Association of Soluble Immune Checkpoint Molecules PD1 and 4‐1BB With CTLA‐4Ig Treatment Response in Early Rheumatoid Arthritis

Abstract

Objective

To investigate whether soluble immune checkpoint molecules in blood are associated with the treatment response to disease‐modifying antirheumatic drugs in early rheumatoid arthritis (eRA).

Methods

This study included 328 Swedish treatment‐naïve patients with eRA from the Nordic Rheumatic Diseases Strategy Trials and Registries (NORD‐STAR) study. Patients were randomized into four treatment groups: methotrexate (MTX) combined with CTLA‐4Ig (n = 90), anti–tumor necrosis factor (n = 83), anti–interleukin‐6 receptor (n = 76), or prednisolone (n = 79). The primary outcome was remission, defined by Clinical Disease Activity Index (CDAI) ≤2.8, assessed at 24 and 48 weeks. Plasma levels of soluble programmed cell death‐1 (sPD1) and soluble 4‐1BB (s4‐1BB) were measured by enzyme‐linked immunosorbent assay at baseline and at weeks 24 and 48 after treatment initiation.

Results

High baseline levels of sPD1 and s4‐1BB were both associated with CDAI remission at 24 weeks (odds ratio 1.31, 95% confidence interval [CI] 1.04–1.66 and odds ratio 1.50, 95% CI 1.07–2.11, respectively) in patients treated with CTLA‐4Ig with MTX, but not in any other treatment groups. Furthermore, baseline levels of sPD1 or s4‐1BB together with proportions of PD1⁺ T follicular helper (TFh) cells predicted treatment response to CTLA‐4Ig with MTX after 48 weeks, achieving 90% and 100% positive predictive value, respectively.

Conclusion

High plasma levels of sPD1 and s4‐1BB are associated with good response to CTLA‐4Ig with MTX therapy in patients with eRA. A combination of sPD1 or 4‐1BB levels and the proportions of PD1⁺ TFh cells in blood at baseline has potential for predicting remission after CTLA‐4Ig treatment.

INTRODUCTION

Rheumatoid arthritis (RA) is a chronic systemic autoimmune disease that affects approximately 1% of the population. ¹ In RA, inflammation primarily targets synovial tissue, leading to pain, swelling, and irreversible joint damage. Although advancement in treatment, particularly the introduction of biologic disease‐modifying antirheumatic drugs (bDMARDs), has greatly improved clinical outcomes for many patients with RA, a substantial proportion still fail to achieve remission within the first year of treatment with these therapies. ² , ³ This highlights the need for biomarkers that can guide the optimal choice of biologic therapy.

Immune checkpoints (ICPs), which include both costimulatory and coinhibitory pathways, are important in regulating immune cell activation and maintaining self‐tolerance. ⁴ In RA, CD4⁺ T cells in the synovium express high levels of ICP molecules, such as the costimulatory receptor 4‐1BB⁵ and the coinhibitory receptor programmed cell death‐1 (PD1). ⁶ , ⁷ In inflammatory conditions, including responses to tumor necrosis factor (TNF) or interleukin‐6 (IL‐6), these ICP receptors can be cleaved from the T cell surface. ⁸ Several soluble ICPs, including soluble PD1 (sPD1) and soluble 4‐1BB (s4‐1BB), are elevated in the blood of patients with early RA (eRA) compared to healthy individuals. ⁵ , ⁹ These soluble receptors bind to their ligands on antigen‐presenting cells (APCs), including dendritic cells, macrophages, and B cells, which disrupts the regulatory signaling between APCs and T cells. ⁸ , ¹⁰

We previously demonstrated that high proportions of circulating CD4⁺ T cells expressing PD1 predict clinical response to CTLA‐4Ig combined with methotrexate (MTX) in patients with eRA after 24 weeks of treatment. ¹¹ Another study showed that patients with higher baseline s4‐1BB levels had better Disease Activity Score 28 with C‐reactive protein (DAS28‐CRP) response to MTX or anti‐TNF with MTX after two years of treatment than those with lower levels. ⁵ However, whether baseline levels of sPD1 or s4‐1BB predict response to other bDMARDs has never been studied. Given that CTLA‐4Ig inhibits CD4⁺ T cell activation, our hypothesis is that circulating levels of sPD1 or s4‐1BB might be associated with remission rates following CTLA‐4Ig treatment. The aim of the present study was to investigate whether baseline plasma levels of the ICP molecules sPD1 or s4‐1BB are associated with the treatment response in patients with eRA after 24 and/or 48 weeks of either CTLA‐4Ig, anti‐TNF, anti‐IL6 receptor (anti‐IL6R), or prednisolone combined with MTX.

