Serum levels of ferritin and transferrin serve as prognostic factors for mortality and survival in patients with end-stage liver disease: A propensity score-matched cohort study

Abstract

Background

Patients with end-stage liver disease are known to suffer from a significantly high risk of mortality, but accurate prediction of the course of disease is challenging.

Objective

The study aim was to evaluate the independent prognostic and clinical importance of serum levels of ferritin and transferrin for 90-day survival of patients with liver disease.

Methods

Patients with end-stage liver disease treated during a 2-year period were enrolled retrospectively in a single-centre study. Unmatched and propensity score matching (PSM) analyses were applied.

Results

The study cohort comprised 286 patients with end-stage liver disease, of which 22.9% died during the observational period. High serum ferritin levels and low serum transferrin levels were associated significantly with increased 90-day mortality in the unmatched (p < 0.001) and PSM study population (p = 0.017). Serum levels of ferritin and transferrin had high prognostic capability to predict 90-day survival similar to the Model for End-stage Liver Disease. Patients with serum ferritin values >1030.5 µg/l had a 50% risk of dying within 11 days after measurement, which translated up to a 90-day mortality of 83%.

Conclusion

Serum levels of ferritin and transferrin have independent and excellent capabilities to determine prognosis in patients with end-stage liver disease. Ferritin measurements can reliably identify those with high mortality in daily practice.

Keywords

Serum ferritin serum transferrin 90-day survival end-stage liver disease prognostic markers

Introduction

Patients with end-stage liver disease (ESLD) show increased morbidity and mortality, and up to 50% of subjects with decompensated cirrhosis die within 2 years.¹ Subsequently, biological markers and scoring systems have been developed to assess the prognosis of such patients. However, even the most routinely applied scores suffer from significant inaccuracy to predict survival in up to 20% of cases.^2–4

Serum ferritin and transferrin are widely available parameters known to predict survival in patients with various types of liver diseases,^5–7 including patients undergoing liver transplantation.⁸ Although the release mechanism of ferritin is incompletely understood, it has been postulated that it leaks from damaged hepatocytes.¹

Ferritin is an acute-phase protein and elevated serum levels are associated with a wide range of clinical conditions including diabetes mellitus,^9,10 chronic kidney disease,¹¹ inflammation^12,13 and malignancy.¹⁴ Patients suffering from liver diseases often present multiple of these factors. Whether serum levels of ferritin and transferrin can be used independently to predict survival and mortality in patients with ESLD if propensity score matching (PSM)¹⁵ is applied is unknown.

The Model for End-stage Liver Disease (MELD) score is the preferred tool to predict 3-month survival in patients with ESLD.⁴ Several studies have evaluated the possibility for additions to enhance the predictive capability of the MELD score,¹⁶ including levels of sodium¹⁷ and albumin.¹⁸ However, no studies have evaluated whether serum levels of ferritin or transferrin can improve the predictability of the MELD score.¹⁹ Furthermore, the MELD score is based on several parameters and must be calculated, which might limit its frequent use as a screening tool in clinical practice.

To address this lack of evidence, we pursued two goals. The primary endpoint was the evaluation of the independent prognostic value of serum levels of ferritin and transferrin to predict 90-day mortality in patients with ESLD, using PSM. The secondary endpoint was to assess the addition of serum ferritin and transferrin to the MELD score to improve its predictive capability for ESLD patients.

Methods

Study design

This retrospective study was carried out at the Department of Medicine B for Gastroenterology and Hepatology of the University Hospital Muenster (Muenster, Germany). The study protocol was approved by the Ethics Committee of the Westphalian Wilhelms-University of Muenster and the Medical Council of Westphalia-Lippe, Germany (approval: April 2016); informed patient consent was not required due to the retrospective study design. This study conformed to the 1975 Helsinki Declaration as reflected in prior approval by our institution’s Human Research Committee.

Data collection

Data of all patients aged ≥18 years hospitalized for ESLD from 1 January 2016 to 1 December 2017 were retrieved from the electronic medical records (EMRs) of University Hospital Muenster. ESLD patients included patients with decompensated cirrhosis and acute-on-chronic liver failure. Only patients in whom the serum ferritin level was determined at least once during hospitalization were included. Patients who had undergone solid-organ transplantation before and during the observational period were excluded from analyses.

