Serum C-X-C motif chemokine ligand 9, interleukin 8, and interleukin 22 as key biomarkers in pediatric inflammatory bowel disease

Abstract

BACKGROUND

The diagnosis of inflammatory bowel disease (IBD) involves clinical, endoscopic, and radiologic evaluation. Endoscopic procedures, particularly in pediatrics, require general anesthesia and carry potential risks.

AIM

To investigate whether serum biomarkers can differentiate between pediatric patients with and without IBD. Secondary objectives included identifying biomarkers that distinguish Crohn’s disease (CD) from ulcerative colitis (UC) and assessing their predictive value for progression to biologic therapy.

METHODS

Pediatric patients undergoing diagnostic colonoscopy at British Columbia Children’s Hospital between December 2017 and June 2022 were enrolled. Blood samples were collected at colonoscopy, and demographic clinical data, laboratory, and histopathologic evaluation were obtained. An exploratory screen of 50 biomarkers was undertaken in a subset of patients (54 IBD, 41 controls) using Legendplex^TM flow cytometry kits to identify candidates. A refined panel of 12 serum biomarkers was subsequently selected and a supervised learning model was developed to classify patients.

RESULTS

The study included 246 pediatric patients, who had a median age of 13.03 years and were 37.4% female (103 CD, 52 UC, 91 controls). In univariate analyses, C-X-C motif chemokine ligand 9 (CXCL9) was the only biomarker significantly elevated in IBD vs controls (P < 0.001). A multivariable model achieved an area under the receiver operating characteristic of 0.861 for distinguishing IBD from controls. Interleukin 8 (IL-8) emerged as a key biomarker alongside CXCL9 and IL-22 in the model. The random forest model identified CXCL9 with the greatest diagnostic accuracy (area under the curve [AUC] = 0.81), followed by IL8 and IL22 (AUC = 0.737 and 0.68, respectively). CXCL9 and IL-18 showed higher levels in CD (P = 0.016), whereas CXCL1 levels predicted progression to biologic therapy within 1 year (P = 0.039). However, the model did not effectively predict disease subclassification or progression to biologic therapy.

CONCLUSION

Serum biomarkers, particularly CXCL9, IL-8, and IL-22, can aid in the diagnosis of pediatric IBD. CXCL9 and IL18 were found to be significant predictors of CD, and CXCL1 differed between patients requiring biologic therapy vs those who did not.

Key Words: C-X-C motif chemokine ligand 9; Interleukin 18; Interleukin 22; C-X-C motif chemokine ligand 1; Pediatric inflammatory bowel disease

Core Tip: The diagnosis of pediatric inflammatory bowel disease (IBD) usually requires invasive endoscopy under anesthesia, which carries risks in children. In this study, we evaluated serum cytokines as potential non-invasive biomarkers to distinguish pediatric IBD from non-IBD controls. Among over 250 pediatric patients, C-X-C motif chemokine ligand 9 (CXCL9), interleukin 8 (IL-8), and IL-22 emerged as key markers with strong discriminatory capacity. CXCL9 and IL-18 differentiated Crohn’s disease from ulcerative colitis, whereas CXCL1 levels were associated with the need for biologic therapy within 1 year. These findings suggest that specific serum biomarkers may support earlier diagnosis and aid in patient stratification.

INTRODUCTION

The current gold standard for the diagnosis of inflammatory bowel disease (IBD) is based on clinical and endoscopic evaluation as well as histologic and radiologic evaluation[1]. However, endoscopic evaluation can be uncomfortable and carry some risks, especially in the pediatric population that requires general anesthesia to perform the procedure[2].

Classic inflammatory biomarkers such as hemoglobin, C-reactive protein (CRP), erythrocyte sedimentation rate, and albumin are part of the diagnostic approach for IBD. In clinical practice, these biomarkers are commonly utilized; however, they lack the specificity and sensitivity required to independently and accurately diagnose IBD[3]. Fecal calprotectin stands as the most extensively studied non-invasive marker for IBD[4,5]. The incorporation of fecal calprotectin alongside symptom assessment currently enhances the diagnostic precision of pediatric IBD (PIBD). However, it is not yet precise enough to obviate the necessity for endoscopies[6]. Consequently, there remains a need to discover new and effective markers for the early diagnosis of PIBD.

