Carbon nanoparticle-Fe (II) complex for ferroptosis therapy in refractory solid tumors: a phase I trial

Abstract

We developed a novel carbon nanoparticle-Fe (II) complex prepared via suspension injection (CNSI-Fe) to induce cancer cell ferroptosis through direct intratumoral delivery of ferrous iron. Preclinical studies have demonstrated that CNSI-Fe has promising antitumour effects, thereby prompting this first-in-human, single-arm, open-label, dose-escalation phase I trial in patients with refractory solid tumors. Primary objectives included assessments of safety and tolerability and the determination of dose-limiting toxicity (DLT) and the maximum tolerated dose (MTD). Secondary objectives included the determination of preliminary antitumour efficacy and pharmacokinetics. Applying a 3 + 3 design across five doses (30–150 mg), a total of 19 enrolled patients received CNSI-Fe treatment. The MTD was not reached, and only one patient (in the 90 mg cohort) experienced DLT events. For all treated patients (n = 19), the common adverse events included injection site pain (89.5%), hypertension (52.6%), elevated serum iron levels (47.4%), and anemia (47.4%). Pharmacokinetic analysis revealed rapid iron absorption (peak time: 0.15–0.41 h) and clearance (half-life: 1.96–5.06 h), with peak serum concentrations ranging from 3520.33 to 13961.67 ng/mL. Preliminary efficacy estimates revealed an overall response rate of 10.5% and a disease control rate (DCR) of 84.2%. Notably, the DCR reached 100% in patients with lymph node and sarcoma lesions. Overall, CNSI-Fe demonstrated a favorable safety profile and promising antitumour activity (ClinicalTrials.gov identifier: NCT06048367), supporting further investigations in a phase II trial.

Introduction

Ferroptosis-targeted cancer therapy has emerged as a promising strategy for eradicating tumors.^1–6 Ferroptosis inducers (FINs) trigger a non-apoptotic, iron-dependent form of of regulated cell death characterized by lethal accumulation of lipid peroxides. This mechanism enhances cancer cell sensitivity to radiotherapy and overcomes acquired drug resistance.^6,7 For instance, glutathione peroxidase 4 (GPX4) inhibitors have been shown to resensitize TKI-resistant tumor cells, restoring therapeutic efficacy.² While the synergistic effects of FINs with radiotherapy and TKIs are well documented,⁸ the interplay between ferroptosis and immune regulation is more complex, being both context-dependent and bidirectional. On one hand, ferroptosis functions as a form of immunogenic cell death, triggering antitumor immunity. GPX4 inhibitors such as RSL3 and N6F11 have demonstrated the ability to enhance antitumor immune responses and potentiate immune checkpoint inhibitors in preclinical models.^9,10 On the other hand, ferroptosis can also exert immunosuppressive effects. Emerging evidence indicates that ferroptotic tumor cells release the ferroptosis suppressor GPX4, which binds to zona pellucida protein 3 on dendritic cells (DCs). This interaction activates the cAMP-dependent protein kinase (cAMP-PRKA) signaling cascade, inhibits DC glycolysis and maturation, and impairs antitumor T cell priming.¹¹ Moreover, ferroptosis may promote the release of oxidized lipids from pathologically activated neutrophils (PMN-MDSCs), leading to suppressed T cell function and reduced responsiveness to immune checkpoint blockade.¹²

To harness the therapeutic potential of ferroptosis, a broad spectrum of FINs has been developed, targeting key regulators of iron and lipid metabolism, including system Xc⁻ inhibitors, glutathione-depleting agents, tumor-specific GPX4 degrader, ferroptosis-suppressor-protein 1 (FSP1) inhibitors, ferric ion donors, nuclear factor-erythroid 2 related factor 2 (Nrf2) inhibitors, and thioredoxin reductase 1 (TXNRD1) inhibitors.^13–17 However, nearly all small-molecule FINs remain in preclinical stages due to inherent limitations such as poor aqueous solubility, chemical instability, rapid clearance, and off-target toxicities. These challenges reflect broader pharmacokinetic and physicochemical limitations inherent to small-molecule FINs as a class, and highlight the need for innovative delivery strategies to unlock their clinical potential.

Nanotechnology provides a compelling avenue to overcome these barriers. Encapsulation of hydrophobic or labile FINs within nanocarriers enables enhanced stability, tumor-selective delivery, and controlled release, thereby transforming mechanistically promising but druggability-limited agents into viable clinical candidates. Preclinically, ferroptosis-inducing nanoformulations have demonstrated marked advantages over conventional small molecules.¹⁸ Beyond improving pharmacokinetics, such nanoplatforms can be engineered for multifunctionality: co-loading chemotherapeutics, incorporating targeting ligands, or enabling combination with other treatment modalities. This versatility offers spatiotemporal precision in drug action within the tumor microenvironment (TME) and addresses key translational hurdles.¹⁹ In particular, nanoparticle-mediated delivery of ferrous iron or modulation of iron homeostasis has been widely shown to trigger ferroptosis and confer therapeutic benefit.^20,21 The convergence of nanotechnology and oncology has catalyzed innovation in cancer therapy, with carbon-based nanomaterials (CBNs) gaining prominence due to their exceptional structural, electronic, and optical properties.^22–25 Carbon nanoparticles (CNPs) exhibit high surface area, electrical and thermal conductivity, and chemical stability-attributes that confer superior performance at the physicochemical, biological, and therapeutic levels.²² Recent advances in CBN engineering, including graphene oxide derivatives, graphdiyne, and functionalized carbon nanotubes, have enabled the design of TME-responsive systems capable of drug release, photothermal ablation, and immunomodulation.^26–30 While conventional applications have focused on nanocarrier-mediated delivery of chemotherapeutics such as paclitaxel or doxorubicin,¹⁸ a new paradigm has emerged wherein the nanoparticle itself constitutes the active therapeutic agent. Among these, the carbon nanoparticle-Fe (II) complex prepared via suspension injection (CNSI-Fe) exemplifies this paradigm shift. Unlike passive delivery systems, CNSI-Fe functions as an integrated theranostic platform-combining scalable synthesis, dual-modal imaging, and tumor-selective iron delivery to induce ferroptosis directly.³¹ This agent is designed to actively modulate the TME by delivering ferrous iron to tumor sites, bypassing systemic toxicity and enhancing therapeutic precision.³² Preclinical studies have demonstrated robust anti-tumor activity across colorectal, breast, and pancreatic cancer models, with enhanced efficacy when combined with chemotherapy, photothermal therapy, or immunotherapy.^27,31–33

