As islet transplantation becomes an acceptable clinical modality for restoring normoglycemia in type 1 diabetic patients, there is a crucial need for noninvasive assessment of the fate of the grafts. In spite of the success of the Edmonton Protocol, a significant graft loss occurs due to immunological and nonimmunological events immediately after transplantation. Allogeneic rejection in graft recipients is one of the major reasons for islet death and graft failure. Therefore, monitoring the islet rejection using reliable noninvasive methods would significantly aid in clinical assessment of graft success. We have previously developed a method to detect transplanted islets noninvasively using magnetic resonance imaging (MRI). For this procedure, human pancreatic islets are labeled with an MRI contrast agent that enables their visualization on magnetic resonance images. In our present study, we not only detected labeled human islets in a preclinical intrahepatic model of human islet transplantation in mice but also showed that islet rejection can be monitored noninvasively and repeatedly in real time by MRI. In addition, in this study, we have adapted, for islet cell labeling, a Food and Drug Administration–approved commercially available contrast agent, Feridex, that is used clinically for liver imaging. We believe that this agent, in combination with our preclinical model of islet transplantation, will facilitate the transition of imaging immune rejection to clinical trials.
- FDA, Food and Drug Administration
- MRI, magnetic resonance imaging
- TUNEL, transferase-mediated dNTP nick-end labeling
Pancreatic islet transplantation has recently emerged as an effective therapy and potential cure for patients with type 1 diabetes (1–3). Studies in experimental models and recent clinical trials have shown that islet transplantation may represent a superior alternative to insulin injections since transplantation, unlike insulin therapy, results in normalization of metabolic control (4). For clinical transplantation, islets are infused in the liver through the portal vein while the subject is under local anesthesia. For an average-size person (70 kg), a typical transplant requires about 1 million islets, extracted from two donor pancreases (5). Allogeneic human islet rejection and/or recurrence of autoimmunity are the two major challenges hampering successful transplantation. Significant improvement in this area was achieved by researchers from the University of Alberta, who introduced an improved antirejection regimen, which became known as the Edmonton Protocol (1). In a 5-year follow-up report from this group, it is emphasized that the results, though promising, still point to the need for improving islet engraftment and preserving islet function (6).
It is clear at this point that immunological and nonimmunological events, which take place immediately after transplantation, lead to significant graft loss. Consequently, the fate of islets after transplantation is mostly unknown. In spite of the sophisticated immunosuppression regimen, a significant number of pancreatic islets die during the first 10–14 days after the procedure. In fact, even in syngeneic models of islet transplantation, a 60% loss of transplanted islets was reported in this early postprocedure period (3). The causes for islet death, besides immune rejection, are numerous and include mechanical injury, ischemia, nonalloantigen-specific inflammatory events in the liver after transplantation, and recurrent autoimmunity. Mediators of islet dysfunction include the inflammatory cytokines interleukin-1β, interferon-γ, and tumor necrosis factor-α (7–9). Islets exposed to allogeneic blood undergo instant blood-mediated inflammatory reaction elicited by tissue factor, which is produced by the endocrine cells and involves coagulation and complement activation (10–12). In addition, there is evidence of “metabolic” rejection of intraportal islet transplants involving lipotoxic destruction of islets by hepatic lipids (13,14). Furthermore, elevated levels of apoptosis have been shown in pancreatic islets exposed to chronic hyperglycemia immediately after transplantation (15). At present, however, immune rejection of pancreatic islets presents the biggest challenge in transplantation. A considerable fraction of islet transplants undergo chronic failure due to immune rejection, despite the novel use of rapamycin-based immunosuppressive agents and other means of prevention of allogeneic human islet rejection (4). Regardless of the specific immunosuppressive strategy used after transplantation, there is a critical need for monitoring islet rejection using reliable noninvasive methods.
