Anti-inflammatory agents are used routinely in clinical islet transplantation in an attempt to promote islet engraftment. Infliximab, and more recently etanercept, is being used to neutralize tumor necrosis factor alpha, but this tenet is based on limited preclinical data. One group has promoted the potential of combined etanercept with an IL-1 receptor antagonist, anakinra in a small clinical study, but without strong preclinical data to justify this approach. We therefore sought to evaluate the impact of combined anakinra and etanercept in a marginal islet mass transplant model using human islets in immunodeficient mice. The combination of anakinra and etanercept led to remarkable improvement in islet engraftment (control 36.4%; anakinra 53.9%; etanercept 45.45%; anakinra and etanercept 87.5% euglycemia, p < 0.05 by log-rank) compared to single-drug treated mice or controls. This translated into enhanced metabolic function (area under curve glucose tolerance), improved graft insulin content and marked reduction in beta-cell specific apoptotis (0.67% anakinra + etanercept vs. 23.5% control, p < 0.001). These results therefore strongly justify the combined short-term use of anakinra and etanercept in human islet transplantation.
Inflammatory injury of transplanted human islets is recognized by most clinical islet programs as a major remediable target, as judged by the large proportion of clinical centers that routinely and empirically administer anti-inflammatory treatments posttransplant. Since the early days of islet transplantation, antioxidants (pentoxiphylline, multivitamins), anti-inflammatories (aspirin, nonsteroidal anti-inflammatory agents), and relatively expensive and specific antitumor necrosis factor-alpha (anti-TNF alpha) blockade with infliximab or more recently etanercept, has been a keystone component of peritransplant management of at least 80% of islet transplants performed, according to the Clinical Islet Transplant Registry (CITR) summary reports (1). The routine use of anti-TNF strategies has been based solely on one preliminary report in a syngeneic mouse islet transplant model, but without formal supportive data in a more relevant human islet transplant model (2). The routine clinical adoption of anti-TNF-alpha strategies is therefore empiric, and lacks supportive data from randomized controlled trials. Hering et al. reported high rates of single donor islet transplant engraftment and insulin independence, with etanercept being a major component of the regimen (3,4). Most recently, one center has advocated the combined use of anti-TNF alpha blockade together with an IL-1 receptor antagonist (IL-1ra), but with only three patients treated, the results are encouraging but difficult to interpret due to limited power (5). We therefore set out to investigate the potential synergy of a soluble dimeric TNF-alpha receptor fusion protein (etanercept) when used alone and in combination with an IL-1Ra (anakinra), when given to chemically diabetic immunodeficient mice receiving marginal mass human islets. The purpose of this study is therefore to provide additional preclinical data to determine the appropriate utility of these agents in clinical islet transplantation.
Materials and Methods
Animals and reagents
Immunodeficient B6-RAG−/− mice (B6.129S7-Rag1tm1Mom/J) were obtained from the Jackson Laboratories (Bar Harbor, ME, USA) and housed under specific pathogen-free conditions. BALB/c mice were also obtained from the Jackson Laboratories but housed under conventional conditions. All animals were cared for according to the guidelines of the Canadian Council on Animal Care, and ethical approval was obtained from the animal welfare committee at the University of Alberta. All reagents were obtained from Sigma Aldrich (Oakville, Ontario, Canada) unless otherwise specified. Anakinra was purchased from Biovitrum (Stockholm, Sweden). Etanercept (Amgen Inc., Thousand Oaks, CA, USA) was purchased commercially from the University of Alberta hospital pharmacy.
Mouse islet isolation
Mouse islets were isolated using established protocols with minor modifications (6). In brief, mouse pancreata were digested with collagenase (1.0 mg/mL in Hanks' buffered saline solution [HBSS]) and purified with Histopaque-density centrifugation (Sigma Aldrich, Oakville, ON, USA). Handpicked islets were washed with HBSS then placed in short-term culture in Connaught Medical Research Laboratories (CMRL-1066) medium supplemented with 10% fetal bovine serum, l-glutamine (100 mg/L), penicillin (112 kU/L), streptomycin (112 mg/L) and HEPES (25 mmol/L). Islets were cultured for a maximum of 2 h before transplantation.
