J Cutan Pathol 2009: 36: 1244–1254 Copyright © 2009 John Wiley & Sons A/Sdoi: 10.1111/j.1600-0560.2009.01283.xJohn Wiley & Sons. Printed in Singapore Journal of

Cutaneous Pathology

Gadolinium-Induced NephrogenicSystemic Fibrosis Is Associated withInsoluble Gd Deposits in Tissues: InVivo Transmetallation Confirmed byMicroanalysisBackground: Nephrogenic systemic fibrosis (NSF) is an extremelydebilitating systemic fibrosing disorder affecting renal failure patients.The association of NSF with gadolinium (Gd) containing magneticresonance contrast agents was noted in 2006. Gd deposition in skinbiopsies was demonstrated shortly thereafter.Methods: We used automated scanning electron microscopy(SEM)/energy dispersive x-ray spectroscopy for in situ quantitativeanalysis of insoluble Gd-containing deposits, recordingmulti-elemental composition and spatial distribution of detectedfeatures.Results: Gd was detected in all 29 patients (53 of 57 skin biopsies)with NSF, biopsied from 2 weeks to 3 years after Gd exposure. Gdconcentration ranged from 1 to 2270 cps/mm2 and was detectedpredominantly in the deep dermis and subcutaneous fibrous septa. Gdwas found associated with Ca, P and sometimes Fe or Zn. Patientswith sequential biopsies showed persistence or increase of Gd intissues (6 of 11). Transmission electron microscopy (TEM) identifiedthe intracellular deposits in fibrocytes and macrophages.Conclusions: The demonstration of insoluble tissue deposits of Gdwith co-associated elements clearly confirms in vivo transmetallationand dissociation of soluble Gd-chelates. Toxic Gd3+ may triggerfibrosis under permissive conditions, e.g., in renal insufficiency.Pathologists and clinicians need to be aware of this serious butpreventable disease.

Thakral C, Abraham JL. Gadolinium-Induced Nephrogenic SystemicFibrosis Is Associated with Insoluble Gd Deposits in Tissues: In VivoTransmetallation Confirmed by Microanalysis.J Cutan Pathol 2009; 36: 1244–1254. © 2009 John Wiley & SonsA/S.

Charu Thakral and JerroldL. AbrahamDepartment of Pathology, SUNY UpstateMedical University, Syracuse, NY, USA

Jerrold L. Abraham, MD, Department of Pathology,SUNY Upstate Medical University, 750 E. Adams St.,Syracuse, NY 13210, USATel: + 315 464 4750Fax: + 315 464 7130e-mail: abrahamj@upstate.edu

Accepted for publication February 5, 2009

IntroductionIn the past decade, the medical field has witnessedthe emergence of a novel disease, nephrogenicsystemic fibrosis (NSF). Initially recognized in skinand termed nephrogenic fibrosing dermopathy, NSF

is now known to be an extremely debilitating multi-systemic fibrosing disorder.1 Our knowledge andunderstanding of NSF has evolved in terms ofclinical manifestations, as well as etiology. Clinically,NSF starts with pain, swelling, pruritus, and skin

1244

Gd induced nephrogenic systemic fibrosis

Fig. 1. The forearm of the patient manifesting the thickened, brawnyplaques of NSF. Later extreme, deep contractures developed inthe hands, severely limiting the movements. (from Thakral et al.,4

Contrast Media Mol Imaging, by permission).

erythema, usually beginning on the hands and feet,and extending proximally. With time, the skindevelops nodules, patches or confluent regions ofhyperpigmentation with associated skin thickening,brawny induration and/or peau d’orange changes.2

Sometimes patients may develop deep bone pain,myalgia, muscle weakness, extensive joint stiffness,and contractures (Fig. 1). Extracutaneous fibrosisinvolving the heart, lungs, diaphragm, skeletalmuscle, liver, genitourinary tract, and central nervoussystem has been reported.3

Histologically, there is extensive dermal fibrosiswith an altered pattern of collagen bundles with sur-rounding clefts and increased number of spindle cells(Fig. 2). These dermal spindle cells are positive forCD34 and procollagen-1 surface markers, identicalto ‘circulating fibrocytes’, which are known to beassociated with scar formation and wound healing.5

Mucin is variable in the dermis and there is no sig-nificant inflammation. The subcutaneous septa aremarkedly thickened with fibrosis extending throughthe lobular septa into the underlying fascia and mus-cle. Occasionally, CD68+ histiocytes are present,and multinucleated giant cells, foci of osteoid depo-sition and/or calcification can be seen. Factor XIIIapositive dendritic cells are noted in a few cases.

