Introduction

Behcet's disease is a chronic immune-mediated multisystemic vasculitis featured by recurrent oral aphthous ulcers, genital ulcers, uveitis, and skin lesions.1 The incidence of Behcet's disease and proportion of related gastrointestinal involvement were reported as 7.5–420/100 000 person years and 2.8–60%, respectively, along the “ancient silk road” (from East Asia to the Mediterranean Basin).2-8 Patients with Behcet's disease can also suffer from gastrointestinal symptoms like diarrhea, intestinal bleeding, and perforation, which was of great importance as they were associated with worse prognosis.9 The cumulative possibility of surgery for intestinal Behcet's disease was reported to be 23.9%, 31.6–32.4%, and 44.4% at 2, 5, and 10 years, respectively, and cumulative hospitalization rates were 59.0% and 69.2% at 5 and 10 years, respectively.10, 11 Thus, timely and effective treatment is very important to improve prognosis of these patients.

Conventional treatment of Behcet's disease included 5-ASA, corticosteroids, and immunomodulators. And the nonresponse rate to conventional therapy remains high. It was reported that 38–66.7% of patients with intestinal Behcet's disease achieved complete response after corticosteroids and immunomodulators. However, among complete responders, 25% and 43–49% patients suffered from relapse of disease at 2 and 5 years, respectively,12-14 leading to enormous health damage and economic loss. However, consensus on the treatment of intestinal Behcet's disease was still lacking. Anti-TNF agents, widely used in immune-mediated inflammatory diseases such as inflammatory bowel disease, were suggested to be effective in the treatment of intestinal Behcet's diseases by some studies and have been used in the induction and maintenance of clinical remission and mucosal healing in patients with Behcet's disease. The most common adverse drug reaction (ADR) of anti-TNF agents included infusion reaction and infection, which should be monitored during the course of treatment. Anti-TNF agents could be a promising therapy for intestinal Behcet's disease.

However, no randomized clinical trials have been conducted to evaluate the effectiveness and safety of anti-TNF agents in the treatment of intestinal Behcet's diseases. And the retrospective nature, small sample size, and heterogeneity limited the quality of evidence of some studies. Therefore, we conducted this systematic review and meta-analysis. By summarizing current evidence, we evaluated the effectiveness (proportion of clinical or endoscopic remission) and safety (incidence of ADRs) of anti-TNF agents in the treatment of intestinal Behcet's disease. The results of our study would provide reference for the decision in arrangement of anti-TNF agents for patients with intestinal Behcet's disease.

Methods

This study was registered on the PROSPERO registry (CRD42020170722, CRD42020223588). And the results of this systematic review were reported according to the PRISMA statement.

Inclusion and exclusion of studies The criteria of inclusion and exclusion of potentially relevant studies were described as follows based on the “PICOs” (participants, intervention, control, outcome, study design) principle. Participants (P): Patients diagnosed as intestinal Behcet's disease. Intervention (I): Anti-TNF agents, including infliximab, adalimumab, etanercept, golimumab, and certolizumab. Control (C): None. Outcomes (O): The outcomes of this study included efficacy outcome and safety outcome. Efficacy outcomes were evaluated based on the proportions of (i) clinical remission (complete response), (ii) marked clinical improvement (partial response), (iii) endoscopic remission (mucosal healing), (iv) marked endoscopic improvement, (v) corticosteroids discontinuation, and (vi) recovery of C-reactive protein (CRP) levels. These were evaluated at 3-, 6-, 12-, 24-month follow-up. Safety outcome was evaluated by (i) ADR of any kind, (ii) infusion reaction, and (iii) infection, with at least 10-week follow-up. Study design (S): Retrospective or prospective cohort studies, clinical trials, randomized clinical trials, and consecutive case series with sample size > 5 were eligible for inclusion. Review, systematic review, letters, editorials, recommendations, and case report would be excluded. Literature search and screening

We conducted searches for eligible studies on PubMed, Embase, and Cochrane Library, and the last search was on February 9, 2021. The detail of our search strategy was shown in Table S1. No restrictions were set on publication types. The references of eligible studies were also viewed.

Two investigators (ZMY and LTT) independently judged the eligibility of the potentially eligible studies based on title, abstract, and full texts according to the criteria described as above. Disagreement between the two investigators would be discussed with an expertise (YH).

