REVIEWS Deferasirox nephrotoxicity—the knowns and unknowns Juan Daniel Díaz-García, Angel Gallegos-Villalobos, Liliana Gonzalez-Espinoza, Maria D. Sanchez-Niño, Jesus Villarrubia and Alberto Ortiz Abstract | In 2005, the oral iron chelator deferasirox was approved by the FDA for clinical use as a first-line therapy for blood-transfusion-related iron overload. Nephrotoxicity is the most serious and frequent adverse effect of deferasirox treatment. This nephrotoxicity can present as an acute or chronic decrease in glomerular filtration rate (GFR). Features of proximal tubular dysfunction might also be present. In clinical trials and observational studies, GFR is decreased in 30–100% of patients treated with deferasirox, depending on dose, method of assessment and population studied. Nephrotoxicity is usually nonprogressive and/or reversible and rapid iron depletion is one of several risk factors. Scarce data are available on the molecular mechanisms of nephrotoxicity and the reasons for the specific proximal tubular sensitivity to the drug. Although deferasirox promotes apoptosis of cultured proximal tubular cells, the trigger has not been well characterized. Observational studies are required to track current trends in deferasirox prescription, assess the epidemiology of deferasirox nephrotoxicity in routine clinical practice, explore the effect on outcomes of various monitoring and dose-adjustment protocols and elucidate the long-term consequences of the different features of nephrotoxicity. Deferasirox nephrotoxicity can be more common in the elderly; thus, specific efforts should be dedicated to investigate the effect of deferasirox use in this group of patients. Díaz-García, J. D. et al. Nat. Rev. Nephrol. 10, 574–586 (2014); published online 22 July 2014; doi:10.1038/nrneph.2014.121
Introduction
Escuela Superior de Medicina del Instituto Politécnico Nacional, Avenida Salvador Díaz Mirón s/n, 11340 Ciudad de México, México (J.D.D.‑G.). IIS-Fundación Jiménez Díaz-UAM, Avenida Reyes Católicos 2, 28040 Madrid, Spain (A.G.‑V., L.G.‑E.). IDIPAZ, Paseo de la Castellana 261, 28046 Madrid, Spain (M.D.S.‑N.). Hospital Ramon y Cajal, Carretera de Colmenar Viejo km. 9,100, 28034 Madrid, Spain (J.V.). Unidad de Diálisis, Fundación Jiménez Díaz-IRSIN, Avenida Reyes Católicos 2, 28040 Madrid, Spain (A.O.). Correspondence to: A.O. [email protected]
Deferasirox (also known as ICL670 and marketed as Exjade®, Novartis Pharma AG, Switzerland) is a potent and specific oral iron chelator that belongs to the N‑substituted bis-hydroxyphenyl-triazole class of triden‑ tate iron chelators (Figure 1).1 The drug was approved as a first-line therapy for blood-transfusion-related iron overload by the FDA in 2005 and the European Medicines Agency (EMA) in 2006. Since this approval, >150,000 patient-years of exposure have occurred.2–6 Deferasirox was designated an ‘orphan medicine’ by the EMA in March 2002.7 Nephrotoxicity identified by an increased serum creatinine (sCr) level, which is observed in more than one in 10 patients, is acknowledged as the most common adverse effect of deferasirox.5 As of December 2013, deferasirox remained a medicine under additional monitoring status because of its nephrotoxic potential. According to the EMA, deferasirox is indicated in patients aged ≥2 years with chronic iron overload caused by transfusion-dependent or nontransfusion-dependent β‑thalassaemia major, other forms of thalassaemia or anaemia.8 The EMA states that treatment should only be initiated if evidence exists of chronic iron overload, that is, after the transfusion of ≥100 ml/kg of packed red blood cells (for example, ≥20 units for an individual weigh‑ ing 40 kg) or when serum ferritin levels are >1,000 μg/l. Deferasirox is contraindicated in patients with impaired Competing interests The authors declare no competing interests.
574 | OCTOBER 2014 | VOLUME 10
renal function; however, the FDA and the EMA propose different glomerular filtration rate (GFR) threshold values for contraindication (Table 1).8–10 The distribution of deferasirox use in routine clinical practice according to underlying diagnosis and age groups is currently unclear. Negative iron balance can be achieved in most patients with a median deferasirox dose of 20 mg/kg per day (approximately 1,500 mg for an individual weighing 80 kg).11 The maximum dose approved by the FDA and the EMA is 40 mg/kg per day.8,9,12–13 Such high doses are reported to be effective with a comparable safety profile to doses of 3 months, includ‑ ing increased sCr levels resulting in an estimated GFR (eGFR) 30 mg per day) or other features of kidney injury, such as proximal tubular injury.30 Unfortunately, the dura‑ tion of renal dysfunction or other features of kidney injury is generally not available in reports of deferasirox nephrotoxicity; thus, the incidence of CKD is unknown. Proximal tubular injury and Fanconi syndrome Fanconi syndrome is a generalized proximal tubular dysfunction that results in glycosuria, hyperphospha‑ turia, aminoaciduria, low-molecular-weight proteinuria,
NATURE REVIEWS | NEPHROLOGY
VOLUME 10 | OCTOBER 2014 | 575 © 2014 Macmillan Publishers Limited. All rights reserved
REVIEWS Table 1 | Recommendations for deferasirox treatment Initial assessment
Monitoring
Contraindications
Dose adjustment
Problems in routine clinical practice
Measure sCr in duplicate
Measure sCr weekly during the first month and at least monthly thereafter
Initial dose: 10–20 mg/kg per day; if Ccr 40–60 ml/min, reduce starting dose by 50%
Estimation of Ccr by Cockcroft–Gault is obsolete
Estimate Ccr by Cockcroft– Gault formula
Monthly monitoring for proteinuria
Ccr 60 ml/min/1.73 m2
FDA
Mentions renal tubular damage but no monitoring
Proximal tubular dysfunction not assessed No action suggested for proteinuria changes
EMA Measure sCr in duplicate
Measure sCr
Ccr 33% above the average of the pretreatment measurements and estimated Ccr* decreases 3 mg/mmol or >30 mg per day
Yes
No
No
Proteinuria
Pathological when >15 mg/mmol or >150 mg per day
Yes
Yes, urine protein:creatinine ratio >0.6 mg/mg (60 mg/ mmol)
Yes
Markers of renal tubular function
No specific cut-off value provided
Yes; for example, renal tubular acidosis, renal potassium wasting, renal magnesium wasting, Fanconi syndrome, nonalbumin proteinuria
No
Glycosuria in patients without diabetes; low serum potassium, phosphate, magnesium or urate; phosphaturia; aminoaciduria
