Leucine-rich alpha-2 glycoprotein as a predictor of primary non-response to anti-TNF-α therapy in biologic-naïve Egyptian IBD patients

Abstract

Background

Anti-tumor necrosis factor-α (anti-TNF-α) agents are a cornerstone in inflammatory bowel disease (IBD) therapy, yet primary non-response remains a significant practical challenge. Leucine-rich alpha-2 glycoprotein (LRG) has been recognized as a promising marker for disease activity. This study aimed to evaluate the predictive significance of pre-treatment serum LRG for response to anti-TNF-α therapy in biologic-naïve Egyptian IBD patients.

Methods

In this prospective cohort study, 100 biologic-naïve IBD adult patients (50 Crohn’s disease [CD], 50 ulcerative colitis [UC]) and 100 healthy controls were enrolled. Patients received induction therapy with adalimumab or infliximab. Clinical, biochemical, and endoscopic evaluations were conducted at baseline and at week 24. Response was defined by clinical indices and endoscopic improvement, while biochemical normalization was evaluated as a secondary, supportive parameter.

Results

IBD patients had significantly elevated baseline LRG levels compared to the healthy controls (p < 0.001). Responders’ baseline LRG was substantially lower than that of non-responders in both UC (26.53 vs. 34.88 µg/mL; p = 0.008) and CD (26.03 vs. 35.07 µg/mL; p = 0.006). Receiver operating characteristic analysis revealed a cut-off of > 29 µg/mL for predicting non-response, yielding sensitivities of 74.2% and 80.0% with specificities of 69.57% and 65.71% for UC and CD, respectively and negative predictive values of 85.7% for UC and 88.5% for CD. Multivariate regression confirmed baseline LRG as an independent predictor of non-response.

Conclusions

Baseline serum LRG levels > 29 µg/mL demonstrated moderate discriminatory ability for predicting primary non-response to anti-TNF-α therapy in Egyptian IBD patients. LRG may serve as a useful adjunctive biomarker for pre-treatment risk estimation and recognizing patients who may require alternative therapeutic strategies.

Introduction

Inflammatory bowel disease (IBD), which encompasses both ulcerative colitis (UC) and Crohn’s disease (CD), is a chronic illness characterized by gastrointestinal tract inflammation. It is distinguished by its recurring pattern of exacerbations and remissions. Around the world, the number of instances is growing, with 4.9 million individuals affected, leading to escalating healthcare expenses [1]. The data that is currently available indicates that the burden of IBD in Egypt is rising, despite the lack of precise population-based prevalence estimates. Elbadry et al. [2], reported the largest nationwide cohort (1,104 patients), with ulcerative colitis accounting for 81% of cases. While Hamad et al. [3], reported comparable epidemiological patterns in the Nile Delta with higher prevalence among rural residents. IBD arises from intricate interactions among environmental stressors, genetic susceptibility, and inappropriate immunological reactions directed against the gut microbiota. The intricacy of this pathogenesis renders it impossible to adopt an efficient root-cause therapy or a single, definitive diagnostic modality [4].

Adopting a “treat-to-target” approach, where the primary aim is to attain endoscopic mucosal healing (MH), has replaced older IBD therapy approaches, leading to enhanced long-lasting outcomes and a decrease in the need for surgeries. Per the STRIDE-II guidance, the principal long-term therapy objectives for individuals with IBD include establishing clinical recovery, ensuring endoscopic healing, upgrading life quality, and preventing disabilities [5].

The management of moderate-to-severe IBD has been revolutionized by biologic medicines, especially anti-tumor necrosis factor-alpha (anti-TNF-α) medications like infliximab and adalimumab [6]. Despite their efficacy, therapeutic failure remains a significant clinical challenge. Approximately 30% of recipients experience primary non-response (PNR) through induction, and as much as 50% of initial responders forfeit response later on (“secondary” loss of response) [7]. Additionally, the use of anti-TNF-α is expensive in low resource communities. It also carries risks of infusion reactions, infections, hepato-splenic T-cell lymphoma, demyelinating disease, cardiomyopathy worsening, paradoxical psoriasis, liver dysfunction, and autoimmune diseases like systemic lupus erythematosus [8].

Consequently, predictive biomarkers are desperately needed to detect IBD patients’ subsets that have a higher chance of responding. To prevent useless exposure to immunosuppression and lower healthcare costs, it is crucial to predict anti-TNF response. Moreover, ileocolonoscopy remains the benchmark for appraising disease activity [9]; however, it is invasive, expensive, and time-consuming for periodic monitoring [10].

Although clinical phenotypes (duration, location, and severity), biochemical markers (albumin, C-reactive protein (CRP), and haemoglobin), and therapeutic background (past surgery or biologic exposure) have been assessed as predictors of anti-TNF outcomes, they are frequently imprecise [11]. In this context, leucine-rich alpha-2 glycoprotein (LRG) has recently emerged as a potentially useful biomarker [12, 13].

Neutrophils, intestinal epithelial cells, hepatic cells, and macrophages synthesize LRG, a fifty-kDa glycoprotein, when exposed to proinflammatory cytokines involving IL-6, TNF-α, IL-1β, IL-22, and IL-6. Since LRG biosynthesis is not reliant just on IL-6, it can react to a wider range of inflammatory pathways than CRP [14, 15].

Additionally, serum LRG concentrations in IBD populations may be associated with disease-related activity, alongside endoscopic activity. It’s still questionable if baseline LRG levels can foretell endoscopic and clinical outcomes and contribute to the selection of optimal biologics for IBD sufferer [14].

Notwithstanding these advances, limited information exists regarding the prognostic utility of LRG in populations of the Middle East. Therefore, this study sought to analyze the predictive validity of serum LRG in forecasting response to anti-TNF-α therapy among Egyptian naive-to-biologics IBD patients.

