Histopathological assessment of GIF involves sampling affected tissue, which often requires a full-thickness skin biopsy that contains deep dermal and subcutaneous tissues. Biopsies from other organs can also be analyzed for evidence of systemic fibrosis. Regardless of the organ affected, several histological features are hallmarks of GIF. Reticular bundles of dermal collagen are thickened and surrounded by clefts, and dermal mucin is increased with intact elastic fibers (Fig. 15.2). CD68⁺ factor XIIIa⁺ dendritic cells and spindle cells that have a CD34⁺ CD45RO⁺ type I procollagen⁺ phenotype infiltrate the dermis [3, 5]. Lesions of established GIF are less cellular than those of early disease and contain abundant collagen. Dystrophic calcification is a well-recognized histologic characteristic of late GIF lesions, and osseous metaplasia and calcified sclerotic bodies have been suggested to be histologic features highly specific for GIF [4, 68, 69]. However, in the absence of characteristic clinical features, it may be difficult to distinguish GIF from scleromyxedema based only upon histopathological analysis [70].

To better conceptualize the etiopathogenesis of GIF, it is important to understand the chemical and molecular properties that distinguish among the various GBCAs. Gd is a lanthanide series rare earth metal with seven unpaired electrons, which makes it highly paramagnetic and an ideal contrast agent for MRI. In solution, however, trivalent Gd⁺³ (“free gadolinium”) is highly toxic [71, 72]. There are multiple mechanisms by which Gd is toxic, including blockage of voltage-gated calcium channels [73], formation of inorganic Gd-phosphate precipitates [74], and induction of cytokines associated with tissue fibrosis (discussed later) [19].

To reduce the potential toxicity of Gd when used as a contrast agent, it is bound to an inorganic carrier molecule (“chelate”). It is this chelate that distinguishes among the various GBCAs approved for use in humans (Fig. 15.1). In 2006, five GBCAs were commercially available in the United States: formulated gadodiamide (gadodiamide with 5 % excess sodium caldiamide chelate, Omniscan^®), gadoversetamide (OptiMARK^®), gadopentetate dimeglumine (Magnevist^®), gadobenate dimeglumine (MultiHance^®), and gadoteridol (ProHance^®). Additional GBCAs were available outside of the United States at that time or have become commercially available since 2006: gadoxetate disodium (Eovist^®, Primovist^®), gadofosveset trisodium (Vasovist^®, ABLAVAR^®), gadobutrol (Gadovist^®), and gadoterate meglumine (Dotarem^®).

GBCAs can be categorized by whether they have a linear or macrocyclic structure and whether they have an ionic or nonionic charge (Fig. 15.1). These two properties determine the kinetic stability and thermodynamic stability of each GBCA and largely explain the propensity for Gd to disassociate from its chelate to become toxic free Gd⁺³. Those GBCAs that are linear and nonionic (formulated gadodiamide and gadoversetamide) are the least stable. In contrast, macrocyclic GBCAs (gadoteridol, gadobutrol, and gadoterate meglumine), whether ionic or nonionic, are the most stable GBCAs. Intermediate in stability are those GBCAs that are linear but ionic (gadopentetate dimeglumine, gadobenate dimeglumine, gadoxetic acid disodium, and gadofosveset trisodium). These properties are noteworthy when considering that the vast majority of reported GIF cases developed after patients had been exposed to the least stable GBCA: linear nonionic formulated gadodiamide.

All GBCAs are large molecules that remain extracellular. Most are highly water soluble and excreted by the kidneys [75]. Two GBCAs (gadobenate dimeglumine and gadoxetic acid disodium) are more lipophilic, such that a sizeable fraction is excreted via the hepatobiliary system [76]. In patients with intact renal function, the half-life (t_½) of formulated gadodiamide is 70 min: 95 % of a single dose is eliminated in the urine after 72 h [77]. However, the clearance of gadodiamide declines dramatically in patients with impaired renal function. The mean t_½ of formulated gadodiamide is 34.3 h in non-dialysis-dependent patients with stage 5 CKD and 52.7 h in patients receiving CAPD [78]. Only an average of 68.0 % of formulated gadodiamide is cleared after one HD treatment, and only an average of 72.3 % of the original amount is removed after four HD sessions [78]. In patients with GFR <20 ml/min, <70 % of a single intravenous gadopentetate dimeglumine dose was recovered in urine by 48 h after administration [79]. Further, Gd tissue deposition has been observed in biopsies of multiple organs from patients with GIF, even years after the last known GBCA exposure [18]. Notably, in patients with normal renal function, Gd has been shown to deposit and accumulate in the brain and bone, with higher bone Gd content after the administration of formulated gadodiamide than of an equivalent dose of gadoteridol [26, 27, 80].

After their administration, GBCAs persist for many more hours in patients with advanced renal disease than in individuals with normal renal function. In patients with compromised renal function who have received a GBCA, free Gd is more likely to dissociate from its chelate and be released as Gd⁺³, especially with the less stable (high-risk) agents. Altered phosphate levels may affect the release of Gd⁺³ from unstable linear chelates, but published data are inconsistent [11, 81, 82]. When a GBCA dissociates, not only does it release free Gd⁺³ but also it releases the empty chelate, which then is able to associate with other cations, such as Fe⁺² and Zn⁺². This process is called transmetallation [83].

An early study described the toxicity of gadopentetate dimeglumine in 151 patients with renal disease [84]. Mostly, only mild adverse effects were reported. However, this was a retrospective chart review with short-term follow-up (30 days for outpatient MRIs or until hospital discharge for inpatient MRIs). Only 71 of the 151 patients (47 %) had serum creatinine values >2.5 mg/dL, and only 15 patients (10 %) were receiving HD. GIF had not yet been described in the medical literature at the time of this study. Thus, it is not surprising that neither cutaneous nor systemic fibrosis was reported in this relatively small study with a brief duration of follow-up.

