Clinical and biological significance of circulating miRNAs in chronic pancreatitis patients undergoing total pancreatectomy with islet autotransplantation

Abstract Background Specific microRNAs (miRNAs) were elevated in chronic pancreatitis (CP) patients during islet infusion after total pancreatectomy (TPIAT). We aimed to identify circulating miRNA signatures of pancreatic damage, predict miRNA‐mRNA networks to identify potential links to CP pathogenesis and identify islet isolation and transplantation functional outcomes. Methods Small RNA sequencing was performed to identify distinct circulating miRNA signatures in CP. Plasma miRNAs were measured using miRCURY LNA SYBR green quantitative real‐time polymerase chain reaction assays. Correlation analyses were performed using R software. The miRNA target and disease interactions were determined using miRNet and the miRNA enrichment and annotation tool. Results Alterations were found in circulating miRNAs in CP patients compared to healthy controls. Further studies were conducted on 12 circulating miRNAs enriched in the pancreas, other tissues and other diseases including cancer and fibrosis. Approximately 2888 mRNAs in the pancreas were their targets, demonstrating interactions with 76 small molecules. Three miRNAs exhibited interactions with morphine and five exhibited interactions with glucose. The miRNA panel targeted 22 genes associated with pancreatitis. The islet‐specific, acinar cell‐specific and liver‐specific miRNAs were elevated at 6 h after islet infusion and returned to baseline levels 3 months after TPIAT. Circulating levels of miRNAs returned to pre‐transplant levels 1‐year post‐transplant. Circulating miRNAs measured before and 6 h after islet infusion were directly or inversely associated with metabolic outcomes at 3 and 6 months post‐transplant. Conclusions miRNAs may contribute to CP pathogenesis, and elevated circulating levels may be specific to pancreatic inflammation and fibrosis, warranting further investigation.


K E Y W O R D S
biomarkers, chronic pancreatitis, circulating miRNAs, islet autotransplantation

BACKGROUND
Chronic pancreatitis (CP), a pathologic, progressive fibroinflammatory syndrome of the pancreas, is characterized by chronic abdominal pain that significantly affects the quality of life. 1 In patients with refractory CP, total pancreatectomy with islet autotransplantation (TPIAT) improves pain and quality of life and prevents brittle diabetes. 2,3TPIAT is considered only after other surgical interventions have failed. 2Clinical practitioners face challenges in determining the timing of surgical interventions due to the disease's unpredictable and highly varied clinical course of the disease. 4,55][6] Most importantly, when exocrine pancreatic insufficiency and/or endocrine dysfunction are diagnosed, the damage caused by persistent inflammation and fibrotic replacement of pancreatic parenchyma is irreversible.Even if endocrine function is intact, an inflamed and fibrotic pancreas may affect islet isolation outcomes negatively.As optimal islet yield is critical in achieving insulin independence post-TPIAT, 7 knowledge of disease progression will help in determining the timing of surgery.
After TPIAT, routine diagnostic tests cannot predict islet stress and damage before islet graft failure.Thus, there is an urgent need to develop metrics to determine the status of pancreas inflammation, fibrosis and islet damage.Circulating microRNAs (miRNAs) are ideal candidates as biomarkers of pancreatic inflammation and islet damage.They are stable in circulation and can be detected easily using polymerase chain reaction (PCR) technologies.][14][15] Further analysis of circulating miRNA profiles in CP patients is needed to identify key miRNA signatures with excellent diagnostic potential.As miRNAs are crucial for post-transcriptional regulation of gene expression, they may play important roles in initiating and perpetuating pancreatic inflammation and fibrosis. 16The involvement of miRNAs in activating rat pancreatic stellate cells, which play a central role in fibrosis, was reported previously. 17 systematic review and meta-analysis of miRNA-mRNA regulation networks in CP revealed hsa-miR-324-5p and NOTCH3 (neurogenic locus notch homolog protein 3), COX5A (cytochrome C oxidase subunit 5a), THBS1 (thrombospondin 1) and KARS (lysyl-tRNA synthetase) as highrisk markers with high prediction accuracy.18 Further investigations of miRNA-mRNA networks in the pancreas and their contributions to the pathogenesis of CP are warranted.
Our objectives are to (1) identify differentially altered circulating miRNAs in CP; (2) identify the miRNA-mRNA networks in the pancreas; and (3) identify associations with metabolic outcomes before and after TPIAT.In this study, we profiled circulating miRNAs in CP patients compared to healthy controls using small RNA sequencing.We investigated the potential interactions of differentially expressed miRNAs in circulation in CP with mRNA networks in the pancreas.We further assessed whether the selected miRNAs targeted genes identified as associated with risk for pancreatitis using publicly available datasets (genome-wide association studies).We also investigated the associations of the selected miRNA candidates with clinical measures before and after TPIAT.

