Nephrogenic systemic fibrosis: review of 408 biopsy-confirmed cases.

Nephrogenic systemic fibrosis (NSF) has now been virtually eliminated by the discovery of its association with gadolinium-based contrast agents (GBCAs) and the consequent reduced use of GBCA-enhanced magnetic resonance imaging (MRI) in severe renal failure patients. This review of 408 biopsy-confirmed cases shows how to minimize NSF risk when performing GBCA-enhanced MRI or magnetic resonance angiography. The absence of any NSF cases in patients less than 8 years old or greater than 87 years old suggests that infants and elderly patients are already protected. Limiting GBCA dose to a maximum of 0.1 mMol/kg, dialyzing dialysis patients quickly following GBCA administration, delaying administration of GBCA in acute renal failure until after renal function returns or dialysis is initiated, and avoiding nonionic linear GBCA in renal failure patients, especially when there are pro-inflammatory conditions, appear to have reduced NSF risk to the point where safe GBCA-enhanced MRI is possible in most patients.

Introduction

In nephrogenic systemic fibrosis (NSF), an over-reaction to some biological trigger in patients with acute or severe chronic renal insufficiency leads to extensive fibrosis involving skin, subcutaneous tissues and, sometimes, internal organs. Its association with gadolinium-based contrast agents (GBCAs), especially at high doses,[1] has led to the hypothesis that GBCAs can be part of the trigger. Dialysis,[2] edema,[3] hyperphosphatemia,[4] epoetin[5] and pro-inflammatory conditions[6] are additional important risk factors. Black Box FDA warnings, European EMEA and other regulatory actions have led to renal function screening for MRI patients and using less or no GBCA when the glomerular filtration rate (GFR) is estimated to be less than 30 mL/min. Dialysis patients are scheduled to have gadolinium-enhanced MRI scans just before a routine dialysis appointment to help remove the GBCA as quickly as possible. Since the establishment of these warnings and regulatory actions, NSF has nearly completely disappeared. Within the last 3 years, almost no new cases with symptom onset have been reported.

These new GBCA policies[78] have deprived many renal-failure patients of GBCA-enhanced imaging, often substituting alternative imaging studies that may have more serious risks such as from ionizing radiation, iodinated contrast or the risk of failing to make a correct diagnosis. To better understand NSF risk so that risk-benefit trade-offs in using GBCAs can be analyzed appropriately, there has been a review of biopsy-confirmed cases.[9] This article updates Prince's review to include 408 biopsy-confirmed cases of NSF individually reported in the literature.[1610-106]

Clinical Features of NSF

NSF presents with dermal pain, thickening and hardening, especially in the lower extremities. The extent ranges from small patches of skin to extensive areas of the body but universally sparing the face. The thickened, peau d’orange skin may acquire a “cobblestone” texture, dimpling or a woody aspect, which are often accompanied by mild-to-moderate edema. Sharp pains, burning or itching may occur in affected areas. These lesions often encompass joints. Joint contractures (n= 115) or limited range of motion (n= 23) were present in 138 of the 227 patients for whom these data were available. Five additional patients reportedly had “stiffness” without contractures. This suggests that over one third of the patients had a mild form of NSF without contractures or limited range of motion. Scleral plaque or injection was described in 20 cases; otherwise no facial involvement was reported. Some of the imaging features included soft tissue activity on bone scanning (n= 8), skin thickening on mammography (n= 4), dermal calcification (n= 9) or inflammatory changes on CT scans (n= 14). Internal organ involvement was described in 16 cases.

Demographic Features of NSF Patients

Among the 386 patients whose gender was reported , males (n= 202) and females (n= 184) appeared to be equally affected. Mean age was 51 years (8 to 87 years — see Figure 1). There were no cases reported in neonates or toddlers even though neonates with low GFR (due to immature kidneys) commonly receive high doses of GBCA, especially for imaging congenital heart disease. This indicates that very young patients are somehow protected. One possible reason is that the immune system at this age is not sufficiently developed to overreact to the NSF trigger.

