sun Yat-sen University Cancer Center (Approval Number: SL-B2024-008-02). Informed consent was obtained per age-specific ethical guidelines: For children aged ≥ 8 y, written informed consent was obtained from legal guardians and the children. For children aged < 8 y, written informed consent was provided by legal guardians. Meanwhile, the study details were explained to the children in age-appropriate language, with verbal acknowledgment of understanding. All consent and assent procedures were conducted without coercion, and participants or their guardians were informed of the right to withdraw from the study at any time without penalty. Treatment protocol Patients received UCB NK cell infusions (100 million/kg): 1st–4th weekly, 5th–8th every 2 weeks. For anti-GD2 therapy, some patients received a naxitamab beta-based regimen: cycle 1 included naxitamab (3 mg/kg/d on days 1, 3, and 5) and granulocyte‒macrophage colony‒stimulating factor (5 mg/kg/d from days −4 to day 0 and 10 mg/kg/d from days 1 to 5) once every 3 weeks for 5 cycles. Some patients received a dinutuximab-based regimen: cycle 1 included dinutuximab beta (10 mg/m2/d from days 0 to 9) and granulocyte‒macrophage colony‒stimulating factor (250 mg/m2/d from days −3 to 10) once every 3 weeks for a total of 5 cycles. Safety and clinical response were assessed from day 84, with follow-up until September 2025. Diagnosis and histology were determined using the International Neuroblastoma Risk Group Staging System and International Neuroblastoma Pathology Classification, respectively. Risk stratification was performed according to the Children’s Oncology Group Risk Group System. For case 2 aged 4 y old, written informed consent for study participation and for publication of de-identified clinical data was obtained from the parents; the child received simplified explanations and provided verbal acknowledgment. For case 1 aged 8 y old, written informed consent for study participation and for the publication of de-identified clinical data was obtained from the child and parents. Animal studies All the animal experiments were approved by the Institutional Animal Care and Use Committee of Sun Yat-sen University Cancer Center (Approval Number: L102012023080P) and conducted in compliance with the American Veterinary Medical Association Guidelines. Sample sizes were determined based on preliminary experiments 12 and similar published studies, 26 , 27 and no formal power calculation was performed. All the mice were purchased from Vital River, China. and maintained under specific pathogen-free conditions. All animal studies were blinded during data measurement. Euthanasia was performed using pentobarbital sodium administered via intraperitoneal injection at a dose of 150 mg/kg body weight. Prior to and following injection, anesthesia depth was monitored by assessing loss of the righting reflex, absence of the pedal withdrawal reflex, and cessation of the corneal reflex. Euthanasia was confirmed by permanent loss of spontaneous respiration and heartbeat, with no recovery observed within 5 minutes. All procedures were executed by researchers trained by the Institutional Animal Care and Use Committee. Four-week-old NOG male mice were used for all in vivo experiments. 1 × 106 Fluc-Kelly cells suspended in 20 µL PBS were surgically injected beneath the renal capsule of the left kidneys of a total of 30 mice at day −7. The mice were randomly divided into 6 groups, with 5 mice per group at day 0. From day 1, the mice were treated twice weekly for 2 weeks with 1 × 107 UCB NK or PB NK cells (intravenous, 4 times in total) and IL-2 (20,000 unit/mouse, intraperitoneal injection) every other day for 2 weeks (7 times in total) alone or in combination with the anti-GD2 Ab (60 µg/mouse, every other day for 5 times in total). Tumor growth was assessed weekly by bioluminescence imaging using the IVIS-Spectrum system and Living Image 3.0 software (PerkinElmer, USA). The experimental endpoints for the mice included lethargy, poor grooming, weight loss reaching 15%–20% of the baseline, and/or hind-limb paralysis according to the Institutional Animal Care and Use Committee criteria. The body weight at day 1, the day of the initial treatment, served as the baseline. All the mice were euthanized after a maximum observation period of 50 d after administration. No formal power calculation was performed for determining the animal sample sizes. Statistical analyses Statistical analyses were performed using the SPSS version 20.0 (SPSS, Chicago, IL, USA) and GraphPad Prism version 8.0. The χ2 test was used to assess the relationships between NKp46 and clinicopathological characteristics. Survival analysis was performed using Kaplan–Meier survival curves, and differences in survival times among subgroups were analyzed using the log-rank test. Univariate and multivariate Cox regression analyses were performed to explore the associations between the characteristics and NKp46 with the outcomes. One-way analysis of variance and an unpaired t-test were used to determine the statistical significance between experimental groups. p < 0.05 was considered statistically significant, *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001. Results NK cell infiltration is correlated with good response to anti-GD2 therapy and increased survival in patients with NB We first assessed NK cell infiltration by NKp46 staining in the tumor tissues of patients with NB (Figure 1A and B) and found