Hypertension, but not BMI, is predictive of increased pancreatic lipid content and islet dysfunction

Abstract

Pancreatic steatosis is thought to be a negative risk factor for pancreas transplantation outcomes. Despite considering donor body mass index (BMI) and the visualization of intercalated fat as indicators of donor pancreas lipid content, transplant surgeons do not use a quantitative method to directly measure steatosis when deciding to transplant a pancreas. In this study, we utilized non-diabetic human pancreata donated for research to measure the pancreatic and islet-specific lipid content in order to determine which clinical markers correlate best with lipid content. Interestingly, we found that BMI and age correlate with increased pancreatic lipid content (Panc-LC) in men, but not women. Our findings further suggest that total Panc-LC correlates with an increase in islet lipid content (Islet-LC) for both men and women. We noted that pancreata donated from individuals with a history of hypertension have increased Panc-LC independent of donor BMI or sex. Moreover, we identify hypertension as a risk factor for reduced islet function following islet isolation. Together, our findings emphasize differences in pancreas graft quality related to pancreatic and islet lipid content, which may not be predicted by assessing BMI alone, but may be influenced by a donor history of hypertension.

1. Introduction

In 2014, the global burden of diabetes was estimated at 422 million individuals, almost quadrupling from 1980.¹ In 2015, it was estimated that 711 million individuals were obese² and it is well accepted that obesity is a significant risk factor for diabetes.³ Injectable insulin and numerous oral medications are commonly used to treat diabetes;⁴ however, these treatments do not perfectly mimic or correct endogenous insulin secretion. Pancreas or islet transplantation remain the only clinically available treatments capable of replacing lost or dysfunctional beta cells, and restoring euglycemia without the need for exogenous insulin.⁵

Over the past 10 years, the number of annual pancreas transplants have decreased, and currently, only ~10% of all deceased donors provide a suitable quality pancreas for transplantation. Approximately 25% of pancreata recovered for intended transplant are discarded.⁶ These discarded organs represent potential life-saving treatments for thousands of recipients on the waiting list, but are not used due to a variety of criteria including donor age, BMI, and damage to the organ. Statistically, discard rates steadily rise as donor age and BMI increase, with discard of over 80% from donors with a BMI >35 kg/m² and 72% of donors above age 50.⁶ Though the reasons for organ discard are myriad, major concerns arise related to steatosis-associated inflammation and ischemia-reperfusion injury, leading to subsequent thrombosis and graft failure.^7–9 Pancreatic steatosis is the accumulation of fat specifically within the parenchyma of the organ, however the ability to determine whether a graft is suitable for transplantation and to qualify the degree of steatosis in pancreata prior to transplantation is highly subjective. Thus, assessment based on visualization, and unclear relationships between BMI and pancreatic steatosis may result in the unnecessary discard of otherwise healthy organs.⁸ To better utilize donor pancreata and reduce discard rates, an improved understanding of how donor selection criteria, such as age and BMI, as well as other donor parameters, actually correlate with steatosis of the pancreas would be beneficial. Furthermore, identifying better demographic correlates for pancreatic steatosis, and understanding how lipid infiltration into the pancreas impacts the endocrine function of the organ, may aid in these decisions.

To this end, several studies have attempted to characterize pancreatic steatosis in humans using histopathology^10,11 and in vivo imaging techniques, including magnetic resonance imaging (MRI),^12–20 magnetic resonance spectroscopy (MRS),^21–24 ultrasonography,^25–33 computerized tomography (CT),^34–36 and combinations of these.^37,38 Although MRI and MRS remain the most sensitive imaging options for estimating donor pancreas fat content in vivo, these techniques are time-consuming and impractical to incorporate into the donor workflow.⁸ Furthermore, ultrasonography does not have the ability to discern steatosis from fibrosis, which precludes its use for accurately characterizing pancreas fat.^8,39 Finally, CT imaging for pancreatic steatosis lacks sufficient specificity and sensitivity.^8,40