Plasma RNA extraction
Cell-free and exosomal RNA were extracted using the miRNeasy Serum/Plasma advanced kit (Qiagen) following the manufacturer's instructions.After stabilization using MS2 RNA (bacteriophage MS2, Millipore Sigma), spiking with Unisp6 (Qiagen) and lysis, plasma proteins were removed.After binding nucleic acids to the spin column and washing, miRNA was eluted in nuclease-free water containing an RNAse inhibitor.MS2 RNA, UniSp6 and RNAse inhibitors were avoided for small RNA sequencing.

Small RNA sequencing analysis
Small RNA integrity and quality were assessed using an Agilent 2100 Bioanalyzer.To profile circulating miRNAs, a miRNA library was constructed using the TruSeq Small RNA Library Prep Kit (Illumina Inc.) following the manufacturer's instructions.Following quality control, the libraries were sequenced using a HiSeq 2500 Genome Analyzer (Illumina Inc.), with a sequencing depth of 2 million reads per sample and a read length of 50 bp.Fastq files were generated from base-calling files using Illumina bcl2fastq2 software for data analysis.After trimming 3′adapter sequences using Cutadapt, 19 reads of length less than 12 nucleotides were excluded.The resulting trimmed reads were aligned to a publicly available database containing published human miRNA sequences and annotations (miRbase, release 22) using the quantifier.plmodule from miRDeep2. 20The count data were analysed using DeSeq2 for normalization and differential analysis. 21For principal component analysis and hierarchical clustering, the heatmap function package for non-negative matrix factorization was used (R software, version 3.6.3). 22miR-NAs with p < .05were considered to be differentially expressed.

Absolute quantification of circulating miRNAs
After RNA extraction from plasma and miRNA (2 μL/sample) conversion to cDNA (miRCURY LNA RT kit, Qiagen), quantitative real-time PCR (qPCR, cDNA dilution 1:40) was performed (QuantStudio 7 Flex realtime PCR system, Applied Biosystems) using miRCURY LNA miRNA PCR assay system following the manufacturer's instructions.A standard curve using specific miRNA mimics (miRNA mimics, Qiagen) and melting curve analysis were included in all reactions.The UniSp6 cycle threshold value (±0.2) was used as a quality control for inclusion in the data analysis.Patient information was protected by relabeling all samples and blinding the investigator.All samples were assayed in triplicate.QuantStudio™ 7 Flex software was used for data analysis (Applied Biosystems).

miRNA-mRNA network analysis
Interactions of miRNAs with mRNAs in the pancreas (464 datasets) or small molecules were assessed using miRNet (version 2.0). 24To identify miRNA interactions with genes associated with a risk for pancreatitis, we compiled a list of predicted targets of the selected miRNAs using Tar-getScan (version 8.0). 25 We searched the genome-wide association studies catalogue (National Human Genome Research Institute-European Bioinformatics Institute, NHGRI-EBI) for genes associated with risk for pancreatitis.Based on 25 published studies, 187 variants (112 mapped genes) were associated with pancreatitis.We compared these lists of mRNA targets of miRNAs and pancreatitisassociated genes (13 lists) using a web-based "multiple list comparator" tool (molbiotools.com) 26to identify potential interactions with genes associated with risk for pancreatitis.

