Quantitative assessment of pancreas R2* in cirrhosis with systemic iron overload: relationship with serum iron parameters and metabolic complications

Abstract

Background

Although pancreatic iron deposition is linked to metabolic dysfunction in hereditary hemochromatosis and transfusional iron overload, its significance in cirrhosis with secondary iron overload remains unclear. We quantitatively assessed pancreatic R2* in cirrhotic patients with and without systemic iron overload using magnetic resonance imaging (MRI)-based R2* relaxometry, evaluated its correlation with hepatic iron content and serum iron parameters, and examined its association with 5-year metabolic outcomes.

Methods

This retrospective observational cohort study included 170 cirrhotic patients who underwent magnetic resonance elastography (MRE) with multi-echo proton density fat fraction (PDFF)/R2* mapping between January 2018 and October 2024. Pancreatic and liver R2* values were measured. Patients were categorized into iron overload and normal iron status based on iron parameters [transferrin saturation (TSAT) >45% and ferritin >200 µg/L in women and TSAT >50% and ferritin >300 µg/L in men]. Correlations between pancreatic R2*, serum iron parameters, liver function markers, and MRI-derived measures were analyzed. The 5-year cumulative incidence of metabolic complications including dyslipidemia, prediabetes, and type 2 diabetes mellitus (T2DM) was assessed.

Results

Pancreatic R2* values were higher in patients with iron overload compared to those with normal iron status, but this difference was observed only in patients with alcoholic liver cirrhosis {26.2 [interquartile range (IQR), 19.5–37.9] vs. 34.7 [IQR, 24.8–345.3] s⁻¹, P=0.001}. Among patients with iron overload, pancreatic R2* values correlated with serum ferritin {adjusted r=0.175 [95% confidence interval (CI): 0.016–0.389], P=0.014} and total iron-binding capacity (TIBC) [adjusted r=−0.175 (95% CI: −0.369 to −0.015), P=0.018] but showed no correlation with liver iron concentration (LIC) [adjusted r=0.292 (95% CI: −0.030 to 0.558), P=0.058]. In this subgroup, pancreatic R2* values also correlated with higher Child-Pugh [adjusted r=0.371 (95% CI: 0.039–0.645), P=0.014] and Model for End-Stage Liver Disease (MELD) scores [adjusted r=0.383 (95% CI: 0.098–0.622), P=0.011]. However, pancreatic R2* did not predict the development of dyslipidemia, prediabetes, or T2DM (all P>0.05).

Conclusions

Pancreatic R2* values correlated with serum iron parameters and liver disease severity, but not with LIC or future metabolic complications.

Introduction

Iron overload occurs in 32–37% of cirrhotic patients without hereditary hemochromatosis (HH) (1,2), resulting from hypersplenism-related hemolysis, transfusions, enhanced intestinal absorption, reduced hepcidin levels, and portosystemic shunting (3,4). Hepatic siderosis promotes hepatocarcinogenesis and is associated with decompensation and reduced survival (5).

Magnetic resonance imaging (MRI) R2*-based relaxometry is the noninvasive gold standard for quantifying tissue iron content, with excellent reproducibility for detecting and monitoring iron distributions across multiple organs (6). While the liver is the primary iron storage site, severe systemic iron overload can affect extrahepatic organs including the pancreas (7).

Pancreatic iron accumulation has garnered particular attention due to its metabolic implications. In HH, type 2 diabetes mellitus (T2DM) occurs in 12–33% of unselected patients and up to 50–85% in advanced stages, primarily through iron-mediated oxidative stress, impaired β-cell function, and decreased insulin secretion (8,9). Similarly, in transfusion-dependent thalassemia major, iron accumulation in adipocytes reduces adiponectin secretion, contributing to insulin resistance alongside direct pancreatic iron toxicity (10,11). Clinical studies have consistently linked elevated pancreatic R2* values with T2DM in HH, thalassemia, and sickle cell disease (8-13).

Despite reported associations between pancreatic iron deposition and metabolic dysfunction in HH and transfusional iron overload, it is unclear whether pancreatic iron accumulation in cirrhosis correlates with hepatic iron content, whether it reflects systemic iron burden or hepatic dysfunction, and whether it confers an increased risk of metabolic complications.

Therefore, this study aimed to: (I) quantitatively assess pancreatic R2* values in cirrhotic patients with and without systemic iron overload using magnetic resonance elastography (MRE) with multi-echo proton density fat fraction (PDFF)/R2* mapping; (II) examine correlations between pancreatic R2* and serum iron parameters, liver function markers, and MRI-derived measures; and (III) investigate whether iron overload status is associated with increased 5-year cumulative incidence of metabolic complications, specifically dyslipidemia, prediabetes, and T2DM. We present this article in accordance with the STROBE reporting checklist (available at https://qims.amegroups.com/article/view/10.21037/qims-2026-1-0419/rc).

Methods

Study patients

This retrospective observational cohort study included a cohort of patients with liver cirrhosis evaluated at Severance Hospital, comprising both liver transplantation (LT) candidates undergoing preoperative work-up and non-candidates referred for hepatologic evaluation. Cirrhosis was diagnosed on clinical and/or radiologic grounds, based on characteristic imaging features such as nodular liver surface, caudate-to-right lobe hypertrophy, and signs of portal hypertension including splenomegaly, recanalized paraumbilical vein, or other spontaneous portosystemic collaterals and/or clinical and laboratory evidence of chronic liver disease, with or without hepatic decompensation (14). This study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. This study was approved by the Institutional Review Board (IRB) of Severance Hospital (IRB No. 4-2025-1600). The requirement for informed consent was waived due to the retrospective nature of the study. Although MRE with multi-echo PDFF/R2* mapping (hereafter referred to as MRE) is not part of routine standard of care for patients with cirrhosis, it is routinely performed at Severance Hospital in LT candidates to assess liver stiffness, portal hypertension, and parenchymal iron or fat content. MRE was not performed in a limited number of cases, including when patients declined MRI due to insurance coverage issues or when deceased donor LT or emergent living donor LT was performed without sufficient time for preoperative MRI. For non-candidates, MRE was performed at the hepatologist’s discretion based on a clinical need for the same assessments and patient preference, with full informed consent. Using electronic medical records, 561 patients who underwent MRE between January 2018 and October 2024 were identified. The exclusion criteria were as follows: (I) no clinically or radiologically diagnosed cirrhosis; (II) known diseases affecting liver iron content, including HH and blood transfusion ≥6 months; (III) diagnosis of dyslipidemia, impaired glucose tolerance (IGT) or type 1 diabetes mellitus (T1DM) or T2DM; (IV) receipt of iron supplement within one month prior to MRE; (V) poor quality of R2* images or artifacts; (VI) absence of laboratory results, including serum iron panel, within one week prior to MRE; and (VII) lack of analyzable liver or pancreas such as polycystic liver and renal disease, pancreatic agenesis, post-operative status, or incomplete coverage outside the scan field. Based on these exclusion criteria, a total of 170 patients were included in the analysis (Figure 1).