Risk of bias assessment

The eligible studies we included were retrospective or prospective cohort studies. As these studies did not set control groups, we could not use Newcastle-Ottawa Scale to assess the risk of bias. Thus, a modified tool based on the Agency for Healthcare Research and Quality tool was used for risk of bias assessment. The criteria of this tool were described in detail in Table S2.

Data extraction Two investigators (ZMY and LTT) extracted the following data independently and double checked by another investigator (BXY). Basic information of the article: author, year of publication, registration ID, study design, country, and sample size. Baseline characteristics: age, sex, disease activity or severity, and refractory to conventional therapy or not. Intervention: therapeutic regimen of anti-TNF agents, including drug generic name, dosage, course, and drug combination. Outcomes: As efficacy and safety outcomes mentioned above. As for the efficacy outcomes, clinical remission was defined as Disease Activity Index for Intestinal Behcet's Disease score < 20, or disappearance of clinical symptoms related to intestinal Behcet's disease. Marked improvement referred to a decrease in the Disease Activity Index for Intestinal Behcet's Disease score of more than 20 points from the baseline, or a significant improvement of gastrointestinal symptoms judged by the investigators. Mucosal healing, or endoscopic remission, was defined as disappearance of active intestinal ulcers (healing or scarring) confirmed by endoscopy. Marked endoscopic improvement was defined as ≥ 1/2 reduction of intestinal ulcers or marked improvement judged by endoscopists. The follow-up time of these studies ranged from 12 weeks to 2 years. Corticosteroids discontinuation referred to reduction in the dose of corticosteroids to zero. Recovery of CRP was defined as CRP level reduced to the normal range. The proportions of efficacy outcomes were calculated at 3, 6, 12, and 24 months (corresponding to 12–14, 24–30, 52–54, and 100–108 weeks). For safety outcome, the combined incidence of (i) any kind of ADRs, (ii) infusion reaction, and (iii) infection was calculated. Statistical analysis

As the studies included were single-arm cohort studies, in this meta-analysis, pooled proportions were calculated using inverse variance methods and random-effect model. In order to improve the normality of data distribution, the original data were calculated after logarithmic conversion. Heterogeneity was considered to be significant if I2 was equal to or larger than 50% or two-side P < 0.1. Because the tools for assessing reporting bias might be not suitable for meta-analysis calculating pooled proportions and incidences, the reporting bias was not assessed in this study. Subgroup analysis was performed based on specific type of anti-TNF agents (infliximab, adalimumab, etanercept, golimumab, and certolizumab). Sensitivity analysis was performed using influence analyses based on the leave-one-out-method, in which we calculated the pooled proportion each time leaving out one study. Statistical analyses were performed in r platform (Version 4.0.2, R Foundation for Statistical Computing, Vienna, Austria, 2018, https://www.R-project.org/) using the package “meta.”

Results General information of included studies

The literature research found 828 results. According to titles and abstracts, 58 studies were regarded to be potentially relevant. These studies were assessed by full-text, and 13 studies were included in the final quantitative synthesis. A flow diagram demonstrated the process of studies selection (Fig. 1). The characteristics of the included studies were shown in Table S3. All of the included studies were single-arm cohort studies.

Flow diagram demonstrating the process of selection of studies. [Color figure can be viewed at wileyonlinelibrary.com]

A total of 739 patients with intestinal Behcet's disease were recruited in the 13 eligible studies included. The diagnostic criteria of included studies were based on international or domestic criteria. Sample sizes ranged from 11 to 471. Except for one study conducted in France, the other 12 studies were conducted in East Asia (Japan, China or Korea), and race of participants was not reported in most studies. Patients were treated with infliximab in five studies, adalimumab in three studies, etanercept in one study, and with more than one type of anti-TNF agents in four studies.

All of the eligible studies included patients with moderate to severe intestinal Behcet's disease or unreported disease activity and patients who did not respond well to conventional therapy. Regimen of infliximab and adalimumab were consistent between studies, and concomitant medication was common in these cohorts. Thus, subgroup analysis was performed based on specific type of anti-TNF agents (infliximab, adalimumab, or both) only. The risk of bias assessment was listed in Table S4. As the included studies were single-arm cohort studies, none of them reported blinding methods. Of the 13 included studies, 4 had predefined protocols. And in 4 of 13 studies, less than 90% of participants completed the follow-up. In the other aspects, no high risk of bias was identified in the included studies.

