Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2024 May 3;63(6):1720-1732.
doi: 10.1093/rheumatology/kead466.

Lymphocyte subset phenotyping for the prediction of progression to inflammatory arthritis in anti-citrullinated-peptide antibody-positive at-risk individuals

Innocent Anioke  1   2 , Laurence Duquenne  1   3 , Rekha Parmar  1 , Kulveer Mankia  1   3 , Farag Shuweihdi  4 , Paul Emery  1   3 , Frederique Ponchel  1
Affiliations

Lymphocyte subset phenotyping for the prediction of progression to inflammatory arthritis in anti-citrullinated-peptide antibody-positive at-risk individuals

Innocent Anioke et al. Rheumatology (Oxford). .

Abstract

Objectives: Inflammatory arthritis (IA) is considered the last stage of a disease continuum, where features of systemic autoimmunity can appear years before clinical synovitis. Time to progression to IA varies considerably between at-risk individuals, therefore the identification of biomarkers predictive of progression is of major importance. We previously reported on the value of three CD4+T cell subsets as biomarkers of progression. Here, we aim to establish the value of 18 lymphocyte subsets (LS) for predicting progression to IA.

Methods: Participants were recruited based on a new musculoskeletal complaint and being positive for anti-citrullinated-peptide antibody. Progression (over 10 years) was defined as the development of clinical synovitis. LS analysis was performed for lymphocyte lineages, naive/memory subsets, inflammation-related cells (IRC) and regulatory cells (Treg/B-reg). Modelling used logistic/Cox regressions.

Results: Of 210 patients included, 93 (44%) progressed to IA, 41/93 (44%) within 12 months (rapid progressors). A total of 5/18 LS were associated with progression [Treg/CD4-naïve/IRC (adjusted P < 0.0001), CD8 (P = 0.021), B-reg (P = 0.015)] and three trends (NK-cells/memory-B-cells/plasmablasts). Unsupervised hierarchical clustering using these eight subsets segregated three clusters of patients, one cluster being enriched [63/109(58%)] and one poor [10/45(22%)] in progressors. Combining all clinical and LS variables, forward logistic regression predicted progression with accuracy = 85.7% and AUC = 0.911, selecting smoking/rheumatoid-factor/HLA-shared-epitope/tender-joint-count-78 and Treg/CD4-naive/CD8/NK-cells/B-reg/plasmablasts. To predict rapid progression, a Cox regression was performed resulting in a model combining smoking/rheumatoid factor and IRC/CD4-naive/Treg/NK-cells/CD8+T cells (AUC = 0.794).

Conclusion: Overall, progression was predicted by specific LS, suggesting potential triggers for events leading to the development of IA, while rapid progression was associated with a different set of subsets.

Keywords: RA; T cells; biomarkers; immunological techniques; laboratory diagnosis; lymphocytes; rheumatic diseases.

PubMed Disclaimer

Figures

Figure 1.
Figure 1.
Frequency of lineage and lymphocytes subsets in at-risk progressors vs non-progressors. LS were analysed by flow cytometry and data displayed as violin plots (each dot representing a patient) for Progressor (Prog, n = 93) and non-Progressors (Non-Prog, n = 117). Star (*) indicates LS that were normalised as previously described [15]. P-value corrected for multiple testing (MWU test) are indicated when significant and # designate trends
Figure 2.
Figure 2.
Unsupervised hierarchical clustering of the 8 subsets associated with progression to IA (n = 210). An unsupervised hierarchical clustering algorithm was applied to log transformed frequencies for 8 LS and results are displayed as a heat-map of data. This clustering algorithm builds relationships between LS frequencies based on spearman rank correlations, and segregated patients into 3 clusters (I, II and III), annotated with the % of progressors. The first group of LS (plasmablasts only) shows particularly high frequencies in patient cluster-II . The 2nd group with 3 subsets (naïve CD4+T cells, B-reg, and NK CD56high) defined Cluster-II. The third group with 3 subsets (Memory B-cells, Treg and CD8+T cells) allows to define both Cluster I and II. The last group (IRC-CD4+T only) shows exclusively high frequencies in cluster-III with lower frequencies in Cluster-I and II. The proportion of progressors to IA in the 3 clusters was significantly different (P < 0.0001). The bar with shades of colours (right hand-side) indicated the frequency observed for each LS from highest to the lowest
Figure 3.
Figure 3.
Performances of the models. (A) AUROC graphical representation of the logistic regression models. Binary logistic regression models of the occurrence of progression to inflammatory arthritis (IA) were constructed using Model 1 (Clinical data only) for 10 parameters (thin line), Model 2 (flow-data only) for 18 subsets (dotted line) and Model 3 (clinical + flow data, thick line). Model 1 (AUC = 0.744 95%CI 0.678–0.810) was inferior to Model 2 (AUC = 0.862, 0.814–0.910) and Model 3 still showed added value (AUC = 0.911, 0.871–0.951). (B) Survival curve based on classification using Model 3. Survival plot analysis was performed after patients were dichotomised for high-risk (black line, n = 65/210) and low-risk (grey line, n = 145/210) based on individual probability (>0.80%) calculated from the logistic regression. (C) Variables contribution to Model 6: This showed the relative importance order of the predictors in the model with Treg as the most discriminating biomarker for rapid progression followed by naiveCD4, smoking, CD8, RF, IRC-CD4 and finally NK-CD56dim cells. (D) Survival curve based on the Cox regression for rapid progression. Survival plot analysis was performed after patients were dichotomised for high-risk (black line, n = 22/158) and low-risk (grey line, n = 136/158) based on individual hazard (>2) calculated from the Cox regression. (E) Overall performance of the prediction model using 1 to 4 flow cytometry panels. Individual participants’ probability for progression was dichotomised into high/low-risk groups (based on 80% specificity) in five logistic regression models including the demographic/clinical data only first and then, sequentially adding data from 1, 2, 3 and then 4 flow-cytometry panels. Numbers of patients in both risk groups are displayed against the number of progressors (black bars) and non-progressors (open bar)
See this image and copyright information in PMC

Similar articles

See all similar articles

Cited by

References

    1. Hunt L, Emery P. Defining populations at risk of rheumatoid arthritis: the first steps to prevention. Nat Rev Rheumatol 2014;10:521–30. - PubMed
    1. Gerlag DM, Raza K, van Baarsen LGM et al. EULAR recommendations for terminology and research in individuals at risk of rheumatoid arthritis: report from the Study Group for Risk Factors for Rheumatoid Arthritis. Ann Rheum Dis 2012;71:638–41. - PMC - PubMed
    1. Finckh A, Liang MH, van Herckenrode CM, de Pablo P. Long-term impact of early treatment on radiographic progression in rheumatoid arthritis: a meta-analysis. Arthritis Rheum 2006;55:864–72. - PubMed
    1. Nam JL, Hunt L, Hensor EM, Emery P. Enriching case selection for imminent RA: the use of anti-CCP antibodies in individuals with new non-specific musculoskeletal symptoms - a cohort study. Ann Rheum Dis 2016;75:1452–6. - PubMed
    1. Mankia K, Siddle H, Di Matteo A et al. A core set of risk factors in individuals at risk of rheumatoid arthritis: a systematic literature review informing the EULAR points to consider for conducting clinical trials and observational studies in individuals at risk of rheumatoid arthritis. RMD Open 2021;7:1–16. - PMC - PubMed

Substances