Nephrogenic systemic fibrosis: Possible mechanisms and imaging management strategies

TWO GROUPS OF CONTRAST AGENTS have been developed and successfully introduced into clinical MRI practice worldwide. These are the gadolinium chelates (1-4) and the superparamagnetic particles (5), the former being the more widely used. Since the early development work on gadopentetate dimeglumine (Magnevist) (6) in the early 1980s in Berlin and London and its licensing in the United States and several European countries in the late 1980s, this prototypical agent, and its several subsequently developed competitors (Fig. 1; Table 1) have been adopted into clinical practice. Most are nonspecific, extracellular fluid space (ECF) markers with pharmacokinetics essentially identical both to each other and to those of the iodinated X-ray agents (7); some are designed either for hepatic uptake (Multihance/gadobenate; Primovist/gadoxetate) or for binding to albumin to achieve blood pool characteristics specifically for use in magnetic resonance angiography (MRA) (Vasovist/gadobutrol); these consequently have modified pharmacokinetics. Structures of the commercially available gadolinium chelates. The upper row of agents are cyclic complexes, the middle row are linear/acyclic complexes, and the lower row are more specialized agents with liver uptake (Multihance and Primovist) or albumin binding to achieve longer intravascular lifetime (Vasovist). Diagnostic agents should be safe and well tolerated and potentially less harmful than therapeutic agents; this criterion appeared from the beginning to be more than satisfied by the Gd chelates (8), which seemed to be associated with both a very low incidence of anaphylactoid reactions (4, 9) and with little nephrotoxic potential (10). Indeed, they have been recommended by several authors as X-ray agents in computed tomography (CT) and digital subtraction angiography (DSA) in patients deemed to be at risk of either anaphylactoid reaction or contrast agent–induced nephrotoxicity (CIN) if iodinated X-ray agents were used (10, 11). “Double-” and “triple-dosing” in both MRI and CT became common and, even though done “out of license,” especially in MRA, caused little concern among radiologists even when carried out in sicker patients, including those with impaired renal function. The high tolerance and safety profile of the Gd chelates is, at least partly, a function of the low doses recommended and generally used but it is now understood that their use in higher doses, and consequently higher osmotic loads, may give them a nephrotoxic potential (10). In addition to the phenomenon of CIN with the Gd chelates, we have seen the emergence of a new condition, nephrogenic systemic fibrosis (NSF), which is associated with, if as yet not definitively proven to be caused by, Gd chelate administration to patients with end-stage renal failure (ESRF) and perhaps other serious medical conditions (12-26). NSF, first recognized in 1997 by Cowper et al (12), but not initially linked to Gd chelates, is characterized by thickening, induration, and tightening of the skin. The distal extremities are most commonly involved but the trunk may also be involved. Some patients manifest only nonprogressive local disease, which may make shunt access problematic, but others develop a progressive form of the disease that may involve lungs, heart, liver, kidneys, skeletal muscle, and diaphragm. Fibrotic changes lead to joint contractures and limit mobility, which is contributory to death in many patients by way of hypostatic pneumonia, especially when there is also diaphragm involvement (12-14, 16, 24). There is no apparent gender predisposition. More than 300 cases have now been reported and, given the fact that reporting systems are voluntary and given the difficulties of follow-up in many patients undergoing imaging, this is likely to be an underestimate. A number of therapeutic approaches have been suggested but none have been proven consistently effective (26). Grobner (15) was the first to propose a connection with Gd chelate administration in 2006. Onset is within a few days to a few months of administration. Gadolinium has now been found in the skin in biopsies from some patients (22, 23). This finding has strengthened the hypothesis that “free”1 gadolinium liberated from the complex by a process of transmetallation, a process in which gadolinium is replaced in the chelate by other ions, plays a direct role in the pathogenesis of the disease. Furthermore, Kimura et al (27) have demonstrated that Gd chelate administration is associated with an increase of zinc and copper excretion in urine; and Idée et al (28) have reported transient increases in serum iron levels after injection of the Gd chelate, gadodiamide. These observations are further indirect support for the idea of “transmetallation,” the displacement of gadolinium from the chelate by other ions (21). “Free” gadolinium is toxic (1, 2, 6, 8, 29-31)—the very reason that necessitated its incorporation into Gd-chelate complexes. Inside cells it is capable of engendering DNA hydrolysis. Liberated Gd3+ largely forms Gd(OH)3 and Gd(PO4)3, which are of low solubility and may be phagocytosed (23) and cause activation of foreign body and fibrous reactions (25). Cowper and Bucala (32) have emphasized a role for