Markers of decreased GFR sCr
No specific cut-off value provided
Should be reported
>33% over baseline or >ULN (in duplicate)
>33% over baseline (in duplicate)
Ccr (estimated by Cockcroft– Gault formula)‡
Not used
Not used
Yes
Yes; dose adjustment if it falls >Cu2+>>Zn2+>Fe2+>>Mg 2+>Ca2+.42
Epidemiology
Studies that have assessed the epidemiology of defera‑ sirox nephrotoxicity used various threshold levels of sCr, serum cystatin C, inulin clearance and tubular dysfunc‑ tion to define kidney injury. Depending on the method used, the reported incidence of nephrotoxicity ranges from 33% increase over baseline on more than two consecutive occasions) were observed in approximately one-third of patients treated with deferasirox.22–24 Treatment was usually continued and the episodes of adverse effects frequently resolved, either spontaneously or after dose interruption or adjustment. The renal abnormalities were mostly reversible, which might be interpreted as an argument to dismiss the potential dangers of deferasirox nephrotoxicity. However, even transient deferasiroxinduced decrements in eGFR could potentially precipi‑ tate serious toxicity of concomitant kidney-excreted or nephrotoxic drugs. The initial safety assessment by the EMA evaluated 652 patients who were treated with deferasirox in four trials, mostly at initial doses of 20–30 mg/kg per day.42 An increase in sCr levels by >33% over baseline was more common in deferasirox-treated patients (24–38%) than in deferoxamine-treated patients (6–14%).42 In 2–17% of patients, deferasirox dose interruption or adjust‑ ments were required because of changes in sCr levels. In 2013, an updated safety evaluation by the EMA analysed
NATURE REVIEWS | NEPHROLOGY
VOLUME 10 | OCTOBER 2014 | 577 © 2014 Macmillan Publishers Limited. All rights reserved
REVIEWS Table 3 | Proximal tubular dysfunction related to deferasirox Mean patient age (years)
Sex
Dose (mg/kg per day)
Time to renal dysfunction (months)*
Time to resolution (weeks)‡
Associated increase in sCr levels
Clinical presentation
Additional information
Reference
Paediatric patients 1.6 1.5 1.9
F F F
25 25 25
NR
No resolution
No No Yes
Fanconi syndrome
Three cycles of induction with tubulotoxic agents for neuroblastoma
37
12.4 ± 3.9
NR
25
NR
NR
Yes
Two patients with Fanconi syndrome; nine patients with at least one sign of proximal tubular dysfunction
NA
38
15 8
M F
20 20
6 6
NR 2
No No
Fanconi syndrome
NA
35
6
M
30
18
2 (partial)
No
Fanconi syndrome
NA
36
16.5
M
29
27
5
No
Fanconi syndrome
NA
39
8.5 11 8
F F M
33 30 30
8 24 36
0.5 8 6
No No No
Fanconi syndrome
NA
40
77
M
24
1
4
Yes
Fanconi syndrome
NA
34
78
M
24
1
4
Yes
Fanconi syndrome
NA
27
28
M
15§
36
0.5 (partial)
No
Hypophosphataemic osteomalacia
NA
31
32
F
38
33
0.5
No
Fanconi syndrome
NA
40
Adults
*From start of deferasirox. ‡From discontinuation of deferasirox. §Estimated from a dose of 1,000 mg per day for a 70 kg patient. Abbreviations: NA, not available; NR, not reported; sCr, serum creatinine.
2,102 patients with β‑thalassaemia major (45% of whom were adults) who were exposed to deferasirox (mean dose 22.4–25.7 mg/kg per day) in six trials.8,44 In five of these studies, an extension phase of up to 5 years was reported.44 After 1 year of therapy, mean sCr levels had increased by 21% and estimated creatinine clearance had decreased by 11% (to –22 ml/min). Renal function deterioration was more severe in patients aged >18 years than in younger patients (decreases in estimated cre‑ atinine clearance of 13.2% versus 9.9%, respectively; P 160 ml/min. At 1 year, 0.6% of patients had a creatinine clearance 50 mg/mmol. In 22 published, randomized clinical trials that included a total of 2,119 deferasirox-treated patients, the overall incidence of nephrotoxicity defined as an increase in sCr levels was 22% (n = 471; Figure 2a and Supplementary Table 2 online).2,9,12,22–24,43,45–59 Similarly, in 16 published clinical practice studies that included a total of 1,373 patients, sCr levels increased in 18% of participants (n = 242; Figure 2b and Supplementary Table 3 online).37,38,40,60–72 However, in clinical trials and 578 | OCTOBER 2014 | VOLUME 10
in clinical practice reports, the incidence of nephro toxicity is higher when defined by a >33% increase in sCr levels over baseline (36% and 28.5%, respectively) than when defined by an increase in sCr levels above the ULN (7.2% and 6.3%, respectively; Figure 2). The former defi‑ nition might be a more sensitive indicator of renal dys‑ function than an increase above the ULN. Incremental changes in sCr levels from 53.04–79.56 μmol/l to 71.60– 106.96 μmol/l (0.60–0.90 mg/dl to 0.81–1.21 mg/dl), that is a >33% increase, represent a decrease in GFR of ~30–40 ml/min/1.73 m2, depending on the age and sex of patients, but do not exceed the ULN values in many laboratories. Thus, incidence of nephrotoxicity depends on the definition of increased sCr levels used, as well as the dose of deferasirox and patient age (Figure 3). In randomized clinical trials and clinical practice studies, the mean age was ≤18 years for 59% (n = 2,051) of patients, 19–64 years for 24% (n = 851)and ≥65 years for 17% (n = 590). In randomized controlled trials, nephrotoxicity incidence was higher in patients with a mean age of ≥65 years than in younger patients when defined as sCr levels above the ULN (Figure 3a), but not when defined as sCr levels above ULN or >33% above baseline (Figure 3d). This difference disappeared when data from randomized clinical trials and clinical practice studies were pooled and all combinations of nephrotoxi city definitions were considered (Figure 3f). However, only one randomized controlled trial involved patients aged ≥65 years and, as monitoring of adverse events in
www.nature.com/nrneph © 2014 Macmillan Publishers Limited. All rights reserved
REVIEWS a 100
b
90
Incidence (%)
80 70 60 50 40 30 20 10 0 >33% >ULN and >ULN
>33%
>33% or ULN
Definition of increase in sCr levels
>33% >ULN and >ULN
>33%
>33% or ULN
Definition of increase in sCr levels
Figure 2 | Incidence of increased sCr levels in patients treated with deferasirox. Pooled data from a | published randomized clinical trials2,9,12,22–24,43,45–59 and b | clinical practice studies.37,38,40,60–72 In four randomized clinical trials and three clinical practice studies, the criteria used to define increased sCR levels were not stated.37,48,54,59,65,69 Abbreviations: sCr, serum creatinine; ULN, upper limit of normal.