In this study, we report the first-in-human evaluation of CNSI-Fe, a ferroptosis-inducing nanomedicine designed for intratumoral administration in patients with advanced solid tumors. We comprehensively assessed its safety, tolerability, pharmacokinetics, and preliminary antitumor activity. This work bridges the critical gap between preclinical innovation and clinical application, establishing CNSI-Fe as a transformative platform in ferroptosis-based cancer therapy. The findings not only support the continued clinical development of CNSI-Fe but also provide a framework for the rational design of next-generation ferroptosis-inducing nanomedicines.

Results

Generation of CNSI-Fe with strong antitumour potency in preclinical models

CNSI-Fe was synthesized by combining carbon nanoparticles with ferrous sulfate to form stable complexes (199.4 ± 4.9 nm in diameter; polydispersity index = 0.137) (Supplementary Fig. 1). The CNSI-Fe was stable for 6 h (Supplementary Table 1). CNSI-Fe triggered ferroptosis-mediated antitumour activity in vitro, significantly suppressing 4T1 cell proliferation while increasing intracellular iron and hydroxyl radical production. CNSI-Fe treatment increased both hydrogen peroxide and malondialdehyde (MDA) levels but decreased peroxidase activity and the content of reduced glutathione (GSH). In vivo data demonstrated that CNSI-Fe inhibited tumor growth across six xenograft models, including colon cancer, breast cancer, and pancreatic cancer models. These findings demonstrate that CNSI-Fe induces ferroptosis through iron-dependent lipid peroxidation, resulting in broad-spectrum antitumour activity and favorable safety characteristics in multiple preclinical cancer models (Supplementary Figs. 2–4).

Trial design and patient characteristics

This is a first-in-human phase I study of the dose-escalation administration of CNSI-Fe in patients with refractory solid tumors. The study employed a classic “3 + 3” dose-escalation design, with five predefined dose levels, namely, 30, 60, 90, 120, and 150 mg. To assess the safety and tolerability of CNSI-Fe, we established primary objectives, including the incidence of dose-limiting toxicity (DLT) and the determination of the maximum tolerated dose (MTD), and secondary objectives, including assessments of the efficacy and pharmacokinetics of CNSI-Fe. The eligibility criteria were as follows: age 18–80 years; histologically or cytologically confirmed advanced solid tumor refractory to standard therapies or lacking effective treatment options; at least one measurable lesion; lesions accessible for intratumoral injection (direct or imaging-guided); and an Eastern Cooperative Oncology Group (ECOG) performance status of 0 or 1.

The median age was 55 years (range: 44–73 years), with a balanced sex distribution (male:female ratio, 1.4:1). The participants were histologically confirmed to have treatment-refractory advanced solid malignancies, including sarcoma (26.3%, n = 5), head and neck carcinoma (21.1%, n = 4), and non-small cell lung cancer (15.8%, n = 3). Metastatic disease (stage IV) was present in 84.2% (16/19) of the participants, and all the participants had received at least two prior systemic therapies. The demographic and clinical characteristics of the treated patients are summarized in Table 1.

Table 1.

Patient demographics and baseline clinical characteristics

Characteristics	Number of patients
Median age, y (range)	55 (44–73)
Sex, n (%)
Male	11 (57.9)
Female	8 (42.1)
ECOG performance status, n (%)
0	12 (63.2)
1	7 (36.8)
Tumor type, n (%)
Sarcoma	5 (26.3)
Head and neck neoplasm	4 (21.1)
Pancreatic cancer	2 (10.5)
Non-small cell lung cancer	3 (15.8)
Ovarian cancer	1 (5.3)
Breast cancer	2 (10.5)
Vulvar cancer	2 (10.5)
Clinical stage, n (%)
I (High risk)	1 (5.3)
III	2 (10.5)
IV	16 (84.2)
Previous lines of therapy, n (%)
2	3 (15.8)
3	4 (21.1)
≥4	12 (63.2)

Treatment

This study followed the Consolidated Standards of Reporting Trials (CONSORT) guidelines, and participant progression through the intervention phases is detailed in Fig. 1. CNSI-Fe was delivered via ultrasound- or computed tomography-guided intratumoral administration on day 1 of each 21-day cycle. Dose escalation followed a standard “3 + 3” design, with DLTs evaluated during cycle 1 according to the Common Terminology Criteria for Adverse Events (CTCAE) v5.0 criteria. Patients who achieved stable disease or better per the Response Evaluation Criteria in Solid Tumors (RECIST) 1.1 criteria were eligible for cycle 2 injections. Safety monitoring was continued for 28 days post-treatment. If no DLTs were observed, the dose was escalated to the next level. If two or more DLTs occurred at a given dose, the dose was de-escalated. Patients were administered intratumoral CNSI-Fe at five escalating dose levels: 30 mg (n = 3), 60 mg (n = 4), 90 mg (n = 6), 120 mg (n = 3), and 150 mg (n = 3). In the 90 mg dose regimen, the cohort was expanded to include six patients following the occurrence of two DLT events in one participant: grade 3 injection site pain and grade 4 hyperlipasemia.

