Association between TNF-α polymorphisms and responsiveness to TNF-α blockers in ankylosing spondylitis and psoriatic arthritis: a meta-analysis

Abstract

To investigate the association between tumor necrosis factor-alpha (TNF-α) polymorphisms and the responsiveness to anti-TNF-α therapy in patients with ankylosing spondylitis (AS) and psoriatic arthritis (PsA). A comprehensive literature search of the PubMed/Medline, Embase, and Web of Science databases was performed to identify relevant published studies. Meta-analysis was performed to assess the relationship between specific TNF-α polymorphisms (-308 A/G, + 489 A/G, -238 A/G, -857 C/T, or -1031 C/T) and the responsiveness of patients with AS or PsA to anti-TNF-α therapy. The study was registered in PROSPERO (CRD42023472655). The analysis incorporated data from 11 comparison studies within 9 articles, involving 611 patients (453 responders and 158 non-responders). Meta-analysis revealed a significant association between the TNF-α -308 G allele and a positive response to TNF-α blockers (odds ratio [OR] 4.221 [95% confidence interval (CI) 1.691–10.54]; p = 0.002). Stratification according to ethnicity demonstrated this association in both the European and Asian populations. Disease-specific meta-analyses indicated an association between the TNF-α -308 G allele and a favorable response to TNF-α blockers in those with AS and PsA. However, the TNF-α + 489 GG genotype did not exhibit a consistent association with the response in PsA, although a single study suggested an association in AS. Furthermore, TNF-α -857 C and − 238 G alleles were associated with a positive response to TNF-α blockers in PsA. No association was found between the TNF-α -1037 TT genotype and response in PsA. Results of this meta-analysis provide evidence supporting a significant association between the TNF-α -308 G allele and an increased responsiveness to TNF-α blockers in AS and PsA. It also suggests that TNF-α -857 C and − 238 G alleles may influence TNF-α blocker responsiveness in PsA.

Introduction

Tumor necrosis factor-alpha (TNF-α) is a critical cytokine involved in inflammatory and immune responses^1,2. It has become an important target for the treatment of autoimmune and inflammatory diseases including ankylosing spondylitis (AS) and psoriatic arthritis (PsA)³. AS and PsA are both chronic inflammatory conditions that affect the musculoskeletal system, leading to pain, stiffness, and reduced quality of life in those affected^4–6. In these disorders, TNF-α plays a central role in promoting inflammation, making it an ideal target for therapeutic intervention(s)⁷.

The effectiveness of anti-TNF-α therapies, such as biologics, has been well established in the management of AS and PsA⁴. However, it has been observed that treatment responses can vary significantly among patients⁸. This variability has led researchers to explore the influence of genetic factors, specifically TNF-α polymorphisms, on treatment outcomes. These genetic variations, particularly those located in the promoter region of the TNF-α gene, are of great interest in the context of predicting individual responses to anti-TNF-α therapies.

The complex interplay between genetics, inflammation, and therapeutic response has led to a growing body of research investigating the association between TNF-α polymorphisms and the efficacy of anti-TNF-α therapies. This line of inquiry has provided valuable insights into the potential of genetic markers for predicting an individual’s response to these treatments. The − 308 A/G polymorphism has been of particular interest, with studies suggesting its potential as a genetic marker for treatment responsiveness. In the realm of genetic research, polymorphisms at specific positions within the TNF-α gene, such as -308, + 489, -238, -857, and − 1031, have been the subject of rigorous investigation^9–17. The central question is whether these polymorphisms have the potential to influence patient responsiveness to anti-TNF-α therapy. Variants in the promoter region of the TNF-α gene (e.g., -308 A/G, -238 A/G, -857 C/T, -1031 C/T) can influence transcriptional activity and cytokine production. For instance, the − 308 A allele has been associated with higher TNF-α expression, which may contribute to heightened inflammatory responses and altered sensitivity to TNF-α inhibitors^18,19. Similarly, the − 238 and − 857 variants have been reported to modulate transcription factor binding, potentially affecting TNF-α gene regulation¹⁹. Although not located in the promoter region, this variant may influence mRNA stability or splicing, thereby affecting TNF-α expression and downstream inflammatory pathways¹⁸. These polymorphisms were chosen because they are among the most frequently studied TNF-α variants in autoimmune and inflammatory diseases, with prior evidence suggesting their potential role in modulating treatment response^18,19. Understanding this association can be a key step toward personalized treatment strategies for patients with AS and PsA.

