Serum iron parameters in the early post-transplant period and infection risk in kidney transplant...

Serum iron parameters in the early post-transplant

period and infection risk in kidney transplant

recipients

M. Fern�andez-Ruiz, F. L�opez-Medrano, A. Andr�es, J.M. Morales,C. Lumbreras, R. San-Juan, N. Polanco, E. Gonz�alez, J.M. Aguado.Serum iron parameters in the early post-transplant period andinfection risk in kidney transplant recipients.Transpl Infect Dis 2013. All rights reserved

Abstract: Background. The impact of iron metabolism on the riskof infectious complications has been demonstrated in variousimmunosuppressed populations. However, no previous studies haveassessed this potential association in kidney transplant (KT)recipients.Methods. We prospectively analyzed 228 patients undergoing KT atour institution from November 2008 to February 2011. Serum ironparameters (iron level, ferritin, total iron-binding capacity,unsaturated iron-binding capacity, transferrin, and transferrinsaturation) were assessed within the first 2 weeks aftertransplantation (median interval, 3 days; interquartile [Q1–Q3]range, 1–6 days), and before the occurrence of the first infectiousepisode (median interval, 26 days; Q1–Q3 range, 11–76 days).Primary outcome was the occurrence of any episode of infectionduring the first year. Multivariate-adjusted hazard ratios (aHRs)were estimated by Cox regression models.Results. Patients with ferritin level ≥500 ng/mL had higherincidence rates (per 1000 transplant-days) of overall infection(P = 0.017), bacterial infection (P = 0.002), and bloodstreaminfection (P = 0.011) during the first post-transplant year. One-yearinfection-free survival rate was lower in these recipients (26% vs. 41%;P = 0.004). On multivariate analysis, after adjusting for potentialconfounders, ferritin emerged as an independent predictor of overallinfection (aHR [per unitary increment], 1.001; P = 0.006), andbacterial infection (aHR [per unitary increment], 1.001; P = 0.020).Conclusion. Monitoring of serum iron parameters in the early post-transplant period may be useful in predicting the occurrence ofinfection in KT recipients, although further studies should be carriedout to confirm this preliminary finding.

M. Fern�andez-Ruiz1, F. L�opez-Medrano1, A. Andr�es2,J.M. Morales2, C. Lumbreras1,R. San-Juan1, N. Polanco2,E. Gonz�alez2, J.M. Aguado11Unit of Infectious Diseases, Hospital Universitario “12 deOctubre,” Instituto de Investigaci�on Hospital “12 deOctubre” (i+12), School of Medicine, UniversidadComplutense, Madrid, Spain, 2Department of Nephrology,Hospital Universitario “12 de Octubre,” Instituto deInvestigaci�on Hospital “12 de Octubre” (i+12), School ofMedicine, Universidad Complutense, Madrid, Spain

Key words: kidney transplantation; infection; iron-binding capacity; serum iron parameters; ferritin;transferrin

Correspondence to:Mario Fern�andez-Ruiz, MD, Unit of InfectiousDiseases, Hospital Universitario “12 de Octubre,”Centro de Actividades Ambulatorias, 2ª planta,bloque D. Avda. de C�ordoba, s/n, Madrid 28041,SpainTel: +34 913908000Fax: +34 914695775E-mail: [email protected]

Received 10 December 2012, revised 5 March2013, accepted for publication 11 April 2013

DOI: 10.1111/tid.12137Transpl Infect Dis 2013: 0: 1–12

Iron is a critical factor for both the survival andvirulence of most pathogenic microorganisms (1). Thestrategy of limiting iron availability to invading bacteriaand fungi represents an evolutionary highly conserveddefense mechanism in vertebrates, and exemplifies theso-called nutritional immunity (2). The iron homeosta-sis is strongly controlled in the human host, sincemost of the element appears complexed with variouscarrier proteins (such as transferrin or lactoferrin) or it

is located inside the cells as part of the heme group.Thus, free iron concentration in tissue fluids isextremely low (about 10�18 M), which is far less thanwhat is needed for the survival of the invadingpathogen (3). Conditions leading to iron overloadmarkedly increase the susceptibility to certain bacte-rial microorganisms (4, 5). This deleterious effect isparticularly evident for invasive fungal infection (IFI)(6, 7). The seminal evidence for this association

1

© 2013 John Wiley & Sons A/S

Transplant Infectious Disease, ISSN 1398-2273

came from patients on maintenance hemodialysis(HD) who developed fulminant and commonly fatalmucormycosis while being treated with deferoxamine(an iron chelator derived from Streptomyces pilosus) foriron overload (8). Further studies revealed thatdeferoxamine acts as a siderophore for Mucorales,delivering available iron directly to the fungus andpromoting its growth (9). In addition, in vitro andclinical studies have demonstrated that iron overloaddecreases the proliferative capacity and activity ofCD4+ helper T cells, enhances CD8+ suppressor Tcells, and impairs antibody production (10, 11). Ironalso exerts a negative effect on neutrophil phagocyticfunction (12).Some previous studies have assessed the impact of

iron metabolism on the risk of infectious complicationsin immunosuppressed patients, mostly in hematopoi-etic stem cell transplant (HSCT) recipients (13–17) and,more recently, in the setting of orthotopic liver trans-plantation (OLT) (18, 19). Almost half of all mainte-nance HD patients exhibit moderate hyperferritinemia(serum level >500 ng/mL) that mirrors not onlyincreased iron stores but also underlying inflammationand malnutrition (20). Moreover, it has been reportedthat HD patients with an iron-replete status are exposedto an increased risk of bloodstream infection (BSI) (21)and infection-related mortality (22). To our knowledge,no previous studies have analyzed the potential value ofiron storage indices in predicting the occurrence ofinfectious complications after kidney transplantation(KT). Hence, in this study we explored the associationbetween serum iron parameters and infection in acohort of KT recipients.

Materials and methods

Study population

Since November 2008, all consecutive patients aged18 years or older undergoing single or double KT atour institution are being included in an ongoingimmune status assessment, as described elsewhere(23). All patients are prospectively followed up forinfection and other major complications during at least1 year (unless death or graft loss occurred earlier). Weexcluded patients with human immunodeficiency virusinfection, known pre-transplant primary immunodefi-ciencies, and those dying or suffering from graft losswithin the first post-transplant week. The local ClinicalResearch Ethics Committee approved the study proto-col, and written informed consent was obtained from allparticipants.