Efficacy of anti-tumor necrosis factor agents in the treatment of intestinal Behcet's disease Clinical remission (complete response)

Results of pooled proportions of clinical remission were shown in Figure 2. Clinical remission was achieved in 61% (95%CI 48–78%) of patients at Month 3, 51% (95%CI 40–66%) of patients at Month 6, 57% (95%CI 48–67%) at Month 12, and 38% (95%CI 16–88%) at Month 24. And the I2 were 82%, 55%, 43%, and 78% respectively, indicating a moderate statistical heterogeneity and random-effect model was required.

Forest plots for subgroup analysis of clinical remission after specific type of anti-TNF agent treatment. (a) Clinical remission after 3-month anti-TNF agent treatment; (b) clinical remission after 6-month anti-TNF agent treatment; (c) clinical remission after 12-month anti-TNF agent treatment; (d) clinical remission after 24-month anti-TNF agent treatment. ADA, adalimumab; CI, confidence interval; IFX, infliximab.

As shown in Figure 2, subgroup analysis was performed based on the specific type of anti-TNF agents (infliximab, adalimumab, or both) to explore the efficacy of each drug. The pooled clinical remission proportions at 3-, 6-, 12-, and 24-month follow-up after the first dose of infliximab were 64% (95%CI 52–100%), 50% (95%CI 38–65%), 55% (95%CI 41–74%), and 38% (95%CI 15–92%), respectively. And for adalimumab, the pooled proportions of clinical remission were 58% (95%CI 52–64%), 39% (95%CI 13–100%), 58% (95%CI 43–80%), and 35% (95%CI 8–100%), respectively. Heterogeneity was not significant between studies on infliximab (I2 = 0–55%), while moderate to high level of heterogeneity (I2 = 50–87%) existed between studies on adalimumab.

The sensitivity analyses (influence analyses) about clinical remission were shown in Figure S1. The results were stable after omitting any one of the included studies.

Marked improvement (partial response)

The pooled proportions of marked improvement in gastrointestinal manifestations, or partial response, after 3-, 6-, 12-, and 24-month treatment with anti-TNF agents, were 74% (95%CI 60–91%), 68% (95%CI 52–87%), 64% (95%CI 54–75%), and 43% (95%CI 28–66%), respectively (Fig. 3). And the heterogeneity was low or moderate (I2 = 0–69%).

Forest plots for subgroup analysis of marked clinical improvement after specific type of anti-TNF agent treatment. (a) Marked clinical improvement after 3-month anti-TNF agent treatment; (b) marked clinical improvement after 6-month anti-TNF agent treatment; (c) marked clinical improvement after 12-month anti-TNF agent treatment; (d) marked clinical improvement after 24-month anti-TNF agent treatment. ADA, adalimumab; CI, confidence interval; IFX, infliximab.

The results of subgroup analysis based on type of anti-TNF agents were shown in Figure 3. The partial response rate to adalimumab alone was reported as 45% (95%CI 28–73%) at Month 6, 60% (95%CI 42–86%) at Month 12, and 40% (95%CI 23–68%) at Month 24. Heterogeneity between studies on adalimumab was not evaluated as only one study was included in each time point. The combined proportion of partial response to infliximab was 82% (95%CI 72–93%) at Month 3, 71% (95%CI 51–100%) at Month 6, 65% (95%CI 50–84%) at Month 12, and 50% (95%CI 25–100%) at Month 24. Heterogeneity between studies on infliximab varied (I2 = 0%, 76% and 47%).

The sensitivity analyses (influence analyses) about marked clinical improvement were shown in Figure S2. The results were stable after omitting any one of the studies.

Endoscopic remission (mucosal healing)

As shown in Figure 4, the pooled proportions of mucosal healing, or endoscopic remission, after anti-TNF agents administration were 66% (95%CI 50–86%) at Month 3, 82% (95%CI 48–98%) at Month 6, 65% (95%CI 51–81%) at Month 12, and 69% (95%CI 39–100%) at Month 24. Significant statistical heterogeneity existed between studies (I2 = 58–83%).

Forest plots for subgroup analysis of endoscopic remission after specific type of anti-TNF agent treatment. (a) Endoscopic remission after 3-month anti-TNF agent treatment; (b) endoscopic remission after 6-month anti-TNF agent treatment; (c) endoscopic remission after 12-month anti-TNF agent treatment; (d) endoscopic remission after 24-month anti-TNF agent treatment. ADA, adalimumab; CI, confidence interval; ETN, etanercept; GLM, golimumab; IFX, infliximab.