Figure 1.

Flow chart illustrating patient selection for the study. MRE, magnetic resonance elastography.

Clinical and demographic profile including age, sex, body mass index (BMI), etiology of liver disease, Child-Pugh class/score, Model for End-Stage Liver Disease (MELD) score, history of prior transfusion or variceal bleeding, and pre-MRE laboratory results, were collected. Severity of ascites was categorized according to the grading criteria established by the International Ascites Club (15). Information on smoking and alcohol use was collected and categorized as current, former, or never smoker/drinker. For current and former drinkers, the amount of alcohol intake was additionally recorded. Excessive alcohol consumption was defined as a daily alcohol intake of more than 30 g for men and 20 g for women (16). The iron load for each patient was expressed as mg/month, accounting for intravenous (IV) iron sucrose and oral iron therapy; the latter calculated under the assumption of 10% gastrointestinal absorption. The cumulative iron exposure was then derived by summing IV iron, oral iron, and transfusion-related iron, with each unit of packed red blood cells estimated to contribute 200 mg of iron (17). Additionally, for LT recipients, the type of maintenance immunosuppressive therapy and the use of maintenance steroid therapy ≥6 months after transplantation were recorded because of their potential catabolic and/or hyperglycemic effects.

MRE

All liver MRE examinations were performed on a 3.0-T MRI scanner (Discovery MR750W, GE Medical Systems, Milwaukee, WI, USA) using a 32-channel torso coil. A 19 cm passive pneumatic driver was placed over the right costal margin and connected to a waveform generator via pneumatic tubing to deliver 60 Hz vibrations. Four axial slices, encompassing mid-portion of right hepatic lobe around the main portal vein level, were obtained in a single 14-s breath-hold. Wave images and elastograms were generated automatically; cross-hatched regions (95% confidence threshold) were excluded from analysis.

The acquisition parameters for the spin-echo planar MRE sequences were: field of view (FOV), 380 mm × 380 mm; matrix size, 64×64; section thickness, 8 mm; repetition time (TR) of 1,000 ms; echo time (TE) of 62 ms; and flip angle 90°. T2-weighted images were obtained with the following parameters: FOV, 380 mm × 380 mm; matrix size, 320×256; section thickness, 8 mm; TR of 540 ms; TE of 80 ms; and flip angle of 90°. For PDFF and R2* mapping, sequence parameters were as follows: FOV, 400 mm × 320 mm; matrix size, 160×128; section thickness, 6 mm; TR, 7.24 ms; TE, minimum full (0.9–4.4 msec); number of echoes, 6; and flip angle, 4°. PDFF and R2* sequences were obtained during a single breath-hold (<17 s) using commercially available chemical shift-encoded pulse sequence [Iterative Decomposition of water and fat with Echo Asymmetry and Least-squares estimation (IDEAL) IQ; GE Healthcare, Milwaukee, WI, USA], which automatically corrects for fat-water signal interference (18).

Quantification of liver stiffness and liver/pancreatic fat fraction and R2*

Liver stiffness was measured by placing a freehand region of interest (ROI) within the right hepatic lobe avoiding lesions, vessels, poor wave propagation and cross-hatching. At least four ROIs were placed per patient, and the mean value was used.

Liver R2* and fat fractions were measured by placing circular ROIs in each Couinaud segment except segment I, and the mean of the remaining eight segments was used (17). Segment I (caudate lobe) was excluded to minimize susceptibility and flow-related artifacts arising from its proximity to the inferior vena cava and portal vein, as well as partial volume effects related to its small size and anatomic variability in cirrhotic livers (19). Pancreatic R2* and fat fraction were measured by placing one circular ROI in the body and one in the tail of the pancreas, ensuring adequate distance from organ boundaries (Figure S1). All ROIs were drawn to avoid vascular structures, ducts and motion or pulsation artifacts. The mean of the body and tail measurements was calculated.

Quantification of liver stiffness, and hepatic and pancreatic fat fraction and R2* values was independently performed by two board-certified radiologists (J.H.P. and J.P., each with 5 years of post-residency experience in abdominal radiology), and consensus values were used for statistical analyses.

Calculation of liver iron concentration (LIC)

LIC was estimated from liver R2* using the equation by Wood et al. (20):

$$LIC\left(mg\text{/}gdryweight\right)=0.025\times R2{\text{*}}_{1.5T}+0.202$$ [1]

Since the original formula was based on 1.5-T MRI, R2* values from 3.0-T scans were converted to 1.5-T equivalents (21):

$$R2{\text{*}}_{3.0T}=2\times R2{\text{*}}_{1.5T}-11$$ [2]

Study endpoints

The primary endpoint was the correlation between pancreas R2* values and serum iron panel as well as MRE-derived measures including liver stiffness, LIC, liver fat fraction, and pancreatic fat fraction. The secondary endpoint was the 5-year cumulative incidence of metabolic complications, specifically dyslipidemia, prediabetes, T2DM, and prediabetes or T2DM, between patients with normal iron status and those with iron overload.

The presence of iron overload was suspected based on iron parameters [transferrin saturation (TSAT) >45% and ferritin >200 µg/L in women and TSAT >50% and ferritin >300 µg/L in men] (22). Since specific cut-offs for non-hereditary iron overload in cirrhosis have not been established, we adopted the thresholds recommended for HH, which have also been suggested for use in patients with chronic liver disease (23). Both TSAT and ferritin were considered, as ferritin may increase with inflammation and TSAT may be falsely elevated when total iron-binding capacity (TIBC), the denominator of the calculation, is reduced. Patients were followed from the date of MRE to the first diagnosis of each metabolic complication or last clinical follow-up, whichever occurred first. Follow-up continued until October 31, 2025.

