Preliminary IL-1 Family Cytokine Signature for Crohn’s Disease Onset in Pediatric Juvenile Idiopathic Arthritis

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

This study investigates cytokine changes during the conversion from juvenile idiopathic arthritis (JIA) to Crohn’s disease (CD) and proposes that elevated levels of the IL‑1 family—particularly IL‑1α, IL‑1β, and IL‑36α—can serve as early biomarkers for predicting clinical onset of CD.  

1. Only 4 patients with JIA eventually converted to CD in this study. Although statistical significance was achieved, the small sample size introduces high chance effects and a serious risk of overfitting. The authors are advised to soften their claims. These findings should be described as preliminary predictive signatures or exploratory biomarkers.  
2. The authors are encouraged to clearly specify in the Methods section how these 4 converters were diagnosed. Did they undergo ileocolonoscopy and upper gastrointestinal endoscopy? Was transmural inflammation or characteristic granuloma confirmed histologically? Pathological and endoscopic details of the diagnostic workup are important.  
3. Standardize the use of abbreviations. Ensure all abbreviations are defined at first mention in both the abstract and main text. Correct erroneous abbreviations such as “CRB” in the abbreviation list and unify terminology and abbreviation formatting throughout the manuscript.  
4. Performance validation of the predictive model is insufficient. The near‑perfect diagnostic performance derived from only 4 conversion events was based solely on direct calculation from a single dataset, without internal cross‑validation, let alone external validation. Such unvalidated “perfect performance” lacks clinical credibility in studies of clinical prediction models.

Author Response

Comment 1: Only 4 patients with JIA eventually converted to CD in this study. Although statistical significance was achieved, the small sample size introduces high chance effects and a serious risk of overfitting. The authors are advised to soften their claims. These findings should be described as preliminary predictive signatures or exploratory biomarkers.

Response 1: We thank the reviewer for this thoughtful and important observation. We fully agree that the small size of the converter subgroup (N = 4) limits generalizability and introduces a risk of overfitting. As noted in the Limitations, these findings represent a hypothesis-generating pilot observation requiring validation in larger, multicenter prospective cohorts.

In response to this comment, we have revised the manuscript to more consistently frame the IL-1 family signature as a preliminary predictive signature and exploratory biomarker. Specific textual adjustments include:

1) In “Abstract”, we replaced the phrase: "A baseline IL-1 family signature may predict CD onset..." by a softer and less ambitious statement: "A baseline IL-1 family signature may represent a preliminary predictive signature for CD onset..." (page 1, lines 35-36).

2) In the “Discussion” section, we replaced the phrase: "we propose a new clinical screening paradigm…" by “we propose that this exploratory biomarker framework could inform future screening strategies” (page 10, lines 330-332).

3) Also in the revised “Conclusions”, we stated “identifies a preliminary biomarker signal warranting further validation” instead of " identifies a specific biomarker “early-warning' system" (page 13, lines 508-509).

4) We also replaced the phrase "predictive biomarker" by either “exploratory biomarker candidate” (page 10, line 379) or “preliminary predictive signature” (page 1, line 36) depend on the context.

Comment 2: The authors are encouraged to clearly specify in the Methods section how these 4 converters were diagnosed. Did they undergo ileocolonoscopy and upper gastrointestinal endoscopy? Was transmural inflammation or characteristic granuloma confirmed histologically? Pathological and endoscopic details of the diagnostic workup are important.

Response 2: In response to the reviewer's comment, we added the following statement to the subsection “4.2 Study Cohort”: the diagnosis of CD in patients with JIA was established according to the standard diagnostic algorithm approved by the Scientific and Practical Council of the Russian Ministry of Public Health in 2024 [51]. Briefly, the diagnostic workup included ileocolonoscopy with biopsy of the terminal ileum, followed by esophagogastroduodenoscopy. The endoscopic and histological findings met the diagnostic criteria for Crohn’s disease with ileitis and colitis (L3) including documentation of non-caseating granulomas and/or transmural inflammation where present, according to the Paris classification (2010) [52].

For your convenience, please, see page 11 (lines 411-418)

Comment 3: Standardize the use of abbreviations. Ensure all abbreviations are defined at first mention in both the abstract and main text. Correct erroneous abbreviations such as “CRB” in the abbreviation list and unify terminology and abbreviation formatting throughout the manuscript.

Response 3: In response to this comment, we verified all abbreviations used in Abstract and Main text of the manuscript. Specifically, “CRB” was replaced by “CRP” (page 15, Abbreviations). “HC” relocated to the last paragraph of the Introduction section (page 3, line 128). “PsA” relocated from page 3 to page 2 (line 84). We have also verified that all abbreviations are now defined at first mention in both the Abstract and main text.

