Effectiveness of ustekinumab in postoperative Crohn's disease management: Evidence from a Chinese multicenter cohort

Abstract

BACKGROUND

Postoperative recurrence (POR) of Crohn’s disease (CD) is a major therapeutic challenge.

AIM

To evaluate ustekinumab (UST) for the POR management in CD in a Chinese multicenter cohort.

METHODS

This retrospective analysis included postoperative CD patients with ≥ 2 high-risk features receiving UST at four tertiary centers. Patients were stratified into a first-line group (biologic-naïve, n = 45) and a second-line group (prior biologic failure, n = 26). The primary outcome was endoscopic improvement (≥ a 1-point reduction in the Rutgeerts score). Multivariate generalized estimating equations (GEE) were used to identify therapeutic response predictors.

RESULTS

Overall, 81.7% achieved endoscopic improvement, with 36.6% achieving complete remission. Clinical remission reached 91.5% with a significant Harvey-Bradshaw Index reduction (5.5 ± 2.8 to 2.1 ± 1.7, P value < 0.001). The GEE multivariate analysis identified first-line UST therapy (OR = 1.727, 95%CI: 1.042-2.862, P = 0.034) and shorter disease duration (OR = 0.979, 95%CI: 0.967-0.992, P = 0.001) as independent predictors of endoscopic improvement. Significant improvements were observed in nutritional parameters (i.e., albumin and hemoglobin) and inflammatory markers (i.e., C-reactive protein and fecal occult blood positivity). Adverse events occurred in 9.9% of patients. These were all mild-to-moderate severity.

CONCLUSION

UST demonstrated significant effectiveness and a favorable safety profile for the management of POR in CD. These findings support its use as an effective therapeutic option, particularly under a first-line setting, and provide valuable evidence to optimize postoperative management strategies for CD recurrence in Chinese populations.

Key Words: Crohn's disease; Ustekinumab; Postoperative recurrence; Rutgeerts score; Endoscopic improvement

Core Tip: This multicenter study demonstrated that ustekinumab (UST) effectively addressed the management of postoperative recurrence in patients with Crohn's disease in China. Results showed high rates of endoscopic (81.7%) and clinical remission (91.5%) at 24 weeks. A multivariate analysis identified first-line UST use and shorter disease duration as independent predictors of superior outcomes. The favorable safety profile supports UST as a potent therapeutic option, particularly when initiated early as a first-line biologic strategy for postoperative management.

INTRODUCTION

Crohn's disease (CD) is a chronic, relapsing inflammatory bowel disease characterized by transmural inflammation that can affect any segment of the gastrointestinal tract[1]. While its global burden is significant, the rising incidence and prevalence of CD in China present a growing clinical challenge. Additionally, emerging evidence suggests potential distinctions in its etiopathogenesis and presentation compared to those of Western populations[2-4]. Intestinal resection remains a crucial intervention for the management of complications such as strictures or penetrating disease[1]. Population-based studies from Europe have indicated that five-year surgical rates range from 12%-27%, with 30%-50% of patients with CD eventually requiring surgery during their disease course[5]. Consistent with this pattern, a large Chinese cohort study reported that 41.9% of patients with CD underwent at least one CD-related surgery during a median follow-up of 102 months[6].

Surgery, is not curative. Postoperative recurrence (POR) represents a fundamental challenge in long-term CD management[7]. Endoscopic recurrence rates can reach 70%-90% within the first year after ileocolonic resection without prophylactic therapy[7]. The recurrence process typically begins with histological inflammation, progresses to endoscopic lesions, and may eventually manifest as clinical or surgical recurrence. Early ileocolonoscopy within 6-12 months after surgery is therefore crucial to identify endoscopic predictors of clinical recurrence and to adjust treatment strategies accordingly[8,9]. The Rutgeerts scoring system is a valuable risk stratification tool, and even mild ileal lesions predicting future clinical recurrence[7,10]. These remains establish an urgent need for effective prophylactic strategies in the postoperative setting. Anti-TNF-α agents have been a cornerstone for this purpose; but approximately one-fifth of patients with CD still experience primary non-response to anti-TNF therapy. Also, an additional 10%-15% experience treatment failure annually, creating a significant therapeutic gap[11].