Dyslipidemia was defined as fasting low-density lipoprotein (LDL) cholesterol levels ≥100 mg/dL, triglyceride (TG) levels ≥150 mg/dL, high-density lipoprotein (HDL) cholesterol levels <40 mg/dL in men and <50 mg/dL in women, or the use of lipid-lowering medication (24). T2DM was defined according to the American Diabetes Association’s criteria (25), as one of following: hemoglobin A1C (HbA1c) ≥6.5%, fasting plasma glucose (FPG) ≥126 mg/dL, 2-hour plasma glucose (PG) ≥200 mg/dL during a 75 g oral glucose tolerance test, or taking antidiabetic medication(s) (26). Prediabetes was defined as one of the following: HbA1c 5.7–6.4%, FPG 100–125 mg/dL, or 2-hour PG 140–199 mg/dL (25).

Statistical analysis

Continuous variables were analyzed using the Mann-Whitney U-test. Categorical variables were evaluated using the Chi-squared test unless more than 20% of cells had expected frequencies below 5, in which case Fisher’s exact test was applied. Correlations were assessed using Spearman rank correlation. Subgroup analyses were performed based on iron status, and partial Spearman correlation analysis accounting for pancreatic fat fraction was also performed to exclude confounding effect of pancreatic fat fraction on R2* values. Bland-Altman difference plots were used to assess agreement and bias among interobserver measurements of liver and pancreatic fat fraction and R2* values. Time-to-event analyses for metabolic complications were performed using Kaplan-Meier curves and compared with the log-rank test. Cox proportional hazards models were used to estimate hazard ratios (HRs) and 95% confidence intervals (CIs). In a sensitivity analysis, we stratified patients into quartiles of pancreatic R2* and assessed its association with metabolic outcomes using Cox-proportional hazards models, adjusting for age, sex, BMI, and LIC. Subgroup analyses were conducted for LT recipients and patients with alcoholic liver cirrhosis. In subgroups with a limited number of events or quasi-complete separation, a Firth-penalized likelihood approach was applied to obtain bias-reduced estimates. In the LT recipient subgroup, to assess the potential confounding effects of immunosuppressive regimen and maintenance steroid use after transplantation, multivariable Firth penalized Cox models were fitted with adjustment for these variables. All statistical analyses were performed using R version 4.5.0 (R Foundation for Statistical Computing). A two-tailed P<0.05 was considered statistically significant.

Results

Patient characteristics

Among 170 cirrhotic patients analyzed, 76 (45%) patients underwent MRE as part of the preoperative evaluation for LT, while 94 (55%) patients underwent MRE at the hepatologist’s discretion for fibrosis assessment or based on patient preference. Of these, 44 (25.9%) had iron overload according to serum iron parameters. Consistent with systemic iron overload, these patients had higher serum ferritin, serum iron, and TSAT levels, but lower TIBC (Table S1). No significant difference in C-reactive protein, a proinflammatory marker, was observed between the two groups.

Baseline patient characteristics are summarized in Table 1. Patients with iron overload had a higher prevalence of alcoholic cirrhosis and a greater proportion of current or former drinkers. They also exhibited impaired liver function, reflected by higher total bilirubin (P=0.019), lower albumin (P=0.005), and a trend toward prolonged prothrombin time (PT) (P=0.056). Both Child-Pugh and MELD scores were significantly higher in this group. In addition, patients with iron overload demonstrated higher LIC (P=0.014) and pancreas R2* values (P=0.001).

Table 1. Baseline patient characteristics.

Variables	Normal iron (n=126)	Iron overload (n=44)	P value
Age (years)	57.0 (50.0; 62.0)	58.0 (43.0; 62.0)	0.853
Sex			0.999
Men	87 (69.0)	30 (68.2)
Women	39 (31.0)	14 (31.8)
BMI (kg/m2)	24.5 (22.3; 27.0)	25.9 (22.7; 29.3)	0.096
Smoker			0.638
Current smoker	11 (8.7)	5 (11.4)
Former smoker	39 (31.0)	16 (36.4)
Never smoker	76 (60.3)	23 (52.3)
Drinker			0.048*
Current drinker	4 (3.2)	4 (9.1)
Former drinker	52 (41.3)	24 (54.5)
Never drinker	70 (55.6)	16 (36.4)
Excessive alcohol consumption	23 (41.1)	15 (53.6)	0.394
Hypertension	31 (24.6)	4 (9.1)	0.048*
Coronary artery occlusive disease	10 (7.9)	1 (2.3)	0.338
Chronic kidney disease	3 (2.4)	0 (0.0)	0.713
Erythropoietin stimulating agent therapy	2 (1.6)	0 (0.0)	0.977
Darbepoetin dose in the month of MRI (μg/week)	0 (0; 0)	0 (0; 0)	0.555
IV iron sucrose therapy	1 (0.8)	0 (0.0)	0.999
IV iron sucrose dose (mg/month)	0 (0; 0)	0 (0; 0)	0.566
Oral iron therapy	11 (8.7)	3 (6.8)	0.937
Oral iron therapy dose (mg/month)	0.0 (0.0; 0.0)	0.0 (0.0; 0.0)	0.757
Transfused patients	32 (25.4)	4 (9.1)	0.039*
Number of transfusions	0.0 (0.0; 1.0)	0.0 (0.0; 0.0)	0.015*
Average number of RBC per transfusion in transfused patients	0.0 (0.0; 1.0)	0.0 (0.0; 0.0)	0.019*
Iron in RBC packs transfused per month in transfused patients (mg/month)	0.0 (0.0; 200.0)	0.0 (0.0; 0.0)	0.016*
Iron total (IV + oral + iron in RBC) (mg/month)	0.0 (0.0; 600.0)	0.0 (0.0; 0.0)	0.025*
Previous variceal bleeding episode	15 (11.9)	3 (6.8)	0.511
Average number of variceal bleeding episodes per month	0.0 (0.0; 0.0)	0.0 (0.0; 0.0)	0.701
Underlying liver disease			0.045*
Alcoholic	15 (11.9)	11 (25.0)
HBV	64 (50.8)	20 (45.5)
HCV	11 (8.7)	2 (4.5)
MASH/MASLD	24 (19.0)	3 (6.8)
Others	12 (9.5)	8 (18.2)
AST (IU/L)	34.0 (25.0; 46.0)	33.5 (27.0; 50.5)	0.621
ALT (IU/L)	24.0 (16.0; 40.0)	23.0 (13.0; 30.5)	0.157
Total bilirubin (mg/dL)	1.0 (0.6; 1.6)	1.5 (0.8; 2.3)	0.019*
γ-GT (IU/L)	39.5 (25.0; 72.0)	37.0 (25.5; 80.5)	0.701
Albumin (g/dL)	4.0 (3.4; 4.5)	3.2 (2.7; 4.3)	0.005**
PT (INR)	1.0 (1.0; 1.2)	1.1 (1.0; 1.4)	0.056
Platelet count (1,000/μL)	124.0 (68.0; 182.0)	92.0 (70.0; 158.0)	0.443
Ascites			0.899
None	104 (82.5)	36 (81.8)
Mild	13 (10.3)	4 (9.1)
Moderate	9 (7.1)	4 (9.1)
Encephalopathy	9 (7.1)	7 (15.9)	0.157
Child-Pugh Class			0.008**
A	87 (69.0)	19 (43.2)
B	33 (26.2)	20 (45.5)
C	6 (4.8)	5 (11.4)
Child-Pugh score	5.0 (5.0; 7.0)	7.0 (5.0; 8.5)	0.006**
MELD score	8.0 (7.0; 11.0)	10.0 (7.0; 13.5)	0.024*
Liver iron concentration (mg/g dry weight)	1.4 (1.3; 1.7)	1.5 (1.3; 2.2)	0.014*
Liver iron concentration (µmol/g dry weight)†	25.1 (23.3; 30.4)	26.9 (23.3; 39.4)	0.014*
Liver stiffness (kPa)	4.5 (3.0; 6.7)	4.0 (3.2; 10.6)	0.467
Liver fat fraction (%)	2.4 (1.6; 6.3)	2.6 (1.6; 3.9)	0.618
Pancreas R2* (1/s)	27.1 (22.8; 31.4)	29.3 (28.0; 38.8)	0.001**
Pancreas fat fraction (%)	2.2 (0.9; 4.6)	2.1 (0.8; 4.0)	0.395