Given the variety of techniques and study populations, data on the potential clinical factors associated with pancreatic steatosis are conflicting. To complicate this, some studies also incorrectly include accumulated visceral fat around the exterior of the organ in their measurements, either due to technical difficulty to distinguish, or by ambiguous definition of steatosis. MRI, MRS, ultrasonography, and CT studies of live patients have correlated different clinical parameters with increased pancreas fat infiltration, including diabetes,^{16,21,23,24,30,32,36} BMI/obesity,^{12,14,16,18,21,24,25,27,29,31–35} heavy alcohol use,²⁵ metabolic syndrome,^28,31,33 increased age,^{14,15,24,26,29,32,35,37} hypertension,^26,29,33 visceral adipose tissue,^{12,14,18,26,28} race/ethnicity,^14,22 waist circumference,^12,18 male sex,^18,26,29 dyslipidemia,^{14,20,29,31,33} and hyperglycemia.^29,33 In particular, BMI has had mixed results regarding its relationship with pancreas fat quantified by imaging. Several studies have suggested that high BMI or obese subjects have higher pancreas fat content;^{12,14,21,24,25,31–33} however, other studies have reported no association between BMI and pancreas fat²³ and others have reported sex-specific findings in which increased BMI and pancreas fat have a stronger relationship in men.^18,20

It is also important to understand whether pancreatic steatosis impacts islet function. Unfortunately, few studies of pancreatic steatosis have simultaneously evaluated islet function, and other studies’ present findings that are inconsistent.^{12,14,16,20,22–24,37} In a study employing MRS and standard oral glucose tolerance tests, increased pancreas fat content as assessed by imaging was suggested to be associated with a decline in beta cell function in non-diabetic Caucasian males; the authors suggested it may correlate with the onset of type 2 diabetes.²³ In another study, pancreatic fat, measured by MRS, was found to be predictive of beta cell dysfunction, as assessed by insulin secretion and insulin sensitivity in a multiethnic sample of men and women.²² However, several other studies have reported no association between pancreatic fat and islet function.^14,20,24,37

Animal and cell culture studies have the ability to examine the effect of lipids on islet function on a more mechanistic level, albeit with less clinical relevance. Studies of minipigs and mice have shown that a high-fat diet promotes enlarged islets, beta cell damage through oxidative stress, insulin resistance, and increased pancreatic steatosis.^41–43 Mouse islets cultured in media with elevated fatty acid levels resulted in increased intracellular triglyceride storage, as evidenced by Oil Red O (ORO)-positive staining, coincident with morphological changes in the islets and reduced glucose-stimulated insulin secretion (GSIS) performance.⁴⁴ Moreover, a study of isolated human islets co-cultured with pancreatic adipocytes showed that adipocyte-secreted cytokines (IL-6, IL-8, and MCP-1) impair GSIS through inflammation of the islet milieu.⁴⁵

Given the contradictory findings regarding pancreatic fat content in human imaging studies, we sought to quantify human pancreatic lipid content (Panc-LC) using a protocol to extract lipids directly from donor tissue. Furthermore, because animal and cell culture studies have suggested that lipids disrupt islet function, and modern imaging studies have not yet explored islet-specific steatosis, we histologically quantified islet lipid content (Islet-LC). We correlated these data with islet stimulation indices (SI) following isolation to investigate how a variety of clinical donor parameters, correlate with islet function. We hypothesized that BMI is not an accurate predictor of Panc-LC and Islet-LC, and evaluated whether other clinical parameters correlate with increased pancreatic and islet steatosis.

2. Materials and Methods

2.1. Pancreas Procurement and Donor Demographics

Donated non-diabetic human pancreata (n=55, age 16–80 years) were procured for transplantation or research. Following organ recovery, the pancreas was allocated for research if deemed unfit for transplantation due to vascular damage, no suitable recipient, non-ideal age or BMI, and other circumstances. Pancreata were obtained through University of Wisconsin Organ and Tissue Donation services with consent obtained for research from next of kin. The use of human pancreata for research was approved by the University of Wisconsin-Madison Health Sciences Institutional Review Board. Organs were received within 24 hours of procurement and processed to remove extra-pancreatic adipose and vascular tissue.