Since its initial description as a scleromyxedema-like disease in dialysis patients, NSF has been reportedin both chronic and acute kidney disease patients.6

Initially no causative agents were identified. Severalfactors including erythropoietin, dialysis membranes,and thromboembolic events were studied, looking forassociation with this new disease. In January 2006,Grobner suggested a relationship between gadolin-ium (Gd)-containing contrast agents and NSF.7 Hereported a case series of five patients, all of whomreceived a contrast medium containing Gd prior tothe manifestations of NSF. Since then, other casereports and case control studies in the dermatologyand nephrology literature have strengthened the linkbetween exposure to Gd and NSF.8,9

Gadolinium is a trivalent ion with high toxicity. Itcan inhibit calcium-activated enzymes, block voltage-gated calcium channels and can induce macrophageapoptosis.10,11 The paramagnetic properties and along relaxation time make Gd a very useful agentfor contrast in MR imaging. Gd ions are bound withorganic molecules to provide a stable and, therefore,theoretically non-toxic complex. The Gd-chelatecomplex is soluble and almost all the commerciallyavailable contrast agents are excreted renally.12 How-ever, they differ in their dissociation constants; thislikely accounts for the varying probability of differentcontrast agents to induce NSF.

The exact mechanisms of development of NSF arenot completely clear at present, yet several piecesof the puzzle fit together in a biologically plausiblemanner. Figure 3 shows the suggested pathogenesisbased on the current understanding of this disease.Gd chelates are administered intravenously andexcreted unchanged by passive glomerular filtration(t1/2 = 1.5 h with normal renal function). In renalinsufficiency, the clearance of Gd chelates is reducedand its elimination half-life may exceed 30 h.12

The increased retention time of Gd chelates in thebody disturbs the equilibrium between ‘chelated Gd’and ‘free Gd’. Importantly, it allows free Gd ionto dissociate from its chelated form. The factorsinfluencing this equilibrium are complex but includethe stability of the chelate, pH, presence of competingmetals (e.g., Ca, Zn, Fe, Cu) which may favor‘transmetallation’, serum calcium, serum phosphateand pro-inflammatory conditions.13

Transmetallation is recognized as a key factorin the development of NSF. It is defined as thedisplacement of Gd by other cations (Ca, Fe, Zn)from the chelate with liberation of free Gd3+ ions.13

The released Gd3+ can form soluble complexes withligands present in the plasma. When the complexingability of the ligands present in plasma is exceeded,‘insoluble’ colloidal precipitates are formed with

1245

Thakral & Abraham

(a) (b)

(c)

Fig. 2. Skin biopsy. (A) Low-power image, H&E. Dense dermal fibrosis is present under a normal epidermis. The infiltrative process is deep,extending into subcutaneous interlobular septa. (B) High power image, H&E. Disorganized collagen bundles are separated by large clefts andsurrounded by numerous spindle cells (fibrocytes). Minimal perivascular inflammatory infiltrate is present. C, CD34 stain. Note the prominentpositive staining of fibrocytes in the dermis.

tissue anions. Thus with increasing concentrationof Gd3+ ions in the blood, insoluble precipitates withhydroxides, carbonates and phosphates are formed.14

Microanalytical techniques are ideally suited foridentification and chemical and morphologic char-acterization of elements in tissue. The technique of

energy dispersive x-ray spectroscopy (EDS) combinedwith scanning electron microscopy (SEM) is one suchsystem that allows identification of elemental deposits(soluble and insoluble) in cryo/fresh frozen tissues,and insoluble deposits in fixed and processed tis-sues. Utilizing SEM/EDS, we identified insoluble

Soluble Gd-chelates NOT

completely cleared in renal insufficiency

Release of TOXIC Gd3+

Gd3+

precipitates with anions (PO4

2−) as INSOLUBLE tissue deposits

Target or trigger for CIRCULATING

FIBROCYTES

IV Gd-containing contrast agents

FIBROSIS

Transmetallation

Fig. 3. Schematic of the developmentof NSF following administration of Gd-containing contrast agents in patientswith renal insufficiency. The proposedschema is based on the existing knowledgeof pharmacokinetics of Gd-chelates andcurrent understanding of NSF.