In the realm of contemporary research of IBD, various biomarkers have undergone evaluation and analysis over the past two decades, remaining integral to ongoing clinical research[7,8]. During this period, the proposition of a regulatory cytokine network with significant implications for IBD progression has gained increasing prominence. Cytokines, encompassing interleukins (ILs), interferons (IFNs), growth factors, chemokines, and the tumor necrosis factor family, are soluble signaling molecules crucial to the adaptive and innate immune systems, playing a pivotal role in the pathophysiology of IBD[9,10]. Acting on both local and systemic levels, it is hypothesized that the levels of circulating cytokines could potentially serve as differentiators between patients with and without IBD, functioning either diagnostically or prognostically[11,12].

Recently, a limited number of cytokines known to play crucial roles in the IBD have been explored as potential key biomarkers for diagnosing IBD[13-16]. However, there is a notable scarcity of studies focused on cytokine levels in patients with PIBD. Furthermore, there is a lack of understanding regarding serum cytokines that could differentiate between Crohn's disease (CD) and ulcerative colitis (UC) in patients with PIBD prior to the initiation of treatment[17].

Our objective was to explore the potential of serum biomarkers in distinguishing between pediatric patients with and without IBD. Additionally, we identified biomarkers that could differentiate between CD and UC, and correlate with disease severity as secondary aims.

MATERIALS AND METHODS

Study population

Pediatric patients undergoing their first diagnostic colonoscopy at British Columbia Children's Hospital (Vancouver, BC, Canada) between December 2017 and June 2022 were invited to participate in the study.

Inclusion criteria: (1) Pediatric patients (ages 7-18 years) undergoing their first diagnostic colonoscopy; and (2) Patients diagnosed with IBD and non-IBD controls (e.g., functional abdominal pain, irritable bowel syndrome, polyps, infection or other inflammatory disease).

Exclusion criteria: (1) Patients with an inconclusive diagnosis following colonoscopy and histopathology; (2) Patients who had previously undergone colonoscopy or had prior established IBD diagnosis; or (3) Patients who were not within the above age range.

Sample collection and biomarker analysis

A 2 mL blood sample was collected from each participant at the time of colonoscopy. Demographic and clinical data were retrieved for all patients. The initial analysis included 95 samples, which were evaluated using Legendplex^TM flow cytometry kits to measure a panel of 50 biomarkers.

Based on an initial screening and extensive literature review, a refined panel of 12 serum biomarkers was selected for further investigation (see Supplementary Table 1 and Supplementary Figures 1 and 2 for the biomarker list and initial analysis results). A customized Legendplex^TM flow cytometry kit was then designed to specifically measure these 12 biomarkers (Table 1).

Table 1 Univariate analysis of biomarker intensities: Comparison of inflammatory bowel disease vs non-inflammatory bowel disease patients, mean ± SD.

Biomarker	IBD (n = 155)	Non-IBD (n = 91)	P value
CCL2	101 ± 42.9	108 ± 43.7	0.239
CCL5	1650 ± 1010	1710 ± 665	0.526
CCL7	249 ± 135	244 ± 120	0.761
IL-8	27.1 ± 151	8.93 ± 4.63	0.138
TSLP	269 ± 123	265 ± 132	0.812
CCL3	45.6 ± 15.7	44.0 ± 13.1	0.409
CXCL1	347 ± 289	306 ± 275	0.267
CXCL2	200 ± 197	196 ± 225	0.892
CXCL9	3980 ± 3250	1440 ± 1460	< 0.001
IL-18	169 ± 187	134 ± 216	0.204
IL-22	52.8 ± 163	31.9 ± 13.6	0.113
TARC	326 ± 461	331 ± 241	0.913

IBD: Inflammatory bowel disease; CCL: C-C motif ligand; CXCL: C-X-C motif chemokine ligand; IL: Interleukin; TARC: Thymus and activation-regulated chemokine; TSLP: Thymic stromal lymphopoietin.

Blood samples were analyzed using the customized Legendplex^TM kit, and a supervised learning model was developed to assess its predictive ability regarding the following outcome measures: (1) Differentiation between IBD and non-IBD; (2) Distinction between CD and UC among patients with IBD; and (3) Prediction of disease progression. For prognostic analysis of treatment escalation, we defined a “favorable outcome” as remaining both steroid-free and biologic-free at the 1-year follow-up. This definition was based on medication use data recorded in the medical chart: Being steroid-free within 1 year of diagnosis without requiring biologic therapy vs those who required biologic treatment.