Pancreatic donor risk index (PDRI) was calculated for all donors excluding the terms for pancreas preservation time and the adjustment for pancreas-after-kidney transplant from the equation.⁴⁶ Pancreatic lipid content was determined using the methodology below. Donor clinical parameters, including age, BMI, sex, ethnicity, HbA1c, amylase, lipase, creatinine, cause of death, cold ischemia time (CIT), and health history were excerpted from the United Network for Organ Sharing (UNOS) donor chart and recorded. Elements of our search included the examination of smoking, drug, and alcohol abuse as well as any well-documented history of chronic medical conditions including gastrointestinal disease (GID), hyperlipidemia (HLD), hypertension (HTN) and coronary artery disease (CAD). All donors were selected with no history or symptoms of pre-diabetes, T1DM, T2DM, or pancreatitis.

2.3. Lipid Extraction and Quantification

Tissue samples were collected from the central parenchyma of the neck of the pancreas, to avoid collecting any extra-pancreatic adipose tissue from the surface of the organ. Pancreatic lipid content (Panc-LC) was quantified using a version of Folch lipid extraction as previously described.⁴⁷ This modified Folch method was not significantly different for lipid quantification than the standard Folch Method, which uses 2:1 chloroform:methanol for lipid extraction (Fig. S1A,C).⁴⁸ Pure adipose tissue was used as a positive control, and contained >90% lipid by dry weight (Fig. S1B).

2.4. Histology and Image Quantification

Fresh pancreas samples were treated with 30% sucrose overnight, embedded in OCT, and stored at −80°C. For staining, 10 µm sections were cut and fixed in 3.7% formaldehyde for 1 hour. Sections were permeabilized with 0.5% Triton X-100/PBS for 5 minutes, and stained with insulin primary antibody (Sigma I2018, 1:10,000, 30 min), washed, and then secondary antibody (Abcam ab150117, 1:2,000, 30 min). Following rinses with 1x PBS, the ORO working solution was applied for 30 minutes. The sections were then stained with DAPI (ThermoFisher, D1306) in 1% BSA for 1 minute (adapted from Koopman et al.).⁴⁹ Samples were imaged using a Zeiss Axiovert 200M microscope and an Olympus BX51TF microscope. Image quantification for Islet-LC was performed in ImageJ by tracing the Ins⁺ area and measuring the ORO-positive area within each islet. Acinar lipid content quantification was performed by tracing the Ins-negative area surrounding the islets. Islet-LC was measured for 19–38 islets per donor. Acinar lipid content was measured on 7–15 images per donor.

2.5. Islet Isolation and GSIS

Human pancreata from non-diabetic donors were recovered for research following confirmed donation consent and consent for research use of organs. Islets were isolated using the previously described method (Hanson, et al.) and cultured in CMRL 1066 Supplemented (Corning) with 2.5% human serum albumin at 37°C in a 5% CO₂ incubator for 24 to 36 hours prior to GSIS.⁵⁰ The cold ischemia time (CIT) for all donors ranged from 2 hours to 12.5 hours, with a median CIT of 6 hours.

Islets were assayed in vitro for response to glucose stimulation according to a previously published protocol.⁵¹ Briefly, triplicate islet samples were incubated for 1 hour in media containing 2.8mM glucose and a sample of supernatant was collected. The same islets were then incubated for 1 hour in media containing 28mM glucose and a sample of supernatant was collected. The insulin concentration of supernatant samples was measured using a human insulin ELISA kit (Mercodia). The SI is calculated as the ratio of insulin concentrations from the 28mM glucose incubation supernatants to the corresponding 2.8 mM glucose incubation supernatants.

2.6. Statistical Analyses

Statistical analyses were performed using Prism 6 for Windows (GraphPad Software, Inc.). For pairwise comparisons, independent unpaired two-sample t-tests were performed except when population proportions were compared. In these instances, a Chi-square test of independence was performed. All correlations were assessed using Pearson’s Correlation and R² and p-values were reported. A p-value < 0.05 was considered significant, and Prism’s suggested significance classification scheme was followed (**p< 0.01) and (* p < 0.05).