Data are presented as n (%) or median (25^th; 75^th percentiles). ^†, values for LIC are also provided in μmol/g dry weight. *, P<0.05; **, P<0.01. ALT, alanine transaminase; AST, aspartate transaminase; BMI, body mass index; γ-GT, gamma-glutamyl transferase; HBV, hepatitis B virus; HCV, hepatitis C virus; INR, international normalized ratio; IU, international unit; IV, intravenous; LIC, liver iron concentration; MASH, metabolic dysfunction-associated steatohepatitis; MASLD, metabolic dysfunction-associated steatotic liver disease; MELD, Model for End-stage Liver Disease; MRI, magnetic resonance imaging; PT, prothrombin time; RBC, red blood cell.

Subgroup analysis by liver disease etiology showed that pancreatic R2* differed significantly between groups only among patients with alcoholic liver cirrhosis (P=0.001, Figure S2). In overall cohort, liver stiffness, liver fat fraction, or pancreatic fat fraction did not differ between iron status groups. Moreover, patients with iron overload had a lower frequency and amount of transfusion, suggesting that systemic iron overload was primarily driven by endogenous rather than transfusional iron accumulation.

Interobserver agreement analysis showed a mean difference of −0.85 (95% CI: −17.76 to 16.07) for liver R2*, 0.04 (95% CI: −0.06 to 0.12) for pancreatic R2*, 0.08 (95% CI: −6.88 to 7.05) for liver fat fraction, and −0.39 (95% CI: −6.76 to 5.97) for pancreatic fat fraction. Bland-Altman plots demonstrated that the majority of observations lay within the 95% limits of agreement, supporting acceptable interobserver agreement (Figure S3).

Correlations between pancreatic R2* values and laboratory parameters and MRE-derived measures

In the overall cohort, pancreatic R2* values significantly correlated with indices of systemic iron status, showing a positive correlation with serum ferritin (r=0.169, P=0.028) and a negative correlation with TIBC (r=−0.194, P=0.018) (Table 2). With respect to liver function, pancreatic R2* values correlated positively with the Child-Pugh scores (r=0.163, P=0.034) and MELD scores (r=0.186, P=0.015), and with markers of impaired liver synthetic function, including lower albumin (r=−0.172, P=0.025) and prolonged PT (r=0.151; P=0.049). Among metabolic parameters, pancreatic R2* showed a negative correlation with serum total cholesterol (r=−0.213, P=0.029) and a positive correlation with pancreatic fat fraction (r=0.260; P<0.001).

Table 2. Correlations between pancreas R2* values and clinical, laboratory parameters, and MRE-derived measures.