clinical practice studies is less stringent than in ran domized controlled trials, safety data in this age group are insufficient. The most widely used mean dose of def‑ erasirox was 20–30 mg/kg per day; this dose was used in studies comprising 80% of patients (n = 2,810). Overall, the incidence of nephrotoxicity in this dose group was higher than at lower doses. No published data are avail‑ able from randomized controlled trials with deferasirox doses of 30 mg/kg per day (Figure 3g–j). In a 1‑year, phase III study in patients with β‑thalassaemia aged ≥2 years who received regular blood transfusions, dose-dependent increases in sCr levels were observed in 38% of patients who received defera sirox, but only in 14% of patients who received deferoxa mine.22 Increases in sCr levels were sometimes transient, generally within the normal range, and did not go above twice the ULN. Dose reduction was required in 13% of all patients because of increased sCr levels; these levels returned to baseline in only 25% of these patients and remained consistently high or fluctuating in 75% of patients.22 In a 4‑year extension study, increase in sCr levels >33% over baseline and above the ULN occurred in 11.2% of patients.46 A 1‑year, phase II trial of deferasirox in regularly trans‑ fused patients with a diverse range of anaemias reported an increase in sCr levels >33% above baseline in 73 of 184 participants (40%).12 Increases in sCr levels occurred at the start of therapy, were nonprogressive and generally remained within the normal range. Deferasirox dose was reduced in 32 patients (17%) and the drug was discon‑ tinued in four patients (2%). sCr levels declined follow‑ ing dose reduction in 14 patients and did not further increase in the remaining patients. In a 1‑year study in patients aged ≥2 years with β‑thalassaemia major and iron overload, most patients initiated deferasirox therapy at a dose of 20 mg/kg per day.2 73 of the 233 participants (31%) had two consecu‑ tive increases in sCr levels >33% above baseline that did not exceed the ULN and an additional six patients (2.5%) had two consecutive increases in sCr levels >33% above
baseline that did exceed the ULN. Deferasirox dose was reduced in nine patients (3.8%). Nephrotoxicity incidence is lower when defined as an increase in sCr levels above the ULN than when defined as an increase in sCr levels >33% above base‑ line. However, this too is a conservative definition of kidney injury. In a phase II randomized clinical trial of deferasirox versus deferoxamine in patients with β‑thalassaemia major and iron overload, no participant had consecutive measurements of sCr levels above the ULN.9 Increases in urinary β2‑microglobulin levels were usually transient and low-grade (33% above baseline was observed in 46 (64%) patients and deferasirox was discontinued in seven (11%) patients.60 Serum cystatin C levels are a more sensitive indica‑ tor of changes in GFR than are sCr levels. Measurement of serum cystatin C levels disclosed 100% (P 6 years with β‑thalassaemia major who were treated with defera‑ sirox, inulin clearance values disclosed a mean reduc‑ tion in GFR of 19.7 ± 11.3%.38 The observation of altered serum cystatin C levels and inulin clearance values con‑ firms the presence of deferasirox-dependent nephro toxicity and excludes the possibility of modulation of tubular secretion of creatinine as a cause for increased sCr levels in deferasirox-treated patients. Information on the occurrence of proximal tubular dysfunction is available in one-third of the studies in approximately 3,500 patients receiving deferasirox pub‑ lished to date.23,51,52,56,57,61,63,67,71 Proximal tubular dysfunc‑ tion was evidenced as Fanconi syndrome in nine of 78 patients (11.5%), increased urinary β2‑microglobulin
NATURE REVIEWS | NEPHROLOGY
VOLUME 10 | OCTOBER 2014 | 579 © 2014 Macmillan Publishers Limited. All rights reserved
REVIEWS
Incidence (%)
a 100 90 80 70 60 50 40 30 20 10 0
b
12 665
172
0
Incidence (%)
e
837
536
Incidence (%) Incidence (%)
86
2,051
851
0
977
229
590
167 ≤18 19–64 ≥65
≤18 19–64 ≥65
≤18 19–64 ≥65
Age group (years)
Age group (years)
Age group (years)
h
i
12 132
220
717
0
j 100 90 80 70 60 50 40 30 20 10 0
237
f
423
g 100 90 80 70 60 50 40 30 20 10 0
Molecular mechanisms
184
48
d 100 90 80 70 60 50 40 30 20 10 0
tubulopathy was mostly reported in children and adolescents with β‑thalassaemia and serum ferritin levels ULN. b,h | Data from randomized controlled trials that defined nephrotoxicity as sCr levels >33% above baseline. c,i | Data from randomized controlled trials that defined nephrotoxicity as sCr levels >ULN and >33% above baseline. d,j | Data from randomized controlled trials that defined nephrotoxicity as sCr levels >ULN or >33% above baseline. e,k | Overall data from clinical practice studies that used any definition of increased sCr levels. f,l | Pooled data from all randomized controlled trials and clinical practice studies that used any definition of increased sCr levels. The numbers above the columns represent the pooled numbers of patients. Abbreviations: sCr, serum creatinine; ULN, upper limit of normal.
levels in 100 of 146 patients (68.5%) and pathological proteinuria in 27 of 816 patients (3.3%).23,51,52,56,57,61,63,67,71 Two patients experienced a generalized proximal tubu lar dysfunction, whereas nine patients presented with at least one sign of proximal tubular dysfunction, such as glycosuria, increased urinary β2‑microglobulin levels, hypophosphataemia or hypercalciuria. 38 Proximal 580 | OCTOBER 2014 | VOLUME 10
The molecular mechanisms of deferasirox nephrotoxicity remain unclear (Figure 4). To prevent or treat defera‑ sirox toxicity two questions should be answered: first, does deferasirox accumulate in proximal tubular cells, and, if so, why? Second, what are the molecular mecha‑ nisms of deferasirox toxicity? Specifically, does toxicity relate to the mechanism of action of the drug (that is, is toxicity related to iron deficiency) or are there addi‑ tional toxic mechanisms that might be dissociated from the mechanism of action of the drug?
