Improved stabilisation of human pancreas biopsies: Comparative evaluation of RNA preservation methods

Abstract

BACKGROUND

The pancreas is particularly vulnerable to rapid post-retrieval degradation due to its high endogenous RNase and enzymatic activity. This presents a major challenge in both pancreas research and clinical transplantation for accurate transcriptomic analyses. Accurate molecular profiling is increasingly important for evaluating graft quality and characterising injury. Although RNA preservation strategies have been reported in animal tissue studies, their comparative performance in human pancreatic tissue remains under-reported.

AIM

To evaluate biopsy preservation methods to validate an optimal approach capable of stabilising human-pancreas biopsies for high-quality gene expression analysis.

METHODS

Four clinically declined human pancreata were regionally sampled (36 biopsies per pancreas). Tissue was preserved using either: (1) Immediate RNA isolation; (2) Snap freezing; (3) RNAlater submersion; or (4) RNAlater injection. Immediate samples were extracted the same day with snap frozen and RNAlater samples being snap frozen and freeze-thawed prior to extraction using a spin-column protocol. RNA concentration and purity were assessed by Nanodrop and RNA integrity number (RIN) generated using a TapeStation. Statistical analyses were conducted using R-Studio.

RESULTS

All samples bar one yielded RNA of acceptable concentration (25-500 ng/μL) and purity (260/280 ~2.00, 260/230 2.00-2.2). RIN values varied significantly. Both RNAlater injection (7.1 ± 1.10; adjusted P value = 0.0089) and RNAlater submersion (7.1 ± 1.13; adjusted P value = 0.0058) produced significantly higher RIN scores compared to snap freezing (3.7 ± 1.21). RNAlater preserved samples exceeded the minimum RIN threshold for downstream transcriptomic analysis (RIN ≥ 7.0). No significant difference was observed between RNAlater techniques or immediate isolation (5.0 ± 1.18).

CONCLUSION

RNAlater based preservation by submersion or injection provided superior stabilisation of human pancreatic RNA compared with conventional snap freezing. RNAlater preserves the transcriptome at the moment of tissue acquisition, minimising degradation post biopsy. Despite the wide adoption of RNAlater usage, this study provides confirmation that RNAlater reliably supports high-quality RNA extraction from biopsies filling the current gap in pancreas-specific evidence. Implementation of this method may enable more accurate evaluation of graft injury, improve biomarker development, and support high-quality biobanking of human pancreas biopsies for future studies.

Key Words: Pancreas; RNA; RNA integrity numbers value; RNA isolation; Human biopsy; Organ transplantation; Molecular biology

Core Tip: Obtaining high-quality RNA from human pancreatic tissue remains a persistent challenge in transplant research due to its high endogenous RNase activity and unavoidable pre-analytical ischemia. Using 144 biopsies from four deceased donor pancreata, we systematically evaluated tissue preservation strategies. Our findings demonstrate that both tissue submersion and direct injection of RNAlater yield significantly superior RNA integrity compared with conventional methods. This study provides specific human evidence for a simple and scalable strategy to ensure reliable biobanking and downstream molecular analysis of human pancreatic tissue.

INTRODUCTION

Pancreatic tissue poses particular challenges for molecular studies because of its exceptionally high content of enzymes including ribonucleases, proteases and deoxyribonucleases[1-7]. Enzymes are released rapidly after pancreatic explant and lead to swift autolytic degradation[1]. Even brief periods between retrieval and stabilisation can therefore be highly detrimental, especially in human transplant research where organs or biopsies often arrive in the laboratory after an uncontrollable period of ischaemia. During this interval, ongoing enzymatic activity can markedly reduce RNA concentration and compromise integrity, limiting the reliability of downstream applications such as reverse transcription quantitative polymerase chain reaction and RNA sequencing. Consequently, careful optimisation of pre-analytical handling and preservation protocols is essential if RNA extracted from pancreatic biopsies is to represent the true transcriptional state of the tissue.

RNA integrity is commonly assessed using the RNA integrity number (RIN), with higher values indicating better preservation of rRNA bands and by extension, overall transcript stability. While many human tissues yield RIN values compatible with high-throughput transcriptomic analyses using standard snap-freezing protocols, pancreatic samples often exhibit comparatively low RIN scores, reflecting inescapable degradation unless dedicated stabilisation strategies are employed[5].