3. Results

3.1. Pancreas donor demographics and graft quality

The donor demographics of 55 non-diabetic human pancreata recovered for Panc-LC analysis are described (Table 1). The median age was 53 years (range: 16–80), the median BMI was 26.3 (range: 16.9 – 48.8), 92.7% were Caucasian, 52.7% were men. The donors exhibited a wide range of Panc-LC, ranging from 4.7% to 70.2% with a median of 25.1%. Comparing men and women, only PDRI was significantly different; this is not surprising as the equation for PDRI adds additional value for female donors. Mean creatinine and HbA1c are within normal ranges, suggesting normal kidney function and glucose homeostasis in this cohort of non-diabetic donors.

Table 1. Correlation of pancreatic donor demographics and graft quality.

Graft quality is expressed by PDRI and pancreatic lipid content, which was determined using the Folch method. Values are reported with ± 1SD or the number of patients (N) for population proportion data.

	All Donors	Men	Women	P-value
N-value	55	29	26	- -
Age (Years)	49.09 ±15.24	44.97 ±14.24	53.69 ±15.27	0.065 (ns)
BMI (kg/m2)	27.28 ±6.04	27.38 ±5.02	27.17 ±7.10	0.727 (ns)
Race (% Caucasian)	92.73 (N= 51)	93.10 (N= 27)	92.31 (N= 24)	0.976 (ns)
CIT (hours)	11.68 ±6.85	12.84 ±7.75	10.43 ±5.61	0.142 (ns)
PDRI	2.22 ±1.30	1.73 ±0.96	2.76 ±1.43	0.007 **
% Smoker	61.82 (N= 34)	58.62 (N= 17)	65.38 (N= 17)	0.750 (ns)
% Heavy Alcohol Use	25.45 (N= 14)	20.69 (N= 6)	30.77 (N= 8)	0.459 (ns)
% HTN	43.64 (N= 24)	44.83 (N= 13)	34.62 (N= 9)	0.482 (ns)
% HTN (>5 years)	23.64 (N= 13)	27.59 (N= 8)	19.23 (N= 5)	0.525 (ns)
%CAD	5.45 (N= 3)	0.00 (N= 0)	11.54 (N= 3)	0.067 (ns)
%HLD	12.73 (N= 7)	10.34 (N= 3)	15.38 (N= 4)	0.601 (ns)
%GID	14.55 (N= 8)	17.24 (N= 5)	11.54 (N= 3)	0.580 (ns)
Death Cause (% Trauma)	23.64 (N= 13)	27.59 (N= 8)	19.23 (N= 5)	0.580 (ns)
% DBD	83.64 (N= 46)	79.31 (N= 23)	88.46 (N= 23)	0.711 (ns)
Creatinine (mg/dL)	1.09 ±0.76	1.04 ±0.50	1.15 ±0.98	0.521 (ns)
HbA1C	5.35 ±0.44	5.28 ±0.53	5.38 ±0.32	0.394 (ns)
Amylase (u/L) (n=48)	80.92 ±89.57	65.77 ±59.24	92.95 ±113.7	0.453 (ns)
Lipase (u/L) (n=48)	52.58 ±88.55	48.62 ±84.26	58.95 ±97.17	0.515 (ns)
Pancreatic Lipid Content (%)	27.84 ±14.39	28.48 ±15.38	27.18 ±13.56	0.707 (ns)

Abbreviations: CIT=Cold Ischemia Time, HTN=Hypertension, CAD=Coronary Artery Disease, HLD=Hyperlipidemia, GID=Gastrointestinal Disease, PDRI=pancreatic donor risk index, DBD=Donor After Brain Death.

ns: not significant (p-value>0.05),

^**p-value<0.01

3.2. Pancreas lipid content is correlated with BMI and age in men but not women

We quantified Panc-LC using lipid extraction to evaluate trends with donor BMI and age, two well-established donor factors correlating with pancreas transplant outcomes. Upon examination of population-level trends for all donors, there was no significant correlation between BMI and age (p=0.197, data not shown), which indicates that these variables should be investigated separately for their relationship with Panc-LC. Among all donors, there was a significant correlation between age and Panc-LC (p=0.037) (Fig. 1A), and a significant correlation between BMI and Panc-LC for all donors (p=0.008) (Fig. 1B). However, when we separated donors according to sex, we observed that age versus Panc-LC (p=0.02), and BMI versus Panc-LC (p=0.009) were both significantly correlated in men (Fig. 2A–B) but were not significantly correlated in women (p=0.42 and p=0.20, respectively) (Fig. 2 C–D).

