The development of NSF requires: (1) GBCA exposure in the setting of (2) reduced renal function. In its ionic form, Gd³⁺ is a known toxin due to its ability to block voltage-gated calcium channels [21]. The high number of unpaired electrons in Gd³⁺ has the advantage of modifying proton relaxation time, a parameter that made it an ideal magnetic resonance imaging (MRI) contrast agent, providing the toxicity could be avoided [21]. Binding the Gd³⁺ to an organic ligand (chelate) seemed to enable the safe use of Gd in this context [9, 22]. Chelated gadolinium was initially believed to be safe for use in all patients, even those with kidney disease, and provided an alternative to nephrotoxic iodine-based contrast agents [11]. The tight bond between Gd and its ligand, and the relatively short half-life of GBCAs in healthy individuals, made these agents safe for use in imaging studies [9]. As a result, the FDA approved GBCAs for use as a contrast agent in MRI studies. However, because GBCAs were considered to have such a high safety profile, these agents were also routinely employed in MRA, an off-label use that required dosing at 2–3 times the FDA approved dosage for routine MRI examinations [23].

Nine GBCAs are currently available (Fig. 12.5), all of which are approved for use in the United States (Table 12.1) [24]. Following the initial publications linking gadolinium exposure to NSF and subsequent observations that suggested breakdown of the gadolinium-chelate complex (dechelation) might be partly to blame, additional studies began to focus on prolonged GBCA stability [9, 10, 24–26]. The risk for developing NSF is highest among patients who had received one of the three high-risk linear GBCAs, with essentially no or extremely rare verified cases of NSF following the use of the more stable macrocyclic agents, implicating a key role for stability of GBCAs in the pathogenesis of NSF [24].

Currently available GBCAs enclose a single Gd atom within a proprietary organic ligand in either a linear or macrocyclic arrangement. Macrocyclic chelates enclose the Gd atom within a carbon ring. This arrangement confers stability, as multiple bonds must be broken simultaneously in order to release ionic gadolinium. Conversely, linear chelates wrap around the gadolinium atom, and can be sequentially unzipped from Gd by breaking single bonds. Both linear and macrocyclic chelates can also be either “ionic” or “non-ionic,” the latter generally regarded as being less stable [24].

Three GBCAs in particular have been implicated in the development of NSF: Gadodiamide (Omniscan™—GE Healthcare Inc., Princeton, NJ), gadopentetate dimeglumine (Magnevist^®—Bayer Healthcare, Wayne, NJ), and gadoversetamide (OptiMARK™—Mallinckrodt, St. Louis, MO) [10, 24]. Of these three agents, two (gadodiamide and gadoversetamide) utilize a linear non-ionic chelate, and one (gadopentetate dimeglumine) utilizes a linear ionic chelate [10].

Almost all GBCAs are cleared through renal dependent mechanisms and reduced GFR is a known prerequisite for the development of NSF [27]. In healthy subjects, the elimination half-life for most GBCAs is 1.3 h ± 0.25 h. In patients with reduced kidney function, the half-life of GBCAs was found to be prolonged up to 34.3 h ± 22.9 h in CKD stage 4, suggesting that reduced renal function results in an extended exposure to GBCA during which an adverse reaction might occur [28]. In patients with ESRD, GBCA will only be cleared when dialysis is performed, and it can take up to four hemodialysis sessions for removal of one GBCA dose [28].

Although a majority of patients have developed NSF in the setting of ESRD, approximately 20 % of patients experience disease onset in the setting of AKI or CKD stages 4 or 5 [24]. A model for calculating risk in patients (Fig. 12.6) with varying levels of CKD has been developed, based upon an estimated risk of 2.4 % for developing NSF in ESRD patients exposed to a high risk GBCA. Utilizing estimated prevalence for CKD in the US and distribution of NSF among patients with various levels of kidney function based on reports in the literature, the model predicts that patients with CKD 4 have a 1 in 2,492 risk of developing NSF after exposure to GBCA. The risk for patients with CKD stage 3 drops to between 1 in 100,000 and 1 in 700,000, suggesting that a patient with CKD stage 3 who is exposed to a single dose of a stable GBCA is not at an increased risk for developing NSF [18, 24]. For patients with ESRD, peritoneal dialysis-dependent patients were found to be at a 7.5 % increased risk of developing NSF as compared to hemodialysis-dependent patients [6]. Finally, only 2–18 % of patients with reduced renal function who are exposed to GBCAs ultimately develop NSF, suggesting that additional factors contribute to the pathogenesis of this disease [24, 26, 29–33].

