|Home | About | Journals | Submit | Contact Us | Français|
Fifteen thousand youths are diagnosed yearly with type 1 diabetes mellitus. Pancreatic islet transplantation has been shown clinically to provide short-term (~1 year) insulin independence. However, challenges associated with early vascularization of transplanted islet grafts and long-term islet survival remain. We utilized dynamic contrast enhanced magnetic resonance imaging (DCE MRI) to monitor neovascularization of islets transplanted into the right lobe of the liver in a syngeneic mouse model. The left lobe received no islets and served as a control. DCE data were analyzed for temporal dynamics of contrast (gadolinium) extravasation and the results were fit to a Tofts two-compartment exchange model. We observed maximal right lobe enhancement at seven days post-transplantation. Histological examination up to 28 days was used to confirm imaging results. DCE-derived enhancement strongly correlated with immunohistochemical measures of neovascularization. To our knowledge these results are the first to demonstrate, using a FDA approved contrast agent, that DCE MRI can effectively and non-invasively monitor the progression of angiogenesis in intraportal islet grafts.
Type 1 diabetes mellitus is a costly and growing medical problem, particularly within the pediatric population where an estimated 15,000 youths are diagnosed with type 1 diabetes annually.1 The current standard of care involves insulin administered by injection or pump, but exogenous insulin treatment cannot perfectly control blood glucose levels despite advances in continuous glucose monitoring. Cumulative lifetime exposure to hyperglycemic episodes can lead to complications such as retinopathy, nephropathy, cardiovascular disease and peripheral neuropathy.2,3 Islet cell transplantation has recently emerged as a promising therapeutic approach for providing long-term (>1 yr) insulin independence with the advent of the Edmonton Protocol.4 More recently, an international trial showed that insulin independence could be achieved in more than 50% of transplanted patients, and that 80% of these patients were insulin-independent after one year.5
Despite these promising results, insulin independence is usually not sustainable, with 85–90% of patients requiring insulin injections by 5 years.6 The failure of long-term insulin independence can be attributed to islet loss due to: (1) the immediate low rate of engraftment,7,8 (2) progressive immune rejection,9,10 (3) islet toxicity due to continued immunosuppressive drugs,11–13 (4) low rate of engraftment due to extreme changes in the local tissue environment caused by the transplantation,14 and (5) increased hypoxia from initial lack of islet vascularity.15,16 In a healthy pancreas, islets account for 1–2% of pancreatic mass but receive 5–10% of blood flow.17 After transplantation into the liver, islets are prone to hypoxia18 with an estimated 50–70% of islets being lost in the immediate post-transplantation period that can then lead to functional impairment of the remaining islets.19–21 A tool to monitor the revascularization process is critical to advancing islet transplantation.
While standard MRI approaches can be used to detect the presence and size of islet grafts, visualization of islet graft revascularization remains unexplored. New imaging modalities, such as dynamic contrast-enhanced (DCE) MRI, are optimized to examine physiological parameters such as blood flow and capillary permeability and could be utilized to determine new vessel formation in transplanted islets. DCE MRI was first developed to characterize tumor vascularization22,23 using rapid and repeated collection of images before, during, and after contrast administration, such as with gadolinium.24,25 The temporal dynamics of contrast arrival within the vasculature and consequent leakage into the extra-vascular space have been modeled (two compartments) by Tofts et al.26 The physiological parameters extracted using this model can be used to observe neovascularization at the transplant site.
Previously, we described a timeline for neovascularization after subcapsular kidney islet transplantation and demonstrated the feasibility of using DCE MRI to detect and quantify changes in vascularization over 28 days.27,28 We have extended DCE MRI to monitor vascularization of intraportal islet transplants over 28 days post transplantation. Neovascularization was confirmed by immunohistochemical evaluation of transplanted islets and host liver tissue.
MRI was used to monitor the progression of new vessel formation in the right liver lobe at post-operative day (POD) 3, 7, 14 and 28. High resolution T1-weighted imaging assisted in localization of the right (transplanted) and left (control) liver lobes (Fig. 1A), and was followed by placement of a single DCE slice at the same level. Rapidly acquired DCE images immediately before and after a single bolus injection of gadolinium (~0.04 cc) showed fast enhancement and slower wash out. Another high resolution T1 scan post-contrast indicated that equilibrium in enhancement was reached within 40 minutes after injection. The time course of right lobe enhancement in this subject at POD 14 showed that 4.3 minutes after contrast injection, a peak value of 2.2 was reached (Fig. 1B).