Fig. 1.

CONSORT flow diagram

Clinical outcomes

This phase I clinical trial (December 2022–September 2024) included 20 eligible participants, 19 of whom received treatments with CNSI-Fe. The treatments consisted of a single cycle for 12 participants (63.2%) and two cycles for 7 participants (36.8%), with a minimum 1-month follow-up period. All 19 treated participants underwent safety evaluations. For the DLT assessment, one participant (T02002) discontinued treatment early because of a treatment-related serious adverse event (grade 3 hypersensitivity reaction) that occurred during administration. Since the administered dose did not reach 80% of the planned 60 mg dose (48 mg), this participant was excluded from the DLT evaluation, resulting in a total of 18 participants being assessed for DLT. For the efficacy assessment, one participant (T01006) was in poor physical condition and did not undergo efficacy evaluation, resulting in only 18 participants being included in the efficacy assessment.

Safety

Among all 19 treated participants (18 evaluable), two DLT events were observed in a 55-year-old patient with pancreatic cancer (90 mg cohort, 1/18 evaluable patients, 5.6%, Supplementary Table 2). The first event was grade 3 injection site pain, which resolved following pharmacological management. This adverse reaction, which was definitively attributed to the investigational drug, resulted in permanent treatment discontinuation and a prolonged hospital stay (>72 h), meeting the criteria for a serious adverse event (SAE). The second DLT event manifested as grade 4 hyperlipasemia, which remained unresolved throughout the observation period without specific intervention. In accordance with the “3 + 3” escalation rules, since one DLT was observed among the initial three participants enrolled at this dose level, three additional participants were enrolled at the same dose. As no DLTs occurred among these additional participants, dose escalation proceeded to the 120 mg and 150 mg cohorts. No DLTs were observed in the higher-dose groups, precluding the identification of the MTD.

Treatment-related adverse events (AEs) occurred in all participants (n = 19), with 73.7% (14/19) experiencing grade 3 or higher events. The most common AEs were injection site pain (89.5%), hypertension (52.6%), hyperferremia (47.4%), and anemia (47.4%) (Table 2). SAEs occurred in eight participants (42.1%), including malignant tumor progression (n = 3) and cancer pain (n = 2). Notable SAEs included grade 3 hypersensitivity reaction (n = 1), grade 4 upper gastrointestinal hemorrhage (n = 1), and grade 5 malignant tumor progression-related death (n = 2; one possibly unrelated and the other unrelated to treatment). AEs of special concern included one case of hypersensitivity reaction (5.3%). Additionally, 8 cases of suspected and unexpected serious AEs were reported (42.1%).

Table 2.

Treatment-related AEs by grade and system

General	Any grade	Grade ≥ 3
Injection site pain	17 (89.5)	4 (21.1)
Injection site swelling	2 (10.5)	0
Weakness	7 (36.8)	0
Pyrexia	6 (31.6)	0
Dizziness	5 (26.3)	0
Headache	2 (10.5)	0
Dorsalgia	2 (10.5)	0
Electrolyte imbalance
Elevated serum iron	9 (47.4)	0
Hypokalemia	2 (10.5)	1 (5.3)
Hyponatremia	7 (36.8)	1 (5.3)
Hematologic
Anemia	9 (47.4)	1 (5.3)
Decreased LYC	6 (31.6)	1 (5.3)
Decreased WBC	4 (21.1)	0
Decreased NEU	3 (15.8)	0
Elevated WBC	2 (10.5)	0
Elevated NEU	2 (10.5)	0
Elevated PLT	4 (21.1)	0
Gastrointestinal
Emesis	7 (36.8)	1 (5.3)
Nausea	7 (36.8)	0
Constipation	2 (10.5)	0
Oral hypoesthesia	2 (10.5)	0
Anorexia	4 (21.1)	0
Respiratory
Hemoptysis	2 (10.5)	0
Cough	2 (10.5)	0
Dyspnea	4 (21.1)	1 (5.3)
Pneumothorax	2 (10.5)	0
Hepatobiliary
Elevated ALT	4 (21.1)	0
Hypoalbuminemia	5 (26.3)	0
Hypertriglyceridemia	4 (21.1)	1 (5.3)
Cardiovascular
Abnormal T wave	2 (10.5)	0
Elevated heart rate	2 (10.5)	0
Elevated blood pressure	10 (52.6)	8 (42.1)

LYC lymphocyte, WBC white blood cell, NEU neutrophil, PLT platelet, ALT alanine aminotransferase

With respect to serious adverse events, grade 4 upper gastrointestinal bleeding manifested as the subject vomiting ~200 mL of dark red gastric contents, with the patient’s hemoglobin concentration decreasing to 52 g/L. The relationship to the drug was considered possibly unrelated. The grade 4 lipase elevation manifested as an increase in lipase concentration to 391 IU/L 28 days after drug administration, peaking at 1332 IU/L on day 40 after administration. The relationship with the drug was considered possibly related. The investigator reported that the patient had no symptoms, such as acute abdominal pain, nausea and vomiting, or fever, and amylase levels were not elevated—only the lipase level was elevated. Therefore, pancreatitis was not considered.