Statistical analysis
All analyses were performed using R software (version 4.2.2) 22 and Bioconductor 3.16. 27Raw miRNA data were transformed to the common log (log to base 10) scale for all analyses.Associations of log 10 miRNA levels with mea- sures on continuous scales (e.g., C-peptide or islet-yield measures) were determined using linear regression.The data were corrected for multiple comparisons for computing FDR-adjusted p-values with and without bootstrapping (empirical distributions, bootstrapped resampling with 1000 replications).In each comparison, a clinical parameter, a time point and 12 miRNAs were considered, and p-values were averaged over all replications.Graph-Pad PRISM (version 9.0) was used for receiver operating characteristic curve analysis.A p-value of < .05 was considered statistically significant.For multiple comparisons, a p-value of < .1 was considered statistically significant.

Distinct circulating miRNA profiles of CP patients
The experimental design for the study is provided in Figure 1A.To profile circulating miRNAs in patients with CP, we isolated small RNAs from plasma collected from patients (n = 18) admitted for TPIAT and normal healthy donors (n = 6) and performed small RNA-seq analyses.Quality control revealed that healthy donor samples (n = 3) exhibited low levels of circulating miRNAs, leading to insufficient reads (<2 million reads) and, hence, *Small RNA was isolated from patient plasma samples and subjected to small RNA sequencing.Differentially expressed circulating miRNAs with high base mean in chronic pancreatitis patients (n = 15) compared to healthy controls (n = 3) are listed.Out of six healthy controls included in the study, three samples did not contain measurable levels of circulating miRNAs and were excluded from the study.FC, fold change; p adj , p-value adjusted for multiple comparisons.
were excluded from further analyses.Principal component analyses revealed that the circulating small RNA profiles of healthy donors were clustered, while those of patients were variable (Figure 1B).In the patient group, three samples were excluded because of high variability (principal component analysis), poor library quality and insufficient reads for analyses (<2 million reads).A heatmap showing the differential circulating miRNAs in patients with CP compared to healthy donors is depicted in Figure 1C.Small RNA-seq analysis revealed 803 miRNAs in circulation, of which 53 were significantly expressed in patients with CP (p < .05, Figure 1C).When adjusted for multiple comparisons, these did not reach statistical significance (Table 1).

Tissue expression and localization of selected miRNA candidates
We performed overrepresentation analysis using the miEAA 23 to assess expression in diseases, organs and tissues and their cellular localization (Figure 2).Of the 12 selected miRNA candidates, 11 have been identified in extracellular vesicles called exosomes and 6 have been identified in microvesicles.Most importantly, 7 of these miRNA candidates have been associated with Argonaut 2 (Ago2) complexes in circulation, forming functional miRNA-mediated silencing complexes (miRISC/Ago2) (Figure 2A).Selected miRNA candidates have also been identified in the nucleus, and except for hsa-miR-216a-5p, all miRNA candidates have been observed in microvesicles, cytoplasm and circulation (Figure 2B).Supporting Information Table S1 lists tissues enriched with the selected miRNA candidates (miRNet version 2.0, function explorer tool). 24nterestingly, except hsa-let-7e-5p, all miRNA candidates were significantly enriched in the pancreas (Supporting Information Table S1).These miRNAs were overrepresented in several pathways, including bacterial invasion of epithelial cells, Parkinson's disease pathogenesis, type 2 diabetes, apoptosis and notch signalling (Figure 2C) (KEGG).These selected miRNA candidates have also been implicated in several diseases (Supporting Information Table 2).