circulating fibrocytes in the pathophysiology of NSF. These fibrocytes migrate to sites of inflammation where they secrete growth factors and cytokines, and contribute to matrix production in connective tissue (33). It has been noted that transforming growth factor β, a potent stimulus for production of type I collagen by some cell types and a mediator of interstitial fibrosis, can induce fibrocytes to differentiate into myofibroblasts, which do indeed represent a small proportion of the spindle cells found in NSF. Others have recorded increased levels of transforming growth factor β in skin and muscle in some NSF patients (34). However, although the idea that “free” gadolinium released as a consequence of relative chelate instability is the initiator, and notwithstanding the elaboration of ideas such as the above, the precise pathway from the putative toxic insult of free gadolinium to the syndrome of NSF remains unclear. To complicate matters somewhat, so far the great majority of NSF cases have occurred after administration of one specific agent, gadodiamide (Omniscan). Recent data from Leiner (35) suggests that some 327 cases have been associated with gadodiamide (in more than 40 million administrations), 43 cases with gadopentetate (in more than 90 million administrations), nine cases with gadoversetamide (in more than 6 million administrations), and one case with gadoteridol (in more than 13 million administrations). No unconfounded cases (confounded cases are those in which more than one agent had been use in the patient in a short time interval) were reported with gadoterate (Dotarem), gadobenate (Multihance), gadofosveset (Vasovist), gadoxetate (Primovist), or gadobutrol (Gadovist). However, some of these agents have been used relatively little and there have been several examples of confusion over use of different agents in the same patient during a short time interval; but any hypothesis concerning dechelation of gadolinium must embrace an explanation for an apparent hierarchy of chelate stabilities. It is true that patients have received a Gd chelate on one occasion with no ill effect, only to develop NSF apparently in association with another administration at a later date. Not all patients with ESRF develop NSF when exposed to Gd chelates. Some patients with NSF have no proven history of Gd chelate exposure (though this may be a documentation issue). In some patients the temporal correlation between chelate administration and development of NSF is much longer than in others, raising doubts about a true connection. Gadolinium has not been found to be present in all skin biopsies in NSF cases and it may be a “passenger” rather than causative agent. Other factors than gadolinium have been associated with the development of NSF, such as: erythropoietin; inflammation on a variety of bases; inhibitors of angiotensin converting enzyme (36); induced antibodies against phospholipids (37); dialyzate fluid (or a contaminant) (24); erythropoietin (24); and hepatorenal syndrome or perioperative liver transplant (24). It is not known whether these play a central role, or are cofactors or adjuvants. A large proportion of cases of NSF have been in contrast-enhanced (CE) MRA (CE-MRA), in which high doses are commonly used, lending support to the idea that this may be a dose-related phenomenon. Indeed, it is interesting to note that the onset of first reports of NSF in 1997 (13) coincided closely with the success and burgeoning use of CE-MRA. Until more is known, a sensible conservative working hypothesis is that Gd3+, liberated from Gd chelates, is implicated in the development of NSF, that it may not be the only possible causative agent, that there may be other factors or cofactors, and that it should be considered in terms of the preinflammatory of ESRF milieu in which it is commonly observed. Our challenge is to determine the true pathophysiology of this disease, including its relationship to gadolinium, and to develop a plan to minimize risk associated with CE-MRI. Below, some aspects of chemical structure and chelate stability and possible mechanisms of NSF development are explored and possible imaging management strategies are considered. To understand Gd chelates and their stability it is first necessary to examine some chemistry. Gd3+, with its seven unpaired electrons in the 4f orbital yields a large paramagnetic moment. The symmetric s-state of Gd(III) is a good environment for electron spins, leading to a much slower electronic relaxation rate. Dysprosium, another of the lanthanides or “rare earth” elements, has a bigger magnetic moment, due to orbital contributions to total electron angular momentum, than Gd(III), but the asymmetry of these electronic states leads to very rapid electron spin relaxation and much reduced effectiveness. Gadolinium, therefore, suggested itself early on as a contrast agent for MRI. However, “free” Gd3+ is known to be toxic, especially if intracellular, with 50% of the lethal dose (LD50) on the order of 0.5 mmol/kg (8, 38). This is the same order of magnitude as the dose currently used in clinical practice and so chelation was used to bind the gadolinium. Certain polycarboxylic acids, such as diethylene triamine pentaacetic acid (DTPA) (Fig. 1), the prototypical “ligand” in this field, coordinate metal ions strongly (2, 