Comment 4: Performance validation of the predictive model is insufficient. The near‑perfect diagnostic performance derived from only 4 conversion events was based solely on direct calculation from a single dataset, without internal cross‑validation, let alone external validation. Such unvalidated “perfect performance” lacks clinical credibility in studies of clinical prediction models.

Response 4: We thank the reviewer for this important methodological observation. We fully agree that perfect separation metrics derived from a single dataset with only four conversion events do not support clinical deployment, and that internal cross-validation and external replication are essential for any formal clinical prediction model.

We would clarify that this study was designed as a hypothesis-generating longitudinal pilot to identify a preliminary biomarker signal, not as a clinical prediction model intended for immediate use. The Fisher’s Exact Test and PCA were employed strictly for exploratory cohort separation to highlight a distinct cytokine trajectory, not to estimate diagnostic accuracy for clinical application. We acknowledge that reporting "100% sensitivity/specificity" in this context may inadvertently imply clinical readiness, which was not our intent.

In response, we have revised the manuscript to:

1) Remove performance metrics that suggest clinical prediction model validation (e.g., "100% sensitivity/specificity") and replace them with descriptive phrasing such as "complete cohort separation in this pilot sample" (page 13, lines 506-507).

2) Explicitly state that this work represents an exploratory biomarker discovery phase (page 10, lines 330-332; page 11, 382-384). Before it can be considered for clinical use, the findings must be tested internally for robustness and validated independently in larger multicenter cohorts.

3) Temper all claims regarding clinical utility, framing the IL-1 signature strictly as a preliminary signal warranting further methodological validation.

In addition, our statistical team confirmed that formal cross-validation was not performed due to the pilot nature and small N. Accordingly, we added the following statement on page 10 to the “limitations”: The complete cohort separation observed here may reflect overfitting and should be validated in larger, multicenter prospective trials to establish universal clinical thresholds (lines 354-356).

These revisions ensure the manuscript accurately reflects its exploratory scope and aligns with current standards for biomarker discovery studies.

Reviewer 2 Report

Comments and Suggestions for Authors

This manuscript addresses a clinically relevant question: whether cytokine profiling can identify JIA patients at risk of developing Crohn's disease before gastrointestinal symptoms emerge. The longitudinal design, inclusion of disease-control groups, and mechanistic discussion represent real strengths. However, the circular subgroup definition is a fundamental methodological flaw and the consistently overcautious language needed to reflect an N=4 converter group is absent throughout. The authors should explicitly reframe this as a hypothesis-generating pilot study, address the subgroup definition methodology, report JIA subtypes, justify cytokine normalisation, and substantially temper their interpretive claims.

Detailed comments are listed below

TITLE

The title overstates the findings. "Predict" cannot be claimed from a pilot cohort of four converters.

ABSTRACT

"Statistical analysis used non-parametric tests, Fisher's Exact Test, and exploratory principal component analysis", the specific non-parametric tests used should be named here for transparency.

INTRODUCTION

1. Reference 2 (Michail et al., 2013) concerns genetic variants in CD, not the clinical description of age of onset and disease course described in lines 45–55. This citation appears misplaced, a clinical epidemiology reference is needed here.
2. There is no discussion of existing literature on cytokine profiling in JIA-to-IBD conversion specifically. If no such studies exist, this should be explicitly stated as the gap being addressed. The introduction would benefit from briefly surveying what is currently known about the gut-joint axis at the molecular level in paediatric populations.

RESULTS

1. The identification of four "converters" at baseline was predicated on elevated levels of IL-1α, IL-1β, and IL-36α; subsequently, Fisher's Exact Test was employed to confirm their progression to CD. This approach introduces a fundamental circularity into the methodology. By defining the subgroup post hoc based on the exact cytokine elevations intended to validate the predictive signature, the study compromises its analytical integrity. Threshold definition absent: No cytokine cut-off values are defined for what constitutes "significantly elevated" IL-1α, IL-1β, or IL-36α. Without defined thresholds, the signature Consequently, the reported 100% sensitivity and specificity do not constitute a validated biomarker discovery but are rather an artifact of the study design. This represents the most critical methodological deficiency within the manuscript and necessitates immediate, direct rectification.
2. Sample size disproportion: The selection criteria for JIA enrolment are not described in sufficient detail to evaluate this.
3. IL-18 not shown at baseline: IL-18 is measured at 12-month follow-up (Figure 2) but its baseline levels across cohorts are not presented. The absence of baseline IL-18 data is a significant gap , it is unknown whether IL-18 was already elevated in converters at baseline.
4. The figure 2 legend describes "PsA – psoriasis (N=20)" — psoriasis and psoriatic arthritis are distinct conditions. This should read "psoriatic arthritis."
5. The figure3 title reads "Principle component analysis" — the correct term is "Principal component analysis." This typographical error also appears in the Figure 3 heading at line 161.
6. The font on the figures is too small to be read.