Materials and methods

Study design

This prospective cohort study carried out at the Department of Hepatology, Gastroenterology, and Infectious Diseases; Kafrelsheikh University Hospital, a tertiary referral centre at Kafr EL-Sheikh governorate, Egypt. Participants were recruited from May 2024 to May 2025, and each participant was monitored for an additional 6 months following recruitment. The study protocol was in strict accordance with the ethical principles described in the 2013 revision of the 1975 declaration of Helsinki and authorized by the Kafrelsheikh University, Faculty of Medicine’s Institutional Review Board (IRB) (IRB NO: KFSIRB200-169). Before participating, all individuals provided their written informed consent.

Study population

This study enrolled 100 adult patients consecutively, ranging in their age from 18 to 65 years, who had been diagnosed with IBD, which includes UC and CD. Standard clinical, biochemical, endoscopic, and histological criteria were used to make the diagnosis [16].

The criteria for inclusion in the present study were patients with active luminal disease who failed on conventional therapy and required biologic treatment; patients who were biologic-naïve, meaning they had no prior exposure to anti-TNF agents, anti-integrins, or anti-interleukins; and patients deemed suitable for therapy with anti-TNF agents.

The following were the exclusion criteria: indeterminate colitis, severe hepatobiliary or renal dysfunction, history of malignancy or other gastrointestinal diseases (e.g., celiac disease), active severe infection (e.g., tuberculosis, sepsis), pregnancy, and another systemic autoimmune disease, like rheumatoid arthritis that may independently alter serum LRG levels. To determine the baseline biomarker reference ranges, an age and sex-matched control group of healthy volunteers recruited from prospective blood donors and athletes undergoing routine check-ups with normal laboratory investigation and imaging with no prior history of autoimmune or gastrointestinal disorders were enrolled.

Sample size

The sample size estimation was calculated by using G*Power software version 3.1.9.7 (Heinrich-Heine-Universität Düsseldorf, Düsseldorf, Germany). Assuming the effect size of the predictive value of LRG to is 0.15, using alpha error of 0.05. and study power of 80% across two groups with two measurements [17]. The minimum required sample size is 90 patients for each group. Considering for 10% dropout ratio, the sample increased to 100 for each group. As a result, 100 IBD patients (50 UC and 50 CD cases) and an equivalent cohort of 100 healthy controls made up our final enrolment.

Data gathering and follow-up

For all enrolled patients, data on sociodemographic factors such as age, sex, city, in addition to medical details like the type of IBD (UC versus CD), body mass index (BMI), disease behaviour, severity, lesion location, illness duration, and prescribed medications, were collected prospectively. Clinical, biochemical, and endoscopic evaluations were conducted at the start (week 0) and again at week 24.

1. Clinical activity was evaluated using the Harvey–Bradshaw Index (HBI) for CD [18]and the Partial Mayo Score (pMayo) [19] for UC.

2. Laboratory assessments and biomarker evaluations involved collecting peripheral venous blood samples from all participants at the start and at week 24. Standard automated laboratory methods were used to analyze routine haematological and biochemical parameters, such as complete blood count (CBC), CRP, serum albumin, and fecal calprotectin. Serum LRG was obtained by centrifuging the serum at 2000–3000 rpm for 20 min within an hour of collection. The samples were then aliquoted and stored at either − 20 °C or -80 °C. LRG levels were measured using a commercially available Enzyme-Linked Immunosorbent Assay kit (code number DZE201121907, SUNRED, China) in compliance with the manufacturer’s instructions.

3. Endoscopic and histologic evaluation: At the start point and again at week 24, ileocolonoscopy was performed using high-definition video colonoscopes (Olympus, Tokyo, Japan) with narrow-band imaging (NBI) potential. Before the procedure, bowel preparation involved a polyethylene glycol-based electrolyte solution. The Simple Endoscopic Score for Crohn’s Disease (SES-CD) [20] was used for patients with CD, while the Mayo Endoscopic Subscore (MES) was applied to UC cases [21]. For a more detailed examination of the mucosa in UC cases, the Ulcerative Colitis Endoscopic Index of Severity (UCEIS) [22] was employed. Additionally, The Montreal classification was applied [23]. Histological scoring: Biopsy samples were taken from the most inflamed sections of the bowel. A senior gastrointestinal pathologist, who was blinded to the study, assessed histological activity using the Geboes score (GS) for UC cases [24].

Treatment protocol

Induction therapy was administered to patients using either infliximab (5 mg/kg IV at weeks 0, 2, and 6, with further maintenance doses every eight weeks) or adalimumab (160 mg subcutaneously at week 0, 80 mg at week 2, and 40 mg every two weeks). The selection between either infliximab or adalimumab was guided by national IBD management protocol, considering both clinical viability and patient preference for the mode of administration. Since a large previous literature shows that infliximab and adalimumab have acceptable therapeutic efficacy in biologic-naïve groups [25].

Response definition [5]

Patients were classified at week 24 into two groups (Responders and Non-responders) according to a composite of clinical and endoscopic criteria, consistent with the STRIDE-II treat-to-target recommendations [5].

Responders

Patients who achieved both clinical response and endoscopic response at week 24. Clinical response was defined as a decrease in the Harvey–Bradshaw Index (HBI) of ≥ 3 points for Crohn’s disease or a reduction in the pMayo score of ≥ 30% (or ≥ 1 point) for ulcerative colitis while endoscopic response was defined as a reduction of more than 50% in the Simple Endoscopic Score for Crohn’s Disease (SES-CD) for CD or in the Ulcerative Colitis Endoscopic Index of Severity (UCEIS) for UC.

Non-responders

Patients who did not meet the above combined clinical and endoscopic response criteria.