Ustekinumab (UST) is a fully human monoclonal antibody that targets the shared p40 subunit of interleukin (IL)-12 and IL-23. It has emerged as a highly effective therapeutic for moderate-to-severe CD[12]. Its efficacy in inducing and maintaining remission in biologic-naïve and biologic-experienced patients has been firmly established in the landmark UNITI/IM-UNITI trials[12-14]. Consequently, UST is a promising candidate for POR mitigation. However, the existing body of literature regarding the application of UST, specifically in the postoperative setting, remains relatively scarce and heterogeneous[15-20]. This evidence gap is particularly pronounced in Chinese populations, where differences in genetic background, gut microbiota, and clinical practice may influence treatment outcomes[9]. Current studies are often limited by small sample sizes, single-center designs, or a lack of comprehensive multivariate analyses that examine the response predictors (Supplementary Table 1).

We therefore conducted this multicenter, retrospective cohort study with the primary objective of evaluating the real-world efficacy and safety of UST for POR management in a Chinese population. We aim to rigorously assess its impact on both clinical and endoscopic outcomes and to identify key patient- and disease-related factors associated with treatment success, thereby informing optimal patient selection and clinical decision-making.

MATERIALS AND METHODS

Study design

This multicenter, retrospective cohort study was conducted at four geographically diverse tertiary medical centers in China. These consisted of institutions from the north (the Seventh Medical Center of Chinese PLA General Hospital), the south (the Nanfang Hospital of Southern Medical University), the northeastern (the Shengjing Hospital of China Medical University), and the southwestern (the First Hospital of Chongqing Medical University). The study period spanning from October 2024 to January 2025. The study protocol received ethical approval (Approval No. S2024-093-01) and the informed consent was waived.

Eligible patients were adults (≥ 18 years) with CD who had underwent intestinal resection and initiated UST for postoperative management. The inclusion criteria were: (1) At least two high-risk POR factors (prior intestinal surgery, penetrating disease, active smoking, perianal disease, extensive lesions, or residual activities); (2) UST initiation ≥ 2 weeks post-surgery; (3) Completion of ≥ 3 UST doses; and (4) Available follow-up colonoscopy and clinical data approximately week 24 post-UST initiation. Patients were stratified into two groups based on their postoperative biologic treatment history: The first-line group, was defined as patients for whom UST was the first biologic agent initiated after the index surgery (regardless of the pre-operative biologic exposure), and the second-line group, was defined as patients who had received and failed at least one other biologic agent (e.g., anti-TNFs) after the index surgery prior to switching to UST. Patients with incomplete electronic medical records (EMRs) were excluded.

Treatment protocol and data collection

UST was administered according to the following standard regimen: An intravenous induction dose of approximately 6 mg/kg, followed by a 90 mg subcutaneous injection at week 8. Subsequent maintenance therapy involved subcutaneous 90 mg injections, with all patients in this cohort receiving their third dose at week 16, aligning with an eight-week interval. Treatment decisions, included the choice of UST as first- or second-line therapy, and they were made at the discretion of the treating gastroenterologist based on individual patient profiles.

Data were systematically extracted from the EMRs of the participating centers using a standardized case report form. The collected variables included the following: (1) Demographics and history: Age, sex, body mass index (BMI), smoking status, and comorbidities; (2) Disease characteristics: Disease duration, Montreal classification (i.e., age at diagnosis: A1 ≤ 16, A2 17-40, A3 > 40; location: L1/L2/L3/L4; and behavior: B1/B2/B3), history of perianal disease, and extraintestinal manifestations; (3) Treatment history: Number and timing of previous CD-related surgeries, prior exposure to biologics or immunomodulators, and the interval from the most recent surgery to the initiation of UST; and (4) Laboratory parameters: C-reactive protein (CRP), erythrocyte sedimentation rate (ESR), albumin, hemoglobin, and fecal occult blood test (FOBT) results at baseline and follow-up.

Outcome measures

The primary outcome was endoscopic improvement, that was defined as a reduction of at least one point in the Rutgeerts score from baseline to the week 24 follow-up colonoscopy. A centralized consensus review was conducted to minimize the inter-observer variability inherent to multicenter studies. De-identified endoscopic images and relevant clinical records were collated and reviewed jointly by a panel of senior inflammatory bowel disease specialists from the four participating centers (Tian F, Zhou XC, Jia Y, and Zhao XM) via a secure online conferencing platform. The panel verified the Rutgeerts scores according to standard definitions. In instances where the retrospective EMR description was ambiguous or differed from the image review, the final score was determined by the panel’s consensus. Secondary outcomes encompassed multiple dimensions of treatment responses: (1) Clinical efficacy, assessed through changes in the Harvey-Bradshaw Index (HBI) score, with clinical remission defined as HBI ≤ 4 and active disease as an HBI > 4; (2) Laboratory parameters that included changes in inflammatory markers (i.e., CRP and ESR), nutritional indicators (i.e., albumin and hemoglobin), and FOBT positivity; and (3) A safety profile, that was evaluated through the adverse events (AEs) incidence and characteristics documented during the follow-up period.