Shown in Figure 4 were the results of subgroup analysis according to different types of anti-TNF agents. Only one study reported the proportion of mucosal healing after adalimumab treatment as 55% (95%CI 37–82%) at Month 12 and 50% (95%CI 32–78%) at Month 24. Pooled proportions of mucosal healing after treatment with infliximab were 78% (95%CI 65–93%) at Month 3, 82% (95%CI 48–98%) at Month 6, and 81% (95%CI 65–100%) at Month 12. And statistical heterogeneity was not significant between these studies on infliximab (I2 = 0% and 31%). Patients treated with etanercept were reported by one study to have an endoscopic remission proportion of 89% (95%CI 77–100%) at Month 24.

The sensitivity analyses (influence analyses) about mucosal healing were shown in Figure S3. The pooled results of mucosal healing at Months 3, 12, and 24 were stable after omitting any one of the studies. The number of studies reporting the proportion of mucosal healing at Month 6 was too few to conduct a meaningful influence analysis.

Marked endoscopic improvement

Figure 5 showed the pooled proportions of marked endoscopic improvement (≥ 1/2 reduction in ulcers or judged by endoscopist), which were 56% (95%CI 30–80%) at Month 3 and 58% (95%CI 52–66%) at Month 12. Subgroup analysis and sensitivity analysis provided little additional information due to limited number of eligible studies.

Forest plots for subgroup analysis of marked endoscopic improvement after specific type of anti-TNF agent treatment. (a) Marked endoscopic improvement after 3-month anti-TNF agent treatment; (b) marked endoscopic improvement after 12-month anti-TNF agent treatment. ADA, adalimumab; CI, confidence interval; ETN, etanercept; GLM, golimumab; IFX, infliximab.

Recovery in C-reactive protein level

As shown in Figure S4, the pooled proportions of recovery of CRP level (reduced into normal range) were 77% (95%CI 56–91%) at Month 3, 62% (95%CI 49–78%) at Month 6, 38% (95%CI 20–59%) at Month 12, and 79% (95%CI 54–94%) at Month 24. Subgroup analysis and sensitivity analysis provided little additional information due to limited number of eligible studies.

Discontinuation of corticosteroids

The pooled proportion of corticosteroids discontinuation was shown in Figure S5. Corticosteroids were discontinued in 65% (95%CI 59–72%) of patients at Month 12 and 30% (95%CI 12–54%) at Month 24. Because of the limited number of studies, subgroup analysis and sensitivity analysis did not provide additional information.

Safety of anti-tumor necrosis factor agents in the treatment of intestinal Behcet's disease

A total of seven studies reported the incidence of ADR of anti-TNF agents in the treatment of intestinal Behcet's disease. Five studies were about infliximab and reported the number of patients that had ADRs. The other two studies were about adalimumab and reported the incidence of ADRs in units of events/100 person-years. As the type of drug and measurement of incidence were different between these two groups of studies and consistent within each group, the pooled incidence of any kind of ADR was calculated separately by each group. The data of safety profile were collected at the last follow-up from studies included.

The pooled proportion of any kind of ADR was 22% (95%CI 7–69%) related to infliximab, as shown in Figure 6. And the pooled proportions of infusion reactions and infection were 12% (95%CI 5–29%) and 21% (95%CI 6–80%), respectively, related to infliximab. However, the heterogeneity was generally significant (I2 = 83%, 29%, and 85%, respectively). The incidence of ADR reported by studies related to adalimumab used in the treatment of intestinal Behcet's disease was summarized in Table S5. The majority of the ADR reported was acute or delayed infusion reaction or infection. For infliximab, severe infection that led to hospitalization was reported in two cases.15, 16 Other ADRs reported included one case of cataract.17 Drug discontinuation related to ADR was reported in three cases.16 No death related to ADR occurred.

Forest plots for safety of infliximab in the treatment of intestinal Behcet's disease. (a) The pooled proportion of overall adverse drug reactions; (b) the pooled proportion of infusion reaction; (c) the pooled proportion of infection. CI, confidence interval.