Study design

To be evaluable for this study, patients had to have anassessment of serum iron parameters (iron level,ferritin, total iron-binding capacity [TIBC], unsaturatediron-binding capacity [UIBC], transferrin, and transfer-rin saturation) within the first 2 post-transplant weeks.If a patient had >1 measurement during this period, theclosest one to the transplant date was considered foranalysis. Therefore, serum iron parameters wereassessed only once. We excluded those patients inwhom the first infectious episode occurred beforeserum sampling. In order to reduce the potential biasinduced by the acute-phase reactant nature of many ofthe serum iron parameters, we also excluded thepatients in which the sampling was performed on thesame calendar day of the infection onset. TIBC wascalculated as serum iron level plus UIBC, and transfer-rin saturation was calculated as serum iron leveldivided by TIBC. The primary study outcome was theoccurrence of at least 1 episode of infection during thefirst post-transplant year (either overall or bacterialinfection, as defined below). Asymptomatic bacteriuriaand upper respiratory tract infections not requiringadmission were excluded. The secondary study out-comes were all-cause and infection-related mortalityduring the first post-transplant year. A number of pre-transplant, perioperative, and post-transplant variableswere prospectively recorded. Graft function was esti-mated using the Modification of Diet in Renal Disease(MDRD-4) equation (24).

Immunosuppression and antimicrobial prophylaxis

All the recipients of grafts from donation after circula-tory death donors received induction therapy withintravenous (IV) rabbit anti-thymocyte globulin (ATG)(1.25 mg/kg daily for 5–7 days) followed by thedelayed introduction of a calcineurin inhibitor (CNI)from day 6 in order to minimize the risk of delayed graftfunction (ATG-containing sequential therapy). Recipi-ents at high immunological risk – those with peak panelreactive antibody >50%, second KT in case the first graftwas lost to rejection within 2 years, or those receiving athird or fourth kidney graft irrespective of humanleukocyte antigen sensitization – also received ATGinduction for 1–3 days with the early initiation of theCNI from day 0. Finally, basiliximab induction (20 mgon days 0 and 4) was used in those patients at high riskof CNI-related nephrotoxicity because of advancedage or pre-transplant comorbidities, with the delayedintroduction of the CNI from day 5 (basiliximab-

2 Transplant Infectious Disease 2013: 0: 1–12

Fern�andez-Ruiz et al: Post-transplant ferritin and infection

containing sequential therapy). Maintenance immuno-suppression consisted of tacrolimus (0.1 mg/kg daily,adjusted to a target level of 10–15 ng/mL for the firstmonth, and 5–10 ng/mL for maintenance); mycophen-olate mofetil (500–1000 mg twice daily) or mycophen-olic acid (360 mg twice daily); and prednisone (1 mg/kg daily with progressive tapering).All patients received a single dose of IV cefazolin

intraoperatively. Prophylaxis for Pneumocystis jiroveciipneumonia with trimethoprim-sulfamethoxazole(160 mg/800 mg 3 times weekly) was administeredduring the first 9 months after transplantation. In thosepatients at high risk for cytomegalovirus (CMV)disease (serologic mismatch [donor positive/recipientnegative] or induction with ATG), either IV ganciclovir(5 mg/kg daily) or oral valganciclovir (900 mg daily)was administered during the first 2 weeks, and long-term prophylaxis was performed with valganciclovir for3 months. Preemptive therapy was not systematicallypracticed in intermediate-risk patients (CMV-seroposi-tive recipients) in our institution during the studyperiod.

Definitions

Bacterial infection included both culture-proven andprobable episodes (the latter ones defined by sugges-tive clinical features with prompt response to empiricantibiotic therapy in the absence of microbiologicaldocumentation). All the cases of viral, fungal, andparasitic infection had to be microbiologically provenby culture, serology, antigen assays, molecular meth-ods, and/or histopathology. We considered bacterial orfungal BSI to be significant according to the criteria ofthe Centers for Disease Control and Prevention (25).Catheter-related BSI was defined according to thecriteria proposed by the Infectious Diseases Societyof America (26). Surgical site infection was defined as aninfection occurring within 30 days of surgery thatinvolved skin or subcutaneous or deep soft tissues atthe surgical wound site, with the isolation of the samemicroorganism in at least 2 aseptically obtained cul-tures. Lower respiratory tract infection was defined bythe simultaneous presence of fever plus cough, dysp-nea, and/or newly appearing signs of pulmonarydisease on physical examination (usually cracklesand/or wheezing), while the diagnosis of pneumoniawas based on the aforementioned criteria plus a newpulmonary infiltrate on chest radiography. Lower uri-nary tract infection included either cystitis – defined bythe acute onset of dysuria, frequency, and/or urinaryurgency associated with significant bacteriuria (an

appropriately collected urine culture showing bacterialgrowth >105 colony-forming units/mL) – or acuteprostatitis – defined by the presence of fever, pyuria,and prostatic tenderness detected by gentle digitalrectal exploration. The diagnosis of acute pyelonephritiswas based on the simultaneous presence of fever andsignificant bacteriuria and/or BSI, along with 1 or moreof the following: lumbar or graft pain, chills, and/orcystitis (27). Intraabdominal infection was defined bythe presence of a fluid collection requiring antibiotictherapy and percutaneous or surgical drainage with apositive culture. CMV disease included either viralsyndrome or end-organ disease, and was definedfollowing the criteria proposed by Ljungman et al.(28). IFI was defined per the criteria proposed by theEuropean Organization on Research and Treatment inCancer and the Mycoses Study Group (29). Onlyproven and probable cases of IFI were included. Overallinfection was defined as any of the aforementionedcategories of infection. Infection-related mortality wasassessed by an infectious diseases physician notdirectly involved in the care of the patient (M.F.-R.)on the basis of the following criteria: (a) persistentisolation of pathogens from the infection sites at death;(b) death before resolution of symptoms and signs ofinfection; or (c) death occurring within 14 days afterthe diagnosis of infection without other attributablecauses. Acute graft rejection was suspected in case of anelevation of the serum creatinine and diagnosed byhistological examination, if possible (30). If renal biopsywas not technically possible, “intention-to-treat” epi-sodes that respond to anti-rejection therapy were alsotaken into account. Graft loss was defined as (a) primarynon-function in those grafts that never regained func-tion, with the patient remaining dialysis dependent; (b)return to long-term dialysis; or (c) retransplantation.