Findings from these investigations are providing valuable insights into the potential of genetic markers to predict and improve the efficacy of anti-TNF-α therapies. Understanding the nuanced variations across different polymorphisms, diverse patient populations, and various clinical conditions is essential in this context^20–22. The present study aimed to explore the association between TNF-α -308 A/G, + 489 A/G, -238 A/G, -857 C/T, or -1031 C/T polymorphisms and the responsiveness to anti-TNF-α therapy in patients diagnosed with AS and PsA.

Materials and methods

Assessing study quality and publication bias

Study quality was scored using the Newcastle-Ottawa Scale (NOS)²³. The full score is 9 stars, and scores from 6 to 9 stars indicate high methodological quality. Although funnel plots are often used to detect publication bias, funnel plotting requires diverse study types with varying sample sizes, and interpretation of the plots involves subjective judgment. Considering this, publication bias was evaluated using Egger’s linear regression test²⁴, which measures funnel plot asymmetry using a natural logarithmic scale of odds ratios (ORs).

Identification of eligible studies and data extraction

A comprehensive literature search was conducted across PubMed/Medline, Embase, and Web of Science to identify relevant studies published up to October 2023. The search strategy combined controlled vocabulary (e.g., MeSH terms) and free-text keywords, including “tumor necrosis factor,” “TNF-α,” “polymorphism,” “TNF blocker,” “etanercept,” “infliximab,” “adalimumab,” “golimumab,” “certolizumab,” “ankylosing spondylitis,” and “psoriatic arthritis.” Boolean operators (“AND,” “OR”) were applied to maximize sensitivity, and the reference lists of retrieved articles were manually screened to capture additional eligible studies. No language or geographic restrictions were imposed.

Inclusion criteria were as follows: (1) Original studies (case-control, cohort, or clinical trial) reporting on the association between TNF-α polymorphisms (− 308 A/G, + 489 A/G, − 238 A/G, − 857 C/T, − 1031 C/T) and response to anti-TNF therapy. (2) Patients diagnosed with AS or PsA according to established classification criteria. (3) Availability of sufficient genotype or allele frequency data to calculate ORs with 95% confidence intervals (CIs). (4) Independent datasets, ensuring no overlap in patient populations. Exclusion criteria included: (1) Reviews, editorials, case reports, or conference abstracts without full data. (2) Studies lacking extractable genotype–response data. (3) Duplicate publications or overlapping cohorts (in such cases, the most comprehensive dataset was retained). (4) Studies investigating polymorphisms outside the predefined loci of interest. Data extraction was performed independently by two investigators using a standardized form. Extracted information included: first author, year of publication, country of study, disease type (AS or PsA), sample size, type of TNF inhibitor used, follow-up duration, response criteria applied, and genotype/allele distributions of the TNF-α polymorphisms in responders and non-responders. Any discrepancies were resolved by discussion and consensus.

Evaluation of statistical associations

Overall estimates of the contrast between the common allele and the minor allele (i.e., allelic effect) and the recessive model of the TNF-α polymorphisms regarding responsiveness to TNF inhibitors were calculated. The point estimates of risk, ORs, and corresponding 95% confidence intervals (CIs) were computed for each study. Cochran’s Q test was used to assess both within- and between-study variation and heterogeneity. This test evaluated the null hypothesis that all studies investigated the same effect. The effect of heterogeneity was quantified using the I² statistic, which ranges from 0 to 100%, and estimated the proportion of total variability in the point estimates attributed to heterogeneity rather than chance^25,26. I² values of 25, 50, and 75% were categorized as low, moderate, and high estimates, respectively^27,28. Two models were considered: the fixed effects model, assuming a consistent genetic effect across all studies, and the random effects model, accounting for substantial diversity among studies^29,30. Statistical analyses were performed using Comprehensive Meta-Analysis Program (BioStat, Englewood, NJ, USA).

Assessing study quality and publication bias

Study quality was assessed using the NOS²³. A full score of nine stars indicated high methodological quality within a score range of 6 to 9 stars. Publication bias was detected using Egger’s linear regression test, which measures funnel plot asymmetry using a natural logarithmic scale of ORs. This approach was chosen due to the limitations of funnel plots, which require diverse study types with varying sample sizes and involve subjective interpretation²⁴.