Biochemical improvement (normalization of CRP to under upper limit of normal (< 6 mg/l) and reduction in fecal calprotectin) was assessed as a supportive measure but was not required for definition of response. This definition ensures that the primary outcome remains independent of the inflammatory biomarkers under investigation.

Statistical analysis

Statistical analyses were performed with IBM SPSS software version 27.0 (Armonk, NY: IBM Corp, released 2020). The Kolmogorov-Smirnov test determined the normality of quantitative data, which were reported as mean ± standard deviation (SD) for normally distributed continuous variables or median (interquartile range [IQR]/range) for non-normally distributed data. Categorical data were expressed as counts (%). Differences between groups were analyzed using the Chi-square test, with Fisher’s Exact or Monte Carlo adjustments used when expected frequencies fell below 5 in > 20% of cells.

For quantitative variables comparing two independent groups (i.e., Responders vs. Non-responders), the student’s t-test was utilized for parametric data, and the Mann-Whitney U test was applied for non-parametric data. When comparing quantitative variables across more than two study groups (such as comparing UC patients, CD patients, and healthy controls), we used the Welch ANOVA test followed by the Games-Howell post hoc test for parametric data with unequal variances, and the Kruskal-Wallis test for non-parametric data.

Receiver operating characteristic (ROC) analysis was used to evaluate the diagnostic performance of LRG in terms of specificity, sensitivity, Negative Predictive Value (NPV), Positive Predictive Value (PPV), and Area Under the Curve (AUC).

For logistic regression analyses, univariate logistic regression was performed separately for the ulcerative colitis and Crohn’s disease cohorts to identify factors associated with primary non-response. Variables demonstrating a significant association in the univariate analysis (p < 0.05), together with variables considered clinically relevant a priori, were considered for inclusion in the multivariable logistic regression models. Given the modest sample size and limited number of primary non-response events, the number of covariates included in each multivariable model was deliberately restricted in accordance with the recommended events-per-variable principle to minimize overfitting and ensure model stability. The final multivariable models included the most clinically relevant variables while maintaining an appropriate events-per-variable ratio. Adjusted odds ratios (ORs) with 95% confidence intervals (CIs) were reported, and a two-sided p value < 0.05 was considered statistically significant.

Results

Patient characteristics

Totally 122 cases were observed over the study period. Out of them, twenty-two patients were excluded. Ten patients did not continue their laboratory and endoscopy follow-ups, and twelve patients experienced side effects from anti-TNF that prevented them from continuing their therapy (such as psoriasiform skin lesions, severe infusion reactions, and reactivation of tuberculosis).

Among 200 subjects in the study, 100 healthy subjects, 50 UC patients, and 50 CD patients, demographically, the cohorts were well comparable; neither the age distribution (p = 0.241) nor the sex distribution (p = 0.582) showed statistically significant discrepancies. Both CD (28.74 ± 10.90 µg/mL) and UC cases (29.20 ± 8.67 µg/mL) had significantly higher initial serum LRG values than controls (7.0 ± 1.29 µg/mL; p < 0.001) (Table 1). A prominent rise in this biomarker in active IBD was indicated by the lack of a significant difference in baseline LRG levels between the UC and CD groups (p = 0.942).

Table 1.

Comparison between the three studied groups according to demographic data and LRG

	Ulcerative colitis (n = 50)	Crohn's disease (n = 50)	Control (n = 100)	P
Sex
Male	23 (46.0%)	26 (52.0%)	55 (55.0%)	0.582
Female	27 (54.0%)	24 (48.0%)	45 (45.0%)
Age (years)	32.50 (26.0 – 42.0)	39.0 (18.0 – 45.0)	37.0 (25.50–48.50)	0.241
LRG	29.20 ± 8.67	28.74 ± 10.90	7.0 ± 1.29	>0.001*
Sig. bet. groups	p1=0.942, p2>0.001*, p3>0.001*

LRG leucine-rich alpha-2 glycoprotein, IQR Inter-quartile range, SD Standard deviation

p: p-value for comparing the three studied groups

p₁: p-value for comparing Ulcerative colitis and Crohn's disease groups

p₂: p-value for comparing Ulcerative colitis and Control groups

p₃: p-value for comparing Crohn's disease and Control groups

*: Statistically significant at p ≤ 0.05

Determinants of anti-TNF treatment response in ulcerative colitis

Thirty-four (68%) of UC cases were categorized as responders at week 24, and sixteen (32%) as non-responders. According to baseline analysis, the BMI of non-responders was substantially higher than that of responders (30.14 ± 5.02 vs. 26.14 ± 5.13; p = 0.013), whereas disease duration and age of onset were comparable in the two groups (Table 2).

Table 2.

Relation between anti-TNF treatment response and different parameters in ulcerative colitis group (n = 50)