miRNA networks in the pancreas and possible links to CP
We analysed miRNA:mRNA/small molecule networks in the pancreas to identify key players in the pathogenesis of CP using miRNet (version 2.0) 23 (Figure 3A, B).Exploration of miRNA interactions with small molecules revealed 72 compounds, including glucose, morphine, vitamin D3, activin A and metformin, with 152 interactions (Figure 3A).In the pancreas, our analyses revealed 7938 genes as targets of 11 out of 12 miRNAs, with 16 222 interactions (Figure 3B).To further identify miRNA networks with genes associated with a risk for pancreatitis, we compiled lists of predicted targets of these miRNAs using TargetScan (Version 8.0). 25 We then compared these lists to 187 variants (mapping to 112 genes) identified as associated with pancreatitis in genome-wide association studies using the NHGRI-EBI catalogue.The commonalities between genes associated with pancreatitis and miRNAs are provided in Figure 3C.These possible miRNA-gene networks in the context of pancreatitis are depicted in Figure 4. Genes with at least two interactions with selected miRNA candidates are indicated in red.

Pre-and post-operative patient characteristics
To determine whether levels of selected circulating miRNA profiles are altered in patients after TPIAT, we included 40 patients from a total of 200 patients who had follow-up plasma samples and data available at admission for TPIAT, 6 h after islet infusion and 3 months and 1 year after TPIAT.We included the 15 patients from our small RNA sequencing study in these follow-up studies.Preand post-operative characteristics of the selected patient cohort are provided in Tables 2 and 3, respectively.Table 2 includes patient demographics, disease duration and aetiology of pancreatitis, metabolic profile and information on islet isolation.Table 3 provides functional outcomes at 3, 6 months and 1 year after transplantation including haemoglobin A1c, glycosylated haemoglobin (HbA1c), fasting C-peptide, glucose, plasma glucagon levels and insulin usage.At 6 months after TPIAT, plasma glucagon levels were not available.

The ability of hsa-miR-125b-5p and hsa-122-5p to distinguish CP patients from healthy donors
For receiver operating characteristic analysis, we compared the circulating miRNA levels of 19 healthy donor samples (10 female and 9 male donors aged 26−54; Supporting Information Table S3) and 40 CP patient plasma samples collected before TPIAT.We observed that hsa-miR-375, hsa-miR-221-3p, hsa-miR-200c-3p, hsa-miR-125b-5p, hsa-miR-99b-5p, hsa-let-7e-5p and hsa-miR-122-5p could distinguish between healthy donors and CP patients significantly (Figure 5).Of these, hsa-miR-122-5pand hsa-miR-125b-5p exhibited higher area under the curve F I G U R E 4 miRNA networks with genes identified as associated with risk for pancreatitis.A total of 112 mapped genes corresponding to 187 variants were identified as associated with risk for pancreatitis in genome-wide association studies.Of these, predicted interactions of miRNAs with genes are shown.Genes with 2 or more interactions with miRNAs are depicted in red.

Varied profiles of circulating miRNAs before and after TPIAT
To assess levels of circulating miRNAs before and after TPIAT, we performed a time-course analysis of selected circulating miRNAs at admission, 6 h after islet infusion and 3 months and 1 year after TPIAT.Absolute levels of circulating miRNAs (fmol/mL) are provided in Figure 6.Circulating levels of hsa-miR-375 (islet-specific, p < .001, Figure 6A), hsa-miR-216a-5p (acinar cell-specific, p < .001, Figure 6B) and hsa-miR-122-5p (liver-specific, p < .001, Figure 6C) were elevated at 6 h after islet infusion.Interestingly, islet-specific and acinar cell-specific miRNA levels returned to baseline 3 months after TPIAT.At 3 months, although not significant, hsa-miR-122-5p was slightly elevated but returned to baseline at 1 year after TPIAT, suggesting gradual improvement in liver inflammation after TPIAT.Levels of other circulating miRNAs gradually increased from 3 months and remained elevated at 1 year post-TPIAT (p < .05,< .01,< .001, Figure 6).

