Deferasirox (DFX) is an iron chelator approved for the treatment of iron overload (Porter and Viprakasit, 2014), and widely used in patients with transfusion-dependent anemia, such as thalassemia, sickle cell disease (SCD), and myelodysplastic syndromes (MDS). This excess iron is stored in the form of free iron and deposits into various organs, commonly the liver, heart, and endocrine glands, leading to organ damage and consequence complications (Porter and Viprakasit, 2014). Serum ferritin (SF) level generally correlates with the amount of stored iron in the body and is frequently used to monitor the body’s iron (World Health Organization, 2020; World Health Organization, 2021). Other indicators reflecting the iron level accumulated in the organs are liver iron concentration (LIC) (Telfer et al., 2000), cardiac T2* and R2* (1/T2*) (Ghugre et al., 2006; Bayraktaroğlu et al., 2011), which are measured using magnetic resonance imaging. The reduced T2* value is correlated with higher iron content, deterioration of left ventricular function, and risk of death (Borgna-Pignatti et al., 2004; Bayraktaroğlu et al., 2011; Fucharoen et al., 2014). Severe iron overload is defined as LIC >15 mg Fe/g liver dry weight, cardiac T2* MRI <10 ms, or SF > 2,500 ng/mL (Modell et al., 2000; Porter and Viprakasit, 2014; Shenoy et al., 2014). However, chelation therapy is started when SF exceeded 1,000 ng/mL or LIC >7 mg Fe/g liver dry weight to prevent the consequences of iron overload and prolonged survival (Fucharoen et al., 2014; Porter and Viprakasit, 2014).

As the pharmacological activity of DFX lowers excess iron, DFX binds to the iron. Then, the DFX-iron complex is predominantly eliminated by the hepatobiliary system and excreted via the fecal route (Waldmeier et al., 2010; Chalmers and Shammo, 2016). DFX induces negative iron balance, a decrease in SF, and reduced iron accumulation in the liver and heart (Cappellini et al., 2006; Pennell et al., 2011). In addition, DFX decreases the risk of organ damage due to reactive oxygen species (ROS) resulting from an increase in labile plasma iron and non-transferrin-bound iron (Galanello et al., 2012). Glucuronidation is the primary metabolic pathway for DFX, mainly by UDP glucuronosyltransferase family 1 member A1 (UGT1A1) and, to a lesser extent, by UDP glucuronosyltransferase family 1 member A3 (UGT1A3) (Novartis Pharmaceuticals Canada Inc., 2022). DFX and its metabolites are transported from the liver hepatocyte to the biliary system via multidrug resistance-associated protein 2 (MRP2), then excreted into the feces (Waldmeier et al., 2010). The cytochrome P450 enzymes (CYPs) play minor roles in the oxidative metabolism of DFX by CYP1A and CYP2D6.

Previous studies have reported variability in response to DFX ranging from 25.7% to 67.7% (Cappellini et al., 2006; Porter et al., 2008; Taher et al., 2011; Viprakasit et al., 2013; Cusato et al., 2015; Allegra et al., 2019). The number of pharmacogenetic studies explaining the genetic effects on interindividual differences in DFX responses has been increasing. Several single nucleotide polymorphisms (SNPs) involved with pharmacokinetics and pharmacodynamics of DFX and thalassemia disease have been investigated. To date, none of these studies have provided comprehensive information on the associations between DFX and their genetic responses. It is still unclear which SNPs influence DFX outcomes. This study aims to determine the associations between genetic polymorphisms and DFX outcomes, including pharmacokinetic (PK), iron-chelating efficacy, and adverse drug reactions (ADRs).

A total of 507 articles were initially identified, and 84 duplicates were removed. The remaining articles were screened through the titles and abstracts and refined using the inclusion and exclusion criteria. This resulted in 21 full texts being assessed for eligibility. Of these, 13 studies were included in qualitative synthesis, and seven studies were included in the quantitative synthesis. A PRISMA flow diagram of the process is depicted in Figure 1.

To our knowledge, this is the first study to comprehensively evaluate the influence of genetic polymorphisms on clinical and pharmacological responses of DFX using a systematic review and meta-analysis approach. All the included studies investigated candidate polymorphisms in genes with known functions related to DFX pharmacokinetics/pharmacodynamics and thalassemia disease. All meta-analysis results demonstrated no observed heterogeneity. Our findings revealed the significant association between ABCC2 rs2273697 the increased C_max of DFX, supporting the previously reported associations with other PK outcomes (Cusato et al., 2015; Allegra et al., 2017). This meta-anaysis confirmed the association between rs2273697 and the lower Vd of DFX which was described by Cusato et al., 2016 and Allegra et al., 2017. The previous studies also reported associations between rs2273697 and the toxicity of methotrexate, which is eliminated via the ABCC2 (MRP2) transporter (Izzedine et al., 2006; Ranganathan et al., 2008). Conversely, ABCC2 rs717620 showed associations with PK parameters leading to inadequate response (Cao et al., 2020). Indeed, in vitro assays demonstrated the opposite functions of these missense variants, supporting our finding, that is, rs2273697 (1249G>A) decreased transport activity (Wen et al., 2017) but rs717620 (−24C>T) increased promotor function of ABCC2, which could result in higher expression of this transporter (Nguyen et al., 2013).

