|Home | About | Journals | Submit | Contact Us | Français|
The effectiveness of pancreatic islet isolation must be maximized to make islet cell transplantation (ICT) a standard therapy. We have performed 100 human islet isolations at Baylor Research Institute including islet isolations for research, for clinical allogeneic transplantation, and for autologous islet transplantation. In this study, we analyzed the results of these isolations. First, we assessed 79 islet isolations using brain-dead donors to determine variables associated with successful islet isolation. Univariate logistic regression analysis revealed that seven variables influenced the success of islet isolation for allogeneic ICT: cause of death, mechanism of death, techniques for pancreas procurement and preservation, heavy fatty infiltration, collagenase type, dilution time, and islet purification method. Multivariate regression analysis revealed that only the current isolation protocol, the Baylor Islet Isolation Method (BIIM)—with its four required elements of pancreas procurement by the team, pancreatic ductal injection, the two-layer method with perfluorocarbon, and density-adjusted density gradient purification—had a significant positive impact on successful islet isolation (P = 0.02). Second, we compared allogeneic and autologous ICT using the BIIM. There were no significant differences in islet yields between allogeneic and autologous ICT using the BIIM; total islet yield after purification was 628 ± 84 × 103 IE in allogeneic ICT vs. 576 ± 49 × 103 IE in autologous ICT (P = 0.59). This retrospective study revealed that the BIIM provided favorable outcomes for both autologous and allogeneic ICT.
Islet cell transplantation (ICT) is a promising treatment for type 1 diabetic patients who have damaged insulin secretory capacity due to autoimmunity (1, 2). It was reported that islet recipients could achieve insulin independence, improve glycemic control and quality of life, and prevent chronic diabetic complications, although these findings were based on a limited number of study cohorts (3–6). Recently our survey revealed that >85% of type 1 diabetic patients wanted to become insulin independent, and >75% believed that transplantation should replace insulin injection therapy (7). However, several issues remain before ICT can be a standard treatment, including the use of multiple donors to achieve and/or maintain long-term insulin independence and the adverse effects of immunosuppressants (3, 5, 8, 9). The modification of islet isolation has helped improve the amount and quality of isolated islets, which is one factor for successful ICT. Since February 2007, we have implemented a modified islet isolation method based on the Kyoto islet isolation method (KIIM), a method developed for non–heart-beating donors in Japan (10, 11), to improve the efficacy of ICT.
ICT using the recipient's own islets is also useful to maintain excellent glycemic control in patients with chronic pancreatitis who undergo a total pancreatectomy (autologous ICT) (12, 13). A cytoprotective strategy during islet isolation might be more important in autologous ICT, which uses pancreata with chronic inflammation. Baylor Regional Transplant Institute has performed autologous ICT since October 2006 and has used a modified islet isolation method, called the Baylor Islet Isolation Method (BIIM), since December 2007. We have already reported 17 cases of autologous ICT and demonstrated excellent graft function and improvement of intractable abdominal pain (14).
In February 2010, the total number of islet isolations performed by Baylor Research Institute reached 100. Herein, we review all islet isolations with two objectives: 1) to determine variables associated with successful islet isolation for allogeneic ICT and 2) to investigate the differences in donor characteristics and the islet isolation outcome between allogeneic and autologous ICT.
All records of pancreatic islet isolations performed at Baylor Research Institute (Dallas, TX) from January 2005 to February 2010 were reviewed on the basis of previous reports (15–19). First, we conducted univariate logistic regression analysis followed by multivariate analysis using data from islet isolations with brain-dead donors in order to identify factors that influence the success of islet isolation. Successful islet isolations were defined as islet yields >300,000 islet equivalents (IEs). This amount allows approximately 5000 IE per kg of body weight, since the average body weight of islet recipients is approximately 60 kg, according to the latest annual report of the Collaborative Islet Transplant Registry (20). Four isolations using brain-dead donors were excluded from this analysis due to technical failures of islet isolation. Second, we compared donor characteristics and outcomes of islet isolations using the same method between autologous and allogeneic transplantations (Figure). The institutional review board of Baylor Research Institute approved this study.
Data on characteristics of pancreas donors were obtained from donor charts provided by organ procurement organizations (LifeGift, Fort Worth; Southwest Transplant Alliance, Dallas) or from patient records for autologous ICT. Categories of donor characteristics were defined based on the United Network for Organ Sharing/Organ Procurement and Transplantation Network data collection system (21). For donor cause of death, three groups were used for statistical analysis: cerebrovascular/stroke, head trauma, and other. Similarly, for mechanism of death, three categories were used: blunt injury, intracranial hemorrhage/stroke, and other. Donor body mass index (BMI) was calculated using the formula of kg/m2, and body surface area was calculated using Mosteller's formula of √[height (cm) × weight (kg)/3600] (22).