Variables	All patients						Normal iron						Iron overload
	Spearman correlation analysis			Partial Spearman correlation analysis†			Spearman correlation analysis			Partial Spearman correlation analysis†			Spearman correlation analysis			Partial Spearman correlation analysis†
	r (95% CI)	P value		r (95% CI)	P value		r (95% CI)	P value		r (95% CI)	P value		r (95% CI)	P value		r (95% CI)	P value
Age (years)	−0.100 (−0.254 to 0.052)	0.196		−0.022 (−0.131 to 0.076)	0.682		−0.105 (−0.284 to 0.064)	0.243		−0.117 (−0.294 to 0.055)	0.193		−0.110 (−0.41 to 0.185)	0.476		−0.068 (−0.38 to 0.252)	0.667
BMI (kg/m2)	0.147 (−0.005 to 0.280)	0.056		0.163 (0.059 to 0.272)	0.002**		0.013 (−0.160 to 0.181)	0.888		0.018 (−0.153 to 0.202)	0.843		0.395 (0.100 to 0.634)	0.008**		0.386 (0.095 to 0.638)	0.011*
Serum ferritin (ng/mL)	0.169 (0.013 to 0.317)	0.028*		0.232 (0.103 to 0.357)	<0.001***		−0.024 (−0.206 to 0.159)	0.792		−0.025 (−0.214 to 0.166)	0.785		0.184 (0.018 to 0.384)	0.012*		0.175 (0.016 to 0.389)	0.014*
Serum iron (μg/dL)	0.054 (−0.108 to 0.224)	0.513		0.074 (−0.061 to 0.218)	0.293		−0.105 (−0.286 to 0.075)	0.279		−0.117 (−0.297 to 0.07)	0.226		0.316 (−0.002 to 0.598)	0.050		0.224 (−0.112 to 0.539)	0.176
Total iron-binding capacity (mg/L)	−0.194 (−0.365 to −0.031)	0.018*		−0.226 (−0.349 to −0.084)	0.001**		−0.074 (−0.278 to 0.136)	0.445		−0.04 (−0.245 to 0.175)	0.68		−0.184 (−0.344 to −0.022)	0.015*		−0.175 (−0.369 to −0.015)	0.018*
Transferrin saturation (%)	0.150 (−0.010 to 0.313)	0.069		0.183 (0.042 to 0.324)	0.009**		−0.028 (−0.213 to 0.159)	0.769		−0.048 (−0.231 to 0.157)	0.621		0.297 (0.017 to 0.541)	0.066		0.266 (−0.038 to 0.527)	0.106
C-reactive protein (mg/L)	−0.008 (−0.157 to 0.142)	0.916		0.054 (−0.056 to 0.158)	0.313		−0.145 (−0.317 to 0.043)	0.105		−0.125 (−0.288 to 0.052)	0.163		0.265 (−0.025 to 0.517)	0.083		0.259 (−0.025 to 0.52)	0.093
AST (IU/L)	0.007 (−0.161 to 0.159)	0.931		0.141 (0.020 to 0.248)	0.009**		−0.037 (−0.212 to 0.16)	0.683		−0.072 (−0.248 to 0.117)	0.422		0.118 (−0.240 to 0.454)	0.447		0.154 (−0.234 to 0.521)	0.324
ALT (IU/L)	−0.099 (−0.254 to 0.051)	0.201		−0.017 (−0.121 to 0.085)	0.758		−0.049 (−0.213 to 0.115)	0.584		−0.111 (−0.264 to 0.043)	0.217		−0.141 (−0.438 to 0.198)	0.362		−0.123 (−0.432 to 0.235)	0.431
Total bilirubin (mg/dL)	0.133 (−0.036 to 0.287)	0.085		0.160 (0.048 to 0.266)	0.003**		−0.043 (−0.225 to 0.126)	0.636		−0.023 (−0.202 to 0.137)	0.796		0.450 (0.181 to 0.666)	0.002**		0.435 (0.158 to 0.672)	0.004**
γ-GT (IU/L)	0.042 (−0.100 to 0.193)	0.586		0.105 (−0.004 to 0.206)	0.051		0.098 (−0.085 to 0.284)	0.275		0.139 (−0.048 to 0.317)	0.123		−0.187 (−0.486 to 0.102)	0.225		−0.146 (−0.465 to 0.169)	0.351
Albumin (g/dL)	−0.172 (−0.320 to −0.023)	0.025*		−0.157 (−0.263 to −0.046)	0.003**		−0.034 (−0.21 to 0.149)	0.702		−0.072 (−0.25 to 0.103)	0.424		−0.402 (−0.639 to −0.129)	0.007**		−0.432 (−0.656 to −0.169)	0.004**
PT (INR)	0.151 (−0.012 to 0.288)	0.049*		0.133 (0.020 to 0.239)	0.013**		0.016 (−0.174 to 0.206)	0.858		0.024 (−0.168 to 0.216)	0.794		0.360 (0.086 to 0.584)	0.016*		0.354 (0.069 to 0.593)	0.019*
Platelet count (1,000/mL)	−0.094 (−0.232 to 0.063)	0.225		−0.116 (−0.224 to −0.005)	0.03*		0.001 (−0.158 to 0.174)	0.995		−0.026 (−0.193 to 0.147)	0.773		−0.368 (−0.583 to −0.087)	0.014*		−0.363 (−0.601 to −0.076)	0.017*
Child-Pugh score	0.163 (−0.003 to 0.327)	0.034*		0.185 (0.072 to 0.295)	<0.001***		0.041 (−0.135 to 0.224)	0.649		0.066 (−0.122 to 0.254)	0.465		0.319 (0.021 to 0.591)	0.035*		0.371 (0.039 to 0.645)	0.014*
MELD score	0.186 (0.035 to 0.326)	0.015*		0.217 (0.102 to 0.323)	<0.001***		0.047 (−0.134 to 0.229)	0.604		0.069 (−0.102 to 0.24)	0.445		0.397 (0.125 to 0.625)	0.007**		0.383 (0.098 to 0.622)	0.011*
Liver stiffness (kPa)	0.008 (−0.15 to 0.144)	0.914		0.054 (−0.045 to 0.166)	0.31		−0.046 (−0.21 to 0.129)	0.608		0.017 (−0.159 to 0.196)	0.851		0.071 (−0.200 to 0.371)	0.648		0.147 (−0.161 to 0.488)	0.347
Liver fat fraction (%)	0.019 (−0.136 to 0.171)	0.802		0.025 (−0.079 to 0.129)	0.637		0.042 (−0.137 to 0.227)	0.641		−0.043 (−0.22 to 0.132)	0.633		0.079 (−0.176 to 0.347)	0.61		0.077 (−0.228 to 0.381)	0.622
Liver iron concentration (mg/g dry weight)	0.113 (−0.055 to 0.264)	0.143		0.202 (0.062 to 0.332)	0.002**		0.005 (−0.184 to 0.196)	0.952		0.008 (−0.174 to 0.191)	0.931		0.349 (0.016 to 0.601)	0.02*		0.292 (−0.03 to 0.558)	0.058
Number of transfusions	−0.049 (−0.218 to 0.114)	0.524		−0.059 (−0.159 to 0.045)	0.269		−0.075 (−0.261 to 0.118)	0.402		−0.082 (−0.288 to 0.107)	0.364		0.317 (0.014 to 0.544)	0.036*		0.25 (−0.05 to 0.471)	0.106
Iron total (IV + oral + iron in RBC) (mg/month)	−0.043 (−0.218 to 0.115)	0.581		−0.059 (−0.165 to 0.045)	0.272		−0.078 (−0.27 to 0.093)	0.384		−0.075 (−0.273 to 0.137)	0.406		0.307 (0.049 to 0.507)	0.043*		0.237 (−0.021 to 0.458)	0.126
Average number of variceal bleeding episodes per month	−0.006 (−0.173 to 0.146)	0.943		−0.02 (−0.149 to 0.12)	0.714		−0.038 (−0.225 to 0.164)	0.671		−0.013 (−0.236 to 0.2)	0.888		0.119 (−0.063 to 0.298)	0.442		0.071 (−0.143 to 0.273)	0.652
Serum triglyceride (mg/dL)	0.029 (−0.187 to 0.241)	0.772		−0.002 (−0.142 to 0.138)	0.972		0.074 (−0.175 to 0.317)	0.538		0.119 (−0.141 to 0.367)	0.322		−0.009 (−0.414 to 0.393)	0.962		−0.016 (−0.405 to 0.378)	0.935
Serum total cholesterol (mg/dL)	−0.213 (−0.378 to −0.035)	0.029*		−0.129 (−0.252 to −0.002)	0.06		−0.235 (−0.417 to −0.022)	0.044*		−0.203 (−0.412 to 0.016)	0.086		−0.144 (−0.438 to 0.19)	0.439		−0.187 (−0.523 to 0.202)	0.323
Serum LDL (mg/dL)	−0.113 (−0.316 to 0.100)	0.32		−0.108 (−0.269 to 0.062)	0.202		−0.153 (−0.389 to 0.094)	0.261		−0.141 (−0.377 to 0.114)	0.304		0.041 (−0.382 to 0.444)	0.853		0.054 (−0.346 to 0.474)	0.812
Serum HDL (mg/dL)	−0.182 (−0.361 to 0.021)	0.073		−0.047 (−0.187 to 0.106)	0.516		−0.186 (−0.409 to 0.043)	0.127		−0.191 (−0.415 to 0.047)	0.119		−0.117 (−0.495 to 0.292)	0.546		−0.11 (−0.564 to 0.32)	0.577
Serum lipase (U/L)	0.134 (−0.068 to 0.313)	0.152		0.057 (−0.074 to 0.215)	0.445		0.124 (−0.121 to 0.36)	0.262		0.116 (−0.137 to 0.364)	0.295		0.158 (−0.178 to 0.479)	0.396		0.105 (−0.2 to 0.426)	0.581
Serum amylase (U/L)	−0.007 (−0.214 to 0.199)	0.94		0.003 (−0.143 to 0.146)	0.972		−0.027 (−0.253 to 0.22)	0.807		−0.014 (−0.252 to 0.221)	0.901		0.159 (−0.207 to 0.475)	0.394		0.095 (−0.272 to 0.443)	0.617
Serum HbA1c (%)	−0.208 (−0.438 to 0.038)	0.071		−0.188 (−0.349 to 0)	0.032*		−0.305 (−0.553 to −0.055)	0.023*		−0.277 (−0.527 to 0.014)	0.041*		0.127 (−0.384 to 0.579)	0.593		0.118 (−0.365 to 0.564)	0.629
Pancreas fat fraction (%)	0.260 (0.113 to 0.398)	<0.001***		–	–		0.284 (0.118 to 0.447)	0.001**		–	–		0.251 (−0.073 to 0.541)	0.101		–	–