Statistical analysis

Quantitative data were shown as the mean � standarddeviation or the median with interquartile ranges(IQR). Qualitative variables were expressed as absoluteand relative frequencies. Categorical variables werecompared using the v2 test and Fisher’s exact test, asappropriate. The Student’s t-test (or Mann–WhitneyU-test when the assumption of normality did not hold)was applied for continuous variables. Serum ironparameters were explored as continuous variables,and subsequently categorized as binary variablesaccording to the median and 75-percentile values.Owing to its notable hour-to-hour within-subjectvariability (19, 20), we did not include serum iron

Transplant Infectious Disease 2013: 0: 1–12 3


levels in the analysis. Unadjusted hazard ratios (HRs)with 95% confidence intervals (CIs) were estimated foreach serum iron parameter and for time to develop-ment of the first episode of overall infection andbacterial infection during the first post-transplant year.Separate multivariate-adjusted HRs were calculated forthe most predictive markers by using stepwise back-ward Cox proportional hazards models that includedthose variables found to be significant at P ≤ 0.1 in theunivariate analysis. We entered serum albumin level atthe time of the assessment of iron parameters into allthe Cox-regression models in an attempt to account forthe role of ferritin as an acute-phase reactant (31, 32).The number of packed red blood cells units transfusedduring the perioperative period was also forced into themodel because of its clinical relevance (19). The overalland infection-free cumulative survival curves wereplotted by the Kaplan–Meier approach and comparedusing the log-rank test. All significance tests were2-tailed and considered as significant at P < 0.05. Wealso used the Bonferroni correction in order to avoidthe possible inflation of P-values owing to multiplecomparisons. The Bonferroni correction method is aconservative approach that sets the a value for eachcomparison equal to the fixed a value divided by thetotal number of comparisons. All the significance testswere 2-tailed. Statistical analysis was performed usingSPSS version 15.0 (SPSS, Inc., Chicago, Illinois, USA)and EPIDAT version 3.1 (Conselleria de Sanidade,Xunta de Galicia, Spain).

Results

A total of 265 KT recipients were originally enrolled inthe immune status assessment between November2008 and February 2011. Twenty-eight patients (10.6%)were excluded because of the lack of appropriate dataon serum iron parameters, as detailed in the flowdiagram (Fig. 1). Therefore, 228 patients with completeserum iron parameters, whose demographic and clin-ical characteristics are shown in Table 1, were finallyincluded in this study. Serum iron parameters wereassessed at a median interval of 3 days (IQR, 1–6 days)after transplantation, and at a median interval of26 days (IQR, 11–76 days) before the occurrence ofthe first infectious episode.The median follow-up period was 494 days (IQR,

401.2–747.7 days), and 191 patients (83.7%) achieved afollow-up of at least 12 months. Thirteen patients (5.7%)died through the first post-transplant year at a medianof 77 days from transplantation (IQR, 44.5–186 days),with 6 deaths (46.1%) considered to be infection-related.

One-year death-censored graft survival was 93.0%. Fifty-six patients (24.6%) experienced a total of 61 episodesof acute graft rejection during the first year (mean,1.1 � 0.3 episodes per patient), including 19 acutehumoral episodes (31.1%). A total of 147 patients(64.5%) suffered from 331 episodes of infection duringthe first post-transplant year (incidence rate, 4.2episodes per 1000 transplant-days). The distribution ofclinical syndromes and agents is shown in Table 2.As detailed in Table 3, recipients with post-transplant

infectious complications had higher ferritin levels andlower TIBC, UIBC, and transferrin levels (as inverseindicators of total body iron status) compared with theremaining cohort. Such differences reached statisticalsignificance for ferritin levels regarding both overall(P = 0.035) and bacterial infection (P = 0.010), as wellas for TIBC and UIBC regarding bacterial infection(P = 0.014 and 0.025, respectively). When the conser-vative Bonferroni correction was applied (P < 0.010),only differences in ferritin levels according to theoccurrence of bacterial infection remained as signifi-cant. No statistical differences in serum iron parame-ters were found according to all-cause mortality at1 year (data not shown).

265 patients initially included(November 2008 to February 2011)

28 patients (10.6%) excluded due to:

no assessment of serum iron parameters within the first 2 weeksoccurrence of infection before or at the time of serum sampling

228 patients analyzed

198 patients (86.8%) with one-year clinical follow-up

30 patients (13.2%) did not completeone-year follow-up:

13 patients (5.7%) died9 patients (3.9%) returned to dialysis due to loss of graft function8 patients (3.5%) lost to follow-up

Fig. 1. Patient flow diagram.



On multivariate analysis and after adjusting for thosecovariates found to be significant in the univariateanalysis (Table 4) and other clinically relevant vari-ables (i.e., number of transfused packed red blood cellsunits and previous transplant), serum ferritin level (HR[per unitary increment], 1.001; 95% CI, 1.000–1.001;P = 0.006), and UIBC (HR [per unitary increment],0.997; 95% CI, 0.994–0.999; P = 0.015) emerged asindependently associated with a higher risk of overallinfection. These associations did not meaningfullychange in a sensitivity analysis in which patients weredichotomized according to the time interval elapsedbetween the measurement of iron parameters and theonset of the first infection (≤14 or >14 days). Inaddition, a similar result was found between serumferritin level and bacterial infection (HR [per unitaryincrement], 1.001; 95% CI, 1.000–1.001; P = 0.020)(Table 5). By using the median value of serum ferritinas a cutoff (500 ng/mL), adjusted HRs for overall andbacterial infection among patients with increased ironstatus within the first 2 post-transplant weeks were1.627 (95% CI, 1.155–2.292; P = 0.005) and 1.661 (95%CI, 1.126–2.451; P = 0.011), respectively.We further explored the predictive value of serum

ferritin dichotomized by the median value. Patientswith serum levels ≥500 ng/mL experienced highercumulative incidences of both overall infection (72.8%vs. 58.8%; P = 0.030) and bacterial infection (59.8% vs.41.9%; P = 0.008) at the end of the first post-transplantyear as compared with the rest of the cohort. We alsofound a near-significant trend toward a higher meannumber of infectious episodes per recipient in thisgroup (1.65 � 1.61 vs. 1.32 � 1.61 episodes;P = 0.054). Likewise, statistically significant differencesin the same direction were detected in the incidencerates (expressed as number of episodes per 1000transplant-days) of overall infection (4.99 vs. 3.82;P = 0.017), bacterial infection (3.52 vs. 2.28;P = 0.002), and BSI (0.85 vs. 0.38; P = 0.011), with a