Human islet isolation
Pancreata were retrieved from multiorgan deceased donors after aortic cross-clamp and infusion of histidine–tryptophan–ketoglutarate (HTK) solution. Islets from four separate human islet isolations were isolated according to a modified Ricordi's semiautomated technique (7,8). Briefly, the pancreas was distended with collagenase NB1 supplemented with neutral protease (Serva Electrophoresis GMbH) and digested in a Ricordi chamber. When free islets were released, tissue digest was collected and further purified on a cell sorter (Model 2991, Cobe, Lakewood, CO, USA) using a continuous density gradient (9). Human islets were processed in all cases with intent for clinical transplantation, but made available for this study when the islet yield fell short of the minimal mass required for clinical transplantation, and where specific research consent allowed for use of human islets. Permission to use human islets for these studies was granted by the Health Research Ethics Board of the University of Alberta. Human islets were cultured overnight in CMRL-1066 supplemented with insulin–selenium–transferrin and insulin-like growth factor-2 (IGF-2) at 37°C before transfer to the laboratory and transplantation into diabetic mouse recipients.
Streptozotocin was administered to recipient mice to induce diabetes (Balb/C: 220 mg/kg intraperitoneally (i.p.); B6-RAG−/-: 180 mg/kg i.p). Animals were considered diabetic after two consecutive blood glucose measurements ≥20 mmol/L using a OneTouch Ultra glucometer (Lifescan Canada, Burnaby, B.C). For mouse islet studies, a marginal mass of 150 islets were implanted into the kidney subcapsular space. In the human islet studies, in four separate replicates with different human islet donors, an equivalent marginal mass of 1500 human islet equivalents (IE) were implanted beneath the kidney capsule, as described previously (10). Transplant recipients were divided into four groups and treated with (1) anakinra 100 mg/kg i.p. daily for 7 days (n = 12 for human islets, n = 11 for mouse islets); (2) etanercept 5 mg/kg ip on days 0, 3, 7 and 10 (n = 11 for human islets, n = 16 for mouse islets) to mimic the current clinical dosing schedule introduced by Hering et al., and now adopted by several islet transplant programs worldwide (3,11); (3) IgG (control antibody) 5 mg/kg i.p. on days 0, 3, 7 and 10 (n = 11 for human islets, n = 11 for mouse islets); (4) combined treatment with anakinra and etanercept at the doses above (n = 16 for human islets, n = 13 for mouse islets). Of note, to control for potential differences in islet quality between different human islet donors, islets from each of the four human donors were distributed across experimental groups in paired analysis. Blood glucose of recipients was monitored daily by tail-vein glucometer readings.
Glucose tolerance tests
Transplanted mice were fasted for 16–20 h and injected intraperitoneally with 50% dextrose at 2 g/kg body weight (intraperitoneal glucose tolerance test, IPGTT). Blood glucose levels were analyzed at baseline, 5, 15, 30, 60, 90 and 120 min postinjection.
Graft insulin content
Islet grafts were recovered from the kidney capsule and stored at −80°C for bulk analysis. Extraction was performed in acid-ethanol by homogenization and ultrasonic cell membrane disruption. Insulin concentration of the neutralized extract was measured using a commercial ELISA kit (Alpco Diagnostics, Windham, NH, USA).
Human islet culture and viability
A portion of each human islet isolation (n = 4 islet preparations) was separated into four groups and placed in culture in CMRL-1066 at 37°C. The medium was supplemented with (1) anakinra 10 ug/mL; (2) etanercept 100 uM; (3) IgG-control; or (4) anakinra and etanercept. Islets were counted using dithizone at the beginning of culture and at the 48-h point. Viability was assessed using Syto green/ethidium bromide, counting 100 islets under fluorescence light microscopy as previously described (Cedarlane laboratories and Sigma-Aldrich, ON, Canada) (12,13).
Combined etanercept and anakinra substantially improves marginal mass human islet engraftment in immunodeficient mice
Immunodeficient, diabetic B6-Rag−/− mice received a marginal mass human islet graft (1500 IE) under the kidney capsule and one of the four treatments defined above. Those receiving anakinra and etanercept had the highest rate of diabetes reversal (87.5%) (two consecutive nonfasting blood glucose readings < 11 mmol/L), which was significantly higher (p < 0.05 by log-rank analysis) than mice receiving anakinra (53.8%), etanercept (45.5%) or IgG control (36.4%) (Figure 1). The use of etanercept or anakinra as single agents failed to significantly enhance marginal mass human islet engraftment in this model.