Results

Inclusion of studies in the meta-analysis

A total of 416 studies were retrieved in the electronic and manual literature searches. After careful review of the title and abstract details, 12 studies were selected for full-text examination (Fig. 1). Three studies were excluded for duplicate data³¹, insufficient data for analysis³², or data related to other polymorphisms³³. Consequently, nine studies fulfilled the predefined inclusion criteria^9–17 (Fig. 1). Notably, one of these eligible studies contained data regarding three different groups¹⁴, which were treated as separate entities. Thus, 11 distinct comparative studies were included in the meta-analysis. These studies had varying sample sizes, ranging from 16 to 99 participants per study, resulting in a combined dataset of 611 patients. This dataset encompassed 453 responders and 158 non-responders, with six studies focused on AS and five on PsA. Detailed characteristics of the selected studies are summarized in Table 1.

Fig. 1.

Flow-diagram illustrating the study selection process.

Table 1.

Characteristics of studies included in the meta-analysis.

Study	Country	Ethnicity	Disease	TNF inhibitor	Numbers		Response criteria	Studied polymorphism	Follow-up period
					R	NR
Liang, 20239	China	Asian	AS	Adalimumab	54	26	ASAS	TNF-α -308 A/G, -238 A/G, -857 C/T, + 489 A/G, -1031 C/T	24 weeks
Wang, 202210	China	Asian	AS	E, I, A, G, C	82	5	ASDAS	TNF-α -308 A/G	3 months
Ma, 201711	China	Asian	AS	rhTNFR-Fc	29	17	ASAS	TNF-α -308 A/G	12 weeks
Simone, 201512	Italy	European	PsA	Etanercept	59	38	PASI	TNF-α -308 A/G, -238 A/G, -857 C/T	3 months
Manolova, 201413	Bulgaria	European	AS	E, A	6	11	ASAS	TNF-α -308 A/G	6 months
Murdaca-1, 201414	Italy	European	PSA	Etanercept	26	11	DAS28	TNF-α -308 A/G, -238 A/G	6 months
Murdaca-2, 201414	Italy	European	PSA	Infliximab	10	20	DAS28	TNF-α -308 A/G, -238 A/G	6 months
Murdaca-3, 201414	Italy	European	PSA	Adalimumab	13	3	DAS28	TNF-α -308 A/G, -238 A/G	6 months
Tong, 201215	China	Asian	AS	I, rhTNFR-Fc	91	8	ASAS	TNF-α -308 A/G, -238 A/G, -857 T/C	6 months
Vasilopoulos, 201216	Greece	European	PS	E, I, A	63	17	PASI	TNF-α -857 C/T	20 weeks
Seitz, 200717	Switzerland	European	AS	E, I, A	20	2	BASDAI	TNF-α -308 A/G	24 weeks

AS: ankylosing spondylitis, PsA: psoriatic arthritis, I: infliximab, E: etanercept, A: adalimumab, G: Golimumab, C: Certolizumab, rhTNFR-Fc: TNF-α receptor II–IgG Fc fusion protein, PASI: Psoriasis Area and Severity Index, DAS28: Disease activity score 28, HBI: Harvey-Bradsaw Index, ASAS: Assessment of SpondyloArthritis International Society criteria, BASDAI: Bath Ankylosing Spondylitis Disease Activity Index.

Association between TNF-α -308 A/G, + 489 A/G polymorphisms, and responsiveness to TNF-α blockers

Meta-analysis revealed a significant association between the TNF-α -308 G allele and a positive response to TNF-α blockers (OR 4.221 [95% CI 1.691–10.54]; p = 0.002) (Table 2). When stratified according to ethnicity, this association was observed in both European and Asian populations (Table 2; Fig. 2). Furthermore, disease-specific meta-analyses indicated that the TNF-α -308 G allele was associated with a favorable response to TNF-α blockers in AS and PsA (OR 10.89 [95% CI 3.585–33.05]; p < 0.001; OR 2.451 [95% CI 1.324–4.535]; p = 0.004, respectively) (Table 2). However, the analysis found no association between the TNF-α + 489 GG genotype and the response to TNF-α blockers in PsA. A single study suggested an association between the TNF-α + 489 GG genotype and the response to TNF-α blockers in AS (Table 2). Given that multiple polymorphisms and subgroup analyses were evaluated, we acknowledge the potential for type I error due to multiple testing. The − 308 A/G polymorphism was designated as the primary variant of interest based on its established functional role and prior evidence, and the observed association remained statistically robust (p = 0.002). Post hoc power analysis indicated that the overall association between the TNF-α -308 G allele and treatment response was detected with approximately 87% power.