Accumulation in kidney To find out why the proximal tubules are sensitive to deferasirox toxicity, we need clues from studies of other specific proximal tubular toxins, such as cidofovir and tenofovir.74 Proximal tubular cells are specialized for transmembrane transport; they are rich in mitochon‑ dria that provide energy for transport processes and in membrane transporters that contribute to excretion of organic anions and cations and recover huge amounts of nutrients and metabolites from the glomerular ultra filtrate. Drugs that are excreted by tubular secretion have rapid access to proximal tubular cells and intracellular concentrations of these drugs are higher in proximal tubular cells than in other cell types. In addition, mito‑ chondrial toxicity might have devastating effects on tubular cell function, resulting in features of Fanconi syndrome, as a consequence of insufficient energy pro‑ duction to support transport processes. Mitochondrial dysfunction could adversely affect proximal tubular cell survival, potentially triggering renal failure.75 Although differences exist between the molecular mechanisms of nucleotide-analogue-related nephrotoxicity (which is thought to result from depletion of mitochondrial DNA) and the known biological activity of deferasirox, the common manifestations of nephrotoxicity induced by these agents supports the hypothesis that defera‑ sirox might injure mitochondria, but not necessarily by damaging mitochondrial DNA. The lipid solubility of deferasirox enables it to easily enter many cells.76 Deferasirox is 99% protein bound, mainly to albumin.8 This high level of protein binding is compatible with tubular secretion rather than glomerular filtration in healthy kidneys. Tubular secretion is con‑ sistent with proximal tubular nephrotoxicity. However, if deferasirox was secreted by the proximal tubules, thera‑ peutic manipulation of nephrotoxicity by modulating the activity of hypothetical deferasirox transporters would be possible. Nevertheless, urine iron elimination does not increase with deferasirox treatment and just 4–8% of the drug is eliminated in urine. This phenomenon argues against substantial tubular secretion. In rats, orally or intravenously administered defera‑ sirox accumulates in the liver.42 This finding is expected because the liver metabolizes and secretes the drug.
www.nature.com/nrneph © 2014 Macmillan Publishers Limited. All rights reserved
REVIEWS
2 Blood circulation
?
Tubular reabsorption of albumin-bound deferasirox?
?
3 Tubular secretion?
Active recovery of ultrafiltrate molecules Tubular lumen
MRP2 5 Low ATP generation Fe3+ Deferasirox
Nucleus
Transporters?
6 Apoptosis
4
Mitochondrion
1 Transporters?
Interstitial space Blood
Figure 4 | Molecular mechanisms of deferasirox toxicity. Although the exact molecular mechanisms of deferasirox toxicity are unclear, several possibilities exist. (1) Proximal tubular cell toxicity is usually a consequence of increased intracellular accumulation of the nephrotoxin owing to active transport by one of the array of molecular transporters expressed by proximal tubular cells; thus, transport of deferasirox into cells might have a role. However, deferasirox is highly membrane permeable and, to date, no such transporters facilitating deferasirox entry have been conclusively identified in proximal tubular cells. (2) Deferasirox could accumulate in the tubular cells through the circulation and exposure of the tubular cells to deferasirox from both the tubular lumen (apical or luminal side) and the blood (basolateral side). Nevertheless, deferasirox is highly albumin bound and unlikely to be filtered into the tubular lumen in great amounts by normal kidneys. (3) The renal excretion of deferasirox (4–8%) might result from proximal tubular secretion (this mechanism has not yet been explored). Alternatively, urinary deferasirox might result from failure of proximal tubules to reabsorb minute amounts of filtered albumin-bound deferasirox. (4) Proximal tubular cells contain a high number of mitochondria, which are key regulators of cellular iron stores and require iron-containing proteins to generate ATP. Deferasirox cytotoxicity was prevented in cultured cells by the addition of iron and is more common in patients responding well to the therapy in terms of iron unloading. Thus, iron chelation might lead to mitochondrial dysfunction and (5) impairment of ATP generation and proximal tubular transport, leading to Fanconi syndrome. (6) Furthermore, mitochondrial injury may activate mitochondria-driven apoptosis, which could lead to tubular cell death. Abbreviation: MRP2, multidrug resistance-associated protein 2.
However, deferasirox also accumulates in the renal cortex, which is unexpected.42 After exposing rats to oral radioactive deferasirox, the highest radioactive deferasirox concentrations were observed in the liver (765 nmol/g) and renal cortex (631 nmol/g). The concen‑ tration of deferasirox in the renal medulla (245 nmol/g) was comparable to that in the blood (213 nmol/g). 42 Accumulation of the drug in the renal cortex is consistent
with its toxicity and with the high vascularization of this tissue. However, how deferasirox gets into the cells and whether the accumulation of the drug in proximal tubular cells is higher than that in other cell types is unknown. Although most circulating deferasirox is the unmetabolized molecule, the question of whether def‑ erasirox or a metabolite accumulates in the renal cortex should also be answered. Physiological albuminuria is 65 years, pre-existing renal or comorbid condi‑ tions such as type 2 diabetes mellitus, medicinal drugs that depress renal function (such as NSAIDs) or co- administration of ciclosporin (Box 2). 49,95,96 These factors might be more common in the elderly or might be associated with disease-causing iron overload as reported for β‑thalassaemia.97 Following an FDA investi gation, a boxed warning has been issued stating that
www.nature.com/nrneph © 2014 Macmillan Publishers Limited. All rights reserved
