Treatment-related changes in serum neutrophil gelatinase-associated lipocalin (NGAL) in psoriatic arthritis: results from the PIPA cohort study

Objectives Obesity and psoriatic arthritis (PsA) have a complicated relationship. While weight alone does not cause PsA, it is suspected to cause worse symptoms. Neutrophil gelatinase-associated lipocalin (NGAL) is secreted through various cell types. Our objective was to assess the changes and trajectories in serum NGAL and clinical outcomes in patients with PsA during 12 months of anti-inflammatory treatment. Method This exploratory prospective cohort study enrolled PsA patients initiating conventional synthetic or biological disease-modifying anti-rheumatic drugs (csDMARDs/bDMARDs). Clinical, biomarker, and patient-reported outcome measures were retrieved at baseline, and 4 and 12 months. Control groups at baseline were psoriasis (PsO) patients and apparently healthy controls. The serum NGAL concentration was quantified by a high-performance singleplex immunoassay. Results In total, 117 PsA patients started a csDMARD or bDMARD, and were compared indirectly at baseline with a cross-sectional sample of 20 PsO patients and 20 healthy controls. The trajectory in NGAL related to anti-inflammatory treatment for all included PsA patients showed an overall change of −11% from baseline to 12 months. Trajectories in NGAL for patients with PsA, divided into treatment groups, showed no clear trend in clinically significant decrease or increase following anti-inflammatory treatment. NGAL concentrations in the PsA group at baseline corresponded to the levels in the control groups. No correlation was found between changes in NGAL and changes in PsA outcomes. Conclusion Based on these results, serum NGAL does not add any value as a biomarker in patients with peripheral PsA, either for disease activity or for monitoring.

Psoriatic arthritis (PsA) is a heterogeneous disease with multiple clinical manifestations, such as peripheral arthritis, axial involvement, enthesitis, dactylitis, and dermatological manifestations such as psoriasis (PsO) with or without nail involvement (1).It is a chronic immune-mediated inflammatory disease in which the body's immune system attacks otherwise healthy cells and tissues.One of the most evident parts of the treatment strategy for patients with PsA is pharmacotherapy with anti-inflammatory treatment.Although the aetiology and pathogenesis of PsA are still not fully understood, tumour necrosis factor-α (TNF-α), a proinflammatory cytokine, plays a central role by regulating several inflammatory pathways and immune functions.More recent discoveries on the role of interleukin (IL)-17 and IL-23 in PsA pathogenesis support the use of drugs with these two cytokines as targets (2).The recommended treatment, depending on the clinical manifestations, includes conventional synthetic diseasemodifying anti-rheumatic drugs (csDMARDs) and biological disease-modifying anti-rheumatic drugs (bDMARDs).Among patients with PsA, obesity is more prevalent than in non-PsA patients (3).Adipose tissue is involved not only in fat storage but also in releasing adipokines, such as pro-inflammatory lipocalins (4).Because of the inflammatory aspects of adipose tissue, it has been hypothesized that obesity could function as an independent risk factor for developing inflammatory rheumatic diseases (5)(6)(7).It has also been hypothesized that obesity could affect disease activity and treatment response (8).Previous studies have found that neutrophil gelatinase-associated lipocalin (NGAL) is closely related to obesity (9,10).
NGAL is a glycoprotein secreted by various cells, e.g. during inflammation.It is synthesized from epithelial cells and chondrocytes.Previous studies have concluded that NGAL concentrations are significantly higher in patients with PsA than in healthy controls, and NGAL has appeared as a possible useful biomarker (11,12).The studies in PsA patients found no immediate correlation with disease activity.However, IL-17 inhibitor and TNF-α inhibitor treatment significantly lowered NGAL in PsA patients after 6 months of treatment (13).Understanding of the role of NGAL in PsA is still limited, but it has previously been recognized to contribute to the pathogenesis of PsO by modulating neutrophil function (14,15).The heterogeneity of PsA challenges accurate and early diagnosis.Delayed diagnosis is associated with poorer physical function and increased risk of invalidity.Early diagnosis and accurate therapeutic intervention, such as using DMARDs, before the onset of structural damage, can prevent permanent joint injury and invalidity (16).C-reactive protein (CRP) is one of the most frequently used biomarkers for inflammation and part of several core outcome measures for PsA.However, despite active inflammatory disease, many patients with PsA do not have elevated CRP.Ogdie et al (17) concluded that there exists an unmet need for biomarkers that can identify PsA early and monitor the treatment response.
The primary objective of this study was to assess the changes and trajectories in NGAL in relation to initiating an anti-inflammatory treatment among patients with PsA.As a secondary objective, we wanted to test the diagnostic value of NGAL in patients with PsA compared with NGAL observed in patients with PsO and in healthy controls.

