Serum ferritin concentration and transferrin saturation before liver transplantation predict decreased long-term recipient survival


  • Tobias J. Weismüller,

    1. Department of Gastroenterology, Hepatology and Endocrinology, Hannover Medical School, Hannover, Germany
    2. Integrated Research and Treatment Center—Transplantation (IFB-Tx), Hannover Medical School, Hannover, Germany
    Search for more papers by this author
  • Gabriele I. Kirchner,

    1. Department of Internal Medicine I, Regensburg University Hospital, Regensburg, Germany
    Search for more papers by this author
  • Marcus N. Scherer,

    1. Department of Surgery, Regensburg University Hospital, Regensburg, Germany
    Search for more papers by this author
  • Ahmed A. Negm,

    1. Department of Gastroenterology, Hepatology and Endocrinology, Hannover Medical School, Hannover, Germany
    Search for more papers by this author
  • Andreas A. Schnitzbauer,

    1. Department of Surgery, Regensburg University Hospital, Regensburg, Germany
    Search for more papers by this author
  • Frank Lehner,

    1. Department of General, Visceral and Transplant Surgery, Hannover Medical School, Hannover, Germany
    Search for more papers by this author
  • Jürgen Klempnauer,

    1. Department of General, Visceral and Transplant Surgery, Hannover Medical School, Hannover, Germany
    Search for more papers by this author
  • Hans J. Schlitt,

    1. Department of Surgery, Regensburg University Hospital, Regensburg, Germany
    Search for more papers by this author
  • Michael P. Manns,

    1. Department of Gastroenterology, Hepatology and Endocrinology, Hannover Medical School, Hannover, Germany
    2. Integrated Research and Treatment Center—Transplantation (IFB-Tx), Hannover Medical School, Hannover, Germany
    Search for more papers by this author
  • Christian P. Strassburg

    Corresponding author
    1. Department of Gastroenterology, Hepatology and Endocrinology, Hannover Medical School, Hannover, Germany
    2. Integrated Research and Treatment Center—Transplantation (IFB-Tx), Hannover Medical School, Hannover, Germany
    • Professor of Gastroenterology and Hepatology, Department of Gastroenterology, Hepatology and Endocrinology, Hannover Medical School, Carl Neuberg Straße 1, 30655 Hannover, Germany
    Search for more papers by this author
    • fax: +49 511 532 4896

  • Potential conflict of interest: Dr. Schlitt advises and is on the speakers' bureau of Roche and Novartis. He received grants from Wyeth.

  • This study was supported by the German Federal Ministry of Education and Research through the Integrated Research and Treatment Center—Transplantation (reference number: 01EO0802) to Hannover Medical School, Hannover, Germany.


Serum ferritin (SF) concentration is a widely available parameter used to assess iron homeostasis. It has been described as a marker to identify high-risk patients awaiting liver transplantation (LT) but is also elevated in systemic immune-mediated diseases, metabolic syndrome, and in hemodialysis where it is associated with an inferior prognosis. This study analyzed whether SF is not only a predictor of liver-related mortality prior to LT but also an independent marker of survival following LT. In a dual-center, retrospective study, a cohort of 328 consecutive first-LT patients from Hannover Medical School, Germany (2003-2008, follow-up 1260 days), and 82 consecutive LT patients from Regensburg University Hospital, Germany (2003-2007, follow-up 1355 days) as validation cohort were analyzed. In patients exhibiting SF ≥365 μg/L versus <365 μg/L prior to LT, 1-, 3-, and 5-year post-LT survival was 73.3% versus 81.1%, 64.4% versus 77.3%, and 61.1% versus 74.4%, respectively (overall survival P = 0.0097), which was confirmed in the validation cohort (overall survival of 55% versus 83.3%, P = 0.005). Multivariate analyses identified SF ≥365 μg/L combined with transferrin saturation (TFS) <55%, hepatocellular carcinoma, and the survival after LT (SALT) score as independent risk factors for death. In patients with SF concentrations ≥365 μg/L and TFS <55%, overall survival was 54% versus 74.8% in the remaining group (P = 0.003). In the validation cohort, it was 28.6% versus 72% (P = 0.017), respectively. Conclusion: SF concentration ≥365 μg/L in combination with TFS <55% before LT is an independent risk factor for mortality following LT. Lower TFS combined with elevated SF concentrations indicate that acute phase mechanisms beyond iron overload may play a prognostic role. SF concentration therefore not only predicts pre-LT mortality but also death following LT. (HEPATOLOGY 2011;)

