Systemic intravenous delivery of immunotherapies can cause off-target immune-related toxicities in multiple organs. Local delivery directly into tumor lesions aims to increase intratumoral drug concentration, improve local bioavailability, modulate the tumor microenvironment, and reduce systemic exposure and toxicity. Intratumoral therapy may also elicit systemic antitumor responses (abscopal effects) by activating effector immune cells that traffic to distant sites. This pooled prospective analysis sought to evaluate the safety and preliminary systemic effects of CT-guided intratumoral injection of immunotherapeutic agents in patients with advanced solid tumors.
This study is a pooled analysis of multiple prospective, single-arm, non-randomized, open-label clinical trials conducted at the Second Affiliated Hospital (Panyu Campus) of Guangzhou Medical University. The Ethics Committee approved the protocol. ClinicalTrials.gov identifiers for included trials are reported in the original publications and supplementary materials. Data were collected from March 1, 2018, to December 31, 2024, and all participants provided written informed consent.
Eligible adults (≥18 years) had histologically confirmed advanced solid tumors refractory or intolerant to standard therapies and at least one measurable lesion accessible for imaging-guided puncture. Key exclusions included severe organ failure, serious comorbid conditions, prior similar interventions, inability to comply with procedures, and history of severe allergic reactions to study drugs. The pooled cohort comprised 169 patients (after excluding 33 screened patients who underwent biopsy only). Mean age was 55 years (range 20–87); 66.86% were male. ECOG scores ranged from 0 to 3. Cancer types included lung (26.63%), liver (24.26%), gastrointestinal (17.75%), pancreatic (9.47%), and other histologies.
Experienced interventional physicians performed all procedures under CT guidance. Needle caliber selection (commonly 21G or 23G) depended on lesion size and skin-to-lesion distance. The needle was positioned into the lesion center and the planned dose administered; large lesions received multiple needles to optimize distribution. Visualisation of intratumoral distribution used a 10-fold dilution of iodinated contrast (e.g., ioversol or iodixanol) mixed with immunotherapy in selected cases. Intratumoral immune checkpoint inhibitor (ICI) dosing was standardized to half the approved intravenous dose per drug label. CAR-T cell therapy was administered intratumorally at 1 × 10^6 cells/kg. Treatments were scheduled every three weeks (Q3W) until progression, unacceptable toxicity, or withdrawal. Local anesthesia (1% lidocaine) and continuous vital sign monitoring were used.
Safety (primary endpoint) was assessed using CTCAE v5.0. Adverse events (AEs) were recorded at each follow-up and adjudicated for relation to the injection procedure or the agent. Clinical data included exams, ECOG performance, vitals, and laboratory tests. Tumor assessments occurred approximately every three weeks using consistent imaging and measurement techniques. Treatment-related AEs were documented on case report forms.
Secondary endpoints included progression-free survival (PFS) and overall survival (OS). Tumor responses were assessed per RECIST v1.1 and classified as complete response (CR), partial response (PR), stable disease (SD), or progressive disease (PD). Objective response rate (ORR) was CR plus PR; disease control rate (DCR) was CR, PR, or SD. PFS was time from intratumoral injection to radiographic progression or death; OS was time from injection to death from any cause.
Analyses were descriptive due to the single-arm, exploratory nature and heterogeneity across trials, agents, and tumor types. Continuous non-normally distributed variables were summarized as medians (ranges) with 95% confidence intervals where reported; categorical variables were reported as percentages. Median follow-up was estimated by the reverse Kaplan–Meier method. Survival curves and summaries were generated using GraphPad Prism 10.
A total of 878 CT-guided intratumoral injection procedures were performed across 169 patients. Target lesion mean length was 3.7 cm (range 1.0–31.5 cm). Most frequently used needle sizes were 21G (40.89%) and 23G (45.90%); larger-gauge needles (17–20G) were used in 8.88% of procedures. Injection sites included liver (30.07%), abdominal cavity (13.10%), lung (12.76%), subcutaneous (9.80%), and other locations. Drug leakage into surrounding tissue occurred in 106 procedures (12.07%), typically confined to the peritumoral region and not requiring further intervention; no severe complications such as organ infarction or major bleeding were attributed to leakage.
Among 169 patients, 15 (8.88%) experienced grade 3–4 treatment-related adverse events (10 grade 3; 5 grade 4). No treatment-related deaths occurred. The most common any-grade AEs were fever (60.36%), local puncture-site pain (54.44%), chills (18.34%), and vomiting (14.79%). Most AEs were mild, transient (often resolving within 24 hours), and managed with symptomatic treatment when necessary. Selected grade 3–4 events included fever (13 patients grade 3–4 combined), local pain (7 grade 3), chills (4 grade 3–4), hypotension (5 grade 3–4), immune-related pneumonitis (4 grade 3–4), immune-related rash (3 grade 3), and respiratory failure (3 grade 3). Laboratory abnormalities such as leukocytopenia were reported less frequently.
Efficacy outcomes were descriptive. Best overall responses in the pooled cohort were 4 patients (2.37%) with CR, 15 (8.88%) with PR, 142 (84.02%) with SD, and 8 (4.73%) with PD. The ORR was 11.24% and the DCR was 95.27%. Median PFS was 3.6 months (95% CI, 3.1–4.1 months) and median OS was 8.8 months (95% CI, 8.2–9.3 months). No formal comparative statistics between subgroups or regimens were performed.
Exploratory descriptive subgroup analyses were conducted with a minimum reporting threshold of 10 cases. Tumor-type subgroups meeting the threshold included lung (n=45), liver (n=41), gastrointestinal (n=30), and pancreatic (n=16). Treatment-regimen groups meeting the threshold included PD-1/CTLA-4 inhibitors (n=75), PD-L1/CTLA-4 inhibitors (n=53), and CAR-T cells (n=21). Because of heterogeneity and the single-arm design, no formal statistical comparisons were made and subgroup results were reported descriptively.
This pooled prospective cohort indicates that CT-guided intratumoral injection of ICIs or CAR-T cells is feasible across a range of tumor sites and has an acceptable safety profile, with most adverse events being mild and self-limited. Drug leakage occurred in a minority of procedures and was generally clinically inconsequential. Descriptive efficacy signals included an ORR of 11.24% and a high DCR (95.27%), with median PFS of 3.6 months and median OS of 8.8 months in this heavily pretreated population. The authors conclude that intratumoral injection may mitigate systemic toxicity while achieving local drug delivery and potential systemic antitumor effects, but emphasize that further research is required to validate therapeutic efficacy and to standardize procedural protocols and dosing.