The study was approved by the British Columbia Children’s Hospital ethics committee (Approval No. H10-01760). All parents or legal guardians of participants signed an informed consent. The manuscript was written according to the STROBE Statement checklist of items.

Statistical analyses

All analyses were conducted in R (v4.2.2) using the tidymodels ecosystem. Two-sided P < 0.05 was considered statistically significant. Continuous variables are summarized as median (interquartile range) and categorical variables as count (percentage). Groupwise comparisons of biomarker expression were performed using Student’s t-tests.

To identify biomarkers associated with IBD status, we fit separate logistic regression models for each analyte, adjusting for age, sex, and ethnicity. From each model, we extracted the type II Wald χ² statistic for the biomarker term. Percentile bootstrap resampling (5000 replicates) was used to derive P values and confidence intervals.

We then trained a random forest classifier (1000 trees) using the ranger engine. The number of variables sampled at each split (mtry) and the minimum node size (min_n) were tuned via grid search. Model training and evaluation used 5-fold cross-validation repeated 10 times, stratified by diagnosis. Final performance was assessed on held-out folds, presented as area under the receiver operating characteristic (AUROC).

Secondary outcomes were analyzed analogously, using t-tests for expression differences, bootstrap-ranked χ² screening, and tailored random forest classifiers.

RESULTS

We initially performed an exploratory screen of 50 biomarkers to identify useful candidates for further study in a subset of 95 patients with IBD (n = 54) and without-IBD (n = 41). The median age of participants was 13.4 years, and 39 (41%) were female. In the initial subset, five biomarkers, namely C-X-C motif chemokine ligand 9 (CXCL9, MIG), CXCL1, thymus and activation-regulated chemokine (also known as CC motif chemokine ligand 17), IL22 and IL8, were found to show differences between IBD and nonIBD. CXCL9 showed the highest discriminatory power in both univariate tests and random forest models (Supplementary Figures 1 and 2). Based on these findings, these five biomarkers, along with seven additional analytes were measured in all participants (Supplementary Table 1).

A total of 255 pediatric patients were recruited to the study, of whom 9 were excluded due to inconclusive diagnosis, leaving 246 pediatric patients for the analyses. The median age of participants was 13.03 years, and 92 (37.4%) were female. IBD was diagnosed in 155 patients (63.0%), with 103 (66.4%) classified as having CD (Table 2). Comparison of biomarker concentrations between IBD and nonIBD control subjects revealed that CXCL9 was significantly elevated in IBD (mean 3980 ± 3250 pg/mL vs 1440 ± 1460 pg/mL in controls; P < 0.001). IL22 and IL8 were higher in IBD (means of 52.8 ± 16.3 pg/mL and 27.1 ± 151 pg/mL, respectively) compared with controls (31.9 ± 13.6 pg/mL and 8.93 ± 4.63 pg/mL), but the differences were not statistically significant (P = 0.113 and P = 0.138, respectively). Other analytes showed no significant differences (Table 1).

Table 2 Demographic and disease characteristics, n (%).

	All (n = 246)	IBD (n = 155)	Non-IBD (n = 91)
Age at diagnosis median (IQR; year)	11.63 (8.56-13.95)	11.15 (9.17-15.08)	11.68 (8.48-13.83)
Female	44 (34.37)	10 (35.71)	34 (34)
Crohn's disease	103 (41.9)	103 (66.4)	0
UC	46 (18.7)	46 (29.7)	0
IBD-U	6 (2.4)	6 (3.9)	0
Ethnicity
Western European	175 (71.1)	106 (68.4)	69 (75.8)
South Asian	32 (13)	22 (14.2)	10 (10.9)
Other	39 (15.8)	27(17.4)	12 (13.2)
Biologic treatment at 1 year		48 (39)
Activity score at diagnosis
Remission		4 (2.58)
Mild		96 (61.93)
Moderate-severe		50 (32.25)
Missing		5 (3.2)
Crohn's disease (n = 103)
Behavior
B1		93 (90.3)
B2		6 (5.8)
B3		4 (3.9)
Location
L1		17 (16.7)
L2		30 (39.4)
L3		55 (53.4)
L4		50 (48.5)
Perianal disease		34 (33)
UC/IBD-U (n = 52)
Location
E1		13 (25)
E2		12 (23.1)
E3		10 (19.2)
E4		17 (32.7)
Diagnosis
IBS/functional abdominal pain			38 (41.8)
Infection			6 (6.6)
Other inflammatory disease			16 (17.6)
Polyp			12 (13.2)
Other			19 (20.9)