Figure 1. Pancreas lipid content is correlated with BMI and age for all donors.

A-B) Correlations including BMI, Age, and Panc-LC in all pancreatic donors (men and women combined) indicate a significant correlation for BMI and Panc-LC. The 95% confidence interval is shaded around the line of best fit. The p-values were determined using a Pearson correlation. Donor BMI and Age are extracted from the donor chart in the UNOS system. Error bars indicate ± 1 SD among technical replicates for Panc-LC. N=55 donor pancreata.

Figure 2. Pancreas lipid content is correlated with BMI and age in men only.

A-D) Correlations including BMI, Age, and Panc-LC in sex-stratified donor populations as labeled above indicate significant correlations for men. The 95% confidence interval is shaded around the line of best fit. The p-values were determined using a Pearson correlation. Error bars indicate ± 1 SD among technical replicates for Panc-LC. Donor BMI and Age are extracted from the UNOS system. N=29 men; N=26 women donor pancreata.

3.3. Pancreas lipid content is correlated with increased islet lipid content

To assess whether Panc-LC may have implications for pancreas quality and endocrine function, we assessed islet quality within a subset of donor organs. When pancreas sections were co-stained for insulin, to identify islets, and ORO, to stain lipids, there were distinct differences in the lipid localization among the islet and acinar compartments. Specifically, there were high concentrations of lipid droplets found within the islets of high lipid content pancreata, but not in lower Panc-LC organs (Fig. 3). To further investigate how lipid distribution varied among pancreas donors, we quantified the ORO staining content within the islet and acinar compartments. Islet-LC was found to strongly correlate with total Panc-LC (p=0.0002) (Fig. 4A) and this was the case regardless of sex (Fig. S2A–B). Furthermore, the ratio of lipids in the islet compared to acinar regions (Isl/Ac) also increased with total Panc-LC (p<0.0001) (Fig. 4B), regardless of sex (Fig. S2C–D). This indicates that high lipid content pancreata contain islets that have significant lipid deposits compared to lower lipid content pancreata. This is visibly distinguishable upon direct histological comparison (Fig. 3).

Figure 3. Islets from steatotic donors (high lipid pancreata) are enriched for lipids within islets.

A-F) Sections from a high lipid pancreas (70.2% Panc-LC) and an average lipid pancreas (26.0% Panc-LC) were embedded in OCT were stained for insulin (green), ORO (red), and DAPI (blue) and imaged at 20X magnification. A) Representative merged image showing ORO-positive tissue overlapping with insulin-positive tissue in a pancreas with high Panc-LC. B) Representative merged image depicting ORO-positive tissue separate from insulin-positive tissue (localized to the acinar) in a pancreas with average Panc-LC. C-D) Insulin channel showing the localization of islets in both pancreata. E-F) ORO channel showing the localization of lipids in both pancreata. Scale bar in B = 100μm.

Figure 4. Panc-LC is significantly correlated with Islet-LC and the Isl/Ac lipid ratio.

A-B) Correlations including Panc-LC, Islet-LC, and an Isl/Ac lipid ratio for a subset of donors. Islet-LC tends to appear in high Panc-LC pancreata. The 95% confidence interval is shaded around the line of best fit, and the p-values were determined using a Pearson correlation. Error bars represent ± 1SD. N=35 donors.

3.4. Donor hypertension is predictive of increased pancreas and islet lipid content and islet dysfunction

A population of donors with a moderate BMI range were found to have a broad range for total Panc-LC, with a subpopulation that contained higher-than-average lipid content, and a different subpopulation that exhibited below-average lipid content (Fig. 5A). To identify additional donor metrics that may correlate with this striking difference, we examined the medical histories available through the UNOS system. We failed to uncover any trends based on smoking, alcohol or drug abuse, or GID, among the two groups (data not shown). In contrast, within the two subsets of donors, we found that 8 of 9 donors in the above average Panc-LC group had a history of HTN, while no donors in the below average group had a history of HTN. There was also an increased presence of CAD and HLD in the high Panc-LC group, but the overall prevalence of these diseases in our donor population was too low to further investigate.