1246

Gd induced nephrogenic systemic fibrosis

Gd-containing deposits in paraffin embedded skinbiopsies involved with NSF.15 Further work led usto develop and apply a novel automated SEM/EDSmethod for in situ quantification of Gd-containingdeposits in tissues involved with NSF.16 This method-ology also provides information about the spatialdistribution and the association of other elements inGd-containing deposits. Our goal was to quantify theamount of detectable Gd, identify the tissue distribu-tion and localization of Gd deposits, and examine therole of transmetallation and association of detectedGd with the course of disease.

Materials and methodsFifty-seven tissue samples (skin biopsies) obtainedfrom 29 patients with a clinical and histopathologi-cal diagnosis of NSF were studied using SEM/EDS.Forty-one of these biopsies were from our analysesof the Danish series of Marckmann et al., four otherblocks were from cases reported,4,6,17 and the remain-ing were referred from other sources. The tissueswere processed as routine diagnostic histologic speci-mens including fixation with 10% buffered formalin,processing with alcohol, xylene, and embedding inparaffin wax. Tissue processing eliminates solublecomponents and allows examination of only remain-ing insoluble deposits. We prepared a surface-cutsection from the paraffin blocks for histological com-parison with the SEM/EDS analysis of the freshly cutsurface. Freshly cut paraffin blocks were next placedin the SEM and examined under variable pressuremode. This allows examination of non-conductivesamples without any additional preparation.

In the SEM, the primary beam of focused electronsinteracts with the sample resulting in emissionof many signals, including secondary electrons,backscattered electrons and x-rays. Backscatteredelectron images display atomic number contrastbased on elemental composition and EDS displayscharacteristic peaks for each element correspondingto the possible transitions in its electron shells inresponse to interaction with the electron beam.We utilized the Automated Feature Analysis (AFA)application software of the ASPEX� SEM; thedetailed methodology has been described earlier.16

Briefly, the area to be analyzed was selected andthe elements of interest were defined. We analyzeddermis and subcutaneous tissue for Na, Mg, K,Ca, Fe, Ti, Si, Al, P and Gd. The standardizedoperating conditions (20 keV accelerating voltage, 0.5nA specimen current, 15–16 mm working distanceand 0.15 torr pressure in the sample chamber) wereemployed for analysis of all the paraffin embeddedtissues.

The relatively high atomic number inorganicmaterials were revealed in the backscattered images inthe low atomic number organic background of tissueand paraffin. The x-ray spectra were collected for 30 sor a maximum of 7000 x-ray counts. These spectrawere compared with standard reference spectra ofvarious elements collected under similar standardizedconditions.The x-ray spectra provide data on all theelements associated in a detected feature and alsotheir relative percentages. We calculated counts persecond (cps) for each element in the analyzed features.The concentration of each element (per unit area) wasexpressed by dividing the sum of cps of that elementby the total area scanned as cps/mm2.

Random search of the tissue allows in situ detectionand semi-quantitative morphometric analysis with aspatial resolution of less than 1 μm. We have identi-fied deposits as small as 0.25 μm by this method. Thedetection limit for this SEM/EDS method is approxi-mately 0.1–1.0% by weight in a given area probed bythe electron beam; smaller or locally lower concen-tration Gd-containing deposits cannot be detected.However, it can search much greater tissue volumesthan can be searched using transmission electronmicroscopy (TEM) ultrathin sections (<100 nm) withelectron energy loss spectroscopy.18

Ultrastructural analysis to further characterizeand localize the tissue Gd was undertaken usingTEM. Portions of paraffin embedded tissue fromtwo tissue samples were reprocessed for epoxy resinembedding, with semi-thin sections (0.5 μm) forSEM imaging and thin sections for TEM. Tissueswere deparaffinized and post fixed with osmiumtetroxide and embedded in epoxy resin. Some sectionswere examined without any additional staining andothers were post stained with lead citrate and uranylacetate. Semi-thin sections for SEM were mountedon carbon discs and analyzed using backscatteredelectron imaging and EDS.

ResultsAutomated analysis by SEM/EDS revealed detect-able Gd in 53 of the 57 skin tissues obtained from29 NSF patients. Every patient had at least onebiopsy showing Gd. There was a wide range oftissue Gd concentration, ranging from 1 to 2270cps/mm2 (Fig. 4). Gd was never identified alone. Gdwas always associated with Na, Ca and P in tissue.Sometimes Fe or Zn was also detected in these co-precipitated deposits (Fig. 5). Overall, these depositsin the tissue seem to be apatite-like complexes withvarious elements incorporated in them.