Study endpoint

The primary endpoint of our study was to ascertain the prognostic capability of serum levels of ferritin and transferrin for 90-day survival of patients with ESLD. For survival analyses, the study population was subdivided into groups of short-term (≤90 days) and long-term (>90 days) survivors. Overall survival was documented by the treating physician during hospitalization and/or by routine follow-ups undertaken by our aftercare service, which noted a patient’s death in the EMR. ‘Overall survival’ was defined as the time between the highest level of serum ferritin recorded up to the patient’s death. Demographic, clinical and biochemical markers were recorded.

Clinical parameters were recorded as dichotomized variables to allow for precise matches using PSM. For collection of biochemical data, the highest level of serum ferritin recorded during hospitalization was documented. All other biochemical data were recorded within 24 h of the highest serum level of ferritin found in the patient’s EMR to ensure valid correlation with other biochemical and clinical parameters.

Statistical analyses

Statistical distribution of variables was assessed using the Shapiro–Wilk test. Continuous variables were evaluated using Student’s t-test and Mann–Whitney U test. For categorical variables, the chi-square test or Fisher’s exact tests was applied. Receiver operating characteristic (ROC) curves for 90-day survival were created for variables of interest, and the optimal cutoff points were determined using Youden’s index. To predict survival, binary logistic regression models were created. For survival analyses, Kaplan–Meier curves and the log-rank test were used.

Cox proportional hazard models using the original parameters and formula of the MELD score were performed, with separate additions of serum levels of ferritin and transferrin. The hazard ratios (HRs) of serum ferritin/transferrin, calculated in these models, were used as multipliers for log-transformed serum ferritin/transferrin values. These newly created parameters were added to the original MELD score⁴ formula and ROC curves were created based on these models. The latter were compared with each other using the DeLong method.²⁰

Cases and controls were matched using PSM.¹⁵ Matching parameters included aetiology of liver disease, diabetes mellitus and inflammation (defined as presence of clinical signs and increased levels of inflammatory markers).^10,13 The matching tolerance was 0.05 with a matching ratio of 1:1.

Results are stated as mean ± standard deviation or median (1^stquartile–3^rdquartile) depending on variable distribution.

HRs were calculated for 95% confidence intervals and p < 0.05 was considered statistically significant. Statistical analyses were performed using SPSS v.25.0 (IBM, Armonk, NY, USA) and R v3.5.3 (R Foundation for Statistical Computing, Vienna, Austria).

Results

Study population

A total of 487 patients were treated for liver disease during the 2-year recruitment period. Of those 487 patients, 190 patients were excluded from analysis due to absence of serum ferritin measurement during hospitalization, and 11 who had undergone solid-organ transplantation were also excluded.

Finally, 279 patients were included in the analysis: 59.5% were male and 40.5% female. The median age of patients was 58 (49–65) years. The most frequent reason for hospital admission was an ascitic decompensation (35.6%), an underlying infection (33%), followed by hepatic encephalopathy (19.3%) and hepatorenal syndrome (15.4%). In total, 64 (22.9%) patients died during the study period. Of those, 36 died within ≤ 90 days. The median serum ferritin of the study population was 222 (73–632) µg/l. The median transferrin was 179 (120–256.8) mg/dl (Table 1). Patients with an active alcohol abuse had a significantly increased serum ferritin ( p = 0.013, Supplemental Figure 1).

Table 1.

Overview of the demographic characteristics and clinical and laboratory data of the patients with end-stage liver disease (n = 286).