REVIEWS patients with high-risk myelodysplastic syndromes and underlying hepatic impairment are at increased risk of deferasirox nephrotoxicity.16 In paediatric and young patients, body weight ≥40 kg and the UGT1A1*6 genotype were risk factors for nephrotoxicity. 96 Deferasirox is mainly metabolized through glucuronidation by liver glucuronosyltrans‑ ferase UGT1A1 and is eliminated mainly into the bile (84%) through MRP2.8 The UGT1A1 polymorphisms UGT1A1*28 and UGT1A1*6 could also be risk factors, as they are known to be associated with irinotecan (a drug that inhibits DNA unwinding)-related toxicity.98–100 Like deferasirox, irinotecan undergoes metabolic detoxifica‑ tion by hepatic UGT1A1 to an inactive metabolite.101 Higher doses and rapid iron depletion is also associated with nephrotoxicity.22,82
Clinical challenges and recommendations
As nephrotoxicity is a recognized adverse effect of defera sirox, recommendations by regulatory agencies are in place to minimize the effect on patients (Table 1). These recommendations include contraindication or dose reduction for individuals with impaired renal function, as well as close monitoring and dose adjustment or discon‑ tinuation of therapy if nephrotoxicity develops. However, different criteria are applied by the EMA and the FDA, and some confusing language has been introduced. For example, these agencies refer to estimated creatinine clearance values but the formula currently used estimates GFR. Moreover, agencies measure creatinine clearance in ml/min, which could markedly differ from creatinine clearance expressed as ml/min/1.73 m2. The Cockcroft–Gault formula that estimates cre‑ atinine clearance is obsolete and requires the input of patient weight. Automatic estimation of eGFR by the Modification of Diet in Renal Disease (MDRD) formula has been implemented in some countries and is provided whenever sCr levels are measured, as this value can be calculated from administrative data and sCr analy‑ ses (Table 2).30 Recommendations on dose-reductions by the EMA are based on eGFR values
Introduction
Escuela Superior de Medicina del Instituto Politécnico Nacional, Avenida Salvador Díaz Mirón s/n, 11340 Ciudad de México, México (J.D.D.‑G.). IIS-Fundación Jiménez Díaz-UAM, Avenida Reyes Católicos 2, 28040 Madrid, Spain (A.G.‑V., L.G.‑E.). IDIPAZ, Paseo de la Castellana 261, 28046 Madrid, Spain (M.D.S.‑N.). Hospital Ramon y Cajal, Carretera de Colmenar Viejo km. 9,100, 28034 Madrid, Spain (J.V.). Unidad de Diálisis, Fundación Jiménez Díaz-IRSIN, Avenida Reyes Católicos 2, 28040 Madrid, Spain (A.O.). Correspondence to: A.O. [email protected]
Deferasirox (also known as ICL670 and marketed as Exjade®, Novartis Pharma AG, Switzerland) is a potent and specific oral iron chelator that belongs to the N‑substituted bis-hydroxyphenyl-triazole class of triden‑ tate iron chelators (Figure 1).1 The drug was approved as a first-line therapy for blood-transfusion-related iron overload by the FDA in 2005 and the European Medicines Agency (EMA) in 2006. Since this approval, >150,000 patient-years of exposure have occurred.2–6 Deferasirox was designated an ‘orphan medicine’ by the EMA in March 2002.7 Nephrotoxicity identified by an increased serum creatinine (sCr) level, which is observed in more than one in 10 patients, is acknowledged as the most common adverse effect of deferasirox.5 As of December 2013, deferasirox remained a medicine under additional monitoring status because of its nephrotoxic potential. According to the EMA, deferasirox is indicated in patients aged ≥2 years with chronic iron overload caused by transfusion-dependent or nontransfusion-dependent β‑thalassaemia major, other forms of thalassaemia or anaemia.8 The EMA states that treatment should only be initiated if evidence exists of chronic iron overload, that is, after the transfusion of ≥100 ml/kg of packed red blood cells (for example, ≥20 units for an individual weigh‑ ing 40 kg) or when serum ferritin levels are >1,000 μg/l. Deferasirox is contraindicated in patients with impaired Competing interests The authors declare no competing interests.
574 | OCTOBER 2014 | VOLUME 10
renal function; however, the FDA and the EMA propose different glomerular filtration rate (GFR) threshold values for contraindication (Table 1).8–10 The distribution of deferasirox use in routine clinical practice according to underlying diagnosis and age groups is currently unclear. Negative iron balance can be achieved in most patients with a median deferasirox dose of 20 mg/kg per day (approximately 1,500 mg for an individual weighing 80 kg).11 The maximum dose approved by the FDA and the EMA is 40 mg/kg per day.8,9,12–13 Such high doses are reported to be effective with a comparable safety profile to doses of 3 months, includ‑ ing increased sCr levels resulting in an estimated GFR (eGFR) 30 mg per day) or other features of kidney injury, such as proximal tubular injury.30 Unfortunately, the dura‑ tion of renal dysfunction or other features of kidney injury is generally not available in reports of deferasirox nephrotoxicity; thus, the incidence of CKD is unknown. Proximal tubular injury and Fanconi syndrome Fanconi syndrome is a generalized proximal tubular dysfunction that results in glycosuria, hyperphospha‑ turia, aminoaciduria, low-molecular-weight proteinuria,
NATURE REVIEWS | NEPHROLOGY
VOLUME 10 | OCTOBER 2014 | 575 © 2014 Macmillan Publishers Limited. All rights reserved
REVIEWS Table 1 | Recommendations for deferasirox treatment Initial assessment
Monitoring
Contraindications
Dose adjustment
Problems in routine clinical practice
Measure sCr in duplicate
Measure sCr weekly during the first month and at least monthly thereafter
Initial dose: 10–20 mg/kg per day; if Ccr 40–60 ml/min, reduce starting dose by 50%
Estimation of Ccr by Cockcroft–Gault is obsolete
Estimate Ccr by Cockcroft– Gault formula
Monthly monitoring for proteinuria
Ccr 60 ml/min/1.73 m2
FDA
Mentions renal tubular damage but no monitoring
Proximal tubular dysfunction not assessed No action suggested for proteinuria changes
EMA Measure sCr in duplicate
Measure sCr
Ccr 33% above the average of the pretreatment measurements and estimated Ccr* decreases 3 mg/mmol or >30 mg per day
Yes
No
No
Proteinuria
Pathological when >15 mg/mmol or >150 mg per day
Yes
Yes, urine protein:creatinine ratio >0.6 mg/mg (60 mg/ mmol)
Yes
Markers of renal tubular function
No specific cut-off value provided
Yes; for example, renal tubular acidosis, renal potassium wasting, renal magnesium wasting, Fanconi syndrome, nonalbumin proteinuria
No
Glycosuria in patients without diabetes; low serum potassium, phosphate, magnesium or urate; phosphaturia; aminoaciduria
Markers of decreased GFR sCr
No specific cut-off value provided