Iron homeostasis monitoring revealed transient hyperferremia in nine participants (47.4%), with no patients meeting clinical iron poisoning criteria. Hematologic changes included anemia (47.4%, 1 grade 3 event), lymphocytopaenia (31.6%, 1 grade 3 event), and platelet count fluctuations (21.1% elevation). No grade 4 hematologic toxicity or treatment-related organ failure was observed. Cardiovascular-related SAEs included hypertension (52.6%, 8 grade ≥3), hypotension (10.5%), and tachycardia (10.5%). Cardiac events included T-wave abnormalities (10.5%) and palpitations (10.5%). Gastrointestinal events included vomiting (36.8%, 1 grade ≥3) and nausea (36.8%). Respiratory events included dyspnea (21.1%). Hepatic events included hypoalbuminemia (26.3%), elevated alanine aminotransferase (ALT) levels (21.1%), and decreased blood albumin levels (10.5%).

Clinical efficacy

Antitumour efficacy was assessed for both the injected lesions and overall tumor burden using the revised RECIST version 1.1. Among the 19 treated patients, 18 were evaluable for efficacy. The best responses for the injected lesions and overall tumor lesions are summarized in Supplementary Table 3. With respect to the injected lesions, two patients achieved a partial response (PR) in the 30 mg dose and 60 mg dose groups, resulting in an objective response rate (ORR) of 10.5% (95% confidence interval [CI], 1.3–33.1). Stable disease (SD) was observed in 14 patients (one in the 30 mg group, two in the 60 mg group, five in the 90 mg group, three in the 120 mg group, and three in the 150 mg group), yielding a disease control rate (DCR) of 84.2% (95% CI, 60.4–96.6). For overall tumor lesions, one patient achieved a PR in the 30 mg dose group, resulting in an ORR of 5.3% (95% CI, 0.1–26.0). Twelve patients (one in the 30 mg group, three in the 60 mg group, four in the 90 mg group, two in the 120 mg group, and two in the 150 mg group) achieved SD, with a DCR of 68.4% (95% CI, 43.5–87.4).

The antitumour efficacy of the injected lesions was superior to that of the overall tumor lesions. A comprehensive assessment of the injected lesions for all 19 treated patients is presented in Fig. 2a. The percentage change in injected lesion size relative to baseline over the course of the study is depicted in Fig. 2b, and the best efficacy outcomes for injected lesions are shown in Fig. 2c. Among the 18 patients evaluable for efficacy, the injected lesion was a lymph node in three patients, with a DCR of 100%. In 5 patients with sarcoma, the DCR also reached 100%, as illustrated in Fig. 2d.

Fig. 2.

Preliminary antitumour activity of CNSI-Fe in patients with advanced solid tumors. a Swimmer plot of 19 patients who received CNSI-Fe. b Spider plot showing changes from baseline in the injected lesions of the 18 evaluable patients. c Waterfall plot illustrating the best percentage change from baseline in the injected lesions of the 18 evaluable patients. d The best response of the injected lesion was evaluated according to tumor type. Among the 18 efficacy-evaluable patients, 3 had lymph nodes as the injected lesion, and 5 had sarcoma. CR complete remission, PR partial response, SD stable disease, PD progressive disease, NE no evaluation, LOC lip and/or oral cancer, PTC papillary thyroid carcinoma, MELTVR myoepithelioma-like tumors of the vulvar region, LPS liposarcoma, PC pancreatic cancer, NCC nasal cavity cancer, OS osteosarcoma, IDBC invasive ductal breast carcinoma, VSCC vulvar squamous cell carcinoma, STS soft tissue sarcoma; spindle cell sarcoma, DDLPS dedifferentiated liposarcoma, LUAD lung adenocarcinoma, OSC ovarian serous cystadenocarcinoma, LSCC lung squamous cell carcinoma, NPC nasopharyngeal carcinoma

Two notable cases are highlighted: (1) A patient with myoepithelioma-like tumors of the vulvar region (MELTVR) and biopsy-confirmed lymph node metastasis received a single 60 mg dose of CNSI-Fe injected into the right inguinal lymph node. Despite discontinuation due to hypersensitivity and the administration of less than 80% of the prescribed dose, the treatment achieved complete remission (CR), with the lymph node size decreasing from 16.4 mm to 5.0 mm on CT (Fig. 3a). Whole-genome and exome sequencing revealed somatic mutations and copy number variations in the tumor, including in two ferroptosis-related genes (HERC2 and AKR1C1) (Supplementary Fig. 5). (2) A patient with papillary thyroid carcinoma and lymph node metastasis received two 30 mg doses of CNSI-Fe injected into the right sternocleidomastoid deep lymph node. Following injection, the lymph node exhibited marked necrosis, with the short-axis diameter decreasing by 30.1% (from 15.6 mm to 10.9 mm) on day 42 and further decreasing to 9.5 mm by day 71. Remarkably, CT scans on day 425 and day 520 revealed complete resolution of the lymph node metastasis, indicating a CR (Fig. 3b).

Fig. 3.

CT images of two notable cases. a CT images of a MELTVR patient. b CT images of a papillary thyroid carcinoma patient

Pharmacokinetics

Changes in mean serum iron concentrations from baseline following intratumoral CNSI-Fe injection are depicted in Fig. 4a (n = 19 during cycle 1) and Fig. 4b (n = 7 during cycle 2). The calculated average pharmacokinetic parameters are summarized in Supplementary Table 4, whereas individual pharmacokinetic parameters for each patient during cycle 1 and cycle 2 are provided in Supplementary Table 5. Following the intratumoral injection of CNSI-Fe, serum iron levels increased rapidly, with the maximum serum concentration (C_max) ranging from 3520.33 to 13,961.67 ng/mL (7.8 times the normal value). The time to reach peak concentration (T_max) was between 0.15 and 0.41 h, indicating rapid absorption of iron ions into the bloodstream. The half-life (t_1/2) ranged from 1.96 to 5.06 h, and the clearance rate (CL) ranged from 1985.00 to 12,456.56 mL/hour, suggesting efficient clearance and redistribution of iron ions into tissues, with the liver serving as the primary storage organ.