For the standard method, a donated pancreas was procured and shipped by a remote center team. The harvested pancreas was preserved in chilled University of Wisconsin (UW) solution. The following modifications were recorded: “pancreas procurement by islet team” when surgeons who belonged to our team procured the pancreas, “ductal injection” when approximately 1 mL/g of chilled preservation solution (ET-Kyoto solution, Otsuka Pharm Fact Inc., Naruto, Japan, or similar solution) was injected into the main pancreatic duct (10, 11, 23, 24), and “oxygen-charged static two-layer method (TLM)” when oxygenated perfluorocarbon (PFC) and preservation solution (25) were used for pancreas preservation. Cold ischemia time was defined as the time from cross-clamp to the start of perfusion.
All isolations were performed based on the modified Ricordi method (26). Islets were purified with the COBE 2991 cell processor (CaridianBCT, Inc., Lakewood, CO) with a continuous density gradient using Ficoll, iodixanol, or both (27). Islet viability after purification was evaluated with fluorescein diacetate (10 μmol/L) and propidium iodide (15 μmol/L) staining (28). The average viability of 50 islets was calculated. Islet yield was assessed using dithizone staining (Sigma Chemical Co., St. Louis, MO) (2 mg/mL) and converted into a standard number of islet equivalents (IE, diameter standardizing to 150 μm) (29).
Since the KIIM was originally developed for non–heart-beating donors, we modified the method for brain-dead donors (Table (Table11). The modified method consists of four factors—pancreas procurement by the islet team, ductal injection with preservation solution, the TLM with oxygenated PFC, and density-adjusted continuous gradient purification with iodixanol—which together define the BIIM. The BIIM has also been applied for autologous ICT (14). When comparing isolations for allogeneic ICT and those for autologous ICT, we included only cases that met all four criteria for the BIIM to avoid confounding effects due to differences of pancreas preservation and islet isolation methods.
Numbers and percentages were reported for categorical variables, while means and standard errors were calculated for continuous variables. P values < 0.05 were considered statistically significant. All statistical analyses were performed by JMP (version 8.0; SAS Institute).
To find possible factors associated with successful islet isolation of brain-dead donors, univariate logistic regression analysis was performed preliminarily, calculating odds ratios (OR) and P values for each variable. For continuous variables, OR and P values were calculated using variables above the 75th percentile and below the 25th percentile. Afterwards, selected factors with statistical significance (P < 0.05) in the univariate logistic regression model were evaluated by multivariate logistic regression analysis, which allows correction of covariates as predictors of islet isolation success. Hosmer-Lemeshow statistics were used to assess the goodness of fit to islet isolation success (30).
Allogeneic and autologous ICT were compared by evaluating continuous variables by Student's t test and categorical data by Fisher's exact test.
Characteristics of brain-dead donors. In all 79 cases, pancreata were obtained from brain-dead donors who had no past history of diabetes mellitus. Cerebrovascular/stroke as cause of death and intracranial hemorrhage/stroke as mechanism of death were significantly associated with unsuccessful islet isolation (OR = 0.29, P = 0.01 and OR = 0.36, P = 0.04, respectively) (Table (Table22). Since 88% (30/34) of cases of cerebrovascular/stroke also had intracranial hemorrhage as mechanism of death (Fisher's exact test P < 0.0001), cerebrovascular/stroke was input for further multivariate regression analysis. Gender had a marginally significant correlation with islet isolation success (OR of women = 2.64, P = 0.053).
Variables for pancreas procurement, preservation, and islet isolation. Pancreas procurement by the islet team had a significant positive impact on successful islet isolation (OR = 7.14, P = 0.0001) (Table (Table33). Both ductal injection and the TLM at pancreas preservation were also found to be statistically significant (OR = 8.57, P < 0.0001 and OR = 4.20, P = 0.03, respectively). All pancreata were intact and not edematous. When heavy fat infiltration was observed, the islet isolation was more likely to be unsuccessful (OR = 0.35, P = 0.04).
For digestion, Liberase HI and Collagenase NB were found to be significant variables (P = 0.001 and P = 0.04, respectively), although Liberase HI was only used before March 2007 and Collagenase NB was only used after May 2007 (Table (Table44). A long dilution time significantly correlated with success (OR = 5.82, P = 0.03). Continuous gradient purification was performed at all islet isolations. For density gradient of purification, density-adjusted iodixanol was significantly associated with success (OR = 3.24, P = 0.02), while Ficoll was less successful (OR = 0.36, P = 0.03).
In May 2007, we started to implement the BIIM. Cases that involved the BIIM were associated with successful islet isolation (OR = 8.64, P = 0.0002). To avoid colinearity among the modified techniques, the BIIM was used for further multivariate analysis instead of the following variables: pancreas procurement by the islet team, ductal injection, the TLM, and density-adjusted density gradient for purification. Success rates were significantly higher for the BIIM than for other isolation methods: 27/30 (90.0%) and 25/49 (51.0%), respectively (P = 0.0005, Fisher's exact test).