DISCUSSION

1. Overclaiming from N=4 data (Major): The discussion repeatedly presents findings as established facts rather than pilot observations. Phrase such as "we propose a new clinical screening paradigm" dramatically overstate what four patients can support. The language throughout must be substantially tempered to reflect the exploratory nature of the data.
2.  Much of the mechanistic discussion is drawn entirely from external literature and is not supported by any data from this study. The authors measure serum cytokine concentrations only , they have no mechanistic data.
3.  The lengthy paragraph on biologics, and surgical management reads as a textbook summary that is disconnected from the study's findings

MATERIALS AND METHODS

1. how the four "high-cytokine" JIA patients were identified at baseline is never explicitly described. The absence of this information is a critical gap in the methods section.
2. JIA has at least seven ILAR-classified subtypes, each with distinct mmunological profiles. The subtype distribution of the JIA cohort and specifically the four converters is not reported.
3. Normalising serum cytokine concentrations to total serum protein is an unconventional approach. Standard practice is to report absolute concentrations (pg/mL) based on a serum volume dilution. The rationale for protein normalisation must be explicitly justified, and the impact of this choice on the reported group differences should be discussed.
4. The methods state that "Baseline differences in cytokine levels across the four cohorts (JIA, CD, JIA+CD, PsA, and HC)" were evaluated, this lists five groups, not four. This inconsistency reflects confusion between the baseline and follow-up analytical frameworks.
5. Abbreviation error: In the abbreviations table, "CRB" is listed for C-reactive protein — the correct abbreviation is CRP.

 

Author Response

Comments 1-5

THE REVIEWER’S OPENING STATEMENT:

This manuscript addresses a clinically relevant question: whether cytokine profiling can identify JIA patients at risk of developing Crohn's disease before gastrointestinal symptoms emerge. The longitudinal design, inclusion of disease-control groups, and mechanistic discussion represent real strengths. However, the circular subgroup definition is a fundamental methodological flaw and the consistently overcautious language needed to reflect an N=4 converter group is absent throughout. The authors should explicitly reframe this as a hypothesis-generating pilot study, address the subgroup definition methodology, report JIA subtypes, justify cytokine normalisation, and substantially temper their interpretive claims. Detailed comments are listed below.

Response: We thank the reviewer for this balanced and thoughtful assessment. We appreciate the recognition of clinical relevance, longitudinal design, and mechanistic discussion. We also fully agree with the five specific recommendations and have addressed each as follows:

Comment 1: Reframe as hypothesis-generating pilot

Response 1: We have revised the Abstract, Discussion, and Conclusions to explicitly describe this work as a pilot biomarker discovery study (page 11, lines 385-387) and to frame the IL-1 signature as a preliminary/exploratory biomarker candidate requiring validation (page 1, line 36; page 8, lines 241-243; page 10, lines 330-332).

Comment 2: Address circular subgroup definition.

Response 2: We clarify in the Methods that the "converter" subgroup was identified post-hoc through baseline cytokine screening, not pre-defined. We have added a methodological note acknowledging this exploratory identification and its implications for interpretation among the limitations on page 11 (lines 400-405).

Comment 3: Report JIA subtypes

Response 3: We have added a breakdown of JIA subtypes (polyarticular, oligoarticular, and pauciarticular) for both converters and non-converters in Supplementary Table 1. In the Methods section, we clarify: JIA subtypes were classified according to clinical practice at our center, distinguishing oligoarticular (3–4 joints) and pauciarticular (1–2 joints) presentations, as defined in the Union of Pediatricians of Russia clinical guidelines [53] (page 11, lines 409-412).

Comment 4: Justify cytokine normalization

Response 4: We have expanded the Methods to clarify that cytokine levels were normalized to total protein concentration to account for inter-individual variation in serum protein content, a standard practice in serum biomarker studies [56] (page 12, lines 470-472). We acknowledge that serum measurements do not fully capture tissue-level microenvironments and have stated this limitation to the Discussion: the current study focused on serum concentrations of cytokines. Although serum provides a non‑invasive window into systemic inflammation, it does not always accurately reflect the localized microenvironment of the intestinal mucosa [40-42] or synovial fluid [43]. Future studies would provide a deeper mechanistic understanding of the IL-1/IL-31 axis at the tissue level (page 10, lines 362-366).