Fig. 4.

Serum iron concentration-time curves for cycles 1 and 2. Changes in the mean serum iron concentration from baseline to 72 h post-administration (a) of 19 patients who received intratumoral injections of CNSI-Fe at different dosages during cycle 1 and (b) of seven patients during cycle 2. Data are represented as means ± SDs. c, d Serum iron ion concentrations of 10 patients whose C_max exceeded 5000 ng/mL

Within the dose range of 30–150 mg per person in cycle 1, serum iron ion C_max and the area under the curve (AUC_0-t and AUC_0-∞) showed a nonlinear relationship with dose, with slopes of −0.0642 (90% CI, −0.9052 to 0.7767), 0.2310 (90% CI, −0.5491 to 1.0112), and 0.3393 (90% CI, −0.4179 to 1.0964), respectively. This nonlinearity may be related to tumor characteristics and vascularity, in contrast with the preclinical findings (Supplementary Fig. 6).

Among the 19 patients, serum iron concentrations exceeded 5000 ng/mL in 10 patients, occurring in 8 following cycle 1 and in 2 following cycle 2 (Fig. 4c, d). In our study, despite markedly elevated serum iron concentrations in 10 patients, including patient T01016, whose serum iron concentration reached 35,179 ng/mL, the symptoms associated with iron poisoning were mild and did not necessitate deferoxamine chelation therapy. This may be attributed to the distinct pharmacokinetic properties of CNSI-Fe, such as a short t_1/2, swift clearance, and limited systemic exposure.

Discussion

In this study, we present a first-in-class phase I trial to evaluate the potential of CNSI-Fe, which is a ferroptosis-inducing agent, for treating patients with refractory solid tumors. The CNSI-Fe demonstrated a manageable safety profile and encouraging clinical activity, with a PR as the best overall response. The administration of CNSI-Fe at doses up to 150 mg demonstrated clinical tolerability in three trial participants, with one patient achieving cumulative serum iron concentrations exceeding 35,000 ng/mL. Pharmacokinetic profiling revealed rapid systemic clearance characterized by a shortened elimination t_1/2. While >20 classes of FINs have been explored preclinically, CNSI-Fe is the first FIN to achieve clinical translation. Our CNSI-Fe solves long-standing barriers to FIN development-systemic toxicity, drug instability, and poor bioavailability-by leveraging tumor-targeted carbon nanotechnology for precision iron delivery. It also offers critical insights into the potential of ferroptosis-based therapies and presents proof-of-concept evidence supporting their translational application in cancer treatment.

In terms of tolerability, two DLT events occurred in patient T01011, who received 90 mg of CNSI-Fe; these events manifested as injection site pain and elevated lipase levels. The most common adverse events were transient injection site pain (89.5%) and hypertension (52.6%). Critically, no grade 4 or higher hematologic or hepatorenal toxicity was observed, aligning with the localized mechanism of action. Similarly, transient hyperferremia (47.4%) was not associated with clinically significant iron toxicity, which is attributable to the unique pharmacokinetics of nanoparticle-delivered Fe²⁺. This was mechanistically supported by nonlinear pharmacokinetics, such as for serum iron, which demonstrated rapid elevation and clearance (distinct from oral iron kinetics), enabling high intratumoral retention without systemic overload. Consequently, while 47.4% of the subjects exhibited significant serum iron increases (some reaching diagnostic thresholds for iron poisoning), no serious intolerance reactions occurred, further corroborating the favorable safety profile. However, hypersensitivity reactions during intratumoral injection necessitate vigilance for potential iron overload/poisoning symptoms (e.g., dizziness, nausea, vomiting, diarrhea, abdominal pain, and shock). Overall, CNSI-Fe-related adverse events were primarily hematologic toxicities, and the safety profile was consistently favorable. In terms of efficacy, the DCR was 68.4%, and a PR was observed in one patient, underscoring the therapeutic potential of CNSI-Fe. Lymph node-injected lesions demonstrated a 100% DCR, likely because of enhanced nanoparticle retention within the reticuloendothelial system. To better understand the underlying mechanism of this patient’s sensitivity to the ferroptosis inducer CNSI-Fe, we analyzed her driver genes and identified two ferroptosis-related genes, HERC2³⁴ and AKR1C1,³⁵ among the 40 established driver genes. Pancancer analysis revealed that her 8 driver genes, including NBPF8P, ANKRD36C, ANKRD36, SORCS2, KMT2C, TCF7L2, MAP2K3, and PPM1F, are positively associated with 5 ferroptosis-promoting genes (ACSL4, ALOX12, BECN1, ATF3, and ALOX5) and negatively associated with 2 ferroptosis-suppressing genes (GPX4 and FXN). Crucially, these driver gene signatures are associated with ferroptosis sensitivity, offering the first clinical framework for patient stratification in this emerging field.

Certain cancer types, such as triple negative breast cancer (TNBC), clear cell renal cell carcinoma with elevated levels of polyunsaturated ether phospholipids, therapy-resistant mesenchymal-type cancers, and epidermal growth factor receptor-tyrosine kinase inhibitor (EGFR-TKI)-resistant lung cancers, are susceptible to ferroptosis.^2,36 Future clinical trials and practice may focus on leveraging ferroptosis driver gene profiles to identify tumor types and specific patients who could benefit from ferroptosis-targeted therapy. The results of the clinical trials are encouraging; however, the ORR was notably low at only 10.5%. This low efficacy metric may be partly attributable to intratumoral liquefaction, which could undermine the accuracy of therapeutic efficacy evaluation. To address this specific challenge in future research, optimizing injection techniques, implementing combination therapies, and refining product design are recommended. Furthermore, a significant underlying challenge is the intrinsic resistance of cancer cells to ferroptosis. This resistance is mediated through multiple protective mechanisms, including robust endogenous antioxidant systems, suboptimal cellular acidity, and limited availability of hydrogen peroxide.³⁷ Despite this resistance, emerging evidence indicates that ferroptosis inducers may reshape the tumor microenvironment by inducing immunogenic cell death in malignant cells or reprogramming myeloid populations (monocytes/macrophages and neutrophils).^12,38,39 Collectively, these findings underscore the necessity of developing synergistic ferroptosis and immunotherapy approaches.