All patients had received at least one dose of Gdcontaining contrast agent. The time period from Gdexposure to performance of biopsy varied from 2

1247

Thakral & Abraham

Gd concentration in 57 biopsies

0.00

0.50

1.00

1.50

2.00

2.50

3.00

3.50

4.00

0 10 20 30 40 50 60

biopsy ranking

log

Gd

cp

s/m

m2

Fig. 4. Plot of Gd concentration in the 57 analyzed skin biopsies, ranked in order of increasing Gd concentration. Gd cps/mm2 is shown ona log scale. Note the four biopsies in which Gd was not detected are included at the zero axis in this log display. The median value is close tolog10 = 2.0 (=100 Gd cps/mm2).

Fig. 5. SEM backscattered electron image of cut surface of paraffin-embedded skin biopsy involved with NSF. Higher atomic number features(brighter) are detected in dermis and subcutaneous tissue in a background of paraffin. The tissue processing allows in situ detection only ofremaining insoluble features. EDS spectra of individual deposits reveals Gd, P, Ca, Na and sometimes Fe.

1248

Gd induced nephrogenic systemic fibrosis

.1

.1

.1

.2

.2

.2

.3

.3

.3

.4

.4

.4

.5

.5.5

.6

.6

.6

.7

.7

.7

.8

.8

.8

.9

.9

.9

P

Gd

CaP

Bx 1 = redBx 2 (7 mths after Bx1) = greenBx 3 (38 mths after Bx1) = blue

Fig. 6. Ternary diagram showing distribu-tion of relative concentration of Gd, Caand P in Gd-containing deposits in threebiopsies from same NSF patient [biopsy 1= red, biopsy 2 (7 months after biopsy 1) =green, biopsy 3 (38 months after biopsy 1)= blue]. Note the gradual relative increasein Gd% in detected features over the periodof 3 years (from Thakral et al.4).

weeks to 3 years. Out of 29 patients, 11 patientshad sequential biopsies over time. Seven of theseeleven showed an increase or persistence of tissue Gdconcentration with no further exposure to Gd. Thetissue deposits showed varying ratios of Gd to Ca indifferent biopsies, with some deposits relatively richin Gd or Ca, and some having similar amounts ofGd and Ca. A ternary diagram presents the relativeconcentrations of the three elements in the tissue,and clearly depicts the varying proportions of Gd,Ca, and P in different biopsies. Figure 6 displays theternary presentation of Gd, Ca and P in one of theseverely affected patient with NSF. It is interestingto note that there was a gradual increase in relativeconcentration of Gd in tissue deposits compared toCa and P over a period of 3 years and no intervalexposure to Gd. We have noted such increase in Gdand/or decrease in Ca in more than one case.

Another significant finding was the distribution ofGd in tissue. The overlay mapping clearly demon-strates the non-uniform distribution, with the majorityof Gd deposits found more in the subcutaneousfibrous septa than in the superficial dermis (Fig. 7).Ultrastructural analysis confirmed the depositionof Gd both in extra and intracellular sites. TEMidentified Gd containing deposits in tissues anddefined these dense deposits in intracellular loca-tion. The CD34-positive elongated spindle cells withoval nuclei (fibrocytes) were identified and multipledense deposits were demonstrated in their cytoplasm

(Fig. 8). These deposits appear to be aggregates ofco-precipitated elements, deposited in an irregularfashion, both at the cellular and tissue scale. Thesize of these dense cytoplasmic deposits varied from100 to 400 nm, some forming even larger aggre-gates (Fig. 9). Some of the deposits were seen in themono- and multi-nucleated macrophages (Fig. 10).Semi-thin sections of tissue on carbon discs observedwith SEM backscattered electron imaging and EDSconfirm that these dense intra-cytoplasmic depositscontain Gd along with Ca and P (Fig. 8 and 10).

DiscussionWith rapidly growing epidemiologic evidence, tissueGd analysis and ultrastructural studies, it is gainingwider recognition that this mysterious disease is essen-tially the manifestation of toxicity of Gd released fromGd containing contrast agents.19 Tissue Gd analysisby our method clearly supports this. First, as thedetectable tissue Gd is insoluble, the Gd-chelate,which is soluble, must have dissociated. It is, infact, indirect but strong evidence of transmetallation.Second, the liberated Gd3+ formed insoluble tissuedeposits, as predictable from the pharmacokinetics ofGd. Some may propose that Gd deposition is a mereassociation and its presence or persistence might notbe a matter of concern. The strong epidemiologic evi-dence with an indisputable temporal association anddose response relationship argues against that.20,21 In

1249

Thakral & Abraham

Fig. 7. Histology (H&E image, left) and distribution map (SEM image, right) of detected Gd in the same tissue. Note the uneven distribution ofdetectable Gd features (blue), with most of detected features in deep dermis and subcutaneous fibrous septa. (Modified from Abraham et al.17).