Study design
This study was designed as a prospective cohort study emanating from the Parker Institute's psoriatic arthritis cohort (the PIPA cohort) (18).The cohort was approved by the Danish Ethics Committee (H-15009080), registered at ClinicalTrials.gov(NCT02572700), and conducted in accordance with the General Data Protection Regulation (2012-58-0004) approved by the Capital Region of Denmark.The specific study objectives and the prespecified statistical analysis plan (SAP) are provided in the Supplementary material (Appendix 1).

Participants
We planned to obtain data on 120 PsA patients, 20 PsO patients, and 20 healthy controls (corresponding to an allocation ratio of 3:1) recruited between September 2015 and September 2021.Written informed consent was obtained from all included participants.For the PIPA cohort, eligible PsA patients were aged ≥ 18 years, fulfilled the ClASsification criteria for Psoriatic ARthritis (CASPAR), and presented with peripheral PsA manifestations (defined as the joints of the extremities) (19), and were scheduled to start a csDMARD [methotrexate (MTX), sulfasalazine (SZZ), or leflunomide (LEF)] or a bDMARD [TNF-α inhibitor (adalimumab, infliximab, etanercept, or certolizumab), IL-17 inhibitor (secukinumab or ixekizumab), or IL-23p19 inhibitor (risankizumab)].Inclusion criteria for the control groups were (i) cutaneous PsO patients without arthritis and no systemic treatment; and (ii) healthy controls.Exclusion criteria for all groups were pregnancy, peripheral neuropathy, demyelinating disease, stroke within 3 months from baseline, and other inflammatory rheumatic diseases, besides PsA.Furthermore, participants were not eligible if they could not pause glucocorticoids, centrally acting analgesics, and non-opioid analgesics at 21 days, 7 days, and 1 day before baseline, respectively.For this study, we additionally required that all participants had samples in the corresponding biobank.Baseline visits were defined as the time window from 14 days before until 10 days after initiating or switching csDMARD/bDMARD.All participants attended a baseline visit and PsA patients further attended follow-up visits at 4 and 12 months (duration from baseline), respectively, where assessments and biosampling were conducted.

Variables and outcome measures
Clinical assessments and blood samples.Trained healthcare professionals performed the interview and clinical examination, which consisted of patient demographics [age, sex, higher education, work status, body mass index (BMI), number of comorbidities (diabetes, hypertension, cardiovascular disease, osteoporosis, cancer, anxiety/depression), and PsO or PsA disease duration], 66/68 swollen/tender joint count (SJC66/TCJ68), Spondyloarthritis Research Consortium of Canada Enthesitis Score (SPARCC), the number of fingers and toes affected by dactylitis, nails (number involved), and Psoriasis Area Severity Index (PASI) in psoriasis vulgaris, and medical profile.
Blood samples for measuring high-sensitivity CRP and kidney function [estimated glomerular filtration rate (eGFR)] were collected and processed by trained laboratory technicians and analysed according to standard procedures.NGAL was measured in serum retrieved from whole blood centrifuged at 2000 g for 15 min, aliquoted into Eppendorf tubes and stored at −80°C until analysis.The concentration was quantified by a high-performance singleplex immunoassay (R-PLEX Human NGAL/LCN2 Assay with Sector™ plates; Meso Scale Discovery®) according to the manufacturer's protocol (Meso Scale Discovery [MSD], Maryland, USA).A dilution factor of 500 × was established in an initial pilot to ensure NGAL measurements within the instrument detection range.All samples were run in duplicate.If the coefficient of variation (CV) exceeded 20%, samples were rerun.
Patient-reported outcomes, disease activity, and response criteria.Patient global disease, pain, and fatigue scores evaluated on a visual analogue scale (VAS) from 0 to 100 mm.Data from the Health Assessment Questionnaire (HAQ) to calculate the Disability Index (HAQ-DI) were obtained from the electronic questionnaire accessible on computer touch screens at the study site.Disease Activity in Psoriatic Arthritis (DAPSA) was calculated using SJC66 and TJC68, the patient's VAS global and VAS pain scores, and CRP.A DAPSA score of ≤ 4 represents remission, ≤ 14 represents a state of low disease activity, ≤ 28 represents moderate disease activity, and > 28 high disease activity (20).The American College of Rheumatology (ACR) response criteria for treatment response are based on the percentage of improvement in SJC and TJC, and at least three of the following five parameters: patient assessment, physician assessment, pain scale, physical function, and CRP.The outcome is dichotomous based on the percentage of improvement that the patient may fulfil.