6). The coordination is between the amino nitrogens and one of the oxygen atoms of each of the carboxyl groups; the second oxygen in each carboxyl group points away from metal and participates in hydrogen bonding with water, thereby contributing to solubility. The extra electrostatic component provided by the charged carboxyl (COO-) moiety assists. Initially this charge is delocalized on the COO- group but comes to reside on the coordination bond of oxygen. Gd3+ has nine sites at which such coordination may take place and, when the DTPA ligand is used, eight are occupied by the three N atoms and five O atoms in the carboxyl groups. Such coordinate bonds form and reform continuously. The ninth site is available for fast exchange of water molecules between the “inner sphere” of and Such a Gd or Gd is as a coordination with points of such as are to as Such are capable of as well as in that they the central in a (1, the Gd3+ in this in reduced ligand all the Gd coordination sites the of and even of on a as by the though this is the fact that more which increases the correlation time and the development of this agent, several others have 1; 1) the commercially available gadolinium chelates are in The of these is into cyclic and linear/acyclic types (1, The cyclic chelates in represent the most of the complexes. The Gd3+ is to N atoms in a symmetric and to three or O atoms provided by carboxyl In the case of acid all involved in coordination are in carboxyl and the is a strongly in and in the is the same for the fact that one of the is in a less effective The is the more The or chelates in are all less than their cyclic (Fig. first and there are between in the complex the Gd3+ is to three and five in carboxyl in the on the other of the carboxyl are replaced by now the coordination is with three but with only three in carboxyl groups; in the now less effective coordination than the carboxyl they Such may be was developed as a agent such are not the be on to be less than the in is less than any cyclic agent. The are In to another has been the is a little more than the former by of its carboxyl groups. both are the is that the more agent have a to bind to a variety of molecules and perhaps to of stability are the stability and the The stability is the when the agent is in Table some for several that there are of several of magnitude between some of the the is as as (Omniscan). This given the very high of even the lower which a very high stability and very low at least in the of free an of ligand the is into the to bind any Gd3+ should it be liberated from the The widely of ligand in of the in a of relative In the is very different from that in there is for gadolinium in ligand binding with a variety of other ions such as and in such is as It is to an to stability that any possible in A further to the of NSF is that is lower instability increases and transmetallation, or In not only is there a time during which such may take place but there may also be a lower than The of the stability of some of the Gd chelates used in MRI and in been much explored and many of the have been Liberated Gd3+ is found in and The stability of the cyclic over the was and within the complex group were noted many In more gadolinium has been found to in and has been in tissue of patients with NSF (22, 23). et al (23) have gadolinium in of 13 tissue from seven patients with NSF were exposed to they were to gadolinium in tissue to months after No gadolinium was in a tissue from any patient NSF. Other found in the of NSF patients large of and zinc and this in itself may is their that in all cases the gadolinium was gadolinium is It is possible that the free gadolinium is liberated and cells as the or or it is possible that the higher of gadodiamide the complex to be into the cell where it is likely to be by and a low at which it be to be very are more than linear/acyclic complexes. have been associated with all the cases of NSF they are less (though this fact is only in the of renal the complex there are in which has been associated with or more of all cases of NSF, is less in than of its available coordination sites are with less binding by group In of a at low in ESRF and in with other ions, the Gd3+ is It is not whether the gadolinium is released of or cells but it is possible that the of gadodiamide may also play a role in into doses of the agents increase the The that more such as and may free Gd3+ to more where the metal and local as well as that circulating fibrocytes to the in circulating fibrocytes induce a process of The of other cofactors are unclear. hepatic in it may be that if hepatic function is an if a not very effective for excretion of the agents in ESRF is is to the liver, if its function is may be capable of as a remains about a number of the less commonly used agents such as (Multihance), gadoxetate (Primovist), and gadofosveset (Vasovist). The first are designed for increased liver uptake and and the is designed to bind to it an effective blood pool agent for use in binding to a large a complex with it is also more effective and may be used at lower thereby are linear/acyclic and from the ideas not be considered first in patients, but it may be that for the liver agents in patients with liver function a is as whether the binding to and from albumin of and increased liver