	Response		p
	Responders (n = 34)	Non-responders (n = 16)
Body Mass Index	26.14 ± 5.13	30.14 ± 5.02	0.013*
Disease duration	4.0 (3.0 – 6.0)	3.0 (2.0 – 5.0)	0.241
Age of onset	27.0 (21.0 – 37.0)	26.0 (21.50 – 33.50)	0.909
Biologic Type
Infliximab	30(88.2%)	14(87.5%)	FEp= 1.000
Adalimumab	4(11.8%)	2(12.5%)
Concomitant corticosteroid	2(5.9%)	3(18.8%)	FEp= 0.311
Concomitant thiopurines	9 (26.5%)	2 (12.5%)	FEp= 0.466
pMayo score at Week 0
Remission (<2)	0 (0.0%)	0 (0.0%)	FEp=0.650
Mild disease (2 - 4)	0 (0.0%)	0 (0.0%)
Moderate (5 - 7)	5 (14.7%)	1 (6.3%)
Severe (>7)	29 (85.3%)	15 (93.8%)
pMayo score at Week 24
Remission (<2)	3 (8.8%)	0 (0.0%)	MCp=>0.001*
Mild disease (2 - 4)	19 (55.9%)	1 (6.3%)
Moderate (5 - 7)	12 (35.3%)	1 (6.3%)
Severe (>7)	0 (0.0%)	14 (87.5%)
Hemoglobin
Week 0	12.0 ± 1.52	11.42 ± 2.12	0.275
Week 24	12.25 ± 1.23	11.58 ± 2.24	0.275
CRP
Week 0	8.0 (4.0 – 18.0)	19.0 (11.0 – 21.50)	0.020*
Week 24	3.60 (2.0 – 6.0)	9.50 (5.0 – 16.0)	0.003*
Fecal calprotectin
Week 0	876.5(577.0 – 1202.0)	979.0(802.5 – 1655.0)	0.231
Week 24	210.0 (47.0 – 520.0)	677.5(502.5 – 1029.0)	>0.001*
Albumin
Week 0	4.33 ± 0.49	4.26 ± 0.39	0.617
Week 24	4.41 ± 0.38	4.77 ± 4.22	0.078
LRG
Week 0	26.53 ± 6.21	34.88 ± 10.49	0.008*
Week 24	13.68 ± 5.37	21.56 ± 5.55	>0.001*
UCEIS
Week 0	7.0 (6.0 – 8.0)	7.50 (7.0 – 8.0)	0.059
Week 24	2.0 (2.0 – 5.0)	7.0 (7.0 – 8.0)	>0.001*
Montreal (disease extent)
E1 proctitis	4 (11.8%)	0 (0.0%)	MCp=0.152
E2 left sided colitis	18 (52.9%)	6 (37.5%)
E3 pancolitis	12 (35.3%)	10 (62.5%)
Geboes score
Week 0	4.0 (4.0 – 4.20)	4.0 (4.0 – 4.20)	0.859
Week 24	2.0 (1.0 – 3.0)	4.0 (4.0 – 4.20)	>0.001*

LRG leucine-rich alpha-2 glycoprotein, UCEIS Ulcerative Colitis Endoscopic Index of Severity, IQR Inter quartile range, SD Standard deviation, t Student t-test,  χ² Chi square test, FE Fisher Exact test, MC Monte Carlotest, U Mann Whitney test

*: Statistically significant at p ≤ 0.05

Most patients in both groups received infliximab (88.2% of responders vs. 87.5% of non-responders; p = 1.000), whereas the remaining patients received adalimumab. There were no statistically significant differences between the groups; concomitant corticosteroid medication was taken in 5.9% of responders compared to 18.8% of non-responders (p = 0.311) and concomitant thiopurines in 26.5% versus 12.5% (p = 0.466).

Non-responders exhibited significantly greater baseline disease activity, evidenced by higher pMayo scores (p < 0.001), CRP (p = 0.007), and UCEIS scores (p = 0.030). Notably, baseline median serum LRG was appreciably higher in non-responders than responders (39.5 vs. 26.53 µg/mL; p = 0.008) (Table 2).

According to the Montreal classification, at baseline, extensive colitis (E3) was more common among non-responders (62.5%) versus responders (35.3%). This variation, nevertheless, did not reach statistical significance (p = 0.152).

Following 24 weeks of anti-TNF therapy, responders displayed marked clinical and endoscopic recovery. All responders achieved remission or attained mild disease activity, whereas 50% of non-responders remained in the severe illness category. This clinical improvement in responders coincided with sharp reductions in CRP, fecal calprotectin, and UCEIS scores. Importantly, post-treatment median LRG levels decreased in responders to 13.68 µg/mL but in non-responders it was 20.5 µg/mL (p < 0.001), confirming its use as a dynamic indicator of anti-TNF treatment response.

Determinants of treatment response in Crohn’s disease

In the CD cohort, 35 patients (70%) were responders and 15 (30%) non-responders. Failing to respond was linked to significantly longer disease duration (median 4.0 opposed 2.0 yrs; p = 0.011), higher baseline BMI (28.31 ± 6.52 vs. 23.39 ± 3.61; p = 0.013), and a higher prevalence of perianal disease (46.7% vs. 5.7%; p = 0.002) (Table 3).

Table 3.

Relation between anti-TNF treatment response and different parameters in Crohn’s disease group (n = 50)

