Patients with CD were enrolled between October 2020 and April 2022 at the Department of Gastroenterology, the First Affiliated Hospital of Soochow University (Suzhou, China). Participants (>18 years old) had a histopathological diagnosis of CD and received scheduled therapy. The treatment regimen included the administration of 5 mg/kg IFX at weeks 0, 2, and 6 and then every 8 weeks during maintenance therapy. A stable dose of 5-aminosalicylic acid (5-ASA) was allowed for at least 4 weeks. The study excluded patients who had been administered IFX for more than 2 years and had comedications including immunomodulators (azathioprine, 6-mercaptopurine) and steroids. Electronic medical records were used to collect information about weight, age, sex, smoking history, disease phenotype and faecal calprotectin (fCal) values. Within the 24 h before IFX infusion, a total of 2 mL of peripheral whole blood was collected for genetic analysis as well as TLI and ATI measurements. Meanwhile, the Crohn’s disease activity index (CDAI) was recorded. Clinical remission (CR) was defined as a CDAI less than 150 points. The normalization of fCal was defined as less than 250 μg/g. Patients provided written and informed consent. The research was approved by the Research Ethics Committee of the First Affiliated Hospital of Soochow University.

Separation of plasma from whole blood was performed to detect TLI. TLI were analyzed with an in-house developed and validated ELISA method. High binding 96-well plates were coated overnight with TNF-α (300-01A, PeproTech, United Kingdom) at 4°C. Plasma samples were diluted in PBS and incubated for 2 h at 37°C in a shaker. Human anti-infliximab, clone AbD19376_hIgG1 (HCA216P, Bio-Rad, United Kingdom), was used as the detecting agent. The lower and upper limits of quantification were 0.50 μg/mL and 40.00 μg/mL, respectively.

ATI was detected by a bridging ELISA method. The analytical method was validated according to current industry practice (Myler et al., 2021). Plasma samples and controls were diluted in glycine (100 mM, pH 2.5) at a minimum dilution of 1:20 for 30 min at RT in a shaker. Then, 50 μL of the diluted samples or controls and 95 μL master mixture containing 0.1 μg/mL each of biotin-IFX, HRP-IFX and 5 μL of Tris (1.5 M, pH 9.0) were added to a 96-well polypropylene plate and then incubated for 2 h at RT in a shaker. Immediately after, 100 μL of a mixture was moved to a streptavidin matrix coated 96-well plate and incubated at RT for 1 h in a shaker. The plate was washed three times, followed by the addition of 3,3’,5,5’-tetramethylbenzidine substrate and 2 M H₂SO₄. A negative control (NC) of pooled plasma from IFX-naïve CD patients as well as low and high positive controls containing 100 and 20, 000 ng/mL rabbit anti-IFX polyclonal antibodies prepared in NC, was included in each plate. The screening cut point based on a panel of plasma samples from 51 IFX-naïve CD patients, was determined to be 1.16 (S/N). Based on the analysis of 51 IFX-naïve plasma samples from CD patients without or with IFX (20 μg/mL), the confirmatory cut point was determined to be a decrease in signal of more than 17.21%. The titer cut point was determined to be 1.34. The minimum significant ratio was determined to be 3. ATI titers of 1:20 and ≥1:60 were considered a low titer and a high titer, respectively. Drug-tolerance at low positive control was ≤12.50 μg/mL IFX. The relative sensitivity of this assay was determined to be 50 ng/mL (for validation parameters, see Supplementary Table S1).

DNA was extracted from 250 μL of whole blood using a DNA Extraction Kit (Takara, Japan). The concentration and purity of DNA were calculated with the OD-1000 (OneDrop^TM, China). Thirteen primer pools of SNPs, namely, HLA-DQA1 (rs2097432), FCGR3A (rs396991), FCGR2A (rs1801274), PTPRC (rs10919563), KLRC1 (rs7301582), HLA-E (rs1264457), IL-17RA (rs4819554), ATG16L1 (rs10210302), TRAF1 (rs3761847), TNF (rs1800629), TNFRSF1A (rs767455), TNFRSF1B (rs1061622) and CCNY (rs12777960) (Supplementary Table S2), were designed. Library preparation was performed using a two-step polymerase chain reaction (PCR). These reaction mixtures contained 1 μL of DNA, 0.4 mM dNTPs, 3 μM of each primer, Taq Buffer (with MgCL₂) and 0.04 U Taq DNA polymerase. Twenty-five μL of reaction mixture was used for PCR. The first PCR conditions were 95°C for 5 min, followed by 10 cycles of 94°C for 30 s, 63°C for 30 s, and 72°C for 30 s. The second PCR conditions were 30 cycles of 95°C for 30 s, 58°C for 30 s, and 72°C for 30 s, with a final extension at 72°C for 10 min. All PCR reagents were provided by ThermoFisher (Waltham, MA) and Sangon Biotech (Shanghai, China). Paired-end sequencing of the library was performed on HiSeq XTen sequencers (Illumina, San Diego, CA).

