MRI‐guided interventional natural killer cell delivery for liver tumor treatment

Abstract While natural killer (NK) cell‐based adoptive transfer immunotherapy (ATI) provides only modest clinical success in cancer patients. This study was hypothesized that MRI‐guided transcatheter intra‐hepatic arterial (IHA) infusion permits local delivery to liver tumors to improve outcomes during NK‐based ATI in a rat model of hepatocellular carcinoma (HCC). Mouse NK cells were labeled with clinically applicable iron nanocomplexes. Twenty rat HCC models were assigned to three groups: transcatheter IHA saline infusion as the control group, transcatheter IHA NK infusion group, and intravenous (IV) NK infusion group. MRI studies were performed at baseline and at 24 h, 48 h, and 8 days postinfusion. There was a significant difference in tumor R2* values between baseline and 24 h following the selective transcatheter IHA NK delivery to the tumors (P = 0.039) when compared to IV NK infusion (P = 0.803). At 8 days postinfusion, there were significant differences in tumor volumes between the control, IV, and IHA NK infusion groups (control vs. IV, P = 0.196; control vs. IHA, P < 0.001; and IV vs. IHA, P = 0.001). Moreover, there was a strong correlation between tumor R2* value change (∆R2*) at 24 h postinfusion and tumor volume change (∆volume) at 8 days in IHA group (R 2 = 0.704, P < 0.001). Clinically applicable labeled NK cells with 12‐h labeling time can be tracked by MRI. Transcatheter IHA infusion improves NK cell homing efficacy and immunotherapeutic efficiency. The change in tumor R2* value 24 h postinfusion is an important early biomarker for prediction of longitudinal response.


Introduction
Hepatocellular carcinoma (HCC) is the fifth most common malignancy in the world and the fourth leading cause of cancer death in the United States [1,2]. Surgical resection and transplantation are potentially curative treatments for HCC, but only 10-15% of patients are candidates [1,2]. The overall survival (OS) benefits from current liver-directed transcatheter approaches remain relatively modest [2][3][4].
While natural killer (NK) cell-based adoptive transfer immunotherapy (ATI) holds great promise for clinical cancer treatment, only modest clinical success has been achieved thus far using NK-ATI therapies in cancer patients [5,6]. Two challenges hurdle for NK-ATI in HCC patients include: (1) inadequate homing efficiency of NKs to the targeted tumors and (2) lack of well-established noninvasive tools for predicting the NK-ATI response [7][8][9].
We used a clinically applicable approach combining the FDA-approved drugs heparin, protamine, and ORIGINAL RESEARCH MRI-guidedinterventionalnaturalkillercelldeliveryfor livertumortreatment ferumoxytol to form HPF nanocomplexes for magnetic NK labeling so that NK cell biodistribution could be visualized in vivo with advanced magnetic resonance imaging (MRI) following transcatheter intrahepatic arterial (IHA) local delivery.
The purposes of our study was to test the hypotheses in a rat model of hepatocellular carcinoma: (1) clinically applicable labeled NK cells can be tracked with MRI, (2) transcatheter IHA NK infusion improves NK cell homing efficacy to targeted tumors, and (3) serial MRI monitoring of NK cell migration to targeted tumors can serve as an early biomarker for prediction of longitudinal response.

MaterialsandMethods
All studies were approved by our institutional animal care and use committee and were performed in accordance with the NIH guidelines.
HPFnanocomplexlabelingofLNKs HPF nanocomplexes (with 0, 50, 100, or 200 μg/mL ferumoxytol) in 0.5 mL phosphate-buffered saline (PBS) were added to each flask (5 × 10 6 LNKs in 10 mL culture medium), separately. Each flask was then incubated for 12 h. Labeled cells were harvested and washed with PBS and then with 10 U/mL heparin to remove the residual HPF [12]. Cells were then fixed for Prussian blue (Invitrogen, Carlsbad, CA) staining and transmission electronic microscopy (TEM) (FEI Company, Hillsboro, OR). Cell labeling efficiency (%) = number of positive-staining cells/number of total cells in per ten fields (20×) (S.Z. and Z.Z.). TEM images of LNKs were used to verify intracellular iron uptake. A separate set of samples was used to quantify iron content within each cell using inductively coupled plasma mass spectrometry (ICP-MS, Thermo Fisher Scientific Inc., Waltham, MA), as previously described [14] (L.Z. and Z.Z.).