Multiple synergistic strategies involving the combination of ferroptosis induction with immune checkpoint blockade (ICB) to treat advanced solid tumors have been established. Specifically, preclinical models revealed that (1) targeting GPX4 inhibitors enhances ICB responsiveness in TNBC and colon cancer^10,40,41; (2) low-dose arachidonic acid improves the efficacy of ICB in melanoma⁴²; (3) targeting CPT1A induces ferroptosis and synergizes with ICB in lung cancer⁴³; (4) engineered cyst(e)inase formulations potentiate the effects of ICB in ovarian carcinoma⁴⁴; and (5) ferroptosis-inducing nanoparticle systems significantly increase the efficacy of ICB therapeutics.^45–48 Building on this research, CNSI-Fe can increase CD8⁺ T cells and M1-type macrophages in the tumor microenvironment while suppressing M2 polarization; this dual mechanism suggests their potential as a novel “ferro-immuno-nanomedicine” (Supplementary Fig. 7). To date, there are no prior human data on ferroptosis therapeutics; this registered clinical trial provides the first foundational evidence for a disruptive cancer treatment modality and extends beyond oncology to nanotechnology, immunometabolism and translational drug development.

CNSI-Fe administered by intratumoral injection is effective for treating ferroptosis-sensitive solid tumors, but this single-arm study has limitations and requires further evaluation through randomized phase II/III trials. In addition, the heterogeneous cohort and lack of biomarker-driven enrollment might restrict the validation of the generalizability of the CNSI-Fe. Considering the synergistic effects of CNSI-Fe and immunotherapy, combination treatment may represent a promising strategy and warrants further clinical investigation.

Materials and methods

Preparation and characterization of CNSI-Fe

Carbon nanoparticle suspension injection (CNSI) and ferrous sulfate for injection were produced under GMP conditions according to the laboratory-scale preparation method we previously reported,³³ with some necessary adjustments. Briefly, 1.00 kg of carbon nanoparticles, 0.06 kg of sodium citrate, 0.40 kg of sodium chloride, and 0.18 kg of poloxamer were added to water for injection to achieve a total weight of 20.52 kg, and a CNSI with a concentration of 50 mg/mL was prepared, with 2 mL filled in each vial. Using 2.24 kg of FeSO₄·7H₂O, followed by dilution with water for injection to achieve a total weight of 30.96 kg, ferrous sulfate was prepared for injection as a lyophilized powder with 30 mg of ferrous ions per vial. CNSI-Fe was freshly prepared by dissolving ferrous sulfate for injection with CNSI at the designated concentration immediately before clinical use.

To characterize the physicochemical properties of CNSI-Fe, we employed a comprehensive suite of analytical techniques. Field-emission transmission electron microscopy (TEM; Talos F200S, Thermo Fisher Scientific, USA) was utilized to analyze the morphological characteristics and dispersion state. The particle size distribution was determined by dynamic light scattering (DLS; Zetasizer Nano ZS90; Malvern Instruments, UK). Surface functional groups were identified using Fourier transform infrared spectroscopy (FTIR; Nicolet iS50, Thermo Fisher Scientific, USA). The characteristic peaks of the carbon materials were examined by confocal Raman spectroscopy (LabRAM HR Evolution, Horiba Scientific, France). The elemental composition was precisely analyzed through X-ray photoelectron spectroscopy (XPS; K-Alpha, Thermo Fisher Scientific, USA). We also measured the stability of CNSI-Fe within 6 hours of preparation, with evaluation indicators including appearance, pH, particle size and polydispersity index (PDI), zeta potential, and Fe²⁺ and Fe³⁺ contents.

Study design and participants

This first-in-human, single-arm, open-label, dose-escalation trial employed a 3 + 3 design to evaluate the safety, pharmacokinetics (PK), and preliminary efficacy of intratumorally administered CNSI-Fe in patients with refractory solid tumors. Conducted across three Chinese clinical centers (West China Hospital of Sichuan University, Shandong First Medical University Affiliated Cancer Hospital, and Enshi Tujia and Miao Autonomous Prefecture Central Hospital) from December 2022 to September 2024, the study protocol received approval from institutional ethics committees (Approval No.: 2022 Clinical Trial (Western Medicine) Review No. 238, SDZLEC2023-203-01, and 2023-003-01; ClinicalTrials.gov: NCT06048367; NMPA CTR: CTR20222235). All participants provided written informed consent prior to enrollment.

The key inclusion criteria were as follows: age 18–80 years; histologically/cytologically confirmed advanced solid tumor; disease progression after standard therapy or lack of effective standard treatment; ≥1 measurable lesion per RECIST v1.1; intratumoural injection (direct or imaging-guided) was possible; ECOG performance status ≤1; and adequate hematologic and organ function.

The exclusion criteria included iron metabolism disorders, hollow organ perforation risk at the injection site, and local skin abnormalities (ulceration, necrosis, hemorrhage) potentially interfering with drug administration.