PATIENTS AND METHODS

Patients

This post hoc study included 328 Swedish DMARD‐naïve patients with eRA who were included in the Nordic Rheumatic Diseases Strategy Trials and Registries (NORD‐STAR) study (registration number: EudraCT 2011‐004720‐35; ClinicalTrials.gov NCT01491815), a phase 4, investigator‐initiated, observer‐masked clinical trial in which patients were randomized to one of four treatment arms. ² , ³ All participants met the American College of Rheumatology/EULAR 2010 classification criteria for eRA and had received no previous DMARD treatment. Inclusion criteria were ≥18 years of age, ≥2 swollen joints (66‐joint count), ≥2 tender joints (68‐joint count), a DAS28‐CRP of >3.2, and a self‐reported symptom duration of less than 24 months. Participants had to be positive for either rheumatoid factor (RF) or anti–citrullinated protein antibodies (ACPAs) or have a CRP level of ≥10 mg/L. Study participants were recruited at 11 rheumatology clinics in Sweden, and all participants gave informed consent to participation. The study was conducted in compliance with the Helsinki Declaration and approved by regional ethics committees of Stockholm (Dnr. 2011/2069–31/4 and amendment Ö‐37‐2023; measurement of ICP molecules in blood) and Gothenburg (Dnr. 691–12 and amendment T270‐13; measurement of T cell subpopulations in blood). The data supporting the conclusions of this article will be shared on reasonable request to the corresponding author TS or AR.

Study format and clinical evaluation

The complete NORD‐STAR study protocol ¹² and primary clinical outcome results at 24 and 48 weeks have been published elsewhere. ² , ³ In brief, all patients were randomized into four treatment arms. All patients received MTX escalated to 25 mg/week within the first 4 weeks, combined with either CTLA‐4Ig (abatacept), anti‐TNF (certolizumab‐pegol), anti‐IL6R (tocilizumab), or oral prednisolone (starting at 20 mg per day, which was reduced to 5 mg within 9 weeks and stopped at 36 weeks). Oral steroids were not allowed, except for one two‐week rescue treatment, in patients who received bDMARDs. Disease activity was assessed before treatment initiation (baseline) and 24 and 48 weeks after initiation by the Clinical Disease Activity Index (CDAI). CDAI was calculated as the sum of swollen joint counts in 28, tender joint counts in 28, patient global score of disease activity and physician global score of disease activity. The primary clinical outcome was achievement of remission defined as CDAI ≤ 2.8. ACPA and RF positivity was determined at baseline by multiplexed anti–cyclic citrullinated peptide test (BioPlex from BioRad) and nephelometry (Beckman Coulter), respectively, and positivity was determined according to cutoff levels in the local clinical immunology laboratories. Blood samples analyzed in this study were drawn at baseline (within 1–2 weeks after RA diagnosis, but before initiation of treatment) and after 24 and 48 weeks of treatment.

Analysis of soluble ICP molecules

The concentrations of sPD1 and s4‐1BB were measured in plasma from all 328 patients with eRA at baseline and at 24 and 48 weeks of treatment using DuoSet enzyme‐linked immunosorbent assay kit (R&D Systems) according to manufacturer's instructions. No interference by RF was found when patient plasma samples were analyzed with different concentrations of HeteroBlock (Omega Biologicals, Inc).

Statistical analysis

Differences in soluble ICP molecule concentrations between groups were analyzed by nonparametric Mann‐Whitney U‐test or Kruskal‐Wallis test with Dunn's test for multiple comparisons when more than two groups were analyzed, as described in figure legend 1 and 2 and supplementrary figure legend 1. Pairwise comparisons of longitudinal data were performed using Friedman's nonparametric test with Dunn's test for multiple comparisons. Patients with missing data for either week 24 or week 48 were not included in the pairwise comparisons. Spearman's correlation coefficient was applied to assess the correlation between soluble ICP molecule concentrations and disease activity measures or T cell proportions (GraphPad Prism, version 10.2.0). Association between soluble ICP molecule levels and remission (yes or no) at weeks 24 or 48 was assessed using binary logistic regression analyses and presented as unadjusted or adjusted odds ratios (ORs), that is, the change in the odds of remission for a one‐unit increase in the predictor variable. Adjustments were made for age and baseline erythrocyte sedimentation rate (ESR), which were the two potential confounders that contributed significantly to the model (R with RStudio PBC). Receiver operating characteristic (ROC) curves with area under the curve (AUC) was used to test whether levels of sPD1, s4‐1BB, or combined with proportions of PD1⁺ TFh cells could predict remission (yes or no) at weeks 24 or 48. Positive predictive values (PPVs) and negative predictive values were calculated based on the remission rate with cutoff values set to the probability that maximizes the proportion of correctly classified patients for each outcome (ie, remission at 24 or 48 weeks; R and Prism). In all logistic regression models soluble ICP molecule concentrations were log‐transformed as their distributions were skewed. Statistical significance was defined as P ≤ 0.05.

