Systemic biomarkers of treatment response to methotrexate in people with painful knee osteoarthritis: a biological substudy of the PROMOTE randomised controlled clinical trial
File(s)
Accepted version
Author(s)
Watt, Fiona
Liang, Jiahao
Type
Journal Article
Abstract
Objective: Stratification of therapeutic responses may help identify efficacious therapies for osteoarthritis (OA). In the PROMOTE randomised trial, participants with
elevated baseline high-sensitivity C-reactive protein (hs-CRP) showed greater pain reduction after methotrexate treatment. We set out to interrogate a broader panel of
serum/plasma inflammatory response markers relevant to methotrexate actions as potential biomarkers of therapeutic effect. Our objectives were to: (i) characterize changes in these systemic markers during methotrexate treatment; determine whether
(ii) baseline levels or (iii) changes in any marker during treatment were associated with treatment response; and (iv) compare these findings with the more established clinical inflammatory marker, hs-CRP.
Design: Plasma/serum samples from participants in PROMOTE’s biological substudy were analysed for 35 inflammatory markers at baseline (pre-treatment) and at 6-months (post-treatment), by MesoScale V-plex multiplex assay. Those with paired biological and clinical data at both baseline and 6-months were included in the
substudy analysis set. Relationships between markers and overall data structure were assessed by Pearson correlation and Principal Component analysis. Associations between markers (baseline levels or change over time) and change in average knee
pain severity in past week (numerical rating scale, NRS) were evaluated by univariable linear regression, adjusting for baseline age, sex, and body mass index. Least Absolute Shrinkage and Selection Operator (LASSO) regression with bootstrap resampling enabled marker selection. Benjamini Hochberg correction adjusted for multiple testing (Padj).
Results: 87 participants with paired blood marker and clinical data were eligible for substudy analysis.18/35 markers were quantifiable and analysed. Systemic IL-8 and TNF-α levels decreased (Padj=0.015, 0.048 respectively) while IL-15 increased (Padj=0.033) with methotrexate treatment over 6-months. Analysing within this active treatment randomised arm, higher baseline IFN-γ was associated with greater reduction in NRS pain change (0.66 [0.01, 1.31], P=0.047), as was decreasing TNF-α over 6-months (2.25 [0.00, 4.5], P=0.049). LASSO identified higher IFN-γ, lower plasma IL-15 and IL-16, and younger age as the most important baseline predictors
of pain improvement. hs-CRP was highly selected by LASSO for treatment response in both arms. In a secondary univariate treatment arm-by-biomarker interaction analysis, of the 19 markers, only hs-CRP showed consistent effects in adjusted models
(at baseline, coeffic. 2.34 [0.53, 4.15], P=0.001; change over 6-months, (0.36 [0.06, 0.66], P=0.018).
Conclusions: Blood measurement of IFN-γ, TNF-α, IL-15 and IL-16 as well as hs-CRP could act as potential markers to stratify the treatment response by average knee pain to methotrexate in knee osteoarthritis.
elevated baseline high-sensitivity C-reactive protein (hs-CRP) showed greater pain reduction after methotrexate treatment. We set out to interrogate a broader panel of
serum/plasma inflammatory response markers relevant to methotrexate actions as potential biomarkers of therapeutic effect. Our objectives were to: (i) characterize changes in these systemic markers during methotrexate treatment; determine whether
(ii) baseline levels or (iii) changes in any marker during treatment were associated with treatment response; and (iv) compare these findings with the more established clinical inflammatory marker, hs-CRP.
Design: Plasma/serum samples from participants in PROMOTE’s biological substudy were analysed for 35 inflammatory markers at baseline (pre-treatment) and at 6-months (post-treatment), by MesoScale V-plex multiplex assay. Those with paired biological and clinical data at both baseline and 6-months were included in the
substudy analysis set. Relationships between markers and overall data structure were assessed by Pearson correlation and Principal Component analysis. Associations between markers (baseline levels or change over time) and change in average knee
pain severity in past week (numerical rating scale, NRS) were evaluated by univariable linear regression, adjusting for baseline age, sex, and body mass index. Least Absolute Shrinkage and Selection Operator (LASSO) regression with bootstrap resampling enabled marker selection. Benjamini Hochberg correction adjusted for multiple testing (Padj).
Results: 87 participants with paired blood marker and clinical data were eligible for substudy analysis.18/35 markers were quantifiable and analysed. Systemic IL-8 and TNF-α levels decreased (Padj=0.015, 0.048 respectively) while IL-15 increased (Padj=0.033) with methotrexate treatment over 6-months. Analysing within this active treatment randomised arm, higher baseline IFN-γ was associated with greater reduction in NRS pain change (0.66 [0.01, 1.31], P=0.047), as was decreasing TNF-α over 6-months (2.25 [0.00, 4.5], P=0.049). LASSO identified higher IFN-γ, lower plasma IL-15 and IL-16, and younger age as the most important baseline predictors
of pain improvement. hs-CRP was highly selected by LASSO for treatment response in both arms. In a secondary univariate treatment arm-by-biomarker interaction analysis, of the 19 markers, only hs-CRP showed consistent effects in adjusted models
(at baseline, coeffic. 2.34 [0.53, 4.15], P=0.001; change over 6-months, (0.36 [0.06, 0.66], P=0.018).
Conclusions: Blood measurement of IFN-γ, TNF-α, IL-15 and IL-16 as well as hs-CRP could act as potential markers to stratify the treatment response by average knee pain to methotrexate in knee osteoarthritis.
Date Acceptance
2026-07-01
Citation
Osteoarthritis and Cartilage
ISSN
1063-4584
Publisher
Elsevier
Journal / Book Title
Osteoarthritis and Cartilage
Copyright Statement
Copyright This paper is embargoed until publication. Once published the Version of Record (VoR) will be available on immediate open access.
License URL
Publication Status
Accepted