DISCUSSION
One of the factors in achieving post-transplantation insulin independence is optimal islet yield. 7Due to the unpredictable clinical course of the disease and the lack of predictive tools, disease stage and islet isolation outcomes cannot be predicted before TPIAT.Although currently available tests can provide information on metabolic mea- sures, including C-peptide and glucose levels, islet stress and dysfunction cannot be monitored before an irreversible loss of function before and after TPIAT.In this study, we identified distinctive circulating miRNA signatures of pancreatic damage, predicted miRNA-mRNA networks to identify potential links to CP pathogenesis and identified associations of circulating miRNAs with islet isolation and post-transplantation functional outcomes (Graphical Abstract).
Our observations of elevated levels of hsa-miR-148a-3p, hsa-miR-221-3p and hsa-miR-122-5p in CP patients are consistent with other studies. 12,14,15Since hsa-miR-375 is an islet damage marker, 8 reduced levels in circulation may indicate intact islet function and survival before TPIAT in these patients.Elevated levels of hsa-miR-375 at 6 h after islet infusion may indicate islet stress and damage immediately after islet isolation and transplantation.The patients excluded from analyses did not have inflamed pancreas, with islet yields ranging between 161 577 IEQs (4.0 mL tissue volume) and 198 947 IEQs (6.0 mL tissue volume).Although three healthy control samples were also excluded from the analysis, the extent of pancreatic inflammation, fibrosis and damage might influence the quality of circulating miRNAs.For further investigations, we selected hsa-miR-99b-5p, hsa-miR-148a-3p, hsa-miR-221-3p, hsa-let-7e-5p, hsa-miR-122-5p and hsa-miR-375 based on base mean, FC and p values.Our previous sequencing analysis also identified these selected miRNAs as being differentially released from isolated human islets in exosomes under proinflammatory and hypoxic conditions and in TPIAT patients after islet infusion. 10Although our sequencing analysis is limited by a small sample size, consistency with our previous observations suggests that these miRNA signatures may be significant in CP and islet function.15]29 We included these additional miRNAs in our follow-up analysis because we previously identified these miRNAs in islet cultures as differentially altered in proinflammatory conditions 10 or CP . 15Our current time-course analysis allows us to study their profiles following TPIAT and assess whether they were associated with metabolic outcomes.As observations from independent studies may be inconsistent due to differences in disease duration, the status of pancreas inflammation and fibrosis, specimen collection and storage, and techniques for measuring circulating miRNAs, we included these additional miRNAs in our studies to better understand the profiles of circulating miRNAs before and after TPIAT.
The candidate miRNAs must be disease-and pancreasspecific to be non-invasive circulating miRNA predictors of pancreatic inflammation, islet yield and post-TPIAT insulin independence.Thus, we assessed whether these selected miRNAs were pancreas-specific and/or CPspecific using overrepresentation analysis.Association with Ago2 or exosomes and microvesicles stabilizes the miRNAs in circulation. 30Although the selected miRNAs were significantly enriched in the pancreas (except for hsa-let-7e-5p), our current data revealed that other tissues might contribute to the circulating miRNA pool.These miRNAs were also differentially altered in circulation in other diseases.Thus, these selected miRNA candidates are neither pancreas-specific nor CP-specific.However, they are enriched in the pancreas and elevated/reduced in circulation in CP patients, which may indicate their involvement in disease-specific processes, warranting further investigations.
Pathway analysis revealed an overrepresentation of these miRNAs in the pathways indicated in Figure 2C.In the context of CP, activation of Notch signalling plays a vital role in pancreas regeneration and repair. 31Gap junction proteins regulate pancreatic stellate cell proliferation, function and activation. 17Trans-endothelial migration of leukocytes is critical in initiating and perpet-uating pancreatic inflammation. 32A previous report on the miRNA: mRNA regulation network in CP was based on their tissue expression profile, collected from the Gene Expression Atlas.In this independent study, analysis of miRNA-mRNA pathway regulating networks identified the involvement of NOTCH, TGF-β, wingless-related integrated site, extracellular matrix and ubiquitin-mediated proteolysis pathways. 18These common findings emphasize the involvement of miRNAs in CP pathogenesis.Further investigations are necessary to identify interactions of miRNAs with mRNAs that regulate leukocyte migration, gap junction proteins, TGF-β and notch signalling pathways and their relevance to miRNAs in circulation to understand their contributions to disease progression.
