Pancreatic fatty replacement as risk marker for altered glucose metabolism and cardiac iron and complications in thalassemia major

Objectives This multicenter study assessed the extent of pancreatic fatty replacement and its correlation with demographics, iron overload, glucose metabolism, and cardiac complications in a cohort of well-treated patients with thalassemia major (TM). Methods We considered 308 TM patients (median age: 39.79 years; 182 females) consecutively enrolled in the Extension-Myocardial Iron Overload in Thalassemia Network. Magnetic resonance imaging was used to quantify iron overload (IO) and pancreatic fat fraction (FF) by T2* technique, cardiac function by cine images, and to detect replacement myocardial fibrosis by late gadolinium enhancement technique. The glucose metabolism was assessed by the oral glucose tolerance test. Results Pancreatic FF was associated with age, body mass index, and history of hepatitis C virus infection. Patients with normal glucose metabolism showed a significantly lower pancreatic FF than patients with impaired fasting glucose (p = 0.030), impaired glucose tolerance (p < 0.0001), and diabetes (p < 0.0001). A normal pancreatic FF (< 6.6%) showed a negative predictive value of 100% for abnormal glucose metabolism. A pancreatic FF > 15.33% predicted the presence of abnormal glucose metabolism. Pancreas FF was inversely correlated with global pancreas and heart T2* values. A normal pancreatic FF showed a negative predictive value of 100% for cardiac iron. Pancreatic FF was significantly higher in patients with myocardial fibrosis (p = 0.002). All patients with cardiac complications had fatty replacement, and they showed a significantly higher pancreatic FF than complications-free patients (p = 0.002). Conclusion Pancreatic FF is a risk marker not only for alterations of glucose metabolism, but also for cardiac iron and complications, further supporting the close link between pancreatic and cardiac disease. Key Points • In thalassemia major, pancreatic fatty replacement by MRI is a frequent clinical entity, predicted by a pancreas T2* < 20.81 ms and associated with a higher risk of alterations in glucose metabolism. • In thalassemia major, pancreatic fatty replacement is a strong risk marker for cardiac iron, replacement fibrosis, and complications, highlighting a deep connection between pancreatic and cardiac impairment.


Introduction
The introduction of oral iron chelation therapy and of the T2* magnetic resonance imaging (MRI) technique for the non-invasive quantification of myocardial iron overload (MIO) has significantly increased the life expectancy of patients with thalassemia major (TM), needing lifelong regular blood transfusions [1][2][3].With an aging TM population, diabetes mellitus (DM) is an increasing issue.Age per se is one of the most important risk factors in the development of hyperglycemia, leading to both deficiency of insulin secretion and insulin resistance [4].In TM, these processes are aggravated/accelerated by pancreatic [5] and hepatic [6] IO.
In TM, the prevalence of pancreatic iron is close to 90% and a normal global pancreas T2* value was found to have a negative predictive value of 100% for disturbances of glucose metabolism [7,8].However, pancreatic iron has a low specificity for glucose dysregulation [8], which depends on IO severity and duration.
It has been hypothesized that, after the death of the pancreatic cells caused by the cytotoxic iron effect, a progressive fatty replacement of the pancreatic parenchyma may occur [9].MRI represents the best imaging technique for evaluating the deposition of ectopic fat [10,11].However, pancreatic fatty replacement and its clinical correlations in TM have been little explored.Midiri et al [9] measured the pancreas-to-fat signal intensity ratio (SIR) in T1-weighted, T2-weighted, and T2*-weighted sequences in 20 TM patients and noticed an increased SIR, attributed to a progressive substitution of the parenchyma by inert adipose tissue, in 3 (15%) patients.Papakonstantinou et al [12] used the signal intensity change between in-phase (water + fat) and opposed-phase (water -fat) images, named pancreatic signal index (PSI), as an index for pancreatic fat.A PSI > 20% was considered indicative of pancreatic fatty replacement, diagnosed in 45% of the 31 included TM patients.The fatty replacement was more frequent in diabetic than in non-diabetic patients (77% vs 20%) [12].However, a negative PSI was detected in 9 patients, due to the predominant effect of iron deposition.Indeed, the ability of this technique to quantify fat is corrupted by multiple confounding factors, especially in the presence of low-fat fractions (FF) and high iron levels [13].Huang et al [14] measured pancreatic FF by using the iterative decomposition of water and fat with echo asymmetry and the least-squares estimation algorithm in 40 pediatric TM patients.Pancreatic FF was associated with pancreatic iron and altered glucidic metabolism.
One attractive approach to quantify the FF by MRI is to take advantage of the conventional gradient-echo multiecho T2* images used for iron quantification.The pancreas T2* is obtained from these images by fitting the signal to an appropriate decay model, and the fat generates a sinusoidal signal fluctuation over-imposed to the exponential decay [15].An appropriate fitting model can be used to separate the fat signal from the water contribution [16,17].This approach, introduced and largely used for the assessment of hepatic FF, has been extended to the pancreas [18,19].In a cohort of 71 patients with different iron overload diseases, pancreatic FF was found associated with pancreatic iron and exocrine function [18].
To the best of our knowledge, no data are available about the association between pancreatic FF and MIO in TM patients.Moreover, since pancreatic steatosis may have a causative effect and contribute to the development of diabetes and diabetes was shown to increase the risk for heart failure (HF), hyperkinetic arrhythmias, and myocardial fibrosis independently of MIO [20], a profound link between pancreatic fat and heart disease could be foreseen.
This multicenter study aimed to assess the extent of pancreatic fatty replacement and its correlation with demographics, iron overload, glucose metabolism, and cardiac complications in a cohort of well-treated TM patients.