In vivo magnetic resonance imaging (MRI) is a powerful modality that allows for acquiring images at near-microscopic resolution with simultaneous detection of anatomical and physiological parameters (16). It can provide important information regarding graft location and survival after transplantation. Recently, we have reported on the in vivo imaging of human pancreatic islets labeled with a magnetic resonance contrast agent and transplanted under the kidney capsule in a mouse model of type 1 diabetes (17). We showed that, using this model, labeled pancreatic islets could be reliably detected in vivo by MRI. Furthermore, we demonstrated that labeled islets were capable of restoring normoglycemia in diabetic animals with the same efficiency as their unlabeled counterparts. The basis for our present report builds upon our previous findings with the ultimate goal of translating this method into clinical practice. Therefore, the goal of this study was twofold. First, for clinical application, we needed to adopt the clinically relevant model of intrahepatic transplantation and demonstrate the possibility of in vivo detection of transplanted islets in this organ. Second, using this model, we intended to detect an input of immune rejection after transplantation in immunocompetent animals by in vivo MRI. As mentioned above, there are multiple reasons for islet death after transplantation. Here, we specifically focused on immune rejection since it presents the biggest challenge in clinical islet transplantation (4). In our previous work, we used superparamagnetic iron oxide nanoparticles synthesized in our laboratory as a magnetic resonance contrast agent for islet labeling (17). Here, we utilized the Food and Drug Administration (FDA)-approved commercially available contrast agent Feridex, which is routinely used in clinics for liver imaging. We adapted it for islet labeling, which we believe would facilitate the transition of this imaging method into clinical practice.
As a result of this study, we demonstrated that intrahepatic islet transplantation could be visualized using in vivo MRI in immunocompetent and immunocompromised mice. Furthermore, islet death was detected by MRI in both models immediately after transplantation. However, there was a clear acceleration of this process in immunocompetent animals pointing to severe immune rejection. We anticipate that these studies could make a significant impact on allotransplantation and the development of new and improved immunosuppressive regiments.
RESEARCH DESIGN AND METHODS
Labeling human pancreatic islets with an FDA-approved commercially available contrast agent.
Human pancreatic islets were obtained through The Islet Cell Resource Center at the National Institutes of Health and the Islet Distribution Program at the Juvenile Diabetes Research Foundation. Islets were shipped overnight to Massachusetts General Hospital, handpicked and cultured in Miami Medium #1 culture media (Cellgro; Mediatech, Herndon, VA), and supplemented with 20 mg/l ciprofloxacin hydrochloride (Fisher Scientific, Pittsburgh, PA) and 10 mg/l l-glutathione (Sigma, St. Louis, MO).
For labeling experiments, we used superparamagnetic iron oxide nanoparticles Feridex (ferumoxide; Advanced Magnetics, Cambridge, MA), which are FDA approved for human use as a liver imaging contrast agent. Here, we utilized it for labeling human pancreatic islets. One thousand human pancreatic islets were incubated overnight with Feridex in culture medium (200 μg iron/ml). After incubation, islets were washed with culture medium three times and used for in vitro studies or islet transplantation.
To calculate the amount of probe associated with the islets, we utilized an iron binding assay. One hundred pancreatic islets were incubated overnight with increasing concentrations of Feridex (10–300 μg iron/ml). After incubation, islets were washed three times with PBS, resuspended in 6N HCl, and incubated at 70°C for 30 min. The iron content of labeled islets was determined using a total iron reagent set (Pointe Scientific, Canton, MI). Incubation with each iron concentration was repeated five times.
Insulin secretion and viability of Feridex-labeled islets.
Insulin secretion was evaluated using static incubation of labeled and nonlabeled islets at low and high glucose concentrations. Briefly, islets were washed in CMRL 1066 media without glucose (Cellgro), supplemented with 0.5% BSA, radioimmunoassay grade (Sigma), 25 mmol/l HEPES, and 1.7 mmol/l glucose and preincubated for 60 min in the same medium. The medium was then discarded and replaced by fresh medium containing 2.5 mmol/l glucose for 1 h for basal secretion, followed by an additional 1-h incubation at 16.7 mmol/l glucose. Supernatants were collected and frozen for insulin assays. Thereafter, islets were washed with PBS and extracted with 0.18 N HC1 in 70% ethanol for 24 h at 4°C. The acid-ethanol extracts were collected for determination of insulin content and to normalize the values measured for insulin secretion. Insulin concentration was measured using a human insulin enzyme-linked immunosorbent assay kit (Mercodia, Uppsala, Sweden). A stimulation index was calculated as the ratio of stimulated to basal insulin secretion normalized by the insulin content. Nonlabeled islets from the same batch served as controls in all experiments. Results are representative of three separate experiments.
Islet viability after labeling was assessed by a standard 3-[4,5-dimethylthiazol-2-yl]-2,5-diphenyltetrazolium bromide assay (Promega, Madison, WI) and compared with that of unlabeled islets.