Statistical analysis was performed using GraphPad Prism® version 9.0 and IBM SPSS Statistics for Mac, version 26.0. The TLI below the lower limits of quantification were imputed by 0.25 μg/mL. The mean and standard deviation or the median and interquartile range (IQR) were used to describe continuous clinical and demographic variables. Categorical variables were expressed as percentages. The Mann‒Whitney U test (two groups) and the Kruskal–Wallis H test (more than two groups) were used to compare continuous non-normality variables. The distribution of 13 SNPs within 13 genes was tested for Hardy-Weinberg equilibrium using the chi-squared goodness-of-fit test. For all SNPs, two genetic models were used: the additive model, which supposes that the effect of the risk allele increases with the addition of each allele copy; and the dominant model, which analyzes the heterosis. A chi-squared test and logistic regression were performed to evaluate the association and multiplicative interactions between SNPs and ATI formation, respectively. All reported p values were 2-sided, and a significant p < .05 was considered a statistically significant difference.

A total of 104 patients with CD were included. The patients’ demographic and clinical characteristics are shown in Table 1. A total of 73.08% of patients were male. The median duration of IFX treatment was 10 months (IQR: 5–16.5). The median CDAI point was 87.6 (IQR: 68.2–123.5). The median fCal value was 64.5 (IQR: 30–344.5). The median TLI in CD patients was 2.89 μg/mL (IQR: 1.58–5.07). The TLI in 10 patients was less than 0.50 μg/mL (10/104, 9.62%).

Eighty-three of the 104 patients (79.81%) achieved clinical remission (CR). The median TLI in patients with CR was higher than that in patients without CR (3.80 vs. 1.50 μg/mL, p < .001, Figure 1A). The median TLI in patients who achieved CR and fCal normalization was higher than that in patients who did not (4.40 vs. 1.90 μg/mL, p = .02, Figure 1B).

The TLI ranged from undetectable to 11.93 μg/mL. The occurrence of ATI was 53.85% (56/104). The titers of ATI showed a wide range from low to high titers. The relationship between TLI and titers of ATI was evaluated. The median TLI in patients with high-titer ATI (1.15 μg/mL, IQR: 0.25–2.39, Figure 2) was significantly lower than that in patients with low-titer ATI (2.95 μg/mL, IQR: 2.25–4.28, p = .03, Figure 2). A significant difference in TLI was also observed between the high-titer group and the negative-ATI group (4.48 μg/mL, IQR: 2.65–6.64, p < .001, Figure 2). No significant differences in TLI were observed between the negative-ATI group and the low-titer group (p = .17, Figure 2).

The frequencies of genotypes for 13 SNPs within 13 genes in 104 patients were in line with Hardy-Weinberg equilibrium (HWE, p > .05, Supplementary Table S3 in the Supplement). The association between 13 SNPs and ATI formation under the dominant model and additive model are shown in Supplementary Table S4. Variant carriers of the SNP (rs2097432 in HLA-DQA1*05) under the dominant model had an higher rate of ATI formation (GG and AG: 27/38 = 71.05%, AA: 29/66 = 43.94%, p = .01, Table 2). There was also a significant association between rs396991 in FCGR3A and ATI production under the dominant model. Carriers of the variant SNP (rs396991 in FCGR3A) had a higher probability of producing ATI (CC and AC:37/57 = 64.91%, AA: 19/47 = 40.43%, p = .01, Table 2). No association was found between other SNPs and ATI production (Table 2; Supplementary Table S5). In logistic regression, the HLA-DQA1*05 rs2097432 GG and GA genotypes increased the risk of ATI formation compared to the AA genotype (GG and AG vs. AA, OR 2.94, 95% CI 1.19–7.30, p = .02, Table 3). Patients carrying the CC and AC genotypes of rs396991 in FCGR3A were associated with a higher frequency of ATI formation (CC and AC vs. AA, OR 2.94, 95% CI 1.24–6.96, p = .01, Table 3). According to the number of the variants in rs2097432 and rs393991, a greater proportion of patients with two variants showed ATI production (two variants vs. no variant, 17/21 = 80.95% vs. 9/30 = 30.00%, OR 9.92, 95% CI 2.59–37.87, p = .001; single variant vs. no variant, 30/53 = 56.60% vs. 9/30 = 30.00%, OR 3.04, 95% CI 1.18–7.88, p = .02, Figure 3 and Table 3).