that patients who achieved a complete response (CR) following GD2 Ab administration showed significant NK cell infiltration compared with those with a non-CR (Figure 1C). High NK cell infiltration tended to correlate with increased event-free survival (Figure 1D). In a cohort of 131 patients with NB (Supplementary Table S1), high NK cell infiltration was associated with favorable event-free and overall survival (Figure 1E and F). Univariate and multivariate Cox regression analyses demonstrated that high NK cell infiltration was an independent prognostic factor for event-free and overall survival in patients with NB (Table 1). Association of NKp46 expression with the event-free survival and overall survival in clinical NB patients. (A) Representative photomicrographs of low or high expression levels of NKp46 by immunohistochemistry staining in tumor tissues of patients with NB. (B) The workflow showing ImageJ analysis for the quantification of NKp46 immunohistochemistry staining. (C) The proportion of NKp46 low or high expression in patients with NB who achieved a complete response or non-complete response after receiving anti-GD2 antibody treatment (n = 27); (D) Kaplan‒Meier survival curves of event-free survival in patients with NB who received anti-GD2 antibody treatment stratified according to different NKp46 expression (low expression, n = 13; high expression, n = 14); (E and F) Kaplan‒Meier survival curves of event-free survival and overall survival in patients with neuroblastoma (n = 131) stratified according to different NKp46 expression (low expression, n = 65; high expression, n = 66). NB, neuroblastoma. Table 1.Univariate and multivariate cox regression analysis for event-free survival and overall survival among 131 patients with neuroblastoma.ParametersEvent-free survivalOverall survivalUnivariate analysisMultivariate analysisUnivariate analysisMultivariate analysisHR (95% CI) p HR (95% CI) p HR (95% CI) p HR (95% CI) p Gender MaleReference–––Reference–––Female0.967 (0.598–1.566)0.893––0.643 (0.323–1.278)0.207–– Age <18 monthsReference–––Reference–––≥18 months1.079 (0.602–1.933)0.799––0.811 (0.373–1.761)0.596–– MYCN status Non-amplifiedReference–Reference–Reference–Reference–Amplified2.482 (1.327–4.644)0.0041.600(0.832–3.078)0.1592.619 (1.199–5.719)0.0161.635 (0.716–3.737)0.244Unknown1.079 (0.618–1.883)0.7900.774(0.436–1.374)0.3811.203 (0.578–2.507)0.6210.759(0.338–1.706)0.505 COG risk group Low riskReference–Reference–Reference–Reference–Intermediate risk2.732 (1.108–6.733)0.0292.790(1.130–6.884)0.0261.041 (0.259–4.185)0.9551.114(0.274–4.525)0.880High risk4.061 (2.005–8.227)0.0004.222(2.037–8.750)0.0002.608 (1.093–6.226)0.0312.620(1.039–6.609)0.041 NKp46 NegativeReference–Reference–Reference–Reference–Positive0.541 (0.340–0.861)0.0100.535(0.333–0.859)0.0100.426 (0.217–0.839)0.0140.451(0.227–0.899)0.024Note: COG, Children's Oncology Group. Univariate and multivariate cox regression analysis for event-free survival and overall survival among 131 patients with neuroblastoma. Note: COG, Children's Oncology Group. UCB NK cells mediated stronger ADCC against NB cells than PB NK cells under the action of anti-GD2 Ab Next, the GD2-positive Kelly cell line was selected for subsequent experiments (Supplementary Figure S1A). The fluorescently labeled anti-GD2 Ab was almost completely bound to NB cells (Supplementary Figure S1B). GD2 antigens to the surface of NB cells increased in a concentration-dependent manner (Supplementary Figure S1C). A low dose of 2.5 µg/mL anti-GD2 Ab alone showed significant cytotoxicity to NB cells, which gradually increased with increasing anti-GD2 Ab concentration (Supplementary Figure S1D and E). Therefore, we selected a low dose of <2.5 µg/mL to exclude the direct cytotoxicity of the anti-GD2 Ab against NB cells in the subsequent NK-mediated ADCC assay. UCB-derived and PB-derived NK cells were expanded and activated in vitro, with a CD3-CD56+ cell proportion of over 90% and CD16 expression of over 90% (Supplementary Figure S2A and B). UCB NK cells exhibited stronger cytotoxicity than PB NK cells against both GD2-positive and GD2-negative NB cells. Notably, the addition of the anti-GD2 Ab significantly enhanced the cytotoxicity of both NK cell populations selectively against GD2-positive NB cells but not against GD2-negative NB cells (Figure 2A and B, Supplementary Figure S3A, Supplementary Movies 1–6). xCELLigence real-time cell analysis showed that both GD2-positive and GD2-negative NB cells were significantly killed upon the addition of UCB and PB NK cells, with cytotoxicity further increasing as the effector-to-target (E:T) ratio increased (Figure 2C and D, Supplementary Figure S3B and C). A stronger ADCC effect of UCB and PB NK cells selectively against GD2-positive NB cells was observed when combined with an anti-GD2 Ab. Collectively, these results indicated that UCB NK cells demonstrated a greater killing capacity than PB NK cells and that the ADCC effect of anti-GD2 Ab mediated by UCB NK cells was also greater. Anti-GD2 antibody mediates enhanced UCB NK cellular cytotoxicity against NB cells compared to PB NK cells. (A–D) PB NK cells from three patients with high-risk NB and UCB NK cells were expanded and activated, respectively. Representative immunofluorescence images of PB NK cells or UCB NK cells co-cultured with GD2-positive Kelly cells and in the presence and absence of anti-GD2 antibody (1 µg/mL) at 0, 4, and 8 h. Effector: target ratio = 