Procedures

Dose escalation followed a 3 + 3 sequential cohort design with planned doses of 30, 60, and 90 mg according to the order in which informed consent was obtained. DLT criteria triggered protocol-specified adjustments: If ≥2 DLTs occurred at 90 mg, de-escalation to 75 mg would occur. If no MTD was identified at 90 mg, the Safety Review Committee (SRC) would evaluate escalation to 120 mg or 150 mg on the basis of cumulative safety, PK/pharmacodynamic data, and efficacy data. Treatment cycles spanned 21 days, with CNSI-Fe administered by imaging-guided (ultrasound/CT) intratumoral injection on day 1. The DLT observation window covered cycle 1. Patients deriving clinical benefit per the investigator’s judgment could receive a second injection in cycle 2.

Study assessments

Primary endpoints: Safety/tolerability, including DLT incidence and MTD determination. Secondary endpoints: PK profile of serum iron ions and preliminary evaluation of antitumour activity. Safety evaluations included the following: AE monitoring per CTCAE v5.0; physical examinations; vital signs; laboratory tests; ECOG performance status; and twelve-lead electrocardiogram (12-lead ECG) and echocardiography. DLTs were defined as follows: Grade ≥3 nonhaematologic toxicity; Grade 4 hematologic toxicity; and any AE leading to treatment discontinuation. MTD was defined as the highest dose at which ≤1/6 of the patients experienced a DLT.

PK analysis

Serum iron concentrations were quantified via inductively coupled plasma mass spectrometry (ICP-MS) at predefined intervals (before injection to 72 h after injection). Key PK parameters (C_max, T_max, AUC_0-t, AUC_0-∞, t_1/2) were derived using noncompartmental analysis for cycle 1 and cycle 2.

Efficacy evaluation

Tumor response was assessed by using modified RECIST ver1.1 and RECIST ver1.1. Initial assessment: Weeks 3-4 of cycle 1. Follow-up assessment: Weeks 7–8 for cycle 2 recipients. The objective response rate (ORR; complete [CR] + partial response [PR]) and disease control rate (DCR; CR + PR + stable disease [SD]) were calculated with two-sided 95% CIs (Clopper–Pearson method).

Statistical analysis

The safety population included all patients who received ≥1 CNSI-Fe dose. The efficacy population was composed of patients with ≥1 posttreatment tumor assessment. AE incidence and drug-related adverse reactions (ADRs) were recorded. SAS^® software version 9.4 or higher was used to perform the statistical analyses.

In vitro antitumour effects

To evaluate the antiproliferative effects of CNSI-Fe on cells and investigate the underlying mechanism, 4T1 TNBC cells were seeded in three 12-well plates (5 × 10⁴ cells/well), incubated for 24 h and then divided into nine groups (triplicate wells per group), namely, the control, CNSI control, and seven CNSI-Fe groups, with increasing concentrations of Fe²⁺ (6.25–400 μg/mL). After the culture medium was removed, fresh medium, CNSI-containing medium, and CNSI-Fe-containing medium ([Fe²⁺]: 6.25–400 μg/mL) were added to the corresponding plates, followed by a 48-h incubation. The cells were then collected and counted to calculate proliferation rates. For mechanistic studies, cells from the control, CNSI, and CNSI-Fe groups were subjected to comprehensive analyses.

Intracellular iron ions

To measure the amount of intracellular iron ions, the cells were placed in a microwave digestion tube and treated with 2 mL of nitric acid and 0.5 mL of hydrogen peroxide. The mixture was then subjected to microwave digestion at 240 °C for 30 min using an EXPEC 790S instrument (Expec Technology Development Co., Hangzhou, China). After cooling, the digested samples were transferred and diluted to a final volume of 50 mL. The iron ion concentration was subsequently determined using an Agilent 7850 ICP-MS (Agilent Technologies, USA).

Oxidative stress markers

To measure hydroxyl radicals, cells were lysed using an ultrasonic cell disruptor (SCIENTZ-IID, Scientz Biotechnology Co., Ningbo, China). After dimethyl-α-phenyl-N-tert-butylnitrone (DMPO) was added, electron spin resonance (ESR, JES-FA200, JEOL, Japan) analysis was performed.

Oxidative stress markers, including hydrogen peroxide (H₂O₂), peroxidase (POD), GSH, and MDA, were measured according to the instructions provided by the manufacturers of the respective assay kits (Solarbio Biotechnology Co., Beijing, China).

To measure GPX4 activity, the collected cells were completely lysed by using RIPA lysis buffer. The lysates were subsequently centrifuged, and the supernatant was retained. Protein samples were predenated and then loaded on sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) gels. After separation by electrophoresis, the protein bands in the gels were electrotransferred onto polyvinylidene fluoride (PVDF) membranes. The membranes were blocked with 5% skim milk for 1 h, followed by incubation with primary antibodies against GPX4 (Product No. GB124327; Wuhan Servicebio Technology Co., Ltd., China) and β-actin (Product No. GB15001; Wuhan Servicebio Technology Co., Ltd) at 4 °C for 3 h. The membranes were subsequently incubated with a secondary antibody (Product No. GB23301; Wuhan Servicebio Technology Co., Ltd., China) for 30 min. After the membranes were washed, chemiluminescence detection and analysis were performed. Data are presented as the mean ± SD. Statistical significance was defined by using Student’s t test, and p < 0.05 was considered significant.

Biological safety

In the 4T1 TNBC tumor-bearing mouse model, CNSI-Fe was administered via intratumoural injection twice. The mice in the control and CNSI-Fe groups were euthanized 21 days after the first administration of CNSI-Fe. Whole blood was collected for hematological parameter analysis by using a fully automatic blood cell analyzer (BC-2800vet, Mindray, China). The collected serum samples were utilized for biochemical index analysis by using a fully automatic biochemical analyzer (Chemray 800, Rayto, China). Simultaneously, the livers, hearts, lungs, spleens, and kidneys of mice were harvested, fixed in 10% (v/v) formalin, embedded in paraffin, sectioned, and finally stained with hematoxylin and eosin (H&E). The sections were then examined and photographed by an Eclipse Ci-L optical microscope (Nikon, Japan) for histopathological analysis.