Statistical methods
Categorical data were summarized by numbers and percentages, and continuous data by the mean and standard deviation (sd).All inferential statistics are presented as two-sided 95% confidence intervals (95% CI) and p values.We did not apply explicit adjustments for multiplicity; rather, we analysed and interpreted the secondary objectives in a hierarchical order.Missing data were handled indirectly, and statistically modelled using mixed models for analysing repeated measurements (21).Contrasts between specific time points and medication groups (csDMARDs, and TNF-α, IL-17, and IL-23 inhibitors) were estimated based on the least squares means derived from the mixed effects models (i.e. at 4 and 12 months from baseline).
In the repeated-measures mixed effects models, the primary analyses for changes and trajectories in NGAL were based on the intention-to-monitor (ITM) population, i.e. all enrolled patients with relevant baseline measures collected.Missing data were handled indirectly and statistically modelled using repeated-measures linear mixed effects models, assuming that data are missing at random.Accordingly, participants belonging to a specific medication group (csDMARDs, and TNFα, IL-17, and IL-23 inhibitors) at baseline were followed, assessed, and analysed as members of that group, irrespective of their adherence to the planned course of treatment.To evaluate changes in the continuous outcome variables [BMI, CRP, eGFR, SJC66/ TJC68, DAPSA, PASI, SPARCC, number of fingers and toes affected by dactylitis, nails (number involved), patients' VAS global, VAS pain, VAS fatigue, and functional disability (evaluated by HAQ-DI)], we used similar analysis models as for the primary objective.The associations between changes in the aforementioned outcomes and changes in NGAL in patients with PsA were evaluated with Spearman's correlations.All data were analysed using SAS version 9.4 (SAS Institute, Cary, NC, USA).

Results
Figure 1 shows the flow of participants through the stages of the cohort study, depicting the study's design, milestones, and conduct.In total, 206 PsA patients, 39 PsO patients, and 20 healthy controls were assessed for eligibility; 117 PsA patients, 20 PsO patients, and 20 healthy controls were included at baseline between September 2015 and September 2021.Over the 12 month longitudinal period, 117 and 87 PsA patients completed the 4 and 12 month follow-up visits, respectively.
As presented in Table 1, demographic characteristics were generally well balanced across the groups, with the exception of mean disease duration and percentage in full-time work.Differences in the clinical characteristics were observed, as expected, with higher numbers of SJC, TJC, enthesitis, and dactylitis in the PsA group and higher PASI and number of nails affected in both the PsA and PsO group compared to the healthy controls.The PsA and PsO groups further rated their VAS global, VAS pain, and VAS fatigue higher than the healthy control group.Serum NGAL concentrations were nearly identical at baseline in patients with PsA compared to others (patients with PsO and healthy controls).Serum CRP levels were moderately higher in the group of patients with PsA.

Changes in NGAL in PsA patients
Figure 2(A) illustrates the NGAL trajectory from baseline to 12 months for all included patients with PsA independent of treatment group, based on estimates from repeated-measures linear mixed models.The mean serum NGAL concentration at baseline was 427 845 pg/mL.After 4 months of treatment, the mean concentration decreased by 7% to 398 034 pg/ mL, and after 12 months of treatment, the concentration further decreased by 4% to 382 504 pg/mL.In total, we observed an overall change of −11% after 12 months of DMARD treatment in all patients with PsA. Figure 2(B) depicts the NGAL trajectories from baseline to 12 months for the PsA patients divided into medication groups, based on estimates from repeated-measures linear mixed models.In the csDMARD group, we observed a decrease from baseline to 12 month followup of 12%.For patients in the TNF-α inhibitor group, the change from baseline to 12 month follow-up was −2%.In the IL-17 inhibitor group, in contrast, we observed an increase of almost 21% after 12 months of treatment.The most notable decrease from baseline to 12 month follow-up was observed in the IL-23(p19) group, with a decrease of 23%.
Table 2 presents the changes in NGAL and secondary outcomes from the baseline to follow-up for all treatment groups in the ITM population.Changes in NGAL concentrations were −19 456 and −34 985 pg/mL at 4 and 12 months, respectively, from a mean NGAL level of 427 850 pg/mL at baseline in the patients with PsA.After 4 months, 57 of the 117 included patients with PsA (49%) had achieved an ACR 20% response (ACR20), and at 12 months, 40 patients (34%) had achieved ACR20.At 12 months, CRP had decreased from a mean value at baseline of 7.  S1 reports nonresponder imputation by replacing missing data with baseline observation carried forward on all changes in NGAL and secondary outcomes from the baseline to follow-up, for all treatment groups in the ITM population presented in Table 2 (Supplementary material, Appendix 2).Changes in NGAL and PsA outcomes from baseline to follow-up based on the ITM population with adjustment for propensity score (the propensity of having PsA) were determined (not reported), but did not alter the results in Table 2.
The associations between changes in the PsA outcomes and changes in the serum NGAL concentration in patients with PsA evaluated with Spearman's correlations are shown in Table 3.All Spearman correlation coefficients were close to 0.0, indicating no correlation between changes in NGAL and changes in PsA outcomes.The Spearman correlation test results for associations between serum NGAL concentration level and BMI at baseline were, for the PsA group, Spearman correlation coefficient = 0.09 (p = 0.366); PsO group, Spearman correlation coefficient = 0.42 (p = 0.071); healthy control group, Spearman correlation coefficient = 0.44 (p = 0.052); and for all included patients at baseline, Spearman correlation coefficient = 0.12 (p = 0.140).

