The MTHFR (Methylenetetrahydrofolate reductase) gene encodes a 77-kDa MTHFR enzyme which participates in folate metabolism, specifically in the conversion of 5,10-methylenetetrahydrofolate to 5-methyltetrahydrofolate (5-MTHF), and this process is crucial for the synthesis of nucleotides and DNA. Several studies, including meta-analyses, have reported an association between the two most common SNPs of the MTHFR gene, namely c.667C>T (rs1801133) and c.1298A>C (rs1801131), with MTX toxicity (Song et al., 2014; Dwivedi et al., 2020; Yang et al., 2023; Martusevich et al., 2020; Von Feldt et al., 2006; Juster-switlyk et al., 2017; Zhao et al., 2016; Campbell et al., 2016). In c.667C>T polymorphism, alanine is substituted with valine due to a change from C to T at nucleotide 677 (Rosenberg et al., 2002). Meanwhile, in c.1298A>C polymorphism, A is replaced with C at position 1,298, leading to the substitution of alanine to glutamine (Sharaki et al., 2018). Both SNPs lead to the lower enzymatic activity of MTHFR, affecting MTX pharmacodynamics (Van Der Put et al., 1998). Consequently, there is a decrease in the production of 5-MTHF, which serves as a methyl donor in the re-methylation of homocysteine to methionine (Figure 1). This leads to homocysteine accumulation in the blood, a condition known as hyperhomocysteinemia. Elevated homocysteine levels have been associated with increased toxicity of MTX and elevated cardiovascular risks, such as coronary artery calcification, high blood pressure, and many more (Von Feldt et al., 2006; Gande et al., 2023). The exact mechanism underlying MTX toxicity caused by lower MTHFR enzymatic activity remains unclear and requires further investigation.

ATIC gene is located at chromosome 2q35 and encodes aminoimidazole carboxamide adenosine ribonucleotide transformylase (ATIC) that participates in the de novo purine synthesis and transforms aminoimidazole carboxamide adenosine ribonucleotide (AICAR) into formyl- AICAR. MTX mechanisms of action includes inhibiting ATIC after entering the cells, causing AICAR intracellular accumulation (Figure 1). This leads to the release of adenine into extracellular which inhibits the functions of several immune cells, such as monocytes, T-lymphocytes, and NK cells, initiating anti-inflammatory activities (Lee and Bae, 2016). The most commonly explored ATIC polymorphism is ATIC c.347C>G, with the C to G variation prompting the change of threonine to serine at position 116 of gene. Several studies showed that patients with the ATIC c.347G allele had a higher risk of MTX-related toxicity, specifically GI toxicity (Muralidharan et al., 2016; Grabar et al., 2010; Huang et al., 2020; Londono et al., 2020). Other studies on the same polymorphism did not report a relationship between gene and toxicity, but associated ATIC c.347C>G with the efficacy or non-responsiveness of MTX (Lee and Bae, 2016; Sha et al., 2022). A meta-analysis consisting of nine comparative studies showed that ATIC c.347C>G polymorphism might be associated with MTX toxicity in Caucasians, compared to Asian patients. This scientific finding remains uncertain and tends to be associated with a higher frequency of the allele in Caucasians (Lee and Bae, 2016).

TPMT gene is responsible for encoding thiopurine S-methyltransferase (TPMT), an enzyme that contributes to thiopurine drugs metabolism in cells. TPMT catalyzes the S-methylation of thiopurine including AZA into inactive and non-toxic forms. There are several alleles of the TPMT gene (TPMT*2 to TPMT*20) which affect TMPT activity, but the most frequently studied polymorphisms associated with AZA toxicity are TPMT*2 c.238G>C (rs1800462), TPMT*3B c.460G>A (rs1800460), and TPMT*3C c.719A>G (rs1142345). TPMT*2, TPMT*3B, and TPMT*3C are more common in Caucasians and Africans, while TPMT*3C is the most frequent in the Asian population (Gu et al., 2023). TPMT*2 c.238G>C were reported to prompt the substitution of alanine to proline. TPMT*3B c.460G>A and TPMT*3C c.719A>G are responsible for the deficiency or the loss of TPMT activity leading to AZA-related toxicity because of a decrease in the drug metabolism (Figure 2) (Steponaitiene et al., 2016). The reduction of AZA metabolism causes excessive accumulation of cytotoxic compounds, TGNs, inducing severe toxicity such as bone marrow toxicity and GI manifestations (Murugesan et al., 2009).