Figure 5. Donor history of hypertension is predictive of increased pancreas and islet lipid content independent from changes in BMI.

A) Correlation plots for Panc-LC and BMI indicate two distinct populations of donors with differences in Panc-LC independent of BMI. Donors were divided into groups with no history of HTN (N=27) and history of HTN (<5 yr, N=8) (5+ yr, N=14) B) Average BMI and C) Panc-LC for the three groups. D) Average Islet-LC in no HTN (N=15) and HTN (<5 yr, N=8) (5+ yr, N=11) groups. Error bars indicate ± 1SD, (** p<0.01, *** p<0.001, as compared to the no the HTN groups).

To examine the correlation of HTN with Panc-LC, donors were categorized into three groups: donors with no history of HTN (N=27), donors with <5 year (N=8) and 5+ year histories of HTN (N=14) (Fig. 5B–D). Donors with unclear history of HTN were excluded to prevent confounding effects. When comparing the three groups, we found that the average BMI was indistinguishable (ANOVA, p=0.17) (Fig. 5B), but there were significant differences with regard to Panc-LC (Fig. 5C) and Islet-LC (Fig. 5D) for both HTN groups, compared to the non-HTN group. We also observed this trend for both men and women when analyzed separately (Fig. S3).

Using this information, we performed a retrospective study of isolated islets from 95 pancreata donated for research purposes, assessing the stimulation index (SI) during static GSIS. These organs were donated for islet isolation and distributed through the Integrated Islet Distribution Program (IIDP), and therefore were by necessity different donors from the pool used to study Panc-LC. Donor history of HTN was collected using UNOS records, and donors were categorized into three groups based on this history: no HTN, <5 years HTN, and 5+ years of HTN history. Again, donors with unclear/unknown records of HTN were excluded from the dataset. The results indicate that donors with a 5-year or longer documented history of HTN had a significantly lower average SI (SI=1.90, p=0.016), compared to donors with no history of HTN (SI=3.14). The SI in islets from the group of donors with a less than 5 year history of HTN was lower than the control group, but was not significantly different (SI=2.86, p=0.529) (Fig. 6C). Furthermore, the SI showed no correlation with donor BMI (p=0.670) (Fig. 6A), or the CIT of the organ before islet isolation (p=0.61) (data not shown). The islet SI was also not significantly different between discrete groups of donors with BMI <30 and obese donors with BMI >35 (p=0.811) (Fig. 6B).

Figure 6. Donor history of hypertension, but not BMI, is predictive of islet dysfunction.

A) Among all islet donors, there is no significant correlation between islet function (SI) and BMI (N=95). B) Donors with BMI <30 (N=29) and BMI >35 (N=35) have indistinguishable islet function by GSIS. C) Donors were divided into groups with no history of HTN (N=58) and history of HTN (<5 yr, N=20) (5+ yr, N=14). Donors with 5+ years of documented hypertension had significantly lower SI following islet isolation. Error bars indicate ± 1SD, (*p<0.05, as compared to the no HTN group).

4. Discussion

Steatosis is considered a critical factor for determining the suitability of donor pancreata for transplantation, but many pancreata from high BMI donors can be successfully transplanted. Given the need to be good stewards of precious donated organ resources and the rising prevalence of diabetes worldwide, it is imperative that available pancreata are maximally utilized. Understanding how the measured lipid content correlates with other associated donor parameters could improve our ability to discern lesser quality pancreata using non-invasive clinical parameters to determine whether the pancreas should be transplanted. To our knowledge, we are the first to directly quantify the lipid content of human pancreata, in order to better understand the clinical correlates and functional impact of lipid accumulation in deceased donor pancreata.