1:1. Kelly cells was transduced with green fluorescent protein (green), NK cells was stained with CellTracker dye (red) (A); apoptosis analysis of GD2-positive Kelly cells and SK-N-BE (2) cells co-cultured with PB NK cells or UCB NK cells in the presence and absence of an anti-GD2 antibody (1 µg/mL) for 2 h. Effector:target ratio = 1:1 (B); analysis of the killing effect of PB NK cells from three patients with high-risk NB and UCB NK cells on GD2-positive Kelly and SK-N-BE (2) cells using xCELLigence real-time cytotoxicity assay at an effector:target ratio of 1:1 or 5:1. Normalized cell index of Kelly and SK-N-BE (2) at 95 h in c is shown as statistical graphs (c and d). Comparisons were determined by one-way ANOVA (B–D). The bars are the mean ± SD (*p < 0.05, **p < 0.01, ****p < 0.0001, ns, not significant; NB, neuroblastoma; PB, peripheral blood; UCB, umbilical cord blood). UCB NK cells display higher expression levels of cytotoxic molecules than PB NK cells, which is further enhanced upon the addition of anti-GD2 Ab The expression levels of CD107a, perforin, granzyme B, interferon-gamma (IFN-γ) and tumor necrosis factor (TNF-α) in UCB NK cells were significantly higher than those in PB NK cells (Figure 3A and B, Supplementary Figure S4). Upon co-culture with tumor cells, CD107a expression in UCB and PB NK cells increased, whereas the expression of perforin, granzyme B, IFN-γ, and TNF-α were inhibited (Figure 3A and B, Supplementary Figures S4 and S5). NK cells seem to exert cytotoxicity following direct contact with NB cells, but their function might be concomitantly suppressed. Notably, the expression of all these molecules significantly increased after the addition of the GD2 Ab; however, the increase in PB NK cells was less pronounced than that in UCB NK cells (Figure 3A and B, Supplementary Figures S4 and S5). These data demonstrate that the functional suppression of UCB NK cells mediated by NB cells can be effectively reversed by the anti-GD2 Ab, whereas the function of PB NK cells was not fully restored, even when combined with the anti-GD2 Ab. Anti-GD2 antibody mediates enhanced production of cytotoxic factors, increased expression of surface-activating receptors and decreased expression of surface-inhibiting receptors by UCB NK cells compared to PB NK cells. (A–D) Ex vivo expanded PB NK cells from three patients with high-risk NB or UCB NK cells directly co-cultured with Kelly cells in the presence and absence of anti-GD2 antibody (1 µg/mL) for 4 h. Effector: target ratio = 2:1. Flow cytometry analysis was performed for the expression of CD107a, granzyme B, perforin (A), IFN-γ, TNF-α (B), NKp30, NKp44, NKG2D (C), CD158, NKG2A, and ILT2 (D) in PB NK cells or UCB NK cells. The frequency of positive cells is shown as statistical graph (right). Comparisons were determined by one-way ANOVA (A–D). The bars are the mean ± SD (*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001, ns, not significant; PB, peripheral blood; UCB, umbilical cord blood). Anti-GD2 Ab mediates a more significant “high activating-low inhibitory” receptor phenotype of UCB NK cells compared with PB NK cells The functional potency of NK cells depends on the balance between their surface activating and inhibitory receptors. 28 For the activating receptors (Figure 3C, Supplementary Figure S6A–C), the expression levels of NKp30, NKp44, and NKG2D in UCB NK cells were higher than those in PB NK cells both before and after co-culture with NB cells. Upon co-culture with NB cells, NKp30 expression in UCB NK cells significantly increased, and a further increase was observed following the addition of the anti-GD2 Ab; however, this phenomenon was not observed in PB NK cells. Meanwhile, anti-GD2 Ab led to a significant enhancement of NKp44 expression in UCB NK cells co-cultured with NB cells compared with those in UCB NK cells alone, whereas such an increase in PB NK cells was negligible. NKG2D expression in both UCB and PB NK cells was increased by the addition of the anti-GD2 Ab during co-culture with NB cells. Regarding the inhibitory receptors (Figure 3D, Supplementary Figure S6D–F), UCB NK cells exhibited higher expression of CD158 and NKG2A but lower expression of ILT2 than PB NK cells. The expression levels of these inhibitory receptors slightly decreased when NK cells were co-cultured with NB cells, and a further decrease was observed following the addition of the anti-GD2 Ab. Taken together, these results suggest that the addition of anti-GD2 Ab contributes to elevated levels of activating receptors but reduces the levels of inhibitory receptors in UCB and PB NK cells. Notably, UCB NK cells exhibited a more significant “high activating-low inhibitory” phenotype and more robust response to anti-GD2 Ab than PB NK cells. Adoptive transfer of UCB NK cells improves anti-GD2 Ab efficacy in mouse model of NB To evaluate the antitumor efficacy and safety of UCB NK cells in vivo, we designed a subcutaneous xenograft NB model using NOG mice (Supplementary Figure S7). Mice receiving UCB NK cell infusion exhibited slower tumor growth and prolonged survival time. There was no difference in body weight between the mice in the two groups, and no pathological damage was observed in the lungs, liver, kidneys, spleen, or heart of the mice receiving the UCB NK cell infusion. In addition, a high DNA copy number of the infused UCB NK cells was detected in crucial organs and injection sites. UCB NK cells were detectable in the PB, tumor