Demographic and clinical characteristics of the 228 kidney transplantrecipients included

Variable N %

Age of recipient, years, mean � SD 54.8 � 14.9

Gender: male 138 60.5

Pretransplant comorbidities

Diabetes mellitus 61 26.8

Heart disease 60 26.3

Chronic liver disease 12 5.3

Chronic lung disease 31 13.6

Previous solid organ transplantation 51 22.4

Etiology of underlying ESRD

Glomerulonephritis 45 19.7

Diabetic nephropathy 41 18.0

Hypertensive nephropathy 29 12.7

Policystosis 28 12.3

Chronic interstitial nephropathy 21 9.2

Unknown 19 8.3

Other 45 19.7

Baseline serostatus

Hepatitis B virus 1 0.4

Hepatitis C virus 26 11.4

CMV status D+/R� 18 7.9

CMV status D�/R� 3 1.3

Pre-transplant renal replacement therapy

Hemodialysis 185 81.1

Continuous ambulatory peritoneal

dialysis

27 11.8

Age of donor in years, mean � SD 52.1 � 17.4

Type of donor

Donation after brain death 151 66.2

Donation after circulatory death 67 29.4

Living donor 10 4.4

Number of HLA mismatches,

mean (IQR)

4 (4–5)

Cold ischemia time in h, mean � SD 16.8 � 6.7

Intraoperative transfusion

requirement

5 2.2

Packed red blood cells units,

mean (IQR)

2 (1–2)

Induction therapy

None 42 18.4

Basiliximab 80 35.1

Anti-thymocyte globulin 105 46.1

Investigational new drug (alefacept) 1 0.4

Table 1 Continued

Variable N %

Primary immunosuppression scheme

Tacrolimus, MMF, and steroids 201 88.2

Tacrolimus, azathioprine, and steroids 25 11.0

SD, standard deviation; ESRD, end-stage renal disease; CMV,

cytomegalovirus D, donor; R, recipient; HLA: human leukocyte

antigen; IQR, interquartile range; MMF, mycophenolate mofetil.

Table 1



non-significant trend for lower urinary tract infection(1.28 vs. 0.83; P = 0.072). No significant differenceswere seen in the incidence rates of CMV disease (1.02vs. 1.02; P = 0.929), end-organ CMV disease (0.23 vs.0.08; P = 0.184), and IFI (0.09 vs. 0.06; P = 0.886). Theinfection-free survival was significantly lower in recip-ients with higher level of serum ferritin (1-year survivalrates: 26% vs. 41%, respectively; log-rank P = 0.004)(Fig. 2A).By using the 75-percentile (700 ng/mL) as an alter-

native cutoff value for serum ferritin, we also founddifferences in the incidence rates (in number ofepisodes per 1000 transplant-days) of overall infection(5.2 vs. 3.9; P = 0.023), bacterial infection (3.6 vs. 2.4;P = 0.006), and BSI, although the latter did not achievestatistical significance (0.9 vs. 0.5; P = 0.060).Finally, those recipients with serum ferritin levels

≥500 ng/mL experienced a higher all-cause mortalityrate during the first post-transplant year compared withthe remaining patients, although without statisticalsignificance (1-year survival rates: 91% vs. 96%, respec-tively; log-rank P = 0.105) (Fig. 2B). Such a differencebecame significant when only infection-related mortal-ity was considered (1-year survival rates: 94% vs. 99%,respectively; log-rank P = 0.029).

Discussion

In this study, we demonstrated an independent associ-ation between the serum ferritin levels assessed in thevery early post-transplant period, and the incidence ofoverall and bacterial infection during the first year afterKT. In addition, we also found an inverse association

Distribution and etiology of the 331 episodes of post-transplantinfection

N %

Clinical syndromes1

Lower urinary tract infection 82 24.8

CMV viral syndrome 63 19.0

Non catheter-related bloodstream infection 44 13.3

Incisional surgical site infection 36 10.9

Acute pyelonephritis 32 9.7

Pneumonia and LRTI 31 9.4

Skin and soft-tissue infection 20 6.0

Catheter-related bloodstream infection 13 3.9

Intraabdominal infection 11 3.3

Clostridium difficile-associated diarrhea 10 3.0

Colitis 10 3.0

BK polyomavirus nephropathy 7 2.1

Osteoarticular infection 5 1.5

Esophagitis 3 0.9

Central nervous system infection 2 0.6

Visceral leishmaniasis 2 0.6

Others2 4 1.2

Microorganisms

Bacteria

Enterococcus species 32 9.7

Coagulase-negative staphylococci 12 3.6

Staphylococcus aureus 2 0.6

Viridans group streptococci 2 0.6

Escherichia coli 67 20.2

Klebsiella species 10 3.0

Other Enterobacteriaceae 20 6.0

Pseudomonas species 22 6.6

Other non-fermenting gram-negative bacilli 1 0.3

C. difficile 10 3.0

Mycobacterium tuberculosis complex 2 0.6

Polymicrobial 4 1.2

Other 3 0.9

Probable bacterial infection

(no microbiological diagnosis)

20 6.0

Viruses

CMV 79 23.7

Herpes simplex virus 1 and 2 10 3.0

Varicella-zoster virus 3 0.9

Other 13 3.9

Table 2 Continued

N %

Fungi

Candida species 14 4.2

Aspergillus fumigatus 2 0.6

Mucormycosis 1 0.3

Parasites (Leishmania donovani complex) 2 0.6

1The total number may be less than the sum of each syndrome

because >1 infection was simultaneously present in some patients

(i.e., bloodstream infection secondary to another documented

infection).2Infective endocarditis, orchiepididimitis, chorioretinitis, and

disseminated nocardiosis (1 case each).

CMV, cytomegalovirus; LRTI, lower respiratory tract infection.