Comment 5: Temper interpretive claims

Response 5: Throughout the manuscript, we have replaced definitive language (e.g., "predictive biomarker," "early-warning system") with cautious phrasing (e.g., "preliminary predictive signature," "exploratory biomarker candidate," "warrant further validation").

For your convenience, please see: page 1, lines 35-36; page 6, lines 184-186; page 8, lines 241-243; page 11, lines 382-384).

TITLE

Comment 6: The title overstates the findings. "Predict" cannot be claimed from a pilot cohort of four converters.

Response 6: We thank the reviewer for this important observation. We agree that the verb "predict" may not adequately reflect a level of clinical validation not yet supported by this pilot study. In response, we have revised the title to more accurately describe the exploratory nature of these findings.

Original: "IL-1 Family Cytokine Signature Predicts Crohn's Disease Conversion in Pediatric Juvenile Idiopathic Arthritis"

Revised: "Preliminary IL-1 Family Cytokine Signature for Crohn's Disease Onset in Pediatric Juvenile Idiopathic Arthritis" (page 1, lines 2 and 3).

ABSTRACT

Comment 7: "Statistical analysis used non-parametric tests, Fisher's Exact Test, and exploratory principal component analysis", the specific non-parametric tests used should be named here for transparency.

Response 7: We thank the reviewer for this suggestion. We have revised the Methods sentence in the Abstract to specify the non-parametric tests employed:

Original:

"Statistical analysis used non-parametric tests, Fisher's Exact Test, and exploratory principal component analysis."

Revised:

" Statistical analysis used Mann‑Whitney U tests for two‑group comparisons, Kruskal‑Wallis test with Dunn's post‑hoc for multi‑group comparisons, Fisher's Exact Test for categorical outcomes, and exploratory principal component analysis (PCA) " (page 1, lines 26-28).

This revision provides the requested transparency and maintains the conciseness of the Abstract and IJMS formatting requirements.

INTRODUCTION

Comment 8

1. Reference 2 (Michail et al., 2013) concerns genetic variants in CD, not the clinical description of age of onset and disease course described in lines 45–55. This citation appears misplaced, a clinical epidemiology reference is needed here.

Response 8: To address this reviewer's comment, we replaced the paper by Michail et al. (Reference 2) with a clinical resource that specifically describes the presentation of Crohn's disease in pediatric patients, including age of onset and typical disease course. The revised citation now appears on page 3 (line 57):

[2] Greenspon JY, Lipskar AM, Villalona GA, Chatoorgoon K. Crohn Disease. Pediatric Surgery NaT. American Pediatric Surgical Association; n.d. Available from: https://www.pedsurglibrary.com/apsa/view/Pediatric-Surgery-NaT/829045/all/Crohn_Disease. Accessed April 30, 2026.

This reference provides the clinical epidemiology context appropriate for the statement regarding pediatric CD presentation.

 

Comment 9:

1. There is no discussion of existing literature on cytokine profiling in JIA-to-IBD conversion specifically. If no such studies exist, this should be explicitly stated as the gap being addressed. The introduction would benefit from briefly surveying what is currently known about the gut-joint axis at the molecular level in paediatric populations.

Response 9:

We thank the reviewer for this important suggestion. We agree that explicitly framing the literature landscape strengthens the manuscript's rationale. We have revised the Introduction to address both points raised:

1) Regarding cytokine profiling in JIA-to-CD conversion:

To our knowledge, no longitudinal study has evaluated baseline serum cytokine signatures as predictors of Crohn's disease onset in pediatric JIA cohorts. Existing literature primarily consists of case reports that document IBD emergence following biologic therapy rather than investigating predictive cytokine trajectories. For example, IL-1 antagonists, TNF inhibitors, and anti-IL-17 agents have each been associated with new-onset IBD in JIA patients [14, 15, 17-19], highlighting cytokine modulation as a trigger but leaving baseline predictive patterns uncharacterized.

2) Regarding the molecular gut‑joint axis in pediatric populations:

We have added a brief survey of current understanding.

Literature on pediatric populations regarding cytokine profile changes associated with the development of CD in patients with juvenile idiopathic arthritis (JIA) remains limited. Existing studies are predominantly case reports (e.g., [14]) that focus on clinical presentation and symptom management rather than investigating underlying molecular events, including the dynamic expression of cytokines. This complexity is further compounded by the dynamic nature of pediatric rheumatic phenotypes, which frequently evolve over time. For example, systemic JIA may transition from an IL-1‑driven acute phase to an IL-17‑driven chronic state [15].