LabeledNKcellviabilityandcytotoxicity functionassays
The cell viability assay methods were described in previous studies [14,15]. Aliquots with 5 × 10 5 cells each were resuspended in 100 μL annexin V binding buffer (BD Pharmingen, San Diego, CA) and incubated with 5 μL FITC-conjugated annexin V reagent (BD Pharmingen), either with or without 10 μL of 50 μg/mL propidium iodide solution (PI, BD Pharmingen) for 15 min in the dark. These cells were then analyzed using a FACScalibur flow cytometer (BD Biosciences, San Jose, CA), yielding percentages of apoptotic (annexin V positive, PI negative), necrotic, and viable cells. The study was repeated with 0, 50, 100, or 200 μg/mL ferumoxytol labeling protocol, and cell viabilities were assessed at 3 days after the 12 h labeling period (each group, N = 6).
NK cells were labeled with HPF nanocomplexes (50 μg HPF/mL with 10 6 NK cells for 12 h labeling). McA target cells were labeled with carboxyfluorescein succinimidyl ester (CFSE, Sigma-Aldrich, St. Louis, MO) to discriminate target cells from effector cells. Labeled and unlabeled or HPF-labeled NK effector cells were co-cultured with CFSE labeled McA target cells at different effectors to target (E/T) ratios at 25:1, 12.5:1, and 6.2:1 in 96-well U-bottomed microplates, separately. Additionally, two sets of controls containing only effector cells or only target cells were prepared. After overnight co-culture, the cells were harvested and stained with 7-AAD (BD Pharmingen) and pacific blue Annexin V for 15 min in the dark. These cells were then analyzed using a FACScalibur flow cytometer. All conditions were performed five times independently. NK cytotoxicity (%) is calculated as the number of cells positive for both CFSE and Annexin V divided by the total number of CFSE positive cells, after subtracting the spontaneous apoptosis (%) in target cell only controls (B.W., B.Z., Z.Z., and L.Q.) [16].

Animalmodel
Twenty male buffalo rats (body weight 250-300 g, Charles River Laboratories, Wilmington, MA) were anesthetized. A mini-laparotomy was performed, and 2 × 10 6 McA cells in 0.2 mL PBS were injected in the left median and left lateral lobes. Tumors were allowed to grow for 7 days to reach a size typically ≥5 mm in diameter verified by MRI. These rats were separated into three groups: transcatheter IHA saline infusion as the control group, transcatheter IHA LNK infusion group, and intravenous (IV) LNK infusion group (each group N = 6, 12 tumors, a total of 36 tumors included) (Z.S., T.L., X.W., M.F., and Z.Z.) [17,18].

TranscatheterIHALNKlocalinfusion
Transcatheter IHA infusion methods were described in previous studies [17][18][19]. In brief, the rat was anesthetized, and a mini-laparotomy was performed. The common hepatic artery was clamped to temporarily prevent bleeding, and a 4-0 suture was used to permanently ligate the gastroduodenal artery to prevent backwards flow. A 24-gauge microcatheter was placed in common branch of the proper hepatic artery with placement confirmed by digital subtraction angiography (DSA) guidance (Omnipaque, GE Healthcare, Pittsburgh, PA). Catheter was placed in the left hepatic artery, and 0.1 mL of heparin was infused before infusing 4.0 × 10 6 labeled LNKs (50 μg/mL HPF labeling 12 h) suspended in 0.3 mL PBS. After LNK infusion, abdominal incisions were closed. For the transcatheter IHA saline infusion control group, 0.5 mL of PBS was infused, comparable in volume to the LNK suspension. All other procedures were done by the exact same method as transcatheter IHA LNK infusion group (Z.S., T.L., X.W., M.F., and Z.Z.). IVLNKsystemicdelivery 4 × 10 6 labeled LNKs (50 μg/mL HPF labeling 12 h) in 0.5 mL of PBS were infused via tail vein followed by 0.2 mL of saline flush and 0.1 mL of heparin lock flush.

Assessmentoftherapeuticresponseusing T2Wimages
The volume of each tumor was measured on axial T2W images as previously described [20]. These measurements were performed for both baseline images and images collected at 8 days postinfusion. The change of tumor volume (∆volume) was calculated using the formula: ∆volume = volume 8 days − volume at baseline (Z.S. and X.W.).

Histology
After completion of MRI 8 days postinfusion, all rats were euthanized. Livers were harvested and fixed in 10% formalin, and then, tissues were embedded in paraffin. Sections including tumor tissues were sliced (4 μm) for CD56 (Anti-CD56, Becton Dickinson, San Jose, CA) immunohistochemistry (IHC) staining [17]. Resulting histology slides were reviewed by a pathologist at Pathology Core Facility (L.L., MD, who had more than 10 years of experience).