Statistical analysis

Statistical analyses were performed using SPSS (version 26.0, IBM Corp., Armonk, NY, United States) for data management and analyses, and R (version 4.2.0, R Foundation for Statistical Computing, Vienna, Austria) for visualization. Descriptive statistics was used to summarize the baseline characteristics, with continuous variables reported as means ± SD or medians (interquartile range), and categorical variables as n (%). Between-group comparisons utilized t-tests, Mann-Whitney U tests, χ² tests, or Fisher's exact tests as appropriate. Paired analyses (i.e., paired t-test/Wilcoxon signed-rank test for continuous variables; McNemar's test for categorical variables) assessed changes from baseline to week 24. Due to the strict inclusion criteria that was required for the available follow-up data, missing data for the primary endpoint were negligible; therefore, a complete case analysis was performed without imputation.

A multivariate analysis was performed using generalized estimating equations (GEE) to identify factors independently associated with the primary outcome of endoscopic improvement. The GEE model was selected specifically to account for the potential clustering of outcomes within patients from the same medical center. This is a crucial consideration for multicenter cohort studies[21]. The model utilized a binomial distribution, a logit link function, and an exchangeable working correlation structure. Variable selection was based on a combination of clinical relevance and statistical screening. Specifically, the model was adjusted for the following: (1) The full Montreal classification (i.e., age, location, behavior, and perianal disease) to characterize the disease phenotype; (2) Potential confounders that showed baseline imbalances between groups, namely, the disease duration and extraintestinal manifestations; and (3) The baseline endoscopic severity (i.e., the pre-treatment Rutgeerts score). The treatment line (first-line vs. second-line) was the primary exposure of interest. While variables such as time from surgery were considered, they were excluded from the final multivariate model to avoid multicollinearity with the treatment line. Results are presented as ORs with 95%CIs. A two-sided P value of < 0.05 defined statistical significance.

RESULTS

Baseline characteristics of study population

A total of 71 postoperative CD patients treated with UST were included in this multicenter retrospective analysis. Baseline characteristics stratified by treatment lines are detailed in Table 1. The cohort had a mean age of 35.4 ± 11.8 years with male predominance (74.6%). Significant differences emerged between first-line (63.4%, 45/71) and second-line (36.6%, 26/71) groups. The disease durations was longer in second-line patients (median 9.5 years vs 4.0 years, P < 0.001). Ileal disease (L1) was more common in second-line patients (46.2% vs 20.0%), while ileocolonic disease (L3) predominated in first-line patients (71.1% vs 42.3%, P = 0.045). Extraintestinal manifestations were more frequent with second-line therapy (23.1% vs 4.4%, P = 0.024). All of the second-line patients had prior biologic exposure vs 11.1% in the first-line patients (P < 0.001). Notably, the vast majority of patients received UST as biologic monotherapy during the post-operative follow-up period. Concomitant corticosteroids, immunomodulators, and 5-ASA were generally discontinued upon UST initiation to strictly evaluate its therapeutic efficacy, and thus were not analyzed as confounding variables. The surgery-to-UST interval was significantly shorter in first-line patients (median 2.0 months vs 16.5 months, P < 0.001, Figure 1).

Figure 1 Timing of ustekinumab initiation after surgery by treatment line. A: The distribution highlights the significantly shorter median time from surgery to ustekinumab (UST) initiation in the first-line treatment group compared to that of the second-line group; B: Bar chart showing the proportion of patients receiving early (≤ 6 months) vs late (> 6 months) UST initiation after intestinal resection, stratified by the treatment line (first-line vs second-line UST). UST: Ustekinumab.

Table 1 Baseline characteristics of patients with postoperative Crohn's disease treated with ustekinumab (n = 71).