excretion in patients with hepatic function make it is The fact that if unconfounded cases of NSF have been found with these agents may to their relatively little use rather than to the role of “free” Gd3+ in the pathogenesis of NSF, the correlation with Gd chelate administration in patients with or hepatic or other possible associated to be and must be a concern in management of the imaging of patients to be at it is also to that it is not to the It should be that over million patients have been exposed to gadolinium chelates the 1980s, with only a any at Indeed, some million have been exposed to the agent not but associated with NSF, and some million to the agent most associated with NSF. cases have all been in a small group of patients with the risk factor (ESRF) and with other possible risk factors or cofactors erythropoietin administration in higher dose or hepatic as There are that even when the Gd chelates are to such patients, the incidence of NSF is apparently This is a challenge much of MRI clinical including the and the European for for of the European the and the of the European of the for in the European for in and and the for have on NSF and on to with the and have for case documentation and in the management has been to be in the use of and in patients with renal disease and in liver and in and to one of they should only be used after very The that all the gadolinium should be used with but they and as involved. However, they have that the is a and that should be with all agents no among the agents is now This a from the should be in any agent, there are specific chelates that may present a risk in patients with ESRF both on the of chemical and clinical This to be the from which the has now The renal function of the patient is the is there is no cause for the is renal no Gd chelate should be necessary for a MRI in the In such a case a cyclic agent should be or of is, of but is the same as a risk for CIN with iodinated X-ray contrast agents the is known and is but there is a reported cases of NSF have been in patients with the most renal in all patients with any serious of impaired renal function and, if a contrast agent is to be a cyclic one be The renal function is not The is, are we to of an renal function in all but patients a Gd To do so be to difficulties into the of and is, it may be not a patient in the to a of a history of renal renal and of other nephrotoxic or it if not that a patient with ESRF as an and such a this is essentially the by the in the United in its Some may to take a more but may administration of more as a all cases in which renal function is not known or is known to be higher than recommended doses should be In and to one of renal function is an all gadolinium should be used with and cyclic are to be The of other imaging should of be considered. In many CT be an effective However, contrast is to achieve CT imaging, and CT is to be any more that MRI. In patients with function there are concerning CIN to be in In the most likely imaging to is not such a In a the that it may be to to NSF in patients with renal have received gadolinium contrast It is true that of Gd chelates may be by both and three of more than of a chelate days of only so only the former is a patient is on for ESRF this management may be though it is not proven to be effective in development of NSF. the patient is the of involve of its which may well be than the risk of NSF. the patient is at risk for NSF. the of a MRI or other imaging with or their contrast it an MRI is considered to be a cyclic agent should be used. the is but the risk to be low but the should be and a cyclic agent should be used if the is with the case is an there should be no in the if A cyclic complex should be used. In cases the provided history should be and by a concerning history of renal disease, renal renal and of potentially nephrotoxic no of renal disease the may and any agent may be used. any of this concern about renal the should be an is In and to one of renal function is an and all gadolinium should be used with patients at risk Gd chelates may be considered for this is of proven in Gd chelates from the it is not proven to be effective as against NSF. should not be in any patient not on it as it is associated with a of its It is to a of and in both that gadolinium chelates are among the agents and that even in patients the of NSF to be less than NSF is a very serious and sensible must be by gadolinium chelates. The pathophysiology of NSF has not been but the strongly suggests a role for gadolinium liberated from the chelate in a of in with other ions, time in the body in and a low if there is hepatic another for excretion of the chelate is The of such a of Gd3+ is well understood and the relative of the chelates are from an of their chelates are more than linear/acyclic chelates, of stability is used, and some linear/acyclic chelates are less than such are out by in clinical practice. is and the of any of the Gd chelates in patients with lower than as recommended by the most sensible one to the must be done with contrast a cyclic complex should be used. The by the patient renal function is not known this be by a history from each and patient for a renal function is not cases of NSF of be and reported to the and to the NSF (26). I of London and of the of for very