Twenty-four hours after transplantation, mice were killed and grafts assessed for apoptotic beta cells (n = 3 per group, processed in triplicate) (Figure 2). Grafts from mice in the control group had 23.47% apoptotic beta cells within the graft, which was significantly more than mice treated with anakinra (7.85%), etanercept (12.44%) or combined anakinra and etanercept-treated mice (0.67%) (p < 0.001, ANOVA). The combination group (anakinra + etanercept) demonstrated significantly lower beta-cell apoptosis than the etanercept group (p < 0.05, bonferroni post hoc analysis).
After 1 month, all transplanted immunodeficient mice were fasted overnight and underwent an IPGTT (Figure 3A). Area under the curve analysis for the IPGTT is displayed in Figure 3B. Mice receiving anakinra and etanercept displayed significantly improved (p < 0.05) glucose tolerance when compared to control mice. Forty-eight hours later, mice were killed and grafts recovered for insulin content analysis. Mice receiving combined anakinra and etanercept had significantly higher mean insulin content in their human islet grafts (17 927 mIU/mL) compared to mice receiving anakinra (10 689 mIU/mL), etanercept (11 881 mIU/mL) or IgG control (10 874 mIU/mL) (p < 0.05, ANOVA).
Anakinra and etanercept are not toxic to human isletsin vitro
Four separate human islet preparations were cultured for 48 h at 37 °C in CMRL supplemented with (1) anakinra; (2) etanercept; (3) IgG2 isotype-specific control; and (4) anakinra and etanercept. At 24 h of culture, there was no difference in the percentage of islet equivalents remaining (Figure 4A) (anakinra: 73.6%, etanercept: 75.2%, anakinra and etanercept: 82.7%, control: 76.8%; p > 0.05 by ANOVA). There was also no difference in the viability of islets at 24 h obtained with the SYTO green/EtBr technique (anakinra: 62.57%, etanercept: 72.43%, anakinra & etanercept: 72.43%, control: 63.97%; p > 0.05 by ANOVA). After a further 24 h there was no significant difference in islets remaining or viability between the groups (Figure 4B) (anakinra: 55.36% IE remaining, 80.8% viability; etanercept: 51.29% IE remaining, 75% viability; A&E: 66.13% IE remaining, 81.4% viability; control: 39.75 IE remaining, 70.4% viability p > 0.05 for both IE remaining and viability by ANOVA).
Combined anakinra and etanercept improves marginal mass islet engraftment in syngeneic mouse islet transplantation
To further test the impact of these anti-inflammatory agents in a syngeneic mouse marginal mass model, we found that the combination of anakinra and etanercept displayed the highest rate of diabetes reversal (61.5%) followed by those receiving etanercept alone (31.8%), control (27.3%) and the anakinra group (18.2%) (p < 0.05 for anakinra and etanercept vs. anakinra by log-rank analysis) (Figure 5). Administration of either etanercept or anakinra given alone however failed to demonstrate significant benefit over control.
After 1 month, recipient syngeneic mice were fasted overnight and underwent glucose tolerance testing (Figure 6A). Area under the curve (AUC) was calculated (Figure 6B) for each curve in Figure 6A. There was no significant difference in AUC between nonhyperglycemic mice in each of the groups (P > 0.05 by one-way ANOVA). Forty-eight hours after glucose tolerance testing, all mice were killed and their graft analyzed for insulin content (Figure 6C). There was no significant difference between the groups (anakinra 151.1 ng/mL, etanercept 131.2 ng/mL, anakinra and etanercept 186 ng/mL, control 103.3 ng/mL; p > 0.05 by ANOVA)
The use of cocktails of different anti-inflammatory agents has become integrated into the routine peritransplant management of clinical islet transplant patients, often without robust supportive appropriate preclinical or clinical data, and with substantial cost of therapy. The justification for this empiric approach has been based on (1) prohibitive cost of randomized controlled trials, (2) lack of adequate power based on limited numbers of clinical islet transplants, and (3) temptation of centers to adopt any potential therapy that might have positive impact in clinical islet transplantation, irrespective of the presence or absence of supportive data. It is estimated that approximately 700 subjects have received islet transplants worldwide. By contrast, the phase III trials required to justify safety and efficacy of anti-TNF-alpha and IL1Ra in rheumatoid arthritis involved large numbers of patients per arm, clearly not feasible in islet transplantation given the relative shortage of donor transplant organs (14–16).