Imageanalysis
For MRI examinations, image analyses were performed using MATLAB (2011a, MathWorks, Natick, MA). T2*/ R2* maps were generated. Using both the T2W MR Images as anatomical reference to identify the tumor boundaries, region of interest (ROI) was drawn to completely cover the viable tumors to measure R2* values; four slices were included for each tumor for R2* measurements. ∆R2* values were calculated (∆R2* = tumor R2* at 24 h postinfusion − tumor R2* at baseline) (X.W. and Z.S.). CD56 stained slides (4 slices/each tumor) were scanned at a magnification of 20× and digitized using the TissueFAXS system (TissueGnostics, Los Angeles, CA). These acquired images were analyzed using the HistoQuest Cell Analysis Software (TissueGnostics) package to quantify the total number of LNK cells within each specimen (T.L., X.W., and Z.S.).

Statisticalanalysis
Data are presented as mean ± standard deviation (SD) as indicated. One-way ANOVA was used to compare in vivo R2* and tumor volume among animal groups (control, IHA, and IV) at baseline. t-tests were used to compare R2* and tumor volume of different groups measured at different time-points (baseline, 24 h, 48 h, and 8 days after infusion). Changes of R2* (∆R2*) and tumor volume (∆volume) were also compared among groups at different time points using t-test. Pearson's correlation coefficient (r) was also applied to assess the relationship between ∆R2* measurements and ∆volume measurements. Statistical analysis was performed using SAS 9.4 (SAS Institute, Inc., Cary, NC) by a biostatistician (K.L.).

NKcelllabelingefficiencyandironcontent ineachcell
Microscopic observation of Prussian blue-stained LNKs revealed extensive, nonuniform uptake of iron particles (Fig. 1A-D). The iron content of the unlabeled cells was significantly lower than that of the labeled cell groups (Fig. 1E). Labeling efficiency measurements using Prussian blue assays are 0% for control with HPF, 87 ± 7% for 50 μg/mL HPF group, 92 ± 5% for 100 μg/mL HPF group, and 95 ± 11% for 200 μg/mL HPF group. Uptake of iron nanoparicles was confirmed in labeled LNKs, but was not observed in unlabeled LNKs by TEM (Fig. 2A). The internalization of iron nanoparticles in the cytoplasm was confirmed and was not observed in the LNK cell membrane (Fig. 1B-C) with 0, 50, 100, or 200 μg/mL.

NKcellviabilityandcytotoxicityfunction
At 3 days after the 12-h labeling period, there were no significant differences in the percentages of apoptotic cells in all groups, while significant increases in percentages of necrotic cells were observed in 100 μg/mL and 200 μg/   mL compared to control group F ( Fig. 3A and B). For cytotoxicity function assay, there was no significant difference between labeled and unlabeled NK cells at all target effector ratios (Fig. 3C). The percentage of early apoptosis and late apoptosis of target cells caused by unlabeled and labeled NK cells also showed no significant difference at all ratios, and representative flow dot plots of (Fig. 3C) at 25:1 (E/T) are shown in Figure 3D.

InvitroMRIofphantoms
The T2* signal-to-noise ratio (SNR) of the LNK cell phantoms containing different numbers of LNKs labeled with 50 μg/mL HPF markedly decreased with increasing cell number (Fig. 4A). SNR values decreased in proportion to the number of labeled LNKs included in each cell phantom in Figure 4B. R2* values increased in proportion to the number of labeled LNKs included in each cell phantom in Figure 4C.

InvivoquantitativeMRImeasurements
One rat displayed no growth (no tumor detected) and thus was excluded from the study. Another rat was excluded because of poor health at baseline imaging time-point.
Following tumor implantation at 7 days, tumor volume was measured in T2W images. The mean baseline tumor volumes for control, IV, and IHA groups were 65.69 ± 18.10 mm 3 , 64.35 ± 18.26 mm 3 , and 64.07 ± 11.53 mm 3 , respectively. The tumors were depicted with relatively homogeneous high signal intensity before the delivery of labeled NKs in baseline T2W and T2*W images. Representative axial T2W images, T2*W images, and R2* maps from one of the IHA group animals were shown (Fig. 5A).
Identification of the IHA and its subsequent catheterization was confirmed by DSA, and the success rate was 100%. For IHA LNK group, both 24 h and 48 h postinfusion signal intensities within the tumor nodules became heterogeneous. Figure 5B shows: (1) There were no significant differences in baseline tumor R2* values of the three groups. A significant difference in R2* values was found between baseline and 24/48 h after IHA LNK cell infusion within tumor tissues, but there was no significant difference in tumor R2* values between the 24 h and 48 h follow-up period and (2) there were no significant differences in R2* values between baseline and 24/48 h; between 24 h and 48 h after IV LNK cell infusion, but increasing R2* value trends were observed between baseline and 24/48 h. However, there were significant differences in tumor R2* values between control, IV, and IHA groups at both 24 h and 48 h postinfusion time-points (Fig. 5B).