Parameter	Av. Data	Result
Demographic
Age (years)^a	279	58 (49–65)
Gender (male/female)	279	166 (59.5%)/113 (40.5%)
Survivors/Non-Survivors	279	215 (77.1%)/64 (22.9%)
Aetiology of liver disease
Alcohol abuse	279	84
NASH/ASH/NAFLD	279	30/6/6
Cryptogen	279	32
Hepatitis B/C/E	279	5/24/1
PSC/PBC	279	18/7
Drug related	279	6
Autoimmune hepatitis	279	8
Secondary sclerosing cholangitis	279	8
Others	279	44
Clinical conditions
AD/ACLF	279	252/27
Inflammation (pos./neg.)	278	69/209
Ascites (pos./neg.)	278	73/205
Hepatic encephalopathy (pos./neg.)	278	45/233
Hepatorenal syndrome (pos./neg.)	278	37/241
Anaemia (pos./neg.)	279	62/217
Comorbidities
Coronary heart disease (pos./neg.)	278	20/258
Hypertension (pos./neg.)	279	130/149
Diabetes (pos./neg.)	279	71/208
Chronic kidney disease (pos./neg.)	279	65/214
Hepatocellular cancer (pos./neg.)	279	42/237
Alcohol abuse (neg./active/history of)	271	171/30/70
Scores predicting survival
CHILD A/B/C	244	91/103/50
MELD-Score^a	249	14 (9–20)
AD-Score^a	203	47.3 (4.6–55.2)
Biomedical characteristics
WBC (10⁹/l)^a	275	6 (4.4–8.3)
Sodium^a	271	139 (136–142)
Albumin (g/dl)^a	182	3.4 (2.8–4.1)
Ferritin (µg/l)^a	279	222 (73–632)
Transferrin (mg/dl)^a	250	179 (120–256.8)
Transferrin saturation (in %)^a	246	32 (18.3–59)
Iron (µg/dl)^a	267	77 (50.5–119.5)
CRP (mg/dl)^a	185	2 (0.7–4)
AST (U/l)^a	243	57 (37–115.5)
ALT (U/l)^a	249	42 (26–87)
GGT (U/l)^a	249	120 (51–278)
Creatinine (mg/dl)^a	245	0.8 (0.8–1.4)
Bilirubin (mg/dl)^a	255	1.6 (0.7–3.7)
INR^a	251	1.26 (1.1–1.6)

Median (1^st–3^rd quartile).

NASH: Non-alcoholic steatohepatitis; ASH: alcoholic steatohepatitis; NAFLD: non-alcoholic fatty liver disease; PSC: primary sclerosing cholangitis; PBC: primary biliary cholangitis; AD: acute decompensation; ACLF: acute-on-chronic liver failure.

Comparison of clinical and laboratory characteristics of the unmatched study population based on 90-day survival

Significant differences concerning clinical conditions directly related to liver disease, such as hepatorenal syndrome (p < 0.001), inflammation (p < 0.001), ascites (p = 0.024) and hepatic encephalopathy (p < 0.001) were noted. Furthermore, comorbidities such as coronary heart disease (p = 0.018) and chronic kidney injury (p < 0.001) showed significant differences between short- and long-term survivors. Similarly, scoring systems such as the Child–Pugh score (p < 0.001), the MELD score (p < 0.001) and the AD-Score²¹ (p < 0.001) differed as well (Supplemental Table 1).

Short-term survivors had significantly higher serum levels of ferritin (861 (244.3–1876.5) vs. 190 (63.5–456) µg/l, p < 0.001) and lower serum levels of transferrin (102.1 (±46.4) vs. 198.5 (137.5–262) mg/dl, p < 0.001) (Figure 1(a, b). Apart from 90-day survival, patients with hyperferritinaemia (>400 µg/l) had a significant shortening of survival (p = 0.02; Supplemental Figure 2 and Supplemental Table 3).

Figure 1.

Serum levels of ferritin and transferrin in propensity score-matched patients with end-stage liver disease (ESLD) based on 90-day survival (A/B) and Kaplan–Meier analyses of 90-day survival of the entire study population based on PSM cutoffs of optimal serum levels of ferritin and transferrin (C/D). (a) Serum levels of ferritin were significantly higher in the short-term survival group (≤90 days) compared with the long-term survival group (>90 days, p < 0.001). The Y-axis is limited to a value of 5000 to allow for better visualization. (b) Serum levels of transferrin were significantly lower in the group of patients with short-term survival (≤90 days) compared with patients with long-term survival ( > 90-day survival; p < 0.001). (c) and (d) Kaplan–Meier analysis revealed that the optimal cutoff values (transferrin = 177 mg/dl and ferritin = 1030.5 µg/l) significantly differentiated short-term (≤90 days) survivors from long-term (>90 days) survivors (p < 0.001). Coloured areas represent the confidence interval and vertical lines indicate 50% median survival.