Figure 1

Effect of age on NSF incidence: For 391 patients reported in the literature whose age is available, the number of NSF patients in each decade of age is shown in solid black bars. Open bars show age distribution for patients undergoing high-dose MR examinations with GFR <30 mL/min at Columbia College of Physicians and Surgeons over 10 years. Note that peak NSF incidence occurs in the 51-60 years age group. No NSF cases are reported in patients less than 8 years old or greater than 87 years old. Open bar with black dots shows the number of infants less than 1 year old receiving high-dose GBCA at Columbia. The absence of NSF cases in infants in spite of their immature kidneys and a large number receiving high dose indicates infants are not at risk of developing NSF. A reduced incidence for age >70 years even though there are many more at-risk patients (GFR <30 mL/min) receiving high doses of GBCA suggests that elderly patients are also protected.

NSF incidence peaks in the decade corresponding to age 51 to 60 years and decreases at older ages even though elderly patients have more severe renal disease and more per capita GBCA-enhanced MRIs. This may reflect a less active immune system and decreased ability to mount an inflammatory response in the elderly. Indeed the oldest reported patient with NSF was only 87 years old in spite of there being many >90-year-olds with low GFR receiving high doses of GBCA. Thus infants and patients greater than 90 years of age appear to have zero risk of developing NSF.

In 227 patients, race was reported: Caucasian (n= 126), Black (n= 54), Hispanic (n= 7), Chinese (n= 8), Malay (n= 1), unspecified Asian (n= 1), Indian (n= 7), Vietnamese (n= 2), Japanese (n= 19) and Brazilian (n= 2). Lower incidence of NSF in China in spite of a large population may relate to their avoiding double- and triple-dose GBCA, which had become common in western countries during the NSF peak.[107]

NSF and GBCA

In 77 of the 100 papers, history of GBCA administration was investigated. Data and clarifications were also provided by the authors of 49 papers via e-mail; accordingly, data on the relationship between GBCA and NSF was available for 313 patients for this review update. A total of 280 (89%) patients were noted to have had GBCA injections prior to NSF symptom onset. The existence of NSF in patients without prior gadolinium exposure is one reason why the relationship of GBCA with NSF remains just an association and GBCA is not considered to be the proven cause of NSF.

The time interval between GBCA injection and onset of NSF was available in 225 patients, ranging from the onset being on the same day to 3 years. When intervals between GBCA exposure and symptom onset that were greater than 1 year were excluded (i.e., no relationship for 14 cases with an interval >1 year), the mean interval between GBCA exposure and symptom onset was 62 days.

Data on incidence of NSF in patients with renal failure who were exposed to GBCA is shown in Table 1. The wide range of incidence, from 0% to 18%, suggests that many variables affect NSF risk.

Table 1

Incidence of NSF in renal-failure patients exposed to GBCA

Renal Function

Renal dysfunction was reported for all NSF patients. Highest NSF risk was described for GFR <15 mL/min, i.e., chronic kidney disease (CKD) stage 5.[108] In 385 patients, it was possible to determine whether or not the patient was on dialysis around the time of developing NSF, including hemodialysis (n= 205), peritoneal dialysis (n= 37), both (n= 13), continuous veno- venous hemofiltration (CVVH) (n= 4) or unspecified (n= 67). Thus 85% (326/385) of NSF patients were on dialysis, indicating that dialysis is a major risk factor. For 59 NSF patients presumably not on dialysis, GFR was reported to range from 0 (anuric) to 54 mL/min with a mean value of 16 mL/min. For 3 patients in whom estimated GFR was greater than 30 mL/min, there was acute renal failure, indicating that the true GFR was actually lower. One patient with an estimated GFR greater than 30 mL/min had got done the GFR measurement 1 month prior to GBCA injection. Another chronic renal failure patient with an estimated GFR of 54 mL/min (stage 3 CKD) at the time of GBCA injection developed NSF more than 2 years after that injection, suggesting that the NSF trigger in that patient may have been a different kind of, as yet undiscovered, exposure.