Our analysis predicted 152 interactions of miRNAs with 72 small molecules including glucose, morphine, vitamin D3, activin A and metformin.Small molecules can either repress or promote miRNA transcription. 33In healthy individuals treated with hydromorphone and oxycodone, hsa-miR-221-3p and let-7 family were upregulated in circulation. 34Given that the let-7 family of miRNAs plays integral roles in opioid tolerance, 35 elevated levels of hsa-let-7e-5p and others in circulation in our cohort of CP patients may be associated with opioid treatment and be unrelated to disease characteristics.Our analysis also predicted interactions of hsa-miR-148a-3p, hsa-miR-200c-3p, hsa-miR-29b-3p, hsa-miR-375 and hsa-miR-122-5p with glucose.Previously, we observed that isolated human islets released these miRNAs into culture media in proinflammatory and hypoxic conditions. 10Elevated levels of hsa-miR-148a-3p were also observed in autoantibody-positive, non-diabetic, recent-onset type 1 diabetic and longstanding type 1 diabetic individuals. 36Our analysis also predicted 22 mRNA targets of the selected miRNAs with a risk for pancreatitis (Figure 4).Considering all our observations, the selected miRNAs may be involved in specific disease processes in CP.Their circulating levels may indicate subtle alterations in islet function in CP patients even though preoperative metabolic measures indicated optimal islet function and reflect specific unexplored disease states.Future investigations should focus on experimental valida- tion of these predicted interactions of miRNAs with small molecules, genes associated with risk for pancreatitis and pathways linked to CP pathogenesis.These experiments will help us understand why these miRNAs are elevated in circulation in CP patients and establish their contributions to CP pathogenesis.
Receiver operating characteristic analysis revealed that hsa-miR-125b-5p and hsa-miR-122-5p distinguished CP patients from healthy controls with good diagnostic accuracy.Although a subset of other miRNAs could significantly discriminate CP patients from healthy controls, AUC scores were sub-optimal, ranging between .65 and .7.As hsa-miR-122-5p is a liver-abundant miRNA, it is unclear as to why it is elevated in CP patients and discriminated CP patients from healthy controls.Whether perpetual pancreatic inflammation and fibrosis affect hepatocyte function and activate common progenitor cells is unknown.Another possibility is that chronic narcotics usage may induce liver damage in these patients without clinical manifestation.Further investigations are necessary to understand whether our findings are specific to CP by comparing circulating miRNA profiles in patients with CP to those with hepatitis or chronic narcotics usage.
In our previous study, POST (Prospective Observational Cohort Study of TPIAT), involving 139 participants from 9 institutions, 6 we did not follow up on profiles of the circulating miRNAs after TPIAT.In our current study, we did not validate our sequencing observations in an independent cohort of CP patients.We conducted further studies using patient samples (n = 40, including patients from our sequencing analysis) collected before and after TPIAT.One of the objectives of our current study was to investigate whether circulating levels of selected miR-NAs were altered following TPIAT.As expected, islet-, acinar cell-and liver-specific miRNAs were increased at 6 h after islet infusion, indicating islet and acinar  cell stress and liver inflammation or damage immediately in the post-operative period.These miRNAs returned to baseline levels 3 months following TPIAT, reflecting recovery.Time-course analysis during and after islet infusion revealed that circulating islet stress and damage miRNA levels peak at the end of islet infusion and normalize to baseline levels 7 days after transplantation. 10As these miRNAs measured at 1-year post-TPIAT were not associated with functional outcomes, it is unclear whether they reflect subtle changes in islet graft function and survival.When adjusted for multiple comparisons, circulating miRNAs were not associated with the duration of disease symptoms, consistent with previous observations in the POST study cohort.While circulating miRNAs did not correlate with age and body mass index in our current study, circulating hsa-miR-200c-3p, hsa-miR-148a-3p and hsa-miR-221-3p were significantly higher in adults compared to paediatric patients in the POST study. 6We did not compare profiles of circulating miRNAs between patients of different aetiologies and with or without exocrine insufficiency because sample size and statistical power would be reduced by sub-group analyses.Nevertheless, levels of circulating miRNAs could be influenced by pancreatic tissue volume and severity of the disease, warranting further investigations. 6 we did not measure pancreatic tissue levels of these miRNAs, it is unclear whether the elevated levels in circulation are due to increased tissue miRNA levels in CP.Interestingly, hsa-miR-375 measured 1 year after transplantation was associated with islet yield, indicating islet survival 1 year after transplantation.Of the 63 significant associations, 33 remained significant after correction for multiple comparisons (23 associations, BS p adj < .05).The strong inverse associations of selected miRNA candidates measured before TPIAT and 6 h after infusion with post-TPIAT metabolic outcomes highlight their potential as predictors of islet survival and function after transplantation.Although the patients were on insulin therapy to maintain HbA1c levels under 6.5%, C-peptide was detectable and in the normal range in the fasted state at 3 months and up to 1 year after transplantation.Plasma glucagon was also detectable (at sub-physiological levels) 1 year after transplantation.As exogenous insulin therapy may influence circulating miRNA levels apart from metabolic measures, future studies should test our hypothesis by comparing the associations of circulating miRNAs between patients with and without insulin independence 1 year after transplantation.
Weak correlations observed in our study could be due to the cohort heterogeneity and the small sample size.