Clinical and laboratory characteristics of the study population based on 90-day survival using PSM

After performing PSM, the presence of coronary heart disease (p = 0.047) remained significantly different between short-term (≤90 days) and long-term (>90 days) survivors. Comparable to the previous unmatched analysis, serum levels of ferritin (861 (244.3–1876.5) vs. 251.5 (85.8–995.3) µg/l, p = 0.017) differed significantly between groups. Short-term survivors had significantly higher MELD scores (26.2 (±8.6) vs. 20 (±7.5), p = 0.003; Supplemental Table 2).

Prognostic capability of ferritin and transferrin serum levels to predict 90-day survival in patients with ESLD

The prognostic capability of serum levels of ferritin and transferrin to predict 90-day survival of patients with ESLD was assessed in unmatched and PSM populations. Serum ferritin and transferrin levels differed significantly between short-term (≤90 days) and long-term (>90 days) survivors for the unmatched (p < 0.001) and PSM populations (p = 0.017). ROC curves were created and optimal cutoff values calculated. For ferritin, the cutoff value was 754 µg/l in the unmatched and 1030.5 µg/l in the PSM group. For transferrin, the cutoff value was 141.5 mg/dl in the unmatched and 177 mg/dl in the PSM group (Table 2).

Table 2.

Binary logistic regression models for the 90-day survival of patients with end-stage liver disease in the unmatched and the propensity score-matched study population.

Binary logistic regression models for 90-day survival	Unmatched analysis			Propensity score-matched analysis
Binary logistic regression models for 90-day survival	AUROC	p-value	Youden index	AUROC	p-value	Youden index
Ferritin (µg/l)	0.75 (0.65–0.84)	0.001	754	0.66 (0.54–0.79)	0.043	1030.5
Transferrin (mg/dl)	0.84 (0.78–0.90)	<0.001	141.5	0.61 (0.52–0.75)	0.041	177
MELD score	0.87 (0.8–0.93)	<0.001	16.5	0.71 (0.58–0.83)	<0.001	27.5

AUROC: Area under receiver operating characteristics.

Kaplan–Meier survival analyses were done for the unmatched study population using these optimal PSM cutoffs. Differences between survival groups proved to be significant (p < 0.001). Most strikingly, 50% of all patients who had serum ferritin levels of ≥1030.5 µg/l died within a median of 11 days. In addition, patients with serum ferritin levels ≥1030.5 µg/l had an 83% 90-day mortality, which exceeded the mortality of a MELD score of 40 points. A serum transferrin level of ≤ 177 mg/dl differed significantly between survival groups, and 50% of patients with this level died within a median of 28 days. (Figure 1(c, d)).

Serum levels of ferritin and transferrin enabled prediction of 90-day survival in binary logistic regression models using PSM and unmatched study populations. The Area Under the Receiver Operating Characteristics (AUROC) of these models was comparable to the AUROC of the MELD score (Table 2).

Addition of values of serum levels of ferritin and transferrin to the MELD score to predict survival in ESLD patients

Cox proportional hazard models were created using MELD-score parameters (levels of bilirubin, creatinine and INR) with separate additions of serum levels of ferritin and transferrin.

A HR for ferritin of 1.0 and of 1.004 for transferrin was obtained. Using these HRs, values for serum levels of ferritin and transferrin were added to the MELD score, and ROC curves were created.²⁰ The serum level of transferrin failed to reach significance within the model created (p = 0.099), while the serum ferritin level was borderline significant (p = 0.045) (Figure 2). The model combining the MELD score parameters and ferritin decreased the AUROC by 0.11% compared with the AUROC of MELD score alone. Similarly, the addition of serum transferrin to the MELD score decreased the AUROC by 2.36% compared with MELD score alone. Both resulting AUROCs were similar to the AUROC of the original MELD score (ferritin p = 0.817, transferrin p = 0.891; Figure 3).

Figure 2.