Orthotopic liver transplantation (LT) represents the ultimate therapeutic option for an array of progressive liver diseases that lead to irreversible liver failure. The evaluation of candidates for LT assesses the need of the potential graft recipient, the availability of a graft for transplantation, and estimates a favorable outcome of the procedure. Need is currently assessed with the Model of End-Stage Liver Disease (MELD) to predict 3-month mortality from different liver diseases, and to assign and define the priority for liver graft allocation.1, 2 This is necessary in view of the unfortunate shortage of organ grafts in most areas where LT is routinely available.

MELD employs international normalized ratio (INR), serum creatinine, and serum bilirubin for its calculation. High MELD scores indicate a high degree of mortality and morbidity and thus a need for organ replacement, but they also appear to affect post-LT survival.3-6 The correlation of MELD with inferior outcome after LT was not reported in all studies and settings.7, 8 It appears to be influenced by indication for LT such as hepatitis C virus infection and cholestatic and noncholestatic diseases,9-11 and shows low predictive abilities in some studies with a c-statistic of 0.54-0.58.12, 13 To refine the accuracy of MELD for allocation, additional parameters have been studied, such as serum sodium14-16 and serum ferritin (SF).17 Both parameters are easily determined and universally available in routine clinical chemistry laboratories. However, they can also indicate patient morbidity, which may influence prognosis and outcome following LT. This has been described for impaired renal function (as a part of MELD parameters),5 serum sodium,18 as well as for elevated SF and prognosis in hemodialysis patients,19-21 hematological diseases,22, 23 and iron overload prior to LT.24 An ideal instrument for the prediction of urgent need for LT would encompass only such parameters not also associated with an inferior prognosis following LT.

Several studies have analyzed data to define prognostic models associated with outcome following LT, which include only pre-LT recipient factors (age, serum creatinine, cholinesterase; SALT [survival after LT] score),25 or recipient, donor, and surgery-related data (survival outcomes following LT [SOFT] score).26 SALT reached a c-statistic of 0.79 (MELD = 0.57; 6-month post-LT survival) in an LT cohort with a mean MELD of 14.5.25 SOFT, developed in a large cohort with a mean MELD of 20.6, showed superior outcome prediction than MELD (c-statistic for SOFT = 0.7; for MELD = 0.63; 3-month post-LT survival) with the main variables being previous LT and pre-LT life support.26 In 2010, SF was reported as a prognostic parameter in patients on the waiting list.17 This observation is interesting, because SF not only represents a parameter for iron homeostasis27 but has also been linked to systemic inflammatory and cytokine-mediated processes spanning conditions including metabolic syndrome,28 rheumatological disease,29, 30 and hemodialysis,19 in which it is associated with increased mortality.19 We observed that patients with high SF concentrations before LT exhibited an inferior survival following LT.31 In this study, we therefore analyzed survival, SF, and transferrin saturation (TFS) in two independent LT cohorts. The results suggest that elevated SF in combination with low TFS prior to LT is an important predictor of mortality following LT.


AIH, autoimmune hepatitis; HCC, hepatocellular carcinoma; ICU, intensive care unit; INR, international normalized ratio; LT, liver transplantation; MELD, Model for End-Stage Liver Disease; NPV, negative predictive value; PBC, primary biliary cirrhosis; PPV, positive predictive value; PSC, primary sclerosing cholangitis; SALT, survival after liver transplantation; SD, standard deviation; SF, serum ferritin; TFS, transferrin saturation.

Patients and Methods

Patient Population and Laboratory Investigations

In this retrospective cohort study, all consecutive adult patients with chronic end-stage liver disease, who underwent a first LT at Hannover Medical School, Hannover, Germany, between January 1, 2003, and April 1, 2008, were included. After the exclusion of patients with fulminant liver failure (n = 38), multiple organ transplantation (n = 39), living donor LT (n = 16), and 16 patients with a diagnosis of hemochromatosis, 354 patients remained to be analyzed.

Laboratory data on the last clinic visit prior to the day of LT were obtained from the patient's medical documentation record. Based on these data, we calculated the MELD and the SALT scores as described.1, 25 Serum sodium was measured immediately prior LT in addition to the documentation of demographics and etiologies of liver diseases.