(a)

(b)

(c)

Fig. 8. (A) TEM (thin section) image shows two elongated spindle cells (fibrocytes) containing electron dense deposits in the cytoplasm. Highermagnification detail of the area in box is shown in (B). The dense deposits contain Gd is confirmed by semi-thin section, SEM backscatteredelectron image (C). Here ‘‘+’’ represents one such feature in cytoplasm of spindle cells analyzed by SEM. EDS spectrum reveals detectable Gdand co-associated elements (P, Ca, Na).

1250

Gd induced nephrogenic systemic fibrosis

(a) (b)

Fig. 9. TEM, thin section. (A) Demonstrates irregular dense deposits in cytoplasm of fibrocytes. The detail of area in box is shown in (B).Visualized deposits range from 100 to 400 nm, with formation also of larger aggregates.

addition, Sieber et al. have categorically establishedthe link between the administration of Gd-containingcontrast media and the development of NSF-like lesions in a preclinical experimental setting.22

Recently, Edward et al. demonstrated in vitro that NSFfibroblasts synthesize excess levels of hyaluronan andcollagen.23 They also showed that gadodiamide, themost common Gd containing contrast agent linkedto NSF, stimulates fibroblast growth. Their initialexperimental design did not allow discriminationbetween the effect of intact gadodiamide and freeGd3+ released from the GdCA, but subsequent stud-ies have shown that GdCl3 also stimulates fibroblastgrowth (personal communication, M. Edward).

In the current understanding of the complex patho-physiology of NSF, at least two things are certain.First, kidney disease is an absolute requirement. Sofar, there has not been a single case reported withnormal renal function. Stages 4 and 5 chronic kid-ney disease, as well as acute kidney injury, are highrisk states. The estimated incidence of NSF in renalinsufficiency patients is 4.3 cases per 1000 patient-years, while some studies report its prevalence amongpatients with stage 5 chronic kidney disease up to18%.24,25 Second, exposure to Gd-containing con-trast agents is essential. Although there have been fewreported cases of NSF without Gd exposure, it cannotbe stated with certainty if those cases did not have

cryptic Gd exposure, or if all truly represent NSF.26,27

The variable stability among contrast agents is clearlyimportant.28

Our results demonstrate that Gd may persist intissue for long periods of time.4,17 The potential effectof such long term retention is still unknown. Thismay be the cause of variable onset of NSF and/orcontribute to the chronicity of the disease. There isclinical evidence that a small amount of Gd is retainedin bones after exposure to Gd-containing contrastagents in patients with normal renal function.29 Itremains to be seen if single and/or cumulative dosesof Gd-containing contrast agents can lead to Gd-induced NSF, albeit more slowly in patients withmoderate loss of or even normal renal function.

Gd is not the first and/or the only environmentalagent that is linked to fibrosing disorders. Toxicexposures including occupational and iatrogenic(aniline-denaturated rapeseed oil, L-tryptophan,polyvinyl chloride, bleomycin, carbidopa) have beenassociated with cutaneous fibrosing disorders.30 Inmost cases, even when the etiology is known orsuspected, the precise pathogenetic mechanismsleading to skin and tissue fibrosis are not alwaysunderstood. It has been postulated that in NSF,circulating fibrocytes are the effector cells and Gd actsas the trigger for the recruitment of these cells. Theinitial description of circulating fibrocytes identifies

1251

Thakral & Abraham

N

N

Fig. 10. SEM, backscattered electron image, semi-thin section. Showscytoplasmic Gd deposits in multi-nucleated macrophage (N-nucleus).Gd is associated with P and Ca in these deposits as revealed by EDSspectrum of analyzed (+) feature.

them as bone-marrow-derived mesenchymal cellsthat circulate in the blood. The characteristic stainingprofile and the absence of mitosis in NSF supportsthe suggestion that circulating fibrocytes are involvedin dermal fibrosis in NSF.31 However, it is not clearwhether local fibroblastic proliferation plays a role inthe pathogenesis.