Table 2.

Meta-analysis of TNF-α -308 A/G, + 489 A/G polymorphisms with responsiveness to TNF-inhibitors in patients with AS and PsA.

TNF polymorphism	Population type	No. of studies	Total sample size		Test of association			Test of heterogeneity
			R	NR	OR	95% CI	P-val	Model	P-val	I2 (%)
TNF-α -308 A/G polymorphism GG vs. GA + CC (Recessive)	Overall	8	302	86	2.660	0.762–9.282	0.125	R	0.012	61.1
	European	5	75	47	1.629	0.618–4.294	0.324	F	0.106	47.6
	Asian	3	227	39	2.035	0.498–8.319	0.323	R	0.006	80.7
	AS	5	282	69	3.504	0.475–25.85	0.219	R	0.008	70.9
	PsA	3	108	72	1.658	0.533–5.158	0.383	F	0.122	52.5
G vs. A	Overall	8	896	360	4.221	1.691–10.54	0.002	R	0.043	51.7
	European	5	374	224	2.690	1.487–4.867	0.001	F	0.205	32.4
	Asian	3	522	136	1.60	3.215–41.86	< 0.001	F	0.105	55.5
	AS	4	680	216	10.89	3.585–33.05	< 0.001	F	0.201	33.8
	PsA	4	2160	144	2.451	1.324–4.535	0.004	F	0.196	36.0
TNF-α + 489 A/G polymorphism GG vs. GA + CC (Recessive)	Overall	4	103	60	0.884	0.280–2.791	0.833	R	00081	55.5
	European	3	49	34	1.723	0.639–4.648	0.282	F	0.872	0
	Asian	1	54	26	0.282	0.106–0.751	0.011	NA	NA	NA
	AS	1	54	26	0.282	0.106–0.751	0.011	NA	NA	NA
	PsA	3	49	34	1.723	0.639–4.648	0.282	F	0.872	0
G vs. A	Overall	3	98	68	1.961	0.896–4.291	0.092	F	0.208	36.3
	European	3	98	68	1.961	0.896–4.291	0.092	F	0.208	36.3
	Asian	NA	NA	NA	NA	NA	NA	NA	NA	NA
	AS	NA	NA	NA	NA	NA	NA	NA	NA	NA
	PsA	3	98	68	1.961	0.896–4.291	0.092	F	0.208	36.3

AS: ankylosing spondylitis, PsA: psoriatic arthritis, R: random effects model, F: fixed effects model, R: responder, NR: non-responder, OR: odds ratio, CI: confidence interval, P-val: P-value, I²: between-study variability attributable to heterogeneity, NA: not available.

Fig. 2.

Odds ratios and 95% confidence intervals from individual studies and combined data, assessing the relationship between the G allele of tumor necrosis factor (TNF)-α-308 A/G polymorphisms and responsiveness to TNF-inhibitors in Europeans and Asians (A) and individuals with ankylosing spondylitis (AS) and psoriatic arthritis (PsA) (B).

Association between TNF-α -857 C/T, -238 A/G, -1031 C/T polymorphisms and responsiveness to TNF-α blockers

Meta-analysis revealed that the TNF-α -857 C and − 238 G alleles were associated with a positive response to TNF-α blockers in PsA (OR 2.238 [95% CI 1.319–3.798]; p = 0.003; OR 4.237 [95% CI 1.538–11.67]; p = 0.005) (Table 3). However, no association was observed between the TNF-α -1037 TT genotype and the response to TNF-α blockers in PsA (Table 3). Associations identified for other polymorphisms (e.g., -857 C/T and − 238 A/G) should be considered exploratory, as their significance may not persist under more stringent correction methods. Subgroup analyses for PsA demonstrated power estimates of ~ 85% for the − 857 C allele and ~ 80% for the − 238 G allele, supporting the adequacy of the sample size for these associations.”

Table 3.

Meta-analysis of TNF-α -857 C/T, -238 A/G, -1031 C/T polymorphisms with responsiveness to TNF-inhibitors in patients with AS and PsA.