Based on results of univariate logistic regression analysis (Tables (Tables22–4) and considerations about colinearity as mentioned above, several variables were included for further multivariate analysis: cerebrovascular/stroke as cause of death, heavy fatty infiltration in the pancreas, Roche Liberase HI and Serva Collagenase NB as collagenase, dilution time, and the BIIM. Based on the Hosmer-Lemeshow statistic, the final model had a good fit (P = 0.25). Only the BIIM was a significant variable among the selected items (OR = 5.91, P = 0.02) (Table (Table55).
This part of the study compared the five cases of allogeneic ICT and the six cases of autologous ICT that employed the BIIM. The six autologous ICT patients had had symptoms of chronic pancreatitis for 5.6 ± 0.7 years. Before autologous ICT, all had regularly used narcotics for pain management (morphine equivalent dose : 339 ± 96 mg/day) and had undergone endoscopic sphincterotomy or endoscopic stent replacement while complaining of severe abdominal pain of 7.3 ± 0.8 on a visual analog scale (with 0 indicating no pain and 10 indicating severe pain) (31). None had had surgical operations for chronic pancreatitis or had a history of diabetes mellitus. Four cases of chronic pancreatitis were idiopathic, one was related to pancreatic divisum, and one was related to alcohol.
Donor characteristics. There were no significant differences between allogeneic and autologous ICT in donor characteristics, except for age. The ages of donors for allogeneic ICT were significantly higher than those for autologous ICT (P = 0.04, Table Table66).
Outcomes of islet isolation. There was no significant difference in pancreas weight between allogeneic and autologous ICT, but cold ischemia time in allogeneic ICT was significantly longer than that in autologous ICT (P = 0.0002, Table Table77). There were no significant differences between the two groups in islet yield after digestion or after purification. The purity of allogeneic ICT was significantly higher than that of autologous ICT (P = 0.0004). In addition, tissue volume after purification for autologous ICT was significantly higher than that for allogeneic ICT (P < 0.0001).
Size of isolated islets. Smaller-sized islets were observed more frequently in isolations for autologous ICT than in those for allogeneic ICT, which was evidenced by the significantly higher ratio of 100 to 149 μm islets for autologous ICT (P = 0.03) (Table (Table88).
This report analyzed the first 100 human pancreatic islet isolations completed at Baylor Research Institute. Markmann et al indicated that approximately 100 human islet isolations were necessary before initiating clinical islet transplantation (32). We initiated clinical islet transplantation at Baylor Regional Transplant Institute before performing 50 human islet isolations. In addition, we have had a high success rate of clinical islet isolation and achieving insulin independence after islet transplantation (23). Those excellent results were mainly due to adaptation of the KIIM (10, 11, 23, 24).
In modifying the KIIM (10, 11), we eliminated the in situ cooling system (33) because it was designed to minimize warm ischemic injury after cardiac arrest, and the BIIM was developed for brain-dead donors without warm ischemic injury. We also eliminated ulinastatin for ductal preservation because ulinastatin was not available in the United States and the effect of trypsin inhibitors on islet isolation was controversial (34, 35). Therefore, we used Kyoto solution or a similar solution without ulinastatin for ductal preservation in the BIIM. To perform ductal preservation, at least one surgeon from the islet team was involved in the pancreas procurement. We used the oxygen-charged static TLM and the standard Ricordi method for both the KIIM and the BIIM; those elements were both stable and reliable and required no modification. Density-adjusted density gradient was originally developed for islet purification using brain-dead donors (36), and it was also beneficial for non–heart-beating donors (10, 11). Therefore, density-adjusted density gradients were used for both the KIIM and the BIIM.
In the first part of the study, variables associated with successful isolation were investigated because one important element of successful ICT is maximization of islet yields (15–19, 24). In the second part, differences between allogeneic and autologous ICT were investigated using cases that implemented the BIIM.
Previous reports have demonstrated that deceased donor age, BMI, cause of death, and length of hospitalization affected islet isolation outcomes (15–19). In this study, cause of death had statistical significance in univariate analysis, while the other donor characteristics did not reach statistical significance. This might be explained by the fact that we accept obese donors for islet isolation. In fact, in our study, the average donor body weight, BMI, and body surface area in the unsuccessful isolation group were higher than in previous reports and close to that of others' successful group.
In the first part of the study using univariate regression analysis, it was found that the modified techniques for pancreas preservation, including ductal injection with chilled preservation solution and the TLM with oxygenated PFC, were associated with islet isolation success. According to the latest annual reports of the Collaborative Islet Transplant Registry and Islet Cell Resource Centers, University of Wisconsin solution alone was used for approximately 50% to 60% of pancreas procurements (20, 37). Previous papers reported that the TLM as well as ductal injection significantly improved the outcome of islet isolation (23, 25, 38, 39). Pancreas procurement by the islet team made it possible to perform these special techniques smoothly.