tissue, and injection site until 20 d post-infusion. Next, an orthotopic NB model was designed to investigate the therapeutic effect of UCB or PB NK cell infusion in combination with an anti-GD2 Ab (Figure 4A). As shown in Figure 4B and C, PB NK cells alone moderately inhibited tumor growth, whereas UCB NK cells alone severely inhibited tumor growth compared with the control group. Mice treated with either PB NK cells or UCB-NK cell infusion combined with anti-GD2 Ab showed a significant reduction in tumor burden compared with mice treated with single treatment. Notably, the combination treatment with UCB NK cells and anti-GD2 Ab significantly improved survival and led to sustained complete regression in all five mice until day 50 post-treatment (Figure 4D, Supplementary Table S2), suggesting a great benefit of UCB NK cell infusion, particularly when combined with anti-GD2 Ab. None of the treatments resulted in changes in body weight or other toxicities (Supplementary Figure S8A and Supplementary Table S3). Anti-tumor effect of the anti-GD2 antibody combined with PB NK cells or UCB NK cells in vivo. (A) Schematic diagram of the experimental protocol of the orthotopic model; (B) bioluminescence images of different groups of mice in A during treatment on day 0, 7, 14 and 28 after tumor cell implantation; (C) bioluminescence of tumor growth for the mice in A at the indicated time points. (D) Kaplan‒Meier survival curves for overall survival of the mice in the different groups in A; (E‒H) mice were euthanized at day 5 post the last NK cell infusion, and simultaneously fresh whole blood and tumor samples were collected. Flow cytometry analysis of the absolute CD3-CD56+ NK cells number in different groups of mice in peripheral blood in subcutaneous model (E); flow cytometry analysis of the ratio of CD45+CD3−CD56+ NK cells relative to total cells in the single-cell suspension of mouse tumor tissues of different groups in the subcutaneous model (F); immunohistochemistry analysis for the granzyme B in the mouse tumors of different groups in the subcutaneous model (G); gene ontology enrichment analysis based on the RNA-sequencing data of mouse tumors in the subcutaneous model (H). Bars are the mean ± SD bars are the mean ± SD (*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001, ns, not significant; CR, complete response; PB, peripheral blood; UCB, umbilical cord blood). The number of circulating NK cells was dramatically higher in the group of combined UCB NK cells and anti-GD2 Ab treatment compared with other groups in another subcutaneous xenograft NB model (Figure 4E). Intratumoral NK cell infiltration (Figure 4F) and the expression of granzyme B in tumor tissue (Figure 4G) were also significantly increased. As shown in Figure 4H and Supplementary Figure S6B, gene set enrichment analysis of the RNA-sequencing data revealed that combination treatment with UCB NK cells and anti-GD2 Ab had a strong enrichment of several immune activation signatures and a strongly negative correlation with NK cell tolerance induction (Figure 4H) and the tumor growth factor (TGF) beta signaling pathway (Supplementary Figure S8B). Taken together, these data indicate that UCB-NK cell infusion effectively improved the efficacy of the anti-GD2 Ab and activated the TME in mice. We also found moderate enrichment of programmed death-1 (PD-1)/programmed death-ligand 1 (PD-L1) signaling in xenografts derived from the group receiving UCB NK cells combined with the anti-GD2 Ab. PD-1 and PD-L1 expressions on UCB-NK cells were upregulated after co-culturing with NB cells, and PD-L1 expression on the surface of tumor cells was significantly increased in vitro (Supplementary Figure S9A–C). The anti-PD-1/PD-L1 Ab slightly enhanced the tumor cell-killing ability of UCB NK cells combined with the anti-GD2 Ab, with no statistical significance (Supplementary Figure S9D). Cytotoxicity of UCB NK cells combined with anti-GD2 Ab against primary NB cells We detected the expression of GD2 in primary NB cells isolated from 3 patients with high-risk NB, with inherent GD2 expression and a positive rate over 80% (Figure 5A). UCB NK cells exhibited potent cytotoxicity against primary NB cells, and higher GD2 expression by tumor cells correlated with greater cytotoxic superiority of UCB NK cells in combination with anti-GD2 Ab (Figure 5B and C). Anti-GD2 antibody combined with UCB NK cells exhibit robust killing effect on primary NB cells. (A–C) Primary GD2-positive NB cells were isolated from three patients with high-risk NB. Flow cytometry analysis was performed for GD2 expression (A). Apoptosis analysis of patient primary GD2-positive NB cells co-cultured with UCB NK cells in the presence and absence of anti-GD2 antibody (1 µg/mL) for 2 h (B and C). Effector:target ratio = 1:1. Bars are the mean ± SD (*p < 0.05, ns, not significant; NB, neuroblastoma; UCB, umbilical cord blood). Combined UCB NK cell infusion and anti-GD2 Ab is safe and may offer clinical benefit in patients with relapsed/refractory NB To evaluate the clinical efficacy of UCB NK cell infusion and its combination with anti-GD2 Ab, we conducted a clinical trial in patients with high-risk, relapsed/refractory NB (NCT06631391). A total of 8 patients were enrolled between December 2024 and April 2025 and included in the analysis (Supplementary Table S4). All 8 patients had measurable evaluable lesions at baseline. The median follow-up time was 14.0 (range, 