Discussion

In this study, we summarized the current evidence on the efficacy and safety of anti-TNF agents, including infliximab, adalimumab, etanercept, and golimumab in the treatment of intestinal Behcet's diseases. The result of our study showed that after administration of anti-TNF agents for intestinal Behcet's disease, clinical remission (complete response) was achieved in 61% (95%CI 48–78%), 51% (95%CI 40–66%), 57% (95%CI 48–67%), and 38% (95%CI 16–88%) of patients at 3, 6, 12, and 24 months, respectively. And endoscopic remission (mucosal healing) was achieved in 66% (95%CI 50%–86%), 82% (95%CI 48%–98%), 65% (95%CI 51%–81%), 69% (95%CI 39%–100%) of patients at 3, 6, 12, and 24 months, respectively. These results favored the use of anti-TNF agents in the treatment of intestinal Behcet's diseases.

Anti-TNF agents were proved to be of good efficacy and safety in the treatment of inflammatory bowel disease, psoriasis and ankylosing spondylitis.18-21 Behcet's disease shared some similar clinical manifestations and pathophysiology with inflammatory bowel disease, which might explain why medications for inflammatory bowel disease could also be effective in treating intestinal Behcet's disease.22, 23 And anti-TNF agents was reported to be of good efficacy and acceptable safety in treating Behcets' disease-associated uveitis.24-26 Gastrointestinal involvement was considered more dangerous to patients with Behcet's disease, as the prognosis was usually worse.9 Previous studies have shown that anti-TNF agents, including infliximab and adalimumab, led to the relief of gastrointestinal symptoms and ulcers in gastrointestinal tract in patients with Behcet's disease.15-17, 27-36 However, these studies were single-arm cohort studies. No randomized clinical trials have been conducted in this field. Sample sizes of most of these studies were small. Level of evidence of individual studies was not strong enough for guiding clinical practice. However, there was still no systematic review and meta-analysis combining these results to provide more convincing evidence on the efficacy and safety of anti-TNF agents in the treatment of intestinal Behcet's disease.

This study is the first systematic review and meta-analysis to evaluate the safety and efficacy of anti-TNF agents in the treatment of intestinal Behcet's disease. According to the Evidence-based diagnosis and clinical practice guidelines for intestinal Behçet's disease 2020 edited by Intractable Diseases, the Health and Labour Sciences Research Grants, anti-TNF agents were strongly recommended for the treatment of intractable intestinal Behcet's disease, including complete or incomplete types of intestinal Behcet's disease refractory to conventional therapies.37 Our findings supported this point statistically. Patients included in these cohort studies who received anti-TNF agents for intestinal Behcet's disease mostly had poor response to conventional therapies. Thus, among these patients with refractory diseases, a clinical or endoscopic remission rate of about 60% suggests that anti-TNF agents are promising therapeutic options that should be recommended. The incidence of overall ADRs in these patients was 22% (95%CI 7–29%) for infliximab and reported by two studies as 30.7 and 28.7 events per 100 patient years for adalimumab. The majority of reported ADRs was acute or delayed infusion reaction and infections. Severe ADRs were seldomly reported. The safety of anti-TNF agents was acceptable, especially in the condition of refractory intestinal Behcet's disease, when conventional therapies failed to control the moderate to severe disease. The cumulative possibility of surgery of intestinal Behcet's disease was reported to be 23.9% at 2 years of disease duration, indicating a high risk of severe complications among these patients, especially those with refractory disease.10, 11 Therefore, the benefit of using anti-TNF agents for these patients was considered to overcome the potential harm.

The outcomes of conventional therapies for intestinal Behcet's diseases were reported to be unsatisfactory, which called for advances in medication of intestinal Behcet's diseases. According to Han et al., 39.8% and 49.1% of patients with intestinal Behcet's disease required moderate to high-dose corticosteroids at disease duration of 1 and 4 years, respectively.38 And at 1 year after first dose of corticosteroids, 48.1% of patients with intestinal Behcet's disease were in prolonged response, 35.2% in corticosteroids dependency, and 7.4% received surgery.39 Previous studies reported a complete remission rate of 38–66.7% after 8-month conventional medication and cumulative relapse rates of 24.9%, 43.0% at Years 2 and 5 among patients with intestinal Behcet's disease.12 According to Jung et al., the cumulative relapse rates of intestinal Behcet's disease were 5.8%, 28.7%, 43.7%, and 51.7%, respectively, during thiopurine maintenance therapy at 1, 2, 3, and 5 years after achieving clinical remission.14 The high nonresponse and relapse rates, and the side effect related to long-term use of corticosteroids or immunomodulators suggested that conventional therapies for intestinal Behcet's diseases were still inadequate to achieve optimal therapeutic effect. Biologics, especially anti-TNF agents, should be a promising option for the treatment of intestinal Behcet's disease. Our study suggested that a clinical remission proportion of 38–61% and mucosal healing proportion of 66–82% reflected appreciable efficacy of anti-TNF agents in treating refractory intestinal Behcet's diseases. Because of the higher price of biologics, in real world, anti-TNF agents were used mainly in patients who were refractory to conventional therapy or dependent to corticosteroids. And our study was based on patients with refractory intestinal Behcet's diseases, as 10 of 13 studies reported concomitant medication at baseline and 6 of 13 studies declared that patients enrolled were refractory to conventional therapy, and the other studies did not declare corresponding baseline information. Therefore, the results of our study were able to predict the use of anti-TNF agents in treating intestinal Behcet's diseases in real setting.