To measure apoptosis in Feridex-labeled and unlabeled islets, we used a fluorometric caspase-3 assay kit (Sigma). Human pancreatic islets were incubated with 200 μg iron/ml Feridex for 0, 1, 2, 4, 6, and 24 h, lysed, and assayed according to the manufacturer’s protocol. Incubation with 1 μg/ml staurosporine served as a positive control.
In vitro imaging of Feridex-labeled islet phantoms.
To show that labeling with Feridex produces a change in signal intensity on T2-weighted images, we performed MRI of islet phantoms. Labeled and nonlabeled pancreatic islets were placed in Eppendorf tubes (200 islets/tube) and allowed to settle on the bottom by gravity sedimentation. MRI was performed using a 9.4T Bruker horizontal bore scanner (Billerica, MA) equipped with ParaVision 3.0.1 software. Acquisition of T2-weighted spin echo pulse sequences was based on the following protocol: TR/TE = 2,000/15, 30, 45, 60, 75, 90, 105, 120 ms; FoV = 4 × 4 cm2; matrix size = 256 × 256; resolution = 156 × 156 μm; slice thickness = 1 mm; and imaging time of 8 min 32 s.
Mice and islet transplantation procedure.
All animal experiments were performed in compliance with institutional guidelines and approved by the subcommittee on research animal care at the Massachusetts General Hospital.
Pancreatic islet transplantation was performed using immunocompromised (NOD.scid, n = 12) and immunocompetent (Balb/C, n = 8) mice. One thousand Feridex-labeled human islets were infused through the portal vein in anesthetized animals. Control animals were transplanted with nonlabeled islets.
In vivo MRI of transplanted islets.
In vivo MRI was performed using a 4.7T Bruker horizontal bore scanner (Billerica, MA) equipped with ParaVision 3.0.1 software. Imaging was performed on days 1, 2, 3, 4–5, 6–7, 10–11, and 14–15 after transplantation The imaging protocol consisted of T2*-weighted gradient echo pulse sequences with the following parameters: TR/TE = 200/8 ms, number of averages = 32, FOV = 3.2 × 3.2 cm2, matrix size = 256 × 256, resolution = 0.125 × 0.125 mm2, slice thickness = 0.5 mm, and a total scan time of 27 min 18 s.
Labeled pancreatic islets (or islet clusters) appeared as distinct dark signal voids in the liver. To compare islet death in immunocompetent and immunocompromised animals, these signal voids were manually scored in the livers of all imaged animals (13 slices each) for each day of imaging. On every day of imaging, after each MRI session, we killed one of the imaged mice to use for ex vivo histology. This study was performed by two independent investigators (N.V.E. and S.L.). The number of islets on day 1 was considered 100%. As a result, our data are based on time-course longitudinal measurements of the same set of animals, where this set was reduced by one on each subsequent day of imaging.
To analyze our imaging findings at the microscopic level, we excised livers from the animals at each day of imaging. Mouse liver samples were embedded in Tissue-Tek OCT Compound (Sakura Finetek, Tokyo, Japan), snap frozen in liquid nitrogen, and axial sections prepared at a thickness of 7 μm.
The in vivo immune rejection of transplanted islets was evaluated semi-quantitatively by immunohistochemical staining of liver sections with antibodies targeting various types of immune cells. Cryosections were fixed in 4% paraformaldehyde and incubated with primary antibody for 1 h followed by biotinylated rabbit anti-rat IgG (H+L) secondary antibody (Vector) and Vectastain Elite ABC kit (Vector). Bound peroxidase was developed with a 3,3′-diaminobenzidine SIGMAFAST kit (Sigma). Some sections were processed for Prussian Blue staining to identify the presence of iron in transplanted islets. Briefly, sections were immersed in Prussian Blue solution containing 5% potassium ferrocyanide (ACROS Organics, Fairlawn, NJ) and 5% hydrochloric acid (Aldrich, Milwaukee, WI) for 30 min and counterstained with nuclear fast red (Sigma-Aldrich). Other sections were counterstained with hematoxylin. The primary antibodies used for this study are summarized in Table 1.
To determine which types of cells were labeled with Feridex in the islet, we performed colocalization studies by staining for islets hormones, resident macrophages, and iron nanoparticles. For these studies, we used mouse anti-human CD68 (Y1/82A) antibody (BD Pharmingen, San Diego, CA), rabbit anti-human insulin polyclonal antibody (Santa Cruz Biotechnology, Santa Cruz, CA), rabbit anti-porcine glucagon antibody, and rabbit anti-human somatostatin polyclonal antibody (MP Pharmaceuticals, Irvine, CA), followed by corresponding secondary biotinylated goat anti-mouse or anti-rabbit IgG (H+L) (Vector). Prussian Blue staining was performed as described above.