TNF polymorphism	Population type	No. of studies	Total sample size		Test of association			Test of heterogeneity
			R	NR	OR	95% CI	P-val	Model	P-val	I2 (%)
TNF-α -857 C/T polymorphism CC vs. CT + TT (Recessive)	Overall	3	208	51	1.703	0.375–7.744	0.491	R	0.019	74.9
	European	1	63	17	3.417	1.113–10.49	0.032	NA	NA	NA
	Asian	2	145	34	1.181	0.145–9.650	0.877	R	0.068	70.0
	AS	2	145	34	1.181	0.145–9.650	0.877	R	0.068	70.0
	PsA	1	63	17	3.417	1.113–10.49	0.032	NA	NA	NA
C vs. T	Overall	3	426	126	2.307	1.438–3.702	0.001	F	0.853	0
	European	2	244	110	2.238	1.319–3.798	0.003	F	0.614	0
	Asian	1	182	16	2.606	0.907–7.484	0.075	NA	NA	NA
	AS	1	182	16	2.606	0.907–7.484	0.075	NA	NA	NA
	PsA	2	244	110	2.238	1.319–3.798	0.003	F	0.614	0
TNF-α -238 A/G polymorphism GG vs. GA + CC (Recessive)	Overall	2	117	19	1.660	0.255–10.79	0.596	F	0.555	0
	European	1	26	11	0.739	0.028–19.55	0.856	NA	NA	NA
	Asian	1	91	8	2.457	0.281–24.05	0.440	NA	NA	NA
	AS	1	91	8	2.457	0.281–24.05	0.440	NA	NA	NA
	PsA	1	26	11	0.739	0.028–19.55	0.856	NA	NA	NA
G vs. A	Overall	3	352	114	3.827	1.523–9.616	0.004	F	0.493	0
	European	2	170	98	4.237	1.538–11.67	0.005	F	0.275	16.1
	Asian	1	182	16	2.360	0.259–21.53	0.447	NA	NA	NA
	AS	1	182	16	2.360	0.259–21.53	0.447	NA	NA	NA
	PsA	2	170	98	4.237	1.538–11.67	0.005	F	0.275	16.1
TNF-α -1031 C/T polymorphism CC vs. CT + TT (Recessive)	Overall	2	145	34	1.139	0.235–5.524	0.872	R	0.084	66.5
	European	NA	NA	NA	NA	NA	NA	NA	NA	NA
	Asian	2	145	34	1.139	0.235–5.524	0.872	R	0.084	66.5
	AS	NA	NA	NA	NA	NA	NA	NA	NA	NA
	PsA	2	145	34	1.139	0.235–5.524	0.872	R	0.084	66.5
C vs. T	Overall	1	182	16	3.233	1.145–9.127	0.027	NA	NA	NA
	European	NA	NA	NA	NA	NA	NA	NA	NA	NA
	Asian	1	182	16	3.233	1.145–9.127	0.027	NA	NA	NA
	AS	1	182	16	3.233	1.145–9.127	0.027	NA	NA	NA
	PsA	NA	NA	NA	NA	NA	NA	NA	NA	NA

AS: ankylosing spondylitis, PsA: psoriatic arthritis, R: random effects model, F: fixed effects model, R: responder, NR: non-responder, OR: odds ratio, CI: confidence interval, P-val: P-value, I²: between-study variability attributable to heterogeneity, NA: not available.

Assessment of study quality, heterogeneity, and publication bias

Quality assessment scores for each study ranged from 5 to 8. Notably, between-study heterogeneity was identified in meta-analyses of TNF-α polymorphisms, except for the − 238 A/G polymorphism (Tables 2 and 3). While funnel plots, which are typically used to detect publication bias, were challenging to interpret due to the limited number of studies in the meta-analysis, Egger’s regression analysis indicated no evidence of publication bias for the TNF-α polymorphisms addressed (Egger’s regression test p-values > 0.1) (Fig. 3). The study was registered in PROSPERO (CRD42023472655).

Fig. 3.

Funnel plot for studies investigating the links between the G allele of tumor necrosis factor (TNF)-α-308 A/G polymorphism and responsiveness to TNF-inhibitors in individuals with ankylosing spondylitis (AS) and psoriatic arthritis (PsA). The Egger’s regression p-value is 0.355, indicating the absence of significant publication bias.