However, our study has several limitations. First, this is a meta-analysis of observational studies, as no RCT has been conducted. Meta-analysis of observational studies had some inherent risk of bias. Lack of randomization at enrollment increases the risk of selection bias. Control or adjustment of confounding factors was difficult in observational studies. Also, without registration, results of unpublished investigations were seldomly found, which led to increased reporting bias. Although 6 of the 13 included studies were retrospective studies, the outcomes in our study were objective index, which eliminated the impact of recall bias. In this meta-analysis, we calculated the pooled proportion. By now, no clinical trial with control group was conducted to compare the efficacy and safety of anti-TNF agents with conventional therapies in treating intestinal Behcet's disease. Although the nature of included studies in this meta-analysis did not allow calculation based on odds ratio or relative risk, our meta-analysis still provided valuable information by summarizing the results of cohort studies, which providing a more convincing evidence than any of the previous studies. Second, the sample sizes of most of the studies included were relatively small, leading to a considerable deviation of the pooled proportion. However, the low prevalence of intestinal Behcet's diseases made it hard to recruit a large cohort. Although the sample size might be unsatisfactory, the result of our study was still valuable and meaningful for supporting the use of anti-TNF agents in patients with intestinal Behcet's diseases. Third, follow-up time and some outcome measures were not strictly identical across studies. For example, the time-point of 3-month follow-up was set as week 10, 12, or 14 in different studies. Nonetheless, as the treatment for intestinal Behcet's disease was a long-term process, and small difference in the time when patients were tested was common and acceptable in observational studies, we considered the discrepancy in time point of follow-up did not impact the reliability of the results. Regarding the choice of fixed or random effect model for statistical calculation, we used random effect model in this meta-analysis, as we presumed that the effect size was different among studies, and heterogeneity among studies was considerable.40 For meta-analysis based on observational studies, random effect model is recommended while fixed effect model is not suitable.41 Although one opinion was that the combination of results based on studies with small sample sizes should be performed using fixed effect model, no consensus had been reached to approve this view. What is more, random effect model gave more balanced relative weights for each study and wider confidence interval.40 Therefore, our choice of random effect model was more suitable for this meta-analysis summarizing studies with small sample size and approximate quality. Most of these limitations mentioned arose from the nature of studies included, and the real-world condition of intestinal Behcet's disease cohort, which could not be resolved by statistical methods.

Despite these limitations, our study provides valuable reference on the efficacy and safety of anti-TNF agents in the treatment of intestinal Behcet's disease, which was evaluated comprehensively by proportions of clinical or endoscopic remission, marked clinical or endoscopic improvement, corticosteroids discontinuation and recovery of CRP level, as well as the incidence of overall ADR and infusion reactions. The results of this meta-analysis indicated that anti-TNF agents were promising therapy with good efficacy and safety for intestinal Behcet's disease. And randomized, controlled, multicenter trials are needed for more convincing evidence guiding the use of anti-TNF agents in treating intestinal Behcet's disease.

Conclusion

In summary, our study suggested that anti-TNF agents had appreciable efficacy and acceptable safety in the treatment of intestinal Behcet's disease. And we call for randomized, controlled, multicenter trials to provide more convincing evidence.

Acknowledgments

We thank National Natural Science Fund, Beijing Natural Science Fund, the CAMS Innovation Fund for Medical Science, and National Health and Family Planning Commission of the People's Republic of China for the funding support.