medwireNews: Analysis of pretreatment follicular helper T (TFH) cell subpopulations could help to identify which patients with newly diagnosed type 1 diabetes are likely to respond to abatacept immunotherapy, suggests research published in Nature Immunology.
The study authors, led by Lucy Walker from University College London in the UK, analyzed cryopreserved peripheral blood mononuclear cell samples from 50 participants of the Type 1 Diabetes TrialNet TN09 trial. This study previously demonstrated that abatacept slows reduction in beta-cell function relative to placebo among patients with new-onset disease, but Walker et al note that “it was clear that some individuals benefited more than others” from treatment with the CTLA-4–immunoglobulin fusion protein.
In the present study, flow cytometry analysis revealed that the proportion of TFH cells decreased significantly from baseline to the 1- and 2-year follow-up points among abatacept-treated patients, but not among those treated with placebo. The researchers note that T cells expressing the markers ICOS and CXCR5 were “[t]he major cell population contributing to this separation.”
Walker and colleagues then used an ensemble machine-learning method with flow cytometry data from pretreatment samples to build a predictive model of abatacept response based on T-cell phenotype. This was done based on samples from the 10 participants with the best abatacept response and the 10 participants with the worst response, as indicated by C-peptide levels at the 2-year follow-up.
In area under the receiver operating characteristic curve analysis, this model correctly distinguished between abatacept responders and non-responders on 81% of occasions.
These findings suggest “that TFH cells are a sensitive biomarker of costimulation blockade […] and reveal that pretreatment TFH profiles can be used to predict response to abatacept,” write the researchers.
They add that “[t]he two features that emerged as being most important in predicting C-peptide retention after abatacept treatment were ICOS+ TFH […] and CXCR5+ naïve cells.”
The study authors explain that because abatacept has only been trialed once in patients with new-onset type 1 diabetes, they were not able to test their model in an independent dataset and therefore “sought an alternative means of validation.” This was done using the machine-learning algorithm CellCnn, a data-driven approach designed to identify rare cell subtypes associated with disease status.
CellCnn validation demonstrated that two specific clusters of TFH cells at baseline – ICOS+PD-1+ TFH and ICOS+PD-1– TFH – were decreased in response to abatacept treatment. Higher pretreatment frequencies of ICOS+ TFH were associated with poor clinical response, whereas higher frequencies of ICOS–PD-1– TFH cells were associated with a good response.
“ICOS appears to be the most discerning cellular marker associated with preservation of beta cell function after abatacept treatment as assessed by two independent approaches,” says the team, adding that data from a mouse model in their study provided “additional support” for this hypothesis.
Walker et al conclude: “It is important to confirm these findings in a separate cohort of patients and explore their wider applicability.
“For example, it remains to be established whether the T cell phenotypes we identified can predict the response to abatacept in other clinical settings, such as rheumatoid arthritis, or whether they are specific to [type 1 diabetes].”
medwireNews is an independent medical news service provided by Springer Healthcare Ltd. © 2020 Springer Healthcare Ltd, part of the Springer Nature Group