This study is a secondary analysis of the Chronic Inflammatory Disease, Lifestyle and Treatment Response (BELIEVE) study, and methods are described in Overgaard et al. [25]. The study complies with the Declaration of Helsinki, and was approved by the Regional Committees on Health Research Ethics for Southern Denmark in 2016 (S-20160124) as well as the Danish Data Protection Agency (2008-58-035). Each participant signed an informed consent before inclusion to the study. The study was registered at Clinical.Trials.gov (NCT03173144), and a protocol was published before the initiation of the study [26]. A statistical analysis plan (SAP) was specified before initiating statistical analyses (Supplementary File). Findings are reported according to the STROBE statement [27].

As previously described, this prospective study took place at nine Danish clinical centres [25, 26]. Patients with CIDs who were about to initiate treatment or change biologic agents were invited to participate. The included CIDs were: rheumatoid arthritis (RA), psoriatic arthritis (PsA), psoriasis (PSO), axial spondyloarthritis (AxSpA), Crohn’s disease (CD), and ulcerative colitis (UC). To be eligible, participants had to be minimum 18 years of age, and able to understand and sign an informed consent.

Treatment response was based on specific disease activity measures for the individual CIDs. At the time of inclusion, all patients had active disease, according to individual CID guidelines, to initiate treatment or change to other biologic agents. Secondary outcomes included participant-reported changes in health-related quality of life as well as the physician’s global assessment and C-reactive protein (CRP) levels.

Groups were defined based on the serum MFAP4 levels, detected using the AlphaLISA technique, as described in Wulf-Johansson, H., et al. [22]. Serum samples were collected at baseline, before treatment initiation, and again 14–16 weeks following initiation of treatment. The primary outcome was the proportion of patients with a positive treatment response to biologic therapy (14–16 weeks) after treatment initiation as a group as well as individual CIDs. The specific criteria for positive treatment response varied across the CID conditions [26]: CD: clinical remission, defined as Harvey-Bradshaw Index of 4 or less; UCUC: clinical remission, defined as Mayo Clinic Score of 2 or less (with no individual sub-score of > 1); RA: clinical response, defined as at least a 20% improvement according to the criteria of the American College of Rheumatology (ACR20) [28]; AxSpa: clinical response, defined as at least a 20% improvement according to the Assessment of Spondyloarthritis International Society (ASAS20) [29]; PsA: clinical response, defined as at least a 20% improvement according to the criteria of ACR20; PSO: clinical response, defined as at least a 75% improvement in Psoriasis Area and Severity Index (PASI 75).

Secondary generic outcomes included changes from baseline to follow-up in measures of health-related quality of life (SF-12; the physical and mental component summaries (PCS and MCS, respectively), the short health scale consists of four components: symptom burden, functional status, disease-related burden and general wellbeing), C-reactive protein, and physician’s global assessment. Additionally, the proportion of patients continuing biologic treatment beyond the follow-up period served as a secondary outcome measure. Other secondary non-generic outcomes listed at clinical.trials.gov included changes from baseline to follow-up in disease scores (e.g., ∆HBI score, ∆Mayo Clinic score, ∆tender joint count, ∆swollen joint count, ∆PASI score, etc., see SAP in supplementary).

Participants’ blood samples were collected at inclusion prior to initiating therapy with biologics. Participants with missing blood samples at baseline were excluded from the analysis population. The examination program at inclusion also included health-related quality of life questionnaires, as well as clinical assessment. After 14–16 weeks of treatment, the participants were re-evaluated clinically for treatment response, blood samples were obtained, and health-related quality of life questionnaires were answered again. The inclusion period was between September 1st 2017 and March 31st 2020. Participants with available blood samples at baseline, were included in the intention to treat (ITT) population. They were stratified into two categories: the upper tertile of serum MFAP4 (High MFAP4) and a combined group joining the middle and lower tertiles of serum MFAP4 (Other MFAP4). The cut-off was pragmatically decided, assuming that a high MFAP4 level indicated high inflammatory disease activity, that could be blocked by biological treatment.