Antitumour effects of CNSI-Fe in vivo

CNSI-Fe treatment of COLO205 colon cancer, HCC1954 breast cancer and MDA-MB-231 triple-negative breast cancer

COLO205 colon cancer, HCC1954 breast cancer, and MDA-MB-231 TNBC cells were inoculated into the right upper limbs of BALB/c nude mice. When the tumor volume reached 100–150 mm³, the mice were divided randomly into 6 groups (10 each), namely, the control group, CNSI group, 5‑fluorouracil (5-Fu)- or paclitaxel (PTX)-positive control group, and low-, medium-, and high-dose CNSI-Fe groups. The tumors were intratumourally injected with 50 μL of normal saline, 50 μL of CNSI, or 50 μL of CNSI-Fe. The concentration of CNSI was 50 mg/mL, and the ferrous ion concentrations of CNSI-Fe were 3.75, 7.5, and 15 mg/mL. The injections were administered twice per week for a total of two doses. The 5-Fu group received intraperitoneal injections at a dosage of 15 mg/kg for 21 consecutive days. The PTX group received intravenous injections at a dosage of 15 mg/kg once per week for a total of three doses. The tumor volume was measured twice per week after treatment using the formula: tumor volume = length × width²/2. The data are presented as the means ± SEMs.

CNSI-Fe treatment of Mia PaCa-2, Capan-2 and AsPC-1 pancreatic cancer cells

Mia PaCa-2, Capan-2, and AsPC-1 pancreatic cancer cells were separately inoculated into the right upper limbs of BALB/c nude mice. When the tumor volume reached 100–150 mm³, the mice were divided randomly into 4 groups, namely, the control group, PTX plus gemcitabine (GEM)-positive control group, medium-dose group, and high-dose CNSI-Fe groups (8 mice each). The tumors were intratumourally injected with 50 μL of normal saline or 50 μL of CNSI-Fe. The ferrous ion concentrations of CNSI-Fe were 7.5 and 15 mg/mL. The injections were administered twice per week for a total of two doses. The PTX plus GEM group received intravenous injections of PTX at a dosage of 20 mg/kg and intraperitoneal injections of GEM at a dosage of 120 mg/kg once per week for a total of four doses. The tumor volume was measured twice per week after treatment using the formula: tumor volume = length × width²/2. The data are presented as the means ± SEMs.

Flow cytometry detection

CT26. WT colon cancer tumor-bearing mice were intratumourally injected with CNSI-Fe twice. On day 17 after the first administration of CNSI-Fe, the mice in the control and CNSI-Fe groups were euthanized. Tumors were collected and prepared into single-cell suspensions for flow cytometry staining with corresponding antibodies (as shown in Supplementary Table 6). After staining, the samples were analyzed on a flow cytometer (FACS LRSfortessa™, BD, USA). FACS data were evaluated using Kaluza 2.1 software to assess the levels of relevant immune cell infiltrates in the samples.

PK detection after administration of CNSI-Fe

CT26. WT colon cancer tumor-bearing mice were divided into three groups and intratumourally injected with CNSI-Fe at volumes of 30 μL, 50 μL, and 60 μL. The ferrous ion concentration of CNSI-Fe was 15 mg/mL. Serum samples were collected from the mice before administration and at 0.033, 0.083, 0.25, 0.5, 1, 2, 4, 8, and 24 h after administration. The iron ion concentrations in the serum were detected using ICP-MS (8900; Agilent Technologies, USA), and drug concentration-time curves were plotted.

Animal experiment ethical approval

All animal experiments described in this study were conducted in accordance with the relevant institutional guidelines and regulations, and were formally approved by the Institutional Animal Care and Use Committees (IACUCs) of Shanghai Medicilon Inc. (Approval Nos.: YRY2101P, YRY2103P, YRY2104P, YRY2110P, YRY2111P, and YRY2112P), Crown Bioscience (Taicang) Co., Ltd. (Approval No.: AN-2204-05-726), West China-Frontier Pharma Tech Co., Ltd. (Approval No.: IACUC-A2019012-K001-03), and Sichuan Enray Pharmaceutical Sciences Company (Approval No.: YRLL-2022-A05).

Author contributions

Conception or design of the work: Y. Wang, P. Xie, Y. Huang, and Y. Chen. Acquisition, analysis or interpretation of data: H. Shi, X. Li., S. Wu, L. Lai, Z. Gou, Z. Liao, Q. He, W. Guo, Q. Dang, Z. Ding, Y. Chen, Z. Lin, and Y. Wen. Manuscript drafting and revision: H. Shi, X. Li, S. Wu, J. Xu, C. Huang, and Y. Wang. All authors have read and approved the article.

Data availability

Deidentified participant data are available in the ClinicalTrials.gov database (NCT06048367). The study protocol and all of the data generated and analyzed during this study are included in our manuscript and Supplemental Materials. Whole-genome sequencing (WGS) and deep whole-exome sequencing (WES) data generated from all samples have been deposited in the Genome Warehouse, National Genomics Data Center (NGDC), under accession number PRJCA022106.

Competing interests

P.X. is the CMO of Sichuan Enray Pharmaceutical Sciences Company. Y.H., and Y.C. are employees of Sichuan Enray Pharmaceutical Sciences Company. Y.W. is the editorial board member of the journal and he was not involved in the peer review or the decision making of this article. The other authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at https://doi.org/10.1038/s41392-026-02929-5.