Study population
We considered 308 β-TM patients (182 females, median age: 39.79 years) consecutively enrolled in the Extension-Myocardial Iron Overload in Thalassemia (E-MIOT) project.The E-MIOT is an Italian network constituted of 66 thalassemia centers and 11 MRI sites, where MRI exams are performed using homogeneous, standardized, and validated procedures [21][22][23].All centers are linked by a shared database, collecting all clinical, laboratory, and instrumental data.
Moreover, with the aim of defining the upper limit of pancreatic FF, we included 20 healthy subjects (10 females, median age: 35.76 years) without pancreatic diseases, alterations of glucidic metabolism, or known conditions/treatments which could affect pancreatic iron content or fat.
The study complied with the Declaration of Helsinki and was approved by the ethical committees of all the MRI sites.All subjects gave written informed consent.

MRI
All patients underwent MRI using conventional clinical 1.5 T scanners from three main vendors.
Five or more axial slices including the whole pancreas [24], a mid-transverse hepatic slice [25], and basal, medium, and apical short-axis views of the left ventricle (LV) [21,26] were acquired with T2* multiecho gradient-echo sequences.T2* image analysis was performed using a custom-written, previously validated software (HIPPO MIOT®) [27].Three regions of interest (ROIs) were manually drawn over the pancreatic head, body, and tail, encompassing parenchymal tissue and avoiding large blood vessels or ducts [15].For each ROI, the mean value of the signal intensity along all echo times was calculated.Each obtained decay curve was fit to a multipeak fat model in order to estimate both T2* and FF [16,18,19].Global pancreatic FF/T2* values were evaluated as the mean of FF/T2* values from the three regions.Hepatic T2* values were calculated in a circular ROI [28] and converted into liver iron concentration (LIC) [29,30].The myocardial T2* distribution was mapped into a 16-segment LV model, according to the American Heart Association standardized segmentation (six equiangular segments in the basal and medium slices and four in the apical slice) [31].For each segment, the mean value of the signal intensity along all the echo times was calculated and the assessed decay curve was fit to the single exponential model.In heavily iron-overloaded hearts, a truncation model was applied to delete the late points with a low signal-to-noise ratio [32].An appropriate correction map was applied to correct for susceptibility artifacts [27].The global heart T2* value was the mean of all segmental values.
Steady-state free precession cine images were acquired in sequential 8-mm short-axis slices from the atrio-ventricular ring to the apex to quantify biventricular function parameters in a standard [33] and reproducible [34] way.The analysis was based on the manual recognition of the endocardial and epicardial borders of the wall, at least in the end-diastolic and end-systolic phases in each slice.The papillary muscles were delineated and were considered myocardial mass rather than part of the blood pool.For the calculation of end-diastolic and end-systolic volumes (EDV and ESV, respectively), no geometric assumption of the ventricle shape was needed.The stroke volume index (SVI) was calculated as the difference between the EDV index (EDVI) and ESV index (ESVI).The ejection fraction (EF) was given by the ratio between the SVI and the EDVI [33].
Short-axis and long-axis late gadolinium enhancement (LGE) images were acquired by a fast gradient-echo inversion recovery sequence 10-18 min after gadobutrol (Gadovist®; Bayer) intravenous administration (0.2 mmol/ kg) to detect replacement myocardial fibrosis [35,36].LGE images were not acquired in patients with renal problems or refusing the contrast medium.LGE was considered present when visualized in two different views [37].