	Response		p
	Responders (n = 35)	Non-responders (n = 15)
Body Mass Index	23.39 ± 3.61	28.30 ± 6.52	0.013*
Disease duration	2.0 (1.0 – 3.0)	4.0 (2.50 – 6.0)	0.011*
Biologic Type
Infliximab	29(82.9%)	13(86.7%)	FEp= 1.000
Adalimumab	6(17.1%)	2(13.3%)
Concomitant corticosteroid	4(11.4%)	3(20%)	FEp= 0.415
Concomitant thiopurines	9  (25.7%)	4 (26.7%)	FEp= 1.000
Age of onset	20.0 (16.75 – 33.0)	28.0 (17.25 – 35.50)	0.334
HBI at week 0
Remission (<2)	0(0.0%)	0(0.0%)	FEp=0.004*
Mild disease (2 - 4)	0(0.0%)	0(0.0%)
Moderate (5 - 7)	32 (91.4%)	8 (53.3%)
Severe (>7)	3 (8.6%)	7 (46.7%)
HBI at week 24
Remission (<2)	24 (68.6%)	1 (6.7%)	MCp=<0.001*
Mild disease (2 - 4)	7 (20.0%)	0 (0.0%)
Moderate (5 - 7)	4 (11.4%)	11 (73.3%)
Severe (>7)	0 (0.0%)	3 (20.0%)
Hemoglobin
Week 0	11.45 ± 1.69	11.27 ± 1.84	0.745
Week 24	11.93 ± 1.36	11.39 ± 1.86	0.328
CRP
Week 0	8.0 (5.0 – 18.25)	15.0 (10.0 – 25.50)	0.011*
Week 24	3.0 (2.0 – 5.0)	6.0 (3.75 – 16.50)	0.006*
Fecal calprotectin
Week 0	398.0 (128.0 – 578.0)	560.0 (331.0 – 692.5)	0.150
Week 24	73.0 (38.50 – 175.0)	385.0 (262.5 – 568.0)	>0.001*
Albumin
Week 0	4.58 ± 0.43	4.17 ± 0.51	0.006*
Week 24	4.68 ± 0.37	4.19 ± 0.53	0.001*
LRG
Week 0	26.03 ± 10.62	35.07 ± 8.99	0.006*
Week 24	14.29 ± 5.69	20.0 ± 6.07	0.002*
SES CD at Week 0
Inactive (0 - 2)	0 (0.0%)	0 (0.0%)	FEp=0.003*
Mild (3 - 6)	0 (0.0%)	0 (0.0%)
Moderate (7 - 15)	25 (71.4%)	4 (26.7%)
Severe (≥16)	10 (28.6%)	11 (73.3%)
SES CD at Week 24
Inactive (0 - 2)	4 (11.4%)	0 (0.0%)	MCp=<0.001*
Mild (3 - 6)	24 (68.6%)	0 (0.0%)
Moderate (7 - 15)	7 (20.0%)	5 (33.3%)
Severe (≥16)	0 (0.0%)	10 (66.7%)
Disease distribution
L1 ileal	13 (37.1%)	1 (6.7%)	MCp=0.047*
L2 colonic	21 (60.0%)	14 (93.3%)
L3+ L4 ileocolonic	1 (2.9%)	0 (0.0%)
Disease behavior
B1 (non-penetrating non- strictring)	32 (91.4%)	9 (60.0%)	MCp=0.023*
B2 (structuring)	2 (5.7%)	4 (26.7%)
B3 (penetrating)	1 (2.9%)	2 (13.3%)
Perianal diseases
No	33 (94.3%)	8 (53.3%)	FEp=0.002*
Yes	2 (5.7%)	7 (46.7%)

LRG leucine-rich alpha-2 glycoprotein, SES-CD The Simple Endoscopic Score for Crohn’s Disease, HBI Harvey–Bradshaw Index, IQR Inter quartile range, SD Standard deviation, t Student t-test, χ² Chi square test, U Mann Whitney test, p p value for comparing between the studied groups

*: Statistically significant at p ≤ 0.05

Infliximab was used in 82.9% of responders vs. 86.7% of non-responders (p = 1.000). Concomitant corticosteroid use was identified in 11.4% of responders against 20% of non-responders (p = 0.415), and thiopurines were utilised in 25.7% versus 26.7% (p = 1.000).

Higher HBI and SES-CD scores indicated a considerably higher initial illness severity in non-responders (p < 0.001). Additionally, pre-anti-TNF therapy serum LRG concentrations were considerably elevated among CD non-responders than responders (35.07 ± 8.99 vs. 26.03 ± 10.62 µg/mL; p = 0.006), a finding consistent with the UC cohort. At week 24, when 66.7% of non-responders remained in the severe SES-CD category, 68.6% of responders experienced clinical remission (p < 0.001).

Predictive performance of baseline LRG and other biomarkers

ROC analysis was performed to evaluate the ability of baseline LRG to predict non-response (Fig. 1). LRG demonstrated moderate discriminatory performance in both cohorts. In UC patients, LRG yielded an AUC of 0.728 (95% CI: 0.539–0.917; p = 0.010), CRP showed statistically significant performance but slightly lower than LRG (AUC 0.704, 95% CI: 0.594–0.860, p = 0.021). Fecal calprotectin did not significantly predict non-response (AUC 0.606, p = 0.232) (Table 4; Fig. 1A).

Fig. 1.

ROC curve for LRG, CRP and Fecal calprotectin in the prediction of non-responders. A: Ulcerative colitis group. B: Crohn’s disease group

Table 4.

ROC curve for LRG, CRP and Fecal calprotectin in prediction of non-responders at week 0 (n= 16 vs. n= 34) in ulcerative colitis group

Before (0 week)	AUC	p	95% C.I	Cut off	Sensitivity	Specificity	PPV	NPV
LRG	0.728	0.010*	0.539 – 0.917	>29	74.19	69.57	52.3	85.7
CRP	0.704	0.021*	0.594 – 0.860	>12	62.50	61.76	43.5	77.8
Fecal calprotectin	0.606	0.232	0.449 – 0.762	>889	56.25	52.94	36.0	72.0

LRG leucine-rich alpha-2 glycoprotein, AUC Area Under a Curve, p value Probability value, CI Confidence Intervals, NPV Negative predictive value, PPV Positive predictive value

*: Statistically significant at p ≤ 0.05

In CD patients, LRG yielded an AUC of 0.753 (95% CI: 0.613–0.894; p = 0.005). CRP exhibited comparable performance (AUC 0.729, 95% CI: 0.589–0.869, p = 0.011) whereas fecal calprotectin continued to be non-significant (AUC 0.630, p = 0.150) (Table 5; Fig. 1B).

Table 5.