RESULTS

sPD1 levels in blood decrease after treatment with MTX in combination with CTLA‐4Ig or anti‐TNF for 48 weeks

Baseline demographic, clinical characteristics, and levels of sPD1 and 4‐1BB were comparable across the four treatment arms (Supplementary Table S1). First, we investigated the effect of treatment with MTX combined with CTLA‐4Ig, anti‐TNF, anti‐IL6R, or prednisolone on sPD‐1 and s4‐1BB concentrations in blood from 328 patients with eRA. Treatment with CTLA‐4Ig and anti‐TNF, but not anti‐IL6R or prednisolone, significantly reduced sPD1 levels between baseline and week 48, whereas s4‐1BB levels were unaffected by the treatments (Supplementary Figure S1A and B). CDAI significantly decreased from baseline to week 48 in all treatment arms (Supplementary Figure S1C). Additionally, double seropositive (ACPA⁺RF⁺) patients had higher sPD1 levels compared to ACPA⁺RF⁻ patients but not compared to ACPA⁻RF⁺ or to ACPA⁻RF⁻ patients, whereas s4‐1BB levels did not differ between seropositive and seronegative patient groups (Supplementary Figure S1D and E). At baseline, neither sPD1 nor s4‐1BB levels correlated with disease activity measures (Supplementary Table S2).

Association of sPD1 and s4‐1BB with remission after treatment with CTLA‐4Ig in patients with eRA

Next, we examined whether baseline sPD1 levels were associated with achieving remission (CDAI ≤ 2.8, yes or no) at weeks 24 and 48 in each treatment group. In the CTLA‐4Ig with MTX arm, but not in the other treatment arms, baseline sPD1 levels were significantly higher in patients who achieved remission at 24 weeks compared to those who did not (Figure 1A–D). Univariable logistic regression revealed that increased baseline sPD1 levels were associated with higher odds of achieving remission at 24 weeks (OR 1.31, 95% confidence interval [CI] 1.04–1.66) but not at 48 weeks (OR 1.21, 95% CI 0.96–1.54). This association persisted after adjustment for age and baseline ESR (OR 1.32, 95% CI 1.03–1.68; Figure 1E) and after exclusion of seronegative (ACPA⁻RF⁻) patients (OR 1.27, 95% CI 1.00–1.66 and OR 1.21, 95% CI 0.95–1.59, for remission at weeks 24 and 48, respectively). In ROC analysis, baseline sPD1 together with age and baseline ESR predicted remission both at week 24 (solid line, AUC 0.76, 95% CI 0.66–0.86) and week 48 (dotted line, AUC 0.72, 95% CI 0.61–0.82) in patients treated with CTLA‐4Ig with MTX (Supplementary Figure S2A).

Figure 1.

Baseline sPD1 levels in patients who did or did not achieve remission (Clinical Disease Activity Index [CDAI] ≤ 2.8) after 24w or 48w of treatment with MTX combined with (A) CTLA‐4Ig (abatacept, n = 90), (B) anti‐TNF (certolizumab‐pegol, n = 83), (C) anti‐IL6R (tocilizumab, n = 76), or (D) prednisolone (n = 79). Mann‐Whitney U‐test. (E) Logistic regression presented as ORs with 95% CIs. The outcome variable was remission (CDAI score ≤ 2.8, yes or no) at 24w or 48w of CTLA‐4Ig with MTX treatment, and the response variable was baseline plasma levels of sPD1. Adjustment was made for age and baseline erythrocyte sedimentation rate. 24w, 24 weeks; 48w, 48 weeks; anti‐IL6, anti–interleukin‐6 receptor; anti‐TNF, anti–tumor necrosis factor; CI, confidence interval; MTX, methotrexate; PD1, programmed cell death‐1; OR, odds ratio; Rem, remission; sPD1, soluble programmed cell death‐1.