Table 2



between UIBC values and these outcomes. Althoughany assumptions about causality should be viewed withcaution, our findings overall suggest that recipientswith increased body iron stores might be exposed to ahigher risk of post-transplant infectious complications,and are in line with those reported for patients withend-stage renal disease (ESRD) on HD therapy (21, 22)and other immunosuppressed hosts (13–19). More-over, higher serum ferritin levels were associated withan increased infection-related mortality at 1 year. Ofnote, our study is the first, to our knowledge, thatfocused on analyzing the predictive value of serum ironparameters in the specific population of KT recipients.Iron overload is an emerging predisposing factor for

infection after OLT (33). Stainable iron in the hepaticexplant specimen has been found to be an independentpredictor of post-transplant IFI (18) and Staphylococcusaureus BSI (34). Although direct measurement of tissueiron concentration in biopsy specimens (liver or bonemarrow) constitutes the gold standard for the measureof total body iron storage, its invasiveness limits its usein daily clinical practice (35). Thus, a number of serumparameters (such as ferritin levels or transferrin satu-ration ratio) have emerged as an alternative and easilyavailable approach. A recent study, based on 109 OLTrecipients, showed that increased serum iron parame-

ters assessed within the first week after the procedurewere an independent risk factor for post-transplantinfection and 1-year mortality (19). In addition, otherstudies have also demonstrated the prognostic value ofbaseline ferritin levels (as a surrogate marker of ironburden) in patients with hematologic malignanciesundergoing HSCT (13, 14, 16, 17, 31, 32). In a largecohort comprising more than 1200 HSCT recipients,Garc�ıa-Vidal et al. (16) found that iron overload,defined as serum ferritin >2000 ng/mL, was an inde-pendent risk factor only for early (<40 days aftertransplantation) but not for late or very late IFI. Serumferritin levels >700 ng/mL have also been linked to a4-fold increase in the risk of bacterial infection (17).In the present cohort, ferritin and UIBC were found to

act as the strongest predictors of infection among theserum iron parameters analyzed, in keeping with thestudy by Chow et al. (19). Recipients with serum ferritinlevels ≥500 ng/mL at baseline had a higher incidence ofoverall infection, bacterial infection, and BSI during thefirst post-transplant year, as well as a lower infection-free survival rate, compared with those with levelsbelow the cutoff. After adjusting for a number ofpotential confounders, including serum albumin levelas inverse acute-phase reactant, high ferritin levelsincreased risk of infection by 60% in the multivariate

Serum iron parameters in kidney transplant recipients according to the occurrence of infection during the first post-transplant year

Iron marker

(mean � SD) No (n = 81) Yes (n = 147) P-value1 Unadjusted HR2 (95% CI) P-value

Overall infection

Ferritin, ng/mL 466.7 � 343.6 599.3 � 479.7 0.035 1.001 (1.00–1.00) 0.023

TIBC, lg/dL 224.7 � 59.1 211.5 � 43.5 0.079 0.997 (0.99–1.00) 0.111

UIBC, lg/dL 165.5 � 65.8 150.5 � 60.7 0.085 0.998 (0.99–1.00) 0.120

Transferrin, mg/dL 177.1 � 145.5 154.2 � 51.8 0.180 0.998 (0.99–1.00) 0.998

Transferrin saturation,% 29.1 � 19.8 30.4 � 21.2 0.796 1.002 (0.99–1.01) 0.679

Iron parameter

(mean � SD) No (n = 116) Yes (n = 112) P-value1 Unadjusted HR2 (95% CI) P-value

Bacterial infection

Ferritin, ng/mL 487.9 � 410.1 618.8 � 461.5 0.010 1.001 (1.00–1.00) 0.022

TIBC, lg/dL 224.1 � 55.1 207.9 � 42.5 0.014 0.996 (0.99–1.00) 0.032

UIBC, lg/dL 165.0 � 62.3 146.3 � 62.2 0.025 0.997 (0.99–1.00) 0.038

Transferrin, mg/dL 170.7 � 122.7 153.6 � 57.0 0.158 0.998 (0.99–1.00) 0.257

Transferrin saturation,% 28.5 � 19.4 31.4 � 21.9 0.412 1.004 (0.99–1.01) 0.332

1Bonferroni adjustment, a = 0.010.2Per unitary increment.

SD, standard deviation; HR, hazard ratio; CI, confidence interval; TIBC, total iron-binding capacity; UIBC, unsaturated iron-binding capacity.

Table 3



model. It should be noted that the physiopathologicalimplication of hyperferritinemia in KT recipients mightdiffer from that in other types of immunosuppressedpatients. The so-called malnutrition-inflammation com-plex syndrome associated with ESRD on maintenanceHD leads to erythropoietin hyporesponsiveness andrefractory anemia (36). It has been suggested thatmoderately high levels of serum ferritin are not a meresurrogate of total body iron stores but also a compositeindicator of inflammation, malnutrition, liver dysfunc-tion, malignancy, and other factors (20). Even in thesetting of uniform iron-supplementation practices, hy-perferritinemia constitutes a reliable short-term markerof death and hospitalization during maintenance HD(37). In this line, serum ferritin levels >700 ng/mLpredicted both all-cause and infection-related 1-yearmortality in a cohort of diabetic HD patients (22), whichis a finding similar to ours. As suggested by Jenq et al.(22), this predictive value of ferritin levels for determin-

ing the hazard of death might lie, at least partially, in itsrole as a marker of inflammation status. On the otherhand, it has been suggested that iron overloadmay exerta deleterious effect on the atherosclerotic process byincreasing oxidative stress and promote b2 microglobu-lin amyloidosis in HD patients (38), thus directlycontributing to worsening outcome.Despite the complex multifactorial nature of this

parameter, the 2004 European Best Practice Guidelinesfor the management of anemia in patients with ESRDrecommend using serum ferritin as the standard test toevaluate iron stores, including those under erythropoi-etin-stimulating agent (ESA) therapy (39). Moreover,serum ferritin levels have demonstrated a good positivecorrelation with both transferrin saturation ratio (22)and tissue iron stores (40). Current guidelines adviseagainst IV iron administration to ESRD patients withferritin levels >500 ng/mL, as such high values mayreflect iron overload (41, 42).