Characteristic	Overall (n = 71)	First-line UST (n = 45)	Second-line UST (n = 26)	P value
Demographics
Age (years)	35.4 ± 11.8	34.2 ± 11.0	37.5 ± 12.9	0.260
Male sex	53 (74.6)	35 (77.8)	18 (69.2)	0.425
Occupation				0.763
Employee	35 (49.3)	22 (48.9)	13 (50.0)
Student	11 (15.5)	8 (17.8)	3 (11.5)
Unemployed	25 (35.2)	15 (33.3)	10 (38.5)
Comorbidities
Tuberculosis history	10 (14.1)	8 (17.8)	2 (7.7)	0.209
Hepatitis B history	3 (4.2)	3 (6.7)	0 (0.0)	0.248
Cancer history	2 (2.8)	1 (2.2)	1 (3.8)	0.690
Disease characteristics
Disease duration (years)	5.0 (2.0-10.0)	4.0 (2.0-6.0)	9.5 (5.3-17.0)	< 0.001
Montreal age classification				0.916
A1 (≤ 16 years)1	7 (9.9)	4 (8.9)	3 (11.5)
A2 (17-40 years)	51 (71.8)	33 (73.3)	18 (69.2)
A3 (> 40 years)	13 (18.3)	8 (17.8)	5 (19.2)
Montreal location				0.045
L1 (ileal)	21 (29.6)	9 (20.0)	12 (46.2)
L2 (colonic)	7 (9.9)	4 (8.9)	3 (11.5)
L3 (ileocolonic)	43 (60.6)	32 (71.1)	11 (42.3)
L4 (upper GI)	0 (0.0)	0 (0.0)	0 (0.0)
Montreal behavior				0.727
B1 (non-stricturing, non-penetrating)	6 (8.5)	3 (6.7)	11 (24.4)
B2 (stricturing)	47 (66.2)	31 (68.9)	16 (61.5)
B3 (penetrating)	18 (25.4)	11 (24.4)	7 (26.9)
Perianal disease	24 (33.8)	16 (35.6)	8 (30.8)	0.681
Extraintestinal manifestations	8 (11.3)	2 (4.4)	6 (23.1)	0.024
Smoking history	9 (12.7)	8 (17.8)	1 (3.8)	0.087
Alcohol history	4 (5.6)	4 (8.9)	0 (0.0)	0.153
History of biologic use
Previous biologic exposure				< 0.001
None	41 (57.7)	40 (88.9)	1 (3.8)
Infliximab only	20 (28.2)	4 (8.9)2	16 (61.5)
Adalimumab only	3 (4.2)	0 (0.0)	3 (11.5)
Infliximab and adalimumab	4 (5.6)	1 (2.2)2	3 (11.5)
Infliximab and vedolizumab	2 (2.8)	0 (0)	2 (7.7)
Infliximab, adalimumab and vedolizumab	1 (1.4)	0 (0.0)	1 (2.8)
Treatment timing				< 0.001
Time from surgery to UST (months)	1.0 (1.0-17.0)	2.0 (1.0-6.0)	16.5 (6.0-36.0)

¹The “A1” classification in the Montreal system refers to the “age at diagnosis” (≤ 16 years), not the patient’s age at enrollment. These seven patients were diagnosed with Crohn's disease during childhood but were adults (≥ 18 years) when they were enrolled in this study for ustekinumab treatment.

²Five patients in the first-line group had exposure to infliximab prior to surgery but received ustekinumab as their first biologic agent during the postoperative period.

Continuous variables are presented as mean ± SD, and non-continuous variables are presented as medians (quartile). Categorical variables are presented as n (%). P values from t-test, χ² test, Fisher's exact tests or non-parametric tests, as appropriate. UST: Ustekinumab; GI: Gastrointestinal.

Comprehensive efficacy of UST treatment

UST treatment resulted in significant improvements for the total cohort (n = 71). Their endoscopic evaluations demonstrated substantial improvements, with the mean Rutgeerts score decreasing from 2.7 ± 1.1 to 1.1 ± 1.1 (P < 0.001). Notably, 81.7% of patients achieved endoscopic improvement, including 36.6% who attained complete endoscopic remission (Rutgeerts i0; Figure 2A). Clinically, the UST treatment induced a robust response. This was evidenced by a pronounced reduction in the mean HBI score from 5.5 ± 2.8 to 2.1 ± 1.7 (P < 0.001, Figure 2B).

Figure 2 Changes in Rutgeerts score, Harvey-Bradshaw Index, and endoscopic improvement by subgroup following ustekinumab treatment. A and B: Box plots that demonstrated the significant reduction in both endoscopic Rutgeerts scores and clinical Harvey-Bradshaw index (HBI) scores following ustekinumab (UST) treatment. The statistical significance was determined using paired analyses (P < 0.001 for Rutgeerts score; P < 0.001 for HBI score); C: Histogram illustrating endoscopic improvement rates (defined as ≥ 1-point reduction in the Rutgeerts score) across key patient subgroups, including the treatment line (i.e., first-line vs second-line UST) and disease characteristics based on the Montreal classification. The overall endoscopic improvement rate was 81.7% (58/71 patients). Together, these panels indicate substantial improvement in endoscopic and clinical disease activity following UST treatment, with differential responses observed across the patient subgroups. UST: Ustekinumab.