Remission: Pediatric ulcerative colitis activity index (PUCAI)[29]/pediatric Crohn's disease activity index (PCDAI)[30] < 10; Mild: PUCAI 10-34, PCDAI 10-30; Moderate-severe: PCDAI > 30, PUCAI > 35. IBD-U: Inflammatory bowel disease unspecified; IBS: Irritable bowel syndrome; IQR: Interquartile range; PCDAI: Pediatric Crohn's disease activity index; PUCAI: Pediatric ulcerative colitis activity index; UC: Ulcerative colitis.

A multivariable model using all 12 biomarkers achieved an AUROC of 0.861 for distinguishing IBD from controls (Figure 1A). Variable importance rankings, from our random forest model identified CXCL9 > IL-22 > IL8 (Figure 1B). We then fitted separate logistic regression models for each biomarker, adjusting for age and sex, and generated ROC curves to assess their individual performance (Figure 1C). CXCL9 showed the greatest diagnostic accuracy (area under the curve [AUC] = 0.81), followed by IL8 (AUC = 0.737) and IL22 (AUC = 0.68).

Figure 1 Area under the receiver operating characteristic curve. A: Primary classification model discriminating between patients with inflammatory bowel disease (IBD) and without IBD including all 12 biomarkers; B: Variable importance in the final random forest model; C: Discriminative ability of each biomarker while adjusting for age and sex. AUC: Area under the curve; CI: Confidence interval; ROC: Receiver operating characteristic; CCL: C-C motif ligand; CXCL: C-X-C motif chemokine ligand; IL: Interleukin; TARC: Thymus and activation-regulated chemokine.

To evaluate whether the panel could differentiate between IBD subtypes, we compared the concentrations of all 12 analytes between CD (n = 103) and UC (n = 46). Only CXCL9 and IL18 differed significantly (both P = 0.016), with higher levels in CD. Although IL22 and IL8 were elevated overall in IBD, they did not distinguish the subtypes (Table 3). A multivariable model using the full biomarker panel classified CD vs UC poorly (AUROC = 0.60).

Table 3 Univariate analysis of biomarker intensities: Comparison of Crohn's disease vs ulcerative colitis, mean ± SD.

	CD (n = 103)	UC (n = 46)	P value
Female, n (%)	32 (31.1)	21 (45.7)	0.125
Ethnicity, n (%)			0.02
Western European	77 (74.8)	25 (54.3)
South Asian	9 (8.74)	11 (23.9)
Other	17 (16.5)	10 (21.7)
Age (year)	12.5 ± 2.71	13.6 ± 3.07	0.031
Biomarker
CCL2	97.0 ± 36.8	111 ± 52.6	0.113
CCL5	1643 ± 1139	1624 ± 665	0.897
CCL7	259 ± 144	245 ± 111	0.501
IL-8	33.2 ± 186	14.4 ± 9.54	0.307
TSLP	268 ± 119	290 ± 130	0.325
CCL3	45.7 ± 15.5	43.9 ± 9.96	0.399
CXCL1	318 ± 253	385 ± 353	0.248
CXCL2	193 ± 155	203 ± 261	0.814
CXCL9	4468 ± 3433	3171 ± 2763	0.016
IL-18	190 ± 218	128 ± 91.0	0.016
IL-22	59.5 ± 199	40.2 ± 16.9	0.331
TARC	293 ± 180	403 ± 802	0.362

CCL: C-C motif ligand; CD: Crohn's disease; CXCL: C-X-C motif chemokine ligand; IL-8: Interleukin 8; TSLP: Thymic stromal lymphopoietin; UC: Ulcerative colitis.

We also explored whether baseline biomarker levels could predict which children would require biologic therapy during the first year after diagnosis. Among the 123 participants with follow-up data, 75 remained biologicfree (61%). CXCL1 was the only analyte that differed significantly between these groups (mean 415 ± 330 pg/mL vs 297 ± 257 pg/mL; P = 0.039), with lower levels associated with remaining steroidfree and biologicfree (Table 4). CXCL9, IL22, and IL8 were comparable between groups. A predictive model combining all biomarkers achieved an AUROC of roughly 0.60, suggesting the limited ability of baseline cytokine levels alone to forecast early biologic use.