SF (immunochemical assay; Roche Diagnostics, Mannheim, Germany),32 TFS, and serum iron were routinely measured at the time of evaluation for LT. In 92.7% of the 354 patients fulfilling the inclusion criteria, pretransplant SF was available; therefore, 328 of 354 patients remained for further analyses. The mean time from transplant evaluation measurement of SF and TFS to the day of LT was 393 ± 575 days.

The end of the study period was April 1, 2010, so that all patients were followed either until death or for at least 2 years after LT.


The primary endpoint of this study was patient survival at the end of follow-up. Secondary endpoints were patient survival at 1, 3, and 5 years after LT, graft survival, and time spent in the intensive care unit (ICU) following LT. ICU time was defined as total days spent in the ICU during hospitalization from the time of LT to discharge from the medical center.

Validation Cohort

In order to validate our hypothesis and results that patients with an increased SF prior to LT exhibit a reduced long-term posttransplant survival, we studied all consecutive adult patients, who received a first LT at Regensburg University Hospital Transplant Center, Regensburg, Germany, between January 1, 2003, and December 31, 2007. SF was available from patient medical records in 59% of the 139 patients, fulfilling the inclusion and exclusion criteria mentioned above, which resulted in 82 patients remaining for the analyses.

Statistical Analyses

Variables were expressed as mean ± standard deviations and medians. Categorical variables were compared with chi-square test, and continuous variables with the Mann–Whitney U test. Patient survival was determined by Kaplan–Meier survival analysis, and different groups were compared by log-rank test. With regard to our local laboratory's normal reference range, which is based on an evaluation of different SF assays in a standard population,32 365 μg/L was chosen as a preselected cutoff value for SF. Cox proportional hazard ratios for death were estimated for univariate and multivariate models. All tests were two-tailed, and a P value < 0.05 was considered significant. Statistical analyses were performed using SPSS version 13.0 for Windows (SPSS Inc., Chicago, IL) software.


Study Population and Outcome

The study cohort comprised 328 LT patients (61.3% were males) with a mean age of 48.8 ± 10.9 years (range = 20.2-68.8 years) at the time of LT. The main indications for LT were alcoholic liver disease (20.7%), hepatitis C (18%), hepatocellular carcinoma (HCC) (17.1%), primary sclerosing cholangitis (PSC) (15.5%), and hepatitis B (12.2%). A split graft was used for primary transplantation in 10.7%. Mean cold ischemia time was 671 ± 161 minutes. The mean MELD score at last reevaluation before LT was 15.1 ± 7.3 (Table 1).

Table 1. Demographics, Clinical Data, Graft Data, and Laboratory Data of the Entire Study Population, and Comparison of These Parameters Between the Two SF Subgroups
CharacteristicStudy Population n = 328Subgroups According to SFP Value
SF < 365 μg/L (n = 238)SF≥ 365 μg/L (n = 90)
  1. For categorical variables, data are given as percentages. For continuous variables, data are given as mean ± SD and median. Mann–Whitney U test for continuous parameters and chi-squared test for categorical variables. AIH, autoimmune hepatitis; BILI, bilirubin; CHE, cholinesterase; CIT, cold ischemia time; CREA, serum creatinine; CRP, c-reactive protein; PBC, primary biliary cirrhosis.