Although shifts in cytokine profiles are increasingly recognized as potential triggers for CD in patients with underlying inflammatory disorders such as JIA, the precise molecular mechanisms driving this phenomenon remain largely elusive [16]. Pharmacological evidence supports this link: Hügle et al. [15] reported new-onset IBD following treatment with IL-1 antagonists, whereas multiple studies describe IBD emergence after the therapy with TNF-α blockers across various JIA subtypes [17-19]. Similarly, Fauny et al. [20] documented IBD development following anti-IL-17 therapy.

In the latter case, the authors hypothesized that the neutralization of IL-17A may compromise the integrity of intestinal epithelial barrier, thereby precipitating mucosal inflammation. Further mechanistic insights come from studies highlighting IL-24, a downstream gene of the IL-17 signaling pathway [21], which acts as a regulator of mucosal inflammation by modulating the expression of membrane-bound mucins (MUC1, MUC3, and MUC4) [22]. Due to these critical cytokine shifts remain poorly characterized, cytokine-targeted biologics should be used with caution in patients predisposed to IBD. Conversely, monitoring baseline cytokine profiles associated with subclinical CD may provide a valuable tool for early risk stratification prior to or during the initial phases of biologic therapy (page 4, lines 90-115).

 

RESULTS

Comment 10:

1. The identification of four "converters" at baseline was predicated on elevated levels of IL-1α, IL-1β, and IL-36α; subsequently, Fisher's Exact Test was employed to confirm their progression to CD. This approach introduces a fundamental circularity into the methodology. By defining the subgroup post hoc based on the exact cytokine elevations intended to validate the predictive signature, the study compromises its analytical integrity. Threshold definition absent: No cytokine cut-off values are defined for what constitutes "significantly elevated" IL-1α, IL-1β, or IL-36α. Without defined thresholds, the signature Consequently, the reported 100% sensitivity and specificity do not constitute a validated biomarker discovery but are rather an artifact of the study design. This represents the most critical methodological deficiency within the manuscript and necessitates immediate, direct rectification.

Response 10:

We thank the reviewer for highlighting this critical methodological point. We agree that the perception of circularity and the absence of pre-specified cytokine thresholds must be addressed transparently. We have implemented the following clarifications and revisions:

1) Clarification of Study Chronology to Eliminate Circular Reasoning

The converter subgroup was not defined by baseline cytokine levels. All JIA patients were enrolled prospectively, baseline serum was collected, and patients were followed clinically for 12 months (page, 11, lines 400-405). CD diagnosis was established independently using endoscopic and histological criteria during follow-up (page, 11, lines 404-410). Baseline cytokine profiling was then analyzed retrospectively to identify molecular patterns that distinguished the four patients who subsequently developed CD from non-converters. Fisher’s Exact Test and PCA were employed strictly for exploratory retrospective pattern recognition, not for prospective validation or circular confirmation (page, 10, lines 330-332).

2) Absence of Pre-defined Thresholds

We confirm that no clinical cut-off values were established for IL-1α, IL-1β, or IL-36α. This study was designed as a hypothesis-generating pilot biomarker discovery effort, not as a diagnostic threshold validation study. We acknowledge that applying clinical decision thresholds requires prospective definition and independent validation (page 10, lines 354-356).

3) Concrete Manuscript Revisions

In direct response to this comment, we have made the following changes:

(1) We removed all claims of "100% sensitivity and specificity" from the Abstract, Results, and Discussion, replacing them with descriptive phrasing such as "complete exploratory cohort separation in this pilot sample" (page 13, lines 515-516).

(2) We added a methodological clarification in the Methods section (page 11, lines 400-410):

"Notably, the JIA cohort included four individuals who subsequently received a secondary diagnosis of comorbid CD during the 12-month follow-up period. Accordingly, the converter subgroup (N = 4) was identified retrospectively: all JIA patients were enrolled with baseline serum collection, followed by 12-month clinical follow-up during which CD diagnosis was established independently using endoscopic/histological criteria. These findings should be interpreted as hypothesis-generating and require validation in pre-specified, multicenter prospective cohorts".

(3) We expanded the Limitations section (page 10, lines 356-361) to directly address this concern:

"The cytokine pattern was identified retrospectively after clinical conversion was confirmed. Because no pre-defined cut-offs were established and the analysis was performed post hoc, the observed separation represents a hypothesis-generating signal rather than a validated diagnostic threshold. Prospective studies with pre-specified thresholds and independent cohorts are required before clinical consideration can be justified."