The application of anti-TNF-alpha blockade became integrated into clinical islet transplantation largely based on a single publication by Farney et al. using syngeneic mouse islet and not human islet transplants in mice (2). Hering et al. promoted the clinical use of etanercept in their single-donor islet transplant series (3) as part of several innovations both in islet preparation and in peritransplant management (17). Oberholzer et al. also adopted routine use of etanercept in their single donor transplant series (11). In Edmonton, we initiated a randomized controlled trial for the investigation of an alternative anti-TNF-alpha medication (infliximab 10 mg per kg per day for 14 days), but abandoned this trial after the first 12 subjects were treated, as we found no significant difference in rates of single-donor transplant success, and recognized that a large number of subjects would be required to detect a small difference in outcome (Shapiro et al., unpublished data). The Miami group also investigated infliximab in clinical islet transplantation, and found no significant clinical benefit (18).
The agent IL1Ra (anakinra) has also been found to be highly effective in controlling inflammation in rheumatoid arthritis in over 2000 subjects, as reviewed in a recent comprehensive Cochrane Review (19). Larsen et al. found improved glycated hemoglobin and enhanced C-peptide secretion in patients with type 2 diabetes (20). Matsumoto et al. adopted the combined strategy of anakinra and etanercept in three subjects undergoing clinical islet transplantation, and while the early results were promising, they were unable to sustain this trial through lack of funding. Of concern, Health Canada and the Food and Drug Administration issued a health warning regarding the chronic coadministration of these two agents in patients with rheumatoid arthritis, as with 24-week therapy there was a higher incidence (7%) of serious infections compared with monotherapy (1.8%) (21,22). Clearly the short-term, 1-week administration of anakinra and etanercept in the islet transplant setting is different from the chronic dosing required in rheumatoid arthritis, but islet transplant patients are also subjected to profound T-cell depletional inductional and potent immunosuppressive maintenance therapies; thus this potential risk does merit consideration.
We therefore set out to revisit the potency of both anti-TNF-alpha and IL1Ra therapies in a preclinical model of human islet transplantation using the marginal mass approach in immunodeficient mice, principally to assess efficacy in a stringent, relevant, islet engraftment model.
The principal findings of the current study are that the combination of both anakinra and etanercept led to marked and meaningful improvement in the engraftment of both human and mounse islet transplants in mice. Each of these agents when used alone failed to produce statistical significance, but synergy in response to the combination of therapies led to positive impact, supporting the approach of Matsumoto et al.
IL-1, TNF-alpha and interferon-gamma are generally regarded as the most toxic cytokines mediating islet injury in laboratory models (23–28). IL-1 causes detrimental effects on pancreatic islets including decreased insulin secretion, and islet death (26). The binding of IL-1 to its receptor (IL-1R) leads to the activation of NF-κB and the regulation of multiple genes including IL-1, IL-6 and TNFα (29,30). Activation of NF-κB also leads to the expression of inducible nitric oxide synthase (iNOS) and the subsequent production of nitric oxide (NO), a potent cause of β-cell apoptosis (25,26,31). TNFα can kill β-cells directly through binding to surface receptors, which contains a death domain, and works together with IL-1 to stimulate NO production (32). Preventing the action of these cytokines on transplanted islets has therefore been hypothesized to protect the engrafting β-cell mass in the early posttransplant period, where over 60% of the infused islet mass is destroyed within hours to days posttransplant (33,34).
To better understand the underlying mechanisms associated with the positive improvement in islet engraftment in the current study, we found that the dominant protective effect was mediated through a significant reduction in early (24 hour) posttransplant apoptosis when mice are treated with either anakinra or etanercept, with the combination producing a more profound effect, leading to improved 30-day metabolic reserve. Clearly there could be several other mediators favoring islet survival in our study that may be unrelated to the effects upon apoptosis.