Age at LT (years)48.8 ± 10.9, 49.647.9 ± 11.2, 49.151.4 ± 9.8, 51.40.01
Male sex201 (61.3%)138 (58%)63 (70%)0.046
 Alcoholic cirrhosis68 (20.7%)42 (17.6%)26 (28.9%)0.025
 HCC56 (17.1%)37 (15.5%)19 (21.1%)n.s.
 Hepatitis C59 (18%)38 (16%)21 (23.3%)n.s.
 Hepatitis B40 (12.2%)31 (13%)9 (10%)n.s.
 PSC51 (15.5%)46 (19.3%)5 (5.6%)0.002
 PBC20 (6.1%)16 (6.7%)4 (4.4%)n.s.
 AIH18 (5.5%)18 (7.6%)00.007
Graft parameter:    
 CIT (minutes)671 ± 161, 664682 ± 165, 676638 ± 143, 641n.s.
 Split graft35 (10.7%)29 (12.2.%)6 (6.7%)n.s.
Pretransplant biochemistry:    
 CREA (μmol/L)89.4 ± 83, 7280.6 ± 48.1, 69.5112.7 ± 135.8, 800.001
 BILI (μmol/L)90.1 ± 150.2, 4168.8 ± 123.2, 36.5146.3 ± 195.1, 69<0.001
 INR (ratio)1.45 ± 0.46, 1.331.39 ± 0.43, 1.311.59 ± 0.52, 1.450.001
 CHE (kU/L)3.6 ± 2.15, 3.13.86 ± 2.22, 3.322.92 ± 1.8, 2.64<0.001
 Na (mmol/L)136.3 ± 5.4, 137137.1 ± 4.9, 138134.1 ± 5.9, 135<0.001
 SF (μg/L)293.7 ± 436.7, 130106.9 ± 91.3, 74.5787.7 ± 581.6, 595<0.001
 TFS (%)48.4 ± 30.6, 4141.7 ± 28.3, 3564.7 ± 29.8, 66<0.001
 Iron (μmol/L)22.4 ± 13.4, 2021.5 ± 12.9, 1924.9 ± 14.3, 230.020
Pretransplant scores:    
 MELD15.1 ± 7.3, 13.913.8 ± 6.4, 1318.5 ± 8.4, 16.7<0.001
 SALT0.97 ± 0.99, 1.210.81 ± 0.99, 1.001.37 ± 0.88, 1.48<0.001

Mean follow-up of the entire study cohort was 1260 days (range = 1-2653 days), follow-up for surviving patients was 1639 days (732-2653) days. During follow-up, retransplantation was necessary in 46 patients (14%), and 96 patients died (29.3%). The 1-, 3-, and 5-year survival rates were 79%, 73.8%, and 70.7%, respectively. Main reasons for death, either alone or in combination, were sepsis (58.5%), multiorgan failure (56.4%), malignancy (21.3%), graft failure (12.8%), bleeding (12.8%), cardiac (10.6%), or cerebral (10.6%) diseases.

Serum Ferritin and Baseline Characteristics

According to the reference range of our central chemistry laboratory, 365 μg/L was used as a cutoff value to divide the study cohort into a low-SF group (n = 238) and a high-SF group (n = 90). Clinical characteristics and biochemical parameters differed significantly between both groups (Table 1): high-SF patients were older with a higher proportion of males, and greater incidence of cirrhosis because of alcoholic disease. In contrast, none of the patients with autoimmune hepatitis (AIH) and only 9.8% of the patients with PSC had a pre-LT SF levels ≥365 μg/L. Parameters reflecting more advanced liver disease such as MELD score, serum sodium, and SALT score, which correlate with outcome after LT, and the parameters used to calculate these scores were significantly higher in the high-SF group. In accordance with this, the mean waiting time from first transplant evaluation and measurement of SF and TFS, until the day of LT was significantly shorter in the high-SF group (290 versus 432 days, P < 0.001). However, neither cold ischemia time nor the percentage of split-organ LTs differed between both groups.

Serum Ferritin and Outcome Following Transplantation

Patients with a SF ≥365 μg/L had a significantly reduced 3-year, 5-year, and overall survival (Fig. 1A). In these patients, longer postoperative ICU times were observed (Table 2). However, overall long-term graft survival did not differ between both groups.

Figure 1.

(A) Study cohort (n = 328): Kaplan–Meier analysis of long-term recipient survival after LT. Patients with a SF ≥ or <365 μg/L were compared by log-rank test. Vertical bars = censored cases. (B) Study cohort (n = 275): Kaplan–Meier analysis of long-term recipient survival. Patients with a SF ≥ or <365 μg/L and a TFS ≥ or <55% were compared by log-rank test. Vertical bars = censored cases.

Table 2. Outcome Parameters of the Entire Study Population and Comparison Between the Two SF Subgroups (Log-Rank Test, Mann–Whitney U Test)
ParameterStudy population n = 328Subgroups According to SFP Value
SF <365 μg/L (n = 238)SF ≥365 μg/L (n = 90)
Overall recipient survival70.7%74.4%61.1%0.0097
Overall graft survival62.5%64.3%57.8%n.s.
Time in intensive care unit (days), mean ± SD, median23.9 ± 34.8, 1122.8 ± 36, 926.7 ± 31.4, 150.005

When analyzed as a continuous variable, surviving patients had a significantly lower SF prior to LT (264.7 ± 377.1 μg/L versus 363.6 ± 551 μg/L, P = 0.014), although there was no significant difference regarding serum iron concentration (22 μmol/L versus 23.3 μmol/L) or TFS (48% versus 49.5%). The surviving patients also had a lower pretransplant MELD score (14.4 ± 6.5 versus 16.8 ± 8.6, P = 0.025) and a lower pretransplant SALT score (0.81 ± 0.98 versus 1.34 ± 0.93, P < 0.001). In addition, a Kaplan–Meier analysis of patients who underwent LT, excluding those with HCC (n = 272), showed similar results, and also a significantly reduced overall survival of patients with a SF ≥365 μg/L (62.0% versus 78.6%, P = 0.002; data not shown).

