Transplantation

Diabetes is a disease of insufficient β-cell mass. In type 1 diabetes, this deficiency is nearly absolute (¹), with poorly understood autoimmune mechanisms targeting pancreatic islets for destruction (²). Conversely, type 2 diabetes is marked by a decreased functional β-cell mass (³), characterized by a combination of β-cell dysfunction and insulin resistance (⁴). Consequently, therapies aimed at restoring β-cell mass are the focus of ongoing research, especially in light of evidence that tight blood glucose control can prevent and even reverse the secondary complications of diabetes (^{5, 6}).

Solid organ pancreas transplantation has been established as the gold standard with respect to reversal of diabetic hyperglycemia (⁷), although the shortage of available donor organs has limited the clinical application of this therapeutic approach. Islet transplantation, it was thought, might represent a means to compensate for limited organ availability. Thus, given that diabetes only becomes evident with a reduction in β-cell mass greater than or equal to 50% (¹), it was suggested that islet graft recipients would accordingly require far fewer islets to restore normoglycemia than would normally be found in the whole pancreas. The potential of improved donor:recipient ratios, coupled with the original notion that isolated islets were less immunogenic than the whole organ, led to significant efforts in the fields of islet isolation and transplantation. The landmark Edmonton protocol, which proved conclusively that diabetic hyperglycemia could be reversed with islets from two to three donors (⁸), was followed by reports of restoration of normoglycemia in recipients of islets from a single donor (⁹). However, recent data suggest that the half-life of graft function remains a constraint (¹⁰).

One benefit of islet transplantation programs has been the increasing availability of high quality human islets for research. This has been instrumental in recent investigative efforts that have demonstrated that cultured islets may be induced to dedifferentiate, proliferate, and redifferentiate, thereby providing a potential pathway to improving islet yield per donor organ (^11–15). Thus, while suitable donor pancreata are considered initially for solid-organ transplantation, studies to identify ways to enhance islet isolation, survival, and cell mass remain an important preoccupation. In fact, studies suggest that organs best suited for solid-organ grafts may differ from those that produce high islet yields (^16–18).

Before the implementation of a human islet replacement program at our institution, we have benefited from having received and processed cadaveric human donor pancreata for islet isolation for research purposes for a decade. In this report, we describe the donor and isolation variables that we have identified as predictors of clinical islet isolation success.

MATERIALS AND METHODS

Organ Procurement and Islet Isolation

Pancreata from adult human cadaveric organ donors were obtained through the local organ procurement organization. Islets were isolated according to established protocols (¹⁹). Briefly, organs were flushed in situ with cold University of Wisconsin (UW) organ preservation solution and removed en bloc with the spleen and a stapled loop of duodenum. Organs were transported on ice to the islet isolation facility.

On receipt, organs were dissected free of spleen, duodenum, and extraneous tissue, including fat. Next, the main pancreatic duct was identified, cannulated, and perfused for 10 min with a chilled solution containing Liberase HI or CI (Roche Diagnostics, Laval, QC) or Serva Collagenase (Helixx, Scarborough, ON). The distended organ was then cut into pieces, placed in a closed recirculating system (Bio-Rep, Miami, FL), and heated to 37°C to activate the enzyme blend. During the course of enzymatic and mechanical dissociation of the gland, samples were taken and evaluated in real-time, using dithizone (Sigma, St. Louis, MO). After the appearance of islets unembedded in acinar tissue, the circulation circuit was cooled and the digestate collected in wash solution (Mediatech, Herndon, VA). After washing, the digestate was pooled and incubated on ice in UW solution to allow recovery.

Tissues were separated by continuous Ficoll (Biochrom AG, Berlin, Germany) density gradient centrifugation, using a 1.077 to 1.100 g/mL gradient established in a cell processor (COBE DCT, Denver, CO). Aliquots determined to contain more than or equal to 70% free islets at a purity greater than or equal to 70% were collected, washed, and counted as islet equivalents (IE=islet with diameter of 150 μm). Isolations were considered successful if the final yield was greater than or equal to 250,000IE.