Acute vs. chronic renal failure

In 224 patients for whom the description in the literature allowed discrimination between acute and chronic renal failure, 61 (27%) had acute renal failure or acute deterioration of chronic renal failure. Acute renal failure has also been reported as an NSF risk factor with odds ratio of 13.4.[49] However, in these 61 patients with acutely deteriorating renal function at the time of GBCA administration, only 19 (31%) patients out of those for whom data on mobility was available had contractions. Follow-up data, available for 37 of the NSF patients, related to acute renal failure shows that 2 (5%) had a complete resolution of NSF and 17 (46%) showed improvement. This likely reflects the tendency of NSF to improve with restoration of renal function that may occur when acute renal failure resolves. Thus although acute renal failure increases NSF risk, the ensuing NSF clinical course may not be as debilitating.

Timely Effective Dialysis

Although 326 patients were on dialysis at the time of developing NSF, a number of factors may have reduced the effectiveness of dialysis for those patients who developed NSF. In a disproportionately large number of NSF cases, peritoneal dialysis (n= 37) or CVVH (n= 4) was utilized, both of which are known to be less effective at rapidly removing GBCA compared to hemodialysis.[109110] In 54 patients, it was specifically reported that GBCA-enhanced abdominal magnetic resonance angiography (MRA) was performed, which can assess for renal artery stenosis as the cause of the renal failure prior to initiating dialysis. When initiating dialysis, the first sessions are of lower intensity, to help the patient acclimate to the stress of hemodialysis. In addition, for patients awaiting initiation of dialysis, the interval between GBCA exposure for MRA and first dialysis may be longer than the usual 2 days between sessions for chronic dialysis patients. In 7 patients, GBCA-enhanced MRA was used to evaluate dialysis fistulae. Effective dialysis following their GBCA-enhanced MRA was less likely if the dialysis fistula was malfunctioning.

For 66 patients in whom the interval between GBCA administration and dialysis could be determined, dialysis was performed the same day in 16 patients, a day later in 13 patients, 2 days later in 9 patients and ≥3 days later in 28 patients. Thus most NSF patients on dialysis may have had a delay between administration of GBCA and receiving of dialysis; may have had peritoneal dialysis (PD) or CVVH, both of which are less effective at removing GBCA; or may have had lower-quality dialysis due to reduced fistula function or just beginning dialysis. For 7 of the patients dialyzed within 2 days of GBCA exposure, the authors specifically indicated that it was a low-intensity dialysis, which could not be expected to remove as much gadolinium as standard hemodialysis.[38111] These data support using prompt hemodialysis to reduce NSF risk in dialysis.

Kidney Transplantation

In 36 (46%) of the 79 NSF patients with history of renal allograft, the renal transplant was failing. In most of these 36 cases, renal transplant artery MRA with high-dose GBCA was being performed to determine if there was a correctable renal transplant artery stenosis.

Liver Disease

Liver disease was described in only 26 patients, even though FDA warnings have singled out liver-disease patients for extra caution. It seems, however, that regulatory warning relating to liver disease may have been overly cautious, since a recent paper by Mazhar shows that liver disease confers no additional NSF risk beyond the risk related to renal dysfunction.[112]

GBCA Dose

Several articles had listed wrong GBCA doses; these were corrected by corresponding with authors. In 280 cases with data on GBCA dose, 35 (12.5%) patients appeared to have received a standard dose of GBCA (0.1 mMol/kg) in the MRI exam most immediately preceding development of NSF, and 245 (87.5%) patients received high-dose GBCA. The mean total dose was estimated to be 48 mL. Data from individual case series showing the increased NSF risk with higher GBCA doses is listed in Table 2.