Therapyresponseandhistology
The efficacy of the LNK cell therapy was evident as inhibition of tumor growth in the IHA group when compared to that in control/IV animals 8 days after infusion (∆volume: control vs. IHA, P < 0.001; IV vs. IHA, P = 0.001). There were no significant differences in therapeutic responses between the control group and IV group animals 8 days after infusion (P = 0.196) (Fig. 6A). No significant differences were observed in ∆volume between the control  LNKs measured within tumors for the IHA group than for the IV systemic delivery group and control group. There were also a significantly larger number of positively stained LNKs measured in the IV group than those in the control group.

Discussion
Our study demonstrated the following: (1) clinically applicable HPF-labeled NKs can be tracked with MRI, (2) transcatheter IHA NK infusion offers the potential to radically augment NK cell homing efficiency to liver tumors, and (3) the change of tumor R2* value at 24 h postinfusion is an important early biomarker for prediction of longitudinal response.
NK-ATI is a promising approach for the treatment of solid tumors including HCC [21,22]. However, conventional intravenous (IV) delivery of NKs may limit the number of cells that ultimately reach the target tumor(s) and surrounding liver tissues [17,22,23]. Because hepatic arteries supply ~90% of the blood flow in HCC [24], the goal of transcatheter IHA delivery is to selectively infuse NKs to targeted HCC lesions for improved therapeutic efficacy. Transcatheter IHA infusion methods were described in previous studies [17][18][19] which used human NK cell line (NK-92MI) with FeOlabel Texas Red (Genovis AB, Lund, Sweden) nanoparticles labeling. However, FeOlabel Texas Red nanoparticles are not approved by FDA for clini settings. This study design has a strong focus on clinical translation. All methods and materials proposed in this study are approved by the FDA for direct clinical uses, and our results are directly applicable to humans.
Furthermore, in vivo noninvasive tracking of NK cell migration to tumors will be critical in predicting early therapeutic responses [25][26][27][28]. We used a clinically applicable approach of combining FDA-approved drugs to form HPF nanocomplexes for magnetic NK labeling so that NKs could be visualized by MRI [12,29]. The HPF nanocomplexes have similar biochemical properties to iron nanoparticles, which have been used previously to label cells through the iron metabolic pathway [12,29]. We optimized the labeling protocol, which showed significant uptake of HPF with minimal effect on cell viability and cytotoxicity. Our observations were in agreement with previous studies [12,29]. In our study, the doses of 50, 100, and 200 μg/mL HPF were used for NK labeling, resulting in high labeling efficiency. However, iron uptake per cell was dependent upon HPF concentrations. A significant increase in cell death was found in both 100 μg/mL and 200 μg/mL HPF labeling groups when compared to the control group. Therefore, the dose of 50 μg/mL HPH was selected for our study. The 12-h labeling time with 50 μg/mL HPf resulted in 87 ± 7% labeling efficiency, with no effect on NK cell viability and cytotoxicity. This labeling approach and 12-h labeling period may be more conducive to clinical translation during NK-ATI trials. This limited-duration labeling period could potentially avoid NK function alterations and cell damage.
In vivo MRI shows extensive HPF-labeled NKs homing to the targeted tumor shortly after transcatheter IHA NK delivery. A tumor R2* increase was observed within 24 h of NK delivery and correlated strongly with therapeutic responses (tumor growth inhibition 8 days after NK infusion). Quantification of NK measurements showed that the highest amounts of CD56-positive NKs in tumors were present in the IHA NK group, while much fewer were present in the IV NK group with the least NKs evident in the IHA saline group.
Our study had several limitations. MR imaging of HPFlabeled NK homing to targeted tumors was performed at only two time-points after injection (24 h and 48 h), and the study protocol examined only one NK dose that was selected from our initial experiences. Additional studies may be valuable to determine the optimal doses and postinjection time intervals for follow-up imaging measurements.
In summary, our study demonstrated that (1) NK cells labeled with clinically applicable approaches can be tracked with MRI, (2) transcatheter IHA NK infusion improves NK cell homing efficacy to targeted tumors, and (3) serial MRI monitoring of NK cell migration to targeted tumors may serve as an important early biomarker for prediction of longitudinal response.