Our direct method of measuring Panc-LC may have advantages over the variety of imaging studies that have resulted in inconsistent findings related to pancreas steatosis. Imaging may not be the most translatable technique for organ donation, but organ biopsies are routinely used to assess donated livers and kidneys prior to transplant. Because steatosis is also a liver transplant concern, crudely estimating the degree of hepatic steatosis by biopsy has become common practice. We collected biopsies of pancreata for ORO staining (Fig. S4) and lipid extraction (Fig. S5) to demonstrate that clinical biopsies may be useful in pre-transplant donor selection. We found that the lipid content values obtained from biopsies were consistent with those obtained from larger pieces of tissue (Fig. S5), but the lipid extraction takes several days and is therefore not a translatable method to apply to organ donation and transplantation practices. On the other hand, pancreas graft biopsies could potentially be obtained at the time of recovery and immediately processed to estimate the degree of steatosis of the acinar and islet regions.^52–54 Although performing pancreas graft biopsies is not common, several studies indicate they are associated with very low complication rates.^55–57

Previous studies have found that donor BMI has minimal impact on pancreas transplant outcomes, whereas others have suggested that high BMI is associated with increased risk of technical failure. Humar et al. found that BMI may have a stronger impact on the technical success of the transplant, and minimal impact on the longer term glycemic outcomes following a successful surgery.⁵⁸ Alhamad et al. similarly found that mildly obese donors (BMI=30–35 kg/m²) had similar allograft failure rates as pancreata from donors with a BMI from 20–30 kg/m², but donors above 35 kg/m² were associated with increased graft failure.⁵⁹ Finger et al. also showed that if a donor possesses one risk factor for technical failure, such as high BMI alone, grafts experience nearly equal technical successes to ideal grafts without any risk factors and normal BMI.⁶⁰ On the other hand, Axelrod et al. showed that a high BMI, older age and other factors contribute to lower 1 year graft survival rates.⁴⁶

Our current data demonstrate that BMI and Panc-LC were found to be significantly correlated for men, but not women. This finding could reflect sex-related differences in systemic fat distribution, as has been postulated by a variety of studies.⁶¹ Such an observation could be useful clinically to broaden the BMI threshold for transplanting female pancreata while restricting the threshold for male pancreata. While our study lacks transplant outcomes data, and the aforementioned recommendation is tentative, it does align with the current pre-determined pancreas acceptance policy at the University of Wisconsin. These sex-specific differences may be one reason that various population-level studies including both men and women have only found modest trends relating Panc-LC with other variables. Taken together our data suggest that BMI and age play a sex-specific role in pancreatic steatosis, which has been reported in only a few prior imaging studies.^18,20,26,29 Additionally, we found a significant correlation between Panc-LC and Islet-LC, which has not previously been reported. Moreover, Panc-LC is significantly correlated with the ratio of islet to acinar lipids (Isl/Ac), which suggests that as the pancreas becomes progressively steatotic, islets become intracellularly enriched with lipids. This is different than what was previously observed in a mouse study, where both islets and acinar cells accumulated lipids equally as the pancreas became steatotic through exposure to a high fat diet.⁶²

We found that donors with a history of HTN had increased Panc-LC and Islet-LC, regardless of sex. These data indicate that a history of HTN independently contributes to an increase in Panc-LC and Islet-LC. We also made an effort to correlate HLD and CAD as other potential markers for increased Panc-LC, but our tissue bank had too few donors with these conditions to make firm conclusions. Interestingly, we also found that in many long-duration HTN donors with high Panc-LC, lipids were concentrated within the islets. The differential accumulation of fat in islets versus acinar tissue of the human pancreas has important implications in the field of pancreas and islet transplantation.

Our study also suggests that HTN is associated with decreased in vitro functional performance of isolated islets. It would be speculative to suggest that HTN would be associated with worse glycemic control in the long-term following pancreas transplant. In support of this hypothesis, however, a recent study analyzing a database of ~25,000 pancreas transplants found that a donor history of HTN had an independent, negative impact on overall recipient and graft survival.⁶³ Borrowing from kidney transplant research, the authors postulated that the negative effect of HTN on pancreas graft survival could occur through vascular damage, which results in poor organ blood supply.⁶³ However, based on our findings, we additionally propose that HTN could contribute to inferior long-term pancreas graft glycemic outcomes. Because HTN is associated with an increase in Panc-LC and Islet-LC independently from sex, it may represent a more useful marker than BMI for pancreas and islet steatosis, and possibly graft function post-transplantation. Further research is needed before donor HTN could be used as strong screening criterion for pancreas donation. However, these data do suggest that pancreata from donors with higher BMI and no history of HTN may be significantly less likely to be steatotic, and could be used for transplant.