Table 2

Effect of GBCA dose on risk of NSF in case-control studies

NSF risk is reduced by at least an order of magnitude by limiting the dose to 0.1 mMol/kg. However, 0.1 mMol/kg becomes “high dose” when a second injection is necessary due to nondiagnostic results with a poorly timed or extravasated first injection.[1] Thus dose reduction also requires having the best technologists and nurses performing the examinations using the best possible equipment for at-risk patients to minimize the need for repeat injections. MRA exams in the past used high dose to make up for equipment limitations; with state-of-the-art equipment, MRA is diagnostic with standard, 0.1 mMol/kg, dose or sometimes less.[113] FDA approval of gadofosveset trisodium, a high-relaxivity blood pool contrast agent, also allows diagnostic MRA with very low doses even with older equipment.

Type of GBCA

Only 63 papers indicated type of GBCA, which included 263 patients who were most likely exposed. Many articles indicated a failure to maintain accurate patient records of GBCA administration, hence the GBCA type had to be inferred from purchase-related or formulary records. There were also errors in reporting GBCA type, with an erratum in at least one paper[114] and several corrections on GBCA type made after correspondence with author. Another issue was the uncertainty about patient exposures at outside institutions.

Where information on type of GBCA was provided, gadodiamide (n= 210), gadopentetate dimeglumine (n= 30), gadoversetamide (n= 5), gadobutrol (n= 3) and multiple agents, including gadobenate and gadoteridol (n= 15), were described; although the accuracy and completeness of this data were questionable.[1] The large proportion of cases with nonionic agents (gadodiamide and gadoversetamide) [Table 3], which have lower in vitro stability compared to the ionic (gadopentetate dimeglumine, gadoxetate, gadofosveset and gadobenate) and macrocyclic (gadoteridol, gadobutrol) agents, has led to the hypothesis that lower chelate stability may contribute to greater NSF risk. However, this does not necessarily reflect overall risk for individual patients because nonionic contrast agents have fewer serious allergic-type adverse events[115-117] and fewer deaths (see FDA AERS database).

Table 3

Comparison of NSF risk with gadodiamide (nonionic) and that with gadopentetate dimeglumine (Gd:DTPA, ionic)

Type of MRI exam

MRI exam type for 195 patients was reported to be MRA in 100 (51%) patients. This reflects the common use of high doses for MRA exams [Table 4]. Abdominal MRI (n= 34; 17%), including liver and magnetic resonance cholangiopancreatography (MRCP), was the second most common MR exam type, which also likely reflects high doses for liver MRI prior to 2008.

Table 4

Types of GBCA-enhanced examinations preceding NSF

Pro-inflammatory Events

Patients with pre-existing pro-inflammatory conditions are hypothesized to be at increased risk because inflammation stimulates circulating fibrocytes, which are then available for diversion to sites where NSF is triggered. Data on ′pro-inflammatory events’ at the time of GBCA injection, including recent major surgery (n= 81), acute thrombosis (n= 58), infection (n= 44), myocardial infarction (n= 2), antiphospholipid syndrome (n= 12) and active systemic lupus erythematosus (SLE) (n= 15), was available in 147 cases. In addition, Golding and Provenzale noted that infection increased the risk of NSF in dialysis patients by 25-fold.[118]

Erythropoetin

Erythropoetin (epoetin) is commonly given to renal-failure patients to stimulate the bone marrow to produce more red corpuscles. As epoetin is known to be pro-inflammatory and to stimulate other cell lines in the bone marrow, the possibility that it represents a factor contributing to risk has been suggested by several authors.[5114] In 82 patients for whom medication details were provided, 66 (80%) were on epoetin, including many on high doses.

Acidosis

Extra-positively charged protons at lower pH (acidosis) may compete with gadolinium-binding sites on the chelator, weakening the strength of the gadolinium-chelate bond.[55119] Of the 43 patients with data on either blood gas or bicarbonate results near the time of GBCA administration, 23 (53%) were noted to be acidotic.