Univariate analysis of risk factors for overall and bacterial infection (excluding serum iron parameters)

Unadjusted HR 95% CI P-value

Overall infection

Age of recipient, years1 1.030 1.018–1.042 0.000

Number of pre-transplant comorbidities1 1.286 1.101–1.503 0.001

Glomerulonephritis as underlying ESRD 0.669 0.432–1.037 0.072

Donation after circulatory death 0.588 0.402– 0.860 0.006

Cold ischemia time, hours1 1.031 1.006–1.056 0.016

Basiliximab as induction therapy 1.593 1.144–2.220 0.006

Serum albumin, g/dL1,2 0.665 0.468–0.944 0.023

Requirement of surgical reintervention within first month 2.226 1.396–3.551 0.001

AR within first month 1.734 1.147–2.623 0.009

Graft function (estimated by MDRD-4 equation) at month 1, mL/min1 0.979 0.970–0.987 0.000

Bacterial infection

Age of recipient, years1 1.028 1.014–1.042 0.000

Number of pre-transplant comorbidities1 1.300 1.089–1.551 0.004

Previous solid organ transplantation 1.432 0.942–2.177 0.093

Glomerulonephritis as underlying ESRD 0.631 0.376–1.058 0.081

Donation after circulatory death 0.605 0.388–0.945 0.027

Serum albumin, g/dL1,2 0.583 0.386–0.881 0.010

Requirement of surgical reintervention within first month 2.527 1.534–4.163 0.000

AR within first month 1.813 1.152–2.851 0.010

Graft function (estimated by MDRD-4 equation) at month 1, mL/min1 0.978 0.968–0.988 0.010

1Per unitary increment.2At the time of the assessment of iron parameters.

CI, confidence interval; ESRD, end-stage renal disease; HR, hazard ratio; AR, acute graft rejection; MDRD, Modification of Diet in Renal Disease.

Table 4



It has been long recognized that iron overloadincreases the pathogenicity of siderophilic bacteria,such as Klebsiella pneumoniae, Yersinia species, orVibrio vulnificus (4, 5), with anecdotal case reports inKT recipients (43). We found a higher incidence of BSIin KT recipients with increased ferritin levels. Most ofthese episodes were caused by enterobacteria, Pseudo-monas aeruginosa, and coagulase-negative staphylo-

cocci. Acquired transfusional iron overload in HDpatients has been associated with a greater risk ofBSI (44). Teehan et al. (21) reported a 2.5-fold higherHR for BSI among HD patients with an iron-repletestatus, as defined by the Dialysis Outcome QualityInitiative Anemia Work Group criteria (41) (ferritinlevel ≥100 ng/mL and transferrin saturation ratio≥20%), before initiating IV iron therapy.

Multivariate analysis of serum iron parameters as predictors of infection during the first post-transplant year (Cox proportional hazards models)

Overall infection Bacterial infection

Serum iron marker Adjusted HR1,2 95% CI P-value Adjusted HR1,3 95% CI P-value

Ferritin, ng/mL 1.001 1.000–1.001 0.006 1.001 1.000–1.001 0.020

TIBC, lg/dL 0.998 0.994–1.002 0.282 0.997 0.992–1.002 0.200

UIBC, lg/dL 0.997 0.994–0.999 0.015 0.996 0.993–0.999 0.006

1Per unitary increment.2Model adjusted for recipient age, number of pre-transplant comorbidities, glomerulonephritis as underlying ESRD, donation after circulatory

death, cold ischemia time, number of perioperatively transfused packed red blood cells units, basiliximab as induction therapy, reoperation within

first month, AR within first month, graft function at month 1 (estimated by MDRD-4 equation), and serum albumin.3Model adjusted for recipient age, number of pre-transplant comorbidities, previous solid organ transplantation, glomerulonephritis as underlying

ESRD, donation after circulatory death, number of perioperatively transfused packed red blood cells units, reoperation within first month, AR

within first month, graft function at month 1 (estimated by MDRD-4 equation), and serum albumin.

HR, hazard ratio; CI, confidence interval; TIBC, total iron-binding capacity; UIBC, unsaturated iron-binding capacity; ESRD, end-stage renal

disease; AR, acute graft rejection; MDRD, Modification of Diet in Renal Disease.

Table 5

A B

Fig. 2. Kaplan–Meier curves during the first post-transplant year stratified by serum ferritin levels: (A) infection-free survival (log-rank test;

P = 0.004); (B) patient survival (log-rank test; P = 0.105).



It has been proven in vitro that iron enhancesendothelial cell activation in response to CMV infection(45), which may be inhibited by iron chelators such asdeferoxamine (46). In the clinical arena, Chow et al.(19) also found a relationship between serum ferritinand the occurrence of CMV disease after OLT. In ourcohort, those recipients with serum ferritin ≥500 ng/mL experienced a slightly higher incidence rate ofCMV end-organ disease (0.23 vs. 0.08 episodes per1000 transplant-days), but this did not achieve statisticalsignificance. Finally, despite the broad evidence on theassociation between iron overload and IFI (6, 7, 15, 16,18), we failed to demonstrate the impact of serum ironparameters on this outcome in our study, although thelow number of events should be noted (3 invasivecandidiasis, 3 pulmonary aspergillosis, and 1 mucormy-cosis in the whole cohort).Some limitations of this study are worth addressing.

As mentioned, the use of serum ferritin as a surrogatemarker of iron storage in ESRD patients is subject toinherent limitations, although its availability and non-invasiveness explain its widespread use in the litera-ture (22, 37, 44). Serum iron parameters wereassessed at a single baseline time point with nosequential follow-up samples. A number of patientsundergoing KT during the study period in ourinstitution lacked appropriate serum samples for theanalysis (10.6%). Only 12 and 4 patients had serumferritin levels >1000 and 2000 ng/mL, respectively,precluding the analysis of higher cutoff values thatmay have been more consistently related to ironoverload (20, 47). We did not take into account theimpact on that issue of IV iron or ESA therapiesadministered before transplantation. Our sample sizemight be underpowered to detect differences inspecific infectious syndromes and microorganisms(i.e., CMV disease or IFI).Finally, the assessment of hepcidin levels would be

highly valuable in order to confirm this potentialassociation between iron overload and infection in KTrecipients. The hormone hepcidin mediates the sys-temic regulation of dietary iron absorption anddecreases the availability of free circulating iron toinvading microorganisms in response to various inflam-matory signals, mainly interleukin-6 (35). Therefore,hepcidin measurements could reliably distinguishwhether serum iron parameters are increased becauseof iron overload or ongoing infection/inflammation.Notwithstanding all these limitations, the plausibility

of our results is supported by the independent associ-ation also found between infection and UIBC, the timeinterval elapsed from the serum iron assessment to thediagnosis of infection, and the inclusion of serum

albumin levels as inverse acute-phase reactant in all themultivariate models. Serum albumin exhibits a strongand negative correlation with C-reactive protein in HDpatients (47). Moreover, no patients with active infec-tion underwent transplantation, thus minimizing therole of serum ferritin as a simple acute-phase reactantin response to an incipient inflammatory process.In conclusion, our study suggests that the assess-

ment of serum ferritin levels in the very early periodafter KT may identify a group of patients at higherrisk of infectious complications. Regardless of theprecise biological significance of this parameter – areflection of total body iron stores or a composite ofunderlying inflammation, malnutrition, and iron over-load – future investigations should confirm thesepreliminary findings, and further examine the role ofiron metabolism in the pathogenesis of infection in KTrecipients.