Cox proportional hazard ratios for MELD parameters vs. serum levels of ferritin and transferrin for 90-day survival. When added separately to the MELD Score, serum levels of ferritin and transferrin showed a hazard ratio close to 1, indicating that, within the model, both parameters were inferior compared with the MELD score parameters INR, bilirubin level and creatinine level.

Figure 3.

Comparison of the area under the receiver operating characteristics (AUROC) of the MELD score vs. the MELD score + ferritin level vs. the MELD score + transferrin level. Comparison of all three combinations revealed a similar AUROC, but MELD + ferritin level and MELD + transferrin level, showed no significant improvement compared with MELD alone (addition of ferritin, p = 0.817, addition of transferrin p = 0.891).

Discussion

This is the first study to show that serum levels of ferritin and transferrin are independent significant prognostic factors for mortality in patients with ESLD when PSM is applied. Furthermore, ferritin and transferrin levels can satisfactorily predict 11-day and 90-day mortality in ESLD patients and may, therefore, serve as simple and powerful screening tools in daily practice.

Serum levels of ferritin and transferrin are routinely applied laboratory markers that are frequently increased in ESLD patients.^5–7 Both parameters are associated with increased mortality in patients with ESLD.^6,12,22 In clinical settings, interpretation of serum ferritin levels can be challenging because they may exhibit a wide range, including extremely high values in some disorders (e.g. haemophagocytic lymphohistiocytosis, brucellosis).^23–25 Furthermore, a multitude of other conditions, related and unrelated to liver disease, might influence overall serum ferritin levels concomitantly.^10,26

Some scholars have evaluated serum levels of ferritin and transferrin as prognostic survival markers for patients with ESLD.^7,25,27,28 Anastasiou et al. found serum levels of ferritin and transferrin to be strong predictors of 90-day mortality in patients with an acute-on-chronic liver failure.²⁷ Viveiros et al. reported comparable results, stating that a serum level of transferrin < 180 mg/dl might be a risk factor for mortality in patients with liver cirrhosis.⁷ In other studies, serum levels of ferritin > 5000 µg/l²⁵ and of > 300 µg/l²⁸ have been associated with higher mortality in patients with ESLD. However, those results were not determined based on optimal cutoff values but upon upper reference limits or the need for specific analyses of serum levels of ferritin and transferrin.^25,28 None of those studies corrected for a potential bias due to conditions that concomitantly increase baseline serum levels of ferritin and transferrin levels. Therefore, the importance of serum levels of ferritin and transferrin as independent prognostic parameter for survival in ESLD patients has not been studied previously.^7,25,27,28

Our study was the first one addressing this lack of evidence. Besides a standard unmatched analysis, we provided a PSM evaluation of serum levels of ferritin and transferrin in patients with ESLD. The unmatched analysis revealed that serum levels of ferritin and transferrin may be prognostic survival markers in these patients. Patients who died within 90 days exhibited significantly higher levels of ferritin and lower levels of transferrin in serum than those who did not decease within 90 days. The cutoff values we used to differentiate between long-term and short-term survivors were comparable to those described by other studies (≥754 µg/l for ferritin, ≤ 141.5 mg/dl for transferrin).^7,25,27,28 In the PSM analysis, our study showed (for the first time) that optimal serum ferritin and transferrin cutoffs had a strong and independent prognostic value for patient survival (≥1030.5 µg/l for ferritin, ≤ 177 mg/dl for transferrin). Eliminating a potential bias by adjusting for conditions known to increase overall serum ferritin demonstrated the prognostic value of ferritin and transferrin for survival prediction in patients with ESLD. Moreover, adjusting for known confounders could result in more exact cutoff values.

The MELD score represents one of the most reliable scores to determine the mortality of patients with ESLD.²⁹ Several proposals have been made to add additional parameters to the MELD score to improve its predictive capability, including serum levels of sodium and albumin; however, none has been approved for daily practice.^17,18,30 We showed, for the first time, that the addition of ferritin and transferrin did not further enhance the predictive capability of the MELD score.

In summary, serum levels of ferritin and transferrin are important, independent and MELD score equipotent predictors for mortality in patients with ESLD.