Chemical RNA stabilisation reagents, such as RNAlater, have been widely adopted as an alternative to snap freezing alone[8]. They permeate tissue and denature RNases thereby inhibiting further degradation after collection. Griffin et al[5] systematically addressed this problem by developing and evaluating a simplified method for obtaining high-quality RNA from rat and porcine pancreas. They compared four approaches: Immediate RNA isolation, immersion of tissue in RNAlater, snap freezing and injection distention of the tissue with RNAlater. In their study, injection with RNAlater produced RIN values that were comparable to immediate processing and clearly superior to both immersion and snap freezing. The present study aims to systematically assess these four preservation strategies in human pancreas tissue, using quantitative measures of RNA integrity to assess whether the benefit observed in animal tissue transfers to human donor organs.

MATERIALS AND METHODS

Organ acquisition

Human donor pancreata that were declined for transplant and with appropriate research consent were utilised in this study, Research Ethics Committee, approval No. 16/NE/0230. Accepted pancreata were procured and packaged as if for clinical transplant and immediately transported from donor hospital to laboratory by certified medical transport. Upon arrival organs were dissected within our cold room at 4 °C, with spleen, duodenum, surrounding fat, and vessels removed and then divided into head, body and tail regions.

Tissue preservation

Three biopsies from each region were collected for each of the 4 preservation methods being assessed: Immediate isolation, snap frozen, injected with RNAlater (Invitrogen™, United Kingdom) and submersed in RNAlater (Figure 1). For immediate extraction, biopsies were dissected into approximately 1 cm³ pieces and placed in 4 °C University of Wisconsin preservation solution (Bridge to Life Ltd, United States) within sterile cryovials. Samples were transported between our laboratories on ice and total RNA was subsequently extracted from the entire tissue sample according to the protocol described below. Snap frozen, biopsies were dissected into approximately 1 cm³, transferred into cryovials and immersed in an isopentane-dry ice slurry to achieve rapid freezing. These samples were then stored in -80 °C until extraction.

Figure 1 Schematic overview of the RNA presentation and quality assessment workflow for human pancreatic tissue biopsies. Generated using BioRender and manually refined by the authors. Donor pancreata (n = 4) were dissected into head, body and tail regions with three biopsies per region allocated to four preservation methods: Immediate RNA isolation, snap freezing in an isopentane-dry ice slurry, submersion in RNAlater, or injection with RNAlater (144 biopsies). Both RNAlater conditions were incubated at room temperature for 15 minutes before snap freezing and storage at -80 °C. All samples underwent RNA extraction using the RNeasy Mini Kit (Qiagen) with bead-based homogenisation. RNA concentration and purity were assessed by NanoDrop and integrity by Agilent 4200 TapeStation following dilution to 250 ng/μL.

RNAlater samples were either submersed or injected and subsequently submersed. Submersed biopsies were cut into 1 cm long thin slices, and two slices were placed in each cryovial containing 1.2 mL RNAlater. For the injection protocol, approximately 1 cm³ tissue samples were serially injected with a total of 1 mL RNAlater using a 16 gauge needle with injections distributed across the specimen to maximise intraparenchymal coverage and tissue distension. Each injected sample was then transferred to a cryovial containing 1.2 mL RNAlater. Both were incubated at room temperature for 15 minutes to allow diffusion of the preservative into the tissue, then snap frozen and stored at -80 °C.

RNA extraction

Samples were dissected in a sterile, RNase-free petri-dish using a sterile scalpel. Fresh tissue was dissected over crushed ice and stored samples over dry ice. 25-35 mg of tissue was weighed for homogenisation and placed in 350 μL of Trizol (Invitrogen™, United Kingdom). Homogenisation was performed at 20 Hz for 2 minutes using a bead-based homogeniser (Qiagen, TissueLyser II) and rotated before a further 2 minutes of homogenisation. RNA was extracted from tissue using the Qiagen RNeasy mini kit (Qiagen, Hilden, Germany), by spin column technique as per manufacturers protocol. RNA concentration was measured using a NanoDrop One (Thermo Fisher Scientific) and RIN assessed by Agilent 4200 TapeStation (Agilent Technologies, United States).