Acknowledgements:

Presentation: This study was partially presented at the24th International Congress of The TransplantationSociety, Berlin (July 15–19, 2012).Funding sources: This study was supported by the

Spanish Ministry of Science and Innovation (Fondo deInvestigaciones Sanitarias [FIS] project 11/01538) andthe Fundaci�on Mutua Madrile~na de Investigaci�onM�edica (FMM Grant 2010/0015). Mario Fern�andez-Ruiz holds a research training contract “Rio Hortega”(CM11/00187) from the Spanish Ministry of Scienceand Competitiveness (Instituto de Salud Carlos III).Authors contributions: M.F.-R. designed research,

collected data, analyzed data, and wrote the paper;F.L.-M. designed research, analyzed data, and revisedand completed the final draft of the manuscript; J.M.A.designed research and revised and completed the finaldraft of the manuscript. E.G. and N.P. collected data;A.A. collected data and revised and completed the finaldraft of the manuscript; R.S.-J. analyzed data andrevised and completed the final draft of the manuscript;and J.M.M. and C.L. revised and completed the finaldraft of the manuscript.Conflicts of interest: All authors, no conflicts.

References

1. Weinberg ED. Iron availability and infection. Biochim BiophysActa 2009; 1790 (7): 600–605.

2. Skaar EP. The battle for iron between bacterial pathogens andtheir vertebrate hosts. PLoS Pathog 2010; 6 (8): e100094.



3. Bullen JJ, Rogers HJ, Spalding PB, Ward CG. Natural resistance,iron and infection: a challenge for clinical medicine. J MedMicrobiol 2006; 3 (Pt 3): 251–258.

4. Bullen JJ, Spalding PB, Ward CG, Gutteridge JM.Hemochromatosis, iron and septicemia caused by Vibrio

vulnificus. Arch Intern Med 1991; 151 (8): 1606–1609.5. Vento S, Cainelli F, Cesario F. Infections and thalassaemia. Lancet

Infect Dis 2006; 6 (4): 226–233.6. Caroline L, Rosner F, Kozinn PJ. Elevated serum iron, low

unbound transferrin and candidiasis in acute leukemia. Blood1969; 34 (4): 441–451.

7. Artis WM, Fountain JA, Delcher HK, et al. A mechanism ofsusceptibility to mucormycosis in diabetic ketoacidosis:transferrin and iron availability. Diabetes 1982; 31 (12):1109–1114.

8. Boelaert JR, Fenves AZ, Coburn JW. Deferoxamine therapy andmucormycosis in dialysis patients: report of an internationalregistry. Am J Kidney Dis 1991; 18 (6): 660–667.

9. Boelaert JR, de Locht M, Van Cutsem J, et al. Mucormycosisduring deferoxamine therapy is a siderophore-mediated infection.In vitro and in vivo animal studies. J Clin Invest 1993; 91 (5):1979–1986.

10. Walker EM Jr, Walker SM. Effects of iron overload on theimmune system. Ann Clin Lab Sci 2000; 30 (4): 354–365.

11. Eiselt J, Kielberger L, Sedl�ackov�a T, Racek J, Pazdiora P. Highferritin, but not hepcidin, is associated with a poor immuneresponse to an influenza vaccine in hemodialysis patients.Nephron Clin Pract 2010; 115 (2): c147–c153.

12. van Asbeck BS, Marx JJ, Struyvenberg A, Verhoef J. Functionaldefects in phagocytic cells from patients with iron overload. JInfect 1984; 8 (3): 232–240.

13. Alt�es A, Remacha AF, Sureda A, et al. Iron overload mightincrease transplant-related mortality in haematopoietic stem celltransplantation. Bone Marrow Transplant 2002; 29 (12):987–989.

14. Pullarkat V, Blanchard S, Tegtmeier B, et al. Iron overloadadversely affects outcome of allogeneic hematopoietic celltransplantation. Bone Marrow Transplant 2008; 42 (12): 799–805.

15. Kontoyiannis DP, Chamilos G, Lewis RE, et al. Increased bonemarrow iron stores is an independent risk factor for invasiveaspergillosis in patients with high-risk hematologic malignanciesand recipients of allogeneic hematopoietic stem celltransplantation. Cancer 2007; 110 (6): 1303–1306.

16. Garcia-Vidal C, Upton A, Kirby KA, Marr KA. Epidemiology ofinvasive mold infections in allogeneic stem cell transplantrecipients: biological risk factors for infection according to timeafter transplantation. Clin Infect Dis 2008; 47 (8): 1041–1050.

17. Kanda J, Mizumoto C, Ichinohe T, et al. Pretransplant serumferritin and C-reactive protein as predictive factors for earlybacterial infection after allogeneic hematopoietic celltransplantation. Bone Marrow Transplant 2011; 46 (2): 208–216.

18. Alexander J, Limaye AP, Ko CW, Bronner MP, Kowdley KV.Association of hepatic iron overload with invasive fungal infectionin liver transplant recipients. Liver Transpl 2006; 12 (12):1799–1804.

19. Chow JK, Werner BG, Ruthazer R, Snydman DR. Increasedserum iron levels and infectious complications after livertransplantation. Clin Infect Dis 2010; 51 (3): e16–e23.

20. Kalantar-Zadeh K, Kalantar-Zadeh K, Lee GH. The fascinating butdeceptive ferritin: to measure it or not to measure it in chronickidney disease? Clin J Am Soc Nephrol 2006; 1 (Suppl 1): S9–S18.

21. Teehan GS, Bahdouch D, Ruthazer R, Balakrishnan VS, SnydmanDR, Jaber BL. Iron storage indices: novel predictors of bacteremia

in hemodialysis patients initiating intravenous iron therapy. ClinInfect Dis 2004; 38 (8): 1090–1094.