In situ apoptosis detection.
To evaluate levels of apoptosis in transplanted pancreatic islets, we performed a terminal deoxynucleotidyl transferase-mediated dNTP nick-end labeling (TUNEL) assay on frozen liver sections (Apoptag Fluorescein In Situ Apoptosis Detection kit; Chemicon International, Temecula, CA) according to the manufacturer’s protocol. The nuclei were counterstained with VECTASHIELD Mounting Medium with DAPI (Vector), and slides were examined using a Nikon Eclipse 50i fluorescence microscope equipped with an appropriate filter set (Chroma Technology, Rockingham, VT). Images were acquired using a charged-coupled device camera with near-infrared sensitivity (SPOT 7.4 Slider RTKE; Diagnostic Instruments, Sterling Heights, MI) and analyzed using SPOT 4.0 advanced version software (Diagnostic Instruments). The percentage of cells undergoing apoptosis within each islet was calculated from the number of islet nuclei positive for DNA fragmentation versus the total number of cells present as observed by two independent investigators (N.V.E. and S.L.).
The data from the iron internalization assay were expressed as means ± SD. The results from the time course of the iron internalization assay were analyzed by ANOVA. For the MRI data, the number of dark signal voids in the livers of immunocompromised and immunocompetent transplanted animals was compared for each day of imaging by a paired Student’s t test (SigmaStat 3.0; Systat Software, Richmond, CA). A value of P < 0.05 was considered statistically significant.
Human pancreatic islet labeling with the FDA-approved contrast agent Feridex.
To test whether human pancreatic islets can be labeled with Feridex, we incubated them with the contrast agent and subjected them to MRI. We compared labeled and unlabeled islets and found a significant decrease in signal intensity on T2-weighted images in labeled islets (Fig. 1A). By contrast, we observed no differences in signal intensity when we imaged the layer of medium directly above the islets. These results indicate that the signal indeed originated from Feridex-labeled human pancreatic islets and that the labeling produced a change in signal intensity on T2-weighted images. Iron uptake by islet cells resulted in 1.21 ± 0.36 to 18.26 ± 1.36 pg iron/islet cell, considering that an average pancreatic islet consists of 2,000 islet cells (Fig. 1B). As we show later, iron distribution within islet cells was not uniform, and these numbers reflect only average iron accumulation. These numbers were similar to those obtained with the “in-house” made nanoparticles used in our previous study (17) and to the reported levels of nonspecific iron uptake by other specialized cells (19–22). Time course studies showed an overall increase in the islet iron content with increasing incubation time, with a maximum at 12 h (P < 0.001) (Fig. 1C). The results of the retention assay indicated that there was no iron loss in labeled islets in culture (Fig. 1D). We did not attempt to investigate iron retention for longer periods of time since we generally receive islets between 3 and 4 days of culture and have, on average, another 3 days to perform experiments while they are still viable. However, it is our experience that labeled islets retained their label in vivo for up to 188 days after transplantation under the kidney capsule (17).
Preservation of islet viability and function after labeling with Feridex is crucial for future clinical transplantation. Therefore, we assessed islet viability after overnight exposure to the contrast agent and found that it remained unchanged compared with nonlabeled cultured islets (P > 0.05). In addition, there was no elevated level of apoptosis in pancreatic islets after 24 h exposure to Feridex compared with nonlabeled cultured islets (online appendix Fig. 1 [available at http://diabetes.diabetesjournals.org]). Glucose-stimulated insulin secretion, as seen by a stimulation index, was unchanged in labeled versus nonlabeled islets (5.9 ± 2.0 vs. 6.2 ± 2.0, respectively; P > 0.05). Based on these experiments, we concluded that labeling of human islets with Feridex does not compromise their viability and function.