Discussion
The primary objective of this study was to assess the levels of and changes in serum NGAL, a potential biomarker, in relation to anti-inflammatory treatment in patients with PsA.The study did not observe a decrease in serum NGAL in relation to anti-inflammatory treatment overall.During the 12 month follow-up period, we found that patients with PsA treated with csDMARDs and TNF-α inhibitors showed no clinically significant decreases or increases in serum NGAL concentration levels after 4 and 12 months of treatment.Patients treated with IL-17 inhibitors, on the other hand, showed a remarkable increase of 21%.No obvious explanation or reason for this increase was identified.Patients treated with IL-23p19 comprised the only group that showed a decrease, although with reservations owing to the relatively small sample size.We further found that levels of serum NGAL in 117 patients with active PsA were similar at baseline compared to the 40 participants in the group of 'others' (patients with PsO and healthy controls).No association was observed between serum NGAL concentration level and BMI in either group at baseline.Our results are in contrast to previous studies investigating NGAL in patients with PsA.Jang et al (22) demonstrated that serum concentrations of lipocalin-2 (LCN2) were associated with obesity.Colak et al (11) found that NGAL concentrations were significantly higher in patients with PsA than in healthy controls, and Sokolova et al (13) further found that IL-17 inhibitor and TNF-α inhibitor treatment significantly lowered serum NGAL after 6 months compared to baseline levels in patients with PsA.None of these findings was substantiated in this study.
NGAL has also previously been investigated in other inflammatory conditions, such as PsO, spondyloarthritis, inflammatory bowel disease, and lupus nephritis, with   current study.Furthermore, the study is based on a clinical cohort, which means that blood samples were collected as part of appointments at the outpatient clinic and therefore not at exactly the same time of day.It is known that diurnal variation plays a role when it comes to biomarkers.We did not have repeated data on BMI, and hence our study was not able to determine whether weight loss occurred during the 12 month treatment period and influenced the changes in NGAL.

Conclusion
Based on the results of this study, serum NGAL does not add any value as a biomarker in patients with peripheral PsA, either for disease activity or for monitoring.
8 mg/L by −4.6 mg/ L. SJC decreased by 3.4 and 4.7 after 4 and 12 months, respectively.The decreases for TJC were 6.0 and 7.8 at 4 and 12 months, respectively.The changes in disease activity (DAPSA score) decreased from a mean level of 36.0 at baseline by −14.2 and −18.3, respectively, at 4 and 12 months.A general reduction was already observed at 4 months in all three patient VAS assessments: global, pain, and fatigue.Table

Table 2 .
Changes in neutrophil gelatinase-associated lipocalin (NGAL) and psoriatic arthritis (PsA) outcomes from baseline to follow-up based on the intention-to-monitor (ITM) population.

Table 1 .
Baseline characteristics in the intention-to-monitor (ITM) population.
(27)response assessment in PsO, a disease closely associated with PsA, including serum NGAL.They reported that NGAL, owing to its pro-inflammatory mediation, plays essential roles in the pathogenesis of PsO.They further describe higher levels of NGAL in both serum and skin lesions of patients with PsO compared to controls.With reservations for our relatively small sample size, we did not find the serum concentration of NGAL in the group of patients with PsO to deviate from the healthy controls or the patients with PsA.Abella et al(27)concluded that NGAL was a potential biomarker for rheumatic diseases.However, the findings of our study did not reflect this.The relatively small sample size and lack of follow-up in the control groups should be considered as limitations of the

Table 3 .
Spearman's correlation measures for the strength and direction of monotonic associations between any combination of change in any outcome and change in neutrophil gelatinaseassociated lipocalin (NGAL) after 12 months.-reactive protein; eGFR, estimated glomerular filtration rate; SJC, swollen joint count; TJC, tender joint count; DAPSA, Disease Activity in Psoriatic Arthritis; PASI, Psoriasis Area Severity Index; SPARCC, Spondyloarthritis Research Consortium of Canada Enthesitis Index; VAS, visual analogue scale; HAQ-DI, Health Assessment Questionnaire Disability Index.