Hyperphosphatemia and Zinc

Elevated serum phosphorus is common in renal failure. For 86 patients whose serum phosphorus data was available, it was found that serum phosphorus level was elevated to a mean of 6.8 mg/dL (2.3 mmol/L) (normal, 2.5 to 4.5 mg/dL, 0.8 to 1.5 mmol/L). Hyperphosphatemia increases the risk of phosphate binding and precipitation of Gd +++when it is transiently released from the chelator, thereby preventing re-association. Although no authors had data on zinc, this cation is known to compete with gadolinium for chelator-binding sites, an action which promotes release of free gadolinium ions, i.e., transmetallation.

Outcome

Of the 248 patients for whom follow-up data was available, 55 experienced NSF symptom improvement, and cure was noted in 5 cases. Partial (n= 8) or complete (n= 3) improvement occurred following restoration of renal function in 11 cases. One NSF patient was reported to have substantial improvement after renal transplant but then deteriorated after allograft rejection.[67] Clinical course was stable in 63 patients. Death was noted in 71 patients but was attributed to NSF in only 3 cases. Cardiopulmonary disease (n= 20), infections (n= 7), general anesthesia (n= 1), SLE (n= 1), lymphoma (n= 1), renal cell carcinoma (n= 1), myeloma (n= 1), toxic megacolon (n= 1), mesenteric ischemia (n= 1), severe stroke (n= 1), multi-organ involvement (n= 2), sudden death (n= 1) and ′unknown’ (n= 30) were the other causes of death reported. In 25 patients, the symptoms were progressive. In another 30 patients, the authors only mentioned “alive” without further details.

Risks of Alternative Procedures

If a contrast-enhanced CT is utilized in place of GBCA-enhanced MRI, there can be risks of radiation-induced cancer, nephrotoxicity, allergic reaction to iodinated contrast and lower diagnostic utility of CT compared to contrast-enhanced MR for some conditions. Younger patients, in particular, face enormously higher risks from radiation compared to older patients, and this needs to be considered when trading off the risk of NSF with the risk of radiation from alternative x-ray-based imaging procedures. If a biopsy is performed instead of characterizing a lesion with GBCA-enhanced MR, the risks of biopsy need to be weighed against the risk of NSF. Significant hemorrhage and other complications occur with liver biopsy in 0.9% to 3.7% of the cases,[120] and the risk of death is 0.1% to 0.01%.[121] Additional risks of biopsy are failure to make a diagnosis, making the wrong diagnosis due to missing the lesion or seeding of the needle track with metastases.[122]

Screening to Identify Patients at Risk

As 81% of NSF patients were on dialysis, this is the most important risk factor for identifying at-risk patients. Renal transplant is another important risk factor. For all inpatients, existing serum creatinine data should be checked to identify patients with GFR <30 mL/min. Although NSF is very rare in outpatients who are not on dialysis, serum creatinine should also be checked in outpatients when severe renal impairment is discovered in the MR patient-safety questionnaire. Acute renal failure is readily detected by asking patients if they are able to urinate.

Discussion and Conclusion

Most institutions appear to have had no NSF cases.[1] Changes in GBCA use since the association between NSF and GBCA was reported in 2006 have nearly eliminated new cases, as reported to the FDA, European Medicines Agency (EMEA) and manufacturers. Thus it appears that attention to NSF risk factors can allow safe administration of GBCA in most patients, including patients with chronic renal insufficiency or on dialysis.

These data compiled from 408 cases reported in the literature suggest that reductions in risk as high as an order of magnitude can be attained with each of the following: 1) avoiding high dose (>0.1 mmol/kg); 2) avoiding nonionic linear chelates in dialysis patients and those with GFR <30 mL/min, especially in the setting of pro-inflammatory conditions; 3) dialyzing quickly after GBCA administration for patients already on dialysis; 4) avoiding GBCA in acute renal failure, especially while serum creatinine is rising. The youngest and oldest reported patients of NSF are 8 years and 87 years of age, respectively, which suggests that infants and very old, more than 87 years of age, individuals are a protected population that is not at risk. Understanding these risk factors can refine practice patterns to allow safe GBCA-enhanced MRI in most patients.