Efficacy outcomes were further stratified by treatment lines according to the baseline differences (Table 2). In the first-line group (postoperative biologic-naïve), the UST treatment induced profound improvements across all clinical, endoscopic, and laboratory domains. The mean Rutgeerts score decreased significantly from 2.7 ± 1.1 to 0.9 ± 1.1 (P < 0.001), with an endoscopic improvement rate of 86.7% (39/45). Notably, 77.8% (35/45) of patients in this group achieved or maintained endoscopic remission (Rutgeerts < i2) at week 24. Clinically, the mean HBI score dropped from 5.6 ± 3.0 to 1.8 ± 1.5 (P < 0.001), and the proportion of patients with active disease fell dramatically from 71.1% to 4.4% (P < 0.001). Furthermore, all of the laboratory parameters, including inflammatory markers (i.e., CRP and ESR) and nutritional indices (i.e., albumin, hemoglobin and BMI), showed statistically significant normalization (all P < 0.01). In the second-line group (rescue therapy), UST also demonstrated effectiveness, particularly in endoscopic and clinical disease control. This result was despite a more refractory disease phenotype. The mean Rutgeerts score decreased from 2.7 ± 1.0 to 1.4 ± 1.1 (P < 0.001), and 73.1% (19/26) of patients achieved endoscopic improvement. The HBI score significantly improved from 5.2 ± 2.5 to 2.8 ± 2.0 (P < 0.001), with active disease rates reducing from 73.1% to 15.4% (P < 0.001). However, unlike the first-line group, improvements in systemic inflammatory and nutritional biomarkers (i.e., CRP, ESR, albumin and emoglobin) did not reach statistical significance (P > 0.05) in this subgroup. This result likely reflected the higher inflammatory burden and chronicity associated with second-line patients. The outcomes at week 24 were then compared, and the first-line group achieved significantly lower absolute HBI scores (1.8 vs 2.8, P = 0.019) and Rutgeerts scores (0.9 vs 1.4, P = 0.039) compared to the second-line group. The endoscopic improvement rate was numerically higher in the first-line group (86.7% vs 73.1%), but this difference was not statistically significant (P = 0.154).

Table 2 Efficacy outcomes before and after ustekinumab treatment (n = 71).

Parameter	First-line group (n = 45)		Second-line group (n = 26)		P value4
	Before UST	After UST	Before UST	After UST
Clinical outcomes
HBI score	5.6 ± 3.0	1.8 ± 1.5	5.2 ± 2.5	2.8 ± 2.0	0.019
P value1		< 0.001		< 0.001
Active disease2	32 (71.1)	2 (4.4)	19 (73.1)	4 (15.4)	0.110
P value1		< 0.001		< 0.001
Endoscopic outcomes
Rutgeerts score	2.7 ± 1.1	0.9 ± 1.1	2.7 ± 1.0	1.4 ± 1.1	0.039
P value1		< 0.001		< 0.001
Endoscopic improvement3	N/A	39 (86.7)	N/A	19 (73.1)	0.154
Endoscopic remission (< i2)	N/A	35 (77.8)	N/A	15 (57.7)	0.074
BMI (kg/m2)	18.7 ± 2.8	20.1 ± 3.0	20.7 ± 4.2	21.8 ± 3.7	0.037
P value1		< 0.001		0.003
Laboratory parameters
CRP (mg/L)	7.1 ± 9.2	3.7 ± 3.8	7.8 ± 11.1	4.6 ± 4.0	0.378
P value1		0.008		0.152
ESR (mm/hour)	20.2 ± 16.7	12.3 ± 13.7	19.1 ± 18.1	13.3 ± 17.2	0.558
P value1		0.005		0.124
Albumin (g/L)	40.2 ± 14.7	43.7 ± 14.6	40.7 ± 4.0	41.2 ± 3.6	0.438
P value1		< 0.001		0.615
Hemoglobin (g/L)	110.5 ± 21.8	125.2 ± 21.9	119.1 ± 21.1	125.2 ± 25.8	0.734
P value1		< 0.001		0.088
Fecal occult blood positive	29 (64.4)	9 (20.0)	14 (53.8)	8 (30.8)	0.306
P value1		< 0.001		0.092

¹Paired t-test or McNemar test comparing week 24 to the baseline within each group.

²Active disease: Harvey-Bradshaw Index > 4.

³Endoscopic improvement was defined as a decrease in the Rutgeerts score of ≥ 1 point.