The data reported for the association of TMPT alleles and AZA-induced hepatotoxicity were conflicting. A meta-analysis showed that TPMT variations were not associated with hepatotoxic (Liu et al., 2015). Other studies found a higher risk of hepatotoxicity due to these polymorphisms, hence individualized AZA dosing could minimize the risk (Sheu et al., 2022). There was no correlation between the TPMT genotype and leukopenia incidence (p = 0.95) in Chinese autoimmune patients (Fei et al., 2018a). In the case of myelotoxicity, the data reported were conflicting because the rate of myelosuppression was significantly higher in Chinese patients with TPMT*2 than TPMT*3B and TPMT*3C polymorphisms, which could lead to clinical failure of AZA treatment (Fei et al., 2018a). Significant relationship was not observed between myelotoxicity and the TPMT polymorphisms (p = 0.973) in 70 Chinese patients receiving AZA (Su et al., 2020). Therefore, the presence of TPMT polymorphisms alone may not serve as a universally reliable predictor for AZA-induced myelotoxicity, particularly in certain populations such as East Asians, where additional genetic variants (e.g., NUDT15) have shown greater clinical relevance.

NUDT15 gene encodes NUDT15 enzymes, which function to dephosphorylate thiopurine triphosphate to monophosphate. This is included among the metabolism pathways of thiopurine drugs, such as AZA. The most common alleles studied and found to have an association with AZA-related toxicity consisted of NUDT15*3. This variant c.415C>T (rs116855232) causes the change of arginine to cysteine at position 139 and leads to function loss of NUDT15 enzyme due to a decrease in the thermal stability (Yang et al., 2014). The loss of NUDT15 function induces the excessive amount of thiopurine triphosphate and increases the number of TGNs incorporated into DNA and RNA, leading to severe AZA-related cytotoxicity. Among all polymorphisms related to AZA toxicity, NUDT15*3 was reported to have the strongest association with myelosuppression and leukopenia in patients, specifically Asians (Song et al., 2014; Martusevich et al., 2020; Wang et al., 2022). This is due to the frequency of the allele being higher in Asians population than in Caucasians (Tanaka and Saito, 2021). The genetic screening of NUDT15*3 gene has been implemented for personalizing AZA doses to decrease the risk of AZA-induced leukopenia in China (Wang et al., 2022).

ITPA gene encodes ITPA which contributes to thiopurine metabolism, such as NUDT15 and TPMT enzymes. This has a role in preventing toxicity by restricting the accumulation of toxic thiopurine metabolites, 6-thioinosine-5-triphosphate, through conversion to TIMP (Nagamine et al., 2012). Gene variant of interest is ITPA c.94C>A (rs1127354), a missense mutation reported to be responsible for the decrease or even the loss of ITPA enzymatic activity (Cao and Hegele, 2002; Sumi et al., 2002). This impairs AZA metabolism process, leading to toxicity in patients with SLE and other diseases requiring AZA treatment (Nagamine et al., 2012). Compared to TPMT variants and NUDT15, ITPA c.94C>A was not associated with myelosuppression and hepatotoxicity (Steponaitiene et al., 2012; Chen et al., 2019), but influenza-like symptoms, digestive intolerance, pancreatitis, and arthralgia (Nagamine et al., 2012; Chen et al., 2021). The exact reason and mechanism of the phenomena remain inconclusive.

ABCC4 gene is responsible for encoding MRP4, an ATP-binding cassette transporter that functions as a transmembrane efflux pump to transfer 6-TGN out of the cell. ABCC4 c.2269G>A (rs3765534) variant is suspected to be related to the side effects of leukopenia in patients given thiopurine because of 6-TGN accumulation (Song et al., 2014; Martusevich et al., 2020). MRP4/ABCC4 c.2269G>A (rs3765534) decreases MRP4 function, might be responsible for myelosuppression (Milosevic et al., 2018), and is rarely found in the Caucasian race. However, the allele has been found at a higher frequency among the Asian population, including 14.7%–23% in the Japanese and 8.3% in the Han Chinese (Juster-switlyk et al., 2017; Zhao et al., 2016; Campbell et al., 2016).

GST genes encode glutathione S-transferase (GST), an important enzyme involved in the detoxification of cyclophosphamide metabolites. GST catalyzes the conjugation of reactive metabolites, such as phosphoramide mustard, into less toxic compounds like 4-glutathionyl-cyclophosphamide (Dirven et al., 1994). A reduction or loss of GST activity can lead to the accumulation and prolonged exposure to these toxic metabolites, thereby increasing the risk of adverse effects and CYC-related toxicity (Hajdinák et al., 2020; Conklin et al., 2015). The well-known GST genes variants associated with occurrence of adverse effects-related to CYC are GSTM1, GSTP1 and GSTT1 (Audemard-Verger et al., 2016; Hahn et al., 2010). GSTA1 is not included in this discussion, as its polymorphisms have been reported to influence the efficacy of CYC treatment rather than its toxicity or adverse effects (Indrawijaya et al., 2024; Wang et al., 2015).

GSTM1 gene encodes Glutathione S-transferase Mu 1 (GSTM1), a key phase II detoxification enzyme that catalyzes the conjugation of glutathione to electrophilic compounds, including toxic metabolites of cyclophosphamide such as acrolein and phosphoramide mustard (NCBI, 2025a) GSTM1 null genotype is the common polymorphism in GSTM1. Individuals with this genotype do not produce functional GSTM1, leading to impaired detoxification capacity and accumulation of acrolein and phosphoramide mustards. There is a strong evidence that the GSTM1 null genotype increases the risk of adverse effects from cyclophosphamide, regardless of other patient characteristics (age, gender, kidney function, and total CYC dose), with the odds ratio was 3.345, compared to those with functioning GSTM1 gene (Audemard-Verger et al., 2016). While some studies have reported a significant association between the GSTM1 null genotype and increased CYC-related adverse effects, particularly when adjusted for clinical variables, others have found no relationship in the context of short-term high-dose regimens (Zhong et al., 2006). These discrepancies may reflect differences in study design, treatment protocol, population genetics, or definitions of toxicity.

GSTP1 gene encodes glutathione S-transferase Pi 1, an enzyme involved in the phase II detoxification of reactive drug metabolites, including those generated during cyclophosphamide (CYC) metabolism. GSTP1 catalyzes the conjugation of electrophilic CYC byproducts with glutathione, facilitating their elimination (NCBI, 2025b; Hayes et al., 2005). A commonly studied variant of this gene is Ile105Val polymorphism (rs1695), which involves a single nucleotide change from A to G, resulting in the substitution of isoleucine (Ile) with valine (Val) at codon 105 of the enzyme (Hasni et al., 2016). This amino acid change alters the structure and function of GSTP1, potentially reducing its catalytic efficiency (Gorukmez et al., 2016). As GSTP1 is only active in its dimer form, both heterozygous and homozygous variant causes reduction in GTSP1 activity, causing an increase of acrolein and phosphoramide mustard concentration. Although increased levels of phosphoramide mustard are associated with improved cyclophosphamide efficacy in cancer patients, in SLE patients this increase has been linked to a higher incidence of myelotoxicity, particularly at higher cyclophosphamide doses. In contrast, gastrointestinal (GI) toxicity tends to be more frequent at lower doses of cyclophosphamide in SLE, likely due to its non-linear pharmacokinetic profile, which results in greater biliary excretion of toxic metabolites following low-dose administration compared to high-dose administration (Zhong et al., 2006).

GSTT1 gene encodes for the enzyme glutathione S-transferase theta 1, which also plays a role in the detoxification of toxic metabolites of CYC. A common polymorphism involves a homozygous deletion of the GSTT1 gene, referred to as the GSTT1 null genotype, resulting in the absence of functional enzyme activity. While the GSTT1 null genotype has been associated with increased chemotherapy-related toxicities in oncology settings (Cho et al., 2010; Wang et al., 2016; Aguiar et al., 2012), this variant does not appear to significantly affect the risk of cyclophosphamide-related adverse effects—such as myelosuppression and gastrointestinal toxicity—in SLE patients (Zhong et al., 2006). This may be due to higher, more frequent doses of CYC, and different patient characteristics in cancer patients compared to SLE patients.

CYP genes are responsible for encoding a family of Cytochrome P450 enzymes which activate CYC to the active metabolite, 4-OH-CPA. It is generated mainly by CYP2B6, CYP3A4, and CYP2C9, with additional contributions from CYP2C19, CYP2A6, and CYP2C8. Genetic polymorphisms in these enzymes can significantly influence CYC’s pharmacokinetics, leading to interindividual differences in efficacy and toxicity. Excessive amounts or prolonged duration of active metabolites exposure to body cells may be associated with CYC-related adverse effects, such as ovarian toxicity. Among these enzymes, the most studied polymorphism variants are CYP2B6, which has been studied in relation to CYC metabolism and toxicity. Three notable variants of CYP2B6 include c.516G>T (rs3745274), c.785A>G (rs2279343), and −750T>C (rs4802101). The c.516G>T variant, which defines the CYP2B6*6, results in the substitution of glutamine with histidine at codon 172 (Q172H), leading to reduced enzyme expression and enzyme (Abuelsoud et al., 2021). The c.785A>G (rs2279343) variant causes a lysine-to-arginine substitution at position 262 (K262R) and often co-occurs with c.516G>T as part of the CYP2B6*6, further contributing to impaired metabolic activity (Tran et al., 2008). This may reduce the formation of active metabolites, potentially lowering toxicity but also decreasing efficacy. Meanwhile, the −750T>C (rs4802101) variant is located in the promoter region, which may influence gene expression levels, though its clinical impact remains inconclusive (Zanger and Klein, 2013). For instance, a study involving 116 Chinese patients, individuals carrying the C allele had a significantly lower incidence of gastrointestinal toxicity and leukocytopenia compared to those with the wild-type TT genotype (Shu et al., 2016). However, other studies have reported that there was no association between CYP2B6 polymorphisms and reduced CYC-related adverse effects (Kumaraswami et al., 2017; Singh et al., 2007).

CYP3A4 is another major to CYC metabolism. One notable polymorphism of CYP3A4 is CYP3A4*1B, characterized by an A>G transition in the 5′-flanking region of the gene. This variant has been associated with altered gene expression and enzymatic activity (Su et al., 2010). While there is still no study investigating the functional impact of this variant on CYC-related adverse effects in SLE patients, a study on cancer patients found no significant association between CYP3A4*1B and the pharmacokinetics of CYC or its active metabolite, 4-OH-CPA (Ekhart et al., 2008). CYP2C9 also plays a role in cyclophosphamide (CYC) metabolism, but contributes only minimally to its activation into 4-hydroxycyclophosphamide (4-OH-CPA) (Griskevicius et al., 2003).

CYP2C19, CYP2A6, and CYP2C8 play ancillary roles in CYC metabolism (Muñiz et al., 2022). Among these, the CYP2C19*2 is one of the most extensively studied variants due to its high frequency in many populations, especially in Asian population, and its association with reduced enzymatic activity (Audemard-Verger et al., 2016). The CYP2C19*2 c.681G>A (rs4244285) in exon 5 is responsible for a decrease in CYP enzyme activity. There are conflicting data regarding whether CYP2C19 is associated with a higher risk of CYC toxicity. Several studies stated that the presence of this allele is related to ovarian toxicity and the risk of CYC treatment failure (Kumaraswami et al., 2017; Lee et al., 2016; Ngamjanyaporn et al., 2011). However, a study associated CYP2C19*2 presence with a lower risk of toxicity, specifically ovarian toxicity in Indian patients (Singh et al., 2007). Another study reported that CYP2C19*2 had no association with CYC-related toxicity (Audemard-Verger et al., 2016). Further investigation is required to understand more about the reason for these results. Additionally, studies on genetic polymorphisms of CYP2A6 and CYP2C8 in the context of CYC metabolism are limited, particularly in patients with systemic lupus erythematosus (SLE). Thus, their clinical relevance in modulating CYC efficacy or toxicity remains unclear and warrants further investigation.