The average time to peak contrast enhancement within the right liver varied significantly only between POD 3 and POD 14 (p < 0.05). The average time to peak was 2.1 ± 0.2 mins at POD 3, 3.5 ± 0.2 mins at POD 7, 3.8 ± 0.6 mins at POD 14, and 2.8 ± 0.5 mins at POD 28 (Fig. 2). However, the magnitude and duration of the enhancement was dependent on the POD time point. Area under the curve (AUC) analysis showed marked increases at POD 7 and POD 28 (Fig. 2B) that trended towards significance (p < 0.09). The enhancement in the control lobes was not significantly different between POD periods, validating the use of the left liver lobe as a control. Similarly, no increased enhancement of muscle tissue was observed, validating consistent contrast injection dose and rate (data not shown). Comparing the maximum relative tissue concentration change between the right and left livers lobes revealed a 30% increase at POD 7, followed by a small change of 8% at POD 14, and a resurgent increase to 26% at POD 28 (Fig. 2C).
Immunostaining for insulin verified the presence of insulin-containing islets at all time points (Fig. 3A–D). von Willebrand Factor (vWF) staining for new vessels (Fig. 3E–H) showed scant staining at POD 3 (Fig. 3E) with a larger increase in small new vessels in close proximity to islets at POD 7 (Fig. 3F). At POD 14, vessel numbers were decreased (Fig. 3G). However, by POD 28, the number of vWF positive vessels again increased (Fig. 3H). High magnification reveals typical peri-islet microvessel morphology at all time points (Fig. 3I–L).
Quantification of vWF stained peri-islet microvessels showed the extent and time course of islet angiogenesis (Fig. 4). An average of 1.6 new vessels per islet was found at POD 3, 3.0 at POD 7, 2.3 at POD14 and 3.2 at POD 28. The number of microvessels at POD 7 and POD 28 were significantly increased compared to POD 3 by more than 80% (p < 0.05). There was no significant increase in new vessels at POD 14 as compared to POD 3.
The correlation between microvessel number and DCE MRI AUC was plotted and a linear least squares regression revealed a strong correlation (r2 = 0.863, p < 0.001). The positive slope of the regression line indicates that increases in contrast enhancement are highly and significantly correlated with increases in new peri-islet microvessels (Fig. 5).
To our knowledge, this is the first demonstration of DCE MRI using the FDA approved contrast agent Gd-DTPA to establish a timeline of vascularization in intra-portal islet transplants. We report the following novel findings: (1) non-invasive DCE MRI suggests a peak of new vessel formation in intraportal islet grafts at post-operative day (POD) 7, (2) immunohistochemical analysis shows a significantly increased number of new peri-islet vessels at POD 7 and POD 28 versus POD 3, and (3) DCE MRI findings strongly correlate with histology/immunohistochemistry results.
Peak enhancement at POD 7 is consistent with the increased rate of new vessel formation seen histologically by us as well as others.29 Of note, the lack of von Willebrand staining within the islets is presumably due to an already mature though disrupted intra-islet donor-derived vasculature.17 By POD 14, DCE-derived enhancement and new vessel numbers decreased, suggestive of a mature islet vasculature. This finding in our non-diabetic model is consistent with the timeline of revascularization described in a diabetic model.30 There was a secondary elevated enhancement in the right liver again at POD 28. Though it was not statistically significant, the increased enhancement appears to be related to the process of angiogenesis as histological assessment revealed a concordant increase in new vessel numbers. Previous subcapsular studies27,28 found that similar numbers of animals resulted in significant DCE changes reflective of angiogenesis while in the current study there was a clear trend towards increased DCE that did not reach significance. A potential explanation could be that the subcapsular transplants were confined to focal regions in contrast to the intraportal transplants that where spread out throughout the entire liver, thus making the angiogenic process more diffuse.
We considered whether islet death as opposed to angiogenesis was the cause of increased enhancement as islet number slowly wanes with time.10,31 A decreased islet mass due to necrosis and/or apoptosis at POD 28 might have caused an inflammatory response and resulted in increased vessel permeability, leading to the increased enhancement seen in the right lobe. However, in another study, histological staining of islet grafts for apoptosis (TUNEL), necrosis (H&E), and HIF-1α revealed essentially no positive stain at this late POD 28 time point 32. Whether the islet neovascularization process is different in non-diabetic mice, whether there is a paring back of islet vasculature in between the two phases of angiogenesis, or whether this reduction may be responsible for the late second phase are all yet to be determined.