ROC curve for LRG, CRP and Fecal calprotectin in prediction of non-responders at week 0 (n= 15 vs. n= 35) in Crohn’s disease group

Before (0 week)	AUC	p	95% C.I	Cut off	Sensitivity	Specificity	PPV	NPV
LRG	0.753	0.005*	0.613 – 0.894	>29	80.0	65.71	50.0	88.5
CRP	0.729	0.011*	0.589 – 0.869	>11.3	66.67	60.0	43.5	81.5
Fecal calprotectin	0.630	0.150	0.466 – 0.793	>420	60.0	62.86	40.9	78.6

LRG leucine-rich alpha-2 glycoprotein, AUC Area Under a Curve, p value Probability value, CI Confidence Intervals, NPV Negative predictive value, PPV Positive predictive value

*: Statistically significant at p ≤ 0.05

A cutoff value of > 29 µg/mL provided sensitivities of (74.2% and 80.0%) and specificities (69.57% and 65.71%) for UC and CD, respectively. The negative predictive value was high in both groups (85.7% for UC and 88.5% for CD), (Tables 4 and 5) indicating that baseline LRG levels below this threshold were highly effective at ruling out future non-response.

Univariate logistic regression revealed baseline LRG as a significant predictor of anti-TNF non-responder in both UC (Crude Odds Ratio (COR) = 1.131; p = 0.003; Table 6) and CD (COR = 1.095; p = 0.011; Table 7). Elevated BMI emerged as a prominent risk factor in both groups. In the CD sample, longer disease duration, HBI and perianal disease were additional significant predictors.

Table 6.

Univariate and multivariable logistic regression analyses identifying predictors of anti-TNF non-responders in the ulcerative colitis group

	Ulcerative colitis
Variable	Univariate		Multivariable
	p	COR (95% C.I.)	p	AOR (95% C.I.)
LRG at 0 week	0.003*	1.131 (1.042–1.229)	0.006*	1.127 (1.034–1.227)
Body Mass Index	0.020*	1.167 (1.024–1.329)	0.046*	1.155 (1.003–1.330)
Albumin	0.609	0.708 (0.188–2.662)	0.983	1.032 (0.054–19.765)
pMayo score at Week 0	0.405	2.586 (0.277–24.189)	0.711	0.710 (0.116–4.340)
CRP	0.579	1.007 (0.983–1.032)	—	—
Disease duration	0.693	0.976 (0.868–1.099)	—	—
Biologic Type (Adalimumab vs. Infliximab)	0.941	1.071 (0.175–6.560)	—	—
Concomitant thiopurines	0.277	0.397 (0.075–2.100)	—	—
Concomitant corticosteroid	0.178	3.692 (0.551–24.730)	—	—

LRG leucine-rich alpha-2 glycoprotein, CRP C-reactive protein, COR Crude Odd`s ratio, AOR adjusted odds ratio, C.I Confidence interval, pMayo Partial Mayo Score

*: Statistically significant at p ≤ 0.05

—: variable not entered into the multivariable model

Table 7.

Univariate and multivariable logistic regression analyses identifying predictors of anti-TNF non-responders in the Crohn’s disease group

	Crohn’s disease
Variable	Univariate		Multivariable
	p	COR (95% C.I.)	p	AOR (95% C.I.)
LRG at 0 week	0.011*	1.095 (1.021–1.174)	0.024*	1.108 (1.014–1.211)
Body Mass Index	0.005*	1.250 (1.069–1.463)	—	—
Perianal disease	0.003*	14.437 (2.506–83.168)	0.081	6.442 (0.793–52.365)
Harvey–Bradshaw Index at 0 week	0.001*	1.423 (1.147–1.766)	0.042*	1.310 (1.009–1.699)
CRP	0.040*	1.052 (1.002–1.105)	—	—
Albumin	0.011*	0.142 (0.032–0.639)	—	—
Disease duration	0.016*	1.503 (1.079–2.094)	—	—
Biologic type (adalimumab vs. infliximab)	0.737	0.744 (0.132–4.190)	—	—
Concomitant thiopurines	0.944	1.051 (0.266–4.145)	—	—
Concomitant corticosteroid	0.429	1.937 (0.376–9.974)	—	—

LRG leucine-rich alpha-2 glycoprotein, CRP C-reactive protein, COR Crude Odd`s ratio, AOR adjusted odds ratio, C.I Confidence interval

*: Statistically significant at p ≤ 0.05

—: variable not entered into the multivariable model

Multivariable modeling validated baseline LRG being an independent predictor of anti-TNF non-responder. For UC group (Table 6) After adjusting for BMI, albumin, and clinical activity (pMayo score), LRG was still substantially related to non-response (Adjusted Odds Ratio (AOR) = 1.127; 95% CI: 1.034–1.227; p = 0.006) Similarly, for CD group (Table 7), baseline LRG stayed independently significant (AOR = 1.108; 95% CI: 1.014–1.211; p = 0.024) (even when factoring in baseline clinical severity (HBI) and perianal disease.

Discussion

Anti-TNF agents remain a keystone in the management of moderate-to-severe IBD owing to their proven ability to induce mucosal healing and alter the natural history of the disease [4]. However, the implementation of a standard “trial-and-error” prescribing policy implies that a significant number of patients, approximately one-third, are exposed to ineffective immunosuppression and its associated risks [7]. Furthermore, confirming response often relies on repeated endoscopic assessment, which is invasive, associated with adverse events, cost-intensive, and not accepted by all patients. That is why attempts to validate non-invasive predictors to detect the likelihood of therapeutic failure have been undertaken [10].

The rationale for assessing baseline serum LRG to predict primary non-response to anti-TNF therapy stems from the accumulating evidence supporting its use as a surrogate for endoscopic activity [26, 27] and the fact that LRG quantification is an unbiased, non-invasive, reproducible metric that can be easily assessed in an outpatient setting.