*, P<0.05; **, P<0.01; ***, P<0.001. ^†, partial Spearman correlation analysis accounting for pancreatic fat fraction. ALT, alanine transaminase; AST, aspartate transaminase; BMI, body mass index; CI, confidence interval; γ-GT, gamma-glutamyl transferase; HbA1c, glycated hemoglobin A1c; HDL, high-density lipoprotein; INR, international normalized ratio; IU, international unit; IV, intravenous; LDL, low-density lipoprotein; LIC, liver iron concentration; MELD, Model for End-stage Liver Disease; MRE, magnetic resonance elastography; PT, prothrombin time; RBC, red blood cell.

Subgroup analysis revealed that most of these associations were driven by the iron overload group. In patients with iron overload, pancreatic R2* values significantly correlated with serum ferritin (r=0.184, P=0.012) and TIBC (r=−0.184, P=0.015), as well as with higher Child-Pugh (r=0.319; P=0.035) and MELD scores (r=0.397; P=0.007), and markers of impaired liver function including higher total bilirubin (r=0.450, P=0.002), lower albumin (r=−0.402, P=0.007), lower platelet count (r=−0.368, P=0.014). Pancreatic R2* also correlated with BMI (r=0.395, P=0.008), number of transfusions (r=0.317; P=0.036), total iron supplementation (r=0.307; P=0.043), and LIC (r=0.349, P=0.020).

Conversely, in patients with normal iron status, pancreatic R2*inversely correlated with total cholesterol (r=−0.235, P=0.044) and HbA1c (r=−0.305; P=0.023), and positively correlated with pancreatic fat fraction (r=0.284; P=0.001). After adjustment for pancreatic fat fraction, pancreatic R2* values did not significantly correlate with any variables in the normal iron group. In contrast, in patients with iron overload, pancreatic R2* values did not significantly correlate with the number of transfusions (P=0.106), total iron supplementation (P=0.126), or LIC (P=0.058) after adjustment; however, correlations with other variables remained significant (all P<0.05).

Iron overload and development of metabolic complications

The 5-year cumulative incidences of dyslipidemia, prediabetes, T2DM and the combined outcome of prediabetes or T2DM did not differ significantly between iron status groups in the overall cohort (Figure 2), among LT recipients (Figure 3), or within the subgroup of patients with alcoholic liver cirrhosis (Figure 4). The median time to diagnosis and results of the Cox proportional hazards analyses are summarized in Table 3. When patients were stratified by pancreatic R2* quartiles, there were no significant increasing trends in the cumulative incidence of metabolic outcomes (Table S2). In the Cox proportional hazards models adjusting for age, sex, BMI, and LIC, no significant associations were observed between pancreatic R2* quartiles and any metabolic outcome (Tables S3,S4). Among LT recipients, there were no significant differences in the type of maintenance immunosuppressive therapy and steroid use after transplantation between the normal iron and iron overload groups (Table S5). Furthermore, even after accounting for these factors, no significant differences were observed in the 5-year cumulative incidences of metabolic complications according to iron status (Table S6).