Assessment of glucose metabolism
To assess the disturbances of glucose metabolism, patients not already diagnosed with diabetes performed an oral glucose tolerance test (OGTT) within 3 months from the MRI study at the reference thalassemia center.
Baseline (after overnight fasting) blood assessments of glucose and insulin were performed.Patients were given 1.75 g/kg (maximum dose = 75 g) of glucose solution; glucose and insulin were measured at 60 and 120 min.In patients without known diabetes, we used the homeostasis model assessment of insulin resistance (HOMA-IR) index to assess the insulin resistance [HOMA-IR = (glucose X insulin)/405] [38].

Diagnostic criteria
The upper limit of pancreatic FF in healthy individuals was defined on log-transformed data as mean + 2standard deviations (SD).
A fasting plasma glucose (FPG) < 100 mg/dL and 2-h glucose < 140 mg/dL were considered normal glucose tolerance (NGT).Impaired fasting glucose (IFG) was diagnosed in presence of FPG levels between 100 and 126 mg/ dL.Impaired glucose tolerance (IGT) was defined by 2-h plasma glucose between 140 and 200 mg/dL, with a FPG < 126 mg/dL.DM was defined by FPG ≥ 126 mg/dL or 2-h glucose ≥ 200 mg/dL during an OGTT or random plasma glucose ≥ 200 mg/dL with classic symptoms of hyperglycemia [41].
The metabolic syndrome was defined by the presence of at least 3 of the following criteria: (1) waist circumference ≥ 102 cm in men or ≥ 88 cm in women; (2) high-density lipoprotein < 40 mg/dL in men and < 50 mg/dL in women or on drug treatment; (3) triglycerides ≥ 150 mg/dL or on drug treatment; (4) systolic blood pressure ≥ 130 mmHg or diastolic blood pressure ≥ 85 mmHg or on anti-hypertensive medication; (5) FPG ≥ 100 mg/dL or on treatment for diabetes [42].
HF was diagnosed by clinicians based on symptoms, signs, and instrumental findings according to the AHA/ACC guidelines [43].Arrhythmias were diagnosed only if ECG-documented and requiring specific medication [44].Pulmonary hypertension (PH) was diagnosed if the trans-tricuspidal velocity jet was > 3.2 m/s [45].The term "cardiac complications" included HF, arrhythmias, and PH clinically active at the time of the MRI.
Due to the non-normal distribution, continuous variables were represented with median and 25th and 75th percentiles.Categorical variables were expressed as frequencies and percentages.
The Friedman test was used to evaluate whether FF was different among the three pancreatic regions.Comparisons between groups were made by Wilcoxon's signed rank test (for 2 groups) or the Kruskal-Wallis test (for more than 2 groups).The Bonferroni post hoc test was used for multiple comparisons between pairs of groups.Correlation analysis was performed by using Spearman's test.
The χ 2 test was used for the comparison of non-continuous variables.
To determine the best cutoff for discriminating the presence of a specific condition, the maximum sum of sensitivity and specificity was calculated from receiver-operating characteristic (ROC) curve analysis.
Univariate and stepwise multivariate regression analyses were performed to identify determinants of global pancreas FF.Multivariate regression was performed using only variables with a p value < 0.05 in univariate regression analyses.The collinearity of variables tested in the multivariate model was assessed using the variance inflation factor (inflated if > 5) and its tolerance statistic (inflated if < 0.20).
A 2-tailed p < 0.05 was considered statistically significant.
The global pancreatic FF in healthy subjects was 1.85 (1.02-3.26)%, and the upper limit of the pancreatic FF was 6.6%.
The global pancreatic FF was significantly correlated with body mass index (R = 0.838; p < 0.0001), but it was not associated with age (R = − 0.030; p = 0.900).