β-Cell Volume and β-Cell Mass

Tissue samples were taken from the tail of the pancreas before isolation, and stored at 4°C in phosphate-buffered formalin. After routine processing and embedding, 4 μm sections were cut and dewaxed in xylene and petroleum ether. Slides were permeabilized by incubation in Tris-buffered saline with 0.1% Triton X-100 for 5 min, before antigen retrieval by heating for 10 min in 0.1 M citrate buffer, pH 6. Slides were then incubated in blocking buffer (Zymed, San Francisco, CA) for 15 min before overnight incubation at 4°C with an α-insulin primary antibody (Dako, Carpinteria, CA). After washing in phosphate-buffered saline, staining was developed using a biotinylated broad-spectrum secondary antibody, streptavidin-conjugated horseradish peroxidase (HistoStain Plus kit; Zymed, San Francisco, CA), and 3,3′-diaminobenzidine substrate (Dako). Sections were analyzed using a Zeiss Axioskop 40 microscope and Northern Eclipse v6.0 (Empix Imaging, Mississauga, ON). Briefly, individual insulin⁺ structures were traced and thresholded to determine insulin⁺ area (²⁰). Total pancreatic tissue area was similarly determined, and β-cell volume was calculated relative to total pancreatic area. β-Cell mass was calculated by normalizing to pancreatic weight.

Statistical Analysis

Variables of interest were identified based on the literature (^{16–18, 21–29}) and empirical observations. Donor variables included age, sex, height, weight, body mass index (BMI=weight/height²) and body surface area (BSA=√[height×weight/3600]), as well as a routine biochemical blood screen. Procurement variables included length of donor hospitalization, organ procurement organization coordinator, procurement team, cause of death, periprocurement hemodynamic measures, and use of vasopressors. Organ and isolation variables included assessment (yes/no) of fatty infiltration and pancreas quality (capsule damage, edema, or fibrosis); pancreatic weight; degree of distension after perfusion; enzyme type, lot, and specific activity; duration of cold ischemia and enzymatic digestion; and the proportion of the yield derived from islets with diameters <150 μm (small islets). There was little turnover over the 5-year study period with respect to laboratory members involved with islet isolations.

Results are presented as mean±SEM. Statistical significance between means was determined using two-tailed Student's t tests. Linear regression analysis was performed to relate continuous variables with islet yield, whereas χ² analyses were used to assess categorical variables with respect to isolation success. Stepwise logistic regression analysis was performed to establish a simple probability model to predict isolation success. Differences were considered significant at P<0.05.

RESULTS

Islet Isolations

From January 2002 to June 2007, the Centre for Pancreatic Diseases at the McGill University Health Centre received 242 donor organs for islet isolation. Of these, 19 (7.9%) were from donors diagnosed with type 2 diabetes, and were excluded from this study. A further 52 organs were not processed for isolation because of donor- or procurement-related issues. As a result, 171 human islet isolations were available for analysis.

The average yield was 223,000±14,000IE, although 18 isolations did not produce any islet yield with a purity of greater than or equal to 70%. The majority (71.3%) of isolations produced yields of greater than or equal to 100,000IE, including 38.0% with yields greater than or equal to 250,000IE, considered to be successful (⁸).

The number of organs received annually increased sharply at the beginning of the study period, but then leveled off, as did the average islet yield (Fig. 1A). Increased organ numbers were associated with increased isolation success, and better discrimination of organ quality. This, in turn, led to a concomitant decrease in the number of isolations performed (Fig. 1B).

To identify factors that predict islet yield, we performed linear regression analysis between continuous variables and islet yield. Of the variables considered, donor age, height, weight, BMI, and BSA were positively correlated with islet yield, though not strongly, whereas the duration of enzymatic digestion and proportion of the yield composed of small islets were negatively correlated with islet yield (Table 1). Based on these, we determined variable values that would correlate with a yield of 250,000 IE.

To better understand the effect of these continuous variables on isolation outcome, we compared the mean values associated with successful and unsuccessful isolations. In this analysis, the effect of BMI was no longer significant (Table 2). However, with respect to the other continuous variables identified, the mean values associated with successful and unsuccessful isolations were found to separate around the calculated values associated with a yield of 250,000IE. Put differently, the predicted values calculated in Table 1 seem to serve well as threshold values for isolation success.

We next analyzed the average yields associated with categorical variables, and identified male donors and local procurement team as associated with higher islet yields. A similar trend was identified for pancreatic fatty infiltration (Fig. 1C). Male donors tended to be taller (171.8±0.7 vs. 160.5±0.9 cm, P<0.001) and heavier (78.8±1.6 vs. 67.8±1.8 kg, P<0.001) than female donors, resulting in increased BSA (1.94±0.02 vs. 1.73±0.03 m², P<0.001) but no difference in BMI (26.9±0.5 vs. 26.2±0.6 kg/m², ns). All these variables were associated with increased islet yield (Table 1). With respect to the effect of the procurement team, subjective assessment of ½organ quality, by blinded observers, was linked to local procurement (58.7% vs. 32.4%, P=0.007), whereas cold ischemia time did not differ between local and other procurement teams (7.6±0.5 vs. 7.3±0.5 hr, ns). In fact, all procurement teams in this study are based in the same city, suggesting that the observed effect is unrelated to logistic considerations.