Our study had a few limitations. First, we reported data from a retrospective analysis. Nevertheless, we collected a comprehensive and detailed dataset and, furthermore, we applied PSM with a small number of matching parameters in combination with a 5% tolerance level to ensure precise matches.³¹ Second, we presented data from a single-centre study and our study lacks a validation cohort; however, we reported data from a large set of patients and included all patients treated for liver disease at our tertiary referral centre over the recruitment period. Third, we could not confirm that serum levels of ferritin and transferrin were influenced by latent infections. Nevertheless, inflammation influences serum levels of ferritin and transferrin.¹² Therefore, we applied PSM using inflammation as one of the matching parameters.

Conclusion

Serum levels of ferritin and transferrin are important and independent prognostic factors for mortality in patients with ESLD. Besides their excellent predictive capability for 90-day mortality, patients with serum ferritin levels ≥1030.5 µg/l had a 50% risk of dying within a median of 11 days. The advantage of using these parameters is to identify ESLD patients at a high risk of death without the need for score calculation.

Supplemental Material

UEG891283 Supplemental Material - Supplemental material for Serum levels of ferritin and transferrin serve as prognostic factors for mortality and survival in patients with end-stage liver disease: A propensity score-matched cohort study

Supplemental material, UEG891283 Supplemental Material for Serum levels of ferritin and transferrin serve as prognostic factors for mortality and survival in patients with end-stage liver disease: A propensity score-matched cohort study by Jörn Arne Meier, Arne Bokemeyer, Friederike Cordes, Valentin Fuhrmann, Hartmut Schmidt, Anna Hüsing-Kabar and Iyad Kabar in United European Gastroenterology Journal

Footnotes

Funding

This study received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.

Declaration of conflicting interests

The authors declare that there are no conflicts of interest.

Author contributions

J.A. M., A.B., A. H.-K. and I.K. designed the study, analysed the data and wrote the manuscript. F.C., V.F. and H.S. analysed the data and contributed to the writing of the manuscript. All the authors edited the scientific content of the manuscript and approved the final version prior to submission.

Informed consent

As approved by the ethics board, informed patient consent was not required for this study because of its retrospective design.

Ethics approval

This study was approved by the Medical Council of Westphalia-Lippe and the ethics board of the Westphalian Wilhelms-University of Münster, Germany (date of approval: April, 2016). This study conforms to the 1975 Declaration of Helsinki as reflected in the prior approval by the institution's human research committee.

ORCID iDs

Arne Bokemeyer

Friederike Cordes

Supplemental material

For access to the underlying study materials, contact the corresponding author.

References

D'Amico

Garcia-Tsao

Pagliaro

. Natural history and prognostic indicators of survival in cirrhosis: A systematic review of 118 studies. J Hepatol 2006; 44: 217–231.

Durand

Valla

. Assessment of prognosis of cirrhosis. Semin Liver Dis 2008; 28: 110–122.

Peng

Guo

. Child-Pugh versus MELD score for the assessment of prognosis in liver cirrhosis: A systematic review and meta-analysis of observational studies. Medicine (Baltimore) 2016; 95: e2877–e2877.

Singal

Kamath

. Model for end-stage liver disease. J Clin Exp Hepatol 2013; 3: 50–60.

Oikonomou

Goulis

Souladiopoulos

, et al. High serum ferritin is associated with worse outcome of patients with decompensated cirrhosis. Ann Gastroenterol 2017; 30: 217–224.

Umer

Makki

Kiran

, et al. Serum ferritin as a predictor of 30 days mortality in patients of decompensated chronic liver disease. J Ayub Med Coll Abbottabad 2017; 29: 415–418.

Viveiros

Finkenstedt

Schaefer

, et al. Transferrin as a predictor of survival in cirrhosis. Liver Transpl 2018; 24: 343–351.

Weismuller

Manns

Strassburg

. Ferritin and liver allocation? Impact on mortality not only on the waiting list but also after orthotopic liver transplantation should be considered. Hepatology 2010; 52: 392–393. author reply 393.

Jehn

Clark

Guallar

. Serum ferritin and risk of the metabolic syndrome in U.S. adults. Diabetes Care 2004; 27: 2422–2428.

10.

Ford

Cogswell

. Diabetes and serum ferritin concentration among U.S. adults. Diabetes Care 1999; 22: 1978–1983.