Discussion

Results of the present meta-analysis revealed a significant association between the TNF-α -308 G allele and an increased responsiveness to TNF-α blockers in both AS and PsA. This finding aligns with those of previous studies, indicating that this specific polymorphism can serve as a valuable genetic marker for predicting treatment outcomes. The OR of 4.221 indicates that patients carrying the − 308 G allele have a substantially higher likelihood of responding positively to anti-TNF-α therapy, indicating that carriers of this allele are over four times more likely to respond positively to TNF-α blockers compared to non-carriers. This substantial effect size underscores the potential of genotyping to guide personalized treatment decisions. From a cost-effectiveness perspective, pharmacogenomic screening prior to initiating TNF-α inhibitor therapy could reduce unnecessary exposure to expensive biologics in likely non-responders. Real-world data suggest that the number needed to treat (NNT) for TNF inhibitors in AS and PsA is approximately 2, with annual costs exceeding $40,000 per patient to achieve minimal clinically important improvement. By identifying genetic responders, TNF genotyping could optimize resource allocation, reduce trial-and-error prescribing, and improve long-term outcomes. We have added a short cost-effectiveness estimation: assuming a genotyping cost of approximately USD 40 per patient and a non-responder rate of ~ 30%, selective use of genotyping in high-risk patients could reduce unnecessary drug exposure and associated costs (anti-TNF therapy ~ USD 15,000/year). Furthermore, stratification according to ethnicity revealed that this association was consistent among Europeans and Asians, emphasizing its robustness across diverse populations. Additionally, a disease-specific meta-analysis revealed that the − 308 G allele is associated with improved responsiveness in both AS and PsA. Patients with AS and this allele had an OR of 10.89, whereas those with PsA had an OR of 2.451, suggesting that this genetic variant has a substantial impact on enhancing treatment outcomes. These findings underscore the potential clinical relevance of this polymorphism for guiding therapeutic decisions in patients with these conditions. In contrast, the meta-analysis did not reveal a significant association between the TNF-α + 489 GG genotype and response to TNF-α blockers in PsA. However, one study indicated a potential association with AS, emphasizing the need for further research and replication to clarify this discrepancy. The study also revealed an association between the TNF-α -857 C allele and the − 238 G allele and response to TNF-α blockers in PsA, indicating that these polymorphisms may influence treatment outcomes in this condition. In rheumatoid arthritis (RA), several studies have investigated the association between TNF-α -308 G/A polymorphism and response to TNF inhibitors, with mixed results^34,35. Some reports suggest that carriers of the − 308 G allele may exhibit better clinical outcomes, while others found no significant correlation, highlighting the complexity of genetic influences in RA³⁵. Similarly, in inflammatory bowel disease—including Crohn’s disease and ulcerative colitis—research has explored TNF-α promoter polymorphisms, with certain alleles (e.g., -308 G and − 857 C) showing potential predictive value for response to infliximab and adalimumab, although findings remain inconsistent across populations³⁶. These observations reinforce the relevance of TNF-α genotyping across multiple immune-mediated conditions and support the broader applicability of pharmacogenomic approaches to optimize anti-TNF therapy.

The interpretation of our findings should take into account the heterogeneity in response criteria (e.g., BASDAI50, ASAS20/40, PsARC) and the allowance of concomitant medications across studies. These differences may have influenced treatment outcomes and contributed to between-study variability. Although random-effects models were applied to mitigate heterogeneity, residual confounding remains possible, underscoring the need for standardized response definitions and careful control of background therapies in future pharmacogenetic studies.

The TNF-α -308 G allele is associated with heightened production of TNF-α mRNA^37,38. In individuals with AS and PsA carrying this G allele, there is increased expression of TNF-α. This heightened expression of TNF-α may contribute to the chronic inflammation evident in these conditions. TNF-α is a key pro-inflammatory cytokine, and its overproduction can lead to joint inflammation and damage. Results of this study also suggest that TNF-α -857 C and − 238 G alleles may influence the responsiveness to TNF-α blockers in PsA. These alleles may modulate the balance between pro-inflammatory TNF-α and anti-inflammatory transforming growth factor-beta (TGF-β)³⁹. TGF-β is regulated by TNF-α, and an imbalance between these cytokines can affect the inflammatory response. The presence of specific alleles may lead to altered cytokine profiles, thereby affecting the efficacy of TNF-α blocker therapy. An important consideration in interpreting our findings is the issue of multiple testing. While our primary analysis focused on the − 308 A/G polymorphism, which demonstrated a strong and biologically plausible association with treatment response, additional analyses of other loci were exploratory in nature. Although several of these secondary associations reached nominal statistical significance, they may be subject to inflation of type I error and should therefore be interpreted with caution. Replication in larger, independent cohorts and validation through functional studies will be essential to confirm these preliminary signals.