Table 4 Univariate analysis of biomarker intensities: Comparison of biological treatment initiation by 1 year.

	No biologic treatment (n = 75)	Biologic treatment (n = 48)	P value
Female, n (%)	33 (44.0)	12 (25.0)	0.052
Ethnicity, n (%)			0.56
Western European	50 (66.7)	34 (70.8)
South Asian	13 (17.3)	5 (10.4)
Other	12 (16.0)	9 (18.8)
Age (year)	13.0 ± 3.07	12.5 ± 2.39	0.353
Biomarker
CCL2	104 ± 37.1	105 ± 54.1	0.937
CCL5	1722 ± 770	1798 ± 1479	0.742
CCL7	283 ± 138	269 ± 108	0.532
IL-8	40.5 ± 217	15.1 ± 13.4	0.316
TSLP	308 ± 106	286 ± 128	0.323
CCL3	44.2 ± 11.9	46.2 ± 17.3	0.498
CXCL1	297 ± 257	415 ± 330	0.039
CXCL2	176 ± 204	241 ± 201	0.083
CXCL9	4087 ± 3470	4102 ± 2828	0.979
IL-18	137 ± 123	214 ± 280	0.08
IL-22	64.9 ± 233	43.8 ± 16.0	0.438
TARC	379 ± 641	288 ± 146	0.238

CCL: C-C motif ligand; CXCL: C-X-C motif chemokine ligand; IL: Interleukin; TARC: Thymus and activation-regulated chemokine; TSLP: Thymic stromal lymphopoietin.

DISCUSSION

A personalized approach is essential for the effective management of PIBD, emphasizing the importance of tailoring treatment strategies to individual patients. Recent studies have highlighted the potential utility of serum biomarkers in predicting both diagnosis and therapeutic outcomes in IBD. Given the critical importance of early intervention and disease control, particularly in CD[1], there is growing interest in identifying biomarkers capable of predicting disease progression. While the top-down therapeutic approach has demonstrated efficacy irrespective of serum biomarker profiles or disease severity, an alternative strategy may be warranted. Specifically, there is a need to identify biomarkers that can stratify patients based on their likelihood of achieving favorable outcomes without progression to biologic therapy. In this study, we explored the potential of serum biomarkers to differentiate between distinct clinical outcomes. Specifically, we assessed their ability to distinguish between pediatric patients with and without IBD, to differentiate between the two major PIBD subtypes, and to identify patients who can successfully navigate the first-year post-diagnosis without requiring biologic therapy.

In our study, CXCL9 and IL-22 were identified as key discriminatory biomarkers for IBD, reflecting their crucial roles in immune response and inflammation. Although IL-22 and IL-8 did not reach statistical significance in univariate analyses when comparing IBD to non-IBD patients, both emerged as important contributors within the multivariable classification model. This distinction underscores the difference between single-analyte performance and model-based feature importance: While IL-22 and IL-8 alone showed only modest discriminatory capacity, their combined contribution with other cytokines, particularly CXCL9, improved the overall performance of the diagnostic panel (AUROC = 0.861). These findings suggest that IL-22 and IL-8 may function synergistically within broader biomarker networks rather than serving as robust stand-alone markers.

CXCL9, known for its involvement in T-cell trafficking[18], and IL-22, associated with mucosal immunity and intestinal epithelial barrier integrity[19-21], showed robust diagnostic performance, suggesting a high degree of reliability. IL-22, was already found to be a key biomarker in treatment response in longitudinal studies[22], but was disappointing as a therapeutic target in a Phase 2 study[23]. CXCL9 has recently been identified as a potential predictor of IBD in a study comparing CXCL9 levels in active patients with IBD to non-IBD controls and assessing their association with disease severity[24]. These findings, combined with the results of our study, underscore the need for further investigation into the role of CXCL9 as a biomarker in IBD.