Pretransplant SF ≥365 μg/L showed a specificity of 76.3% and a negative predictive value (NPV) of 74.4%, but only a low sensitivity of 36.5% and a positive predictive value (PPV) of 38.9% for death after LT in the long-term follow-up, resulting in an accuracy of 64.6%. The accuracy of SF ≥365 μg/L as a predictive parameter for outcome was even lower in patients who underwent transplantation because of alcoholic cirrhosis, HCC, or hepatitis C (Table 3). In contrast, for patients with cholestatic liver diseases (PSC in particular), specificity and NPV were greater than 85%, resulting in a good accuracy, although sensitivity was low.

Table 3. Sensitivity, Specifity, PPV and NPV, and Accuracy for SF ≥365 μg/L Used as a Test for Long-Term Survival After LT (Follow-Up at Least 2 Years)
Cutoff SF 365 μg/LSensitivitySpecifityPPVNPVAccuracy
  1. PBC, primary biliary cirrhosis; SSC, secondary sclerosing cholangitis.

Whole Study Population36.5%76.3%38.9%74.4%64.6%
TFS < 55%31.3%85.8%45.5%76.8%70.86%
 Alcoholic cirrhosis32.1%57.5%34.6%54.8%47.1%
 Hepatitis C30%61.5%28.6%63.2%50.9%
 Hepatitis B27.3%79.3%33.3%74.2%65%
 Cholestatic (PSC, PBC, SSC, Caroli disease)23.1%85.3%23.1%85.3%75.3%

In view of the difference in waiting time between the low-SF and high-SF groups, we analyzed whether this impacted the observed results. Pre-LT waiting time was significantly longer in survivors (423 versus 321 days, P = 0.014). However, when both waiting time and SF ≥365 μg/L where entered into a logistic regression model, SF ≥365 μg/L remained as the only independent parameter. To exclude an influence of elevated SF levels that rose immediately prior LT due to acute-phase reactions, an additional analysis excluding those 45 cases with an SF measurement obtained less than 60 days before LT was undertaken. Even in this analysis, the results were confirmed to demonstrate a significantly superior overall survival of the low-SF group (n = 216) versus the 67 patients with pretransplant SF ≥365 μg/L (74.1% versus 61.2%, P = 0.027).

To further identify confounding variables that influence the effect of pretransplant SF on long-term survival, we analyzed the parameters differing between the group of recipients, which exhibited a correct correlation of SF with outcome and the ones that did not (Table 4). The group of patients (n = 212, 64.6%) in which long-term outcome was accurately predicted by pretransplant SF (cutoff 365 μg/L) was significantly younger, and had significantly lower MELD and SALT scores prior to LT. In addition, c-reactive protein and the MELD parameters creatinine, bilirubin, and INR were lower, whereas serum sodium and cholinesterase were higher in this group of patients. Interestingly, these patients also exhibited a significantly lower mean pre-LT TFS, whereas their serum iron concentrations did not differ.

Table 4. Comparison of Different Clinical Parameters of Recipients with a Correct Correlation of SF with Outcome with Those Without a Correct Prediction of Outcome
ParameterData Entry RateCorrelation of SF with Overall Recipient Survival (n = 212)No Correlation of SF with Overall Recipient Survival (n = 116)P Value
 n (%)meanmean 
  1. BILI, bilirubin; CHE, cholinesterase; CREA, serum creatinine; CRP, c-reactive protein.