Binding of IL-1 to its receptor leads to intracellular changes which can negatively affect β-cells, including the activation of, MAPKs (ERK, p38, JNK), NF-κB, protein kinase Cδ and the induction of Fas expression (35–38). Activation of the former three leads to iNOS expression and subsequent β-cell apoptosis while the latter increases cell sensitivity to Fas ligand (FasL) and downstream caspase activation. TNFα binds to its receptor, which contains an intracellular death domain, leading to downstream Nf-κB activation and direct activation of the caspase cascade (39). In addition, TNFα strongly potentiates the cytotoxic effects of IL-1 on β-cells by synergistically augmenting the MAPK signaling pathway (40). This may explain the synergistic effect of the combination of IL-1 and TNFα blockade in this study.
We did not observe any detrital effect on islet survival when human islets were exposed to relevant concentrations of anakinra or etanercept in vitro, but in these studies we did not specifically challenge the human islets in vitro with addition of toxic cytokine cocktails (2,23,41,42). We did not anticipate a protective effect therefore in our in vitro studies; however, there was not a detrimental effect on islet viability or survival.
Interestingly, we found more positive impact of combined anakinra and etanercept in human islets than in the syngeneic mouse islet transplants. The exact reason for this is unclear, but this likely reflects the fact that human islets have undergone much more prolonged periods of injury during brain death procurement, cold ischemic transportation, more prolonged isolation, purification and culture, than in the simpler mouse islet preparation model (7,43–48).
While the results of this study are promising, there are evident limitations that preclude direct extrapolation to the clinical setting. We clearly acknowledge that the current studies are restricted to islet implantation in the renal subcapsular space and not the intraportal site in mice, and that any impact of the combination of Anakinra and Etanercept observed in these experiments may or may not be applicable to the intraportal site, which is generally considered to harbor a more intensive inflammatory milieu. Our rationale for not formally testing the intraportal site reflects our previous experience with marginal mass intraportal islet transplantation in mice, where we observed less consistent response both with mouse and human islets transplanted in this site. We recognize that this deviates from the clinical setting, where the intraportal site is current favored and provides a more efficient environment for islet implantation. While technically possible, assessment of insulin content and apoptosis is easier to measure in the renal subcapsular site where islets are aggregated, and is more challenging in the intrahepatic site. Clearly the intraportal site may be associated with a different profile of dynamic cytokine exposure compared with the renal subcapsular space, that may have either masked or enhanced potential differences in therapeutic efficacy, but this was not explored further. We did not directly quantify the cytokine milieu in the peritransplant site or in the serum of the transplanted mice, which may have revealed additional data. Based on the known mechanisms of action of anakinra and etanercept, we would anticipate that there would be reduced local expression of both IL-1 and TNF-alpha, but this was not directly measured in the current studies. We therefore assume but cannot assign direct causality to combined blockade of IL-1 and TNF-alpha leading to reduced apoptosis and improved islet mass survival that was clearly established by these studies. Clinically, Froud et al. demonstrated marked perturbations in TNF-alpha levels in the peritransplant period (49–51).
In summary, the current study provides important, positive supportive data to justify the combined use of both TNF-alpha blockade and IL1Ra therapy for short-term use in clinical islet transplantation. Conversely, the study questions the efficacy of etanercept monotherapy as currently adopted in most clinical islet transplant programs. Clearly, randomized controlled clinical trials are needed to fully address this issue, but will be challenging if not impossible to implement in the present environment of limited funding and limited clinical transplant activity.
AMJS is supported by a Senior Clinical Scientist award from Alberta Innovates-Healthcare Solutions (AIHS). MDM is supported by an AIHS Clinical Fellowship and the Clinician Investigator Program from the University of Alberta. Infrastructure support for the clinical islet transplant program and the clinical islet isolation laboratory comes from Alberta Health Services, the Diabetes Research Institute Foundation of Canada (DRIFCan), and from grant support from the Juvenile Diabetes Research Foundation (JDRF), the National Institutes of Health (NIH) and the National Institute of Allergy, Immunology and Diabetes (NIAID) through the Collaborative Islet Transplant Trials (CIT) Group.
The authors of this manuscript have no conflicts of interest to disclose as described by the American Journal of Transplantation.