Pancreatic FF in TM: distribution and correlation with demographics
The patient's demographic and clinical characteristics are summarized in Table

Pancreatic FF and hepatitis C virus (HCV) infection
On the basis of the presence of HCV antibodies and ribonucleic acid, a categorization into three groups was performed: negative patients (group 0; 39.2%), patients who eradicated the virus spontaneously or after treatment with antiviral therapy (group 1; 55.9%), and patients with chronic HCV infection (group 2; 4.9%).Patients in group 0 were significantly younger than patients in group 1 [29.17(18.The frequency of diabetes was significantly lower in group 0 versus both group 1 (6.3% vs 20.6%; p = 0.003) and group 2 (6.3% vs 21.4%; p = 0.048).

Predictors of global pancreas FF
Past or active HCV infection, serum ferritin levels, body mass index, and pancreatic iron levels were the strongest predictors of global pancreas FF (F = 51.68;p < 0.0001) (Table 2).No variable was excluded from the multivariable model due to excessive collinearity (tolerance statistic < 0.20 and/or variance inf lation factor > 5).

Pancreatic FF and glucose metabolism
Thirty-nine patients were already diagnosed with diabetes, and 261 were tested for blood glucose.8.3% of the patients showed IFG, 6.3% IGT, and 16.0%DM.
Patients with normal glucose metabolism showed a significantly lower global pancreas FF than patients with IFG [20.27 (5.2A).
The 74.4% of the patients with a normal glucose metabolism had fatty replacement.All patients with altered glucose metabolism had pancreatic fatty replacement and IO.A normal global pancreas FF showed a negative predictive value of 100% for altered glucose metabolism.

Pancreatic FF and iron overload in other organs
No significant correlation was found between global pancreatic FF and LIC values, but the 138 (44.8%) patients with hepatic IO showed a significantly higher global pancreatic FF than patients without hepatic IO [26.95 (15.97-39.42)% vs 23.95 (6.99-34.72)%; p = 0.018].
The global pancreas FF showed a negative correlation with global heart T2* values (R = − 0.249; p = 0.009) and a significant positive correlation with the number of segments with pathological T2* (R = 0.207; p < 0.0001).All 24 (7.8%) patients with significant MIO had pancreatic fatty replacement.A normal global pancreas FF showed a negative predictive value of 100% for MIO.Patients with significant MIO showed a significantly higher global pancreatic FF than patients without significant MIO [42.13 (27.36-55.88)% vs 24.37 (9.07-35.54%;p < 0.0001] (Fig. 3A).

Pancreatic FF and cardiac function
No correlation was detected between global pancreatic FF and left (R = − 0.021; p = 0.713) or right (R = − 0.079; p = 0.179) ventricular ejection fractions.