The biological rationale lies in the inflammatory complexity of IBD. LRG is induced not only by TNF-α but also by IL-1β, IL-6, and IL-22 [13].The actual mechanism of significant LRG elevation among non-responders may be due to the dominance of TNF-independent inflammatory pathways (specifically the IL-6/STAT3 axis) even before treatment initiation [14]. Thus, elevated LRG identifies a patient subset whose inflammation is driven by cytokines that anti-TNF agents cannot effectively block, making it a logically sound candidate for pre-treatment stratification.

In our cohort of IBD patients, baseline demographic features, regarding age and gender distribution, were largely comparable between the studied groups. Similar data were observed in studies focusing on biomarker prediction in IBD [12, 15]. On the other hand, there were highly statistically significant differences in indices of metabolic status and inflammatory burden. Some of the examined clinical parameters showed significant differences between responder and non-responder groups, namely, the higher BMI. In our cohort, non-response was significantly driven by high BMI in UC and CD. This complied with Kennedy et al. [28], who demonstrated in a multicentre cohort study that pretreatment BMI in the CD population was linked to increased PNR. Also, Kurnool et al. [29], demonstrated that elevated BMI negatively influenced the response to anti-TNF medications.

The underlying mechanism likely involves a synergistic effect: obesity not only fosters a chronic proinflammatory state, but also accelerates the proteolytic clearance of immunoglobulins, where higher body weight directly correlates with rapid drug elimination [11, 30].

In our CD cohort, as regards the impact of illness duration on therapy response, non-response was substantially related to longer disease duration. These results align with the findings of a prior meta-analysis by Hamdeh et al. [31], thereby reinforcing the concept of a crucial “window of opportunity” for early intervention. The accumulation of fibrotic bowel injury may be the real mechanism behind this observation. Because early, inflammatory-stage illness is more sensitive to anti-inflammatory drugs than established fibrosis, there is a gap whereby anti-TNF medications decrease inflammation but are unable to repair structural damage [32].

We looked at the relationship between the disease duration and therapy nonresponse in our UC population and found no statistically significant difference among nonresponders. This result is consistent with Iborra et al. [33], who observed no correlation between a long disease duration and anti-TNF nonresponse. In contrast to Beswick et al. [34], who demonstrated that shorter disease duration is linked to an optimized response to anti-TNF drugs. This discrepancy may result from distinct clinical phenotypes: our study included ambulatory patients with moderate-to-severe active luminal illness, which may react differently to treatment than the Acute Severe Colitis (ASC) cases that Beswick et al. studied. Long-standing illness frequently denotes a refractory phenotype in the crucial setting of ASC, possibly involving structural changes that are less amenable to rescue treatment.

Perianal illness was more frequent among non-responders (46.7% vs. 5.7%), indicating that this subgroup is a “difficult-to-treat” phenotype, consistent with findings of Faggiani et al. [35], who categorized perianal involvement as a robust feature, often necessitating higher trough levels of anti-TNF medications to achieve fistula closure. Moreover, clinical manifestations in these aggressive phenotypes differed statistically between responders and non-responders, suggesting that perianal disease represents a distinct biological entity that may require optimized dosing or combination therapy rather than standard induction regimens.

We found that pretreatment LRG concentrations were greater in active UC and CD populations than in healthy people (p < 0.001), which is consistent with previous research. Additionally, Shimoyama et al. [26], showed a clear correlation between LRG and endoscopic activity in CD (r = 0.820), surpassing CRP (r = 0.600) in the detection of small intestinal lesions. In the PLANET study, similarly, Shinzaki et al. [13], confirmed LRG as a sensitive marker in UC as well. They showed a strong correlation with the MES (r = 0.648) as opposed to CRP (r = 0.503), especially in patients with active endoscopic disease but normal CRP levels. We found no discernible difference in LRG levels between our UC and CD subjects (p = 0.942), which is intriguing since it supports the idea that LRG represents the systemic inflammatory burden shared by both phenotypes rather than disease-specific pathways.

The most clinically impactful finding of our study is the capacity of baseline LRG to stratify risk for anti-TNF non-response. We identified an empirically derived cut-off of > 29 µg/ml demonstrated moderate discriminatory performance, provided acceptable sensitivities (74.2% and 80.0%) and specificities (69.57% and 65.71%) for UC and CD respectively, and yielded a NPV of nearly 86–88% for both UC and CD. This implies that patients with LRG levels below this threshold are associated with a high probability of response (NPV 85.7–88.5%), supporting the usefulness of anti-TNF as a rational first-line strategy for this subgroup. Also, the positive predictive value (PPV) hovered at approximately 50% -52% for both clinical phenotypes. This means an elevated LRG level alone does not guarantee treatment failure. Instead, it flags high-risk patients who may require closer clinical monitoring or faster dose adjustments.

Similarly, our findings align with, yet distinctively extend, the observations from Shinzaki et al.‘s PLANET study [13], which demonstrated that LRG levels < 16 µg/ml were highly specific for endoscopic remission (Specificity 78.4%); they did not primarily utilize baseline LRG to predict primary non-response. In contrast, our study leveraged pretreatment LRG levels (> 29 µg/ml) to identify patients destined to fail induction. The difference in cut-offs, 16 µg/ml for remission monitoring in PLANET versus 29 µg/ml for failure prediction in our cohort—highlights the dual utility of LRG: lower thresholds define deep remission; however, significantly higher thresholds signal a refractory, high-inflammatory phenotype likely driven by TNF-independent pathways.

Similarly, Karashima et al. [36], recently demonstrated that an early reduction in serum LRG at week 1, rather than baseline levels, predicted clinical remission at week 8 and endoscopic improvement within one year in UC patients commencing induction therapy, including a subgroup receiving anti-TNF agent. Thus, our data complements and extend these studies by validating LRG not just as a “target” for treatment (week 24), but as a “gatekeeper” for drug selection (week 0).