Statistical analysis

Statistical analysis and visualisation of data was performed in R version 4.5.1 (2025-06-13). RIN values were averaged for all samples per donor and preservation method to account for biological replicates. Data distribution of preservation methods was assessed using Shapiro-Wilk test for normality and Levene’s test for variance. To assess overall differences between RIN and preservation methods a one-way analysis of variance followed by Tukey’s Honestly Significance Difference test was applied (^aP < 0.05, ^bP < 0.01).

RESULTS

Donor demographics

To assess RNA preservation methods of pancreatic tissue, four human donor pancreata declined for transplantation were included in this study. Donor characteristics are summarised in Table 1. Donors had a median age of 32 years and a median Body Mass Index of 22.25 kg/m² (interquartile range: 20.95-26.3). The cohort comprised two male and two female donors. Causes of death were hypoxic brain damage (n = 2) and intracranial haemorrhage (n = 2). Three donors were donation after circulatory death (DCD) and one was donation after brain death. Median cold ischaemia time (CIT) was 14.65 hours (interquartile range: 10.09-16.87 hours). One donor (P003) underwent normothermic regional perfusion (NRP) prior to donation. A total of 144 biopsies were collected across four preservation methods (36 biopsies per method), with three biopsies taken from each of the head, body and tail regions per donor per method (Figure 1).

Table 1 Demographic donor data.

Donor	Age	BMI (kg/m2)	Cause of death	Donation	Sex	CIT (hour)	NRP
1	21	23.4	Hypoxic brain damage	DCD	Female	15.32	No
2	20	21.1	Hypoxic brain damage	DBD	Male	13.98	No
3	68	20.8	Intracranial haemorrhage	DCD	Female	6.2	Yes
4	43	29.2	Intracranial-unclassified	DCD	Male	18.42	No

BMI: Body mass index; CIT: Cold ischemia time; NRP: Normothermic regional perfusion; DCD: Donation after circulatory death; DBD: Donation after brain death.

RNA yield and purity

RNA was successfully extracted from 143 of 144 biopsies. One immediate sample failed to meet the minimum concentration threshold (donor 2, tail region 1) and was omitted from analysis. All remaining samples yielded RNA within the acceptable concentration range of 25-500 ng/μL and demonstrated satisfactory purity with A260/280 ratios of approximately 2.00 and A260/230 ratios within the range of 2.00-2.20, indicating minimal protein and organic solvent contamination. There were no significant differences in RNA concentration between preservation methods (Figure 2). However, when examining results by region there was no significant difference between RIN (Figure 3A) but, there was a significantly higher yield of RNA isolated from the head compared to the tail region (Figure 3B).

Figure 2 Effect of tissue preservation method on RNA yield from human pancreatic tissue. Bar chart showing mean ± SD RNA concentration (ng/mg tissue) per region per donor (n = 4) of extracted from human pancreatic tissue preserved by four methods: Immediate processing, Snap freezing, Submerged and Injected with RNAlater. One-way analysis of variance applied to test significance. ANOVA: Analysis of variance.

Figure 3 Regional variation in RNA yield and integrity across the human pancreas. ^aAdjusted P value < 0.05. Bar plots showing the mean ± SD of the yield (ng/mg tissue) and integrity (RNA integrity number score) of RNA extracted from the head, body and trail regions from across all donors (n = 4) and preservation methods. One-way analysis of variance with Tukey’s post-hoc used to assess statistical significance. A: RNA integrity number across the pancreatic head, body and tail regions; B: RNA yield (ng/mg) across pancreatic head, body and tail regions.

RIN results

RIN values were averaged for each donor within each preservation method prior to statistical analysis, resulting in an effective sample size of n = 4 per group with 3 degrees of freedom for between-group comparisons. This approach accounted for the non-independence of biopsies obtained from the same donor with the same preservation technique. Clear differences in RNA quality were observed between preservation approaches (Figure 4). Immediate extraction displayed moderate RNA integrity, with discernible but partially degraded ribosomal RNA peaks. In contrast, snap frozen samples exhibited pronounced RNA degradation, characterised by reduced 18S and 28S ribosomal peaks and increased baseline signal. Samples preserved in RNAlater by injection and submersion demonstrated the highest RNA integrity, with well-defined 18S and 28S ribosomal RNA peaks and minimal baseline degradation.