It is currently unclear as to the basis for the bi-phasic response (DCE and vessel number) after intraportal transplantation. Several potential reasons include: (1) use of a non-diabetic model, where hyperglycemia-associated stimuli may be missing, (2) secondary signaling from pro-angiogenic factors might be transiently blunted by hypoxia-inducible factors at POD 14, and (3) accelerated but unsustained angiogenesis in intraportal transplantation (7 days earlier than subcapsular)27,28 may potentially lead to a secondary local hypoxic event that further stimulates angiogenesis. However, only future work evaluating the molecular angiogenic profiles at these time points will be able to definitively resolve this interesting finding.
The use of untransplanted liver lobes as controls helped to avoid intersubject variability and was justified as the dynamics of enhancement did not vary significantly over time. Any effects that may have been caused by the transient (~1 min) ischemia from clamping of the left portal vein branches during islet transplantation were not evident on DCE MRI at POD 3 and beyond. One potential limitation of infusing islets into only the right liver lobe was the possibility that initial portal thrombosis from islet clusters and/or macrophages would lower the blood flow through the right lobe and concomitantly increase flow to the other lobes.33 Such a differential in flow could have been manifest as decreased delivery of contrast agent to and decreased enhancement of the right lobe. Activation of the extrinsic coagulation pathways may have also been another source of right lobe specific thrombosis.34 However, by the earliest imaging time point at POD 3, no right lobe thromboses were evident as right lobe enhancement was statistically no different than the control lobe, thus validating the lobe-specific transplant model.
Another advantage of this approach is its independence from islet labeling. The clinical utility of MRI in islet transplantation has recently been demonstrated,35 but visualization of islets required labeling with superparamagnetic iron oxide (SPIO). Micro-positron emission tomography (PET) is another imaging modality that has been used to track islet survival.36–38 However, neither PET nor MRI without contrast is able to resolve the details of vascular structures. MRI with an experimental long circulating contrast agent (PGC-GdDTPA-F) has been used to monitor the vascular changes around native pancreatic islets.39 In earlier studies we have established a timeline for islet revascularization in subcapsular kidney grafts.27,28 In this study we have shown that MRI with a clinical contrast agent can be used to follow the vascularization of syngeneic murine islets transplanted into the liver. These studies suggest that DCE MRI might eventually be applicable in humans and useful in evaluating therapeutic strategies for increasing the efficiency of engraftment in clinical islet transplantation.
Adult female Balb/c mice weighing 25–30 g were purchased (Charles River, Wilmington, MA) and housed under specific pathogen-free conditions with a 12-hr light/dark cycle and had free access to food and water. All care and handling of animals was in accordance with Loma Linda University Institutional Animal Care Use Committee which approved all experimental protocols.
Islets were isolated by collagenase digestion of the pancreas and separated from exocrine tissue by discontinuous Ficoll density gradient centrifugation and then hand-picked as previously described.27 Iron labeling of islets was performed by overnight co-culture of freshly isolated islets in Feridex (Advanced Magnetics Inc., Cambridge MA)-supplemented medium at 200µg iron/ml. Under general inhalation anesthesia (2% isoflurane), a midline incision was made in recipient mice, the bowel was moved out of the abdominal cavity to expose the portal vein at the level of hepatic hilar to ileocecal veins. The left portal vein was temporarily clamped, the ileocecal vein was punctured with a 25-gauge needle (Becton Dickinson, NJ, USA) and 800 syngeneic islet equivalents (IEQ = islets with a diameter of 150 µm) were slowly injected into the right liver lobe. The needle was pulled out with pressure from a cotton swab to stop bleeding. Animals were allowed to recover for 72 hours prior to the first imaging time point.
MRI was performed at 3 (n = 6), 7 (n = 4), 14 (n = 5) and 28 (n = 4) days post transplantation (POD). Each mouse was imaged prior to and after-contrast injection using a T1-weighted sequence of the entire liver. DCE MRI was performed on a slice through which both right and left liver lobes were visible. All MRI data were collected on a Bruker Advance 11.7 T MRI (8.9 cm bore) with a 3.0 cm (ID) volume radiofrequency coil (Bruker Biospin, Billerica MA). Mice were lightly anesthetized using isoflurane (3% induction, 1% maintenance). A tail vein catheter was inserted and fastened to the tail for infusion of gadolinium DTPA contrast (Gd-DTPA, -BMA, Gadodiamide, 0.8 mmol/kg, Omniscan, Amersham Health, Princeton NJ). Body temperature was maintained at 35–37 ± 1°C using a thermostat-controlled heated water cushion placed under the mouse while respiration was monitored with a MR-compatible pressure transducer on a Biopac MP150 (Goleta CA).