All categorical data, including threshold values for continuous variables, estimated by linear regression, were analyzed for associations with isolation success (Table 3). Again, BMI was no longer associated with isolation success, although BSA remained significant. The only new categorical variable to be identified was that of fatty infiltration of the pancreas, which was more common in successful isolations.

Logistic Regression

To establish a model to predict isolation success, we performed stepwise logistic regression analysis, using the factors identified from the preceding analyses of both categorical and continuous variables. Working backward to provide a simplified model for predicting isolation success, we identified three variables that, when combined, allowed for an accuracy of 72.3% with respect to predicting a successful isolation (Table 4).

Two-Layer Preservation and Serine Protease Inhibition

In the interest of maintaining an up-to-date isolation protocol, we assessed two recent improvements to the standard protocol. First, we assessed the two-layer method (TLM) of pancreas preservation (n=4), using perfluorocarbons as described (³⁰). Although this technique is designed for situations of longer preservation times, cold ischemia was kept relatively constant (6.9±1.3 vs. 7.6±0.3 hr) to study the direct effect on islet yield. Use of the TLM did not affect isolation yield (Fig. 1C), suggesting that there is no drawback to the use of this method in cases where longer cold ischemia times may be an issue.

Next, to examine the effect of serine protease inhibition, a commercially available inhibitor was added to the diluted enzyme blend, as described (³¹). The inhibitor was used in only three isolations and then abandoned due to empirical observations of unsuccessful isolation. Indeed, average yield was significantly reduced relative to isolations performed using the same enzyme lot, but without the inhibitor added (Fig. 1C).

Enzyme Lot

The use of Liberase HI has become the accepted standard with respect to pancreatic digestion for clinical islet isolation (³²). Each lot of Liberase has a distinct enzymatic profile, as assessed by Wunsch and Caseinase activity. While most isolations (93.0%) were performed using Liberase HI, a total of 22 lots were used for a minimum of three isolations each. The importance of enzyme lot was observed empirically, and indeed lot number was a strong predictor of isolation success (P=0.009). Moreover, while statistical analyses suggested that the duration of enzymatic digestion is a predictor of both isolation yield (Table 1) and success (Table 2), duration of enzymatic digestion tended to be associated with enzymatic lot (P=0.048), such that enzyme lot is a strong predictor of a digestion time of more than or equal to 12 min (P=0.004). There was no association, however, between Wunsch or Caseinase activity, and digestion length or isolation yield.

Enzyme Type

Based on experience, we identified young donors (≤25 years) as a population with suboptimal isolation results; with islet yield being significantly reduced (101,000±27,000 vs. 234,000±15,000 IE, P=0.006). Although decreased β-cell mass likely affects yield in young donors (³³), we hypothesized that Liberase CI might provide a better enzymatic composition for the retrieval of islets from organs of young donors, based on a tissue architecture that is comparable with canine pancreata. We tested the effect of Liberase CI on a limited number of donor organs (n=8), including young donors (n=4). Although the overall average yield did not differ between isolations performed with Liberase HI and CI, the average yield from young donors was significantly higher when Liberase CI was used (Table 5).

Pancreatic β-Cell Mass and Islet Size

We routinely obtained biopsies from the tail of donor organs before islet isolation, and we assessed β-cell volume and β-cell mass in a limited number of randomly selected organs (n=41). There was a significant correlation between isolation yield and β-cell volume, as determined by linear regression analysis (Table 1). Moreover, successful isolations were achieved from organs with significantly increased β-cell volume (Table 2).

The proportion of small islets (diameter <150 μm) seemed to be a strong predictor of both isolation yield (Table 1) and success (Table 2). This variable was identified as a surrogate for digestion quality, given that overdigested islets tend to be smaller. It is not surprising that isolations that produced a higher proportion of small islets also had lower overall yields, given that yields were calculated based on IEs. However, analysis of average in situ insulin⁺ cross-sectional area, a measure of islet size before digestion, shows no correlation with the proportion of small islets. This suggests that yields comprised of small islets are a result of the isolation itself. Moreover, the proportion of small islets was associated with enzyme lot (P=0.015), suggesting that overdigestion is a characteristic of certain enzyme lots.