In the CTLA‐4Ig with MTX treatment arm, baseline s4‐1BB levels were significantly higher in patients who achieved remission at both week 24 and week 48 compared to those who did not (Figure 2A), whereas there were no differences in s4‐1BB levels between these groups in any of the other treatment arms (Figure 2B–D). Univariable logistic regression revealed that increased baseline log s4‐1BB levels were associated with higher odds of achieving remission both at 24 weeks (OR 1.50, 95% CI 1.07–2.11) and at 48 weeks (OR 1.44, 95% CI 1.01–2.07), but this association was lost after adjustment (Figure 2E). In ROC analysis, baseline s4‐1BB together with age and baseline ESR predicted remission at week 24 (solid line, AUC 0.73, 95% CI 0.63–0.83) and at week 48 (dotted line, AUC 0.71, 95% CI 0.61–0.82; Supplementary Figure S2B). A logistic regression model combining baseline sPD1 and s4‐1BB levels was not performed due to a strong correlation between the levels of these two molecules (Figure 3A).

Figure 2.

Baseline s4‐1BB levels in patients who did or did not achieve remission (Clinical Disease Activity Index [CDAI] ≤ 2.8) after 24w or 48w of treatment with MTX combined with (A) CTLA‐4Ig (abatacept, n = 90), (B) anti‐TNF (certolizumab‐pegol, n = 83), (C) anti‐IL6R (tocilizumab, n = 76), or (D) prednisolone (n = 79). Mann‐Whitney U‐test. (E) Logistic regression presented as ORs with 95% CIs. The outcome variable was remission (CDAI score ≤ 2.8, yes or no) at 24w or 48w of CTLA‐4Ig with MTX treatment, and the response variable was baseline plasma levels of s4‐1BB. Adjustment was made for age and baseline erythrocyte sedimentation rate. 24w, 24 weeks; 48w, 48 weeks; anti‐IL6R, anti–interleukin‐6 receptor; anti‐TNF, anti–tumor necrosis factor; CI, confidence interval; MTX, methotrexate; PD1, programmed cell death‐1; OR, odds ratio; Rem, remission; s4‐1BB, soluble 4‐1BB.

Figure 3.

Correlation between (A) plasma levels of sPD1 and s4‐1BB at baseline (n = 328) and (B) plasma levels of sPD1 or s4‐1BB and blood proportions of PD1⁺ TFh cells at baseline (n = 60). ρ indicates Spearman's correlation coefficient. (C and D) Receiver operating characteristic (ROC) curves showing the diagnostic value of baseline PD1⁺ TFh cell proportions alone (dotted black line, n = 17) or combined with baseline sPD1 (solid black line, n = 17) or s41‐BB levels (solid gray line, n = 17) for achieving remission at 24w and 48w in patients treated with MTX and CTLA‐4Ig (abatacept). The ROC curve for the model combing s4‐1BB and PD1⁺ TFh (solid gray line) in (D) is only representative, as this model could not converge due to nearly perfect separation. AUC, area under the curve; CI, confidence interval; ICP, immune checkpoint; NPV, negative predictive value; PD1, programmed cell death‐1; PPV, positive predictive value; Rem, remission; s4‐1BB, soluble 4‐1BB; sPD1, soluble programmed cell death‐1; TFh, T follicular helper.

Baseline sPD1 or s4‐1BB in combination with PD1 ⁺ TFh cells in blood predicting remission in patients with eRA treated with CTLA‐4Ig

We have previously demonstrated ¹¹ that the proportion of PD1⁺ TFh cells in blood predicts remission at week 24 in a subset of 17 patients treated with CTLA‐4Ig with MTX, who are also part of this study (Supplementary Table S3). Because no multicollinearity was observed between the proportions of PD1⁺ TFh cells and sPD1 or s4‐1BB levels at baseline (Figure 3B), we analyzed whether the addition of baseline sPD1 or s4‐1BB levels could enhance the predictive value of PD1⁺ TFh cell proportions for remission after CTLA‐4Ig with MTX treatment. ¹¹ For remission at 24 weeks, combining sPD1 or s4‐1BB with PD1⁺ TFh cell proportions did not improve the predictive values compared to PD1⁺ TFh cell proportions alone (Figure 3C). However, for remission at 48 weeks, inclusion of sPD1 in the model improved the AUC value (0.94, 95% CI 0.84–1.0), as well as sensitivity (90%), specificity (86%), and PPV (90%) compared to PD1⁺ TFh cell proportions alone (Figure 3D). Combining baseline s4‐1BB levels with PD1⁺ TFh cell proportions resulted in nearly perfect separation (Supplementary Figure S3) with an AUC value of 1, thus improving the predictive value (Figure 3D).