Two additional factors must be considered regarding the use of GBCAs during the early 2000s. First, since its introduction in 1994 [85], GBCA-enhanced MR angiography was being performed with increasing frequency. In this procedure, patients often were administered twice or three times the standard 0.1 mmol/Kg dose of GBCA used for MRI [85]. Second, according to the FDA Office of Surveillance and Epidemiology (OSE), two GBCAs dominated the United States market during the mid-2000s: gadopentetate dimeglumine had an approximately 50 % share of the GBCA market, and formulated gadodiamide followed with a market share of almost 40 % [56]. The vast majority of patients who were reported to have developed GIF had been exposed to formulated gadodiamide, the least stable of the various GBCAs; many fewer patients were reported to have developed GIF after exposure to the slightly more stable GBCA gadopentetate dimeglumine. Thus, its thermodynamic instability, rather than its market share, likely accounted for the perceived greater propensity of formulated gadodiamide to cause GIF.

Many molecular, cellular, and animal studies have provided the framework to help understand the mechanism by which high-risk GBCAs likely cause GIF (Fig. 15.3). Many lines of evidence support the hypothesis that, with prolonged tissue exposure such as that which occurs in patients with significant renal dysfunction, formulated gadodiamide and other high-risk GBCAs dissociate into free Gd⁺³ and chelate [86]. Investigators have used microanalytical scanning electron microscopy/energy dispersive X-ray spectroscopy (SEM/EDS), synchrotron X-ray fluorescence (SXRF) microscopy, extended X-ray absorption fine structure (EXAFS) spectroscopy, and inductively coupled plasma-mass spectrometry (ICP-MS) to characterize and to quantify Gd in insoluble deposits found in tissue obtained from GIF lesions [17, 18]. In these deposits, Gd⁺³ cations associate with complexes of inorganic phosphate, calcium, and sodium [87].

The amount of free Gd⁺³ released is the primary determinant of tissue toxicity [88]. Several factors favor the release of free Gd⁺³ from its chelate, including an acidic environment (as might be found in the phagolysosome of macrophages) [89] and transmetallation, whereby other cations, such as Fe⁺², Zn⁺², Ca⁺², and Cu⁺², may displace Gd⁺³ from its chelate [95]. The increased mobilization of iron that is observed after exposure of patients with dialysis-dependent renal disease to formulated gadodiamide provides evidence that transmetallation occurs in patients at risk for developing GIF [96].

In lesional tissue of GIF patients, mRNA for TGBβ1 (a pro-fibrotic cytokine) is increased [5]. In vitro, the release of free Gd⁺³ from GBCAs promotes the production of TGBβ1 and other pro-fibrotic cytokines, chemokines, and growth factors by monocyte-derived macrophages and peripheral blood monocytes [19, 90]. GBCAs also stimulate fibroblast proliferation and increased synthesis of extracellular matrix components, such as hyaluronic acid, fibronectin, and type I and III collagens [93, 94]. Formulated gadodiamide produces these pro-fibrotic effects at much lower concentrations than the more stable macrocyclic GBCAs [97, 98]. These effects are likely mediated by signaling through toll-like receptors (TLR) 4 and 7 [99] with activation of nuclear factor-kB (NF-κB) [90]. Free Gd⁺³, gadodiamide, and gadopentetate dimeglumine each activate the NLRP3 inflammasome in vitro to produce IL-1β; these molecules preferentially activate pro-fibrotic IL-4-polarized M2 macrophages [91].

Animal models of GIF demonstrate pro-fibrotic effects of various GBCAs, especially gadodiamide. Early studies, conducted before the first cases of GIF appeared, revealed toxic effects of GdCl₃ in mice, but these were due mostly to mineral deposition in vascular and reticuloendothelial tissues [72]. Rats given high doses of gadodiamide developed skin ulceration, an effect that was attributed to abnormal zinc metabolism [100]. After the association between GBCAs and GIF was recognized, investigators revisited animal models, administering GBCAs intravenously to rats that had or had not undergone partial nephrectomy [20, 21, 101, 102]. Although animal models do not recapitulate human GIF perfectly, several common themes emerged from these studies. First, formulated gadodiamide induced cutaneous changes in rats that were similar histologically to those of GIF, displaying increased cellularity and tissue fibrosis with increased collagen deposition and dermal thickening. Second, formulated gadodiamide was the only one of the eight GBCAs tested that induced these changes to any significant degree. Third, among rats that had received GBCAs, the highest tissue concentrations of Gd were detected in tissues of those that received formulated gadodiamide [20, 101, 102]. TGF-β1 was detected in lesional skin from rats treated with gadodiamide but not from those treated with gadoteridol [101], as it had been in the skin of patients with GIF [5].

Previously, we and others have applied Bradford Hill criteria [28] to contend that the relationship between high-risk GBCAs and GIF is not just an association but rather implies causation [103–106] Todd and Kay [23]. We readdress this line of reasoning in Table 15.2, updating the nine criteria with additional data presented in this review. With regard to strength of association, a meta-analysis of seven epidemiologic studies calculated an odds ratio of 27 for the association between GBCA exposure and the subsequent development of GIF [103], which is comparable to the odds ratio associating heavy cigarette smoking with lung cancer [28]. Consistency also has been fulfilled, since many investigators working at different institutions in different countries have each identified similarly strong associations between exposure to high-risk GBCAs and GIF [8, 10, 11, 14, 31]. Specificity is fulfilled by the absence of convincing cases of GIF occurring without known GBCA exposure; Gd has been detected in lesional skin even from patients with GIF who recalled no prior GBCA exposure [64].