11.

Kalantar-Zadeh

Lee

. The fascinating but deceptive ferritin: To measure it or not to measure it in chronic kidney disease? Clin J Am Soc Nephrol 2006; 1(Suppl 1): S9–S18.

12.

Kell

Pretorius

. Serum ferritin is an important inflammatory disease marker, as it is mainly a leakage product from damaged cells. Metallomics 2014; 6: 748–773.

13.

Depalma

Hayes

Chow

, et al. Ferritin levels, inflammatory biomarkers, and mortality in peripheral arterial disease: A substudy of the Iron (Fe) and Atherosclerosis Study (FeAST) Trial. J Vasc Surg 2010; 51: 1498–1503.

14.

Kalousova

Krechler

Jachymova

, et al. Ferritin as an independent mortality predictor in patients with pancreas cancer. Results of a pilot study. Tumour Biol 2012; 33: 1695–1700.

15.

Kuss

Blettner

Borgermann

. Propensity score: An alternative method of analyzing treatment effects. Dtsch Arztebl Int 2016; 113: 597–603.

16.

Cholongitas

Marelli

Shusang

, et al. A systematic review of the performance of the model for end-stage liver disease (MELD) in the setting of liver transplantation. Liver Transpl 2006; 12: 1049–1061.

17.

Luca

Angermayr

Bertolini

, et al. An integrated MELD model including serum sodium and age improves the prediction of early mortality in patients with cirrhosis. Liver Transpl 2007; 13: 1174–1180.

18.

Myers

Shaheen

Faris

, et al. Revision of MELD to include serum albumin improves prediction of mortality on the liver transplant waiting list. PLoS One 2013; 8: e51926–e51926.

19.

Walker

Stuart

Ryan

, et al. Serum ferritin concentration predicts mortality in patients awaiting liver transplantation. Hepatology 2010; 51: 1683–1691.

20.

DeLong

Clarke-Pearson

. Comparing the areas under two or more correlated receiver operating characteristic curves: A nonparametric approach. Biometrics 1988; 44: 837–845.

21.

Jalan

Pavesi

Saliba

, et al. The CLIF Consortium Acute Decompensation score (CLIF-C ADs) for prognosis of hospitalised cirrhotic patients without acute-on-chronic liver failure. J Hepatol 2015; 62: 831–840.

22.

Hagström

Nasr

Bottai

, et al. Elevated serum ferritin is associated with increased mortality in non-alcoholic fatty liver disease after 16 years of follow-up. Liver Int 2016; 36: 1688–1695.

23.

Arica

Silfeler

Arica

, et al. Brucellosis with very high ferritin levels: report of five cases. Hum Exp Toxicol 2012; 31: 104–106.

24.

Allen

Kozinetz

, et al. Highly elevated ferritin levels and the diagnosis of hemophagocytic lymphohistiocytosis. Pediatr Blood Cancer 2008; 50: 1227–1235.

25.

Beer

Vadakara

. Etiologies and short-term mortality in patients with ultraelevated serum ferritin. South Med J 2015; 108: 574–578.

26.

Ledesma

Hurtado-Roca

Leon

, et al. Association of ferritin elevation and metabolic syndrome in males. Results from the Aragon Workers’ Health Study (AWHS). J Clin Endocrinol Metab 2015; 100: 2081–2089.

27.

Anastasiou

Kälsch

Hakmouni

, et al. Low transferrin and high ferritin concentrations are associated with worse outcome in acute liver failure. Liver Int 2017; 37: 1032–1041.

28.

Al-Freah

Kriese

Foxton

, et al.

The association of pretransplant ferritin level with waiting list and post-transplant survival. Does ferritin actually predict outcome?

Transpl Int 2013; 26: 1070–1079.

29.

Kamath

Wiesner

Malinchoc

, et al. A model to predict survival in patients with end-stage liver disease. Hepatology 2001; 33: 464–470.

30.

Asrani

Kamath

. Model for end-stage liver disease score and MELD exceptions: 15 years later. Hepatol Int 2015; 9: 346–354.

31.

King

Nielsen

. Why propensity scores should not be used for matching. Political Analysis 2019; 27.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.53 MB