The study data were collected, pseudo-anonymized and managed using Research Electronic Data Capture database (REDCap) tools hosted at the Open Patient Data Explorative Network – Odense University Hospital storage facility (OPEN) [30]. Patients were included and evaluated after signed informed consent (baseline). Attending physicians assessed the disease activity of the participants and filled out CID-specific standardised data forms on disease activity at baseline and follow-up. These data forms were the Harvey Bradshaw Index (HBI, CD), the Mayo Clinic Score (UC), the Simple Clinical Colitis Activity Index (SCCAI, UC), the Simplified Disease Activity Index (SDAI, RA and PsA), the 46 joint count (RA), the Bath Ankylosing Spondylitis Metrology Index (BASMI, axSpA), the Psoriasis Area and Severity Index (PASI, PsO), the 66/68 joint count (PsA), as well as a global assessment (all). Qualified study personnel collected project-relevant clinical information from the patient’s medical records.

Patients reported on the generic quality of life measures [the Short Form Health Survey-12 (SF-12) and the Short Health Scale] as well as CID-specific quality of life measures [RA and PsA; Health Assessment Questionnaire Disability Index (HAQ-DI), PsO; Dermatology Life Quality Index (DLQI)]. Furthermore, patients reported disease activity scores relevant for AxSpA; Bath Ankylosing Spondylitis Functional Index (BASFI) and Bath Ankylosing Spondylitis Disease Activity Index (BASDAI). Smoking status was registered via electronic questionnaire at baseline and again at follow-up. More details are available in the statistical analysis plan (SAP) (see Supplementary File 1).

The study was approved by The Regional Committees on Health Research Ethics for Southern Denmark (S-20160124), and the processing of personal data was notified to and approved by the Region of Southern Denmark and listed in the internal record (18/13682) cf. Art 30 of the EU General DATA Protection Regulation. As reported by Overgaard et al. [25], two patient associations (the Danish Colitis-Crohn’s Association and the Danish Psoriasis Association) and three patient representatives diagnosed with RA were involved in developing recruitment plans as well as the design of the study and communication about the study to patient members.

Statistical analyses followed the pre-specified statistical analysis plan (SAP) based on the analysis outlined in the original protocol and the first study published on this cohort [25, 26]. The treatment response was explored in relation to baseline MFAP4 levels, stratified into the upper tertile, 33.3% (High MFAP4) and lumping the middle and lower tertile, 66.67% (Other MFAP4). Standardised differences (StdD) were calculated to compare the distribution of baseline covariates between groups [31]: a standardised difference above 0.5 StdD-units was considered indicative of a potentially significant imbalance and difference between exposure groups at baseline. P-values were calculated using the Student’s t-test, Wilcoxon rank sum, or chi-squared test, as appropriate.

The independent contribution of the CIDs to the primary composite outcome was visualised in a forest plot: In the ITT population (using a simplistic non-responder imputation for missing outcome data), the ORs and 95% confidence interval (95% CI) of clinical responses in the groups within each CID were calculated and pooled the estimates using random-effects meta-analysis (STATA version 16.1, the “metan” package) as heterogeneity across CIDs was anticipated. Differences in the proportions of participants responding between groups were analysed using two different logistic regression analysis models: (i) in the “CID adjusted model,” an adjustment was made only for CID; whereas, in the adjusted model (ii) an adjustment for CID, sex, age, smoking status (ordinal scale: never, former, occasional, and current), as well as BMI category (ordinal scale: underweight, normal, overweight, and obese) was made, which were a priori considered potential confounding variables (See SAP).

The estimates for the continuous outcomes were reported as least squares (LS) means with standard errors. All P-values and 95% CIs were estimated and reported as two-sided; a P < 0.05 was considered potentially indicative of a statistically significant finding.

To assess the predictive value of serum MFAP4 levels on clinical response on a continuous scale (i.e., independent of arbitrary tertile thresholds), a receiver operating characteristic (ROC) curve analyses in an attempt to improve the threshold of serum MFAP4 levels best predicting a positive treatment response. The ROC curve was created by plotting the true positive rate (sensitivity) against the false positive rate (1 - specificity) at various threshold serum MFAP4 values. Each point on the curve represents a different threshold for classifying positive and negative instances. Youden’s index was used to find the optimal threshold for all CIDs and individual groups.

The datasets generated during and analysed during the current study are available from the corresponding author on reasonable request. Relevant authorities e.g., the Danish Data Protection Agency, must approve the data requestors, in adherence to GDPR regulations.