Still, one question remains unanswered. How doesGd induce or promote fibrosis? As NSF is fairly rare,a number of permissive factors are likely required forGd exposure to initiate fibrosis. The permissive fac-tors may include vascular injury and an inflammatorystate, or metabolic derangements (increased serumphosphate, increased serum calcium, metabolic aci-dosis, high dose erythropoietin and iron overload).The combination of these events allows Gd3+ to dis-sociate from its chelate. Free Gd3+ can leak out ofdysfunctional blood vessels and precipitate in tissuesalong with Ca and P. Our findings of insoluble Gddeposits with co-associated elements (Ca, P and Fe)support this hypothesis. Such co-deposition may fur-ther promote oxidative stress and inflammation. In arecent review, Perazella suggested that macrophagesmay phagocytose Gd3+ and produce local profibroticcytokines, reactive oxygen species and signals thatattract circulating fibrocytes to the tissues.32 These

cells then transform into spindle cells and promotefibrosis.

We identified Gd in the cytoplasm of fibrocytesand macrophages. To our knowledge, this is the firstdemonstration that Gd deposits are actually presentin the cytoplasm of fibrocytes. The intracellular Gdin fibrocytes correlates with the in vitro and in vivo Gd-chelate metabolism and dissociation experimentalmodel by Franano et al.33 They demonstrated thatGd-chelates are endocytosed in intracellular vesiclesand have a high propensity to dissociate under theacidic environment provided by lysosomes. In anin vitro study, Mizgerd et al. also demonstrated Gdin phagolysosomes after exposure of rat alveolarmacrophages to GdCl3.11 It is likely that the observedGd in fibrocytes may represent dissociation anddeposition in lysosomes.

We have also described Gd-Ca-P co-depositionalong the basement membranes of vessels andsweat glands.15 The perivascular-dispersed Gddeposition in deparaffinized tissue specimens ofNSF as well as Gd deposition in the vicinityof sebaceous glands has been demonstrated.18,34

Both calciphylaxis and osseous metaplasia havebeen described in NSF.35,36 It is not apparent atthis time if calciphylaxis related to renal diseasebears any association, direct or indirect, with Gddeposition. Clinically, Marckmann et al. found thathigher serum concentrations of ionized calcium andphosphate increase the risk of gadodiamide-relatedNSF in renal failure patients.9 Since renal insufficientpatients frequently have deranged metabolic profile,the hypothesis that calciphylaxis and NSF sharea common pathogenetic link warrants furtherinvestigation.

A high index of suspicion is required clinically torecognize NSF. Common fibrosing disorders to beconsidered in the differential diagnosis include scle-roderma, scleromyxedema, and eosinophilic fasciitis.The distribution and the quality of skin involvement,and the association with particular concurrent dis-eases or specific laboratory parameters can be ofsubstantial help in refining the diagnosis.30 None ofthe histologic features in NSF is absolutely specific.Histologically, it can be difficult to distinguish scle-romyxedema from NSF as the spindle cells havesimilar CD34/procollagen dual positive profile.31

However, scleromyxedema usually displays pools ofmucin, lymphocytic inflammatory infiltrate and doesnot involve the subcutaneous tissue. The sparingof face and a lack of gammopathy in NSF alsodifferentiates it from scleromyxedema. Sclerodermais a systemic disorder, has positive autoantibodies(such as SCL-70 and ANA); and morphologically,

1252

Gd induced nephrogenic systemic fibrosis

is less cellular, and displays collagen homogeniza-tion and lymphoplasmacytic inflammatory infiltrate.Importantly, the absence of collagen clefts andCD34 positive fibrocytes distinguishes it from NSF.31

Eosinophilic fasciitis, an inflammatory fibrosing con-dition of the fascia, may manifest peripheral bloodand tissue eosinophilia, and polyclonal hypergamma-globulinemia. Both eosinophilic fasciitis and NSFmay display widening of the interlobular subcu-taneous septa, however, inflammation is absent inNSF.30

The cutaneous changes in NSF predominantlyaffect dermis and subcutaneous tissue. In most cases,a deep, full-thickness skin biopsy is essential fordiagnosis.31 This is of importance in Gd detection andquantitation as well. We have consistently observedthat Gd deposits are identified more in deep dermisand subcutaneous septa – thus superficial biopsiesmay not accurately identify and/or quantify theamount of Gd.