Similar to our previous findings with related iron oxides (17), we observed that within the pancreatic islet, Feridex localized to all islet cells including β-, α-, and δ-cells, as well as to islet macrophages (Fig. 2A). Most of the label was associated with insulin-producing β-cells since these cells represent the largest cell population in the islet. Feridex does not have any specificity toward islet cells, and cellular uptake is most likely associated with conventional fluid phase endocytosis. These findings are in agreement with previous results regarding iron uptake by various cell types shown by us (17) and other investigators (20,23–25). Labeling of pancreatic islets with Feridex had a heterogeneous pattern, and some of the islets labeled more efficiently than the others with overall efficiency ranging from 10 to 70% (Fig. 2B). These numbers represent only a histological cross section of a particular islet and might not reflect an overall volumetric accumulation of the iron. In addition, we did not find any correlation between islet size and labeling efficiency. It seems that this process depends more on the ”freshness“ of the particular islet isolation with islets being labeled more efficiently immediately to 2–3 days after isolation. Detailed investigation on this matter, however, is outside the scope of this manuscript and is currently underway in our laboratory. It is noteworthy that the average labeling efficiency achieved in this and similar studies was enough to detect a single islet by MRI in islet phantoms (17).
In vivo imaging of immune rejection after transplantation.
To test the feasibility of in vivo MRI of intrahepatic human islet grafts, we transplanted Feridex-labeled islets in immunocompetent (Balb/c) and immunocompromised (NOD.scid) mice. We found that islets transplanted into the liver appeared as dark hypointense foci representing single islets and/or, possibly, islet clusters (Fig. 3A), whereas nonlabeled islets did not cause any change in signal intensity on T2*-weighted images (Fig. 3B). These signal voids were detectable in both models for the duration of the study, although the number of these voids changed with time. To evaluate the number of islets/islet clusters in the liver, we manually scored the number of signal voids found on each slice. The validity of this method was first confirmed by creating a calibration curve where the known number of transplanted islets directly correlated to the number of dark voids representing labeled islets/islet clusters found on T2*-weighted images (R2 = 0.997; Fig. 3D). Since islet death has been reported in islet grafts during the first 10–14 days after transplantation (9,15,26,27), we performed MRI starting on the 1st day after transplantation. Visual coregistration of matching slices on magnetic resonance images showed that the disappearance of signal voids in Balb/c mice was more pronounced than in NOD.scid mice over time (Fig. 3C). Quantitative analysis of relative islet loss was performed for immunocompromised and immunocompetent animals based on 13 slices covering the entire liver. This analysis showed that the number of islets in both models started to decline immediately after transplantation. The rate of islet loss gradually decreased during the course of the study with plateau between days 10 and 14 (Fig. 3E). This immediate decrease is consistent with islet death after transplantation due to mechanical injury, ischemia, and nonantigen-specific inflammatory events (27). However, it is evident from Fig. 3E that immunocompetent mice exhibited a significantly higher rate of islet disappearance on magnetic resonance images compared with immunocompromised animals, especially pronounced on day 10, and resulting in a 20% difference in relative islet number by 14 days after transplantation, presumably due to the input from severe immune rejection. To correlate our MRI data of islet death with apoptotic rates, we performed a TUNEL assay on excised livers. In NOD.scid and Balb/c mice, apoptosis was most pronounced on day 1 (12.1 and 18.6%, respectively). The number of apoptotic cells in immunodeficient mice decreased to 1.8% by day 4 and stayed at this level for the duration of the study. In contrast, the number of apoptotic cells in immunocompetent mice gradually decreased to 3.3% on days 4–5 followed by severe antigen-specific rejection, which destroyed up to 13.2% of the islets on days 10–11 (Fig. 3F). The apoptosis level gradually decreased by day 14 since by that time, almost all cells in the islets were destroyed. This rejection was evident from our correlative immunohistochemical studies that showed that the higher rate of islet death detected by MRI in immunocompetent mice was due to significant immune cell infiltration. As evident from Table 2, this infiltration in Balb/c mice started in the first days after transplantation. The transplanted pancreatic islets in these mice contained CD8+ and CD4+ T-cells, B-cells, macrophages (Fig. 4A), and neutrophils (Table 2). The level of infiltration by these cells increased with time and peaked on days 10–11 in agreement with the TUNEL assay results. In contrast, immunocompromised animals showed only marginal presence of immune cells in islets after transplantation except for CD68+ invading macrophages, which are unaffected in animals with the scid mutation (28) (Fig. 4B). Evidently, in NOD.scid mice, islet death occurred mostly from the input from infiltrating macrophages and from cell damage from mechanical injury, ischemia, and nonantigen-specific inflammatory events (27). Similar results were obtained by other investigators studying immune rejection in allogeneic transplants in different species (29–31).