A high resolution pre/post-contrast T1 composed of a TR/TE of 837/10 ms, a 2562 matrix, 3 cm field of view (FOV), and 4 averages for a total acquisition time of 14 minutes. Twenty coronal slices were collected with a 0.8 mm thickness and interleaved by 0.8 mm. These T1 images were visually evaluated to identify the liver lobes. A single DCE acquisition slice was then placed through a section of liver that included the transplanted right lobe. The DCE sequence acquired one image slice through the liver using a TR/TE = 250/6.4 ms, 1282 matrix, 3 cm FOV, one average for an acquisition time of 32 sec/image and a total acquisition time of 32 minutes with 60 images collected.
DCE analysis utilized the temporal dynamics of contrast enhancement that were quantified using JIM software (Thorpe Waterville, UK). Briefly, regions of interest (ROIs) (right liver, left liver and muscle) were identified on the DCE images based on T1 images (Fig. 1). Kinetic analysis used a bidirectional two-compartment model based on the equations of Tofts et al.26 Briefly, contrast agent that is injected into the blood pool extravasates into different tissues with various permeabilities. This process can be described by equations that model the dynamics of contrast agent exchange back and forth between the blood and tissue compartments. DCE MRI tissue gadolinium concentration curves extracted from JIM software were normalized for inter- and intra-animal comparisons and then averaged and curve fit. Area under the curve (AUC) values were calculated by integrating the area using the trapezoidal method.
After imaging, the liver was recovered and placed in 10% formalin for 24 hrs and then the right lobe(s) was embedded in paraffin. The orientation of liver tissue blocks was kept consistent and five µm serial sections were taken at four different levels through the lobes at 400 µm intervals. Sections were deparaffininized in xylene and hydrated. Sections were stained with Hematoxylin and Eosin (H&E) and immunohistochemistry for insulin and von Willebrand Factor (anti-vWF) was performed to identify the presence of insulin in the islets and peri-islet neovasculature.40,41 To restore antigen immunoreactivity for anti-vWF, tissue was treated with Proteinase K (Dako, Carpinteria, CA) for 2 minutes followed by blockade of endogenous peroxidase activitiy by treatment of 0.1% hydrogen peroxide for 30 minutes. Nonspecific binding was blocked with 10% goat serum for 30 minutes. Specimens were incubated with either the guinea pig anti-insulin antibody (1:100; Dako) or rabbit anti-vWF (1:500; Abcam, Cambridge, MA) for 90 minutes at room temperature. Biotinylated anti-rabbit IgG antibody treatment for 30 minutes was followed by streptavidin-conjugated horseradish peroxidase treatment for an additional 30 minutes (Vectastain Elite ABC kit; Vector Laboratories, Burlingame, CA). Bound peroxidase was developed with 3-3'-diaminobenzidine (DAB: brown; Dako) for vWF or aminoethylcarbozole chromogen (AEC+:red; Dako) for insulin and counterstained with hematoxylin.
Insulin-postive islets were identified in liver sections by brightfield microscopy, images were captured (Zeiss Axio Imager A1) and islet area was manually outlined (ImageJ, NIH). Islets larger than 1,800 µm2 underwent microvessel quantification on subsequent vWF stained sections (n = 14 at POD3, n = 28 at POD 7, n = 64 at POD14, n = 19 at POD28). New microvessels that stained positive for vWF were counted by a blinded observer at higher magnification (200–400×). Inclusion criteria for islet-specific microvessels were: (1) localization, vessels were in contact with or within the islet and (2) morphology, vessels must be less than 40 µm along the longest axis.
Statistical evaluation including repeated measures ANOVA (RM ANOVA) was performed using Sigmastat software (SPPS, Chicago IL) and differences among experimental groups were considered significant for p < 0.05. Pairwise comparisons were made using the Holm-Sidak method and post-hoc t-tests. Data were expressed as the mean ± standard error of the mean (SEM).
We are grateful for the technical assistance provided by John Chrisler, Pete Hayes and Tom Lechuga. Special thanks also to Dr. William Pearce. This work was supported by NIH/NIDDK grant number 1-R01-DK077541-03 (E.H.).