 

Comment 11:

1. Sample size disproportion: The selection criteria for JIA enrolment are not described in sufficient detail to evaluate this.

Response 11:

We thank the reviewer for this important request for methodological clarity. We have expanded the Methods section to provide a complete description of JIA enrollment criteria and to contextualize the sample size distribution.

 

1) Enrollment criteria added to Methods (page 11, line 423 –page 12, 433):

Children aged 6–16 years with confirmed diagnoses were eligible: JIA (including psoriatic JIA) per ILAR criteria [54]; CD per Paris pediatric IBD classification [52] and PsA (non-JIA) per CASPAR criteria [55]. Inclusion required active disease requiring systemic therapy, no prior concurrent autoimmune/autoinflammatory conditions (beyond index diagnosis), availability of a baseline serum sample, and 12-month follow-up commitment. Healthy controls were age- and sex-matched individuals without chronic inflammatory conditions. Exclusion criteria were the following: concurrent conditions beyond the index diagnosis, active infection/malignancy, recent immunosuppression, or incomplete data. Consecutive enrollment occurred at Filatov Clinical Institute of Children's Health, Sechenov University, between 13 February 2024 and 12 March 2024.

2) Regarding sample size disproportion:

The observed distribution (N = 25 total JIA patients; 4 converters, 21 non-converters) reflects the natural incidence of CD conversion in pediatric JIA cohorts rather than selective enrollment. All eligible patients meeting the above criteria during the recruitment window were included; no post-hoc exclusions were applied. We acknowledge that the small converter subgroup limits statistical power and generalizability, a limitation explicitly stated in the manuscript (Limitations section, page 10, lines 351-354). This pilot cohort was designed for hypothesis generation, not definitive validation of biomarkers.

 

Comment 12:

1. IL-18 not shown at baseline: IL-18 is measured at 12-month follow-up (Figure 2) but its baseline levels across cohorts are not presented. The absence of baseline IL-18 data is a significant gap , it is unknown whether IL-18 was already elevated in converters at baseline.

Response 12:

We thank the reviewer for this important observation. We agree that baseline IL-18 data would strengthen interpretation of its role in the conversion trajectory.

Clarification of assay scope:

We have clarified in the Methods section (4.4 ELISA) that baseline cytokine profiling was limited to the IL-1 family axis (IL-1α, IL-1β, and IL-36α) and IL-6 based on prior mechanistic hypotheses. IL-18, IL-27, IL-31, and IL-37 were assessed at 12-month follow-up to explore longitudinal dynamics but were not included in the baseline discovery panel due to sample volume constraints and the focused exploratory design of this pilot phase. This limitation is also explicitly acknowledged in the Discussion.

1) To be specific, we added a clarification to the Methods section (page 12, lines 472-477):

Baseline cytokine profiling was limited to the IL-1 family axis (IL-1α, IL-1β, and IL-36α) and IL-6 based on prior mechanistic hypotheses. IL-18, IL-27, IL-31, and IL-37 were assessed at 12-month follow-up to explore longitudinal dynamics but were not included in the baseline discovery panel due to sample volume constraints and the focused exploratory design of this pilot phase.

2) Expanded the Limitations section (page 10. Lines 366-370) to explicitly acknowledge this gap:

Baseline IL‑18 levels were not assessed in this pilot cohort. Consequently, it remains unclear whether IL‑18 elevation preceded clinical conversion or emerged during disease progression. Prospective studies utilizing broader baseline cytokine panels are needed to clarify the temporal relationship between IL‑18 dynamics and CD onset in JIA patients.

 

Comment 13:

1. The figure 2 legend describes "PsA – psoriasis (N=20)" — psoriasis and psoriatic arthritis are distinct conditions. This should read "psoriatic arthritis."

Response 13:

We thank the reviewer for this important correction. We agree that psoriasis and psoriatic arthritis are distinct clinical entities, and we have revised the legend of figure 2 accordingly (page 6, lines 171-172). The text now reads: "PsA – psoriatic arthritis (N = 20)." This revision ensures terminological precision throughout the manuscript and all figure legends.

 

Comment 14:

1. The figure 3 title reads "Principle component analysis" — the correct term is "Principal component analysis." This typographical error also appears in the Figure 3 heading at line 188 (page 7).

Response 14:

We thank the reviewer for identifying this typographical error. We have corrected "Principle component analysis" to "Principal component analysis" in the Figure 3 legend (previously line 161). These instances have been updated to ensure terminological consistency throughout the manuscript.