7.7–16.4) months. Treatment-related safety was favorable in all 8 patients, the most common adverse events were grade 1–2 fever and fatigue, and no grade 3–4 toxicities or treatment-related serious adverse events occurred. No disease progression events were observed in any patient, and the median EFS time was 14.0 (range, 7.7–16.4) months. The best overall response included 3 cases of CR, 4 cases of partial response (PR), and 1 case of stable disease. The CR rate was 37.5% (3/8), the objective response rate was 87.5% (7/8), and the disease control rate was 100% (8/8). Two representative cases of relapsed/refractory NB were presented herein. Case 1 was an 8-y-old boy initially diagnosed with low-risk NB originating from the cervical region, with MYCN non-amplification and an 11q deletion. He achieved CR following first-line treatment, which did not include anti-GD2 therapy. However, the patient relapsed in the bone marrow and multiple bone sites 18 months after treatment discontinuation. Despite four subsequent lines of chemotherapy, residual lesions persisted in the bone marrow and right femur, indicating resistance to chemotherapy. The patient then received UCB NK cell infusion combined with naxitamab-based anti-GD2 therapy (Figure 6A), achieving an overall partial response (PR), with CR in the bone marrow and PR in the lesions of the right femur (Figure 6B and C). The patient is currently receiving post-maintenance treatment and has maintained continuous PR until the last follow-up. Combined UCB NK cell infusion and anti-GD2 therapy achieves good clinical efficacy in two patients with relapsed/refractory NB. (A) Treatment protocol; (B) comparison of case 1's 123I-Meta-Iodobenzylguanidine scintigraphy images showing the lesions in the right femur before and after combined UCB NK cell infusion and anti-GD2 therapy. The red arrows indicate the tumor lesions; (C) comparison of case 1's bone marrow cytological smear images before and after combined UCB NK cell infusion and anti-GD2 therapy. Red arrows indicate tumor cells; (D) comparison of case 2's 123I-meta-iodobenzylguanidine scintigraphy images before combined UCB NK cell infusion and anti-GD2 therapy and 18F-meta-fluorobenzylguanidine positron emission tomography/computed tomography scan images after combined UCB NK cell infusion and anti-GD2 therapy, showing the lesions in the skull. Red arrows indicate the tumor lesions; (E) the percentage of donor UCB NK cells in total NK cells from the peripheral blood of case 1 at the indicated time points following the first single infusion was assessed by DNA chimerism assay; (F) the relative frequencies of donor UCB NK cells and recipient NK cells from the peripheral blood of case 2 at the indicated time points following the first single infusion were assessed by flow cytometry analysis, and donor UCB NK cells and recipient NK cells were identified according to HLA-Bw4 expression; (G) heatmap showed the in vivo phenotype of infused UCB NK cells and recipient NK cells from the peripheral blood of case 2 at the indicated time points following the first infusion were assessed by flow cytometry analysis. The median log2 values from three independent experiments are presented as a heatmap; no correction for multiple comparisons was applied; (H and I) for the ADCC assay (left), PBMCs derived from case 1 and case 2 at baseline and after 4th and 8th UCB NK cell infusions were co-incubated with Kelly cells in vitro at an E: T ratio of 1:1 or 5:1 for 4 h. For the CDC assay (right), plasma derived from case 1 and case 2 at baseline and after 4th and 8th UCB NK cell infusions were co-incubated with Kelly cells in a serum: cell culture medium ratio of 1:20 or 1:4 for 4 h. Flow cytometry was performed to detect the apoptosis of Kelly cells. ADCC, antibody-dependent cellular cytotoxicity; CDC, complement-mediated cytotoxicity; E:T, effector:target; S:M, serum: medium; PB, peripheral blood; UCB, umbilical cord blood. Case 2 was a 4-y-old boy initially diagnosed with high-risk NB in the left adrenal gland, with MYCN non-amplification and 11q deletion. The patient achieved a CR following induction, consolidation, and maintenance therapy with naxitamab. Unfortunately, 16 months after treatment discontinuation, the patient relapsed involving the bone marrow, multiple bones, and soft tissue lesions of the skull. Following second-line therapy with six courses of irinotecan plus temozolomide, the overall response was minimal, with CR in both bone marrow, PR of skull soft tissue lesions, and stable disease of residual lesions in multiple bones. 123I-meta-iodobenzylguanidine scintigraphy revealed the maximum standardized uptake value in the skull lesions. Subsequently, the patient received a UCB NK cell infusion combined with dinutuximab-based anti-GD2 therapy (Figure 6A). A follow-up 18F-meta-fluorobenzylguanidine positron emission tomography/computed tomography scan revealed a significant reduction in all tumor lesions, with a maximum standardized uptake value of 9.5 in the skull lesion (Figure 6D). However, pathological biopsy of this lesion confirmed that there was no tumor. The attending physician determined that the patient had achieved CR in the bone marrow, bones, and skull soft tissue lesions. This evidence suggests that UCB NK cell infusion may overcome anti-GD2 Ab resistance and that combining UCB NK cell infusion with GD2 Ab-based salvage therapy demonstrates favorable efficacy in children with relapsed/refractory NB. In