⁴Comparison refers to the difference between the first-line and second-line groups at week 24 (or the rate of response) using independent t-tests or χ² tests.

Data are presented as mean ± SD for continuous variables or n (%) for categorical variables. UST: Ustekinumab; HBI: Harvey-Bradshaw Index; BMI: Body mass index; ESR: Erythrocyte sedimentation rate; CRP: C-reactive protein; N/A: Not applicable.

Endoscopic improvement predictors

A GEE model was constructed to identify the factors independently associated with endoscopic improvement (i.e., decrease in Rutgeerts score), adjusting for confounders such as disease duration and extraintestinal manifestations (Table 3). Patients who received the first-line UST therapy had significantly higher odds of achieving endoscopic improvement compared to those who received it as a second-line agent (OR = 1.727, 95%CI: 1.042-2.862, P = 0.034). Furthermore, several Montreal classification components were significantly associated with the outcome. Patients in the A2 (OR = 11.149, 95%CI: 1.237-100.482, P = 0.032) and A3 (OR = 2.953, 95%CI: 1.637-5.329, P < 0.001) age groups had significantly higher of improvement odds compared to that of the A1 group. Regarding disease behavior, both stricturing (B2; OR = 3.764, 95%CI: 1.031-13.742, P = 0.045) and penetrating (B3; OR = 11.692, 95%CI: 1.107-123.521, P = 0.041) phenotypes were positive predictors compared to the non-stricturing, non-penetrating (B1) phenotype. Conversely, a longer disease duration was negatively associated with the likelihood of endoscopic improvement (OR = 0.979 per year, 95%CI: 0.967-0.992, P = 0.001). The disease location (L1-L3), perianal disease presence, extraintestinal manifestations, and the pre-treatment Rutgeerts score were not independent predictors in this model. The endoscopic improvement rates across these key subgroups are further illustrated in Figure 2C.

Table 3 Multivariate generalized estimation equations analysis of factors associated with a Rutgeerts score improvement (n = 71).

Variable	β	Robust SE	Wald χ²	OR (95%CI)	P value
Intercept	-3.709	1.351	7.532	0.025 (0.002-0.346)	0.006
Treatment line
First-line UST therapy (vs second-line)	0.546	0.258	4.498	1.727 (1.042-2.862)	0.034
Montreal classification
Age A2 (17-40 years) vs A1 (≤ 16 years)	2.411	1.122	4.621	11.149 (1.237-100.482)	0.032
Age A3 (> 40 years) vs A1 (≤ 16 years)	1.083	0.301	12.927	2.953 (1.637-5.329)	< 0.001
Location L2 (colonic) vs L1 (ileal)	1.171	1.014	1.334	3.225 (0.442-23.527)	0.248
Location L3 (ileocolonic) vs L1 (ileal)	1.342	1.022	1.725	3.827 (0.517-28.348)	0.189
Behavior B2 (stricturing) vs B1 (non-stricturing, non-penetrating)	1.325	0.661	4.023	3.764 (1.031-13.742)	0.045
Behavior B3 (penetrating) vs B1 (non-stricturing, non-penetrating)	2.459	1.203	4.179	11.692 (1.107-123.521)	0.041
Perianal disease (yes vs no)	-1.208	1.165	1.074	0.299 (0.030-2.934)	0.300
Other clinical factors
Extraintestinal manifestations (yes vs no)	0.613	0.459	1.783	1.847 (0.750-4.544)	0.182
Disease duration (years)	-0.021	0.007	10.431	0.979 (0.967-0.992)	0.001
Pre-UST Rutgeerts score	0.607	0.331	3.369	1.835 (0.960-3.509)	0.066

Model information: (1) Dependent variable: Rutgeerts score improvement (0 = no, 1 = yes); (2) Analysis method: Generalized estimating equations; and (3) Reference categories: Second-line, Age A1, location L1, behavior B1, no perianal disease and no extraintestinal manifestations. UST: Ustekinumab.

Safety and tolerability

The UST safety profile in this postoperative CD population was favorable. AEs over the study period were reported in 7 out of 71 patients (9.9%). The incidence was comparable between the first-line and second-line treatment groups (11.1% vs 7.7%; Figure 3A). The AEs spectrum included upper respiratory infection (2.8%, 2/71), renal impairment (2.8%, 2/71), liver function abnormality (2.8%, 2/71), and hand numbness (1.4%, 1/71; Figure 3B). All of the reported AEs were graded as mild to moderate in severity. Critically, no serious AEs, opportunistic infections, hepatitis B reactivations, malignancies, or treatment discontinuations attributable to AEs were observed.