Figure 4 RNA integrity profiles of human pancreatic tissue across four preservation methods. The X-axis shows RNA fragment size (nucleotides) and the Y-axis shows normalised fluorescence intensity. Vertical markers left to right denote the lower internal size standard and the 18S and 28S ribosomal RNA subunit peaks. RNA integrity number (RIN) scale: 1 = fully degraded, 10 = fully intact. Representative electropherograms and traces displaying RNA integrity profiles of preserved human pancreatic tissue. A: Immediate processing (RIN = 5.0); B: Snap freezing (RIN = 2.8); C: Submersion in RNAlater (RIN = 7.8); D: RNAlater injection (RIN = 8.0). RIN^e: RNA integrity number^e; nt: Nucleotides.

Comparison of RIN by preservation method had a significant effect on RIN values (P = 0.003; Figure 5). Snap frozen samples had significantly lower RIN values compared with both the Submerged [adjusted P value (P_adj) = 0.0058] and the Injected methodology (P_adj = 0.0089). Although no other pairwise comparisons reached statistical significance, trends towards differences were observed between the immediate isolation and RNAlater methods (P_adj = 0.079 and P_adj = 0.119). While these findings did not meet the threshold for statistical significance in the current study, the observed variation may warrant further investigation in larger cohort with greater statistical power. There was also no significant difference observed between Submerged and Injected methods (P_adj = 0.995).

Figure 5 RNA integrity number of human pancreatic tissue across tissue preservation methodologies. ^bAdjusted P value < 0.01. Box plot showing the mean distribution of RNA integrity number scores from human pancreatic tissue (n = 4 biologically independent donors) preserved by four methods: Immediate processing, Snap freezing, Submerged storage in RNAlater, and RNAlater injection. One-way analysis of variance and Tukey’s post hoc used to assess statistical significance. RIN: RNA integrity number.

DISCUSSION

This study systematically compared four different RNA preservation strategies to human pancreatic biopsies obtained from deceased donor organs. This setting presents major challenges for molecular analysis of the pancreas because donor physiology, peri and postmortem affects during organ procurement can all compromise RNA quality. Both RNAlater based preservation methods produced significantly higher RNA integrity than snap freezing without any chemical stabilisation, with the RNAlater methods in our cohort being statistically indistinguishable from one another. These results indicate that early chemical stabilisation ensures transcript quality at the point of procurement.

The poor RNA integrity from snap-frozen biopsies is well explained by digestive gland nature of the pancreas. The exocrine pancreas contains substantially higher RNase content, estimated to be approximately 180000-fold greater than that of the liver[7]. This enzymatic burden means that even brief exposure to un-stabilised conditions is sufficient to expose the tissue to autolytic degradation. What is more notable in this study is the underperformance of biopsies processed on the same day of tissue arrival to the lab. In previous animal studies, Griffin et al[5] reported immediate isolation to be the equivalent to RNAlater injection. While we achieved no significant difference in our results the average RIN was notably reduced. A critical distinction between the two studies is the ischaemic context of the tissue at the point of biopsy. Griffin et al[5] biopsies were processed within 10 minutes of euthanasia, minimising any prefixation cellular injury. In the present study donor organs arrived after substantial CIT introducing a prolonged period of subcellular injury prior to RNA extraction.

There are contrasting reports on the effect of ischemia timings on the quality of RNA from the pancreas. The Genotype-Tissue Expression Consortium[9] identified a drop in RNA quality with longer ischemia times across multiple tissue types, while Philips et al[10] did not find pancreas transport times to be significant in their multivariable model[9,10]. At the cellular level, the impact of CIT on human donor pancreas has been characterised in detail by Kattner et al[11] and Dyson et al[12] who developed and validated an electron microscopy scoring system for acinar and endocrine stress. They reported, even at relatively short CIT intervals, ultrastructural changes including mitochondrial swelling and endoplasmic reticulum dilation. These ultrastructural changes in pancreas cell organelles highlight the compromised state of membranes and organellar stress in which RNases become free within cells. As our present cohort was insufficiently powered to formally test comparison between RIN and CIT future larger studies will enable the assessment between preservation groups.