We speculate that high Panc-LC and longstanding HTN may be early markers of metabolic syndrome. One mechanism which may connect these conditions with mild islet dysfunction could be through hypoxia-induced lipid accumulation, increased lipid peroxidation, or lipotoxicity. Hypertension induces vasoconstriction of small vessels and reduced microvascular density, which impairs tissue perfusion.⁶⁴ In cultured cells and many organs, including the liver and islets, hypoxia and hypoxia-inducible factors have been associated with altered lipid metabolism and intracellular lipid droplet accumulation.^65–68 Several animal and cell culture studies have identified that intracellular lipid storage within islets causes a marked reduction in islet function, through inhibition of insulin production and secretion, as well as cellular toxicity.^{41–45,69,70} Consequently, HTN may contribute to chronic tissue hypoxia resulting in lipid accumulation and islet dysfunction. Thus, HTN may be an early marker of metabolic syndrome and insulin resistance.

Our study establishes a correlation between HTN and pancreas and islet lipid accumulation, but questions still remain. A better understanding of how a history of HTN in the donor affects pancreas transplant outcomes would be beneficial. Future studies could utilize pancreas transplant outcomes data to evaluate the effects of donor HTN history on recipient glycemic control as well as pancreas graft and patient survival following transplantation. Furthermore, it is possible that the pancreas and islet conditions observed in our study are the result of a treatment for HTN, rather than HTN itself. However, donor records in the current study also do not contain sufficient data to assess the treatment of HTN, or compliance with treatment. In support of this hypothesis, previous studies have concluded that certain HTN treatments, such as beta-blockers or thiazide diuretics, may have adverse glycemic effects,⁷¹ but we are unable to assess this within our data set. It is also possible that other important disease conditions co-existed with HTN in this donor population that more directly contributed to increased Panc-LC and Islet-LC than did HTN.

With respect to islet function, we raise the question of whether or not islet lipid accumulation is retained following islet isolation and culture, and whether this may also affect islet transplant outcomes, as well as in vitro islet studies. Further studies are necessary to answer these questions. Furthermore, we recommend additional study before these findings are firmly integrated into clinical decision making and donor selection.

In conclusion, by directly quantifying the pancreatic lipid content, our study demonstrates that men and women may have differential lipid accumulation profiles in the pancreas based on age and BMI. We found that men are at higher risk to have increased Panc-LC with increased age and BMI, but this trend was not found for women. For both sexes, very high Panc-LC was associated with an increased lipid enrichment specifically in islets, which could be an indication of poor islet health and function and increased potential for damage after an ischemia-reperfusion injury. Both total Panc-LC and Islet-LC were found to be significantly increased in a population of donors with histories of HTN. Islets isolated from donors with a history of HTN also were shown to have reduced GSIS function. Moving forward, further investigation into improving the identification of steatotic pancreata prior to transplantation is necessary to better utilize available donor organs. It is also important to understand how islet function may be affected in donors and patients with high Panc-LC and Islet-LC and whether these factors impact pancreas transplant outcomes.

Abbreviations:

BMI

Body Mass Index

Panc-LC

Pancreas Lipid Content

Islet-LC

Islet Lipid Content

MRI

Magnetic Resonance Imaging

MRS

Magnetic Resonance Spectroscopy

CT

Computerized Tomography

GSIS

Glucose Stimulated Insulin Secretion

SI

Stimulation Index

IIDP

Integrated Islet Distribution Program

ORO

Oil Red O

PDRI

Pancreatic Risk Donor Index

UNOS

United Network for Organ Sharing

CIT

Cold Ischemia Time

GID

Gastrointestinal Disease

HLD

Hyperlipidemia

HTN

Hypertension

CAD

Coronary Artery Disease