Figure 3 Ustekinumab safety profile in postoperative Crohn's disease. A: Bar chart comparing the adverse event (AE) rates between the first-line and second-line ustekinumab treatment groups, showing comparable safety profiles; B: Pie chart detailing the distribution of AE types among the seven reported events, including upper respiratory infection, renal impairment, liver function abnormality, and hand numbness. All of the AEs were mild to moderate in severity. UST: Ustekinumab.

DISCUSSION

POR in CD remains a paramount challenge in clinical gastroenterology[7]. UST has established its efficacy in moderate-to-severe CD[12-14], but robust data on its application in the postoperative setting, particularly among Chinese populations, have been notably limited. Our multicenter, retrospective cohort study provided compelling evidence that UST is a highly effective and safe therapeutic option for POR management in Chinese patients. The significant improvements observed in endoscopic activity, clinical symptoms, and laboratory parameters, coupled with a favorable safety profile, position UST as a pivotal agent in the postoperative management arsenal. Notably, the vast majority of our patients received UST as biologic monotherapy without concomitant corticosteroids or immunomodulators during follow-up, the observed efficacy can be directly attributed to UST. This approach minimized the confounding effects of co-medication. Furthermore, our multivariate analysis delineated specific patient profiles that were most likely to benefit from this therapy, offering valuable insights for treatment personalization.

The demographic and disease characteristics of our cohort aligned with the recognized phenotype of CD that requires surgical intervention[17-20]. Our cohort existed predominantly of young males with a high prevalence of ileocolonic involvement (L3, 60.6%) and stricturing behavior (B2, 66.2%). The significant difference in disease duration and prior biologic exposure between the first-line and second-line groups accurately reflected real-world clinical practice, where patients progressed to second-line biologics often have a longer, more treatment-refractory disease course[18,19,22]. Our findings confirmed and extended the observations from smaller European studies regarding postoperative UST use[15-19]. The dramatic reduction in the median Rutgeerts score from 2.7 to 1.1, with 81.7% of patients achieving endoscopic improvement, is a cornerstone finding. This objective endoscopic efficacy is powerfully corroborated by parallel, statistically significant improvements in clinical activity (HBI score reduction from 5.5 to 2.1) and key systemic biomarkers (i.e., CRP, ESR, albumin and hemoglobin). This multi-dimensional response underscored the capacity of UST to fundamentally alter the postoperative disease course by addressing the underlying inflammatory factors.

A critical contribution of our multivariate GEE analysis was the identification of clinically relevant factors that can predict of endoscopic improvement following UST treatment in postoperative CD. Our data revealed that first-line UST therapy was an independent positive predictor of the endoscopic response. This supports a “top-down” therapeutic strategy[23] in the postoperative setting, where early intervention with high-efficacy biologics may fundamentally alter disease progression and reduce complications. It is important to interpret this finding in the context of clinical practice, where the “first-line” status is intrinsically linked to earlier intervention. The first-line patients in our cohort had significantly shorter disease durations and shorter intervals from surgery to treatment initiation compared to these of the second-line patients. There was the inherent collinearity between biologic-naïve status and early timing, the superior outcomes observed in the first-line group likely reflected the synergistic benefit of an early, first-line intervention strategy, rather than the biologic-naïve status alone. The observed effect size was consistent with the emerging real-world evidence that suggests that UST demonstrated robust efficacy in biologic-naïve patients[23,24]. This result was potentially due to an unaltered immune landscape that responded more effectively to the IL-12/23 pathway inhibition. Conversely, we found a negative association between a longer disease duration and endoscopic improvement. This reinforces the concept of a “therapeutic window of opportunity” early in the disease course[22], underscoring the importance of timely intervention after surgery[7,19]. Each additional year of disease duration was associated with a 2.1% reduction in the odds of endoscopic improvement. This highlighted how cumulative bowel damage and disease chronicity may progressively diminish treatment responsiveness[7]. This finding strongly advocated for the early UST introduction during the postoperative management algorithm to maximize the therapeutic benefit.

One of the most intriguing findings was the superior endoscopic response observed in patients with stricturing (B2) and penetrating (B3) disease behaviors compared to the purely inflammatory (B1) phenotype. This result is seemingly counterintuitive, but this may be explained by several factors. First, patients with complicated phenotypes often undergo more definitive surgical resections, which, when combined with subsequent biologic prophylaxis, may lead to more pronounced endoscopic improvements[7,23]. Second, it is plausible that UST is particularly effective at suppressing specific inflammatory pathways (e.g., IL-12/23 mediated Th1/Th17 responses) that drive these complicated phenotypes. Some studies have suggested that UST may promote transmural healing, and this could be particularly relevant in phenotypes characterized by deeper tissue involvement[15,16,18]. These findings challenged conventional perceptions and warrants further investigation into the phenotype-specific efficacy of UST.