Overall, the results of this study suggested that the disappearance of pancreatic islets on magnetic resonance images was due to islet death, which was confirmed by histological studies. This event was more pronounced in immunocompetent animals compared with immunocompromised animals due to severe immune rejection in the former.
In our next set of experiments, we attempted to trace the history of the iron label after transplantation and as a function of islet death. As seen in Fig. 5A, immediately after transplantation in the liver, islets retained Feridex and, therefore, were visible on magnetic resonance images. However, as immune rejection and other factors were affecting their viability, transplanted islet cells underwent apoptosis releasing Feridex and decreasing its content significantly. Figure 5B shows a pancreatic islet from an immunocompetent mouse 10 days after transplantation, which was infiltrated with immune cells but was still maintaining some of the label allowing for its detection by MRI. Interestingly, the remaining label in the islet was not internalized by invading cells but stayed associated with islet cells. At the same time, as islets were dying, they released their content into the liver parenchyma where it was internalized and processed by Kuppfer cells (Fig. 5C). By 14 days after transplantation, however, there was no evidence of iron associated with Kuppfer cells or liver parenchyma, suggesting rapid clearance of the label consistent with the known pharmacokinetics of the compound (32) (Fig. 5D). From an imaging perspective, iron released by dying islet cells in small amounts diffuses throughout the liver and sparsely distributes over a larger area and, therefore, is not expected to create “false- positives” on magnetic resonance images. By contrast, Feridex concentrated upon internalization within viable pancreatic islet cells creates a signal void due to compartmentalization of the label in a membrane-enclosed environment (33,34). The persistence of iron within Kuppfer cells is short lived due to its rapid processing by these cells and its release into the physiologic iron pool (32). Therefore, the only entity within the liver, which would retain iron at a high concentration and with a steady time-course would be viable islet grafts. Overall, based on these results, we have demonstrated that the labeling of human pancreatic islets with an FDA-approved contrast agent and their imaging by MRI is feasible and represents a promising strategy for the in vivo tracking of transplanted islet fate over time as well as for the detection of islet death immediately after transplantation.
Type 1 diabetes results from autoimmune destruction of the insulin-producing β-cells (35). In diabetic patients, daily treatment with exogenous insulin is required in order to achieve normoglycemia. However, because of difficulties in achieving physiological control of blood glucose concentrations, chronic and degenerative complications still occur in a marked number of patients. Pancreatic islet transplantation has recently emerged as an attractive alternative to insulin injection and as a promising clinical modality, which can render patients with type 1 diabetes insulin independent (1,2). In general, islet transplants are highly susceptible to inflammatory damage and to ischemia and anoxia, as a consequence of the transplant procedures. During harvesting, islets are stripped from their blood vessels. Consequently, the perfusion of the graft is severely compromised until revascularization is established (4). However, with the establishment of blood flow, the islets become a target for ischemia-reperfusion injury, in addition to inherent ischemia and anoxia, during the period of compromised blood flow, coagulation, and thrombosis (36). The major challenge, however, lies in allograft rejection, which occurs after transplantation. Clearly, for islet transplantation to become widely applicable, development of successful and safe strategies for inducing tolerance to islet transplants and reducing immune rejection is a necessity. Successful immunosuppressive regimens provide hope for continous improvements in achieving insulin independence (2). To evaluate the success of islet transplants, repeated noninvasive monitoring of surviving grafts is needed. MRI, with its high resolution and availability of contrast agents, seems to be an ideal modality to perform such monitoring.
We have previously shown that in vivo MRI of transplanted islets is possible, providing that islets are labeled with a contrast agent (17). Here, we built on our previous findings and performed in vivo imaging of transplanted islets using Feridex, which is currently used in clinics as a contrast agent for the detection of liver lesions. Similar to nanoparticles used in our proof-of-principle studies (17), Feridex consists of a superparamagnetic iron oxide core covered with a dextran coat. When used for hepatic MRI, the iron oxide nanoparticles are phagocytosed and accumulate in the endosomes of Kuppfer cells and reticuloendothelial cells (18). Iron oxide particles are nontoxic, biodegradable, and have been used in the clinic as intravenous contrast agents. In addition, we utilized the clinically relevant intrahepatic model of islet transplantation. Currently, this is the only clinically established protocol (2) that, together with the use of an FDA-approved contrast reagent, serves as a solid base for transferring our studies into clinical trials.