Age at LT, years328 (100%)47.152.02<0.001
TFS (%)275 (83.8%)44.5%55.5%0.004
Iron (μmol/L)307 (93.6%)22.1322.85n.s.
CRP (mg/L)181 (55.2%)12.6720.950.007
CREA (μmol/L)328 (100%)88.990.40.043
BILI (μmol/L)328 (100%)80.6107.30.042
INR (ratio)327 (99.7%)1.41.54<0.001
CHE (kU/L)318 (97%)3.843.160.008
Serum sodium (mmol/L)322 (98.2%)136.6135.50.032
MELD327 (99.7%)14.416.50.002
SALT318 (97%)0.811.25<0.001

Serum Ferritin Plus Low Transferrin Saturation and Outcome Following Transplantation

Because of the highly significant lower TFS values in patients with a correct correlation of SF ≥365 μg/L and outcome, we stratified the patients of the high-SF group and the low-SF group according to their TFS with 55% as the optimal cutoff point, which we identified by receiver operating curve analysis. The overall survival of the 242 patients with either SF <365 μg/L or with SF ≥365 μg/L but TFS ≥55% was 74.8%, which was significantly (P = 0.003) better than the overall survival of 54.5% of the 33 patients with SF ≥365 μg/L and TFS <55% (Fig. 1B). Notably, the mean waiting time from measurement of SF and TFS to LT was longer (425 days; not significant) in the 33 patients with SF ≥365 μg/L and TFS <55% than in the 48 patients with SF ≥365 μg/L but TFS ≥55% (209 days). The graft survival was also lower in the high-SF group with TFS <55% (65.3% versus 51.5%), but this difference was not significant (log-rank test: P = 0.07). In addition, the post-LT ICU time was longer (26.8 versus 24 days) but not statistically significant (Mann–Whitney U test: P = 0.34). There were no significant differences regarding the causes of mortality between both groups.

Pretransplant SF ≥365 μg/L plus TFS <55% exhibited a specificity of 91% and a NPV of 74.8% for death after LT in the long-term follow-up (Table 5), but sensitivity (19.7%) and the PPV (44%) were low. The prognostic accuracy of SF ≥365 μg/L was also improved by the combination with TFS <55% in the other subgroups.

Table 5. Sensitivity, Specifity, PPV and NPV, and Accuracy for SF ≥365 μg/L Plus TFS <55% Used as a Test for Long-Term Survival After LT (Follow-Up at Least 2 Years)
SF ≥365 μg/L Plus TFS <55%SensitivitySpecificityPPVNPVAccuracy
  1. PBC, primary biliary cirrhosis; SSC, secondary sclerosing cholangitis.

Whole study population19.7%91%45.5%74.8%71.3%
 Alcoholic cirrhosis9.1%89.5%33.3%63%60%
 Viral (hepatitis B and/or C)8.7%90%25%72%67.5%
 Cholestatic (PSC, PBC, SSC, Caroli disease)18.2%91.2%28.6%85.2%79.4%

Univariate and Multivariate Cox Proportional Hazard Models of Long-Term Outcome

To identify predictive parameters for long-term outcome following LT, we assessed hazard ratios of etiology of liver disease, sex, the predictive MELD and SALT scores, serum sodium, and SF >365 μg/L plus TFS <55% in univariate and multivariate Cox proportional hazard models (Table 6). In univariate analyses, a history of alcoholic cirrhosis, presence of HCC prior to LT, MELD score and SALT score, and a SF >365 μg/L plus TFS <55% before LT were significant risk factors for overall mortality. PSC was a significant protective factor. We then further entered these significant variables into a multivariate Cox proportional hazard model, which identified the SALT score, presence of HCC, and a SF > 365 μg/L plus TFS <55% as three independent parameters predicting higher mortality following LT.

Table 6. Cox Proportional HRs with 95% CIs Were Estimated for Overall Mortality
VariableUnivariate AnalysisMultivariate Analysis
HR (95%CI)P ValueHR (95%CI)P Value
  1. Parameters that were significant in the univariate model were further analyzed in a multivariate model. CI, confidence interval; HR, hazard ratio.

Male sex1.32 (0.86-2.01)0.20  
Alcoholic cirrhosis1.73 (1.11-2.69)0.0151.4 (0.83-2.35)0.20
HCC2.09 (1.33-2.29)0.0011.98 (1.12-3.48)0.018
Hepatitis C1.20 (0.73-1.97)0.46  
Hepatitis B0.92 (0.49-1.73)0.81  
PSC0.37 (0.17-0.81)0.0121.2 (0.49-2.95)0.69
MELD1.04 (1.02-1.07)0.0011.02 (0.98-1.06)0.44
SALT1.77 (1.38-2.28)<0.0011.56 (1.09-2.24)0.016
SF >365 μg/L and TFS <55%2.28 (1.29-4.02)0.0041.9 (1.07-3.37)0.029
Serum sodium0.97 (0.93-1.01)0.087  

Analysis of a Second Validation Cohort

The independent validation cohort from Regensburg Transplant Center differed from the Hannover study cohort: 75.6% of the patients were male. The most common etiologies were alcoholic-induced cirrhosis (51.2%), hepatitis C (18.3%), HCC (15.9%), hepatitis B (6.1%), and PSC (6.1%).