Regarding patient age, our analysis indicated that adult patients (particularly A2, 17-40 years) demonstrated better endoscopic outcomes compared to the pediatric-onset (A1) group. This result may reflect differences in disease biology, medication compliance, or the distinct immune profiles of pediatric-onset CD[25-27]. However, the exceptionally wide confidence interval for the A2 group necessitates a cautious interpretation and likely reflects the limited subgroup sample size. Future studies with larger pediatric and A1 populations are required to validate this relationship. Notably, factors such as disease location, perianal disease, and extraintestinal manifestations did not significantly influence endoscopic outcomes in our cohort. This suggested that the efficacy of UST may be broadly applicable across various disease localizations. This would reinforce its utility in a heterogeneous patient population. The lack of a negative impact from perianal disease is particularly encouraging, as these patients often present a therapeutic challenge[28].

The favorable safety profile observed in our cohort was characterized by a 9.9% incidence of predominantly mild AEs. Crucially, all of the reported AEs were classified as mild to moderate (grade 1 or 2) according to the Common Terminology Criteria for Adverse Events, and none required therapy discontinuation. The absence of serious safety signals aligned with the established UST long-term safety data[15]. This reassuring risk-benefit profile is particularly relevant for a prophylactic or treatment strategy intended for long-term use in a postoperative population that is, by definition, in a state of clinical remission.

Our study has several limitations inherent to its retrospective design, including potential selection bias and unmeasured confounding factors, despite adjustment using the GEE. While this multicenter cohort is a significant real-world dataset in the context of postoperative UST use in China, the sample size remains relatively modest. Consequently, the statistical power to detect differences within smaller subgroups was limited, and the findings regarding specific phenotypes with small numbers (i.e., reflected by wide confidence intervals in the multivariate analysis) should be interpreted as exploratory. Second, detailed surgical variables, such as the specific type of anastomosis (e.g., side-to-side vs end-to-end), were not consistently available in the retrospective records. However, all of the surgeries were performed at tertiary referral centers following the standard bowel-sparing principles, minimizing the surgical technique as a major confounder. Additionally, as noted above, the first-line therapy group represented an early intervention strategy, and most patients had established endoscopic recurrence at baseline (mean Rutgeerts score of 2.7). This distinguished this study from a pure primary prophylaxis setting beginning at i0. Finally, the 24-week follow-up was sufficient to demonstrate the early endoscopic efficacy, but it preclude an assessment of long-term durability and impact on surgical recurrence rates.

Based on these findings, we propose the following for clinical practice. First, UST could be considered as a first-line biologic option for high-risk postoperative CD patients, given its superior efficacy in this position and excellent safety profile. Second, treatment decisions can be individualized based on the predictors of response. Biologic-naïve patients with shorter disease durations and those with stricturing or penetrating disease behavior appeared to derive the greatest benefit from UST. Third, the high efficacy rate supported a proactive postoperative management strategy, that initiates UST early after surgery and confirms the endoscopic response thereafter. Future research should prioritize prospective, randomized controlled trials that directly compare UST with anti-TNF agents under a postoperative setting. Additionally, efforts should focus on the development of integrated clinical-serological predictive models to further optimize patient selection and validate the predictors identified in this study.

CONCLUSION

In this multicenter retrospective study of Chinese patients with postoperative CD, UST demonstrated promising efficacy for POR management, as evidenced by significant improvements in endoscopic Rutgeerts scores, clinical activity, and relevant laboratory parameters, along with a favorable safety profile. These findings support its consideration as a valuable therapeutic option in the postoperative setting, particularly as a first-line biologic agent. While the retrospective design and sample size limitations require cautious interpretation, this study provides valuable real-world evidence for the application UST in this clinical context. This warrants further validation through larger prospective studies.

ACKNOWLEDGEMENTS

We extend our special gratitude to Professor Shi-Rong Li, Chief Physician of the Department of Gastroenterology at the Seventh Medical Center of the Chinese PLA General Hospital, for his guidance and assistance throughout this study. We would like to extend our special thanks to Yu-Jie Wang (Assistant Researcher, Medical Data Science Center, Beijing Tsinghua Changgung Hospital) for providing professional guidance and reviewing the statistical analysis of this study.