Should be reported
>33% over baseline or >ULN (in duplicate)
>33% over baseline (in duplicate)
Ccr (estimated by Cockcroft– Gault formula)‡
Not used
Not used
Yes
Yes; dose adjustment if it falls >Cu2+>>Zn2+>Fe2+>>Mg 2+>Ca2+.42
Epidemiology
Studies that have assessed the epidemiology of defera‑ sirox nephrotoxicity used various threshold levels of sCr, serum cystatin C, inulin clearance and tubular dysfunc‑ tion to define kidney injury. Depending on the method used, the reported incidence of nephrotoxicity ranges from 33% increase over baseline on more than two consecutive occasions) were observed in approximately one-third of patients treated with deferasirox.22–24 Treatment was usually continued and the episodes of adverse effects frequently resolved, either spontaneously or after dose interruption or adjustment. The renal abnormalities were mostly reversible, which might be interpreted as an argument to dismiss the potential dangers of deferasirox nephrotoxicity. However, even transient deferasiroxinduced decrements in eGFR could potentially precipi‑ tate serious toxicity of concomitant kidney-excreted or nephrotoxic drugs. The initial safety assessment by the EMA evaluated 652 patients who were treated with deferasirox in four trials, mostly at initial doses of 20–30 mg/kg per day.42 An increase in sCr levels by >33% over baseline was more common in deferasirox-treated patients (24–38%) than in deferoxamine-treated patients (6–14%).42 In 2–17% of patients, deferasirox dose interruption or adjust‑ ments were required because of changes in sCr levels. In 2013, an updated safety evaluation by the EMA analysed
NATURE REVIEWS | NEPHROLOGY
VOLUME 10 | OCTOBER 2014 | 577 © 2014 Macmillan Publishers Limited. All rights reserved
REVIEWS Table 3 | Proximal tubular dysfunction related to deferasirox Mean patient age (years)
Sex
Dose (mg/kg per day)
Time to renal dysfunction (months)*
Time to resolution (weeks)‡
Associated increase in sCr levels
Clinical presentation
Additional information
Reference
Paediatric patients 1.6 1.5 1.9
F F F
25 25 25
NR
No resolution
No No Yes
Fanconi syndrome
Three cycles of induction with tubulotoxic agents for neuroblastoma
37
12.4 ± 3.9
NR
25
NR
NR
Yes
Two patients with Fanconi syndrome; nine patients with at least one sign of proximal tubular dysfunction
NA
38
15 8
M F
20 20
6 6
NR 2
No No
Fanconi syndrome
NA
35
6
M
30
18
2 (partial)
No
Fanconi syndrome
NA
36
16.5
M
29
27
5
No
Fanconi syndrome
NA
39
8.5 11 8
F F M
33 30 30
8 24 36
0.5 8 6
No No No
Fanconi syndrome
NA
40
77
M
24
1
4
Yes
Fanconi syndrome
NA
34
78
M
24
1
4
Yes
Fanconi syndrome
NA
27
28
M
15§
36
0.5 (partial)
No
Hypophosphataemic osteomalacia
NA
31
32
F
38
33
0.5
No
Fanconi syndrome
NA
40
Adults
*From start of deferasirox. ‡From discontinuation of deferasirox. §Estimated from a dose of 1,000 mg per day for a 70 kg patient. Abbreviations: NA, not available; NR, not reported; sCr, serum creatinine.
2,102 patients with β‑thalassaemia major (45% of whom were adults) who were exposed to deferasirox (mean dose 22.4–25.7 mg/kg per day) in six trials.8,44 In five of these studies, an extension phase of up to 5 years was reported.44 After 1 year of therapy, mean sCr levels had increased by 21% and estimated creatinine clearance had decreased by 11% (to –22 ml/min). Renal function deterioration was more severe in patients aged >18 years than in younger patients (decreases in estimated cre‑ atinine clearance of 13.2% versus 9.9%, respectively; P 160 ml/min. At 1 year, 0.6% of patients had a creatinine clearance 50 mg/mmol. In 22 published, randomized clinical trials that included a total of 2,119 deferasirox-treated patients, the overall incidence of nephrotoxicity defined as an increase in sCr levels was 22% (n = 471; Figure 2a and Supplementary Table 2 online).2,9,12,22–24,43,45–59 Similarly, in 16 published clinical practice studies that included a total of 1,373 patients, sCr levels increased in 18% of participants (n = 242; Figure 2b and Supplementary Table 3 online).37,38,40,60–72 However, in clinical trials and 578 | OCTOBER 2014 | VOLUME 10
in clinical practice reports, the incidence of nephro toxicity is higher when defined by a >33% increase in sCr levels over baseline (36% and 28.5%, respectively) than when defined by an increase in sCr levels above the ULN (7.2% and 6.3%, respectively; Figure 2). The former defi‑ nition might be a more sensitive indicator of renal dys‑ function than an increase above the ULN. Incremental changes in sCr levels from 53.04–79.56 μmol/l to 71.60– 106.96 μmol/l (0.60–0.90 mg/dl to 0.81–1.21 mg/dl), that is a >33% increase, represent a decrease in GFR of ~30–40 ml/min/1.73 m2, depending on the age and sex of patients, but do not exceed the ULN values in many laboratories. Thus, incidence of nephrotoxicity depends on the definition of increased sCr levels used, as well as the dose of deferasirox and patient age (Figure 3). In randomized clinical trials and clinical practice studies, the mean age was ≤18 years for 59% (n = 2,051) of patients, 19–64 years for 24% (n = 851)and ≥65 years for 17% (n = 590). In randomized controlled trials, nephrotoxicity incidence was higher in patients with a mean age of ≥65 years than in younger patients when defined as sCr levels above the ULN (Figure 3a), but not when defined as sCr levels above ULN or >33% above baseline (Figure 3d). This difference disappeared when data from randomized clinical trials and clinical practice studies were pooled and all combinations of nephrotoxi city definitions were considered (Figure 3f). However, only one randomized controlled trial involved patients aged ≥65 years and, as monitoring of adverse events in
www.nature.com/nrneph © 2014 Macmillan Publishers Limited. All rights reserved
REVIEWS a 100
b
90
Incidence (%)
80 70 60 50 40 30 20 10 0 >33% >ULN and >ULN
>33%
>33% or ULN
Definition of increase in sCr levels
>33% >ULN and >ULN
>33%
>33% or ULN
Definition of increase in sCr levels
Figure 2 | Incidence of increased sCr levels in patients treated with deferasirox. Pooled data from a | published randomized clinical trials2,9,12,22–24,43,45–59 and b | clinical practice studies.37,38,40,60–72 In four randomized clinical trials and three clinical practice studies, the criteria used to define increased sCR levels were not stated.37,48,54,59,65,69 Abbreviations: sCr, serum creatinine; ULN, upper limit of normal.