Beyond diagnosis, CXCL9 and IL-18 were found to be significant predictors of CD, with P values of 0.016 for both. This differentiation holds clinical importance given the distinct management strategies required for CD and UC. This observation is consistent with the T helper 1 (Th1)-skewed inflammatory profile of CD, in which IL-18 promotes IFN-γ production and CXCL9 recruits CXCR3⁺ T cells, amplifying mucosal Th1 responses[25]. A relatively old pediatric study similarly reported elevated systemic IL-18 in children with CD but not in those with UC, supporting its potential as a CD-specific biomarker[26]. CXCL9, produced predominantly by activated macrophages in response to IFN-γ, has also emerged as one of the most upregulated chemokines in the serum of patients with CD, reinforcing its discriminatory role. Indeed, in one large multiplex study, CXCL9 and IL-18 showed among the strongest associations with CD status, underscoring that a Th1-associated cytokine profile is a hallmark of CD[27]. However, further validation in larger, independent cohorts is necessary to solidify these findings.

The study also explored biomarkers predictive of disease progression, identifying low CXCL1 as a potential marker for defense from progression to biologic treatment within 1 year of diagnosis (P = 0.039). High CXCL1 reflects an intense innate inflammatory response, and neutrophil chemoattraction and activation, which often correlates with severe, refractory disease. Notably, a large serologic study identified CXCL1 as a top discriminator of CD vs healthy controls[27], and general IBD research has shown that CXCL1 is elevated during active flares and in patients with extensive inflammation. Moreover, Fang et al[24] observed that active IBD (both UC and CD) is marked by significantly increased M1 macrophages expressing CXCL9/CXCL10 and higher chemokine levels, which decline in remission. This underscores that chemokine elevation tracks with inflammatory burden. Clinically, pediatric patients who ultimately require biologics often present with more systemic inflammation at diagnosis (e.g., high CRP, hypoalbuminemia)[28]. CXCL1 may be a surrogate for this “high-inflammatory” state. Thus, elevated CXCL1 early in disease could flag a robust innate immune activation and mucosal neutrophil infiltrate, portending a more aggressive course. Despite this, the overall prediction model for subclass differentiation and treatment progression exhibited limited accuracy. This highlights the complexity of IBD progression and suggests a need for integrating additional biomarkers and clinical factors to enhance predictive capabilities.

This study represents one of the largest pediatric cohorts evaluated for serum cytokine biomarkers in IBD to date and addresses an important gap in non-invasive diagnostics. However, despite its promising findings, it has several limitations. While the study included a relatively large cohort, it was conducted at a single center, potentially limiting the generalizability of findings to broader populations with varying demographics and healthcare settings. Patient demographics, disease phenotypes, and clinical management protocols at British Columbia Children’s Hospital may not fully reflect those in other geographic regions or healthcare systems. For example, the ethnic composition of our cohort was predominantly Western European, which may influence biomarker distributions and limit applicability to more diverse populations. Likewise, local practice patterns, such as the thresholds for initiating biologic therapy, could affect the observed associations between cytokine levels and treatment outcomes. Multicenter validation across varied populations will therefore be essential to confirm the diagnostic and prognostic utility of CXCL9, IL-22, IL-18, and CXCL1 in PIBD. Additionally, the study’s cross-sectional nature restricts the ability to infer causation or track changes in biomarker levels over time. A longitudinal study design would better capture the dynamic nature of disease progression and response to treatment. Additionally, in our prognostic analysis of progression to biologic therapy within 1 year, we defined a favorable outcome as being both steroid-free and biologic-free at follow-up, assuming this reflected clinical remission. However, we did not have objective confirmation of remission (e.g., endoscopic or histologic findings), and some patients may not have been in true clinical remission at the time of data collection.

CONCLUSION

The results of this study offer several practical applications and research avenues such as early identification, early differentiation, monitoring and treatment prediction. Our findings suggest that these biomarkers may serve as a useful triage tool in clinical practice, helping to prioritize which patients should undergo invasive procedures. Children without elevated cytokine levels may be less likely to require urgent endoscopic evaluation, whereas those with elevated biomarkers could be flagged for earlier referral. This approach could potentially reduce unnecessary endoscopies, while ensuring that patients with suspected IBD still receive appropriate diagnostic work-up and phenotyping. Biomarkers such as CXCL9 and IL-18 can also aid in early differentiation between CD and UC, facilitating timely and tailored treatment strategies. While the predictive model for disease progression requires improvement, CXCL1 offers a starting point for refining predictive tools. Incorporating additional biomarkers or longitudinal clinical data could enhance model accuracy. Future research should validate these findings across larger, diverse cohorts and investigate the integration of these biomarkers into routine clinical practice.