Donor heterogeneity within our cohort adds a further layer of complexity. Age spanned a wide range (20-68 years), whereby age-related changes in pancreatic composition, including increased adipose infiltration and acinar atrophy may independently influence tissue RNA yield and integrity in ways that are difficult to disentangle from procurement-related variables[13]. Further, our cohort comprised three DCD and one donation after brain death with cause of death being identified as a significant predictor of pancreatic RNA quality[10]. DCD procurement introduces an unavoidable warm ischemic interval with the duration well established to negatively impact the quality and viability of donor pancreas[14-17]. One donor (donor 3) also underwent NRP prior to procurement, which adds a distinct and under characterised source of variability. While literature on the effect of the NRP remains limited, with regards to pancreas transplant outcome, in our study it introduces a perfusion related biological confounder that cannot be explored with a single donor.

The RIN values obtained with our thin-slice submersion protocol exceeded those previously reported for RNAlater preserved samples. Griffin et al[5] demonstrated that bulk immersion in RNAlater produced degraded RNA in piglet pancreas with reduced 18S and 28S ribosomal bands comparable to snap frozen tissue. Our protocol decreased the diffusion distances for RNase inhibition during submersion replicating the intended advantage of needle-based techniques without the histological disruption associated with RNAlater perfusion[5]. Additionally, Philips et al[10] identified the prioritisation of RNA vial preservation over tissue block preparation as the single most important controlled variable for RNA quality, a practice we employ within our sampling protocol. However, submersion requires no specialist equipment, does not risk histological disruption and is more readily scalable within standard procurement logistics. Injection may offer faster intraparenchymal penetration, but requires a needle and carries a small risk of architectural disruption as previously reported[5].

Analysis across pancreatic regions revealed no significant difference in RNA integrity between the head, body and tail. This is in contrast to Philips et al[10], who observed significantly lower quality RNA in body and tail regions compared to the head. The absence of a regional difference in our study could be due to the smaller sample size or alternative tissue stabilisation method. However, the pancreatic head in our study did yield significantly greater total RNA per milligram of tissue than the tail (P_adj = 0.027). A finding that is biologically coherent as the head is disproportionately compromised of RNA-rich acinar cells, while the tail contains a higher proportion of endocrine islets per unit mass[18].

To our knowledge this is the first study to systematically compare these four RNA preservation methods in deceased donor pancreas tissue. The use of four donor pancreata while modest in number permitted dense systematic sampling across both preservation conditions and anatomical regions simultaneously yielding 144 biopsies. While, preserving sufficient statistical power to detect significant effects of preservation method. Collectively these findings establish sliced biopsies in RNAlater submersion as an operationally simple and highly effective RNA preservation strategy for deceased donor pancreatic tissue. This method can be extended to clinical biopsies without concern over regional differences of integrity. The protocol requires no specialist equipment or additional costs and is compatible with standard procurement logistics, outperforming traditional snap freezing and offering beneficial logistical timings than immediate RNA isolation. This approach has been embedded within established national biobanking infrastructure the Quality in Organ Donation biobank which utilises the submersion protocol described here (https://quod.org.uk/tag/panc). Preliminary unpublished data from our group suggest that downstream NanoString transcriptomic analysis is achievable from samples preserved using this submersion protocol[19]. The characterisation of RNA integrity and yield across preservation conditions, processing intervals, and anatomical regions reported here establishes that donor pancreas biopsies preserved by submersion can reliably support downstream transcriptomic analyses, overcoming a longstanding technical barrier in the molecular study of human pancreatic tissue.

CONCLUSION

This study demonstrates that RNAlater-based preservation, through either tissue submersion or injection, provides significantly improved RNA integrity compared with conventional snap freezing in deceased donor pancreatic tissue. Submersion of thin-sliced biopsies represents the most practical approach for routine preservation, achieving comparable RNA integrity to injection whilst requiring no specialist equipment, and integrating readily within existing biopsy workflows. These findings suggest that adoption of RNAlater-based protocols within biobanking infrastructure may enhance the quality and reproducibility of downstream transcriptomic analyses in human pancreatic tissue, and provide a rigorous basis for validation in larger prospective multi-donor studies.