Figure 2.

Kaplan-Meier curves showing the cumulative incidence of (A) dyslipidemia, (B) prediabetes, (C) T2DM, and (D) prediabetes or T2DM in all patients with liver cirrhosis, stratified by iron status (normal iron vs. iron overload). Solid lines represent cumulative incidence estimates, and dotted lines indicate 95% confidence intervals. T2DM, type 2 diabetes mellitus.

Figure 3.

Kaplan-Meier curves showing the cumulative incidence of (A) dyslipidemia, (B) prediabetes, (C) T2DM, and (D) prediabetes or T2DM in the subgroup of liver transplantation recipients, stratified by iron status (normal iron vs. iron overload). Solid lines represent cumulative incidence estimates, and dotted lines indicate 95% confidence intervals. T2DM, type 2 diabetes mellitus.

Figure 4.

Kaplan-Meier curves showing the cumulative incidence of (A) dyslipidemia, (B) prediabetes, (C) T2DM, and (D) prediabetes or T2DM in the subgroup of patients with alcoholic liver cirrhosis, stratified by iron status (normal iron vs. iron overload). Solid lines represent cumulative incidence estimates, and dotted lines indicate 95% confidence intervals. T2DM, type 2 diabetes mellitus.

Table 3. Kaplan-Meier and Cox-proportional hazards analysis of metabolic outcomes by iron overload status.

Variables	Normal iron	Iron overload	P value
All patients
No. of patients/total patients	126/170 (74.1%)	44/170 (25.9%)
No. of patients who developed outcome
Dyslipidemia	24/126 (19.0%)	5/44 (11.4%)	0.351
Prediabetes	32/126 (25.4%)	11/44 (25.0%)	0.999
T2DM	25/126 (19.8%)	10/44 (22.7%)	0.848
Prediabetes or T2DM	46/126 (36.5%)	18/44 (40.9%)	0.735
Median days to diagnosis (95% CI)
Dyslipidemia	1,570.0 (1,470.0–1,658.1)	1,646.3 (1,493.9–1,759.6)	0.272
Prediabetes	1,431.7 (1,309.0–1,547.4)	1,433.2 (1,208.7–1,630.3)	0.971
T2DM	1,504.9 (1,385.5–1,615.6)	1,465.4 (1,262.2–1,655.5)	0.679
Prediabetes or T2DM	1,241.5 (1,103–1,382.9)	1,178.6 (933.1–1,407.3)	0.595
Hazards ratio (95% CI)
Dyslipidemia	1	0.587 (0.256–1.342)	0.308
Prediabetes	1	0.987 (0.499–1.954)	0.963
T2DM	1	1.167 (0.546–2.495)	0.622
Prediabetes or T2DM	1	1.159 (0.659–2.037)	0.555
Liver transplant recipients
No. of patients/total patients	53/76 (69.7%)	23/76 (30.2%)
No. of patients who developed outcome
Dyslipidemia	23/53 (43.4%)	5/23 (21.7%)	0.124
Prediabetes	20/53 (37.7%)	10/23 (43.5%)	0.830
T2DM	22/53 (41.5%)	10/23 (43.5%)	0.999
Prediabetes or T2DM	31/53 (58.5%)	17/23 (73.9%)	0.307
Median days to diagnosis (95% CI)
Dyslipidemia	1,240.8 (1,039.4–1,434.4)	1,483.1 (1,204.3–1,750.7)	0.107
Prediabetes	1,178.3 (950.0–1,398.9)	1,150.6 (804.5–1,468.9)	0.815
T2DM	1,166.4 (939.6–1,377.2)	1,137.0 (790.6–1,455.4)	0.836
Prediabetes or T2DM	828.6 (600.2–1,057.4)	663.4 (369.0–976.6)	0.347
Hazards ratio (95% CI)
Dyslipidemia	1	0.461 (0.208–1.019)	0.113
Prediabetes	1	1.095 (0.506–2.367)	0.764
T2DM	1	1.082 (0.506–2.311)	0.781
Prediabetes or T2DM	1	1.325 (0.712–2.468)	0.330
Alcoholic liver cirrhosis
No. of patients/total patients	15/26 (57.7%)	11/26 (42.3%)
No. of patients who developed outcome
Dyslipidemia	4/15 (26.7%)	3/11 (27.3%)	0.999
Prediabetes	3/15 (20.0%)	5/11 (45.5%)	0.337
T2DM	5/15 (33.3%)	4/11 (36.4%)	0.999
Prediabetes or T2DM	6/15 (40.0%)	6/11 (54.5%)	0.736
Median days to diagnosis (95% CI)
Dyslipidemia	1,415.9 (1,059.6–1,726.3)	1,399.0 (973.0–1,825.0)	0.971
Prediabetes	1,516.5 (1,159.1–1,825.0)	1,062.5 (579.4–1,529.2)	0.164
T2DM	1,306.5 (901.1–1,651.5)	1,278.4 (838.8–1,698.1)	0.879
Prediabetes or T2DM	1,229.3 (823.6–1,589.3)	935.5 (457.4–1,408.7)	0.373
Hazards ratio (95% CI)
Dyslipidemia	1	1.028 (0.229–4.609)	0.928
Prediabetes	1	2.653 (0.639–11.010)	0.172
T2DM	1	1.108 (0.295–4.163)	0.845
Prediabetes or T2DM	1	1.661 (0.516–5.340)	0.361

CI, confidence interval; No., number; T2DM, type 2 diabetes mellitus.

Discussion

To the best of our knowledge, this is the first study to investigate the relationship between pancreatic R2* values, liver function and systemic iron indices in cirrhotic patients with and without iron overload. All other studies assessing pancreatic R2* were largely limited to patients with transfusion-related iron overload such as those with thalassemia major, or hereditary disorder such as HH.