Findings of the present study have several clinical implications. First, identifying patients with the TNF-α -308 G allele can potentially aid in personalized treatment strategies because they are more likely to respond favorably to anti-TNF-α therapy. Kroeger et al. demonstrated that carriers of the − 308 A allele exhibited approximately 2–3 fold higher TNF mRNA expression compared to -308G homozygotes⁴⁰. Similarly, Wilson et al. reported a ~ 2.5-fold increase in TNF secretion associated with the − 308 A allele⁴¹. This can optimize treatment decision-making and potentially reduce healthcare costs associated with ineffective therapies. Second, association of the TNF-α -857 C and − 238 G alleles with treatment responsiveness in PsA suggests that a broader range of genetic markers may play a role in guiding treatment choices. Understanding these genetic factors could lead to more precise and tailored therapies, ultimately improving patient outcomes and quality of life. Although this meta-analysis provides valuable insights, it had its limitations. The studies included in the analysis may vary in terms of patient populations, study designs, and methodologies, which may have introduced heterogeneity. Additionally, the sample sizes in some sub-analyses, such as the + 489 GG genotype, were small, potentially affecting the robustness of the findings.

This study has several limitations. The relatively small number of studies for certain polymorphisms, heterogeneity in study design, response criteria, and concomitant medications, as well as the potential for publication bias and multiple testing, should be considered when interpreting our findings. Furthermore, most included studies were conducted in European and Asian populations, which may limit generalizability. These limitations highlight the need for larger, multi-ethnic, prospective studies with standardized response definitions to validate the predictive role of TNF-α polymorphisms in anti-TNF therapy. While both AS and PsA are spondyloarthritides that share overlapping pathophysiological features and therapeutic approaches, combining them introduces heterogeneity. We therefore performed additional subgroup analyses separating AS and PsA studies, and have presented these results. Although the statistical power was limited, the direction of association for the − 308 polymorphism remained consistent across subgroups. We conducted Egger’s regression analysis for each TNF-α polymorphism, which yielded p-values > 0.1, indicating no statistically significant evidence of publication bias. However, the limited number of included studies may reduce the sensitivity of funnel plot interpretation and regression-based tests, potentially masking subtle bias. To address the “winner’s curse”—the tendency for initial studies to overestimate effect sizes—we emphasize that our meta-analysis incorporated multiple studies across diverse populations, which helps mitigate this concern. Nonetheless, the observed associations, particularly those with large effect sizes (e.g., OR 4.221 for the − 308 G allele), should be interpreted with caution until validated in larger, prospective cohorts. Although our meta-analysis was adequately powered for the primary association (-308 G allele) and reasonably powered for subgroup findings (-857 C and − 238 G alleles), these estimates are post hoc and conditional on the observed effect sizes. Therefore, replication in larger, prospective cohorts remains essential to confirm the stability and generalizability of these associations.

In conclusion, this meta-analysis demonstrated a significant association between the TNF-α -308 G allele and an increased responsiveness to TNF-α blockers in AS and PsA, and also indicated the potential influence of TNF-α -857 C and − 238 G alleles on TNF-α blocker responsiveness in PsA. Conversely, no such association was found for the TNF-α + 489 GG genotype. These findings provide valuable insights into the genetic factors that influence the efficacy of TNF-α blocker therapies in AS and PsA. Future research should aim to replicate these findings in larger and more diverse patient cohorts, and explore the potential interactions between these genetic markers and other clinical factors that may affect treatment responsiveness. Advances in our understanding of the genetic underpinnings of treatment outcomes in AS and PsA hold promise for more effective personalized therapies for patients living with these conditions.

Author contributions

YH Lee was involved in conception and design of study, acquisition of data, analysis and/or interpretation of data, drafting the manuscript, revising the manuscript critically for important intellectual content. GG Song was involved in conception and design of study, analysis and interpretation of data, and drafting the manuscript.

Data availability

All data generated or analysed during this study are included in this article. Further enquiries can be directed to the corresponding author.

Declarations

Competing interests

The authors declare no competing interests.