clinical practice studies is less stringent than in ran domized controlled trials, safety data in this age group are insufficient. The most widely used mean dose of def‑ erasirox was 20–30 mg/kg per day; this dose was used in studies comprising 80% of patients (n = 2,810). Overall, the incidence of nephrotoxicity in this dose group was higher than at lower doses. No published data are avail‑ able from randomized controlled trials with deferasirox doses of 30 mg/kg per day (Figure 3g–j). In a 1‑year, phase III study in patients with β‑thalassaemia aged ≥2 years who received regular blood transfusions, dose-dependent increases in sCr levels were observed in 38% of patients who received defera sirox, but only in 14% of patients who received deferoxa mine.22 Increases in sCr levels were sometimes transient, generally within the normal range, and did not go above twice the ULN. Dose reduction was required in 13% of all patients because of increased sCr levels; these levels returned to baseline in only 25% of these patients and remained consistently high or fluctuating in 75% of patients.22 In a 4‑year extension study, increase in sCr levels >33% over baseline and above the ULN occurred in 11.2% of patients.46 A 1‑year, phase II trial of deferasirox in regularly trans‑ fused patients with a diverse range of anaemias reported an increase in sCr levels >33% above baseline in 73 of 184 participants (40%).12 Increases in sCr levels occurred at the start of therapy, were nonprogressive and generally remained within the normal range. Deferasirox dose was reduced in 32 patients (17%) and the drug was discon‑ tinued in four patients (2%). sCr levels declined follow‑ ing dose reduction in 14 patients and did not further increase in the remaining patients. In a 1‑year study in patients aged ≥2 years with β‑thalassaemia major and iron overload, most patients initiated deferasirox therapy at a dose of 20 mg/kg per day.2 73 of the 233 participants (31%) had two consecu‑ tive increases in sCr levels >33% above baseline that did not exceed the ULN and an additional six patients (2.5%) had two consecutive increases in sCr levels >33% above
baseline that did exceed the ULN. Deferasirox dose was reduced in nine patients (3.8%). Nephrotoxicity incidence is lower when defined as an increase in sCr levels above the ULN than when defined as an increase in sCr levels >33% above base‑ line. However, this too is a conservative definition of kidney injury. In a phase II randomized clinical trial of deferasirox versus deferoxamine in patients with β‑thalassaemia major and iron overload, no participant had consecutive measurements of sCr levels above the ULN.9 Increases in urinary β2‑microglobulin levels were usually transient and low-grade (33% above baseline was observed in 46 (64%) patients and deferasirox was discontinued in seven (11%) patients.60 Serum cystatin C levels are a more sensitive indica‑ tor of changes in GFR than are sCr levels. Measurement of serum cystatin C levels disclosed 100% (P 6 years with β‑thalassaemia major who were treated with defera‑ sirox, inulin clearance values disclosed a mean reduc‑ tion in GFR of 19.7 ± 11.3%.38 The observation of altered serum cystatin C levels and inulin clearance values con‑ firms the presence of deferasirox-dependent nephro toxicity and excludes the possibility of modulation of tubular secretion of creatinine as a cause for increased sCr levels in deferasirox-treated patients. Information on the occurrence of proximal tubular dysfunction is available in one-third of the studies in approximately 3,500 patients receiving deferasirox pub‑ lished to date.23,51,52,56,57,61,63,67,71 Proximal tubular dysfunc‑ tion was evidenced as Fanconi syndrome in nine of 78 patients (11.5%), increased urinary β2‑microglobulin
NATURE REVIEWS | NEPHROLOGY
VOLUME 10 | OCTOBER 2014 | 579 © 2014 Macmillan Publishers Limited. All rights reserved
REVIEWS
Incidence (%)
a 100 90 80 70 60 50 40 30 20 10 0
b
12 665
172
0
Incidence (%)
e
837
536
Incidence (%) Incidence (%)
86
2,051
851
0
977
229
590
167 ≤18 19–64 ≥65
≤18 19–64 ≥65
≤18 19–64 ≥65
Age group (years)
Age group (years)
Age group (years)
h
i
12 132
220
717
0
j 100 90 80 70 60 50 40 30 20 10 0
237
f
423
g 100 90 80 70 60 50 40 30 20 10 0
Molecular mechanisms
184
48
d 100 90 80 70 60 50 40 30 20 10 0
tubulopathy was mostly reported in children and adolescents with β‑thalassaemia and serum ferritin levels ULN. b,h | Data from randomized controlled trials that defined nephrotoxicity as sCr levels >33% above baseline. c,i | Data from randomized controlled trials that defined nephrotoxicity as sCr levels >ULN and >33% above baseline. d,j | Data from randomized controlled trials that defined nephrotoxicity as sCr levels >ULN or >33% above baseline. e,k | Overall data from clinical practice studies that used any definition of increased sCr levels. f,l | Pooled data from all randomized controlled trials and clinical practice studies that used any definition of increased sCr levels. The numbers above the columns represent the pooled numbers of patients. Abbreviations: sCr, serum creatinine; ULN, upper limit of normal.
levels in 100 of 146 patients (68.5%) and pathological proteinuria in 27 of 816 patients (3.3%).23,51,52,56,57,61,63,67,71 Two patients experienced a generalized proximal tubu lar dysfunction, whereas nine patients presented with at least one sign of proximal tubular dysfunction, such as glycosuria, increased urinary β2‑microglobulin levels, hypophosphataemia or hypercalciuria. 38 Proximal 580 | OCTOBER 2014 | VOLUME 10
The molecular mechanisms of deferasirox nephrotoxicity remain unclear (Figure 4). To prevent or treat defera‑ sirox toxicity two questions should be answered: first, does deferasirox accumulate in proximal tubular cells, and, if so, why? Second, what are the molecular mecha‑ nisms of deferasirox toxicity? Specifically, does toxicity relate to the mechanism of action of the drug (that is, is toxicity related to iron deficiency) or are there addi‑ tional toxic mechanisms that might be dissociated from the mechanism of action of the drug?