Mean follow-up time of this cohort was 1355 days (range = 30-2821 days) and 1689 days (range = 1054-2821 days) for surviving patients. During follow-up, a retransplantation was necessary in four patients (5%), and 25 patients died (30.5%). The 1-, 3-, and 5-year recipient survival rates were 81.7%, 76.8%, and 72%.

We could assign 40 of the 82 patients to the high-SF group (SF ≥ 365 μg/L); these patients had a significantly higher SF (1224.7 ± 1751.3 μg/L versus 100.7 ± 84.9 μg/L, P < 0.001), a significantly higher TFS (70.9% ± 35.9% versus 39.5% ± 27.9%, P < 0.001), but serum iron concentrations did not differ (116.4 ± 52.2 μmol/L versus 101.5 ± 67.2 μmol/L, P = 0.087). the 1-, 3-, and 5-year survival rates (70%, 60%, and 57.5% versus 92.9%, 92.9%, and 85.7%) as well as the overall survival (55% versus 83.3%) were significantly decreased in the high-SF group (Fig. 2A). The Cox proportional hazard ratio for overall mortality of SF >365 μg/L was estimated as 3.24 (95% confidence interval = 1.35-7.79, P = 0.009).

Figure 2.

(A) Validation cohort (n = 82): Kaplan–Meier analysis of long-term recipient survival after LT. Patients with a SF ≥ or <365 μg/L were compared by log-rank test. Vertical bars = censored cases. (B) Validation cohort (n = 75): Kaplan–Meier analysis of long-term recipient survival. Patients with a SF ≥ or <365 μg/L and a TFS ≥ or <55% were compared by log-rank test. The actual 1-, 3-, and 5-year survival rates are given next to the curves. Vertical bars = censored cases.

TFS data were available in 39 of 40 patients of the high-SF group. A total of 14 patients showed a SF ≥365 μg/L and a TFS <55%, and their overall survival was only 28.6%. This was significantly lower than the 72% overall survival of the 25 patients with SF ≥365 μg/L but TFS ≥55% (P = 0.017), the 87.5% overall survival of the eight patients with SF <365 μg/L and TFS ≥55% (P = 0.008), and the 82.1% overall survival of the 28 patients with SF <365 μg/L and TFS <55% (P < 0.001; Fig. 2B). The Cox proportional hazard ratio for overall mortality of SF >365 μg/L and TFS <55% was estimated as 4.83 (95% confidence interval = 2.09-11.16, P < 0.001).


SF and TFS are routinely available biochemical parameters usually employed to assess iron homeostasis as part of the clinical work-up of iron storage diseases such as hemochromatosis.27 However, SF is also elevated in other conditions, including diabetes mellitus,20 hemodialysis,19 metabolic syndrome,28 advanced liver diseases,33, 34 adult-onset Still's disease,30, 35 Behcet's disease,36 and other inflammatory conditions.37, 38

In a recent study by Walker et al., elevated SF was identified as a prognostic marker for liver-related mortality and clinical events in patients on the waiting list.17 Because SF is elevated in many conditions and appears to have a prognostic role, it appeared plausible that SF may also be a predictor of mortality and outcome after LT. We therefore studied a cohort of 328 consecutive patients who underwent first LT from Hannover Medical School that was stratified according to SF equal or above, or below 365 μg/L, reflecting the upper limit of normal of biochemical SF determinations. The analysis demonstrated a significant difference of overall recipient (74.4% versus 61.1%) but not graft survival following LT. This difference was significant at 3 and 5 years after LT. An independent validation cohort of 82 LT patients from Regensburg University Hospital confirmed this finding (83.3% versus 55%), with significant differences after 1, 3, and 5 years. As expected, the high-SF group exhibited more males, more alcoholic cirrhosis, less PSC and AIH, and higher MELD and SALT scores in addition to higher TFS and serum iron values. Patients with high pre-LT SF values required significantly longer ICU treatment indicating a higher degree of morbidity. The study and validation cohorts differed regarding indications for LT. More alcoholic cirrhosis (50.2% versus 20.7%) as well as a lower number of viral hepatitis and autoimmune diseases were present in the validation cohort. High numbers of alcoholic cirrhosis and associated high SF levels is a likely explanation why SF already significantly predicted mortality after the first year following LT in the validation cohort. The confirmation of this observation in two differing cohorts strengthens the observed predictive role of SF.