DISCUSSION

We hypothesized that circulating levels of sPD1 or s4‐1BB might serve as biomarkers for remission after treatment with the Th cell costimulation blocker CTLA‐4Ig. Here, we found that elevated baseline levels of sPD1 and s4‐1BB were associated with achieving remission following MTX combined with CTLA‐4Ig treatment, but not with anti‐TNF, anti‐IL6R, or prednisolone. Furthermore, combining sPD1 or s4‐1BB levels with PD1⁺ TFh cell proportions predicted response to CTLA‐4Ig with MTX treatment for 48 weeks.

To our knowledge, this is the first study to investigate the relationship between sPD1 and treatment response in eRA. In an adjusted regression model, we found a relationship between higher sPD1 levels at baseline and increased odds of achieving remission in the CTLA‐4Ig treatment arm. Previous studies, including our own, have shown that higher baseline PD1⁺ TFh cell proportions can predict response to CTLA‐4Ig treatment. ¹¹ , ¹³

Although we found no correlation between sPD1 levels and proportions of PD1⁺ TFh at baseline, the predictive capacity of sPD1 could partly derive from circulating TFh cells. The lack of correlation also suggests that sPD1 may originate from other cellular sources. Indeed, in patients with eRA, PD1 overexpression by activated CD4⁺ T cells has been documented in both lymph nodes, including PD1⁺ TFh, and synovial tissue, including PD1⁺ T peripheral helper cells. ⁶ , ⁹ , ¹⁴ In line with this, sPD1 levels are higher in synovial fluid compared to plasma in paired samples from patients with established RA, ¹⁵ , ¹⁶ which raises the possibility that sPD1 could leak from the inflamed synovium into peripheral blood. Of note, sPD1 do not appear to have similar suppressing properties as membrane bound PD1 because the soluble form has been shown to reduce PD1‐mediated suppression of CD4⁺ T cells by monocytes in vitro. ⁸ This could partly explain our results that high sPD1 is associated with remission when treated with a therapy that blocks T cell costimulation.

Regarding s4‐1BB, a previous study including patients with eRA has reported that higher baseline s4‐1BB levels are associated with better responses to MTX monotherapy and/or anti‐TNF with MTX treatment after two years but not after one year. ⁵ Our study supports that s4‐1BB levels do not predict treatment response to anti‐TNF with MTX treatment at the one‐year follow‐up. Given the short follow‐up period in the present study, we cannot rule out the possibility of an association between s4‐1BB and treatment response to anti‐TNF at a later time point. Instead, we observed that baseline s4‐1BB levels were higher in patients who achieved remission after weeks 24 and 48 of CTLA‐4Ig with MTX therapy compared to those who did not. The role of s4‐1BB in eRA is still unclear; however, it appears to have antagonistic activity, including inhibition of T cell proliferation and IL‐2 production, ¹⁰ and may be released by T cells to control their own degree of activation. ¹⁷ It is tempting to speculate that patients with high levels of s4‐1BB molecules have increased CD4⁺ T cell activity in the joint. Alternatively, patients with high s4‐1BB levels have suppressed T cell activity owing to the antagonistic properties of s4‐1BB. Although a causal relationship between s4‐1BB levels and CD4⁺ T cell activity is not demonstrated in the present study, our results that high levels s4‐1BB is associated with remission only after CTLA‐4Ig therapy suggest that levels of s4‐1BB are associated with T cell activity.

Because first‐line treatment with CTLA‐4Ig with MTX in eRA induced the highest remission levels at week 48 compared to MTX and glucocorticoids in the NORD‐STAR cohort, ³ reliable biomarkers for selecting the patients who have the highest chance of treatment success might make immediate start with CTLA‐4Ig feasible. Here, the combination of sPD1 or s4‐1BB levels with PD1⁺ TFh cell proportions could provide a PPV of 90% or 100%, respectively. However, it should be noted that these results only apply to a population with a similar prevalence of remission. Moreover, owing to the relatively small sample size in the analysis combining sPD1 and s4‐1BB with and PD1⁺ TFh cell proportions, this finding should be interpreted cautiously and needs to be replicated in a larger cohort.

AUTHOR CONTRIBUTIONS

All authors contributed to at least one of the following manuscript preparation roles: conceptualization AND/OR methodology, software, investigation, formal analysis, data curation, visualization, and validation AND drafting or reviewing/editing the final draft. As corresponding author, Ms Selldén confirms that all authors have provided the final approval of the version to be published and takes responsibility for the affirmations regarding article submission (eg, not under consideration by another journal), the integrity of the data presented, and the statements regarding compliance with institutional review board/Declaration of Helsinki requirements.

Supporting information

Appendix S1: Supplementary Information

Disclosure form: