Tumor Necrosis Factor alpha inhibitors (TNFi) are widely used in the clinical setting to treat severe autoimmune diseases as they have shown promising efficacy and safety (1). As the name suggests, TNFi are designed to block the action of Tumor Necrosis Factor alpha (TNFα), which is one of the most potent proinflammatory cytokines. TNFα has been associated with the pathogenesis of several autoimmune diseases (ADs) including rheumatoid arthritis (RA) (2–4), inflammatory bowel disease (IBD) (5–7), psoriasis (PS) (8, 9), and ankylosing spondylitis (AS) (10), serving as a driver of chronic inflammation (11). TNFα is a type II transmembrane protein produced mainly by macrophages and to a lesser extent by T and B lymphocytes, natural killer cells, and neutrophils (12, 13). It exists in a membrane-bound form (mTNFα) and a soluble form (sTNFα) (14). Both forms of TNFα bind to specific receptors known as TNFR1 (also known as TNFRSF1A or p55) and TNFR2 (also known as TNFFSRF1B or p75); however, sTNFα has less affinity to TNFR2 (15). The transmembrane form of TNF is the prime activating ligand of the 80 kDa tumor necrosis factor receptor (16). TNFR1 is universally present on the membranes of almost all nucleated cells while TNFR2 is primarily present on membranes of immune cells, endothelial cells, and some tumor cells (17, 18). Upon binding TNFR1 and TNFR2, TNFα initiates a complex cascade of molecular signals for various biological functions and different cellular responses including inflammation, cell death, cell proliferation, and differentiation (19, 20). The signaling cascade initiated upon the activation of TNFR1 by TNFα leads to the activation of the Nuclear factor kappa-light-chain-enhancer of activated B cells (NF-κB) (21) pathway and the mitogen-activated protein kinases (MAPK) pathway (22), which subsequently induce inflammation, immune response, tissue degeneration, and cell survival and proliferation. Furthermore, the binding of TNFα to TNFR1 can mediate cell death via apoptosis or necroptosis through activating caspase-8 or mixed lineage kinase domain-like protein (MLKL), respectively. In contrast to TNFR1, TNFR2 lacks a death domain and thus does not prompt cell death directly. However, similar to TNFR1, the activation of TNFR2 stimulates NF-κB, MAPKs, and protein kinase B (AKT) pathways, promoting cell proliferation, tissue regeneration, and inflammatory responses against pathogens (23). TNFα drives inflammation by stimulating the secretion of other inflammatory cytokines, activating immune cells, and amplifying its own expression through a positive feedback loop (24). The rationale behind TNFα being a suitable target in ADs lies in its central role in inflammation and that it is “at the apex” of the signaling cascade of the pro-inflammatory cytokines (25), along with its implication in the pathogenesis of several ADs. In addition, early experiments proving the efficacy of the first TNFi, infliximab, in AD patients (26), opened the door for developing and improving anti-TNF treatment as a therapeutic approach in several ADs gradually ( Figure 1 ).

Adalimumab, Infliximab, Certolizumab pegol, Golimumab, and Etanercept are the five TNFi approved by the US Food and Drug Administration (FDA) (27). The timeline for the development of TNFi has been comprehensively discussed in previous reviews (20, 28), from which we adapted and modified the figure presented here ( Figure 1 ). Apart from Etanercept, which is a fusion protein that serves as a receptor for TNFα, the other four TNFi are monoclonal antibodies (mAbs). Each TNFi has two functional regions: the constant region (Fc) and the variable region (Fab), except for Certolizumab pegol, which lacks the constant region. When the Fab region of the mAb binds to TNFα, it prevents TNFα from interacting with its receptors, which is the therapeutic objective of this family of compounds (13). Although these inhibitors have been used since the past decade, the complete mechanism of action is still unclear. Previous studies have reported that TNFi neutralizes TNFα, induces direct and indirect apoptosis, modulates the immune system, induces Fc-dependent apoptosis, and promotes outside-to-inside signaling. The inflammation is reduced by the immediate neutralization reaction driven by the TNFi, which inactivates the proinflammatory cytokine TNFα (29). Nonetheless, these inhibitors appear to have more complex activities than simple blocking, owing to the complexity of the TNFα signaling. TNFi have substantially improved the treatment course of autoimmune diseases. However, response to TNFi is significantly variable, with up to 40% of patients not experiencing a positive clinical outcome. This lack of response can be classified as primary, where patients fail to respond from the outset, or secondary, where the response diminishes over time despite initial effectiveness (30).

Pharmacogenomic (PGx) variants influence how individuals respond to drugs. Variable responses to medications among ethnic groups have been attributed to the diversity of PGx variants. Previous studies highlighted that approximately 20-30% of the variability in drug efficacy and toxicity can be better elucidated by exploring PGx variants associated with drug response (31). Genetic variants have been identified in pharmacogenes that affect both pharmacokinetics and pharmacodynamics of specific drugs. Moreover, previous studies have identified that more than 90% of patients carry at least one variant in pharmacogenes that prompts a change in dosage or drug selection (32, 33). After identifying genetic loci associated with drug responses, it becomes imperative to analyze the distribution of these variants in populations, particularly because some variants, such as the human leukocyte antigen (HLA) variants, increase the risk of severe adverse effects in response to some medications. This review discusses the pharmacokinetics, pharmacodynamics, and pharmacogenomics aspects of TNFα inhibitors, with a particular focus on HLA variants.

TNF inhibitors are protein compounds that target either membrane-bound or soluble TNF (34). These medications have complex pharmacokinetics and pharmacodynamics features due to the high affinity to their pharmacological target, compared to small-molecule medicines (35). Since TNFi are large protein molecules with low membrane permeability, they are poorly absorbed in the gastrointestinal tract. Therefore, they are parenterally administered and distributed via the lymphatic system before being metabolized and broken down into small molecules. Biologics that target soluble TNF, such as Etanercept, have a linear elimination profile, whereas biologics that target membrane-bound antigen have a non-linear elimination profile (36). The non-linear elimination of the latter category of biologics is related to target-mediated drug disposal (TMDD). One important pharmacokinetic property of TNF inhibitors is half-life (37). Etanercept has the lowest half-life of 3–5.5 days, while all other inhibitors have half-lives of ~ two weeks (38). The frequency of administration of TNFi is associated with its half-life, wherein a short half-life warrants more frequent administration. Infliximab is injected intravenously, while all other TNF inhibitors are used both intravenously and subcutaneously (39). The pharmacokinetics of TNFi is influenced by several variables as detailed in Figure 2 .

The response to TNFi is influenced by pharmacokinetic variables. Males appear to have higher clearance of TNFi compared to females (40). In general, the sex-related pharmacokinetic differences are attributed to molecular and physiological differences between males and females. To our knowledge, the available evidence on the TNFi pharmacokinetic discrepancies between males and females is observational rather than investigating the mechanisms in effect (41, 42). One plausible factor for higher drug clearance in males is that females naturally have more body fat (43); the logic behind it is that the blood circulation in fat tissues is poor, thus having more fat composition contributes to variations in drug kinetics (44). Another possible reason is the difference in innate and adaptive immune responses between males and females; for instance, the peripheral blood mononuclear cells and neutrophils (45) in men were shown to produce more TNFα than in women after stimulation by lipopolysaccharide; hence, when TNFi is bound to its target, it is cleared faster. On the contrary to normal levels of body fat, excess body fat is correlated with lower serum levels of TNFi, whereby studies have shown that people with high body mass index (BMI) have lower trough levels of TNFi compared to individuals with normal weight (46). Evidence on the mechanism behind the inverse association of obesity and serum TNFi levels is scarce; however, some studies attributed it to the amplified inflammatory status induced by obesity and to a phenomenon known as TNF sink or antigen sink (47, 48). Briefly, in an inflammatory state, the anti-inflammatory drug, in this case, TNFi, can bind to all TNF in the blood, and the unbound TNFi will be sequestered by the saturated TNF via a pattern described as a sink; studies found that this phenomenon leads to rapid clearance of TNFi (48). Furthermore, low albumin levels (49) and high levels of inflammatory biomarkers, such as C-reactive protein (CRP) (50), are associated with higher clearance of TNFi. To our knowledge, the link between hypoalbuminemia and the reduced clearance of TNFi is not clearly elaborated. It is well-established that albumin is the most abundant serum protein and that low albumin levels are associated with severe inflammatory status. One of its key roles is binding drugs and other molecules in the blood; although it does not bind large molecules, such as TNFi, it was shown to have a protective effect against the degradation of drugs, including TNFi (49, 51). This could explain why low albumin level is linked to higher clearance of TNFi. On the other hand, high CRP levels reflect the presence of inflammation, and its production is usually triggered by enhanced TNFα production as is the case in systemic inflammatory diseases; therefore, a high CRP level is linked to high serum TNFα levels, which consequently lead to accelerated clearance of TNFi (50). Another well-established factor implicated in TNFi kinetics is the development of anti-drug antibodies (ADAs), whereby lower trough levels of TNFi were reported in presence of ADAs (52). ADAs can neutralize TNFα by binding to the pharmacologically active site of the TNFi, or they can bind to TNFi without neutralizing it; both cases lead to enhanced clearance of TNFi (53–55). On the contrary, some studies found that the concomitant intake of immunomodulators (56) as well as the pegylation of TNFi (57) are linked to reduced clearance and higher drug levels. The use of concomitant immunosuppressives could prevent ADAs production and consequently lead to prolonged presence of TNFi in serum and slower clearance (58, 59). More studies are required for conclusive evidence on some of these variables, such as the concomitant use of immunomodulators (60). In terms of PEGylation, the hydrophilic nature of PEG molecules improves the solubility of therapeutic proteins, such as TNFi, and promotes higher accumulation at target sites through enhanced permeability (61, 62). In fact, PEGylation can also protect therapeutic proteins, such as TNFi, from proteolytic degradation, thereby reducing their clearance (63). Clearance of TNFi has also been shown to vary according to differences in the neonatal Fc receptor (FcRn) (64) and Fcγ receptors (65). For instance, two alleles out of five in a variable number of tandem repeats in FcRn, showed lower levels of TNFi upon initiating the treatment, whereby patients with VNTR2/VNTR3 genotype had lower levels of infliximab or adalimumab in their serum compared to patients with VNTR3/VNTR3 genotype (64). The Fcγ receptors might accelerate clearance through an increased binding of the Fc portions of therapeutic mAbs to high affinity FcgRs inducing their elimination and resulting in a rapid decrease in serum drug levels (66). Subcutaneous administration of TNFi was also found associated with variabilities in absorption and, consequently, the trough levels of the drug (67). The primary limitations of subcutaneous administration include the restriction on the volume that can be delivered (typically no more than 1 mL) compared to the intravenous route, and the heterogeneity in dosage absorption among patients, ranging from 50% to 100%, which result in greater pharmacokinetic variability between individuals and doses (60). The subcutaneous administration also slows down the absorption of the drug due to longer processing of foreign bodies in the skin (60). One study reported more stable drug levels upon subcutaneous infliximab administration compared to intravenous infliximab (68), while another study showed that intra-articular injection of TNFi was more effective than subcutaneous TNFi administration (69).

Monoclonal TNFi antibodies interact with membrane bound TNFα, while the fusion proteins like Etanercept engage with free soluble TNF. These mechanisms prohibit TNFα from binding to its own receptors on the inflammatory cells. The interaction of TNFi with TNFα is known as TNFα neutralization, by which downstream inflammatory cytokines, such as IL-1 and IL-6, are subsequently suppressed, and the inflammatory processes subside (70). Although TNFi appear to have a selective and precise anti-inflammatory action, they can cause significant side effects such as infections and malignancies (70). This could be due to TNFα’s involvement in normal physiological processes, including tumoricidal activities and host defense. Because monoclonal antibodies can be directed against a wide range of soluble or membrane-bound targets, they can exert their pharmacological effects through several ways. Soluble TNFα exists as a circulating cytokine and is involved in systemic inflammation (71). mAbs such as infliximab, adalimumab, and golimumab primarily target soluble TNFα, neutralizing its pro-inflammatory effects by preventing its interaction with TNF receptors on immune cells. This leads to the inhibition of downstream signaling pathways, particularly the NF-κB and MAPK pathways, which are critical for the activation of immune responses, inflammation, and the recruitment of other immune cells to sites of infection or injury (72, 73). On the other hand, membrane-bound TNFα is expressed on the surface of certain immune cells, such as T-cells, macrophages, and dendritic cells (74). mAbs like etanercept, which acts as a soluble TNF receptor fusion protein, bind to both soluble and membrane bound TNFα (75). By blocking membrane-bound TNF, etanercept can modulate immune cell interactions differently than mAbs targeting only soluble TNFα. This may impact the activation of various immune responses, such as cell-mediated immunity, through the reverse signaling mechanism of membrane-bound TNF, which can influence the activation of both pro-inflammatory and anti-inflammatory pathways (29, 76). Targeting soluble TNFα can primarily dampen systemic inflammation, while blocking membrane-bound TNFα could alter immune cell functions more directly by affecting cell-cell communication and immune cell activation (77).

Moreover, the TNFi are administered parenterally and are absorbed into the bloodstream through mechanisms such as diffusion and facilitated transport at the injection site. Intravenous (IV) administration of TNFi offers 100% bioavailability since the drug is delivered directly into the bloodstream, leading to a quicker onset of action (78, 79) ( Figure 3 ). In contrast, subcutaneous (SC) administration typically results in slower absorption and lower bioavailability, but it allows for more gradual drug delivery over time (80). Studies comparing the two routes, such as those involving infliximab (IV) and adalimumab (SC), have shown comparable therapeutic efficacy, although IV formulations are more commonly associated with infusion-related reactions (IRRs) including fever, chills, and shortness of breath (81). SC administration, on the other hand, is generally linked to local injection site reactions, such as pain or redness (82), but carries a lower risk of IRRs. Additionally, while both delivery methods activate the immune system similarly, the faster delivery associated with IV administration may lead to a more immediate immune response, whereas SC administration offers a more gradual immune modulation (83).

Some rare reports have been published that describe the adverse reactions of TNFi treatment, including tuberculosis (TB), heart failure, skin cancer, lymphoma, and the occurrence of demyelinating diseases (65). Some recent studies reported that treatment with TNFi does not increase the chance of cancer in RA patients. A study reported that all TNFi are effective; however, Etanercept is better in terms of safety compared to other TNF blockers as minimal side effects were reported (84). Another study unraveled the effectiveness and safety of anti-TNFα therapy for Juvenile Idiopathic Arthritis and reported that Adalimumab has good efficacy and safety compared to Infliximab. In contrast, Infliximab has higher efficacy as compared to Etanercept. However, Golimumab and Certolizumab have lower efficacy than these three drugs (1). A higher risk of serious infections was associated with Infliximab, Certolizumab, and Adalimumab, whereas Etanercept contributes to a lower risk of discontinuation due to adverse reactions (85). A study involving a cohort of over 42 spondyloarthritis (SpA) patients found that anti-TNF therapy demonstrated acceptable safety and a positive response rate in SpA patients (86). The safety profiles of TNFi vary among ankylosing spondylitis and rheumatoid arthritis patients. A meta-analysis conducted on ankylosing spondylitis patients found that Adalimumab and Infliximab have good clinical outcomes with Infliximab being more effective compared to Adalimumab (87). There is limited evidence on golimumab, but one study observed low infection rate during treatment with golimumab in a patient that had RA (88).

About 30% of patients initially fail to respond to anti-TNFα therapy (primary non-responders) and up to 50% of patients lose their response during the course of treatment (secondary non-responders). The substantial proportions of patients experiencing failure of TNFi treatment whether upon initiation or at a later stage highlight the need for biomarkers that can predict response to these inhibitors. One possible clinical biomarker is serum calprotectin, which is also known as S100A8/A9 or MRP8/14 complex. A study conducted in three cohorts, one with Infliximab, one with Adalimumab, and one with Rituximab showed that TNFi responders had considerably greater baseline serum calprotectin levels than non-responders in rheumatoid arthritis patients. The study reported that the MRP8/14 (myeloid-related protein) levels were frequently higher in those who responded to targeted therapy, regardless of the mechanism of TNFi (89). Another plausible predictor could be the baseline TNFα levels; in a study conducted on 36 Crohn’s disease patients, it was found that primary non-response was related to higher TNF levels at baseline (56). Apart from clinical biomarkers, the links between genetic variants and response to TNFi were also assessed to identify genetic biomarkers. The genetic profiles of primary non-responders (PNR) have been investigated by previous studies by targeting the genes that are associated with cytokines and their receptors. For instance, associations were identified between non-response to TNFi and different genetic variants in TNFRI, IL13R2, VNTR2/VNTR3, IL23R, MAPK and FcYRIIIa (90). The secondary failure of response has been attributed in some cases to toxicity and immunogenicity. The risk of developing anti-drug antibodies is associated with several factors, including patient-related and treatment-related factors. The latter involves administration route, dosage, frequency, concomitant use of immunomodulators, and past TNFi treatment. While the individual-based factors that have been linked to immunogenicity include genetic susceptibility (91, 92), disease activity (93), BMI (56, 93, 94), and smoking status (56, 95),from a genetic perspective, the link between HLA markers in particular and immunogenicity is well-documented.

Adverse drug reactions (ADRs) are a major concern in healthcare. The presence of specific alleles in the human leukocyte antigen (HLA) system has been linked to the development of idiosyncratic ADRs. Understanding these connections is critical for prioritizing patient safety and customizing drug therapies. The HLA gene is the most extensively studied genetic biomarker for predicting therapeutic outcomes of rheumatoid arthritis (RA). The HLA-DRB1 gene, a disease susceptibility marker, accounts for 30-50% of the genetic risks in RA. A distinct sequence at positions 70-74, termed shared epitope, is associated with the onset and pathology of RA (96). Shared epitope positivity was found to be linked to progressive joint destruction, particularly when valine is present at position 11 of HLA-DRB1. In terms of TNFi efficacy, a study using a UK RA cohort found significant disease activity improvement with TNFi therapy when valine, lysine and alanine were at positions 11, 71, and 74 of HLA-DRB1, respectively (97). Moreover, the HLA-DQA1*05 allele has been recognized as a risk factor for the development of antibodies against anti-TNF agents. HLA-DQA1*05:01 and its extended haplotypes were associated with infliximab immunogenicity, but not adalimumab whereas HLA-DQA1*05:05 and its extended haplotypes were associated with immunogenicity to both adalimumab and infliximab, with a stronger effect for adalimumab (98). Similarly, the HLA-DRB9 g.32465390 G>T (rs2395185) G variant was also associated with IBD risk and primary non-response to Infliximab in a subset of patients with an age at diagnosis under 21 years (99). Moreover, other HLA variants such as HLA-DQA1 g.32622994 T>A/T>C (rs2097432) and HLA-DRB1 g.32465390 G>T (rs2395185) were associated with long-term anti-TNF drug response in children with IBD, highlighting their potential as predictive biomarkers for therapy outcomes (100). Previous studies recommend testing HLA-DRB1*3, HLA-DQA1*05:01, and HLA-DQA1*05:05 to predict immunogenicity to anti-TNF drugs; prior genetic testing for these variants could aid in therapy selection and preventive strategies (101). Moreover, another study reported that HLA-B rs41563412-GCA (p.Ser33LeufsTer9) homozygous carriers did not respond to adalimumab whereas HLA-DRB1 rs1071752-C (p.Gln125Gln) heterozygous and homozygous carriers showed response to infliximab (102). These two variants are situated in the exonic and intronic regions of the HLA-DRB1 and HLA-B genes, respectively; both are an integral part of the adaptive immune system. These genes encode pivotal proteins for presenting antigens to CD8+ (in MHC-1) and CD4+ T-cells (MHC-II) in a dependent manner, shaping the altered adaptive immune response observed in Crohn’s disease (CD) patients. While genetic factors influencing clinical CD phenotypes may diverge from those related to anti-TNF response, the HLA region stands as a prominent risk locus for CD. Numerous genetic association studies have underscored the significant association between variants in HLA-B and HLA-DRB1 with the development of CD. In contrast to CD4+ T-cells, antigen presentation to CD8+ T-cells follows MHC-I restrictions. Specifically, the HLA-B g. 31356966 (rs41563412) variant, located in the intronic region of the HLA-B gene, aligns with this MHC-I family. Notably, the sole homozygous carrier of the pathogenic HLA-B rs41563412-GCA variant, featuring a frameshift mutation (p. Ser33LeufsTer9), exhibited unresponsiveness to anti-TNF therapy. This suggests the potential involvement of this loss-of-function variant in modulating the CD8+ cytotoxic immune response, emphasizing its role in therapeutic outcomes (102).

Pharmacogenomics studies principally focusing on discovering variants in the HLA region and their associations with TNFi response, are reviewed in Table 1 (98, 100, 103–110, 115). The assessment of genetic profiles of autoimmune patients is essential for identifying such HLA variants to enhance clinical diagnosis and implement personalized treatment.

Several clinical and genome-wide association studies investigated the link between genetic variants and response to TNFi across various autoimmune diseases ( Table 2 ). A meta-analysis, focused on spondyloarthropathy, psoriasis and Crohn’s disease, identified that the G allele in TNF -238G>A (rs361525), TNF -308G>A (rs1800629) and the C allele in TNF -857C>T(rs1799724) showed better response to TNFα blockers in the European population, but failed to show a significant impact in the Asian population (119, 144). Another meta-analysis conducted on individuals of European population assessed the association between SNPs and response to anti-TNF-α therapy across major autoimmune diseases, including psoriasis, rheumatoid arthritis, inflammatory bowel disease, and spondyloarthritis. They identified six SNPs in the FCGR2A (rs1801274), FCGR3A (rs396991), TNF (rs361525, rs1800629, rs1799724), and TNFRSF1B (rs1061622) genes that were significantly associated with treatment response, primarily within disease-specific subgroups. However, no single pharmacogenetic marker was consistently predictive across all conditions, highlighting the need for further research to identify robust biomarkers for guiding anti-TNF therapy (120).

In RA, a study conducted on 1752 patients treated with Infliximab in the UK found that the FTO g.54026293G>A (rs7195994) was related to Infliximab response (116). Another study examined variants in NKG2D in 280 RA patients undergoing anti-TNF therapy. NKG2D g.10379727A>G(rs2255336) and NKG2D g.10372766C>G (rs1049174) variants showed significant associations of CC or GG genotypes to poor response (p=0.003, p=0.004 respectively). Moreover, the GG genotype was associated with non-response in etanercept-treated patients after 12 weeks (124). The genetic link to the response to TNFi has also been investigated in IBD patients. A study of an Italian cohort reported that IBD patients possessing a variant in FCGR3A exhibited diminished clinical response by the end of the induction period. Moreover, they observed a remarkable correlation between the FCGR3A variant and median Infliximab levels during maintenance therapy, revealing that patients with the wild-type genotype demonstrated elevated Infliximab levels compared to those with the variant allele. Additionally, individuals with the variant allele displayed an increased likelihood of developing antidrug antibodies (145). Another longitudinal study conducted on 132 Crohn’s disease patients identified that SNPs including TLR2 g.153688371T>C(rs1816702) and TLR2 c.597T>C (rs3804099) were linked with long-term response to Infliximab, whereas IL6 c.6331T>C (rs10499563) correlated with supratherapeutic and infratherapeutic Infliximab levels (117). They also identified that genotypes AG and GG in TNFRSF1B has been associated with lower response to adalimumab and infliximab. In pediatric Crohn’s disease patients treated with Infliximab, studies reported that genotypes TNFRSF1B CC>CT (rs3397) was associated with the anti-Infliximab antibody production, suggesting its potential in identifying patients prone to immunogenicity (146). More variants were recently associated with long-term response to infliximab or adalimumab in IBD pediatric patients, including rs10508884 (CXCL12), rs2241880 (ATG16L1), and rs6100556 (PHACTR3); these three SNPs were linked to poor long-term response to TNFi (92). Certain other variants were associated with disease type or drug type; for instance, rs6908425 (CDKAL1) was associated with poor response in CD, while rs2188962 (IRF1-AS1), rs2241880 and rs6100556, were associated with worse outcomes in ulcerative colitis (UC). Moreover, rs2241880 was linked to poor response to both infliximab and adalimumab, whereas rs6100556 correlated with poor response to infliximab specifically (92). Furthermore, in a genetic association study cohort comprised of 474 IBD patients with European ancestry, two loci were found to be significantly associated with response to TNFi and were replicated with a p <1 x 10–⁰³ in a validation cohort: rs116724455 in TNFSF4/18 and rs2228416 in PLIN2. Allele C of rs116724455, located on chromosome 1 near the TNFSF4 and TNFSF18 genes, was strongly associated with nonresponse to TNFi (OR = 19.9, p = 4.79×10−⁸). Likewise, allele T of rs2228416 on chromosome 9 near the PLIN2 and HAUS6 genes, was also linked to nonresponse (OR = 5.25, p = 5.24×10−⁶) (147). Four other SNPs showed suggestive association with the response to TNFi; those include rs762787 (allele T) near LTF, CCR5, and CCRL2, rs9572250 (allele G) near KLHL1, rs144256942 (allele G) near PROX1 and RPS6KC1 and rs523781 (allele G) near RORB and TRPM6 (147). In a very recent study, certain genetic variants were associated with the likelihood of achieving steroid-free remission (SFR) in a small cohort of European IBD patients (83 patients) treated with infliximab or adalimumab (148). Patients carrying the GG genotype of rs1800629 (p = 0.025) and the AA genotype of rs1061624 (p = 0.029) had significantly higher odds of achieving SFR compared to those who did not. In contrast, patients with the AA genotype of rs361525 or the CC genotype of rs767455 were significantly less likely to achieve SFR. Moreover, the allele A of rs1800629 was more frequent in patients who discontinued TNFi treatment and was linked to a higher risk of early therapy interruption (148).

Similar to RA and IBD patients, studies have identified genetic variants linked to variable response to TNFi in psoriasis patients; one example is the TNFRSF1B c.196T>G (rs1061622) SNP, which was associated with a higher risk of poor response to TNFi in psoriasis patients of European ancestry (139). Remarkably, the G allele of rs1061622 was found more frequent in patients carrying the allele HLA-CW6*0602 (149). In a more recent study, a candidate variant, rs1991820, was identified via a retrospective study approach involving 1849 psoriasis patients (149). Although it did not reach genome-wide significance, rs1991820 (an intronic variant of KLK7 gene) showed a promising association with positive response to TNFi (p = 1.30 × 10^–6). The study found that rs1991820 is significantly correlated with elevated expression of KLK7 in the skin tissue of psoriatic patients and that TNFi treatment gradually reduced KLK7 expression levels (149). In a study involving 312 ankylosing spondylitis patients treated with Etanercept, patients with specific genotypes (CYP2C9*3, CYP2D6*10, and CYP3A5*3) showed lower joint swelling, erythrocyte sedimentation rate, and C-reactive protein levels after 24 weeks of Etanercept treatment. The CYP2D6*10/*10 and CYP3A5*3/*3 diplotypes were associated with higher efficacy scores, suggesting a potential correlation between genetic variations and Etanercept treatment response in these patients (142). In one study of 74 Italian Behçet syndrome patients receiving anti-TNFα therapy, the TNF -308G>A (rs1800629) variant was found to be associated with treatment response; GG genotype prevalence was significantly higher among responders (86.2%) compared to non-responders (56.3%) (150). Well-characterized cohorts with larger samples size are needed for further validation of TNF -308G>A (rs1800629) as a predictive biomarker for TNFi response. A study on an Italian cohort found a significant association between the TNF -308G>A (rs1800629) and clinical remission without steroids in pediatric patients receiving infliximab therapy. Additionally, the study identified a potential link between carriers of the HLA-DQA1*05 and a higher risk of developing anti-TNFα immunogenicity (151). A previous study conducted on Psoriasis patients from the European population reported that variants in the TNF -857C and TNFRSF1B 676T genes were associated with positive response when treated with Etanercept. Moreover, no significant associations were found between variants in these genes and treatment outcomes for infliximab or adalimumab (121). Another study conducted on European population reported that variants in IL1B (rs1143623, rs1143627), LY96 (rs11465996), TLR2 (rs11938228, rs4696480) and TLR9 (rs352139) were associated with response to Adalimumab, Infliximab and Etanercept (122). Furthermore, a recent study conducted on 738 European patients with IBD using a candidate gene approach identified 19 functional polymorphisms in genes involved in NFκB activation via TLR pathways, TNF-α signaling, and NFκB-regulated cytokines as significant predictors of response to anti-TNF therapy. The study highlighted that a genetically strong TNF-mediated inflammatory response was associated with favorable treatment outcomes, while also suggesting alternative cytokine targets such as IL-1β, IL-6, and IFN-γ for non-responders (136). These findings suggest that pharmacogenetic monitoring could become an essential tool in tailoring anti-TNF therapies for both pediatric and adult populations.

Autoimmune disease treatment poses significant clinical and economic challenges, including decisions on when to start targeted therapy and which drug class to select. The decision of prescribing TNFi therapies is often driven by general guidelines (described below), pricing incentives, and habitual prescribing patterns (152). One study reported that 98.8% of participating rheumatologists (n= 248) expressed interest in a predictive test for inadequate TNFi response, noting that such a tool would influence their treatment decisions and patient management. The study also highlighted that 71% of rheumatologists were concerned that inadequate response to TNFi therapies result in patients paying for ineffective treatments. The current prescribing practices entail patients switching between multiple TNFi drugs before switching to drugs with different mechanisms of action (152, 153). This approach raises healthcare costs, extends high disease activity, and negatively impacts quality of life. According to the guidelines of the American College of Rheumatology, the physician’s decision to initiate or switch biologic DMARDs should consider a patient’s disease activity, prior DMARD use, and comorbidities (154). Studies investigating the predictors of biologic treatment in RA patients have shown variable findings (130). Some findings suggest that biologic initiation is influenced by previous glucocorticoid or non-biologic DMARD use (155). Other studies highlight the association between sociodemographic factors, such as lower income and older age, and accessibility to biologics. The differences in study design, population, and disease stage have led to inconsistent conclusions. While most research efforts in this area focused on initiating biologics, evidence on switching between biologics is scarce. When the optimal treatment is delayed, patients are subjected to a poor quality of life and several risks, including adverse events, disease progression, and chronic pain. The clinical applications of pharmacogenomics of TNFi entail personalized drug prescription and dosage based on the genetic profile of the patient, prediction of response prior to treatment initiation, and prediction of the risk of adverse events and immunogenicity before they occur, aiming to achieve personalized treatment and better health outcomes.

As discussed in the previous section, several limitations need to be addressed before achieving a valid personalized approach for patients on TNFi. Concisely, the main factors to be addressed are reliability, standardization, and sufficient statistical power to discern authentic and robust associations. Investigating a sole aspect is not enough to explain the variability in response to TNFi. Therefore, future study designs must develop models that integrate genetic, clinical, demographic, and environmental data to capture the intricate heterogeneity among patients and achieve robust accuracy, specificity, and sensitivity values. This comes hand-in-hand with increasing sample size and studying diverse populations. As a matter of fact, in 2020 the pharmacogenomics GWAS represented around 7% only of all published GWAS, and the majority of these studies predominantly focused on European populations (181). The GWAS on response to TNFi is even more scarce; hence, large-scale GWAS of response to TNFi are needed in diverse and less represented populations. Moreover, patient stratification is another area that needs improvement, whereby previous studies primarily aimed to distinguish between responders and non-responders whilst grouping patients of various response, such as optimal response and moderate response, in one category under “good responders”. This might have been due to limited sample size, which reiterates the necessity of larger cohorts to achieve significant results. Similarly, it is equally essential to predict primary or secondary non-response to TNFi. As a matter of fact, secondary non-response results in many cases from immunogenicity-related causes, and models that predict immunogenicity are still lacking. Hence, patients who will develop a secondary failure of TNFi might be faultily categorized as responders. Designing models that can predict immunogenicity and secondary non-response would be helpful for the physicians in prescribing a less immunogenic TNFi or other complementary treatments that can prevent the probability of developing antidrug antibodies. In fact, the temporal change in response to TNFi necessitates longitudinal study designs to assess the response to treatment, particularly because a significant proportion of patients tend to lose their response to TNFi almost a year after the initiation of the treatment. Whilst most studies would evaluate patients’ response within 3 to 6 months, more studies focusing on the long-term response are required to capture the variability in response spectrum among patients and achieve a tailored treatment plan. Furthermore, from a precision medicine approach, future investigation of the mechanisms behind the failure of response to TNFi is also crucial, especially in the absence of cure for autoimmune diseases, whereby non-responders to TNFi suffer from a poor quality of life.

Emerging technologies such as machine learning (ML), single-cell genomics, and multi-omics, hold great promise in revolutionizing the prediction of TNFi response in patients. Previous studies have employed machine learning models to predict TNF inhibitor use in rheumatologic conditions (165–167). These studies typically analyze baseline clinical and laboratory data to identify patients likely to benefit from early TNF inhibitor therapy. ML approaches such as artificial neural networks (ANN) (165), have demonstrated superior predictive performance compared to traditional methods like logistic regression and support vector machines. ML with its ability to process large, complex datasets, can uncover non-obvious patterns within genetic, clinical, and environmental data (182). By training models on patient-specific characteristics, ML algorithms can predict who is more likely to respond to TNFi treatment, allowing for more personalized and effective therapeutic strategies (166).

Moreover, single-cell genomics enables the study of individual cells within heterogeneous tissue samples, offering detailed insights into immune cell dynamics at the molecular level. This technology is especially valuable in identifying rare or previously undetected cell populations that may play critical roles in drug response or resistance (183). By isolating and analyzing immune cells from patients on TNFi therapy, single-cell genomics can pinpoint specific immune pathways or cell types contributing to treatment efficacy or adverse effects, ultimately guiding better-targeted therapies. Furthermore, multi-omics approaches, which integrate data from genomics, transcriptomics, proteomics, and metabolomics (184), provide a comprehensive view of the biological processes that underlie drug response. These approaches can help identify novel biomarkers that predict treatment outcomes and uncover complex interactions between genes, proteins, and metabolites that influence how patients respond to TNFi. By combining data across multiple biological layers, multi-omics can reveal the full complexity of TNFi treatment responses, helping to tailor therapies to individual patients based on their unique molecular profiles. Together, these technologies represent the future of TNFi response prediction, offering the potential for more accurate, personalized treatment plans that improve patient outcomes while minimizing adverse effects. Moreover, the successful integration of TNFi response prediction into clinical practice requires collaboration between regulatory agencies, clinicians, and patients. Regulatory agencies like the FDA and EMA must validate predictive biomarkers and diagnostic tools, ensuring they meet safety and efficacy standards. Clinicians need to incorporate these tools into daily practice, interpreting complex data to make informed decisions tailored to individual patients. Patients must be educated on how genetic testing, and predictive models can help in tailoring a personalized treatment plan and adhering to it is part of the therapy; this is important to ensure informed consent and active participation in the decision-making process. Ultimately, close collaboration among these stakeholders will ensure the effective and safe implementation of personalized TNFi therapies.

Although TNFα inhibitors have improved the quality of life of many patients with autoimmune diseases, a significant proportion of patients remain in a progressive disease state and poor quality of life due to the failure of anti-TNFα therapy. Previous evidence demonstrated that the variability in response to TNFi is associated with genetic variants. Therefore, integrating a pharmacogenomic therapeutic approach seems to be promising for personalized treatment plans. As scientific research advances in this area, identifying robust genetic biomarkers to predict the response and the risk of adverse reactions to TNFi would ultimately improve the quality of care for patients with autoimmune diseases.