general, UCB NK cell infusion and anti-GD2 Ab combination therapy was well-tolerated, and no treatment-related adverse events were observed in these two patients. In vivo persistence and phenotype of donor UCB NK cells To determine persistence after the first single infusion of donor UCB NK cells in the two patients, a DNA chimerism assay was performed on case 1, as both the donor and recipient were HLA-Bw4-negative. For case 2, a flow cytometry assay was performed because the first infusion was HLA-Bw4-mismatched with the recipient, with later donors HLA-Bw4-matched with the recipient. The DNA chimerism assay showed a high level of UCB NK cells on day 1 after the first infusion in case 1, followed by a sharp decrease on day 7 and a gradual increase from days 7 to 21 (Figure 6E). Flow cytometry analysis showed that the in vivo expansion of donor UCB NK cells in case 2 peaked 21 d after the first infusion and remained >20% on day 56 (Figure 6F). Notably, UCB NK cells seemed to show a second expansion on day 49 (Figure 6F). For the in vivo phenotype (Figure 6G), compared with endogenous NK cells, adoptively transferred UCB NK cells maintained a higher expression of multiple activating receptors, including NKp30, NKp46, NKp44, CD16, and CD69, and memory-like NK cell markers, including IL7R and CXCR6, but lower expression of multiple inhibitory receptors, including ILT2, CD158, NKG2A, PD1, TIM3, and LAG3, from 1 to 56 d after the first infusion. The activating receptors on donor UCB NK cells continuously increased from days 1 to 14 and then gradually decreased from days 21 to 56; in contrast, the expression of inhibitory receptors was relatively stable over time post-infusion. These results suggest that donor UCB NK cells sustain a long-term, stable phenotypic profile featuring high activation, low inhibition, and memory-like traits in vivo. Notably, PD-1 expression in donor UCB NK cells began to decrease on day 35 and remained low on day 56, underscoring the prominent capacity of donor UCB NK cells to resist functional exhaustion in vivo. However, donor UCB NK cells exhibited lower NKG2D expression than recipient NK cells. We also evaluated the ADCC effect on patient-derived peripheral blood mononuclear cells and the CDC effect on patient-derived serum at baseline and after UCB NK cell infusion. Compared with those at baseline, both the ADCC and CDC effects dramatically increased after multiple infusions (Figure 6H and I). Combined UCB NK cell infusion and anti-GD2 Ab elicits a “hotter” TME To comprehensively assess the serum milieu of the two patients, we used a semiquantitative human cytokine Ab array to profile 120 different cytokines in paired serum samples before and after treatment (Supplementary Figure S10A). Gene Ontology (GO) biological process enrichment analysis showed the upregulation of lymphocyte chemotaxis, chemokine-mediated signaling pathways, and neutrophil and granulocyte chemotaxis after treatment (Figure 7A). Furthermore, post-treatment samples showed increased expression of NK-related, macrophage-related, neutrophil-related, and Th1-related cytokines and chemokines (Figure 7B). A protein–protein interaction network identified CXCL10 as a hub in post-treatment serum (Figure 7C). Moreover, GO molecular function analysis demonstrated a marked enrichment of chemokine receptor binding after combined therapy, particularly CXCR3 (Supplementary Figure S10B). This pattern implied the engagement of chemokine ligand‒receptor axes, such as CXCL10‒CXCR3, in the post-treatment serum. Collectively, these data support chemokine-mediated recruitment and immune activation in patient sera after UCB NK cell plus anti-GD2 therapy. Serum cytokine profiles and tumor transcriptomic signatures of patients before and after combined UCB NK cell infusion and anti-GD2 therapy. (A) Gene Ontology biological process enrichment of differential proteins based on semiquantitative human cytokine antibody array in the serum samples of case 2 before (day 0) and after (day 98) combined UCB NK cell infusion and anti-GD2 therapy; (B) line plots of the serum cytokines and chemokines grouped as macrophage-related, NK-related, Th1-related and neutrophil-related factors at baseline and after combined UCB NK cell infusion and anti-GD2 therapy in case 1 (red circles) and case 2 (blue squares). No correction for multiple comparisons was performed; (C) the hub genes of the protein–protein interaction network in the posttreatment serum of case 1 and case 2; (D) gene ontology biological process enrichment of the differentially expressed genes based on bulk RNA-sequencing of case 2's paired tumor tissues obtained before and after 8 cycles of UCB NK cell infusion combined with anti-GD2 therapy. (E) Heatmap of gene expression representing NK-cell dysfunction, stress, activation, and cytotoxicity based on bulk RNA-sequencing of case 2's paired tumor tissues obtained before and after combined UCB NK-cell infusion and anti-GD2 therapy. UCB, umbilical cord blood. By RNA sequencing analysis, tumor specimens of case 2 after treatment showed coordinated activation of innate and adaptive immune responses (Figure 7D), with the top terms included lipid metabolic process, NK T-cell activation, T-helper 1 cell cytokine production, cellular response to tumor necrosis factor, positive regulation of neutrophil chemotaxis, response to macrophage colony-stimulating factor, and regulation of NK cell-mediated cytotoxicity directed against tumor cell targets. Meanwhile, the significant enrichment of Fc-gamma receptor signaling, C3b/C4b binding, and complement activation implied the enhanced effects of ADCC and CDC after treatment. Moreover, comparative analysis of NK cell-related gene expression demonstrated upregulation of cytotoxicity and activation signatures but downregulation of dysfunction and stress signatures (Figure 7E), suggesting that intratumoral NK cells adopted a more activated functional state after UCB NK cell infusion and anti-GD2 combined therapy. Discussion Treatment of high-risk and relapsed/refractory NB remains a major challenge in pediatric oncology. 