In our cohort, baseline clinical activity scores HBI for CD and pMayo score for UC displayed high baseline severity among non-responders. However, when subjected to multivariable regression, the pMayo score did not emerge as an independent predictor of primary non-response in the ulcerative colitis group (p = 0.711), whereas the HBI remained independently significant for the Crohn’s disease cohort (p = 0.042).

Furthermore, even after adjusting for BMI, albumin, and baseline clinical activity scores (partial Mayo score in UC and Harvey-Bradshaw Index in CD), baseline LRG continued to be an independent predictor of primary non-response in both UC (AOR 1.127, p = 0.006) and CD (AOR 1.108, p = 0.024). This indicates that LRG is a valuable supplementary biomarker for pre-treatment risk stratification in biologic-naïve IBD patients, since it offers additional prognostic information beyond conventional clinical assessment techniques.

Therefore, patients with pretreatment LRG levels above the cut-off of 29 µg/mL may benefit from a more intensive and proactive management strategy during anti-TNF induction. In clinical practice, these patients might be identified for more frequent clinical evaluations, serial biomarker tests (CRP and faecal calprotectin), and an earlier endoscopic evaluation at weeks 8–12 instead of waiting until the traditional week-24 assessment. To prevent irreversible bowel damage, this strategy would enable the early identification of primary non-responders and enable timely treatment optimisation, such as dose escalation, immunomodulator addition, or switching to an alternative biologic class (e.g., anti-integrin or anti-IL-12/23 agents). Early multidisciplinary discussion regarding surgical options may be necessary in certain high-risk patients, such as high BMI, lengthy disease duration, or perianal involvement to prevent prolonged exposure to ineffective therapy. Thus, incorporating baseline LRG into risk-stratification algorithms may provide a more customised “treat-to-target” approach, which may enhance clinical results and the use of medical resources.

In our UC patient, median LRG levels decreased in responders to 13.68 µg/mL but in non-responders it was 20.5 µg/mL (p < 0.001) supporting its utility as a dynamic marker of therapeutic response. Also, in Crohn’s patients, responders had lower LRG levels (14.29 µg/mL) than non-responders (20 µg/mL) (p = 0.002) further supporting its utility as a dynamic marker of therapeutic response.

Consistent with the findings by Kono et al. [12], and the PLANET study [13], at week 24, we found that responders’ serum LRG levels significantly decreased, but non-responders’ levels remained elevated. This dynamic change validates LRG not just as a baseline predictor, but as a reliable monitoring tool for “treat-to-target” strategies. By reflecting the suppression of deep inflammation, LRG may offer complementary utility to CRP, particularly in determining whether true remission has been achieved.

Several constraints need to be noted. First, the small sample size and single-center design warrant caution; our derived cut-off requires extra validation in larger multicentre cohorts. Furthermore, because of the small sample size according to the current recommendations for logistic regression modelling, including large number of covariates would substantially increase the risk of model over fitting and unstable coefficient estimates and hence multiple significant factors in univariate analysis were not included in the multivariate analysis model. Second, we were unable to distinguish between pharmacokinetic failure (rapid clearance) and pharmacodynamic failure (mechanistic escape) since we did not quantify anti-TNF trough levels. Future research should integrate baseline inflammatory stratification with therapeutic drug monitoring induction. In such an algorithm, baseline LRG could inform biologic selection, while early trough levels guide dose optimization.

In conclusion, baseline serum LRG serves as a modest non-invasive marker that can complement existing pre-treatment evaluations. High baseline levels (> 29 µg/mL) act as a helpful clinical indicator for potential failure of anti-TNF induction in Egyptian patients. Its moderate yet consistent discriminatory performance, demonstration of a biological gradient, and mechanistic linkage to TNF-independent inflammation support its integration into precision treatment frameworks. Implementing LRG testing in routine clinical practice could help stratify patients, guiding clinicians to consider alternative biologics for those at high risk of non-response, thereby optimizing resource allocation and improving long-term outcomes.

Abbreviations

Abbreviation

Definition

IBD

Inflammatory bowel disease

UC

Ulcerative colitis

CD

Crohn’s disease

anti-TNF-α

anti-tumor necrosis factor-alpha

PNR

primary non-response

LRG

leucine-rich alpha-2 glycoprotein

CRP

C-reactive protein

IRB

Institutional Review Board

BMI

Body mass index

HBI

Harvey–Bradshaw Index

pMayo

Partial Mayo Score

CBC

Complete blood count

SES-CD

Simple Endoscopic Score for Crohn’s Disease

MES

Mayo Endoscopic Subscore

UCEIS

Ulcerative Colitis Endoscopic Index of Severity

GS

Geboes score

SD

Standard deviation

IQR

Interquartile range

ROC

Receiver operating characteristic

NPV

Negative Predictive Value

PPV

Positive Predictive Value

AUC

Area Under the Curve

ASC

Acute Severe Colitis

Authors’ contributions

IA, YE, ME, SE, DS proposed the concept of the study. IA, YE, ME, SE, HE, DS, ME searched the literature. IA, YE, ME, SE revised the study design. IA, YE, ME, SE, HE, ME evalauted patients and nalysed the data. DS revised the laboratory work up. IA, YE, ME, SE, DS analyized the results. IA, YE, ME, SE wrote the intial draft. All authors read, revised and accepted the final manuscript.

Funding

No funding was received in any aspect of this study.

Data availability

All data supporting the findings of this study are available within the paper and its Supplementary Information.

Declarations

Ethics approval and consent to participate

The study protocol was approved by the Kafrelsheikh University, Faculty of Medicine Institutional Review Board (IRB NO: KFSIRB200-169) and conformed to the 2013 revision of the Declaration of Helsinki. Written informed consent was obtained from all enrolled subjects.

Consent for publication

Not Applicable.

Competing interests

The authors declare no competing interests.