F I G U R E 1
Circulating miRNA profiles in chronic pancreatitis.(A) Experimental design.Plasma miRNAs were isolated and subjected to small RNA sequencing analysis.(B) Principal component analysis plot.(C) Heatmap representation of circulating miRNA sequencing data (differential expression analysis based on the negative binomial model) from patients with chronic pancreatitis and healthy controls.Colour codes for fold change are indicated in the heat map.

F I G U R E 2
Tissue expression and localization of selected miRNA candidates and overrepresentation in KEGG pathways.Selected miRNAs were input into miRNA Enrichment and Annotation Tool software for overrepresentation analysis in (A) association in circulation; (B) cellular localization; and (C) KEGG pathways.Heat maps illustrate -log 10 (p value) for corresponding miRNAs.

F
I G U R E 3 miRNA-mRNA networks in the pancreas.(A) Clusters reveal distinct miRNA-compound interaction networks in the pancreas based on expression data available in the literature.(B) Clusters reveal distinct miRNA-mRNA interaction networks in the pancreas based on expression data available in the literature.(C) Jaccard similarity plot showing common targets of selected miRNA candidates in the pancreas and genes identified as associated with risk for pancreatitis.

TA B L E 1 Differentially expressed circulating miRNAs in patients with chronic pancreatitis.* Circulating miRNA Base mean log 2 FC FC p Value p adj Value Elevated miRNAs in circulation
Preoperative patient characteristics.
TA B L E 2 Metabolic outcomes at 3, 6 and 12 months after total pancreatectomy with islet autotransplantation.* TA B L E 3*Follow-up data were not available for all patients included in the study.NA, data not available.

miRNAs 6 h after islet infusion Metabolic measures pre-TPIAT Metabolic measures 3 months post-TPIAT Metabolic measures 6 months post-TPIAT Parameter R p p adj BS p adj Parameter
Association of circulating miRNAs at 6 h after islet infusion with metabolic measures before and after TPIAT.Circulating TA B L E 4 PP, postprandial; ND, association not detected; R, Pearson's correlation coefficient; TPIAT, total pancreatectomy with islet autotransplantation; p adj , FDR adjusted p-value; BS p adj , Bootstrapped FDR adjusted p-value.