22. Jenq CC, Hsu CW, Huang WH, Chen KH, Lin JL, Lin-Tan DT.Serum ferritin levels predict all-cause and infection-cause 1-yearmortality in diabetic patients on maintenance hemodialysis. Am JMed Sci 2009; 337 (3): 188–194.

23. Fern�andez-Ruiz M, L�opez-Medrano F, Varela-Pe~na P, et al.Monitoring of immunoglobulin levels identifies kidney transplantrecipients at high risk of infection. Am J Transplant 2012; 12 (10):2763–2773.

24. Levey AS, Bosch JP, Lewis JB, Greene T, Rogers N, Roth D. Amore accurate method to estimate glomerular filtration rate fromserum creatinine: a new prediction equation. Modification of Dietin Renal Disease Study Group. Ann Intern Med 1999; 130 (6):461–470.

25. Garner JS, Jarvis WR, Emori TG, Horan TC, Hughes JM. CDCdefinitions for nosocomial infections, 1988. Am J Infect Control1988; 16 (3): 128–140.

26. Mermel LA, Allon M, Bouza E, et al. Clinical practice guidelinesfor the diagnosis and management of intravascular catheter-related infection: 2009 Update by the Infectious Diseases Societyof America. Clin Infect Dis 2009; 49 (1): 1–45.

27. Rubin RH. Infection in the organ transplant recipient. In: RubinRH, Young LS, Russell P, eds. Clinical Approach to Infection inthe Compromised Host (3rd edn). New York: Plenum Publishing,1981; 629–705.

28. Ljungman P, Griffiths P, Paya C. Definitions of cytomegalovirusinfection and disease in transplant recipients. Clin Infect Dis 2002;34 (8): 1094–1097.

29. De Pauw B, Walsh TJ, Donnelly JP, et al. Revised definitions ofinvasive fungal disease from the European Organization forResearch and Treatment of Cancer/Invasive Fungal InfectionsCooperative Group and the National Institute of Allergy andInfectious Diseases Mycoses Study Group (EORTC/MSG)Consensus Group. Clin Infect Dis 2008; 46 (12): 1813–1821.

30. European Expert Group on Renal Transplantation (EBPG);European Renal Association (ERA-EDTA); European Society forOrgan Transplantation (ESOT). European best practiceguidelines for renal transplantation (part 1). Nephrol DialTransplant 2000; 15 (Suppl 7): 1–85.

31. Armand P, Kim HT, Cutler CS, et al. Prognostic impact ofelevated pretransplantation serum ferritin in patients undergoingmyeloablative stem cell transplantation. Blood 2007; 109 (10):4586–4588.

32. Lim ZY, Fiaccadori V, Gandhi S, et al. Impact of pre-transplantserum ferritin on outcomes of patients with myelodysplasticsyndromes or secondary acute myeloid leukaemia receivingreduced intensity conditioning allogeneic haematopoietic stemcell transplantation. Leuk Res 2010; 34 (6): 723–727.

33. Singh N, Sun HY. Iron overload and unique susceptibility of livertransplant recipients to disseminated disease due to opportunisticpathogens. Liver Transpl 2008; 14 (9): 1249–1255.

34. Singh N, Wannstedt C, Keyes L, et al. Hepatic iron content andthe risk of Staphylococcus aureus bacteremia in liver transplantrecipients. Prog Transplant 2007; 17 (4): 332–336.

35. Fleming RE, Ponka P. Iron overload in human disease. N Engl JMed 2012; 366 (4): 348–359.

36. Kalantar-Zadeh K, Ikizler TA, Block G, Avram MM, Kopple JD.Malnutrition-inflammation complex syndrome in dialysis patients:causes and consequences. Am J Kidney Dis 2003; 42 (5): 864–881.

37. Kalantar-Zadeh K, Don BR, Rodriguez RA, Humphreys MH.Serum ferritin is a marker of morbidity and mortality inhemodialysis patients. Am J Kidney Dis 2001; 37 (3): 564–572.



38. Valenti L, Valenti G, Como G, et al. HFE gene mutations andoxidative stress influence serum ferritin, associated with vasculardamage, in hemodialysis patients. Am J Nephrol 2007; 27 (1):101–107.

39. Locatelli F, Aljama P, B�ar�any P, et al. Revised European BestPractice Guidelines for the management of anaemia in patientswith chronic renal failure. Nephrol Dial Transplant 2004; 19(Suppl 2): ii1–ii47.

40. B�ar�any P, Eriksson LC, Hultcrantz R, Pettersson E, Bergstr€om J.Serum ferritin and tissue iron in anemic dialysis patients. MinerElectrolyte Metab 1997; 6: 273–276.

41. National Kidney Foundation. K/DOQI clinical practice guidelinesand clinical practice recommendations for anemia in chronickidney disease. Am J Kidney Dis 2006; 47 (Suppl 3): S16–S45.

42. National Institute for Health and Clinical Excellence (NICE).Anaemia management in people with chronic kidney disease(CG114), 2011. Availabel at: http://guidance.nice.org.uk/CG114(accessed on 25 July 2012).

43. Fakir M, Wong T, Toupance O, Lavaud S, Wynckel A, Chanard J.Yersinia enterocolitica septicemia in a renal transplant recipient

on oral iron supplementation. Transplantation 1995; 60 (2):206–208.

44. Boelaert JR, Daneels RF, Schurgers ML, Matthys EG, Gordts BZ,Van Landuyt HW. Iron overload in haemodialysis patientsincreases the risk of bacteraemia: a prospective study. NephrolDial Transplant 1990; 5 (2): 130–134.

45. Kartikasari AE, Georgiou NA, de Geest M, et al. Iron enhancesendothelial cell activation in response to cytomegalovirus orChlamydia pneumoniae infection. Eur J Clin Invest 2006; 36 (10):743–752.

46. Cinatl J, Scholz M, Weber B, et al. Effects of desferrioxamine onhuman cytomegalovirus replication and expression of HLAantigens and adhesion molecules in human vascular endothelialcells. Transpl Immunol 1995; 3 (4): 313–320.

47. Kalantar-Zadeh K, Rodriguez RA, Humphreys MH. Associationbetween serum ferritin and measures of inflammation, nutritionand iron in haemodialysis patients. Nephrol Dial Transplant 2004;19 (1): 141–149.



Serum iron parameters in the early post-transplant period and infection risk in kidney transplant...

Documents

Transcript of Serum iron parameters in the early post-transplant period and infection risk in kidney transplant...