Regarding the major metabolism pathway of DFX (Waldmeier et al., 2010), several polymorphisms in UGTs revealed significant associations with PK parameters, efficacy, and toxicity outcomes. However, some variants in this gene demonstrated inconsistent clinical associations between different populations or outcomes. UGT1A1*6 and *28, missense variants leading to reduced enzyme activity (Beutler et al., 1998), showed clinical significance as pathogenic variants for a hyperbilirubinemia condition called Gilbert’s syndrome (Landrum et al., 2017), which is associated with irinotecan toxicity in Asians (Han et al., 2014; Atasilp et al., 2016). For DFX, UGT1A1*6 was a risk factor of creatinine elevation indicating renal toxicity (Lee et al., 2013); but UGT1A1*28 influenced the decrease in steady-state DFX concentrations (Mattioli et al., 2015). In addition, UGT1A1*80, an intron variant in a strong LD with UGT1A1*28 (Gammal et al., 2016), was associated with the improved PK profile and efficacy outcomes in Caucasian (Cusato et al., 2015; Cusato et al., 2016; Allegra et al., 2017; Allegra et al., 2019), but with the reduced PK profile in Chinese (Chen et al., 2020). It should be noted that the distribution of this variant in the Chinese population reported in Chen et al., in 2020 was not in HWE. Another possibility is that a single SNP might not explain all the effects. Due to different genetic variability and distribution across populations, one SNP may be a part of different haplotypes leading to distinct phenotypes and outcomes (Sai et al., 2004). Interestingly, our meta-analysis revealed that rs3806596 (A > G), a non-coding variant in UGT1A3 (Howe et al., 2020), was associated with reduced DFX AUC, as well as improved efficacy by lowering SF levels. These associations could not be directly compared due to the difference in the grouping of genotypes (AG/GG vs. AA for AUC and GG vs. AA/AG for SF level). In addition, a significant association between rs3806596 and increased AUC was found in Caucasian thalassemia patients (Cusato et al., 2016), whereas the meta-analysis included a population combining Chinese healthy adults (Chen et al., 2020) with Caucasian pediatrics (Allegra et al., 2017). Previous research found that hepatic glucuronidation activity in children was lower than in adults (Strassburg et al., 2002), indicating less dependence on UGTs for DFX metabolism. Moreover, the pathology of thalassemia may affect the pharmacokinetic of drugs metabolized by glucuronidation (Temsakulphong et al., 2004). Due to various confounding factors that could affect glucuronidation, as mentioned above, we suggested further research to study the genetic associations between UGT variants and DFX responses in a specific population concerning these factors.

As the relationship between vitamin D deficiency and iron deficiency or β-thalassemia has been indicated (Napoli et al., 2006; Mogire et al., 2022), variants in genes related to vitamin D were investigated for their impact on DFX responses. Among these, two variants in CYP24A1 showed associations with the reduced PK parameters in adult and pediatric patients with β-thalassemia. The meta-analysis confirmed the previously reported associations of rs2248359 with C_trough and C_min, and of rs2585428 with C_trough and the half-life of DFX (Allegra et al., 2018b). Additionally, rs2585428 was associated with a lower value of cardiac T2*, indicating the inferior efficacy of DFX in removing cardiac iron. (Allegra et al., 2018c). Furthermore, rs2248359 is located in the transcription factor binding site in the promotor region, and rs2585428 is in the intron of CYP24A1 (Howe et al., 2020). To date, there has been no reports on the direct function of these two variants. It is interesting to further investigate the function of these variants and validate their associations in different ethnic groups. Furthermore, rs4646536 (C allele) and rs10877012 (T allele) in CYP27B1 significantly decreased PK and the efficacy of DFX (Allegra et al., 2018a; Allegra et al., 2019). These two intron variants were in complete LD and exhibited a similar function to increase 1α-hydroxylase, a gene product of CYP27B1 involved with vitamin D metabolism (Clifton-Bligh et al., 2011). Interestingly, the 1000 Genomes Project reported the disparity of the frequencies of rs4646536 (C allele) and rs10877012 (T allele) among different ethnicities (Auton et al., 2015), that is, 32% in Europeans but 64%–65% in East Asians. These results suggest the significance of variations in genes involved with vitamin D metabolism, especially in East Asian populations.

This study has some limitations. First, due to the different genotype grouping in previous studies, some significant associations could not be pooled in our meta-analysis. Second, the polygenic effect on drug responses could not be analyzed because we could not access to the individual data in previous publications. Although the mechanism underlying DFX as iron chelation is simple, the target organs in which iron is chelated include the bloodstream and the liver and heart (Porter and Viprakasit, 2014). DFX pharmacokinetics is also complicated, including its metabolism via several pathways and enterohepatic circulation (Waldmeier et al., 2010). In addition, personal conditions, such as age, underlying disease, physiological alteration, and organ impairment, may affect drug properties (Whittaker et al., 2018; García-Cortés and García-García, 2022), drug efficacy and toxicity (Veyssier, 1992). Therefore, the response to DFX can be influenced by various genetic polymorphisms involving pathophysiology and disease progression. These cause difficulties in the interpretation and evaluation of the consequences of genetic variants on DFX PK, efficacy, and toxicity outcomes.

Polymorphisms in ABCC2 should be considered for both increasing and decreasing DFX responses. The impacts of ABCC2 rs2273697 and rs71762 on DFX responses and transporter function have been well established by clinical and in vitro studies, respectively, and should be evaluated for clinical use. We recommend validating the controversial associations through clinical studies (e.g., polymorphisms in UGT genes) and in vitro methods (e.g., CYP24A1 rs2248359 and rs2585428). These associations need to be further clarified in terms of ethnic differences that may possibly affect allele frequency, the physiological difference between children versus adults or between healthy persons versus patients, and the pharmacological mechanism by which genetic variation influences DFX response. Furthermore, to integrate the total influence of functional genetic polymorphisms, polygenic association analysis should be performed and ultimately used to predict drug responses.