A plausible explanation for the predictive value of SF would be an increased iron load contributing to mortality after LT. SF would not only reflect increased hepatic iron but also iron accumulation in extrahepatic sites relevant to overall mortality. Cardiac iron deposition was observed in transvenous endomyocardial biopsies of 64% of patients with substantial hepatic iron staining.33 Moreover, studies examining iron load in explanted livers have shown an inferior survival in individuals with iron overload. A reduced 5-year post-LT survival of 40% in 37 patients with hepatic iron overload compared to 62% in age-matched controls was reported.24 In another study, hepatic iron overload (>40 μmol/g) was associated with a 1- and 5-year unadjusted survival of 74% and 63%, compared to 80% and 72% in patients with normal iron content (<40 μmol/g) (not significant).39 A multicentric study reported that patients without hemochromatosis but with hepatic iron overload had a reduced 5-year post-LT survival of 63% versus 72% in the overall LT cohort.40

Although iron load is a determinant of the course and severity of liver diseases the question remains why iron overload would affect 5-year survival of patients in whom the liver as a relevant site of iron storage has been eliminated. In univariate and multivariate analyses, independent parameters associated with increased mortality were analyzed. The presence of HCC, the SALT score (age, serum creatinine, cholinesterase),25 and SF ≥365 μg/L in combination with TFS <55% were identified as independent variables. In LT patients exhibiting SF ≥365 μg/L and TFS <55%, an overall survival of 54.5% in comparison to 74.8% in the remaining group was observed and confirmed in the validation cohort (28.6% versus 72%). These data indicate that with TFS below 55% the elevation of SF is associated with a higher risk of post-LT mortality.

Ferritin is also an acute phase protein elevated in response to immune-mediated and infectious stimuli, which may thus represent a surrogate marker for a general predisposition for morbidity and mortality. In our study, c-reactive protein levels were compared and found to be lower in the group in which SF correlated well with overall recipient survival (Table 4). Generally, elevated SF need not be linked to c-reactive protein levels in acute phase responses.41, 42 In addition, advanced liver diseases can contribute to a low c-reactive protein level response by reduced hepatic protein synthesis. In patients treated with interferon alpha-2b decreased c-reactive protein and significant elevations of SF were reported.43 This indicates a differential activation of acute phase markers such as c-reactive protein and SF, which is likely to be responsible for high SF, i.e., in adult-onset Still's disease29, 30 and other conditions. In patients undergoing hemodialysis and those with metabolic syndrome, elevated SF without elevations of TFS44, 45 has been observed, and SF levels have been associated with inferior prognosis.19, 21 Therefore, SF and TFS are not only markers for iron overload but can indicate an activation of acute phase and possibly other mechanisms35, 36 that influence mortality. In our study cohorts, liver biopsy material was not available to correlate histological iron load with the biochemical data. However, an analysis of the National Health and Nutrition Examination Survey (NHANES) 1999-2002 reported that even modest elevations of SF were associated with reduced cardiovascular fitness in young male subjects,46 and that SF may represent a morbidity-associated parameter. Against this background, the finding that elevated SF in addition to lower levels of TFS are predictive for mortality and morbidity may not indicate systemic iron overload. One limitation of this retrospective study is that no measurements of iron metabolism parameters were performed or were available after LT, which should be studied in future analyses to observe whether elevated SF persists after LT in patients with decreased survival. In addition, it may be of interest to reanalyze the pretransplant situation in other studies17 to assess whether there is also a difference between patients with high or low TFS and elevated SF regarding mortality on the waiting list. This may contribute to potential pre-LT therapeutic strategies.

In conclusion, we show that SF elevations before LT predict an increased mortality following LT. This risk is highest in patients with SF ≥365 μg/L and TFS <55%, which was identified as an independent parameter. The incorporation of SF into score models assessing the urgency of LT and placement on the waiting list should therefore be viewed with caution to avoid incorporating a parameter that may simultaneously be associated with inferior outcome.