 

Comment 15:

1. The font on the figures is too small to be read.

Response 15:

We thank the reviewer for this helpful feedback. To improve readability, we have redesigned all figures with a 50% increase in font size. All figure panels have been updated in the revised manuscript and uploaded as high-resolution files to ensure clarity in both print and digital formats.

For your convenience, please see pages 5 (Figure 1), 6 (Figure 2), and 7 (Figure 3) in the revised manuscript.

 

DISCUSSION

Comment 16:

1. Overclaiming from N=4 data (Major): The discussion repeatedly presents findings as established facts rather than pilot observations. Phrase such as "we propose a new clinical screening paradigm" dramatically overstate what four patients can support. The language throughout must be substantially tempered to reflect the exploratory nature of the data.

Response 16:

We thank the reviewer for this critical observation. We have systematically revised the Discussion, Abstract, and Conclusions to replace definitive claims with cautious, pilot-appropriate phrasing. Specifically, we removed "we propose a new clinical screening paradigm" and replaced it with "we propose that this exploratory biomarker framework could inform future screening strategies for JIA" (page 10, lines 330-332). All assertive language has been tempered to consistently reflect the exploratory nature of this N = 4 converter subgroup.

 

Comment 17:

1. Much of the mechanistic discussion is drawn entirely from external literature and is not supported by any data from this study. The authors measure serum cytokine concentrations only, they have no mechanistic data.

Response 17:

We thank the reviewer for this important observation regarding mechanistic interpretation. We agree that serum cytokine data alone cannot establish tissue-level causality. In response, we have added a clarifying statement at the end of the first Discussion paragraph:

"We emphasize that these interpretations derive from serum cytokine measurements alone and do not constitute direct mechanistic evidence. Rather, they align with prior literature linking IL-1/IL-31 axis dysregulation to barrier dysfunction, generating testable hypotheses for future tissue-level investigations" (page 7, lines 206-210).

Additionally, we have tempered definitive mechanistic phrasing throughout (e.g., replacing "is rooted in" with "may be preceded by" – page 8, lines 259-261; and "prime the mucosa" with "may be associated with increased mucosal susceptibility" – page 7, lines 217-220). These revisions ensure external literature is cited for contextual hypothesis generation, not as mechanistic proof derived from our pilot cohort.

 

Comment 18:

1. The lengthy paragraph on biologics, and surgical management reads as a textbook summary that is disconnected from the study's findings

Response 18:

We thank the reviewer for this helpful observation. We agree that the detailed summary of pediatric CD management, while clinically relevant, was not sufficiently integrated with our biomarker findings and risked distracting from the study's core message. In response, we have substantially condensed this paragraph and reframed it to explicitly connect current therapeutic strategies to the potential clinical utility of the proposed IL-1/IL-36 signature.

Original paragraph (in full):

"Management of pediatric CD follows a risk-stratified approach centered on mucosal healing, growth preservation, and corticosteroid minimization. EEN is the preferred first-line induction therapy for active luminal disease, achieving 80% remission rates with superior mucosal healing compared to corticosteroids [38,39]. For EEN non-responders, corticosteroids (with budesonide preferred for ileocecal disease) serve as an induction bridge. Although maintenance traditionally involves thiopurines or methotrexate, anti-TNF biologics (infliximab or adalimumab) are already increasingly utilized as early first-line therapy for high-risk features, such as perianal involvement, penetrating disease, or severe growth failure [23,39]. Ustekinumab and vedolizumab serve as effective third-line options for patients who have failed anti-TNF therapies [40,41]. Surgery (resection or strictureplasty) is reserved for complications or refractory localized disease to facilitate catch-up growth [42]. Finally, proactive therapeutic drug monitoring (TDM) and treat-to-target (T2T) strategies are essential to optimizing long-term outcomes in children [43]."

 

Revised, integrated text:

" The proposed IL-1/IL-36 signature could inform risk-stratified treatment selection in pediatric JIA patients at risk for CD conversion. For example, early identification of high-risk patients might support preemptive referral for exclusive enteral nutrition (EEN) or consideration of 'dual-action' biologics (e.g., anti-IL-1 or anti-TNF-α agents) that target both synovial and intestinal inflammation [23,38,39]. This biomarker-guided approach aligns with current treat-to-target strategies while potentially enabling earlier intervention before irreversible mucosal damage occurs (page 10, lines 344-350)".

 

Comment 19:

MATERIALS AND METHODS

1. how the four "high-cytokine" JIA patients were identified at baseline is never explicitly described. The absence of this information is a critical gap in the methods section.