1 , 2 For the first time, we conducted a complete translational study of UCB-derived NK cells combined with anti-GD2 Ab in NB, spanning preclinical validation of in vitro function and animal models for clinical application in two relapsed/refractory patients. Thus, this study provides a novel and effective “off-the-shelf” combination strategy for NB immunotherapy using UCB NK cells. Previous studies have reported the promoting effect of NK cells on anti-GD2 Ab efficacy. 16 , 17 Several ongoing clinical trials have evaluated haploidentical or autologous NK cell-based immunotherapies combined with anti-GD2 Abs for NB, such as NCT00877110, NCT01576692, NCT01857934, NCT02573896, and NCT03242603. However, the superiority of UCB NK cells in NB has not yet been clarified. This study systematically confirmed the unique functional advantages and molecular characteristics of UCB NK cells over patient-derived PB NK cells, as well as the superior capacity of UCB NK cells to mediate anti-GD2-dependent cytotoxicity against GD2-positive NB cells. In the present study, 43% of CR patients showed low NKp46 expression, while 33% of non-CR patients exhibited high NKp46 levels, indicating that anti-GD2 therapeutic efficacy is not fully determined by NK cell infiltration alone. It has been reported that key factors, including GD2 expression on tumor cells, 27 , 29 infiltration of macrophages and neutrophils, 11 Fcγ receptor diversity on NK cells, 30 and other immune components in the TME induced by radiation therapy 31 , significantly influence the anti-GD2 response. Recent studies have also demonstrated possible correlations between CD4+ and CD8+ T cells and NK cells 12 , 32 or the anti-GD2 response, 33 and further investigations of T cell subsets will be included in our future studies to characterize the broader TME of NB. Mechanistically, our findings revealed that UCB NK cells exhibit unique phenotypic profiles. First, UCB NK cells constitutively expressed higher levels of cytotoxic molecules than PB NK cells, which further increased upon anti-GD2 treatment. Second, UCB NK cells sustained a stable and favorable activation–inhibition receptor balance. Elevated CD158 and NKG2A expression in UCB NK cells under basal conditions may be attributed to their immature phenotypes. 25 Upon contact with NB cells, UCB NK cells exhibited a modest upregulation of activating receptors and a subtle downregulation of inhibitory receptors. Importantly, the anti-GD2 Ab further amplifies the activation–inhibition balance of UCB NK cells, whereas PB NK cells display only modest responsiveness to the anti-GD2 Ab. Indeed, donor UCB NK cells in case 2 retained higher expression of activating receptors and lower expression of inhibitory receptors. Notably, a marked reduction in PD-1 expression on day 35 post infusion underscored the superior resistance of donor UCB NK cells to functional exhaustion. Third, the donor UCB NK cells demonstrated prolonged persistence in vivo. Donor UCB NK cells showed a stably high expression of memory-like markers in vivo, in contrast to recipient NK cells. In this study, the results of flow cytometry and DNA chimerism assays were generally consistent, with the maximum expansion of donor UCB NK cells at 21 d post-infusion in vivo. In line with a previous study, 34 flow cytometry might be more sensitive. The high DNA chimerism rate of UCB NK cells on day 1 post-infusion in case 1 may be attributed to slow clearance of donor DNA. Prior studies demonstrated that unlicensed NK cells lacking inhibitory KIRs specific for self-HLA played a critical role in mediating ADCC against NB following anti-GD2 Ab treatment, as HLA class I on NB cells selectively suppressed licensed NK cell function. 35 , 36 The immature phenotype of UCB NK cells may reflect a high proportion of unlicensed subsets, which contributes to their robust ADCC response to anti-GD2 Ab, though formal KIR/KIR-ligand genotyping was not performed in this exploratory study. Importantly, this study revealed that UCB NK cells and anti-GD2 Ab combination therapy remodeled into a “hot” immune TME. RNA sequencing of xenografts revealed enrichment of immune activation signatures, suggesting a potential association between immune cell activation and metabolic reprogramming. 37-40 The identification of CXCL10 as a hub molecule in post-treatment serum is also notable. CXCL10 may serve as a predictive biomarker for treatment response. Apart from efficacy, a key distinction of this study was the favorable safety profile of UCB NK cells. In contrast to GD2-targeted T-cell therapies 41 or haploidentical NK cells, 17 this favorable safety profile is particularly valuable for pediatric patients. Anti-PD-1/PD-L1 antibodies did not significantly enhance the killing of NB cells by UCB NK cells combined with anti-GD2 Ab in vitro, which may be explained by the simple in vitro model, the short duration of NK cell function, and other inhibitory signals. 