DISCUSSION

Although much progress has been made during the past 35 years in the development of islet transplantation as a cellular therapy for diabetes, outcomes are highly variable between centers, and successful outcomes are difficult to predict (³⁴). Moreover, sustained graft function seems to be difficult to maintain for reasons that remain to be fully elucidated (¹⁰).

There are many variables associated with the islet transplant procedure, including those related to the donor, the pancreatic tissue, the isolation itself and, of course, the recipients. In the field of islet isolation, many attempts have been made to identify the critical variables that would predict high yields of high quality islets (^{16–18, 21–29}). However, the utility of such studies rests on the identification of variables that can be used to screen donor organs before the isolation procedure. To this end, a recent study reported on identifying donor variables predicting pancreatic weight (³⁵), which serves as a surrogate for islet yield (¹⁸). We did not observe such a correlation in our analyses (P=0.758). In fact, a review of the literature addressing isolation success indicates that few predictors are consistently identified, though these include donor age (^{16, 18, 23}), weight and BMI (^{17, 18, 22, 23, 26}), pancreatic quality (^{18, 24–26, 28}), and enzyme lot and digestion duration (^{18, 21, 23, 27, 29}); the same factors found in our predictive model. Other groups have also reported an effect of increased cold ischemia (^{16, 21, 36}), which we did not observe. Although islet yield and function are not necessarily related, we have previously analyzed islet viability and function (^{37, 38}), and our values are in keeping with those reported by other leading islet isolation facilities (³⁴).

A simple examination of the islet isolation procedure indicates that there are two requirements for a successful islet isolation; a large number of islets within the pancreas, and the ability to efficiently isolate them. Studies of donor and isolation variables are in fact identifying predictors of one or the other. Hence, donor variables associated with islet yield, namely age, weight, and BMI, are also associated with β-cell volume (^{33, 39}). We therefore sought to establish directly the association between β-cell volume and isolation yield. In fact, this is the first report, to our knowledge, comparing β-cell volume and islet isolation yield from the same donor organs. Linear regression analysis of a limited number of samples demonstrated significance, and successful isolations were associated with donor organs with increased β-cell volume. Thus, while we anticipate that further analyses will indicate a strong correlation between β-cell volume and isolation yield, the influence of other isolation variables is such that an expanded sample size is required.

Although the presence of islets within the pancreas is paramount, this fact alone does not guarantee their successful isolation. As we observed, enzyme type and lot are consistently identified as key variables in determining isolation success (^{18, 21, 23, 27, 29}). In fact, the issue of enzymatic activity is so critical to successful islet isolation that one report suggests that the selection between different enzyme preparations be based on general donor characteristics (²⁹). It is in keeping with this opinion that we propose the use of a different enzymatic preparation, Liberase CI, for young donors. Optimization of islet yield in this donor population is all the more important given that islets isolated from young donors seem to be the most functional (²³).

Because tissue quality is of critical importance for either whole pancreas or islet transplantation, methods to maintain or improve tissue preservation after organ harvest have engendered much attention, one such approach is the so-called TLM (³⁰). Although our experience with organs procured using this method has been limited, we can confirm that the TLM allows for successful islet isolation. While this technique of organ preservation now rivals UW as the standard approach to pancreatic preservation for the purposes of islet isolation (³⁴), the time and costs associated lead us to suggest that this technique only be applied for procurements in which the duration of cold ischemia may be excessively prolonged.

In this report, we analyzed variables to identify predictors of islet yield. Stepwise logistic regression provided a simple model for predicting isolation success based on donor sex and age and enzymatic digestion time. Analysis of β-cell volume suggests a correlation with islet yield, explaining the importance of donor variables that have been associated with β-cell volume (^{33, 39}). However, our results, and a review of the literature, suggest that enzyme type and lot are in fact valuable predictors of isolation success, likely more so than donor variables. Moreover, we have also observed, empirically, an increase in the consistency of recent lots of available enzymes, suggesting that one of the key factors dictating isolation success may, in fact, be less of an issue moving forward. Finally, the results of continuing retrospective analysis of this type has allowed our laboratory to be more selective with respect to organs received, leading to improved cost-effectiveness. This selectivity, coupled with tailoring of the isolation procedure to specific variables, has led to ever increasing isolation success. However, the fact remains that organs for clinical islet isolation are in short supply, while the predictive value of isolation variables is limited. Taken together, islet isolation should be attempted whenever a suitable organ becomes available.

ACKNOWLEDGMENTS

The authors thank Québec-Transplant for coordination of organ availability, all those involved with islet isolations, and Prof. James A. Hanley for help with statistical analysis.

REFERENCES