Accumulation in kidney To find out why the proximal tubules are sensitive to deferasirox toxicity, we need clues from studies of other specific proximal tubular toxins, such as cidofovir and tenofovir.74 Proximal tubular cells are specialized for transmembrane transport; they are rich in mitochon‑ dria that provide energy for transport processes and in membrane transporters that contribute to excretion of organic anions and cations and recover huge amounts of nutrients and metabolites from the glomerular ultra filtrate. Drugs that are excreted by tubular secretion have rapid access to proximal tubular cells and intracellular concentrations of these drugs are higher in proximal tubular cells than in other cell types. In addition, mito‑ chondrial toxicity might have devastating effects on tubular cell function, resulting in features of Fanconi syndrome, as a consequence of insufficient energy pro‑ duction to support transport processes. Mitochondrial dysfunction could adversely affect proximal tubular cell survival, potentially triggering renal failure.75 Although differences exist between the molecular mechanisms of nucleotide-analogue-related nephrotoxicity (which is thought to result from depletion of mitochondrial DNA) and the known biological activity of deferasirox, the common manifestations of nephrotoxicity induced by these agents supports the hypothesis that defera‑ sirox might injure mitochondria, but not necessarily by damaging mitochondrial DNA. The lipid solubility of deferasirox enables it to easily enter many cells.76 Deferasirox is 99% protein bound, mainly to albumin.8 This high level of protein binding is compatible with tubular secretion rather than glomerular filtration in healthy kidneys. Tubular secretion is con‑ sistent with proximal tubular nephrotoxicity. However, if deferasirox was secreted by the proximal tubules, thera‑ peutic manipulation of nephrotoxicity by modulating the activity of hypothetical deferasirox transporters would be possible. Nevertheless, urine iron elimination does not increase with deferasirox treatment and just 4–8% of the drug is eliminated in urine. This phenomenon argues against substantial tubular secretion. In rats, orally or intravenously administered defera‑ sirox accumulates in the liver.42 This finding is expected because the liver metabolizes and secretes the drug.
www.nature.com/nrneph © 2014 Macmillan Publishers Limited. All rights reserved
REVIEWS
2 Blood circulation
?
Tubular reabsorption of albumin-bound deferasirox?
?
3 Tubular secretion?
Active recovery of ultrafiltrate molecules Tubular lumen
MRP2 5 Low ATP generation Fe3+ Deferasirox
Nucleus
Transporters?
6 Apoptosis
4
Mitochondrion
1 Transporters?
Interstitial space Blood
Figure 4 | Molecular mechanisms of deferasirox toxicity. Although the exact molecular mechanisms of deferasirox toxicity are unclear, several possibilities exist. (1) Proximal tubular cell toxicity is usually a consequence of increased intracellular accumulation of the nephrotoxin owing to active transport by one of the array of molecular transporters expressed by proximal tubular cells; thus, transport of deferasirox into cells might have a role. However, deferasirox is highly membrane permeable and, to date, no such transporters facilitating deferasirox entry have been conclusively identified in proximal tubular cells. (2) Deferasirox could accumulate in the tubular cells through the circulation and exposure of the tubular cells to deferasirox from both the tubular lumen (apical or luminal side) and the blood (basolateral side). Nevertheless, deferasirox is highly albumin bound and unlikely to be filtered into the tubular lumen in great amounts by normal kidneys. (3) The renal excretion of deferasirox (4–8%) might result from proximal tubular secretion (this mechanism has not yet been explored). Alternatively, urinary deferasirox might result from failure of proximal tubules to reabsorb minute amounts of filtered albumin-bound deferasirox. (4) Proximal tubular cells contain a high number of mitochondria, which are key regulators of cellular iron stores and require iron-containing proteins to generate ATP. Deferasirox cytotoxicity was prevented in cultured cells by the addition of iron and is more common in patients responding well to the therapy in terms of iron unloading. Thus, iron chelation might lead to mitochondrial dysfunction and (5) impairment of ATP generation and proximal tubular transport, leading to Fanconi syndrome. (6) Furthermore, mitochondrial injury may activate mitochondria-driven apoptosis, which could lead to tubular cell death. Abbreviation: MRP2, multidrug resistance-associated protein 2.
However, deferasirox also accumulates in the renal cortex, which is unexpected.42 After exposing rats to oral radioactive deferasirox, the highest radioactive deferasirox concentrations were observed in the liver (765 nmol/g) and renal cortex (631 nmol/g). The concen‑ tration of deferasirox in the renal medulla (245 nmol/g) was comparable to that in the blood (213 nmol/g). 42 Accumulation of the drug in the renal cortex is consistent
with its toxicity and with the high vascularization of this tissue. However, how deferasirox gets into the cells and whether the accumulation of the drug in proximal tubular cells is higher than that in other cell types is unknown. Although most circulating deferasirox is the unmetabolized molecule, the question of whether def‑ erasirox or a metabolite accumulates in the renal cortex should also be answered. Physiological albuminuria is 65 years, pre-existing renal or comorbid condi‑ tions such as type 2 diabetes mellitus, medicinal drugs that depress renal function (such as NSAIDs) or co- administration of ciclosporin (Box 2). 49,95,96 These factors might be more common in the elderly or might be associated with disease-causing iron overload as reported for β‑thalassaemia.97 Following an FDA investi gation, a boxed warning has been issued stating that
www.nature.com/nrneph © 2014 Macmillan Publishers Limited. All rights reserved
REVIEWS patients with high-risk myelodysplastic syndromes and underlying hepatic impairment are at increased risk of deferasirox nephrotoxicity.16 In paediatric and young patients, body weight ≥40 kg and the UGT1A1*6 genotype were risk factors for nephrotoxicity. 96 Deferasirox is mainly metabolized through glucuronidation by liver glucuronosyltrans‑ ferase UGT1A1 and is eliminated mainly into the bile (84%) through MRP2.8 The UGT1A1 polymorphisms UGT1A1*28 and UGT1A1*6 could also be risk factors, as they are known to be associated with irinotecan (a drug that inhibits DNA unwinding)-related toxicity.98–100 Like deferasirox, irinotecan undergoes metabolic detoxifica‑ tion by hepatic UGT1A1 to an inactive metabolite.101 Higher doses and rapid iron depletion is also associated with nephrotoxicity.22,82
Clinical challenges and recommendations
As nephrotoxicity is a recognized adverse effect of defera sirox, recommendations by regulatory agencies are in place to minimize the effect on patients (Table 1). These recommendations include contraindication or dose reduction for individuals with impaired renal function, as well as close monitoring and dose adjustment or discon‑ tinuation of therapy if nephrotoxicity develops. However, different criteria are applied by the EMA and the FDA, and some confusing language has been introduced. For example, these agencies refer to estimated creatinine clearance values but the formula currently used estimates GFR. Moreover, agencies measure creatinine clearance in ml/min, which could markedly differ from creatinine clearance expressed as ml/min/1.73 m2. The Cockcroft–Gault formula that estimates cre‑ atinine clearance is obsolete and requires the input of patient weight. Automatic estimation of eGFR by the Modification of Diet in Renal Disease (MDRD) formula has been implemented in some countries and is provided whenever sCr levels are measured, as this value can be calculated from administrative data and sCr analy‑ ses (Table 2).30 Recommendations on dose-reductions by the EMA are based on eGFR values
