All animal experiments were performed under Institutional Animal Care and Use Committee (IACUC) approved protocol (IACUC approval no: 20–129 LF). Eighteen Yorkshire pigs were divided into 3 groups: controls (n = 6), MSC group (n = 6), and MSC with nanocarrier (MSC + NC, n = 6). All animals underwent RFA of liver. The test groups (MSC and MSC + NC) had infusion of MSCs or MSCs with nanocarrier, respectively, via the hepatic artery 24 h after ablation. The control group had no MSC infusion.

Power calculation assumed that the number of ablations within a pig will be four and the expected rates of liver tissue regeneration in MSC and MSC + NC group will be 5% and 85%, respectively. These numbers were chosen according to our previous experience and similar numbers were used in other liver ablation studies [14]. We also assumed that correlation among ablation sites within a pig is 0.5, thus 6 pigs in each group will give 96.3% power to detect difference of 80%. Statistical analysis for all studies was performed using the IBM SPSS software. The independent sample T-test was used for analysis of continuous variables and one-way ANOVA tests were used for categorical variables with significance level set at 0.05.

The procedures for the preparation of nanocarrier-coated stem cells are detailed in Online supplementary materials.

All animal procedures were performed by a board-certified interventional radiologist with swine expertise. RFA was conducted using the Covidien Cool-Tip ablation system (Covidien, Boulder, Colorado) under US guidance on the CT table with general anesthesia. After confirming anesthesia depth, an initial CT scan was done. Based on liver lobe size and anatomy, 2 to 4 ablations of 2 cm diameter were planned per animal. A Covidien Cool-Tip, 11-gage ablation probe was inserted under US guidance, and RFA was performed for 6 min at 80 Watts in impedance control mode. Immediately after, a contrast-enhanced triple-phase CT scan of the liver was obtained using Omnipaque-180 (388 mg of iohexol per mL) for IV contrast (1.4 mL/kg).

The control animals had no additional procedures after ablation. The MSC and MSC + NC groups underwent intraarterial stem cell delivery the day after ablation under general anesthesia. After confirming anesthesia depth, a 5-French vascular sheath was placed in the common femoral artery. A 5-French Cobra catheter was advanced into the abdominal aorta, and a microcatheter (2.8 French Progreat, Terumo, Japan) was used to catheterize the proper hepatic artery. The MSC group received approximately 1 × 106 plain MSCs in 5 mL of PBS, injected into the hepatic artery. The MSC + NC group received approximately 1 × 106 MSCs coated with nanocarriers. Hemostasis was achieved by manual compression after catheter removal.

All animals survived for 30 days, except the first two control animals. One died the day after ablation, and the other was euthanized due to severe distress. Necropsy revealed large areas of ablation but no direct cause of death, likely due to device malfunction related to power settings.

All animals had blood tests (liver function, renal function, and complete blood count) on days 0 (prior to ablation), 15, and 30. On day 30, they received a contrast-enhanced liver CT scan under anesthesia, were euthanized, and underwent necropsy for liver pathology analysis.

The ablation cavity on contrast enhanced CT was defined as the area that does not enhance compared to the rest of the parenchyma. The ablation cavity volume was measured by CT using 3D Slicer (Brigham Women’s Hospital, Boston, MA), a free, open-source medical image computing software [15]. Two board-certified radiologists with at least 5 years of experience independently contoured the ablation cavities. Both reviewers were blinded to the identity of the groups. The reduction in ablation cavity volume at 1 month was used as a surrogate marker of liver regeneration, which was compared between the groups [16].

Cell proliferation at the margins of the ablation cavity was measured by immunohistochemical staining of formalin-fixed paraffin-embedded (FFPE) tissue sections using the proliferation marker Ki67 (Clone K2, Lot# 70,036, Leica Biosystems Inc, Buffalo Grove, IL). The HepPar-1 immunohistochemical stain (Clone OCH1E5, Lot# 69,056, Cell Marque Corp, Rocklin, CA) was used to confirm and localize areas of hepatocyte-specific regeneration. The GFP-transduced cells were identified from FFPE tissue sections using a fluorescent microscope. For all markers, the three areas with the highest count of positive cells (“hotspots”) were selected at medium and high magnifications for each ablation cavity sample. ImageJ software (https://imagej.nih.gov/ij/) quantified positive cells in the three hotspots, and the counts were averaged to obtain a final count per ablation cavity. GFP positivity was graded for all specimens using a 1–4 Likert scale based on GFP positive cells per HPF, as visually analyzed by a blinded pathologist. Likert score was defined as- score 1: < 25% cells/HPF are GFP+, 2: 25–50% of cells/HPF are GFP+, 3: 50–75% of cells/HPF are GFP+, 4: 75–100% of cells/HPF are GFP+. A similar Likert scoring was also used for HepPar+ cells. All histopathological evaluations were performed by a board-certified pathologist who was blinded to the identity of the groups.

The liver function tests on day 15 and 30 were also compared between test and control groups. Complete blood count, renal function tests, and prothrombin time/international normalized ratio (PT/INR) were monitored for each group to identify any systemic effect or toxicity.