In summary, our study clearly demonstrates thattoxic Gd ions are released from soluble Gd chelatesand precipitate as insoluble deposits in tissues.Ultrastructural evidence confirms the deposition ofGd both extra- and intra-cellularly (macrophages,fibrocytes). Superficial biopsies may not detect Gdwhich is detectable in deeper biopsies. This is themost likely explanation for the inability to detectGd deposits in four of the biopsies. Nevertheless,to date we have detected Gd deposition in all NSFpatients whose skin biopsies we have analyzed. Thequantitative identification of insoluble Gd depositsprovides some important links in the understandingof development of NSF and also raises concern aboutpotential effects of long term retention of Gd intissues.

Since the spectrum of NSF is not limited to or exclu-sive to one field of medicine; awareness about its ini-tial presentation in multiple disciplines – nephrology,rheumatology, dermatology, radiology and pathol-ogy – would allow early and accurate recognition ofthis disease. In theory, if the etiology is the releaseof free Gd3+ from Gd-containing contrast agents,NSF should be nearly totally preventable in thefuture.

AcknowledgementsThe authors thank M. Barcza for the special preparation of tissues forTEM examination. Funded by the Department of Pathology, SUNYUpstate Medical University.

Presented at the US and Canadian Academy of Pathology Meeting,Denver, Colorado, USA, March 2008.

References

1. Cowper SE, Robin HS, Steinberg SM, Su LD, Gupta S, LeBoitPE. Scleromyxoedema-like cutaneous diseases in renal-dialysispatients. Lancet 2000; 356(9234): 1000.

2. Marckmann P, Skov L, Rossen K, Thomsen HS. Clinicalmanifestation of gadodiamide-related nephrogenic systemicfibrosis. Clin Nephrol 2008; 69(3): 161.

3. Ting WW, Stone MS, Madison KC, Kurtz K. Nephrogenicfibrosing dermopathy with systemic involvement. Arch Dermatol2003; 139(7): 903.

4. Thakral C, Alhariri J, Abraham JL. Long-term retention ofgadolinium in tissues from nephrogenic systemic fibrosis patientafter multiple gadolinium-enhanced MRI scans: case report andimplications. Contrast Media Mol Imaging 2007; 2(4): 199.

5. Cowper SE, Bucala R. Nephrogenic fibrosing dermopathy:suspect identified, motive unclear. Am J Dermatopathol 2003;25(4): 358.

6. Kalb RE, Helm TN, Sperry H, Thakral C, Abraham JL,Kanal E. Gadolinium-induced nephrogenic systemic fibrosis ina patient with an acute and transient kidney injury. Br J Dermatol2008; 158(3): 607.

7. Grobner T. Gadolinium – a specific trigger for the developmentof nephrogenic fibrosing dermopathy and nephrogenic systemicfibrosis? Nephrol Dial Transplant 2006; 21(4): 1104.

8. Moreno-Romero JA, Segura S, Mascaro JM Jr., et al. Nephro-genic systemic fibrosis: a case series suggesting gadolinium as apossible aetiological factor. Br J Dermatol 2007; 157(4): 783.

9. Marckmann P, Skov L, Rossen K, Heaf JG, Thomsen HS.Case-control study of gadodiamide-related nephrogenic systemicfibrosis. Nephrol Dial Transplant 2007; 22(11): 3174.

10. Evans CH. Biochemistry of the lanthanides. Plenum Press, 1990.11. Mizgerd JP, Molina RM, Stearns RC, Brain JD, Warner AE.

Gadolinium induces macrophage apoptosis. J Leukoc Biol 1996;59(2): 189.

12. Ersoy H, Rybicki FJ. Biochemical safety profiles of gadolinium-based extracellular contrast agents and nephrogenic systemicfibrosis. J Magn Reson Imaging 2007; 26(5): 1190.

13. Idee JM, Port M, Raynal I, Schaefer M, Le Greneur S, Corot C.Clinical and biological consequences of transmetallation inducedby contrast agents for magnetic resonance imaging: a review.Fundam Clin Pharmacol 2006; 20(6): 563.

14. Abraham JL, Thakral C. Tissue distribution and kinetics ofgadolinium and nephrogenic systemic fibrosis. Eur J Radiol2008; 66(2): 200.

15. Boyd AS, Zic JA, Abraham JL. Gadolinium deposition innephrogenic fibrosing dermopathy. J Am Acad Dermatol 2007;56(1): 27.

16. Thakral C, Abraham JL. Automated scanning electronmicroscopy and x-ray microanalysis for in situ quantificationof gadolinium deposits in skin. J Electron Microsc (Tokyo) 2007;56(5): 181.