42 Further investigation of triple combination therapy is warranted in GD2-positive NB. 43 Our study has several important limitations. First, the clinical study included only 8 patients, and the results should be considered anecdotal and exploratory; long-term follow-up is required to evaluate the durability of clinical responses and delayed toxicities. A larger prospective cohort study is essential to confirm the clinical efficacy and safety of this regimen. Second, immunocompetent mouse models are needed to better recapitulate the human TME, which may account for the different expression patterns of NKG2A and NKG2D observed in vitro versus in vivo. Third, a broader range of GD2-positive NB cell subtypes, including the noradrenergic and mesenchymal subtypes, are needed for preclinical validation. Fourth, CXCL10 requires further validation in a larger patient cohort and long-term efficacy studies to confirm its value as a predictive biomarker. Additionally, we did not perform KIR/KIR-ligand genotyping to characterize licensed and unlicensed NK cell subsets, limiting direct comparison to prior studies. 35 , 36 In conclusion, our study identified UCB NK cells as a superior effector source for NB immunotherapy, characterized by a highly activating-low inhibitory phenotype, strong effector function, long-term in vivo persistence, resistance to functional exhaustion, and favorable safety profiles in preclinical models and clinical cases. Moreover, UCB NK cells combined with anti-GD2 Ab showed remarkable synergistic efficacy in mice and achieved significant clinical benefits in patients with relapsed/refractory NB, thereby providing the first clinical evidence of combined UCB NK cell and anti-GD2 therapy. Our findings support UCB NK cells as a unique therapeutic option that integrates safety and efficacy and as a promising adjuvant to improve the efficacy of anti-GD2 therapy for NB. Supplementary Material Supplementary material New Supplementary materials and methods.docx Supplementary material New CONSORT checklist 20260509.docx Supplementary material New Author Checklist 20260509.docx New Supplementary Figures and Tables.docx New Supplementary Figures and Tables.docx Funding Statement This work was supported by the National Key Research and Development Program of China under Grant No. 2022YFC2705005; the National Natural Science Foundation of China under Grant Nos. 82203754, 82573714, and 82302940; the Natural Science Foundation of Guangdong Province, China under Grant No. 2024A1515012866; and the Guangdong Basic and Applied Basic Research Foundation under Grant No. 2026A1515010153. Acknowledgments We sincerely thank the neuroblastoma patients and their families for their support. We also acknowledge the China Cord Blood Bank of Shandong Province for providing umbilical cord blood natural killer cells. Disclosure of potential conflicts of interest No potential conflicts of interest were disclosed. Data availability statement RNA-sequencing data have been deposited in the Gene Expression Omnibus (GEO) under accession numbers GSE322849 for mouse tumor xenografts and GSE324016 for human tumor samples. Key flow cytometry datasets are available in a public repository of the Research Data Deposit. Other data supporting the findings of this study are available from the corresponding author upon reasonable request. Ethics approval statement This study was approved by the Ethics Committee of Sun Yat-sen University Cancer Center (Approval Number. SL-B2024-008-02) and conducted by the latest version of the Declaration of Helsinki. Written informed consent was obtained from the patients before they participated in the study. All experiments involving humans were carried out in accordance with the Code of Ethics of the World Medical Association (Declaration of Helsinki). All experiments involving animals were approved by the Institutional Animal Care and Use Committee of Sun Yat-sen University Cancer Center (Approval Number: L102012023080P) and conducted in compliance with the American Veterinary Medical Association Guidelines. Abbreviations antibody antibody-dependent cellular cytotoxicity complement-mediated cytotoxicity complete response effector-to-target disialoganglioside GD2 gene ontology human leukocyte antigen interferon-gamma immunohistochemistry neuroblastoma natural killer peripheral blood programmed cell death protein 1 programmed death-ligand 1 partial response tumor microenvironment tumor necrosis factor-alpha umbilical cord blood Supplementary material Supplemental data for this article can be accessed at https://doi.org/10.1080/2162402X.2026.2679315. References Associated Data Supplementary Materials Supplementary material New Supplementary materials and methods.docx Supplementary material New CONSORT checklist 20260509.docx Supplementary material New Author Checklist 20260509.docx New Supplementary Figures and Tables.docx New Supplementary Figures and Tables.docx Data Availability Statement RNA-sequencing data have been deposited in the Gene Expression Omnibus (GEO) under accession numbers GSE322849 for mouse tumor xenografts and GSE324016 for human tumor samples. Key flow cytometry datasets are available in a public repository of the Research Data Deposit. Other data supporting the findings of this study are available from the corresponding author upon reasonable request.