Tracking T1D Progression in At-Risk Individuals Through Transcriptional Networks

Researchers identified gene expression signatures associated with risk of type 1 diabetes (T1D).

By analyzing longitudinal blood transcriptomes of samples from 400 individuals enrolled in The Environmental Determinants of Diabetes in the Young (TEDDY) study prior to type 1 diabetes (T1D) onset or islet autoimmunity, researchers concluded results indicate the disease is characterized by early and longitudinal changes in gene expression.

The findings, published in Science Translational Medicine, inform both the immunopathology of T1D progression and help to facilitate prediction of its course, authors wrote.

Although the exact cause of T1D is unknown, the strong genetic association of Human leukocyte antigen (HLA) and other immune variants, in addition to progressive development of pancreatic islet b cell autoantibodies (IAbs), point to an autoimmune pathogenesis.

Using data from TEDDY—which aims to identify gene-environment interactions causing T1D in high-risk infants—investigators conducted transcriptional network analyses of gene expression microarray data to identify early changes in whole blood gene expression in healthy infancy or those tracking with progression to both islet autoimmunity and T1D. In addition, to estimate individual risk of T1D progression, researchers built a predictive model which included gene expression and islet autoantibodies. Age-matched healthy controls were also included in the study.

Data from matched case control cohorts, including 2013 whole blood transcriptomes sampled longitudinally, were divided depending on if the individual developed islet autoimmunity or T1D.

“As co-expression patterns were preserved in disease and control groups, we constructed a co-expression network based on all samples considered together and applied linear mixed modeling (lmm) to whole blood modular eigengenes from this combined network,” researchers explained.

Taken together, data from this portion of the study demonstrated marked gene expression changes during healthy infancy, “highlighting the dynamic context in which autoimmune diseases such as T1D occur.”

By stratifying the T1D cohort by target of first appearing autoantibody, researchers were able to compare modular gene expression changes in autoantibodies to insulin (IAAs)-first and GAD antibody (GADA)–first subgroups with controls. This analysis showed:

  • Whereas disease-specific longitudinal changes were not apparent in the T1D cohort on the whole, distinct gene expression signatures showed clear age-independent association with time to T1D onset in IAbs subgroups.
  • Among IAAs cases, 1 dominant signature (IAAsig) showed an early increase in expression with a later secondary increase before diagnosis, a pattern not seen in most matched controls.
  • In GADA cases, a group of 4 closely correlated signatures termed together as GADAsig showed an age-independent decrease toward T1D onset; a pattern absent from matched controls and occurring closer to diagnosis in contrast to the earlier increases in the IAAsig.

Further enrichment analyses of IAAsig and GADAsig revealed “IAAsig genes showed strong, specific enrichment for natural killer (NK) cell–specific transcripts, transcription factors, and kinases and correlated with deconvoluted percentage of NK cells, and to a lesser but still significant extent with CD4+ memory T cells…comparison with deconvolved cell type frequencies indicated strongest associations with reduced percentage of CD4+ memory T cells and NK cells, with a relative increase in an activated NK phenotype.”

This finding indicates early stages of T1D pathogenesis are associated with different immune cell trajectories involving NK and CD4+ memory T cells, researchers wrote.

To determine if specific changes in gene expression occur around the onset of islet autoimmunity, analyses showed that among 50 modular signatures showing significant associations with its onset, 1 dominant signature (IAsig) was associated with seroconversion in both GADA and IAAs subgroups. Six more signatures also showed a significant age-independent association with islet autoimmunity onset in both groups.

To validate their findings, investigators assessed an independent cohort of IAbs and T1D cases which included 356 samples from 58 individuals. Here, the NK signature increased toward T1D onset with a later decline in matched controls who did not develop the disease, confirming an independent association of an NK cell-enriched transcriptional signature with both IAbs seroconversion and rate of progression to T1D, authors explained.

Further analyses also revealed that in early infancy 4 modules were associated with the rate of subsequent progression toward T1D. Enrichment and deconvolution correlation analyses of the modules “indicate early high expression tumor necrosis factor (TNF)–enriched monocyte signature and early low expression of a B lymphoblast signature were associated with slower progression to T1D onset.”

Using a multivariate Bayesian joint model, investigators tested serial prediction of T1D over a fixed time of 12 months (using all cumulative data available at each time point) and prediction at 1.5 years of age over a serially increasing future horizon. “Gene expression measures provided greatest support for prediction when measured early (up to 18 months) to predict T1D risk over a longer time horizon (up to 5 years in this dataset),” they wrote.

Although the association of the NK transcriptional signature was validated, the mechanism linking NK cells to insulitis warrants further investigation, marking a limitation to the study.

“Incorporating gene expression signatures alongside patterns of islet autoimmunity seroconversion facilitates robust prediction of individual risk, validated in an independent cohort,” authors concluded. “This creates the potential for early monitoring of at-risk infants for T1D onset, facilitating the prevention of severe complications such as ketoacidosis, effective trialing of preventive therapies, or the identification of targets for immuno-modulation.”

Reference

Xhonneux L, Knight O, Lernmark Å, et al. Transcriptional networks in at-risk individuals identify signatures of type 1 diabetes progression. Sci Transl Med. Published online March 31, 2021. doi:10.1126/scitranslmed.abd5666