Unlike these conditions where pancreatic R2* values show a strong positive correlation with liver R2* or LIC (27,28), our study found no significant association between pancreatic R2* and LIC in cirrhotic patients. Instead, pancreatic R2* showed closer correlations with systemic serum iron parameters, particularly ferritin and TIBC. A possible explanation for the observed discordance, in which pancreatic R2* correlated with systemic iron markers whereas liver R2* or LIC did not, may be altered iron trafficking in advanced cirrhosis. Portosystemic shunting may reduce effective first-pass hepatic sequestration of circulating iron, allowing a greater fraction of iron to enter systemic circulation rather than being retained in the liver. Concurrently, impaired hepatocellular function may result in reduced hepcidin synthesis, increasing circulating iron availability, reflected by elevated TSAT. In this setting, extrahepatic organs susceptible to iron deposition, such as the pancreas, may more accurately reflect systemic iron exposure than hepatic iron-weighted metrics such as LIC. Consistent with this hypothesis, patients with iron overload in our cohort showed higher Child-Pugh and MELD scores, along with evidence of impaired liver function, including hyperbilirubinemia, hypoalbuminemia, and prolonged PT. A prior study has similarly reported that iron overload in cirrhosis is associated with hepatic decompensation and reduced survival (5). Additionally, BMI correlated with pancreatic R2* values among patients with iron overload; however, this association likely reflects fluid retention, ascites, and peripheral edema accompanying hepatic decompensation, rather than reflecting true increase in adiposity (29).

Interestingly, among various etiologies, only alcoholic cirrhosis showed significant differences in R2* values between patients with iron overload and those with normal iron status. Chronic alcohol exposure is known to suppress hepcidin expression and enhance intestinal iron absorption, thereby increasing systemic iron availability and promoting parenchymal iron deposition (30-32). In line with this mechanism, the median pancreatic R2* values in patients with alcoholic cirrhosis and iron overload were higher than those observed in other etiologic groups.

In addition to these correlations, the present study investigated whether iron overload status was associated with an increased risk of metabolic complications. Cirrhosis is commonly accompanied by insulin resistance and IGT, which may progress to hepatogenous diabetes in advanced stages (33). These abnormalities arise from decreased hepatic clearance of insulin and increased portosystemic shunt, leading to systemic hyperinsulinemia and progressive pancreatic β-cells dysfunction (33). Both insulin resistance and β-cell dysfunction worsen with greater hepatic decompensation and correlate with higher Child-Pugh class (34-36). In this cohort, approximately 90% of patients had Child-Pugh class A or B cirrhosis with MELD scores <15, representing a population at relatively low risk for hepatogenous diabetes and an appropriate setting to evaluate the independent influence of iron overload.

None of the patients had dyslipidemia, prediabetes, or T2DM at the time of MRE and the occurrence of these disorders was retrospectively assessed during up to 5-year of follow-up. Although associations between elevated pancreatic R2* values and T2DM have been reported in various iron overload conditions, including HH (37), thalassemia (10,11), and sickle cell disease (13), no significant difference in the risk of developing T2DM was observed between cirrhotic patients with and without iron overload in the present cohort, even among patients with alcoholic liver cirrhosis. Similarly, the risk of dyslipidemia, prediabetes and the composite endpoint of prediabetes or T2DM were comparable between the two groups.

In addition, no study to date has evaluated the degree to which pancreatic iron deposition is alleviated after LT in cirrhosis. Prior MRI investigations in patients with systemic iron overload who underwent hematopoietic stem cell transplantation have shown that organ iron accumulation, including the pancreas, often persists after transplantation (38,39). In our subgroup analysis of LT recipients, pre-transplant iron overload was not associated with post-transplant metabolic complications, even after adjustment for maintenance immunosuppression and steroid use. However, subgroup sample sizes were small, and these findings require confirmation in larger cohorts.

This study has several limitations. First, it was a retrospective, single-center analysis, which may have introduced selection bias. About 44% of all cirrhotic patients in this cohort underwent MRE as part of preoperative evaluation for LT; therefore, the findings may not be fully generalizable to the broader population of cirrhotic patients who are not transplant candidates. Second, histologic confirmation of pancreatic iron deposition was not available since pancreatic biopsy was not a routinely performed in cirrhotic patients. Third, because no validated threshold for abnormal pancreatic R2* values exist at 3.0-T MRI, the cohort was dichotomized according to iron overload status defined by serum iron parameters rather than pancreatic or liver R2* measurements. Liver R2* was not used for stratification given the absence of a significant correlation between hepatic and pancreatic R2* values in our cohort. Fourth, R2* values were obtained from the pancreatic body and tail. Although a previous study (28) reported significant regional differences in R2* between the pancreatic head and tail, this variation is likely of limited clinical relevance, as the same study suggested that iron accumulation begins in the tail. Moreover, another study reported no regional differences in pancreatic R2* values in healthy subjects (40). Lastly, we required concurrent elevation of both TSAT and ferritin to minimize misclassification due to reduced TIBC from impaired hepatic transferrin synthesis; however, some residual confounding from this effect cannot be entirely excluded in a cirrhotic cohort.

Conclusions

In this retrospective cohort of cirrhotic patients, pancreatic R2* values measured at 3.0-T MRI correlated more closely with systemic serum iron indices, particularly ferritin and TIBC, than with hepatic iron concentration, a pattern likely reflecting altered iron trafficking in advanced cirrhosis. Notably, this systemic-pancreatic association was most prominent in patients with alcoholic cirrhosis, consistent with the known effects of chronic alcohol exposure on hepcidin suppression. Despite these biologically plausible correlations, iron overload status was not associated with an increased 5-year risk of dyslipidemia, prediabetes, or T2DM, even among liver transplant recipients. These findings suggest that, in cirrhotic patients, pancreatic R2* may serve as a surrogate marker of systemic rather than hepatic iron status, but does not appear to carry independent prognostic value for metabolic complications within the time frame examined. Clinically, our results imply that routine measurement of pancreatic R2* on multiparametric liver MRI is unlikely to refine the metabolic risk stratification of cirrhotic patients beyond conventional iron indices, although larger prospective studies with longer follow-up are warranted to confirm these observations and clarify the role of pancreatic iron deposition.

Supplementary

The article’s supplementary files as

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by Institutional Review Board of Severance Hospital, Seoul, Korea (IRB No. 4-2025-1600). and individual consent for this retrospective analysis was waived due to the nature of the study.

Data Sharing Statement

Available at https://qims.amegroups.com/article/view/10.21037/qims-2026-1-0419/dss

qims-16-08-650-dss.pdf ^{(68.9KB, pdf)}

DOI: 10.21037/qims-2026-1-0419