ALDH encodes Aldehyde Dehydrogenase (ALDH), an enzyme that contributes to CYC metabolism and detoxification, such as GST. ALDH converts aldophosphamide to a non-toxic metabolite, carbophosphamide. The presence of two variants (ALDH3A1*2 and ALDH1A1*2) is known to reduce the activity of the ALDH enzyme, which affects the detoxification capacity of cyclophosphamide (CYC), theoretically increasing the risk of toxicity from CYC (Ekhart et al., 2008). A study involving 113 Caucasian patients receiving high-dose chemotherapy with a combination of CYC, thiotepa, and carboplatin showed that the ALDH1A1*2 variant, located in the promoter region and potentially having significant gene regulatory effects, was associated with an increased risk of liver toxicity and hemorrhagic cystitis (Ekhart et al., 2008). Another study at Michigan Hospital involving 846 patients receiving CYC-based chemotherapy regimens indicated that the presence of the ALDH1A1 c.1234A>G (rs8187996) variant was actually associated with a reduced risk of ≥3 toxicity or the need for treatment modification due to toxicity. The rs8187996 variant is located in the intron (non-coding) region, which has a minor effect on gene regulation (Hwang et al., 2022). The differences in the location and functional effects of the genetic variants studied, as well as the differences in chemotherapy regimens used, may explain the differing results between the two studies.

IMPDH encodes an enzyme called inosine 5′-monophosphate dehydrogenase 1 (IMPDH), which functions to produce the guanosine required in the lymphocytes proliferation process. MPA works by inhibiting IMPDH activity, while IMPDH1 and IMPDH2 variants influence the incidence of MMF toxicity. The variant of IMPDH1 -106C>A (rs2278294) and c.1575G>A (rs2228075) were strongly associated with infection susceptibility and GI intolerance, such as nausea, vomiting, and diarrhea (Ohmann et al., 2010; Shu et al., 2021). IMPDH1 encodes inosine monophosphate dehydrogenase type I, a key enzyme involved in the de novo synthesis of guanine nucleotides, which is essential for the proliferation of T and B lymphocytes. Mycophenolic acid (MPA), the active metabolite of mycophenolate mofetil (MMF), exerts its immunosuppressive effects by inhibiting this enzyme. Certain IMPDH1 polymorphisms, such as nonsynonymous variants including rs2278294 and rs2228075, have been reported to reduce enzymatic activity through altered protein stability and impaired tetramer formation, without significantly affecting mRNA levels, indicating a post-translational regulatory mechanism. This reduction in IMPDH1 activity may enhance the pharmacodynamic effects of MPA, thereby intensifying immunosuppression and predisposing patients to infection. Additionally, enhanced local MPA effects in the gastrointestinal tract may contribute to increased risk of nausea, vomiting, and diarrhea. Ethnic differences in the distribution of IMPDH1 haplotypes may also partially explain interindividual variability in toxicity profiles (Wu et al., 2010). Meanwhile, IMPDH2 c.3757T>C (rs11706052) was associated with bone marrow toxicity due to an increase in IMPDH activity (Neerman and Boothe, 2003).

UGT encodes UDP-glucuronosyltransferases (UGTs), such as UGT1A9 and UGT1A8, which convert MPA to the inactive MPAG. However, one variant of UGT known as UGT2B7 produces a minor highly reactive metabolite, AcMPAG, which is associated with drug toxicity (Bernard et al., 2006). A UGT2B7 variant called UGT2B7 -900A>G (rs7438135) was associated with a higher risk of leukopenia and anemia. In this variant, adenine (A) at coding DNA position 900 is replaced by guanine (G), causing a decrease in UGT2B7 activity and leading to an accumulation of toxic metabolite, AcMPAG. Another genetic variant, UGT2B7 c.802C>T (rs7439366), was identified with contributions to increased susceptibility to infections, particularly Pneumocystis carinii pneumonia (Shu et al., 2021).

SLCO1B1 or solute carrier OATs family member 1B1, is a gene that encodes a membrane transporter protein primarily found in the liver. This transporter plays a crucial role in facilitating various endogenous and exogenous substances, including MMF, across cell membranes. SLCO1B1 participates in the clearance of numerous drugs, and the most common alleles studied include SLCO1B1 c.521T>C (rs4149056) that causes the substitution of valine to alanine at position 174. The substitution leads to a decrease in the ability of the transporter to facilitate MPA intake, increasing MPA plasma concentration. SLCO1B1 c.521T>C was associated with MMF-induced anemia (Shu et al., 2021), but a previous study did not find a relationship between this polymorphism and toxicity (Neerman and Boothe, 2003).