As a first step, we established the suitability of Feridex for labeling human pancreatic islets. Specifically, we confirmed that islets could be labeled with Feridex without altering their function and causing toxicity. Furthermore, we showed that labeled islets were readily detectable on T2-weighted magnetic resonance images due to the presence of the label, a finding that was confirmed microscopically. Based on our in vitro results, we performed in vivo MRI of transplanted pancreatic islets in immunocompromised and immunocompetent mice. Severe immune rejection in immunocompetent animals was evident from MRI as the gradual disappearance of the islet-associated label. The rate of islet death in these animals was significantly higher than in NOD.scid mice and correlated with the loss of signal voids on magnetic resonance images. Interestingly, using our in vivo imaging technique, we were able to confirm previous findings regarding early islet death after transplantation, which were exclusively based on histological assessment. Namely, in vivo MRI showed that this process started immediately after the transplantation procedure in both animal models and continued for the duration of the study. In this study, we deliberately avoided using diabetic animals to exclude the effect of glucose toxicity on islet death (15); instead, we focused on the input from immune rejection, which was noted as the major challenge in islet transplantation (4). However, we believe that the input of glucose toxicity is an extremely important factor and deserves special consideration, which we are currently pursuing in a separate study.
Our findings are in drastic contrast with the recent publication by Kriz et al. (37), where no islet death was reported in syngeneic animals up to 6 weeks after transplantation. Furthermore, these authors claimed that islet death in allogeneic transplants began as late as 1 week after transplantation. We believe that this cannot be the case since islet death in both allogeneic and syngeneic models has been detected immediately after transplantation in previous studies and is well documented (9,15,26,27). The authors may not have taken into consideration islet distribution in the entire liver since they counted dark voids in only three slices on magnetic resonance images. This provides very limited coverage of the liver, especially taking into account the imaging parameters used in this study (slice thickness 2 mm with slice separation of 3 mm). In addition, we could not support the authors’ claim that iron was only localized to islet macrophages, which actually contradicts their own previous findings (38). If all of the iron label was localized to macrophages, then calculating the iron concentration per cell based on their reported iron accumulation (38) and the known number of resident macrophages in an islet (39), the average macrophage would contain an unrealistic 10–20 ng of iron. The possibility of iron being trapped between islet cells, as suggested by Jirak et al. (38), should be confirmed experimentally.
Our previously reported observations (17), and the data presented here, suggest that labeling with iron oxides involves all islet cell types and that massive accumulation of the label limited to a small volume of the islet causes the dark contrast on magnetic resonance images that we observed. Overall, the application of contrast agents helps to overcome a natural low sensitivity of MRI, which seems to be the most suitable modality for islet imaging. MRI is advantageous in comparison to other imaging modalities because it does not utilize damaging ionizing radiation, has tomographic capabilities, can deliver the highest resolution images in vivo, and has unlimited depth penetration (16). We believe that in the current study, we have established a convincing case for utilizing the FDA-approved contrast agent Feridex for in vivo MRI of immune rejection of labeled grafts, pending approval by appropriate regulatory agencies. We believe that these studies will be translated into clinical trials in the near future. Pancreatic islet transplantation holds hope for patients with type 1 diabetes. Therefore, the ability to follow the fate of the graft is crucial for the success of this procedure.
This work was supported in part by the National Institutes of Health Grants DK071225 and DK072137 (to A.M.).
We thank the following Islet Cell Resource Centers for providing human pancreatic islets: Diabetes Research Institute, Miami, Florida; City of Hope National Medical Center, Duarte, California; Puget Sound Blood Center, Seattle, Washington; Human Islet Laboratory, University of Pennsylvania, Philadelphia, Pennsylvania; College of Medicine, University of Tennessee, Memphis, Tennessee; College of Physicians and Surgeons, Columbia University, New York, New York; Joslin Diabetes Center, Boston, Massachusetts; University of Minnesota, Minneapolis, Minnesota; Children’s Hospital of Pittsburgh, Pittsburgh, Pennsylvania; and University of Alberta, Edmonton, Canada.
We acknowledge John Moore, BS, for his exceptional support in animal surgery and Guangping Dai, PhD, for his assistance with in vivo MRI.
Z.M. and J.P. contributed equally to this work.
Additional information for this article can be found in an online appendix at http://diabetes.diabetesjournals.org.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Received April 11, 2006.
- Accepted June 16, 2006.