17. Abraham JL, Thakral C, Skov L, Rossen K, Marckmann P.Dermal inorganic gadolinium concentrations: evidence for invivo transmetallation and long-term persistence in nephrogenicsystemic fibrosis. Br J Dermatol 2008; 158(2): 273.

18. Schroeder JA, Weingart C, Coras B, et al. Ultrastructuralevidence of dermal gadolinium deposits in a patient withnephrogenic systemic fibrosis and end-stage renal disease. ClinJ Am Soc Nephrol 2008; 3(4): 968.

19. Todd DJ, Kay J. Nephrogenic systemic fibrosis: an epidemic ofgadolinium toxicity. Curr Rheumatol Rep 2008; 10(3): 195.

1253

Thakral & Abraham

20. Thomsen HS, Marckmann P, Logager VB. Nephrogenicsystemic fibrosis (NSF): a late adverse reaction to some of thegadolinium based contrast agents. Cancer Imaging 2007; 7: 130.

21. Kay J. Gadolinium and nephrogenic systemic fibrosis: theevidence of things not seen. Cleve Clin J Med 2008; 75(2): 112.

22. Sieber MA, Pietsch H, Walter J, Haider W, Frenzel T,Weinmann HJ. A preclinical study to investigate the devel-opment of nephrogenic systemic fibrosis: a possible rolefor gadolinium-based contrast media. Invest Radiol 2008;43(1): 65.

23. Edward M, Quinn JA, Mukherjee S, et al. Gadodiamidecontrast agent ‘activates’ fibroblasts: a possible cause ofnephrogenic systemic fibrosis. J Pathol 2008; 214(5): 584.

24. Deo A, Fogel M, Cowper SE. Nephrogenic systemic fibrosis:a population study examining the relationship of diseasedevelopment to gadolinium exposure. Clin J Am Soc Nephrol2007; 2(2): 264.

25. Rydahl C, Thomsen HS, Marckmann P. High prevalence ofnephrogenic systemic fibrosis in chronic renal failure patientsexposed to gadodiamide, a gadolinium-containing magneticresonance contrast agent. Invest Radiol 2008; 43(2): 141.

26. Wahba IM, Simpson EL, White K. Gadolinium is not the onlytrigger for nephrogenic systemic fibrosis: insights from two casesand review of the recent literature. Am J Transplant 2007;7(10): 2425.

27. Anavekar NS, Chong AH, Norris R, Dowling J, GoodmanD. Nephrogenic systemic fibrosis in a gadolinium-naive renaltransplant recipient. Australas J Dermatol 2008; 49(1): 44.

28. Thomsen HS, Marckmann P. Extracellular Gd-CA: Differencesin prevalence of NSF. Eur J Radiol 2008; 66(2): 180.

29. White GW, Gibby WA, Tweedle MF. Comparison ofGd(DTPA-BMA) (Omniscan) versus Gd(HP-DO3A) (Pro-Hance) relative to gadolinium retention in human bone tissueby inductively coupled plasma mass spectroscopy. Invest Radiol2006; 41(3): 272.

30. Boin F, Hummers LK. Scleroderma-like fibrosing disorders.Rheum Dis Clin North Am 2008; 34(1): 199.

31. Cowper SE, Rabach M, Girardi M. Clinical and histologicalfindings in nephrogenic systemic fibrosis. Eur J Radiol 2008;66(2): 191.

32. Perazella MA. Tissue deposition of gadolinium and developmentof NSF: a convergence of factors. Semin Dial 2008; 21(2): 150.

33. Franano FN, Edwards WB, Welch MJ, Brechbiel MW, GansowOA, Duncan JR. Biodistribution and metabolism of targetedand nontargeted protein-chelate-gadolinium complexes: evi-dence for gadolinium dissociation in vitro and in vivo. MagnReson Imaging 1995; 13(2): 201.

34. Singh M, Davenport A, Clatworthy I, et al. A follow-up of fourcases of nephrogenic systemic fibrosis: is gadolinium the specifictrigger? Br J Dermatol 2008; 158(6): 1358.

35. Edsall LC, English JC 3rd, Teague MW, Patterson JW. Calci-phylaxis and metastatic calcification associated with nephrogenicfibrosing dermopathy. J Cutan Pathol 2004; 31(3): 247.

36. Lewis KG, Lester BW, Pan TD, Robinson-Bostom L. Nephro-genic fibrosing dermopathy and calciphylaxis with pseudoxan-thoma elasticum-like changes. J Cutan Pathol 2006; 33(10): 695.

1254