The density-adjusted continuous density gradient purification with iodixanol, which reduces the force to islets during centrifugation by its low viscosity (10, 40, 41), significantly improved islet isolation. Of note, the modified techniques mentioned above were initiated in early 2007, and there was a possibility that these variables interact with each other in determining islet isolation success. To avoid confounding effects with the other variables, multivariate logistic regression analysis was performed for selected variables on the basis of results of univariate analysis. Multivariate logistic regression analysis revealed that the BIIM, our current isolation protocol, was an independent predictor of success. The combination of multiple modified techniques including procurement by the islet team, ductal injection, the TLM, and density-adjusted iodixanol-based continuous purification could contribute to improve islet isolation.
Heavy fatty infiltration was found to be a negative predictor of isolation success in univariate analysis. However, previous studies demonstrated that fatty infiltration of the pancreas was beneficial for isolation success (15, 18). This discrepancy is most likely due to the association between fatty pancreas and high BMI. In our series, most pancreata were from high-BMI donors; therefore, the positive effect of high BMI was not apparent. On the contrary, although fatty infiltration inhibited collagenase delivery and activity, the positive effect of high BMI might have overcome this disadvantage if pancreata from both low- and high-BMI donors were accepted.
In univariate analysis, collagenase type was associated with the outcome of islet isolation. However, as described in the Results section, Liberase HI was used for old isolations and Collagenase NB was used for recent isolations, where modified pancreas preservation techniques and islet isolation methods were implemented. Therefore, careful interpretation is needed for univariate analysis of collagenase types, and in multivariate analysis no significant effects were observed based on collagenase type.
A longer dilution time was also associated with islet isolation success. Earlier, it was shown that a short digestion time was associated with high islet yield due to avoidance of overdigestion (16). In this study, digestion time was shorter in the high islet yield group even though it did not reach statistical significance. A longer dilution time might reflect underdigestion, which should increase islet yield due to avoidance of overdigestion as long as the amount of undigested pancreas is within an acceptable range.
Multivariate logistic regression analysis revealed that only the BIIM had a significant impact on successful islet isolation with brain-dead donors. This suggests that the BIIM could overcome other factors to achieve successful islet isolation.
Next, we compared variables related to donor characteristics and islet isolation between allogeneic and autologous ICT where the BIIM was fully performed. This analysis can clarify the differences of donor variables and isolation outcomes between allogeneic and autologous ICT, while avoiding effects of different islet isolation methods. There were no significant differences in donor characteristics except for age. The difference of donor age between allogeneic and autologous ICT might be due to donor selection for allogeneic ICT. It should be noted that donors are selected for whole pancreas transplantation prior to ICT, and younger donors are favorable for whole pancreas transplantation (42). In addition, there was a significantly longer cold ischemia time for allogeneic ICT. On the other hand, pancreata for autologous islet isolation had inevitable chronic pancreatitis. It was shown that chronic pancreatitis inhibited pancreas digestion, which resulted in poor islet yield (16). Older donor age and prolonged cold ischemia time were negative factors for allogeneic ICT, and chronic pancreatitis was a negative factor for autologous ICT. Interestingly, islet yields both after digestion and after purification did not have significant differences between the two groups when the BIIM was applied. Therefore, the BIIM should be a favorable technique for both allogeneic and autologous ICT.
Final tissue volume was significantly different between the two groups, so purity also was significantly different. This result can be explained by the difference of our protocols for maximum volume of final product: 10 mL for allogeneic ICT and 15 mL for autologous ICT. All islet preparations for both allogeneic and autologous ICT were safely transplanted into the liver, except for one autologous ICT case where 602,709 IE was infused into the liver and 117,067 IE into the intraperitoneal cavity (14), and no portal vein thrombosis was observed in our cohort.
Interestingly, smaller islets were observed in isolation for autologous ICT compared with allogeneic ICT. Pancreata with chronic inflammation would be easily overdigested since a previous report suggested that overdigested islets became smaller compared with pre- and post-isolation islet sizes (18). Therefore, careful digestion is necessary to avoid overdigestion, especially for autologous ICT.
In summary, we reviewed 100 pancreatic islet isolations at Baylor Research Institute. The BIIM was found to be a predictor of islet isolation success in multivariate logistic regression analysis. Comparison of allogeneic and autologous ICT revealed that the BIIM provided favorable outcomes for both.
The authors thank Ms. Sara Shah for the grammatical editing of the manuscript and Ms. Yoshiko Tamura and Mr. Greg S. Olsen for their technical support.