Discussion
The clinical significance of pancreatic fat replacement has gained considerable attention in recent years, and we explored the extent of pancreatic fatty replacement and its clinical correlates in well-treated TM patients.The obtained upper limit of pancreatic FF is in close agreement with that one (6.2%)recommended by a recent meta-analysis including 9 studies, for a total of 1209 healthy individuals studied by MRI, although with different methods [46].
Eighty-two percent of our TM patients had a pancreatic fatty replacement, and this impressive prevalence is in line with the Pfifer's study [18].Moreover, as in the above-mentioned study, we did not detect significant differences among the regional pancreatic FF values.
We found a weak association between pancreatic fatty replacement and aging and we confirmed that fat accumulation could occur also in patients < 10 years of age [14].As expected, patients with metabolic syndrome showed a significantly higher pancreatic FF.Our finding about the increased pancreatic FF among splenectomized TM patients could be explained by the fact that the spleen acts as storage for non-toxic iron [47] and lipids [48].Midiri et al [9] did not find a correlation between pancreatic fatty replacement and age or history of splenectomy, Fig. 2 A Global pancreas FF in the four groups of patients identified on the basis of the alterations of glucose metabolism.The horizontal dotted line represents the upper limit of normal for global pancreas FF.B ROC curve analysis of global pancreas FF to predict the alterations of glucose metabolism Fig. 3 A Global pancreas FF in patients without and with myocardial iron overload.B Global pancreas FF in patients without and with replacement myocardial fibrosis.In both graphs, the horizontal dotted line represents the upper limit of normal for global pancreas FF likely due to the use of a semi-quantitative approach and the significantly smaller study population [9].
As previously shown in smaller cohorts of adults [18] and pediatric TM patients [14], pancreatic fatty replacement was correlated with pancreatic IO.All patients with fatty replacement had also pancreatic iron overload, confirming the hypothesis of the substitution of pancreatic cells, died because of the cytotoxic effects of iron, with adipose tissue [9].A global pancreas T2* < 20.81 ms predicted the presence of fatty replacement with good sensitivity and specificity.This is the first study showing an association between HCV infection and higher levels of fat accumulation in the pancreas.HCV can be present in human pancreatic β-cells [49] and it can lead to acinar cell apoptosis and irreversible replacement by adipocytes both directly [50] and indirectly through the increase in pancreatic iron load [8].
In non-diabetic patients, pancreatic FF was significantly associated with plasma glucose levels but not with HOMA-IR.In TM patients, insulin resistance is caused by hepatic IO that interferes with the suppression of hepatic glucose production from insulin, and by iron deposition in the muscle that decreases the glucose uptake [6], while a reduced insulin secretion can be present also in normoglycemic patients [51].Moreover, the use of indices based on fasting glucose and insulin concentrations has its limitations [38].Patients with normal glucose metabolism showed a significantly lower global pancreas FF than patients with IFG, IGT, and DM; a normal global pancreas FF value showed a negative predictive value of 100% for disturbances of glucose metabolism.We introduced a cutoff of 15% for the prediction of abnormal glucose metabolism, which matches well with the cutoff of 18% found in TM pediatric patients for the discrimination between IFG and normal glucose function [14].The low specificity of our cutoff can be the consequence of a latency time between the fatty replacement of pancreatic parenchyma and the overt disease.Pancreatic FF resulted superior to pancreatic iron in discriminating patients with altered glucose metabolism.This is the first study demonstrating a significant association between MIO and pancreatic FF and revealing that, in addition to normal global pancreas T2* values [7,8,52], also a normal FF had a 100% negative predictive value for significant MIO.
Pancreatic FF was not associated with cardiac function, which can be impaired by different causes besides iron [34], but it was increased in patients with replacement myocardial fibrosis.This finding further reinforces the crucial role of HCV infection in cardiac impairment, by causing myocardial fibrosis both directly [35,53] and indirectly through the development of DM [20,54].As myocardial fibrosis was shown to be the strongest CMR predictor and DM the strongest clinical predictor for HF and cardiac complications in TM [36], it is not surprising, although never demonstrated before, that pancreatic fatty replacement was associated with the development of cardiac complications.

Limitations
We did not perform MR spectroscopy, which is generally considered the gold standard for non-invasive fat and metabolite quantification [46].
Due to its high risk, no histological confirmation was performed for iron and fat content measurements by MRI.

Conclusions
In TM, pancreatic fatty replacement is a frequent clinical entity, predicted by a pancreatic T2* < 20.81 ms and associated with the risk of developing alterations in glucose Fig. 4 A Global pancreas FF in patients without and with cardiac complications.The horizontal dotted line represents the upper limit of normal for global pancreas FF.B ROC curve analysis of global pancreas FF to predict the presence of cardiac complications metabolism.Pancreatic fatty replacement is a strong risk marker for cardiac iron, replacement fibrosis, and complications, highlighting a deep connection between pancreatic and cardiac impairment.Since pancreatic FF can be easily obtained by the same T2* sequence employed for iron overload assessment, it should be included in the routine MRI assessment of TM patients for early identification of patients at high risk for glucose dysregulation and cardiac damage.

Table 1
Demographic, clinical, and MRI data of the patients HCV hepatitis C virus, N number, FF fat fraction, MRI magnetic resonance imaging, LIC liver iron concentration

Table 2
Univariate and multivariate linear regression analysis for the prediction of the global pancreas FF