 

Response 19:

We thank the reviewer for this important request for methodological clarity. We confirm that "high-cytokine" status was not a pre-specified selection criterion for enrollment or analysis. Rather, the four JIA patients who later converted to CD were identified retrospectively during data analysis as having baseline serum levels of IL-1α, IL-1β, and IL-36α that were visually and statistically distinct from the non-converter JIA cohort and healthy controls.

Respectively, we clarified it in the “Methods”

"Accordingly, the converter subgroup (N = 4) was identified retrospectively: all JIA patients underwent baseline serum collection prior to a 12-month clinical follow-up, during which CD diagnosis was established independently using standardized endoscopic and histological criteria. Their baseline cytokine profiles (IL-1α, IL-1β, and IL-36α) were descriptively characterized as "elevated" relative to the JIA non-converter median and healthy control range; no pre-specified numerical thresholds were applied. This retrospective, hypothesis-generating approach is consistent with the pilot design of this study. Consequently, these findings should be interpreted as exploratory signals requiring validation in pre-specified, multicenter prospective cohorts (page 11, lines 402-410)."

We would also clarify why this framing is appropriate:

1) In pilot biomarker discovery studies, exploratory pattern recognition (e.g., visual inspection of distributions, comparison to reference ranges) is standard prior to prospective threshold validation.

2) Pre-specified cut-offs require independent cohorts and ROC analysis — steps reserved for future validation phases.

Cumulatively, these revisions ensure full transparency regarding the retrospective, exploratory nature of the "high-cytokine" observation while preserving the scientific validity of the pilot signal.

 

Comment 20:

1. JIA has at least seven ILAR-classified subtypes, each with distinct mmunological profiles. The subtype distribution of the JIA cohort and specifically the four converters is not reported.

Response 20:

We thank the reviewer for this important observation. Detailed information on JIA subtypes for the full cohort and the four converters has been added to Supplementary Table 1. Due to the small converter subgroup (N = 4), subtype-specific analysis remains exploratory. However, this transparency enables readers to assess potential immunological heterogeneity and informs future validation studies.

 

Comment 21:

1. Normalising serum cytokine concentrations to total serum protein is an unconventional approach. Standard practice is to report absolute concentrations (pg/mL) based on a serum volume dilution. The rationale for protein normalisation must be explicitly justified, and the impact of this choice on the reported group differences should be discussed.

Response 21:

We thank the reviewer for this observation. Cytokine concentrations were normalized to total serum protein (pg/mg) because clinical conditions, including systemic inflammation, hydration status, and acute-phase responses, directly alter total serum protein levels in pediatric patients. Normalization to mg protein controls for this biological variability and strengthens internal consistency for exploratory comparisons within this heterogeneous cohort. We also acknowledge that absolute concentrations (pg/mL) facilitate cross-study benchmarking and future validation studies will report both metrics.

 

Comment 22:

1. The methods state that "Baseline differences in cytokine levels across the four cohorts (JIA, CD, JIA+CD, PsA, and HC)" were evaluated, this lists five groups, not four. This inconsistency reflects confusion between the baseline and follow-up analytical frameworks.

Response 22:

We thank the reviewer for identifying this inconsistency. We confirm that the study enrolled four primary cohorts: JIA, CD, PsA, and HC. The "JIA+CD" group represents a retrospective subgroup of the JIA cohort (N = 4 converters) identified during follow-up, not a fifth independent baseline group. We have corrected the Methods section to reflect this hierarchical analytical framework.

Corrected text in the Methods section (4.6 Statistics, page 13)

"Baseline differences in cytokine levels across the four primary cohorts (JIA, CD, PsA, and HC) were evaluated using the Kruskal-Wallis test, followed by Dunn’s post-hoc test with Šidák correction for multiple comparisons. Due to the small size of the JIA converter subgroup (N = 4), its baseline cytokine profile was analyzed descriptively to explore patterns associated with subsequent CD conversion (page 13, lines 495-499)".

 

Comment #23:

1. Abbreviation error: In the abbreviations table, "CRB" is listed for C-reactive protein — the correct abbreviation is CRP.

Response #23

We thank the reviewer for identifying this typographical error. We have corrected "CRB" to "CRP" (C-reactive protein) in the Abbreviations table and verified that all instances of this abbreviation throughout the manuscript now use the standard acronym (page 15).

 

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors The manuscript has been improved to warrant publication in IJMS.

Reviewer 2 Report

Comments and Suggestions for Authors

No further comments