In 2003, circulating fibrocytes (CF) were identified as the likely effector cells in NSF [34]. CFs were first described in 1994 as a subpopulation of circulating leukocytes (Fig. 12.7) with the ability to differentiate into fibroblast-like cells to facilitate wound repair at the local level [35]. However, in addition to normal wound repair, CFs have also been implicated in the propagation of several fibrosing conditions [36]. In 2003, the observation that the spindle-shaped cells seen in biopsies from patients with NSF stained for both CD34 and procollagen I (Fig. 12.8) led to the hypothesis that aberrant CF activity was to blame [34, 36]. Why this population of cells was emerging from the circulation and homing to the particular sites of fibrosis in NSF was unclear [34].

Shortly after the epidemiologic link between GBCAs and NSF was identified, a cell culture study by Edward et al. explored some of the key differences between NSF and control fibroblasts as well as the effects of gadolinium on fibroblast function [37]. The study showed that fibroblasts from patients with NSF demonstrated higher levels of hyaluronan and collagen production than those from controls. Additionally, control fibroblasts demonstrated increased production of these matrix materials after exposure to the serum of patients with NSF. Finally, when control fibroblasts were exposed to gadodiamide itself, they were shown to have increased growth, matrix production, and an enhanced ability to differentiate into myofibroblasts [37]. Additional studies confirmed the stimulatory role of gadolinium on the proliferation of fibroblasts [38] and the in vitro pro-fibrotic activity of the gadolinium-chelate complex [39].

To date, most of the animal studies of NSF were planned and executed by contrast manufacturers. Until recently, these depended on a sub-totally nephrectomized rat model [22, 40–43]. However, this model did not fully capture the clinical picture present in NSF. A subtotal nephrectomy leaves the animal with a diminished GFR (as compared to an absent GFR in ESRD patients) and is not accompanied by the same degree of hyperphosphatemia typically observed in ESRD patients [22]. Nevertheless, these models were collectively able to demonstrate that the GBCA gadodiamide was capable of producing histopathological changes in the 5/6 nephrectomized animals similar to those found in human patients with NSF (Fig. 12.9) [44]. Additionally, deposition of gadolinium in the skin of these animals was measured and found to be highest for animals dosed with gadodiamide and other linear agents, when compared to levels found in animals exposed to macrocyclic agents [22].

More recently, Fretellier et al. have been able to achieve a more accurate rat model replication of NSF pathology [22]. This group’s model utilized rats on an adenine-enriched diet for 8, 14, or 16 days to produce differing degrees of renal failure within the context of hyperphosphatemia. Rats with differing degrees of renal failure were then exposed to either gadodiamide or saline. Serial biopsies were obtained to allow for histologic examination and quantification of gadolinium within the tissue using inductively coupled plasma mass spectrometry. This trial demonstrated that rats with more severe kidney dysfunction had more severe profibrotic skin lesions and systemic toxicity. Additionally, rats that developed the most severe pathology were also found to have higher gadolinium levels by mass spectrometry than rats without cutaneous changes. Akin to humans, affected rats demonstrated overexpression of profibrotic cytokines including transforming growth factor beta-1 (TGF-β1) and tissue inhibitor of metalloproteinases-1 (TIMP-1) [22].

Another trial by this group examined the effects of exposing the rats with the greatest degree of kidney dysfunction to each of the available GBCAs. As opposed to the gadodiamide trial with differing degrees of renal compromise, these rats did not produce cutaneous changes, likely due to the high degree of toxicity and early mortality in this trial [22]. However, the authors were able to perform relaxometry studies that revealed that both linear GBCAs (gadodiamide and gadopentetate) underwent slow, gradual dechelation (defined by a release of the gadolinium ion from its organic chelate) in the adenine-induced renal failure rats. Conversely, all macrocyclic agents remained stable and did not produce a measurable degree of dechelation in vivo. To date, this model has most closely mimicked the pathologic process of NSF in vivo and provided additional insight into possible mechanisms at play behind the role of gadolinium in NSF [22].